I I I I . |~ ’1 _ . '1'. ' ‘1 A ‘1. i 3. 1' . - laid .. ' 2‘ “i 11% I... ‘ ': 9‘ )1 ‘ aphil ':.1;:-‘; l ‘ “ I. i: :- QIQ‘}_, ._I 3 ‘ . ~11 " ’ ‘ ' . . .3 i II ‘ 1‘1 ‘-:. ' l o . Li I. , .: 5‘ ‘ 1:: i 13.; 1 ':-.° 1 I T. ‘ t. o fid‘ I '... . 1111 ’ O V ' :I‘. ,' ‘_ 3' I ‘ ml. . ~‘ 3.51 . ‘u‘ atffigfi; 1 L ‘1! :._';; ‘ 1:15.:- 1 I" I I ' 6% :3 1' 1 E'c"::";l.'H ‘59“ .- ‘01:?“ 1.913111: '1' 12:36 “III 1 II‘I; 5115‘sH IIIHYII IIU YIII! ‘ MI 1111 11 - {15:1 $111.11qu .I1I11‘tIIIY11IIl 1 ..,.1 .. I'II ‘II11II1IIIIIIII’I1II1IIIIIII‘1II1 III? IIIIIIm III!ls III I?“ v _ n W -u...~ - .4 _. _..r— ....- ,w .. .. ”q— ”. , a . -——<: I .. '- ' ”-1 :o— r - - ~— '- w w H . .— w -' . - .. M'- - cow —— M .. “— .. a...” — cur "‘ - ”m . ”II ”I II 1 ”I 1“]11VIV‘IEI' I'M LII; 1I1[.1 "I- II 1 IF'MI I11 . ’_ - __.._....~.. . a 't n a W _ . M... u. -A v «.“o— F. - o no I. 1. , 11 5511",» 1 1‘ . 111I ‘1 111111.111 111 .1 \"I 'v‘I .II'I- .Y '1. 1111: 1‘11. II“ 1II1‘ 1 m ”III 'h I IEIL‘I (it; ,1': 'YIII" [LIIO 11‘ “1 5%? III II :I‘ 131113 1.1,! IWI Kim I; IITYHIII‘III VII, I‘I‘II IIIIIII "v— I . . . . . 1.... . .— . "fimz W . *‘z . a . - . N v. . ...._._.. . ,-- _.. . ”.2 . . .-. . - -. I - . . . .— ' “V. . ”- w “to“. . .- W mohl . ’ . - c . . . I o W .M M c u v y. no crow-o“ M *N ..=- "a“. .c-o‘ .— u w .A- .—. cu~c-‘ 0 . “a”... nu— -.- a. v ....... 1‘1 31“,}.1. .(J‘iI [1 “‘SI II 11 INI‘IIII I111 —. M ‘D W. M ‘QOP‘ -.- 95“ ' a . - .MQ—c— .mer u I. ’v “."flm ,. , .. u .m; 0‘ ~,. 1 new. «.- u. ”cow. , m... . . 4,.— N 1 . . ‘ - w! '7' uuuuu {11111 1Y5? I!" ‘3’ .III’II ' I 5"} SJ“ .— _ ~’.. _.. m Jc-o-d. . .. 0 ‘-av “v...- n a... 1- “I511” “215' “1‘ 1.. 1: ’ r1111111111' {11“15Y1Y1Y11111Y11a1:111“ I III 11 11 1L“. 3.1“ IUI‘HIII 1111 11111" I111“; "31”” £21:th 3‘” u . ‘IIFIIIIIIIIII‘II'II . . . . r. O. :21... W- . - . . . . n t-us v c. ““ '3 .._I WWII 11111111111. Date 0-7639 5-]4-86 *3L;.L*a£15**';i.‘l I I 9 ‘1- .. _ ‘1‘ 2 * __“F f ‘ 4 1.; ii. '5': ‘8 S I It. Or ‘ 1,1“?a'v‘af-in . -_ «a .14; __o “ ‘i' l-' a o O _ If .'1'\‘ -.:MV£‘ 525.7 This is to certify that the thesis entitled X-Ray Fluorescence Spectroscopy of CdTe Substrates presented by Daniel R. Hines has been accepted towards fulfillment of the requirements for Masters degree in Physics ‘ or professor MS U is an Affirmative Action/Equal Opportunity Institution MSU LIBRARIES ‘— RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. APR 2 o my; § ' liik X-RAY FLUORESCENCE SPECTROSCOPY OF CdTe SUBSTRATES BY Daniel Ray Hines A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Physics and Astronomy 1985 70570.9 3 ABSTRACT X-RAY FLUORESCENCE SPECTROSCOPY OF CdTe SUBSTRATES BY Daniel Ray Hines CdTe is a commonly used substrate material for epitaxial growth of Hg1_x Cdx Te which has gained tremendous attention in the infrared detector industry. It has been demonstrated that the properties of the Hg1_x Cdx Te epilayer depends on defects and impurities present in the substrate. Therefore, the study of these: defects and impurities in CdTe substrates is essential. We have performed x-ray fluorescence spectroscopy in conjunction with electron microscopy on CdTe substrates in order tc>study micro- chemical inhomogeneities. The sample preparation necessary for electron microscopy has been developed. Six CdTe substrates obtained through the Rockwell International Science Center were prepared and studied for stoichiometry and impurities. Te deposits and impurity deposits (especially Fe) are reported in addition to stoichiometric variations as large as 20%. ACKNOWLEDGEMENTS I would like to thank Dr. Jerry Cowen for his continuous encouragement auuitnotivation throughout this research. His enthusiasm and guidance kept me going through many discouraging times. Dr. Cowan's constructive criticism and spelling input is also sincerely appreciated. I wish to thank all the people at the Rockwell International Science Center especially Dr. Paul Newman, Dr. J. Bajaj, and Eun-Hee Cirlin for their support throughout my visit to Thousand Oaks, California. I would like to emnwms my gratitude to the professors of the solid state group and the support staff, especially Vivian Shull, for their generous advice and assistance. Last, but certainly not least, I would like to thank the department secretaries and my fellow graduate students (especially Sami Mahmood) for their friendship and support throughout my graduate career. The financial support and CdTe sample received from The Rockwell International Science Center are gratefully acknowledged. II. III. IV. TABLE OF CONTENTS INTRODUCTION History of Infrared Detector Materials CdTe Crystal Growth Hg _ Cd Te Epilayer Growth Epiléyer§Substrate Interface CdTe Substrates Studied at Michigan State University X-RAY FLUORESCENCE SPECTROSCOPY Electron Microscopy X-ray Production and Detection X-ray Data Corrections SAMPLE PREPARATION Introduction CdTe Top-Surface Preparation and Etching CdTe Back-Side Preparation and Etching Jet Thinning RESULTS Data Acquisition CdTe 3-268 CdTe 2-36 CdTe A—2 CdTe 19N9 CdTe 3531 CdTe: Cu 2303 CONCLUSION Analysis of CdTe substrates Future Work LIST OF REFERENCES APPENDIX A CdTe Electron Microscope Sample Preparation Step-by~Step Procedure Page \OOU'IUN-‘d 11 ll 16 19 28 28 32 35 35 37 37 H1 117 5A 5A 5A 67 7A 75 77 77 LIST OF TABLES Raw Data Correction Factors for Ka x-ray Lines Parameters Characterizing the Six CdTe Sam pl es Page 25 71 Figure 10. 11. 12. 13. 14. 15. LIST OF FIGURES Energy band gap versus maximum detector wave length. Included points represent a number of common materials. (From Stevens, Ref. 3) Hg _ Cd Te energy band gap as a function of Cd concentration (From Kruse, Ref. 6) Composition profile of epitaxial Hg _ Cd Te on a CdTe substrate as a function of deptR (From Micklethwaite, Ref. 9) Schematic view of various signal detectors used in analytical electron microscopy Console and column of the V0 H8501 Scanning Transmission Electron Microscope Schematic View of the VG H8501 Scanning Transmission Electron Microscope column Sample area cross-section of the V0 H8501 Scanning Transmission Electron Microscope Energy Level scheme illustrating x-ray production. Cross-sectional view of Links detector system (not to scale) CdTe 2-36 x-ray spectrum at zero tilt angles CdTe 2-36 x-ray spectrum at optimum tilt angles X-ray production volume for thin film and bulk samples (From Williams and Edington, Ref. 22) CdTe: Cu 2303 x-ray spectrum enumerating the x-ray lines of Cd and Te. Beam-sample detector geometry CdTe 2-36 x-ray spectrum showing 8r contamination Page 12 13 114 15 17 18 20 21 22 2M 27 3O Figure 16. 