<%BANNER%>

Implantation and Activation of Ultra-Shallow Boron in Germanium

Permanent Link: http://ufdc.ufl.edu/UFE0044910/00001

Material Information

Title: Implantation and Activation of Ultra-Shallow Boron in Germanium
Physical Description: 1 online resource (188 p.)
Language: english
Creator: Yates, Bradley R
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2012

Subjects

Subjects / Keywords: activation -- boron -- cluster -- dopant -- germanium -- hall -- implantation -- ion -- semiconductor
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: The physical scaling associated with integrated circuits is exhausting the properties of Si and requires that advanced materials be used for future device generations.  Ge is widely regarded as a possible replacement for Si due to its enhanced mobility and reduced contact resistance.  Due to the extensive use ofSixGe1-x  in current devices, the implementation of Ge into future devices could be considered a mere evolution from Si rather than a revolutionary change.However, the information regarding technologically relevant ultra-shallow dopant implants into Ge and their associated activation behavior is currently sparse and must be fully understood prior to implementation into future devices. The activation behavior of ultra-shallow B+ implants in Ge has been investigated using micro Hall effect and micro four point probe techniques.  It has been observed that the activation behavior of ultra-shallow B+ implants are anomalous in that the electrically active dopant fraction is independent of the implanted B+ fluence for both crystalline and pre-amorphized Ge. Ion beam analysis techniques have been employed which have confirmed that a small fraction of B is located substitutionally and the substitutional fraction does not increase appreciably with thermal processing for temperatures= 600°C.  Activation is observed to increase with increasing B+ energy which is attributed to the effect of the surface proximity and its associated effect on vacancy annihilation.  The activation behavior is further explained through the largely immobile B atoms which have a distinctly low probability of recombining with a vacant site due to the overwhelmingly large population of interstitials created during implantation as simulated by SRIM.  The excess interstitial population increases competition for B recombination on a vacant lattice site and thereby reduces B activation. B+ implantation at increased energy increases the number of vacancies created and reduces the effect of the surface on vacancy annihilation which explains the observed increase in activation. The observed activation behavior in this work is a strong departure from what has been observed previously for B in Si.  The results suggests that previous activation and dopant solubility models implemented for activation in Si do not apply for shallow B+ implantation in Ge.
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 Bradley R Yates.
Thesis: Thesis (Ph.D.)--University of Florida, 2012.
Local: Adviser: Jones, Kevin S.
Local: Co-adviser: Law, Mark E.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2012
System ID: UFE0044910:00001

Permanent Link: http://ufdc.ufl.edu/UFE0044910/00001

Material Information

Title: Implantation and Activation of Ultra-Shallow Boron in Germanium
Physical Description: 1 online resource (188 p.)
Language: english
Creator: Yates, Bradley R
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2012

Subjects

Subjects / Keywords: activation -- boron -- cluster -- dopant -- germanium -- hall -- implantation -- ion -- semiconductor
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: The physical scaling associated with integrated circuits is exhausting the properties of Si and requires that advanced materials be used for future device generations.  Ge is widely regarded as a possible replacement for Si due to its enhanced mobility and reduced contact resistance.  Due to the extensive use ofSixGe1-x  in current devices, the implementation of Ge into future devices could be considered a mere evolution from Si rather than a revolutionary change.However, the information regarding technologically relevant ultra-shallow dopant implants into Ge and their associated activation behavior is currently sparse and must be fully understood prior to implementation into future devices. The activation behavior of ultra-shallow B+ implants in Ge has been investigated using micro Hall effect and micro four point probe techniques.  It has been observed that the activation behavior of ultra-shallow B+ implants are anomalous in that the electrically active dopant fraction is independent of the implanted B+ fluence for both crystalline and pre-amorphized Ge. Ion beam analysis techniques have been employed which have confirmed that a small fraction of B is located substitutionally and the substitutional fraction does not increase appreciably with thermal processing for temperatures= 600°C.  Activation is observed to increase with increasing B+ energy which is attributed to the effect of the surface proximity and its associated effect on vacancy annihilation.  The activation behavior is further explained through the largely immobile B atoms which have a distinctly low probability of recombining with a vacant site due to the overwhelmingly large population of interstitials created during implantation as simulated by SRIM.  The excess interstitial population increases competition for B recombination on a vacant lattice site and thereby reduces B activation. B+ implantation at increased energy increases the number of vacancies created and reduces the effect of the surface on vacancy annihilation which explains the observed increase in activation. The observed activation behavior in this work is a strong departure from what has been observed previously for B in Si.  The results suggests that previous activation and dopant solubility models implemented for activation in Si do not apply for shallow B+ implantation in Ge.
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 Bradley R Yates.
Thesis: Thesis (Ph.D.)--University of Florida, 2012.
Local: Adviser: Jones, Kevin S.
Local: Co-adviser: Law, Mark E.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2012
System ID: UFE0044910:00001


This item has the following downloads:


Full Text

PAGE 1

1 IMPLANTATION AND ACTIVATION OF ULTRA SHALLOW B ORON IN G ERMANIUM By BRADLEY R YATES 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 2012

PAGE 2

2 2012 Bradley R. Yates

PAGE 3

3 To Mom

PAGE 4

4 ACKNOWLEDGMENTS The culmination of this project is an exciting time for me and many thanks need to be extended for all of the help that I have received alo ng the way. While I am earning the degree from the University of Florida, it seems like the majority of this work was completed at other institutions. I have travelled and sent samples around the world, but this work would not have been completed if it w as not for those closest giving me their support. To begin, I w ould like to thank my mother, family and friends for always being there for me during my academic career The support received was probably not outwardly recognized, but its effect was impo rtant nonetheless. I would also like to acknowledge all of the help that my advisor, Dr. Kevin Jones, has given me along the way. Foolishly, I had turned down his initial offer to join his group just to plead with him one year later for admittance. For this, I will always be grateful. I also appreciate all of the help given to me by Dr. Mark Law, Dr. Toshi Nishida, Dr. Gerald Bourne, Dr. Luisa Amelia Dempere and Dr. Rajiv Singh for sitting on my supervisory committee I also need to thank Dr. John Talbo tt for his support of my academic endeavors ; this will never be forgotten and I will be sure to pay it forward. I would also like to thank all of the members of the group that have helped me along the way. Everyone from Pat Whiting with his seemingly nev er ending supply of academic memes to Nick Rudawski with his imitations of th ose that shall go written with Nick in mind. All of my interactions here at the University of F lorida have been positive and have been instrumental in achieving this goal.

PAGE 5

5 I would also like to thank the countless collaborators that have helped me with this project. All of the folks at CAPRES have been undoubtedly a large part of this work and can not go unmentioned; Peter Nielsen, Dirch Petersen, Rong Lin, and Ole Hansen have all assisted with acquiring data as well as writing manuscripts. I would also like to acknowledge Lucia Romano of the University of Catania and Barney Doy le of Sandia Nationa l Laboratories for their assistance with ion beam analysis. Alex Kontos of Varian Semiconductor/Applied Materials, Rob Elliman at Australian National University, and Russell Gwilliam of the University of Surrey are also acknowledge d for their help with al l of the ion implants used for this project. Dr. Ignacio Martin Bragado and Jose Luis Gomez Sellis at IMDEA have provided simulations for this work and their efforts are greatly appreciated. The project would not have been possible without the funding p rovided by Intel. I also would like to acknow ledge the Major Analytical Instrumentation Center and the Nanoscale Research Facility at the University of Florida and the Advanced Materials Processing and Analysis Center at the University of Central Florida for the use of their facilities in completing this work.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 9 LIST OF FIGURES ................................ ................................ ................................ ........ 10 LIST OF ABBREVIATIONS ................................ ................................ ........................... 14 ABSTRACT ................................ ................................ ................................ ................... 15 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 17 1.1 Technological Motivation ................................ ................................ .................. 17 1.2 Objective and Statement of Thesis ................................ ................................ ... 18 1.3 Background Information ................................ ................................ .................... 18 1.3.1 Ion Implantation, Amorphization and Solid Phase Epitaxial Growth ........ 18 1.3.2 Activation of B in Ge ................................ ................................ ................ 23 1.3.3 Diffusion of B in Ge ................................ ................................ .................. 29 1.3.4 Activation and Diffusion of B in Si ................................ ............................ 30 1.3.5 Summary ................................ ................................ ................................ 32 2 EXPERIMENTAL TECHNIQUES ................................ ................................ ............ 39 2.1 Materials Processing ................................ ................................ ......................... 39 2.1.1 Ion Implantation ................................ ................................ ....................... 39 2.1.2 Dicing Saw ................................ ................................ .............................. 40 2.1.3 Thermal Proces sing ................................ ................................ ................. 40 2.2 Electrical Characterization ................................ ................................ ................ 43 2.2.1 Four Point Probe ................................ ................................ ..................... 43 2.2.2 Hall Effect ................................ ................................ ................................ 44 2.2.3 Micro Hall Effect & Micro Four Point Probe ................................ ............. 47 2.3 Structural Characterization ................................ ................................ ............... 49 2.3.1 Transmission Electron Microscopy ................................ .......................... 49 2.3.1.1 TEM Sample Preparation ................................ ................................ ..... 51 2.3.2 Secon dary Ion Mass Spectrometry ................................ .......................... 53 2.3.3 Rutherford Backscattering Spectrometry ................................ ................. 55 2.3.4 Nuclear Reaction Analysis ................................ ................................ ....... 56 2.3.5 Elastic Recoil Detection ................................ ................................ ........... 57 2.4 Summary ................................ ................................ ................................ .......... 58 3 ACTIVATION OF ULTRA SHALLOW BORON IMPLANTS IN GERMANIUM ........ 71

PAGE 7

7 3.1 Activation of B + Implants in Ge ................................ ................................ .......... 71 3.2 Experimental Details ................................ ................................ ......................... 73 3.3 Characterization of Residual Implanted Dose ................................ ................... 75 3.4 Hall Effect Characterization ................................ ................................ .............. 76 3.5 Nuclear React ion and Channeling Analyses ................................ ..................... 79 3.6 Summary ................................ ................................ ................................ .......... 81 4 EFFECT OF IMPLANT ENERGY ON B ACTIVATION ................................ ........... 89 4.1 Low Energy B + Implantation in Ge ................................ ................................ .... 89 4.2 Experimental Details ................................ ................................ ......................... 89 4.3 Effect of Increased B + Implant En ergy ................................ .............................. 90 4.4 Role of Implant Energy ................................ ................................ ..................... 93 4.5 Summary ................................ ................................ ................................ .......... 97 5 THERMAL STAB ILITY OF BORON ACTIVATION IN GERMANIUM ................... 106 5.1 Activation Stability of B in Ge ................................ ................................ .......... 106 5.2 Experimental Details ................................ ................................ ....................... 107 5.3 Isothermal Annealing ................................ ................................ ...................... 108 5.4 Activation Thermal Stability Between 400 600 C ................................ ............ 112 5 .5 High Temperature Anneals ................................ ................................ ............. 116 5.5.1 B Concentration Profile Following High T Annealing ............................. 117 5.5.2 Reduction of Active Carriers Fo llowing High T Annealing ..................... 119 5.5.3 Significance of the Lack of B Diffusion ................................ .................. 119 5.6 Theory for B Inactivity ................................ ................................ ..................... 121 5.7 Simulation of Activation Behavior ................................ ................................ .... 126 5.8 Summary ................................ ................................ ................................ ........ 128 6 IMPLANT RELATED DAMAGE IN GE ................................ ................................ 147 6.1 Introduction ................................ ................................ ................................ ..... 147 6.2 Experimental Details ................................ ................................ ....................... 149 6.3 Projec ted Range Damage ................................ ................................ ............... 150 6.4 End of Range Damage ................................ ................................ ................... 153 6.5 Discussion ................................ ................................ ................................ ...... 154 6.6 Summary ................................ ................................ ................................ ........ 155 7 CONCLUSIONS AND FUTURE WORK ................................ ............................... 164 APPENDIX : EFFECT OF IMPLANT CONDITIONS ON B ACTIVATION .................... 168 A.1 Variable Implantation Conditions ................................ ................................ .... 168 A.2 Experimental Methods ................................ ................................ .................... 16 8 A.3 Results and D iscussion ................................ ................................ .................. 169 A.4 Summary ................................ ................................ ................................ ........ 171 LIST OF REFERENCES ................................ ................................ ............................. 176

PAGE 8

8 BIOGRAPHICAL S KETCH ................................ ................................ .......................... 188

PAGE 9

9 LIST OF TABLES Table page 3 1 Channeling yields and corresponding substitutional B fraction s as measured by channeling analysis using nuclear reactions along three axes (<100>, <110>, and <111>) and Hall effect me 60s. ................................ ................................ ................................ ..................... 83 5 1 Inactive B determined from micro Hall effect and the corresponding displaced Ge obtained from RBS characterization for c Ge samples annealed for various times. ................................ ................................ ................................ ... 130

PAGE 10

10 LIST OF FIGURES Figure page 1 1 Schematic of the ion solid interactions during ion implantation. ......................... 34 1 2 Evidence of the channe ling tail in c Ge. ................................ ............................ 35 1 3 Schematic displaying th e amorphization process in Ge. ................................ .... 36 1 4 Evidence of transient B diffusion in G e ................................ ............................... 37 1 5 Typical B concentration profile following trans ient enhanced diffusion in Si. ...... 38 2 1 Schematic of four point probe configuration used to determine sheet resistance, R S ................................ ................................ ................................ .... 60 2 2 Sample geometry for a four point probe measurement ................................ ...... 60 2 3 Illustration of the electrons passing through a solid in the presence of a magnetic field. ................................ ................................ ................................ .... 61 2 4 Schematic of the van der Pauw contact scheme ................................ ................ 61 2 5 Illustration of the contact scheme typically used for Hall effect measurements. ................................ ................................ ................................ 62 2 6 Leakage current into the n type substrate ca n dominate a measurement if the metallurgical junction is not sufficient. ................................ ................................ 62 2 7 Decreased horizontal length scale of measurement probes reduces the junction leakage. ................................ ................................ ................................ 63 2 8 Current paths near contacts for M4PP measurements under the influence of a magnetic field. ................................ ................................ ................................ 64 2 9 The presence of a cleaved boundary gives rise to a resistance difference, R ...................... 65 2 10 Schematic diagram of a TEM. ................................ ................................ ............ 66 2 11 Representation of typical damage layer produced by FIB milling ....................... 67 2 12 Typical data pr oduced from c RBS experiment. ................................ ................. 68 2 13 Illust ration of a nuclear reaction between a proton and a B atom as it produces an particle ................................ ................................ ........................ 69

PAGE 11

11 2 14 Representation of particle counts produced by analyzing B a toms in different confirmations ................................ ................................ ........................ 69 2 15 Experimental setup of an ERD experiment ................................ ......................... 70 3 1 The effect of B + backscattering for implantation into Ge and Si ......................... 84 3 2 Percent of B ions backscattered as a function of implant energy ...................... 85 3 3 Electrical characteristics as a function of B + fluence implanted at 2 keV into 1h. ................................ ................................ ................................ ...................... 86 3 4 HR XTEM micrographs of a crystalline Ge sample B + implanted at 2 keV to 5.010 15 cm 2 as implanted in c Ge ................................ ................................ .... 87 3 5 B concentration profiles of a pre amorphized Ge sample B + implanted at 2 keV to doses of 5.010 13 or 5.010 15 cm 2 as implanted (dashed line) and after an ............ 88 4 1 SRIM simulations depicting the shift in the B concentration profile produced with increasing implant energy. 78 ................................ ................................ ........ 99 4 2 Measured sheet resistance of samples B + implanted at 2, 4, and 6 keV to fluences ranging from 5.010 13 to 5.010 15 cm .... 100 4 3 Measured sheet number of samples B + implanted at 2, 4, and 6 keV to fluences ranging from 5.010 13 to 5.010 15 cm .... 101 4 4 Measured drift mobility of samples B + implanted at 2, 4, and 6 keV to fluences ranging from 5.010 13 to 5.010 15 cm ... 103 4 5 Measured percentage of electrically active B at variable implant energies. A 35 keV implant corresponds to an R P of 90 nm ................................ ................ 103 4 6 Measured percentage of electrically active B in Si at variable implant energies. ................................ ................................ ................................ ........... 104 4 7 Vacancies created per incoming B ion as a function of implant energy as simulated by SRIM. 78 ................................ ................................ ........................ 105 5 1 Sheet resistance data obtained for samples implanted at 2 keV with a fluence of 1 .010 15 cm and subsequently annealed at 40 ......... 131 5 2 Sheet resistance line scan acquired across c Ge sample annealed at 400 C for 30s. ................................ ................................ ................................ ........... 132 5 3 Sheet number data obtained for samples implanted at 2 keV with a fluence of 1 .010 15 cm ............. 133

PAGE 12

12 5 4 Dri ft mobility for a c Ge samples implanted at 2 keV with a fluence of 1 .010 15 cm ............ 134 5 5 RBS channeling spectra for c Ge samples implante d at 2 keV to a fluence of 1 .010 15 cm ............. 135 5 6 Sheet resistance data obtained for samples implanted at 2 keV with fluences ranging from 5. 010 13 to 5.010 15 cm ................................ ................................ ............................. 136 5 7 Change in sheet resistance for 2, 4, and 6 keV B + implants to fluences ranging from 5.010 13 to 5.010 15 cm between anneali ................................ ................................ ................................ ... 137 5 8 Electrical activation characteristics as a function of anneal temperature for samples B + implanted at 2, 4, and 6 keV to a fluence of 5.0 10 15 cm into c Ge and PA Ge. ................................ ................................ ................................ 138 5 9 HR XTEM images of samples B + implanted at 2 keV to a fluence of 5.010 15 cm 2 into c and D) into PA ................................ ......... 139 5 10 Sheet resistance for samples B + implanted at 6 keV to a fluence of 5.010 15 cm 2 into c Ge and PA Ge. ................................ ................................ .............. 140 5 11 Sheet number for samples implanted at 6 keV to a fluence of 5.010 15 cm 2 ... 140 5 12 B concentration profiles for samples B + implant ed at 6 keV to a fluence of 5.010 15 cm 2 into c Ge as ................................ ................................ ................... 141 5 13 B concentration profiles for samples B + implanted at 6 keV to a fluence of 5.010 15 cm 2 into PA Ge as im ................................ ................................ ................... 141 5 14 HR XTEM micrograph of a pit formed in Ge following thermal processing. The presence of the native oxide surrounding th e pit is evident. ...................... 142 5 15 SIMS characterization of a sample B + implanted at 35 keV to a fluence of 2.010 15 cm 2 following various thermal treatments. The horizontal shift in the chemical pro ........ 143 5 16 Simulation displaying the cumulative vacancy population as a function of radial distance from an implanted B ion at 2 keV. ................................ ............. 144 5 17 Kinetic Monte Carlo simulations displaying the substitutional and interstitial boron concentrations following annealing at 400 C for 60s for a 2 keV B + implant ................................ ................................ ................................ .............. 145

PAGE 13

13 5 18 Kinetic Monte Carlo simulations displaying the substitutional and interstitial boron concentrations following B + implantation at 2 keV to a fluence of 5. 010 13 cm 2 C ................................ .............................. 146 6 1 Bright field XTEM image taken with g 220 (3g) diffraction condition of projected range damage created by a 2 keV B + implant to a fluence of 1. 010 15 cm 2 following annealing at 500 .................... 157 6 2 HR XTEM image of projected range damage created by a 2 keV B + implant to a fluence of 1.010 15 cm 2 followi ng annealing at 500 ................. 158 6 3 Defect orientations of projected range damage in Ge. ................................ ..... 159 6 4 XTEM images of projected range damage created by a 2 keV B + implant to a fluence of 1.010 15 cm 2 following annealing at 500 ................................ ..... 160 6 5 Bright field PT EM images taken with g 220 (3g) diffraction condition of projected range damage created by a 2 keV B + implant to a fluence of 1.010 15 cm 2 following annealing at 500 ................................ ................. 161 6 6 XTEM images taken with g 220 (3g) diffraction condition of EOR damage created by a 1 MeV Ge + implant to a fluence of 2.010 15 cm 2 following annealing at 330 ................................ ............................... 162 6 7 EOR dama ge created by a 1 MeV Ge + implant to a fluence of 2.010 15 cm 2 as a function of annealing time at 330 ................................ ......................... 163 A 1 Measured sheet resistanc e ( R s ) after annealing at 400 C for 60 s as a function of beam curren t ................................ ................................ ................... 172 A 2 Measured sheet number ( n s ) and percent elec trical activation as a function of beam current implanted at 2 keV to a fluence of 5.010 14 cm into crystalline Ge after annealing at 400 C for 60s. ................................ ............................... 173 A 3 Measured drift mobility ( D ) as a function of beam current implanted at 2 keV to a fluence of 5.010 14 cm into crystalline Ge after annealing at 400 C for 60s. ................................ ................................ ................................ ................... 174 A 4 HR XTEM micrograph of an as implanted crystalline Ge sa mple B + implanted at 2 keV to 5.010 14 cm at a beam current of 6.4 mA ................................ .... 175

PAGE 14

14 LIST OF ABBREVIATI ONS 4PP Four point probe Ge Amo r phous Ge c Ge Crystalline Ge CMOS Complementary metal oxide semiconductor ERD Elastic recoil detection FIB Focused ion b eam HR XTEM High resolution transmission electron m icroscopy IBA Ion beam analysis M4PP Micro four poi nt p robe MHE Micro Hall effect NRA Nuclear reaction a nalysis n S Sheet n umber PA Ge Preamorphized Ge q Elementary charge, 1.602 10 19 C RBS Rutherford b ackscattering R P Projected range R S Sheet r esistance TEM Transmission electron m icroscopy SIMS Secondar y ion m ass s pectrometry SPEG Solid phase epitaxial growth SRIM Stopping range of ions in matter D Drift m obility V H Hall voltage

PAGE 15

15 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy IMPLANTATION AND ACTIVATION OF ULTR A SHALLOW BORON IN GERMANIUM By B radley R. Yates December 2012 Chair: Kevin S. Jones Cochair: Mark E. Law Major: Materials Science and Engineering The physical scaling associated with integrated circuits is exhausting the properties of Si and requires that advanced materials be used for future device generations. Ge is widely regarded as a possible replacement for Si due to its enhanced mobility and reduced contact resistance. Due to the extensive use of Si x Ge 1 x in current devices, the implementatio n of Ge into future devices could be considered a mere evolution from Si rather than a revolutionary change. However, t he information regarding technologicall y relevant ultra shallow dopant implants into Ge and their associated activation behavior is curre ntly sparse and must be fully understood prior to implementation into future devices. The activation behavior of ultra shallow B + implants in Ge has been investigated using micro Hall effect and micr o four point probe techniques. It has been observed that the activation behavior of ultra shallow B + implants are anomalous in that the electrically active dopant fraction is independent of the implanted B + fluence for both crystalline and pre amorphized Ge Ion beam analysis techniques have been employed whic h have confirmed that a small fraction of B is located substitutionally and the substitutional fraction does not increase appreciably with thermal processing for

PAGE 16

16 temperatures 600 C Activation is observed to increase with increasing B + energy which is attributed to the effect of the surface proximity and its associated effect on vacancy annihilation. The activation behavior is further explained through the largely immobile B atoms which have a distinctly low probability of recombining with a vacant site due to the overwhelmingly large population of interstitials created during impla ntation as simulated by SRIM. The excess interstitial population increases competition for B recombination on a vacant lattice site and thereby reduces B activation. B + implantation at increased energy increases the number of vacancies created and reduces the effect of the surface on vacancy annihilation which explains the observed increase in a ctivation The observed activation behavior in this work is a strong departure from what has been observed previously for B in Si. The results suggests that previous activation and dopant solubility models implemented for activation in Si do not apply for shallow B + implantation in Ge.

