Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The effect of short term anneals on the cathodoluminescence of GaAs Third, Christine Elizabeth 1990

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-UBC_1991_A1 T53.pdf [ 15.03MB ]
Metadata
JSON: 831-1.0078584.json
JSON-LD: 831-1.0078584-ld.json
RDF/XML (Pretty): 831-1.0078584-rdf.xml
RDF/JSON: 831-1.0078584-rdf.json
Turtle: 831-1.0078584-turtle.txt
N-Triples: 831-1.0078584-rdf-ntriples.txt
Original Record: 831-1.0078584-source.json
Full Text
831-1.0078584-fulltext.txt
Citation
831-1.0078584.ris

Full Text

T H E EFFECT OF SHORT T E R M A N N E A L S O N THE C A T H O D O L U M I N E S C E N C E OF GaAs By CHRISTINE E L I Z A B E T H THIRD B . A . S c , The University of British Columbia, 1981 M . A . S c , The University of British Columbia, 1984 A THESIS SUBMITTED IN P A R T I A L F U L F I L L M E N T OF THE REQUIREMENTS FOR THE D E G R E E OF DOCTOR OF PHILOSOPHY IN THE F A C U L T Y OF G R A D U A T E STUDIES D E P A R T M E N T OF M E T A L S A N D M A T E R I A L S ENGINEERING We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH C O L U M B I A December 1990 © Christine Elizabeth Third, 1990 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of Metals & Mater ia ls Engineering The University of British Columbia Vancouver, Canada Date December 18, 1990  DE-6 (2/88) i i A B S T R A C T This study examines the effect of furnace annealing and rapid thermal annealing (RTA) on the room-temperature cathodoluminescence (CL) image of liquid encapsulated Czochralski (LEC) GaAs substrates. Furnace annealed samples were heated in a tube furnace for 20 minutes at tem-peratures from 600 to 950 °C. R T A annealed samples were heated in a commercial R T A furnace for 5 s at temperatures from 650 to 950 °C. The times used for both methods are typical of those used for post-ion implantation annealing although selected samples were R T A annealed at times from 10 to 160 s. The temperature range examined has been extended beyond typical post-ion implantation anneal temperatures to investigate the effect of temperature on the substrate. Examination of R T A annealed GaAs using C L has not been reported previous to this investigation. The C L images of the annealed samples are compared with those of the as-received (un-annealed) material. The C L images of the L E C GaAs wafers prior to annealing have dark spots which correspond to the location of dislocations with regions of higher C L intensity surrounding them. These regions of higher C L intensity are referred to as 'halos'. The remaining material has a C L intensity lower than these halos. The dislocations in L E C GaAs form into cellular networks to reduce the strain energy in the crystal. When viewed at low magnification the overlap of the halos makes the cell walls appear bright and the cell interiors appear dark in a C L image. The furnace annealed substrates show an increase in C L intensity in the interior of the cells. The halos are still present at the cell walls but a region of low C L intensity persists outside the halos making the cell walls appear dark with a brighter interior. This behaviour was seen in all the furnace annealed material although the contrast decreases with increasing anneal temperature. i i i The R T A anneal samples show similar behaviour to the furnace annealed samples at tem-peratures below 800 °C. Above this temperature the halos are no longer noticeable in the surface C L images, although the region of lower C L intensity can be seen along some cell walls. When a cleaved cross-section is examined using C L , there are regions of higher C L intensity adjacent to both surfaces. These regions typically extend from 100 to 200 pm in from the surface and are nearly uniform in depth. The centre region of the sample appeared the same as the as-received material with dislocation spots, surrounding halos and low C L intensity in the interiors of the cells. The bright regions seen in the cross-section C L images of the R T A samples were examined using photoluminescence at liquid helium temperatures. This investigation found a correlation between the bright regions and the presence of Cu. In addition, the amount of residual Cu on the surface has a significant affect on the depth of the bright bands. Low residual Cu levels results in shallower band depths than high residual Cu levels. It is proposed that the presence of the Cu acts to increase the recombination rate thus increasing the brightness of the C L image. iv T A B L E O F C O N T E N T S ABSTRACT ii TABLE OF CONTENTS iv TABLE OF TABLES ix TABLE OF FIGURES x ACKNOWLEDGMENTS xvii C H A P T E R 1 I N T R O D U C T I O N 1 1.1 DOPING 2 1.2 POST-IMPLANTATION ANNEALING 5 1.3 CATHODOLUMINESCENCE 6 1.3.1 Carrier generation 7 1.3.2 Recombination and Emission 7 1.3.3 Cathodoluminescence Resolution and Contrast 9 1.4 SUMMARY 9 C H A P T E R 2 L I T E R A T U R E R E V I E W 12 2.1 NATURE OF DEFECTS IN SEMICONDUCTORS 12 2.1.1 Dislocations 13 2.1.2 Impurities 13 2.1.3 Native Defects 15 2.2 EFFECT OF DEFECTS ON ELECTRICAL PROPERTIES 17 2.2.1 Dislocations 17 2.2.2 Point Defects 19 2.3 ANNEALING STUDIES 21 2.3.1 Post Ion-Implant Annealing 23 2.4 CATHODOLUMINESCENCE STUDIES 25 2.5 SUMMARY 28 C H A P T E R 3 H E A T T R E A T M E N T 31 3.1 S A M P L E P R E P A R A T I O N 31 3.2 F U R N A C E A N N E A L I N G 32 3.3 R A P I D T H E R M A L A N N E A L I N G 33 3.3.1 Series 1 34 3.3.2 Series 2 34 3.3.3 Series 3 34 3.3.4 Series 4 35 3.3.5 Series 5 35 3.3.6 Series 6 35 3.3.7 Series 7 35 3.3.8 Series 8 36 3.3.9 Series 9 36 3.3.10 Series 10 37 3.3.11 Series 11 37 3.3.12 Series 12 37 3.3.13 Series 13 38 3.3.14 Series 14 38 C H A P T E R 4 S A M P L E C H A R A C T E R I S A T I O N 47 4.1 C A T H O D O L U M I N E S C E N C E 47 4.1.1 C L Apparatus 47 4.1.2 C L Procedure 48 4.1.3 B e l l Northern C L Analys is 49 4.2 P H O T O L U M I N E S C E N C E 50 4.2.1 B e l l Northern P L Analys is 50 4.2.2 S imon Fraser P L Analys is 51 v i 4.3 O P T I C A L T R A N S I E N T C U R R E N T S P E C T R O S C O P Y 52 4.3.1 Basis of Measurement 52 4.3.2 O T C S System Description 53 4.3.3 O T C S Exper imental 53 C H A P T E R 5 R E S U L T S 57 5.1 G E N E R A L O B S E R V A T I O N S 57 5.1.1 Surface Effects 57 5.2 F U R N A C E A N N E A L I N G R E S U L T S 59 5.2.1 Surface C L : 59 5.2.2 Cross-section C L : 60 5.2.3 Effect of cooling rate 60 5.3 R A P I D T H E R M A L A N N E A L I N G R E S U L T S 61 5.3.1 R T A Test Series 1 61 5.3.2 R T A Test Series 2 63 5.3.3 R T A Test Series 3 64 5.3.4 R T A Test Series 4 65 5.3.5 R T A Test Series 5 65 5.3.6 R T A Test Series 6 66 5.3.7 R T A Test Series 7 66 5.3.8 R T A Test Series 8 69 5.3.9 R T A Test Series 9 70 5.3.10 R T A Test Series 10 70 5.3.11 R T A Test Series 11 71 5.3.12 R T A Test Series 12 71 5.3.13 R T A Test Series 13 72 5.3.14 R T A Test Series 14 72 v i i 5.4 B E L L N O R T H E R N C L R E S U L T S 72 5.5 P H O T O L U M T N E S C E N C E R E S U L T S 73 5.5.1 B e l l Northern P L Results 73 5.5.2 B N R Scanning P L Results 75 5.5.3 S imon Fraser Univers i ty P L Results 75 5.6 O T C S R E S U L T S 77 C H A P T E R 6 M A T H E M A T I C A L M O D E L 130 6.1 M O D E L A S S U M P T I O N S 130 6.2 G O V E R N I N G E Q U A T I O N S 131 6.3 M O D E L P A R A M E T E R S 134 6.4 V E R I F I C A T I O N O F T H E M O D E L 138 6.5 M O D E L R E S U L T S 138 6.6 S T R E S S C A L C U L A T I O N S 139 6.7 C O N C L U S I O N 141 C H A P T E R 7 DISCUSSION 152 7.1 F U R N A C E A N N E A L I N G 152 7.2 R T A A N N E A L I N G 152 7.2.1 Cross-section Results 153 7.3 T H E S O U R C E O F C O P P E R In G a A s 153 7.4 D I F F U S I O N O F C O P P E R I N G a A s 155 7.5 E F F E C T O F V A C A N C I E S O N B R I G H T B A N D S 159 7.6 O T C S R E S U L T S 160 7.7 M A T H E M A T I C A L M O D E L R E S U L T S 162 7.8 T H E E F F E C T O F C O P P E R O N L U M I N E S C E N C E 162 7.9 C O N C L U S I O N 164 C H A P T E R 8 S U M M A R Y A N D C O N C L U S I O N S 168 C H A P T E R 9 F U T U R E W O R K 170 B I B L I O G R A P H Y 172 A P P E N D I X A L I S T O F A C R O N Y M S 181 A P P E N D I X B P R I N T O U T O F M A T H E M A T I C A L M O D E L P R O -G R A M 182 A P P E N D I X C C A L C U L A T I O N O F T H E R M A L E X P A N S I O N C O E F F I C I E N T , Y O U N G ' S M O D U L U S A N D POISSON'S R A T I O 187 ix TABLE OF TABLES TABLE I Typical Impurity Levels in Undoped LEC GaAs 30 TABLE II Summary of Furnace Anneals 39 TABLE HI Summary of RTA Test Series 40 TABLE IV List of Band Depths for RTA Test Series 2 79 TABLE V List of Band Depths for RTA Test Series 12 79 TABLE VI Calculation of Diffusion Coefficient Using Equation 7.9 .... 165 TABLE OF FIGURES Figure 1.1 Graph showing variation in EPD across LEC wafer 10 Figure 1.2 Electron beam interaction with sample 10 Figure 1.3 Intrinsic and extrinsic transitions of electrons 11 Figure 3.1 Location of primary and secondary flats for (100) wafer 43 Figure 3.2 Sample layout used for wafers from boule A 43 Figure 3.3 Schematic of RTA furnace 44 Figure 3.4 Temperatures during a typical RTA anneal cycle 45 Figure 3.5 Sample layout for RTA series 1 and 2 45 Figure 3.6 Thermal history showing modified heating and cooling rates 46 Figure 4.1 Schematic of pole piece showing mounting of CL detector. 55 Figure 4.2 Schematic of polishing jig used for taper samples 55 Figure 4.3 Configuration of OTCS measurement system 56 Figure 4.4 Schematic of a ring-dot OTCS electrode structure 56 Figure 5.1 CL image showing spots and halos 80 Figure 5.2 CL image showing a dark circular area 80 Figure 5.3 CL image of the as-received material from boule A 80 Figure 5.4 CL image of sample A s32-2 81 Figure 5.5 Higher magnification photo of cell wall 81 Figure 5.6 CL image of sample A s32-5 81 Figure 5.7 CL image of sample A s32-7 81 Figure 5.8 CL image of sample A s32-8 82 Figure 5.9 CL image of sample A s33-8 82 Figure 5.10 Cross-section CL image of sample A s32-4 82 Figure 5.11 Cross-section CL image of sample A s33-2 82 xi Figure 5.12 Cross-section CL image of sample A s33-2 83 Figure 5.13 Cross-section CL image of sample A s32-5 83 Figure 5.14 Cross-section CL image of sample A s32-7 83 Figure 5.15 Cross-section CL image of sample A s32-8 83 Figure 5.16 Surface CL image of sample A s33-6 84 Figure 5.17 Surface CL image of sample A s33-9 84 Figure 5.18 Surface CL image of sample A s33-8 84 Figure 5.19 Surface CL image of sample A s33-5 84 Figure 5.20 CL image of as-received sample from boule B 85 Figure 5.21 CL image of sample B s69-l, with large cell size 85 Figure 5.22 CL image of sample B s69-l, with small cell size 85 Figure 5.23 CL image of sample B s69-3. 86 Figure 5.24 CL image of sample B s69-2 86 Figure 5.25 Lower magnification CL image of sample B s69-2 86 Figure 5.26 Cross-section CL image of as-received sample from boule B 87 Figure 5.27 Cross-section CL image of sample B s69-l 87 Figure 5.28 Cross-section CL image of sample B s69-3 87 Figure 5.29 Typical cross-section CL image of RTA anneal (T>850) 88 Figure 5.30 Cross-section CL image of bare sample B s68-2 88 Figure 5.31 CL image of exposed edge of sample B s68-6 88 Figure 5.32 Cross-section CL image showing bump in bright band 88 Figure 5.33 Surface CL image of sample B s68-5 89 Figure 5.34 Cross-section CL image of sample B s68-5 89 Figure 5.35 Cross-section CL image of sample A s32-12 89 Figure 5.36 Surface CL image of sample C s59-lB 90 Figure 5.37 Cross-section CL image of sample C s59-lB 90 xii Figure 5.38 Cross-section CL image showing slip bands 90 Figure 5.39 Surface CL image of sample A s32-l 91 Figure 5.40 Surface CL image of sample A s32-9 91 Figure 5.41 Cross-section CL image of sample A s32-9 91 Figure 5.42 Cross-section CL image of sample C s59-3A 92 Figure 5.43 Cross-section CL image of sample C s59-3B 92 Figure 5.44 Cross-section CL image of sample C s59-3C 92 Figure 5.45 Cross-section CL image of sample C s59-lB 92 Figure 5.46 Cross-section CL image of sample C s37-l 93 Figure 5.47 Cross-section CL image of sample C s37-2 93 Figure 5.48 Cross-section CL image of sample C s37-3 93 Figure 5.49 Cross-section CL image of sample from supplier A 94 Figure 5.50 Cross-section CL image of sample from supplier B 94 Figure 5.51 Low magnification CL image of sample from supplier C. .. 94 Figure 5.52 Low magnification CL image of sample from supplier D. .. 94 Figure 5.53 Low magnification CL image from supplier E with dark region at centre 95 Figure 5.54 Low magnification CL image of sample from supplier E without dark region at centre 95 Figure 5.55 Low magnification CL image of (111) sample from supplier F 95 Figure 5.56 Low magnification CL image of (100) sample from supplier F 95 Figure 5.57 Higher magnification CL image of irregular feature in Figure 5.56 96 Figure 5.58 Secondary electron image of area shown in Figure 5.57. ... 96 Figure 5.59 CL cross-section image of (100) sample D s34-l 96 Figure 5.60 CL cross-section image of (110) sample D vl0-l 97 Figure 5.61 CL cross-section image of thinned (110) sample D vlO-2. 97 Figure 5.62 CL cross-section image of upper sample in face to face anneal 97 Figure 5.63 CL cross-section image of lower sample in face to face anneal 97 Figure 5.64 CL cross-section image of sample E s9-l 98 Figure 5.65 CL cross-section image of sample E s89-l 98 Figure 5.66 CL cross-section image of sample E s9-2 98 Figure 5.67 CL cross-section image of sample E s89-2 98 Figure 5.68 CL cross-section image of sample C s35-5 99 Figure 5.69 CL cross-section image of sample C s35-6 99 Figure 5.70 CL cross-section image of sample C s35-4. 99 Figure 5.71 CL cross-section image of sample C s32-3 100 Figure 5.72 CL cross-section image of sample C s32-4 100 Figure 5.73 CL cross-section image of sample C s32-5 100 Figure 5.74 CL cross-section image of sample C s32-6 100 Figure 5.75 CL cross-section of sample C s33-2 101 Figure 5.76 BNR CL spectra from bright region of sample B s68-4 102 Figure 5.77 BNR CL spectra from centre region of sample B s68-4 102 Figure 5.78 BNR PL spectra from as-received sample from boule A. ... 103 Figure 5.79 BNR PL spectra from sample A s33-2 104 Figure 5.80 BNR PL spectra from sample A s32-5 104 Figure 5.81 BNR PL spectra from sample A s32-8 105 Figure 5.82 BNR PL spectra from sample A s32-12 105 Figure 5.83 BNR PL spectra from as-received sample from boule B. ... 106 Figure 5.84 BNR PL spectra from sample B s68-4 106 Figure 5.85 BNR PL spectra from sample B s68-3 107 xiv Figure 5.86 BNR PL spectra from sample B s68-5 107 Figure 5.87 BNR PL spectra from sample C s59-3B 108 Figure 5.88 BNR PL spectra from sample C s59-3C (implant face) 109 Figure 5.89 BNR PL spectra from sample C s59-3C faack face) 109 Figure 5.90 BNR colour PL map of sample B s69-2 110 Figure 5.91 BNR colour PL map of sample A s32-10 I l l Figure 5.92 BNR colour PL map of sample A s33-2 112 Figure 5.93 Schematic diagram of tapered sample B s69-6 113 Figure 5.94 SFU PL spectra along taper of sample B s69-6 114 Figure 5.95 SFU PL spectra from surface of sample B s68-l 115 Figure 5.96 SFU PL spectra from centre of sample Bs68-1 115 Figure 5.97 SFU PL spectra from surface of sample A s32-4 116 Figure 5.98 SFU PL spectra from centre of sample A s32-4 116 Figure 5.99 SFU PL spectra from centre of sample A s32-5 117 Figure 5.100 SFU PL spectra from surface of sample D vl0-l 118 Figure 5.101 SFU PL spectra from centre of sample D vl0-l 118 Figure 5.102 SFU PL spectra from surface of sample C s35-4 119 Figure 5.103 SFU PL spectra from centre of sample C s35-4 119 Figure 5.104 SFU PL spectra along taper of sample C s35-6 120 Figure 5.105 SFU PL spectra from surface of sample E s9-l 121 Figure 5.106 SFU PL spectra from centre of sample E s9-l 121 Figure 5.107 SFU PL long wavelength spectra from surface of sample E s9-l 122 Figure 5.108 SFU PL long wavelength spectra from centre of sample E s9-l 123 Figure 5.109 Raw OTCS data for dot 1 124 Figure 5.110 Normalized OTCS data for dot 1 124 XV Figure 5.111 Photocurrent data for dot 1 124 Figure 5.112 Raw OTCS data for dot 2 125 Figure 5.113 Normalized OTCS data for dot 2 125 Figure 5.114 Photocurrent data for dot 2 125 Figure 5.115 Raw OTCS data for dot 3 126 Figure 5.116 Normalized OTCS data for dot 3 126 Figure 5.117 Photocurrent data for dot 3 126 Figure 5.118 Raw OTCS data for dot 4, using expanded temperature range 127 Figure 5.119 Normalized OTCS data for dot 4, using expanded tem-perature range 127 Figure 5.120 Photocurrent data for dot 4, using expanded tempera-ture range 127 Figure 5.121 Raw OTCS data for dot 5, using expanded temperature range 128 Figure 5.122 Normalized OTCS data for dot 5, using expanded tem-perature range 128 Figure 5.123 Photocurrent data for dot 5, using expanded tempera-ture range 128 Figure 5.124 Raw OTCS data for dot 5, using same temperature range as dots 1 to 3 129 Figure 5.125 Normalized OTCS data for dot 5, using same tempera-ture range as dots 1 to 3 129 Figure 6.1 Stepwise approximation of the spectral emissivity of GaAs. 142 Figure 6.2 Calculated emissivity of GaAs 142 Figure 6.3 Nitrogen heat transfer coefficient as a function gap width. 143 Figure 6.4 Calculated versus measured cooling rates 143 xvi Figure 6.5 Measured versus calculated heating rate for Si emissivi-ty=0.7, T=1625 K 144 Figure 6.6 Measured versus calculated heating rate for Si emissivi-ty=0.5, T=1725 K 144 Figure 6.7 Calculated temperature profile for Si emissivity=0.7 at t=1.5s 145 Figure 6.8 Calculated temperature profile for Si emissivity=0.7 at t=3s 145 Figure 6.9 Calculated temperature profile for Si emissivity=0.5 at t=1.5s 146 Figure 6.10 Calculated temperature profile for Si emissivity=0.5 at t=3.0 s 146 Figure 6.11 Calculated temperature profile for GaAs emissivity=0.3 with Si emissivity=0.7 147 Figure 6.12 Calculated temperature profile for GaAs emissivity=0.3 with Si emissivity=0.5 147 Figure 6.13 Calculated temperature profile for GaAs emissivity=0.5 with Si emissivity=0.7 148 Figure 6.14 Calculated temperature profile for GaAs emissivity=0.5 with Si emissivity=0.5 148 Figure 6.15 Coordinate system used for stress calculations 149 Figure 6.16 Parabolic approximation to the temperature in Figure 6.8 150 Figure 6.17 Calculated stresses for temperature profile in Figure 6.16 150 Figure 6.18 Parabolic approximation to the temperature in Figure 6.11 151 Figure 6.19 Calculated stresses for temperature profile in Figure 6.18 151 Figure 7.1 Plot of the square of the bright band depth (x2) versus annealing time for the results shown in Tables IV and V 167 xvii A C K N O W L E D G M E N T S I would like to thank my supervisors, Dr. Fred Weinberg and Dr. Lawrence Young, for their invaluable support and advice throughout the course of this thesis. I also would like to thank the Science Council of British Columbia and the Keyes Foundation for providing financial support for this project. My thanks to Johnson-Matthey for providing most of the gallium arsenide used in this study. I am very grateful to Mary Mager for her assistance in the running of the ETEC SEM. I indebted to Serge Milaire, our electronics technician, and Ed Armstrong, our stores manager, for their time and effort during this project. My special thanks to my friends Pearl Lee, Willie Fajber, Dave Macquistan, Cliff Mui and Chris Parfeniuk for their boundless comments and assistance. Finally, I would like to thank my family for their continued support in all my efforts. To my special friend, Glenn Roemer, words do not adequately encompass my thanks for his enduring support in my personal and professional life. 1 C H A P T E R 1 I N T R O D U C T I O N To a metallurgist, heat treatment is one of the most powerful techniques for modifying the physical properties of metals. In the electronics industry, heat treatment is also used to improve the material properties but it is the electrical rather than the mechanical properties that are the focus of the treatment. In particular, annealing of GaAs is used in device fabrication and to improve the starting material. Gallium arsenide (GaAs) has been studied extensively in recent years because of the potential advantages that it has to offer. GaAs has higher electron mobility than silicon (Si) which is the industry standard for most semiconductor devices. The higher mobility means GaAs can be used to make faster devices. In addition to high speed, GaAs can be made semi-insulating which can make device isolation easier. GaAs also has lower power dissipation, the ability to work at higher temperatures and better radiation resistance than Si. In spite of these advantages GaAs has had limited use due to problems with the inconsistent electrical performance of the GaAs devices. This unreli-ability has been attributed to defects in the GaAs crystal. The most prominent defects in GaAs are dislocations, with average dislocation densities of 104 -105/cm2. Dislocations are more easily formed in GaAs than Si because the critical resolved shear stress of GaAs is about 7x smaller than that for Si. A typical distribution of dislocations across an LEC GaAs wafer is shown in Figure 1.1. At present, crystal growth techniques for GaAs cannot produce dislocation-free material without doping. 2 Initial studies of the nonuniforrnity of electrical properties showed that these nonuniformities followed the dislocation distribution. More recent work has suggested that the point defect distribution affects the electrical properties and it is the effect of the dislocations on the distribution of these defects that accounts for the observed relationship. The interaction between these defects and the dislocations is dominated by the thermal history of the material. Consequently the relationship between the heat treatments and the crystal defects must be understood for the reliable production of GaAs devices. The heat treatments used to improve the uniformity of the electrical properties usually involve annealing the boule for one or more hours after growth. A more frequently encountered form of heat treatment of GaAs is employed after a process known as ion implantation. Ion implantation is used to selectively dope areas of the semiconductor. After the ions have been implanted the material must be annealed to remove the damage caused by the ion beam and to allow the implanted ions to enter lattice sites where they can act as dopants. These anneals are short compared to the anneals used to improve the electrical properties. 1.1 DOPING Dopingis necessary to the operation of a semiconductor device. The purpose of doping is to change the concentration of carriers to favour one type of carrier. In semiconductors there are two kinds of charge carriers: electrons and holes. The hole is the empty state left when an electron leaves the valence band. Unlike most metals, the movement of these empty states can represent charge flow. This ability results from the fact that the valence band is normally fully occupied. 3 If an electron leaves the valence band, the resulting empty state may be filled by a neighbouring electron moving into the empty state. This move will then leave a hole where the electron used to be. When an electric field is applied the hole will appear to move in the opposite direction to the electron flow hence it is considered as a positive charge carrier. In semiconductors, the valence band and the conduction band are separated by an energy gap. This gap is small enough that thermal energy can excite electrons into the conduction band, producing hole-electron pairs. Unlike metals, increasing temperature will increase the conductivity of the semicon-ductor. The electron flow occurs in the conduction band and the hole flow in the valence band. In a semiconductor with no doping the number of holes and electrons must be equal. The use of a dopant which adds electrons to the conduction band will produce a material with more negative than positive carriers. Material so doped is referred to as n-type semiconductor. Conversely, a dopant that removes electrons from the valence band will increase the positive carriers, producing p-type semiconductor. When an n-type region is brought into contact with a p-type region the resulting diffusion leads to the p-n junction. It is the behaviour of this junction that makes semiconductors a useful electrical material. For most devices it is necessary to produce specific areas of n- and p-type doping. One technique for creating these doped regions is diffusion. Dopants can be diffused from the gas phase or from a layer deposited on the semiconductor surface. However, this process is not very satisfactory for the manufacture of small scale devices because it is difficult to control the lateral spread of the dopant in the semiconductor. In addition, the diffusion process usually requires 4 that the semiconductor substrate be heated to temperatures in the range of 700 to 1000 °C to achieve high diffusion rates. For a compound semiconductor such as GaAs, heating at these temperatures can result in the degradation of the semiconductor surface due to the low vapour pressure of arsenic. Another major technique used for the formation of the device layers is the growth of doped epitaxial layers. The quality of the epitaxial layer often exceeds the underlying substrate quality, particularly for GaAs. However, the use of more than one dopant requires more than one deposition. As well, it is difficult to produce selected areas of doping on the substrate. As a result the production of planar devices is not practical with these techniques. Another disadvantage is that the epitaxial techniques with the greatest control, such as molecular beam epitaxy, require longer growth times reducing the throughput of devices. Ion implantation on the other hand, can readily produce areas of selected doping and can create mixed doping areas in the same layer. Ions of the desired impurity are accelerated in vacuo to high energies. The resultant ion beam can be rastered across the semiconductor substrate. The ions usually penetrate the lattice to a depth ranging from 0.1 to 2 microns. A major disadvantage of ion implantation is the lattice damage produced by the ion beam. The extent of damage depends on the energy, dose and atomic weight of the implanted species and the temperature of the substrate. Annealing has been shown to partially recover this damage and produce viable devices. Ongoing research is aimed at improving device yields from this process. Much of this research has been focussed on studying the effect of the annealing on the electrical properties of the final devices. Analysis of these results is difficult since the electrical properties can be affected by other pro-5 cessing steps. For example, problems with the ion implantation process could be incorrectly attributed to inadequacy in the annealing procedure. More effort needs to be directed at understanding the behaviour of the substrate during the annealing process. 