<%BANNER%>

Amorphization and Solid Phase Epitaxial Growth of Germanium

Permanent Link: http://ufdc.ufl.edu/UFE0044891/00001

Material Information

Title: Amorphization and Solid Phase Epitaxial Growth of Germanium
Physical Description: 1 online resource (160 p.)
Language: english
Creator: Darby, Blake L
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2012

Subjects

Subjects / Keywords: amorphization -- cmos -- germanium -- implantation -- solid-phase-epitaxy
Materials Science and Engineering -- Dissertations, Academic -- UF
Genre: Materials Science and Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The substrate orientation dependence on SPEG was studied for the first time in Ge.  The velocity showed a strong dependence on substrate orientation and the 001 direction displayed a velocity 16 times greater than the 111 direction.  Cross sectional (XTEM) and plan view transmission electron microscopy (PTEM) revealed stacking fault and twin defect formation in the 111 orientation with densities ~1×108 cm-2, where all other orientations showed only hairpin dislocations with densities of ~3×108 cm-2. Unlike Si, Ge had considerably less{111} defects upon SPEG, which contributed to higher normalized velocities as afunction of orientation. Multidimensional (2D) SPEG was studied for the first time in Ge.  Pattern induced stress was found to promote mask edge defect formation, which became more pronounced with decreasing pattern width.  The 1D substrate orientation dependence on SPEG was then used to construct a model for 2D SPEG using the Florida Level-set Object Oriented Process Simulator (FLOOPS).  Mask edge defect formation was accurately simulated in FLOOPS by using a curvature factor of 8×10-8cm.  The evolution of the a/c interface matched well with simulations for both convex and concave interfaces. 2D SPEG was also studied for non-planar Ge substrates.  The effect of a free surface was studied for trench structures by passivating the surface with an oxide.  The presence of an oxide hindered the SPEG process by forcing the Ge to template off the SiO2.  These defects were effectively eliminated by creating a free surface via an HF etch. Ge was found to be less defective than its Si counterpart and the results show promise for using Ge in non-planar finFET structures.Overall, the SPEG process in Ge was shown to be less defective than Si in terms of mask edge defects, {111} defects, trench corner stacking faults, and EOR; however, a high dose regime exists where a porous layer forms within the amorphous Ge. This suggested that implant conditions should be chosen to avoid this high dose regime.  All of these studies lead to a much better understanding of ion implantation and SPEG in Ge.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Blake L Darby.
Thesis: Thesis (Ph.D.)--University of Florida, 2012.
Local: Adviser: Jones, Kevin S.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2012
System ID: UFE0044891:00001

Permanent Link: http://ufdc.ufl.edu/UFE0044891/00001

Material Information

Title: Amorphization and Solid Phase Epitaxial Growth of Germanium
Physical Description: 1 online resource (160 p.)
Language: english
Creator: Darby, Blake L
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2012

Subjects

Subjects / Keywords: amorphization -- cmos -- germanium -- implantation -- solid-phase-epitaxy
Materials Science and Engineering -- Dissertations, Academic -- UF
Genre: Materials Science and Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: The substrate orientation dependence on SPEG was studied for the first time in Ge.  The velocity showed a strong dependence on substrate orientation and the 001 direction displayed a velocity 16 times greater than the 111 direction.  Cross sectional (XTEM) and plan view transmission electron microscopy (PTEM) revealed stacking fault and twin defect formation in the 111 orientation with densities ~1×108 cm-2, where all other orientations showed only hairpin dislocations with densities of ~3×108 cm-2. Unlike Si, Ge had considerably less{111} defects upon SPEG, which contributed to higher normalized velocities as afunction of orientation. Multidimensional (2D) SPEG was studied for the first time in Ge.  Pattern induced stress was found to promote mask edge defect formation, which became more pronounced with decreasing pattern width.  The 1D substrate orientation dependence on SPEG was then used to construct a model for 2D SPEG using the Florida Level-set Object Oriented Process Simulator (FLOOPS).  Mask edge defect formation was accurately simulated in FLOOPS by using a curvature factor of 8×10-8cm.  The evolution of the a/c interface matched well with simulations for both convex and concave interfaces. 2D SPEG was also studied for non-planar Ge substrates.  The effect of a free surface was studied for trench structures by passivating the surface with an oxide.  The presence of an oxide hindered the SPEG process by forcing the Ge to template off the SiO2.  These defects were effectively eliminated by creating a free surface via an HF etch. Ge was found to be less defective than its Si counterpart and the results show promise for using Ge in non-planar finFET structures.Overall, the SPEG process in Ge was shown to be less defective than Si in terms of mask edge defects, {111} defects, trench corner stacking faults, and EOR; however, a high dose regime exists where a porous layer forms within the amorphous Ge. This suggested that implant conditions should be chosen to avoid this high dose regime.  All of these studies lead to a much better understanding of ion implantation and SPEG in Ge.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Blake L Darby.
Thesis: Thesis (Ph.D.)--University of Florida, 2012.
Local: Adviser: Jones, Kevin S.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2012
System ID: UFE0044891:00001


This item has the following downloads:


Full Text

PAGE 1

1 AMORPHIZATION AND SOLID PHASE EPITAXIAL GROWTH OF GERMANIUM By BLAKE LEONARDI DARBY A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE O F DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2012

PAGE 2

2 2012 Blake Leonardi Darby

PAGE 3

3 To my family and friends who have made this dissertation possible

PAGE 4

4 ACKNOWLEDGMENTS I would like to acknowledge my family for stressing the importance of a n education and always being there for me. I would also like to thank my close friends for offering support and making my graduate experience an enjoyable one. This dissertation would not have been possible without these people. I also want to thank Lud ovic Godet for hiring me as an intern at Varian Semiconductor Equipment Associates (VSEA) and giving me freedom to pursue my own ideas in research and development. Ludo always took the time to answer my questions and gave me extraordinary exposure within the company. I would also like to acknowledge Alex Kontos at VSEA and Rob Elliman at Australian National University for doing the implants for this research. I would also like to thank my advisor Kevin Jones and my committee members. I originally came to knew of any openings for part time research, he immediately offered me a position to join his group. I am also grateful for his trust in me and will ingness to allow me to pursue research inte rests, even if they were slightly off topic. Last, but not least I would like to acknowledge the past and present SWAMP group members: Brad Yates, Nick Rudawski, Ray Holtzworth, Nick Vito, Aaron Li nd, Pat Whiting, Sidan Jin, Dan Gostovic Ashish Kumar, and Saurabh Morarka This group has provided numerous thoughtful discussions and created a n environment that constantly pushed one another to learn new things

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 3 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ .......... 9 LIST OF ABBREVIATIONS ................................ ................................ ........................... 13 ABSTRACT ................................ ................................ ................................ ................... 14 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ ..... 16 1.1 Motivation and Technological Relevance of Ge SPEG ................................ .... 16 1.2 Objective and Statement of Research ................................ ............................. 17 1.3 Ion Induced Amorphization ................................ ................................ .............. 17 1.4 I ntroduction to the SPEG Process ................................ ................................ ... 19 1.5 Temperature Dependence of SPEG ................................ ................................ 21 1.6 Dopant and Composition Dependence on SPEG ................................ ............ 21 1.6.1 Electrically Active Impurities ................................ ................................ ... 21 1.6.2 Electrically Inactive Impurities ................................ ................................ 22 1.6.3 Composition Dependence ................................ ................................ ...... 23 1.7 Stress Dependence of SPEG ................................ ................................ .......... 24 1.7.1 Hydrostatic Stress Dependence ................................ ............................. 24 1.7.2 Uniaxial Stress Dependence ................................ ................................ .. 24 1.8 The Effect of /c Interface Roughness ................................ ............................. 26 1.9 Orientation Dependence of SPEG ................................ ................................ ... 27 1.10 Effect of Implant Conditions ................................ ................................ ........... 29 1.11 SPEG Mechanisms ................................ ................................ ........................ 30 1.12 SPEG for Patterned Si Substrates ................................ ................................ 31 1.12.1 Pinned Interface ................................ ................................ ................... 31 1.12.2 Unpinned Interface ................................ ................................ ............... 33 1.12.3 SPEG for Non planar Surfaces in Si ................................ ..................... 33 1.13 Defect Formation Upon SPEG ................................ ................................ ....... 35 1.13.1 End of Range Defects ................................ ................................ .......... 35 1.13.2 Regrowth Related Defects ................................ ................................ .... 36 2 EXPERIMENTAL AND SIMULATION TECHNIQUES ................................ ............. 43 2.1 Material Processing ................................ ................................ ......................... 43 2.1.1 Electron Beam Lithography ................................ ................................ .... 43 2.1.2 Reactive Ion Etching ................................ ................................ ............... 44 2.1.3 Temperature Calibration and Annealing ................................ ................. 44 2.1.4 XTEM Sample Preparation ................................ ................................ ..... 45

PAGE 6

6 2.1.5 PTEM Sample Preparation ................................ ................................ ..... 45 2.2 Material Characterization ................................ ................................ ................. 46 2.2.1 Transmission Electron Microscopy ................................ ......................... 46 2.2.2 Amorphous Layer Measurements and Velocity Calculations .................. 48 2.2.3 Scanning Electron Microscopy ................................ ............................... 48 2.2.4 Atomic Force Microscopy ................................ ................................ ....... 49 2.2.5 Tencor Stress Measurement Technique ................................ ................. 50 2.2.6 Raman Spectroscopy ................................ ................................ ............. 51 2.3 Simulation Techniques ................................ ................................ ..................... 52 2.3.1 Level Set Methods and FLOOPS ................................ ........................... 52 3 AMORPHIZATION OF GERMANIUM AND SUBSTRATE ORIENTATION DEPENDENCE ON SPEG ................................ ................................ ...................... 59 3.1 Amorphization and TDD of G ermanium ................................ ........................... 59 3.2 Substrate Orientation Dependence of Ge ................................ ........................ 60 3.2.1 Introduction ................................ ................................ ............................. 60 3.2.2 Experimental ................................ ................................ .......................... 61 3.2.3 Results ................................ ................................ ................................ ... 62 3.2.4 Discussion ................................ ................................ .............................. 66 3. 2.5 Conclusion ................................ ................................ .............................. 67 4 MULTIDIMENSIONAL SPEG AND FLOOPS SIMULATIONS OF PATTERNED GERMANIUM ................................ ................................ ................................ ......... 75 4.1 Introduction ................................ ................................ ................................ ...... 75 4.2 Experimental ................................ ................................ ................................ .... 77 4.2.1 Sample Processing ................................ ................................ ................. 77 4.2.2 Wafer Curvature Experiment ................................ ................................ .. 78 4.2.3 Raman Spectroscopy Experiment ................................ .......................... 79 4.2.4 Implementation of Level Set Methods ................................ .................... 79 4.3 Results and Discussion ................................ ................................ .................... 80 4.3.1 Pinned Interface Structure ................................ ................................ ...... 80 4.3.2 Unpinned Interface Structure ................................ ................................ .. 83 4.3.3 Comparison of Si and Ge ................................ ................................ ....... 84 4.3.4 Wafer Curvature Measurement Data ................................ ...................... 85 4.3.5 Raman S pectroscopy Data ................................ ................................ ..... 86 4.4 Conclusions ................................ ................................ ................................ ..... 87 5 SPEG WITH NON PLANAR SURFACES AND TRENCH EDGE DEFECT FORMATION IN GERMANIUM ................................ ................................ .............. 98 5.1 Introduction ................................ ................................ ................................ ...... 98 5.2 Experimental ................................ ................................ ................................ .... 99 5.3 Results ................................ ................................ ................................ ........... 100 5.4 Discussion ................................ ................................ ................................ ..... 101 5.5 Conclusions ................................ ................................ ................................ ... 102

PAGE 7

7 6 THE INFLUENCE OFIMPLANT ENERGY AND DOSE ON SPEG ....................... 107 6.1 Introduction ................................ ................................ ................................ .... 107 6.2 Experimental ................................ ................................ ................................ .. 107 6.3 Results ................................ ................................ ................................ ........... 108 6.4 Discussion ................................ ................................ ................................ ..... 110 6.5 Conclusion ................................ ................................ ................................ ..... 112 7 HIGH DOSE SELF ION IMPLANTATION IN GERMANIUM ................................ .. 115 7.1 Mechanisms and Proposed Theories for Porous Formation in Bulk Ge ......... 115 7.1.1 Introduction ................................ ................................ ........................... 115 7.1.2 Experimental ................................ ................................ ........................ 116 7.1.3 Results ................................ ................................ ................................ 117 7.1.4 Discussion ................................ ................................ ............................ 119 7.1.5 Conclusions ................................ ................................ .......................... 121 7.2 High Dose Ion Implantation in Sputtered and Evaporated Ge ....................... 121 7.2.1 Introduction ................................ ................................ ........................... 121 7.2.2 Experimental ................................ ................................ ........................ 122 7.2.3 Results ................................ ................................ ................................ 122 7.2.4 Discussion ................................ ................................ ............................ 124 7.2.5 Conclusions ................................ ................................ .......................... 126 8 SUMMARY AND FUTURE WORK ................................ ................................ ........ 134 8.1 Overview of Results ................................ ................................ ....................... 134 8.2 Future Work ................................ ................................ ................................ ... 136 APPENDIX: FLOOPS SCRIPT FOR PATTERNED STRUCTURES ........................... 138 LIS T OF REFERENCES ................................ ................................ ............................. 140 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 160

PAGE 8

8 LIST OF TABLES Table page 3 1 Amorphous layer depth measured by XTEM along with critical damage and TDD calculated using SRIM. ................................ ................................ ............... 60

PAGE 9

9 LIST OF FIGURES Figure page 1 1 Schematic of an ion implanter. ................................ ................................ ........... 39 1 2 A schematic of the a/c interface evolving over time in the SPEG process. ......... 39 1 3 Schematic showing the number of amorphous atoms needed to attach to 2 crystalline atoms at different orientations. ................................ .......................... 40 1 4 Plot of regrowth vs. substrate orientation for Si measured at 550 C ................ 40 1 5 A WBDF XT EM image of EOR defects in Ge. ................................ ................... 41 1 6 XTEM images of hairpi n dislocation fo rmation in Ge. ................................ ........ 41 1 7 PTEM images of stacking fault defects in (111) Ge ................................ ........... 42 2 1 Schematic of a RIE chamber. ................................ ................................ ............. 54 2 2 Schematic of the tube furnace used for annealing experiments. ........................ 54 2 3 SEM images at 52 of the FIB procedure for XTEM sample preparation ........... 55 2 4 SEM and ion beam images at 52 of the FIB procedur e for PTEM sample preparation. ................................ ................................ ................................ ....... 56 2 5 Schematic of a TEM column. ................................ ................................ .............. 57 2 6 Schematic of the Tencor film stress measurement system. ............................... 58 3 1 Photograph of the 2 polishing stubs used for obtaining different Ge orientations. ................................ ................................ ................................ ........ 69 3 2 XTEM micrographs of an annealing sequence at 330 C of the 0 (001) Ge orientation. ................................ ................................ ................................ .......... 69 3 3 C for the 3 main orientations in Ge. ................................ ................................ ............. 70 3 4 Measured SPEG velocities for Ge at different orientations at 330 C. ................. 70 3 5 SPEG velocities for Ge and Si norm alized to the [001] direction. ...................... 71 3 6 XTEM images for an isochronal anneal of all 8 orientations done at 330 C for 11 hrs. ................................ ................................ ................................ ................ 71

PAGE 10

10 3 7 PTEM images of all 8 orientations taken un der various 2 beam conditions. ...... 72 3 8 PTEM images of (111) Ge a nnealed at 330 C for 7.5 days taken at different zone axes ................................ ................................ ................................ ........... 72 3 9 PTEM images of (111) Ge annealed at 330 C for 7.5 days taken under different g vectors.. ................................ ................................ ............................. 73 3 10 XTEM images of (111) Ge annealed at 330C for 7.5 days showing inclined twin defects .. ................................ ................................ ................................ ....... 74 3 11 XTEM images of (111) Ge annealed at 330 C for 85 hrs showing defects paralle to growth interface ................................ ................................ ................. 74 4 1 Schematic of wafer processing for multidirectional growth ............................... 89 4 2 XTEM image of the continuous 165 nm Si 3 N 4 film used for wafer curvature measureme nts. ................................ ................................ ................................ ... 90 4 3 Plan view SEM image of the 3 nitride pattern s for Raman measurements. ........ 91 4 4 The Ge orientation dependence on SPEG velocities normalized to the [001] direction (0) with a fourth order polynomial fit. ................................ .................. 92 4 5 XTEM images of nitride stressed patterned Ge implanted with 90 keV 510 14 Ge + /cm 2 and annealed at 330C with corresponding FLOOPS simulations ...... 92 4 6 XTEM images of unstressed patterned Ge implanted with 90 keV 510 14 Ge + /cm 2 and annealed at 330C w ith corresponding FLOOPS simulations ..... 93 4 7 XTEM images of patterned Ge implanted with 300 keV 510 14 Ge + /cm 2 and annealed at 330C w ith corresponding FLOOPS simulations ........................... 93 4 8 FLOOPS simulations for pinned inte rface structures for Si at 500 C and Ge at 330 C ................................ ................................ ................................ ............. 94 4 9 XTEM images for unpinned interface structures in Ge and Si ............................ 94 4 10 Evolution of residual tensile stress in the Si 3 N 4 film upon annea ling at 330 C as calculated from wafer curvature measurements. ................................ .......... 95 4 11 Raman spectroscopy data showing the unstressed peak near 300.7 cm 1 and a progressive shift for smaller nitride line widths ................................ ............... 96 4 12 The effect of line width on substrate stress calculated from equation 4 5. ......... 97 4 13 XTEM images of different nitride pattern widths implanted with 90 keV, 510 14 Ge/cm 2 and annealed at 330 C for 335 minutes ................................ ... 97

PAGE 11

11 5 1 Schematic of a finFET device structure. ................................ ........................... 103 5 2 A schematic of wafer processing for the trench Ge structures. ........................ 104 5 3 An XTEM annealing sequence for the uncoated trench structure at 330 C. .... 104 5 4 XTEM images of the trench structures annealed at 400 C for 1 d ay. ............... 105 5 5 High resolution XTEM image o f the stacking faults for the SiO 2 coated sample. ................................ ................................ ................................ ............. 106 5 6 Si trench structure pinned against an SiO 2 layer. ................................ ............ 106 6 1 XTEM micrographs of the solid phase epitaxial growth process at 330 C of (001) Ge self implanted at 90 keV to a dose of 2.010 15 cm 2 .......................... 113 6 2 XTEM and PTEM images of (001) Ge self implanted at 90 keV to a dose of 2.010 15 cm 2 ................................ ................................ ................................ ... 113 6 3 Amorphous layer thickness versus anneal ing time behavior at 330 C of self implanted (001) Ge. ................................ ................................ .......................... 114 6 4 The solid phase epitaxial growth velocity at 330 C of self implanted (001) Ge. ................................ ................................ ................................ .................... 114 7 1 Ge thickness and depth of peak vacancy concentration (R d ) as determined by simulations. ........ 127 7 2 XTEM micrographs illustrati ng the evolution of porous Ge with dose at 130 keV and corresponding plan view SEM micrographs. ................................ ...... 128 7 3 AFM dose sequence showing the change in surface morphology as a function of dose ................................ ................................ ................................ 128 7 4 Graph of RMS roughness vs. implant energy measured by AFM. Dose was kept constant at 210 15 Ge + /cm 2 ................................ ................................ ..... 129 7 5 Graph of RMS rough ness vs. dose measured by AFM. ................................ .. 129 7 6 Implant damage map for self implants into Ge. ................................ ................ 130 7 7 XTEM images of a 130 keV 110 16 Ge + /cm 2 implant into evaporated and sputtered Ge with zoomed in images of the evaporated and sputtered film microstructures ................................ ................................ ................................ 131 7 8 XTEM micrographs illustrating the evolution of sputtered porous Ge with dose at 130 keV and corresponding plan view SEM micrographs. ................... 132

PAGE 12

12 7 9 Pore diameter histograms for c rystalline (001) Ge, sputtered Ge, and evaporated Ge for doses ranging from 1.010 16 1.010 17 Ge + /cm 2 with a constant implant energy of 130 keV. ................................ ................................ 132 7 10 Porous formation schematic for crystalli ne and evaporated Ge. ...................... 133 7 11 Porous formation schematic for sputtered Ge. ................................ ................. 133

PAGE 13

13 LIST OF ABBREVIATION S Ge amorphous germanium amorphous crystalline interfa ce AFM atomic force microscopy BF brightfield c Ge crystalline germanium CMOS complimentary metal oxide semiconductor DF darkfield E beam electron beam EOR end of range FFT fast fourier transform FLOOPS florida level set object oriented process simulator P TEM plan view transmission electron microscopy RIE reactive ion etching SAD selected area diffraction SEM scanning electron microscopy SPEG s olid phase epitaxial growth. SRIM the stopping and range of ions in matter TDD threshold damage density TEM trans mission electron microscopy WBDF weak beam dark field XTEM cross sectional transmission electron microscopy

PAGE 14

14 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements f or the Degree of Doctor of Philosophy AMORPHIZATION AND SOLID PHASE EPITAXIAL GROWTH OF GERMANIUM By Blake Darby December 2012 Chair: Kevin Jones Major: Materials Science and Engineering The substrate orientation dependence on SPEG was studied for the first time in Ge The velocity showed a strong dependence on substrate orientation and the [001] direction displayed a velocity 16 times greater than the [111] direction. Cross sectional (XTEM) and plan view transmission electron microscopy (PTEM) reveal ed stacking fault and twin defect formation in the [111] orientation with densities ~1 10 8 cm 2 where all other orientations showed only hairpin dislocations with densities of ~3 10 8 cm 2 Unlike Si, Ge had considerably less {111} defects upon SPEG, whic h contributed to higher normalized velocities as a function of orientation. Multidimensional (2D) SPEG was studied for the first time in Ge. Pattern induced stress was found to promote mask edge defect formation which became more pronounced with decreasi ng pattern width The 1D substrate orientation dependence on SPEG was then used to construct a model for 2D SPEG using the Florida Level set Object Oriented Process Simulator (FLOOPS) Mask edge defect formation was accurately simulated in FLOOPS by usin g a curvature factor of 8 10 8 cm The evolution of the /c interface matched well with simulations for both convex and concave interfaces

PAGE 15

15 2D SPE G was also studied for non planar Ge substrates. The effect of a free surface was studied for trench structures by passivating the surface with an oxide. The presen ce of an oxide hindered the SPEG process by forcing the Ge to template off the SiO 2 These defects were effectively eliminated by creating a free surface via an HF etch. Ge was found to be less defective than its Si counterpart and the results show promi se for using Ge in non planar f inFET structures. Overall, the SPEG process in Ge was shown to be less defective than Si in terms of mask edge defects, {111} defects, trench corner stacking faults, and EOR; however, a high dose regime exists where a porous layer forms within the amorphous Ge. This suggeste d that implant conditions should be chosen to avoid this high dose regime. All of these studies lead to a much better understanding of ion implantation and SPEG in Ge.

PAGE 16

16 CHAPTER 1 INTRODUCTION 1.1 Motivation and Technological Relevance of Ge SPEG The semiconductor industry has made enormous strides in order to keep pace s roughly every 2 years [ 1 ] In order to achieve this goal, companies like Intel have scaled down device lengths, changed device geometries, and implemented stress to achieve the desired electronic properties. The predominant substrate material, Si, is currently reaching its fundamental limits and researchers are turning to new materials, like Ge for higher performance. Ge has superior electronic properties, such as higher electron and hole mobilities, and als o allows for less aggressive annealing due to its lower recrystallization temperature. With this new interest in Ge for CMOS devices, it is important to understand the defects that form from ion implantation and the solid phase epitaxial growth (SPEG) pro cess. Although the first transistor was made from Ge in 1954, researchers quickly abandoned the material and started using Si due to its beneficial oxide properties and low cost. Recently, there has been a push to return to Ge as an alternative source d rain material in CMOS devices. Ge has many beneficial properties such as higher electron and hole mobilities than Si. Ge also has a lower melting temperature, which would allow for less aggressive annealing steps. Moreover, the low diffusivity of B in G e makes the creation of ultra shallow junctions in Ge realizable [ 2 ] It has also been shown that high electrical activation of B can be achieved with Ge substrates that have been pre amorphized [ 3 ] The need for a pre amorphization step warrants research on

PAGE 17

17 the SPEG process for Ge. The SPEG process has been studied in great detail for Si substrates, but basic knowledge of the SPEG process for Ge remains unknown. 1.2 Objective and Statement of Research The goal of this work is to explore SPEG and defect formation for ion implantation in Ge More specifically, the following aspects have been studied. The substrate orientation dependenc e on Ge SPEG and subsequent defect formation Multidimensional SPEG around patterned structures and the effect of silicon nitride induced stress on defect formation for patterned Ge structures Mult idimensional SPEG around non planar surfaces and t he effect of a free surface during SPEG on the formation of defects in trench structures The effect of high dose implantation in Ge and proposed theory a nd models 1.3 Ion Induced Amorphization Ion implantation is one of the most common methods for introducing dopant atoms into semiconductor materials It is a process by which ions are accelerated at a semiconductor target and come to rest in interstitial sites within the lattice of the material. The main advantages include precise control over the depth and amount of dopants uniformity, and reproducibility The depth is controlled through implant energy whereas the amount of dopant ions is controlled through the dose. Once the ions enter the lattice, they are slowed by both electronic and nuclear stopping mechanisms. Electronic stopping refers to the interaction of the incoming ion

PAGE 18

18 with the electron cloud of the target material and can be modeled wi th the following equation: (1 1) w here is the electronic stopping power as a function of energy, is a variable associated with the ion and target materials, and is the energy of the incoming ion. The nuclear stopp ing mechanism refers to the interaction of the incoming ion with the atomic nucleus of the target material and can be modeled with the following equation: (1 2) where is the nuclear stopping power as a function of energy, is the ion atomic number, is the target material atomic number, is the ion mass, and is the target material mass. The nuclear stopping mechanism dominates for lo w energy implants and implants with high ion masses, while the electronic stopping mechanism dominates for high energy (MeV) implants and implants with low ion masses. Both electronic and nuclear stopping mechanisms need to be accounted for when determini ng the total energy loss as a function of depth into the sample. This can be modeled in the following equation: (1 3) where is the atomic density of the target material. This combination of nuclear and electronic stopping is what determines the range or depth that the ions travel into the material. The ion projected range is often referred to as R p which follows a Gaussian distribution. As the ion loses energy through nuclear and electronic stopping, it

PAGE 19

19 transfers some of this energy to the lattice of the target material causing Frenkel pair defects. A Frenkel pair refer s to a pair of interstitial and vacancy point defects and is created when a crystalline atom is displaced off its lattice site. For this work, it is assumed that 15 eV of energy is required to displace a Ge atom off its lattice site. The distribution of vacancies created also follows a Gaussian type and is referred to as R d Once a critical concentration of point defects is created, the semiconductor material turns amorphous during a first order phase transformation. This critical density is dependent on the ion mass, dose, implant energy, and temperature of the substrate. In this work, the critical density is referred to as the threshold damage density. Oftentimes, the target material is pre amorphized with a heavy ion species prior to implanting with dopant ions. This effectively reduces channeling and has been shown to increase electrical activation upon SPEG. 1.4 Introduction t o t he S PEG P rocess Once a critical amount of damage is created through ion implantation, an amorphous layer is created thr ough a first order phase transformation. The damage can be recovered through a process called solid phase epitaxial growth (SPEG) where the amorphous material adopts the crystalline structure of the underlying seed crystal [ 4 ] This occurs on the atomistic level through the nu cleation of an atomic ledge, followed by the lateral propagation of this ledge [ 5 ] On a local scale, this growth interface is expected to be atomically sharp. SPEG is different than vapor or liquid phase epitaxy in that solid phase epitaxy invol ves a solid solid growth interface. This means that SPEG is no t concerned with the flux of atoms to the /c interface and surface diffusion does not play a role in the growth process [ 6 ] SPEG is also quite diffe rent that other solid solid crystallization models,

PAGE 20

20 such as grain growth in that SPEG is limited in one dimension. In classical nucleation and growth, grains take on preferred shapes due to differences in surface energy. This brings multiple grains and crystallographic orientations into play. 1D SPEG is limited to only one crystallographic orientation, so the growth is limited only by the geometry of the sample and not surface energy. SPEG is a thermally activated process in which the amorphous layer re crystallizes from the /c interface up to the surface as seen in Fig ure 1 2 SPEG for Si has an activation energy of 2.7 eV which is significantly higher than the 2.17 eV reported for Ge [ 7 11 ] Consequently, the recrystallization temperatures for Si are typically much higher than Ge. SPEG is technologically important in that it is one of the most common methods to achieve electrical activation of dopant atoms in se miconductors. In this process, a pre amorphization implant is typically used to create a continuous amorphous layer. This is achieved through heavy ion implantation, such as Ge + This allows for reduced channeling during the dopant implant, which is adv antageous for the realization of ultra shallow junctions. It is also well understood that dopant atoms transition from interstitial to substitutional sites during the SPEG process, resulting in electrical activation. SPEG effectively increases the activa tion percentage for Si and Ge, which further increases the value of this process. Many factors can influence the SPEG rate, such as stress, temperature, crystallographic orientation, electrically active impurities, and electrically inactive impurities. T he following sections will discuss the dependencies most important for this body of work.

