<%BANNER%>

Ion Implantation Induced Defect Formation and Amorphization in the Group IV Semiconductors

Permanent Link: http://ufdc.ufl.edu/UFE0021224/00001

Material Information

Title: Ion Implantation Induced Defect Formation and Amorphization in the Group IV Semiconductors Diamond, Silicon, and Germanium
Physical Description: 1 online resource (197 p.)
Language: english
Creator: Hickey, Diane Patricia
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: crystal, diamond, germanium, implantation, implants, ion, semiconductor, silicon, single
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: Silicon, which has been the workhorse of the semiconductor industry for the past several decades, is now being enhanced with other Group IV elements, such as carbon (silicon carbide) and germanium (silicon-germanium strained channels in transistors), to accentuate properties of silicon for various nanoelectronic devices. However, there is little understanding of the relationship between ion implantation and defect evolution in two of the three corners of the Group IV phase diagram. In particular, the rod-like {311} defect is theorized to be unique to the diamond crystal structure elements. Due to its ability to affect dopant diffusion, the {311} defect is well studied in silicon. However, few studies of germanium and none of diamond have analyzed extended defect formation and evolution using transmission electron spectroscopy. Using ion implantation to induce amorphization is a technological process step in Si devices and potentially for diamond nano-electronics. Defects associated with crystal regrowth in Ge and diamond are not well known. My research studies the formation conditions of extended defects and amorphization in carbon and germanium after ion implantation. Ion implantation damage in diamond-cubic single-crystal silicon (Si), germanium (Ge) and diamond was produced by Si^+ implantation at 1 MeV to a dose of 1 ? 10^14 cm^-2 and 1 ? 10^15 cm^-2. Damage in Si and Ge was produced by Si^+ implantation at 40 keV to a dose of 1 ? 10^14 cm^-2 and 1 ? 10^15 cm^-2, and amorphizing damage in diamond was produced by Si^+ implantation at 1 MeV to a dose of 3 and 7 ? 10^15 cm^-2. All implants were carried out at room temperature. For non-amorphizing implants (10^14 Si^+ cm^-2) into Ge, dot-like defects formed immediately upon implantation and were stable up to temperatures of 650 degrees Celsius. The activation energy of these defects was determined to be approximately 0.2 +/- 0.1 eV. For amorphizing implants (10^15 Si^+ cm^-2) into Ge and upon solid-phase epitaxial regrowth, the same types of defects seen in Si were also seen in Ge. However, in Ge, the end-of-range defects were the least stable, dissolving at temperatures around 650 degrees Celsius. The activation energy for the dissolution of end-of-range defects in Ge is approximately 0.4 +/- 0.1 eV. For diamond, non-amorphizing Si^+ implantation ( < 3 ? 10^15 Si^+ cm^-2) and subsequent annealing to temperatures as high as 1350 ?C for 24 hours resulted in no implantation related defects. Amorphizing implants (greater than or equal to 3 ? 10^15 Si^+ cm^-2) into diamond revealed a threshold for amorphization at approximately 6 ? 10^22 vacancies cm^-3, higher than previously reported values. The diamond did not convert to graphite immediately after implantation, but instead resulted in an amorphous carbon layer, with a density of 1.3 +/- 0.1 ? 10^23 atoms cm^-3, which is approximately 25% less dense than diamond and 15% more dense than graphite. After annealing at 1350 degrees Celsius for 24 hours, the amorphous carbon converted to graphite. These results provide one of the first comparative studies of these Group IV semiconductor?s defect evolution, thus potentially providing more accurate design of simulation programs and future electronic devices.
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 Diane Patricia Hickey.
Thesis: Thesis (Ph.D.)--University of Florida, 2007.
Local: Adviser: Jones, Kevin S.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2009-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021224:00001

Permanent Link: http://ufdc.ufl.edu/UFE0021224/00001

Material Information

Title: Ion Implantation Induced Defect Formation and Amorphization in the Group IV Semiconductors Diamond, Silicon, and Germanium
Physical Description: 1 online resource (197 p.)
Language: english
Creator: Hickey, Diane Patricia
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2007

Subjects

Subjects / Keywords: crystal, diamond, germanium, implantation, implants, ion, semiconductor, silicon, single
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: Silicon, which has been the workhorse of the semiconductor industry for the past several decades, is now being enhanced with other Group IV elements, such as carbon (silicon carbide) and germanium (silicon-germanium strained channels in transistors), to accentuate properties of silicon for various nanoelectronic devices. However, there is little understanding of the relationship between ion implantation and defect evolution in two of the three corners of the Group IV phase diagram. In particular, the rod-like {311} defect is theorized to be unique to the diamond crystal structure elements. Due to its ability to affect dopant diffusion, the {311} defect is well studied in silicon. However, few studies of germanium and none of diamond have analyzed extended defect formation and evolution using transmission electron spectroscopy. Using ion implantation to induce amorphization is a technological process step in Si devices and potentially for diamond nano-electronics. Defects associated with crystal regrowth in Ge and diamond are not well known. My research studies the formation conditions of extended defects and amorphization in carbon and germanium after ion implantation. Ion implantation damage in diamond-cubic single-crystal silicon (Si), germanium (Ge) and diamond was produced by Si^+ implantation at 1 MeV to a dose of 1 ? 10^14 cm^-2 and 1 ? 10^15 cm^-2. Damage in Si and Ge was produced by Si^+ implantation at 40 keV to a dose of 1 ? 10^14 cm^-2 and 1 ? 10^15 cm^-2, and amorphizing damage in diamond was produced by Si^+ implantation at 1 MeV to a dose of 3 and 7 ? 10^15 cm^-2. All implants were carried out at room temperature. For non-amorphizing implants (10^14 Si^+ cm^-2) into Ge, dot-like defects formed immediately upon implantation and were stable up to temperatures of 650 degrees Celsius. The activation energy of these defects was determined to be approximately 0.2 +/- 0.1 eV. For amorphizing implants (10^15 Si^+ cm^-2) into Ge and upon solid-phase epitaxial regrowth, the same types of defects seen in Si were also seen in Ge. However, in Ge, the end-of-range defects were the least stable, dissolving at temperatures around 650 degrees Celsius. The activation energy for the dissolution of end-of-range defects in Ge is approximately 0.4 +/- 0.1 eV. For diamond, non-amorphizing Si^+ implantation ( < 3 ? 10^15 Si^+ cm^-2) and subsequent annealing to temperatures as high as 1350 ?C for 24 hours resulted in no implantation related defects. Amorphizing implants (greater than or equal to 3 ? 10^15 Si^+ cm^-2) into diamond revealed a threshold for amorphization at approximately 6 ? 10^22 vacancies cm^-3, higher than previously reported values. The diamond did not convert to graphite immediately after implantation, but instead resulted in an amorphous carbon layer, with a density of 1.3 +/- 0.1 ? 10^23 atoms cm^-3, which is approximately 25% less dense than diamond and 15% more dense than graphite. After annealing at 1350 degrees Celsius for 24 hours, the amorphous carbon converted to graphite. These results provide one of the first comparative studies of these Group IV semiconductor?s defect evolution, thus potentially providing more accurate design of simulation programs and future electronic devices.
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 Diane Patricia Hickey.
Thesis: Thesis (Ph.D.)--University of Florida, 2007.
Local: Adviser: Jones, Kevin S.
Electronic Access: RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2009-08-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2007
System ID: UFE0021224:00001


This item has the following downloads:


Full Text

PAGE 1

1 ION IMPLANTATION INDUCED DEFECT FORMATION AND AMORPHIZATION IN THE GROUP IV SEMICONDUCTORS: DIAMOND, SILICON AND GERMANIUM By DIANE P. HICKEY A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2007

PAGE 2

2 2007 Diane P. Hickey

PAGE 3

3 For my parents, Genevieve and Howard Hickey

PAGE 4

4 ACKNOWLEDGMENTS I acknowledge and thank the members of my dissertation committee. Dr. Kevin S. Jones, my advisor, has been an excel lent guide throughout the gradua te experience, and has provided many insightful thoughts into the workings of science outreach and the semiconductor industry. Dr. Mark Law, the Software Analysis of Ma terials Processing (SWAMP ) Group co-advisor, has significantly helped in the organi zation of this dissertation, and al so has put an indelible mark on my memory from the graduation ceremony for my Masters degree, when he said, Youre a Gator now. Dr. Paul Holloway has answered ma ny questions about vacuum systems from my original project on sili con micro-electrical-mechanical syst ems, and was a wonderful advisor while I was chair of the stude nt chapter of AVS Science a nd Technology Society. Dr. Stephen Pearton, an Australian from Tasm ania, taught the first class that I took at the University of Florida, and through that class assu red me that the decision to retu rn to graduate school was the right decision. Finally, Dr. Scott Thompson ha s provided excellent i ndustrial council and technical knowledge, and provided refreshing firsthand knowledge of the ways in research and industry interact. Dr. Rob Elliman, of the Australian National Un iversity, provided the implants for this project, as well has been an excellent advisor who answered questions by email and was available for long discussions during MRS meetings. Dr. E.E. Haller, of Lawrence Berkley Laboratory and the Department of Materials Science and Engineering at the University of California Berkeley, provided the high-purity germanium for this study, and co-authored the germanium papers that resulted from this wo rk. Dr. Rob Chodelka was the original industrial science advisor for the diamond work presente d here. The Gemesis Corporation graciously provided single-crystal diamond wafers for this st udy. Most of the analytical equipment used for this study is housed at the Major An alytical Instrumentation Center at the University of Florida.

PAGE 5

5 Several members of its staff were instrumental in the completion of this work, including Dr. Jerry Bourne and Kerry Siebein. Ms. Siebein also co-authored one of the papers resulting from this study. Dr. Gerhard Fuchs and his group provide d the furnace used for the high-temperature diamond anneals. Special thanks to Brandon W ilson for the operation an d maintenance of the furnace, and especially his time during the anneals. Raman studies in this work were performed by Dr. David Hahn, of the University of Flor idas Mechanical and Aerospace Engineering Department. Electron Energy Loss Spectroscopy was performed at the Univ ersity of Central Florida. Financial support for this work was partia lly provided by the Semiconductor Research Corporation contract #00057787. Sp ecial thanks are extended to the Applied Materials corporation, and Dr. Susan Felch, for the Gra duate Student Fellowship. Smaller monetary fellowships were also provided by the University of Floridas Association of Academic Women and the Office of the President, as well as the AVS Manufacturing Science and Technology Group. There are many other faculty at the University of Florida who have been instrumental to the papers and presentations I created during my tenure here. Dr. Simon Philpott provided insight during general defect discussi ons regarding germanium and diamond, as well as added insight into certain simulations. Dr. Susan Sinnott serv ed as co-advisor to the AVS Science and Technology student chapter, and al so provided letters of recommenda tion that assisted in landing various fellowships. Dr. Debbie Treise served as my advisor for the Science Communication Masters degree program in the College of Journalism. Dr. Rob Fox was always available to answer questions pertaining to t echnical electrical engineering or Auburn Football. Others at the University of Florida have made it a pleas ure to spend time here. Ludie Harmon, Matthew

PAGE 6

6 Walters and Teresa Stephens have been absolute joys to work with in the Materials Science and Engineering department. The SWAMP Group has been an excellent group of people with whom to work. They include Dr. Mark Clark, Dr. Renata Camillo-Cas tillo, Dr. Erik Kuryliw, Dr. Robert Crosby, Dr. Russ Robison, Dr. Kevin Gable and Dr. Jennette Jaques. Current members of the SWAMP group continue to provide support and encouragement throughout the process, including Michelle Phen, Sam Moore, Leah Eldemen, Dan Gostovec, Sidan Jin, Saurabh Morarka, and Aijie Chen. Special thanks is given to Ni ck Rudawski who, in addition to being a current SWAMP group member, also kindly edited this document Sherry Huo and David Kundinger are two undergraduates working with the group, and I wish them the best in their future endeavors. Thanks go to William Tre Harrison who was my first research partner during the microelectrical-mechanical systems proj ect, as well as an excellent volleyball partner. Special thanks to Zachary Bryan, who has worked with me fo r two years in his research education for undergraduates position, and prepared most of the plan-views presented in this work, as well as co-authored two papers. Many friends encouraged me in the pursuit of an engineering Ph.D. degree. Emmalyn Fuller-Garvett, Stephanie Bailey, Danielle Gigl io-Danas, Jennifer Hussey-Kilgo, Hope Blass, Katie Allen, Amy Anderson-Whigham and Beth Miller all provided encouragement throughout the process. A couple of my good friends encouraged me to do well in life, but not go to grad school, presumably because they themselves were in the midst of their dissertation writing Dr. V. Carol Chancey and Dr. Karla McInnis are specially acknowledged for support and insight. Anna Fuller, Stephen Tedeschi, Abby Queale, Dr Carol Kem and Mike Davis, who have been a wealth of inspiration, fun and enco uragement during this experience.

PAGE 7

7 I have been blessed with the most wonderful family, who has encouraged and supported me throughout my life. Nothing that I have ever achieved or will achieve could have been done without them. For this, I would lik e to thank my all of my aunts, uncles, cousins, and closer family, Anthony Huffstetler, Brandon Huffstetle r, Ryan Hickey, Karen Young, and especially my siblings who have always looked out for th eir baby sister Lisa Hickey, Susan Huffstetler and Howard Hickey. And most of all, to my parents, Howard and Genevieve Hickey, without whose love and support none of this would be possible.

PAGE 8

8 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ........11 LIST OF FIGURES................................................................................................................ .......12 ABSTRACT....................................................................................................................... ............15 CHAPTER 1 INTRODUCTION & MOTIVATION...................................................................................17 Interest in Alternative Se miconductor Base Materials...........................................................17 The Cause and Problem of Extended Defects........................................................................18 Objective of Research.......................................................................................................... ...20 2 LITERATURE REVIEW.......................................................................................................25 Introduction................................................................................................................... ..........25 General Concepts............................................................................................................... .....25 Ion Implantation..............................................................................................................25 Defects in the Crystalline Lattice....................................................................................29 Amorphization.................................................................................................................32 Solid Phase Epitaxial Regrowth......................................................................................35 Simulation Process Tools................................................................................................35 Relevant Literature for Ge, SiGe, SiC, and C.........................................................................38 Germanium Defects and Amorphization......................................................................38 Defects in Pure Ge....................................................................................................38 Amorphization, Regrowth and Lack of EOR Damage of Ge..................................40 SiGe Alloy Defects and Amorphization.......................................................................41 Silicon Carbide (SiC)Ion Implantation and Amorphization.........................................42 Diamond........................................................................................................................ ..43 Ion Implantation in Diamond...................................................................................43 Controversies in Diamond Literature.......................................................................47 Properties of the Elemen tal Crystal Systems..........................................................................49 3 ANALYTICAL TECHNIQUES............................................................................................51 Introduction................................................................................................................... ..........51 Sample Preparation for TEM..................................................................................................52 Transmission Electron Microscopy........................................................................................58 Energy Dispersive X-Ray Spectroscopy................................................................................61 Electron Energy Loss Spectroscopy.......................................................................................62 Raman Spectroscopy............................................................................................................. .64

PAGE 9

9 Stopping and Range of Ions in Matter (SRIM)......................................................................66 4 RESULTS FROM NON-AMORPHIZI NG IMPLANTS INTO GERMANIUM..................85 Introduction................................................................................................................... ..........85 Experimental Design............................................................................................................ ..86 Results........................................................................................................................ .............87 Discussion..................................................................................................................... ..........89 Defect Density Analysis..................................................................................................89 Defect Dissolution Rate...................................................................................................90 Defect Location Analysis................................................................................................91 Conclusion..................................................................................................................... .........92 5 RESULTS FROM AMORPHIZING IMPLANTS INTO GERMANIUM..........................102 Introduction................................................................................................................... ........102 Experimental Design............................................................................................................104 Results........................................................................................................................ ...........104 High Energy Implant 1 MeV......................................................................................104 Low Energy Implant 40 keV......................................................................................106 Discussion..................................................................................................................... ........108 Defects........................................................................................................................ ...108 Trapped Interstitials and Activation Energy of 1 MeV EOR Defects in Ge.................110 Threshold Damage Density for Ge................................................................................111 Conclusion..................................................................................................................... .......112 6 RESULTS FROM NON-AMORPHIZING ME V IMPLANTS INTO DIAMOND............125 Introduction................................................................................................................... ........126 Experimental Design............................................................................................................127 Preliminary Experiments...............................................................................................127 Presented Experiment....................................................................................................127 Results........................................................................................................................ ...........129 As-implanted Comparison: Orie ntation, Dose, Temperature........................................129 Annealing: Low Dose....................................................................................................130 Annealing: High Dose...................................................................................................130 Annealing: Implant Temperature Effect........................................................................131 Discussion..................................................................................................................... ........132 Conclusion..................................................................................................................... .......134 7 RESULTS FROM AMORPHIZING ME V IMPLANTS INTO DIAMOND......................145 Introduction................................................................................................................... ........145 Experimental Design............................................................................................................146 Results........................................................................................................................ ...........147 TRIM Results................................................................................................................147 Swelling of Diamond Experiment.................................................................................148

PAGE 10

10 Density of As-implanted Amorphous ( ) C Layer.......................................................149 Annealing Ion-implantation Induced Amorphous C.....................................................150 Determination of the Graphitic Nature of Annealed Diamond.....................................152 EELS Results of the Amorphous Region......................................................................153 Annealed Amorphous-Crystalline Interfaces................................................................154 Discussion..................................................................................................................... ........154 Conclusion..................................................................................................................... .......157 8 SUMMARY AND FUTURE WORK..................................................................................171 Summary........................................................................................................................ .......171 Non-amorphizing Si+ Implants into Diamond and Ge, as Compared to Si...................171 Amorphizing Si+ Implants into Diamond and Ge, as Compared to Si..........................172 Future Work.................................................................................................................... ......173 APPENDIX A SURFACE INITIATED DEFECT S OF ANNEALED DIAMOND....................................177 B BORON IMPLANTATION INTO DIAMOND..................................................................182 Introduction................................................................................................................... ........182 Experimental Design............................................................................................................184 Results & Discussion........................................................................................................... .185 LIST OF REFERENCES.............................................................................................................187 BIOGRAPHICAL SKETCH.......................................................................................................197

PAGE 11

11 LIST OF TABLES Table page 1-1 The mobility of carriers in semiconduc tors, as reported in the September 6, 2002 issue of Science............................................................................................................... ...22 2-1 Relevant properties of the Group IV di amond-cubic semiconductors and graphite..........50 3-1 Major Raman peaks from diamond, graphite and related materials. 7...............................80 5-1 Simulated damage values from current experiments.......................................................124 6-1 Implantation parameters...................................................................................................137 7-1 Simulated damage values from current experiments.......................................................170

PAGE 12

12 LIST OF FIGURES Figure page 1-1 Cross-sectional TEM image of Intel Pentium transistors showing the appearance of dislocation loop defects. Courtesy of Intel and Scott Thompson......................................23 1-2 A comparison of the reported {311} defect in (a) Si and (b) Ge.......................................24 3-1 In-situ lift-out of diamond wedge, as vi ewed using the ion beam which is at a 52 angle to the stage............................................................................................................. ...69 3-2 The sample thickness before (a) and after (b) the thinning process..................................70 3-3 A TEM image of an as-implanted (110) single crystal diamond implanted with a 1 MeV 1x1014 Si+ cm-2 at a temperature of 30 C...............................................................71 3-4 XTEM of (100) orient ed diamond after a 1 MeV 1x1015 cm-2 Si+ implant (Ti = 30 C) annealed at 1050 C for 10 min: (a) XTEM image (labels denote platinum, carbon coating and surface) (b) SADP of top region (c) SADP of bulk.......................................72 3-5 HRTEM image showing the middle of a damaged diamond layer from a (001)oriented diamond implanted at 77 K with 1 MeV, 115 Si+/cm2...................................73 3-6 Schematic diagram of the signals cr eated during transmission of a high-energy electron beam through a thin solid sample. ......................................................................74 3-7 Single-crystal spot tr ansmission electron diffracti on patterns for the diamond structures (diamond, Si and Ge) for the (a ) [001] and (b) [011] zone axis, and the hexagonal crystal structure for the (c) [2110] and (d) [0110] zone...................................75 3-8 De-excitation mechanisms for an atom that has under gone K-shell ionization: (a) emission of a characteristic K X-ray and (b) emission of a KLM Auger electron. ........76 3-9 TEM and EDS characterization of sample.........................................................................77 3-10 The low-loss and K-ionization regions of the energy-loss spectra of three allotropes of carbon as detected using an EELS system. ...................................................................78 3-11 Screenshot of data displayed from the EELS system used for this study..........................79 3-12 Raman spectra of diamond.................................................................................................81 3-13 Raman spectra of carbon allotropes...................................................................................82 3-14 Comparison of simulation data to experimental images....................................................83 3-15 Comparison of simulation data to experimental images....................................................84

PAGE 13

13 4-1 WBDF-XTEM images of (a ) dot-like defects in Ge Si+ implanted at 1 MeV with a dose of 1 1014 cm-2 following annealing at 45 0 C for 10 min and (b) {311} defects in Si with the sample implantati on following annealing at 850 C for 10 min.....94 4-2 High resolution BF-TEM image of a dot-like defect seen in Ge Si+ implanted at 1 MeV with a dose of 1 1014 cm-2 after annealing for 10 min at 450 C...........................95 4-3 WBDF-XTEM images of Ge and Si Si+ implanted at 1 MeV with a dose of 1 1014 cm-2 after 10 min of annealing at various temperatures.....................................................96 4-4 WBDF images of Si Si+ implanted at 40 keV with a dose of 1 1014 cm-2 after annealing for 10 min at 750 C:(a) XTEM and (b) PTEM................................................97 4-5 WBDF-XTEM images of Ge Si+ implanted at 40 keV with a dose of 1 1014 cm-(a) as implanted, and annealed for 10 min at (b) 350 (c) 450 (d)550 (e)650 (f)750 C.........98 4-6 WBDF images of Ge Si+ implanted at 40 keV with a dose of 1 1014 cm-2 and annealed for 10 min at 550 C:(a) XTEM and (b) PTEM.................................................99 4-7 Microstructure evolutio n of Ge and Si after annealing for 10 min at various temperatures................................................................................................................... ..100 4-8 Plot of the logarithm of vs. 1/kT for 1 MeV Si+ implantation into Ge after 10 min of annealing................................................................................................................... ...101 5-1 WBDF-XTEM images taken of Ge Si+ implanted at 1 MeV to a dose of 1 1015 cm-2 (a) as-implanted and (b) afte r 10 min of annealing at 350 C..........................................114 5-2 HRTEM images of Ge Si+ implanted at 1 MeV to a dose of 1 1015 cm-2 asimplanted...................................................................................................................... ....115 5-3 WBDF XTEM images of Ge Si+ implanted at 1 MeV to a dose of 1 1015 cm-2 after annealing for 10 min at various temperatures..................................................................116 5-4 XTEM images of Ge Si+ implanted at 40 keV to a dose of 1 1015 cm-2 at 303 K(a) as-implanted and after anneali ng for 10 min at (b) 350 and (c) 380 C for 10 min.........117 5-5 HR-XTEM images taken of Ge Si+ implanted at 40 keV to a dose of 1 1015 cm-2......118 5-6 WBDF-XTEM images taken of Ge Si+ implanted at 40 keV to a dose of 1 1015 cm2 at 303 K and annealed for 10 min at (a) 450 (b) 550 (c) 650 and (d) 750 C................119 5-7 WBDF-XTEM images taken of Si Si+ implanted at 40 keV to a dose of 1 1015 cm-2 at 303 K and annealed for 10 min at 750 C (a) XTEM and (b) PTEM..........................120 5-8 Graphical representations of the temper ature dependence of the Ge EOR (a) defect density, (b) defect size, (c ) dissolution rate and (d ) trapped interstitials.........................121

PAGE 14

14 5-9 XTEM images of Ge Si+ implanted at 1 MeV to a dose of 1 1015 cm-2 at 303 K.........122 5-10 XTEM images of Ge Si+ implanted at 40 keV to a dose of 1 1015 cm-2 at 303 K........123 6-1 WBDF-XTEM images of (001)-oriented single-crystal diamond Si+ implanted at 1 MeV to a dose of 1 1015 cm-2 at (a) 77K and (b) 303K................................................138 6-2 BF TEM images of (011)-oriented diamond Si+ implanted at 1 MeV to a dose of 1 1014 cm-2 Si+ at 303K.......................................................................................................139 6-3 WBDF-XTEM image of (001)-oriented diamond Si+ implanted at 1 MeV to a dose of 1 1014 cm-2 at 303K after annealing at 1250 C for 1 hr..........................................140 6-4 BF TEM image of (110)-oriented diamond Si+ implanted at 1 MeV to a dose of 1 1015 cm-2 at 303K.............................................................................................................141 6-5 WBDF-XTEM images of (001)-oriented diamond Si+ implanted at 1 MeV to a dose of 1 1015 cm-2 at 303K..................................................................................................142 6-6 WBDF-XTEM images of (001)-oriented diamond Si+ implanted at 1 MeV to a dose of 1 1015 cm-2 at 77K....................................................................................................143 6-7 HRTEM image of damaged diamon d layer in (001)-oriented diamond Si+ implanted at 1 MeV to a dose of 1 1015 cm-2 after annealing at 1350 C for 24 hrs.....................144 A-1 WBDF XTEM images of single-crystal diamond Si+ implanted at 1 MeV to a dose of 3 1015 cm-2 at 303 K after annealing for 24 hrs at 1350 C..........................................178 A-2 WBDF XTEM images of single-crystal diamond Si+ implanted at 1 MeV to a dose of 1 1015 cm-2 at 77 K after annealing for 24 hrs at 1350 C............................................179 A-3 Images of single-crystal diamond Si+ implanted at 1 MeV to a dose of 1 1014 cm-2 at 303 K after annealing for 2 hrs at 950 C....................................................................180 A-4 BF-XTEM images of single-crystal diam ond un-implanted, after annealing for 2 hrs at 950 C...........................................................................................................................181 B-1 WBDF-XTEM images of (001)-oriented diamond B+ implanted at 2 MeV to a dose of 1 1015 cm-2 at 303K as implanted, and af ter annealing (a) in flowing N2 ambient at 1000 C for 10 min and (b) in vacuum at 1350 C for 24 hrs......................................186

PAGE 15

15 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ION IMPLANTATION INDUCED DEFECT FORMATION AND AMORPHIZATION IN THE GROUP IV SEMICONDUCTORS: DIAMOND, SILICON AND GERMANIUM By Diane P. Hickey August 2007 Chair: Kevin S. Jones Major: Materials Science and Engineering Silicon, which has been the workhorse of the semiconductor industry for the past several decades, is now being enhanced with other Group IV elements, such as carbon (silicon carbide) and germanium (silicon-germanium st rained channels in transistors) to accentuate properties of silicon for various nanoelectronic devices. Ho wever, there is little understanding of the relationship between ion implantation and defect evolution in two of th e three corners of the Group IV phase diagram. In particular, the rod-li ke {311} defect is theorized to be unique to the diamond crystal structure elements. Due to its ability to affect dopant diffusion, the {311} defect is well studied in silicon. However, few st udies of germanium and none of diamond have analyzed extended defect formation and evol ution using transmission electron spectroscopy. Using ion implantation to induce am orphization is a techno logical process step in Si devices and potentially for diamond nano-electro nics. Defects associated with crystal regrowth in Ge and diamond are not well known. My research studies the formation conditions of extended defects and amorphization in carbon and ge rmanium after ion implantation. Ion implantation damage in diamond-cubic si ngle-crystal silicon, germanium and diamond was produced by Si+ implantation at 1 MeV to a dose of 1 1014 cm-2 and 1 1015 cm-2. Damage in Si and Ge was produced by Si+ implantation at 40 keV to a dose of 1 1014 cm-2 and 1 1015

PAGE 16

16 cm-2, and amorphizing damage in diamond was produced by Si+ implantation at 1 MeV to a dose of 3 and 7 1015 cm-2. All implants were carried out at room temperature. For non-amorphizing implants (1014 Si+ cm-2) into Ge, dot-like defects formed immediately upon implantation and were stable up to temperatures of 650 C. The activation energy of these defects was determined to be approximately 0.2 0.1 eV. For amorphizing implants (1015 Si+ cm-2) into Ge and upon solid-phase epitaxial regrowth, th e same types of defects seen in Si were also seen in Ge. However, in Ge, the end-of-range defects were th e least stable, dissolvi ng at temperatures around 650 C. The activation energy for th e dissolution of end-of-range defects in Ge is approximately 0.4 0.1 eV. For diamond, non-amorphizing Si+ implantation (< 3 1015 Si+ cm-2) and subsequent annealing to temper atures as high as 1350 C for 24 hours resulted in no implantation related defects. Amorphizing implants ( 3 1015 Si+ cm-2) into diamond revealed a threshold for amorphization at approximately 6 1022 vacancies cm-3, higher than previously reported values. The diamond did not convert to graph ite immediately after im plantation, but instead resulted in an amorphous carbon layer, with a density of 1.3 0.1 1023 atoms cm-3, which is approximately 25% less dense than diamond and 15% more dense than graphite. After annealing at 1350 C for 24 hours, the amorphous carbon convert ed to graphite. These results provide one of the first comparative studies of these Gr oup IV semiconductors defect evolution, thus potentially providing more accurate design of simula tion programs and future electronic devices.