1.2 POST-IMPLANTATION ANNEALING The heat treatment of gallium arsenide is more complicated than that for silicon because it is a compound. As a result there are more native defects that can be formed in the substrate. As well, the vapour pressure of arsenic can lead to surface degradation during heat treatment without adequate protection of the surface. Two techniques have been employed for post-implantation annealing. One method uses a standard tube furnace with forming gas or nitrogen for the ambient gas. For gallium arsenide typical annealing practice uses temperatures around 850 °C for 15 -20 minutes. The wafer is usually capped with silicon dioxide or silicon nitride to inhibit degradation of the wafer surface. An alter-native to capping is to use an arsine atmosphere. However this approach requires additional safety precautions due to the toxicity of this gas. Until recently furnace annealing was the standard annealing technique for post-implantation anneals. The alternative annealing procedure is known as rapid thermal annealing (RTA) or rapid thermal processing (RTP). The process usually employs inco-herent light, though laser light or electron beams can be used, to rapidly heat the semiconductor to the annealing temperature. The sample is then held at the annealing temperature for 5 to 10 seconds. Sample cooling is achieved by 6 a combination of radiation and convection. The RTA anneal temperatures are higher than those used for furnace annealing, typically 900 to 1000 °C. However the times are significantly shorter so that there is less time for diffusion of the dopant and outdiffusion of arsenic from the semiconductor surface. Because of the shorter times, capping the wafer is not always considered necessary. 1.3 CATHODOLUMINESCENCE The electrical behaviour of the substrate, after implanting and annealing is related to the final structure of the crystal. It is therefore important to understand the behaviour of the substrate during this annealing process. Most studies of post-implantation annealing have examined the change in the elec-trical properties after annealing or have looked at the behaviour of the implanted species. But what of the behaviour of the substrate itself? Cathodoluminescence (CL) images are influenced by the defect distribution. The changes in the CL image after annealing will therefore reflect the changes in the defect distribu-tion. To relate the CL images to the crystal defect structure some understanding of the CL process is necessary. The name cathodoluminescence (CL) refers to the fact that the lumines-cence from the sample is a result of the recombination of carriers excited by an electron beam. The energy from the electron beam excites electrons from the valence band into the conduction band or into energy levels lying in the gap. Light is emitted when the electron recombines with a hole in the valence band. Luminescence can also be generated by excitation with light instead of an electron beam. This process is termed photoluminescence. 7 1.3.1 Carrier generation When an electron beam strikes a sample the electrons from the beam will travel into the sample. Figure 1.2 shows the emissions produced by the inter-action of the e-beam with a sample. Some electrons are backscattered out of the sample but the remaining electrons move through the sample losing their energy to the lattice. The range of these electrons is determined primarily by the electron beam energy, E b and the sample density. The excitation volume increases with the penetration depth. The carriers generated in this volume, G, are given by, G=Eb(\-i)IEi (1.1) where is the ionization energy, i.e. the energy required to produce an electron-hole pair and y is the fraction of energy lost due to backscattered electrons. The generation rate, g, is given by, g=GIb/e (1.2) where Ib is the beam current and e is the electron charge. 1.3.2 Recombination and Emission The light emitted from the sample is produced when the electron-hole pairs recombine. This recombination process is most likely to emit light if the sample has a direct band gap. A direct transition means that the electron returns to the valence band without needing to change its momentum. Indirect band gap materials require a change in crystal momentum for transition between the conduction and valence band therefore a photon and a phonon would be needed 8 for radiative recombination. This radiative process has a very low probability so that the electron transition usually takes place by means of nonradiative processes. However, even in direct band gap materials nonradiative recombination can take place. An electron may relax to the ground state by means of multiple phonon processes. The energy for electron transition may be absorbed by another electron (the Auger effect) which may gain sufficient energy to leave the semiconductor or may transfer the energy to other electrons. Nonradiative recombination may occur at surface states or defects. Recombination centres can result in nonradiative recombination. These centres have energy levels that lie within the band gap. They can trap both types of carriers increasing the probability of recombination. Some of these centres result in radiative recombination of the carriers. The luminescence from the sample depends on the ratio of radiative and nonradiative recombination centres and their relative capture cross-sections. The light emitted from the sample is separated into two classes, intrinsic and extrinsic emission. Intrinsic emission refers to light resulting from a direct transition from the conduction band into the valence band. Extrinsic emission refers to light emitted as a result of transitions between an impurity level and either band or between impurity levels. Figure 1.3 illustrates possible trans-itions. These transitions result in different wavelengths of light so that the spectra of the emitted light can be analysed to identify impurities. However, unlike X-ray spectra which result from inner shell transitions, the transitions 9 for each element are not unique. They depend on the electronic structure of neighbouring atoms. Use of CL or PL spectra for impurity identification depends on the existence of previously established correlations. 1.3.3 Cathodoluminescence Resolution and Contrast The carriers, once created may diffuse before recombination takes place. Unlike the secondary electron signal the emitted light comes from the excited region which is determined by the beam voltage and is much larger than the incident electron beam. Consequently the carrier diffusion length and the beam voltage determine the resolution of the signal. The contrast in CL images arises from a number of factors. The presence of defects can alter the radiative efficiency thereby reducing or increasing the CL intensity. Thickness variations can be important when observing epitaxial layers. Anything that affects the internal reflection of light at the sample surface can affect the amount of light leaving the specimen. Variations in the wave-length of the light emitted from the specimen may lead to contrast variations depending on the detector sensitivity. The presence of internal electric fields can also affect the CL contrast. 1.4 S U M M A R Y For the production of viable devices the annealing behaviour must be fully established. In particular the effect of the annealing on the substrate itself needs to be investigated in more detail. This study examines the effect of short term annealing, using cathodoluminescence to characterize the changes due to annealing. Implanted samples have been examined but the major emphasis has been on the behaviour of the unimplanted gallium arsenide. 10 Figure 1.1 Graph showing variation i n K O H etch pit density, which is related to the dislocation density, across a wafer Electron Figure 1.2 Electron beam interaction with semiconductor sample. 11 a) b) c) d) Conduction Band Valence Band Figure 1.3 Excited electron relaxing to lower energy state: a) Conduction band to valence band transition, b) Donor level to valence band, c) Conduction band to acceptor level, and d) Donor level to acceptor level transition. Transition a) represents the intrinsic process and b), c) and d) are extrinsic processes. 12 C H A P T E R 2 L I T E R A T U R E R E V I E W Semiconductors normally experience high temperatures during crystal growth. As well, some crystals are annealed after growth to modify the electrical properties. This process usually involves heating the entire boule at high temperature for several hours. The time and temperature used depend on the manufacturer. The heat treatment used to anneal ion implant damage and activate the implant involves relatively short times at slightly lower tempera-tures. Ion implantation is crucial to the formation of devices on GaAs so that almost all devices are subject to these short term anneals during processing. The thermal history of GaAs influences the defect distribution. This dis-tribution will in turn influence the luminescence behaviour of the material. The defects may change the carrier lifetimes affecting the overall CL intensity or they may introduce new luminescence peaks in the CL spectrum. 2.1 N A T U R E O F D E F E C T S I N S E M I C O N D U C T O R S Any imperfection in the crystal lattice is defined as a defect. The type of crystalline defects found in metal systems are also found in semiconductors, such as interstitials, vacancies and dislocations. Defects are introduced into the crystal lattice during growth and device processing steps. Sometimes these defects are intentionally introduced and are vital to the operation of the device. The defects considered in this investigation are those defects found in electronic grade wafers. Macrodefects, such as twins, which preclude the use of the material for device manufacture will not be discussed. 13 2.1.1 Dislocations A dislocation is an extended defect of which there are essentially two types-the edge dislocation and the screw dislocation. In GaAs the dislocations are predominantly 60° dislocations. The 60° refers to the angle between the line of the dislocation and the Burger's vector. High resolution electron microscopy has shown that those dislocations are normally dissociated into a 30° and a 90° partial.1 Early work in the deformation of GaAs2 ,3 established that the slip system is {111}<110> and the Burger's vector is usually a/2<110>. In GaAs, the dislocations are often designated as a or (3 dislocations depending on whether the extra half plane ends on a row of Ga or As atoms, respectively. Dislocations in the crystal produce stresses in the surrounding lattice. When a large number of dislocations are present the lattice will favour a con-figuration that reduces the strain energy. In GaAs the dislocations form into cellular networks, probably by a combination of climb and glide processes, with the cell walls tending to be aligned along {110} planes. The cell diameters are usually a few hundred microns, the diameter decreasing with increasing dis-location density. In addition to the cellular structure, lineage dislocations have been observed.4 Stirland et al.5 further distinguished other dislocation structures referred to as sheets and streamers which were observed both on samples etched with the AB reagent and in IR absorption micrographs. 2.1.2 Impurities Since impurities represent a discontinuity in the crystal lattice they are a form of point defect. Impurities may be located at interstitial or substitutional sites. The amount of strain associated with the impurity depends on its location 14 and size relative to the lattice atoms. Sometimes the impurities will form defect complexes with other point defects. These complexes are difficult to identify because the nature of the interactions must be inferred from indirect observa-tion. Impurities may be intrinsic or extrinsic. Intrinsic or residual impurities are those impurities that are present in the refined material. Extrinsic impu-rities are intentionally introduced to alter the electronic or crystalline beha-viour. Intentional doping can reach relatively high concentrations. Donor impurities are those atoms which contribute electrons to the semiconductor and acceptor atoms capture a valence electron and so produce a hole in the valence band. Typical donor or n-type impurities in GaAs are Si, Se, S, Sn and Te. Acceptor or p-type dopants include Zn, Be, Mg and Cd. Isoelectronic impurities such as In, B, N and Al do not affect the electrical properties but are sometimes introduced to modify the material parameters especially the dislocation density. Typical impurity levels for undoped LEC GaAs are given in Table I, com-piled from references 6, 7, 8, 9,10, and 11. The values were determined from secondary ion mass spectroscopy (SIMS) measurements although spark source mass spectroscopy (SSMS) has been used by some researchers for detection of bulk impurities. For reference, 4x1014 atoms/cm3 is approximately equal to 10 ppba (part per billion atomic). Carbon in substitutional As positions is the main residual acceptor in GaAs. The carbon concentration can affect the compensation of carriers that results in the semi-insulating behaviour of GaAs. 1 2 1 4 1 3 The most common technique for measuring carbon in GaAs is local vibrational mode (LVM) infrared absorption, although PL and CL spectra have been used for comparative mea-15 surements. PL measurements have shown that the 1.49ev peak, associated with carbon, is lower at dislocations.15,4,16,17 CL measurements have also shown that the carbon peak is affected by the presence of dislocations4. An impurity that is usually found in horizontal Bridgman grown GaAs is Cr. Cr is added to produce semi-insulating material by compensating for the Si that is present in the GaAs. The Si comes from the quartz boats used to contain the GaAs during crystal growth. Si contamination was also a problem in liquid encapsulated Czochralski (LEC) crystal growth but the use of pyrolytic boron nitride crucibles has eliminated the need for Cr doping. Because of its fast diffusion rate Cu is another impurity that is of interest in GaAs. Cu can be present in the starting material but processing improve-ments have limited the contributions from this source. At present, most Cu contamination comes from the chemicals used to treat the wafer surface18 and from the quartz in the annealing furnace.19 Larrabee20 looked at the residual Cu 6 4 left on (111) oriented GaAs after washing in nitric acid. His measurements indicated that approximately 3 X 1013 Cu atoms/cm2 were left on the surface from solutions containing only 1.8 ppm Cu. 2.1.3 Native Defects In addition to impurities, point defects include antisite defects, excess atoms and vacancies. Antisite defects are only present in compound semicon-ductors, such as GaAs where an arsenic atom is found in a gallium site or a gallium atom is found in an arsenic site, designated AsG a or GaA s, respectively. Excess atoms are usually present because the open nature of the zincblende lattice permits self-interstitials with lattice strain energies comparable to that 16 produced by a vacancy. Vacancies are lattice spaces left by missing atoms in the lattice and are present in crystals due to entropy. The normal equilibrium number of vacancies is given by: n=Nexp(-Ev/kBT) (2.1) where n is the number of vacancies, N is the number of lattice atoms, E v is the energy required to move an atom from a lattice site to a site on the surface, kB is Boltzmann's constant, T is the temperature. The number of vacancies present is often higher than the equilibrium value if the crystal is formed at high temperatures then cooled faster than the time required for excess vacancies to diffuse out of the crystal. Excess vacancies in the lattice can lead to the formation of dislocation loops and enhance dislocation climb. A major defect in GaAs is a deep level donor known as EL2. This deep level is responsible for the semi-insulating behaviour of LEC GaAs and is therefore of considerable interest to researchers. However, identification of the defect associated with EL2 has not been clearly established although the general consensus is that a complex including the antisite defect AsG a is involved.38,23,24 This hypothesis would explain why EL2 is highest in As-rich melts.25,13 Deep level transient spectroscopy (DLTS), Electron Spin Resonance and optical behaviour (esp. the photoquenching effect) have been used to identify the source of EL2 however different samples have shown different behaviour. These dif-17 ferences have been explained by assuming that EL2 is not a single defect but a family of defects.26,27,28 The number and type of atoms present in the complex with the AsG a determines the response. 2.2 EFFECT OF DEFECTS ON ELECTRICAL PROPERTIES A defect implies an imperfection in the lattice but its presence is not nec-essarily undesirable. Some defects, such as dopants are intentionally introduced to alter certain properties of the semiconductor. To produce the desired characteristics, the defects must be carefully controlled. Therefore the effect of defects on electrical properties should be well established. The electrical properties are influenced by the behaviour of the free carriers in the matrix. The number of available carriers can be increased by the addition of dopants or it can be decreased by defects which trap carriers for long periods of time. The bulk electrical properties such as resistivity and mobility can reflect the behaviour of the defect distribution on the free carriers. 2.2.1 Dislocations Dislocations can be intentionally introduced to getter harmful impurities.29 The technique, known as extrinsic gettering, involves damaging the back surface of the wafer then heat treating the wafer. The post damage heat treatments have been found to be more effective at low temperatures, 450 -550 °C. Improvements of 30-40 % in the transconductance have been reported. The most improvement was noted for high EPD (Etch Pit Density) substrates.30 Early evidence on the effect of dislocations showed no clear evidence of dislocation-limited yield.31 Studies on silicon bipolar transistors32,33 found that 18 the presence of dislocations produced after ion implantation did not affect the device performance unless a dislocation looped down to intersect the emit-ter/base or collector/base junctions. However a number of studies have found a relationship between dislocations and certain electrical properties. Resistivity measurements across wafers have shown that the resistivity has an M-shape which is the inverse of the etch pit density (EPD) distrib-ut ion. 3 4 , 3 5 , 3 6 , 3 7 Some studies have found that the mobility exhibits a W-shape 3 8 , 3 9 , 3 4 across a wafer but others have not shown any correlation. 3 8 , 3 5 Annealing studies have shown that the resistivity and mobility profiles across a wafer are flattened after heat treatment. 3 7 , 4 0 , 4 1 The effect of dislocations on device behaviour has been examined using field effect transistors (FETs) fabricated on GaAs wafers. The mean value and the scatter in the threshold voltage, V t h , are used for wafer comparisons. The variation in the threshold voltage across the wafer has been found to be similar to the variation in the etch pit density (EPD). 4 2 Takebe et a l . 4 3 have shown that V t h shifts to more negative values as the E P D increases. More specifically Miyazawa et a l . 4 4 found that V t h was markedly affected when a dislocation was within 20-30 um of the F E T channel. Not all studies have found a correlation between V t h and individual dis-locations. Winston et a l . 4 5 did not find a correlation between the proximity of an etch pit from the F E T and V t h . Rather they found that the uniformity of the dislocation density affected the scatter of the threshold voltage. 4 6 Samples with uniform high density of dislocations showed the lowest scatter and those with a W-shaped E P D across the wafer showed the highest scatter. Indium-doped 19 low-dislocation GaAs had a scatter that fell between these two cases. Similarly Takebe et a l . 4 3 noted that the scatter in the threshold voltage, aV t h , was low for low or high E P D material. Honda et a l . 3 5 found that the sheet carrier concentration was directly correlated with the E P D . Other researchers have found that the sheet carrier concentration increases when a dislocation is within 75 |im of the Hal l chip. 4 7 , 4 8 , 4 9 Hyuga et a l . 5 0 counted the number of dislocations within this 75 pm radius. When the E P D of these dislocations exceeded 5X10 4 /cm 2 , the sheet carrier concentration no longer increased but remained constant. P L investigation attributed this increase to a decrease in silicon on acceptor sites.5 1 This decrease was related to the suppression of V A g which contributed to an increase in the E L 2 concentration, as observed. Shinohara et a l . 5 2 found that for as-grown molecular beam epitaxy (MBE) layers of GaAs grown on a GaAs substrate the sheet carrier concentration showed no correlation with dislocations. A later paper 5 3 found the same was true for Si-doped M B E layers but not for Si-implanted layers. For Si-implanted layers the sheet carrier concentration increased around dislocations. The effect was attributed to the gettering of impurities during the post-implant anneal. Investigation with P L showed that the intensities of the 1.49 and 0.8 eV peaks did become nonuniform after the anneal but no correlation with dislocations was observed. 2.2.2 Point Defects Impurities are intentionally introduced to change semiconductor proper-ties. However the presence of certain impurities has been shown to have a 20 detrimental effect on electrical properties. For instance, C r doping has been found to reduce the mobility in GaAs. 7 , 6 4 , 5 5 Boron, which is considered to be an isoelectronic dopant, has been found to correlate with the sheet carrier con-centration for an In-doped crystal. 5 6 The resistivity of undoped GaAs has been found to decrease with the boron concentration.5 7 Residual impurities can also affect the electrical behaviour. C u has been shown to decrease the drain current. 1 8 T in et al . 6 0 found a copper related deep level using optical current transient spectroscopy. Chen et a l . 6 1 found a linear relationship between the carbon concentration and V t h . As the carbon concen-tration increased the threshold voltage decreased. Growth striations have been found to cause large variations in the source-drain current in gateless F E T structures. 1 4 2 In one study by Dobrilla et al . . 6 3 the EL2 distribution showed a stronger correlation with the performances of F E T devices than the EPD . Two wafers •were examined. In one wafer the EL2 pattern was similar to the E PD distri-bution. In the other, the E P D was uniform and the EL2 pattern was W-shaped. In the first wafer, the threshold voltage showed the same pattern as the EL2 and E P D patterns. However, for the second wafer, the threshold voltage was most negative where the EL2 concentration was the highest. The E P D pattern did not correlate with the threshold voltage. A later paper by Miyazawa and Wada 6 4 noted that the EL2 concentration should not directly affect F E T per-formance since it is a deep donor and should be electrically non-active in the n-channel layer. The use of Si0 2 caps was assumed responsible for the more negative threshold voltages due to the presence of V G a which would decrease V A s resulting in more Si on donor sites. 21 Obokata et al. found that material grown far-off stoichiometry, either Ga-or As-rich, showed nonuniform resistivity and mobility. Terashima et al. 6 6 found that the resistivity decreased as the As concentration increased but the resistivity of Ga-rich crystals decreased by a factor of 102 after annealing. Sato et al. 6 7 found that crystals grown off stoichiometry exhibited wide variation in resistivity, activation efficiency and uniformity of carrier concentration. In addition, Miyairi et al. 6 8 attributed their observed nonuniformity in the resistivity to non-stoichiometric effects not to a non-uniform impurity distri-bution. This dependence on the stoichiometry is attributed to the change in the native point defect distribution. 2.3 A N N E A L I N G S T U D I E S Many early annealing studies were aimed at the removal of dislocations. In metal systems annealing usually provides the dislocations with enough energy to move and therefore annihilate or leave the matrix. In GaAs the cellular dislocation networks that form are very difficult to remove once formed. Annealing whole ingots or wafers at temperatures from 650° to 950°C does not significantly reduce the dislocation density although Leigh et al.4 reported that the cellular network was finer and more uniform after ingot annealing. Other studies have not reported any change in the dislocation pattern but rather have shown an improvement in the electrical properties. Long term annealing of either the boule or wafer can improve the uniformity of the resistivity.69,42,40,70,41 Long term annealing has also been shown to improve the uniformity of V t h . 