<%BANNER%>

Novel Synthesis Methods for Preparation of Ceria Abrasives for Chemical Mechanical Planarization Applications in Semicon...

Permanent Link: http://ufdc.ufl.edu/UFE0042340/00001

Material Information

Title: Novel Synthesis Methods for Preparation of Ceria Abrasives for Chemical Mechanical Planarization Applications in Semiconductor Processing
Physical Description: 1 online resource (220 p.)
Language: english
Creator: Oh, Myoung
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: abrasive, ceria, cmp, semiconductor, slurry
Materials Science and Engineering -- Dissertations, Academic -- UF
Genre: Materials Science and Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: As the device design rule decreased, ceria-based slurries have been widely used instead of silica-based slurries in a variety of chemical mechanical planarization (CMP) applications for multilevel integrated circuit (IC) manufacture, since these slurries address many of the important issues resulting from the use of silica-based slurries. However, ceria (CeO2) abrasives usually induce higher scratch level than silica particles due to its cubic crystalline structure, irregular shape, and poor dispersion stability in slurry. Therefore, this article is intended to establish novel synthetic methods of ceria abrasives leading to lower scratch level on wafer surface and ultimately present the direction of CMP abrasive for future technology nodes in order to meet the ever more challenging defectivity requirements. To accomplish these aims, this article introduced the 4 types of novel synthetic methods for the formation of ceria abrasives. The ceria abrasives were synthesized by solution growth method, grain control method, core/shell composite method, and thermal decomposition method. In this investigation, the influences of solvent type and suspension pH on the formation of ceria particles were intensively investigated. The size of ceria particles was controlled by adjusting the reaction parameters of each method without additional mechanical milling and filtration. The relationships between dielectric property of the solvent and morphological properties were also discussed in terms of the supersaturation of solution and electrostatic attraction mechanism. The resultant particles were characterized with field emission scanning electron microscopy (FESEM), transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS), X-ray diffraction (XRD), Thermogravimetric and differential thermal gravimetry (TGA/DTG), Fourier transform infrared (FTIR) spectroscopy, Brunauer-Emmett-Teller (BET), light scattering instruments and zeta potential measurements. In order to investigate the effects of the synthesized ceria abrasives on CMP performance, CMP tests were carried out with the ceria-based slurry formulated by dispersing the synthesized ceria particles with anionic organic polymer. The effects of the synthesized ceria abrasives in CMP slurry were investigated for silicon dioxide and silicon nitride CMP process. The polishing behaviors of ceria abrasives were discussed in terms of morphological properties and mechanical abrasion of the ceria particle. In this CMP evaluation, material removal rate, selectivity, wafer uniformity, and defectivity of the polished wafer were measured by metrology tools, which are used in current integrated circuit (IC) fabrication plants in order to support polishing results obtained by this investigation.
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 Myoung Oh.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Singh, Rajiv K.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-06-30

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042340:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042340/00001

Material Information

Title: Novel Synthesis Methods for Preparation of Ceria Abrasives for Chemical Mechanical Planarization Applications in Semiconductor Processing
Physical Description: 1 online resource (220 p.)
Language: english
Creator: Oh, Myoung
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: abrasive, ceria, cmp, semiconductor, slurry
Materials Science and Engineering -- Dissertations, Academic -- UF
Genre: Materials Science and Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: As the device design rule decreased, ceria-based slurries have been widely used instead of silica-based slurries in a variety of chemical mechanical planarization (CMP) applications for multilevel integrated circuit (IC) manufacture, since these slurries address many of the important issues resulting from the use of silica-based slurries. However, ceria (CeO2) abrasives usually induce higher scratch level than silica particles due to its cubic crystalline structure, irregular shape, and poor dispersion stability in slurry. Therefore, this article is intended to establish novel synthetic methods of ceria abrasives leading to lower scratch level on wafer surface and ultimately present the direction of CMP abrasive for future technology nodes in order to meet the ever more challenging defectivity requirements. To accomplish these aims, this article introduced the 4 types of novel synthetic methods for the formation of ceria abrasives. The ceria abrasives were synthesized by solution growth method, grain control method, core/shell composite method, and thermal decomposition method. In this investigation, the influences of solvent type and suspension pH on the formation of ceria particles were intensively investigated. The size of ceria particles was controlled by adjusting the reaction parameters of each method without additional mechanical milling and filtration. The relationships between dielectric property of the solvent and morphological properties were also discussed in terms of the supersaturation of solution and electrostatic attraction mechanism. The resultant particles were characterized with field emission scanning electron microscopy (FESEM), transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS), X-ray diffraction (XRD), Thermogravimetric and differential thermal gravimetry (TGA/DTG), Fourier transform infrared (FTIR) spectroscopy, Brunauer-Emmett-Teller (BET), light scattering instruments and zeta potential measurements. In order to investigate the effects of the synthesized ceria abrasives on CMP performance, CMP tests were carried out with the ceria-based slurry formulated by dispersing the synthesized ceria particles with anionic organic polymer. The effects of the synthesized ceria abrasives in CMP slurry were investigated for silicon dioxide and silicon nitride CMP process. The polishing behaviors of ceria abrasives were discussed in terms of morphological properties and mechanical abrasion of the ceria particle. In this CMP evaluation, material removal rate, selectivity, wafer uniformity, and defectivity of the polished wafer were measured by metrology tools, which are used in current integrated circuit (IC) fabrication plants in order to support polishing results obtained by this investigation.
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 Myoung Oh.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Singh, Rajiv K.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2011-06-30

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042340:00001


This item has the following downloads:


Full Text

PAGE 1

NOVEL SYNTHESIS METHODS FOR PR EPARATION OF CERIA ABRASIVES FOR CHEMICAL MECHANICAL PLANARIZATION APPLICATIONS IN SEMICONDUCTOR PROCESSING By OH MYOUNG HWAN 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 2010 1

PAGE 2

2010 Oh Myoung Hwan 2

PAGE 3

To my lovely wife, Minyoung and daughter, Yujin 3

PAGE 4

ACKNOWLEDGMENTS I am sincerely appreciative of my adv isors, family members, colleagues, and friends whose support made it possible for me to complete my doctoral studies and research. At this time, I would like to take a moment to acknowledge my professor, Dr. Rajiv K. Singh for challenging and guiding me th rough this research and for always being present as a mentor in sci ence and in life. His support, guidance, caring, and patience allowed me to complete this research and to grow as a scientist. Also, I would like to acknowledge my committee, Dr. Stephen Pearton, Dr. David Norton, Dr. Hassan ElShall, and Dr. Chang-Won Park, for their ad vice and support. In addition, special thanks are extended to the faculty and staff at t he Particle Engineering Research Center and Major Analytical Instrumentation Center, espec ially Dr. Valentine Craciun, Eric Lambers, and Kerry Siebein for always preparing my samples to exact specification and for providing access to their instrumentation. I am deeply indebted to my family member s for all their help which have truly made it possible for me to reach my goals and obtain my dreams. I would like to thank my grandmother, Dalwha Choi, for her life lessons and encouragement. I would especially like to thank father and mot her, Bonggeun Oh, and Jungsoon Yoo, for their great love and full support for me. I would also like to thank my father-in-law and mother-in-law, Soonbong Jang and Hyeonsook Lee, from the bottom of my heart for generous love and encouragement. I thank my brother, Kyounghwan Oh, for his deep affection for Yujin. I extend special thanks to brother-in-law and his family, Junsoo Jang, Yeonok Park, Uin Jang, Soyun Jang. Addi tionally, I would like to acknowledge to president of LG Chem Research Park and Director of Corporate R&D center, Dr. Jin4

PAGE 5

Nyoung Yoo and Dr. Junguk Choi. I could not have done this without their full support and trust. I also thank my past and present group members, Sejin Kim, Taekon Kim, Jaeseok Lee, Sushant Gupta, Aniruddh Kh anna, Balasundaram K annan, Jungbae Lee, and Jinhyung Lee, for their contributions to this research with valuable discussion and my old and present friends, Jinuk Kim, Sanghyun Eom, Chanwoo Lee, Donghyun Kim, Kangtaek Lee, Inkook Jun, Donghwa Lee, Kyeongwon Kim, Sungwon Choi, Jihun Choi, dongjo Oh, Byungwook Lee, Seonhoo Kim, Sangj un Lee, Dongwoo Song, Minki Hong, Myonghwa Lee who not only helped and encouraged me in this research but also made my graduate study years lots of fun in Gai nesville. I would also like to thank Heesung Yoon for his helpful research discussion. I express my most sincere appreciation to my wife and daughter, Minyoung and Yujin, whose endless love, encouragement and support made me the person I am today. 5

PAGE 6

TABLE OF CONTENTS page ACKNOWLEDG MENTS .................................................................................................. 4LIST OF TABLES .......................................................................................................... 11LIST OF FI GURES ........................................................................................................ 12LIST OF ABBR EVIATION S ........................................................................................... 17ABSTRACT ................................................................................................................... 20 CHAPTER 1 INTRODUC TION .................................................................................................... 22Research Rationale ................................................................................................ 22Scope of the Re search ........................................................................................... 242 LITERATURE REVIEW .......................................................................................... 28Chemical Mechanical Pl anarization (CMP) ............................................................. 28CMP Proc ess ................................................................................................... 28CMP Slu rry ....................................................................................................... 29CMP of Dielectrics .................................................................................................. 30Oxide CM P ....................................................................................................... 30Process of ox ide CM P ............................................................................... 30Mechanism of oxide CM P .......................................................................... 31Relationship between particles and wafer .................................................. 31Shallow Trench Isolat ion (STI) CMP ................................................................ 32Process of STI CM P .................................................................................. 32Mechanism of STI CM P ............................................................................. 33Mechanism of silicon dioxide using ceria par ticles ..................................... 34Mechanism of silicon nitride using ceria par ticles ...................................... 35High selectivity ceri a-based slu rry .............................................................. 37Ceria Abra sive ........................................................................................................ 38Advantages of Ceri a Abrasi ve .......................................................................... 38Disadvantages of Ce ria Abrasi ve ..................................................................... 39Key Quality Issues .................................................................................................. 40Removal Rate ................................................................................................... 41Uniformity and Planarity ................................................................................... 42Global Planar ization ......................................................................................... 42Selectiv ity ......................................................................................................... 44Surface Defect ivity ........................................................................................... 45 6

PAGE 7

3 EXPERIMENTAL FACILITY AND PROCEDURES ................................................. 58Introducti on ............................................................................................................. 58Sample Prepar ation .......................................................................................... 58Wafers ....................................................................................................... 58Abrasive pa rticle s ....................................................................................... 59Slurries ....................................................................................................... 60CMP Equipment ...................................................................................................... 60CMP Polis hers .................................................................................................. 60Slurry Delivery System ..................................................................................... 60Characterizati on and Met hod .................................................................................. 61Abrasives .......................................................................................................... 61Microstructure and shape .......................................................................... 61Physical proper ties ..................................................................................... 62Surface and chemical properties ................................................................ 62Slurries ............................................................................................................. 63Electrical potential (Zeta Potent ial) ............................................................ 63Particle size distribut ion ............................................................................. 64Polished Wafer ................................................................................................. 65Film thickness measurem ent ..................................................................... 65Selectivity between silic on dioxide and nitride ........................................... 66Oxide CMP within-wafer nonuniformity (W IWNU) ...................................... 67Defectivity monitoring by wafer defect scattering analysis ......................... 684 NOVEL METHOD TO CONTROL THE SIZE OF SINGLE CRYSTALLINE CERIA PARTICLES BY HYDROTHERMAL METHOD AND ITS CMP PERFORMANCE .................................................................................................... 75Introducti on ............................................................................................................. 75Materials and Methods ............................................................................................ 77Abrasives .......................................................................................................... 77Preparation of sol-type ceria precursor ...................................................... 77Hydrothermal synthesis of ceria parti cles .................................................. 78CMP Evalua tion ................................................................................................ 78Preparation of ceriabased slu rries ............................................................ 78CMP tools and c onsumables ..................................................................... 78Characteriza tion ............................................................................................... 79Abrasives ................................................................................................... 79Ceria-based slurry ...................................................................................... 79CMP perform ance ...................................................................................... 79Results and Discussion ........................................................................................... 80Preparation of Ceria Particles ........................................................................... 80Influence of solvent type on ceri a particle characteristics .......................... 80Effect of the precipitation participa ting anions on nucleation and growth ... 83Effect of hydrothe rmal conditi ons ............................................................... 84CMP Perfo rmance ............................................................................................ 85Ceria abras ives .......................................................................................... 85 7

PAGE 8

Characteristics of ceria abr asive before and after CMP ............................. 87Polishing per formance ............................................................................... 87Conclusi ons ............................................................................................................ 91Synthesis of Ceria Particles by Hydrothe rmal Met hod ..................................... 91CMP Evalua tion ................................................................................................ 915 POLISHING BEHAVIORS OF SPHERICAL CERIA ABRASIV ES ON SILICON DIOXIDE AND SILICON NITRIDE CM P ............................................................... 109Introducti on ........................................................................................................... 109Materials and Methods .......................................................................................... 111Abrasives ........................................................................................................ 111Preparation of as-prepared particles by hydrotherma l method ................ 111Preparation of ceria abrasive particles by solid state reaction (flux method) ................................................................................................ 111CMP Evaluati on .............................................................................................. 112Preparation of ceria-ba sed slurri es .......................................................... 112CMP tools and cons umables ................................................................... 112Characteriza tion ............................................................................................. 113Abrasives ................................................................................................. 113Ceria-based Sl urry ................................................................................... 113Polishing of wafers ................................................................................... 113Results and Discussion ......................................................................................... 114Ceria Abrasives .............................................................................................. 114Morphological pr operties .......................................................................... 114Crystalline stru cture ................................................................................. 114Effects of molten salt and as-prepared par ticle ........................................ 115CMP Evaluati on .............................................................................................. 116Characteristics of ceria abras ives before and a fter CMP ......................... 116Polishing Te st .......................................................................................... 117Conclusion s .......................................................................................................... 120Ceria Abrasives .............................................................................................. 120CMP Perform ance .......................................................................................... 1216 PREPARATION AND CHARACTERISTICS OF THE CERIA COATED SILICA PARTICLES AND ITS CM P PERFORMAN CE ..................................................... 134Introducti on ........................................................................................................... 134Materials and Methods .......................................................................................... 135Abrasives ........................................................................................................ 135Preparation of monodispersed silica parti cles .......................................... 135Preparation of ceria precursors ................................................................ 136Preparation of ceria coat ed silica parti cles ............................................... 137Preparation of Ceriabases Slu rry .................................................................. 138CMP Evaluati on .............................................................................................. 138Characteriza tion ............................................................................................. 139Results and Discussion ......................................................................................... 140 8

PAGE 9

Ceria Coated Silica Particle s .......................................................................... 140Morpholog y .............................................................................................. 140Crystalline phase ..................................................................................... 141XPS spectra of the ceria coating on silica particles .................................. 142Electrokinetic behavior ............................................................................. 143Control of thickness of the ceri a coating on silica particles ...................... 145Size control of the ceria c oating on silica par ticles ................................... 145CMP Evaluati on .............................................................................................. 146Effect of pH .............................................................................................. 146Effect of down pressure ........................................................................... 147Wafer roughness (WIWNU) ..................................................................... 147Conclusion s .......................................................................................................... 148Ceria Coated Silica Particle s .......................................................................... 148CMP Evaluati on .............................................................................................. 1497 SYNTHESIS OF SPHERICAL CERIA PARTICLES BY THERMAL DECOMPOSITION METHOD AND ITS CMP PERFORM ANCE .......................... 167Introducti on ........................................................................................................... 167Materials and Methods .......................................................................................... 168Preparation of Spherical Ceria Abrasives ....................................................... 168CMP Evaluati on .............................................................................................. 169Preparation of ceria-ba sed slurri es .......................................................... 169CMP tools and cons umables ................................................................... 169Characteriza tion ............................................................................................. 170Abrasives ................................................................................................. 170Ceria-based sl urry .................................................................................... 170CMP perform ance .................................................................................... 170Results and Discussion ......................................................................................... 171Properties of Spherical Ceri um Carbonate Pr ecursor .................................... 171Influence of solvent type on particle mor phology ..................................... 171Effect of dielectric const ant on particle mo rphology ................................. 173CMP Materials ................................................................................................ 174Preparation of ceri a abrasives ................................................................. 174Characteristics of ce ria-base slu rry .......................................................... 175CMP Evaluati on .............................................................................................. 176Effects of calcination temperature on physical properties of ceria abrasives .............................................................................................. 176Effects of suspension pH on oxide and nitrid e CMP ................................ 178Conclusion s .......................................................................................................... 180Synthesis of Cerium Carbonates .................................................................... 180Synthesis of Ceria Abrasives .......................................................................... 180Preparation of Ceriabased Slurry .................................................................. 181CMP Evaluati on .............................................................................................. 181 9

PAGE 10

8 A COMPARISION OF CMP PERFORMANCE IN THE CERIA ABRASIVES SYNTHESIZED VIA VAR IOUS METHOD S .......................................................... 196Introducti on ........................................................................................................... 196Materials and Methods .......................................................................................... 196Sample Prepar ation ........................................................................................ 196Preparation of ceri a abrasives ................................................................. 196Preparation of ceria-ba sed slurri es .......................................................... 197CMP tools and cons umables ................................................................... 197Characteriza tion ............................................................................................. 197Abrasives ................................................................................................. 197Ceria-based sl urry .................................................................................... 198CMP perform ance .......................................................................................... 198Results and Discussion ......................................................................................... 198Comparison in Polishi ng Removal Ra te ......................................................... 198Comparison in WIWNU .................................................................................. 199Abrasive Effects on Defectivit y ....................................................................... 200Conclusion ............................................................................................................ 2019 CONCLUSION S ................................................................................................... 207Solution Growth Abrasives .................................................................................... 208Grain Control Ab rasives ........................................................................................ 209Core/shell Composit e Abrasives ........................................................................... 209Solid State Ab rasives ............................................................................................ 210Comparison of Polis hing Behavio r ........................................................................ 212APPENDIX: DIELECTRIC CONSTANTS OF MIXED SOLUTION OF SOME ORGANIC SOLVENT AND WATER AT ROOM TEMPERATURE ....................... 213LIST OF REFE RENCES ............................................................................................. 214BIOGRAPHICAL SK ETCH .......................................................................................... 220 10

PAGE 11

LIST OF TABLES Table page 2-1 Guiding principles for slurry design in chemical mechanical planarization ......... 574-1 Comparison of slurries used in this study ......................................................... 1074-2 The results of the CMP evaluation. ................................................................... 1085-1 Comparison of slurries used in this study. ........................................................ 1325-2 The results of the CMP evaluation. ................................................................... 1336-1 The results of removal rate for the ceria coated silic a particle s ........................ 1667-1 Dielectric constants of mixed solven t, zeta potentials and morphologies of cerium carbonate compounds with the ratio of ethanol to wate r ....................... 1958-1 Comparison of ceria abrasives used in this study ............................................. 2058-2 The results of the CMP evaluation .................................................................... 206 11

PAGE 12

LIST OF FIGURES Figure page 2-1 Schematic for CMP proce ss. .............................................................................. 482-2 Schematics of ideal oxide ILD CMP. .................................................................. 492-3 Oxide removal me chanism by CMP ................................................................... 502-4 Shallow trench isolatio n (STI) CMP process ...................................................... 512-5 Zeta potential of the oxide/nitride substrates and ceria abrasive as a function of pH. .................................................................................................................. 522-6 Silicon nitride SN2 hydrolysis reac tion schem e. .................................................. 532-7 (a) Zeta potential of the oxide/ni tride substrates and ceria abrasive as a function of pH and (b) the formation of passivation layer on the surface of STI structure with anionic organic polymer. .............................................................. 542-8 The comparison on (a) removal rate for oxide substrate with different abrasives and (b) Mohs hardness of ceria and materials to be polished during CM P......................................................................................................... 552-9 The schematics of su rface defecti vity. ................................................................ 563-1 Layout of the SKW-1 pattern wafer: (a) pattern density and pitch size layout, (b) mask floor plan, and (c) cross-sectional view ................................................ 703-2 Schematic diagram of rotational CM P tool ......................................................... 713-3 Schematic illustration of a slurry deliv ery system ............................................... 723-4 Diagram of film thickness m easurement system us ing NanoS pec. .................... 733-5 Schematic illustration of light scatteri ng analysi s ................................................ 744-1 XRD patterns of ceria particles synthesized from the mi xture of water and different alcohols; (a) ethylene glycol, (b) methanol, (c) 1, 4-buthylene glycol, (d) ethanol ......................................................................................................... 934-2 FETEM photomicrographs of ceria particles obtained by hydrothermal method using a new type of ceria precursor. ...................................................... 944-3 FESEM photographs of ceria particles prepared from the mixture of water and different alcohols; (a) ethanol, (b) 1,4-buthylene glycol (c) methanol, (d) ethylene gl ycol. ................................................................................................... 95 12

PAGE 13

4-4 (a) Average particle size and (b) cr ystallites sizes of ceria particles synthesized with different dielec tric constants of alc ohols .................................. 964-5 FESEM photographs of ceria particles prepared with different concentrations of potassium hydroxide; (a) 0.5 M, (b) 1.0 M, and (C) 1.5 M. ............................. 974-6 FESEM photographs of ceria particles prepared from different concentrations of nitric acid in hydrothermal conditions at 230 oC for 12 hr. ; (a) pH 4, (b) pH 2.5, (c) pH 0.5 and (d) pH 0.5. ............................................................................ 984-7 Crystallites size for ceria particles prepared from different pH at (a) 150 oC, (b) 200 oC and 230 oC. ....................................................................................... 994-8 FESEM photographs of the ceri a particles prepared with different hydrothermal conditions; (a) pH 3.0 at 220oC, (b) pH 3.0, (c) pH 1.5 and (d) pH 0.5 at 230oC, respecti vely. .......................................................................... 1004-9 XRD patterns and the (111) peaks analyzed to confirm grain size of the ceria abrasives dispersed in ceria-based slurry (a) A, (b) B, (c) C and (d) D. ........... 1014-10 FETEM micrographs and of ceria abrasive with average parti cle diameters of (a) 62 nm (slurry A) and (b ) 232 nm (slurry D). ................................................. 1024-11 Particle size distribution of ceriabased slurry used in this st udy. ..................... 1034-12 FESEM photographs of ceria abrasiv es (a) before and (b) after oxide CMP process. ............................................................................................................ 1044-13 Results of CMP field evaluation for removal rate and se lectivity. ..................... 1054-14 Results of CMP field evaluation for within-wafer nonuniformity (WIWNU) of silica film .......................................................................................................... 1065-1 Schematic diagram of ex perimental pr ocedure. ............................................... 1225-2 FESEM photographs of the ceria abrasives prepared with different calcination conditions; (a) slurry A, (b) B, (c) C and d(c) D ............................... 1235-3 (a) XRD patterns and (b) the (111) peaks analyzed to confirm crystallite size of the ceria abrasives dispersed in slu rry (a) A, (b) B, (c) C and (d ) D ............. 1245-4 The variation of crystallite size as a function of the co ncentration of grain growth accelerator ............................................................................................ 1255-5 FETEM micrographs of the ceria abras ives prepared with different cerium precursor; (a) cerium hydroxide, (b) ce rium nitride, (c) cerium chloride and (d) cerium diox ide .................................................................................................. 1265-6 Particle size distribution of ceria slurries as function of abrasive size. .............. 127 13

PAGE 14

5-7 FESEM photographs of ceria abrasives (a) before and (b) after polishing ....... 1285-8 Results of CMP field evaluation for removal rate and selectivity ...................... 1295-9 TGA curves of the ceria abrasives dried from (a) slurry A and (b) slurry D ...... 1305-10 Within-wafer non uniformity (WIWNU) of oxi de film. ......................................... 1316-1 FESEM images of silica core particles obtained by modified Stber method. .. 1506-2 (a) FESEM and (b) HRTEM micrographs for the surface condition of coated particle and FESEM micrographs for ceri a coated silica particles prepared by (c) precursor B and (d) precur sor A, respec tively. ............................................ 1516-3 XRD patterns of the synthesized particles; (a) bare silica particles, (b) ceria coated silica particles prepared by prec ursor B, and (c) pr ecursor A ............... 1526-4 XPS survey spectrum of ceri a coated silica par ticles ....................................... 1536-5 XPS spectra of O 1s peaks of ce ria coated silica particles. .............................. 1546-6 XPS Ce 3d multiplex of ceri a coated silica pa rticles ......................................... 1556-7 FESEM photographs for ceria coated silica particles prepared at different pH (a) 3.2, (b) 6.8 and (c) 9. 7 ................................................................................. 1566-8 Electrophoretic mobility for (a) silica particles (b) ceria coated silica particles and (c) ceria par ticles ....................................................................................... 1576-9 Scheme of the formation mechanism of ceria coated silica particles at different pH. ...................................................................................................... 1586-10 FESEM micrographs of ceria coated silica particles prepared by different concentration of ceria precursors (a) 0.0 ml, (b) 1.0 ml, (c) 2.0 ml, (d) 4.0 ml and (e) 8.0 ml. .................................................................................................. 1596-11 The variations of (a) coating thickness and (b) average particle size for samples obtained by changing the concent ration of ceria pr ecursors .............. 1606-12 FESEM micrographs of the silica particles with different size; (a) 105 nm, (b) 214 nm, (c) 332 nm, a nd (d) 442 nm ................................................................ 1616-13 FESEM micrographs of the ceria coated silica particles obtained from different core silica particles with diffe rent size; (a) 146 nm, (b) 256 nm, (c) 334 nm, and (d) 384 nm. .................................................................................. 1626-14 Results of CMP field evaluation for removal rate as fu nction of pH .................. 163 14

PAGE 15

6-15 Results of CMP field ev aluation for removal rate as function of CMP pressure. ......................................................................................................................... 1646-16 The result for within-wafer non uniformity (WIWNU) of oxide film ..................... 1657-1 XRD patterns of cerium compos itions produced under precipitation conditions with differ ent solven t. ....................................................................... 1837-2 FESEM micrographs of cerium ca rbonate compounds obtained by using pure water as solvent ................................................................................................ 1847-3 FESEM micrographs of spherical ceri um carbonate particles prepared from the mixture of water and differ ent alcohols; (a) methanol (CH3OH), (b) ethanol (C2H5OH), (c) 2-propanol (C3H8O), and (d) 1, 4-butandiol (C4H10O2). 1857-4 FESEM micrographs of cerium ca rbonate compounds prepared by various ratio of ethanol to water: (a) 0, (b) 1, (c) 3, and (d) 5 ........................................ 1867-5 The XRD pattern of (a ) cerium carbonate prepared by using mixed solvent of ethanol and water and (b) ceria abr asives obtained from thermal decomposition of the cerium carbonate at 700 oC. ........................................... 1877-6 FESEM micrographs of (a) as-prepared particles of ceria abrasives and (b) ceria abrasives obtained from thermal decomposition at 700 oC. ..................... 1887-7 Relationship between surface area and cr ystalline size of ceria abrasives as a function of calcinati on temperatur e. ............................................................... 1897-8 Electrokinetic behavior of silica, ceria and ceria with surface active agent added as a function of suspension pH .............................................................. 1907-9 The changes in particle size distribution of ceria-based solvent as a function of suspensi on pH .............................................................................................. 1917-10 The CMP evaluation for removal rate of oxide and nitride films as a function of calcination te mperatur e ................................................................................ 1927-11 Results of CMP field evaluation for removal rate and selectivity ...................... 1937-12 The CMP evaluation for removal rate of silicon oxide wafer as a function of suspension pH. ................................................................................................. 1948-1 FESEM micrographs of various ki nds of spherical ceria abrasives synthesized by variety methods; (a) hy drothermal method, (b) flux method, (c) surface-induced precipitation method, and (d) thermal decomposition. ...... 2028-2 Particle size distribution of ceria-based slurries. ............................................... 2038-3 Comparison of different ceria abr asives on surface defectivity. ........................ 204 15

PAGE 16

A-1 The dependence of dielectric constant on the composition of different alcohols and water ............................................................................................ 213 16

PAGE 17

LIST OF ABBREVIATIONS Viscosity m Micrometer (1 X106 cm) A Specific surface area Angstrom (1 X1010 cm) AW Active trench BET Brunauer-Emmett-Teller CMP Chemical mechanical planarization D Crain size dBET Average particle size determined by BET DLVO Derjaguin Landau Verwey Overbeek dSEM Average particle size determined by FESEM dXRD Crystalline size estimated from XRD patterns FESEM Field emission scanning electron microscopy FTIR Fourier transform infrared h Hour IC Integrated circuit IEP isoelectric point ILD Interlayer dielectric K Degrees Kelvin kV Kilo voltage LOCOS Local oxidation of silicon 17

PAGE 18

LPCVD Low-pressure chemical vapor deposition M Molarity mA Milliampere mg Milligram min Minute mL/min Milliliter per minute MRR Material removal rate nm Nanometer (1 X 109 cm) oC Degrees Celsius PAA Poly acrylic acid PECVD Plasma enhanced chemical vapor deposition psi Pound per square inch rpm Rate per minute SSA Specific surface area STI Shallow trench isolation TEM Transmission electron microscopy TG/DTA Thermogravimetric and di fferential thermal gravimetry TW Trench width ULSI Ultra large scale integrate WIWNU Within-wafer nonuniformity wt % Weight percent XPS X-ray photoelectron spectroscopy 18

PAGE 19

XRD X-ray diffraction Half-width of the diffraction peaks Dielectric constant Diffraction angle Wavelength Zeta potential 19

PAGE 20

Abstract of Dissertation Pr esented to the Graduate School of the University of Florida in Partial Fulf illment of the Requirements for t he Degree of Doctor of Philosophy NOVEL SYNTHESIS METHODS FOR PR EPARATION OF CERIA ABRASIVES FOR CHEMICAL MECHANICAL PLANARIZATION APPLICATIONS IN SEMICONDUCTOR PROCESSING By Oh Myoung Hwan December 2010 Chair: Rajiv. K. Singh Major: Materials Science and Engineering As the device design rule decreased, ceria-based slurries have been widely used instead of silica-based slurries in a variety of chemical mechanical planarization (CMP) applications for multilevel in tegrated circuit (IC) manufacture, since these slurries address many of the important issues result ing from the use of silica-based slurries. However, ceria (CeO2) abrasives usually induce higher scratch level than silica particles due to its cubic crystalline structure, irr egular shape, and poor dispersion stability in slurry. Therefore, this article is intended to establish novel synthet ic methods of ceria abrasives leading to lower scratch level on wafer surface and ultimately present the direction of CMP abrasive for future technol ogy nodes in order to meet the ever more challenging defectivity requirements. To accomplish these aims, this article in troduced the 4 types of novel synthetic methods for the formation of ceria abrasives The ceria abrasives were synthesized by solution growth method, grain control method, core/shell composite method, and thermal decomposition method. In this investigation, the influences of solvent type and suspension pH on the formation of ceria parti cles were intensively investigated. The 20

PAGE 21

21 size of ceria particles was controlled by adjusting the reaction parameters of each method without additional mechanical milling and filtration. The relationships between dielectric property of the solvent and mor phological properties were also discussed in terms of the supersaturation of solution a nd electrostatic attraction mechanism. The resultant particles were characterized with field emission scanning electron microscopy (FESEM), transmission electron microscopy (TEM), X-ray photoelectron spectroscopy (XPS), X-ray diffraction (XRD), Thermogravim etric and differential thermal gravimetry (TGA/DTG), Fourier transform infrared ( FTIR) spectroscopy, Brunauer-Emmett-Teller (BET), light scattering instrument s and zeta potential measurements. In order to investigate the effects of the synthes ized ceria abrasives on CMP performance, CMP tests were carried out with the ceria-based slurry formulated by dispersing the synthesized ceria particles with anionic organic polymer. The effects of the synthesized ceria abrasives in CMP slu rry were investigated for silicon dioxide and silicon nitride CMP process. The polishing b ehaviors of ceria abrasives were discussed in terms of morphological properties and mec hanical abrasion of the ceria particle. In this CMP evaluation, material removal rate, selectivity, wafer uniformity, and defectivity of the polished wafer were measured by metr ology tools, which are used in current integrated circuit (IC) fabricat ion plants in order to support polishing results obtained by this investigation.