17. 18. 19. 20. 21. 22. 23. 2“. 25. 26. 27. 28. 29. 30. 31. 32. 33. 3A. 35. Top-surface of CdTe disc with center perforation Spinner for photoresist application CdTe piece after top-surface etching CdTe: Cu 2303 x-ray spectrum showing K a peak integrated counts CdTe 2-36 x-ray spectrum showing region of interest (ROI) definitions Example of a x-ray map for CdTe: Cu 2303 (Cd map on the left and Te map on the right) X-ray map of Te deposit in CdTe 3~268 (Cd map on the left and Te map on the right) CdTe 3-268 x-ray spectrum of Te deposit shown in Figure 22 X-ray map of Te deposit in CdTe 3-268 (Cd map on the left and Te map on the right) CdTe 3-268 x-ray spectrum of Te deposit shown in Figure 211 CdTe 3-268 x-ray spectrum of region away from Te deposits CdTe 2-36 x-ray spectrum of a typical region X-ray map of Fe deposit in CdTe “-2 (Cd map on the left and Fe map on the right) CdTe ”-2 x-ray spectrum of Fe deposit shown in Figure 28 X-ray map of Zr deposit in CdTe A-2 (Cd map on the left and Zr map on the right) CdTe 4-2 x-ray spectrum of Zr deposit shown in Figure 30 CdTe A-2 x-ray spectrum of region away from impurity deposits CdTe 19A9 x-ray spectrum of a typical region CdTe 3531 x-ray spectrum of a typical region X-ray map of Cu deposit in CdTe: Cu 2303 Page 31 33 3A 38 39 H0 H2 “3 AM 45 H6 A8 M9 50 51 52 53 55 56 57 36. 37. 38. 39. M0. M1. N2. “3. 1m. 1:5. A6. 147. ’48. 49. (Cd map on the left and Cu map on the right) CdTe: Cu 2303 x-ray spectrum of Cu deposit shown in Figure 35 x-ray map of Fe deposit in CdTe: Cu 2303 (Cd map on the left and Fe map on the right) CdTe: Cu 2303 x-ray spectrum of Fe deposit shown in Figure 37 X-ray map of Fe deposit in CdTe: Cu 2303 (Cd map on the left and Fe deposit on the right) CdTe: Cu 2303 x-ray spectrum of Fe deposit shown in Figure 39 CdTe: Cu 2303 x-ray spectrum of Ti trace CdTe: Cu 2303 x-ray spectrum of Al trace CdTe: Cu 2303 x-ray spectrum of region away from impurity deposits SEM micrograph of etched edge (magnification - 10,000x) SEM micrograph of cleaved edge (magnification - 10,000x) X-ray detector efficency curves CdTe piece with photoresist protective layer prior to semistatic backside etching Teflon basket used to hold CdTe piece during semistatic backside etching CdTe disc mounted with Apiezon M grease on holder for South Bay model 550C jet thinner 58 59 6O 61 62 6M 65 66 68 69 72 8O 82 8A INTRODUCTION History of Infrared Detector Materials CdTe is a commonly used substrate material for epitaxial growth of Hg1_x Cdx Te which has gained tremendous attention in the infrared detector industry. It has been demonstrated that the properties of H81-x Cdx Te epilayers depend on defects and 1’2 Therefore, the study of impurities present in the substrate. these defects and impurities in CdTe substrates is essential. Let us put into perspective the role of II-VI semiconductors, namely CdTe and Hg1_x Cdx Te, by first presenting a brief history of infrared detector materials. Experimentalists in the 1800's first studied infrared radiation using thermocouples.3 Temperature measurements were recorded as a current change resulting from heating or cooling of a circuit Junction made from two dissimilar metals. As materials research developed it was discovered that a number of intrinsic band gap semiconductors could also be used to study infrared radiation” Ir: general, a semiconductor of energy band gap, EE’ is related to the longest wavelength, Amax’ that can be detected by E=° (1 where h is Planck's constant and co is the speed of light in a vacuum.’4 For an intrinsic semiconductor, incident radiation of wavelength Amax or less excites electrons from states near the top of the valence band into states near the bottom of the conduction band. Such excitations change the properties (i.e. electrical conductivity or photovoltage) of the material“. These changes can be recorded to measure the incident radiation. The first intrinsic band gap materials used for infrared detector devices were elemental and compound semiconductors. These semiconductors have discrete band gaps and are well established in the 1-8 pm wavelength range as shown in Figure 1. The need soon arose, however, for infrared detectors that could be "tuned" to specific cut-off wavelengths and also cover the 8-12 pm wavelength range. It is possible to achieve such band gap tunability with ternary compounds where the concentration of the cations can be varied. One of the most promising ternary semiconductor materials for intrinsic infrared detectors is Hg1_x Cdx Te which has an energy band gap that varies with concentration, x, as shown in Figure 2. In terms of fabricating electro-optical device arrays, epitaxially grown Hg1__x Cdx Te has shown certain advantages over bulk grown H81-x Cdx Te. For instance, some geometries available with epilayer devices can avoid surface contamination problems inherent to bulk devicess. Also, for large scale production, small epilayers can be grown quite easily while small bulk pieces must be cut from large boules which is a slow process. However, there is Energy band gap (eV) 1.0 0.9 _ 0.8 __ 0.7 __ 0.6 .— 0.5 .— 0.4 __ 003 _ 0.2 F" 0.1 p_. 