<%BANNER%>

Chemical Vapor Deposition of WNxCy Thin Films for Diffusion Barrier Application

Permanent Link: http://ufdc.ufl.edu/UFE0021186/00001

Material Information

Title: Chemical Vapor Deposition of WNxCy Thin Films for Diffusion Barrier Application
Physical Description: 1 online resource (299 p.)
Language: english
Creator: Ajmera, Hiral Mahesh
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: barrier, carbide, copper, cvd, diffusion, films, metallization, mocvd, nitride, thin, tungsten
Chemical Engineering -- Dissertations, Academic -- UF
Genre: Chemical Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The speed and reliability of integrated circuits has been improved significantly by replacing the Al/SiO2 based metallization scheme with a Cu/low-k based one. This study focused on chemical vapor deposition (CVD) of ultra-thin WNxCy (tungsten carbonitride) diffusion barrier films using novel tungsten metal organic precursors. The precursors used for deposition of WNxCy thin film were Cl4(RCN)W(NC3H5) (3a, R = CH3; 3b, R = Ph), W(NiPr)Cl3iPrNC(NMe2)NiPr (4) and (CH3CN)Cl4W(NNMe2). The thin films deposited in a aerosol assisted, vertical CVD reactor were evaluated for their material properties and diffusion barrier efficacy. Successful thin film deposition of WNxCy by MOCVD utilizing 3a,b, H2 and NH3 was achieved in the temperature range of 450 to 750 degree C. The lowest temperature for film growth was 450 degree C. The WNxCy films consisted of W, N, C and O as determined by AES. The Cl content of the film was below the detection limit of XPS (ca. 1 at. %). As compared to films deposited with 3a,b in H2, those deposited with 3a,b, H2 and NH3 had a higher N content. The higher nitrogen content in films deposited with NH3 resulted in higher film resistivity with the lowest measured resistivity of 1690 micro ohm-cm for the film grown at 550 degree C. The apparent activation energy for the film growth from 3a,b, H2 and NH3 in the kinetically controlled growth regime was 0.34 eV, which is significantly higher than the activation energy of 0.15 eV observed for film growth from 3a,b in H2. Successful WNxCy thin film deposition could be achieved for 4 with coreactant NH3 + H2, only NH3, and only H2. The lowest temperature for which film growth occurred was 300 degree C with either NH3 + H2 or only NH3 coreactant(s). Comparison of films deposited with coreactant(s) NH3 + H2, only NH3, and only H2 shows that the best films, in terms composition, resistivity, and microstructure, are deposited using only NH3 as a coreactant. The films deposited with NH3 at 400 degree C had a high N content of 28 at. % and were X-ray amorphous. The films also showed very good conformality in a 0.2 micrometer wide, 2:1 aspect ratio feature. Diffusion barrier testing showed that films deposited at 400 degree C with only NH3 coreactant were able to prevent Cu diffusion after annealing at 500 degree C for 30 min in vacuum. Chemical vapor deposition of WNxCy was achieved using 5 with H2 as a coreactant. It was determined that H2 coreactant is essential for deposition of WNxCy films from 5. The lowest growth temperature for 5 was 300 degree C. Films consisted of W, N, C, and O as determined by AES and no Cl impurity was detected by XPS. The film N content was significantly higher for films deposited from 5 as compared to those deposited from 3a,b or 4. An Amorphous film microstructure was observed for deposition below 550 degree C. The apparent activation energy for the film growth in the kinetically controlled growth regime was 0.52 eV. The films showed a high resistivity compared to the bulk resistivity of pure W2N and W2C. Diffusion barrier testing showed that films deposited at 350 and 400 degree C were able to prevent bulk Cu diffusion after annealing at 500 degree C in vacuum for 30 min. CVD thin film deposition from 3a,b, 4, and 5 highlights the importance of precursor selection and deposition parameters (e.g., coreactant selection, deposition temperature) on the film properties and diffusion barrier performance. Detailed film characterization and preliminary diffusion barrier testing revealed that films deposited with 3a,b and NH3 exhibited the most promise for diffusion barrier applications. To aid the precursor screening process and help understand the mechanism of precursor fragmentation prior to the growth studies, quantum mechanical (QM) calculations using density functional theory were carried out. Statistical mechanics along with QM calculations were employed to determine the energy barrier of potential reaction pathways which would lead to the deposition of WNxCy thin film. QM calculations for fragmentation of precursor 5 showed that the first step of precursor fragmentation was dissociation of the CH3CN ligand, followed by removal of the Cl ligands by either sigma-bond metathesis or reductive elimination.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Hiral Mahesh Ajmera.
Thesis: Thesis (Ph.D.)--University of Florida, 2007.
Local: Adviser: Anderson, Timothy J.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021186:00001

Permanent Link: http://ufdc.ufl.edu/UFE0021186/00001

Material Information

Title: Chemical Vapor Deposition of WNxCy Thin Films for Diffusion Barrier Application
Physical Description: 1 online resource (299 p.)
Language: english
Creator: Ajmera, Hiral Mahesh
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: barrier, carbide, copper, cvd, diffusion, films, metallization, mocvd, nitride, thin, tungsten
Chemical Engineering -- Dissertations, Academic -- UF
Genre: Chemical Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The speed and reliability of integrated circuits has been improved significantly by replacing the Al/SiO2 based metallization scheme with a Cu/low-k based one. This study focused on chemical vapor deposition (CVD) of ultra-thin WNxCy (tungsten carbonitride) diffusion barrier films using novel tungsten metal organic precursors. The precursors used for deposition of WNxCy thin film were Cl4(RCN)W(NC3H5) (3a, R = CH3; 3b, R = Ph), W(NiPr)Cl3iPrNC(NMe2)NiPr (4) and (CH3CN)Cl4W(NNMe2). The thin films deposited in a aerosol assisted, vertical CVD reactor were evaluated for their material properties and diffusion barrier efficacy. Successful thin film deposition of WNxCy by MOCVD utilizing 3a,b, H2 and NH3 was achieved in the temperature range of 450 to 750 degree C. The lowest temperature for film growth was 450 degree C. The WNxCy films consisted of W, N, C and O as determined by AES. The Cl content of the film was below the detection limit of XPS (ca. 1 at. %). As compared to films deposited with 3a,b in H2, those deposited with 3a,b, H2 and NH3 had a higher N content. The higher nitrogen content in films deposited with NH3 resulted in higher film resistivity with the lowest measured resistivity of 1690 micro ohm-cm for the film grown at 550 degree C. The apparent activation energy for the film growth from 3a,b, H2 and NH3 in the kinetically controlled growth regime was 0.34 eV, which is significantly higher than the activation energy of 0.15 eV observed for film growth from 3a,b in H2. Successful WNxCy thin film deposition could be achieved for 4 with coreactant NH3 + H2, only NH3, and only H2. The lowest temperature for which film growth occurred was 300 degree C with either NH3 + H2 or only NH3 coreactant(s). Comparison of films deposited with coreactant(s) NH3 + H2, only NH3, and only H2 shows that the best films, in terms composition, resistivity, and microstructure, are deposited using only NH3 as a coreactant. The films deposited with NH3 at 400 degree C had a high N content of 28 at. % and were X-ray amorphous. The films also showed very good conformality in a 0.2 micrometer wide, 2:1 aspect ratio feature. Diffusion barrier testing showed that films deposited at 400 degree C with only NH3 coreactant were able to prevent Cu diffusion after annealing at 500 degree C for 30 min in vacuum. Chemical vapor deposition of WNxCy was achieved using 5 with H2 as a coreactant. It was determined that H2 coreactant is essential for deposition of WNxCy films from 5. The lowest growth temperature for 5 was 300 degree C. Films consisted of W, N, C, and O as determined by AES and no Cl impurity was detected by XPS. The film N content was significantly higher for films deposited from 5 as compared to those deposited from 3a,b or 4. An Amorphous film microstructure was observed for deposition below 550 degree C. The apparent activation energy for the film growth in the kinetically controlled growth regime was 0.52 eV. The films showed a high resistivity compared to the bulk resistivity of pure W2N and W2C. Diffusion barrier testing showed that films deposited at 350 and 400 degree C were able to prevent bulk Cu diffusion after annealing at 500 degree C in vacuum for 30 min. CVD thin film deposition from 3a,b, 4, and 5 highlights the importance of precursor selection and deposition parameters (e.g., coreactant selection, deposition temperature) on the film properties and diffusion barrier performance. Detailed film characterization and preliminary diffusion barrier testing revealed that films deposited with 3a,b and NH3 exhibited the most promise for diffusion barrier applications. To aid the precursor screening process and help understand the mechanism of precursor fragmentation prior to the growth studies, quantum mechanical (QM) calculations using density functional theory were carried out. Statistical mechanics along with QM calculations were employed to determine the energy barrier of potential reaction pathways which would lead to the deposition of WNxCy thin film. QM calculations for fragmentation of precursor 5 showed that the first step of precursor fragmentation was dissociation of the CH3CN ligand, followed by removal of the Cl ligands by either sigma-bond metathesis or reductive elimination.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Hiral Mahesh Ajmera.
Thesis: Thesis (Ph.D.)--University of Florida, 2007.
Local: Adviser: Anderson, Timothy J.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021186:00001


This item has the following downloads:


Full Text





CHEMICAL VAPOR DEPOSITION OF WNxC, THIN FILMS FOR DIFFUSION BARRIER
APPLICATION





















By

HIRAL M. AJMERA


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2007

































O 2007 Hiral M. Ajmera


































To my great grandmother Jeevatiben Z. Ajmera, grandparents Chunilal and Heeraben Ajmera,
and my parents Mahesh and Neela Ajmera









ACKNOWLEDGMENTS

I thank the chair of my advisory committee, Dr. Timothy J. Anderson, for his continued

support and guidance throughout the course of my study at the University of Florida. My

advisory committee members, Dr. Lisa McElwee-White, Dr. David Norton and Dr. Fan Ren,

also provided excellent guidance in my work. The work on the design, assembly and testing of

the CVD copper reactor would not have been possible without the support of Jim Hinnant

(Chemical Engineering, UF), Dennis Vince (Chemical Engineering, UF), Rob Holobof (A&N

Corporation), and Chuck Rowland (MicroFabritech). The excellent facilities at Major Analytical

Instrumentation Center along with its extremely helpful staff (Wayne Acree, Dr. Valentin

Craciun, Kerry Siebein, and Rosabel Ruiz) were instrumental in obtaining film characterization

results. I thank my co-workers Dr. Omar J. Bchir, Andrew T. Heitsch, Karthik Boinapally, Dr.

Seemant Rawal and Dr. Kee-Chan Kim for their assistance. Last but not the least, I thank my

parents, Neela and Mahesh Ajmera, my brother, Abhijit Ajmera and my wife, Kinj al Ajmera for

their unconditional love and support throughout my education.











TABLE OF CONTENTS


page

ACKNOWLEDGMENT S .............. ...............4.....

LI ST OF T ABLE S ................. ...............9................


LIST OF FIGURES ................. ...............11................


AB S TRAC T ........._._ ............ ..............._ 16...


CHAPTER

1 INTRODUCTION ................. ...............19......... .....


1.1 Background ................ ... ... ...............1
1.2 Device Scaling and Moore' s Law ..........._...__.......... ...............21...
1.3 Interconnects Challenges ................ ...............22................
1.3.1 Delay in Interconnects................ .............2
1.3.2 Electromigration in Interconnects ................. ...............23......._.._....
1.3.3 Power Consumption in Interconnects ................. .... ...............2
1.4 Transition from Al-W-SiO2 to Copper low-k Metallization Scheme............................24
1.4.1 RC Delay in Aluminum Interconnects ................. ............_ ......... .....25
1.4.2 Electromigration in Aluminum Interconnects ................. ................. ......25
1.4.3 Copper Interconnects .............. ...............26....
1.4.4 Low-k Dielectrics ................. ...............27........... ....
1.5 Challenges to Copper Metallization............... .............2
1.5.1 Diffusion of Copper in Silicon ................. ...............28...............
1.5.2 Patterning of Copper .............. ........ ...............30
1.5.3 Dielectric Integration and Reliability ................. ...............32........... ...
1.5.4 El ectromigrati on ................. ...............33......... ....
1.6 Problem Statement ................. ...............35..............
1.6.1 Limitations of PVD ................. ............. ......... ...............35.
1.6.2 Drawbacks of TaN Deposition via CVD or ALD .............. .. .........__ .......36
1.6.3 Desired Properties of Diffusion Barrier Film and its Deposition Process ........37
1.7 Hyp othe si s..........._.... ...............38....._... ....

2 LITERATURE REVIEW .............. ...............46....


2.1 Diffusion in Thin Films............... ......... ..............4
2.2 Refractory Carbides and Nitrides as Diffusion Barriers .............. ....................4
2.3 Deposition Methods for Diffusion Barrier Films ................. ................ ......... .50
2.3.1 Physical Vapor Deposition ................. ...............50................
2.3.2 Chemical Vapor Deposition ................. .......... .... ............... 52....
2.4 Characterization and Testing of Diffusion Barrier Films ................. ............. .......57
2.4.1 Diffusion Barrier Film Characterization .............. ...............57....
2.4.2 Diffusion Barrier Testing ................. ......... ...............61. ....











2.5 Titanium Nitride as Diffusion Barrier............... .......... ...........6
2.5.1 Chemical Vapor Deposition of Titanium Nitride............... ...............64
2.5.2 Atomic Layer Deposition of Titanium Nitride............... ...............71
2.6 Tantalum Nitride as Diffusion Barrier ........._.__........._.._._ ...._.. ...........7
2.6.1 Chemical Vapor Deposition of Tantalum Nitride ........._.__...... ..._.._.........78
2.6.2 Atomic Layer Deposition of Tantalum Nitride ........._.._ .... ....._ ............85
2.7 Tungsten Nitride as Diffusion Barrier............... ............ ..........9
2.7.1 Chemical Vapor Deposition of Tungsten Nitride .............. ....................9
2.7.2 Atomic Layer Deposition of Tungsten Nitride .............. ....................9
2.8 Tungsten Nitride Carbide as Diffusion Barrier............... ...............98
2.8.1 Atomic Layer Deposition of WNxC,.................. ...............98...............
2.8.2 Chemical Vapor Deposition of WNxCy.................. ................ ......... .101

3 EXPERIMENTAL TECHNIQUES FOR THIN FILM DEPOSITION AND
CHARACTERIZATION ................. ...............111................


3.1 Description of the CVD Reactor ................. ...............111..............
3.2 Modification of the CVD Reactor ................. ..... .... .. ..... ............. .... ........11
3.2.1 Installation of H2 and N2 Purifiers in the H2 and N2 Delivery Lines ..............11 1
3.2.2 Reconfiguration of the Gas Lines Downstream of the Reactor ..........._..........1 12
3.2.3 Modification to the Line for Delivery of NH3 to the Reactor .......................1 13
3.2.4 Replace MKS Mass Flow and Pressure Programmer Display.............._._._.....113
3.3 Operating Procedure for CVD Deposition ................. ...............114..............
3.4 Deposition Parameters .................. ... ........ ...............114......
3.5 Dimensionless Numbers for CVD Reactor ................. ...............114..............
3.5.1 Prandtl Number ................. ...............115................
3.5.2 Knudsen Number ................. ...............116...............
3.5.3 Reynolds Number ................. ...............117...............
3.5.4 Peclet Number ................. ...............118................
3.5.5 Grashof Numb er ........._._ ...... .... ...............118..
3.5.6 Rayleigh Number ............... .. .......... ...............119......
3.6 Techniques Used for Thin Film Characterization ................. ......... ................11 9
3.6.1 X-Ray Diffraction ................. ...............119...............
3.6.2 Auger Electron Spectroscopy............... .............12
3.6.3 X-Ray Photoelectron Spectroscopy .............. ...............123....
3.6.4 Scanning Electron Microscopy .............. ...............124....
3.6.5 Four Point Probe .............. ...............125....
3.7 Diffusion Barrier Testing ................. ...............125......... .....


4 WNxC, THIN FILM DEPOSITION from Cl4(RCN)W(NC3HS), NH3 AND H2 ................135

4.1 Precursor Synthesis ................. ...............135...............
4.2 Film Composition .............. ...............136...
4.3 X-Ray Diffraction Measurement .............. ...............138....
4.3.1 Film Crystallinity .............. ...............138....
4.3.2 Lattice Parameter .............. ...............140....
4.3.3 Polycrystal Grain Size. ................ ...._.._ ...............141 ....












4.4 Film Growth Rate ................. ...............142...............
4.5 Film Resistivity ................. ...... .. ........ ... .. ......... ............14
4.6 Comparison of Films Deposited from 3a,b and 2a,b ...._.._ .............. .... ...........143
4.7 Conclusions ..........._...__........ ...............144.....


5 DEPOSITION OF WNxCy USING W(N'Pr)Cl3 zPr(NMe2)N'Pr] AND H2 .........................156


5. 1 Precursor Synthesis ...._.._ ................ ...............156 .....
5.2 Film Growth ................. ...............157................
5.3 Film Composition .............. ...............157....
5.4 X-Ray Diffraction of Films ................. ...............159._____.....
5.5 Lattice Parameter .............. ...............162....
5.6 Grain Size............... ...............163.
5.7 Film Growth Rate ........._..... ......___ .. ...............164..
5.9 Atomic Bonding from XPS Measurement ................. ...............165..............
5.8 Film Resistivity ................. ...............167...............
5.9 Diffusion Barrier Testing ................. ...............168......... .....
5-10 Conclusions ................. ...............170................


6 DEPOSITION OF WNxC, FROM W(N'Pr)Cl3 ['PrNC(NMe2)N'Pr]: EFFECT OF NH3 CO-
REACTANT ON FILM PROPERTIES .............. ...............194....


6.1 Film Growth Studies .............. ...............194....
6.2 Film Composition .............. ...............194....
6.3 Film Crystallinity .............. ...............196....
6.4 Lattice Parameter .............. ...............197....
6.5 Grain Size ................. ...............198................
6.6 Film Growth Rate ................. ...............199...............
6.7 Atomic Bonding ................. ...............200................
6.8 Film Resistivity ................. ...............203...............
6.9 Diffusion Barrier Testing ................. ...............204......... .....
6.9.1 Barrier Testing for 1A............... ...............204..
6.9.2 Barrier Testing for IB .............. ...............206....
6-10 Film Conformality................ ............20
6-11 Conclusions ................. ...............208................


7 DEPOSITION OF WNxCy FROM (Ch3CN)Cl4W(NNMe2) FOR DIFFUSION BARRIER
APPLICATION ................. ...............23. 1..............


7. 1 Precursor Synthesis ................. ...............23. 1......... ....
7.2 Film Growth by CVD ................. ...............23. 1......... ...
7.3 Film Composition .............. ...............232....
7.4 Film Crystallinity .............. ...............233....
7.5 Lattice Parameter .............. ...............234....
7.6 Grain Size ................. ...............235................
7.7 Film Growth Rate............... ...............235.
7.8 Atomic Bonding from XPS............... ...............236..











7.9 Film Resistivity ........._.... ....._.. ...............238.....
7.10 Diffusion Barrier Testing .........._...._ ...............239..._.._ .....
7-11 Conclusions ................. ...............240......_ ......


8 QUANTUM MECHANICAL CALCULATIONS FOR PREDICTING PRECURSOR
FRAGMENT AT ION .........._.._ .......... ...............257....


8.1 Computational Methodology .............. ...............257....
8.2 Geometry Optimization............... .............25
8.3 Acetonitrile Cleavage............... ...............25
8.4 Chlorine Cleavage ..........._..__...... ..._ ...............259....
8.4.1 Reaction of Sa w ith H 2............. ... ...... ..... ..........5
8.4.2 Reaction Pathways for Removal of a Chloride from MI-1 ...........................260
8.4.3 Reaction Pathways for Removal of Chloride from MI-2 and RI-2 ................ .260
8.4.4 Reaction Pathways for Removal of the Fourth Chloride from Sa .................261
8.5 Conclusion .............. ...............262....


9 CONCLUSION AND FUTURE WORK ............__......___....._ ...........27


9.1 Diffusion Barrier Testing ............_...... ._ ...............273..
9.1.1 Film Thickness Control ............ .....___ ...............273..
9. 1.2 Contamination Control ................. ...............274........... ...
9.1.3 Realistic Diffusion Barrier Testing ................. ...............275........... ...
9.2 Diffusion Barrier Integration .............. ...............275....

APPENDIX: DENSITY FUNCTIONAL THEORY ................. ...............277...............


A. 1 Exchange-Correlation Functional (Exc) ................ ...............279........... ...
A. 1.1 Exchange Energy Density .............. ...............279............ ...
A. 1.2 Correlation Energy Density .............. ...............280....
A.2 Density Gradient Correction ............... .... ...............281.
A.2.1 Correction to Exchange Functional: .......... ......... ......... ....281
A.2.2 Correction to Correlation Functional ........................ ...............28
A.3 Similarity Between Hartree-Fock (HF) and DFT Methodologies .............. ................282
A.4 Comparison of DFT to Molecular Orbital Theory ................. .......... ...............282
A.4.1 Computational Efficiency .............. ...............282....
A.4.2 Energetics ................. ...............283................
A.4.3 Geometries .............. ...............283....
A.5 The Future of DFT ................. ...............284.......... ...


LIST OF REFERENCE S ................. ...............286...............


BIOGRAPHICAL SKETCH .............. ...............299....










LIST OF TABLES


Table page

1-1 Bulk resistivity at room temperature of the five most conductive elemental metals.........39

1-2 Dielectric constant of materials with low-k value ........._._.._......_.. ....._.._.......3

2-2 Techniques used for characterization of diffusion barrier thin films............... ............... 103

2-3 Precursors used for deposition of TiNx thin films by CVD ................ ............. .......104

2-4 Precursors used for deposition of TiNx thin films by ALD .............. .....................104

2-5 Precursors used for deposition of TaNx thin films by CVD .............. .....................105

2-6 Precursors used for deposition of TaNx thin films by ALD ................. ............. .......106

2-7 Precursors used for deposition of WNx thin films by CVD ................. ............. .......107

2-8 Precursors used for deposition of WNx thin film by ALD ................. ............ .........107

3-1 Process parameters for CVD of thin films from precursors 3a,b, 4 and 5 .........._..........127

3-2 Knudsen number for precursors 3a,b, 4 and 5 with H2 or N2 carrier gas for different
deposition temperature ...........__............ ...............127....

3-3 Reynolds number for H2 and N2 carrier gas for different deposition temperature ..........128

3-4 Grashof number for H2 and N2 carrier gas for different deposition temperature .............128

3-5 Reference XRD peaks and their relative intensity for Si, P-W2N, P-W2C, a-W2C and
WO3 from JCPDS for Cu Ku radiation ................. ...............129........... ..

3-6 Elemental relative sensitivity factors for AES quantification along with position of
peaks in dN(E)/d(E) spectra used for quantification............... ............12

4-1 Bond distances (A+) and angles (degrees) for Cl4(CH3CN)W(NC3H5) (3a).....................146

5-1 Selected bond distances (A+) and bond angles (0) for 4............... ...............172...

5-2 Analysis results from the deconvolution of the XPS peak of W for films deposited
from 4 on Si(100) substrate at 400, 500, 600 and 700 oC............... ..................17

5-3 Analysis results from the deconvolution of XPS peaks of C, N and O for films
deposited from 4 on Si(100) substrate at 400, 500, 600 and 700 oC .............. ................174

6-1 Molar flow rates used for deposition of WNxCv thin film from precursor 4. ................. .209










6-2 Analysis results from deconvolution of XPS peaks of W for films deposited by
procedure 1A on Si(100) substrate between 300 and 700 oC. ............. ....................21

6-3 Analysis results from deconvolution of XPS peaks of C, N and O for films deposited
by procedure 1A on Si(100) substrate between 300 and 700 oC ................. ..............2 11

6-4 Analysis results from deconvolution of XPS peaks of W for films deposited by
procedure 1B on Si(100) substrate between 300 and 700 oC. ............. ....................21

6-5 Analysis results from deconvolution of XPS peaks of C, N and O for films deposited
by procedure 1B on Si(100) substrate between 300 and 700 oC .............. ...................213

6-6 Apparent sputter rate of barrier film measured by dividing barrier film thickness by
the time required to sputter the film during AES depth profiling .............. ...............2 14

7-1 Molar flow rates of reactants in CVD reactor for deposition from precursor 5 .............242

7-2 Analysis results from deconvolution of XPS peak of W for films deposited from 5 on
Si(100) substrate between 300 and 700 oC. ................ ...............243.............

7-3 Analysis results from deconvolution of XPS peaks of C, N and O for films deposited
from 5 on Si(100) substrate between 300 and 700 oC. ............. ......................4

8-1 Selected bond lengths (A+) and bond angles (0) for precursor 5 obtained from X-ray
crystallographic measurement and DFT calculation............... ..............26

8-2 Selected bond lengths (A+) and bond angles (0) for 5a obtained from DFT calculation ..263










LIST OF FIGURES


Figure page

1-1 Cross-sectional view of metal-oxide-semiconductor field effect transistor (MOSFET) ...40

1-2 A cross-sectional view of metallization scheme ................. ...............40........... ..

1-3 Scaling of transistors in the last 3 decades for Intel Corporation microprocessors ...........41

1-4 The device delay as a function of different device generations............._ .........._.._. ...42

1-5 Effect of feature size scaling on total capacitance.. ............._ ............_ ......_ .....42

1-6 Damaged Al interconnect line showing the formation of hillock and void due to
el ectromi grati on ................. ...............43..._.... .....

1-7 Electromigration lifetime for aluminum interconnects at different current densities........43

1-8 Band diagram of copper impurity in silicon. A) p-type silicon. B) n-type silicon .......44

1-9 Simplified steps in damascene processing for copper deposition. ............. ...................44

1-10 Schematic of one ALD cycle forming a monolayer of film ................. ......................45

2-1 Arrhenius plot for CVD showing mass transfer limited and reaction limited growth
regimes. G is the film growth rate and T is the deposition temperature in K.................108

2-2 Characteristic ALD growth rate vs. T curve showing different growth regimes.............109

2-3 Characteristic ALD plot showing the exposure time required to achieve self-limiting
growth ................ ...............110................

3-1 Schematic of CVD reactor used for thin film deposition .............. .....................130

3-2 Line diagram of the carrier gas feed line. A) Before installation of gas purifier. B)
after installation of gas purifier ................. ...............131........... ...

3-3 Flow diagram of reactor downstream. A) Before reconfiguration. B) After
reconfiguration ................. ...............132................

3-4 Design of new cold trap made of stainless steel ................. ............. ........ .......133

3-5 Schematic of the old and new NH3 feed port. .....__.....___ ...........___.......13

4-1 Thermal ellipsoids diagram of the molecular structure of Cl4(CH3CN)W(NC3HS)
(3a). Thermal ellipsoids are plotted at 50% probability .............. ....................14










4-2 Composition of films deposited from 3a,b (with and without ammonia) and 2a,b
with NH3 at different deposition temperature as determined by AES on Si (100)
substrate. ............. ...............147....

4-3 X-ray photoelectron spectroscopy patterns of films deposited at 450, 600 and 750 oC
from 3a,b and ammonia on Si (100) substrate ................. ...............148........... .

4-4 X-ray diffraction spectra of films deposited on Si (100) substrate from 3a,b and
ammonia............... ...............149

4-5 Grazing incidence XRD spectra of films deposited on Si (100) substrate from 3a,b
and ammonia ............. ...............150....

4-6 Thickness of film (obtained from X-SEM) at different deposition temperature for
films deposited with 3a,b and ammonia on Si(100) substrate ................. ................ ...151

4-7 Lattice parameters for films grown from 3a,b on Si (100) substrate with and without
am m onia. ................. ...............152..............

4-8 Average grain size at different deposition temperature for films grown from 3a,b
with and without ammonia on Si (100) substrate. ............. ...............153....

4-9 Scanning electron microscope images for films grown from 3a,b with ammonia. A)
Film deposited at 450 oC. B) Film deposited at 650 oC .............. ....................15

4-10 Arrhenius plot of log of film growth rate vs. inverse temperature for deposition from
3a,b (with and without ammoniam) and 2a,b with ammonia............... ................15

4-11 Change in film resistivity with deposition temperature for films grown on Si (100)
from 3a,b (with and without ammonia) and 2a,b with ammonia............... .................5

5-1 Thermal ellipsoids diagram of the molecular structure of
W(N'Pr)Cl3 ['Pr(NMe2)N'Pr]. .......... ...............175......

5-2 Composition of films deposited from 4 and 2a,b on Si (100) substrate at different
deposition temperature as determined by AES after 0.5 min of sputtering ...................176

5-3 N/W and C/W atomic ratios for films deposited from 4 and 2a,b on Si (100)
substrate at different deposition temperature as determined by AES after 0.5 min of
sputtering............... ..............17

5-4 X-ray photoelectron spectroscopy patterns for films deposited from 4 at 400, 500,
600 and 700 oC on Si(100) surface ................. ...............178.............

5-5 X-ray diffraction patterns for films deposited from 4 on Si (100) substrate ................... 179

5-6 Deconvolution of XRD peaks for films deposited from 4 at 700 OC on a Si (100)
substrate. .......... ...............180......










5-7 Deconvolution of XRD peaks for films deposited from 4 at 750 oC on a Si (100)
sub state. .......... ...............18 1.....

5-8 Grazing incidence X-ray diffraction spectra for films deposited from 4 on Si (100)
substrate .............. ...............182....

5-9 Lattice parameters for films grown from 4 and 2a,b on Si (100) sub state. ................... .1 83

5-10 Average grain size at different deposition temperature for films grown from 4 and
2a,b on Si (100) substrate ................ ...............184..............

5-11 Scanning electron microscope images for films grown from 4 on Si (100) substrate
between 400 and 750 oC in 50 oC increments .............. ...............185....

5-12 Arrhenius plot of log of film growth rate vs. inverse temperature for deposition from
4 and 2a,b on a Si(100) sub state ................. ...............186............

5-13 Deconvolution of the XPS peaks of W, N, C and O for film deposited from 4 at 400,
500, 600 and 700 oC on Si (100) substrate .............. ...............187....

5-14 Change in film resistivity with deposition temperature for films grown on Si (100)
from 4 and 2a,b .........._ __...... ._ ...............188...

5-15 Pre-anneal AES depth profile of Cu (100 nm)/ WNxCy (45 nm)/Si (substrate) stack
for WNxC, film deposited at 450 oC (using precursor 4). ............. ....................18

5-16 Auger electron spectroscopy profiles of Cu/WNxC,/Si stack for WNxCy deposited at
450 oC using 4 and annealed at A) 200 OC B) 400 OC C) 500 OC D) 600 OC. ..............190

5-17 X-ray diffraction plots for Cu/WNxC,/Si stack (WNxCy films deposited from 4 at 450
oC) before annealing and after annealing at 200, 400, 500 and 600 oC...........................191

5-18 Auger electron spectroscopy depth profiles of Cu (100 nm)/WNxCy (55 nm)/Si (100)
stack for WNxC, film deposited at 500 OC (using precursor 4) and annealed in
vacuum for 30 min at A) 200 OC B) 400 OC C) 500 OC D) 600 OC............... ................192

5-19 X-ray diffraction plots for Cu/WNxC,/Si stack (WNxC, films deposited with 4 at 500
oC) before annealing and after annealing at 200, 400, 500 and 600 oC...........................193

6-1 Composition of films deposited by procedures 1A, 1B and IC on Si (100) substrate
at different deposition temperature as determined by AES after 0.5 min of sputtering ..215

6-2 X-ray diffraction patterns for films deposited on Si (100) substrate by procedure 1A ...216

6-3 X-ray diffraction patterns for films deposited on Si (100) substrate by procedure 1B....217

6-4 Lattice parameters for films grown by procedures 1A and IC on Si (100) substrate. ...218










6-5 Average grain size at different deposition temperature for films grown by procedures
1A and IB on Si (100) substrate. .............. ...............219....

6-6 Scanning electron microscope images for films grown by procedure 1A at different
deposition temperature on Si (100) sub state ................. ...............220.......... .

6-7 Scanning electron microscope images for films grown by procedure 1B at different
deposition temperature on Si (100) sub state ................. ...............221.......... .

6-8 Arrhenius plot for growth using procedures 1A, 1B and IC for growth on a Si(100)
substrate .............. ...............222....

6-9 X-ray photoelectron spectroscopy measurements for films deposited by procedure
1A between 300 and 700 oC on Si (100) substrate ................ ................ ......... .223

6-10 X-ray photoelectron spectroscopy measurements for films deposited for between 300
and 700 oC by procedure 1B on Si (100) substrate............... ...............22

6-11 Change in film resistivity with deposition temperature for films deposited by
procedures 1A, 1B and IC on Si (100) substrate ................. ...............225...........

6-12 Pre- and post-anneal AES depth profiles of Cu (100 nm)/ WNxCy /Si (substrate)
stacks for WNxCy films deposited by procedure 1A. ............ ...............226.....

6-13 Pre- and post-anneal XRD measurements of Cu (100 nm)/ WNxCy /Si (substrate)
stacks for WNxCy film deposited by procedure 1A at 300, 350 and 400 oC ...................227

6-14 Pre- and post-anneal AES depth profiles of Cu (100 nm)/ WNxCy /Si (substrate)
stacks for WNxCy film deposited by procedure IB. .......... .......... ............28

6-15 Pre- and post-anneal XRD measurements of Cu (100 nm)/ WNxCy /Si (substrate)
stack for WNxCy film deposited by procedure 1B at 300, 350 and 400 oC .....................229

6-16 Scanning electron microscope images of WNxCy thin films deposited on FSG
dielectric. The feature size is 0.2 Clm with aspect ratio of 2:1 .............. ...................230

7-1 Composition of films deposited from 5 on Si (100) substrate at different deposition
temperature as determined by AES after 0.5 min of sputtering ............. .................245

7-2 X-ray photoelectron spectroscopy measurements for films deposited from 5 on Si
(100) substrate at different deposition temperature ................. ................. ..........246

7-3 X-ray diffraction patterns for films deposited from 5 between 300 and 700 oC on Si
(100) substrate............... ...............24

7-4 Lattice parameters for films grown from 5 on Si (100) substrate. Error bars indicate
uncertainty in determination of peak position for lattice parameter calculation. ............248










7-5 Average grain size for films grown from 5 at different deposition temperature on Si
(100) substrate. ........... ..... .._ ...............249..

7-6 Scanning electron microscope images for films deposited from 5 between 300 and
700 oC on Si (100) substrate ..........._ ..... ..__ ...............250.

7-7 Arrhenius plot for deposition from 5 on a Si(100) substrate ........._.._.. .............. ....252

7-8 X-ray photoelectron spectroscopy measurements for films deposited from 5 between
300 and 700 oC on Si (100) substrate .............. ...............253....

7-9 Change in film resistivity with deposition temperature for films deposited from 5 on
Si (100) substrate .............. ...............254....

7-10 Pre- and post-anneal AES depth profile of Cu/WNxC, /Si (substrate) stack for
WNxCv film deposited from 5. ............ ...............255.....

7-11 Pre- and post-anneal XRD measurement of Cu (100 nm)/ WNxC, /Si (substrate)
stack for WNxCv film deposited from 5 at 350 and 400 oC ................ ............. .......256

8-1 Geometry of 5. A) X-ray crystallographic determination. B) Geometry
optimization by DFT calculation .............. ...............264....

8-2 Cleavage of acetonitrile ligand from precursor 5. The figure shows optimized
geometries for 5a and CH3CN ........... ......._._ ...............264..

8-3 Calculated transition state and products for reductive elimination and o-bond
metathesis from Sa. ............ ...............265.....

8-4 Possible reaction pathways for chloride cleavage by reductive elimination and o-
bond metathesis from MI-1 intermediate. ............ ...............266.....

8-5 Possible reaction pathways for reductive elimination and o-bond metathesis from
M I-2 (cis and trans).rt~t~rtrt~t~rtrt~ ........... ...............267.....

8-6 Possible reaction pathway for reductive elimination from RI-2. R = reductive
elimination, T = transition state and I = reaction intermediate ................. ........_........268

8-7 Possible reaction pathways for reductive elimination and/or o-bond metathesis from
M I-3 and RI-3. ........... ...............269.....

8-8 The AH values with respect to the A~H (5a) for reaction pathway through intermediate
M I2-cis. ............ ...............270.....

8-9 The AH values with respect to the A~H (5a) for reaction pathway through intermediate
M I2-trans ............ ...............271.....

8-10 The AH values with respect to the A~H (5a) for reaction pathway through intermediate
R I2. ............ ...............272.....









Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

CHEMICAL VAPOR DEPOSITION OF WNxC, THIN FILMS FOR DIFFUSION BARRIER
APPLICATION

By

Hiral M. Ajmera

August 2007

Chair: Timothy J. Anderson
Major: Chemical Engineering

The speed and reliability of integrated circuits has been improved significantly by

replacing the Al/SiO2 based metallization scheme with a Cu/low-k based one. This study

focused on chemical vapor deposition (CVD) of ultra-thin WNxC, diffusion barrier films using

novel tungsten metal organic precursors. The precursors used for deposition of WNxC, thin film

were Cl4(RCN)W(NC3H5) (3a, R = CH3; 3b, R = Ph), W(N'Pr)Cl3 ['PrNC(NMe2)N'Pr] (4) and

(CH3CN)Cl4W(NNMe2) (5). The thin films deposited in a aerosol assisted, vertical CVD reactor

were evaluated for their material properties and diffusion barrier efficacy.

Successful thin film deposition of WNxC, by MOCVD utilizing 3a,b, H2 and NH3 WaS

achieved in the temperature range of 450 to 750 oC. The lowest temperature for film growth was

450 oC. The WNxC, films consisted ofW, N, C and O as determined by AES. The Cl content of

the film was below the detection limit of XPS (ca. I at. %). As compared to films deposited with

3a,b in H2, those deposited with 3a,b, H2 and NH3 had a higher N content. The higher nitrogen

content in films deposited with NH3 resulted in higher film resistivity with the lowest measured

resistivity of 1690 C1D-cm for the film grown at 550 OC. The apparent activation energy for the

film growth from 3a,b, H2 and NH3 in the kinetically controlled growth regime was 0.34 eV,









which is significantly higher than the activation energy of 0. 15 eV observed for film growth from

3a,b in H2

Successful WNxCv thin film deposition could be achieved for 4 with coreactant(s) NH3

H2, only NH3, and only H2. The lowest temperature for which fi1m growth occurred was 300 oC

with either NH3 + H12 Of Only NH3 COreactant(s). Comparison of films deposited with

coreactant(s) NH3 + H12, only NH3, and only H12 shows that the best films, in terms composition,

resistivity, and microstructure, are deposited using only NH3 aS a COreactant. The films

deposited with NH3 at 400 OC had a high N content of 28 at. % and were X-ray amorphous. The

films also showed very good conformality in a 0.2 Clm wide, 2:1 aspect ratio feature. Diffusion

barrier testing showed that films deposited at 400 OC with only NH3 COreactant were able to

prevent Cu diffusion after annealing at 500 OC for 30 min in vacuum.

Chemical vapor deposition of WNxCv was achieved using 5 with H12 aS a, COreactant. It was

determined that H12 COreactant is essential for deposition of WNxCv films from 5. The lowest

growth temperature for 5 was 300 oC. Films consisted of W, N, C, and O as determined by AES

and no Cl impurity was detected by XPS. The film N content was significantly higher for films

deposited from 5 as compared to those deposited from 3a,b or 4. An Amorphous film

microstructure was observed for deposition below 550 oC. The apparent activation energy for

the film growth in the kinetically controlled growth regime was 0.52 eV. The films showed a

high resistivity compared to the bulk resistivity of pure W2N and W2C. Diffusion barrier testing

showed that films deposited at 350 and 400 oC were able to prevent bulk Cu diffusion after

annealing at 500 OC in vacuum for 30 min.

CVD thin film deposition from 3a,b, 4, and 5 highlights the importance of precursor

selection and deposition parameters (e.g., coreactant selection, deposition temperature) on the









film properties and diffusion barrier performance. Detailed film characterization and preliminary

diffusion barrier testing revealed that films deposited with 3a,b and NH3 exhibited the most

promise for diffusion barrier applications.

To aid the precursor screening process and help understand the mechanism of precursor

fragmentation prior to the growth studies, quantum mechanical (QM) calculations using density

functional theory were carried out. Statistical mechanics along with QM calculations were

employed to determine the energy barrier of potential reaction pathways which would lead to the

deposition of WNxCv thin film. QM calculations for fragmentation of precursor 5 showed that

the first step of precursor fragmentation was dissociation of the CH3CN ligand, followed by

removal of the Cl ligands by either sigma-bond metathesis or reductive elimination.









CHAPTER 1
INTRODUCTION

1.1 Background

In the last few decades, electronic devices have become an integral part of our day-to-day

life. From personal computers to laptops, cell phones to PDAs, almost all the electronic devices

contain integrated circuits (ICs). The exponential growth in functionality of these ICs is the

result of continued miniaturization of the devices in circuits, leading to an increase in chip

functionality and decrease in cost.

Semiconductor devices can be built on various substrates, which include the elemental

semiconductors silicon and germanium, and compound semiconductors such as GaN and GaAs.

While compound semiconductors are used for specialized applications such as communication

ICs, most of the ICs today use silicon as the substrate. The abundance of silicon ICs is primarily

because SiO2 can be grown easily on silicon substrate. SiO2 is alSo an excellent insulator that

can easily passivate the silicon surface.

The manufacturing of ICs can be divided in two phases: front-end-of-line (FEOL)

processing and back-end-of-line (BEOL) processing. FEOL is the initial processing on

semiconductor substrate to build devices such as transistors and resistors. Major FEOL

processes include epitaxial silicon film deposition, dopant implantation, diffusion, oxidation and

deposition, and associated patterning steps. BEOL processing includes all the steps involved in

connecting the active components of the circuitry via metal wiring. BEOL processing includes

deposition of contacts, insulators, metal lines and vias, dielectric and associated patterning steps.

The deposition of bonding sites for chip to package connection and wafer dicing are included in

BEOL processing.









The basic structures deposited during FEOL processing of ICs are resistors, capacitors,

diodes, transistors, fuses, and conductors.l Transistors are the building block of an IC. They act

as a switch or an amplifier in an IC. There are many different types of transistors, the most

prominent being field effect transistor (FET) and bipolar junction transistor (BJT). Figure 1-1

shows a metal gate metal-oxide-semiconductor field effect transistor (MOSFET). In a MOSFET,

when a sufficient voltage is applied to the gate metal, current flows between the source and drain

through the channel region. By controlling the gate voltage, the transistor can be turned on or

off. MOS devices are voltage amplifiers, whereas BJTs are current amplifiers. MOS transistor

circuits have the advantage of small area requirement and higher yield in processing. Unlike

BJTs, the MOS transistor is normally in "off' mode and hence it is ideal for low power

application. Moreover, lower power requirement results in low heat generation in MOS

transistors. MOS transistor circuits require fewer processing steps resulting in lower cost per

transistor. Because of these advantages, the maj ority of devices used in ICs today employ

MOSFETs. The CMOS (complimentary metal-oxide-semiconductor) FET is formed in both p-

channel and n-channel MOS transistor. CMOS uses lower power than similar circuits and is

most popular in low power application.

Various devices deposited during FEOL processing are connected to each other via a three

dimensional network of metal lines known as interconnects. The application of metals in ICs is

known as metallization. The metal lines are separated from each other by dielectric materials.

Figure 1-2 shows a cross-section of a three-level metallization scheme.

Interconnects can be 'local' interconnects or 'global' interconnects. Local interconnects

primarily refer to interconnects that connect the gate, source, and drain in MOSFET. Typical

local interconnects are doped polysilicon and refractory metal silicide. Local interconnects can









have higher resistivity because the current in them travels for very short distances. But local

interconnects need to be able to withstand higher processing temperature. Global interconnects,

on the other hand, travel long distances and need to have lower resistivity. Typical global

interconnect metals are aluminum and copper.

1.2 Device Scaling and Moore's Law

Moore's Law is an empirical observation that the number of components per chip (or chip

functionality) doubles every 24 months. The law is attributed to Gordon Moore, co-founder of

Intel Corporation. Figure 1-3 shows how the number of transistors per microprocessor has

roughly doubled every 24 months over the last three decades.

The scaling of transistors in the earlier years was achieved primarily through the

improvements in the FEOL processes. Between 1970 and 1990, the speed of IC was primarily

determined by the speed of the transistors. Much of the speed increase achieved in transistors

was by scaling the transistor dimensions such as oxide film thickness and channel length and

width. Scaling of the device resulted in a higher transistors density per chip for a given chip

area, lower the cost per transistor and increasing speed. The disadvantage of scaling is the higher

current density in the transistors. Figure 1-4 shows how overall device performance varies with

the scaling of the device.2 For earlier generation devices with large feature size, the gate delay

dictated the overall device performance. With the shrinking of device dimensions, interconnect

delay becomes much more important. In fact, interconnect delay is the dominant factor in

determining the device speed for sub-100 nm devices.









1.3 Interconnects Challenges


1.3.1 Delay in Interconnects

The time delay in an interconnect circuit depends on two factors: resistance of the

conductor and the capacitance of the dielectric material. For a MOS circuit, the RC delay is

defined in terms of the circuit response.3


Vout- (t=Vut( a) -x (1-1)


The output voltage of the circuit at time t is Vout (t) maximum output voltage is Vout (max), the

resistance of the metal line is R, and the total capacitance of the circuit is C. The resistance of

interconnect line is defined in terms of its length L, width W, thickness tM, and resistivity p.


R= p
Wt, (1-2)

The capacitance between silicon substrate and metal line (metal to ground), CMI-G, is given by

Equation 1-3.

WL
CM-G 0
tD (1-3)

e and tD are the dielectric constant of the dielectric material and thickness of the dielectric line

respectively, and so is the permittivity of free space. The capacitance between two metal lines

(Cn/r-) separated by distance D is given by Equation 1-4.4

t,L
C,, = eco
D (1-4)

The total capacitance, C, can then be calculated from Equations 1-3 and 1-4.

tW t, 1 5
C = K, (CM-G M-M) = KIecoL 15
tD D










KI is a constant that depends on the fringing field effect. The RC delay is calculated from

Equations 1-2 and 1-5.

K,eeWto pL'2 tW t~k 1
RC + (1-6)


As the critical device dimension shrinks, CanI- dominates the overall capacitance C (Figure 1-5).5

The RC delay can then be written as Equation 1-7.

K,esopL2
RC I (1-7)
WD

For local interconnects the shrinking of the transistor leads to a corresponding decrease in

L, W and D. Therefore, the actual delay in the local interconnect does not increase with device

scaling. However, the chip area increases with each successive technology node. This leads to

an increase in the length of global interconnects and a corresponding increase in the RC time

delay.

1.3.2 Electromigration in Interconnects

Electromigration is the phenomenon by which high current density in the interconnect

lines leads to the movement of the metal atoms in the direction of the electron flow. Most

present day microprocessors operate at high temperature (ca. 100 oC). At high temperature, the

metal atoms in the interconnect have higher mobility. When high energy electrons transfer their

momentum to these mobile metal atoms, there is a movement of metal atoms in the direction of

electron flow. This atom movement leads to formation of voids at the cathode and hillocks at the

anode (Figure 1-6)

Electromigration, which causes failure of the circuit, is one of the most important concerns

for interconnect reliability. As the interconnect dimensions shrink with each successive

generation, the current density in the interconnects also increases. The scaling of interconnects










requires higher electromigration resistance in interconnect materials. Electromigration becomes

a maj or concern for aluminum-based metallization scheme. Figure 1-7 shows that the mean time

to failure for aluminum interconnect sharply decreases if the current density in the interconnect

exceeds 3 MA/cm2

1.3.3 Power Consumption in Interconnects

The power consumption in an interconnect operating at frequency f and voltage V is

given by Equation 1-8.

P = CV2f (1-8)

The shrinking of interconnect dimensions causes an increase in parasitic capacitance, thereby

increasing the power required to operate the IC. The power requirement in ICs is increasingly

becoming an important issue because of the popularity of battery operated mobile devices. Since

the improvements in power storage capacity of batteries has not increased as fast as the power

requirement of ICs, there is an increased focus on developing lower power ICs. To reduce the

power requirement of interconnects, it becomes important to decrease the capacitance of the

inter-metal dielectric.

1.4 Transition from Al-W-SiO2 to Copper low-k Metallization Scheme

Until the 1990's, aluminum had been the preferred metal for interconnect lines and Sio2

was the preferred dielectric. Tungsten was deposited in the vias to connect different levels of

aluminum lines. The Al-W-SiO2 based metallization scheme has many advantages.6 Aluminum

can be easily deposited with evaporation or sputtering techniques with film resistivity close to

the bulk resistivity of 2.7 CID-cm. Aluminum is easy to pattern as it can be readily etched by dry

or reactive ion etching. Aluminum also has good adhesion to the SiO2 surface. Addition of 0.5

at. % copper to aluminum increases the electromigration resistance of aluminum interconnects.









1.4.1 RC Delay in Aluminum Interconnects

One of the maj or problems with the Al-W-SiO2 based interconnect scheme is that RC

delay in these interconnects, which is a significant factor in determining the speed of an IC. The

interconnect delay surpassed the transistor delay below 0.25 Clm minimum feature size (Figure 1-

4). To leverage the gains made in transistor speeds and density, a reduction in interconnect delay

is required. Increase in signal noise and higher power requirement for narrower global

interconnect lines necessitated use of new dielectric material with dielectric constant lower than

that of SiO2 dielectricc constant k = 3.9).

1.4.2 Electromigration in Aluminum Interconnects

Another major problem with the aluminum-based interconnect scheme is reliability. In

1960s, it was discovered that electromigration can damage the aluminum lines and compromise

the interconnect reliability.' Electromigration of aluminum atoms at the Al-W interface results

in movement of aluminum atoms away from the interface by grain boundary diffusion. The

vacancies created by aluminum atoms due to electromigration accumulate to form a void. If the

size of the void becomes larger than the diameter of the via, an opening in the circuit is created.

This phenomenon, which leads to the failure of the interconnect, is referred to as "wear-out"

mode of failure. The "wear-out mode" in Al-W-SiO2 metallization scheme results in a decrease

in lifetime of multilevel metallization by a factor of 50 as compared to single level aluminum

interconnect.8 A short-term solution to electromigration problem in Al-W-SiO2 metallization

was to use small amount of solute (1 4 at. % copper) in aluminum lines. Ames et al.9 found that

the addition of 4 at. % copper in aluminum strips increased the electromigration resistance by a

factor of 70 as compared to pure aluminum strips under similar testing conditions of 175 OC

temperature and 4 MA/cm2 current density. Copper in aluminum forms Al2Cu precipitates along

the grain boundaries of aluminum crystallites. While the exact mechanism behind the increase in









electromigration resistance of aluminum by addition of copper is not completely understood, it is

believed that copper may either decrease the concentration of vacancies in grain boundaries or

increase the activation energy of aluminum diffusion through grain boundaries.

The use of copper as solute extended the use of aluminum as the preferred material for

interconnects. However, with further increase in current density in interconnects, even the

aluminum doped with copper could not provide sufficient electromigration resistance. Sakimoto

et al.10 studied the mean time to failure (MTTF) of aluminum (0.5 at. % copper) at 200 OC. The

MTTF showed a dramatic decrease at current densities of 3 MA/cm2 and higher. Higher current

densities in interconnects demand a new interconnect material that has a higher resistance to

electromigration than aluminum doped with copper.

1.4.3 Copper Interconnects

One way to reduce the RC delay in interconnects is to use a material with higher electrical

conductivity as interconnect. Table 1 shows the conductivity of the 5 most conductive metals.

The metal with lowest bulk resistivity is silver. But silver shows low resistance to

electromigration. Gold also has a lower bulk resistivity compared to aluminum, but gold is much

more expensive and the performance gain in terms of resistivity is minimal. Copper with bulk

resistivity close to silver is a much better conductor than aluminum. Moreover, copper has a

melting temperature (1083 OC), which is much higher than that of aluminum (660 oC). Atomic

diffusion is slower in copper owing to its high melting temperature, and hence is expected to

have a better electromigration resistance as compared to aluminum. The semiconductor industry

has gradually moved towards copper metallization for global interconnects because of lower

resistivity and superior electromigration resistance of copper. IBM was the first maj or chip

manufacturer to announce successful implementation of copper interconnects in 1997.









1.4.4 Low-k Dielectrics

Another way of reducing the RC time delay is to use dielectric material with a lower

dielectric constant. The dielectric constant of a material depends on the polarizability of its

constituent atoms or molecules; the higher the polarizability, the higher the dielectric constant.

SiO2 has a dielectric constant of 3.9. For SiO2 based materials, low dielectric constant film can

be obtained by replacing the Si-O bonds in the film with Si-C or Si-F bonds, which have a much

lower polarizability as compared to Si-O bonds. The C and F atoms are incorporated in the film

by introducing an organic precursor or a fluorinated silicon precursor source during the plasma

enhanced chemical vapor deposition (PECVD) of SiO2.

Another important class of materials that has been investigated for dielectric application

includes silsesquioxane (SSQ) group based materials. Since the cubic SSQ cells occupy a much

larger volume than tetrahedral SiO2, the k-value of SSQ is lower than that of SiO2.

Silsesquioxane unit cell contains Si-O bonds in a cubic arrangement. In hydrogen terminated

SSQ (HSSQ), unit cells are connected to each other via oxygen atoms, whereas the cube corners

are terminated by hydrogen. In methyl terminated SSQ (MSSQ), the cubes are connected by -

CH2 group, while cube corners are terminated by -CH3 group.

Non-polar organic compounds have also been investigated for dielectric application.

Aliphatic compounds have the lowest k-value but they are thermally unstable at temperature

higher than 300 oC.11 Aromatic compounds and cross-linked polymers have a higher thermal

stability. Table 1-2 lists the important polymeric materials that have been studied for their

potential application as dielectric material in ICs.

Decreasing the density of polarizable molecules is an effective way of reducing the k-

value, however, it often comes at the expense of mechanical strength. Density can be decreased

by introducing micro/meso pores in the dielectric film. Most materials have some inherent










porosity, though the pore size might be very small. This inherent porosity is referred to as

constitutive porosity. To increase the k-value of dielectric film, intentional doping of thermally

unstable molecules is used. Once these unstable molecules are evaporated or decomposed by

thermal treatment, they leave behind a matrix of void spaces or pores resulting in a decrease in k-

value of the film. These void causing agents are referred to as porogens and are very effective in

lowering the k-value of dielectric film. The porosity that is increased after the growth of the

film, by agents such as porogen or by selective etching of the film, is referred to as subtractive

porosity. For silica based films deposited by PECVD, CHx is used as porogen. Xerogels and

aerogels both have ultra low-k values primarily due to their subtractive porosity.

Presently, many IC manufacturers have successfully integrated PECVD deposited carbon

doped oxide (CDO or SiOC) into their metallization schemes in 90 nm devices with effective k-

value of 2.7 2.9. While acceptance of copper as the metal of choice for interconnects has been

ubiquitous, the semiconductor industry has not yet singled out a single dielectric material for 45

nm devices. Carbon doped oxides have been preferred over other low-k materials for 65 nm

devices, however, industry is still evaluating different materials and processes for a suitable low-

k dielectric (< 2.5) at the 45 nm technology node.

1.5 Challenges to Copper Metallization

The introduction of new materials has led to new challenges in fabrication of ICs. These

challenges are related to the processing of the new material or the integration of new material

with other components of the IC. Copper metallization has introduced many processing and

integration related challenges and this section highlights the most important challenges.

1.5.1 Diffusion of Copper in Silicon

In a transistor, interstitial impurities such as Cu, Ni, Fe, and Li can be incorporated during

various processing steps. The tolerance level of these impurities in the device depends on the









energy level of the impurity in band structure of the device, device dimension, diffusivity, and

interaction of the impurity with silicon and dopants. If an impurity has an energy level lying

deep in the band gap of the semiconductor, such impurities can be detrimental to the device

performance. These 'deep impurities' can act as alternate recombination centers and hence

decrease diffusion length, increase junction leakage, and decrease minority carrier life time.3

Moreover, the tolerable amount of defects in silicon decreases as the device dimension shrinks.

Copper is an extremely fast diffuser in silicon. The diffusion coefficient of copper in

silicon at 500 oC is 2.0 x 105 cm2/SOC.14 Furthermore, copper is mobile in silicon even after

annealing. Copper can travel significant distances within the silicon wafer even at room

temperature. The interaction of copper in silicon depends on the doping of the silicon substrate.

In n-type silicon, copper primarily forms neutral or negatively charged precipitates. Since

interstitial copper is positively charged in silicon", there is either no barrier or electrostatic

attraction between interstitial copper and copper precipitates. Hence, in n-type silicon, copper

readily forms Cu3Si precipitates. These copper precipitates can act as recombination centers for

minority carriers and significantly lower minority carrier lifetime.16 In p-type silicon, copper

precipitates are positively charged species." There is electrostatic repulsion between the

positively charged copper in precipitates and positively charged interstitial copper, which acts as

a barrier in nucleation and precipitation of copper in p-type silicon. The out diffusion of copper

to the silicon surface is also dependent on the type of doping of silicon. If the surface of a p-type

Si wafer is free of native oxide, copper diffuses from the bulk to the surface of the wafer. Since

surface segregated copper can easily be etched by a H202 and HF mixture, it is inferred that the

copper does not form Cu3Si precipitates. For n-type silicon wafer, out diffusion occurs only

when the wafer is heated to 400 OC.









The primary reason why copper is extremely undesirable in silicon is that it forms deep

levels in the band gap of silicon (Figure 1-8). If copper exists in the triplet state, it forms three

acceptor levels in the silicon band gap at 0.24, 0.37 and 0.52 eV.3 Once copper has diffused in

silicon, the high diffusivity of copper in silicon ensures that the detrimental effects of copper

spreads throughout the device. Copper is also detrimental if it diffuses into the dielectric film.

Copper in dielectrics decrease the breakdown voltage of the dielectric and also increases the

parasitic capacitance, thus negating the benefits achieved from low-k dielectric film.

To prevent diffusion of copper into dielectric films, researchers have investigated various

thin films that can act as diffusion barrier between copper and the dielectric. Industry presently

employs a TaN/Ta bilayer as diffusion barrier/adhesion promoter. TaN has good diffusion

barrier properties and has good adhesion to SiO2. Ta, on other hand, shows good adhesion to

copper and TaN films. Ta also promotes the growth of copper (111) crystals during physical

vapor deposition (PVD) copper seed deposition. It is well known that copper (111) has higher

electromigration resistance as compared to other crystal orientations. With the scaling of

interconnects, the barrier film also has to shrink. This is a maj or challenge in copper

metallization because the PVD process currently employed to deposit the barrier film can not be

extended for future generations of interconnects. The challenges in barrier deposition are

discussed in detail in section 1.6.

1.5.2 Patterning of Copper

One of the main challenges in initial adoption of copper interconnects in ICs was its

patterning scheme. Since copper does not form any volatile by-products at low temperature,

reactive ion etching (RIE) can not be employed for patterning copper.'" This problem was

addressed by a radical change in patterning scheme for interconnect layers. For aluminum

metallization, aluminum is deposited first, followed by patterning via lithography and etching










steps. The SiO2 dielectric was then deposited into the aluminum pattern. To implement copper

metallization, the scheme of deposition is reversed. The dielectric film is deposited first, and

then the lines and vias for the metals are patterned in the dielectric via lithography and etching

steps, followed by deposition of copper in the lines and vias. The film stack is then subj ected to

chemical-mechanical polishing step (CMP) to planarize the copper layer and remove the barrier

layer on the dielectric surface. This approach of deposition is often referred to as the 'dual-

damascene' patterning scheme. Figure 1-9 shows the basic steps in the dual-damascene

processing scheme. It should be noted that since the purpose of the figure is to highlight the

main steps in dual damascene processing, deposition of other layers such as diffusion barrier,

dielectric etch-stop layers and capping layers during the dual damascene processing are not

included in the figure.

In the dual damascene scheme, either the trench can be formed first followed by the via, or

the via can be patterned first followed by the trench. In 'trench first approach', patterning of the

via results in 'pooling' of the photoresist in the trenches. The 'pooling' increases the thickness

of dielectric in the trenches, making it difficult to etch the vias through the thicker dielectric

layer. Hence, industry has adopted the 'via first approach' for patterning the dielectric layer for

devices with minimum feature size of 0.25 Clm or lower.

The detailed description of steps involved in copper interconnects are as follows:

* The copper line is coated with the first etch-stop layer (usually SiN or SiCN). Dielectric film
(layer 1) is deposited on the etch-stop layer, followed by deposition of the second etch-stop
layer. Another dielectric film (layer 2) is deposited on the second etch-stop layer followed
by the deposition of the third etch-stop layer.

* Mask is applied to pattern the vias (for 'via first approach'). Anisotropic etch cuts through
the top etch-stop and dielectric layers to form the vias. This process is optimized to stop at
the etch-stop layer at the bottom of via. The photoresist is then removed.

* Mask is applied to pattern the trenches. Anisotropic etch cuts through the third etch-stop and
second dielectric layer to form the trenches. The photoresist deposited at the bottom of the









via prevents over etching of the via during the trench etch process. Low energy etching step
removes the mask from bottom of the via. The mask layer is then removed.

* Diffusion barrier and adhesion promoter thin films are deposited by PVD process (TaN/Ta
bilayer).

* Copper thin film, which acts as 'seed' layer for subsequent electrochemical plating process,
is deposited by PVD. The trenches and vias are filled with bulk copper by electrochemical
plating.

* Excess copper over the barrier film is removed by CMP that stops at the first etch-stop layer.

As the feature size shrinks further, new processes and/or materials are expected to be

introduced in the dual damascene patterning scheme. For instance, the integration of ultra low-k

dielectric films in the interconnects would require additional steps such as pore sealing to close

the meso pores in the dielectric film prior to barrier film deposition.l

1.5.3 Dielectric Integration and Reliability

Presently, industry has successfully integrated CDOs in the interconnect stack. These

dielectric films have k-value in the range of.2.7 to 2.9. But as device dimension shrinks further,

films with lower k-value will be required to decrease the parasitic capacitance and power

requirements. Introduction of highly porous films lowers the dielectric constant of films but it

comes at the cost of mechanical stability.19 Moreover, future interconnects will need processes

such as CVD or ALD to deposit conformal diffusion barrier films. Gaseous reactants used in

CVD or ALD process can easily diffuse through the interconnected pores in the dielectric film,

thereby increasing the parasitic capacitance of the dielectric film. An ideal dielectric film should

have uniformly distributed pores that are not interconnected to one another. These isolated pores

prevent the diffusion of precursor molecules used in CVD and ALD process through the

dielectric film and aid uniform deposition of barrier film. But as the porosity of dielectric film is

increased to lower the k-value of dielectric film, these pores become more interconnected to one

another. These interconnected pores act as pathways for diffusion of precursor molecules









through the dielectric film. Interconnected pores also make it difficult to deposit uniform barrier

film that can prevent copper diffusion into the dielectric. Various processes such as plasma

exposure have been proposed to plug or block the open pores on the surface of the dielectric

film.

Adhesion of the dielectric film to the diffusion barrier is also of concern for new dielectric

films that are being investigated for future interconnects. Failure modes such as time-dependent

dielectric breakdown (TDDB) that were not previously observed for Sio2 based interconnects,

have become reliability issues for low-k dielectrics because of film adhesion problems.20

In summary, lowering of the k-value comes at the expense of mechanical stability, ease of

integration with barrier film deposition process, and adhesion and interconnect reliability.

Extensive research is ongoing to find suitable processes and materials for deposition low-k

dielectric film and its integration in the dual damascene processing scheme.

1.5.4 Electromigration

Numerous studies have shown that copper has better electromigration resistance as

compared to aluminum. However, as the dimensions of the interconnect shrink, electromigration

in copper can result in poor reliability of interconnects. Moreover, the electromigration

mechanism in aluminum interconnect is very different from that in copper. While

electromigration in aluminum interconnect occurs via the 'wear-out' mode of failure videe

supra), electromigration in copper interconnects is primarily a surface migration mode. Surface

migration has lower activation energy as compared to grain boundary diffusion. Copper

electromigration occurs via surface diffusion because copper does not adhere well with dielectric

films such as SiO2.21 The top surface of trench is in direct contact with the dielectric and the

surface diffusion of copper atoms occurs at this interface.









One of the approaches that has found success in limiting electromigration in copper is

alloying of copper metal. The solute atoms in the copper alloy block the kink sites on the surface

of the dielectric forming the copper line, thereby hindering the surface diffusion of copper and

increasing the electromigration resistance. Solutes such as Sn and Zr have shown significant

improvements in electromigration resistance when added to copper.22 The downside of alloying

copper is that the RC time delay increases as the resistance of copper alloy is higher than that of

pure copper metal. Moreover, current copper bulk deposition in vias and trenches is done via an

electrochemical plating process. Addition of solute in copper by the electrochemical plating

process would require significant modifications to the existing process and further research needs

to be done in order to adopt this approach.

Another approach that has found wide acceptance in increasing electromigration resistance

of copper is the use of cap layer to separate the copper-dielectric interface. An ideal cap layer

would have good adhesion to both the copper and dielectric films. Good adhesion between

copper and cap layer can significantly reduce electromigration failure in interconnect lines.

Recent studies have investigated various cap layers such as Ta, Ta/TaN, Pd, SiNx, SiCxN,Hz and

CoWP.23-26 The disadvantage of using a cap layer is that it increases the effective k-value of the

dielectric fi1m as the cap layer reduces the area of dielectric film.

Industry is believed to introduce copper alloy for improving electromigration resistance of

copper because this approach is considered fairly low risk. In the long term, it is believed that

capping layers can alleviate electromigration in copper interconnects for several generations.20

Moreover, capping layers such as CoWP can be selectively deposited on copper line by

electroless plating and hence the deposition of capping layer will not require patterning or

etching steps. The primary reason why implementation of capping layer is pushed to future










generations is because there is skepticism of introducing additional layer to the interconnect

stack. Additional layers in interconnect stack introduce new processes, materials) and

integration related challenges that need to be addressed.

1.6 Problem Statement

1.6.1 Limitations of PVD

Of all the future challenges to the scaling of copper metallization discussed in the previous

section, one of the most important challenges is the scaling of the diffusion barrier film. With

the shrinking dimensions of interconnect lines, the thickness of diffusion barrier needs to be

reduced. In fact, according to the roadmap for future interconnects published by 'International

Technology Roadmap of Semiconductors' (ITRS), a consortium of industry and academic

institutions, the thickness of barrier layer needs to be scaled down from 72 to 37 A+ by year 2007

and to 24 A+ by year 2013.27 The scaling of barrier film is required for two reasons. First, as

copper line and via dimensions shrink, the relative volume that the barrier film occupies in the

line or via increases resulting in decrease in available area for copper. This could increase both

the RC time delay of the interconnect and the current density in the copper lines and vias.

Secondly, the barrier forms a part of electrical circuit where a via contacts the line. The thinner

the barrier, the lower is the resistance to electron transport between via and line. Extensive

research is being done to find a suitable process along with new material for deposition of low

resistivity ultra-thin barrier film with good conformality and adhesion to copper and dielectric

layers.

Physical vapor deposition, which is presently employed to deposit diffusion barrier film, is

believed to be incapable of depositing ultra-thin films with conformal coverage. The PVD

process is a 'line of sight' process and the directional nature of the deposition process results in

excess deposition at the edges and poor sidewall coverage inside the trenches and vias. Ionized









PVD (I-PVD), a variant of PVD, employs a high density plasma to ionize the metal ions during

deposition. These ions are then directed perpendicular to the wafer surface via magnets to obtain

better sidewall coverage.28 While I-PVD has helped in extending the usefulness of PVD for

deposition of diffusion barrier film, future generations of diffusion barrier films would require

deposition techniques that have a significantly better step coverage than I-PVD.

Chemical vapor deposition (CVD) is a technique that employs gaseous reactants to deposit

thin films. Films deposited by CVD have much better step coverage than the best variant PVD

process. Atomic layer deposition (ALD), a variant of CVD process, also has excellent step

coverage and precise control over film thickness. Similar to CVD, ALD employs gaseous

reactants but the delivery of the reactants is done in a sequential manner. One ALD cycle

consists of4 steps (Figure 1-10). In the first step, the substrate is exposed to reactant A, which

adsorbs the first layer on the substrate surface by chemisorption and subsequent layers by

physisorption. In the next step, excess reactant A is purged from the reactor and the reactant

molecules that are physisorbed on the chemisorbed layer are removed from the reactor. In the

third step, a co-reactant B is introduced into the reactor. Co-reactant B reacts with the

chemisorbed A to form a monolayer thick film. In the last step, excess B is purged from the

reactor. If chemisorption occurs in a self-limiting mode, one ALD cycle should ideally be able

to deposit a monolayer film. Atomic layer deposition is believed to be capable of extremely

uniform film deposition in very high aspect ratio features.

1.6.2 Drawbacks of TaN Deposition via CVD or ALD

Another challenge to scaling of the diffusion barrier is the material. While PVD TaN has

been very successful so far as a diffusion barrier, future interconnects require films deposited

with CVD or ALD. Many investigators have attempted the growth of TaN thin films via CVD

and ALD. But the growth of the resistive Ta3N5 phase during attempts to grow TaN by CVD









results in an increase in resistivity of the film.29 Presently, many new materials are being

investigated for diffusion barrier application, the details of which are included in the next

chapter.

1.6.3 Desired Properties of Diffusion Barrier Film and its Deposition Process

The following are the main requirements of a diffusion barrier film:

* Prevent copper diffusion under device operating conditions
* Good adhesion with copper, dielectric, and etch-stop layers
* Low resistivity (< 400 CIR-cm)
* Good mechanical and thermal stability
* CMP removal rate of the barrier film similar to that of copper

Together with the material, there are certain process requirements for deposition of the

diffusion barrier film:

* Capable of ultra-thin film deposition (few nm thick)
* Excellent conformality
* Low deposition temperature (< 500 oC)
* No pin-holes in the barrier film

Film properties that are considered beneficial for diffusion barrier film include:

1 Amorphous microstructure: Diffusion through the barrier film primarily occurs via grain
boundary diffusion. One of the ways of hindering the diffusion of copper through the
barrier film is to deposit a single crystal barrier film. However, thermal requirements and
lattice mismatch make deposition of a single crystal film impractical. The next best option
is to deposit an amorphous barrier film.30

2 Sufficient N content: Transition metal nitrides have been found to have good diffusion
barrier properties. The primary reason, it is believed, is the presence of excess N at the
grain boundaries of the film. Since N does not form a stable compound with copper, the
presence of N hinders the diffusion of copper through the grain boundaries. So, an
effective diffusion barrier should have sufficient N to "stuff" the grain boundaries to
prevent copper diffusion.

3 Low impurities incorporation: Oxygen impurities form oxides that have a high resistivity.
So, incorporation of O has to be minimal in the diffusion barrier film. In addition,
impurities such as Cl and F are also undesirable because they increase the film resistivity
and adversely affect the adhesion of barrier film with copper.









1.7 Hypothesis

Thin film WNxCy deposited by metal-organic chemical vapor deposition (MOCVD) would

be capable of meeting the material and process requirements expected from a good diffusion

barrier film. WNxCy, a ternary transition metal compound, is expected to be able to be grown

with an amorphous microstructure because the presence of a third element in the binary

transition metal nitride matrix disrupts the crystal lattice to favor the growth of amorphous film.

Carbon in the film is expected to lower the film resistivity of WNxCy because WCx is more

conductive than WNx.

An important focus of the present work is deposition of WNxCy thin films using metal

organic precursors. Films deposited by single-source metal organic precursors should help to

lower the deposition temperature. Low deposition temperature from these films should also aid

the deposition of amorphous films. Precursors used for MOCVD should help in preventing

incorporation of halides in the film. Previous studies have shown that WNx films have excellent

diffusion barrier properties.31,32,33 It is anticipated that WNxCy films deposited with the metal

organic precursors used in the present work would also have excellent diffusion barrier

properties. The WNxCy films deposited by MOCVD are expected to have good thermal and

mechanical stability, good adhesion to dielectric and copper films, and a CMP removal rate

similar to that of copper.

Since the deposition technique used in the present work is CVD, it is expected that the

process would yield films with excellent conformality. As discussed earlier, films deposited by

CVD have a much better conformality than those deposited by PVD. Film thickness control of

CVD is expected to be similar to that of PVD. The precursors that are successful in CVD

deposition of WNxCy films can then be used to attempt film growth by ALD to get even better

conformality and precise thickness control.










Table 1-1. Bulk resistivity at room temperature of the five most conductive elemental metals
Metal Bulk Resistivity (CIG-cm)
Silver (Ag) 1.6
Copper (Cu) 1.7
Gold (Au) 2.4
Aluminum (Al) 2.6
Molybdenum (Mo) 5.2






Table 1-2. Dielectric constant of materials with low-k value
Material Dielectric Reference
Constant (k)


Silicon-based
SiF
Si-C
Silsesquioxane (SSQ) based
HSSQ
MSSQ
Polymer based
Polyimides
Parylene-N
Parylene-F
Teflon (PTFE)
Highly porous oxides
Xerogel s/Aerogels


3.2 3.4 11
2.7 2.8 11

2.8 -3.0 12
2.7 2.9 13


3.0 4.0
2.6
2.2
1.9 2.1


< 2.5 12



















Figure 1-1. Cross-sectional view of metal-oxide-semiconductor field effect transistor
(MO SFET)


-Contacts

4( Drain
PI. Channel


Contacts


a I C hannel~ri


Sou rce


Sou rce


Figure 1-2. A cross-sectional view of metallization scheme























































r ooe

I I I II1

1971 1980 1990 2000 200


10,000,000,000




1,000,000,000 -

1)000 -




100,000,000 -




10,000,000 -




1000,000






10,000 -


2,300


Number of t~ranistors doubling every 18 months.







Number of trnanistors doubling every 2~4 months, .


a Itnium 2
(9 MB cochel

-'Itanium 2


Pentium 4
itanium
* Pntium Ill


'Pentium II
Pentium

S 486



,' '286


i,


Figure 1-3. Scaling of transistors in the last 3 decades for Intel Corporation microprocessors

(Wikipedia, http://en.wikipedia. org/wiki/Moores_1aw, May, 2007)


Moore's Law






















..,......


.... ... ....

rrel~7o~R*b O.by

''' '" ''' '
Gne ~~iuR LUn
1... .s .~
r r
.r
-d
~areI~a~
4:t~q ~~ a S~r .~ .... dC~K
~. is-.' R& Ik~s, ;Cu
-"-'"
~.~,--~- ~~.---is--'"~
-------c-~~-~-~--t'~~ I
IS 136 r~a; Irl IrB: I


I


_I


I
i

5"
as
~25;
i
g,

rT
r6
P
I


Figure 1-4. The device delay as a function of different device generations.2 Adopted from M. T.
Bohr, "Interconnect scaling the real limiter to high performance ULSI," Proceedings
of IEEE International Electron Devices Meeting, 241 (1995). 01995 IEEE.

Reproduced with permission.


I.a c


r.J :


C~ "1 "4 r


1L


Figure 1-5. Effect of feature size scaling on total capacitance.5 Adopted from S. P. Jeng, R. H.
Havemann, and M. C. Chang, "Process integration and manufacturability issues for

high performance multilevel interconnect", Mat. Res. Soc. Symp. Proc. 337, 24

(1994). O 1994, Materials Research Society, Pittsburgh, PA. Reproduced with
permission.


























108

107

10"



104

104

103


Figure 1-7. Electromigration lifetime for aluminum interconnects at different current densities.10
Adopted from M. Sakimoto, T. Fujii, H. Yamaguchi, and K. Eguchi, "Temperature
measurement of Al metallization and the study of Black' s model in high current
density", Proceedings of the IEEE Reliability Physics Symposium, Las Vegas, NV,
333 (1995). O 1995 IEEE. Reproduced with permission.


Figure 1-6. Damaged Al interconnect line showing the formation of hillock and void due to
electromigration. Adopted from K. Wetzig and C. M. Schneider, M~etalBased Thin
Films for Electronics, 2nd Edition, 231 (Wiley-VCH, 2006), C Wiley-VCH Verlag
GmbH & Co. KGaA. Reproduced with permission.


1 10
J[MAk~nr]l










n-type silicon
EC --


Cui


Cui


Ev


Figure 1-8. Band diagram of copper impurity in silicon. A) p-type silicon. B) n-type siliconl3
Adopted from A. A. Istratov, C. Flink, and E. R. Weber, "Impact of the unique
physical properties of copper in silicon on characterization of copper diffusion
barriers", Physica Status Solidi B: Basic Research 222, 261 (2000). O Wiley-VCH
Verlag GmbH & Co. KGaA. Reproduced with permission.


JLD


p-type silcon

E C -...........


EC


I~p____~ LClu~J~ .~~~





'----;-----~wF~nr -~- -.;
ILD ~unZ~C`['\ .`t


A, B


Figure 1-9. Simplified steps in damascene processing for copper deposition. A) Single
damascene copper deposition. B) Dual-damascene copper deposition.35 Adopted
from D. T. Price, R. J. Gutmann, and S. P. Murarka, "Damascene copper
interconnects with polymer ILDs", Thin Solid Films 308-309, 525 (1997). O
Elseweir. Reproduced with permission.
























Gas Phase


Gas Phase










1. Adsorption of reactant A on
substrate


Gas Phase










3. Introduction of reactant B.
A + B *+ barrier film


Gas Phase









4. Purging of reactant B


2. Chemisorption of A on substrate and
purging of excess A in gas phase


Figure 1-10. Schematic of one ALD cycle forming a monolayer of film









CHAPTER 2
LITERATURE REVIEW

2.1 Diffusion in Thin Films

To prevent diffusion of copper into silicon by depositing a barrier film between the two

materials, it is important to understand the fundamental mechanism of diffusion through thin

films. When two dissimilar materials are put in contact with one another, atoms from one

material diffuse into the other and vice versa because of the difference in concentration gradient.

Fick' s law (Eq. 2-1 for 1D diffusion) gives a relation between the atomic flux at position x and

the local concentration gradient of diffusing species.

J=-DdC\
dx (2-1)

The atomic concentration is C and atomic flux in direction x is J. For solids the diffusion

coefficient, D, is most often expressed by Equation 2-2.


D = Do e RT
(2-2)

Where, Do is a constant which is independent of temperature. It depends on the thermodynamic

and kinetic properties of interaction between the two materials. Ea is the activation energy for

diffusion, R is the universal gas constant, and T is the temperature in K. The activation energy

for diffusion depends on the mechanism of diffusion. Diffusion can occur for example by lattice

diffusion, surface diffusion, grain boundary diffusion or dislocation diffusion. There is a

hierarchy of diffusion rates for these mechanisms of diffusion. Experimental observations for

diffusion reveal that the activation energy for diffusion through a material depends on its

absolute melting temperature (Eq. 2-3).

Ea acA Tm(2-3)









The proportionality constant, A, depends on many factors such as lattice structure and material

interactions. The melting temperature (in Kelvin) of the film through which the material is

diffusing is Tm. The values of A for surface, grain boundary, dislocation and lattice diffusion are

13, 17, 25 and 34 J/(gmol.K), respectively and the corresponding values of pre-exponential term

Do are 0.014, 0.3, 2.1 and 0.5 cm2/SOC, TOSpectively.36 Since the diffusion of material across

surfaces is being considered, the surface diffusion can be ignored in the discussion. It is quite

evident that the activation energy for diffusion is lower for grain boundary diffusion as compared

to lattice diffusion. Since the diffusion coefficient varies exponentially with activation energy,

the diffusion coefficient for grain boundary diffusion is several orders of magnitude higher than

that for lattice diffusion.

Since both lattice and grain boundary diffusion depend on temperature, a cross over occurs

where the mechanism of diffusion changes from one to the other. The temperature at which this

crossover occurs is known as the Tammann temperature. It is widely believed that the Tammann

temperature is between one-half or two-thirds of the melting temperature Tm of the solid.37 Since

almost all the potential diffusion barrier films have high melting temperature, grain boundary

diffusion is the predominant mechanism by which diffusion occurs in these films. The problem

becomes severe for thin film diffusion barriers because the grain boundaries in thin films are of

the order of thickness of the film for polycrystalline films.

The discussion on various diffusion mechanisms reveals two important properties of a

good diffusion barrier film: high melting temperature and defect-free microstructure. A high

melting temperature material is expected to have a higher activation energy for diffusion. The

primary reason why diffusion barrier research has focused on transition metal compounds is

because of the high melting temperature of transition metals along with their compounds such as









nitrides and carbides. The melting temperature of some transition metals and their compounds

are listed in Table 2-1.

A perfect diffusion barrier would have no grain boundaries or micro-structural defects. A

defect-free single crystal would be the best diffusion barrier because of the absence of grain

boundaries. However, lattice mismatch and thermal budget limitations make the growth of

single crystal diffusion barriers diffacult.38 The next best solution is amorphous films as these

films have no long range order. Hence, one of the primary structural requirements for diffusion

barrier is to have amorphous microstructure. It should be noted that the material has to be

perfectly amorphous. Nano-crystalline film, which has grain boundaries of the order of few

nanometers, is worse than polycrystalline fi1m as it provides a network of pathways through

which copper could diffuse.

2.2 Refractory Carbides and Nitrides as Diffusion Barriers

The maj ority of research towards finding diffusion barrier materials for copper

metallization has focused on transition metals and their compounds. In particular, transition

metal nitrides and carbides have been extensively researched as diffusion barrier to copper. To

understand what makes these refractory carbides and nitrides good candidates for diffusion

barrier application in copper metallization, a brief survey of the general properties of these

compounds is presented.

Transition metal carbides and nitrides can have a significant concentration of vacancies.

These compounds can have significantly varying C or N content with appreciable vacancy

concentration. Therefore, these compounds generally deviate from stoichiometric formulations.

Carbon and nitrogen are interstitially located in the transition metal sublattice. Two kinds of

bonding exist in these compounds, metal-to-metal and metal-to-nonmetal (C or N). The metal-









to-metal bonding is 'metallic' whereas the metal-to-nonmetal bonding is covalent. The metal-to-

oxygen bonding is presumed to be ionic.

Transition metals and their compounds have many properties that make them good

candidates for diffusion barrier application. Carbides and nitrides of transition metals have

extremely high melting temperatures (2000 4000 oC) and are therefore referred to as 'refractory

carbides and nitrides'. For instance, TaC has a melting temperature of 3983 OC, which is the

highest melting temperature of any known material.39 Since the activation energy for diffusion

depends on the melting temperature of the compound videe supra), refractory carbides and

nitrides can be expected to be good barriers against copper diffusion.

Diffusion barrier films should show good mechanical stability because thin films deposited

by vacuum processes are almost always under stress.40 The processes subsequent to deposition

of barrier films also involve thermal cycling and contribute to the stress levels in the barrier film.

Poor mechanical strength could cause fracture of the film and lead to diffusion of copper through

macroscopic gaps in the barrier film. Refractory carbides and nitrides are also known for their

excellent mechanical stability. These compounds, such as TiN and WC, are extremely hard and

have found application as wear-resistant coatings. While the exact reason for the extreme

hardness of these materials is not known, it is believed that the presence of either C or N in

transition metal stiffens the metal matrix resulting in increased hardness of refractory carbides

and nitrides as compared to corresponding transition metals.39 Refractory nitrides and carbides

are brittle and they undergo minimal plastic deformation at low temperature (< 800 oC).

However, these compounds become plastic at higher temperature.

Another important property of the barrier film is its electrical resistivity. In dual

damascene processing, one trench and via feature is separated from the other by the barrier film.









The barrier film forms part of the interconnect stack and hence low resistivity of barrier film is

desired. Most refractory carbides and nitrides are metallic in nature and have low resistivity.

Moreover, the resistivity of these compounds shows little temperature dependence. The

resistivity of refractory carbide or nitride thin film depends on factors such as porosity, non-

metal to metal ratio and impurity concentration. Oxygen impurity is always present in refractory

carbides and nitrides due to the high affinity of these compounds for oxygen. Oxygen in the

lattice can scatter electrons and increase the resistivity of the film. It should be noted that even

though refractory carbides have low resistivity in general, a large variation in resistivity values

for the same refractory carbide or nitride is observed because of factors discussed above.

2.3 Deposition Methods for Diffusion Barrier Films

Thin films can be prepared by various techniques such as evaporation, PVD, CVD and

ALD. Films for diffusion barrier applications have been predominantly deposited by PVD and

CVD or one of their variants. The deposition technique is an important consideration for

diffusion barrier film because many film properties depend strongly on the deposition process.

Properties such as adhesion, microstructure, impurity concentration, within wafer uniformity,

and conformality depend on the deposition process. This section details the advantages and

limitations of most common deposition techniques for diffusion barrier film deposition.

Important processing parameters for these deposition techniques and their general implications

on film properties are also discussed.

2.3.1 Physical Vapor Deposition

In general, PVD includes deposition processes such as evaporation, sputtering, pulsed laser

deposition, and high velocity oxygen fuel. For diffusion barrier film deposition, the most

prominent PVD technique used is sputtering. In sputtering, ions generated by a plasma are

accelerated towards the target (substrate). The sputtered target atoms deposit on the substrate









resulting in the formation of thin film. Plasma generation can be done either by radio frequency

(RF) or direct current (DC). Radio frequency sputtering is used for a non-conducting target,

whereas DC sputtering is used for conducting targets. When a magnetron is used to trap the

electrons in the plasma closer to the target for low pressure plasma sustainability, the process is

referred to as magnetron sputtering. For deposition of a metallic substance, pure metal is used as

the target. To deposit alloys such as WT\x, a reactive gas (e.g., N2) is USed along with an inert

gas to incorporate nitrogen into the metal. This process is often referred to as reactive ion

sputtering.

Important processing parameters for PVD are deposition pressure, applied voltage, target

to substrate distance, partial pressure of reactive gas (for reactive ion sputtering), substrate

temperature, and bias voltage on the substrate. The most important factor affecting the film

properties is the energy of sputtered particles at the substrate surface. Applied voltage, target-to-

substrate distance, chamber pressure and bias voltage on the substrate determine the energy of

the sputtered particles impacting the surface of the substrate. Applied voltage determines the

energy of the sputtered particles ejected from the target surface. The deposition pressure

determines the mean free path of the sputtered particles. The higher the mean free path, the

higher the kinetic energy of the particles deposited on the substrate. The target-to-substrate

distance, chamber pressure, and the bias on the substrate determine how much of the energy of

the sputtered particle is lost due to collisions prior to reaching the substrate surface. All these

factors can influence the film adhesion, porosity, crystallinity, texture, and substrate damage.

The substrate temperature can also influence the crystallinity and texture of the deposited film.41

The partial pressure of reactive gas used in reactive ion sputtering can influence the

incorporation of reactant atoms in the deposited films.42









The advantages of sputtering are good film adhesion, uniform film deposition, low

contamination and ease of deposition. The stoichiometry of alloys deposited by reactive ion

etching can also be altered easily be changing the partial pressure of reactant gas. The maj or

disadvantage of this technique is that it has poor conformality for ultra-thin barrier film

deposition. While I-PVD can extend the use of PVD technique by improving the conformality of

films deposited by PVD28,43, new deposition techniques will be required in the future to deposit

highly conformal films in high aspect ratio features. Another disadvantage of PVD is the

columnar microstructure of films deposited by PVD. Columnar microstructure provides short

pathways for copper to diffuse along grain boundaries.

Although it is well acknowledged that PVD cannot be extended to deposition of ultra thin

barrier films, researchers have used the technique extensively because PVD is one of the easiest

techniques to test new materials for diffusion barrier application. Unlike CVD which requires

suitable precursors for deposition of thin films, PVD only needs a suitable target of good purity

for deposition. A review of PVD films is not included in this section because PVD process is not

capable of depositing ultra thin barrier films for future interconnects. Moreover, films of a

particular material deposited by PVD films can have markedly different properties than the same

material deposited by CVD because of the differences in deposition mechanism.

2.3.2 Chemical Vapor Deposition

In CVD, gaseous reactants or precursors are used to deposit thin films onto the substrate.

For deposition of diffusion barrier film, the different variants of CVD that have been

predominantly used are plasma enhanced CVD (PECVD), low pressure CVD (LPCVD), metal-

organic CVD (MOCVD), aerosol assisted CVD (AACVD), atomic layer CVD (ALCVD or

ALD) and plasma assisted ALD (PAALD).










The important processing parameters for a CVD reactor are reactor pressure, precursor(s)

selection and their carrier gas flow rates, substrate temperature, and for PECVD and PAALD,

power of plasma. The reactor pressure determines the number of collisions that involve the

precursor molecules in the gas phase, which is important for determining the rate of homogenous

gas phase reactions. The reactor pressure together with carrier gas flow rates and substrate

temperature determine the flow dynamics inside the reactor. The thermal energy supplied to the

substrate aids the precursor decomposition on the substrate surface and is important to

determining the rate of heterogeneous reactions.

A characteristic growth curve for thermal CVD is shown in Figure 2-1. Two distinct

growth rate regimes are visible in the curve. At low deposition temperature, there is an

exponential increase in film growth rate with deposition temperature. This behavior is observed

when the rate determining step for the film growth is thermally activated, most commonly by

surface reaction, hence this growth rate regime is often termed 'reaction limited' or kineticallyy

controlled'. The growth rate in this growth regime, G, is given by Equation 2-4.

-Ea
G = Ae RT(24

The activation energy for the film growth is Ea, proportionality constant is A, the universal gas

constant is R and the absolute temperature of deposition is T. The slope of the plot of In(G) vs.

1/T gives the value of Ea.

At higher deposition temperature, the film growth rate is proportional to Tm, where m has a

value of approximately 1.5. In this growth regime, the rate determining step is the mass transfer

of the precursor to the substrate surface and hence it is often termed 'mass transfer controlled'

growth regime. This growth regime is sensitive to the reactor pressure and flow dynamics of the

reactor. By reducing the reactor pressure, the growth rate increases because the boundary layer









thickness decreases with decrease in pressure. A decrease in boundary layer thickness increases

the mass transfer of the reactants to the substrate surface and increases the fi1m growth rate.

A CVD process uses a heated substrate to aid the decomposition of the precursor(s) on the

substrate. In PECVD, a plasma is used to aid the decomposition of the precursor(s). The

primary advantage of PECVD is that the plasma facilitates the deposition of films at low

temperature as compared to thermal CVD. This is particularly important when halides such as

WF6 or WCl6 arT USed as precursors because halide precursors usually require very high

temperature for decomposition. The disadvantage of PECVD is that plasma gives directionality

to the atoms, molecules and/or ions and this has an adverse effect on the conformality of

deposited film. Another disadvantage of PECVD is that excessive fragmentation of the

precursor leads to incorporation of impurities such as halides into the film. Halide impurities are

undesirable in diffusion barrier films since they can increase the resistivity of the fi1m.

Moreover, halides present at the interface are known to create adhesion problems between the

barrier and copper.44 Excessive fragmentation can also lead to deposition of undesirable

compounds. For instance, PECVD of WNx has W also present in the film. Precise control of the

precursor fragmentation due to the plasma is difficult. The plasma can also result in damage to

the substrate surface.

In LPCVD, the deposition of thin films is carrier out at sub-atmospheric pressures. Low

pressure increases the mean free path of the precursor(s), which in turn decreases the collision

rate of precursor molecules and carrier gas in the gas phase prior to reaching the substrate

surface. Low pressure also improves the within wafer uniformity of the film. Most CVD

processes used in the semiconductor industry are LPCVD processes.









In MOCVD, a metal-organic precursor is used for deposition of thin film. As compared to

CVD, which uses inorganic precursors, MOCVD has many advantages. Inorganic precursors

predominantly use halogen containing precursors such as WCl6, which are corrosive. Moreover,

the byproducts of the halogen containing precursors are HCI or HF, which are also very

corrosive and can etch the underlying substrate. Films deposited by halogen containing

precursors have shown pitting because of the etching action of precursor or byproducts videe

infr~a). In general, halogen containing precursors are hazardous and need special handling. In

MOCVD, the precursor is mostly non-corrosive. The metal organic precursor may or may not

have halide containing ligands. Most of the by-products of metal-organic precursors that do not

contain halides are non-corrosive and hence do not cause pitting of the substrate. Even for

halogen containing metal-organic precursors, the likelihood of pitting is less because of the

presence of non-corrosive organic byproducts. A limitation of MOCVD is that sometimes

carbon containing ligands can also incorporate into the film. The C impurity can be either

beneficial or not depending on application of the thin film. For diffusion barrier application,

particularly for WNx, a small amount of C is beneficial because it can increase the

recrystallization temperature of the film. But greater amounts of C deposition can lead to

significant increase in film resistivity and hence is not desirable for diffusion barrier film.

Overall, MOCVD is a very good technique for diffusion barrier film deposition because of its

low deposition temperature, non-corrosive precursor, minimal halide contamination, and

minimal substrate damage.

AACVD is a variant of CVD where solid precursor is dissolved in a solvent and delivered

to the reactor by aerosol generation. The aerosol generation is achieved by a nebulizer and the

aerosol is transferred to the reactor by carrier gas. The advantage of using AACVD is that solid










precursors with low vapor pressure can be delivered to the CVD reactor. The disadvantage of

AACVD is that the solvent could participate in the homogenous or heterogeneous reactions,

thereby altering the fi1m properties.

ALD (or ALCVD) is a limiting case of CVD in which the substrate is exposed to the

precursors in a sequential manner to achieve self-limiting growth videe supra). The major

advantage of the ALD process is the precise thickness control and excellent conformality. Since

the growth of film occurs layer by layer, an ALD process can give very good thickness control

and the conformality is also excellent. In fact, researchers have shown that even in features with

200:1 aspect ratio, the ALD process achieves excellent conformality.45 Another important

advantage of ALD is that since the precursors are introduced in the reactor alternately, there is no

homogenous reaction of precursors.

The two plots that are characteristic of an ALD process are growth rate versus temperature

and growth rate versus precursor exposure time (Figures 2-2 and 2-3). In Figure 2-2, the growth

rate increases linearly with temperature at lower temperature. Since the precursor adsorption and

reaction between precursor and surface are thermally activated processes, the thermal energy

available for precursor reaction at low deposition temperature is insufficient to achieve complete

saturation or reaction.46 The second growth regime is characterized by film growth rate

independent of temperature. This region is often referred to as the 'process window' of ALD

because the fi1m growth rate is insensitive to substrate temperature. The third growth regime is

characterized by decomposition of the precursor resulting in a 'CVD-like' growth. Figure 2-3

shows how the growth rate of film varies with exposure time. Initially when complete saturation

of the surface is not achieved, fi1m growth rate increases with an increase in the exposure time.

Once complete saturation is achieved, the growth rate is independent of precursor exposure time,









unless sufficient time is given for chemical reaction between the chemisorb layer and any gas

phase species, including impurities.

2.4 Characterization and Testing of Diffusion Barrier Films

Diffusion barrier films need to be characterized to understand the factors that affect the

diffusion barrier performance of the film. The barrier film also needs to be tested for its ability

to prevent copper diffusion. This section summarizes various techniques that have been used for

diffusion barrier film characterization and testing.

2.4.1 Diffusion Barrier Film Characterization

There are a number of properties of the deposited film that are very important. These

include microstructure, composition, density, resistivity, thickness, surface roughness, bonding

between different atoms and adhesion of barrier film to copper and dielectric (low-k or SiO2).

Various techniques used to analyze these film properties are listed in Table 2-2. This section

details the efficacy of each technique in analyzing various film properties without discussing the

guiding principles of each technique. Detailed explanation of principles of analysis techniques

used in the present work are given in the Chapter 3.

X-ray diffraction (XRD) is the most common technique used to determine the crystallinity

and microstructure of thin films. For ultra thin films, grazing incidence XRD (GIXD) is used

because it has a small depth of resolution. These techniques can also determine the texture of the

films. Electron diffraction is another highly sensitive technique that can determine the

crystallinity in thin films. Imaging techniques such as SEM and TEM can provide information

on grain structure of the films. High resolution images have been used by many researchers to

determine whether the films have equiaxed or columnar microstructure.

Film composition can be determined by Auger electron spectroscopy (AES), X-ray

photoelectron spectroscopy (XPS), secondary ion mass spectroscopy (SIMS), energy dispersive










X-ray spectroscopy (EDS), Rutherford back scattering (RBS) and elastic recoil detection analysis

(ERDA). Auger electron spectroscopy and XPS are very effective techniques in determining the

elements present in the deposited film. All the elements in the periodic table, except H and He,

can be detected by using AES and XPS. However, film composition determination requires the

calibration to identify the relative sensitivity factors needed for quantification in both AES and

XPS. Pure elemental sensitivity factors can be used for quantification to obtain an estimate of

film composition but elemental sensitivity factors do not account for the matrix effect. Both

techniques have a detection limit of approx. 1 at. %. Both AES and XPS are surface sensitive

techniques with sampling depth of few tens of nanometers. Both these techniques have been

widely used for film composition determination and depth profiling of diffusion barrier films.

Secondary ion mass spectroscopy can be used for determination of film content, but this

technique is not used for determination of film composition. Its main use is in determination of

trace impurities in the film. The technique can detect impurity concentrations of ppm or even

ppb levels. The technique can also detect presence of H and He in the film. Energy dispersive

spectroscopy is another technique that has been used to determine elements present in the film,

however it can determine film composition only semi-quantitatively. This technique can detect

elements with atomic number greater or equal to 6. Since EDS has poorer sensitivity to lighter

elements in a heavy element matrix, this technique is not commonly used for transition metal

nitrides and carbides. The sensitivity of this technique is 1 at. %. Rutherford back scattering

(RBS) has been widely used for film composition determination of barrier films. This technique

has the advantage of not requiring standards for film composition determination. The detection

limit of RBS is about 4 at. % and hence this technique is not suitable for trace impurity detection.

Elastic recoil detection analysis is a complimentary technique to RBS and it can detect lighter









elements such as H in heavier element matrix. This technique has been widely used to determine

H element composition in deposited films. This technique has a detection limit of 0. 1 at. % and

a depth of resolution of few tens of nanometers.

The common techniques used to determine fi1m thickness are cross sectional scanning

electron microscopy (X-SEM) and transmission electron microscopy (TEM) imaging. Both

these techniques provide an accurate measure of film thickness. Film thickness from

profilometer measurements requires depositing fi1m on a section of the substrate and proofing

the step created by the film on the substrate. The accuracy of this technique depends on the

resolution of the profilometer, which is usually few tens of angstroms. Rutherford back

scattering and XRR can also provide measurement of film thickness, but these techniques are

indirect techniques. The accuracy of the measurement by RBS and XRR depends on the

accuracy of the variables used in simulation.

Two techniques that have been prominently used to determine film density are RBS and

XRR. Both these techniques are indirect techniques for density measurement. Rutherford back

scattering uses atomic density and the thickness obtained from either SEM or TEM analysis to

simulate the value of film density. Film density can be obtained from XRR by simulating the

fringes in the spectrum. Simulation of XRR pattern requires accurate values of film thickness

and roughness. Accuracy of both XRR and RBS in determining film density depends on the

accuracy of parameters used in the simulation.

The most common technique for determination of diffusion barrier resistivity is four point

probe. Four point probe determines the value of sheet resistance and this value when multiplied

with the film thickness (from SEM or TEM) gives the resistivity of the film. Bonding between

elements present in the film can be determined by XPS analysis. The change in bonding state of









an atom results in a shift in the binding energy observed in XPS spectra. This shift in binding

energy can be used to determine the atomic bonding of a particular element in the film. X-ray

photoelectron spectroscopy is the only technique that can give the bonding information of atoms

in thin films.

Surface roughness of the film can be determined by various techniques such as AFM,

stylus profilometer, scanning tunneling microscopy (STM), optical profilometer, SEM, TEM and

X-ray reflectivity (XRR). Of these techniques, AFM and stylus profilometer are contact

techniques because the probe used for roughness measurement is in contact with the surface.

Non-contact techniques used for surface roughness measurement include STM, optical

profilometer, SEM, TEM and XRR.

Among the contact measurement techniques, AFM provides the best surface resolution of a

fraction of an A~ngstrom. Atomic force microscopy can give 3-dimensional images of the

scanned surface and provide values of root-mean squared (RMS) surface roughness and average

grain size. The probe used in AFM is a cantilever that scans across the surface. A stylus

profilometer uses a diamond stylus to scan across a surface and provide 2-dimensional image of

the surface. The resolution of a stylus profilometer is few tens of nanometer. Stylus

profilometer is commonly used to measure uniformity of film in trenches or vias, whereas AFM

is commonly used to measure surface roughness of a blanket film. Since both AFM and stylus

profilometer are contact measurement techniques, they can damage the surface because of

scratching by the probe.

Non-contact techniques include imaging techniques such as SEM and TEM. These

techniques are good for qualitative determination of surface roughness but not suitable to obtain

RMS roughness. An optical profilometer uses light scattering to probe the surface roughness.









This technique has a good spatial resolution of few nanometers and can provide 3-dimensional

images as well as RMS roughness of the film. The disadvantage of this technique is that the

surface has to be reflective. Since most of the diffusion barrier films are metallic in nature, this

technique can be used to measure surface roughness of diffusion barrier films. X-ray reflectivity

is an indirect method for surface roughness measurement and it requires accurate values of film

thickness and density to simulate fringes in the XRR spectra. X-ray reflectivity is not commonly

used for surface roughness measurement. Scanning tunneling microscopy uses a very sharp

probe which tunnels current through the sample surface to obtain images of the surface with

atomic scale resolution. This technique requires very careful sample preparation and the imaging

area is very small. Hence this technique is not commonly used to measure surface roughness of

diffusion barrier films.

The best method to determine the adhesion strength of a barrier to copper and dielectric is

the four point bend test. This test measures the adhesion energies of interfaces in a multilayer

thin film stack. Another method that is commonly used to determine the adhesion of barrier film

qualitatively is the Scotch tape test. This test uses an adhesive tape that is applied to the thin film

and the tape is pulled to determine if the thin film adheres to the substrate. The Scotch tape test

is primarily used for preliminary evaluation of barrier film adhesion to copper and dielectric.

2.4.2 Diffusion Barrier Testing

Various direct and indirect methods are used to test the effectiveness of diffusion barrier

film in preventing copper diffusion. There are numerous techniques that can detect copper

diffusion through the barrier into silicon and the most effective technique to detect trace amount

of copper in silicon is electrical characterization. For a MOS test structure, electrical

characterization techniques that are used to detect trace amount of copper in silicon or SiO2 are

flat band voltage shift in CV measurement, I-V measurement and triangular voltage sweep









(TVS).47 For a Schottky contact test structure, the transient ion drift (TID) method can be used

to detect and quantify the amount of copper that has diffused into silicon through the barrier after

the film has been subj ected to thermal stress or bias temperature stress.

Ideally, a diffusion barrier test should involve integration of the barrier in actual chip

followed by electrical characterization to determine the barrier performance. However,

preliminary diffusion barrier testing is often done on test structures such as a MOS capacitor to

check for diffusion of copper through the barrier film. In a MOS test structure, copper metal

dots are deposited on barrier/ SiO2/ Si stack. The test structure is then stressed at high

temperature (150 300 oC) under a bias of few MV/cm. This accelerated diffusion test is often

referred to as bias temperature stress (BTS). Electrical characterization is then done to detect

any diffusion of copper through the barrier into the oxide and silicon. In CV measurement, the

flat band voltage is measured before and after BTS. If copper has diffused through the barrier

film, the flat band voltage shifts indicating that the barrier film has failed. The I-V measurement

can also be done on a MOS capacitor after BTS. In an I-V measurement, if the leakage current

density of the capacitor is higher than 10-6 A/cm2 for the applied electric field of 1.2 MV/cm,

then the capacitor is considered a failed capacitor.48 A leaked capacitor would mean that copper

has diffused through the barrier film. In TVS measurement, a positive voltage is applied to the

copper electrode and is slowly made negative at a constant rate. If copper has diffused through

the barrier, the mobile copper ions are inj ected back into the oxide from the Si-SiO2 interface.

This results in a jump in the measured current and copper diffusion through the barrier can be

detected. Among the three techniques, i.e. flat band voltage shift in CV measurement, I-V

measurement and TVS, the TVS technique is the most sensitive in detecting the presence of

copper in silicon.4









In the TID technique, a Schottky barrier test structure is made by depositing barrier film on

p-type Si. Copper metal dots are then deposited on the barrier and encapsulated by a protection

layer.49 If the Schottky test structure is quenched after thermal stressing, most of the copper in

silicon remains as an interstitial for a few hours.' The interstitial copper ions in silicon tend to

drift out when high electric field is applied. The resulting change in electric charge distribution

induces a transient capacitance signal, which can be correlated to the copper ion concentration in

silicon. This method has a detection limit of 1012 at/cm3.

For preliminary evaluation of diffusion barrier efficacy, researchers have used other

simpler techniques that do not require fabrication of a device. In one such test, a Cu/barrier/Si

(or SiO2) structure is thermally stressed by annealing at temperatures between 200 and 800 oC.

The higher the temperature at which the barrier shows copper diffusion into silicon, the better is

the diffusion barrier performance. The diffusion of copper through the barrier is then detected by

various techniques such as four point probe, depth profiling (by AES, XPS or SIMS) and XRD.

Four point probe measurement gives the sheet resistance of the Cu/barrier/Si(SiO2) stack. When

copper diffuses through barrier film, there is a sharp increase in sheet resistance. Depth profiling

by AES, XPS or SIMS uses Ar' sputtering to sample through the Cu/barrier/Si(SiO2) stack and

detect if copper has diffused through the barrier into silicon. The disadvantage of depth profiling

is that copper can be knocked forward the barrier film during the Ar' sputtering and the resulting

artifact could give a false positive. A solution to the 'knock-on' problem with sputtering is to

etch the copper layer after thermal stressing and prior to depth profiling. Since bulk copper has

been etched, the 'knock-on' effect due to sputtering can be avoided. The detection limit for

copper by depth profiling depends on the technique used for elemental detection. Secondary ion

mass spectroscopy provides the best resolution in detecting the presence of copper (detection









limit ca. 1015 atoms/cm3). Both AES and XPS can detect copper if its concentration in silicon is

ca. I at. %. Pre- and post anneal XRD measurements can provide information on diffusion of

copper through the barrier and into the silicon. Copper forms Cu3Si in silicon at a relatively low

temperature of 150 oC. The annealing temperature at which Cu3Si peaks appear in the XRD

spectra indicates the lowest temperature at which the barrier film has failed. All these

techniques, i.e. sheet resistance measurement, XRD, four point probe measurement and depth

profiling, have poorer resolution in detecting copper in silicon than electrical characterization.

However, these techniques can be very helpful in preliminary screening of diffusion barrier films

as these techniques are much simpler than electrical characterization.

Most of the research on diffusion barriers has been focused on refractory nitrides and

carbonitrides of Ti, Ta and W videe supra). The following sections review the work done on

refractory nitrides and carbonitride thin films of Ti, Ta and W deposited by CVD and ALD

techniques with primary focus on the suitability of these films for diffusion barrier application.

2.5 Titanium Nitride as Diffusion Barrier

2.5.1 Chemical Vapor Deposition of Titanium Nitride

Table 2-3 shows the inorganic and metal organic precursors that have been used to deposit

TiNx thin films by CVD. TiNx has been deposited using the inorganic precursors TiCl4 and Til4

as Ti source and NH3 or N2 plaSma as N source. Thermodynamically, the reaction between TiCl4

and NH3 is feaSible at temperature as low as 300 oC, but appreciable film growth rate is observed

only at 400 OC by thermal CVD.S1 Two different processes, LPCVD and APCVD (atmospheric

pressure chemical vapor deposition) have been used for thermal reaction of TiCl4 and NH3. Both

APCVD and LPCVD deposit nearly stoichiometric TiN films with N:Ti ratio close to 1.