PAGE 22

CHAPTER 1 INTRODUCTION Research Rationale As the minimum feature size of micr oelectronic devices deceases, the newly developed planarization technique and new cons umable materials have been utilized in modern semiconductor fabrication industry.1 To continually satisfy more demanding devices, chemical mechanical planarization (C MP) has become one of the most critical semiconductor fabrication technologies becau se it offers a superior means for global and local planarization. Global planarization which is ess ential to produce a multilevel integrated circuit (IC) device is achieved by reducing topographic vari ation at the wafer scale.2 Without CMP, it would be impossibl e to fabricate complex, dense, and miniaturizing multilevel IC devices. Over the past years, CMP has significantly advanced both in the development of more sophisticated processing tools and in the formulation of novel slurries to further enhance process performance.3,4 Despite these advancements, the fundamental knowledge of the effects of the numerous CMP process variables on polishing perform ance is not clear, due to the lack of understanding of the substrate to be polished, the slurry that provides the chemistry and abrasives for mechanical removal and pad interactions involved. The lack of this understanding is a significant barrier to the development of next-generation CMP technologies. Therefore, m any research has been investigate to understand the nature of substrate-slurry-pad interactions that occur during t he CMP process. The CMP has been used for interlayer dielectric (ILD) planarization, shallow trench isolation (STI) structure, and damascene technologies. For these various CMP processes, the characteristics of slurry particles are critical in determining the 22

PAGE 23

planarization performance of CMP. During the past decade, silica (SiO2) particles were traditionally used as CMP abrasives to remove deposited oxide topography at technology nodes of 90 nm and higher. Thus, previous studies for the planarization performance were focused on mechanical abrasion between silica particles and the substrate to be polished and chemical modifica tion in silica-based slurry in order to increase the removal ratio of oxide to nitri de layer in STI CMP. Ho wever, at the 65 nm technology node and at below nodes, ceria (CeO2) particles are being introduced in a variety of CMP applications for IC manufacture since ceria particles have the capability of achieving higher removal rates and global planarization than silica particles and ceria-based slurry can more easily be contro lled by additives in the slurry formulation.5 Recently, the demand of ceria particles as abrasives has been rapidly increased in semiconductor fabrication industry. Therefore, the fundamental knowledge for the characteristics of ceria particles as CMP abr asive is required in order to enhance the ability of current semiconductor devices. Compared with other abrasives used in CM P slurry, ceria particles is commonly used in the ultra large scale integrated (ULSI) circuit structure due to the effective removal rate for oxide film and the softness of the particles. Recently, ceria-based slurry has been used in CMP of STI structures consis ting of silicon dioxide and silicon nitride (Si3N4) deposition due to its high selectivity over nitride.6 For the STI-CMP process, the use of high selectivity slurries is very impor tant to halt the polishi ng at the nitride stop layer and reduce the amount of defects such as erosion and dishing. The structural properties, chemical aspects and morphological characteristics of the ceria abrasives have been identified as the im portant parameters that in fluence the CMP performance 23

PAGE 24

such as oxide removal rate, removal se lectivity, CMP-induced defects and wafer uniformity.7 Therefore, many approaches to control these properties of ceria abrasives have been extensively investigated. Even though the characteristics of ceria parti cles significantly a ffect the quality of CMP process, it is not easy to manufactu re ceria particles as CMP abrasives. The commercial method for synthesis of ceria parti cles involves thermal decomposition of cerium salts such as cerium carbonate and cerium hydroxide. This method leads to very porous ceria particles with high surface area, inducing softness and high chemical reactivity to oxide films.8 However, the size and the s hape of ceria abrasives are very limited since particle growth is difficult to c ontrol during calcination process. To achieve the desired particle size and the uniform parti cle size distribution, mechanical milling and filteration is required. Other methods for preparing ce ria abrasives are liquid phase processes. These methods can lead to ceria abras ives with desirable morphological characteristics by manipulating reaction par ameters. However, the size of ceria abrasives is limited to less t han 100 nm. Use of these small si ze particles results in low removal rates of target layers during CMP. Therefore, a new method to overcome these problems of ceria abrasives is required. Scope of the Research The research presented in this dissertati on is intended to establish novel synthetic methods of preparing ceria abrasives for ILD and STI CMP with precise morphologies and chemical composition. The ov erall objective of this research is to investigate the effects of abrasive material properties on pol ishing removal rate and wafer defectivity by using different kinds of ceria particles obtain ed from a variety of synthesis methods. On the basis of results obtained fr om this study, this article will ultimately present the 24

PAGE 25

direction of CMP abrasive for future technol ogy nodes in order to meet the ever more challenging defectivity require ments. To achieve these aims, the research presented in this dissertation was devoted on a variety of preparation methods of ceria particles. For CMP performance evaluation, the effect of th e resultant ceria abrasives on the removal rate, the oxide-nitride removal selectiv ity and within-wafer nonuni formity (WIWNU) was investigated. A synopsis of the efforts consti tuting this study is organized as follows. Chapter 2 reviews the liter ature on the CMP process in cluding main components in planarization performance. The procedure and mechanism for ILD and STI CMP was addressed in detail. From these backgr ounds, the need of ceria abrasives was emphasized in CMP application and motivati ons for the implement ation of the CMP process using ceria abrasives were discusse d. Furthermore, the important issues for CMP evaluation were summarized and discussed in terms of abrasive characteristics. New approaches for developing more e ffective abrasives are introduced. Chapter 3 was devoted to the studies conducted on the preparation and analysis techniques of consumables used in this st udy. For abrasive particles, the synthesized particles were analyzed by XRD, FESEM, TEM, XPS, FTIR, TG/DTA and BET. For slurry, slurry stability, mean particle size, and zeta potential measurements was reported using a variety of light scatteri ng instruments. For CMP evaluation, film thickness, selectivity, nonuniformity, and defectivity of the polished wafer were measured by metrology tools using in current integrated circuit (IC) fabrication plants in order to support polishing results obtained by th is investigation. Brief descriptions of measurement principles for each facility are also presented in this chapter. 25

PAGE 26

Chapter 4 investigated the effects of si ngle crystalline ceria abrasives on silicon dioxide and silicon nitride CMP process. T he single crystalline ceria particles were synthesized by heating peptized ceria sol as precursor under hydrothermal conditions. In this chapter, the relationships between diel ectric property of the solvent and particle size were investigated in terms of the supersaturation of solute. In addition, the influences of precipitation participating anions (OH-) and acidic hydrothermal medium on crystallites size of ceria particles were st udied. Furthermore, the polishing behavior of the single crystalline ceria abrasives was discu ssed in terms of morphological properties of the abrasive particle. Chapter 5 discussed the effects of spher ical ceria abrasives on planarization performance. The ceria abrasives were prepared by the flux method, using potassium hydroxide (KOH) as the grain growth accelera tor. In this chapter CMP test was carried out with four types of ceria-based slurry fo rmulated by dispersing the ceria abrasives with different particle size in order to determine how the re moval rate, removal selectivity, and wafer surface roughness of oxide and nitride films depend on the abrasive size and particle size distribution in slurry. Chapter 6 presented studies conducted on the synthesis of monodispersed ceria coated silica particles and its CMP performance. The coated particles were prepared by the surface-induced precipitation method, in which a new type of ceria coating precursor was deposited on the surfac e of spherical silica particles via electrostatic attraction route. The ceria coat ing precursor was synthesized by the sol-gel technique, which employs ethanol as a solvent. In this chapter, the effects of solvent type and solution pH on the formation of ceria coating layer were investigated. CMP test 26

PAGE 27

was performed with the different types of slurries with 146 nm of abrasive size controlled by using 135 nm of colloidal silic a particles. The suspension pH effects were investigated as a functi on of the applied head load an d explained in terms of absorption/repulsion behavior between abrasives and materials to be polished. Chapter 7 introduced a novel method to synthesize the spherical ceria particles via two-step procedure. In firs t step, spherical cerium carbonate particles were prepared via simple precipitation met hod using alcohol/water mixed solvent. In second step, the ceria particles were obtained by subsequent thermal decomposition of the precursor. After calcination, the resultant particles we re used as abrasives of ceria-based slurry without mechanical milling and filter ation. In this chapter, the effects of physicochemical solvent properties on the crystalline phase, microstructures and morphological properties of particles were investigated. In addition, the effects of suspension pH in slurry on polishing performance we re discussed in terms of elec trostatic repulsive forces. Chapter 8 investigated the effects of abrasive material properties on polishing removal rate and wafer defectivity by using different kinds of ceria particles obtained from previous chapters. In this chapter, the effects of the br ittle behavior of ceria abrasives and particle size distribution of slurry on wafer surface were discussed. Chapter 9 summarized the conclusions of this study and offered some suggests for future research. 27

PAGE 28

CHAPTER 2 LITERATURE REVIEW Chemical Mechanical Planarization (CMP) Chemical mechanical planarization (CMP) is an abrasive process using chemical agents and a circular action to polish the surface of the wafe r smooth. Planarization is the process of smoothing and planning surface. CMP can be also referred to chemical mechanical polishing that causes planarizatio n of surface. Howe ver, the meaning of polishing is different from the meaning of pl anarization. Polishing generally refers to smoothing the surface not necessa rily planar. Thus, the primar y function of CMP is to planarize individual layers in complex integrated circuits. The slurry is the very important key player among the CMP consumables providing both chemical and mechanical effects. CMP Process A schematic of a typical CMP process is illustrated in Figure 2-1. The wafer is held on a rotating carrier force down and is pressed against a polishing pad attached to a rotating disk, while chemically and mechanically active slurries are applied. CMP slurry contains abrasive silica or ceria particles suspended in an aqueous medium. Both mechanical action of the abrasive particles and the chemical action of slurry constituents remove material from the wa fer surface. Planarization results because material is removed faster from protrudi ng regions on the surface than from recessed regions. The general requirements of CMP can be summarized as follows: First, there is a need for high removal rates of the mate rial to be polished to achieve the needed throughput. Second, the selectivity of the slurry must be sufficiently high so that only the material of interest is polished. Third, the polished surface must exhibit excellent 28

PAGE 29

topographical uniformity. Finally, local dishing and erosion effects must be minimized to satisfy the die-level flatness requirements of sub-0.3 micron devices. Therefore, in order to meet the requirements, it is necessary to understand in detail the nature of contact between chemical-mechanical consumable and individual films. Among these consumable, CMP slurry is one of the most crucial element s to improve the quality of multilevel interconnect networks. It is genera lly agreed that CMP slurries should be designed to optimize for specific applications.1 CMP Slurry CMP slurry is typically contained with su spended abrasive particles, an oxidizing agent, corrosion inhibitor, and other additives including dispersants.1 During the CMP process, the abrasives in the slurry and the rotating polishing pad provide the mechanical action that remo ves material on the surfac e layer. The chemical components of the slurry accelerate polishi ng and can be mixed to select specific substances on the surface of the wafer. The types of CMP slurry is categorized by the target materials polished in CMP process. Abrasives and chemical components are also changed by the properties of layers polished during CMP process. Abrasives in the slurry play the very important role of transfe rring mechanical energy to target material. Silica or ceria particles are commonly used as abrasive of oxide CMP process and alumina particles are us ed in metal CMP process.1 During CMP process, it has been known that the abrasive particle size and si ze distribution have an enormous impact on the evolution of microscratches. Over t he past 10 years, the c hemical property of surface and the hardness of abrasive have been identified as the important parameters, which affect removal rate, selectivity, and the quality of surface polished during CMP process. Chemical components in the slurry can be designed for specific functions by 29

PAGE 30

the addition of oxidizers or adjusti ng the pH of the chemical vehicle.9 It is generally understood that additives modifies the su rface to be polished and yields a softer and porous complex layer, which is then removed by mechanical force in the process. Moreover, dispersion agent is used to provide a stable dispersion of abrasive. Therefore, the quality of polishing, which is critical to yield, depends upon the quality and consistency of the CMP slurry. CMP slurry must continually improve to meet much higher performance specifications demande d by the trend of new types of CMP technique and the introduction of noble materials. CMP of Dielectrics CMP is commonly employed for both t he front and back end processing of integrate circuit (IC) devices due to its unique global planarizat ion capability. This process includes interlayer dielectric (ILD), shallow trench isolation (STI), pre-metal dielectric (PMD), and copper CMP. Slurry is specifically modifi ed for each CMP process to improve the polishing performance such as removal rate, removal selectivity, global planarization, and minimized defectivity. Am ong these CMP processes, this paper will deal with oxide used as ILD and STI CMP to understand the fundamentals of slurry design for CMP. In this part, I discuss the procedure and mechanism for these CMP process. Oxide CMP Process of oxide CMP Oxide planarization is probably the most co mmon of all CMP processes. Inter-level dielectrics (ILD) are routinel y planarized prior to the deposit ion of the next metal layer. Oxides layers vary thickness, but genera lly between 5000 and 10,000 of dielectric material is removed during CMP. Figure 2-2 shows the schemat ic for the ILD CMP.10 30

PAGE 31

Mechanism of oxide CMP Oxide removal does not occur as a result of physical abrasive action alone. Rather, it is the result of chemical reactions in which oxide bonds form between the slurry particles and the wafer surface.11,12 The physical abrasive action then comes into play as the moving slurry particles break these bonds and move away. Although much research is still being conducted in this area, it is known that the process proceeds along these lines, as illustrated in Figure 2-3: 1. Hydroxylation formation of hy drogen bonds between oxides on the wafer surface and the slurry particles 2. Formation of hydrogen bonds between slurry and wafer 3. Dehydration (expulsion of H2O) 4. Breaking of bonds as the slurry particles are forced along The most common slurries used for ILD CMP are silica-based and ceria-based. These slurries generally have particles wh ich range in size from 30 ~ 150 nm. Relationship between particles and wafer An understanding of the nature of contac t between particles and the wafer to be polished is essential to maintain the strict process requirements for manufacturing current and future generation integrated circuit (IC) chips. Especially in oxide CMP, particles play an important role in achi eving desired CMP performance such as high material removal rate, low surface defects, and local global planarizat ion via mechanical abrasion and chemical modifi cation of the wafer surface.10 In spite of its importance, the effect that the slurry particles have no po lishing performance is not clear. For polishing of copper or ferrite, it was suggested that the polishing rate is proportional to particle size and solids loading.13,14 Cook presented data suggesting that the polishing rate is 31

PAGE 32

independent of particle size for glass polishing.12 Izumitani suggested that the polishing rate decreases with increasing particle size.15 Singh suggests two polishing mechanism in silica CMP.16 One is a contact area based mechanism by which 3/13/1 0 CA (2-1) where A is the contact area, C0 is the particle concentration (the number of particles) and is the particle diameter (abrasive si ze). In this model, the polishing rate increases with an increase in particle conc entration and a decrease in particle size, which was observed during tungsten CMP.17 The other is an indentation volume based mechanism by which 3/43/1 0 CV (2-2) where V is the indentation volume. Accordi ng to this indentation volume based mechanism, the polishing rate increase wit h decreasing particle concentration and increasing particle size. This mechanism was observed via silica polishing experiments. Shallow Trench Isolation (STI) CMP Process of STI CMP The shallow trench isolation (STI) pr ocess is one of the most important applications of CMP. This process has emerged as the pr imary technique for advanced ultra large scale integration (ULSI) tec hnologies. This process was developed as an alternative to traditional local thermal oxidation processes (LOCOS). The LOCOS process has a major drawback known as the birds beak phenomenon. A birds beak defect occurs due to the diffusive nat ure of the oxide growth process.1 As the oxide grows vertically downward into the underlying silicon, it also grows horizontally to the sides and underneath the silicon nitride mask, thus encroaching into the active device 32

PAGE 33

regions. This becomes more of a problem at geometries below 0.25 m. A secondary benefit of STI is that it can generally be done faster and at lower temperatures than LOCOS. Referring to Fig. 2-4, the STI processes begins with the growing of a very thin oxide layer (100 ~ 200 ) sometimes calle d the pad oxide. Then, a thicker (500 ~ 1,500 ) layer of CVD nitride is deposited on top of the pad oxide. These layers are then patterned with photoresist and trench is et ched into the substrate. After the trench is etched, a thin oxide layer is grown on the trench sidewalls and bottom to smooth out the corners and to serve as a liner. Finally, 8,000 ~ 11,000 of CVD oxide is deposited to fill the trench. This oxide is t hen planarized using CMP. With the oxide serving as an etch stop, the nitride layer is then stripped away to expose the active device regions. One of the key issues of this process is the selectiv ity of the nitride vs. oxide. Results ranging from 5:1 to 175:1 have been reported. Mechanism of STI CMP STI is a specific CMP application which ge nerally requires the selective removal of silicon dioxide to silicon nitride on a pattern ed wafer substrate. In this case etched trenches are overfilled with a dielectric (silicon dioxide) which is polished using the silicon nitride barrier film as a stop laye r. The process ends with clearing the silicon dioxide from the barrier film while minimizi ng the removal of exposed silicon nitride and trench silicon dioxide. This requires slurry ca pable of achieving a high relative ratio of silicon dioxide material removal to silicon nitride removal (high selectivity slurry). Ceriabased suspensions have received considerable attention in STI applications because of their ability to achieve high selectivity.18,19 33

PAGE 34

Mechanism of silicon diox ide using ceria particles One mechanism of silica glass polishing us ing ceria particles previously proposed by Cook involves proton abstraction from s ilica followed by reaction with Ce-OH to form a Si-O-Ce bond.13 Cook described the nature of c hemical interaction leading to the accelerated removal rate with ceria abrasives, listed below: 1. Water penetrates into the glass surface 2. Water reacts with the surface, which l eads to the dissolution under particle load 3. Abrasives adsorb some dissolution products and leave fr om the substrate 4. Some dissolution products redeposit onto the substrate 5. Surface dissolution happens between particle impacts It is hypothesized that the formation of a strong Si-O-Ce bond leads to break of SiO-Si bond on wafer surface because the free energy of the formation of cerium oxide (Hf = -260 kcal/mole) is much less than the free energy of the fo rmation of silicon dioxide (Hf= -216 kcal/mole).17,18 Maximum material removal happens when a neutrally charged ceria particle approaches a silica s ubstrate with negative surface charges to form surface chemical bonds in aqueous environments. 2OHMHOHM(pH < pHIEP) (2-3) HOMOHM (pH > pHIEP) (2-4) where M-OH is the neutral hydroxyl group and pHIEP is pH value at zero charge of ceria abrasives (isoelectric point), -M-Ois the deprotonated surface groups and -MOH2 + is the protonated surface groups, respecti vely. Figure 2-5 shows the variation of zeta potential with pH value for substrates (oxide and nitride) and ceria abrasives. At pH < pHIEP, silica substrate has a negative surface charge and the ceria abrasives have a 34

PAGE 35

positive surface charge, leading to absorption between two materials. On the other hand, with the increasing pH of the solution, the ceria surface becomes more negatively charged and the silica surface also has a negat ive surface charge, leading to repulsion between two materials. The remo val of material from the su rface of silica-glass during the polishing process is attr ibuted to a temporary attach ment (through surface chemical bonds) of ceria particles to the silica-glass surface. OHCeOSiOHCeOSi (2-5) The material removal occur when a silic a tetrahedron structure is broken from the silica substrate because the strength of Ce-O bonding is greater than Si-O bonding.19 CeOSiOSiCeOSiOSi (2-6) Mechanism of silicon nitride using ceria particles The proposed mechanisms for ceria-based sl urry on silicon nitride substrate may be slightly more complex than pr oposed for silicon dioxide substrate.22-24 As shown in Fig. 2-5, the surface charge of silicon nitride has a functionally difference in acidic pH region. Particularly, at pH 5 ~ 6, the surface of silicon nitr ide have a positive charge due to the presence of protonat ed amine groups which are no t present on silicon oxide substrate.25-27 However, hydrolysis reactions on the surface of silicon nitride occur readily in aqueous solutions which liberate ammonia and generate silica-like surface structures. Such a hydrolysis reaction depict ed by Eq. 2-7 would favor formation of reactive surface silanol groups and surfac e charge at pH 5 ~ 6 would also become negative. 3 2 2 43436NHSiOOHNSi (2-7) 35

PAGE 36

Therefore, the kineti cs of the hydrolysis reaction on silicon nitride affects surface reactivity and functionality with ceria abrasives. Laarz et al.28 proposed an acidic catalyzed pathway for hydrolysis of silicon nitride in aqueous environments. Fig. 2-6 depicts hydrolysis reaction scheme for a nucleophilic displacement reaction (SN2) of silicon nitride, listed below: 1. Protonation of surface amine 2. Coordination of water 3. Concerted water insertion and Si-N cleavage 4. Proton transfer to amine leaving group 5. Continued hydrolysis at Si center A water molecule can coordinate to the silicon via an SN2 insertion known in organic chemistry as a nucleophi lic displacement reaction (SN2 = substitution, nucleophilic, bimolecular). In this pathway, an amine is liberated as a leaving group after the first water insertion. This reaction introduces a hydroxyl group into silicon and the resulting silicon is more electropositive and sterically less hindered. These electronic and steric considerations induce that subsequent hydrolysis should proceed more quickly than the initial water insertion. In this reaction, three proton transfer reactions occur29: (1) from the solution to a surface ami ne Si-NR2, (2) from the surface amine to the adjacent water, and (3) from the prot onated silanol to t he solution amine. R represents the subsurface neighboring atom covalently bonded to silicon (most likely nitrogen). R represents the surf ace atom covalently bonded to nitrogen (either silicon or hydrogen). Based on this mechanism, mole cules which can compete for surface 36

PAGE 37

protons should have an impact on hydrolysis rates, affect Si3N4 surface functionality, and ultimately influence reactivity with ceria abrasives. High selectivity ceria-based slurry The surface potential for oxide and nitr ide are affected by the suspension pH, dispersants and organic additives during STI CM P process. In order to improve the selectivity and uniformity, an anionic acrylic polymer is commonly used to passivate the surface of the nitride film during STI-CM P, which prevents ceria abrasives from contacting the film surface. Hirai et al.30 explained the selective absorption mechanism of acrylic polymers on silicon oxide and silic on nitride layers in water-based system. They showed that the charac teristics of the passivation layer are determined by the acrylic polymers and suspension pH during CM P process. Moreover, many researchers reported that the selective adsorption is attr ibuted to the difference in surface charge between silicon oxide and silicon nitride layers.31-33 Generally, the silicon oxide layer and the abrasives in ceria-based slurry with acrylic polymer have a negative surface charge at pH 3.0, while the silicon nitride layer has a positive surface charge at pH 3.0.30-35 Figure 2-7(a) shows the variation of zeta potential as a function of pH value for substr ates (oxide and nitride) and ceria abrasives including acrylic polymer. Philipossian et al.32 also proposed a selective adsorption model based on the zeta potential of ceria-bas ed slurry with anionic organic polymer in terms of high selectivity. Fi gure 2-7(b) represent the formation of passivation layer on the surface of silicon nitride layer and elec trophoretic behavior of each material with anionic organic polymer during STI CMP process. The attraction/repulsion reaction between ceria abrasives and ox ide/nitride layers resu lts from the different electrophoretic mobility as a function of su spension pH. These behaviors affect CMP 37

PAGE 38

performance such as material removal rate of substrate and removal selectivity between silicon oxide and silicon nitride layer. Anionic acrylic polymer is commonly used to improve the removal selectivity by formatting passivation layer on the surface of the silicon nitride. Ceria Abrasive Abrasives in the slurry play the very important role of tr ansferring mechanical energy to the surface of substrates dur ing CMP. Among these abrasives, ceria as abrasive has received considerable attent ion in CMP process due to its chemical functions leading to high removal rate, silicon oxide to silicon nitride selectivity, and lower solid content in slurry. However, there are some problems to be worked out. Advantages of Ceria Abrasive As mentioned earlier, ceria particles receiv ing intense attention as a main slurry component for CMP process in semiconductor manufacturing industry due to the effective removal rate for oxide film and the so ftness of the particles.14 Fig. 2-8(a) compares the removal rate of oxide layers between fumed silica, colloidal silica, and ceria as a function of normalized polishing stress.19 The removal rate with ceria-based slurry is greater than that with silica-based slurry This is attributed to the fact that ceria abrasive exhibits a chemical reactivity for oxide layer leading to acceleration of the removal rate during oxide CMP. As a re sult, the chemical bonding between ceria and oxide layer can be rapidly removed by t he mechanical force generated by pressed pad and abrasive, and this physicochemical reac tion lead to the high removal rate of a silicon dioxide film by ceria abrasive. Moreover the hardness of ceria is lower than that of substrates to be polished during CMP process as shown in Fi g. 2-8(b). From this fact, it is expected that the scrat ches on the surface of wafer will be decreased by the lower 38