PAGE 21

21 1.5 Temperature Dependence o f SPEG The SPEG velocity increases with temperature through the following Arrhenius relationship: (1 4) where is a temperature independent prefactor on the order of 1 10 7 m/s = 2.17 eV is the activation energy for Ge =8.62 10 5 is the absolute temperature [ 11 15 ] The activation energy and velocity prefactor can be found from the slope and y intercept of the Arrhenius line. 1.6 Dopant and C omposition D ependence on SPEG 1.6 .1 E lectricall y A ctive I mpurities The effect of dopa nt atoms on SPEG velocity depends on whether the impurity is electrically active or inactive. Electrically active dopants like B, Al, As, and P tend to increase SPEG velocity in Si and Ge, which has been explained by a Fermi level shifting model [ 11 16 22 ] This model relates structural changes at the interface to the Fermi level. It follows the assumption that the ratio of charged to uncharged kink sites present at the /c interface is Fermi level dependent. If we consider a doped semiconductor, then we can express the enhancement in velocity, as: (1 5) where is the intrinsic velocity, is the concentration of dopant, is the intrinsic carrier concentration at T, is a degeneracy factor, is the Fermi level of the intrinsic material, while is the energy level at the /c interface from the charged site. One can see that a greater shift away from the intrinsic Fermi level results in a larger increase in SPEG velocity.

PAGE 22

22 SPEG rate enhancement in Ge does not occur until 5 10 19 cm 3 for As, 1 10 19 cm 3 for Al, and 2 10 19 cm 3 where typical enhancements a re on the order of 1.5 5 x [ 11 17 ] A compensating effect was found for Ge regions which contained equal concentrations of n and p type dopants. Ge SPEG was enhanced for P concentrations between 1 10 1 8 5 10 19 cm 3 while a SPEG reduction occurred above 4 10 20 cm 3 due to supersaturation and P precipitation [ 23 ] Ho et al. studied the effect of dopant enhanced regrowth for different crystallographic orientations an d observed a constant enhancement factor among the [100], [110], and [111] directions for both B and P type dopants [ 24 ] While a tenfold enhancement was observed, no TEM analysis was performed to confirm whether crystalline quality was improved from the dopants. The effect of electrically active dopants was also studied for patterned Si SPEG. It was determined that dopants enhanced SPEG v elocity in an isotropic fashion, which confirms that the interface curvature effect is independent of the electronic effect that controls the enhancement in SPEG velocity [ 25 ] 1. 6 .2 E lectrically Inactive I mpurities Some common electrically inactive impurities that affect SPEG include H O and C. Atomic H can be introduced at the surface from the oxidation reaction or from water vapor in the ambient [ 11 26 ] Once introduced, H can quickly diffuse through the amorphous layer to the /c interface. It is believed that H segreg ates on the amorphous side of the /c interface during SPEG, causing a reduction in velocity by passivating crystallization sites [ 10 11 27 ] However, the presence of H is believed to be less pronounced in Ge relative to Si since H diffuses much slower in Ge and Ge does not

PAGE 23

23 form a stable oxide. Nonetheless, a con centration of 6 10 18 H/cm 3 has been observed to cause a 3.7 fold reduction in SPEG velocity in Ge [ 11 ] The presence of O and C impurities can retard SPEG velocities in Si and G e [ 28 ] Since the Fermi level shifting model does not apply to electrically inactive impurities, it is assumed that these impurities negatively affect the migration kinetics since additional energy is needed to incorporate an impurity atom in non su bstitutional sites within the lattice [ 29 30 ] C and O impurities in Si have shown to promote the formation of twins on {111} planes, and was found to be highly dependent on solute concentration. Impurities were found to affect the interface stabili ty, leading to non planar /c interfaces and twin formation [ 31 ] 1. 6 .3 C omposition D ependence SPEG rate measurements have been conducted using time resolved reflectivity (TRR) of SiGe alloys of varying composi tion. Arrhenius plots showed a decrease in activation energy from 2.64 2.19 eV for increasing Ge content [ 14 ] Interestingly, a linear relationship was not observed between the activation energy and composition [ 32 33 ] This was explained in terms of 2 components which affect activation energy: a bond breaking energy and a bond reorientation energy. I t was concluded that variation s in the reorientation energy was non linear with composition due to varying amounts of strain in the lattice. The activation energy reached a maximum for alloys near Si 0.8 Ge 0.2 compositions before decreasing down to 2.17 eV for increasing Ge content Oth er impurities such as Ni or In have shown to increase SPEG in Si by more than a factor of 300 [ 34 36 ] This was attributed to metal diffusion to the a/c interfa ce forming nuclei for SPEG to take place

PAGE 24

24 1.7 Stress D ependence of SPEG 1.7 .1 H ydrostatic S tress D ependence The effect of hydrostatic stress on SPEG velocity has been studied by several authors using a high pressure diamond anvil cell [ 15 37 43 ] For both Si and Ge, the application of hydrostatic stress in the GPa range increases the velocity of SPEG. T his enhancement ca n be explained thermodynamically since the SPEG process is characterized by a negative activation volume. This means that the transition from the amorphous state to the activated state is characterized by a negative volume change. This volume change is i ndependent of dopant concentration and dopant type. The negative sign implies that the velocity is enhanced by the application of pressure which can be up to 2 orders of magnitude for ~5 GPa of stress. The stress dependence on SPEG velocity can be expre ssed through the following equation: (1 6) where is the pre exponential, is the activation energy, is pressure, and is the activation volume [ 15 ] From this equation, one can mathematically see how a negative activation volume would be enhanced by pressure. Aziz et al. have measured the activation volume for Si and Ge to be 0.28 and 0.45 respectively where is the crystalline atomic volume The fact that Ge has a more negative activation volume than Si implies that the application of pressure has a larger effect on SPEG kinetics for Ge. 1.7 .2 U niaxial S tress D epen dence In contrast to hydrostatic stress, i t is well known that the application of uniaxial stress can cause drastically different changes in the SPEG velocity Carter and Aziz conducted 3 point bending experiments to study the effect of non hydrostatic st ress

PAGE 25

25 on Si SPEG [ 44 46 ] The experiments were done by elastically bending Si with an amorphous layer present, while annealing the sample to measure SPEG rates. The conclusion from the experiments was that in plane tension enhanced SPEG kinetics, while in plane compression slowed SPEG kinetics. Aziz was able to describe stressed SPEG using the following equation: ( 1 7) where is the activation volume tensor and is an arbitrary stress. The expression is similar to the expression for hydrostatic pressure, except the numerator of the exponent is replaced with an activation strain tensor ( The activation volume is given by: The activated state for stressed SPEG was described as expansion in the two in plane directions and a contraction in the direction normal to the surfac e large enough to make the overall volume change negative. This model was expanded upon by Rudawski, who showed finite limits to the velocity under compressive stress [ 47 ] W afer bending experiments for ultra thin Si have shown that compression retards SPEG up to values of 0.5 GPa at which point the velocity plateaus. Interestingly, the application of tensile stress d id not appreciably affect SPEG velocities eve n up to 1.5 GPa of applied stress The relative decrease in SPEG velocity with compression was shown to be independent of annea ling temperature A dual timescale model was proposed for SPEG, which consisted of nucleation of an atomic ledge and propagatio n of the ledge along the interface.

PAGE 26

26 Considering this atomistic model, it was proposed that i n plane compression retard s the velocity by inc reasing the time required for a crystalline island to propagate along that ledge. Uniaxial s tressed SPEG for Ge subs trates remains widely unstudied. The application of in plane tensile stress has shown to reduce threading dislocations in SiGe alloys layers grown in Si by means of a strain stabilization technique [ 48 ] Another study show ed that high in plane stresses can be generated in Ge arising from the incorporation of dopant atoms that differ in size from the host lattice [ 9 ] ; however, the calculated stresses from this experiment were low compared with uniaxially stressed samples. Another study showed that in plane compressive stresses were generated during Kr + implantation into Ge, where the generated stresses increased with dose at a fixed implant energy [ 49 ] The generated st resses led to dramatic swelling of the Ge surface, but this study lacked a SPEG correlation. 1.8 The E ffect of /c I nterface R oughness While planar unstressed SPEG is known to occur with root mean square (rms) roughness of less than 2 nm, stressed SPEG ca n result in increased interface roughness [ 50 ] SPEG was studied in strained SiGe layers and it was determined that rms roughness of the /c interface can increase up to 80 nm as the interface proceeds up to the surface [ 50 ] The SPEG velocity was reduced in areas of increased roughness, which also led to defective growth. It was also demonstrated that B doping can minimize the interface roughness by increasing the number of kink sites available for nucleation. The influence of stress on interface roughness and SPEG was further studied in Si, where it was shown that compressive stress increased /c interface roughness. The

PAGE 27

27 application of tensile stress was similar to the unstressed case, where no interface roughening occurred [ 51 ] When interface rough ness did exist, it was found to increase over time [ 45 52 54 ] This was linked to an local conce ntration of stress near the trough of the /c interface which is less than that of the apex This gives rise to a reduction in local interfacial mobility when under compressive stress. A kinetic model showed that the rate of roughening increased for smaller wavelengths of roughness [ 52 ] Several researchers have theorized that even [100] SPEG is dominated by nanofaceting and regrowth along [111] terraces [ 12 55 56 ] Molecular dynamic simulations at different temperatures have shown that the propensity to facet changes with temperature and could affect interface roughness [ 12 ] While this effect has not been directly correlated with SPEG velocity, it may be responsible for changes in {111} defect formation. Rechtin et al. studied the eff ect of temperature on defect formation in (111) oriented Si. At low temperatures, a slower regrowth velocity and reduced nucleation rate yielded lower number densities of {111} defects. At higher temperatures, the size of the defects parallel to the (111 ) surface was observed to increase [ 57 ] 1.9 Orientation D ependence of SPEG The orientation dependence on SPEG refers to the effect of crystallographic orient ation of the substrate on the velocity. The substrate orientation dependence on Si SPEG was measured by Csepregi annealing implanted wafers cut at different angles away from the (001) [ 58 ] T he Rutherford Bac kscattering (RBS) technique was then used to measure the SPEG ve locities at 550 C The highest SPEG velocity was found to exist for the [001] direction, which was nearly 25 times greater than the [111] and 3

PAGE 28

28 times greater than the [110]. All orientations had the same activation energy, so the variation in velocity derives from the pre exponential factor ( ). A bond rearrangement model was used to explain the orientation dependence [ 16 59 ] An amorphous atom can recrystallize only if it is attached to 2 undistorted crystalline atoms. For diamond cubic crystal structures, there are different numbers of amorpho c) interface that are needed to attach to a crystalline atom with 2 undistorted bonds depending on the orientation. This number is 1, 2, and 3 for [001], [011], and [111], respectively as seen in Figure 1 3 [ 60 ] This gives rise to a minimum velocity at the [111] orientation and a maximum velocity at the [001] orientation. The entire orientation dependence for Si can be seen in Figure 1 4. Linear velocities exist for the [100] and [110] directions, but the [111] direction displayed bimodal growth for Si The slow regime in the first 100 nm was attributed to a higher density o f twin defects, as observed in angled X TEM experiments [ 57 ] The fast regime of the bimodal growth for [111] Si was marked by a lower density of larger twins, but a highly non uniform /c interface. The amount of dechanneling observed in RBS for [111] was much greater than in [100] and [110] oriented samples. The higher amount of dechanneling was believed to be related to the amount of residual defects in the regrown layer. This high concentration of defects was also observed within 16 of the 111 axis, through RBS measurements. These defects were characterized as large twins using electron diffraction in TEM [ 57 ] It has been hypothesized that the presence of defects retard the SPEG velocity near the [111] orientation [ 61 ] and recent kinetic lattice Monte Carlo simula tions have

PAGE 29

29 addressed the influence of defects on velocity [ 60 62 65 ] Substrate orientation mode ls that neglect the influence of defects tend to overestimate experimental velocities [ 61 ] while models that do take into account defect formation match better with experimental velocities [ 60 65 ] Thus far, there are only velocities for 3 different substrate orientations for Ge in the literature [ 59 66 ] Of these velocities, the normalized [111] velocity matches closely with that of Si, but the normalized [110] velocity is nearly double compared with Si. Investigati on of the entire orientation dependence is warranted in this area if Ge will be incorporated into next generation devices. Knowledge of the substrate orientation dependence on SPEG is essential for understanding multidirectional growth and subsequent defe ct formation [ 25 47 67 ] 1.10 Effect of Implant C ondition s The most common implant variables include ion type, dose, dose rate (flux), implant temperature, and implant energy. Ion irradiation usually induces amorphization of the target material, but can also induce recrystallization of the amorphous layer depen ding on sample temperature and ion beam parameters [ 68 70 ] Ion beam induced crystallization is a process by which the amorphous layer can recrystallize upon imp lantation due to the production of point defects at the /c interface typically from a high energy implant [ 71 ] This promotes a dynamic rearrangement of danglin g bonds at the interface, contributing to SPEG. It was shown that regrowth rate decreases with increasing dose rate and increases with increasing ion mass [ 71 ] This implies that ion induced recrystallization occurs by mobile point defects created at the /c interface [ 69 72 75 ] Ion induced r ecrystallization only occurs in a temperature range where the thermal process is inhibited and follows an Arrhenius temperature dependence with an

PAGE 30

30 activation energy of ~ 0.3 eV for Si. The critical temperature is referred to as the reversal temperature (T R ) since amorphization is reversed and crystallization begins. This critical temperature is near 200 C for Si [ 76 ] Thus far, there is limited knowledge as to how implant conditions aff ect SPEG in Ge [ 77 79 ] Ion induced recrystallization is less apparent in Ge, possibly due to its low threshold for amorphization. 1. 11 SPEG M echanisms Several t heories have been proposed for what governs the SPEG process. Suni et al. has proposed a model based on the number of mobile point defects at the /c interface [ 17 ] Based on this correlation, a higher number of vacancies at the /c interface would enhance the bond breaking process, which precedes atomic re arrangement during recrystallization. This argument is substantiated by the fact that the activation energy for epitaxial growth matches well with the energy of vacancy formation for Si and Ge [ 6 17 61 80 81 ] I n this case, the rate of defects diffusing to the interface should change with time or thickness of the amorphous layer, but we know this is not observed experimentally [ 6 ] Presumably, the rate of defect diffusion should be isotropic in amorphous materials and cubic crystals, yet a substrate orientation dependence is observed with SPEG [ 59 66 81 82 ] In addition, it was later calculated that the self diffusion coefficient of Ge is too low to account for the growth rate [ 15 ] Other theories have proposed that kink sites or dangling bonds at the interface control the SPEG process. In this theory, dangling bonds help reconstruct the random network of the amorphous solid into a crystalline network [ 83 ] This theory seems plausible since the number of charged kink sites increases with doping, as does the SPEG rate. The mobility of dangling bonds is also increased with pressure since the transition state

PAGE 31

31 has a lower local volume (negativ e activation volume). The most commonly accepted theory of SPEG is a bond rearrangement at the /c interface since the Si Si bond strength (2.5 eV) corresponds well to the activation energy of SPEG [ 83 84 ] The amorp hous phase is composed of a combination of fivefold and sevenfold rings, which need to conform to a sixfold ring of the diamond crystal substrate. In this theory, bond breaking and bond forming is the process by which sixfold rings are achieved [ 5 ] 1. 12 SPEG for Patterned Si Substrates Although 1D SPEG is often studied in the literature, 2D SPEG is much more applicable to device processing since patterned wafer s are often used. A 2D interface involves regrowth in multiple directions and is therefore more complicated than 1D SPEG. The substrate orientation dependence on SPEG gives rise to an overlap of regrowth fronts, creating what is known as a mask edge defe ct. The following sections will highlight some of the relevant work done for Si substrates so that it can be compared with Ge in the research chapters 1 12.1 P inned I nterface A pinned /c interface is defined as one which has only convex curvature and intersects the surface. Cerva and Kusters were the first to report mask edge defects in pinned interface structures in Si [ 85 ] The origin of the defect was linked with the anisotropy of the substrate orientation dependence, which caused the evolving [001] and [110] fronts to overlap. The defect was noticed to form with in the regrown layer and proceed up the surface. Saenger et al. studied a similar structure in (001) and (011) Si substrates and presented a nanofacet model of SPEG [ 55 86 ] In this model, crystallization proceeds quickly, until the growth is faceted on slow {111} planes. Eventually, mask edge defects form on these {111} planes, which are 54.7 away from

PAGE 32

32 the (00 1) surface and 35.3 away from the (011) surface [ 87 ] This facet model was later supported by lattice kinetic Monte Carlo simulations showing similar evolution [ 63 64 ] The effect of pattern induced stress was studied in (001) Si by Rudawski et. al and was crea ted through successive implants of 1 10 15 Si + /cm 2 at 20 and 60 keV into a silicon nitride patterned sample [ 47 ] It was determined that the presence of the nitri de mask significantly affected the mask edge defect formation. When the nitride was removed prior to annealing, mask edge defects formed within the regrown layers. However, the presence of the nitride significantly reduced the defect formation. This was attributed to a tensile stress created by the nitride. In all cases, a band of EOR defects formed near the original /c interface. Further investigations of this stress effect revealed that compressive stresses enhanced mask edge defect formation, whi le unstressed samples reduced defect formation. Stress was applied to patterned samples by wafer bending during the anneal [ 47 ] Defect formation for the compre ssed samples imp lies that ratio of the [100] to [110] was altered due to the stress. Interestingly, applied tension was found to result in defect suppression, similar to the unstressed case. The influence of dopant enhanced SPEG on mask edge defect form ation was studied at a range of energies for As+ implants in Si [ 88 ] I n theory, one could tailor a dopant implant such that the projected range of the dopant ion is near the nucleation site of the mask edge defect, thus suppressing defect formation. However, it was concluded that such doping effects did not significantly in fluence mask edge defect formation.

PAGE 33

33 1.12 .2 U npinned I nterface An unpinned /c interface is defined as one that does not intersect the surface at any point and is therefore no subjected to surface pinning effects. Planar 1D SPEG was studied with patterned SiN overlayers in Si substrates and a reduction in SPEG velocity was found to occur in areas under the pattern [ 89 ] to account for the reduction in velocity, where the rigidity of the SiN could reduce the mobility of Si at oms in the underlying /c interface. The presence of H released from the SiN pattern during the anneal could have also played a role in the reduction of SPEG velocity [ 26 27 61 ] The recrystallization of unpinned interfaces which have both convex and concave curvature (2D) was recently investigated in Si substrates [ 25 47 67 ] This type of interface was created in Si usi ng successive implants of 20, 60 and 160 keV with doses of 1 10 15 Si + /cm 2 1 10 15 Si + /cm 2 and 3 10 15 Si + /cm 2 Upon annealing, mask edge defects began to form and t wo triangular Si regions remained at the surface due to the slow moving [111] and [110] f ronts This caused the a/c interface to facet on {111} planes during SPEG, causing triangular amorphous regions to remain near the mask edge. In all cases, the evolution of the /c interface was found to be independent of annealing temperature. 1 12 3 S PEG for N on planar S urfaces in Si Non planar surfaces are particularly of interest recently since the advent of the FinFET device. These devices allow for increased scalability since they are designed vertically from the surface. They are often referred gate covers three sides of the device, which allows for reduced threshold voltage and steeper sub threshold slopes [ 90 91 ] One of the central challenges in designing a

PAGE 34

34 FinFET device is reducing parasitic resistance, which can stem from the extension and source/drain regions [ 92 ] One of the ways to reduce this resistance is with angled ion implantation and SPEG [ 93 ] This technology has been explored for Si and is known to cause {111} stacking fault formation in the middle of the fin and polycrystalline growth in the upper reg ions of the fin [ 94 95 ] Similar fin structures have also been studied in Ge and are observed to be far less defe ctive than Si, but stacking fault defects become more prominent as the fin width decreases and devices are scaled down [ 96 97 ] Thus far, no research has explored ways to reduce the stacking fault formation in non planar Ge structures. Several researchers have studied the effect of a free surface on stacking fault formation [ 98 100 ] SPEG was studied for non planar trench structures in Si surrounded by SiO 2 The trenches were implanted with 1 10 15 Si + /cm 2 at 40 keV, which created an Si layer pinned by SiO 2 Upon annealing, triangular Si regions pinned against the corner of the trench due to the slow velocity of the [111] front. The fully recrystallized structure resulted in a region of de fective Si in the corner. When the SiO 2 was removed prior to annealing, the amount of corner defects was reduced The results indicated that the presence of the SiO 2 hindered growth at the point of attachment, forcing the Si to break the Si/SiO 2 bond to complete the SPEG process. Saenger et al studied rectilinear patterned structures in Si and found that trench structures aligned in the [100] direction resulted in less stacking faults in the corner regions than trenches aligned in the [110] directions [ 55 98 ] In addition, a 5 hour anneal at 1325 C further reduced defect formation for the [100] aligned samples. If Ge is to be incorporated in

PAGE 35

35 future F inFET devices, ways to control defect formation is essential in order to optimize the performance. 1. 13 Defect F ormation U pon SPEG Ion implantation introduces an excess population of point defects in the target material. Several different types of defec ts can form during the SPEG process. These defects are categorized based on the type of damage created in the crystal. 1. 13 1 E nd of R ange D efects For amorphizing implants, e nd of range ( EOR ) defects form near or just beyond the original /c interface u pon annealing [ 101 102 ] The source of this type of defect is believed to be excess interstitials that come to res t just beyond the In Si, it is well known that these extrinsic defects can cause device leakage. In addition, transient enhanced diffusion (TED) has been linked to EOR dissolution during annealing [ 103 107 ] TED is the anomalous diffusivity enhancement from dopant atoms interacting with defects. TED is well known to cause electrical deactivation in Si, where boron interstitial clusters are formed from excess intersti tials diffusing from the EOR to the doped region [ 108 109 ] Recently, this same phenomenon was observed for B dopants in Ge [ 110 ] The nature of the defects differs greatly between Si and Ge. EOR damage in Ge is characterized by small 5 10 n m interstitial clusters and dislocation loops while these defects can be orders of magnitude greater in size in Si and can also take the form of {311} defects [ 103 111 112 ] The nature of the defects depends greatly on the implant and annealing conditions used. EOR density decreases with decreasing implant temper ature since less interstitials are able to contribute to EOR formation [ 113 114 ] Likewise, increasing the implant temperature has the opposite effect. It has also been

PAGE 36

36 demonstrated that increasing the implant dose and energy can increase the number of interstitials bound in the EOR region by several orders of magnitude [ 115 ] EOR defects und ergo a process called Ostwald ripening, which causes the defects to decrease in density and increase in size throughout the anneal. An example of EOR damage in Ge after a 1MeV, 210 15 Ge + /cm 2 self implant and 11 minute anneal at 330 C is shown in Fig ure 1 5 The small size of EOR damage in Ge has been linked in part to the limited transport of self interstitials in Ge [ 116 ] Self diffusion in Ge is largely driven by vacancies, w hich makes the formation of large dislocation loops more difficult in Ge. In addition, Ge has a lower threshold for amorphization [ 117 ] which limits the popu lation of interstitials that are able to contribute to EOR. Unlike Si, EOR damage is highly unstable in Ge and can dissolve out at temperatures as low as 400 C [ 110 ] 1. 13 .2 R egrowth R elated D efects Regr owth related defects consist of regrowth related defects which form within the amorphous region. One of th e most common type of regrowth related defects are hairpin or spanner disloca tions Hairpin defects nucleate when the advancing /c interface encounters microcrystalline pocket s at a slightly different orientation than the bulk substrate [ 102 ] Hairpin dislocations are more prominent with a rough interface a s this increases the number of pockets which serve as nucleation sites (Figure 1 6 (a)) Results have shown that low tempera ture (liquid nitrogen) implantation can reduce hairpin formation by creating a smoother /c interface [ 118 ] It has also been theorized that hairpins nucleate from dislocation loops near the /c interface, where the two ends of the truncated loops act as nucleation sites [ 119 ] An example of a hairpin dislocation in Ge is shown in Fig ure 1 6 (b) This type of defect nucleates at the /c interface and

PAGE 37

37 propagates up towards th e su In Si, hairpins were diffraction contrast analysis [ 120 ] Stacking fault defects are another type of regrowth related defect observed for Si and Ge. Stacking faults are 2D defects that are characterized by faulty stacking sequences. An intrinsic stacking fault has a stacking sequence of ABCB CABC, and is formed by an agglomeration of vacancies. An extrinsic stacking fault has a stacking sequence of ABCAABCA, and is formed by an agglomeration of interstitials. At the boundary of the stacking fault is a Shockley partial dislocation, which has vector equal to a/6[112] for FCC crystals. Although the perfect dislocation for FCC is of the [110] type, it is more energetically favorable for this dislocation to split into partials of the [112] type as seen in the following equation. (1 5) Stacking faults are best imaged in a dynamical 2 beam condition in TEM. This results in a 2 fringe system, known as stacking fault fringes. The nature of stacking faults can be determined from the outer thickness fringes following the two beam bright field/displaced aperture dark field imaging procedure of Williams and Carter [ 121 ] where intrinsic stacking faults start with a white fringe and extrinsic stacking faults start wi th a dark fringe for a 220 type g vector In Si and Ge, these defects appear on the close packed (111) planes. An example of an intrinsic stacking fau lt in (111) Ge is shown in Fig ure 1 7 Twin defects are another type of regrowth related defect observed in TEM for Si and Ge [ 60 82 94 96 97 120 122 ] Tw in defects can be thought of as a single

PAGE 38

38 crystal re gion within another single crystal, where the two crystals share common lattice points. The stacking sequence for a twin is AB(C)BACB, where (C) represents the twin boundary between the two regions. In Si and Ge, twinning occurs on (111) planes during SP EG since there is a high probability that one of the three amorphous atoms will crystallize along the ledge in a defective manner [ 57 59 ] Occasionally, twin planes are repeated in a periodic fashion known as microtwins. Twins and microtwins are typically observed through HRTEM, but can also be characterized through electron diffraction. Twinning causes the ap pearance of satellite spots occurring 1/3 the distance between adjacent (111) spots. In Si, it is well known that the presence of stacking faults and microtwins can cause device leakage [ 12 3 ] and high sheet resistance in spreading resistance profiling [ 124 126 ] TEM evidence suggests that these defects are stable even after 20 minutes at 1000 C. I n this regard it is valuable to understand the formation of these defects in Ge. Thus far, there is no documentation of SPEG for [111] Ge and subsequent stacking fault and microtwin defect formation.