Deposition by APCVD gives more uniform growth but the film contains a large amount of Cl

impurity. At 400 oC, the films contain as high as 50+20 at. % Cl impurity. Increasing the









deposition temperature leads to decrease in Cl content and films deposited at 620 oC have Cl

content of 5 + 2 at. %.51 Film resistivity has a direct correlation with the Cl content of film. The

resistivity of film decreases from 6000 CIR-cm for films grown at 400 OC to 200 CIR-cm for films

grown at 620 oC. When LPCVD is used to deposit TiN films from TiCl4 and NH3, the film

impurity content and resistivity are better than those deposited by APCVD. Films deposited by

LPCVD show 3 4 at. % Cl impurity for deposition at 450 oC and high quality films with

impurity content below the detection limit of AES and RBS are deposited at 650 oC.52 However,

O and H content in the film increases with increasing deposition temperature. Because of low Cl

content, films deposited with LPCVD have a lower resistivity. Resistivity of the film decreases

from 290 CIR-cm for films grown at 450 oC to 110 CIa-cm for films grown at 700 OC. The XRD

spectra of films deposited below 600 oC show that TiN films are textured with (200) preferred

orientation. However, the films deposited at 700 OC show a preferred orientation of (111).53

Diffusion barrier testing by leakage current testing on p~n junction of TiN/TiSi2 bilayer

deposited with CVD showed that the bilayer was able to prevent copper diffusion after annealing

at 550 oC for 30 min in N2 CHVITOnment.54 Annealing above 550 oC caused an increase in

leakage current suggesting that copper had diffused through the barrier film.

The main challenge for the LPCVD TiN process from TiCl4 and NH3 is the high Cl content

of film for low temperature deposition. It is believed that Cl impurities in the film have high

diffusivity. If these impurities diffuse to the interface, corrosion and adhesion problems can

result. Increasing deposition temperature does reduce Cl content, but higher temperature

processing is not desirable because of the thermal instability of low-k dielectrics above 450 oC.

Yokoyama et al.53 tried to use high temperature anneal in H2 CHVITOnment to remove Cl impurity

from the film. The process did reduce the Cl content of film from 5.7 at. % to 2.7 at. % after









annealing at 1000 OC for 30 min., but the complete Cl elimination from the fi1m was not

achieved. The N:Ti ratio of the film also decreased from 1.02 to 0.97 after the annealing

process, suggesting that the Cl and N are lost from the fi1m as NH4C1. Chlorine remains in the

film after annealing due to the low volatility of TiNxCly even at 1000 OC. Besides the Cl

problem, LPCVD from TiCl4 deposits polycrystalline films that have columnar microstructure.

As discussed earlier, columnar microstructure provides short diffusion pathways for copper

diffusion and is highly undesirable. The process has problems with particulates because NH3

reacts with TiCl4 to form NH4C1, a white solid. TiCl4 alSO reacts with NH3 at room temperature

to form TiCl4.2NH3, which is also a solid.5 This problem cannot be solved without changing the

precursor chemistry of deposition.

To address the Cl impurity concern, two different approaches have been attempted. One is

the use of an alternate halide source such as Til4 and the other is the use of N2 plaSma as N

source. Til4 approach is believed to lower the halide content of film because the Ti-I bond is

weaker than the Ti-Cl bond. The heat of formation of Ti-I at 298 K is -92 kcal/mol whereas the

heat of formation of Ti-Cl is -192 kcal/mol at 298 K.56 Moreover, since I is a larger element

compared to Cl, it would require much higher activation energy as compared to Cl to diffuse out

of TiNx and into the Si. TiNx deposition by LPCVD from Til4 and NH3 shows that the films

deposited are nearly stoichiometric with N:Ti ratio of 1.06: 1 at 430 oC deposition

temperature.56,57 The I and O impurities in films are less than 2 and 1 at. %, respectively. The

density of the film is also higher by approx. 10 % as compared to the density of PVD TiNx film.

The film has resistivity as low as 100 CIR-cm and the step coverage in a 0.5 Clm 3:1 AR feature is

90%. The Til4 prOCOSs does yield better films as compared to films deposited using TiCl4.









However, the films are still polycrystalline and amorphous film growth could not be realized.

Diffusion barrier testing for copper metallization was not reported for films deposited with Til4.

PECVD has been used to deposit TiNx using TiCl4 and H2 + N2 plaSma.58,59 When RF H2

+ N2 plaSma is used, violet colored films were deposited below 300 oC.59 The authors argued

that the films were probably TiCl3. A minimum temperature of 300 oC was required to deposit

TiNx films. The films deposited at and above 300 oC were TiNx. The films were slightly Ti-rich

and the lowest Cl content was 3 at. % at 500 oC deposition temperature. Increasing the H2 flOW

resulted in an increase in Cl content, whereas an increase in the N2 flow decreased the Cl content

of the film. Other properties of the film deposited with plasma assistance were similar to those

for film deposited by LPCVD process. While the use of plasma did allow deposition to occur at

lower deposition temperature, higher temperature close to 500 OC was still needed to deposit

films with minimal Cl impurity and low resistivity. So, the plasma process did not show any

significant advantage over the LPCVD process.

Tetrakis (dimethylamido) titanium (TDMAT) has been used as a precursor for TiN

deposition. When TDMAT is used in an inert atmosphere as a single source precursor, TiNxCv

films are deposited.60 Films can be deposited at temperature as low as 250 oC.61 Film deposited

at 300 oC has 27 at. % Ti, 17 at. % N, 11 at. % O and 35.2 at. % of C. X-ray photoelectron

spectroscopy analysis of the film shows that the C in the film exists in both carbidic form

(bonded to Ti) and amorphous or organic form.62 Films are also resistive, with the resistivity

varying from 1500 to 5000 CIa-cm. The films showed an 'aging' effect as evidenced by the post

growth increase in resistivity with time due to oxygen and moisture incorporation in the film.

Diffusion barrier testing of the film shows that the barrier fails after vacuum annealing at 500 OC.

The poor performance of the barrier films was attributed to the porous nature of the film.63









When TDMAT is used with H2 and NH3, f11ms were nearly stoichiometric with N:Ti ratio

of 1.64 Since TDMAT undergoes transamination reaction in solution, NH3 WaS used as co-

reactant to aid the transamination reaction in gas phase. The transamination reaction also lowers

the C incorporation in the films.65 The films have a low C content of less than 3 at. % and the O

content of the film varied between 4 and 9 at. %. The use of NH3 has a significant effect on the

lowest deposition temperature for film growth. Films could be deposited at temperatures as low

as 30 oC. The films also showed an 'aging' effect as the resistivity of the film increased by 20 %

within 24 hours of exposure to the ambient. The films were quite conductive and the resistivity

varied between 400 to 1500 CIa-cm for as deposited films. However, the films were

polycrystalline and showed columnar microstructure.64 Moreover, films deposited with NH3

showed poor step coverage and the films were less conformal at higher deposition temperature.66

The conformality of the film was 30 % at 450 oC deposition temperature. So, the process with

NH3 is HOt suitable for diffusion barrier application.

Hydrazine (N2H4) has been used as a co-reactant with and without NH3 to deposit TiNx

thin films using TDMAT.67 N2H4 has a higher N:H ratio and lower dissociation temperature

(140 oC) as compared to NH3. When N2H4 is USed with TDMAT to deposit thin films, the N:Ti

ratio of film can be modulated by changing the N2H4:TDMAT ratio. The films contain very high

O content (ca. 36 at. % at 100 OC deposition temperature). The films also show the 'aging' effect

and the film resistivity more than doubles after 150 min exposure to ambient. The resistivity of

the as deposited film is 3000 CIR-cm for deposition at 200 OC. The 'aging' effect is attributed to

the diffusion and subsequent reaction of oxygen with the unsaturated Ti atoms. To improve the

film properties, a mixture of N2H4 and NH3 WaS used to deposit thin films. The films deposited

with N2H4 and NH3 were polycrystalline and nearly stoichiometric TiN with O content below the









detection limit of RBS. The authors argued that while N2H4 WaS a better reducing agent, NH3

was a better transamination reagent. So, the fi1ms deposited with N2H4 and NH3 had the best

fi1m properties. The resistivity of the fi1m was ca. 1000 CIR-cm so the use of NH3 with N2H4

couldn't lower the film resistivity. No diffusion barrier testing on these fi1ms has been reported

so far.

Plasma has been used both during growth and post-growth to improve the properties of

film deposited with TDMAT. When the N2 plaSma treatment is used post-growth, films show an

increase in N:Ti ratio.68,69 The C content of the film decreases from 23 at. % to 3 at. % upon

exposure to N2 plaSma. The films also show a low resistivity of 720 CIR-cm and no 'aging'

effect is observed for plasma exposed films. The decrease in resistivity due to plasma exposure

is believed to be because of a decrease in (N+C+O):Ti ratio because N, C and O act as scattering

centers for electrons.69 The plasma treatment is helpful only if done in situ. Once the films are

exposed to the ambient, subsequent plasma exposure does not cause a dramatic improvement in

film properties. Plasma exposure also increases the film density. The plasma exposure,

however, does not have any effect on diffusion barrier performance of the film. Films that are

exposed to N2 plaSma after deposition fail at the same temperature of 550 oC as the films that are

not exposed to plasma.

Tetraki s(ethylmethylamido) titanium (TDEAT) has been used as a Ti precursor to deposit

TiN, films with and without co-reactant NH3. When the precursor is used without NH3, a

minimum temperature of 350 oC is required for thin film deposition.70 For films deposited at

450 oC, the film has a N:Ti ratio of 1. The films have significant C content (20 at. %) suggesting

that the films are actually TiNxC,. The films have a high O contamination of 18 at. % possibly

because of the post growth diffusion of O in porous TiNxC, films. The film density has a value









of 3.0 g/cc, which is low as compared to the bulk TiN density of 5.4 g/cc. The fi1m shows an

'aging effect', whereby the fi1m resistivity increases with exposure time to the ambient because

of the adsorption of O in the fi1m. The resistivity of film deposited at 420 oC increases from

1500 CIR-cm to ca. 8000 CIR-cm after exposure to air for 24 hr.68 To improve the fi1m resistivity,

in situ N2 plaSma exposure was employed. The plasma process densifies the films and

incorporates N into the fi1m as evidenced by the increase in N content from 23 to 44 at. % after

plasma exposure. The C content of the fi1m also shows a small decrease from 30 to 23 at. %

after plasma exposure. The O in the film is decreased to below 3 at. % by plasma exposure

resulting in a decrease in film resistivity to below 1000 CIR-cm. The film does not show an

'aging effect' as the film resistivity increases by only 4 % after air exposure for 24 hr. Since

Wang et al.68 deposited TiNx for application as a barrier for W in plugs, no copper diffusion

testing was performed.

To improve the properties of films deposited with TDEAT, NH3 has been used as a co-

reactant. With NH3, TDEAT shows an increase in growth rate by a factor of 2 3 as compared

to deposition without NH3-70 It is believed that this increase in growth rate is because of

transamination. The films are slightly N-rich with N:Ti ratio of 1.2: 1 for deposition at 425 oC.

The C and O contamination in the film is ca. 0.5 at. %. The films deposited with NH3 have a

higher density of 4.2 g/cc (at 4250C) as compared to that for films deposited without NH3.

Increasing deposition temperature, pressure or NH3 flow rate results in a decrease in film density

and resistivity. However, increase in TDEAT flow results in a decrease in film density and

resistivity.n The lowest resistivity of 200 CIR-cm was observed for deposition at 425 oC. The

drawbacks of using NH3 aS CO-reactant are the poor conformality and particle generation due to

high precursor reactivity. Poor conformality is observed with NH3 because NH3 addition









increases the growth rate of film and the growth mechanism changes from surface reaction

limited to mass transfer limited. A process operated in the mass transfer limited growth regime

is expected to give poor conformality.64 IHCTreSe in deposition temperature further increases the

reactive sticking coefficient and worsens the conformality.7 The conformality of film deposited

with NH3 at 425 OC in a 0.45 Clm 3 AR contact hole is only 30 %. So, the process with NH3 1S

not suitable for deposition of highly conformal diffusion barrier film.

To improve the property of film deposited with TDEAT, Yun et al.72 USed N2 and H12

plasma to deposit thin films via PECVD. When N2 plaSma is used to deposit TiNx, the films

were C-rich (45 at. %) and the Ti and N content of the film was 25 and 13 at. %, respectively.

The films had a resistivity of 1500 1700 CIR-cm. When H12 plaSma was used instead of N2

plasma, the films show low C content (17 at. %), but the film N content was also reduced (6 at.

%). The film resistivity was between 1200 and 1500 CIa-cm. Overall, the use of plasma with

TDEAT does not result in deposition of good quality films with low resistivity and favorable

composition.

2.5.2 Atomic Layer Deposition of Titanium Nitride

Table 2-4 shows the precursors that have been employed for deposition of TiNx thin films

by ALD. Two main inorganic Ti sources used to deposit TiNx are TiCl4 and Til4. Since Ti in

these precursors is in the +4 oxidation state, reducing agents are needed to reduce Ti to the +3

oxidation state in TiN. The reducing agents that have been used are NH3, Zn,

dimethylhydrazine, tert-butylamine and allylamine.

The deposition using TiCl4 and NH3 has been attempted between 350 and 500 oC. The

growth by ALD mode is nonlinear for first 50 cycles because of the initial nucleation and growth

on the SiO2 substrate.72,172 The temperature process window, which is characteristic of an ALD

process, is not observed for TiCl4 and NH3 chemistry. The growth rate per cycle increases with









increase in deposition temperature. Films deposited were polycrystalline and showed dense

columnar structure.73,74 The ratio of Ti:N was ca. 1:1, but the films contained Cl contamination.

Similar to CVD deposition, ALD showed a decrease in Cl contamination with an increase in

deposition temperature." The resistivity of film decreased from 250 to 50 CIa-cm as the

deposition temperature was increased from 350 to 500 OC. There is a direct correlation between

Cl contamination in the film and the resistivity of the film. Increase in the Cl contamination

increases film resistivity." At least 100 nm thick barrier film was required for complete surface

coverage of TiNx on SiO2 as determined by low energy ion scattering spectroscopy (LEISS).

The adhesion of TiN film with copper and Sio2 WaS excellent. The adhesion energy values for

TiN on SiO2 and copper were 60 and 25 J/m2, respectively as determined by the four point bend

test.76 In BTS testing of 5 nm thick TiN film, the barrier film did not fail after stressing at 2

MV/cm at 200 OC for various times.

There are 3 critical drawbacks of TiNx films deposited with TiCl4 and NH3. First, the films

are polycrystalline and have columnar microstructure. As discussed earlier, polycrystalline films

can act as fast diffusion pathways for copper and hence are undesirable. The films also had

columnar microstructure, which provides shorter pathways for copper diffusion. Second, the

deposition temperature to obtain good quality TiNx films with minimal Cl contamination is close

to 500 oC, which is relatively high since the low-k dielectrics are thermally unstable at this

temperature. Higher deposition temperature for TiCl4 is needed because the Ti-Cl bond is very

strong (bond energy of 429 kJ/mol). Moreover, higher temperature is also needed because at low

temperature, NH3 reacts with Cl adsorbed on surface to form NH4Cl and temperature above 370

oC is required to sublime NH4 -.74 Third, HCI formed as a byproduct of the reaction can









severely etch the substrate and cause pitting. Severe copper pitting has been observed when

TiNx is grown on copper.73,74,77

One of the ways to reduce deposition temperature is to use PAALD. Plasma assisted ALD

using TiCl4 and NH3 has been investigated by Elers and coworkers.7 Similar to thermal ALD,

PAALD also did not show the characteristic ALD process window for growth temperature. The

growth rate of films deposited with PAALD was significantly higher than that for thermal ALD

process. Film deposited by PAALD had lower Cl contamination as compared to films deposited

with ALD and the lowest Cl contamination of 1.2 at. % was observed at 400 OC deposition

temperature. The change of plasma power in PAALD resulted in a change in N:Ti ratio in the

fi1m. The N:Ti ratio increased with an increase in plasma power. Such modulation of

composition is not possible for thermal ALD, where composition is determined only by growth

temperature. While PAALD did have some advantage over thermal ALD, the process still was

not able to deposit Cl-free films. Moreover, the fi1ms were polycrystalline with columnar

microstructure. Furthermore, the use of halide chemistry also meant that the process would

cause pitting on the substrate surface.

To overcome the problems related to corrosive byproducts of the reaction, additional

reducing agent such as Zn has been used in the ALD process.75,79 When Zn is used as a reducing

agent, there is no difference in growth rate per cycle. However, the Zn does influence the film

texture. When Zn is used as reducing agent along with NH3, the films have (111) preferential

orientation, whereas without Zn the preferred orientation of the film is (100). No Zn impurity is

found in the TiNx films.7 The film composition with or without Zn is similar except Cl content,

which is higher for films deposited without Zn. Films deposited with Zn also had about 5 times

lower resistivity than those deposited without Zn. Even though very good film qualities are









obtained with Zn pulses in TiNx ALD, this process is not suitable for diffusion barrier deposition

because Zn is an electrically active impurity in silicon and its incorporation in active areas of a

transistor can severely affect the device performance.

Since 1,1 -dimethylhydrazine (DmHy) is a much stronger reducing agent than NH3, ALD

TiNx using TiCl4 and DmHy has been investigated.so The films deposited had a very high Cl

content, up to 23 at. % at 200 OC deposition temperature. The films were also more resistive

because of the increased Cl and C content in the film. The lowest resistivity obtained was 500

CIR-cm at 350 oC deposition temperature. The films were weakly polycrystalline. The use of

DmHy did not show any significant improvements over NH3 aS CO-reactant. In fact, the

resistivity and impurity content of films deposited was worse when DmHy was used.

tert-Butylamine and allylamine have also been investigated as reductive N sources to lower

the deposition temperature of TiNx ALD using TiCl4-81 Depositions were done between 400 and

500 oC. Films deposited with tert-butylamine had a low growth rate of 0.03 A/cycle. Addition

of NH3 to tert-butylamine increased the growth rate to 0.15 A/cycle. Film resistivity was similar

to those deposited with NH3 only and the contamination level of C and H was also below 1 at. %.

By using tert-butylamine, the deposition temperature could be lowered from 500 to 400 OC to

obtain low resistivity TiNx films with minimal Cl contamination (1.4 at. %). However, the

process could not deposit amorphous TiNx films.

When allylamine was used as the reducing agent, the films had higher C and Cl

contamination of 7 and 6 at. %, respectively.8 This resulted in an increase in film resistivity to

360 CIR-cm. Addition of NH3 to allylamine did not have any significant effect, although the Cl

contamination decreased to 4 at. % and C contamination increased to 9 at. %. As with tert-

butylamine, films deposited with allylamine were weakly polycrystalline. Additional diffusion









barrier testing of TiNx films deposited with either tert-butylamine or allylamine has not been

reported yet.

TiNx ALD has been attempted using Til4 and NH3-7982,83 The ALD process did not show

the characteristic temperature process window as the film growth rate increased with increase in

deposition temperature.83 The films were Ti-rich and the N content of film increased from 20 at.

% at 350 oC to 45 at. % at 500 oC. Impurity content of the film, as measured by RBS, showed

that the films had low I content (2 at. % at 350 oC and 0 at. % at 400 OC). The films were

heavily contaminated with O and the film O content decreased with increase in deposition

temperature. Films deposited at 350 oC showed O content of 40 at. %, while those deposited at

500 oC showed O content of 10 at. %. The TiNx films were polycrystalline and the preferred

orientation was dependent on deposition temperature. Films deposited below 425 OC showed

preferred orientation of (100), whereas films deposited above 450 oC showed preferred

orientation of (111). The film resistivity for deposition at 400 OC was 380 CIR-cm. Overall, high

porosity and the presence of impurities such as I and O made the films deposited by this process

unsuitable for diffusion barrier deposition.

To overcome the issues related to halogen impurities in films deposited using inorganic

precursors, organometallic Ti precursors have been investigated for deposition of ALD TiNx thin

films. Tetrakis(ethylmethylamido) titanuim (TEMAT), or Ti[N(C2H5)(CH3 14, has been used

with NH3 to deposit ALD TiNx thin films. Self-limiting growth of TiNx has been reported using

TEMAT.84,85 Unlike ALD from halide precursors, ALD from TEMAT showed the characteristic

ALD process window. While one report established the ALD process window for TEMAT

between 150 and 220 oC,s another report located the ALD process window between 170 and

210 oC.84 The deposition temperature for TEMAT is considerably lower than that for inorganic









Ti precursors. The growth rate per cycle is 5 A/cycle, which is about 1.6 ML/cycle for TiNx

fi1m. It has been argued that the fi1m growth of more than 1 ML/cycle is possible if NH3 Can

react with TEMAT adsorbed on the surface and TEMAT can react with chemisorbed NH3. Film

growth during the individual pulses of both TEMAT and NH3 WOuld result in film growth rate

greater than 1 ML/cycle.84 Unlike films grown with inorganic Ti precursors, fi1ms grown with

TEMAT had amorphous microstructure. While fi1ms deposited with inorganic Ti precursors had

a N:Ti ratio close to 1:1, films deposited with TEMAT had N:Ti ratio of 0.55 suggesting that the

film deposited is actually Ti2N. C and H contamination in the fi1m was 4and 6 at. %,

respectively for films deposited at 200 OC. The film resistivity varied between 230 CIR-cm for

films grown at 160 oC to 8000 CIa-cm for films grown at 200 OC. Diffusion barrier testing for

Ti2N films showed that the barrier was able to prevent bulk copper diffusion after annealing at

600 oC in vacuum (as determined by XRD of film post-anneal). After annealing at 650 oC,

Cu3Si peaks appeared in XRD suggesting that copper had penetrated through the barrier film.

Ti2N films deposited by TEMAT showed many properties expected from a good diffusion

barrier film. To evaluate if Ti2N is effective in preventing trace copper diffusion, additional

electrical testing of barrier film is required.

TDMAT has also been used with NH3 to deposit TiNx films via ALD. Unlike TEMAT,

TDMAT did not show the characteristic temperature process window for ALD. The reaction of

TDMAT and NH3 is HOt self limiting. Precursor decomposition study by FTIR indicated weak

dependence of growth rate on TDMAT exposure time because of incomplete desorption of

residual Ti(N(CH3)2)x species.86 Mechanistic study by FTIR also indicated that the film growth

occurs via transamination between TDMAT and NH3. Similar to TEMAT, TDMAT was also

able to deposit TiNx thin films at relatively low deposition temperature. At 180 oC deposition









temperature, the growth rate of TiNx was 2 A/cycle.8 Films were Ti-rich and the N:Ti ratio of

ca. 0.5 indicated the presence of Ti2N. The C contamination in the fi1m was less than 10 at. %

but the fi1ms had a very high O contamination which decreased with increase in deposition

temperature.86 The fi1ms were highly porous (> 35% porosity) which explained the low density

of Ti2N fi1m. The highest film density was 3 g/cc while the bulk density of stoichiometric TiN is

5.2 g/cc. High porosity results in an increase in O diffusion through the barrier film and a

corresponding increase in film resistivity. The lowest resistivity of 1.4 x 104 CIR-cm was

observed at 240 oC deposition temperature. High porosity and extremely high resistivity of Ti2N

films deposited by TDMAT makes this process unsuitable for diffusion barrier application.

To improve the Ti2N film quality, post-growth rapid thermal nitriding and plasma exposure

have been explored." Rapid thermal nitriding at 500 OC resulted in a net decrease in organic C

in the film (ca. 5 at. %) and the films were amorphous even after nitriding. Increasing the

nitriding temperature to 700 OC resulted in formation of macroscopic cracks in the barrier film.

Post-growth plasma exposure decreases the film C content below the detection limit of XPS.

Since higher N content was observed on the surface of Ti2N than in bulk, N most likely gets

incorporated in the film during plasma exposure. The plasma exposure lowered the film

resistivity from 2 x 104 to 3000 CIR-cm due to lowering of C content and minimal O

incorporation. The plasma exposure also resulted in the densification of the Ti2N surface. The

only disadvantage of the plasma process was that it changed the film microstructure from

amorphous to weakly polycrystalline as indicated by the Ti2N peak in the XRD spectrum.

TiNx film growth by PAALD using TDMAT with remote N2, N2 + H2 and H2 plaSma has

been investigated.88,89 The use of plasma was aimed at reducing the film C content. Since the

plasma is generated ex situ, the likelihood of surface damage by plasma is also minimized.









Unlike thermal ALD of TiNx using TDMAT and NH3, the PAALD process did show a

characteristic ALD temperature window of 200 300 oC where the film growth rate did not

change with deposition temperature. Plasma also resulted in dense films which produced low

levels of O impurities (ca. 10 at. %). The C content of the film was also < 10 at. %. The film

resistivity was between 300 and 500 CIa-cm. Film resistivity was lowest for films deposited

using the N2 plaSma process, followed by those deposited using H2 + N2 plaSma. The highest

resistivity was observed for films deposited with H2 plaSma. The C impurities also showed a

similar trend for different plasma exposures with the lowest C content observed for films

deposited with N2 plaSma. While thermal ALD affords 100% conformal coverage in high aspect

ratio features, the step coverage for PAALD process was 95 % in 1:10 aspect ratio feature.

Diffusion barrier testing of 19 nm TiNx films deposited with different plasma exposures showed

that copper was able to diffuse through the film after annealing at 500 OC as witnessed by the

emergence of Cu3Si peaks in XRD spectra. The failure temperature of barrier film was same

irrespective of whether N2, H2 + N2 or N2 plaSma was used for film deposition.89 Even though

the PAALD film was able to decrease the film resistivity as compared to thermal ALD, the

copper diffusion barrier performance of the film was not satisfactory.

2.6 Tantalum Nitride as Diffusion Barrier

2.6.1 Chemical Vapor Deposition of Tantalum Nitride

CVD of TaNx has employed both inorganic and metal organic precursors as the Ta source

(Table 2-5). The three main inorganic precursors that have been employed to deposit TaNx thin

films are TaClS, TaFS and TaBrS. TalS has not been employed for CVD because it requires very

high temperature to obtain sufficient vapor pressure for delivery in a CVD reactor.90 TaClS, TaFS

and TaBrS are polycrystalline solids at room temperature with sufficient vapor pressure at









elevated temperature for delivery in a CVD reactor. The N sources employed with these Ta

precursors have been either N\H3 al0He Or N\H3 with H2

Chemical vapor deposition of TaNx from TaClS and NH3 at the low deposition temperature

of 350 oC is not suitable for diffusion barrier application because the films had very high

resistivity (> 10,000 CIR-cm) and high Cl contamination (4.5 at. %).90 Since the Ta-Cl bond is

very strong, higher energy is required to cleave the bond. When TaNx films are deposited using

TaFS and NH3 at 350 oC, films have a relatively low resistivity of 1650 CIR-cm. However,

considerable F contamination (4.0 at. % F) was observed and the step coverage of the film was

poor (75%).90 The poor step coverage for TaNx films deposited by TaFS can be explained by

considering the high growth rate of deposition. Films deposited with TaFS had a growth rate of

1.5 nm/sec, which is high for a typical CVD process.90 TaNx deposition from thermal CVD

using TaCl4 and TaFS has not been pursued further, possibly because of the high temperature

required to deposit TaNx films with low resistivity and acceptable halogen contamination.

TaBrS has been studied as a precursor for CVD of TaNx using H2 and NH3 CO-reactants. It

has been argued that since Br is a larger atom than Cl and F, impurity incorporated in the film

would have lower diffusivity in TaNx film and the impurity would not influence the diffusion

barrier property of the film. Moreover, since the bond energy of Ta-Br in TaBrS is lower than

that for Ta-Cl in TaClS, lower deposition temperature would be required for the TaBrS and NH3

reaction. Chen et al.91 reported deposition of TaNx films between 350 and 500 oC using TaBrS,

H2 and NH3. The films were N-rich with N:Ta ratio varying between 1.75:1 and 1.87:1 for

deposition between 350 and 500 oC, indicating that the N:Ta ratio was not very sensitive to the

reaction temperature. The N-rich film deposited with TaBrS indicates that film is possibly Ta3NS

and not TaN. The O and C contamination in the film was below the detection limit of AES.









Films contained Br as an impurity and the Br content in the fi1m decreased with increase in

deposition temperature. The highest Br content of 5 at. % was observed for deposition at 350 oC

while the Br content in film was below the detection limit of AES for deposition at 500 OC. The

fi1m crystallinity was strongly dependent on deposition temperature. Films deposited at 350 and

425 OC were amorphous whereas films deposited at 500 OC were polycrystalline with the peak

positions in XRD spectrum indicating the presence of tetragonal Ta3N5 and hexagonal TaN

phases. Film resistivity also showed a strong dependence on deposition temperature with the

lowest resistivity of 5040C1R-cm observed at 5000C. The resistivity of the fi1ms decreased with

increase in deposition temperature. To ascertain the diffusion barrier efficacy of TaNx fi1ms,

Kaloyeros et al.92 performed extensive barrier testing on TaNx films deposited at 425 oC. The

Cu/TaNx/Si stack was annealed between 450 and 650 oC to determine the temperature at which

barrier film failed. From post-anneal XRD analysis, it was confirmed that the barrier film was

able to prevent bulk diffusion of copper up to 650 oC annealing as no Cu3Si peaks were observed

in the XRD spectrum. X-ray photoelectron spectroscopy depth profile also confirmed that

copper had not diffused through the barrier film. However, when the Seeco etch pit test was

performed to identify trace copper diffusion through the barrier, films annealed at 600 and 650

oC showed etch pits, suggesting that a trace amount of copper had actually diffused through the

barrier film. To compare diffusion barrier efficacy of films deposited by PVD and CVD, the

authors also deposited PVD TaNx films with the same N:Ta ratio. TaNx films deposited by PVD

did not fail the Seeco etch pit test after annealing at 650 oC. The difference in diffusion barrier

efficacy of CVD TaNx and PVD TaNx was attributed to the polycrystalline phase difference

between CVD and PVD films. The PVD TaNx films exhibited hexagonal TaN phase, which has

a N:Ta ratio of 1:1. Since the PVD films had a 1.78:1 N:Ta ratio, the excess N in PVD films









might 'stuff' the grain boundaries thereby preventing copper diffusion. For CVD films, the

predominant phase is believed to be Ta3NS. The emergence of Ta3NS peaks in XRD spectra

when CVD TaNx film is annealed at 650 oC further supports this assumption.92 In Ta3N5, the

N:Ta ratio is 1.67:1, which means that there is very little excess N to 'stuff' the grain boundaries

in CVD TaNx films. This difference is believed to be responsible for the difference in diffusion

barrier efficacy of CVD and PVD TaNx films with similar N:Ta ratio. Overall, the primary

concerns for TaNx films deposited from TaBrS are their higher resistivity at acceptable deposition

temperature, lower recrystallization temperature and deposition of Ta3NS phase instead of the

desired TaN phase. A possible approach that might result in lowering recrystallization

temperature, decrease in resistivity and also possibly disrupt the growth of Ta3NS would be to

grow a ternary compound such as TaNxCv using TaBrS, NH3 and a C source. The ternary

compound could lower the recrystallization temperature and C in the film could disrupt the

growth of Ta3NS phase.

PECVD of TaNx has been attempted with TaCls90'93, TaF593 and TaBr5.9394 Hillman et al.90

used N2 plaSma to deposit TaNx thin films. The films had a low halogen contamination with the

lowest halogen content of I at. % observed for film deposited with TaCls. The resistivity of film

deposited with TaCls, TaFS and TaBrS was 395, 615 and 710 CIR-cm, respectively for deposition

at 350 oC. The film composition and texture were not reported. The diffusion barrier testing of

films deposited by these three precursors was done by depositing 20 nm barrier film and

annealing the Cu (160 nm)/ TaNx (20 nm)/ Si stack at 550 oC for 30 min in N2 CHVITOnment.

Films deposited from TaClS, TaBrS and TaFS failed after anneal and copper in silicon was

detected by SIMS depth profiling. Hence, PECVD process using halide precursors was not able

to deposit good quality diffusion barrier films.









Metal organic sources have also been explored as Ta source for CVD of TaNx. The metal

organic sources that have been explored so far include tert-butylimido tris(diethylamido)

tantalum (TBTDET), pentakis(dimethylamido) tantalum (PDMAT), pentakis(diethylamido)

tantalum (PDEAT), ethylimidoethyl(C,N) tris(diethylimido) tantalum, and

tertiaryamylimidotris(dimethylamido)tantau (TAIMATA). Tsai et al.95-97 have extensively

studied deposition of CVD TaNx films from TBTDET. The precursor was used as a single-

source precursor between 450 650 oC deposition temperature. Films deposited were

polycrystalline with the N:Ta ratio close to 1. The C and O impurities in the film were less than

10 and 5 at. %, respectively at 600 OC deposition temperature. X-ray photoelectron spectroscopy

analysis indicated that C present in the film was in carbide phase. The resistivity of film

decreased from 10,000 CIR-cm at 500 OC deposition temperature to 600 CIR-cm at 650 oC

deposition temperature. Diffusion barrier testing on 120 nm TaNx films deposited at 650 oC

revealed that trace amount of Cu could diffuse through the barrier film after the Cu/TaNx/Si (p~n

diode) structure was annealed at 550 oC and bulk copper diffusion occurred at 600 OC as

evidenced by the formation of Cu3Si.97 The authors also deposited 120 nm PVD TaNx film to

compare the diffusion barrier performance of CVD TaNx films vis-a-vis PVD TaNx films. The

PVD barrier film showed trace copper diffusion at 600 OC and bulk copper diffusion at 650 oC.

The authors argued that the PVD TaNx showed better diffusion barrier performance than CVD

TaNx primarily because PVD TaNx films had a preferred orientation of (111), which has a higher

thermal stability than the (200) orientation observed for CVD TaNx films. However, the film

texture should not play such a significant role in diffusion barrier performance because the CVD

films are polycrystalline and show both (111) and (200) reflections in XRD spectra.95 A more

plausible explanation for poorer barrier performance of CVD films is the columnar









microstructure of the films deposited by CVD at higher temperature. Overall, CVD TaNx films

deposited using TBTDET without any co-reactant had several limitations including high

deposition temperature, polycrystalline microstructure with columnar grains, high resistivity and

poor barrier performance. The use ofNH3 or H2 CO-reactants might decrease the deposition

temperature and improve film properties of TaNx, but such an approach has not been explored

yet.

Fix et al.98 USed PDMAT with NH3 to deposit TaNx films between 200 and 400 oC

deposition temperature. The films deposited with PDMAT had amorphous microstructure.

Charging effects were observed when TaNx films were observed under electron microscope

suggesting that the films were non-conducting. The films were N-rich with a N:Ta ratio of 1.7:. i

The film had significant H content as measured by hydrogen forward recoil spectroscopy. The O

and C contamination in the film was less than the detection limit ofRBS (typically 3 4 at. %).

The resistivity of the film was greater than 106 CIR-cm and is indicative of the growth of the non-

conducting Ta3N5 phase. The growth of highly resistive Ta3N5 makes the PDMAT precursor

unsuitable for diffusion barrier deposition.

PDEAT has been used with and without ammonia to deposit TaNx films. When PDEAT

was used without NH3, the films deposited between 275 and 400 oC were Ta-rich films with

significant C content (> 20 at. %) indicating that the films were actually TaNxCv.99 X-ray

diffraction spectra indicated that the films deposited above 275 OC were polycrystalline with

peak positions corresponding to TaN(111) phase. The films deposited at low deposition

temperature were very resistive (p > 104 CIR-cm) and the lowest resistivity of 6000 CIR-cm was

observed for film deposited at 400 OC. Cho et al.100 investigated the use of NH3 aS a CO-reactant

to deposit TaNx films. The film deposited at 325 OC was polycrystalline with TaN(111) as the









dominant phase. The addition of NH3 dramatically increased the film N content and the N:Ta

ratio for films deposited with NH3 WaS 1.75:1. The addition ofNH3 alSO decreased the C content

of the film to below 10 at. %. The Seeco etch pit test on 50 nm thick barrier film showed that the

barrier film was able to prevent copper diffusion after annealing at 600 OC. The comparison

between films deposited with and without NH3 indicated that films deposited with NH3 WeTO

better at preventing copper diffusion. It was argued that finer grain structure together with dense

microstructure of films deposited with NH3 WaS responsible for enhancing the diffusion barrier

quality. The maj or drawbacks of films deposited with PDEAT and NH3 were the high resistivity

(6000 CIR-cm) and poor conformality of the films.

When a mixture of EtN= Ta[NEt2 3 and Ta[N(Et2)l 5WaS used with H12 to deposit thin films

between 375 and 500 oC, the resulting films were C-rich and the N content of the film was less

than 1 at. %.101 However, when a mixture of EtN= Ta[NEt2 3 and (Et2N)3Ta[(NEt-CMeH] was

used without any co-reactant to deposit thin films between 500 and 600 oC, the films had a

significant N and C with the Ta:C:N ratio of 1:1:1 at 600 OC.102 The XRD measurement of films

showed peaks corresponding to polycrystalline cubic TaN even though the film had significant C

content. The report did not mention film resistivity and no diffusion barrier testing was done on

the films.

TAIMATA has been used with NH3 to deposit barrier films at 250 oC.103 The report

focused on the barrier efficacy and its integration with SiOC:H dielectric. Film characterization

was not discussed in detail. Bias temperature stress testing at 0.75 MV/cm and 350 oC on 10 and

15 A+ thick barrier film showed that the TaNx film was able to prevent copper diffusion. The

precursor shows promise as a diffusion barrier but a detailed analysis of films deposited from

TAIMATA is yet to be reported.









Overall, the biggest challenge for CVD TaNx deposition is the growth of insulating Ta3N5

phase. Since Ta is in the +5 oxidation state in both the precursor and the deposited film, the use

of a stronger reducing agent such as dimethylhydrazine might help in reduction of Ta to the +3

oxidation state to form conducting TaN films.

2.6.2 Atomic Layer Deposition of Tantalum Nitride

The precursors utilized for ALD TaNx have been the same as those used for CVD TaNx

(Table 2-6). TaClS has been used with N\H3 to deposit TaNx thin films by thermal ALD.104,105

The films deposited with the TaClS and NH3 are N-rich and polycrystalline. The XRD spectrum

for films deposited at 400 OC show the presence of the tetragonal Ta3N5 phase. The film growth

rate is 0.22 0.24 A/cycle and the temperature window for the ALD process is 300 500 oC.

The films contain less than 5 at. % Cl and less than 2.5 at. % H impurity. Similar to CVD, ALD

using TaCls and NH3 shows that the Cl impurity in TaNx films decreases with increasing

deposition temperature. The resistivity of the film varies from 0.5 x 106 200 x 106 CIR-cm.

The film is highly resistive because of the presence of the non-conducting Ta3N5 phase. Similar

to CVD, ALD from TaCls also shows that the reducing power of NH3 1S not sufficient to reduce

Ta from the +5 oxidation state in TaCls to the + 3 oxidation state in TaN.

To aid the reduction of Ta, Zn has been utilized as an additional reducing agent in the ALD

process for deposition using TaCls.1os When a Zn pulse is used between the TaClS and NH3

pulses in the ALD process, the film growth rate is 0.2 A/cycle at 400 oC and 0.15 A/lcycle at 500

oC. The characteristic temperature window of ALD could not be established because the growth

rate was not constant between 400 and 500 oC. The films deposited were cubic TaN and

impurity content in the film was lower than that for films deposited without a Zn pulse. The

impurity content of film was 0. 1 4.0 at. % Cl, 3.0 4.0 at. % O, less than 0.5 at. % H and less

than 0.5 at. % Zn. The films were quite conductive and the lowest resistivity of the film was 900









CIR-cm. The main challenges for deposition with Zn as a reducing agent are the polycrystalline

films deposited with the ALD process, incorporation of electrically active Zn impurity in the

films and higher deposition temperature. As discussed earlier, Zn is an electrically active

impurity and its diffusion in the device can cause severe degradation of device performance.

Deposition below 400 oC is not possible with Zn because Zn has a low volatility and its delivery

below 400 oC is not possible.

There are a number of other reducing agents that have been used with and without NH3 to

deposit TaNx from TaClS and TaBrS. These reducing agents include DmHyso, tert-butyl amine

(with and without NH3) and allylamine (with and without NH3).106 When DmHy was used as a

reducing agent, no significant improvement in film properties was achieved as compared to NH3

based ALD. The film deposited from TaClS and DmHy was insulating Ta3N5 and the resistivity

of the film was too high to be measured by the standard four point probe technique.

tert-Butyl amine has been used with and without NH3 to deposit TaNx films from TaClS

and TaBrS.106 When tert-butyl amine is used without NH3, deposition with TaCls has a higher

growth rate and yields films with lower resistivity as compared to TaBrS. Films deposited with

TaClS are amorphous at 350 oC deposition temperature but polycrystalline TaN films are

deposited above 400 oC. The films with TaCls and tert-butyl amine show Cl impurity between 3

and 8 at. %. Film impurity content and resistivity decreases with increase in deposition

temperature. Similar to CVD, low temperature ALD utilizing halide precursors gives highly

resistive films with high impurity content. Good quality films with resistivity below 2700 CIR-

cm are obtained at 500 OC deposition temperature but such high temperature is not suitable for

barrier film deposition. To improve the film quality, NH3 is USed along with tert-butyl amine as









a reducing agent. The process does show improvement with respect to impurity content in the

film and resistivity. However, amorphous film deposition could not be obtained for this process.

When allylamine is used with and without NH3 to deposit TaNx films from TaClS, the films

deposited were polycrystalline TaN.106 The films were highly resistive with lowest resistivity of

18000 CIR-cm. Thus, the process with allylamine did not show any significant improvement

over the tert-butyl amine process.

Plasma assisted ALD of TaNx films using TaClS and H2 + N2 plaSma has been reported.l07

Depending on the partial pressure ofN2 in plaSma, the film N content could be varied to deposit

Ta-rich Ta2N films, stoichiometric TaN films or N-rich Ta3N5 films. The growth rate increased

with deposition temperature indicating that no characteristic ALD process window was observed

for PAALD process. The TaN films deposited were quite conductive with resistivity varying

between 350 and 400 CIR-cm. Films had a high H impurity of about 11 at. % and low Cl content

of less than 3 at. %. Similar to thermal ALD, PAALD also showed decrease in Cl impurity with

increase in deposition temperature. Overall, PAALD was able to deposit good quality films that

might be suitable for the diffusion barrier application. However, the process could not yield

amorphous films.

Metal organic precursors such as TBTDET, TDMAT and TAIMATA have been used to

deposit TaNx films by ALD. Chemical vapor deposition using these precursors deposited films

that had high C and O impurities and high resistivity. The ALD process, if used with the same

CVD precursors, is expected to yield films with low C content and dense microstructure at lower

deposition temperature.

When TBTDET is used with NH3 to deposit TaNx films in ALD mode, amorphous TaNx

films are deposited between 150 and 300 oC. The films deposited are Ta-rich and the N:Ta ratio









is 0.73:1. The films contained low C contamination (< 5 at. %) but had a higher level of O

impurity (ca. 17 at. %). The films were highly resistive (> 106 CIR-cm at 260 oC) despite high Ta

content in the film. High resistivity could possibly result from the formation of Ta3N5, high O

contamination and/or low density of the film (3.6 g/cc). The resistivity of the film decreases to

15000 CIR-cm when the deposition temperature is increased to 450 oC.los Choi et al.109 reported

that the temperature window for the ALD process is between 200 and 250 oC, but the growth

rate versus temperature curve shows that no saturation region can be conclusively determined.

The growth rates of 1.1 and 0.4 A/cycle have been reported for ALD using TBTDET and

NH3-109,110 If deposition is done in the ALD process window and the growth occurs through the

ALD mode, the growth rate per cycle should be independent of all other reactor conditions such

as reactor pressure. The discrepancy in the growth rates reported by Choi et al. 109 and Park et

al.110 suggests that the film growth from TBTDET might actually not be in 'ALD mode'.

Overall, TBTDET is not a suitable precursor for diffusion barrier deposition because of the

deposition of highly insulating films.

PAALD of TaNx films from TBTDET and H radical has been reported.no0 This ALD

process is unique in that it employs TBTDET as a single source precursor as H radicals are used

for reduction of TBTDET. The intent behind using H radicals is that all the N atoms attached to

Ta atoms via single bonds are cleaved by H radical while the N atom attached to Ta via a double

bond survives the H radical exposure, leading to the formation of stoichiometric TaN film. The

PAALD process showed a growth rate of 0.8 A/cycle. The resulting TaNx films are Ta-rich with

N:Ta ratio of 3:5. The films deposited had high C content (20 at. %) and it could be argued that

the films were actually TaNxCv instead of TaNx. The films were polycrystalline and had low

resistivity of 1000 CIR-cm. The films were considerably denser than those deposited with









thermal ALD. Overall, the plasma process improved the film property of films deposited using

TBTDET. The process, however, deposited polycrystalline films. No diffusion barrier testing

was reported for films deposited via this process.

Thermal ALD from PDMAT and NH3 has been used to deposit TaNx films.111-114 The

process shows that the growth rate is constant at 0.3 A/cycle between 225 and 300 oC, indicating

that the growth in this temperature window is truly in ALD mode.114 The film composition and

microstructure strongly depends on the substrate used for deposition.113 The films deposited on

Si, SiO2 and Cu are N-rich, with N:Ta ratio of 2:1 for deposition at 275 00.111,113 However,

when the same process is used to deposit TaNx films on Ta, the films are Ta-rich with N:Ta ratio

of 1:2. The films have a high level ofH impurity that can be removed by annealing the film at

600 oC for 1 hr. The C and O impurity in the film is 2 and 5 at. %, respectively. Even though

the films have high N content, the microstructure analysis indicates the presence of cubic TaN

phase. The excess N is believed to reside in Ta vacancies in the Ta sublattice.ll The films

deposited are nanocrystalline with average grain size of 4 nm. 10 nm thick TaNx film was able

to prevent Cu diffusion after annealing at 750 oC for 30 min in vacuum. The biggest drawback

of the film was its extremely high resistivity, which could not be measured by standard four

point probe technique. Overall, the TaNx films deposited with PDMAT and NH3 in ALD mode

not suitable for barrier application primarily because of its extremely high resistivity.

To deposit more conductive films using PDMAT, PAALD has been investigated. 112,114,115

Kim et al.11 investigated the effect of plasma on the deposition of TaNx films. When H2 plaSma

is used, the film growth rate saturates between 225 and 300 oC, indicating that ALD growth

mode is achieved with PDMAT and H2 plaSma. The deposited film is Ta-rich, with N:Ta ratio

between 0.7 and 0.8. The film C and O content is ca. 15 and 15 at. %, respectively. The films









are conductive with resistivity of 350 CIR-cm observed for deposition at 250 oC. The

microstructure analysis by TEM reveals that the film contains 2 -3 nm cubic TaN crystallites in

an amorphous matrix. Diffusion barrier testing of the film shows that 3 nm TaNx film is able to

prevent copper diffusion after annealing at 750 oC in vacuum. This process could be a potential

candidate for deposition of diffusion barrier film. When N2 + H2 plaSma is used with PDMAT,

the film N:Ta ratio varies between 1.2 and 1.6, depending on the partial pressure of N2 in the

plasma. A sharp decrease in C and O content is also observed, with the impurity content

decreasing with increasing N2 partial pressure. But the additional N in the film results in an

increase in film resistivity and the lowest resistivity of ca. 800 CIR-cm was observed at low N2

partial pressure. Diffusion barrier testing for TaNx film shows that 0.6 nm barrier film is able to

prevent copper diffusion after annealing at 800 OC. From the diffusion barrier standpoint, N2

H2 plaSma yields better films than films deposited with H2 plaSma. Overall, PAALD from

PDMAT provides various options to modify film composition and further investigation on

integration of TaNx films could provide additional information on suitability of the film for

diffusion barrier application. The effect of plasma exposure on underlying layers along with

conformality of the film also needs to be investigated.

Metal organic precursor TAIMATA has also been investigated with NH3 CO-reactant for

deposition of TaNx thin films by ALD.109 Films deposited by thermal ALD between 150 and

175 OC exhibit a growth rate that is constant at 0.2 A/cycle implying that the growth is self-

limiting in this temperature window. The films deposited are N-rich with N:Ta ratio of 1.31.

The films had minimal C contamination (< 5 at. %) but the O contamination was high (ca. 20 at.

%). The XRD spectra indicate the presence of the insulating Ta3N5 phase. The films deposited

with thermal ALD had extremely high resistivity and hence are not suitable for diffusion barrier









application. When H2 plaSma is used instead of NH3 to deposit TaNx films by PAALD,

conductive films with low resistivity of 366 CIR-cm were deposited for growth at 250 oC.116 The

low resistivity of PAALD TaNx films is probably because of a more favorable N:Ta ratio of 0.71

and the absence of insulating Ta3N5 phase. The films also had a high C content of about 15 at.

%. The growth rate per cycle is 1.2 A/cycle. Films deposited by PAALD are polycrystalline

with the XRD peak position indicating the presence of cubic TaN phase. The conformality of

the films deposited with PAALD was 94% achieved in a 10:1 aspect ratio 0.25 Clm via. The

maj or drawback of PAALD using TAIMATA is that the process could not deposit amorphous

TaNx films.

To summarize, both inorganic and metal-organic precursors have been used to deposit

TaNx films by ALD. The biggest challenge for inorganic precursors such as TaClS is the

formation of the insulating Ta3N5 phase instead of the conductive TaN phase. Moreover, the

films deposited with halide precursors contain halogen impurities and the use of stronger

reducing agents or plasma could not completely eliminate the halogen incorporation in the film

for low deposition temperature. The use of metal-organic precursors was also more likely to

deposit Ta3NS phase at low temperature and the use of plasma was essential to obtain conducting

TaNx thin films with a favorable N:Ta ratio. Further investigation needs to be done on PAALD

to achieve the deposition of TaNx films on different substrates and determine if plasma has any

adverse effects on the substrate.

2.7 Tungsten Nitride as Diffusion Barrier

2.7.1 Chemical Vapor Deposition of Tungsten Nitride

Table 2-7 shows the precursors that have been used to deposit WNx thin films by CVD.

WCl6 has been used with NH3 and H2 to deposit WNx thin films for application in catalysis.""

However, the lowest deposition temperature for this precursor is 500 oC, so this precursor has









not been explored for diffusion barrier application. Another inorganic precursor, WF6, has been

explored for deposition of WNx films. This precursor, when used with NH3 aS CO-reactant,

results in the formation of WF6:NH3 adduct even at temperature as high as 600 oC.119 The films

are highly resistive and the lowest resistivity of 3000 CIR-cm is observed for films deposited at

350 oC.120 To prevent the adduct formation, H2 1S used as an additional reducing agent. The

resulting process deposits polycrystalline WNx films and the lowest deposition temperature is

450 oC. Because ofF impurity in the films, relatively high deposition temperature and high film

resistivity, this process also has not been further investigated for diffusion barrier film

deposition. A modification to the process is the use of SiH4 as a reducing agent.120 The lowest

deposition temperature for this process is 385 oC. The resulting films are W-rich and the N:W

ratio decreases with an increase in SiH4 partial pressure. X-ray diffraction measurement

indicates that the films are polycrystalline and the peaks in XRD spectra indicate the presence of

P-W2N. Addition of SiH4 aS CO-reactant also decreases the film resistivity to below 600 CIR-cm

and film resistivity decreases with an increase in SiH4 partial pressure. Diffusion barrier testing

revealed that 6 nm thick barrier film can prevent Cu diffusion after annealing at 450 oC for 4

hours. However, the study by Gonohe found that adhesion of the WNx film to low-k dielectric

film was extremely poor.120 X-ray photoelectron spectroscopy analysis showed that the presence

of F at the WNx/ low-k interface caused the delamination of the WNx film. The presence of F at

the interface was due to the reaction of WF6 with the dielectric surface. Chemical pretreatment

of the dielectric film prevented the reaction of WF6 with the dielectric film and no F impurity

was detected at the barrier-dielectric interface. Details of the chemical pretreatment process

were not discussed in the paper.









PECVD using WF6 as tungsten source has been used to deposit WNx thin films. NH3

plasma, NH3 + H2 plaSma and N2 + H2 plaSma have been used as the N source. When deposition

is done with NH3 plaSmal21-12, the resulting WNx films are N-rich. The film composition can be

varied by changing the NH3:WF6 feed ratio, deposition temperature and plasma power. X-ray

diffraction measurement reveals that the films are polycrystalline with peak positions indicating

the presence of P-W2N. Gas phase reaction between reactants can cause particle formation

during deposition and optimization of the NH3:WF6 flow ratio is necessary to avoid this problem.

Films can be deposited at temperature as low as 250 oC, however the films have a high F

contamination at low deposition temperature. The lowest F contamination of 0. 1 at. % is

obtained for deposition at 625 OC.121 The maj or problem of using this chemistry to deposit WNx

thin film is the formation of an interfacial W layer about 5 nm thick due to the high reactivity of

WF6 and Si. This can also result in etching of the substrate.122 On SiO2 surface, a WO3 layer is

first formed prior to the deposition of WNx film. Another issue with the PECVD process is the

poor conformality. Film conformality of only 33% is obtained for deposition in a 0. 14 Clm 9: 1

aspect ratio feature.123 Hence this process is not suitable for diffusion barrier deposition.

PECVD of WNx using WF6 has also been done using both NH3 and H2 plaSma instead of

just NH3 plaSma as discussed previously.124-130 The advantage of PECVD is that by altering the

NH3:WF6 ratio and the plasma power, the resulting WNx films could be deposited as

stoichiometric W2N, W-rich WNx or N-rich WNx.130 Films can be deposited at temperature as

low as 150 oC.129 An increase in the NH3:WF6 ratio results in a decrease in growth rate, increase

in crystallinity and better film adhesion to the substrate. The films have low resistivity and the

resistivity of the film depends, among other factors, on the film N content and deposition

temperature. Due to higher F incorporation at low deposition temperature, the film resistivity









decreases with increase in deposition temperature. Film resistivity as low as 90 CIR-cm has been

reported for WNx fi1ms deposited by this process.125 Particle formation during deposition has

also been reported and the NH3:WF6 flow ratio has to be optimized to avoid this problem.129 The

main drawback of this process is that pitting of the Si surface is observed due to the reaction of

WF6 with the underlying Si substrate.128

WF6 has been used with N2 + H2 plaSma to deposit WNx thin fi1ms.124-130 N2 1S used as the

N source to prevent particle problems associated with the use of NH3. By changing the plasma

power, N2 H2 ratio and deposition temperature, fi1ms with varying N:W ratio can be deposited by

PECVD. As compared to the NH3 plaSma process, the N2 plaSma process requires higher

deposition temperature to deposit WNx fi1ms. The film resistivity depends strongly on the N:W

ratio and resistivity for stoichiometric W2N and W-rich WNx 61ms varies between 200 and 400

CIR-cm.131 The stoichiometric W2N and W-rich WNx 61ms are amorphous while N-rich WNx

fi1ms are polycrystalline. Similar to the NH3 plaSma process, the N2 plaSma process also resulted

in the presence of F at the barrier substrate interface. The F contamination at the interface was

more severe for SiO2 substrate than Si substrate.32 The F contamination resulted in poor

adhesion of the barrier film with Si and Sio2. The use of N2 plaSma could not prevent the

deposition of highly undesirable F impurity at the barrier substrate interface.

The three main metal organic precursors that have been used to deposit WNx thin films are

bi s(tert- butylimi do)bi s(tert-butyl ami do) tungsten (TB TBW), tungsten hexacarb onyl and

pentacarbonyl tungsten(1 -methylbutylisonitrile). Depositions with TBTBW precursor have been

attempted using inert gas argon and co-reactant NH3-31,132,133 Films deposited with NH3 have a

lower C content and higher N content as compared to those deposited without NH3-132 The

lowest deposition temperature for this precursor is 450 oC. When the deposition is done with Ar









carrier gas at 600 oC, the film contains ca. 70 at. % W, 10 15 at. % N, 10 15 at. % C and 5 at.

% O as measured by AES. The apparent activation energy for film growth in the kinetically

controlled growth regime is 0.9 eV.31 Films were polycrystalline and the XRD spectra indicated

the presence of P-W2N in the film.133 The film was highly resistive at low deposition

temperature and the resistivity of film decreased with increase in deposition temperature. The

lowest resistivity of film was 620 CIR-cm for deposition at 600 OC. The possible reaction

pathways for this precursor have also been investigated by Crane and coworkers.134 Because of

the relatively high deposition temperature, high resistivity of the film and polycrystalline

microstructure, TBTBW is not a good precursor candidate for deposition of diffusion barrier thin

films.

Tungsten hexacarbonyl has been successfully used to deposit WNx thin films with NH3 aS

co-reactant.135-137 Films can be deposited at temperature as low as 200 oC. Kelsey et al.135

reported that the films were amorphous for deposition below 275 OC as measured by XRD and

TEM. However, Lee et al.136 reported that the films deposited at 250 oC were weakly

polycrystalline with the peak position indicating the presence of P-W2N. This discrepancy could

result from the different deposition conditions used for the deposition of WNx films. The

apparent activation energy for film growth is 1 eV for growth in the kinetically controlled growth

regime. The N:W ratio for the film was 1:1, however the presence of the W2N polycrystalline

phase in XRD spectra suggests that the films were actually W2N polycrystals embedded in a N-

rich amorphous WNx matrix. The films had C and N impurities of less than 10 at. % and the

impurity content of the film decreased with increase in deposition temperature. Low C impurity

in the film was possibly due to weak coordinate covalent bonds between CO and W in the

precursor. Kelsey et al.135 reported that the films were also highly conductive and had their









lowest resistivity of 123 CIR-cm for deposition at 400 OC, while Lee et al.136 reported film

resistivity of ca. 900 CIR-cm for the same deposition temperature. This discrepancy in resistivity

could be due to the different film thicknesses used by Kelsey et al. and Lee et al. While Kelsey

et al. measured resistivity for relatively thick WNx film (50 nm), Lee et al. measured resistivity

for a much thinner film (15 nm). As discussed earlier, resistivity could increase for thinner films

because of the increase in electron scattering due to surface and grain boundary scattering. The

process deposited highly conformal films with conformality of 90 % observed for deposition of

50 nm thick films in a 0.25 Clm via with an aspect ratio of 4:1.135 To determine the diffusion

barrier effectiveness, a Cu/WNx (15nm)/Si stack was annealed in Ar environment for 1 hour.136

The post-anneal XRD and sheet resistance measurements indicated that the film was able to

prevent copper diffusion after 600 oC annealing. After annealing at 620 oC, the XRD spectra

showed the emergence of Cu3 Si peaks indicating bulk diffusion of copper through the barrier

film. Tungsten carbonyl is hence a good candidate for diffusion barrier deposition. Further tests

are needed to evaluate the effectiveness of the WNx film in preventing trace copper diffusion.

Another precursor, pentacarbonyl tungsten (1 -methylbutylisonitrile), has also been used

with NH3 to deposit WNx thin films.138 The films were stoichiometric W2N with amorphous

microstructure. Detailed film characterization or diffusion barrier testing for this precursor has

not been reported.

2.7.2 Atomic Layer Deposition of Tungsten Nitride

Table 2-8 shows the precursors that have been used to deposit WNx films by ALD. George

and coworkersl39,140 flTSt reported the deposition of WNx thin films by ALD using WF6 and NH3

as precursors. The films were deposited between 327 and 527 oC. Minimum growth

temperature for ALD growth was 327 oC because below this temperature, WFx:NH3 adduct

formation was observed by FTIR due to incomplete reduction of WF6 by NH3. X-ray









photoelectron spectroscopy analysis showed that the films had a N:W ratio of 1:3. The films had

low C and O contaminations of ca. 5 and 3 at. %, respectively. No F contamination was detected

by XPS. The ALD process had a deposition rate of 2.5 A/cycle. The characteristic ALD process

window was observed between 325 and 525 OC.140 X-ray diffraction measurement showed that

the films deposited were polycrystalline with the peak positions indicating the presence of P-

W2N. The average grain size of crystallites calculated from the peak broadening of P-W2N peaks

in XRD spectra was 1 10 A+. The films deposited have high resistivity of 4500 CIR-cm. To

deposit WNx films with lower film resistivity, Kim et al.141 USed diborane (B2H6) as an additional

reducing agent along with NH3. The use of B2H6 resulted in films with significantly lower

resistivity (< 450 CIR-cm). The growth rate of the film was 2.8 A/cycle and the N:W ratio was

0.82. The XRD measurement revealed that the film contained a two phase mixture of P-W2N

and 6-WN. The density of the film was 15 g/cc, which is close to the bulk density of both P-

W2N and 8-WN. X-ray photoelectron spectroscopy measurement indicated that in addition to

WNx, the film also contained the oxynitride phase WNxOv. The source of oxygen in the film was

not discussed. No diffusion barrier testing was reported for these films.

As discussed earlier, a maj or drawback of the process using WF6 is that it can react with

the underlying Si substrate to form a thin pure tungsten film before the deposition of WNx. This

could increase the total thickness of the barrier film and is not desirable. To prevent or minimize

the reaction of WF6 with the underlying Si substrate, Sim et al.142,143 USed pulsed plasma

enhanced ALD (PP-ALD). Plasma was used during the NH3 pUlSe only because if plasma is

applied during the WF6 pUlSe, W particles could be generated and subsequently incorporated into

the film. The process was able to deposit WNx film on a Si substrate without the formation of









pure W interface. However, the films also contained F as an impurity though the F content was

not quantified by the authors.

The only metal-organic precursor that has been used to deposit WNx films by ALD is

bis(tert-butylimido)-bis(dimethylamido)tunstn45,144 The precursor, when used with NH3 in

ALD mode, deposits WNx thin films. No ALD process window was observed between the

deposition temperature of 250 and 300 oC. The N:W ratio was 1:1 and films did not contain C

impurity as indicated by AES and XPS measurements. X-ray diffraction and HRTEM

measurements indicated that the films were amorphous. The films were stable after annealing at

700 oC. Subsequent heating at 725 OC resulted in decomposition of film into pure W and N2 gaS.

The growth rate was 1 A/cycle. The films were very smooth with a RMS roughness of 3.3 A+.

The ALD process also demonstrated 100 % step coverage in a 200: 1 aspect ratio feature. WN

film as thin as 1.5 nm was able to prevent copper diffusion after annealing at 600 OC. The only

drawback of this film was its high resistivity (1500 4000 CIR-cm). The resistivity of the film

could be reduced by annealing the film at 700 OC in forming gas, but annealing at such high

temperature is not desirable for diffusion barrier application.

2.8 Tungsten Nitride Carbide as Diffusion Barrier

2.8.1 Atomic Layer Deposition of WNxC,

Numerous studies on ALD WNxC, growth have been reported. Most of the reports of

WNxCy by ALD have used ASM microchemistry's PulsarTM ALD deposition tool. Almost all of

the reports on ALD WNxC, have employed halide chemistry with WF6 as the tungsten source,

NH3 as the nitrogen source and triethyl boron (TEB) as the carbon source. The composition of

films deposited was W:N:C = 55:15:30.145 The concentration of impurities such as O, F and B in

the film was low (< 0.5 at. % O, < 0.5 at. % F and < 0.5 at. % B).146 However, F was detected at

the interface of Si and WNxC,.89 This is undesirable since halogens present at the interface can









severely affect adhesion of the films. In fact, the adhesion energy of WNxCy with SiO2 is 1.5

J/m2 as measured by the four point bend test." The low adhesion strength, however, did not

cause delamination of the film during the CMP process.76 Despite the use of a halide precursor,

the deposition of WT\xCy did not show any pitting on copper and WT\xC, had good adhesion with

copper.145 The WNxC, film was nanocrystalline with the peak position indicating the presence of

WNx and WCx or a single WNxC, phase. The average crystallite size was between 3 and 7

nm.147 Unlike TiN films deposited by ALD, WNxC, films did not exhibit a columnar grain

str-ucture, which is desirable because a columnar structure could provide short diffusion paths for

copper to diffuse. The density of the film was 15.37 g/cc, which is close to the bulk value. The

dense microstructure of WNxCy was believed to be responsible for minimal O diffusion through

the film after exposure to air. The film was stable after annealing at 700 OC. After annealing at

800 oC, a multiphase mixture containing metallic W, WC and WCx emerges as evidenced by the

corresponding peaks in the XRD spectrum. The N from the film is lost as N2 gas after annealing

at 800 oC. While WCx reacts with Si to form W5Si3 at 700 00148, HO Silicide formation is

observed for WNxC, films even after annealing at 800 oC.89 The growth rate of the film was 0.8

- 0.9 A/cycle. The resistivity of the film was 300 400 CIR-cm for films deposited between 300

and 400 oC.

Growth of ALD WNxC, film was sensitive to the surface on which the films were

deposited.149 This is particularly important because in dual damascene structures, the diffusion

barrier film is deposited simultaneously on three different surfaces, (i.e. copper, dielectric (Sio2

or low-k dielectric) and the etch stop layer (SiC or Si3N4)). A transient growth regime is

observed up to film thickness of 5 nm for growth on SiO2 and SiC, however deposition on Si3N4

shows a much shorter transient regime. Surface treatment has been proposed to speed up the









nucleation and growth of ALD films.'so WNxC, film deposited on PECVD SiO2 pretreated with

NH3 plaSma showed complete surface coverage after 40 ALD cycles whereas WNxCy deposited

on untreated PECVD SiO2 required 50 cycles to achieve complete surface coverage.

The WNxC, film was compatible with dielectric films such as SiO2 and SilKTM and etch

stop layers such as SiC and Si3N4-145 However, for ultra low-k films, the precursor diffuses

through the pores of the dielectric. Diffusion of the precursor through both HSQ and MSQ

dielectric films with pore size of4 5 nm has been reported.73 Nitrogen plasma was able to seal

the pores of these dielectric films and prevent the diffusion of the precursor(s) into the dielectric

film. Surface treatment of SilKTM dielectric by 02 inductively coupled plasma (ICP) in an

oxygen rich environment and reactive ion etch (RIE) done in a nitrogen rich environment have

been investigated to seal the pores of the dielectric film. Hydrophilic groups created on the

dielectric surface during surface treatment are expected to enhance ALD growth. N2 RIE was

able to densify the dielectric surface and close the pores of the dielectric film. However, since

the Ol ICP does not have any ion bombardment, the dielectric films remained semi-permeable

after surface treatment. WNxC, films on plasma treated polymer were smoother, more so for

film deposited on an N2 RIE treated surface than film grown on Ol ICP treated surface. Films

deposited by ALD were continuous after 10 nm for untreated polymer, 3.5 nm for the Ol ICP

treated surface and 1.4 2.3 nm for the N2 RIE treated surface.

The thermal stability of ALD WNxCy as measured by XRD showed that the film failed

after annealing at 700 OC as evidenced by the formation of Cu3Si.147 However, the etch pit test,

which can detect trace diffusion of copper through the barrier film, showed that the diffusion of

copper through the WNxC, film occurs after annealing at 600 OC for 30 min.89 Electromigration

resistance of copper deposited on WNxC, has been compared with copper deposited on PVD Ta,









which is the liner material currently used in industry.76 For the electromigration test performed

at 325 OC and 2 mAmp current, copper film deposited on WNxC, has better electromigration

resistance than copper film deposited on PVD Ta. The dominant failure mechanism of copper

deposited on WNxC, in dual damascene structure was voiding in M2 trench starting at the via

and propagating along the capping layer. Bias temperature stress tests done at 300 OC and 0.7

MV/cm bias showed that 2.7 nm WNxC, film failed after 30 min stress, while 5 nm WNxCy

barrier did not fail. These reports confirm that WNxCy can prevent copper diffusion under

normal operating conditions and has good electromigration resistance.

A major integration problem of WNxCy barrier film was its galvanic corrosion by the H202

based slurry used in CMP of copper. The large potential difference between Cu and WNxC, in

H202 CaUSes the corrosion of WNxCy by H202. The corrosion occurs by dissolution of W as

tungsten oxide. To overcome this problem, a CMP slurry based on HNO3 has been proposed.152

The use of HNO3 based slurry can reduce the WNxCy corrosion. Additives such as

monosaccharides or organic acids can be used to slow the excessive loss of copper by HNO3.

2.8.2 Chemical Vapor Deposition of WNxC,

WNxCy is a very promising material for diffusion barrier application as discussed above.

However, the use of WF6 leads to F deposition at the barrier-substrate interface89 leading to poor

adhesion of the barrier film to the substrate. Moreover, WF6 has problems related to its handling

and storage. An alternate method for the deposition of WNxCy is thus desirable. Three different

metal organic precursors have been used to deposit WNxCy thin films [Cl4(RCN)W(NPh) (la, R

= CH3 and lb, R = Ph), Cl4(RCN)W(NC'Pr) (2a, R = CH3 and 2b, R = Ph) and

Cl4(RCN)W(NC3H5) (3a, R = CH3 and 3b, R = Ph)].153-157 From the study of films deposited

using these precursors, it was evident that the N-C imido bond strength has a significant effect on

the N content of the deposited film. Films deposited with la,b were W-rich and poor in N









because the strong N-C imido bond results in the cleavage of W-N bond. Films deposited with

2a,b and 3a,b have a higher N content due to relatively weaker N-C bond strength, even though

the N content was below the desirable levels for diffusion barrier application.

Previous studies on WNxCv deposition by MOCVD using la,b, 2a,b and 3a,b have shown

that the films have low N content. As discussed earlier, excess N in diffusion barrier films is

desirable because it improves the diffusion barrier performance by 'stuffing' the grain

boundaries. Another concern for deposition using precursors la,b, 2a,b and 3a,b was that the

lowest temperature at which films could be deposited was 450 oC or higher. It is desirable to

deposit WNxCv films at even lower temperature as low-k dielectrics are thermally unstable above

450 oC. In the present work, further investigation is done to deposit WNxCv thin films with

alternate chemistries to deposit films with higher N content at temperature below 450 oC. To

improve film N content, two different approaches have been used. One is the use of precursors

with higher N:W rati o such as W(N'Pr)Cl3 ['PrNC(NMe2)N'Pr] (4) and (CH3 CN)Cl4W(NNMe2)

(5). The other approach is to use co-reactant NH3 to increase film N content.

Chapter 4 investigates the use of NH3 with 3a,b to increase the film N content. Chapter 5

discusses the results for film deposition using 4 and H2. Chapter 6 discusses the results obtained

from thin film deposition using 4 and NH3 (with and without H2). The results obtained from thin

films deposited with 5 and H12 are discussed in Chapter 7. To facilitate the understanding of the

gas phase reactions of 5, Chapter 8 discusses results from quantum mechanical calculations using

density functional theory. Chapter 9 provides a summary of work reported in this manuscript

and discusses future work that would enhance our understanding of WNxCv thin film deposition

using MOCVD.










Table 2-1. Melting temperature of Ti, Ta, and W and their nitrides and carbides39
Transition Metal Tm ( OC)


Titanium
Tantalum
Tungsten
Titanium Nitride
Tantalum Nitride
Tungsten Nitride
Titanium Carbide
Tantalum Carbide
Tungsten Carbide


1677
2997
3380
2949
3093
< 800 for WN
3067
3983
2776


Table 2-2. Techniques used for characterization of diffusion barrier thin films
Film Property Characterization Technique Acronym(s)
Microstructure X-ray diffraction XRD
Grazing Incidence XRD GIXD, GIXRD
Scanning electron microscopy SEM
Electron diffraction in transmission electron
microscopy TEM


AES
XPS, ESCA
EDS, EDX
SIMS
RBS
ERDA
SEM
TEM


Composition






Film Thickness





Density

Resistivity
Surface roughness




Atomic Bonding
Film Adhesion


Auger electron spectroscopy
X-ray photoelectron spectroscopy
Electron dispersive X-ray spectroscopy
Secondary ion mass spectroscopy
Rutherford back scattering
Elastic recoil detection analysis
Scanning electron microscopy
Transmission election microscopy
Profilometer
Rutherford backscattering
X-ray reflectivity
X-ray reflectivity
Rutherford backscattering
Four point probe
Atomic force microscopy
Optical profilometer
Stylus profilometer
X-ray reflectivity
X-ray photoelectron spectroscopy
Four point bend test, Scotch tape test


RB S
XRR
XRR
RBS

AFM



XRR
XPS










Table 2-3. Precursors used for deposition of TiNx thin films by CVD


PECVD Inorganic TiCl4 H12, N2 plaSma 58, 59
Precursor
Metal-organic tetrakis (diethylamido) H12 or N2 plaSma 72
Precursor titanium



Table 2-4. Precursors used for deposition of TiNx thin films by ALD
Technique Ti N source Reference
source
ALD Inorganic Precursor TiCl4 NH3 73, 74, 77, 83, 86, 172
TiCl4 Zn + NH3 75, 79
TiCl4 dimethylhydrazine 80
TiCl4 tert-butylamine 81
TiCl4 allylamine 81
Til4 NH3 79, 81, 83
Metal-organic TDMAT NH3 86, 87, 167, 168
Precursor
TEMAT NH3 84, 85, 167
PAALD Inorganic Precursor TiCl4 NH3 78
Metal-organic TDMAT remote N2 /H2 88, 89
Precursor plasma


Reference
51-55, 158-
160

56, 57
60-62, 68, 69,
72, 161
68, 69, 162-
164

60, 64-66, 165

67

70, 64, 68,
166
64, 66, 70, 71


Technique
CVD


Ti source
TiCl4


N source
NH3


Inorganic
Precursor


Metal-organic
Precursor


tetrakis (dimethylamido)
titanium (TDMAT)
tetrakis (dimethylamido)
titanium

tetrakis (dimethylamido)
titanium
tetrakis (dimethylamido)
titanium
tetrakis (diethylamido)
titanium (TDEAT)
tetrakis (diethylamido)
titanium


H2 or N2 + POSt-
growth plasma
exposure
NH3


hydrazine

inert gas

NH3




















Metal-organic
Precursor


NH3


NH3

NH3

H2


Table 2-5. Precursors used for deposition of TaNx thin films by CVD


PECVD Inorganic TaClS N2 90, 93
Precursor

TaFS N2 90
TaBrS N2 90
TaBrs N2 + H12 94
TaBrs N2 90


Technique


CVD


Ti source


TaClS


Reference


90, 169


N source


NH3


NH3
NH3
H2 + NH3


Inorganic
Precursor


TaFS
TaBrS
TaBrs
tert-butylimido tris(diethylamido)
tantalum (TBTDET)
pentaki s(dimethyl ami do) tantalum
(PDMAT)
pentaki s(di ethyl ami do) tantalum
(PDEAT)
tert- amylimi dotri s(dim ethyl ami do)
tantalum (TAIMATA)
ethylimi doethyl(C,N)
tris(diethylimido) tantalum


90
90

91, 92
95-97


98, 170


99, 100

103

91










Table 2-6. Precursors used for deposition of TaNx thin films by ALD
Technique Ta source N source Reference
ALD Inorganic TaClS NH3 104, 105
Precursor
TaClS Zn + NH3 105
TaClS 1,1 dimethyl hydrazine 80
TaClS tert-butylamine (+NH3) 106
or allylamine (+ NH3)
TaBrS tert-butylamine (+NH3) 106
Metal- tert-butylimido NH3 108-110
organic tri s(di ethyl ami do)
Precursor tantalum (TBTDET)
pentaki s(dim ethyl ami do) NH3 111-114
tantalum (PDMAT)
tertiaryamylimidotris NH3 109, 116
(dimethylamido)
tantalum (TAIMATA)
PAALD Inorganic TaClS N2 H2 plaSma 107
Precursor
Metal- tert-butylimido H radicals 110, 171
organic tri s(di ethyl ami do)
Precursor tantalum (TBTDET)
pentaki s(dimethyl ami do) H2/N2 plaSma 112, 114, 115
tantalum (PDMAT)
tertiaryamylimidotris H12 plaSma 116
(dimethylamido)
tantalum (TAIMATA)





















tungsten hexacarbonyl NH3 135-137

tungsten pentacarbonyl NH3 138
1 -methylbutyli sonitrile

PECVD Inorganic WF6 NH3 121-123
plasma
Precursor WF6 H12 + NH3 124-130
plasma
WF6 H12 + N2 32, 131
plasma


Precursor bis(dimethylamido) tungsten

Inorganic WF6 NH3 plaSma 142, 143
Precursor


Table 2-7. Precursors used for deposition of WNx thin films by CVD
Technique W source N source Reference
CVD Inorganic WCl6 N\H3 + H12 117, 118
Precursor WF6 N\H3 + H12 119
WF6 NH3 + 120
SiH4 + H12
Metal organic tert-butylimino tert- NH3 or Ar 31, 132-134


Precursor


butylamido tungsten


Table 2-8. Precursors used for deposition of WNx thin film by ALD
Technique W source N source Reference
ALD Inorganic WF6 NH3 139, 140
Precursor WF6 N\H3 and B2H6 141
PAALD Metal organic bis(tert-butyido H 5 4










Reaction limited
growth regime


Mass tra nsfer l limited
growth regime


1/T
Figure 2-1. Arrhenius plot for CVD showing mass transfer limited and reaction limited growth
regimes. G is the film growth rate and T is the deposition temperature in K













Precursor
decom position


Process
Window


Incomplete
satu ration/
reaction


Deposition Tempe rature
Figure 2-2. Characteristic ALD growth rate vs. T curve showing different growth regimes











Incomplete
sat urat io n


Exposure Time
Figure 2-3. Characteristic ALD plot showing the exposure time required to achieve self-limiting
growth


Self-limiting growth









CHAPTER 3
EXPERIMENTAL TECHNIQUES FOR THIN FILM DEPOSITION AND
CHARAC TERIZATION

3.1 Description of the CVD Reactor

All the depositions of WNxCv were done in a custom CVD reactor (Figure 3-1). Metal

organic precursor used for CVD deposition was dissolved in benzonitrile solvent and loaded into

a syringe. The solution was delivered into a nebulizer at a constant flow rate of 4 mL/hr. The

nebulizer contains a quartz plate that converts the precursor-solvent mixture into aerosol and the

aerosol is carried to the reactor via carrier gas (H2, H2 + NH3 or N2 + NH3). The aerosol/ carrier

gas mixture was delivered to the reactor through heated delivery tube. The delivery tube was

heated to prevent condensation and accumulation of condensates on the tube wall. The substrate

was heated by RF generator and the RF power was adjusted to change the temperature of the

substrate. A detailed description of the system has been reported previously.173

3.2 Modification of the CVD Reactor

There were 4 maj or changes made in the experimental setup:

* Installation of H2 and N2 pUriflerS in the H2 and N2 delivery lines
* Reconfiguration of the gas lines downstream of the reactor
* Modification to the line for delivery of NH3
* Replacement ofMKS mass flow and pressure programmer/ display

3.2.1 Installation of H2 and N2 Purifiers in the H2 and N2 Delivery Lines

The precursors used for CVD deposition are oxygen sensitive. Hence, the CVD reactor

should contain a minimal amount of oxygen during film growth. Ultra high purity gases (H2 and

N2) obtained from Praxair contain a few ppm of oxygen and water vapor. The intent behind

installing purifiers was to reduce the impurity levels of carrier gas from ppm level to ppb level.

Figure 3-2 shows the change in feed line configuration before and after the installation of gas

purifiers. The purifiers were made by Saes Corporation (model # PS11-MC1N for N2










purification and PS11-MC1H for H2 purification). The purifier uses activated catalyst to adsorb

gases such as 02, H20, CO and CO2. The installation of the purifier requires that they be

installed vertically with the gas inlet from the top. The gas lines were purged prior to installation

of purifier since exposure to atmosphere can poison the catalyst in the purifier. The purifier also

contained a particle filter that removes particles with size larger than 0.003 Clm in diameter. The

purifier was connected to the lines by VCR fittings. The purifier is designed to operate at a

maximum flow of 5000 sccm with nominal flow of 500 sccm.

3.2.2 Reconfiguration of the Gas Lines Downstream of the Reactor

One of the maj or operational problems for the CVD system was the failure of the throttle

valve. The throttle valve uses a small disc with an o-ring inside a cylindrical bore to modulate

the conductance through the valve. To control the reactor pressure, a throttle valve controller

adjusts the disc position in the valve based on the difference between pressure input from the

Baratron pressure gauge and the set point for the reactor pressure. The main problem with the

throttle valve was the swelling of the o-ring on the disc due to absorption of benzonitrile solvent.

Since the cold trap, intended to remove condensables such as solvent vapor, was installed

downstream of the throttle valve, the disc was exposed to the benzonitrile solvent. The swelling

of the disc' s o-ring due to benzonitrile absorption resulted in rupture of the o-ring.

Consequently, the throttle valve was not able to control the reactor pressure. The frequency of

throttle valve failure was once every four to six experiments.

To alleviate the problem of throttle valve failure, the downstream of the reactor was

reconfigured (Figure 3-3). A cold trap was installed between the reactor and throttle valve so

that benzonitrile could be condensed and removed from effluent gas to minimize the exposure of

throttle valve o-ring to benzonitrile. As a result of this modification, the frequency of throttle

valve failure was reduced to once every 50 experiments.










The glass cold trap was replaced by a new stainless steel cold trap (Figure 3-4). The

stainless steel cold trap was designed such that its total internal volume was same as that of the

glass cold trap. The glass cold trap had no port for removing the condensate collected at the

bottom of cold trap. The stainless steel cold trap had an inclined bottom to collect the

benzonitrile and other condensables near the purge port. After every experiment, about 1 2 ml

of solvent (primarily benzonitrile) condensed in the cold trap. Before starting new experiment,

the solvent collected in the cold trap was removed by opening the purge valve at the bottom of

the cold trap.

3.2.3 Modification to the Line for Delivery of NH3 to the Reactor

When NH3 WaS used as a co-reactant, the gas was delivered using a secondary NH3 lin

labeled as 'Old NH3 feed port' in Figure 3.5. Films deposited when NH3 WaS used as co-reactant

were quite non-uniform in thickness. The primary reason for this non-uniformity was the

inhomogeneous distribution of NH3 because of the proximity of the NH3 feed port to the

substrate surface. To alleviate this problem, a modification was made in the system near the H12

inlet line at the nebulizer. The line was split to allow for the introduction of NH3 along with the

H2. The new port from which NH3 WaS introduced is labeled 'new NH3 feed port in Figure 3-5.

This modification allowed the NH3 gas to mix properly with the H12 carrier gas prior to the

exposure of the gas mixture on the substrate. Films deposited after this modification had

uniform film thickness as measured by SEM.

3.2.4 Replace MKS Mass Flow and Pressure Programmer Display

The old MKS mass flow and pressure programmer (MKS 147) failed to control the flow

rate of NH3 and had to be replaced. The manufacturing of MKS 147 pressure and flow controller

has been discontinued by MKS Inc. So, a new MKS mass flow and pressure programmer

display (MKS 647B) was installed in the system. The MKS 647B mass flow and pressure










programmer/display has a LCD display while the old MKS 147 had a CRT display. Most of the

other features of MKS 647 B are similar to that of MKS 147. However, the new controller

required rewiring of the transducer input into the controller.

3.3 Operating Procedure for CVD Deposition

The operating procedure for CVD deposition has been described previously. 173 The only

change made to the procedure was that after the growth, the substrates were taken out of the

CVD chamber only after the substrate temperature reached room temperature. The old

procedure allowed for the substrates to be unloaded from the reactor once the substrate

temperature was below 75 oC. The procedure was modified primarily to minimize the diffusion

of oxygen into the deposited film.

3.4 Deposition Parameters

Table 3-1 shows the flow rates of precursor, solvent and carrier gas used for CVD growth.

For a particular set of experiments, all the deposition parameters except substrate temperature

were held constant. Depositions at different growth temperature were done to understand the

effect of substrate temperature on film properties such as growth rate, crystallinity, resistivity,

atomic bonding and efficacy as a diffusion barrier.

3.5 Dimensionless Numbers for CVD Reactor

Dimensionless numbers can provide valuable insight into the effect of various process

parameters such as temperature and pressure on reactor flow pattern. This section discusses the

calculation results for important dimensionless numbers for the CVD Eilm growth for the reactor

and deposition conditions discussed previously. The calculation of dimensionless numbers has

been done by using the thermophysical properties of carrier gas alone unless specified otherwise.









3.5.1 Prandtl Number

The Prandtl number, Pr, is the ratio of momentum diffusivity to thermal diffusivity.

Pr-
CIkC (3-1)

The kinematic viscosity is v, thermal diffusivity is a, viscosity is CI, thermal conductivity is k and

specific heat at constant pressure is c,. Chapman-Enskog theory can be used to estimate the

Newtonian viscosity.174

(MT)O 5
y=(2.669x10- 2~b
~L (3-2)

The molecular weight is M (kg/kgmol), temperature is T (K), collision diameter is a (A+) and

dimensionless collision integral is Op,. Opis a function of dimensionless temperature, kBT/E. For

H2 and N2 carrier gases used for CVD deposition, the value of E/kB are 38.0 and 91.5

respectively. Since the depositions were done between 300 and 750 oC, the value of

dimensionless temperature is greater than 6.0. For dimensionless temperature greater than 6.0,

02, is given by Equation 3-3.'7


n =1.1615(kT 18
e (3-3)

For N2 and H2, the value of a is 3.681 and 2.915 respectively.176 Substituting these values in

Equation 3-2, the value of CI for H2 and N2 carrier gases be calculated.

11H2=4.5864XTO.687 (3-4)


"2 =4.5864xTO.687 (3-5)

The thermal conductivity for a polyatomic gas is given by modified Eucken correlation (Eq. 3-

6).174










k_ P 1.32 cM- +1.77
cLRM cMR :I(3-6)

The Universal gas constant (8314 J/mol-K) is R. Since depositions in CVD reactor were done at

high temperature and low pressure, the ideal gas assumption is valid. For an ideal polyatomic

gas, c, is given by Equation 3-7.

7R
2M (3-7)

Substituting the values of c,, CI, R and M in Equation 3-6, the value of k can be calculated (Eq. 3-

8 and 3-9).


kH2 -4.6742 x10-3 TO6487 (3-8)

k, 2=.8995 x10-3 TO6487 (3-9)
From Equations 3-4, 3-5, 3-7, 3-8 and 3-9, the value calculated for Pr is 0.690 for both H2 and N2

carrier gases. The value of Pr indicates that the thermal boundary layer and the velocity

boundary layer for both H2 and N2 carrier gases are important for thermal diffusion

consideration.

3.5.2 Knudsen Number

Knudsen number (Kn) determines if the continuum approach can be used for the fluid flow

in CVD reactor. For a dilute gas mixture in which species B is the predominant gas, Knudsen

number is given by Equation 3-10.m7

-0 5
Kn=1.758x 10-3 B;M M ost,+,
(3-10)

The characteristic length of the reactor is L, molecular weight of dilute species A (precursor

3a,b, 4 or 5) is MA, molecular weight of species B (H2 or N2) is MB, the collision diameter of

species A is oA and the mole fraction of species B is xB. For the CVD reactor used for film









growth, the diameter of the reactor (0.0597 m) was used as the characteristic length of the

reactor. The present work used 3 different precursors, 3a,b, 4 and Sa,b with N2 and/or H12 Carrief

gases. The approximate values of oA USed for 3a,b, 4 and Sa,b were obtained from X-ray

crystallography. The distance between furthest atoms in the X-ray crystallographic structure of

all three precursor molecules was ca. 10 A+ and this value was used as an approximation of GA. It

should be noted that the X-ray crystallographic structure does not account for the substitution of

acetonitrile ligand by benzonitrile when the solid precursor is dissolved in benzonitrile solvent.

T is calculated by averaging the temperature of the heated impinging jet (50 OC) and the

temperature of the substrate (between 300 and 750 oC). Table 3-2 shows the values of Knudsen

number for precursors 3a,b, 4 and 5 at different deposition temperature. For Kn < 0.01,

continuum approach is valid for analyzing the system. Since for all the different precursors and

carrier gases the Kn number is of the order of 10- continuum approach is valid.

3.5.3 Reynolds Number

The Reynolds number (Re) is the ratio of inertial to viscous force. Reynolds number for a

CVD system is given by Equation 3-11.m7


Re=3.747x105 O5(-1


The cross-sectional area of the reactor is Aes (m2) and the mass flow rate of the carrier gas is m

(kg/s). Reynolds number is calculated using carrier gas only because the molar flow rate of

carrier gas is at least 3 orders of magnitude higher than that of the precursor. Aes for the reactor

is 2.8 x 10-3 m2 and mass flow rate for H12 and N2 are 3.41 x 10-7 kg/s and 4.77 x 10-6 kg/s

respectively. Table 3-3 shows the value of Re for H12 and N2 at different deposition temperature.

Since the value of Re is less than 2000, the flow through the reactor is laminar.









3.5.4 Peclet Number

The Peclet number (Pe) is a dimensionless number related to the rate of advection of flow

to its thermal diffusion. It is calculated by multiplying Reynolds number with Prandtl number.

Forced convective heat transfer can be neglected if Peo.s << 1. Table 3-3 shows the values of Pe

and Peo.s calculated for H2 and N2 carrier gases at different deposition temperature. Since

(Re*Pr~o.s for H2 and N2 is HOt << 1, forced convective heat transfer through the reactor can not

be neglected.

3.5.5 Grashof Number

The Grashof number (Gr) is the ratio of buoyancy force to vi scous force on a fluid. This

dimensionless number is helpful in determining the occurrence of turbulence for free convection.

If the Grashof number is less than 10 no turbulence effects are expected for free convection.

Grashof number for the CVD reactor is given by Equation 3-12.m7


Gr-1.9903 x 104 4Mo ~2\iiT 42(-


The characteristic temperature difference, AT, is calculated by subtracting the impinging j et

temperature from the temperature of the substrate. Table 3-4 summarizes the Grashof number

calculated for H2 and N2 at different deposition temperature. For H2, the value of Gr is less than

105 indicating that turbulence is not expected for free convection. However, Gr for N2 is

significantly higher than 105 suggesting that turbulence is expected for free convection when N2

is used as carrier gas. Experimentally, when N2 is USed as carrier gas, condensates of

benzonitrile are observed in the view port. Since the view port is located above the shower head,

benzonitrile condensation on view port is possible only if longitudinal convective rolls are

present in the reactor. However, when H2 is USed as carrier gas, no such condensation is









observed on the view port. This observation along with the value of Grashof number for N2

confirms that buoyancy effects are important when N2 is USed as a carrier gas.

3.5.6 Rayleigh Number

The Rayleigh number (Ra) is the ratio of thermal flux by free convection to that by

diffusion. Rayleigh number is defined as the product of Grashof number and Prandtl number.

Table 3-4 summarizes the Rayleigh number calculated for H2 and N2 carrier gas for different

deposition temperature. Natural convective heat transfer can be neglected if (Ra*Pr)0.25<1

Since the value of (Ra*Pr)0.25 is HOt << 1 for both H2 and N2 carrier gas throughout the

deposition temperature, natural convective heat transfer cannot be neglected. Longitudinal

convective rolls have been observed in horizontal CVD reactor for Ra > 1708.m7 For both H2

and N2, the value of Ra is greater than 1708 indicating that convective rolls are present in the

reactor. Furthermore, the convective rolls are expected to be more severe for N2 carrier gas

because the Ra value for N2 is two orders of magnitude higher than 1708. Experimentally, the

condensation of benzonitrile on the view port confirms that the problem of recirculation is more

severe for N2 aS COmpared to H2 carrier gas.

3.6 Techniques Used for Thin Film Characterization

3.6.1 X-Ray Diffraction

The fi1m crystallinity was examined by Philips APD 3720, operating from 5 85

26 degrees with Cu Ku radiation. X-rays were generated at 40 kV voltage and 20 mA current.

Monochromatic Cu Ku radiation was obtained by filtering out the Cu KP radiation using a Ni

fi1ter. When the fi1ms were extremely thin (few nm thick), film crystallinity was measured by

GIXD technique using Phillips MRD X'Pert system. Cu Koc radiation, generated at 45 kV and 40

mAmp (1.8 kW), was used for the GIXD analysis. The angle of incidence for measurements was

20 and the step size was 0.020 per step. The peaks obtained from the XRD and GIXD spectra









were compared with reference spectra in the JCPDS database" to determine the crystalline

phase(s) present in the film. Table 3-5 lists the reference peak positions and their relative

intensities for P-W2N, P-W2C, a-W2C, WO3 and Si."7

The broadening of peak in XRD spectra can be used to calculate the grain size of the

crystallites by Scherrer's equation (Eq. 3-13).179

t = 0.9h / (B cos 6) (3-13)

The wavelength of X-rays used for measurement is h, the angle of incidence is 6, and the

broadening of the diffraction line measured as full width at half maximum (FWHM) is B. For P-

W2N or P-W2C peaks, the most intense (111) diffraction peak was used to determine FWHM.

The peak position in an XRD spectrum can also be used to calculate the lattice parameter

of the unit cell. The inter-planar spacing, d, is obtained from Bragg's Law.


d-
2sin6 (3-14)

The value of h for Cu Kal radiation is 1.5405 A+. The lattice parameter, a, for a cubic cell is

given by Equation 3-15.

1 1
---- (h2+k212)
da2 (3-15)

The Miller Indices are h, k and 1. For a hexagonal crystal, the lattice parameter is given by

Equation 3-16.

14 1
22(h2+hk+k2) (2 2
d" 3a' c (3-16)

For hexagonal crystal, two different peak positions are required to solve for lattice parameters a

and c.









3.6.2 Auger Electron Spectroscopy

Film composition was determined by AES. In AES, energetic electrons are used to knock

out core electron of atoms close to the surface of the thin fi1m. The excited atom relaxes to lower

energy state by either Auger emission or X-ray fluorescence. In Auger emission, the excited

atom relaxes to lower energy state by transitions of an electron from higher energy level to the

inner shell. The excess energy from the transition causes another electron from higher energy

level to ej ect from the atom. The kinetic energy of the ej ected Auger electron can be used to

identify the energy levels for that particular transition. Determination of the energy levels can

then be used for elemental identification. In AES spectra, the number of electrons ej ected is

plotted versus the kinetic energy and the data are differentiated for better resolution. Elemental

identification is then done by comparing the peak positions in the AES spectrum to reference

spectral lines of different elements.

Auger electron spectroscopy measurements were done using a Perkin-Elmer PHI 660

Scanning Auger Multiprobe. A 5 kV acceleration voltage and 50 nA beam current was used for

the Auger analysis. The beam diameter was 1 Clm and the angle of incidence was 450. The AES

spectra were obtained between 50 and 2050 eV with a step size of 1 eV. Prior to AES

measurement, the sample surface was cleaned by sputter etching for 30 sec using Ar+ ions. The

etch rate for the sputtering was calibrated at 100 A/min using a Ta203 standard. The ion gun for

sputtering was operated at 15 mA current at Ar pressure of 1.5 x 10-3 Pa. The beam voltage of

the ion gun was 5 keV.

A general expression for determination of quantifieation of atomic concentration in given

by Equation 3-17.



N CI~/S~ (3-17)









where,

Xa = concentration of any constituent a in the sample

Na = number of atoms per unit volume of constituent a

CNi = sum of number of atoms per unit volume of all the constituents of the film

la = Auger intensity of a particular transition for constituent a

l~i = sum of Auger intensities of all the constituents of the film

Sa = relative sensitivity factor for constituent a

Si = relative sensitivity factor of constituent i

Auger intensity is calculated from peak-to-peak height in the numerically differentiated

EdN(E)/dE spectrum. The spectrum is numerically differentiated according to the Savitzky-

Golay differentiation algorithm using five differentiation points. The relative sensitivity factor is

calculated by comparing the peak intensity of a pure element with that of a reference standard

(typically the Cu LMM transition obtained at 10 keV). Table 3-6 lists the relative sensitivity

factors of C, N, O and W used for quantification in AES.lso The table also lists the position of

peaks used for quantification of different elements. It should be noted that the calculation of

atomic concentration using Equation 3-17 does not take the matrix effect into account. If an

element is present in a matrix of other elements, its properties such as Auger transition

probability, secondary ionization by scattered electrons and inelastic mean free path are different

from that of pure element. This change in sensitivity factor is often referred to as the matrix

effect. The elemental composition obtained by using pure elemental sensitivity factors can have

significant error because of the matrix effect. While the absolute value of film composition

might not be accurate, the film composition values can be helpful in comparing films deposited









at different process conditions. Accurate calculation of film composition requires the calibration

of the sensitivity factors using a WNxCv standard of known composition.

The change in elemental composition of the film with depth was measured by AES depth

profile. In this technique, the sample surface is sputtered and then analyzed to identify the

intensity of peaks of elements present in the film. This process is repeated to obtain the depth

profile for different elements. An AES depth profile can be obtained in two different ways: 3-

point depth profiling and survey depth profiling. The main difference between these two

techniques is the manner in which peak intensity is calculated. In survey depth profiling, an

entire survey of a particular peak is taken to determine the peak intensity. In 3 -point depth

profiling, a peak is identified by measuring the peak at 3 points. Two of the three points

correspond to the baseline of the peak and the third point corresponds to the highest point of the

peak. From these three points, the peak intensity is calculated. The advantage of 3-point depth

profiling over survey profiling is that since only 3 points of a peak are measured, the

measurement is much faster and a better depth resolution can be achieved. The disadvantage of

this technique is that precise elemental composition can not be determined. In the present work,

3-point technique was used for depth profiling.

3.6.3 X-Ray Photoelectron Spectroscopy

XPS analysis was done to obtain bonding information of atoms in thin films. Surface

analysis by XPS is done by irradiating the sample surface with monoenergetic soft X-rays and

measuring the energy of the electrons ej ected from sample by photoemission process. The kinetic

energy of an emitted electron, KE, is given by Equation 3-18.

KE = hv BE cps (3-18)

The energy of the photon is hy, the binding energy of atomic orbital from which the electron is

generated is BE and the work function of the spectrometer is cps. Since each element has a









unique binding energy, XPS can be used to determine the elements present in the sample.

Depending on the bonding of a particular atom in the sample, the binding energy of that atom

shifts. This shift can be used to obtain the bonding information of an element in the sample.

Table 3-7 lists the binding energies of atoms in different compounds of W, C, N and O along

with pure elemental binding energies of polymeric and graphitic carbon.

X-ray photoelectron spectroscopy measurements were taken using a Perkin Elmer PHI

5600 system. X-ray photoelectron spectroscopy spectra were taken using monochromatic Mg

Koc radiation with the X-ray source operating at 300 W (15 kV voltage and 20 mA current). The

sample surface was sputter etched for 15 min using Ar+ ions to remove surface contaminants.

The etch rate for the XPS system was calibrated at 10 A/min using Ta203 standard. The ion gun

for sputtering was operated at 15 mA current at a Ar pressure of 1.5 x 10-3 Pa. The beam voltage

of the ion gun was 5 keV.

3.6.4 Scanning Electron Microscopy

Film thickness measurement was done using SEM. In SEM, an electron beam is attracted

towards substrate by a positive bias on the substrate. Secondary electrons are generated from the

interaction of electron beam with the substrate surface. Secondary electrons have low energy (2

- 5 eV) and are generated very close to the surface. By mapping the coordinates with the

intensity of the secondary electron, an image is constructed. For the present study, JEOL JSM

6400 and JEOL JSM 6335F scanning electron microscopes were used for measuring film

thickness. To generate electron beam, JEOL JSM 6400 microscope uses thermionic emission

whereas the JEOL JSM 6335F uses field emission technique. In thermionic emission, a tungsten

filament was heated to 2400 oC to generate electrons. In field emission, electrons were

generated by applying high bias of 15 kV on a single crystal tungsten tip. For measurements of









less than 40,000x magnification, SEM JEOL 6400 was used. To obtain resolution higher than

40,000x, a field emission SEM JEOL JSM 6335F was used. To measure film thickness, the

substrate was cleaved and the cross-section of the substrate was viewed under the SEM.

3.6.5 Four Point Probe

Film sheet resistance was measured by using a four point probe. A four point probe

consists of four sharp metallic probes lined with one another. Current is passed through the outer

probes and the voltage is measured by the inner probes. Based on the V and I measured, the

sheet resistance is calculated using Equation 3-19.


R =4.53x V
I (3-19)

The sheet resistance of deposited films was measured using an Alessi Industries four-point

probe. The resistivity of the film, p, is then calculated by Equation 3-20.

p=Rs xt (-0
The thickness of the film, t, is obtained from SEM imaging.

3.7 Diffusion Barrier Testing

For diffusion barrier testing, copper film was deposited on the WNxCy thin film using Kurt

J. Lesker CMS-18 multi-target sputter system. Prior to the deposition of copper on the barrier

film, the film was exposed to air for 1 2 hr. The base pressure of the sputter system was 10-6

Torr and deposition was done at 5 mTorr with Ar used as sputter deposition gas. The forward

sputtering power for Cu target was 250 W and the film growth rate was 240 A/min. The

thickness of copper deposited on the barrier film was 100 A+. After the copper deposition, the

Cu/WNxC,/Si stack was annealed in the same sputter chamber with the chamber pressure of ca.

10-6 Torr. The annealing temperature was varied between 200 and 600 oC.









To check for diffusion of copper through the barrier film, two different procedures were

used. Pre- and post-anneal XRD measurement were done to check for bulk diffusion of copper

through the silicon. X-ray diffraction was done using the same procedure described earlier. If

the copper diffuses through the silicon, copper forms a stable compound Cu3Si. A Cu3Si peak in

the post-anneal XRD spectrum would indicate the diffusion of copper into the silicon and the

failure of the barrier film. Another method used to detect diffusion of copper in silicon was 3 -

point AES depth profile. An AES depth profile can indicate the presence of copper in the barrier

as well as silicon. The advantage of this technique over XRD is that it has higher sensitivity (1

at. %). The disadvantage of 3-point AES depth profile is that during sputtering, the Ar+ ion can

knock copper atoms into the barrier film and also into silicon if the barrier film is very thin. This

effect of sputtering is commonly referred as the 'knock-on' effect. The 'knock-on' effect can

lead to artifacts in the depth profile. To avoid this problem especially for ultra thin diffusion

barrier films, the copper film was etched off after annealing. The copper etching was done by

dipping the Cu/WNxC,/Si stack in 50 % HNO3 SOlution for 3 seconds. The 3-point depth profile

was done following the etching of the copper layer.










Table 3-1. Process parameters for CVD of thin films from precursors 3a,b, 4 and 5
Process Parameter Precursor and Carrier Gas
3a,b+N\H3+H2 4 + H2 4+N\H3+H2 4+N\H3+N2 5 + H2
(Ch. 4) (Ch.5) (Ch.6) (Ch.6) (Ch.7)
Temperature (OC) 450 750 400-750 300 750 300 750 300 750
Pressure (Torr) 350 350 350 350 350
H2 Flow Rate (sccm) 1000 1000 1000 0 1000
N2 Flow Rate (sccm) 0 0 0 1000 0
NH3 Flow Rate (sccm) 25 0 30 30 0
Precursor conc. 11.2 9.0 9.0 9.0 7.4
(mg/mL benzonitrile)


Table 3-2. Knudsen number for precursors 3a,b, 4 and 5 with H2 or N2 carrier gas for different
deposition temperature
Deposition Knudsen Number
Temperature (OC) Precursor 3a,b Precursor 4 Precursor 5
H2 N2 H2 N2 H2 N2
300 .8310-3.77x10-' 1.05x10-' 3.42x10-' 1.16x10-' 3.76x10
350 1.94 x10-7 3.98 x10-7 1.11 x10-7 3.61 x107 1.23 x10-7 3.97 x107
400 2.04 x 10-7 4. 19 x10-7 1.17 x10-7 3.8 sx10-7 1.29 x10-7 4.1 lx 107
450 2. 14x10-7 4.4x10-7 1.23x10-7 4x10-7 1.35x10-7 4.39x107
500 2.24x 10-7 4.61x10-7 1.29x10-7 4. 19x10-7 1.42x10-7 4.6x107
550 2.34x10-7 4.82x10-7 1.35x10-7 4.38x10-7 1.48x10-7 4.8sx107
600 2.45x10-7 5.03x10-7 1.4x10-7 4.57x10-7 1.55x10-7 5.01x107
650 2.55x10-7 5.24x10-7 1.46x10-7 4.76x10-7 1.61x10-7 5.22x107
700 2.65x10-7 5.45x10-7 1.52x10-7 4.95x10-7 1.68x10-7 5.43x107
750 2.75x10-7 5.66x10-7 1.58x10-7 5.14x10-7 1.74x10-7 5.64x107












Reynolds Number
H2 N2
0.62 4.23
0.60 4.08


Peclet Number


N2
2.92
2.82
2.72
2.64
2.56
2.49
2.42
2.36
2.30
2.24


0.58
0.56
0.55
0.53
0.52
0.50
0.49
0.48


3.95
3.82
3.71
3.60
3.51
3.41
3.33
3.25


Table 3-4. Grashof number for H2 and N2 Carrie g fo dfen epsto merue


(Ra*Pr)UL
H2 N2
7 19
7 18
7 18
7 18
7 17
6 17
6 17
6 16
6 16
6 15


H2
3718
3533
3304
3059
2816
2583
2365
2164
1980
1812


N2
170503
162032
151514
140299
129147
118470
108472
99240
90790
83098


I


0.43
0.41
0.40
0.39
0.38
0.37
0.36
0.35
0.34
0.33


Grashof Numb er
H2 N2
5388 247106
5120 234829
4788 219585
4433 203332
4081 187170
3744 171695
3428 157206
3136 143827
2869 131580
2626 120432


Rayleigh Number


Deposition
Temperature (OC)
300
350
400
450
500
550
600
650
700
750


Table 3-3. Reynolds number for H2 and N2 carrier gas for different deposition temperature


(Re*Pr~o.5
H2
0.66
0.64
0.63
0.62
0.61
0.61
0.60
0.59
0.58
0.57


= (e>o.5
N2
1.71
1.68
1.65
1.62
1.60
1.58
1.56
1.53
1.52
1.50


Deposition
Temperature (OC)
300
350
400
450
500
550
600
650
700
750













Table 3-5. Reference XRD peaks and their relative intensity for Si, P-W2N, P-W2C, a-W2C and
WO3 from JCPDSus8 for Cu Ku radiation
Si P-W2N P-W2C a-W2C WO3
26 Intensity 26 Intensity 26 Intensity 26 Intensity 26 Intensity
28.44 100 37.73 100 36.98 40 34.52 25 23.64 100
47.30 55 43.85 47 42.89 100 38.03 22 33.64 69
56.11 30 63.73 33 62.03 30 39.57 100 41.46 20
69.13 6 76.52 44 74.20 50 52.30 17 48.43 16
76.38 11 80.59 13 78.23 10 61.86 14 54.57 36
69.79 14 60.25 22
72.84 2 70.75 6
74.98 12 75.81 11
75.98 10 78.06 2
81.33 2 80.87 3


Table 3-6. Elemental relative sensitivity factors for AES quantification along with position of
peaks in dN(E)/d(E) spectra used for qluantificationiso
Element Relative sensitivity Position of peak used for
factor quantification (eV)
C 0.14 272
N 0.23 381
O 0.4 510
W 0.08 1736










Table 3-7. Binding energy of C, N, O and W in different compounds
Compound Binding Energy (eV) Reference
Cls Nls 01s W3 f-

W2N -- 397 -- 32.8-33 122, 181, 182

W2C 283.4-283.6 -- -- 31.4-31.7 183-185
WC 282.7-283.5 -- -- 31.8-32.4 183-186
WO2 -- -- 530.2 32.8-33.0 183-186

W205 -- -- -- 34.4-34.7 186, 187
WO3 -- -- 530.3-531.0 35.5-36.0 186-190

Pure W -- -- -- 31.2- 31.8 186, 188, 189, 191, 194
Amorphous C 284.5-285.2 -- -- -- 183
Graphitic C 284.4-284.5 -- -- -- 192, 193
Diamond-like C 285.3 -- -- -- 192, 193


Dissolved
Precursor from


IThermocouple


INebulizerl


* Not to scale


Figure 3-1. Schematic of CVD reactor used for thin film deposition











Carrier gas to
reactor


Carrier gas in
reactor


N, H 2
Source Source
Line Line


H2
Soumce
Line


Line


Figure 3-2. Line diagram of the carrier gas feed line. A) Before installation of gas purifier. B)
after installation of gas purifier












Pirani Gauge


Vacuum Gauge


Vacuum Pump



Pirani Gauge


Vacuum Gauge


Lli BVacuum Pump
Drain Port


Figure 3-3. Flow diagram of reactor downstream. A) Before reconfiguration. B) After
reconfiguration








Gas Inlet
Male VCR

SGas Outlet
Male VCR

















Drain Port


Figure 3-4. Design of new cold trap made of stainless steel


































Figure 3-5. Schematic of the old and new NH3 feed port


Precursor+
solvent


* Not to scale









CHAPTER 4
WNxCv THIN FILM DEPOSITION FROM Cl4(RCN)W(NC3H5), N\H3 AND H2

Our group has previously reported growth of WNxCv thin films from 3a,b. Films grown

from 3a,b below 550 oC were amorphous and had a low resistivity at lower deposition

temperature (287 CIR-cm at 450 oC). However, films had low N content even though the low N-

C imido bond dissociation energy for 3a,b is expected to result in higher N content in the film.