PAGE 39

hardness of ceria abrasives. Additionally, ce ria-based slurry has received considerable attention in STI CMP because of its ability to improve removal selectivity. The oxide-tonitride removal selectivity is usually enhanced by adding of an acrylic polymer as additive to water-based slurry with ceria abras ives. Thus, removal selectivity is affected by the molecular weight, the concentration of acrylic polymer, and the morphological properties of ceria abrasive. Furthermore, alt hough ceria is a relatively soft material, it has long been used to polish harder glass subs trates effectively. Compared with as high as 30 wt % for conventional colloidal silica abrasives and 12.5 wt % for fumed silica abrasives, ceria-based slurries typically contain less than 1 wt % solid content.36 This will induce a considerable reduction in manufacturing cost and solid waste discharge. Therefore, ceria particle can provide excellent CMP performance owing to high polishing efficiency for silicon dioxide film and lower hardness. Therefore, the ceria particle as abrasive for CMP slurry has been widely investigated to improve the quality of CMP process. Disadvantages of Ceria Abrasive In spite of many advantages of ceria abrasiv e, this contains critical disadvantages leading to serious defects on substrates dur ing CMP process. Usually, commercial ceria abrasives for CMP slurry were synthesized by thermal decomposition of the cerium salt such as cerium carbonate and cerium hydroxide.8 This method offers certain advantages, such as the higher chemical ac tivity and the brittle property of ceria abrasive due to the high porosity of the surface.35 However, the size and morphology of the ceria particles are very limited in that particle growth is difficult to control.38-40 A large number of oversized particles in the distribution tend to gi ve high scratch counts on the polished wafer.41 Also, these particles need a complicated milling process to regulate 39

PAGE 40

the size distribution. In or der to overcome this problem, many approaches to control these properties of ceria particles have been extensively investigated by using liquid phase processes, such as precipitation method,42 hydrothermal method,43-45 sol-gel method,46 and electrochemical method.47 These is the attractive methods since particles with the desired size and morphology can be produced by carefully manipulating parameters such as solution pH, concentration reaction temperature, time, and the type of solvent. Besides, these processes can directly synthesize well-crystallized particles without post-heat treatment. However, the size of ceria particles synthesized by using liquid phase process was limited to less than ~ 100 nm. These particles lead to the low removal rate in the CMP process. Moreover, the ceria abrasive in CMP slurry has easily sedimented because ceria is too dense to remain suspended in solution. The settling behavior is the different characteristics of ce ria abrasive with respect to colloidal silica. The specific gravity of ceria and colloidal silica is about 7.13 g/cm3 and 2.2 g/cm3, respectively.48 The particle settling is much more se vere for ceria-based slurry than that for silica. The sedimentation of ceria abr asive induces an unstable polishing rate for changeable solid contain during CMP and hard aggregates resulting from poor dispersion stability creates surface scrat ches on the polished f ilm. Therefore, the broader particle size distribution and the sedimentation aggregates have a bad influence on the quality of t he polished films during CMP pr ocess. Many approaches have been extensively investigated to overcome these problems. Key Quality Issues The most important issues in slurry perfo rmance for CMP relate to removal rate, global planarity, surf ace topography (dishing and erosion), surface defectivity (including roughness, scratches, dents, and delamin ating), and particle contamination.10 To 40

PAGE 41

develop a methodology for designing slu rry formulations, one must be able to understand the mechanisms active during CMP processing. Table 2-1 lists some of the most important fundamental par ameters that must be optimized in order to achieve acceptable characteristics in CMP slurry. Removal Rate A high removal rate is an essential aspect of a CMP performance. The removal rate is the amount of material removal by CM P in a given time frame. It is calculated according to the Pres ton equation, MRR = KpP0V, where MRR is the material removal rate, P0 the down pressure, V the relati ve velocity of water, and Kp a constant representing the effect of other remaini ng parameters, and the amount is usually expressed in /min.49 Removal rates depend on the f ilm being removed, type of pad and slurry being used, amount of downforce and relative velocity of the wafer carrier and polishing platen.50-52 Especially, the removal rate of dielectrics can be affected by:51 1. The size and distribution of the abrasives in slurry 2. The number of abrasives 3. The pH of the slurry 4. Pre-CMP film stress Increasing any of these properties will usuall y result in an increased removal rate. However, it has been reported that in some instances raising the pH does not necessarily increase the removal rate. In fact, for some types of slu rries reducing the pH can slightly increase the removal rate. The removal rate is affected by the size and concentration of slurry abrasives due to frictional force between the abrasives and the wafer as mentioned in CMP of dielectrics sect ion. The type of oxid e, thermal or CVD, also has an effect on the removal rate. Another factor which affects the removal rate is 41

PAGE 42

the topography on the surface. Material is typically removed at a higher rate on small and isolated features. On lar ger or tightly spaced features the removal rate can be reduced drastically. This will also impact the uniformity of removal. Uniformity and Planarity Uniformity and planarity are two closely related but distinctly different topics. Uniformity is the measure of film thickness (or removal rate ) variations across the wafer. Planarity is more a measure of overall die flatness. In ot her words, a given wafer may exhibit acceptable planarity but not be very uniform. All of the process variables mentioned can have direct effect on uniformity Controlling the process is the key and much work and research still needs to be done. Uniformity is the standard deviation of thickness removal rate measurements and it is expressed as a percentage of the average thickness removed. Planarity, on th e other hand, is simply a measurement of the degree of flatness and can be expressed as a percentage (planarity across the wafer) or as a specific number. Both of t hese issues are complicated by a variety of factors, such as topography spacing. The topography not only affects uniformity across the entire wafer, but it can cause pr oblems within each specific IC device. Global Planarization Global planarization refers to the ability of the CMP abrasives to rapidly planarize pattern-dependent and large-scale surface mo rphologies. The lateral dimensions of surface topography can range from nanometers to several millimeters, due to large pattern size or gentle topographic variations The CMP process is typically conducted on films deposited on patterned surfaces that as a result have significant surface topography. The pitch of the pattern (the sum of the width of the patterned lines and the spacing between them) as well as its density c an vary significantly across the die, which 42

PAGE 43

results in different local polishing pressures across the patterned surface. The variations in local polishing pressure lead to varying re moval rates and thus to varying amounts of material removed before global planarizatio n can be achieved. Compared with CMP, chemical etching decreases surface planar ity and increases surface roughness, while mechanical polishing can enhance the planarity but only at a low removal rate and at the expense of a poor surface finish. A key condition for global planarization is the formation of a very thin passivating surfac e layer that is subsequently removed by mechanical component of the slurry. The thi ckness of this layer is commonly under 2 nm. The removal rate of thin passivated surface layer is greater at the highest regions of the wafer surface than at the lowest regions, due to differenc es in local pressure in these regions. If the passivated layer is thi nner than the difference in height between the highest and the lowest regions significant pl anarization is expected to occur with CMP. In the case of dielectric CMP, it is genera lly believed that by controlling the pH in the alkaline regions, a thin hy drated surface layer is achieved.12 The role of the hydrated surface layer formed under alkaline pH conditions is to soften the surface so that higher removal rate can be obtained. It is speculated that the thickness and properties of the soft, gel-like layer depend on the pH as well as on the contact pressure. The removal rate of silica under purely mechanical conditions (at low to neutral pH) is less than a factor of two lower than those obtained under alkaline pH conditions. This indicates that oxide CMP is more mechanical in nature than metal CMP. High-planarity polishing is typically observed for slurries that ex hibit linear variation in removal rate with a change in applied pressure ( Papp). The planarization capability of slurry is related to the sensitivity of t he removal rate to high and low regions on the 43

PAGE 44

wafer. z is the distance between highest and lowe st regions, and then the planarity of the removal rate can be directly related to dRR/ d z, or the rate of change in the removal rate with the variation in surf ace height. This parameter can be further expressed as10 zd dP dP dRR zd dRR (2-8) where P is the local pressure on the polis hing surface, which is directly proportional to Papp. The first term in the product is strongly dependent on the characteristics of the slurry, while the second term is dependent on the mechanical properties of the pad. The values of dRR/ dP are typically enhanced by having a thin layer that exhibits pressure-dependent ma terial-removal characteristics and harder pad that transfer a significant portion of the app lied pressure directly to the abrasives. Selectivity Another important criterion fo r STI CMP is selectivity, wh ich represents the ratio of material removal rate (MRR) of silicon oxide to silicon nitride. Generally, a high selectivity value is desired because the CMP process needs to stop once the silica layer is removed. To enhance selectivity, the s ilica layer is typically polished by applying chemical/mechanical action and ensuring that chemicals do not extend their chemical assisted synergistic effects to the underlying layer, caus ing it to be mechanically removed. For STI CMP, conventional silicabase slurries achieve removal selectivity value of silica-to-silicon nitr ide layer in the range of 3 ~ 4.10 These low values can lead to extensive loss of nitride thickness, espec ially for large pattern density variations across the die. Recently, the removal selectivity value has been significantly increased due to reduced mechanical and chemical effects on the silicon nitride layer with the use 44

PAGE 45

of ceria-based slurries. A preferred STI proc ess can be achieved by driving the removal rate of the protective nitri de layer as low as practical while maintaining a reasonable rate for the fill oxide. Additionally, by suppressing the nitride removal rate, issues associated with pattern dependent nonunifo rmity with CMP can be reduced or minimized. Thus, selected additive and acid ic polymer can be added to ceria-based slurry. Purely mechanical action on the underly ing layer can result in high defectivity, especially in soft materials such as low-k dielectrics. Methods to reduce the mechanical component of the slurry, for ex ample, by the use of even smaller particles or softer abrasives, may be required in the future. Surface Defectivity Another important aspec t of CMP processing is surface defectivity. Defectivity issues include surface scratches, indentations, surface roughness, dishing, particle adhesion, and corrosion. Figure 2-9 shows the CMP defectivity for wafer surface. Among these defects, surface scratches are ty pical defects of the CMP process and are produced mainly due to the aggregates of sl urry. One of the key ways to control microscratching is to control the size of the abrasive particles in the slurry and the size distribution of the particles (especially the proportion of lager particles in the slurry). The indentation depth and stresses are related to the mechanical properties of the interacting surfaces, the particle size, and the layer thickness. If one assumes elastic contact between the particles and t he surface, the indentation depth as a function of particle size is given by10 3/224 3 KE Papp (2-9) 45

PAGE 46

where K is the particle fill factor at the surf ace and E is the Youngs modulus of the surface layer. Thus, equation 2-9 clearly shows that indentation depth is directly proportional to particle size. Also, both t he length and depth of microscratches are expected to increase as particle size increa ses. Besides the aggregates in the slurry lead to other complicating factors during CMP process. These include the timedependent aggregation of particles pH, drift, and issues related to the long-term stability of the slurry. Aggregation issues can lead to larger overall particle size. Depending on the shape of the aggregates and the aggregate strength, the wafer can be subjected to higher contact stresses, resulti ng in increased defectivity. If the slurry is unstable, the particles can settle on the wafer surface, resulting in a higher density of microscratches and increased particle adhesion t hat may be difficult to eliminate during post-CMP cleaning. To prevent large abrasiv es in slurry, filters can be used on the slurry line. The drawbacks of filters in the sl urry line, however, are that the slurry flow rate may decrease as the filter is being clogged and the removal rate can decrease as the filter approaches its end of life. A choice of a filter with the correct membrane pore size can prevent the shift in removal rate The pore size chosen must be significantly larger than the slurry particle size. Additionally, in order to decrease in wafe r scratches, the ideal particle is one that should be softer than the substrate material in choosing the abrasive for a particular CMP process. For ceria-based slurry, ceri a abrasives induce defects on the wafer surface because of the large, abrasive agglo merated particles resulting from poor dispersion stability. During CMP, the agglom erated particles easily stick to the wafer surface, which results in residual parti cles and scratches on the wafer surface.53-56 In 46

PAGE 47

order to reduce the surface defects, parti cle agglomeration should be prevented by the stabilization of the ceria suspension. In gener al, the dispersion stability can be improved by adding an adequate polymeric dispersant. Ho wever, the stabilizat ion of the ceria suspension using a polymeric dispersant has limitations in successfully eliminating the agglomerated particles which can induce scratc hes on the wafer surface. Therefore, a new method which can prevent the agglomerat ion of the ceria particle must be proposed in terms of intrinsic properties of ceria particles. The primary problem which occurs during STI is known as dishing. Dishing occurs when the oxide in the trench is polis hed at a higher rate t han the nitride. Dishing is a result of the fact th at pressure and pad flexibility combine to polish the recessed areas. If the pad were complete ly stiff, then theor etically a completely planar surface would be the result. In the subsequent HF stri pping of the nitride layer, some of the trench oxide will also be removed. The danger here is that the trench oxide could ultimately be reduced to a lower level than t hat of the active silicon. This causes a problem known as poly wraparound when the poly silicon gates are deposited. CMP processes for STI, including pad and slurry selection, must be optimized in order to avoid these problems. 47

PAGE 48

Slurry Feed Backing Film Retainer Ring Wafer Pad Table Conditioner Down Force Oscillation Wafer Rotation Head Figure 2-1. Schematic for CMP process 48

PAGE 49

Figure 2-2. Schematics of ideal oxide ILD CMP 49

PAGE 50

Figure 2-3. Oxide removal mechanism by CMP1,12 50

PAGE 51

Figure 2-4. Shallow trench isolation (STI) CMP process 51

PAGE 52

024681012 -80 -60 -40 -20 0 20 40 60 80 Zeta potential (mV)pH Silicon oxide Cerium oxide Silicon nitride Figure 2-5. Zeta potential of the oxide/nitride substrat es and ceria abrasive as a function of pH29 52

PAGE 53

Figure 2-6. Silicon nitride SN2 hydrolysis reaction scheme28 53

PAGE 54

(a) (b) Figure 2-7. (a) Zeta potential of the oxide/ nitride substrates and ceria abrasive as a function of pH29 and (b) the formation of passi vation layer on the surface of STI structure with ani onic organic polymer 54

PAGE 55

(a) (b) Figure 2-8. The comparison on (a) removal rate for oxide substrate with different abrasives1 and (b) Mohs hardness of ceri a and materials to be polished during CMP 55

PAGE 56

Figure 2-9. The schematics of surface defectivity 56

PAGE 57

57 Table 2-1. Guiding principles for slurry design in chemical me chanical planarization10 CMP issues CMP parameters Removal rate Rapid formation of the thin surface layer Control of the mechanical/i nterfacial properties of the surface layer Stress induction by abrasion to remove the surface layer Indentation-based wear Fracture/delamination-based removal Global planarization Formation of a thin passivated surface layer Minimization of chemical etching Minimization of mechanical polishing Selectivity Top-layer chemical/mechanical polishing Bottom-layer mechanical polishing Reduction of mechanical component in slurry Surface defectivity Rapid formation of a thin surface layer Minimization of mechanical polishing Control of particle size and hardness Control of particle size distribution

PAGE 58

CHAPTER 3 EXPERIMENTAL FACILITY AND PROCEDURES Introduction This chapter is devoted to the studies conducted on the prepar ation and various analysis methods of consumables used in this investigation, the polishing experiment, and the measurement for CMP performance. Especially, a ccurate measurement for consumables and wafers after CMP experim ent plays a significant role to an understanding the nature betwe en abrasive particles and wafer to be polished. Therefore, various analysis t ools were employed in this investigation. For abrasive particles, the synthesized particles were analyzed by XRD, FESEM, TEM, XPS, FTIR, TG/DTA and BET. For slurry, slurry stabilit y, mean particle size, and zeta potential measurements are reported usin g a variety of light scatte ring instruments. For CMP evaluation, film thickness, selectivity, nonuniformity, and defectivity of the polished wafer were measured by metrology tools us ing in current integrated circuit (IC) fabrication plants in order to support polishi ng results obtained by this investigation. Brief descriptions of measurement principles for each facility are also presented in this chapter. Sample Preparation Wafers For blanket wafer test, silicon dioxide films of 2 m thick were formed on a 5-in. ptype silicon substrates with (001) orientation by plas ma enhanced chemical vapor deposition (PECVD). Silicon ni tride films were deposited wit h the thickness of 7000 by using low-pressure chemical vapor deposit ion (LPCVD). These wafers were obtained from Seyoung semi-tech. For the patterned ca se, the SKW-1 pattern wafer designed by 58

PAGE 59

SKW Associates (Santa Clara, CA) was used for characterization with respect to the pattern density and pitch size.57 The STI mask consisted of 4 mm 4 mm density and pitch structures dividing the 20 mm 20 mm die into 5 rows and 5 columns. Figure 3-1 illustrates the specially des igned layout of the SKW-1 pa ttern wafer, including the pattern density and pitch size layout, the mask floor plan, and a cross-sectional view. In the density structure, density is defined as the trench area over the total area or expressed as ][ AWTW TW density (3-1) where TW is trench width and AW is active trench. The pa ttern density is varied systematically from 0% to 100% in increments of 10%, with a fixed pitch of 100 m. The density structures are fabricat ed in a random layout to plac e high-density regions next to low-density regions. In the pitch structure, the density is fixed with the same trench width and space (50%), and the pitch is va ried from 1 to 1000 m, with vertically oriented lines. Abrasive particles In this experiment, differ ent types of particles were used as abrasives for CMP slurry. The particles were synthesized by in duced-surface precipitation, hydrothermal, flux, and thermal decomposition method. T he resultant particles have different characteristics in terms of morphology, cryst alline structure, mechanical hardness, and thermal stress etc. The details for preparati on methods of particles were explained in each chapter. 59

PAGE 60

Slurries Different slurries were formulated by dispersing abrasives each with different abrasives in deionized water containing an anionic organic polymer (Poly acrylic acid, PAA; Mw 4000, LG Chem.) as dispersant. The PAA was 2 wt% based on the total weight of the used abrasives. For each slurry pH was adjusted to 6.5 ~ 6.7 by adding ammonium hydroxide (NH4OH). The solid loading of the us ed abrasives was fixed to 2.0 wt%. CMP Equipment CMP Polishers Polishing tests were performed on a ro tary type CMP machine (GNP POLI 400, G&P technology) for one minute with each of the slurries. IC 1000/SUBA IV stacked pads (supplied by Rodel Inc.) were utilized as CMP pads. The polishing pressure, applied as a down force, was 280 g/cm2. The relative velocity between the pad and the wafer was 90 rpm. During polishing, the slurry is continuously stirred by a magnetic bar, and pumped to the pad-wafer interface at a flow rate of 100 mL/min. The pad was conditioned before each polis hing run with a grid-abrade diamond pad conditioner manufactured by TBW. The diamond conditioner minimizes pad glazing. Figure 3-2 shows the schematic of the CMP tools us ed in this experiment. The average polishing data for removal rate were calculated by performing the same test more than three times in order to support the validity of t he results from the st atistical viewpoint. Slurry Delivery System Slurry performance on CMP process can be influenced by the complex delivery systems that are employed in the Lab. Pumps used for re-circulation and filters used to remove large particles and aggregates, defect causing particles can impart increased 60

PAGE 61

shear to the slurry and provide enough energy to the particles that they surmount their repulsive force barriers and come into c ontact. Figure 3-3 shows a slurry delivery system, which consisted of filtering system, pumping devices and a slurry delivery loop. CMP profile II filter capsule (Pall Microelectronics Inc.) was used as a filter to remove aggregates and larger particles. The filter size was 3.0 m and the slurry on the tank was circulated three times by using diaphragm pump. Characterization and Method Abrasives Microstructure and shape Morphological characteristics of the abrasiv e particles have been identified as the important parameters that influence CMP performance. The morphology and size of particles were analyzed by a field emissi on scanning electric microscopy (FESEM, JEOL JSM-6335F) at an accelerating volt age of 15 kV. The samples for FESEM analysis were prepared by dropping the dispersed samples on the sample holder and then deposited with carbon coat ing under vacuum condition. The average primary particle size was determined by FESEM micrographs with counting more than 100 particles. The average particle size ( dSEM) can be calculated as T i SEMN d d (3-2) where the di is identified with the indi vidual particle size and the NT is the total number of particles measured from SEM microscopy. The large magnified microscopy and linear structures of the precipitates were observed by a transmission electron microscopy (TEM, JEOL TEM-2010F) at an accelerating voltage of 120 kV. The samples for TEM analysis were prepared by dispersing the final particles into distilled 61

PAGE 62

water under ultrasonic treatment and t hen the dispersion was dropped on carboncoated copper grids. Physical properties X-ray diffraction is an extremely important technique in the field of materials characterization to obtain information on an atomic scale from both crystalline and noncrystalline (amorphous) materials. Espec ially, the crystalline size of abrasive particles has influence on CMP performance su ch as removal rate and the defects of film polished. The crystal structure was identified through x-ray diffraction (XRD, Philips APD 3720) using CuK radiation ( =0.154 nm in this study). The accelerating voltage and the applied current were 40 kV and 20 mA respectively. The crystalline size ( dXRD) of samples can be estimated from XRD pa tterns by applying full-width half-maximum (FWHM) of characteristic peak to Scherrer equation as5 cos 9.0XRDd (3-3) where dXRD is grain size, is the wavelength of x-rays, is the half-width of the diffraction peaks, and is the diffraction angle. For ceria particles, the broadening of the (111) peak in XRD was analyzed to confirm the primary grain size of particles. Thermogravimetric and differential thermal gravimetry (TGA/DTG, Mettler Toledo 851) analysis was performed in an air flow of 100 ml/min at a heating rate of 10 oC min-1 from room temperature to 800 oC. The typical sample quantity was between 80 100 mg. Surface and chemical properties The chemical composition of the samp les was determined by X-ray photoelectron spectroscopy (XPS, Physical Electronics Perkin-Elmer PHI 5100). The XPS measurements were performed with a non-monochromatic MgK or AlK source at a 62

PAGE 63

base pressure of 5 10-1 mbar. The analysis chamber included an aluminum anode. As the sample were insulating, the energy ca libration was achieved by setting the binding energy of carbon at 284.6 eV. Fourier transfo rm infrared (FTIR) spectroscopy (Thermo Nicolet Magna 760) was used to determine the chemical interaction between the core particles and coating materials. The FT-IR s pectra for the samples of investigation were carried out in the range 4000 500 cm-1. Infrared absorption spectra were recorded for KBr disk containing the synthesized particl es and the FT-IR cell was purged for 20 min prior to spectral collection. FTIR spectr oscopy was used with a dry air purge. The specific surface area of ceria abrasives was determined by Brunauer-Emmett-Teller (BET, Micrometric ASAP 2010) method using nitr ogen adsorption/desorption at 77 K. All samples degassed at 200 oC prior to measurement. The BET surface area was determined by the multipoint BET method wit h the adsorption data in the relative pressure ( P / P0) range of 0.05.25. Assuming that the samples are spherically and noporous, the corresponding particle size (dBET) can be estimated as A dBET6 (3-4) where A is specific surface area and is the true density of sample. For ceria abrasives, is 7.28g/cm3. Slurries Electrical potential (Zeta Potential) Electrophoresis measurements were used to obtain the electrophoretic mobility and isoelectric point (IEP) of the prepared particles. These electrostatic behaviors of particles play an important role in the preparation of the c oated particles and dispersion stability of abrasive particles in CMP sl urry. The zeta potential (type Brookhaven 63

PAGE 64

ZetaPlus) of particles was measured by utilizi ng an electrophoresis method to measure the electrophoretic velocity ( E) of colloidal particles.58 When the electrical double layer ( 1/k ) is much smaller than particle size, the Smoluchowski equation can be applied to measure the zeta potential ( ). 0EE (3-5) where and are the values of viscosity and diel ectric constant in the solution and E0 is the electric field. If t he electrical double layer (1/k ) is much larger than the particle size, the Huckel equation can be applied, and expressed as 02 3 EE (3-6) An electrolyte solution (1 mM) was used to keep the ionic strength constant while the pH value was varied by adding 0.01 N KOH or HCl into solution. The pH of the suspension was taken as the isoelectric point (IEP) at which the zeta potential was zero. Particle size distribution Average particle size of abrasive parti cles in CMP slurry was measured by dynamic light scattering method (type H oneywell Microtrac UPA 150). The light scattering technique has been commonly employed for measuring particle size below 1 m. This technique can be divided into two groups, dynamic light scattering (or quasielastic light scattering or photon correla tion spectroscopy) and low angle laser light scattering (or laser diffraction). Dynamic light scattering can quickly and accurately measure the particle size range from 0.8 nm to 6.5 m. When the particle size is smaller than a submicron, colloidal particles have a behavior of Brownian motion due to thermal 64

PAGE 65

vibration. Thus, for a monodisperse distribution of spherical particles, an intensity of autocorrelation function is employed fo r analyzing the intensity fluctuations. )exp()(2tDq tg (3-7) where g(t) is the normalized autocorrelation function, and D q and t are the Brownian diffusion coefficient (Stokes-Einstein relation), scattering vector, and delay time of the autoco rrelation function. d TK DB3 (3-8) where KB is Boltzmann constant, T is the absolute temperature, is the liquid viscosity, and d is the particle diameter. Thus, sma ller particles move more rapidly and have more intensity fluctuations than bigger particles. The laser diffraction has been commonly employed for detecting large particles. When a particle size is larger than incident wavelength, the di ffraction phenomenon occurs while light beam interacts with particles. The smaller particles have a hi gher angle of diffraction than bigger particles.59 Polished Wafer Film thickness measurement Thickness measurements of films before and after CMP were measured by using a Nanometrics NanoSpec 6100. The NanoSpec is an in strument for measuring the thickness of optically transparent thin films ( photoresist or oxide etc) on silicon wafers. It uses reflectometry or measurement of refl ected light to determine film thicknesses based on interference effects. The basic opera ting principle is that the intensity of monochromatic reflected light depends st rongly on film thickness because of interference (the film thicknesses are comparabl e to the wavelength of the incident light). The machine uses a computer-controlled gr ating monochromator and a photomultiplier 65

PAGE 66

tube detector to measure the reflected opt ical spectrum (over the 350 to 800 nm wavelength band) from a bare silicon referenc e wafer and from the wafer under test. Given an index of refraction for a thin f ilm and the two measured spectrums, the computer will analyze the interference pattern to determine film thickness. The equation for describing the interference is )( 20 fs ig n (3-9) where X0 is film thickness, is the wavelength (in vacuum) of the incident radiation, s is the relative phas e shift at the SiO2/Si interface, f is the relative phase shift at the air/SiO2 interface, ni is the index of refrac tion of the thin film, and g is the order of the interference. The intensity is a maximum when the bracketed term is an integer and a minimum when it is an integer plus 1/2. In this experiment, the thickness of each silicon dioxide film was measured wit h the NanoSpec using a refractive index of 1.46. To determine a film thickness, the instrument scans the film measuring the change in intensity as a function of wavelength. The spec tral fluctuation in intensity results from constructive and destructive interference of the electromagnetic waves as they travel through the film and reflect off the underlying substrate. From the spectral intensity distribution, the thickness of the deposit ed film can be computed. Figure 3-4 shows schematically the principles of film thickness measurement using NanoSpec 6100. Selectivity between silicon dioxide and nitride Chemical mechanical planarization (CMP) has enabled replacement of previously used device isolation method, local oxidatio n of silicon (LOCOS), by STI. The major requirements of the STI CMP step are: comple te removal of silicon dioxide over silicon nitride layer (nitride layer acts as a CMP stop and protects the active device area); 66