PAGE 17

17 CHAPTER 1 INTRODUCTION 1.1 Technological Motivation The semiconductor industry has evolved tremendously over the preceding half century. One of the industry pioneers, Gordon Moore, originally claimed that the number of components in an integrated circuit will double every year, but has since been taken to mean a doubling every 1.5 1 In an faced with physically scaling down the size of components built on integrated circuits. However, silicon, whose natural abundance and stable native oxide has propelled it as the traditional workhorse of the semiconductor industry, is reaching its physical scaling limits. To continue the scaling, the industry has recently chosen to shift away from planar processing towards three dimensional device st ructures. However, the need for advanced semiconducting materials for future device generations still looms on the horizon. Ge rmanium is regarded as one of the possible replacement materials for Si in future device generations due to its enhanced electron and hole mobility in comparison to Si. 2 Ge was studied extensively several decades ago and the first transistor was actually constructed from Ge. The recent implementation of high gate dielectric has circumvented the issues regarding the unstable native oxide of Ge. In order for Ge to be realized as a replacement for Si in future generation devices, further fundamental understanding of the ac tivation behavior of B must be realiz ed. At present, there is a significant dearth of information available regarding technologically relevant ultra shallow B + implants in Ge The few reports available only

PAGE 18

18 mention a single implant or anneal condition with little analysis of the general act ivat ion behavior 1. 2 Objective and Statement of Thesis T here is significant technological motivation to understand the electrical activation behavior of ultra shallow B + implants in Ge. The research discussed in this document seeks to understand the el ectrical characteristics of implanted B in Ge as well as the effects the ion implantation process has on the substrate. T he following aspects will be given significant attention. The a ctivation behavior of ultra shallow ion implanted behavior will be inv estigated using micro Hall effect and micro four point probe techniques. Ion beam analysis will be used to provide further evidence of activation results. The thermal stability of B activation in Ge will be investigated for a wide range of annealing time s and temperatures. The formation and evolution of implant related damage and its effect on activation will also be investigated. 1.3 Background Information 1.3.1 Ion Implantation, Amorphizati on and Solid Phase Epitaxial G rowth Ion implantation involves the use of an accelerated beam of charged particles directed towards a solid in order to effect change on the surface properties of the material. It has been used extensively by the semiconductor industry for several decades in order to alter the electric al properties of the material by altering the chemical composition of the near surface regions. It is the preferred method for introducing high concentrations of dopant atoms in precise locations. A substantial benefit of ion implantation is its ability to self align the placement of the dopant. For example, implantation over a masked substrate will not allow for incorporation of dopant below the masked region while implanting dopant where the surface is bare. Its self aligning

PAGE 19

19 capabilities enable the a ccurate and highly repeatable control of dopant profiles which have proven invaluable for the processing of semiconductor devices. 3 6 During ion implantation, a gaseous source of the intended dopant is ionized accelerated and mass separated based on its mass to charge ratio. The ionization process will create a series of ion s with the same atomic mass, but with varying charge based on the number of electrons removed. The desired ion charge is typically chosen based on the fraction of the total beam to create the highest number of ions per second or in simpler terms, beam cur rent. Once the desired charge is separated from the beam, it is steered towards the substrate where it eventually comes to rest or is implanted in the target. The dopant of mass, m is ionized to obtain a charge, q and is accelerated by a potential diff erence, V The accelerating voltage can vary from a few eV to several hundred MeV, but is typically near the 1 100 keV range for typical semiconductor applications. The accelerating voltage imparts kinetic energy and dictates the ion velocity using the c lassical mechanics equation: (1 1) Once the accelerated ion impacts the surface, a series of ion solid interactions is set in motion. As the ion reaches the target surface, it is likely that the ion will displace target atoms through sputteri ng or the accelerated ion may be rejected from the solid through backscattering. For sputtering, the fraction of ions sputtered increases increasing ion mass or namely the ratio of atomic mass of the ion and the target; the opposite is true for ion backsc attering. Once the ion enters the solid, the ion slows through a combination of electronic (elastic) and nuclear (inelastic) scattering

PAGE 20

20 processes. Near the surface, the ion slows primarily through electronic interactions between the charge associated wit h the ion and the electronic clouds of target atoms. As the ion travels further in to the solid, it is continually slowed by electronic means and begins to lose a larger fraction of its kinetic energy through nuclear collisions. Electronic stopping is dom inant for ion travel near the surface where the ion is generally slowed while nuclear stopping begins to dominate as the ion travels deeper into the solid as the ions are travelling at a slower velocity. The probability of nuclear scattering increases as the energy and velocity of the incoming ion is decreased. Due to this, nuclear stopping dominates near the end of the ion range while electronic stopping is more prominent near the surface; however, both mechanisms are active throughout the entire ion ran ge. The energy lost to the ion solid interactions are described by: (1 2) where dE/dx is the energy lost through the depth of the sample, S n is the nuclear stopping power, S e is the electronic stopping power, and N is the number of atoms in the solid per unit volume. Both elastic and inelastic scattering mechanisms are important for loss of ion energy, but nuclear collisions are the primary source of ion stopping and radiation damage to the lattice. The displaced target atoms, or knock on atoms, have kinetic energy which is subseq uently lost through nuclear collisions with other host atoms. This pyramid process is known as a cascade and creates a number of vacancy and interstitial (Frenkel) pairs in the target material. Once all kinetic energy is lost by electronic and nuclear me ans, the ion becomes implanted in the target A schematic representation of the ion solid interaction is shown in Figure 1 1

PAGE 21

21 The implanted ion comes to rest in the solid at a controllable depth for a given set of implantation conditions. The total dista nce travelled through the solid is called the ion range, R while the perpendicular distance travelled from the surface is termed the projected range, R P These definitions hold true for single ions; however, an implanted fluence for semiconductor applica tions to a rough approximation, is typically over 5 .010 1 2 ions. The implantation process is largely statistical in which two identical ions implanted at identical conditions into the same substrate can have radically different paths through the target. For given conditions, the average projected range of an implanted dose creates a Gaussian distribution of dopants with respect to depth in the sample. The peak of the profile is the average R P for a given set of implant conditions. A side effect of th e ion implantation process is the introduction of implant related damage to the material. The point defects created during the implantation process have been shown to have deleterious effects on dopant activation, dopant diffusion, and device leakage. 7 15 For a low fluence, isolated point defects may form in the crys tal lattice Upon annealing these point defects may coalesce to form extended defects which are referred to as sub amorphization defects or the occurrence of defects when ion implantation damage is insufficient to create an amorphous layer. 16 As the ion fluence is increased, a proportionally larger amount of Frenkel pairs are created and damaged regions may coalesce. The damage accumulation is assumed to be l inear with implanted fluence until a certain value is obtained. A critical threshold of damage imparted to the crystalline lattice exists which upon reaching the crystal relaxes to an amorphous state which renders the lattice devoid of any long range orde r. The transition to an amorphous solid is deemed a first order phase transition. 17

PAGE 22

22 The critical fluence required for amorphization is deemed the amorphization threshold and varies for d ifferent ion species. The amorphization process is directly dependent on the creation of point defects and therefore dependent on ion mass and energy, ion beam current, implant angle, substrate temperature, etc., where heavier ions create denser damage cas cades and the amorphization occ urs at a reduced fluence. 18,19 T he amorphization threshold can be determined by alternate means, namely through the accumulation of vacancies created during the implantation process called the threshold damage density. In doing so, the amorphization threshold can be desc ribed in terms of damage created to allow for better prediction of the amorphous layer depth independent of the implanted species. 20 The f ormation of an amorphous layer has been shown to be advantageous for semiconductor processing. The amorphization of the semiconductor prior to dopant implantation or preamorphization, has several advantages over dopant implantation alone following a subs equent annealing step. Following annealing, preamorphization enables increased dopant activation, a reduction of imp lant damage, and shallower dopant concentration profiles. Self implants allow for amorphization with a reduced implant fluence which decre ases implantation time and are typically used prior to the dopant implant. The amorphization threshold for Ge has been well studied and occurs at a fluence of approximately 5 .010 1 3 cm 2 for a 120 keV Ge + implant at room temperature 20,21 This amorphization threshold corresponds to a thre shold damage density of approximately 3 .010 20 keV/cm 3 22 In addition, amorphization allows for shallower dopant profiles in comparison to implants into crystalline substrates. When crystalline substrates are irradiated, ions

PAGE 23

23 have a tendency to channel down open crystallographic axes. These pathways allow for deeper penetration into the solid then the same implant into a randomly ordered or amorphous solid. In doing so, a channeling tail is created which effectively increases the dopant depth into the solid. The channeling tail is evident in a plot of a B implant into crystalline and preamorphized Ge as shown in Figure 1 2 During annealing, the amorphous layer regrows from the crystalline seed below through a process called solid phase epitaxial growth (SPEG) Figure 1 3 displa ys a schematic of the introduction of an amorphous Ge ( Ge) layer and the subsequent regrowth of that layer from the crystalline (001) Ge below. During this process, an increased fraction of dopant atoms are incorporated into substitutional sites renderi ng them electrically active. The SPEG process allows for approximately an order of magnitude increase in electrical activation in Ge 23,24 During regrowth the SPEG process allows for a reduction of point defects upon regrowth due to the direct rearrangement of bon ds as the crystalline interface advances through the amorphous layer ; however, the region just beyond the amorphous crystalline interface is highly damaged and is often the source of point defects responsible for extended defect formation known End of Ran ge (EOR) damage. The EOR defects are denoted as being defect s that occur when implant ation damage is sufficient to produce an amorphous layer 16 1.3. 2 Act iva tion of B in Ge Germanium is a rather resistive material with intrinsic Ge having a resistivity on the order of 50 cm. 25 However, the introduction of a minute concentration of dopant atoms can alter the resistivity of the material drastically by sever al orders of magnitude. Herein lies the beauty of semiconducting materials. However, the mere presence of

PAGE 24

24 dopant atoms in the lattice is sufficient to effect change to the electrical properties of the material. For a dopant to be electrically active in Ge, it mus t be located on a substitutional lattice site. The process of finding a lattice site is typically completed by pairing with a vacant lattice site or replacing an interstitial from its site through a kick out mechanism. Once situated on a substitutional si te, the B atom has acts as an acceptor in which it accepts an electron from the Ge crystal to complete the fourth of its covalent bonds to the lattice and binds a hole which can be ionized to aid in conduction of the lattice. A semiconductor that is doped with an impurity that acts an acceptor is considered to be p type. By increasing the number of B atoms situated on substitutional sites, the conductivity of the crystal is affected positively. The conductivity, of a material is defined by: (1 3) where q is the fundamental charge, p and n are the concentrations of holes and electrons and h and e are the hole and electron mobilities respectively. For B implantation in Ge, the acceptor concentration far outweighs a ny intrinsic electron contribution. The conductivity then becomes: (1 4) Upon first inspection of this equation, it appears that by increasing the concentration of active dopants, the conductivity of the material will decrease. To a first a pproximation, this is true; however, there is a finite limit on the number of B atoms that can be incorporated in solution with Ge. 26,27 The maximum chemical solubility of B in Ge is reported to be approximately 5.5 10 1 8 cm 3 as determined from diffusion experiments. 28

PAGE 25

25 30 From these experiments, it has been shown that B atoms in Ge are relatively immobile for all practical thermal treatments. For example, Uppal et al. implanted B + at 20 keV to a fluence of 6.0 10 1 4 cm 2 and annealed for 3 h in the range of 675 800 C following which no significant diffusion was observed. 30 Despite the low solubility of B in Ge, the ion implantation process often allows for an active concentration of carriers above the reported solubility l imit 23,31 In effect, the non equilibrium ion implantation process introduces a metastable increase in the solubility of B atoms. The activation of B in Ge following ion imp lantation was studied several decades ago and has just recently regained interest due to Si reaching its physical scaling limits. For low fluence implants implanted to a high energy, it has been shown that a large fraction of implanted B ions are substitu tional and active immediately following implantation and prior to any annealing step. 32,33 Recently, the investigations into the activation of high fluence B + implants in Ge ha ve been underway. 23,31 Specifically, several authors have shown that pream orphization allows for an increased fraction of the implanted dopant as compared to dopant implants alone. I n some cases electrical solubility was reported to concentrations as high as 5.5 10 20 cm 3 and a dditional authors have reported similar values. 24,31,34,35 It was observed that the preamorphization energy and subsequent depth of the resulting amorphous l ayer has an effect on activation where a shallow amorphous layer allowed for dynamic annealing during B + implantation. 35 Dynamic annealing occurs when B ions traverse the amorphous crystalline interface and allows for regrowth of the amorphous layer during dopant implantation. To prevent dynamic

PAGE 26

26 annealin g, the preamorphization implant must be sufficiently deep to not allow for a significant fraction of B ions to interact with the amorphous crystalline interface. However, the maximum active concentration allowed by B + implantation in crystalline Ge witho ut a preamorphization step is not as well documented. The electrical solubility for B + implants into c Ge vary by approximately an order of magnitude. Satta et al. 34 has determined the maximum activation to be 1 2 10 19 cm 3 while other authors have reported values in excess of 1 0 10 20 cm 3 24,36 It has been speculated that that electrical solubility of B + implants in Ge is directly related to the damage imparted into the crystal lattice during the implantation process. 23 For 35 keV B + implants, it has been shown that increasing the fluence to 2 8 10 1 6 cm 2 yields similar activation values as produced by usin g a preamorphizing implant. Conversely, implanting B + at liquid nitrogen temperature reduces the dynamic annealing and subsequently increases the damage production which enables similar activation values as preamorphized Ge for reduced fluences. 23 It has been elucidated that sub microscopic amorphous pockets form with increasing B + fluence which increase the activatio n upon annealing 23,24,37 The wide range of reported electrical solubilities of B + impl ants into c Ge raises questions on whether electrical solubility arguments used for activation in Si apply as the literature does not suggest a single solubility value for Ge. There have been reports of the formation of a B Ge cluster which renders impla nted B atoms inactive. 34,37 39 These reports were observed in crystalline Ge only and not when a preamorphization implant is used. The inactive cluster has been sho wn to affect approximately 50% of the implanted fluence. 37 The inactive complex was studied by Impellizzeri et al. with Rutherford backscattering spectrometry to determine

PAGE 27

27 the ratio of the number of displaced Ge atoms to the inactive B atoms. From this investigation, it was determined that the B Ge cluster has the stoichiometry of 8: 1 (Ge:B) 37 Bisognin et al. investigated the same sample set with high resoluti on X ray diffraction and determined that each inactive B atom has a positive volume expansion of 14.8 1.7 3 which they conclude is due to the formation of a B Ge cluster with a stoichiometry of 8:1 39 A few studies have investigated the the rmal stability of B activation in Ge and have found that the activation shows little change with additional annealing. 23,34,40 Bruno et al. investigated the change in sheet resistance for samples implanted at 35 keV to various fluences A slight decrease in sheet resistance was reported fo r c Ge while a slight increase was observed for PA Ge samples. 23 Although fewer data points were presented, similar behavior was reported by other authors. 34,40 For all reports regarding the thermal stability of B implants in Ge, no anomalous activation behavior was discussed and activation was reported as being remarkably stable across the investi gated annealing range. Electrical data for samples annealed at high temperature ( >600 C) is not available in great quantities which is likely due to the surface desorption of the native Ge oxide. 41 Satta et al studied the sheet resistance of both P and B implants in Ge following f lash lamp annealing. Results showed little change in activation with flash lamp annealing for 20 ms at high temperature as opposed to an anneal at 500 C for 1h. 42 The lack of sheet resistance variation observed could be attributed to the short time spent at high temperature.

PAGE 28

28 It has been reported that the activation of high energy B + implants in Ge exhibits full activation if implanted to a concentration below the electrical so lubility and subjected 31 If th e concentration of implanted B exceeds the electrical solubility, the dopants are rendered electrically inactive. 31 Experiments investigating the thermal stability of B activation have shown that the activation is remarkably stable showing very little change with annealing for temperatures up to 23 However, there is a significant lack of information regarding th e activation of ultra shallow B + implants in Ge. Satta et al. reported values for sheet resistance for 6 keV B implants in Ge. From these values, the electrical solubility was determined to be 1 2 10 1 9 / cm 3 and 2 4 10 20 / cm 3 for crystalline and preamorph ized Ge, respectively. 34 However, these values were obtained by measuring sheet resistance and not directly measuring the active concentration 43,44 Other reports encompass only a few data points and follow a similar procedure. 40,45 To accurately characterize the activation behavior, the carrier density and drift mobility should be measured directly using the Hall effect. However, prior to this work, only two report s have investigated the activation following B implantation in Ge using the Hall effect. Hellings et al. implanted B + to a fluence of 8.0 10 1 4 / cm 2 at 2.4 keV into preamorphized Ge and determined that nearly 46 However, this was a systemat ic study of ultra shallow junction formation and no other B activation values were provided. Bennett et al. used a single 500 eV B + implant to a fluence of

PAGE 29

29 1.0 10 1 5 / cm 2 into PA Ge and determined the maximum solubility to be 2 0 10 20 / cm 3 using a differe ntial Hall effect technique. 47 It is clear that further information regarding the activation of ultra shallow B + implants in Ge is spar s e at this present time. The available literature on this subject is sparse which creates discrepancies when c omparing results from different experiments. The literature presently available does not allow for a comprehensive understanding of B activation in Ge. 1.3.3 Diffusion of B in Ge It has been established that B diffuses through an interstitial mediated m echanism in Ge. The high formation energy of an interstitial in Ge limits the ability for B to diffuse. It is well established that B does not appreciably diffuse under typical processing conditions. 28 30,34,35 However, recent reports have given evidence that excess interstitial populations may spur B diffusion to lengths greater than those observed under equilibrium conditions. It has been repor ted by Napolitani et al that a transient diffusion component exists for B following implantation. Figure 1 4 displays the transient B diffusion behavior observed. In this work, a Ge + implant was used to create a band of damage in close proximity to a B doped layer. Following annealing, it was evidenced that the B doped layer exhibited a diffusion behavior which slowed with increasing annealing temperature. The decreased diffusion was attributed to the removal of implant damage with increasing annealing temperature. 48 This behavior is similar to, but with a much smaller magnitude of the well researched transient enhanced diffusion beh avior of B in Si which will be discussed in the following section.

PAGE 30

30 In addition, there have been other reports of abnormal diffusion of B in Ge; however, these were observed under extreme conditions such as during proton irradiation 49 52 diffusion through thick amorphous Ge layers 53 o r following oxygen precipitation after implantation 54 While these reports provide insight to the interstitial mediated mechanism of B diffusion in Ge, they are not directly relevant to conventional integrated circuit processing. Fortunately, under typical processing conditions, it is well agreed that diffusion is not a concern for B + implants in Ge. 28,34,35 1.3. 4 Activation and Diffusion of B in Si In silicon, the most common p type dopant is boron due t o its high dopant solubility 55 As such, the activation and diffusion properties of ion implanted B in Si have been stu died extensively over the last several decades. 56 Similar to what has been reported previously for B + implants in Ge, activation in Si has been approached by invoking an electrical solubility limit above which dopants are inactive. However, t he behavior of B in Si is dominated by the dopant interact ion with point defects present in the crystal through the ion implantation process. Following implantation, damage to the crystalline lattice can vary from excess point defects to the complete amorphization of the surface region depending on the implant conditions. Upon annealing, these excess point defects strongly affect the activation and diffusion of B in Si. In the case of non amorphizing implants, excess interstitials coalesce to form defects near the projected range. In the case of amorphizing i mplants, the region just beyond the amorphous crystalline interface is heavily defective as it should be assumed to be just below the damage level required to transition to the amorphous state which spurs the formation of extended defects

PAGE 31

31 The activation o f B in Si is largely influenced by interaction s with excess interstitials present in the lattice. 57 Upon defect dissolution, a wave of excess interstitials migrates through the crystal where they interact with B atoms to form electrically inactive B interstitial complexes commonly referred to as BICs. 58 60 The formation of BICs can restri ct the activation of B in Si to a level well below the solubility. 58,61,62 the sheet resistance of the sample which upon annealing, increases before decreas ing towards a reduced value. 13,63 The incre ase in sheet resistance is attributed to cluster formation while the reduction is due to cluster dissolution and an increase in B solubility with increasing temperature. Similar to the activation behavior of B in Si, diffusion is also influenced by dopan t defect interactions. It has been widely observed that during annealing, implanted B exhibits diffusion is typically much faster than what is predicted by equilibrium diffusion models. 64 66 It has also been observed that this diffusivity enhancement decreases with annea ling time and hence is called transient enhanced diffusion or TED. 7,8,57,64,65,67 Figure 1 5 displays the enhanced diffusion of a B doped layer following an Si implant at 40 keV to a fluence of 1.0 10 1 5 / cm 2 adapted from Stolk et al 68 This amorphizing implant created a band of highly defective band of Si that is rich in excess interstitials. The observed observed diffusion is typical for TED in Si and is much more significant than that observed in Ge. The enhanced diffusion has been shown to be due to interstitials released from extended defect dissolution kicking out B and allowing it to diffuse through the lattice. TED has also been observed in the absence of extended defects. 69 The transient

PAGE 32

32 nature of the enhanced diffusion is created by the reduction of excess interstitials available duri ng the annealing pro cess. 70 Several techniques have been adapted to reduce the excess interstitial concentration in Si as well as its ef fects on activation and diffusion. The use of co implantation with C or F to sequester interstitials through a so called point defect engineering mechanism has been successful in reducing TED. 71 73 In addition, vacancy engineering involves a high energy (a few MeV) implant to introduce an ex cess concentration of vacancies near the surface with excess interstitials driven further into the bulk due to forward momentum transfer. The excess vacancies invoke recombination with excess interstitials and have been shown to reduce TED in Si. 74 77 1.3. 5 Summary Ion implantation is the preferred method for dopant incorporation into semiconductor devices due to its ability to efficiently locate well defined dopant quantities has earned. However, ion implantation is a non equilibrium process that introduces damage to the lattice. The damage to the crystalline lattice can vary from excess point defects to the complete amorphization of the surface region depending on the implant conditions The activation and diffusion beha vior of B in Si and Ge have been reviewed. For B in Ge, It has been established that B does not diffuse any appreciable amount in Ge under normal processing conditions. Complete activation has been observed with the electrical solubility limit as high as 5.5 10 20 / cm 3 follow ing a preamorphization implant. An inactive B Ge cluster with the stoichiometry 1:8 was reported to occur for implants into crystalline Ge in which approximately half of the dose is rendered inactive. B

PAGE 33

33 activation has been reported to be remarkably stable with no significant change during annealing. In contrast, B in Si is susceptible to anomalous transient diffusion labeled TED which is characterized by diffusion far in excess of equilibrium values that decreases with time. TED is d ue to the excess interstitial population produced during extended defect dissolution diffusing and subsequently kicking out B from their lattice site allowing them to diffuse through the lattice. TED decreases as the extended defects evolve and the excess interstitial population decreases. Similarly, B activation in Si is controlled by the release of excess interstitials from extended defects forming interacting with B atoms to form inactive BICs. It has been shown that BIC formation and dissolution is a thermal process and that activation of 100% of th e B concentration can be achieved given proper annealing conditions up to a certain temperature dependent dopant solubility 55 It should be noted that both the activation and d iffusion behavior of B in Si as compared to Ge are widely different. The activation behavio r of B in PA Ge is re ported to fully activate following a n anneal at 360 C for 1h and maintain its activation with increased annealing. In contrast, the activation and diffusion behavior of B in Si is dependent on dopant defect interactions and as such, varies as a function of annealing and subsequent defect environment.

PAGE 34

34 Figure 1 1 Schematic of the i on solid inter actions during ion implantation

PAGE 35

35 Figure 1 2 Evidence of the channeling tail in c Ge. Concentration profiles resulting from a B + implant at 6 keV to a fluence of 5 .010 15 cm 2 into c Ge and PA Ge.

PAGE 36

36 Figure 1 3 Schematic displaying the amorphization process in Ge. Crystalline Ge (c Ge) is self implanted at 120 keV to a fluence of 2 .010 1 4 cm 2 to amorphize surface layer. Sample is annealed and amorphous layer is regrown.

PAGE 37

37 Figure 1 4 E vidence of transient B diffusion in Ge. Reprinted with permission from E. Napolitani, G. Bisognin, E. Bruno, M. Mastromatteo, G.G. Scapellato, S. Boninelli, D. De Salvador, S. Mirabella, C. Spinella, A. Carnera, and F. Priolo, Appl. Phys. Lett. 96 201906 Copyright 2010. American Institute of Physics.