Film deposited with 3a,b at 450 oC had only 4 at. % N and the highest N concentration of 11 at.

% was for film deposited at 550 oC. A nitrogen deficient diffusion barrier film is not desirable

because such films can easily form grain boundaries that can act as pathways for copper

diffusion. In addition to having low N content, films deposited with 3a,b had high O content (16

at. % at 450 oC and 11 at. % at 525 OC). When ammonia was used as a co-reactant for

deposition from precursor 2a,b, the WNxC, films had higher N content and lower O content.l5

In this chapter, deposition of WNxC, thin films from 3a,b and ammonia is discussed with

primary focus on diffusion barrier application. The intent of using ammonia as co-reactant is to

increase the N content and decrease the O content so that films deposited with 3a,b are more

suitable for diffusion barrier application. The data for films deposited with 3a,b and H2 (without

ammonia) previously reported have been included to evaluate the effect of ammonia addition

on film properties. The films deposited with 3a,b and ammonia are also compared with those

deposited with 2a,b and ammonia to evaluate the effect of C3H5 and zPr substituents in tungsten

imido precursors on film properties.


4.1 Precursor Synthesis

Standard Schlenk and glovebox techniques were employed in the synthesis of 3a.

Tungsten oxychloride was prepared by a slightly modified literature method.195 Tungsten

oxychloride (1.229 g, 3.60 mmol) was added to a solution of allyl isocyanate (0.366 g, 4.41









mmol) in heptane (60 mL) in a sealed pressure vessel and the resulting mixture was heated at

110 oC for 36 h. After the solvent was removed in vacuo, the reddish brown residue was

dissolved in ca. 10 mL of CH3CN. The resulting solution was stirred for two hours and the

solvent was removed under reduced pressure. Brown residue from the mixture was washed with

5 x 10 mL of toluene and the extracts concentrated to approximately 5 mL. The product was

precipitated by adding hexane. The orange-brown solid was filtered and washed with hexane to

afford 0.974 g (64 % yield) of the imido complex. The benzonitrile complex

Cl4(PhCN)W(NC3HS) (3b) was produced in situ by substitution of the acetonitrile ligand of 3a

with benzonitrile, which was utilized as the solvent for the deposition experiments. Figure 4-1

shows the X-ray crystallographic structure of 3a. "7 Table 4-1 lists important bond distances and

bond angles for 3a.

4.2 Film Composition

Figure 4-2 shows the AES results for films deposited in the presence and absence ofNH3

at temperature ranging from 450 to 750 oC. The Auger spectra for films deposited with

ammonia indicate the presence ofW, N, C, and O in the films. Note that the AES measurements

for films deposited with ammonia were done after sputtering for 30 sec whereas those for films

deposited without ammonia were measured after 2 min of sputtering. Sputter etching can change

the near surface composition of compounds due to preferential sputtering.196 In fact, tungsten

nitride has been reported to undergo preferential erosion of nitrogen during sputter etching.18s9

Despite preferential sputtering, the film composition obtained from Auger measurements can be

very useful in comparative analysis of films deposited at different temperature.

For films deposited with ammonia at 450 oC, the W concentration was 51 at. %, which

increases to 61 at. % for growth at 475 oC. This is accompanied by a decrease in N










concentration from 23 at. % to 17 at. %. Above 450 oC, the W concentration gradually

decreases, accompanied by an increase in C content in the films.

The film deposited with ammonia at 450 oC shows the highest N concentration of 23 at. %.

As the deposition temperature is increased, the N concentration in the film decreases. The

lowest N concentration of 8 at. %. is observed for films grown at 750 oC.

The carbon concentration remains relatively steady at ca. 20 at. % for growth between 450

and 500 oC. Between 500 and 600 oC, the C concentration increases twofold and continues to

rise with growth temperature. Increased incorporation of C from decomposition of the precursor

ligands and the benzonitrile solvent is consistent with such an increase in C concentration.

Oxygen incorporation in the film remains below 6 at. % throughout the deposition

temperature range. While the O concentrations at the lower end (450 500 oC) and higher end

(650 750 oC) of the temperature range are slightly higher than that at the center (525 600 oC),

the O concentration remains fairly constant over the range of deposition temperature. Oxygen is

incorporated into the films either from residual gas in the reactor during growth or from exposure

of film to atmosphere after growth.156

In comparison to films grown in the absence of ammonia, deposition with ammonia as co-

reactant resulted in a decrease in W concentration, increase in N concentration, and a marked

decrease in O concentration (Figure 4-2). The increase in N concentration for depositions with

ammonia is quite dramatic in the low temperature range, as exemplified by the effect for growth

at 450 oC, where the addition of ammonia increases the N concentration from 4 at. % to 23 at. %.

The increase in N concentration suggests that the ammonia molecule reacts with 3a,b either in

the gas phase (homogeneous reaction) or on the substrate (heterogeneous reaction), resulting in

higher N incorporation in the film. Below 650 oC, C incorporation for depositions with









ammonia was higher than that for depositions without ammonia. Even at low temperature, when

C deposition from the solvent is lowl56, the C content of films deposited with ammonia is higher

than for films deposited without ammonia.

Due to the presence of chlorine in the precursor complex 3a, the possible presence of

chlorine in the fi1ms was of interest. This analysis could not be carried out by Auger

spectroscopy because the W NNN peak overlaps with the Cl LMM peak at 180 eV. Thus, XPS

data were collected for films deposited at 450, 600 and 750 oC (Figure 4-3) to ascertain if the

fi1ms contained chlorine. No chlorine peaks were observed for either the Cl 2s or Cl 2p3/2 at 270

eV and 199 eV, respectively, confirming that the chlorine content in the films was lower than the

detection limit of XPS (ca. I at. %). Fe 2p peaks at 723 eV and 711 eV in the XPS spectra are

artifacts that arise from the sample holder (which contains Fe) due to small sample sizes.

4.3 X-Ray Diffraction Measurement

4.3.1 Film Crystallinity

Figure 4-4 shows the XRD spectra for films deposited using 3a,b and ammonia. The three

prominent peaks in the spectra at 33.05, 61.70 and 69.15 degrees 26 correspond to standard

reference peaks of Si(200) Ka, Si(400) KP and Si(400) Ku respectively. The XRD spectrum at

650 oC showed a small peak at 38.45 degrees 26 indicating the presence of crystalline phase of

P-W2N or P-W2C. The deposition at all other temperature except 650 oC showed no peaks

corresponding to P-W2N or P-W2C, indicating that the films were X-ray amorphous.

If films are very thin, the XRD measurements cannot detect a crystalline phase because of

the low intensity of the diffracted X-ray beam. For very thin polycrystalline films, the XRD

measurements done in the Bragg-Bretano geometry (6-26 scans) do not show diffraction peaks

because of low intensity signal from the film and the XRD spectrum is dominated by the high

intensity substrate peak. In GIXD, the scan is done at grazing incidence; hence most of the










signal from the diffraction is generated from the top 1 2 nm of the thin film. The substrate

peak, if present, do not dominate the GIXD spectrum. The depth of penetration of X-rays can be

increased by increasing the angle of incidence in the GIXD measurement. Grazing incidence

XRD measurements were done for films deposited with 3a,b and ammonia to determine if the

films were amorphous as indicated from the XRD measurements by Bragg-Bretano geometry.

Figure 4-5 shows the grazing incidence XRD spectra for films deposited between 450 and 700

oC. Figure 4-6 shows the thickness of films for which grazing incidence XRD spectra were

taken. The XRD spectrum of the film deposited at 450 oC indicates that it is X-ray amorphous.

At 500 oC, there are three peaks that emerge at 37.47, 43.49 and 63.71 260 degrees. All three

peaks lie between the standard diffraction peaks for P-W2N and P-W2C, suggesting the existence

of separate P-W2N and P-W2C phases or the presence of a single WNxC, phase. As the

deposition temperature is increased from 500 to 550 oC, the same three peaks appear, though the

intensity of these peaks is lower than that for films deposited at 500 OC. In the same temperature

range, the tungsten concentration changes from 59 at. % to 49 at. %. A lower tungsten

concentration would lead to decrease in the W2N phase, W2C phase or both, resulting in a

decrease in the film crystallinity. The XRD spectrum for film deposited at 600 OC shows the

highest level of crystallinity. As the deposition temperature is increased from 550 to 600 OC, the

nitrogen concentration decreases while the carbon concentration increases. An increase in

crystallinity would suggest an increase in the P-W2C phase. The films deposited at 650, 700 and

750 oC show a decrease in crystallinity, evidenced by decrease in peak intensity. At higher

temperature, an increase in amorphous C deposition would hinder the formation of crystallites,

which is consistent with the decrease in film crystallinity at higher deposition temperature.










Peaks at 650, 700 and 750 oC are much broader possibly due to increased film strain at high

temperature.

We have previously reported XRD spectra for films deposited with 3a,b without

ammonia.m" The films deposited without ammonia are amorphous below 550 oC. At and above

550 oC, the film crystallinity increases with deposition temperature. Films deposited with

ammonia do not show this consistent increase in crystallinity with deposition temperature. In

fact, for higher temperature depositions with ammonia, the film crystallinity actually decreases.

This observation reinforces the assertion that the ammonia co-reactant alters the mechanism of

film deposition.

4.3.2 Lattice Parameter

Figure 4-7 shows the change in lattice parameter with deposition temperature for growth

with and without ammonia. The P-WT\xCv (111) peak position from XRD spectra was used to

calculate the lattice parameter. All the films deposited with ammonia between 500 and 750 oC

have a lattice parameter between the standard lattice parameters for P-W2N (111) and P-W2C

(111). The changes in lattice parameter with deposition temperature could result from either

uniform strain or compositional variation. Since the WNxCv films are not highly ordered,

compositional variation rather than uniform strain is believed to be the primary factor affecting

lattice parameter. The lattice parameter decreases with an increase in deposition temperature for

depositions with ammonia between 500 and 700 oC, whereas it increases with an increase in

deposition temperature for depositions without ammonia. A decrease in lattice parameter could

result from a decrease in N and/or C in interstitial sublattice and an increase in the vacancy

concentration. For films deposited with ammonia, the N concentration decreases with deposition

temperature whereas the C concentration increases. One possible explanation for the decrease in









lattice parameter with increase in deposition temperature is that the variation in N concentration

in the films is primarily responsible for the lattice parameter variation and the additional carbon

deposited at higher temperature incorporates outside the WNxCv polycrystals.

4.3.3 Polycrystal Grain Size.

The grain size (t) was calculated using the Scherrer equation videe supra). The most

intense (111) diffraction peak was used to determine FWHM. Figure 4-8 shows the change in

grain size with temperature for films deposited with and without ammonia. The average grain

size calculation shows that films deposited with 3a,b and ammonia are nanocrystalline between

500 and 750 oC deposition temperature, with the average grain size ranging from 37 to 60 A+. A

previous study has reported the growth of nanocrystalline WNxCv between deposition

temperature of 225 and 400 oC.197 In the present work, we found that the films deposited with

3a,b and ammonia were amorphous at 450 oC deposition temperature and nanocrystalline growth

is observed at deposition temperature of 500 oC and higher. At a deposition temperature of 500

oC, the average grain size of crystallites for film deposited with ammonia is 56 A+. The average

grain size remains almost unchanged at 550 oC. Between 550 oC and 700 oC, the average grain

size gradually decreases from 58 A+ to 39 A+. As the deposition temperature increases, the excess

C deposited from the solvent/precursor fragmentation could hinder the growth of polycrystals,

resulting in a decrease in grain size.

The comparison of grain size for experiments with and without ammonia shows that for

deposition at temperature 600 oC and below, the average grain size is higher for films deposited

with ammonia than for those deposited without ammonia. At 600 oC, the grain size for film

grown with ammonia is similar to that for film grown without ammonia within the margin of









error. At 650 oC, the grain size for film deposited with ammonia is slightly lower than the grain

size for films deposited without ammonia.

4.4 Film Growth Rate

Growth rates were estimated by dividing film thickness (measured from X-SEM) by

deposition time. Figure 4-9 shows X-SEM images for films grown at 450 and 650 oC. The

growth rate for 3a,b with ammonia ranged from ca. 4 A/min at 450 oC to ca. 17 A/min at 750 oC.

It should be noted that for growth using 3a,b without ammonia, the concentration of 3a,b in

benzonitrile was 7.5 mg/mL (1.7 x 10-2 moles/L) whereas for growth using 3a,b with ammonia,

the 3a,b concentration in benzonitrile was 11.2 mg/mL (2.66 x 10-2 moles/L). Because the

introduction of ammonia as co-reactant decreases the film growth rate, a higher concentration of

3a,b was necessary to obtain films that were sufficiently thick for materials characterization.

This trend is different from that previously observed for growth from 2a,b, where the use of

ammonia as co-reactant resulted in an increase in growth rate at lower temperature and a

decrease at higher temperature.m' Figure 4-10 shows Arrhenius plots of growth rate for

experiments with and without ammonia. The apparent activation energy (Ea) for film growth

from 3a,b and ammonia was 0.34 eV, which is significantly higher than the 0.15 eV activation

energy reported for depositions with 3a,b precursor without ammonia.

4.5 Film Resistivity

Figure 4-11 shows film resistivity at different deposition temperature for films grown with

and without ammonia. For depositions with ammonia, the resistivity is 5820 CIR-cm for 450 oC

deposition temperature and increases to 7190 CIR-cm for 500 oC. Interestingly, the film

resistivity has the lowest value of 1690 CIR-cm for growth at 550 oC, even though the film

deposited at that temperature has lower W content and higher N content. Resistivity increases to









12650 CIR-cm at 600 OC and continues to increase with an increase in deposition temperature

until it reaches its highest value of 24178 CIR-cm at 700 OC. The decrease in W concentration

and concomitant increase in C concentration between 550 and 700 oC is believed to be the reason

for the increase in film resistivity. If the C in films exists as WCx, it would decrease the film

resistivity. But an increase in resistivity with increase in C content of film could possibly result

from the incorporation of additional C at the grain boundary (outside the WNxC, polycrystal) at

higher deposition temperature. The film resistivity was considerably higher for films deposited

with ammonia compared to films deposited without ammonia. This behavior is expected since

ammonia tends to increase N content in the films and higher N content leads to higher film

resistivity.

4.6 Comparison of Films Deposited from 3a,b and 2a,b

The comparison of films deposited from 3a,b and ammonia with those deposited from 2a,b

and ammonia" can provide insight into the effect of different ligands (C3H5 and zPr) on film

properties. Films deposited at 450 oC from 3a,b and ammonia have similar W and N content,

higher C content and lower O content as compared to films deposited with 2a,b and ammonia at

same temperature (Figure 4-2). At 500 and 600 oC deposition temperature, films from 2a,b

exhibit higher N content and lower C content as compared to films deposited with 3a,b and

ammonia at the same temperature. The comparison of film crystallinity for films deposited with

3a,b and ammonia with those deposited with 2a,b and ammonia shows that while 3a,b results in

crystalline films at and above 500 oC, films deposited with 2a,b do not show crystallinity until

600 oC and higher. This trend in crystallinity can be explained by considering that the film

growth rate for films deposited with 2a,b and ammonia is considerably higher (17 23 A/min)

than that for films deposited with 3a,b and ammonia (4 17 A/min), especially at lower










deposition temperature. Higher growth rate could shorten the time for crystallites to grow via

surface diffusion, resulting in amorphous film growth.

The comparison of film resistivity (Figure 4-1 1) shows that at 450 and 500 oC deposition

temperature, films grown with 3a,b and ammonia have a significantly lower resistivity than

those grown with 2a,b and ammonia. Interestingly, films deposited with both precursors have

the lowest resistivity for deposition at 550 oC. Above 550 oC, there is a dramatic increase in

resistivity of films deposited with 3a,b and ammonia while the resistivity of films grown with

2a,b and ammonia levels off.

Evaluation of all films grown from 3a,b or 2a,b and ammonia shows that film deposited

with 3a,b and ammonia at 450 oC is the best candidate for diffusion barrier application because

of its high N content, low O content, low deposition temperature, amorphous microstructure and

lowest resistivity among the amorphous films grown from 3a,b or 2a,b with ammonia.

Diffusion barrier testing is required to ascertain if this film is suitable for diffusion barrier

application.

4.7 Conclusions

Successful thin film deposition of WNxCv by MOCVD utilizing 3a,b, H2 and ammonia has

been achieved. When used as a co-reactant with the allylimido complexes 3a,b, ammonia

significantly alters the composition, crystallinity, and resistivity of the resulting films as

compared to films grown without ammonia. Films deposited with ammonia had significantly

higher carbon and nitrogen content and substantially lower oxygen content. No chlorine was

detected in the XPS spectra of films grown over the entire deposition temperature range. While

films grown without ammonia showed an increase in film crystallinity with increasing deposition

temperature, those grown with ammonia exhibited more complex behavior, with crystallinity

peaking for growth at 600 OC. The activation energy for the film growth from 3a,b and ammonia









is estimated to be 0.34 eV, which is significantly higher than the activation energy of 0. 15 eV

reported for film growth from 3a,b without ammonia. Films deposited with ammonia have

higher resistivity as compared to films deposited without ammonia because of a higher N and C

concentration coupled with a lower W concentration. The lowest resistivity for films deposited

with ammonia was the value of 1690 C1a-cm obtained from growth at 550 oC.

Films deposited with 3a,b and ammonia have higher C content and lower O content than

those grown from 2a,b and ammonia. The N content of film grown with 3a,b and ammonia is

similar to that for film grown with 2a,b and ammonia at 450 and 700 oC. Films grown with 2a,b

and ammonia below 600 oC were amorphous, whereas deposition from 3a,b and ammonia

resulted in amorphous film below 500 oC. At temperature below 550 oC, films deposited with

3a,b and ammonia had a significantly lower resistivity as compared to films deposited with 2a,b

and ammonia. But above 550 oC, films deposited with 3a,b and ammonia had a much higher

resistivity as compared to those deposited with 2a,b and ammonia.

From a diffusion barrier application standpoint, it has been demonstrated that ammonia can

be used with 3a,b to increase the N content of WNxC, films. A corresponding increase in film

resistivity is also seen and some optimization may be needed to strike a balance between higher

N content and lower film resistivity.











Table 4-1. Bond distances (A) and angles (degrees) for Cl4(CH3CN)W(NC3HS) (3a)
Bond Distance/angle Bond Distance/angle
W-N(1) 1.687(9) W-Cl(4) 2.351(9)
W-N(2) 2.308(8) N(1)-C(1) 1.508(17)
W-Cl(1) 2.339(10) C(1)-C(2)' 1.51(2)
W-Cl(2) 2.317(8) C(2)'-C(3)' 1.36(3)
W-Cl(3) 2.324(9) N(2)-C(4) 1.130(12)
C(1)-(1)-W167(2) N(1)-W-N(2) 175.8(15)
N(1)-W-Cl(2) 90.6(8) N(2)-W-Cl(2) 86.0(7)
N(1)-W-Cl(3) 93.6(7) N(2)-W-Cl(3) 84.0(6)
Cl(2)-W-Cl(3) 88.7(4) C(4)-N(2)-W 175(3)
N(1)-WCl(1)102.4(8) N(2)-W-Cl(1) 81.1(7)
C()'C1)N()114.2(21) C()-()-()120(2)
Cl(3)-W-Cl(1) 90.18(13) N(2)-C(4)-C(5) 178(2)


Figure 4-1. Thermal ellipsoids diagram of the molecular structure of Cl4(CH3CN)W(NC3HS)
(3a). Thermal ellipsoids are plotted at 50% probabilitym5
























la4b+NH3
401 -0 )- aqb without NH3
--y-- 2agb+NH3
30







10




300 400 500 600 700 801

Deposition Temperature (C)

40

-e- lab +NH3
--0~- laqbwithoutNH3
301 I- --t-- 2a4b+NH3







O,


d








-- in,b+NH3
-la,b without NH3


-f- 2a,b+NH3

400 500 600

Deposition Temperature (C)


700 800


--- lab + NH3 0
-4- without NH3 lrc
--2a4b+NH3 ,














400 450 500 550 600 650 700 750
Deposition Temperature ("C)


300 400 500 600

Deposition Temperature ("C)


700 800


Figure 4-2. Composition of films deposited from 3a,b (with and without ammonia" ) and 2a,b

with ammonia"' at different deposition temperature as determined by AES on Si

(100) substrate.


60






3 0













0 s



350


--
o
--0













1.6e+5


1.4e+5 -( C Auger

O Auger
1.2e+5 -N Auger

S1.0e+5 -Fe 2p3/2 O 1s W4320
I i lil450 oC W 4p
8.0e4 ,, ~ )N 1s W 4d W 4f
6 .0e+4 --. ..



S4.0e+4 ~C750 oC I-;~y

2.0e+4-

0.0


1000 800 600 400 200 0

Binding Energy (eV)




Figure 4-3. X-ray photoelectron spectroscopy patterns of films deposited at 450, 600 and 750 oC
from 3a,b and ammonia on Si (100) substrate.
























600 oC

550 *C








10 20 30 40 50 60 70 80

20 Degrees

Figure 4-4. X-ray diffraction spectra of films deposited on Si (100) substrate from 3a,b and
ammomia
























149


















~ u~rv~cr.I. .~ 750 oC

700 oC




C ~m lnsl'l 5~r4~u ~ .. 6000C:

550 0C

500 oC




30 40 50 60 70

26 Degrees
Figure 4-5. Grazing incidence XRD spectra of films deposited on Si (100) substrate from 3a,b
and ammonia. The solid and dashed vertical lines indicate the location of reference
peaks for standard powder diffraction of P-W2N and P-W2C respectively











0: 30


0.25


S0.20




- 0.10- -E

LL.



(c)
0.00
400 450 500 550 600 650 700 750 800

Deposition Temperature (oC)

Figure 4-6. Thickness of film (obtained from X-SEM) at different deposition temperature for
films deposited with 3a,b and ammonia on Si(100) substrate











4.26


4.24 Without Ammonia

4.22-


g 4.20-

2 4.18-


.2 4.16-


4.14
WNo.5
4.12-


4.10
450 500 550 600 650 700 750 800

Deposition Temperature (oC)
Figure 4-7. Lattice parameters for films grown from 3a,b on Si (100) substrate with and without
ammonia. The solid line at 4. 126 A+ corresponds to the standard lattice parameter for
P-W2N. The dashed line at 4.236 A+ corresponds to standard lattice parameter for P
-W2C (1 11). Error bars indicate uncertainty in determination of peak position for
lattice parameter calculation.





























I I I I I I
500 550 600 650 700 750 8


320 I
450


Deposition Temperature (oC)

Figure 4-8. Average grain size at different deposition temperature for films grown from 3a,b
with and without ammonia' on Si (100) substrate.



1 pm 1 pm



SI substrate Film
SI substrate Fl






Figure 4-9. Scanning electron microscope images for films grown from 3a,b with ammonia. A)
Film deposited at 450 oC. B) Film deposited at 650 oC


+ With Ammonia
+ Without Ammonia














S3a,b with NH,
O 3a,b without NH,
y 2a,b with NH,


Deposition Temperature (oC)


750 700 650 600 550


450


O1 -


0.9


1000/T (K-1)

Figure 4-10. Arrhenius plot of log of film growth rate vs. inverse temperature for deposition
from 3a,b (with and without ammoniam) and 2a,b with ammoniam5














*


60000


E~ 1 I 3a,b + NH,

E ,000- j- 3a,b without NH,
-y-- -2a,b + NH,
O
40000-


30000-


P~20000-


10000-



400 450 500 550 600 650 700 750 800

Deposition Temperature ( "C )

Figure 4-11. Change in film resistivity with deposition temperature for films srown on Si (100)
from 3a,b (with and without ammoniam)7 an ab ihamoi









CHAPTER 5
DEPOSITION OF WNxCv USING W(N'Pr)Cl3 zPr(NMe2)N'Pr] AND H2

Transition metal guanidinate and amidinate complexes have been used to deposit a variety

of materials including TiCxNv, Fe, Co, Ni, Cu, and TaN198-201 by CVD and/or ALD. The success

with these complexes suggested preparation of guanidinato derivatives of 2a,b as precursor

candidates. This chapter discusses the deposition of WNxCv thin films using

W(N'Pr)Cl3 ['PrNC(NMe2)N'Pr] (4), a guanidinato derivative of 2a,b, and evaluates its potential

for deposition of diffusion barrier films. The fi1m properties obtained from 4 are compared to

those from 2a,b to assess the effect of the guanidinato ligand. The WNxCv fi1ms were also

evaluated as diffusion barriers by coating them with PVD Cu and annealing the Cu/WNxC,/Si

stack in vacuum at different temperature.


5.1 Precursor Synthesis202

A 250 mL Schlenk tube was charged with LiNMe2 (0.223 g, 4.38 mmol) and 100 mL of

Et20. The resulting colorless suspension was cooled to 0 OC. 1,3-diisopropylcarbodiimide (0.69

mL, 4.4 mmol) was added drop wise to the suspension at 0 OC. After warming the reaction

mixture to room temperature over 2 h, the resulting cloudy solution of lithium guanidinate

reagent was cannula-transferred into a 250 mL Schlenk tube containing a solution of

W(NCH(CH3>2>C 4(OEt2) (2.00 g, 4.38 mmol) in Et20 (100 mL) at -78 oC. The reaction mixture

was stirred for 20 min at -78 OC and then warmed to room temperature overnight in the absence

of light. The solvent was removed from the reaction mixture in vacuo and extraction into Et20

(200 mL) followed by filtration yielded a dark amber solution. Et20 was removed in vacuo to

give crude 4 as a dark amber powder. Pure 4 crystals (1.27 g, 56%) were obtained by

recrystallization from a toluene solution layered with hexane at -20 OC. Figure 5-1 shows the X-










ray crystallographic structure of precursor 4 and Table 5-1 lists important bond distances and

bond angles for 4 obtained from the X-ray structure.

5.2 Film Growth

The concentration of 4 in benzonitrile was 9.0 mg/mL and the flow rate of precursor

solution was 4 mL/hr. The H2 flow rate was held constant 1000 sccm and reactor pressure was

350 Torr. Detailed description of reaction conditions is provided in Chapter 3. The lowest

temperature at which appreciable fi1m growth was observed from precursor 4 was 400 oC. Films

were generally smooth and had a shiny metallic surface with fi1m color varying from golden

(deposition below 500 oC) to shiny black (deposition at and above 400 oC).

5.3 Film Composition

Figure 5-2 shows AES results for films deposited with precursors 4 and 2a,b using H2 aS

co-reactant and carrier gas. The AES spectra indicated the presence ofW, N, C, and O in films

deposited with precursor 4. For films deposited with 4 at 400 OC, the W concentration is 53 at.

%. As the deposition temperature is increased to 450 oC, the W content reaches its highest value

of 66 at. %, then gradually decreases to 37 at. % for deposition at 700 OC, reflecting an increase

in C content as the deposition temperature rises. The N content of films deposited with precursor

4 has its highest value of 12 at. % for deposition at 400 OC. The N content of the film decreases

as deposition temperature is increased from 400 to 650 oC, then remains almost unchanged

between 650 and 750 oC. The variation in the final film N content with growth temperature is

likely influenced by several factors, including N volatilization, competition between C and O for

bonding with W sites, and precursor decomposition pathways and rates.

Both C and O were detected in significant amounts in all films. The C content of film

deposited with precursor 4 decreases from 20 at. % at 400 OC deposition temperature to 14 at. %

at 450 oC. At growth temperatures between 450 and 700 oC, the C content of film continuously









increases and reaches its highest value of 57 at. % at 700 OC. Figure 5-3 shows a monotonic

increase in C/W ratio with deposition temperature between 450 and 700 oC. In a previous study

using precursor 2a,b, it has been demonstrated that the extent of C incorporation depended upon

the solvent (1,2-dichlorobenzene or benzonitrile). 15 It is possible, of course, that the precursor is

also involved in the C incorporation mechanism directly through its decomposition pathway or

indirectly by reaction with the solvent or one its decomposition products. The O content of film

deposited with 4 at 400 OC is 15 at. % and it gradually decreases to 3.0 at. % at 600 OC. Between

600 and 750 oC deposition temperature, the O content of film is ca. 3.0 at. %. Oxygen is

believed to be incorporated into the films either from residual gas (oxygen or water vapor) in the

reactor during growth or from atmospheric oxygen exposure post-growth."

It is desirable for barrier application to deposit the film at the lowest temperature that gives

suitable properties, which includes amorphous microstructure, a high value of N/W atomic ratio

and low oxygen content. The comparison of films deposited from 4 and 2a,b reveals that the

observed lowest deposition temperature for 4 is one experimental increment (50 oC) lower that

the lowest growth temperature of 450 oC for 2a,b. No film deposition was observed for 4 at 350

oC as measured by AES. The comparison of film composition at 450 oC shows that film

deposited from 4 has higher W and N content, similar C content and significantly lower O

content as compared to film deposited with 2a,b using the same solvent. For deposition at and

above 500 oC, the N and O content of films deposited with 4 and 2a,b are quite similar. Since

the number ofN atoms per W atom in 4 is double that in 2a,b, it might be expected that films

deposited with 4 would have a higher N/W ratio as compared to those deposited with 2a,b. The

highest N/W ratio for films deposited with 4 is 0.24 at 400 OC, which is considerably higher than

the highest N/W ratio Of 0. 16 observed for films deposited with 2a,b at the higher temperature









450 oC. This difference in N/W ratio is significant and larger than other values observed for la,b

and 3a,b. Since the difference in O content between films deposited from 4 and 2a,b is

significant only at temperature below 500 oC, it is believed that films deposited below 500 oC by

4 are more dense that those deposited from 2a,b. Higher density in the film would hinder O

diffusion and decrease subsequent O incorporation. Films deposited with 4 have a significantly

lower W content and a higher C content as compared to films deposited with 2a,b for deposition

temperature above 500 oC. At higher deposition temperature, decomposition of C containing

ligands and the benzonitrile solvent results in C incorporation in the film.

As discussed earlier, Cl impurity is highly undesirable in the diffusion barrier film. Since

the precursor 4 has three Cl atoms per molecule, there is a possibility of Cl incorporation into the

film. X-ray photoelectron spectroscopy measurements were done to detect Cl in the thin film

because AES was not capable of detecting Cl impurity in W containing film videe supra). The

detection limit of XPS technique is ca. I at. %. Figure 5-4 shows the XPS profiles of films

deposited at 400, 500, 600 and 700 oC. The sample surface was sputter cleaned with Ar' ions for

5 min prior to the measurement. The spectra do not show any Cl peak indicating that the Cl

content of the film was below the detection limit of the XPS.

5.4 X-Ray Diffraction of Films

Figure 5-5 shows the XRD patterns for films deposited with 4 between 400 and 750 oC.

X-ray diffraction spectra for films deposited at 400 and 450 oC show no peaks attributable to the

film, but only peaks associated with the substrate (Si(200) Ka, Si(400) KP, and Si(400) Ku

reflections at 33.10, 61.75 and 69.200 20, respectively). The absence of other peaks in these two

spectra suggests that films deposited at 400 and 450 oC are X-ray amorphous. The XRD pattern

for the film deposited at 500 OC shows the emergence of crystallinity as evidenced by the two

broad peaks at 37.74 and 44.400. These peaks lie between the standard peak position of P-W2N









[37.74 26 for (111)phase and 43.85 o 26 for (200) phase] and P-W2C [36.98 26 for (111) phase

and 42.89 o 26 for (200) phase], indicating the presence of either the solid solution P-WNxCv or a

physical mixture of P-W2N and p-W2C-178 Films deposited at 550, 600, and 650 oC show

evidence of increased crystallinity with four peaks observed at approximately 37.50, 43.70,

63.55 and 75.050, corresponding to the (1 11), (200), (220) and (31 1) phases of P-WNxCv,

respectively. The (111) peaks become sharper at higher deposition temperature, suggesting an

increase in grain size. The XRD spectra for films deposited at 700 and 750 oC show convoluted

peaks, which are deconvoluted as shown in Figures 5-6 and 5-7, respectively. Deconvolution of

XRD peaks for film deposited at 700 OC affords 5 peaks at 35.40, 37.60, 38.20, 40.00 and

43.100. The two peaks at 37.60 and 43.100 match the standard P-W2N (111) and P-W2N (200)

peaks at 37.75 and 43.850, respectively. The three peaks at 35.40, 38.20 and 40.000 are close to

the peaks of the binary compounds a-W2C (100), a-W2C (002) and a-W2C (101) with reference

values of 34.52, 38.03 and 39.570, respectively.l7 The deconvolution of XRD spectra for films

deposited at 700 OC indicates the presence of a physical mixture of P-W2N and a-W2C phases.

Deconvolution of XRD spectra for films deposited at 750 oC also show 5 peaks at 34.90,

37.60, 38.80, 39.80 and 43.100. The peaks at 37.60 and 43.100 match the standard P-W2N (111)

and P-W2N (200) peaks reasonably well. The other three peaks at 34.90, 38.80, and 39.80 again

match reasonably well with the standard a-W2C (100), a-W2C (002), and a-W2C (101) peaks

values, respectively. Thus the film deposited at 750 oC is a binary mixture of P-W2N and a-W2C

phases with neither of the two phases showing any preferred orientation.

As discussed in Chapter 4, GIXD is more sensitive in detecting crystalline phases in thin

films because of its larger interaction volume. As discussed in Chapter 4, XRD measurements

for films grown with 3a,b and ammonia showed no peaks while the GIXD spectra showed peaks









corresponding to P-W2N or P-W2C. Hence, to ascertain crystallinity of films grown from 4 with

greater accuracy, GIXD measurements were done (Figure 5-8). Grazing incidence XRD

measurements confirmed that the films deposited by 4 at 400 and 450 oC are X-ray amorphous

as no peaks corresponding to the thin film were observed. The films deposited between 500 and

650 oC are polycrystalline with the peak position indicating the presence of a solid solution of P-

WNxCv or a mixture of P-W2N and P-W2C. For deposition at 700 and 750 oC, the GIXD spectra

show peak convolution similar to that observed in XRD spectra. Hence the GIXD results

reaffirmed the results obtained from XRD.

For XRD measurements, the X-rays penetrate completely through the thin film and into the

substrate as indicated by the large Si peak observed in the XRD spectra. The intensity of the

diffracted beam would depend on both the crystallinity of the film as well as the thickness of the

film assuming all other parameters such as film density are constant. Since the film thickness is

different for deposition at different temperature, XRD measurements cannot be used to compare

crystallinity for films deposited at different growth temperature. But for GIXD measurement, X-

rays penetrate only the top 1 2 nm of the film because of the shallow angle of incidence.

Hence, even if the films are of varying thickness, GIXD would have a constant interaction

volume if all the films are thicker than 2 nm and the depth of penetration of X-rays is same for

films deposited at different growth temperature. Since the smallest film thickness for growth

from 4 is 45 nm, GIXD measurements can be used to compare film crystallinity of films for

deposition at different temperature assuming film density is constant. Figure 5-8 shows that the

intensity of the WNxCv peak increases as deposition temperature is increased from 500 to 600

oC, indicating that the film crystallinity increases. This increase in crystallinity is probably

because increased energy is available for crystallite formation and surface diffusion with an









increase in deposition temperature. As deposition temperature is increased from 600 to 6500C,

the WNxCv peak intensity decreases, suggesting a decrease in film crystallinity. As shown from

XPS analysis videe infra), additional C deposited at 700 OC segregates to form both graphitic and

diamond-like C. It is possible that the C segregation starts at 650 oC resulting in a decrease in

surface diffusivity of the adsorbed species during film growth. The decrease in surface

diffusivity causes a decrease in film crystallinity for deposition at 650 oC. The film crystallinity

decreases further at 700 OC deposition temperature as suggested by a decrease in peak intensity

of convoluted W2N and W2C peaks. The XPS spectrum for deposition at 700 OC shows that there

is large amount of elemental C (diamond-like and graphitic) present in the films, further

supporting the assertion that C segregation results in a decrease in film crystallinity. At 750 oC

growth temperature, the peak intensity increases again but the increase is observed for the peak

corresponding to W2C suggesting that the crystallinity increase is due to the growth of W2C

crystalline phase (Figure 5-7).

5.5 Lattice Parameter

Since P-W2N and P-W2C have the same crystal structure with only a 2.7% lattice

mismatch, it is possible that a solid solution of significant extent is formed. As described

elsewhere, XRD peak shifts can be used to estimate the combined extent of the solid solution and

residual strain.m' The (1 11) peak of P-WNxCv or P-W2N and P-W2C has the highest intensity

and thus was used for calculating the lattice parameter. Figure 5-9 shows the change in lattice

parameter with deposition temperature for films deposited with 4 and 2a,b. For films deposited

with 4, the lattice parameter is 4.12 A+ for deposition at 500 OC, which is nearly identical to the

literature value of 4. 126 A+ for P-W2N.m7 As the deposition temperature increases from 500 to

650 oC, there is a concomitant increase in the estimated lattice parameter. Changes in the lattice

parameter can result from an increase in uniform strain in the film or a change in film









composition. Uniform strain can be introduced by differential thermal expansion. Since the

fi1ms grown from 4 are highly disordered and have small grains (< 50 A+, vide post), minimal

uniform strain is expected to exist in these films. Thus, the increase in lattice parameter between

500 and 650 oC is attributed to the increase in C content and decrease in N content. For film

deposited with 4 at 700 and 750 oC, the measured lattice parameter for the P-W2N peak is 4.13

and 4. 14 A+, respectively. The lattice parameters of a-W2C for film grown at 700 oC are: a = 2.97

A+ and c = 4.68 A+, while the values are: a = 2.92 A+ and c = 5.18 A+ for the film grown at 750 oC.

Comparison of films deposited with 4 and 2a,b shows that for films grown between 500

and 600 oC, those deposited with 4 have a slightly larger lattice parameter than those deposited

with 2a,b. This is consistent with the higher C content in the films deposited from 4. The lattice

parameter for films deposited from 4 and 2a,b increases with deposition temperature between

500 and 650 oC. This increase in lattice parameter value with deposition temperature is due to an

increase in C content with temperature for films deposited with both 4 and 2a,b.

5.6 Grain Size

As discussed in Chapter 3, the average grain size for polycrystalline materials can be

estimated using the Scherrer equation if it is assumed that XRD peak broadening is due to the

crystallite size distribution. Again, since the (111) reflection was most intense, it was used to

determine FWHM and estimate the change in grain size with temperature for films deposited

with 4 (Figure 5-10).

For films grown at 500 oC, the average grain size of crystallites for material deposited

from 4 is 25 A. Between 500 and 650 oC, the average grain size gradually increases from 25 S,

to 34 A+. Higher surface diffusivity of atoms at higher deposition temperature facilitates

nucleation and grain growth. At 700 and 750 oC, a two phase mixture of a-W2C and P-W2N

phases is formed. The average grain size for the a-W2C phase, calculated from the (101) peak, is









30 and 42 A+ for deposition at 700 and 750 oC, respectively. The average grain size for P-W2N

phase, calculated from (111) reflection, is 39 and 47 A+ for deposition at 700 and 750 oC,

respectively. The comparison of grain size for 4 and 2a,b shows that the average grain size for

films deposited with 4 is lower than that for films deposited by 2a,b over the entire deposition

temperature range.

5.7 Film Growth Rate

The growth rate for films grown from 4 was calculated from the measured film thickness

(X-SEM) and deposition time. Figure 5-11 shows X-SEM images for films grown between 400

and 750 oC in 50 oC intervals. The growth rate varied from 3 a/min to 24 8/min, with the

lowest growth rate observed at 400 OC and highest growth rate observed at 700 OC. There is an

exponential increase in film thickness between 400 and 600 oC. As the deposition temperature is

increased from 600 to 700 OC, there is a small increase in growth rate from 21 A+ to 24 A+. The

growth rate decreases to 17 A/lmin as the deposition temperature is increased to 750 oC. To

determine whether a further decrease in growth rate is observed at higher temperature, growth

was attempted at 800 OC. There was no hard film deposited at 800 OC, however there was a

carbon-like deposit observed on the substrate surface, which could easily be removed from the Si

substrate by a cotton swab.

Figure 5-12 shows Arrhenius plots of growth rate for deposition from 4 and 2a,b in the

presence of hydrogen. The growth rate for films deposited with 4 is lower than that for films

deposited with 2a,b between 400 and 650 oC. The Arrhenius plot for 4 is consistent with film

growth between 400 and 600 oC being surface reaction limited with a change to mass transfer

limited between 600 and 700 oC. The decrease in growth rate at 750 oC and no film growth at

800 oC is possibly because of homogenous decomposition of the precursor above 750 oC. While

the transition from surface reaction limited growth to mass transfer limited growth occurs









between 550 and 600 oC for 2a,b, 4 shows the same transition at 600 OC. The activation energy

for 4 calculated from the Arrhenius relationship for film growth using 4 is 0.54 eV, which is

significantly lower than 0.84 eV activation energy reported for 2a,b. The value of activation

energy for 4 is within the typical values between 0.5 and 1.0 eV observed for CVD growth in

surface reaction limited regime.

5.9 Atomic Bonding from XPS Measurement

Figure 5-13 shows the XPS measurements for films deposited from 4 at 400, 500, 600, and

700 oC and Table 3-7 (Chapter 3) summarizes the reference peaks of elemental C, elemental W

and different compounds of W, N, C and O obtained from the literature. Table 5-2 and 5-3

summarizes the peak parameters from peak fitting analysis done for the XPS peaks of W, C, N

and O. The deconvolution of W 4 f- 2 and 4f5/3 peaks was done by using the peak separation

between 4 f- 2 and 4f5/2 peaks of 2. 1 eV and peak area ratio 4 f- 2: 4f5/2 of 4:3. The deconvolution

of W 4f and 4p peaks for film deposited at 400 OC indicates the presence of three different

binding states ofW. The first W 3 f- : peak at 3 1.9 eV lies between the reference W peaks of

W2N (~33.0 eV) and W2C (~ 31.5 eV), indicating that the W atom is bonded to both C and N.

The second 4 f- 2 peak at 34.0 eV lies between the reference WO2 peak at 32.9 eV and W205 peak

at 34.6 eV, suggesting that the phase of tungsten oxide is WOx with the value of x between 2.0

and 2.5. The third W 4 f- 2 peak at 35.7 eV corresponds to the W present in WO3. As the

deposition temperature is increased from 400 to 700 OC, the intensity of the W peak

corresponding to WNxCy increases whereas the intensity of W peak corresponding to both WOx

and WO3 decreases.

For deposition at 400 OC, the N 1s peak is at 397.8 eV and corresponds well with the

reference N 1s peak for WNx film.119 As the temperature is increased to 500 OC, a broad

shoulder which is centered at 399.2 eV develops. This broad peak indicates that a small amount









of N is present in amorphous state most probably at grain boundaries.119 The relative intensity of

the amorphous N peak around 399.5 eV increases as the deposition temperature is increased

from 500 to 600 OC. This indicates that although the overall N content of film decreases as

deposition temperature is increased from 500 to 600 OC, the relative % of N present in the

amorphous state increases with respect to that bonded to W. The XPS spectra at 700 OC shows

no N 1s peak even though the AES measurements indicate the presence of N because of the poor

detection limit of XPS for detecting light elements (N) in a heavy element matrix (W).

For deposition at 400 OC, a small broad C 1s peak at 283.8 eV is observed in the XPS

spectrum. This peak position is slightly higher than the reference C 1s peak for W2C at ~ 283.5

eV. At 500 oC, the C 1s peak can be deconvoluted into two different peaks at 283.5 eV and

284.9 eV. The former peak corresponds to C bonded to W and the latter peak reflects C present

in amorphous phase. As the deposition temperature is increased from 500 to 600 OC, the peak

intensity of both carbidic and amorphous C increases. The C 1s spectrum for deposition at 700

oC can be deconvoluted into three distinct peaks at 283.6 eV, 284.5 eV, and 285.5 eV. The first

peak at 283.6 corresponds to carbidic carbon bonded to W. The second and third peaks at 284.5

and 285.5 eV reflect the presence of graphitic carbon (sp2 bonded C) and diamond-like C (sp3

bonded C). Previous studies on diamond-like C films have shown the presence of C as both

graphitic C with a peak at ca. 284.5 eV and diamond-like C with a broad peak at ca. 285.3

eV.192,193 The presence of graphitic and diamond-like C for films deposited at 700 OC shows that

besides phase separation of a-W2C, C separates from the WNxCv as graphitic and diamond-like



The O 1s peak for deposition at 400 OC can be deconvoluted into two peaks at 530.8 and

531.8 eV. The peak at 530.8 eV corresponds to O bonded to W in WOx or WO3. Since O is









present as 02-, the O peak does not show peak separation between the two different bonding

states in WOx and WO3. The other peak at 531.8 eV corresponds to O present in WNxCy.

Previous study on tungsten oxynitride has shown that XPS spectra for O 1s show a peak at 531.7

eV for O present in WNx matrix.203 The O 1s peaks for films deposited at 500, 600 and 700 oC

can be similarly deconvoluted to two peaks corresponding to O present as WOx or WO3 and O

present in the WNxCy matrix.

To summarize the XPS results, film deposited at 400 OC shows presence of the WNxCy

with a significant amount of WOx and WO3. At 500 oC, the bulk of the fi1m is WNxCy with

small amounts ofWOx and WO3. Additionally, film deposited at 500 OC shows the presence of

both C and N in an amorphous state probably at grain boundaries. Incidentally, the lowest

temperature at which the presence of amorphous C and N is observed is 500 oC, which is also the

temperature at which crystallinity is first observed. The film deposited at 600 OC consists of

WNxCy, a small amount ofWOx/WO3 together with amorphous N and C. At 700 oC, the film

shows the presence of WNxC,, graphitic and diamond-like C and a small amount of WOx/WO3-

5.8 Film Resistivity

Figure 5-14 shows the film resistivity at different deposition temperature for films grown

with precursors 4 and 2a,b. For depositions with 4 at 400 OC, the resistivity is 1150 CLR-cm.

The resistivity decreases to 980 CLR-cm for deposition at 450 oC due to higher W content and

lower C and N content of the film. Between 450 and 600 oC deposition temperature, the film

resistivity gradually increases from 980 to 6857 CLR-cm. This increase is attributed to a decrease

in W content of the film. The resistivity of films deposited at 700 and 750 oC is ca. 2400 CLR-cm.

The XRD spectra indicate the formation of a two phase mixture of P-W2N and a-W2C at these

temperatures and XPS measurement shows the separation of C to form graphitic C and diamond-

like C. The phase separation at 700 and 750 oC results in segregation of excess amorphous C as









a-W2C, graphitic C and diamond-like C resulting in decrease of film resistivity. The decrease in

resistivity as deposition temperature is increased from 600 to 650 oC could result from the onset

of phase separation at 650 oC, even though the phase separation is not clearly seen in the XRD

spectrum at 650 oC.

Comparison of resistivity for films deposited with 4 and 2a,b reveals that while the lowest

resistivity obtained with 2a,b is 750 CLR-cm at 4500 oC, the lowest resistivity obtained with 4 is

980 CLR-cm at 450 oC. The resistivity of films deposited with 2a,b increases throughout the

deposition temperature, whereas the resistivity of films deposited with 4 show a steep increase

between 450 and 600 oC followed by a decrease in resistivity between 600 and 750 oC because

of the phase separation of W2N and W2 -

5.9 Diffusion Barrier Testing

To determine the effectiveness of diffusion barrier deposited with 4, barrier films deposited

at 450 and 500 oC were coated with 100 nm PVD Cu. The thickness of films deposited at 450

and 500 was 45 and 55 nm, respectively. The Cu/barrier/Si stack was annealed in vacuum at

temperature ranging from 200 to 600 OC for 30 min to determine the temperature at which Cu

diffuses through the barrier film into the Si substrate. After annealing, three-point AES depth

profile and XRD measurements were done to detect copper diffusion through the barrier film.

Figure 5-16 shows the depth profile for a post-anneal Cu/WNxC,/Si stack for WNxCy films

deposited from 4 at 450 oC. For anneal at 200 OC for 30 min, the AES depth profile shows that

the Cu/WNxCy interface is similar to that for films that were not annealed (Figure 5-15),

suggesting that no bulk copper diffusion had occurred. As the anneal temperature is increased to

400 oC, there is slight mixing of the Cu/WNxCy interface, however, the copper has not diffused

through the barrier film. For 500 oC anneal, there is further diffusion of Cu in the WNxCy film,

but the barrier film is able to prevent complete diffusion of Cu through it. After annealing at 600









OC, the AES profile shows that there is complete diffusion of copper through the barrier film.

Intermixing is also observed for the WNxCy/Si interface. The depth profiling shows that 45 nm

barrier film of WNxCy deposited from 4 at 450 oC is able to prevent bulk Cu diffusion when

annealed at 500 OC for 30 min.

Figure 5-17 shows the XRD plots for films deposited from 4 at 450 oC before annealing

and after annealing at 200, 400, 500 and 600 oC. No peak(s) corresponding to WNxCy is

observed because film deposited at 450 oC is X-ray amorphous. As discussed in the literature

review chapter, the recrystallization temperature is important for diffusion barrier film because

diffusion along grain boundaries is believed to be the primary cause for barrier failure.

Amorphous W2N film deposited by PVD recrystallizes after annealing at 600 OC.182 However,

for WNxC, film deposited by 4, the film does not recrystallize even after annealing at 600 OC as

seen from the XRD pattern. This is expected since ternary compounds tend to have higher

recrystallization temperature as compared to binary compounds.

The pre-anneal XRD profile shows peaks at 31.1, 50.75 and 74.55 26 degrees and these

peaks correspond to Cu (111), Cu (200) and Cu (220) reflections, respectively. After annealing

at 200 oC, the intensity of Cu peaks increases because of grain growth in the Cu film. No peak

corresponding to CuSix compound is observed indicating that there is no bulk diffusion of copper

after annealing at 200 OC. As the annealing temperature is increased, the intensity of Cu peaks

further increases because of grain growth, however, no other peaks is observed besides peaks

corresponding to Cu and Si substrate suggesting that bulk phase copper diffusion did not occur

even after annealing at 600 OC. This result contradicts the AES measurement which shows that

Cu has diffused through the barrier film and into Si substrate after annealing at 600 OC. This

apparent contradiction occurs because XRD can detect CuSix compound formation only after









sufficient amount of the silicide has formed. If a small amount of Cu has diffused through the

barrier at the onset of barrier failure, XRD is not able to detect small quantities of CuSix whereas

AES can detect Cu in barrier film or Si even if the Cu content is ca. I at. % because of the higher

sensitivity of AES. As discussed in the Chapter 2, XRD and four point probe are the least

sensitive techniques for detecting copper diffusion through the barrier. Previous studies have

also shown that even though the barrier film has failed at lower annealing temperature, XRD

only detects the barrier failure at much higher anneal temperature.63

Figure 5-18 shows the 3 point AES depth profile for a post-anneal Cu/WNxC,/Si stack for

WNxCv films deposited from 4 at 500 OC. After annealing at 200 OC and 400 oC for 30 min, the

Cu-WNxCv interface is similar to that for films that were not annealed, suggesting that no bulk

diffusion of copper has occurred. For 500 oC annealing, slight mixing of Cu-WNxCv interface is

evident in the AES depth profile. Copper diffuses through the barrier film after annealing at 600

oC and significant intermixing of Cu and WNxCv is observed in AES depth profile. The AES

depth profiling shows that 55 nm barrier film of WNxCv deposited with 4 at 500 OC is also able

to prevent Cu diffusion when annealed at 500 OC for 30 min. Figure 5-19 shows the XRD

measurements on a Cu/WNxCv/Si stack before annealing and after annealing at 200, 400, 500

and 600 oC. Similar to film deposited at 450 oC, the film deposited at 500 OC do not show

recrystallization of the barrier film. The XRD spectra do not indicate formation of CuSix even

after annealing at 600 OC for the reason discussed above, even though the barrier film has failed.

5-10 Conclusions

It has been demonstrated that the mixed imido guanidinato complex

W(N'Pr)Cl3 ['PrNC(NMe2)N'Pr] (4) can be used in an aerosol assisted CVD system to deposit

WNxCv thin films. A comparison of effect of Cl4(CH3CN) and Cl3('Pr(NMe2)N'Pr ligands on

film properties is presented by comparing the film composition, crystallinity, lattice parameter,









grain size, growth rate and resistivity of films deposited with 4 and 2a,b. The AES spectra

showed the presence ofW, N, C and O in this material. When compared with films grown with

2a,b, the films grown with 4 have a higher C/W ratio and a similar N/W ratio. The O content of

films deposited with 4 is also significantly lower than that for films deposited with 2a,b,

especially at lower deposition temperature. The lowest growth temperature for 4 is 400 oC,

which is 50 oC lower than that for 2a,b. For both 4 and 2a,b, films grown at and above 500 oC

are crystalline. From the diffusion barrier application standpoint, films deposited with 4 at 450

and 500 oC were able to prevent copper diffusion after annealing at 500 OC for 30 min in

vacuum. Since the film deposited from 4 at 450 oC is amorphous, has the lowest resistivity of

980 CLR-cm and can prevent Cu diffusion after annealing at 500 OC, films deposited from 4 at

450 oC are candidates for diffusion barrier application.












Bond Length Bond Angle
(A) (degree)
W-N1 2.247(4) N1-W-N3 163.23(18)
W-N2 1.961(4) N2-W-Cl3 155.81(13)
W-N3 1.702(4) Cll-W-Cl2 167.30(5)
W-Cll1 2.3752(15) N2-W-N3 101.44(19)
W-Cl2 2.3819(16) N1-W-N2 61.88(16)
W-Cl3 2.3833(14) W-N1-C1 90.3(3)
C1-N2 1.399(6) N1-C1-N2 107.8(4)
C1-N1 1.294(6) W-N2-C1 100.0(3)
C1-N4 1.373(6) W-N3-C10 168.4(8)


Table 5-1. Selected bond distances (+)


and bond angles (0) for 4










Table 5-2. Analysis results from the deconvolution of the XPS peak of W for films deposited from 4 on Si(100) substrate at 400, 500,
600 and 700 oC
Deposition WNxCy WOx (2.0< x <2.5) WO3
temperature Peak % Area FWHM Peak % Area FWHM Peak % Area FWHM
Position Position Position
(eV) (eV) (eV) (eV) (eV)
400 W3 f- 31.95 20.4 1.61 34.01 17.0 1.84 35.74 15.9 2.00
W4f5/2 33.90 15.3 1.94 36.01 11.1 1.97 37.94 11.9 1.81
W4p3/2 37.10 3.8 2.15 39.27 2.8 1.64 41.04 1.8 1.20
500 W3 f- 31.89 40.4 1.36 33.51 8.7 1.52 37.61 4.8 1.96
W4f5/2 34.06 30.3 1.48 35.61 6.6 2.17 37.49 1.0 1.75
W4p3/2 35.51 6.3 2.02 38.81 1.3 1.27 41.11 0.6 2.16
600 W3 f- 32.01 38.2 1.30 33.72 7.8 1.61 35.31 7.4 1.78
W4f5/2 34.10 29.8 1.50 35.59 5.9 2.01 37.41 5.6 2.44
W4p3/2 37.40 3.0 1.98 38.89 0.9 1.41 40.71 1.4 3.45
700 W3 f- 32.05 36.3 1.28 33.54 9.9 1.61 35.33 7.5 2.14
W4f5/2 34.15 27.2 1.33 35.64 7.4 1.86 37.53 5.6 2.15
W4p3/2 37.45 2.7 2.04 38.94 2.0 1.22 40.83 1.4 2.10










Table 5-3. Analysis results from the deconvolution of XPS peaks of C, N and O for films deposited from 4 on Si(100) substrate at
400, 500, 600 and 700 oC
Deposition C 1s Nls 01s
temperature Peak % Area FWHM Peak % Area FWHM Peak % Area FWHM
Position Position Position

(eV) (eV) (eV) (eV) (eV) (eV)
400 Peak 1 283.83 100.0 2.29 397.81 100.0 1.83 530.85 15.9 2.00
Peak 2 -- -- -- -- -- -- 531.85 11.9 1.81
500 Peak 1 283.51 73.7 1.48 397.71 84.1 1.50 530.77 66.2 1.59
Peak 2 284.95 26.3 2.19 399.25 15.9 1.77 532.11 33.8 2.03
600 Peak 1 283.59 72.3 1.54 397.73 64.6 1.73 530.57 49.4 1.46
Peak 2 285.06 27.7 1.98 399.49 35.4 2.87 531.69 50.6 2.02
700 Peak 1 283.60 19.2 1.53 -- -- -- 530.77 52.4 1.81
Peak 2 284.50 44.2 1.64 -- -- -- 531.91 47.6 2.48
Peak 3 285.50 36.6 2.43---------









































Figure 5-1. Thermal ellipsoids diagram of the molecular structure of
W(N'Pr)Cl3 ['Pr(NMe2)N'Pr]. Thermal ellipsoids are drawn at 40 % probability, and
the hydrogen atoms have been omitted for clarity.202
















Cf


60-4
~- 2 a, b

50-
E
iT 40-


E
S40

S30




20


S30

S20

10


ic~O,~,


300 400 500 600
Deposition Temperature ("C)


700 800 300 400 500 600
Deposition Temperature ("C)


700 800


70
4
60 1--- 2a,b


50 0

S40 -




a20 -
10 -


4 O
-4 2a,b


E 40




U 20

8 1


P-O


0C
700 800 300


300 400 500 600
Deposition Temperature ("C)


400 500 600
Deposition Temperature ("C)


700 800


Figure 5-2. Composition of films deposited from 4 and 2a,b on Si (100) substrate at different

deposition temperature as determined by AES after 0.5 min of sputtering


-0 2 a,b





































-


0.25--



0.20-



0.15-



0.10-



0.05 -



0.00 -
30[


e4
- 4 -2a,b


e-- o


400 500 600 700


*4
- 4 2ab


400 500

Deposition


600

Temperature (oC)


Figure 5-3. N/W and C/W atomic ratios for films deposited from 4 and 2a,b on Si (100)
substrate at different deposition temperature as determined by AES after 0.5 min of
sputtering















Y i 1700 *C



1 ~4,- L~~-c~C~r 1600 06







400 oC


1000 800 600 400 200 0

Binding Energy (eV)


Figure 5-4. X-ray photoelectron spectroscopy patterns for films deposited from 4 at 400, 500,
600 and 700 oC on Si(100) surface
























178

















SDep. T, Film ThickneSS Si (400 KP



65 C, 316 nm








oC. 55 nm


I I5 II

20 40 60 80

20 Degrees

Figure 5-5. X-ray diffraction patterns for films deposited from 4 on Si (100) substrate





32 34 36 38 40 42 44 46 48

26 Degrees

Figure 5-6. Deconvolution of XRD peaks for films deposited from 4 at 700 OC on a Si (100)
substrate. The dashed line represents raw data, solid lines represent individual peaks
of (3-W2N and a-W2C USed for deconvolution and the bold solid line represents the
profile calculated from individual peaks.













rro 7510 "C






I

.- liil I




QIll


32 34 36 38 40 42 44 46 48

26 Degrees

Figure 5-7. Deconvolution of XRD peaks for films deposited from 4 at 750 oC on a Si (100)
substrate. The dashed line represents raw data, solid lines represent individual peaks
of P-W2N and a-W2C USed for deconvolution and the bold solid line represents the
profile calculated from individual peaks.















2wC (200) 1 p W,C (220)

IS) 1750 "C



el 650 "C
600 "C

550 oC

c, 500 oC

450 oC
400 "C

30 40 50 60 70

26 Degrees



Figure 5-8. Grazing incidence X-ray diffraction spectra for films deposited from 4 on Si (100)
sub state












4.30


S4.20-




X 4.15 --,



4.10-



4.05 111
450 500 550 600 650 700 750 800

Deposition Temperature (oC)

Figure 5-9. Lattice parameters for films grown from 4 and 2a,b on Si (100) substrate. The solid
line at 4. 126 A+ corresponds to the standard lattice parameter for P-W2N. The dotted
line at 4.236 A+ corresponds to standard lattice parameter for P-W2C (111). Error
bars indicate uncertainty in determination of peak position for lattice parameter
calculation.












90 -


80 -


70 -


60 -
50-




40 -


1 (P-WNxC, or P-WNx / cc-WCx phase)
1 (P-W,N phase)
1 (uc-W,C phase)


- 4 2ab (p-WNxC or p-WNx / a-WCxphase)









/


450


500


550


600


650


700


750


800


Deposition Temperature (oC)

Figure 5-10. Average grain size at different deposition temperature for films grown from 4 and
2a,b on Si (100) substrate





























































Figure 5-11. Scanning electron microscope images for films grown from 4 on Si (100) substrate
between 400 and 750 oC in 50 oC increments



185












- I


Deposition Tem peratu re (oC)
700 650 600 550 500


400


8 0 75;0


3.0


2 -

3.0 -


2.5 -


2.0 -






0.5 -


3 5


4
O 2a~b





No growt


1.0 1.1 1.2 1.3 1.4 1.5


10001T (K^1)


Figure 5-12. Arrhenius plot of log of film growth rate vs. inverse temperature for deposition
from 4 and 2a,b on a Si(100) substrate












Amorphous N--NiWxCN1









400 C



404 402 400 398 396 394 392

Binding Energy (eV)

O 1s
O in WNxC O in WOx/WOs







400 -C



536 534 532 530 528 526
Binding Energy (eV)


42 40 38 36 34 32 30 28
Binding Energy (eV)


290 288 286 284 282 280 278
Binding Energy (eV)


Figure 5-13. Deconvolution of the XPS peaks of W, N, C and O for film deposited from 4 at
400, 500, 600 and 700 oC on Si (100) substrate




























187













+- 4


16000~

14000 -

12000 -

10000 -
800-

8000 -
400-

6000 -


400


500 600 700

Deposition Temperature ( oC )


800


Figure 5-14. Change in film resistivity with deposition temperature for films grown on Si (100)
from 4 and 2a,b























I I.. I



Spute Tie(mn






interface. min














........ B
-- Nu










'10 20 30 40
Sputer Time (min)



D-





10"' 20 30 4
Sputter ime (min


I_ ..4 .1


10 20 30 40 0

Sputter Time (min)


-- 1 C o






) 10 0 30 40 0
Sputte Time(min)


Figure 5-16. Auger electron spectroscopy depth profiles of Cu (100 nm)/ WNxCv (45 nm)/Si
(100) stack for WNxCv film deposited at 450 oC (using precursor 4) and annealed in
vacuum for 30 min at A) 200 OC B) 400 OC C) 500 OC D) 600 OC.





















500 oC Anneal



400 oC Anneal



200 oC Anneal



LNo Anneal

20 40 60 80

20 (Deg rees)

Figure 5-17. X-ray diffraction plots for Cu/WNxC,/Si stack (WNxCy films deposited from 4 at
450 oC) before annealing and after annealing at 200, 400, 500 and 600 oC

























191
















1 ---- cu









D 10 20 30 40
Sputter Time (min)



I.,. .. c


1 Cu


I -
I~......- cu





Ci 10 20 30 40
Sputter Time (min)




D oc
--- N
-----
Cu
Is


10 20 30
Sputter Time (min)


Figure 5-18. Auger electron spectroscopy depth profiles of Cu (100 nm)/WNxC, (55 nm)/Si
(100) stack for WNxC, film deposited at 500 OC (using precursor 4) and annealed in
vacuum for 30 min at A) 200 OC B) 400 OC C) 500 OC D) 600 OC.


10 20 30 40 0
Sputter Time (min)





















500 oC Anneal



400 oC Anneal



200 oC Anneal



No Anneal

20 40 60 80

20 (Deg rees)

Figure 5-19. X-ray diffraction plots for Cu/WNxC,/Si stack (WNxCy films deposited with 4 at
500 oC) before annealing and after annealing at 200, 400, 500 and 600 oC

























193









CHAPTER 6
DEPOSITION OF WNxCv FROM W(N'Pr)Cl3 zPrNC(NMe2)N'Pr] : EFFECT OF NH3 CO-
REACTANT ON FILM PROPERTIES

Chapter 5 discussed the growth of WNxC, films from 4 and H2 CO-reactant. When 4 is

used with H2 aS CO-reactant, the films deposited by CVD are W rich and deficient in N. The

highest N content of 12 at. % is observed for deposition from 4 at 400 OC. To increase the film

N content, the effect of NH3 CO-reactant on deposition of WNxC, using 4 was investigated. The

effect of NH3 CO-reactant on lowest deposition temperature, film crystallinity, growth rate and

resistivity was also studied. Two different experiments were done, one using a mixture of NH3

and H2 aS CO-reactants (designated as 1A) and other using a mixture of NTH3 and N2 aS CO-

reactants (designated as IB). Table 6-1 shows the molar flow rate of reactants for 1A and IB.

Films deposited for 1A and IB are compared to those deposited with only H2 aS CO-reactant

(designated as IC) to understand the effect of NH3 (with or without H2) on film properties.

6.1 Film Growth Studies

Films grown by procedures 1A and IB were generally smooth and had a shiny metallic

surface with film color varying from golden to shiny black. The lowest temperature at which

appreciable film growth was observed for 1A and IB was 300 oC. The use ofNH3 (with or

without H2) l0WeTS the deposition temperature by 100 oC (two experimental increments).

Addition ofNH3 aS CO-reactant lowers the energy barrier for film growth resulting in film

deposition at lower temperature.

6.2 Film Composition

Figure 6-1 shows AES results for 1A, 1B and IC. For films deposited with co-reactants

NH3 + H2 (procedure 1A), the W content varies between 51 and 59 at. % between 300 and 650

oC. The N content in the film is relatively constant at ca. 18 at. % between 300 and 400 oC with

the highest value of 21 at. % observed for deposition at 450 oC. Between 450 and 750 oC, the N


194









content gradually decreases from 21 at. % to 10 at. %. The sources of N in the fi1m are 4, NH3

and benzonitrile. The decrease in N content with increase in deposition temperature could result

from cleavage of N containing ligands from the precursor, lower incorporation of N from NH3

and benzonitrile or volatilization ofN from the film at higher deposition temperature.

Volatilization of N from WNxCv fi1ms has been reported to occur at 700 OC,89 hence the decrease

in N content between 450 and 700 oC is due to decrease in N incorporation from precursor

and/or NH3. The C content of the fi1m monotonically increases from 11 at. % at 350 oC to 43 at.

% at 750 oC deposition temperature. The increased C content of the fi1m is due to incorporation

of C from the precursor and/or benzonitrile into the film. The O content of the film decreases

from 17 at. % at 350 oC to 4 at. % at 500 OC and remains constant at ca. 4 at. % between 500 and

750 oC. Oxygen can be incorporated into the film during growth or post-growth. The sources of

O during film growth are residual gases (oxygen and water vapor) in the reactor and O impurity

in the precursor (usually in the ppm levels). Oxygen can also be incorporated into the film due

to post-growth exposure to air. The XPS results videe infr~a) show that O in the film is

incorporated both during growth and post-growth. The O incorporated during film growth forms

WOx (2 < x < 2.5)/WO3 whereas the O incorporated due to post-growth exposure to air is

incorporated in the WNxCv lattice.

For films deposited with only NH3 CO-reactant (procedure IB), the W and O content of

film decreases and N and C content of film increases between 300 and 500 oC. Between 500 and

750 oC, the W and O content of the film remains relatively unchanged at ca. 39 at. % and 3 at. %

respectively, the N content of film decreases from 30 at. % to 15 at. % and the C content of the

film increases from 30 to 49 at. %. This observation suggests that at 500 OC deposition

temperature, there is a major change in mechanism of reaction for IB. Deposition temperature









higher than 500 oC triggers decrease in N content possibly due to cleavage of N containing

ligand(s) in the precursor or lower N incorporation from NH3.

Comparison of composition of films deposited for 1A, 1B and IC shows that the most

prominent effect of using NH3 aS CO-reactant (with or without H12) is the significant increase in

film N content throughout the deposition temperature range. In general, the film N content

shows a twofold increase for 1A and a threefold increase for IB as compared to IC. While

previous studies on 2a,b and 3a,b have shown an increase in film N content when N\H3 + H12 is

used as co-reactant, the work on precursor 4 shows that the film N content can be further

increased by using only NH3 aS CO-reactant.

6.3 Film Crystallinity

Figure 6-2 shows the XRD patterns for films deposited for 1A between 300 and 750 oC.

The XRD spectra for films deposited between 300 and 450 oC show no peaks attributable to the

film, but only peaks associated with the substrate (Si(200) Ka, Si(400) KP, and Si(400) Ku

reflections at 33.10, 61.75 and 69.200 26, respectively). The absence of other peaks in these two

spectra suggests that films deposited for 1A below 500 oC are X-ray amorphous. The XRD

pattern for the film deposited at 500 OC shows the emergence of crystallinity as evidenced by the

two broad peaks at 37.26 and 43.690. These peaks lie between the standard peak position of P-

W2N [37.74 26 for (111) phase and 43.85 o 26 for (200) phase and P-W2C [36.98 26 for (111)

phase and 42.89 o 26 for (200) phasem, indicating the presence of either the solid solution P-

WNxC, or a physical mixture of P-W2N and P-W2C. Since the intensity of the most intense

reflection (111) is small, the other peaks corresponding to (220) and (311) reflections at 63.55

and 75.050 are not clearly visible in the XRD spectra. Figure 6-3 shows the XRD patterns for

films deposited for IB between 300 and 750 oC. The XRD spectra for IB show no peaks


196









attributable to the film, but only peaks associated with the Si substrate suggesting that films

deposited are X-ray amorphous throughout the deposition temperature range.

The comparison of films deposited for 1A, 1B and IC shows that the films are amorphous

for deposition below 500 oC irrespective of the co-reactant used. But at and above 500 oC, 1A

and IC result in formation of polycrystalline phases while 1B fi1ms are amorphous. This

suggests that when H2 is USed as co-reactant for 1A and IC, the reaction mechanism at

temperature above 450 oC encourages the formation of crystallites, resulting in formation of

polycrystalline thin films. When only NH3 is USed as co-reactant for IB, the reaction mechanism

hinders the formation of crystallites resulting in deposition of amorphous films throughout the

deposition temperature range.

6.4 Lattice Parameter

Figure 6-4 shows the change in lattice parameter with deposition temperature for 1A and

IC. The (1 11) peak of P-WNxCv has the highest intensity and thus was used for calculating the

lattice parameter. For 1A, the lattice parameter is 4.17 A+ for deposition at 500 OC, which is

between the literature value of lattice parameter for P-W2N (4.126 A+) and P-W2C (4.236 A+). As

the deposition temperature is increased from 500 to 550 oC, the lattice parameter decreases to

4.15 A+. Since the lattice mismatch between P-W2N and P-W2C is Only 2.7 %, it is possible that

W, N and C form a solid solution. For a WNxCv solid solution, an increase in interstitial

concentration (C, N and/or O) in the non-metal sublattice would result in an increase in lattice

parameter while an increase in vacancy concentration in the non-metal sublattice would result in

a decrease in lattice parameter. Transition metal nitrides and carbides are known to have

appreciable vacancy concentration on both non-metal and metal sublattices.39 Since the

(C+N+O)/W ratio remains unchanged between 500 and 550 oC, the decrease in lattice parameter

suggests an increase in vacancy concentration for deposition at 550 oC. Between 550 and 700









OC, the lattice parameter gradually increases from 4. 15 to 4. 19 A+. In the same temperature range,

the (C+N+O)/W ratio increases primarily because of an increase in C content while the N and O

content of the film decreases. The additional C can be incorporated into the W sublattice in

carbidic form or outside the W sublattice in amorphous form. The increase in lattice parameter

between 550 and 700 oC suggests that there is a net increase in the amount of C present in the W

sublattice as the temperature is increased from 550 to 700 OC. As the deposition temperature is

increased from 700 to 750 oC, the lattice parameter decreases from 4.19 to 4.15 A+ even though

the C content of the film increases suggesting either a net decrease in amount of C present in W

sublattice because of phase separation of C to form graphitic/ diamond-like C videe infra) or an

increase in vacancy concentration.

Comparison of films deposited with procedures 1A and IC shows that for films grown at

500 oC, 1A results in a higher lattice parameter than IC because of the higher (C+N+O)/W ratio.

For deposition between 550 and 650 oC, 1A results in a higher lattice parameter than IC even

though the (C+N+O)/W ratio for 1A is smaller than that by procedure IC indicating that films

deposited for IC have a higher vacancy concentration in that temperature range. For deposition

between 700 and 750 oC, lC affords a lower lattice parameter than 1A because IC results in

phase separation of a-W2C CaUSing a decrease in C content in the W sublattice of WNxCv phase.

6.5 Grain Size

For polycrystalline materials, the XRD peak broadening could be due to crystallite size

distribution, instrumental broadening or uniform strain in the film. Line profile analysis can be

used to deconvolute the effect of crystallite size distribution from the effect of uniform strain on

XRD peak broadening. Line profile analysis of XRD spectra for 1A indicated that the peak

broadening was primarily due to crystallite size distribution and the effect of uniform strain on

peak broadening was negligible. If the XRD peak broadening is entirely due to crystallite size









distribution, Scherrer's equation can be used to calculate the average crystallite size. Since the

(111) reflection was most intense, it was used to determine FWHM and estimate grain size for

1A and IC. Figure 6-5 shows the average crystallite size for 1A and IC using Scherrer's

equation. For 1A, the average grain size is approx. 48 A+ for deposition between 500 and 650 oC.

Average grain size increases slightly to 55 A+ for deposition at 700 OC and drops again to 50 A+

for deposition at 750 oC. Overall, the average grain size measurement indicates that the films

deposited by 1A at and above 500 oC are nanocrystalline and the effect of deposition temperature

on average grain size is minimal. Comparison of films deposited using 1A and IC shows that

films deposited for 1A have a larger average grain size than those deposited for IC over the

entire deposition temperature range.

6.6 Film Growth Rate

The growth rate for 1A and IB was calculated from film thickness measurement using X-

SEM. Figures 6-6 and 6-7 show X-SEM images for films grown at different deposition

temperature for 1A and IB respectively. For 1A, the growth rate varied from 3 to 14 1/min

between 300 and 750 oC. Figure 6-8 shows the Arrhenius plot for films grown for 1A, 1B and

IC. For films grown between 300 and 550 oC by 1A, the growth rate increases logarithmically

with inverse of temperature, indicating that the growth is in the kinetic controlled growth regime

and surface reaction is the rate limiting step. As discussed earlier, the slope of an Arrhenius plot

can be used to calculate the value of Ea. For 1A, the apparent activation energy is 0.45 eV.

Between 550 and 750 oC, the Arrhenius plot for 1A shows a small negative slope. Growth for

1A between 550 and 750 oC is in the diffusion limited growth regime and the negative slope is

probably because of homogenous gas phase decomposition of the precursor, which leads to

decrease in growth rate with increase in deposition temperature. For IB, the Arrhenius plot

shows that film growth between 300 and 400 oC is in the kinetically controlled growth regime


199









and the apparent activation energy for film growth is 0.27 eV. Between 400 and 750 oC, the film

growth rate for IB is almost constant indicating that the growth is in the diffusion limited growth

regime.

Comparison of film growth for 1A, 1B and IC shows that the rates of films growth by IC

have the highest value for Ea of 0.54 eV while the growth for 1A has the lowest value of Ea of

0.27 eV. While the Ea value for IB is higher than 1A, the growth rate for IB is significantly

higher than that for 1A suggesting that the addition of H2 to NH3 significantly increases the film

growth rate. Overall, the selection of co-reactant(s) has a significant impact on the film growth

rate and the lowest deposition temperature for precursor 4. The temperature at which the growth

mechanism changes from kinetically controlled to diffusion limited is also different for 1A, 1B

and IC because of difference in mechanisms of film growth and the changes in hydrodynamic

properties for different co-reactant mixtures.

6.7 Atomic Bonding

Figure 6-9 shows the results of XPS peak deconvolution and Tables 6-2 and 6-3 show the

peak fit parameters obtained from deconvolution for films deposited by 1A. The reference peaks

for elemental C, elemental W and different compounds of W, N, C and O are listed in Table 3 -7

(Chapter 3). The deconvolution of W 4 f- 2 and 4f5 3 peaks was done by using the peak separation

between the 4 f- 2 and 4f5 2 peaks of 2. 1 eV and peak area ratio 4 f- 2: 4f5 2 of 4:3. The

deconvolution of W 4f and 4p peaks for film deposited by 1A at 300 OC indicates the presence of

three different bonding states of W. The first W 4 f- 2 peak at 32. 1 eV lies between the reference

W peaks of W2N (~33.0 eV) and W2C (~ 3 1.5 eV), indicating that the W atom is bonded to both

C and N. The second 3 f- o peak at 33.7 eV lies between the WO2 peak at 32.9 eV and W20s peak

at 34.6 eV, suggesting that the phase of tungsten oxide is WOx (2 < x < 2.5). The third W 4 f-

peak at 35.7 eV corresponds to W present in WO3. As the deposition temperature is increased


200









from 300 to 400 OC, the W peak corresponding to WNxCv increases whereas the intensity of W

peak corresponding to WOx and WO3 decreases, which is consistent with the AES results that

shows a decrease in O content of the film in the same deposition temperature range. For

deposition at 500 OC and above for 1A, the W peaks can be deconvoluted into only WNxCv and

WOx peaks and there is no evidence of the presence of WO3. Comparison of W peaks for 1A

and IC shows that while IC shows evidence of WO3 throughout the deposition temperature

range, 1A shows the presence of WO3 only below 500 oC suggesting that use of NH3 aS CO-

reactant suppresses the formation of WO3 phase.

The N peak for 1A at 300 OC can be deconvoluted into two peaks. The first N 1s peak at

397.7 eV corresponds well with the reference N 1s peak for WNx film.119 The second peak at

399.9 eV suggests that N is also present as amorphous N.119 Between 300 and 700 oC, the N

peak can be similarly deconvoluted into a dominant peak around 397.5 eV and a broad shallow

peak around 399.5 eV implying that the bulk of the N is bonded to W and a small amount of N is

present in amorphous phase.

Between 300 and 500 oC, the C 1s peak for 1A can be deconvoluted into two peaks around

283.6 eV and 285.4 eV. The former peak corresponds to C bonded to W (reference peak at

283.5 eV) and the latter peak reflect C present as amorphous carbon (reference peak between

284.5 and 285.2 eV). For deposition at 600 and 700 oC, the C 1s spectra can be deconvoluted

into three distinct peaks at 283.5 eV, 284.5 and 285.8 eV. The first peak corresponds to carbidic

carbon bonded to W. The second and third peaks at 284.5 and 285.5 eV reflect the presence of

graphitic carbon (sp2 bonded C) and diamond-like C (sp3 bonded C) respectively. Previous

reports have indicated that the XPS peaks for graphitic C and diamond-like C appear at ca. 284.5

eV and ca. 285.3 eV respectively. 192,193 The presence of graphitic and diamond-like C for films









deposited at 600 and 700 oC shows that C separates from the WNxC, phase as has been reported

previously for IC. While films deposited by IC show evidence of phase separation of C to form

graphitic and diamond-like C only at 700 OC, films deposited for 1A and IB show the initiation

of similar process at 600 oC suggesting that the use ofN\H3 CO-reactant (with or without H12)

lowers the initiation temperature for C phase separation.

The O 1s peak for deposition between 300 and 700 oC can be deconvoluted into two peaks

at ca. 530.7 and ca. 532.0 eV. The peak at 530.7 eV corresponds to O bonded as WOx or WO3.

The other peak at 532.0 eV corresponds to O present in WNxC,. As discussed Chapter 6, O

could be incorporated into the film either during film growth because of oxygen and water vapor

impurity in the reactor or after the film growth because of exposure to atmosphere. The presence

of WOx/WO3 phase suggests that residual gas in reactor and/or O impurity in the precursor

react(s) with the precursor to form WOx/WO3 during the film growth. The source of O bound to

W in WNxCy is most likely atmospheric O that diffuses through the film when the film is

exposed to atmosphere. In summary, deconvolution of the O 1s peak suggests that O is

incorporated in the film during growth as WO,/WO3 and post growth as O loosely bound to W in

WNxCy.

XPS results for 1A show that films deposited between 300 and 400 oC are WNxCy with

significant amount of WOx and WO3 along with small amounts of amorphous N and C. For

deposition between 400 and 500 oC, the bulk of the film is WNxCy with a small amount of WOx

and increased amounts of amorphous C and N. The film deposited at 600 and 700 oC consists of

WNxCy, a small amount of WOx along with graphitic and diamond-like C.

XPS measurements for IB (Figure 6-10, Tables 6-4 and 6-5) show that there is no maj or

difference in the bonding states of W, N and O for 1A and IB. The only major difference is seen









for the bonding state of C, especially for deposition below 500 oC. For deposition at 300, 350

and 400 oC, the C 1s peak for 1A shows that C is present in both carbidic and amorphous phases,

whereas the C 1s peak for IB shows that C is present in only carbidic form. For IB, the broad

amorphous C peak at 284.8 eV appears only at 500 OC deposition temperature. This observation

suggests that the use of only NH3 aS CO-reactant favors the deposition of carbidic C, particularly

at low deposition temperatures.

6.8 Film Resistivity

Figure 6-11 shows the variation of film resistivity with deposition temperature for films

grown using procedures 1A, 1B and IC. For 1A at 300 OC, the resistivity is 4182 CLR-cm. As

the deposition temperature is increased from 300 to 600 OC, the film resistivity gradually

increases to 8395 CIR-cm possibly due to increase in C content of the film. Film resistivity peaks

for deposition at 650 oC and decreases thereafter. Deposition for IB shows that low resistivity

(393 CIR-cm) films are deposited at 300 OC. For growth between 350 and 400 oC, there is a

sharp increase in film resistivity because of the decrease in W content of the film. Between 400

and 550 oC resistivity increases due to the increase in C and N content of the film. Above 500

oC, resistivity decreases for IB because of a decrease in N content of the film.

Comparison of film resistivity for 1A, 1B and IC shows that the co-reactant used for

deposition of WNxC, from 4 can have a significant effect on resistivity. When either only H2 Of

only NH3 arT USed as co-reactants, low resistivity films are deposited at low deposition

temperature. However, when both H2 and NH3 arT USed together for deposition, high resistivity

films are deposited throughout the deposition temperature range. A number of factors could

affect the film resistivity including metal to non-metal ratio, crystallinity, grain size, density and

bonding state of non-metals; the choice of co-reactants affect each of these factors due to

different reaction mechanisms for each co-reactant mixture.









6.9 Diffusion Barrier Testing

To determine the effectiveness of diffusion barrier film deposited by procedures 1A and

IB, barrier films deposited at 300, 350 and 400 oC were coated with 100 nm PVD Cu. Prior to

the deposition of Cu thin film, the barrier film was exposed to atmosphere for approximately 1-2

hourss. The Cu/barrier/Si stack was annealed in vacuum at 500 OC for 30 min. After annealing,

three-point AES depth profile and XRD measurements were done to detect copper diffusion

through the barrier film.

6.9.1 Barrier Testing for 1A

Figure 6-12 shows the depth profile of pre- and post-anneal Cu/WNxCv (1A)/Si stacks for

WNxCv films deposited at 300, 350 and 400 oC. The thickness of the film deposited at 300, 350

and 400 oC was 37, 75 and 75 nm respectively. For deposition at 300 OC, the pre-anneal depth

profile (Figure 6-12 A) indicates that the copper has already penetrated through the barrier into

the Si substrate. This is an artifact resulting from the 'knock-on' effect of sputtering videe

supra). The post-anneal profile (Figure 6-12 B) for deposition at 300 OC shows that the Cu has

penetrated deeper into the barrier film, indicating that the barrier film was unable to prevent Cu

diffusion. For deposition at 350 oC, the comparison of pre- and post-anneal depth profile (Figure

6-12 C and D) shows that copper diffuses through the barrier film after annealing. For

deposition at 400 OC, the pre-anneal depth profile shows sharp interfaces between Cu/ barrier and

barrier/Si. After annealing at 500 OC for 30 min, Cu has further penetrated into the barrier but

not completely, indicating that 75 nm WNxCv film (lA) was able to prevent Cu diffusion through

the barrier film after vacuum annealing at 500 OC. Intermixing is seen in the depth profile for

both Cu/ barrier interface and the barrier/ Si interface.

A number of factors could influence the diffusion barrier performance of WNxCv including

film density and composition. For 1A, the film composition is relatively constant for deposition


204









at 300, 350 and 400 oC. To determine the density of the films for different deposition

temperature, an indirect method is the comparison of their relative sputter rate. For films with

similar composition, denser films would have a lower sputter rate as compared to less dense

films and vice versa. Table 6-6 shows the sputter rate for the barrier film deposited by procedure

1A at 300, 350 and 400 oC. Comparison of the relative sputter rate for 1A shows that film

deposited at 300 and 350 oC have a higher sputter rate and hence lower density as compared to

film deposited at 400 OC. So, the better diffusion barrier performance of film deposited at 400

oC is primarily because of higher film density.

Figure 6-13 shows the pre- and post-anneal XRD measurements for films deposited by

procedure 1A. For 300 oC, the pre-anneal depth profile shows five peaks at 43.50, 50.70, 61.75,

69.30 and 74.60 2-6 degrees. The peaks at 43.50, 50.70 and 69.30 degrees correspond to

Cu(111), Cu(220) and Cu(220) phases whereas the peaks at 61.75 and 69.30 degrees corresponds

to Si(400) ka and Si(400) kp respectively. The XRD profile also indicates that PVD copper

deposition on WNxCv shows a preferential orientation of (111). This is highly desirable because

the Cu(111) phase has a better electromigration resistance as compared to other crystalline

phases. The post anneal XRD profile shows an increase in the intensity of peaks corresponding

to copper suggesting grain growth of Cu crystallites due to annealing. No peaks corresponding

to Cu3Si are observed in the post-anneal XRD profile even though the AES depth profile shows

that copper has diffused through the barrier film. This apparent contradiction is due to the poor

sensitivity of XRD in detecting small amount of Cu3Si crystallites. The pre- and post-anneal

XRD spectra for barrier film deposited for 1A at 350 and 400 oC show results similar to those of

film deposited at 300 OC. No evidence of Cu3Si is seen in post-anneal XRD spectra for barrier









film deposited at 350 and 400 oC suggesting that no bulk diffusion of Cu has occurred after

annealing.

6.9.2 Barrier Testing for 1B

Figure 6-14 shows the depth profile for pre- and post-anneal Cu/WNxC,/Si stacks for

WNxC, films deposited by procedures IB at 300, 350 and 400 oC. The thickness of WNxC, film

from deposition at 300, 350 and 400 oC was 12, 23 and 45 nm respectively. Since the films

deposited by IB are much thinner than those deposited by 1A, the 'knock-on' effect of sputtering

was anticipated to be worse. Etching off the Cu layer prior to AES depth profiling can eliminate

the artifact arising from the 'knock-on' effect. The Cu layer was etched-off using 50 % HNO3

solution before AES depth profiling. For deposition by procedure 1B at 300 OC, the post-anneal

profile shows that Cu has diffused through the barrier layer into the Si. Comparison of the

barrier/ Si interfaces for the pre- and post-anneal depth profiles shows no evidence of

intermixing at this interface. For deposition at 350 oC, a trace amount of copper can be seen in

the post-anneal depth profile indicating that the barrier film was not effective in preventing

copper diffusion. For deposition at 400 OC, the post-anneal AES depth profile shows that there is

no copper diffusion through the barrier film suggesting that the barrier film was able to prevent

Cu diffusion after annealing at 500 OC. No intermixing at the barrier/ Si interface is observed.

Overall, for both 1A and IB films, barrier films deposited at 300 OC are not able to prevent Cu

diffusion whereas barrier films deposited at 400 OC are able to prevent Cu diffusion after

annealing at 500 OC. To identify the factors) that make film deposited at 400 OC effective as

diffusion barriers, the sputter rate of films deposited at 300, 350 and 400 oC by procedure 1B

was calculated from the AES depth profile as discussed above (Table 6-6). The sputter rate for

films deposited by IB is expected to be higher than that for films deposited for 1A because of

lower W content in the films deposited for IB. Unlike films deposited for 1A, films deposited


206









for IB show variation in W content for deposition at 300, 350 and 400 oC, hence the sputter rates

for these films cannot be directly correlated to film density because film composition influences

sputter rates. As expected, the sputter rate for film deposited at 400 OC is lower than that that for

film deposited at 300 and 3 50 oC because of the higher non-metal (N+C+O) to W ratio for film

deposited at 400 OC. Hence, no conclusive evidence of effect of density on diffusion barrier

performance could be obtained from sputter rates for films deposited for IB. Another important

factor that influences diffusion barrier performance is the film composition. Film deposited at

400 oC has significantly higher N content as compared to films deposited at 300 and 350 oC.

One of the reasons for the better diffusion barrier performance of film deposited at 400 OC for IB

could be its higher N content.

Figure 6-15 shows the pre- and post-anneal XRD measurements for barrier films deposited

by procedure IB. Similar to Figure 6-13 for 1A, Figure 6-15 for IB shows no evidence of

copper diffusion for films deposited at 300, 350 and 400 oC. The only peaks observed

correspond to copper and Si, suggesting that no bulk diffusion of copper has occurred through

the barrier film after annealing. Similar to films deposited by 1A, films deposited by IB also

show preferential deposition of Cu(l111) phase when copper is deposited by PVD.

6-10 Film Conformality

As discussed in Chapter 2, the conformality of barrier film is of utmost importance for

deposition in high aspect ratio features. Since the film deposited by procedure 1B at 400 OC

showed good diffusion barrier performance and high N content, conformality test was done for

deposition at 400 OC with only NH3 aS CO-reactant. Deposition was done on patterned chips

provided by Samsung. The dielectric material in the test chip was FSG and the feature size was

0.2 Clm with aspect ratio of 2:1. Figure 6-16 shows the X-SEM image of the feature after

diffusion barrier deposition using 4 and only NH3 CO-reactant. The film thickness for deposition









at 400 oC on FSG dielectric is 5 nm, which is significantly lower than the 45 nm thick films

deposited on Si substrate at the same deposition temperature. This implies that nucleation on

FSG substrate for IB takes longer time than on Si substrate. The film is highly conformal

visually, however, a precise number for conformality could not be calculated because of the

insufficient resolution of the SEM images.

6-11 Conclusions

It has been demonstrated that the mixed imido guanidinato complex

W(N'Pr)Cl3 ['PrNC(NMe2)N'Pr] (4) can be used with N\H3 + H12 (procedure 1A) and N\H3 + N2

(procedure IB) to deposit WNxCv thin films in an aerosol assisted CVD system. The use ofNH3

lowers the minimum growth temperature for 4 by 100 oC. The use of NTH3 with or without H12 aS

co-reactant has significant effect on film composition, growth rate, crystallinity, grain size and

resistivity. Comparison of films deposited with procedures 1A, 1B and IC shows that the best

films in terms are deposited by IB because of the lower resistivity of films at low deposition

temperature, amorphous film deposition throughout the deposition temperature range and higher

N content. Diffusion barrier testing shows that films deposited at 400 OC for both 1A and IB are

able to prevent Cu diffusion after annealing at 500 OC for 30 min in vacuum. Films deposited for

IB at 400 oC also show good conformality in a 0.2 Clm 2:1 aspect ratio feature.









Table 6-1. Molar flow rates used for deposition of WNxCy thin film from precursor 4.
Reactant Molar flow rate (mol/min)
1A IB
4 1.16 x 10-6 1.16 x 10-6
Benzonitrile 6.47 x 10-4 6.47 x 10-4

H2 4.09 x 10-2
NH3 1.02 x 10-3 1.02 x 10-3
N2 -- 4.09 x 10-2


209










Table 6-2. Analysis results from deconvolution of XPS peaks of W for films deposited by procedure 1A on Si(100) substrate between
300 and 700 oC.

Deposition WNxC, WOx (2< x <2.5) WO3
temperature Peak Peak Peak
position % Area FWHM position % Area FWHM position % Area FWHM
(e V) (e V) (e V) (e V) (e V) (e V)
300 W3 f- 32.12 10.50% 1.51 33.71 15.20% 1.78 35.75 26.90% 2.29
W4f5/2 34.22 7.90% 1.70 35.81 11.40% 1.95 37.95 20.10% 2.23
W4p3/2 37.52 1.60% 1.52 39.11 2.30% 2.20 41.15 4.00% 2.57
350 W3 f- 31.91 18.70% 1.46 33.79 17.90% 1.99 35.61 15.60% 2.01
W4f5/2 33.91 14.00% 1.67 35.89 13.40% 2.21 37.75 11.70% 1.85
W4p3/2 37.31 2.80% 2.60 39.19 3.60% 1.72 41.01 2.30% 2.26
450 W3 f- 32.05 33.90% 1.39 33.88 14.60% 1.97 35.36 3.30% 1.51
W4f5/2 34.15 27.10% 1.68 35.98 10.90% 2.24 37.46 2.50% 1.71
W4p3/2 37.45 5.10% 2.34 39.28 2.20% 1.67 40.76 0.50% 1.53
550 W4 f- 32.00 41.40% 1.39 34.06 13.00% 1.97 -- -- --
W4f5/2 34.20 31.10% 1.56 36.06 9.10% 2.44 -- -- --
W4p3/2 37.40 4.10% 2.26 39.46 1.30% 2.49 -- -- --
650 W3 f- 31.99 32.60% 1.25 33.38 18.70% 2.05 -- -- --
W4f5/2 34.19 23.50% 1.28 35.48 15.00% 2.02 -- -- --
W4p3/2 37.39 6.50% 1.72 38.78 3.70% 1.96 -- -- --
750 W3 f- 31.98 33.10% 1.32 33.48 17.40% 1.84 -- -- --
W4f5/2 34.27 24.80% 1.48 35.58 13.90% 2.05 -- -- --
W4p3/2 37.38 8.30% 1.95 38.88 2.60% 1.50 -- -- --










Table 6-3. Analysis results from deconvolution of XPS peaks of C, N and O for films deposited by procedure 1A on Si(100) substrate
between 300 and 700 oC
Deposition C 1s Nls 01s
temperature Peak Peak Peak
position % Area FWHM position % Area FWHM position % Area FWHM
(eV) (eV) (eV) (eV) (eV) (eV)
300 Peak 1 283.59 58.30% 1.7 397.69 60.40% 1.64 530.9 64.00% 1.51
Peak 2 285.35 41.70% 2.22 399.97 39.60% 2.1 531.93 36.00% 2.12
350 Peak 1 283.43 59.50% 1.21 397.79 82.90% 1.56 530.65 64.80% 1.48
Peak 2 284.82 40.50% 1.42 399.77 17.10% 1.25 531.9 35.20% 1.71
400 Peak 1 283.54 73.30% 1.26 397.64 80.10% 1.58 530.65 64.80% 1.48
Peak 2 285 26.70% 1.64 399.39 19.90% 2.26 531.9 35.20% 1.71
500 Peak 1 283.48 73.20% 1.36 397.68 75.60% 1.63 530.61 63.40% 1.53
Peak 2 284.81 26.80% 1.67 399.75 24.40% 2.1 532.17 36.60% 1.98
600 Peak 1 283.43 55.50% 1.41 397.64 75.20% 1.69 530.63 62.60% 1.66
Peak 2 284.57 28.20% 1.61 399.42 24.80% 1.95 531.98 37.40% 2.05
Peak 3 286.19 16.30% 2.28 -- -- -- -- -- --
700 Peak 1 283.55 44.70% 1.63 397.73 79.30% 1.59 530.78 64.10% 1.62
Peak 2 284.6 28.50% 1.73 399.04 20.70% 2.1 532.06 35.90% 1.94
Peak 3 285.73 26.80% 2.3 -- -- -- -- -- --










Table 6-4. Analysis results from deconvolution of XPS peaks of W for films deposited by procedure 1B on Si(100) substrate between
300 and 700 oC.

Deposition WNxC, WOx (2< x <2.5) WO3
temperature Peak Peak Peak
position % Area FWHM position % Area FWHM position % Area FWHM
(e V) (e V) (e V) (e V) (e V) (e V)
300 W3 f- 31.82 17.40% 1.5 33.44 20.20% 2.17 35.4 14.10% 2.04
W4f5/2 33.92 13.10% 1.55 35.54 15.10% 2.17 37.55 11.30% 1.96

W4p3/2 37.22 2.60% 1.52 38.84 4.00% 1.46 40.7 2.10% 1.91
350 W3 f- 32.01 36.50% 1.59 33.71 9.40% 1.6 35.2 7.00% 1.72
W4f5/2 34.11 26.30% 1.51 35.81 7.10% 1.64 37.3 5.30% 1.88
W4p3/2 37.41 5.50% 2.19 39.11 1.90% 1.22 40.6 1.10% 2.02
400 W3 f- 31.89 38.50% 1.47 33.56 9.80% 1.95 35.5 4.30% 1.57
W4f5/2 34.04 28.90% 1.47 35.66 7.40% 1.94 37.6 3.20% 1.88
W4p3/2 37.29 5.80% 2.31 38.96 1.50% 1.15 40.9 0.60% 1.8
500 W3 f- 31.91 37.00% 1.28 33.41 15.70% 2.1 -- -- --
W4f5/2 34.11 27.70% 1.4 35.51 11.80% 1.98 -- -- --
W4p3/2 37.31 5.50% 1.68 38.81 2.40% 2.18 -- -- --
600 W3 f- 31.94 36.10% 1.29 33.39 16.50% 2.12 -- -- --
W4f5/2 34.14 27.10% 1.36 35.49 12.40% 2.01 -- -- --

W4p3/2 37.34 5.40% 1.84 38.79 2.50% 2.39 -- -- --
700 W3 f- 31.93 34.50% 1.22 33.24 18.20% 1.97 -- -- --
W4f5/2 34.13 25.80% 1.26 35.34 13.60% 1.84 -- -- --
W4p3/2 37.33 5.20% 1.81 38.64 2.70% 2.11 -- -- --










Table 6-5. Analysis results from deconvolution of XPS peaks of C, N and O for films deposited by procedure 1B on Si(100) substrate
between 300 and 700 oC
Deposition C 1s Nls 01s
Peak Peak Peak
temperature position % Area FWHM position % Area FWHM position % Area FWHM
(eV) (eV) (eV) (eV) (eV) (eV)
300 Peak 1 283.54 100.00% 1.53 397.55 71.50% 1.43 530.75 71.10% 1.5
Peak 2 -- -- -- 398.96 28.50% 1.96 531.97 28.90% 1.84


350 Peak 1 283.46 100.00% 1.57 397.58 83.30% 1.5 530.57 67.40% 1.48
Peak 2 -- -- -- 399.2 16.70% 1.52 531.85 32.60% 1.93


400 Peak 1 283.39 85.70% 1.55 397.5 87.20% 1.58 530.46 70.20% 1.58
Peak 2 284.85 14.30% 1.62 399.25 12.80% 1.73 531.85 29.80% 1.96


500 Peak 1 283.32 64.60% 1.33 397.5 82.70% 1.48 530.39 64.20% 1.56
Peak 2 284.72 35.40% 2.08 398.91 17.30% 1.58 531.64 35.80% 1.93


600 Peak 1 283.42 61.80% 1.5 397.49 69.00% 1.53 530.45 64.40% 1.54
Peak 2 284.5 25.80% 1.77 398.99 31.00% 2 531.89 35.60% 1.81
Peak 3 286 12.40% 1.5 -- -- -- -- -- --
700 Peak 1 283.33 45.70% 1.29 397.54 70.30% 1.39 530.33 57.50% 1.55
Peak 2 284.42 31.40% 1.57 398.82 29.70% 1.56 531.66 42.50% 1.86
Peak 3 285.93 23.00% 2.46 -- -- -- -- -- --










Table 6-6. Apparent sputter rate of barrier film measured by dividing barrier film thickness by
the time required to sputter the film during AES depth profiling
Deposition
Temperature
(oC) 1A IB
300 pre-anneal 74 80
post-anneal 82 68
350 pre-anneal 93 74
post-anneal 150 70
400 pre-anneal 65 94
post-anneal 60 88


214









































































Figure 6-1. Composition of films deposited by procedures 1A, 1B and IC on Si (100) substrate

at different deposition temperature as determined by AES after 0.5 min of sputtering


80












20 yN F


---H1+NHz(1A)
-4- Honly (iC)
--0-- NHgoaly(1B)


~l Hz+NH,(1A)

--0-- NHzonly(iB)


Z


E
-
u
.e
c40
.o
c
e
E
m
eZO
o
u


o
zoo


300 400 500 400

Deposition Temperature ('C)


700 800


700 180 200 300 400 500 (00

Deposition Temperature (*C)


200 300 400 500 600

Deposition Temperature (*C)




































10 20 30 40 50 60 70 80

20 Degrees


Figure 6-2. X-ray diffraction patterns for films deposited on Si (100) substrate by procedure 1A


216




































10 20 30 40 50 60 70 80

20 Degrees

Figure 6-3. X-ray diffraction patterns for films deposited on Si (100) substrate by procedure 1B












4.30


4.20





-1-0.0 i~~





4.10 -0.5
450 500 550 600 650 700 750 800

Deposition Tem peratu re (*C)

Figure 6-4. Lattice parameters for films grown by procedures 1A and IC on Si (100) substrate.
Error bars indicate uncertainty in determination of peak position for lattice parameter
calculation.





























~~ H2+ NH, (1A)
- H2 only (1C)


70 -


60 -



50-
40 -


30 -


20 -


c**


a


I
~_--Y


'10


450


500 550 600 650


700


750


800


Deposition Temperature (*C)

Figure 6-5. Average grain size at different deposition temperature for films grown by procedures
1A and IB on Si (100) substrate.


219
























































Figure 6-6. Scanning electron microscope images for films grown by procedure 1A at different
deposition temperature on Si (100) substrate





220
























































Figure 6-7. Scanning electron microscope images for films grown by procedure 1B at different
deposition temperature on Si (100) substrate






221










Deposition Temperature (oC)
750 700 650 600 550 500 450 400 350 300
111 I II I I I i I

eH, 2 N H, (1 A)
yH2 Only (1C)
-1 NH, only (1B)


"r~ ~" ~- "~o"~~ 1~90m~""Bgrowth regirne


I I I I
1.0 1 .2 1.4 1.6 1


Diffusion controlled
growth regime


10001T (K-1)

Figure 6-8. Arrhenius plot for growth using procedures 1A, 1B and IC for growth on a Si(100)
sub state





































402 400 398

Binding Energy (eV)


W in W0x(4f,,r +-I W in WNxCV









300 C


O in WNxC, TO in WOx/WO,




400 C









534 532 530 52

Binding Energy (eV)


c


I
n
L

s.
c
m

E


eAcmorphous N--
600 "C

500 "C



-400 oC

_350r "
300_"C


=N in WNC,


42 40 38 36 34 32 30 28 404

Binding Energy (eV)


I


396


Graphitic C

Diamond-likC






AlmorphousC Chi


h
Y
c







C
v,

E


- C in WNxC

700 "C


600 C



3500 C
400 "


290 288 286 284 282 280 278

Binding Energy (eV)


Figure 6-9. X-ray photoelectron spectroscopy measurements for films deposited by procedure

1A between 300 and 700 oC on Si (100) substrate









































223




















t--N in WNx v


W in W0x(4f i NC


4f ,W in O






500 C


Amorphous N





500 "C

400 "C

350 "C


A

c


m
Z
rr
L



ur
c
ar
E


C-6~P


30 28 404


42 40 38 36 34 32

Binding Energy (eV)




Graphitic C +-C in

Diamond-likte C i














290 288 286 284 282

Binding Energy (eV)


h
V)
e


m
L
rr
L
~cC


vr
e
o


O in WNxC, -+
700 *0


500'C
400 "C

350"C


+-0 in WOx/WO,


280 278


Figure 6-10. X-ray photoelectron spectroscopy measurements for films deposited for between

300 and 700 oC by procedure 1B on Si (100) substrate


































224


S400 398 396

Binding Energy (eV)


534 532 530 528

Binding Energy (eV)















20000

18000-
SH2 + NH, (lA)
E~ 16000 H2 Only (1C)
fi 400- 1- -~- NH3 only (l B)

3* 12000-



8000-




.r 2000 -7

20-



200 300 400 500 600 700 800

Deposition Temperature ( "C)

Figure 6-11. Change in film resistivity with deposition temperature for films deposited by
procedures 1A, 1B and IC on Si (100) substrate

























Cu- A
.--- ----- N






,WN C (J75nm) .




O 5 10 15 2 21
Sputter Timer (min)


Cu~,- -. 1
E ...... .
,-*W



SI


0 5 10 15 2
Sputter Time (min)



Cu --













0 5 10 15 20 25

Sputter Time (min)


Cu


B /_i
...... c


Cu
.... ....
F~-~


I j

s,
i
\I ~Ncc-rsnmr /
~-~LsW_


0 5 10 15 20 25 30 0 5 10 15 20 25 30


Sputter Timne (min)


Sputter Time (min)


Figure 6-12. Pre- and post-anneal AES depth profiles of Cu (100 nm)/ WNxCv /Si (substrate)
stacks for WNxCv films deposited by procedure 1A. A) Barrier deposition at 300 OC,
no annealing. B) Barrier deposition at 300 OC, annealing at 500 OC. C) Barrier
deposition at 350 oC, no annealing. D) Barrier deposition at 350 oC, annealing at 500
oC. E) Barrier deposition at 400 OC, no annealing. F) Barrier deposition at 400 OC,
annealing at 500 OC.


226


F








































0 10 20 30 40 50 60 70 80

20Degrees

Figure 6-13. Pre- and post-anneal XRD measurements of Cu (100 nm)/ WNxCy /Si (substrate)
stacks for WNxC, film deposited by procedure 1A at 300, 350 and 400 oC


































D



~~v~ nm)




-~L:.''''-"l''i,-~r~+h~
`"


1 2 3 4 5
Sprrttw Time (min)




F
o
WNCy (45 nm)
~ j

X


II 1 2
Sputter Tnme (min)


2
Sputter Time (min)


2 3
~Spurtter Time (min)


Spurtter Time (min) Spuitter Tlme (min)


Figure 6-14. Pre- and post-anneal AES depth profiles of Cu (100 nm)/ WNxC, /Si (substrate)
stacks for WNxC, film deposited by procedure IB. After vacuum annealing and prior
to AES depth profiling, the Cu layer in the Cu/WNxC, /Si stack was etched-off. A)
Barrier deposition at 300 OC, no annealing. B) Barrier deposition at 300 OC,
annealing at 500 OC. C) Barrier deposition at 350 oC, no annealing. D) Barrier
deposition at 350 oC, annealing at 500 OC. E) Barrier deposition at 400 OC, no
annealing. F) Barrier deposition at 400 OC, annealing at 500 OC.

