PAGE 17

17 CHAPTER 1 INTRODUCTION & MOTIVATION Interest in Alternative Se miconductor Base Materials Silicon (Si) has been the main semiconductor ba se material for the past several decades. Although germanium (Ge) was the first transistor material,1 Sis low cost and stable native oxide made it the more attractive material for microele ctronic devices. Silicons low cost and decades of research ensure its continued dominance over the next few years, but a theoretical limit to simple scaling does exist. In order to stay on track with Moores Law,2 many technologies are being explored, intended to bot h extend silicons life time and, in some devices, to replace silicon. The 2005 Interna tional Technology Roadmap for Semiconductors3 highlights that research into new base materials is extrem ely important for the semiconductor industry. Two attractive base materials ar e carbon (C), in the diamond form,4 and Ge.1 Both have higher carrier mobilities than Si (Table 1-1). Both are elemental crystals, and do not suffer the complications ion-implantation into higher-elec tron mobility III-V semic onductors like gallium arsenide. Additionally, with the advent of high-k dielectric gate oxides, the need for a stable native oxide in these materials becomes less impo rtant. Ge and C are already being used as important semiconductor materials when alloyed with silicon. Silicon carbide (SiC) is one of the hardest materials known and produced today, and s ilicon-germanium (SiGe) has been used to increase performance in sub-100 nanometer (nm) transistors. As the semiconductor industry explores other materials besides Si, leveraging decades of Si research could speed development time for new materials. Currently, in the pursuit for ultrashallow junction devices, extended defects are a ma jor concern in Si because defects can impact device performance.5 Figure 1-1 shows a transistor with di slocation loop defects apparent in the transmission electron microscope (TEM) image. The question remains as to whether these same

PAGE 18

18 extended defects would form in alternate material devices, and, if so, would the defects be as detrimental to device performance as in Si. The Cause and Problem of Extended Defects Extended defects arise after ion implantation. Ion implantation is necessary to dope the base material to impart properties desirable in the electronic device Extended defects are irregularities in the crystal that are large agglom erations of point defect s, such as interstitial atoms or vacancies, or a combination of both. Co mmon extended defects in Si, resulting from the agglomeration of interstitial at oms, are the {311} and dislocati on loop defects. The {311} defect is the source of the dislocation loop. 6 These defects have affected Si device performance for many years, inhibiting carrier mobility and therefore increasing leakage current. The {311} defect is unique to semiconductors. This defect has never been observed in metals, and is theorized to be the stable inte rstitial defect in all di amond-structure crystals.7 However, it has only been reported in two elem ents Si and Ge. In Si, the {311} defect has formed during electron irradiation7,8 and for a variety of ion impla ntation regimes at both high and low energies. 9 However, the {311} defect has onl y been reported in Germanium in a handful of situations electron irradiation and lig ht element implantation. 5,8,10-12 As yet, the defect in Ge has not been demons trated using a typical dopant atom. In addition, the direct compar ison of a {311} defect in Si to a {311} defect in Ge raises questions. For example, Figure 1-2(a) shows an example of a high-reso lution (HR) TEM image of a {311} defect in Si. It shows a well-organized defect struct ure. Figure 1-2(b), is a HRTEM image of a {311} defect in Ge on the same scal e, created after a hydrogen (H) implant. Notice the jagged appearance of the Ge {311} as compared to the Si {311}. Since these differences have been seen in pure Ge versus pure Si, Crosby et al. sought to understand the {311} defect formation in SiGe a lloys. He found that the defect formation

PAGE 19

19 changed as a function of alloy concentration. In SixGe1-x alloys with a Ge concentration less than 25%, {311} defects continued to form; above 25 % Ge content, {311} defects did not appear. Crosby concluded that the weaker bond strength of Ge prevented the formation of {311} defects. 13 Additional differences in defect formation have been seen recently: an analysis of 10, 20 and 30% Ge concentrations in SiGe alloys reve aled near surface vacancy nanocavities with increasing Ge concentration.14 Logically proceeding from Crosbys conclu sion that the weaker bond strength of Ge prevents {311} formation, an element with a stronger bond strength, like C, should encourage {311} defect formation. However, it is well known that {311} formation in Si is actually inhibited by the presence of C. Ra ther than bond strength, this phenom ena is attributed to the size difference of C to Si (C atoms are much smalle r than Si), which allo ws interaction with {311} defect-forming interstitials and essentially traps the intersti tials preventing the atoms from migrating to agglomerate.15 However, if the entire crystal was C, rather than just an implant, the {311} defect is still theorized to be the stable interstitial defect.7 Advances in man-made single-crystal diamond development have progressed to a point in whic h, due to diamonds desirable qualities, diamond as a substrate for microelectr onics is a growing possibility.4 Diamond may also form the {311} defect as an interstitial precipitate, but due to the difficulty of creati ng appropriate TEM samples this remains unconfirmed. Additionally, one scenar io that creates extended defects in Si is the regrowth from an amorphous state. Using an amorphizing implant prior to a doping implant has become common practice in industry to prevent the diffusion of the dopant species away from the channel region and to increa se dopant activation. Therefore, defect formation during this process is also of interest. However, past a certain damage point, diamond will not transition

PAGE 20

20 back to a diamond cubic lattice but will instead undergo a phase transition to graphite upon annealing.16 Solid phase epitaxital regrowth, the direct transition to graphite, or extended defects have not been previously demonstrated in di amond using direct imaging techniques, like TEM. Therefore, there is no imaging evidence of extended defects in diamond. Objective of Research These differences highlight the fact th at even though these are all Group IV semiconductors, behavior can vary greatly wh en processed using standard semiconductor techniques. The theory that the {311} defect is the stab le interstitial defect in all diamond-cubic structure elements, and the similarities inherent in being in the same column of the periodic table, allow a hypothesis that the same defects should occur in all three materials after the same implant conditions. However, the recent research di scussed in the previous paragraphs indicates that this hypothesis may not be true in all situations. My research studied the defect formation, evolution and amorphization occurring after ion implantation to determine whether the same defect s occur in Ge and C as in Si under the same implant conditions. The expected result is that the same defects do form, and if so, leveraging the collected knowledge of Sis extended defects will save development time and research costs into devices based on C and Ge. If the expected result of similar defects is not found, the results of this research could be applied to adjust cu rrent simulation tools to save development time in potential future electronic devices. The research documented in this publication is organized in the following manner. Chapter One provides the motivation and the objectives of this research, while Chapter Two reviews the current state of the literature. Chapter Three provides an overview of the characterization techniques used in this st udy, including a novel method for creating cross-sectional TEM samples from single crystal diamond substrat es. Leveraging the research in non-amorphizing

PAGE 21

21 implants into SiGe by Crosby, et al., Chapter Four presents the expe rimental results of nonamorphizing implants into pure Ge and Si, and di scusses the differences in the formation and evolution of the extended defects. Extended de fects are also known to occur in Si for amorphizing implants, therefore Chapter Five pres ents the experimental results of amorphizing implants into Ge, and discusses the Si similarities and differences in the results, in the frame of other current research. Knowing the similarities a nd differences presented in the previous results for Si and Ge, the results of the non-amor phizing implants in diamond are provided and discussed in Chapter Six. This covers implants below and to the verge of creating a continuous amorphous layer in diamond. Chapter Seven presents the results from amorphizing implants into single-crystal diamond, compares the results to those from Ge and Si, and presents the experimental results of amorphization thresholds Chapter Eight summarizes the results based on the original hypothesis that the same defects that form in Si would also form in Ge and C. Since that hypothesis did not hold true for all scenar ios, recommended future work is discussed.

PAGE 22

22 Table 1-1. The mobility of carriers in semi conductors, as reported in the September 6, 2002 issue of Science. The diamond values we re achieved using plasma-grown single crystal diamond.4 Semiconductor Mobility, cm2 V-1 s-1 ( ) Maximum electric field, V cm-1 (Em) Bandgap, eV (Eg) Diamond (C) 4500 (electron) 3800 (hole) 107 5.5 Silicon carbide (SiC) 700 (electron) 3.0 106 3.26 Gallium nitride (GaN) 2000 (electron) 3.0 106 3.0 Gallium arsenide(GaAs) 8500 (electron) 4.0 105 1.42 Silicon (Si) 1500 (electron) 450 (hole) 3.7 105 1.12 Germanium (Ge) 3900 (electron) 1900 (hole) 2.0 105 0.66

PAGE 23

23 Figure 1-1. Cross-sectional TEM image of Intel Pentium transistors showing the appearance of dislocation loop defects. Courtesy of Intel and Scott Thompson. Dislocation Loops

PAGE 24

24 Figure 1-2. A comparison of the re ported {311} defect in (a) Si 17and (b) Ge.11 (b) (a)

PAGE 25

25 CHAPTER 2 LITERATURE REVIEW Introduction Presented here are descriptions of general concepts associat ed with extended defects and amorphization in silicon (Si), and relevant literature reviews that describe the state of current knowledge in germanium (Ge), carbon (C) and their alloys with Si. If th e reader is already familiar with the concepts of ion implantation damage in a crystal, annealing kinetics and the difficulty of modeling the crystalline-to-amorphous transition, then the first section of this chapter may be superfluous. The second part of the chapter will review relevant studies of extended defects, rising from both non-amorphizi ng and amorphizing implants, in Si, Ge, C and their alloys. General Concepts Ion Implantation Ion implantation has been the preferred me thod of introducing dopants into semiconductor crystals for decades, due to its many advantag es: reproducibility, accura te dose control, and tailored doping profiles. As such it is one of the most important tools in Si processing.18 Ion implantation is the forced entry of ions into a substrate through its surface. The ions travel through the lattice intera cting with the lattice atoms though el astic collisions with electron clouds and inelastic electronic or nuclear collisions, wh ich disrupts the lattice st ructure. If the energy transferred during the inelastic collision is greater than the displacement energy (Ed) of an atom, that atom will be dislodged from its lattice pos ition. Stopping of the implanted ion is dominated by nuclear collisions at low energy implants, wh ile electronic stopping is dominant for high energy implants.19 Nuclear collisions can be mode led using the following equation:

PAGE 26

26 2 1 1 2 / 1 3 / 2 2 3 / 2 1 2 1 15) ( 10 8 2 ) ( m m m Z Z Z Z E Sn (2-1) where Sn(E) is the nuclear stopping powe r as a function of energy, Z1 the ion atomic number, Z2 the substrate atomic number, m1 the ion mass, and m2 the substrate mass. Nuclear stopping contributes most of the damage after ion implantation. The second stopping mechanism, electronic stopping, can be modeled as: 2 / 1) ( kE cv E Sion e (2-2) where Se(E) is the electronic stopping power as a function of energy, E is the energy of the ion, v is the velocity of the ion, and c and k depend on the ion, substrate and the stopping mechanism. Nuclear stopping increases with increasing ma ss (m) and decreasing energy (E); electronic stopping increases with increasing implant energy (E ), as shown in Equations 2-1 and 2-2. The rate at which an ion loses its energy is repres ented by the following differential equation, which takes into account both stopping mechanisms, and the atomic density of the substrate (N): )] ( ) ( [ E S E S N dx dEe n (2-3) In this work, the projected range and damage peak of the implant will be an integral part of estimating damage in the crystal and also for determining threshold parameters for amorphization. To begin to determine the projected range of an implant, the range (R) of one ion, in some measure of distance, may be determin ed by rearranging the above differential equation and integrating over the distance: 00 0) ( ) ( 1E e n RE S E S dE N dx R (2-4) As the implantation process typically implants above 1010, and even as high as 1020 ions cm-2, the process of determining the projected range and straggle is of a statistical nature. The ion concentration as a function of depth is represented to the first order by:

PAGE 27

27 2 22 ) ( exp ) (P P pR R x C x C (2-5) where Cp is the peak concentration at the cen ter of the Gaussian distribution, RP is the projected range of the ions into the substrate, and RP is the first standard deviation about RP, also known as ion straggle. Dose (Q) is the total number of implanted ions, and can be expressed in terms of Cp and RP: p PC R Q 2 (2-6) The implant profile may not always be Gaussian in nature due to ion channeling. Channeling is the phenomena by which the ion traject ory is in line with atomic rows, and experiences a slower rate of energy loss. This results in a non-Gaussian distribution of ions and damage in the lattice. Channeling may be reduced or avoided by titling th e substrate or amorphizing the crystal prior to implantation. Amorphization will be discussed in a future section. In this study, channeling was avoided by tilting the (011) and (001) crystals seve n degrees from normal. Four moments are used to fully describe th e distribution of implan ted dopants within a crystal: projected ion range RP), ion straggle ( RP), skewness ( ) and kurtosis ( ). Skewness describes the asymmetry of the profile about the Gaussian peak, and kurtosis describes the contribution of the profile tail on the flatness of the profile shape. A ll four moments can be mathematically represented in terms of dose and concentration as a function of depth: dx x xC Q RP1 (2-7) dx x C R x Q RP P 2) ( 1 (2-8) 3 3) (P PR Q dx x C R x (2-9)

PAGE 28

28 4 4) (P PR Q dx x C R x (2-10) Thus far, the stopping of the ion by energy tran sfer has been described. If the energy per collision transferred to an atom equals or exceeds the atoms displacement energy (ED), the lattice atom will be displaced from its lattic e site. If the energy transferred exceeds the ED, the displaced host atom then may continue to inter act (by electronic or nucle ar stopping) with other host atoms in the same manner as the incoming ions. This is called a collision cascade, and hundreds to thousands of these lattice atoms ma y be displaced per incoming ion depending on the ions mass and energy. Each displacement leaves a vacancy, and each atom that does not come to rest on a lattice site is an interstitial atom. A vacan cy and an interstitial are known as a Frenkel pair. For room temperat ure implants, it is assumed th at all Frenkel pairs recombine during implantation, so only interstitials remain, in the amount of the original dose of the ion implant.20 Both interstitials and vacancies naturally exis t in any crystal above absolute zero according to the relation: kT Q PPN N exp (2-11) where NP is the concentration of point defects, N is the concentration of lattice sites, QP is the formation energy of the point defect, k is Boltzma nns constant, and T is te mperature. For Si at room temperature, the formation energy for inte rstitials is 2 eV, and vacancies is 4.4 eV.20 The simulation program commonly known as SR IM, the Stopping and Range of Ions in Matter,21 and its calculation program (commonly known as TRIM, for the Transport of Ions in Matter) is often used for simulating the damage caused by ion implantation. The TRIM program is a Monte-Carlo simulation which requires input pa rameters such as ion type, ion energy, target

PAGE 29

29 substrate, target composition, density and Ed. The TRIM program does not take into account channeling effects, phase transformation of the ma terial or the possibility that defects can be annihilated during the process, a nd therefore always overestimates the damage. Even with these drawbacks, it remains an excellent estimator for the damage created during ion implantation.19 Defects in the Crystalline Lattice The dislodged atoms or the implanted ions ma y either come to rest on a vacant lattice position (fill a vacancy) or occupy a non-lattice position (become an interstitial). These types of defects are known as point def ects. Other point defects which may exist after ion implantation and annealing are substitutiona l atoms (foreign atoms which s it on a host lattice site) and interstitial atoms (either self-interstitials or foreign interstitials). At a high enough density, the point defects can agglomerate into defect co mplexes, commonly known as extended defects. Defects that are interstitial in na ture are referred to as extrinsic defects; those that are created by vacancies are intrinsic defects.19 To repair the defects created during ion implantation, annealing is used. The word a nneal comes from a Middle Englis h word analen which meant to set on fire. Annealing is the process of heating the crys tal to remove damage by adding energy into the system so that the defects become mobile. Upon annealing at moderate temperatures, point defects can interact to form extended defects. Extended annealing time or annealing at higher temperatures will result in the defects evolving so much that the atoms find lower-energy sites to occupy in the lattice and effectively dissolve the defects, thereby restoring the crystal structure.19 These extended defects may include dislocati on loops, stacking faults, rod-like inte rstitial defects, clusters and others. Jones et al. classi fied extended defects which occur in Si after ionimplantation and annealing into five types. Th ese defect types are known as subthreshold, endof-range (EOR), regrowth related, clamshell (or zipper), and solubility related defects.9

PAGE 30

30 A visual representation of the principal damage features in Si as a function of dose and annealing temperatures, for MeV energy implants was presented by Libertino. 22 No similar type representations seem to exit for Ge or C. For this study, the dose ranges into the crystals will be in the range of 1014-1015 cm-2, and annealing temperatures will encompass and expand beyond those for defect formation in Si. This study fo cuses on the {311} and extended defects, occurring in the 1014-1015 cm-2 dose range in Si, which have been industrially unfavorable for Si complementary metal oxide semiconductor (CMOS) processing. {311} defect in the Si crystalline lattice. One particular defect that has been especially relevant to dopant motion in silicon is the rod-like {311} extended defect.17 Si and Ge have been the only elements to display this unique de fect; it has never been seen in metals.5 This defect is the preferred interstitial precipita te in silicon, and Salisbury concl uded that the {311} rather than the {111} is the normal plane for self-interstitial preci pitation for all diamond structure crystals.7 The {311} defect is a planar defect that re sides on the {311} habit plane and grows linearly in the <110> direction.23 The {311} defect has three distinct stages of evolution: formation, growth and dissolution.24 The defect is formed from an accumulation of point defects into clusters, known as sub-microsco pic interstitial clusters (SMICs ), since the clusters are not discernible by conventional TEM methods.25,26 Next, the clusters grow in the <110> direction and finally, at a high enough thermal budget, the {311} defects will dissolve in the matrix by ejecting interstitials.17 The exact mechanism of the {311} fo rmation has been studied at length, and various models have been proposed in the literature.5,7 In silicon, the {311} defect is known to affect diffusion of dopants17and are the source of the formati on of another form of extended defects, dislocation loops.6 These dislocation loops are know n to affect the electrical characteristics of silicon devices.5

PAGE 31

31 Much is known about the {311} defect in si licon. A high resolution TEM image of a {311} defect was shown in Figure 1-2. The {311} defect is an average of 40 wide and varies in length. It grows in the <110> direction, and will trap 26 interstitials per nm. As annealing progresses, {311} defects will increa se in size but decrease in density.17 There is no length dependence on the {311} defect evolution the formation energy is constant. Unlike loops, the formation energy does not decrease as the size of the defect increases. The activation energy of a {311} has been estimated to be approximately 3.6 eV, which was derived from annealing curves.17 Various experimental conditions can creat e the {311} defect. A ccording to Jones, Prussin and Weber, the {311} defect appears du ring various doping regimes. When the implant does not exceed the amorphization threshold, the {311} defects ar e the primary extended defect. When the damage exceeds the amorphization thres hold, upon annealing, the {311} defect will develop in the region of the EOR.9 Using the activation energy of the {311} defect, the equivalent defect state fo r various times or temperatures can be determined.27 The {311} defect occurs at both low and high energies. Defects created by Si ion implants into Si in the range of 20 keV to 150 keV (rela tively low energies) have been mapped according to dose. The {311} defect was also seen after a Si+ implantation at 1.2 MeV implant of 5 1013 cm-2 into Si and annealed at 680 C for one hour,26 and also 4 MeV implants of 1 1014 Si+ cm-2 into Si annealed at 800 C for 15 min,28 both of which are rela tively high energies. Besides ion implantation, the {311} defect ha s also been observed in unimplanted Si resulting from electron beam irradiation.7,8,29,30 Salisbury and Lorettos work with electronirradiated silicon led to the dete rmination that the (113) rather th an the (111) is the normal plane for precipitation of self -interstitials in diamond-structure crystals.

PAGE 32

32 Impact on Si doping. Control over dopant atom placement is one of the strongest benefits of ion implantation. However, dissolution of {3 11} defects during anneali ng gives rise to an unwanted side effect tran sient enhanced diffusion (TED ). During {311} dissolution, interstitials are emitted, and assi st in the motion of dopant atoms away from the intended target depth in the bulk material.17 By moving atoms away from the targeted channel area, TED directly and negatively aff ects device performance. {311} defects are also the pr ecursors to extended defects which negatively impact device performance. Dislocation loops are two dimensional circular precipita tes of interstit ials inserted between two consecutive {111} planes. Roberts on et al. showed that {311} defects are the preferential site for disloca tion loop formation after an amor phizing implantation regime when the defects occur in the EOR.31 Additionally, the {311} defect formation has been known to be inhibited or enhanced by various dopants. The formation of the {311} is strongly enhanced by the presence of arsenic (As), but the dissoluti on remains unaffected.32 The density of interst itials in {311} defects decreases linearly with the increasing presence of C, which fully suppresses the TED nature of boron (B) in Si.15 The normal characterization method for {311} defect detection is TEM. Amorphization If the concentration of point defects exceeds a cr itical value during implantation, a phase transformation occurs, changing the crystalline solid into an amorphous solid. As mentioned before, light ions cause less damage than heavier ions. The damage tracks left in a material are called a displacement spikes, which are very linear w ith the track of the atom in light ions but consists of a volume of material with heavie r ions. Morehead and Crowder assumed that, in crystalline Si, the displacements spikes relax to form amorphous carbon, and when these spikes overlap, a continuous amorphous layer is formed.33 In Si, the amorphiza tion process has been

PAGE 33

33 modeled as a first-order phase transformation. Dose has a large effect on amorphization. For light ions, higher doses are needed to create a continuous amorphous layer, but for heavier ions, overlapping collision cascades allow for lowe r amorphization doses. Additionally, after ion implantation, three profiles may occur in Si:34 the first is a profile that will not have an amorphous phase change upon implantation, the second is a buried amorphous layer and the third is a continuous amorphous layer.27 Implantation temperature can have significan t impact on whether am orphization will occur in Si. A room temperature implant of a Si+ implantation at 1.25 MeV to a dose of 1 1016 cm-2 into Si produced an amorphous layer approxima tely 400 nm thick, while a dose of 2.5 1015 cm2 did not create any amorphous layer.34 The same ion and energy implant at liquid nitrogen temperatures, 80 K, for a much lower dose of 5 1014 Si+ cm-2 into Si, resulted in a larger amorphous layer, nearly 1400 nm thick.34 Low temperature implants are thought to freeze in point defects, while room temperature implants assume that all Frenkel pairs recombine during implantation.20 The ion dose rate, if too high, may heat up the sample during implantation, but this can be avoided by using a Waycool station.35 Additionally, the morphology of the damage after amorphization at low temperatures is differe nt than that at room temperature or higher. Pockets of crystalline material were shown to be trapped in the amorphous layer near the -c interface of a liquid nitrogen temperature Si+ implant into Si. These crys tallites were not seen in the same implant at room temperature.34 At sufficiently high implant temperatures, amorphization can be avoided.20 Amorphization changes the density of a materi al. By nature of disrupting the lattice, swelling occurs in the crystal. In Si, the densit y change is minimal, and difficult to observe by

PAGE 34

34 XTEM. Using computational simulations, the amorphous phase of the Si material was determined to be approximately 1.76% less dense than crystalline Si.36 Amorphization is a difficult phenomenon to model due to the fact that it is affected by ion dose, dose rate, energy, mass of the incoming ion, mass of the target ion, and temperature. Christel, Gibbons, and Sigmon suggested that displacing 10% of the Si lattice at low temperatures (14% at room temper ature) would result in amorphization.37 They used a transport equation, and in conjunction with other work, used it to determine the number of silicon recoils produced per unit volume at each depth. Howeve r, backscattering measurements were used to determine interface points, and not XTEM, which allo ws some room for error. Prior to Christel, energy-to-volume was put forth as a model to equalize the different factors that affect amorphization, with a general value of 6-10 1020 keV/cm3 determined.37 Since then, three parameters have been commonly used to descri be amorphization in Si: critical damage, using either recoils or vacancies per volume (Dcrit or DC, respectively), threshold damage density (TDD), and displacements per atom (DPA). Thres hold damage density is reported in terms of energy per volume, and has been identified for P+ implants into Si over a wide range of energies to a dose of 1 1015 cm-2 as 1.0 1021 keV cm-3, which is slightly higher than the previous model.35 The ion implantation community favors re porting the amount of damage needed to induce amorphization at liquid nitr ogen temperature in displacements per atom (DPA), a unitless parameter. For Si, the DPA is 0.3. For all the previous methods, a key point is the depth at which the measurement or calculation is performed. For a buried amorphou s layer, there are tw o amorphous-crystalline ( c) interfaces: one near the surface and one near th e bulk. As the energy of the implant increases, the surface -c interface becomes increasingly wide r. For El-Ghor et al.s 1.25 MeV Si+

PAGE 35

35 implantation into Si, the authors used a range between 0.3 0.9 m depth to define the -c interface.34 The bulk interface is normally quite abr upt, due to the sharp decreace of implant damage in the EOR region. Christel and Prussi n both used the bulk interfacial measurement to determine the amorphization thresholds in Si, th ough both used different methods to determine the position of the interface.35,37 Solid Phase Epitaxial Regrowth Annealing at sufficient temperatures will cause the reorganization back into the crystalline structure, starting at the interface between the amorphous region and the crystalline bulk substrate. This is called solid phase epitaxial regrowth (SPER). Three kinds of defects may occur in the lattice during SPER: end-of-range (EOR), clamshell, and regrowth-related defects. Regrowth-related defects look lik e hairpins, clamshell defects are a line of defects extended horizontally across the sample a nd are created when the two adva ncing fronts meet, while the EOR defects are furthest from the surface, and consist of rod-like {311} and dislocation loop defects.34 The EOR defects have been well studied.36,38Defect evolution both before and after SPER is important because defects can infl uence dopant motion and mobility within the crystal.17 These defects directly affect dopant electrical activation, without which the semiconductor could not be used as an electrical device. Simulation Process Tools Many computation methods have been used to model aspects of def ects in Si. This is important because the cost of crea ting and testing devices for Si is expensive, and would be more so for devices made from Ge or C. Additionally the development of phys ics-based models that accurately describe defect evolution can be as similated into process simulators, which have become an essential step for device optimization.39,40 The computational methods that have been

PAGE 36

36 used to model defects are ab initio, molecular dynamics, kine tic Monte Carlo simulations and continuum models. Continuum process simulators. By far, the physical based continuum process models are the most useful to industry due to the fact th at the models are computationally less expensive than other techniques.39 In continuum process simulators, the physics of the system is calculated as a series of differential equations for each particle type. Typically these equations are continuity equations, where each particle gain or loss is formulation in terms of its generation and recombination rates and the diffusion flux.40,41 The benefits of this system are that it is fast, it can model a wide range of size implants and temp erature ranges and times and is able to be compared to experimental evidence.39,42 The {311} continuum model solves for the total number of inte rstitials in the {311} defects and the total number of the {311} defects. The evolution of {311} defect s unfaulting into loops at the EOR damage was modeled by Avci et al.43 using the Florida Object Oriented Process Simulator (FLOOPS).39 Recently, Zechner et al.44 proposed a physics based model using five reaction equations to model the formation and grow th of the {311} defect over a wide range of implantations, doses and annealing temperatur es. This simulation matched experimental evidence. The {311} defect is now well-modeled for a wide range of implantation regimes for silicon using the continuum process simulators. The continuum methods are the best for industrial applications. However, these simulations are based on what can be ascertained from experimental evidence. In the case of Ge and C, little or no experimental evidence exists to be able to create a co ntinuum model. In order to compare Si to Ge and C, a more basic mode l, albeit requiring a sl ower method with more computational power, might be needed.

PAGE 37

37 Other simulation tools. For silicon, the actual mechanis ms that control formation of the {311} have been attempted to be modeled at the atomic level, and these simulations might prove useful in comparing the three el ements of C, Si and Ge. The {311} defect has been modeled using first principals calculations (ab initio). From these calculations, the formation energy for the {311} defect was determined to be 0.7 eV, and additionally, growth in the [011] direction was predicted from the model.45 This is much lower than the experimentally determined value of 3.6 eV. The ab initio authors estimated the sm allest stable planar {311} defect contained approximately 40 interstitials with a width of 19 and a length or 27 .45 Although these simulations can give insight into the atomic in teractions of defects, the simulations are not practical for use either for indus trial applications or for comp aring to experimental results. The structure and mechanism of the {311} def ect has also been studied using the tightbinding (TB) method. These calculations determined that the {311} defect could be calculated as stable with small bond distortions and with smalle r energies per intersti tial atom than a selfinterstitial in bulk Si.46 However, due to the computationa l power required, TB studies only analyze approximately 103-104 atoms and are not comparable to experimental results. These methods provide insight for making si mulations of larger atom counts more accurate. Physical parameters determined from AB and TB studies can be fed into continuum and kinetic Monte Carlo simula tions. Kinetic Monte Carlo (KMC) computer codes take the dynamics of defects (such as diffusion, emission from cluster, and recombination) into account. These types of simulations have also been called atomistic simula tions and take into account rate theory and atomic diffusion. The amorphization a nd recrystallization proce sses in silicon have been modeled at the atomic level. KMC codes di ffer from continuum models in that they keep track of the location of every frenkel pair prov ided by binary collision approximation simulations

PAGE 38

38 (such as SRIM)21 or molecular dynamics calculations. Parameters that define the atomic interactions (such as diffusivities, binding energi es, capture radius) are all derived from ab initio or MD calculations. However, as KMC simula tions can only model up to a million atoms at approximately 10-3 seconds,47 these are not appropriate for industrial use. However, since atomistic simulations rely on physical parameters such as bond energy, and simulations in silicon and carbon have already been used in at omistic simulations of grain boundaries,48 it is theoretically possible to simulate the formation of a {311} def ect in a lattice of either Si, Ge or C with an atomistic simulation. Relevant Literature for Ge, SiGe, SiC, and C Germanium Defects and Amorphization Germanium was the original base material in the first transistors, and was the dominant material for semiconductors in the 1960s. Due to it s inability to form a stable native gate oxide, Ge was surpassed by Si as the semiconductor material of choice. Ges native gate oxide dissolves in waters, and its growth is non-uniform a nd sensitive to humidity. However, due to the development of high-k dielectrics instead of native oxides for use as a gate material, Ge is being seriously considered as an alte rnative channel material for ma instream CMOS devices. Although defects in Ge have not been as th oroughly studied as in Si, Ge is currently being used in the SiGe alloy to strain the channel (which increases carr ier mobility) in current generation transistors. Of the group IV elements, Si and Ge are the only tw o which are fully miscible in each other, and form a solid solution across all compositional per centages. Research so far indicates that ion implantation and dopant interaction in Ge seem s to be fundamentally different than Si.1 Defects in Pure Ge Defects have been viewed in Ge for as long as Si, but due to its comparative lack of interest, the study has not been as intense. The {311} defect has nevertheless been seen in pure

PAGE 39

39 Ge for decades. Early studies focused on electron irradiation in Si and Ge, while later studies have focused on light ions. And example of a r ecent {311} defect created in pure Ge was shown in Figure 1-2. In that example, the Ge was implanted with H+ (15-80 keV, 5 1015 7 1016 cm1) and annealed at temperature between 280-400 C for 30-60 minutes.11 Electron irradiation. Electron irradiation has produced the {311} extended defect in Ge. 10,29 Electron irradiated Ge showed small interstitial loops on { 311} planes, which grew in the <110> direction upon annealing. At 450 C for ten minutes, the defects began to transform, and became perfect dislocation loops when the ten minute anneals reached 500 and 600 C.10 In a separate study, rod-like defects along the {111} and {311} planes were also seen in Ge electron irradiated at temperatures be tween 400-500 C along with defect s found in Si irradiated at 650700 C.8 Hydrogen (H) and helium (He) ion implantation. Ion implantation of H+ has also produced the {311} extended defect in Ge. In 20 05, Akatsu et al. studi ed the extended defects after H+ implantation into pure Si and Ge (5 1015 H+cm-2 and 7 1016 H+cm-2 at 50 and 80 keV).11 After annealing at temper atures ranging from 280 400 C and times between 30-60 minutes, a large density of {311} defects were ob served in Ge but not in Si. The Ge formed spherical nano-cavities as well as the {311} defect but these nanocavities were not observed in Si either. This was attrib uted to the differences in chemical and physical interactions between the hydrogen ions and the respective crystal.11 However, previous knowledge of Si defect evolution shows that microstructural evolution in Si is unusual at temperatures lower than 400 C; therefore, a direct comparison at these temperat ures, as compared to the melting point of each material, is a poor comparison.