7 1 , 7 2 22 The cooling rate after annealing has been shown to change the resistivity of LEC GaAs. Slowly cooled samples show low resistivity, p, as compared to quenched samples which show high resistivity. Look et al. 7 3 found that this process was reversible. As grown material with p=2.5 Q cm could be quenched after holding at 950 °C to produce p= 9.4 X l O 6 Qcm then changed to 7.8 Qcm after reheating followed by slow cooling. The behaviour of impurities during heat treatment is important to the final device properties. The phenomena of thermal conversion is attributed to dif-fusion effects. In Cr-doped GaAs, the outdiffusion of Cr can lead to the devel-opment of a conductive layer near the surface.74,75 The phenomenon of thermal conversion of SI material to p-type behaviour after annealing is attributed to outdiffusion of EL2. This conclusion was based on the results of a number of studies which found that the stoichiometry of the crystal growth melt affected the stability of the sample. Ga-rich crystals, which have low EL2 distributions, exhibit this conversion whereas As-rich or near stoichiometric crystals do not.76,1 ,77 Annealing flattens out the W-shape of the EL2 profile across a wafer. Etching reveals no noticeable change in the dislocation distribution supporting the theory that EL2 is not governed by the dislocations.41 Long term annealing results in the EL2 concentration becoming more uniform and decreasing to an almost constant value near 1.6X1016/cm3. This decrease removes the difference between As-rich and near-stoichiometric EL2 concentrations.78 Another study found that annealing removes the difference between undoped LEC, Indium-doped LEC and horizontal Bridgman wafers.79 23 Other researchers consider that the carbon concentration is more impor-tant to the thermal stability than the stoichiometry.8014 Since carbon is the main residual acceptor in GaAs, the carbon concentration and the ionized EL2 concentration have been found to be linearly related.12,13,14 Obokata et al. 8 0 found that when the carbon concentration exceeded a certain value, the material showed thermal conversion during annealing. Efforts to reduce the carbon concentration in GaAs have included coating the graphite furniture with alu-minum nitride or pyrolytic boron nitride.81 The importance of the carbon concentration has led to a number of studies examining the change in the carbon concentration with annealing. For these studies the intensity of the 1.49 eV peak in the PL spectra has been observed. This peak is generally accepted to be associated with carbon although one paper has refuted this claim.17 The intensity of the peak along the boule was found to become more uniform after annealing, both axially and radially.82 ,83 ,84 2.3.1 Post Ion-Implant Annealing During ion implantation the impact of the high energy ion beam causes damage to the crystal lattice. This damage must be removed by annealing at elevated temperatures. In addition the heat treatment provides energy for the incorporation of dopants atoms onto lattice sites where they will be electrically active. The extent of activation is determined by the annealing conditions and depends on the efficiency of the impurity atom rearrangement and the removal of implant damage. Most of these annealing studies are aimed at optimizing the implant activation. 24 The type of cap used to protect the surface has been shown to affect the activation efficiency. For example, Si implants have shown better activation using Si0 2 caps than Si 3 N 4 but Se implants have shown better activation with Si 3 N 4 . 5 7 The use of Si0 2 caps has been shown to allow Ga to outdiffuse and this appears to aid the activation of Si. 8 5 However it does not aid Se which is also an n-type implant. Because of the high electron mobility of GaAs, n-type implants are desirable for device applications. Problems have occurred with thermal conversion due to Cr redistribution during annealing.86,87,47 The diffusion of Cr has been shown to be quite complex.88 The diffusion of the p-type implants, Zn and Cd has been shown to be complex as well. The diffusion is inhibited by the presence of Ga vacancies which are thought to trap the fast moving interstitial component of the dopants resulting in lower diffusion rates.89 Diffusion problems with implants can be minimised by using rapid thermal annealing (RTA). Arai et al.9 0 showed that the same results were obtained for samples with or without caps using RTA annealing as opposed to furnace annealed samples using Cr-doped GaAs. The activation of RTA samples has been shown to be comparable to furnace annealed samples.91 RTA samples have even shown better transconductance than furnace annealed samples.92 Hiramoto et al. 9 3 looked at the sheet carrier concentration of RTA samples annealed at various temperatures. They found a peak in the sheet carrier concentration at 900 °C when the hold time was 0 s. The peak shifted down to 850 °C for an anneal time of 20 s. The decrease above the peak temperature was attributed to an increase in the concentration of the GaA s antisite defect. Similar results were reported by Kuzuhara et al. 9 4 where the peak temperature 25 was found to be 1000 °C for a 2 s anneal. They also tried varying the anneal time keeping the temperature fixed and found the best time was 4 s for a 950 °C anneal. 2.4 C A T H O D O L U M I N E S C E N C E S T U D I E S The relationship between point defects and dislocations has been examined by a number of researchers using cathodoluminescence (CL). Dislocations, which act as highly nonradiative centres, can be seen as black spots in the CL image. The luminescence from the remaining material surrounding the dislo-cation is influenced by the point defect distribution.95 The cellular pattern of the dislocations in LEC GaAs can be seen in CL images. In as-grown LEC material, the dislocations are usually surrounded by bright zones, from 40 to 100 microns in diameter. These bright zones, often referred to as halos, are thought to represent regions that have been denuded of nonradiative defects because the defects are gettered to the dislocation core. These bright regions overlap along the cell walls making the cell walls appear bright when viewed at low magnification. The luminescence from the interior of the cells is lower in intensity appearing dark against these bright zones but not as dark as the dislocation spots. After annealing at 950 °C for 14 hours, Dussac et al. 9 6 found that the CL contrast decreased but the variations of the .83 and .82 p.m peaks across the dislocation cells were still present. Chin et al. 9 5 found that heat treating LEC GaAs could result in a reversal of the contrast because the brightness of the interior was higher than the region surrounding the cell walls. Similarly PL contrast reversal was reported by Lohnert.139 Chin et al. postulated that a 26 relatively fast diffusing defect involving an As vacancy was responsible for the increased brightness. From their experiments they estimated a diffusion coefficient was 0.7 xl0'5 cm2/s at 750 °C. Kimura, Hunter and Olsen97 used CL to observe diffusion in In-alloyed GaAs. The samples they used were 20 X 30 mm, 3 mm thick, polished on both sides. After annealing in an evacuated quartz ampoule, the samples were angle-lapped and CL micrographs taken. Diffusion fronts were clearly visible in a sample annealed at 830 °C for 20 minutes. A bright zone was observed at the surface and edges of the specimen. No diffusion front could be seen in the sample annealed for 80 minutes and the sample was very bright far away from the surface. A recent study by Sekiguchi and Sumino98 examined the effect of the cooling rate on the CL image. Samples that were rapidly quenched after annealing for 24 hours did not have halos around the dislocation spots. Subsequent annealing at temperatures between 700 - 750 °C caused the bright zones to develop around the dislocations. Annealing temperatures above 800 or below 700 °C did not produce the bright zones. No electrical measurements were reported in this study. Ishii et al. 9 9 found that the presence or absence of well-defined cell networks affected the scatter of the threshold voltage, V t h . High EPD GaAs with no cell networks had low crVth. The highest aV t h was reported when cell networks were present. Examination by CL showed sharp contrast in areas where the cell networks were more pronounced and was more diffuse where the networks were weaker. 27 Miyazawa et al.7 1 reported results of long time furnace anneals at 800 °C. Wafers with a 1500 angstroms Si 3 N 4 cap were annealed for 8-24 hrs. Again there was little effect on the dislocation density regardless of time. Observations of CL intensity across the wafer showed a reduction of sharp peaks after annealing. These peaks were associated with the denuded zone around the dislocations. With increasing time not only did the peaks become less noticeable but the average CL intensity increased. Miyazawa et al.7 1 also noted that annealing changed the critical distance for a dislocation. The critical distance was defined as the distance between a FET channel and the nearest dislocation pit beyond which the threshold voltage no longer showed a marked decrease. A previous paper100 had established a correlation between FET performance and nonuniformities in the CL pattern. The affect of the dislocation was attributed to the denuded zone around the dislocation. The critical distance was observed to be about 20-30 urn which was roughly the same as the denuded zone observed in the CL pattern. For an annealed wafer the critical distance changed to around 40 um and the effect of the dislocation was not as pronounced. Miyazawa and Wada64 attributed the results of Winston et al., which did not show any proximity effect, to the fact that in high EPD material the denuded zones would overlap therefore obscuring the proximity effect. As well, Miyazawa felt that the Si0 2 caps used by Winston et al. would result in more negative V t h overall, again obscuring the proximity effect. The effect of dislocations in epitaxial layers is very different than in LEC GaAs. Although the dislocation pattern from the substrate is replicated in an epitaxial layer grown by MOCVD, these layers have shown good activation and 28 values for V t h that are higher than for LEC material. Examination with CL showed a different pattern than for LEC material.102 The MOCVD material did not show any bright zones around the dislocations, either before or after annealing. Cathodoluminescence has also been used to examine ion-implanted material. The beam voltage is varied to change the penetration depth of the beam. When the luminescence is plotted versus the beam voltage the curves resemble the implant profiles. Using a mathematical model, Cone and Hen-gehold103 attempted to use this technique for a quantitative assessment of Mg-implanted GaAs. 2.5 SUMMARY In summary the preceding sections have shown the following, 1) Point defects play a crucial role in the electrical behaviour of GaAs. 2) Dislocations are important because of their influence on the distri-bution of the point defects which produces nonuniform electrical prop-erties. 3) Heat treatment modifies the electrical properties probably due to the redistribution of the point defects. 4) The cathodoluminescence image of GaAs is markedly affected by heat treatment. 5) The redistribution of impurities during post-implantation annealing can lead to thermal conversion or other undesirable effects. 29 6) The use of rapid thermal annealing minimizes the diffusion that occurs during post-implantation annealing. 7 ) No study has reported the effect of rapid thermal annealing on the C L behaviour of GaAs. TABLE I Typical Impurity Levels in Undoped LEC GaAs B 1013-1017 atoms/cm3 Acceptor Be 1013-10u " Impurities Cr 1014 Fe 1 0 u 1 0 i 6 » Mg 1 0 i 3 _ 1 0 i 5 Mn 1014-1015 " C 1 0 i 4 . 1 0 i 6 .. Zn 2X1015 * Cu 4X1014 * Donor S 1014-1016 " Impurities Se 10n-1015 " Si 1014-1016 " Te 1012-1014 " 0 1 0 i 5 . 1 0 i 6 .. * Only one reference found for these elements, ref 8 31 CHAPTER 3 HEAT TREATMENT The heat treatment of gallium arsenide serves one of two functions. The first form of heat treatment likely to be encountered is the boule annealing aimed at improving the uniformity of electrical properties. Many GaAs suppliers now include this anneal as a standard treatment. The second form of heat treatment is the short term post-implantation anneal. In this study it is the second type of anneal that has been of interest because it is encountered in almost all processing for GaAs devices. 3.1 S A M P L E P R E P A R A T I O N Test samples were either scribed or cleaved from 3" L E C GaAs wafers, 0.06 cm thick, polished to a mirror finish on both faces. The (100) orientation of the wafer is indicated by two flats that are ground into the as-grown boule before cutting. For (100) wafers the primary flat and the secondary flat are at right angles to each other, as shown in Figure 3.1. The flats are parallel to (110) planes. Scribed samples were prepared using a Tempress scribing machine in Electrical Engineering, U . B . C . Because of the limited travel of the diamond head, it was necessary to halve a 3" wafer before it could be scribed with this unit. To do this, the wafer was scribed with a diamond pencil along the diameter. A glass slide was placed under one half of the wafer with one edge close to the scribed line. Another slide was placed on top of the wafer, above the other slide. By pressing at edge of the wafer that was furthest from the scribe line the wafer would break along the scribed line. A similar procedure was used after scribing the samples by machine. When scribing the samples, either by hand or machine, 32 the scribe line was always parallel to the trace of a (110) plane on the sample surface. The wafers break easily along these lines since the (110) planes are cleavage planes in GaAs. Scribing the samples produced damage along the scribed edge of the sample. To avoid this damage some samples were cleaved from the wafer without scribing. Cleaved samples were prepared by first making a small nick at the edge of the wafer. The wafer was then turned over and placed on a soft surface such as a rubber sheet. A slight pressure was applied to the surface above the nick. A cleavage line would propagate from the nick to the other edge. This procedure was repeated until a sample of the desired size was obtained. Some samples were coated with Si 3 N 4 using plasma enhanced chemical vapour deposition (PECVD) to minimize loss of arsenic from the sample. To prepare for Si 3 N 4 capping wafers were etched in buffered HF, rinsed in deionized water then etched in a 10 percent solution of ammonium hydroxide. After loading into the chamber the wafer was exposed to an ammonia plasma for 1 min. prior to deposition. The Si 3 N 4 was deposited at 300 °C for 6 min at approx. 110 angstroms/min, using a mixture of SiH4/NH3/He. 3.2 F U R N A C E A N N E A L I N G Two wafers from boule A were used. Figure 3.2 shows the layout used for cutting the samples from the wafers. Pieces #1, #3, #10 and #12 were not used for furnace annealing. The average sample size was 24 mm X18 mm (430 mm2). All furnace annealed samples were coated with Si 3N 4 . In addition to the nitride a 2" silicon wafer was placed on the exposed surface of the samples during annealing to suppress loss of arsenic from the samples. 33 To measure thermal history of the sample, a thermocouple (TC) was wired to the quartz boat containing the sample. The performance of a bare TC was compared to a TC that was cemented between two pieces of GaAs (approx. 3X3 mm square). The bare TC read higher than the embedded TC but was subject to wide swings in temperature due to intermittent contact with the quartz boat. Therefore the embedded TC was wired to the boat for the furnace measurements. A temperature transverse along the furnace axis was made to establish the optimum position for samples to be located. Samples were placed on a quartz boat with the proximity wafer then pushed into position using a glass rod. Samples were annealed for 20 min. at various temperatures (600 - 950 °C) in a tube furnace with flowing nitrogen gas. To examine the role of cooling rate some samples were cooled in the furnace, by turning off the furnace while the samples were still in position. The effect of annealing time was not investigated. A summary of the furnace anneal tem-peratures used is given in Table II. 3.3 R A P I D T H E R M A L A N N E A L I N G Samples were annealed in a Heatpulse 210 commercial RTA unit, shown schematically in Figure 3.3. The sample is heated by incoherent light provided by the two banks of halogen lamps. A typical anneal cycle is shown in Figure 3.4. The power to the lamps is controlled by a microprocessor. The temperature is monitored by a thermocouple embedded in the 4" Si wafer on which the sample rests. The Si wafer is supported at the edges by pins attached to a quartz tray. A circular opening in the quartz tray is provided to expose the underside of the Si wafer to the radiation from the lamps. Nitrogen gas is supplied to the chamber at the rate of 2 1/2 1/min. 34 Annealing tests in the present investigation were carried out in the tem-perature range of 650 to 950 °C with annealing times from 5 to 160 s. A total of fourteen test series were carried out to investigate different aspects of the annealing process, as summarised in Table III. Details of these tests are given in the following sections. 3.3.1 Ser ies 1 Material from boule B was used for these anneals since there was insuf-ficient material from boule A, which was used for the furnace anneals. Samples were also slightly larger in size, 20 x 28mm. Samples were scribed from wafer B s69 (indicating boule B, slice 69) as shown in Fig 3.5. The samples were annealed for 5 s at temperatures from 650 - 950 °C. These samples were coated with Si 3 N 4 to suppress As loss and to duplicate the surface treatment of the furnace annealed samples. 3.3.2 Ser ies 2 This series was undertaken to investigate the effect of annealing time. il Samples were scribed from wafer 68 of boule B, using the same sample layout as series 1. The wafer was not etched prior to use. Samples were annealed without nitride at 850 °C for 5,10 and 18 s and at 950 °C for 5 and 10 s. The samples annealed for 5 s were compared with the nitrided samples at 850 and 950 °C to determine what effect, if any was caused by the absence of the nitride. 3.3.3 Ser ies 3 To compare the behaviour of boule A, used for the furnace anneals and boule B, used for the RTA anneals, a sample from boule B was furnace-annealed 35 at 850 °C for 20 min. Samples from borne A were RTA annealed at 950 °C for 5 s. Samples from a third boule with lower dislocation density, boule C were also RTA annealed at 950 °C for 5 s. 3.3.4 Series 4 To examine the importance of the thermal history of the samples, one sample was RTA annealed at 950 °C for 5 s then furnace annealed at 800 °C for 20 min. Another sample was RTA annealed at 800 °C for 5 s followed with an RTA anneal at 950 °C for 5 s. Samples were taken from wafer A s32. 3.3.5 Series 5 To examine the effect of ion implantation on the formation of the bands a quarter wafer was implanted with an n + (Si) implant. Two pieces from this sample were annealed at 850 and 950 °C, respectively, for 5 s. Samples were taken from wafer C s59. No nitride was used for these samples. 3.3.6 Series 6 Three pieces from wafer 37 of boule C were annealed to study the effect of sample size, a one-quarter wafer piece, a 29mm x 14mm piece and a 23mm x 6mm piece. The samples were annealed at 950 °C for 5 s. No nitride was used for these anneals. 3.3.7 Series 7 This series examined the effect of RTA on wafers from different suppliers. Wafers from six suppliers were annealed at 950 °C for 5 s (including a (111) Bridgman wafer). Included in this series was a wafer from supplier A, the 36 supplier who provided boules A, B, C and D. This wafer from boule E had been etched in KOH and was included because all of the other suppliers' samples had been previously etched in KOH to reveal the dislocations. Annealing was preceded by chemical polishing to remove these pits. Chemical polishing con-sisted of etching in a 1:1:1 solution of hydrogen peroxide, ammonium hydroxide and deionized water followed by a polish in a 5 percent bromine-methanol solution. The polishing rate of the bromine-methanol solution depends on the bromine content which is exhausted by the polishing process and evaporation. The samples were not coated with nitride. 3.3.8 Series 8 To investigate the effect of wafer orientation slices were taken from boule D - one cut parallel to the growth direction in the (110) orientation, from the shoulder region (vlO) and one cut perpendicular to the growth direction, in the normal (100) orientation (s34). All pieces were provided in the as-sawn condi-tion. Prior to annealing the samples were chemically polished to remove mechanical damage. The (110) slice was noticeably thicker than the (100) slice even after polishing. To verify that the sample thickness did not affect the band formation another sample from the (110) slice was thinned to be close in thickness to the (100) sample and annealed. All samples were annealed at 950 °C for 5 s. The samples were not coated with nitride. 3.3.9 Series 9 Since the annealing behaviour of GaAs is thought to be related to the loss of As, two quarter wafers were annealed face to face at 950 °C for 5 s. This 37 configuration provides three surfaces for comparison - the top exposed face, the shared GaAs face and the face adjacent to the Si wafer. The samples were not coated with nitride. 3.3.10 Ser ies 10 To investigate annealing behaviour that may be related to variations within a boule, samples from wafers at the seed and tail end of boules E were annealed at 950 °C for 5 s. Chemically polished samples from the same wafers were also annealed to examine the effect of chemical polishing. The samples were not coated with nitride. 3.3.11 Ser ies 11 Samples from boule C were treated to change the amount of residual Cu on the surface. It has been reported that treating GaAs with KCN suppresses the Cu peak observed in PL spectra after annealing GaAs.18 On the basis of this observation one sample was immersed for 15 s in a 1 percent solution of KCN (to which sodium carbonate had been added to suppress the loss of the cyanide) to reduce the residual Cu. Another sample was immersed for 15 s in a copper sulphate solution (3 grams of CuS0 4 dissolved in 60 ml H20) to increase the residual Cu on the surface. These samples, and a control sample from the same wafer, were annealed at 950 °C for 5 s. 3.3.12 Ser ies 12 This series was undertaken to re-examine the results from the second series. The samples were taken from boule C since there was no more available 38 material from boule B. The samples were coated with nitride to allow longer annealing times. All sample were annealed at 850 °C. Samples were annealed for 5, 20, 80 and 160 seconds. 3.3.13 Series 13 This series examined the effect of annealing temperature in a narrow temperature range. Samples from boule C were annealed at 700, 725,750, 775, 800, 825 and 850 °C for 5 s. The samples were not coated with nitride. 3.3.14 Series 14 This series examined the effect of the cooling rate on the RTA samples. A quarter-wafer sample from boule C was annealed using a program that con-trolled the power to the lamps. The resulting anneal cycle is shown in Figure 3.6. This sample was not coated with nitride. T A B L E II Summary of Furnace Anneals S A M P L E T E M P E R A T U R E T I M E C O M M E N T S °C s • A s33-7 600 1200 A s32-2 650 II A s32-4 700 tt A s33-2 750 n A s32-5 800 II A s32-7 850 ti A s32-8 900 tt A s33-8 950 II A s33-6 600 II A s33-5 950 tt Furnace cooled A s33-9 600 II Furnace cooled 40 Table III Suinmary of R T A Test Series S E R I E S S A M P L E T E M P E R A T U R E T I M E C O M M E N T S °C s 1 B s69-l 650 5 B s69-3 800 B s69-2 850 B s69-4 875 B s69-5 900 B s69-6 950 2 B s68-l 850 5 B s68-4 II 10 B s68-2 it 18 B s68-3 950 5 B s68-6 II 10 3 B s68-5 850 1200 Furnace annealed A s32-10 950 5 A s32-12 950 5 C S59-1B 950 5 4 A s32-l 950 5 +800 1200 Furnace annealed A s32-9 800 5 +950 5 41 Table III (Continued) SERIES SAMPLE TEMPERATURE TIME COMMENTS °C s 5 C s59-3B 850 5 With Si Implant C s59-3C 950 5 II C S59-1B 950 5 No Implant 6 C s37-l 950 5 C s37-2 II II C s37-3 it tt 7 Supplier A 950 5 Supplier B II I I Supplier C it tt Supplier D II n Supplier E t i t i Supplier F tt tt 8 D s34-l 950 5 D vlO-1 II II (110) orientation D vlO-2 t i t i tt 9 C s36-l 950 5 Top wafer C s36-2 II it Bottom wafer 42 Table III (Continued) S E R I E S S A M P L E T E M P E R A T U R E T I M E C O M M E N T S °C s 10 E s 9 - 1 950 5 E s89-l i i II E s 9 - 2 II II Chemical ly Polished E s89-2 II II II 11 C s35-5 950 5 C s35-6 II II Treated w i t h K C N C s35-4 tt tt Treated w i t h C u S 0 4 12 C s3 -1 850 5 C s 3 -2 n 20 C s 3 -3 tt 80 C s 3 -4 II 160 13 C s32-l 700 5 C s32-2 725 t i C s32-3 750 it C s32-4 775 tt C s32-5 800 t i C s32-6 825 it C s33-l 850 II 14 C s33-2 950 5 Slow Cool 43 Figure 3.1 Diagram of wafer showing primary and secondary flats for (100) orientation. Figure 3.2 Sample layout used for wafers from Boule A. V 44 Water Cooled Reflective Housing Halogen Lamps Quartz Chamber GaAs Sample 4" Si Wafer Thermocouple Quartz Wafer Tray Figure 3.3 Schematic of R T A furnace. (Note: this diagram is not to scale.) 45 TIME (sees) Figure 3.4 Temperature during a typical RTA anneal cycle. Figure 3.5 Sample layout for RTA series 1 and 2. 46 THERMAL HISTORY OF SAMPLE C S33-2 1 0 0 0 -i 1 0 0 -\ 1 | 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 2 0 4 0 6 0 8 0 1 0 0 1 2 0 1 4 0 1 6 0 1 8 0 T I M E (s) Figure 3.6 Thermal history of sample with modified heating and cooling rates. To slow down the heating and cooling rates the R T A furnace was pro-grammed to adjust the power to the lamps rather than use the temperature control mode of the furnace. As a result the temperature profile is more rounded than a typical anneal cycle as shown in Figure 3.4. The small hump in the cooling curve was caused by a brief period when the power level was held fixed before returning to a constant decrease in the power. 47 C H A P T E R 4 S A M P L E C H A R A C T E R I S A T I O N The predominant characterisation technique used in this study was cathodoluminescence (CL). Photoluminescence (PL) and optical current spec-troscopy (OTCS) were used to supplement the information obtained from the C L analysis. The aim of the study was to observe changes in the luminescence pattern after annealing and to relate these changes to the electrical behaviour of the substrate. 4.1 C A T H O D O L U M I N E S C E N C E 4.1.1 C L A p p a r a t u s A n E T E C S E M was used for the C L imaging. To provide the C L signal, a solid state silicon detector is used in place of the backscatter detector. The detector is mounted by means of brackets located underneath the pole piece. A schematic of the mounting arrangement is shown in Figure 4.1 in which the detector is seen to be mounted coaxial with the electron beam. At magnifications higher than 20X the detector did not interfere with the beam raster. However, at the 20X magnification the edge of the centre hole in the detector could be seen in both the C L and S E images. A working distance of 25mm was used for optimum C L imaging. By using the same mounting as the backscatter detector the C L system produced minimal interference with the collection of the normal secondary electron (SE) signal. The detector was partitioned into four quadrants; the signals from each quadrant could be independently turned on, off or inverted. To prevent backscattered electrons from interfering with the C L signal the 48 detector was provided with a high purity quartz window. The signal from the C L detector was passed through the same preamp and main amplifier used for the backscatter detector. The C L signal is generated by the interaction of the electron beam with the specimen. To obtain sufficient C L signal for imaging it was necessary to maximize the beam current reaching the sample. As a result very large aper-tures were used, 1000 micron diameter for the spray aperture and 400 micron diameter for the final aperture. Focussing the beam to a small diameter results in the loss of total beam current. Therefore a low current (approx. 