LDeposition -400 oC, 500"64 a oC

Deposition 400 "C, Nlo An leal





Deposition 350 oC, No Anel




Deposition 300 oC, No Anneal


20 40 60 80

2 0(Deg rees)

Figure 6-15. Pre- and post-anneal XRD measurements of Cu (100 nm)/ WNxCy /Si (substrate)
stack for WNxC, film deposited by procedure 1B at 300, 350 and 400 oC


229























































Figure 6-16. Scanning electron microscope images of WNxCy thin films deposited on FSG
dielectric. The feature size is 0.2 Clm with aspect ratio of 2:1






230









CHAPTER 7
DEPOSITION OF WNxCy FROM (CH3CN)Cl4W(NNMe2) FOR DIFFUSION BARRIER
APPLICATION

Chapters 4, 5 and 6 discussed two different approaches for increasing film N content. The

first approach used NH3 aS a CO-reactant with precursor 3 and 4 to increase the film N content.

The second approach used metal-organic precursor 4 which contains more N atoms per W atom

(Chapter 5). Following the same approach, (CH3CN)Cl4W(NNMe2) (5) has been used to deposit

WNxCy thin films for diffusion barrier application. Scheper et al.204 have demonstrated that

diorganohydrazido(2-) titanium complexes can be used for the deposition of titanium nitride thin

films by CVD. This chapter discusses the characterization of the films deposited from 5 using

CVD and evaluates the effectiveness of the deposited films in preventing diffusion of copper.


7.1 Precursor Synthesis

A Schlenk flask was charged with WCl6 (2.10 g, 5.30 mmol) and methylene chloride (40

ml). 1,1 -Dimethylhydrazine (0.4 ml, 5.3 mmol) was added to the flask under vigorous stirring at

-78 oC. After 10 min, the solvent was removed in vacuo while warming to room temperature.

After removing the solvent, acetonitrile (15 mL) was added to the flask and the mixture was

stirred for 30 min. The solvent was removed in vacuo and solids from the reaction were extracted

with 2x15 mL of methylene chloride. The combined extracts were filtered and the volume

reduced to 10 mL. The product was precipitated from the mixture by adding vigorously stirred

pentane (200 mL) maintained at 0 OC. The orange product 5 was filtered off as a microcrystalline

powder and dried in vacuo.

7.2 Film Growth by CVD

When CVD growth was attempted at 400 OC from 5 in an inert atmosphere with N2 aS

carrier gas, the resulting film contained W and O while no C or N was detected by AES.









However, when H2 WaS used as carrier gas at the same deposition temperature, the deposited film

contained W, N, C and O as determined by AES. Hence, the H2 CO-reactant is essential for

deposition of WNxCv thin film from 5. Appreciable film growth was observed for 5 at

deposition temperature as low as 300 oC when H2 WaS used as co-reactant. This temperature is

the lowest growth temperature among all the precursors 1 5 when only H2 is USed as co-

reactant. The film characterization and testing reported in this chapter were performed on films

deposited using 5 and H2 aS CO-reactant. Table 7-1 provides the molar flow rate of reactants used

in the MOCVD reactor. The molar flow rate of 5 was same as that used for deposition using

precursor 4. The color of the deposited film varied from golden brown at low deposition

temperature to grayish at higher deposition temperature.

7.3 Film Composition

Figure 7-1 shows AES results for films deposited using 5 between 300 and 700 oC. The

AES results showed that films deposited from 5 contained W, N, C and O in varying

concentration for deposition between 300 and 650 oC. The films deposited between 300 and 650

oC are W rich, with W content varying between 51 and 63 at. %. The W content of film

decreases to 34 at. % for deposition at 700 OC because of excessive incorporation of C in the

film. The N content in the film is 11 at. % for deposition at 300 OC. As the deposition

temperature is increased from 300 to 350 oC, the film N content increases dramatically to 24 at.

%, which is also the highest value of N content obtained for deposition from 5 throughout the

deposition temperature range. For deposition between 350 and 750 oC, the film N content

decreases from 24 at. % to 5 at. %. The sources ofN in the film are precursor 5 and/or

benzonitrile. The decrease in N content between 350 and 650 oC is due to a decrease in N

incorporation from the precursor and benzonitrile. For deposition at 700 OC, the AES

measurement shows that the film N content is below the detection limit ofAES. This indicates









that deposition temperature higher than 650 oC leads to cleavage of N containing ligand from the

precursor, resulting in deposition of WCx thin film.

The C content of the film remains constant at ca. 9 at. % for deposition between 300 and

400 oC. Between 400 and 600 oC, the C content of film monotonically increases from 9 at. % to

41 at. % and remains unchanged at 650 oC. The increased C content of the film is due to

increased incorporation of C from the precursor and/or benzonitrile into the film with an increase

in deposition temperature.m' Increasing the deposition temperature further to 700 OC causes a

sharp increase in film C content to 63 at. %. The O content of the film decreases from 19 at. %

at 300 oC to 15 at. % at 350 oC. In Chapter 6, it has been argued that the O in the film is

incorporated during growth as well as post-growth. The XPS results videe infra) show that O in

the film is incorporated both during growth and post-growth. For growth between 450 and 600

oC, the O content of the film decreases from 9 at. % to 2 at. % and remains constant at 2 at. %

between 600 and 700 oC.

Due to the presence of chlorine in 5, the possible presence of chlorine in the films was of

interest. As discussed earlier, XPS was used to detect the presence of Cl in the film because the

W NNN peak in AES spectra overlaps with the Cl LMM peak at ca. 180 eV. Figure 7-2 shows

the XPS data for films deposited using 5 between 300 and 700 oC. No peaks were observed for

either Cl 2s or Cl 2p3/2 at 270eV and 199 eV, respectively, confirming that the chlorine level in

the films was lower than the detection limit of XPS (ca. I at. %).

7.4 Film Crystallinity

Figure 7-3 shows the XRD patterns for films deposited from 5 between 300 and 700 oC.

The XRD spectra for films deposited between 300 and 500 oC show no peaks attributable to the

film, but only peaks associated with the substrate (Si(200) Ka, Si(400) KP, and Si(400) Ku

reflections at 33.10, 61.75 and 69.200 26, respectively). The absence of other peaks in these









spectra suggests that films deposited up to 500 OC are X-ray amorphous. The XRD pattern for

the film deposited between 550 and 650 oC shows the emergence of crystallinity as evidenced by

the two broad peaks that lie between the standard peak positions of P-W2N [37.74 26 for (1 11)

phase and 43.85 o 26 for (200) phase] and P-W2C [36.98 26 for (111) phase and 42.89 o 26 for

(200) phase],m" indicating the presence of either the solid solution P-WNxCv or a physical

mixture of P-W2N and P-W2C. Since the intensity of most the intense (1 11) reflection is small,

the other peaks corresponding to (220) and (311) reflections at 63.55 and 75.050 are not clearly

visible in the XRD spectra. For deposition at 700 OC, the two peaks at 37.20 and 42.55 o 26

correspond to P-W2C because the films contain only W, C and O as indicated by AES

measurement.

7.5 Lattice Parameter

Figure 7-4 shows the change in lattice parameter with deposition temperature for films

deposited from 5. The (1 11) peak of P-WNxCv or P-W2N and P-W2C has the highest intensity

and thus was used for calculating the lattice parameter. For film deposited at 550 oC, the lattice

parameter has a value of 4.13 A+, which is similar to the literature value of lattice parameter for P-

W2N (4.126 A+). As the deposition temperature is increased from 550 to 600 OC, the lattice

parameter value remains almost unchanged at 4. 12 A+ even though the C content of the film

increases by 9 at. % in the same temperature range. If the additional C were incorporated in the

W sublattice, the lattice parameter value would have increased. An observation to the contrary

suggests that the additional C is incorporated outside the W sublattice in amorphous state.

Between 600 and 650 oC, the lattice parameter increases from 4.12 to 4.16 A+. In the same

temperature range, the film composition remains relatively unchanged. The increase in lattice

parameter could be due to increase in interstitials (C, N and/or O) or a decrease in vacancies in

the W sublattice. As deposition temperature is increased to 700 OC, the lattice parameter









increases further to 4.18 A+. At 700 oC, the film is P-W2C but the lattice parameter of 4.18 A+ is

much lower than the standard lattice parameter of 4.236 A+ for P-W2C pOSsibly due to strain in

the film.

7.6 Grain Size

As discussed earlier, XRD peak broadening could be used to determine average crystallite

size for polycrystalline films using Scherrer' s equation. Since the (111) reflection was most

intense, it was used to determine FWHM and estimate average grain size. Figure 7-5 shows the

average crystallite size for films deposited from 5. The average grain size is 45 A+ for deposition

at 550 oC. Between 550 and 650 oC, the average grain size increases with an increase in

deposition temperature. With an increase in deposition temperature, the surface diffusivity of

adsorbed species increases resulting in an increase in grain growth. The average grain size

decreases to 30 A+ at 700 OC because of the formation of P-W2C instead of P-WNxCv. Overall,

the average grain size measurement indicates that the films deposited from 5 at and above 550 oC

are nanocrystalline with average grain size varying between 30 and 66 A+.

7.7 Film Growth Rate

The growth rate for films deposited from 5 was determined by dividing the film thickness

measured using X-SEM by deposition time. Figure 7-6 shows X-SEM images for films grown

at different deposition temperature from 5. The growth rate increased from 1 A/min at 300 oC to

4 8Jmin for deposition at 400 oC. Between 450 and 650 oC, the growth rate gradually decreases

from 13 A/min to 10 A/min. For deposition at 700 OC, the growth rate increased significantly to

38 A/min suggesting a change in growth mechanism at this temperature. Figure 7-7 shows the

Arrhenius plot for films grown for deposition from 5. Film growth rate increased exponentially

with temperature between 300 and 450 oC, indicating that the growth is in the kinetically

controlled growth regime and that surface reaction is the rate limiting step. Using the Arrhenius









equation, the apparent activation energy calculated for the kinetically controlled growth regime

is 0.52 eV. For film deposited between 450 and 650 oC, growth is in the diffusion limited

growth regime. Typically, growth in the diffusion limited growth regime shows a small positive

slope. The small negative slope in the Arrhenius plot for growth from 5 suggests that some

precursor is decomposed via homogenous reactions resulting in a decrease in growth rate in the

diffusion limited growth regime. The transition from kinetically controlled to diffusion limited

growth occurs between 450 and 500 oC.

7.8 Atomic Bonding from XPS

Figure 7-8 shows the deconvolution for the XPS peaks of W, N, C and O for films

deposited from 5 between 300 and 700 oC. The corresponding peak fitting parameters obtained

from deconvolution are listed in Table 7-2 and Table 7-3. The reference peaks of elemental C,

elemental W and different compounds of W, N, C and O are listed in Table 3-7 (Chapter 3). The

deconvolution of the W 4f and 4p peaks at 300 OC indicates the presence of three different

binding states of W. The first W 3 f- : peak at 3 1.8 eV lies between the reference W peaks of

W2C (ca. 31.5 eV) and W2N (ca. 33.0 eV), indicating that the W atom is bonded to both C and N.

The second 3 f- o peak at 33.4 eV lies between the reference WO2 peak at 32.9 eV and reference

W205 peak at 34.6 eV, suggesting that the stoichiometry of tungsten oxide is WOx (2 < x < 2.5).

The third W 4 f- 2 peak at 35.2 eV corresponds to W present in WO3. These results are similar to

those obtained for films deposited from 4 by procedures 1A, 1B and IC (Chapter 6). With a

increase in deposition temperature, intensity of the W peak corresponding to WOx and WO3

decreases, which is consistent with the AES results that shows a decrease in O content of the film

in the same deposition temperature range. For deposition at 700 OC, the W peaks can be

deconvoluted into WNxCv and WOx peaks without a contribution from WO3. The N peak for

deposition at 300 OC can be deconvoluted into two peaks. The first N 1s peak at 397.5 eV









corresponds well with the reference N 1s peak for N in WNx film.119 The second peak at 398.8

eV suggests that N is also present in amorphous state.119 Between 300 and 600 oC, the N peak

can be similarly deconvoluted to a dominant peak around 397.5 eV and a broad shallow peak

around 398.5 eV implying that the bulk of N is bonded to W and a small amount of N present in

amorphous phase. The film deposited at 700 OC shows no N 1s peak confirming that the film is

WCx.

Between 300 and 700 oC, the C 1s peak can be deconvoluted into two peaks around 283.4

eV and 285.0 eV. The former peak corresponds to C bonded to W (reference peak at 283.5 eV)

and the latter peak reflects C present in amorphous phase (reference peak between 284.5 and

285.2 eV). Between 300 and 500 oC, the C 1s peak deconvolution shows that the C in the film is

predominantly bonded to W with some C present in amorphous state. However, for deposition at

600 and 700 oC, the ratio of amorphous C to carbidic C increases dramatically suggesting an

increase in ligand and/ or benzonitrile decomposition and subsequent incorporation of

amorphous C in the film.

The O 1s peak for deposition between 300 and 700 oC can be deconvoluted into two peaks

at 530.5 and 531.7 eV. The peak at 530.5 eV corresponds to O bonded to W in WOx or WO3.

The other peak at 531.7 eV corresponds to O bonded to W in WNxCv. As discussed earlier, O

could be incorporated into the film either during film growth because of O and water vapor

impurity in the reactor or after the film growth because of exposure of film to atmosphere. The

presence of WOx/WO3 suggests that residual gas (oxygen, water vapor) in reactor and/or O

impurity in the precursor react with the precursor on the film surface during film growth. The

source of O bound to W in WNxCv is most likely atmospheric O that diffuses through the film

when the film is exposed to atmosphere. With an increase in deposition temperature, the peak









intensity of O decreases for both O present in WOx/WO3 and O loosely bound to W in WNxCv.

The O present in WOx/WO3 decreases with an increase in deposition temperature because of the

increase in film thickness with deposition temperature or densification of film. The peak

intensity of O loosely bound to W in WNxCv decreases with increase in deposition temperature

because of decreased O diffusion in the film from atmosphere as a result of denser and thicker

films deposited at higher temperatures. In summary, deconvolution of 1s peak of O suggests that

O is incorporated in the film during growth as WOx/WO3 and post growth as O loosely bound to

W in WNxCv.

XPS results indicate that films deposited at 300 and 400 oC are WNxCv with significant

amount of WOx and WO3 along with presence of small amounts of amorphous N and C. For

deposition at 500 and 600 oC, the bulk of the film is WNxCv with a small amount of WOx and

increased amount of amorphous C. The film deposited at 700 OC consists of WCx along with a

small amount of WOx and a significantly higher amount of amorphous C.

7.9 Film Resistivity

Figure 7-9 shows the variation of film resistivity with deposition temperature for films

deposited from 5. Films with the lowest resistivity of 844 CIR-cm are obtained for deposition at

300 oC. This value is considerably higher than bulk resistivities of pure W2N and W2C.

Previous study on ALD WNxCv has also shown that films with a low resistivity of 300 400

CIR-cm can be deposited at low deposition temperature (300 350 oC).145 A number of factors

influence the resistivity of transition metal nitrides and carbides including metal to non-metal

ratio, vacancies in metal and non-metal sublattice, bonding states of C and N, presence of

contaminants such as O, film crystallinity and porosity. The high film resistivity for film

deposited from 5 could be because of interplay between some or all of these factors. As the

deposition temperature is increased from 300 to 350 oC, the film resistivity increases sharply to









2250 CIR-cm possibly due to the increase in N content of the film. Between 3 50 and 400 oC,

there is a sharp increase in film resistivity because of the decrease in W content of the film.

Between 450 and 600 oC, film resistivity gradually decreases from 2800 to 1200 CIR-cm. The

film deposited at 700 OC shows the highest resistivity of 4400 CIR-cm even though AES results

indicate that the film is WCx, which has low bulk resistivity. The film deposited at 700 OC is C

rich and high resistivity of the film is believed to be due to the presence of excessive amorphous

C in the film as indicated by the XPS spectrum.

7.10 Diffusion Barrier Testing

To determine the effectiveness of diffusion barrier film deposited from 5, barrier films

deposited at 350 and 400 oC were coated with 100 nm PVD Cu. Prior to the deposition of Cu

thin film, the barrier film was exposed to atmosphere for approximately 1 2 hourss. The

Cu/barrier/Si stack was annealed in vacuum at 500 OC for 30 min. After annealing, three-point

AES depth profile and XRD measurements were done to detect copper diffusion through the

barrier film. While XRD measurements were done on the Cu/barrier/Si stack, the AES

measurements were done on a barrier/Si stack obtained after removing the Cu layer by etching

with dilute HNO3. The Cu layer was removed to prevent the knock-on effect of sputtering videe

supra).

Figure 7-10 shows the depth profile of pre- and post-anneal Cu/WNxC,/Si stacks for

WNxC, films deposited at 350 and 400 oC. The thickness of the film deposited at 350 and 400

oC was 50 and 60 nm respectively. For deposition at 350 oC, the pre-anneal depth profile

(Figure 7-10 A) shows the background signal for Cu when there is no Cu diffusion. The pre-

annealed stack has a sharp barrier/Si interface. The post-anneal AES depth profile (Figure 7-10

B) shows that the Cu signal is similar to that observed in the pre-anneal depth profile suggesting

that the barrier film was able to prevent Cu diffusion into the Si. The barrier/Si interface is sharp









in the post-anneal AES depth profile with no indication of mixing at this interface. The XRD

measurement of the pre- and post-annealed film stack shown in Figure 7-11 shows peaks

corresponding to Cu and Si with no indication of presence of Cu3Si. Thus, both AES and XRD

measurements indicate that 50 nm barrier film deposited at 350 oC was able to prevent Cu

diffusion after annealing at 500 OC in vacuum for 30 min. Furthermore, the XRD profile shows

that PVD copper deposition on WNxCv has a preferred orientation of (111). This is highly

desirable because the Cu(111) phase has a better electromigration resistance as compared to

other crystalline phases of Cu.

For deposition at 400 OC, the comparison of pre- and post-anneal depth profile (Figures 7-

10 C and D) shows that copper has not diffused through the barrier film after annealing. The

post-anneal depth profile shows a sharp barrier/Si interface indicating that no intermixing has

occurred at this interface. The XRD profile shown in Figure 7-11 also indicates that the post-

anneal film stack exhibits no peaks corresponding to Cu3Si. Thus, barrier film deposited at 400

oC is able to prevent Cu diffusion after annealing at 500 OC in vacuum. As with film deposited

at 350 oC, WNxCv film deposited at 400 OC also favors deposition of (111) oriented Cu film for

deposition by PVD.

7-11 Conclusions

It has been demonstrated that the tungsten hydrazido complex (CH3CN)Cl4W(NNMe2) (5)

can be used with H12 in an aroOSol assisted CVD system to deposit WNxC, thin films. The H12 CO-

reactant is essential for deposition of WNxCv films as deposition with 5 in inert atmosphere (N2

carrier gas) resulted in deposition of WOx films. The lowest growth temperature for 5 was 300

oC, which is the lowest among precursors 1 5. Films deposited with 5 consisted of W, N, C and

O as determined by AES and no Cl impurity was detected by XPS. The film N content was

significantly higher for films deposited from 5 as compared to films deposited from 1 4.










Amorphous film deposition was observed for deposition below 550 oC. Films were quite

resistive as compared to bulk resistivity of pure W2N and W2C. Diffusion barrier testing shows

that films deposited at 350 and 400 oC were able to prevent bulk Cu diffusion after annealing at

500 oC in vacuum for 30 min.









Table 7-1. Molar flow rates of reactants in CVD reactor for deposition from precursor 5
Molar flow rate
Reactant.
(mol/mmn)
Precursor (5) 1.16 x 10-6
H2 4.09 x 10-2
Benzonitrile (solvent) 6.47 x 10-4










Table 7-2. Analysis results from deconvolution of XPS peak of W for films deposited from 5 on Si(100) substrate between 300 and
700 oC.

WNxC, WOx (2 < x < 2.5) WO3
Deposition Peak Peak Peak
temperature position % Area FWHM position % Area FWHM position % Area FWHM
(e V) (e V) (e V) (e V) (e V) (e V)
300 W3 f- 31.76 27.70% 1.49 33.36 14.10% 1.53 35.18 10.80% 1.96
W4f5/2 33.86 20.80% 1.64 35.51 10.60% 1.8 37.28 8.10% 2.09
W4p3/2 37.16 4.20% 2.49 38.76 2.10% 1.26 40.58 1.60% 1.82
400 W3 f- 31.67 35.10% 1.49 33.63 13.90% 2.74 35.09 3.60% 1.55
W4f5/2 33.77 26.30% 1.51 35.73 10.40% 2.35 37.19 2.70% 1.62
W4p3/2 37.07 5.30% 3.31 39.03 2.10% 1.7 40.49 0.50% 1.69
500 W3 f- 31.6 41.50% 1.38 33.4 7.50% 1.93 34.87 5.30% 1.52
W4f5/2 33.7 31.10% 1.36 35.5 5.70% 1.88 36.97 4.00% 2.32
W4p3/2 37 4.10% 2.78 38.8 0.80% 1.32 40.27 0.50% 1.5
600 W3 f- 31.92 42.00% 1.35 34.18 5.70% 1.71 35.4 5.20% 1.3
W4f5/2 33.98 33.60% 1.39 36.28 4.30% 2.09 37.5 3.90% 3.33
W4p3/2 37.32 4.20% 2 39.58 0.60% 1.19 40.8 0.50% 1.5
700 W4 f- 31.76 39.50% 1.24 33.4 12.50% 2.38 -- -- --
W4f5/2 33.92 29.60% 1.34 35.5 9.40% 2.05 -- -- --
W4p3/2 37.16 7.10% 2.22 38.8 1.90% 1.68 -- -- --










Table 7-3. Analysis results from deconvolution of XPS peaks of C, N and O for films deposited from 5 on Si(100) substrate between
300 and 700 oC.

Deposition C 1s Nls 01s
temperature Peak Peak Peak
position % Area FWHM position % Area FWHM position % Area FWHM
(eV) (eV) (eV) (eV) (eV) (eV)
300 Peak 1 283.44 77.40 1.78 397.50 79.40 1.40 530.63 73.00 1.43
Peak 2 285.16 22.60 2.10 398.75 20.60 1.73 531.72 27.00 1.65
400 Peak 1 283.35 75.80 1.20 397.50 77.70 1.38 530.54 65.70 1.32
Peak 2 284.81 24.20 1.68 398.79 22.30 2.10 531.62 34.30 1.55
500 Peak 1 283.33 84.30 1.35 397.52 79.50 1.29 530.49 68.10 1.34
Peak 2 285.09 15.70 1.54 398.89 20.50 1.87 531.61 31.90 1.54
600 Peak 1 283.45 57.80 1.43 397.62 83.20 1.64 530.45 59.20 1.45
Peak 2 284.76 42.20 2.49 399.28 16.80 2.10 531.77 40.80 1.81
700 Peak 1 283.47 73.30 1.98 -- -- -- 530.57 63.30 1.66
Peak 2 285.11 26.70 2.70 -- -- -- 531.76 36.70 1.54














I


WV
-I- -C


60 -

0-1




0-1


40 -


I


L~ ~


200


300


400


700


500


600


800


Deposition Temperature (oC)


Figure 7-1. Composition of films deposited from 5 on Si (100) substrate at different deposition
temperature as determined by AES after 0.5 min of sputtering





















k ~-c, 1700 *C

L 600 *C


500 "C


400 0C


300 "C


1000 900 800 700 600 500 400 300 200 100 0

Binding Energy (eVr)


Figure 7-2. X-ray photoelectron spectroscopy measurements for films deposited from 5 on Si
(100) substrate at different deposition temperature


246


















% ..~, ... _3t~ Y 1700 oC

R 1650 oC





'B 500 oC
__ i _450 oC|

400 oC



300 "C

10 20 30 40 50 60 70 80

20 Degrees

Figure 7-3. X-ray diffraction patterns for films deposited from 5 between 300 and 700 oC on Si
(100) substrate













4.30




4.25- |-- C




S4.20-




4.15-

p-W,N


4.10
500 550 600 650 700 750

Deposition Temperature (oC)

Figure 7-4. Lattice parameters for films grown from 5 on Si (100) substrate. Error bars indicate
uncertainty in determination of peak position for lattice parameter calculation.





















S40




> 20





500 550 600 650 700 750

Deposition Temperature (oC)

Figure 7-5. Average grain size for films grown from 5 at different deposition temperature on Si
(100) substrate.


249


















ltl Clm--' 1 1400 "C I ltl um- II1 dr;n r I


Figure 7-6. Scanning electron microscope images for films deposited from 5 between 300 and
700 oC on Si (100) substrate


250


500"CI JclClm-











































Figure 7-6. Continued...









Deposition Tem perature (oC)


750 700 650 600 550
I I I


450 400


-- t







Diffusion controlled
growth regime


Kinetically controlled
growth regime


`I


10001T (K-1)

Figure 7-7. Arrhenius plot for deposition from 5 on a Si(100) substrate
















O in WO, :I W 4f, 4p
a ~(2
SO in WO k=-l- 0"


E X~ 600 "C







300 oC

42 40 38 36 34 32 30 28

Binding Energy (eV)


B
c


E




u,




288


286 284 282

Binding Energy (eV)


N 18 ~o in WNxc O1
B iAmorphous N N in WNxC ? O in WO, or WO

600 C 500 "C

500 oC 40 sooc

400 oC 30 "




300 oC

404 402 400 398 396 534 532 530 528

Binding Energy (eV) Binding Energy (eV)


Figure 7-8. X-ray photoelectron spectroscopy measurements for films deposited from 5 between
300 and 700 oC on Si (100) substrate












5000



0 4000





2000











200 300 400 500 600 700 800

Deposition Tem peratu re ( "C )

Figure 7-9. Change in film resistivity with deposition temperature for films deposited from 5 on
Si (100) substrate


254







































---w


WN.Ey (50 nm)


40 50 0 10 20 30

Sputter Time (mnin)


Si
r

I
I

I

4C~


10 20 39

Sputter Time (min)


40 50


O 10 20 30
Sputter Time (min)


40 50


10 20 30

Sputter Time (min)


40 50


Figure 7-10.


Pre- and post-anneal AES depth profile of Cu/WNxC, /Si (substrate) stack for
WNxC, film deposited from 5. After vacuum annealing and prior to AES depth
profiling, the Cu layer in the Cu/WNxC,/Si stack was removed by etching. A)
Barrier deposition at 350 oC, no annealing. B) Barrier deposition at 350 oC,
annealing at 500 OC. C) Barrier deposition at 400 OC, no annealing. D) Barrier

deposition at 400 OC, annealing at 500 OC.





















SDeposition 350 oC, 500 oG Anneal





10ostin 20 30, 40 060708






stc o Dxyfl epositedn frm5a 350 and 400 oC


256









CHAPTER 8
QUANTUM MECHANICAL CALCULATIONS FOR PREDICTING PRECURSOR
FRAGMENTATION

Chemical vapor deposition is a very complex process to model because it is a not an

equilibrium process. Moreover, the numerous reaction pathways accessible to the precursor

molecule both in gas phase as well as on the substrate surface make fundamental understanding

of CVD processes all the more challenging. Computational chemistry can be used to study

possible reaction pathway(s) for metal-organic precursors. In this chapter, homogenous

decomposition of precursor 5 has been investigated using density functional theory (DFT) and

statistical mechanics. Based on the calculation results, possible reaction pathways for precursor

5 have been proposed.


8.1 Computational Methodology

A brief description of density functional theory is presented in Appendix A. All the

calculations were done using Gaussian 03W software. For calculating binding energy of

transition metal compounds such as hydrides, the Hartree Fock theory is not very accurate until

quite high levels of theory are used to account for the electron correlation. In contrast, DFT has

been shown to predict geometries and binding energies in transition metal compounds more

accurately.205 The B3LYP functional, which has been quite accurate in predicting structure and

energy of transition metal compounds, was used for geometry optimization and energy

calculations. The calculations were done using a split basis set (LanL2DZ for W and 6-31G(d)

for all other elements). To reduce the computational time, the effective core potential LanL2DZ

was used for W.206 The molecular structures were visualized using Molden and gOpenMol.207,208

Thermodynamic properties were calculated using the statistical mechanics package in the

Gaussian 03W software.









8.2 Geometry Optimization

Figure 8.1 shows the structure of 5 obtained from X-ray crystallographic determination and

the computationally optimized geometry. As expected, the experimental and calculated

geometries of 5 are quite similar. Table 8.1 shows the comparison of selected bond lengths and

bond angles obtained from X-ray crystallographic measurement and computational geometry

optimization. The calculated structure of 5 shows octahedral coordination at the W center.

Comparison of experimentally obtained and calculated bond lengths listed in Table 8-1 show that

except for the W-N1 bond length, the bond lengths obtained from calculation are larger than

those obtained experimentally. The DFT calculations are done for gas phase species whereas the

experimentally obtained structure is for solid precursor. The differences in bond angles could

arise because of crystal packing forces that are not accounted for in gas phase DFT calculations.

8.3 Acetonitrile Cleavage

Previous study on la,b, 2a,b and 3a,b has shown that dissociation of the acetonitrile ligand

from the precursors is facile at growth temperature (> 450 oC).209 To study the cleavage of

acetonitrile ligand from precursor 5, AGf and AHf were calculated for 5, CH3CN and

Cl4W(NNMe2) (5a). The AG and AH values for cleavage of acetonitrile from 5 were then

calculated using Equations 8-1 and 8-2.

AG~ = AG (5a)+ AG (CH3CN) -AG, (5) (8-1)

AH = AHf (5a) + AH (CH3CN) AHf (5) (8-2)

Figure 8-2 shows the optimized structures for 5a and CH3CN and Table 8-2 lists selected

bond lengths and bond angles for Sa. The calculated values for AG and AH can be assumed to

be approximately equal to AG8 and AH8 because the transition states for endothermic reactions










should be product-like. The values for AG and AH for acetonitrile cleavage are 6.68 and 14.92

kcal/mol respectively. These values are slightly higher that those obtained for la 3a reported

previously,209 suggesting that the acetonitrile cleavage reaction for precursor 5 has a higher

energy barrier as compared to the other precursors.

8.4 Chlorine Cleavage

One of important findings from film growth characterization discussed in Chapter 7 is that

even though precursor 5 contains four Cl atoms per W atom, the film deposited using 5 showed

no evidence of Cl in the XPS measurement indicating that Cl incorporation in the film is below

the detection limit of XPS (ca. I at. %). Hence, possible reaction pathways that can lead to Cl

cleavage have been explored computationally. It should be noted that these calculations are done

for homogenous gas phase reactions and the thermodynamic properties are calculated at 298.15

K and 1 atm.

8.4.1 Reaction of 5a with H2

Two mechanistic pathways for the removal of Cl from Sa have been explored. The first

pathway involves loss of Cl2 from Sa by reductive elimination. The AH8 value for reductive

elimination of Cl2 from Sa is 95.92 kcal/mol (Figure 8-3). The magnitude of this value suggests

that the reductive elimination pathway is not accessible to Sa. It should also be noted that the

calculated values of AGo for reductive elimination product (RI-1) indicated that the triplet was

more stable than the singlet, a possible indication of spin crossover problems with reductive

elimination. The second pathway involved reaction of transition metal complex with H2.

Possible pathways for reaction of H2 with transition metals include oxidative addition,

coordination of H2 followed by transfer of an acidic proton and o-bond metathesis. However,

neither oxidative addition nor coordination of H2 is available for a do complex such as 5.









Previous work has shown that a transition state could be found for o-bond metathesis of H12 with

the Cl4W(N'Pr).209 The calculated AH8 value for the a bond metathesis of H12 with 5a was 35.55

kcal/mol, indicating that this pathway should be accessible under the growth conditions. The

AH8 calculated for 5a is very similar to the values reported previously for Cl4W(N'Pr).209

8.4.2 Reaction Pathways for Removal of a Chloride from MI-1

Figure 8-4 shows reaction pathways for reductive elimination and o-bond metathesis of

MI-1 to replace a second chloride ligand with hydride. For reductive elimination of HCI from

MI-1, the value of AH8 is 37.26 kcal/mol. This value is quite low as compared to those

previously reported for calculations on 2a.209 For o-bond metathesis, the reaction can proceed to

replace the chloride atom at either a cis or the trans position. The search for transition states for

both cis and transrt~t~rt~t~rt~t~rt~ configurations and subsequent calculation of~H8 showed that the values of

activation energy for both these reaction pathways are quite similar. The formation of cis isomer

MT-2 cis is slightly more favorable energetically as compared to the trans isomer MT-2 trans.t~t~rt~t~rt~t~rt~

Overall, the AH8 values for reductive elimination and o-bond metathesis for both cis and transrt~t~rt~t~rt~t~rt~

configurations are quite similar. This result is different from that for 2a, which showed that the

reductive elimination was energetically less favorable as compared to o-bond metathesis for

second chloride cleavage.

8.4.3 Reaction Pathways for Removal of Chloride from MI-2 and RI-2

Since the second chloride reaction discussed above could form MI-2 cis, MI-2 transrt~t~rt~t~rt~t~rt~ or RI-

2, possible reaction pathways for these intermediates have been explored. Figure 8-5 shows the

different reductive elimination and o-bond metathesis pathways for MI-2 cis and trans isomers.

Figure 8-6 shows the reductive elimination of HCI from RI-2. Similar to RI-2, the RI-3 structure

is more stable as a triplet than a singlet. The AH8 value for reductive elimination of HCI from

RI-2 was 43.07 kcal/mol. To get a better idea of reaction surface for pathways involving MI-2










cis, MI-2 trans and RI-2, the reactions that lead to the fourth chloride cleavage needs to be

studi ed.

8.4.4 Reaction Pathways for Removal of the Fourth Chloride from Sa

Possible reaction pathways for MI-3 and RI-3 are shown in Figure 8-7. The reductive

elimination of HCI from MI-3 has an energy barrier of 74.64 kcal/mol. The o-bond metathesis

from MI-3 has a smaller energy barrier (AH8 = 38.09 kcal/mol). The o-bond metathesis reaction

for RI-3 has a AH8 value of 48.26 kcal/mol.

Since the energy barriers for second chloride cleavage by o-bond metathesis and reductive

elimination are quite similar videe supra), a comparative study of the energy barriers for third

and fourth chloride cleavage reactions is required to ascertain which of these pathways would be

more favorable if the reaction were to proceed by sequentially removing all chlorides. Figures 8-

8, 8-9 and 8-10 shows the AH values with respect to the AH (5a) for different reaction pathways

for intermediates MI2-cis, MI2-transrtrt~rtrt~rtrt~r and RT2. Benchmarking the AH values to AH (5a) allows

the comparison of different reaction pathways corresponding to MI2-cis, MI2-trans and RT2

intermediates. The comparison of different reaction pathways passing through intermediates

MI2-cis, MI2-trans and RI2 shows that the removal of second chloride at the transrt~t~rt~t~rt~t~rt~ position by a

bond metathesis to form MI2-transrtrt~rtrt~rtrt~r intermediate is most favorable energetically, followed by

removal of third and fourth chloride also by a bond metathesis to form MI4. The difference

between the energy values for third and fourth chloride cleavage for reaction pathways through

intermediates MI2-cis and MI2-trans is small (ca. 5 kcal/mol), hence it is possible that reaction

pathway involving intermediate MI2-cis would also be accessible to 5a if the reaction were to

proceed via sequential elimination of all four chlorines in the gas phase.

The above mentioned calculation gives relative comparison of different pathways that can

lead to sequential removal of all Cl from the precursor via either a bond metathesis or reductive









elimination. Realistically, the values of activation energy for the removal of third and fourth Cl

atoms from Sa are high enough to preclude the availability of these pathways under the reaction

conditions (deposition temperature of 300 oC or higher).

8.5 Conclusion

Computational chemistry has been used to assess the reaction pathways for precursor 5.

The facile dissociation of acetonitrile ligand from 5, which has observed experimentally for

reaction in solution, is also observed for calculations done for gas phase reaction. Various

reaction pathways for Cl cleavage from Sa have also been explored. For the first Cl cleavage

from Sa, o-bond metathesis is energetically favored and the activation energy value suggests that

this pathway would be accessible to Sa. For the second Cl cleavage, both o-bond metathesis and

reductive elimination pathways have similar energy barriers. However the high activation

energy (ca. 50 kcal/mol) for the second Cl cleavage for both o-bond metathesis and reductive

elimination pathways indicates that these pathways may or may not be accessible to intermediate

MI-1. Because of very high energy barriers for the third and fourth Cl cleavages, these pathways

would not be accessible to 5a in the gas phase.

While the present work focuses on homogenous reactions for acetonitrile dissociation from

5 and sequential Cl cleavage from Sa, further work is needed to understand heterogeneous

reaction pathways of 5, 5a and other reaction intermediates. The investigation of heterogeneous

reactions requires computationally intensive calculations and is a topic for future research.










Table 8-1. Selected bond lengths (A+) and bond angles (0) for precursor 5 obtained from X-ray
crystallographic measurement and DFT calculation


W1 Cl 2.388 2.347(16)
W1 N 1.749 1.769(5)
N N2 1.289 1.271(8)
W1 -N3 2.268 2.224(7)
N2 Cl 1.467 1.43 8(7)
N W1 Cl 96.8 95.9(4)
N3 W1 -Cla 83.24 84.07(4)
W1 N -N2 178.0 180.0
a Average value for four equivalent chlorides.


Table 8-2. Selected bond lengths (A+) and bond angles (0) for 5a obtained from DFT calculation


Bond
W1 Cla
W1 Nl
Nl N2
C W1 -N1a
W1 N -N2


a Average value for four equivalent chlorides.


Bond


Calculated Experimental


Calculated
2.351
1.734
1.293
102.1
177.5

















































AH* 15.62 kcallmol
AG* 6.68 kcallmol
AS* 27.66 callmol-K<


AB


Figure 8-1. Geometry of 5. A) X-ray crystallographic determination. The thermal ellipsoids
diagram are drawn at 50 % probability. Hydrogen atoms are omitted for clarity. B)
Geometry optimization by DFT calculation


Figure 8-2. Cleavage of acetonitrile ligand from precursor 5. The figure shows optimized
geometries for 5a and CH3CN


CH.,CN

















-10.08 kcal/mol


95.92 kcallmol


High spin state
RI-1


RT-1


3 5.55 kcallmoli


-20.30 kcallmol


MT-1


MI-1


Figure 8-3. Calculated transition state and products for reductive elimination and o-bond
metathesis from Sa. M = metathesis, R = reductive elimination, T = transition state
and I = reaction intermediate


Cl.,


HCI














HCI


-6.97 kcal/mol


RI-2 (high spin)


RT-2


-20.04 kcallmol







-19.02 kcallmol


35.59
kcallmor






36.35
kcal/mol


HCI






HCI


MT2 -cis


M12 -cis


MT2 trans


Figure 8-4. Possible reaction pathways for chloride cleavage by reductive elimination and o-
bond metathesis from MI-1 intermediate. M = metathesis, R = reductive elimination,
T = transition state and I = reaction intermediate


37.26 kcal/mol


MI-1

S+ H, ~





MI 2-trans


-13.62 1Cl






62.82


-13.2 8HCI


RI3


RT3-trn

r55.69


MI 2-cis


M T3-trans


-20.71


M13


Figure 8-5. Possible reaction pathways for reductive elimination and o-bond metathesis from
MI-2 (cis and trans).rt~t~rtrt~t~rtrt~ M = metathesis, R = reductive elimination, T = transition state
and I = reaction intermediate


M T3-cis














43.07 kcallmol

+ H2


-13.62 kcallmol
HCI


RT3


Figure 8-6. Possible reaction pathway for reductive elimination from RI-2. R = reductive
elimination, T = transition state and I = reaction intermediate


RI-2 (high spin)


RI3(high spin)













48.26 kcallmol

+ H2


-29.51 kcallmol


-HCI


MT5


74.64 kcallmol


-5.46 kcallmol


RT4


6.29 kcallmol

-HCI


38.09 kcallmol


+ H2


MT"4


MI4


Figure 8-7. Possible reaction pathways for reductive elimination and/or o-bond metathesis from
MI-3 and RI-3. M = metathesis, R = reductive elimination, T = transition state and I
= reaction intermediate


RI3(high spin)


RI4 (high spin)















RT4
~1~8j~25 R14
iSSITs



WTTd
~,~\ Aa~r
BC~


~55~9T R1.1

I 85.84


' -


140


40 -


I
I
I
I
I

: MT~'I

I, \
r
Ir
5a
I
1


,DI
I\r
RT3-cr~

R13
~3~S~bWi '
,17419 r
,~' --7
II
I~


~:~S~P

tJ~1112


M1.1f


-


20


O


Reaction Coordinates



Figures 8-8. The A~H values with respect to the AH (5a) for reaction pathway through
intermediate MI2-cis. The energy values are in kcal/mol


R117-5
~ilglas
t

t
t


d
X





Reaction Coordinates


Figures 8-9. The A~H values with respect to the AH (5a) for reaction pathway through
intermediate MI2-trans.trt~rtrt~rtrt~r The energy values are in kcal/mol


-


140 -

_120
o

-4100 -
o


6j 0 -
Ss-

S4 0 -

20 1
0-


MBT5
,W2.5


RT4



MVT4


RT-1


4 *i


\RI
1 RT3~-trans 9



:: I -')
MT? trans : 3 ~'MI!.-*
.81.a ? 4

12d s


i I


5a Y


15.25












MTS
:TK25s

R1 4
gg5~


RT3


RI3


140 -

,120
o
100 -
-

80 -
S8-


I 40
-


20 -

0-


Reaction Coordinates


Figures 8-10. The A~H values with respect to the AH (5a) for reaction pathway through
intermediate RI2. The energy values are in kcal/mol


RT-1

I- S




M fT-1

t -









CHAPTER 9
CONCLUSION AND FUTURE WORK

The films deposited using 3a,b, 4 and 5 show many properties that are essential for

diffusion barrier application. First and foremost, the work demonstrates that WNxCy can be

effective in preventing Cu diffusion. Films deposited at low temperature from 4 and Sa,b are

able to prevent bulk Cu deposition after annealing at 500 OC for 30 min in vacuum. Secondly,

precursor 5 could be used to deposit films at temperatures as low as 300 oC indicating that

MOCVD is capable of depositing films at temperature acceptable for diffusion barrier

application. Third, the films showed good adhesion and were thermally and mechanically stable

after annealing upto 500 OC. Fourth, WNxC, film promotes the growth of (111) oriented PVD

Cu, which is important to obtain Cu films with high electromigration resistance.


This chapter discusses the additional experiments needed to evaluate the potential of

WNxCy deposition from metal organic precursors for diffusion barrier application.

9.1 Diffusion Barrier Testing

The diffusion barrier testing reported in the present work gives preliminary results on the

effectiveness of barrier film in preventing bulk Cu diffusion. However, these tests need further

refinement in three critical areas: film thickness control, contamination control and trace copper

detection under realistic testing conditions.

9.1.1 Film Thickness Control

The barrier films tested in present work used relatively thick films (20 80 nm) for

evaluating their efficacy as diffusion barriers. Future copper metallization would require barrier

films that are thinner than 10 nm. Atomic layer deposition shows the most promise for

deposition of ultra thin barrier films with excellent conformality. Since 4 and 5 have shown

promising results for films deposited by MOCVD, these precursors should be used in ALD mode









to meet the thickness and conformality requirements for future barrier films. Actual barrier

testing on these ultra-thin films is essential for providing accurate evaluation of the barrier

material as well as the deposition process.

There are a number of challenges that need to be addressed to deposit ALD films using

precursors 4 and 5. Since precursors 4 and 5 are solids with low vapor pressure, their delivery in

the reactor requires dissolution of the precursor in a suitable solvent such as benzonitrile. For

MOCVD growth, the precursor solvent mixture was delivered to the reactor using a nebulizer

system. The ALD system needs to be modified so that it can use the nebulizer system for

delivery of the precursor solution. For ALD, the nebulizer system should be capable of

withstanding 1 Torr operating pressure. The nebulizer system used in present work requires

relatively high pressure of 350 Torr. However such high pressures are not suitable for ALD, so

the nebulizer system also needs to be redesigned so that it is capable of operating at lower

pressures (~ 1Torr).

9.1.2 Contamination Control

AES studies done on films deposited from 4 and 5 have shown that films deposited at low

deposition temperature (< 500 oC) contain a significant amount of O contamination. In the

presence of a high level of contamination, the diffusion barrier performance of the film could be

significantly altered. The XPS studies have shown that the source of O is not only the residual O

in the reactor, but also post-growth O incorporated during exposure of the film to the

atmosphere. To reduce the O incorporated during film growth, it is essential to grow film in a

clean system. This might be possible in the new ALD reactor system, which could be purged

and kept under inert atmosphere so that the reactor is not exposed to atmosphere after every

experiment. The transfer chamber attached to the ALD reactor could then be used to load the

samples into the ALD system. To reduce the O incorporated due to post-growth exposure to









atmosphere, the barrier films need to be coated with Cu in situ. In situ Cu deposition would

eliminate or at least minimize the O contamination from atmosphere. The CVD reactor has been

connected to the ALD reactor through the transfer chamber so that in situ deposition of Cu is

now possible. Future experiments should be performed using this new system to ascertain the

'true' properties and diffusion barrier capabilities of the barrier film.

9.1.3 Realistic Diffusion Barrier Testing

The diffusion barrier testing reported in present work accelerated the diffusion process by

annealing the Cu/barrier/Si stack in vacuum at 500 OC. However, the driving force for diffusion

in actual ICs is not just temperature but also bias. Hence, a more realistic test to access the

diffusion barrier efficacy would be bias temperature stress. To detect Cu diffusion, the present

work used AES depth profiling and XRD measurements. These techniques are good for

detecting bulk Cu diffusion but not trace Cu diffusion. The detection of Cu should be done using

electrical characterization such as p'n junction diode because electrical characterization is able to

detect trace Cu diffusion through the barrier film.

9.2 Diffusion Barrier Integration

At present, numerous processes and materials have shown good promise for fulfilling the

future requirements of diffusion barrier thin film. However, significant work needs to be done

on integration of these barrier films into actual interconnect. While a barrier film might be

excellent in preventing Cu diffusion, there are additional requirements on the barrier film and

deposition process. The barrier film needs to have good adhesion to adj acent layers and that

include Cu, low-k dielectric and etch stop layers. While the present work reported film growth

only on Si substrate, additional work needs to be done to determine adhesion of the WNxC, film

to Cu, low-k dielectric and etch stop layers. The four point bend test is an excellent technique to

ascertain the adhesion of thin films quantitatively. These tests need to be performed on the









barrier film deposited on different substrates. The barrier deposition process also needs to be

compatible with the overall metallization deposition scheme. This includes low deposition

temperature and pin-hole free film deposition. Lastly, the barrier deposition process which

employs gaseous reactants should be compatible with ultra low-k dielectrics. Since ultra low-k

dielectric films are highly porous, the gaseous reactants used in CVD and ALD process could

diffuse through the dielectric film, thereby increasing its the dielectric constant and negating any

benefits obtained from adoption of ultra low-k dielectric film. The integration of CVD or ALD

with ultra low-k dielectric films is a challenge for future metallization schemes and pore sealing

techniques for ultra low-k films like plasma exposure need to be explored.









APPENDIX
DENSITY FUNCTIONAL THEORY

The energy of a multi-particle system, E, is given by the Schriidinger equation (Eq. A-1).

EYP = HY (A-1)

YP is the wave function and H is the Hamiltonian operator. The Hamiltonian operator in the

Schriidinger equation depends on the positions and atomic number of the nuclei and the total

number of electrons. The dependence of the Hamiltonian on the total number of electrons

suggests that a useful physical observable would be the electron density p, since, integrated over

all space, it gives the total number of electrons N.


N = p(r)dr A

The assignment of nuclear atomic numbers is also available from the density, since for each

nucleus A located at an electron density maximum rA,



= -2Z ,P(r A) (A-3)



Given a known density, one could form the Hamiltonian operator, solve the Schriidinger

equation, and determine the wave functions and energy eigenvalues. Hohenberg and Kohn

proposed two theorems that revolutionized the DFT methodology.210

* Hohenberg-Kohn existence theorem: the non-degenerate ground state density must
determine the external potential (the charges and positions of the nuclei), and thus the
Hamiltonian and the wave function.

* Hohenberg-Kohn Variational Theorem: the density obeys the variational principle.

Kohn-Sham self-consistent field methodology takes as a starting point a fictitious system

of non-interacting electrons that have for their overall ground state density the same density as

some real system of interest where electrons do interact. Since the electron density determines









the position and atomic numbers of nuclei, these quantities are necessarily identical in the non-

interacting and in the real systems. The energy functional is then divided into specific

components to facilitate further analysis.

E(p) = Tni[p(r)] + Vne[p(r)] + Vee[p(r)] + AT[p(r)] + AVee[p(r)] (A-4)

Tni is the kinetic energy of non-interacting electrons, Vne is the nuclear-electron interaction, Vee

is electron-electron repulsion, AT is the correction to the kinetic energy deriving from the

interacting nature of electrons and AVee includes all non-classical corrections to electron-electron

repulsion energy. The above equation can be rewritten as Equation A-5.





(A-5)

N is the number of electrons in the system and Exc is the exchange-correlation energy (a

parameter that accounts for the terms AT and AVee). The density is given by Equation A-6.


a = 1(X:,z, )(A -6 )


If orbitals X minimize E, then Equation A-7 can be used to calculate E.

h," Zi i i (A-7)

The Kohn-Sham one electron operator is defined by Equation A-8.

hK n cei Zk J' p 7') ,r + Vse (A-8)



V sc (A-9)


Vxe is the so-called functional derivative. For the determination of Kohn-Sham (KS) orbitals, a

basis set is used, similar to the Hartree Fock method. The KS orbitals are expressed within a









basis set of functions {Qi}. The individual orbital coefficients are determined by solution of a

secular equation (as in the Hartree Fock method).

Spin correction is made in DFT by using individual functionals of a and P densities. The

spin densities at any position are expressed in terms of i, the normalized spin polarization.


r(r) = (A-10)


A.1 Exchange-Correlation Functional (Exc)

The exchange-correlation functional accounts for:

* coulombic correlation energy associated with electron-electron repulsion
* quantum mechanical exchange which arises from antisymmetry requirement
* correction for classical self-interaction energy
* difference in kinetic energy between a fictitious non-interacting system and a real system.

In practice, most modern functionals do not try to compute (4) explicitly. The effect of (4)

is either ignored completely or it is incorporated using 'hole function' with kinetic energy

correction.

A.1.1 Exchange Energy Density

Hohenberg and Kohn showed that if the electron density varies extremely slowly with

position, then Exc(p) is accurately given by Equation A-11.

E;" cDA xcE(p)dr (A-1 1)

Exe (= Ex+ E ) is the exchange plus correlation energy per electron in a homogenous electron gas.

Slater proposed that the exchange hole about any position could be approximated as a sphere of

constant potential with a radius depending on the magnitude of the density at that position. With

this approximation, the Exis determined by Equation A-12.


ex =(P) =\3 IP4/3(r)dr (A-12)









The Slater approach takes the value of a = 1. The LDA (local density approximation)

approach takes a value of a = 2/3 while the X, model most typically uses a = %/. The LDA,

Slater and X, methods can be extended to account for spin polarization using the equation A-13.


F, [Pr), l = xOFP~r~l FxL~U\II f(1 + r)4 + (1 r)4/ 2A13
ExE~f),1= E ~lf)+IS ~lf)-E Elill 2(21' -1) (-3



Ex is the exchange energy for system of uniform electron gas with electrons of like spin and e

is the exchange energy calculated using equation derived by Slater. Systems that have spin

polarization (e.g. open shell systems) must use the spin polarized formalism and it is generally

referred to as 'local spin density approximation' (LSDA).

A.1.2 Correlation Energy Density

Even for a simple uniform electron gas, no analytical derivation has been reported.

However, by using Monte Carlo techniques, the total energy of uniform electron gases of several

different densities have been calculated to a very high accuracy.211 By subtracting the exchange

energy density for each case, the correlation energy density can then be calculated. The

functional is given by Equation A-15.

rs 2b 1 4c-bZ
In + tan
A rg +b~S~ +c J4c-b; 2 +bh

In +J ~ tan >
xo2 +bxo + c r, +b+c4c-b 2 +b~
(A-14)



4xip(r): (A-15)

A, xo, b, c are different empirical constants for i = 0 and i = 1. Vosko et al. proposed several

different fitting schemes, and two functional forms (VWN and VWN5) have been widely









used.212 LSDA calculations that employ a combination of Slater exchange and the VWN

correlation energy expression are referred to as using SVWN method.

A.2 Density Gradient Correction

The LSDA approximation assumes a uniform electron gas as the source of energy

expression. It is assumed that the exchange-correlation energy density at every position in space

for the molecule is the same as it would be for the uniform electron gas having the same density

as is found at that position. However, in a molecule, this approximation is not true. A plausible

way to increase the accuracy of the correlation functional for molecular calculations is to make it

depend not only on the local value of density but also on the gradient of density. This approach

is referred to as 'gradient corrected' or 'generalized gradient approximation' (GGA). The

correction term is added to the LDA functional.


EG~l[~) cDs~ x 4/c rVp (A-16)


The subscript x/c indicates that the same functional holds for either exchange or correlation.

A.2.1 Correction to Exchange Functional:

The most popular GGA exchange functional has been developed by Becke (usually

abbreviated by 'B').213 The 'B' functional has a single empirical parameter which is optimized

by fitting to the exactly known exchange energies of the six noble gas atoms He through Rn.

Several other exchange functionals that have been reported are CAM(B), FT9 and mPW.214-216

There is very little data available in literature on the comparison of these different functionals for

actual molecules.

A.2.2 Correction to Correlation Functional

Various functionals developed by researchers for GGA correction to the correlation

functional include PW86, PW91, B95 and LYP.217-220 The PW86 functional includes one










empirical parameter fitted for the neon gas atom. The PW91 and B95 functionals use a different

expression from the general expression described above and do not contain any empirical

parameter. LYP does not correct the LDA expression but computes the full correlation energy.

It contains four empirical parameters fitted to the He atom. Of all the correlation functionals,

LYP is the only one that provides an exact cancellation of the self-interaction error in one-

electron systems. The LYP functional has proved more robust for systems that are more prone to

over delocalization because of its effective cancellation of self-interaction error.

The common nomenclature of expressing the functionals used in calculation is combining

the acronyms for correlation and exchange functionals. For instance, BLYP calculation

combines Becke's GGA exchange with the GGA correlation functional of Lee, Yang and Parr.

In literature, the most widely used combinations are BP86, B3LYP and BPW91. Results from

these three levels of theory are fairly comparable.

A.3 Similarity Between Hartree-Fock (HF) and DFT Methodologies

The two main similarities between HF and DFT techniques are:

* Mathematically, both use basis sets to obtain orbitals
* The kinetic energy and nuclear attraction components of matrix elements are identical.

There is a key difference between HF and DFT methodologies. Density functional theory

contains no approximations, it is exact. It is only required to know Exe as a function of p. Since

the exact nature of Ex is not known, the relevant equations are solved approximately. In

contrast, HF is a deliberately approximate theory.

A.4 Comparison of DFT to Molecular Orbital Theory205

A.4.1 Computational Efficiency

The formal scaling of DFT has been noted to be in principle no worse than N3, where N is

the number of basis functions used to represent KS (Kohn-Sham) orbitals. This is better than HF









which scales by a factor of N4. The scaling factor refers to how computational time increases

with the number of basis functions. For programs that use the same algorithms to carry out HF

and DFT calculations, the cost of a DFT calculation on a moderately sized molecule (~ 15

atoms) is double that of the HF calculations with the same basis set. However, if the program is

optimized for DFT, the calculation cost is much lower for DFT as compared to HF method.

A.4.2 Energetics

For a given average level of accuracy, hybrid DFT methods are the most efficient in

predicting atomization energies as compared to more expensive methods. Of the currently

available DFT models, GGA models offer a maj or improvement over the older LSDA model.

Moreover, the P86 functional should be avoided. There is no clear preference amongst the

remaining functionals, other than noting that the hybrid functional tends to perform better than

pure functionals.

In calculating binding energies in transition metal compounds such as hydrides, the

molecular orbital method is not very accurate until quite high levels are used to account for the

electron correlation. In contrast, DFT has been shown to predict binding energies in transition

metal compounds more accurately.

A.4.3 Geometries

Analytical first derivatives are available for almost all density functionals, and as a result

geometry optimization can be carried out with facility. The performance of most functionals is

usually quite good when it comes to predicting minimum energy structures. Accuracies in bond

angles for DFT average about 1 degree, the same as is found for HF and MP2.

For transition metals, DFT has proven to be a tremendous improvement over HF and

post-HF methods, particularly when the metal atom is coordinatively unsaturated. The narrow

separation between filled and empty d-block orbitals typically leads to enormous non-dynamical









correlation problems with an HF treatment, and DFT is much less prone to analogous problems.

Even in cases of a saturated coordination sphere, DFT methods typically significantly

outperform HF and MP2.

Density functional theory has shown weakness in predicting geometries for van der Waals

complexes and ionic complexes. Hydrogen bonds are somewhat too short as a rule, and most

charge transfer complexes have their polarities overestimated.

A.5 The Future of DFT

The DFT methodology has become very popular for various calculations in quantum

chemistry. Density functional theory has the advantage of allowing for correlation effects to be

included in a calculation that takes roughly the same time as a HF calculation, which does not

include correlation. However there are some drawbacks of the DFT method and these are the

areas that are currently being actively investigated. There is no systematic way of improving the

DFT calculations. There is no clear roadmap for the systematic improvement of the Exe

functionals. It is certain that a lot of research effort in the future is going to be on the

formulation of these functionals.

Density functional theory also has not been widely accepted as the method of choice for

excited states. The development of DFT formalisms to handle excited states remains a subject of

active research. Since DFT uses approximate functionals, KS DFT is not variational and can

yield an energy below the true ground state energy. Calculations with gradient corrected

functionals are size consistent. The currently used KS DFT functionals often do not give good

results for activation energies of reactions. Although KS DFT yields good results for most

molecular properties with the presently available functionals, KS DFT cannot match the

accuracy that methods like CCSD(T) and QCISD(T) can achieve. Of course, CCSD(T) and

QCISD(T) are limited to dealing with small molecules, whereas DFT can handle rather large









molecules. In summary, DFT has emerged as one of the promising fields in the field of quantum

calculations. It is most certainly going to be an area of active research in the years to come.









LIST OF REFERENCES


P. Van Zant, M~icrochip Fabrication: A Practical Guide to Semiconductor Processing,
(McGraw-Hill, 2000).

2M. T. Bohr, "Interconnect Scaling The real limiter to high performance ULSI,"
Proceedings of IEEE International Electron Devices Meeting, 241 (1995).

3 S. P. Murarka, I. V. Verner, and R. J. Gutmann, Copper-Fund amentald~~~dd~~~d M~echanisms for
Microelectronic Applications (John Wiley and Sons, New York, NY, 2000).

4 R. I. M. Dang, and N. Shigyo, IEEE Electron Devices Lett. EDL-2, 196 (1981).

SS. P. Jeng, R. H. Havemann, and M. C. Chang, Mat. Res. Soc. Symp. 337, 24 (1994).

SG. H. Schnable, and R. S. Keen, Proceedings of the IEEE 57, 1570 (1969).

SR. V. Penny, Physics and Chemistry of Solids 25, 335 (1964).

SK. N.Tu, J. Appl. Phys. 94, 5451 (2003).

SI. Ames, F. M. d'Heurle, and R. E. Horstmann, IBM J. Research and Development 14,
461 (1970).

10 M. Sakimoto, T. Fujii, H. Yamaguchi, and K. Eguchi, Proceedings of the IEEE Reliability
Physics Symposium, Las Vegas, NV, 333, April (1995).

11 K. Maex, M. R. Baklanov, S. H. Brongersma, and Z. S. Yanovitskaya, J. Appl. Phys. 93,
8793 (2003).

12 M. Michael, R. E. Todd, J. -H. Zhao, C. Hu, T. Cho, and P. S. Ho, Annual Review of
Materials Science 30, 645 (2000).

13 F. Iacopi, C. Zistl, C. Jehoul, Z. S. Tokei, Q. T. Le, A. Das, C. Sullivan, G. Prokopowicz,
D. Gronbeck, M. Gallagher, J. Calvert, and K. Maex, Microelectron. Eng. 64 351
(2002).

14 A. A. Istratov, C. Flink, and E. R. Weber, Physica Status Solidi B: Basic Research 222,
261 (2000).

15 E. R. Weber, Appl. Phys. A30, 1 (1983).

16 A. A. Istratov, H. Hedemann, M. Seibt, 0. F. Vyvenko, W. Schroter, T. Heiser, C. Flink,
H. Hieslmair, and E. R. Weber, J. Electrochem. Soc., 145, 3889 (1998).









'7C. Flink, H. Feick, S. A. McHugo, W. Seifert, H. Hieslmair, T. Heiser, A. A. Istratov, and
E. R. Weber, Physical Rev. Lett. 85, 4900 (2000).

IsB. J. Howard, and C. Steinbruchel, Appl. Phys. Lett. 59, 914 (1991).
19 D. Shamiryan, T. Abell, F. Iacopi, and K. Maex, Mater. Today 7, 34 (2004).

20 L. Peters, 'Alers visits interconnect reliability concerns', Semiconductor Intemnational
S ept. (200 5), http:.//www.reed-el ectroni cs. com/semi conductor/arti cl e/
CA6252222?pubdate=9%2Fl1%2F2005, last accessed May (2007).

21 C. L. Gan, C. V. Thompson, K. L. Pey, W. K. Choi, H. L. Tay, B. Yu, and M. K.
Radhakrishnan, Appl. Phys. Lett. 79, 4592 (2001).

22 C. -K. Hu, K. L. Lee, D. Gupta, and P. Blauner, Mat. Res. Soc. Symp. Proc. 428, 43
(1996).

23 C. -K. Hu, L. Gignac, E. Liniger, B. Herbst, D. L. Rath, S. T. Chen, S. Kaldor, A. Simon,
and W. -T. Tseng, Appl. Phys. Lett. 83, 869 (2003).

24 C. -K. Hu, L. Gignac, and R. Rosenberg, Microelectron. Reliability 46, 213 (2005).

25 E. G. Liniger, and C. Dziobkowski, Thin Solid Films 513, 295 (2006).

26 A. V. Vairagar, Z. Gan, V. Shao, S. G. Mhaisalkar, H. Li, K. N. Tu, Z. Chen, E. Zschech,
H. J. Engelmann, and S. Zhang, J. Electrochem. Soc. 153, G840 (2006).

27 International Technology Roadmap for Semiconductors, www.itrs.net, (2005), last
accessed May (2007).

28 S. M. Rossnagel, and J. Hopwood, Appl. Phys. Lett. 63, 3285 (1993).

29 A. Paranjpe, R. Bubber, L. Velo, G. Shang, S. Gopinath, J. Dalton, and M. Moslehi,
Proceedings of IEEE International Interconnect Technology Conference, San Francisco,
CA, 119, May (1999).

30 A. E. Kaloyeros, and E. Eisenbraun, Annual Rev. in Material Science 30, 363 (2000).

31M. H. Tsai, S. C. Sun, H. T. Chiu, and S. H. Chuang, Appl. Phys. Lett. 68, 1412 (1996).

32 R. Ecke, S. E. Schulz, M. Hecker, and T. Gessner, Microelectron. Eng. 64, 261 (2002).

33S. Bystrova, A. A. I. Aarnink, J. Holleman, and R. A. M. Wolters, J. Electrochem. Soc.
152, G522 (2005).

34 K. Wetzig and C. M. Schneider, Metal Ba~sed Thin Films for Electronics, 2nd Edition,
figure 4.1l(b) page 231 (Wiley-VCH, 2006).









35 D. T. Price, R. J. Gutmann, and S. P. Murarka, Thin Solid Films 308-309, 525 (1997).

36 N. A. Gj ostein, Di~fuision (American Society for Metals, Metal Park, Ohio 1973).

37 M. A. Nicolet, Thin Solid Films 52, 415 (1978).

38 B. Movehan, and A. Demchishin, Fiz. Metal. Metalloved. 28, 653 (1969).

39 L. E. Toth, Transition Metal Calrbides and Nitrides, (Academic Press, New York and
London, 1971).

40 J. A. Thornton, and D. W. Hoffman, Thin Solid Films 171, 5 (1989).

41 N. Hellgren, K. Maca, E. Broitman, M. P. Johansson, L. Hultman, and J-E. Sundgren, J.
Appl. Phys. 88, 524 (2000).

42 M. Uekubo, T. Oku, K. Nii, M. Murakami, K. Takahiro, S. Yamaguchi, T. Nakano, and
T. Ohta, Thin Solid Films 286, 170 (1996).

43 D. Mao and J. Hopwood, J. Appl. Phys. 96, 820 (2004).

44 J. A. T. Norman, J. Phys. IV France 11, 497 (2001).

45 J. S. Becker, S. Suh, S. Wang, and R. G. Gordon, Chem. Mater. 15, 2969 (2003).

46 M. Ritala, M. Leskela, Handbook of Thin Film Materials, vol. 1, 103, (Academic Press
2001).

47 G. W. Book, K. Pfeifer, and S. Smith, Microelectron. Eng. 64, 255 (2002).

48 H. Kizil, and C. Steinbruchel, Thin Solid Films 449, 158 (2004).

49 T. Heiser, C. Brochard, and M. Swaaneen, Mat. Res. Soc. Symp. Proc. 612, D7.3.1
(2000).

so T. Heisner, S. McHugo, H. Hieslmair, and E. R. Weber, Appl. Phys. Lett. 70, 3576
(1997).

51S. R. Kurtz, and R. G. Gordon, Thin Solid Films 140, 277 (1986).

52 A. Sherman, J. Electrochem. Soc. 137, 1892 (1990).

53 N. Yokoyama, K. Hinode, and Y. Homma, J. Electrochem. Soc. 138, 190 (1991).

54 J. C. Hu, T. C. Chang, L. J. Chen, Y. L. Yang, S. Y. Chen, and C. Y. Chang, J. Vac. Sci.
Technol. A 17, 2389 (1999).









55M. J. Buiting, A. F. Otterloo, and A. H. Montree, J. Electrochem. Soc. 138, 500 (1991).

56C. Faltermeier, C. Goldberg, M. Jones, A. Upham, D. Manger, G. Peterson, J. Lau, and A.
E. Kaloyeros, J. Electrochem. Soc. 144, 1002 (1997).

57C. Faltermeier, C. Goldberg, M. Jones, A. Upham, A. Knorr, A. Ivanova, G. Peterson, and
A. E. Kaloyeros, J. Electrochem. Soc. 145, 676 (1998).

58N. J. Archer, Thin Solid Films 80, 221 (1981).

59N. J. Ianno, U. Ahmed, and D. E. Englebert, J. Electrochem. Soc. 136, 276 (1989).

o0 A. Intemann, H. Koerner, and F. Koch, J. Electrochem. Soc. 140, 3215 (1993).

61K. Sugiyama, S. Pac, Y. Takahashi, and S. Motojima, J. Electrochem. Soc. 122, 1545
(1975).

62 R. M. Fix, R. G. Gordon, and D. M. Hoffman, Chem. Mater. 2, 235 (1990).

63 K. -C. Park, K. -B. Kim, I. J. M. M. Raaijmakers, and K. Ngan, J. Appl. Phys. 80, 5674,
(1996).

64 S. C. Sun, and M. H. Tsai, Thin Solid Films 253, 440 (1994).

65R. M. Fix, R. G. Gordon, and D. M. Hoffman, Chem. Mater. 3, 1138 (1991).

66I. J. Raaijmakers, Thin Solid Films 247, 85 (1994).

67C. Amato-Wierda, and D. A. Wierda, J. Mater. Res. 15, 2414 (2000).

68C. -K. Wang, L. -M. Liu, M. Liao, H. -C. Cheng, and M. -S. Lin, Jpn. J. Appl. Phys. 35,
4274 (1996).

69 S. -H. Kim, D. -S. Chung, K. -C. Park, K. -B. Kim, and S. -H. Min, J. Electrochem. Soc.
146, 1455 (1999).

7o I. J. Raaijmakers, and J. Yang, Appl. Surf. Sci. 73, 31 (1993).

n1 J. M. Bulger, C. S. Whelan, A. Dumont, M. Kuhn, and J. Clark, J. Vac. Sci. Technol. B
17, 410 (1999).

72 J. -Y. Yun, and S. -W. Rhee, Thin Solid Films 312, 24 (1998).

73 W. Besling, A. Satta, J. Schumacher, T. Abell, V. Sutcliffe, A. -M. Hoyas, G. Beyer, D.
Gravesteijn, and K. Maex, Proceedings of IEEE International Interconnect Technology
Conference, Burlingame, CA, 288, June (2002).









74 K. -E. Elers, V. Saanila, P. J. Soininen, W. -M. Li, J. T. Kostamo, S. Haukka, J. Juhanoja,
and W. F. A. Besling, Chem. Vap. Deposition 8, 149 (2002).

75M. Ritala, M. Leskela, E. Rauhala, and P. Haussalo, J. Electrochem. Soc. 142, 2731
(1995).

76S. Smith, G. Book, W. M. Li, Y. M. Sun, P. Gillespie, M. Tuominen, and K. Pfeifer,
Proceedings of the IEEE International Interconnect Technology Conference, San
Francisco, CA, 135, June (2003).

77S. Smith, W. -M. Li, K. -E. Elers, and K. Pfeifer, Microelect. Eng. 64, 247 (2002).

78K. -E. Elers, J. Winkler, K. Weeks, and S. Marcus, J. Electrochem. Soc. 152, G589
(2005).

79 P. M~rtensson, M. Juppo, M. Ritala, M. Leskela, and J. -0. Carlsson, J. Vac. Sci.
Technol. B 17, 2122 (1999).

so M. Juppo, M. Ritala, and M. Leskela, J. Electrochem. Soc. 147, 3377 (2000).

81M. Juppo, P. Alen, M. Ritala, T. Saj avaara, J. Keinonen, and M. Leskela, Electrochem.
Solid State Lett. 5, C4 (2002).

82 M. Juppo, A. Rahtu, and M. Ritala, Chem. Mater. 14, 281 (2002).

83 M. Ritala, M. Leskela, E. Rauhala, and J. Jokinen, J. Electrochem. Soc. 145, 2915 (1998).

84 J. -S. Min, Y. -W. Son, W. -G. Kang, S. -S. Chun, and S. -K. Kang, Jpn. J. Appl. Phys.
37, 4999 (1998).

85D. -J. Kim, Y. -B. Jung, M. -B. Lee, Y. -H. Lee, J. -H. Lee, and J. -H. Lee, Thin Solid
Films 372, 276 (2000).

86 J. W. Elam, M. Schuisky, J. D. Ferguson, and S. M. George, Thin Solid Films 436, 145
(2003).

87J. -H. Yun, E. -S. Choi, C. -M. Jang, and C. -S. Lee, Jpn. J. Appl. Phys. 41, L418 (2002).

88J. Y. Kim, Y. Kim, and H. Jeon, Jpn. J. Appl. Phys. 42, L414 (2003).

8"J. Y. Kim, S. Seo, D. Y. Kim, H. Jeon, and Y. Kim, J. Vac. Sci. Technol. A 22, 8 (2004).

"o J. Hillman, J. Hautala, and S. Caliendo, Proceedings of Advanced Metallization
Conference, San Diego, CA, 313, Oct. (2000).









91 X. Chen, G. G. Peterson, C. Goldberg, G. Nuesca, H. L. Frisch, and A. E. Kaloyeros, J.
Mater. Res. 14, 2043 (1999).

92 A. E. Kaloyeros, X. Chen, T. Stark, K. Kumar, S. -C. Seo, G. G. Peterson, H. L. Frisch,
B. Arkles, and J. Sullivan, J. Electrochem. Soc. 146, 170 (1999).

93 A. Y. Ganin, L. Kienle, and G. V. Vajenine, Eur. J. Inorg. Chem. 16, 3233 (2004).

94 X. Chen, H. L. Frisch, A. E. Kaloyeros, B. Arkles, and J. Sullivan, J. Vac. Sci. Technol. B
17, 182 (1999).

95 M. H. Tsali, S. C. Sun, H. T. Chiu, C. E. Tsai, and S. H. Chuang, Appl. Phys. Lett. 67,
1128 (1995).

"6M. H. Tsai, S. C. Sun, C. P. Lee, H. T. Chiu, C. E. Tsai, S. H. Chuang, and S. C. Wu,
Thin Solid Films 270, 531 (1995).

97 M. H. Tsali, S. C. Sun, C. E. Tsai, S. H. Chuang, and H. T. Chiu, J. Appl. Phys. 79, 6932
(1996).

98R. M. Fix, R. G. Gordon, and D. M. Hoffman, Chem. Mater. 5, 614 (1993).

""G. C. Jun, S. L. Cho, and K. B. Kim, Jpn. J. Appl. Phys. 37, L30 (1998).

100 S. -L. Cho, K. -B. Kim, S. -H. Min, and H. -K. Shin, Mat. Res. Soc. Symp. Proc. 514, 531
(1998).

101 H. Machida, A. Hoshino, T. Suzuki, A. Ogura, and Y. Shshita, J. Cryst. Growth 237-239,
586 (2002).

102 H. T. Chiul, and W. P. Chang, J. Mater. Sci. Lett. 11, 96 (1992).

103 J. W. Hong, M. L. Lee, K. I. Choi, Y. Chung, S. W. Lee, G. H. Choi, S. T. Kim, U. -I.
Chung, J. T. Moon, and B. -I. Ryu, Proceedings of the IEEE 2005 International 141,
June (2005).

104 L. Hiltunen, M. Leskela, M. Makela, L. Niinistoi, E. Nykanen, and P. Soininen, Thin Solid
Films 166, 149 (1988).

1os M. Ritala, P. Kalsi, D. Riihela, K. Kukli, M. Leskela, and J. Jokinen, Chem. Mater. 11,
1712 (1999).

1o' P. Alen, M. Juppo, M. Ritala, and M. Leskela, J. Mater. Res 17, 107 (2002).

107 H. Kim, A. J. Kellock, and S. M. Rossnagel, J. Appl. Phys. 92, 7080 (2002).









10s E. Eisenbraun, O. van der Straten, Y. Zhu, K. Dovidenko, and A. Kaloyeros, Proceedings
of the IEEE International, Burlingame, CA, 207, June (2001).

109 K. I. Choi, B. H. Kim, S. W. Lee, J. M. Lee, W. S. Song, G. H. Choi, U. -I. Chung, and J.
T. Moon, Proceedings of the IEEE International Interconnect Technology Conference,
6th, Burlingame, CA, 129, June (2003).

110 J. -S. Park, H. -S. Park, and S. -W. Kang, J. Electrochem. Soc. 149, C28 (2002).

'1 Y. Y. Wu, A. Kohn, and M. Eizenberg, J. Appl. Phys. 95, 6167 (2004).

112 A. Furuya, H. Tsuda, and S. Ogawa, J. Vac. Sci. Technol. B 23, 979 (2005).

113 A. Furuya, N. Ohtsuka, N. Ohashi, S. Kondo, and S. Ogawa, J. Vac. Sci. Technol. A 24,
103 (2006).

114 W. J. Maeng, S. -J. Park,and H. Kim, J. Vac. Sci. Technol. B 24, 2276 (2006).

"1 H. Kim, C. Detavenier, O. van der Straten, S. M. Rossnagel, A. J. Kellock, and D. -G.
Park, J. Appl. Phys. 98, 014308-1 (2005).

116 J. W. Hong, K. I. Choi, Y. K. Lee, S. G. Park, S. W. Lee, J. M. Lee, S. B. Kang, G. H.
Choi, S. T. Kim, U. -I. Chung, and J. T. Moon, Proceedings of the IEEE International,
141, June (2005).

'1 M. Nagai, and K. Kishida, Appl. Surf. Sci. 70/71, 759 (1993).

Il M. Nagai, N. Hirano, and S. Omi, Jpn. J. Appl. Phys. 39, 4558 (2000).

119 T. Nakajima, K. Watanabe, and N. Watanabe, J. Electrochem. Soc. 134, 3175 (1987).

120 N. Gonohe, Materials Transactions 43, 1585 (2002).

121 S. D. Marcus, and R. F. Foster, Thin Solid Films 236, 330 (1993).

122 C. Meinier, C. Monteil, C. Savall, F. Palmino, J. Weber, R. Berjoan, and J. Durand, Appl.
Surf. Sci. 125, 313 (1998).

123 A. R. Ivanova, C. J. Galewski, C. A. Sans, T. E. Seidel, S. Grunow, K. Kumar, and A. E.
Kaloveros, Mat. Res. Soc. Symp. 564, 321 (1999).

124 C. W. Lee, Y. T. Kim, and S. -K. Min, Appl. Phys. Lett. 62, 3312 (1993).

125 C. W. Lee, Y. T. Kim, C. Lee, J. Y. Lee, S. -K. Min, and Y. W. Park, J. Vac. Sci.
Technol. B 12, 69 (1994).









126 C. W. Lee, Y. T. Kim, and J. Y. Lee, Appl. Phys. Lett. 64, 619 (1994).

127 Y. T. Kim, C. W. Lee, and S. -K. Min, Jpn. J. Appl. Phys. 32, 6126 (1993).

128 C. S. Kwon, D. J. Kim, C. W. Lee, Y. T. Kim, and I. -H. Choi, Mat. Res. Soc. Symp.
Proc. 355, 441 (1995).

129 M. F. Bain, B. M. Armstrong, and H. S. Gamble, J. Mat. Sci. 14, 329 (2003).

130 J. Lin, A. Tsukune, T. Suzuki, and M. Yamada, J. Vac. Sci. Technol. A 16, 611 (1998).

131 S. Ganguli, L. Chen, T. Levine, B. Zheng, and M. Chang, J. Vac. Sci. Technol. B 18, 237
(2000).

132 H. -T. Chiul, S. -H. Chuang, Mat. Res. Soc. Symp. Proc. 250, 317 (1992).

133 H. -T. Chiul, S. -H. Chuang, Mat. Res. Soc. Symp. Proc. 8, 1353 (1993).

134 E. L. Crane, H. -T. Chiu, and R. G. Nuzzo, J. Phy. Chem. B 105, 3549 (2001).

135 J. E. Kelsey, C. Goldberg, G. Nuesca, G. Peterson, and B. Arkles, J. Vac. Sci. Technol. B
17, 1101 (1999).

136 B. H. Lee, and K. Yong, J. Electrochem. Soc. 151, C594 (2004).

137 S. Jeon, K. Yong, S. -G. Park, and S. -W. Rhee, J. Vac. Sci. Technol B 24, 1428 (2006).

138 R. G. Gordon, S. Barry, R. N. R. Broomhall-Dillard, V. A. Wagner, and Y. Wang, Mat.
Res. Soc. Symp. Proc. 612, D9121 (2000).

139 J. W. Klalus, S. J. Ferro, and S. M. George, Appl. Surf. Sci. 162-163, 479 (2000).

140 J. W. Klalus, S. J. Ferro, and S. M. George, J. Electrochem. Soc. 147, 1175 (2000).

141 S. -H. Kim, J. -K. Kim, N. Kwak, H. Sohn, J. Kim, S. -H. Jung, M. -R. Hong, S. H. Lee,
and J. Cooling, Electrochem. Solid-State Lett. 9, C54 (2006).

142 H. S. Sim, S. Kim-II, and Y. T. Kim, J. Vac. Sci. Technol. B 21, 1411 (2003).

143 C. W. Lee, and Y. T. Kim, J. Vac. Sci. Technol. B 24, 1432 (2006).

144 J. S. Becker, and R. G. Gordon, Appl. Phys. Lett. 82, 2239 (2003).

145 W. -M. Li, K. Elers, J. Kostamo, S. Kaipio, S. Huotari, M. Soininen, P. J. Soininen, M.
Tuominen, S. Haukka, S. Smith, and W. Besling, Proceedings of the IEEE International
Interconnect Technology Conference, 191, June (2002).









146 W. -M. Li, M. Tuominen, S. Haukka, H. Sprey, and I. J. Raaijmakers, Solid State
Technology 106, 103 (2003).

147 S. -H. Kim, S. S. Oh, K. -B. Kim, D. -H. Kang, W. -M. Li, S. Haukka, and M. Tuominen,
Appl. Phys. Lett. 82, 4486 (2003).

148 S. J. Wang, H. Y. Tsai, S. C. Sun, and M. H. Shiao, J. Electrochem. Soc. 148, G500
(2001).

149 A. M. Hoyas, Y. Travaly, J. Schumacher, T. Sajavaara, C. M. Whelan, B. Eyckens, O.
Richard, S. Giangrandi, B. Brij s, W. Vandervorst, K. Maex, J. P. Celis, A. M. Jonas, and
A. Vantomme, J. Appl. Phys. 99, F64 (2006).

150 K. -S. Kim, M. -S. Lee, S. -S. Yim, H. -M. Kim, K. -B. Kim, H. -S. Park, W. Koh, W. -M.
Li, M. Stokhof, and H. Sprey, Appl. Phys. Lett. 89, 081913-1 (2006).

A. M. Hoyas, J. Schumacher, C. M. Whelan, M. R. Baklanov, L. Carbonell, J. P. Celis,
and K. Maex, J. Vac. Sci. Technol. B 23, 1551 (2005).

152 D. Emur, V. Terzieva, J. Schumacher, V. Sutcliffe, C. M. Whelan, and K. Maex, J.
Electrochem. Soc. 152, B512 (2005).

153 0. J. Bchir, K. M. Green, M. S. Hlad, T. J. Anderson, B. C. Brooks, C. B. Wilder, D. H.
Powell, and L. McElwee-White, J. Organomet. Chem. 684, 338 (2003).

154 0. J. Bchir, S. W. Johnston, A. C. Cuadra, T. J. Anderson, C. G. Oritz, B. C. Brooks, D.
H. Powell, and L. McElwee-White, J. Cryst. Growth 249, 262 (2003).

O. J. Bchir, K. C. Kim, T. J. Anderson, V. Craciun, B. C. Brooks, and L. McElwee-White,
J. Electrochem. Soc. 151, G697 (2004).

156 0. J. Bchir, K. M. Green, M. S. Hlad, T. J. Anderson, B. C. Brooks, and L. McElwee-
White, J. Cryst. Growth 261, 280 (2004).

1 O. J. Bchir, K. M. Green, H. M. Ajmera, E. A. Zapp, T. J. Anderson, B. C. Brooks, L. L.
Reitfort, D. H. Powell, K. A. Abboud, and L. McElwee-White, J. Am. Chem. Soc. 127,
7825 (2005).

1 R. I. Hegde, R. W. Fiordalice, E. O. Travis, and P. J. Tobin, J. Vac. Sci. Technol. B 11,
1287 (1993).


1 R. Leuterecker, B. Froischle, U. Cao-Minh, and P. Ramm, Thin Solid Films 270, 621
(1995).









160 P. -T. Liu, T. -C. Chang, J. C. Hu, Y. L. Yang, and S. M. Sze, J. Electrochem. Soc. 147,
368 (2000).

161 A. Paranjpe, and M. IslamRaja, J. Vac. Sci. Technol. B 13, 2105 (1995).

162 M. Danek, M. Liao, J. Tseng, K. Littau, D. Saigal, H. Zhang, R. Mosley, and M.
Eizenberg, Appl. Phys. Lett. 68, 1015 (1996).

163 D. -H. Kim, S. -L. Cho, K. -B. Kim, J. J. Kim, J. W. Park, and J. J. Kim, Appl. Phys. Lett.
69, 4182, (1996).

164 J. Bonitz, S. E. Schulz, and T. Gessner, Microelectron. Eng. 70, 330 (2003).

165 G. S. Sandhu, S. G. Meikle, and T. T. Doan, Appl. Phys. Lett. 62, 240 (1993).

166 J. -Y. Yun, M. -Y. Park, and S. -W. Rhee, J. Electrochem. Soc. 146, 1804 (1999).

167 J. Y. Kim, G. H. Choi, Y. D. Kim, Y. Kim, and H. Jeon, Jpn. J. Appl. Phys. 42, 4245
(2003).

J. W. Elam, C. A. Wilson, M. Schuisky, Z. A. Sechrist, and S. M. George, J. Vac. Sci.
Technol. B 21, 1099 (2003).

169 K. Hieber, Thin Solid Film 24, 157 (1974).

170 D. W. Sloan, P. M. Blass, and J. M. White, Appl. Surf. Sci. 143, 142 (1999).

1 X. Zhao, N. P. Magtoto, and J. A. Kelber, Thin Solid Films 478, 188 (2005).

172 G. Beyer, A. Satta, J. Schuhmacher, K. Maex, W. Besling, O. Kilpela, H. Sprey, and G.
Tempel, Microelectron. Eng. 64, 233 (2002).

173 0. J. Bchir, Dissertation, University of Florida (2005).

174 R. C. Reid, J. M. Prausnitz, B. E. Poling, The Properties of Gases and Liquids, 4th
Edition, (McGraw-Hill, New York, NY 1987).

S. A. Gokoglu, PhD Thesis, Yale University (1982).

176 R. B. Bird, W. E. Stewart, and E. N. Lightfoot, Transport Phenomenoa, (John Wiley and
Sons, New York, NY 1960).

1 M. A. Kuczmarski, J. Wide Bandgap Mater. 7, 192 (2000).

Database of Joint Committee on Powder Diffraction Pattern (JCPDS), (2006).









179 B. D. Culity, and S. R. Stock, Elements ofX-ray Diffa~ction, 3rd Edition, (Prentice Hall,
Upper Saddle River, NJ, 2001).

Iso Handbook ofAuger Electron Spectroscopy, edited by K. D. Childs, B. A. Carlson, L. A.
LaVanier, J. F. Moulder, D. F. Paul, W. F. Stickle, and D. G. Watson, (Physical
Electronics, Eden Prairie, 3rd Edition, 1995).

1 G. Soto, W. de la Cruz, and M. H. Farias, J. Electron Spectroscopy and Related
Phenomena 135, 27 (2004).

182 Y. G. Shen, and Y. W. Mai, Material Sci. Eng. A 288, 47 (2000).

183 J. Luthin, and C. Linsmeier, J. Nuc. Mater. 290, 121 (2001).

184 Y. M. Sun, S. Y. Lee, A. M. Lemonds, E. R. Engbrecht, S. Veldman, J. Lozano, J. M.
White, J. G. Ekerdt, I. Emesh, and K. Pfeifer, Thin Solid Films 397(1-2), 109 (2005).

M. B. Zellner, and J. G. G. Chen, Catal. Today 99(3-4), 299 (2005).

186 A. Katrib, F. Hemming, P. Wehrer, L. Hilaire, and G. Maire, J. Electron Spectroscopy
and Related Phenomena 76, 195 (1995).

1 D. Gogova, K. Gesheva, A. Kakanakova-Georgieva, and M. Surtchev, Eur. Physical
Journal Appl. Phys. 11, 167 (2000).

G. E. McGuire, G. K. Schweitzer, and T. A. Carlson, Inorg. Chem. 12, 2450 (1973).

189 J. L. Alay, H. Bender, G. Brij s, A. Demesmaeker, and W. Vandervorst, Surface and
Interface Analysis 17, 373 (1991).

190 J. -M. Giraudon, P. Devassine, J. -F. Lamonier, L. Delannoy, L. Leclercq, and G.
Leclercq, J. Solid State Chemistry 154, 412 (2000).

191 R. J. Colton, and J. W. Rabalais, Inorg. Chem. 15, 236 (1976).

192 H. Nakazawa, T. Kawabata, M. Kudo, and M. Mashita, Appl. Surf. Sci. 253, 4188 (2007).

193 N. Yasui, H. Inaba, and S. Sasaki, e-Journal of Surf. Sci. Nanotech. 4, 129 (2007).

194 Handbook ofX-ray Photoelectron Spectroscopy, edited by J. F. Moulder, W. F. Stickle, P.
E. Sobol, and K. D. Bomber, (Physical Electronics Inc., 1995).

1 S. F. Pedersen, and R. R. Schrock, J. Am. Chem. Soc. 104, 7483 (1982).

D. Briggs, and M. P. Seah, Practical Surface Analysis Vol. 1,(John Wiley & Sons, 1992).









197 K. -E. Elers, V. Saanila, W. -M. Li, P. J. Soininen, J. T. Kostamo, S. Haukka, J. Juhanoja,
and W. F. A. Besling, Thin Solid Films 434, 94 (2003).

198 C. J. Carmalt, A. C. Newport, S A. O'Neill, I. P. Parkin, A. J. P. White, and D. J.
Williams, Inorg. Chem. 44, 615 (2005).

1 B. S. Lim, A. Rahtu, and R. G. Gordon, Nature Mater. 2, 749 (2003).

200 Z. W. Li, S. T. Barry, and R. G. Gordon, Inorg. Chem. 44, 1728 (2005).

201 A. Baunemann, D. Rische, A. Milanov, Y. Kim, M. Winter, C. Gemel, and R. A. Fischer,
Dalton Transactions 18, 3051 (2005).

202 C. B. Wilder, L. L. Reitfort, K. A. Abboud, and L. McElwee-White, Inorg. Chem. 45, 263
(2006).

203 P. Prieto, and R. E. Kirby, J. Vac. Sci. Techno. A 13, 2819 (1995).

204 J. T. Scheper, P. J. McKarns, T. S. Lewkebandara, and C. H. Winter, Mater. Sci.
Semicond. Process 2, 149 (1999).

205 C. J. Cramer, Essentials of Computational Chemistry, (Wiley Interscience, 2004).

206 P. J. Hay, and W. R. Wadt, J. Chem. Phys. 82, 270 (1985).

207 L. Laaksonen, J.Mol. Graph. 10, 33 (1992).

208 G. Schaftenaar, and J. H. Noordik, J. Computat. -Aided Mol. Design 14, 123 (2000).

209 Y. S. Won, Y. S. Kim, T. J. Anderson, L. L. Reitfort, I. Ghiviriga, and L. McElwee-
White, J. Am. Chem. Soc. 128, 13781 (2006).

210 P. Hohenberg, and W. Kohn, Phys. Rev. 136, B864 (1964).

211 D. M. Ceperley, and B. J. Alder, J. Phys. Rev. Lett. 45, 566 (1980).

212 S. J. Vosko, L. Wilk, and M. Nusair, Can. J. Phys. 58, 1200 (1980).

213 A. D. Becke, J. Chem. Phys. 38, 3098 (1988).

214 G. J. Laming, and N. C. Handy, J. Chem. Phys. 99, 8756 (1993).

215 M. Filatov, and W. Thiel, Mol. Phys. 91, 847 (1997).

216 C. Adamo, and V. Barone, J. Chem. Phys. 108, 664 (1998).









217 J. P. Perdew, Phys. Rev. B 33, 8822 (1986).

218 J. P. Perdew, and Y. Wang, Phys. Rev. B 45, 13244 (1992).

219 A. D. Becke, J. Chem. Phys. 104, 1040 (1996).

220 C. Lee, W. Wang, R. G. Parr, Phys. Rev. B 37, 785 (1988).









BIOGRAPHICAL SKETCH

Hiral M. Ajmera was born in Mumbai, India. He is the son of Mahesh and Neela Ajmera,

and has a brother, Abhijit. He is married to Kinjal Ajmera. He did his schooling from Kendriya

Vidyalaya, Rajkot in 1995 and received his bachelor of engineering (chemical) degree from

Maharaj a Sayajirao University, Baroda, India in 1999. He attended Rensselaer Polytechnic

Institute from 1999 to 2000 and received his MS degree in chemical engineering in 2001. Upon

graduation, he worked for two years at Applied Materials Inc. as a process engineer. He joined

NuTool in 2003 as a process engineer and worked there for 8 months. In August 2003, he joined

the PhD program in Chemical Engineering at University of Florida. Upon graduation, he plans

to work for Intel Corporation in Portland, Oregon.