PAGE 38

38 Figure 1 5 Typical B concentration profile following transient enhanced diffusion in Si. Reprinted from Stolk et al Implantation and transient boron diffusion: the role of the self interstitial Nucl. Instrum. Methods Phys. Res. B 96 187 195. Copyright ( 1995 ), with permission from Elsevier

PAGE 39

39 CHAPTER 2 EX PERIMENTAL TECHNIQUES This chapter serves to outline the various methods used to process and characterize the samples used for the experi ments in this work. It is not intended to be a thorough explanation for the layman, but rather a brief introduction and explanation of the experimental details to assist the reader in understanding the steps taken in this work. 2 .1 Materials P rocessing 2 1.1 Ion Implantation In this work, ion implantation was used for the introduction of all dopant atoms into the Ge lattice. It was also used to induce the formation of an amorphous layer prior to dopant implantation by using a self implant with a fluence a bove the amorphization threshold of Ge; commonly referred to as a preamorphization implant. All implants were completed using commercial ion implantation tools under high beam current conditions. B oron implants were completed using fluences ranging from 5 .010 1 3 5 .010 1 5 cm 2 at energies ranging from 1 6 keV. The corresponding projected range of the implants varied from 4.0 to 16.5 nm as simulated by SRIM. 78 Ge amorphization was completed using self implants with a fluence of 2 .010 1 4 Ge + /cm 2 at 120 keV which amorphized the substrate to a depth of approximately 100 nm as verified by transmission electron microscopy. The preamorphization conditions were chosen to encompass the peak of the implanted dopant concentration profile for all investigate d B + implant conditions. The inclusion of the B profile within the amorphous layer was done to reduce dynamic annealing and regrowth during B + implantation. 35

PAGE 40

40 2 .1.2 Dicing Saw Group IV semiconductors are notab ly brittle which makes sample preparation difficult. Sectioning of wafers into research sample s is typically completed using a high speed wire saw or diamond coated wire blade. The blade thickness is on the order of 100 m and the blade speed is approximately 25 krpm. The high speed blade enables the dicing of brittle semiconductors into small sa mples but creates kerf or chips along the edge of the samples. Kerf can be reduced by changing dicing parameters such as blade type or spindle speed, but can never be removed. Edge k erf is a major problem for samples to be characterized using the micro Hall effect as a clean edge is of utmost importance. To circumvent this issue, samples were diced with the implanted surface down and programming the saw to not cut the last 50 m of the wafer. In doing so, the wafer was not diced completely, but rather, the dicing was used to simulate a deep scribe to enable a clean cleave of the remaining 50 m of material. The technique enables the manufacture of small samples with edges tha t are much sharper and cleaner than capable by dicing or cleaving alone. 2 .1.3 Thermal Processing Following ion implantation, thermal processing is necessary to activate the implanted dopant and to remove any residual implant or regrowth related defects from the crystal lattice. 79 Activation, dopant diffusion, defect ev olution and solid phase epitaxial growth (SPEG) are all thermally activated processes that follow an Arrhenius behavior. With all of the processes occurring simultaneously, it is necessary to find an ideal combination of temperature and time, or thermal b udget, that produces the best combination of favorable material properties. For example, while a high temperature anneal for a long time may provide the best dopant activation, 55 it will also increase

PAGE 41

41 dopant diffusion or may allow for the formation of extended defects that will be detrimental to device properties. It is this balancing act that has warranted the advancement of annealing technology towards temperatures in excess of the melting temperature, T m for ultra fast times with femtosecond laser pulses. 80 For Si device processing, the primary focus towards these thermal budgets was driven by reducing transient enhanced diffusion (TED) which is a non equilibrium diffusion process that is driven by dopant defect i nteractions during annealing. 57,67,81 TED is characterized by dopant concentration profiles that extend to much greater depths than those created immediately after implantation or even those ex pected from typical diffusion processes. In addition, equilibrium diffusion of B in Ge is not significant which reduces the need for extreme thermal processing techniques. 28 30 Although TED of B is not an important concern with Ge, rapid thermal annealing at increased temperature for shorter times allows for decreased processing times whic h is more relevant for industry as they reduce the total time required to process a device. The reduced ti me devoted to processing each has a direct economic benefit. In this work, samples were thermally processed using two methods: standard furnace annealing and rapid thermal annealing. Standard furnace anneals were completed in a quartz tube furnace which is characterized by a heating coil that is categorized as having a long sample ramp up and cool down period which is on the order of several minutes. Tube furnace anneals wer e completed with N 2 flowing at a rate of approximately 1 L/min to reduce surface oxidation. The sample was placed in a

PAGE 42

42 pre heated quartz boat and inserted to the center of the tube. The temperature was not measured directly at the sample, but rather the temperature of the boat was monitored with a thermocouple prior to beginning the annealing process Due to the long duration of tube furnace anneals for those anneals completed at 500 C or higher samples were capped with SiO 2 to resist surface desorption. In contrast to the long anneals completed with standard annealing, rapid thermal annealing (RTA) employs the use of high powered incandescent tungsten lamps to control the temperatu re. The lamps are placed above and below the sample on the outside of an ambient controlled quartz chamber which allows for rapid heating and cooling with rates on the order of 100 C/s. The sample is placed on a carrier wafer which is then placed on a qu artz tray that is only in contact with the wafer by three small pins to reduce thermal conductivity. Due to quartz being transparent to the radiation produced by the lamps, RTA anneals are considered a cold wall process because only the sample and carrier wafer are heated during processing. The lack of chamber heating reduces chamber contamination and allows for short down times between processing runs. S amples were processed in an N 2 ambient with actual annealing times on the order of 60 s with a ramp rate of approximately 50 C/s The processing temperature was monitored in real time with the use of a thermocouple sandwiched between two carrier wafers rather than bonding the thermocouple directly to the carrier wafer. The carrier wafer sandwich was em ployed to reduce the risk of carrier wafer breakage during processing. During processing, the thermal stress of quick ramp rates is significant and capable of fracturing large samples which is exacerbated by bonding the

PAGE 43

43 thermocouple to the wafer. The tem perature of the sample is not directly measured, but error in the actual processing temperature is believed to be due less than 3%. Due to the short duration of the anneals, no special surface preparation was employed for RTA anneals at or below 600 C. Above this temperature, an SiO 2 capping layer was employed to reduce Ge surface desorption. 41 An AG Heatpulse 4100 was used for all RTA anneals in this work. 2 .2 Electrical Characterization 2 .2.1 Four Point Probe A four point probe (4PP) test is a quick and simple testing procedure that is capable of producing the electrical resistivity of semiconducting materials. As the name suggests, the test is conducted with four equidistant probes thro ugh which two of the probes a current is passed and through the two remaining probes a voltage is sensed. There are three probe configurations used for 4PP meas urements, but the most common is the C configuration in which current is passed between 1 and 4 with volta ge probe configuration is displayed in Figure 2 1 In this configuration, the electrical resistivity can be easily acquired if a few geometrical measurements are reasonably well known or if a few assump tions are made. As seen in Figure 2 2 f or a sample with a n area, A length, L resistivity is related to the resistance R by the following equation: ( 2 1 ) By separating sample thickness, t and width, w from the area term and combining th e resistivity and thickness values, the resistance can be defined by a sheet resistance, R S In doing so, the resistance can be described independent of sample thickness which is

PAGE 44

44 advantageous when describing doped layers created through an implantation process in which layer thicknesses are not always known. The relation between resistivity and sheet resistance is as follows: (2 2 ) The units of sheet resistance are ohms ( ), but it is typically represented as ohms/square ( between bulk resistance Assuming an equidistant probe spacing of s and a sample size in which L w >> s and t << s the sheet r esistance is then measured from: (2 3 ) where I is the current applied through contacts 1 and 4, V is the voltage se nsed between contacts 2 and 3, and t is the thickness of the conductin g layer. However, as mentioned previously, for ion implanted samples, the thickness of the conducting layer is not always readily apparent. In addition, the reporting of sample resistivity may obscure the true electrical behavior of the conducting layer because it is a function of thickness; ie, a thin layer with the same R S as a thick layer will be much more conductive in comparison. Therefore, sheet resistance is typically reported instead of resistivity values for ion implanted substrates. 2 .2.2 Hall Effect Although four point probe yields a quick and simple measure of sheet resistance, further characterization needs to be completed to fully comprehend the electrical behavior of implanted dopant atoms. The sheet resistance does not yield any informati on regarding the total number of electrical carriers or active dopants, commonly referred to as sheet number ( n S ) or the drift mobility ( D ) of charge carriers in the

PAGE 45

45 semiconductor. The sheet resistance is related to the sheet number and drift mobility by:: ( 2 4 ) where R S is related to the integral of the elementary charge q D and n S from the surface to the depth of the implanted layer t. Hall effect characterization enables the determination of whether any change in R S is governed by n S or D and is crucial for full understanding of the electrical characteristics of the implanted layer. The Hall effect enables the characterization of n S or D by measuring the electrical properties of a semiconductor in the presence of a magnetic field. 82 The governing principle behind the Hall effect is the Lorentz force which states that when an electron with the elementary charge, q is driven by an electric field to a velocity, v in a direction perpendicular to an a pplied magnetic field, B it experiences a force equal to ( q v B ) a cting normal to both directions. Con versely, the same is true for holes where the charge is reversed. By applying a force to the charge carriers as they flow through the semiconductor, the path of the carriers is bent towards one side of the sample which creates a measurable electrical potential denoted the Hall voltage, V H An example of the Lorentz force in an n type se miconductor is shown in Figure 2 3 Typical Hall effect measuremen ts ar e conducted on square samples 1 cm on a side with contacts placed at the corners. For p type Si or Ge samples, the use of InGa eutectic is an excellent contact material. The corner contact configuration also enables the characterization of the sheet resistance through the van der Pauw technique 83,84 For both Hall and van der Pauw techniques, electrical contac ts should be less than 1/10 th the length of the sample and placed near the corner to reduce measurement

PAGE 46

46 error. 85 90 An illustration of the typical contac t placement is given in Figure 2 4 The R S can be calculated by measuring characteristic resistances through pairs of parallel contacts. For example passing a cur rent through contacts 1 and 2 creates a measurable potential in contacts 3 and 4. (Figure 2 5 ) The characteristic resistance for this measurement is denoted R 12,34 and is calculated by dividing the measured potential by the applied current. The sheet resi stance of the sample can be calculated by the following equation: (2 5 ) where the R 12,34 and R 23,41 values are measured which allows for the calculation of sheet resistance. For Hall measurements completed in the van der Pauw configuration, a defined current on the order of 1 mA is passed through a pair of diagonal contacts (contacts 1 and 3 in Figure 2 5 ) in the presence of a perpendicular magnetic field on the order of 1 tesla which is typically produced by an electromagnetic magnet. The magnetic field deflect s the carriers which creates the Hall voltage which is measured through the other pair of diagonal contac ts (contacts 2 and 4 in Figure 2 5 ). The process is completed iteratively for all contact combinations and measured values averaged to minimize measur ement noise. The magnitude of the Hall voltage, V H is relates both the applied current and magnetic field through the following relation: (2 6 )

PAGE 47

47 where I is the applied current, B is the applied magnetic field, q is the elemen tary charge of the carrier, and n H is the Hall sheet carrier density. From the V H measurement in which all terms are well known, it is possible to calculate the Hall sheet carrier density. Similar to sheet resistance which is independent of thickness, th e sheet carrier density (or sheet number) is the number of electrically active carriers in the sample independent of sample thickness and is given in the units of (cm 2 ) Once n H is known, the Hall mobility can be calculated from the following equation: (2 7 ) Due to the complex band structure of Ge, and the presence of neutral and ionized impurities creating scattering sites, a correction factor needs to be applied to the Hall values obtained to determine the true drift mobil ity and sheet number. 91,92 The Hall scattering factor, r H relates Hall values to sheet number and drift mobility value by: ( 2 8 ) ( 2 9 ) where n s is the sheet carrier density and d is the drift mobility The scattering factor can be determined through simulation or determined empirically. 44,93 96 The measurements completed in this work were corrected using a scattering factor of 1.21 as determined empirically by Mirabella et al for high dose B + implantation into Ge. 31 2 .2.3 Micro Hall Effect & Micro Four Point Probe T he length scales associated with modern and future IC devices are ever decreasing which creates issues with device metrology. 97 The narrow band gap of Ge (0.66 eV) compared to that of Si (1.21 eV) 25 in conjunction with the shallow and box like ion profiles creates an environment primed for junction leakage from the implanted

PAGE 48

48 layer 98 103 The increased leakage current is significant for next generation devices, but also for device metrology which has driven the need for advanced metrology techniques. An example of leakage current affecting a Hall effect measurement is disp layed in Figure 2 6 A micro Hall effect and micro four point probe tool is a characterization method that is capable of measuring the electrical characteristics of ultra shallow profiles. Its operation is very similar to that of conventional 4PP and Ha ll effect characterization in that four contacts are used, but with the a few caveats The probe spacing is on the order of 20 m as opposed to 1 mm which works to reduce or eliminate any influence from junction leakage. In essence, by decreasing the hori zontal probe spacing, the vertical characterization depth is also decreased as seen in Figure 2 7 104 By decreasing the probe spacing to the m scale, the techniques have earned the names micro four point probe (m4PP) and the micro Hall effect (MHE). For 4PP characterization, the conversion from conventional to m4P P is rather straightforward in that the only change to the experimental method is that not only is C configuration used, but configuration A and B is also used to reduce measurement error. With reduced probe spacing, characterized volume shifts further t owards the surface and leakage current effects are reduced. 104 107 However, the conversion of Hall effect measurements to the m scale are not as st raight forward. As opposed to the traditional van der Pauw configuration used for conventional Hall effect measurements, MHE measurements utilize the same in line probes used for M4PP samples. 108 For conventional Hall effect measurements, the use of in line contacts wou ld not be feasible as the measur ement relies on the deflection of

PAGE 49

49 the current to create the Hall voltage. In order for M HE measurements to utilize in line probes, it relies on the presence of an insulating boundary such as a cleaved edge In addition to the influence of a magnetic field, the cleaved edge alters the current flow near the injection point which is detectable using the contact probes in the B configuration as seen in Figure 2 8 The cleaved edge creates a resistance difference R Figure 2 9 gives evidence of the measurable resistance difference near the edge which is non existent a mere 60 m from the edge. Once the R term is measured, the Hall sheet resistance, R H can be extracted from the following equation: ( 2 1 0 ) where s is the probe spacing and y 0 is the distance from the sample edge. Once R H is known, t he drift mobility and sheet number can be calculated from: ( 2 1 1 ) ( 2 1 2 ) w here the constants are the same as those used in conventional Hall effect calculations. In this work, all samples were characterized using a CAPRES microRSP M 150 at CAPRES in Kgs. Lyngby, Denmark. 2 .3 Structural Characterization 2 .3.1 Transmissio n Electron Microscopy The transmission electron microscope (TEM) is an indispensable tool for studying semiconductors. It provides valuable information regarding ion implantation processing

PAGE 50

50 of semiconductors and produces corroborating information for elec trical data A TEM relies on a high energy electron beam that is highly focused through a series of electromagnetic lenses towards an electron transparent specimen. Figure 2 1 0 shows a schematic diagram of a TEM. The interaction of the accelerated electr on beam with the thin sample is capable of providing a wealth of information regarding the crystal orientation, chemical compositi on, and microstructural imaging to be brief. The significant advantage of TEM characterization over other methods is that the instrument has a tremendous spatial resolution owing to the ultra small wavelength of accelerated electrons. 109,110 According to the de Broglie principle regarding wave particle dua lity, 111 an electron s wavelength, h and its momentum, p by: (2 1 3 ) It is possible to calculate the momentum of the accelerated electron because the momentum of the electron can be determined from: (2 1 4 ) where m 0 is the mass of an electron, is the electron velocity, q is the f undamental charge, and V is the accelerating voltage of t he microscope. Substituting Equation 2 15 into Equation 2 1 3 and excluding any relativistic effects, the wavelength of an accelerated electron as a function of voltage can be determined by: ( 2 1 5 ) Assuming a typical accelerating voltage of 200 kV, the electron wavelength is on the order of 2.7 pm. However, the resolution of contemporary microscopes are not close to

PAGE 51

51 this value due to the inability to manufacture a perfect elec tromagnetic lens which introduces aberrations during operation. However, modern day instruments are capable of sub Angstrom resolution which enables the characterization of nanometer scale features to be commonplace. 2 .3.1.1 TEM Sample Preparation For T EM characterization to be effective, the sample must be thin enough to be transparent to the beam. With increasing sample thickness, the beam undergoes additional elastic and inelastic scattering processes which decrease spatial resolution. Due to increa sed scattering in the sample, a fraction of electrons will lose a small amount of their initial energy which is typically on the order of 10 eV. Due to the difference in energy, chromatic aberration will be introduced which will greatly reduce the resolut ion of the TEM. 110 Unfortunately, scattering increases with increasing atomic mass whic h makes the sample preparation for Ge much more difficult than for Si. For high resolution imaging of Ge, the sample thickness should typically be under 5 0 nm. 112 The preparation of electron transparent semiconductor samples is not a trivial task. It is commonly more involved and time consuming th an the TEM characterization that follows. In general, there are two primary orientations for TEM characterization of semiconductors: 1. Cross section or XTEM and 2. Plan view or PTEM. In both orientations, it is critical to protect the surface as most io n implant processes place the region of interest (ROI) within 1 m of the surface. Modern sample preparation relies heavily on ion beam preparation of TEM samples with both focused ion beam (FIB) and broad ion beam (BIB) milling tools widely used. In this work, plan view TEM samples were prepared using a broad ion b eam (or ion mill) to produce electron transparent samples. PTEM samples were created by

PAGE 52

52 dicing a 1 2 mm rectangle of the material followed by the mechanical thinning the back side of the sample to a thickness of approximately 10 m as monitored by optic al microscopy. A piece of Si was polished on the same stub to use a thickness reference because Si is well known to be transparent to infrared when approximately 10 m thick. Once thinned, a 3 mm Cu grid with oval opening was affixed to the back of the sa mple using M Bond 610. The samples were then polished using a Fischione 1010 ion mill at an accelerating voltage of 3kV, a beam current of 5 mA, and a milling angle of 15 Cross sectional TEM samples were prepared using a focused ion beam (FIB). A FIB is essentially an SEM which has a high current focused Ga + beam mounted at 52 to the electron column which enables a sample to be milled from the surface of interest. In order to protect the sample from Ga + implantation and damage, an approximately 100 nm thick layer of carbon is thermally evaporated ex situ prior to beginning sample preparation. In situ Pt is deposited on top of the ROI to further protect from ion milling damage. Sample preparation proceeds by milling trenches on both sides of the ROI wh ich progress closer to each other until the sample is thinned to approximately 1 m thick lamella. The sample is then transferred in situ to a Cu grid where is it thinned to electron transparency. As thinning progresses, beam energy and current is decreased to reduce the amount of damage imparted to the sample. In comparison to a b road ion beam system, FIB preparation is advantageous for its site specificity, ease and quickness of operation, and low volume of material needed for sample preparation. However, a significant downfall of the FIB is the incredible difficulty of creating a PTEM sample which typically results in low quality. In general, FIB samples are of lower quality than other traditional methods due to the high current

PAGE 53

53 and high ion energy which creates a layer of damaged or amorphous material on both sides of the sampl e. The damaged layer can be reduced by decreasing the ion energy for the final milling steps, but the resulting sample is still hindered by a damaged surface. Typical edge damage of ion milled and FIB milled samples is depicted in Figure 2 1 1 Despite the low energy polishing mill used in the FIB, it is evident that ion milled samples exhibit less surface damage. However, both methods provide adequate samples when proper care and attention is given during the preparation. 2 .3. 2 Secondary Ion Mass Spec trometry Secondary ion mass spectrometry, or SIMS, is a destructive chemical analysis technique that is used to quantify minute dopant and impurity concentrations in semiconductors with a sensitivity to levels as low as 1 10 1 3 cm 3 113,114 The ability of SIM S to detect minute impurity concentrations lends itself wel l to semiconductor characterization which is the biggest proponent of its use In addition, it is capable of providing a dopant concentration depth profile from the surface to a depth of several microns with a resolution of approximately 1 nm. SIMS prof iling of B in Ge is carried out by accelerating a focused beam of primary Cs or O ions at a fixed flux towards the surface of interest. The impact of the primary ion at the surface induces the ejection of surface atoms from the surface of the material thr ough a sputtering process. A fraction of these ejected atoms are ionized and are called secondary ions. The primary and secondary ions that escape the material are then passed through a mass analyzer which separates the ions based on mass and charge by v arying the magnetic field. Once the secondary ion s ar e separated, they are sent to a mass spectrometer and the counts collected. The mass

PAGE 54

54 spectrometer allows for the detection of all elements and provides an elemental profile of the impurities in the sam ple. The primary ion beam sputtering of the material creates a crater in the surface which can be measured ex situ to determine the sputtering rate and profiling depth. The combination of secondary ion counts and sputtering rate is used to create the do pant depth profile. However, t he construction of the concentration profile is not trivial due to differences in sputtering and ionization yields between the primary ions and the impuri ty and substrate atoms. 115 The differences in sputter and ionization yield is a functio n of the surface chemistry and primary ion. Because of these differences, the use of a calibration standard is required for accurate depth profiling Typically, for ion implanted samples an as implanted sample is used as a calibration standard by averag ing the ion signal with the ratio of the implanted fluence to the sputtered depth 116 In addition, the near Gaussian shape and the assumed 1% implant uniformity provides further confirmation that the impl anted species is being measured rather than some other interfering species. The use of an as implanted sample is assumed to be valid because both samples should have the identical chemical compositions in the absence of any dopant out diffusion. Fortunat ely, this is not an issue for B in Ge under typical processing conditions An important issue for characterizing dopant profiles of ultra shallow implants is the presence of profile artifacts produced during the sputtering of a surface or interface. The se issues arise due to differences in dynamically changing ion yields in these near surface and near interface regions and exhibit themselves as an ano malous change in concentration unique to each sample characterized. These effects can be reduced by

PAGE 55

55 usin g a p rimary ion with increased mass decreased sputtering angle or by evaporating a thin film of the substrate material on the surface 117,118 ; however, the complete removal is not possible Due to the presence of surface artifacts, SIMS profiling of ultra shallow B in Ge is not capable of producing absolute fluence meas urements, but is satisfactory for comparison between as implanted and proce ssed concentration profiles. 2 .3. 3 Rutherford Backscattering Spectro metry Rutherford backscattering spectroscopy, or RBS, is an ion beam analysis technique that takes advantage of the inherent structure of a crystalline material to monitor the compositio n and structure of the surface of a sample. 119 The technique uti lizes a beam of light ions, typically H + or He 2+ that is accelerated to an e nergy in excess of 0.5 MeV and directed towards a target precisely aligned along a crystal axis The primary ions enter the sample to a given depth and collide with a target atom F ollowing collision with the sample, ions are backscattered and collected by an energy sensitive detector. A plot of backscattered ion counts versus energy is then created from which information regarding the crystal structure is determined. For examp le, Figure 2 1 2 displays the channeling yield for a (001) Ge sample amorphized to a depth of 100 nm (black) and an as received (001) Ge wafer. The randomized structure of the amorphized sample increases ion counts significantly over the pristine sample. In contrast, the pristine Ge sample allows for ion channeling down the (001) axis which reduces ion counts. The peak for the virgin crystal is due to increased scattering at the surface due to the presence of a surface oxide. The energy spread of the inc reased damaged counts of the amorphized sample corresponds to the thi ckness of the amorphous layer in which energy of the backscattered particles directly corresponds to

PAGE 56

56 a distance or depth travelled through the crystal. From the difference in counts betw een two samples, the number of displaced Ge atoms can be determined. One of the primary drawbacks of RBS is the low sensitivity for light elements, which often requires the combination of other nuclear based methods like nuclear reaction analysis (NRA) o r elastic recoil detection analysis (ERD) which will be discussed in following sections. In addition, its chemical sensitivity 120 and depth resolution 119 are orders of magnitude less than that of other techniques. However, its ability to accurately monitor crystal damage is an advantage that few other techniques offer. 2 .3. 4 Nuclear Reaction Analysis Nuclear reaction analysis, o r NRA, is a technique that provides quantitative information regarding the chemical concentration and lattice location of light impurities in semiconductors. Similar to RBS, it utilizes the interactions of accelerated light ions and target nuclei to chara cterize the target. 119 However, the primary advantage of NRA cha racterization is its ability to quantify absolutely the concentration and location of light elements such as B. During NRA characterization the primary ion undergoes a characteristic nuclear reaction with the impurity atom of interest where the reaction byproducts are counted. For example, a typical reaction induced protons reacting with boron is described by 11 B(p, ) 8 Be where the reaction produces an alpha particle and are directly proportional to the B concentration in the target. (Figure 2 1 3 ) More significant to this work by analyzing the target along known orientations such as < 001 >, < 011 >, and <111> it is possible to determine if the dopant is sitting in a preferred crystallographic orientation.

PAGE 57

57 To complete NRA and lattice location characteriz ation of B lattice location in Ge, several standards are needed to complete the experiment. Namely, a sample with B randomly distributed in amorphous Ge simulates the yield for a completely random or off lattice distribution of B and a pristine (001) Ge w afer simulates the yield for a particles are shown in Figure 2 18. Using these spectra as a reference, it is possible to determine the fraction of B that is substitutio nal in an unknown sample through the use of the following equation: ( 2 1 6 ) where f s (B) is the fraction of substitutional B, B is the ratio of the off lattice and random yield, and Ge is ratio of the number of displace d atoms measured in the pristine (001) sample to the number of displaced atoms measured in the random sample. The normalized channelling yield ( Ge and B for Ge and B atoms, respectively) is defined as the ratio of the aligned yield to the yield of ra ndomly directed beam. B i s obtained from the particle yield normalized to the random yield and is proportional to the fraction of B displaced out of the lattice. Ge i s measured just below the surface peak of th e backscattered proton spectrum. The characterization is completed along the three maj or axes which enables the determination of lattice location of displaced B atoms. 2 .3. 5 Elastic Recoil Detection Elastic Recoil Detection, or ERD, utilizes similar principles to those discussed in preceding sections on RBS and NRA where an accelerated io n impacts a target and the reaction products measured. NRA relies on elastic scattering between the primary ion and the target atom for the ability to monitor elemental concentration and depth profiling

PAGE 58

58 of light impurity atoms in heavy atomic mass targets However, in contrast to RBS and NRA, ERD typically utilizes a heavy ion such as Si, to forward scatter light recoil ions such as B, towards a detector for characterization. The forward scattered yield of light ions is directly proportional to the che mical composition in the solid which allows for quantitative elemental analysis. 119 Similar to RBS, ERD allows for absolute quantification of dopant concentration in the absence of a calibration standard. In addition, it sensitivity to light elements makes it an ideal candidate to determine the chemical concentration of B in Ge. In an ERD experiment, a beam of heavy atomic mass ions are accelerated toward a target consisting of a matrix of heavy atomic mass atoms with an unknown concentration of impurity atoms of interest. The primary ion is angled at grazing incide nce with the detector also placed at grazing exit incidence. A thin mylar foil on the order of 10 2 15 The light atomic mass of the impurity atoms allows for easy penetration through the foil whilst stopping the primary ions as well as any forward scattered matrix atoms. The detected counts allow for the conversion into an elemental areal concentration. 121 2 .4 S ummary In this work, a combination of electrical and structural characterization yielded information regarding the properties of ion implanted B in Ge following thermal processing. I on implantation was used to introduce a precise amount of B into (001) Ge substrates. Elastic Recoil Detection was used to quantify the residual implanted B fluence as a function of energy. It was shown that a significant fraction of implanted dopant was lost due to ion backscattering in good agreement with simulations.

PAGE 59

59 Secon dary ion mass spectrometry was used to monitor the B concentration profile as a function of annealing temperature. Electrical characterization was carried out with micro Hall effect and micro four point probe techniques to determine the sheet resistance, s heet carrier density and the drift mobility of B + implanted samples. Nuclear reaction analysis and Rutherford backscattering spectrometry was used to yield information regarding the lattice location of B following processing to corroborate electrical data

PAGE 60

60 Figure 2 1 Sch ematic of four point probe configuration used to determine sheet resistance, R S Figure 2 2 Sample geometry for a four point probe measurement

PAGE 61

61 Figure 2 3 Illustration of the electrons passing through a solid in the p resence of a magnetic field. Figure 2 4 Schematic of the van der Pauw contact scheme

PAGE 62

62 Figure 2 5 Illustration of the contact scheme typically used for Hall effect measurement s Current is passed through contacts 1 and 3 and V H is measured th rough contacts 2 and 4. Figure 2 6 Leakage current into the n type substrate can dominate a measurement if the metallurgical junction is not sufficient

PAGE 63

63 Figure 2 7 Decreased horizontal length scale of measurement probes reduces the junction leakage

PAGE 64

64 Figure 2 8 Current paths near contacts for M4PP measurements under the influence of a magnetic field A ) In the bulk, current travels tangentially a round current injection contact. B ) N ear an insulating boundary, the current fields in the sample under go a detectable shift

PAGE 65

65 Figure 2 9 The presence of a cleaved boundary gives rise to a resistance difference R during micro Hall effect measurements. At A ) 4 m from the cleaved edge which is not evident in B ) at 60 m from the cleaved edge.