PAGE 40

40 In addition to H, ion implantation of He has produced the {311} defect in pure Ge. Hutchinson et al. implanted Ge with 1.23 MeV He ions (1 1016 He+/cm2) at room temperature, performed no anneal, and observed voids (pro bably He bubbles) and {311} defects. A model was suggested explaining the development of th e {311} defects incorporating hexagonal Ge, 5, 7 and 8-member rings.12 Additionally, rod shaped def ects and {311} defects were al so observed when implanting He+ into Ge at a dose of 1 1016 cm-2 at an implant temperature of 20 C and at 600 keV, after a 12 minute anneal at 640 C. These defects lack th e rigid zigzag pattern normally seen in Si. These defects were observed for all annealing temperatures betw een 450 C and 650C; however, above 550 C the rod-shaped defects be gan to turn into dislocation loops. 49,50 Supposedly, rod like defects, along the {111} and {311} planes, were also seen in Ge implanted with He ions at a dose of 1 1016 at 500 keV with no anneal, however, no micrograph was provided. The {111} rod like defects were attributed to the split <100> interstiti al configuration. 8 Amorphization, Regrowth and La ck of EOR Damage of Ge All of the above implantation experiments in Ge were below the amorphization threshold. Ge will amorphize, as does Si, and will fully regr ow at temperatures around 400C. The regrowth rate of Ge is approximately 2 min-1 at 350C and 13 min-1 at 380C.51 Unlike Si, {311} defects resulting from EOR damage for amorphi zing implants has not been reported. In 2005, Chao reported amorphizing Ge with a 45 keV, 1x1015 Ge+cm-2 ion implant into pure Ge, resulting in an amorphous layer 1700 deep. No EOR defects were reported upon regrowth.52 Satta et al. did not observe any EOR damage in Ge after a 40 keV implant and SPER, either.53-55 Swelling of the crystal is associated with amor phization due to a decrease in density in the amorphous layer. The swelling in Ge was measured by step height versus implant temperature during 1 MeV Ge+ implants into Ge.56 For doses of 1 1015 Ge+ cm-2, and room temperature

PAGE 41

41 implants, the crystal swells approximately 2%.56 Non-intuitively, ion-implantation can also assist with crystalline regrowth af ter amorphization in Si and Ge, due to dynamic annealing.57 With extremely high energy and heavy ions, a void-f illed phase is also formed alongside the amorphous phase in Ge. For fluences exceeding 1 1017 atoms/cm2 into pure Ge, a 30% decrease in density has been reported.56,58 SiGe Alloy Defects and Amorphization Crosby et al. studied the { 311} defect formation in Si1-xGex from Ge concentrations ranging from 0 to 50%. He noted that although the { 311} defect was formed in SiGe alloys up to 5% Ge, it was not formed when the alloy surpassed a 25% Ge content.13 The decreased binding energy and the number of Si inte rstitials effectively increased the activation energy for {311} formation.41,59 Tan has suggested that a 15% or above distortion of the Si lattice would result in the destabilization of the {311} defect configuration.5 This would mean that a 16-20% Ge content would be required to destabilize the {311} defect.60 Recently, vacancy nanoclusters have been detected in ion-implanted SiGe. These clusters have been found using Copper (Cu) gettering and positron analysis techniques. The vacan cy clusters are located near the surface in SiGe with Ge percentages of 10-30%.14 Amorphization is a phenomenon by which the crys tal lattice loses l ong range order after ion-implantation past a critical da mage point. It is associated with a reduction in material density, approximately 1.5-2.1% in SiGe allo ys, which causes the crystal to swell.61 Many factors influence whether amorphization will occur, incl uding implant temperature, dose weight, mass of the ion, mass of the target atoms, energy of the ion and dose rate. In SiGe alloys, the threshold for amorphization is noted to decr ease with increasing Ge content.14 Ge & SiGe summary. The {311} defect has been observe d in SiGe alloys under 25% Ge, and in electron-irradiated and light-element implanted pure Ge. Annealing temperatures

PAGE 42

42 necessary to produce this defect are significantly lower than Si in the 350-650 C range. No reports of {311} defects or dislocation loops resulting fr om EOR damage in amorphizing implants have been reported. Silicon Carbide (SiC)Ion Impl antation and Amorphization Silicon carbide is a wide band-gap semiconductor, extremely hard and is being considered as an ideal material for use in high-tempera ture, high-power, and highfrequency applications.62 SiC has a relatively low elec tron carrier mobility (700 cm2 /V s) as compared to Si (1500 cm2 /V s) and single crystal pl asma deposited diamond (4500 cm2 /V s).4 SiC has been used as a base material for growing CVD diamond. However, the grown diamond is not of device quality due to its polycrystalline nature. SiC has also been simulated by implanting high doses of Si into diamond.19 SiC, while a fascinating technological materi al, will only be briefly discussed here as it relates to the present study. Ion implantation in SiC has been studied by e xperimental and computation techniques. The displacement energy (Ed) for SiC is slightly more complicated than for single-element crystals. The Ed in SiC for a carbon atom is approximately 20 eV, and for a silicon atom, 25-40 eV.62 Ion implantation of Si into diamond in high fluences (1017-1018 ions/cm2) at elevated temperatures (900 C) have produced large grains of SiC. Th e low dose results in layers of epitaxial SiC stripes, while the higher dose produces large grains of SiC; however, the diamond bulk material next to the grains become s amorphous in the process.63 Others have us ed high-fluence Si+ implanted into diamond to create SiC.64 Devanathan et al. studie d the amorphization of SiC by ion implantation. It was determined that amorphi zation is onset after 0.2 displacements per atom (DPA) is achieved. The decrea se in density accompanying amorphization is about 15%, while previous studies put the decrea se at approximately 12 4%.62 Ishimaru reported the DPA for SiC was 0.5 for an elevated temperature of 350 K.65,66

PAGE 43

43 Diamond Diamond, in many crystallographic forms but es pecially single-crystal diamond, is gaining academic and industrial interest.67,68 The creation of synthetic diamond was first reported in 1955,69,70 and like SiC, is expected to be an im portant semiconductor for high power and high frequency electronic applications.71 An attractive diamond quality fo r electronic devices is the high carrier mobilities demonstrated in single crystal plasma deposited diamond.72 The popularity of diamond for electronic applicatio ns became evident in the mid-to-late 1990s through the assembly of a numbe r of symposiums in the ma terials research community.73 Most current interest in diamond is focused on diam ond films, which are a CVD-grown combination of sp2 and sp3 bonded C, with the possibility of H as well. Diamond is being used in MEMS devices,74,75 AFM tips,76 FETs,71,77 Diodes,68 pn junctions,78 and there is also development of metal contacts to diamonds for electronic applications.79 Currently, ion implantation of singlecrystal diamond is of great inte rest to the MEMS community fo r creating single crystal diamond microstructures. Ion Implantation in Diamond Although TEM experiments are sparse, much is already known about ion implantation in diamond. Both natural diamonds and synthetic diamonds (created by both high temperature high pressure (HTHP) and chemical vapor depos ition (CVD) methods) have been used in experiments. Since the HPHT diamonds have the same structure and defects as natural diamond,19 studies of both kinds of diamonds will be pr esented. In general, ion implantation into diamond behaves in a similar fashion to silicon. A few key points of interest in the literature include a debatable Ed for SRIM21 simulations, implant temperat ure considerations, and the possibility of a direct graphiti zation transformation after implan tation over the threshold limit.

PAGE 44

44 Sub-graphitizing implants and repairing anneals. The literature has reported that implant conditions that create less than approximately 1022 vac/cm3 will not graphitize the single crystal diamond (not film) upon annealing,16 though a recent study puts the number at 9 1022 vac/cm3.75 According to Hall measurements, furn ace annealing at a temperature up to 900C appears to repair the implantation damage to the crystal80. Kalish agreed that below the critical dose, the material will gradually anneal back to diamond, starting at annealing temperatures as low as 330 C. The defects dissolve at higher te mperatures, and for annealing above 1000 C, the diamond crystal structure will be fully restored.81 This temperature was determined by density measurements at the temperature at which the crystal restores its original density after annealing.81 Additionally, it is reported that damage to the lattice can be repaired almost immediately by modulating the impl antation temperature. Below a 50 C implantation temperature, vacancy and interstitial movement is inhibited. Above 50 C, interstitial diffusion begins to annihilate vacancies, and above 500 C, vacancies diffuse and begin to form extended defects.82 Even a perfect crystal of di amond reverts to graphite when annealed at over 1800 C in an excellent vacuum.83 Activation energy and mobility of interstitials in diamond. As just reported, below an implant temperature of 320 K, both interstitials and vacancies are frozen in. Above 320 K but below 800 K, interstitials are mobile but vacanci es are not. Above 800 K, both interstitials and vacancies are mobile.84 That is why the critical dose in diamond also relies on the implant temperature.85 It has been determined, using the work by Moorehead and Crowder as a basis then adding data from Kalish, Prawer and Prins, th at the activation energy for the diffusion of an interstitial in diamond is 0.2 eV, up to temperatures of 700 K.85 For more information of selfinterstitial energy levels in diam ond, see the work by Davis and Smith.86

PAGE 45

45 Doping of diamond. Vavilov implanted B+ into diamond in the 1970s, and (after etching off the graphitized layer) meas ured the p-type conduction. Br aunstein and Kalish repeated similar experiments.83 Two methods proven successful when doping B+ are the Prins-suggested cold implantation rapid annealing (CIRA)19 and low dose drive in (LODDI).87 The CIRA method has been used to dope diamond with B+, C+ N+ and O+.88 Optimization of the B+ CIRA process led to good quality diamond and hol e mobilities in excess of 1000 cm2 /V s.89 p-type doping of diamond. MeV-ion implantation has shown to be a promising method for doping of diamond. Lower energi es (<640 keV) and low doses (~1013 cm-2) of implanting boron has been tried using the CIRA method, followed by hour-long anneals at 1100 C then 10 min at 1450 C.90 The implantation of 2 MeV B+, followed by furnace annealing at 1450 C, was used as a justification of the high pressure theory.83 An activation energy of the boron acceptors was determined to by 0.372 eV.91 Uzan-Saguy later published a doping efficiency of 2% for 2 MeV B+ atoms at a dose of 1 1015 cm-2 implanted at 77 K. No diffusion of B+ was observed in that study.89 In 2004, Vogel et. al implanted mimicked the experiment, but at room temperature, implanting 2 MeV B+ atoms into natural diamond samples and achieved 30% activation of the implanted boron atoms for a dose of 1 1014 cm-2 and 15% for a 1 1015 cm-2 dose, and claimed it to be the highest at that time for ion-implanted diamond.92 That experiment used annealing temperatures up to 1650 C in a furnace under vacuum. It was determined that annealing at temperatures above 1500 C for 15 min did not im prove the quality of the implanted layer, but instead decreased it. Low residual defects afte r the 1500 C anneal were determined by sheet conductivity and mobility measurements.92The present research performed an XTEM study of this implant regime, and the resu lts are reported in Appendix B.

PAGE 46

46 n-type doping of diamond. While boron doping has been met with some success in diamond, n-type doping of diamond is proving to be more difficul t than p-type doping. Nitrogen is a n-type dopant in diamond, but lies appr oximately 1.7eV below the conduction band, making it a deep donor. Phosphorus, a common n-type dopant in Si, imparts four times as much damage when implanted into diamond as does boron, and this is a major problem in doping diamond. Codoping with hydrogen has been able to increase the conductivity slightly.83 Damage also has an effect; n-type conductivity has been reported in high-fluence Si+-implanted diamond.93 The presence of defects does not seem to enhance diffusion of light dopants in diamond like it does in Si, up to temperatures of 1300 K. However, slig ht diffusion of boron has been reported above 1300 K.94Additionally, pand n-type doping of diam ond has also been performed by growing doped diamond by CVD processes.95 Defects in diamond. The {311} defect has never been reported in diamond, even though Salisbury concluded that the {311} rather than the {111} is the nor mal plane for self-interstitial precipitation for all diamond structure crystals.7 Additionally, ion implan tation has been studied in diamonds for over three decades,82 but reports of any kind of defect are sparse. This is because, in the past, creating a TEM sample of single crystal diamond has been very difficult, due to diamonds high bond strength.96 TEM is the best method to pursue a number of current research questions in diamond, especially defect formation and evolution as the result of ion implantation and annealing.19 Recently, use of the focused ion beam (FIB) has led to the ability to create TEM cross sections of diamonds in a more timely fashion, and that method will be presented later in this paper.

PAGE 47

47 The (100) split interstitial has been calculated to be the stable defect in diamond after ion implantation.81,97 Additionally, after ion implantation, intrinsic defects can be electronically active.83 Electrical conduction in diamond is impe ded by the high density of residual defects.83 Amorphization vs graphitization of diamond. It is well known that past a critical damage value, the diamond lattice is too damaged to repair itself.83Above this critical dose, the diamond converts to a form of amorphous carbo n, supposedly stable up to an annealing temperature of approximately 430 C.When the annealing temperature exceeds this temperature, the crystal begins to graphitize by forming sp2 bonded nanoclusters.81 However, Prins theorized that the amorphous region was not truly amor phous, but is actually a vacancy rich diamond crystal lattice.98 In a later paper, Prins stated that re laxation to the graphite state might occur before a real amorphization (suc h as in silicon) takes place.99 In 1999, Kalish et al. reported that co mplete graphitization is reached at 600 C (after a 20 min anneal) when conductivity becomes metallic, due to an overlapping of the localized states.81 However, the density of the amorphous layer does not equal to that of graphite until approximately 800 C with a sharp drop in density occurring near 500 C.81 Graphitization in the literature is a term that sometimes mean s that the electrical properties of the damaged diamond are metallic, and not that a true graphite material has formed. This process has been simulated by computational methods; molecular dynamic images of the evolution from damaged diamond to graphitic planes were demonstrated by Kalish et al.81,100 Controversies in Diamond Literature Critical dose threshold for amorphization. The critical implantation dose to effect graphitization upon annealing was orig inally determined to be 1 1022 vacancies (vac) cm-3 by Uzan-Saguy et al. in 1995.16 This threshold has been widely referenced in the literature. The method of determination was to implant an amorphizing dose into th e crystal, and after

PAGE 48

48 annealing, etch in boiling acids. The depth of et chable material, assumed to be graphite, was determined by using a buried marker layer. In 2 005, Brunetto et al. theorized the threshold to be just slightly higher, at approximately 2 1022 vac cm-3.101 In 2006, Olivero et al. estimated the damage threshold to be approximately 9 1022 vac cm-3.75 And Prawer noted that the threshold is extremely dependent on implantation temperat ure, and states that graphitization can be completely avoided for high e nough implantation temperatures.102 Displacement energy. In the original study of diam ond that set the graphitization threshold at 1022 vac cm-3, the Ed of one carbon atom was determined to be 45 eV.16 However, other values have been determined in the literature, ranging from 52 eV 81using molecular dynamics to 55 eV 98 by volume expansion methods. Using these various Ed values in SRIM21simulations results in little difference in terms of projected damage range (Rd) and peak damage depth (Rp) in diamond. Displacement spikes. Displacement spikes in diamond ha ve been a controversy in the literature.83,102 Some argue that displacement spikes fr om the ion trajectory relax to form graphite-like material.102 However, it has been theorized th at by using MeV ions, the cap of undamaged diamond keeps the damaged layer under e nough pressure so that this prevents the buried damage layer from relaxi ng to the graphitic structure.89,103 Raman studies have been used to provide justification. Transmission electron microscopy studi es of single crystal diamond. The few TEM studies that do exist have been either plan-vie w or diamonds thinned by oxidation thinning, and have been used to study the defects in HPHT cr ystals. In 2001, plan view TEM studies were done on ion-implanted natural diamond.104 Yin has reported on the de fects in a lab-grown high pressure high temperature (HPHT) diamond crysta l studied using plan view TEM as one method.

PAGE 49

49 96,105,106 Additional TEM studies of HTHP single crystal diamond whiskers 107 and particles 108 have been conducted. However, these sample pr eparation methods were not area selective in nature. Additionally, none of these studies looked at defect creation or evolution. Planar defects have been seen in polycrystalline diamond fi lms, and studied by HRTEM and scanning TEM; these defects consisted mostly of microtwins.109 The only reported cross-sectional TEM study of diamond seems to be by Adams et al.,110 who studied ion beam sputtering and damage in a single crystal diamond. Properties of the Elem ental Crystal Systems As a comparative study between three elementa l crystal systems, the current work used many published properties of three elements involved: C, Si and Ge. Some of these constants and properties are summarized in Table 2-1 for c onvenience and reference. Si, the most widely studied of the three systems, is listed first. In ge neral, Sis property values are in between those of C and Ge, which makes sense as it is between the two in the fourth Group of the periodic table. It is interesting to note, though, that both carrier mobilities are actually lower in Si than in Ge and Diamond. Also listed are some references for graphite, as also being a C crystal system, and these values are used in Chapters 6 and 7 when studying the evolution of damaged crystalline diamond.

PAGE 50

50 Table 2-1. Relevant properties of the Group IV diamond-cubic semiconductors and graphite. Property Si Ge Diamond Graphite Reference Atomic Weight 28.09 72.61 12.01 12.01 111 Melting Point, C 1414 938.25 4440-4773 (melting point, 12.4 GPa) 44734489 (triple point, 10.3 MPa) 111 19 Lattice parameter (a), 5.4282 5.6580 3.567 a=2.461 c = 6.708 112, 19 Tetrahedral Covalent Radius, 1.17 1.22 0.77 n/a 111 Atomic Radius, pm 118 123 77 77 111 Space group Fd3m Fd3m Fd3m P63/mmc 19,113 Bond Dissociation Energy, D298/ kJ mol Si-Si 310 Ge-Ge 264.4 6.8 Ge-Si 297 C-C 595.63 C-Si 447 C-Ge 455.7 11 114 Atomic concentration, atoms/cm3 5.00e22 4.42e22 1.76e23 1.14e23 19,113 Packing fraction for diamond crystal structure 34% 34% 34% n/a 113 Minimum Room Temp. Energy (Band) Gap (eV) 1.12 0.67 5.4 -0.04 115 19 Hole Mobility (room temp), cm2/V s (CRC, pg 12-82 500 1820 1400 n/a 115 Electron Mobility (room temp), cm2/V s 1900 3800 1800 n/a 115 Displacement Energy (ED), eV 15 15 50, 52 SiC: 20 (C) 25-40 (Si) 28 21 101,116 62

PAGE 51

51 CHAPTER 3 ANALYTICAL TECHNIQUES Introduction Transmission electron spectroscopy is one of th e best analytical techniques for comparing the defect formation and amorphization phenomen a of ion-implanted car bon (C) and germanium (Ge) to that of silicon (Si), because direct imaging of the crystal is possible. The most challenging aspect of TEM analysis is the sample preparation, and therefore it shall be treated first in this chapter. Few TEM studies of C in its diamond cu bic form exist, due to the historically difficult preparation technique.96 In order to study ion-im planted diamond, the author created a technique to make relatively quick cross-sectional (XTEM) sa mples of diamond using a focused-ion beam (FIB). This technique was published in the Journa l of Vacuum Science Technology-A,117 and its presentation won the AVS Manufacturing Science and Technology Group Fellowship Grant at the 2005 AVS Internati onal Symposium for best paper as well as Grand Prize in the Materials Characterization Po ster Division at the 2006 Joint Symposium of the FLAVS and Florida Soci ety for Microscopy. After sample preparation of all three materials has been discussed, TEM, as well as the insitu tools of EDS and EELS, will be described. Raman optical scattering was also used to confirm the results of damage to the diamond cr ystal, attempting to determine amophization or graphitization. The simulation tool, SRIM-2006 (T he Stopping and Range of Ions in Matter)21 has been used extensively in this work to si mulate the damage in each crystal created by the incoming ion implantation process. A short descript ion of the benefits and limitations of each of these analytical techniques will be presented.

PAGE 52

52 Sample Preparation for TEM Transmission electron microscopy imaging requi res samples normally less than 200 nm in thickness, and approximately 50 nm of less for hi gh-resolution imaging. The physics of imaging, and why a thin material is requ ired, is discussed in the next section. There are many different sample preparation techniques for different materials.112 Only two will be discussed here, which are appropriate for the semiconductors Si and Ge : cross-sectional (X) TEM sample preparation by automated focused ion beam milling (FIB) scripts and plan-view (P) TEM sample preparation using chemical and mechanical polishing tech niques. Then a novel method of creating XTEM samples for single-crystal diamond will be presented. Silicon & Ge XTEM sample preparation. Using an FEI dual beam FIB, and the provided automated script for creating XTEM samples of Si, XTEM samples of both Si and Ge were produced.118 This is essentially a milling technique in which a source, usually gallium (Ga), ions are accelerated towards a target and sputter ta rget ions from the sample. Mill rates can be adjusted through adjustment of th e dose rate of the ion. Since the Ga ion beam also creates damage of the sample, the bulk material is firs t coated with carbon to protect the surface from initial views using the ion beam. Approximately 150 nm of carbon is sufficient. Once the sample is in the FIB chamber (which is under vacuum), the sample is oriented along the (111) cleave planes (in the <110> direction) so that the fi nal XTEM sample will be viewed in the <110> direction. Platinum (Pt) is then deposited over the surface area of choice, to protect the surface from future milling steps. The carbon and Pt laye rs will be visible in the XTEMs presented in this work. Milling continues until a material sample approximately 15 m 2 m 0.1 m is freed from the bulk. The sample is remove d ex-situ and placed on a copper-supported carbonfilm backed grid.

PAGE 53

53 Silicon & Ge PTEM sample preparation. To view a sample from the top-down in the near surface region, rather than a cross-section, a PTEM method is used. First, a 3-mm section is milled from the bulk, and the surf ace is crystal-bonded to a metal chuck. The material is then mechanically polished using alumni slurry to approximately 0.2-0.4 m thick. At that point, appropriate etch chemistries are used per each mate rial to create an electron transparent region in the center of the 3-mm disk: for Si, a 25:75 buffe red hydrofluoric acid (49% HF): nitric acid (HNO3) solution is used, and for Ge, a 15:85 48% HF: HNO3 mixture is appropriate.112 The HNO3 oxidizes the semiconductor, as shown here with Si: Si(s) + 4 HNO3(l) 3SiO2(s) + 4NO (g) + 2H20(l) (3-1) Then, the HF etches the oxide as shown again with Si: SiO2(s) + 6HF(l) H2SiF6(l) + 2H20(l) (3-2) This is the same mechanism responsible fo r HF etching glass (a s it is also SiOx), therefore, HF is always stored in polyethylene or Teflon cont ainers. Great care should be taken with both chemicals as both are highly corrosive. After aci d etching, the samples are briefly rinsed in water, air or nitrogen dried, a nd are ready for imaging in the TEM.112 Single crystal diamond XTEM sample preparation. Due to the difficulty in creating TEM samples from single crystal diam ond, few studies have been reported.96 In the 1960s, TEM samples were prepared from natural diamond by ox idation thinning, which is flowing oxygen at 750C or carbon dioxide at 1350C over the diamond substrate, th ereby gas etching the crystal.119 However, this process does not protect the su rface and is not selec tive of the area. Other disadvantages of this method are that the sample mu st be boiled in an acid mixture to remove the surface layer of carbon that forms and such high te mperatures may anneal out defects present in the sample.120 In 2001, plan view TEM studies were performed on ion-implanted natural

PAGE 54

54 diamond.104 Yin has reported on the defects in a la b-grown high pressure high temperature (HPHT) diamond crystal studied using TEM as one method.96,105,106 Additional TEM studies of HTHP single crystal diamond whiskers 107 and particles 108 have been conducted. However, these sample preparation methods were not area selective in nature. Recently, Adams 110et al studied ion beam sputtering and damage in a single crystal diamond. The focused ion-beam has been used to crea te TEM samples in single crystal silicon and other materials faster than traditional methods.118 Following FIB sample preparation, ex-situ liftout is common using microman ipulation tools, but in-situ liftout is also possible and preferable for the study of ion-implanted diamond. Ex-situ liftout of diamond specimens is not preferable for two reasons: (1) the film comm only on the back of TEM grids is carbon, thus potentially compromising any EELs spectrum analys is of the ion-implan ted diamond and (2) if the sample is not thin enough for high resolution wo rk, it can not be thinned further in the future. This may become an issue if the source material has been processed further. In addition, the diamond surface must be protected prior to creating the FIB crosssection, otherwise the gallium ion beam could produce unrelated cr ystal damage as deep as 30 nm.118 This method was developed to create XTEM samples from HPHT single crystal diamonds, approximately 5 x 5 x 1 mm3, provided by The Gemesis Corporation. The diamonds in this study were ion-implanted, and after im plantation, a 1000 layer of SiO2 was either PECVD or sputter deposited on the implanted surface, with the thermal exposure of the diamond being 310 C for one minute during PECVD deposition. The purpose of the SiO2 layer was to compositionally separate the carbon of th e crystalline diamond from the conduc tive carbon coating applied next. Approximately 150 nm of carbon was evaporat ed onto the sample, for conductivity and protection of the sample surface.

PAGE 55

55 Process steps. Using a FEI Strata DB235 dual beam scanning electron microscope (SEM) and focused ion beam (FIB), a sample area was selected. A 20 m x 2 m x 1.2 m platinum box was deposited onto the chosen area. Using a 5000pA ion beam sourced by gallium, a trench six microns deep was milled away on three sides of the platinum box. T ilting 45, an undercut using a 3000pA ion beam released the sample wedge from the substrate bulk. The Omniprobe Autoprobe 200TM in-situ lift-out system was then inserted along with the platinum gas injection needle, and placed on the corner of the sample. Pl atinum was deposited to attach the sample to the needle tip. The sample was then lifted out an d retracted with the Omniprobe needle, as shown in Figure 3-1. The bulk sample was then remove d from the chamber. An Omniprobe copper grid was inserted and adjusted in the FIB. The platinum needle and the Omniprobe needle, with sample still attached, were reinserted into the FI B, maneuvered to the copper gird, and attached to the grid with platinum, using a 300 pA ion beam. Once attached, the Omniprobe needle was cut away from the sample using a 5000 pA focuse d ion beam. Both the platinum needle and the Omniprobe needle were then retracted. At this point, the sample was approximately 15 m x 2 m x 2 m; the thickness of the sample is shown in Figure 3-2(a). Tilting the stage normal to the ion column, the sample was thinned, first us ing a 300 pA focused ion beam, then 100 pA and final thinning was completed with a 50 pA beam. All ion beam energies described here were 30 keV. The sample was thinned from a +1 and -1 to normal, and alternating scan rotation setting of +1 and -1, so that the e nd of the sample would be thin moving to a thicker (approximately 200 nm) region at the base section attached to the grid. Figure 32(b) shows the thickness of the sample after the thinning pro cess, approximately 50 nm toward the end, and approximately 200 nm near the base (area attached to grid). The sample can then be viewed in the TEM without further processing.

PAGE 56

56 Results of the method. The single crystal diamond TEM samples produced by this method have been imaged in both regular re solution TEM and high resolution TEM (HRTEM) with good results. Figure 3-3 shows an image obt ained on a JEOL TEM 200 CX clearly showing the layers of platinum, carbon, and silicon oxide above the diamond crystal, and also shows the implant damage from the 1x1014 Si/cm2 dose as it was implanted (w ithout any anneal). Figure 34 (a) shows an XTEM image from a damaged diamond implanted with 1MeV, 1x1014 Si+/cm2, and annealed for 10 minutes at 1050 C. Select ed area diffraction imaging is shown to be possible from the heavily damaged region [Fig. 3-4(b)] as well as the bulk undamaged region [Fig. 3-4(c)]; both SADP show the characteristic pattern of a diamond-crysta l structure as viewed in the <011> beam direction. The method also produces good results when imaged under HRTEM conditions, as shown in Fi gure 3-5. As carbon is a less ma ssive ion than Si or Ge, a clear on-axis HRTEM image is more difficult to obt ain due to its smaller atomic size. However, individual rows of atoms are ab le to be imaged even in a heavily damaged region of a diamond implanted at 77 K with 1 MeV, 115 Si+/cm2. This process does indeed allow all TEM analysis as in traditional cros s-section techniques. Advantages of method. In creating these TEM samples, the total time per sample was less than three hours, a significant reduction compared to traditional sample preparation techniques. In addition to the time savings, another advant age over traditional sample preparation methods that is particularly good for rare substrate material like single crys tal diamond is that very little material is needed, and the source of the sample can be reprocessed, increasing research efficiently and controlling costs at the same tim e. This method differs from other FIB methods by accounting for the nuances of an alyzing single crysta l diamond. Carbon coating is often used to make insulating and semi-conducting samples c onductive for use in the FIB. However, as the

PAGE 57

57 point of this research will be in part, to study the phase tran sformation of crystalline diamond, having a carbon layer deposited di rectly onto the diamond (carbon) substrate could compromise characterization methods such as electron en ergy loss spectroscopy (EELS). To avoid this compromise, a layer of SiO2 was deposited to differentiate the two layers. The top layer, in black [Fig. 3-3], shows the platinum deposited to prot ect the surface from the Ga ion beam milling currents. The three layers of carbon coating, which protect the surface from the Ga ion beam damage, are clearly visible beneath the platinum The next layer is si licon oxide, deposited to separate the carbon deposition from the carbon sample. Finally, the HPHT single crystal diamond is visible with the implant damage ap pearing as a dark band. Other FIB sample preparation methods choose not to use in-s itu lift out, but inst ead use an ex-situ micromanipulation technique. This method involve s removing the sample from the FIB after it has been released from the bulk, and then transf erring the sample from the substrate to a grid by static forces. However, the grid is often backed by a carbon film. If this method were used for the single crystal diamond, characteri zation methods such as EELS could again be compromised due to the presence of carbon film behind the carbon sample. In addition, the precision of this method allows accuracy and easy variation in the orientation of the cross-se ction fabricated. This can be an issue since the traditional orientation of a s ilicon substrate, i.e. (100), is not always known for single crystal diamonds produced vi a the HTHP method. Site specific plan view samples are possible but require a different technique and will not be discussed here. Possible improvements to method. Although this method has proved sufficient, it can be improved. HRTEM lattice images do show that the diamond lattice can be observed; however image resolution can be improved by using lower ion beam energy on the final thinning cuts to impart less damage from the Ga ion beam. This method used a 30 keV beam during the entire

PAGE 58

58 thinning process. A direct 30 keV beam, according to SRIM (using a displacement energy of 45 eV per carbon atom16) created a damage peak at approxi mately 15 nm. However, the thinning beam was not direct on, but only at a glancing angle, so the damage can be assumed to be less. Two characterization techniques we re additionally used to determ ine that FIB damage was kept to a minimum. HRTEM showed that, even on the 1 1015 Si+/cm2 implant, the sample stayed crystalline throughout its entire damaged regi on. Additionally, electron dispersive x-ray spectrometry (EDS) showed Ga (the ion used in this FIB process) was highly present in the platinum deposition layer (as is to be expected) but only in background amount s in the rest of the sample. Although these methods show that the Ga ion beam did not do extreme damage, even less damage could be achieved by switching to a lower energy ion beam for the final thinning cuts. For example, a direct 5 keV Ga ion beam would have a damage peak only 5 nm into the diamond lattice (according to SRIM21) as compared to 15 nm caused by a direct 30 keV beam. Additionally, oxygen gas FIB accessories, available fr om FEI, would allow for faster mill rates. Transmission Electron Microscopy The sample preparation process for tr ansmission electron microscopy (TEM) is challenging (as discussed in the last section). TEM provides high lateral spatial resolution, as high as 0.2 nm point-to-point for some instru ments. Image and diffraction patterns may be obtained from the same sample. The signals used in image and diffraction modes are also used in other TEM in-situ characterization tools such as EDS and EELS. The TEM has an incredible range of magnifications, from 50 to 106 increase in magnification, made possible by the small wavelength of highly-energetic elec trons. Thin samples insure few scattering events to the probe electrons, making signal re solution quite accurate.112,121 Electrons, normally in the energy range of 100-400 keV, are generated by either a field emission gun or a tungsten filament. Using electromagnetic focusing lenses, a beam of coherent

PAGE 59

59 and ideally monochromatic electrons is focused to a thin foil sample.121 Upon passing through a solid sample, the electrons interact with the fo il in a number of ways, as shown in Figure 3-6.122 For TEM, the elastically scattere d electrons (Bragg scattering) generates the bulk of the signal used for imaging and diffraction pattern genera tion. The principles of image and diffraction pattern creation in the TEM are base d on Braggs law of diffraction: n d sin 2 (3-3) where d is the inter-planar spacing (listed in Table 2-1 for the crystal systems used in this study), is the diffraction angle, n is an interger and is the wavelength of the incident electron.122 For diffraction patterns, the intensit y and distribution of the diffract ed spots provides information such as lattice type, the lattice constant (listed in Table 2-1 for the crystal system used in this study) and crystallographic orientation.120 The planar spacing (d) can be determined for a particular crystal system by usi ng the lattice constant (a) and M iller indices (h,k,l) as shown: 2 2 2l k h a d (3-4) The diffraction patterns for the diamond crystal systems, like Si, Ge and carbon in the diamond crystal structure, are shown in Figure 3-7; th e pattern arising from looking at a (001)-surface oriented plan view is shown in Figure 3-7(a), an d from a (011)-oriented cross section is shown in Figure 3-7(b). The diffraction pattern for hexagonal crystal structures is shown in Figure 3-7(c) and (d), which both contain the most intensely diffr acted plane in the graphite crystal system, the (0002) plane.112,123,124For diamond-like carbon films, TEM is a widely-used tool for differentiating between the differe nt crystalline phases of carbon.120Crystalline materials will produce a spot pattern, as already shown; polycryst alline materials display a distinct ring pattern, and amorphous materials produce a hazy ring pattern.121 Normally, selected area diffraction patterns (SADP) are imaged directly from the inte raction with the foil. However, in this study,

PAGE 60

60 two TEM instruments were used. For imaging in the FEI 2010 TEM, HRTEM images were used to generate Fast Fourier Transf orm (FFT) patterns, which look like SADP and provide the same crystallographic information, but are not generated directly from electron-foil interaction. The method of generation, either direct SADP or FFT will be designated on provided figures in this study. In imaging mode, the TEM operator may select to view the transmitted electrons or the diffracted electrons. Using centered bright field (B F) mode, only the transmitted electrons will be viewed on phosphorus screen. Therefore, areas of th e crystal that have scattered electrons due to irregularities in the plane will appear dark, such as damaged crystalline regions or defects. BF TEM is useful for measurements (because the sa mple is not tilted). In weak-beam dark field (WBDF) mode, the sample is brought to a two-beam condition, so that only a certain plane is aligned to the transmitted beam. WBDF TEM micros copy is most useful for imaging defects in the lattice.112 One of the benefits of TEM is that small sample sizes may be used, which is extremely beneficial when working with rare HPHT di amond crystals. Additionally, crystallographic information may be obtained and correlated with the imaged samples, as will be shown not only with SADP but also EELS analysis. The major drawback to TEM analysis has already been discussed in detail: the intense effort involved in making a suffi ciently thin (less than 200 nm) sample. Another drawback of TEM imaging is that three-dimensional data provided by scattered electrons is projected onto a tw o-dimensional phosphorus screen, giving TEM little field of view. Three dimensional data must be derived from tilt ing the sample or adjusting the focal point to different depths in the thin foil.112 Additionally, diffraction techniqu es are not sensitive to small

PAGE 61

61 amounts of amorphous material, and therefore othe r techniques such as Raman and EELS must be used to confirm the presence of amorphous carbon.120 Energy Dispersive X-Ray Spectroscopy Energy-Dispersive X-Ray Spectroscopy (EDS) sy stems can detect X-rays from all the elements in the periodic table above beryllium if sufficient quantities are present, approximately 1-2% by weight. Therefore, it is an excellent tool for detecting chemical species. The main advantages of an EDS system are its sp eed of data collect ion and ease of use.125 The main disadvantage for this study is the weight pe rcent detection limit: dopant atoms and background nitrogen (in the diamond samples) were presen t below the resolution of the EDS system. EDS systems are present in many electron column instru ments; for this study, an EDS system attached to an FEI 2010 TEM was used. EDS uses the X-ray signals produced from highenergy electron interact ion with a solid, as shown in Figure 3-6. X-rays are produced as a result of ionization of an atom where an innershell electron is removed during the interaction w ith the electron. In order to return the ionized atom to its ground state, an electron from an out er shell fills the inner-shell spot, and in the process, must release energy equal to the diffe rence in potential energy between the two shells. As shown in Figure 3-8, the excess energy (which is unique to each atomic transition) is released in the form of a characteristic X-ray [Fig. 3-8(a)] or an Auger electron [Fig. 3-8(b)].122 In the case of electrons with many filled shells, the ejection of one electron from an inner shell may begin a cascade which produces many signals from a single electron-atom interaction.125 For this research, EDS was used to confirm two experimental parameters: first, that a deposited oxide layer did indeed separate th e carbon sample from th e conductive carbon layer applied to protect the surface from the Ga ion beam during sample preparation, and second to confirm that the Ga ion beam did not deposit Ga ions in the surface layer of interest.