PAGE 67

minimal nitride erosion to prevent damage to the underlying silicon material in active areas; and minimal dishing of the trench oxide. Among these requirements, the relationship between removal rate of oxide and nitride as st op layer is one of the most important issue in STI-CMP process. Usually this has been designated as selectivity, which represents the ratio of material remo val rate (MRR) of silicon oxide to silicon nitride: nitride siliconofrate removal Material silicaofrate removal Material ySelectivit (3-10) In this experiment, the effects of various abrasive particles on removal selectivity were investigated. Oxide CMP within-wafer nonuniformity (WIWNU) The primary purpose of using CMP is to pl anarize the surface of film. WIWNU is of critical important in ev aluating equipment, developi ng and comparing process, monitoring process performance. For instanc e, the edge effects, namely, the rapid variation of the material removal rate at the edge, require an ex clusion of the wafer edge after CMP. This reduces the yields of the process. The uneven material removal rates across the wafer will bring the over-polis hing in the faster removal regions in the shallow trench isolation (STI) and copper damascene processes. This causes a degeneration of the circuit performance in that area. In addition, WIWNU will bring a systematic variation of the circuit performance across th e wafer. A better understanding of the formation mechanism of the WIWNU will be able to increase the yields and help to optimize the circuit performances. T he WIWNU can be expressed as follows:50-52 %100) (min max avgMRR MRR MRR WIWNU (3-11) 67

PAGE 68

where, MRRmax is maximum material removal rate, MRRmin minimum material removal rate, and MRRavg is average material removal ra te. The non-uniform removal rate across the wafer in CMP can be attributed to the uneven distribution of a number of parameters such as the temperature and slurry distributions. However, if the distributions of the pressure P and velocity V are the two major contributors, equation 310 can be revised by Prestons Equation of material removal rate MRR = KePV+ MRR0:50-52 %100 ])()[( %100) )( )( )( (min max 0 0 min 0 max avg e avg e e eMRR PV PVK MPR PVK MPR PVKMPR PVK WIWNU (3-12) where Ke and MRR0 are two experimental fitting par ameters. In this experiment, the pressure and velocity was fixed in order to investigate the effects of abrasive particles on WIWNU during CMP process. T herefore, the WIWNU of material removal rate was measured by equation 3-10. Defectivity monitoring by wafe r defect scattering analysis A common technique for measuring defects on unpatterned wafers employs a laser to scan across the entire wafer. This is designated as wafer defect scattering analysis. If a defect is present, light is sca ttered away at the point of incidence. A photomultiplier tube collects the scatted ligh t, whose magnitude is proportional to the size of the particle. Laser scanning can also be used to detect defects on patterned wafers using dark field light scattering. Sca ttered laser light is detected using chargecoupled diode (CCD) camera. Through an image processor and the intensity calculations, defects can be detected and quant ified on the wafer. Another method of 68

PAGE 69

defect detection is digital image comparison. The tool works by com paring a pixel of one die to that of the prec eding and succeeding die. If there di fference in contrast value, that coordinate is flagged as a defect. In this expe riment, defects of pattern wafer after CMP were measured by using a KLA-Tencor puma 91XX. This system uses a UV/visible light source to illuminate defect types. Figure 3-5 shows schematically the principle of wafer defect scattering analysis.60 This system is as follows: a UV laser beam is illuminated onto the wafer surface from above, scattered light from a defect is collected by a receiving lens, and the scattered light is converted to an electrical signal by a detector. The wafer is set on the rotating st age, and by moving the stage in the radial direction while it rotates, the whole wafer surface can be inspected at high speed. And by fixing an encoder to the st age, positional data of the wa fer defect can be obtained. 69

PAGE 70

Figure 3-1. Layout of the SKW-1 pattern wafer: (a) pattern density and pitch size layout, (b) mask floor plan, and (c) cross-sectional view61 70

PAGE 71

Figure 3-2. Schematic diagram of rotational CMP tool 71

PAGE 72

Figure 3-3. Schematic illustrati on of a slurry delivery system 72

PAGE 73

Figure 3-4. Diagram of film thickness measurement system using NanoSpec 73

PAGE 74

Figure 3-5. Schematic illustra tion of light scattering analysis60 74

PAGE 75

CHAPTER 4 NOVEL METHOD TO CONTROL THE SIZE OF SINGLE CRYSTALLINE CERIA PARTICLES BY HYDROTHERMAL METHOD AND ITS CMP PERFORMANCE Introduction This chapter is devoted to the studi es on synthesis of single crystalline ceria particles by using hydrothermal method and its CMP performance. During the past decade, the usual method for synthesizing ceri a abrasives is thermal decomposition of the cerium salt such as ceri um carbonate and cerium hydroxide.8 These ceria abrasives have polycrystalline structure with an easily br ittle system, which affects a high removal rate of oxide films during polishing.35 However, the size and the shape of ceria abrasives are very limited since particle growth is difficult to control during calcination process. The irregular shape and aggregates lead to defects on the surface of film during CMP process. To achieve the desired particle size and the uniform particle size distribution, thermal technique requires a co mplicated combination of mechanical milling process and filtration system.41 Other methods for preparing ceria abrasives are liquid phase processes.62,63 Other methods for preparing ce ria abrasives are liquid phase processes, such as precipitation method, 42 hydrothermal method, 43-45 sol-gel method46 and electrochemical method. 47 Among these approaches, hydrothermal synthesis is an attractive method for the prepar ation of crystalline ceramic ox ide particles and has been employed for the synthesis of fine particles at relatively low temperature. The hydrothermally synthesized particles have e xcellent homogeneity and particle uniformity without post-heat treatment. However, the si ze of ceria particles obtained from conventional hydrothermal synthesis is lim ited to less than 100 nm leading to a low removal rate during CMP process. 75

PAGE 76

To overcome this problem, this research introduced a new type of ceria precursor in hydrothermal synthesis. This precursor was synthesized by alkoxide method, which employs alcohols with different dielectric pr operties as solvent. Over the past years, several researches using the mixed solvent system showed the possibility to increase the controllability of cera mic morphology. Park et al.64 introduced the alcohol/water mixed solvent to thermally hydrol yze titanium tetrachloride (TiCl4) in the preparation of spherical titania (TiO2) particles. The DLVO ( Derjaguin Landau Verwey Overbeek ) theory was proposed to explain the effect of the alcohol/water mixed solvent on the morphology of the parti cles. Fang et al.65 found that the morpholog y of titania particles was controlled by the dielectric constant of the solvents, which can be regulated by changing the volume ratio of npropanol to water. Hu et al.66 employed the alcohol/water mixed solutions for synthesizing zirconia par ticles. They found that the dielectric property of the mixture affe cts the nucleation and growth of the zirconia particles. However, these reports about the mixed solv ent system were focused on formation of metal oxide through precipitation reaction. In addition, synthesis of single crystalline ceria particles of more than 100 nm under solution phase has not been reported yet. Furthermore, previous studies for the po lishing performance of ceria abrasives have limited to polishing behaviors of polycrystalli ne ceria particles for silica layer and have mainly focused on the chemical interactions between ceria abrasives and silica layer during CMP process. Therefore, the first purpose of the present work is to co ntrol the morphology of the ceria crystallites under hydrothermal conditions using a new type of ceria precursor obtained by precipitation method and investigate the effects of the dielectric property of 76

PAGE 77

organic solvent on formation of ceria parti cles. In addition, the influences of the hydrothermal temperature and an acidic hydrothermal medium on particle characteristics have been studied. This study will provide the new approach to modulate the particle size of ceria crystallites usi ng hydrothermal method. The second purpose of the present work is to investigate the effe cts of single crystalline ceria abrasives on silicon oxide and silicon nitride CMP. The singl e crystalline ceria particles as abrasives for ceria-based slurry have seldom been applied to CMP processing. In this chapter we synthesized the single crystalline ceria par ticles of different mean sizes using hydrothermal method. For CMP performance evaluation, the ef fect of single crystalline ceria abrasives on the removal rate, the ox ide-nitride removal selectivity and withinwafer nonuniformity (WIWNU) was investigated. Materials and Methods Abrasives Preparation of sol-type ceria precursor Cerium (III) nitrat e hexahydrate (Ce(NO3)3H2O) and potassium hydroxide (KOH) were used as the starting materials for t he ceria precursor. Cerium (III) nitrate hexahydrate and potassium hydroxide were s eparately dissolved in distilled water with a desired concentration and then mixed with the various kinds of alcohols. The volume ratio of alcohol to water was kept at 2:3. Alcohols, including methanol (CH3OH), ethanol (C2H5OH), ethylene glycol (C2H6O2) and 1, 4-butylene glycol (C4H10O2) were used to investigate the influence of the dielectric constant of the mixed solvents. The reaction was carried out at a temperature of 50 oC with stirring rate of 100 rpm for 12 h. Air was bubbled into the precipitation reactor with passage through a gas distributor as an oxidizer. The precipitated ceria precurso r was separated via centrifugation and then 77

PAGE 78

redispersed in distilled water under continuous stirring. The we ight ratio of distilled water to a precipitated ceria precur sor was kept 5:1. The pH of the suspension solution was adjusted at pH 0.5 to 4.0 by addi ng concentrated nitric acid (HNO3). After reaction, light yellow solution was obtained. Hydrothermal synthesis of ceria particles The sol-type ceria precursor was put into an autoclave with a reaction chamber of 100 cm3. Three quarters of the volume of the chamber was filled with the light yellow solution. The hydrothermal reactions were carried out at 150 to 230 oC for 6 h, corresponding to a pressure range from 200 to 800 psi. After the hydrothermal reaction, the synthesized particles were washed wit h distilled water three times and were subsequently dried at 90 oC. CMP Evaluation Preparation of ceria-based slurries Different ceria-based slurries were formu lated by dispersing abrasives each with different primary particle size in DI wate r containing an anionic organic polymer (Poly acrylic acid, PAA; Mw 4000, LG Chem.) as di spersant. 2 wt% of PAA based on the total weight of the ceria abrasives was added. For each slurry, pH was adjusted to 6.5 ~ 6.7 by adding ammonium hydroxide (NH4OH). The solid loading of ceria abrasives was fixed to 2.0 wt%. CMP tools and consumables Silicon dioxide film of 2 m thick was grown on a 5-in. p-type silicon substrates with (001) orientation by plasma enhanced chemical vapor deposition (PECVD). The silicon nitride films were deposited by usi ng low-pressure chemical vapor deposition (LPCVD). Polishing tests were performed on a rotary type CMP machine (GNP POLI 78

PAGE 79

400, G&P technology) for one minute with eac h of the ceria-based slurries. IC 1000/SUBA IV stacked pads (supplied by R odel Inc.) were utilized as CMP pads. The downforce was 4 psi and the rotation s peed between the pad and the wafer was 100 rpm. The slurry flow rate was 100 mL/min. Characterization Abrasives The crystal structure and grain size was identified by x-ray diffraction (XRD) using CuK radiation. The grain size was estimated by the Scherrer equation according to the formula D = 0.9 / ( cos ), where D is the grain size, is the wavelength of x-rays, is the half-width of the diffraction peaks, and is the diffraction angle. The broadening of the reflection from the (111) plane was used to calculate the grain size. The morphology and size of the precipitate parti cles were examined by high resolution transmission electron microscope (HRTEM) and field emission scanning electron microscope (FESEM). The aver age primary particle size was calculated by measuring ca. 100 particles from FESEM micrographs. The specific surf ace area (SSA) of the ceria abrasives was determined by Brunauer-Emmett-Teller (BET) method using nitrogen adsorption/desorption at 77 K. Ceria-based slurry The abrasive size distribution of slurry was measured using light scattering method (UPA 150, Microtrac Inc.). CMP performance The film thickness on the wafers before and after CMP was measured using spectroscopic reflectometry (Nanospec 6100, Nanometrics) to calculate the removal rate. In this experiment, the WIWNU was defined as the standard deviation of remaining 79

PAGE 80

thickness divided by the average of the rema ining thickness after the CMP process. The average polishing data for removal rate was carried by performing the same tests more than three times in order to support the valid ity of the results fr om the statistical viewpoint. Results and Discussion Preparation of Ceria Particles Influence of solvent type on ceria particle characteristics The XRD patterns of the synthesized particles are shown in Fig. 4-1. The 0.5 M of cerium (III) nitrate hexahydrat e solution and the 0.5 M of pot assium hydroxide solution were mixed with the various kinds of alcohol s in precipitation reaction. The synthesized ceria precursors were hy drothermally treated at 230 oC after adjusting to pH 3.0. The dielectric constant of alcoho ls decreases in the following order: ethylene glycol (41.4) > methanol (33.0) > 1, 4-butylene glycol (31.9) > ethanol (25.3).67,68 As shown in Fig. 4-1, the major reflections associated with fluorit e structure of ceria can be observed on all specimens regardless of the kinds of alc ohols used to prepare the ceria precursors. Fig. 4-2 shows the HRTEM image of the ceria particles which were synthesized from the mixed solvent of water and et hylene glycol. They appear to be single crystalline structure bas ed on the fact that the lattice fringes corresponding to reflections are clearly observed. Fig. 4-3 shows t he FESEM micrographs of ceria particles prepared in different mixed solvents. The synthesized particles were prepared without hard aggregates. It is interesting to note t hat the ceria particles prepared using higher dielectric constants show bigger morphol ogy in spite of using same processing parameters and steps. The average particle size of the ceri a particles increased in proportion to dielectric constant of alcohol us ed in the precipitation reaction as shown in 80

PAGE 81

Fig. 4-3. These results indicate that the alcohol affects the physi cal properties of the reaction medium without changi ng the reaction paths and arr angements of the crystal structure. This can be quantitatively seen from Fig. 4-4. The curve (a) depicts the average particle size whereas curve (b) shows the crystallites size with respec t to the dielectric constant of alcohols used in the mixed solvent. It is found that the size and the crystallites size of ceria particle increased wit h increase in the dielectric constant. These results indicate that the solv ent type may affect properties of the synthesized particles because different alcohols show different di electric constant and different affinity forwards water. The dielectric constant of a solvent is the quantitative measur e of its ability to decrease attraction between two oppositely char ged ions. The dielectric constant is defined by the free energy for the coul omb interaction between two charges. The relationship between the concentration of a saturated solution in equilibrium and dielectric constant can be expressed as following.69 )(4 exp)exp(0 2aakT ezz kT Xi l (4-1) where the Xl is identified with the solubility of solute in any solvent. i is the difference in energy when going from the as sociated state to the dissociated state of two systems. a+ and aare ionic radii of ions charged z+ and zand is the dielectric constant of the medium and e represent s the elementary charge (e = 1.602 X 10-19 C). As can be seen from the equation (1), the solu bility of the solute is larger when the dielectric constant of the solution is higher. Th is infers that the solubility is larger as 81

PAGE 82

increasing the dielectric c onstant and can be adjusted by changing the dielectric constant of solution. According to the classical nucleation theory,70 the particle size is dependent on the supersaturation of solute, which affects strongly the nucleation rate. This can be expressed in the form of t he Arrhenius velocity equation.70 2 33 23)(ln3 16 exp STk AJ (4-2) where J is the nucleation rate A is the rate constant, is interfacial tension between the solute and the solution, is the number of ion formula units, k is the Boltzmann constant and S is the supersaturati on of solute. This equation indicates that the increase in supersaturation induces a great number of nuclei due to the rapid increase of the nucleation rate. Additionally, the relations hip between solubility and the radius of nuclei can be expressed by the basic Gibbs-Thomson equation.70 rRT V S X Xm l 2 lnln (4-3) where X is the concentration of solution, S is the supersaturation of solute, is the density of the solid, Vm is the molar mass of the solid in the solution and R is the gas constant. With increasing solubility of the solute the supersaturation of the solute is decreased. Moreover, the nuclei radius is proportional to the solubility of the solute. From (1) to (3) equations, it can be found that the dielectric constant of solution affects nucleation rate and the radius of nuclei. Moreover, the size of crystallites can be modulated by varying the solvent type. 82

PAGE 83

In this work, the nucleation rate has been modulated by changing of the dielectric constant of mixed solvent. A number of nuclei are decreased with increasing the dielectric constant of alcoho ls used in precipitation reacti on. Therefore, the crystallites size and the morphology of the ceria particles increased with the incr ease of dielectric constant of the solvent. Effect of the precipitati on participating anions on nucleation and growth Fig. 4-5 shows FESEM micrographs of the ceria particles synthesized from different ceria precursors which were precipit ated in the mixed solvent of ethylene glycol and water using different concentrations of potassium hydroxide. The hydrothermal reactions were carried out at 230 oC for 6 h. As shown in Fig. 4-5, it is found that the size of ceria particles was strongly dependant on the potassium hydroxide concentration. Particle size was decreased with the increas e in potassium hydroxide concentration. The formation of ceria particles involv es a series of chemical reaction71 and the whole process can be classified into four stages.72 During these stages, hydrated cerium hydroxide complexes were generated and dehydrated. Precipit ation participating anions (OH-) were generated via the hydrolysis of potassium hydroxide with the molecular water of cerium sa lt. Precipitation participatin g anions were exhausted by hydration of tetraval ent cerium ions (Ce4+) and subsequently caused the decrease in the pH. Moreover, the formation of crystalline particles includes usually two steps: nucleation and growth. In order to prepare the particles wit h designed size, both steps should be controlled. In nucl eation step, the in crease of nuclei induces the smaller particle size. In growth step, secondary nucleation occurs in high supersaturation because crystal growth has a lower energy barrier than that of the nucleation. 83

PAGE 84

In order to investigate the effect of precipitation partic ipating anions on the formation of ceria particles, 0.5 M and 1.5 M of potassium hydroxide solution added in cerium salt solution and the pH of these so lutions was kept 7.8 and 11.6, respectively. The pH of the solution was changed to 4.4 and 9.1 after precipitation for 12 h, respectively. It is attributed to the fact that precipitatio n participating anions under basic solution are more slowly consumed during growth of nuclei. This result indicates that increasing precipitation participating anions induces continuously the large number of nuclei and the decrease of solubility which help to decrease the growth rate. Under basic condition, the solubility product is much higher than the solubility constant, meaning the supersaturation (S) is very large.73 spK OHCe S ]][[3 (4-4) where Ksp is the solubility constant of Ce(OH)3. A high supersaturation induces a great number of nuclei due to secondary nucleat ion. Therefore, the crystallites size and the particle size of ceria decrease with increasing precipitation participating anions. Effect of hydrothermal conditions The effect of an acidic hydrothermal medium on the formation of ceria particles is shown in Fig. 4-6. Ceria precursor was synt hesized using the mixed solvent of ethylene glycol and water with 0.5 M of potassium hydroxide solution. With decreasing the pH of the solution, the size of the ceria particle s increased under hydrothermal condition. This result is related to Ostwald ripe ning phenomena in the liquid phase system .74 Wu et al. found that the acidity of hydr othermal medium played a key ro le in dissolution of smaller particles, which directly influences the struct ure. In this work, the increase in hydrogen ions led to a sizable increase in the solubility of the ceria precursor. This implies that the 84

PAGE 85

solute diffuses quickly and the crystal growth is more rapid with dissolution of the cerium precursor in an acidic medium. The effect of the hydrothermal treatment temperature on the crystalline size is summarized in Fig. 4-7. The crystallites size was increased with an increase in hydrothermal treatment temper ature. Under hydrothermal c onditions, the concentration of the hydrogen ion (H+) is increased with increasing temperature.43,74 It appeared that higher temperature in hydrother mal reaction promotes the crystal growth in the ceria particles from the hydrated cerium hydrox ide complexes according to the dissolutionprecipitation mechanism. CMP Performance Ceria abrasives Fig. 4-8 shows the four types of abrasive particles prepared in different hydrothermal conditions as described in Table 4-1. The primary particle sizes determined in the FESEM examination were 62, 116, 163, and 232 nm for slurry A, B, C and D, respectively. The well-dispersed particles of spherical shape as shown in Fig. 48(a) were transformed into square shape as a result of grain growth which is clearly seen from FESEM image (Fig. 4-8(d)). These images indicate that the primary particle size increases with hydrothe rmal temperature and strength of acidic medium and the morphology of abrasives can be controlled by changing the hydrothermal conditions, which affect the grain grow th of ceria crystallites. The major reflections associated with c ubic fluorite structure of ceria can be observed from the XRD pattern as shown in Fig. 4-9. Sharp intensity peaks are observed for ceria abrasives with increased primary particle size. The average grain size of various abrasives were calculated from the Scherrer equation by using the line85

PAGE 86

broadening of the (111) peak in XRD pattern. The grain size gradually increased from 29 nm to 66 nm as the hydrothermal te mperature and the acidity of hydrothermal medium were increased. This result coincides with the trend of increasing primary particle size in the FESEM images shown in Fig. 4-8. Fig. 4-10 shows the high resolution TE M images of two samples with average particle diameters of 62 nm and 232 nm, respec tively. As shown in the Fig. 4-10(a), the sample with diameter of 62 nm appears to be well-defined crystallites based on the fact that the lattice fringes corresponding to the (111) reflections are clearly observed from the crystal orientation. For 232 nm ceria particle, the TEM image (Fig. 4-10(b)) shows homogenous single phase of the lattice fringes without disorder and defects in the lattice. These results indicate that ceria abrasives used in this study have a single crystalline structure regardless of grain growth and particle size. Well-crystalline ceria (CeO2) particles were synthesized by using sol-type ceria precursor under hydrothermal conditions at pH 0.5 to 4.0. The ceria precursor was prepared by chemical precipit ation in mixed solution of water and alcohols, including methanol, ethanol, 1,4-butylene glycol and ethylene glycol (EG), separately. The resultant particles exhibit cubic fluorite structure with size ranged from 20 to 400 nm. The particle sizes were determined by X RD and SEM analyses. The results showed that the crystallites size and the morphol ogy of the hydrothe rmal ceria particles increased with an increase in the dielectric constant of alco hols used in precipitation reaction and hydrothermal treatment tem perature and a decrease in the pH of hydrothermal medium, which affect nucleat ion rate and crystal growth. Consequently, 86

PAGE 87

the size of ceria particles was easily c ontrolled in the range fr om 20 nm to 400 nm without post-heat treatment. Characteristics of ceria abrasive before and after CMP To evaluate the effects of single crystalli ne ceria abrasives, Polishing tests for four types of slurries with different abrasive parti cle size were preformed. Fig. 4-11 shows the particle size distribution of different slurries without PAA dispersant. The size distribution of secondary parti cle size determined by light scattering method were 178, 273, 326, and 484 nm for slurry A, B, C and D, respectively. The dispersed particle size is much larger than the crystallite size esti mated by X-rays and the primary particle size calculated by FESEM. This mismatch in size is due to extensive overlapping of the ceria particles in water-based solution.75 As shown in Fig. 4-11, the increase in size of the primary particles led to broader particle size distribution in water-based solution. Fig.412 shows the FESEM images of ceria abr asives before and after silicon dioxide polishing for slurry D. The ceria abrasives after polishing were washed with distilled water three times via centrifugation. Accord ing to Fig. 4-12, no definitive difference between both abrasives before and after polishing can be seen except that square edges of some abrasives were changed to round edges after polishing. This indicates that the single crystalline ceria abrasives syn thesized in this study are less brittle and dont fracture upon applied pressure during polishing. Polishing performance Fig. 4-13 shows the results of CMP field evaluation and quantitat ive results of the slurry are presented in the Table 4-2. For oxide CMP process, it is clear that the removal rate increases with increasing size of ceria abrasives. The polishing of oxide film is mainly affected by the chemical c ontribution of ceria particles and mechanical 87

PAGE 88

factors, such as the CMP conditions, morp hological characteristics of abrasives and particles size distribution. During the polishing of the oxide film, ceria abrasives exhibit a chemical tooth property.12 As a consequence, chemical interaction between the silica film and the ceria abrasives occur during CMP process and Si-O-Ce bonding is formed on the surface of silica film. The Si-O -Ce bonds can be rapidly removed by the mechanical force generated by pressed pad. This physicochemical reaction leads to the high removal rates of oxide film. Furthermo re, polishing behavior is closely associated with shape and size of the ceria abrasives. As shown in Fig. 4-8, the shape of the small abrasives (Fig. 4-8(a) and (b)) is spherical, whereas the large abrasives (Fig. 4-8(a) and (b)) have a square shape with sharp edges. The sharp edge and large size of abrasive particles can induce higher local pressure to generate more frictional force during polishing. The removal rate of nitride film increas es with the increase in the abrasive size. According to previous report,76 the removal rate of nitride f ilm is affected by the physical properties of ceria-based slurry systems and t he amount of surfactant adsorbed on the film surface. In order to im prove the selectivity and uniformity, an anionic acrylic polymer is generally used to passivate the surface of the nitride film during STI-CMP, which prevents ceria abrasives from contacting the film surface.30 In this study, the amount of polymer added was same for all the slurries. T herefore, it seems t hat the increase in removal rate of the nitride film is related to the mechanical factors rather to the effect of adsorbed polymer passivation layer. These mechanical factors are influenced by several physical parameters of the CMP process, such as morphological aspects of the abrasives, crystallite size of the ab rasives and the CMP conditions. 88

PAGE 89

The removal selectivity can be calculated by comparing the removal rate between oxide and nitride film. As shown in Fig. 4-13, the removal selectivity showed a transition behavior at the slurry C. The removal rate of oxide film rapidly increased from 1174 /min at 62 nm to 2369 /min at 163 nm, whereas it slowly increased to 2674 /min for 232 nm. This phenomenon is explained by two related factors: the contact area reduction17 and the particle surface activity.76 In the case of oxide CMP process, the polishing behavior should be also considered from the viewpoint of the contact area between the abrasives and the film surface. According to contact area mechanism,16,54 the removal rate increases with decreasing abrasive size and increasing solid loading, due to the increase in contact area between the abrasives and the f ilm surface. At a fixed solid loading, the number of ceria abrasives in slurry decrease as the abrasive size increases which leads to relatively low remo val rate for the target material during CMP process even though the abrasive size increas es. For nitride film, the removal rate increased slightly and did not vary much with increasing abrasive size, which can be explained by the relationship between additive polymer and abrasive size. The additive polymer can be more easily attached on the su rface of the small abrasives than on the surface of large ones, due to high surface acti vity and specific surface area of the small abrasives. The slurry with small abrasives can induce a relatively high removal rate for nitride film in comparison with its abrasive size, since the passivation layer is insufficiently formed on the nitride film su rface as polymer is largely adsorbed on the particle surface. As described in Table 1, the specific surface area decreased with increasing abrasive size. This means that remo val rate for nitride film can be relatively increased in spite of decrease in the abrasive size. As a result, the removal rate of 89

PAGE 90

nitride film showed relatively low increase in rates with increase in the abrasive size. Such behavior is suggestive of existence of an optimum size of ceria abrasives for high removal selectivity. Additionally, the results for surface uniformity of the oxide films ar e shown in Fig. 4-14. The slurry shows a higher WIWNU for t he oxide film with increase in the size of ceria abrasives. This polishing behavior is attributed to a broader particle size distribution of the slurry and the shape of the large abrasives The distribution data in Fig. 4-11 shows that the particle size distri bution broadens with the increase in abrasive size. The broader particle size distribution of large abrasives can cause different removal rates between the center and the edge of the wafer due to t heir limited mobility on the wafer surface. This result is also consistent with the obser vations of Moudgil et al.54 They investigated the polishing mechanis m of slurry with non-uniform particle size distribution, which not only created surfac e deformation but also changed the polishing removal rate. In this report, the sharp edge of the abrasives can be regarded as another factor for roughness of wafer. The film abraded by the s harp edge has a higher local pressure to generate more friction force duri ng polishing. This can induce a significant increase in the local surface roughness caused by pit formation on the wafer surface.15 As a result, the surface uniformity shows det erioration with increase in abrasive size. Therefore, it seems that surf ace uniformity of oxide film is related to the mechanical factors and morphological properties of ceri a abrasives and particle uniformity of the ceria-based slurry. Interestingl y, there exists an optimum ab rasive size distribution at which enhanced removal rates and selectivity are observed. 90

PAGE 91

Conclusions Synthesis of Ceria Particles by Hydrothermal Method Well-crystalline ceria (CeO2) particles were synthesized by using sol-type ceria precursor under hydrothermal conditions at pH 0.5 to 4.0. The ceria precursor was prepared by chemical precipit ation in mixed solution of water and alcohols, including methanol, ethanol, 1,4-butylene glycol and ethylene glycol (EG), separately. The resultant particles exhibit cubic fluorite structure with size ranged from 20 to 400 nm. The particle sizes were determined by X RD and SEM analyses. The results showed that the crystallites size and the morphol ogy of the hydrothe rmal ceria particles increased with an increase in the dielectric constant of alcohols used in precipitation reaction and hydrothermal treatment tem perature and a decrease in the pH of hydrothermal medium, which affect nucleat ion rate and crystal growth. Consequently, the size of ceria particles was easily c ontrolled in the range fr om 20 nm to 400 nm without post-heat treatment. CMP Evaluation In this study, we investigated the effect s of single crystalline ceria abrasives on polishing performance during silicon dioxide and silicon nitride CMP. The abrasive size was directly controlled by varying hydrothermal conditions without post-heat treatment and mechanical milling process. The result ant abrasives have a single crystalline phase regardless of the size of particles. The resu lts showed that the single crystalline ceria abrasives were not easily broken-down by mechanical force between abrasives and film surface during polishing. With increasing abrasive size, the removal rate of silicon dioxide and silicon nitride films increased. On the other hand, the surface uniformity deteriorated with increasing abrasive size, due to a broader particle size distribution of 91