0 1 2 3 4 5 6 7 8 9 10 Wavelength (um) Figure 1. Energy band gap versus maximum detector wavelength. Included points represent a number of common materials. (From Stevens, Ref. 3) Energy band gap (eV) .4 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 HgTe CdTe Cd concentration (x) Figure 2. Hg1_xCdee Energy band gap as a function of Cd concentration (From Kruse, Ref. 6) much debate concerning what substrate produces the best epilayer. At the present time CdTe is one of the most widely used because of the rather small (0.3% for x = 0.2) lattice mismatch between the (111) crystal orientation of CdTe and HgO 8 Cd Te. Also, CdTe 0.2 substrates can be grown in single crystal form and are transparent to infrared radiation (E8 = 1.6 eV) which is important for some device geometries. CdTe Crystal Growth7 CdTe single crystals are grown by the Bridgman method. Equal atomic amounts of Cd and Te are loaded into a sealed ampoule which is lowered through a vertical furnace. A temperature gradient is set-up to ensure directional crystalization of the CdTe at the bottom of the ampoule. The upper portion of the growth chamber is held at the higher temperature to prevent sublimation and vapor growth at the top of the ampoule. The growth shape of the liquid- solid interface is determined by the temperature isotherms in the ampoule. Hg x Cdx Te Epilayer Growth 1- H81-x Cdx Te epilayers are most often grown on (111) oriented single crystal CdTe substrates by liquid phase epitaxy (LPE). Other epitaxial growth processes such as molecular beam epitaxy (MBE), vapor phase epitaxy (VPE), and organometallic vapor phase epitaxy (OMVPE) are in the experimental stages but at present are too expensive to be used for large scale production.8 An epilayer is defined as a smooth, continuous single crystal film grown on a substrate such that the film crystal structure corresponds to and is determined by that of the surface of the single crystal substrate.” A melt of Hg1_x Cdx Te is formed just above the liquidus temperature at one end of a closed quartz ampoule. At the other end a CdTe substrate is mounted and externally cooled. Once the ampoule reaches thermal equilibrium, it is inverted to allow the Hg1_x Cdx Te melt to flow onto the substrate and crystalize. After the epilayer is obtained, the ampoule is inverted back to its original position to decant the remaining H81-x Cdx Te.9 Epilayer-Substrate Interaction During LPE growth, there is substantial interaction at the epilayer-substrate interface. Not only is there a mixing of Hg1_x Cdx Te and CdTe (see Figure 3) but also any defects or impurities in the first 10 to 20 um of the substrate should be expected to affect the epilayer. It is these defects or impurities in the substrate that are believed to degrade infrared detector devices and hence provide the motivation for this project. Defects or impurities which propagate into the epilayer appear as electronic levels in the energy band gap causing excessive noise and shorter carrier lifetimes in a device.10 Composition (Z CdTe) 100 90 80 70 60 50 40 30 20 10 X Substrate K 0/0 CdTe — h— Epilayer 10 0 10 20 30 40 50 60 70 80 90 100 Distance (um) Figure 3. Composition profile of epitaxial Hg1_xCdee on CdTe substrate as a function of depth. (From.Micklethwaite,Ref. 9) These electronic levels in the energy band gap were studied by groups at Grenoble, Rockwell International, California Institute of Technology, Honeywell, Westinghouse, and Hughes Aircraft using both optical and electronic techniques. Molva et. al.11’12’13 characterized impurities (especially Cu and Ag) in CdTe using low temperature (AK) photoluminescence studies of exciton recombinations. Bajaj et. al.1u, using the same techniques at 77K have characterized CdTe substrates which are both "good" and "bad" (in terms of the quality of LPE H31-x Cdx Te grown on them) by studying the broad donor-acceptor transitions. S.H. Shin et. al.1_5 saw Te deposits in CdTe substrates using Auger Spectroscopy. Such deposits are detrimental to CdTe crystallinity which affects Hg1_x Cdx Te epilayer quality. Hunter et. al.16 also used jphotoluminescence of Hg1_x Cdx Te to show how surface concentration gradients can be mapped. Polla et. al.?7’18 studied deep level transient spectroscopy of H81-x Cdx Te to characterize defect energy levels relative to the band gap. Scanning electron microscopy of defects at the epilayer- substrate interface have been studied by James et. al.19 Also, Wood et. at.20 used transmission electron microscopy to characterize epilayer-substrate interfaces for both defects and chemistry. Wood used x-ray fluorescence spectroscopy to study Hg concentrations in the interface region of a HgO 9 Cdo Te epilayer on CdTe. 2 Despite all the work that has been done related to epitaxial detector devices, to date it is not clearly understood exactly how a substrate interacts with an epilayer to degrade a device. Fume experimental work is needed to investigate the chemistry of substrates and the defect propagation at the epilayer-substrate interface. CdTe Substrate Studies at Michigan State University Because very little is known about the micro-chemistry of CdTe substrates used for epilayer growth, we performed x-ray fluorescence spectroscopy in conjunction with electron microscopy. The x-ray fluorescence spectroscopy available in a scanning transmission electron microscope is a powerful analytic tool for studying micro- chemical inhomogeneities on a scale of several hundred angstroms for impurity concentrations as small as 0.05 atomic percent. Special sample preparation is necessary for our electron microscopy studies and has been developed in collaboration with Eun- Hee Cirlirrafl: the Rockwell International Science Center. A sample preparation technique has been developed to produce self-supporting 3mm discs cm CdTe having a microscopic perforation with edges 10003 or less in thickness. Rockwell International has provided a random selection of CdTe substrates - three Rockwell grown samples CdTe 3-268, CdTe 2-36, and CdTe ‘1-2; and three II-VI Inc. grown samples CdTe 19119 (a "good" substrate), CdTe 3531 (a "bad" substrate), and CdTe: Cu 2303 (doped 10 in the melt with ~1016 Cu). We prepared and studied these samples for stoichiometry and impurities. 11 II. X-RAY FLUORESCENCE SPECTROSCOPY Electron Microscopy In an electron microscope, when beam electrons interact with an electron transparent sample, information concerning chemistry, crystallinity, thickness, and topology is generated. The processing of this information can be accomplished by properly positioning the appropriate detector system along the beam axis as shown in Figure A. Signal generation and processing will be discussed for x- ray fluorescence spectroscopy as generated in a VG H8501 Scanning Transmission Electron Microscope (see Figure 5). This microscope is equipped with a high resolution, field emission elemtron gun to provide a 100 Kev electron beam. A double condenser lens system is used for increased probe forming flexibility along with a single, high excitation objective lens. The microscope is also equipped with a cold stage which cools the sample area to -136°C and a tilt stage with orthogonal angle ranges of 0x = i 60° and E)y = i 60°. The x-ray micro-analysis system is a Links Si(Li) detector interfaced to a Tracor-Northern 2000 computer. These basic features of the microscope are illustrated in Figures 6 and 7. Electron Energy Loss | | Detector l BF ‘\ DF DF Bright and Dark Field Detectors X-ray Detector Specimen Back Scattered ' Electron Detector Secondary HI— -—? Electron , Detector A Primary Electron Beam Figure 4. Schematic view of various signal detectors used in analytical electron microscopy. 