PAGE 39

39 Figure 1 1 Schematic of an ion implanter. Figure 1 2 A schematic of the a/c interface evolving over time in the SPEG process.

PAGE 40

40 Figure 1 3 Schematic showing the number of amorphous atoms needed to attach to 2 crystalline atoms at different orientations. Figure 1 4. Plot of regrowth vs. substrate or ientation for Si measured at 550 C [ 59 ]

PAGE 41

41 Figure 1 5 A WBDF XTEM image of EOR defects in Ge. The sample was implanted at 1 MeV with 21 0 15 Ge + /cm 2 and annealed at 330 C for 11 minutes. Figure 1 6 XTEM image s of hairpin dislocation formation in Ge. A morphous and microcrystalline 15 Ge + / cm 2 implant (a) and an XTEM image of (110) Ge implanted at 1MeV, 110 15 Ge + / cm 2 annealed at 330 C for 885 min showing hairpin dislocations nucleating at the a/c interface (b).

PAGE 42

42 Figure 1 7 PTEM images of stacking fault defects in (111) Ge implanted with 110 15 Ge + / cm 2 at 1MeV and annealed at 330 C for 7.5 days. A 2 beam BF image of a n intrinsic stacking fault (a) and an axial DF image of the same stacking fault in (b). bottom and top of the sample, respectively.

PAGE 43

43 CHAPTER 2 EXPERIMENTAL AND SIM ULATION TECHNIQUES The following sections will give background information to some of the most important material processing and characterization techniques used in this body of work. The exact recipes and procedures of these techniques will be elucidated in the experimental sections of subsequent chapters. 2.1 Material Processing 2.1.1 Electron B eam L i thography Electron beam (E beam) lithography is a proc ess that uses a focused beam of electrons to form patterns in a resist material. Unlike traditional photolithography, which is limited by the wavelength of light, E beam lithography is limited only by the forward scattering of electrons. This allows for the creation of much finer patterns with higher precision, but E beam lithography requires a vacuum and more time for processing [ 127 ] T he system typically cons ists of an electron gun which generates the electron beam and an optical column which focuses t he beam with a system of lenses. T he sample surface is the target of the electron beam which contains e ither positive or negative resist. Positive resist softe ns upon exposure, while negative resist h ardens upon exposure. After exposure, the sample is then immersed in a developer solution of MIBK (methyl isobutyl ketone), which exposes the pattern. Despite the high precision of E beam lithography, there are som e disadvantages to the technique. Since an electron beam is required, the system must be held in a vacuum and is expensive. This scanning of the E beam also takes longer than traditional photolithography, making it impractical for high volume production. Another disadvantage of E beam lithography is electron scattering. When electrons enter the

PAGE 44

44 resist, the backscattered electrons spread out and cover a much larger volume than the initial focused beam. When the beam scans 2 areas that are close to one a nother, the overlap of scattered electrons causes unwanted etching, known as the proximity effect. The line widths in this work were spaced several hundred nanometers apart to minimize this effect. 2.1.2 Reactive I on E tching Re active ion etching (RIE) i s a dry etch technique which results in anisotropic etching of the surface The system consists of a cylindrical vacuum chamber with a sample platform in the middle. Gases such as SF 6 CHF 3 and C 4 F 8 are fed in thr ough the top of the chamber where a pla sma is created by a radiofrequency (RF) electromagnetic field The chemically active plasma is then directed at the sample due to charge build up on the plat en and then pumped out through the chamber bottom. As the ions reach the surface of the wafer, th ey not only get directed by an electric field, but also chemically react with the surface This leads to an anisotropic etch of patterned wafers and is advantageous over wet etches, which typically result in isotropic etches Parameters such as gas flow, power, pressure, and time can be adjusted to control the etch. A schematic of a typical RIE chamber can be seen in Fig ure 2 1. 2.1.3 Temperature C alibration and A nnealing To activate the SPEG process, the samples in this study were annealed in a quartz sample boat within a tube furnace. The tube furnace consisted of a 3 foot glass tube with a gas inlet on one end and an end cap on the other. All anneals were done with the sample boat positioned in the center of the furnace as seen in Fig ure 2 2 The temperature was calibrated before each anneal using a thermocouple, which rested on top of the sample boat. To avoid measurement errors, samples within each experiment

PAGE 45

45 were annealed together. Temperatures ranging from 330 400 C were explored in this stud y and the error in all thermocouple readings was estimated to be 1 C 2.1.4 XTEM Sample P reparation A focused ion beam (FIB) was used to prepare site specific samples for cross sectional transmission electron microscopy in this work. The FIB works by a focused beam of Ga + ions accelerated at 30 keV towards the area of interest. A carbon rod evaporator system was used to deposit ~200 nm of C on the sample to protect the surface from Ga + damage. In addition, a ~8 2 1.5 m layer of Pt was deposited in sit u to further protect the sample during preparation as seen in Fig ure 2 3 ( a ) Trenches were milled using the 5000 pA aperture as seen in Fig ure 2 3 ( b ) and then progressively thinned with smaller currents until the sample was ~500 nm thick. At this point, the stage was rotated to 0 and undercut using the 300 pA aperture as seen in Fig ure 2 3 ( c ) Further sample thinning with the 100 pA and 50 pA apertures were used until the sample thickness reached ~100nm as seen in Fig ure 2 3 ( d ) During the final mills the stage was rotated to 51 and 53 to preserve Pt at the surface. At this point, the samples were cut out placed onto a copper grid for imaging. 2.1.5 PTEM S ample P reparation Traditionally, plan view samples are made by a standard polish and etch process, but this method consumes several mm 2 of implanted material. S ince Ge material was in short supply for this research, p lan view transmission electron microscopy samples were prepared using the FIB technique, which only consumes several m 2 of material. In this technique, the bulk Ge sample was mounted edge on so that the implanted surface was facing away from the ion beam. Then, Pt was deposited

PAGE 46

46 near the edge of the implanted surface. Care was taken to deposit the Pt at 51 instead of 52 to avoid Pt damage on the implanted surface. Then, a trench was milled on one side of the sample as seen in Fig ure 2 4 ( a ) Once the sample w as thinned to ~1 m of thickness the stage was rotated to 0 where the undercut and one sidecut were made, as seen in Fig ure 2 4 ( b ) The sample was then welded to an Omniprobe needle using Pt and transferred to a Cu Omniprobe grid as seen in Fig ure 2 4( c ) ( d ) respectively At this point, the samples were continually thinned from the bulk side down to ~100 nm th ickness. 2.2 Material Characterization 2.2.1 T ransmission E lectron M icroscopy TEM is the most frequently used technique for microstructural characterization in this work. TEM uses a focused beam of electrons generated by a thermionic or field emission source and accelerated with 200 keV at the sample of interest. Considering that electrons contain a wave p article duality, the wavelength ( ) of such electrons can be calculated from the following equation: (2 1) where =6.62 6 10 34 constant =9.109 10 3 1 kg is the mass of an electron, =1.602 10 19 C is the electric charge of an electron, and is the accelerating voltage. Under 200 keV, =2.5 10 12 m, which gives rise to the ability to obtain nanometer scale features in the TEM. Following electron generation, t he electrons are guided by electromagnets through the TEM column which consists of various lens systems that focus the electron beam onto the sample. The condenser lens system serves to control the formation of

PAGE 47

47 the initial beam, the objective lens system then focuses the beam onto the sample itself, and then the projector lens system expands the beam onto a phosphor screen for viewing. An illustration of a TEM column and its componen ts is shown in Fig ure 2 5. The most common method to image amorphous layers is on axis brightfield mode, where the objective aperture is aligned around the zone axis of the bulk crystal. Due to the difference in electron scattering from crystalline Ge an d Ge, a phase contrast occurs. This allows for the measurement and characterization of amorphous layers The weak beam dark field technique is another common technique for imaging defects in crystals. The sample is typically tilted away from the [110] zo ne axis along a kikuchi band. In the SAD pattern the diffracted and transmitted spots will overlap the kikuchi band, creating a 2 beam condition. The vector from the transmitted to diffracted spot is referred to as the g vector. Once the 2 beam conditi on is aligned in both brightfield and dark field mode, the incident beam is tilted in dark field mode such that the transmitted beam is translated in the objective aperture is placed over this weak beam. When ima ging dislocations and stacking faults under 2 beam conditions, the defect will only be visible if the following equation is satisfied: (2 2) where vector analysis can be done to determine for different defects.

PAGE 48

48 2.2.2 Amorphous L ayer M easurements a nd V elocity C alculations In order to measure SPEG velocities, the amorphous layer must be quantified. This was done by taking XTEM images at various annealin g times and importing the image into ImageJ analysis software. For each sample, the amorphous layer thickness was measured from the surface to the beginning of the /c interface. By plotting amorphous layer thickness as a function of annealing time, a linear fit could be used to determine the rate (slope) of SPEG. Since SPEG velocity is essentially a slope, the following equation was used to determine the standard deviation ( ) of the velocities. (2 3) w here is the value of the amorphous layer thickness, is the calculated value for the amorphous layer thickness based on the slope, is the annealing time, and is the mean annealing tim e. The standard deviation is expressed as positive and negative error bars throughout this work. 2.2.3 Scanning E lectron M icroscopy Scanning electron microscopy is a common technique for imaging sample surfaces. Similar to TEM, an electron beam is g enerated by thermionic or field emission means and then directed at the area of interest. Unlike TEM, electrons in an SEM system are directed to different detectors which give rise to different methods of contrast. Secondary electrons are formed from ine lastic collisions between electrons and atoms in the sample. This method of imaging is typically used for resolving small features on the sample surface. Alternatively, a backscattered detector can be used to provide information on the atomic number (Z) of the sample. Backscattered electrons

PAGE 49

49 are formed from elastic collisions with atoms in the specimen, thus samples with higher (lower) Z number will produce more (less) elastic collisions and form a brighter (darker) contrast. The accelerating voltage fo r SEM is typically much less than 40 keV and the majority of samples in this work were imaged using 5 keV. Usually a conductive coating is deposited on the surface of the sample to prevent charging, but this was not necessary for this work due to the high conductivity of Ge. 2.2.4 Atomic F orce M icroscopy AFM is a technique where a Si cantilever is scanned across the surface of the sample to measure the height and roughness of surface features. When the cantilever tip is brought in close proximity of th e surface, forces such as Van der Waals, chemical bonding, electrostatic forces, and magnetic forces result in the deflection of the cantilever. By focusing a laser on the top surface of the cantilever, the deflection of the cantilever can be used to dete rmine a force ( (2 4) where is the stiffness of the cantilever and is the distance the cantilever is bent. There are 2 basic modes of AFM: contact and tapping mode. Contact mode involves a cantilever that is constantly in contac t with the sample surface. The contact force is set by a piezoelectric positioning element. As the tip scans back and forth across the sample, the height is constantly adjusted to maintain a constant deflection. In this fashion, the height of the surfac e features can be inferred from the movement of the tip. Tapping mode is a nother AFM technique which uses an oscillating cantilever, vibrating at its resonant frequency. As the tip is scanned across the sample surface, it lightly taps the surface. The c hanges in h eight result in changes in the vibrating

PAGE 50

50 frequency. Tapping mode results in high resolution images s ince the tapping action imparts a vertical force and minimizes shear forces. The AFM for this work was done in tapping mode with Si cantilevers that had a 2nm tip radius and a resonant frequency of 320 kHz. 2.2.5 Tencor Stress M easurement T echnique The Tencor stress measurement technique is a highly accurate way to measure the radius of curvature of a wafer caused by the stress of a thin film. The system measures the radius of curvature through a laser which maps the height of the sample at various points (Fig ure 2 6) By measuring the sample before and after the deposition of a thin film, the residual stress ( ) can be calculated by the change in curvature. (2 5) where is the elastic modulus (1.03 10 11 Pa for Ge), is ratio (0.26 for Ge), is the substrate thickness (3.5 10 4 m for this work), is the substrate radius of curvature (m ), and is the film thickness (m). Residual stress is a function of intrinsic stress and thermal stress as denoted in the following equation: (2 6) The thermal component can be expressed as: (2 7) where and and are the differences in thermal expansion coefficients and difference between the deposition and measurement temperatur es, respectively [ 128 ]

PAGE 51

51 2.2.6 Raman S pectroscopy Raman spectroscopy is a technique for studying the vibrational and electronic properties of semiconductors using the inelastic scattering of light. In this technique a focused laser is used to illuminate the sample The light that is scattered from the sample can either occur without a change in energy from the incident photon (Raleigh), or with a change in energy (Raman). Th is light is then collected with a lens and directed to an interference filter to obtain the Raman spectrum. The amount of energy lost is seen as a change in wavelength of the incident photon. Since only about 0.001% of incoming photons produce Raman scattering, the main challenge of this technique is dif ferentiating Raman from the intense stray light that produces Raleigh scattering. This is usually accomplished with a Raleigh filter, which increases the signal to noise ratio. Raman spectroscopy is typically used for identifying phases, chemical composi tion, and crystal structure but the technique can also be used to measure stress [ 129 135 ] If we consider a monochromatic laser light with angular frequency t he light is then scattered inelastically by the crystal. The Raman effect is given by: (2 8) where is the angular frequency of the interacting phonon. The inelastically scattered photons will thus display a Raman shift equal to the frequency of the lattice vibration This method is unique in the fact that it is the only non destructive, spectroscopic technique that can offer information about stress. When a crystalline semicond uctor is under compressive stress, the reduction in lattice spacing causes the frequency of oscillation to decrease. This can be seen i f we consider a one dimensional lattice with a

PAGE 52

52 2 masses and positioned alternately, the frequencies of osc illation can be expressed as: (2 9 ) where is a force constant, is the wavenumber, and is the lattice constant [ 136 ] From this equation, it is evident that a smaller lattice constant results in a shift to lower frequencies. The opposite is true for tensile stresses; therefore, t he Raman spectrum shift s to lower frequencies when under compressive stress and higher frequencies when subjected to tensile stress [ 137 ] This shift in the Raman spectrum is linear with applied stress and can be expressed for Ge by: (2 10 ) where is the residual stress and is the change in frequency from the unstressed to stressed conditions. Most recently, Raman spectroscopy has been used to quantify stresses in patterned Si devices caused by mask edge defect formation [ 135 138 ] This work uses Raman spectroscopy to do a similar experiment, but with patterned Ge substrates. 2.3 Simulation Techni ques 2.3.1 Level S et M ethods and FLOOPS The level set method is a numerical method that can accurately monitor the evolution of curved interfaces by embedding the position of the interface as a higher order equation. The level set method has a distinct a dvantage over traditional techniques that incur positive feedback error as the interface progresses. Level set methods have been used to simulate etching, deposition, surface diffusion, and the propagation of an /c interface. The level set approach mini mizes error by embedding

PAGE 53

53 the interface position within a higher dimensional function. This is done by first initializing the interface in a Cartesian grid using the following equation: (2 1 1 ) w here is the higher dimensional funct ion. The function then evolves using an equation that contains the velocity of the interface. (2 12 ) where is the velocity of the interface, is the time derivative of and is the spatial derivative of In this fashion, the value of is c alculated for each time instant The level set technique is especially useful in this work, where the epitaxial growth front moves in two dimensions simultaneously around a curved interface.

PAGE 54

54 Figure 2 1. Schematic o f a RIE chamber. A RIE consists of two electrodes (1 and 4) that create an electric field (3) meant to accelerate ions (2) toward the surface of the samples (5). Figure 2 2 Schematic of the tube furnace used for annealing experiments.

PAGE 55

55 Figure 2 3. S EM images at 52 of the FIB procedure for XTEM sample preparation showing platinum deposition (a), trench milling (b), sample thinning and undercut (c), and final thinned sample (d).

PAGE 56

56 Figure 2 4. SEM and ion beam images at 52 of the FIB procedure for PT EM sample preparation. A SEM image of Pt deposition at sample edge with trench mill is shown in (a), SEM image of the undercut (b), Ion beam image of omniprobe liftout (c), and SEM image of sample mounted on omniprobe grid (d).

PAGE 57

57 Figure 2 5. Schematic of a TEM column.

PAGE 58

58 Figure 2 6. Schematic of the Tencor film stress measurement system. The radius of curvature (R) is measured by scanning multiple points across the wafer. The film thickness (t f ) and substrate thickness (t s ) are indicated above.

PAGE 59

59 CHAPTER 3 AMORPHIZATION OF GER MANIUM AND SUBSTRATE ORIENT ATION DEPENDENCE ON SPEG 3.1 Amorphization and TDD of G ermanium This section provides a summary of the amorphous layer depths measured by XTEM for implants that produced a continuous amorphous layer in this work. By performing a f ull damage density simulation using the SRIM code [ 139 ] the vacancy profile can be overlaid onto the XTEM images to determine the at the interface This can then be used to determine the critical damage (vacancies/ cm 3 ) by multiplying by the d ose The critical damage is a measure of the critical vacancy population needed for amorphization to occur. In this way, the TDD (keV/cm 3 ) can be calculated using the following equation: (3 1) where is the displacement ene rgy of Ge (15 eV). Since TDD is independent of energy and dose, a wide range of implant conditions can be directly compared. As seen in Table 3 1 the average TDD for Ge was calculated to be 2.7 1.2 10 22 keV/cm 3 which is significantly less than the repo rted TDD for Si (7.5 1. 5 10 22 keV/cm 3 ) This result indicates that it is easier to amorphize Ge than Si, which is consistent with literature findings [ 117 ] The average critical damage was determined to be 1.80.8 10 22 vac /cm 3 which is roughly half of the atomic density for Ge (4.42 10 22 /cm 3 ).

PAGE 60

60 Table 3 1 Amorphous layer depth measured by XTEM along with critical damage and TDD calculated using SRIM. All s ubstrates are (001) Ge for this table. Implant Energy (keV) Dose (Ge/cm 2 ) (nm) Critical Damage (vac/cm 3 ) TDD (keV/cm 3 ) 20 5.0E+13 18.1 1.03E+22 1.54E+20 20 1.0E+14 21.8 1.09E+22 1.63E+20 30 2.0E+15 44.8 1.41E+22 2.11E+20 60 2.0E+15 73.6 2.14E+22 3.21E+20 90 5.0E+14 81.0 3.15E+22 4.72E+20 90 2.0E+15 107.0 1.26 E+22 1.88E+20 120 2.0E+15 131.4 1.40E+22 2.10E+20 150 1.0E+14 118.9 9.13E+21 1.37E+20 150 2.0E+14 128.5 1.54E+22 2.31E+20 150 5.0E+14 138.8 1.50E+22 2.25E+20 150 1.0E+15 145.8 2.11E+22 3.16E+20 150 2.0E+15 160.5 1.73E+22 2.60E+20 300 5.0E+14 234.3 3 .89E+22 5.83E+20 1000 1.0E+15 828.6 1.70E+22 2.55E+20 1000 2.0E+15 856.7 2.14E+22 3.21E+20 AVG 1.80E+22 2.70E+20 STD DEV 8.09E+21 1.21E+20 3.2 Substrate Orientation Dependence of Ge 3 2. 1 Introduction Amorphization caused by ion implant ation and subsequent solid phase epitaxial growth (SPEG) [ 37 ] is a common technique used to dope the source and drain regions of field effect transistors (FETs) [ 140 ] With the renewed interest in G e as an alternative material in complementary metal oxide semiconductor devices [ 2 141 144 ] it is important to under stand the recrystallization process and defects that form for this material. The SPEG orientation dependence for Si has been well studied, but relatively little knowledge is known for Ge. The SPEG process for Si shows a clear dependence on orientation wh ere the regrowth in the [001] direction is about 25 times gr eater than

PAGE 61

61 the [111] and about 3 times greater than the [011] [ 24 59 81 82 ] It is believed that the difference in SPEG rate is attributed to the number of amorphous atoms at the c ) interface that are needed to attach to a crystalline atom with 2 undistorted bonds. This number is 1, 2, and 3 for [001], [011], and [111], respectively [ 6 0 62 65 ] It is also theorized that twin defect formation affects SPEG. Csepregi et al. has noted high defect densities within 16 of the [111] orientation for Si [ 59 ] and Monte Carlo simulations match reasonably well with the experiments [ 60 ] The entire orien tation dependence has been measured from [001] to [011] for Si, but so far, only 3 relative velocities along the major indices have been reported for Ge [ 59 66 ] The goal of this work is to measure the SPEG orientation dependence for Ge and characterize the resulting defect structures upon crystallization. In this regard, this work will also attempt to compare the SPEG process of Ge with past work done for Si. 3 .2 .2 Experimental 15 Ge + /cm 2 at 1MeV using a 5SDH 4 tandem accelerator at Austra lian National University. The low background doping (<1 10 1 7 As /cm 3 ) for these Ge wafers was not expected to affect SPEG velocities in this experiment [ 9 ] A set of (001) Ge wafers were mechanically polished to a mirror finish at angles of 15, 25, 40, 54.7, 70, and 80 away from t he [001] and then implanted normal to the polished surface with the same implant (Fig ure 3 1) In this way, the polished surface was normal to the ion beam during i mplant. For all samples, this implant resulted in a continuous amorphous layer extending a pproximately 800 nm from the surface. The samples were then annealed in a tube furnace with flowing N 2 at 330C and the amorphous depths were measured at various times via cross sectional

PAGE 62

62 transmission electron microscopy (XTEM). Plan view transmission el ectron microscopy (PTEM) was also used to characterize and quantify defect formation of the fully recrystallized structures for all orientations. An FEI DB235 focused ion beam (FIB) was used to prepare both XTEM and PTEM samples. A JEOL 2010F microscope operated at 200 kV was used for imaging XTEM samples, while a JEOL 200CX microscope operated at 200 kV was used for imaging PTEM samples. 3.2.3 Results Fig ure 3 2 shows an XTEM annealing sequence of the SPEG process for (001) Ge (0 ). The distance from t he surface to the /c interface marks the thickness of the amorphous layer, which gives rise to a SPEG velocity over many anneals. The regrowth of the amorphous layer was then measured at 330 C in this fashion for all 8 orientations. The progression of the /c interface for the 3 main directions ( [ 001 ] [ 111 ] and [ 011 ] ) is graphed in Fig ure 3 3. The SPEG velocities were relatively linear for these orientations. Unlike Si, which had a bimodal growth regime for the [111] the SPEG velocity for [111] Ge was constant thro ughout the annealing sequence [ 59 82 ] A linear regression analysis was performed for all 8 Ge orientations, shown in Fig ure 4 4. The measured velocity for [001] was the fastest among the different orientations at 1.04 nm/min, which corresponds well to previous reports of 0.93 nm/min at 330 C [ 145 ] The measured velocities can then be normalized to the [001] as seen in Fig ure 3 5. are graphed on the same plot. Due to the difference in activation en ergies between Si and Ge, the recrystallization temperature for Si was 550 C. The shape of the

PAGE 63

63 orientation dependence for Ge mimics that of Si reported in the literature [ 59 ] except the normalized values were higher for Ge compared to Si. The [001] velocity for this work was measured to be 16 times faster than the [111] velocity, and only 1.4 times faster than the [011] velocity. This is significantly faster than the normalized data for Si. The normalized Ge velocities obtained by Csepregi at 331.5 C compare well to what was observed in this work. The absolute velocities; however, differed by a factor of 3 This is likely due to temper ature calibration errors since both sets of Ge samples had relatively low doping levels [ 9 11 ] The absolute velocities from other literature reports match ed well with the present work, validating the results [ 11 14 112 ] Fi g ure 3 6 shows XTEM images for an isochronal anneal for all 8 orientations. The difference in amorphous layer thickness for an 11 hr anneal at 330 C illustrates a clear orientation dependence. Type III defects were uncommon to find within the regrown lay ers which is evidenced in the clean regrowth in Fig ure 3 6 Interestingly, no significant difference in roughness of the /c interface was observed among the 8 orientations (Fig ure 3 6). This is unique to Ge, since an increase in interface roughness was observed for [111] Si relative to other Si orientations [ 82 ] In order to characterize defect formation, PTEM samples were made for all 8 orientations ( Fig ure 3 7) Anneal time velocity in order to image the nearly recrystallized layers 2 beam BF conditions were chosen to maximize diffraction contrast from h airpin dislocations These dislocations were observed within the recrystal lized layers for all Ge orientations in this work with an average density of 2.710 8 1.610 8 cm 2 The density did not vary significantly among the different orientations. Hairpins nucleate when the advancing /c interface

PAGE 64

64 encounters microcrystalline pockets at a slightly different orientation than the bulk substrate [ 102 ] c interface [ 57 ] It is not surprising that similar hairpin concentrations exist for different orientations since each sample had the same implant, and presumably c interface morphology. In addition to hairpin dislocations, stacking faults and twins were observed in the (111) orientation only ( 54.7 ) For Si substrates, Csepregi et al. noticed a high concentration of twins within 16 of the [111] axis [ 59 ] but this was not the case for Ge. PTEM images indicate that these stacking faults and twins lie on the 3 inclined {111} planes, as se en in Fig ure 3 8 The density of these defects was only ~110 8 cm 2 and was not great enough to produce twin spots in diffraction, as observed in Si [ 57 ] U pon tilting to the [110] zone axis t he defects either appear wider or narrower as seen in Fig ure 3 8( b ) The angle between the set of defects that appear narrower is 109.5 which is the angle between the {111} planes in the [110] zone. The set of defe cts that appear wider is inclined 35.3 to the [110] zone which also resides on a {111} plane This confirms the twins and stacking faults form on inclined {111} planes Fig ure 3 9 shows a 2 beam tilting sequence to identify the b urgers vector of the def ects in the 54.7 Ge samples The group of defects that are highlighted in Fig ure 3 9( a ) disappear when g b=0 is satisfied as seen in Fig ure 4 9(b ) ( d) By tilting to 3 unique g vectors, the fault vector for these defects was determined to be of the a/6 [2 11 ] type, where a is the lattice constant of Ge (0.565 nm). In this manner, the defects residing on other 2 inclined (111) planes were determined to have fault vectors of a/6 {211} type as well