PAGE 62

62 The layer of silicon dioxide (SiO2) was deposited on the surface of the crystal, either by plasma enhanced chemical vapor deposition (PECVD) or e-beam sputtering. The PECVD SiO2 adhered better to the crystal surface than the sputter deposited layer; however, at high temperature annealing cycles, the PECVD layer would flake off. The main drawback of the PECVD layer was that it exposed the sample to a thermal budget of 310 C for 3 minutes. For diamond and Si, this is a relatively low thermal budget as compar ed to their melting (and in the case of silicon) regrowth temperatures. However, this was an i ssue for the Ge samples which show some defect evolution and regrowth in the 300-400 C range Sputter deposition was performed at room temperature in the Benton Clean Room Laboratory at the University of Florida. However, it flaked off the sample after even initial anne aling sequences. In order to characterize the exisitence of the SiO2 layer, and to verifiy the aforementi oned lack of Ga ion damage to the diamond crystal surface during sample prepara tion, energy dispersive X-Ray Spectroscopy (EDS) was employed. Figure 3-9 shows the path of the line scan across an ion-implanted diamond XTEM created using the aforementioned sample technique and the equivalent EDS spectrum. As shown, no Ga ion damage is present in the sample, and the SiO2 layer does provide a sufficient separation distance between the carb on coating and carbon sample. The implanted Si ions or the as-grown HPHT nitrogen contaminant atoms were not detectable using EDS, due to its low concentration in the crystal. Electron Energy Loss Spectroscopy In this study, electron ener gy loss spectroscopy (EELS) was used to determine the bonding in the amorphous and crystalline regions of the ion-implantation damaged diamond. In EELS, high-energy electrons are used to probe the electronic structure of the solid through columbic interactions. EELS obtains information from eith er the inner-shell electr ons (K-shell) or from those in outer shells (L, M, N) Electrons (normally of energies of ~5 to 2000eV) traveling

PAGE 63

63 through a solid can lose energy in many ways, as shown by Figure 3-6.120,122 For EELS, the main losses used are plasmon scatteri ng, single-particle electron exc itations (signals from valance electrons) and ioniza tion of atoms (signals from core elect rons). For carbon, EELS can determine the changes in three bonding conf igurations: diamond, graphite and amorphous carbon. For very thin samples, such as CVD grown films, both lo w loss (signals from valence electrons) and high loss (singles from core elect rons) can be used to charac terize the bonding of the solid.120 For slightly thicker samples (greater than 50 nm), low loss signal is obscured by the contribution from plasmon scattering.126 Identification of each carbon phase comes fr om the high energy electron losses due to excitation of core electrons, also known as K-shell electrons, as shown in Figure 3-8.122 These loss peaks are observed as specifically-shaped st ructures in the K-shell ionization loss spectra. These appear at higher energies than the thres hold energy for K-shell excitation due to plasmon creation and interband transitions. The position a nd shape of this region of the loss spectra is very dependant on the density of states in the conduction band (f or diamond) or above the Fermi level (for graphite). These finely defined spec tra can be used to distinguish between diamond, graphite and amorphous carbon, as shown in Fig. 3-10.126 The diamond spectrum consists of three peaks, at approximately 294, 299 and 306 eV (after aligni ng the zero loss peak at zero). The graphite spectra has two almost identically intense pe aks at 286 and 294 eV (again, dependent on the location of the zero-loss peak). Finally, the amorphous peaks occur at the same energy levels as graphite, 286 and 294 eV, however, th e first peak is almost half the intensity of the second peak, as shown in Fig.3-10.126 Reported literature says th at the percentages of sp2/sp3 bonding can be determined by EELS, and is used extensively to characterize chemical vapor deposition (CVD)-grown diamond

PAGE 64

64 films;127 however this requires extremely thin samples on the order of 10-20 nm.126 A drawback of the FIB method of sample preparation leaves the diamond too thick or, in the extremely thin regions, too damaged. Even though final thinning of the sample is performed using a glancing angle of the Ga ion beam to thin the sides of the XTEM sample, some lateral damage (~5 nm) can occur. Therefore, for a sample approximate ly 100 nm thick, as li ttle as 90 nm may be crystalline. To confirm that the amorphous spect rum, a FIB damaged region near the tip of the diamond sample was probed. EELS analysis wa s performed on a FEI Technai TEM at the University of Central Florida. Figure 3-11 s hows the screen-shot from the EELS software package on the Technai system. On the upper left, a BF TEM image is shown, with the line scan shown. The point at which the EELS spectrum is taken is designated by a red cross (in the FIB damaged region), and the spectrum is shown. No te the characteristic amorphous K-loss peaks (compare to the standard spectrum in Fig. 310). EELS data is presented in Chapter 7 to determine the bonding in damaged diamond after amor phization and before and after annealing. Raman Spectroscopy Raman spectroscopy has been the most widely us ed tool for the examination of the types of bonding present in carbon films,120 because it is sensitive to the details of atomic arrangement and chemical bonding. Raman is able to distingui sh various types of car bon: diamond, graphite, amorphous carbon and hydrogenated carbon, as well as stressed diamond (as grown on a substrate) and defect structures.128 Raman is extremely sensitive to graphitic phases of carbon due to its high scattering centers. Raman spectroscopy is able to detect different bonding in materials due to the relationship that each bond vibrates with only on e fundamental vibrational frequency.129An intense monochromatic light beam, norma lly created by a laser, impi nges on a sample. The chemical bonds are distorted by the electric field of the inci dent radiation. When th e radiation passes, the

PAGE 65

65 bonds relax and most of the stored energy is reradiated at the same freque ncy as the incident light. A small portion of the stored energy is transferred to the sample, which excites the virbrational modes of the chemical bonds. These vibrational energies ar e subtracted from the energy of the incident beam and appear as Rama n lines. The separation of the Raman lines from the frequency of the incident light is a direct measure of the vi brational frequencies of the bonds in the sample.129 Both the position and the width of the Raman peaks provide information on the type of carbon present, the structural perfection and the inte rnal stress state, as shown in Table 3-1 which lists the various peaks and widths of Raman peaks produced by various forms of carbon.120 The standard spectrum for natural single-crystal diamond is shown in Figure 3-12(a). The parameter of full width at half-maximum (FWHF) is a measure of the peaks width (in cm-1) taken at half of the maximum intensity of the peak, and vari es with the type of bonding detected. Raman spectrum taken from an unimplanted HPHT diam ond used in this study is shown in Figure 312(b). This spectrum has the characteristic tall, skinny peak (a low FWHM of 6 cm-1); however, it is shifted slightly by 2.5cm-1 from the standard diamond peak. Peak shifting may be caused by a number of stresses to the crystal: foreign atoms, stress or defects.128 For the diamonds used in this study, the shift may be due to the nitroge n impurity present in the HPHT samples, obvious due to the yellow-orange color of the diamond wafers. In this study, the Raman system used a He-N e Laser at 632.81 nm with a spot size of approximately 5 m, located in Dr. David Hahns lab in the Department of Mechanical Engineering at the University of Florida. Spec tra of various carbons are shown in Figure 3-13 (a), 127 and can be compared to a spectrum from a gr aphite standard using this system, shown in Figure 3-13(b). The spectrum contains the two charac teristic peaks of graphite, as listed in Table

PAGE 66

66 3-1. Note the low intensity of the signal (the y-axis) due to the opaquene ss of the graphite. All Raman spectra used in this study were taken from samples being placed on a glass slide. In order to confirm that the glass slide did not contribute to the signal, the spectrum from the glass slide (unfocused) is presented in Figure 3-13(c). The low intensity and consistent background signal confirm that the glass slide did not detrimenta lly affect the spectra used in this study. The main drawback of the Raman method is the spot size, especially for this study. The 5 m spot size was larger than the band of amorphous carbon material in the diamond XTEM samples, so although cross-section analysis wa s performed, it must be noted that both the crystalline and amorphous regions were represented in a single spectrum. Bulk measurements of the implanted diamonds were also used to determ ine the crystalline or amorphous nature of the material. A smaller spot size would be ideal for analysis of the current experiment. However, nano-Raman systems are extremely new, and have only recently begun use as a characterization tool in the research arena, a nd may become a future tool for study of XTEM ion-implanted diamond.127 Stopping and Range of Ions in Matter (SRIM) The Stopping and Range of Ions in Matter (SRIM)21 is a widely used tool to simulate ionimplantation damage in crystals. Introduced in 1989, SRIM has been upgraded approximately every five years. The version of SRIM used for this study was the 2003 version for experiments performed before 2006, and the 2006 version fo r experiments since. However, no significant changes were made to the 2006 version of SR IM concerning the elements or stopping powers involved in this study. SRIM is known to be very accurate for simulating the ion-implantation damage in crystals.19 It has been used to study the threshold damage in Si,35 the threshold damage in diamond,16 and countless other studies into the phe nomena arising in semiconductors after ion-

PAGE 67

67 implantation. Advantages of SRIM is that it is easy to use and widely a dopted in the community for estimation of ion-implanted damage. However, the limitations of SRIM may be significant, depending upon the experiment. First of all, SRIM only simulates damage into crystalline systems. SRIM does not simulate damage into an amorphous material. Also, it does not estimate when a material becomes amorphous. SRIM does not take into account dynamic annealing, which is the heating of the lat tice that takes place during implan tation, either due to implant temperature or heating by a high dose rate. Ther efore, SRIM simulates low temperature (liquid N2) temperature implants better than highe r temperature (room and above) implants. As few to no XTEM studies of ion-implan ted diamond are reported, it was important in this study to confirm that SRIM does simulate the damage into diamond after non-amorphizing ion-implantation well, before beginning any seri ous experiments that used SRIM results to hypothesize about defect formation or amorphi zation in diamond. XTEM diamond samples would be used to compare SRIM results to the experi mental results. First of all, alterations to the SRIM defaults of carbon as a target material were necessary, since the default is for the graphite form of the carbon crystal. The paramete rs adjusted were density (2.253 g/cm3 to 3.5 g/cm3, graphite to diamond respectively) and displacement energy, Ed (28 to 50 eV, graphite to diamond respectively). All simulati ons were run using the Full Ca scade mode. For a detailed explanation for the choice of disp lacement energy, see Chapter 2. All as-implanted TEM images of the nonamorphizing implants matched well to the TRIM simulations. Figure 3-14(a) shows th e plotted SRIM results of a 1 MeV Si+ implant into diamond, plotting vacancies, interstitials and fina l ion depths. Figure 4-14(b) is a bright-field (BF) TEM image of (011)-oriente d diamond implanted with 1 1014 Si+/cm2 at 303 K. BF TEM imaging conditions were used as to not intr oduce tilt into the image. The ion-implantation

PAGE 68

68 induced damage band matches well with the simulations prediction. The damage for this dose and energy seems to be centered on the projected range, rather than the damage peak. The (100)oriented diamond with the same dose ha d similar results. A higher dose, 1 1015 Si+/cm2, and its simulation results are shown in Figure 3-15(a) for a (001)-oriented diamond. The SRIM simulation is the same as it was for the lower dos e, since dose is not taken into account in the implantation dose (only energy and species). Howe ver, with the order-of-magnitude higher dose, the damage band is much clearer in the on-axis (011) BF TEM image (Fig. 3-15(b)). The damage band is centered equally between the damage peak and the projected rang e. Although the damage is quite clear, the crystal di d not create a continuous buried amorphous region. This was confirmed by HRTEM imaging (Fig. 3-15(c)), which revealed a crystalline lattice image even in the middle of the damage band. Further confir ming the imaging results is the fast Fourier transform (FFT) selected-area di ffraction pattern of th e Figure 3-15(d) image showing a perfect diamond pattern. These results confirm that SRIM is an accurate simulation tool for nonamorphizing ion-implants into diamond, and has been used extensively in this study.

PAGE 69

69 Figure 3-1. In-situ lift-out of diamond wedge, as viewed using the ion beam which is at a 52 angle to the stage. The diam ond sample is approximately 3 m deep on the far side, and the surface of the sample has been protected by a layer of platinum.

PAGE 70

70 (a) (b) Figure 3-2. The sample thickness before (a) and after (b) the thinning process.

PAGE 71

71 Figure 3-3. A TEM image of an as-implanted (110) single crystal diamond implanted with a 1 MeV 1x1014 Si+ cm-2 at a temperature of 30 C. Platinum Carbon SiO 2 Single Crystal Diamond Surface Implant Damage

PAGE 72

72 Figure 3-4. XTEM of (100) or iented diamond after a 1 MeV 1x1015 cm-2 Si+ implant (Ti = 30 C) annealed at 1050 C for 10 min: (a) XTEM image (labels denote platinum, carbon coating and surface) (b) SADP of top region (c) SADP of bulk. (a) (b) (c)

PAGE 73

73 Figure 3-5. HRTEM image showing the middl e of a damaged diamond layer from a (001)oriented diamond implanted at 77 K with 1 MeV, 115 Si+/cm2. No continuous amorphous layer exists; pockets of amorphous material reside in a crystalline matrix.

PAGE 74

74 Figure 3-6. Schematic diagram of the signals created during transmission of a high-energy electron beam through a thin solid sample. 122

PAGE 75

75 Figure 3-7. Single-crystal s pot transmission elec tron diffraction patterns for the diamond structures (diamond, Si and Ge) for the (a ) [001] and (b) [011] zone axis, and the hexagonal crystal structure for the (c) [ 2110] and (d) [0110] zone Asterisks (*) indicate spots occurring in the pattern from double diffraction, cro sses () indicate the positions of forbidden reflections, but occu rring in the pattern by double diffraction, open circles ( ) indicate additional weaker spots; z is the zone axis and B is the beam direction. 112

PAGE 76

76 Figure 3-8. De-excitation mech anisms for an atom that has undergone K-shell ionization: (a) emission of a characteristic K X-ray and (b) emission of a KLM Auger electron. 122

PAGE 77

77 (a) (b) Figure 3-9. TEM and EDS char acterization of sample. (a) TEM image of a (001)-oriented diamond implanted with 1 MeV, 1 1015 Si+cm-2 at room temperature, and annealed at 1050 C for 10 minutes, with the EDS line scan data overlaid. (b) EDS spectra of the line scan of the TEM sample showi ng relative counts of elemental species.

PAGE 78

78 Figure 3-10. The low-loss and K-ionization regions of the energy-loss spectra of three allotropes of carbon as detected using an EELS system. 126

PAGE 79

79 Figure 3-11. Screenshot of data displayed from the EELS system used for this study.

PAGE 80

80 Table 3-1. Major Raman peaks from di amond, graphite and related materials. 120 Peak Position Type of Carbon Description ~1140 cm-1 Small size (<0.1 m) cubic diamond Occasionally observed in diamond films with very small grain sizes (<0.1 m) 1315-1326 cm-1 Hexagonal diamond Broad band, observed in shock wave produced diamond 1332 cm-1 Cubic diamond First order peak with full width at half maximum of 1.9 cm-1 for natural diamond 1345 cm-1 Amorphous carbon Broad band. It becomes a shoulder of the 1550 cm-1 band when the material is hydrogenated 1355 cm-1 Microcrystalline graphite Ob served in materials with small grain sizes 1550 cm-1 Amorphous or diamond-like carbon Broad band 1580 cm-1 Graphite First order peak 2458 cm-1 Cubic diamond Second order peak 2710 cm-1 Microcrystalline graphi te Second order peak 3240 cm-1 Graphite Second order peak

PAGE 81

81 (a) (b) Figure 3-12. Raman spectra of diamond. (a) Raman spectrum of natural diamond 128 (b) Peak from bulk un-implanted diamond used in our study. Peak is at 1335 cm-1 with a FWHM of 6 cm-1, which is a 2.5 cm-1 shift from the reported diamond peak. The shift may be due to the nitrogen impurity (givi ng rise to the yellow-orange color of the diamond).

PAGE 82

82 (a) (b) (c) Figure 3-13. Raman spectra of carbon allotrope s. (a) Comparison of typical Raman spectra in carbons. 127 (b) Graphite standard There are peaks at 1335 cm-1 (FWHM= 49cm-1) and 1587 cm-1 (FWHM= 23 cm-1). (c) Glass slide background spectra.

PAGE 83

83 Figure 3-14. Comparison of simulation data to e xperimental images. (a) SRIM plotted data of damage created by Si+ implantation at 1 MeV into diamond, showing total vacancies, interstitials and ion range. (b) BF X TEM image of (011)-oriented diamond Si+ implanted at1 MeV to a dose of 1 1014 cm-2.

PAGE 84

84 Figure 3-15. Comparison of simulation data to e xperimental images. (a) SRIM plotted data of damage created by Si+ implantation at 1 MeV into diamond, showing total vacancies, interstitials and ion range. (b) BF-XTEM image of Si+ implantation at1 MeV to a dose of 1 1015 cm-2 into (001)-oriented diamond. (c) HRTEM of damaged area. (d) FFT of damaged diamond HRTEM image.

PAGE 85

85 CHAPTER 4 RESULTS FROM NON-AMORPHIZI NG IMPLANTS INTO GERMANIUM The literature review in Chap ter 2 presented current contr oversies in the differences between defect formation in germanium (Ge) and silicon (Si). Both crysta ls have been reported to form the interstitial {311} defect for non-amorphizing implants but only in light ion or electron irradiation in Ge. 10-12,29,49 Additionally, recent research into SiGe alloys have resulted in results that showed that with increasing Ge content, {311} defects do not form 13 and that possible vacancy defects are creat ed in the near surface region. 14 To explore the Ge {311} controversy, cro ss-sectional transmissi on electron microscopy (XTEM) was used to study defect formation a nd evolution in (001) Ge and Si wafers Si+ implanted at 40 keV and 1 MeV with doses of 1 1014 cm-2. As expected, upon annealing, {311} extended defects form and subsequently dissolve at the projected range for nonamorphizing implants into Si. In Ge, however, no {311} defect formation is observed for the 1 MeV non-amorphizing Si+ implant after annealing at temperatures between 350 850 C. Instead, small dot-lik e defects are observed after 1 MeV Si+ implantation into Ge, which dissolve upon annealing between 650 750 C for 10 minutes. Some evidence of dot-like defects is also observed in the lower energy implant regime, however these defects seem to be just at the level of detection for XTEM. Introduction There is interest in replacing the channel mate rial of transistors with Ge instead of Si because of its significantly higher carrier mobilities. 115 Recent advances in the use of high-k dielectrics have assisted in this regard by reducing the need for a stable Ge-based gate oxide. Additionally, SiGe alloys are being used in the cu rrent generation of devices to induce strain in the transistor channel, which in turn increases carrier mobility.130 Ion-implantation remains the

PAGE 86

86 preferred method of introducing dopants into the extension region of transistors next to the gate.131 However relatively little knowledge of the formation and evolutio n of ion-implantation related defects in Ge exists Early studies of electron 10,29 or light ion (hydrogen (H+) and helium (He+)) 11,12,49 implants into Ge reported {311} defect fo rmation. As recently as 2005, Akatsu et al. reported observing {311} defect s in bulk Ge implanted with He+ at doses between 5 1015 and 7 1016 cm-2, at energies between 15 and 80 keV, and annealed at temperatures between 280 C and 400 C for 30 to 60 min. 11 The {311} defect, however, has not been reported in Ge for any ion species heavier than He+. Recent studies of amorphizing implants into Ge have not reported any rod-like extended defects for low energy implants. 54 To summarize, extended defects have been reported for light ion and electron irradiation but have not been seen for low energy implants using typical dopant atoms. It is well known that in Si, extended defect formation occurs for a wide range of implant and annealing conditions. 9 Recent studies in SiGe alloys suggest that above 25% Ge, no {311} defects are observed, though only alloys were studied. 13 Hence, the purpose of this chapter is to understand extended defect formation and evolution in bulk Ge after non-amorphizing Si+-implantation at high energy. Experimental Design Undoped (001) Czochralski grown high-purity Ge wafers and undoped (001) Czochralski grown Si wafers were Si+ implanted at 1 MeV and 40 keV with doses of 1 1014 cm-2. Samples were held at 30 C during implantation and tilted 7 relative to the incide nt ions to avoid ion channeling. In Si, this implant regime is known to produce easily visible {311} extended defects after annealing. The 1 MeV Si+ implanted wafers were capped with 100 nm of silicon dioxide (SiO2) via plasma-enhanced chemical vapor depos ition (PECVD) at 300 C for 3 min to prevent possible oxidation during furnace annealing. Su bsequent annealing was conducted in a Lindberg tube furnace for 10 minutes in a flowing nitrogen (N2) environment at temperatures ranging from

PAGE 87

87 350 C to 850 C for Ge, and 650 C to 950 C for Si. This broad range of temperature for Ge encompasses those below the amorphous regrowth temperature of 400 C132 and near the melting point of 938 C111. An FEI focused-ion beam and ex-situ li ft-out system was used to prepare the Si and Ge XTEM samples, while traditional polis hing methods and etch chemistry appropriate to each material was used to prepare plan-view TEM (PTEM) samples. 133 Transmission electron microscopy analysis was performed on a JEOL 200CX microscope using g220 weak-beam darkfield (WBDF) imaging conditions and a JEOL 2010 microscope using on-axis bright-field (BF) imaging conditions for high-resolution imaging. Results Upon Si+ implantation at 1 MeV, and without any annealing, WBDF XTEM imaging revealed small dot-like defects in Ge extending from the surface to 1.1 m below the surface. In Si, however, no discernable defect s were observed prior to ann ealing. Upon annealing, Ge defects and Si defects display radically different mo rphology, as seen on the same scale using WBDF-XTEM imaging. In Ge, defects appear as dot -like defects as presented in Figure 4-1(a). These defects are too small to produce a clear phase contrast image when imaged using BFHRTEM imaging, shown in Figure 4-2. The defect s are an average size of approximately 12 nm in diameter over all annealing condi tions. However, the defects in Si appear as rod-like defects, elongated along <110> directions, a nd in this study were at their coarsest after annealing for 10 min at 850 C (60% of the melting point of Si (TMP Si)), as shown in Figure 4-1(b). These rodlike defects are commonly referred to as {311} defects as the defects have {311}-type habit planes. 13 Microstructural evolution for 1 MeV Si+ implantation into Ge and Si is shown in Figure 43. The projected range (Rp) of the damage in Ge is pred icted to be approximately 0.91 m below the surface, while the damage peak (Rd) should occur at approximately 0.72 m, according to

PAGE 88

88 simulations.21 This is in good agreement with TEM an alysis, as the dens ity of defects was greatest between Rd and Rp. The simulation also agrees with the experiment for the implant into Si, with Rp and Rd of approximately 1.19 m and 1.04 m, respectively. In Ge, the dot-like defects which appear im mediately after implanta tion are shown after a 350 C anneal for 10 min in Figure 4-3 (a). Th e defects decrease in dens ity after the 450 and 550 C anneals for 10 min and remain until finally dissolving at temperatures between 650 and 750 C, as shown in g Figures 4-3 (b)(d). The defect s decrease in density but their average diameter remains at approximately 12 nm. However, in Si, after annealing for 10 min at 750 C, extremely small rod-like defects begin to appear in densities right at the limit of detect ion, but are visible in the micrograph presented in Figur e 4-3(e). After 10 min at 800 C, the defects have reduced in number but increased in size, as shown in Figure 4-3(f). This trend continues at 850 C, where the defects have grown into large, easily visible {311} extrinsic defects, averaging approximately170 nm in size with some as larg e as 300nm, as evident by the micrographi in Figure 4-3(g). Presented in Figur e 4-3(h), after 10 min at 950 C, the defects in Si have dissolved. For the lower energy implants, neither Si nor Ge shows any defects after initial implantation as observed using WBDF XTEM an d PTEM methods. Upon annealing, Si forms the characteristic {311} defects, visible bot h in WBDF-XTEM and PTEM imaging conditions, as shown in Figure 4-4 (a) and (b), respectively. The majority of the defects lie in the region between RD (~ 36nm) and RP (~ 68nm), though some defects ex tend to the end-of-range (EOR) of the implant, as observed in Figure 4-4( a). In contrast, the Ge implanted with Si+ at 40 keV, shown in Figure 4-5, shows no defects in WB DF-XTEM for any annealing temperature, though for some annealing temperatures 450 and 550 C, dot-like defects are vi sible in PTEM, presented in Figure 4-6(b). The fact that the defects are ob served in the PTEM micrographs and not in the

PAGE 89

89 XTEM micrographs, presented in Fi gure 4-6(a), coupled with the lack of repeatability in viewing these dot-like defects at various anneal temper atures, may indicate that these dot-like defects seen in the Si+ implanted at 40 keV Ge samples are artifacts of the PTEM sample preparation process. Discussion Si defects have been well documented to follow a specific process. Theoretical submicroscopic interstitial clusters (SMICs) evolve into small {311} defects, and with increasing thermal budgets, larger defects grow at the expense of smaller defects. 17 In low dose Si+ implantation regimes, like 1 1014 cm-2, and with a high enough thermal budget, like 950 C, the large defects become unstable and dissolve.17 However, the evolution of the defects in Ge, under this same implant energy and dose regime, is qui te different, as shown in Figure 4-7. Though the density of the defects decline with increasing ann eal temperature (similar to Si, Fig. 4-7(a)), the size of the defects stays constant in sharp contrast to the incr easing size of the defects in Si. Defect Density Analysis Plan-view quantification of Si+ implantation at 1 MeV is not possible because the defects are deeper than the Ge extinction distance for TEM analysis. 112 However, defect count analysis is possible in the XTEM samples because these samples were prepared using a standard automatic FIB process, so all samples are approximately the same thickness of approximately 100 nm. The density of defects can be derived from these XTEM images by the following method: each sample was made using the same automated XTEM procedure on a dual beam FEI Focused Ion Beam instrument, which resulted in samples uniformly thick, approximately 100nm. Images of defects were then taken at vari ous magnifications, always using the WBDF g220 imaging condition. Two different scientists inde pendently counted and measured the defects, with no significant discrepancy between the two scientists. Usi ng this method, the resolution

PAGE 90

90 limit for the detection of small defects ( 10 nm in diameter) is approximately 1010 defects/cm3. Ge defect counts af ter anneals of 750 C and 850 C for 10 minutes are right at this detection limit, while Si defects after anneals of 650 C and 950 C were below this resolution (no defects were observed). Using the above method, defect densities in bot h Ge and Si were determined. In Ge, after annealing at 350 C for 10 minut es, a defect density of 2 1012 cm-3 was observed. At 450 C, the density of defects declin ed to approximately 5 1011 cm-3, and then leveled off at approximately 2 1011 cm-3 after anneals at 550 and 650 C. Defects in Ge dissolved completely at 750 and 850 C, and the defect count s in these images were right at the detection limit of this method. This is gr aphically represented in Figure 47(a). Interestingly, the average size of the defect remains the same, approxima tely 10 15 nm throughout the microstructural evolution, as evidenced by Figure 4-7(b). The Ge defects after 10 min at 450 C may be small loops. However, either the small size or the offaxis habit plane preven ted obvious identification using on-axis high-resolution phase contrast imaging. No {311} defects were observed in the high-resolution images. Defect Dissolution Rate Assuming an Arrhenius relations hip between temperature and the dissolution rate of the Ge defects, an activation energy and time consta nt were derived. Begi nning with the following two formulas, where D is the dissolution rate of the dot-like defects and is a function of time and temperature, D0 is the number of defects at time zero, t is time and is the time constant: / 0) (te D T t D (4-1) is more explicitly given as: kT Eae/ 0 (4-2)

PAGE 91

91 where is a pre-exponential factor, Ea is the activation energy, k is Boltzmanns constant and T is the absolute temperature. From the defect analysis, the number of defects after implantation, but before anneal ing was approximately 1.18 1012 cm-2. However, this makes the activation energy calculation difficult, as the firs t value after annealing is approximately 1.92 1012 cm-2, larger than the initial count. For the sake of calculation, D0 is estimated to be slightly larger than the 350 C quantity. Interesting, the increase, and then decrease, in visible defects between the as-implanted state and subsequent anneals may indicate that more than one mechanism is contributing to the defect evolution. Thus, at a given temperature may be estimated using equation (4-1). The value of Ea is obtained by plotting the logarithm of versus 1/kT and determining the slope, as pr esented in Figure 4-8. Doing this, Ea is approximately 0.2.1 eV. Defect Location Analysis Defect location within each material differs between Ge and Si. In the 1 MeV Si+ implant into Ge, defects are seen immediately after im plantation and stretch from the surface to the predicted projected range of 0.91 m. Defects in the same implant into Si are not seen until after annealing for 10 min at 750 C, and are concentrated ar ound and below the predicted RP of 1.19 m. In Ge, the defect density peaks at depths shallower than the projected range, whereas in Si where the majority of defects reside below the pr ojected range. Additionally, defects in Ge seem to dissolve more rapidly near the surface, wherea s in Si, the ripening process of each defect is not affected by its proximity to the surface, as shown in Figure 4-3. This phenomenon may be due to a difference in the stability of the v acancy population in the near surface region of Ge versus Si.134 This hypothesis gains merit in light of a recent report that in SiGe, MeV implants into 30 atomic % Ge concentration SiGe alloys showed near-surface vacancy nanocavities,

PAGE 92

92 confirmed by positron analysis. 14 Crosby et al. also noted differences in defect formation for increasing concentrations in SiGe alloys. The expe riment used SiGe, with 0, 2, 5, 25, 35, and 50 atomic % Ge, with Si+ implantation at 40 keV, with a dose of 1 1014 cm-2 and annealed at 750 C for various times. They noted {311} defect formation, similar to Fig.4-4 (b) for low concentrations of Ge. At 25 atomic % Ge a nd subsequent higher concentrations, no {311} defects were observed, but instead dislocation loops formed immediately. No XTEM analysis was reported in the previous study to compare to the current results. Th e results of this study agrees with Crosbys presented trend as no {311} defects were visible in Si+ implanted pure Ge. Future work focusing on the defect formation in lower energy implants into pure Ge in necessary to confirm the lack of defects shown in this study. Conclusion Unlike the original hypothesis proposed in the introduction, Ge does not display the characteristic behavior of Si for non-amor phizing implants. Although {311} defects were reported for H+ implantation in Ge11 and for other electron or light ion implants, 10,12 no {311} defects, platelets, or any type of planar defects were observed fo r heavier ion implants. This is consistent with Crosby et al.s observations of d ecreasing density of {311} de fects in SiGe alloys with increasing Ge content. 13 Crosby suggested that the bond strength111 differences between Si and Ge may cause the {311} defect to become thermodynamically unfavorable at high Ge content. Unlike Si, the location of the defects in Ge is closer to the surface implying some of the defects may be intrinsic. In addition, the size di stribution of the defects in Ge remains relatively constant throughout the annealing process. Thes e results suggest that {311} defect formation may be unique to Si. Future work regarding extended defects in Ge could focus on characterizing the dot-like defects and clarifying the forma tion mechanism. Chronological stud ies could be used to better

PAGE 93

93 quantify the activation energy of the dot-like defects. Additiona lly, positron analysis could be used to confirm a vacancy mechanism at work for potential intrinsic near-surface defects as opposed to extrinsic defects toward s and beyond the projected range. The results of this chapter show that, for non-amorphizing implants, defects do form in Ge but are fundamentally different than those seen in Si. Amorphization of Si is a processing step that has gained importance, but also results in stable defects that may be detrimental for device performance. The next chapter, then, will focus on whether these detrimental defects also form after amophizing implants into Ge, and whether the stab ility of the defects diffe rs from that of Si.