PAGE 92

the abrasives in slurry and the morphology of the large abrasives. In addition, the removal selectivity showed a transition at the slurry C (with particle size of 163 nm). Considering these polishing behaviors of single crystalline ceria abra sives, it was found that there exists an optimum abrasive size fo r optimum removal rate and selectivity in silicon dioxide and silicon nitride CMP. 92

PAGE 93

Figure 4-1. XRD patterns of ceria particles synthesized from the mixture of water and different alcohols; (a) ethylene glycol, (b) methanol, (c) 1,4buthylene glycol, (d) ethanol 93

PAGE 94

Figure 4-2. FETEM photomicrographs of ce ria particles obtained by hydrothermal method using a new type of ceria precursor 94

PAGE 95

Figure 4-3. FESEM photographs of ceria particles prepared from the mixture of water and different alcohols; (a) ethanol, (b) 1,4-buthylene glycol (c) methanol, (d) ethylene glycol 95

PAGE 96

Figure 4-4. (a) Average particle size and (b) crystallites sizes of ceria particles synthesized with different dielectric constants of alcohols 96

PAGE 97

Figure 4-5. FESEM photographs of ce ria particles prepared with different concentrations of potassium hydroxide; (a) 0.5 M, (b) 1.0 M, and (C) 1.5 M 97

PAGE 98

Figure 4-6. FESEM photographs of ceri a particles prepared from different concentrations of nitric acid in hydrothermal conditions at 230 oC for 12 hr. ; (a) pH 4, (b) pH 2.5, (c ) pH 0.5 and (d) pH 0.5 98

PAGE 99

Figure 4-7. Crystallites size for ceria par ticles prepared from di fferent pH at (a) 150 oC, (b) 200 oC and 230 oC 99

PAGE 100

Figure 4-8. FESEM photographs of the ceria particles prepared with different hydrothermal conditions; (a) pH 3.0 at 220oC, (b) pH 3.0, (c) pH 1.5 and (d) pH 0.5 at 230oC, respectively 100

PAGE 101

Figure 4-9. XRD patterns and the (111) pea ks analyzed to confirm grain size of the ceria abrasives dispersed in ceria-based slurry (a) A, (b) B, (c) C and (d) D 101

PAGE 102

Figure 4-10. FETEM micrographs and of ceri a abrasive with average particle diameters of (a) 62 nm (slurry A) and (b) 232 nm (slurry D) 102

PAGE 103

Figure 4-11. Particle size distribution of ceria-based slurry used in this study 103

PAGE 104

Figure 4-12. FESEM photographs of ceria ab rasives (a) before and (b) after oxide CMP process 104

PAGE 105

Figure 4-13. Results of CMP field eval uation for removal rate and selectivity 105

PAGE 106

Figure 4-14. Results of CMP field evaluation for within-wafer nonuniformity (WIWNU) of silica film 106

PAGE 107

Table 4-1. Comparison of slurries used in this study Samples Hydrothermal conditions Primary particle Size (SEM, nm) Grain size (XRD, nm) Slurry mean size (UPA, nm) Surface area (m2/g) pH Temp.(oC) Slurry A 3.0 220 62 29 178 22.12 Slurry B 3,0 230 116 40 273 16.37 Slurry C 1.5 230 163 45 326 11.44 Slurry D 0.5 230 232 66 484 7.48 107

PAGE 108

Table 4-2. The result s of the CMP evaluation Samples Oxide removal rate (/min) Nitride removal rate (/min) Selectivity WIWNU of oxide film (%) Slurry A 1174 105 46 3 25.3 0.5 3.0 Slurry B 1794 125 47 4 38.0 0.6 5.5 Slurry C 2369 194 52 4 45.2 0.1 8.1 Slurry D 2674 201 62 5 43.4 0.5 11.89 108

PAGE 109

CHAPTER 5 POLISHING BEHAVIORS OF SPHERICAL CE RIA ABRASIVES ON SILICON DIOXIDE AND SILICON NITRIDE CMP Introduction In Chapter 4, the stru cture and size of ceria abrasiv es were controlled by liquid phase method. However, the shape of ceri a particles was transformed into square shape with increasing in the size of particles because of it cubic fluorite structure. The square shape may lead to serious defects on the wafer during CMP process. Thus, other methods will investigate on how to cont rol the shape and size of ceria abrasives. As mentioned in Chapter 4, the commercial method for synthesis of ceria abrasives involves thermal decomposition of cerium salts such as cerium carbonate and cerium hydroxide. This method leads to ve ry porous ceria particles with high surface area, inducing softness and high chemical reacti vity to oxide films. However, the size and the shape of ceria abrasives are very limited since particle growth is difficult to control during calcination process. To achieve the desired particle size and the uniform particle size distribution, mechanical milling and filtration is required. This leads to expensive installation cost. To overcome this problem, flux method is proposed to synthesize the ceria abrasives with a well-defined morphology in this chapter. This method consists of adding precursor in the required rati o to a molten salt mixture often close to an eutectic stoichiome try, which accelerates the kinetics of the formation of ceria particles.77 Reactions in molten salts pr ovide an original method for the preparation of solid. This method has been used in the past to synthesize a number of compounds (binary and ternary oxide, sulfi des) at relatively low temperature, which would otherwise require in excess of 1100 oC for conventional solid state synthesis. Molten salt mixtures either serve the role of a solvent with no direct participation in the 109

PAGE 110

reaction or may actually enter into reaction with the oxide additives. In cases where the molten salt serves merely as a solvent medium, its sole purpose is to accelerate the kinetics by enhancing diffusion, since coefficient s in the liquid state are lower than those in the solid state. Reactions are presumed to occur by the dissoluti on of constituents, reaction of the constituents in solution, and precipitation of the required compounds upon exceeding the solubility limit. The ceria particles obtained by this method have the several advantages over other methods such as narrow size distribution, desirable characteristics including very fine size, high chemical purity and good chemical homogeneity.78 However, this method has seldom been applied to the synthesis of ceria abrasives for CMP slurry. Furthermore, the polishing performance of s pherical ceria abrasives synthesized using flux method has not been reported so far for the silica and s ilicon nitride films. In this chapter, ceria particles with spher ical shape were synthesized using the flux method. Potassium hydroxide was employed as molten salt to accelerate the growth of ceria crystallites. The effe cts of the molten salts and reaction conditions on the formation of ceria particles were invest igated. Field emissi on scanning electron microscopy (FESEM), transmission electron microscopy (TEM), X-ray techniques, and surface area analysis (BET) were used to study the morphology and particle size distribution of the solid produc ts. To evaluate the size effects of ceria abrasives on CMP performance, the size of abras ives was controlled by adjusting reaction parameters. For CMP performance evaluation, the effect of spherical ceria abr asives on the removal rate, the oxide-nitride removal selectivity and the within-wafer non uni formity (WIWNU) was investigated. 110

PAGE 111

Materials and Methods Abrasives Preparation of as-prepared particles by hydrothermal method Cerium (III) nitrat e hexahydrate (Ce(NO3)3H2O) and potassium hydroxide (KOH) were used as the starting materials for synthesis of ceria abrasive particles. 0.5 M of cerium (III) nitrate hexahydrate and 1.0 M of potassium hydroxide were separately dissolved in mixed solvent of ethylene glycol (C2H6O2) and deionized (DI) water. The volume ratio of ethylene glycol to water was kept at 2:3. The reaction was carried out at a temperature of 50 oC with stirring rate of 100 rpm fo r 12 h. Air was bubbled into the precipitation reactor with passage thr ough a gas distributor as an oxidizer. The precipitated substance was separated via cent rifugation and then redi spersed in distilled water under continuous stirring. The weight ratio of distilled water to a precipitated substance was kept 5:1. The suspension solution was put into an autoclave. The hydrothermal reaction was carried out at 230 oC for 6 h. Preparation of ceria abr asive particles by solid st ate reaction (flux method) After the hydrothermal reaction, the pr ecipitated particles were washed with distilled water three times via centrifugat ion and then uniformly wetted in a potassium hydroxide/water mixed solution. The concen tration of potassium hydroxide was 0.1 ~ 2.0 wt% depending on the total we ight of the precipitated part icles. The wetted particles were sintered for 2 hr at 800 900oC temperature. The synthesized particles were washed with distilled water until the ion con ductivity of the washed solution was less than 0.5 S. The schematic diagram of experiment al procedure was shown in Fig. 5-1. 111

PAGE 112

CMP Evaluation Preparation of ceria-based slurries Different ceria-based slurries were formu lated by dispersing abrasives each with different primary particle size in DI wate r containing an anionic organic polymer (Poly acrylic acid, PAA; Mw 4000, LG Chem.) as di spersant. 2 wt% of PAA based on the total weight of the ceria abrasives was added. For each slurry, pH was adjusted to 6.5 ~ 6.7 by adding ammonium hydroxide (NH4OH). The solid loading of ceria abrasives was fixed to 2.0 wt%. In other to investigate the adsorption characteristics of ceria abrasives for the additive polymer, slurry A and D were dried at 80 oC for 24 hours. The weight loss of the abrasives dried from slurries was measured by thermo gravimetric analysis (TGA). TGA was performed in an air flow of 100 ml/min at a heating rate of 10 oC/min from 30 oC to 600 oC. CMP tools and consumables Silicon dioxide film of 2 m thick was grown on a 5-in. p-type silicon substrates with (001) orientation by plasma enhanced chemical vapor deposition (PECVD). The silicon nitride films were deposited by usi ng low-pressure chemical vapor deposition (LPCVD). Polishing tests were performed on a rotary type CMP machine (GNP POLI 400, G&P technology) for one minute with eac h of the ceria-based slurries. IC 1000/SUBA IV stacked pads (supplied by R odel Inc.) were utilized as CMP pads. The downforce was 4 psi and the rotation s peed between the pad and the wafer was 100 rpm. The slurry flow rate was 100 mL/min. 112

PAGE 113

Characterization Abrasives The crystal structure and grain size were identified through x-ray diffraction (XRD) using CuK radiation. The crystallite size was estimated by the Scherrer equation according to the formula D = 0.9 / ( cos ), where D is crystallite size, is the wavelength of x-rays, is the half-width of the diffraction peaks, and is the diffraction angle. The broadening of the reflection from the (111) plane was used to calculate the crystallite size. The morphology and sizes of the abrasive particles were also examined by field emission scanning electron micro scope (FESEM). The average primary particle size was calculated by measuring ca. 100 particles from FESEM micrographs. The specific surface area (SSA) of ceria abrasives was determined by Brunauer-EmmettTeller (BET) method using nitrogen adsorption/desorpt ion at 77 K. Ceria-based Slurry The abrasive size distribution of slurry was measured using light scattering method (UPA 150, Microtrac. Inc.). For the light sca ttering measurements, ceria particles were dispersed in deionized water without dispers ant, using an ultrasonic probe for 15 min. Polishing of wafers The film thickness on the wafers bef ore and after CMP was measured using spectroscopic reflectometry (Nanospec 6100, Nanometrics) to calculate the removal rate. In this experiment, the WIWNU was defined as the standard deviation of remaining thickness divided by the average of the rema ining thickness after the CMP process. The average polishing data for removal rate was carried by performing the same tests more than three times in order to support the valid ity of the results fr om the statistical viewpoint. 113

PAGE 114

Results and Discussion Ceria Abrasives Morphological properties Fig. 5-2 shows the FESEM images of abr asive particles prepared in different calcination conditions as described in Tabl e 5-1. The primary par ticle sizes determined using the FESEM micrographs were 84, 166, 295, and 417 nm for slurry A, B, C and D, respectively. As shown in Fig. 5-2, the ceri a particles consisted of crystalline grains with a well-defined morphology. The size of spher ical ceria particles increased with the increase in concentration of potassium hy droxide and the calci nation temperature. These images indicate that the primary particle size can be controlled by changing the calcination conditions, which influence t he crystal growth of ceria particles. Crystalline structure Fig. 5-3 shows the X-ray diffraction pattern s of the prepared particles. As shown in Fig. 5-3(a), the characteristic peaks corre sponding to (111), (200), (220), (311), and (222) planes are located at 2 = 28.51, 33.06, 47. 48, 56.20, and 59.05o, respectively. They show very close to the ones with cubic fluorite structure of ce ria crystal in JCPDS database. The sharp intensity peaks were observed for ceria abrasives with larger primary particle size as shown in Fig. 5-3(b) The crystallite size of ceria abrasives was calculated from the Scherrer equation using t he line-broadening of the (111) peak in the XRD pattern. The crystallite size gradually increased from 38 nm to 88 nm as the calcination temperature and the concentration of additive were increased. This result coincides with the trend of increasing primary particle size in the FESEM images shown in the Fig. 5-2. 114

PAGE 115

Effects of molten salt a nd as-prepared particle The synthesis method employed in this study is characterized by a low sintering temperature process, using gr ain growth accelerator. The eutectic mixed solvents as an accelerator is used to promote the kinet ics by enhancing diffusion, due to their low melting temperature. In this study, potassium hydroxide/ water mixed solvent was used as the accelerator to reduce sintering tem perature. Considering that the conventional sintering of ceria requires high temperatures leading to partial reduction above 1200 oC,79 potassium hydroxide/water mixed solvent might offer a liquid phase to promote interdiffusion on contact surface between t he smaller particle and the larger ones during the sintering process. Fig. 5-4 shows this dependence of the crys tallite size on the concentration of potassium hydroxide at a fixed temperature (850oC). It can be obs erved that an increase in the crystallite size was seen wit h addition of KOH at same calcination temperature. These results indicate that potassium hydroxide strongly affects the growth rate of ceria particle s at relatively low temperat ure. Therefore, the physical properties and the morphology of ceria abras ives could be controlled by manipulating the concentration of potassium hydroxide in this system. Additionally, cerium dioxide particles obt ained by hydrothermal method were used as the precursor in this study, instead of cerium salts such as hydroxide, nitride, and chloride. Fig. 5-5 shows the FESEM images of the particles prepared with different cerium precursors. Compared with other particles as shown in Fi g. 5-5(a), (b), and (c), it can be observed that the uniformity of ceria par ticles (Fig. 5-5(d)) prepared using oxide is superior. This result is attributed to t he fact that in case of cerium dioxide as a precursor, direct grain growth of crystallite is involved du ring heat treatment. However, 115

PAGE 116

in case of other precursors (hydroxide, ni tride and chloride), the formation of cerium dioxide crystallites involves two steps via thermal decomposition followed by grain growth. Thermal decomposition of such pr ecursors involve series of chemical reactions80 leading to retarded growth of ceria par ticles. Furthermore, it is difficult to achieve uniform ceria particle morphology because of the high surface energy and chemical reactivity of the volatile cerium compounds. Therefore, it seems that the direct formation to ceria induces the absence of hard aggregates and surface necking in the ceria crystallites. CMP Evaluation Characteristics of ceria abr asives before and after CMP To investigate the CMP performance using spherical ceria abrasives, we performed polishing test for the fo ur types of slurries with di fferent abrasive size. Fig. 56 shows the particle size distribution of these slurries without surfactant addition. The average particle sizes determined by light scattering method were 165, 278, 483, and 742 nm for slurry A, B, C and D, respectively. The dispersed particle size is much larger than the crystallite size estimated by X-rays and the primary particle size calculated by FESEM. This observation is due to the extensive overlapping of the ceria particles in water-based solution.75 The general tendency when using light scattering (LS) size distributions of particles is to oversize the coarse end of the distribution because LS uses the longer axis of an elongated particl e to determine the mean particle diameter. Another reason for obtaining a larger LS size is that the fundamental size distribution, as derived via LS, is based on volume; in other words, if a sample consists of an equal number of two sizes of particle, e.g., 50 nm and 100 nm, the volume of the 100 nm particles is 8 times larger than that of the 50 nm particles. Hence, as a volume 116

PAGE 117

distribution, the larger particl es represent most of the tota l volume, which yields a mean particle size by volume that is larger than that obtained by population. As shown in Fig. 5-6, the increase in the size of the prim ary particle led to broader particle size distribution in water-based solution. Fig.5-7 shows the FESEM images of ce ria abrasives before and after silicon dioxide polishing for slurry C. The ceria abrasives after polishing were washed with distilled water three times via centrifugation. According to Fig. 5-6, we can confirm that the ceria abrasives are brittle and break during polishing process. This indicates that the ceria abrasives used in this study are easily broken-down by applied pressure and shear force during polishing. Polishing Test Removal rate. Table 5-2 summarizes the quantitative results of the polishing test. It is clear that the removal ra te of the oxide films increased with increasing size of ceria abrasives. The removal rate of oxide film is mainly influenced by chemical contribution of ceria abrasives and mechanical factors, su ch as the CMP conditions, morphological characteristics of abrasives and particles size distribution. During the polishing of the oxide film, ceria abrasives exhibit a chemic al tooth property which accelerates the polishing removal rate of oxide film.12 As a consequence, the Si-O-Ce bonds can be rapidly removed by the mechanical forc e generated by pressed pad and abrasives, and this physicochemical reaction lead to the high re moval rate of oxide film. In this work, it was found that the removal rate of oxide f ilm was essentially dependent on the size of ceria abrasives. As with the removal rate of ox ide film, the removal rate of nitride film increased with increasing in the crystallite size of ceria abrasives. However, the removal rate of nitride film is affected by the ph ysical properties of ceri a-based slurry systems 117

PAGE 118

and the amount of surfactant adsorbed on the film surface.5 In order to improve the selectivity and uniformity, an anionic acrylic polymer is commonly used to passivate the surface of the nitride film during STI-CM P, which prevents ceria abrasives from contacting the film surface.76 In this study, the amount of the polymer was maintained constant for all slurry. Therefor e, it seems that the increase in the removal rate of nitride film is relate to the mechanica l factors rather that to the effect of pa ssivation layer of the polymer absorbed on the film surface. Thes e mechanical factors are influenced by several physical parameters of the CMP process, such as morphological aspects of the abrasives, crystallite size of the abrasives and the CMP conditions. Selectivity. It is well known that the remo val rate increases with increasing particle size due to mechanical indentation. However, the increase in removal rate shows lower slope between slurry C and D as com pared to that from slurry A to C as shown in Fig. 5-8. This led to a transit ion in removal select ivity between oxide and nitride film at slurry C. This result is attr ibuted to two related fa ctors: the contact-area reduction and the particle surface activity. In the case of oxide CMP process, the removal rate is mainly affected by contact area between the abrasives and the film surface. According to contact-area mechanism,16 the removal rate increases with decreasing abrasive size and increasing solid loading, due to the increase in contactarea between the abrasives and the film surface. At a fixed solid loading, the number of ceria abrasives in slurry decrease as the abrasive size increases. This implies the decrease in contact area between ceria abrasiv es in slurry D and oxide film during CMP processing. Additionally, large abrasives wit h spherical shape lead to the decrease in interfacial contact with film su rface, due to their rolling motion.4 Furthermore, relatively 118

PAGE 119

lower removal rate are attributed to t he inhomogeneous distribution of slurry on the wafer. For the slurry D, the broader size di stribution induces a loss in the frictional occasion of smaller abrasives by bigger abrasives and a local friction interaction of bigger abrasives on the wafer. The frictional force between the abrasives and the wafer is decreased by using slurry D because the fr ictional force is directly proportional to the contact area.81 These polishing behaviors would lead to a relatively lower increase in removal rate for the oxide wafer during CMP. In this study, such behavior can be clearly seen from the change in oxide removal rate fr om slurry A to C and slurry C to D. As shown in Fig. 5-8, the remo val rate rapidly increased from 1662 /min for 84 nm to 3990 /min for 295 nm, whereas it slightly incr eased to 4343 nm for 417 nm in spite of size effect of bigger abrasives during CMP process. In case of the ni tride film CMP, the polishing behavior can be explained by the relationship between additive polymer and abrasive size. The additive polymer can be more easily attached on the surface of the small abrasives than on the surface of la rge ones, due to high surface activity and specific surface area of the small abrasives. The slurry with small abrasives can induce a relatively high removal rate for nitride film since the passivation layer is insufficiently formed on the nitride film surface as polymer is largely adsorbed on the particle surface. As described in Table 5-1, the specific surface area dec reased with increasing abrasive size. Moreover, the compositional changes associated with the additive polymer adsorbed on the surface of ceria abrasives we re investigated with thermal analysis. Fig. 5-9 presents the TGA curves of the ceria abrasives dried fr om slurry A and D. The TGA curves of the ceria abrasives show two weig ht losses (curve (a) and (b)). The initial weight loss below t he temperature of 100 oC can be attributed to the evaporation of 119

PAGE 120

physically absorbed water in the air. The second weight loss observed at 200 ~ 260 oC is related to the decomposition of the additive polymer adsorbed on the abrasive surface. As shown in Fig. 5-9, final weig ht loss for two abrasives was different because the additive polymer content adsorbed on abrasive surface was different. This means that removal rate for nitride films can be rela tively increased in spite of decrease in the abrasive size. As a result, the removal rate of nitride film showed relatively low increase in rate with increase in the abrasive size. Therefore, these polis hing behaviors for oxide and nitride films result in the transition of removal selectivit y, suggestive of existence of an optimum size of spherical ceria abras ives for high removal selectivity. WIWNU. The results for surface uniformity of the oxide f ilms are shown in Fig. 510. As the size of ceria abrasives increas ed, the slurry had a higher WIWNU for the oxide film. This polishing behavior is attributed to a broader pa rticle size distribution of the slurry with large abrasives. The distribut ion data in Fig. 5-6 showed that the slurry has a rather-wide particle size distribution as abrasive particles increases. The broader particle size distribution of large abrasives can cause a different removal rate between the center and the edges of t he film surface due to their limited mobility on the wafer surface. Consequently, the surface uniformity deteriorated with increasing abrasive size during CMP process. It seems that surface uniformity of oxide film is related to the mechanical factors and the morphol ogical factors of ceria abras ives on the film surface. Conclusions Ceria Abrasives The ceria abrasives were prepared by the flux method, using potassium hydroxide (KOH) as the grain growth accelerator. The synthesized particles consisted of crystalline grains with a welldefined morphology. The size of spherical ceria particles 120

PAGE 121

increased with the increase in c oncentration of potassium hy droxide and the calcination temperature. Considering the sintering te mperature of ceria, potassium hydroxide strongly affected the growth rate of ceria particles at rela tively low temperature. The physical properties and the morphology of ceria abrasives could be controlled by manipulating the concentration of potassium hydroxide in this system. The FESEM analysis showed that ceria particles obtaine d from oxide phase is superior to other particles in terms of unifo rmity and shape. This method proposed in this study is very simple and can lead to well-cryst alline particles with desirable characteristics, including very fine size, narrow size distribution, high purity, and good chemical homogeneity. CMP Performance In this study, we investigated the effect s of spherical ceria abrasives on polishing performance during silicon dioxide and silicon nitride CMP. The size of the ceria abrasives was controlled by changing the calcination conditions without mechanical milling process. With increasi ng abrasive size, the removal rate of silicon dioxide and silicon nitride films increased. On the other hand, the su rface uniformity deteriorated after CMP process, due to a wide particle si ze distribution of the slurry with large abrasives. In addition, the removal selectivity showed a transition behavior at the slurry C (with particle size of 295 nm). This result indicates that there exists an optimum for removal selectivity as a function of abrasive size in the used slurry system. Therefore, we concluded that the control of abrasive size and particle si ze distribution of spherical ceria abrasives is an important paramete r for high removal selectivity and surface uniformity in the CMP process. 121

PAGE 122

Figure 5-1. Schematic diagram of experimental procedure 122

PAGE 123

Figure 5-2. FESEM photographs of the ceria abrasives prepared with different calcination conditions; (a) slu rry A, (b) B, (c) C and d(c) D 123

PAGE 124

Figure 5-3. (a) XRD patterns and (b) the (111) peaks analyzed to confirm crystallite size of the ceria abrasives dispersed in slurry (a) A, (b) B, (c) C and (d) D 124

PAGE 125

Figure 5-4. The variation of crystallite size as a function of the concentration of grain growth accelerator 125

PAGE 126

Figure 5-5. FETEM micrographs of the ceri a abrasives prepared with different cerium precursor; (a) cerium hydroxide, (b) ce rium nitride, (c) cerium chloride and (d) cerium dioxide 126

PAGE 127

Figure 5-6. Particle size distribution of ceria slurries as functi on of abrasive size 127

PAGE 128

Figure 5-7. FESEM photographs of ceria abr asives (a) before and (b) after polishing 128

PAGE 129

Figure 5-8. Results of CMP field evaluation for removal rate and selectivity 129

PAGE 130

Figure 5-9. TGA curves of t he ceria abrasives dried from (a) slurry A and (b) slurry D 130

PAGE 131

Figure 5-10. Within-wafer non uni formity (WIWNU) of oxide film 131

PAGE 132

Table 5-1. Comparison of slurries used in this study Samples Calcination conditions Primary particle Size (SEM, nm) Grain size (XRD, nm) Slurry mean size (UPA, nm) Surface area (m2/g) Molten salt (wt%) Temp. (oC) Slurry A 0.5 800 84 38 165 20.53 Slurry B 0.2 850 166 43 278 14.06 Slurry C 0.5 850 295 57 483 12.70 Slurry D 0.5 900 417 88 742 5.58 132

PAGE 133

Table 5-2. The result s of the CMP evaluation Samples Oxide removal rate (/min) Nitride removal rate (/min) Selectivity WIWNU of oxide film (%) Slurry A 1661 77 79 6 21 0.9 7.9 Slurry B 3390 103 117 8 29 1.0 11.0 Slurry C 3989 127 6 32 1.3 14.6 Slurry D 4343 155 13 28 1.4 16.7 133

PAGE 134

CHAPTER 6 PREPARATION AND CHARACTERISTICS OF THE CERIA COATED SILICA PARTICLES AND ITS CMP PERFORMANCE Introduction This chapter presents studies conducted on synthesis of the ceria coated silica particles and its CMP performance. According to rec ent reports, the homogeneous precipitation coating method using electrostatic attraction process has been extensively investigated to control the shape and uniformit y of particles. In particular, core-shell composites with spherical silica microspher es have been wisely investigated to provide novel properties that are not found in the single metal oxide.81 Moreover, silica particle as core material offers many advantages such as high specific surface, dispersion stability, narrow particle size distribution, hi gh mechanical strength and controllable size of the particles.83,84 Therefore, many researchers have investigated methods for producing the ceria coated silica particles. Howe ver, in previous works, the synthesized particles experienced severe aggregation du e to hydrogen bonding from water during precipitation process.72 Moreover, the resulting dis persion contains both the ceria coated silica particles and nano-sized ceria pa rticles due to the detachments of ceria coating on the surface of core particles. In addition, in some cases, synthesized particles need post-heat treatment to ac quire a well-crystalline ceria coating.85 However, post-heat treatment often lead to hard aggregat ion and incomplete coating surface for the particles. Furthermore, the experimental re sults showing control of the ceria coating thickness not been reported so far. To overcome these problems, cerium alcoholate complex are proposed as a new precursor of ceria coating on the surface of s ilica particles in this study. As mentioned in chapter 4, the solvent type used in reac tion affected physical properties of ceria 134