PAGE 66

66 Figure 2 1 0 Schematic diagram of a TEM

PAGE 67

67 Figure 2 1 1 Representation of typical damage layer produced by FIB milling A ) at 7 kV with Ga + beam and B ) ion milling at 3 kV with Ar + beam

PAGE 68

68 Figure 2 1 2 Ty pical data produced from c RBS experiment. (001) Ge crystal a morphized to a depth of 100 nm with a 100 keV Ge + implant to a fluence of 2 .010 1 4 cm 2 and a virgin substrate

PAGE 69

69 Figure 2 1 3 Illustration of a nuclear reaction between a proton and a B a tom as it produces an particle Figure 2 1 4 Representation of particle counts produced by analyzing B atoms in different confirmations ; randomly oriented (grey), fully substitutional (green) and off lattice (red).

PAGE 70

70 Figure 2 1 5 Experimental setup of an ERD exper iment

PAGE 71

71 CHAPTE R 3 ACTIVATION OF ULTRA SHALLOW BORON IMPLAN TS IN GERMANIUM 3 .1 Activation of B + Implants in Ge As the scaling associated with traditional Si metal oxide semiconductor (MOS) devices reaches fundamental limits, changes in MOS design or materia l composition are becoming necessary. Ge is a promising alternative material for next generation MOS devices as its increased carrier mobility makes it an attractive replacement for Si. 2 In addition, its reduced melting temperature may allow for less aggressive annealing recipes which could prove advantageous for process integration of hig h device structures. In the case of B, it is known that the equilibrium chemical solid solubility limit in Ge 28 is 5.510 18 cm 3 homologous temperature. 55,122 However, electrical activation in excess of this value has been reported for B + implantation into single crystal (c) as well as pre amorphized (PA) material. 23,31,34,35,37 ,46,123 For B + implants at 35 keV into PA Ge, complete activation was observed for all doses up to 7.6 10 15 cm 2 implying active B concentrations as high as 5.7 10 20 cm 3 31 In the case of PA Ge, annealing to induce solid phase epitaxial growth (SPEG) of the amorphized layer results in the incorporation of a concentration of substitutional B greater than the equilibrium chemical concentration; analogous behavior has been similarly reported for the case of B + implantation into Si. 124 Although B activation has been characterized for deep B + implantation int o both c Ge and PA Ge, the activation of low energy B + implantation in Ge remains poorly understood and characterized. In the few studies which utilized low energy B +

PAGE 72

7 2 implantation, the activation level, if determined, was calculated indirectly by fitting resistance values to a mobility model. 23,45,123 This model was developed using Hall effect measurements and may suffer from possible error introduced from Hall scattering factor assumptions. 44 In order to accurately characterize the activation behavior, the sheet carrier density and mobility shou ld be measured directly using a Hall effect technique. Characterization of shallow implant activation is very challenging and in some cases impossible with conventional four point probe and Hall effect techniques partially due to high junction leakage. 104,125,126 Due to the smaller band gap of Ge compared to Si and the larger number of intrinsic carriers 2 junction leakage is increased for Ge 98,127 which exacerbates the characterization challenges. Recently, instruments and techniques have been developed which perform four point probe and Hall effect measurements using probes with m scale spacing as described in detail elsewhere 104,107,108,128 ; these micro four point probe (M4PP) and micro Hall effect (MHE) measurements have been shown to greatly reduce the effects of junction leakage 104,125,126 and have previously been used for successful charact erization of active shallow dopants in Ge. 46,129 In addition to the electrical characterization of ultra shallow B + implants, the chemical dose following ion implantation can not be assumed to be equal to the implanted dose due to ion solid interactions reducing the total amount of implanted B. Due to the relatively l ight atomic mass relative to Ge, the B ions are highly susceptible to backscattering which is caused by elastic collisions between the implanted B + and the target Ge matrix which expels B from the lattice (Figure 3 2 ) The effect of

PAGE 73

73 backscattering is incr eased with reducing ion mass and ion energy which creates an environment primed for dose loss for ultra shallow B + implants into Ge. As shown in Figure 3 1, for a 2 keV B + implant, greater than 20% of the implanted fluence is expected to be lost to backsc attering in Ge while only 6% is lost for Si. 78 Due to the susceptibility of ultra shallow B + implants to backscatter, it is necessary to measure the residual chemical dose for accurate quantification of the implanted dopants. Following implantat ion and annealing, it is also important to structurally determine the location of residual B atoms to fully understand the activation behavior. It has been shown previously that Ge is p type immediately following B + implantation with no annealing treatmen t. 33 Deep level transient spectroscopy (DLTS) characterization was used and it was determined that the p type activation was due to implant damage related acceptor centers and not as implanted substitutional B atoms. Nuclear reaction ana lysis (NRA) is an ion beam analysis technique that is able to determine the lattice location of B following implantation in Ge. In addition, the lattice location results obtained by using NRA are useful for corroborating electrical data if results are at ypical. In this chapter the activation of low energy B + implantation into c Ge and PA Ge was studied using M4PP and MHE measurements. Ion beam analysis techniques and tra nsmission electron microscopy are used to further explain the electrical behavior ob served upon annealing. 3 .2 Experimental Details Czochralski grown n work. A set of c Ge samples was produced by B + implantation at 2 keV to doses of 5.010 13 5.010 15 cm 2 while a set of PA Ge samples was produced by first performing Ge + imp lantation at 120 keV to a dose of 210 14 cm 2 before the same B + implantation

PAGE 74

74 step. In the case of PA Ge samples, a continuous amorphous layer extending ~100 nm from the surface was produced as verified by high resolution cross sectional transmission elec tron microscopy (HR XTEM). Samples were annealed in N 2 ambient XTEM was used to image the microstructure of the specimens before and after annealing with samples prepared using focused ion beam (FIB) milling. Secondary ion mass spectrometry (SIMS) was perfor med on selected samples to determine B concentration profiles before and after annealing. Sheet resistance, Hall sheet number ( n H ), and Hall mobility ( H ) were measured using a CAPRES microRSP M 150 M4PP with Au coated probes, a probe spacing of 20 a permanent magnet with a magnetic flux density of 0.475 T. Hall sheet number and mobility values were adjusted to obtain the carrier sheet number ( n s ) and drift mobility ( d ) by using a scattering factor ( r H ) of 1.21 as determined empirically by Mirabel la et al for high dose B + implantation into Ge. 31 The carrier density and drift mobility are related to the Hall values by n s = n H r H and d = H / r H respectively. It has been speculated that a large fraction of the implanted B + fluence is lost to ion backscattering. 24,130 To characteri ze the as implanted chemical dose of ultra shallow B + implants in Ge, a set of variable energy samples were implanted at 2, 4, and 6 keV and characterized using elastic recoil detection (ERD). Any losses due to backscattering would be independent of impla nted dose; therefore, a dose of 5.010 15 cm 2 was used to increase measurement counts a nd decrease experimentation time for all implant energies investigated. ERD characterization was performed using the 11 B ( 28 Si, 11 B) reaction with a 28 MeV Si 4+ beam wi

PAGE 75

75 scattered Si ions and to allow the recoiled B a toms to enter the detector. The B counts enabled the calculation of the areal density of the implanted boron 121 Nuclear Reaction Analysis (NRA) is an ion beam technique to detect B atoms in Ge by measuring the yield of 11 8 Be 15,131 134 NRA and channelling measurements along the <100>, <110> and <111> axes were performed using a proton energy of 650 keV on a set of B + implanted samples. The particle detector was placed at 160 with respect to the inci dent beam direction and it was covered with a 10 m thick aluminised mylar film to prevent backscattered protons to reach the detector. A second detector was placed at 165 and was used to detect protons backscattered from Ge atoms and to perform the Ge sa mple alignment procedure. The normalized channelling yield ( Ge and B for host Ge and B atoms, respectively) is defined as the ratio of the yield obtained when the beam is oriented down a channelling axis to the yield obtained in a randomly oriented or amorphous crystal. B is proportional to the fractio n of B that is sitting in off lattice positions. Ge is a measure of the off lattice Ge atoms with respect to a perfect crystal. I f the off lattice B atoms do not have a preferential position, the n si milar B values will be detected along the main crystal axes Given the measured B values, the apparent substitutional fraction f s is defined as : (3 1) 3 .3 Characterization of Residual Implanted Dose To confirm the residual implanted dose of ultra shallow B + implants in Ge, samples as implanted to a dose of 5.010 15 cm 2 were characterized using ERD. The

PAGE 76

76 residual implanted dose for samples implanted at 2, 4, and 6 keV was found to be 3.8410 15 cm 2 3.8810 15 cm 2 and 4.1210 15 cm 2 respectively. The deviation from the implanted dose is significant as the loss is in excess of 20% of the implanted dose for the lowest implant energy. As speculated in previous reports, it is believed that ion backscatter is a large source of dose loss and at first inspection could seemingly reduce the activation of the B + implant. 24 Due to the low atomic mass relative to Ge and the 1/E 2 dependence of backscattering, boron is highly susceptible to ion backscattering during low energy implantation which reduces the chemical dose before any other processing is completed. Tak ing into account that samples were characterized as implanted, it is assumed that the deviation from implanted dose is due completely to backscattering losses during implantation. Boron is known to diffuse very slowly in Ge 28 30 and no further significant dose loss is expected due to surface desorption following annealing at 400 135 Figure 3 2 shows the percentage of implanted B + lost to backscattering as a function of implant energy as measured with ERD plotted in conjunction with SRIM simulations. 78 The simulations compare favorably with the dose loss values expe rimentally determined through ERD and confirm that a large fraction of the implanted dose is lost to ion backscattering. Given that backscattering is an energy dependent phenomenon, it is assumed that this behavior is identical for lower doses. The experi ment confirms that SRIM simulations are sufficient for estimating the retained implanted dose for ultra shallow B + implants in Ge. 3 .4 Hall Effect Characterization Figure 3 3 presents measured R s and n s values as a function of implanted B + dose for both c Ge and PA Ge samples following annealing at 40 C for 1 h. As shown

PAGE 77

77 in Figure 3 3 (a), the R s value decreases with increasing B + dose for both c Ge and PA Ge samples with c Ge samples exhibiting higher sheet resistance values compared to the PA Ge samples. The reduced R s values for the PA Ge samples are explained by the increased solubility and B incorporation during SPEG as previously reported by others for different implant conditions. 23,34,35 The carrier sheet number is plotted as a function of implanted B + dose for both c Ge and PA Ge samples after adjustment with a Hall scattering f ac tor of 1.21 as shown in Figure 3 3 (b); the PA Ge samples exhibited higher n s values compared to c Ge samples, similar to previous reports. 23,34,35 More interestingly, for B + doses less than 5.010 15 cm 2 the percent activation for both c Ge and PA Ge samples is relatively independent of dose at ~7 and ~30 %, respectively. At a dose of 5.010 15 cm 2 the difference in percent activation is much smaller, ~10 and 15 % for c Ge and PA Ge samples, respectiv ely. Figure 3 4 shows HR XTEM images of a c Ge sample B + implanted at 2 keV to a dose of 5.010 15 cm 2 As shown in Figure 3 4 (a), a damage layer extending 180.5 nm from the surface is evident, which matches well with the expected range 24 of the implant. At higher m agnifications, as shown in Figure 3 4 (b), amorphous pockets are evident within the damaged layer. Thus, the smaller difference in R s between c Ge and PA Ge for the case of a B + dose of 5.010 15 cm 2 is likely due to enhanced incorporation of substitutional B within the amorphous regions during SP EG upon annealing. To better understand the activation behavior, samples B + implanted to doses of 5.010 13 and 5.010 15 cm 2 into c Ge and PA Ge were analyzed using SIMS before and after annealing to determine B concentrat ion profiles, as shown in Figure 3 5 The measured B concentration profiles (both before and after annealing) were very similar to

PAGE 78

78 as implanted simulations. 78 The lack of dopant diffusion is in excellent agreement with previous reports. 34 The maximum active concentration was estimated using the SIMS profile and the corresponding measured n s val ues presented in Figure 3 2 ( B ). For c Ge and PA Ge samples B + implanted to a dose of 5.010 15 cm 2 the maximum active B concentrations were estimated at 2.010 20 and 4.010 20 cm 3 respectively; the value for PA Ge is similar to prior results while the results for c Ge are the highest reported 23,31,34,35,46,123 The observed activation behavior for both c Ge and PA Ge sample s is very interesting; incomplete activation is observed for even the lowest B + fluenc es in both sample sets. Based on the reported maximum active B concentration of 5 .5 10 20 cm 3 31 full activation should be expected for B + implantation at 2 keV to a dose of 510 13 cm 2 for both c Ge and PA Ge, yet the measured per cent activation in each case was ~7 and ~30 %, respectively. It should be further noted that this behavior implies that a single electrical solubility limit does not exist for these shallow implants in Ge. This is unexpected and may have resulted from ch emical dose loss or a unique dopant clustering/precipitation that scales with B + fluence Boron is not known to diffuse readily or suffer dose loss through the surface upon annealing at 400 C and Is thus not expected to be a contributor to the reduction o f active carriers 28 30 One contributing factor for the low levels of activation could be due to the shallow nature of the implant, even slight surface oxidation is capable of consuming a non n egligible portion of the implanted B + dose. HR XTEM analysis revealed the presence of a 3.00.2 nm thick surface oxide on all as impla nted samples, as shown in Figure 3 3 As per simulations 78 ~8 % of the implanted dose is rendered inactive via the

PAGE 79

79 presence of a GeO x layer of this thickness. Finally, it should be noted that the percent of dose rendered inactive due to both oxidation and ion backscattering should be independent of implanted dose; however, the combination of losses due to backscat tering and surface oxidation can only account for ~28% of dose loss. Thus, the bulk of the inactive fraction cannot be explained by these factors. Another possible source of inactive B is due to the formation of an inactive cluster or complex which has b een observed and extensively studied in Si. 13,67 In the case of Si, for a given processing condition, full activation is expected for low B + doses until a certain concentration threshold is reached after which clustering occurs. However, for the presented data, incomp lete activation is observed even for the lowest B + dose with the percent activation remaining relatively constant with increasing dose for doses <5.010 15 cm 2 Work by Impellizzeri et al 37 revealed incomplete activation for B + implantation at 35 keV into c Ge, which was attributed to the formation of a B Ge cluster; however, incomp lete activation observed in this work is much more pronounced and is observed in both c Ge and PA Ge samples. The physical explanation behind the observed anomalous activation behavior is unclear; however, the close proximity of the surface is a possible contributing factor, since it is known that defect production and annihilation can be influenced by surface proximity. 136 Sample characterization using structural techniques i s necessary to corroborate the electrical data and to gather further data regarding the anomalous activation behavior. 3 .5 Nuclear Reaction and Channeling Analyses It is expected that the formation of an inactive complex would be dependent on the impla nted fluence or overall B concentration with respect to a solubility limit. If the B concentration were to exceed this limit, clustering and inactive dopants would be

PAGE 80

80 expected; below this limit, substitutional and active dopants would be the case. This b ehavior is what is typically observed for dopant implantation and activation in Si. It should be noted that this behavior is much different from what has been reported previously regarding B clustering in Ge 34,37 39 for two reasons: 1. Although not as prominent, dose independent clustering also occurred in ultra shallow B + implants in PA Ge and 2. The behavior is independent of fluence and the activation percentage is fixed even for a 5.010 13 B + cm 2 implant for both c Ge and PA Ge. Since the electrical activation behavior deviates far from what has been observed previously in Ge as well as in Si, select samples were structurally characterized through channeling an alyses utilizing nuclear reactions to determine the substitutional fraction of B after processing. Table 3 1 shows the fraction of substitutional B as measured using channeling Ge and PA Ge, low and high fluence samples were characterized to obtain structural data on samples that would be expected to be below an d above electrical solubility, respectively. For all characterized samples, the normalized channeling yield ( B ) obtained along the <100>, <110>, and <111> orientations are all approximately equal. This suggests that the non substitutional B fraction is randomly distributed throughout the lattice. The substitutional fraction for each sample was obtained by using Equation 3 1 and averaging the channeling yield obtained along each crystal orientation. The substitutional fractions obtained using channeling analyses and electrical measurements agree favorably for all characterized samples. For even a modest B + fluence of 1.010 14 cm 2 (peak B concentration of approximately 6.010 19 cm 3 ) 78 the substitutional fractions as measured by Hall and channel ing analyses are in agreement at approximately 10%.

PAGE 81

81 The data obtained from NRA and channeling analyses confirm the efficacy of electrical measurements of ultra shallow B + implants in Ge and suggest that electrical results obtained using different process ing conditions, ie different implant fluences or annealing thermal budgets, should be considered to be accurate. 3 6 Summary In summary, residual implanted dose following ultra shallow B + implantation was characterized using ERD and is in good agreement with values obtained through Monte Carlo simulations. The results permit the use of SRIM simulations for the evaluation of backscattering loss. T he activation of B + implantation at 2 keV into crystalline and pre amorphized Ge was studied using micro four point probe and micro Hall effect measurements. For B + doses of 5.0 10 13 5.010 15 cm 2 pre amorphized samples exhibited greater activation compared to crystalline samples followi for 1 h. In the case of B + implantation to a dose of 5.010 15 cm 2 the discrepancy in activation between crystalline and pre amorphized samples wa s much smaller; this was attributed to solid phase epitaxial growth within amorphous pock ets formed in crystalline samples as a result of only B + implantation. Notably for both crystalline and pre amorphized samples, the measured percent activation was approximately independent of implanted dose; this behavior is in stark contrast to reporte d activation behavior of shallow B + implantation in Si and deeper B + implantation in Ge. In conjunction with Hall measurements, samples were characterized using NRA and have determined that a large fraction of the residual implanted fluence is off lattice and electrically inactive. It should be noted that these activation values are well below reported electrical solubility. These results suggest the possibility of the formation of a dose independent B Ge cluster which is unique to ultra

PAGE 82

82 shallow B + impla ntation in Ge; the physical explanation of this anomalous behavior is unclear, though the close proximity of the surface and reduced B + implant energy may be a contributing factor.

PAGE 83

83 Table 3 1. Channeling yields and corresponding substitutional B fraction s as measured by channeling analysis using nuclear reactions along three axes (<100>, 60s. B + Fluence (/cm 2 ) B <100> B <110> B <111> NRA a Hall b c Ge 1.010 14 0.920. 05 0.900.05 0.910.05 9 .0 11.9 c Ge 1.010 15 0.810.01 0.830.01 0.840.01 18.3 12 .0 c Ge 5.010 15 0.910.01 0.890.01 0.920.01 11 .0 11.7 PA Ge 1.010 15 0.410.01 0.390.01 0.430.01 62 .0 53.9 PA Ge 5.010 15 0.760.01 0.750.01 0.790.01 24.6 19 .0 a Determined by averaging B values from <100>, <110>, and <111> orientations b Ratio of carrier sheet density divided by the residual fluence

PAGE 84

84 Fig ure 3 1 The effect of B + backscattering for implantation into Ge and Si as simulated by SRIM.

PAGE 85

85 Figure 3 2 Percent of B ion s backscattered as a function of implant energy into Ge as simulated by SRIM and experimentally determined through ERD for a 5.010 15 cm implant into Ge.

PAGE 86

86 Figure 3 3 Electrical characteristics as a function of B + fluence implanted at 2 keV into crys talline and preamorphized Ge, respectively, 1 h A ) Measured sheet resistance ( R s ) and B ) sheet number ( n s ) as a function of B + dose impla nted

PAGE 87

87 Figure 3 4 HR XTEM micrographs of a crystalline Ge sample B + implanted at 2 ke V to 5.010 15 cm 2 as implanted in c Ge showing: A ) a surface GeO x layer and a damaged layer ex tending 18 nm from surface and B ) amorphous pockets and lattice defects in close proximity to the surface.

PAGE 88

88 Figure 3 5 B concentration profiles of a pre am orphized Ge sample B + implanted at 2 keV to doses of 5.010 13 or 5.010 15 cm 2 as implanted (dashed line) and horizontal dotted lines indicate the estimated maximum active B concentration for both implant conditions.

PAGE 89

89 CHAPTE R 4 EFFECT OF IMPLANT EN ERGY ON B ACTIVATION 4.1 Low Energy B + Implantation in Ge This work is the first report of the anomalous activation behavior observed for shallow B + implants in c Ge and PA Ge which raises the question as to why it has not been observed previously. A minute body of research exists regarding the electrical behavior of shallow B + implants in Ge which may explain how this behavior has gone unnoticed In addition, the experiments that have been completed typically measured sheet resistance of a single sample or small sa mple set implanted to a high fluence which does not yield significant information regarding carrier activation. 34,38,40,45 The use of a single high fluence implant is capable of masking the incomplete activation observed for low energy implant s in this work by pushing the peak concentration well above electrical solubility values reported in the literature. To further understand the fluence independent activation behavior of B in Ge, a systematic study of increasing implant energy as a functi on of fluence is necessary. In this chapter the activation of low energy B + implantation into c Ge and PA Ge was studied using M4PP and MHE measurements. The effect of increasing B + implant energy and its effect of increasing point defect formation is s tudied. The consequence of increasing the distance between the surface and the B profile on electrical activation is also investigated. 4 .2 Experimental Details Czochralski grown n type (001) wafers (resistivity experiments. A set of c Ge samples was produced by B + implantation at 2 4, and 6 keV to doses of 5.010 13 5.010 15 cm 2 while a set of PA Ge samples was produced

PAGE 90

90 by first performing Ge + implantation at 120 keV to a dose o f 210 14 cm 2 before the same B + implantation step. Samples were annealed in N 2 60s to activate the implanted B. Sheet resistance, Hall sheet number ( n H ), and Hall mobility ( H ) were measured using a CAPRES microRSP M 150 M4PP with Au coated probes, a probe spacing of 20 number and mobility values were adjusted to obtain the carrier sheet number ( n s ) and drift mobility ( d ) by using a scattering factor ( r H ) of 1.21 as determined empirically by Mirabella et al for high dose B + implantation into Ge. 31 The carrier density and drift mobility are related to the Hall values by n s = n H r H and d = H / r H respectively. 4.3 Effect of Increased B + Implant Energy Figure 4 1 shows profiles for 2, 4, and 6 keV B + implants t o a fluence of 5.010 13 cm 2 as simulated using SRIM. 78 The profiles show that as B + implant energy is increased, the B profile is pushed further from the surface as well as broadened. In doing so, the peak B concentration is reduced. A fluence of 5.010 13 cm 2 was simulated to show that the majority of the B concentration is below the reported electrical solubility of B in Ge for the low fluence implants. The maximum reported electrical solubility of B in c Ge is approximately 2 .010 20 cm 3 as reported in the previous chapter. Figure 4 2 shows the sheet resistance ( R S ) for samples implanted at 2, 4, and 6 keV to B + fluences ranging from 5.010 13 to 5.010 15 cm 2 for 60s. In Figure 4 2(A) and 4 2 ( B ), it is evident that R S decreases with increasing fluence and energy for both c Ge (diamonds) and PA anneal, the minimum R S for PA Ge and c Ge,

PAGE 91

91 respectively. The lower R S values for PA Ge with respect to c Ge have been documented previously 6,15,20 and are due to increased B incorporation upon SPEG. The decrease in R S with increasing implant energy can be explained by the incr ease in the fraction of implanted B with a concentration below the solubility limit. SRIM simulations show that the projected range of the implant shifts from 9.5 to 24.0 nm as the implant energy is increased from 2 to 6 keV. In doing so, the B peak wide ns at higher energies and allows for a larger fraction of dopant activation for identical fluences. The B concentration profiles following B + implantation to a fluence of 5.010 1 3 cm 2 is shown in Figure 4 1 It should be noted that the fluence used for t his figure is only for representation of the effect of energy as similar behavior would be observed for increased or decreased B + fluence. In Figure 4 3 ( A ) and 4 3 ( B) sheet number (as adjusted using r H = 1.21) is plotted as a function of implant fluence. It is evident that the decrease in R S with increasing fluence is due to an increase in the overall number of active dopants. With increasing implant energy, a larger fraction of implanted dopant is incorporated into the lattice for reasons described in t he previous paragraph. For 5.010 15 B + cm 2 implants at 6 keV, the n s obtained was 7.410 14 and 2.210 15 cm 2 for c Ge and PA Ge, respectively. The corresponding activation value, defined as the ratio of sheet number divided by the residual implanted flu ence, was 18 and 52% for c Ge and PA Ge, respectively. These low activation values are not entirely surprising given the peak B concentration, which was simulated by SRIM to be approximately 1.410 21 cm 3 and thus is well above solubility values reported i n the literature. 24,31,34

PAGE 92

92 Figure 4 4 (A) and 4 4 (B), t he drift mobility ( as adjusted using r H = 1.21) is plotted as a function of B + implant fluence for both c Ge and PA Ge samples It can be seen that drift mobility is higher for c Ge with respect to PA Ge samples across the investigated range. In additio n, the data shows a trend of decreasing mobility with increasing B + fluence. Comparing this data with the sheet number values in Figure 4 3 (A) and 4 3 (B), these trends can be explained by the increase in the number of active dopants for these conditions as ionized carriers exhibit a stronger effect on carrier mobility than neutral dopant atoms. In conjunction with the anomalous activation behavior of ultra shallow B + implants in Ge that has been reported previously, 24 a large discrepancy between the implanted and active fluence exists for both c Ge and PA Ge. For the lowest B + fluence of 5.0 10 13 cm 2 implanted at 2 keV, the peak B concentration as simulated by SRIM 78 is expected to be only 3 .010 1 9 cm 3 which is lower than the reported solubility of B in both c Ge and PA Ge. 24 Despite the low concentration, only a small fraction of dopant is e similar results. 24 The decrease from complete activation is not due to any electrical solubility argument. The large concentration of inactive B observed in c Ge and PA Ge is well behaved across the investigated energy range and i s intriguing due to its fluence independent nature. The sheet number values obtained increase as a function of energy which suggests that inactivity may increase when the boron profile is located near the surface.