PAGE 94

94 Figure 4-1. WBDF-XTEM images of (a) dot-like defects in Ge Si+ implanted at 1 MeV with a dose of 1 1014 cm-2 following annealing at 450 C for 10 min and (b) {311} defects in Si with the sample implantation following annealing at 850 C for 10 min.

PAGE 95

95 Figure 4-2. High resolution BF-TEM imag e of a dot-like def ect seen in Ge Si+ implanted at 1 MeV with a dose of 1 1014 cm-2 after annealing for 10 min at 450 C.

PAGE 96

96 Figure 4-3. WBDF-XTEM images of Ge Si+ implanted at 1 MeV with a dose of 1 1014 cm-2 after 10 min of annealing at (a) 350 (b) 450 (c) 650 (d) 750 C and Si with the same implantation after 10 min of anneali ng at (e) 750 (f) 800 (g) 850 and (h) 950 C.

PAGE 97

97 Figure 4-4. WBDF images of Si Si+ implanted at 40 keV with a dose of 1 1014 cm-2 after annealing for 10 min at 750 C:(a) XTEM and (b) PTEM.

PAGE 98

98 Figure 4-5. WBDF-XTEM images of Ge Si+ implanted at 40 keV with a dose of 1 1014 cm-(a) as implanted, and annealed for 10 min at (b) 350 (c) 450 (d)5 50 (e)650 (f)750 C

PAGE 99

99 Figure 4-6. WBDF images of Ge Si+ implanted at 40 keV with a dose of 1 1014 cm-2 and annealed for 10 min at 550 C:(a) XTEM and (b) PTEM.

PAGE 100

100 Figure 4-7. Microstructure e volution of Ge and Si after annealing for 10 min at various temperatures: (a) defect density versus temp erature and (b) average defect size versus temperature. The TEM detection limit in (a) is approximately 1010 cm-3.

PAGE 101

101 Figure 4-8. Plot of the logarithm of vs. 1/kT for 1 MeV Si+ implantation into Ge after 10 min of annealing. The value of Ea is approximately 0.2 eV 0.1 eV.

PAGE 102

102 CHAPTER 5 RESULTS FROM AMORPHIZING IMPLANTS INTO GERMANIUM The previous chapter showed that Germanium (Ge) does not form {311} defects in either the low or high energy Si+ implantation regime in which these defects form in Si. The plus-one model, which states for every ion implanted there should be an inte rstitial remaining after initial implantation, only holds for a sub-amorphizing implant, and not an amorphizing implant. However, extended defects, such as {311} def ects and dislocation loops, are seen in Si for amorphizing implants after annealing. In addi tion, for buried amorphous layers, Si has shown other defects as well, including regrowth-related, clamshell a nd end-of-range (EOR) defects. 9 This chapter presents the results from amor phizing implant experiments. Ge which is Si+ implanted at 1 MeV at a dose of 1 1015 cm-2 creates a buried am orphous layer that, upon regrowth, exhibits EOR, regrowth-related, and clamshell defects. Unlike Si, no planar {311} defects are observed. The minimal EOR defects are small dot-like defects and are very unstable, dissolving between 450 and 550 C. This is in c ontrast to Si, where the EOR defects are very stable. The amorphous layer results in both re growth-related defects a nd clamshell defects, which were more stable than the EOR damage. Ge implanted with Si+ at 40 keV at the same dose creates an amorphous layer th at extends from the surface to approximately 70 nm deep. No defects were seen upon regrowth in Ge, while {311} defects and loops are observed in Si. Introduction The potential benefits of usi ng Ge as a channel material we presented in Chapter 4, and need not be repeated. Despite the renewed intere st in Ge, relatively little knowledge of the formation and evolution of ion-implantation rela ted defects in Ge exists. Amorphization is a process step that, in Si, assi sts in the activation and (non)diffusion of dopants. However, the stable EOR defects that form during SPER in Si can be detrimental to device performance and

PAGE 103

103 affect production yield. Theref ore, it is an important phenom ena to study when looking at alternate channel materials. Amorphization is a phenomenon by which the crys tal lattice loses l ong range order after ion-implantation past a critical da mage point. It is associated with a reduction in material density, approximately 1-2% in Si, Ge and SiGe alloys, which causes the crystal to swell.56,58,61,66 Many factors influence whether amorphization will occur, including implant temperature, dose weight, mass of the ion, mass of the target atoms, en ergy of the ion and dose rate. Normally, adding energy to the crystal in the form of heat recr ystallizes the damaged material, and the crystal regrows from the bulk seed base and eventually repairs the entire lattice. Non-intuitively, ionimplantation can also assist with crystalline regrowth after amor phization in Si and Ge, due to heating from the ion-beam.57 With extremely high energy and h eavy ions, a void-filled phase is also formed alongside the amorphous phase in Ge.58 In Si, amorphization results in EOR defects th at are initially obser ved as planar {311} defects and evolve into stable dislocation loops upon annealing.6 End-of-range defects can dominate dopant diffusion and leakage in junc tions formed using amorphizing implants. In addition, a buried amorphous layer has been show n to produce regrowth-related defects as well as clamshell defects in Si.9 Recent transmission electron microscopy (TEM) studies of dopant implants into Ge reported no EOR damage forma tion for amorphizing implants.52-55,135 However these were lower energy implants and the EOR defect density is known to increase with increasing implant energy. 136 Extended defects in Ge, reported as {311} defects, have been observed after nonamorphizing light ion implantation 11 and electron-irradiation, 10 though no published reports exist on the formation of stable regrowth related defects or clamshell defects in implanted Ge.

PAGE 104

104 Experimental Design Undoped (001) Czochralski grown high-purity Ge wafers were Si+ implanted in a tandem accelerator at 1 MeV and 40 keV at 30 C to a dose of 1 1015 cm-2 with 7 of tilt. The wafer implanted at 1 MeV was capped with 100 nm silicon dioxide (SiO2) via plasma-enhanced chemical vapor deposition at 300 C for 3 mi n to prevent possible oxidation during furnace annealing. Subsequent annealing wa s conducted for 10 min in a nitrogen (N2) ambient at temperatures between 350 850 C in a Lindbe rg tube furnace. Solid phase epitaxial recrystallization (SPER) of ion-implantation induced amorphous layers in Ge is known to occur at approximately 400 C. 51,132 After annealing, cross-sect ion TEM samples (XTEM) were prepared using a FEI Dual Beam Focused Ion Beam and Omniprobe lift-out system. Plan-view TEM (PTEM) samples were prepared of the samples implated at 40 keV using traditional techniques and appropria te etch chemistries. 133 Plan-view TEM samples of the samples implanted at 1 MeV were not performed because the projected range of the implant is 9400 (940nm), and the extinction distance in Ge for the 220 reflection is 452 (45nm); therefore, defects seen at or beyond the projected range of the implant could not be imaged via PTEM 21,112 Transmission electron microscopy analysis was performed on a JEOL 200CX microscope using g220 weak beam dark field (WBDF) imaging co nditions and a JEOL 2010 microscope using onaxis bright-field (BF) high-resolution imaging. Results High Energy Implant 1 MeV The implant conditions created a buried am orphous layer starti ng approximately 0.1 m below the surface and extending to a depth of 1.2 m as shown in Fig. 5-1(a). The WBDF image shows that the upper and lower amorphous-crystalline ( -c) interfaces are rough. This agrees well with the rough -c Ge interface previously repor ted after an amorphizing implant. 53,54

PAGE 105

105 Transmission electron microscopy analysis shows th at the near surface region is not completely crystalline but rather contains pockets of crystalline material embedded in the amorphous Ge as shown in Figure 5-2(a). High resolution TEM latti ce imaging confirmed the crystalline nature of the pockets of embedded Ge in the near surfac e region. Transmission el ectron microscopy of the as-implanted lower -c interface also shows pockets of crystalline Ge embedded in the amorphous phase along the rough interface region, s hown in Fig.52(b). The transition region for the upper interface is approximately 140 nm, while the lower interface is much more abrupt at 65 nm. After annealing the sample at 350 C for 10 min, approximately 100 nm of regrowth was observed, as shown in Figure 5-1(b). This ag rees well with Csepregi et al.s reported 350 C regrowth rate of (001) Ge of 9 nm/min. 132 This regrowth created a significantly smoother -c interface. End-of-range damage is observed after and consists of small dot-like defects that extend beyond the -c interface. End-of-range dislocation loops in implanted Ge have not been previously reported. Planarization of the bulk interface occurred, while no well-defined crystalline advancing front developed in the near -surface region. Figure 53(a) shows that after annealing at 450 C for 10 min, the Ge has completely recrystallized. Regrowth-related defects, clamshell (or zipper) defects, and EOR defects are observed. Regrowth-related defects, which appear as threading dislocations with a primarily vertical line direction in the image, are the result of imperfect solid phase epitaxial recrystallization. Thes e defects are known to nucleate at either a rough interface or when the solid phase epitaxial growth process proceeds around an embedded crystalline region. 137 The density of regrowth-related defects appears greater in the layer that nucleated at the surface. This may be th e result of the wider transition region near the surface.

PAGE 106

106 Clamshell defects occur when the two advanci ng crystalline growth planes meet and are approximately parallel to the surface. 9 Previous studies have suggested these defects in Si have a Burgers vector parallel to the surface and thus have a pure shea r character. Transmission electron microscopy g b analysis using many two-beam c onditions indicated these dislocations in Ge do not display shear nature. Instead the disl ocation appears to be a mixture of 60 and 90 perfect dislocations. The EOR def ects are approximately 1.1 m below the surface and appear to be loops similar to what is seen in Si. 9 These dislocation loops appear to evolve from small loops. No {311} defects are ever observed in the defect evolution process. This suggests that, unlike in Si where it has been shown that {311} defects unfault into dislocation loops, in Ge the loops appear to skip this step and evolve directly fr om submicroscopic clusters to dislocation loops. 6 If {311} defects do form in the EOR region and unfault into loops, the defects must do this before becoming visible in WBDF imaging conditions. Regrowth-related damage is still apparent after 10 min at 550 C, as shown in Figure 53(b), though a reduction in EOR dama ge density occurred from the 450 C anneal, indicating that the EOR defects are relatively unstable in Ge. After the 650 C anneal, no EOR and virtually no clamshell defects remain, as displayed in Figur e 5-3(c). Only the regrowth-related defects are still visible in the XTEM image. Therefore, it ca n be concluded that regr owth-related defects are the most stable in high-energy implants into Ge. Very few regrowth-related defects remain after anneals at 750 C as presented in Fi gure 5-3(d), and at 850 C (not presented) no defects are detected (defect density < 108 cm-2). Low Energy Implant 40 keV The low energy implant conditions formed an amorphous layer that extends from the surface to approximately 70 nm belo w the surface, as shown in Figure 5-4(a). Similar to the high energy implant, the amorphous-cry stalline interface is rough. Annealing for 10 min at 350 C

PAGE 107

107 initiated SPER, with approximately 15 20 nm of regrowth, shown in Figur e 5-4(b). This results in a regrowth rate of approximately 1.5 -2.0 nm /min. This is significantly higher than Zellamas reported rate of 0.2 nm/min51 at 350 C and but only slightly highe r than Csepregis reported SPER rate of 1.0 nm/min132 in (001) orientation at 350 C. Additionally, the interface has planarized during regrowth. Compared to the rough as-implanted interface, which has a range of approximately 10 nm, as shown in Figure 5-5(a) the regrown interface ha s planarized to less than 3 nm in the interface regi on, as shown in Figure 5-5(b). Annealing for 10 min at 380 C, shown in Figure 5-4 (c), fully recrystallizes the amorphous region. The minimum regrowth rate po ssible for this time and temperature is 70 /min, which is still higher than Zellamas 1979 reported 13.2 /min. Ho wever, it is possible that the regrowth rate is higher than the 70 / min, as full recrystallization was achieved and no final interface exists. The HRTEM image also s hows a Ge oxide (GeOx) layer has formed on the surface during annealing, as noted on Figure 5-4(c) GeOx is known to vary in thickness, due to its formation being highly dependent on humidity. Though the samples were annealed in a flowing high-purity N2 atmosphere, trace amounts of water vapor and oxygen are known to exist in the system, and may have cont ributed to the GeOx formation. The resulting regrown crystalline layer is defe ct-free. No EOR defects or regrowth related defects, as seen in the MeV implant, are seen in HR imaging, shown in Figure 5-4. Subsequent 10 min anneals, at 450, 550, 650 and 750 C, shown in Figure 5-6 (a) (d) respectively, show no defects in the regrown re gion, though all show the GeOx layer. The 450 650 C oxide layers appear to be a consistent th ickness, which indicates a limiting mechanism in the diffusion of oxygen from the ambient to the bulk Ge. After the 750 C anneal for 10 min, the surface of the Ge wafer, which originally was highly reflectiv e, became visibly clouded, possibly indicating the

PAGE 108

108 presence of a significant oxide layer. GeOx is known to be water soluble, and the presence of a surface oxide layer on the 750 C sample was confirmed when a water-based solvent was applied to the surface of the sample, removing the dull surface layer and re vealing reflective Ge underneath. Transmission electron microscopy of th is sample confirms that the sample was preferentially etched, as indicate d by etch pits visible in Fig. 5-6(d). The pit wall normals are at an approximate 54 angle to the wafer normal, indicati ng the pit normals are {111} planes. surface, indicating etching along the (111) plane in the <110> direction. Pyramidal etch pits are common in diamond-cubic systems. The etch pits were created during annealing, because the carbon-coating prior sample preparation filled in the pit, as evidenced in Figure 5-6(d). TEM of Ge, Si+ implanted at 40 keV, revealed no defects. However, Si wafers implanted similarly to the Ge, displayed the characteris tic {311} extended defect formation and loop evolution during annealing. Figures 5-7 (a) and (b) show TEM images of Si+ implanted Si annealed at 750 C for 10 min, showing defects in both cros s-section [Fig. 5-7(a)] and plan view, respectively. The defects are concentrated in the EOR regi on, approximately 60-100nm below the surface. Discussion Defects The type and stability of defects found in Ge differ from what has be en previously reported in Si. The type of defects in the high ener gy case mimic Si defects resulting from the recrystallization of a buried amorphous layer, but the stability of these defects is quite unlike Si. Additionally, few EOR defects were observed during the Ge regrow th process in the high energy implant. Finally, none were seen in the lower energy implant. In the 1 MeV implant, the existing Ge EOR defects were the first defects to dissolve between 450 and 550 C. The clamshell defects ann ealed out between 550 and 650 C. The

PAGE 109

109 regrowth-related defects, also known as hairpin di slocations, were clearly visible after 10 min at 450 C and did not dissolve below 750 C, while the defects lack the morphology of the hairpin dislocation seen in Si. 137 The line direction of the regrowth de fects in Si is observed to be very similar to what is observed in these samples. In Ge, the line direction appears to move but this may be related to the increased density of disl ocation in the upper regrowth layer. A g b analysis of the Ge regrowth defects conf irmed that the Burgers vectors are perfect with 110 2 a b Jones et al reported that in Si, hairpin dislocations were less stable than clamshell defects, which were in tu rn less stable than EOR defects. 9 In Si, EOR defects are the most stable defect, while we find that in Ge, EOR defects are the least stable of the three types of defects. It has been shown in Si that re ducing the threshold damage density (TDD) for amorphization either by increasing the implant temp erature or increasing the dose rate results in fewer EOR defects.136 The decrease in EOR defect density an d stability in Ge, as compared to Si, may be a result of fewer interstitials in the EOR available to form extended defects. As previously noted, the TDD in Ge is less than in Si, making it easier to amorphize Ge than Si using the same implant. Therefore, for the same damage profile, more of the total damage profile is within the amorphous range for Ge, leaving less interstitials in the EOR region available for defect formation. No defects were seen in the low energy as-imp lanted or annealed TEM images. This agrees with previous low energy amorphization studies of a 6 keV amorphization investigation by Satta et al. in 2005.54 The absence of EOR defects may be th e result of (1) lower energy implants leaving less interstitials past the a-c interface to participate in ex tended defect formation, (2) the bond-strength differences between Si and Ge making EOR defects less stable, and/or (3) the defects are below the resolution of TEM. The firs t explanation, lower energy implants leave less

PAGE 110

110 interstitials in the EOR region to participate in defect formation, is a phenomenon seen in Si. As the energy of the implant is lowered, less EOR defects are observed. 36 Extremely low energy implants can avoid EOR defects entirely. The second explanation re garding bond strength was postulated by Crosby et al. to explain the de crease in {311} formation with increasing Ge content in SiGe alloys. 13 The third explanation, that any extended defects which may occur are below the resolution of the TEM, is unlikely as the a-c interface region was thoroughly examined using HRTEM imaging in the search for EOR defects. Trapped Interstitials a nd Activation Energy of 1 MeV EOR Defects in Ge Using data from the XTEM images, an activation energy for the dissolution rate of the EOR defects was determined. It should be not ed that there are many assumptions in this calculation as well as an extremely small sample size. The initial assumption made in order to perform the calculation is that the EOR defects seen in Ge are faulted Frank loops that are circular in nature and reside on the (111) plane. The defect s appear to behave like the Si EOR loops, in the fact that the defects decrease in de nsity, as shown in Figur e 5-8(a), but increase in size, as shown in Figure 5-8(b) with each higher temperature annealing step, a phenomenon known as Otswald Ripening. Using this data, and analogous calculations as shown in the previous chapter, the activation energy (Ea) is determined to be approximately 0.4 0.1 eV, as determined using Figure 5-8(c). However, it shoul d be noted that this is an extremely low number of data points, and furt her study should be conducted to re fine the result. The number of trapped interstitials in the defects appears to stay almost constant, at approximately 1 14 cm-2, as shown in Figure 5-8(d). This is also consistent with Otswald ripening, where the large defects grow at the expense of small defects.

PAGE 111

111 Threshold Damage Density for Ge An approximate 2% decrease in density acco mpanies amorphization in Si, Ge and SiGe alloys in this dose regime. 56,58,61,66 Using XTEM, the thickness of the amorphous layer (ta) can be determined, giving a thickne sses of approximately 1050 nm and 73nm for the 1 MeV and 40 keV Si+ implantation, as shown in Figures 7-8 (a ) and 7-9(a), respectively. Performing a mass balance calculation using the de nsities of both the amorphous ( a) and crystalline ( c) layers, and the known amorphous layer thickness (ta), the thickness of the crystalline layer (tc) that was converted to amorphous material can be obtained, using a ta = c tc (5-1) tc = ( a/ c) ta = .98 ta (5-2) This simple calculation results in a convert ed crystalline layer (c-layer) thickness of approximately 1030 nm and 72 nm for 1 MeV and 40 keV Si+ implantation, respectively. At first this does not seem like a significant change. However, the significance of the reduction is important to determine threshold damage densities By comparing the dept hs of the converted clayer, and using simulated damage profiles, the damage necessary to amorphize Ge can be determined. Figures 5-9(b) and 5-10(b) both sh ow the simulated vacancy profile determined superimposed on the XTEM image. 21 The damage matches well with the amorphous layer depth. Then, the calculated c-layer thickness is used to determine the damage thresholds, as shown in Fig. 5-9(c) and 5-10(c), plotting TDD and critical damage (DC) in Ge versus depth for 1 MeV and 40 keV Si+ implantation. It is important to use the converted c-layer thickness, rather than the amorphous layer thickness, to determine the correct value, as simulations do not take into account amorphization or the swelling cause d by it. The literature ha s reported the damage necessary to amorphize a semiconductor by TDD (f avored by the semicondutor community), DC

PAGE 112

112 (favored by the diamond community) and displace ments per atom (DPA) (favored by the ion implantation community). All three can be derive d from simulations, and all three are reported here. Additionally, the values for amorphization from 40 keV Si+ implantation into Si are also provided. It should be noted that no 1 MeV amor phizing implant into Si was performed in this study. Table 5-1 summarizes the results. The DC necessary to amorphize Si is analogous to displacing every atom off its lattice site. 20 That agrees with the determination here that the DC for Si is approximately 5 1022 vacancies/cm3. The DC determined for Ge is roughly 4 2 1022 vacancies/cm3, which is approximately the atomic density of Ge, 4.42 1022 cm-3. In fact, all three Ge values are slightly lower than those calculated for Si. These re sults indicate that Ge is easier to amorphize than Si. This is logical, as Ge is known to have weaker bond strength than Si.114 Conclusion Transmission electron microscopy was used to characterize the exte nded defects arising from annealing Si+ implantation induced amorphized Ge. Although {311} defects have been previously reported in Ge afte r light ion implantation and el ectron-irradiation no {311} defects were observed over a wide range of anneali ng temperatures. In th is study, end-of-range dislocation loops were observed, possibly due to the high implant energy, as these defects have not been seen in previous lower-energy studies In addition, regrowth -related defects and clamshell defects were also observed, with regrowth-related defects exhibiting the greatest stability. Amorphization thresholds were determined for Ge, confirming that Ge is easier to amorphize than Si. The threshold defect density of Ge determined from these experiments is approximately 6 3 1020 keV/cm3. The defects shown during solid phase epitaxial regrowth in Ge agree well with what is seen in Si, which is different than the previ ous chapter. It would seem that the hypothesis

PAGE 113

113 presented previously, that the same defects would a ppear in Ge that do in Si is correct. However, there is some difference in behavior. In Si, th e end-of-range dislocation loops are very stable; while in Ge, the end-of-range defects are the leas t stable defects, annealing out at 650 C for 10 min. In Si, {311} defects are seen in the end-of-ra nge, while these defects have not been seen in Ge. Future work should focus on computation simu lation to see if bond strength really is the deciding factor in the stab ilization of the {311} defect. Thus far, the defects arising after ion-impl antation in Ge have been compared to the defects in Si. For amorphizing implants, simila r type defects occurred in both semiconductors, though the stability of these defects differ. In non-amorphizing implants, Si+ implantation into Ge results in dot-like defects as compared to the {311}defects in Si. The next experiment will focus on non-amorphizing implants into another potential alternate channel material: single crystal diamond.

PAGE 114

114 Figure 5-1. WBDF-XTEM images taken of Ge Si+ implanted at 1 MeV to a dose of 1 1015 cm2 (a) as-implanted and (b) af ter 10 min of annealing at 350 C.

PAGE 115

115 Figure 5-2. HRTEM images of Ge Si+ implanted at 1 MeV to a dose of 1 1015 cm-2 asimplanted (a) surface crystalline-to-a morphous transition region and (b) EOR amorphous-to-crystalline transition region.

PAGE 116

116 Figure 5-3. WBDF XTEM images of Ge Si+ implanted at 1 MeV to a dose of 1 1015 cm-2 after annealing for 10 min at (a) 450 (b) 550 (c) 650 and (d) 750 C. Types of defects are labeled next to the images.

PAGE 117

117 Figure 5-4. XTEM images of Ge Si+ implanted at 40 keV to a dose of 1 1015 cm-2 at 303 K(a) as-implanted and after anneali ng for 10 min at (b) 350 and (c) 380 C for 10 min.

PAGE 118

118 Figure 5-5. HR-XTEM images taken of Ge Si+ implanted at 40 keV to a dose of 1 1015 cm-2 (a) as implanted, with a rough a-c interf ace approximately 10 nm thick and (b) after annealing at 350 C for 10 min.

PAGE 119

119 Figure 5-6. WBDF-XTEM images taken of Ge Si+ implanted at 40 keV to a dose of 1 1015 cm2 at 303 K and annealed for 10 min at (a) 450 (b) 550 (c) 650 and (d) 750 C.

PAGE 120

120 Figure 5-7. WBDF-XTEM images taken of Si Si+ implanted at 40 keV to a dose of 1 1015 cm-2 at 303 K and annealed for 10 min at 750 C (a) XTEM and (b) PTEM.

PAGE 121

121 Figure 5-8. Graphical representa tions of the temperature depende nce of the Ge EOR (a) defect density, (b) defect size, (c) dissolution rate and (d) trapped interstitials.

PAGE 122

122 Figure 5-9. XTEM images of Ge Si+ implanted at 1 MeV to a dose of 1 1015 cm-2 at 303 K(a) as-implanted and (b) as-implanted with simulated total vacancies. (c) Plot of TDD ( ) and Dc ( ) versus depth as predicted by simulations.

PAGE 123

123 Figure 5-10. XTEM images of Ge Si+ implanted at 40 keV to a dose of 1 1015 cm-2 at 303 K(a) as-implanted and (b) as-implanted with simulated total vacancies. (c) Plot of TDD ( ) and Dc ( ) versus depth as predicted by simulations.

PAGE 124

124 Table 5-1. Simulated damage values from current experiments. Material, Implant Energy, Temperature & Dose Threshold Damage Density (keV/cm3) Critical Damage (vacancies/cm3) DPA (unitless) Ge 1000 keV, 303 K, 1 1015 Si+ cm-2 7.0 2 1020 4.5 1.5 1022 0.8 0.2 Ge 40 keV, 303 K, 1 1015 Si+ cm-2 Bulk Interface Measurement 5.5 1.5 1020 3.5 1 1022 0.9 0.1 Si 40 keV, 303 K, 1 1015 Si+ cm-2 Bulk Interface Measurement 7.5 1.5 1020 5.0 1 1022 1.0 0.2

PAGE 125

125 CHAPTER 6 RESULTS FROM NON-AMORPHIZING MEV IMPLANTS INTO DIAMOND Thus far, the defects created from ion-implanta tion in silicon (Si) have been compared to those in germanium (Ge). However, carbon (C) was also presented in Chapter 1 as a potential base material, though little dir ect transmission electron micros copy (TEM) research has been performed on single-crystal diamond. Additionally, there has been a growing interest in utilizing synthetic diamond for engineering applications since diamond was first synthesized by Bundy.70 The development over the past few years of increased production of both chemical vapor deposition (CVD) grown and high-pressure-hig h-temperature (HPHT) man-made diamonds, along with reported high carrier mobilities 72 and the demonstration of diamond transistors80 have all driven this increasing interest.4 In order to be an effective semiconductor, single-crystal diamond must be precisely doped to increase con ductivity without graphi tizing the diamond. Ion implantation has been used for decades to dope si licon for use in microelectronics. Similarly, ion implantation has been studied for over 30 years in diamond. 82,138 However, with the small amount of TEM analysis, repair of the crystal after ion implan tation and the identification of associated defects has been derived from elect rical measurements and spectrometry analysis techniques. Direct imaging of the tr ansformation has not been captured. This chapter presents the results of non-amorphizing Si+ implantation into single crystal diamond, and the evolution of the damage upon annealing. Unlike Si, no {311} defects are observed. Unlike Ge, no dot-like defects are obser ved. Only a slight c ontrast in the damaged crystal is apparent in the TEM images after im plantation. After annea ling, the contrast is reduced, but no extended defects are observed to form. An implant temperature comparison of a Si+ implantation at 1 MeV with a dose of 115 Si+cm-2 revealed that, upon annealing, implantation at 77 K creates buried discontinuous amorphous regions but implantation at 303 K

PAGE 126

126 anneals to a perfect crystal. This indicates th at this implant is right at the verge of the amorphization threshold. Introduction In Si, evolution of defects related to ion-implantation has been studied by TEM. Transmission electron microcopy would be an excel lent method to further pursue a number of research questions in C, especially defect formation and evolution as the result of ion implantation and annealing, and the phase tran sformation from crystalline C to amorphous C. Transmission electron microscopy samples requi re a thickness between 50 300 nm to be electron transparent. Creating these samples is typically a very time-consuming process because of the properties of diamond.{Yin, 2000 #36} Mech anical polishing and ion milling methods are typically used to create samples from C VD-grown diamond films, and high resolution TEM (HRTEM) has been used to image CVD-grown diamond film defects.120 Due to the difficulty in creating TEM samples from single crystal di amond, few studies have been reported. 96 In the 1960s, TEM samples were prepared by oxidation thinning, by flowing oxygen at 750 C or carbon dioxide at 1350 C over the diamond subs trate, thereby gas etching the crystal.119 However, this process does not protect the su rface and is not selec tive of the area. Other disadvantages of this method are that the sample mu st be boiled in an acid mixture to remove the surface layer of carbon that forms and such high te mperatures may anneal out defects present in the sample.96 In 2001, plan view TEM (PTEM) studies we re performed on ion-implanted natural diamond.104 Yin has reported on the def ects in a lab-grown HPHT di amond crystal studied using TEM as one method. 96,106 Additional TEM studies of HTHP single crystal diamond whiskers 107 and particles 108 have been conducted. However, these sa mple preparation methods were not area selective in nature.