1.6 amps) was used for the condenser lens. Minor adjustments to the condenser current were used to optimize the C L signal. C L images could only be obtained using 30 K e V for the accelerating voltage. 4.1.2 C L Procedure Al l samples in this study were examined using C L imaging. The samples were prepared by mounting on a standard S E M stub using carbon paint (DAG). Cross-section samples were mounted on double-sided sticky tape to expose one of the cleaved edges then the remaining tape and top surface of the mounting stub were painted with carbon paint to ensure sample grounding. Once the carbon paint had set the samples were placed in the S E M . After pumping down to sufficient vacuum (< 2x10"4 Torr) the beam voltage was turned on. To ensure optimal signal the filament position was adjusted to achieve proper saturation, using the S E signal. After obtaining adequate saturation, the S E image was adjusted to provide a coarse focus of the image. The C L 49 detector was then turned on, normally using all four quadrants of the detector. The ETEC SEM possesses two image screens allowing simultaneous viewing of the SE and CL images. The focus of the CL image differs from that of the SE image because the signal is obtained from a greater depth in the specimen. Initially the SE image was focussed then the focus of the CL image was optimized using a combination of the focus control, CL brightness and contrast controls and the z axis of the stage. Because the focus control changes the beam current reaching the sample, the CL brightness has to be increased or decreased to maintain a visible signal on the CRT screen. The CL contrast determines the sensitivity of the signal to the changes in signal and sometimes had to be changed to obtain a meaningful image. The z axis is adjusted to maintain the sample at the 25 mm working distance. 4.1.3 Be l l N o r t h e r n C L Ana l ys i s Only one sample was examined using the CL system at the Bell Northern Research laboratory in the time available. The Cambridge Instruments SEM at Bell Northern is used only for CL analysis. The light from the sample is collected by means of an elliptical mirror and focussed onto an optical fibre. The light is passed through a diffraction grating spectrometer then detected with either a silicon or cooled germanium detector. The germanium detector provides greater sensitivity to the longer wavelength light up to 1.6 microns. This system is capable of providing either a CL image or CL spectra from the sample. A cross-section sample was examined at room temperature. The silicon detector was used for the initial examination of the sample. The CL image 50 obtained from this system was very noisy so no pictures were taken of the sample. Images of the sample were generated on the screen using only part of the light spectrum to compare with the full spectrum integrated light signal. The silicon detector was then replaced with the germanium detector to take C L spectra from the sample. Spectra were taken from the near surface region and the centre region of the sample. 4.2 P H O T O L U M I N E S C E N C E Photoluminescence (PL) is very similar to cathodoluminescence except that light is used to generate the electrons-hole pairs. As a result the energy used is much lower than CL . Some P L systems use a tunable laser to provide the light allowing a selection of energies but most systems use fixed wavelengths that provide energy in excess of the band gap. A special form of P L called selective pair luminescence (SPL) uses wavelengths smaller than that needed for the gap transition to eliminate the band to band transitions and therefore more clearly show the intergap phenomena. 4.2.1 Bel l Northern P L Analysis Two photoluminescence apparatus were used at Bell Northern Research. The standard apparatus uses a HeNe laser focussed to a 250um spot to excite the sample, using approx. 7 mW. The light emitted is analysed using a dif-fraction grating spectrometer. The scanning P L equipment also uses a HeNe laser, focussing on a 10 pm spot. The laser power at the spot is estimated to be about 2 uW. With the scanning P L system, the sample is moved in very small steps to obtain scans across the sample. The samples are kept at liquid helium temperature for P L systems. 51 The standard P L apparatus was used to examine the wafer surface. Cross-section samples could not be examined as the spot size was too large. The spectra were recorded on a strip chart recorder. Both as-received and annealed material were examined from the furnace and R T A annealed samples. Three cross-section and two surface samples were examined using the scanning P L equipment. 4.2.2 Simon Fraser PL Analysis The P L equipment at Simon Fraser University uses an A r laser to excite the sample. The sample is cooled to liquid helium temperature. A cylindrical lens is used to focus the laser on the sample and to provide a strongly elliptical spot. The spot size is estimated to be 2.5 mm 2 . The light is analysed using Fourier Transform Spectroscopy (FTS). To use the large spot size required an increase in the exposed width of the bright bands which could be achieved by taper polishing at a shallow angle. To taper polish, the nitride cap, i f present, was removed from the annealed sample and the sample mounted on a polishing jig, shown in Figure 4.2, which main-tained a fixed taper during the mechanical polishing. Coarse polishing was done on SiC paper starting with the 240 grit, followed by 400 grit and 600 grit papers. The samples were then polished using an alumina slurry starting at a 25 urn particle size and ending at 5 um size. After mechanical polishing, the samples were removed from the polishing jig and then chemically polished. A chemical etch using a 1:1:1 solution of sodium hydroxide, hydrogen peroxide and de-ionised water was followed by chemical polishing using a 5% solution of bromine in methanol. The sample was swabbed constantly during the 52 bromine-methanol polish to provide a mild mechanical action. 4.3 O P T I C A L T R A N S I E N T C U R R E N T S P E C T R O S C O P Y 4.3.1 Basis of Measurement Optical transient current spectroscopy (OTCS) is based on another char-acterisation technique called deep-level current spectroscopy (DLTS). D L T S measures the decay of the capacitance after the application of a voltage pulse and requires the presence of a depletion region. The pulse reduces the width of the depletion region allowing trap levels to fill. After the pulse the depletion region returns to the equilibrium width and the traps in the depletion region release the carriers to return to normal occupancy. Optical transient current spectroscopy measures the current decay after subjecting a sample to light. This technique is able to look at trap levels in semi-insulating material. The light provides sufficient energy to excite carriers into the trap levels. When the light is turned off these traps emit the carriers while returning to equilibrium. The rate of emission is unique to the trap and is dependent on the temperature. The decay of the current is exponential based on the activation energy of the traps. The current is measured at two times, tx and r^ , after the light is turned off. By scanning through a temperature range the current difference between t x and t 2 will go through a peak at a temperature corresponding to the trap energy. Identification of a trap depends on the cor-relation of this activation energy and a trap level. 53 4.3.2 OTCS System Description The sample is placed on a ceramic stage in a low temperature microprobe station manufactured by MMR. Cooling is achieved by gas expansion. A MMR K20 Temperature Controller is used to control the temperature of the stage by adjusting the power to a strip heater located beneath the stage. Adjustable probes are used to make contact with the electrodes on the sample. The sample is illuminated by a GaAlAs LED (Stanley H2000). The wavelength from this LED is centred at 660 nm. The current supplied to the LED is 18 mA. At this current the measured illumination is approximately 170 u.W/cm2 (according to Figure 4.10 from ref 104 ). A control board is used to pulse the light from the LED. The frequency is determined by a frequency generator. The configuration of the OTCS system is shown in Figure 4.3. The electrode bias is applied to the ring electrode. The current from the dot electrode is input to a preamp then input into a box referred to as the "grey box". The dark current transient from this box is sampled using a boxcar averager (EG&G, model 162) at a predetermined and t2. The boxcar averager changes the measured current difference into a voltage which is input to the computer program. The photo-current and temperature are recorded by the control program during a sample scan. Measurements scans are performed after selecting the desired temperature range, temperature increment and hold time. 4.3.3 OTCS Experimental The Cr ring-dot electrodes were approximately 1000 angstroms thick. The electrode pattern is obtained using a predesigned mask containing various sizes 54 of ring-dot structures. The ring refers to the area left exposed and the dot refers to the centre Cr electrode, as shown in Figure 4.4. The bias voltage is applied to a probe which makes contact with the outlying Cr electrode. The current response is measured by the probe positioned on the dot electrode. The electrode pattern was applied to the taper section of a sample to obtain O T C S spectra in the bright band and centre region of the sample.. Five different ring-dot structures were probed. There was some difficulty with making good contact with the probes, possibly due to the taper of the sample. With the pulse frequency set at 5 Hz the boxcar averager was calibrated to sample the dark current transient at 5 and 25 ms. 5 5 ELECTRON BEAM Pole Piece Detector Housing Mounting Bracket Silicon Detector Quartz Window 777777777777777771 Gallium Arsenide Sample Figure 4.1 Schematic of pole piece showing mounting of C L detector. POLISHING JIG SIDE VIEW 1 t ± B O T T O M VIEW -Handle Retaining Ring i Specimen Stage Figure 4.2 Schematic of polishing jig used for taper samples. The screws are used to adjust the tilt of the specimen stage. The retaining ring maintains the jig in a vertical postion when polishing. Note - this diagram is not to scale. Frequency Generator Power Supply #1 Grey Box Photocurrent Computer Dark Current Transient LED Board Trigger Input Output Boxcar Averager I Dot Ring Applied Electrode Electrode ' Bias LED Low Temperature Station Figure 4.3 Configuration of OTCS measurement system. Figure 4.4 Schematic of a ring-dot OTCS electrode structure. 57 CHAPTER 5 RESULTS 5.1 GENERAL OBSERVATIONS In a typical CL image of the as-received LEC GaAs the dark spots due to the dislocations vary from 3 to 5 microns in diameter. Surrounding these spots are the regions of higher CL intensity referred to as halos, as shown in Figure 5.1. The remaining material has a CL intensity lower than the halos though not necessarily as low as the dislocation spots. As previously mentioned, most of the dislocations form cellular networks in the GaAs. The walls of the cells, when seen in a CL image, appear brighter due to the overlap of the halos. The cell interiors appear dark. The average size of the cells depends on the dislo-cation density which varies across the wafer and from boule to boule. 5.1.1 Surface Effects The CL image is affected by more than just the bulk of the sample. Although most of the CL signal comes primarily from the first few microns of the sample the behaviour of the surface can affect the emitted light. The presence of dust on the surface prevents light from reaching the detector so that dust particles appear as black spots in the CL image. These spots usually appear much darker than spots due to the presence of dislocations. The secondary electron image is used to determine if the spots are due to dust particles. The presence of other contaminants can also affect the CL image but they may not appear in the secondary electron image. The presence of silicon nitride or a thin metallization layer will not be detected in the CL image as long as the 58 film is uniform. Sometimes the CL intensity is increased by the presence of contaminants. In some photos this effect can been seen as irregular bright blotches in the CL image. Contaminants such as carbon can decrease the CL intensity by increasing the amount of internal reflection. Carbon is deposited on the sample surface due to the cracking of the residual diffusion pump oil which decreases the CL intensity across the entire sample. Because the deposition is uniform it is not apparent in the CL images. However the CL intensity decreases with time and itis necessary to increase the brightness to maintain a visible level. Occasionally the deposition rate is high enough that the CL intensity decreases as the CL image is being recorded resulting in a decrease in the intensity from the top to bottom of the CL image. In a number of CL images a circular dark area is observed, as seen in Figure 5.2. The size of this area, which is not present upon initial examination of a sample, appears to be fixed at a diameter of approximately 400 microns which corresponds to the size of the final aperture. The CL intensity from this area decreases with time and the location of the area is permanent once formed. The cause of this area is not known although it could be a region of enhanced carbon deposition or beam damage. Scratches on the sample surface will scatter the light from the sample so that most of the light is not emitted normal to the sample surface. This reduces the light reaching the detector. As a result scratches appear dark in the CL image. The cleavage steps on the cross-section samples also scatter emitted light and, like scratches, appear as dark lines in the CL image. 59 5.2 F U R N A C E A N N E A L I N G R E S U L T S 5.2.1 Sur face C L : The cell interiors appear dark and the cell walls bright in the as-received material from boule A shown in Figure 5.3. Annealing at 650 °C for 20 min causes a change in the appearance of the C L image. The background appears brighter and the cell walls appear dark with a brighter region in the center of this dark area, as shown in Figure 5.4. Initially, it would appear that the material has undergone a contrast reversal. However examination of the cell wall area at higher magnification reveals that the dark spots and halos are still present, as shown in Figure 5.5. It appears that the background intensity has increased everywhere except near the cell walls. Samples annealed at 700, 750 and 800 °C appear very similar to the 650 °C sample. The dark areas are still lying along the cell walls in the sample annealed at 800 °C, as shown in Figure 5.6. The dark areas are not as sharply defined after annealing at 850 °C (Figure 5.7). Annealing at 900 °C produces dark areas that are much more diffuse and discontinuous (Figure 5.8). Increasing the anneal temperature to 950 °C the dark areas become very difficult to define, as shown in Figure 5.9. The contrast setting of this photo was increased by approximately 50 percent over the pre-vious photos, in order to delineate the dark areas. Many dislocations appear to have neither halos nor dark areas around them. However dislocations lying inside a dark area do appear to have a weak halo. Overall, the contrast between all these features is very low. 60 5.2.2 Cross-section CL: The cross-section of the 700 °C sample showed an uneven dark band in the centre of the wafer, as shown in Figure 5.10. A dark region could be seen in the centre of the sample annealed at 750 °C, as shown in Figure 5.12, but other sections did not have any such regions, as shown in Figure 5.11. It is also interesting to note that the dark areas seen next to the cell walls in the surface C L images are not apparent in the cross-section. Occasional dark regions were found in the centre area of the 800 °C sample but most of the cross-section had no dark regions present (Figure 5.13). No dark region was found in the centre of the cross-section of samples annealed at 850 °C, as shown in Figure 5.14. As the temperature increases, the cross-section becomes more non-uniform in C L intensity. Figure 5.15 shows the sample annealed at 900 °C where the C L intensity is very irregular. 5.2.3 Effect of cooling rate To compare the effect of cooling rate, samples were annealed and then removed from the furnace to cool (normal cooling) or the samples were left in the furnace to cool with the furnace turned off (furnace cooling). After annealing at 600 °C the halos and dark areas along the cell walls are slightly more pro-nounced in the furnace cooled sample, as shown by comparing the surface C L image of the normally cooled sample (Figure 5.16) and that of the furnace cooled sample (Figure 5.17), both taken at the same contrast setting. The cross-section C L images appear very similar to one another. 61 The effect of cooling rate is more pronounced in the samples annealed at 950 °C. The halos are barely discernible and the dark areas around the cell walls are very weak in the normally cooled sample, as shown in Figure 5.18. The halos are much more distinct in the furnace cooled sample which is shown in Figure 5.19. 5.3 R A P I D T H E R M A L A N N E A L I N G R E S U L T S 5.3.1 R T A T e s t Ser ies 1 The as-received C L image of boule B is slightly different from that of boule A, the boule used for the furnace anneals. The halos are larger in boule B making the cell walls easier to distinguish, as shown in Figure 5.20. Annealing at 650 °C for 5 sec produces a change in the C L image similar to the furnace annealed samples. The interiors of the cells have brightened with dark areas remaining near the cell walls, as shown in Figure 5.21. However in areas with higher dislocation density than that shown in Figure 5.21 the C L image was almost the same as for the as-received material, as demonstrated in the higher dislocation density area shown in Figure 5.22. The density of dis-locations, seen as spots in the C L image, was 5.7 x lO 4 spots/cm2 in Figure 5.21, and 1.2 x 10 5 spots/cm2 in Figure 5.22. Similar behaviour was seen in the sample annealed at 800 °C. Looking at Figure 5.23, the lower right corner appears similar to the as-received material and the upper left shows areas of brightening. Increasing the anneal temperature to 850 °C produces a much more noticeable change. The dark spots, associated with the dislocations are still 62 present but no halos can be seen and the background appears brighter than the as-received material, as shown in Figure 5.24. Looking at a lower magnification image some dark areas can be seen where there is a high degree of clustering of the dislocations (Figure 5.25, taken at the same magnification as Figures 5.21 and 5.22). These CL patterns were typical of all material RTA annealed at or above 850 °C. To determine if the change in the CL images was uniform throughout the sample thickness, cross-section CL images were taken of (110) cleavage planes. Cleavage lines can be seen on many of the cross-sections, appearing as dark straight or curved lines connecting the top and bottom surface of the sample. Except where indicated all cross-sections photos include both surfaces of the sample. The CL image of the cross-section of the as-received wafer appears the same as the image of the surface apart from the black lines due to the cleavage marks, as shown in Figure 5.26. The dislocations spots and halos are clearly present. The sample that was annealed at 650 °C appears the same as the as-received sample, as shown in Figure 5.27. There is some evidence of brighter areas near the sample surface in the cross-section of the sample annealed at 800 °C but most of the image is similar to the as-received material, as shown in Figure 5.28. The large dark circle at the lower edge of the sample is an artifact of the CL system, as mentioned previously, and is not related to the sample. In all the samples annealed at 850 °C or above, there are regions of higher CL intensity, near the surfaces of the wafer, that are fairly uniform in depth, 63 forming sharply delineated bands. The centre region appears similar to the as-received material. Figure 5.29 shows a typical cross-section C L image of these samples. 5.3.2 R T A Tes t Ser ies 2 A l l samples in this series were annealed without nitride. Chin et a l . 9 5 observed that the brightening that occurred in their furnace annealed samples was suppressed by the presence of a silicon nitride film therefore it was thought that the bright band might be larger if the wafers were annealed without nitride. This was not found to be the case when comparing the samples annealed at 850 and 950 °C for 5 s with and without nitride. The cross-section of a sample annealed without nitride shown in Figure 5.30 has bands of nearly the same depth as the sample annealed at the same temperature with nitride. The depths of the bands in this test series are summarized in Table IV. The cross-section images that have been shown up to this point have all been of cleaved surfaces that were produced after annealing rather than an edge that was exposed during annealing. Examination of the exposed edge of the samples annealed at 850 °C revealed that the bright bands were visible and were of the same depth as the bands seen in the internal cross-sections. However the exposed edge of the sample annealed at 950 °C for 10 s has bright bands that are larger and very irregular, as shown in Figure 5.31. A feature that was occasionally observed in samples annealed with or without nitride was a semi-circular "bump" in the bright band which extends almost halfway into the wafer, as shown in Figure 5.32. Cause of this phe-nomenon is not known. 64 5.3.3 RTA Test Series 3 The purpose of this test series was to compare the behaviour of the boule used for furnace annealing, boule A, and that of the boule used for R T A annealing, boule B. A piece of boule B, B s68-5 was furnace annealed at 850 °C for 20 min. The surface C L image of this sample has the same general pattern as the other furnace annealed samples showing the spots and halos and the dark areas along the cell walls, as shown in Figure 5.33. The cross-section C L image of the same sample does not have any dark regions at the centre of the wafer, as shown in Figure 5.34. The samples from boule A, R T A annealed with nitride at 950 °C for 5 s, appeared the same in the surface C L images as the samples from boule B. The cross-section C L image has bright bands that are more than twice the depth of the bands observed in the samples from boule B that were annealed at the same time and temperature, as shown in Figure 5.35. A sample from a third boule, C, with a lower dislocation density, was also R T A annealed at 950 °C for 5 s. The as-received C L image had fewer spots than either boule A or B because of the lower dislocation density. After annealing, the surface C L image is similar to the other samples, as shown in Figure 5.36. The bands in the cross-section C L image have a depth comparable to those seen in boule B, as shown in Figure 5.37. The C L image of the same sample near the edge of the sample shows pronounced slip bands. Compared to Figure 5.37, the presence of the slip dislocations has not noticeably affected the uniformity or depth of the bands, as shown in Figure 5.38. 65 5.3.4 RTA Test Series 4 The purpose of this test series was to observe the effect of multiple anneals. The surface C L image of sample A s32-l, which was R T A annealed at 950 °C for 5 s then furnace annealed at 800 °C for 20 min appears the same as an 800 °C furnace anneal sample, as shown in Figure 5.39. The cross-section C L image (not shown) appeared the same as the normal 800 °C furnace anneal sample, i.e. with occasional dark regions in the centre but mostly uniform across the section. Sample A s32-9, which was R T A annealed at 800 °C for 5 s then R T A annealed at 950 °C for 5 s appeared similar to a furnace annealed sample in the surface C L image, as shown in Figure 5.40. However, Figure 5.41,the C L cross-section image, shows broad bands, similar to Figure 5.35, which shows another sample from this boule that was R T A annealed at 950 °C for 5 s. The bands in this combination annealed sample showed large variations in depth, not seen in the other R T A anneals from this boule. r 5.3.5 RTA Test Series 5 Ion implantation creates a heavily damaged layer in a half micron layer near the wafer surface. To study the effect of this damage layer on the bright band formation, a quarter wafer was implanted with an n + implant. It was difficult to obtain a surface C L image of the implanted face before annealing due to the implant damage. The cross-section C L image of the implant sample before annealing appears the same as the as-received material, as shown in Figure 5.42. 66 After annealing at 850 °C for 5 sec the bright bands can be seen, as shown i n Figure 5.43. The is no significant difference in the depth of the bands, between the implanted face and the non-implanted face. Similarly there does not appear to be any significant difference between the bands in sample C s59-3C, which was annealed at 950 °C for 5 sec (Figure 5.44). The bands in this sample are comparable in depth to those seen in the unimplanted sample from the same wafer as the previous samples, C s59-lB, which was annealed at 950 °C for 5 s, as shown Figure 5.45. 5.3.6 RTA Test Series 6 The sample size does not noticeably affect the depth of the bright bands. This can be seen when comparing the cross-section C L images of the smallest sample that was annealedinthe size test, shown in Figure 5.46 with the mid-size sample, shown in Figure 5.47 and the quarter wafer piece, shown in Figure 5.48. 