PAGE 65

65 Fig ure 3 10 shows XTEM images of the completely recrystal lized 54.7 sample. The weak beam dark field (WBDF) image of Fig ure 3 10( a ) illustrates the inclined nature of the twins residing on (111) planes, residing 70.5 from the surface. Hairpin dislocations are also evident in this image, spanning vertically w ithin the recrystallized layer. Fig ure 3 10( b ) shows a higher magnification image of a twinned region, with an inset of the [011] zone fast fourier transform (FFT). The pattern shows satellite spots appearing at positions 1/3 the distance between the {11 1} matrix spots, which is char acteristic of twins on { 111 } planes. Also evident from this micrograph is a widening of the twin as it progresses up to the surface. The twin widens from 18 to 23 atomic columns in Fig ure 3 10(a) and widens up to 44 atomic columns near the surface (not shown) The twin changes thickness near the core of the twinning partial dislocation s [ 146 ] The thickness of the twin is therefore defined by the number of { 111 } planes that have been sheared by partial dislocations. shape in PTEM, as seen in Fig ure 3 8 Stacking faults arranged in a 2 layer structure were sometimes observed within the twin itself evident in the bottom of Fig ure 3 10( b ) Similar polytype structures have been obser ved to form during SPEG on {111} planes in Ge [ 97 ] Stacking faults in the 54.7 orientation were also observed to form parallel to the (111) surface. These defects were usually very small (~10 20nm) in length and less than 10 monolayers thick. These defects appeared at a constant concentrat ion throughout the regrow n layer but only for the 54.7 orientation. A high resolution XTEM image of these defects is shown in Fig ure 3 11. Unlike Si, no large twins were observed to form parallel to the surface. Due to the small size and weak diffract ion

PAGE 66

66 contrast from these parallel stacking faults, they were not able to be resolved or quantified by PTEM [ 147 ] It is important to note that recrystallization temperature for this work was chosen to reflect the same fraction of the melting temperature (T m ) as work done for Si in the literature (~0.35 T m ). In this regard, the effect of temper ature on defect formation should be relatively the same for Ge and Si. 3.2.4 Discussion The use of TEM in this work offered the advantage to measure SPEG velocity as well as defect concentration. This allowed for correlating a structure property relation ship for Ge SPEG. While hairpin dislocations were observed for all orientations, their density did not vary significantly among the orientations. Elghor at al. have shown that hairpin concentrations of ~110 11 cm 2 can cause a 30% reduction in SPEG rate [ 118 ] but this density is roughly 3 orders of magnitude l arger than what was observed in this work. Moreover, the [001] SPEG velocity compared well with previous SPEG experiments in Ge, where no hairpin dislocations were observed [ 145 ] This evidence supports the conclusion that the low density of hairpin dislocations did not contribute to the orientation dependence on SPEG for this work. Previous studies have shown 2 velocity regimes for Si [111], where the initial 1 50 nm of SPEG is 3 times slower than the remaining growth [ 24 59 82 ] This was c interface, followed by a lower density of larger twins near the surface [ 57 ] I n contrast, only one [111] velocity existed for Ge The constant SPEG velocity in the [111] is likely a reflection of the constant twin and stacking fault concentration throughout the regrown layer observed in XTEM The density of inclined stacking faul ts and twins in this work was estimated to be ~110 13 cm 3 (assuming a sample thickness of 200 nm), where the density in Si was

PAGE 67

67 reported as high as ~110 16 cm 3 [ 57 ] It follows that a lower density of stacking faults and twins could lead to a less evident change in the [111] SPEG velocity. It could also be t hat the transition between the phases occurred so rapidly that it was not observed or not great enough to produce an appreciable change in velocity. Recent theories have suggested that the SPEG velocity for orientations close to the [111] is limited by the formation of twin defects [ 60 65 ] Compared with Si, this work showed a smaller concentration o f defects for Ge in such orientations [ 57 ] The higher normalized SPEG velocities near the [111] seem to confirm this theory. Thus, it is likely that the geo metrical effect of amorphous atoms bonding at the interface controls the overall shape of the or ientation dependence in Figure 3 5, while twin defect concentration influences the degree of curvature around the [111] orientations. Since Ge has smaller conc entrations of twins than Si, the normalized velocities are higher than those in Si. The reason for a decreased concentration of defects along {111} Ge could stem from a difference in stacking fault energies compared with Si First principles calculations have shown that the stacking fault energy of Si ranges from 26 33 mJm 2 while Ge is 46 56 mJm 2 [ 148 ] The larger stacking fault energy means that the defect would be harder to form, which correlates well with experimental results. 3.2.5 Conclusion The solid phase epitaxial growth process has been studied at 330C by transmission electron microscopy (TEM) for Ge wafers polished at 10 15 increments from the [001] to [011] orientations. The velocity showed a strong dependence on substrate orientation with the [001] direction displaying a veloci ty 16 times greater than the [111] direction. Cross sectional (XTEM) and plan view transmission electron

PAGE 68

68 microscopy (PTEM) revealed stacking fault and twin defect formation in the [111] orientation where all other orientations showed only hairpin dislocat ions. The twin defects formed from Ge SPEG were comparatively less dense than what has previously been reported for Si, and unlike Si, Ge [111] SPEG showed a constant SPEG velocity throughout the entire annealing sequence. The structural results indicate d that low defect densities on {111} planes gave rise to higher normalized SPEG velocities for Ge. The decreased defect densities in Ge could result from a larger stacking fault energy compared with Si.

PAGE 69

69 Figure 3 1. Photograph of the 2 polishing stubs used for obtaining different Ge orientations. Implants were done normal to the top surface of the polishing stubs. XTEM samples were fibbed perpendicular to the edge of the wafer. Figure 3 2. XTEM micrographs of an annealing sequence at 330 C o f the 0 (001) Ge orientation. Sample was implanted at 1 MeV with 1 10 15 Ge + /cm 2 (a) and annealed for 30 minutes (b), 150 minutes (c), and 330 minutes (d).

PAGE 70

70 Figure 3 C for the 3 main orientations in Ge. Figure 3 4. Measured SPEG velocities for Ge at different orientations at 330 C. Each data point represents measurements from roughly 5 XTEM samples.

PAGE 71

71 Figure 3 5. SPEG velocities for Ge and Si normalized to the [001] direction. Values for this work were measured by TEM at 330 C, while Csepregi measured Si SPEG at at 550 C and Ge SPEG at 331.5 C by RBS [ 59 ] Fi gure 3 6. XTEM images for a n isochronal anneal of all 8 orientations done at 330 C for 11 hrs. The [111] direction (54.7 ) is noticeably the slowest orientation, while [001] (0 ) is the fastest. The orientations are 0 (a), 15 (b), 25 (c), 40 (d), 54. 7 (e), 70 (f), 80 (g), and 90 (h).

PAGE 72

72 Figure 3 7. PTEM images of all 8 orientations taken under various 2 beam conditions. Orientations and annealing times were 11 hrs for 0 (a), 14 hrs for 15 (b), 46.3 hrs for 25 (c), 71.5 hrs for 40 (d), 7.5 day s for 54.7 (e), 71.5 hrs for 70 (f), 71.5 hrs for 80 (g), and 27.1 hrs for 90 (h). Figure 3 8. PTEM images of (111) Ge annealed at 330 C for 7.5 days taken with B= [111] (a), and B= [110] (b). The red and white arrows serve to guide the eye to ide ntical regions in the sample. The red (white) arrow points to a defect that becomes wider (narrower).

PAGE 73

73 Figure 3 9. PTEM images of (111) Ge annealed at 330 C for 7.5 days taken under different g vectors. The highlighted box in (a) indicates a set of def ects with a [211] fault vector. This set of defects disappear when gb=0 is satisfied (b d).

PAGE 74

74 Figure 3 10 XTEM images of (111) Ge annealed at 330C for 7.5 days. A WBDF image with B= [011] of the regrown layer containing hairpin dislocations and twins (a). A HRTEM multibeam image of a twinned region with an FFT inset (b). Figure 3 11. XTEM images of (111) Ge annealed at 330 C for 85 hrs. A low magnification image of the recrystallized layer (a), and a high resolution image of the stacking faults para llel to the surface with an FFT showing streak s (b).

PAGE 75

75 CHAPTER 4 MULTIDIMENSIONAL SPE G AND FLOOPS SIMULATIONS OF PATTERNED GE RMANIUM 4 .1 Introduction Solid phase epitaxial growth (SPEG) is a common method of achieving high dopant activation for ultra shall ow junctions in Ge [ 2 3 149 ] This process describes the layer by l ayer crystallization of an amorphous layer, which typically takes place in the source and drain regions of CMOS devices. As silicon nears the end of its roadmap, Ge is an attractive alternative material due to its higher free carrier mobility and dopant ac tivation. One dimensional (1D) SPEG has been well documented for Ge [ 11 15 37 66 145 ] but little work has been done on 2D SPEG in Ge substrates. The velocity of the amorphous c) interface for Ge is known to be thermally activated and obeys the Arrhenius type relationship given by : (4 1) where is a temperature independent prefactor, = 2.17 eV is the activation energy, =8.6210 [ 7 8 15 37 150 ] c interface, dopant impurities, and applied stress [ 10 11 17 25 151 ] Unlike 1D SPEG for blanket implants, implants around the source and drain c interfaces. This is essentially a three dimensional (3D) process, but in the case of one of the dimensions of the structure being very long, it can be simplified to a two dimensional (2D) process. In this study, a line pattern was chosen where the length (dimension into the page for all

PAGE 76

76 the figures) is very long (hundreds of microns), which is a good assumption for many devices. The crystallographic orientation dependence is important to consider for 2D SPEG. The normalized regrowth velocity for Ge has been measured by TEM from [001] to [011], as discussed in the previous chapter The SPEG velocity as a function of can be expressed as : ( 4 2) where is the value of along the [001] and is temperature independent and can be fit using a fourth order polynomial fit. For Ge, The [001] regrowth velocity is 16 times faster than the slowest regrowth direction of [111] and 1.4 times faster than the [011] direction. The SPEG velocity of the [00 1] direction is much faster than that of the [111] and [011] This causes the vertical and lateral epitaxial fronts to meet when a masked implant recrystallizes. The [111] front becomes pinched off, resulting in what is known as a mask edge defect [ 55 85 86 ] These defects are highly sensitive to stress, and tend to be more pronounced when formed under compressive stress [ 152 ] From a modeling perspective, it is important to be able to predict the formation of mask edge defects as they can affect the short channel mobility and drive current of devices [ 138 153 155 ] The goal of this work is to study and simulate the 2D SPEG process for Ge in order to gain an under standing of how Ge crystallizes and the mask edge defects that form during this process.

PAGE 77

77 4 .2 Experimental 4 .2.1 Sample P rocessing For this work, a (001) Ge wafer was patterned with 110 nm of Si 3 N 4 using a plasma enhanced chemical vapor deposition (PECVD) tool at 300 C The wafer was then spin coated with 250 nm of PMMA A4 (polymethylmethacrylate, anisole 4) resist and baked at 170C for 30 min. E beam lithography was used to create line regions on the wafer aligned along [011] directions using a 10 keV electron beam from a working distance of 7 mm For 250 nm of resist, an exposure dose of 100 C/cm 2 was used. Since PMMA A4 is a positive resist, the exposed resist was removed when treated with a developer solution of 1:3 MIBK (methyl isobutyl ketone). Reactive ion etching (RIE) was used to expose 3 different line structures. The different li ne structures consisted of ~1 5 0 nm ~350nm and ~530nm wide silicon nitride lines as seen in Fig ure 4 3 For clarity, the simulation results in this chapter will focus on the 350nm sized lines. The etch parameters consisted of gas flow= 30 sccm SF 6 RF1=2 00W, RF2= 1W, pressure= 5mT, time=60 sec. The wafer was purposely under etched by ~10 nm to prevent damage to the Ge surface. A schematic of the wafer processing for this experiment is illustrated in Fig ure 4 1. One set of patterned samples was implanted at 90 keV with a dose of 510 14 Ge + /cm 2 c interface under the mask edge, as seen in the cross sectional transmission electron microscopy (XTEM) image in Fig ure 4 5( a ) Another set of samples wa s implanted at 300 keV with a dose of 510 14 Ge + /cm 2 which produced an amorphous layer 160 nm under the mask and 235 nm in the

PAGE 78

78 exposed areas as seen in the XTEM image in Fig ure 4 7( a ) c interface is not in contact with the surface an d is therefore not subjected to any surface pinning. The samples were annealed in a tube furnace at 330C in N 2 atmosphere for 44 335 min. Anneal times were chosen to see different points during the regrowth process when the regrowth evolution showed im portant changes. A set of the samples with interfacial pinning had the Si 3 N 4 removed prior to annealing via a 5 minute etch in hydrofluoric acid, as seen in Fig ure 4 6( a ) This was done to observe the effect of c interface evolution. It was determined from previous work that the presence of the Si 3 N 4 c interfaces since s imulations have shown that the stress from the silicon nitride is concentrated within the first 100 nm of the surfac e [ 47 67 ] An FEI DB235 focused ion beam (FIB) was used to prepare XTEM samples and a JEOL 2010F was used to image the 2D SPEG process. 4 .2.2 Wafer Curvature E xperiment Wafer curvature measurements are a commonly used technique for understanding the stresses of thin films, as described in the introduction chapter. For this experiment, a Tencor thin film stress measurement tool was used to map the intrinsic curvature of a bare 4 inch Ge wafer. Then, a 165 nm layer of Si 3 N 4 was deposited via PECVD onto the Ge wafer (Fig ure 4 3) The wafer was then heated from room temperature to 330 C at a ramp rate of 15 C per m inute while wafer curvature measurements were taken every 12 minutes. The wafer was held at this temperature for 300 minutes, which is approximately the time of the longest SPEG anneal.

PAGE 79

79 4 .2.3 Raman Spectroscopy E xperiment 3 different Si 3 N 4 pattern spac ings ( 530 nm 350nm and 150nm ) were chosen as the area s of interest as seen in Fig ure 4 3 The silicon nitride patterned sample was annealed at 330C for 335 minutes, resulting in a stressed /c interface. A nother patterned sample was annealed with the silicon nitride removed, resulting in an unstressed /c interface. The 530 nm, 350nm, and 150nm patterned regions were analyzed for both the stressed and unstressed samples. A Horiba Jobin Yvon LabRAM Aramis Raman system equipped with a CCD detector was used for analyzing substrate stress at room temperature The setup consisted of a 50mW 532 nm laser with a 50 um confocal aperture and 2400 g/mm grating. This resulted in a ~ 1 um spot size for the laser and penetration depth of about 20 nm [ 156 ] Data acquisition consisted of 10 second exposures, which were averaged 3 times to form a spectrum. Peak shifts were determined by fitting the Raman spectrum with a Lorentzian f it [ 157 ] 4 .2.4 Implementation of Level Set M ethods As mentioned in the introduction chapt er, the Level Set Method is a very accurate numerical technique to simulate the progression of an /c interface. The 2D SPEG process was modeled using level set techniques and implemented in FLOOPS [ 158 ] Level set simulations were used to track the evolution of the propagating interface, where the interfa c interface [ 159 ] For the FLOOPS simulations of the patterned Ge structures, Equation ( 4 2) was modified to be linearly dependent on interfacial curvature via [ 67 160 ] (4 3)

PAGE 80

80 where is the curvature factor with units of length and where is the radius of curvature at the mask edge (r~80nm). For the pinned interface nitride patterned samples, =8 10 8 cm was used, while =310 7 cm was used for the etched samples with a pinned interface. The unpinned interface samples were simulated with a curvature factor of =110 6 cm. Eq uation ( 4 3) was used for level set simulations of the 2D SPEG process at 330C Fig ure 4 4 shows the SPEG velocities as a function of Ge orientation dependence with a fourth order polynomial fit. The data has been normalized to the [001] direction, which was measured to be ~ 1 nm/min at 330C. The orientation angle is meas ured from the [001] direction so that 90 corresponds to the [011] direction. Given the dataset, the fourth order is the least order polynomial that yields a reasonable fit. The polynomial for orientation is given by: (4 4) where = 1.4 7 10 7 = 3.22 10 5 = 1.92 10 3 = 1 51 10 2 = 1.01 and is the angle in degrees. is valid between 0 and 90 degrees, which w as sufficient for modeling the 2D SPEG from the [001] to [110] directions. Since is unitless, the units of are degrees 1 Also, since the orientation dependence is independent of temperature [ 59 ] the regrowth shapes simulated in this experiment would also be independent of temperature. 4 .3 Results and Discussion 4 .3.1 P inned Interface S tructure Fig ure 4 5 shows XTEM images of the annealing sequence at 330C for pinned interface structures with silicon nitride present during the anneal. After 44 minutes, the

PAGE 81

81 c interface squares off (Fig ure 4 5( b ) ) and after 135 minute s, mask edge defects form (Fig ure 4 5( c ) ). The m ask edge defects remain after complete crystallization at 335 minutes (not shown). Similar mask edge defects were observed in Si and were determined to be dislocations aligned along [011] type directions [ 47 85 161 ] Fig ure 4 6 shows XTEM images of the annealing sequence at 330C for pinned interface structure with the silicon nitride etched prior to the anneal. After 44 minutes, c interface becomes obtuse (Fig ure 4 6(b) ) and eventually crystallizes up to the surface defect free (Fig ure 4 6( d ) ). This result indicates that the stress from the silicon nitride c interface shape, resulting in mask edge defect formation. Wafer curvature measurements revealed that the stress in the silicon nitride was tensile, which would impart a compressive stress on the substrate. This is consistent wi th previous work, which found that compressive stresses facilitate the c interface [ 47 152 ] Interestingly, the removal of this film relieved enough stress to eliminate mask edge defect formation in the patterned Ge. While mask edge defects have been known to cause device leakage [ 123 ] recent work has shown that mask edge defects can actually be beneficial if positioned correctly along the edges of the channel [ 135 138 153 ] Due to the vacancy type nature of the mask edge defect the regions surrounding the defect are put in a state of ten sion. This implies that annealing with (without) a nitride mask would be beneficial for nMOS (pMOS) planar devices. Morarka et al. have suggested that the role of the applied stress on Si SPEG evolution may be accounted for by simply changing the curvatu re factor [ 151 ]

PAGE 82

82 S imu lations with larger curvature factors matched reasonably well with experiments under tensile stress, and simulations with smaller curvature factors matched reasonably well with experiments under compressive stress. This work has shown that unstressed (str essed) Ge SPEG can be simulated with a curvature factor of A=310 7 cm (A=810 8 cm), which further supports that the silicon nitride caused a compressive stress in the substrate. This compressive substrate stress is quantified throu gh Raman spectroscopy is section 4 .3. 5 It is interesting to note that the presence of a silicon nitride stress reduced mask edge defects for Si [ 152 ] but enhanced mask edge defect formation for Ge in this work. One possible explana tion for this discrepancy is that the nature of stress in the silicon nitride could be different. This work has shown that a tensile nitride mask produces compressive stresses in the substrate, thus enhancing defect formation. It follows that a compressive nitride mask would produce tensile stress in the substrate and reduce defect formation. The fact that Olson (this work) observed faster (slower) SPEG with the mask present substantiates this argument [ 152 ] It is also possible that mask edge defect formation was suppressed in the Si case due to the location of nucleation. For Si, lateral straggle results in less amorphization under the mask due to a higher amorphization threshold. This results in mask edge defect nucleat ion further away from the mask edge. Conversely, the low amor phization threshold for Ge results in more amorphization under the masked region. This is an important difference between Si and Ge since it has been demonstrated that the stress concentrates at the mask edge and changes sign as well [ 130 131 162 164 ] This would imply that the same tensile

PAGE 83

83 stress could result in tensile or compressive substrate stress depending on exactly where the defect nucleates. 4 .3.2 U npinned Interface S tructure Fig ure 4 7 c interface. Unlike the structures with con vex ure 4 5(a) and 4 6(a) the as im planted structure shown in Fig ure 4 7( a ) c interface with both concave ( 0 ) and convex ( c interfaces. A negative value would yield smaller SPEG velocities for concave interfaces according to Equation (4 3). In fact, this is exactly what is observed experimentally since the interface curvature decreases over time. Th is implies that interfaces with positive grow faster than interfaces with negative Throughout the annealing sequence, the FLOOPS simulations matched well with the XTEM images. This shows that Eq uation ( 4 3) can be used to simulate both concave and c onvex interfaces for Ge. Fig ure 4 7( d ) shows the recrystallized structure for the unpinned interface free of mask edge defects. c interface in the unpinned case had less curvature than the pinned case, sugg esting that the formation of mask edge defects is also highly dependent on the initial interfacial curvature. This is also why a different curvature factor needed to be used. T he defect free nature of th is sample could also be due to the absence of stres s near the initial c interface Simulations have shown that the stress from the nitride is concentrated within the first 100 nm of the surface [ 47 ] which coul d explain why the deep implant in this case recrystallized free of defects. It is interesting to note that analogous structures in Si formed mask edge defects in the unpinned case. The triangular amorphous regions near the surface in Si were not

PAGE 84

84 observed in Ge. This stems from the different SPEG dependence on substrate orientation, notably the higher [110] velocity in Ge. The more isotropic SPEG orientation dependence for Ge leads to a more uniform and defect free regrowth up to the surface. Unlike simi lar reports for Si [ 67 ] there was minimal diffraction contrast from en d of range damage in Ge throughout this annealing sequence. This is consistent with reports in the literature, suggesting a small size and density of EOR in Ge [ 111 112 165 ] Since Ge has a reduced threshold for amorphization (as determined in chapter 3), it makes sense that less interstitials would contribute to defects in the EOR. 4 .3.3 Comparison of Si and Ge Mask edge defect formation is important for creating tensile stress for planar nMOS devices [ 138 153 ] ; however these defects may be undesirable for pMOS devices where a compressive stress is wanted [ 166 ] Since Ge has gained attention as an alternative source drain material it is important to be able to predict ma sk edge defect formation in these structures to optimize performance. Figure 4 8 shows FLOOPS simulations for identical Si and Ge structures. The curvature factor for the simulations is also kept constant at 8 10 8 cm for both structures so the only di fference is the substrate orientation factor for Si and Ge It is evident from the simulations that /c interface is less pinched than its Si counterpart. This means that Ge would be less prone to mask edge defect formation. While this may be inconvenie nt for utilizing stress memorization technology in nMOS Ge devices it may be advantageous for the use of Ge in pMOS devices since mask edge defects would be undesirable in this case. From this comparison, it is clear that the orientation

PAGE 85

85 factor plays an important role in mask edge defect formation. Since Ge has a more isotropic orientation dependence on SPEG, the structure recrystallizes with less defects. Perhaps a more dramatic comparison of SPEG orientation dependence between Ge and Si lies in the unp inned interface structure. Figure 4 9 (a) (b) shows annealed samples for Ge and Si Both structures had similar initial /c interfaces where the difference in amorphous depth under the masked and unmasked areas was 75 nm and 100 nm for Ge and Si respecti vely. Upon annealing, Ge crystallizes defect free, whereas Si forms well defined mask edge defects and end of range defects. Since the initial interface was deeper in the unpinned case, there is more room for the [100] front to overwhelm the other orient ations in the case of Si. This leads to the creation of amorphous regions faceted on {111} planes. This was not observed in unpinned Ge structures due to a more isotropic orientation dependence. 4 .3.4 Wafer Curvature Measurement D ata The wafer curvatur e method measures the residual stress of the deposited nitride as discussed in chapter 2 Fig ure 4 10 shows the residual tensile stress of the nitride during thermal cycling from room temperature to 330 C. The film stress initially decreases due to ther mal relaxation of the film. The decrease is minimal (<20 MPa), which is primarily due to the small difference between the deposition and anneal temperatures. This essentially minimized the thermal component of the residual stress as seen in Equation (2 7 ). The tensile stress in the nitride then increases over the 300 minutes at the anneal temperature, which is then retained in the film upon cooling. Low temperature deposited nitrides (300 C) have a smaller surface diffusivity, which ultimately creates a relatively porous microstructure. Since the nitride deposition

PAGE 86

86 involves silane (SiH 4 ) and ammonia (NH 3 ), i t is p ossible that upon annealing, residual gases are removed from the nitride, causing it to densify [ 167 170 ] This irreversible change in microstructure would cause a permanent increase in the tensile stress generated in the film as seen in Fig ure 4 10 Ultimately, this graph shows that stress was still re tained in the film at the anneal temperatures studied in this work. It is important to note that several thermal cycles were performed during the annealing of patterned Ge samples since XTEM samples needed to be made while the nitride stress was only mea sured for one thermal cycle However, it can be assumed that the stress of the nitride did not increase for further thermal cycling based on comparison with literature results [ 167 170 ] 4 .3.5 Raman Spectroscopy D ata In addition to wafer curvature measurements, Raman Spectroscopy was used as a second technique to measure stress. As described in the introduction chapter, Raman spectroscopy can be used to measure str ess in semiconductor crystals by measuring the shift of vibrational modes within the lattice The stress for Ge scales linearly with frequency shift according to the following equation: (4 5) where is the residual stress and is the change in frequency from the unstressed to stressed conditions [ 137 ] For this experiment, a 532 nm laser was focused on a section of patterned Ge with (stressed) and without the nitride (unstressed). The Raman peak for the 3 unstressed Ge patterns averaged at 300. 8 0.1 cm 1 which compares well with the reported Ge peak at 300.7 c m 1 [ 137 ] T he Raman peak shifted for the nitride patterned samples, indicating pattern induced stres s in the substrate

PAGE 87

87 (Fig ure 4 11 ) The Raman shifts and corresponding compressive stresses for the large, medium, and small patterns were 300. 5 0.01 cm 1 ( 10.0 MPa) 300. 2 0.01 cm 1 (21.6 MPa) and 299 9 0.02 cm 1 (31.3 MPa) respectively. This increase in substrate stress with decreasing pattern spacing is consistent with similar studies in Si [ 164 ] T he calculated stress using Eq uation (4 5 ) is graphed in Fig ure 4 1 2 which shows a linear relationship between line width and substrate stress. Th e magnitude of the calculated stresses from Raman compares well with the results from wafer curvature calcula tions, suggesting that the stress in the implanted region is similar in magnitude to the stress in the film, but opposite in sign [ 89 162 163 171 ] Fig ure 4 1 3 shows XTEM images of the fully recrystallized patterned structures. Mask edge defects form for the 350 nm and 150 nm patterned samples but not for the 530 nm pattern. This supports the conclusion that the mask edge defects are stress induced. The limitation of using Raman spectroscopy to measure stress is the spatial resolution. In this case, the beam spo t of the laser is roughly an order of magnitude larger than the feature size, so line scans could not be performed across the patterned structures. Other techniques, such as nano beam diffraction (NBD) have recently been employed to gather high spatial re solution (10nm) data of in plane stresses around patterned structures [ 135 172 174 ] Unlike Raman each element of the strain tensor can be solved for in TEM. 4 .4 C onclusions The 2D SPEG process for Ge was studied using TEM and modeled using level set c interface shape upon SPEG. The unstressed patterned Ge SPEG was virtually free of mask edge defects, while t he stressed patterned Ge formed a mask edge defect upon SPEG. The

PAGE 88

8 8 c interface was found to recrystallize defect free as well. The ability to c interface curvature shows promise for use of Ge in devi ce structures. FLOOPS simulations were accurate in predicting the c interface and the curvature factor was modified to account for the c interface. Wafer curvature experiments me asured tensile film stresses, which caused compressive substrate stresses of the same magnitude as measured by Raman spectroscopy. Moreover, there was a measured stress dependence that scaled inversely with nitride pattern width. As the nitride pattern w idth increased to 530 nm the s tress in the substrate was insufficient for mask edge defects form ation This supports the theory that the presence of compressive stresses promotes mask edge defect formation in Ge. In comparison to Si, mask edge defects a re less likely to form in Ge since Ge has a more isotropic SPEG orientation dependence. Simulations were found to match reasonably well with experimental data to support this theory.