PAGE 93

93 With decreasing implant energy, the con centration prof ile shifts towards the surface as shown in Figure 4 1. This increases the concentration of B above the solubility limit and also reduces the activation of B. The simulated concentration profile and the measured sheet number were used to de termine the corresponding maximum active B concentration as described elsewhere. 34 For the highest fluence across all B + energies investigated, t he average a ctive concentration was determined to be 1.3 0.3 10 20 and 4.5 0.5 10 20 B cm 3 for c Ge and PA Ge and correlates well with values in the literature. 23,24,31,34 However, it needs to be stressed that the inactivity exhibited for these ultra shallow B + implants is not due to electrical solubility constraints. Completing the same procedure for the 5 .0 10 10 cm 2 fluence returns an electrical solubility value well below the l owest report ed electrical solubility values. This suggests that the activation behavior of these ultra shallow B + implants in Ge do not follow traditional electrical solubility rules in which all dopant above the solubility concentration is inactive while all dopants below the solubility concentration is electrically active. Rather, the observed behavior suggests that B activation is dependent on the local B concentration in which the electrical solubility limit mimics the chemical dopant profile decrease d by a factor less than unity. 4.4 Role of Implant Energy To investigate the effect of increasing implant energy the electrical activation values obtained by variable B energy implants into c Ge to a fluence of 5.010 1 4 cm w ere used. Data from Bruno et al. 23 and Suh et al. 36 which used 35 k eV and 60 keV B + implants, respectively, is used to gain further insight into the effect of implant energy and surface proximity. Indeed, increasing the

PAGE 94

94 implant energy while implanting a fixed fluence widens the B profile and decreases the peak concentrat ion. However, it has been shown previously in this work that the fraction of electrically active dopant atoms is approximately constant as a function of fluence and not a function of electrical solubility which is typically invoked for implants into Si. I n addition, the effect of varying annealing conditions is not expected to change activation values appreciably as will be discussed in the following chapter. The activation results as a function of variable B energy is plotted in Figure 4 5. It is evid ent that activation is observed to strongly increase for the ultra shallow implants completed in this work. However, upon fitting the low energy implants to the data obtained from other works it is evident that the increase in activation observed with in creasing energy begins to level off. Extrapolating the fit, complete activation is not expected until an implant energy of approximately 1 90 keV is used. It should also be noted that this fit predicts 0% activation as the B + energy approaches 0 keV. How ever, this finding is not entirely unreasonable if the surface truly is a sink for vacancies which would create the greatest interstitial super saturation as the B energy approaches 0 keV. Similar results of surface proximity have been reported by Priolo e t al. for low energy B + implants into Si. Figure 4 6 shows a representation of the percentage of electrically active B atoms as a function of the implant R P for a fixed fluence of 1 .010 1 3 cm It is evident that a sharp decrease in electrical activatio n is evident for low energy implants. The authors were able to show that high energy Si implants used to enrich the surface with vacancies were successful in increasing the activation of these low energy implants by a factor of 2.5 Successive high energ y Si + implants were

PAGE 95

95 successful in further increasing the B activation. Therefore, t his behavior of incomplete activation for low energy B + implants in Si was attributed to a decrease in the vacancy population in the near surface region which increased the formation of inactive B complexes. 137 In recent publications, it has been shown that the Ge surface may act as a sink for vacancies while reflecting interstitials. In work by Bracht et al. B doped layers grown by MBE showed no change as a function of proximity to the sur face during annealing. Simulations of the diffused B profile showed that a homogenous interstitial supersaturation existed. Conversely, P profiles showed a reduction of diffusion during proton irradiation. Considering that B and P diffuses through inter stitial and vacancy mechanisms, respectively, it can be concluded that the near surface region may be an interstitial rich environment following implantation. 50 Other work by Scapellato et al. have also concluded that the Ge surface does not allow for inte rstitial recombination through the use of diffusion studies of B doped layers grown by MBE. 54 In this work, B doped Ge layers were implanted with oxygen at 235 keV to a fluence of 4 .010 1 4 cm which has an R P of approximately 500 nm. Following annealing at 650 C for various times, it was shown that B profiles diffused homogenously for times less than and equal to120 min through a form of enhanced diffusion due to oxygen precipitation. From t hese results, it was suggested that the homogenous B diffusion further proves that the surface does not allow for interstitial recombination, but rather reflects them to the bulk. 54

PAGE 96

96 In addition to increasing the distance from the surface, increasing the B + implant energy also increases the number of vacancies created during implantation. As the B ion comes to rest, it loses its energy to the lattice through nuclear collisions which result in the creation of target vacancies and interstitials. If the B ion energy is increased, the amount of energy lost to nuclear collisions, and therefore the creation of point defects, increased. Figure 4 7 displays the num ber of vacancies created per incoming B ion for variable energy B + implantation in Ge as simulated by SRIM. 78 The observed increase in activation with increasing energy could also be attributed to the introduction of a larger vacancy population. For an off lattice B atom to become electrically active, it must recombine with a lattice vacancy. By increasing the implant energy, the number of vacancies created is increased which increases the probability that an off lattice B atom will react with a vacant. It should also be mentioned that the Ge surface is known to undergo a radical transformation into a nanoporous structure following high dose implantation. 138 142 This structure is attributed to a barrier between interstitial vacancy recombination and the formation of vacancy clusters that increase in size with fluence. 141,143 Although high dose Ge + implants were not used in this work, the effects of this barrier to interstitial vacancy recombination may still be affecting the activation results. It has been shown that Ge + implants to a fluence of 2 .010 15 cm at energies from 30 to 120 keV were not observed to c reate the nanoporous structure, but SPEG was slowed near the surface. 144 The reduction of SPEG veloc ity was increased as the energy of the implant was reduced Despite the lack of porous formation, the reduction of SPEG velocity was

PAGE 97

97 attributed to the presence of sub microscopic voids formed due to the barrier of interstitial vacancy recombination in the near surface region. 144 By increasing the implant energy, the effect of surface proximity is reduced ; however, the number of vacancies created is also increased Both of these properties may explain the increase in activation observed in this work for B + implants at an energy between 2 and 6 keV. Although clustering has been previously observed for 35 keV B + implants in Ge in which approximately 50% of implanted B was rendered active, 37,39 the quantity of inactive B is much more significant for ultra shallow implants as exhibited in this work. The work presented here is the first systematic report of ultra shallow B activation in Ge which may explain why this phenomen on has not been observed previously. The results suggest that the role of implant energy has a significant effect on the electrical activation of shallow B + implants in Ge. 4 5 Summary In this chapter, it has been shown that moderate increases in the per centage of activated dopants occurs with increasing B + implant energy. However, as was observed in the previous chapter, for a fixed implant energy, activation remains relatively independent of implanted B + fluence. The surface proximity and associated point defect populations was speculated to have an impact on B activation as increasing implant energy deepens the B profile from the surface For a B + implant to a fluence of 5.010 1 3 cm the percent of electrically active dopants as determined through micro Hall effect was found to be 1 0.1, 18.5 and 2 5 1% at an energy of 2, 4, and 6 keV, respectively. Given these implant conditions, the peak B concentration investigated is approximately 3 .010 19 cm for the 2 keV implant and decreases with increasing implant energy. Full activation would be

PAGE 98

98 expected for all of the investigated energies. As such, it has been determined that the effect of increasing activation with increasing implant energy is due to the creation of a larger vacancy population as well as the B profile being further removed from the surface and thereby reducing its effects on vacancy annihilation

PAGE 99

99 Figure 4 1. SRIM simulations depicting the shift in the B concentration profile produced with increasing implant energy. 78

PAGE 100

100 Figure 4 2. Measured sheet resistance of samples B + implanted at 2, 4, and 6 keV to fluences ranging from 5.010 13 to 5.010 15 cm in A) c Ge and B) PA Ge

PAGE 101

101 Figure 4 3. Measured sheet number of samples B + implanted at 2, 4, and 6 keV to fluences ranging from 5.010 13 to 5.010 15 cm T he dotted line represents complete ac tivation for A) c Ge and B) PA Ge

PAGE 102

102 s

PAGE 103

103 Figure 4 4. Measured drift mobility of samples B + implanted at 2, 4, and 6 keV to fluences ranging from 5.010 13 to 5.010 15 cm T he dotted line represents complete activation A) c Ge an d B) PA Ge Figure 4 5. Measured percentage of electrically active B at variable implant en ergies A 35 keV implant corresponds to an R P of 90 nm. T he crossed and dotted data point s are from Bruno et. al. 23 and Suh et al. 36 respectively.

PAGE 104

104 Figure 4 6 Measured percentage of electrically active B in Si at variable implant energies. Data is adapted from Priolo et al. 137

PAGE 105

105 Figure 4 7 Vacancies created per incoming B ion as a function of implant energy as simulated by SRIM. 78

PAGE 106

106 CHAPTER 5 THERMAL STABILITY OF BORON ACTIVATION IN GERMANIUM 5 .1 Activation Stability of B in Ge In recent years, se veral studies have investigated the electrical behavior of ion implanted B in both crystalline (c Ge) and preamorphized (PA Ge) Ge. 23,31,35 37,145,146 Similar to Si it has been shown that preamorphization increases dopant activation during the solid phase epitaxial growth (SPEG) process. 35 However, the majority of the experiments published in the literature have used high energy B + implants that are not directly relevant for ultra shallow junctions. Limited dat a exists regarding the thermal stability of B activation in Ge which can be explained by the adequate stability evidenced in the available reports. Impellizzeri et al. has shown that 35 keV B + implants exhibit stable activation following a modest activati on anneal. They have shown that of boron activation 31 37 Bruno et al. has shown similar results for c Ge and PA Ge samples in which activation remains stable for anneals up to 550 C for 1h. However, although not explicitly mentioned in this work a qualitative decrease in R S is evident for c Ge samples while a similar increase is evident for PA Ge samples. 23 Similarly, Simoen et al. has shown 4.5 keV B + implantation to a fluence of 1 .010 1 5 cm 2 and 1 .010 1 6 cm 2 into c Ge returns stable R S values following RTA anneals between 400 and 600 C. 40 Satta et al. reported similar results for 6 keV B + implantation to a fl uence of 3 .010 1 5 cm 2 in which R S remains stable for RTA anneals up to 600 C. 38 Similar to the work by Bruno, 23 the results shown by Simoen 40 and

PAGE 107

107 Satta 38 et al. displayed a decrease in R S with increasing annealing temperature for c Ge samples. Previously, i t has been shown that the activation of 2 keV B + implants Ge has an anomalous activation b ehavior which is characterized by an incomplete activation independent of implanted fluence for both c Ge and PA Ge. The behavior is believed to be due to a B Ge cluster formation which renders a large fraction of the implanted fluence inactive. 24 Although far less pronounced, the presence of B Ge clusters has been reported previously, but has only been observed for implants into c Ge. 34,37 39 For Si, the formation and evolution of boron interstitial clusters is well characterized and understood, 56,59,70 but to date, a comprehensive study has not been completed for B + implants in Ge. In this work, a systematic study of the effect of isochronal annealing on the electrical activation of ultra shallow B + implants in Ge is presented. High temperature anneals are used to determine if the activation is stable as temperatures approach the melting temperature of Ge. Transmission electron microscopy and secondary ion mass spectroscopy (SIMS) is used to fur ther explain the electrical behavior observed upon annealing. 5 .2 Experimental Details Experiments were performed on Czochralski grown n type Ge (001) wafers with cm. Samples were B + implanted at 2, 4, and 6 keV with fluences ranging from 5.010 13 to 5.010 15 cm 2 An identical set of PA Ge samples was produced by first implanting a Ge + fluence of 2.010 14 cm 2 at 120 keV prior to B + implantation to produce an a morphized surface layer to a depth of 100 nm as verified by

PAGE 108

108 high resolution cross sectional transmission electron microscopy (HR XTEM). The beam current was fixed at 1.1 mA for all B + Samples were processed in a Heatpulse 4100 rapid thermal annealer (RTA) in an N 2 ambient at 400 Select samples were isothermally annealed at 400 C for times ranging fr om 30s to 5 h. Additional samples were underwent high temperature isochronal anneals. Micro Hall effect measurements were used for their ability to accurately measure the electrical properties of ultra shallow junctions. 12 6,147,148 Micro Hall effect characterization was completed using a CAPRES microRSP M 150 M4PP fitted with Au flux density of 0.475 T. Hall sheet number ( n H ) and mobility va lues ( H ) were adjusted to obtain the carrier sheet number ( n s ) and drift mobility ( d ) by using a scattering factor ( r H ) of 1.21 as determined empirically. 31 The carrier density and drift mobility are related to the Hall values by n s = n H r H and d = H / r H respectively. 5.3 Isothermal Annealing For this experiment, a small sample subset was implanted at 2 keV to a fluence of 1 .010 15 cm 2 into c Ge and PA Ge. The electrical behavior of these samples was monitored as a function of annealing at 400 C for times ranging from 30s to 5h. For short annealing times ( 960s), samples were annealed using an RTA while longer anneals were completed in a tube furnace. Both annealing environments utilized flowing N 2 Samples were processed concurrently mea ning that an RTA sample annealed for 960s received all ramp up and ramp down cycles accumulated during all shorter annealing durations. RBS characterization was completed with a 2 MeV He +

PAGE 109

109 beam at normal incidence was used to monitor the number of displace d Ge atoms as a function of annealing time. Figure 5 1 shows the sheets resistance values obtained for isothermal annealing sequence at 400 C for B + implants at 2 keV to a fluence of 1 .010 15 cm 2 In Figure 5 1 (A), the data for c Ge samples is displayed. It is evident that sheet resistance decreases for times less than 120s and increases thereafter. The changes in sheet resistance are sl ight as the minimum R S measured was 430.8 and 473.9 /sq, respectively for 120s and 5h anneals. Figure 5 1 (B) shows the sheet resistance values obtained for PA Ge samples. It should be noted that 60s at 400 C is the minimum time necessary to complete th e SPEG process which explains why data for the PA Ge series does not begin prior to this annealing time. In stark contrast to what is observed for c Ge, PA Ge samples show minimal change as a function of annealing time. To provide further evidence that the R S values do in fact decrease for short annealing times, a line scan comprising 40 individual measurements as a function of distance were completed on the sample annealed for 30s as shown in Figure 5 2. It is evident that the deviation from the mean i s slight as the standard deviation is 1.76 /sq across the entire 4 mm scan. From this line scan data, it can be assumed that every data point is reproducible and the trends observed are accurate. Figure 5 3 displays the sheet number values obtained fol lowing the isothermal annealing sequence. In Figure 5 3 (A), it is clear that c Ge samples show a strong decrease in active carriers with increasing annealing time. Conversely, Figure 5 3 (B) shows little change in active carriers for PA Ge. It should b e stated that the observed deactivation for c Ge samples is significant in that slightly over 2 0 % of the active

PAGE 110

110 dopants become electrically inactive with further annealing; however, the overall fraction of active dopants remains near 10% for all annealing times investigated. The deactivation observed is not believed to be due to any sort of chemical dose loss as the PA Ge samples were processed identically and did not exhibit any significant dose loss. Comparing the R S and sheet number data, it becomes c lear that the increase in R S for c Ge samples annealed longer than 120s is due to a deactivation of dopant atoms. The decrease in R S observed for short annealing times is believed to be due to annealing of implant damage which allows for an increase in t he drift mobility. Figure 5 4 shows the drift mobility for c Ge samples annealed for times ranging from 30 to 1920s. It is clear that a quick uptick in drift mobility is evident for short annealing times which corresponds with removal of implant damage After this increase, the drift mobility values begin to stagnate. The removal of implant damage and subsequent increase in drift mobility coupled with the deactivation of B atoms explains the R S trends observed in Figure 5 1 (A). Similar to the fluence independent activation documented in previous chapters, t he deactivation observed for c Ge samples with increasing annealing times is believed to be due to the large interstitial population that exists following ion implantation which may spur the for mati on of an inactive cluster. The displaced Ge population can be measured using RBS which may provide insight into the deactivation behavior of the c Ge samples. Figure 5 5 (A) shows the RBS spectra acquired for c Ge samples B + implanted at 2 keV to a fluen ce of 1 .010 15 cm 2 following furnace annealing for 1, 3, and 5h. A virgin Ge sample is shown as well for comparison. Figure 5 5 (B) highlights the surface

PAGE 111

111 region of the investigated samples. The area under the surface peak displayed in Figure 5 5 (B) is directly proportional to the displaced Ge concentration. From this characterization, it is evident that the number of displaced Ge atoms does not change appreciably with annealing de spite the deactivation observed as the surface peak s are nearly overlaid. It is, however, evident that there is a population of Ge atoms that are displaced from their lattice sites as the peaks are larger than that of a virgin crystal. The number of displaced Ge atoms were calculated and tabulated in Tabl e 5 1 in conjunction with the inactive number of B atoms as determined from the micro Hall effect characterization. It is clear that the number of displaced Ge atoms remains relatively constant with annealing time across the investigated range. Following a 1h anneal, 4.9 3 10 15 Ge/cm 2 atoms are displaced in comparison to 4. 56 10 15 Ge/cm 2 following a 5h anneal. To gain perspective on the deactivation behavior, the ratio of displaced Ge atoms to electrically inactive B atoms (displaced Ge:inactive B) was de termined and found to be approximately 6.66. The Ge:B ratio is relatively constant across the annealing range which thereby fails to yield information regarding the deactivation observed for c Ge samples. It should be noted that the RBS technique is una ble to determine the confirmation of displaced Ge; whether it is situated as a unique point defect, an interstitial cluster, or an inactive Ge:B complex among other possibilities. This shortcoming of RBS limits the characterization ability of inactive B i n Ge However, the data obtained shows that a substantial fraction of Ge is situated off lattice despite a

PAGE 112

112 interstitial Ge population may be a driving force in the inactivity of ultra shallow B in Ge. 5. 4 Ac tivation Thermal Stability Between 400 600 C Figure 5 6 shows the sheet resistance values obtained for 2 keV c Ge (Figure 5 6 ( A )) and PA Ge (Figure 5 6 ( B 60s. It is evident that large changes in R S exi st for samples implanted at a reduced fluence while minimal changes exist for those samples implanted at an increased fluence. The trends hold true for both c Ge and PA Ge, but the R S data trends in opposite directions for both c Ge and PA Ge. For c Ge s amples, with increasing annealing temperature, a reduction in R S is observed while for PA Ge samples, an increase in R S is evident. Figure 5 7 for samples implanted at 2, 4, and 6 keV Rather than presenting all measured data, the relative change in R S was used to highlight the trend observed for all implant energ ies while maintaining a concise plot. Interestingly, with increasing annealing temperature, it was apparent that R S decreased for all c Ge implant conditions and increased fo r all PA Ge implant conditions. As implant fluence was decreased, the relative c hanges in R S became more prominent for both c Ge and PA Ge. For the lowest influence implanted at 2 keV in c Ge, R S decreased 33.70% while for the highest fluence R S decreased only 4.86%. A trend of increasing R S w as observed between the lowest and highe st implanted fluences for PA Ge as well. Bruno et al. reported the thermal stability of high energy B activation following 35 23 The data appears to follow a similar trend to what is observed for ultra shallow implants in this

PAGE 113

113 work in which R S decreases for c Ge and increases for PA Ge. How ever, it appears to occur to a lesser extent which is not surprising as the data presented in this work suggests that the trend decreases with implant energy. For example, for a B fluence of 5.010 13 cm 2 into c Ge, the decrease in R S is 34% and 26% for 2 keV and 6 keV, respectively. The trends in R S behavior are intriguing and can be explained by examining the components of the resistance term itself. Figure 5 8 shows the thermal evolution of active carriers and drift mobility for a B + implant to a flue nce of 5.010 15 cm 2 into c Ge and PA no significant change in activation was observed across the investigated temperature range for both c Ge and PA Ge samples. Previous reports ha ve shown that the activation of B in Ge is remarkably stable. 23,34,36,40 However, Pancier a et al. have reported that the dopant defect interactions involved with end of range dissolution has an effect on activation values. 146 The observed changes were slight (approximately 10% change in activation/ deactivation) and suggests that dopant defect interactions in Ge behave much differently from that which has been extensively studied in Si in which large fluctuations in activation are observed upon annealing. 149 In Figure 5 8 (A ), n s is observed to increase with increasing annealing temperature for PA Ge samples; conversely, n s values slightly decreased for c Ge samples. Similar to the work by Panciera et al., 146 the observed changes in activation are subtle and do not have significant effect on the overall activation value. The changes in activation for conditi ons investigated were on the order of 10%. A signif icant

PAGE 114

114 anneal for 60s. These results suggest that the B Ge cluster responsible for the B inactivity in both c Ge and PA Ge is stable at elevated temperatures. Drift m obility exhibited a contrasting behavior in that values decreased for PA Ge and increased for c Ge samples with increasin g temperature as shown in Figure 5 8 (B ). The drift mobility decreased with increasing implant energy and annealing temperature for PA G e samples which can be explained by the increase in the number of active dopants. It is known that the impact of ionized dopants on mobility is much more significant than that of neutral dopants due to the effects of coulombic scattering. 2 The increase in D for c Ge is explained by the reduction of microstructural damage and subsequent redu ction in scatter ing centers as evidenced in Figure 5 9 It should be mentioned that the increase in D for c Ge is not as significant as expected based on reported mobility models for high B concentrations in Ge. 31 This results suggests that some other mobility degradation mechanism is at work for c Ge samples. The culprit is likely the presence of B Ge clusters which scatter carriers and reduce their effective drift mobility. The se mobility trends give further evidence that the inactive B is not in the lattice as a free interstitial, but rather likely in a B Ge complex. Fig ure 5 9 shows the microstructure of samples B + implanted at 2 keV to a fluence of 5.010 15 cm 2 into c Ge and PA Ge after annealing for 400 is well known that B + implants into c Ge are characterized by a defective microstructure that is centered near the projected range ( R P ) of the implant. 34,37,40,45 In the case of c Ge, the layer is not characterized by discernible extended defects, but rather a highly defective microstructure distinguished by inho mogeneous contrast which diminished with increasing annealing temperature. The inability to observe unique defects may be

PAGE 115

115 due to several factors. The samples may not have been subjected to a sufficient thermal budget to allow the formation of extended de fects or if already formed, they may be very small and high in concentration which limits the ability to view individual defects. 150 Further work was completed regarding B + implant related defects in c Ge and will be discussed in later a chapter In the case of PA Ge, the i Ge) layer was approximately anneal for 60s as shown in Figure 5 9 ( D ) In addition, no implant related defects were found for any annealing condition of PA Ge. However, the formation of extended defects resulting from the SPEG process is not expected for low Ge + implant fluences and is not 34,38,45,151 The observed activation behavior in both c Ge and PA Ge is certainly unique and a far departure from what has been observed previously for B + implants in Si. The ultra shallow nature of the implants in this work suggests that there may be a correlation with surface proximity but the effect may not have been captured within the investigated B + energy regime The Hall effect results by Mirabella et al. for 35 keV B + implantation which report full activation following a modest 360 C for 1h further suggests that the cluster behavior observed is a function of implant energy. It has been suggested that there is a barrier to point defect recombination at the Ge surface which has been shown to spur the formation of a nanoporous structu re. 141 However, the clustering behavior was prominent also for samples implanted at 6 keV and any correlation between surface proximity and clustering that exists was not apparent in this work.