13 Figure 5. Console and column of the VG H8501 Scanning Transmission Electron Microscope. 14 ,_.._i_, ,_ - 7. _ _, . .- - . i- l..- “h. -7 1 _. - N .2. fi- . 1 i ELECTRON SPECTROMETER AND BRIGHT FIELD DETECTOR DIFFRAC'T ION SCREEN COLLECTOR APERTURE 3 _ ANNULAR DARK FIELD DETECTOR J “BLOCK Pm WITH SPECIMEN OUTGASSING FACILITY MICRODIFFRACI'ION AND BEAM TILT COIL ASSEMBLY -—-—- OBJECTIVE S'I'IGMATOR DIFF'RACI'ION APERTU'RE EV DOUBLE CONDENSER LENS svsm CONDENSER STIGMA’TOR AND GUN ‘: V ‘ ALIGNMENT COIL ASSEMBLY \ GUN ISOLATION VALVE HIGH BRIGHTNESS FIELD EMISSION GUN 3"" ..u- AD-J 1i“ Figure 6. Schematic view of VG H3501 Scanning Transmission Electron Microscope column. 15 ‘g’T’TTT Tilt cartridge Stage ‘3‘ Plate \ \ COld / Plate \\\\V~« Objective Lens "\ Pole Piece X—ray Detector __—_“-Snecimen a Figure 7. Sample area cross— section of VG H3501 Scanning Transmission Electron Microscope. 16 X-ray Production and Detection .In an electron microscope, beam electrons interact with atomic electrons via energy transfer. An atomic electron gains energy and is excited out of its ground state leaving a vacancy. This vacancy is filled by a higher energy level atomic electron which fluoresces in the x-ray energy range in order to shed its excess energy. X-rays produced from L shell, M shell, or N shell atomiczexiectrons which fill a vacancy in the K shell are referred to as Ka’ KB and KY x-rays respectively. In addition, x-rays produced from M shell or N shell atomic and electrons which fill a vacancy in the L shell are referred to as La and L x-rays respectively. Figure 8 illustrates B the excitations involved in x-ray production. The K0 and La x-rays typically have the highest production cross sections due to the nature of dipole interactions. Each atom has a different set of energy levels, hence a unique x-ray fluorescence spectrum. It is this uniqueness that makes x-ray fluorescence spectroscopy a powerful tool for chemical analysis. The x-rays generated by the beam-sample interaction are detected by a 30mm2 cross-sectional area Si(Li) detector system whicni consists of an 8 pm 8e window, 2003 gold layer, 10008 Si dead layer, and a 3mm Si active region as shown in Figure 9. The x-ray detector is liquid nitrogen cooled and mounted at a 13° angle off the zero tilt plane of the sample. During x-ray data acquisition, the sample is tilted toward the detector at the tilt setting of 17 N shell M shell F/LY L H/ B ,—’I. a rd i Y L shell K X-ra f’rfi' Y y KB f")' Atomic e— r” x f a. .r ‘ Y K shell e- beam Figure 8. Energy level scheme illustrating x-ray production. 18 Be window Si dead layer —*| 811m H 1000A|~— L F(—’ 3 mm 0 200 A Si active region Au layer Cross- section view of Links detector system. Figure 9. (not to scale) 19 0x = -20° and 0y a 20°. These tilt angles provide the highest (La/Kc) peak ratios for both Cd and Te x-ray lines as shown in figures 10 and 11. This insures that the system absorption is minimized and that the sample-detector geometry is maximized. X-ray Data Corrections When a 20 8 diameter electron beam impinges on a sample, the beam spreads rapidly in the fashion indicated in Figure 12. The sample volume in which the x-rays are produced depends strongly on the sample thickness. 'The integrated count rate of characteristic Ka or La x-rays from elemnm.A (or 8) which have been detected is a complicated function of sample parameters, detector efficency, and beam-sample- detector geometries. For thin film samples of thickness dt this function can be linearized to the expression shown in equation 2.21 IA dt = 6A kA 8A n CA dt (2 where IA + Kn or La integrated count rate of element A 6A + absorption correction of element A in the sample k + fluorescence parameter (k-factor) for x-ray line of element A e + detector efficency for x-ray line of element A n + number of electrons bombarding dt C + weight percent concentration of element A 20 sa&.av om..¢m oas.am ' I I O O O I. I. ll OIQLPQ. E‘fll' .moncm uafiu onus um asuuooam Smutx onlw oHvo .oH ouswwm so..vN .... «.3. OD. @900 >mx yummzm 2:4: 52: 8...: 8.... Sum “0 0000 o O o co 0‘ o ’0 00’! c on. JO. 8 0‘4. (0‘ 0 CW 0 O '"roo . O “or. fi.’00|"$.~ a H00 C at o m 0 O U w H.000 O N UGO °° momm ssaflfiq UODZE-‘(D 21 (”In '90 m. ("£330. ammov P >mx rummzm sas.sm .Lwc a _ b.L .mufiwcm uawu Baaaumo um Esuuoomm mmutx menu .HH ouswfim I I 11 J 1111'! 11111 smmov ass-a 4. rte PL i o. N 0 od'.““ . team 0 00 ISGGH . team” m o IQQSN D U a E laamN UODZE-‘(D 22 20 A Thinfilm sample 0 O 100 A 21000 A _____ _..._._.\..._._._._..... ..._._ \ \ \ Absorption \\ event \ \ \ \ 1 Bulk sample \ 1 um :3 l l l I I I \‘\__ __,”I / Figure 12. X-ray production for thin film and bulk samples. (From Williams and Edington, Ref. 22) 23 In using equation 2 to calibrate the integrated count rate of an x-ray line from an element of unknown weight percent concentration, a "standard" x-ray line from a known weight percent concentration must be included in the sample and appear in each spectrum. This is not a trivial task. However, a standardless technique can be used to determine a weight percent concentration 21 ratio between two elements in the same sample. The equation for this standardless technique is HIH (DJ:- (3 I'D > X CD > 00 > 0' O > 03 Furthermore, for thin film samples (thickness ~10003) it will be shown that absorption corrections associated with the standardless technique (GA/68) are at most a few percent and can be neglected. Thus for standardless, thin film samples the weight percent concentration ratio for two elements is O x on m w H (11 0I w > X P m 3:. H 00 1:» All weight percent concentration ratios will be done using K0: x-ray lines because of their superior resolution in comparison with the La x-ray lines. This resolution is important for the Cd (Aall8) and Te (Au-52) lines as shown in Figure 13. The information necessary for k-factor and detector efficency calculations of Cd, Te and the impurities found in the CdTe samples has been published by 23 Zaluzec and is listed in Table 1. .09 com to you moCHH Smulx wcfiumuoE3co sauuooam Smuux momm so "meow .m~ ocsxwm >mz xoxmzm . soa.em 332.3m ama.v~ ass.am som.sa ana.aa aom.e ass ..FLC.