PAGE 135

particles.86 Moreover, nanocrystalline particles produced by alkoxide method shows much more reactivity than those produced by aqueous routes, since the organic solvents of alkoxide act as the dispersant for solid parti cles during precipitation and prevent from aggregating during drying process.87 Therefore, the first objective of this study is to estab lish a novel synthetic method of preparing monodispersed ceria coated silica particles with precise morphologies and chemical composition. The second objective is to deposit well-crystalline ceria coating on the surface of the silica par ticles without post-heat treatment. The third objective is to use the synthesized coated particles as abrasives for ceria-based slurry on CMP process. To achieve these purposes, ce rium ethanolate solvate complex as a new coating precursor was prepared by alkoxi de method, which helps to decrease the colloidal interaction and affects the crystallinit y of ceria coating. This study showed that the ceria coated particles prepared by the new precursor were uniformly synthesized without the formation of hard aggregate as compared to those obtained by conventional method. Additionally, the influences of va rious precipitation conditions on coating characteristics were investigated. Finally, the effect of the synthesized ceria coated silica particles in CMP slurry was investigated on silicon dioxide polishing process. Materials and Methods Abrasives Preparation of monodispersed silica particles Monodispersed spherical silica particles we re synthesized in a semi-batch reactor by modified Stber method.84 In a typical preparation, 2 ~ 5 ml of tetraethyl orthosilicate (TEOS) was fed into the reactor, in which a mixture of de-ionized water (1 ~ 3.5 ml), ethanol (10 ~ 80 ml) and ammonia (1 ~ 3.4 ml ) was premixed. The size of core silica 135

PAGE 136

particles can be controlled by adjusting diffe rent concentration of materials used. The reactants were agitated vigorously by stirring for 3 hours at a constant temperature of 25 oC. The silica particles were collect ed and washed repeatedly with ethanol. The particles were then dried at 50 oC for 24 hours. Preparation of ceria precursors To investigate the effects of solvent type on precursor characteristic, two kinds of ceria precursor were prepared via different methods using alcohol-based solvent and water-based solvent. The sufficiently high reactivity of ethanol (as alcohol-based solvent) and water can allow direct reacti on between cerium meta l and the solvent by heating the suspension under reflux. Synthesis of cerium ethanolate complex (designated as precursor A). The cerium alcoholate was prepar ed via alkoxide method. Cerium nitrate hexahydrate (Ce(NO3)2H2O) and potassium hydroxide (KOH) were used as the starting materials. Cerium nitrate hexahydrate (1 g) and potassium hydroxi de (1 g) were separately dissolved in 20 ml ethanol under vigorous stirring for 24 hours at 50 oC. Hydrogen peroxide (30 % H2O2, 0.05 ml) was added to mixing solution as the oxidizing agent. Eqn. 6-1 describes the chemical reaction: OHKNO CHOCHCe KOH OHCHCHOHNOCereflux 2 3 223 23 22382) ( 2) (26)( (6-1) The precipitated ceria precursor was separ ated via centrifugation. The supernatant solution obtained from the centrifugation wa s discarded and the 1 g of the precipitate was redispersed in 20 ml distilled water under continuous stirring. The pH of the suspension solution was adjusted to 0.1 by adding concentrated nitric acid (HNO3). Nitric acid was used to obtain peptized ceria sol. Finally, the solution was heated to 40 136

PAGE 137

oC and kept at this temperature for 2 hour s. The transparent li ght yellow solution obtained was cooled down to room temperature. Synthesis of cerium hydroxide comp lex (designated as precursor B). The cerium hydroxide was prepared via precipitation method. Cerium nitrate hexahydrate (Ce(NO3)2H2O) and potassium hydroxide (KOH) were used as the starting materials. Cerium nitrate hexahydrate (1 g) and potassium hydroxi de (1 g) were separately dissolved in 20 ml distilled water under vigorous stirring for 24 hours at 50 oC. Hydrogen peroxide (30 % H2O2, 0.05 ml) was added to mixing solution as the oxidizing agent. Eqn. 6-2 describes the chemical reaction: OHKNO OHCe KOHOHOHNOCereflux 2 3 2 2 22382)( 2)(26)( (6-2) The precipitated ceria precursor was separ ated via centrifugation. The supernatant solution obtained from the centrifugation wa s discarded and the 1 g of the precipitate was redispersed in 20 ml distilled water under continuous stirring. The pH of the suspension solution was adjusted to 0.1 by adding concentrated nitric acid (HNO3). Nitric acid was used to obtain peptized ceria sol. Finally, the solution was heated to 40 oC and kept at this temperature for 2 hour s. The transparent li ght yellow solution obtained was cooled down to room temperature. Preparation of ceria coated silica particles For the synthesis of silica particles coated with ceria, 0.1 g of the silica particles was dispersed in 30 ml distilled water. 1.0 ~ 8.0 ml of ceria precursor was added to the solution. After stirring for 30 min, the pH of the solution was adjusted to 3.0 ~ 10.0 by adding 0.5 mol/L ammonium hydroxide (NH4OH) solution. The final mixed solution was stirred for 4 hours at a temperature of 60 oC. The coated particles were washed through 137

PAGE 138

several decantation and redis persion cycles by centrifugatio n. The resulting particles were then dried at 50 oC for 24 hours. In order to co mpare with ceria particles, the precipitated ceria precursor was washed wi th distilled water three times by repeated centrifugation. After washing, the precipitate was dried at 90 oC. Preparation of Ceria-bases Slurry Different slurries were formulated by dispersing the synthesized ceria coated silica particles in deionized water containing an an ionic organic polymer (Poly acrylic acid, PAA; Aldrich Mw: 50,000) as dispersan t. The PAA was 0.5 wt% based on the total weight of the used abrasives. The pH was adjusted to 3.0 ~ 9.0 by adding ammonium hydroxide (NH4OH). The solid loading of the used abrasives was fixed to 2.0 wt%. CMP Evaluation Silica (SiO2) wafer for all experiments hav e a nominal thickness of 2 m. Plasma enhanced chemical vapor deposition (PECVD) is used to the deposit the oxide layers on 1 mm thick, p-type silicon subs trates with (001) or ientation. A small sample (0.8 in 0.8 in) is cut from the silica wafer and then is mounted on the stainless steel cylinder with a plastic adhesive. The silica wafe rs will be obtained from Silicon Quest International. Polishing experiments is performed on a Struers Rotopol 31 and TegraPol 35 tabletop polishers. The TegraPol 35 polisher allows control of platen velocity and pressure down to 3 ~ 7 psi. The ceria po lishing slurry is delivered by a Struers Multidoser unit that is connected to t he polisher. The head and platen speed is 60 rpm and the center of the sample to be polished is placed at a distance of 8.9 cm from the center of the table. An IC 1000/Suba IV stacked pads (Rodel Inc.) is used for all polishing tests. During polishing, the slurry is continuously stirr ed by a magnetic bar, 138

PAGE 139

and pumped to the pad-wafer inte rface at a flow rate of 100 mL/min by the Multidoser unit. The pad is conditioned before each polishing run with a grid-abrade diamond pad conditioner manufactured by TBW. The diamond conditioner minimizes pad glazing and ensures reproducible pad condi tions for each polish run. Characterization Ceria coated silica abrasives. The morphology and the size of the silica particles coated with ceria were examined by fiel d emission scanning electric microscope (FESEM) at an accelerating voltage of 15 kV and high resolution transmission electron microscope (HRTEM) at an accelerating voltage of 120 kV. The average particle size of the silica particles coated with ceria was calc ulated by measuring ca. 100 particles from FESEM micrographs. The crystal structure of the particles was determined by x-ray diffraction (XRD) using CuK (1.54 ) radiation. The X-ray photoelectron spectroscopy (XPS) measurements were performed with a non-monochromatic MgK source at a base pressure of 5 10-1 mbar. For consistency, all bi nding energies are reported with reference to the C 1s transition at 284.6 eV. The electrophoresis measurements were performed to obtain the electroph oretic mobility and the isoel ectric point (IEP) of the prepared particles. The electrophoretic mobili ty of the particles was determined as a function of the pH. Slurry stability. The abrasive size distribution of slurry was measured using light scattering method (UPA 150, Microtrac Inc.) as a function of pH of suspension. Polished wafer. The film thickness on the wa fers before and after CMP was measured using Filmetrics F20 (Filmetrics Inc.) to calculate the removal rate. In this experiment, the WIWNU was defined as the standard deviation of remaining thickness 139

PAGE 140

divided by the average of t he remaining thickness after the CMP process. The average polishing data for removal rate was carri ed by performing the same tests more than three times in order to support the validity of the results fr om the statistical viewpoint. Results and Discussion Ceria Coated Silica Particles Morphology Fig. 6-1 presents the FESEM micrographs of the silica core particles. It can be seen from the Fig. 6-1 that the uncoated silica particles have a monodispersed spherical shape and a very smooth surface. Fig. 6-2(a) shows the F ESEM micrograph of the ceri a coated silica particles prepared using the 1.0 ml of prec ursor A at pH 6.8. As shown in Fig. 6-2(a), the coated particles have rough surface in contrast with the smooth surface of silica core particles. To better visualize the nature of ceria coatin g, a HRTEM micrograph of the ceria coated silica particles is presented in Fi g. 6-2 (b). It can be found from the Fig. 6-2( b) that ceria coating with a thickness about 3 ~ 8 nm was depos ited on the surface of silica particles. Moreover, as shown in Fig. 6-2 (c) and (d ), the difference in morphology between two specimens prepared with precursor A and B are very interesting. The coated particles obtained from the precursor B (Fig. 6-2(c)) were severely aggregated. It appears that this result is mainly due to the bridging of adjacent particles with water by hydrogen bonding and subsequent high capillary forces during the drying process.86 On the other hand, the coated particles synthesized from ceria precursor were monodispersed without hard aggregates (see Fig. 6-2( d)). This is attributed to the fact that the particles prepared by the alkoxide method generally show much higher reactivity than those via aqueous routes because organic solvent used in alkoxide method act as a dispersant 140

PAGE 141

for particles during precipitation and preven t them from aggregating during drying. This result is consistent with the study reported by Ikegami et al.87, which shows that the Al(OH)3 and Y(OH)3 wet precipitates show better dispersi ty in alcohols. In our work too, alcohol helps to decrease the colloidal attraction forces bet ween particle surfaces during precipitation process and hence increase the dispersity of the ceria coated silica particles. Crystalline phase XRD patterns of the bare silica particles and ceria coated silica particles synthesized from the two precursors are shown in Fig. 6-3. The curve (a) represents the XRD spectra of the bare silica particles while curve (b) and (c) show XRD pattern of the coated silica particles obtained from precursor A and B, respectively. As shown in Fig. 6-3, the major reflections associated with fl uorite structure of ceria coating can be observed from curve (b) and (c) in Fig. 6-3. The characteristic peaks corresponding to (111), (220), and (311) planes are located at 2 = 28.53 47.47 and 56.22 respectively. They show very close to the ones with cubic fluorite structure of ce ria crystal in JCPDS database. However, the ceria coating synthesized from precursor A show higher peak intensity as compared to that prepared by precursor B. It is interesting to note that the ceria coating synthesized by using ceria prec ursor shows better crystallinity in spite of the processing temperature being same as used in the conventional precipitation method. Additionally, the XRD analysis of t he coated particles shows very broad peaks of ceria fluorite structure due to a thickness of ceria layer being below 10 nm and ceria coating was formed on the amorphous silica core. 141

PAGE 142

XPS spectra of the ceria coating on silica particles To further confirm the existence of ceria coating on the surface of silica particles, XPS analysis was carried out. The silica particles coated with ceria were prepared using 2.0 ml of ceria precursor at pH 6.8. Fig. 6-4 shows the XPS survey spectrum of the ceria coated silica particles obtained. It can be found that the sample contains Ce, Si, C and O elements. The photoelectron peaks for Si 2P, Si 2S, Ce 3d5/2, and Ce 3d3/2 are detected at 104.4 eV, 164.2 eV, and 860 eV ~ 920 eV with reference to the C 1s transition at 284.6 eV, respectively. XPS analysis was also used to determine the oxidation state of cerium ions coated on the surface of silica particles. The O 1s core spectra are very informative in terms of structural details of the ceria coating. O 1s spectra of core silica particles and the coated particles are shown in Fig. 6-5. The binding energy of O 1s peaks for pure silica is 533.0 eV. As seen from the Fig. 6-5, it is clear that O 1s peak can be deconvoluted into two peaks associated with silica and ceria. For the silica particles coated with ceria, there are two mixed O 1s peaks at 532.4 eV and 534.8 eV, respectively. According to the data published by National Institute of Standar ds and Technology of USA, the peak at 532.4 eV is attributed to oxygen in silica par ticles. Whereas, the other peak at 534.8 eV is associated with the ceria coating on silica surface. After ceria coating, the binding energy of oxygen for the silica was slightly dow nward shifted. The chemical shift of O 1s peak can be explained by the insertion of Ce4+ cations into the tetr ahedral sites of the silica network to form Ce-O-Si bonds and the greater electronegativity of Si4+ to that of Ce4+. This indicates that the ceria coating layer and core silica particle is connected through the Ce-O-Si chemical bonding at the interface. 142

PAGE 143

To confirm the purity of ceria coating deposited on silica particles, the XPS Ce 3d multiplex is shown in Fig. 5. The Ce 3d peak consists of two main features in the range of 870 ~ 930 eV, corresponding to the Ce 3d3/2 and 3d5/2 components due to the spinorbit coupling. In previous works,88,89 both Ce(III) and Ce(IV) show the 3d5/2 and 3d3/2 multiplexes. However, Ce(III) shows only two peaks (P1 and P2) for each component (3d5/2 and 3d3/2) whereas Ce(IV) shows three peaks (P 1, P2 and P3) associated with the initial state of tetrav alent cerium. In our work the pr esence of the initial state of tetravalent cerium is further substantiated by the evidence of an third peak at a binding energy of 918 eV. The presence of three structures (P1, P2 and P3) confirms the coating of ceria on the su rface of silica particles. Electrokinetic behavior Fig. 6-7 shows the FESEM micrographs of the ceria coated silica particles prepared using the 1.0 ml of prec ursor A at different pH. At pH 3.2, the ceria layer was not observed on the surface of silica particles (F ig. 6-7(a)). This indicates that the ceria coating was not formed on the surface of silica particles. At pH 6.8, the coated particles were observed and the ceria layer was uni formly deposited on the surface of silica particles (Fig. 6-7(b)). It is found that this pH condition is favorable for the chemical bond formation between the ceria precursor and s ilica particles. At pH 9.7, the prepared particles were covered with a massive ceria laye rs (Fig. 6-7(c)). This is attributed to the fact that at pH 9.7, ceria precursor becomes unstable and ce ria precipitates. Additionally, the pH is near the ceria precip itate IEP leading to formation of ceria particle aggregates. In Fig. 6-8, curves show the variation of zeta potential with pH value for the silica particles and the pr ecipitated precursor A, respectively. As show n by the curve (c), the precipitated precursor A has the IEP at pH 8 ~ 10. In our work coating process for the 143

PAGE 144

ceria coated silica particles include three step s; the formation of ge l by ceria sol, the formation of gel-like layer on the core parti cles and crystallization of ceria in this layer.90,91 This coating behavior is qualitatively in fluenced by the pH value of reaction medium due to variation in the electrostati c forces between ceria precursor and silica core particles as shown in Fig. 6-9. At pH 3.2, the ceria layer was not deposited on the surface of silica particles bec ause of the weak electrostati c attraction between ceria sol and core particles. Besides, participating anions (OH-) cause the precipitation of ceria particles in liquid phase reaction,72,92 it is probable that the formation of the gel was not induced in spite of adding ammonium hydroxi de due to the acidic condition at pH 3.2. On the other hand, at pH 6. 8, silica particles have a negative surface charge and the precipitated ceria precursor has a positive surface charge with a significant difference in magnitude of charge between the two materials as shown in Fig. 6-8. This indicates that the precipitated ceria precurso r would be expected to experi ence a strong electrostatic attraction towards the negatively changed surf ace of silica particles leading to deposition of ceria layer on the surface of silic a particles. Further, at pH 9.7, the pH of ceria precursor solution is near the IEP of precipitated precursor A. Under these conditions, ceria particles aggr egate to form thick layer due to unstable conditions and weak interaction between the ceria precipitat e and the silica particles. Moreover, in Fig. 6-8, curve (b) shows the variations of zetapotential with pH values for the ceria coated silica particles. The coated particles were obtained by adding 4 ml of ceria precursor at pH 6.7. As shown in Fig. 68, it can be found that the IEP of the coated particles shifted from silica particles toward the similar val ue of ceria particles and the electrophoretic mobility of the coated particles showed a si milar trend with that of ceria particles. 144

PAGE 145

Control of thickness of the cer ia coating on silica particles Fig. 6-10 shows the FESEM micrographs of the ceria coated silica particles prepared with the different concentration of precursor A at pH 6.6 ~ 6.9. The coated particle sizes determined in the FESEM examination gradually increased with the increase in precursor concentration. The vari ations of the average particle size and the ceria coating thickness with t he amount of ceria precursor ar e shown in Fig. 6-11. The average particle size was 289, 302, 338, and 402 nm for the 0, 1, 2, and 4 ml of the precursor added, respectively. After adding 8 ml precursor, the average particle size could not be calculated, since massive ceria layers were deposited on the surface of silica particles. The average ceria coating thickness was 6.5 nm after adding 1 ml ceria precursor. With increasing t he concentration of ceria prec ursor, the coating thickness gradually increased about 14 nm per additional 1 ml precur sor. Therefore, the thickness of ceria coating could be cont rolled from 6.5 nm to maxi mum 56.5 nm by changing the concentration of ceria precursor in this system. Size control of the ceria coating on silica particles Figure 6-12 shows the FESEM micrographs of the silica particles synthesized by different experiment conditions. It is found that the size of colloidal silica particles could be controlled by adjusting the r eaction parameter such as tem perature, solvent type, pH, and the concentration of TEOS. In this expe riment, the average size s of silica particles determined by FESEM were 105 nm, 214 nm, 332 nm, and 442 nm, respectively. Figure 6-13 shows the FESEM micrographs of the ceria coated silica particles, which have each different size according to the size of colloidal silica particles used. The average sizes of the synthesized particles determined by FESEM were 146 nm, 256 nm, 334 nm, 145

PAGE 146

and 384 nm, respectively. Therefor e, the size of the ceria c oated silica particles can be controlled by the size of core silica particles. CMP Evaluation To investigate the effects of the ceria coated silica particles on CMP performance, I carried out CMP test with the slurry incl uding the silica particles coated with the coating thickness of 7 ~ 12 nm under t he different CMP conditions. Table 6-1 summarizes the quantitative results of the polis hing test as a function of suspension pH and down pressure. Effect of pH The results for removal rate of the oxide films as a f unction of suspension pH are shown in Fig 6-14. The downforce was 7 psi. As the pH of ceria-based slurry increased, the slurry had a lower removal rate for the oxide film. This polishing behavior is attributed to electrophoretic mobility and absorption behavior between abrasives and silica film as a function of suspension pH. As s hown in Fig. 6-8, the pH at the respective IEP can be identified as the point of zero charge (pHpzc). The pHpzc of ceria abrasives occurs at pH 7 ~ 8. Also, the pHpzc of silica film is at about 1.5 ~ 2.8.93 This value is close to core silica particles reported in this experiment. At pH values below the pHpzc of ceria abrasives (~ pH 7), the surface of ceria becomes posit ively charged and the surface of silica film becom es negatively charged, leading to absorption between two materials. However, with the increasing pH of the solution, the ceria surface becomes more negatively charged and the silica surf ace also has a negative surface charge, leading to repulsion between two materials. It seems that these in teraction behaviors affected the removal rate of silica film. T herefore, the removal rate of silica film decreased with the increase in pH of ceria-based slurry. 146

PAGE 147

Effect of down pressure The results for removal rate of the oxid e films as function of CMP pressure are shown in Fig 6-15. For each slurry, pH wa s adjusted to 2.9 ~ 3. 1 by adding ammonium hydroxide. As shown in Fig. 6-15, the re moval rate increased with increasing in the pressure during CMP process. This mechan ical behavior is explained by Prestons equation. The abrasion/wear-bas ed Prestons equation MRR = KpP0V, where MRR is the material removal rate, P0 the down pressure, V the relati ve velocity of water, and Kp a constant representing the effect of other remaining parameters, has been widely used in CMP process control and consumable dev elopment for integrated circuit (IC) fabrication and manufacturing. It reflects the influence of mechanicals on the material removal rate. From this equation, it is found t hat the removal rate is proportional to the down pressure (P0). Therefore, material removal rate of oxide increased with the increase in down pressure. Wafer roughness (WIWNU) Fig. 6-16 shows the results for surface unifo rmity of the oxide films as function of pH. The downforce was 5 psi. As show n in Fig. 6-16, WIWNU decreased with increasing the suspension pH. This polishing behavior is attributed to a broader particle size distribution and the agglomerated particle behavior of ceria slurri es as function of suspension pH. The slurry was formulated by dispersing ceria abrasives in DI water containing an anionic organic polymer. Generally, the acid ic suspension had greater particle size and broader size distribution of abrasive ceria than thos e of the neutral or alkaline suspensions.29 The broader particle size distribution of large abrasives can cause a different removal rate between the c enter and the edges of t he film surface due 147

PAGE 148

to their limited mobility on the wafer surfac e. Consequently, the surface uniformity deteriorated with the aggregated abrasives duri ng CMP process. Conclusions Ceria Coated Silica Particles Monodispersed ceria coated silica particles were prepared by surface-induced precipitation method using new type of ceria coating precurso r. In order to investigate the effects of solvent type on the physical and chemical proper ties of ceria coating, the cerium ethanolate complex was used as t he precursor A and wa s compared with the precursor B obtained from water-based solvent. The pH of the ceria treatment solution had significantly influence on the formation of the coating on the particles due to the electrostatic attraction between the silica particles and the sol type ceria precursor. The ceria coating was uniformly formed on the surfac e of silica particles at a pH of about 7.0. The electrophoretic mobility for coated particl es proved that the ceria layers were deposited on the surface of silica particles FESEM and TEM micrographs showed that the resultant coated particles prepared from precursor A exhibited spherical shape without the formation of hard aggregates as co mpared to that prepared from precursor B. The XRD analysis of the coated particles revealed that the crystalline ceria coating was formed without post-heat treatment. XRD re sults also revealed a difference in crystallinity of ceria coati ng obtained from precursor A and B. The XPS investigation for the O 1s and Ce 3d photoelectron lines showed the pure ceria (Ce(IV) oxide) coating on the silica particles which was chemically bonded with the silica particles as evidenced by the additional peak at a binding energy of 918 eV. The coating thickness of particles was controlled by adjusting the amount of ce ria coating precursor. As the precursor concentration was increased, the coating thic kness gradually increased from 6.5 to 56.5 148

PAGE 149

nm (maximum thickness). Additionally, the size of the ceria coated silica particles could be easily controlled by the size of core silic a particles. This means that abrasive size can be controlled according to CMP proc ess conditions without mechanical milling process and filteration system. CMP Evaluation In this study, the resultant particles were used as CMP abrasives for ceria-based slurry to investigate the effects of t he ceria coated silica particles on polishing performance during oxide CMP. The primary particle size was controlled about 146 nm without mechanical milling process by depositing the ceria coating on the surface of colloidal silica particles with size 135 nm. The removal rate of silicon dioxide film strongly depended upon suspension pH. With in creasing suspension pH, the removal rate of silica film decreased. On the other hand, surface uniformity deteriorated as the suspension pH decreases. These polishing behaviors were related to electrophoretic mobility of abrasives for silica film to be pol ished, since surface charge of the abrasives can be changed by suspension pH. Thus, it means that absorption/repulsion behavior between abrasives and materials to be polished plays an important role in polishing performance during CMP process. Additionally, the removal rate of silica film increased with increasing down pressure in terms of mechanical aspect for CMP condition. 149

PAGE 150

Figure 6-1. FESEM images of silica core par ticles obtained by modified Stber method 150

PAGE 151

Figure 6-2. (a) FESEM and (b ) HRTEM micrographs for the su rface condition of coated particle and FESEM micrographs for ceri a coated silica particles prepared by (c) precursor B and (d) precursor A, respectively 151

PAGE 152

Figure 6-3. XRD patterns of the synthesized particles; (a) bare silica particles, (b) ceria coated silica particles prepared by precursor B, and (c) precursor A 152

PAGE 153

Figure 6-4. XPS survey spectrum of ceria coated silica particles 153

PAGE 154

Figure 6-5. XPS spectra of O 1s peaks of ceria coated silica particles 154

PAGE 155

Figure 6-6. XPS Ce 3d multiplex of ceria coated silica particles 155

PAGE 156

Figure 6-7. FESEM photographs for ceria coat ed silica particles prepared at different pH (a) 3.2, (b) 6.8 and (c) 9.7 156

PAGE 157

Figure 6-8. Electrophoretic mobility for (a) s ilica particles (b) ceria coated silica particles and (c) ceria particles 157

PAGE 158

Figure 6-9. Scheme of the formation mechan ism of ceria coated silica particles at different pH 158

PAGE 159

Figure 6-10. FESEM micrographs of ceria c oated silica particles prepared by different concentration of ceria precursors (a) 0.0 ml, (b) 1.0 ml, (c) 2.0 ml, (d) 4.0 ml and (e) 8.0 ml 159

PAGE 160

Figure 6-11. The variations of (a) coating thickness and (b) average particle size for samples obtained by changing the c oncentration of ceria precursors 160

PAGE 161

Figure 6-12. FESEM micrographs of the silica particles with different size; (a) 105 nm, (b) 214 nm, (c) 332 nm, and (d) 442 nm 161

PAGE 162

Figure 6-13. FESEM microgr aphs of the ceria coated si lica particles obtained from different core silica particles with diffe rent size; (a) 146 nm, (b) 256 nm, (c) 334 nm, and (d) 384 nm 162

PAGE 163

369 0 30 60 90 120 150 180 210 240 270 Removal rate (min)pH Figure 6-14. Results of CMP field evaluation for removal rate as function of pH 163

PAGE 164

2345678 0 50 100 150 200 250 Removal rate (A/min)Pressure (psi) Figure 6-15. Results of CMP field evaluation for removal rate as function of CMP pressure 164

PAGE 165

369 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 WIWNU (%)pH Figure 6-16. The result for within-wafe r non uniformity (WIWNU) of oxide film 165

PAGE 166

Table 6-1. The results of the removal rate for the ceria coated silica particles Pressure (psi) pH9 (/min) pH6(/min) pH3(/min) 3 16.6 24.2 42.4 5 46.4 63.0 95.2 7 67.6 147.2 245.8 166

PAGE 167

CHAPTER 7 SYNTHESIS OF SPHERICAL CERIA PA RTICLES BY THERMA L DECOMPOSITION METHOD AND ITS CMP PERFORMANCE Introduction The shape-controlled synthesis of nanomaterials has been considered as important research subject in numerous applic ations, such as in ceramics, catalysis, additive in paint, recording materials, and m any others. The change in morphology of nanocrystalline materials will alter the properties which we re formerly thought to be constant for given materials. As ment ioned in chapter 4 and 5, commercial ceria abrasives is commonly prepared by thermal decomposition method. Although the ceria abrasives obtained from thermal decompos ition have a higher porosity leading to softness and high chemical reactivity to oxide films, the size and t he shape of the ceria abrasives are very limited since particle growth is difficult to control during calcination process. Therefore, a new method which can control the morphological properties of ceria particle should be proposed. In chapter 4, solvent effects on the shape, size, and crystalline phase for ceria particles under hydrothermal precipitation condition had be en investigated. Also, many research results have been reported that the morphology of solid particles is influenced by the dielectric properties of alcohols as solvent in view of the thermodynamics of reaction system, nucleation kinetics, and particle interaction potentials.64,65 Therefore, the advantage of the use of organi c solvent lies in that it can adjust the growth habit of the ceria particles, leading to the formation of final products with desirable morphology and change the micro-environm ent of the reaction. Considering the framework of precursors or starting ce rium salt tends to remain immediately after the thermal decomposition,94 cerium carbonate as precursor is 167