PAGE 89

89 Figure 4 1. Schematic of wafer processing for multidirectional growth Deposition of Si 3 N 4 and PM MA resist (a), exposure of Si 3 N 4 using E beam lithography (b), and etching of the Si 3 N 4 using RIE (c).

PAGE 90

90 Figure 4 2. XTEM image of the continuous 165 nm Si 3 N 4 film used for wafer curvature measurements.

PAGE 91

91 Figure 4 3. Plan view SEM image of the 3 nitride patterns with widths of 530nm (a), 350nm (b), and 150 nm (c). The red circle indicates the approximate laser spot size for Raman measurements.

PAGE 92

92 Figure 4 4. The Ge orientation dependence on SPEG velocities normalized to the [001] direction (0) with a fou rth order polynomial fit. The [111] and [011] directions exist at 54.7 and 90, respectively. Figure 4 5. X TEM images of nitride stressed patterned Ge implanted with 90 keV 510 14 Ge + /cm 2 and annealed at 330C for (a) 0, (b) 44, (c) 135, and (d) 235 minutes. The corresponding FLOOPS simulations are shown below in (e) through (h) using a curvature factor of A=810 8 cm. A mask edge defect is produced in the stressed case.

PAGE 93

93 Figure 4 6. X TEM images of unstressed patterned Ge implanted with 90 keV 51 0 14 Ge + /cm 2 and annealed at 330C for (a) 0, (b) 44, (c) 135, and (d) 235 minutes. The corresponding FLOOPS simulations are shown below in (e) through (h) using a curvature factor of A=310 7 cm. No mask edge defect is produced in the unstressed case. Figure 4 7. X TEM images of patterned Ge implanted with 300 keV 510 14 Ge + /cm 2 and annealed at 330C for (a) 0, (b) 44, (c) 235, and (d) 335 minutes. The corresponding FLOOPS simulations are shown below in (e) through (h) using a curvature factor of A=1 10 6 cm. No mask edge defect is produced in the unpinned case.

PAGE 94

94 Figure 4 8. FLOOPS simulations for pinned interface structures for Si at 500 C (a) (d) and Ge at 330 C (e) (h). The curvature factor was kept constant for both structures at 8 10 8 cm as w ell as the /c initial interface. Figure 4 9. XTEM images for unpinned interface structures in Ge annealed for 335 minutes at 330 C (a), and Si annealed for 600 minutes at 500 C. The white dashed line indicates the initial /c interface for Ge in (a). Reprinted and modified with permission from S. Morarka, N.G. Rudawski, M.E. Law, K.S. Jones, R.G. Elliman, J. Appl. Phys. 105 (2009), Copyright [ 2009 ], American Institute of Physics

PAGE 95

95 Figure 4 10 Evolution of residual tensile stress in the Si 3 N 4 film up on annealing at 330 C as calculated from wafer curvature measurements. Arrows indicate the heating and cooling cycle. The sample was held at temperature for 300 minutes.

PAGE 96

96 Figure 4 11 Raman spectroscopy data showing the unstressed peak near 300.7 cm 1 a nd a progre ssive shift for smaller nitride line widths

PAGE 97

97 Figure 4 1 2 The effect of line width on substrate stress calculated from equation 4 5 Figure 4 13 XTEM images of different nitride pattern widths implanted with 90 keV, 510 14 Ge/cm 2 and annea led at 330 C for 335 minutes: 530 nm (a), 350 nm (b), and 150 nm (c). The mask edge defect becomes more apparent as the pattern spacing decreases.

PAGE 98

98 CHAPTER 5 SPEG WITH NON PLANAR SURFACES AND TRENCH EDGE DEFE CT FORMATION IN GERMANIUM 5 .1 Introduction Th e previous chapter discussed 2D SPEG in patterned substrates with a planar surface. Recent CMOS device structures have moved to more com plicated 3D structures, such as the f inFET (Fig ure 5 1) In this light, it is often useful to understand how SPEG beha ves around non planar surfaces. SPEG in Si fin structures consists of epitaxial growth in the lower portion of the fin stacking fault formation in the middle of the fin, and polycrystalline homogenous nucleation in the upper portion of the fin [ 94 96 97 ] The stacking fault defects are known to be highly stable, even after rapid thermal annealing (RTA) at 600 C [ 97 ] It is therefore desirable to understand the mechanism of this defect formation and control it, if possible. It is known that 2D SPEG from amorphized trench like structures in Si can result in defective regrow th in the [111] direction [ 55 98 175 ] In addition, the pr esence of an oxide on the (110) face of the trench can noticeably affect defect formation in the corner of the trench [ 99 ] It is believed that having an oxide present near the trench edge hinders SPEG by forcing Si O bonds to rearrange at the /c interface. Thus, annealing without SiO 2 near the sidewall results in a less defective structure. So far, there has not been an analogous study for the effect of a surface oxide on trench edge defect formation in Ge. In contrast to Si, t he thermally grown oxide for Ge is known to be highly unstable and growth rates vary throughout the literature [ 96 176 180 ] While the oxidation of Si can be predicted well by the Deal Grov e model [ 181 ] the oxidation of Ge behaves very differe ntly. The growth of GeO (Ge +2 ) is more

PAGE 99

99 preferable than GeO 2 (Ge + 4 ) and above 450 C, the monoxide state tend s to desorb from the surface, resulting in substrate loss For this reason, all anneals in this experiment are kept below this temperature. 5 .2 Experimental Similar to the previous chapter, a (001) Ge wafer was patterned with 110 nm of Si 3 N 4 using a plasma enhanced chemical vapor deposition (PECVD) tool at 300 C The wafer was then spin coated with 250 nm of PMMA A4 (polymethylmethacrylate, aniso le 4) resist and baked at 170C for 30 min. E beam lithography was used to create line regions on the wafer aligned along [011] directions using a 10 keV electron beam from a working distance of 7 mm For 250 nm of resist, an exposure dose of 100 C/cm 2 was used. Since PMMA A4 is a positive resist, the exposed resist was removed when treated with a developer solution of 1:3 MIBK (methyl isobutyl ketone). Reactive ion etching (RIE) was used to purposely over etch and create a trench structure with a hei ght to width ratio of 0.25. The etch parameters consisted of gas flow= 30 sccm SF 6 RF1=200W, RF2=1W, pressure= 5mT, time=150 sec. A schematic of this wafer processing is depicted in Fig ure 5 2. The samples were implanted at 30 0 keV with a dose of 510 14 Ge + /cm 2 whic h produced an amorphous layer 230 nm deep, as seen in the cross sectional transmission electron m icroscopy (XTEM) image in Fig ure 5 3(a) Samples were etched for 5 minutes in HF to remove any native oxide on the surface. One set of samples was immediately coated with a 90nm layer of SiO 2 via PECVD, while another set was left uncoated The two sets of samples were t hen annealed in a tube furnace in a reducing ambient of 5% H 2 95% Ar for times ranging from 88 minutes to 1 day at

PAGE 100

100 temperatures 400 C. Times and temperatures were chosen to monitor key changes in the evolution of the /c interface with TEM. An FEI DB235 focused ion beam (FIB) was used to prepare XTEM samples and a JEOL 2010F was used to image the structures. 5 .3 Res ults The as implanted structure created /c interface that was unpinned from the surface (Fig ure 5 3( a ) ) similar to what was observed in the previous chapter. The RIE process resulted in an overetched surface, which created some irregularities in t he co rner regions of the trench; h owever, this did not significantly affect the shape of the original /c interface. The evolution of the /c interface can be seen in Fig ure 5 3( b ) ( d ) After 665 minutes at 330 C, amorphous Ge remains in the corner regions. T he /c interface then facet s to {111} planes in the corner regions as seen in Fig ure 5 3( d ) This type of faceting is similar to what has been observed for analogous structures using a Si substrate [ 98 ] In order to completely recrystallize the structure, the anneal temperature was i ncreased to 400 C. The fully recrystallized structures of the SiO 2 coated and uncoated samples are seen in Fig ure 5 4 For the coated sample, the amorphous layer regrew epitaxially about halfway up the sidewall while the upper 50 nm was marked by defect ive growth. The defects in the corner regions were characteristic of stacking faults lying on {111} planes (Fig ure 5 4( b ) ). A high resolution XTEM image of the stacking faults formed in the coated sample is shown in Fig ure 5 5 Interestingly, the remova l of native Ge oxide from an HF etch eliminated defect formation all together in the uncoated sample shown in Figure 5 4 (c) (d). This is different than what was

PAGE 101

101 observed in Si structures, where the uncoa ted trench Si sample still displayed stacking fault formation upon SPEG [ 99 ] 5 .4 Discussion Since defects wer e only observed in the oxide coated samples, it is believed that the presence of the oxide influences the defect formation. Since SPEG in the [111] has the slowest velocity [ 59 82 ] the a/c interface gets pinned in the corner regions From there, the a/c interface must then template from an oxide, which further inhibits the epitaxial process. The fact that th e amorphous layer regrew defect free partially up the sidewal l is an interesting finding and was not observed for analogous Si structures reported in the literature [ 99 ] In Si, stacking faults were observed to form in the entire trench corner where the /c interface was pinned. Also, unlike Si, SPEG with a free surface prevented the formation of stacking faults in Ge. This finding supports similar conclusions where SPEG in Ge fins was found to be less defective than Si [ 94 96 97 ] This difference between Si and Ge is consistent with the results from chapter 3 w here Ge had a lower {111} defect concentration than Si. The reason for that Ge is less defective than Si could be due to a higher stacking fault energy for Ge, as discussed in chapter 3 This implies that it is harder for the 3 amorphous Ge atoms to rear range in a faulted configuration along the {111} crystalline ledge than Si. Alternatively, it could be that crystalline bonds are easier to form in the case of Ge. From a bond breaking perspective, it makes sense that the Ge / Si O 2 bond is slightly less co valent and thus weaker than the Si O 2 bond, thus allowing for easier regrowth along the coated edge for Ge Due to the low annealing temperatures employed in this work, it is possible that a strong bond was never created between the Ge/SiO 2 Research has shown that

PAGE 102

102 anneals up to 700 800 C are needed to strengthen the Ge/SiO 2 bond [ 182 ] I n this case, the amorphous layer would recrystallize before the Ge/SiO 2 bond strengthened allowing for easier SPEG. 5 .5 Conclusion s SPEG was investigated for self amorphizing implants in non planar Ge substrates. The trench corner defects that formed during SPEG result from pinning the /c interface against an oxide layer. Upon removal of this surface oxide, Ge SPEG improved and corner defects did not form. Compared with sim ilar trench structures in Si, Ge proved to be less defect ive upon SPEG The reason that Ge is less defective tha n Si could be related to a higher stacking fault energy for Ge or weak bonding at the Ge/SiO 2 interface This work shows encouraging results for using SPEG to create highly doped Ge Fin FET structures.

PAGE 103

103 Figure 5 1. Schematic of a f inFET device structur e

PAGE 104

104 Figure 5 2. A schematic of wafer processing for the trench Ge structures. Deposition of Si 3 N 4 and PMMA resist (a), exposure of Si 3 N 4 using E beam lithography (b), and overetching of the Si 3 N 4 using RIE (c). Figure 5 3. An XTEM annealing sequence for the uncoated trench structure at 330 C. The as implanted structure with 510 14 Ge + /cm 2 is shown in (a), along with annealed samples of 88 minutes (b), 235 minutes (c), and 665 minutes (d).

PAGE 105

105 Figure 5 4. XTEM images of the trench structures annealed a t 400 C for 1 day. The SiO 2 coated trench (a) and a zoomed in image of the SiO 2 coated corner (b). The uncoated trench (c) and zoomed in image of the etched corner in (d).

PAGE 106

106 Figure 5 5. High resolution XTEM image of the stacking faults for the SiO 2 coate d sample. Figure 5 6 Si trench structure pinned against an SiO 2 layer. Implanted with 1 10 15 Si + /cm 2 at 40 keV (a), and annealed at 700 C for 1 minute with SiO 2 present (b), and with SiO 2 removed (c). Reproduced and modified with permission from Burbur e, Electrochem. Solid State Lett.10, H184 (2007). Copyright 2007 by the Electrochemical S ociety. Reprinted by permission o f the Electrochemical Society.

PAGE 107

107 CHAPTER 6 THE INFLUENCE OFIMPL ANT ENERGY AND DOSE ON SPEG 6 .1 Introduction Recently, Ge has rece ived renewed interest as an alternative source/drain and channel material in CMOS devices due to higher dopant activation and free carrier mobility compared to Si [ 14 4 183 184 ] However, the knowledge base and understanding of CMOS related processing of Ge is st ill relatively small compared to that of Si. Pre amorphization using self implantation to create a continuous amorphous results in higher activation of subsequently implante d dopants during SPEG [ 2 141 142 185 ] Moreover, understanding the SPEG process is crucial to optimizing the performance of CMOS devices [ 25 47 ] The SPEG process has been studied extensively for Si and is known to be influenced by many variables [ 7 8 10 17 18 28 51 59 80 81 150 186 ] though comparatively minimal similar research has been performed for Ge. A key difference between Ge and Si is that Ge is known to become highly porous at doses above 410 15 cm 2 [ 49 187 192 ] Such behavior suggests the Ge phase can be altered with dose or implant energy [ 193 ] which may possibly lead to dependence of the SPEG kinetics on self implantation conditions, in contrast with Si where no such dependence is observed [ 7 ] The goal of this chapter is to investigate the effects of implantation conditions on Ge SPEG. 6 .2 Experimental Two sets of (001) Ge samples with background B concen trations of 5.010 17 cm 3 were self implanted at room temperature using a VIISta 900XP ion implanter. The first set of samples was implanted at a fixed energy of 150 keV with doses of 1.010 14

PAGE 108

108 2.010 15 cm 2 while the second set was implanted at a dose o f either 1.010 14 or 2.010 15 cm 2 330C in N 2 sectional transmission electron microscopy (XTEM) and plan view transmission electron microsc opy (PTEM) samples, prepared by focused ion beam (FIB) milling, were used to investigate the evolution of the SPEG process and provide a quantitative means of measuring the growth velocity, similarly as described elsewhere [ 51 ] 6 .3 Results Figure 6 1 shows a sequence of XTEM micrographs depicting the growth process of a sample implanted at 90 keV to a dose of 2.010 15 cm 2 The as impla nted structure shown in F igure 6 Ge layer 1073 nm thick. With Ge layer has crystallized and reduced in thickness 6610, 365, and 155 nm, respectively. For some samples implanted to a dose of 2.010 15 cm 2 voids 156nm in diameter spaced randomly a few hundred nm apart appeared just below the surface and tended to swell the surface Ge layer depths were measured in non swelled regions to obtain the most accurate measurement of SPEG kinetics. Figure 6 2 (a) shows an XTEM micrograph of a sample self implanted at 90 keV to a dose of 2.010 15 cm 2 exhibiting clusters of small voids just below the surface and an Ge layer 1073 nm thick (as measured in regions without voids). PTEM i maging o f the same sample, shown in Figure 6 2 (b), indicates an average of 94 voids per cluster. Approximately 0.07% of the surface of this sample was covered by voids. Afte r annealing the sample in Figure 6 2 (b) at 330 C for 176 min, the amorphous lay er crystallized, but the void clusters remain with the same area distribu tion and size as

PAGE 109

109 shown in Figure 6 2 (c). It should be noted that trapped gas is likely not a possible explanation for t he formation of voids, since only self implants were performed in this experiment. This is consistent with prior reports indicating the stability of porous regions formed from high dose self implantation upon annealing [ 141 191 192 194 197 ] Behavior sim il ar to that presented in Figure 6 2 was also observed for other samp les self implanted at 30 120 keV to a dose of 2.010 15 cm 2 It is apparent from Figure 6 2 that voids were not present uniformly across the entire sample, but were clustered randomly. This observation is surprising considering the dose uniformity is e stimated at <1% across the sample [ 198 ] Ge layer thickness versus time for samples implanted at 150 keV with doses of 1.010 14 15 cm 2 is shown in Figure 6 3(a), and for samples implanted with energies of 20 150 keV with a dose of 1.010 14 or 2.010 15 cm 2 shown in Figu re 6 3(b). The average growth velocity from each set of thickness versus time data was calculated using least squares regression analysis from 2 2 to 88 min as shown in Figure 6 4. The linear regression analysis was performed on data from 22 to 88 min in order to reduce the error in SPEG velocity calculations from the initial surface. For samples implanted at 150 keV with doses of 1.010 14 15 cm 2 presented in Figure 6 4(a), the growth velocity was nearly identical for all doses with an average growth velocity of 0.930.02 nm/min, which agrees well with the values reported in the literature for similar implant conditions [ 11 14 112 ] In terms of energy dependence, the growth kinetics of samples implanted at a dose of 1.010 14 cm 2 showed little variation with energy. However, for samples implanted with a dose of

PAGE 110

110 2.010 15 cm 2 the growth velocity clearly decreased with decreasing implant energy. This behavior differs considerably from that of self amorphized Si, whe re the SPEG kinetics are independent of implantation conditions [ 7 ] 6. 4 Discussion I t is interesting that implantation conditions affect SPEG kinetics for self amorphized Ge when since no such effect has been observed in Si. It is known that H and/or O contamination in the near surface region during annealing can slow SPEG [ 7 11 37 66 ] though the depth into which such contamination is known to occur (several Ge layer thicknesses used in this work. For a given implant condition, the growth rate was independent of the growth interface dep th [ 32 112 ] which again contradicts the notion of contamination reducing the growth velocity. Additionally, the presence of electrically active dopants [ 11 17 19 ] alters growth kinetics but this effect is typically only ob served with dopant levels exceeding 4.010 18 cm 3 which is much higher than the background concentrations used in this work. Also, this background concentration should be uniform throughout the wafer for all samples. A possible explanation of the variati on of SPEG with implant energy is implant induced stress since the presence of stress at the growth interface is known to alter SPEG kinetics [ 47 51 199 200 ] One study showed that compressive stresses in the plane of the growth interface are generated durin g Kr+ implantation into Ge, where the generated stresses increased with dose at a fixed implant energy [ 49 ] While the implanted ion in the present study is different, a reduction of the implant energy at a fixed dose leads to a higher density of implanted ions, similarly to the previous work. Strain could be attributed to an increase in three and five fold configurations upon

PAGE 111

111 implantation, leading to an increase in av erage bond length [ 201 203 ] Additionally, the generation of compressive stresses resulting from implantation is consistent with current models of stress altered SPEG [ 51 ] where in plane compressive stresses tend to retard growth kinetics. This is further supported by the observation that growth retardation was not observ ed at low doses (corresponding to lower damage densities) since less stress would be generated at lower doses. Ge to a porous structure at high damage densities may support the observed implantation dependent SPEG kinetics being influenced by stress [ 49 187 192 198 ] As stated earlier, high dose/low implant energy conditions generated sparse, but microscopic voids (a precursor to th e porous structure) and possibly, these voids are preceded by smaller, submicroscopic voids. As suggested by Mayr and Averback [ 49 ] the presence of these voids is res ponsible for the generation of in plane compressive stresses during implantation, and an incident ion transferring energy to the substrate over a smaller volume increases the probability of void formation. Thus, at high doses and low implant energies, the volume over which the energy is deposited decreases, resulting in a higher probability of void fo rmation in Ge network. The in plane compressive stress generated from void formation then could possibly slow the growth kinetics in regions between voids [ 199 ] Finally, it should be noted that in the case of group IV semiconductors, it is known that the amorphous phase can exist in a so e where short range order is not maintained and bond lengths/angles are excessively distorted [ 201 203 206 ] bond breaking/rearrangement in the growth interface, which mediates SPEG would be

PAGE 112

112 affected [ 5 ] Ge from the unrelaxed state is expected to occur very rapidly at the thermal budget used in this work [ 204 206 ] A structural relaxation argument also d oes not account for the fact that the pores and voids in Ge are highly stable upon annealing [ 141 191 192 194 197 ] Thus, voids (and void generated stresses) remain even after short range order is restored and bond lengths/angles relaxed in the non voided materi al. 6.5 Conclusion The effects of implantation energy and dose on Ge solid phase epitaxial growth kinetics were studied using (001) Ge substrates self keV and doses of 110 14 15 cm 2 It was shown that implant conditions generating the gr eatest ion density (low energy/high dose) tended to produce the greatest Ge to become highly porous at high damage densities, it was postulated that the implant produced both microscopic a nd sub microscopic voids, resulting in stress and altering the growth kinetics.

PAGE 113

113 Figure 6 1. XTEM micrographs of the solid phase epitaxial growth process at 330 C of (001) Ge self implanted at 90 keV to a dose of 2.010 15 cm 2 : a) the as implanted s tructure, b) after annealing for 44 min, c) after annealing for 88 min, and d) after annealing Figure 6 2 (001) Ge self implanted at 90 keV to a dose of 2.010 15 cm 2 : a) XTEM micrograph of the as implanted structure (surface indicated by the dotted li ne), b) PTEM micrograph of the as implanted structure (inset diffraction pattern indicates sample is amorphous) and c) PTEM micrograph of the sample in b) following annealing at 330C for 176 min (inset diffraction pattern indicates sample is single crysta l). Red arrows in parts b) and c) indicate the same void clusters in each sample.

PAGE 114

114 F igure 6 3 Amorphous layer thickness versus annealing time behavior at 330 C of self implanted (001) Ge: a) samples implanted at 150 keV with doses of 1.010 14 2.010 15 cm 2 and b) samples implanted with energies of 30 150 keV with a dose of 2.010 15 cm 2 Figure 6 4 The solid phase epitaxial growth velocity at 330 C of self implanted (001) Ge: a) the effect of implanted dose for samples implanted at 150 keV and b) the effect of implant energy for samples implanted at a dose of 1.010 14 or 2.010 15 cm 2

PAGE 115

115 CHAPTER 7 HIGH DOSE SELF ION I MPLANTATION IN GE RMANIUM 7.1 Mechanisms and Proposed Theories for Porous Formation in Bulk Ge 7.1.1 Introduction There i s renewed interest in Ge as an alternative channel material in complementary metal oxide semiconductor devices due to its higher free carrier mobility and dopant activation compared to Si. However, the evolution of damage in ion implanted Ge as a function of implantation conditions remains poorly understood. It is known that for a critical dose, Ge undergoes a crystalline (c Ge) to amorphous ( Ge) phase transition [ 112 ] and at significantly higher doses exhibits voiding within the Ge layer forming a porous structure with surface cavitation [ 187 190 192 198 ] However, the threshold ion implantation conditions for void formation remain basically unknown. Over the past 30 years, there has been much debate as to what m echanism governs formation of the porous structure in ion implanted Ge. Currently, there are two main theories of void formation for Ge: vacancy clustering and so called on of Ge point defects during ion implantation [ 207 ] where once a critical poin t defect population is created by ion implantation, excess vacancies cluster into pores [ 188 192 202 208 212 ] in order to minimize the dangling bond density. In contrast, the microexplosion theory is based on the creation of voids through pressure waves and therm al spikes caused by the overlap of ion cascades [ 213 215 ] In principle, it is possible to determine which theory better models void formation by selecting appropr iate implant conditions and observing the resulting microstructure after implantation. If vacancy clustering is the governing mechanism, then varying depth and

PAGE 116

116 concentration of the vacancy profile should have an effect on the size and depth of the voids. If the microexplosion theory governs the formation process, then a small fraction of implanted ions (<0.1%) would produce microexplosions that result in voids. This implies that dose is the critical parameter that controls the void formation process, whi ch should occur at the surface regardless of the implant energy [ 213 215 ] In this work, the influence of ion dose and implant energy on void formation in ion impl anted Ge is investigated in an attempt to better understand the threshold conditions for the formation of a porous microstructure as well as which theory best explains void formation in Ge. 7.1.2 Experimental Two sets of (001) Ge samples with background B concentrations of 5.010 17 cm 3 were self implanted at room temperature using a VIISta 900XP ion implanter with beam 2 The first set of samples was implanted at imp lant 300 keV with doses ranging from 1.010 13 15 cm 2 while the second set was impl 150 keV at a fixed dose of 2.010 15 cm 2 The samples were then annealed at 330C in a tube furnace in N 2 176 min. A third set of samples was self implanted at 130 keV with doses betwee n 1.010 16 17 cm 2 using a 5SDH 4 tandem accelerator at Australian National University. Implantation for this set was performed at room 2 Ge layers and voids were characterized using a JE OL 2010F transmission electron microscope at 200 kV in cross section (XTEM) and plan view (PTEM). An FEI DB235 focused ion beam (FIB) was used to prepare both XTEM and PTEM samples via a 30 keV Ga + beam. Scanning electron microscopy (SEM) in the FIB was used to characterize the surface morphology

PAGE 117

117 of the samples at 5 k e V. A Multimode Nanoscope III was used for Atomic force microscopy (AFM) to characterize surface roughness as a function of energy and dose. 7.1.3 Results As discussed in chapter 6 rando m clusters of voids were observed for implant conditions of 30 120 keV to a dose of 2.010 15 Ge + /cm 2 At this dose, the average void diameter was measured to be 156nm via PTEM. These void clusters covered roughly 0.07% of the surface with 94 voids per cluster Interestingly, the average depth of the voids (measured by XTEM) in samples self implanted at energies of 30 120 keV to a dose of 2.010 15 cm 2 appears to be independent of implant energy, as shown in Fig ure 7 1 This is in contrast to the approx Ge layer thickness and depth of the vacancy concentration profile peak (R d ) with implant energy predicted with SRIM [ 139 ] Furthermore, there is a complete lack of voids upon increasing the self implantation energy to 150 keV. As the implant dose is increased, the voids transfo rm into open pores. Figure 7 2 presents the evolution of the porous microstructure with dose at implant energy of 130 keV. It is evident that the amorphous layer thickness remains relatively constant over the dose sequence, while the thickness of the por ous region increases with dose, as shown in Figure 7 2 (a) (c). The pore diameter also increases with dose as shown in the SE M micrographs presented in Figure 7 2 (d) (f), which is consistent with literature reports [ 187 208 ] Interestingly, the pores at the surface appear open at a dose of 1.010 16 cm 2 but an increasing portion of the pores get covered by a surfa ce layer at 3.010 16 and 1.010 17 cm 2 as seen in Figure 7 2 (d)