__£FIE§%P_pp_LPLAL__EA A 0000' 0000 I 0* 00." a .o N mu . 4...... m "w 0 . m o ...\w . M BQM o H m. P 5 § “.0 NIH— M . s c. o .. . o . “ ox ... _ _ . _ m. n B. ... a _ . o _ . .. A Q . . U _ .o _ a A 4 . 2 _ _ _ . . + . m _ .H. _ . _ o . A 4. _ . . . + _ _ _ _ _ *a . . I o a '3 I 19 I) in 0232920 U 25 Table 1. Raw data correction factors for Ka x-ray lines Element kA-factor EA ;§E_;EQ A A Cd 0.3685 0.8A356 --- Te 0.2697 0.67661 1.7 Fe 2.0890 0.98032 0.15 Cu 1.7H93 0.98933 0.18 Zr 0.6717 0.99087 0.”? A1 ”.8280 0.70u18 0.10 Ti 2.473A 0.95053 0.13 26 Let us now consider absorption effects. When an atomic electrwnn fluoresces, the generated x-ray can be absorbed by another atomic electron elsewhere in the sample, hence decreasing the measured intensity. This effect is dependent on the beam-sample- detector geometry, sample topology, and sample thickness (see Figure 111). The microscopic sample topology is not known exactly. The best that can be done is to assume that a uniform wedge shaped edge results fauna the jet thinning process. It is also not possible to determine the exact thickness of an edge. However, if an edge is thin enough to give a good bright field image, the thickness is on the order of 1000 A or less. Absorption calculations for the standardless technique determined for Cd and Te K0: lines, with the additional assumption that CCd + CTe = 1, show corrections of approximately 1% for a CdTe sample thickness of 10 um. Hence absorption terms for Cd and Te in standardless, thin film CdTe samples are negligible and equation A can indeed be employed. X-ray detector Figure 14. 27 4 e- beam Beam- sample— detector geometry. Sample Zero tilt plane 28 III. SAMPLE PREPARATION Introduction In electron microscopy, sample preparation is by no means a trivial procedure. The desired end result is a region of the sample less then 10008 in thickness which is physically and chemically equivalent to the bulk material from which the sample was prepared. Frw this reason, it is important to develop special handling and etching techniques which do not contaminate or alter the microscopic characteristics of the sample in the region which is to be studied. Typically semiconductor, single crystal materials are soft, yet brittle, in comparison with metals; thus, classical polishing techniques tend to create and propagate defects. This needs to be avoided. Also, when dealing with binary and ternary compounds, preferential etchinggmmst be gguarded against. For these reasons, static and semistatic chemical etching techniques have been developed for use with semiconductor materials. For CdTe and Hg1_x Cdx Te it has been determined that bromine in methanol (PHHMi) or hydrobromic acid (H8r) are the most effective chemical etchants. The bromine bonds with the cation to form a cation-8r salt which is carried away along with the Te by the methonal (or hydrobromic acid). The concentration for the maximum etching rate is about 11% 8r iiltnethonal (mx Nomuzu euaam UQ 003290) 31 1 mm |———————l Figure 16. Top surface of CdTe disc with center perforation. 32 CdTe Top-Surface Preparation and Etching Starting with a 600 pm thick bulk piece of CdTe,eniinitial etch is performed to remove ~25 pm from the top surface. This step is also done prior to LPE growth as it removes any mechanical surface damage, leaving a clean CdTe surface. A photoresist masking technique is then used to obtain the proper disc geometry. Shipley 1375 photoresist has been chosen because it has good adhesive properties and is insoluble in both HBr and water. 1375 photoresist is.a.light sensitive, viscous liquid which undergoes a positive photographic process when exposed to ultraviolet light. A uniform thin layer, which is necessary for good results, can be applied with an eyedropper while spinning the CdTe sample at 3000 rpm. We have designed a simple spinner system which utilizes a scotch tape sample [nounting technique (see Figure 17). The photoresist mask itself is a thin piece of clear plastic on which is mounted an array of 3mm discs which are opaque to ultraviolet light. Note that the array side of the mask is referred to as the top side. (Once exposed and developed, a well shaped 3mm photoresist disc will be left on the top surface of the CdTe sample to protect a circular region from the etchant. A Br/HBr solution is used to etch a 150 um plateau into the CdTe as shown in Figure 18. This plateau becomes the desired disc after the back side is etched away. 33 7///// //////////////// / V/[Z/ CH /////fl uble sticky tape 0 D 3000 rpm motor 7/////// / /////////////////// ist application. photores for 17. Spinner Figure 34 / Top surface —.‘_ Plateau 150 pm 600 um CdTe \ 'Backside Figure 18. CdTe after top surface etching. 35 CdTe BackSide Preparation and Etching We have developed a semistatic procedure for back side etching. The etching process of 8r in methonal (or hydrobromic acid) exhibits a "melting" characteristic which attacks sides and edges as well as surfaces. The more the back surface is etched away, the faster the sides etch inward. This action tends to destroy the disc before it can be retrieved. To counteract this tendency, a thin bead of photoresist is painted on the sides and back edge in addition to the protective layer applied to the CdTe top surface. This forces the etching action to cut its own edges allowing the entire unprotected back side to "melt" uniformly to the point where the disc appears intact. Due to the large amount of CdTe which must be etched away, the stirring action of a small magnetic "peanut" and stirring plate are employed to continually refresh the etchant in contact with the sample. Exceptionally good results have been obtained with this method, however it typically takes 6 to 9 hours to remove the back side. Jet Thinning The last step in sample preparation is to jet thin from the back side of the CdTe disc such that