PAGE 168

expected to affect the shape and size of ceria particles afte r heat treatment. If monodispersed spherical cerium carbonate particles are prepared, ceria particles obtained from the cerium carbonates should have the following properties: (1) spherical shape, (2) narrow particle size distribution, (3) mechanical softness with lager surface area, and (4) chemical properties From these concepts, in th is chapter, spherical ceria particles were prepared by thermal decom position method using spherical cerium carbonate as as-prepared particles. I hav e developed a two-st ep procedure to synthesize ceria crystallites. A cerium car bonate precursor was firs t prepared via simple precipitation method using alcohol/water mixed solvent, and ceria particles was obtained by subsequent thermal decomposition of the precurso r. The resultant particles were used as abrasives of ceria-based slurry. In order to investigat e the size effects of ceria abrasives on CMP performance, the size of cerium carbonate particles were modulated by changing the volume ratio of al cohol to water of mixed solvent and the dielectric constant of solv ent. For CMP performance evaluation, the effect of spherical ceria abrasives on the removal rate, the oxide-nitride removal selectivity and the withinwafer non uniformity (WIWNU) was investigated. Materials and Methods Preparation of Spheri cal Ceria Abrasives Cerium (III) nitrat e hexahydrate (Ce(NO3)3H2O) and ammonium carbonate ((NH4)2CO3) were used as the starting materials for ceria particles. Both regents were of analytical grade purity and were used without fu rther purification. Cerium nitrate and ammonium carbonate were separately dissolved in distilled water. The concentration of both chemicals was adjusted to 5 M. Th is aqueous solution was then mixed with alcohols to adjust the volume ratio of alc ohol to water to 0, 1, 3, and 5. The final 168

PAGE 169

concentration of both chemicals was 1 mol/L, respectively. Alcohols, including methanol (CH3OH), ethanol (C2H5OH), 2-propanol (C3H8O), and 1, 4-butandiol (C4H10O2) were used to investigate the influence of the pr operties of solvent. The cerium carbonate particles were obtained by dripping the ceri um nitride solution into ammonium carbonate solution. The carbonate precipitation was carried out at a temperature of 75 oC for 24 hour with magnetic stirring. After the precip itation reaction, the precipitates were washed with distilled water three times by repeated centrifugatio n and then dried at room temperature with flowing nitrogen. The dried particles were calcined in for 90 min at 500 800oC temperature to obtain ceria particles. CMP Evaluation Preparation of ceria-based slurries Different ceria-based slurries were formu lated by dispersing abrasives each with different primary particle size in DI wate r containing an anionic organic polymer (Poly acrylic acid, PAA; Mw 4000, LG Chem.) as di spersant. 2 wt% of PAA based on the total weight of the ceria abrasives was added. For each slurry, pH was adjusted to 3 ~ 9 by adding ammonium hydroxide (NH4OH). The solid loading of ce ria abrasives was fixed to 2.0 wt%. CMP tools and consumables Silicon dioxide film of 2 m thick was grown on a 5-in. p-type silicon substrates with (001) orientation by plasma enhanced chemical vapor deposition (PECVD). The silicon nitride films were deposited by usi ng low-pressure chemical vapor deposition (LPCVD). Polishing tests were performed on a rotary type CMP machine (GNP POLI 400, G&P technology) for one minute with eac h of the ceria-based slurries. IC 1000/SUBA IV stacked pads (supplied by R odel Inc.) were utilized as CMP pads. The 169

PAGE 170

downforce was 4 psi and the rotation s peed between the pad and the wafer was 100 rpm. The slurry flow rate was 100 mL/min. Characterization Abrasives The crystal structure and grain size was identified by x-ray diffraction (XRD) using CuK radiation. The grain size was estimated by the Scherrer equation according to the formula D = 0.9 / ( cos ), where D is the grain size, is the wavelength of x-rays, is the half-width of the diffraction peaks, and is the diffraction angle. The broadening of the reflection from the (111) plane was used to calculate the grain size. The morphology and size of the precipitate parti cles were examined by high resolution transmission electron microscope (HRTEM) and field emission scanning electron microscope (FESEM). The aver age primary particle size was calculated by measuring ca. 100 particles from FESEM micrographs. The specific surf ace area (SSA) of the ceria abrasives was determined by Brunauer-Emmett-Teller (BET) method using nitrogen adsorption/desorption at 77 K. Ceria-based slurry The abrasive size distribution of slurry was measured using light scattering method (UPA 150, Microtrac Inc.). CMP performance The film thickness on the wafers bef ore and after CMP was measured using spectroscopic reflectometry (Nanospec 6100, Nanometrics) to calculate the removal rate. In this experiment, the WIWNU was defined as the standard deviation of remaining thickness divided by the average of the rema ining thickness after the CMP process. The average polishing data for removal rate was carried by performing the same tests more 170

PAGE 171

than three times in order to support the valid ity of the results fr om the statistical viewpoint. Results and Discussion Properties of Spherical Cerium Carbonate Precursor Influence of solvent type on particle morphology Figure 7-1 shows the XRD pa tterns of cerium carbonate compounds synthesized at 75 oC with different kinds of solvents. All except the particle obtained by using isopropanol as solvent exhi bited a pure orthorhombic CeOHCO3 structure with variable crystallinity in JCPDS database (a=5 .015, b=8.565, c=7. 337 symmetry group Pmcn ). On the other hand, t he particles obtained from isopropanol consisted of a mixture of products including CeO2 and CeOHCO3, the former being responsible for the stronger peaks. This solvent type has a signi ficant effect on the phase formation with different affinities for water. It is wellknown that, physicochemical solvent properties, such as polarity, viscosity, dielectric pr operties, and softness, will strong influence the solubility and transport behavior of the precursors. Under various experim ental conditions, the microstr uctures of cerium carbonate compounds can be changed to different shape such as spherical, oval and spindleshape. Figure 7-2 shows FESEM micrographs of cerium carbonate obtained from water-based solvent. As shown in Fig. 7-2, the cerium carbonate compounds are composed of oval-like particles with the si ze of 300 ~ 500 nm. On the other hand, the cerium carbonate compounds obtained by using mixed solvent are composed of spherical agglomerates with uniform size di stribution as shown in Fig. 7-3. This difference in morphology of the resultant particles indicates t hat the colloidal stability of the precipitation particles in the mixed solvent is different from that pure water solvent. 171

PAGE 172

This result can be understood by considerin g the parameters determining the colloidal stability.95-97 Generally, the maximum repulsive forc e can be estimated from the eq. 7-1 for electrostatically stabilized particles. 2 02 r rF (7-1) where 0 is the permittivity in free space, r is the dielectric constant of the continuous phase, is the Debye-Huckel parameter, is the particle diameter, is the particle surface potential.98,99 Under the constant ionic strength of the solvent, the maximum repulsive force depends on the parti cle surface potential, the dielectric constant and the particle surface potentia l. According to DLVO (Derjaquin Landau Verwey Overbeek) theory, the energy barrier between each particle which inhibit agglomeration can be also expressed as 2 02 12 r bA V (7-2) where A is the effective Hamaker constant. The effective Hamaker constant depends on the dispersion medium. Considering that the mixed solvent is composed of water and alcohol, the mixed solvent may no t greatly influence the effective Hamaker constant. Therefore, the ionic strength is supposed to be constant in the solvent. From these conditions, the magnitude of repulsive force and the barrier energy is determined by the dielectric constant, the surface potential and particle size. Additionally, Figure 7-3 shows the morphologies of cerium carbonate compounds prepared from different mixed solvent. The pr ecipitates obtained from mixed solvents are composed of spherical secondary agglomerates with small primar y particles of nanometer scale. From these results, it is thought that the particle is grown by the aggregation of fine primary par ticle in mixed solvents. It is interesting to note that the 172

PAGE 173

size of the resultant particles wa s varied by the kinds of solvents. Generally, the primary particles nucleated from solutions are k nown to grow by aggregation or molecular addition with small sub-units. The particle grow th after nucleation can be also affected by the types of solvents, because the particl e interaction potential is different in each solvent. Thus, the secondary particles ar e composed of the aggr egation of sub-unit particles with fine size. In this experiment, the size of the secondary particles increased as the number of carbon in alcohol used as mixed solvent increased. Effect of dielectric consta nt on particle morphology Table 7-1 summarized the dielectric constants of mixed solvent using ethanol and water and the zeta potentials and the morphol ogies of particles obtained from each solvent. The dielectric constants and the zeta potentials decrease as the ratio of ethanol to water increases.67,68 For the pure water, the dielectric constant of water and the zeta potential of the synthesized particles are very high. According to Eq 7-2, the potential energy barrier is relatively high. Under this condition, primary particles may be relatively stable against the aggregation. On the other hand, in the case of the mixed solvent with a higher ratio of 3 ~ 5, the zeta potential and t he dielectric constant are low. So, the zeta potential of the synthesized particles and the dielectric constant of mixed solvent are very low. The precipitated cerium car bonate compounds have a low potential energy barrier and maximum repulsive force. For this condition, primary particles may be relatively unstable as the nucleation occurs However, the colloid al stability of the precipitates increases as the size of the aggregates increases because the potential barrier and the maximum repulsive force increases with the increase in particle size. Figure 7-4 shows the FESEM micrographs of the precipitates prepared using mixed solvent of ethanol and water as a function of the ratio of et hanol to water. The resulting 173

PAGE 174

particles are composed of small and well-def ined particles with spher ical shape. These results indicate that the co mposition of the mixed solvent affects the morphology of particles, which is changed by the dielectric properties of solvent. CMP Materials Preparation of ceria abrasives Figure 7-5 shows the XRD pattern of as -prepared particles of ceria abrasives (cerium carbonate particles) and ceria abrasiv es obtained from thermal decomposition of the cerium carbonate particles at 700 oC. The as-prepared particles were synthesized by using mixed solvent of ethanol and water and the ratio of the solvent was 3. As shown in curve (b), the characteristic peaks corresponding to (111), (200), (220), (311), and (222) planes are located at 2 = 28.54, 33.12, 47.45, 56.32, and 59.02o, respectively. They show very close to the ones with cubic fluorite structure of cerium oxide in JCPDS database. It implies t hat the cerium carbonate compounds are completely transformed into a pure crystalline cerium oxide at 700 oC. Figure 7-6 shows the FESEM micrographs of as-prepared particles of ceria abrasives (cerium carbonate particles) and ceria abrasives obtained from thermal decomposition of the cerium carbonate particles at 700 oC. As shown in Fig. 7-5, the asprepared particles exhibits a rough surface and a spherical shape wit h diameter of 50 ~ 300 nm. After thermal decomposition, the re sultant particles show spherical shape regardless of the crystallizat ion to oxide by heat treatm ent. On the other hand, the surface of particles was rougher due to emissi on of the resultant gases such as carbon dioxide, ammonia, nitride, etc. Figure 7-7 shows the relationship bet ween surface area and crystalline size of ceria abrasives as a function of the calcinat ion temperature. The overall trend reveals 174

PAGE 175

that the surface area of ceria abrasives was reduced and the crystalline size was enhanced with increasing in calci nation temperature. This resu lt attributes to the fact that the sintering mechanisms generally lead to bonding and growth of necks between each particle, so the strength of the particles compact increases during sintering.100 This result corresponds to the result for BET and crystalline size of ceria abrasives mentioned in chapter 5. Characteristics of ceria-base slurry The electrokinetic behaviors of the ceria particles and ceria particles with PAA as a function of pH were investi gated to identify the polishing behavior in oxide CMP. These results are shown in Fig. 7-8. The electrokinetic behavior of each particle is reflected in the interaction between the ceria-based slu rry and the silica wafer. The electrophoretic mobility of all components is strongl y dependent on the suspension pH. The electrophoretic mobility of silica is negative above pH = 2.2, which is the isoelectric point (pHIEP) of silica. The electrophoretic mobilit y of silica decreases with increasing suspension pH. This is attributed to a comp ression of the electrical double layer due to both the dissolution of the Si ion, resulting in an increase of ionic silicate species in solution, and the presence of alkaline ionic species.93 For ceria abrasives, the pHIEP is at about pH 7, and a slightly positive-charged surface below this pH region. However, the pHIEP of the ceria abrasives shifted toward the acidic pH region with additive polymer (PAA). There are two reasons for this behavior. First, the ionization of nearsurface segments partially screens the char ge on the particles, thereby decreasing the shear plane potential. Sec ond, the presence of polymer chains may disturb the hydrodynamic plane of shear, shifting it further out from the particle surface. Because the potential decreases expo nentially with distance,102 the modified shear plane will 175

PAGE 176

experience a lower potential. Above the pHIEP, the electrophoretic mobility of ceria abrasives increased with suspension pH up to the saturation point. Saturation occurs in the pH region of 6, resulting in a negativ ely charged particle whose electrokinetic behavior is essentially no longer dependent on the suspension pH. Figure 7-9 shows the changes in particle size distribution of ceria-based slurry as a function of suspension pH. As shown in Fi g. 7-9, the acidic suspension had broader size distribution of ceria particles and bigger particle size than those of the alkaline or neutral suspensions. Ceria-based slurry in acidic condition became unstable because of lower electrostatic repulsive forces. On the other hand, it is observed that the ceriabased slurry have a better dispersion stability in neutral and alkaline suspension due to higher surface potential. CMP Evaluation Effects of calcination temp erature on physical properties of ceria abrasives Figure 7-10 shows the results for removal rate of oxide and nitride blanket wafers. To evaluate the effects of calcination tem perature on CMP performanc e, four kinds of ceria abrasives were prepared by calcination at 500 ~ 800 oC temperature. The suspension pH of all slurries were adjusted at 6.8 ~ 7.1. As shown in Fig. 7-10(a), it is observed that the ceria-based sl urry had a higher removal rate for the silica wafer as the calcination temperature increased. Especially, the removal rate of oxide film rapidly increased in high calcination temperature above 700 oC. On the other hand, the removal rate of nitride film increased slightly and did not vary much with increasing the calcination temperature as shown in Fig. 7-10(b). Figure 7-11 summarizes the quantitative result s of the polishing test as a function of calcination temperature. The removal rate of oxide film is mainly affected by its 176

PAGE 177

chemical interaction with the ceria abrasives, and by physical parameters including the physical properties of the particles and the mechanical gri nding factor. The surface of silica film is soluble in neutral and alkaline pH solutions due to the dissolution of Si ions. Interaction between the ceria abrasives and the su rface of oxide film occurs in this pH region and SiOCe bonds are formed on the surf ace of oxide film during polishing. As a consequence, the Si-O-Ce bonds can be rapidly removed by the mechanical force generated by pressed pad and abrasives, and this physicochemical reaction lead to the high removal rate of oxide film. In this experiment, the suspension pH was maintained constant for all slurries. Therefore, it seem s that the increase in the removal rate of oxide film is relate to the mechanical factor s rather that to the e ffect of the chemical interaction. These mechanical factors are influenced by several physical parameters of the CMP process, such as morphological aspec ts of the abrasives, cr ystallite size of the abrasives and the CMP conditions. As seen from Fig. 7-7, the crystalline size of ceria abrasives increased as the calcination temperature increased. The removal rate can be significantly affected by the crystalline size of ceria abrasives because indentation is related to the mechanical properties of abrasive hardness and abrasive size. The removal rate of nitride film is affect ed by the physical properties of ceria-based slurry systems and the amount of surfac tant adsorbed on t he film surface.5 The surface of the nitride film during po lishing is passivated with an anionic acrylic polymer in the slurry, which prevents the ceria abrasives from contacting the film surf ace. In this study, the amount of the polym er was also maintained constant for all slurry. Therefore, it seems that the increase in the removal rate of nitride film is relate to the mechanical 177

PAGE 178

factors rather that to the e ffect of passivation la yer of the polymer absorbed on the film surface. The removal selectivity can be calculated by comparing the removal rate between oxide and nitride film. As shown in Fig. 711, the removal selectivity increases with calcination temperature, from 27.8 at 500 oC to 37.4 at 800 oC. This behavior was different with the CMP result s for removal selectivity shown in chapters 4 and 5. In chapters 4 and 5, the removal selectivity showed a transiti on behavior and the existence of optimum abrasive size in slurry. This result attributes that the ceria abrasives used in this experiment have a relatively high su rface area in spite of high calcination temperature. For nitride removal rate, passi vation layer might be insufficiently formed on the nitride film surface due to a largely ads orbed polymer on the particle surface. This adsorption behavior for additive polymer leads to the steady increase in removal rate of nitride. Moreover, the removal rate of oxi de film rapidly increased in high calcination temperature. Therefore, in this experim ent, the removal selectivity increased as calcination temperature increased. Effects of suspension pH on oxide and nitride CMP Figure 7-12 shows the results for remova l rate of oxide and nitride and Table 7-2 summarizes the quantitative results of the polis hing test as a function of suspension pH. The ceria-based slurries were prepared by using ceria abrasives calcined at 700 oC. As shown in Fig. 7-12, it is observed that the ceria abrasives dispersed in neutral and alkaline conditions had high removal rates for silicon dioxide layer and low within-wafer nonuniformity (WIWNU). It is well known that the removal rate of oxide film is dependent on two CMP parameters, the me chanical grinding and the c hemical interaction. The surface of the oxide film is soluble in alka line pH solutions due to the instability of the 178

PAGE 179

silica at this pH. As mentioned in previous section, the interaction between the ceria abrasives and the oxide film occurs during the polishing of the silica blanket wafer, and then SiOCe bonds are formed on the film surf ace. SiOCe bonding on the surface is reported to be the dominant mec hanism in the chemical interaction between silica film and the ceria abrasives.12 Mechanical removal of the SiOSi bonds causes O-Si-O or Si-(OH)4 monomer to be removed as a lump, which is released by the ceria abrasives during CMP process. Therefore, the removal rate of oxide film showed higher value at alkaline region of pH 9 ~ 10 because silicon on the oxide film dissolved, which caused the oxide film to become softer. The reduction in the removal rate of oxide film at t he acidic region is related to the chemical solubility, a decrease in the frictional interactions, and lubrication by polymer adsorption. The dissolution process of silica film in aqueous solution is mainly due to the hydrolysis of SiOS i bonds. However, at acidic region, the Si ions on the silica film seldom dissolve, which results in t he low chemical solubility of the film surface. Thus, the removal rate of oxide film in acidic suspension was lower than in other conditions. Additionally, as shown in Fig. 79, broader particle size distribution in slurry was observed in the acidic slurry because of lower surface potential of ceria abrasives. The presence of agglomerated particles l eads to a decrease in the contact area between the oxide film to be polished and the ceria abrasives. Furthermore, the frictional force between the oxide film and the ceria abrasives is decreased due to the presence of the agglomerated particles because the frictional force is directly proportional to the contact area. Therefore, the removal ra te of oxide film decreased with acidic ceria-based slurry. 179

PAGE 180

The slurry shows a higher WIWNU for the oxide film with acidic ceria-based slurry as shown in Fig. 7-12. This polishing behavior is attributed to a broader particle size distribution of the slurry. The distribution data in Fig. 7-9 shows that the particle size distribution broadens with the decrease in suspens ion pH. The inhomogeneous distribution in acidic pH regi on leads to local frictional interactions of the agglomerated particles, causing different removal rates between the center and the edge of the wafer due to their limited mobility on the wafer surface. Therefore, it seems that surface uniformity of oxide film is re lated to the particle uniformity of the ceria-based slurry. Conclusions Synthesis of Cerium Carbonates Spherical cerium carbonate compounds were synthesized by using mixed solvent of alcohol and water. The obtained results indicated that physicochemical solvent properties had a significant effect on t he crystalline phase, microstructures and morphological properties of the resultant particles. FE SEM results revealed the morphology of cerium carbonates could be changed by the type of solvent. In this experiment, the particles obtained from pure water were composed of oval-like particles with the size of 300 ~ 500 nm. In contrast, the particles pr epared by using alcohols had a spherical shape with uniform size distributio n. FESEM analysis also showed that the size of the cerium carbonates increased as the number of carbon in alcohol used as mixed solvent increased. Moreover, the size of the precipitant s decreased with the increase in the ratio of ethanol to water. Synthesis of Ceria Abrasives Spherical ceria abrasives were prepared by thermal decomposition of the synthesized spherical cerium carbonates at 700 oC. XRD result revealed that the cerium 180

PAGE 181

carbonates were completely transformed in to a pure crystalline cerium oxide at 700 oC. From FESEM analysis, ceria particles after calcination process exhibited spherical shape regardless of the crystallization to ox ide by heat treatment. On the other hand, the surface of particles was rougher due to emission of the resultant gases such as carbon dioxide, ammonia, ni tride, etc. BET and XRD resu lts showed that the surface area of ceria abrasives was reduced and the crystalline size was enhanced with increasing in calcination temperature. Preparation of Ceria-based Slurry Ceria-base slurries were prepared by dispersing the synthesized particles and additive polymer into DI water. T he obtained results showed that the pHIEP of ceria abrasives shifted toward the acidic pH region with additive polymer. Moreover, the acidic suspension had broader size distribution of ceria particles and bigger particle size than those of the alkaline or neutral suspensions in terms of electrostatic repulsive forces. CMP Evaluation In this study, the effect of the calcination temperature on the physical properties of synthesized ceria particles and suspension pH on CMP performance were investigated. The CMP results revealed that the ceria-bas ed slurry had a higher removal rate for the silica and nitride wafer as the calcination te mperature increased. Moreover, the removal selectivity increased as calcination temperat ure increased. For suspension pH, the ceria abrasives dispersed in neutral and alkaline co nditions had high removal rates for silicon dioxide layer because the surface of the oxi de film is soluble in alkaline and neutral pH solutions. In addition, the surface uniformity deteriorated at acidic pH solution because of a broader particle size distribution of the ceria abrasives in the slurry and the 181

PAGE 182

presence of agglomerated abrasives. Furthermo re, the surface uniformity deteriorated with acidic pH solution due to inhomogeneous di stribution leading to local frictional interactions of the agglomerated particles. 182

PAGE 183

Figure 7-1. XRD patterns of cerium co mpositions produced under precipitation conditions with different solvent 183

PAGE 184

Figure 7-2. FESEM micrographs of ceri um carbonate compounds obtained by using pure water as solvent 184

PAGE 185

Figure 7-3. FESEM micrographs of spherical cerium carbonate particles prepared from the mixture of water and differ ent alcohols; (a) methanol (CH3OH), (b) ethanol (C2H5OH), (c) 2-propanol (C3H8O), and (d) 1, 4-butandiol (C4H10O2) 185

PAGE 186

Figure 7-4. FESEM micrographs of cerium carbonate compounds prepared by various ratio of ethanol to water: (a ) 0, (b) 1, (c) 3, and (d) 5 186

PAGE 187

Figure 7-5. The XRD pattern of (a) ceri um carbonate prepared by using mixed solvent of ethanol and water and (b) ceria abrasives obtained from thermal decomposition of the cerium carbonate at 700 oC 187

PAGE 188

Figure 7-6. FESEM micrographs of (a) as-p repared particles of ceria abrasives and (b) ceria abrasives obtained from thermal decomposition at 700 oC 188

PAGE 189

Figure 7-7. Relationship betw een surface area and crystalline si ze of ceria abrasives as a function of calcination temperature 189

PAGE 190

Figure 7-8. Electrokinetic behavior of silic a, ceria and ceria with surface active agent added as a function of suspension pH 190

PAGE 191

Figure 7-9. The changes in particle size dist ribution of ceria-based solvent as a function of suspension pH 191

PAGE 192

Figure 7-10. The CMP evaluation for remova l rate of oxide and nitride films as a function of calcin ation temperature 192

PAGE 193

Figure 7-11. Results of CMP field eval uation for removal rate and selectivity 193

PAGE 194

Figure 7-12. The CMP evaluation for removal rate of silicon oxide wafer as a function of suspension pH 194

PAGE 195

Table 7-1. Dielectric constants of mixed so lvent, zeta potentials and morphologies of cerium carbonate compounds with the ratio of ethanol to water22,23 Ratio of ethanol to water at 20 oC Dielectric constant of solvent at 20 oC Zeta potential at 20 oC Morphology 0 80.01 18.3 Spindle and spherical 1 56.53 11.2 Spindle and spherical 3 49.50 4.8 Spherical 5 42.46 3.6 Spherical 195

PAGE 196

CHAPTER 8 A COMPARISION OF CMP PERFORMANCE IN THE CERIA ABRASIVES SYNTHESIZED VIA VARIOUS METHODS Introduction As shown in previous chapters, t here are many methods to manufacture ceria particles. Most of methods introduced in this paper give rise to spherical ceria particles because it is thought that slurries formulated using such particles will be required for future CMP technology in order to meet the ever more challenging defectivity requirements. On the other hands certain methods lead to ceri a particles with irregular shape due to its cubic crystalline structure. However, ther e is no substantive evidence that the round particles are s uperior to the irregular particles in effective CMP slurry. Although the slurries containing spherical particles are promoti ng, it has not been possible to convincingly tie in morphology wit h removal rate or def ectivity. In fact, surface chemistry or particle size distribut ion of ceria abrasives may appear to be more important than morphological properties considering the polishing mechanism for removal rate of oxi de and nitride layer. In this chapter, I intend to investigate the effects of abrasive material properties on polishing removal rate and wafer defectivity by using different kinds of ceria particles obtained from previous chapters. Materials and Methods Sample Preparation Preparation of ceria abrasives Four types of ceria particles obta ined from different me thods, including solution growth method (hydrothermal method), grai n control method (flux method), core-shell composition method (surface-induced precipitation method), and solid state method 196

PAGE 197

(thermal decomposition method), were used as abrasives for ceria-based slurry at pH 6 ~ 7. The size ceria abrasive with diameter of 140 ~ 170 nm was controlled by adjust the reaction parameters of each method. Preparation of ceria-based slurries Different ceria-based slurries were formu lated by dispersing abrasives each with different primary particle size in DI wate r containing an anionic organic polymer (Poly acrylic acid, PAA; Mw 4000, LG Chem.) as di spersant. 2 wt% of PAA based on the total weight of the ceria abrasives was added. For each slurry, pH was adjusted to 6 ~ 7 by adding ammonium hydroxide (NH4OH). The solid loading of ce ria abrasives was fixed to 2.0 wt%. CMP tools and consumables Silicon dioxide film of 2 m thick was grown on a 5-in. p-type silicon substrates with (001) orientation by plasma enhanced chemical vapor deposition (PECVD). The silicon nitride films were deposited by usi ng low-pressure chemical vapor deposition (LPCVD). Polishing tests were performed on a rotary type CMP machine (GNP POLI 400, G&P technology) for one minute with eac h of the ceria-based slurries. IC 1000/SUBA IV stacked pads (supplied by R odel Inc.) were utilized as CMP pads. The downforce was 4 psi and the rotation s peed between the pad and the wafer was 100 rpm. The slurry flow rate was 100 mL/min. Characterization Abrasives The crystal structure and grain size was identified by x-ray diffraction (XRD) using CuK radiation. The grain size was estimated by the Scherrer equation according to the formula D = 0.9 / ( cos ), where D is the grain size, is the wavelength of x-rays, 197

PAGE 198

is the half-width of the diffraction peaks, and is the diffraction angle. The broadening of the reflection from the (111) plane was used to calculate the grain size. The morphology and size of the precipitate par ticles were examined by field emission scanning electron microscope (FESEM). The average primary particle size was calculated by measuring ca. 100 particles from FESEM micrographs. The specific surface area (SSA) of t he ceria abrasives was determined by Brunauer-Emmett-Teller (BET) method using nitrogen adsor ption/desorption at 77 K. Ceria-based slurry The abrasive size distribution of slurry was measured using light scattering method (UPA 150, Microtrac Inc.). CMP performance The film thickness on the wafers bef ore and after CMP was measured using spectroscopic reflectometry (Nanospec 6100, Nanometrics) to calculate the removal rate. Defectivity of wafer after CMP was measured by using a KLA-Tencor puma 91XX. This system uses a UV/visible light s ource to illuminate defect types. The average polishing data for removal rate was carri ed by performing the same tests more than three times in order to support the validity of the results from the statistical viewpoint. Results and Discussion Comparison in Polishing Removal Rate Fig. 8-1 shows FESEM micrographs of the four types of ceria abrasives prepared from various methods as described in Tabl e 8-1. The size of ceria particles was controlled by adjusting the reac tion parameters of each method As shown in Fig. 8-1, all abrasives except sample A have a round shape with diameter of 140 ~ 170. The sample A shows irregular morphology posse ssing shape edges and wide size variations 198