PAGE 118

118 interface since the ions travel through different thicknesses of material depending on whether a surface layer is present. Fig ure 7 3 shows a sequence o f 1 m 2 AFM scans for Ge self implanted at various dose s. Fig ure 7 3( a ) shows an image of the Ge surface after forming a continuous amor phous layer with a 210 14 Ge + /cm 2 implant. The RMS roughness increased from 1.0 to 1.7 nm at the dose threshold for por e cluster formation (210 15 Ge + /cm 2 ) (Fig ure 7 3( b ) ) This increase in roughness is a precursor to porous formation in Ge [ 216 ] U pon increasing the dose to 1 10 16 Ge + /cm 2 the Ge surface turned porous and the RMS roughness increased to 16 nm. Implant energy di d not have a significant effect on surface roughness in the range of 30 150 keV (Fig ure 7 4 ), but the roughness did scale with increasing implant dose (Fig ure 7 5 ). The roughness reached a maximum near 1 10 16 Ge/cm 2 and then slowly decreased above this d ose which is consistent with literature findings [ 196 ] At sufficiently high doses, the roughness decreases due to the formation of a surface layer overtop some of the pores, as seen in the SEM image of Fig ure 7 2( e ) ( f ) Figure 7 6 shows a damage map for self implantation in Ge. The threshold dose for the formation of a continuous amorphous layer in Ge is 5.010 13 cm 2 [ 112 ] while the threshold implant dose for void formation was determined to be 2.010 15 cm 2 with an implant energy of 120 keV. No clusters of voids were observed in XTEM or PTEM above this implant energy or below this dose. This threshold dose compar es well with the void threshold doses for ions of similar mass, such as As + and Ga + [ 217 ]

PAGE 119

119 7.1.4 Discussion If void formation is governed by a vacancy cluste ring mechanism alone one would expect a uniform void distribution due the high uniformity of the ion implantation process. This work has shown that at a dose of 2.010 15 cm 2 approximately 0.07% of the surface is covered with clusters of voids. However, it is known that once the dose is increased to 4.010 15 cm 2 the entire surface is covered with voids [ 187 ] By comparison, the microexplosion theory suggests ~0.1% of incoming ions overcome the crit ical energy to form a microexplosion. However, this cannot explain void formation alone since a two fold increase in dose leads to an increase in surface coverage by roughly three orders of magnitude. The lateral range [ 139 ] (~15 nm) of a 90 keV Ge + ion into Ge is roughly equal to the average diameter of a singl e void (156nm), which means that the clusters of voids seen in Fig ure 6 2 cannot be formed by a single ion. In addition, the voids observed in this work were several orders of magnitude larger than those predicted with single ion molecular dynamics simul ations [ 213 215 218 ] These results indicate that neither the vacancy clustering or microexplosion theory can sole ly explain void formation. It is possible that the initial microexplosion serves as a nucleation point for vacancy clustering; once a single void is formed, the formation energy decreases for neighboring voids, resulting in a cluster. Therefore, the numbe r of voids a cluster contains would increase with dose, resulting in even more nucleation points. In this manner, the percentage of the surface covered in voids would increase nonlinearly with dose after the initial void formation. Furthermore, it is kno wn that the individual void size increases with dose [ 187 208 ] which strengthens the argument that voids nuclea te

PAGE 120

120 through a microexplosion mechanism, and then a vacancy clustering mechanism could govern the growth process. Additionally, the fact that voids were observed in the near surface region well above R d indicates that vacancy clustering alone cannot explai n void formation. If vacancy clustering solely governed void formation, then the void depth dependence on ion energy, shown in Figure 7 1 should be centered on the vacancy R d Instead, the void depth is roughly the same value for all ion energies. In t erms of the microexplosion explanation, there is a critical density of cascades required for a void to form. As the depth of the cascade increases, the cascade volume increases as well, and thus the critical energy to produce a microexplosion increases ra pidly with the depth of the cascade below the surface [ 215 ] This could possibly explain the observation of voids at the same distance from the surfac e, regardless of implant energy. It is possible that self implantation at 150 keV to a dose of 2.010 15 cm 2 produces a cascade density just below the critical value, thus resulting in no void formation at the surface. Upon increasing the dose into the po rous regime, several factors determine the surface morphology, including sputtering, redeposition, swelling, and ion beam annealing. It is likely that a combination of these factors contribute to the surface layer formation and su rface roughness as shown in Figure 7 3 (d) (f). Mayr et al. showed that high dose implantation in Ge results in compressive stresses, which plateau around 500 MPa [ 49 ] These high compressive stresses lead to the increase in surface roughness with dose seen in AFM. Interestingly, the amorphous depth remains relatively constant, whereas the depth of the porous layer increases with dose. It is speculated that

PAGE 121

121 vacancies continue to cluster with increasing dose, leaving the interstitials to migrate out from the surface, causing the surface to swell. 7.1.5 Conclusions This work has shown that void formation in ion implanted Ge does not occur uniformly across the surface of the samples. Rather, void formation at the threshold implant conditions exists in random clusters, which may be explainable via a combination of both the vacancy clustering and microexplosion theories of void formation. Once the voids nucleate through a microexplosion mechani sm, the voids grow into open pores through a vacancy clustering mechanism. A implantation map diagram for amorphous and porous formation in Ge has been presented. Since the voids form after amorphous threshold, but do not require annealing to nucleate, th ey do not fit into any of the 5 defect types presented by Jones et al [ 102 ] This work suggests that implan t conditions must be chosen carefully in any type of Ge based device processing; since common p and n type dopants (Ga + and As + ) have similar masses to Ge+. Dopant doses that approach 2.010 15 cm 2 could result in void formation at low implant energies w hich cannot be removed via annealing. In addition, the work presented in chapter 6 revealed that the presence of void clusters reduces SPEG velocities by exerting in plane compressive stress on the amorphous layer. These effects make high dose implantati on in Ge undesirable for device processing. 7.2 High Dose Ion Implantation in Sputtered and Evaporated Ge 7.2.1 Introduction Although high dose implantation is not desirable in the semiconductor industry, there are other applications which might take adv antage of such high surface area porous materials. Recently, there is renewed interest in porous Ge due to several

PAGE 122

122 emerging applications, such as high performance lithium ion battery anodes [ 219 ] gas sensors [ 220 ] thermoelectrics [ 221 222 ] and thermal insulators for MEMS devices [ 223 ] For many of these applications, d eposited Ge films present several advantages over single crystal Ge includi ng lower cost and the ability to use a variety of substrates In addition, the ability to control and predict the final mic rostructure is important for different application. The previous chapter proposed that the porous formation in crystalline Ge i s governed by a nucleation and growth process. Since the formation of an amorphous layer precedes the formation of the porous la yer, the number of nucleation sites within this amorphous layer should be critical for the evolution of the final microstructure. This chapter focuses on qualitatively changing the initial microstructure of deposited Ge to test the theory of nucleation an d growth for porous formation in ion implanted Ge. 7.2.2 Experimental For this experiment 300 500 nm layer s of Ge were sputtered and evaporated on to thermally grown layer s of SiO 2 Sputtering was done using a KJL CMS18 tool, while a Temescal E beam e vaporator was use d for the evaporation. The samples were implanted with doses ranging from 1 10 16 1 10 17 Ge + /cm 2 at 130 keV. Implantation was performed at room temperature using 5SDH 4 tandem accelerator with a beam 2 An F EI DB235 focused ion beam (FIB) was used to prepare XTEM samples and a JEOL 2010F was used for imaging. S canning electron microscopy (SEM) w as used to characterize and quantify sample microstructure. 7.2.3 Results Fig ure 7 7 shows a cross sectional view of the evaporated and sputtered Ge substrates implanted with 130 keV 1 10 16 Ge + /cm 2 Similar to (001) crystalline Ge, i t is

PAGE 123

123 clear that columnar por es develop in the evaporated Ge; however, spherical pores form for the sputtered Ge. Close r examination of the film microstructure through XTEM revealed spherical pores with 1nm diameter within the sputtered film (Fig ure 7 7( d ) ) while the evaporated film consisted of a void less amorphous layer (Fig ure 7 7( b ) ) It is likely that the small pores in the sputter ed film act as nucleation sites during the implant process, whereby vacancies agglomerate to form larger spherical pores [ 216 ] For the implanted sputtered film, the void size varied as a function of depth, where larger voids were found near the peak of the dam age profile (R d =40 nm) and smaller voids were found at the end of the projected range. The structure of the evaporated Ge samples was similar to the (001) crystalline Ge shown in the previous chapter. This is likel y due to a low concentration of nuclea tion sites with in the starting s ubstrates. This resulted in a columnar growth of the pores from the surface downwards. This implies that the nucleation and growth process is highly dependent on the initial microstructure of the film [ 208 209 ] Fig ure 7 8 shows the evolution of the pores formed in the sputtered Ge as a function of dose. The depth of the poro us layer increases with dose (Fig ure 7 8(a) (c) ), which is expected since the vacancy concentration increases with dose. The amount of swelling in the sputtered film was not able to be accurately measured since the thickness of the deposited layer varied by tens of nanometers across the sample. As the porous depth increased with dose, so did th e size of the pores as seen in the plan view SEM images (Fig ure 7 8(d) (f) ). The reduced contrast from secondary electrons is due to the presence of a ~7nm surfac e layer for the implanted sputtered Ge. This

PAGE 124

124 surface layer was also observed for the evaporated Ge implants as seen in Fig ure 7 7(a) The pore diameters of the different Ge substrates were measured using plan view SEM images assuming a circular pore sha pe. The distributions of the crystalline, sputtered, and evaporated substrates are shown as a function of dose for a 130 keV Ge + implant in Fig ure 7 9. The distributions follow a positively skewed Gaussian profile and become wider with increasing dose. The substrates which form columnar pores, i.e. crystalline and evaporated Ge, create larger pores than the spherical pores for sputtered Ge at the same dose. This indicates that the final pore size of ion induced porous Ge is highly dependent on the start ing substrate. 7.2.4 Discussion It is known that porous Ge forms within an amorphous layer, regardless of the starting substrate, thus it is difficult to define vacancies in an amorphous solid [ 197 ] Ho wever, there is still a difference in the amount of deposited energy within the amorphous layer as simulated with SRIM [ 139 ] For purposes of explanation, vacancies will be discussed in the sense of damage created by the ion profile. The difference in pore morphology for sputtered and evaporated films for the s ame implant conditions highlights the importance of nucleation sites for the formation of porous Ge. When a Ge film does not have pre existing nucleation sites, voids nucleate at the surface and then grow in a columnar shape from the surface downward s as demonstrated with the schematic in Fig ure 7 10 Once the voids a t the surface nucleate, it is possible that vacancies migrate from the bulk to the bottom of the void in order to minimize the dangling bond density. This results in anisotropic growth of c olumnar voids, as seen with the (001) Ge and evaporated Ge.

PAGE 125

125 The presence of ~1 nm voids existing within the sputtered film effectively create d a sp herical pore morphology with in the implanted layer An abundance of pre existing voids likely causes a com petition between nucleation sites. This results in isotropic growth of spherical pores, as seen with the sputtered Ge. A schematic of this process is illustrated in Fig ure 7 11. The final size of each spherical pore then depends on the vacancy concentr a tion at different depths caused by the ion profile It is interesting to note that crystalline, evaporated, and sputtered Ge all nucleate spherical voids at first, but these voids turn into columnar voids for crystalline and evaporated Ge for increasing do se. The location of the nucleation sites likely cause this change in morphology. The nucleation sites for crystalline Ge were highly localized near the surface. From here, the vacancies from the bulk cluster to the bottom of the void and the forward mom entum of the ion beam elongates the void to create a columnar shape. If the entire amorphous film has a uniform concentration of pre existing voids, such as in the case of sputtered Ge, then the voids expand isotropically as vacancies cluster on all sides of the voids. Further research is warranted to investigate the formation of the thin Ge surface layer. Interestingly, this layer forms for both sputtered and evaporated films but not for (001) Ge. This layer is not a result of pore morphology since both evaporated and (001) Ge formed columnar pores. The thickness of the surface layer is consistent with literature reports under different implant conditions [ 208 209 219 ] Since the thi ckness of the surface layer does not evolve with dose or energy it is possible that surface recombination forms an equil ibrium thickness. Thus, it seems plausible that deposited films have a different IV recombination rate than crystalline Ge.

PAGE 126

126 7.2.5 Conclusions This work has shown that the formation of porous Ge follows a nucleation and growth model. The morphologies of porous Ge can b e significantly altered by chang ing the nature of the initial nucleation sites. The nucleation sites grow with increasing dose due to increased vacancies. This work has shown that the size and location of the initial nucleation sites ultim ately determine whether the Ge will for columnar or spherical pore morphologies. This implies that careful engineering of the starting substrate could allow for different porous structures for different applications. Th e presented work shows promise for using deposited substrates for porous applications since both columnar and spherical pores were achieved with deposited Ge.

PAGE 127

127 Figure 7 Ge thickness and depth of peak vacancy concentration (R d ) as determined by simulations [ 139 ] Error bars indicate the average minimum and maximum depth of the voids versus implant energy for the samples self implanted to a dose of 2.010 15 cm 2 No voids were observed at 150 keV with doses of 1.010 14 2.010 15 cm 2

PAGE 128

128 Figure 7 2 XTEM micrographs illustrating the evolution of porous Ge with dose at 130 keV: (a) 1.010 16 Ge + / cm 2 (b) 3.010 16 Ge + / cm 2 (c) 1.010 17 Ge + / cm 2 (surface indicated by d otted line) and corresponding plan view SEM micrographs (d) 1.010 16 Ge + / cm 2 (e) 3.010 16 Ge + / cm 2 (f) 1.010 17 Ge + / cm 2 Figure 7 3 AFM dose sequence showing the change in surface morphology from 210 14 Ge + /cm 2 150 keV (a), 210 15 Ge + /cm 2 150 keV (b), and 110 16 Ge + /cm 2 130keV. The z scale is the same for all images.

PAGE 129

129 Figure 7 4. Graph of RMS roughness vs. implant energy measured by AFM. Dose was kept constant at 210 1 5 Ge + /cm 2 Figure 7 5 Graph of RMS roughness vs. dose measured by AFM. The imp lant energy was 150 keV between 110 14 Ge + /cm 2 210 15 Ge + /cm 2 and was 130 keV between 110 16 Ge + /cm 2 210 15 Ge + /cm 2

PAGE 130

130 Figure 7 6 Implant damage map for self implants into Ge The threshold dose values are 5.010 13 cm 2 for amorphization, 2.010 15 cm 2 for void formation, and 4.010 15 cm 2 for porous structure formation. Boxed symbol represents crystallinity, filled symbols represent continuous amorphization, half filled symbols represent void clustering, and open symbols represent porous formation. All implants were done at room temperature [ 111 112 117 187 208 217 224 ]

PAGE 131

131 Figure 7 7. XTEM images of a 130 keV 110 16 Ge + / cm 2 implant into evaporated (a) and sputtered Ge (c) with zoomed in images of the evaporated and sputtered film microstructures in (b) and (d), respectively.

PAGE 132

132 Figure 7 8. XTEM micrographs illustrating the evolution of sputtered porous Ge wi th dose at 130 keV: (a) 1.010 16 Ge + / cm 2 (b) 3.010 16 Ge + / cm 2 (c) 1.010 17 Ge + / cm 2 and corresponding plan view SEM micrographs (d) 1.010 16 Ge + / cm 2 (e) 3.010 16 Ge + / cm 2 (f) 1.010 17 Ge + / cm 2 Figure 7 9. Pore diameter histograms for crystalline (001 ) Ge, sputtered Ge, and evaporated Ge for doses ranging from 1.010 1 6 1.010 17 Ge + / cm 2 with a constant implant energy of 130 keV.

PAGE 133

133 Figure 7 10 Porous formation schematic for crystalline and evaporated Ge. Ion implantation creates clusters of voids jus t below the surface at doses near 210 15 Ge + /cm 2 (a). Voids then elongate in the direction of the ion beam as vacancies cluster at the bottom of the voids (b). Upon further increasing the dose, the porous layer continues to elongate anisotropically (c). Figure 7 1 1 Porous formation schematic for sputtered Ge. The as deposited film contains many nucleation sites (voids) from the deposition process (a). Upon implantation, the voids act as vacancy sinks and expand isotropically (b). The pores then enla rge with dose, and the largest pores are centered on R d (c).

PAGE 134

134 CHAPTER 8 SUMMARY AND FUTURE W ORK 8.1 Overview of Results The presented work has given insight into the SPEG process and defect formation for Ge. Overall, Ge was determined to have less defec ts ( {111} type, EOR, mask edge defects, and corner defects in trench structures) than Si for similar structures and implant conditions. In contrast to Si, Ge exhibits a porous regime for high dose, heavy ion implants. These defects were not recoverable d uring the SPEG process. The orientation dependence on SPEG was studied for the first time in Ge using TEM. This not only allowed for velocity measurements, but also provided information on the types of defects created upon SPEG. Ge showed significantly l ess {111} type defects than what has been reported in the literature for Si. Since t hese defects only appear ed on {111} planes for SPEG on (111) substrate s it was speculated that the defects arose from faulty bond rearrangement on {111} planes and contr ibute to higher normalized SPEG velocities for Ge Patterned substrates were created through E beam lithography and RIE, and for the first time, 2D SPEG for was studied for both pinned and unpinned /c interfaces in Ge The evolution of the /c interface was found to heavily depend on the amount of pattern induced stress. As the amount of pattern induced stress increased for pinned interfaces, the mask edge defect became more pronounced. The mask edge defect was found to disappear if the mask was stripped prior to annealing. Again, Ge was found to be less prone to mask edge defect formation than Si which was attributed to a more isotropic SPEG orientation dependence.

PAGE 135

135 The SPEG orientation dependen ce was then used to model 2D SPEG for patterned Ge substrates. Level set techniques were used to simulate SPEG in patterned substrates by implementing the orientation dependence along with a curvature factor The curvature factor was found to correlate with substrate stress and could be altered to accurately simulate the evolution of the /c interface and mask edge defect formation. 2D SPEG for non planar Ge substrat es was studied in trench Ge structures. The effect of a free surface was studied for these structures by passivating the surface with an oxide. The presence of an oxide hindered the SPEG process by forcing the Ge to template off the SiO 2 This resulted in {111} stacking faults in the corner regions of the structure. The amount of corner defects in this work was qualitatively less than similar experiments done with Si substrates. Interestingly, the defects were eliminated by creating free surfaces via a HF etch. Ge was found to be less defective than its Si counterpart and the results show promise for using Ge in non planar FinFET structures. 1D SPEG has been studied at a variety of implant conditions and was determined to decrease in velocity for low e nergy, high dose self implants. This decrease in velocity correlated well with the observation of void clusters within the amorphous layer. This suggests that the voids played a role in the decreased SPEG velocity and that implant conditions should be ch osen to avoid this regime. The mechanism of porous formation in Ge was explored and for the first time, the nucleation of void clusters near the threshold conditions was documented in XTEM and PTEM. It was proposed that voids nucleate through a microexpl osion mechanism, and then grow with dose following a nucleation and growth model. Implantation into deposited Ge further confirmed this

PAGE 136

136 theory and a physical model was presented for porous formation. The nature of the nucleation sites ultimately determin ed the final microstructure of porous Ge. 8.2 Future Work The fact that Ge ha s a different SPEG substrate orientation dependence than Si is intriguing This work has shown that SPEG for Ge is significantly less defective than its Si counterpart. This could be adva ntageous for adopting Ge in CMOS technology since Ge would have to be grown in source and drain regions. It would therefore be interesting to observe the crystal quality of grown Ge on Si as a function of crystal orientation. E xperimentally the size of the mask edge defect in Ge was smaller relative to Si. This finding warrants more research since the formation of mask edge defects is relevant to strain memorization techniques for device processing. It would be interesting to see whether th e mask edge defects in Ge are more /less dependent on stress than Si. This could be tested by creating identical patterned structures on ultrathin Si and Ge substrates and then recrystallizing implanted regions under tensile and compressive stresses by usi ng a bending apparatus. It would also be interesting to see if the mask edge defect in Ge could be enhanced by creating a deep implant, which was pinned to the surface. This would allow more room for the [110] and [100] fronts to overlap each other, thus creating a longer defect. This type of interface engineering has proved to enhance the amount of channel stress in Si [ 138 ] but has not yet been tested for Ge. Further research is also necessary to correlate quantitative stress to the curvature factor Currently, it is unknown whether the relationship between curvatu re factor and stress is linear, since there is a limited set of experiments and simulations presented in

PAGE 137

137 this work. It would be desirable to have this relationship for both tensile and compressive stress in order to accurately predict the defect formation in FLOOPS. Further research is also warranted in quantifying the nitride induced stress with NBD instead of Raman spectroscopy. Raman spectroscopy has limited spatial resolution due to the size of the beam, but point scans could be performed using NBD a cross the pattern width. This could offer information on the magnitude of the different stress tensors in addition to how stress changes spatially.

PAGE 138

138 APPENDIX FLOOPS SCRIPT FOR PA TTERNED STRUCTURES ## grid for curvature structure.. line x loc=0.0 spac=0.05 tag=Top line x loc=0.11 spac=0.002 line x loc=0.28 spac=0.002 line x loc=0.4 spac=0.05 tag=Bottom line y loc= 0.35 tag=left spac=.05 line y loc= 0.25 spac=.002 line y loc= 0.1 spac=.002 line y loc=0.1 tag=right spac=.05 region germanium xlo=Top xhi=Bo ttom ylo=left yhi=right init #plot.2d grid #source ~/scripts/params #foreach Curv {0 1e 8 5e 8 1e 7 2e 7 3e 7 6e 7 8e 7 1e 6 2e 6} { #pdbSetDouble Si Curv $Curv pdbSetDouble Si Curv 8e 8

PAGE 139

139 SPER temp=330 time=4000 spac=0.002 conc=0 level_mat=Level grid=10 movie.plot plot_name=curv_2e 7 #}

PAGE 140

140 LIST OF REFERENCES [1] G. E. Moore, "Cramming More Components Onto Integrated Circuits," Proceedings of the IEEE, vol. 86, pp. 82 85, 1998. [2] A. Satta, E. Simoen, T. Clarysse, T. Janssens, A. Benedetti, B. De Jaeger, M. Meuris, and W. Vandervorst, "Diffusion, activation, and recrystallization of boron implanted in preamorphized and crystalline germanium," Applied Physics Letters, vol. 87, Oct 24 2005. [3] B. R. Yates, B. L. Darby, N. G. Rudawski, K. S. Jones, D. H. Petersen, O. Hansen, R. Lin, P. F. Nielsen, and A. Kontos, "Anomalous activation of shallow B+ implants in Ge," Materials Letters, vol. 65, pp. 3540 3543, 2011. [4] G. L. Olson, "Kinetics and Mechanisms of Solid Phase Ep itaxy and Competitive Processes in Silicon," MRS Online Proceedings Library, vol. 35, pp. null null, 1984. [5] N. G. Rudawski and K. S. Jones, "Atomistic considerations of stressed epitaxial growth from the solid phase," Scripta Materialia, vol. 61, pp. 32 7 330, Aug 2009. [6] O. Hellman, "Topics in solid phase epitaxy: Strain, structure and geometry," Materials Science and Engineering: R: Reports, vol. 16, pp. 1 42, 1996. [7] J. A. Roth, G. L. Olson, D. C. Jacobson, and J. M. Poate, "KINETICS OF SOLID PHASE EPITAXY IN THICK AMORPHOUS SI LAYERS FORMED BY MEV ION IMPLANTATION," Applied Physics Letters, vol. 57, pp. 1340 1342, Sep 1990. [8] G. L. Olson and J. A. Roth, "Kinetics of solid phase crystallization in amorphous silicon," Materials Science Reports, vol 3, pp. 1 77, 1988. [9] B. C. Johnson, T. Ohshima, and J. C. McCallum, "Dopant effects on solid phase epitaxy in silicon and germanium," Journal Of Applied Physics, vol. 111, p. 034906, 2012. [10] B. C. Johnson and J. C. McCallum, "Dopant enhanced solid p hase epitaxy in buried amorphous silicon layers," Physical Review B, vol. 76, Jul 2007. [11] B. C. Johnson, P. Gortmaker, and J. C. McCallum, "Intrinsic and dopant enhanced solid phase epitaxy in amorphous germanium," Physical Review B, vol. 77, p. 12, Jun 2008. [12] M. Posselt and A. Gabriel, "Atomistic simulation of amorphous germanium and its solid phase epitaxial recrystallization," Physical Review B, vol. 80, Jul 2009.

PAGE 141

141 [13] E. P. Donovan, F. Spaepen, D. Turnbull, J. M. Poate, and D. C. Jacobson, "CALOR IMETRIC STUDIES OF CRYSTALLIZATION AND RELAXATION OF AMORPHOUS SI AND GE PREPARED BY ION IMPLANTATION," Journal Of Applied Physics, vol. 57, pp. 1795 1804, 1985. [14] T. E. Haynes, M. J. Antonell, C. A. Lee, and K. S. Jones, "COMPOSITION DEPENDENCE OF SOLI D PHASE EPITAXY IN SILICON GERMANIUM ALLOYS EXPERIMENT AND THEORY," Physical Review B, vol. 51, pp. 7762 7771, Mar 1995. [15] G. Q. Lu, E. Nygren, M. J. Aziz, D. Turnbull, and C. W. White, "PRESSURE ENHANCED SOLID PHASE EPITAXY OF GERMANIUM," Applied Phy sics Letters, vol. 56, pp. 137 139, Jan 1990. [16] T. K. Chaki, "Solid phase epitaxy. Effects of irradiation, dopant, and pressure," physica status solidi (a), vol. 142, pp. 153 166, 1994. [17] I. Suni, G. Goltz, M. A. Nicolet, and S. S. Lau, "EFFECTS OF E LECTRICALLY ACTIVE IMPURITIES ON THE EPITAXIAL REGROWTH RATE OF AMORPHIZED SILICON AND GERMANIUM," Thin Solid Films, vol. 93, pp. 171 178, 1982. [18] J. S. Williams and R. G. Elliman, "ROLE OF ELECTRONIC PROCESSES IN EPITAXIAL RECRYSTALLIZATION OF AMORPHOU S SEMICONDUCTORS," Physical Review Letters, vol. 51, pp. 1069 1072, 1983. [19] E. Simoen, A. Brugere, A. Satta, A. Firrincieli, B. Van Daele, B. Brijs, O. Richard, J. Geypen, M. Meuris, and W. Vandervorst, "Impact of the chemical concentration on the solid phase epitaxial regrowth of phosphorus implanted preamorphized germanium," Journal Of Applied Physics, vol. 105, May 2009. [20] L. E. Mosley and M. A. Paesler, "Electronic contribution to the recrystallization growth velocity asymmetry in doped silicon," Journal Of Applied Physics, vol. 57, pp. 2328 2330, 1985. [21] L. Csepregi, E. F. Kennedy, T. J. Gallagher, J. W. Mayer, and T. W. Sigmon, "Reordering of amorphous layers of Si implanted with [sup 31]P, [sup 75]As, and [sup 11]B ions," Journal Of Applied P hysics, vol. 48, pp. 4234 4240, 1977. [22] D. D'Angelo, L. Romano, I. Crupi, E. Carria, V. Privitera, and M. G. Grimaldi, "Role of the strain in the epitaxial regrowth rate of heavily doped amorphous Si films," Applied Physics Letters, vol. 93, pp. 231901 3, 2008. [23] E. Simoen, A. Brugere, A. Satta, A. Firrincieli, B. Van Daele, B. Brijs, O. Richard, J. Geypen, M. Meuris, and W. Vandervorst, "Impact of the chemical concentration on the solid phase epitaxial regrowth of phosphorus implanted preamorphized g ermanium," Journal Of Applied Physics, vol. 105, pp. 093538 3, 2009.