a microscopic perforation is produced on the top surface. The edge of this perforation is thin enough to be analyzed in the electron microscope. We have employed a South Bay model 550C jet thinner modified for chemical etching. It has an automatic LED sensor system which can be set to stop the 36 jet action at the point when the microscopic perforation is made in the CdTe disc. The sample is mounted in the holder with Apiezon M grease in such a way that only the center 1mm area is exposed to the etchant. This allows a well shaped dimple to form. Also, the grease protects the covered CdTe area from the etching solution. The adhesive properties of the grease are adequate to hold the CdTe dhuafdrmly in place against the jet action. Apiezon M is also insoluble in Br/methonal but readily soluble in toluene. A step-by-step procedure of the above sample preparation can be found in Appendix A. 37 IV. RESULTS Data Acquisition CdTe substrates were analyzed for stoichiometry and impurities using both quantitative and qualitative x-ray fluorescence spectroscopy. For quantitative analysis, spot acquisitions (using a 203 beam diameter) of 100 seconds per spectrum were studied. Integrated K0 peak intensities are typically on the order of several hundred counts. An example of a spot acquired x-ray fluorescence spectrum is shown in Figure 19. Qualitative analysis was performed using x-ray fluorescence mapping techniques. The scanned sample area under study was assigned a 128 x 128 pixel matrix. A region of interest (ROI) was defined at the x-ray energy of the Ka line for each elemental map as shown in Figure 20. As the beam scans the sample area, data is acquired at each spot for 0.07 seconds and is stored in a bin corresponding to the appropriate pixel. Two main points of information are obtained using x-ray mapping. First, any gross impurity is clearly identifiable and second, an increase it: thickness is observed as an increase in intensity. For each sample mapping area the intensity scale limits are defined on the Cd map. A linear relationship between pixel counts and a color scale is assigned where the minimum counts in a pixel corresponds to black .and the maximum counts in a pixel corresponds to white on the x-ray map. An example of a x-ray fluorescence map is shown in Figure 21. 38 .m o w acaoo U umu mus“ xmoo 8M mcaaonm coauoavoua Smulx momm so “memo .a~ ouawfim >mx wommzm S .3 ... .2 . .. W...Wm_ ..wmmw._fia O O -1141” '3 m 0“ . O. m D ‘ O O a .. o N m m o m . _%m.. m U 00 m m m H. O ‘M i. 9 s an Ma 0 Q @QNNQHH m o 00 i can u eoH as . 4 m m m a FA: 2 a o _ o woo . sore“ owns" 9mm. . .. a m fl e .cowuwcfimop aHomv umopuucfi mo cofiwou wow3ocm Eauuooam mmutx onlm memo >mx woxmzm .oN orgasm 39 ,.as.a sea e. 3:3.3m asm.vm asa.am aoa.s~ saa.aa uoa.e ans n. .PP_LP_1W%h_F_|—b___f o o {a JW .I iliflfliv at.” n r .0 I I O. .4. am . N as ........ . a m a . ma . Amev Hem .n < . a ...N. U . Aeoo Hom + 00 _ . _ w _ . _ . _ . . . q . _ 09 ......—....._._....___.+.___..-.._.. 1 53 :3 I G 3 ID 0032920 40 ILL'TE CU 2 1R} : i JOOC-C ..-.. L119 IF‘P scan Figure 21. Example of an x-ray map for CdTe: Cu 2303 (Cd map on the left and Te map on the right). 111 The six CdTe samples that have been analyzed are discussed below. CdTe 3-268 Areas that are high in Te and low in Cd were seen in CdTe 3- 268. These Te deposits were oval in shape and on the order of 10002 along the long axis. They were typically found in clusters of 2-5 deposits which seemed to be randomly distributed throughout the sample. Examples of the Te deposits are shown in Figures 22 and 2A with spot acquired spectra for these regions shown in Figures 23 and 25 respectively. Figure 26 illustrates a spot acquired x-ray fluorescence spectrum obtained in a region away from Te deposits. The average weight percent concentration ratio of Te to Cd determined from four such spectra for CdTe 3-268 was found to be 0.95 i 0.05. 42 See Figure 23 \ Figure 22. X-ray map of Te deposits in CdTe 3-268 (Cd map on the left and Te map on the right). 43 E‘IIJ .NN owswfim :« o3oam ufimoooo we no aswuommm Smulx wowlm capo .mm ouswfim an: r. >mx wommzm as can . a o 8. mp._. an“ 0 II. 1 m a m s «4...... AU— U h J O m- w ' NJ if a .s 4 a a o 4 O 4*- m euaam m e B to ma n «6 n Tluao v o .... UODZE-‘UJ 44 r 7’1‘3C ’10 IF'P an! SeeFigure 25 ———— Figure 24. X-ray map of Te deposit in CdTe 3—268 (Cd map on the left and Te map on the right). 45 .qm ouswwm a“ :3onm ufimoaov we no Esuuoomm Smuux womum whom .mm ouawwm >mx sommzm ssmwvc _ won an. _ as“ an .m. «a and em so. «a sad SH 0 . o .11.. 11.11.? m .s o . m a a no . N m o a o coo wfigumnew.mnuAMMI u am..v 383.3 a In D o e C a O O U in « ..fiv. .1 v . a“ B o _ o 00 .v. m L. . $ . euaam # m 4 e 0 b. m B sienna UODZE-‘UJ 46 .mufimoaov 0H Scum hmam cowmou mo asuuooam mmutx momlm capo >mx Mummzm ....a . . . .s .m.. s..a m.. as.“ Fe. _ Mom 3.. as m «N . N- . 3 ll ‘ O. ‘ ’0 O a r .... m a m . N m . a. . a o 1 80 26 + Nae u Amml o .om ouswam ow. 0? .JuIVquw “ 3 . .u 4 U m to at u H .3: _ 9 Nu “ . _ . . . _ _ _ co — . . a. _ .. + . '0 - . . . _ . . . _ _ . _ . . . _ 4. Q . o _ . . o _ m . UODZE-‘UI M7 CdTe 2-36 No distinguishing features such as Te or impurity deposits were seen in CdTe 2-36. Figure 27 illustrates a spot acquired x-ray fluorescence spectrum obtained from this sample. The average weight percent concentration ratio of Te to Cd determined from four such spectra for CdTe 2-36 was found to be 0.8” i 0.02. CdTe “-2 Deposits of Fe and Zr were found in CdTe 11-2. The Fe deposit shown in Figures 28 and 29 and the Zr deposit shown in figures 30 and 31 were found in the same area. Several other Fe deposits were observed while no other Zr deposits were seen. Figure 32 illustrates a spot acquired x-ray fluorescence spectrum obtained in a region away from impurity deposits. The analyzed samples incurred extensive damage when mounted in the microscope to the point where a meaningful, thin film, average weight percent concentration ratio of Te to Cd could not be obtained. However the ratio from Figure 32 did meet the thin film approximation and was found to be 010 itLOB. .ooammu Hmowmhu m mo aauuoomm Smutx omnm capo .nm owswam . >m¥ wommzm ssa.a¢ am..vm aaaoam ..m..vN as...sN ..m..vd “8.3:." ..m..v asaoa L P P _ _ l - I 1 -fi ’ a O Iii O 0 a .l.. In.rullllrssvlrtlfihJIIXIIr. «I m '0 O a m m .0 ‘ O on. * new 0 N m m w m ’ 0‘". o a . m o H 00 N...