5.3.7 RTA Test Series 7 To determine i f the bright band phenomenon was present in GaAs wafers produced by other manufacturers, samples from a range of different suppliers were R T A annealed at 950 °C for 5 s. The cross-section C L image of the sample from supplier A appeared bright through the thickness of the specimen, as shown in Figure 5.49. This result was surprising since the samples from boules A, B and C, also from this supplier, exhibited the bright bands. Possibly the bands have met everywhere at the centre. This sample differs from the previous samples from this supplier in that it had been chemically polished to remove etch pits produced by K O H , like all other material from the suppliers in this 67 test series. Further tests were performed later on other material from this boule to determine i f the surface treatment affected the behaviour. These results will be presented later, in accordance with the order of the test series. In the C L cross-section image of an annealed sample from supplier B bright bands can be seen near both surfaces but there is a difference between the average depth of the bands, as shown in Figure 5.50. This difference may be related to the fact that the wafers from this supplier are polished on one side only. Also both bands are more irregular in depth than the samples shown in previous figures from supplier A. Similarly, the sample from supplier C also has irregular depth of bands as well as a difference in the average depth between the top and bottom, as shown in Figure 5.51. The sample from supplier D has a discontinuous dark region in the centre of the wafer, as shown in Figure 5.52. In some areas this dark region was long enough that it appeared as though there were two individual bright bands near each surface. Perhaps the areas without this dark region represents regions where the two bands have met. After anneahng the sample from supplier E also showed irregular dark regions near the centre, as shown in Figure 5.53. However most of the cross-section appeared as shown in Figure 5.54, with no dark regions at the centre. In addition to the dark spots associated with the dislocations this sample had areas that appeared as large blotches. Where these blotches intersected the sample surface there was a sharp decrease in the C L intensity. The cause of these features is not known but may be related to the doping of the original boule. The material is believed to be Bridgman-grown therefore some Cr doping may be expected since silica crucibles are normally used for this mode of growth. 68 The C L cross-section image of the (111) Bridgman-grown sample from supplier F after annealing has brighter regions near the surface of the sample but they are very irregular in depth, as shown in Figure 5.55. There is an area of lower C L intensity at the inner edge of these bands. This behaviour was not observed in the (100) sample from the same supplier that was also annealed. As can be seen in Figure 5.56 the bright bands are present but there is no region of lower C L intensity lying adjacent to the bright region. It is interesting to note that the bands at the left side of the photo appear to be overlapping but become indistinguishable as separate bands at the extreme left. The width of the bands was very nonuniform as can be seen in this photo. A higher magni-fication photo of the centre region of Figure 5.56, showing the irregular shape in the upper bright band contains a large dark feature near the lower left area of the dark shape, as shown in Figure 5.57. The secondary electron image, shown in Figure 5.58 clearly shows that there was an inclusion located at this point. The other smaller dark circle located near the centre of the C L cross-section may also be an inclusion that was not present in the secondary image because it is beneath the surface. It should be noted that all the material used in this series is approximately 6 years old. Current material from the same suppliers may behave differently due to changes in boule growth practices or wafer preparation. 5.3.8 RTA Test Series 8 A n interesting phenomenon was observed in several of the R T A samples. The bright bands were visible in the C L image of the cleaved edge that was exposed during annealing. It was not possible to examine all samples along the exposed edge but in all samples that were examined this was found to be true. 69 Since it was thought that the bright bands were formed by some form of diffusion it was not clear why the exposed edge was not uniformly bright. To determine i f the bright band formation exhibited any orientation dependence, (100) and (110) samples from the same boule were annealed then examined using CL . The (110) orientation was chosen since it corresponds to the orientation of the cleaved faces. After annealing, the bright bands in the (100) sample, D s34-l, are fairly uniform in depth and are approximately 150 microns deep, as shown in Figure 5.59. In the centre region of this image there is a line of dislocations extending from the top surface to the bottom. Where this line intersects the bright band the band depth is slightly larger. By contrast, the bands in the (110) sample, D v l 0 - l , are very shallow, as shown in Figure 5.60. Since this sample was much thicker than the (100) sample another sample, D vlO-2, was thinned to be closer in size then annealed. The bands are still shallow and irregular in depth, as shown in Figure 5.61. It should be noted that both the (100) and (110) were provided in the as-sawn condition then chemically polished to remove approximately 40 microns of material from each surface. 5.3.9 RTA Test Series 9 Since the formation of the bands was thought to be related to the loss of arsenic, this test was performed to provide three different surfaces to examine. Two quarter wafers were annealed, face to face, under the assumption that the shared faces would suppress the normal loss of As in the other sample. Out-diffusion of As from the exposed face of the top wafer would not be suppressed. 70 The lower face that was exposed to the Si wafer would be expected to have As loss somewhere between these two extremes. In the CL cross-section of the upper sample, C s36-l, there is no apparent difference in size between the upper band, which is adjacent to the exposed face and the lower band which is adjacent to the shared face, as shown in Figure 5.62. Similarly for the lower sample, the upper band, adjacent to the shared face, appears the same as the lower band which is adjacent to the Si face, as shown in Figure 5.63. 5.3.10 RTA Test Series 10 There was some question whether the initial stoichiometry of the melt might affect the formation of the bright bands. To examine this effect, samples from the seed and tail region of two boules were annealed. Normally in Czochralski growth, the seed end of a boule sees a different stoichiometry of the melt than the tail region. Comparing the cross-section CL image of the seed sample, E s9-l, and the tail sample, E s89-l, the average depth of the bands is very similar although the depth of the bands is more irregular in the tail sample, as shown in Figures 5.64 and 5.65. It is interesting to note that the bands are clearly evident since the other sample from this boule, sample E s34-l, which was annealed in series #7 was bright across the entire cross-section (see Figure 5.49). The only difference between these wafers was that the previous sample came from a wafer that had been KOH etched then chemically polished to remove the dislocation pits. To examine if the chemical polishing had any effect on the band formation, samples from the same seed and tail wafers were chemically polished then annealed. The effect of this procedure was to produce deeper bands in both the seed and 71 tail samples, as shown in Figures 5.66 and 5.67, respectively. In the tail sample an area can be seen where the two bands have met in the centre, becoming indistinct as individual bands. 5.3.11 RTA Test Series 11 To further investigate the effect of surface treatment, samples were treated to adjust the level of residual C u on the surface of the samples. The control sample for this test series did not receive any surface treatment prior to annealing. The cross-section C L image of this sample, C s35-5, has bright bands that are nearly uniform and average 130 microns in depth, as shown in Figure 5.68. The sample treated with the K C N solution, to reduce the residual C u level, has bright bands that are very irregular and average 25 microns, as shown in Figure 5.69. The sample treated with the C u S 0 4 solution has bands that extend almost to the centre of the wafer, as seen in Figure 5.70. 5.3.12 RTA Test Series 12 The series examined the time dependence of the band depth using longer annealing times than possible than in test series #2. The depth of the bands does not appear to increase very much with time as can be seen from the results shown in Table V . The depths given are the average of the two band depths. 5.3.13 RTA Test Series 13 The samples in this series were annealed over a narrow temperature range at 25 °C intervals starting at 700 °C and ending at 850 °C. The samples annealed at 700, 725 and 750 °C did not show any sign of bright bands, as shown in the cross-section C L image of the 750 °C annealed sample shown in Figure 5.71. 72 The sample annealed at 775 °C had a weak bright band that can be seen inFigure 5.72, as did the sample annealed at 800 "Cthatis showninFigure 5.73. The samples annealed at 825 and 850 °C had distinct bright bands, as shown in the C L image of the sample annealed at 825 °C that is shown in Figure 5.74. 5.3.14 R T A T e s t Series 14 The sample in this test was annealed using a modified R T A cycle to reduce the heating and cooling rates. The cross-section C L image of this sample has bright bands that are slightly deeper and the decrease in C L intensity at the edge of the band is less abrupt than other samples annealed using the standard annealing cycle, as shown in Figure 5.75. 5.4 B E L L N O R T H E R N C L R E S U L T S Time permitted only one sample to be examined using the B N R C L system. This was considered sufficient since the examination was undertaken to com-pare the C L image obtained using a different C L arrangement. The C L system at Bell Northern was used to image the cross-section of the sample but no photos were taken. The bright bands were clearly visible, both in the total integrated light image and in the monochromatic image. The monochromatic image was taken using a narrow wavelength range centred on the main peak at .89 um. C L spectra were taken from a bright region and from the darker centre region. The spectra were digitized from the original plots. The spectra of the bright area only shows the carbon peak, as shown in Figure 5.76. Similarly the carbon peak is the only major peak present in the spectra from the dark region, 73 shown in Figure 5.77. The main difference between these regions is the height of the peak. The small hump on the side of the main peak in the spectra from the dark area was found to disappear with longer integration time. 5.5 P H O T O L U M I N E S C E N C E R E S U L T S 5.5.1 Bell Northern P L Results The C L spectra shown in Figures 5.76 and 5.77 were taken at room tem-perature. To obtain more detailed spectra, P L spectra were taken at liquid helium temperatures using the single spot P L system at BNR. The spectra from the single spot P L apparatus were digitized from the original chart recordings and as a result are not as smooth as the original data. The P L peak at 911 nm (1.36 eV) is associated with the presence of C u . 1 0 5 1 0 8 This peak and the L O phonon replica at 936 nm were seen in all the heat treated samples, as shown i n Figures 5.78 - 5.89. The peak height did not show any correlation with the furnace anneal temperature. The sample from boule A, furnace annealed at 900 °C (Figure 5.81), had a C u peak the same height as the sample from boule A, R T A annealed at 950 °C (Figure 5.82). This peak height is higher than those for the other furnace annealed samples. It is possible that the chart scale was incorrectly recorded leading to an unexpectedly high value. The full chart deflection was recorded as 300 uV for the C u peak range in this sample whereas in the other furnace annealed samples the maximum deflection in this wavelength range was recorded as 100 uV. The 300 uV maximum was used for recording the wavelength range from 800 to 860 nm for all the furnace annealed samples from boule A. 74 The as-received samples from boule A and B both had weaker signals than the heat treated samples and do not show a C u peak, as shown in Figures 5.78 and 5.83 respectively. Only the plasma peak due to the HeNe is visible at 917 nm. The only sample with a weaker signal was the furnace annealed sample from boule B, shown in Figure 5.86. The signal was so weak it was necessary to use larger slit widths in the diffraction grating spectrometer. The C u peak is present but the plasma peak at 917 nm is still visible. In all other samples where the C u peak was observed the peak was large enough to conceal the plasma peak. The R T A annealed sample from Boule A (Figure 5.82) showed much higher peak heights of the exciton and carbon peaks, located near 820 and 830 nm respectively than any of the furnace annealed samples (Figures 5.79 to 5.81). The R T A annealed samples from Boule B (Figures 5.84 to 5.85) had higher exciton and carbon peaks than any sample from boule A. The implanted samples from boule C (Figures 5.87 to 5.89) showed even larger peak heights. The largest signal was seen in the implanted sample C s59-3B which was annealed at 850 °C for 5 s, as shown in Figure 5.87. 5.5.2 BNR Scanning PL Results Images were taken with the scanning P L system at B N R to compare the P L images at liquid helium temperature with the C L images taken at room temperature. The signal from the as-received sample was too weak to take images of the surface or of the cross-section. The scanning P L samples of the surface of the annealed samples (not shown) did not show any variation in the C u peak. The cross-section P L images show a variation in the intensity of the 810 nm peak across the samples, as shown in Figure 5.90 to 5.92. The warm 75 colours indicate the highest PL intensities although the scale is relative and cannot be used to compare samples. Both RTA samples had brighter regions near the surface. The furnace annealed sample has irregular bright regions similar to the bright bands seen in the CL image of the sample annealed at 700 °C, shown in Figure 5.10. No variation in the Cu level was detected across the cross-sections but the sample temperature may have been too high to see the peak. The PL operator believed that the thickness of the sample could have prevented the cold finger from adequately cooling the sample edge. 5.5.3 Simon Fraser University PL Results The PL spectra from BNR were taken only from the surface of the samples. To obtain PL spectra from deeper regions of the samples pieces were taper polished and examined using the PL system at Simon Fraser University. The taper provided a large surface area which was needed for the PL spot. Spectra were taken from the locations shown in Figure 5.93, along the taper of sample Bs69-6 which had been RTA annealed at 950 °C for 1 Os. The resulting spectra show the progressive decrease of the Cu peak into the depth of the sample (Figure 5.94).' The presence of the Cu peak corresponds to the bright band as shown in the drawing in Figure 5.93. The location of the bright band was established by exarnining one half of the taper using CL imaging at UBC prior to the PL examination of the other half. The peaks near .82 and .83 pm also decrease in height with sample depth. PL spectra of sample B s68-l which had been RTA annealed at 850 °C for 5 s (taken only at the surface and centre of the taper) show that the Cu peak is present at the surface but not in the 76 centre region ,as shown in Figures 5.95 and 5.96. The C u peak is present at the surface and centre regions of the furnace annealed samples from boule A, as shown in Figures 5.99 to 5.98. The C L intensity of both samples was bright near the centre region although both samples showed irregular dark regions at the centre of the sample, as shown in Figures 5.10 and 5.13. The P L spectra were probably taken at the edge of a bright region. The C u peak is present at the surface but not in the centre region of the sample D vlO-1 ((110) orientation), as shown in Figures 5.100 and 5.101. The C L cross-section image of this sample has bright bands near the surface and a darker region in the centre, as shown in Figure 5.60. The C u peak was present at the surface and centre regions of sample C s35-4, as shown in Figures 5.102 and 5.103. This sample, which had been treated with a C u S 0 4 solution prior to annealing, had bright bands that extended through most of the sample, as shown in Figure 5.70. Sample C s35-6, which had been treated with a K C N solution prior to annealing, had a very weak C u peak which dropped off rapidly from the surface and was not present in the centre region of the taper, as shown in Figure 5.104. This sample had very shallow bright bands, as shown in Figure 5.69. The sample from boule E also showed no C u in the centre region of the sample, as seen in Figures 5.105 and 5.106. The longer wavelength spectra of this sample taken with the InSb detector are shown in Figures 5.107 and 5.108. The bright region shows a lower peak between 1.6 and 2.2 urn than the centre region. This broad peak does not correspond with a known defect but may be 77 related to a deep level trap. 5.6 O T C S R E S U L T S O T C S spectra were taken to distinguish between the deep levels present in the bright band and those in the centre region. Two sizes of ring-dot structures were used. Ring-dot #1, #2 and #3 had a dot size of280 pm with a ring diameter of480 pm. Ring-dot #4 and #5 had a dot size of 200 pm with a ring diameter of 430 pm. Ring-dot #1 and #5 were located at the top of the taper in the bright band. Ring-dot #3 and #4 were in the centre of the taper in the darker centre region. Ring-dot #2 was at the edge of the taper but it was not possible to determine if this location corresponded to a bright region. The O T C S plots shown are not scaled to the same maximum and minimum so that the peaks can be seen clearly. To account for possible problems with probe contact the raw data was divided by the photocurrent to provide the normalized data since the magnitude of the photocurrent should reflect the effect of the contact resistance. Only the reverse bias (i.e. dot negative) scans are shown since the forward bias did not provide additional information. A l l dots showed a peak at 350 °C in the raw data plots, as shown in Figures 5.109 to 5.125. This peak was even more pronounced in the normalized plots. Dots #1 and #5 show a peak at 270 °C although it is more pronounced in the data from dot #1. A negative peak near 310 °C can be seen in the data for dots #1 and #5 and near 280 °C for dots #2,#3 and #4. The negative peak at 310 °C was present in a scan for dot #5 taken from 250 to 380 °C, shown in Figure 5.124 but not in the expanded temperature range scan. Dots #4 and #5 were scanned 78 over a larger temperature range than the other dots. The normalized data for dot #5 shows a peak at 90 °C which is not present in the normalized data for dot #4. Only one rate window was used in this study therefore the activation energy for these peaks cannot be calculated. 79 T A B L E IV List of Band Depths for R T A Test Series 2 Anneal Temperature Anneal Time Depth of Band CO (seconds) (microns) 850 5 90 ±8 850 10 122+5 850 18 137 ±10 950 5 100+12 950 10 145 ±10 T A B L E V List of Band Depths for R T A Test Series 12 Anneal Temperature Anneal Time Depth of Band CC) (seconds) (microns) 850 5 104 ±15 850 20 130 ±10 850 80 120 ±15 850 160 155 ±10 80 Figure 5.1 CL image showing spots and halos. Figure 5.2 CL image showing occurence of a dark circular area. This area is an artifact of the CL system and is not related to the sample. Figure 5.3 CL image of the as-received material from boule A, before furnace annealing, showing the dislocation spots and the surrounding halos. 81 Figure 5.4 CL image of sample Figure 5.5 Higher magnification of pre-A s32-2 which has been furnace vious photo showing a dark area annealed at 650 °C for 20 min- along cell wall. The dislocation utes. spots and surrounding halos are still present. Figure 5.6 CL image of sample Figure 5.7 CL image of sample A s32-7, A s32-5, which was furnace which was furnace annealed at 850 annealed at 800 °C for 20 min- °C for 20 minutes, utes. 82 Figure 5.8 CL image of sample Figure 5.9 CL image of sample A s33-8 A s32-8 which was furnace which was furnace annealed at 950 annealed at 900 °C for 20 min- °C for 20 minutes. Photo was taken utes. at higher contrast setting than previous photos. Overall contrast is very weak. Figure 5.10 Cross-section CL image Figure 5.11 Cross-section CL image of of sample A s32-4 which was sample A s33-2 which was furnace furnace annealed at 700 °C. annealed at 750 °C, without any dark region at the centre of the sample. 83 Figure 5.12 Cross-section CL image Figure 5.13 Cross-section CL image of of sample A s33-2 which was sample A s32-5 which was furnace furnace annealed at 750 °C, annealed at 800 °C, showing irreg-showing a dark region in the ular dark region at centre of wafer, centre of the sample. Figure 5.14 Cross-section CL image Figure 5.15 Cross-section CL image of of sample A s32-7 which was sample A s32-8 which was furnace furnace annealed at 850 °C. annealed at 900 °C. The small bright spots are due surface con-tamination. Figure 5.16 Surface CL image of Figure 5.17 Surface CL image of sample sample A s33-6 which was fur- A s33-9 which was furnace nace annealed at 600 °C with annealed at 600 °C, cooled inside normal cooling. the furnace. Figure 5.18 Surface CL image of Figure 5.19 Surface CL image of sample sample A s33-8 which was fur- A s33-5 which was furnace annealed nace annealed at 950 °C with at 950 °C, cooled inside the furnace, normal cooling. 85 Figure 5.20 As-received CL image of material used for RTA anneals. Figure 5.21 CL image of sample RTA Figure 5.22 CL image of sample B s69-l B s69-l which was annealed at which was RTA annealed at 650 °C 650 °C for 5 s (dislocation den- for 5 s, in an area with higher dis-sity = 5.7 X10 4 spots/cm2). location density (dislocation density = 1.2 X l O 5 spots/cm2). 86 Figure 5.23 C L image of sample Figure 5.24 C L image of sample B s69-2 B s69-3 which was R T A which was R T A annealed at 850 °C annealed at 800 °C for 5 s. for 5 s. Figure 5.25 Lower magnification C L image of sample B s69-2 which was R T A annealed at 850 °C for 5 s, showing dark areas where there is a high degree of dislocation clustering. 87 Figure 5.26 As-received cross-section CL image of material used for RTA anneals. Dark lines are cleavage marks. Both surfaces of the wafer are shown. Figure 5.27 Cross-section CL image Figure 5.28 Cross-section CL image of of sample B s69-l which was sample B s69-3 which was RTA RTA annealed at 650 °C for 5 s. annealed at 800 °C for 5 s. 88 150 Microns Figure 5.29 Typical cross-section C L image of samples R T A annealed at or above 850 °C for 5 s, showing bands of higher C L intensity near the sample sur-faces. Figure 5.30 Cross-section C L image of bare sample B s68-2 which was R T A annealed at 850 °C for 18 s. Figure 5.31 C L image of exposed cleavage edge of bare sample Bs68-6 which was R T A annealed at 950 °C for 5 s. Figure 5.32 Cross-section C L image of R T A sample showing "bump" in a bright band. 89 Figure 5.33 Surface CL image of Figure 5.34 Cross-section CL image of sample B s68-5 which was fur- sample B s68-5 which was furnace nace annealed at 850 °C for 20 annealed at 850 °C for 20 min. min. Figure 5.35 Cross-section CL image of sample A s32-l 2 which was RTA annealed at 950 °C for 5 s. 90 Figure 5.36 Surface C L image 950 °C for 5 s. Figure 5.37 Cross-section C L image of sample C s59-lB which was RTA annealed at 950 °C for 5 s. e C S59-1B which was RTA annealed at 150 Microns Figure 5.38 Cross-section C L image of sample C s59-lB which was RTA annealed at 950 °C for 5 s, showing slip bands. 91 Figure 5.39 Surface CL image of sample A s32-l which was RTA annealed at 950 °C for 5 s then furnace annealed at 800 °C for 20 minutes. Figure 5.40 Surface CL image of sample A s32-9 which was RTA annealed at 800 °C for 5 s then RTA annealed at 950 °C for 5 s. Figure 5.41 Cross-section CL image of sample A s32-9 which was RTA annealed at 800 °C for 5 s then RTA annealed at 950 °C for 5 s. 92 Figure 5.42 CL image of cross-section Figure 5.43 CL cross-section image of of implanted sample C s59-3A implanted sample C s59-3B which before annealing. was RTA annealed at 850 °C for 5 s. Figure 5.44 Cross-section CL image Figure 5.45 CL image of sample C s59-lB of implanted sample C s59-3C with NO implant, after RTA which was RTA annealed at 950 annealing at 950 °C for 5 s. °C for 5 s. 93 Figure 5.46 CL cross-section image of smallest piece annealed in size test, sample C s37-l. Figure 5.47 CL cross-section of mid- Figure 5.48 CL cross-section of quarter size piece annealed in size test, wafer annealed in size test, sample sample C s37-2. C s37-3. 94 Figure 5.49 Low magnification cross-section CL image of sam-ple from supplier A (boule E). Figure 5.51 Low magnification cross-section CL image of sam-ple from supplier C; Figure 5.50 Low magnification cross-section CL image of sample from supplier B. Figure 5.52 Low magnification cross-section CL image of sample from supplier D, showing dark area in centre. 95 Figure 5.53 Low magnification Figure 5.54 Low magnification cross-cross-section CL image of sam- «, section CL image of sample from pie from supplier E showing supplier E j showing area without a dark area in centre. dark region at centre. Figure 5.55 Low magnification Figure 5.56 Low magnification cross-cross-section CL image of (111) section CL image of (100) sample sample from supplier F. from supplier F. 96 Figure 5.57 Higher magnification Figure 5.58 Secondary electron image of cross-section CL image of area shown in previous CL photo, irregular feature in mi ddle of previous photo. Figure 5.59 CL cross-section of (100) sample from orientation test, sample D s34-l, which was RTA annealed at 950 °C for 5 s. Note the change in band depth near a line of dislocations that extends from the top to bottom surface. 97 Figure 5.60 C L cross-section of (110) sample from orientation test, sample D vlO-1 which was R T A annealed at 950 °C for 5 s. Figure 5.61 C L cross-section of thinned (110) sample from orientation test, sample D v l 0 - 2 which was R T A annealed at 950 °C for 5 s. 375 Microns ^ — Shared Face Figure 5.62 C L cross-section of upper sample i n the face to face annealing test, sample C s36-l which was R T A annealed at 950 °C for 5 s. 375 Microns Figure 5.63 C L cross-section of lower sample i n the face to face annealing test, sample D s36-2 which was R T A annealed at 950 °C for 5 s. 98 Figure 5.64 CL cross-section image Figure 5.65 CL cross-section image of of sample from seed end of sample from tail end of boule, boule, sample E s9-l which was E s89-l which was RTA annealed at RTA annealed at 950 °C for 5 s. 950 °C for 5 s. Figure 5.66 CL cross-section image Figure 5.67 CL cross-section image of of sample from seed end of sample from tail end of boule, boule, E s9-2 which was RTA E s89-2 which was RTA annealed at annealed at 950 °C for 5 s, after 950 °C for 5 s, after chemical chemical polishing. polishing. 