PAGE 142

142 [24] K. T. Ho, I. Suni, and M. A. Nicolet, "Substrate orientation dependence of enhanced epitaxial regrowth of silicon," Journal Of Applied Physics, vol. 56, pp. 1207 1212, 1984. [25] S. Morarka, N. G. Rudawski, M. E. Law, K. S. Jones, and R. G. Elliman, "Effect of n and p type dopants on patterned amorphous regrowth," Journal of Vacuum Science & Technology B, vol. 28, pp. C1F1 C1F5, Jan 2010. [26] D. J. Pyke, J. C. McCallum, and B. C. Jo hnson, "Hydrogen refinement during solid phase epitaxy of buried amorphous silicon layers," Journal Of Applied Physics, vol. 108, p. 044901, 2010. [27] B. C. Johnson and J. C. McCallum, "Modeling the effect of hydrogen infiltration on the asymmetry in arse nic enhanced solid phase epitaxy in silicon," Journal Of Applied Physics, vol. 96, pp. 2381 2385, 2004. [28] E. F. Kennedy, L. Csepregi, J. W. Mayer, and T. W. Sigmon, "INFLUENCE OF O 16, C 12, N 14, AND NOBLE GASES ON CRYSTALLIZATION OF AMORPHOUS SI LAYER S," Journal Of Applied Physics, vol. 48, pp. 4241 4246, 1977. [29] C. G. Van de Walle, F. R. McFeely, and S. T. Pantelides, "Fluorine silicon reactions and the etching of crystalline silicon," Physical Review Letters, vol. 61, pp. 1867 1870, 1988. [30] M. Ramamoorthy and S. T. Pantelides, "Enhanced modes of oxygen diffusion in silicon," Solid State Communications, vol. 106, pp. 243 248, 1998. [31] J. Narayan, "Interface structures during solid phase epitaxial growth in ion implanted semiconductors and a cry stallization model," Journal Of Applied Physics, vol. 53, pp. 8607 8614, 1982. [32] P. Kringhoj and R. G. Elliman, "SOLID PHASE EPITAXIAL CRYSTALLIZATION OF STRAIN RELAXED SI1 XGEX ALLOY LAYERS," Physical Review Letters, vol. 73, pp. 858 861, Aug 1994. [33 ] F. Corni, S. Frabboni, G. Ottaviani, G. Queirolo, D. Bisero, C. Bresolin, R. Fabbri, and M. Servidori, "Solid‐phase epitaxial growth of Ge‐Si alloys made by ion implantation," Journal Of Applied Physics, vol. 71, pp. 2644 2649, 1992. [34] B Mohadjeri, J. Linnros, B. G. Svensson, and M. stling, "Nickel enhanced solid phase epitaxial regrowth of amorphous silicon," Physical Review Letters, vol. 68, pp. 1872 1875, 1992. [35] E. Nygren, A. P. Pogany, K. T. Short, J. S. Williams, R. G. Elliman, and J. M. Poate, "Impurity stimulated crystallization and diffusion in amorphous silicon," Applied Physics Letters, vol. 52, pp. 439 441, 1988.

PAGE 143

143 [36] J. S. Williams and R. G. Elliman, "Substitutional solid solubility limits during solid phase epitaxy of io n implanted (100) silicon," Applied Physics Letters, vol. 40, pp. 266 268, 1982. [37] G. Q. Lu, E. Nygren, and M. J. Aziz, "PRESSURE ENHANCED CRYSTALLIZATION KINETICS OF AMORPHOUS SI AND GE IMPLICATIONS FOR POINT DEFECT MECHANISMS," Journal Of Applied Ph ysics, vol. 70, pp. 5323 5345, Nov 1991. [38] E. Nygren, M. J. Aziz, D. Turnbull, J. M. Poate, D. C. Jacobson, and R. Hull, "Effect of pressure on the solid phase epitaxial regrowth rate of Si," Applied Physics Letters, vol. 47, pp. 232 233, 1985. [39] T. K. Chaki, "Hydrostatic pressure enhanced solid phase epitaxy," Philosophical Magazine Letters, vol. 63, pp. 303 307, 1991/06/01 1991. [40] K. Tanaka, "Amorphous Ge under pressure," Physical Review B, vol. 43, pp. 4302 4307, 1991. [41] K. Tanaka, "Crystalli zation of amorphous Ge," Solid State Communications, vol. 76, pp. 213 215, 1990. [42] Q. Z. Hong, J. G. Zhu, J. W. Mayer, W. Xia, and S. S. Lau, "Solid phase epitaxy of stressed and stress relaxed Ge Si alloys," Journal Of Applied Physics, vol. 71, pp. 176 8 1773, 1992. [43] T. K. Chaki, "Mechanism of Stress Enhanced Solid Phase Epitaxy," MRS Online Proceedings Library, vol. 237, pp. null null, 1991. [44] M. J. Aziz, P. C. Sabin, and G. Q. Lu, "The activation strain tensor: Nonhydrostatic stress effects on c rystal growth kinetics," Physical Review B, vol. 44, pp. 9812 9816, 1991. [45] W. Barvosa Carter, M. J. Aziz, L. J. Gray, and T. Kaplan, "Kinetically Driven Growth Instability in Stressed Solids," Physical Review Letters, vol. 81, pp. 1445 1448, 1998. [46] W. Barvosa Carter and M. J. Aziz, "Time resolved measurements of stress effects on solid phase epitaxy of intrinsic and doped Si," Applied Physics Letters, vol. 79, pp. 356 358, Jul 2001. [47] N. G. Rudawski, K. S. Jones, S. Morarka, M. E. Law, and R. G. Elliman, "Stressed multidirectional solid phase epitaxial growth of Si," Journal Of Applied Physics, vol. 105, Apr 2009. [48] J. F. Sage, W. Barvosa Carter, and M. J. Aziz, "Strain stabilized solid phase epitaxy of Si -Ge on Si," Journal Of Applied Physics vol. 99, p. 113529, 2006.

PAGE 144

144 [49] S. G. Mayr and R. S. Averback, "Ion irradiation induced stresses and swelling in amorphous Ge thin films," Physical Review B, vol. 71, p. 8, Apr 2005. [50] D. D'Angelo, A. M. Piro, A. Terrasi, M. G. Grimaldi, S. Mirabella, and C. Bongiorno, "Interface roughening and defect nucleation during solid phase epitaxy regrowth of doped and intrinsic Si[sub 0.83]Ge[sub 0.17] alloys," Journal Of Applied Physics, vol. 101, pp. 103508 6, 2007. [51] N. G. Rudawski, K. S. Jones, and R. Gw illiam, "Stressed solid phase epitaxial growth of ion implanted amorphous silicon," Materials Science & Engineering R Reports, vol. 61, pp. 40 58, May 2008. [52] W. Barvosa Carter, M. J. Aziz, A. V. Phan, T. Kaplan, and L. J. Gray, "Interfacial roughening during solid phase epitaxy: Interaction of dopant, stress, and anisotropy effects," Journal Of Applied Physics, vol. 96, pp. 5462 5468, 2004. [53] A. V. Phan, T. Kaplan, L. J. Gray, D. Adalsteinsson, J. A. Sethian, W. Barvosa Carter, and M. J. Aziz, "Model ling a growth instability in a stressed solid," Modelling and Simulation in Materials Science and Engineering, vol. 9, p. 309, 2001. [54] J. F. Sage, W. Barvosa Carter, and M. J. Aziz, "Morphological instability of growth fronts due to stress induced mobil ity variations," Applied Physics Letters, vol. 77, pp. 516 518, 2000. [55] K. L. Saenger, K. E. Fogel, J. A. Ott, D. K. Sadana, and H. Yin, "An examination of facet formation during solid phase epitaxy of line shaped amorphized regions in (001) and (011) S i," Journal Of Applied Physics, vol. 101, p. 104908, 2007. [56] A. Gouye, I. Berbezier, L. Favre, G. Amiard, M. Aouassa, Y. Campidelli, and A. Halimaoui, "Low temperature solid phase epitaxy for integrating advanced source/drain metal oxide semiconductor s tructures," Applied Physics Letters, vol. 96, p. 063102, 2010. [57] M. D. Rechtin, "An electron microscopy study of defect structures in recrystallized amorphous layers of self ion irradiated 111 silicon," Philosophical magazine. A, Physics of condensed matter. Defects and mechanical properties, vol. 37, pp. 605 620, 1978. [58] L. Csepregi, J. W. Mayer, and T. W. Sigmon, "Chaneling effect measurements of the recrystallization of amorphous Si layers on crystal Si," Physics Letters A, vol. 54, pp. 157 158, 1975. [59] L. Csepregi, E. F. Kennedy, J. W. Mayer, and T. W. Sigmon, "SUBSTRATE ORIENTATION DEPENDENCE OF EPITAXIAL REGROWTH RATE FROM SI IMPLANTED AMORPHOUS SIA," Journal Of Applied Physics, vol. 49, pp. 3906 3911, 1978.

PAGE 145

145 [60] I. Martin Bragado, "Importance of twin defect formation created by solid phase epitaxial growth: An atomistic study," Scripta Materialia, vol. 66, pp. 186 189, 2012. [61] J. C. Bourgoin and R. Asomoza, "Solid phase growt h of silicon and germanium," Journal of Crystal Growth, vol. 69, pp. 489 498, 1984. [62] I. Martin Bragado and B. Sklenard, "Understanding Si(111) solid phase epitaxial regrowth using Monte Carlo modeling: Bi modal growth, defect formation, and interface t opology," Journal Of Applied Physics, vol. 112, p. 024327, 2012. [63] I. Martin Bragado and V. Moroz, "Modeling of {311} facets using a lattice kinetic Monte Carlo three dimensional model for selective epitaxial growth of silicon," Applied Physics Letters, vol. 98, p. 153111, 2011. [64] I. Martin Bragado and V. Moroz, "Facet formation during solid phase epitaxy regrowth: A lattice kinetic Monte Carlo model," Applied Physics Letters, vol. 95, p. 123123, 2009. [65] I. Martin Bragado, "{111} local configuratio ns: The main source of silicon defects during solid phase epitaxial regrowth modeled by lattice kinetic Monte Carlo," Applied Physics Letters, vol. 98, p. 233109, 2011. [66] L. Csepregi, R. P. Kullen, J. W. Mayer, and T. W. Sigmon, "REGROWTH KINETICS OF AM ORPHOUS GE LAYERS CREATED BY GE 74 AND SI 28 IMPLANTATION OF GE CRYSTALS," Solid State Communications, vol. 21, pp. 1019 1021, 1977. [67] S. Morarka, N. G. Rudawski, M. E. Law, K. S. Jones, and R. G. Elliman, "Modeling two dimensional solid phase epitaxial regrowth using level set methods," Journal Of Applied Physics, vol. 105, Mar 1 2009. [68] M. C. Ridgway, R. G. Elliman, and J. S. Williams, "Ion beam induced epitaxial crystallization of NiSi[sub 2]," Applied Physics Letters, vol. 56, pp. 2117 2119, 1990. [69] L. Pelaz, L. A. Marques, and J. Barbolla, "Ion beam induced amorphization and recrystallization in silicon," Journal Of Applied Physics, vol. 96, pp. 5947 5976, 2004. [70] T. K. Chaki, "Ion irradiation induced solid phase epitaxial growth," Philosoph ical Magazine Letters, vol. 59, pp. 223 227, 1989/05/01 1989. [71] J. Linnros and G. Holmen, "Dose rate dependence and time constant of the ion beam induced crystallization mechanism in silicon," Journal Of Applied Physics, vol. 62, pp. 4737 4744, 1987.

PAGE 146

146 [7 2] R. G. Elliman, S. T. Johnson, A. P. Pogany, and J. S. Williams, "Ion beam induced epitaxial crystallisation of silicon," Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms, vol. 7 8, Part 1, pp. 310 315, 1985. [73] J. S. Williams, W. L. Brown, R. G. Elliman, R. V. Knoell, D. M. Maher, and T. E. Seidel, "The Kinetics and Microstructure of Ion Beam Induced Crystallisation of Silicon," MRS Online Proceedings Library, vol. 45, pp. null null, 1985. [74] R G. Elliman, J. S. Williams, D. M. Maher, and W. L. Brown, "Kinetics, Microstructure And Mechanisms of Ion Beam Induced Epitaxial Crystallization of Semiconductors," MRS Online Proceedings Library, vol. 51, pp. null null, 1985. [75] J. Linnros, R. G. Elli man, and W. L. Brown, "The Competition between Ion Beam Induced Epitaxial Crystallization and Amorphization in Silicon: The Role of the Divacancy," MRS Online Proceedings Library, vol. 74, pp. null null, 1986. [76] R. G. Elliman, J. S. Williams, S. T. John son, and E. Nygren, "Ion Beam Induced Amorphization and Crystallization Processes in Silicon and GaAs," MRS Online Proceedings Library, vol. 74, pp. null null, 1986. [77] R. G. Elliman, M. C. Ridgway, J. S. Williams, and J. C. Bean, "Ion beam induced epita xial crystallization of GexSi1 x/Si structures," Applied Physics Letters, vol. 55, pp. 843 845, 1989. [78] N. Kobayashi, H. Katsumata, Y. Makita, M. Hasegawa, N. Hayashi, H. Shibata, S. Uekusa, and S. Hishita, "Ion Beam Induced Epitax y and Solid Phase Epitaxy of Sigec on Si Formed by Ge and C Ion Implantation and their Structural and Optical Properties," MRS Online Proceedings Library, vol. 388, pp. null null, 1995. [79] N. Kobayashi, "Structural characterization of ion beam induced ep itaxially crystallized thin layers of III V and IV IV semiconductors," Thin Solid Films, vol. 270, pp. 307 313, 1995. [80] L. Csepregi, E. F. Kennedy, T. J. Gallagher, J. W. Mayer, and T. W. Sigmon, "REORDERING OF AMORPHOUS LAYERS OF SI IMPLANTED WITH P 31 AS 75, AND B 11 IONS," Journal Of Applied Physics, vol. 48, pp. 4234 4240, 1977. [81] L. Csepregi and E. F. Kennedy, "REGROWTH OF AMORPHOUS SI LAYERS," Journal Of The Electrochemical Society, vol. 124, pp. C287 C288, 1977. [82] L. Csepregi, J. W. Mayer, and T. W. Sigmon, "Regrowth behavior of ion implanted amorphous layers on <111> silicon," Applied Physics Letters, vol. 29, pp. 92 93, 1976. [83] F. Spaepen and D. Turnbull, "Kinetics of Motion of Crystal Interfaces," in AIP 1979.

PAGE 147

147 [84] F. A. Cotton and G. Wilkinson, Advanced Inorganic Chemistry N.Y.: Interscience Publishers, 1972. [85] H. Cerva and K. H. Kusters, "Defect formation in silicon at a mask edge during crystallization of an amorphous implantation layer," Journal Of Applied Physics, vol. 66, pp. 4723 4728, 1989. [86] K. L. Saenger, K. E. Fogel, J. A. Ott, C. Y. Sung, D. K. Sadana, and H. Yin, "Mask edge defects in hybrid orientation direct Si bonded substrates recrystallized by solid phase epitaxy after patterned amorphization," Journal Of Applie d Physics, vol. 101, p. 084912, 2007. [87] S. M. Hu, "Stress related problems in silicon technology," Journal Of Applied Physics, vol. 70, pp. R53 R80, 1991. [88] N. G. Rudawski, K. S. Jones, and R. G. Elliman, "Influence of As on the formation of mask edg e defects during stressed solid phase epitaxy in patterned Si wafers," Journal of Vacuum Science & Technology B: Microelectronics and Nanometer Structures, vol. 26, pp. 435 438, 2008. [89] K. L. Saenger, K. E. Fogel, J. A. Ott, J. P. d. Souza, and C. E. Mu rray, "Effects of patterned, stressed SiN overlayers on Si solid phase epitaxy," Applied Physics Letters, vol. 92, p. 124103, 2008. [90] X. Shiying and J. Bokor, "Sensitivity of double gate and FinFETDevices to process variations," Electron Devices, IEEE T ransactions on, vol. 50, pp. 2255 2261, 2003. [91] D. Nirmal, P. Vijayakumar, D. M. Thomas, B. K. Jebalin, and N. Mohankumar, "Subthreshold performance of gate engineered FinFET devices and circuit with high k dielectrics," Microelectronics Reliability [9 2] J. Kedzierski, I. Meikei, E. Nowak, T. S. Kanarsky, Z. Ying, R. Roy, D. Boyd, D. Fried, and H. S. P. Wong, "Extension and source/drain design for high performance FinFET devices," Electron Devices, IEEE Transactions on, vol. 50, pp. 952 958, 2003. [93] L. Witters, N. J. Son, I. Ferain, T. San, R. Singanamalla, C. Kerner, N. Collaert, K. De Meyer, and M. Jurczak, "Treshold voltage modulation in FinFET devices through Arsenic Ion Implantation into TiN/HfSiON gate stack," in SOI Conference, 2007 IEEE Intern ational 2007, pp. 31 32. [94] R. Duffy, M. J. H. V. Dal, B. J. Pawlak, M. Kaiser, R. G. R. Weemaes, B. Degroote, E. Kunnen, and E. Altamirano, "Solid phase epitaxy versus random nucleation and growth in sub 20 nm wide fin field effect transistors," Applie d Physics Letters, vol. 90, p. 241912, 2007.

PAGE 148

148 [95] E. Lampin and C. Krzeminski, "Regrowth of oxide embedded amorphous silicon studied with molecular dynamics," Journal Of Applied Physics, vol. 109, p. 123509, 2011. [96] M. Shayesteh, R. Duffy, B. McCarthy, A. Blake, M. White, J. Scully, R. Yu, V. Djara, M. Schmidt, N. Petkov, and A. M. Kelleher, "Germanium Fin Structure Optimization for Future MugFET and FinFET Applications," ECS Transactions, vol. 35, pp. 27 34, April 25, 2011 2011. [97] R. Duffy, M. Shayes teh, B. McCarthy, A. Blake, M. White, J. Scully, R. Yu, A. M. Kelleher, M. Schmidt, N. Petkov, L. Pelaz, and L. A. Marques, "The curious case of thin body Ge crystallization," Applied Physics Letters, vol. 99, p. 131910, 2011. [98] K. L. Saenger, J. P. d. Souza, K. E. Fogel, J. A. Ott, C. Y. Sung, D. K. Sadana, and H. Yin, "A study of trench edge defect formation in (001) and (011) silicon recrystallized by solid phase epitaxy," Journal Of Applied Physics, vol. 101, p. 024908, 2007. [99] N. Burbure, N. G. R udawski, and K. S. Jones, "Effect of oxide on trench edge defect formation in ion implanted silicon," Electrochemical and Solid State Letters, vol. 10, pp. H184 H185, 2007. [100] H. Kawarada, T. Ueno, I. Ohdomari, and Y. Kunii, "Characterization of roughne ss and defects at an Si/SiO[sub 2] interface formed by lateral solid phase epitaxy using high resolution electron microscopy," Journal Of Applied Physics, vol. 63, pp. 2641 2644, 1988. [101] M. Omri, C. Bonafos, A. Claverie, A. Nejim, F. Cristiano, D. Alqu ier, A. Martinez, and N. E. B. Cowern, "Is there an effect of the proximity of a ''free surface'' on the formation of End Of Range defects?," Nuclear Instruments & Methods In Physics Research Section B Beam Interactions With Materials And Atoms, vol. 120, pp. 5 8, Dec 1996. [102] K. S. Jones, S. Prussin, and E. R. Weber, "A systematic analysis of defects in ion implanted silicon," Applied Physics A: Materials Science & Processing, vol. 45, pp. 1 34, 1988. [103] T. Noda, "Evolution of end of range damage and transient enhanced diffusion of indium in silicon," Journal Of Applied Physics, vol. 91, pp. 639 645, 2002. [104] P. A. Stolk, H. J. Gossmann, D. J. Eaglesham, D. C. Jacobson, C. S. Rafferty, G. H. Gilmer, M. Jaraiz, J. M. Poate, H. S. Luftman, and T. E. Haynes, "Physical mechanisms of transient enhanced dopant diffusion in ion implanted silicon," Journal Of Applied Physics, vol. 81, pp. 6031 6050, 1997.

PAGE 149

149 [105] A. Agarwal, H. J. Gossmann, D. C. Jacobson, D. J. Eaglesham, M. Sosnowski, J. M. Poate, I. Yamada J. Matsuo, and T. E. Haynes, "Transient enhanced diffusion from decaborane molecular ion implantation," Applied Physics Letters, vol. 73, pp. 2015 2017, 1998. [106] D. J. Eaglesham, P. A. Stolk, H. J. Gossmann, and J. M. Poate, "Implantation and transien t B diffusion in Si: The source of the interstitials," Applied Physics Letters, vol. 65, pp. 2305 2307, 1994. [107] D. R. Lim, C. S. Rafferty, and F. P. Klemens, "The role of the surface in transient enhanced diffusion," Applied Physics Letters, vol. 67, p p. 2302 2304, 1995. [108] B. J. Pawlak, R. Surdeanu, B. Colombeau, A. J. Smith, N. E. B. Cowern, R. Lindsay, W. Vandervorst, B. Brijs, O. Richard, and F. Cristiano, "Evidence on the mechanism of boron deactivation in Ge preamorphized ultrashallow junctions ," Applied Physics Letters, vol. 84, pp. 2055 2057, 2004. [109] F. Cristiano, N. Cherkashin, P. Calvo, Y. Lamrani, X. Hebras, A. Claverie, W. Lerch, and S. Paul, "Thermal stability of boron electrical activation in preamorphised ultra shallow junctions," M aterials Science and Engineering: B, vol. 114 115, pp. 174 179, 2004. [110] F. Panciera, P. F. Fazzini, M. Collet, J. Boucher, E. Bedel, and F. Cristiano, "End of range defects in germanium and their role in boron deactivation," Applied Physics Letters, vo l. 97, p. 012105, 2010. [111] S. Koffel, N. Cherkashin, F. Houdellier, M. J. Hytch, G. Benassayag, P. Scheiblin, and A. Claverie, "End of range defects in Ge," Journal Of Applied Physics, vol. 105, p. 3, Jun 15 2009. [112] A. Claverie, S. Koffel, N. Cherka shin, G. Benassayag, and P. Scheiblin, "Amorphization, recrystallization and end of range defects in germanium," Thin Solid Films, vol. 518, pp. 2307 2313, 2009 2010. [113] Y. Erokhin, "Device scaling and performance improvement: Advances in ion implantati on and annealing technologies as enabling drivers," in Junction Technology (IWJT), 2012 12th International Workshop on 2012, pp. 12 17. [114] J. R. Liefting, J. S. Custer, R. J. Schreutelkamp, and F. W. Saris, "Dislocation formation in silicon implanted a t different temperatures," Materials Science and Engineering: B, vol. 15, pp. 173 186, 1992. [115] K. S. Jones and D. Venables, "The effect of implant energy, dose, and dynamic annealing on end‐of‐range damage in Ge+‐implant ed silicon," Journal Of Applied Physics, vol. 69, pp. 2931 2937, 1991.

PAGE 150

150 [116] E. Napolitani, G. Bisognin, E. Bruno, M. Mastromatteo, G. G. Scapellato, S. Boninelli, D. D. Salvador, S. Mirabella, C. Spinella, A. Carnera, and F. Priolo, "Transient enhanced di ffusion of B mediated by self interstitials in preamorphized Ge," Applied Physics Letters, vol. 96, p. 201906, 2010. [117] S. Koffel, P. Scheiblin, A. Claverie, and G. Benassayag, "Amorphization kinetics of germanium during ion implantation," Journal Of Ap plied Physics, vol. 105, p. 013528, Jan 1 2009. [118] M. K. Elghor, O. W. Holland, C. W. White, and S. J. Pennycook, "STRUCTURAL CHARACTERIZATION OF DAMAGE IN SI(100) PRODUCED BY MEV SI+ ION IMPLANTATION AND ANNEALING," Journal of Materials Research, vol. 5, pp. 352 359, Feb 1990. [119] C. Carter, W. Maszara, D. K. Sadana, G. A. Rozgonyi, J. Liu, and J. Wortman, "Residual defects following rapid thermal annealing of shallow boron and boron fluoride implants into preamorphized silicon," Applied Physics Lette rs, vol. 44, pp. 459 461, 1984. [120] C. W. Nieh and L. J. Chen, "Cross sectional transmission electron microscope study of residual defects in BF[sup + ][sub 2] implanted (001)Si," Journal Of Applied Physics, vol. 60, pp. 3114 3119, 1986. [121] D. B. Will iams and C. B. Carter, Transmission Electron Microscopy: A Textbook for Materials Science. Spectrometry : Plenum Press, 1996. [122] N. A. Sobolev, "Defects and their influence on the luminescence of rare earth ions implanted in single crystal Si," Physica B : Condensed Matter, vol. 308 310, pp. 333 336, 2001. [123] S. Onishi, A. Ayukawa, K. Tanaka, and K. Sakiyama, "A Mechanism of the Sidewall Process Induced Junction Leakage Current of LDD Structure," Journal Of The Electrochemical Society, vol. 138, pp. 143 9 1443, 1991. [124] V. F. Drobny and J. Rubalcava, "Doping characteristics of BF2+ implants in ⟨100⟩ and ⟨111⟩ silicon," Electron Devices, IEEE Transactions on, vol. 48, pp. 1661 1666, 2001. [125] G. Queirolo, C. B resolin, D. Robba, M. Anderle, R. Canteri, A. Armigliato, G. Ottaviani, and S. Frabboni, "Low temperature dopant activation of BF<sub>2</sub> implanted silicon," Journal of Electronic Materials, vol. 20, pp. 373 378, 1991. [126] C. W. Nieh and L. J. Chen, "Cross sectional transmission electron microscope study of residual defects in BF[sup + ][sub 2] implanted (111) Si," Journal Of Applied Physics, vol. 62, pp. 4421 4425, 1987.