PAGE 199

between each particle. In the ca se of sample C, the grain size of particles could be not measured by Scherrer equation because t he particles compose of core/shell composites. Table 8-2 summarizes the quantitative results of removal rate and WIWNU of oxide wafer after CMP process. For oxide remo val rate, the slurries including sample B and D showed higher removal rare than other slu rries. This result attributes to the fact that the ceria particles obtained from solid state method are brittle and easily brokendown by applied pressure and shear forc e during CMP process. As mentioned previously, the physicochemical reaction by Si-O-Ce bonding leads to the high removal rate of oxide film. At this poi nt, it is thought that the brok en fragments of ceria abrasives promote more Si-O-Ce bonds onto oxide film and increase contact area between abrasives and wafer surface. Thus, the oxide layer can be more rapidly removed by the brittle behavior of ceria abrasives. In chapt er 5 and 7, they were commonly synthesized via thermal pyrolysis (calcination), leading to softness and high chemical reactivity to oxide film. Moreover, using FESEM results in Fig. 4-12 and 5-7, we could confirm that the ceria abrasives obtained from solid st ate method are easily broken-down by applied pressure and shear force during polishing, wh ile the ceria abrasives synthesized from solution state method are less brittle and don t fracture upon applied pressure during polishing. Therefore, it was found that the removal rate of oxide film is dependent on synthesis method of ceria abrasives. It seems that the brittle behavior of ceria abrasives plays an important role in the physicochem ical reaction mechanism for oxide CMP. Comparison in WIWNU For WIWNU, the slurries includi ng sample C and D showed a lower WIWNU for the oxide film. This result is related to parti cle size distribution of the slurry. Fig. 8-2 199

PAGE 200

shows the particle size distribution of the ceria abrasives used in this experiment. Although the size of four types of ceria abrasives were simila rly controlled to diameter of 140 ~ 170 nm, there was difference in the particle size distribution of slurries as shown in Fig. 8-1 and 8-2. Sample C and D have a narrow particles size distribution, while sample A and B have a relatively boarder parti cle size distribution. The slurry with narrow particle size distribution and uni form particle size can uniformly impose interaction force of ceria abrasives over the whole wafer surfac e. This can enhance wafer uniformity after CMP process. On the other hand, the broader particle size distribution of large abrasives can cause different removal rates between the center and the edge of the wafer due to thei r limited mobility on the wafer surface. This result is also consistent with the WIWNU results of chapter 4 and 5. Therefore, uniformity of ceria abrasives and narrow particle size dist ribution should be optimized to realize the global planarization. Abrasive Effects on Defectivity Fig. 8-3 shows a comparison on tota l scratch defect after polishing in CMP process using various ceria abrasives. As shown in Fig. 8-3, the ceria abrasives obtained from solid state method exhibit su perior defect performance over the ceria abrasives prepared from solution state method. It seems that this result is related to the morphological properties and crystalline stru cture of ceria abrasives. As mentioned previously, the ceria-base slurry with non-uniform particle size distribution not only created surface deformation but also changed the polishing removal rate. For solution growth ceria, the sharp edge of the abrasives can be regarded as another factor for roughness of wafer. The film abraded by the sharp edge has a higher local pressure to generate more friction force during polishing. This behavior can induce serious defects 200

PAGE 201

by pit formation on the wafer surface. Mo reover, the ceria abrasives are not easily broken and dont fracture upon applied pressure during CMP process. This means that the particles have a higher hardness leading to the scratch defects on the wafer surface. On the other hand, the abrasives obtained from solid state method have brittle properties on wafer surface to be polished. It is thought that this brittle behavior induces the reduction in the number of scratch defect on wafer surface. Theref ore, it seems that surface defectivity of oxide film is related to the mechanical factors (crystalline structure) and morphological properties of ceria abrasives used during CMP. Conclusion To compare the polishing properties of ceria abrasives according to synthesis method, CMP tests for four types of ceria-ba sed slurry were preformed. The synthesis methods were classified by solution state method, including hydrothermal method and surface-induced precipitation method, and solid state method, including thermal decomposition method and flux method. For ox ide removal rate, the result showed that the brittle behavior of ceria abrasives pla ys an important role in the physicochemical reaction mechanism for oxide CMP. This br ittle behavior also affe cted the reduction in defectivity of wafer surface. The result for WIWNU showed that the slurry with narrow particle size distribution and uniform particle size help to improve wafer uniformity after CMP process. 201

PAGE 202

Figure 8-1. FESEM micrographs of various kinds of spherical ceria abrasives synthesized by variety methods; (a) hydr othermal method, (b) flux method, (c) surface-induced precipitation met hod, and (d) thermal decomposition 202

PAGE 203

Figure 8-2. Particle size distribution of ceria-based slurries 203

PAGE 204

Figure 8-3. Comparison of different ceria abrasives on surface defectivity 204

PAGE 205

Table 8-1. Comparison of ceria abrasives used in this study Sample Synthesis method Primary particle Size (SEM, nm) Grain size (XRD, nm) Slurry mean size (UPA, nm) Surface area (m2/g) A Hydrothermal (Solution state reaction) 163 45 326 11.14 B Flux (Solid state reaction) 166 43 278 14.06 C Precipitation coating (Solution state reaction) 146 198 37.63 D thermal decomposition (Solid state reaction) 158 29 242 22.02 205

PAGE 206

Table 8-2. The result s of the CMP evaluation Samples Oxide removal rate (/min) WIWNU of oxide film (%) Slurry A 2369.4 10.1 Slurry B 3390.9 11.0 Slurry C 245.8 0.765 Slurry D 2824.5 6.02 206

PAGE 207

CHAPTER 9 CONCLUSIONS Ceria abrasive particles are becoming more widely used in a variety of CMP applications for multilevel IC manufacture. However, only a few years ago, most facilities traditionally used silica particles fo r all IC applications, mainly to remove deposited oxide topography fo r bulk oxide removal for ILD and STI applications. Chemical additives in CMP slurries were formulated with fumed or colloidal silica particles in order to improve the removal selectivity in STI CMP. As the device design rule decreased, ceria particles have been used instead of silica particles, since ceriabased slurries address many of the issues resu lting from the use of silica-based slurries. Recently, the ceria-based slurries are bei ng introduced not only in ILD but also STI CMP step in order to improve removal rate, chip uniformity, and selectivity. In spite of its advantages, ceria particles usually induce higher scratch level than silica particles due to its cubic crystalline structure, irregular shape, and poor dispersion stability in slurry. Especially, flash memory devices such as NAND or NOR type are more sensitive to CMP scratch than conven tional DRAM devices because of the lack of their redundancy. Therefore, it is very important in CMP process to reduce CMP scratch level and increase device yiel d. Additionally, although an understanding of nature of mechanical interaction between ceria particl es and substrates to be polished is essential to maintain the strict process requirements for manufacturing current and future generation IC devices, the fundamental knowledge of its mechanical behaviors on planarization performance is not clear. To accomplish this aim, this research developed novel ceria particles for CMP abrasives and introduced novel technology to synthesize these particles. Furthermore, 207

PAGE 208

this article was intended to provide a fundam ental basis for the directions of ceria particles needed for present-day and ne xt generation CMP technologies. Solution Growth Abrasives Well-crystalline ceria particles were synt hesized by heating peptized ceria sol as precursor under hydrothermal conditions. The morphology and the crystallites size of ceria particles were controlled by varying t he dielectric property of solvent used in preparation of the ceria prec ursor. The synthesized particles exhibit cubic fluorite structure with size ranged from 20 to 400 nm without the formation of hard aggregates. In this work, the relationships between dielectric property of the solvent and particle size were discussed in terms of the supersaturati on of solute. In addition, the influences of precipitation participating anions (OH-) and acidic hydrothermal medium on crystallites size of ceria particles were investigated. For CMP performance evaluation, the effect s of single crystalline ceria abrasives in CMP slurries were investigated for s ilicon dioxide and silicon nitride CMP process. The size of ceria abrasives was controlled by varying hydrothermal reaction conditions. Polishing removal rate was measured with f our slurries, with different mean primary particle size of 62 nm, 116 nm, 163 nm and 232 nm. The polishing results showed that the single crystalline ceria abrasives were not easily broken-down by mechanical force during CMP process. It was found that the removal rate of ox ide and nitride film strongly depend upon abrasive size, whereas the surfac e uniformity deteriorates as abrasive size increases. The observed polishing result s confirmed that ther e exists an optimum abrasive size (163 nm) for maximum removal selectivity between oxide and nitride films. The polishing behavior of the si ngle crystalline ceria abrasives was discussed in terms of morphological properties of the abrasive particle. 208

PAGE 209

Grain Control Abrasives The effects of spherical ceria abrasives in CMP slurries were investigated on silicon dioxide and silicon nitride polishing proc ess. The ceria abrasives were prepared by the flux method, using potassium hydrox ide as the grain growth accelerator. The primary particle size of the ceria abrasives was controlled in the range of ~ 84 417 nm by changing the concentration of potassium hydroxide and the calc ination temperature without mechanical milling process. The removal rate of silicon dioxide film strongly depended upon abrasive size up to an optimum abrasive size (295 nm) after CMP process. However, the surface uniformity deteriorated as abrasive size increases. The observed polishing results confirmed that t here exists an optimum abrasive size (295 nm) for maximum removal selectivity between oxide and nitr ide films. In this study, polishing behaviors of the spherical ceri a abrasives were discussed in terms of morphological characteristics. Core/shell Composite Abrasives Monodispersed ceria coated silica particles were prepared using a peptized ceria sol as coating precursor. The ceria coating precursor was synthesized by alkoxide method, which employs ethanol as solvent. T he resulting particles were characterized with scanning electron microscopy (SEM), tr ansmission electron microscopy (TEM), Xray photoelectron spectroscopy (XPS), X -ray diffraction (XRD) and zeta potential measurements. It was found that crystalline ce ria coating was formed on the surface of silica particles at temperature of 60 oC without requirement for the post-heat treatment. The thickness of ceria coating was controlled by changing the concentration of the coating precursor. The influence of soluti on pH on the formation of ceria coating was investigated in terms of electrostatic attraction mechanism. The apparent surface 209

PAGE 210

coverage was estimated using an isoelectric point (IEP) data. In this study, the resultant particles were used as CMP abrasives for ceri a-based slurry to investigate the effects of the ceria coated silica particles on polis hing performance during oxide CMP. The primary particle size was controlled about 146 nm without mechanical milling process by depositing the ceria coating on t he surface of colloidal silica particles with size 135 nm. The removal rate of silicon dioxide film strongly depended upon suspension pH. With increasing suspension pH, the removal rate of silica film decreased. On the other hand, surface uniformity deteriorated as the suspension pH decreases. These polishing behaviors were related to electrophoretic m obility of abrasives for silica film to be polished, since surface charge of the abrasives can be changed by suspension pH. Thus, it means that absorption/repulsion behavior between abrasives and materials to be polished plays an important role in polishing performance dur ing CMP process. Additionally, the removal rate of silica film increased with increasing down pressure in terms of mechanical aspect for CMP condition. Solid State Abrasives The ceria abrasives were developed via tw o-step procedure; simple precipitation method using mixed solvent and thermal decom position method using spherical cerium carbonate compounds. In the first step, s pherical cerium carbonate compounds were synthesized by using mixed solvent of alc ohol and water. The obtained results indicated that physicochemical solvent properties had a significant effect on the crystalline phase, microstructures and morphological properties of the resultant parti cles. FESEM results revealed the morphology of cerium carbonates could be changed by the type of solvent. In this experiment, the parti cles obtained from pure water were composed of oval-like particles with the size of 300 ~ 500 nm. In contrast, t he particles prepared by using 210

PAGE 211

alcohols had a spherical shape with uniform size distribution. FESEM analysis also showed that the size of the cerium carbonates increased as the number of carbon in alcohol used as mixed solvent increased. Moreover, the size of the precipitants decreased with the increase in the ratio of ethanol to water. In the second step, spherical ceria abrasives were prepared by thermal decomposition of the synthesized spherical cerium carbonates at 700 oC. XRD result revealed t hat the cerium carbonates were completely transformed into a pure crystalline cerium oxide at 700 oC. From FESEM analysis, ceria particles after calc ination process exhibited spherical shape regardless of the crystallizat ion to oxide by heat treatm ent. On the other hand, the surface of particles was rougher due to emissi on of the resultant gases such as carbon dioxide, ammonia, nitride, etc. BET and XR D results showed that the surface area of ceria abrasives was reduced and the crysta lline size was enhanced with increasing in calcination temperature. In CMP test, the effect of the calc ination temperature on t he physical properties of synthesized ceria particles and suspension pH on CMP performance were investigated. The CMP results revealed that the ceria-bas ed slurry had a higher removal rate for the silica and nitride wafer as the calcination te mperature increased. Moreover, the removal selectivity increased as calcination temperat ure increased. For suspension pH, the ceria abrasives dispersed in neutral and alkaline co nditions had high removal rates for silicon dioxide layer because the surface of the oxi de film is soluble in alkaline and neutral pH solutions. In addition, the surface uniformity deteriorated at acidic pH solution because of a broader particle size distribution of the ceria abrasives in the slurry and the presence of agglomerated abrasives. Furthermo re, the surface uniformity deteriorated 211

PAGE 212

with acidic pH solution due to inhomogeneous di stribution leading to local frictional interactions of the agglomerated particles. Comparison of Polishing Behavior To compare the polishing properties of ce ria abrasives according to synthesis method, CMP tests for four types of ceria-ba sed slurry were preformed. The synthesis methods were classified by solution state method, including hydrothermal method and surface-induced precipitation method, and solid state method, including thermal decomposition method and flux method. For ox ide removal rate, the result showed that the brittle behavior of ceria abrasives pla ys an important role in the physicochemical reaction mechanism for oxide CMP. This br ittle behavior also affe cted the reduction in defectivity of wafer surface. The result for WIWNU showed that the slurry with narrow particle size distribution and uniform particle size help to improve wafer uniformity after CMP process. 212

PAGE 213

APPENDIX A DIELECTRIC CONSTANTS OF MIXED SO LUTION OF SOME ORGANIC SOLVENT AND WATER AT ROOM TEMPERATURE 020406080100 0 10 20 30 40 50 60 70 80 90 Methanol Ethanol n-Propanol Isopropanol 1,4-Buthanol Etylene glycol GlycerolVol. % of alcoholDielectric constant ( Figure A-1. The dependence of dielectric c onstant on the composition of different alcohols and water.68 213

PAGE 214

LIST OF REFERENCES 1. J. M. Steigerwald, S. P. Murarka, and R. J. Gutmann, Chemical Mechanical Planarization of Microelectric Materials, John Wiley & Sons. INC., New York (1997). 2. C. F. Lin, W. T. Tseng, and M. S. Feng, J. Electrochem. Soc ., 146 1984 (1999). 3. K. D. Beyer, IBM Micronews 5 40 (1999). 4. W. Choi, J. Abiade, S. Lee, and R. K. Singh, J. Electrochem. Soc ., 151(8) G512 (2004). 5. H. -G. Kang, H. -S. Park, U. Paik, T. Katoh, and J. -G. Park, J. Mater. Res., 22(3) 777 (2007). 6. J. H. So, D. J. Lee, Y.-J. Cho, M.-H, Ahn, I.-Y. Lee, S.-J. Je on, S. Kim, and S.-M. Yang, in Proceedings of the Kore an CMP-UGM Conference The Korean Institute of Electrical and Electronic Mate rial Engineers, Paper 2 (2005). 7. S. K. Kim, Y. H. Kim, U. Paik, T. Katoh, and J. -G. Park, J. Electrochem. Soc., 154(7) H642 (2007). 8. P. Janos, M. Petrak, Preparation of ceriabased polishing powders from carbonates, J. Mater. Sci. 26 (1991) 4062-4066. 9. Y. N. Prasad and S. Ramanathan, J. Electrochem. Soc ., 9(12) G337 (1999). 10. R. K. Singh, S. M. Lee, K. S. Choi, G. B. Basim, W. Choi, Z. Chen, and B. M. Moudgil, Mater. Res. Bull., 27, 752 (2002). 11. M. J. Cumbo, D. Fairhurst, S. D. Jacobs, and B. E. Puchebner, Applied Optics 34, 3743 (1995). 12. L.M. Cook, J. Non-Cryst. Solids, 120 152 (1990). 13. R. Jairath, M. Desai, M. Stell, R. Tolles, and D. Scherber-Brewer, Mater. Res. Soc. Symp. Proc. 337 121(1994). 14. Y. Xie and B. Bhushan, Wear 200 281 (1996). 15. T. Izumitani, Polishing lapping and diamond gr inding of optical glasses in Treaise on Material Science and Technology, M. Tomozawa and R. Dorrmus, Editors, p. 115, Academic Press, New York (1979). 16. U. Mahajan, M. Biel mann, and R. K. Singh, Mater. Res. Soc. Symp. Proc ., 566 27 ( 1999). 17. M. Bielmann, Masters Thesis, University of Flor ida, Gainesville, FL (1998). 214

PAGE 215

18. S. D. Kim, I. S. Hwang, H. M. Park, J. K. Rhee, and C. W. Nam, J. Vac. Sci. Technol. B, 20, 918 (2002). 19. J. W. Lee, B. U. Yoon, S. Hah, and J. T. Moon, Mater. Res. Soc. Symp. Proc., 671 M5.3.1 (2001). 20. O. Kubaschewski, E. Evans, and C. Alcock, Metallurgical Thermochemistry Pergamon Press, Oxford (1967). 21. K. Osseo-Asare, J. Electrochem. Soc ., 149 G651 (2002). 22. M. Jiang, N. O. Wood, and R. Komanduri, J. Eng. Mater. Technol., 120 30 (1998). 23. H. Zhu, A. B. Tessarotto, V. A. Greenhut, D. E. Niesz, and R. Sabia, Ceram. Trans ., 102 259 (1999). 24. P. W. Carterz and T. P. Johns, Electrochem Solid-State Lett., 8(8) G218 (2005). 25. L. Bergstrom and E. Bostedt, Colloids Surf ., 49, 183 (1990). 26. S. Mezzasalma and D. Baldovino, J. Colloid Interface Sci ., 180 413 (1996). 27. C. Galassi, F. Bertoni, S. Ardizzone, and C. L. Bianchi, J. Mater. Res ., 15, 155 (2000). 28. E. Laarz, B. V. Zh mud, and L. Bergstrom, J. Am. Ceram. Soc ., 83, 2394 (2000). 29. S. K. Kim, S. Lee, U. Pa ik, T. Katoh, and J. G. Park, J. Mater. Res., 18, 2163 (2003). 30. W. G. America and S. V. Babu, Electrochem. Solid-State Lett., 7 G327 (2004). 31. L. Wang, W. M. Sigmund, and F. Aldinger, J. Am. Ceram. Soc., 83 697 (2000). 32. A. Philipossian and M. H anazono, Tribology and fluid dynamics characterization of cerium oxide slurries, www.innovativeplanarization.com (2001). 33. S. K. Kim, P. H. Yoon, U. Paik, T. Katoh, and J. G. Park, Jpn. J. Appl. Phys., 43 7427 (2004). 34. Y.Z. Hu, R.J. Gu tmann, and T.P. Chow, J. Electrochem. Soc., 145, 3919 (1998). 35. S. D. Kim, I. S. Hwang, and H. M. Park, J. Vac. Sci. Technol. B, 20, 918 (2002). 36. Y. Zhuang, D. King, T. Kido, and A. Philipossian, Jpn. J. Appl. Phys., 44, 30 (2005). 37. M. Jun, S. Atsushi, Y. Masato, H. Keizou, A. To ranosuke, and O. Yuuto, US patent 6615499 (2003). 38. H. G. Kang, T. Ka toh, D. H. Kim, U. Paik, and J. G. Park, Jpn. J. Appl. Phys. 44, 238 215

PAGE 216

(2005). 39. M. Jun, S. Atsushi, Y. Masato, H. Keizou, A. Tor anosuke, O. Yuuto, US patent 6615499 (2003). 40. Y. Masato, A. Toranosuke, T. Hiroki, K. Yasushi, M. Ju n, T. Kiyohito, and O. Yuuto, US patent 6863700 (2005). 41. E. Remsen, S. Anjur, D. Boldrige, M. Kamiti, S. Li, T. Johns, and C. Dowell, J. Electrochem. Soc., 153(5) G453 (2006). 42. E. Matjevic, and W. Hsu, J. Colloid. Interface Sci., 118 506 (1987). 43. M. Hirano, Y. Fukuda, H. Iwata, y. Hotta, and M. Inagaki, J. Am. Ceram. Soc., 83, 1287 (2000). 44. M. Hirano, and E. Kato J. Am. Ceram. Soc., 79, 777 (1996). 45. M. Hirano, and E. Kato J. Am. Ceram. Soc., 82, 786 (1999). 46. X. Chu, W. Chung, and L. D. Schmidt, J. Am. Ceram. Soc., 76, 2115 (1993). 47. Y. Zhow, R. J. Philips, and J. A. Switzer, J. Am. Ceram. Soc., 78, 981 (1995). 48. J. H. Park, H. Cui, S. H. Yi, and J. G. Park, J. Mater. Res., 23(12) 3323 (2008). 49. F. Preston, J. Soc. Grass Technol ., 11, 214, (1927). 50. J. Luo and D. A. Dornfeld, IEEE Trans. Semicond. Manuf ., 14(2) 112 (2001). 51. J. Luo and D. A. Dornfeld, IEEE Trans. Semicond. Manuf ., 16(1) 45 (2003). 52. Y. Moon, Ph.D. Thesis, University of California Berkeley, CA (1999). 53. T. Detzel, S. Hosali, A. Sethuraman, J. F. W ang, and L. Cook, in Proceedings of the CMP-MIC Conference, IMIC, 2007, p. 202. 54. G. B. Basim, J. J. Ad ler, U. Mahajan, R. K. Singh, and B. M. Moudgil, J. Electrochem. Soc., 147 3523 (2000). 55. T. L. Neo, E. S. Y. Shang, and C. M. Chong, in Proceedings of the International Symposium on Semiconductor Manufac turing, IEEE, 2001, p. 321. 56. Y. J. Seo, G. U. Kim, and W. S. Lee, Microelectron. Eng., 71, 209 (2004). 57. D. Boning, B. Lee, C. Oj i, D. Ouma, T. Park, T. Sm ith, and T. Tugbawa, Mater. Res. Soc. Symp. Proc., 566 P5.5, (2000). 58. R. J. Hunter, Foundations of colloid science Oxford University Press, (2001). 216

PAGE 217

59. I. D. Morrison and S. Ross, Colloidal suspensions: Suspensions, Emulsions, and Foams John Wiley and Sins, New York, (2002). 60. H. Ota, M. Hachiya, Y. Ichiyasu,, and T. Kurenuma, Hitachi Review 55(2) 78 (2006). 61. http://www.testwafer.com/PDF/skw1-200_spec.pdf 62. S.H. Lee, Z.L. Lu, S.V. Babu, and E. Matijevic, J. Mater. Res ., 17(10) 2744 (2002). 2744-2749. 63. N.C. Wu, E.W. Shi, Y.Q. Zheng, W.J. Li, J. Am. Ceram. Soc ., 85(10), 2462 (2002). 64. H. K. Park, D. K. Kim, C. H. Kim, J. Am. Ceram. Soc ., 80, 743 (1997). 65. C. S. Fang and Y. W. Chen, Mater. Chem. Phys., 78, 739 (2003). 66. M.Z.C. Hu, E. A. Pa yzaut, C.H. Byers, J. Colloid. Interface. Sci., 222 20 (2000). 67. D. R. Lide, CRC handbook of Chemistry and Physics, 87th edn ., CRC press, New York (2007). 68. G. Akerlof, J. Am. Chem. Soc ., 54, 4125 (1932). 69. J. N. Israelachvili, Intermolecular and Surface Forces, 2nd edn ., Academic press, London (1992). 70. J. W. Mullin, Crystallization, 3rd edn ., Butterworth-Heinemann, London (1992). 71. P. L. Chen and I. W. Chen, J. Am. Ceram. Soc ., 76, 1577 (1993). 72. J. G. Li, T. Ikegami, J. H. Lee, and T. Mori, Acta Mater., 49, 419 (2001). 73. X. D. Zhou, W. Huebn er, and H. U. Anderson, Appl. Phys. Lett ., 80, 3814 (2002). 74. N. C. Wu, E. W. Shi, Y. Q. Zheng, and W. J. Li, J. Am. Ceram. Soc ., 85, 2462 (2002). 75. T. Allen, Particle si ze measurement, Vol. 1, 5th ed., Chapman and Hall, New-York (1997). 76. Y.H. Kim, S.K. Kim, N.S. Kim, J.G. Park, and U. Paik, Ultramicroscopy 108 1292 (2008). 77. A. R. West, Solid State Chemistry and its Applications Wiley, New York (1995). 78. F. Bondioli, A. B onamartini Corradi, C. Leonel li, and T. Manfredini, Mater. Res. Bull ., 34, 2159 (1999). 217

PAGE 218

79. C. Kleinlogel and L. J. Gauckler, Adv. Mater., 13(14) 1081 (2001). 80. H. Wakita and S. Kinoshita, Bull. Chem. Soc. Jpn ., 52(2) 428 (1979). 81. U. Mahajan, M. Be lmann, and R.K. Singh, Electrochem. Solid-State Lett ., 2 46 (1999). 82. Y. X. Li, X. Z. Zhou, Y. Wang, and X. Z. You, Mater. Lett ., 58, 245 (2004). 83. A. Yoshida, The colloidal chemistry of silica, advance in chemistry series, vol. 234, Oxford Univ. Press, Oxford (1994). 84. W. Stber, A. Fink, and E. Bohn, J. Colloid Interface Sci. 26, 62 (1968). 85. X. Song, N. Jiang, Y. Li, D. Xu, and G. Qiu, Mater. Chem. Phys ., 110 128 (2008). 86. M. S. Kaliszewski and A. H. Heuer, J. Am. Ceram. Soc ., 73, 1504 (1990). 87. T. Ikegami and N. Satio, J. Ceram. Soc. jpn .,104 469 (1996). 88. N. Thromat, M. Gautier-Soyer, and G. Bordier, Suf. Sci. 345 290 (1996). 89. S. Gavarini, M. J. Guitte t, P. Trocellier, M. Gautier-S oyer, F. Carrot, and G. Matzen, J. Nuclear Mater ., 322 111 (2003). 90. B. Cheng, Li Zhao, J. Yu, and X. Zhao, Mater. Res. Bull ., 43, 714 (2008). 91. J. Yu, Li Zh ao, and B. Cheng, Mater. Chem. Phys ., 96, 311 (2006). 92. X.D. Zhou, W. Huebner and H. U. Anderson, Appl. Phys. Lett ., 80(20) 3814 (2002). 93. P. Suphantharida and K. Osseo-Asare, J. Electrochem. Soc ., 151(10) G658 (2004). 94. T. Murata, F.S. Howell, and K. Itatani, J. Mater. Sci ., 36, 1277 (2001). 95. I. S. Seog, C. H. Kim, J. Mater. Sci ., 28, 3277 (1993). 96. Y.T. Moon, H.K. Park, D.K. Kim, I.S. Seog, and C.H. Kim, J. Am. Ceram. Soc ., 78, 1103 (1995). 97. R.J. Hunter, Foundat ion of colloid science, Vol. 1, p. 443, Clarendon press, Oxford, UK, 1987. 98. J.L. Look, C.F. Zukoski J. Colloid. Interface. Sci. 153 469 (1992). 99. J.L. Look, C.F. Zukoski J. Am. Ceram. Soc. 78 21 (1995). 100. M.N. Rahaman, Ceramic processing, CRC Press, New York (2007). 218

PAGE 219

219 101. V.A. Hackley and S.G. Malghan, J. Mater. Sci ., 29, 4420 (1994). 102. J. Scheutjens and G. J. Fleer, J. Phys. Chem ., 83, 1619 (1979).

PAGE 220

BIOGRAPHICAL SKETCH Oh, Myoung Hwan was born in Suwon, S outh Korea. In February of 2001, he received Bachelor of Engineering from chemical engineering of Korea University in Seoul, South Korea. He star ted his graduate study at Korea University in March of 2001, where he pursued Master of Engineering in c hemical engineering under the advisory of Professor Kwan-young Lee. In January of 2003, he joined as researcher at chemical mechanical planarization (CMP) team in LG Chem Research Park in Daejeon, South Korea, where he helped develop the ce ria-based slurry for STI CMP. He was accepted to the graduate program in the Department of Materials Science and Engineering (MSE) in August of 2007. He began pursuing his doctorate degree in MSE and joined Dr. Singhs research group in October 2007. His dissertation research focused on synthesis of ceramic powder fo r CMP slurry and its CMP performance. He graduated from University of Florida with a doctorate degree in MSE with electronic materials in December 2010. 220