99 150 Microns Figure 5.68 CL cross-section image Figure 5.69 CL cross-section image of of control sample, C s35-5 which sample C s35-6 which was treated was RTA annealed at 950 °C for with KCN solution, RTA annealed 5 s. at 950 °C for 5 s. The bright bands are diminished in depth compared to the untreated sample. \ ' / " 150 Microns Figure 5.70 CL cross-section image of sample C s35-4 which was treated with CuS04 solution, RTA annealed at 950 °C for 5 s. The depth of the bright bands is significantly larger than the untreated sample. 100 Figure 5.71 CL cross-section image Figure 5.72 CL cross-section image of of sample, C s32-3, RTA sample C s32-4, RTA annealed at annealed at 750 °C for 5 s. 775 °C for 5 s. Figure 5.73 CL cross-section image Figure 5.74 CL cross-section image of of control sample, C s32-5, RTA sample C s32-6, RTA annealed at annealed at 800 °C for 5 s. 825 °C for 5 s. 1 0 1 Figure 5.75 C L cross-section of sample C s33-2, which has been R T A annealed using a modified heat cycle to reduce heating and cooling rates. 102 B N R C L S P E C T R A - B R I G H T B A N D CO _ l o 0 . 8 1 1 . 2 1 . 4 W A V E L E N G T H (Microns) 1 . 6 Figure 5.76 B N R C L spectra from bright region of cross-section of sample Bs68-4. B N R C L S P E C T R A - C E N T R E R E G I O N 2 - i 1 . 9 -1 . 8 -1 . 7 -0 . 2 -0 . 1 H 1 1 1 1 1 1 1 0 . 8 1 1 . 2 1 . 4 1 . 6 W A V E L E N G T H (Microns) Figure 5.77 B N R C L spectra from centre region of cross-section of sample Bs68-4. 103 BOULE A - AS-RECEIVED 0.1 -i 0.09 -0.08 -0.07 -E 0.06 -800 820 840 860 880 900 920 940 960 980 W A V E L E N G T H (nm) Figure 5.78 B N R P L spectra from surface of as-received sample from boule A (i.e. unannealed sample). The slit width of the diffraction grating spec-trometer was 500 pm for the wavelengths from 800 to 860 nm. Because of the weaker signal a larger slit width of 1500 pm was used for the wavelengths from 860 to 980 nm. The slit widths were the same for all the B N R P L spectra except where noted. 104 BOULE A - FA: 750 DEG C 0.5 - i 1 0.4 -£ 0.3 -800 820 840 860 880 900 920 940 960 980 W A V E L E N G T H (nm) Figure 5.79 B N R P L spectra from surface of sample A s33-2 which was furnace annealed at 750 °C for 20 min. BOULE A - FA: 800 DEG C 0.5 - i 1 0.4 -0.3 -0.2 -800 820 840 860 880 900 920 940 960 980 W A V E L E N G T H (nm) Figure 5.80 B N R P L spectra from surface of sample A s32-5 which was furnace annealed at 800 °C for 20 min. tn 105 BOULE A - FA: 900 DEG C 0 . 5 - i 1 0 . 4 -> E 0 .3 -Q. 8 0 0 8 2 0 8 4 0 8 6 0 8 8 0 9 0 0 9 2 0 9 4 0 9 6 0 9 8 0 WAVELENGTH (nm) Figure 5.81 B N R P L spectra from surface of sample A s32-8 which was furnace annealed at 900 °C for 20 min. CO BOULE A - RTA: 950 DEG C, 5 S 0 . 5 0 . 4 0 . 3 -0 . 2 0.1 -V I ' I — i 1 1 1 1 1 1 r 8 0 0 8 2 0 8 4 0 8 6 0 8 8 0 9 0 0 9 2 0 9 4 0 9 6 0 9 8 0 WAVELENGTH (nm) Figure 5.82 B N R P L spectra from surface of sample A s32-12 which was R T A annealed at 950 °C for 5 s. 106 BOULE B - AS-RECEIVED 5-E O tn 0.1 0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 -i _ n i i u n u - r w w v i i u u u i r W A V E L E N G T H (nm) Figure 5.83 B N R P L spectra from surface of as-received sample from boule B (i.e. unannealed sample). BOULE B - RTA: 850 DEG C, 10 S > E C3 tn 860 880 900 W A V E L E N G T H (nm) 960 980 Figure 5.84 B N R P L spectra from surface of sample B s68-4 which was R T A annealed at 850 °C for 10 s. 107 BOULE B - RTA: 950 DEG C, 5 S > E O CO 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 l 1 1 f i 1 1 — i i n P 1 1 1 1 1 1 800 820 840 860 880 900 920 940 960 980 W A V E L E N G T H (nm) Figure 5.85 B N R P L spectra from surface of sample B s68-3 which was R T A annealed at 950 °C for 5 s. BOULE B - FA: 850 DEG C > E (3 CO 0.1 0.09 0.08 0.07 0.06 0.05 0.01 -0 -• / 1 1 1 1 1 1 I 1 1 1 1 800 820 840 860 880 900 920 W A V E L E N G T H (nm) 940 960 980 Figure 5.86 B N R P L spectra from surface of sample B s68-5 which was furnace annealed at 850 °C for 20 min. Because the signal from this sample was very weak a slit width of 1000 pm was used for the 800 to 860 nm wave-length region and 2000 pm for the 860 to 980 nm wavelength region. 108 BOULE C - RTA: 850 DEG C 5 S IMPLANTED FACE 800 820 840 860 880 900 920 940 960 980 W A V E L E N G T H (nm) Figure 5.87 B N R P L spectra from surface of implanted sample C s59-3B which was R T A annealed at 850 °C for 5 s. The signal from 860 to 980 nm is shown at 10X the actual value to better show the C u peak. 109 > E CD CO BOULE C - RTA: 950 DEG C, 5 S FRONT FACE - IMPLANTED SAMPLE M "- r i 1 r 8 8 0 9 0 0 9 2 0 W A V E L E N G T H (nm) T r 9 4 0 9 6 0 9 8 0 Figure 5.88 B N R P L spectra from top surface of implanted sample C s59-3C which was R T A annealed at 950 °C for 5 s. The signal from 860 to 980 nm is shown at 10X the actual value to better show the C u peak. Figure 5.89 B N R P L spectra from back surface of implanted sample C s59-3C which was R T A annealed at 950 °C for 5 s. The signal from 860 to 980 nm is shown at 10X the actual value to better show the C u peak. Figure 5.91 BNR scanning P L image of RTA annealed sample from boule A . 113 Figure 5.93 Schematic diagram of tapered sample Bs69-6,showing approximate locations of P L spectra. < co PL SPECTRA ALONG TAPER J L B I I I I I I I I I I 1 1 L 0.8 0.82 0.84 0.86 0.88 0.9 0.92 0.94 0.96 0.98 WAVELENGTH (Microns) Q_ LU Q O < LU DC O Figure 5.94 SFU PL spectra from sample B s69-6 showing progressive decrease of Cu peak with increasing distance from the surface. 115 BOULE B - RTA: 850 DEG C, 5 S BRIGHT R E G I O N 1 - i 0.9 -0.8 0.82 0.84 0.86 0.88 0.9 0.92 0.94 0.96 0.98 1 W A V E L E N G T H (Microns) Figure 5.95 S F U P L spectra from surface of sample B s68-l which was R T A annealed at 850 °C for 5 s. BOULE B - RTA 850 DEG C, 5 S C E N T R E R E G I O N 1 0.9 -0.8 -0.7 -< 0.6 -0.8 0.82 0.84 0.86 0.88 0.9 0.92 0.94 0.96 0.98 1 W A V E L E N G T H ( M i c r o n s ) Figure 5.96 S F U P L spectra from centre region of sample B s68-l which was R T A annealed at 850 °C for 5 s. 116 Figure 5.97 S F U P L spectra from surface of sample A s32-4 which was furnace annealed at 700 °C for 1200 s (20 min). Figure 5.98 S F U P L spectra from centre of taper of sample A s32-4 which was furnace annealed at 700 °C for 1200 s (20 min). The C u peak is still present in the centre the taper. 117 < z BOULE A - FA: 800 DEG C C E N T R E REGION -| 1 1 1 r " 1 — " 1 1 — " i 1 1 1 1 1 1 1 1 r 0.8 0.82 0.84 0.86 0.88 0.9 0.92 0.94 0.96 0.98 W A V E L E N G T H (Microns) Figure 5.99 S F U P L spectra from centre of taper of sample A s32-5 which was furnace annealed at 800 °C for 1200 s (20 min). The C u peak is present in the centre of the taper which corresponds to the centre region of a cross-section. 118 <2 (A (110) ORIENTATION BRIGHT REGION Peak Height = .837 WAVELENGTH (Microns) Figure 5.100 S F U P L spectra from surface of sample D vlO-1 which was R T A annealed at 950 °C for 5 s. £2 c/> 0.2 0.19 0.18 0.17 0.16 0.15 0.14 0.13 0.12 0.11 0.1 0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0 (110) ORIENTATION CENTRE REGION WAVELENGTH (Microns) Figure 5.101 S F U P L spectra from centre region of sample D vlO-1 which was R T A annealed at 950 °C for 5 s. 1 119 2 0.4 0.35 TREATED WITH COPPER SULPHATE N E A R S U R F A C E REGION -i 1 1 1 1 1 1 1 1 1 i 1 1 1 1 1 1 1 r 0.8 0.82 0.84 0.86 0.88 0.9 0.92 0.94 0.96 0.98 W A V E L E N G T H (Microns) Figure 5.102 S F U P L spectra from surface of sample C s35-4 which was treated with C u S 0 4 solution before annealing. o tn TREATED WITH COPPER SULPHATE C E N T R E R E G I O N 0.4 0.35 0.3 0.25 0.2 0.15 0.1 H 0.05 Peak height = 2.26 n I I I I I I I I I I I I I I i i I T " 0.8 0.82 0.84 0.86 0.88 0.9 0.92 0.94 0.96 0.98 W A V E L E N G T H (Microns) Figure 5.103 S F U P L spectra from centre region of sample C s35-4 which was treated with C u S 0 4 solution before annealing. Notice C u peak is still present in the centre region. T R E A T E D WITH KCN < < o 0.15 0.14 0.13 0.12 0.11 0.1 0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 0 Peak Height A 2.06 B 1.8 C 0.8 D 0.79 Location Along Taper A B C D -U.H 0.84 0.86 0.88 0.9 0.92 0.94 WAVELENGTH (Microns) 0.96 0.98 Figure 5.104 SFU PL spectra from taper region of sample C s35-6 which was treated with K C N solution prior to annealing. The spectra for regions A, B and C have been shifted vertically by an arbitrary amount. The Cu peak is very weak and drops off rapidly with distance from the surface. 121 o CO BOULE E - RTA: 950 DEG C, 5 S B R I G H T R E G I O N i f' i—T~^I ~n—i—i 'i . 8 0 . 8 2 0 . 8 4 0 . 8 6 0 . 8 8 0 . 9 i 1 1 1 1 — i i 0 . 9 2 0 . 9 4 0 . 9 6 0 . 9 8 W A V E L E N G T H (Microns) Figure 5.105 S F U P L spectra from surface of sample E s9-l which was R T A annealed at 950 °C for 5 s. < z o CO 3 2 . 8 2 . 6 2 . 4 2 . 2 2 1 .8 1.6 1.4 1.2 1 0 . 8 0 .6 0 .4 0 . 2 0 B O U L E E - RTA: 950 DEG C, 5 S C E N T R E R E G I O N J 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 r 8 0 . 8 2 0 . 8 4 0 . 8 6 0 . 8 8 0 . 9 0 . 9 2 0 . 9 4 0 . 9 6 0 . 9 8 W A V E L E N G T H (Microns) Figure 5.106 S F U P L spectra from centre region of sample E s9-l which was R T A annealed at 950 °C for 5 s. ZD < O CO B O U L E E - RTA: 950 D E G C, 5 S BRIGHT REGION 1.2 1.4 1.6 1.8 2 2.2 W A V E L E N G T H (Microns) 2.6 Figure 5.107 Full wavelength PL spectra from surface of sample E s9-l which was RTA annealed at 950 °C for 5 s. This spectra was taken with an InSb detector to show spectra at longer wavelengths. to o CO B O U L E E - RTA: 950 DEG C , 5 S CENTRE REGION 0.28 0.26 0.24 0.22 0.2 -0.18 -0.16 -0.14 -0.12 -0.1 -0.08 0.06 0.04 0.02 0 j|lL-/^ 0.8 1.2 1.4 1.6 1.8 WAVELENGTH (Microns) 2.6 Figure 5.108 Full wavelength PL spectra from centre region surface of sample E s9-l which was RTA annealed at 950 °C for 5 s. This spectra was taken with an InSb detector to show spectra at longer wavelengths. to L O RAW O T C S DATA: DOT 1, REV BIAS R4 280 310 330 TEMPERATURE (K) Figure 5.109 Raw O T C S data for dot 1. NORMALIZED DATA: DOT 1, REV BIAS R4 2SO 310 330 TEMPERATURE <K) Figure 5.110 Normalized O T C S data for dot 1. P HOTOCURRENT: DOT 1, REV BIAS R4 5 C C O C 3 290 310 330 TEMPERATURE (K) Figure 5.111 Photocurrent behaviour for dot 1. RAW O T C S DATA: DOT 2, REV BIAS RI TEMPERATURE(K) Figure 5.112 Raw O T C S data for dot 2. NORMALIZED DATA: DOT 2, REV BIAS RI 2SO 310 330 TEMPERATURE (K) Figure 5.113 Normalized O T C S data for dot 2. PHOTOCURRENT: DOT 2, REV BIAS RI 5 o.i o.; o.< o.i o.. o.; o.: 280 310 330 TEMPERATURE (K) Figure 5.114 Photocurrent behaviour for dot 2. RAW OTCS DATA: DOT 3, REV BIAS TEMPERATURE (K) Figure 5.115 Raw O T C S data for dot 3. NORMALIZED DATA: DOT 3, REV BIAS O.I 6 O.-IS 0.14 O.I 3 0.12 0.11 O.I 0.09 -O.OS -O.OT 0.06 O.OS 0.04 -0.03 0.02 -O.OI O -O.OI -0.02 -0.03 31O 330 =MPERATURE (K) Figure 5.116 Normalized O T C S data for dot 3. PHOTOCURRENT: DOT 3, REV BIAS 290 310 330 TEMPERATURE (K) Figure 5.117 Photocurrent behaviour for dot 3. 127 RAW OTCS DATA: DOT 4, REV BIAS 200 240 2SO TEMPERATURE (K) Figure 5.118 Raw O T C S data for dot 4 using expanded temperature range. NORMALIZED DATA: DOT 4, REV BIAS 0.3S 0.3 0.2S 0.2 O.IG 200 240 280 TEMPERATURE <K) Figure 5.119 Normalized O T C S data for dot 4 using expanded temperature range. PHOTOCURRENT: DOT 4, REV BIAS 2.2 2.1 - y — \ 2 f X . 1.8 - / 1.8 / >. 1.7 / \ 1 . 6 / \ 1.S — / \^ 1.4 - / ED 1 . 3 -cc 1.2 — / >. 3 1.1 / X f\ 1 o 0.8 V. n_ o.a >. O.T \ 0 . 6 x ^ o.s 0.4 0 . 3 0.2 " ' ' . _^  O.I " O 1 1 1 1 1 1 1 1 1 1 1 1 1 1 200 240 280 TEMPERATURE (K) Figure 5.120 Photocurrent behaviour for dot 4 using expanded temperature range. 128 RAW OTCS DATA: DOT S, REV BIAS ISO 200 240 280 TEMPERATURE (K) Figure 5.121 Raw O T C S data for dot 5 using expanded temperature range. NORMALIZED DATA: DOT 5, REV BIAS 0.32 0.3 A 0.2S / \ 0.2B / \ 0.24 / s < 0.22 / 0.2 / ca ca O.IB / UZE O.I 6 / ^ \ / 0.14 / \ / o 0.12 \ / O.I \ / 0.08 - \^_^ / o.oe 0.04 0.02 ^ — O 1 1 1 1 i 1 i 1 1 1 1 1 1 1 200 240 2SO TEMPERATURE <K) Figure 5.122 Normalized O T C S data for dot 5 using expanded temperature range. PHOTOCURRENT: DOT 5, REV BIAS T E M P E R A T U R E (K ) Figure 5.123 Photocurrent behaviour for dot 5 using expanded temperature range. 129 RAW O T C S DATA: DOT 5, REV BIAS RI 2SO 310 330 TEMPERATURE (K) Figure 5.124 Raw O T C S data for dot 5 using same temperature range as dots 1,2 and 3. NORMALIZED DATA: DOT 5, REV BIAS RI 0.2 O.IS -O.I 6 O.I 4 0.12 O.I o.oa o.oe 0.04 0.02 o -0.02 -0.04 280 310 330 TEMPERATURE (K) Figure 5.125 Normalised data for dot 5 using reduced temperature range. 130 C H A P T E R 6 M A T H E M A T I C A L M O D E L A typical annealing cycle using the Heatpulse furnace consists of 4 to 5 s for the sample to heat to temperature, 5 to 10 s at temperature then approxi-mately 60 s to cool the sample. Compared to normal furnace annealing which takes typically 20 min., the entire cycle is very brief. Although the samples are only 600 microns thick the rapid rise in temperature may create thermal gra-dients through the sample thickness. Such gradients, i f present, could lead to thermal stresses in the sample that could influence the annealing behaviour. Since it is impractical to measure the temperatures directly the only other means to investigate this possibility is a mathematical model. 6.1 M O D E L A S S U M P T I O N S The model presented in this chapter is a one-dimensional heat transfer model to calculate the temperatures in the GaAs sample and the Si support wafer during sample heating. Because the model only addresses the problem of gradients through the sample thickness it was felt that only one-dimensional heat flow needed to be considered. Other models that have assumed a uniform temperature through the wafer thickness have shown that significant thermal gradients were only present at the periphery of the wafer . 1 0 9 1 1 0 The tempera-tures away from the edge of the GaAs were essentially uniform. Therefore it does not seem necessary to model temperatures parallel to the wafer surface. Variations in these planes would be due to local irregularities which cannot be accounted for without using an extremely complex 3-D model. 131 Other model assumptions are as follows: 1) Heat transfer at the exposed GaAs and Si surfaces occurs by radiation from the halogen lamps and convection to the ambient gas. 2) The halogen lamps are assumed to be a planar blackbody source at a fixed temperature. 3) The GaAs and Si are assumed to be gray and diffuse. 4) There exists a thermal contact resistance at the GaAs - Si interface. 5) The quartz walls of the chamber are transparent to the radiation from the lamps. 6.2 G O V E R N I N G E Q U A T I O N S The temperatures in the solid were solved for using the one-dimensional, time-dependent heat equation: „ dT (6.1) where k, p and C p are the thermal conductivity, density and heat capacity of the solid. The above equation assumes that these values are constant; however the thermal conductivity is a strong function of temperature over the temper-ature range examined, 400 - 950 °C. To account for this discrepancy the thermal conductivity was calculated for each node at every time step. 132 The equations used at the boundaries are as follows: 1 )The exposed GaAs surface: 4 4 dTG dTG (6.2) A EGc(T4B - T4G) - hcA (TG ~Ta)-kGA-^ = pGCpGVG -jj- \ s u l f a c e where ZQ is the emissivity of the GaAs surface, o is the Stefan-Boltzmann constant, T B is the temperature of the halogen lamps, h c is the convective heat transfer coefficient, and T a is the ambient gas temperature. 2)The GaAs side of the GaAs-Si interface: dTG dTG (6.3) hinterfaceA (TG ~ Ts) — kQA l ^ e i / a c e - PGCpGVG \interfact where hinterface is the interfacial heat transfer coefficient. 3)The Si side of the GaAs-Si interface: dTs_ _ dTs_ (6.4) hinterfaoA (Ts ~ -Ts) — ^ \interface~ Ps^Ps^S \intetface 4) The exposed Si surface 4 4 dTs dTs (6.5) A £sa(T4B ~ T4S) - hcA (Ts ~Ta)~ksA-^ = psCpsVs -± \staface 133 To solve these equations the Crank-Nicholson finite difference method was employed. Using this method equation 6.1 becomes: (r,.+1-2r,.+r1._1) , (T^-IT^T^) (r-r> (6.6) -k k r = pCp—:—-Ax2 Ax2 Ar T' indicates the temperature at time t + At and T indicates the temperature at time t. The Crank-Nicholson method uses both T' and T on the left-hand side of the equation to avoid the stability problems which arise when solving for T' using the explicit finite difference technique. Using the explicit technique the mesh size, Ax and time step At are subject to the stability criterion: Ar pCp (6.7) Ax2< 2k If a fine mesh size is desired then a very small time step must be used. This problem is avoided in the Crank-Nicholson method. Applying the Crank-Nicholson approach to the boundary conditions yields the following equations: 1) GaAs surface EG (Tl-Tt) hc (T'.-T'. + T.-TJ^ kG (T'2-T\ + T2-Tx) T\-Tx (6.8) pcCpo Ax poCpa 2Ax pGCpG 2(Ax)2 Ar 2) GaAs interface W ~ V'i+i-T'i + T^-Td kG (T'j-r";_t + T;-Tj_t) ^T' j -T, (6.9) pGCpG 2Ax pGCpG 2(AX)2 Ar 134 3) Si interface h ^ A T ' i - r ^ + Tj-T;^) ks (T'i+1-T'i+2 + Ti + 1-Ti+7) _T'i + l-Ti+1 (6.10) psCps 2Ax psCps 2(Axf Ar 4) Si surface eo- (Jt-n) he (j'N-ra+TN-Ta) ks (ry 1-r i V+T i V_ 1-r J V)_r i V-r j y (6.11) psCps Ax psCps 2Ax psCps 2(Ax)2 Ar These equations yield a matrix of coefficients that can be used to solve for the values of T' based on the known values of T. The resultant matrix is tridiagonal and can be solved using a simple subroutine found in reference 111. A copy of the computer program used to solve these equations can be found in Appendix B. 6.3 M O D E L P A R A M E T E R S Before the preceding equations can be solved values must be obtained for the following parameters: kG, ks, pG, ps, CpG, Cp s > EQ, es, T B , and T a . The expressions for kG and k s as a function of temperature were taken from reference 112. Values for pG, CpG, p s and Cp s were taken from references 113,114, and 115. Choosing a single value for the emissivity of GaAs is difficult because of the strong spectral dependence of the emissivity. The most abrupt change in the emittance of GaAs occurs near the wavelength corresponding to the band gap energy. At longer wavelengths the emittance is influenced by the free carrier concentration. The free carrier concentration is a function of the temperature and the initial doping of the material. In Jordan's paper116 on the emittance of 135 n-type GaAs the calculated value of the emissivity showed a weak temperature dependence for material with low doping and did not change much for longer wavelengths (<4 um). Since the samples in this study are semi-insulating, the only doping present is the intrinsic level of dopants and therefore is very low. Using this result the spectral dependence of the emittance was approximated as shown in Figure 6.1. Using this stepwise approximation for the emissivity an effective value for the emissivity can be calculated as follows 1 1 7: •W*. = £i • FiO-W + ez • F ( ^ r -X 2 T) + e 3 . F ^ I * — ) (6.12) where F(\T - X^T) is the fraction of blackbody radiation between \T and X2T. To account for some temperature dependence of e, the value of \ was calculated based on the known temperature dependence of the band gap. Figure 6.2 shows the resultant effective emissivity as a function of the temperature of the GaAs and the source temperature. Two values for the emissivity of Si, 0.5 and 0.7, were found in the litera-ture 1 1 2 , 1 1 5 . Based on the information available it was difficult to determine which was the more accurate of the two therefore model calculations were performed for both values and compared to assess the impact of this parameter. The GaAs sample and the Si susceptor cannot be in complete contact. It is likely that there is a small air gap to inhibit heat conduction between the two materials. A heat transfer coefficient for the gap can be calculated by dividing the thermal conductivity of the gas by the gap width. Thermal conductivity of nitrogen gas as a function of temperature was obtained from reference 119. Figure 6.3 shows the variation in this coefficient with the gap width. A small 136 gap was seen between the GaAs sample and the Si wafer but the gap was no more than half the thickness of the GaAs sample at the widest point. Therefore a worst case depth of 300 microns was assumed for the gap width to calculate a heat transfer coefficient associated with the gap. In addition to conduction across the gap there can be radiative exchange between the GaAs and the Si. The formula for radiative interchange between infinite parallel planes is given by118, A ~~ (I/Eg + — 1) h ) where T G and T s are the temperatures of the GaAs and Si interfaces respectively. The radiative heat transfer coefficient, h,. is therefore given by: 1 (TG-T$) (6.14) ' (Veo + l/Es-lfiTo-Ts) Calculating values of h, in the temperature range of interest using equation 6.14 it was found that the value of h, was given approximately by the expression h~2.95 x 10"12T3. Therefore this expression was used in the model for h,.. Combining the heat transfer coefficients due to the air gap and the radiative exchange gave the gap heat transfer coefficient. No value for h c was found in the literature. Typical values for forced convection as given by Kreith120 vary from .002 - .02 W/cm2oC. Wilson112 examined the effect of h,. on the surface temperature of the GaAs in his model using values from .03 to .07 W/m2°C. These values seemed prohibitively high compared to the range suggested by Kreith120. To determine an approximate 137 value for h c , the cooling rate of the Si was calculated using various values of h,.. Using a value of .05 W/m 2 oC the cooling rate is too fast. The value of .0025 W/m 2 o C showed the best fit with the experimental data therefore it was used for both the GaAs and Si surfaces in the model, as shown on Figure 6.4. This graph shows the result of the calculated cooling from 760 to 460 K( approx. 500°C to 200°C) due to convection only compared with the cooling rate measured by the thermocouple in the Si wafer. The value for h,. was chosen using an arbitrary value for the ambient temperature, T a . The furnace design did not permit monitoring the gas tem-perature and insufficient data was available to calculate the temperature. A value of 473 K (200°C) was used for the cooling calculations. The same value was used for the model although it may be expected that the gas temperature would rise during the heating cycle. Using a fixed value for the ambient tem-perature could lead to overestimating the heat loss due to convection which would slow down the heating rate. However since the value of h c is low this inaccuracy is expected to be small. To determine a value for the bulb temperature the measured heating rate was compared to the model calculation of the heating rate. The measured heating rate was taken using the output of the thermocouple that was embedded in the Si wafer. Heating rates were calculated using both values of the emissivity of Si. The bulb temperature that provided the closest match to the measured heating rate was 1625 K for e S i = 0.7 and 1725 K for e S i = 0.5, as shown in Figures 6.5 and 6.6, respectively. 138 6.4 V E R I F I C A T I O N O F T H E M O D E L The dimensions of the GaAs and Si wafers make it impossible to verify the predicted temperatures by direct measurement. Therefore the model results were compared to calculated solutions for simplified boundary conditions. First the model prediction of the steady state surface temperature of the silicon was compared to the calculated temperature for the case of a fixed temperature of900 K at the one surface of the Si with a heat flow of937.5 W/cm 2 applied to the other surface. The model solution was i n exact agreement with the calculated value. The next test was a slight variation of this case with the surface temper-ature being calculated with one surface held at 900 K and convective cooling on the other face with h c = .05 and the ambient temperature = 700 K . The model prediction was i n good agreement with the calculated result. These calculations indicate that there was no logic flaw i n the formulation of the model equations but they do not imply that the model results are exact. The uncertainty i n the real boundary conditions makes it difficult to establish the validity of the model. 6.5 M O D E L R E S U L T S The model calculations of the temperatures through the thickness of the GaAs and Si at t=l .5 s show that the temperature distribution i n the GaAs has a slight U-shape, as shown i n Figure 6.7. The temperature gradient is higher between the centre of the GaAs wafer and the interface than between the centre and the surface of the GaAs. The shape of the profile is the same at t= 3.0 s but 139 the average temperature is higher, as shown in Figure 6.8. Using the lower value of 0.5 for the Si emissivity results in a slight increase in the curvature in the GaAs solid, as shown in Figures 6.9 and 6.10. Because of the uncertainty in the validity of the calculated value of the emissivity of the GaAs, some model calculations were performed for arbitrary values of E Q ^ . Using ^^=0.3, the results show that the U-shape is even more pronounced and the GaAs surface temperature is nearly the same as the interface temperature, as shown in Figures 6.11 and 6.12. Using = 0.5 the surface temperature of the GaAs is higher than the surface temperature of the Si and the U-shape is still present, as shown in Figures 6.13 and 6.14. 6.6 STRESS CALCULATIONS The model results show the temperature profile across the GaAs sample and the Si wafer. In addition to the temperature gradients it is useful to know the stresses associated with these profiles. To calculate the stress the GaAs was assumed to be a thin plate of uniform thickness with the temperature a function of a; only, using the coordinate system shown in Figure 6.15. For this case, the stress in the y-direction is given by, 1 2 1 aET 1 f* rrrJ 3x rrr . (6.15) a =-- + - — oJETdx +— aETxdx y 1 - v 2c(l-v)J-c 2c\l-v)J-c where a is thermal expansion coefficient, E is the Young's Modulus, v is Poisson's ratio and c = \ x the thickness of the sample. At 1000 K, a = 8.5x10"6 K"1, E = 7.7xl0 3 M P a and v = .3106 (The calculations for these values are shown in Appendix C). 