PAGE 151

151 [127] S. o. P. o. I. Engineers, S. Equipment, and M. International, Materials and Device Characterization in Micromachining : SPIE, 2000. [128] E. Irene, "Residual stress in silicon nitride films," Journal of Electronic Materials, vol. 5, pp. 287 298, 1976. [129] C. Jian and I. De Wolf, "Theoretical and experimental Raman s pectroscopy study of mechanical stress induced by electronic packaging," Components and Packaging Technologies, IEEE Transactions on, vol. 28, pp. 484 492, 2005. [130] I. D. Wolf, H. E. Maes, and S. K. Jones, "Stress measurements in silicon devices through Raman spectroscopy: Bridging the gap between theory and experiment," Journal Of Applied Physics, vol. 79, pp. 7148 7156, 1996. [131] S. Mayuzumi, S. Yamakawa, D. Kosemura, M. Takei, Y. Tateshita, H. Wakabayashi, M. Tsukamoto, T. Ohno, A. Ogura, and N. Nag ashima, "Channel Stress Enhancement Characteristics for Scaled pMOSFETs by Using Damascene Gate With Top Cut Compressive Stress Liner and eSiGe," Electron Devices, IEEE Transactions on, vol. 56, pp. 2778 2784, 2009. [132] E. Anastassakis, A. Pinczuk, E. Bu rstein, F. H. Pollak, and M. Cardona, "Effect of static uniaxial stress on the Raman spectrum of silicon," Solid State Communications, vol. 8, pp. 133 138, 1970. [133] D. Olego and M. Cardona, "Pressure dependence of Raman phonons of Ge and 3C SiC," Physic al Review B, vol. 25, pp. 1151 1160, 1982. [134] F. Cerdeira, C. J. Buchenauer, F. H. Pollak, and M. Cardona, "Stress Induced Shifts of First Order Raman Frequencies of Diamond and Zinc Blende Type Semiconductors," Physical Review B, vol. 5, pp. 580 593, 1972. [135] M. Takei, H. Hashiguchi, T. Yamaguchi, D. Kosemura, K. Nagata, and A. Ogura, "Channel strain measurements in 32nm node CMOSFETs," Technical report of IEICE. SDM, vol. 111, pp. 43 48, 2011/10/13 2011. [136] S. S. M, Physics of Semiconductor Devi ces : Wiley Eastern Limited, 1981. [137] R. G. Sparks and M. A. Paesler, "Micro Raman analysis of stress in machined silicon and germanium," Precision Engineering, vol. 10, pp. 191 198, 1988. [138] L. Kwan Yong, L. Hyunjung, R. Choongryul, S. Kang Ill, K. U ihui, K. Seokhoon, C. Jongwan, O. Kyungseok, J. Hee Kyung, S. Chulgi, K. Tae Ouk, C. Jinyeong, L. Seunghun, S. Yangsoo, Y. Hong Sik, P. Junghyun, L. Kwanheum, K. Wookje, L. Eunha, S. Sang Pil, K. Chung Geun, K. Sang Bom, C. Siyoung, and C. Chilhee, "Novel stress memorization technology (SMT) for high electron mobility enhancement of gate last high k/metal gate devices," in Electron Devices Meeting (IEDM), 2010 IEEE International 2010, pp. 10.1.1 10.1.4.

PAGE 152

152 [139] J. F. Ziegler, "SRIM 2003," Nuclear Instruments & Methods In Physics Research Section B Beam Interactions With Materials And Atoms, vol. 219, pp. 1027 1036, Jun 2004. [140] A. Bousetta, J. A. v. d. Berg, D. G. Armour, and P. C. Zalm, "Si ultrashallow p[sup + ]n junctions using low energy boron implanta tion," Applied Physics Letters, vol. 58, pp. 1626 1628, 1991. [141] A. Satta, E. Simoen, T. Janssens, T. Clarysse, B. De Jaeger, A. Benedetti, I. Hoflijk, B. Brijs, M. Meuris, and W. Vandervorst, "Shallow junction ion implantation in Ge and associated defe ct control," Journal Of The Electrochemical Society, vol. 153, pp. G229 G233, 2006. [142] E. Simoen, A. Satta, A. D'Amore, T. Janssens, T. Clarysse, K. Martens, B. De Jaeger, A. Benedetti, I. Hoflijk, B. Brijs, M. Meuris, and W. Vandervorst, "Ion implantat ion issues in the formation of shallow junctions in germanium," Materials Science In Semiconductor Processing, vol. 9, pp. 634 639, Aug Oct 2006. [143] C. O. Chui, K. Gopalakrishnan, P. B. Griffin, J. D. Plummer, and K. C. Saraswat, "Activation and diffusi on studies of ion implanted p and n dopants in germanium," Applied Physics Letters, vol. 83, pp. 3275 3277, Oct 20 2003. [144] D. Kuzum, A. J. Pethe, T. Krishnamohan, and K. C. Saraswat, "Ge (100) and (111) N and P FETs With High Mobility and Low T Mobili ty Characterization," Ieee Transactions On Electron Devices, vol. 56, pp. 648 655, Apr 2009. [145] B. L. Darby, B. R. Yates, N. G. Rudawski, K. S. Jones, and A. Kontos, "Self implantation energy and dose effects on Ge solid phase epitaxial growth," Nuclear Instruments & Methods In Physics Research Section B Beam Interactions With Materials And Atoms, vol. 269, pp. 20 22, Jan 2011. [146] L. Kovarik, R. Unocic, J. Li, and M. Mills, "The intermediate temperature deformation of Ni based superalloys: Importance of reordering," JOM Journal of the Minerals, Metals and Materials Society, vol. 61, pp. 42 48, 2009. [147] R. N. Kyutt, N. A. Sobolev, Y. A. Nikolaev, and V. I. Vdovin, "Defect structure of erbium doped 1 1 1 silicon layers formed by solid phase epitaxy," Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms, vol. 173, pp. 319 325, 2001. [148] P. J. H. Denteneer and W. v. Haeringen, "Stacking fault energies in semiconductors from first principles calculations," Journal of Physics C: Solid State Physics, vol. 20, p. L883, 1987. [149] Y. L. Chao, S. Prussin, J. C. S. Woo, and R. Scholz, "Preamorphization implantation assisted boron activation in bulk germanium and germanium on insulator," Applied Physics Letters, vol. 87, 2005.

PAGE 153

153 [150] J. M. Poate, S. Coffa, D. C. Jacobson, A. Polman, J. A. Roth, G. L. Olson, S. Roorda, W. Sinke, J. S. Custer, M. O. Thompson, F. Spaepen, and E. Donovan, "AMORPHOUS SI THE ROLE OF MEV IMPLANTATION IN ELUCIDATING DEFECT AND THERMODYNAMIC PROPERTIES," Nuclear Instruments & Methods In Physics Research Section B Beam Interactions With Materials And Atoms, vol. 55, pp. 533 543, Apr 1991. [151] S. Morarka, S. Jin, N. G. R udawski, K. S. Jones, M. E. Law, and R. G. Elliman, "Interface stability in stressed solid phase epitaxial growth," Journal of Vacuum Science & Technology B: Microelectronics and Nanometer Structures, vol. 29, p. 041210, 2011. [152] C. R. Olson, E. Kuryliw B. E. Jones, and K. S. Jones, "Effect of stress on the evolution of mask edge defects in ion implanted silicon," Journal of Vacuum Science & Technology B: Microelectronics and Nanometer Structures, vol. 24, pp. 446 449, 2006. [153] E. W. Cory, M. C. Step hen, D. Hemant, G. Oleg, and Y. L. Mark, "Modeling of NMOS performance gains from edge dislocation stress," pp. 34 34, 2011. [154] S. Flachowsky, A. Wei, R. Illgen, T. Herrmann, Ho, x, J. ntschel, M. Horstmann, W. Klix, and R. Stenzel, "Understanding Strai n Induced Drive Current Enhancement in Strained Silicon n MOSFET and p MOSFET," Electron Devices, IEEE Transactions on, vol. 57, pp. 1343 1354, 2010. [155] N. Tamura and Y. Shimamune, "45 nm CMOS technology with low temperature selective epitaxy of Si Ge," Applied Surface Science, vol. 254, pp. 6067 6071, 2008. [156] N. Fukata, K. Sato, M. Mitome, Y. Bando, T. Sekiguchi, M. Kirkham, J. i. Hong, Z. L. Wang, and R. L. Snyder, "Doping and Raman Characterization of Boron and Phosphorus Atoms in Germanium Na nowires," ACS Nano, vol. 4, pp. 3807 3816, 2010/07/27 2010. [157] J. Laane and W. Kiefer, "Determination of frequency shifts by Raman difference spectroscopy," The Journal of Chemical Physics, vol. 72, pp. 5305 5311, 1980. [158] M. E. Law and S. M. Cea, "C ontinuum based modeling of silicon integrated circuit processing: An object oriented approach," Computational Materials Science, vol. 12, pp. 289 308, 1998. [159] J. A. Sethian, "Evolution, implementation, and application of level set and fast marching met hods for advancing fronts," Journal of Computational Physics, vol. 169, pp. 503 555, May 2001. [160] B. Drosd and J. Washburn, "A new technique for observing the amorphous to crystalline transformation in thin surface layers on silicon wafers," Journal Of Applied Physics, vol. 51, pp. 4106 4110, 1980.

PAGE 154

154 [161] H. Cerva and W. Bergholz, "IMPLANTATION DEFECTS BELOW MASK EDGES IN SILICON STRUCTURE AND EFFECT ON DEVICE," Journal Of The Electrochemical Society, vol. 140, pp. 780 786, Mar 1993. [162] I. D. Wolf, J Vanhellemont, A. Romano Rodriguez, H. Norstrom, and H. E. Maes, "Micro Raman study of stress distribution in local isolation structures and correlation with transmission electron microscopy," Journal Of Applied Physics, vol. 71, pp. 898 906, 1992. [163] I. D. Wolf, "Micro Raman spectroscopy to study local mechanical stress in silicon integrated circuits," Semiconductor Science and Technology, vol. 11, p. 139, 1996. [164] I. De Wolf, "Raman spectroscopy: About chips and stress," Spectroscopy Europe, vol. 1 5, pp. 6+8+10+12 13, 2003. [165] S. Boninelli, G. Impellizzeri, A. Alberti, F. Priolo, F. Cristiano, and C. Spinella, "Role of the Ge surface during the end of range dissolution," Applied Physics Letters, vol. 101, p. 162103, 2012. [166] N. Yasutake, A. Az uma, T. Ishida, K. Ohuchi, N. Aoki, N. Kusunoki, S. Mori, I. Mizushima, T. Morooka, S. Kawanaka, and Y. Toyoshima, "A high performance pMOSFET with two step recessed SiGe S/D structure for 32 nm node and beyond," Solid State Electronics, vol. 51, pp. 1437 1443, 2007. [167] K. S. Chen, X. Zhang, and S. Y. Lin, "Intrinsic stress generation and relaxation of plasma enhanced chemical vapor deposited oxide during deposition and subsequent thermal cycling," Thin Solid Films, vol. 434, pp. 190 202, 2003. [168 ] P. Paduschek, C. Hpfl, and H. Mitlehner, "Hydrogen related mechanical stress in amorphous silicon and plasma deposited silicon nitride," Thin Solid Films, vol. 110, pp. 291 304, 1983. [169] M. P. Hughey and R. F. Cook, "Stress development kinetics in pl asma enhanced chemical vapor deposited silicon nitride films," Journal Of Applied Physics, vol. 97, p. 114914, 2005. [170] M. Martyniuk, J. Antoszewski, C. A. Musca, J. M. Dell, and L. Faraone, "Stress Response of Low Temperature PECVD Silicon Nitride Thin Films to Cryogenic Thermal Cycling," in Optoelectronic and Microelectronic Materials and Devices, 2004 Conference on 2004, pp. 381 384. [171] C. E. Murray, K. L. Saenger, O. Kalenci, S. M. Polvino, I. C. Noyan, B. Lai, and Z. Cai, "Submicron mapping of s ilicon on insulator strain distributions induced by stressed liner structures," Journal Of Applied Physics, vol. 104, p. 013530, 2008.

PAGE 155

155 [172] K. Usuda, T. Irisawa, T. Numata, N. Hirashita, and S. Takagi, "Characterization of in plane strain relaxation in st rained layers using a newly developed plane nano beam electron diffraction (plane NBD) method," Semiconductor Science and Technology, vol. 22, p. S227, 2007. [173] J. P. Liu, K. Li, S. M. Pandey, F. L. Benistant, A. See, M. S. Zhou, L. C. Hsia, R. Schamper s, and D. O. Klenov, "Strain relaxation in transistor channels with embedded epitaxial silicon germanium source/drain," Applied Physics Letters, vol. 93, p. 221912, 2008. [174] S. Mayuzumi, S. Yamakawa, D. Kosemura, M. Takei, K. Nagata, H. Akamatsu, H. Wak abayashi, K. Amari, Y. Tateshita, M. Tsukamoto, T. Ohno, A. Ogura, and N. Nagashima, "Mobility and Velocity Enhancement Effects of High Uniaxial Stress on Si (100) and (110) Substrates for Short Channel pFETs," Electron Devices, IEEE Transactions on, vol. 57, pp. 1295 1300, 2010. [175] K. L. Saenger, J. P. d. Souza, K. E. Fogel, J. A. Ott, A. Reznicek, C. Y. Sung, D. K. Sadana, and H. Yin, "Amorphization/templated recrystallization method for changing the orientation of single crystal silicon: An alternativ e approach to hybrid orientation substrates," Applied Physics Letters, vol. 87, p. 221911, 2005. [176] P. W. Loscutoff and S. F. Bent, "REACTIVITY OF THE GERMANIUM SURFACE: Chemical Passivation and Functionalization," Annual Review of Physical Chemistry, v ol. 57, pp. 467 495, 2006. [177] R. J. Kaiser, S. Koffel, P. Pichler, A. J. Bauer, B. Amon, L. Frey, and H. Ryssel, "Germanium substrate loss during thermal processing," Microelectronic Engineering, vol. 88, pp. 499 502, 2011. [178] N. Ioannou, D. Skarlato s, C. Tsamis, C. A. Krontiras, S. N. Georga, A. Christofi, and D. S. McPhail, "Germanium substrate loss during low temperature annealing and its influence on ion implanted phosphorous dose loss," Applied Physics Letters, vol. 93, p. 101910, 2008. [179] S. R. M. d. Silva, G. K. Rolim, G. V. Soares, I. J. R. Baumvol, C. Krug, L. Miotti, J. F. L. Freire, M. E. H. M. d. Costa, and C. Radtke, "Oxygen transport and GeO[sub 2] stability during thermal oxidation of Ge," Applied Physics Letters, vol. 100, p. 191907, 2012. [180] Y. Oniki, H. Koumo, Y. Iwazaki, and T. Ueno, "Evaluation of GeO desorption behavior in the metal/GeO[sub 2]/Ge structure and its improvement of the electrical characteristics," Journal Of Applied Physics, vol. 107, p. 124113, 2010. [181] B. E. Deal and A. S. Grove, "General Relationship for the Thermal Oxidation of Silicon," Journal Of Applied Physics, vol. 36, pp. 3770 3778, 1965.

PAGE 156

156 [182] A. Sakai, "Characterization of wafer bonded substrates for advanced channels in Si based MOSFET," in Solid S tate and Integrated Circuit Technology (ICSICT), 2010 10th IEEE International Conference on 2010, pp. 1517 1520. [183] A. Toriumi, T. Tabata, C. H. Lee, T. Nishimura, K. Kita, and K. Nagashio, "Opportunities and challenges for Ge CMOS Control of interfa cing field on Ge is a key," Microelectronic Engineering, vol. 86, pp. 1571 1576, Jul Sep 2009. [184] F. Bellenger, B. De Jaeger, C. Merckling, M. Houssa, J. Penaud, L. Nyns, E. Vrancken, M. Caymax, M. Meuris, T. Hoffmann, K. De Meyer, and M. Heyns, "High F ET Performance for a Future CMOS GeO2 Based Technology," Ieee Electron Device Letters, vol. 31, pp. 402 404, May 2010. [185] S. Mirabella, G. Impellizzeri, A. M. Piro, E. Bruno, and M. G. Grimaldi, "Activation and carrier mobility in high fluence B implant ed germanium," Applied Physics Letters, vol. 92, p. 251909, 2008. [186] W. O. Adekoya, M. Hageali, J. C. Muller, and P. Siffert, "ON THE KINETICS OF SOLID PHASE REGROWTH AND DOPANT ACTIVATION DURING RAPID THERMAL ANNEALING OF IMPLANTATION AMORPHIZED SILICO N," Journal Of Applied Physics, vol. 64, pp. 666 676, Jul 1988. [187] I. H. Wilson, "THE EFFECTS OF SELF ION BOMBARDMENT (30 500 KEV) ON THE SURFACE TOPOGRAPHY OF SINGLE CRYSTAL GERMANIUM," Journal Of Applied Physics, vol. 53, pp. 1698 1705, 1982. [188] H. Huber, W. Assmann, S. A. Karamian, A. Mucklich, W. Prusseit, E. Gazis, R. Grotzschel, M. Kokkoris, E. Kossionidis, H. D. Mieskes, and R. Vlastou, "Void formation in Ge induced by high energy heavy ion irradiation," Nuclear Instruments & Methods In Physics Research Section B Beam Interactions With Materials And Atoms, vol. 122, pp. 542 546, Feb 1997. [189] H. Huber, W. Assmann, R. Grotzschel, H. D. Mieskes, A. Mucklich, H. Nolte, and W. Prusseit, "Void formation and surface rippling in Ge induced by high en ergetic Au irradiation," Materials Science Applications Of Ion Beam Techniques, vol. 248 pp. 301 312, 1997. [190] B. R. Appleton, O. W. Holland, J. Narayan, O. E. Schow, J. S. Williams, K. T. Short, and E. Lawson, "CHARACTERIZATION OF DAMAGE IN ION IMPLA NTED GE," Applied Physics Letters, vol. 41, pp. 711 712, 1982. [191] E. M. Lawson, K. T. Short, J. S. Williams, B. R. Appleton, O. W. Holland, and O. E. Schow, "ANOMALOUS NEAR SURFACE EFFECTS IN ROOM TEMPERATURE IMPLANTED GERMANIUM," Nuclear Instruments & Methods in Physics Research, vol. 209, pp. 303 307, 1983.

PAGE 157

157 [192] O. W. Holland, B. R. Appleton, and J. Narayan, "ION IMPLANTATION DAMAGE AND ANNEALING IN GERMANIUM," Journal Of Applied Physics, vol. 54, pp. 2295 2301, 1983. [193] G. Peto, J. Kanski, and G. Holmen, "PROPERTIES OF ULTRAHIGH VACUUM SELF IMPLANTATION INDUCED AMORPHOUS GERMANIUM," Applied Physics Letters, vol. 55, pp. 692 693, Aug 1989. [194] I. B. Khaibullin, G. G. Zakirov, M. M. Zaripov, T. Lohneb, L. Pogny, G. Mezey, M. Fried, E. Ktai, F. P szti, A. Manuaba, and J. Gyulai, "Effect of Heavy Ion Implantation and Laser Annealing on the Structural Properties of Germanium," physica status solidi (a), vol. 94, pp. 371 377, 1986. [195] L. M. Wang and R. C. Birtcher, "AMORPHIZATION, MORPHOLOGICAL INS TABILITY AND CRYSTALLIZATION OF KRYPTON ION IRRADIATED GERMANIUM," Philosophical Magazine a Physics of Condensed Matter Structure Defects and Mechanical Properties, vol. 64, pp. 1209 1223, Dec 1991. [196] T. Steinbach, J. Wernecke, P. Kluth, M. C. Ridgway, and W. Wesch, "Structural modifications of low energy heavy ion irradiated germanium," Physical Review B, vol. 84, p. 104108, 2011. [197] B. L. Darby, B. R. Yates, N. G. Rudawski, K. S. Jones, A. Kontos, and R. G. Elliman, "Insights for void formation in ion implanted Ge," Thin Solid Films, vol. 519, pp. 5962 5965, 2011. [198] B. R. Appleton, O. W. Holland, D. B. Poker, J. Narayan, and D. Fathy, "MORPHOLOGICAL INSTABILITIES AND ION BEAM MIXING IN GE," Nuclear Instruments & Methods In Physics Research Secti on B Beam Interactions With Materials And Atoms, vol. 7 8, pp. 639 644, 1985. [199] S. Decoster and A. Vantomme, "Implantation induced damage in Ge: strain and disorder profiles during defect accumulation and recovery," Journal Of Physics D Applied Physics vol. 42, Aug 21 2009. [200] N. G. Rudawski, K. S. Jones, and R. Gwilliam, "Kinetics and morphological instabilities of stressed solid solid phase transformations," Physical Review Letters, vol. 100, Apr 2008. [201] C. J. Glover, M. C. Ridgway, K. M. Yu, G. J. Foran, D. Desnica Frankovic, C. Clerc, J. L. Hansen, and A. Nylandsted Larsen, "Structural relaxation induced bond length and bond angle changes in amorphized Ge," Physical Review B, vol. 63, Feb 2001. [202] M. C. Ridgway, C. J. Glover, K. M. Yu, G. J. Foran, C. Clerc, J. L. Hansen, and A. N. Larsen, "Ion dose dependent microstructure in amorphous Ge," Physical Review B, vol. 61, pp. 12586 12589, May 2000.

PAGE 158

158 [203] M. C. Ridgway, C. J. Glover, I. D. Desnica Frankovic, K. Furic, K. M. Yu, G. J. Foran, C. Clerc, J. L. Hansen, and A. N. Larsen, "Implantation induced disorder in amorphous Ge: Production and relaxation," Nuclear Instruments & Methods In Physics Research Section B Beam Interactions With Materials And Atoms, vol. 175, pp. 21 25, Apr 2001. [204] M. L. Theye, A. Gheorghiu, M. Gandais, and S. Fisson, "STRUCTURAL RELAXATION AND CRYSTALLIZATION OF AMORPHOUS GE FILMS," Journal of Non Crystalline Solids, vol. 37, pp. 301 323, 1980. [205] S. Roorda, W. C. Sinke, J. M. Poate, D. C. Jacobson, S. Dierker, B S. Dennis, D. J. Eaglesham, F. Spaepen, and P. Fuoss, "STRUCTURAL RELAXATION AND DEFECT ANNIHILATION IN PURE AMORPHOUS SILICON," Physical Review B, vol. 44, pp. 3702 3725, Aug 1991. [206] C. E. Bouldin, R. A. Forman, M. I. Bell, and E. P. Donovan, "STRUC TURAL RELAXATION IN ION DAMAGED AMORPHOUS GERMANIUM," Physical Review B, vol. 44, pp. 5492 5496, Sep 1991. [207] R. Kogler, A. Mucklich, W. Skorupa, A. Peeva, A. Y. Kuznetsov, J. S. Christensen, and B. G. Svensson, "Excess vacancies in high energy ion impl anted SiGe," Journal Of Applied Physics, vol. 101, p. 033508, Feb 2007. [208] L. Romano, G. Impellizzeri, M. V. Tomasello, F. Giannazzo, C. Spinella, and M. G. Grimaldi, "Nanostructuring in Ge by self ion implantation," Journal Of Applied Physics, vol. 107 p. 084314, Apr 2010. [209] G. Impellizzeri, L. Romano, L. Bosco, C. Spinella, and M. G. Grimaldi, "Nanoporosity Induced by Ion Implantation in Germanium Thin Films Grown by Molecular Beam Epitaxy," Applied Physics Express, vol. 5, p. 035201, 2012. [210] L. Ottaviano, A. Verna, V. Grossi, P. Parisse, S. Piperno, M. Passacantando, G. Impellizzeri, and F. Priolo, "Surface morphology of Mn+ implanted Ge(100): A systematic investigation as a function of the implantation substrate temperature," Surface Science, vol. 601, pp. 2623 2627, Jul 2007. [211] Y. J. Chen, I. H. Wilson, W. Y. Cheung, J. B. Xu, and S. P. Wong, "Ion implanted nanostructures on Ge(111) surfaces observed by atomic force microscopy," Journal of Vacuum Science & Technology B, vol. 15, pp. 809 8 13, Jul Aug 1997. [212] S. Rubanov and P. R. Munroe, "The application of FIB milling for specimen preparation from crystalline germanium," Micron, vol. 35, pp. 549 556, 2004. [213] M. Ghaly, K. Nordlund, and R. S. Averback, "Molecular dynamics investigatio ns of surface damage produced by kiloelectronvolt self bombardment of solids," Philosophical Magazine a Physics of Condensed Matter Structure Defects and Mechanical Properties, vol. 79, pp. 795 820, Apr 1999.

PAGE 159

159 [214] J. C. Kim, D. G. Cahill, and R. S. Averba ck, "Surface defects created by 20 keV Xe ion irradiation of Ge(111)," Surface Science, vol. 574, pp. 175 180, Jan 2005. [215] P. Bellon, S. J. Chey, J. E. Vannostrand, M. Ghaly, D. G. Cahill, and R. S. Averback, "SURFACE DAMAGE PRODUCED BY 20 KEV GA BOMBA RDMENT OF GE(001)," Surface Science, vol. 339, pp. 135 141, Sep 1995. [216] B. R. Yates, B. L. Darby, R. G. Elliman, and K. S. Jones, "Role of nucleation sites on the formation of nanoporous Ge," Applied Physics Letters, vol. 101, p. 131907, 2012. [217] R. J. Kaiser, S. Koffel, P. Pichler, A. J. Bauer, B. Amon, A. Claverie, G. Benassayag, P. Scheiblin, L. Frey, and H. Ryssel, "Honeycomb voids due to ion implantation in germanium," Thin Solid Films, vol. 518, pp. 2323 2325, Feb 2010. [218] J. C. Kim and Y. W Kim, "Kinetics of bulk point defects in the growth of nanocavities in crystalline Ge," Applied Surface Science, vol. 229, pp. 19 23, May 2004. [219] N. G. Rudawski, B. L. Darby, B. R. Yates, K. S. Jones, R. G. Elliman, and A. A. Volinsky, "Nanostructured ion beam modified Ge films for high capacity Li ion battery anodes," Applied Physics Letters, vol. 100, p. 083111, 2012. [220] D. Sun, A. E. Riley, A. J. Cadby, E. K. Richman, S. D. Korlann, and S. H. Tolbert, "Hexagonal nanoporous germanium through surfa ctant driven self assembly of Zintl clusters," Nature, vol. 441, pp. 1126 1130, 2006. [221] J. H. Lee and J. C. Grossman, "Thermoelectric properties of nanoporous Ge," Applied Physics Letters, vol. 95, p. 013106, 2009. [222] Y. He, D. Donadio, and G. Galli "Morphology and Temperature Dependence of the Thermal Conductivity of Nanoporous SiGe," Nano Letters, vol. 11, pp. 3608 3611, 2011/09/14 2011. [223] H. Foll, J. Carstensen, and S. Frey, "Porous and nanoporous semiconductors and emerging applications," J. Nanomaterials, vol. 2006, pp. 25 25, 2006. [224] B. Stritzker, R. G. Elliman, and J. Zou, "Self ion induced swelling of germanium," Nuclear Instruments & Methods In Physics Research Section B Beam Interactions With Materials And Atoms, vol. 175, pp. 193 1 96, Apr 2001.

PAGE 160

160 BIOGRAPHICAL SKETCH Blake Darby grew up in Virginia Beach, VA and graduated from the International Baccalaureate program at Princess Anne High School He went on to study Materials Science and Engineering at Carnegie Mellon University, where he completed internships at Nucor Steel and Arcelormittal. While at CMU, Blake competed on the golf team and lettered all 4 years. After graduating in 2008, he went to the University of Florida and earned his degree in Materials Science an d Engineering in 2010. While pursuing a Ph.D. at UF, Blake completed an internship at Varian Semiconductor Equipment Associates in the summer of 2011, where he worked on the deposition and applications of porous semiconductors for lithium ion batteries. dissertation topic is about CMOS technology, he helped file 3 patent applications for lithium ion batteries during his graduate career. In his spare time, Blake enjoys playing golf, tennis, violin, and mandolin. After graduatio n, Blake plans on accepting a job offer with Intel, where he will make it rain in Portland, Oregon.