PAGE 1

1 CHEMICAL VAPOR DEPOSITION OF WN x C y THIN FILMS FOR DIFFUSION BARRIER APPLICATION By HIRAL M. AJMERA A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2007

PAGE 2

2 2007 Hiral M. Ajmera

PAGE 3

3 To my great grandmother Jeevatiben Z. Ajmera, grandparents Chunilal and Heeraben Ajmera, and my parents Mahesh and Neela Ajmera

PAGE 4

4 ACKNOWLEDGMENTS I th ank the chair of my advisory committee, Dr. Timothy J. Anderson, for his continued support and guidance throughout the course of my study at the University of Florida. My advisory committee members, Dr. Lisa McElwee White, Dr. David Norton and Dr. Fan Ren also provided excellent guidance in my work. The work on the design, assembly and testing of the CVD copper reactor would not have been possible without the support of Jim Hinnant (Chemical Engineering, UF), Dennis Vince (Chemical Engineering, UF), Rob Holobof (A&N Corporation), and Chuck Rowland (MicroFabritech). The excellent facilities at Major Analytical Instrumentation Center along with its extremely helpful staff (Wayne Acree, Dr. Valentin Craciun, Kerry Siebein, and Rosabel Ruiz) were instrumenta l in obtaining film characterization results. I thank my co workers Dr. Omar J. Bchir, Andrew T. Heitsch, Karthik Boinapally, Dr. Seemant Rawal and Dr. Kee Chan Kim for their assistance. Last but not the least, I thank my parents, Neela and Mahesh Ajmera, my brother, Abhijit Ajmera and my wife, Kinjal Ajmera for their unconditional love and support throughout my education.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 4 LIST OF TABLES ................................ ................................ ................................ ........................... 9 LIST OF FIGURES ................................ ................................ ................................ ....................... 11 ABSTRACT ................................ ................................ ................................ ................................ ... 16 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .................. 19 1.1 Background ................................ ................................ ................................ ................... 19 1.2 ................................ ................................ ................. 21 1.3 Interconnects Challenges ................................ ................................ ............................ 22 1.3.1 Delay in Interconnects ................................ ................................ ....................... 22 1.3.2 Electromigration in Interconnects ................................ ................................ ..... 23 1.3.3 Power Consumption in Interconnects ................................ ............................... 24 1.4 Transition from Al W SiO 2 to Copper low k Metallization Scheme ............................ 24 1.4.1 RC Delay in Aluminum Interconnects ................................ .............................. 25 1.4.2 Electromigration in Aluminum Interconnects ................................ ................... 25 1.4.3 Copper Interconnects ................................ ................................ ........................ 26 1.4.4 Low k Dielectrics ................................ ................................ .............................. 27 1.5 Challenges to Copper Metallization ................................ ................................ .............. 28 1.5.1 Diffusion of Copper in Silicon ................................ ................................ .......... 28 1.5.2 Patterning of Copper ................................ ................................ ......................... 30 1.5.3 Diele ctric Integration and Reliability ................................ ................................ 32 1.5.4 Electromigration ................................ ................................ ................................ 33 1.6 Problem Statement ................................ ................................ ................................ ........ 35 1.6.1 Limitations of PVD ................................ ................................ ........................... 35 1.6.2 Drawbacks of TaN Deposition via CVD or ALD ................................ ............. 36 1.6.3 Desired Properties of Diffusion Barrier Film and its Deposition Process ........ 37 1.7 Hypothesis ................................ ................................ ................................ ..................... 38 2 LITERATURE REVIEW ................................ ................................ ................................ ....... 46 2.1 Diffusion in Thin Films ................................ ................................ ................................ 46 2.2 Refractory Carbides and Nitrides as Diffusion Barriers ................................ ............... 48 2 .3 Deposition Methods for Diffusion Barrier Films ................................ .......................... 50 2.3.1 Physical Vapor Deposition ................................ ................................ ................ 50 2.3.2 Chemical Vapor Deposition ................................ ................................ .............. 52 2.4 Characterization and Testing of Diffusion Barrier Films ................................ ............. 57 2.4.1 Diffusion Barrier Film Characterization ................................ ........................... 57 2.4.2 Diffusion Barrier Testing ................................ ................................ .................. 61

PAGE 6

6 2.5 Titanium Nitride as Diffusion Barrier ................................ ................................ ........... 64 2.5.1 Chemical Vapor Deposition of Titanium Nitride ................................ .............. 64 2.5.2 Atomic Layer Deposition of Titanium Nitride ................................ .................. 71 2.6 Tantalum Nitride as D iffusion Barrier ................................ ................................ .......... 78 2.6.1 Chemical Vapor Deposition of Tantalum Nitride ................................ ............. 78 2.6.2 Atomic Layer Deposition of Tantalum Nitride ................................ ................. 85 2.7 Tungsten Nitride as Diffusion Barrier ................................ ................................ ........... 91 2.7.1 Chemical Vapor Deposition of Tungsten Nitride ................................ ............. 91 2.7.2 Atomic Layer Deposition of Tungsten Nitride ................................ ................. 96 2.8 Tungsten Nitride Carbide as Diffusion Barrier ................................ ............................. 98 2.8.1 Atomic Layer Deposition of WN x C y ................................ ................................ 98 2.8.2 Chemical Vapor Deposition of WN x C y ................................ ........................... 101 3 EXPERIMENTAL TECHNIQ UES F OR THIN FILM DEPOSIT ION AND CHARACTERIZATION ................................ ................................ ................................ ...... 111 3.1 Description of the CVD Reactor ................................ ................................ ................. 111 3.2 Modification of the CVD Reactor ................................ ................................ ............... 111 3.2.1 Instal lation of H 2 and N 2 Purifiers in the H 2 and N 2 Delivery Lines .............. 111 3.2.2 Reconfiguration of the Gas Lines Downstream of the Reactor ...................... 112 3.2.3 Modification to the Line for Delivery of NH 3 to the Reactor ......................... 113 3.2.4 Replace MKS Mass Flow and Pressur e Programmer Display ........................ 113 3.3 Operating Procedure for CVD Deposition ................................ ................................ .. 114 3.4 Deposition Parameters ................................ ................................ ................................ 114 3.5 Dimensionless Numbers for CVD Reactor ................................ ................................ 114 3.5.1 Prandtl Number ................................ ................................ ............................... 115 3.5.2 Knudsen Numbe r ................................ ................................ ............................ 116 3.5.3 Reynolds Number ................................ ................................ ........................... 117 3.5.4 Peclet Number ................................ ................................ ................................ 118 3.5.5 Gr ashof Number ................................ ................................ .............................. 118 3.5.6 Rayleigh Number ................................ ................................ ............................ 119 3.6 Techniques Used for Thin Film Characterization ................................ ....................... 119 3.6.1 X Ray Diffraction ................................ ................................ ........................... 119 3.6.2 Auger Electron Spectroscopy ................................ ................................ .......... 121 3.6.3 X Ray Photoelectron Sp ectroscopy ................................ ................................ 123 3.6.4 Scanning Electron Microscopy ................................ ................................ ....... 124 3.6.5 Four Point Probe ................................ ................................ ............................. 125 3.7 Diffusion Barrier Testing ................................ ................................ ............................ 125 4 WN x C y THIN FILM DEPOSITION from Cl 4 (RCN)W(NC 3 H 5 ), NH 3 AND H 2 ................. 135 4.1 Precur sor Synthesis ................................ ................................ ................................ ..... 135 4.2 Film Composition ................................ ................................ ................................ ...... 136 4.3 X Ray Diffraction Measurement ................................ ................................ ................ 138 4.3.1 Film Crystallinity ................................ ................................ ............................ 138 4.3.2 Lattice Parameter ................................ ................................ ............................ 140 4.3.3 Polycrystal Grain Size. ................................ ................................ .................... 141

PAGE 7

7 4.4 Film Growth Rate ................................ ................................ ................................ ........ 142 4.5 Film Resistivity ................................ ................................ ................................ ........... 142 4.6 Comparison of Films Deposited from 3a,b and 2a,b ................................ ................. 143 4.7 Conclusions ................................ ................................ ................................ ................. 144 5 DEPOSITION OF WN x C y USING W(N i Pr)Cl 3 [ i Pr(NMe 2 )N i Pr] AND H 2 ......................... 156 5.1 Precursor Synthesis ................................ ................................ ................................ ..... 156 5.2 Film Growth ................................ ................................ ................................ ................ 157 5.3 Film Composition ................................ ................................ ................................ ....... 157 5.4 X Ray Diffraction of Films ................................ ................................ ......................... 159 5.5 Lattice Parameter ................................ ................................ ................................ ........ 162 5.6 Grain Siz e ................................ ................................ ................................ .................... 163 5.7 Film Growth Rate ................................ ................................ ................................ ........ 164 5.9 Atomic Bonding from XPS Measurement ................................ ................................ .. 165 5.8 Film Resistivity ................................ ................................ ................................ ........... 167 5.9 Diffusion Barrier Testing ................................ ................................ ............................ 168 5 10 Conclusions ................................ ................................ ................................ ................. 170 6 DEPOSITION OF WN x C y FROM W(N i Pr)Cl 3 [ i PrNC(NMe 2 )N i Pr] : EFFECT OF NH 3 CO REACTANT ON FILM PRO PERTIES ................................ ................................ ............... 194 6.1 Film Growth Studies ................................ ................................ ................................ ... 194 6.2 Film Composition ................................ ................................ ................................ ....... 194 6.3 Film Crystallinity ................................ ................................ ................................ ........ 196 6.4 Lattice Parameter ................................ ................................ ................................ ........ 197 6.5 Grain Size ................................ ................................ ................................ .................... 198 6.6 Film Growth Rate ................................ ................................ ................................ ........ 199 6.7 Atomic Bonding ................................ ................................ ................................ .......... 200 6.8 Film Resistivity ................................ ................................ ................................ ........... 203 6.9 Diffusion Barrier Testing ................................ ................................ ............................ 204 6.9.1 Barrier Testin g for 1A ................................ ................................ ..................... 204 6.9.2 Barrier Testing for 1B ................................ ................................ ..................... 206 6 10 Film Conformality ................................ ................................ ................................ ....... 207 6 11 Conclusions ................................ ................................ ................................ ................. 208 7 DEPOSITION OF WN x C y FROM (Ch 3 CN)Cl 4 W(NNMe 2 ) FOR DIFFUSION BARRIE R APPLICATION ................................ ................................ ................................ .................... 231 7.1 Precursor Synthesis ................................ ................................ ................................ ..... 231 7.2 Film Growth by CVD ................................ ................................ ................................ 231 7.3 Film Composition ................................ ................................ ................................ ....... 232 7.4 Film Crystallinity ................................ ................................ ................................ ........ 233 7.5 Lattice Parameter ................................ ................................ ................................ ........ 234 7.6 Grain Size ................................ ................................ ................................ .................... 235 7.7 Film Growth Rate ................................ ................................ ................................ ........ 235 7.8 Atomic Bonding from XPS ................................ ................................ ......................... 236

PAGE 8

8 7.9 Film Resistivity ................................ ................................ ................................ ........... 238 7.10 Diffusion Barrier Testing ................................ ................................ ............................ 239 7 11 Conclusions ................................ ................................ ................................ ................. 240 8 QUANTUM MECHANICAL C ALCULATIONS FOR PRED IC TING PRECURSOR FRAGMENTATION ................................ ................................ ................................ ............ 257 8.1 Computational Methodology ................................ ................................ ...................... 257 8.2 Geometry Optimization ................................ ................................ ............................... 258 8.3 Acetonitrile Cleavage ................................ ................................ ................................ .. 258 8.4 Chlorine Cleavage ................................ ................................ ................................ ....... 259 8.4.1 Reaction of 5a with H 2 ................................ ................................ .................... 259 8.4.2 Reaction Pathways for Removal of a Chloride from MI 1 ............................. 260 8.4.3 Reaction Pathways for Removal of Chloride from MI 2 and RI 2 ................. 260 8.4.4 Reaction Pathways for Removal of the Fourth Chloride from 5a .................. 261 8.5 Conclusion ................................ ................................ ................................ .................. 262 9 CONCLUSION AND FUTUR E WORK ................................ ................................ ............. 273 9.1 Diffusion Barrier Testing ................................ ................................ ............................ 273 9.1.1 Film Thickness Control ................................ ................................ ................... 273 9.1.2 Contamination Control ................................ ................................ .................... 274 9.1.3 Realistic Diffusion Barrier Testing ................................ ................................ 275 9.2 Diffusion Barrier Integration ................................ ................................ ...................... 275 A PPENDIX: DENSITY FUNCTIONAL T HEORY ................................ ................................ .. 277 A.1 Exchange Correlation Functional (E xc ) ................................ ................................ ....... 279 A.1.1 Exchange Energy Density ................................ ................................ ............... 279 A.1.2 Correlation Energy Density ................................ ................................ ............ 280 A.2 Density Gradient Correction ................................ ................................ ....................... 281 A.2.1 Correction to Exchange Functional: ................................ ............................... 281 A.2.2 Correction to Correlatio n Functional ................................ .............................. 281 A.3 Similarity Between Hartree Fock (HF) and DFT Methodologies .............................. 282 A.4 Comparison of DFT to Molecular Orbital T heory ................................ ...................... 282 A.4.1 Computational Efficiency ................................ ................................ ............... 282 A.4.2 Energetics ................................ ................................ ................................ ........ 283 A.4.3 Geometries ................................ ................................ ................................ ...... 283 A.5 The Future of DFT ................................ ................................ ................................ ...... 284 LIST OF REFERENCES ................................ ................................ ................................ ............. 286 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ....... 299

PAGE 9

9 LIST OF TABLES Table page 1 1 Bulk resistivity at room temperature of the five most conductive elemental metals ......... 39 1 2 Dielectric constant of materials with low k value ................................ .............................. 39 2 2 Techniques used for characterization of diffusion barrier thin films ............................... 103 2 3 Precursors used for deposition of TiN x thin films by CVD ................................ ............. 104 2 4 Precursors used for deposition of TiN x thin films by A LD ................................ ............. 104 2 5 Precursors used for deposition of TaN x thin films by CVD ................................ ............ 105 2 6 Precursors used for deposition of TaN x thin films by ALD ................................ ............. 106 2 7 Precursors used for deposition of WN x thin films by CVD ................................ ............. 107 2 8 Precursors used for deposition of WN x thin film by ALD ................................ ............... 107 3 1 Process parameters for CVD of thin films from precursors 3a,b 4 and 5 ...................... 127 3 2 Knudsen number for precursors 3a,b 4 and 5 with H 2 or N 2 carrier gas for different d eposition temperature ................................ ................................ ................................ ..... 127 3 3 Reynolds number for H 2 and N 2 carrier gas for different deposition temperature .......... 128 3 4 Grashof number for H 2 and N 2 carrier gas for different deposition temperature ............. 128 3 5 W 2 W 2 C W 2 C and WO 3 ................................ ................................ ............ 129 3 6 Elemental relative sensitivity factors for AES quantification along with position of peaks in dN(E)/d(E) spectra used for quantification ................................ ........................ 129 4 1 Bond distances () and angles (degrees) for Cl 4 (CH 3 CN)W(NC 3 H 5 ) ( 3a ) ..................... 146 5 1 Selected bond distances ( ) and bond angles () for 4 ................................ .................... 172 5 2 Analysis results from the deconvolution of the XPS peak of W for films deposited from 4 on Si(100) substrate at 400, 500, 600 and 700 C ................................ ................ 173 5 3 Analysis results from the deconvolution of XPS peaks of C, N and O for films deposited from 4 on Si(100) substrate at 400, 500, 600 and 700 C ............................... 174 6 1 Molar flow rates used for deposition of WN x C y thin film from precursor 4 .................. 209

PAGE 10

10 6 2 Analysis results from deconvolution of XPS peaks of W for films deposited by procedure 1A on Si(100) substrate betwe en 300 and 700 C. ................................ ......... 210 6 3 Analysis results from deconvolution of XPS peaks of C, N and O for films deposited by procedure 1A on Si(100) substrate between 300 and 700 C ................................ ..... 211 6 4 Analysis results from deconvolution of XPS peaks of W for films deposited by procedure 1B on Si(100) substrate between 300 and 700 C. ................................ ......... 212 6 5 Analysis results from deconvolution of XPS peaks of C, N and O for films deposited by procedure 1B on Si(100) substrate between 300 and 700 C ................................ ..... 213 6 6 Apparent sputter rate of barrier film me asured by dividing barrier film thickness by the time required to sputter the film during AES depth profiling ................................ .... 214 7 1 Molar flow rates of reactants in CVD reactor for deposition from precursor 5 .............. 242 7 2 Analysis results from deconvolution of XPS peak of W for films deposited from 5 on Si(100) substrate between 300 and 700 C. ................................ ................................ ..... 243 7 3 Analysis results from deconvolution of XPS peaks of C, N and O for films deposited from 5 on Si(100) substrate between 300 and 700 C. ................................ .................... 244 8 1 Selected bond lengths () and bon d angles () for precursor 5 obtained from X ray crystallographic measurement and DFT calculation ................................ ........................ 263 8 2 Selected bond lengths () and bond angles () for 5a obtained from DFT calculation .. 263

PAGE 11

11 LIST OF FIGURES Figure page 1 1 Cross sectional view of metal oxide semiconductor field effect transistor (MOSFET) ... 40 1 2 A cross sectional view of metallization scheme ................................ ................................ 40 1 3 Scaling of transistors in the last 3 decades for Intel Corporation microprocessors ........... 41 1 4 The device delay as a function of different device generations ................................ ......... 42 1 5 Effect of feature size scaling on total capacitance.. ................................ ........................... 42 1 6 Damaged Al interconnect line showing the formation of hillock and void due to e lectromigration. ................................ ................................ ................................ ............... 43 1 7 Electromigration lifetime fo r aluminum interconnects at different current densities. ....... 43 1 8 Band diagram of copper impurity in silicon. A) p type silicon. B) n type silicon ....... 44 1 9 Simplified steps in damascene processing for copper deposition. ................................ ... 44 1 10 Schematic of one ALD cycle forming a monolayer of film ................................ .............. 45 2 1 Arrhenius plot for CVD showing mass transfer limited and reaction limited growth regimes. G is the film growth rate and T is the deposition temperature in K ................. 108 2 2 Characteristic ALD growth rate vs. T curve showing different growth regimes ............. 109 2 3 Characteristic ALD plot showing the exposure time required to achieve self limiting growth ................................ ................................ ................................ .............................. 110 3 1 Schematic of CVD reactor used for thin film deposition ................................ ................ 130 3 2 Line diagram of the carrier gas feed line. A) Before installati on of gas purifier. B) after installation of gas purifier ................................ ................................ ........................ 131 3 3 Flow diagram of reactor downstream. A) Before reconfiguration. B) After reconfiguration ................................ ................................ ................................ ................. 132 3 4 Design of new cold trap made of stainless steel ................................ .............................. 133 3 5 Schematic of the old and new NH 3 feed port ................................ ................................ ... 134 4 1 Thermal ellipsoids diagram of the molecular structure of Cl 4 (CH 3 CN)W(NC 3 H 5 ) ( 3a ). Thermal ellipsoids are plotted at 50% probability ................................ ................. 146

PAGE 12

12 4 2 Composition of films deposited from 3a,b (with and without ammonia) and 2a,b with NH 3 at different deposition temperature as determined by AES on Si (100) substrate. ................................ ................................ ................................ .......................... 147 4 3 X ray photoelectron spectroscopy patterns of films dep osited at 450, 600 and 750 C from 3a,b and ammonia on Si (100) substrate. ................................ ................................ 148 4 4 X ray diffraction spectra of films deposited on Si (100) substrate from 3a,b and ammonia ................................ ................................ ................................ ........................... 149 4 5 Grazing incidence XRD spectra of films deposited on Si (100) substrate from 3a,b and ammonia ................................ ................................ ................................ .................. 150 4 6 Thickness of film (obtained from X SE M) at different deposition temperature for films deposited with 3a,b and ammonia on Si(100) substrate ................................ ......... 151 4 7 Lattice parameters for films grown from 3a,b on Si (100) substrate with and withou t ammonia. ................................ ................................ ................................ .......................... 152 4 8 Average grain size at different deposition temperature for films grown from 3a,b with and without ammonia on Si (100) substrate. ................................ ........................... 153 4 9 Scanning electron microscope images for films grown from 3a,b with ammonia. A) Film deposited at 450 C. B) Film deposited at 650 C ................................ ................. 153 4 10 Arrhenius plot of lo g of film growth rate vs. inverse temperature for deposition from 3a,b ( with and without ammonia 157 ) and 2a,b with ammonia ................................ ......... 154 4 11 Change in film resistivity with deposition temperature for f ilms grown on Si (100) from 3a,b (with and without ammonia) and 2a,b with ammonia ................................ .... 155 5 1 Thermal ellipsoids diagram of the molecular structure of W(N i Pr)Cl 3 [ i Pr(NMe 2 )N i Pr]. ................................ ................................ ........................ 175 5 2 Composition of films deposited from 4 and 2a,b on Si (100) substrate at different deposition temperature as determined by AES after 0.5 min of sputtering ..................... 176 5 3 N/W and C/W atomic ratios for films deposited from 4 and 2a,b on Si (100) substrate at different deposition temperature as determined by AES after 0.5 min of sputtering ................................ ................................ ................................ .......................... 177 5 4 X ray photoelectron spectroscopy patterns for films deposited from 4 at 400, 500, 600 and 700 C on Si(100) surface ................................ ................................ .................. 178 5 5 X ray diffraction patterns for films deposited from 4 on Si (100) substrate ................... 179 5 6 Deconvolution of XRD peaks for films deposited from 4 at 700 C on a Si (100) substrate. ................................ ................................ ................................ ....................... 180

PAGE 13

13 5 7 Deconvolution of XRD peaks for films deposited from 4 at 750 C on a Si (100) substrate. ................................ ................................ ................................ ....................... 181 5 8 Grazing incidence X ray diffraction spectra for films deposited from 4 on Si (100) substrate ................................ ................................ ................................ ........................... 182 5 9 Lattice parameters for films grown from 4 and 2a,b on Si (100) substrate. ................... 183 5 10 Average grain size at different dep osition temperature for films grown from 4 and 2a,b on Si (100) substrate ................................ ................................ ................................ 184 5 11 Scanning electron microscope images for films grown from 4 on Si (100) substrate between 400 and 750 C in 50 C increments ................................ ................................ 185 5 12 Arrhenius plot of log of film growth rate vs. inverse temperature for deposition from 4 and 2a,b on a Si(100) substrate ................................ ................................ ..................... 186 5 13 Deconvolution of the XPS peaks of W, N, C and O for film deposited from 4 at 400, 500, 600 and 700 C on Si (100) substrate ................................ ................................ ...... 187 5 14 Change in film resistivity with de position temperature for films grown on Si (100) from 4 and 2a,b ................................ ................................ ................................ ................ 188 5 15 Pre anneal AES depth profile of Cu (100 nm)/ WN x C y (45 nm)/Si (substrate) stack for WN x C y film deposited at 450 C (us ing precursor 4 ). ................................ ............... 189 5 16 Auger electron sp ectroscopy profiles of Cu /WN x C y /Si stack for WN x C y deposited at 450 C using 4 and annealed at A) 200 C B) 400 C C) 500 C D) 600 C. .............. 190 5 17 X ray diffraction plots for Cu/WN x C y /Si stack (WN x C y films deposited from 4 at 450 C) before annealing and after annealing at 200, 400, 500 and 600 C ........................... 191 5 18 Auger electron spectroscopy depth profiles of Cu (100 nm)/WN x C y (55 nm)/Si (100) stack for WN x C y film deposited at 500 C (using precursor 4 ) and annealed in vacuum for 30 min at A) 200 C B) 400 C C) 500 C D) 600 C. .............................. 192 5 19 X ray diffraction plots for Cu/WN x C y /Si stack (WN x C y films deposited with 4 at 500 C) before annealing and after annealing at 200, 400, 500 and 600 C ........................... 193 6 1 Composition of films deposited by procedures 1A, 1B and 1C on Si (100) substrate at different deposition temperature as determined by AES after 0.5 min of sputtering .. 215 6 2 X ray diffraction patterns for films deposited on Si (100) substrate by procedure 1A ... 216 6 3 X ray diffraction patterns for films deposited on Si (100) substrate by proc edure 1B .... 217 6 4 Lattice parameters for films grown by procedures 1A and 1C on Si (100) substrate. .. 218

PAGE 14

14 6 5 Average grain size at different deposition temperature for films grown by procedures 1A and 1B on Si (100) substrate. ................................ ................................ ..................... 219 6 6 Scanning electron microscope images for films grown by procedure 1A at different dep osition temperature on Si (100) substrate ................................ ................................ ... 220 6 7 Scanning electron microscope images for films grown by procedure 1B at different deposition temperature on Si (100) substrate ................................ ................................ ... 221 6 8 Arrhenius plot for growth using procedures 1A, 1B and 1C for growth on a Si(100) substrate ................................ ................................ ................................ ........................... 222 6 9 X ray photoelectron spectroscopy meas urements for films deposited by procedure 1A between 300 and 700 C on Si (100) substrate ................................ .......................... 223 6 10 X ray photoelectron spectroscopy measurements for films deposited for between 300 and 700 C by procedure 1B on Si (100) substrate ................................ .......................... 224 6 11 Change in film resistivity with deposition temperature for films deposited by procedures 1A, 1B and 1C on Si (100) substrate ................................ ............................. 225 6 12 Pre and post anneal AES depth profiles of Cu (100 nm)/ WN x C y /Si (substrate) stacks for WN x C y films deposited by procedure 1A. ................................ .................... 226 6 13 Pre and post anneal XRD measurements of Cu (100 nm)/ WN x C y /Si (substrate) stacks for WN x C y film deposited by procedure 1A at 300, 350 and 400 C ................... 227 6 14 Pre and post anneal AES depth prof iles of Cu (100 nm)/ WN x C y /Si (substrate) stacks for WN x C y film deposited by procedure 1B. ................................ ..................... 228 6 15 Pre and post anneal XRD measurements of Cu (100 nm)/ WN x C y /Si (substrate) stack for WN x C y film deposited by procedure 1B at 300, 350 and 400 C ..................... 229 6 16 Scanning electron microscope images of WN x C y thin films deposited on FSG dielectric. The feature size is 0.2 m with aspect ratio of 2:1 ................................ ........ 230 7 1 Composition of films deposited from 5 on Si (100) substrate at different deposition temperature as determined by AES after 0.5 min of sputtering ................................ ....... 245 7 2 X ray photoelectron spectroscopy measurements for films deposited from 5 on Si (100) substrate at different deposition temperature ................................ ......................... 246 7 3 X ray d iffraction patterns for films deposited from 5 between 300 and 700 C on Si (100) substrate ................................ ................................ ................................ .................. 247 7 4 Lattice parameters for films grown from 5 on Si (100) substrate. Error bars indicate uncert ainty in determination of peak position for lattice parameter calculation. ............ 248

PAGE 15

15 7 5 Average grain size for films grown from 5 at different deposition temperature on Si (100) substrate. ................................ ................................ ................................ ................. 249 7 6 Scanning electron microscope images for films deposited from 5 between 300 and 700 C on Si (100) substrate ................................ ................................ ............................ 250 7 7 Arrhenius pl ot for deposition from 5 on a Si(100) substrate ................................ ........... 252 7 8 X ray photoelectron spectroscopy measurements for films deposited from 5 between 300 and 700 C on Si (100) substrate ................................ ................................ .............. 253 7 9 Change in film resistivity with deposition temperature for films deposited from 5 on Si (100) substrate ................................ ................................ ................................ ............. 254 7 10 Pre and post anneal AES depth profile of Cu/WN x C y /Si (substrate) stack for WN x C y film deposited from 5 ................................ ................................ ...................... 255 7 11 Pre and post anneal XRD measurement of Cu (100 nm)/ WN x C y /Si (substrate) stack for WN x C y film dep osited from 5 at 350 and 400 C ................................ ............. 256 8 1 Geometry of 5. A) X ray crystallographic determination. B) Geometry optimization by DFT calculation ................................ ................................ ..................... 264 8 2 Cleavage of acetonitrile ligand from precursor 5 The figure shows optimized geometries for 5a and CH 3 CN ................................ ................................ ......................... 264 8 3 Calculated transition state and products for reductive bond metathesis from 5a ................................ ................................ ................................ ........ 265 8 4 bond metathesis from MI 1 intermediate. ................................ ................................ ...... 266 8 5 bond metathesis from MI 2 ( cis and trans ). ................................ ................................ ................................ ...... 267 8 6 Possible reaction pathway for reductive elimination from RI 2. R = reductive elimination, T = transition state and I = reaction intermediate ................................ ........ 268 8 7 bond metathesis from MI 3 and RI 3. ................................ ................................ ................................ ............... 269 8 8 5a) for reaction pathway through intermediate MI2 cis. ................................ ................................ ................................ .......................... 270 8 9 5a) for reaction pathway through intermediate MI2 trans ................................ ................................ ................................ ....................... 271 8 10 5a) for reaction pathway through intermediate RI2 ................................ ................................ ................................ ................................ 272

PAGE 16

16 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy CHEMICAL VAPOR DEPOSITION OF WN X C y THIN FILMS FOR DIFFUSION BARRIER APPLICATION By Hiral M. Ajmera August 2007 Chair: Timothy J. Anderson Major: Chemical Engineering The speed and reliability of integrated circuits has been improved significantly by replacing the Al/S iO 2 based metallization scheme with a Cu/low k based one. This study focused on chemical vapor deposition (CVD) of ultra thin WN x C y diffusion barrier films using novel tungsten metal organic precursors. The precursors used for deposition of WN x C y thin fi lm were Cl 4 (RCN)W(N C 3 H 5 ) ( 3 a R = CH 3 ; 3 b R = Ph) W(N i Pr)Cl 3 [ i PrNC(NMe 2 )N i Pr] ( 4 ) and (CH 3 CN)Cl 4 W(NNMe 2 ) ( 5 ). The thin films deposited in a aerosol assisted, vertical CVD reactor were evaluated for their material properties and diffusion barrier efficac y. Successful thin film deposition of WN x C y by MOCVD utilizing 3 a,b H 2 and NH 3 was achieved in the temperature range of 450 to 750 C. The lowest temperature for film growth was 450 C. The WN x C y films consisted of W, N, C and O as determined by AES. The Cl content of the film was below the detection limit of XPS (ca. 1 at. %). As compared to films deposited with 3a,b in H 2 those deposited with 3a,b H 2 and NH 3 had a higher N content. The higher nitrogen content in films deposited with NH 3 resulted in higher film resistivity with the lowest measured resistivity of 1690 cm for the film grown at 550 C. The apparent activation energy for the film growth from 3 a,b H 2 and NH 3 in the kinetically controlled growth regime was 0.34 eV,

PAGE 17

17 which is signifi cantly higher than the activation energy of 0.15 eV observed for film growth from 3 a,b in H 2 Successful WN x C y thin film deposition could be achieved for 4 with coreactant(s) NH 3 + H 2 only NH 3 and only H 2 The lowest temperature for which film growth occurred was 300 C with either NH 3 + H 2 or only NH 3 coreactant(s). Comparison of films deposited with coreactant(s) NH 3 + H 2 only NH 3 and only H 2 shows that the best films, in terms composition, resistivity, and microstructure, are deposited using on ly NH 3 as a coreactant. The films deposited with NH 3 at 400 C had a high N content of 28 at. % and were X ray amorphous. The films also showed very good conformality in a 0.2 m wide, 2:1 aspect ratio feature. Diffusion barrier testing showed that film s deposited at 400 C with only NH 3 coreactant were able to prevent Cu diffusion after annealing at 500 C for 30 min in vacuum. Chemical vapor deposition of WN x C y was achieved using 5 with H 2 as a coreactant It was determined that H 2 coreactant is ess ential for deposition of WN x C y films from 5 The lowest growth temperature for 5 was 300 C. Films consisted of W, N, C, and O as determined by AES and no Cl impurity was detected by XPS. The film N content was significantly higher for films deposited f rom 5 as compared to those deposited from 3a,b or 4 An Amorphous film microstructure was observed for deposition below 550 C. The apparent activation energy for the film growth in the kinetically controlled growth regime was 0.52 eV. The films showed a high resistivity compared to the bulk resistivity of pure W 2 N and W 2 C. Diffusion barrier testing showed that films deposited at 350 and 400 C were able to prevent bulk Cu diffusion after annealing at 500 C in vacuum for 30 min. CVD thin film depositi on from 3a,b 4 and 5 highlights the importance of precursor selection and deposition parameters (e.g., coreactant selection, deposition temperature) on the

PAGE 18

18 film properties and diffusion barrier performance. Detailed film characterization and preliminary diffusion barrier testing revealed that films deposited with 3a,b and NH 3 exhibited the most promise for diffusion barrier applications. To aid the precursor screening process and help understand the mechanism of precursor fragmentation prior to the grow th studies, quantum mechanical (QM) calculations using density functional theory were carried out. Statistical mechanics along with QM calculations were employed to determine the energy barrier of potential reaction pathways which would lead to the deposi tion of WN x C y thin film. QM calculations for fragmentation of precursor 5 showed that the first step of precursor fragmentation was dissociation of the CH 3 CN ligand, followed by removal of the Cl ligands by either sigma bond metathesis or reductive elimin ation.

PAGE 19

19 CHAPTER 1 INTRODUCTION 1.1 Background In the last few decades, electronic devices have become an integral part of our day to day life. From personal computers to laptops, cell phones to PDA s, almost all the electronic devices contain integrated circuits (ICs). The exponential growth in functionality of these ICs is the result of continued miniaturization of the devices in circuits, leading to an increase in chip functionality and decrease in cost. Semiconductor devices can be built on various s ubstrates, which include the elemental semiconductors silicon and germanium, and compound semiconductors such as GaN and GaAs. While compound semiconductors are used for specialized applications such as communication ICs, most of the ICs today use silicon as the substrate. The abundance of silicon ICs is primarily because SiO 2 can be grown easily on silicon substrate. SiO 2 is also an excellent insulator that can easily passivate the silicon surface. The manufacturing of ICs can be divided in two phases: front end of line (FEOL) processing and back end of line (BEOL) processing. FEOL is the initial processing on semiconductor substrate to build devices such as transistors and resistors. Major FEOL processes include epitaxial silicon film deposition, dopa nt implantation, diffusion, oxidation and deposition, and associated patterning steps. BEOL processing includes all the steps involved in connecting the active components of the circuitry via metal wiring. BEOL processing includes deposition of contacts, insulators, metal lines and vias, dielectric and associated patterning steps. The deposition of bonding sites for chip to package connection and wafer dicing are included in BEOL processing.

PAGE 20

20 The basic structures deposited during FEOL processing of ICs ar e resistors, capacitors, diodes, transistors, fuses, and conductors. 1 Transistors are the building block of an IC. They act as a switch or an amplifier in an IC. There are many different types of transistors, the most prominent being field effect transi stor (FET) and bipolar junction transistor (BJT). Figure 1 1 shows a metal gate metal oxide semiconductor field effect transistor (MOSFET). In a MOSFET, when a sufficient voltage is applied to the gate metal, current flows between the source and drain th rough the channel region. By controlling the gate voltage, the transistor can be turned on or off. MOS devices are voltage amplifiers, whereas BJTs are current amplifiers. MOS transistor circuits have the advantage of small area requirement and higher y ield in processing. Unlike application. Moreover, lower power requirement results in low heat generation in MOS transistors. MOS transistor circuits require fewer pro cessing steps resulting in lower cost per transistor. Because of these advantages, the majority of devices used in ICs today employ MOSFETs. The CMOS (complimentary metal oxide semiconductor) FET is formed in both p channel and n channel MOS transistor. CMOS uses lower power than similar circuits and is most popular in low power application. Various devices deposited during FEOL processing are connected to each other via a three dimensional network of metal lines known as interconnects. The application of metals in ICs is known as metallization. The metal lines are separated from each other by dielectric materials. Figure 1 2 shows a cross section of a three level metallization scheme. s. Local interconnects primarily refer to interconnects that connect the gate, source, and drain in MOSFET. Typical local interconnects are doped polysilicon and refractory metal silicide. Local interconnects can

PAGE 21

21 have higher resistivity because the curr ent in them travels for very short distances. But local interconnects need to be able to withstand higher processing temperature. Global interconnects, on the other hand, travel long distances and need to have lower resistivity. Typical global interconn ect metals are aluminum and copper. 1.2 functionality) doubles every 24 months. The law is attributed to Gordon Moore, co founder of In tel Corporation. Figure 1 3 shows how the number of transistors per microprocessor has roughly doubled every 24 months over the last three decades. The scaling of transistors in the earlier years was achieved primarily through the improvements in the FEOL processes. Between 1970 and 1990, the speed of IC was primarily determined by the speed of the transistors. Much of the speed increase achieved in transistors was by scaling the transistor dimensions such as oxide film thickness and channel length and w idth. Scaling of the device resulted in a higher transistors density per chip for a given chip area, lower the cost per transistor and increasing speed. The disadvantage of scaling is the higher current density in the transistors. Figure 1 4 shows how o verall device performance varies with the scaling of the device. 2 For earlier generation devices with large feature size, the gate delay dictated the overall device performance. With the shrinking of device dimensions, interconnect delay becomes much mor e important. In fact, interconnect delay is the dominant factor in determining the device speed for sub 100 nm devices.

PAGE 22

22 1.3 Interconnects Challenges 1.3.1 Delay in Interconnects The time delay in an interconnect circuit depends on two factors: resista nce of the conductor and the capacitance of the dielectric material. For a MOS circuit, the RC delay is defined in terms of the circuit response 3 (1 1) The output voltage of the circuit at time t is maximum output voltage is the resistance of the metal line is R, and the total capacitance of the circuit is C. The resistance of interconnect line is defined in terms of its length L, width W, thickness t M (1 2) The capacitance between silicon substrate and metal line (metal to ground), C M G is given by Equation 1 3. (1 3) and t D are the dielectric constant of the dielectric material and thickness of the dielectric line respectively, and 0 is the permittivity of free space. The capacitance between two metal lines (C M M ) separated by distance D is given by Equation 1 4. 4 (1 4) The total capacitance, C, can then be calculated from Equations 1 3 and 1 4. (1 5)

PAGE 23

23 K I is a constant that depends on the fringing field effect. The RC delay is calculated from Equat ions 1 2 and 1 5. (1 6) As the critical device dimension shrinks, C M M dominates the overall capacitance C (Figure 1 5). 5 The RC delay can then be written as Equation 1 7. (1 7) For lo cal interconnects the shrinking of the transistor leads to a corresponding decrease in L, W and D. Therefore, the actual delay in the local interconnect does not increase with device scaling. However, the chip area increases with each successive technolo gy node. This leads to an increase in the length of global interconnects and a corresponding increase in the RC time delay. 1.3.2 Electromigration in Interconnects Electromigration is the phenomenon by which high current density in the interconnect lines leads to the movement of the metal atoms in the direction of the electron flow. Most present day microprocessors operate at high temperature (ca. 100 C). At high temperature, the metal atoms in the interconnect have higher mobility. When high energy el ectrons transfer their momentum to these mobile metal atoms, there is a movement of metal atoms in the direction of electron flow. This atom movement leads to formation of voids at the cathode and hillocks at the anode (Figure 1 6) Electromigration, which causes failure of the circuit, is one of the most important concerns for interconnect reliability. As the interconnect dimensions shrink with each successive generation, the current density in the interconnects also increases. The scaling of interconnec ts

PAGE 24

24 requires higher electromigration resistance in interconnect materials. Electromigration becomes a major concern for aluminum based metallization scheme. Figure 1 7 shows that the mean time to failure for aluminum interconnect sharply decreases if the current density in the interconnect exceeds 3 MA/cm 2 1.3.3 Power Consumption in Interconnects The power consumption in an interconnect operating at frequency f and voltage V is given by Equation 1 8. P = CV 2 f (1 8) The shrinking of interconnec t dimensions causes an increase in parasitic capacitance, thereby increasing the power required to operate the IC. The power requirement in ICs is increasingly becoming an important issue because of the popularity of battery operated mobile devices. Sinc e the improvements in power storage capacity of batteries has not increased as fast as the power requirement of ICs, there is an increased focus on developing lower power ICs. To reduce the power requirement of interconnects, it becomes important to decre ase the capacitance of the inter metal dielectric. 1.4 Transition from Al W SiO 2 to Copper low k Metallization Scheme alu minum had been the preferred metal for interconnect lines and SiO 2 was the preferred dielectric. Tungsten was deposi ted in the vias to connect different levels of aluminum lines. The Al W SiO 2 based metallization scheme has many advantages. 6 A luminum can be easily deposited with evaporation or sputtering techniques with film resistivity close to the bulk resistivity o cm. A luminum is easy to pattern as it can be readily etched by dry or reactive ion etching. A luminum also has good adhesion to the SiO 2 surface. Addition of 0.5 at. % copper to aluminum increases the electromigration resistance of aluminum inte rconnects.

PAGE 25

25 1.4.1 RC Delay in Aluminum Interconnects One of the major problems with the Al W SiO 2 based interconnect scheme is that RC delay in these interconnects, which is a significant factor in determining the speed of an IC. The interconnect delay s urpassed the transistor delay below 0.25 m minimum feature size (Figure 1 4). To leverage the gains made in transistor speeds and density, a reduction in interconnect delay is required. Increase in signal noise and higher power requirement for narrower global interconnect lines necessitated use of new dielectric material with dielectric constant lower than that of SiO 2 (dielectric constant k = 3.9). 1.4.2 Electromigration in Aluminum Interconnects Another major problem with the aluminum based interconne ct scheme is reliability. In 1960s, it was discovered that electromigration can damage the aluminum lines and compromise the interconnect reliability. 7 Electromigration of aluminum atoms at the Al W interface results in movement of aluminum atoms away fr om the interface by grain boundary diffusion. The vacancies created by aluminum atoms due to electromigration accumulate to form a void. If the size of the void becomes larger than the diameter of the via, an opening in the circuit is created. This phen W SiO 2 metallization scheme results in a decrease in lifetime of multilevel metallization by a factor of 50 as compared to sing le level aluminum interconnect. 8 A short term solution to electromigration problem in Al W SiO 2 metallization was to use small amount of solute (1 4 at. % copper) in aluminum lines. Ames et al. 9 found that the addition of 4 at. % copper in aluminum str ips increased the electromigration resistance by a factor of 70 as compared to pure aluminum strips under similar testing conditions of 175 C temperature and 4 MA/cm 2 current density. Copper in aluminum forms Al 2 Cu precipitates along the grain boundaries of aluminum crystallites. While the exact mechanism behind the increase in

PAGE 26

26 electromigration resistance of aluminum by addition of copper is not completely understood, it is believed that copper may either decrease the concentration of vacancies in grain boundaries or increase the activation energy of aluminum diffusion through grain boundaries. The use of copper as solute extended the use of aluminum as the preferred material for interconnects. However, with further increase in current density in inter connects, even the aluminum doped with copper could not provide sufficient electromigration resistance. Sakimoto et al. 10 studied the mean time to failure (MTTF) of aluminum (0.5 at. % copper) at 200 C. The MTTF showed a dramatic decrease at current den sities of 3 MA/cm 2 and higher. Higher current densities in interconnects demand a new interconnect material that has a higher resistance to electromigration than aluminum doped with copper. 1.4.3 Copper Interconnects One way to reduce the RC delay in in terconnects is to use a material with higher electrical conductivity as interconnect. Table 1 shows the conductivity of the 5 most conductive metals. The metal with lowest bulk resistivity is silver. But silver shows low resistance to electromigration. Gold also has a lower bulk resistivity compared to aluminum, but gold is much more expensive and the performance gain in terms of resistivity is minimal. Copper with bulk resistivity close to silver is a much better conductor than aluminum. Moreover, co pper has a melting temperature (1083 C), which is much higher than that of aluminum (660 C). Atomic diffusion is slower in copper owing to its high melting temperature, and hence is expected to have a better electromigration resistance as compared to al uminum. The semiconductor industry has gradually moved towards copper metallization for global interconnects because of lower resistivity and superior electromigration resistance of copper. IBM was the first major chip manufacturer to announce successful implementation of copper interconnects in 1997.

PAGE 27

27 1.4.4 Low k Dielectrics Another way of reducing the RC time delay is to use dielectric material with a lower dielectric constant. The dielectric constant of a material depends on the polarizability of its c onstituent atoms or molecules; the higher the polarizability, the higher the dielectric constant. SiO 2 has a dielectric constant of 3.9. For SiO 2 based materials, low dielectric constant film can be obtained by replacing the Si O bonds in the film with S i C or Si F bonds, which have a much lower polarizability as compared to Si O bonds. The C and F atoms are incorporated in the film by introducing an organic precursor or a fluorinated silicon precursor source during the plasma enhanced chemical vapor dep osition (PECVD) of SiO 2 Another important class of materials that has been investigated for dielectric application includes silsesquioxane (SSQ) group based materials. Since the cubic SSQ cells occupy a much larger volume than tetrahedral SiO 2 the k v alue of SSQ is lower than that of SiO 2 Silsesquioxane unit cell contains Si O bonds in a cubic arrangement. In hydrogen terminated SSQ (HSSQ), unit cells are connected to each other via oxygen atoms, whereas the cube corners are terminated by hydrogen. In methyl terminated SSQ (MSSQ), the cubes are connected by CH 2 group, while cube corners are terminated by CH 3 group. Non polar organic compounds have also been investigated for dielectric application. Aliphatic compounds have the lowest k value but t hey are thermally unstable at temperature higher than 300 C. 11 Aromatic compounds and cross linked polymers have a higher thermal stability. Table 1 2 lists the important polymeric materials that have been studied for their potential application as diel ectric material in ICs. Decreasing the density of polarizable molecules is an effective way of reducing the k value, however, it often comes at the expense of mechanical strength. Density can be decreased by introducing micro/meso pores in the dielectri c film. Most materials have some inherent

PAGE 28

28 porosity, though the pore size might be very small. This inherent porosity is referred to as constitutive porosity. To increase the k value of dielectric film, intentional doping of thermally unstable molecules is used. Once these unstable molecules are evaporated or decomposed by thermal treatment, they leave behind a matrix of void spaces or pores resulting in a decrease in k value of the film. These void causing agents are referred to as porogens and are ve ry effective in lowering the k value of dielectric film. The porosity that is increased after the growth of the film, by agents such as porogen or by selective etching of the film, is referred to as subtractive porosity. For silica based films deposited by PECVD, CH x is used as porogen. Xerogels and aerogels both have ultra low k values primarily due to their subtractive porosity. Presently, many IC manufacturers have successfully integrated PECVD deposited carbon doped oxide (CDO or SiOC) into their met allization schemes in 90 nm devices with effective k value of 2.7 2.9. While acceptance of copper as the metal of choice for interconnects has been ubiquitous, the semiconductor industry has not yet singled out a single dielectric material for 45 nm dev ices. Carbon doped oxides have been preferred over other low k materials for 65 nm devices, however, industry is still evaluating different materials and processes for a suitable low k dielectric (< 2.5) at the 45 nm technology node. 1.5 Challenges to Cop per Metallization The introduction of new materials has led to new challenges in fabrication of ICs. These challenges are related to the processing of the new material or the integration of new material with other components of the IC. Copper metallizati on has introduced many processing and integration related challenges and this section highlights the most important challenges. 1.5.1 Diffusion of Copper in Silicon In a transistor, interstitial impurities such as Cu, Ni, Fe, and Li can be incorporated dur ing various processing steps. The tolerance level of these impurities in the device depends on the

PAGE 29

29 energy level of the impurity in band structure of the device, device dimension, diffusivity, and interaction of the impurity with silicon and dopants. If a n impurity has an energy level lying deep in the band gap of the semiconductor, such impurities can be detrimental to the device decrease diffusion length, increase junction leakage, and decrease minority carrier life time. 3 Moreover, the tolerable amount of defects in silicon decreases as the device dimension shrinks. Copper is an extremely fast diffuser in silicon. The diffusion coefficient of copper in silicon a t 500 C is 2.0 10 5 cm 2 /sec. 14 Furthermore, copper is mobile in silicon even after annealing. Copper can travel significant distances within the silicon wafer even at room temperature. The interaction of copper in silicon depends on the doping of the silicon substrate. In n type silicon, copper primarily forms neutral or negatively charged precipitates. Since interstitial copper is positively charged in silicon 15 there is either no barrier or electrostatic attraction between interstitial copper and copper precipitates. Hence, in n type silicon, copper readily forms Cu 3 Si precipitates. These copper precipitates can act as recombination centers for minority carriers and significantly lower minority carrier lifetime. 16 In p type silicon, copper prec ipitates are positively charged species. 17 There is electrostatic repulsion between the positively charged copper in precipitates and positively charged interstitial copper, which acts as a barrier in nucleation and precipitation of copper in p type silic on. The out diffusion of copper to the silicon surface is also dependent on the type of doping of silicon. If the surface of a p type Si wafer is free of native oxide, copper diffuses from the bulk to the surface of the wafer. Since surface segregated c opper can easily be etched by a H 2 O 2 and HF mixture, it is inferred that the copper does not form Cu 3 Si precipitates. For n type silicon wafer, out diffusion occurs only when the wafer is heated to 400 C.

PAGE 30

30 The primary reason why copper is extremely undes irable in silicon is that it forms deep levels in the band gap of silicon (Figure 1 8). If copper exists in the triplet state, it forms three acceptor levels in the silicon band gap at 0.24, 0.37 and 0.52 eV. 3 Once copper has diffused in silicon, the hig h diffusivity of copper in silicon ensures that the detrimental effects of copper spreads throughout the device. Copper is also detrimental if it diffuses into the dielectric film. Copper in dielectrics decrease the breakdown voltage of the dielectric an d also increases the parasitic capacitance, thus negating the benefits achieved from low k dielectric film. To prevent diffusion of copper into dielectric films, researchers have investigated various thin films that can act as diffusion barrier between co pper and the dielectric. Industry presently employs a TaN/Ta bilayer as diffusion barrier/adhesion promoter. TaN has good diffusion barrier properties and has good adhesion to SiO 2 Ta, on other hand, shows good adhesion to copper and TaN films. Ta als o promotes the growth of copper (111) crystals during physical vapor deposition (PVD) copper seed deposition. It is well known that copper (111) has higher electromigration resistance as compared to other crystal orientations. With the scaling of interco nnects, the barrier film also has to shrink. This is a major challenge in copper metallization because the PVD process currently employed to deposit the barrier film can not be extended for future generations of interconnects. The challenges in barrier d eposition are discussed in detail in section 1.6. 1.5.2 Patterning of Copper One of the main challenges in initial adoption of copper interconnects in ICs was its patterning scheme. Since copper does not form any volatile by products at low temperature, reactive ion etching (RIE) can not be employed for patterning copper. 18 This problem was addressed by a radical change in patterning scheme for interconnect layers. For aluminum metallization, aluminum is deposited first, followed by patterning via litho graphy and etching

PAGE 31

31 steps. The SiO 2 dielectric was then deposited into the aluminum pattern. To implement copper metallization, the scheme of deposition is reversed. The dielectric film is deposited first, and then the lines and vias for the metals are p atterned in the dielectric via lithography and etching steps, followed by deposition of copper in the lines and vias. The film stack is then subjected to chemical mechanical polishing step (CMP) to planarize the copper layer and remove the barrier layer o 9 shows the basic steps in the dual damascene processing scheme. It should be noted that since the purpose of the figure is to highlight the main steps in dual damascene processing, deposition of other layers such as diffusion barrier, dielectric etch stop layers and capping layers during the dual damascene processing are not included in the figure. In the dual damascene scheme, either the trench can be formed first followed by the via, or ickness of dielectric in the trenches, making it difficult to etch the vias through the thicker dielectric devices with minimum feature size of 0.25 m or lower. The detailed description of steps involved in copper interconnects are as follows: The copper line is coated with the first etch stop layer (usually SiN or SiCN). Dielectric film (layer 1) is deposited on the etch stop layer, followed by depositio n of the second etch stop layer. Another dielectric film (layer 2) is deposited on the second etch stop layer followed by the deposition of the third etch stop layer. hrough the top etch stop and dielectric layers to form the vias. This process is optimized to stop at the etch stop layer at the bottom of via. The photoresist is then removed. Mask is applied to pattern the trenches. Anisotropic etch cuts through the third etch stop and second dielectric layer to form the trenches. The photoresist deposited at the bottom of the

PAGE 32

32 via prevents over etching of the via during the trench etch process. Low energy etching step removes the mask from bottom of the via. The mas k layer is then removed. Diffusion barrier and adhesion promoter thin films are deposited by PVD process (TaN/Ta bilayer). is deposited by PVD. The trenches and v ias are filled with bulk copper by electrochemical plating. Excess copper over the barrier film is removed by CMP that stops at the first etch stop layer. As the feature size shrinks further, new processes and/or materials are expected to be introduced in the dual damascene patterning scheme. For instance, the integration of ultra low k dielectric films in the interconnects would require additional steps such as pore sealing to close the meso pores in the dielectric film prior to barrier film deposition. 11 1.5.3 Dielectric Integration and Reliability Presently, industry has successfully integrated CDOs in the interconnect stack. These dielectric films have k value in the range of.2.7 to 2.9. But as device dimension shrinks further, films with lower k valu e will be required to decrease the parasitic capacitance and power requirements. Introduction of highly porous films lowers the dielectric constant of films but it comes at the cost of mechanical stability. 19 Moreover, future interconnects will need proc esses such as CVD or ALD to deposit conformal diffusion barrier films. Gaseous reactants used in CVD or ALD process can easily diffuse through the interconnected pores in the dielectric film, thereby increasing the parasitic capacitance of the dielectric film. An ideal dielectric film should have uniformly distributed pores that are not interconnected to one another. These isolated pores prevent the diffusion of precursor molecules used in CVD and ALD process through the dielectric film and aid uniform d eposition of barrier film. But as the porosity of dielectric film is increased to lower the k value of dielectric film, these pores become more interconnected to one another. These interconnected pores act as pathways for diffusion of precursor molecules

PAGE 33

33 through the dielectric film. Interconnected pores also make it difficult to deposit uniform barrier film that can prevent copper diffusion into the dielectric. Various processes such as plasma exposure have been proposed to plug or block the open pores on the surface of the dielectric film. Adhesion of the dielectric film to the diffusion barrier is also of concern for new dielectric films that are being investigated for future interconnects. Failure modes such as time dependent dielectric breakdown (TD DB) that were not previously observed for SiO 2 based interconnects, have become reliability issues for low k dielectrics because of film adhesion problems. 20 In summary, lowering of the k value comes at the expense of mechanical stability, ease of integrat ion with barrier film deposition process, and adhesion and interconnect reliability. Extensive research is ongoing to find suitable processes and materials for deposition low k dielectric film and its integration in the dual damascene processing scheme. 1 .5.4 Electromigration Numerous studies have shown that copper has better electromigration resistance as compared to aluminum. However, as the dimensions of the interconnect shrink, electromigration in copper can result in poor reliability of interconnects Moreover, the electromigration mechanism in aluminum interconnect is very different from that in copper. While electromigration in aluminum vide supra ), electromigration in copper interconnects is primarily a surface migration mode. Surface migration has lower activation energy as compared to grain boundary diffusion. Copper electromigration occurs via surface diffusion because copper does not adhere well with dielectric films such as SiO 2 21 Th e top surface of trench is in direct contact with the dielectric and the surface diffusion of copper atoms occurs at this interface.

PAGE 34

34 One of the approaches that has found success in limiting electromigration in copper is alloying of copper metal. The solut e atoms in the copper alloy block the kink sites on the surface of the dielectric forming the copper line, thereby hindering the surface diffusion of copper and increasing the electromigration resistance. Solutes such as Sn and Zr have shown significant i mprovements in electromigration resistance when added to copper. 22 The downside of alloying copper is that the RC time delay increases as the resistance of copper alloy is higher than that of pure copper metal. Moreover, current copper bulk deposition in vias and trenches is done via an electrochemical plating process. Addition of solute in copper by the electrochemical plating process would require significant modifications to the existing process and further research needs to be done in order to adopt this approach. Another approach that has found wide acceptance in increasing electromigration resistance of copper is the use of cap layer to separate the copper dielectric interface. An ideal cap layer would have good adhesion to both the copper and diel ectric films. Good adhesion between copper and cap layer can significantly reduce electromigration failure in interconnect lines. Recent studies have investigated various cap layers such as Ta, Ta/TaN, Pd, SiN x SiC x N y H z and CoWP. 23 26 The disadvantage of using a cap layer is that it increases the effective k value of the dielectric film as the cap layer reduces the area of dielectric film. Industry is believed to introduce copper alloy for improving electromigration resistance of copper because this app roach is considered fairly low risk. In the long term, it is believed that capping layers can alleviate electromigration in copper interconnects for several generations. 20 Moreover, capping layers such as CoWP can be selectively deposited on copper line by electroless plating and hence the deposition of capping layer will not require patterning or etching steps. The primary reason why implementation of capping layer is pushed to future

PAGE 35

35 generations is because there is skepticism of introducing additional layer to the interconnect stack. Additional layers in interconnect stack introduce new processes, material(s) and integration related challenges that need to be addressed. 1.6 Problem Statement 1.6.1 Limitations of PVD Of all the future challenges to the scaling of copper metallization discussed in the previous section, one of the most important challenges is the scaling of the diffusion barrier film. With the shrinking dimensions of interconnect lines, the thickness of diffusion barrier needs to be reduc institutions, the thickness of barrier layer needs to be scaled down from 72 to 37 by year 2007 and to 24 by year 2013. 27 The scaling of barrier film is required for two reasons. First, as copper line and via dimensions shrink, the relative volume that the barrier film occupies in the line or via increases resulting in decrease in available area for copper. This could increase both the RC time delay of the interconnect and the current density in the copper lines and vias. Secondly, the barrier forms a part of electrical circuit where a via contacts the line. The thinner the ba rrier, the lower is the resistance to electron transport between via and line. Extensive research is being done to find a suitable process along with new material for deposition of low resistivity ultra thin barrier film with good conformality and adhesio n to copper and dielectric layers. Physical vapor deposition, which is presently employed to deposit diffusion barrier film, is believed to be incapable of depositing ultra thin films with conformal coverage. The PVD nd the directional nature of the deposition process results in excess deposition at the edges and poor sidewall coverage inside the trenches and vias. Ionized

PAGE 36

36 PVD (I PVD), a variant of PVD, employs a high density plasma to ionize the metal ions during dep osition. These ions are then directed perpendicular to the wafer surface via magnets to obtain better sidewall coverage. 28 While I PVD has helped in extending the usefulness of PVD for deposition of diffusion barrier film, future generations of diffusion barrier films would require deposition techniques that have a significantly better step coverage than I PVD. Chemical vapor deposition (CVD) is a technique that employs gaseous reactants to deposit thin films. Films deposited by CVD have much better st ep coverage than the best variant PVD process. Atomic layer deposition (ALD), a variant of CVD process, also has excellent step coverage and precise control over film thickness. Similar to CVD, ALD employs gaseous reactants but the delivery of the reacta nts is done in a sequential manner. One ALD cycle consists of 4 steps (Figure 1 10). In the first step, the substrate is exposed to reactant A, which adsorbs the first layer on the substrate surface by chemisorption and subsequent layers by physisorption In the next step, excess reactant A is purged from the reactor and the reactant molecules that are physisorbed on the chemisorbed layer are removed from the reactor. In the third step, a co reactant B is introduced into the reactor. Co reactant B reac ts with the chemisorbed A to form a monolayer thick film. In the last step, excess B is purged from the reactor. If chemisorption occurs in a self limiting mode, one ALD cycle should ideally be able to deposit a monolayer film. Atomic layer deposition i s believed to be capable of extremely uniform film deposition in very high aspect ratio features. 1.6.2 Drawbacks of TaN Deposition via CVD or ALD Another challenge to scaling of the diffusion barrier is the material. While PVD TaN has been very successfu l so far as a diffusion barrier, future interconnects require films deposited with CVD or ALD. Many investigators have attempted the growth of TaN thin films via CVD and ALD. But the growth of the resistive Ta 3 N 5 phase during attempts to grow TaN by CVD

PAGE 37

37 results in an increase in resistivity of the film. 29 Presently, many new materials are being investigated for diffusion barrier application, the details of which are included in the next chapter. 1.6.3 Desired Properties of Diffusion Barrier Film and its Deposition Process The following are the main requirements of a diffusion barrier film: Prevent copper diffusion under device operating conditions Good adhesion with copper, dielectric, and etch stop layers cm) Good mechanical and thermal stability CMP removal rate of the barrier film similar to that of copper Together with the material, there are certain process requirements for deposition of the diffusion barrier film: Capable of ultra thin film deposition (few nm thick) Excelle nt conformality Low deposition temperature (< 500 C) No pin holes in the barrier film Film properties that are considered beneficial for diffusion barrier film include: 1 Amorphous microstructure: Diffusion through the barrier film primarily occurs via gra in boundary diffusion. One of the ways of hindering the diffusion of copper through the barrier film is to deposit a single crystal barrier film. However, thermal requirements and lattice mismatch make deposition of a single crystal film impractical. Th e next best option is to deposit an amorphous barrier film. 30 2 Sufficient N content: Transition metal nitrides have been found to have good diffusion barrier properties. The primary reason, it is believed, is the presence of excess N at the grain boundarie s of the film. Since N does not form a stable compound with copper, the presence of N hinders the diffusion of copper through the grain boundaries. So, an prevent cop per diffusion. 3 Low impurities incorporation: Oxygen impurities form oxides that have a high resistivity. So, incorporation of O has to be minimal in the diffusion barrier film. In addition, impurities such as Cl and F are also undesirable because they in crease the film resistivity and adversely affect the adhesion of barrier film with copper.

PAGE 38

38 1.7 Hypothesis Thin film WN x C y deposited by metal organic chemical vapor deposition (MOCVD) would be capable of meeting the material and process requirements expecte d from a good diffusion barrier film. WN x C y a ternary transition metal compound, is expected to be able to be grown with an amorphous microstructure because the presence of a third element in the binary transition metal nitride matrix disrupts the crysta l lattice to favor the growth of amorphous film. Carbon in the film is expected to lower the film resistivity of WN x C y because WC x is more conductive than WN x An important focus of the present work is deposition of WN x C y thin films using metal organic pr ecursors. Films deposited by single source metal organic precursors should help to lower the deposition temperature. Low deposition temperature from these films should also aid the deposition of amorphous films. Precursors used for MOCVD should help in preventing incorporation of halides in the film. Previous studies have shown that WN x films have excellent diffusion barrier properties. 31,32 ,33 It is anticipated that WN x C y films deposited with the metal organic precursors used in the present work would also have excellent diffusion barrier properties. The WN x C y films deposited by MOCVD are expected to have good thermal and mechanical stability, good adhesion to dielectric and copper films, and a CMP removal rate similar to that of copper. Since the dep osition technique used in the present work is CVD, it is expected that the process would yield films with excellent conformality. As discussed earlier, films deposited by CVD have a much better conformality than those deposited by PVD. Film thickness con trol of CVD is expected to be similar to that of PVD. The precursors that are successful in CVD deposition of WN x C y films can then be used to attempt film growth by ALD to get even better conformality and precise thickness control.

PAGE 39

39 Table 1 1. Bulk resis tivity at room temperature of the five most conductive elemental metals Metal cm) Silver (Ag) 1.6 Copper (Cu) 1.7 Gold (Au) 2.4 Aluminum (Al) 2.6 Molybdenum (Mo) 5.2 Table 1 2. Dielectric constant of materials with low k va lue Material Dielectric Constant ( k ) Reference Silicon based SiF Si C 3.2 3.4 2.7 2.8 11 11 Silsesquioxane (SSQ) based HSSQ MSSQ 2.8 3.0 2.7 2.9 12 13 Polymer based Polyimides Parylene N Pary lene F Teflon (PTFE) 3.0 4.0 2.6 2.2 1.9 2.1 12 12 12 12 Highly porous oxides Xerogels/Aerogels < 2.5 12

PAGE 40

40 Figure 1 1. Cross sectional view of metal oxide semiconductor field effect transistor (MOSFET) Figure 1 2. A cross sectional view of metallization scheme

PAGE 41

41 Figure 1 3. Scaling of transistors in the last 3 decades for Intel Corporation microprocessors ( Wikipedia, http://en.wikipedia.org/wiki/Moores_law, May, 2007)

PAGE 42

42 Figure 1 4. The device delay as a function of different device generations 2 Adopted from M. T. Boh the r eal l imiter to h igh p of IEEE International Elect ron Devices Meeting, 241 (1995). 1995 IEEE. Reproduced with permission. Figure 1 5. Effect of feature size scaling on total capacitance. 5 Adopted from S. P. Jeng, R. H. 337 24 (1994). 1994, Materials Research Society, Pittsburgh, PA. Reproduced with permission.

PAGE 43

43 Figure 1 6. Damaged Al interconnect line showing the formation of hillock and void due to e lectromigration. Adopted from K. Wetzig and C. M. Schneider, Metal Based Thin Fi lms for Electronics 2nd Edition, 231 (Wiley VCH, 2006) Wiley VCH Verlag GmbH & Co. KGaA. Reproduced with permission. Figure 1 7. Electromigration lifetime for aluminum interconnects at different current densities. 10 Adopted from M. Sakimoto, T. Fu 333 (1995). 1995 IEEE. Reproduced with permiss ion.

PAGE 44

44 Figure 1 8. Band diagram of copper impurity in silicon. A) p type silicon. B) n type silicon 13 Adopted from physical properties of copper in silicon on characterization of copper diffusion Wiley VCH Verlag GmbH & Co. KGaA. Reproduced with permission. Figure 1 9. Simplified steps in damascene processing for copper deposition. A) Single damascene copper depos ition. B) Dual damascene copper deposition. 35 Adopted from 308 309 525 (1997). Elseweir. Reproduced with permission.

PAGE 45

45 Figure 1 1 0. Schematic of one ALD cycle forming a monolayer of film

PAGE 46

46 CHAPTER 2 LITERATURE REVIEW 2.1 Diffusion in Thin Films To prevent diffusion of copper into silicon by depositing a barrier film between the two materials, it is important to understand the fund amental mechanism of diffusion through thin films. When two dissimilar materials are put in contact with one another, atoms from one material diffuse into the other and vice versa because of the difference in concentration gradient. 1 f or 1D diffusion) gives a relation between the atomic flux at position x and the local concentration gradient of diffusing species. (2 1) The atomic concentration is C and atomic flux in direction x is J. For solids th e diffusion coefficient, D, is most often expressed by Equation 2 2. (2 2) Where, D 0 is a constant which is independent of temperature. It depends on the thermodynamic and kinetic properties of interaction between the two materials. E a is the activation energy for diffusion, R is the universal gas constant, and T is the temperature in K. The activation energy for diffusion depends on the mechanism of diffusion. Diffusion can occur for example by lattice diffusion, su rface diffusion, grain boundary diffusion or dislocation diffusion. There is a hierarchy of diffusion rates for these mechanisms of diffusion. Experimental observations for diffusion reveal that the activation energy for diffusion through a material depe nds on its absolute melting temperature (Eq. 2 3). (2 3)

PAGE 47

47 The proportionality constant, A, depends on many factors such as lattice structure and material interactions. The melting temperature (in Kelvin) of the film thr ough which the material is diffusing is T m The values of A for surface, grain boundary, dislocation and lattice diffusion are 13, 17, 25 and 34 J/(gmol.K), respectively and the corresponding values of pre exponential term D 0 are 0.014, 0.3, 2.1 and 0.5 c m 2 /sec, respectively. 36 Since the diffusion of material across surfaces is being considered the surface diffusion can be ignored in the discussion. It is quite evident that the activation energy for diffusion is lower for grain boundary diffusion as com pared to lattice diffusion. Since the diffusion coefficient varies exponentially with activation energy, the diffusion coefficient for grain boundary diffusion is several orders of magnitude higher th an that for lattice diffusion. Since both lattice and g rain boundary diffusion depend on temperature, a cross over occurs where the mechanism of diffusion changes from one to the other. The temperature at which this crossover occurs is known as the Tammann temperature. It is widely believed that the Tammann temperature is between one half or two thirds of the melting temperature T m of the solid. 37 Since almost all the potential diffusion barrier films have high melting temperature, grain boundary diffusion is the predominant mechanism by which diffusion occu rs in these films. The problem becomes severe for thin film diffusion barriers because the grain boundaries in thin films are of the order of thickness of the f ilm for polycrystalline films. The discussion on various diffusion mechanisms reveals two impor tant properties of a good diffusion barrier film: high melting temperature and defect free microstructure. A high melting temperature material is expected to have a higher activation energy for diffusion. The primary reason why diffusion barrier research has focused on transition metal compounds is because of the high melting temperature of transition metals along with their compounds such as

PAGE 48

48 nitrides and carbides. The melting temperature of some transition metals and their compounds are listed in Table 2 1. A perfect diffusion barrier would have no grain boundaries or micro structural defects. A defect free single crystal would be the best diffusion barrier because of the absence of grain boundaries. However, lattice mismatch and thermal budget limitat ions make the growth of single crystal diffusion barriers difficult. 38 The next best solution is amorphous films as these films have no long range order. Hence, one of the primary structural requirements for diffusion barrier is to have amorphous microst ructure. It should be noted that the material has to be perfectly amorphous. Nano crystalline film, which has grain boundaries of the order of few nanometers, is worse than polycrystalline film as it provides a network of pathways through which copper co uld diffuse. 2.2 Refractory Carbides and Nitrides as Diffusion Barrier s The majority of research towards finding diffusion barrier materials for copper metallization has focused on transition metals and their compounds. In particular, transition metal ni trides and carbides have been extensively researched as diffusion barrier to copper. To understand what makes these refractory carbides and nitrides good candidates for diffusion barrier application in copper metallization, a brief survey of the general p roperties of these compounds is presented. Transition metal carbides and nitrides can have a significant concentration of vacancies. These compounds can have significantly varying C or N content with appreciable vacancy concentration.Therefore, these comp ounds generally deviate from stoichiometric formulations. Carbon and nitrogen are interstitially located in the transition metal sublattice. Two kinds of bonding exist in these compounds, metal to metal and metal to nonmetal (C or N). The metal

PAGE 49

49 to metal to nonmetal bonding is covalent. The metal to oxygen bonding is presumed to be ionic. Transition metals and their compounds have many properties that make them good candidates for diffusion barrier application. Ca rbides and nitrides of transition metals have extremely high melting temperatures (2000 highest melting te mperature of any known material. 39 Since the activation energy for diffusion depends on the melting temperature of the compound ( vide supra ), refractory carbides and nitrides can be expected to be good barriers against copper diffusion. Diffusion barrier films should show good mechanical stability because thin films deposited by vacuum processes are almost always under stress. 40 The processes subsequent to deposition of barrier films also involve thermal cycling and contribute to the stress levels in the barrier film. Poor mechanical strength could cause fracture of the film and lead to diffusion of copper through macroscopic gaps in the barrier film. Refractory carbides and nitrides are also known for their excellent mechanical stability. These compoun ds, such as TiN and WC, are extremely hard and have found application as wear resistant coatings. While the exact reason for the extreme hardness of these materials is not known, it is believed that the presence of either C or N in transition metal stiffe ns the metal matrix resulting in increased hardness of refractory carbides and nitrides as compared to corresponding transition metals. 39 Refractory nitrides and carbides are brittle and they undergo minimal plastic deformation at low temperature (< 800 C). However, these compounds become plastic at higher temperature. Another important property of the barrier film is its electrical resistivity. In dual damascene processing, one trench and via feature is separated from the other by the barrier film.

PAGE 50

50 Th e barrier film forms part of the interconnect stack and hence low resistivity of barrier film is desired. Most refractory carbides and nitrides are metallic in nature and have low resistivity. Moreover, the resistivity of these compounds shows little tem perature dependence. The resistivity of refractory carbide or nitride thin film depends on factors such as porosity, non metal to metal ratio and impurity concentration. Oxygen impurity is always present in refractory carbides and nitrides due to the hig h affinity of these compounds for oxygen. Oxygen in the lattice can scatter electrons and increase the resistivity of the film. It should be noted that even though refractory carbides have low resistivity in general, a large variation in resistivity valu es for the same refractory carbide or nitride is observed because of factors discussed above. 2.3 Deposition Methods for Diffusion Barrier Films Thin films can be prepared by various techniques such as evaporation, PVD, CVD and ALD. Films for diffusion ba rrier applications have been predominantly deposited by PVD and CVD or one of their variants. The deposition technique is an important consideration for diffusion barrier film because many film properties depend strongly on the deposition process. Proper ties such as adhesion, microstructure, impurity concentration, within wafer uniformity, and conformality depend on the deposition process. This section details the advantages and limitations of most common deposition techniques for diffusion barrier film deposition. Important processing parameters for these deposition techniques and their general implications on film properties are also discussed. 2.3.1 Physical Vapor Deposition In general, PVD includes deposition processes such as evaporation, sputtering pulsed laser deposition, and high velocity oxygen fuel. For diffusion barrier film deposition, the most prominent PVD technique used is sputtering. In sputtering, ions generated by a plasma are accelerated towards the target (substrate). The sputtered target atoms deposit on the substrate

PAGE 51

51 resulting in the formation of thin film. Plasma generation can be done either by radio frequency (RF) or direct current (DC). Radio frequency sputtering is used for a non conducting target, whereas DC sputtering is used for conducting targets. When a magnetron is used to trap the electrons in the plasma closer to the target for low pressure plasma sustainability, the process is referred to as magnetron sputtering. For deposition of a metallic substance, pure metal is used as the target. To deposit alloys such as WN x a reactive gas (e.g., N 2 ) is used along with an inert gas to incorporate nitrogen into the metal. This process is often referred to as reactive ion sputtering. Important processing parameters for PVD are deposition pressure, applied voltage, target to substrate distance, partial pressure of reactive gas (for reactive ion sputtering), substrate temperature, and bias voltage on the substrate. The most important factor affecting the film properties is th e energy of sputtered particles at the substrate surface. Applied voltage, target to substrate distance, chamber pressure and bias voltage on the substrate determine the energy of the sputtered particles impacting the surface of the substrate. Applied vo ltage determines the energy of the sputtered particles ejected from the target surface. The deposition pressure determines the mean free path of the sputtered particles. The higher the mean free path, the higher the kinetic energy of the particles deposi ted on the substrate. The target to substrate distance, chamber pressure, and the bias on the substrate determine how much of the energy of the sputtered particle is lost due to collisions prior to reaching the substrate surface. All these factors can in fluence the film adhesion, porosity, crystallinity, texture, and substrate damage. The substrate temperature can also influence the crystallinity and texture of the deposited film. 4 1 The partial pressure of reactive gas used in reactive ion sputtering ca n influence the incorporation of reactant atoms in the deposited films. 4 2

PAGE 52

52 The advantages of sputtering are good film adhesion, uniform film deposition, low contamination and ease of deposition. The stoichiometry of alloys deposited by reactive ion etching can also be altered easily be changing the partial pressure of reactant gas. The major disadvantage of this technique is that it has poor conformality for ultra thin barrier film deposition. While I PVD can extend the use of PVD technique by improving t he conformality of films deposited by PVD 28 ,43 new deposition techniques will be required in the future to deposit highly conformal films in high aspect ratio features. Another disadvantage of PVD is the columnar microstructure of films deposited by PVD. Columnar microstructure provides short pathways for copper to diffuse along grain boundaries. Although it is well acknowledged that PVD cannot be extended to deposition of ultra thin barrier films, researchers have used the technique extensively because PVD is one of the easiest techniques to test new materials for diffusion barrier application. Unlike CVD which requires suitable precursors for deposition of thin films, PVD only needs a suitable target of good purity for deposition. A review of PVD films is not included in this section because PVD process is not capable of depositing ultra thin barrier films for future interconnects. Moreover, films of a particular material deposited by PVD films can have markedly different properties than the same mater ial deposited by CVD because of the differences in deposition mechanism. 2.3.2 Chemical Vapor Deposition In CVD, gaseous reactants or precursors are used to deposit thin films onto the substrate. For deposition of diffusion barrier film, the different var iants of CVD that have been predominantly used are plasma enhanced CVD (PECVD), low pressure CVD (LPCVD), metal organic CVD (MOCVD), aerosol assisted CVD (AACVD), atomic layer CVD (ALCVD or ALD) and plasma assisted ALD (PAALD).

PAGE 53

53 The important processing par ameters for a CVD reactor are reactor pressure, precursor(s) selection and their carrier gas flow rates, substrate temperature, and for PECVD and PAALD, power of plasma. The reactor pressure determines the number of collisions that involve the precursor molecules in the gas phase, which is important for determining the rate of homogenous gas phase reactions. The reactor pressure together with carrier gas flow rates and substrate temperature determine the flow dynamics inside the reactor. The thermal ene rgy supplied to the substrate aids the precursor decomposition on the substrate surface and is important to determining the rate of heterogeneous reactions. A characteristic growth curve for thermal CVD is shown in Figure 2 1. Two distinct growth rate reg imes are visible in the curve. At low deposition temperature, there is an exponential increase in film growth rate with deposition temperature. This behavior is observed when the rate determining step for the film growth is thermally activated, most comm only by 4. (2 4) The activation energy fo r the film growth is E a proportionality constant is A, the universal gas constant is R and the absolute temperature of deposition is T. The slope of the plot of ln(G) vs. 1/T gives the value of E a At higher deposition temperature, the film growth rate i s proportional to T m where m has a value of approximately 1.5. In this growth regime, the rate determining step is the mass transfer fer growth regime. This gr owth regime is sensitive to the reactor pressure and flow dynamics of the reactor. By reducing the reactor pressure, the growth rate increases because the boundary layer

PAGE 54

54 thickness decreases with decrease in pressure. A decrease in boundary layer thicknes s increases the mass trans fer of the reactants to the substrate surface and i ncreases the film growth rate. A CVD process uses a heated substrate to aid the decomposition of the precursor(s) on the substrate. In PECVD, a plasma is used to aid the decompos ition of the precursor(s). The primary advantage of PECVD is that the plasma facilitates the deposition of films at low temperature as compared to thermal CVD. This is particularly important when halides such as WF 6 or WCl 6 are used as precursors because halide precursors usually require very high temperature for decomposition. The disadvantage of PECVD is that plasma gives directionality to the atoms, molecules and/or ions and this has an adverse effect on the conformality of deposited film. Another di sadvantage of PECVD is that excessive fragmentation of the precursor leads to incorporation of impurities such as halides into the film. Halide impurities are undesirable in diffusion barrier films since they can increase the resistivity of the film. Mor eover, halides present at the interface are known to create adhesion problems between the barrier and copper. 44 Excessive fragmentation can also lead to deposition of undesirable compounds. For instance, PECVD of WN x has W also present in the film. Prec ise control of the precursor fragmentation due to the plasma is difficult. The plasma can also result in damage to the substrate surface. In LPCVD, the deposition of thin films is carrier out at sub atmospheric pressures. Low pressure increases the mean free path of the precursor(s), which in turn decreases the collision rate of precursor molecules and carrier gas in the gas phase prior to reaching the substrate surface. Low pressure also improves the within wafer uniformity of the film. Most CVD proces ses used in the semiconductor industry are LPCVD processes.

PAGE 55

55 In MOCVD, a metal organic precursor is used for deposition of thin film. As compared to CVD, which uses inorganic precursors, MOCVD has many advantages. Inorganic precursors predominantly use ha logen containing precursors such as WCl 6 which are corrosive. Moreover, the byproducts of the halogen containing precursors are HCl or HF, which are also very corrosive and can etch the underlying substrate. Films deposited by halogen containing precurs ors have shown pitting because of the etching action of precursor or byproducts ( vide infra ). In general, halogen containing precursors are hazardous and need special handling. In MOCVD, the precursor is mostly non corrosive. The metal organic precursor may or may not have halide containing ligands. Most of the by products of metal organic precursors that do not contain halides are non corrosive and hence do not cause pitting of the substrate. Even for halogen containing metal organic precursors, the li kelihood of pitting is less because of the presence of non corrosive organic byproducts. A limitation of MOCVD is that sometimes carbon containing ligands can also incorporate into the film. The C impurity can be either beneficial or not depending on app lication of the thin film. For diffusion barrier application, particularly for WN x a small amount of C is beneficial because it can increase the recrystallization temperature of the film. But greater amounts of C deposition can lead to significant incre ase in film resistivity and hence is not desirable for diffusion barrier film. Overall, MOCVD is a very good technique for diffusion barrier film deposition because of its low deposition temperature, non corrosive precursor, minimal halide contamination, and minimal substrate damage. AACVD is a variant of CVD where solid precursor is dissolved in a solvent and delivered to the reactor by aerosol generation. The aerosol generation is achieved by a nebulizer and the aerosol is transferred to the reactor by carrier gas. The advantage of using AACVD is that solid

PAGE 56

56 precursors with low vapor pressure can be delivered to the CVD reactor. The disadvantage of AACVD is that the solvent could participate in the homogenous or heterogeneous reactions, thereby altering the film properties. ALD (or ALCVD) is a limiting case of CVD in which the substrate is exposed to the precursors in a sequential manner to achieve self limiting growth ( vide supra ). The major advantage of the ALD process is the precise thickness control and excellent conformality. Since the growth of film occurs layer by layer, an ALD process can give very good thickness control and the conformality is also excellent. In fact, researchers have shown that even in features with 200:1 aspect ratio, the AL D process achieves excellent conformality. 45 Another important advantage of ALD is that since the precursors are introduced in the reactor alternately, there is no homogenous reaction of precursors. The two plots that are characteristic of an ALD process are growth rate versus temperature and growth rate versus precursor exposure time (Figures 2 2 and 2 3). In Figure 2 2, the growth rate increases linearly with temperature at lower temperature. Since the precursor adsorption and reaction between precurso r and surface are thermally activated processes, the thermal energy available for precursor reaction at low deposition temperature is insufficient to achieve complete saturation or reaction. 46 The second growth regime is characterized by film growth rate because the film growth rate is insensitive to substrate temperature. The third growth regime is VD 3 shows how the growth rate of film varies with exposure time. Initially when complete saturation of the surface is not achieved, film growth rate increases with an increase in the exposure time. Once complete saturation is ach ieved, the growth rate is independent of precursor exposure time,

PAGE 57

57 unless sufficient time is given for chemical reaction between the chemisorb layer and any gas phase species, including impurities. 2.4 Characterization and Testing of Diffusion Barrier Films Diffusion barrier films need to be characterized to understand the factors that affect the diffusion barrier performance of the film. The barrier film also needs to be tested for its ability to prevent copper diffusion. This section summarizes various t echniques that have been used for diffusion barrier film characterization and testing. 2.4.1 Diffusion Barrier Film Characterization There are a number of properties of the deposited film that are very important. These include microstructure, composition, density, resistivity, thickness, surface roughness, bonding between different atoms and adhesion of barrier film to copper and dielectric (low k or SiO 2 ). Various techniques used to analyze these film properties are listed in Table 2 2. This section det ails the efficacy of each technique in analyzing various film properties without discussing the guiding principles of each technique. Detailed explanation of principles of analysis techniques used in the present work are given in the Chapter 3. X ray dif fraction (XRD) is the most common technique used to determine the crystallinity and microstructure of thin films. For ultra thin films, grazing incidence XRD (GIXD) is used because it has a small depth of resolution. These techniques can also determine t he texture of the films. Electron diffraction is another highly sensitive technique that can determine the crystallinity in thin films. Imaging techniques such as SEM and TEM can provide information on grain structure of the films. High resolution image s have been used by many researchers to determine whether the films have equiaxed or columnar microstructure. Film composition can be determined by Auger electron spectroscopy (AES), X ray photoelectron spectroscopy (XPS), secondary ion mass spectroscopy ( SIMS), energy dispersive

PAGE 58

58 X ray spectroscopy (EDS), Rutherford back scattering (RBS) and elastic recoil detection analysis (ERDA). Auger electron spectroscopy and XPS are very effective techniques in determining the elements present in the deposited film. All the elements in the periodic table, except H and He, can be detected by using AES and XPS. However, film composition determination requires the calibration to identify the relative sensitivity factors needed for quantification in both AES and XPS. P ure elemental sensitivity factors can be used for quantification to obtain an estimate of film composition but elemental sensitivity factors do not account for the matrix effect. Both techniques have a detection limit of approx. 1 at. %. Both AES and XPS are surface sensitive techniques with sampling depth of few tens of nanometers. Both these techniques have been widely used for film composition determination and depth profiling of diffusion barrier films. Secondary ion mass spectroscopy can be used fo r determination of film content, but this technique is not used for determination of film composition. Its main use is in determination of trace impurities in the film. The technique can detect impurity concentrations of ppm or even ppb levels. The tech nique can also detect presence of H and He in the film. Energy dispersive spectroscopy is another technique that has been used to determine elements present in the film, however it can determine film composition only semi quantitatively. This technique c an detect elements with atomic number greater or equal to 6. Since EDS has poorer sensitivity to lighter elements in a heavy element matrix, this technique is not commonly used for transition metal nitrides and carbides. The sensitivity of this technique is 1 at. %. Rutherford back scattering (RBS) has been widely used for film composition determination of barrier films. This technique has the advantage of not requiring standards for film composition determination. The detection limit of RBS is about 4 at. % and hence this technique is not suitable for trace impurity detection. Elastic recoil detection analysis is a complimentary technique to RBS and it can detect lighter

PAGE 59

59 elements such as H in heavier element matrix. This technique has been widely use d to determine H element composition in deposited films. This technique has a detection limit of 0.1 at. % and a depth of resolution of few tens of nanometers. The common techniques used to determine film thickness are cross sectional scanning electron m icroscopy (X SEM) and transmission electron microscopy (TEM) imaging. Both these techniques provide an accurate measure of film thickness. Film thickness from profilometer measurements requires depositing film on a section of the substrate and profiling the step created by the film on the substrate. The accuracy of this technique depends on the resolution of the profilometer, which is usually few tens of angstroms. Rutherford back scattering and XRR can also provide measurement of film thickness, but th ese techniques are indirect techniques. The accuracy of the measurement by RBS and XRR depends on the accuracy of the variables used in simulation. Two techniques that have been prominently used to determine film density are RBS and XRR. Both these techn iques are indirect techniques for density measurement. Rutherford back scattering uses atomic density and the thickness obtained from either SEM or TEM analysis to simulate the value of film density. Film density can be obtained from XRR by simulating th e fringes in the spectrum. Simulation of XRR pattern requires accurate values of film thickness and roughness. Accuracy of both XRR and RBS in determining film density depends on the accuracy of parameters used in the simulation. The most common techniqu e for determination of diffusion barrier resistivity is four point probe. Four point probe determines the value of sheet resistance and this value when multiplied with the film thickness (from SEM or TEM) gives the resistivity of the film. Bonding betwee n elements present in the film can be determined by XPS analysis. The change in bonding state of

PAGE 60

60 an atom results in a shift in the binding energy observed in XPS spectra. This shift in binding energy can be used to determine the atomic bonding of a parti cular element in the film. X ray photoelectron spectroscopy is the only technique that can give the bonding information of atoms in thin films. Surface roughness of the film can be determined by various techniques such as AFM, stylus profilometer, scannin g tunneling microscopy (STM), optical profilometer, SEM, TEM and X ray reflectivity (XRR). Of these techniques, AFM and stylus profilometer are contact techniques because the probe used for roughness measurement is in contact with the surface. Non contac t techniques used for surface roughness measurement include STM, optical profilometer, SEM, TEM and XRR. Among the contact measurement techniques, AFM provides the best surface resolution of a fraction of an ngstrom. Atomic force microscopy can give 3 d imensional images of the scanned surface and provide values of root mean squared (RMS) surface roughness and average grain size. The probe used in AFM is a cantilever that scans across the surface. A stylus profilometer uses a diamond stylus to scan acro ss a surface and provide 2 dimensional image of the surface. The resolution of a stylus profilometer is few tens of nanometer. Stylus profilometer is commonly used to measure uniformity of film in trenches or vias, whereas AFM is commonly used to measure surface roughness of a blanket film. Since both AFM and stylus profilometer are contact measurement techniques, they can damage the surface because of scratching by the probe. Non contact techniques include imaging techniques such as SEM and TEM. These techniques are good for qualitative determination of surface roughness but not suitable to obtain RMS roughness. An optical profilometer uses light scattering to probe the surface roughness.

PAGE 61

61 This technique has a good spatial resolution of few nanometers and can provide 3 dimensional images as well as RMS roughness of the film. The disadvantage of this technique is that the surface has to be reflective. Since most of the diffusion barrier films are metallic in nature, this technique can be used to measu re surface roughness of diffusion barrier films. X ray reflectivity is an indirect method for surface roughness measurement and it requires accurate values of film thickness and density to simulate fringes in the XRR spectra. X ray reflectivity is not co mmonly used for surface roughness measurement. Scanning tunneling microscopy uses a very sharp probe which tunnels current through the sample surface to obtain images of the surface with atomic scale resolution. This technique requires very careful sampl e preparation and the imaging area is very small. Hence this technique is not commonly used to measure surface roughness of diffusion barrier films. The best method to determine the adhesion strength of a barrier to copper and dielectric is the four point bend test. This test measures the adhesion energies of interfaces in a multilayer thin film stack. Another method that is commonly used to determine the adhesion of barrier film qualitatively is the Scotch tape test. This test uses an adhesive tape tha t is applied to the thin film and the tape is pulled to determine if the thin film adheres to the substrate. The Scotch tape test is primarily used for preliminary evaluation of barrier film adhesion to copper and dielectric. 2.4.2 Diffusion Barrier Testi ng Various direct and indirect methods are used to test the effectiveness of diffusion barrier film in preventing copper diffusion. There are numerous techniques that can detect copper diffusion through the barrier into silicon and the most effective tech nique to detect trace amount of copper in silicon is electrical characterization. For a MOS test structure, electrical characterization techniques that are used to detect trace amount of copper in silicon or SiO 2 are flat band voltage shift in CV measurem ent, I V measurement and triangular voltage sweep

PAGE 62

62 (TVS). 47 For a Schottky contact test structure, the transient ion drift (TID) method can be used to detect and quantify the amount of copper that has diffused into silicon through the barrier after the fil m has been subjected to thermal stress or bias temperature stress. Ideally, a diffusion barrier test should involve integration of the barrier in actual chip followed by electrical characterization to determine the barrier performance. However, preliminar y diffusion barrier testing is often done on test structures such as a MOS capacitor to check for diffusion of copper through the barrier film. In a MOS test structure, copper metal dots are deposited on barrier/ SiO 2 / Si stack. The test structure is the n stressed at high temperature (150 300 C) under a bias of few MV/cm. This accelerated diffusion test is often referred to as bias temperature stress (BTS). Electrical characterization is then done to detect any diffusion of copper through the barrier into the oxide and silicon. In CV measurement, the flat band voltage is measured before and after BTS. If copper has diffused through the barrier film, the flat band voltage shifts indicating that the barrier film has failed. The I V measurement can al so be done on a MOS capacitor after BTS. In an I V measurement, if the leakage current density of the capacitor is higher than 10 6 A/cm 2 for the applied electric field of 1.2 MV/cm, then the capacitor is considered a failed capacitor. 48 A leaked capacit or would mean that copper has diffused through the barrier film. In TVS measurement, a positive voltage is applied to the copper electrode and is slowly made negative at a constant rate. If copper has diffused through the barrier, the mobile copper ions are injected back into the oxide from the Si SiO 2 interface. This results in a jump in the measured current and copper diffusion through the barrier can be detected. Among the three techniques, i.e. flat band voltage shift in CV measurement, I V measurem ent and TVS, the TVS technique is the most sensitive in detecting the presence of copper in silicon. 48

PAGE 63

63 In the TID technique, a Schottky barrier test structure is made by depositing barrier film on p type Si. Copper metal dots are then deposited on the bar rier and encapsulated by a protection layer. 49 If the Schottky test structure is quenched after thermal stressing, most of the copper in silicon remains as an interstitial for a few hours. 50 The interstitial copper ions in silicon tend to drift out when high electric field is applied. The resulting change in electric charge distribution induces a transient capacitance signal, which can be correlated to the copper ion concentration in silicon. This method has a detection limit of 10 12 at/cm 3 For prelimi nary evaluation of diffusion barrier efficacy, researchers have used other simpler techniques that do not require fabrication of a device. In one such test, a Cu/barrier/Si (or SiO 2 ) structure is thermally stressed by annealing at temperatures between 200 and 800 C. The higher the temperature at which the barrier shows copper diffusion into silicon, the better is the diffusion barrier performance. The diffusion of copper through the barrier is then detected by various techniques such as four point probe depth profiling (by AES, XPS or SIMS) and XRD. Four point probe measurement gives the sheet resistance of the Cu/barrier/Si(SiO 2 ) stack. When copper diffuses through barrier film, there is a sharp increase in sheet resistance. Depth profiling by AES, XPS or SIMS uses Ar + sputtering to sample through the Cu/barrier/Si(SiO 2 ) stack and detect if copper has diffused through the barrier into silicon. The disadvantage of depth profiling is that copper can be knocked forward the barrier film during the Ar + s puttering and the resulting etch the copper layer after thermal stressing and prior to depth profiling. Since bulk copper has ect due to sputtering can be avoided. The detection limit for copper by depth profiling depends on the technique used for elemental detection. Secondary ion mass spectroscopy provides the best resolution in detecting the presence of copper (detection

PAGE 64

64 lim it ca. 10 15 atoms/cm 3 ). Both AES and XPS can detect copper if its concentration in silicon is ca. 1 at. %. Pre and post anneal XRD measurements can provide information on diffusion of copper through the barrier and into the silicon. Copper forms Cu 3 Si in silicon at a relatively low temperature of 150 C. The annealing temperature at which Cu 3 Si peaks appear in the XRD spectra indicates the lowest temperature at which the barrier film has failed. All these techniques, i.e. sheet resistance measurement, XRD, four point probe measurement and depth profiling, have poorer resolution in detecting copper in silicon than electrical characterization. However, these techniques can be very helpful in preliminary screening of diffusion barrier films as these tech niques are much simpler than electrical characterization. Most of the research on diffusion barriers has been focused on refractory nitrides and carbonitrides of Ti, Ta and W ( vide supra ) The following sections review the work done on refractory nitrides and carbonitride thin films of Ti, Ta and W deposited by CVD and ALD techniques with primary focus on the suitability of these films for diffusion barrier application. 2.5 Titanium Nitride as Diffusion Barrier 2.5.1 Chemical Vapor Deposition of Titanium N itride Table 2 3 shows the inorganic and metal organic precursors that have been used to deposit TiN x thin films by CVD. TiN x has been deposited using the inorganic precursors TiCl 4 and TiI 4 as Ti source and NH 3 or N 2 plasma as N source. Thermodynamicall y, the reaction between TiCl 4 and NH 3 is feasible at temperature as low as 300 C, but appreciable film growth rate is observed only at 400 C by thermal CVD. 51 Two different processes, LPCVD and APCVD (atmospheric pressure chemical vapor deposition) have been used for thermal reaction of TiCl 4 and NH 3 Both APCVD and LPCVD deposit nearly stoichiometric TiN films with N:Ti ratio close to 1. Deposition by APCVD gives more uniform growth but the film contains a large amount of Cl impurity. At 400 C, the films contain as high as 5020 at. % Cl impurity. Increasing the

PAGE 65

65 deposition temperature leads to decrease in Cl content and films deposited at 620 C have Cl content of 5 2 at. %. 51 Film resistivity has a direct correlation with the Cl content of film. The cm for films grown at 620 C. When LPCVD is used to deposit TiN films from TiCl 4 and NH 3 the film impurity content and resistivity are better than those deposited by APCVD. Films deposited by LPCVD show 3 4 at. % Cl impurity for deposition at 450 C and high quality films with impurity content below the detection limit of AES and RBS are deposited at 650 C. 52 However, O and H content in the film increases with inc reasing deposition temperature. Because of low Cl content, films deposited with LPCVD have a lower resistivity. Resistivity of the film decreases from 290 cm for films grown at 700 C. The XRD spectra of films deposited below 600 C show that TiN films are textured with (200) preferred orientation. However, the films deposited at 700 C show a preferred orientation of (111). 53 Diffusion barrier testing by leakage current testing on p + n junction of TiN/TiSi 2 bilayer deposited with CVD showed that the bilayer was able to prevent copper diffusion after annealing at 550 C for 30 min in N 2 environment. 54 Annealing abo ve 550 C caused an increase in leakage current suggesting that copper had diffused through the barrier film. The main challenge for the LPCVD TiN process from TiCl 4 and NH 3 is the high Cl content of film for low temperature deposition. It is believed tha t Cl impurities in the film have high diffusivity. If these impurities diffuse to the interface, corrosion and adhesion problems can result. Increasing deposition temperature does reduce Cl content, but higher temperature processing is not desirable beca use of the thermal instability of low k dielectrics above 450 C. Yokoyama et al. 53 tried to use high temperature anneal in H 2 environment to remove Cl impurity from the film. The process did reduce the Cl content of film from 5.7 at. % to 2.7 at. % afte r

PAGE 66

66 annealing at 1000 C for 30 min., but the complete Cl elimination from the film was not achieved. The N:Ti ratio of the film also decreased from 1.02 to 0.97 after the annealing process, suggesting that the Cl and N are lost from the film as NH 4 Cl. Chl orine remains in the film after annealing due to the low volatility of TiN x Cl y even at 1000 C. Besides the Cl problem, LPCVD from TiCl 4 deposits polycrystalline films that have columnar microstructure. As discussed earlier, columnar microstructure provi des short diffusion pathways for copper diffusion and is highly undesirable. The process has problems with particulates because NH 3 reacts with TiCl 4 to form NH 4 Cl, a white solid. TiCl 4 also reacts with NH 3 at room temperature to form TiCl 4 .2NH 3 which i s also a solid. 55 This problem cannot be solved without changing the precursor chemistry of deposition. To address the Cl impurity concern, two different approaches have been attempted. One is the use of an alternate halide source such as TiI 4 and the ot her is the use of N 2 plasma as N source. TiI 4 approach is believed to lower the halide content of film because the Ti I bond is weaker than the Ti Cl bond. The heat of formation of Ti I at 298 K is 92 kcal/mol whereas the heat of formation of Ti Cl is 192 kcal/mol at 298 K. 56 Moreover, since I is a larger element compared to Cl, it would require much higher activation energy as compared to Cl to diffuse out of TiN x and into the Si. TiN x deposition by LPCVD from TiI 4 and NH 3 shows that the films depos ited are nearly stoichiometric with N:Ti ratio of 1.06:1 at 430 C deposition temperature. 56,57 The I and O impurities in films are less than 2 and 1 at. %, respectively. The density of the film is also higher by approx. 10 % as compared to the density o f PVD TiN x film. The film has resistivity as low cm and the step coverage in a 0.5 m 3:1 AR feature is 90%. The TiI 4 process does yield better films as compared to films deposited using TiCl 4

PAGE 67

67 However, the films are still polycrystalline and amorphous film growth could not be realized. Diff usion barrier testing for copper metallization was not reported for films deposited with TiI 4 PECVD has been used to deposit TiN x using TiCl 4 and H 2 + N 2 plasma. 58,59 When RF H 2 + N 2 plasma is used, violet colored films were deposited below 300 C. 59 Th e authors argued that the films were probably TiCl 3 A minimum temperature of 300 C was required to deposit TiN x films. The films deposited at and above 300 C were TiN x The films were slightly Ti rich and the lowest Cl content was 3 at. % at 500 C d eposition temperature. Increasing the H 2 flow resulted in an increase in Cl content, whereas an increase in the N 2 flow decreased the Cl content of the film. Other properties of the film deposited with plasma assistance were similar to those for film dep osited by LPCVD process. While the use of plasma did allow deposition to occur at lower deposition temperature, higher temperature close to 500 C was still needed to deposit films with minimal Cl impurity and low resistivity. So, the plasma process did not show any significant advantage over the LPCVD process. Tetrakis (dimethylamido) titanium (TDMAT) has been used as a precursor for TiN deposition. When TDMAT is used in an inert atmosphere as a single source precursor, TiN x C y films are deposited. 60 Fi lms can be deposited at temperature as low as 250 C. 61 Film deposited at 300 C has 27 at. % Ti, 17 at. % N, 11 at. % O and 35.2 at. % of C. X ray photoelectron spectroscopy analysis of the film shows that the C in the film exists in both carbidic form (bonded to Ti) and amorphous or organic form. 62 Films are also resistive, with the resistivity growth increase in resistivity with time due to oxygen and moistur e incorporation in the film. Diffusion barrier testing of the film shows that the barrier fails after vacuum annealing at 500 C. The poor performance of the barrier films was attributed to the porous nature of the film. 63

PAGE 68

68 When TDMAT is used with H 2 and NH 3 films were nearly stoichiometric with N:Ti ratio of 1. 64 Since TDMAT undergoes transamination reaction in solution, NH 3 was used as co reactant to aid the transamination reaction in gas phase. The transamination reaction also lowers the C incorporat ion in the films. 65 The films have a low C content of less than 3 at. % and the O content of the film varied between 4 and 9 at. %. The use of NH 3 has a significant effect on the lowest deposition temperature for film growth. Films could be deposited at temperatures as low within 24 hours of exposure to the ambient. The films were quite conductive and the resistivity varied between 400 to cm for as deposited films. However, the films were polycrystalline and showed columnar microstructure. 64 Moreover, films deposited with NH 3 showed poor step coverage and the films were less conformal at higher deposition temperature. 66 The confo rmality of the film was 30 % at 450 C deposition temperature. So, the process with NH 3 is not suitable for diffusion barrier application. Hydrazine (N 2 H 4 ) has been used as a co reactant with and without NH 3 to deposit TiN x thin films using TDMAT. 67 N 2 H 4 has a higher N:H ratio and lower dissociation temperature (140 C) as compared to NH 3 When N 2 H 4 is used with TDMAT to deposit thin films, the N:Ti ratio of film can be modulated by changing the N 2 H 4 :TDMAT ratio. The films contain very high O content (c and the film resistivity more than doubles after 150 min exposure to ambient. The resistivity of the diffusion and subsequent reaction of oxygen with the unsaturated Ti atoms. To improve the film properties, a mixture of N 2 H 4 and NH 3 was used to deposit thin films. The films deposited with N 2 H 4 and NH 3 were polycrystallin e and nearly stoichiometric TiN with O content below the

PAGE 69

69 detection limit of RBS. The authors argued that while N 2 H 4 was a better reducing agent, NH 3 was a better transamination reagent. So, the films deposited with N 2 H 4 and NH 3 had the best film properti cm so the use of NH 3 with N 2 H 4 so far. Plasma has been used both during growth and post growth to improve the properties of film deposited with TDMAT. When the N 2 plasma treatment is used post growth, films show an increase in N:Ti ratio. 68,69 The C content of the film decreases from 23 at. % to 3 at. % upon exposure to N 2 plasma. The films also show a low re effect is observed for plasma exposed films. The decrease in resistivity due to plasma exposure is believed to be because of a decrease in (N+C+O):Ti ratio because N, C and O act as scattering centers for electrons. 69 The plasma treatment is helpful only if done in situ Once the films are exposed to the ambient, subsequent plasma exposure does not cause a dramatic improvement in film properties. Plasma exposure also increases the film density. The plasma exposure, however, does not have any effect on diffusion barrier performance of the film. Films that are exposed to N 2 plasma after deposition fail at the same temperature of 550 C as the films that are not exposed to plasma. Tetrakis(ethylmethylamido) titanium ( TDEAT) has been used as a Ti precursor to deposit TiN x films with and without co reactant NH 3 When the precursor is used without NH 3 a minimum temperature of 350 C is required for thin film deposition. 70 For films deposited at 450 C, the film has a N :Ti ratio of 1. The films have significant C content (20 at. %) suggesting that the films are actually TiN x C y The films have a high O contamination of 18 at. % possibly because of the post growth diffusion of O in porous TiN x C y films. The film density has a value

PAGE 70

70 of 3.0 g/cc, which is low as compared to the bulk TiN density of 5.4 g/cc. The film shows an of the adsorption of O in the film. The resistivity of film deposited at 420 C increases from 1 cm after exposure to air for 24 hr. 68 To improve the film resistivity, in situ N 2 plasma exposure was employed. The plasma process densifies the films and incorporates N into the film as evidenced by the increase in N content fro m 23 to 44 at. % after plasma exposure. The C content of the film also shows a small decrease from 30 to 23 at. % after plasma exposure. The O in the film is decreased to below 3 at. % by plasma exposure resulting in a decrease in film resistivity to bel ow 100 cm. The film does not show an Wang et al. 68 deposited TiN x for application as a barrier for W in plugs, no copper diffusion testing was performed. To im prove the properties of films deposited with TDEAT, NH 3 has been used as a co reactant. With NH 3 TDEAT shows an increase in growth rate by a factor of 2 3 as compared to deposition without NH 3 70 It is believed that this increase in growth rate is bec ause of transamination. The films are slightly N rich with N:Ti ratio of 1.2:1 for deposition at 425 C. The C and O contamination in the film is ca. 0.5 at. %. The films deposited with NH 3 have a higher density of 4.2 g/cc (at 425C) as compared to tha t for films deposited without NH 3 Increasing deposition temperature, pressure or NH 3 flow rate results in a decrease in film density and resistivity. However, increase in TDEAT flow results in a decrease in film density and resistivity. 71 The lowest re sis cm was observed for deposition at 425 C. The drawbacks of using NH 3 as co reactant are the poor conformality and particle generation due to high precursor reactivity. Poor conformality is observed with NH 3 because NH 3 addition

PAGE 71

71 increases the growth rate of film and the growth mechanism changes from surface reaction limited to mass transfer limited. A process operated in the mass transfer limited growth regime is expected to give poor conformality. 64 Increase in deposition temperature fu rther increases the reactive sticking coefficient and worsens the conformality. 71 The conformality of film deposited with NH 3 at 425 C in a 0.45 m 3 AR contact hole is only 30 %. So, the process with NH 3 is not suitable for deposition of highly conform al diffusion barrier film. To improve the property of film deposited with TDEAT, Yun et al. 72 used N 2 and H 2 plasma to deposit thin films via PECVD. When N 2 plasma is used to deposit TiN x the films were C rich (45 at. %) and the Ti and N content of the f ilm was 25 and 13 at. %, respectively. The films had a r esistivity of 1500 cm. When H 2 plasma was used instead of N 2 plasma, the films show low C content (17 at. %), but the film N content was also reduced (6 at. cm. Overall, the use of plasma w ith TDEAT does not result in deposition of good quality films with low resistivity and favorable composition. 2.5.2 A tomic L ayer D eposition of Titanium Nitride Table 2 4 shows the precursors that have been employed for deposition of TiN x thin films by ALD. Two main inorganic Ti sources used to deposit TiN x are TiCl 4 and TiI 4 Since Ti in these precursors is in the +4 oxidation state, reducing agents are needed to reduce Ti to the +3 oxidation state in TiN. The reducing agents that have been used are NH 3 Zn, dimethylhydrazine, tert butylamine and allylamine. The deposition using TiCl 4 and NH 3 has been attempted between 350 and 500 C. The growth by ALD mode is nonlinear for first 50 cycles because of the initial nucleation and growth on the SiO 2 substrat e. 72 ,172 The temperature process window, which is characteristic of an ALD process, is not observed for TiCl 4 and NH 3 chemistry. The growth rate per cycle increases with

PAGE 72

72 increase in deposition temperature. Films deposited were polycrystalline and showed dense columnar structure. 73,74 The ratio of Ti:N was ca. 1:1, but the films contained Cl contamination. Similar to CVD deposition, ALD showed a decrease in Cl contamination with an increase in deposition temperature. 75 The resistivity of film decreased from 250 cm as the deposition temperature was increased from 350 to 500 C. There is a direct correlation between Cl contamination in the film and the resistivity of the film. Increase in the Cl contamination increases film resistivity. 75 At least 100 n m thick barrier film was required for complete surface coverage of TiN x on SiO 2 as determined by low energy ion scattering spectroscopy (LEISS). The adhesion of TiN film with copper and SiO 2 was excellent. The adhesion energy values for TiN on SiO 2 and c opper were 60 and 25 J/m 2 respectively as determined by the four point bend test. 76 In BTS testing of 5 nm thick TiN film, the barrier film did not fail after stressing at 2 MV/cm at 200 C for various times. There are 3 critical drawbacks of TiN x films deposited with TiCl 4 and NH 3 First, the films are polycrystalline and have columnar microstructure. As discussed earlier, polycrystalline films can act as fast diffusion pathways for copper and hence are undesirable. The films also had columnar microst ructure, which provides shorter pathways for copper diffusion. Second, the deposition temperature to obtain good quality TiN x films with minimal Cl contamination is close to 500 C, which is relatively high since the low k dielectrics are thermally unstab le at this temperature. Higher deposition temperature for TiCl 4 is needed because the Ti Cl bond is very strong (bond energy of 429 kJ/mol). Moreover, higher temperature is also needed because at low temperature, NH 3 reacts with Cl adsorbed on surface to form NH 4 Cl and temperature above 370 C is required to sublime NH 4 Cl. 74 Third, HCl formed as a byproduct of the reaction can

PAGE 73

73 severely etch the substrate and cause pitting. Severe copper pitting has been observed when TiN x is grown on copper. 73,74,77 One of the ways to reduce deposition temperature is to use PAALD. Plasma assisted ALD using TiCl 4 and NH 3 has been investigated by Elers and coworkers. 78 Similar to thermal ALD, PAALD also did not show the characteristic ALD process window for growth temper ature. The growth rate of films deposited with PAALD was significantly higher than that for thermal ALD process. Film deposited by PAALD had lower Cl contamination as compared to films deposited with ALD and the lowest Cl contamination of 1.2 at. % was o bserved at 400 C deposition temperature. The change of plasma power in PAALD resulted in a change in N:Ti ratio in the film. The N:Ti ratio increased with an increase in plasma power. Such modulation of composition is not possible for thermal ALD, wher e composition is determined only by growth temperature. While PAALD did have some advantage over thermal ALD, the process still was not able to deposit Cl free films. Moreover, the films were polycrystalline with columnar microstructure. Furthermore, th e use of halide chemistry also meant that the process would cause pitting on the substrate surface. To overcome the problems related to corrosive byproducts of the reaction, additional reducing agent such as Zn has been used in the ALD process. 75,79 When Zn is used as a reducing agent, there is no difference in growth rate per cycle. However, the Zn does influence the film texture. When Zn is used as reducing agent along with NH 3 the films have (111) preferential orientation, whereas without Zn the pref erred orientation of the film is (100). No Zn impurity is found in the TiN x films. 75 The film composition with or without Zn is similar except Cl content, which is higher for films deposited without Zn. Films deposited with Zn also had about 5 times low er resistivity than those deposited without Zn. Even though very good film qualities are

PAGE 74

74 obtained with Zn pulses in TiN x ALD, this process is not suitable for diffusion barrier deposition because Zn is an electrically active impurity in silicon and its in corporation in active areas of a transistor can severely affect the device performance. Since 1,1 dimethylhydrazine (DmHy) is a much stronger reducing agent than NH 3 ALD TiN x using TiCl 4 and DmHy has been investigated. 80 The films deposited had a very hi gh Cl content, up to 23 at. % at 200 C deposition temperature. The films were also more resistive because of the increased Cl and C content in the film. The lowest resistivity obtained was 500 cm at 350 C deposition temperature. The films were weakly polycrystalline. The use of DmHy did not show any significant improvements over NH 3 as co reactant. In fact, the resistivity and impurity content of films deposited was worse when DmHy was used. t ert Butylamine and allylamine have also been investigated as reductive N sources to lower the d eposition temperature of TiN x ALD using TiCl 4 81 Depositions were done between 400 and 500 C. Films deposited with t ert butylamine had a low growth rate of 0.03 /cycle. Addition of NH 3 to t ert butylamine increased the growth rate to 0.15 /cycle. Fil m resistivity was similar to those deposited with NH 3 only and the contamination level of C and H was also below 1 at. %. By using t ert butylamine, the deposition temperature could be lowered from 500 to 400 C to obtain low resistivity TiN x films with mi nimal Cl contamination (1.4 at. %). However, the process could not deposit amorphous TiN x films. When allylamine was used as the reducing agent, the films had higher C and Cl contamination of 7 and 6 at. %, respectively. 81 This resulted in an increase in film resistivity to cm. Addition of NH 3 to allylamine did not have any significant effect, although the Cl contamination decreased to 4 at. % and C contamination increased to 9 at. %. As with t ert butylamine, films deposited with allylamine were weakly polycrystalline. Additional diffusion

PAGE 75

75 barrier testing of TiN x films deposited with either t ert butylamine or allylamine has not been reported yet. TiN x ALD has been attempted using TiI 4 and NH 3 79,82,83 The ALD process did not show the characteris tic temperature process window as the film growth rate increased with increase in deposition temperature. 83 The films were Ti rich and the N content of film increased from 20 at. % at 350 C to 45 at. % at 500 C. Impurity content of the film, as measure d by RBS, showed that the films had low I content (2 at. % at 350 C and 0 at. % at 400 C) The films were heavily contaminated with O and the film O content decreased with increase in deposition temperature. Films deposited at 350 C showed O content o f 40 at. %, while those deposited at 500 C showed O content of 10 at. %. The TiN x films were polycrystalline and the preferred o rientation was dependent on dep osition temperature. Films deposited below 425 C showed preferred orientation of (100), where as films deposited above 450 C showed preferred orientation of (111) The film resistivity for deposition at 400 C was 380 cm Overall, high porosity and the presence of impurities such as I and O made the films deposited by this process unsuitable for diffusion barrier deposition. To overcome the issues related to halogen impurities in films deposited using inorganic precursors, organometallic Ti precursors have been investigated for deposition of ALD TiN x thin films. Tetrakis(ethylmethylamido) ti tanuim (TEMAT), or Ti[N(C 2 H 5 )(CH 3 )] 4 has been used with NH 3 to deposit ALD TiN x thin films. Self limiting growth of TiN x has been reported using TEMAT. 84,85 Unlike ALD from halide precursors, ALD from TEMAT showed the characteristic ALD process window. While one report established the ALD process window for TEMAT between 150 and 220 C, 85 another report located the ALD process window between 170 and 210 C. 84 The deposition temperature for TEMAT is considerably lower than that for inorganic

PAGE 76

76 Ti precurso rs. The growth rate per cycle is 5 /cycle, which is about 1.6 ML/cycle for TiN x film. It has been argued that the film growth of more than 1 ML/cycle is possible if NH 3 can react with TEMAT adsorbed on the surface and TEMAT can react with chemisorbed NH 3 Film growth during the individual pulses of both TEMAT and NH 3 would result in film growth rate greater than 1 ML/cycle. 84 Unlike films grown with inorganic Ti precursors, films grown with TEMAT had amorphous microstructure. While films deposited wit h inorganic Ti precursors had a N:Ti ratio close to 1:1, films deposited with TEMAT had N:Ti ratio of 0.55 suggesting that the film deposited is actually Ti 2 N. C and H contamination in the film was 4and 6 at. %, respectively for films deposited at 200 C. cm for cm for films grown at 200 C. Diffusion barrier testing for Ti 2 N films showed that the barrier was able to prevent bulk copper diffusion after annealing at 600 C in vacu um (as determined by XRD of film post anneal). After annealing at 650 C, Cu 3 Si peaks appeared in XRD suggesting that copper had penetrated through the barrier film. Ti 2 N films deposited by TEMAT showed many properties expected from a good diffusion barr ier film. To evaluate if Ti 2 N is effective in preventing trace copper diffusion, additional electrical testing of barrier film is required. TDMAT has also been used with NH 3 to deposit TiN x films via ALD. Unlike TEMAT, TDMAT did not show the characterist ic temperature process window for ALD. The reaction of TDM AT and NH 3 is not self limiting. Precursor decomposition study by FTIR indicated weak dependence of growth rate on TDMAT exposure time because of incomplete desorption of residual Ti(N(CH 3 ) 2 ) x spe cies 86 Mechanistic study by FTIR also indicated that the film growth occurs via transamination between TDMAT and NH 3 Similar to TEMAT, TDMAT was also able to deposit TiN x thin films at relatively low deposition temperature. At 180 C deposition

PAGE 77

77 temper ature, the growth rate of TiN x was 2 /cycle. 87 Films were Ti rich and the N:Ti ratio of ca. 0.5 indicated the presence of Ti 2 N. The C contamination in the film was less than 10 at. % but the films had a very high O contamination which decreased with inc rease in deposition temperature. 86 The films were highly porous (> 35% porosity) which explained the low density of Ti 2 N film. The highest film density was 3 g/cc while the bulk density of stoichiometric TiN is 5.2 g/cc. High porosity results in an incr ease in O diffusion through the barrier film and a corresponding increase in film resistivity. The lowest resistivity of 1.4 10 4 cm was observed at 240 C deposition temperature. High porosity and extremely high resistivity of Ti 2 N films deposited b y TDMAT makes this process unsuitable for diffusion barrier application. To improve the Ti 2 N film quality, post growth rapid thermal nitriding and plasma exposure have been explored. 87 Rapid thermal nitriding at 500 C resulted in a net decrease in organi c C in the film (ca. 5 at. %) and the films were amorphous even after nitriding. Increasing the nitriding temperature to 700 C resulted in formation of macroscopic cracks in the barrier film. Post growth p lasma exposure decreases the film C content belo w the detection limit of XPS. Since higher N content was observed on the surface of Ti 2 N than in bulk, N most likely gets incorporated in the film during plasma exposure. The plasma exposure lowered the film resistivity from 2 10 4 to 3000 cm due to lowering of C content and minimal O incorporation The plasma exposure also resulted in the densification of the Ti 2 N surface. The only disadvantage of the plasma process was that it changed the film microstructure from amorphous to weakly polycrystallin e as indicated by the Ti 2 N peak in the XRD spectrum. TiN x film growth by PAALD using TDMAT with remote N 2 N 2 + H 2 and H 2 plasma has been investigated. 88,89 The use of plasma was aimed at reducing the film C content. Since the plasma is generated ex situ the likelihood of surface damage by plasma is also minimized.

PAGE 78

78 Unlike thermal ALD of TiN x using TDMAT and NH 3 the PAALD process did show a characteristic ALD temp erature window of 200 300 C where the film growth rate did not change with deposition te mperature. Plasma also resulted in dense films which produced low levels of O impurities (ca.10 at. %). The C content of the film was also < 10 at. %. The film cm. Film r esistivity was lowest for films deposited using the N 2 plasma process followed by those deposited using H 2 + N 2 plasma The highest resistivity was observed for films deposited with H 2 plasma The C impurities also showed a similar trend for different plasma exposures with the lowest C content observed for films deposited with N 2 plasma. While thermal ALD affords 100% conformal coverage in high aspect ratio features, the s tep coverage for PAALD process was 95 % in 1:10 aspect ratio feature. Diffusion barrier testing of 19 nm TiN x films deposited with different plasma exposures showed that copper was able to diffuse through the film after annealing at 500 C as witnessed by the emergence of Cu 3 Si peaks in XRD spectra. The failure temperature of barrier film was same irrespective of whether N 2 H 2 + N 2 or N 2 plasma was used for film deposition. 89 Even thou gh the PAALD film was able to decrease the film resistivity as compared to thermal ALD, the copper diffusion barrier performance of the film was not satisfactory. 2.6 Tantalum Nitride as Diffusion Barrier 2.6.1 Chemical Vapor Deposition of Tantalum Nitride CVD of TaN x has employed both inorganic and metal organic precursors as the Ta source (Table 2 5). The three main inorganic precursors that have been employed to deposit TaN x thin films are TaCl 5 TaF 5 and TaBr 5 TaI 5 has not been employed for CVD becau se it requires very high temperature to obtain sufficient vapor pressure for delivery in a CVD reactor. 90 TaCl 5 TaF 5 and TaBr 5 are polycrystalline solids at room temperature with sufficient vapor pressure at

PAGE 79

79 elevated temperature for delivery in a CVD rea ctor. The N sources employed with these Ta precursors have been either NH 3 alone or NH 3 with H 2 Chemical vapor deposition of TaN x from TaCl 5 and NH 3 at the low deposition temperature of 350 C is not suitable for diffusion barrier application because the films had very high cm) and high Cl contamination (4.5 at. %). 90 Since the Ta Cl bond is very strong, higher energy is required to cleave the bond. When TaN x films are deposited using TaF 5 and NH 3 at 350 C, films have a relativ cm. However, considerable F contamination (4.0 at. % F) was observed and the step coverage of the film was poor (75%). 90 The poor step coverage for TaN x films deposited by TaF 5 can be explained by considering the high growt h rate of deposition. Films deposited with TaF 5 had a growth rate of 1.5 nm/sec, which is high for a typical CVD process. 90 TaN x deposition from thermal CVD using TaCl 4 and TaF 5 has not been pursued further, possibly because of the high temperature requi red to deposit TaN x films with low resistivity and acceptable halogen contamination. TaBr 5 has been studied as a precursor for CVD of TaN x using H 2 and NH 3 co reactants. It has been argued that since Br is a larger atom than Cl and F, impurity incorporate d in the film would have lower diffusivity in TaN x film and the impurity would not influence the diffusion barrier property of the film. Moreover, since the bond energy of Ta Br in TaBr 5 is lower than that for Ta Cl in TaCl 5 lower deposition temperature would be required for the TaBr 5 and NH 3 reaction. Chen et al. 91 reported deposition of TaN x films between 350 and 500 C using TaBr 5 H 2 and NH 3 The films were N rich with N:Ta ratio varying between 1.75:1 and 1.87:1 for deposition between 350 and 500 C, indicating that the N:Ta ratio was not very sensitive to the reaction temperature. The N rich film deposited with TaBr 5 indicates that film is possibly Ta 3 N 5 and not TaN. The O and C contamination in the film was below the detection limit of AES.

PAGE 80

80 Fil ms contained Br as an impurity and the Br content in the film decreased with increase in deposition temperature. The highest Br content of 5 at. % was observed for deposition at 350 C while the Br content in film was below the detection limit of AES for deposition at 500 C. The film crystallinity was strongly dependent on deposition temperature. Films deposited at 350 and 425 C were amorphous whereas films deposited at 500 C were polycrystalline with the peak positions in XRD spectrum indicating the presence of tetragonal Ta 3 N 5 and hexagonal TaN phases. Film resistivity also showed a strong dependence on deposition temperature with the lowest resistivity of cm observed at 500C. The resistivity of the films decreased with increase in deposition temperature. To ascertain the diffusion barrier efficacy of TaN x films, Kaloyeros et al. 92 performed extensive barrier testing on TaN x films deposited at 425 C. The Cu/TaN x /Si stack was annealed between 450 and 650 C to determine the temperature at which barrier film failed. From post anneal XRD analysis, it was confirmed that the barrier film was able to prevent bulk diffusion of copper up to 650 C anneal ing as no Cu 3 Si peaks were observed in the XRD spectrum. X ray photoelectron spectroscopy depth profile also confirmed that copper had not diffused through the barrier film. However, when the Seeco etch pit test was performed to identify trace copper dif fusion through the barrier, films annealed at 600 and 650 C showed etch pits, suggesting that a trace amount of copper had actually diffused through the barrier film. To compare diffusion barrier efficacy of films deposited by PVD and CVD, the authors al so deposited PVD TaN x films with the same N:Ta ratio. TaN x films deposited by PVD did not fail the Seeco etch pit test after annealing at 650 C. The difference in diffusion barrier efficacy of CVD TaN x and PVD TaN x was attributed to the polycrystalline phase difference between CVD and PVD films. The PVD TaN x films exhibited hexagonal TaN phase, which has a N:Ta ratio of 1:1. Since the PVD films had a 1.78:1 N:Ta ratio, the excess N in PVD films

PAGE 81

81 er diffusion. For CVD films, the predominant phase is believed to be Ta 3 N 5 The emergence of Ta 3 N 5 peaks in XRD spectra when CVD TaN x film is annealed at 650 C further supports this assumption. 92 In Ta 3 N 5 the N:Ta ratio is 1.67:1, which means that the in CVD TaN x films. This difference is believed to be responsible for the difference in diffusion barrier efficacy of CVD and PVD TaN x films with similar N:Ta ratio. Overall, the primary concerns for TaN x films deposited from TaBr 5 are their higher resistivity at acceptable deposition temperature, lower recrystallization temperature and deposition of Ta 3 N 5 phase instead of the desired TaN phase. A possible approach that might result in lowering re crystallization temperature, decrease in resistivity and also possibly disrupt the growth of Ta 3 N 5 would be to grow a ternary compound such as TaN x C y using TaBr 5 NH 3 and a C source. The ternary compound could lower the recrystallization temperature and C in the film could disrupt the growth of Ta 3 N 5 phase. PECVD of TaN x has been attempted with TaCl 5 90,93 TaF 5 93 and TaBr 5 93,94 Hillman et al. 90 used N 2 plasma to deposit TaN x thin films. The films had a low halogen contamination with the lowest halogen c ontent of 1 at. % observed for film deposited with TaCl 5 The resistivity of film deposited with TaCl 5 TaF 5 and TaBr 5 cm, respectively for deposition at 350 C. The film composition and texture were not reported. The diffusion b arrier testing of films deposited by these three precursors was done by depositing 20 nm barrier film and annealing the Cu (160 nm)/ TaN x (20 nm)/ Si stack at 550 C for 30 min in N 2 environment. Films deposited from TaCl 5 TaBr 5 and TaF 5 failed after ann eal and copper in silicon was detected by SIMS depth profiling. Hence, PECVD process using halide precursors was not able to deposit good quality diffusion barrier films.