140 The temperature profile in Figure 6.8 has the largest temperature gradi-ents in the GaAs sample among Figures 6.7 to 6.9 therefore this temperature distribution should produce the maximum thermal stresses. This temperature profile can be approximated by a parabola given by centred at x = -175 pm. Using T 0 = 1062.96 K and b = 8689 pm gives the parabola shown in Figure 6.16. To solve for the thermal stress it useful to perform a coordinate change to make the integrations easier therefore x is replaced by x' where x'=x + c0. Equation 6.15 then becomes, For the parabola shown cD = 175 pm. The stress profile, calculated using equation 6.17, is shown in Figure 6.17. As expected from the temperature profile the stresses are not symmetric. The thermal stress was also calculated for the nearly symmetric profile shown in Figure 6.11. The parabolic approximation for this profile is shown in Figure 6.18, for T„=l l 37.47 K and b=8431 pm. The resulting stress profile is shown in Figure 6.19. (6.16) (6.17) 141 The critical resolved shear stress can be calculated using the expression given by Jordan, as follows, 1 1 4 At 1000 K, the critical resolved shear stress = 1.63 x 10 7 dynes/cm2 = .163 MPa. The stresses in both cases are large, exceeding the critical resolved shear stress (CRSS) in places. 6.7 CONCLUSION The model results show that the the temperature distribution has a minimum near the centre of the wafer. The exact shape of the distribution depends on the degree of radiative absorption at the surfaces which is not known. In most cases examined the temperature distribution was asymmetric except in the case shown in Figure 6.11. dynes I cm2 (6.18) 142 S T E P W I S E A P P R O X I M A T I O N O F S P E C T R A L E M I S S I V I T Y O F G a A s - f \ Normalised Blackbody Output - / \ " T = 1 7 5 0 K -/ / / / \ \ 0 * I I I I I 2 4 £ C u t o f f WAVELENGTH (Microns) W a v e l e n g t h Figure 6.1 Stepwise approximation of the spectral emissivity of GaAs. The cutoff wavelength depends on the temperature of the GaAs. The spectral output of a 1750 K blackbody is superimposed on this graph for reference. E F F E C T I V E EMISSIVITY O F G a A s A S A F U N C T I O N O F T E M P E R A T U R E 0.145 -1 1 0.105 400 600 800 1000 T E M P E R A T U R E (K) Figure 6.2 Calculated emissivity of GaAs as a function of source temperature and the GaAs temperature. 143 o o CO <£ DC Si 0.45 H E A T T R A N S F E R C O E F F I C I E N T O F N2 G A S A S A F U N C T I O N O F G A P W I D T H 40 80 120 -i 1 160 200 1 r 240 280 G A P WIDTH (M ic rons ) Figure 6.3 Calculated heat transfer coefficient due to the gas in the GaAs-Si air gap as a function of gap width. M e a s u r e d v e r s u s C a l c u l a t e d C o o l i n g C u r v e s Ul DC i 12 16 T I M E (s) Figure 6.4 Calculated cooling rates for various values of the convective heat transfer coefficient, h c, compared to the measured cooling rate. 144 C A L C U L A T E D V E R S U S M E A S U R E D H E A T I N G R A T E g UJ cc => 1 UJ a. S 1 1 5 0 S i E m i s s i v i t y = 0 . 7 T B U L B = 1 6 2 5 K 1 2 3 T I M E ( s ) _ — M e a s u r e d -+— Ca l cu la ted Figure 6.5 Measured heating rate compared to the calculated heating rate with Esi = 0.7, bulb temperature = 1625 K. o. C A L C U L A T E D V E R S U S M E A S U R E D H E A T I N G R A T E 1 1 5 0 S I E m i s s i v i t y = 0 . 5 T B U L B = 1 7 2 5 K T I M E ( s ) M e a s u r e d Ca l cu la ted Figure 6.6 Measured heating rate compared to the calculated heating rate with Eg; = 0.5, bulb temperature = 1725 K. 145 C A L C U L A T E D T E M P E R A T U R E P R O F I L E 8 8 5 -8 8 4 -8 8 3 -8 8 2 -E(K) E(K) 8 8 1 -CC 8 8 0 -DC D_ E 8 7 9 -8 7 8 -8 7 7 -8 7 6 -8 7 5 -S i S U RFA It I T E F I F A C : E f , Gal is S I J R F i V C E T I N IE = ' 1.5 S S i I E M I S s i v n Y = 0 . 7 T B J L B = 161 5 K 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 POSITION (Microns) Figure 6.7 Calculated temperature profile with = 0.7 at time t=1.5 s. C A L C U L A T E D T E M P E R A T U R E P R O F I L E 1 0 7 0 -1 0 6 9 -1 0 6 8 -1 0 6 7 -U J 1 0 6 6 -OC => 1 0 6 5 -OC s 1 0 6 4 -U J i — 1 0 6 3 -1 0 6 2 -1 0 6 1 -1 0 6 0 -S i S U 3 F A < D E It I T E F F A C E G a y i s S I J R F i V C E T I N I E = : J O s S i E M I S i s i v r r Y = D.7 T B U L B = 1 6 : ! 5 K 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 POSITION (Microns) Figure 6.8 Calculated temperature profile with eSi = 0.7 at time t=3.0 s. 146 C A L C U L A T E D T E M P E R A T U R E P R O F I L E cc UJ Q. 8 8 0 8 7 9 8 7 8 8 7 7 8 7 6 8 7 5 8 7 4 8 7 3 8 7 2 8 7 1 8 7 0 S I £ J U R I : A C I IN T E R F A C I E i G a l , s S I J R F J k C E T I M E = 1 .5 s S i E n i s s v i t y = O.S T B I I L B : : 1 7 2 5 K 2 0 0 4 0 0 6 0 0 8 0 0 P O S I T I O N ( M i c r o n s ) 1 0 0 0 Figure 6.9 Calculated temperature profile with e S i = 0.5 at t=1.5 s. C A L C U L A T E D T E M P E R A T U R E P R O F I L E 1 0 7 0 1 0 6 9 1 0 6 8 1 0 6 7 1 0 6 6 1 0 6 5 1 0 6 4 1 0 6 3 1 0 6 2 1 0 6 1 1 0 6 0 I S U R F A \ C E It I T E F F A C E ; G a / i s S I J R F i V C E T I N IE = i I.O s S i E M I S SIVI1 r Y = 3.5 T B J L B = 1 7 5 5 K 2 0 0 4 0 0 6 0 0 POSITION (Microns) 8 0 0 Figure 6.10 Calculated temperature profile with egj = 0.5 at t=3.0 s. C A L C U L A T E D T E M P E R A T U R E P R O F I L E 1 1 4 5 1 1 4 4 1 1 4 3 1 1 4 2 1 1 4 1 1 1 4 0 1 1 3 9 1 1 3 8 1 1 3 7 1 1 3 6 1 1 3 5 < i i S L R F / C E IN T E R F A C E G a / , s S l J R F V ^ C E " G a Vs E i 1 n i s s i wity = : 0 .3 S i E :mis« t iv i ty = O.I T B I J L B = 162 5 K 2 0 0 4 0 0 6 0 0 P O S I T I O N (Microns) 8 0 0 1 0 0 0 Figure 6.11 Calculated temperature profile with = 0.3 and e S i = 0.7. C A L C U L A T E D T E M P E R A T U R E P R O F I L E UJ cc 1 1 6 5 1 1 6 4 1 1 6 3 1 1 6 2 1 1 6 1 1 1 6 0 1 1 5 9 1 1 5 8 1 1 5 7 . 1 1 5 6 1 1 5 5 G a y LS En l i s s i / i t y = 0 .3 S i E m i s s i v i t y = 0.£ T B J L B = 171 !5 K SI S i J R F i \ C E GaA s S I I R F / • .CE IN T E R F A C I E \ -2 0 0 4 0 0 6 0 0 8 0 0 P O S I T I O N ( M i c r o n s ) 1 0 0 0 Figure 6.12 Calculated temperature profile with = 0.3 and e S i = 0.5. C A L C U L A T E D T E M P E R A T U R E P R O F I L E g UJ 1 2 2 0 1 2 1 9 1 2 1 8 1 2 1 7 1 2 1 6 1 2 1 5 1 2 1 4 1 2 1 3 1 2 1 2 1 2 1 1 1 2 1 0 L S S I J R F i I C E S i S I J R F i V C E l is T E R F A C E G a / ! LS E n • i s s h rity = 0 . 5 S i E m i s s i v i t y = 0 . 7 T B I I L B = : 1 6 2 > K 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 POSITION (Microns) Figure 6.13 Calculated temperature profile with =0.5 and e S i = 0.7. C A L C U L A T E D T E M P E R A T U R E P R O F I L E 1 2 6 8 1 2 6 7 1 2 6 6 1 2 6 5 1 2 6 4 1 2 6 3 1 2 6 2 1 2 6 1 1 2 6 0 1 2 5 9 1 2 5 8 G a / L S S I J R F i V C E G a . !Vs E i n i s s i v i t y : = 0 . 5 S i 1 i m i s 5 i v i t y = O. > T B J L B = 171 5 K S i S J R F \ C E ir I T E F I F A C : E 200 4 0 0 6 0 0 8 0 0 P O S I T I O N ( M i c r o n s ) 1000 Figure 6.14 Calculated temperature profile with =0.5 and e S i = 0.5. Top Surface of GaAs Sampl X -X = Lower Surface of GaAs sample Figure 6.15 Coordinate system used for stress calculations. 150 TEMPERATURE PROFILE -300 -100 100 300 POSITION (Microns) Figure 6.16 Parabolic approximation to the temperature distribution in the GaAs shown in Figure 6.8. STRESS PROFILE 0.14 - i 0 200 400 600 POSITION (Microns) Figure 6.17 Calculated stresses for temperature profile shown in Figure 6.16. 151 TEMPERATURE PROFILE 1 1 1 1 1 1 -300 -100 100 300 POSITION (Microns) Figure 6.18 Parabolic approximation to the temperature distribution in the GaAs shown in Figure 6.11. STRESS PROFILE 0.05 0 200 400 600 POSITION (Microns) Figure 6.19 Calculated stresses for temperature profile shown in Figure 6.18. 152 C H A P T E R 7 D I S C U S S I O N 7.1 F U R N A C E A N N E A L I N G The results from the furnace annealing showed two phenomena: 1) the luminescence from the centre of the dislocation cells increased and 2) a dark region developed that was located along the cell walls but outside the halos. The cause of the overall brightening in the cell interiors was not determined but it may be due to the same cause as the brightening observed in the rapid thermal annealing, which will be discussed in the next section. The dark region lying near the cell walls may represent an area where the brightening seen in the cell interior does not take place or it may be associated with the presence of a defect. Photoluminescence analysis did not show any correlation with any known P L peaks - however the B N R scanning P L apparatus does not use a very high powered laser and therefore may not generate sufficient signal for detection. Although this dark region became less noticeable at higher temperature anneals, it did not completely disappear. The fact that this area remained most distinct along cell walls that showed a large number of dislo-cations suggests that the dark area may be associated with the stresses due to the dislocations. 7.2 R T A A N N E A L I N G The results from the R T A anneals were similar to the furnace anneals. The samples annealed below 800 °C showed a brightening in the interior of large cells. In areas of high dislocation density this brightening was not seen perhaps 153 due to the development of the dark region along the cell walls. If the cell size was small enough then the dark regions could overlap, obscuring the brightening effect. At higher temperatures the dark regions were present where the disloca-tions showed a high degree of clustering but were not present around isolated dislocations. Particularly interesting was the fact that no halos could be seen around individual dislocations. 7.2.1 Cross-section Results The most remarkable result from the R T A samples was the behaviour of the luminescence through the thickness of the sample. For samples annealed above 850 °C a band of higher C L intensity could be seen adjacent to both surfaces. These regions, referred to as bright bands, increased slightly in depth with time and temperature. They were observed using the Bell Northern C L system and they could be seen in the scanning P L maps of the intensity of the 810 nm peak across the cross-section. 7.3 T H E S O U R C E O F C O P P E R In GaAs The P L spectra from the taper samples correlated the presence of C u with the bright bands. Consequently samples were treated, prior to R T A annealing, to change the residual C u level on the surface. Since the use of K C N has been reported to remove C u from the surface of GaAs 2 0 , one sample was immersed in a K C N solution. This sample showed distinctly shallower bright bands than the untreated sample from the same wafer. Conversely the sample treated with 154 C u S 0 4 solution showed much deeper bands than the untreated sample. These results clearly support the correlation between the presence of C u and the bright bands. The C u in the untreated samples probably comes from C u diffusing in from the wafer surface. Some C u may come from the quartz in the furnace but the effectiveness of the K C N treatment implies that most of the C u was present before treatment. Therefore the C u must be left from polishing and other surface treatments. The amount of C u left on the surface could be expected to vary among manufacturers and sometimes from boule to boule. The difference in band depth between boules A and B may be due to different amounts of residual Cu. There is a difference in the depth of the bright bands between the (100) sample and the (110) sample. The bands are only 40 to 80 pm deep in the (110) sample compared to 175 to 200 pm in the (100) sample from the same boule. This orientation dependence of the bright bands may reflect a difference in the adsorption of C u on the (100) and (110) surfaces. Since the adsorption of atoms is related to the Fermi-level at the surface the difference in the adsorption may indicate a difference in the Fermi level at these orientations. 1 4 1 The residual C u on the surface is not detected in the P L spectra of the as-received wafers, as shown in the P L spectra of the samples from boules A and B before heat treatment (see Figures 5.78 and 5.83). The C u on the surface is not occupying lattice or interstitial sites therefore the C u does not introduce any detectable impurity levels into the band gap. No attempt was made to 155 measure the C u concentration using the SIMS system since it cannot provide quantitative information without a standard and it has a high detection limit for C u as shown in ref 123. 7.4 D I F F U S I O N O F C O P P E R I N G a A s Because of the depth of the bands and the short annealing times, the dif-fusion of C u must be very fast. C u is known to have a high diffusion rate 1 2 4 which is usually associated with interstitial diffusion. Frank and Turnbul l 1 2 5 found that the diffusion rate of C u was higher in near perfect Ge crystals compared to Ge crystals with low angle grain boundaries. They proposed that for the near perfect crystal the C u diffused purely interstitially. In a crystal with a large number of dislocations where the vacancy concentration is higher the ratio of the concentrations of interstitial to substitutional C u atoms remains fixed and is the same as their ratio of their solubilities. Each atom spends only a fraction of its time in a state of high diffusivity. The result is a diffusion rate slower than the interstitial diffusion rate but faster than expected for normal substitutional diffusion. Kendal l 1 2 6 noted that in a compound the divacancy complex (Schottky defect) is likely to be more mobile than single vacancies and may be present in larger concentrations. The reaction needed to form a substitutional C u impurity in the case of a single vacancy is CUi + Vcit-tCua,, (7.1) whereas for a divacancy the following reaction takes place cui + {vayAt}<r>CuGa+vAt (7.2) 156 If the divacancy does not dissociate readily the interstitial species would be favoured. Evidence for some substitutional diffusion being present comes from the slight enhancement of the band depth near dislocations (see Figures 5.45,5.50, 5.51 and 5.59). Such enhancement would be expected in the case of substitu-tional diffusion since the vacancy concentration is higher at dislocations. 1 4 0 This enhancement is not always observed near dislocations therefore most atoms must be diffusing interstitially. Using the data from Hal l and Racette 1 2 7 (Do=.03, Q=.53 eV) a diffusion coefnent, D, for interstitial C u diffusing into GaAs at 950 °C is calculated to be 1.9X10"4 cm2/s. The equation for a semi-infinite source of C u diffusing into a semi-infinite solid is , 1 2 8 C=C0erfc f ~ ^ (7.3) x where C is the concentration of C u at point x at time t, and C 0 is the initial concentration at the surface. Using this equation a value for x can be estimated assuming that where the bright band ends C « C 0 . When C=.l 56C 0 the argument of the error function is equal to one therefore x2=4Dt. The choice of this point is arbitrary but should give an order of magnitude for the value of x for the diffusion conditions being considered. Using the calculated value for D, x should be 616 pm for an annealing time of 5 s. This value is much larger than a typical bright band depth of 115 pm but the threshold concentration at which the bright bands ends may be higher than assumed in which case x would be less. 157 Plotting the square of the band depths from Tables TV and V versus annealing time does not give a single value for D at 850 °C, as shown in Figure 7.1. The lines do not pass through zero. A reason source for this dis-crepancy may be the fact that the equation 7.3 assumes that there is a semi-infinite source of C u at the surface. In the samples studied the C u concentration is low and may be depleted by the diffusion process. A more accurate equation to use in this case would be the case for a thin planar source into a semi-infinite solid where, 1 2 8 /-  m C = exp f _.2 , \ X K ADt (7.4) where m is the amount of C u which is unknown. Since m is not known this equation cannot be used to estimate x. However, since the source is depleting the driving force for diffusion decreases with time therefore x would be smaller in this case. Assuming the C u concentration at the edge of the band is given by C=C e d g e then according to equation 7.4, at time tx it should be given by, n m f xl \ (7.5) 4Dti where xx is the depth of the bright band at time tx. Similarly at time t2 it should be given by, 158 This equation assumes that the diffusion has occured at the same temperature as equation 7.5 since the same value for D is used in both equations. Assuming that the edge of the band occurs at the same C u concentration for both temperatures then, m ^nDt! exp f 2 \ m -exp ^ x2 ^ ADt2 (7.7) rearranging equation 7.6 gives, 2 2 / " . + __£_ = l n ADti 4Dt2 ^Dti (7.8) rearranging equation 7.7 to solve for D gives, D = ( 2 K4ti 4t2 xl ^ /In (7.9) Since D is constant at a given temperature then using equation 7.9 and the measured band depths at different times should give a constant value for D. However calculating values for D using the band depths shown in Tables IV and V does not give a fixed value for D, as shown in Table VI. In addition the values of D are much lower than expected. There are two reasons that can explain why the diffusion process is not obeying equation 7.7. First, the diffusion coefficient may be dependent on the C u concentration and is decreasing with the decreasing concentration at the surface. Looking at the calculated D values for the data from Table V, the value of D decreases with longer annealing times. 159 Second, the assumption that the band edge occurs at a fixed value of the C u concentration may not be valid therefore equation 7.7 is not valid. This may also be true since the edge of the band appears abrupt in the C L images. It is therfore likely that the edge of the band may represent a region where the C u diffusion is stopped suddenly by the fast cooling rate which lowers the tem-perature to the point where the diffusion is very low. As a result the band edge may represent an region where the C u concentration falls off abruptly and therefore is not represented by a fixed C u concentration. This assumption is supported by the results from the sample that was slow cooled in the R T A test series #14. This sample has a less abrupt transition from the bright region to the darker region than samples R T A annealed at the same time and tempera-ture. Another deviation from standard diffusion behaviour is the appearance of the bright bands in the 775 °C sample although no bands are apparent in the 750 °C sample. The sudden appearance of the bright bands after only 25 °C change in temperature may be related to the behaviour of the C u on the surface. Since the C u is likely to be present in the form of copper oxide on the surface the oxide would have to dissociate for the C u to be free to move in the lattice. The change in temperature from 750 °C to 775 °C may result in an increase in the free energy of the oxide to the point where it is higher than that for the C u and oxygen separately. This change would then favour the dissociation of the oxide. A n increase in the dissociation would increase the C u available for dif-fusion. 160 7.5 E F F E C T O F V A C A N C I E S O N B R I G H T B A N D S If the bright bands are related to the diffusion of C u from the specimen surface then the band depth should not depend on the As vacancy concentration. This is consistent with the observations made in series #9 in which two GaAs wafers were R T A annealed face to face. In the cross-section C L image of the top wafer the bright band depth adjacent to the surface which was in contact with the lower GaAs wafer was approximately the same depth as the band near the exposed surface. The As vacancy concentration at the exposed surface should be higher due to the greater loss than at the shared GaAs surface. In the results of Chin et a l . 9 5 the presence of a silicon nitride cap suppressed the brightening observed in their samples and they attributed this behaviour to the reduction in the As vacancies because of the use of the cap. Although not sensitive to the As vacancies the diffusion of C u could be sensitive to the presence of G a vacancies since the C u occupies G a sites in the lattice. Since nitride caps are known to suppress the outdiffusion of G a it may have been the lower Ga vacancy concentration that resulted in the suppression of the brightening. In the nitrided samples in this study, no suppression was observed but this could be due to the oxygen content of the nitride. Only a very low concentration of O in the nitride (-0.1 atomic percent) will successfully suppress the loss of Ga . 8 5 The oxygen concentration of the nitride used is not known. Another possibility may be that the residual C u on the surface was removed in the cleaning process used by Chin et al. prior to nitriding and therefore less C u was available for diffusion. 161 7.6 O T C S R E S U L T S It was hoped that the O T C S scans would show a peak at 362 K in the ring-dots corresponding to the bright bands but not in the centre of the sample. According to the results of T in et a l . 6 0 , 1 3 0 , a peak at 362 K was found to be related to the presence of Cu. However, using the same rate window as T in et al., a peak at 350 K was present for all ring-dots examined. This peak does not appear to be the same peak observed by T in et al. and could be related to the E L 2 level as suggested by T in et a l . 6 0 The 362 K peak in the results of T in et a l . 6 0 appeared near the surface of samples that had been diffused with Cu. The activation energy of this peak was found to be 0.5 to 0.52 eV which is comparable to the 0.44 eV value for the activation energy of the peaks seen in D L T S studies. 1 3 1" 1 3 3 It should be noted that the C u peak in the D L T S spectra of reference 131 was at a lower temper-ature, near 210 K. The 362 K peak was not present in scans of areas deeper than 70 pm from the surface even though P L spectra still showed the presence of Cu. The authors believed that this peak was related to either a CuGa-VA s or a V A s -CuG a -V A 8 complex. In addition in a control sample where C u was not intentionally introduced, the C u peak at 1.36 eV (911 nm) was found in the P L spectra of the annealed sample. The intensity of the C u peak increased with depth into the sample. The increase in the 1.36 eV P L peak could be due to the decrease in the amount of C u tied up in the Cu-V A s defects. It is possible that this defect complex was not present at all in the samples in this study therefore the 362 K peak was not seen. 162 A peak at 270 K was seen in the ring-dots corresponding to the bright region but not elesewhere. As well a peak around 100 K was seen in the normalized data of the extended temperature scan of the ring-dot from the bright region but not in the same scan for the ring-dot in the centre. The significance of the negative peaks is not clear. The negative peak appears to shift from near 275 K in the ring-dots from the centre region to 310 K in the ring-dots in the bright region. Possibly these represent two different negative peaks. 7.7 M A T H E M A T I C A L M O D E L R E S U L T S The model results show that the temperature in the wafers has a parabolic distribution. In most cases, the centre of the parabola is well away from the centre of the wafer resulting in an asymmetric temperature distribution in the wafer. In most cases the bright bands appear to the same depth at the top and bottom surfaces. If the temperature gradients have influenced the development of the bands then their depths would not be equal. If the temperature profile in the GaAs resembles the profile shown in Figure 6.11 then the thermal stresses are nearly symmetric. In this case the temperature distribution could be affecting the band formation. However this symmetric temperature distribution was only obtained for one set of emissivity values and in all other cases the model implied that the temperature distribution would be asymmetric. 163 7.8 T H E E F F E C T O F C O P P E R O N L U M I N E S C E N C E From the results of this study it appears that the C u is able to enhance the luminescence from the GaAs. The enhancement can be seen at liquid He temperature in the P L scans and at room temperature in the C L images. The C u may be changing C L intensity by decreasing the diffusion length of the carriers. The C L signal, S is related to the minority carrier diffusion length, L, as follows, 1 3 4 1 (7.10) where a is the absorption coefficient. Decreasing the diffusion length will increase the C L signal. Eckstein et a l . 1 3 6 claimed that C u decreased the diffusion length by 20% resulting in an increase in the contrast in their C L images. No mechanism was given for the decrease in diffusion length. The behaviour may be related to the location of the C u in the lattice. The interstitial C u acts as a donor 1 3 7 1 3 7 and the substitutional C u acts as an acceptor. The solubility of the substitutional C u is 30 times the solubility of the interstitial C u . 1 2 7 This solubility ratio depends on the number of vacancies present in the material. Hasegawa 1 3 8 found that C u diffused into epitaxial GaAs did not reduce the carrier concentration unlike diffusion into the bulk-grown GaAs substrate. This effect was attributed to the fact that more C u remained in interstitial sites in the epitaxial material where there are fewer vacancies than in the bulk grown material. 164 In the case of the R T A samples the cooling may be fast enough to freeze in the interstitials. Some interstitials would also be present in furnace annealed samples but to a lesser degree. Since the interstitial C u acts as a donor, its presence will increase the number of electrons in the sample. The effect of the increased electron concentration would be a decrease in the diffusion length of the hole by increasing the probability of recombination taking place, thus increasing the C L intensity. This type of increase in the C L signal can be seen in the implanted sample in Figures 5.88 and 5.89 where the implanted face has a signal that is three times higher than the back face. The substitutional C u has deep acceptor levels in the band gap and therefore is more likely to act as a trap. In both types of annealing the vacancy concentration would be expected to be higher near the dislocations. The regions near the dislocations that had lower C L intensity may have been due to a higher concentration of the substitutional form of C u whereas in the bulk the concentration of substitutional C u is lower therefore it appears brighter. Insufficient data was available on the behaviour of interstitial C u to explore this hypothesis further. More information is needed on the population of the C u interstitials in the bright bands, it would also be interesting to know if the C u interstitial could be related to 1.36eV P L peak or i f there is another peak corresponding to this site. 7.9 CONCLUSION The results from this study have related the brightening observed in the C L images of GaAs after heat treatment to the presence of low levels of Cu. This 165 brightening phenomena is common to both furnace and R T A annealed samples and is believed to due to a decrease in the carrier diffusion length. It has been proposed that the interstitial C u acts to increase the electron concentration by acting as a donor. The brightening that occurs due to the presence of the C u obscures any other changes that may have occurred during annealing. T A B L E VI Calculation of Diffusion Coefficient using Equation 7.9 T E M P E R A T U R E ti x 2 D °C (s) (cm) (s) (cm) (cm2/s) 850 ( T A B L E TV) 5 .0090 10 .0122 8.0x10"7 II 5 .0090 18 .0137 2.1x10 6 it 18 .0122 18 .0137 3.7x10"6 850 ( T A B L E V) 5 .0100 20 .0130 4.6x10 6 II 5 .0100 160 .0155 3.6x10"6 II 20 .0130 160 .0155 1.2xl0"6 167 2.50E-04 2.40E-04 2.30E-04 2.20E-04 2.10E-04 2.00E-04 1.90E-04 1.80E-04 1.70E-04 1.60E-04 1.50E-04 1.40E-04 1.30E-04 1.20E-04 1.10E-04 1.00E-04 9.00E-0S 8.00E-05 • Table IV: 850 C + Table V: 850 C o Table IV: 950 C n i i i i i 1 1 1 1 1 1 1— 40 60 80 100 120 140 160 Time (s) Figure 7.1 Plot of the square of the bright band depth (x2) versus annealing time for the results shown in Tables IY and V. 168 CHAPTER 8 SUMMARY AND CONCLUSIONS To summarize, the main results of this study are: 1) After furnace annealing, the dislocation spots and halos are still present but the C L intensity increases in the centre of the cells. A n area of lower C L intensity persists along the cell walls. 