PAGE 116

116 5.5 High Temperature Anneals The large inactive fraction of B following annealing at 600 C for 60s for nearly 2/3 of the melting temperature of G e in both c Ge and PA Ge samples is intriguing. The changes in electrical values can be deemed significant when comparing relative values between anneals, but on the whole, the fraction of dopant rendered electrically active is relatively minute. It woul d be assumed that any inactive defect complex that may exist would be able to be dissolved into its constituent components, in this case presumably B and Ge, given a heat treatment high enough in temperature. To determine if a high temperature heat treatm ent is capable of dissolving the dopant defect cluster, high temperature isochronal anneals at 10s were completed on samples implanted at 6 keV 5.010 1 4 cm 2 into c Ge and PA Ge. The samples were capped with 100 nm of SiO 2 prior to the heat treatment to r esist the loss of the Ge surface and implanted dopants. 41,152,153 HR XTEM was completed using a JEOL 2010F to image the microstructure of specimens before and after annealing. SIMS was used to verify the chemical profile following annealing. TEM samples were prep ared using a FEI DB235 focused ion beam. Figure 5 10 displays the sheet resistance of c Ge and PA Ge samples implanted at 6 keV to a fluence of 5.010 15 cm 2 and were subjected to anneals for temperatures ranging from 400 to 850 C. R S is relatively stable for temperatures below Figure 5 11 shows the she et number of c Ge and PA Ge samples implanted at 6 keV to a fluence of 5.010 15 cm 2 and were subjected to anneals for temperatures ranging from 400 to 850 C. Similar to the R S data, the sheet number is relatively stable

PAGE 117

117 at low temperatures while signific ant changes occur at increased temperature. For PA Ge samples, anneals in excess of 650 C produced a dramatic decrease in sheet number while for c Ge the decrease is less prevalent and occurs for temperatures in excess of 750 C. 5 .5.1 B Concentration Pr ofile Following High T Annealing To eliminate the possibility of surface or dopant loss, s econdary ion mass spectroscopy (SIMS) was used to verify that the B profile has not changed significantly during heat treatment to allow for direct comparison of elec trical data across all anneals. Figure 5 12 displays the B concentration profiles as obtained from SIMS for samples as implanted, annealed at 650 C for 10s and 800 C for 10s following removal of the SiO 2 and electrical characterization. The annealed profiles are characterized by an anomalous hump extending from the surface to near the projected range of the implant. The behavior is more pronounce d for increasing annealing temperature. The channeling tail of the B profile is similar for all investigated samples. Similar behavior was obtained for PA Ge samples as shown in Figure 5 13 The anomalous SIMS data is explained by the presence of dispe rse pits distributed across the surface of the samples which interferes with the sputtering process during characterization SIMS characterization relies on a standard rate of material removal as a function of time in order to effectively produce the atom ic concentrations present in the material of interest. The presence of an interface or surface inhomogeneities make it difficult for sputtering to occur uniformly as a function of area which translates to non conformities as a function of depth. 118 The pits are explained through the loss of the Ge surface through the formation of GeO and subsequent desorption of GeO. 41,152,153 However, it has been observed that the

PAGE 118

118 surface lo ss occurs only from material desorb from pits and is not a planar process. Figure 5 14 shows a HR XTEM image of a pit formed during thermal processing of Ge prior to removal of the SiO 2 It is evident that the native oxide of Ge still remains which shows that the initial surface is not affected by surface loss in areas where pitting does not occur. The planar area typically measured during SIMS characterization is relatively large and is typically on the order 0.1 cm 2 114 The relatively large characterization volume lends SIMS towards being more of a sampling t echnique representative of the bulk rather than local concentration differences. An obvious benefit of using scaled down metrology tools is that the sensitivity is much greater than conventional tools and enables the characterization of precise locations 106,108,154 The electrical measurements on these high temperature samples were completed in regions that were devoid of pits and were optically pristine. To further confirm that this is true, measurements taken in locations that have even minor surface scratches y ield inconsistent results. In addition, even activation fluctuations created by inhomogeneous laser annealing is easily detected by MHE characterization. 104 If the surface was lost due to GeO desorption, a lateral shift in the B concentration profile would be expected. Published results concerning surface loss from Ge desorption confirm this to be true as well. 152 However, this is not the case for these samples. The profile tail of the as impla nt ed and annealed samples are well aligned. If the surface was lost during GeO desorption, the concentration in the tail region of the profiles would not coincide with each other as is shown in Figure 5 15 In

PAGE 119

119 this figure from Oh et al. it is evident th at surface desorption introduces a significant horizontal shift in the chemical profile following surface desorption of uncapped samples. 152 This shift is not evident in the SIMS profiles displayed in Figure 5 12 and Figure 5 13. The SIMS characterization is unable to show that the peak of the B profile is unchanged, but the data obtained from the tail of the profile proves that B is still located in the substrate at approximately the same depth as expected. The lack of diffusion is in agreement with published values regarding B diffusion in Ge following ion implantation. 28 3 0 5 .5.2 Reduction of Active Carriers Following High T Annealing The equilibrium solid solubility of B in Ge at 800 C is considered to be approximately 5.510 18 cm 3 as measured from high temperature diffusion studies. In these studies the diffused part of the B profile was considered to be below solubility while the non diffused B was considered to be above solubility. 28 30 In addition, phase diagram determinations have shown the solubility of B in Ge to be negligible. 26,27 The analysis of the n S data shows that the increase in R S is due to a reduction in the number of active carriers with increasing temperature. The reduction of carriers can be explained by the metastable electr ical activation created during the ion implantation process. Upon high temperature annealing, the B atoms fall out of solid solution with Ge as it approaches its equilibrium state. 5.5. 3 Significance of the Lack of B Diffusion It is well known that B di ffuses through an interstitial mediated mechanism and is largely controlled by the large formation energy of Ge interstitials. 30,50 Under equilibrium conditions B does not readily diffuse 28 30 but diffusion of B in Ge has been observed when an excess interstitial concentration is present. 48,50,54 The electrical and NRA

PAGE 120

120 characterization completed in prior chapters as well as this chapter confirms that a large fraction of implanted B is located in off lattice positions following annealing. RBS characteriza tion has also shown that a large population of displaced Ge also remains following annealing. However, it is not known whether B atoms can be situated as interstitial B or in a B Ge cluster. The lack of diffusion is certainly a note worthy observation t hat may shed light on the confirmation of off lattice B atoms. Given the high residual interstitial Ge concentration, it could be expected that B diffusion would be observed following an anneal at 800 C for 10s as these interstitials should be mobile given this thermal budget. Transient enhanced diffusion of B in Ge has been documented following a 380 C anneal for 1h in which end of range damage was shown to dissolve thereby releasing interstitials t o induce B diffusion. 48 However, for this work, no diffusion was observed. For boron in Si, it has been observed that the fraction of the B atoms in an inactive complex with Si do not diffuse. 155 Similar behavior could be expected for B in Ge. It can be interpreted that the off lattice B and Ge concentrations as measured by Hall and ion beam analysis techniques are not free interstitials, but rather arranged in a thermally stable B Ge complex. Bisognin et al. and Impellizzeri et al. have repor ted the formation of a B Ge cluster in the ratio of 8:1 (Ge:B) following a 35 keV B + implant and annealing at 360 C for 1h. 37,39 From this information, it can be assumed that the off lattice B and Ge populations are in an electrically inactive and thermally stable complex which explains the lack of diffusion of the ch emical profile observed following high temperature annealing.

PAGE 121

121 5 .6 Theory for B Inactivity The observed activation behavior is unique in that is independent of implanted dose The Hall and channeling analyses data in the previous chapter suggests that a small fraction of the implanted B is located substit ut ionally while the majority of the dopant is located in off lattice positions. Presumably, the dopant is bound to an electrically inactive complex ; notably the formation of a B Ge cluster as has been d ocumented for 35 keV B + implants in Ge 23,37,39 In addition, the minimal change observed with increasing annealing temperature suggests that the inactive complex is thermally sta bl e during subsequent anneals. For PA Ge samples, a modest 60s anneal at 400 C is necessary to complete SPEG and activate the B. However, annealing at increased temperature has not shown any significant change in activation for PA Ge or c Ge. It is not clear whether an inactive B Ge cluster forms during the implantation or duri ng the subsequent annealing step The formation of the well studied B Si complexes occur during the post implantation annealing step. 156 158 For this work the pla ten is held at room temperature during implantation to reduce sample heating but the temperature of the sample specifically the sample surface, is not directly h eld at a set temperature Recent works by Lopez et al. have shown that B + implantation in Ge can induce thermal spikes, or a localized volume near the ion cascade of increased temperature, due to the reduced melting temperature and increased thermal condu ctivity of Ge with respect to Si. 159 These thermal spikes may allow for the formation of the inactive complex during the implantation step. Although to a lesser extent, thermal spikes do occur in Si as well during implant ation, 159,160 but the formation of B Si clusters are not observed until post implantation annealing. If the inactive

PAGE 122

122 complex is formed during post implantation annealing as in Si, the B and point defect diffusivities are impor tant parameters to hypot hesize the formation mechanism of the inactive complex. The diffusivity of atoms or point defects through a solid can be described by the Arrhenius relationship as follows: ( 5 1) where D is the diffusivity, D 0 is the dif fusion coefficient, E A is the activation energy for the diffusion process, k T is temperature. The diffusivity of B in Ge has been reported to be characterized with a pre exponential term of 19 m 2 / s 1 and an activation energy of 4.65 0.3 eV. 30 Similar values have been observed in other works. 28,29 The mechanism for B diffusion is reported to be interstitial mediated 30,50,161 which is speculated to be limited by the formation of Ge interstitials which is reported to have a formation energy of 3.5 eV. 162,163 Data r egarding the diffusivity and equilibrium concentration of self interstitials in Ge i s sparse and very little information is presently known. However, metal tracer diffusion experiments have shown that the vacancy diffusivity and concentration product is m uch greater than that of the interstitial. 164,165 Other reports have documented similar values for the diffusion of Ge self interstitials. 166 168 The calculation of the diffusion length of B atoms for the lowest thermal budget investigated of 400 C for 60s y ields a value of 2.05 10 7 nm The diffusion length of B atoms is minimal and can be considered immobile for the lowest thermal budget investigated in this work.

PAGE 123

123 However, the diffusivity of vacancie s in Ge is quite significant given the investigated the rmal budget. The diffusivity of vacancies in Ge in the temperature range from 650 C to 900 C has been reported to be characterized with a pre exponential term of 2.40 10 3 m 2 / s 1 and an activation energy of 2.65 0.11 eV. Using this Arrhenius relationship, t he corresponding diffusion length of vacancies following 400 C for 60s is 28. 67 nm. The calculated diffusion length of vacancies in Ge is several orders of magnitude larger than that of B or self interstitials. From this information, it can be deduced that vacancies are the primary diffuser which would require vacancies to diffus e to interstitial B atoms to render them substitutional. Using SRIM simulations, a 2 keV B + implant to a fluence of 5.0 10 13 B + /cm 2 in Ge creates a self interstitial population, which to a rough approximation is 5 0 times larger than the number of implan ted B atoms. 78 It should be noted that the defect populations created by SRIM do not take into account any dynamic point defect recombination during the implant process. The number of interstitials and vacancies created during the ion implantation process creates an excess concentration that is in excess of equilibrium values which creates a driving force for point defect recombination. Using the assumptions outlined above, for a n implanted B atom to become substitutional on a lattice site, a vac ancy must travel to the B site due to the inability for B t o diffuse through the lattice satisfying the forward direction of this reaction: ( 5 2 ) where B i is an interstitial B atom, V is the mobile vacancy and B S is the substitutional B atom. Likewise, a vacancy must also travel to the self interstitial to render it substitutional. The reaction can be described by:

PAGE 124

124 ( 5 3 ) where Ge i is an interstitial Ge atom and Ge S is the substitutional Ge atom. It is theorized that the excess number of Ge i created during implantation with respect to B i drives the reaction depicted in Equation 5 3 forward while simu ltaneously restricting the number of vacancies available for recombination with B i atoms. Therefore, the reaction rate described by Equation 5 2 would decrease with an increased interstitial population. For a single mobile vacancy diffusing the lattice, the probability of interacting with an implanted B atom is only a fraction of the probability that it will react with a self interstitial due to the far greater self interstitial population It is this low probability of B i V interaction that is theorize d to restrict the activation of ultra shallow ion implanted B in Ge. It has been observed that increasing the B + implant energy results in an increase in the number of active or substitutional B atoms. R ecent research has shown that the surface is an eff icient sink for vacancies while reflecting interstitials 50,54 For near surface implants, this boundary condition would further reduce the number of vacancies available thereby reducing the forward reaction of Equation 5 2. In addition, it should allow for a supersaturation of Ge i following implantation which would increase the forward reaction depicted in Equation 5 3. Also, as discussed in the previous chapter, reducing the B + implant energy effectively reduces the numbe r of vacancies created which should have a direct correlation on the fraction of B atoms activated during annealing. However, the mere creation of vacancies should not be sufficient enough to describe the activation behavior as ion implantation is a very dynamic process. Indeed, increasing the implant energy

PAGE 125

125 introduces additional point defects into the lattice, but it also broadens the implant damage affected volume. Even though increasing the implant energy introduces a larger number of vacancies, it al so increases the volume over which they are distributed thereby possibly reducing the probability of B i V interaction. The presence of a vacancy sink, such as the surface, or a vacancy clustering mechanism as discussed in the previous chapter further comp licates assumptions regarding vacancy populations. The most significant vacancy population would be located in close proximity to the B ion to reduce these effects. Lopez et al. has used molecular dynamics to simulate the point defect environment surrou nding a B ion following implantation at 2 keV. 169 Figure 5 16 represents the vacancy population as a function of radial distance from the implanted B ion from this work. In Figure 5 16 (B), it is evident that the vacancy population sharply increases as a function of distance until approximately 6.0 nm is reached. The interstitial distribution closely mimicked the vacancy profile. The increase in vacancy population can be explained by the in crease in the volume as the sampling radius is increased as where V is the volume and r is the sampling radius. Increasing the implant energy would certainly increase the cumulative number of vacancies created, but it would also spread the vacancies over a larger volume thereby which would decrea se B i V interaction. The increase in activation observed with increasing B + energy can be ascribed to the introduction of a larger vacancy population; however, it should not be assumed that the mere number of vacancies created should be directly correlate d to the active fraction of B atoms. The quantity of vacancies created does not

PAGE 126

126 take into account the effect of the surface proximity or the vacancy distribution a nd their distance to the B atom which are both important factors to consider. 5.7 Simulati on of Activation Behavior To further test the preceding theory, the use of kinetic Monte Carlo (kMC) simulations were employed. The kMC simulations are capable of calculating the probability that certain atomic steps, for example a B atom migrating to a v acant site, will occur during processing given certain a time and temperature. For these simulations, the point defect equilibrium concentration and diffusivities listed in the previous section were used in addition to the B diffusivity. The vacancy and interstitial populations and distributions created during implantation were simulated using a Monte Carlo based process simulator and input into the kMC simulation as well. The dopant defect reactions investigated in this work were B i + V B S B + I B i I + V null and V + V V 2 The formation of V 2 were also allowed to interact with free vacancies to form larger V clusters. The formation of these larger V clusters reduces the numbers of V B i to render B atoms electrica lly active. The activation of a B + implant at an energy of 2 keV to a fluence of 5.0 10 1 3 and 5.0 10 1 5 cm 2 following was simulated in this work. These two fluences were chosen as they represent the highest and lowest val ues investigated in this work. Figure 5 17 displays the kMC simulations for the samples implanted to a fluence 5.0 10 1 3 and 5.0 10 1 5 cm 2 In Figure 5 17 (A) and Figure 5 17 (B), it can be observed that the simulation s predict an active concentration that is approximately 10% of the total B concentration profile. These results are in excellent agreement with the Hall results presented in this chapter as well as preceding chapters which have shown that the active fract ion is approximately 10% for a 2 keV B +

PAGE 127

127 implant independent of the implanted fluence. By comparing the two investigated fluences, it can be observed that the simulations also predict that a single electrical solubility limit does not exist; rather, the ac tive profile mimics the total B concentration profile. This behavior is in agreement with experimental work reported by Suh et al. in which the active profile mimics the chemical profile further suggesting that a single electrical solubility limit does no t exist for B in Ge. 36 To simulate the thermal stability of the B activation expiermentally observed in this word, simulations were completed as a function of annealing time. Figure 5 18 displays the kMC simulations for the samples implanted to a flue nce of 5.0 10 1 3 cm 2 following In comparing the two simulations, it is evident that prediction is in excellent agreement with the experimental data presen ted earlier in this chapter in which the active fraction is approximately 10% independent of implant fluence or thermal budget for B + implants in c Ge. It should be mentioned that the observed activation behavior could only be simulated by allowing the V + V V 2 defect reaction and thus invoking a mechanism which consumes free vacancies. When the simulations were completed in the absence of these reactions, the predicted active fractions were much higher; near 60% rather than 10%. The simulations provide further evidence that the activation of B in Ge is driven by the availability of vacant sites to accommodate an interstitial B atom. C orrelation between electrical data and the present theory lies in the fact that the creation of point defects would sca le proportionally with the implanted B + fluence. Evidence of the dose independent B inactivity can be observed in Figure 3 3 (B) The

PAGE 128

128 number of vacancies created would also increase with increasing energy which may explain the increased activation. The observed electrical and surrounding theory is a far departure from what is typically observed for B activation in Si; however, the equilibrium point defect concentrations and diffusivities are also much different in that the vacancy is the dominant point d efect in Ge rather than the interstitial as it is in Si. 164,165 In effect, these differences give rise to the B activation differen ces present between Si and Ge. 5 8 Summa ry The electrical activation of ultra shallow B + implants in c Ge and PA Ge was investigated using micro Hall effect techniques following isothermal anneals at 400 C and isochronal anneals between 400 and 85 Following isothermal annealing, it has been shown that a deactivation occurs for c Ge samples while PA Ge samples remain relatively constant. The origins of this deactivation is not clear, but could be assumed to be from the interstitial saturation remaining following implantation. It has also be en shown that systematic changes in R S exist upon annealing for both c Ge (decrease with temperature) and PA Ge (increase with temperature) samples. Corresponding changes in sheet number and drift mobility have been observed to be the source of the systema tic changes of R S However, the changes in sheet number were not significant for temperature below 650 for PA Ge and below 750 for c Ge Above these temperatures, a significant decrease in activation was observed which was attributed to the dopant at oms falling out of solution with Ge as predicted by the equilibrium solubility. A large fraction of implanted dopant was electrically inactive for all investigated conditions which suggests the pr esence of an electrically inactive and thermally stable B G e complex. The lack of diffusivity observed following an anneal at 8 0 0 C for 10s

PAGE 129

129 further suggests that the off lattice B atoms are bound to a B Ge cluster. The slow diffusivity of B in Ge coupled with the overwhelmingly large population of Ge interstitials relative to the implanted B fluence following implantation is given as an explanation for the inability for B interstitial atoms to find a vacant site during annealing. Kinetic Monte Carlo simulations were employed to simulate the activation behavior

PAGE 130

130 Table 5 1. Inactive B determined from micro Hall effect and the corres ponding displaced Ge obtained from RBS characterization for c Ge samples annealed for various times. Annealing Time (h) Inactive B (/cm 2 ) Displaced Ge (/cm 2 ) Ge:B Ratio 1 .0 7.210 14 4.9310 15 6.86 3 .0 7.210 14 4.8710 15 6.80 5 .0 7.210 14 4.5610 15 6.33

PAGE 131

131 Figure 5 1. Sheet resistance data obtained for samples implanted at 2 keV with a fluence of 1 .010 15 cm and subsequently annealed for various times for A) crystalline Ge and B) preamorphized Ge.

PAGE 132

132 Figure 5 2. Sheet resistance line scan acquired across c Ge sample annealed at 400 C for 30s. The precision of the M4PP technique is evidenced.

PAGE 133

133 Figure 5 3 Sheet number data obtained for samples implanted at 2 keV with a fluence of 1 .010 15 cm and subsequently annealed a for various times for A) crystalline Ge and B) preamorphized Ge.

PAGE 134

134 Figure 5 4. Drift mobility for a c Ge samples implanted at 2 keV with a fluence of 1 .010 15 cm and subsequently annealed for various times

PAGE 135

135 Figure 5 5. RBS cha nneling spectra for c Ge samples implanted at 2 keV to a fluence of 1 .010 15 cm and subsequently annealed for various times where A) is the entire spectra and B) is the surface region.

PAGE 136

136 Figure 5 6 Sheet resistance data obtained for sample s implanted at 2 keV with fluences ranging from 5.010 13 to 5.010 15 cm and subsequent annealing at for A) crystalline Ge and B) preamorphized Ge.

PAGE 137

137 Figure 5 7 Change in sheet resistance for 2, 4, and 6 keV B + implants to fluen ces ranging from 5.010 13 to 5.010 15 cm increase and decrease in R S for PA Ge and c Ge, respectively.

PAGE 138

138 Figure 5 8 Electrical activation character istics as a function of anneal temperature for samples B + implanted at 2, 4, and 6 keV to a fluence of 5.010 15 cm into c Ge and PA Ge. M easured A) sheet number and B) drift mobility

PAGE 139

139 Figure 5 9 HR XTEM images of samples B + implanted at 2 keV to a fluence of 5.010 15 cm 2 into c A ) at B ) at C ) and D) into PA Ge

PAGE 140

140 Figure 5 10 Sheet resistance for samples B + implanted at 6 keV to a fluence of 5.010 15 cm 2 into c Ge a nd PA Figure 5 11 Sheet number for samples B + implanted at 6 keV to a fluence of 5.010 15 cm 2 into c Ge and PA C were 60s.

PAGE 141

141 Figure 5 12 B concentration profiles for samples B + implanted at 6 keV to a fluence of 5.010 15 cm 2 into c Ge as Figure 5 13 B concentration profiles for samples B + implanted at 6 keV to a fluence of 5.010 15 cm 2 into PA Ge as

PAGE 142

142 Figure 5 14 HR XTEM micrograph of a pit formed in Ge following thermal processing. The presence of the native oxide s urrounding the pit is evident.

PAGE 143

143 Figure 5 15. SIMS characterization of a sample B + implanted at 35 keV to a fluence of 2 .010 1 5 cm 2 following various thermal treatments. The horizontal shift in the chemical profile of the uncapped sample annealed at is evident J. Oh and J.C. Campbell, J. of Elec tron. Mater. 33 364. Copyright ( 2004). Kindly r eprinted with permission from Springer Science and Business Media.

PAGE 144

144 Figure 5 16. Simulation displaying the cumulative vacancy population as a function of radial distance from an implanted B ion at 2 keV.

PAGE 145

145 Figure 5 17. Kinetic Monte Carlo simulations displaying the substi tut ional and interstitial boron concentration s following annealing at 400 C for 60s for a 2 keV B + implant to a f luence of A) 5.010 1 3 cm 2 and B) 5.010 15 cm 2

PAGE 146

146 Figure 5 18. Kinetic Monte Carlo simulations displaying the substitutional and interstitial boron concentrations following B + implantation at 2 keV to a fluence of 5.010 1 3 cm 2 and annealing at 400 A) 60s and B) 1h

PAGE 147

147 CHAPTER 6 IMPLANT RELATED DAMAGE IN G E 6 .1 Introduction For the past half century, ion implantation has been used extensively in the semiconductor industry due to its ability to readily position high dopant concentrations accura tely due to its self aligning capabilities. Ion implantation has its drawbacks in that it creates lattice displacements which, given the correct conditions, can create extended defects during subsequent annealing. The formation and evolution of ion implant ation related defects and their effects on semiconductors have been studied in depth for several decades. The understanding of these effects on dopant activation and diffusion are crucial for the scaling of semiconductor devices where with electrical acti vation needs to be high and junction depths are becoming increasing shallow. For Si, control of extended defects has proved invaluable for realizing dopant defect interactions and their effects on dopant diffusion and activation. It is well known that th e release of interstitials during defect dissolution allows for subsequent dopant kick out and diffusion in excess of equilibrium conditions. 7 12 This behavior allows for anomalous diffusion of dopants in Si and is known as transient enhanced diffusion (TED) which is characterized by a nega tive activation energy. The process can be inhibited through the implementation of a so called vacancy engineering implant 74 77 or with a co implanted species to act as an interstitial trap to reduce the effects of excess interstitials. 72,170 In addition, it is known that excess interstitials from extended defects can affect electrical activation as well. 13 15 Recently, Ge has garnered research interests as a replacemen t material for Si in microelectronic applications. 127 Its advantageo us electrical characteristics such as

PAGE 148

148 increased hole mobility may allow for continued scaling of microelectronic devices. 1,25 The deviation from po ly Si/SiO 2 gate stacks has enabled the industry to avoid the issues of its unstable oxide. 171 In addition, the relatively low melting point of Ge allow s for reduced processing temperature which is advantageous for the implementation metal gate/high expansive amount of knowledge regarding defect formation in Si, little is understood r egarding implant related extended defects in Ge. Several authors have investigated the amorphization of Ge and subsequent defect formation at the amorphous crystalline interface or end of range (EOR) with contrasting results. A few reports have shown th at no EOR damage was observed, 34,35,38,45,151 while other reports have observed the presence of small dislocation loops at the EOR and reported their properties in detail. 20,150,172 174 Of note, it has been shown that the dissolution of these extended defects and the rel ease of the contained interstitials has an effect on dopant activation. 175 These defects have been shown to be small dislocation loops which dis Similarly, there have been mixed reports of projected range damge produced followin g sub amorphizing implantations. It has been observed that projected range defects do not form following conventional implantation and anneal ing commonly used in semiconductor research. 150,172,176 Although rod like defects have been reported during H + He + and e irradiation at elevated temperature, it needs to be stressed that these defects were not formed using conventional dopant or self implants, but rather high energy irradiations at increased temperature. 177 183

PAGE 149

149 It has been observed that traditional project range defects form in Ge following ultra shallow B + implantation upon annealing. In this work, high resolution cross sectional transmission electron microscopy (HR XTEM) and plan view TEM (PTEM) is used to characterize the formation and evolution of these defects. 6 .2 Experimental Details Experiments were performed on Czochralski grown n type Ge (001) wafers with cm. Samples were B + implanted at 2 keV to a fluence of 1.010 15 cm 2 The beam current was fixed at 1.1 mA for the B + implants and the platen t o resist sample heating For these conditions, the B + implantation is non amorphizing. Samples were annealed in an N 2 ambient at 400 HR XTEM was completed using a JEOL 2010F to image the microstructure of specimens before and after annealing. HR XTEM samples were prepared using a FEI DB235 focused ion beam. Final polishing was completed with a 7 kV Ga + beam. PTEM in a Fischione 1010 ion mill. A s econd set of (001) Ge samples wafers were implanted with 1 MeV Ge + to a fluence of 2 .010 15 cm 2 The implant created amorphous layer extending approximately 870 nm from the surface. Samples were annealed in an N 2 ambient at 330 10 to 214 min. TEM samples were prepared using a FIB and subsequent characterization was completed on a JEOL 200CX. Weak beam dark field imaging was used with the g 400 (3g) diffracting condition. Ge is highly susceptible to FIB dama ge during preparation. 184 The FIB ion damage is difficult to distinguish from Ge + implant related defects which makes quantification di fficult. To circumvent the issue of FIB damage introducing background

PAGE 150

150 noise and skewing the quantification, an image processing program, ImageJ was used for background removal. After the raw image was acquired, the image was converted to binary and the defects highlighted based on a minimum size of ~5 nm. After defects were highlighted, the defect size and population was tabulated as a function of annealing time. 6 .3 Projected Range Damage Figure 6 1 displays a bright field TEM image taken with g 220 ( 3g) diffraction conditions of a sample that was annealed at 500 C for 2h. From this image, small rod like defects with an average width of approximately 3 nm are evident. These typical projected range defects 16 ar e small in nature and are inclined with respect to the (001) Ge surface. Their general appearance is similar to that of {311} type defects which has been observed previously in Ge following electro n or light particle irradiation at elevated temperature. 177 183 However, the observed defects in this present work are much smaller in width as those created d uring elevated temperature irradiations. This is the first reported observation of {311} type defects in Ge following B + implantation. in which they are corrugated across their width These so called zig zag defects were reported for ultra low energy B + implantation into Si for fluences that create a concentration of interstitials in excess of 1% of the number of atoms in the solid 9 In this work by Agarwal et al. it was observed that zig zag defects evolve by unfaulting onto additional (311) planes thereby creating the zig zag morphology. It was also observed that the zig zag defects were significantly more stable during annealing than traditional {311} defects which allowed them to grow to significantly larger lengths as compared to ordinary {311} type defects.