— r a. o. 9.. fl 0 o . U a as m 4 B 4. m .rlaam B .. 9. v00 Q 86 a 35 u «I o H U . # 4 ._. UODZE-‘(D 7.. core 4-22 ton-3010 I" ... E.PFI’JSIONI 2X 1 12 HIM— See Figure 29 fl ’- Eigure 28. X-ray map of Fe deposit in CdTe 4—2 (Cd map on the left and Te map on the right). 50 .mm ouswfim ca caosm uHmooov on GO Esuuooom Smulx th memo .mm ouswfim >mx wummzm as . . . as .am m .va aaaé w .v 55.5 .lm $v. ..wmm mm. m empfb b w _ P.lem . . P _ n _H_ _ w m _ LI. a Is 0 h o O 1 m as 1110’ u .. ....m .. . a m . .. . .\ H. O m 0 O O 9* O O O o . InasN m .. lasso m ... . Isaam cu . 89o ... 3N5 a New m D O lass» m a 1&5” 003280) 51 Elli '(.é:-‘T I}; anaesraw 2x 142tuom»5__ See Figure 31 Figure 30. X-ray map of Zr deposit in CdTe 4—2 (Cd map on the left and Te map on the right). .om ouswfim cw :3osm uHmoooo um mo one Smutx ~18 menu .fim muowwm 52 >mx wommzm. asa.am Nmm.¢N aaa.NN smm.va Naa.ad .Nom.q asa.s _ Eltltctrt .1: _ t .11: Ir. ..a ...lvl. . is!!!” .. m . . .. ll .. on. K m o O 3 III" 00 o m .. .t m . m a v .0. a O H OO-H . a . ma nusaN . N m . tease .O . tease co . uaNNm moo.o N Naa.o u .mmw u . m txaa a uxma UCDZE‘U) 53 .muwmoooo mufiusoEH aoum Scam cowwmu m cw asuuomam woulx N18 capo .Nm ouswwm >mx wommzm . asa.s~ amm.v 335.5 can SN 1MP _ M _ ..M.bi _I_ rlila O {'0 0‘ o m m. . . LI: if .h H. om m 00 U m O. 0000 8 .H. I H O r . a . a u a U u < 4T OO O J- O nj O LT .n_ to . a . I o 0 no.0..am o..Ammw w . steam UODZE-‘(D 54 CdTe 1949 No distinguishing features (such as Te or impurity deposits) were seen in CdTe 1949. Figure 33 illustrates a spot acquired x-ray fluorescence spectrum obtained for this sample. The average weight percent concentration ratio of Te to Cd determined from four such spectra for CdTe 1949 was found to be 0.76 i 0.02. CdTe 3531 Noldistinguishing features (such as Te or impurity deposits) were seen in CdTe 3531. Figure 34 illustrates a spot acquired x-ray fluorescence spectrum obtained from this sample. The average weight percent concentration ratio f0 Te to Cd determined from four such spectra for CdTe 3531 was found to be 0.97 i 0.05. CdTe: Cu 2303 Deposits of Cu and Fe were found in CdTe: Cu 2303 (doped in the Inelt with ~1O16 Cu). The Cu deposit shown in Figures 35 and 36 and the Fe deposit shown in Figures 37 and 38 were found in the same area; while the Fe deposit shown in Figures 39 and 40 was found in a separate area. Fe traces were always found at Cu deposits; while Cu 55 no.o w mn.o .cowwou Hmofimxu m mo Eauuooom Smutx mqmfi oHpo >mx rummzm sss.a~ UO .. Mala 0 EU aom.v~ aas.s~ amm.v Na (:1 063 .mm magmas euaem UODZE-‘UJ 56 E-‘IIJQ .cowmou Hmofioeu m mo Esuuooom emulx Hmmm menu .qm ouswwm OD >mx rummzm as .GN All!“ 1|! ‘1' _ lit. f. (IRS-:1 11 3:: PWmm Maw—.ms 5H “w” m h _ ass a _;_ I a s E H _; I S 8 m H UODZE‘U) 57 (0. CDTECU3 :m-zc/io RIPP in: . :1". {. iv "IggiFZH L ,-« ---“‘—"‘; EXPHEIONI 2X 2.25 MICROS— See Figure 36 \ A Figure 35. X-ray map of Cu deposit in CdTe: Cu 2303 (Cd map on the left and Te map on the right). 58 saa.sv amm.¢m ass.sm amm.vN ass.a~ -pW. _. 1:... 11FIPIANLFIFImWIW, .JIIWIL _l 1.14 51 _ P._UWI4IQ 31‘ PM .mm ouswqm a“ czosm uwmomop no mo asuuooom emutx momm no "menu .om ouswfim >mx Nummzm so vn i_ _ we“: m. aa.sH awm.v $33.5 +11 «1. Li- _ :1 lllul 4|l4|1|1 o o o o ill‘lslllfilla .N 4 8a 886 n 88.0 .... $10! .4 o . Bid sigma UODZE‘U) 59 P‘FIS= 1CIOCU' See Figure 38 Figure 37. X—ray map of Fe deposit in CdTe: Cu 2303 (Cd map on the left and Fe map on the right). 60 .em ousmam a“ :305m uamooov we no asuuuuam emulx momm so "memo .mm ouswam >mx wommzm. . . . . .s smm.va ass.a~ amm.v 598.5 saws—v wmmwmpLEw—_lbel_._llwl_le_lbllPLPFjl ..ilullllJ «.fi . o .. lilfi‘ll. . “w m 0% o oo o co .... a .. m . .. . n we m 00 O 3.0 i e m H o ”e . a . a .TaasN O 4 4 4 . M m . a 4 o. uh a e 4Iasav 4 a . e O O AW $ A H elssao c at 4 4 . 4 O u“ a a 886 n NNNod ... AMw . w o e e a 4 4 fl $Ixsa # 4 . . e 4 m 4 .... s .2. .a .192 o ir—v UODZE‘U) 61 COTE CU 1 lrlv3C 10 {or 315?; '0 .r . . 4 ,,,» g '5: See Figure 40 Figure 39. X—ray map of Fe deposit in CdTe: Cu 2303 (Cd map on the left and Te map on the right). 62 .am muswwh cw :30zm ufimomoo on wo Esuuounm emulx momw no "menu .oc wmswwm >mx wommzm . . . . . .vn aes.an am .e ass.s 53$ fie. _ «mm mm _ _ $3m AML11_ wmm mm _ _ aaw AN. _ ammlg _ _ _ n1e _ _ _ w _ _ _ _ Is 111i! lllahlahlllll o tflflllldtllut III M W. .L. AW .- m m x D .o l- o I .v. m U o N m m U 00 on 5‘00 4 H. II m, m- .. .. . S a . fl r m .. . . :Issm m 4 e .. .T W 11988” u 4 m a. e .T woo :Issma 86 H 36 n 0&1 .4 D O t . m m 4 eleaaN UODZE-‘Ui 63 was run;:found at all Fe deposits. Also, traces of Ti and Al which were unable to be resolved by x-ray mapping were seen as shown in Figures 41 and 42 respectively. Figure 43 illustrates a spot acquired x-ray fluorescence spectrum obtained in a region away from impurity deposits. The average weight percent concentration ratio of Te to Cd determined from four such spectra for CdTe: Cu 2303 was found to be 0.82 i 0.06. 64 .oowuu He mo Bouuooam emulx momm so "oeoo .Hq ouswfim >mx wommzm naa.sv amm.vm ssa.am amm.v~ ssa.s~ smm.vH aaa.aa am¢.v $35.5 .-.L._..1e-1eemtml.l__lul is m . on . . A\..J..uuu H .H. 0" m 00 .00 o .0 .... a w . m ... ..s.... 0 o O O O “ nfi a O U 4 o is . euasm w: "U00 AT 86“ mNéu Al . fie U # ... 4T Hm C see UODZE-‘UJ 7w illl‘ 65 Noo.o n meo.o u A .oomuu H< mo aouuooam emulx menu so >mx wwmmzm uoevo "we ousmwm coo Huwusaafi Eoum hmBm scammu mo Sauuomam xmulx momm so ”mHvo .mc >mx wommzm asiav $33” “9.3m onlvw usséw 59$" “sis... 33¢ . 3. III‘IIllul'll-Iu ! O m 0.0 . o . m 0 a o . m m m w m .. w. m . m s . ‘ O Q 0 U 6 O 6 m coo moguu ongvu nflfl¢ . 