2) R T A annealing above 800 °C causes an increase in the background C L intensity. This increase results in a near uniform C L intensity in the image of the specimen surface, except for the dislocations spots and the areas of lower C L intensity near the cell walls. 3) C L images of the cross-sections of specimens that have been R T A annealed above 800 °C show that the increase in background intensity does not extend throughout the sample thickness. As a result the cross-section C L image has regions of higher C L intensity adjacent to the sample surface. The depth of these bright bands is increased somewhat by increasing annealing time and temperature. 4) The bright bands seen in the R T A annealed samples have been corre-lated with the presence of the 1.36 eV peak in liquid He P L spectra. This peak is associated with the presence of C u in GaAs. 5) The level of residual C u on the surface of the samples has a strong influence on the depth of the bright bands. Decreasing the residual C u level decreases the depth of the bright bands while increasing the C u level increases the depth. 169 6) The orientation of the sample affects the depth of the bright bands. The depth of the bright bands in (110) samples is significantly less than (100) samples from the same boule. 7) According to the model results, the presence of the thermal gradients does not appear to influence the depth of the bright bands since in most cases the thermal gradients are asymmetric but the bright band depth is the same at both surfaces. 8) The increase in the C L intensity has been attributed to a decrease in the minority carrier diffusion length. 9) The influence of the C u on the integrated light C L images makes it difficult to determine the changes that have taken place in the GaAs during annealing. 170 C H A P T E R 9 F U T U R E W O R K Recommendations for future research include, 1) Further study into the origin of the regions of lower C L intensity that persist along the cell walls is needed. These dark regions may be responsible for the detrimental behaviour associated with clusters of dislocations. 2) The use of monochromatic C L images may be more useful in the determination of the behaviour of individual peaks before and after annealing. This may lead to a better understanding of the annealing behaviour of the bulk GaAs. 3) The effect of C u on the presence of the bright bands needs to investigated more fully. Future work should be aimed at determining the mecha-nism of the enhancement of the C L intensity. 4) The difference in the behaviour of the (110) and (100) samples should be examined. This work should try to establish whether this behaviour is due to a fundamental difference in the adsorption between these faces or to an orientation dependence of the diffusion of C u in GaAs. 5) The behaviour of the O T C S spectra needs to be investigated in greater detail. The taper polishing technique should be modified to reduce the level of residual damage after polishing. The size of the ring-dot structures in normal O T C S make it necessary to use a tapered specimen 171 however improvements in the scanning O T C S technique may make it possible to avoid the problems with the ring-dot size and make direct observations of a cleaved edge thus avoiding the polishing damage. 6) Additional research should be performed on the electrical behaviour of the bright bands. In particular a resistivity map of a taper sample with bright bands would be useful in determining the effect of the Cu. 172 B I B L I O G R A P H Y 1 H . E m o r i , T. K i k u t a , T. Inada, T. Obokata, & T. F u k u d a , J p n . J . A p p l . Phys . , v24, L291,(1985). 2 D . La is ter & G . M . J e n k i n s , G . M . , J . Mater . Sci . , v8 ,1218, (1973). 3 V . Swaminath ian & S . M . Copely, J . Amer . Ceram. S o c , v58, 482, (1975). 4 P . A . Le igh , J . P . H a l l , C R . El l io t t B . Wakf ie ld , & M . H . Lyons, Semi-Insulat ing III-V Mater ia ls 1984, 214, (1984). 5 D . J . S t i r land , P . D . Augustus , M . R . Brozel & E . J . Foulkes, Semi-Insulating III-V Mater ia ls 1984, 91, (1984). 6 R . D . F a i r m a n , R.T. Chen , J .R . Ol iver & D.R . Ch'en, I E E E Electron Devices, ED-28 ,135 , (1981). 7 H . M . Hobgood, G . W . Eldridge, D . L . Barrett , & R . N . Thomas, I E E E Trans. Electron Devices, vED-28 ,140 , (1981). 8 J . Krauskopf , J . D . Meyer , B . Weidemann, M . Waldschmidt, K . Bethge, G . Wol f & W . Schutze, Semi-Insulating III-V Mater ia ls 1988,165, (1988). 9 K . E l l io t t , D . E . Holmes, R.T. Chen, & C G . K i r k p a t r i c k , A p p l . Phys . Lett . , v40, 898, (1982). 10 W . M . Duncan , G . H . Westphal , & J . B . Sherer, I E E E Electron Device Let-ters, v E D L - 4 , 1 9 9 , (1983). 11 J . B . Clegg, Semi-Insulating III-V Mater ia ls 1982, 80, (1982). 12 D . E . Holmes, R .T Chen, K . R . E l l io t t , C G . K i r k p a t r i c k , & P . W . Y u , I E E E Trans . Electron Devices, vED-29 ,1045 , (1982). 13 D . E . Holmes, R .T Chen, K . R . E l l io t t , & C G . K i r k p a t r i c k , A p p l . Phys . Lett . , v40, 46, (1982). 14 Y . N a k a m u r a , Y . Ohtsuki , Y . Itoh, J . K i k a w a , & Y . Kashiwayanagi , Inst. Phys . Conf. Ser. No . 79, 49, (1986). 15 K . Watanabe, H . N a k a n i s h i , Y . Y a m a d a & K . Hoshikawa, A p p l . Phys . Lett . , v45, 643, (1984). 16 C A . W a r w i c k & G.T . B r o w n , A p p l . Phys . Lett . , v46, 574, (1985). 173 17 T. Kikuta, J . Katsumata, T. Obokata, & K. Ishida, Inst. Phys. Conf. Ser. No. 74, 47, (1985). 18 T. Hiramoto & T.Ikoma, Semi-Insulating III-V Materials 1988 (Mal-mo,Sweden), 337, (1988). 19 Z. Q. Fang, L. Shan, J . H . Zhao, X .J . Bao, T . E . Schlesinger, A .G . Milnes, G.R. Yang, & W.M. Lau, J . of Electronic Materials, v l8 ,123, (1989). 20 G.B. Larrabee, J . Electrochem. Soc. vl08,1130, (1961). 21 L.C. Kimerling, & J . M . Parsey,Jr., J . of Metals, 60, (1985). 22 P. Gallagher, Masters Thesis, Dept. of Metals and Materials Engineering, University of British Columbia, (1987). 23 S. Schneider, Semi-Insulating Materials 1982,144, (1982). 24 J . Lagowski, J . M . Parsey, M. Kaminska, K. Wada, & H.C. Gatos, Semi-Insulating III-V Materials 1982,154, (1982). 25 D.E. Holmes, R.T Chen, & J . Yang, Appl. Phys. Lett., v42, 419, (1983). 26 M. Taniguchi & T. Ikoma, J . Appl. Phys., v54, 6448, (1983). 27 W. Frank, Inst. Phys. Conf. Ser. No. 79, 217, (1986). 28 T. Ikoma, M. Taniguchi, & Y. Mochizuki, Inst. Phys. Conf. Ser. No. 74, 65, (1985). 29 H.J. Queisser, Defects in Semiconductors II, 323, (1983). 30 F.C. Wang & M . Mujati, I E E E Trans. Electron Devices, vED-32, 2839, (1985). 31 R. Zucca, B .M. Welch, P.M. Asbeck, R.C. Eden & S.I. Long, Semi-Insulating III-V Materials 335, (1980). 32 P. Ashburn, G. Bull , K .H. Nicholas, and G.R. Booker, Solid St. Electron., V20, 731,(1977). 33 G. Bull , P. Ashburn, G.R. Booker, & K.H. Nicholas, Solid St. Electron., v22, 95, (1979). 174 34 T. Matsumura, H . Emori, K. Terashima, & T. Fukuda, Jpn. J . Appl. Phys. Pt 2, v22, L I 54, (1983). 35 T. Honda, Y. Ishii, S. Miyazawa, H. Yamazaki, & Y. Nanishi, Jpn. J . Appl. Phys. pt 2, v22, L270, (1983). 36 K. Matsumoto, H . Morishita, M. Sasaki, S. Nishine, M. Yokogawa, M. Sekinobu, K Tada, & S. Akai, Semi-Insulating III-V Materials 1984, 175, (1984) 37 R. Rumbsy, I. Grant, M.R. Brozel, E.J. Foulkes & R.M. Ware, Semi-Insulating III-V Materials 1984,165, (1984). 38 A. Tamura & T. Onuma, Jpn. J . Appl. Phys., v24, 510, (1985). 39 F.A. Ponce, F.C Wang & R. Hiskes, Semi-Insulating III-V Materials 1984, 68, (1984). 40 T.Obokata, T. Matsumura, K. Terashima, F. Orito, T. Kikuta & T. Fukuda, Jpn. J . Appl. Phys. Pt. 2, v23, L602, (1984). 41 R. Rumsby, R.M. Ware, B. Smith, M. Tyjberg, M.R. Brozel & E.J. Foulkes, I E E E 1983 GaAs IC Symposium, 34, (1983). 42 Y. Nanishi, S. Ishida, T. Honda, H. Yamazaki & S. Miyazawa, Jpn. J . Appl. Phys. Pt. 2, v21, L335, (1982). 43 T. Takebe, M. Shimazu, A. Kawasaki, T. Kotani, R. Nakai, K. Kikuchi, S. Muria, K. Tada, S. Akai & T. Suzuki, Semi-Insulating III-V Materials 1984, 335, (1984). 44 S. Miyazawa, Y. Mizutani,& H. Yamazaki, Jpn. J . Appl. Phys. Pt. 2, v21, L542, (1982). 45 H.V. Winston, A.T. Hunter, H .M. Olsen & R.E Lee, Appl. Phys. Lett., v45, 447, (1984). 46 H.V Winston, A.T. Hunter, H. Kimura, H.M. Olsen, R.P Bryan, R.E. Lee & O.J. Marsh, Inst. Phys. Conf. Ser. No. 74, 497, (1985). 47 C A . Evans, V.R. Deline, T.W. Sigmon, & A. Lidow, Appl. Phys. Lett., v35, 291, (1979). 48 F. Hyuga, Jpn. J . Appl. Phys. pt 2, v24, LI 60, (1985). 175 49 S. Miyazawa & F. Hyuga, I E E E Trans. Electron Devices, vED-33, 227, (1986). 50 F. Hyuga, H . Kohda, H. Takanishi, T. Kobayashi, & K. Hoshikawa, Appl. Phys. Lett., v47, 620, (1985). 51 K. Watanabe, F. Hyuga, H. Nakanishi & K Hoshikawa, Inst. Phys. Conf. Ser. No. 79, 277, (1986). 52 M. Shinohara, F. Hyuga, K. Watanabe & Y. Imamura, J . Appl. Phys., v60, 304, (1986). 53 M . Shinohara, J . Appl. Phys., v61, 365, (1987). 54 E .M. Swiggard, I E E E 1983 GaAs IC Symposium, 26, (1983). 55 H. Yamazaki, Y. Nanishi, S. Miyazawa & Y. Ishii, I E E E 1983 GaAs IC Symposium, 30, (1983). 56 F. Orito, F. Seta, Y. Yamada, T. Ibuka, T. Okano, F. Hyuga & J . Osaka, I E E E 1986 GaAs IC Symposium, 33, (1986). 57 P.K. Willardson & A.C. Beer, eds. Semiconductors and Semimetals, v20 : S.L GaAs. Academic Press, New York:1984. 58 J . Williams & J . M . Poate, eds. Ion Implantation and Beam Processing. Academic Press, New York:1984. 59 J .F . Zeigler, Ion Implantation Science and technology, 2nd ed., Academic Press, New York (1988). 60 C C . T in, C K . Teh & F.L. Weichman, J . Appl. Phys, v62, 2329, (1987). 61 R.T. Chen, D.E. Holmes, & P.M. Asbeck, Appl. Phys. Lett., v45, 459, (1984). 62 J.V. DiLorenzo, A.S Jordan, A.R. Von Neida, & P. O'Conner, Semi-Insulating III-V Materials 1984, 309, (1984). 63 P.Dobrilla, J.S. Blakemore, A .J . McCamant, K.A. Gleason, & R.Y. Koy-ama, Appl. Phys. Lett., v47, 602, (1985). 64 S. Miyazawa & K. Wada, Appl. Phys. Lett., v48, 905, (1986). 176 65 T. Obokata, T. Katsumata & T. Fukuda, Jpn. J . Appl. Phys. Pt. 2, v24, L785, (1985). 66 K. Terashima, S. Washizuka, J . Nishio, A. Okada, S. Yasuami and M. Watanabe, Inst. Phys. Conf. Ser. No. 79, 37, (1986). 67 T. Sato, K. Terashima, H. Emori, S. Ozawa, M. Nakjima, T. Fukuda & K. Ishida, Jpn. J . Appl. Phys. Pt. 2, v24, L488, (1985). 68 H. Miyairi , T. Inada, T. Obokata, M. Nakajima, T. Katsumata, & T. Fukuda, Jpn. J . Appl. Phys. Pt. 2, v24, L729, (1985). 69 Y. Matsumoto & H. Watanabe, Jpn. J . Appl. Phys. Pt 2., v21, L515, (1982). 70 M.R. Brozel, I. Grant, R.M. Ware, & D.J. Stirland, Appl. Phys. Lett., v42, 610, (1983). 71 S.Miyazawa, T. Honda, Y. Ishii, & S. Ishida, Appl. Phys. Lett., v44, 410, (1984). 72 J . Kasahara, M. Arai , & N. Watanabe, Jpn. J . Appl. Phys. pt 2, v25, L85, (1986). 73 D.C. Look, P.W. Yu , W.M. Theis, W. Ford, G. Mathur, J.R. Sizelove, D.H. Lee & S.S. L i , Appl. Phys. Lett., v49,1083, (1986). 74 A . M . Huber, G. Morillot, N.T. L inh, P.N. Favennec, B. Deveaud, & B. Toulouse, Appl. Phys. Lett., v34, 858, (1979). 75 T . J . Magee, K. Shee, P. Ormond, R.J. Blattner, & C A . Evans, Appl. Phys. Lett., v37, 447, (1980). 76 L.B. Ta, H .M. Hobgood, A. Rohatgi & R.N. Thomas, J . Appl. Phys., v53, 5771,(1982). 77 H .M. Hobgood, L.B. Ta, A. Rohatgi, G.W. Eldridge, & R.N.Thomas, Semi-Insulating III-V Materials 1982, 28, (1982). 78 D.E. Holmes, H . Kuwamoto, C G . Kirkpatrick, & R.T. Chen, Semi-Insulating Material 1984, 204, (1984). 79 S. Martin, M. Duseaux & M. Erman, Inst. Phys. Conf. Ser. No. 74, 53, (1985). 80 T. Obokata, H. Okada, Y. Katsumata & T. Fukuda, Jpn. J . Appl. Phys. Pt. 2, v25,L179, (1986). 81 R.N. Thomas, I E E E 1986 GaAs IC Symposium, 29, (1986). 82 M. Yokogawa, S. Nishine, K. Matsumaoto, H . Morishita, H. Fujita & S. Akai , Inst. Phys. Conf. Ser. no. 74, 29, (1985). 83 M. Yokogawa,S. Nishine, M. Sasaki, K. Matsumoto, K. Fujita & S. Akai , Jpn. J . Appl . Phys. Pt. 2, v23, L339, (1984). 84 N. Noto, Y. Kitagawara, T. Takahashi & T. Takenaka, Jpn. J . Appl. Phys. pt 2, v25,L394, (1986). 85 K.V. Vaidyanathan, M.J. Helix, D.J. Wolford, B.G. Streetman, R.J.Blat-tner and C A . Evans, Jr., J . Electrochem. Soc, vl24,1781, (1977). 86 P.K Vasudev, R.G. Wilson & C A . Evans, Appl. Phys. Lett., v37, 309, (1980). 87 P.K Vasudev, R.G. Wilson & C A . Evans, Appl. Phys. Lett., v36, 837, (1980). 88 M.D. Deal & D.A. Stevenson, J . Appl. Phys., v59, 2398, (1986). 89 G.A. Kachurin, N.B. Pridachin & L.S. Smirnov, Phys. Conf. Ser. No 23, 461,(1975). 90 M. Arai , K. Nishiyame, and N. Watanabe, Jpn. J . Appl. Phys., v20, L I 24, (1981). 91 K.D. Cummings, S.J. Pearton, & G.P. Vella-Coleiro, J . Appl. Phys., v60, 163, (1986). 92 M.H. Badawi and J . Mun, Jpn. J . Appl.Phys. Pt 2, v20,125, (1984). 93 T. Hiramoto, Y. Mochizuki, T. Saito, & T. Ikoma, Jpn. J . Appl. Phys. Pt 2, v24, L921,(1985). 94 M. Kuzuhara, H. Kohzu & Y. Takayama, Appl. Phys. Lett., v41, 755, (1982). 95 A.K. Chin, I. Camlibel, R. Caruso, M.S.S. Young, & A.R. Von Neida, J . Appl. Phys., v57, 2203, (1985). 178 96 M. Dussac, M. Dupuy & E. Molva, Defect Recognition and Image Process-ing in III-V Compunds, ed. J.P. Fil lard, (Elsevier Science Publishers: Amsterdam), 209 (1985). 97 H. Kimura, A.T. Hunter, & H.M. Olsen, Inst. Phys. Conf. Ser. No. 79,1, (1986). 98 T. Sekiguchi & K. Sumino, Jpn. J . Appl. Phys. Pt. 2, v26, L I 79, (1987). 99 Y. Ishii, S. Miyazawa, & S. Ishida, I E E E Trans. Electron Devices, vED-31,800, (1984). 100 S. Miyazawa, Y. Ishii, S. Ishida & Y. Nanishi, Appl. Phys. lett, v43, 853, (1983). 101 Y. Sano, H. Nakmura, M. Akiyama, T. Egawa, T. Ishida, & K. Kamini-shi, Jpn. J . Appl. Phys. Pt. 2, v23, L290, (1984). 102 C A . Warwick, S.S. Gil l , P.J. Wright & A .G. Cullis, Inst. Phys. Conf. Ser. No. 76, 365, (1985). 103 M.L. Cone & R.L. Hengehold, J . Appl. Phys., v54, 6346, (1983). 104 D. Hui , PhD thesis, Dept of Electrical Engineering, University of British Columbia, (1989). 105 H.J. Queisser & C.S. Fuller, J . Appl. Phys, v37, 4895 (1966) 106 E. Fabre, phys. stat. sol., v9, 259 (1972). 107 J . van de Ven, W.J.A.M. Hartmann & L.J. Giling, J . Appl. Phys., v60, 3735 (1986). 108 Z.G. Wang, H.P. Gislason & B. Monemar, J . Appl. Phys., v58, 230 (1985). 109 G. Bentini, L. Correrra and C. Donolato, J . Appl. Phys., v56, 2922 (1984). 110 H.A. Lord, I E E E Trans. Semiconductor Manufacturing, v l , 105 (1988). 111 W.H . Press, B.P. Flannery, S.A. Teukolsky, & W. Vetterling, Numerical Recipes, Cambridge Unviersity Press, Cambridge ,p40 (1986). 112 M.R. Wilson, PhD thesis, Dept. of Electrical Engineering, University of Cincinnati, (1986) 179 113 Properties of Silicon, INSPEC, The Institute of Elec. Engineers, London, p35-36, (1988). 114 A.S. Jordan, J . Crystal Growth, v49, 631 (1980). 115 Y.S. Touloukian & D.P. DeWitt, Thermophysical Properties of Matter, v8. Thermal Radiative Properties. Nonmetallic Solids, IFI/Plenum, New York. 116 A.S. Jordan, J . Appl. Phys., v51, 2218, (1980). 117 R. Siegel & J.R. Howell, Thermal Radiation Heat Transfer, McGraw-Hil l , New York (1972) 118 S. Whittaker, Fundamental Principles of Heat Transfer, Pergamon Press, New York, 452, (1977). 119 R.C. Weast, editor, C R C Handbook of Chemistry and Physics, C R C Press, Florida (1982). 120 F. Kreith & W.Z. Black, Basic Heat Transfer, Harper & Row, New York (1980). 121 S.P. Timeshenko & J .N . Goodier, Theory of Elasticity, McGraw-Hill , New York (1974). 122 J.S. Blakemore, J . App. Phys., v53, R123 (1982). 123 D. Macquistan, Masters Thesis, Dept of Metals and Materials Engineer-ing, University of British Columbia, (1989). 124 A .G . Milnes, Advances in Electronics and Electron Physics, v61, Ed . P.W. Hawkes, Academic Press, New York, 64 (1983). 125 F.C. Frank, and D. Turnbull, Phys. Rev. vl04, 617, (1956). 126 D.L. Kendall, Semiconductors and Semimetals, v4, Eds R.K. Willardson & A .C . Beer, Academic Press, New York, 163 (1968). 127 R.N. Hal l , & J . H . Racette, J . Appl. Phys., v35, 379, (1964). 128 R.J. Borg & G.J. Dienes, A n Introduction to Solid State Diffision, Aca-demic Press, San Diego (1988). 129 W.C. Dash, J . Appl. Phys. v27,1193, (1956). 180 130 C C . T in, C K . Teh & F.L. Weichman, J . Appl. Phys, v63, 355, (1988). 131 D.V.Lang & R.A. Logan, J . Appl. Phys., v47,1533 (1976). 132 V. Kumar & L-A Ledebo, J . Appl. Phys., v52, 4866 (1981). 133 N. Kullendorff, L. Jansson & L-A Ledebo, J . Appl. Phys., v54, 3203 (1983). 134 B.G. Yacobi and D.B. Holt, Cathodoluminescence Microscopy of Inor-ganic Solids, Plenum Press, New York (1990). 135 M. Eckstein, A. Jakubowicz,M. Bode, & H.-U. Habermeier, Solid State Phenomena, v6 & 7 ,71, (1989). 136 C S . Fuller, J .D. Struthers, J.A. Ditzenberger, & K B . Wolfstirn. Phys. Rev., n93,1182, (1954). 137 J .C . Severiens & C S . Fuller, Phys. Rev, v92,1322 (1953). 138 F. Hasegawa, J . Appl. Phys., v45,1944 (1972). 139 K. Lohnert, W. Wettling & G. Koschek, Semi-Insulating III-V Materials 1986, 267, (1986). 140 J.P. Stark, Solid State Diffusion. John Wiley & Sons, New York (1976). 141 F.F. VoFkenshtein, The Electronic Theory of Catalysis of Semiconductors, Pergammon Press, Oxford (1963). 142 J . Ding, J.S.C. Chang & M. Bujatti, Applied Physics Letters, v50,1089, (1987). A P P E N D I X A L I S T O F A C R O N Y M S C L Cathodoluminescence C-V Capacitance-Voltage D F Dislocation Free D L T S Deep Level Transient Spectroscopy E B I C Electron Beam Induced Current E L 2 Electron Trap Level E P D Etch Pit Density F A Furnace Anneal F E T Field Effect Transistor GaAs Gall ium Arsenide H B Horizontal Bridgeman IC Integrated Circuit IR Infrared K O H Potassium Hydroxide L E C Liquid Encapsulated Czochralski L P E Liquid Phase Epitaxy L V M Local Vibrational Mode M B E Molecular Beam Epitaxy M E S F E T Metal Semiconductor F E T M O C V D Metal-Organic Chemical Vapour Deposition O T C S Optical Transient Current Spectroscopy P B N Pyrolytic Boron Nitride P L Photoluminescence R T A Rapid Thermal Annealing S E M Secondary Electron Microscope SI Semi-Insulating SIMS Secondary Ion Mass Spectroscopy S i 3 N 4 Silicon Nitride S i 0 2 Silicon Dioxide V L S I • Very Large Scale Integration v t h Threshold Voltage 182 APPENDIX B P R I N T O U T O F M A T H E M A T I C A L M O D E L P R O G R A M C R T A H E A T I N G P R O G R A M C I N T E G E R I N T E R , ISS, L I M 1 , L I M 2 , P , P T A G , N T A G , L T A G , L R E A L K , K 1 , K 2 , D X , D T , T B M A X , S I G -M A , P I N T , L I M , E P S 1 ,EPS2,C1 , C 2 , H I N T R E A L H C O N V , T A M B , E 2 R E A L * 8 TOLD(200),TNEW(200),A(200),B(200),C(200),R(200) C O M M O N D T , P , I T A G C C O P E N I N P U T A N D O U T P U T F I L E S C OPEN(unit=22 , F I L E = ' T I N P U T . D A T ' ) OPEN(unit=33, F I L E = ' T D A T A . D A T ' ) C C E N T E R M E S H S I Z E (DX), T I M E S T E P (DT), M A X . B U L B T E M P ( T B M A X ) , C M A X T I M E F O R H E A T I N G (LIM) , A N D O U T P U T I N T E R V A L (PINT) - T I M E C B E T W E E N P R I N T O U T S C R E A D (22,*) D X , D T , T B M A X , L I M , P I N T R E A D (22,*) H C O N V , T A M B C P = N I N T ( P I N T / D T ) L I M 1 =NINT(LIM/DT) D X 2 = D X * * 2 C C C A L C U L A T E T H E I N T E R F A C E N O D E ( INTER) A N D T H E S i S U R F A C E N O D E (ISS) C C I N T E R = NINT( .06/DX) + 1 ISS = N I N T U 0 5 / D X ) + 1 C C A S S I G N V A L U E S T O P H Y S I C A L C O N S T A N T S (1= V A L U E S F O R G a A s , C 2= V A L U E S F O R Si) C CP1=.383 RH01=5.32 C 1 = R H 0 1 * C P 1 C CP2=.832 RH02=2.32 C 2 = R H 0 2 * C P 2 C C A S S I G N V A L U E F O R E M I S S I V I T Y O F S i READ(22 , * )E1 READ(22 , * )E2 183 C A S S I G N V A L U E F O R S T E F A N - B O L T Z M A N N C O N S T A N T SIGMA=5.67E-12 C C WRITE(33,10) E 2 10 F O R M A T C E M I S S T V I T Y O F S i =\F10.7) WRITE(33,15) H C O N V 15 F O R M A T C C O N V E C T I O N C O E F F I C I E N T = ',F10.7) WRITE(33,20) D X 20 F O R M A T C M E S H S I Z E ( IN C M ) = \F10.7) WRITE(33,30) D T 30 F O R M A T C T I M E S T E P ( IN S E C ) =\F10.7) WRITE(33,40) T B M A X , T A M B 40 F O R M A T C T B M A X A N D T A M B = \2F10.3) C C I N I T I A L I Z E T E M P M A T R L X T O 400 D E G R E E S C (673 D E G R E E S K ) C D O 300 I=1,ISS TOLD(I)=673 300 C O N T I N U E C D O 350 1=1 J S S A(I)=0.0 B(I)=0.0 C(I)=0.0 350 C O N T I N U E C PTAG=0 C N U M = 0 N T A G = 0 C C B E G I N C A L C U L A T I O N C D O 2000 J=1,LIM1 C P T A G = P T A G +1 C C K=2271 . / T O L D ( l )**1.463 B(1)=1/DT + K/(2*C1*DX2) + H C O N V / ( 2 * C l * D X ) C(1)=-K/(2*C1*DX2) R ( l )=EPS1 ( T B M A X , T O L D ( l ) ) * S I G M A * ( T B M A X * * 4 - T O L D ( l )**4)/(Cl * D X ) -+(K/(2*Cl*DX2))*(TOLD(2)-TOLD(l) ) + T O L D ( l ) / D T -- H C O N V * T O L D ( l ) / ( 2 * C l * D X ) + H C O N V * T A M B / ( C l * D X ) C L I M 2 = I N T E R - 1 D O 600 I=2,LIM2 K=2271 . / T O L D ( I ) * * l .463 A(I)=-K/(2*C1*DX2) B(I)=1/DT + K/ (C1*DX2) C(I)=-K/(2*C1*DX2) R(I )=( l /DT-K/(Cl*DX2))*TOLD(I) + K*(TOLD(I+l)+ 184 -T0LD(I -1 ))/(2*Cl *DX2) 600 C O N T I N U E C HINT=1.57E-5*TOLD(INTER)+.00547+2.95E-12*TOLD(INTER)**3 C K I =2271 . / T O L D ( I N T E R ) * * l .463 K 2 = l 521 . / T O L D ( I N T E R ) * * l .226 A(INTER)=-K1/(2*C1 *DX2) B(INTER)=1/DT + (K1+HINT)/(2*C1*DX2) C(INTER)=-HINT/(2*C1 *DX2) R(INTER)=(1/DT-(K1 +HTNT)/(2*C1 * D X 2 ) ) * T O L D ( I N T E R ) + - K 1 * T 0 L D ( I N T E R - 1 ) / ( 2 * C 1 * D X 2 ) + H I N T * T 0 L D ( I N T E R + 1 ) / ( 2 * C 1 * D X 2 ) A( INTER+1 )=-HINT/(2*C2*DX2) B(INTER+1)=1/DT + (K2+HINT)/(2*C2*DX2) C(INTER+1)=-K2/(2*C2*DX2) R(INTER+1 )=( l /DT-(K2+HINT)/(2*C2*DX2))*TOLD(INTER+l) + -K2*TOLD(INTER+2) / (2*C2*DX2) + H I N T * T O L D ( I N T E R ) / ( 2 * C 2 * D X 2 ) C L I M 2 = I N T E R + 2 LIM3=ISS-1 D O 800 I = L I M 2 , L I M 3 K = l 521 . / T O L D ( I ) * * l .226 A(I)=-K/(2*C2*DX2) B(I)=1/DT + K/ (C2*DX2) C(I)=-K/(2*C2*DX2) R(I)=(l /DT-K/(C2*DX2))*TOLD(I) + K*(TOLD(I+l )+ - T O L D ( I - l ))/(2*C2*DX2) 800 C O N T I N U E C C K = l 521 . / T O L D ( I S S ) * * l .226 A(ISS)=-K/(2*C2*DX2) B(ISS)=1/DT + K/(2*C2*DX2) + H C O N V / ( 2 * C 2 * D X ) R( ISS)=E2*SIGMA*(TBMAX**4-TOLD(ISS) * *4 ) / (C2*DX) -+K*(TOLD(ISS-l ) -TOLD(ISS)) / (2*C2*DX2) +TOLD(ISS)/DT -- H C O N V * T O L D ( I S S ) / ( 2 * C 2 * D X ) + H C O N V * T A M B / ( C 2 * D X ) C C S O L V E F O R T N E W C C A L L T R I D A G ( A , B , C , R , T N E W , I S S ) C C C H E C K F O R O U T P U T C O N D I T I O N C IF ( P T A G . E Q . P ) C A L L O U T P U T ( T N E W , I S S , P T A G , N T A G ) C C C H A S T I M E L I M I T B E E N R E A C H E D ? C IF ( ( J *DT) .GE.LIM) G O T O 3000 C C U P D A T E T N E W C C C A L L U P D A T E ( T O L D , T N E W , I S S ) C C 2000 C O N T I N U E C 3000 C O N T I N U E C C A L L O U T P U T ( T N E W , I S S , P T A G , N T A G ) C C L O S E O U T P U T F I L E C CLOSE(22 ) CLOSE(33) C S T O P E N D C C F U N C T I O N E P S K T 1 . T 2 ) R E A L E P S 1 ,F2,T1 , L A M 1 , L A M 2 R E A L * 8 T2 C EG=1.52-(5.4E-4 * T2**2)/(T2 + 204.) L A M 1 = 1 . 2 4 / E G LAM2=3.0 C EPS1=.66*F2(T1,LAM1) + .1 *(F2(TL ,LAM2)-F2(T1 ,LAM1)) C R E T U R N E N D C C F U N C T I O N F 2 ( T , L A M ) R E A L V , L A M , F 2 , T C F2=0.0 V=14388.0/(LAM*T) D O 10 J= l ,3 F2=F2 +.1539897*(EXP(-J*V)*((( J*V+3)*J*V+6)*J*V+6))/J**4 10 C O N T I N U E R E T U R N E N D C S U B R O U T I N E T R I D A G ( A , B , C , R , T , L I M ) R E A L * 8 A(200),B(200),C(200),R(200),T(200),GAM(200) R E A L * 8 B E T I N T E G E R L I M IF (B(l).EQ.O.O) WRITE(* ,5) 5 F O R M A T C E R R O R : E L E M E N T B ( l ) E Q U A L S Z E R O ' ) BET=B(1) T(1)=R(1)/BET D O 20 J = 2 , L I M 186 GAM(J)=C(J-1)/BET BET=B(J ) - A ( J ) * G A M ( J ) IF (B(J).EQ.O.) WRITE(* ,10) J 10 F O R M A T C E R R O R : E L E M E N T N U M B E R ',15, * O F B E Q U A L S ZERO') T( J)=(R( J)-A( J)*T( J-1 ))/BET 20 C O N T I N U E D O 30 J=LIM-1,1,-1 T( J)=T( J ) - G A M ( J+l )*T( J+l) 30 C O N T I N U E R E T U R N E N D C S U B R O U T I N E O U T P U T ( T , L I M I T , M T , N U M ) R E A L * 8 T(200) R E A L D T , P T I N T E G E R L I M I T , M T , N U M , P C O M M O N D T , P , I T A G P T = N U M * P * D T + M T * D T WRITE(33,5) P T 5 F O R M A T C T H I S IS T H E O U T P U T A T T I M E = ',F10.7,' ( in sees)') N=INT(LIMIT/3) D O 10 1=1 , N WRITE(33,* ) T ( 2 * N + I), T ( N + I), T(I) 10 C O N T I N U E IF ( N . E Q . L I M I T ) G O T O 20 L3=N*3 +1 D O 15 J = L 3 , L I M I T WRITE(33,*)T(J) 15 C O N T I N U E 20 MT=0 WRITE(33,* ) I T A G N U M = N U M +1 R E T U R N E N D C S U B R O U T I N E U P D A T E ( T 1 , T 2 , L I M I T ) R E A L * 8 T1(200),T2(200) I N T E G E R L I M I T IF (LTMIT.EQ.O) G O T O 50 D O 20 1=1,LIMIT T1(I)=T2(I) 20 C O N T I N U E 50 C O N T I N U E R E T U R N E N D 187 APPENDIX C Calculation of Thermal Expansion Coefficient, Young's Modulus and Poisson's Ratio The expression for the thermal expansion coefficient is (from reference 114) a = 4.68x10"* +3.82* l(T*r (XT1) (C.I) where T is absolute temperature. At 1000 K a = 8.5 x 10"6 K"1. The temperature dependence of the elastic coefficients is given by, 1 1 4 cn = 12.16* 1011 - 1.39xl08r (dynes/cm2) (C.2) c1 2 = 5.43* 1011 - 5.76* 10V (dynes/cm2) (C.3) C 4 4 = 6.18X10u-7.01A;107T (dynes/cm2) (C.4) At 1000 K, c n =10.77x10 1 1 dynes/cm2, c 1 2 = 4.854x10 n dynes/cm2 and c 4 4 = 5.479x101 1 dynes/cm2. Young's Modulus for the (100) orientation can be calculated from the elastic coefficients using the following expression, 1 2 2 (c11 + 2c 1 2)(c 1 1-c 1 2) 2 (C.5) E = (dynes I cm ) (Cn + C 1 2) at 1000 K, E=7.7 x 10 1 1 dynes/cm2 = 7.7 x 10 3 MPa. Poisson's ratio for the (100) orientation is given by, 1 2 2 cl2 (C.6) v = C l l + C12 at 1000 K, v=.3106 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0078584/manifest

Comment

Related Items