PAGE 151

151 Figure 6 2 displays a HR XTEM image of a sample annealed for 24h at 500 C which displays several zig zag defects in close proximity. The proximity of several zig zag defects following a substantial anneal gives credence to the notion that zig zag defects are also more thermally stable than ordinary {311} defects i n Ge. However, the width of the defects do es not appear to be substantially larger during annealing. Further examination of the defects using HR XTEM shows that not all defects are orientated on a {311} plane. However, a small fraction of defects are actually situated on what appears to be a {511} plane. The origin of this {511} defect is not known as it has not been reported before following implantation into Group IV materials. Indeed, the small defect width makes accurate measure of the angle of the habit plane difficult. Figure 6 3 shows two HR XTEM images of both {311} and {511} type defects and the calculated misorientation with respect to the (001) plane. XTEM images from all anneals were analyzed and defect angle subsequently characterized. Defect angle quantification was aided by the use of computer software, namely ImageJ. 185 I t is apparent that a large fraction of the observed defects lie on a plane that is oriented approximately 25 from the (001) plane while a smaller subset comprising approximately 20% of the total is oriented 16 from the (001) plane. The spread of the me asured defect angles can be attributed to a few factors. The minute dimensions of the defects inhibits the exact measurement as it is difficult to accurately mark the defect end points. In addition, the small length decreases the contrast of the defect ob served in the TEM. For example, the defect are approximately 10nm in length while an XTEM sample is typically on the order of 50nm. A simple approximation yields that 20% of the through plane thickness yields information

PAGE 152

152 regarding the defect, while the o ther 80% of the sample masks information. However, the obsserved defect counts as a function of angle from the (001) plane gives confidence that the defects are a mixture of {311} and {511} defects in a 4:1 ratio. The presence of {511} defects was observ ed during the entire annealing sequence. Figure 6 4 displays a series of select HR XTEM images from samples annealed at 500 C for times ranging from 2 to 36 hr. It should be noted that XTEM images are not entirely representative of the defect concentrat ion due to differences in sample thickness as well as the minute sampling volume. Images were chosen to represent the bulk of the sample. Additional annealing does not appear to create any coarsening of the defects. In opposition to what is observed for defect evolution in Si, the projected range damage in Ge does not evolve to form dislocation loops. 8 In conjunction with what has been reported thus far for EOR damage in Ge, the defects appear to decrease in number and size with further annealing. 20,172 ,174 Figure 6 5 displays PTEM images taken after annealing for 2 and 12h. The defects are approximately 10 2 nm in length along <011> directions for both annealing conditions. A marginal increase in length is observed after annealing for 12h. The d efects decrease in number with increasing annealing time. Defect quantification shows that there are approximately 8.07 10 1 0 cm 2 and 1.24 10 10 cm 2 defects following 2 h and 12 h at 500 C, respectively. Defect dissolution was observed for anneals in ex cess of 500 C for 36 hr. Select samples were also annealed for various times for temperatures between 400 and 600 C. In conjunction with electrical data, it has been shown that {311} defect formation or dissolution does not have any substantial effect on B activation in Ge.

PAGE 153

153 Therefore, i t is assumed that the defects are not primarily composed of B atoms, but rather composed of Ge interstitials. Following dissolution and presumably a release of excess Ge interstitials, no change in the number of active B carriers is observed which suggests that there is minimal interaction between these excess interstitials and B atoms. However, the density and size of the defect is quite small which limits the total number of trapped G e interstitial or B atoms that may b e released during dissolution. 6 .4 End of Range Damage Figure 6 6 shows XTEM images taken with g 220 (3g) diffraction condition of EOR damage created by a 1 MeV Ge + implant to a fluence of 2 .010 15 cm 2 following annealing at 330 defects have formed just below the original amorphous crystalline interface at a depth of approximately 870 nm from the surface. The defects show contrast in the form of approximately spherical white dots Reports of EOR in Ge by Koffel et al. have show n that defects of similar contrast are composed of self interstitials as determined by X ray diffraction techinques. 174 Presumably, the observed defects in this work are composed of self interstitials as well. The small nature of the defects makes direct identification of the defect difficult. Figure 6 7 displays gra phs of defect density and defect size as a function of damage imparted by FIB preparation and makes absolute defect quantification difficult. Background noise was subtracted using ImageJ as described in previous sections. The defect size was determined by measuring the diameter and assuming the defects are spherical in nature. In accordance with previous reports of EOR in Ge, the defects are small as they are no larger than ap proximately 10 nm in diameter at their largest. 20,174 With

PAGE 154

154 increasing annealing time, the defect density decreases with the average defect size increasing for times less than 50 min. For times grea ter than approximately 50 min, the defect size also decreases. The combination of decreasing defect size and density suggests a non conservative evolution of defects where the overall population of trapped self interstitials decreases with time. These fi ndings are in agreement with previous reports of EOR evolution in Ge. 20,174 6 .5 Di scussion Previous works have investigated high fluence B + implantations in pure Ge, but extended defects were not ob served ; rather only a disordered microstructure 38 In addition, Crosby et al. studied the formation of extended defects in Si x Ge ( 1 x ) and found that {311} defects did not form for Ge concentrations grea ter than 25% which was attributed to the reduced bond strength of Ge. 176 The first observation of projected range damage may be attributed to the surfa ce proximity of these shallow B + implants. As has been discussed in previous chapters, the Ge surface is regarded as a vacancy sink while reflecting interstitials which would increase extended defect formation. 50 Similar to previous published results of imp lant related damage in Ge, the observed projected range and EOR defects were much smaller in size than what is typically observed in Si. In addition, the defects were not observed to coarsen appreciably in during annealing. The results suggest that defec t formation and evolution is not as signif icant in Ge as compared to Si where it has been shown to have a profound effect on B activation 13 as well as on B diffusion 67 A possible reason for this occurrence is due to the dominant point defect in Ge being the vacancy rather than the interstitial as it is in Si. 165 It has been shown in pr evious chapters that self interstitials have limited mobility at the investigated

PAGE 155

155 temperatures for this work. It can be concluded that this lack of mobility hinders the ability of interstitials to coalesce into extended defects. Likewise, it can also be concluded that interstitials may have a higher probability of recombining with a mobile vacancy prior to forming an extended defect. Further evidence that may show this to be true was produced by Boninelli et al. where they showed a dependence of EOR for mation and dissolution on Ge implant energy. 186 The increase of EOR damage with increasing energy could be attributed to the separation of interstitial and vacancy profiles during implantation. For high energy implants, the forward momentum of the ion drives interstitials deeper into the solid while an excess vacancy population remains near the surface. This separation of point defects could allow for decreased recombination and increased EOR damage formation. Similar energy dependence on EOR results were found in this work, but were not reported. The reduced amorphization threshold of Ge is likely a primary factor regarding the reduced number and size of EOR defects in Ge. For Ge, the numbe r of vacancies required to produce an amorphous layer is reduced. 20,21 Because of this, there are less self interstitials remaining at the end of range following regrowth. The reduced number of self interstitials available for extended defect formation has a direct influence on the size an d number of EOR defects created. 6 .6 Summary The formation and evolution of projected range and end of range damage in Ge has been investigated. For both projected range and EOR defects in Ge, the damage evolution has been shown to be non conservative w here the total number of trapped interstitials decreases with annealing. The relatively low concentration of implant related

PAGE 156

156 defects in Ge has been attributed to the quick diffusion of vacancies and lack of mobile interstitials in Ge. The first observati on of rod like defects were reported following a non amorphizing 2 keV B + implant at a fluence of 1 .010 15 cm 2 The defects were determined to be primarily of a {311} type with a small subset occupying a {511} plane. Zig zag defects were also observed a cross the annealing range. The rod like defects were observed to not undergo an appreciable increase in length or width during annealing with the average length and width being approximately 10 nm and 3 nm, respectively. No dislocation loops were observe d following the dissolution of rod like defects. The dissolution of the rod like defects did not have an appreciable effect on activation. The formation and evolution of EOR damage in Ge was monitored as a function of annealing time at 330 C following a 1 MeV Ge + implant to a fluence of 2 .010 15 cm 2 Defects were observed to initially increase in size for times less than approximately 50 min and decrease in size for longer anneals. The defect density was observed to decrease with increasing annealing time.

PAGE 157

157 Figure 6 1. Bright field XTEM image taken with g 220 (3g) diffraction condition of projected range damage created by a 2 keV B + implant to a fluence of 1.010 15 cm 2 following annealing at 500 for 2 h times ranging from 2 to 36h

PAGE 158

158 Figur e 6 2 HR XTEM image of projected range damage created by a 2 keV B + implant to a fluence of 1.010 15 cm 2 following annealing at 500 zig zag defects are evident.

PAGE 159

159 Figure 6 3 Defect orientations of projected range damage in Ge. HR XTEM images of A) {311} and B) {511} type defects. C) Histogram displaying their relative abundance.

PAGE 160

160 Figure 6 4 XTEM image s of projected range damage created by a 2 keV B + implant to a fluence of 1.010 15 cm 2 following annealing at 500 for times ranging from 2 to 36 h

PAGE 161

161 Figure 6 5 Bright field PTEM images taken with g 220 (3g) diffraction condition of projected range damage created by a 2 keV B + implant to a fluence of 1.010 15 cm 2 following annealing at 500

PAGE 162

162 Fi gure 6 6 XTEM images taken with g 220 (3g) diffraction condition of EOR damage created by a 1 MeV Ge + implant to a fluence of 2 .010 15 cm 2 following annealing at 330

PAGE 163

163 Figure 6 7 EOR damage created by a 1 MeV Ge + implant to a fluence of 2 .010 15 cm 2 as a function of annealing time at 330 A) average defect density and B) average defect size.

PAGE 164

164 CHAPTER 7 CONCLUSIONS AND FUTU RE WORK The implantation and activation of ultra shallow B + implants in Ge was thoroughly investigated. Prior to this work, very little research was conducted on these technologically relevant ion implants. It can be assumed that the dearth of information presently available prior to this work is related to the difficulty in measuring the electrical prope rties of these ultra shallow layers as evidenced in this document. The advent of the micro H all effect and micro four point probe techniques has pushed the scaling of the metrology tools used to characterize these layers in conjunction with the physical s caling of these implants which has allowed for their accurate characterization. The activation behavior of ultra shallow B + implants in Ge has been determined to follow anomalous activation trends as compared to what has been previously published regardi ng higher energy implants in Ge as well as B activation in Si It has been observed that independent of implanted fluence, a large fraction of the implanted B is rendered inactive for implants into both crystalline and preamorphized Ge. Ion beam analysis was used to verify the micro Hall effect results and it was determined that a large fraction of B is in fact situated in off lattice positions following annealing. The results also suggest that a preferred site for inactive B does not exist. The effect o f annealing has been shown to have little effect in increasing or decreasing the fraction of activated dopant. Similar to results reported for higher energy B + implants in Ge, slight changes in sheet resistance and activation were observed as a function o f annealing temperature between 4 00 and 600 C ; however, the changes did not significantly affect the activation values. High temperature isochronal anneals were

PAGE 165

165 preamor phized and crystalline samples, respectively. The deactivation is believed to be due to the B falling out of solution due to the concentration in excess of the chemical solubility limit. Across all investigated temperatures and times investigated, a larg e fraction of inactive B remained for both crystalline and preamorphized samples The activation behavior of B + implants in Ge has been shown to be independent of the implanted fluence and as such does not exhibit a single electrical solubility across the investigated fluence range. With increasing fluence, the percent of activated dopant atoms remains fixed for a given energy which suggests that the activation behavior of B in Ge is related to the fluence dependent damage imparted to the crystal. The ano malous activation behavior of ultra shallow B + implants is believed to be correlated with the relatively immobile B species in Ge and the relatively large population of Ge interstitials created during implantation The low diffusivity of B in Ge restricts its ability travel to and to combine with a vacant lattice site. The B + implantation process creates far more interstitials per incoming ion which creates an atmosphere of excess interstitials in close proxim ity with the implanted B atom. The surface pr oximity is expected to further increase the interstitial super saturation as th e surface is known to act as a sink for vacancies. The competition between B atoms and interstitials decreases the probability that B atoms will be able to recombine with a vac ant site. For the first time, it has been shown that extended defects form in Ge following sub amorphizing B + implantation as confirmed by cross sectional TEM analysis. The defects were observed to be predominately of the {311} type with a small fraction having a {511} habit plane. In comparison to {311} defects in Si, the defects are smaller in stature and were not observed to coarsen or to evolve into dislocation loops. The

PAGE 166

166 defects were observed to form following an anneal at 500 C for 0.5hr with diss olution occurring after 36 hr. Upon dissolution, no change in activation was observed which suggests that defects are not composed of B atoms. End of range damage produced by amorphizing Ge + implants was investigated. It was observed that end of range damage was formed following a high energy self implantation and subsequent annealing for various times at 330 C. In contrast to Si, the defects were small in stature and not observed to coarsen appreciably, but rather the number of trapped interstitials a ppeared to decrease with annealing time. An increase in size was observed initially for short annealing times with a decrease in size for long anneals. The number of defects continually decreased with increasing annealing times. In this work, t he activa tion and implant related damage results have shown that dopant defect interactions for ultra shallow B + implants in Ge differs greatly from that of Si. With the vacancy being the dominant point defect, the associated physical models of dopant defect inter actions in Ge will need to be adjusted significan tly for future experimentation. The results published in this work have documented several new findings regarding the implantation and activation or ultra shallow B + implants in Ge. Although attempted for t his work, but ultimately not to fruition, the use of vacancy engineering may yield important information regarding the activation of these technologically relevant implants. Vacancy engineering involves a high energy (a few MeV) implant to introduce an ex cess concentration of vacancies near the surface with excess interstitials driven further into the bulk due to forward momentum transfer. It has been used successfully to reduce TED of B in Si. The excess vacancies introduced should allow

PAGE 167

167 for increased ac tivation or decreased projected range damage and would be an interesting experiment to complete. In addition, further implementation of ion beam analysis or positron annihilation spectroscopy could be employed to further analyze the resulting point defec t populations following annealing and during subsequent processing. Understanding the evolution of the point defect population would prove fruitful in grasping the dopant defec t interactions present in Ge. Lastly, the implementation of high resolution x ray diffraction (HRXRD) could be employed to understand the local changes in lattice constant following B + implantation and processing. HRXRD has been employed for several material systems to understand the strain associated with dopants or dopant cluste rs. Monitoring the strain induced by B + implantation and with subsequent processing could allow for the use of simulations to produce information regarding the inactive B concentration.

PAGE 168

168 APPENDIX EFFECT OF IMPLANT CO NDITIONS ON B ACTIVA TION A.1 Variable Implantation Conditions It is well known that a variety of implantation parameters, such as ion mass and energy, target temperature, ion flux, etc. can have a profound effect on the resulting lattice damage and micro structure of the target. 187,188 As evidenced in this document, very little research has been conducted regarding ultra shallow B + implants in Ge and this is especially true in terms of i mplant parameters. Thus, the effects of varying beam current on the clustering and electrical activation behavior of ultra shallow B + implants in Ge is investigated. A. 2 Experimental Methods Experiments were performed on Czochralski grown n type Ge (001) w afers with cm. A set of samples were diced and B + implanted at 2 keV to a fluence of 5.010 14 cm 2 with beam current varying from 0.4 to 6.4 mA. The beam size is estimated to be 180 cm 2 which yields an ion flux range of 1.38 10 13 to 2.2110 14 ions/(s cm 2 ) for 0.4 mA and 6.4 mA, respectively. During B + implantation, the annealer (RTA) in N 2 High resolution cross sectional transmission electron microscopy (HR XTEM) was completed using a JEOL 2010F to image the microstructure of specimens before and after annealing. TEM samples were prepared using a FEI DB235 focused ion beam. Electrical char act erization was completed using a CAPRES microRSP M 150 M4PP with Au a magnetic flux density of 0.475 T. Hall sheet number ( n H ) and mobility values ( H )

PAGE 169

169 were adjusted to obtain the carr ier sheet number ( n s ) and drift mobility ( d ) by using a scattering factor ( r H ) of 1.21 as determined em pirically by Mirabella et al 31 The carrier density and drift mobility are related to the Hall values by n s = n H r H and d = H / r H respectively. A. 3 Results and Discussion In Figure A 1 the sheet resistance R s is plotted as a function of beam current for 2 keV B + implants to a dose of 5.010 14 cm of decreasing R s with increasing beam current is observed across the investigated range. At 6.4 mA, the measured R s va 0.4 mA. Interestingly, the decrease in R s can be explained by an increas e in activation as seen in Figure A 2 At 6.4 mA, the n s was 4.5510 13 cm which is an increase of 76% from the lowest current. In Fig ure A 3 the drift mobility as a function of beam current is shown. Due to the increase in ionized dopants, the drift mobility decreases with current to a minimum of 203.2 cm 2 /V s which is in good agreement with other results in this doping regime. 31 Despite the increase in carriers with increasing beam current, th e overall activation of the samples is far from complete activated with the highest value achieved at 11.4%. This finding is not surprising given the abundance of data garnered from various processing conditions that yielded similar values. The microstru cture of the samples was characterized using HR XTEM both as implanted and post anneal to ascertain the presence of any increased damage created by the high beam current implantation Fig ure A 4 shows an image of the highest beam current sample implanted at 6.4 mA without any subsequent annealing The samples implanted at 0.4 and 6.4 mA both appeared unremarkable with no discernible difference

PAGE 170

170 between the samples as implanted and post anneal. The presence of any amorphization was not observed in the as i mplanted case and extended defects were not observed in the annealed case for any sample. Although not discernible via HR XTEM, the presence of sub microscopic damaged or amorphous regions is certainly possible. The presence of these sub microscopic amorp hous regions would allow for increased B activation upon annealing as is observed in this work. Previous studies have shown that B + implantation into c Ge can yield activation values comparable to PA Ge if the fluence is great enough 24 or if the platen is held at reduced temperature. 23 In both cases, B + implantation induces amorphization which yields increased activation upon annealing. For Si, it is well known that varying the beam current during implantatio n can significantly alter the resulting microstructure through ion beam induced recrystallization or amorphization. 189,190 At elevated temperatures, a subtle change in the ion flux is capable of influencing the point defect environment surrounding the crystalline amorphous interface in such a way that it may recrystallize or further amorphize the layer depending o n an increase or decrease in beam current, respectively However, i t should be noted that the observed increase in active carriers is not believed to be due to any beam heating effects due to the platen being held at room temperature during the implantati on. In addition, the relatively low implant energy and low fluence reduces the total power deposited into the sample. Several studies regarding ion beam induced recrystallization and amorphization have been completed with Ge which have reported similar findings. 191 193 Notably, Sigurd et al have reported that for B + ions in Ge, the lattice disorder produced is ten

PAGE 171

171 times higher than what is observed in Si at room temperature. 191 These results are likely explained through the lower amorphization threshold of Ge as compared to Si. 21 H owever, for light ions such as B, the amorphous layer is not as thick nor is amorphization observed at the projected range of the ion as predicted by simulations for heavier ions, but rather occurs closer to the surface. 20 A. 4 Summary The reduced amorphization threshold of Ge with respect to Si likely explains the observed activation behavior following B + implantation at variable beam curren ts. The dilute damage cascades created during B + implantation does not allow for the formation of a distinct amorphous layer, but locally, it is assumed that amorphous pockets are formed during implantation During annealing, these amorphous pockets allo w for increased B activation.

PAGE 172

172 Figure A 1 Measured sheet resistance ( R s ) after annealing at 400 C for 60 s as a function of beam current implanted at 2 keV to a fluence of 5.010 14 cm into crystalline Ge.

PAGE 173

173 Figure A 2 Measured sheet number ( n s ) and percent electrical activation as a function of beam current implanted at 2 keV to a fluence of 5.010 14 cm into crystalline Ge after annealing at 400 C for 60s.

PAGE 174

174 Figure A 3 Measured drift mobility ( D ) as a function of beam current implante d at 2 keV to a fluence of 5.010 14 cm into crystalline Ge after annealing at 400 C for 60s.

PAGE 175

175 Figure A 4 HR XTEM micrograph of an as implanted crystalline Ge sample B + implanted at 2 keV to 5.010 1 4 cm at a beam current of 6.4 mA showing a 2.9 0. 3 nm surface GeO x layer and no evident implant damage or amorphization present near the projected range, R p A simulation of the B profile is overlaid on the image.

PAGE 176

176 LIST OF REFERENCES 1 G.E. Moore, El ectronics 38 114 (1965). 2 S.M. Sze and K.K. Ng, Physics of Semiconductor Devices (John Wiley and Sons, 2007). 3 J.F. Gibbons, Proceedings of the IEEE 56 295 (1968). 4 J.F. Gibbons, Proceedings of the IEEE 60 1062 (1972). 5 E. Chason, S.T. Picraux, J.M. Poate, J.O. Borland, M.I. Current, T. Diaz de la Rubia, D.J. Eaglesham, O.W. Holland, M.E. Law, C.W. Magee, J.W. Mayer, J. Mel ngailis, and A.F. Tasch, Appl. Phys. Rev. 1997 2 (1997). 6 J.S. Williams, Mater. Sci. and Eng A253 8 (1998). 7 D.J. Eag lesham, P.A. Stolk, H. J. Gossmann, and J.M. Poate, Appl. Phys. Lett. 65 2305 (1994). 8 D.J. Eaglesham, P.A. Stolk, H. J. Gossmann, T.E. Haynes, and J.M. Poate, Nucl. Instrum. Methods Phys. Res. B 106 191 (1995). 9 A. Agarwal, T.E. Haynes, D.J. Eaglesh am, H. J. Gossmann, D.C. Jacobson, J.M. Poate, and Y.E. Erokhin, Appl. Phys. Lett. 70 3332 (1997). 10 A. Agarwal, H. J. Gossmann, D.J. Eaglesham, L. Pelaz, D.C. Jacobson, T.E. Haynes, and Y.E. Erokhin, Appl. Phys. Lett. 71 3141 (1997). 11 A. Agarwal, H J. Gossmann, D.. Eaglesham, L. Pelaz, S.. Herner, D.. Jacobson, T.. Haynes, and R. Simonton, Mater. Sci. Semicond. Process. 1 17 (1998). 12 J. Li and K.S. Jones, Appl. Phys. Lett. 73 3748 (1998). 13 A.D. Lilak, M.E. Law, L. Radic, K.S. Jones, and M. Clark, Appl. Phys. Lett. 81 2244 (2002). 14 N.E.B. Cowern, B. Colombeau, J. Benson, A.J. Smith, W. Lerch, S. Paul, T. Graf, F. Cristiano, X. Hebras, and D. Bolze, Appl. Phys. Lett. 86 101905 (2005). 15 L. Romano, A.M. Piro, V. Privitera, E. Rimini, G. Fortunato, B.G. Svensson, M. Foad, and M.G. Grimaldi, Nucl. Instrum. Methods Phys. Res. B 253 50 (2006). 16 K.S. Jones, S. Prussin and E.R. Weber, Appl. Phys. A 45 1 (1988).

PAGE 177

177 17 G. Hobler and G. Otto, Mater. Sci. Semicond. Process. 6 1 (2003). 18 U. Yarkulov, Rad. Effect. 100 11 (1986). 19 U. Yarkulov, Rad. Effect. 103 135 (1987). 20 A. Claverie, S. Koffel, N. Cherkashin, G. Benassayag, and P. Scheiblin, Thin Solid Films 518 2307 (2010). 21 S. Koffel, P. Scheiblin, A. Claverie, and G. Benassaya g, J. Appl. Phys. 105 013528 (2009). 22 D.P. Hickey, Ion Implantation Induced Defect Formation and Amorphization in the Group IV Semiconductors: Diamond, Silicon and Germanium University of Florida, 2007. 23 E. Bruno, G. Impellizzeri, S. Mirabella, A .M. Piro, A. Irre ra, and M.G. Grimaldi, Mater. Sci. and Eng. B 154 155 56 (2008). 24 B.R. Yates, B.L. Darby, N.G. Rudawski, K.S. Jones, D.H. Petersen, O. Hansen, R. Lin, P.F. N ielsen, and A. Kontos, Mater. Lett. 65 3540 (2011). 25 W. Martienssen and H. Warlimont, editors, Springer Handbook of Condensed Matter and Materials Data 26 L. R. Bidwell, J. of the Less Common Metals 20 19 (1970). 27 R. Olesinski and G. Abbaschian, J of Phase Equilibria 5 476 (1984). 28 S. Uppal, A.F.W. Willoughby, J.M. Bonar, A.G.R. Evans, N.E.B. Cowern, R. Morris, and M.G. Dowsett, J. Appl. Phys. 90 4293 (2001). 29 S. Uppal, A.F.W. Willoughby, J.M. Bonar, A.G.R. Evans, N.E.B. Cowern, R. Morris, and M.G. Dowsett, Physica B 308 310 525 (2001). 30 S. Uppal, A.F.W. Willoughby, J.M. Bonar, N.E.B. Cowern, T. Grasby, R.J.H. Morris, and M.G. Dowsett, J. Appl. Phys. 96 1376 (2004). 31 S. Mirabella, G. Impellizzeri, A.M. Piro, E. Bruno, and M.G. Grimaldi, Appl. Phys. Lett. 92 251909 (20 08). 32 V.M. Gusev, M.I. Guseva, E.S. Ionova, A.N. Mansurova, and C.V. Starinin, Phys. Stat. Sol. A 21 413 (1974). 33 K.S. Jones and E.E. Haller, J. Appl. Phys. 61 7 (1987).