PAGE 127

127 Experimental Design Preliminary Experiments Preliminary experiments on plasma-doped si ngle crystal diamonds were performed to determine appropriate anneali ng conditions to study the elimin ation of damage from diamond after ion implantation. As the literature reports ann eal temperatures from 300 C19 to 1600 C 92, and Si is known to only produce defects in certa in doses and temperature regimes, preliminary experiments were necessary to determine appropriate temperatures. The samples were annealed at 300, 400, 500, 700, and 900 C, with the results from these experiments and recent literature results indicating that temperatures above 900 C are necessary for damage elimination, 92 though some recent literature still contradicts this conclusion. 75 Presented Experiment Gem-quality HPHT single-crystal diamond samples, approximately 5 5 1 mm3, were Si+ implanted according to Table 1. This ion spec ies was chosen to reduce variables between comparing this experiment with the Ge and Si experiments, and to keep from introducing dopant interaction effects due to the el ectrical differences between group elements. It is well known that Group III and IV dopants can affect the formation and dissolution of defects in Si, and this additional effect was not de sired for these experiments.27,32,139 The implantation of Si+ into the diamonds was performed in a tandem accelerator ion implanter, with a dose rate of 0.44 A/cm2 for all experiments. Samples were tilted 7 relative to the incident ions to minimize ion channeling. The original nitrogen (N2) impurity concentration of the diamonds, obvious from the visible light-orange color, is on the order of 1018 1020 cm-3, 140 and was uniform throughout the crystals. According to simulations, 21 the damage should peak (Rd) at approximately 540 nm and the projected range (Rp) of the damage should be approximately 570 nm belo w the surface of the

PAGE 128

128 diamond for 1 MeV Si+ implantation, and 840 nm and 900 nm for 2 MeV Si+ implantation. After implantation, a 1000 layer of silicon dioxide (SiO2) was plasma-enhanced chemical vapor deposition (PECVD) deposited on the (011)-oriented sample surfaces, Si+ implanted at 1 1014 Si/cm2, with the thermal exposure of the diamond be ing 310 C for approximately one minute. A separate 1000 layer of SiO2 was sputter-deposited onto the ( 011)-oriented sample surfaces, Si+ implanted at 1 1015 Si/cm2, with the thermal exposure of the diamond being less than 100 C. The original purpose of the SiO2 layer was to compositionally separate the C of the crystalline diamond from the conductive carbon coating app lied during sample preparation. However, surface etching occurred after initial anneals as C is highly reactive with oxygen (O2) at higher temperatures 141 and this practice of placing an O2 compound next to a reactive C layer was eliminated for further samples. It should be noted that the SiO2 PECVD layer remained on after annealing at 950 C; however, the sputter-deposite d layer peeled off. The PECVD SiO2 layer did remain on the Si+ implanted to a dose of 1 1014 cm-2 sample, and did seem to protect it from surface erosion. This wont work at higher te mperatures, where the Ellingham diagram shows that C-O2 compounds are more stable than Si-O2 compounds, but a nitride layer might be a good alternative. Approximately 150 nm of carbon was evaporated onto all samples, for conductivity and protection of the sample surface during focu sed-ion beam processing. Cross-sectional TEM (XTEM) samples were prepared according to the procedure previously published by the author,117 during which a layer of Platinum was applied to the surface. Both layers, if visible, are noted in all TEM images presented here. Annealing was performed in both tube and gr aphite furnaces. Samples annealed at 1050 C and below were annealed in a Lindberg tube furnace in a high-purity N2 environment. Anneals at 1250 and 1350 C used a graphite chamber furnace at diffusion pump vacuum pressures ( 10-4

PAGE 129

129 Torr) 142 with a residual He environment. Additionally the samples were placed in tantalum (Ta) foil (folded as envelopes), which is an O2 getterer, to additionally pr event accidental oxidation of the diamond surface. The Ta envelopes served th e additional purpose of separating and marking each sample separately. Transmission electron mi croscopy analysis was performed on JEOL 200CX and JEOL 2010 microscopes. Results As-implanted Comparison: Orientation, Dose, Temperature Orientation of the diamond crys tal, whether (011) or (001), did not affect the implant damage severity or range, as detectable using these TEM methods. Changing the dose did significantly change the damage density in the cr ystal. As expected, with increasing dose, there was increasing damage, as already shown between the 1 1014 and 1 1015 cm-2 dose implants, shown in the SRIM section of Chapter 3, in Fi gures 3-14(b) and 3-15(b) respectively. Higher Si+ doses of 3 1015 Si+/cm2 and 7 1015 cm-2 will be shown in Chapter 7. The temperatures chosen for this study, 77 K and 303 K, are in the range that freezes in point defects in diamond. 85 However, it is known that even a room temperature implant can, due to dynamic annealing, achieve significantly differen t results than at liquid nitrogen temperature. As-implanted, the difference between the sample Si+ implanted at 1 MeV to a dose of 1 1015 cm-2 at 77 K and 303 K is shown in WBDF-TEM imaging using a g220 two-beam condition, as presented in Figure 6-1. The range of damage for the 77 K implant is larger than that of the 303 K implant, as evidenced by Figur es 6-1(a) and (b), respectivel y. Additionally, the 77 K implant seems to have a dark band of damage in the midst of the larger damage band. Upon initial imaging, this was thought to be a conti nuous amorphous layer of material, but upon highresolution imaging, was shown to be severely damaged diamond enclosed in a diamond lattice.

PAGE 130

130 Larger differences between these two temperatures appear during the annealing process, as discussed later. Annealing: Low Dose As discussed, preliminary annealing experime nts were used to determine that even low damage in diamond shows little change when ann ealed at temperatures below 950 C, which is 20% of the triple point of diamond-graphite-liqui d, or 56% of the stable diamond temperature at vacuum. 111 (011)-oriented diamond, Si+ implanted at 1 MeV to a dose of 1 1014 cm-2 at 303 K, is shown using BF-XTEM imaging in Figure 6-2 (a ). After annealing at 950 C for 10 min, the damaged layer, although slight, is still visible in Fi gure 6-2 (b). The damage band becomes less noticeable after 40 min as shown in Figure 6-2 (c), but at 120 min of annealing, presented in Figure 6-2, unusual defects extending from the surf ace, begin to appear. Some defects seem to be oriented at approximately 37 to the surf ace, though the exact crystal morphology is unclear. As the defects are intermittent, and begin at the surface, it is not believed that these defects are associated with the implant. The defects will be briefly discussed in Appendix A. Another diamond sample, (001)-oriented and similarly implan ted, annealed at 1250 C for 1 hr is shown using WBDF-XTEM imaging in Figure 6-3 and s hows that damage is still present at this temperature. Annealing: High Dose In the literature, many references were made to the ability of diamond to regain its crystalline form (if the damage threshold was not exceeded beyond amorphization) around 1000 C. 75 Therefore, another annealing sequence was perf ormed just below this threshold to try to capture extended defect formation and e volution. Diamond, (011)-oriented and Si+ implanted at 1 MeV to a dose of 1 1015 cm-2 (an order of magnitude higher th an the previous implant, with RD = 540nm and RP = 570nm) at 303 K, is shown in Figure 6-4 both after implantation, shown in

PAGE 131

131 Figure 6-4(a), and an annealing sequence. No overa ll repair of the lattice has occurred and the damage band is still quite clea r after 10, 40, and120 min of annea ling, shown in Figures 6-4 (b) (d) respectively. However, significant surface etch ing occurred during the annealing procedure, as shown in Figure 6-4 (d). Further annealing times would probably completely eliminate the implant damage. Thus, due to the surface etching by residual O2 in the furnace during annealing, higher-temperature anneals were performed in a graphite chamber under vacuum. Annealing: Implant Temperature Effect As noted in an earlier section, there were slight differences in the damage created after the same implant at different implant temperatures. However, after annealing at 1350 C for 24 hrs, significant differences appear. The specimen implan ted at 303 K, shown in Figure 6-5(a), almost completely recrystallized to undama ged diamond, shown in Figure 6-5(b). In stark contrast, implantati on at 77 K leaves pockets of amorphous material near RP after annealing, as shown in Figures 6-6(a) and (b ). These pockets, in WBDF imaging conditions resembled extended defects as presented in Figure 6-6(c). However, tilting to 50 away from the [110] zone axis and performing BF imaging reveal ed a seemingly ordered and consistently thick layer of defects. However, upon closer review in HRTEM imaging, these were determined to be pockets of amorphous material in a crystalline matrix. As shown in Figure 6-7, HRTEM images taken from the middle of the damaged layer revealed pockets of amorphous material. Although no extended defects were observed in the damage range after the 1350 C anneal of samples implanted at both temperatures, unusu al and intermittent defects extending from the surface occurred in the sample implanted at 77 K. As in the case of the lower dose (1 1014 cm2) Si+ implant, presented in Figure 4-5(d), these defect s are not believed to be associated with the implant because of their intermittent nature and apparent surface orientation. Hence, since these

PAGE 132

132 are defects, though not associated with implantati on (the defects were also seen when annealing unimplanted diamonds), these defects will be discussed in Appendix A. Discussion As-implanted comparison. As implanted, there is no sign ificant difference between the implant into a (011)-oriented or (111)-oriented diamond, which is not different that what is expected from Si. However, there is consider able difference at higher doses. This is not unexpected, but necessary to comment on because dose will have a large effect upon annealing. Importantly, for the same energy, dose rate, i on species, and implant temperature, Si and Ge amorphized, but diamond did not. In previous ch apters, Si+ implantation to a dose of 1 1015 cm-2 dose was discussed in the Ge amorphization chapter. For diamond, this dose did not amorphize the diamond upon implant, in implantation at 303 K or 77 K. However, there is some difference seen on implant between the implant te mperatures. Both created significant projected range damage, but no immediate defects are seen nor is a continuous amorphous layer created upon implant. This shows a major difference between Ge, Si and Diamond since small dot-like defects were seen immediately upon non-amorphizi ng implantation in Ge. No defects are seen upon implant in either Si or Diamond for non-amor phizing implants. This may be related to the mass of the elements (since the lower mass elemen ts show no defects) or the bond energy, since Ge has the lowest bond energy. 111 Also, the lack of as-implanted defects may be related to the ion-implantation process, since Si+ implantation into diamond introdu ces a more massive particle than the host material, indicating nuclear stopping will begin more rapidly. Si ions into Si are approximately the same mass, and Si ions into Ge are much less massive, and nuclear stopping may not occur for some distance into the crystal. 20 Annealing: low dose. The damage created by the lower dose Si+ implantation, to a dose of 1 1014 cm-2, seems to be completely eliminated afte r a 950 C, 4 hour anneal. However, due to

PAGE 133

133 the presence of the surface ini tiated defects, this conclusion may be invalid. The defects may have interacted with the damaged diamond to a ssist in the diffusion of point defects from the damaged area or affected the imaging techni que during TEM, making it difficult to view the contrast from damaged material. WBDF TEM images of specimens after high temperature anneal at 1250 C shows that damage still exists in the crystal. Even for this respectively low dose, higher annealing temperatures (or longer times) are necessary for full repair of the crystalline damage imparted by implantation at 1 MeV. No extended defects were observed at any stage of the annealing process, which is unlike what was seen for non-amorphizing experiments in Ge and Si. Annealing: high dose. For the higher dose in this study, it is shown that annealing at 1050 C, even for times up to 2 hours, is not significant to repair the lattice. Annealing at 1350 C for one hour still reveals residual damage for an impl ant temperature of 303 K. This is in agreement with recent electrical studies that recommend anneal temperatur es around 1500 C to completely repair the crystalline lattice. 92 It should be noted that the referenced study was with a B+ implantation at 2 MeV and the TEM imaging results from an identical implant will be shown in Appendix B. The results here, and of Vogel, 92 contradict other studies that recommend temperatures of 600 1000 C 88 74 are sufficient for lattice repa ir. Again, no extended defects were observed for the higher dose, non-amorphizing implant. Annealing: implant temperature effect. At the verge of amor phization, the implant temperature has a profound affect upon annealing. A Si+ implantation at 303 K at a dose of 1 1015 cm-2, followed by annealing at 1350 C shows no ex tended defects forming, and damage is significantly reduced in the crystal from the as-imp lanted state. However, for implantation at 77 K, the elimination of room-temperature dyna mic annealing allowed enough extra damage to

PAGE 134

134 form within the crystal, which upon annealing at 1350 C, local collapsing of the lattice into amorphous pockets occurred. No extended defects we re observed either in the damage range of the specimens implanted at 303 K implant or around the amorphous pockets of material in specimens implanted at 77 K. The lack of extended defects ag rees with molecular dynamics simulations that predict that the stable defect in diamond is the <110> split inte rstitial, also referre d to as a dumbbell structure, which is a point de fect, not an extended defect. 143 However, this structure is not observable in TEM since it is a point defect. The most likely reason that extended defects are not forming in the diamond lattice is the strong c ovalent bond between the carbon atoms may make other configurations thermodynami cally unstable. The diffusion of self-interstitials in carbon may also be faster than in Si or Ge, which may contribute to the lack of extended defect formation. Further computational analysis is n ecessary to shed definitive light onto the reasoning behind the lack of extended defects in this implantation regime. Conclusion This chapter covered ion implantation experime nts into single crys tal diamonds that did not exceed the crystalline damage level nece ssary to amorphize the crystal. From these experiments, the following conclusions are dr awn. Crystal orientation makes no noticeable difference on ion-implantation damage depth or amount of damage to the diamond. Implantation into (001)-oriented diamonds vers us (011) oriented diamonds yields similar implant results, within the limitations of TEM imaging. The ve rge of diamond amorphization was captured at the 1 MeV, 1 1015 Si+/cm2 implant. Upon implantation, the liqui d nitrogen temperature (77 K) and a room temperature implant (303 K), with al l other parameters being equal, displayed approximately the same depth and amount of dama ge. However, the 77K seemed to have a band of defects in WBDF, but upon cl oser inspection using HRTEM, revealed damaged diamond, and

PAGE 135

135 not an amorphous layer. Upon annealing, the 77 K implant trapped amorphous pockets, while the 303 K implant almost fully recovered the crystal st ructure. To fully anneal the damage from the crystal, 1 hour at 1250 C appears almost sufficient for a 1 1014 Si/cm2 1 MeV, 303 K implant. However, some damage (though slight) still is apparent for the 1 1015 Si/cm2 implant after 24 hrs at 1350 C. Higher annealing temperatures or longer times are necessary to completely anneal out the implant damage. No extended def ects were observed to develop in the heavily damaged regions. However, non-uniform defect dist ribution from the surface did appear in select samples, and is discussed in Appendix A. Sin ce no extended defects were seen in the damaged region, this agrees with molecular dynamics simu lations that determined the stable defect in diamond is the <100> split interstitial. The {311} defect does not seem to be the preferred interstitial defect for this di amond-lattice material, as was th eorized. The verge of critical damage density to convert crystalline diamond to a buried amorphous layer is at the conditions of an implant equal to an en ergy, dose, dose rate, and implan t temperature of 1 MeV, 1x1015 Si+/cm2, 0.44 A/cm2, 77 K. In the previous chapters, the de fect behavior of Ge was compared to that of Si. In this chapter, it has been shown that for non-amor phizing implants, diamond does not display any defects related to the implant damage. In Ge non-amorphizing implants at 1 MeV produced dotlike defects, and in Si, {311} defects. Though it seems that diamond fully recrystallizes upon annealing without showing defects, further st udy is necessary. Higher annealing temperatures could be tried, up to 1700 C, to s ee if defects develop. However, it is unlikely that defects will be visible after a higher annealing temperature as past experience in Si shows that a higher annealing temperature would only speed defect dissolution, not formation. Annealing diamonds above 1700 C destabilizes the crystal, even und er the best vacuum, and the C converts to a

PAGE 136

136 graphitic form.82, 85 This may be the explanation as to w hy defects are not seen in this study of diamonds. Defects appear in Si at approximately 750 C and in Ge at approximately 450 C and both temperatures are about 50% of the resp ective melting points, while diamond has no melting point. There is a triple point which occurs at approximately 4492 C. Thus, 50% of 4492 C is slightly above 2000 C, which is above the dest abilization temperature of diamond. However, even if the diamond was stable, it is unlikely th at {311} defects would form because longer time anneals were used in this study. In Si, defect formation and evolution are related to the time and temperature of the anneal. A {311} defect appears in Si after 5 min at 800 C, 33 min at 750 C, or 47 hours and 14 min at 650 C. 27 Therefore, since almost all th e damage has annealed out of diamond for 1350 C 24 hours, without showing extende d defects, it is unlikely that longer times or higher temperatures would reveal extended defects. Thus far, the three Group IV diamond-cubic semiconductors have not created the same defects after identical anneals. For sub-amor phizing implants, C has not shown any extended defects. Si showed none upon implant, but {311} defects were apparent during the annealing process, and were stable to about 65% of the melting point. Ge showed dot-like defects immediately after implant, some of which were stable up to 70% of its melting point. The last condition that will be studied is amorphizing implants into diamond, and that will be discussed in the next chapter.

PAGE 137

137 Table 6-1. Implantation parameters Parameter Varied from base implant Orientation Implant Temperature, Ti (K / C) Energy (keV) Fluence & Ion, ions/cm2 Base Implant <100> 303 / 30 1000 1 1015 Si Orientation <110> 303 / 30 1000 1 1015 Si Dose <100> 303 / 30 1000 1 1014 Si Dose & Orientation <110> 303 / 30 1000 1 1014 Si Temperature <100> 77 / -196 1000 1 1015 Si Temperature & Energy<100> 77 / -196 2000 1 1015 Si

PAGE 138

138 Figure 6-1. WBDF-XTEM images of (001)-oriented single-crystal diamond Si+ implanted at 1 MeV to a dose of 1 1015 cm-2 at (a) 77K and (b) 303K.

PAGE 139

139 Figure 6-2. BF TEM images of (011)-oriented diamond Si+ implanted at 1 MeV to a dose of 1 1014 cm-2 Si+ at 303K (a) as implanted, and annealed at 950 C for (b) 10 min, (c) 40 min and (d) 120 min.

PAGE 140

140 Figure 6-3. WBDF-XTEM image of (001)-oriented diamond Si+ implanted at 1 MeV to a dose of 1 1014 cm-2 at 303K after annealing at 1250 C for 1 hr.

PAGE 141

141 Figure 6-4. BF TEM image of (110)-oriented diamond Si+ implanted at 1 MeV to a dose of 1 1015 cm-2 at 303K (a) as implanted, and after an nealing at 950 C for (b) 10 min, (c) 40 min and (d) 120 min.

PAGE 142

142 Figure 6-5. WBDF-XTEM images of (001)-oriented diamond Si+ implanted at 1 MeV to a dose of 1 1015 cm-2 at 303K (a) as implanted, (b) an d after annealing in vacuum at 1350 C for 24 hrs.

PAGE 143

143 Figure 6-6. WBDF-XTEM images of (001)-oriented diamond Si+ implanted at 1 MeV to a dose of 1 1015 cm-2 at 77K (a) as implanted and (b) annealed in vacuum at 1350 C for 24 hrs. Detailed views of damage region after annealing imag ed using (c) WBDFXTEM and (d) BF-XTEM (approximately 50 tilt).

PAGE 144

144 Figure 6-7. HRTEM image of damaged di amond layer in (001)-oriented diamond Si+ implanted at 1 MeV to a dose of 1 1015 cm-2 after annealing at 1350 C for 24 hrs.

PAGE 145

145 CHAPTER 7 RESULTS FROM AMORPHIZING ME V IMPLANTS INTO DIAMOND Introduction The previous chapter described the results of experiments involving implants into diamond that were below the threshold to amorphize th e carbon (C), and compared the results to nonamorphizing implants into silicon (Si) and germanium (Ge). This chapter will focus on amorphizing implants into diamond. There is much interest in the amorphization phenomena. For Si, pre-amorphization is used before dopant implants to contain a nd activate the dopant during solid-phase epitaxi al regrowth (SPER),20 and the regrowth process reveals the stable yet detrimental end-of-range extended defects in Si.9 Additionally, amorphiza tion in diamond is of interest for creating micro-electrical-mechani cal systems (MEMS) devices. Dresselhaus and Kalish noted that there are ma ny outstanding questions about th e damage from ion-implantation that could be solved by cross-sectional transm ission electron microscope (XTEM) analysis, and one of the outstanding questions was the struct ure and threshold of ion-amorphized diamond. However, amorphiziation is a challenging pheno menon to model, because it is influenced by implant temperature, energy, dose, dose ra te and ion species. Implanting diamond adds another complication, since it is not the stable form of elemental C at room temperature and atmospheric pressure. Upon exceeding the amorphi zation threshold damage, and annealing at sufficient temperatures, diamond w ill convert to graphite. This tr ansformation has been studied using many techniques, 19,85 but has not been studied using TEM. For the same reason, no extended defects have been seen in diamond using TEM analysis. It has been theorized that the {311} defect is the stable interstitial defect in all diamond-cubic structures 7, though molecular dynamics simulations using diamond predict that th e split interstitial is the stable defect. 144 Additionally, upon amorphization of the diamond crysta l, swelling has been previously observed.

PAGE 146

146 This experiment uses XTEM analysis to document the amorphization and conversion to graphite transition of ion-implanted diamond that has exceeded the threshold for amorphization. The threshold for critical damage is estimated, and compared to previously published results. The density of the amorphous region immediatel y upon implant, and the density decrease during annealing, is calculated. Finally, data are prov ided that diamond does not graphitize immediately upon implantation, but instead becomes amorphous a nd transforms to graphite during annealing. Experimental Design Gem-quality high-pressure, high-temperature single-crystal diamonds, with dimensions of approximately 5 5 1 mm3, were Si+ implanted in a tandem accelerator implanter at a 7 relative to the incident ions to avoid ion cha nneling. (001)-oriented diamond was Si+ implanted at 2 MeV to a dose of 1 1015 cm-2 at 77K to create a buried ma rker layer with point defects frozen-in to avoid migration of th e defects to the region of the next implant. This sample was cut into 4 equal pieces, and one additionally Si + implanted at 1 MeV to a dose of 7 1015 cm-2 at 303K. For a separate experiment, virgin diamond wa s Si+ implanted with 1 MeV to a dose of 3 1015 cm-2 at 303K. The dose rate for all experiments was 0.44 A/cm2. Annealing was performed in a graphite chamber furnace at diffusion pump vacuum pressures ( 10-4 Torr) 142 with a residual helium environment. Cross-sectional transmission electron microscope samples were prepared using the method previously published by the author,117 and outlined in Chapter Three. In that process, layers of carbon and pl atinum are deposited onto the diamond surface, and (when applicable) are noted on images. Tran smission electron micr oscopy analysis was performed on JEOL 200CX and a JEOL 2010 mi croscopes. Electron energy loss spectroscopy (EELS) analysis was performed on a FEI T echnai TEM. Raman spectroscopy using a He-Ne Laser at 632.81 nm with a spotsize of approximately 5 m was performed.

PAGE 147

147 Results TRIM Results The as-implanted profiles of the samples Si+ implanted to doses of 3 and 7 1015 cm-2 implants are shown in Fig. 7-1(a), and 7-2 (a) re spectively. Sections of these samples, as noted on the figures, did undergo the first-order phase transformation to an amorphous structure. As reported in Chapter 3, the TRIM simulated damage profile matched very well with the experimental TEM evidence of damage. Also, TR IM has been used to predict the threshold damage density (TDD) for the amorphization of Si. 35 However, when the surface of the asimplanted profiles for these amorphized samples is lined up with the TRIM damage profile, the profile does not match well with the 3 1015 Si+ /cm2 implant, shown in Figure 7-1(b), and does not match at all to the 7 1015 Si+ /cm2 implant, shown in Figure 7-2(b). Moreover, if the threshold damage density is attempted to be ca lculated using this match, the numbers do not make any physical sense since the lower threshold of TRIM-simulated damage is predicted to be in the midst of the amorphous region. Three plausible explanations were considered. First, the TRIM simulation paraments were incorrect. Second, energy released during the phase change from the diamond crystalline to amorphous carbon phase amorphized surrounding materi al. Third, significan t swelling occurred which is associated with the phase transition. The first explanation wa s thoroughly explored. The ion species (Si+), ion energy (1000 keV) and target spec ies are all set by the experiment. The density of the target was change d from graphite (which is the stable form of carbon at room temperature and pressure) to th at of diamond. The displacement energy was set to 50 eV. This parameter has ranged in the literature between 30-55eV; however, the most recent experiments have identified the displacement energy at 50 eV.101 Simulations were run varying the Ed from 10 eV (which is less than th e 15 eV used for Si and Ge) to 100 eV, with little change in the depth of

PAGE 148

148 the profile. Damage to the profile was changed, but as energy has the most effect on depth, and energy was kept constant, the de pth of the profile did not cha nge significantly (~10-20 nm). Hence, this could not explain the current resu lts. The second explanation of energy release, though plausible, does not seem probable due to the lack of literature evidence to support it. Additionally, it would be difficult to devise an experiment to test the theory. The third explanation, one of swelling of the diamond due to a reduction in the density of the material, seemed a more plausible explanation than the prev ious two put forth. There is sufficient evidence in the literature to support swelling of other semiconduc tors like Si and SiGe,56,58,61,66 but only a 1-2% swelling. There is also substan tial literature on the swelling of SiC 62,66 and diamond, 99,102,145 indicating large percent decreases (approximat ely 12-15% for SiC) in density. To test the swelling theory, a buried marker layer was used. Swelling of Diamond Experiment The buried marker layer used was generated via Si+ implantation at 2 MeV to a dose of 1 1015 cm-2 at 77K. The low temperatur e froze-in point defects,85 avoiding defect migration to the area of the future amorphizing implant. The results from the non-amorphizing Si+ implant are shown in Figure 7-3(a). A WBDF-XTEM image, using g220 imaging conditions, shows the damaged diamond layer approximately 150 200 nm thick and centered on the projected range (RP) at 0.90 m. A BF-XTEM image inset, shown in Figure 7-3(b) shows the damage more clearly, and although the band of damage is quite di stinct, it is not a buried amorphous layer. The non-amorphous nature of the damaged diamond is proven by HR-TEM on-axis [110] lattice imaging, as presented in Figure 7-3(c), which sh ows a diamond lattice with pockets of damaged diamond throughout. In addition, Raman spectroscopy shows the characteristic diamond peak at approximately 1335 cm-1 (FWHM of 5 cm-1 indicating a sharp peak) confirms the undamaged, non-amorphous nature of the sample, as displaye d in Figure 7-4. Howeve r, it should be noted

PAGE 149

149 that the sample volume for this Ra man technique is approximately 500 m into the crystal, while the damaged layer is very small in comparison. Additionally, the crosssection Raman analysis could not be used because the spot size of the laser was as large as the whole sample, including coatings and FIB damage. The next step to confirm that the buried marker layer would not affect the future amorphizing implant is to compare the TRIM damage profiles. Figure 7-5(a) shows the asimplanted marker layer, next to a TRIM-deriv ed simulation of the ion-implantation induced damage profile, shown in Figure Figure 7-5(b) The grey band designate s the area between the damage peak (RD) and RP of the marker layer. The total damage of the 2 MeV implant is an order-of-magnitude below the amorphizing implan t, and centered almost 400 nm deeper than RD of the amorphous implant. Therefore, the marker layer implant would have little to no impact on the next amorphizing implant. Th is is experimentally shown in Figure 7-5(c), in which both implants are visible. The as-implanted marker layer and the amorphizing Si+ implant with a dose of 7 1015 cm-2 are shown in Figures 7-6 (a) and (b). The swelling that occurred during amorphization is marked on Figure 7-6(b), indicating approximate ly 150 nm of swelling. The density of the amorphous layer can now be determined. Density of As-implanted Amorphous () C Layer The amount of crystalline ma terial that underwent amor phization was determined by subtracting the thickness of the swelling at the surface (using the marker layers as guides) from the amorphous thickness [Fig. 7.6]. Note that on-ax is [110] BF images were used for the actual measurement, rather than the WBDF images s hown, to avoid measuring a tilted image and to obtain a more accurate measurement. This analysis determined the thickness of the crystalline region that converted to the amorphous layer, designated tc-a, being 150 nm +/20 nm. A mass

PAGE 150

150 balance of atoms determines the density of the amorphous region, a, using the known density of diamond, diamond, and the thickness of the amorphous layer, ta, as shown here: diamond tc-a = a ta (7-1) (1.754 1023 atoms/cm3) (420 20 nm) = a (570 20 nm) a = [(1.754 1023 atoms/cm3) (420 20 nm)] / (570 20 nm) a = 1.3 0.1 1023 atoms/cm3 For comparison, graphit e= 1.129 1023 atoms/cm3 ( a is ~ 15% more dense than graphite) diamond = 1.754 1023 atoms/cm3 ( a is ~ 25% less dense than diamond) Thus, the density of the as-implanted amorphous carbon ( -C) region seems to be less dense than diamond but slightly more dense th an graphite. In terms of thickness, the asimplanted amorphous layer is 35% (+/10%) th icker than the crysta lline layer that was converted. Annealing Ion-implantation Induced Amorphous C For annealing at high temp eratures, a 1 MeV, 3 1015 Si+ /cm2 at 303K implant into a virgin diamond was used. Fig. 7-7(a) shows a WBDF TEM image, using g220 two-beam imaging conditions, of the as-implanted sample. The am orphous layer of the as-implanted sample is approximately 315 +/25 nm, again taking the m easurement from an untilted, <110> on-axis BF TEM image [Fig. 7-7(d)]. Assuming that the as-i mplanted density is approximately the same for this implant as for the previous implant, the thickness of the layer that converted to amorphous material can be determined: diamond tc-a = a ta (7-2) (1.754 1023 atoms/cm3) tc-a = (1.3 0.1 1023 atoms/cm3) (315 25 nm)