PAGE 82

82 Metal organic sources have also been explored as Ta source for CVD of TaN x The met al organic sources that have been explored so far include tert butylimid o tris(diethylamido) tantalum (TBTDET), pentakis(dimethylamido) tantalum (PDMAT), pentakis(diethylamido) tantalum (PDEAT), ethylimidoethyl(C,N) tris(diethylimido) tantalum and tertiar yamylimidotris(dimethylamido)tantalum (TAIMATA). Tsai et al. 95 97 have extensively studied deposition of CVD TaN x films from TBTDET. The precursor was used as a single source precursor between 450 650 C deposition temperature. Films deposited were po lycrystalline with the N:Ta ratio close to 1. The C and O impurities in the film were less than 10 and 5 at. %, respectively at 600 C deposition temperature. X ray photoelectron spectroscopy analysis indicated that C present in the film was in carbide p hase. The resistivity of film cm at 650 C deposition temperature. Diffusion barrier testing on 120 nm TaN x films deposited at 650 C revealed that trace amount of Cu could diffuse th rough the barrier film after the Cu/TaN x /Si (p + n diode) structure was annealed at 550 C and bulk copper diffusion occurred at 600 C as evidenced by the formation of Cu 3 Si. 97 The authors also deposited 120 nm PVD TaN x film to compare the diffusion barrie r performance of CVD TaN x films vis vis PVD TaN x films. The PVD barrier film showed trace copper diffusion at 600 C and bulk copper diffusion at 650 C. The authors argued that the PVD TaN x showed better diffusion barrier performance than CVD TaN x pri marily because PVD TaN x films had a preferred orientation of (111), which has a higher thermal stability than the (200) orientation observed for CVD TaN x films. However, the film texture should not play such a significant role in diffusion barrier perform ance because the CVD films are polycrystalline and show both (111) and (200) reflections in XRD spectra. 95 A more plausible explanation for poorer barrier performance of CVD films is the columnar

PAGE 83

83 microstructure of the films deposited by CVD at higher temp erature. Overall, CVD TaN x films deposited using TBTDET without any co reactant had several limitations including high deposition temperature, polycrystalline microstructure with columnar grains, high resistivity and poor barrier performance. The use of NH 3 or H 2 co reactants might decrease the deposition temperature and improve film properties of TaN x but such an approach has not been explored yet. Fix et al. 98 used PDMAT with NH 3 to deposit TaN x films between 200 and 400 C deposition temperature. The films deposited with PDMAT had amorphous microstructure. Charging effects were observed when TaN x films were observed under electron microscope suggesting that the films were non conducting. The films were N rich with a N:Ta ratio of 1.7:1. The film ha d significant H content as measured by hydrogen forward recoil spectroscopy. The O and C contamination in the film was less than the detection limit of RBS (typically 3 4 at. %). The resistivity of the film was greater than 10 6 cm and is indicative of the growth of the non conducting Ta 3 N 5 phase. The growth of highly resistive Ta 3 N 5 makes the PDMAT precursor unsuitable for diffusion barrier deposition. PDEAT has been used with and without ammonia to deposit TaN x films. When PDEAT was used without NH 3 the films deposited between 275 and 400 C were Ta rich films with significant C content (> 20 at. %) indicating that the films were actually TaN x C y 99 X ray diffraction spectra indicated that the films deposited above 275 C were polycrystalline with peak positions corresponding to TaN(111) phase. The films deposited at low deposition temperature 4 cm was observed for film deposited at 400 C. Cho et al. 100 investigated the use of NH 3 as a co reactant to deposit TaN x films. The film deposited at 325 C was polycrystalline wi th TaN(111) as the

PAGE 84

84 dominant phase. The addition of NH 3 dramatically increased the film N content and the N:Ta ratio for films deposited with NH 3 was 1.75:1. The addition of NH 3 also decreased the C content of the film to below 10 at. %. The Seeco etch p it test on 50 nm thick barrier film showed that the barrier film was able to prevent copper diffusion after annealing at 600 C. The comparison between films deposited with and without NH 3 indicated that films deposited with NH 3 were better at preventing copper diffusion. It was argued that finer grain structure together with dense microstructure of films deposited with NH 3 was responsible for enhancing the diffusion barrier quality. The major drawbacks of films deposited with PDEAT and NH 3 were the high resistivity (6 cm) and poor conformality of the films. When a mixture of EtN=Ta[NEt 2 ] 3 and Ta[N(Et 2 )] 5 was used with H 2 to deposit thin films between 375 and 500 C, the resulting films were C rich and the N content of the film was less than 1 at. %. 101 However, when a mixture of EtN=Ta[NEt 2 ] 3 and (Et 2 N) 3 Ta[(NEt CMeH] was used without any co reactant to deposit thin films between 500 and 600 C, the films had a significant N and C with the Ta:C:N ratio of 1:1:1 at 600 C. 102 The XRD measurement of films showed pe aks corresponding to polycrystalline cubic TaN even though the film had significant C content. The report did not mention film resistivity and no diffusion barrier testing was done on the films. TAIMATA has been used with NH 3 to deposit barrier films at 2 50 C. 103 The report focused on the barrier efficacy and its integration with SiOC:H dielectric. Film characterization was not discussed in detail. Bias temperature stress testing at 0.75 MV/cm and 350 C on 10 and 15 thick barrier film showed that th e TaN x film was able to prevent copper diffusion. The precursor shows promise as a diffusion barrier but a detailed analysis of films deposited from TAIMATA is yet to be reported.

PAGE 85

85 Overall, the biggest challenge for CVD TaN x deposition is the growth of ins ulating Ta 3 N 5 phase. Since Ta is in the +5 oxidation state in both the precursor and the deposited film, the use of a stronger reducing agent such as dimethylhydrazine might help in reduction of Ta to the +3 oxidation state to form conducting TaN films. 2 .6.2 A tomic L ayer D eposition of Tantalum Nitride The precursors utilized for ALD TaN x have been the same as those used for CVD TaN x (Table 2 6). TaCl 5 has been used with NH 3 to deposit TaN x thin films by thermal ALD. 104,105 The films deposited with the T aCl 5 and NH 3 are N rich and polycrystalline. The XRD spectrum for films deposited at 400 C show the presence of the tetragonal Ta 3 N 5 phase. The film growth rate is 0.22 0.24 /cycle and the temperature window for the ALD process is 300 500 C. The films contain less than 5 at. % Cl and less than 2.5 at. % H impurity. Similar to CVD, ALD using TaCl 5 and NH 3 shows that the Cl impurity in TaN x films decreases with increasing deposition temperature. The resistivity of the film varies from 0.5 10 6 200 10 6 cm. The film is highly resistive because of the presence of the non conducting Ta 3 N 5 phase. Similar to CVD, ALD from TaCl 5 also shows that the reducing power of NH 3 is not sufficient to reduce Ta from the +5 oxidation state in TaCl 5 to the + 3 oxidation state in TaN. To aid the reduction of Ta, Zn has been utilized as an additional reducing agent in the ALD process for deposition using TaCl 5 105 When a Zn pulse is used between the TaCl 5 and NH 3 pulses in the ALD process, the film growth rate is 0.2 /cycle at 400 C and 0.15 /cycle at 500 C. The characteristic temperature window of ALD could not be established because the growth rate was not constant between 400 and 500 C. The films deposited were cubic TaN and impurity content in the fi lm was lower than that for films deposited without a Zn pulse. The impurity content of film was 0.1 4.0 at. % Cl, 3.0 4.0 at. % O, less than 0.5 at. % H and less than 0.5 at. % Zn. The films were quite conductive and the lowest resistivity of the fil m was 9 00

PAGE 86

86 cm. The main challenges for deposition with Zn as a reducing agent are the polycrystalline films deposited with the ALD process, incorporation of electrically active Zn impurity in the films and higher deposition temperature. As discussed earlier, Zn is an electrically active impurity and its diffusion in the device can cause severe degradation of device performance. Deposition below 400 C is not possible with Zn because Zn has a low volatility and its delivery below 400 C is not possible. There are a number of other reducing agents that have been used with and without NH 3 to deposit TaN x from TaCl 5 and TaBr 5 These reducing agents include DmHy 80 tert butyl amine (with and without NH 3 ) and allylamine (with and without NH 3 ). 106 When DmHy was use d as a reducing agent, no significant improvement in film properties was achieved as compared to NH 3 based ALD. The film deposited from TaCl 5 and DmHy was insulating Ta 3 N 5 and the resistivity of the film was too high to be measured by the standard four po int probe technique. t ert Butyl amine has been used with and without NH 3 to deposit TaN x films from TaCl 5 and TaBr 5 106 When tert butyl amine is used without NH 3 deposition with TaCl 5 has a higher growth rate and yields films with lower resistivity as co mpared to TaBr 5 Films deposited with TaCl 5 are amorphous at 350 C deposition temperature but polycrystalline TaN films are deposited above 400 C. The films with TaCl 5 and tert butyl amine show Cl impurity between 3 and 8 at. %. Film impurity content and resistivity decreases with increase in deposition temperature. Similar to CVD, low temperature ALD utilizing halide precursors gives highly resistive films with high impurity content. Good quality films with resistivity bel cm are obtained at 500 C deposition temperature but such high temperature is not suitable for barrier film deposition. To improve the film quality, NH 3 is used along with tert butyl amine as

PAGE 87

87 a reducing agent. The process does show improvement with respect to impurity content in the film and resistivity. However, amorphous film deposition could not be obtained for this process. When allylamine is used with and without NH 3 to deposit TaN x films from TaCl 5 the films deposited were polycrystalli ne TaN. 106 The films were highly resistive with lowest resistivity of cm. Thus, the process with allylamine did not show any significant improvement over the tert butyl amine process. Plasma assisted ALD of TaN x films using TaCl 5 and H 2 + N 2 plasma has b een reported. 107 Depending on the partial pressure of N 2 in plasma, the film N content could be varied to deposit Ta rich Ta 2 N films, stoichiometric TaN films or N rich Ta 3 N 5 films. The growth rate increased with deposition temperature indicating that no characteristic ALD process window was observed for PAALD process. The TaN films deposited were quite conductive with resistivity varying cm. Films had a high H impurity of about 11 at. % and low Cl content of less than 3 at. %. Similar to thermal ALD, PAALD also showed decrease in Cl impurity with incre ase in deposition temperature. Overall, PAALD was able to deposit good quality films that might be suitable for the diffusion barrier application. However, the process could not yield amorphous films. Metal organic precursors such as TBTDET, TDMAT and TA IMATA have been used to deposit TaN x films by ALD. Chemical vapor deposition using these precursors deposited films that had high C and O impurities and high resistivity. The ALD process, if used with the same CVD precursors, is expected to yield films w ith low C content and dense microstructure at lower deposition temperature. When TBTDET is used with NH 3 to deposit TaN x films in ALD mode, amorphous TaN x films are deposited between 150 and 300 C. The films deposited are Ta rich and the N:Ta ratio

PAGE 88

88 is 0. 73:1. The films contained low C contamination (< 5 at. %) but had a higher level of O impurity (ca. 17 at. %). The films were highly resistive (> 10 6 cm at 260 C) despite high Ta content in the film. High resistivity could possibly result from the formation of Ta 3 N 5 high O contamination and/or low density of the film (3.6 g/cc). The resistivity of the film decreases to cm when the depos ition temperature is increased to 450 C. 108 Choi et al. 109 reported that the temperature window for the ALD process is between 200 and 250 C, but the growth rate versus temperature curve shows that no saturation region can be conclusively determined. T he growth rates of 1.1 and 0.4 /cycle have been reported for ALD using TBTDET and NH 3 109,110 If deposition is done in the ALD process window and the growth occurs through the ALD mode, the growth rate per cycle should be independent of all other reactor conditions such as reactor pressure. The discrepancy in the growth rates reported by Choi et al. 109 and Park et al. 110 Overall, TBTDET is not a suitable precursor for diffusi on barrier deposition because of the deposition of highly insulating films. PAALD of TaN x films from TBTDET and H radical has been reported. 110 This ALD process is unique in that it employs TBTDET as a single source precursor as H radicals are used for r eduction of TBTDET. The intent behind using H radicals is that all the N atoms attached to Ta atoms via single bonds are cleaved by H radical while the N atom attached to Ta via a double bond survives the H radical exposure, leading to the formation of st oichiometric TaN film. The PAALD process showed a growth rate of 0.8 /cycle. The resulting TaN x films are Ta rich with N:Ta ratio of 3:5. The films deposited had high C content (20 at. %) and it could be argued that the films were actually TaN x C y inste ad of TaN x The films were polycrystalline and had low cm. The films were considerably denser than those deposited with

PAGE 89

89 thermal ALD. Overall, the plasma process improved the film property of films deposited using TBTDET. The proce ss, however, deposited polycrystalline films. No diffusion barrier testing was reported for films deposited via this process. Thermal ALD from PDMAT and NH 3 has been used to deposit TaN x films. 111 114 The process shows that the growth rate is constant a t 0.3 /cycle between 225 and 300 C, indicating that the growth in this temperature window is truly in ALD mode. 114 The film composition and microstructure strongly depends on the substrate used for deposition. 113 The films deposited on Si, SiO 2 and Cu are N rich, with N:Ta ratio of 2:1 for deposition at 275 C. 111,113 However, when the same process is used to deposit TaN x films on Ta, the films are Ta rich with N:Ta ratio of 1:2. The films have a high level of H impurity that can be removed by anneali ng the film at 600 C for 1 hr. The C and O impurity in the film is 2 and 5 at. %, respectively. Even though the films have high N content, the microstructure analysis indicates the presence of cubic TaN phase. The excess N is believed to reside in Ta v acancies in the Ta sublattice. 111 The films deposited are nanocrystalline with average grain size of 4 nm. 10 nm thick TaN x film was able to prevent Cu diffusion after annealing at 750 C for 30 min in vacuum. The biggest drawback of the film was its ex tremely high resistivity, which could not be measured by standard four point probe technique. Overall, the TaN x films deposited with PDMAT and NH 3 in ALD mode not suitable for barrier application primarily because of its extremely high resistivity. To de posit more conductive films using PDMAT, PAALD has been investigated. 112,114,115 Kim et al. 115 investigated the effect of plasma on the deposition of TaN x films. When H 2 plasma is used, the film growth rate saturates between 225 and 300 C, indicating th at ALD growth mode is achieved with PDMAT and H 2 plasma. The deposited film is Ta rich, with N:Ta ratio between 0.7 and 0.8. The film C and O content is ca. 15 and 15 at. %, respectively. The films

PAGE 90

90 are conductive cm observed for deposition at 250 C. The microstructure analysis by TEM reveals that the film contains 2 3 nm cubic TaN crystallites in an amorphous matrix. Diffusion barrier testing of the film shows that 3 nm TaN x film is a ble to prevent copper diffusion after annealing at 750 C in vacuum. This process could be a potential candidate for deposition of diffusion barrier film. When N 2 + H 2 plasma is used with PDMAT, the film N:Ta ratio varies between 1.2 and 1.6, depending o n the partial pressure of N 2 in the plasma. A sharp decrease in C and O content is also observed, with the impurity content decreasing with increasing N 2 partial pressure. But the additional N in the film results in an increase in film resistivity and th cm was observed at low N 2 partial pressure. Diffusion barrier testing for TaN x film shows that 0.6 nm barrier film is able to prevent copper diffusion after annealing at 800 C. From the diffusion barrier standpoint, N 2 + H 2 plasma yields better films than films deposited with H 2 plasma. Overall, PAALD from PDMAT provides various options to modify film composition and further investigation on integration of TaN x films could provide additional information on suitability of the film for diffusion barrier application. The effect of plasma exposure on underlying layers along with conformality of the film also needs to be investigated. Metal organic precursor TAIMATA has also been investigated with NH 3 co reactant for deposi tion of TaN x thin films by ALD. 109 Films deposited by thermal ALD between 150 and 175 C exhibit a growth rate that is constant at 0.2 /cycle implying that the growth is self limiting in this temperature window. The films deposited are N rich with N:Ta ratio of 1.31. The films had minimal C contamination (< 5 at. %) but the O contamination was high (ca. 20 at. %). The XRD spectra indicate the presence of the insulating Ta 3 N 5 phase. The films deposited with thermal ALD had extremely high resistivity an d hence are not suitable for diffusion barrier

PAGE 91

91 application. When H 2 plasma is used instead of NH 3 to deposit TaN x films by PAALD, cm were deposited for growth at 250 C. 116 The low resistivity of PAALD TaN x films is probably because of a more favorable N:Ta ratio of 0.71 and the absence of insulating Ta 3 N 5 phase. The films also had a high C content of about 15 at. %. The growth rate per cycle is 1.2 /cycle. Films deposited by PAALD are polycrystalline wi th the XRD peak position indicating the presence of cubic TaN phase. The conformality of the films deposited with PAALD was 94% achieved in a 10:1 aspect ratio 0.25 m via. The major drawback of PAALD using TAIMATA is that the process could not deposit a morphous TaN x films. To summarize, both inorganic and metal organic precursors have been used to deposit TaN x films by ALD. The biggest challenge for inorganic precursors such as TaCl 5 is the formation of the insulating Ta 3 N 5 phase instead of the conducti ve TaN phase. Moreover, the films deposited with halide precursors contain halogen impurities and the use of stronger reducing agents or plasma could not completely eliminate the halogen incorporation in the film for low deposition temperature. The use o f metal organic precursors was also more likely to deposit Ta 3 N 5 phase at low temperature and the use of plasma was essential to obtain conducting TaN x thin films with a favorable N:Ta ratio. Further investigation needs to be done on PAALD to achieve the deposition of TaN x films on different substrates and determine if plasma has any adverse effects on the substrate. 2.7 Tungsten Nitride as Diffusion Barrier 2.7.1 Chemical Vapor Deposition of Tungsten Nitride Table 2 7 shows the precursors that have been u sed to deposit WN x thin films by CVD. WCl 6 has been used with NH 3 and H 2 to deposit WN x thin films for application in catalysis. 117,118 However, the lowest deposition temperature for this precursor is 500 C, so this precursor has

PAGE 92

92 not been explored for d iffusion barrier application. Another inorganic precursor, WF 6 has been explored for deposition of WN x films. This precursor, when used with NH 3 as co reactant, results in the formation of WF 6 :NH 3 adduct even at temperature as high as 600 C. 119 The fi lms cm is observed for films deposited at 350 C. 120 To prevent the adduct formation, H 2 is used as an additional reducing agent. The resulting process deposits polycrystalline WN x films and the lowest deposition temperature is 450 C. Because of F impurity in the films, relatively high deposition temperature and high film resistivity, this process also has not been further investigated for diffusion barrier film deposition. A modification to th e process is the use of SiH 4 as a reducing agent. 120 The lowest deposition temperature for this process is 385 C. The resulting films are W rich and the N:W ratio decreases with an increase in SiH 4 partial pressure. X ray diffraction measurement indica tes that the films are polycrystalline and the peaks in XRD spectra indicate the presence of W 2 N. Addition of SiH 4 as co cm and film resistivity decreases with an increase in SiH 4 partial pres sure. Diffusion barrier testing revealed that 6 nm thick barrier film can prevent Cu diffusion after annealing at 450 C for 4 hours. However, the study by Gonohe found that adhesion of the WN x film to low k dielectric film was extremely poor. 120 X ray photoelectron spectroscopy analysis showed that the presence of F at the WN x / low k interface caused the delamination of the WN x film. The presence of F at the interface was due to the reaction of WF 6 with the dielectric surface. Chemical pretreatment of the dielectric film prevented the reaction of WF 6 with the dielectric film and no F impurity was detected at the barrier dielectric interface. Details of the chemical pretreatment process were not discussed in the paper.

PAGE 93

93 PECVD using WF 6 as tungsten sourc e has been used to deposit WN x thin films. NH 3 plasma, NH 3 + H 2 plasma and N 2 + H 2 plasma have been used as the N source. When deposition is done with NH 3 plasma 121 123 the resulting WN x films are N rich. The film composition can be varied by changing the NH 3 :WF 6 feed ratio, deposition temperature and plasma power. X ray diffraction measurement reveals that the films are polycrystalline with peak positions indicating W 2 N. Gas phase reaction between reactants can cause particle format ion during deposition and optimization of the NH 3 :WF 6 flow ratio is necessary to avoid this problem. Films can be deposited at temperature as low as 250 C, however the films have a high F contamination at low deposition temperature. The lowest F contami nation of 0.1 at. % is obtained for deposition at 625 C. 121 The major problem of using this chemistry to deposit WN x thin film is the formation of an interfacial W layer about 5 nm thick due to the high reactivity of WF 6 and Si. This can also result in etching of the substrate. 122 On SiO 2 surface, a WO 3 layer is first formed prior to the deposition of WN x film. Another issue with the PECVD process is the poor conformality. Film conformality of only 33% is obtained for deposition in a 0.14 m 9:1 aspec t ratio feature. 123 Hence this process is not suitable for diffusion barrier deposition. PECVD of WN x using WF 6 has also been done using both NH 3 and H 2 plasma instead of just NH 3 plasma as discussed previously. 124 130 The advantage of PECVD is that by a ltering the NH 3 :WF 6 ratio and the plasma power, the resulting WN x films could be deposited as stoichiometric W 2 N, W rich WN x or N rich WN x 130 Films can be deposited at temperature as low as 150 C. 129 An increase in the NH 3 :WF 6 ratio results in a decrea se in growth rate, increase in crystallinity and better film adhesion to the substrate. The films have low resistivity and the resistivity of the film depends, among other factors, on the film N content and deposition temperature. Due to higher F incorpo ration at low deposition temperature, the film resistivity

PAGE 94

94 cm has been reported for WN x films deposited by this process. 125 Particle formation during deposition has also been reported and the NH 3 :WF 6 flow ratio has to be optimized to avoid this problem. 129 The main drawback of this process is that pitting of the Si surface is observed due to the reaction of WF 6 with the underlying Si substrate. 128 WF 6 has been used with N 2 + H 2 plasma to deposit WN x thin films. 124 130 N 2 is used as the N source to prevent particle problems associated with the use of NH 3 By changing the plasma power, N 2 /H 2 ratio and deposition temperature, films with varying N:W ratio can be deposited by PECVD. As compared to the NH 3 plasma process, the N 2 plasma process requires higher deposition temperature to deposit WN x films. The film resistivity depends strongly on the N:W ratio and resistivity for stoichiometric W 2 N and W rich WN x films varies be tween 200 and 400 cm. 131 The stoichiometric W 2 N and W rich WN x films are amorphous while N rich WN x films are polycrystalline. Similar to the NH 3 plasma process, the N 2 plasma process also resulted in the presence of F at the barrier substrate interface. The F contami nation at the interface was more severe for SiO 2 substrate than Si substrate. 32 The F contamination resulted in poor adhesion of the barrier film with Si and SiO 2 The use of N 2 plasma could not prevent the deposition of highly undesirable F impurity at the barrier substrate interface. The three main metal organic precursors that have been used to deposit WN x thin films are bis( tert butylimido)bis( tert butylamido) tungsten (TBTBW), tungsten hexacarbonyl and pentacarbonyl tungsten( 1 methylbutylisonitril e) Depositions with TBTBW precursor have been attempted using inert gas argon and co reactant NH 3 31,132,133 Films deposited with NH 3 have a lower C content and higher N content as compared to those deposited without NH 3 132 The lowest deposition tempera ture for this precursor is 450 C. When the deposition is done with Ar

PAGE 95

95 carrier gas at 600 C, the film contains ca. 70 at. % W, 10 15 at. % N, 10 15 at. % C and 5 at. % O as measured by AES. The apparent activation energy for film growth in the kinet ically controlled growth regime is 0.9 eV. 31 Films were polycrystalline and the XRD spectra indicated W 2 N in the film. 133 The film was highly resistive at low deposition temperature and the resistivity of film decreased with increase in deposition temperature. The cm for deposition at 600 C. The possible reaction pathways for this precursor have also been investigated by Crane and coworkers. 134 Because of the relatively high deposition temperatur e, high resistivity of the film and polycrystalline microstructure, TBTBW is not a good precursor candidate for deposition of diffusion barrier thin films. Tungsten hexacarbonyl has been successfully used to deposit WN x thin films with NH 3 as co reactant. 1 35 137 Films can be deposited at temperature as low as 200 C. Kelsey et al. 135 reported that the films were amorphous for deposition below 275 C as measured by XRD and TEM. However, Lee et al. 136 reported that the films deposited at 250 C were weakly W 2 N. This discrepancy could result from the different deposition conditions used for the deposition of WN x films. The apparent activation energy for film growth is 1 eV for growth in th e kinetically controlled growth regime. The N:W ratio for the film was 1:1, however the presence of the W 2 N polycrystalline phase in XRD spectra suggests that the films were actually W 2 N polycrystals embedded in a N rich amorphous WN x matrix. The films h ad C and N impurities of less than 10 at. % and the impurity content of the film decreased with increase in deposition temperature. Low C impurity in the film was possibly due to weak coordinate covalent bonds between CO and W in the precursor. Kelsey et al. 135 reported that the films were also highly conductive and had their

PAGE 96

96 cm for deposition at 400 C, while Lee et al. 136 reported film cm for the same deposition temperature. This discrepancy in res istivity could be due to the different film thicknesses used by Kelsey et al. and Lee et al. While Kelsey et al. measured resistivity for relatively thick WN x film (50 nm), Lee et al. measured resistivity for a much thinner film (15 nm). As discussed ear lier, resistivity could increase for thinner films because of the increase in electron scattering due to surface and grain boundary scattering. The process deposited highly conformal films with conformality of 90 % observed for deposition of 50 nm thick f ilms in a 0.25 m via with an aspect ratio of 4:1. 135 To determine the diffusion barrier effectiveness, a Cu/WN x (15nm)/Si stack was annealed in Ar environment for 1 hour. 136 The post anneal XRD and sheet resistance measurements indicated that the film w as able to prevent copper diffusion after 600 C annealing. After annealing at 620 C, the XRD spectra showed the emergence of Cu 3 Si peaks indicating bulk diffusion of copper through the barrier film. Tungsten carbonyl is hence a good candidate for diffu sion barrier deposition. Further tests are needed to evaluate the effectiveness of the WN x film in preventing trace copper diffusion. Another precursor, pentacarbonyl tungsten (1 methylbutylisonitrile), has also been used with NH 3 to deposit WN x thin film s. 138 The films were stoichiometric W 2 N with amorphous microstructure. Detailed film characterization or diffusion barrier testing for this precursor has not been reported. 2.7.2 A tomic L ayer D eposition of Tungsten Nitride Table 2 8 shows the precursors that have been used to deposit WN x films by ALD. George and coworkers 139,140 first reported the deposition of WN x thin films by ALD using WF 6 and NH 3 as precursors. The films were deposited between 327 and 527 C. Minimum growth temperature for ALD grow th was 327 C because below this temperature, WF x :NH 3 adduct formation was observed by FTIR due to incomplete reduction of WF 6 by NH 3 X ray

PAGE 97

97 photoelectron spectroscopy analysis showed that the films had a N:W ratio of 1:3. The films had low C and O conta minations of ca. 5 and 3 at. %, respectively. No F contamination was detected by XPS. The ALD process had a deposition rate of 2.5 /cycle. The characteristic ALD process window was observed between 325 and 525 C. 140 X ray diffraction measurement show ed that the films deposited were polycrystalline with the peak positions indicating the presence of W 2 W 2 N peaks in XRD spectra was 110 The films deposited have high resistivity of 4500 cm. To deposit WN x films with lower film resistivity, Kim et al. 141 used dibo rane (B 2 H 6 ) as an additional reducing agent along with NH 3 The use of B 2 H 6 resulted in films with significantly lower cm). The growth rate of the film was 2.8 /cycle and the N:W ratio was 0.82. The XRD measurement revealed that t W 2 N W 2 WN. X ray photoelectron spectroscopy measurement indicated that in addition to WN x the film also contain ed the oxynitride phase WN x O y The source of oxygen in the film was not discussed. No diffusion barrier testing was reported for these films. As discussed earlier, a major drawback of the process using WF 6 is that it can react with the underlying Si subs trate to form a thin pure tungsten film before the deposition of WN x This could increase the total thickness of the barrier film and is not desirable. To prevent or minimize the reaction of WF 6 with the underlying Si substrate, Sim et al. 142,143 used pu lsed plasma enhanced ALD (PP ALD). Plasma was used during the NH 3 pulse only because if plasma is applied during the WF 6 pulse, W particles could be generated and subsequently incorporated into the film. The process was able to deposit WN x film on a Si s ubstrate without the formation of

PAGE 98

98 pure W interface. However, the films also contained F as an impurity though the F content was not quantified by the authors. The only metal organic precursor that has been used to deposit WN x films by ALD is bis( tert buty limido) bis(dimethylamido)tungsten. 45,144 The precursor, when used with NH 3 in ALD mode, deposits WN x thin films. No ALD process window was observed between the deposition temperature of 250 and 300 C. The N:W ratio was 1:1 and films did not contain C impurity as indicated by AES and XPS measurements. X ray diffraction and HRTEM measurements indicated that the films were amorphous. The films were stable after annealing at 700 C. Subsequent heating at 725 C resulted in decomposition of film into pur e W and N 2 gas. The growth rate was 1 /cycle. The films were very smooth with a RMS roughness of 3.3 The ALD process also demonstrated 100 % step coverage in a 200:1 aspect ratio feature. WN film as thin as 1.5 nm was able to prevent copper diffusi on after annealing at 600 C. The only drawback of this film was its high resistivity (1500 cm). The resistivity of the film could be reduced by annealing the film at 700 C in forming gas, but annealing at such high temperature is not desirabl e for diffusion barrier application. 2.8 Tungsten Nitride Carbide as Diffusion Barrier 2.8.1 Atomic Layer Deposition of WN x C y Numerous studies on ALD WN x C y growth have been reported. Most of the reports of WN x C y r TM ALD deposition tool. Almost all of the reports on ALD WN x C y have employed halide chemistry with WF 6 as the tungsten source, NH 3 as the nitrogen source and triethyl boron (TEB) as the carbon source. The composition of films deposited was W:N:C = 55:15 :30. 145 The concentration of impurities such as O, F and B in the film was low (< 0.5 at. % O, < 0.5 at. % F and < 0.5 at. % B). 146 However, F was detected at the interface of Si and WN x C y 89 This is undesirable since halogens present at the interface c an

PAGE 99

99 severely affect adhesion of the films. In fact, the adhesion energy of WN x C y with SiO 2 is 1.5 J/m 2 as measured by the four point bend test. 77 The low adhesion strength, however, did not cause delamination of the film during the CMP process. 76 Despite the use of a halide precursor, the deposition of WN x C y did not show any pitting on copper and WN x C y had good adhesion with copper. 145 The WN x C y film was nanocrystalline with the peak position indicating the presence of WN x and WC x or a single WN x C y phase. The average crystallite size was between 3 and 7 nm. 147 Unlike TiN films deposited by ALD, WN x C y films did not exhibit a columnar grain structure, which is desirable because a columnar structure could provide short diffusion paths for copper to diffuse. The density of the film was 15.37 g/cc, which is close to the bulk value. The dense microstructure of WN x C y was believed to be responsible for minimal O diffusion through the film after exposure to air. The film was stable after annealing at 700 C. A fter annealing at 800 C, a multiphase mixture containing metallic W, WC and WC x emerges as evidenced by the corresponding peaks in the XRD spectrum. The N from the film is lost as N 2 gas after annealing at 800 C. While WC x reacts with Si to form W 5 Si 3 at 700 C 148 no silicide form ation is observed for WN x C y films even after annealing at 800 C 89 The growth rate of the film was 0.8 0.9 /cycle. The resistivity of the film was 300 cm for films deposited between 300 and 400 C. Growth of ALD WN x C y film was sensitive to the surface on which the films were deposited. 149 This is particularly important because in dual damascene structures, the diffusion barrier film is deposited simultaneously on three different surfaces, (i.e. copper, dielectri c (SiO 2 or low k dielectric) and the etch stop layer (SiC or Si 3 N 4 )). A transient growth regime is observed up to film thickness of 5 nm for growth on SiO 2 and SiC, however deposition on Si 3 N 4 shows a much shorter transient regime. Surface treatment has been proposed to speed up the

PAGE 100

100 nucleation and growth of ALD films. 150 WN x C y film deposited on PECVD SiO 2 pretreated with NH 3 plasma showed complete surface coverage after 40 ALD cycles whereas WN x C y deposited on untreated PECVD SiO 2 required 50 cycles to a chieve complete surface coverage. The WN x C y film was compatible with dielectric films such as SiO 2 and SilK TM and etch stop layers such as SiC and Si 3 N 4 145 However, for ultra low k films, the precursor diffuses through the pores of the dielectric. Diffu sion of the precursor through both HSQ and MSQ dielectric films with pore size of 4 5 nm has been reported. 73 Nitrogen plasma was able to seal the pores of these dielectric films and prevent the diffusion of the precursor(s) into the dielectric film. S urface treatment of SilK TM dielectric by O 2 inductively coupled plasma ( ICP ) in an oxygen rich environment and reactive ion etch ( RIE ) done in a nitrogen rich environment have been investigated to seal the pores of the dielectric film. 151 Hydrophilic grou ps created on the dielectric surface during surface treatment are expected to enhance ALD growth N 2 RIE was able to densify the dielectric surface and close the pores of the dielectric film. However, since the O 2 ICP does not have any ion bombardment, t he dielectric films remained semi permeable after surface treatment. WN x C y films on plasma treated polymer were smoother, more so for film deposited on an N 2 RIE treated surface than film grown on O 2 ICP treated surface. Films deposited by ALD were conti nuous after 10 nm for untreated polymer, 3.5 nm for the O 2 ICP treated surface and 1.4 2.3 nm for the N 2 RIE treated surface. The thermal stability of ALD WN x C y as measured by XRD showed that the film failed after annealing at 700 C as evidenced by the formation of Cu 3 Si. 147 However, the etch pit test, which can detect trace diffusion of copper through the barrier film, showed that the diffusion of copper through the WN x C y film occurs after annealing at 600 C for 30 min. 89 Electromigration resistance of copper deposited on WN x C y has been compared with copper deposited on PVD Ta,

PAGE 101

101 which is the liner material currently used in industry. 76 For the electromigration test performed at 325 C and 2 mAmp current, copper film deposited on WN x C y has better elect romigration resistance than copper film deposited on PVD Ta. The dominant failure mechanism of copper deposited on WN x C y in dual damascene structure was v oiding in M2 trench starting at the via and prop agating along the capping layer. Bias temperature st ress tests done at 300 C and 0 .7 MV/cm bias showed that 2.7 nm WN x C y film failed after 30 min stress, while 5 nm WN x C y barrier did not fail These reports confirm that WN x C y can prevent copper diffusion under normal operating conditions and has good elec tromigration resistance. A major integration problem of WN x C y barrier film was its galvanic corrosion by the H 2 O 2 based slurry used in CMP of copper. The l arge potential difference between Cu and WN x C y in H 2 O 2 causes the c orrosion of WN x C y by H 2 O 2 The c orrosion occurs by dissolution of W as tungsten oxide. To overcome this problem, a CMP slurry based on HNO 3 has been proposed. 152 The use of HNO 3 based slurry can reduce the WN x C y corrosion. Additives such as monosaccharides or organic acids can be used to slow the excessive loss of copper by HNO 3 2.8.2 Chemical Vapor Deposition of WN x C y WN x C y is a very promising material for diffusion barrier application as discussed above. However, the use of WF 6 leads to F deposition at the barrier substrate interfa ce 89 leading to poor adhesion of the barrier film to the substrate. Moreover, WF 6 has problems related to its handling and storage. An alternate method for the deposition of WN x C y is thus desirable. Three different metal organic precursors have been use d to deposit WN x C y thin films [ Cl 4 (RCN)W(N Ph ) ( 1 a R = CH 3 and 1 b R = Ph), Cl 4 (RCN)W(NC i Pr ) ( 2 a R = CH 3 and 2 b R = Ph) and Cl 4 (RCN)W(NC 3 H 5 ) ( 3a R = CH 3 and 3b R = Ph) ]. 153 157 From the study of films deposited using these precursors, it was evident t hat the N C imido bond strength has a significant effect on the N content of the deposited film. Films deposited with 1a,b were W rich and poor in N

PAGE 102

102 because the strong N C imido bond results in the cleavage of W N bond. Films deposited with 2a,b and 3a,b have a higher N content due to relatively weaker N C bond strength, even though the N content was below the desirable levels for diffusion barrier application. Previous studies on WN x C y deposition by MOCVD using 1a,b 2a,b and 3a,b have shown that the fil ms have low N content. As discussed earlier, excess N in diffusion barrier films is boundaries. Another concern for deposition using precursors 1a,b 2a,b and 3a,b wa s that the lowest temperature at which films could be deposited was 450 C or higher It is desirable to deposit WN x C y films at even lower temperature as low k dielectrics are thermally unstable above 450 C. In the present work, further investigation is done to deposit WN x C y thin films with alternate chemistries to deposit films with higher N content at temperature below 450 C. To improve film N content, two different approaches have been used. One is the use of precursors with higher N:W ratio such a s W(N i Pr)Cl 3 [ i PrNC(NMe 2 )N i Pr] ( 4 ) and (CH 3 CN)Cl 4 W(NNMe 2 ) ( 5 ). The other approach is to use co reactant NH 3 to increase film N content. Chapter 4 investigates the use of NH 3 with 3a,b to increase the film N content. Chapter 5 discusses the results for f ilm deposition using 4 and H 2 Chapter 6 discusses the results obtained from thin film deposition using 4 and NH 3 (with and without H 2 ). The results obtained from thin films deposited with 5 and H 2 are discussed in Chapter 7. To facilitate the understan ding of the gas phase reactions of 5 Chapter 8 discusses results from quantum mechanical calculations using density functional theory. Chapter 9 provides a summary of work reported in this manuscript and discusses future work that would enhance our under standing of WN x C y thin film deposition using MOCVD.

PAGE 103

103 Table 2 1. Melting temperature of Ti, Ta, and W and their nitrides and carbides39 Transition Metal T m ( C) Titanium 1677 Tantalum 2997 Tungsten 3380 Titanium Nitride 2949 Tantalum Nitride 3093 T ungsten Nitride < 800 for WN Titanium Carbide 3067 Tantalum Carbide 3983 Tungsten Carbide 2776 Table 2 2. Techniques used for characterization of diffusion barrier thin films Film Property Characterization Technique Acronym(s) Microstructure X ray d iffraction XRD Grazing Incidence XRD GIXD, GIXRD Scanning electron microscopy SEM Electron diffraction in transmission electron microscopy TEM Composition Auger electron spectroscopy AES X ray photoelectron spectroscopy XPS, ESCA Electron disper sive X ray spectroscopy EDS, EDX Secondary ion mass spectroscopy SIMS Rutherford back scattering RBS Elastic recoil detection analysis ERDA Film Thickness Scanning electron microscopy SEM Transmission election microscopy TEM Profilometer Rut herford backscattering RBS X ray reflectivity XRR Density X ray reflectivity XRR Rutherford backscattering RBS Resistivity Four point probe Surface roughness Atomic force microscopy AFM Optical profilometer Stylus profilometer X ray refl ectivity XRR Atomic Bonding X ray photoelectron spectroscopy XPS Film Adhesion Four point bend test, S cotch tape test

PAGE 104

104 Table 2 3. Precursors used for deposition of TiN x thin films by CVD Technique Ti source N source Reference CVD Inorganic Precurso r TiCl 4 NH 3 51 55, 158 160 TiI 4 56, 57 Metal organic Precursor tetrakis (dimethylamido) titanium (TDMAT) 60 62, 68, 69, 72, 161 tetrakis (dimethylamido) titanium H 2 or N 2 + post growth plasma exposure 68, 69, 162 164 tetrakis (dimethyl amido) titanium NH 3 60, 64 66, 165 tetrakis (dimethylamido) titanium hydrazine 67 tetrakis (diethylamido) titanium (TDEAT) inert gas 70, 64, 68, 166 tetrakis (diethylamido) titanium NH 3 64, 66, 70, 71 PECVD Inorganic Precursor TiCl 4 H 2 N 2 plasma 58, 59 Metal organic Precursor tetrakis (diethylamido) titanium H 2 or N 2 plasma 72 Table 2 4. Precursors used for deposition of TiN x thin films by ALD Technique Ti source N source Reference ALD Inorganic Precursor TiCl 4 NH 3 73, 74, 77, 83 86, 172 TiCl 4 Zn + NH 3 75, 79 TiCl 4 dimethylhydrazine 80 TiCl 4 tert butylamine 81 TiCl 4 allylamine 81 TiI 4 NH 3 79, 81, 83 Metal organic Precursor TDMAT NH 3 86, 87, 167, 168 TEMAT NH 3 84, 85, 167 PAALD Inorganic Precursor TiCl 4 NH 3 78 Metal organic Precursor TDMAT remote N 2 / H 2 plasma 88, 89

PAGE 105

105 Table 2 5. Precursors used for deposition of TaN x thin films by CVD Technique Ti source N source Reference CVD Inorganic Precursor TaCl 5 NH 3 90, 169 TaF 5 NH 3 90 TaBr 5 NH 3 90 TaBr 5 H 2 + NH 3 91, 92 Metal organic Precursor tert b utylimi d o tris(diethylamido) t antalum (TBTDET) 95 97 p entakis(dimethylamido) t antalum (PDMAT) NH 3 98, 170 p entakis(diethylamido) t antalum (PDEAT) NH 3 99, 100 tert amylimidotris(dimethylamido) t ant alum (TAIMATA) NH 3 103 e thylimidoethyl(C,N) tris(diethylimido) t antalum H 2 91 PECVD Inorganic Precursor TaCl 5 N 2 90, 93 TaF 5 N 2 90 TaBr 5 N 2 90 TaBr 5 N 2 + H 2 94 TaBr 5 N 2 90

PAGE 106

106 Table 2 6. Precursors used for deposition of TaN x thin films b y ALD Technique Ta source N source Reference ALD Inorganic Precursor TaCl 5 NH 3 104, 105 TaCl 5 Zn + NH 3 105 TaCl 5 1,1 dimethyl hydrazine 80 TaCl 5 tert butylamine (+NH 3 ) or allylamine (+ NH 3 ) 106 TaBr 5 tert butylamine (+NH 3 ) 106 Metal organi c Precursor tert butylimido tris(diethylamido) tantalum (TBTDET) NH 3 108 110 pentakis(dimethylamido) tantalum (PDMAT) NH 3 111 114 tertiaryamylimidotris (dimethylamido) tantalum (TAIMATA) NH 3 109, 116 PAALD Inorganic Precursor TaCl 5 N 2 /H 2 plasma 107 Metal organic Precursor tert butylimido tris(diethylamido) tantalum (TBTDET) H radicals 110, 171 pentakis(dimethylamido) tantalum (PDMAT) H 2 /N 2 plasma 112, 114, 115 tertiaryamylimidotris (dimethylamido) tantalum (TAIMATA) H 2 plasma 116

PAGE 107

107 Table 2 7. Precursors used for deposition of WN x thin films by CVD Technique W source N source Reference CVD Inorganic WCl 6 NH 3 + H 2 117 118 Precursor WF 6 NH 3 + H 2 119 WF 6 NH 3 + SiH 4 + H 2 120 Metal organic Precursor tert butylimino tert butylamido t u ngsten NH 3 or Ar 31, 132 134 t ungsten hexacarbonyl NH 3 135 137 t ungsten pentacarbonyl 1 methylbutylisonitrile NH 3 138 PECVD Inorganic WF 6 NH 3 plasma 121 123 Precursor WF 6 H 2 + NH 3 plasma 124 130 WF 6 H 2 + N 2 plasma 32, 131 Table 2 8. Prec ursors used for deposition of WN x thin film by ALD Technique W source N source Reference ALD Inorganic WF 6 NH 3 139, 140 Precursor WF 6 NH 3 and B 2 H 6 141 PAALD Metal organic Precursor bis( tert butylimido) bis(dimethylamido) tungsten NH 3 45, 144 Inorga nic Precursor WF 6 NH 3 plasma 142, 143

PAGE 108

108 Figure 2 1. Arrhenius plot for CVD showing mass transfer limited and reaction limited growth regimes G is the film growth rate and T is the deposition temperature in K

PAGE 109

109 Figure 2 2. Characteristic ALD growth rate vs. T curve showing different growth regimes

PAGE 110

110 Figure 2 3. Characteristic ALD plot showing the exposure time required to achieve self limiting growth

PAGE 111

111 CHAPTER 3 EXPERIMENTAL TECHNIQ UES FOR THIN FILM DE POSITION AND CHARACTERIZATION 3.1 Descript ion of the CVD Reactor All the depositions of WN x C y were done in a custom CVD reactor (Figure 3 1). Metal organic precursor used for CVD deposition was dissolved in benzonitrile solvent and loaded into a syringe. The solution was delivered into a nebuliz er at a constant flow rate of 4 mL/hr. The nebulizer contains a quartz plate that converts the precursor solvent mixture into aerosol and the aerosol is carried to the reactor via carrier gas (H 2 H 2 + NH 3 or N 2 + NH 3 ). The aerosol/ carrier gas mixture w as delivered to the reactor through heated delivery tube. The delivery tube was heated to prevent condensation and accumulation of condensates on the tube wall. The substrate was heated by RF generator and the RF power was adjusted to change the temperat ure of the substrate. A detailed description of the system has been reported previously. 173 3.2 Modification of the CVD Reactor There were 4 major changes made in the experimental setup: Installation of H 2 and N 2 purifiers in the H 2 and N 2 delivery lines Reconfiguration of the gas lines downstream of the reactor Modification to the line for delivery of NH 3 Replacement of MKS mass flow and pressure programmer/ display 3.2.1 Installation of H 2 and N 2 Purifiers in the H 2 and N 2 Delivery Lines The precurso rs used for CVD deposition are oxygen sensitive. Hence, the CVD reactor should contain a minimal amount of oxygen during film growth. Ultra high purity gases (H 2 and N 2 ) obtained from Praxair contain a few ppm of oxygen and water vapor. The intent behin d installing purifiers was to reduce the impurity levels of carrier gas from ppm level to ppb level. Figure 3 2 shows the change in feed line configuration before and after the installation of gas purifiers. The purifiers were made by Saes Corporation (m odel # PS11 MC1N for N 2

PAGE 112

112 purification and PS11 MC1H for H 2 purification). The purifier uses activated catalyst to adsorb gases such as O 2 H 2 O, CO and CO 2 The installation of the purifier requires that they be installed vertically with the gas inlet from the top. The gas lines were purged prior to installation of purifier since exposure to atmosphere can poison the catalyst in the purifier. The purifier also contained a particle filter that removes particles with size larger than 0.003 m in diameter. The purifier was connected to the lines by VCR fittings. The purifier is designed to operate at a maximum flow of 5000 sccm with nominal flow of 500 sccm. 3.2.2 Reconfiguration of the Gas Lines Downstream of the Reactor One of the major operational proble ms for the CVD system was the failure of the throttle valve. The throttle valve uses a small disc with an o ring inside a cylindrical bore to modulate the conductance through the valve. To control the reactor pressure, a throttle valve controller adjusts the disc position in the valve based on the difference between pressure input from the Baratron pressure gauge and the set point for the reactor pressure. The main problem with the throttle valve was the swelling of the o ring on the disc due to absorpti on of benzonitrile solvent. Since the cold trap, intended to remove condensables such as solvent vapor, was installed downstream of the throttle valve, the disc was exposed to the benzonitrile solvent. The swelling ring due to benzonitril e absorption resulted in rupture of the o ring. Consequently, the throttle valve was not able to control the reactor pressure. The frequency of throttle valve failure was once every four to six experiments. To alleviate the problem of throttle valve fail ure, the downstream of the reactor was reconfigured (Figure 3 3). A cold trap was installed between the reactor and throttle valve so that benzonitrile could be condensed and removed from effluent gas to minimize the exposure of throttle valve o ring to b enzonitrile. As a result of this modification, the frequency of throttle valve failure was reduced to once every 50 experiments.

PAGE 113

113 The glass cold trap was replaced by a new stainless steel cold trap (Figure 3 4). The stainless steel cold trap was designed such that its total internal volume was same as that of the glass cold trap. The glass cold trap had no port for removing the condensate collected at the bottom of cold trap. The stainless steel cold trap had an inclined bottom to collect the benzonitril e and other condensables near the purge port. After every experiment, about 1 2 ml of solvent (primarily benzonitrile) condensed in the cold trap. Before starting new experiment, the solvent collected in the cold trap was removed by opening the purge v alve at the bottom of the cold trap. 3.2.3 Modification to the Line for Delivery of NH 3 to the Reactor When NH 3 was used as a co reactant, the gas was delivered using a secondary NH 3 line 3 NH 3 was used as co reactant were quite non uniform in thickness. The primary reason for this non uniformity was the inhomogeneous distribution of NH 3 because of the proximity of the NH 3 feed port to the substrate surface. To alleviate this problem, a mo dification was made in the system near the H 2 inlet line at the nebulizer. The line was split to allow for the introduction of NH 3 along with the H 2 The new port from which NH 3 3 feed port in Figure 3 5. This modificati on allowed the NH 3 gas to mix properly with the H 2 carrier gas prior to the exposure of the gas mixture on the substrate. Films deposited after this modification had uniform film thickness as measured by SEM. 3.2.4 Replace MKS Mass Flow and Pressure Progr ammer Display The old MKS mass flow and pressure programmer (MKS 147) failed to control the flow rate of NH 3 and had to be replaced. The manufacturing of MKS 147 pressure and flow controller has been discontinued by MKS Inc. So, a new MKS mass flow and p ressure programmer display (MKS 647B) was installed in the system. The MKS 647B mass flow and pressure

PAGE 114

114 programmer/display has a LCD display while the old MKS 147 had a CRT display. Most of the other features of MKS 647 B are similar to that of MKS 147. However, the new controller required rewiring of the transducer input into the controller. 3.3 Operating Procedure for CVD Deposition The operating procedure for CVD deposition has been described previously. 173 The only change made to the procedure was th at after the growth, the substrates were taken out of the CVD chamber only after the substrate temperature reached room temperature. The old procedure allowed for the substrates to be unloaded from the reactor once the substrate temperature was below 75 C. The procedure was modified primarily to minimize the diffusion of oxygen into the deposited film. 3.4 Deposition Parameters Table 3 1 shows the flow rates of precursor, solvent and carrier gas used for CVD growth. For a particular set of experiments, all the deposition parameters except substrate temperature were held constant. Depositions at different growth temperature were done to understand the effect of substrate temperature on film properties such as growth rate, crystallinity, resistivity, atom ic bonding and efficacy as a diffusion barrier. 3.5 Dimensionless Numbers for CVD Reactor Dimensionless numbers can provide valuable insight into the effect of various process parameters such as temperature and pressure on reactor flow pattern. This secti on discusses the calculation results for important dimensionless numbers for the CVD film growth for the reactor and deposition conditions discussed previously. The calculation of dimensionless numbers has been done by using the thermophysical properties of carrier gas alone unless specified otherwise.

PAGE 115

115 3.5.1 Prandtl Number The Prandtl number, Pr, is the ratio of momentum diff usivity to thermal diffusivity. (3 1) The kinematic viscosity is specific heat at constant pressure is c p Chapman Enskog theory can be used to estimate the Newtonian viscosity. 174 (3 2) The molecular weight is is a function of dimensionless temperature, k B H 2 and N 2 B are 38.0 and 91.5 re spectively. Since the depositions were done between 300 and 750 C, the value of dimensionless temperature is greater than 6.0. For dimensionless temperature greater than 6.0, is given by Equation 3 3. 175 (3 3) For N 2 and H 2 176 Substituting these values in Equation 3 2, the value of for H 2 and N 2 carrier gases be calculated. (3 4) (3 5) The ther mal conductivity for a polyatomic gas is given by modified Eucken correlation (Eq. 3 6). 174

PAGE 116

116 (3 6) The Universal gas constant (8314 J/mol K) is R. Since depositions in CVD reactor were done at high temperature and low pre ssure, the ideal gas assumption is valid. For an ideal polyatomic gas, c p is given by Equation 3 7. (3 7) Substituting the values of c p R and M in Equation 3 6, the value of k can be calculated (Eq. 3 8 and 3 9). (3 8) (3 9) From Equations 3 4, 3 5, 3 7, 3 8 and 3 9, the value calculated for Pr is 0.690 for both H 2 and N 2 carrier gases. The value of Pr indicates that the thermal boundary layer a nd the velocity boundary layer for both H 2 and N 2 carrier gases are important for thermal diffusion consideration. 3.5.2 Knudsen Number Knudsen number (Kn) determines if the continuum approach can be used for the fluid flow in CVD reactor. For a dilute ga s mixture in which species B is the predominant gas, Knudsen number is given by Equation 3 10. 177 (3 10) The characteristic length of the reactor is L, molecular weight of dilute species A (precursor 3a,b 4 or 5 ) is M A mol ecular weight of species B (H 2 or N 2 ) is M B the collision diameter of species A is A and the mole fraction of species B is x B For the CVD reactor used for film

PAGE 117

117 growth, the diameter of the reactor (0.0597 m) was used as the characteristic length of the reactor. The present work used 3 different precursors, 3a,b 4 and 5a,b with N 2 and /or H 2 carrier A used for 3a,b 4 and 5a,b were obtained from X ray crystallography. The distance between furthest atoms in the X ray crystallographic structure of all three precursor molecules was ca. 10 and this valu A It should be noted that the X ray crystallographic structure does not account for the substitution of acetonitrile ligand by benzonitrile when the solid precursor is dissolved in benzonitrile solvent. T is calculated by averaging the temperature of the heated impinging jet (50 C) and the temperature of the substrate (between 300 and 750 C). Table 3 2 shows the values of Knudsen number for precursors 3a,b 4 and 5 at different deposition temperature. For Kn < 0.01, continuum approach is valid for analyzing the system. Since for all the different precursors and carrier gases the Kn number is of the order of 10 7 continuum approach is valid. 3.5.3 Reynolds Number The Reynolds number (Re) is the ratio of inertial to viscous force. Reynolds number for a CVD system is given by Equation 3 11. 177 (3 11) The cross sectional area of the reactor is A cs (m 2 ) and the mass flow rate of the carrier gas is (kg/s). R eynolds number is calculated using carrier gas only because the molar flow rate of carrier gas is at least 3 orders of magnitude higher than that of the precursor. A cs for the reactor is 2.8 10 3 m 2 and mass flow rate for H 2 and N 2 are 3.41 10 7 kg/s and 4.77 10 6 kg/s respectively. Table 3 3 shows the value of Re for H 2 and N 2 at different deposition temperature. Since the value of Re is less than 2000, the flow through the reactor is laminar.

PAGE 118

118 3.5.4 Peclet Number The Peclet number (Pe) is a dimens ionless number related to the rate of advection of flow to its thermal diffusion. It is calculated by multiplying Reynolds number with Prandtl number. Forced convective heat transfer can be neglected if Pe 0.5 << 1. Table 3 3 shows the values of Pe and P e 0.5 calculated for H 2 and N 2 carrier gases at different deposition temperature. Since 0.5 for H 2 and N 2 is not << 1, forced convective heat transfer through the reactor can not be neglected. 3.5.5 Grashof Number The Grashof number (Gr) is the rati o of buoyancy force to viscous force on a fluid. This dimensionless number is helpful in determining the occurrence of turbulence for free convection. If the Grashof number is less than 10 5 no turbulence effects are expected for free convection. Grasho f number for the CVD reactor is given by Equation 3 12. 177 (3 12) temperature from the temperature of the substrate. Table 3 4 s ummarizes the Grashof number calculated for H 2 and N 2 at different deposition temperature. For H 2 the value of Gr is less than 10 5 indicating that turbulence is not expected for free convection. However, Gr for N 2 is significantly higher than 10 5 sugges ting that turbulence is expected for free convection when N 2 is used as carrier gas. Experimentally, when N 2 is used as carrier gas, condensates of benzonitrile are observed in the view port. Since the view port is located above the shower head, benzonit rile condensation on view port is possible only if longitudinal convective rolls are present in the reactor. However, when H 2 is used as carrier gas, no such condensation is

PAGE 119

119 observed on the view port. This observation along with the value of Grashof numb er for N 2 confirms that buoyancy effects are important when N 2 is used as a carrier gas. 3.5.6 Rayleigh Number The Rayleigh number (Ra) is the ratio of thermal flux by free convection to that by diffusion. Rayleigh number is defined as the product of Gras hof number and Prandtl number. Table 3 4 summarizes the Rayleigh number calculated for H 2 and N 2 carrier gas for different deposition temperature. Natural convective heat transfer can be neglected if (Ra*Pr) 0.25 << 1. Since the value of (Ra*Pr) 0.25 is n ot << 1 for both H 2 and N 2 carrier gas throughout the deposition temperature, natural convective heat transfer cannot be neglected. Longitudinal convective rolls have been observed in horizontal CVD reactor for Ra > 1708. 177 For both H 2 and N 2 the value of Ra is greater than 1708 indicating that convective rolls are present in the reactor. Furthermore, the convective rolls are expected to be more severe for N 2 carrier gas because the Ra value for N 2 is two orders of magnitude higher than 1708. Experime ntally, the condensation of benzonitrile on the view port confirms that the problem of recirculation is more severe for N 2 as compared to H 2 carrier gas. 3.6 Techniques Used for Thin Film Characterization 3.6.1 X R ay Diffraction The film crystallinity was examined by Philips APD 3720, operating from 5 85 2 rays were generated at 40 kV voltage and 20 mA current. radiation using a Ni filter. When the fi lms were extremely thin (few nm thick), film crystallinity was measured by GIXD technique using radiation, generated at 45 kV and 40 mAmp (1.8 kW), was used for the GIXD analysis. The angle of incidence for measurements w as 2 and the step size was 0.02 per step. The peaks obtained from the XRD and GIXD spectra

PAGE 120

120 were compared with reference spectra in the JCPDS database 178 to determine the crystalline phase(s) present in the film. Table 3 5 lists the reference peak posit ions and their relative intensities for W 2 N, W 2 W 2 C, WO 3 and Si. 178 The broadening of peak in XRD spectra can be used to calculate the grain size of the 13). 179 t = 0.9 / (B cos ) (3 13) T he wave length of X rays used for measurement is the angle of incidence is and the broadening of the diffraction line measured as full width at half maximum (FWHM) is W 2 W 2 C peaks, the most intense (111) diffraction peak was used to determine FWHM. The peak position in an XRD spectrum can also be used to calculate the lattice parameter of the unit cell. The inter (3 14) The value of 1.5405 The lattice parameter, a, for a cubic cell is given by Equation 3 15. (3 15) The Miller Indices are h, k and l. For a hexagonal crystal, the lattice parameter is given by Equation 3 16. (3 16) For hexagonal crystal, two different peak positions are required to solve for lattice parameters a and c.

PAGE 121

121 3.6.2 Auger Electron Spectroscopy Film composition was determined by AES. In AES, energetic electrons are used to knock out core electron of atoms close to the surface of the thin film. The excited atom relaxes to lower energy state by either Auger emission or X ray fluorescence. In Auger emission, the excited atom relaxes to lower energy state by transitions of an electron from higher energy level to the inner shell. The excess energy from the transition causes another electron from higher energy level to eject from the atom. The kinetic energy of the ejected Auger electron can be used to identify the energy levels for that par ticular transition. Determination of the energy levels can then be used for elemental identification. In AES spectra, the number of electrons ejected is plotted versus the kinetic energy and the data are differentiated for better resolution. Elemental i dentification is then done by comparing the peak positions in the AES spectrum to reference spectral lines of different elements. Auger electron spectroscopy measurements were done using a Perkin Elmer PHI 660 Scanning Auger Multiprobe A 5 kV acceleratio n voltage and 50 nA beam current was used for the Auger analysis. The beam diameter was 1 m and the angle of incidence was 45. The AES spectra were obtained between 50 and 2050 eV with a step size of 1 eV. Prior to AES measurement, the sample surface was cleaned by sputter etching for 30 sec using Ar + ions. The etch rate for the sputtering was calibrated at 100 /min using a Ta 2 O 3 standard. The ion gun for sputtering was operated at 15 mA current at Ar pressure of 1.5 10 3 Pa. The beam voltage of the ion gun was 5 keV. A general expression for determination of quantification of atomic concentration in given by Equation 3 17. (3 17)

PAGE 122

122 where, X a = concentration of any constituent a in the sample N a = number of atoms per unit volume of constituent a i = sum of number of atoms per unit volume of all the constituents of the film I a = Auger intensity of a particular transition for constituent a i = sum of Auger intensities of all the constituents of the film S a = relative sensitivity factor for const ituent a S i = relative sensitivity factor of constituent i Auger intensity is calculated from peak to peak height in the numerically differentiated EdN(E)/dE spectrum. The spectrum is numerically differentiated according to the Savitzky Golay differentiat ion algorithm using five differentiation points. The relative sensitivity factor is calculated by comparing the peak intensity of a pure element with that of a reference standard (typically the Cu LMM transition obtained at 10 keV). Table 3 6 lists the r elative sensitivity factors of C, N, O and W used for quantification in AES. 180 The table also lists the position of peaks used for quantification of different elements. It should be noted that the calculation of atomic concentration using Equation 3 17 does not take the matrix effect into account. If an element is present in a matrix of other elements, its properties such as Auger transition probability, secondary ionization by scattered electrons and inelastic mean free path are different from that of pure element. This change in sensitivity factor is often referred to as the matrix effect. The elemental composition obtained by using pure elemental sensitivity factors can have significant error because of the matrix effect. While the absolute value o f film composition might not be accurate, the film composition values can be helpful in comparing films deposited

PAGE 123

123 at different process conditions. Accurate calculation of film composition requires the calibration of the sensitivity factors using a WN x C y s tandard of known composition. The change in elemental composition of the film with depth was measured by AES depth profile. In this technique, the sample surface is sputtered and then analyzed to identify the intensity of peaks of elements present in the film. This process is repeated to obtain the depth profile for different elements. An AES depth profile can be obtained in two different ways: 3 point depth profiling and survey depth profiling. The main difference between these two techniques is the m anner in which peak intensity is calculated. In survey depth profiling, an entire survey of a particular peak is taken to determine the peak intensity. In 3 point depth profiling, a peak is identified by measuring the peak at 3 points. Two of the three points correspond to the baseline of the peak and the third point corresponds to the highest point of the peak. From these three points, the peak intensity is calculated. The advantage of 3 point depth profiling over survey profiling is that since only 3 points of a peak are measured, the measurement is much faster and a better depth resolution can be achieved. The disadvantage of this technique is that precise elemental composition can not be determined. In the present work, 3 point technique was used for depth profiling. 3.6.3 X Ray Photoelectron Spectroscopy XPS analysis was done to obtain bonding information of atoms in thin films. Surface analysis by XPS is done by irradiating the sample surface with monoenergetic soft X rays and measuring the ener gy of the electrons ejected from sample by photoemission process. The kinetic energy of an emitted electron, KE, is given by Equation 3 18. BE (3 18 ) e electron is s Since each element has a

PAGE 124

124 unique binding energy, XPS can be used to determine the elements present in the sample. Depending on the bonding of a particular atom in the sample, t he binding energy of that atom shifts. This shift can be used to obtain the bonding information of an element in the sample. Table 3 7 lists the binding energies of atoms in different compounds of W, C, N and O along with pure elemental binding energies of polymeric and graphitic carbon. X ray photoelectron spectroscopy measurements were taken using a Perkin Elmer PHI 5600 system. X ray photoelectron spectroscopy spectra were taken using monochromatic Mg K radiation with the X ray source operating at 300 W (15 kV voltage and 20 mA current). The sample surface was sputter etched for 15 min using Ar + ions to remove surface contaminants. The etch rate for the XPS system was calibrated at 10 /min using Ta 2 O 3 standard. The ion gun for sputtering was operated at 15 mA current at a Ar pressure of 1.5 10 3 Pa. The beam voltage of the ion gun was 5 keV. 3.6.4 Scanning Electron Microscopy Film thickness measurement was done using SEM. In SEM, an electron beam is attracted towards substrate by a positive bias on the substrate. Secondary electrons are generated from the interaction of electron beam with the substrate surface. Secondary electrons have low energy (2 5 eV) and are generated very close to the su rface. By mapping the coordinates with the intensity of the secondary electron, an image is constructed. For the present study, JEOL JSM 6400 and JEOL JSM 6335F scanning electron microscopes were used for measuring film thickness. To generate electron b eam, JEOL JSM 6400 microscope uses thermionic emission whereas the JEOL JSM 6335F uses field emission technique. In thermionic emission, a tungsten filament was heated to 2400 C to generate electrons. In field emission, electrons were generated by apply ing high bias of 15 kV on a single crystal tungsten tip. For measurements of

PAGE 125

125 less than 40,000x magnification, SEM JEOL 6400 was used. To obtain resolution higher than 40,000x, a field emission SEM JEOL JSM 6335F was used. To measure film thickness, the substrate was cleaved and the cross section of the substrate was viewed under the SEM. 3.6.5 Four Point Probe Film sheet resistance was measured by using a four point probe. A four point probe consists of four sharp metallic probes lined with one another Current is passed through the outer probes and the voltage is measured by the inner probes. Based on the V and I measured, the sheet resistance is calculated using Equation 3 19. (3 19) The sheet resistance of depos ited films was measured using an Alessi Industries four point probe. The resistivity of the 20. (3 20) The thickness of the film, t, is obtained from SEM imaging. 3.7 Diffusion Barrier Testing For diffusion barrier testing, copper film was deposited on the WN x C y thin film using Kurt J. Lesker CMS 18 multi target sputter system. Prior to the deposition of copper on the barrier film, the film was exposed to air for 1 2 hr. The base pressure of the sputter system was 10 6 Torr and deposition was done at 5 mTorr with Ar used as sputter deposition gas. The forward sputtering power for Cu target was 250 W and the film growth rate was 240 /min. The thickness of copper deposited on the barrier film was 100 After the copper deposition, the Cu/WN x C y /Si stack was annea led in the same sputter chamber with the chamber pressure of ca. 10 6 Torr. The annealing temperature was varied between 200 and 600 C.

PAGE 126

126 To check for diffusion of copper through the barrier film, two different procedures were used. Pre and post anneal X RD measurement were done to check for bulk diffusion of copper through the silicon. X ray diffraction was done using the same procedure described earlier. If the copper diffuses through the silicon, copper forms a stable compound Cu 3 Si. A Cu 3 Si peak in the post anneal XRD spectrum would indicate the diffusion of copper into the silicon and the failure of the barrier film. Another method used to detect diffusion of copper in silicon was 3 point AES depth profile. An AES depth profile can indicate the pr esence of copper in the barrier as well as silicon. The advantage of this technique over XRD is that it has higher sensitivity (1 at. %). The disadvantage of 3 point AES depth profile is that during sputtering, the Ar + ion can knock copper atoms into the barrier film and also into silicon if the barrier film is very thin. This lead to artifacts in the depth profile. To avoid this problem especially for ultra t hin diffusion barrier films, the copper film was etched off after annealing. The copper etching was done by dipping the Cu/WN x C y /Si stack in 50 % HNO 3 solution for 3 seconds. The 3 point depth profile was done following the etching of the copper layer.

PAGE 127

127 Table 3 1. Process parameters for CVD of thin films from precursors 3a,b 4 and 5 Process Parameter Precursor and Carrier Gas 3a,b +NH 3 +H 2 (Ch 4) 4 + H 2 (Ch 5) 4 +NH 3 +H 2 (Ch 6) 4 +NH 3 +N 2 (Ch. 6) 5 + H 2 (Ch 7) Temperature (C) 450 750 400 750 300 750 300 750 300 750 Pressure (Torr) 350 350 350 350 350 H 2 Flow Rate (sccm) 1000 1000 1000 0 1000 N 2 Flow Rate (sccm) 0 0 0 1000 0 NH 3 Flow Rate (sccm) 25 0 30 30 0 Precursor conc. (mg/mL benzonitrile) 11.2 9.0 9.0 9.0 7.4 Table 3 2. Knud sen number for precursors 3a,b 4 and 5 with H 2 or N 2 carrier gas for different deposition temperature Deposition Temperature (C) Knudsen Number Precursor 3a,b Precursor 4 Precursor 5 H 2 N 2 H 2 N 2 H 2 N 2 300 1.8310 7 3.7710 7 1.0510 7 3.4210 7 1.16 10 7 3.7610 7 350 1.9410 7 3.9810 7 1.1110 7 3.6110 7 1.2310 7 3.9710 7 400 2.0410 7 4.1910 7 1.1710 7 3.810 7 1.2910 7 4.1810 7 450 2.1410 7 4.410 7 1.2310 7 410 7 1.3510 7 4.3910 7 500 2.2410 7 4.6110 7 1.2910 7 4.1910 7 1.42 10 7 4.610 7 550 2.3410 7 4.8210 7 1.3510 7 4.3810 7 1.4810 7 4.810 7 600 2.4510 7 5.0310 7 1.410 7 4.5710 7 1.5510 7 5.0110 7 650 2.5510 7 5.2410 7 1.4610 7 4.7610 7 1.6110 7 5.2210 7 700 2.6510 7 5.4510 7 1.5210 7 4.9510 7 1.68 10 7 5.4310 7 750 2.7510 7 5.6610 7 1.5810 7 5.1410 7 1.7410 7 5.6410 7

PAGE 128

128 Table 3 3. Reynolds number for H 2 and N 2 carrier gas for different deposition temperature Deposition Temperature (C) Reynolds Number Peclet Number (Re*Pr) 0.5 = (Pe) 0.5 H 2 N 2 H 2 N 2 H 2 N 2 300 0.62 4.23 0.43 2.92 0.66 1.71 350 0.60 4.08 0.41 2.82 0.64 1.68 400 0.58 3.95 0.40 2.72 0.63 1.65 450 0.56 3.82 0.39 2.64 0.62 1.62 500 0.55 3.71 0.38 2.56 0.61 1.60 550 0.53 3.60 0.37 2.49 0.61 1.58 600 0.52 3.51 0.36 2.42 0.6 0 1.56 650 0.50 3.41 0.35 2.36 0.59 1.53 700 0.49 3.33 0.34 2.30 0.58 1.52 750 0.48 3.25 0.33 2.24 0.57 1.50 Table 3 4. Grashof number for H 2 and N 2 carrier gas for different deposition temperature Deposition Temperature (C) Grashof Number Rayle igh Number (Ra*Pr) 0.25 H 2 N 2 H 2 N 2 H 2 N 2 300 5388 247106 3718 170503 7 19 350 5120 234829 3533 162032 7 18 400 4788 219585 3304 151514 7 18 450 4433 203332 3059 140299 7 18 500 4081 187170 2816 129147 7 17 550 3744 171695 2583 118470 6 17 600 3428 157206 2365 108472 6 17 650 3136 143827 2164 99240 6 16 700 2869 131580 1980 90790 6 16 750 2626 120432 1812 83098 6 15

PAGE 129

129 Table 3 5. Reference XRD peaks and their relative intensity for Si, W 2 W 2 W 2 C and WO 3 from JCPDS 178 Si W 2 N W 2 C W 2 C WO 3 Intensity Intensity Intensity Intensity Intensity 28.44 100 37.73 100 36.98 40 34.52 25 23.64 100 47.3 0 55 43.85 47 42.89 100 38.03 22 33.64 69 56.11 30 63.73 33 62.03 30 39.57 100 41.46 20 69.13 6 76.52 44 74.2 0 50 52.3 0 17 48.43 16 76.38 11 80.59 13 78.23 10 61.86 14 54.57 36 69.79 14 60.25 22 72.84 2 70.75 6 74.98 12 75.81 11 75.98 10 78.06 2 81.33 2 80.87 3 Table 3 6. Elemental relative sensitivity factors for AES quantification along with position of peaks in dN(E)/d(E) spectra used for quantification 180 Element Relative sensitivity factor Position o f peak used for quantification (eV) C 0.14 272 N 0.23 38 1 O 0.4 510 W 0.08 1736

PAGE 130

130 Table 3 7 Binding energy of C, N, O and W in different compounds Compound Binding Energy (eV) Reference C1s N1s O1s W4f 7/2 W 2 N -397 -32.8 33 122 181 182 W 2 C 283.4 283.6 --31.4 31.7 183 185 WC 282.7 283.5 --31.8 32.4 183 186 WO 2 --530.2 32.8 33.0 183 186 W 2 O 5 ---34.4 34.7 186 187 WO 3 --530.3 531.0 35.5 36.0 186 190 Pure W ---31.2 31.8 186, 188 189, 191, 194 Amorphous C 284.5 285.2 ---183 Graphitic C 284.4 284.5 ---192 193 Diamond like C 285.3 ---192 193 Figure 3 1. Schematic of CVD reactor used for thin film deposition

PAGE 131

131 Figure 3 2. Line diagram of the carrier gas feed line A ) B efore installation of gas purifier. B) after installation of gas purifie r

PAGE 132

132 Figure 3 3. Flow diagram of reactor downstream. A) Before reconfiguration. B) After reconfiguration

PAGE 133

133 Figure 3 4. Design of new cold trap made of stainless steel

PAGE 134

134 Figure 3 5. Schematic of the old and new NH 3 feed port

PAGE 135

135 CHAPTER 4 WN x C y T HIN FILM DEPOSITION FROM C l 4 (RCN)W(NC 3 H 5 ), NH 3 AND H 2 Our group has previously reported growth of WN x C y thin films from 3a,b 157 Films grown from 3 a,b below 550 C were amorphous and had a low resistivity at lower deposition cm at 450 C). However, films had low N content even though the low N C imido bond dissociation energy for 3a,b is expected to result in higher N content in the film. F ilm deposited with 3a,b at 450 C had only 4 at. % N and th e highest N concentration of 11 at. % was for film deposited at 550 C. A nitrogen deficient diffusion barrier film is not desirable because such films can easily form grain boundaries that can act as pathways for copper diffusion. In addition to having low N content, films deposited with 3 a,b had high O content (16 at. % at 450 C and 11 at. % at 525 C). W hen ammonia was used as a co reactant for deposition from precursor 2 a,b the WN x C y films had higher N content and lower O content 155 In this chapter, deposition of WN x C y thin films fr om 3a,b and ammonia is discussed with primary focus on diffusion barrier application The intent of using ammonia as co reactant is to increase the N content and decrease the O content so that films deposited with 3a,b are more suitable for diffusion barr ier application. The data for films deposited with 3 a,b and H 2 (without ammonia) previously reported 157 have been included to evaluate the effect of ammonia addition on film properties. The films deposited with 3 a,b and ammonia are also compared with tho se deposited with 2a,b and ammonia to evaluate the effect of C 3 H 5 and i Pr substituents in tungsten imido precursors on film properties. 4.1 Precursor Synthesis Standard Schlenk and glovebox techniques were employed in the synthesis of 3a Tungsten oxychlo ride was prepared by a slightly modified literature method. 195 Tungsten oxychloride (1.229 g, 3.60 mmol) was added to a solution of allyl isocyanate (0.366 g, 4.41

PAGE 136

136 mmol) in heptane (60 mL) in a sealed pressure vessel and the resulting mixture was heated a t 110 C for 36 h. After the solvent was removed in vacuo the reddish brown residue was dissolved in ca. 10 mL of CH 3 CN. The resulting solution was stirred for two hours and the solvent was removed under reduced pressure. Brown residue from the mixture was washed with 5 x 10 mL of toluene and the extracts concentrated to approximately 5 mL. The product was precipitated by adding hexane. The orange brown solid was filtered and washed with hexane to afford 0.974 g (64 % yield) of the imido complex. The benzonitrile complex Cl 4 (PhCN)W(NC 3 H 5 ) ( 3b ) was produced in situ by substitution of the acetonitrile ligand of 3a with benzonitrile, which was utilized as the solvent for the deposition experiments. Figure 4 1 shows the X ray crystallographic structure o f 3a 157 Table 4 1 lists important bond distances and bond angles for 3a 4.2 Film Composition Figure 4 2 shows the AES results for films deposited in the presence and absence of NH 3 at temp erature ranging from 450 to 750 C. The Auger spectra for films deposited with ammonia indicate the presence of W, N, C, and O in the films. Note that the AES measurements for films deposited with ammonia were done after sputtering for 30 sec whereas those for films deposited without ammonia were measured after 2 min of sputtering. Sputter etching can change the near surface composition of compounds due to preferential sputtering. 196 In fact, tungsten nitride has been reported to undergo preferential erosion of nitrogen during sputter etching. 189 Despite preferenti al sputtering, the film composition obtained from Auger measurements can be very useful in comparative analysis of films deposited at different temperature. For films deposited with ammonia at 450 C, the W concentration was 51 at. %, which increases to 61 at. % for growth at 475 C. This is accompanied by a decrease in N

PAGE 137

137 concentration from 23 at. % to 17 at. %. Above 450 C, the W concentration gradually decreases, accompanied by an increase in C content in the films. The film deposited with ammonia at 450 C shows the highest N concentration of 23 at. %. As the deposition temperature is increased, the N concentration in the film decreases. The lowest N concentration of 8 at. %. is observed for films grown at 750 C. The carbon concentration remains relatively steady at ca. 20 at. % for growth between 450 and 500 C. Between 500 and 600 C, the C concentration increases twofold and continues to rise with growth temperature. Increased incorpora tion of C from decomposition of the precursor ligands and the benzonitrile solvent is consistent with such an increase in C concentration. Oxygen incorporation in the film remains below 6 at. % throughout the deposition temperature range. While the O con centration s at the lower end (450 500 C) and higher end (650 750 C) of the temperature range are slightly higher than that at the center (525 600 C), the O concentration remains fairly constant over the range of deposition temperature. Oxygen is incorporated into the films either from residual gas in the reactor during growth or from exposure of film to atmosphere after growth. 156 In comparison to films grown in the absence of ammonia, deposition with ammonia as co reactant resulted in a decrease in W concentration, increase in N conce ntration, and a marked decrease in O concentration (Figure 4 2 ). The increase in N concentration for depositions with ammonia is quite dramatic in the low temperature range, as exemplified by the effect for growth at 450 C, where the addition of ammonia increases the N concentration from 4 at. % to 23 at. %. The increase in N concentration suggests that the ammonia molecule reacts with 3a,b either in the gas phase (homogeneous reaction) or on the substrate (heterogeneous reaction), resulting in higher N incorporation in the film. Below 650 C, C incorporation for depositions with

PAGE 138

138 ammonia was higher than that for depositions without ammonia. Even at low temperature, when C deposition from the solvent is low 156 the C con tent of films deposited with ammonia is higher than for films deposited without ammonia. Due to the presence of chlorine in the precursor complex 3 a the possible presence of chlorine in the films was of interest. This analysis could not be carried out b y Auger spectroscopy because the W NNN peak overlaps with the Cl LMM peak at 180 eV. Thus, XPS data were collected for films deposited at 450, 600 and 750 C (Figure 4 3 ) to ascertain if the films contained chlorine. No chlorine peaks were observed for either the Cl 2s or Cl 2p 3/2 at 270 eV and 199 eV, respectively, confirming that the chlorine content in the films was lower than the detection limit of XPS (ca. 1 at. % ) Fe 2p peaks at 723 eV and 711 eV in the XPS spectra are artifacts that arise from the sample holder (which contains Fe) due to small sample sizes. 4.3 X Ray Diffraction Measurement 4.3.1 Film Crystallinity Figure 4 4 shows the XRD spectra for films deposited using 3a,b and ammonia. The three prominent peaks in the spectra at 33.05, 61.70 and W 2 W 2 C. The depositi on at all other temperature except 650 C showed no peaks corresponding to W 2 N or W 2 C, indicating that the films were X ray amorphous. If films are very thin, the XRD measurements cannot detect a crystalline phase because of the low intensity of the di ffracted X ray beam. For very thin polycrystalline films, the XRD measurements done in the Bragg because of low intensity signal from the film and the XRD spectrum is dominated by the high intens ity substrate peak. In GIXD, the scan is done at grazing incidence; hence most of the

PAGE 139

139 signal from the diffraction is generated from the top 1 2 nm of the thin film. The substrate peak, if present, do not dominate the GIXD spectrum. The depth of penetr ation of X rays can be increased by increasing the angle of incidence in the GIXD measurement. Grazing incidence XRD measurements were done for films deposited with 3a,b and ammonia to determine if the films were amorphous as indicated from the XRD measur ements by Bragg Bretano geometry. Figure 4 5 shows the grazing incidence XRD spectra for films deposited between 450 and 700 C. Figure 4 6 shows the thickness of films for which grazing incidence XRD spectra were taken. The XRD spectrum of the film dep osited at 450 C indicates that it is X ray amorphous. At 500 C, there are three peaks that emerge at 37.47, 43.49 and 63.71 2 degrees. All three peaks lie between the standard diffraction peaks for W 2 N and W 2 C, suggesting the existence of separate W 2 N and W 2 C phases or the presence of a single WN x C y phase. As the deposition temperature is increased from 500 to 550 C, the same three peaks appear, though the intensity of these peaks is lower than that for films deposited at 500 C. In the same temperature range, the tungsten concentration changes from 59 at. % to 49 at. %. A lower tungsten concentration would lead to decrease in the W 2 N phase, W 2 C phase or both, resulting in a decrease in the film crystallinity. The XRD spectrum for film deposited at 600 C shows the highest level of crystallinity. As the deposition temperature is increased from 550 to 600 C, the nitrogen concentration decreases while the carbon concentration increases. An increase in crystallinity would suggest an increase in the W 2 C phase. The films deposited at 650, 700 and 750 C show a decrease in crystallinity, evidenced by decrease in peak intensity. At higher temperature, an increase in amorphous C deposition would hinder the formation of crystallites, which is consistent wit h the decrease in film crystallinity at higher deposition temperature.

PAGE 140

140 Peaks at 650, 700 and 750 C are much broader possibly due to increased f ilm strain at high temperature. We have previously reported XRD spectra for films deposited with 3a,b without a mmonia 157 The films deposited without ammonia are amorphous below 550 C. At and above 550 C, the film crystallinity increases with deposition temperature. Films deposited with ammonia do not show this consistent increase in crystallinity with deposit ion temperature. In fact, for higher temperature depositions with ammonia, the film crystallinity actually decreases. This observation reinforces the assertion that the ammonia co reactant alters the mechanism of film depos ition. 4.3.2 Lattice Parameter Figure 4 7 shows the change in lattice parameter with deposition temperature for growth with and without ammonia. The WN x C y (111) peak position from XRD spectra was used to calculate the lattice parameter. A ll the films deposited with ammonia between 5 00 and 750 C have a lattice parameter between the standard lattice parameters for 2 N (111) and 2 C (111). The changes in lattice parameter with deposition temperature could result from either uniform strain or compositional variation. Since the WN x C y films are not highly ordered, compositional variation rather than uniform strain is believed to be the primary factor affecting lattice parameter. The lattice parameter decreases with an increase in deposition temperature for depositions with ammonia b etween 500 and 700 C, whereas it increases with an increase in deposition temperature for depositions without ammonia. A decrease in lattice parameter could result from a decrease in N and/or C in interstitial sublattice and an increase in the vacancy concentration For f ilms deposited with ammonia, the N concentration decreases with deposition temperature whereas the C concentration increases. One possible explanation for the decrease in

PAGE 141

141 lattice parameter with increase in deposition temperature is that the variation in N concentration in the films is primarily responsible for the lattice parameter variation and the additional carbon deposited at higher temperature incorporates outside the WN x C y polycrystals. 4.3.3 Polycrystal Grain Size. The grain size (t) was calculated using the Scherrer equation ( vide supra ). The most intense (111) diffraction peak was used to determine FWHM. Figure 4 8 shows the change in grain size with temperature for films depo sited with and without ammonia. The average grain size calculation sh ows that films deposited with 3a,b and ammonia are nanocrystalline between 500 and 750 C deposition temperature, with the average grain size ranging from 37 to 60 A previous study has reported the growth of nanocrystalline WN x C y between deposition temperature of 225 and 400 C. 197 In the present work, we found that the films deposited with 3a,b and ammonia were amorphous at 450 C deposition temperature and nanocrystalline growth is observed at deposition temperature of 500 C and higher. At a deposition temperature of 500 C, the average grain size of crystallites for film deposited w ith ammonia is 56 The average grain size remains almost unchanged at 550 C. Between 550 C and 700 C, the average grain size gradually decreases from 58 to 39 As the deposition temperature increases, the excess C deposited from the solvent/pre cursor fragmentation could hinder the growth of polycrystals, resulting in a decrease in grain size. The comparison of grain size for experiments with and without ammonia shows that for deposition at temperature 600 C and below, the average grain size is higher for films deposited with ammonia than for those deposited without ammonia. At 600 C, the grain size for film grown with ammonia is similar to that for film grown without ammonia within the margin of

PAGE 142

142 error. At 650 C, the grain size for film depo sited with ammonia is slightly lower than the grain size for films deposited without ammonia. 4.4 Film Growth Rate deposition time. Figure 4 9 own at 450 and 650 C. The growth rate for 3a,b with ammonia ranged from ca. 4 /min at 450 C to ca. 17 /min at 750 C. It should be noted that for growth using 3a,b without ammonia, the concentration of 3a,b in benzonitrile was 7.5 mg/mL (1.7 10 2 m oles/L) whereas for growth using 3a,b with ammonia, the 3a,b concentration in benzonitrile was 11.2 mg/mL (2.66 10 2 moles/L) Because the introduction of ammonia as co reactant decreases the film growth rate, a higher concentration of 3a,b was necessar y to obtain films that were sufficiently thick for materials characterization. This trend is different from that previously observed for growth from 2a,b where the use of ammonia as co reactant resulted in an increase in growth rate at lower temperature and a decrease at higher temperature 155 Figure 4 10 shows Arrhenius plots of growth rate for experiments with and without ammonia. The apparent activation energy (E a ) for film growth from 3a,b and ammonia was 0.34 eV, which is significantly higher than the 0.15 eV activation energy reported for depositions with 3a,b precursor without ammonia. 4.5 Film Resistivity Figure 4 11 shows film resistivity at different deposition temperature for films grown with and without ammonia. For depositions with ammonia the resistivity is 5820 C deposition temperature and increases to 7190 C. Interestingly, the film resistivity has the lowest value of 1690 C, even though the film deposited at that temperature has l ower W content and higher N content. Resistivity increases to

PAGE 143

143 12650 C and continues to increase with an increase in deposition temperature until it reaches its highest value of 24178 C. The decrease in W concentration and conc omitant increase in C concentration between 550 and 700 C is believed to be the reason for the increase in film resistivity. If the C in films exists as WC x it would decrease the film resistivity. But an increase in resistivity with increase in C conte nt of film could possibly result from the incorporation of additional C at the grain boundary (outside the WN x C y polycrystal) at higher deposition temperature. The film resistivity was considerably higher for films deposited with ammonia compared to films deposited without ammonia. This behavior is expected since ammonia tends to increase N content in the films and higher N content leads to higher film resistivity. 4.6 Comparison of F ilms D eposited from 3a,b and 2a,b The comparison of films deposited fro m 3a,b and ammonia with those deposited from 2a,b and ammonia 155 can provide insight into the effect of different ligands (C 3 H 5 and i Pr) on film properties. Films deposited at 450 C from 3a,b and ammonia ha ve similar W and N content, higher C content and lower O content as compared to films deposited with 2a,b and ammonia at same temperature (Figure 4 2 ). At 500 and 600 C deposition temperature, films from 2a,b exhibit higher N content and lower C content as compared to films deposited with 3a,b and am monia at the same temperature. The comparison of film crystallinity for films deposited with 3a,b and ammonia with those deposited with 2a,b and ammonia shows that while 3a,b results in crystalline films at and above 500 C, films deposited with 2a,b do n ot show crystallinity until 600 C and higher. This trend in crystallinity can be explained by considering that the film growth rate for films deposited with 2a,b and ammonia is considerably higher ( 17 23 /min) than that for films deposited with 3a,b a nd ammonia (4 17 /min ), especially at lower

PAGE 144

144 deposition temperature. Higher growth rate could shorten the time for crystallites to grow via surface diffusion, resulting in amorphous film growth. The comparis on of film resistivity (Figure 4 11 ) shows tha t at 450 and 500 C deposition temperature, films grown with 3a,b and ammonia have a significantly lower resistivity than those grown with 2a,b and ammonia. Interestingly, films deposited with both precursors have the lowest resistivity for deposition at 550 C. Above 550 C, there is a dramatic increase in resistivity of films deposited with 3a,b and ammonia while the resistivity of films grown with 2a,b and ammonia levels off. Evaluation of all films grown from 3a,b or 2a,b and ammonia shows that film d eposited with 3a,b and ammonia at 450 C is the best candidate for diffusion barrier application because of its high N content, low O content, low deposition temperature, amorphous microstructure and lowest resistivity among the amorphous films grown from 3a,b or 2a,b with ammonia. Diffusion barrier testing is required to ascertain if this film is suitable for diffusion barrier application. 4.7 Conclusions Successful thin film deposition of WN x C y by MOCVD utilizing 3a,b H 2 and ammonia has been achieved. When used as a co reactant with the allylimido complexes 3a,b ammonia significantly alters the composition, crystallinity, and resistivity of the resulting films as compared to films grown without ammonia. Films deposited with ammonia had significantly h igher carbon and nitrogen content and substantially lower oxygen content. No chlorine was detected in the XPS spectra of films grown over the entire deposition temperature range. While films grown without ammonia showed an increase in film crystallinity with increasing deposition temperature, those grown with ammonia exhibited more complex behavior, with crystallinity peaking for growth at 600 C. The activation energy for the film growth from 3a,b and ammonia

PAGE 145

145 is estimated to be 0.34 eV, which is significantly higher than the activation energy of 0.15 eV reported for film growth from 3a,b without ammonia. Films deposited with ammonia have high er resistivity as compared to films deposited without ammonia because of a higher N and C concentration coupled with a lower W concentration. The lowest resistivity for films deposited with ammonia was the value of 1690 cm obtained from growth at 550 C. Films deposited with 3a,b and ammonia have higher C content and lower O content than those grown from 2a,b and ammonia. The N content of film grown with 3a,b and ammonia is similar to that for film grown with 2a,b and ammonia at 450 and 700 C. Films grown with 2a,b and ammonia below 600 C were amorphous, whereas deposition from 3a,b and ammonia resulted in amorphous film below 500 C. At temperature below 550 C, films deposited with 3a,b and ammonia had a significantly lower resistivity as compare d to films deposited with 2a,b and ammonia. But above 550 C, films deposited with 3a,b and ammonia had a much higher resistivity as compared to those deposited with 2a,b and ammonia. From a diffusion barrier application standpoint, it has been demonstrat ed that ammonia can be used with 3a,b to increase the N content of WN x C y films. A corresponding increase in film resistivity is also seen and some optimization may be needed to strike a balance between higher N content and lower film resistivity.

PAGE 146

146 Table 4 1. Bond distances () and angles (degrees) for Cl 4 (CH 3 CN)W(NC 3 H 5 ) ( 3a ) Bond Distance/angle Bond Distance/angle W N(1) 1.687(9) W Cl(4) 2.351(9) W N(2) 2.308(8) N(1) C(1) 1.508(17) W Cl(1) 2.339(10) C(1) C(2)' 1.51(2) W Cl(2) 2.317(8) C(2)' C(3)' 1.3 6(3) W Cl(3) 2.324(9) N(2) C(4) 1.130(12) C(1) N(1) W 167(2) N(1) W N(2) 175.8(15) N(1) W Cl(2) 90.6(8) N(2) W Cl(2) 86.0(7) N(1) W Cl(3) 93.6(7) N(2) W Cl(3) 84.0(6) Cl(2) W Cl(3) 88.7(4) C(4) N(2) W 175(3) N(1) W Cl(1) 102.4(8) N(2) W Cl(1) 81.1(7) C(2)' C(1) N(1) 114.2(21) C(3)' C(2)' C(1) 120(2) Cl(3) W Cl(1) 90.18(13) N(2) C(4) C(5) 178(2) Figure 4 1. Thermal ellipsoids diagram of the molecular structure of Cl 4 (CH 3 CN)W(NC 3 H 5 ) ( 3a ). Thermal ellipsoids are plotted at 50% probability 157

PAGE 147

147 Figu re 4 2. Composition of films deposited from 3a,b (with and without ammonia 157 ) and 2a,b with ammonia 155 at different deposition temperature as determined by AES on Si (100) substrate

PAGE 148

148 Figure 4 3. X ray photoelectron spectroscopy patterns of fil ms deposited at 450, 600 and 750 C from 3a,b and ammonia on Si (100) substrate.

PAGE 149

149 Figure 4 4 X ray diffraction spectra of films deposited on Si (100) substrate from 3a,b and ammonia

PAGE 150

150 Figure 4 5 Grazing incidence XRD spectra of films deposited on Si (100) substrate from 3a,b and ammonia. The solid and dashed vertical lines indicate the location of reference peaks for standard powder diffraction of W 2 N and W 2 C respectively

PAGE 151

151 Figure 4 6 Thickness of film (obtained from X SEM) at different de position temperature for films deposited with 3a,b and ammonia on Si (100) substrate

PAGE 152

152 Figure 4 7. Lattice parameters for films grown from 3a,b on Si (100) substrate with and without ammonia The solid line at 4.126 corresponds to the standard lattice parameter for 2 N The dashed line at 4.236 corresponds to standard lattice parameter for 2 C (111). Error bars indicate uncertainty in determination of peak position for lattice parameter calculation.