PAGE 178

178 34 A. Satta, E. Simoen, T. Clarysse, T. Janssens, A. Benedetti, B. De Jaeger, M. Meuris, and W. Vandervorst, Appl. Phys. Lett. 87 172109 (2005). 35 Y. L. Chao, S. Prussin, J.C.S. Woo, and R. Scholz, Appl. Phys. Lett. 87 142102 (2005). 36 Yong Seok Suh, M.S. Carroll, R.A. Levy, G. Bisognin, D. De Salvador, M.A. Sahiner, and C.A. King, Electron Devices, IEEE Transactions On 52 2416 (2005). 37 G. Impellizzeri, S. Mirabella, E. Bruno, A.M. Piro, and M.G. Grimaldi, J. Appl. Phys. 105 063533 (2009). 38 A. Satta, E. Simoen, T. Janssens, T. Clarysse, B.D. Jaeger, A. Benedetti, I. Hof lijk, B. Brijs, M. Meuris, and W. Vandervorst, J. Electrochem. Soc. 153 G229 (2006). 39 G. Bisognin, S. Vangelista, M. Berti, G. Impellizzeri, and M.G. Grimaldi, J. Appl. Phys. 107 103512 (2010). 40 ysse, K. Martens, B. De Jaeger, A. Benedetti, I. Hoflijk, B. Brijs, M. Meuris, and W. Vandervorst, Mater. Sci. Semicond. Process. 9 634 (2006). 41 N. Ioannou, D. Skarlatos, C. Tsamis, C.A. Krontiras, S.N. Georga, A. Christofi, and D.S. McPhail, Appl. Phy s. Lett. 93 101910 (2008). 42 Daele, and T. Janssens, Nucl. Instrum. Methods Phys. Res. B 257 157 (2007). 43 V.I. Fistul, M.I. Iglitsyn, and E.M. Ome Sov. Phys. Sol State 4 784 (1962). 44 O.A. Golikova, Sov. Phys. Sol. State 3 2259 (1962). 45 E. Simoen, G. Brouwers, A. Satta, M. L. David, F. Pailloux, B. Parmentier, T. Clarysse, J. Goossens, W. Vandervorst, and M. Meuris, Mater. Sci. Semicond. Process 11 368 (2 008). 46 G. Hellings, E. Rosseel, T. Clarysse, D.H. Petersen, O. Hansen, P.F. Nielsen, E. Simoen, G. Eneman, B. De Jaeger, T. Hoffmann, K. De Meyer, and W. Vandervorst, Microelectron. Eng 88 347 (2011). 47 N.S. Bennett and N.E.B. Cowern, Appl. Phys. Le tt. 100 172106 (2012). 48 E. Napolitani, G. Bisognin, E. Bruno, M. Mastromatteo, G.G. Scapellato, S. Boninelli, D. De Salvador, S. Mirabella, C. Spinella, A. Carnera, and F. Priolo, Appl. Phys. Lett. 96 201906 (2010).

PAGE 179

179 49 M.I. Guseva and A.N. Mansurova, Rad Effect. 20 207 (1973). 50 H. Bracht, S. Schneider, J.N. Klug, C.Y. Liao, J.L. Hansen, E.E. Haller, A.N. Larsen, D. Bougeard, M. Posselt, and C. Wndisch, Phys. Rev. Lett. 103 255501 (2009). 51 E. Bruno, S. Mirabella, G. Scapellato, G. Impellizzer i, A. Terrasi, F. Priolo, E. Napolitani, D. De Salvador, M. Mastromatteo, and A. Carnera, Thin Solid Films 518 2386 (2010). 52 G.G. Scapellato, E. Bruno, A.J. Smith, E. Napolitani, D. De Salvador, S. Mirabella, M. Mastromatteo, A. Carnera, R. Gwilliam, a nd F. Priolo, Nucl. Instrum. Methods Phys. Res. B 282 8 (2012). 53 L.A. Edelman, K.S. Jones, R.G. Elliman, L.M. Rubin, E.G. Seebauer, S.B. Felch, A. Jain, and Y.V. Kondratenko, AIP Conf. Proc. 1066 225 (2008). 54 G.G. Scapellato, S. Boninelli, E. Napol itani, E. Bruno, A.J. Smith, S. Mirabella, M. Mastromatteo, D. De Salvador, R. Gwilliam, C. Spinella, A. Carnera, and F. Priolo, Phys. Rev. B 84 024104 (2011). 55 F.A. Trumbore, J. Bell Syst. Tech. 39 205 (1960). 56 P.M. Fahey, P.B. Griffin, and J.D. P lummer, Rev. Mod. Phys. 61 289 (1989). 57 N.E.B. Cowern, G.F.A. van de Walle, P.C. Zalm, and D.W.E. Vandenhoudt, Appl. Phys. Lett. 65 2981 (1994). 58 L. Pelaz, M. Jaraiz, G.H. Gilmer, H. J. Gossmann, C.S. Rafferty, D.J. Eaglesham, and J.M. Poate, Appl. Phys. Lett. 70 2285 (1997). 59 M.J. Caturla, M.D. Johnson, and T. Diaz de la Rubia, Appl. Phys. Lett. 72 2736 (1998). 60 L. Pelaz, V.C. Venezia, H. J. Gossmann, G.H. Gilmer, A.T. Fiory, C.S. Rafferty, M. Jaraiz, and J. Barbolla, Appl. Phys. Lett. 75 662 (1999). 61 P.A. Stolk, H. J. Gossmann, D.J. Eaglesham, D.C. Jacobson, J.M. Poate, and H.S. Luftman, Appl. Phys. Lett. 66 568 (1995). 62 G. Mannino, N.E.B. Cowern, F. Roozeboom, and J.G.M. van Berkum, Appl. Phys. Lett. 76 855 (2000). 63 J. Y. Jin, J. Liu, U. Jeong, S. Mehta, and K. Jones, J. Vacuum Sci. Technol. B 20 422 (2002).

PAGE 180

180 64 A.E. Michel, W. Rausch, P.A. Ronsheim, and R.H. Kastl, Appl. Phys. Lett. 50 416 (1987). 65 M. Miyake and S. Aoyama, J. Appl. Phys. 63 1754 (1988). 66 S. Solmi, F. Baruffaldi, and R. Canteri, J. Appl. Phys. 69 2135 (1991). 67 N.E.B. Cowern, K.T.F. Janssen, and H.F.F. Jos, J. Appl. Phys. 68 6191 (1990). 68 P.A. Stolk, H. J. Gossmann, D.J. Eaglesham, and J.M. Poate, Nucl. Instrum. Methods Phys. Res. B 96 187 (1995 ). 69 L.H. Zhang, K.S. Jones, P.H. Chi, and D.S. Simons, Appl. Phys. Lett. 67 2025 (1995). 70 J. Appl. Phys 81 6031 (1997). 71 P.A. Stolk, D.J. Eaglesham, H. J. Gossmann, and J.M. Poate, Appl. Phys. Lett. 66 1370 (1995). 72 G. Impellizzeri, J.H.R. dos Santos, S. Mirabella, F. Priolo, E. Napolitani, and A. Carnera, Appl. Phys. Lett. 84 1862 (2004). 73 G. Impe llizzeri, S. Mirabella, F. Priolo, E. Napolitani, and A. Carnera, J. Appl. Phys 99 103510 (2006). 74 A.J. Smith, N.E.B. Cowern, R. Gwilliam, B.J. Sealy, B. Colombeau, E.J.H. Collart, S. Gennaro, D. Giubertoni, M. Bersani, and M. Barozzi, Appl. Phys. Let t. 88 082112 (2006). 75 R. Gwilliam, N.E.B. Cowern, B. Colombeau, B. Sealy, and A.J. Smith, Nucl. Instrum. Methods Phys. Res. B 261 600 (2007). 76 N.S. Bennett, N.E. Cowern, S. Paul, W. Lerch, H. Kheyrandish, A.J. Smith, R. Gwilliam, and B.J. Sealy, in Solid State Device Research Conference, 2008. ESSDERC 2008. 38th European (IEEE, 2008), pp. 290 293. 77 N.E.B. Cowern, A.J. Smith, N. Bennett, B.J. Sealy, R. Gwilliam, R.P. Webb, B. Colombeau, S. Paul, W. Lerch, and A. Pakfar, Mater. Sci. Forum 573 574 295 (2008). 78 J.F. Ziegler, Nucl. Instrum. Methods Phys. Res. B 219 220 1027 (2004). 79 J. Narayan and O.W. Holland, J. Appl. Phys 56 2913 (1984).

PAGE 181

181 80 N. Misra, L. Xu, M.S. Rogers, S.H. Ko, and C.P. Grigoropoulos, Phys. Stat. Sol. C 5 (2008). 81 M. D. Giles, J. Electrochem. Soc. 138 1160 (1991). 82 E.H. Hall, Amer. J. of Math. 2 287 (1879). 83 L.J. van der Pauw, Philips Res. Rep. 13 (1958). 84 L.J. van der Pauw, Philips Tech. Rev. 20 (1958). 85 D.W. Koon, A.A. Bahl, and E.O. Duncan, Rev. Sci. Instrum. 60 275 (1989). 86 D.W. Koon, Rev. Sci. Instrum. 60 271 (1989). 87 D.W. Koon, Rev. Sci. Instrum. 61 2430 (1990). 88 D.W. Koon and C.J. Knickerbocker, Rev. Sci. Instrum. 63 207 (1992). 89 D.W. Koon and C.J. Knickerbocker, Rev. Sci. Instrum. 64 510 (1993). 90 D.W. Koon and C.J. Knickerbocker, Rev. Sci. Instrum. 67 4282 (1996). 91 F. Szmulowicz, W. Mitchel, F.L. Madarasz, and P.M. Hemenger, Measurement and Theory of the Hall Scattering Factor and the Conductivity Mobility in Ultra Pure P T ype Silicon at Low Temperatures (Defense Technical Information Center, 1983). 92 R.S. Popovic, Hall Effect Devices, Second Edition (CRC Press, 2003). 93 F.J. Morin, Phys. Rev. 93 62 (1954). 94 F.J. Morin and J.P. Maita, Phys. Rev. 94 1525 (1954). 95 O.A. Golikova, B.Y. Mo izhez, and L.S. Stilbans, Sov. Phys. Sol. State 3 (1962). 96 J.E. Dijkstra and W.T. Wenckebach, Appl. Phys. Lett. 70 2428 (1997). 97 International Technology Roadmap for Semiconductors (2011). 98 G. Eneman, M. Wiot, A. Brugere, O.S.I. Casain, S. Sonde, D.P. Brunco, B. De Jaeger, A. Satta, G. Hellings, K. De Meyer, C. Claeys, M. Meuris, M.M. Heyns, and E. Simoen, Electron Devices, IEEE Transactions On 55 2287 (2008). 99 G. Eneman, B. De Jaeger, E. Simoen, D.P. Brunco, G. Hellin gs, J. Mitard, K. De Meyer, M. Meuris, and M.M. Heyns, Electron Devices, IEEE Transactions On 56 3115 (2009).

PAGE 182

182 100 G. Eneman, O. Sicart i Casain, E. Simoen, D.P. Brunco, B. De Jaeger, A. Satta, G. Nicholas, C. Claeys, M. Meuris, and M.M. Heyns, in Solid S tate Device Research Conference, 2007. ESSDERC 2007. 37th European (2007), pp. 454 457. 101 G. Eneman, M.B. Gonzalez, G. Hellings, B.D. Jaeger, G. Wang, J. Mitard, K. DeMeyer, C. Claeys, M. Meuris, M. Heyns, T. Hoffmann, and E. Simoen, ECS Trans. 28 143 (2010). 102 T. Krishnamohan, Z. Krivokapic, K. Uchida, Y. Nishi, and K.C. Saraswat, Electron Devices, IEEE Transactions On 53 990 (2006). 103 M.B. Gonzalez, G. Eneman, G. Wang, B.D. Jaeger, E. Simoen, and C. Claeys, ECS Trans. 34 725 (2011). 104 C.L. Petersen, Rong Lin, D.H. Petersen, and P.F. Nielsen, in IEEE International Conference on Advanced Thermal Processing of Semiconductors, RTP. (2006), p. 153. 105 C.L. Petersen, T.M. Hansen, P. Bggild, A. Boisen, O. Hansen, T. Hassenkam, and F. Grey, Sens and Act. A 96 53 (2002). 106 D.H. Petersen, O. Hansen, T.M. Hansen, P.R.E. Petersen, and P. Bggild, Microelectron. Eng. 85 1092 (2008). 107 S. Thorsteinsson, F. Wang, D.H. Petersen, T.M. Hansen, D. Kjr, R. Lin, J. Y. Kim, P.F. Nielsen, and O. Hans en, Rev. Sci. Instrum. 80 053902 (2009). 108 D.H. Petersen, O. Hansen, R. Lin, P.F. Nielsen, T. Clarysse, J. Goossens, E. Rosseel, and W. Vandervorst, in IEEE International Conference on Advanced Thermal Processing of Semiconductors, RTP. (2008), p. 251 109 J.W. Edington, Practical Electron Microscopy in Materials Science (Techbooks, 1991). 110 D.B. Williams and C.B. Carter, Transmission Electron Microscopy: A Textbook for Materials Science (Springer, 2009). 111 L. de Broglie, Nature 112 (1923). 11 2 B. Fultz and J. Howe, Transmission Electron Microscopy and Diffractometry of Materials 3rd ed. (Springer, 2009). 113 P. Williams, Ann. Rev. of Mater. Sci. 15 517 (1985).

PAGE 183

183 114 A. Benninghoven, F.G. Rudenauer, and H.W. Werner, Secondary Ion Mass Spectro metry: Basic Concepts, Instrumental Aspects, Applications and Trends (1987). 115 C.W. Magee and M.R. Frost, J. of Mass Spectromet. Ion Process. 143 (1995). 116 R.P. Gittins, D.V. Morgan, and G. Dearnaley, J. Phys. D 5 (1972). 117 P. Williams and J.E. Baker, Appl. Phys. Lett. 36 842 (1980). 118 P. Williams, J.E. Baker, J.A. Davies, and T.E. Jackman, Nucl. Instrum. Methods Phys. Res. B 191 318 (1981). 119 Y. Wang and M. Nastasi, editors, Handbook of Modern Ion Beam Materials Analysis 2 Volume Set 2n d ed. (Cambridge University Press, 2010). 120 C. Jeynes, Z.H. Jafri, R.P. Webb, A.C. Kimber, and M.J. Ashwin, Surf. and Interface Anal. 25 254 (1997). 121 B.L. Doyle and D.K. Brice, Nucl. Instrum. Methods Phys. Res. B 35 301 (1988). 122 V.E. Borisenko and S.G. Yudin, Phys. Stat. Sol. 101 123 (1987). 123 Yu Lin Chao and J.C.S. Woo, Electron Devices, IEEE Transactions On 54 2750 (2007). 124 M.Y. Tsai and B.G. Streetman, J. Appl. Phys. 50 183 (1979). 125 T. Clarysse, D. Vanhaeren, I. Hoflijk, and W. Vandervorst, Mater. Sci. and Engineer R 47 123 (2004). 126 T. Clarysse, J. Bogdanowicz, J. Goossens, A. Moussa, E. Rosseel, W. Vandervorst, D.H. Petersen, R. Lin, P.F. Nielsen, O. Hansen, G. Merklin, N.S. Bennett, and N.E.B. Cowern, Mater. Sci. and Eng ineer B 154 155 24 (2008). 127 C. Claeys, J. Mitard, G. Eneman, M. Meuris, and E. Simoen, Thin Solid Films 518 2301 (2010). 128 D.H. Petersen, O. Hansen, R. Lin, and P.F. Nielsen, J. Appl. Phys. 104 013710 (2008). 129 G. Hellings, E. Rosseel, E. Sim oen, D. Radisic, D.H. Petersen, O. Hansen, P.F. Nielsen, G. Zschatzsch, A. Nazir, T. Clarysse, W. Vandervorst, T.Y. Hoffmann, and K. De Meyer, Electrochem. Solid State Lett. 14 H39 (2011). 130 R. Wittmann and S. Selberherr, Solid State Electronics 51 98 2 (2007).

PAGE 184

184 131 L. Romano, A.M. Piro, S. Mirabella, M.G. Grimaldi, and E. Rimini, Appl. Phys. Lett. 87 201905 (2005). 132 A.M. Piro, L. Romano, S. Mirabella, and M.G. Grimaldi, Mater. Sci. and Eng B 124 125 249 (2005). 133 L. Romano, A.M. Piro, M.G. Gr imaldi, and E. Rimini, Nucl. Instrum. Methods Phys. Res. B 249 181 (2006). 134 L. Romano, A.M. Piro, S. Mirabella, and M.G. Grimaldi, Phys. Rev. B 81 075210 (2010). 135 R.J. Kaiser, S. Koffel, P. Pichler, A.J. Bauer, B. Amon, L. Frey and H. Ryssel, Mi croelectron. Engineer. 88 499 (2011). 136 A. Agarwal, H. J. Gossmann, D.J. Eaglesham, L. Pelaz, D.C. Jacobson, J.M. P oate, and T.E. Haynes, Mater. Sci. and Engineer. A 253 269 (1998). 137 F. Priolo, G. Mannino, M. Miccich, V. Privitera, E. Napolitani, and A. Carnera, Appl. Phys. Lett. 72 3011 (1998). 138 B.R. Appleton, O.W. Holland, J. Narayan, O.E. Schow, J.S. Williams, K.T. Short, and E. Lawson, Appl. Phys. Lett. 41 711 (1982). 139 O.W. Holland, B.R. Appleton, and J. Narayan, J. Appl. Phys. 54 2 295 (1983). 140 H. Huber, W. Assmann, S.A. Karamian, A. Mcklich, W. Prusseit, E. Gazis, R. Grtzschel, M. Kokkoris, E. Kossionidis, H.D. Mieskes, and R. Vlastou, Nucl. Instrum. Methods Phys. Res. B 122 542 (1997). 141 L. Romano, G. Impellizzeri, M.V. T omasello, F. Giannazzo, C. Spinella, and M.G. Grimaldi, J. Appl. Phys. 107 084314 (2010). 142 B.R. Yates, B.L. Darby, R.G. Elliman, and K.S. Jones, Appl. Phys. Lett. 101 131907 (2012). 143 G. Impellizzeri, L. Romano, B. Fraboni, E. Scavetta, F. Ruffino C. Bongiorno, V. Privitera, and M.G. Grimaldi, Nanotech. 23 395604 (2012). 144 B.L. Darby, B.R. Yates, N.G. Rudawski, K.S. Jones, and A. Kontos, Nucl. Instrum. Methods Phys. Res. B 269 20 (2011). 145 C.O. Chui, K. Gopalakrishnan, P.B. Griffin, J.D. P lummer, and K.C. Saraswat, Appl. Phys. Lett. 83 3275 (2003).

PAGE 185

185 146 F. Panciera, P.F. Fazzini, M. Collet, J. Boucher, E. Bedel, and F. Cristiano, Appl. Phys. Lett. 97 012105 (2010). 147 D.H. Petersen, O. Hansen, R. Lin, and P.F. Nielsen, J. Appl. Phys. 10 4 013710 (2008). 148 D.H. Petersen, O. Hansen, T.M. Hansen, P. Boggild, R. Lin, D. Kjaer, P.F. Nielsen, T. Clarysse, W. Vandervorst, E. Rosseel, N.S. Bennett, and N.E.B. Cowern, J. Vacuum Sci. Technol. B 28 C1C27 (2010). 149 A.F. Saavedra, K.S. Jones, M .E. Law, K.K. Chan, and E.C. Jones, J. Appl. Phys. 96 1891 (2004). 150 D.P. Hickey, Z.L. Bryan, K.S. Jones, R.G. Elliman, and E.E. Haller, J. Vacuum Sci. Technol. B 26 425 (2008). 151 A. Satta, T. Janssens, T. Clarysse, E. Simoen, M. Meuris, A. Benedet ti, I. Hoflijk, B.D. Jaeger, C. Demeurisse, and W. Vandervorst, J. Vacuum Sci. Technol. B 24 494 (2006). 152 J. Oh and J.C. Campbell, J. of Elec tron. Mater. 33 364 (2004). 153 A. Chroneos, J. Appl. Phys. 105 056101 (2009). 154 Fei Wang, D.H. Petersen F.W. Osterberg, and O. Hansen, in Advanced Thermal (2009), pp. 1 6. 155 S. Mirabella, E. Bruno, F. Priolo, D. De Salvador, E. Napolitani, A.V. Drigo, and A. Carnera, Appl. Ph ys. Lett 83 680 (2003). 156 F. Cristiano, X. Hebras, N. Cherkashin, A. Claverie, W. Lerch, and S. Paul, Appl. Phys. Lett 83 5407 (2003). 157 S. Boninelli, S. Mirabella, E. Bruno, F. Priolo, F. Cristiano, A. Claverie, D. De Salvador, G. Bisognin, and E. Napolitani, Appl. Phys. Lett. 91 031905 (2007). 158 M. Ngamo, S. Duguay, F. Cristiano, K. Daoud Ketata, and P. Pareige, J. Appl. Phys. 105 104904 (2009). 159 P. Lpez, L. Pelaz, I. Santos, L.A. Marqus, and M. Aboy, J. Appl. Phys. 111 033519 (2012) 160 M. J. Caturla, T. Daz de la Rubia, L.A. Marqus, and G.H. Gilmer, Phys. Rev. B 54 16683 (1996). 161 P. Delugas and V. Fiorentini, Phys. Rev. B 69 085203 (2004).

PAGE 186

186 162 K. Sueoka and J. Vanhellemont, Mater. Sci. Semicond. Process. 9 494 (2006). 1 63 J. Appl. Phys. 101 036103 (2007). 164 A. Giese, N.A. Stolwijk, and H. Bracht, Appl. Phys. Lett. 77 642 (2000). 165 J. Vanhellemont and E. Simoen, J. Electrochem. Soc. 154 H572 (2007). 166 M. Werner, H. Mehrer, and H.D. Hochheimer, Phys. Rev. B 32 3930 (1985). 167 H.D. Fuchs, W. Walukiewicz, E.E. Haller, W. Dondl, R. Schorer, G. Abstreiter, A.I. Rudnev, A.V. Tikhomirov, and V.I. Ozhogin, Phys. Rev. B 51 16817 (1995). 168 E. Silveira, W. Dondl, G. Abst reiter, and E.E. Haller, Phys. Rev. B 56 2062 (1997). 169 P. Lopez, Personal Communication, University of Valladolid, 2012. 170 D.F. Downey, J.W. Chow, E. Ishida, and K.S. Jones, Appl. Phys. Lett. 73 1263 (1998). 171 P.W. Loscutoff and S.F. Bent, Ann. Rev. of Phys. Chem. 57 467 (2006). 172 D.P. Hickey, Z.L. Bryan, K.S. Jones, R.G. Elliman, and E.E. Haller, Appl. Phys. Lett. 90 132114 (2007). 173 G. Bisognin, S. Van gelista, and E. Bruno, Mater. Sci and Eng. B 154 155 64 (2008). 174 S. Koffel, N. Cherkashin, F. Houdellier, M.J. Hytch, G. Benassayag, P. Scheiblin, and A. Claverie, J. Appl. Phys. 105 126110 (2009). 175 F. Panciera, P.F. Fazzini, M. Collet, J. Boucher, E. Bedel, and F. Appl. Phys. Lett. 97 012105 (2010). 176 R. Crosby, K.S. Jones, M.E. Law, A.N. Larsen, and J.L. Hansen, J. Vacuum Sci. Technol. B 22 468 (2004). 177 C.A. Ferreira Lima and A. Howie, Phil Mag. 34 1057 (1976). 178 S. Furuno, K. Izui, and H. Otsu, Japanese J. Appl. Phys. 15 889 (1976). 179 M. Pasemann, D. Hoehl, A .L. Ase ev, and O.P. Pchelyakov, Phys. Stat. Sol. A 80 135 (1983).

PAGE 187

187 180 A.L. Aseev, V.M. Ivakhnishin, V.F Stas, and L.S. Smirnov, Sov. Phys. Sol. State 25 1786 (1983). 181 H. Bartsch, D. Hoehl, and G. Kstner, Phys. Stat. Sol. A 83 543 (1984). 182 J.L Hutchinson, A.L. Aseev, and L.I. Fedina, in Microscopy of Semiconducting Materials (1993), pp. 41 46. 183 T. Akatsu, K.K. Bourdelle, C. Richtarch, B. Faure, and F. Letertre, Appl. Phys. Lett. 86 181910 (2005). 184 S. Rubanov and P.R. Munroe, Micron 35 549 (2004). 185 C.A. Schneider, W.S. Rasband, a nd K.W. Eliceiri, Nat. Meth. 9 671 (2012). 186 S. Boninelli, G. Impellizzeri, A. Alberti, F. Priolo, F. Cristiano, and C. Spinella, Appl. Phys. Lett. 101 (2012). 187 F.F. Morehead and B.L. Crowder, Rad. Effect. 6 27 (1970). 188 L.A. Christel, J.F. Gibbons, and T.W. Sigmon, J. Appl. Phys. 52 7143 (1981). 189 R.G. Elliman, J.S. Williams, W.L. Brown, A. Leiberich, D.M. Maher, and R.V. Knoell, Nucl. Instrum. Methods Phys. Res. B 19 20, Part 2 435 (1987). 190 R.G. Elliman, S.T. Johnson, A.P. Pogany, and J.S. Williams, Nucl. Instrum. Methods Phys. Res. B 7 8, Part 1 310 (1985). 191 D. Sigurd, G. Fladda, L. Eriksson, and K. Bjrkqvist, Rad. Effect. 3 145 (1970). 192 G. Holmn, S. Peterstrm, A. Burn, a nd E. B gh, Rad. Effect. 24 45 (1975). 193 J. Linnros and G. Holmn, J. Appl. Phys. 62 4737 (1987).

PAGE 188

188 BIOGRAPHICAL SKETCH Bradley Robe rt Yates is the son of Charlene Leyda and Ja mes Yates. He was born in Pittsburgh, Pennsylvania. He was raised in a s mall town located about an hour south of Pittsburgh called Belle Vernon He attended Belle Vernon Area High School where he played football and graduated in June 2004. He attended Carnegie Mellon University in Pittsburgh, Pennsylvania where he double maj ored in Materials Science and Biomedical Engineering in addition to playing defensive end for the Tartans for two years. His professional football hopes were dashed by concussion issues which steered him to attend graduate school following the culmination of his undergraduate career in May 2008. He attended the University of Florida in Gainesville, Florida in August 2008 where he began working for Dr. Kevin Jones on the work discussed herein in addition to work on Group IV anode materials for lithium ion batteries, the formation of nanoporous Ge, and electroless etching of silicon among other side projects. Upon completion of his Ph. D. degree, he will be joining Intel in Hillsboro, Oregon as a n engineer within P ortland Technolog y Development In the future, he p lans to retire young and purchase a large fu e l efficient boat to cruise around the Caribbean Sea.