PAGE 151

151 tc-a = [1.3 0.1 1023 atoms/cm3 (315 25 nm)] / (1.754 1023 atoms/cm3) tc-a = 233 35 nm The thickness of crystalline material that converted to graphite (assuming the as-implanted amorphous density is the same as the previous im plant) is approximately 230 nm. This will be an important value to determine the threshold da mage density necessary to amorphize diamond. The Si+ implanted at 1 MeV to a dose of 3 1015 cm-2 implanted sample was then annealed at 1350 C for 1 hour and 24 hours in vacuum. Th e diamond showed additi onal signs of swelling after the 1 hour and 24 hour anneals, as show n in Figures 7-7(b) and (c), respectively. Approximately 70 nm of additional swelling occurred during the heat treatments, after the initial implantation and conversion to -C. Using the same conservation of atoms mass balance used to determine the density of the as-implanted amorphous layer, the density of the a nnealed layer can also be determined. For this measurement, on-axis [110] TEM imaging was used to get a detailed view of each interface, both as implanted and annealed for 24 hrs, as presen ted in Figures 7-6 (d) and (e). The measured thickness of the as-implanted amorphous layer is 325 15 nm, and the measured thickness after annealing is 390 15 nm. Therefore, the dens ity of the annealed amorphous layer can be determined, assuming that the as-implanted density is approximately the same as for the previous implant: as implnated ta as implanted = aannealed ta-annealed (7-3) aannealed = ( as implnated ta as implanted) / ta-annealed aannealed = [(1.3 0.1 1023 atoms/cm3) (325 15)] / (390 15 nm) aannealed = 1.1 0.2 1023 atoms/cm3 For comparison,

PAGE 152

152 graphite = 1.129 1023 atoms/cm3 Using the density calculation, the amorphous layer seems to have achieved approximately the same density as graphite. To determine the st ructural nature of the annealed amorphous layer, diffraction patterns of the samples were obtained. Determination of the Graphiti c Nature of Annealed Diamond Using the fast Fourier transform (FFT) featur e of the Gatan TEM software package, FFT images (which simulate diffraction patterns) were obtained from the on-axis [110] BF images of the diamond sample Si+ implanted at 1 MeV to a dose of 3 1015 cm-2 at room temperature, as shown in Figure 7-8 (a) with the associated FFT image in Figure 7-8(b), and then annealed for one and 24 hours at 1350 C, shown in Figures 7-8 (c) and (e) respectively, with the associated FFTs in Figures 7-8 (d) and (f). After the 24 hr, 1350 C anneal, the HRTEM image shows ordering of the amorphous layer, which is reflected in the FFT as defined bands with a preferred orientation appearing inside the [111] plan e diffracted spots. Us ing the relationship, rd = l, (7-4) where r is the radius from the center transmitted spot to the area of interest, d being the spacing between planes, being the wavelength of the electron used at the part icular accelerating energy, and l being the camera length in the microscope the distance between planes represented by the preferred orientation parentheses-shaped spots can be determined, and compared to graphite. Distances between the diffraction spots were measured, and using the above relationship, knowing that the product of and l is the same for both the diamond and the graphite, the d spacing of the preferential non-diamond plane can be determined (ddiamond(111) = 2.059): rdia (111) ddia(111) = rgraphite d(graphite) (7-5)

PAGE 153

153 1.8 2.059 = 1.1 d(graphite) d(graphite) = ~3.37 This corresponds well with the reported values of the most intensely diffracted spots in graphite, as reported by JCPDS International Ce ntre for Diffraction Data 1996, which reports the most intense spots for graphite to be the (002) planar spacing of 3.36 123 and 3.38 124 for different reported literature values. For th e hexagonal crystal fou r-index scheme, this corresponds to an (0002) graphite plane, since i = -(h+k). EELS Results of the Amorphous Region An EELS analysis was conducted on the buried amorphous layer resulting from Si+ implantation at 1 MeV to a dose of 3 1015 cm-2 implanted at 303K. Two EELS spectrums are shown in Fig. 7-9 (a-f): the first shows the as-imp lanted spectrum in the near surface region [Fig. 7-9 (a)], the mid-amorphous region [Fig. 7-9( b)], and the bulk crystalline region [Fig. 7-9 (c)], while the second set show the same regions [Fig. 79 (d),(e),and (f), respectively] for the sample annealed at 1350 C for 1 hour. Reference spectrum for diamond, amorphous carbon and graphite can be found in Chapte r 3 Analytical Techniques. As implanted, the near-surface region and bulk region both show the characteristic threepeak diamond configuration, as shown on Figures 79 (a) and (c). Similarly, the same three-peak diamond configuration occurs for the near surface and bulk regi ons of the annealed sample, shown in Figures 7-9 (d) and (f). The transf ormed region in between, for the as-implanted sample, lost the three distinct peaks and instead only has a broad peak. This is indicative of amorphous carbon. Upon annealing, th e broad peak is preceded by a short peak at approximately half the intensity, quite similar to the graphite standard. This is further indication that the diamond is amorphous upon implantation, and b ecomes graphite during annealing. Raman results (not shown) additionally agree with th is conclusion; broad p eaks indicating amorphous

PAGE 154

154 carbon were seen in the as-implanted sample, with defined peaks indicating a crystalline nature seen after annealing. Limitati ons of the spot size of the Ra man technique did not allow for individual cross-secti onal location analysis. Annealed Amorphous-Cry stalline Interfaces The crystalline-amorphous interfaces of the two amorphized samples both have a larger transition region at the surface oriented inte rface than the bulk-oriented interface. Upon annealing for 1350 C for 24 hrs, the bulk interfac e becomes more defined, as shown in Figure 77(e). The surface-oriented transition region, while also becoming more defined, has interesting morphological features. Figure 7-10(a) shows th e on-axis [110] TEM image of the annealed diamond. The surface interface region, when viewed at higher magnification as shown in Figure 7-10(b), reveals a very rough in terface, while the same magnifi cation of the bulk interface is relatively abrupt, as shown in Figure 7-10(c). Us ing HRTEM imaging, it is clear that trapped pockets of crystalline material are in the first several tens of nanometers of the amorphous region, as shown in Figure 7-10(d) Whether these pockets of crysta lline material are stable, or will collapse upon further annealing, is unknown. Discussion The results of these amorphization experi ments bring some clarification to the amorphization and graphitization of diamond after ion-implantation. Diamond transforms to an amorphous carbon (and not graphiti c) state upon surpassing the cr itical damage threshold, confirmed by TEM, EELS and Raman analysis The density of the amorphous carbon is approximately 1.3 0.1 1023 atoms/cm3, which is 15% more dense than graphite, but approximately 25% less dense than diamond. As im planted, the amorphous layer is 35% ( 10%) thicker than its crystalline form before amor phization. And upon annealing for 24 hrs at 1350 C, the amorphous layer has transformed to grap hite, with a densit y of 1.1 0.2 1023 atoms/cm3,

PAGE 155

155 and planar spacings equal to that of graphite. These are the dire ct results of the experiment. Using these results, the threshold fo r amorphization can be determined. Threshold damage density for diamond. To determine the threshold damage density (TDD) while accounting for swelling, the TRIM simulation for the vacancies from 1 MeV, Si+ implantation was transformed fr om vacancies//ion into keV/cm3, using the displacement energy for a diamond atom from its lattice site as 50 eV. 101 These results were plotted for four doses: 1 1014 and 1 1015 Si+ /cm3 (below the amorphization th reshold) and 3 and 7 1015 Si+ /cm2 (above the amorphization thres hold). Since the conversion to amorphous material creates an increase in the thickness of the amorphous layer, TRIM can no longer be ma tched up directly to determine the threshold. However, it need not be a direct match. The im portant interface is the near-surface interface. It is at this point that the crystal has just exceeded the threshold for amorphization. Although this region is typically rougher (wider), whic h adds error to the calculation, the deeper interface is a poor estima tion because atoms that have been implanted past the initial amorphization threshold for the proj ected range have seen two layers, a crystalline layer and an amorphous layer, before reaching a fi nal depth. Therefore, the damage parameters will be determined from the near-surface -c interface. Figure 7-11 is a plot of the threshold damage density, in keV/cm3. This is a parameter previously reported in the literature for Si and Ge. For the diamond community the critical damage (DC) is normally reported, in vacancies/cm3. The ion-implantation community norma lly prefers to report amorphization thresholds using a liquid nitrogen implant a nd reporting parameter displacements per atom (DPA), which is a unitless parameter. The DC and DPA were determined using the same method as the TDD, using the SRIM simulation of damage and the near-surface -c interface. Table 7-1 summarizes the calculated values, and repeats the Si and Ge values from earlier experiments.

PAGE 156

156 Initial analysis of the comp arison between the diamond-determined thresholds and the Si and Ge thresholds seems confusing. The values for TDD and Dc are higher for diamond than for Si and Ge, indicating that it s hould be harder to amorphize diam ond. However, the value of DPA for diamond (~0.14) is significantly less than that of Si and Ge, suggesting that amorphization of diamond is easier than for Si or Ge. The answer to this apparent paradox is in the reporting of the units. DPA is a unitless parameter, and therefor e can be compared across elements. However, TDD and Dc are both reported per volume, making density an important factor. Si and Ge have approximately the same atomic density, 5 and 4.42 1022 atoms/cm3 respectively. Diamonds atomic density is over three times higher, at 17.63 1022 atoms/cm3. Therefore, across the elements, DPA shows the trend th at recognizes it is easier to am orphize diamond then Si or Ge on a per-atom basis. When comparing a per volum e basis, the number of atoms per volume must be taken into consideration. Thus, Dc can be used to compare the relative number of lattice atoms that must be displaced (in simulation terms) in order for amorphization to take place. For Si and Ge, this is approximately equal to the atomic density. For diamond, however, the Dc value of ~ 6 1022 atoms/cm3 is significantly less than half its atomic density of 1.763 1023 atoms/cm3. This also indicates that on a pe r crystal basis, less damage needs to be done to the diamond lattice to allow amorphization to take place, as compared to Si and Ge. The TDD is reported in energy per volume, and as noted, is higher for di amond than for Si and Ge. This, too, can be explained by examining the units of measure. In Si and Ge, the energy to displace an atom from its lattice site is 15 eV, significantly less th an the approximately 50 eV to displace diamond atoms.21,101 Since there are more atoms per volume in diamond, and more energy is needed to displace an atom, more energy would be needed per volume to amorphize diamond. Furthermore, in Chapter 2, it was noted th at simulating the amorphization phenomena is a

PAGE 157

157 challenging task. This supposition is further reinfo rced in this study, shown by the fact that a single parameter cannot be used to completely describe the amorphization trends among various materials. Conclusion Thus far, the results of ion-implanta tion of both non-amorphizing and amorphizing implants into Ge, Si and C have been presen ted. In this final an alysis, the amorphization threshold of diamond has been surpassed. In co mparing the number of displacements per atom necessary in order to amorphize these crystals, diamond has the lowest at approximately 0.4 DPA, as compared to Ge and Si, all at room temp erature. This indicates that on a per atom lattice basis, less damage needs to be done to a diamond cr ystal (per unit cell) than needs to be done to Ge or Si lattice in order to induce amorphizi ation. The measurement of critical damage (Dc) also agrees with this analysis: approximately 6 x 1022 vacancies/cm3 out of 1.7 x 1023 atoms/cm3 in the diamond lattice, approximately 35% of the atoms per unit volume, need to be displaced from the diamond lattice in order to induce amorphization. In Si and Ge, this percentage is on the order of 100% of the atoms in the lat tice in order to i nduce amorphization. However, on a volumetric basis, more energy must be imparted into the same volume of diamond than Si or Ge to amorphize the same am ount of crystal, because of the high energy it takes to displace one carbon atom (ED = 50 eV) versus Si and Ge (ED = 15 eV, for both), in addition to the higher atomic density. Therefore, in order to amorphize all crystals equally, more energy must be imparted into the diamond cr ystal in order to initiate amorphization. In discussing the defects apparent after a buried amorphous layer is formed and regrown, both Si and Ge show three kinds of extended de fects: regrowth-related, clamshell and end-ofrange. No solid phase epitaxial regrowth is obs erved in diamond, and therefore, no regrowth, end-of-range or clamshell defects. No extended defects were seen to form uniformly in the

PAGE 158

158 projected range of the damaged diamond, either in the crystalline areas or in the graphite region upon annealing. Thus, there seem to be no stable extended defects that reside in diamond after ion-implantation. This agrees with computational simulations that the <100 > split-interstitial (a point defect) is the stable in terstitial defect in diamond.

PAGE 159

159 Figure 7-1. (a) WBDF-XTEM image, using g220 imaging conditions, of (001)-oriented singlecrystal diamond Si+ implanted at 1 MeV to a dose of 3 1015 cm-2 at 303K. (b) The plotted TRIM results of total vacancies crea ted by the implant, in both log and linear axis. (c) The TRIM-derived TDD values, matched with the surface of the amorphized diamond.

PAGE 160

160 Figure 7-2. (a) WBDF-XTEM image, using g220 imaging conditions, of (001)-oriented singlecrystal diamond Si+ implanted at 1 MeV to a dose of 7 1015 cm-2 at 303K. (b) The plotted TRIM results of total vacancies crea ted by the implant, in both log and linear axis. (c) The TRIM-derived TDD values, matched with the surface of the amorphized diamond.

PAGE 161

161 Figure 7-3. (a) WBDF-XTEM image of diamond Si+ implanted at 2 MeV to a dose of 1 1015 cm-2 at 77 K. (b) An inset of an on-ax is [110] BF TEM image at the same magnification of the damaged region and (c ) an on-axis [110] BF HRTEM image of the center of the damaged region.

PAGE 162

162 Figure 7-4. Raman spectrum of bulk diamond implanted with 2 MeV, 1 1015 Si+/cm2 at an implant temperature of 77 K, centered at 1400 cm-1. The characteristic sharp diamond peak at approximately 1335 cm-1 is observed. No broad skewed peaks that would indicate radiation damage are seen. Theoreti cal calculations stat e that no peaks should occur above 1400 cm-1 for sp3-bonded carbon structures. 146

PAGE 163

163 Figure 7-5. (a) WBDF TEM image, using g200 imaging conditions, of a (001)-oriented singlecrystal diamond implanted with 2 MeV, 1 1015 Si+ /cm2 at 77 K. (b) The damage density (keV/cm3) produced by a 2 MeV, 1 1015 Si+ /cm2 overlaid with a 1 MeV, 7 1015 Si+ /cm2 implant. (c) WBDF TEM image both implants: 2 MeV, 1 1015 Si+ /cm2 at 77 K and the 1 MeV, 7 1015 Si+ /cm2 implant. The 2 MeV implant serves as a buried marker layer.

PAGE 164

164 Figure 7-6. WBDF TEM images using g 220 imaging conditions of (001)-oriented diamond implanted with (a) 2 MeV, 1 1015 Si+ /cm2 at 77 K showing the crystalline region between the surface and the non-amorphized da mage layer. (b) The same sample as with an additional implant of 1 MeV, 7 1015 Si+ /cm2 at 303K, showing the remaining crystalline regions, se parated by an amorphous region.

PAGE 165

165 Figure 7-7. WBDF-XTEM images of diamond Si + implanted at 1 MeV to a dose of 3 1015 cm2 at 303 K (a) as-implanted, annealed at 1350 C for (b) 1 and (c) 24 hours. Onaxis BF-XTEM images of (d) as-implanted and (e) annealed at 1350 C for 24 hours.

PAGE 166

166 Figure 7-8. HR-XTEM images of diamond Si+ implanted at 1 MeV to a dose of 3 1015 cm-2 at 303K using on-axis [110] BF imaging condi tions, shown with the respective fast Fourier transforms (FFT): (a) as-implant ed HRTEM and (b) FFT, and annealed at 1350 C for 1 hr image (c) and FFT (d) and 24 hrs (e) image and (f) FFT.

PAGE 167

167 Figure 7-9. EELS spectrum of a (001)-oriented diamond Si+ implanted at 1 MeV to a dose of 3 1015 cm-2 at 303K, (a) as-implanted in the ne ar-surface crystallin e region (b) midamorphous layer region and (c) bulk region. After annealing for one hour at 1350 C in the (d) near-surface, (b) mid-amorphous and (c) bulk regions. Note: the zero loss peak on the annealed sample is skewed to +5 eV.

PAGE 168

168 Figure 7-10. HR-XTEM images of (001)-oriented diamond Si+ implanted at 1 MeV to a dose of 3 1015 cm-2 at 303 K and annealed at 1350 C fo r 24 hours, using on-axis [110] BF imaging conditions: (a) Whole amorphous-car bon layer, (b) upper layer, (c) lower layer and (d) detailed view of upper region.

PAGE 169

169 Figure 7-11. SRIM-derived damage density, in keV/cm3, of diamond Si+ implanted at 1 MeV to doses of 1 1014 (), 1 1015 (), 3 1015 (), and 7 1015 Si+ /cm2 ().

PAGE 170

170 Table 7-1. Simulated damage values from current experiments Material, Implant Energy, Temperature and Dose Threshold Damage Density (keV/cm3) Critical Damage (vac/cm3) DPA (unitless) Diamond 1000 keV, 303 K, 3 1015 Si+ cm-2 5 1 1021 7 1 1022 0.3 0.05 Diamond 1000 keV, 303 K, 7 1015 Si+ cm-2 3 0.5 1021 5 1 1022 0.5 0.05 Ge 1000 keV, 303 K, 1 1015 Si+ cm-2 7.0 2 1020 4.5 1.5 1022 0.8 0.2 Ge 40 keV, 303 K, 1 1015 Si+ cm-2 Bulk Interface Measurement 5.5 1.5 1020 3.5 1 1022 0.9 0.1 Si 40 keV, 303 K, 1 1015 Si+ cm-2 Bulk Interface Measurement 7.5 1.5 1020 5.0 1 1022 1.0 0.2

PAGE 171

171 CHAPTER 8 SUMMARY AND FUTURE WORK Summary In our study, the behavior of single crystals of diamond-cubic germ anium (Ge) and carbon (C) after Si+ implantation have been studied, and then co mpared to the behavior of silicon (Si). The purpose of the ion-implantation was to study the defects that occur after implantation and annealing, and to study the evolution and def ect formation after amorphizing implantation and annealing. These are common processe s used in silicon device fabric ation; therefore, if Ge or C are to be considered as base materials, the e ffects of these common processes need to become known. The following pages will summarize the results of this study. Non-amorphizing Si+ Implants into Diamond and Ge, as Compared to Si In Si, the {311} defect and dislocation l oop have been well known to impede device performance. The {311} defect has been reported in Ge for light ion and electron implantation. However, no {311} defects in Ge were observed in this study for non-amorphizing implants (1 MeV & 40 keV, 1014 Si+ cm-2). Dot-like defects appeared in Ge after implantation, and were stable up to approximately 650 C. The activat ion energy of the disso lution of the dot-like defects in Ge was determined to be approximate ly 0.3 0.1 eV, almost an order of magnitude lower than the {311} defect in Si, which has an activation energy for di ssolution of 3.6 eV.17 In diamond, temperatures exceeding 1350 C ar e necessary to completely remove the damage caused by non-amorphizing ionimplantation (1 MeV, < 3 15 Si+ cm-2). This result is in agreement with studies that indicate temper atures of 1500 C are necessary to completely remove damage,92 but contradict other studi es that report damage repair at temperatures of 5001000 C.74,80,81,147 Non-amorphizing implants into singlecrystal diamond did not result in any consistent defects during the annealing process, over temperatures rang ing from 300 to 1350 C

PAGE 172

172 for 24 hrs. Some defects, originating from the surface, did appear. However, as these defects were not uniform across samples, and also app eared in unimplanted di amonds, these defects are assumed to be unassociated with the implantatio n process. Appendix A contains XTEM images and brief descriptions of the observed defects. Amorphizing Si+ Implants into Diamond and Ge, as Compared to Si When the threshold of damage exceeded a ce rtain point, all three diamond-cubic crystal systems converted to amorphous material. Upon a nnealing, however, the three crystal systems behaved differently. Si and Ge both undergo solid phase epitaxial regrowth and evolution of regrowth-related, clamshell and e nd-of-range defects after 1 MeV, 1015 Si+ cm-2 amorphizing implants. However, the most stable defect in Si is the EOR defect; in Ge, it is the regrowthrelated defect. The activa tion energy for the dissolution of { 311} defects in Si is 3.6 eV, while my study determined the activation energy for th e dissolution of EOR defects in Ge to be approximately 0.4 0.1 eV. However, the sample size for the determination of this activation energy was low. The number of trapped interstitials in the EOR defects in Ge stayed relatively constant at 1014 cm-2. Once diamond exceeded the critical value to lose long-range crystal structure (1 MeV, 3 1015 Si+ cm-2), it did not immediately graphitize afte r implantation (as has been reported in some literature), but instead converted to an am orphous carbon state, with a density of 1.3 0.1 1023 atoms cm-3, which is approximately 25% less dense than diamond, and 15% more dense than graphite. The amorphous structure was confir med using selected area diffraction patterns, electron energy loss spectroscopy, and Raman spectroscopy characterization techniques. Upon annealing at sufficiently high temperatures and long times, th e amorphous layer converted to graphite. No SPER was observed in diamond, either for reconstruction back to a diamond lattice, or seed growth of the graphitic layer. The critical threshold de termined for the amorphization of

PAGE 173

173 diamond is approximately 6 1 1022 vacancies cm-3, which is higher than the widely referenced results of 1 1022 vacancies cm-3, 16 but is in the range of recent higher reported thresholds of 2 and 9 1022 vacancies cm-3.74,101 Within this study, the amorphization threshold may be compared across all three materials, however the results depend on the parameter used for threshold determination. Using the number of displacements of host atoms per incoming ion atom (DPA), diamond seems to be the easiest material to amorphize due to its low DPA of a pproximately 0.14. For Si and Ge, this number approached 1.0; all implants were at room temp erature. However, more energy is required to displace the C atoms in the diamond lattice, and fo r that reason, the threshold damage density is largest for diamond, at approximately 4 1021 keV cm-3, as compared to approximately 7 1020 keV cm-3 for Ge and Si, respectively. Interestingly, the number of vacancies cm-3 to amorphize the crystal seems to be approximately constant for all three crystal systems, in the range of 4-7 1021 vacancies cm-3. Ideally, these results should be compared to the historical literature for amorphization criterion. Using multiple data sources over the past twenty years for the materials of Si, Ge, SiGe, SiC, and C, the author attempte d to normalize the data and compare results. This assessment was attempted; however, due to differen ces in temperature, dose rates, calculation parameters, characterization techniques and parame ter definitions, no logical trends in the data were apparent. Future Work The current experiments demonstrate that Ge and diamond do not seem to suffer from the same stable defects which occur in Si after i on implantation and anneal ing. However, as the material behavior differs greatly than Si, the years of Si research cannot be leveraged when engineering these potential new base materials. Although the stab le end-of-range defects do not occur at low energies in Ge, and no extende d defects seem to occur for non-amorphizing

PAGE 174

174 implants into diamond, other roadblocks may ex ist for these potential base materials. More research would be necessary to determine whethe r years of Si technology may be leveraged in order to enable these potential base materials to be viable for electronic devices. The following paragraphs outline further study which could be used to further define and confirm the results of the present research. Ion choice for implantation. As noted in the literature review, the mass of the incoming ion and the mass of the target material help determine the amount of damage caused during the implantation process. In this study, the ion species, Si+, was kept constant across all three target crystal systems (Si, Ge, C) in order to minimi ze variability in the experiment. However, the result is that three different ion damage patterns occurred within each crystal system a heavier ion into a lighter base material for Si+ into C, a lighter ion into a heavier base material for Si+ into Ge and a self-implant into Si. To reduce th e variability in damage, self-implants could be used for another experiment, which could confirm or deny the validity of the defect structures and amorphization thresholds determined in our study. Affect of dopants and other factors on defect formation. As stated in the previous paragraph, Si+ was chosen as the implanted species to minimized variability in the experiment. However, for industrial use, it is important to know the effect of elect rically-active species on defect formation. In Si, dopants are known to affect defect form ation in a number of ways, as presented in Chapter 2. However, it is unknown at this time how dopant species, such as B+, P+ or As+, would affect the dot-like defect formation in Ge. Additionally, other factors have been known to affect defect fo rmation in Si. For example, an oxide layer injects interstitials into the lattice during annealing. This increases the supersaturation of interstitials in the near surf ace area, which could encourage extended defect

PAGE 175

175 formation. In our study, the 40 keV implants into Ge, upon annealing, created a small GeOx layer. It is currently unknown whether the GeOx layer affected the defect formation in this study. Activation energy of defects in Ge. In our study, an initia l estimate of the activation energy for the dissolution of both dot-like defects and end-of-range defects occurring in Ge was determined. However, a small sample size was used to determine the values, and due to the small sample size, the statistical error opportunity is high. In order to confirm the validity of these activation energies, further studies into the form ation and dissolution of these defects would be necessary. Vacancy clusters in the Ge near surface. Preliminary calculations of the interstitial population in non-amorphoizing 1 MeV implants into Ge revealed a substa ntial increas e over the predicted plus-one model used for Si. Additiona lly, the near-surface dot-like defects dissolved faster than the defects centered near the damage peak. The explanation for why Ge is diverging from the plus-one model may be that both v acancy-clusters (in the near-surface region) and interstitial clusters (in the projected range) are both present after implantation at room temperature for high-energy implants. This theory is supported by recent work revealing vacancy clusters in the near surface region after amorphizing implants into SiGe, which increased for higher concentrations of Ge.14 This theory could be tested, us ing the current implanted material, by either positron analysis or copper gettering experiments. Future work regarding extended defects in Ge should focus on characterizing the dot-like defects and clarifying the forma tion mechanism. Chronological stud ies could be used to better quantify the activation energy of th e dot-like defects. Positron anal ysis could be used to confirm a vacancy mechanism at work for potential intrin sic near-surface defects as opposed to extrinsic defects towards and beyond the projected range.

PAGE 176

176 Future work for diamond. The results of our study answ ered many questi ons regarding ion implantation into single crystal diamond which ha ve existed in the literature for some time: a more accurate critical damage threshold, the state of amorphous carbon after implantation, and the non-SPER conversion of amorphous carbon to gr aphite during annealing. Additionally, this study confirmed that anneals higher than 1350 C for 24 hours are required to completely remove the damage caused by ion implantation, even for relatively low doses. However, this study also revealed the previous ly unknown phenomena of trapped crystallites in the amorphous region, which were stable up to temperatures of 1350 C for 24 hrs. Wh ether these pockets of crystalline material continue to be stable, or will collapse up on further annealing, is unknown.

PAGE 177

177 APPENDIX A SURFACE INITIATED DEFECTS OF ANNEALED DIAMOND Throughout this study, annealing of the singl e-crystal diamond samples produced surfaceinitiated defects. These defect s occurred in both implanted and non-implanted specimens. The defects occurred for relatively low anneal temperatures (950 C ) and at high temperatures (1350 C). As the defects were intermittent, while the implant was uniform, these defects are not thought to be associated with the implantation regimes. Howeve r, as they are defects that occurred during the annealing process, and were revealed during XTEM imaging, the images will be presented here. All of the defects seem to originate at the surface, some with a 54 angle between the defect and the surface. That would indicate that the defect is traveling along the (111) plane in the <110> direction. The images presented in the followi ng pages are of defects seen in high dose implants with high temperatur e and long anneal time; mid-dose implants with high temperatures and short anneal times, low do ses with low annealing temperatures and short annealing times, and images of defects s een in an unimplanted annealed diamond.

PAGE 178

178 (a) (b) Figure A-1. WBDF XTEM images of single-crystal diamond Si+ implanted at 1 MeV to a dose of 3 1015 cm-2 at 303 K after annealing for 24 hrs at 1350 C (a) defects in one section of the sample and (b) same magnification of a different section of the same XTEM sample.

PAGE 179

179 (a) (b) Figure A-2. WBDF XTEM images of single-crystal diamond Si+ implanted at 1 MeV to a dose of 1 1015 cm-2 at 77 K after annealing for 24 hrs at 1350 C (a) defects in one section of the sample and (b) higher magnifi cation of a different section of the same XTEM sample.

PAGE 180

180 (a) (b) Figure A-3. (a) WBDFand (b) BF-XTEM images of single-crystal diamond Si+ implanted at 1 MeV to a dose of 1 1014 cm-2 at 303 K after annealing for 2 hrs at 950 C.

PAGE 181

181 Figure A-4. BF-XTEM images of single-crystal diamond un-implanted, afte r annealing for 2 hrs at 950 C.

PAGE 182

182 APPENDIX B BORON IMPLANTATION INTO DIAMOND Introduction As noted in the introduction, the implantation of Boron (B+) is of technological interest for purposes of doping a transistor channel, source or drain region. An electrically well-studied implant in the literature has been repeated here. The B+ implantation is at 2 MeV to a dose of 1 1015 B+/cm2 at an implant temperature of 303 K. Though studied through electrical measurements, the author is unaware of any published XTEM studies. The following summarizes key B+ doping of diamond papers publis hed in the last 15 years. In 1993, Zeidler et al. implanted B+, C+ and a combination of the two elements into natural single-crystal diamond at implant temperatures of 77, 300 or 800 K. They concluded that low temperature implantations provided the best carr ier concentration and mobility. Using a multiplestep, low-temperature implant, B+ implan tation procedure, this Naval Command group fabricated a single-crystal diamond insula ted gate field eff ect transistor. In 1996, Fontaine, Uzan-Saguy, Philosoph and Ka lish, of Israel, used Prins suggested cold-implantation rapid-anneali ng procedure to produce record va lues of high Hall mobility of B+ doped diamond. They implanted at -97 C, ann ealed first at 1050 C and then further annealing at 1450 C for 10 min to achieve a p-type room temperature mobility of 385 cm2/V s. In 1998, the same group (plus Prawer from Australia) improve d the process to achieve a room temperature Hall mobility of 585 cm2/V s, 5% compensation ratio, acceptor concentration of 3.9 0.3 1018 cm-2 and an activation energy of approximately 0.354 eV. In the same year, Prins studied B+ doped diamonds by catholuminence, adding more in sight into the resistance behavior of boron doping in both natural and synthetic diamonds.

PAGE 183

183 In 2002, Prins conducted a rigorous study of implanting C+, N+, B+ and O+ into natural single-crystal diamonds, and determined that O+ and N+ treated diamond donor states were metastable, and annealed out at temperatures about 600 C. However, it should be noted that all three dopant atom implanted diamonds had lo wer electrical resist ivities than the C+ implanted diamonds. The acceptor state in the B+ treated di amonds was determined to be at ~0.37 eV, while the donor states in the N+ and O+ im planted diamonds were at ~0.28 and ~0.32 eV, respectively. In 2004, Vogel et al. (based out of Germa ny and the City University in New York) implanted at the same conditions, except the im plant temperature was 303 K. Using Hall effect and conductivity measurements, after anneals up to 1650 C, they showed clear p+ type conduction, with B+ activation up to 30%, the highest yet reported for ion-implanted diamond. They concluded that the best annealing condi tions for removing damage, and achieving the highest conductivity of 230 cm2/ V s, was at anneals of 1500 C for 15 min. Annealing the samples above 1500 C did not improve, but ac tually reduced, conductivity measurements. In 2006, a research group based in Japan, used natural and HPHT diamonds to study B+ dosage dependencies on structural and electrical properties of doped diamond. A B+ box profile, 30-380 nm below the surface, was created using doses in the range of 2.1 1015 B+/cm2 1.7 1017 B+/cm2. The implant temperature was ~ 400 C a nd the anneal treatment was 1450 C in vacuum for 1 hour. Using the high implant te mperature, the group prevented collapse to graphitization and allowed the diam ond to form a very low resistance p+ diamond layer, with a B concentration of 2.5 1021 cm-3 and resistively of 2.3 m cm at room temperature. The transport properties were shown to be almost temperature independent and p-type carrier concentration was realized in the range of 80-1000 K.