PAGE 153

153 Figure 4 8 Average grain size at diffe rent deposition temperature for films grown from 3a,b with and without ammonia 157 on Si (100) substrate. Figure 4 9. Scanning electron microscope images for films grown from 3a,b with ammonia A ) Film deposited at 450 C B ) Film deposited at 650 C

PAGE 154

154 Figure 4 10. Arrhenius plot of log of film growth rate vs. inverse temperature for deposition from 3a,b ( with and without ammonia 157 ) and 2a,b with ammonia 155

PAGE 155

155 Figure 4 11 Change in film resistivity with deposition temperature for films grown on Si (100) from 3a,b (with and without ammonia 157 ) and 2a,b with ammonia 155

PAGE 156

156 CHAPTER 5 DEPOSITION OF WN x C y USING W(N i P r )C l 3 [ i P r (NM e 2 )N i P r ] AND H 2 Transition metal guanidinate and amidinate complexes have been used to deposit a variety of materia ls including TiC x N y Fe, Co, Ni, Cu, and TaN 198 201 by CVD and/or ALD. The success with these complexes suggested preparation of guanidinato derivatives of 2a,b as precursor candidates. This chapter discusses the deposition of WN x C y thin films using W(N i P r)Cl 3 [ i PrNC(NMe 2 )N i Pr] ( 4 ) a guanidinato derivative of 2a,b and evaluate s its potential for depositi on of diffusion barrier films. The film p roperties obtained from 4 are compared to th ose from 2a,b to assess the effect of the guanidinato ligand The W N x C y films were also evaluated as diffusion barriers by coating them with PVD Cu and annealing the Cu/WN x C y /Si stack in vacuum at different temperature. 5.1 Precursor Synthesis 202 A 250 mL Schlenk tube was charged with LiNMe 2 (0.223 g, 4.38 mmol) and 100 m L of Et 2 O. The resulting colorless suspension was cooled to 0 C. 1,3 diisopropylcarbodiimide (0.69 mL, 4.4 mmol) was added drop wise to the suspension at 0 C. After warming the reaction mixture to room temperature over 2 h, the resulting cloudy soluti on of lithium guanidinate reagent was cannula transferred into a 250 mL Schlenk tube containing a solution of W(NCH(CH 3 ) 2 )Cl 4 (OEt 2 ) (2.00 g, 4.38 mmol) in Et 2 O (100 mL) at 78 C. The reaction mixture was stirred for 20 min at 78 C and then warmed to ro om temperature overnight in the absence of light. The solvent was removed from the reaction mixture in vacuo and extraction into Et 2 O (200 mL) followed by filtration yielded a dark amber solution. Et 2 O was removed in vacuo to give crude 4 as a dark amber powder. Pure 4 crystals (1.27 g, 56%) were obtained by recrystallization from a toluene solution layered with hexane at 20 C. Figure 5 1 shows the X

PAGE 157

157 ray crystallographic structure of precursor 4 and Table 5 1 lists important bond distances and bond ang les for 4 obtained from the X ray structure. 5.2 Film G rowth The concentration of 4 in benzonitrile was 9.0 mg/mL and the flow rate of precursor solution was 4 mL/hr. The H 2 flow rate was held constant 1000 sccm and reactor pressure was 350 Torr. Detail ed description of reaction conditions is provided in Chapter 3. The lowest temperature at which appreciable film growth was observed from precursor 4 was 400 C. Films were generally smooth and had a shiny metallic surface with film color varying from go lden ( deposition below 500 C) to shiny black ( deposition at and above 400 C) 5.3 Film Composition Figure 5 2 shows AES results for films deposited with precursors 4 and 2a,b using H 2 as co reactant and carrier gas The AES spectra indicated the presenc e of W, N, C and O in films deposited with precursor 4 For films deposited with 4 at 400 C, the W concentration is 5 3 at. %. As the deposition temperature is increased to 450 C, the W content reaches its highest value of 6 6 at. % then gradually decre ases to 37 at. % for deposition at 7 00 C, reflecting an increase in C content as the deposition temperature rises The N content of films deposited with precursor 4 has its highest value of 12 at. % for deposition at 400 C. The N content of the film de creases as deposition temperature is increased from 400 to 650 C then remains almost unchanged between 650 and 750 C. The variation in the final film N content with growth temperature is likely influenced by several factors, including N volatilization, competition between C and O for bonding with W sites, and precursor decomposition pathways and rates. Both C and O were detected in significant amounts in all films. The C content of film deposited with precursor 4 decreases from 20 at. % at 400 C depos ition temperature to 1 4 at. % at 450 C. At growth temperatures b etween 450 and 700 C, the C content of film continuously

PAGE 158

158 increases and reaches its highest value of 5 7 at. % at 700 C. Figure 5 3 shows a monotonic increase in C/W ratio with deposition t emperature between 450 and 700 C. In a previous study using precursor 2a,b it has been demonstrated that the extent of C incorporation depended upon the solvent (1,2 dichlorobenzene or benzonitrile). 155 It is possible, of course, that the precursor is also involved in the C incorporation mechanism directly through its decomposition pathway or indirectly by reaction with the solvent or one its decomposition products. The O content of film deposited with 4 at 400 C is 15 at. % and it gradually decreases to 3.0 at. % at 600 C. Between 600 and 750 C deposition temperature, the O content of film is ca. 3.0 at. %. Oxygen is believed to be incorporated into the films either from residual gas (oxygen or water vapor) in the reactor during growth or from atm ospheric oxygen exposure post growth. 155 It is desirable for barrier application to deposit the film at the lowest temperature that gives suitable properties, which includes amorphous microstructure, a high value of N/W atomic ratio and low oxygen content. The comparison of films deposited from 4 and 2a,b reveals that the observed lowest deposition temperature for 4 is one experimental increment ( 50 C ) lower that the lowest growth temperature of 450 C for 2a,b No film deposition was observed for 4 at 3 50 C as measured by AES. The comparison of film composition at 450 C shows that film deposited from 4 has higher W and N content, similar C content and significantly lower O content as compared to film deposited with 2a,b using the same solvent. For de position a t and above 500 C, the N and O content of films deposited with 4 and 2a,b are quite similar. Since the number of N atoms per W atom in 4 is double that in 2a,b it might be expected that films deposited with 4 would have a higher N /W ratio as c ompared to those deposited with 2a,b The highest N/W ratio for films deposited with 4 is 0.24 at 400 C, which is considerably higher than the highest N/W ratio 0f 0.16 observed for films deposited with 2a,b at the higher temperature

PAGE 159

159 450 C This differe nce in N/W ratio is significant and larger than other values observed for 1 a,b and 3 a,b Since the difference in O content between films deposited from 4 and 2a,b is significant only at temperature below 500 C, it is believed that films deposited below 5 00 C by 4 are more dense that those deposited from 2a,b Higher density in the film would hinder O diffusion and decrease subsequent O incorporation. F ilms deposited with 4 have a significantly lower W content and a higher C content as compared to films deposited with 2a,b for deposition temperature above 500 C. At higher deposition temperature, decomposition of C containing ligands and the benzonitrile solvent results i n C incorporation in the film. As discussed earlier, Cl impurity is highly undesira ble in the diffusion barrier film. Since the precursor 4 has three Cl atoms per molecule, there is a possibility of Cl incorporation into the film. X ray photoelectron spectroscopy measurements were done to detect Cl in the thin film because AES was not capable of detecting Cl impurity in W containing film ( vide supra ). The detection limit of XPS technique is ca. 1 at. %. Figure 5 4 shows the XPS profiles of films deposited at 400, 500, 600 and 700 C. The sample surface was sputter cleaned with Ar + io ns for 5 min prior to the measurement. The spectra do not show any Cl peak indicating that the Cl content of the film was below the detection limit of the XPS. 5.4 X Ray D iffraction of Films Figure 5 5 shows the XRD patterns for films deposited with 4 be tween 400 and 750 C. X ray diffraction spectra for films deposited at 400 and 450 C show no peaks attributable to the film, but only pea ks associated with the substrate ( reflections at 33.10, 61.75 and 69.20 2 r espectively ). The absence of other peaks in these two spectra suggests that films deposited at 400 and 450 C are X ray amorphous. The XRD pattern for the film deposited at 500 C show s the emergence of crystallinity as evidenced by the two broad peaks at 37.74 and 44.40 The W 2 N

PAGE 160

160 W 2 C indicating the presence of either the solid solution WN x C y or a physic al mixture of W 2 W 2 C. 178 Films deposited at 550, 600 and 650 C show evidence of increased crystallin ity with four peaks observed at approximately 37.50, 43.70, 63.55 and 75.05 corresponding to the WN x C y respectively. The (111) peaks become sharper at higher deposition temperature, suggesting an increase in grain size. The XRD spectra for films deposited at 700 and 750 C show convoluted peaks which are deconvoluted as shown in Figure s 5 6 and 5 7, respectively. Deconvolution of XRD peaks for film deposited at 700 C affords 5 peaks at 35.40, 37.60, 38.20, 40.00 and 43.10 The two peaks at 37.60 and 43.10 W 2 W 2 N (200) peaks at 37.75 and 43.85 respectively. The three peaks at 35.40, 38.20 and 40.00 are close to the peaks of the binary compounds W 2 W 2 W 2 C (101) with reference values of 34.52, 38. 03 and 39.57 respectively. 178 The deconvolution of XRD spectra for films deposited at 700 C indicates the presence of a physical W 2 W 2 C phases. Deconvolution of XRD spectra for films deposited at 750 C also show 5 peaks at 34.90 37.60, 38.80, 39.80 and 43.10 The peaks at 37.60 and 43.10 match W 2 N (111) W 2 N (200) peaks reasonably well The other three peaks at 34.90, 38.80 and 39.80 again match reasonably well with W 2 W 2 C (002) W 2 C (101) peaks values, respectively. Thus the film deposited at 750 C is a binary mixture of W 2 W 2 C phases with neither of the two phases showing any preferred orientation. As discussed in Chapter 4, GIXD is more sensitive in detecting c rystalline phases in thin films because of its larger interaction volume. As discussed in Chapter 4, XRD measurements for films grown with 3a,b and ammonia showed no peaks while the GIXD spectra showed peaks

PAGE 161

161 W 2 W 2 C. Hence, to asc ertain crystallinity of films grown from 4 with greater accuracy GIXD measurements were done (Figure 5 8). Grazing incidence XRD measurements confirmed that the films deposited by 4 at 400 and 450 C are X ray amorphous as no peaks corresponding to the t hin film were observed. The films deposited between 500 and 650 C are polycrystalline with the peak position indicating the presence of a solid solution of WN x C y W 2 W 2 C. For deposition at 700 and 750 C, the GIXD spectra show peak convolution similar to that observed in XRD spectra. Hence the GIXD results reaffirmed the results obtained from XRD. For XRD measurements, the X rays penetrate completely through the thin film and into the substrate as indicated by the large Si peak observed in the XRD spectra. The intensity of the diffracted beam would depend on both the crystallinity of the film as well as the thickness of the film a ssuming all other parameters such as film density are constant. Since the film thickness is different for deposition at different temperature, XRD measurements cannot be used to compare crystallinity for films deposited at different growth temperature. B ut for GIXD measurement, X rays penetrate only the top 1 2 nm of the film because of the shallow angle of incidence. Hence, even if the films are of varying thickness, GIXD would have a constant interaction volume if all the films are thicker than 2 nm and the depth of penetration of X rays is same for films deposited at different growth temperature. Since the smallest film thickness for growth from 4 is 45 nm, GIXD measurements can be used to compare film crystallinity of films for deposition at differ ent temperature assuming film density is constant. Figure 5 8 shows that the intensity of the WN x C y peak increases as deposition temperature is increased from 500 to 600 C, indicating that the film crystallinity increases. This increase in crystallinity is probably because increased energy is available for crystallite formation and surface diffusion with an

PAGE 162

162 increase in deposition temperature. As deposition temperature is increased from 600 to 650C, the WN x C y peak intensity decreases, suggesting a decre ase in film crystallinity. As shown from XPS analysis ( vide infra ), additional C deposited at 700 C segregates to form both graphitic and diamond like C. It is possible that the C segregation starts at 650 C resulting in a decrease in surface diffusivi ty of the adsorbed species during film growth. The decrease in surface diffusivity causes a decrease in film crystallinity for deposition at 650 C. The film crystallinity decreases further at 700 C deposition temperature as suggested by a decrease in p eak intensity of convoluted W 2 N and W 2 C peaks. The XPS spectrum for deposition at 700 C shows that there is large amount of elemental C (diamond like and graphitic) present in the films, further supporting the assertion that C segregation results in a dec rease in film crystallinity. At 750 C growth temperature, the peak intensity increases again but the increase is observed for the peak corresponding to W 2 C suggesting that the crystallinity increase is due to the growth of W 2 C crystalline phase (Figure 5 7). 5.5 Lattice Parameter W 2 W 2 C have the same crystal structure with only a 2.7% lattice mismatch, it is possible that a solid solution of significant extent is formed. As described elsewhere, XRD peak shifts can be used to estimate the c ombined extent of the solid solution and residual strain. 157 The WN x C y W 2 N and W 2 C has the highest intensity and thus was used for calculating the lattice parameter. Figure 5 9 shows the change in lattice parameter with deposition temperature for films deposited with 4 and 2a,b For films deposited with 4 the lattice parameter is 4.12 for deposition at 500 C, which is nearly identical to the W 2 N. 178 As the deposition temperature increases from 500 to 650 C, there is a concomitant increase in the estimated lattice parameter. Change s in the lattice parameter c an result from an increase in uniform strain in the film or a change in film

PAGE 163

163 composition. Uniform strain can be introduced by differenti al thermal expansion. Since the films grown from 4 are highly disordered and have small grains (< 50 vide post ), minimal uniform strain is expected to exist in these films Thus, the increase in lattice parameter between 500 and 650 C is attributed t o the increase in C content and decrease in N content. For film deposited with 4 at 700 and 750 C, the measured lattice parameter for the W 2 N peak is 4.13 and 4.14 W 2 C for film grown at 700 C are: a = 2.97 and c = 4.68 while the values are: a = 2.92 and c = 5.18 for the film grown at 750 C. Comparison of films deposited with 4 and 2a,b shows that for films grown between 500 and 600 C, those deposited with 4 have a slightly larger lattice paramete r than those deposited with 2a,b This is consistent with the higher C content in the films deposited from 4 The lattice parameter for films deposited from 4 and 2a,b increases with deposition temperature between 500 and 650 C. This increase in lattic e parameter value with deposition temperature is due to an increase in C content with temperature for films deposited with both 4 and 2a,b 5.6 Grain Size As discussed in C hapter 3, the average grain size for polycrystalline materials can be estimated usi ng the Scherrer equation if it is a ssumed that XRD peak broadening is due to the crystallite size distribution. Again, since the (111) reflection was most intense it was used to determine FWHM and estimate the change in grain size with temperature for fi lms deposited with 4 ( Figure 5 10) For films grown a t 500 C, the average grain size of crystallites for material deposited from 4 is 25 Between 500 and 650 to 34 Higher surface diffusivity of atoms at higher deposition temperature facilitates nucleation and grain growth W 2 C and W 2 N phase s is formed. The average grain size for the W 2 C phase, calculated from the (101) peak is

PAGE 164

164 30 and 42 for W 2 N phase, calculated from (111) reflection, is 39 and 47 for deposition at 700 and 750 C, respectively. The comparison of grain size for 4 and 2a,b shows that the average grain size for films deposited with 4 is lower than that for films deposited by 2a,b over the entire deposition temperature range 5.7 Film Growth Rate The growth rate for films grown from 4 was calculated from the measured film thickness ( X SEM ) and deposition time. Figure 5 11 images for films grown between 4 0 0 and 750 C in 50 C intervals. to 24 lowest growth rate observed at 400 C and highest growth rate observed at 700 C. There is an exponential increase in film thickness between 400 and 6 00 C. As the deposition temperature is increased from 6 00 to 7 0 0 C, the re is a small increase in growth rate from 21 to 24 The growth rate decreases to 17 /min as the deposition temperature is increased to 750 C. To determine whether a further decrease in growth rate is observed at higher temperature, growth was attempted at 800 C. There was no hard film deposited at 800 C, however there was a carbon like deposit observed on the substrate surface, which could easily be removed from the Si substrate by a cotton swab. Figure 5 12 shows Arrhenius plots of growth rate for deposition from 4 and 2a,b in the presence of hydrogen. The growth rate for films deposited with 4 is lower than that for films deposited with 2 a,b between 400 and 650 C. The Arrhenius plot for 4 is consistent with film growth between 400 and 6 00 C being surface reaction limited with a change to mass transfer limited between 600 and 700 C. The decrease in growth rate at 750 C and no film gro wth at 800 C is possibly because of homogenous decompos ition of the precursor above 750 C While the transition from surface reaction limited growth to mass transfer limited growth occurs

PAGE 165

165 between 5 50 and 60 0 C for 2a,b 4 shows the same transition at 6 00 C The activation energy for 4 calculated from the Arrhenius relationship for film growth using 4 is 0. 54 eV, which is significantly lower than 0.84 eV activation energy reported for 2a,b The value of activation energy for 4 is within the typical va lues between 0.5 and 1.0 eV observed for CVD growth in surface reaction limited regime 5.9 Atomic Bonding from XPS Measurement Figure 5 13 shows the XPS measurements for films deposited from 4 at 400, 500, 600, and 700 C and Table 3 7 (Chapter 3) summar izes the reference peaks of elemental C, elemental W and different compounds of W, N, C and O obtained from the literature. Table 5 2 and 5 3 summarizes the peak parameters from peak fitting analysis done for the XPS peaks of W, C, N and O. The deconvolu tion of W 4f 7/2 and 4f 5/3 peaks was done by using the peak separation between 4f 7/2 and 4f 5/2 peaks of 2.1 eV and peak area ratio 4f 7/2 : 4f 5/2 of 4:3. The deconvolution of W 4f and 4p peaks for film deposited at 400 C indicates the presence of three diff erent binding states of W. The first W 4f 7/2 peak at 31.9 eV lies between the reference W peaks of W 2 N (~33.0 eV) and W 2 C (~ 31.5 eV), indicating that the W atom is bonded to both C and N. The second 4f 7/2 peak at 34.0 eV lies between the reference WO 2 p eak at 32.9 eV and W 2 O 5 peak at 34.6 eV, suggesting that the phase of tungsten oxide is WO x with the value of x between 2.0 and 2.5. The third W 4f 7/2 peak at 35.7 eV corresponds to the W present in WO 3 As the deposition temperature is increased from 40 0 to 700 C, the intensity of the W peak corresponding to WN x C y increases whereas the intensity of W peak corresponding to both WO x and WO 3 decreases. For deposition at 400 C, the N 1s peak is at 397.8 eV and corresponds well with the reference N 1s peak for WN x film. 119 As the temperature is increased to 500 C, a broad shoulder which is centered at 399.2 eV develops. This broad peak indicates that a small amount

PAGE 166

166 of N is present in amorphous state most probably at grain boundaries. 119 The relative inte nsity of the amorphous N peak around 399.5 eV increases as the deposition temperature is increased from 500 to 600 C. This indicates that although the overall N content of film decreases as deposition temperature is increased from 500 to 600 C, the rela tive % of N present in the amorphous state increases with respect to that bonded to W. The XPS spectra at 700 C shows no N 1s peak even though the AES measurements indicate the presence of N because of the poor detection limit of XPS for detecting light elements (N) in a heavy element matrix (W). For deposition at 400 C, a small broad C 1s peak at 283.8 eV is observed in the XPS spectrum. This peak position is slightly higher than the reference C 1s peak for W 2 C at ~ 283.5 eV. At 500 C, the C 1s peak can be deconvoluted into two different peaks at 283.5 eV and 284.9 eV. The former peak corresponds to C bonded to W and the latter peak reflects C present in amorphous phase. As the deposition temperature is increased from 500 to 600 C, the peak intensi ty of both carbidic and amorphous C increases. The C 1s spectrum for deposition at 700 C can be deconvoluted into three distinct peaks at 283.6 eV, 284.5 eV, and 285.5 eV. The first peak at 283.6 corresponds to carbidic carbon bonded to W. The second a nd third peaks at 284.5 and 285.5 eV reflect the presence of graphitic carbon (sp 2 bonded C) and diamond like C (sp 3 bonded C). Previous studies on diamond like C films have shown the presence of C as both graphitic C with a peak at ca. 284.5 eV and diamo nd like C with a broad peak at ca. 285.3 eV. 192,193 The presence of graphitic and diamond like C for films deposited at 700 C shows that W 2 C, C separates from the WN x C y as graphitic and diamond like C. The O 1s peak for deposition at 400 C can be deconvoluted into two peaks at 530.8 and 531.8 eV. The peak at 530.8 eV corresponds to O bonded to W in WO x or WO 3 Since O is

PAGE 167

167 present as O 2 the O peak does not show peak separation between the two different bonding states in WO x and WO 3 The other peak at 531.8 eV corresponds to O present in WN x C y Previous study on tungsten oxynitride has shown that XPS spectr a for O 1s show a peak at 531.7 eV for O present in WN x matrix. 203 The O 1s peaks for films deposited at 500, 600 and 700 C can be similarly deconvoluted to two peaks corresponding to O present as WO x or WO 3 and O present in the WN x C y matrix. To summariz e the XPS results, film deposited at 400 C shows presence of the WN x C y with a significant amount of WO x and WO 3 At 500 C, the bulk of the film is WN x C y with small amounts of WO x and WO 3 Additionally, film deposited at 500 C shows the presence of bot h C and N in an amorphous state probably at grain boundaries. Incidentally, the lowest temperature at which the presence of amorphous C and N is observed is 500 C, which is also the temperature at which crystallinity is first observed. The film deposite d at 600 C consists of WN x C y a small amount of WO x /WO 3 together with amorphous N and C. At 700 C, the film shows the presence of WN x C y graphitic and diamond like C and a small amount of WO x /WO 3 5.8 Film Resistivity Figure 5 14 shows the film resistiv ity at different deposition temperature for films grown with precursors 4 and 2a,b For depositions with 4 at 400 C, the resistivity is 1150 The resistivity decreases to 980 for deposition at 450 C due to higher W content and lower C and N content of the film. Between 450 and 600 C deposition temperature, the film resistivity gradually increases from 980 to 6857 ncrease is attributed to a decrease in W content of the film. The r esistivity of films deposited at 700 and 750 C is ca. 2400 The XRD spectra indicate the formation of a W 2 N and W 2 C at these temperatures and XPS measuremen t shows the separation of C to form graphitic C and diamond like C. The phase separation at 700 and 750 C results in segregation of excess amorphous C as

PAGE 168

168 W 2 C, graphitic C and diamond like C resulting in decrease of film resistivity. The decrease in re sistivity as deposition temperature is increased from 600 to 650 C could result from the onset of phase separation at 650 C, even though the phase separation is not clearly seen in the XRD spectrum at 650 C. Comparison of resistivity for films deposited with 4 and 2a,b reveals that w hile the lowest resistivity obtained with 2a,b C, the lowest resistivity obtained with 4 is 980 C. The resistivity of films deposited with 2a,b increases throughout the deposition temperature, whereas the resistivity of films deposited with 4 show a steep increase between 450 and 600 C followed by a decrease in resistivity between 600 and 750 C because of the phase separation of W 2 N and W 2 C. 5.9 Diffusion Barrier Testing To determine the effectiveness of diffusion barrier deposited with 4 barrier film s deposited at 450 and 5 0 0 C w ere coated with 100 nm PVD Cu The thickness of films deposited at 450 and 500 was 45 and 55 nm, respectively. The Cu/barrier/Si stack was annealed in vacuum at temperature ranging from 200 to 600 C for 30 min to determine the tempera ture at which Cu diffuses through the barrier film into the Si substrate. After annealing, three point AES depth profile and XRD measurements were done to detect copper diffusion through the barrier film Figure 5 16 shows the depth profile for a post an neal Cu/WN x C y /Si stack for WN x C y films deposited from 4 at 450 C. For anneal at 200 C for 30 min, the AES depth profile shows that the Cu/ WN x C y interface is similar to that for films that were not annealed (Figure 5 15 ), suggesting that no bulk copper d iffusion had occurred. As the anneal temperature is increased to 400 C, there is slight mixing of the Cu/WN x C y interface, however the copper has not diffused through the barrier film. For 500 C anneal, there is further diffusion of Cu in the WN x C y fil m, but the barrier film is able to prevent complete diffusion of Cu through it. A fter annealing at 600

PAGE 169

169 C, the AES profile shows that there is complete diffusion of copper through the barrier film. Intermixing is also observed for the WN x C y /Si interface. The depth profiling shows that 45 nm barrier film of WN x C y deposited from 4 at 450 C is able to prevent bulk Cu diffusion when annealed at 500 C for 30 min. Figure 5 17 shows the XRD plots for films deposited from 4 at 450 C before annealing and after annealing at 200, 400, 500 and 600 C. No peak(s) corresponding to WN x C y is observed because film deposited at 450 C is X ray amorphous. As discussed in the literature review chapter, the recrystallization temperature is important for diffusion barrier film because diffusion along grain boundaries is believed to be the primary cause for barrier failure. Amorphous W 2 N film deposited by PVD recrystallizes after annealing at 600 C. 182 However, for WN x C y film deposited by 4 the film does not recrystalli ze even after annealing at 600 C as seen from the XRD pattern. This is expected since ternary compounds tend to have higher recrystallization temperature as compared to binary compounds. The pre anneal XRD profile shows peaks at 31.1, 50.75 an peaks correspond to Cu (111), Cu (200) and Cu (220) reflections, respectively. After annealing at 200 C, the intensity of Cu peaks increases because of grain growth in the Cu film. No peak corresponding to CuSi x compound is observed indicating that there is no bulk diffusion of copper after annealing at 200 C. As the annealing temperature is increased, the intensity of Cu peaks further increases because of grain growth, however, no other peaks is observed besides peaks corr esponding to Cu and Si substrate suggesting that bulk phase copper diffusion did not occur even after annealing at 600 C. This result contradicts the AES measurement which shows that Cu has diffused through the barrier film and into Si substrate after an nealing at 600 C. This apparent contradiction occurs because XRD can detect CuSi x compound formation only after

PAGE 170

170 sufficient amount of the silicide has formed. If a small amount of Cu has diffused through the barrier at the onset of barrier failure, XRD i s not able to detect small quantities of CuSi x whereas AES can detect Cu in barrier film or Si even if the Cu content is ca. 1 at. % because of the higher sensitivity of AES. As discussed in the Chapter 2, XRD and four point probe are the least sensitive techniques for detecting copper diffusion through the barrier. Previous studies have also shown that even though the barrier film has failed at lower annealing temperature, XRD only detects the barrier failure at much higher anneal temperature. 63 Figure 5 18 shows the 3 point AES depth profile for a post anneal Cu/WN x C y /Si stack for WN x C y films deposited from 4 at 500 C. After annealing at 200 C and 400 C for 30 min, the Cu WN x C y interface is similar to that for films that were not annealed suggesting that no bulk diffusion of copper has occurred. For 500 C annealing, slight mixing of Cu WN x C y interface is evident in the AES depth profile. Copper diffuses through the barrier film after annealing at 600 C and significant intermixing of Cu and WN x C y is observed in AES depth profile The AES depth profiling shows that 5 5 nm barrier film of WN x C y deposited with 4 at 500 C is also able to prevent Cu diffusion when annealed at 500 C for 30 min. Figure 5 19 shows the XRD measurements on a Cu/WN x C y /Si st ack before annealing and after annealing at 200, 400, 500 and 600 C. Similar to film deposited at 450 C, the film deposited at 500 C do not show recrystallization of the barrier film. The XRD spectra do not indicate formation of CuSi x even after annea ling at 600 C for the reason discussed above even though the barrier film has failed 5 10 Conclusions It has been demonstrated that the mixed imido guanidinato complex W(N i Pr)Cl 3 [ i PrNC(NMe 2 )N i Pr] ( 4 ) can be used in an aerosol assisted CVD system to depo sit WN x C y thin films. A comparison of effect of Cl 4 (CH 3 CN) and Cl 3 ( i Pr(NMe 2 )N i Pr ligands on film properties is presented by comparing the film composition, crystallinity, lattice parameter,

PAGE 171

171 grain size, growth rate and resistivity of films deposited with 4 and 2a,b The AES spectra showed the presence of W, N, C and O in this material When compared with films grown with 2a,b the films grown with 4 have a higher C/W ratio and a similar N/W ratio. The O content of films deposited with 4 is also significa ntly lower than that for films deposited with 2a,b especially at lower deposition temperature. The lowest growth temperature for 4 is 400 C, which is 50 C lower than that for 2a,b For both 4 and 2a,b films grown at and above 500 C are crystalline From the diffusion barrier application standpoint, films deposited with 4 at 450 and 500 C were able to prevent copper diffusion after annealing at 500 C for 30 min in vacuum. Since the film deposited from 4 at 450 C is amorphous, has the low est resist ivity of 980 4 at 450 C are candidate s for diffusion barrier application.

PAGE 172

172 Table 5 1. Selected bond distances ( ) and bond angles () for 4 Bond Length () Bond Angle (degree) W N1 2.247(4) N1 W N3 163.23(18) W N2 1.961(4) N2 W Cl3 155.81(13) W N3 1.702(4) Cl1 W Cl2 167.30(5) W Cl1 2.3752(15) N2 W N3 101.44(19) W Cl2 2.3819(16) N1 W N2 61.88(16) W Cl3 2.3833(14) W N1 C1 90.3(3) C1 N2 1.399(6) N1 C1 N2 107.8(4) C1 N1 1.294(6) W N2 C1 100.0(3) C1 N4 1.373(6) W N3 C10 168.4(8)

PAGE 173

173 Table 5 2. Analysis results from the deconvolution of the XPS peak of W for films deposited from 4 on Si(100) substrate at 400, 500, 600 and 700 C Deposition WN x C y WO x (2.0 < x < 2. 5) WO 3 temperature Peak Position % Area FWHM Peak Position % Area FWHM Peak Position % Area FWHM (eV) (eV) (eV) (eV) (eV) 400 W4f 7/2 31.95 20.4 1.61 34.01 17.0 1.84 35.74 15.9 2.00 W4f 5/2 33.90 15.3 1.94 36.01 11.1 1.97 37.94 11.9 1.81 W4p 3/2 37.10 3.8 2.15 39.27 2.8 1.64 41.04 1.8 1.20 500 W4f 7/2 31.89 40.4 1.36 33.51 8.7 1.52 37.61 4.8 1.96 W4f 5/2 34.06 30.3 1.48 35.61 6.6 2.17 37.49 1.0 1.75 W4p 3/2 35.51 6.3 2.02 38.81 1.3 1.27 41.11 0.6 2.16 600 W4f 7/2 32.01 38.2 1.30 33.72 7.8 1.61 35.31 7.4 1.78 W4f 5/2 34.10 29.8 1.50 35.59 5.9 2.01 37.41 5.6 2.44 W4p 3/2 37.40 3.0 1.98 38.89 0.9 1.41 40.71 1.4 3.45 700 W4f 7/2 32.05 36.3 1.28 33.54 9.9 1.61 35.33 7.5 2.14 W4f 5/2 34.15 27.2 1.33 35.64 7.4 1.86 37.53 5.6 2.15 W4p 3/2 37.45 2.7 2.04 38.94 2.0 1.22 40.83 1.4 2.10

PAGE 174

174 Table 5 3. Analysis results from the deconvolution of XPS peaks of C, N and O for films deposited from 4 on Si(100) substrate at 400, 500, 600 and 700 C Deposition C 1s N1s O1s temperature Peak Position % A rea FWHM Peak Position % Area FWHM Peak Position % Area FWHM (eV) (eV) (eV) (eV) (eV) (eV) 400 Peak 1 283.83 100 .0 2.29 397.81 100 .0 1.83 530.85 15.9 2 .00 Peak 2 ------531.85 11.9 1.81 500 Peak 1 283.51 73.7 1.48 397.71 84.1 1.5 0 530. 77 66.2 1.59 Peak 2 284.95 26.3 2.19 399.25 15.9 1.77 532.11 33.8 2.03 600 Peak 1 283.59 72.3 1.54 397.73 64.6 1.73 530.57 49.4 1.46 Peak 2 285.06 27.7 1.98 399.49 35.4 2.87 531.69 50.6 2.02 700 Peak 1 283.6 0 19.2 1.53 ---530.77 52.4 1.81 Pea k 2 284.5 0 44.2 1.64 ---531.91 47.6 2.48 Peak 3 285.5 0 36.6 2.43 ------

PAGE 175

175 Figure 5 1. Thermal ellipsoids diagram of the molecular structure of W(N i Pr)Cl 3 [ i Pr(NMe 2 )N i Pr]. T hermal ellipsoids are drawn at 40 % probability, and the hyd rogen atoms have been omitted for clarity. 202

PAGE 176

176 Figure 5 2 Composition of films deposited from 4 and 2a,b on Si (100) substrate at different deposition temperature as determined by AES after 0.5 min of sputtering

PAGE 177

177 Figure 5 3. N/W and C/W atomic ratio s for films deposited from 4 and 2a,b on Si (100) substrate at different deposition temperature as determined by AES after 0.5 min of sputtering

PAGE 178

178 Figure 5 4. X ray photoelectron spectroscopy patterns for films deposited from 4 at 400, 500, 600 and 700 C on Si(100) surface

PAGE 179

179 Figure 5 5. X ray diffraction patterns for films deposited from 4 on Si (100) substrate

PAGE 180

180 F igure 5 6. Deconvolution of XRD peaks for films deposited from 4 at 700 C on a Si (100) substrate. The dashed line represents raw data, s olid line s represent individual peaks W 2 W 2 C used for deconvolution and the bold solid line represents the profile calculated from individual peaks

PAGE 181

181 Figure 5 7 Deconvolution of XRD peaks for films deposited from 4 at 750 C on a Si (100) s ubstrate. The dashed line represents raw data, solid line s represent individual peaks W 2 W 2 C used for deconvolution and the bold solid line represents the profile calculated from individual peaks

PAGE 182

182 Figure 5 8. Grazing incidence X ray diffra ction spectra for films deposited from 4 on Si (100) substrate

PAGE 183

183 Figure 5 9 Lattice parameters for films grown from 4 and 2a,b on Si (100) substrate. T he solid line at 4.126 corresponds to the standard lattice parameter for 2 N. The dotted line at 4.236 corresponds to standard lattice parameter for 2 C (111). Error bars indicate uncertainty in determination of peak position for lattice parameter calculation

PAGE 184

184 Figure 5 10 Average grain size at different deposition temperature for films grown from 4 and 2a,b on Si (100) substrate

PAGE 185

185 Figure 5 11 Scanning electron microscope images for films grown from 4 on Si (100) substrate between 400 and 750 C in 50 C increments

PAGE 186

186 Figure 5 12 Arrhenius plot of log of film growth rate vs. inverse temper ature for deposition from 4 and 2a,b on a Si(100) substrate

PAGE 187

187 Figure 5 1 3 Deconvolution of the XPS peaks of W, N, C and O for film deposited from 4 at 400, 500, 600 and 700 C on Si (100) substrate

PAGE 188

188 Figure 5 14 Change in film resistivity with deposit ion temperature for films grown on Si (100) from 4 and 2a,b

PAGE 189

189 Figure 5 15 Pre anneal AES depth profile of Cu (100 nm)/ WN x C y ( 45 nm)/Si (substrate) stack for WN x C y film deposited at 450 C (using precursor 4 ) showing the Cu barrier interface.

PAGE 190

190 Figure 5 16 A uger electron spectroscopy depth profile s of Cu (100 nm)/ WN x C y (45 nm)/Si ( 100 ) stack for WN x C y film deposited at 450 C (using precursor 4 ) and anneal ed in vacuum for 30 min at A ) 200 C B ) 400 C C ) 500 C D ) 600 C.

PAGE 191

1 91 Figure 5 17. X ray d iffraction plots for Cu/WN x C y /Si stack (WN x C y films deposited from 4 at 450 C) before annealing and after annealing at 200, 400, 500 and 600 C

PAGE 192

192 Figure 5 18 A uger electron spectroscopy depth profile s of Cu (100 nm)/WN x C y (55 nm)/Si ( 100 ) stack for WN x C y film deposited at 500 C (using precursor 4 ) and annealed in vacuum for 30 min at A ) 200 C B ) 400 C C ) 500 C D ) 600 C.

PAGE 193

193 Figure 5 19. X ray diffraction plots for Cu/WN x C y /Si stack (WN x C y films deposited with 4 at 500 C) before annealing and af ter annealing at 200, 400, 500 and 600 C

PAGE 194

194 CHAPTER 6 DEPOSITION OF WN x C y FROM W(N i P r )C l 3 [ i P r NC(NM e 2 )N i P r ] : EFFECT OF NH 3 CO REACTANT ON FILM PRO PERTIES C hapter 5 discussed the growth of WN x C y films from 4 and H 2 co reactant W hen 4 is used with H 2 as co reactant the films deposited by CVD are W rich and deficient in N. The highest N content of 12 at. % is observed for deposition from 4 at 400 C. To increase the film N content, the effect of NH 3 co reactant on deposition of WN x C y using 4 was investi gated The effect of NH 3 co reactant on lowest deposition temperature, film crystallinity, growth rate and resistivity was also studied Two different experiments were done, one using a mixture of NH 3 and H 2 as co reactants (designated as 1A) and other u sing a mixture of NH 3 and N 2 as co reactants (designated as 1B). Table 6 1 shows the molar flow rate of reactants for 1A and 1B. Films deposited for 1A and 1B are compared to those deposited with only H 2 as co reactant (designated as 1C) to understand th e effect of NH 3 (with or without H 2 ) on film properties. 6.1 Film Growth Studies Films grown by procedures 1A and 1B were generally smooth and had a shiny metallic surface with film color varying from golden to shiny black. The lowest temperature at whic h appreciable film growth was observed for 1A and 1B was 300 C. The use of NH 3 (with or without H 2 ) lowers the deposition temperature by 100 C (two experimental increments). Addition of NH 3 as co reactant lowers the energy barrier for film growth resul ting in film deposition at lower temperature. 6.2 Film Composition Figure 6 1 shows AES results for 1A, 1B and 1C. For films deposited with co reactants NH 3 + H 2 (procedure 1A ) the W content varies between 51 and 59 at. % be tween 300 and 650 C. The N c ontent in the film is relatively constant at ca. 18 at. % b etween 300 and 400 C with the highest valu e of 21 at. % obser ved for deposition at 450 C. Between 450 and 750 C, the N

PAGE 195

195 content gradually decreases from 21 at. % to 10 at. % The sources of N i n the film are 4 NH 3 and benzonitrile. The decrease in N content with increase in deposition temperature could result from cleavage of N containing ligands from the precursor, lower incorporation of N from NH 3 and benzonitrile or volatilization of N from the film at higher deposition temperature. Volatilization of N from WN x C y films has been reported to occur at 700 C, 89 hence the decrease in N content between 450 and 700 C is due to decrease in N incorporation from precursor and/or NH 3 The C content of the film monotonically increases from 11 at. % at 350 C to 43 at. % at 750 C deposition temperature. The increased C content of the film is due to incorporation of C from the precursor and/or benzonitrile into the film. The O content of the film de creases from 17 at. % at 350 C to 4 at. % at 500 C and remains constant at ca. 4 at. % between 500 and 750 C. Oxygen can be incorporated into the film during growth or post growth. The sources of O during film growth are residual gases (oxygen and wat er vapor) in the reactor and O impurity in the precursor (usually in the ppm levels). Oxygen can also be incorporated into the film due to post growth exposure to air. The XPS results ( vide infra ) show that O in the film is incorporated both during growt h and post growth. The O incorporated during film growth forms WO x (2 < x < 2.5)/WO 3 whereas the O incorporated due to post growth exposure to air is incorporated in the WN x C y lattice. For films deposited with only NH 3 co reactant (procedure 1B), the W a nd O content of film decreases and N and C content of film increases between 300 and 500 C. Between 500 and 750 C, the W and O content of the film remains relatively unchanged at ca. 39 at. % and 3 at. % respectively, the N content of film decreases fro m 30 at. % to 15 at. % and the C content of the film increases from 30 to 49 at. % This observation suggests that at 500 C deposition temperature, there is a major change in mechanism of reaction for 1B. Deposition temperature

PAGE 196

196 higher than 500 C trigge rs decrease in N content possibly due to cleavage of N containing ligand(s) in the precursor or lower N incorporation from NH 3 Comparison of composition of films deposited for 1A, 1B and 1C shows that the most prominent effect of using NH 3 as co reactant (with or without H 2 ) is the significant increase in film N content throughout the deposition temperature range. In general, the film N content shows a twofold increase for 1A and a threefold increase for 1B as compared to 1C. While previous studies on 2a ,b and 3a,b have shown an increase in film N content when NH 3 + H 2 is used as co reactant, the work on precursor 4 shows that the film N content can be further increased by using only NH 3 as co reactant. 6.3 Film Crystallinity Figure 6 2 shows the XRD patt erns for films deposited for 1A between 300 and 750 C. The XRD spectra for films deposited between 300 and 450 C show no peaks attributable to the spectra suggests that films deposited f or 1A below 500 C are X ray amorphous. The XRD pattern for the film deposited at 500 C shows the emergence of crystallinity as evidenced by the W 2 W 2 phase and 178 WN x C y W 2 W 2 C. Since the intensity of the most int ense reflection (111) is small, the other peaks corresponding to (220) and (311) reflections at 63.55 and 75.05 are not clearly visible in the XRD spectra. Figure 6 3 shows the XRD patterns for films deposited for 1B between 300 and 750 C. The XRD spec tra for 1B show no peaks

PAGE 197

197 attributable to the film, but only peaks associated with the Si substrate suggesting that films deposited are X ray amorphous throughout the deposition temperature range. The comparison of films deposited for 1A, 1B and 1C shows th at the films are amorphous for deposition below 500 C irrespective of the co reactant used. But at and above 500 C, 1A and 1C result in formation of polycrystalline phase s while 1B films are amorphous. This suggests that when H 2 is used as co reactant for 1A and 1C, the reaction mechanism at temperature above 450 C encourages the formation of crystallites, resulting in formation of polycrystalline thin films. When only NH 3 is used as co reactant for 1B, the reaction mechanism hinders the formation of crystallites resulting in deposition of amorphous films throughout the deposition temperature r ange. 6.4 Lattice Parameter Figure 6 4 shows the change in lattice parameter with deposition temperature for 1A and WN x C y has the highes t intensity and thus was used for calculating the lattice parameter. For 1A, the lattice parameter is 4.17 for deposition at 500 C, which is between the literature value of lattice parameter for W 2 W 2 C (4.236 ). As the deposition t emperature is increased from 500 to 550 C, the lattice parameter decreases to W 2 W 2 C is only 2.7 %, it is possible that W, N and C form a solid solution. For a WN x C y solid solution, an increase in inte rstitial concentration (C, N and/or O) in the non metal sublattice would result in an increase in lattice parameter while an increase in vacancy concentration in the non metal sublattice would result in a decrease in lattice parameter. Transition metal ni trides and carbides are known to have appreciable vacancy concentration on both non metal and metal sublattices. 39 Since the (C+N+O)/W ratio remains unchanged between 500 and 550 C, the decrease in lattice parameter suggests an increase in vacancy concen tration for deposition at 550 C. Between 550 and 700

PAGE 198

198 C, the lattice parameter gradually increases from 4.15 to 4.19 In the same temperature range, the (C+N+O)/W ratio increases primarily because of an increase in C content while the N and O content of the film decreases. The additional C can be incorporated into the W sublattice in carbidic form or outside the W sublattice in amorphous form. The i ncrease in lattice parameter between 550 and 700 C suggests that there is a net increase in the amount of C present in the W sublattice as the temperature is increased from 550 to 700 C. As the deposition temperature is increased from 700 to 750 C, the lattice parameter decreases from 4.19 to 4.15 even though the C content of the film increases sugges ting either a net decrease in amount of C present in W sublattice because of phase separation of C to form graphitic/ diamond like C ( vide infra ) or an increase in vacancy concentration. Comparison of films deposited with procedures 1A and 1C shows that fo r films grown at 500 C, 1A results in a higher lattice parameter than 1C because of the higher (C+N+O)/W ratio. For deposition between 550 and 650 C, 1A results in a higher lattice parameter than 1C even though the (C+N+O)/W ratio for 1A is smaller than that by procedure 1C indicating that films deposited for 1C have a higher vacancy concentration in that temperature range. For deposition between 700 and 750 C, 1C affords a lower lattice parameter than 1A because 1C results in W 2 C causing a decrease in C content in the W sublattice of WN x C y phase. 6.5 Grain Size For polycrystalline materials, the XRD peak broadening could be due to crystallite size distribution, instrumental broadening or uniform strain in the film. Line profile analysis can be used to deconvolute the effect of crystallite size distribution from the effect of uniform strain on XRD peak broadening. Line profile analysis of XRD spectra for 1A indicated that the peak broadening was primarily due to crystallite size distribution and the effect of uniform strain on peak broadening was negligible. If the XRD peak broadening is entirely due to crystallite size

PAGE 199

199 e. Since the (111) reflection was most intense, it was used to determine FWHM and estimate grain size for 1A and 1C. Figure 6 equation. For 1A, the average grain size is approx. 48 fo r deposition between 500 and 650 C. Average grain size increases slightly to 55 for deposition at 700 C and drops again to 50 for deposition at 750 C Overall, the average grain size measurement indicates that the films deposited by 1A at and abov e 500 C are nanocrystalline and the effect of deposition temperature on average grain size is minimal Comparison of films deposited using 1A and 1C shows that films deposited for 1A have a larger average grain size than those deposited for 1C over the en tire deposition temperature range. 6.6 Film Growth Rate The growth rate for 1A and 1B was calculated from film thickness measurement using X SEM. Figure s 6 6 and 6 7 show temperature for 1A and 1B respectively between 300 and 750 C. Figure 6 8 shows the Arrhenius plot for films grown for 1A, 1B and 1C. F or f ilms grow n between 300 and 550 C by 1A the growth rate increases logarithmically with inverse of temperature, indicating that the growth is in the kinetic controlled growth regime and surface reacti on is the rate limiting step. As discussed earlier, the slope of an Arrhenius plot can be used to calculate the value of E a For 1A, the apparent activation energy is 0.45 eV. Between 550 and 750 C, the Arrhenius plot for 1A shows a small negative slope Growth for 1A between 550 and 750 C is in the diffusion lim ited growth regime and the negative slope is probably because of homogenous gas phase decomposition of the precursor, which leads to decrease in growth rate with increase in deposition temperature. For 1B, the Arrhenius plot shows that film growth between 300 and 400 C is in the kinetically controlled growth regime

PAGE 200

200 and the apparent activation energy for film growth is 0.27 eV. Between 400 and 750 C, the film growth rate for 1B is almost constant indicating that the growth is in the diffusion limited gro wth regime. Comparison of film growth for 1A, 1B and 1C shows that the rates of films grow th by 1C have the highest value for E a of 0.54 eV while the growth for 1A has the lowest value of E a of 0.27 eV. While the E a value for 1B is higher than 1A, the gro wth rate for 1B is significantly higher than that for 1A suggesting that the addition of H 2 to NH 3 significantly increases the film growth rate. Overall, the selection of co reactant (s) has a significant impact on the film growth rate and the lowest deposi tion temperature for precursor 4 The temperature at which the growth mechanism changes from kinetically controlled to diffusion limited is also different for 1A, 1B and 1C because of difference in mechanism s of film growth and the change s in hydrodynamic properties fo r different co reactant mixture s. 6.7 Atomic Bonding Figure 6 9 shows the results of XPS peak deconvolution and Table s 6 2 and 6 3 show the peak fit parameters obtained from deconvolution for films deposited by 1A. The reference peaks for e lemental C, elemental W and different compounds of W, N, C and O are listed in Table 3 7 (Chapter 3). The deconvolution of W 4f 7/2 and 4f 5/3 peaks was done by using the peak separation between the 4f 7/2 and 4f 5/2 peaks of 2.1 eV and peak area ratio 4f 7/2 : 4f 5/2 of 4:3. The deconvolution of W 4f and 4p peaks for film deposited by 1A at 3 00 C indicates the presence of three different bonding states of W. The first W 4f 7/2 peak at 32.1 eV lies between the reference W peaks of W 2 N (~33.0 eV) and W 2 C (~ 31.5 eV), indicating that the W atom is bonded to both C and N. The second 4f 7/2 peak at 33.7 eV lies between the WO 2 peak at 32.9 eV and W 2 O 5 peak at 34.6 eV, suggesting that the phase of tungsten oxide is WO x (2 < x < 2.5) The third W 4f 7/2 peak at 35.7 e V corresponds to W present in WO 3 As the deposition temperature is increased

PAGE 201

201 from 300 to 400 C, the W peak corresponding to WN x C y increases whereas the intensity of W peak corresponding to WO x and WO 3 decreases, which is consistent with the AES results that shows a decrease in O content of the film in the same deposition temperature range. For deposition at 500 C and above for 1A, the W peaks can be deconvoluted into only WN x C y and WO x peaks and there is no evidence of the presence of WO 3 Comparison of W peaks for 1A and 1C shows that while 1C shows evidence of WO 3 throughout the deposition temperature range, 1A shows the presence of WO 3 only below 500 C suggesting that use of NH 3 as co reactant suppresses the formation of WO 3 phase. The N peak for 1 A at 3 00 C can be deconvoluted into two peaks. The first N 1s peak at 397. 7 eV corresponds well with the reference N 1s peak for WN x film 119 The second peak at 399.9 eV suggests that N is also present as amorphous N. 119 Between 300 and 700 C, the N p eak can be similarly deconvoluted into a dominant peak around 397.5 eV and a broad shallow peak around 399.5 eV implying that the bulk of the N is bonded to W and a small amount of N is present in amorphous phase. Between 300 and 500 C, the C 1s peak for 1A can be deconvoluted into two peaks around 283. 6 eV and 285.4 eV. The former peak corresponds to C bonded to W (reference peak at 283.5 eV) and the latter peak reflect C present as amorphous carbon (reference peak between 284.5 and 285.2 eV) For depo sition at 600 and 700 C, the C 1s spectra can be deconvoluted into three distinct peaks at 283.5 eV, 284.5 and 285. 8 eV. The first peak corresponds to carbidic carbon bonded to W. The second and third peaks at 284.5 and 285.5 eV reflect the presence of graphitic carbon (sp 2 bonded C) and diamond like C (sp 3 bonded C) respectively Previous reports have indicated that the XPS peaks for graphitic C and diamond like C appear at ca. 284.5 eV and ca. 285.3 eV respectively 192,193 The presence of graphitic a nd diamond like C for films

PAGE 202

202 deposited at 600 and 700 C shows that C separates from the WN x C y phase as has been reported previously for 1C While films deposited by 1C show evidence of phase separation of C to form graphitic and diamond like C only at 700 C, films deposited for 1A and 1B show the initiation of similar process at 600 C suggesting that the use of NH 3 co reactant (with or without H 2 ) lowers the initiation temperature for C phase separation. The O 1s peak for deposition between 300 and 7 00 C can be deconvoluted into two peaks at ca. 530. 7 and ca. 53 2.0 eV. The peak at 530. 7 eV corresponds to O bonded as WO x or WO 3 The other pea k at 532.0 eV corresponds to O present in WN x C y As discussed Chapter 6, O could be incorporated into the film e ither during film growth because of oxygen and water vapor impurity in the reactor or after the film growth because of exposure to atmosphere. The presence of WO x /WO 3 phase suggests that residual gas in reactor and/or O impurity in the precursor react(s) with the precursor to form WO x /WO 3 during the film growth. The source of O bound to W in WN x C y is most likely atmospheric O that diffuses through the film when the film is exposed to atmosphere. In summary, deconvolution of the O 1s peak suggests that O is incorporated in the film during growth as WO x /WO 3 and post growth as O loosely bound to W in WN x C y XPS results for 1A show that film s deposited between 3 00 and 400 C are WN x C y with significant amount of WO x and WO 3 along with small amounts of amorphou s N and C For deposition between 400 and 500 C, the bulk of the film is WN x C y with a small amount of WO x and increased amounts of amorphous C and N The film deposited at 600 and 700 C consists of WN x C y a small amount of WO x along with graphitic and diamond like C. XPS measurements for 1B (Figure 6 10, Tables 6 4 and 6 5) show that there is no major difference in the bonding states of W, N and O for 1A and 1B. The only major difference is seen

PAGE 203

203 for the bonding state of C, especially for deposition bel ow 500 C. For deposition at 300, 350 and 400 C, the C 1s peak for 1A shows that C is present in both carbidic and amorphous phases, whereas the C 1s peak for 1B shows that C is present in only carbidic form. For 1B, the broad amorphous C peak at 284.8 eV appears only at 500 C deposition temperature. This observation suggests that the use of only NH 3 as co reactant favors the deposition of carbidic C, particularly at low deposition temperatures. 6.8 Film Resistivity Figure 6 11 shows the variation of f ilm resistivity with deposition temperature for films grown using procedures 1A, 1B and 1C. For 1A at 3 the deposition temperature is increased from 300 to 600 C, the film resistivity gradually cm possibly due to increase in C content of the film. Film resistivity peaks for deposition at 650 C and decreases thereafter. Deposition for 1B shows that low resistivity cm) films are deposited at 300 C. For growth between 350 and 400 C, there is a sharp increase in film resistivity because of the decrease in W content of the film. Between 400 and 550 C resistivity increases due to the increase in C and N content of the film. Above 500 C, resistivity decreases for 1B because of a decrease in N content of the film. Comparison of film resistivity for 1A, 1B and 1C shows that the co reactan t used for deposition of WN x C y from 4 can have a significant effect on resistivity. When either only H 2 or only NH 3 are used as co reactants, low resistivity films are deposited at low deposition temperature. However, when both H 2 and NH 3 are used togeth er for deposition, high resistivity films are deposited throughout the deposition temperature range. A number of factors could affect the film resistivity including metal to non metal ratio, crystallinity, grain size, density and bonding state of non meta ls; the choice of co reactants affect each of these factors due to different reaction mechanisms for each co reactant mixture.

PAGE 204

204 6.9 Diffusion Barrier Testing To determine the effectiveness of diff usion barrier film deposited by procedures 1A and 1B, barrier films deposited at 300, 350 and 400 C were coated with 100 nm PVD Cu. Prior to the deposition of Cu thin film, the barrier film was exposed to atmosphere for approximately 1 2 hour(s). The Cu/barrier/Si stack was annealed in vacuum at 500 C for 30 min After annealing, three point AES depth profile and XRD measurements were done to detect copper diffusion through the barrier film. 6.9.1 Barrier Testing for 1A Figure 6 1 2 shows the depth profile of pre and post anneal Cu/WN x C y (1A) /Si stack s for WN x C y films deposited at 300, 350 and 40 0 C. The thickness of the film deposited at 300, 350 and 400 C was 37, 75 and 75 nm respectively. For deposition at 300 C, the pre anneal depth profile ( Figure 6 12 A) indicates that the copper has already penetrate d through the barrier into vide supra ). The post anneal profile ( Figure 6 12 B) for deposition at 300 C shows that the Cu has penetrated deeper into the barrier fi lm, indicating that the barrier film was unable to prevent Cu diffusion. For deposition at 350 C, the comparison of pre and post anneal depth profile ( Figure 6 12 C and D) shows that copper diffuses through the barrier film after annealing. For deposit ion at 400 C, the pre anneal depth profile shows sharp interfaces between Cu/ barrier and barrier/Si. After annealing at 500 C for 30 min, Cu has further penetrated into the barrier but not completely, indicating that 75 nm WN x C y film (1A) was able to p revent Cu diffusion through the barrier film after vacuum annealing at 500 C. Intermixing is seen in the depth profile for both Cu/ barrier interface and the barrier/ Si interface. A number of factors could influence the diffusion barrier performance o f WN x C y including film density and composition. For 1A, the film composition is relatively constant for deposition

PAGE 205

205 at 300, 350 and 400 C. To determine the density of the films for different deposition temperature, an indirect method is the comparison of their relative sputter rate. For films with similar composition, denser films would have a lower sputter rate as compared to less dense films and vice versa. Table 6 6 shows the sputter rate for the barrier film deposited by procedure 1A at 300, 350 and 400 C. Comparison of the relative sputter rate for 1A shows that film deposited at 300 and 350 C have a higher sputter rate and hence lower density as compared to film deposited at 400 C. So, the better diffusion barrier performance of film deposited at 400 C is primarily because of higher film density. Figure 6 13 shows the pre and post anneal XRD measurements for films deposited by procedure 1A. For 300 C, the pre anneal depth profile shows five peaks at 43.50, 50.70, 61.75, 69.30 and 74.60 2 degrees. The peaks at 43.50, 50.70 and 69.30 degrees correspond to Cu(111), Cu(220) and Cu(220) phases whereas the peaks at 61.75 and 69.30 degrees corresponds respectively. The XRD profile also indicates that PVD copper depo sition on WN x C y shows a preferential orientation of (111). This is highly desirable because the Cu(111) phase has a better electromigration resistance as compared to other crystalline phases. The post anneal XRD profile shows an increase in the intensity of peaks corresponding to copper suggesting grain growth of Cu crystallites due to annealing. No peaks corresponding to Cu 3 Si are observed in the post anneal XRD profile even though the AES depth profile shows that copper has diffused through the barrier film. This apparent contradiction is due to the poor sensitivity of XRD in detecting small amount of Cu 3 Si crystallites. The pre and post anneal XRD spectra for barrier film deposited for 1A at 350 and 400 C show results similar to those of film depos ited at 300 C. No evidence of Cu 3 Si is seen in post anneal XRD spectra for barrier

PAGE 206

206 film deposited at 350 and 400 C suggesting that no bulk diffusion of Cu has occurred after annealing. 6.9.2 Barrier Testing for 1B Figure 6 14 shows the depth profile for pre and post anneal Cu/ WN x C y /Si stack s for WN x C y films deposited by procedures 1 B at 300, 350 and 40 0 C. The thickness of WN x C y film from deposition at 300, 350 and 400 C was 12, 23 and 45 nm respectively. Since the films deposited by 1B are much thi was anticipated to be worse. Etching off the Cu layer prior to AES depth profiling can eliminate off using 50 % HNO 3 solution before AES depth profiling. For deposition by procedure 1B at 300 C, the post anneal profile shows that Cu has diffused through the barrier layer into the Si. Comparison of the barrier/ Si interfaces for the pre and post anneal depth prof iles shows no evidence of intermixing at this interface. For deposition at 350 C, a trace amount of copper can be seen in the post anneal depth profile indicating that the barrier film was not effective in preventing copper diffusion. For deposition at 400 C, the post anneal AES depth profile shows that there is no copper diffusion through the barrier film suggesting that the barrier film was able to prevent Cu diffusion after annealing at 500 C. No intermixing at the barrier/ Si interface is observed Overall, for both 1A and 1B films, barrier films deposited at 300 C are not able to prevent Cu diffusion whereas barrier films deposited at 400 C are able to prevent Cu diffusion after annealing at 500 C. To identify the factor(s) that make film dep osited at 400 C effective as diffusion barriers, the sputter rate of films deposited at 300, 350 and 400 C by procedure 1B was calculated from the AES depth profile as discussed above (Table 6 6). The sputter rate for films deposited by 1B is expected t o be higher than that for films deposited for 1A because of lower W content in the films deposited for 1B. Unlike films deposited for 1A, films deposited

PAGE 207

207 for 1B show variation in W content for deposition at 300, 350 and 400 C, hence the sputter rates for these films cannot be directly correlated to film density because film composition influences sputter rates. As expected, the sputter rate for film deposited at 400 C is lower than that that for film deposited at 300 and 350 C because of the higher non metal (N+C+O) to W ratio for film deposited at 400 C. Hence, no conclusive evidence of effect of density on diffusion barrier performance could be obtained from sputter rates for films deposited for 1B. Another important factor that influences diffusio n barrier performance is the film composition. Film deposited at 400 C has significantly higher N content as compared to films deposited at 300 and 350 C. One of the reasons for the better diffusion barrier performance of film deposited at 400 C for 1 B could be its higher N content. Figure 6 15 shows the pre and post anneal XRD measurements for barrier films deposited by procedure 1B. Similar to Figure 6 13 for 1A, Figure 6 15 for 1B shows no evidence of copper diffusion for films deposited at 300, 3 50 and 400 C. The only peaks observed correspond to copper and Si, suggesting that no bulk diffusion of copper has occurred through the barrier film after annealing. Similar to films deposited by 1A, films deposited by 1B also show preferential depositi on of Cu(111) phase when copper is deposited by PVD. 6 10 Film Conformality As discussed in Chapter 2, the conformality of barrier film is of utmost importance for deposition in high aspect ratio features. Since the film deposited by procedure 1B at 400 C showed good diffusion barrier performance and high N content, conformality test was done for deposition at 400 C with only NH 3 as co reactant. Deposition was done on patterned chips provided by Samsung. The dielectric material in the test chip was FSG and the feature size was 0.2 m with aspect ratio of 2:1. Figure 6 16 shows the X SEM image of the feature after diffusion barrier deposition using 4 and only NH 3 co reactant. The film thickness for deposition

PAGE 208

208 at 400 C on FSG dielectric is 5 nm, which is significantly lower than the 45 nm thick films deposited on Si substrate at the same deposition temperature. This implies that nucleation on FSG substrate for 1B takes longer time than on Si substrate. The film is highly conformal visually, however, a precise number for conformality could not be calculated because of the insufficient resolution of the SEM images. 6 11 Conclusions It has been demonstrated that the mixed imido guanidinato complex W(N i Pr)Cl 3 [ i PrNC(NMe 2 )N i Pr] ( 4 ) can be used with NH 3 + H 2 (procedure 1A) and NH 3 + N 2 (procedure 1B) to deposit WN x C y thin films in an aerosol assisted CVD system. The use of NH 3 lowers the minimum growth temperature for 4 by 100 C. The use of NH 3 with or without H 2 as co reactant has significant effect on fi lm composition, growth rate, crystallinity, grain size and resistivity. Comparison of films deposited with procedures 1A, 1B and 1C shows that the best films in terms are deposited by 1B because of the lower resistivity of films at low deposition temperat ure, amorphous film deposition throughout the deposition temperature range and higher N content. Diffusion barrier testing shows that films deposited at 400 C for both 1A and 1B are able to prevent Cu diffusion after annealing at 500 C for 30 min in vac uum. Films deposited for 1B at 400 C also show good conformality in a 0.2 m 2:1 aspect ratio feature.

PAGE 209

209 Table 6 1. Molar flow rates used for deposition of WN x C y thin film from precursor 4 Reactant Molar flow rate (mol/min) 1A 1B 4 1.16 10 6 1.16 10 6 Benzonitrile 6.47 10 4 6.47 10 4 H 2 4.09 10 2 -NH 3 1.02 10 3 1.02 10 3 N 2 -4.09 10 2

PAGE 210

210 Table 6 2. Analysis results from deconvolution of XPS peaks of W for films deposited by procedure 1A on Si(100) substrate between 300 and 70 0 C. Deposition WN x C y WO x (2 < x < 2.5) WO 3 temperature Peak position % Area FWHM Peak position % Area FWHM Peak position % Area FWHM ( eV ) ( eV ) ( eV ) ( eV ) ( eV ) ( eV ) 300 W4f 7/2 32.12 10.50% 1.51 33.71 15.20% 1.78 35.75 26.90% 2.29 W4f 5/2 34.2 2 7.90% 1.70 35.81 11.40% 1.95 37.95 20.10% 2.23 W4p 3/2 37.52 1.60% 1.52 39.11 2.30% 2.20 41.15 4.00% 2.57 350 W4f 7/2 31.91 18.70% 1.46 33.79 17.90% 1.99 35.61 15.60% 2.01 W4f 5/2 33.91 14.00% 1.67 35.89 13.40% 2.21 37.75 11.70% 1.85 W4p 3/2 37.31 2.8 0% 2.60 39.19 3.60% 1.72 41.01 2.30% 2.26 450 W4f 7/2 32.05 33.90% 1.39 33.88 14.60% 1.97 35.36 3.30% 1.51 W4f 5/2 34.15 27.10% 1.68 35.98 10.90% 2.24 37.46 2.50% 1.71 W4p 3/2 37.45 5.10% 2.34 39.28 2.20% 1.67 40.76 0.50% 1.53 550 W4f 7/2 32.00 41.40% 1. 39 34.06 13.00% 1.97 ---W4f 5/2 34.20 31.10% 1.56 36.06 9.10% 2.44 ---W4p 3/2 37.40 4.10% 2.26 39.46 1.30% 2.49 ---650 W4f 7/2 31.99 32.60% 1.25 33.38 18.70% 2.05 ---W4f 5/2 34.19 23.50% 1.28 35.48 15.00% 2.02 ---W4p 3/2 3 7.39 6.50% 1.72 38.78 3.70% 1.96 ---750 W4f 7/2 31.98 33.10% 1.32 33.48 17.40% 1.84 ---W4f 5/2 34.27 24.80% 1.48 35.58 13.90% 2.05 ---W4p 3/2 37.38 8.30% 1.95 38.88 2.60% 1.50 ---

PAGE 211

211 Table 6 3. Analysis results from deconvolution of XPS peaks of C, N and O for films deposited by procedure 1A on Si(100) substrate between 300 and 700 C Deposition C 1s N1s O1s temperature Peak position % Area FWHM Peak position % Area FWHM Peak position % Area FWHM (eV) (eV) (eV) (eV) (eV) (eV) 300 Peak 1 283.59 58.30% 1.7 397.69 60.40% 1.64 530.9 64.00% 1.51 Peak 2 285.35 41.70% 2.22 399.97 39.60% 2.1 531.93 36.00% 2.12 350 Peak 1 283.43 59.50% 1.21 397.79 82.90% 1.56 530.65 64.80% 1.48 Peak 2 284.82 40.50% 1.42 399.77 17.10% 1.25 5 31.9 35.20% 1.71 400 Peak 1 283.54 73.30% 1.26 397.64 80.10% 1.58 530.65 64.80% 1.48 Peak 2 285 26.70% 1.64 399.39 19.90% 2.26 531.9 35.20% 1.71 500 Peak 1 283.48 73.20% 1.36 397.68 75.60% 1.63 530.61 63.40% 1.53 Peak 2 284.81 26.80% 1.67 399.75 24.4 0% 2.1 532.17 36.60% 1.98 600 Peak 1 283.43 55.50% 1.41 397.64 75.20% 1.69 530.63 62.60% 1.66 Peak 2 284.57 28.20% 1.61 399.42 24.80% 1.95 531.98 37.40% 2.05 Peak 3 286.19 16.30% 2.28 ------700 Peak 1 283.55 44.70% 1.63 397.73 79.30% 1.59 530.78 64.10% 1.62 Peak 2 284.6 28.50% 1.73 399.04 20.70% 2.1 532.06 35.90% 1.94 Peak 3 285.73 26.80% 2.3 ------

PAGE 212

212 Table 6 4 Analysis results from deconvolution of XPS peaks of W for films deposited by procedure 1B on Si(100) substrate b etween 300 and 700 C. Deposition WN x C y WO x (2 < x < 2.5) WO 3 temperature Peak position % Area FWHM Peak position % Area FWHM Peak position % Area FWHM (eV) (eV) (eV) (eV) (eV) (eV) 300 W4f 7/2 31.82 17.40% 1.5 33.44 20.20% 2.17 35.4 14.10% 2.0 4 W4f 5/2 33.92 13.10% 1.55 35.54 15.10% 2.17 37.55 11.30% 1.96 W4p 3/2 37.22 2.60% 1.52 38.84 4.00% 1.46 40.7 2.10% 1.91 350 W4f 7/2 32.01 36.50% 1.59 33.71 9.40% 1.6 35.2 7.00% 1.72 W4f 5/2 34.11 26.30% 1.51 35.81 7.10% 1.64 37.3 5.30% 1.88 W4p 3/2 3 7.41 5.50% 2.19 39.11 1.90% 1.22 40.6 1.10% 2.02 400 W4f 7/2 31.89 38.50% 1.47 33.56 9.80% 1.95 35.5 4.30% 1.57 W4f 5/2 34.04 28.90% 1.47 35.66 7.40% 1.94 37.6 3.20% 1.88 W4p 3/2 37.29 5.80% 2.31 38.96 1.50% 1.15 40.9 0.60% 1.8 500 W4f 7/2 31.91 37.00% 1 .28 33.41 15.70% 2.1 ---W4f 5/2 34.11 27.70% 1.4 35.51 11.80% 1.98 ---W4p 3/2 37.31 5.50% 1.68 38.81 2.40% 2.18 ---600 W4f 7/2 31.94 36.10% 1.29 33.39 16.50% 2.12 ---W4f 5/2 34.14 27.10% 1.36 35.49 12.40% 2.01 ---W4p 3/2 3 7.34 5.40% 1.84 38.79 2.50% 2.39 ---700 W4f 7/2 31.93 34.50% 1.22 33.24 18.20% 1.97 ---W4f 5/2 34.13 25.80% 1.26 35.34 13.60% 1.84 ---W4p 3/2 37.33 5.20% 1.81 38.64 2.70% 2.11 ---

PAGE 213

213 Table 6 5. Analysis results from deconvolution o f XPS peaks of C, N and O for films deposited by procedure 1B on Si(100) substrate between 300 and 700 C Deposition C 1s N1s O1s temperature Peak position % Area FWHM Peak position % Area FWHM Peak position % Area FWHM (eV) (eV) (eV) (eV) (eV) ( eV) 300 Peak 1 283.54 100.00% 1.53 397.55 71.50% 1.43 530.75 71.10% 1.5 Peak 2 ---398.96 28.50% 1.96 531.97 28.90% 1.84 350 Peak 1 283.46 100.00% 1.57 397.58 83.30% 1.5 530.57 67.40% 1.48 Peak 2 ---399.2 16.70% 1.52 531.85 32. 60% 1.93 400 Peak 1 283.39 85.70% 1.55 397.5 87.20% 1.58 530.46 70.20% 1.58 Peak 2 284.85 14.30% 1.62 399.25 12.80% 1.73 531.85 29.80% 1.96 500 Peak 1 283.32 64.60% 1.33 397.5 82.70% 1.48 530.39 64.20% 1.56 Peak 2 284.72 35.40 % 2.08 398.91 17.30% 1.58 531.64 35.80% 1.93 600 Peak 1 283.42 61.80% 1.5 397.49 69.00% 1.53 530.45 64.40% 1.54 Peak 2 284.5 25.80% 1.77 398.99 31.00% 2 531.89 35.60% 1.81 Peak 3 286 12.40% 1.5 ------700 Peak 1 283.33 45.70% 1 .29 397.54 70.30% 1.39 530.33 57.50% 1.55 Peak 2 284.42 31.40% 1.57 398.82 29.70% 1.56 531.66 42.50% 1.86 Peak 3 285.93 23.00% 2.46 ------

PAGE 214

214 Table 6 6. Apparent sputter rate of barrier film measured by dividing barrier film thickness by t he time required to sputter the film during AES depth profiling Deposition Temperature (C) 1A 1B 300 pre anneal 74 80 post anneal 82 68 350 pre anneal 93 74 post anneal 150 70 400 pre anneal 65 94 post anneal 60 88

PAGE 215

215 Figure 6 1. Composition of films deposited by procedures 1A, 1B and 1C on Si (100) substrate at different deposition temperature as determined by AES after 0.5 min of sputtering

PAGE 216

216 Figure 6 2. X ray diffraction patterns for films deposited on Si (100) substrate by procedure 1A

PAGE 217

217 Figure 6 3. X ray diffraction patterns for films deposited on Si (100) substrate by procedure 1B

PAGE 218

218 Figure 6 4. Lattice parameters for films grown by procedures 1A and 1C on Si (100) substrate. Error bars indicate uncertainty in determination of peak po sition for lattice parameter calculation.

PAGE 219

219 Figure 6 5. Average grain size at different deposition temperature for films grown by procedures 1A and 1B on Si (100) substrate.

PAGE 220

220 Figure 6 6 Scanning electron microscope images for films grown by procedure 1 A at different deposition temperature on Si (100) substrate

PAGE 221

221 Figure 6 7. Scanning electron microscope images for films grown by procedure 1B at different deposition temperature on Si (100) substrate

PAGE 222

222 Figure 6 8 Arrhenius plot for growth using procedu res 1A, 1B and 1C for growth on a Si(100) substrate

PAGE 223

223 Figure 6 9 X ray photoelectron spectroscopy measurement s for film s deposited by procedure 1A between 300 and 700 C on Si (100) substrate

PAGE 224

224 Figure 6 10. X ray photoelectron spectroscopy measure ment s for film s deposited for between 300 and 700 C by procedure 1B on Si (100) substrate

PAGE 225

225 Figure 6 1 1 Change in film resistivity with deposition temperature for films deposited by procedures 1A, 1B and 1C on Si (100) substrate

PAGE 226

226 Figure 6 1 2. Pre and post anneal AES depth profile s of Cu (100 nm)/ WN x C y /Si (substrate) stack s for WN x C y films deposited by procedure 1A. A) Barrier deposition at 30 0 C, no annealing. B) Barrier deposition at 300 C, annealing at 500 C. C) Barrier deposition at 35 0 C, no annealing. D) Barrier deposition at 350 C, annealing at 500 C. E) Barrier deposition at 40 0 C, no annealing. F) Barrier deposition at 400 C, annealing at 500 C.

PAGE 227

227 Figure 6 13. Pre and post anneal XRD measurements of Cu (100 nm)/ WN x C y / Si (substrate) stack s for WN x C y film deposited by procedure 1A at 30 0 350 and 400 C

PAGE 228

228 Figure 6 1 4. Pre and post anneal AES depth profile s of Cu (100 nm)/ WN x C y /Si (substrate) stack s for WN x C y film deposited by procedure 1B. After vacuum annealing and prior to AES depth profiling, the Cu layer in the Cu/ WN x C y /Si stack was etched off. A) Barrier deposition at 30 0 C, no annealing. B) Barrier deposition at 300 C, annealing at 500 C. C) Barrier deposition at 35 0 C, no annealing. D) Barrier depositi on at 350 C, annealing at 500 C. E) Barrier deposition at 40 0 C, no annealing. F) Barrier deposition at 400 C, annealing at 500 C.

PAGE 229

229 Figure 6 15. Pre and post anneal XRD measurements of Cu (100 nm)/ WN x C y /Si (substrate) stack for WN x C y film depo sited by procedure 1B at 30 0 350 and 400 C

PAGE 230

230 Figure 6 16. Scanning electron microscope images of WN x C y thin films deposited on FSG dielectric. The feature size is 0.2 m with aspect ratio of 2:1

PAGE 231

231 CHAPTER 7 DEPOSITION OF WN x C y FROM ( CH 3 CN)C l 4 W(NNM e 2 ) FOR DIFFUSION BARRIE R APPLICATION Chapters 4, 5 and 6 discussed two different approaches for increasing film N content. The first approach used NH 3 as a co reactant with precursor 3 and 4 to increase the film N content. The second approach used metal organic precursor 4 which contains more N atoms per W atom (Chapter 5). Following the same approach, (CH 3 CN)Cl 4 W(NNMe 2 ) ( 5 ) has been used to deposit WN x C y thin films for diffusion barrier application. Scheper et al. 204 have demonstrated that diorganohyd razido(2 ) titanium complexes can be used for the deposition of titanium nitride thin films by CVD This chapter discusses the characterization of the films deposited from 5 using CVD and evaluates the effectiveness of the deposited films in preventing di ffusion of copper. 7.1 Precursor Synthesis A Schlenk flask was charged with WCl 6 (2.1 0 g 5.30 mmol ) and methylene chloride (40 ml) 1,1 D imethylhydrazine (0.4 ml, 5.3 mmol) was added to the flask u nder vigorous stirring at 78 C. After 10 min, the solve nt was removed in vacuo while warming to room temperature. After removing the solvent, acetonitrile (15 mL) was added to the flask and the mixture was stirred for 30 min. The solvent was removed in vacuo and solids from the reaction were extracted with 2x1 5 mL of methylene chloride. The combined extracts were filtered and the volume reduced to 10 m L The product was precipitated from the mixture by adding vigorously stirred pentane (200 m L ) maintained at 0 C. The orange product 5 was filtered off as a micr ocrystal line powder and dried in vacuo 7.2 Film Growth by CVD When CVD growth was attempted at 400 C from 5 in an inert atmosphere with N 2 as carrier gas, the resulting film contained W and O while no C or N was detected by AES.

PAGE 232

232 However, when H 2 was use d as carrier gas at the same deposition temperature, the deposited film contained W, N, C and O as determined by AES. Hence, the H 2 co reactant is essential for deposition of WN x C y thin film from 5 Appreciable film growth was observed for 5 at depositio n temperature as low as 300 C when H 2 was used as co reactant This temperature is the lowest growth temperature among all the precursors 1 5 when only H 2 is used as co reactant. The film characterization and testing reported in this chapter were perf ormed on films deposited using 5 and H 2 as co reactant. Table 7 1 provides the molar flow rate of reactants used in the MOCVD reactor. The molar flow rate of 5 was same as that used for deposition using precursor 4 The color of the deposited film varie d from golden brown at low deposition temperature to grayish at higher deposition temperature. 7.3 Film Composition Figure 7 1 shows AES results for films deposited using 5 between 300 and 700 C The AES results showed that films deposited from 5 contain ed W, N, C and O in varying concentration for deposition between 300 and 650 C. T he films deposited between 300 and 650 C are W rich, with W content varying between 51 and 63 at. % The W content of film decreases to 34 at. % for deposition at 700 C b ecause of excessive incorporation of C in the film. The N content in the film is 11 at. % for deposition at 300 C. As the deposition temperature is increased from 300 to 350 C, the film N content increases dramatically to 24 at. %, which is also the hi ghest value of N content obtained for deposition from 5 throughout the deposition temperature range. For deposition between 350 and 750 C, the film N content decreases from 24 at. % to 5 at. %. The sources of N in the film are precursor 5 and/or benzoni trile. The decrease in N content between 350 and 650 C is due to a decrease in N incorporation from the precursor and benzonitrile. For deposition at 700 C, the AES measurement shows that the film N content is below the detection limit of AES. This in dicates

PAGE 233

233 that deposition temperature higher than 650 C leads to cleavage of N containing ligand from the precursor, resulting in deposition of WC x thin film. The C content of the film remains constant at ca. 9 at. % for deposition between 300 and 400 C. Between 400 and 600 C, the C content of film monotonically increases from 9 at. % to 41 at. % and remains unchanged at 650 C. The increased C content of the film is due to increased incorporation of C from the precursor and/or benzonitrile into the film with an increase in deposition temperature. 155 Increasing the deposition temperature further to 700 C causes a sharp increase in film C content to 63 at. %. The O content of the film decreases from 19 at. % at 300 C to 15 at. % at 350 C. In Chapter 6, it has been argued that the O in the film is incorporated during growth as well as post growth. The XPS results ( vide infra ) show that O in the film is incorporated both during growth and post growth. For growth between 450 and 600 C, the O content o f the film decreases from 9 at. % to 2 at. % and remains constant at 2 at. % between 600 and 700 C. Due to the presence of chlorine in 5 the possible presence of chlorine in the films was of interest. As discussed earlier, XPS was used to detect the pre sence of Cl in the film because the W NNN peak in AES spectra overlaps with the Cl LMM peak at ca. 180 eV. Figure 7 2 shows the XPS data for films deposited using 5 between 300 and 700 C. No peaks were observed for either Cl 2s or Cl 2p 3/2 at 270eV and 199 eV, respectively, confirming that the chlorine level in the films was lower than the detection limit of XPS (ca. 1 at. %). 7.4 Film Crystallinity Figure 7 3 shows the XRD patterns for films deposited from 5 between 300 and 70 0 C. The XRD spectra for films deposited between 300 and 50 0 C show no peaks attributable to the se

PAGE 234

234 spectra suggests that films deposited up to 500 C are X ray amorphous. The XRD pattern for the film deposited between 5 5 0 and 650 C shows the emergence of crystallinity as evidenced by the two broad peaks that lie between the standard peak position s W 2 ] W 2 (200) phase ], 178 WN x C y or a physical W 2 W 2 C. Since the int ensity of most the intense (111) reflection is small, the other peaks corresponding to (220) and (311) reflections at 63.55 and 75.05 are not clearly visible in the XRD spectra. For deposition at 700 C, the two peaks at 37.20 and 42.55 2 correspond t W 2 C because the films contain only W, C and O as indicated by AES measurement. 7.5 Lattice Parameter Figure 7 4 shows the change in lattice parameter with deposition temperature for films deposited from 5 WN x C y W 2 W 2 C has the highest intensity and thus was used for calculating the lattice parameter. For film deposited at 5 5 0 C, the lattice parameter has a value of 4.13 which is similar to the literature value of lattice parameter for W 2 N (4.126 ) As the deposi tion temperature is increased from 5 5 0 to 60 0 C, the lattice parameter value remains almost unchanged at 4.12 even though the C content of the film increases by 9 at. % in the same temperature range. If the additional C were incorporated in the W subla ttice, the lattice parameter value would have increased. An observation to the contrary suggests that the additional C is incorporated outside the W sublattice in amorphous state. Between 60 0 and 65 0 C, the lattice parameter increases from 4.1 2 to 4.1 6 In the same temperature range, the film composition remains relatively unchanged The increase in lattice parameter could be due to increase in interstitials (C, N and/or O) or a decrease in vacancies in the W sublattice. As deposition temperature is increased to 700 C, the lattice parameter

PAGE 235

235 increases further to 4.1 8 At 700 C, the film is W 2 C but the lattice parameter of 4.18 is W 2 C possibly due to strain in the film. 7.6 Grain Size As discussed earlier, XRD peak broadening could be used to determine average crystallite si ze for polycrystalline films using Since the (111) reflection was most intense, it was used to determine FWHM and estimate average grain size. Figure 7 5 shows the average crystallite size for films deposited from 5 T he average gra in size is 45 for deposition at 5 50 C. Between 550 and 650 C, the average grain size increases with an increase in deposition temperature. With an increase in deposition temperature, the surface diffusivity of adsorbed species increases resulting in an increase in grain growth. The average grain size W 2 WN x C y Overall, the average grain size measurement indicates that the films deposited from 5 at and above 5 5 0 C are nanocrysta lline with average grain size varying between 30 and 66 7.7 Film Growth Rate The growth rate for films deposited from 5 was determined by dividing the film thickness measured using X SEM by deposition time. Figure 7 wn at different deposition temperature from 5 The growth rate increased from 1 /min at 300 C to 4 for deposition at 40 0 C. Between 450 and 650 C, the growth rate gradually decreases from 13 /min to 10 /min. For deposition at 700 C, the gro wth rate increased significantly to 38 /min suggesting a change in growth mechanism at this temperature. Figure 7 7 shows the Arrhenius plot for films grown for deposition from 5 Film growth rate increased exponentially with temperature between 300 and 450 C indicating that the growth is in the kinetic ally controlled growth regime and that surface reacti on is the rate limiting step. Using the Arrhenius

PAGE 236

236 equation the apparent activation energy calculated for the kinetically controlled growth regime is 0. 5 2 eV. For film deposited between 4 50 and 6 50 C growth is in the diffusion limited growth regime Typically, growth in the diffusion limited growth regime shows a small positive slope. The small negative slope in the Arrhenius plot for growth from 5 suggests that some precursor is decomposed via homogenous reactions resulting in a decrease in growth rate in the diffusion limited growth regime. The transition from kinetically controlled to diffusion limited growth occurs between 450 and 500 C. 7.8 Atomic Bonding from XPS Figure 7 8 shows the deconvolution for the XPS peaks of W, N, C and O for films deposited from 5 between 300 and 700 C. The corresponding peak fitting parameters obtained from deconvolution are listed in Table 7 2 and Table 7 3 The reference peaks of elemental C, elemental W and different compounds of W, N, C and O are listed in Table 3 7 (Chapter 3). The deconvolution of the W 4f and 4p peaks at 3 00 C indicates the presence of three different binding states of W. The first W 4f 7/2 peak at 31.8 eV lies between the reference W peaks of W 2 C ( ca. 31.5 eV) and W 2 N ( ca. 33.0 eV), indicating that the W atom is bonded to both C and N. The second 4f 7/2 peak at 33.4 eV lies between the reference WO 2 peak at 32.9 eV and reference W 2 O 5 p eak at 34.6 eV, suggesting that the stoichiometry of tungsten oxide is WO x (2 < x < 2.5) The third W 4f 7/2 peak at 35.2 eV corresponds to W present in WO 3 These results are similar to those obtained for films deposited from 4 by procedures 1A, 1B and 1 C (Chapter 6). With a increase in deposition temperature, intensity of the W peak corresponding to WO x and WO 3 decreases, which is consistent with the AES results that shows a decrease in O content of the film in the same deposition temperature range. Fo r deposition at 700 C, the W peaks can be deconvoluted into WN x C y and WO x peaks without a contribution from WO 3 The N peak for deposition at 3 00 C can be deconvoluted into two peaks. The first N 1s peak at 397. 5 eV

PAGE 237

237 corresponds well with the reference N 1s peak for N in WN x film 119 The second peak at 398.8 eV suggests that N is also present in amorphous state. 119 Between 300 and 600 C, the N peak can be similarly deconvoluted to a dominant peak around 397.5 eV and a broad shallow peak around 398.5 e V implying that the bulk of N is bonded to W and a small amount of N present in amorphous phase. The film deposited at 700 C shows no N 1s peak confirming that the film is WC x Between 300 and 700 C, the C 1s peak can be deconvoluted into two peaks arou nd 283. 4 eV and 285.0 eV. The former peak corresponds to C bonded to W (reference peak at 283.5 eV) and the latter peak reflects C present in amorphous phase (reference peak between 284.5 and 285.2 eV) Between 300 and 500 C, the C 1s peak deconvolution shows that the C in the film is predominantly bonded to W with some C present in amorphous state. However, for deposition at 600 and 700 C, the ratio of amorphous C to carbidic C increases dramatically suggesting an increase in ligand and/ or benzonitri le decomposition and subsequent incorporation of amorphous C in the film. The O 1s peak for deposition between 300 and 7 00 C can be deconvoluted into two peaks at 530. 5 and 53 1.7 eV. The peak at 530. 5 eV corresponds to O bonded to W in WO x or WO 3 The o ther peak at 531.7 eV corresponds to O bonded to W in WN x C y As discussed earlier, O could be incorporated into the film either during film growth because of O and water vapor impurity in the reactor or after the film growth because of exposure of film to atmosphere. The presence of WO x /WO 3 suggests that residual gas (oxygen, water vapor) in reactor and/or O impurity in the precursor react with the precursor on the film surface during film growth. The source of O bound to W in WN x C y is most likely atmosp heric O that diffuses through the film when the film is exposed to atmosphere. With an increase in deposition temperature, the peak

PAGE 238

238 intensity of O decreases for both O present in WO x /WO 3 and O loosely bound to W in WN x C y The O present in WO x /WO 3 decreas es with an increase in deposition temperature because of the increase in film thickness with deposition temperature or densification of film. The peak intensity of O loosely bound to W in WN x C y decreases with increase in deposition temperature because of decreased O diffusion in the film from atmosphere as a result of denser and thicker films deposited at higher temperatures. In summary, deconvolution of 1s peak of O suggests that O is incorporated in the film during growth as WO x /WO 3 and post growth as O loosely bound to W in WN x C y XPS results indicate that film s deposited at 3 00 and 400 C are WN x C y with significant amount of WO x and WO 3 along with presence of small amounts of amorphous N and C For deposition at 500 and 6 00 C, the bulk of the film is WN x C y with a small amount of WO x and increased amount of amorphous C The film deposited at 700 C consists of W C x along with a small amount of WO x and a significantly higher amount of amorphous C 7.9 Film Resistivity Figure 7 9 shows the variation of f ilm resistivity with deposition temperature for films deposited from 5 cm are obtained for deposition at 300 C This value is considerably higher than bulk resistivities of pure W 2 N and W 2 C. Previous study o n ALD WN x C y has also shown that films with a low resistivity of 300 400 cm can be deposited at low deposition temperature (300 350 C). 145 A number of factors influence the resistivity of transition metal nitrides and carbides including metal to no n metal ratio, vacancies in metal and non metal sublattice, bonding states of C and N, presence of contaminants such as O, film crystallinity and porosity. The high film resistivity for film deposited from 5 could be because of interplay between some or a ll of these factors. As the deposition temperature is increased from 300 to 35 0 C, the film resistivity increases sharply to

PAGE 239

239 cm possibly due to the increase in N content of the film. Between 350 and 400 C, there is a sharp increase in film resistivity because of the decrease in W content of the film. Between 450 and 600 C, film resistivity gradua cm. The cm even though AES results indicate that the film is WC x which has low bulk resistivity. The film deposited at 700 C is C rich and high resisti vity of the film is believed to be due to the presence of excessive amorphous C in the film as indicated by the XPS spectrum. 7.10 Diffusion Barrier Testing To determine the effectiveness of diffusion barrier film deposited from 5 barrier films deposited at 350 and 4 00 C were coated with 100 nm PVD Cu. Prior to the deposition of Cu thin film, the barrier film was exposed to atmosphere for approximately 1 2 hour(s). The Cu/barrier/Si stack was annealed in vacuum at 500 C for 30 min. After annealing, three point AES depth profile and XRD measurements were done to detect copper diffusion through the barrier film While XRD measurements were done on the Cu/barrier/Si stack, the AES measurements were done on a barrier/Si stack obtained after removing the Cu layer by etching with dilute HNO 3 The Cu layer was removed to prevent the knock on effect of sputtering ( vide supra ). Figure 7 1 0 shows the depth profile of pre and post anneal Cu/WN x C y /Si stack s for WN x C y films deposited at 350 and 40 0 C. The thi ckness of the film deposited at 350 and 400 C was 50 and 60 nm respectively. For deposition at 350 C, the pre anneal depth profile (Figure 7 10 A) shows the background signal for Cu when there is no Cu diffusion. The pre annealed stack has a sharp barr ier/Si interface. The post anneal AES depth profile (Figure 7 10 B) shows that the Cu signal is similar to that observed in the pre anneal depth profile suggesting that the barrier film was able to prevent Cu diffusion into the Si. The barrier/Si interfa ce is sharp

PAGE 240

240 in the post anneal AES depth profile with no indication of mixing at this interface. The XRD measurement of the pre and post annealed film stack shown in Figure 7 11 shows peaks corresponding to Cu and Si with no indication of presence of Cu 3 Si. Thus, both AES and XRD measurements indicate that 50 nm barrier film deposited at 350 C was able to prevent Cu diffusion after annealing at 500 C in vacuum for 30 min. Furthermore, the XRD profile shows that PVD copper deposition on WN x C y has a pre ferred orientation of (111). This is highly desirable because the Cu(111) phase has a better electromigration resistance as compared to other crystalline phases of Cu. For deposition at 400 C, the comparison of pre and post anneal depth profile (Figures 7 10 C and D) shows that copper has not diffused through the barrier film after annealing. The post anneal depth profile shows a sharp barrier/Si interface indicating that no intermixing has occurred at this interface. The XRD profile shown in Figure 7 11 also indicates that the post anneal film stack exhibits no peaks corresponding to Cu 3 Si. Thus, barrier film deposited at 400 C is able to prevent Cu diffusion after annealing at 500 C in vacuum. As with film deposited at 350 C, WN x C y film deposited at 400 C also favors deposition of (111) oriented Cu film for deposition by PVD. 7 11 Conclusions It has been demonstrated that the tungsten hydrazido complex (CH 3 CN)C l 4 W(NNMe 2 ) ( 5 ) can be used with H 2 in an aerosol assisted CVD system to deposit WN x C y th in films. The H 2 co reactant is essential for deposition of WN x C y films as deposition with 5 in inert atmosphere (N 2 carrier gas) resulted in deposition of WO x films. The lowest growth temperature for 5 was 300 C, which is the lowest among precursors 1 5 Films deposited with 5 consisted of W, N, C and O as determined by AES and no Cl impurity was detected by XPS. The film N content was significantly higher for films deposited from 5 as compared to films deposited from 1 4

PAGE 241

241 Amorphous film depositi on was observed for deposition below 550 C. Films were quite resistive as compared to bulk resistivity of pure W 2 N and W 2 C. Diffusion barrier testing shows that films deposited at 350 and 400 C were able to prevent bulk Cu diffusion after annealing at 500 C in vacuum for 30 min.

PAGE 242

242 Table 7 1. Molar flow rates of reactants in CVD reactor for deposition from precursor 5 Reactant Molar flow rate (mol/min) Precursor ( 5 ) 1.16 10 6 H 2 4.09 10 2 Benzonitrile (solvent) 6.47 10 4

PAGE 243

243 Table 7 2. Analysi s results from deconvolution of XPS peak of W for films deposited from 5 on Si(100) substrate between 300 and 700 C. Deposition temperature WN x C y WO x (2 < x < 2.5) WO 3 Peak position % Area FWHM Peak position % Area FWHM Peak position % Area FWHM (e V) (eV) (eV) (eV) (eV) (eV) 300 W4f 7/2 31.76 27.70% 1.49 33.36 14.10% 1.53 35.18 10.80% 1.96 W4f 5/2 33.86 20.80% 1.64 35.51 10.60% 1.8 37.28 8.10% 2.09 W4p 3/2 37.16 4.20% 2.49 38.76 2.10% 1.26 40.58 1.60% 1.82 400 W4f 7/2 31.67 35.10% 1.49 33.63 13 .90% 2.74 35.09 3.60% 1.55 W4f 5/2 33.77 26.30% 1.51 35.73 10.40% 2.35 37.19 2.70% 1.62 W4p 3/2 37.07 5.30% 3.31 39.03 2.10% 1.7 40.49 0.50% 1.69 500 W4f 7/2 31.6 41.50% 1.38 33.4 7.50% 1.93 34.87 5.30% 1.52 W4f 5/2 33.7 31.10% 1.36 35.5 5.70% 1.88 36.9 7 4.00% 2.32 W4p 3/2 37 4.10% 2.78 38.8 0.80% 1.32 40.27 0.50% 1.5 600 W4f 7/2 31.92 42.00% 1.35 34.18 5.70% 1.71 35.4 5.20% 1.3 W4f 5/2 33.98 33.60% 1.39 36.28 4.30% 2.09 37.5 3.90% 3.33 W4p 3/2 37.32 4.20% 2 39.58 0.60% 1.19 40.8 0.50% 1.5 700 W4f 7/2 31.76 39.50% 1.24 33.4 12.50% 2.38 ---W4f 5/2 33.92 29.60% 1.34 35.5 9.40% 2.05 ---W4p 3/2 37.16 7.10% 2.22 38.8 1.90% 1.68 ---

PAGE 244

244 Table 7 3 Analysis results from deconvolution of XPS peaks of C, N and O for films deposited from 5 on S i(100) substrate between 300 and 700 C. Deposition C 1s N1s O1s temperature Peak position % Area FWHM Peak position % Area FWHM Peak position % Area FWHM (eV) (eV) (eV) (eV) (eV) (eV) 300 Peak 1 283.44 77.40 1.78 397.50 79.40 1.40 530.63 73.00 1.43 Peak 2 285.16 22.60 2.10 398.75 20.60 1.73 531.72 27.00 1.65 400 Peak 1 283.35 75.80 1.20 397.50 77.70 1.38 530.54 65.70 1.32 Peak 2 284.81 24.20 1.68 398.79 22.30 2.10 531.62 34.30 1.55 500 Peak 1 283.33 84.30 1.35 397.52 79.50 1.29 530.49 68.1 0 1.34 Peak 2 285.09 15.70 1.54 398.89 20.50 1.87 531.61 31.90 1.54 600 Peak 1 283.45 57.80 1.43 397.62 83.20 1.64 530.45 59.20 1.45 Peak 2 284.76 42.20 2.49 399.28 16.80 2.10 531.77 40.80 1.81 700 Peak 1 283.47 73.30 1.98 ---530.57 63.30 1.66 Peak 2 285.11 26.70 2.70 ---531.76 36.70 1.54

PAGE 245

245 Figure 7 1. Composition of films deposited from 5 on Si (100) substrate at different deposition temperature as determined by AES after 0.5 min of sputtering

PAGE 246

246 Figure 7 2. X ray photoelectron sp ectroscopy measurements for films deposited from 5 on Si (100) substrate at different deposition temperature

PAGE 247

247 Figure 7 3 X ray diffraction patterns for films deposited from 5 between 300 and 700 C on Si (100) substrate

PAGE 248

248 Figure 7 4. Lattice parameter s for films grown from 5 on Si (100) substrate. Error bars indicate uncertainty in determination of peak position for lattice parameter calculation.

PAGE 249

249 Figure 7 5. Average grain size for films grown from 5 at different deposition temperature on Si (100) s ubstrate.

PAGE 250

250 Figure 7 6 Scanning electron microscope images for films deposited from 5 between 300 and 700 C on Si (100) substrate

PAGE 251

251 Figure 7

PAGE 252

252 Figure 7 7 Arrhenius plot for deposition from 5 on a Si(100) substrate

PAGE 253

253 Figure 7 8 X ra y photoelectron spectroscopy measurement s for film s deposited from 5 between 300 and 700 C on Si (100) substrate

PAGE 254

254 Figure 7 9 Change in film resistivity with deposition temperature for films deposited from 5 on Si (100) substrate

PAGE 255

255 Figure 7 1 0. Pr e and post anneal AES depth profile of Cu/ WN x C y /Si (substrate) stack for WN x C y film deposited from 5 After vacuum annealing and prior to AES depth profiling, the Cu layer in the Cu/ WN x C y /Si stack was removed by etching. A) Barrier deposition at 35 0 C no annealing. B) Barrier deposition at 350 C, annealing at 500 C. C) Barrier deposition at 40 0 C, no annealing. D) Barrier deposition at 400 C, annealing at 500 C.

PAGE 256

256 Figure 7 11. Pre and post anneal XRD measurement of Cu (100 nm)/ WN x C y /Si (su bstrate) stack for WN x C y film deposited from 5 at 35 0 and 400 C

PAGE 257

257 CHAPTER 8 QUANTUM MECHANICAL C ALCULATIONS FOR PRED ICTING PRECURSOR FRAGMENTATION Chemical vapor deposition is a very complex process to model because it is a not an equilibrium process. Moreover, the numerous reaction pathways accessible to the precursor molecule both in gas phase as well as on the substrate surface make fundamental understanding of CVD processes all the more challenging. Computational chemistry can be used to study pos sible reaction pathway(s) for metal organic precursors. In this chapter, homogenous decomposition of precursor 5 has been investigated using density functional theory (DFT) and statistical mechanics. Based on the calculation results, possible reaction pa thways for precursor 5 have been proposed. 8.1 Computational Methodology A brief description of density functional theory is presented in Appendix A. All the calculations were done using Gaussian 03W software. For calculating binding energy of transition metal compounds such as hydrides, the Hartree Fock theory is not very accurate until quite high levels of theory are used to account for the electron correlation. In contrast, DFT has been shown to predict geometries and binding energies in transition me tal compounds more accurately. 205 The B3LYP functional, which has been quite accurate in predicting structure and energy of transition metal compounds, was used for geometry optimization and energy calculations. The calculations were done using a split b asis set (LanL2DZ for W and 6 31G(d) for all other elements). To reduce the computational time, the effective core potential LanL2DZ was used for W. 206 The molecular structures were visualized using Molden and gOpenMol. 207,208 Thermodynamic properties w ere calculated using the statistical mechanics package in the Gaussian 03W software.

PAGE 258

258 8.2 Geometry Optimization Figure 8.1 shows the structure of 5 obtained from X ray crystallographic determination and the computationally optimized geometry. As expected, the experimental and calculated geometries of 5 are quite similar. Table 8.1 shows the comparison of selected bond lengths and bond angles obtained from X ray crystallographic measurement and computational geometry optimization. The calculated structure of 5 shows octahedral coordination at the W center. Comparison of experimentally obtained and calculated bond lengths listed in Table 8 1 show that except for the W N1 bond length, the bond lengths obtained from calculation are larger than those obtained experimentally. The DFT calculations are done for gas phase species whereas the experimentally obtained structure is for solid precursor. The differences in bond angles could arise because of crystal packing forces that are not accounted for in gas phase DFT calculations. 8.3 Acetonitrile Cleavage Previous study on 1a,b 2a,b and 3a,b has shown that dissociation of the acetonitrile ligand from the precursors is facile at growth temperature (> 450 C). 209 To study the cleavage of acetonitrile ligand from precursor 5 and were calculated for 5 CH 3 CN and Cl 4 W(NNMe 2 ) ( 5 a ). The and values for cleavage of acetonitrile from 5 were then calculated using Eq uations 8 1 and 8 2. (8 1) (8 2) Figure 8 2 shows the optimized structures for 5a and CH 3 CN and Table 8 2 lists selected bond lengths and bond angles for 5a The calculated values for and can be assumed to because the transition states for endothermic reactions

PAGE 259

259 should be product like. The values for and for acetonitrile cleavage are 6.68 and 14.92 kcal/mol respect ively. These values are slightly higher that those obtained for 1a 3a reported previously, 209 suggesting that the acetonitrile cleavage reaction for precursor 5 has a higher energy barrier as compared to the other precursors. 8.4 Chlorine Cleavage One o f important findings from film growth characterization discussed in Chapter 7 is that even though precursor 5 contains four Cl atoms per W atom, the film deposited using 5 showed no evidence of Cl in the XPS measurement indicating that Cl incorporation in the film is below the detection limit of XPS (ca. 1 at. %). Hence, possible reaction pathways that can lead to Cl cleavage have been explored computationally. It should be noted that these calculations are done for homogenous gas phase reactions and the thermodynamic properties are calculated at 298.15 K and 1 atm. 8.4.1 Reaction of 5a with H 2 Two mechanistic pathways for the removal of Cl from 5a have been explored. The first pathway involves loss of Cl 2 from 5a value for reductive elimination of Cl 2 from 5a is 95.92 kcal/mol (Figure 8 3). The magnitude of this value suggests that the reductive elimination pathway is not accessible to 5a It should also be noted that the calcu RI 1 ) indicated that the triplet was more stable than the singlet, a possible indication of spin crossover problems with reductive elimination. The second pathway involved reaction of transition metal complex with H 2 Possible pathways for reaction of H 2 with transition metals include oxidative addition, coordination of H 2 bond metathesis. However, neither oxidative addition nor coordination of H 2 is ava ilable for a d 0 complex such as 5

PAGE 260

260 bond metathesis of H 2 with the Cl 4 W(N i Pr) 209 2 with 5a was 35.55 kcal/mol, indicating that thi s pathway should be accessible under the growth conditions. The calculated for 5a is very similar to the values reported previously for Cl 4 W(N i Pr) 209 8.4.2 Reaction Pathways for Removal of a Chloride from MI 1 Figure 8 4 shows reaction pathways for r bond metathesis of MI 1 to replace a second chloride ligand with hydride. For reductive elimination of HCl from MI is 37.26 kcal/mol. This value is quite low as compared to those previously reported for calc ulations on 2a 209 bond metathesis, the reaction can proceed to replace the chloride atom at either a cis or the trans position. The search for transition states for both cis and trans showed that th e values of activation energy for both these reaction pathways are quite similar. The formation of cis isomer MT 2 cis is slightly more favorable energetically as compared to the trans isomer MT 2 trans values for reductive elimination bond metathesis for both cis and trans configurations are quite similar. This result is different from that for 2a which showed that the bond metathesis for second chloride cle avage. 8.4.3 Reaction Pathways for Removal of Chloride from MI 2 and RI 2 Since the second chloride reaction discussed above could form MI 2 cis MI 2 trans or RI 2, possible reaction pathways for these intermediates have been explored. Figure 8 5 shows t he bond metathesis pathways for MI 2 cis and trans isomers. Figure 8 6 shows the reductive elimination of HCl from RI 2. Similar to RI 2, the RI 3 structure value f or reductive elimination of HCl from RI 2 was 43.07 kcal/mol. To get a better idea of reaction surface for pathways involving MI 2

PAGE 261

261 cis MI 2 trans and RI 2, the reactions that lead to the fourth chloride cleavage needs to be studied. 8.4.4 Reaction Pathwa ys for Removal of the Fourth Chloride from 5a Possible reaction pathways for MI 3 and RI 3 are shown in Figure 8 7. The reductive elimination of HCl from MI 3 has an energy barrier of 74.64 kcal/mol bond metathesis from MI bond metathesis reaction for RI value of 48.26 kcal/mol. bond metathesis and reductive el imination are quite similar ( vide supra ), a comparative study of the energy barriers for third and fourth chloride cleavage reactions is required to ascertain which of these pathways would be more favorable if the reaction were to proceed by sequentially r emoving all chlorides. Figures 8 8, 8 9 and 8 10 shows the 5a ) for different reaction pathways for intermediates MI2 cis MI2 trans 5a ) allows the comparison of different reaction pathways corresponding to MI2 cis MI2 trans and RT2 inte rmediates. The comparison of different reaction pathways passing through intermediates MI2 cis MI2 trans and RI2 shows that the removal of second chloride at the trans bond metathesis to form MI2 trans intermediate is most favorable energet ically, followed by between the energy values for third and fourth chloride cleavage for reaction pathways through intermediates MI2 cis and MI2 trans is small (ca. 5 kcal/mol), hence it is possible that reaction pathway involving intermediate MI2 cis would also be accessible to 5a if the reaction were to proceed via sequential elimination of all four chlorines in the gas phase. The above mentioned calculation gives relative comparison of different pathways that can lead to sequential removal of all Cl from the precursor via either

PAGE 262

262 elimination. Realistically, the values of activation energy for the removal of third and fourth Cl atoms from 5a are high enough to preclude the availability of these pathways under the reaction conditions (deposition temper ature of 300 C or higher). 8.5 Conclusion Computational chemistry has been used to assess the reaction pathways for precursor 5 The facile dissociation of acetonitrile ligand from 5 which has observed experimentally for reaction in solution, is also ob served for calculations done for gas phase reaction. Various reaction pathways for Cl cleavage from 5a have also been explored. For the first Cl cleavage from 5a bond metathesis is energetically favored and the activation energy value suggests that this pathway would be accessible to 5a bond metathesis and reductive elimination pathways have similar energy barriers. However the high activation bond metathesis and reductive elimination pathways indicates that these pathways may or may not be accessible to intermediate MI 1. Because of very high energy barriers fo r the third and fourth Cl cleavages, these pathways would not be accessible to 5a in the gas phase While the present work focuses on homogenous reactions for acetonitrile dissociation from 5 and sequential Cl cleavage from 5a further work is needed to un derstand heterogeneous reaction pathways of 5 5a and other reaction intermediates. The investigation of heterogeneous reactions requires computationally intensive calculations and is a topic for future research.

PAGE 263

263 Table 8 1. Selected bond lengths () and bond angles () for precursor 5 obtained from X ray crystallographic measurement and DFT calculation Bond Calculated Experimental W1 Cl a 2.388 2.347(16) W1 N1 1.749 1.769(5) N1 N2 1.289 1.271(8) W1 N3 2.268 2.224(7) N 2 C 1 1.467 1.438(7) N 1 W 1 Cl a 96.8 95.9 (4) N3 W1 Cl a 83.24 84.07(4) W 1 N1 N2 178.0 180.0 a Average value for four equivalent chlorides. Table 8 2. Selected bond lengths () and bond angles () for 5 a obtained from DFT calculation Bond Calculated W 1 Cl a 2.3 5 1 W 1 N1 1.7 34 N1 N2 1.2 93 Cl W 1 N1 a 102.1 W 1 N1 N2 17 7.5 a Average value for four equivalent chlorides.

PAGE 264

264 Figure 8 1. Geometry of 5 A) X ray crystallographic determination. The thermal ellipsoids diagram are drawn at 50 % probability Hydrogen atoms are omitted for clarity. B) Geometry optimization by DFT calculation Figure 8 2. Cleavage of acetonitrile ligand from precursor 5 The figure shows optimized geometries for 5a and CH 3 CN

PAGE 265

265 Figure 8 3. Calculated transition state and bond metathesis from 5a M = metathesis, R = reductive elimination, T = transition state and I = reaction intermediate

PAGE 266

266 Figure 8 4 bond metathesis from MI 1 intermediate. M = metathesis, R = reductive elimination, T = transition state and I = reaction intermediate

PAGE 267

267 Figure 8 bond metathesis from MI 2 ( cis and trans ). M = metathesis, R = reductive elimination, T = transition state and I = reaction intermediate

PAGE 268

268 Figure 8 6. Possible reaction pathway for reductive elimination from RI 2. R = reductive elimination, T = transition state and I = reaction intermediate

PAGE 269

269 Figur e 8 bond metathesis from MI 3 and RI 3. M = metathesis, R = reductive elimination, T = transition state and I = reaction intermediate

PAGE 270

27 0 Figures 8 ( 5a ) for reaction pathway through intermediate MI2 cis The energy values are in kcal/mol

PAGE 271

271 Figures 8 5a ) for reaction pathway through intermediate MI2 trans. The energy values are in kcal/mol

PAGE 272

272 Figures 8 10. 5a ) for reaction pathway through intermediate RI2 The energy values are in kcal/mol

PAGE 273

273 CHAPTER 9 CONCLUSION AND FUTUR E WORK The films deposited using 3a,b 4 and 5 show many properties that are essential for diffus ion barrier application. First and foremost, the work demonstrates that WN x C y can be effective in preventing Cu diffusion. Films deposited at low temperature from 4 and 5a,b are able to prevent bulk Cu deposition after annealing at 500 C for 30 min in v acuum. Secondly, precursor 5 could be used to deposit films at temperatures as low as 300 C indicating that MOCVD is capable of depositing films at temperature acceptable for diffusion barrier application. Third, the films showed good adhesion and were thermally and mechanically stable after annealing upto 500 C. Fourth, WN x C y film promotes the growth of (111) oriented PVD Cu, which is important to obtain Cu films with high electromigration resistance. This chapter discusses the additional experiments needed to evaluate the potential of WN x C y deposition from metal organic precursors for diffusion barrier application. 9.1 Diffusion Barrier Testing The diffusion barrier testing reported in the present work gives preliminary results on the effectiveness of barrier film in preventing bulk Cu diffusion. However, these tests need further refinement in three critical areas: film thickness control, contamination control and trace copper detection under realistic testing conditions. 9.1.1 Film Thickness Control The barrier films tested in present work used relatively thick films (20 80 nm) for evaluating their efficacy as diffusion barriers. Future copper metallization would require barrier films that are thinner than 10 nm. Atomic layer deposition shows the most promise for deposition of ultra thin barrier films with excellent conformality. Since 4 and 5 have shown promising results for films deposited by MOCVD, these precursors should be used in ALD mode

PAGE 274

274 to meet the thickness and conformality requirements for future barrier films. Actual barrier testing on these ultra thin films is essential for providing accurate evaluation of the barrier material as well as the deposition process. There are a number of challenges that need to be addressed to deposit ALD films using precursors 4 and 5 Since precursors 4 and 5 are solids with low vapor pressure, their delivery in the reactor requires dissolution of the precursor in a suitable solvent such as benzonitrile. For MOCVD growth, the precursor solvent mixture w as delivered to the reactor using a nebulizer system. The ALD system needs to be modified so that it can use the nebulizer system for delivery of the precursor solution. For ALD, the nebulizer system should be capable of withstanding 1 Torr operating pre ssure. The nebulizer system used in present work requires relatively high pressure of 350 Torr. However such high pressures are not suitable for ALD, so the nebulizer system also needs to be redesigned so that it is capable of operating at lower pressure s (~ 1 Torr). 9.1.2 Contamination Control AES studies done on films deposited from 4 and 5 have shown that films deposited at low deposition temperature (< 500 C) contain a significant amount of O contamination. In the presence of a high level of contami nation, the diffusion barrier performance of the film could be significantly altered. The XPS studies have shown that the source of O is not only the residual O in the reactor, but also post growth O incorporated during exposure of the film to the atmosph ere. To reduce the O incorporated during film growth, it is essential to grow film in a clean system. This might be possible in the new ALD reactor system, which could be purged and kept under inert atmosphere so that the reactor is not exposed to atmosp here after every experiment. The transfer chamber attached to the ALD reactor could then be used to load the samples into the ALD system. To reduce the O incorporated due to post growth exposure to

PAGE 275

275 atmosphere, the barrier films need to be coated with Cu in situ In situ Cu deposition would eliminate or at least minimize the O contamination from atmosphere. The CVD reactor has been connected to the ALD reactor through the transfer chamber so that in situ deposition of Cu is now possible. Future experime nts should be performed using this new system to ascertain the 9.1.3 Realistic Diffusion Barrier Testing The diffusion barrier testing reported in present work accelerated the diffus ion process by annealing the Cu/barrier/Si stack in vacuum at 500 C. However, the driving force for diffusion in actual ICs is not just temperature but also bias. Hence, a more realistic test to access the diffusion barrier efficacy would be bias temper ature stress. To detect Cu diffusion, the present work used AES depth profiling and XRD measurements. These techniques are good for detecting bulk Cu diffusion but not trace Cu diffusion. The detection of Cu should be done using electrical characterizat ion such as p + n junction diode because electrical characterization is able to detect trace Cu diffusion through the barrier film. 9.2 Diffusion Barrier Integration At present, numerous processes and materials have shown good promise for fulfilling the futu re requirements of diffusion barrier thin film. However, significant work needs to be done on integration of these barrier films into actual interconnect. While a barrier film might be excellent in preventing Cu diffusion, there are additional requiremen ts on the barrier film and deposition process. The barrier film needs to have good adhesion to adjacent layers and that include Cu, low k dielectric and etch stop layers. While the present work reported film growth only on Si substrate, additional work n eeds to be done to determine adhesion of the WN x C y film to Cu, low k dielectric and etch stop layers. The four point bend test is an excellent technique to ascertain the adhesion of thin films quantitatively. These tests need to be performed on the

PAGE 276

276 barri er film deposited on different substrates. The barrier deposition process also needs to be compatible with the overall metallization deposition scheme. This includes low deposition temperature and pin hole free film deposition. Lastly, the barrier depos ition process which employs gaseous reactants should be compatible with ultra low k dielectrics. Since ultra low k dielectric films are highly porous, the gaseous reactants used in CVD and ALD process could diffuse through the dielectric film, thereby incr easing its the dielectric constant and negating any benefits obtained from adoption of ultra low k dielectric film. The integration of CVD or ALD with ultra low k dielectric films is a challenge for future metallization schemes and pore sealing techniques for ultra low k films like plasma exposure need to be explored.

PAGE 277

277 APPENDIX DENSITY FUNCTIONAL T HEORY The energy of a multi particle system, E, is given by the Schrdinger equation (Eq. A 1). (A 1) amiltonian operator. The Hamiltonian operator in the Schrdinger equation depends on the positions and atomic number of the nuclei and the total number of electrons. The dependence of the Hamiltonian on the total number of electrons suggests that a usefu all space, it gives the total number of electrons N. (A 2) The assignment of nuclear atomic numbers is also available from the density, since for each nucleus A located at an electron density maximum r A (A 3) Given a known density, one could form the Hamiltonian operator, solve the Schrdinger equation, and determine the wave functions and energy eigenvalues. Hohenb erg and Kohn proposed two theorems that revolutionized the DFT methodology. 210 Hohenberg Kohn existence theorem: the non degenerate ground state density must determine the external potential (the charges and positions of the nuclei), and thus the Hamiltoni an and the wave function. Hohenberg Kohn Variational Theorem: the density obeys the variational principle. Kohn Sham self consistent field m ethodology takes as a starting point a fictitious system of non interacting electrons that have for their overall ground state density the same density as some real system of interest where electrons do interact. Since the electron density determines

PAGE 278

278 the position and atomic numbers of nuclei, these quantities are necessarily identical in the non interacting and in th e real systems. The energy functional is then divided into specific components to facilitate further analysis. ni ne ee ee (A 4) T ni is the kinetic energy of non interacting electrons, V ne is the nucle ar electron interaction, V ee is electron ee includes all non classical corrections to electron electron repulsion energy. The above equa tion can be rewritten as Equation A 5. (A 5) N is the number of electrons in the system and E xc is the exchange correlation energy (a ee ). The density is given by Equation A 6. (A 6) 7 can be used to calculate E. (A 7) The Kohn Sham one electron operator is defined by Equation A 8. (A 8) (A 9) V xc is the so called functional derivative. For the determination of Kohn Sham (KS) orbitals, a basis set is used, similar to the Hartree Fock method. The KS orbitals are expressed within a

PAGE 279

279 i }. The individual orbital coefficients are determined by solution of a secular equation (as in the Hartree Fock method). spin densities at any position are expressed i (A 10) A.1 Exchange C orrelation F unctional (E xc ) The exchange correlation functional accounts for: coulombic correlation energy associated with electron electron repulsion quant um mechanical exchange which arises from antisymmetry requirement correction for classical self interaction energy difference in kinetic energy between a fictitious non interacting system and a real system. In practice, most modern functionals do not tr y to compute (4) explicitly. The effect of (4) correction. A.1.1 Exchange Energy Density Hohenberg and Kohn showed that if the electron density varies extremely s lowly with position, then E xc 11. (A 11) xc x c ) is the exchange plus correlation energy per electron in a homogenous electron gas. Slater proposed that the exchange hole ab out any position could be approximated as a sphere of constant potential with a radius depending on the magnitude of the density at that position. With x is determined by Equation A 12. (A 12)

PAGE 280

280 The The LDA, Slater and X methods can be extended to account for spin polarization using the equa tion A 13. (A 13) is the exchange energy for system of uniform electron gas with electrons of like spin and is the exchange energy calculated using equation derived by Sl ater. Systems that have spin polarization (e.g. open shell systems) must use the spin polarized formalism and it is generally A.1.2 Correlation Energy Density Even for a simple uniform electron ga s, no analytical derivation has been reported. However, by using Monte Carlo techniques, the total energy of uniform electron gases of several different densities have been calculated to a very high accuracy. 211 By subtracting the exchange energy density for each case, the correlation energy density can then be calculated. The functional is given by Equation A 15. (A 14) (A 15) A, x 0 b, c are different empirical constants for i = 0 and i = 1. Vosko et al. proposed several different fitting schemes, and two functional forms (VWN and VWN5) have been widely

PAGE 281

281 used. 212 LSDA calculations that employ a combination of Slater exchange and the VWN correlation energy expression are referre d to as using SVWN method. A.2 Density Gradient Correction The LSDA approximation assumes a uniform electron gas as the source of energy expression. It is assumed that the exchange correlation energy density at every position in space for the molecule is the same as it would be for the uniform electron gas having the same density as is found at that position. However, in a molecule, this approximation is not true. A plausible way to increase the accuracy of the correlation functional for molecular calcu lations is to make it depend not only on the local value of density but also on the gradient of density. This approach correction term is added to the LDA functional (A 16) The subscript x/c indicates that the same functional holds for either exchange or correlation. A.2.1 C orrection to E xchange F unctional: The most popular GGA exchange functional has been developed by Becke (usually 213 by fitting to the exactly known exchange energies of the six noble gas atoms He through Rn. Several other exchange functionals that have been reported are CAM( B), FT9 and mPW. 214 216 There is very little data available in literature on the comparison of these different functionals for actual molecules. A.2.2 C orre ction to Correlation Functional Various functionals developed by researchers for GGA correction to the correlation functional include PW86, PW91, B95 and LYP. 217 220 The PW86 functional includes one

PAGE 282

282 empirical parameter fitted for the neon gas atom. The PW91 and B95 functionals use a different expression from the general expression described above and do not contain any empirical parameter. LYP does not correct the LDA expression but computes the full correlation energy. It contains four empirical parameters fitted to the He atom. Of all the correlation functionals, LYP is the only one that provides an exact cancellation of the self interaction error in one electron systems. The LYP functional has proved more robust for systems that are more prone to over delocalization because of its effective cancellation of self interaction error. The common nomen clature of expressing the functionals used in calculation is combining the acronyms for correlation and exchange functionals. For instance, BLYP calculation In liter ature, the most widely used combinations are BP86, B3LYP and BPW91. Results from these three levels of theory are fairly comparable. A.3 Similarity B etween Hartree Fock ( HF ) and DFT M ethodologie s The two main similarities between HF and DFT techniques ar e: Mathematically, both use basis sets to obtain orbitals The kinetic energy and nuclear attraction components of matrix elements are identical. There is a key difference between HF and DFT methodologies. Density functional theory contains no approximati ons, it is exact. It is only required to know E xc as a the exact nature of E xc is not known, the relevant equations are solved approximately. In contrast, HF is a deliberately approximate theory. A.4 Comparison of DFT to Molecular Orbital Theory 205 A.4.1 Computational E fficiency The form al scaling of DFT has been noted to be in principle no worse than N 3 where N is the number of basis functions used to represent KS (Kohn Sham) orbitals. This is better than HF

PAGE 283

283 which scales by a factor of N 4 The scaling factor refers to how computationa l time increases with the number of basis functions. For programs that use the same algorithms to carry out HF and DFT calculations, the cost of a DFT calculation on a moderately sized molecule (~ 15 atoms) is double that of the HF calculations with the s ame basis set. However, if the program is optimized for DFT, the calculation cost is much lower for DFT as compared to HF method. A.4.2 Energetics For a given average level of accuracy, hybrid DFT methods are the most efficient in predicting atomization e nergies as compared to more expensive methods. Of the currently available DFT models, GGA models offer a major improvement over the older LSDA model. Moreover, the P86 functional should be avoided. There is no clear preference amongst the remaining func tionals, other than noting that the hybrid functional tends to perform better than pure functionals. In calculating binding energies in transition metal compounds such as hydrides, the molecular orbital method is not very accurate until quite high levels are used to account for the electron correlation. In contrast, DFT has been shown to predict binding energies in transition metal compounds more accurately. A.4.3 Geometries Analytical first derivatives are available for almost all density functionals, a nd as a result geometry optimization can be carried out with facility. The performance of most functionals is usually quite good when it comes to predicting minimum energy structures. Accuracies in bond angles for DFT average about 1 degree, the same as is found for HF and MP2. For transition metals, DFT has proven to be a tremendous improvement over HF and post HF methods, particularly when the metal atom is coordinatively unsaturated. The narrow separation between filled and empty d block orbitals ty pically leads to enormous non dynamical

PAGE 284

284 correlation problems with an HF treatment, and DFT is much less prone to analogous problems. Even in cases of a saturated coordination sphere, DFT methods typically significantly outperform HF and MP2. Density funct ional theory has shown weakness in predicting geometries for van der Waals complexes and ionic complexes. Hydrogen bonds are somewhat too short as a rule, and most charge transfer complexes have their polarities overestimated. A.5 The Future of DF T The D FT methodology has become very popular for various calculations in quantum chemistry. Density functional theory has the advantage of allowing for correlation effects to be included in a calculation that takes roughly the same time as a HF calculation, whi ch does not include correlation. However there are some drawbacks of the DFT method and these are the areas that are currently being actively investigated. There is no systematic way of improving the DFT calculations. There is no clear roadmap for the s ystematic improvement of the E xc functionals. It is certain that a lot of research effort in the future is going to be on the formulation of these functionals. Density functional theory also has not been widely accepted as the method of choice for excite d states. The development of DFT formalisms to handle excited states remains a subject of active research. Since DFT uses approximate functionals, KS DFT is not variational and can yield an energy below the true ground state energy. Calculations with gr adient corrected functionals are size consistent. The currently used KS DFT functionals often do not give good results for activation energies of reactions. Although KS DFT yields good results for most molecular properties with the presently available fu nctionals, KS DFT cannot match the accuracy that methods like CCSD(T) and QCISD(T) can achieve. Of course, CCSD(T) and QCISD(T) are limited to dealing with small molecules, whereas DFT can handle rather large

PAGE 285

285 molecules. In summary, DFT has emerged as one of the promising fields in the field of quantum calculations. It is most certainly going to be an area of active research in the years to come.