PAGE 184

184 Experimental Design Gem-quality HPHT single-crystal diamond samples, approximately 5 5 1 mm3, were B+ implanted at 2 MeV to a dose of 1 1015 B+/cm2 at an implant temperature of 303 K. This simulates the implant conditions of previous studies. The implantation of B+ into the diamonds was performed in a tandem accelerator ion implanter, with a dose rate of 0.44 A/cm2, and the samples were tilted 7 relative to the incident ions to minimize ion channeling. The original nitrogen (N2) impurity concentration of the diamonds obvious from the visible light-orange color, is on the order of 1018 1020 cm-3, 140 and was uniform thr oughout the crystals. According to simulations, 21 the damage should peak (Rd) at approximately 1.35 m and the projected range (Rp) of the ions are approximately 1.38 m below the surface of the diamond. After implantation, approximately 150 nm of carbon was evaporated onto all samples, for conductivity and protection of the sample surface during focused-ion beam processing. Cross-sectional TEM (XTEM) samples were prepar ed according to the procedure previously published, 117 during which a layer of Platinum was applied to the surface. Annealing was performed in both tube and gr aphite furnaces. Samples annealed at 1050 C and below were annealed in a Lindberg tube furnace in a high-purity N2 environment. Anneals at 1350 C used a graphite chamber furnace at diffusion pump vacuum pressures ( 10-4 Torr) 142 with a residual helium environment. Additionally, the samples were placed in tantalum (Ta) foil (folded as envelopes), which is an oxygen getterer, to additionally prevent accidental oxidation of the diamond surface. The Ta envelopes se rved the additional purpose of separating and marking each sample. Transmission electron microscopy analysis was performed on JEOL 200CX and JEOL 2010 microscopes.

PAGE 185

185 Results & Discussion WBDF-XTEM imaging of the B+ implant showed that the simulation-predicted RD = 1.35 m and RP = 1.38 m matched well with the experimental damage band. The damage band is still apparent after a 1000 C, 10 min anneal in a tube furnace [F ig. B-1(a)], but after a 1350 C at 24 hour anneal, the crystal structure is almost completely recovered, and the damage band is almost negligible in WBDF imaging. Further an nealing, or higher temperature annealing, would likely remove all semblance of damage from the ion implantation process. This agrees well with Vogels claim that all implant dama ge, and highest ac tivation of the B+ dopant, takes place at 1500 C for 15 min. Though this experiment was not ab le to repeat the high anneal temperatures of Vogels previously repor ted literature, this study is the firs t XTEM analysis of the damage and recovery of single-crystal diamond after B+ implantation.

PAGE 186

186 Figure B-1. WBDF-XTEM images of (001)-oriented diamond B+ implanted at 2 MeV to a dose of 1 1015 cm-2 at 303K as implanted, and af ter annealing (a) in flowing N2 ambient at 1000 C for 10 min and (b) in vacuum at 1350 C for 24 hrs.

PAGE 187

187 LIST OF REFERENCES 1 A.R. Peaker, A. Satta, V. P. Markdevich, E. Simoen, and B. Hamilton, presented at the 16th International Conference on Ion implanta tion Technology, Mars eille, France, 2006 (unpublished). 2 G. E. Moore, Electronics 38, 114-117 (1965). 3 Semiconductor Industry Association, (Sematech, Inc., 2005), Vol. 2005. 4 Gehan A. J. Amaratunga, Science 297, 1657-1658 (2002). 5 T. Y. Tan, Philosophical Magazine A Phys ics of Condensed Matter Strucutre Defects and Mechanical Properties 44, 101-125 (1981). 6 J. H. Li and K. S. Jones, Appl. Phys. Lett. 73, 3748-3750 (1998). 7 I.G. Salisbury and M. H. Loretto, Philosophical Magazine A 39, 317-323 (1979). 8 M. Pasemann, D. Hoehl, A. L. Aseev, and O. P. Pchelyakov, Phys. Status Solidi A-Appl. Res. 80, 135-139 (1983). 9 K. S. Jones, S. Prussin, and E. R. We ber, Applied Physics A-Materials Science & Processing 45, 1-34 (1988). 10 C. A. Ferreira Lima, Howie, A., Philosophical Magazine 34, 1054-1071 (1976). 11 T. Akatsu, Bourdelle, K. K., Richtarch, C., Faure, B., and Letertre, F., Appl. Phys. Lett. 86, 181910-181911 (2005). 12 L. Hutchison, A. L. Aseev, and L. I. Fedina, in Microscopy Of Semiconducting Materials 1993 (1993), pp. 41-46. 13 R. Crosby, K. S. Jones, M. E. Law, A. N. Larsen, and J. L. Hansen, Journal Of Vacuum Science & Technology B 22, 468-470 (2004). 14 R. Kogler, A. Mucklich, W. Skorupa, A. Peev a, A. Y. Kuznetsov, J. S. Christensen, and B. G. Svensson, J. Appl. Phys. 101 (2007). 15 P. A. Stolk, H. J. Gossmann, D. J. Eaglesham, and J. M. Poate, Mater. Sci. Eng. B-Solid State Mater. Adv. Technol. 36, 275-281 (1996). 16 C. Uzan-Saguy, Cytermann, C., Brener, R ., Richter, V., Shaanan, M., and Kalish, R., Applied Physics Letters 67, 1194-1196 (1995). 17 D. J. Eaglesham, Stolk, P. A., Gossmann, H. -J. and Poate, J. M., Applied Physics Letters 65, 2305-2307 (1994).

PAGE 188

188 18 E. Chason, S. T. Picraux, J. M. Poate, J. O. Borland, M. I. Current, T. D. delaRubia, D. J. Eaglesham, O. W. Holland, M. E. Law, C. W. Magee, J. W. Mayer, J. Melngailis, and A. F. Tasch, J. Appl. Phys. 81, 6513-6561 (1997). 19 M.S. Dresselhaus and R. Kalish, Ion Implantation in Diamond, Graphite and Related Materials (Springer-Verlag, Berlin He idelberg New York, 1992), p.202. 20 J. F. Ziegler, Ion Implantation Technology, 6th Edition, 752 pages, (Ion Implantation Technology Co., Yorktown, 1996). 21 J. F. Ziegler, Nucl. Instrum. Methods Phys Res. Sect. B-Beam Interact. Mater. Atoms 219-20, 1027-1036 (2004). 22 S. Libertino, in 17TH International Conference on the Application of Accelerators in Research and Industry edited by AIP Conference Proceedings (AIP, Catania, Italy, 2003), Vol. 680, pp. 750-754. 23 A. Claverie, B. Colombeau, B. De Mauduit, C. Bonafos, X. Hebras, G. Ben Assayag, and F. Cristiano, Applied Physics AMaterials Science & Processing 76, 1025-1033 (2003). 24 G. Z. Pan, K. N. Tu, and S. Prussin, Appl. Phys. Lett. 71, 659-661 (1997). 25 J. L. Benton, S. Libertino, P. Kringhoj, D. J. Eaglesham, J. M. Poate, and S. Coffa, J. Appl. Phys. 82, 120-125 (1997). 26 S. Coffa, S. Libertino, and C. Spinella, Appl. Phys. Lett. 76, 321-323 (2000). 27 P. Keys, Phosphorus-Defect Interactions During Thermal Annealing of Ion Implanted Silicon Dissertation, University of Flor ida, Gainesville, FL (2001), p.137. 28 J. Wong-Leung, S. Fatima, C. Jagadish, J. D. Fitz Gerald, C. T. Chou, J. Zou, and D. J. H. Cockayne, J. Appl. Phys. 88, 1312-1318 (2000). 29 H. Bartsch, D. Hoehl, and G. Kast ner, Phys. Status Solidi A-Appl. Res. 83, 543-551 (1984). 30 I. G. Salisbury and M. H. Loretto, Radiat. Eff. Defects Solids 59, 59-68 (1981). 31 L. S. Robertson, K. S. Jones, L. M. Rubin, and J. Jackson, J. Appl. Phys. 87, 2910-2913 (2000). 32 R. Brindos, P. Keys, K. S. Jones, and M. E. Law, Appl. Phys. Lett. 75, 229-231 (1999). 33 F. F. Morehead, B. L. Crowder, and R. S. Title, J. Appl. Phys. 43, 1112-& (1972).

PAGE 189

189 34 M. K. Elghor, O. W. Holland, C. W. White and S. J. Pennycook, Journal Of Materials Research 5, 352-359 (1990). 35 S. Prussin, D. I. Margolese, and R. N. Tauber, J. Appl. Phys. 57, 180-185 (1985). 36 L. Pelaz, L. A. Marques, a nd J. Barbolla, J. Appl. Phys. 96, 5947-5976 (2004). 37 L. A. Christel, J. F. Gibbons, and T. W. Sigmon, J. Appl. Phys. 52, 7143-7146 (1981). 38 G. Z. Pan, K. N. Tu, and S. Prussin, Appl. Phys. Lett. 68, 1654-1656 (1996); B. Demauduit, L. Laanab, C. Bergaud, M. M. Faye, A. Martinez, and A. Claverie, Nucl. Instrum. Methods Phys. Res. Sect B-Beam Interact. Mater. Atoms 84, 190-194 (1994). 39 M. E. Law and S. M. Cea, Comput. Mater. Sci. 12, 289-308 (1998). 40 L. Pelaz, L. A. Marques, M. Aboy, P. L opez, and J. Barbolla, Comput. Mater. Sci. 33, 92-105 (2005). 41 C. S. Rafferty, G. H. Gilmer, M. Jaraiz, D. Eaglesham, and H. J. Gossmann, Appl. Phys. Lett. 68, 2395-2397 (1996). 42 C. Rafferty and R. K. Smith, Cmes-Com puter Modeling In Engineering & Sciences 1, 151-159 (2000). 43 I. Avci, M. E. Law, E. Kuryliw, A. F. Saavedra, and K. S. Jones, J. Appl. Phys. 95, 24522460 (2004). 44 C. Zechner, N. Zographos, D. Matveev, and A. Erlebach, Mater. Sci. Eng. B-Solid State Mater. Adv. Technol. 124, 401-403 (2005). 45 J. Kim, F. Kirchhoff, J. W. Wilkin s, and F. S. Khan, Phys. Rev. Lett. 84, 503-506 (2000). 46 M. Kohyama and S. Takeda, Phys. Rev. B 51, 13111-13116 (1995). 47 L. A. Marques, L. Pelaz, P. Cast rillo, and J. Barbolla, Phys. Rev. B 71 (2005). 48 P. Keblinski, D. Wolf, S. R. Phillpot, and H. Gleiter, Journal Of Materials Research 13, 2077-2099 (1998). 49 A. L. Aseev, Ivakhnishin, V. M., Stas', V. F., Smirnov, L.S., Soviet Physics Solid State (Translation of Fiz. Tverd. Tela) 25, 1786-1789 (1983). 50 A. L. Aseev, V. M. Ivakhnishin, V. F. St as, and L. S. Smirnov, Fiz. Tverd. Tela 25, 30973103 (1983).

PAGE 190

190 51 K. Zellama, Morhange, J. F., Germain, P ., Bourgoin, J.C., Physica Status Solidi (a) 56, 717-723 (1979). 52 Y. L. Chao, S. Prussin, J. C. S. Woo, and R. Scholz, Appl. Phys. Lett. 87, 142102 (2005). 53 A. Satta, T. Janssens, T. Clarysse, E. Simoen M. Meuris, A. Benedetti, I. Hoflijk, B. De Jaeger, C. Demeurisse, and W. Vandervorst, Journal Of Vacuum Science & Technology B 24, 494-498 (2006). 54 A. Satta, E. Simoen, T. Clarysse, T. Janssens A. Benedetti, B. De Jaeger, M. Meuris, and W. Vandervorst, Appl. Phys. Lett. 87, 172109 (2005). 55 A. Satta, E. Simoen, R. Duffy, T. Janssens, T. Clarysse, A. Benedett i, M. Meuris, and W. Vandervorst, Appl. Phys. Lett. 88, 162118 (2006); A. Satta, E. Simoen, T. Janssens, T. Clarysse, B. De Jaeger, A. Benedetti, I. Hoflijk, B. Brijs, M. Meuris, and W. Vandervorst, J. Electrochem. Soc. 153, G229-G233 (2006). 56 B. Stritzker, R. G. Elliman, and J. Zou, Nu cl. Instrum. Methods Phys. Res. Sect. B-Beam Interact. Mater. Atoms 175, 193-196 (2001). 57 I. Jencic, I. M. Robertson, and J. Skvarc Nucl. Instrum. Methods Phys. Res. Sect. BBeam Interact. Mater. Atoms 148, 345-349 (1999). 58 H. Huber, W. Assmann, S. A. Karamian, A. Mucklich, W. Prusseit, E. Gazis, R. Grotzschel, M. Kokkoris, E. Kossionidis, H. D. Mieskes, and R. Vlas tou, Nucl. Instrum. Methods Phys. Res. Sect. B-B eam Interact. Mater. Atoms 122, 542-546 (1997). 59 S. Solmi, L. Mancini, S. Milita, M. Servi dori, G. Mannino, and M. Bersani, Appl. Phys. Lett. 79, 1103-1105 (2001); P. Alippi a nd L. Colombo, Phys. Rev. B 62, 1815-1820 (2000). 60 J. P. Dismukes, R. J. Paff, and L. Ekstrom, Journal Of Physical Chemistry 68, 3021-& (1964); C. Tzoumanekas and P. C. Kelires, Phys. Rev. B 66 (2002). 61 K. Laaziri, S. Roorda, and J. M. Baribeau, J. Non-Cryst. Solids 191, 193-199 (1995). 62 R. Devanathan, F. Gao, and W. J. Weber, Appl. Phys. Lett. 84, 3909 (2004). 63 B. Pecz, A. Barna, V. Heera, F. Fontaine, and W. Skorupa, in Silicon Carbide And Related Materials, Ecscrm2000 (Trans Tech Publications Ltd, Zurich-Uetikon, 2000), Vol. 353-3, pp. 199-202. 64 H. Weishart, F. Eichhorn, V. Heera, B. P ecz, A. Barna, and W. Skorupa, J. Appl. Phys. 98 (2005).

PAGE 191

191 65 M. Ishimaru, Nucl. Instrum. Methods Phys. Res. Sect. B-Beam Interact. Mater. Atoms 250, 309-314 (2006). 66 M. Ishimaru, I. T. Bae, A. Hirata, Y. Hirotsu, J. A. Valdez, and K. E. Sickafus, Phys. Rev. B 72 (2005). 67 J. Isberg, a, J. Hammersberg a, D. J. Twitchen b and A. J. Whitehead b, Diamond and Related Materials 13, 320-324 (2004); E. Ebert Kohn, W. Aleksov, A. Denisenko, A. Adamschik, M. Schmid, P., Microelectro nics, 2002. MIEL 2002. 23rd International Conference on 1, 59 66 (2002). 68 M. Kubovic Aleksov A., M. Kasu b, P. Schmid, D. Grobe, S. Ertl, M. Schreck, B. Stritzker and E. Kohn, Di amond and Related Materials 13, 233-240 (2004). 69 F. P. Bundy, W. A. Bassett, M. S. Weathers, R. J. Hemley, H. K. Mao, and A. F. Goncharov, Carbon 34, 141-153 (1996). 70 F. P. Bundy, H. T. Hall, H. M. Strong, and R. H. Wentorf, Nature 176, 51-55 (1955). 71 M. Kubovic, M. Kasua, I. Kallfassa, M. Neuburgera, A. Aleksova, G. Koleyc, M. G. Spencerc and E. Kohna, Diamond and Related Materials 13, 802-807 (2003). 72 J. Isberg, J. Hammersberg, E. Johansson, T. Wikstrm, D.J. Twitchen, A.J. Whitehead, S.E. Coe, and G.A. Scarsbrook, Science 297, 1670-1672 (2002). 73 D. L. Dreifus, A. Collins, T. Humphreys, K. Das, and P.E. Phersson, Diamond for Electronic Applications (Materials Research Society, Pittsburge, Pennsylvania, 1995); Y. Bando, M. Akaishi, H. Kanda, Y. Matsui, T. Kobayashi, K. Watanabe, and N. Yamashita, New Semiconducting Materials: Diamond and Related Materials: Proceedings of the 6th NIRIM International Symposium on Advanced Materials (ISAM '99) (National Institute for Research in Inorganic Ma terials, Tsukuba, Japan, 1999). 74 P. Olivero, S. Rubanov, P. Reichart, B. C. Gibson, S. T. Huntington, J. Rabeau, A. D. Greentree, J. Salzman, D. Moore, D. N. Jamieson, and S. Prawer, Adv. Mater. 17, 2427-+ (2005). 75 P. Olivero, S. Rubanov, P. Reichart, B. C. Gi bson, S. T. Huntington, J. R. Rabeau, A. D. Greentree, J. Salzman, D. Moore, D. N. Ja mieson, and S. Prawer, Diamond And Related Materials 15, 1614-1621 (2006). 76 A. Olbrich, Ebersberger, B., Boit, C., Niedermann, Ph., Hanni, W., Vancea, J., and Hoffman, H., Journal of Vacuum Science & Technology B 17, 1570-1574 (1999). 77 K. Ueda, M Kasu, Y. Yamauchi, T. Makimoto, M. Schwitters, D.J. Twitchen, G.A. Scarsbrook, and S.E. Coe, IEEE Electron Device Letters 27, 570-572 (2006); M.

PAGE 192

192 Kubovic, K. Janischowsky, and E. K ohn, Diamond and Related Materials 14, 514-517 (2005). 78 S. Koizumi, K. Watanabe, F. Hasegawa, and H. Kanda, Science 292, 1899-1901 (2001). 79 M. Werner, Semiconductor Science And Technology 18, S41-S46 (2003); M. Werner, R. Job, A. Denisenko, A. Zaitsev, W. R. Fahrner, C. Johnston, P. R. Chalker, and I. M. BuckleyGolder, Diamond And Related Materials 5, 723-727 (1996); M. Yokoba, Koide, Y., Otsuki, A., Ako, F., Oku, T., Murakami, M., J. Appl. Phys. 81, 6815-6821 (1997); P.E. Viljoen, E.S. Lambers, and P.H. Holloway, Journal of Vacuum Science and Technology B 12, 2997-3005 (1994). 80 J. R. Zeidler, Hewett, C. A., Wilson, R. G., Phys. Rev. B 47, 2065 (1993). 81 R. Kalish, Reznik, A., Prawer, S., Saada, D., Adler, J., Phys. Status Solidi A-Appl. Res. 174, 83-89 (1999). 82 J. F. Prins, Semicond. Sci. Technol. 18, S27-S33 (2003). 83 J. F. Prins, Diamond And Related Materials 10, 1756-1764 (2001). 84 J.F. Prins, Mater. Sci. Rep. (Netherlands) 7 (Feb. 1992) 271-364 (1992). 85 R. Kalish and S. Prawer, in Handbook of Industrial Diamonds and Diamond Films edited by M.A. Prelas, F. Popovici, and L. K. Bigelow (New York, 1997), pp. 945-982. 86 G. Davies, B. Campbell, A. Mainwood, M. Newton, M. Watkins, H. Kanda, and T. R. Anthony, Phys. Status Solidi A-Appl. Res. 186, 187-198 (2001); H. E. Smith, G. Davies, M. E. Newton, and H. Kanda, Phys. Rev. B 69 (2004). 87 J.F. Prins, Appl. Phys. Lett. 73, 2308-2310 (1998). 88 J. F. Prins, Diamond And Related Materials 11, 612-617 (2002). 89 C. Uzan-Saguy, R. Kalish, R. Walker, D. N. Jamieson, and S. Prawer, Diamond And Related Materials 7, 1429-1432 (1998). 90 F. Fontaine, C. UzanSaguy, B. Philosoph, and R. Kalish, Appl. Phys. Lett. 68, 2264-2266 (1996). 91 R. Walker, Prawer,S., Jamieson,D. N., Nugent K. W. and Kalish, R., Appl. Phys. Lett. 71, 1492-1494 (1997). 92 T. Vogel, J. Meijer, and A. Za itsev, Diamond And Related Materials 13, 1822-1825 (2004).

PAGE 193

193 93 H. Weishart, V. Heera, and W. Skorupa, J. Appl. Phys. 97 (2005). 94 B. Fizgeer, C. Uzan-Saguy, C. Cytermann, V. Richter, I. Avigal, M. Shaanan, R. Brener, and R. Kalish, Phys. Status Solidi A-Appl. Res. 186, 281-289 (2001). 95 C. Saguy, A. Reznik, E. Baskin, Z. Remes, and R. Kalish, Diamond And Related Materials 13, 722-726 (2004); H. Kato, S. Yamasaki, and H. Okushi, Appl. Phys. Lett. 86 (2005). 96 L.W. Yin, Zou, Z.D., Li, M.S. Geng, G.L., Sun, D. S., Hao, Z. Y. Yao, Z.Y., Cryst. Res. Technol. 35, 1289-1294 (2000). 97 J. F. Prins, Diamond And Related Materials 9, 1835-1839 (2000). 98 J. F. Prins, T. E. Derry, and J. P. F. Sellschop, Phys. Rev. B 34, 8870-8874 (1986). 99 Johan F. and Trevor E. Derry Prins, Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms 166-167, 364-373 (2000). 100 R. Kalish, A. Reznik, K. W. Nugent, and S. Prawer, Nucl. Instrum. Methods Phys. Res. Sect. B-Beam Interact. Mater. Atoms 148, 626-633 (1999). 101 R. Brunetto, Giuseppe A Baratta and Giovanni Strazzulla, Journal of Physics: Conference Series 6, 120-125 (2005). 102 S. Prawer, Kalish, R., Phys. Rev. B 51, 15711 (1995). 103 R. Walker, S. Prawer, D. N. Jamieson, K. W. Nugent, and R. Kalish, Appl. Phys. Lett. 71, 1492-1494 (1997); S. Prawer, D. N. Jamieson, K. W. Nugent, R. Walker, C. UzanSaguy, and R. Kalish, Diam. Films Technol. 8, 195-210 (1998). 104 P.F. Lai, S. Prawer, and L.A. Bursill, Diamond and Related Materials 10, 82-86 (2001). 105 L. W. Yin, M. S. Li, D. S. Sun, and J. J. Cui, Materials Letters 48, 21-25 (2001). 106 L. W. Yin, M. S. Li, D. S. Sun, F. Z. Li, and Z. Y. Hao, Materials Letters 55, 397-402 (2002). 107 Y. Nishibayashi, Y. Ando, H. Furuta, K. Kobashi, H. Furuta, K. Meguro, T. Imai, T. Hirao, and K. Oura, Molecular Crystals And Liquid Crystals 386, 183-188 (2002). 108 F. Langenhorst, J. P. Poirier, and D. J. Frost, Journal Of Materials Science 39, 1865-1867 (2004).

PAGE 194

194 109 R. Erni, B Freitag, P Hartel, H. Muller, P. Tiemeijer, M. Van der Stam, M. Stekelenburg, D. Hubert, P. Specht, and V. Gariba y-Febles, Microscopy and Microanalysis 12, 492497 (2006). 110 D. P. Adams, M. J. Vasile, T. M. Mayer, and V. C. Hodges, Journal Of Vacuum Science & Technology B 21, 2334-2343 (2003). 111 in CRC Handbook of Chemistry and Physics edited by David R. Lide (Taylor & Francis Group, 2005-2006), pp. 4-125. 112 J. W. Edington, Practical Electron Microscopy (1977). 113 J.I. Gersten and F.W. Smith, The Physics and Chemistry of Materials (John Wiley & Sons, Inc., 2001), p.826. 114 Yu-Ran Luo and J. Alistair Kerr, in CRC Handbook of Chemistry and Physics edited by David R. Lide (Taylor & Francis Group, 2005-2006), pp. 9 54-62. 115 L.L. Berger, in CRC Handbook of Chemistry and Physics edited by David R. Lide (Taylor & Francis Group, 2005-2006), pp. 12-82. 116 D. Saada, J. Adler, and R. Kalish, Int. J. Mod. Phys. C 9, 61-69 (1998). 117 D. P. Hickey, E. Kuryliw, K. Siebein, K. S. Jones, R. Chodelka, and R. Elliman, Journal Of Vacuum Science & Technology A 24, 1302-1307 (2006). 118 L. A. Giannuzzi and F. A. Stevie, Micron 30, 197-204 (1999). 119 T. Evans, in Physical Properties of Diamond edited by R. Berman (Clarendon Press, Oxford, 1965), pp. 116-134. 120 W. Zhu, in Diamond Electronic Properties and Applications edited by Lawrence S. Pan and Don R. Kania (Kluwer Academic Publishers, Boston, 1995), pp. 175-240. 121 Kurt E. Sickafus, in Encyclopedia of Materials Characterization edited by C. Richard Brundle, Charles A. Evans Jr., and Shaun Wilson (Butterworth-Heinemann, Boston, 1992), pp. 99-119. 122 R. Brydson, Electron Energy Loss Spectroscopy (BIOS Scientific Publishers Limited, Oxford, 2001). 123 Ferguson, Berry, and Thompson, Mem. Geo. Soc. Am. 85 (1962). 124 Sanc and Plytchna, Foreing Trade Corporation (1990).

PAGE 195

195 125 Roy H. Geiss, in Encyclopedia of Materi als Characterization edited by C. Richard Brundle, Charles A. Evans Jr., and Shaun Wilson (Butterworth-Heinemann, Boston, 1992), pp. 120-134. 126 R.F. Egerton, Electron energy-loss pectroscop y in the electron microscope (Plenum Press, New York, 1986). 127 J. Robertson, Materials Sc ience & Engineering R-Reports 37, 129-281 (2002). 128 D. S. Knight and W. B. White Journal Of Materials Research 4, 385-393 (1989). 129 William B. White, in Encyclopedia of Materials Characterization edited by C. Richard Brundle, Charles A. Evans Jr., and Shaun Wilson (Butterworth-Heinemann, Boston, 1992), pp. 428-441. 130 S. E. Thompson, M. Armstrong, C. Auth, M. Alavi, M. Buehler, R. Chau, S. Cea, T. Ghani, G. Glass, T. Hoffman, C. H. Jan, C. Kenyon, J. Klaus, K. Kuhn, Z. Y. Ma, B. McIntyre, K. Mistry, A. Murthy, B. Obradovi c, R. Nagisetty, P. Nguyen, S. Sivakumar, R. Shaheed, L. Shiften, B. Tufts, S. Tyagi, M. Bohr, and Y. El-Mansy, IEEE Transactions On Electron Devices 51, 1790-1797 (2004). 131 P. R. Chidambaram, C. Bowen, S. Chak ravarthi, C. Machala, and R. Wise, IEEE Transactions On Electron Devices 53, 944-964 (2006). 132 L. Csepregi, R. P. Kullen, J. W. Maye r, and T. W. Sigmon, Solid State Commun. 21, 1019-1021 (1977). 133 K.C. Thompson-Russell and J. W. Edington, in Monographs in Practical Electron Microscopy in Materials Science edited by J. W. Edington (Philips Technical Library, Holland, 1977), Vol. 5, p. 62. 134 V. C. Venezia, T. E. Haynes, A. Agarwal, L. Pelaz, H. J. Gossmann, D. C. Jacobson, and D. J. Eaglesham, Appl. Phys. Lett. 74, 1299-1301 (1999). 135 E. Simoen, A. Satta, M. Meuris, T. Janssens T. Clarysse, A. Bene detti, C. Demeurisse, B. Brijs, I. Hoflijk, W. Vandervorst, and C. Claeys, in Gettering And Defect Engineering In Semiconductor Technology Xi (Trans Tech Publications Ltd, Zurich-Uetikon, 2005), Vol. 108-109, pp. 691-696. 136 K. S. Jones and D. Venables, J. Appl. Phys. 69, 2931-2937 (1991). 137 T. Sands, J. Washburn, R. Gronsky, W. Ma szara, D.K. Sadana, and G.A. Rozgonyi, 13th Intl. Conf. on Defects in Semiconductors, Coronado, 531-537 (1984). 138 V.S. Vavilov, Phys. Status Solidi A (E ast Germany) 31 (16 Sept. 1975) 11-26 (1975); V.S. Vavilov, Radiat. Eff. (UK) 37 (1978) 229-36 (1978).

PAGE 196

196 139 K. S. Jones, M. E. Law, R. Brindos, and P. Keys, in Microscopy Of Semiconducting Materials 2001 (Iop Publishing Ltd, Bristol, 2001), pp. 349-356. 140 P.J. Glielisse, in Handbook of Industrial Diamonds and Diamond Films edited by M.A. Prelas, F. Popovici, and L.K. Bigelow (Mar cel Dekker, Inc., New York, 1997), pp. 49-88. 141 R. T. DeHoff, Thermodynamics in Materials Science (McGraw-Hill), p.326. 142 J.F. O'Hanlon, A User's Guide to Vacuum Technology Third ed. (John Wiley & Sons, Hoboken, 2003). 143 S. Prawer, I. Rosenblum, J. O. Orwa, and J. Adler, Chemical Physics Letters 390, 458461 (2004). 144 D. C. Hunt, D. J. Twitchen, M. E. Newt on, and J. M. Baker, T. R. Anthony, W. F. Banholzer, W. F. Banholzer, S. S. Vagarali, Phys. Rev. B 61, 3863 (2000). 145 E. W. Maby, C. W. Magee, and J. H. Morewood, Appl. Phys. Lett. 39, 157-158 (1981); C. Uzansaguy, V. Richter, S. Prawer, Y. Li fshitz, E. Grossman, and R. Kalish, Diamond And Related Materials 4, 569-574 (1995). 146 J. O. Orwa, K. W. Nugent, D. N. Jamieson, and S. Prawer, Phys. Rev. B 62, 5461-5472 (2000). 147 J. D. Hunn, S. P. Withrow, C. W. White, R. E. Clausing, L. Heatherly, C. P. Christensen, and N. R. Parikh, Nucl. Instrum. Methods Phys. Res. Sect. B-Beam Interact. Mater. Atoms 99, 602-605 (1995).

PAGE 197

197 BIOGRAPHICAL SKETCH Diane Patricia Hickey was born in Summit, New Jersey and moved to Mobile, Alabama at a young age. A graduate of McGill-Toolen High School, she attended Auburn University where she was selected into the Cater Society, recogniz ing her as one of the top fifteen female leaders on campus. She was a technical editor of The A uburn Plainsman, a Camp War Eagle counselor, a founder of the engineering ambassadors' Cupola Engineering Society, a nd member of several honor societies including Mortar Board before gradua ting with a B. S. in chemical engineering in 1996. During that time, she also volunteered for the 1996 Atlanta Summer Olympics. From 1997-2003, Diane worked for Accenture, C-Bri dge and the non-profit Cassiopeia Science Foundation., with volunteer work with Hands-On A tlanta, the Nexus Art Center, the United Way and the 2000 Sydney Summer Olympics. In 2003, she started graduate studies at the University of Florida, receiving her Master of Science in Materials Science and Engineering in 2004. She also pursued graduate work in science comm unication in the College of Journalism. Through journalism and the Materials Research Society, she became involved in science outreach and was selected to attend outreach tr aining sponsored by the National Science Foundation as a part of the Nanoscale Informal Science Education Networ k. She has written and presented her science communication research at seve ral conferences, was awarded best paper at the 2005 AMHCR conference, and recorded a podcast essay for the San Francisco Exploratorium's SmallTalk series. Dianes materials scienc e and engineering research focu s involves using germanium and single-crystal diamond for electron ic devices, and will receive he r doctoral degree in materials science and engineering in August of 2007.