PAGE 286

286 LIST OF REFERENCES 1 P. Van Zant, Microchip Fabrication: A Practical Guide to Semiconductor Processing (McGraw Hill, 2000). 2 M. T. Boh The r eal l imiter to h igh p Proceedings of IEEE International Elect ron Devices Meeting, 241 (1995). 3 S. P. Murarka, I. V. Verner, and R. J. Gutmann, Copper Fundamental Mechani sms for Microelectronic Applications (John Wiley and Sons, New York, NY, 2000). 4 R. I. M. Dang, and N. Shigyo, IEEE Electron Devices Lett. EDL 2 196 (1981). 5 S. P. Jeng, R. H. Havemann, and M. C. Chang, Mat. Res. Soc. Symp. 337 24 (1994). 6 G. H. Schnable, and R. S. Keen, Proceedings of the IEEE 57, 1570 (1969). 7 R. V. Penny, Physics and Chemistry of Solids 25 335 (1964). 8 K. N. Tu, J. Appl. Phys. 94 5451 (2003). 9 ment 14, 461 (1970). 10 M. Sakimoto, T. Fujii, H. Yamaguchi, and K. Eguchi, Proceedings of the IEEE Reliability Physics Symposium, Las Vegas, NV, 333, April (1995). 11 K. Maex, M. R. Baklanov, S. H. Brongersma, and Z. S. Yanovitskaya, J. Appl. Phys. 93 8793 (2003). 12 M. Michael, R. E. Todd, J. H. Zhao, C. Hu, T. Cho, and P. S. Ho, Annual Review of Materials Science 30 645 (2000). 13 F. Iacopi, C. Zistl, C. Jehoul, Z. S. Tokei, Q. T. Le, A. Das, C. Sullivan, G. Prokopowicz, D. Gronbeck, M. Gallag her, J. Calvert, and K. Maex, Microelectron. Eng. 64 351 (2002). 14 A. A. Istratov, C. Flink, and E. R. Weber, Physica Status Solidi B: Basic Research 222, 261 (2000). 15 E. R. Weber, Appl. Phys. A30 1 (1983). 16 A. A. Istratov, H. Hedemann, M. Se ibt, 0. F. Vyvenko, W. Schroter, T. Heiser, C. Flink, H. Hieslmair, and E. R. Weber, J. Electrochem. Soc., 145 3889 (1998).

PAGE 287

287 17 C. Flink, H. Feick, S. A. McHugo, W. Seifert, H. Hieslmair, T. Heiser, A. A. Istratov, and E. R. Weber, Physical Rev. Lett. 8 5 4900 (2000). 18 B. J. Howard, and C. Steinbruchel, Appl. Phys. Lett. 59 914 (1991). 19 D. Shamiryan, T. Abell, F. Iacopi, and K. Maex, Mater. Today 7 34 (2004). 20 tional Sept. (2005), http://www.reed electronics.com/semiconductor/article / CA6252222?pubdate=9%2F1%2F2005, last accessed May (2007). 21 C. L. Gan, C. V. Thompson, K. L. Pey, W. K. Choi, H. L. Tay, B. Yu, and M. K. Radhakrishnan, Appl. Phys. Lett. 79 4592 (2001). 22 C. K. Hu, K. L. Lee, D. Gupta, and P. Blauner, Mat. Res. Soc. Symp. Proc. 428 43 (1996). 23 C. K. Hu, L. Gignac, E. Liniger, B. Herbst, D. L. Rath, S. T. Chen, S. Kaldor, A. Simon, and W. T. Tseng, Appl. Phys. Lett. 83 869 (2003). 24 C. K. Hu, L. Gignac, and R. Rosenberg, Microelectron. Reliability 46 213 (2005). 25 E. G. Liniger, and C. Dziobkowski, Thin Solid Films 513 295 (200 6). 26 A. V. Vairagar, Z. Gan, V. Shao, S. G. Mhaisalkar, H. Li, K. N. Tu, Z. Chen, E. Zschech, H. J. Engelmann, and S. Zhang, J. Electrochem. Soc. 153 G840 (2006). 27 International Technology Roadmap for Semiconductors, www.itrs.net, (2005), last acc essed May (2007). 28 S. M. Rossnagel, and J. Hopwood, Appl. Phys. Lett. 63 3285 (1993). 29 A. Paranjpe, R. Bubber, L. Velo, G. Shang, S. Gopinath, J. Dalton, and M. Moslehi, Proceedings of IEEE International Interconnect Technology Conference, San Fra ncisco, CA,119, May (1999). 30 A. E. Kaloyeros, and E. Eisenbraun, Annual Rev. in Material Science 30 363 (2000). 31 M. H. Tsai, S. C. Sun, H. T. Chiu, and S. H. Chuang, Appl. Phys. Lett. 68 1412 (1996). 32 R. Ecke, S. E. Schulz, M. Hecker, and T. Gessner, Microelectron. Eng. 64 261 (2002). 33 S. Bystrova, A. A. I. Aarnink, J. Holleman, and R. A. M. Wolters, J. Electrochem. Soc. 152 G522 (2005). 34 K. Wetzig and C. M. Schneider, Metal Based Thin Films for Electronics 2nd Edition, figure 4.1(b ) page 231 (Wiley VCH, 2006).

PAGE 288

288 35 D. T. Price, R. J. Gutmann, and S. P. Murarka, Thin Solid Films 308 309 525 (1997). 36 N. A. Gjostein, Diffusion (American Society for Metals, Metal Park, Ohio 1973). 37 M. A. Nicolet, Thin Solid Films 52 415 (1978). 38 B. Movchan, and A. Demchishin, Fiz. Metal. Metalloved. 28 653 (1969). 39 L. E. Toth, Transition Metal Carbides and Nitrides (Academic Press, New York and London, 1971). 40 J. A. Thornton, and D. W. Hoffman, Thin Solid Films 171 5 (1989). 41 N. Hellgren, K. Maca, E. Broitman, M. P. Johansson, L. Hultman, and J E. Sundgren, J. Appl. Phys. 88 524 (2000). 42 M. Uekubo, T. Oku, K. Nii, M. Murakami, K. Takahiro, S. Yamaguchi, T. Nakano, and T. Ohta, Thin Solid Films 286 170 (1996). 43 D. Mao and J. Hopwood, J. Appl. Phys. 96 820 (2004). 44 J. A. T. Norman, J. Phys. IV France 11 497 (2001). 45 J. S. Becker, S. Suh, S. Wang, and R. G. Gordon, Chem. Mater. 15 2969 (2003). 46 M. Ritala, M. Leskel, Handbook of Thin Film Materials, vol. 1 103, (Academic Press 2001). 47 G. W. Book, K. Pfeifer, and S. Smith, Microelectron. Eng. 64 255 (2002). 48 H. Kizil, and C. Steinbruchel, Thin Solid Films 449 158 (2004). 49 T. Heiser, C. Brochard, and M. Swaaneen, Mat. Res. Soc. Symp. Proc. 612 D 7.3.1 (2000). 50 T. Heisner, S. McHugo, H. Hieslmair, and E. R. Weber, Appl. Phys. Lett. 70 3576 (1997). 51 S. R. Kurtz, and R. G. Gordon, Thin Solid Films 140 277 (1986). 52 A. Sherman, J. Electrochem. Soc. 137 1892 (1990). 53 N. Yokoyama, K. H inode, and Y. Homma, J. Electrochem. Soc. 138 190 (1991). 54 J. C. Hu, T. C. Chang, L. J. Chen, Y. L. Yang, S. Y. Chen, and C. Y. Chang, J. Vac. Sci. Technol. A 17 2389 (1999).

PAGE 289

289 55 M. J. Buiting, A. F. Otterloo, and A. H. Montree, J. Electrochem. Soc. 138 500 (1991). 56 C. Faltermeier, C. Goldberg, M. Jones, A. Upham, D. Manger, G. Peterson, J. Lau, and A. E. Kaloyeros, J. Electrochem. Soc. 144 1002 (1997). 57 C. Faltermeier, C. Goldberg, M. Jones, A. Upham, A. Knorr, A. Ivanova, G. Peterson, and A. E. Kaloyeros, J. Electrochem. Soc. 145 676 (1998). 58 N. J. Archer, Thin Solid Films 80 221 (1981). 59 N. J. Ianno, U. Ahmed, and D. E. Englebert, J. Electrochem. Soc. 136 276 (1989). 60 A. Intemann, H. Koerner, and F. Koch, J. Electrochem. Soc 140 3215 (1993). 61 K. Sugiyama, S. Pac, Y. Takahashi, and S. Motojima, J. Electrochem. Soc. 122 1545 (1975). 62 R. M. Fix, R. G. Gordon, and D. M. Hoffman, Chem. Mater. 2 235 (1990). 63 K. C. Park, K. B. Kim, I. J. M. M. Raaijmakers, and K. N gan, J. Appl. Phys. 80 5674, (1996). 64 S. C. Sun, and M. H. Tsai, Thin Solid Films 253 440 (1994). 65 R. M. Fix, R. G. Gordon, and D. M. Hoffman, Chem. Mater. 3 1138 (1991). 66 I. J. Raaijmakers, Thin Solid Films 247 85 (1994). 67 C. Amato Wie rda, and D. A. Wierda, J. Mater. Res. 15 2414 (2000). 68 C. K. Wang, L. M. Liu, M. Liao, H. C. Cheng, and M. S. Lin, Jpn. J. Appl. Phys. 35 4274 (1996). 69 S. H. Kim, D. S. Chung, K. C. Park, K. B. Kim, and S. H. Min, J. Electrochem. Soc. 14 6 1455 (1999). 70 I. J. Raaijmakers, and J. Yang, Appl. Surf. Sci. 73 31 (1993). 71 J. M. Bulger, C. S. Whelan, A. Dumont, M. Kuhn, and J. Clark, J. Vac. Sci. Technol. B 17 410 (1999). 72 J. Y. Yun, and S. W. Rhee, Thin Solid Films 312 24 (1998 ). 73 W. Besling, A. Satta, J. Schumacher, T. Abell, V. Sutcliffe, A. M. Hoyas, G. Beyer, D. Gravesteijn, and K. Maex, Proceedings of IEEE International Interconnect Technology Conference, Burlingame, CA, 288, June (2002).

PAGE 290

290 74 K. E. Elers, V. Saanila, P. J. Soininen, W. M. Li, J. T. Kostamo, S. Haukka, J. Juhanoja, and W. F. A. Besling, Chem. Vap. Deposition 8 149 (2002). 75 M. Ritala, M. Leskel, E. Rauhala, and P. Haussalo, J. Electrochem. Soc. 142 2731 (1995). 76 S. Smith, G. Book, W. M. Li, Y M. Sun, P. Gillespie, M. Tuominen, and K. Pfeifer, Proceedings of the IEEE International Interconnect Technology Conference, San Francisco, CA, 135, June (2003). 77 S. Smith, W. M. Li, K. E. Elers, and K. Pfeifer, Microelect. Eng. 64 247 (2002). 7 8 K. E. Elers, J. Winkler, K. Weeks, and S. Marcus, J. Electrochem. Soc. 152 G589 (2005). 79 P. Mrtensson, M. Juppo, M. Ritala, M. Leskel, and J. O. Carlsson, J. Vac. Sci. Technol. B 17 2122 (1999). 80 M. Juppo, M. Ritala, and M. Leskel, J. Elec trochem. Soc. 147 3377 (2000). 81 M. Juppo, P. Aln, M. Ritala, T. Sajavaara, J. Keinonen, and M. Leskel, Electrochem. Solid State Lett. 5 C4 (2002). 82 M. Juppo, A. Rahtu, and M. Ritala, Chem. Mater. 14 281 (2002). 83 M. Ritala, M. Leskel, E. R auhala, and J. Jokinen, J. Electrochem. Soc. 145 2915 (1998). 84 J. S. Min, Y. W. Son, W. G. Kang, S. S. Chun, and S. K. Kang, Jpn. J. Appl. Phys. 37 4999 (1998). 85 D. J. Kim, Y. B. Jung, M. B. Lee, Y. H. Lee, J. H. Lee, and J. H. Lee, Th in Solid Films 372 276 (2000). 86 J. W. Elam, M. Schuisky, J. D. Ferguson, and S. M. George, Thin Solid Films 436 145 (2003). 87 J. H. Yun, E. S. Choi, C. M. Jang, and C. S. Lee, Jpn. J. Appl. Phys. 41 L418 (2002). 88 J. Y. Kim, Y. Kim, and H. Jeon, Jpn. J. Appl. Phys. 42 L414 (2003). 89 J. Y. Kim, S. Seo, D. Y. Kim, H. Jeon, and Y. Kim, J. Vac. Sci. Technol. A 22 8 (2004). 90 J. Hillman, J. Hautala, and S. Caliendo, Proceedings of Advanced Metallization Conference, San Diego, CA, 313, Oc t. (2000).

PAGE 291

291 91 X. Chen, G. G. Peterson, C. Goldberg, G. Nuesca, H. L. Frisch, and A. E. Kaloyeros, J. Mater. Res. 14 2043 (1999). 92 A. E. Kaloyeros, X. Chen, T. Stark, K. Kumar, S. C. Seo, G. G. Peterson, H. L. Frisch, B. Arkles, and J. Sullivan, J. Electrochem. Soc. 146 170 (1999). 93 A. Y. Ganin, L. Kienle, and G. V. Vajenine, Eur. J. Inorg. Chem. 16 3233 (2004). 94 X. Chen, H. L. Frisch, A. E. Kaloyeros, B. Arkles, and J. Sullivan, J. Vac. Sci. Technol. B 17 182 (1999). 95 M. H. Tsai, S. C Sun, H. T. Chiu, C. E. Tsai, and S. H. Chuang, Appl. Phys. Lett. 67 1128 (1995). 96 M. H. Tsai, S. C. Sun, C. P. Lee, H. T. Chiu, C. E. Tsai, S. H. Chuang, and S. C. Wu, Thin Solid Films 270 531 (1995). 97 M. H. Tsai, S. C. Sun, C. E. Tsai, S. H. C huang, and H. T. Chiu, J. Appl. Phys. 79 6932 (1996). 98 R. M. Fix, R. G. Gordon, and D. M. Hoffman, Chem. Mater. 5 614 (1993). 99 G. C. Jun, S. L. Cho, and K. B. Kim, Jpn. J. Appl. Phys. 37 L30 (1998). 100 S. L. Cho, K. B. Kim, S. H. Min, and H. K. Shin, Mat. Res. Soc. Symp. Proc. 514 531 (1998). 101 H. Machida, A. Hoshino, T. Suzuki, A. Ogura, and Y. Shshita, J. Cryst. Growth 237 239 586 (2002). 102 H. T. Chiu, and W. P. Chang, J. Mater. Sci. Lett. 11 96 (1992). 103 J. W. Hong, M. L. Lee, K. I. Choi, Y. Chung, S. W. Lee, G. H. Choi, S. T. Kim, U. I. Chung, J. T. Moon, and B. I. Ryu, Proceedings of the IEEE 2005 International 141, June (2005). 104 L. Hiltunen, M. Leskel, M. Mkel, L. Niinist, E. Nyknen, and P. Soininen, Thin Solid Films 166 149 (1988). 105 M. Ritala, P. Kalsi, D. Riihel, K. Kukli, M. Leskel, and J. Jokinen, Chem. Mater. 11 1712 (1999). 106 P. Aln, M. Juppo, M. Ritala, and M. Leskel, J. Mater. Res 17, 107 (2002). 107 H. Kim, A. J. Kellock, and S. M. Rossnagel, J. Appl. Phys. 92 7080 (2002).

PAGE 292

292 108 E. Eisenbraun, O. van der Straten, Y. Zhu, K. Dovidenko, and A. Kaloyeros, Proceedings of the IEEE International, Burlingame, CA, 207, June (2001). 109 K. I. Choi, B. H. Kim, S. W. Lee, J. M. Lee, W. S. S ong, G. H. Choi, U. I. Chung, and J. T. Moon, Proceedings of the IEEE International Interconnect Technology Conference, 6th, Burlingame, CA, 129, June (2003). 110 J. S. Park, H. S. Park, and S. W. Kang, J. Electrochem. Soc. 149 C28 (2002). 111 Y. Y. Wu, A. Kohn, and M. Eizenberg, J. Appl. Phys. 95 6167 (2004). 112 A. Furuya, H. Tsuda, and S. Ogawa, J. Vac. Sci. Technol. B 23 979 (2005). 113 A. Furuya, N. Ohtsuka, N. Ohashi, S. Kondo, and S. Ogawa, J. Vac. Sci. Technol. A 24 103 (2006). 11 4 W. J. Maeng, S. J. Park,and H. Kim, J. Vac. Sci. Technol. B 24 2276 (2006). 115 H. Kim, C. Detavenier, O. van der Straten, S. M. Rossnagel, A. J. Kellock, and D. G. Park, J. Appl. Phys. 98 014308 1 (2005). 116 J. W. Hong, K. I. Choi, Y. K. Lee, S G. Park, S. W. Lee, J. M. Lee, S. B. Kang, G. H. Choi, S. T. Kim, U. I. Chung, and J. T. Moon, Proceedings of the IEEE International, 141, June (2005). 117 M. Nagai, and K. Kishida, Appl. Surf. Sci. 70/71 759 (1993). 118 M. Nagai, N. Hirano, and S. Omi, Jpn. J. Appl. Phys. 39 4558 (2000). 119 T. Nakajima, K. Watanabe, and N. Watanabe, J. Electrochem. Soc. 134 3175 (1987). 120 N. Gonohe, Materials Transactions 43 1585 (2002). 121 S. D. Marcus, and R. F. Foster, Thin Solid Films 236 330 (199 3). 122 C. Meinier, C. Monteil, C. Savall, F. Palmino, J. Weber, R. Berjoan, and J. Durand, Appl. Surf. Sci. 125 313 (1998). 123 A. R. Ivanova, C. J. Galewski, C. A. Sans, T. E. Seidel, S. Grunow, K. Kumar, and A. E. Kaloveros, Mat. Res. Soc. Symp. 56 4 321 (1999). 124 C. W. Lee, Y. T. Kim, and S. K. Min, Appl. Phys. Lett. 62 3312 (1993). 125 C. W. Lee, Y. T. Kim, C. Lee, J. Y. Lee, S. K. Min, and Y. W. Park, J. Vac. Sci. Technol. B 12 69 (1994).

PAGE 293

293 126 C. W. Lee, Y. T. Kim, and J. Y. Lee, Appl. Phys. Lett. 64 619 (1994). 127 Y. T. Kim, C. W. Lee, and S. K. Min, Jpn. J. Appl. Phys. 32 6126 (1993). 128 C. S. Kwon, D. J. Kim, C. W. Lee, Y. T. Kim, and I. H. Choi, Mat. Res. Soc. Symp. Proc. 355 441 (1995). 129 M. F. Bain, B. M. Armstrong, and H. S. Gamble, J. Mat. Sci. 14 329 (2003). 130 J. Lin, A. Tsukune, T. Suzuki, and M. Yamada, J. Vac. Sci. Technol. A 16 611 (1998). 131 S. Ganguli, L. Chen, T. Levine, B. Zheng, and M. Chang, J. Vac. Sci. Technol. B 18 237 (2000). 132 H. T. C hiu, S. H. Chuang, Mat. Res. Soc. Symp. Proc. 250 317 (1992). 133 H. T. Chiu, S. H. Chuang, Mat. Res. Soc. Symp. Proc. 8 1353 (1993). 134 E. L. Crane, H. T. Chiu, and R. G. Nuzzo, J. Phy. Chem. B 105 3549 (2001). 135 J. E. Kelsey, C. Goldberg, G. Nuesca, G. Peterson, and B. Arkles, J. Vac. Sci. Technol. B 17 1101 (1999). 136 B. H. Lee, and K. Yong, J. Electrochem. Soc. 151 C594 (2004). 137 S. Jeon, K. Yong, S. G. Park, and S. W. Rhee, J. Vac. Sci. Technol B 24 1428 (2006). 138 R. G. Gordon, S. Barry, R. N. R. Broomhall Dillard, V. A. Wagner, and Y. Wang, Mat. Res. Soc. Symp. Proc. 612 D9121 (2000). 139 J. W. Klaus, S. J. Ferro, and S. M. George, Appl. Surf. Sci. 162 163 479 (2000). 140 J. W. Klaus, S. J. Ferro, and S. M. George, J. Electrochem. Soc. 147 1175 (2000). 141 S. H. Kim, J. K. Kim, N. Kwak, H. Sohn, J. Kim, S. H. Jung, M. R. Hong, S. H. Lee, and J. Colling, Electrochem. Solid State Lett. 9 C54 (2006). 142 H. S. Sim, S. Kim II, and Y. T. Kim, J. Vac. Sci. Techn ol. B 21 1411 (2003). 143 C. W. Lee, and Y. T. Kim, J. Vac. Sci. Technol. B 24 1432 (2006). 144 J. S. Becker, and R. G. Gordon, Appl. Phys. Lett. 82 2239 (2003). 145 W. M. Li, K. Elers, J. Kostamo, S. Kaipio, S. Huotari, M. Soininen, P. J. Soinin en, M. Tuominen, S. Haukka, S. Smith, and W. Besling, Proceedings of the IEEE International Interconnect Technology Conference, 191, June (2002).

PAGE 294

294 146 W. M. Li, M. Tuominen, S. Haukka, H. Sprey, and I. J. Raaijmakers, Solid State Technology 106 103 (2003). 147 S. H. Kim, S. S. Oh, K. B. Kim, D. H. Kang, W. M. Li, S. Haukka, and M. Tuominen, Appl. Phys. Lett. 82 4486 (2003). 148 S. J. Wang, H. Y. Tsai, S. C. Sun, and M. H. Shiao, J. Electrochem. Soc. 148 G500 (2001). 149 A. M. Hoyas, Y. Travaly, J. Schumacher, T. Sajavaara, C. M. Whelan, B. Eyckens, O. Richard, S. Giangrandi, B. Brijs, W. Vandervorst, K. Maex, J. P. Celis, A. M. Jonas, and A. Vantomme, J. Appl. Phys. 99 F64 (2006). 150 K. S. Kim, M. S. Lee, S. S. Yim, H. M. Kim K. B. Kim, H. S. Park, W. Koh, W. M. Li, M. Stokhof, and H. Sprey, Appl. Phys. Lett. 89 081913 1 (2006). 151 A. M. Hoyas, J. Schumacher, C. M. Whelan, M. R. Baklanov, L. Carbonell, J. P. Celis, and K. Maex, J. Vac. Sci. Technol. B 23 1551 (2005). 152 D. Emur, V. Terzieva, J. Schumacher, V. Sutcliffe, C. M. Whelan, and K. Maex, J. Electrochem. Soc. 152 B512 (2005). 153 O. J. Bchir, K. M. Green, M. S. Hlad, T. J. Anderson, B. C. Brooks, C. B. Wilder, D. H. Powell, and L. McElwee White, J. Organ omet. Chem. 684 338 (2003). 154 O. J. Bchir, S. W. Johnston, A. C. Cuadra, T. J. Anderson, C. G. Oritz, B. C. Brooks, D. H. Powell, and L. McElwee White, J. Cryst. Growth 249 262 (2003). 155 O. J. Bchir, K. C. Kim, T. J. Anderson, V. Craciun, B. C. B rooks, and L. McElwee White, J. Electrochem. Soc. 151 G697 (2004). 156 O. J. Bchir, K. M. Green, M. S. Hlad, T. J. Anderson, B. C. Brooks, and L. McElwee White, J. Cryst. Growth 261 280 (2004). 157 O. J. Bchir, K. M. Green, H. M. Ajmera, E. A. Zapp, T. J. Anderson, B. C. Brooks, L. L. Reitfort, D. H. Powell, K. A. Abboud, and L. McElwee White, J. Am. Chem. Soc. 127 7825 (2005). 158 R. I. Hegde, R. W. Fiordalice, E. O. Travis, and P. J. Tobin, J. Vac. Sci. Technol. B 11 1287 (1993). 159 R. Leute recker, B. Frschle, U. Cao Minh, and P. Ramm, Thin Solid Films 270 621 (1995).

PAGE 295

295 160 P. T. Liu, T. C. Chang, J. C. Hu, Y. L. Yang, and S. M. Sze, J. Electrochem. Soc. 147 368 (2000). 161 A. Paranjpe, and M. IslamRaja, J. Vac. Sci. Technol. B 13 210 5 (1995). 162 M. Danek, M. Liao, J. Tseng, K. Littau, D. Saigal, H. Zhang, R. Mosley, and M. Eizenberg, Appl. Phys. Lett. 68 1015 (1996). 163 D. H. Kim, S. L. Cho, K. B. Kim, J. J. Kim, J. W. Park, and J. J. Kim, Appl. Phys. Lett. 69 4182, (1996). 164 J. Bonitz, S. E. Schulz, and T. Gessner, Microelectron. Eng. 70 330 (2003). 165 G. S. Sandhu, S. G. Meikle, and T. T. Doan, Appl. Phys. Lett. 62 240 (1993). 166 J. Y. Yun, M. Y. Park, and S. W. Rhee, J. Electrochem. Soc. 146 1804 (1999). 167 J. Y. Kim, G. H. Choi, Y. D. Kim, Y. Kim, and H. Jeon, Jpn. J. Appl. Phys. 42 4245 (2003). 168 J. W. Elam, C. A. Wilson, M. Schuisky, Z. A. Sechrist, and S. M. George, J. Vac. Sci. Technol. B 21 1099 (2003). 169 K. Hieber, Thin Solid Film 24 15 7 (1974). 170 D. W. Sloan, P. M. Blass, and J. M. White, Appl. Surf. Sci. 143 142 (1999). 171 X. Zhao, N. P. Magtoto, and J. A. Kelber, Thin Solid Films 478 188 (2005). 172 G. Beyer, A. Satta, J. Schuhmacher, K. Maex, W. Besling, O. Kilpela, H. Spr ey, and G. Tempel, Microelectron. Eng. 64 233 (2002). 173 O. J. Bchir, Dissertation, University of Florida (2005). 174 R. C. Reid, J. M. Prausnitz, B. E. Poling, The Properties of Gases and Liquids, 4th Edition (McGraw Hill, New York, NY 1987). 175 S. A. Gokoglu, PhD Thesis, Yale University (1982). 176 R. B. Bird, W. E. Stewart, and E. N. Lightfoot, Transport Phenomenoa (John Wiley and Sons, New York, NY 1960). 177 M. A. Kuczmarski, J. Wide Bandgap Mater. 7 192 (2000). 178 Database of Joint Committee on Powder Diffraction Pattern (JCPDS), (2006).

PAGE 296

296 179 B. D. Culity, and S. R. Stock, Elements of X ray Diffraction, 3rd Edition (Prentice Hall, Upper Saddle River, NJ, 2001). 180 Handbook of Auger Electron Spectroscopy edited by K. D. Childs, B. A. Carlson, L. A. LaVanier, J. F. Moulder, D. F. Paul, W. F. Stickle, and D. G. Watson, (Physical Electronics, Eden Prairie, 3rd Edition, 1995). 181 G. Soto, W. de la Cruz, and M. H. Farias, J. Electron Spectroscopy and Related Phenomena 135 27 (2004 ). 182 Y. G. Shen, and Y. W. Mai, Material Sci. Eng. A 288 47 (2000). 183 J. Luthin, and C. Linsmeier, J. Nuc. Mater. 290 121 (2001). 184 Y. M. Sun, S. Y. Lee, A. M. Lemonds, E. R. Engbrecht, S. Veldman, J. Lozano, J. M. White, J. G. Ekerdt, I. Em esh, and K. Pfeifer, Thin Solid Films 397(1 2) 109 (2005). 185 M. B. Zellner, and J. G. G. Chen, Catal. Today 99(3 4) 299 (2005). 186 A. Katrib, F. Hemming, P. Wehrer, L. Hilaire, and G. Maire, J. Electron Spectroscopy and Related Phenomena 76 195 ( 1995). 187 D. Gogova, K. Gesheva, A. Kakanakova Georgieva, and M. Surtchev, Eur. Physical Journal Appl. Phys. 11 167 (2000). 188 G. E. McGuire, G. K. Schweitzer, and T. A. Carlson, Inorg. Chem. 12 2450 (1973). 189 J. L. Alay, H. Bender, G. Brijs, A Demesmaeker, and W. Vandervorst, Surface and Interface Analysis 17 373 (1991). 190 J. M. Giraudon, P. Devassine, J. F. Lamonier, L. Delannoy, L. Leclercq, and G. Leclercq, J. Solid State Chemistry 154 412 (2000). 191 R. J. Colton, and J. W. Rabal ais, Inorg. Chem. 15 236 (1976). 192 H. Nakazawa, T. Kawabata, M. Kudo, and M. Mashita, Appl. Surf. Sci. 253 4188 (2007). 193 N. Yasui, H. Inaba, and S. Sasaki, e Journal of Surf. Sci. Nanotech. 4 129 (2007). 194 Handbook of X ray Photoelectron Sp ectroscopy edited by J. F. Moulder, W. F. Stickle, P. E. Sobol, and K. D. Bomber, (Physical Electronics Inc., 1995). 195 S. F. Pedersen, and R. R. Schrock, J. Am. Chem. Soc. 104 7483 (1982). 196 D. Briggs, and M. P. Seah, Practical Surface Analysis V ol. 1, (John Wiley & Sons, 1992).

PAGE 297

297 197 K. E. Elers, V. Saanila, W. M. Li, P. J. Soininen, J. T. Kostamo, S. Haukka, J. Juhanoja, and W. F. A. Besling, Thin Solid Films 434 94 (2003). 198 C. J. Carmalt, A. C. Newport, S A. O'Neill, I. P. Parkin, A. J. P. White, and D. J. Williams, Inorg. Chem. 44 615 (2005). 199 B. S. Lim, A. Rahtu, and R. G. Gordon, Nature Mater. 2 749 (2003). 200 Z. W. Li, S. T. Barry, and R. G. Gordon, Inorg. Chem. 44 1728 (2005). 201 A. Baunemann, D. Rische, A. Milanov, Y. Kim, M. Winter, C. Gemel, and R. A. Fischer, Dalton Transactions 18 3051 (2005). 202 C. B. Wilder, L. L. Reitfort, K. A. Abboud, and L. McElwee White, Inorg. Chem. 45 263 (2006). 203 P. Prieto, and R. E. Kirby, J. Vac. Sci. Techno. A 13 2819 (1995) 204 J. T. Scheper, P. J. McKarns, T. S. Lewkebandara, and C. H. Winter, Mater. Sci. Semicond. Process 2 149 (1999). 205 C. J. Cramer, Essentials of Computational Chemistry (Wiley Interscience, 2004). 206 P. J. Hay, and W. R. Wadt, J. Chem. Phys. 82 270 (1985). 207 L. Laaksonen, J. Mol. Graph. 10 33 (1992). 208 G. Schaftenaar, and J. H. Noordik, J. Computat. Aided Mol. Design 14 123 (2000). 209 Y. S. Won, Y. S. Kim, T. J. Anderson, L. L. Reitfort, I. Ghiviriga, and L. McElwee White, J. Am Chem. Soc. 128 13781 (2006). 210 P. Hohenberg, and W. Kohn, Phys. Rev. 136 B864 (1964). 211 D. M. Ceperley, and B. J. Alder, J. Phys. Rev. Lett. 45 566 (1980). 212 S. J. Vosko, L. Wilk, and M. Nusair, Can. J. Phys. 58 1200 (1980). 213 A. D. B ecke, J. Chem. Phys. 38 3098 (1988). 214 G. J. Laming, and N. C. Handy, J. Chem. Phys. 99 8756 (1993). 215 M. Filatov, and W. Thiel, Mol. Phys. 91 847 (1997). 216 C. Adamo, and V. Barone, J. Chem. Phys. 108, 664 (1998).

PAGE 298

298 217 J. P. Perdew, Phys. R ev. B 33 8822 (1986). 218 J. P. Perdew, and Y. Wang, Phys. Rev. B 45 13244 (1992). 219 A. D. Becke, J. Chem. Phys. 104 1040 (1996). 220 C. Lee, W. Wang, R. G. Parr, Phys. Rev. B 37 785 (1988).

PAGE 299

299 BIOGRAPHICAL SKETCH Hiral M. Ajmera was born in Mumbai, India. He is the son of Mahesh and Neela Ajmera, and has a brother, Abhijit. He is married to Kinjal Ajmera. He did his schooling from Kendriya Vidyalaya, Rajkot in 1995 and received his bachelor of engineering (chemical) degree from Maharaja S ayajirao University, Baroda, India in 1999. He attended Rensselaer Polytechnic Institute from 1999 to 2000 and received his MS degree in chemical engineering in 2001. Upon graduation, he worked for two years at Applied Materials Inc. as a process enginee r. He joined NuTool in 2003 as a process engineer and worked there for 8 months. In August 2003, he joined the PhD program in Chemical Engineering at University of Florida. Upon graduation, he plans to work for Intel Corporation in Portland, Oregon.


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID E20101129_AAAAAS INGEST_TIME 2010-11-29T23:51:47Z PACKAGE UFE0021186_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES
FILE SIZE 2187 DFID F20101129_AAAOAG ORIGIN DEPOSITOR PATH ajmera_h_Page_214thm.jpg GLOBAL false PRESERVATION BIT MESSAGE_DIGEST ALGORITHM MD5
b778708b9bc62487b128f28ded8941d7
SHA-1
e701a17af7dbc3b1b2c78b02f00949dbcabba962
2123 F20101129_AAANDN ajmera_h_Page_123.txt
223d43dbcc56cd4865dfff46871b7323
6ccfb1daa3550e6286e4702a951cd215da38ab57
37248 F20101129_AAANQY ajmera_h_Page_090.QC.jpg
55fa93b4b0c3d027ce6f2bd5e6ce553c
db96e91f23b57e0ee8235071cef656b909bc76bc
1053954 F20101129_AAAMGU ajmera_h_Page_118.tif
4a45ed7771418339cda9b6095d71ebae
a5a264282a24f6180896152c217dc48f640d175e
7693 F20101129_AAAOAH ajmera_h_Page_214.QC.jpg
b70f5f0310101f7e9c9716c4029f160c
77dbcdd65a4bf424f0613260c7bab21b03d83faa
2050 F20101129_AAANDO ajmera_h_Page_124.txt
4d04907af366299b2cc53a93fc5806aa
4925690f452cf421bfe9c1d9feace8eb8184bb55
8719 F20101129_AAANQZ ajmera_h_Page_091thm.jpg
d1ada48978fc3014403bc78de10ebb13
3181487dc5a98d327487a483620fea5de3e9b314
F20101129_AAAMGV ajmera_h_Page_119.tif
b0b38f00fcfac6a9c7d9989069c2be34
1730cc03ffb1a08da022cb8e52b384c586dca406
5547 F20101129_AAAOAI ajmera_h_Page_215thm.jpg
9002ebb62d8cc0535459aff9e2e2788f
91f142b28b70ab675a91fe732e02bbcd856e7b4b
1836 F20101129_AAANDP ajmera_h_Page_125.txt
04b1b04c1501a09c72f2f5621954740b
06d22ea5cc12eda90ce15f4537d0c8dec3f7d770
F20101129_AAAMGW ajmera_h_Page_120.tif
7c8031948e53f8850e398d3717bfb329
0f19a6683dc511f7feb820c1fc97eaa733fc03eb
17892 F20101129_AAAOAJ ajmera_h_Page_215.QC.jpg
089419ef9204fab9db81874abce220be
bb9c7adedc9c90085c1a2bd6f891e52f9fda05b9
217 F20101129_AAAMZA ajmera_h_Page_003.txt
92dac4517f4691a39c9ec9c509f6fe31
0bb6c909c5552f3b8a655e1a54ef88dd67bfb1e3
1450 F20101129_AAANDQ ajmera_h_Page_126.txt
4141b87cc5c05fbfbc521d7042f31643
63e541c218b015b2fef7de2bfa2f852febf5b69c
F20101129_AAAMGX ajmera_h_Page_121.tif
28feba0201249b856a10f7b446547679
a34f0107d4c5dd6608bdfaca47a78084437cbe22
5328 F20101129_AAAOAK ajmera_h_Page_216thm.jpg
c2b9885c65fd097dde564402e3df36e9
edcba58fdc5b92455af664b3d63f719b913bef76
1253 F20101129_AAAMZB ajmera_h_Page_004.txt
370ab2f1cd23dad9a3056b05e7379d38
d1c544943e5c7a14b17f474946716f0827a61ed7
1603 F20101129_AAANDR ajmera_h_Page_127.txt
6112ff68dfbe56b880a8b2a0e81406ba
1d865d73ce02dba9c33f1562aa33519834668376
F20101129_AAAMGY ajmera_h_Page_122.tif
b361155241c0d96b54bf3164fc2db583
ddbf0ecef903f3a38608a41c81a16dcc5aee4324
20852 F20101129_AAAOAL ajmera_h_Page_216.QC.jpg
d32e399c46340769c9f36984e5e9903e
90e14ec27d2dc9bd5d88acbb1138ac14471dbd41
3689 F20101129_AAAMZC ajmera_h_Page_005.txt
950b07826c1a5eecace8a34dc7cad859
1a27b03e47f7c4aadb3fbbc57bec8750d50b78d7
1502 F20101129_AAANDS ajmera_h_Page_128.txt
091502a9de4d0326263414f4f12768ce
ff6bc2269fd82e6f6aedc91c4c5e2ba236ea8f51
F20101129_AAAMGZ ajmera_h_Page_123.tif
69bc19500003d8d299d6cbb25f261f26
5b724bc84a99e3ed6b3b259f536a9688f9b2b641
5653 F20101129_AAAOAM ajmera_h_Page_217thm.jpg
e5e65bc84483f0128680ae5286f16103
1514e96b630f9c6216d7d56c93d8c6f46131af81
4278 F20101129_AAAMZD ajmera_h_Page_006.txt
73cec4f3d1b0037891ce929c41b31cc4
8b7eea87b1d526d5db59b272f741de3eaae1a428
1189 F20101129_AAANDT ajmera_h_Page_129.txt
ed1786f4b474a340489dc40c1a5182e1
7bada89d1919829f086365bb93fab41373c7f84c
21727 F20101129_AAAOAN ajmera_h_Page_217.QC.jpg
bac8ba1e57f72c087359bebc010de2ad
c8f06188d993b141f42ce95e5e930a7a4d94023b
3827 F20101129_AAAMZE ajmera_h_Page_007.txt
86b2871d1cd1ac21f9cd06b82dba5f11
8eb0498f41db1ca78c5a2565cbcde605f7332c68
740 F20101129_AAANDU ajmera_h_Page_130.txt
e2416bf852a5ac0b743d80bf9da8dc6f
aa30ebd5d65c01f1fda736d5c366605b9fee2714
27558 F20101129_AAALPA ajmera_h_Page_248.jpg
ac4e7795c4dc711282e0ff4025400d97
eb696a87f278681bb302e9e1ff4df30dbc6518e4
5080 F20101129_AAAOAO ajmera_h_Page_218thm.jpg
c60dca2ee5ac8856ef5aac4c490df8f4
da2d4e8c3cd9b65862d7bb0539826d9f76cca666
3365 F20101129_AAAMZF ajmera_h_Page_008.txt
5d34ee10aa66056d8accdf3fa5db6c09
6373da28f6084ef155d4810d43a5ba594ee3d4ce
306 F20101129_AAANDV ajmera_h_Page_131.txt
ee6b8eee1dd27afaacd47c564b73e7c2
294bbd49c9fef988a355846e6c10c7caf1d106a3
25830 F20101129_AAALPB ajmera_h_Page_249.jpg
91cbd192de8322d49092e7f65560c138
70f3a6aeef6ca5c4878e1eb5fbe351ff3e2f6feb
16854 F20101129_AAAOAP ajmera_h_Page_218.QC.jpg
ed2466e92755249075b47072fdf84407
b7d6f9988272872b4af916faf4b67efe75e7d6c0
2589 F20101129_AAAMZG ajmera_h_Page_009.txt
63f8cf6584d56f60b09f203029c1ec01
cc73b2c9e8d5a9baa1ce83d5ec715f8a60558d24
276 F20101129_AAANDW ajmera_h_Page_132.txt
da5702d67a0f85135746132121009785
a1554a2d15fac718231cdd100532d3b3ae468e92
51742 F20101129_AAALPC ajmera_h_Page_250.jpg
1eecb41334f8e68f7bf5b16351e21def
c71c42836f59da35cc01d4df1f1539e8d557c34a
4210 F20101129_AAAOAQ ajmera_h_Page_219thm.jpg
3db943a50e6ad16d2c776dad1a90e9e0
637d963011341003a8acd3b1288735838002bef4
9397 F20101129_AAANWA ajmera_h_Page_158thm.jpg
9e9d246b39f4715ff4001328f91c1b84
03cca0d99284595815b270bed025afe0bf79cb3e
1739 F20101129_AAAMZH ajmera_h_Page_010.txt
3279603e77f0001bfd4fd8cd40c70673
86992c7718f9b28625c1ad45a98e7bbbd7642f3f
178 F20101129_AAANDX ajmera_h_Page_133.txt
85a212eac8752e218d35526c3709e2a1
d09a2125953b92a61a87aa521d000dd9a6c653b9
27204 F20101129_AAALPD ajmera_h_Page_251.jpg
8063e0889eaa60b0a72631d13abde4c2
23dbdab23138f82707f822e3a46591f6513b5a36
13552 F20101129_AAAOAR ajmera_h_Page_219.QC.jpg
bb543c1189f713c5a8a18c4ded1b901a
7739fc5e4cd33f6cd2f29d5d2bc14b78bc31da39
37820 F20101129_AAANWB ajmera_h_Page_158.QC.jpg
c37179cfb09f243ed60b598afa47fc02
fb1b0bed601521a3ac43781eb98386fb9f68a1b4
2521 F20101129_AAAMZI ajmera_h_Page_011.txt
4eede17decc1c65249ae2bb24e163ab0
c425e4bc31fcc56161f51216d19423c5ef610d39
168 F20101129_AAANDY ajmera_h_Page_134.txt
da1cb36d9c326ffdd2cd5ade39acfbec
107b705eb47a535414cab39527affb2b46388599
28334 F20101129_AAALPE ajmera_h_Page_252.jpg
0f376dbbafa9c5df54849797cc8f29fe
484ab1b01cc79530a01a39192c3e095dbe3001c0
6365 F20101129_AAAOAS ajmera_h_Page_220thm.jpg
743812f974079e0d960d35c16564314a
51b6467f94e7fb672befa34cfdfa2d0c58ef9147
9247 F20101129_AAANWC ajmera_h_Page_159thm.jpg
9d398f8408a42034647a128386cb4459
c5a501de3054f8fae1937c39667fcc9d39f02b33
3076 F20101129_AAAMZJ ajmera_h_Page_012.txt
c1c514a46e464ef64fa5234ded011ed6
fcac11fc8d59ab64e49ca30cb735daceea28d535
2135 F20101129_AAANDZ ajmera_h_Page_135.txt
b42c63dd36675d5329ee451751769095
a2c93af94bd5881fa41f583341a58f7656d66d19
72637 F20101129_AAALPF ajmera_h_Page_253.jpg
93b1f397dce7a4054108a71938e5d87a
604ee566b0be16b421cf0ff15edfc971ef76fe99
18693 F20101129_AAAOAT ajmera_h_Page_220.QC.jpg
cd2a74619e6ef17f9666f7d3b13a8be7
624d3b198a36c43cbcb15048b4e5a366551d787d
39573 F20101129_AAANWD ajmera_h_Page_159.QC.jpg
0c9a08bf42b30b66f2da51ac2995926c
1f872ebe816f07dde8b361b973dac2e1f02aec31
2960 F20101129_AAAMZK ajmera_h_Page_013.txt
028aef2e33d8a4f912ae7145178c9884
e1763313b08164f75ecb21fbe86f5334116d7df8
26269 F20101129_AAALPG ajmera_h_Page_254.jpg
73049356de6c283c0822289f042863af
39c77c7c8d3a266b94f3162eea7444fea0dc9a06
6118 F20101129_AAAOAU ajmera_h_Page_221thm.jpg
7da6f4bd77bd26cda7a29b28d3ca25ae
718b09e97dc5467b00a183cb18af3976f8de0231
9332 F20101129_AAANWE ajmera_h_Page_160thm.jpg
9e4a73ea2d899441646b85ceabdce9be
449ccd55744e7504bfabe64b82b107c094e590c5
3014 F20101129_AAAMZL ajmera_h_Page_014.txt
38b3d5d0de87b1fe0890a25ef2d9c518
3b90e28d19ae157bf8d6ffeb0622c1e1238827e5
78520 F20101129_AAALPH ajmera_h_Page_255.jpg
70f3dc7f26ef7602d387ec32b0a9d5e3
b33a69d8eecb6f5babad6009aced3ae9ec4eb2e6
F20101129_AAAMMA ajmera_h_Page_257.tif
48bee1033f62fd0fe96225f0c47eb7fa
bfe4565a6974c5fc4a1e3e5089fc3df611b31011
18141 F20101129_AAAOAV ajmera_h_Page_221.QC.jpg
234d67ef0f8e6d7dd98e43e9881dadc4
3cffc71e652535c1b07af4a998f1cabcbafa5b7b
38412 F20101129_AAANWF ajmera_h_Page_160.QC.jpg
1dd9bb984b7e3342e95713cba21e6d5f
b1bfbae5e476f455dc3c984e44128aa5420ac866
3183 F20101129_AAAMZM ajmera_h_Page_015.txt
856f831cac708b8a5372e7a6d2e22f82
f46de4cfdb2e0c379c742e3bdb920d4b122363cd
48501 F20101129_AAALPI ajmera_h_Page_256.jpg
aef6119cc3e734ea78658600174b768b
1dd6c331d6e23dbaa0b6220a8e4d6da03eab1077
F20101129_AAAMMB ajmera_h_Page_258.tif
4b111e3ae33872ca90955aa075f31b85
af11c4f43b5f6169f8efd2a11489fda543773acf
5000 F20101129_AAAOAW ajmera_h_Page_222thm.jpg
e45a3b52ec4c79235869230339f437b1
bbeabc02c138383b581b98d01ca38cc1ede05d47
9037 F20101129_AAANWG ajmera_h_Page_161thm.jpg
e0390fe839b437f14c3fc7b9989fa1bb
34eb3ed44ea08aa13caa0f50903df088f1d45d41
1987 F20101129_AAAMZN ajmera_h_Page_016.txt
38fa401961e9c962a643705c436ac021
373913009e247f432010ce8c1623d3328e9342cf
100594 F20101129_AAALPJ ajmera_h_Page_257.jpg
bd16f9a755750090959475f652195b66
7ce2ee9b7104109662b4ab9458cee848ddd4d28b
F20101129_AAAMMC ajmera_h_Page_259.tif
43512da4e38064711e95d9ab79ba554e
6e47207809db65272bf788f4c57d457d6b80d762
16397 F20101129_AAAOAX ajmera_h_Page_222.QC.jpg
6d31ca936f4f80ba92a1d5b63b83e20d
801df09a196fbad1aa61be64858093b5725c6958
37200 F20101129_AAANWH ajmera_h_Page_161.QC.jpg
b4516029afb6f36c672a73bd70efd0a9
0a35c84f4c2f01abde60506e591f24b7ebc196b2
2054 F20101129_AAAMZO ajmera_h_Page_017.txt
c3430e9f29504dc968fb09a0ad19c75f
e9c33287d9814b464c1056259bd03d50e56bba15
97476 F20101129_AAALPK ajmera_h_Page_258.jpg
029a1657e9903474e691225c0bcbca31
a61da80c357ca89d5ab66495957cfed65e5b9227
F20101129_AAAMMD ajmera_h_Page_260.tif
25569c005a58b2ddf446db87e926ffbc
51140db6820a1a7d5d54652b4d81517bdb85e78e
5280 F20101129_AAAOAY ajmera_h_Page_223thm.jpg
1003c5266b6cf065d5137b42c92fc776
00e65f6b53d614073d61bd5cb6684190adb5364a
8899 F20101129_AAANWI ajmera_h_Page_162thm.jpg
8580e1bae7b2b1116db06c3f6be3904b
73225dd8832e6fc8be80a7c6ea68b13dd51b166a
918 F20101129_AAAMZP ajmera_h_Page_018.txt
e34a182a2219182e3bbe6e79063fb878
dca5b378113524a8cf161afc80eb05c65c2c0d50
100754 F20101129_AAALPL ajmera_h_Page_259.jpg
1af20ca1d5242901de32463009e12822
6e6c8d476b960c33cd810a432d034fe7f2128c63
F20101129_AAAMME ajmera_h_Page_261.tif
214fff23f1083d4d832648e663d60736
024b7b96fa352e5c530ca0f1437e8bf56182239c
19857 F20101129_AAAOAZ ajmera_h_Page_223.QC.jpg
386f8781d9e946496ff99bf23b05dc35
ccdd2c8334a78572ddf51b7e81e2681dab773d97
38106 F20101129_AAANWJ ajmera_h_Page_162.QC.jpg
d9b3d60aa53917f5ef2fd621dd9ed5a7
cc9eef6eae2372c48c063b10ff8fba42e5a9e517
1919 F20101129_AAAMZQ ajmera_h_Page_019.txt
9b5576b89843a7b5b49c2789e682b167
e2e525140045ce62d1b808e88c7dbe73f1aa7162
114302 F20101129_AAALPM ajmera_h_Page_260.jpg
286a37211001154ab5d37cb253504c1d
cea3bfb2925af79fab7d577974893b0f1a387b68
F20101129_AAAMMF ajmera_h_Page_262.tif
7176c63797ca28cdda8c099094601206
1e048a93ba0228a56195dbe25a83743cf586a8cc
8637 F20101129_AAANWK ajmera_h_Page_163thm.jpg
5e98600ff34c44ab09a677b2c9b15dfe
19eeabae79d9a103597df74a18c239f58f94d33e
2153 F20101129_AAAMZR ajmera_h_Page_020.txt
72503e6b6b445d8d1d330aa6582a3526
09c4abfdd414523c76d75eea5e6f37698565bd0c
112358 F20101129_AAALPN ajmera_h_Page_261.jpg
03f9132d6d44510c53b467f71adbbfe5
6a3192464ad31d70f3a0f45d52bfc39713cfde95
F20101129_AAAMMG ajmera_h_Page_263.tif
ee4586b1047d85502e2c53ed6a594418
3df16db313742f0f38034134a29caee4267ec0c1
36820 F20101129_AAANWL ajmera_h_Page_163.QC.jpg
0bc7c0f66b061e052faedfa47f7ad980
d384d689d155234b71d773c2d5bf6245ed913b7e
851 F20101129_AAANJA ajmera_h_Page_270.txt
0183fa575a6355c07fdc5705012093a4
70a22a27225291d597d36ed6827eca88847c8351
1793 F20101129_AAAMZS ajmera_h_Page_021.txt
9af378a2102316a2d03b9bb33c6b3e91
a47426b1ce9a116be0855d1957d891be5cb5e033
88221 F20101129_AAALPO ajmera_h_Page_262.jpg
d10a65ac8a44ab9bc1d4caa5ebf3d6c2
2f7056f47f3c3641660e92cce5681dbef532e82e
25271604 F20101129_AAAMMH ajmera_h_Page_264.tif
5ab5e0b0ff1b9fa3a188a28031405150
9d0644b4374e81360767a01b5911400208cec321
8991 F20101129_AAANWM ajmera_h_Page_164thm.jpg
e8a9cd686f3721c7b842b42df8407072
f9d6a4da78e2b79ee31146eac2f6c5f7167923d0
579 F20101129_AAANJB ajmera_h_Page_271.txt
e5761f9b91f81ecb6908705e3431c315
768491ae6cda8678be6cbd534ac103026e1b88ef
1586 F20101129_AAAMZT ajmera_h_Page_022.txt
4d3623e280c550e45ed7d107555b8e8e
6ac8d39c4975e21a66d8267aa72b66c2b93297df
F20101129_AAAMMI ajmera_h_Page_265.tif
1a2832e76a08d6fdb8e96a8dcc6606ec
a11c786d9c59b5582abb3fea78c952de4af9a21a
36233 F20101129_AAANWN ajmera_h_Page_164.QC.jpg
c0a455abe236ada75329867ca4ffdc8f
7666aa2fed817dae17609594443ea7d158515683
449 F20101129_AAANJC ajmera_h_Page_272.txt
16a2746928c0389faaa74de85fc92b8c
141a1a02f37a3c8412389ec6dfe5d553bd687ed5
1805 F20101129_AAAMZU ajmera_h_Page_023.txt
2e043875d791b0abd3c93be71791624b
b2970afbc725422a0c6df3ee7b28692d4ff9dbbb
42515 F20101129_AAALPP ajmera_h_Page_263.jpg
15a01f4ee11e3d947b5afd20da80f4f9
dc1a46bd77790be9f76757147aa6f529990ccb0c
F20101129_AAAMMJ ajmera_h_Page_266.tif
c2eac4896e643b0a9d59ad42e72e64c0
f4ebb9aee67056e8967bb31d43f11f89358cb0b1
9008 F20101129_AAANWO ajmera_h_Page_165thm.jpg
a88ed0d6d96680449111a192d96c2a62
899289014e79f69458c11041894cf0103fb90a7f
2060 F20101129_AAANJD ajmera_h_Page_273.txt
5ebe9c65977d232373d5691e5c349198
7bb18bf0ea851002a2ff7d442732b094e2166487
1890 F20101129_AAAMZV ajmera_h_Page_024.txt
ba4e0bff20ed8caa12b24ff7d86aa422
abe1cf51d3aa04d4319e13fc18830c08002d9bf7
57270 F20101129_AAALPQ ajmera_h_Page_264.jpg
492b4dd79b6df885535a137653569dfc
dbcebde3f9502eabe1e73803e0956c589657e5e9
F20101129_AAAMMK ajmera_h_Page_267.tif
1258bcf2fd256d1aad7e1ef4b98d0c01
92b4ead425a252b769cc1ece095c498b2fb5943a
37228 F20101129_AAANWP ajmera_h_Page_165.QC.jpg
5708f4e41766f55101369b8c1ebb17e1
848613107c6a80eab88c501d09c2472075cdfe03
2127 F20101129_AAANJE ajmera_h_Page_274.txt
aecb2881d8eeaefcc53192396c878c4e
727b8e9d4f862369ea9da1a1ea893aa49e653fab
2199 F20101129_AAAMZW ajmera_h_Page_025.txt
f4f00b7e47c312abd158d2d4bc23f068
d8f71c890e15dc5567c28be1e8b5fe37efae2fc2
42516 F20101129_AAALPR ajmera_h_Page_265.jpg
22a76b17e68ec6b80be39f9b8fbb8b69
6235103d794e2b194c180bf512ada7be3c4df92a
8919 F20101129_AAANWQ ajmera_h_Page_166thm.jpg
302e06e273696efb47459585a8ee7e14
aea3aacf675c4885c9676677be88b73e9f6ca968
2098 F20101129_AAAMZX ajmera_h_Page_026.txt
a973e5c8942c6474bccd1825054a2ce6
dcaa4b56ddde148dcd6ae123118a539585d748d0
48124 F20101129_AAALPS ajmera_h_Page_266.jpg
a84c1e2f7a0a1cc262da6e8f06df80b1
44bd65237d6ae8be4a8c87a2a02c6bcbb466ad78
F20101129_AAAMML ajmera_h_Page_268.tif
f91c4ded85ddcf367c57ece6b7fcdeae
7d611792a28650f3e0c20ab9b9e667999c9eaa71
2211 F20101129_AAANJF ajmera_h_Page_275.txt
d3b252993bde5d8b8e19c1d608b04f53
ded40da9a4d116c2ede2ac3949227aa8f14a7d9c
37064 F20101129_AAANWR ajmera_h_Page_166.QC.jpg
7ea020949df6c5e201142d43054055bc
0b57f61ac1c88a130df9e109cb57a05120291a11
51267 F20101129_AAALPT ajmera_h_Page_267.jpg
681f4efbebb89322d860bd2b8bf8bc24
bd58e123bfc1e13be9f8d3e1a306ddb3efaccbd8
F20101129_AAAMMM ajmera_h_Page_269.tif
e93db2aa239ff8948ef35804ebfbea90
3d0fb15d7cf85fdc93fe80595e1866f38cbfde0c
874 F20101129_AAANJG ajmera_h_Page_276.txt
0faf6f3973d80adc5cba96eef699e143
e823e565e1a87cdd58ebf720a9835a9f17f1cae1
8502 F20101129_AAAOGA ajmera_h_Page_292thm.jpg
aad4dfd0ec9b0f7c02cf0654c3682de7
6bf17f1893a009e75ac89f4ef1eb627e6f3b7d93
9091 F20101129_AAANWS ajmera_h_Page_167thm.jpg
776f5fb04b5e063bf13f4ad80388fa43
a7d83d6fe496bff6879ededb16540eb8b94124b1
2136 F20101129_AAAMZY ajmera_h_Page_027.txt
c74ef9a09ee58966e7c13ff6cea11701
9735a9b7d4ee7db2e02b7698ecff4ad094e95cfd
52121 F20101129_AAALPU ajmera_h_Page_269.jpg
b132fb466125d291d68821f88640e064
9f69a9276ea75d663674f95f88b2ec7318d0402b
F20101129_AAAMMN ajmera_h_Page_270.tif
d11df1d4b3e58ae7f54653facb3e00ff
485e5e5735f007eff21194b1f20d765000dccc24
32473 F20101129_AAAOGB ajmera_h_Page_292.QC.jpg
55af96055eebddded86c51b01b2e2405
720bceedba42c80d6fcd6f35b5f88a6468277872
37411 F20101129_AAANWT ajmera_h_Page_167.QC.jpg
5371326e3526dbf0104c78017a11ab8f
5eaec8e43ddeca551014c49d069bbc64269cc5c9
2225 F20101129_AAAMZZ ajmera_h_Page_028.txt
7793dc2fd08e2d33093d5616e01bfb53
b0e87e5b7921d041625d205c89e845cb6842fde9
35307 F20101129_AAALPV ajmera_h_Page_270.jpg
287eaf86346b7bc0c3bd756046b45e5c
d2b24b2396ff0120390bf4e3565eb617422ca622
F20101129_AAAMMO ajmera_h_Page_271.tif
c5e212a13d705469e2333b0a221219ab
a224ef1971902d4d8f8032813facda7224bf2247
1758 F20101129_AAANJH ajmera_h_Page_277.txt
82391bd4cabc9b88b7c5fcd7a57cfe92
ebb9a819c0be464b9c8c41ebcb10739c1a888dd4
8835 F20101129_AAAOGC ajmera_h_Page_293thm.jpg
d2968a8405442e95c117281d7a9ab766
967faf1724aa07a9668daf3056883a40b6cb3081
34677 F20101129_AAALPW ajmera_h_Page_271.jpg
0ee3e944f08fa000eafa49b838c9d3d0
4b1a5040aa05c529d1d9cec85eb1f16ad464a61c
F20101129_AAAMMP ajmera_h_Page_272.tif
35de805a720ee78b7add73313126bb07
6f18ed33e5252379d86b5610e7472abdfa467ede
1519 F20101129_AAANJI ajmera_h_Page_278.txt
8b78ba01757037fb4e695d8c70ebc262
5040bbd60693b48ca2e13e17f943e8ab35766dd1
9252 F20101129_AAANWU ajmera_h_Page_168thm.jpg
dc1f3b09cf17bc4144a54b53270ff9b2
07ff8d1c4e7436ef2e53f665e0c80bf3a193c3f8
28107 F20101129_AAALPX ajmera_h_Page_272.jpg
c74702e5e6d9c273e6d05787a8b27fd0
a91c387140d81b30ef663f09d2bf747613bf4815
F20101129_AAAMMQ ajmera_h_Page_273.tif
13ba2ecf2b60b1c7a25e9a50b36f3013
1cac98133417c7707034b40c49786f2f81de46fe
1773 F20101129_AAANJJ ajmera_h_Page_279.txt
60813219d0828fe0df940bc09915d78f
2165a77d5cb73ed2e3edc1ea47d256956fb2e365
33982 F20101129_AAAOGD ajmera_h_Page_293.QC.jpg
9a05deb6057e6acdd5f65cb156bc3e11
95a644d8907a6137ab3c9a434d1bee1e59a0171b
37769 F20101129_AAANWV ajmera_h_Page_168.QC.jpg
fb76e9daa445a905f84f26fa782e8ebc
1dbb2e1ca80c6c6fe51bf2b73c1b09dd1d519399
100663 F20101129_AAALPY ajmera_h_Page_273.jpg
14a10939a66fb9ac14de3e24699bf02a
6fc8b6bc0625076adbe9e087bde677b0bda8a2c8
F20101129_AAAMMR ajmera_h_Page_274.tif
27a9ea107410789496a72236fee2c4d9
1e9e744c9c1ffac65504fb77dbe8d8eaad6ccb82
1975 F20101129_AAANJK ajmera_h_Page_280.txt
4cb414394b96dcb96724a22176ca80db
d3b834873c32b618104ed719e70ea2be71a24f50
8709 F20101129_AAAOGE ajmera_h_Page_294thm.jpg
2fe6b8701ff174503e9018d59fed9197
3c50327f62b4bbacd6dc37aed0dce66cacfcee8b
9014 F20101129_AAANWW ajmera_h_Page_169thm.jpg
50338510447265d959847c0c7b594d35
57a21c2a78dbd84341c6d967300b6815f1a8a83c
108749 F20101129_AAALPZ ajmera_h_Page_274.jpg
3c13443eefafa0551ed49d653d89d198
d393c6f96eff401bcce1ae671b53ea83e5556498
F20101129_AAAMMS ajmera_h_Page_275.tif
a78982e2ba5cb8698daee9f01a1bd90e
b74e5129adc1ba9077306f9352666f3d5c62fc3f
1934 F20101129_AAANJL ajmera_h_Page_281.txt
1df5558ff5244c2edf4d3bc73e4cdb6d
f17024a32e089a0441bf7137f7bf51e257451690
31326 F20101129_AAAOGF ajmera_h_Page_294.QC.jpg
ddb880b3c6a691824a86013323dbab4c
ab29af77f528f9c4daea772a01d141150afe18bd
36729 F20101129_AAANWX ajmera_h_Page_169.QC.jpg
3b51b6c7c549dc166e8ef1d27d04411c
f5bf54f14cac8e90d41e5f7e6e7815a2c2885c71
F20101129_AAAMMT ajmera_h_Page_276.tif
9dfb776d905f1466dbaaa95a1a51a70e
d27f5777805ebd5d4986d416786219b58d0cc31c
F20101129_AAANJM ajmera_h_Page_282.txt
39d7d31b6082745b2e7a30baadbe20b2
ac6e2508406aa6cf6c99763243e4f5ecc0454a9b
8332 F20101129_AAAOGG ajmera_h_Page_295thm.jpg
735e38c9bc94a3666f9a6b7862d86d82
872617836a7b8e9f775ad8d9ead1365a2978baad
9085 F20101129_AAANWY ajmera_h_Page_170thm.jpg
f898c9d6fd424ca6125106d7ef780136
6d078943d99fed91a57da4cd214c96c70f2d2d61
F20101129_AAAMMU ajmera_h_Page_277.tif
b01b9061ddc625022d7afebf716b376b
3ece51bbe64ad2abfa1fa7d3f1e8142cbed7cb83
2007 F20101129_AAANJN ajmera_h_Page_283.txt
68e3288cc8c6912996189c3dfc296fd3
7e536d467c78c9661ac25d45fecbad130d3ac1c3
31140 F20101129_AAAOGH ajmera_h_Page_295.QC.jpg
bd1b906a3527533f232196e1a6caa766
2c8314ede90b8e6f1e5f74d6dbaf21faf7ab2595
38605 F20101129_AAANWZ ajmera_h_Page_170.QC.jpg
4183a7e05035e05ed7944fe69265276f
5d80e0aa8c748fe6512879a39e9492927b98f0b2
F20101129_AAAMMV ajmera_h_Page_278.tif
7cc274101322b730ef3181c7d36d62df
e19010ee6d7aabe61bde9400777d7d8e6ce4e0ce
2083 F20101129_AAANJO ajmera_h_Page_284.txt
c9f88da36308aff5e09bd22e808735cc
b3651c374f237639a268a5dca9f608a9c384e3ff
8802 F20101129_AAAOGI ajmera_h_Page_296thm.jpg
c3e70da625e9e9179c2f599ce968a47e
2c1d272b48a6a025597cba13ea263567c8396293
F20101129_AAAMMW ajmera_h_Page_279.tif
bb46c0b37c1b9bb7bf9bbb0928deef19
733794bb31ade98a92aa1573bfca571d65ee7816
209 F20101129_AAANJP ajmera_h_Page_285.txt
3af1f8ca3ab9e2bb22edf804c911fa83
4f460de003ef7d4cf27949490681888101a2d472
33794 F20101129_AAAOGJ ajmera_h_Page_296.QC.jpg
ff3db23bff71aba7001b4358f0db3bcf
a732c7789d65c25b6d846b43cc6c3a3e9e9ca02b
F20101129_AAAMMX ajmera_h_Page_280.tif
8478db360596f0b29f3178506a79ed63
d91267c9d0a6a963cfd0a3d554401ee33dde6438
1863 F20101129_AAANJQ ajmera_h_Page_286.txt
a55b015c7d72f61c1b0f8dd5bd433dbe
00a24d0095e1a5182249974105b38ef14fca0c79
8185 F20101129_AAAOGK ajmera_h_Page_297thm.jpg
0b048d9df98ef0b579fbcf9ce7b2a790
09a01f3203e639da9e930104d64748387a372c03
F20101129_AAAMMY ajmera_h_Page_281.tif
750467a96b6aa6e41ca548d32662cfdf
07a26c64c518e60490db3b67a11fe925d66608c3
2203 F20101129_AAANJR ajmera_h_Page_287.txt
3b10a2f209dd53997c00e6ee4f6cb617
02cdffaa62f1e49321d22365d29e9bb26e83a978
1620 F20101129_AAAOGL ajmera_h_Page_298thm.jpg
a373e6443cbe379a45356f55873ed02e
82cb7971fdabb46f6939cf2437831057980d0f39
F20101129_AAAMMZ ajmera_h_Page_282.tif
83f95843d089e688225023984cd0f683
361ebde36b05ad4a2ec5de188203177ce3926ad4
1755 F20101129_AAANJS ajmera_h_Page_288.txt
734fbd5b69db10abe89f81f7d92c5590
5be8661b026b07d7f1d5db05e8f864e3db33184c
5425 F20101129_AAAOGM ajmera_h_Page_298.QC.jpg
716e12256e4a909b4299a67dd119c3ba
687484e320925a585c87aca5e4bad293950e9956
1912 F20101129_AAANJT ajmera_h_Page_289.txt
eb01cf5e3cb0943a12018acf8aafe4c6
ec8146442394ed2851e2ca67890fa7083979de96
4075 F20101129_AAAOGN ajmera_h_Page_299thm.jpg
6756d532a14b9d407e4a3e664d8140cf
c28ae6f74218ebe191b2179efb082c507a934c38
216698 F20101129_AAALVA ajmera_h_Page_109.jp2
19911c829c7d1ec7cb6baff06cd9a7b8
db830a57942e16741e37d4bd48704e23efbc065a
1929 F20101129_AAANJU ajmera_h_Page_290.txt
fe27126bffb697e73dbe0e7ab29ad448
af7a46bd332c162fad4c7289aaea7ff68cc92643
15897 F20101129_AAAOGO ajmera_h_Page_299.QC.jpg
8985d53a2a3401c9f763dda709a4904e
e2f4b58337c90b7b7ca86f3cb46c53f4d574aede
192878 F20101129_AAALVB ajmera_h_Page_110.jp2
a32a197e8be4228192839fda082aa3b9
2042de9c67751df9c04fdcf40f4adfdf9268fb64
1927 F20101129_AAANJV ajmera_h_Page_291.txt
649c34fb98a734a21546e62aaa385326
f7cf5951404a60e4b7aa89b1836271804c889766
342084 F20101129_AAAOGP UFE0021186_00001.mets FULL
a3098ef88af79d53b9507f55339f9408
56252f33f8d14a002070863edae791b61cb0ae36
108791 F20101129_AAALVC ajmera_h_Page_111.jp2
b3b548dff86b4acf96bb760ca143ee3c
04e573079eaf9020e6529e6f50558e5e16a8e380
2032 F20101129_AAANJW ajmera_h_Page_292.txt
f28f29c4287c39bbad3015c4e501be06
ce6976bc53f5c63a0781f13639449a0b5013e53f
118584 F20101129_AAALVD ajmera_h_Page_112.jp2
2c295d92d1e37310577fe311da8799a1
e427307c0cb01becc1844f80cf36a8d13d568541
2058 F20101129_AAANJX ajmera_h_Page_293.txt
00c11d89f6dbd16cb4476fcc051fa8b4
e91781fcdfba4947ae804c905c0d321317e49b3e
114138 F20101129_AAALVE ajmera_h_Page_113.jp2
c16aa42a5592a3809cf88d8a94e08925
8159eb86ffd65de5298338b68e32e858fa816108
F20101129_AAANJY ajmera_h_Page_294.txt
e02c21f0bd9393fbbb8f0b4404605a8e
cbc32fb82b7bf24d7e3f7041ea2437f648bcee5f
101555 F20101129_AAALVF ajmera_h_Page_114.jp2
9e33519db32548fe88a22601713bc362
b3f80f282f88c29640b7bc4e75f32c02f8240300
1840 F20101129_AAANJZ ajmera_h_Page_295.txt
ef4364eeb18dbfef6baedb79a96046f7
2ca17ee587a84c84ceba79baf5a0a4431b33528c
72922 F20101129_AAALVG ajmera_h_Page_115.jp2
bf88f030bfc543603f454b2a0bcc9477
9f1c8f97347139e66f31c8aed0d88581ddcbbd30
77865 F20101129_AAALVH ajmera_h_Page_116.jp2
201bec0128af7194bfccf57259db1c49
c392105375b61d161fd48416eca535dd99937d4a
32978 F20101129_AAAMSA ajmera_h_Page_115.pro
a940dce6e594ff0e00c48c3434c2732e
5538aff9c58e6664dcae0fd6597df76512828199
107129 F20101129_AAALVI ajmera_h_Page_117.jp2
c625617aabf91c0bc60ed87cb779b02b
bf4c6e1d0ab54d3789c29bba9f6e26d0be755f61
34899 F20101129_AAAMSB ajmera_h_Page_116.pro
0a9f2d77cb89c794fe8333adec43de3e
b9520e56e6d5d81c33f854d18baaed6951bd3cff
101188 F20101129_AAALVJ ajmera_h_Page_118.jp2
bdf9ffc229bc4b863c0d063a85336e76
f32377c8ecbd6b21ec68bed2cbe51362980a5f83
49450 F20101129_AAAMSC ajmera_h_Page_117.pro
55dbf6e52a15100b18532428e88f4227
6fac14315a29d40470dd78031c695027e131e2c0
113162 F20101129_AAALVK ajmera_h_Page_119.jp2
33ac412cf1e2917eb811ab3657fbeca5
68cde265d8ce08bb31b834ed4ff7cef0dc35704f
46619 F20101129_AAAMSD ajmera_h_Page_118.pro
c18cb2a35bc512c249dea09080824ee4
24180fddd60e644a7fcc2d638ec08483afd2d11a
74733 F20101129_AAALVL ajmera_h_Page_120.jp2
e3b4e3ada65d35c2edfa667686f1b012
73e6558d3d55875815816e7d02f49adf4e757991
52073 F20101129_AAAMSE ajmera_h_Page_119.pro
0325a873f67e8f19e0757ccb89d89f2a
1be2edca145350945120fb67fdadbcc7092366d4
116189 F20101129_AAALIA ajmera_h_Page_063.jpg
369b499714dcbdca31bf5819249e3c99
c85c89445923ea2996f78090a814dc5a91fe6836
105014 F20101129_AAALVM ajmera_h_Page_121.jp2
469ac6a0ae836bcaabd0137c9d406a00
0d8f5ebf292f73b68b000616d5fa1462c7e98522
33444 F20101129_AAAMSF ajmera_h_Page_120.pro
725f2f0b6ecf2af8a6ea3218981a11fd
e6a7b867e20fbaf04bdfce16278a845474693771
115545 F20101129_AAALIB ajmera_h_Page_064.jpg
86965f8752efa74b01ca3295c20c36ca
741003decccfdb382caeacbb292ebe4b00307019
98251 F20101129_AAALVN ajmera_h_Page_122.jp2
d5382d57c2e393dbbf0d842f1eeba83d
517ccf110b5bcee2837c2f22c809b308a85da435
48490 F20101129_AAAMSG ajmera_h_Page_121.pro
01fb447b810bea34465829e7a69fbcfc
4d9c3b484fc5605b9bd491cb6f3946fcd404e30d
115801 F20101129_AAALIC ajmera_h_Page_065.jpg
deae3be2df9fa701f0524dac6430059b
a98f673d9fcd680165ae4bac4e64f450b161d401
113892 F20101129_AAALVO ajmera_h_Page_123.jp2
b68c76225dcd192c1ba576d580fc229d
7e7da579b3a6a46bf268c06c70c42a0473ed58c6
46776 F20101129_AAAMSH ajmera_h_Page_122.pro
04a492d10f0ed9aa5d37a6c1e8bd869d
50c1310cb8aaa83e1e19036a7a342ef406a6c2af
9057 F20101129_AAANPA ajmera_h_Page_064thm.jpg
621f712810d19739a61f4c06a4868176
718c8879e9995d19f4d748962cf2dc8d955631a4
107251 F20101129_AAALID ajmera_h_Page_066.jpg
4e8eaa739cf42a8b284eddfc6c180d15
c542a3149d80b1d9ed29953dc39ad791ef2c8d98
114121 F20101129_AAALVP ajmera_h_Page_124.jp2
63faf687a6080f8cfd427fb4b92b39ac
87e1f18a122189c53144c5df77621f3571dcb603
53780 F20101129_AAAMSI ajmera_h_Page_123.pro
dbf2a551ad474a7a01303ef1efebe717
04ba0cf4df019ccf76c3dd2c907a6a8fe7c35bee
38310 F20101129_AAANPB ajmera_h_Page_064.QC.jpg
0316183844ee7d347e7f786003921b9f
174e757b0baa4882005296e7e8e0a59e95318591
113023 F20101129_AAALIE ajmera_h_Page_067.jpg
5e31cc3f7835ebf90efbc2ad0165a7d1
2e734d542752c13a420f765d46977e0f7077dc31
94803 F20101129_AAALVQ ajmera_h_Page_125.jp2
5e486f44236c7d8ba411c44638d49ab7
4788bd793fe67f22237860cc62f6e78f32e3ad8a
52139 F20101129_AAAMSJ ajmera_h_Page_124.pro
0a94a390b531285f59dd16cd3cba8d5c
e27df3e52276b6b480e4a9ea9a95ee14a744a6c5
9239 F20101129_AAANPC ajmera_h_Page_065thm.jpg
353d43d72afb50a6af21c04d8716c5fe
8935999cb8b5084828f865346a795e08b09706e1
114340 F20101129_AAALIF ajmera_h_Page_068.jpg
67e43b5974bdb5ecaa2de2208412a943
b61ac8294f034fe83f6a7ed2fc65e7d630087e0f
79069 F20101129_AAALVR ajmera_h_Page_126.jp2
41e04a18656984bb2b2e8f7801bccd71
99d5de041b744d0f0fda9a2bdb68202ad2604242
44028 F20101129_AAAMSK ajmera_h_Page_125.pro
b973348b422e1c0fd105df6c139e05e5
61130638146076d3cd9f5c19311044f47db2b7c7
37772 F20101129_AAANPD ajmera_h_Page_065.QC.jpg
ff93af05b6b583952b006849bfbc5c8a
0dd46bb1facd7ada66e27c119962b07877fce6eb
107672 F20101129_AAALIG ajmera_h_Page_069.jpg
f445115c21e0fdaec40b72a3f0896aa4
c21bda1389154867592ee90900d16bcadab9a6fc
F20101129_AAAMFA ajmera_h_Page_070.tif
37ee75688fae1a4c3a8008f7357c92f7
86771576caac58f6bdb9cec3aef79583c78e5512
77750 F20101129_AAALVS ajmera_h_Page_127.jp2
20bfb720783676aa73e6a1f3befd6949
9e26f61c65977e5126fe8e0ed5c3cfff9d013799
36499 F20101129_AAAMSL ajmera_h_Page_126.pro
2c2f20a4e32558c323ea9cd746af7b51
8b605d7bb6913c018767dccbd47638491fb0f5b8
8822 F20101129_AAANPE ajmera_h_Page_066thm.jpg
d3336bb8be4fbdc71f9eeb102a4604b9
f287ff9a65b5b4ae66e66a02feddcbea0622ba2c
111009 F20101129_AAALIH ajmera_h_Page_070.jpg
0e993f2d510105ed4a14a4d05cc81688
7e89fb0ebd714836c67f5d95261b70f82d9131cf
F20101129_AAAMFB ajmera_h_Page_071.tif
6e7b69dc7d59dcfc6ad9e5ab6e5b649a
bed26afdd2c458e545f31a85eeedcb67a89f20f9
71798 F20101129_AAALVT ajmera_h_Page_128.jp2
21f8202219e231e8a329e6e0e387e4e5
5c506877b4d86552129499a6e8c6779c01a4d3cc
30265 F20101129_AAAMSM ajmera_h_Page_128.pro
193beb8c24a1b01d259c017840213451
6634764e7f8e0fb99cb2dbad8b23b5ad1115ca16
34825 F20101129_AAANPF ajmera_h_Page_066.QC.jpg
af0c798c7b9b48d25adffc6c8eecf6dc
308c7374885a4f16f87ea50a004152be035fa817
F20101129_AAAMFC ajmera_h_Page_072.tif
bf848d935f7c3f57dfdb258b81104694
466e3429a8d71653970cebe6cf7fa1550eb450f7
54984 F20101129_AAALVU ajmera_h_Page_129.jp2
5e573e503d30f36913cda2b90eae929c
51fa3bfd45d87eb756c7bd49f46f54910251ecbd
21738 F20101129_AAAMSN ajmera_h_Page_129.pro
fe7630507bc07a9c0a10697883577c11
fb35101a6780cfb168e7541657c811f049b0d9f1
9082 F20101129_AAANPG ajmera_h_Page_067thm.jpg
b0f387b8abf6311f5be5db5d84b68ab0
fefe0fff71a4c42affb7ad18774211ed77ae9c8f
110523 F20101129_AAALII ajmera_h_Page_071.jpg
c603493045be3c0585788de522587ad0
eec16a6e2c67e3d8eb96072c2a9d6c5228256dde
F20101129_AAAMFD ajmera_h_Page_073.tif
cbadec42c66bfaf199fb1e0d814d66f4
4a577e10bff8c6a1b7bdd7fc4edb9513e151a291
17823 F20101129_AAAMSO ajmera_h_Page_130.pro
7cbc74a256d72095c47c316a82d4de5c
cd4720b3d9466281d9b580dfb6c985a5873a36c2
37430 F20101129_AAANPH ajmera_h_Page_067.QC.jpg
7a48a022ec1bced4f4852c6bf2dcaa35
38a0eab156c452d1b524c54b6b508190a7813156
108220 F20101129_AAALIJ ajmera_h_Page_072.jpg
b969a367279e0e5b67d7e5434a4c89eb
fc59ca6b7e35b1f9c3b01abc787e3a8115eca094
722435 F20101129_AAALVV ajmera_h_Page_130.jp2
85c63e84160127d3d78296b5f6121e97
7f230b8ad3d6a1b3a311a6cc3ab403af20fcd711
6722 F20101129_AAAMSP ajmera_h_Page_131.pro
b0d324176db0393e752faf7978eeb3c0
d1a68917b79ad460753f44914b130068fe5fbd13
9179 F20101129_AAANPI ajmera_h_Page_068thm.jpg
664ecbecba2dfc84fbad20c6bb70de53
5abe6eba0a2aefae26ec3bc1a4f3e2c637b80326
107726 F20101129_AAALIK ajmera_h_Page_073.jpg
7ff671c022cbc5baf75aa9267bf7e18f
5b57160a3e152ca6467d3bbcd8a54aee5a8e437c
F20101129_AAAMFE ajmera_h_Page_075.tif
73228b335ed2f76953ab931c474cba93
8333640f671c4b90d7f9347981400ecee7d7871f
238082 F20101129_AAALVW ajmera_h_Page_131.jp2
a380ce3b13d6f6f5e712293b5ca63593
ce5b94251aa50d7a935a1f428a5616f46f547026
5671 F20101129_AAAMSQ ajmera_h_Page_132.pro
3daddfcd974a9b75a666b4f798ae7232
f296051cd1f108f0394ed33e7acba3b43e445572
37667 F20101129_AAANPJ ajmera_h_Page_068.QC.jpg
98a9b36b173c8903a9542a3c367a6165
cd1a55d2037cae619b33040ecf2501ce39e9df4c
108903 F20101129_AAALIL ajmera_h_Page_075.jpg
1e25102d1c3b19fec2383bd43c1d3ddb
ccfd62bd224c129533b750c7b651c265c9f3c2a1
F20101129_AAAMFF ajmera_h_Page_076.tif
8ce5af856ef321bf0d951badb3c37ca0
cbf89a2d0813b9b3170fd9f517d044457d99c7da
251945 F20101129_AAALVX ajmera_h_Page_132.jp2
a7e988b54a471d411ab2c6962d5e367b
a94329710cb90a6e3dc8b396970bc430b03f5433
9144 F20101129_AAANPK ajmera_h_Page_069thm.jpg
2740095688eca2f7776cc74cd6185823
bda8bedfdc04a57511dfa86a2af6fe6bd27d254b
113787 F20101129_AAALIM ajmera_h_Page_076.jpg
cbcff19b8190fab84357072134f0820b
2244f09b08e0430bab1c75f43a01e132ff334d4c
F20101129_AAAMFG ajmera_h_Page_077.tif
2f183956d967c57fcd126b432cbfddc6
57b451e8f0e5d48414b17af7d5b0dd9d58e67e7e
229080 F20101129_AAALVY ajmera_h_Page_133.jp2
9979f136d1ea77c900a054277a3fc5ef
631f174fe30b2a809623a8ffc808ee1d8f852111
3280 F20101129_AAAMSR ajmera_h_Page_133.pro
11a8f85b35a6df0a915f90928101e519
49b0f1d280f69be9a73004561949c4a257fe7e4e
35465 F20101129_AAANPL ajmera_h_Page_069.QC.jpg
ba7cf199d6e4842d77151cb2eaaf3fca
196f3265aff8a654e1252bb4e82886017ba39f78
113834 F20101129_AAALIN ajmera_h_Page_077.jpg
b59f8e58477c8220cd92eadd921a51df
a2d4832ca2c325ed0af3d7e9495cf6b58965ca4f
F20101129_AAAMFH ajmera_h_Page_078.tif
8814c44d8489d83453c3f41f4fbce52e
f1338750b3d6b44a9f30c5aca102d72d3f32a58f
283637 F20101129_AAALVZ ajmera_h_Page_134.jp2
546d4e58a881d867b4f2ea9e57979e80
1f63333143bf40b6c14b851263b79cb9a08974f4
2052 F20101129_AAANCA ajmera_h_Page_083.txt
bed39d19ae8b1087d3d275363a8f154e
8e29ac840042e08332a934363087453ad519909d
2738 F20101129_AAAMSS ajmera_h_Page_134.pro
8ad980e013949920d060060f39d515c8
9dc98327e1e1616bda5a9c905f36309dff082e30
8987 F20101129_AAANPM ajmera_h_Page_070thm.jpg
74732d2f0f74bee5cedfd214f857055b
61877317c82c096cc9f7d7c8d885be4e083e3272
109418 F20101129_AAALIO ajmera_h_Page_078.jpg
395dcc3961f08816d4b045db4fdce8fd
973ef459864139b34eff8fa6ae58710126b8fb01
F20101129_AAAMFI ajmera_h_Page_079.tif
6ff8876157c853163dc4939196fb4095
7e4081f3d3054bce7e48bbdcdfd71118aa83dfc6
2055 F20101129_AAANCB ajmera_h_Page_084.txt
f7e3906c0e0dc5217cbcc70c2c8309de
224ec0017f4ad04fb36ff4470617a000f310b035
52158 F20101129_AAAMST ajmera_h_Page_135.pro
a08da21efcb870fb6f6773c577eaae1e
beca139313f8a408891fe28a1950b2a2ca235267
113580 F20101129_AAALIP ajmera_h_Page_079.jpg
c1b50e8072ae66b40ad1256ef73cf0e7
cf308fd028d8a821688cf30a07c7910c95e6ebec
F20101129_AAAMFJ ajmera_h_Page_080.tif
8aa683e0793f8caedcd5e8c0bb7638e1
aa86cb656ed1b40b33d6850ca8c0cb7d7808fb8b
F20101129_AAANCC ajmera_h_Page_085.txt
2d7801b60f221a0f444d4a2f6663baff
94fed70cd5141dc82772587d2fd88446b92769d4
51503 F20101129_AAAMSU ajmera_h_Page_136.pro
51edab18c3d5e54c6ae5e05bf601b420
3656897b551bec3ccadac496cd3716ddd6b311e3
36334 F20101129_AAANPN ajmera_h_Page_070.QC.jpg
6083dae8afe09195b9c0fa0059f63437
c9ca7926ac835a9647b18136ad8790caccc9192b
117075 F20101129_AAALIQ ajmera_h_Page_080.jpg
7733e6f1ffdaf924822812c030357ebb
b0b3fc6671cb34d47ff1f79826d66b4a5b5e7e57
F20101129_AAAMFK ajmera_h_Page_081.tif
9844dc323dd93709348ef227d0e5ea44
6a4103fc7076525a43541512627cf57f9555102d
2047 F20101129_AAANCD ajmera_h_Page_086.txt
78eebe74c6b3ae8a999e874988af2c66
a02d960f490e0faaa0740c54b0b95b5b90dc881e
54023 F20101129_AAAMSV ajmera_h_Page_137.pro
8170bdd587f7475a8b73a9b1c34ee3be
8dea6f4494210341af0e0ef5d1718709cd1966d9
8581 F20101129_AAANPO ajmera_h_Page_071thm.jpg
658f898ea12f1dbbc55b53e7d65f43e2
1eda2a447b0805ed0e2fbc186108374d459d8c37
110159 F20101129_AAALIR ajmera_h_Page_081.jpg
b53dcd10f75f1a32927a50db1ae1a540
8a8e9189d80136e2024ef49503f5cef058204e73
F20101129_AAAMFL ajmera_h_Page_082.tif
4f08caaf0e8977b1b6740c19fdbb2e19
d2e47d7e4a4434913a60ae780b33c106adfd3b33
2012 F20101129_AAANCE ajmera_h_Page_087.txt
0359b4b7ad4a824b185b2661de0facef
b161667fcbbb25101f486edc5a1d90a0f1757ecc
53844 F20101129_AAAMSW ajmera_h_Page_138.pro
f454aa1574cd74c856d2f03f019f6897
0bb1e9719fbd272af012fcccfaef4c1fd3841109
36159 F20101129_AAANPP ajmera_h_Page_071.QC.jpg
46f7805a2daf06e3ef31afbc9d8fccb9
3b406df4f3fc179b88da5896ab5681c7d8be1c0a
114099 F20101129_AAALIS ajmera_h_Page_082.jpg
5193a36453c51ee6c6f05f25b30995c6
e7453ae2243bebd6f50b73468026047a50964777
F20101129_AAAMFM ajmera_h_Page_083.tif
9a0b98f4c8e877c02bb283719001722e
cadac40b1f93945a230de537a853d763b13b93fe
2168 F20101129_AAANCF ajmera_h_Page_088.txt
936d6bc5fb3b2f0ff2af1e2ac970a6d7
19e1a92f7fca6b338d71a818c1f052d80a1fc5a8
53900 F20101129_AAAMSX ajmera_h_Page_139.pro
05a141d0569754b983846b414b9ce217
df0dec3bdd8b82d1e8684a6d3bdd41648706faec
8623 F20101129_AAANPQ ajmera_h_Page_072thm.jpg
10e1bb9ec46518c2412cc8dfed586028
5e6c103d2d6edcfdee778f0c53f7f3c3a4090de9
108369 F20101129_AAALIT ajmera_h_Page_083.jpg
fc6a41b96bb1f5bb22940fe9bdfaf4d8
406757f200486e26e48b3fc9e383430e502ee938
F20101129_AAAMFN ajmera_h_Page_084.tif
b1a67e08b75825145dda84eed8ba1b35
8d079d086e2e6bf71d072c6e0db3c9e7c06bb767
2175 F20101129_AAANCG ajmera_h_Page_089.txt
0cc8a20767bc690cd95ad65ecb2a512b
a51cf409b754b3a8a6a484c393fd790dfc52159d
50687 F20101129_AAAMSY ajmera_h_Page_140.pro
cb8105f21a6bfca1097ad5430b99ded5
80619b0f08351b643ca68c58038456d1fa7c63bf
35175 F20101129_AAANPR ajmera_h_Page_072.QC.jpg
cc274e2c0156b55b342dc76708731f49
fa1e8d360c0b562af6fb31b630444daeffd28a86
F20101129_AAALIU ajmera_h_Page_084.jpg
3aa1c8896d85bc077a0ded28f64e9bca
b06555927e909094788ea75a034d6ad014507a26
2184 F20101129_AAANCH ajmera_h_Page_090.txt
471bbfb7e9fb37e90e42c8a252877831
b48902e68315fce2b9d8b954e3fdbadc94cdec89
50718 F20101129_AAAMSZ ajmera_h_Page_141.pro
c1c8fe17e99823d230f6250b7c2eb69b
d0596b88fc91fb7939c62b2636c8e9a1b381e9c8
34893 F20101129_AAANPS ajmera_h_Page_073.QC.jpg
5dca41bc4c8105375ad26978f8afbf1d
7dda56662fa5cfd41339a2ccf305262f2520f26a
114740 F20101129_AAALIV ajmera_h_Page_085.jpg
012fbbdea333ec61e90dc03d5d1cea09
a1a7450d4af3c723ec2954861862516f850bc4e7
F20101129_AAAMFO ajmera_h_Page_085.tif
90a3cf1f13d3e1f1404016891f917531
662658513e2ed61f5ab00e2b9d391e129e5dc2e6
2082 F20101129_AAANCI ajmera_h_Page_091.txt
32124cdfad70034356dd8734a89a9bfe
55571f8b71ffbf9c5c3461ce7eb31923e3e1c267
9089 F20101129_AAANPT ajmera_h_Page_074thm.jpg
1f43f1bd9c1a6a018036b14746e372f0
da7df9a5c1e403a66f05e24179fc1953fc99f0f9
107062 F20101129_AAALIW ajmera_h_Page_086.jpg
9a11e354cde784ab89e29fdec5270fd8
a99adede1e2f7cd4c666aec016566d8344e6efd4
F20101129_AAAMFP ajmera_h_Page_086.tif
da56a27d9e5f7a9c9d5d8109b279f0df
1f1030d0f37d725321f2e105eb290ca1093b4a7c
2101 F20101129_AAANCJ ajmera_h_Page_092.txt
1ce0b705d8bd307c47009bc8ed03a9c7
f2182779c46ad4d550f69b10be7d4a6af09a22ea
36400 F20101129_AAANPU ajmera_h_Page_074.QC.jpg
0818eb38160ae0bfda1ae3f9a222e885
e049816b6fdbd7885c03d6efab8d13bad090c899
106794 F20101129_AAALIX ajmera_h_Page_087.jpg
f3d8acd8a1306cc62fd32fa7777ac3f6
128a4136f044a78422989b51884b572cccf82f94
F20101129_AAAMFQ ajmera_h_Page_087.tif
42b303285e884b995de852d810827a5d
0bd3c55a7889d662d15f6a16f0aa4607e1b0810a
2220 F20101129_AAANCK ajmera_h_Page_093.txt
03bb347d811a1a13c07ee2c171937c25
791d005b1fb376cfedae8ef80b13ab51ecc72533
8881 F20101129_AAANPV ajmera_h_Page_075thm.jpg
3376aba4ea8ac0b68c816d42d85d2ba8
04a40ae0a3b5ac92161d7cd70611fef5a25d15eb
113527 F20101129_AAALIY ajmera_h_Page_088.jpg
35c4a0721fe2e7ab6c3c389e10830eed
f5f705b71e6c0e80d833b56d00b19cef45e5ce49
F20101129_AAAMFR ajmera_h_Page_088.tif
df842cf1216fec4bd528935a67e108ff
06fbeb6f0c8f4c3e8a45710a68f36a4a91c4cd14
2165 F20101129_AAANCL ajmera_h_Page_094.txt
7d89f211ac3604f7a01d3ec0ea5153b8
0bd105ffbaee78047c7fcd95025f30b73447dc69
35188 F20101129_AAANPW ajmera_h_Page_075.QC.jpg
96fcfd29195b651b07ec2178c62d467c
1c28fe4f47495b4cb50d7ca21a11dcb5cfb5b850
113137 F20101129_AAALIZ ajmera_h_Page_089.jpg
9843994f5665acf6d8582b2d925e57bd
cd7df688b51547df8017f343d03c258ec5391066
F20101129_AAAMFS ajmera_h_Page_089.tif
68d4e8161f29db8429e667c5384d8bd0
4aa2985252a32de4738a3f29eb401b720b9685b9
2125 F20101129_AAANCM ajmera_h_Page_095.txt
bfd2e374b4843c0c3bf13552a65035f7
9f3eb44236c5cf5331113f33e1b8390e5410dd84
36986 F20101129_AAANPX ajmera_h_Page_076.QC.jpg
bfced237e4c7ba1b13440fd19d9f705d
d7014aaeda7e7c8f58299f3cd8d8e1ebbf4e6e4c
F20101129_AAAMFT ajmera_h_Page_090.tif
f1af9258181b38201274a21e8159a9e8
e753de0ff11a1871ab82b2c2b30fe75ea072f3c7
F20101129_AAANCN ajmera_h_Page_096.txt
23215827d75e9416d7b5c62e80846ece
d5a780c76c5f26cb4005026be667d013cb335caf
9214 F20101129_AAANPY ajmera_h_Page_077thm.jpg
b0f329301dd8ca22657600ccd364ed0c
08fc6c4ca173d5c3e26f7f5913600da1c1c0379e
F20101129_AAAMFU ajmera_h_Page_091.tif
22fc05937a7c1576310b76597b45a19b
ee7f6622744dc8f29674398c75159385e8931e23
2113 F20101129_AAANCO ajmera_h_Page_097.txt
a84ab4c568869401f2dd52ceab505ac8
b554b6b416a5d9056e17e46ab233cda5fd85c58e
37448 F20101129_AAANPZ ajmera_h_Page_077.QC.jpg
8f1d464430a9ca31ac4d3318e599a453
176ae719fa75e41cc2ec55064b67dd11113ac371
F20101129_AAAMFV ajmera_h_Page_092.tif
df604da9eb810f3fa45671957a6d8fbf
42cefef5bcdc4e7b04f012b14ea4e28fe606bd83
2112 F20101129_AAANCP ajmera_h_Page_098.txt
7bc129c3ebb17c46455c0c94e45ad454
3d9bc655bc769387c436ec9b7b366000166e4382
F20101129_AAAMFW ajmera_h_Page_094.tif
71b3a87aea39497a4897caa4edace7bf
696d0b270b74ead8b7bb2fdf6953f80d4f58a5e9
55698 F20101129_AAAMYA ajmera_h_Page_275.pro
1efce66269ca566178c83605fe5ea79b
cbb45770aaddcc25ba06f1b3cfb33f8dd9f0c74a
2150 F20101129_AAANCQ ajmera_h_Page_099.txt
dba35d29ac28a67c64dd4025af9e66ff
358f0fbbc14e10cd78ecba8ae50614a7a49942d8
F20101129_AAAMFX ajmera_h_Page_095.tif
629998b9be372cc6401dc927347a4fea
b2b86a00a6af2ef08f13f44736d92df36a58de63
22060 F20101129_AAAMYB ajmera_h_Page_276.pro
d5903de15a5c9b1ca8e944806a2dfef5
0f0f2bf4cca93caed20650088b62f17590d4fd0d
2178 F20101129_AAANCR ajmera_h_Page_100.txt
43ab654863065fab1e8a83b0e29b73c6
844cd3467c509422ac7b137f937e5e9e506ace98
F20101129_AAAMFY ajmera_h_Page_096.tif
f6120b71b2cfd9eb2731bd6e2c6b52ec
ef79baa9dc90fe498aef9b9e2767f9bff9dd46bc
41712 F20101129_AAAMYC ajmera_h_Page_277.pro
767b215ebcf2d5a8c229e0a689a8038a
aca0526a3245746fc11ef126bae94e1388df6e61
2198 F20101129_AAANCS ajmera_h_Page_101.txt
e41f1d6f1e6cedd73b050718bae4b94a
d5032e1e9d77cb933d0d4db5c72e1f6ac62d8a9c
F20101129_AAAMFZ ajmera_h_Page_097.tif
83740d738f17e8257cc017fe9446ef82
b5a282185fc63e57266c2b6d120d926ba68fd7cf
34267 F20101129_AAAMYD ajmera_h_Page_278.pro
4666b0e74545004464004b2429c55605
73218d8132e0491b5de962d87e8da21b02e9c150
1601 F20101129_AAANCT ajmera_h_Page_103.txt
9cdef1855abeed4203afd63a2648f95d
6b7164cd0cae485d02ff133f1cd4ccacf65dc301
41521 F20101129_AAAMYE ajmera_h_Page_279.pro
f146bc8ec3134a6e6e1060fdd3d74f6e
4e6da96014e60ba481a1a07726fd0f4a9e80b94b
1592 F20101129_AAANCU ajmera_h_Page_104.txt
8a62d3815784351003075ab8a1c57cda
8a9189d54dba68ec5e99b9c09a86917d9f0276b4
44394 F20101129_AAALOA ajmera_h_Page_222.jpg
b15c6f8af0d54c708305c9ed1ff22601
de5efc3851300d4faa29b635933e59f796ebdf70
40995 F20101129_AAAMYF ajmera_h_Page_280.pro
b84981d22fa30e14d9a7f15a383aa27b
65f7de6dc1aa57d8f4a0209345e24a194f0b6262
888 F20101129_AAANCV ajmera_h_Page_105.txt
7994f5465aef899ede3b87e190695c55
6efad85e9a489ee6f2a35180fd2e18d4822705ac
66217 F20101129_AAALOB ajmera_h_Page_223.jpg
202add46f8d11550ba18fa1140cb9a95
80d044e3e882b356c02c1ddab728d35298e38035
46984 F20101129_AAAMYG ajmera_h_Page_281.pro
f81f390d8eaf5b12e511b24ebf249e02
04a992c89dcca02b886d7986e775ff3458c1d25c
1299 F20101129_AAANCW ajmera_h_Page_106.txt
23d05f109857fb9e0cd1680c33ed8a4f
d1dccacd1a2b7046d007325ee14dfad965605afe
73009 F20101129_AAALOC ajmera_h_Page_224.jpg
40ba0db41cb737f69e39e8e0ff817201
858e6452ee701064bc832af73893baac3cd06b02
6725 F20101129_AAANVA ajmera_h_Page_145thm.jpg
ccf4314677c76106a5004f1375c4100c
14a9ff923c603fbce7695049e0e698408ccaa1ac
50887 F20101129_AAAMYH ajmera_h_Page_282.pro
cbd20c26702b47473d5571c3ad2a780f
1ba8cec6df7fc1fc7b86b66a839b0c5a3f4edad0
F20101129_AAANCX ajmera_h_Page_107.txt
5f37e0eebd2963f80261573f4f8566f3
0caf92421b9b4e505e98120b51a869a05f131bfa
39574 F20101129_AAALOD ajmera_h_Page_225.jpg
1b71b442e417be4d86375ace74798f34
b6dae9da15efb6c4ecd4bcf1716bbbc39697946e
27469 F20101129_AAANVB ajmera_h_Page_145.QC.jpg
b974385bea838ef4c26baa967693035c
68ff5119bd47249f42a8a61e4e4101f8928aecc1
50708 F20101129_AAAMYI ajmera_h_Page_283.pro
18ad620f39c9dfd724ca2dfb1d517816
5000f47f32de1da11512bed1c4ef20c49ee95aef
320 F20101129_AAANCY ajmera_h_Page_108.txt
a53b5a1f61d7d5bd3826b7c9441c6a31
ab11e54fd34a727c762e1eee44aa9780bd52e6e8
86258 F20101129_AAALOE ajmera_h_Page_226.jpg
5b603ab78faeda3a8fae94e800aad35f
e6a0bc5aed617a5d55e2e15696edfb31c08b81e1
4549 F20101129_AAANVC ajmera_h_Page_146thm.jpg
e6b44ac1a05180541f2b7c0a867243bd
fd876fcc2df284a161f9558c4aa33459cf7aa743
52027 F20101129_AAAMYJ ajmera_h_Page_284.pro
b8751ff65cb26553adc4b1606b5e0f4d
6b6cc8bc5e362b0442bb4656680b096fbe9ea2dc
258 F20101129_AAANCZ ajmera_h_Page_109.txt
4eb302db0360c340162e3e642d46a268
16f341a083406f1a301aaaf04619e46ca8f03f29
53501 F20101129_AAALOF ajmera_h_Page_227.jpg
cfa7d443fd42c23e19b5ab4b9ea38bf9
7f3830d87260a575cb50643f0d82561072f4dea5
17782 F20101129_AAANVD ajmera_h_Page_146.QC.jpg
1d62d649c2ca291e33520e704b25db87
1896b24176fefab1f93964422d7371cbe1f5df24
5213 F20101129_AAAMYK ajmera_h_Page_285.pro
b1b3c960132d18b15dd139ef136c9aff
3223a050c0b2ddc203e6f3747f33ad57b8010c3f
101529 F20101129_AAALOG ajmera_h_Page_228.jpg
533698af376695e5609126241b2646e6
c303904a37d3608295187633cd10501e95e50dac
4950 F20101129_AAANVE ajmera_h_Page_147thm.jpg
0205f73002d1875a741eadc2d9664f34
3580d68a4c54be623579eecc837c76ce184b45ff
46363 F20101129_AAAMYL ajmera_h_Page_286.pro
07825b6d31bde6544c9cdca17a2e175a
befa5a207b08700cc010ae7cbdb582b361f94c9e
61230 F20101129_AAALOH ajmera_h_Page_229.jpg
b4857e6f03c0d3efcb767323e2d78cd4
2dd33e481ec69e18517e3f2410693818b39f7a11
F20101129_AAAMLA ajmera_h_Page_230.tif
27d1d6a296a42f4f4227b9ac584d6505
16b23c7c5d8a9b99b8527a668929db27f5c36830
17364 F20101129_AAANVF ajmera_h_Page_147.QC.jpg
343a3cd6d2d96234fa01845d0a963f60
9d4b0b0d91c0c45f72711518300b5965d25714d6
55482 F20101129_AAAMYM ajmera_h_Page_287.pro
b76a74fcedae2c5620f07672c3d933ba
27b1bad49ab9ea7238ee25fdd84d3509ba1cdf93
34527 F20101129_AAALOI ajmera_h_Page_230.jpg
7ee242eec263d0e7a6846bae3ae1519f
10cdb28a6942f4a2f07d661b5ec4eeb238ae6494
F20101129_AAAMLB ajmera_h_Page_231.tif
48a839ec247b085b5054cc69da00a81d
66a20cb02d597f44f0e48361597ca518af8bba20
4063 F20101129_AAANVG ajmera_h_Page_148thm.jpg
8eae4609daa43bffb6efdac1382e1a65
fa22decbee4c90bd97fa8451059dbde604316848
44171 F20101129_AAAMYN ajmera_h_Page_288.pro
bb4a44da2d49e101b13a8eec86588209
eca63be95eea0d3663c0f2cb43029de635e07c83
103274 F20101129_AAALOJ ajmera_h_Page_231.jpg
affd79b00a25dc4456ac1cc0b8d4d847
d1415c1283bc1ef7a27d62e376618918de4dad5e
F20101129_AAAMLC ajmera_h_Page_232.tif
3c6288adc30dae99a8a2431bb07b8833
87afa4ad7f7ce82774b5868e1b6178d5aef958aa
13664 F20101129_AAANVH ajmera_h_Page_148.QC.jpg
bfd139ae49440ac872b1389f10588c06
d41418928ed18ff4370534f391ebedbb211b3826
47916 F20101129_AAAMYO ajmera_h_Page_289.pro
13fb5258f405915141da02ad349f8423
0dc59900cd537324d79ac37a97e5b14a93932acc
112659 F20101129_AAALOK ajmera_h_Page_232.jpg
33a8ef35d6a5d7a3b9b6c25c0da0ae95
7c649fac6216c34444ebdc41082a85cbc25b3115
F20101129_AAAMLD ajmera_h_Page_23