0 Fill UQ Fuasm UODZE-‘UJ 67 V. CONCLUSION Analysis of CdTe Substrates Before any meaningful analysis can be done, we need to understand what effects, if any, the chemical etching used in sample preparation has on measured stoichiometry. X-ray fluorescence studies were performed on CdTe 3531 for both etchedznuicleaved regions. When a sample is mounted in the electron microscope, a substantial amount of damage can result because the CdTe is single crystal and the edges of the perforation are extremely thin. Due to the smoothness or Jaggedness of a thin edge seen in the microscope, it is possible to distinguish between etched and cleaved edges. SEM micrographs of both etched and cleaved edges are shown in Figures an and 145 respectively. A cleaved edge was chemically etched only on the top surface hence is expected to be representative of CdTe substrates. No chemical difference has been observed between these two types of edges where average weight percent concentration ratios of Te to Cd were 0.92 i 0.06 for etched regions and 0.91 i 0.06 for cleaved regions. Thus it seems reasonable to expect our analysis is meaningful in terms of bulk CdTe substrates. 68 Figure 44. SEM micrograph of etched edge (magnification= 10,000x) 69 Figure 45. SEM micrograph of cleaved edge. (magnification= 10,000x) 70 A. list of the six CdTe samples that were analyzed along with the distinguishing features seen and the average weight percent concentration ratios of Te to Cd for regions away from any Te or impurity deposits is shown in Table 2. These CdTe substrates were all expected to be reasonably stoichiometric; however, we observed as much as 20% variation and all samples appear on the Cd rich side. We attribute these Cd rich, nonstoichiometric measuramuns to a systematic error associated with detector efficency. No specific information is available for the detector efficency of our Si(Li) system, hence x-ray fluorescence correction calculations were done for a.runninal detector. The Cd and Te Ka lines fall on the tail of the efficency curve as shown in Figure 46; thus, variations in.the curve fwn~ our system versus a nominal system could be significant. Figure 46 also shows a detector efficency curve generated by assuming the average weight percent concentration of Te to Cd for CdTe 2-36 to be 1.0. Based on this, our results could be expected to increase by about 10% so that the concentration ratios of the six CdTe substrates would be reasonable. Regardless of the actual weight percent concentration ratios, the difference between samples appears significant. Our studies of CdTe substrate samples revealed numerous impurity deposits in both doped and undoped substrates. Stoichiometric differences between "good" and "bad" CdTe substrates (in terms of the quality of LPE Hg1__x Cdx Te grown on them) were also seen. 71 Table 2. Parameters Characterizing the six CdTe samples. Sample Distinguishing Average Weight Percent Features Concentration Ratio of Te to Cd in regions away from any deposits CdTe 3-268 Te deposits 0.95 i 0.05 CdTe 2—36 ---- 0.8“ i 0.02 CdTe “-2 Fe and Zr deposits 0.89 i 0.08* CdTe 1949 ---‘ 0.76 i 0.02 CdTe 3531 ---- 0.97 i 0.05 CdTe:Cu 2303 Fe and Cu deposits 0.82 i 0.06 Ti and Al traces *determined from one spot. Efficency 1.0 0.95 0.9 0.85 0.8 0.75 0.7 0.65 0.6 0.55 0.5 45 Figure 46. 72 Assuming stoichiometry of CdTe 2-36 | l l l I l I I l I I I I I l I l l Nominal detector I I l I 11 [Ii 1 ll 46 47 48 49 50 51 52 53 54 55 Atomic Number X-ray detector efficency curves. 73 Among the Rockwell grown samples, 10002 Te deposits were identified in CdTe 3-268, Fe and Zr deposits (approximately the same size) were identified in CdTe lI-2, and no deposits were identified in CdTe 2-36. A 10% variation in stoichiometry was observed between these three samples. Among the II-VI Inc. grown samples, Ferand Cu deposits approximately 10003 in size along with traces of Ti and Al were identified in the Cu doped sample CdTe: Cu 2303. No deposits were ixbntified in CdTe 1949 or CdTe 3531 however a 20% variation in stoichiometry was observed between these two samples-ea"good" substrate and a "bad" substrate respectively. These results should be compared with other optical and electronic analysis performed on the same samples. Such comparisons may lead to a more complete understanding of how CdTe substrates affect infrared detector device performance. Much work has been done with impurity doped (especially Cu) samples in order to identify the energy levels of the inunuvity in CdTe. We showed that in a Cu doped sample, significant amounts of Fe were also introduced; hence, any optical or electronic analysis must account for the Fe along with the Cu. In general for any doped (or even undoped) sample, one should investigate the micro-chemistry before any meaningful conclusions are made as to how impurities interact with the host material. 7‘4 Our work illustrates the capabilities of xrray fluorescence spectroscopy as a tool for chemical analysis, especially transition metal impurity studies. The Scanning Transmission Electron Microscope can also be used to study defect densities (via bright field imaging) and crystallinity (via micro-diffraction) associated with these1umnudties; hence it is a powerful and versitile instrument for the characterization of semiconductors important to the infrared detector industry. Future Work With our present sample preparation technique it is possible to study the chemistry, crystallography and defects of both substrates and epilayers. By removing different amounts of the epitaxy sample surface during the initial etch, the entire epilayer, the epilayer- substrate interface, and several depths of the substrate itself can be studied. Such experiments should be performed for substrates with and without epilayers to develop a better understanding of epilayer-substrate interactions in terms of defect propagation and impurity diffusion. Comparing these results with optical and electronic analysis performed on the same materials would provide a rather detailed understanding of infrared, semiconductor detectors. 10. 11. 12. 13. 14. 150 16. LIST OF REFERENCES T. J. McGee and G. R. Woolhouse, MCT Workshop in Dallas, Feb. 1983. R. E. DeWames, M. Hinnrichs, G. M. Williams, J. Bajaj, S. H. Shin, and E. H. Cirlin, MCT Workshop in San Diego, Oct. 1985. ' N. B. Stevens, Semiconductors and Semimetals, vol. 5, chapt. 7, Academic Press, New York and London, 1970. D. Long and J. L. Schmit, Semiconductors and Semimetals, vol. 5, chapt. 5, Academic Press, New York and London, 1970. J. Bajaj, private communication. P. W. Krusen Semiconductors and Semimetals, vol. 18, chapt. 1, Academic Press, New York and London, 1981. K. Zanio, Semiconductors and Semimetals, vol. 13, Chapt. 1, Academic Press, New York and London, 1978. C. C. Wang and S. H. McFarlane, III, Characterization of Epitaxial Semiconductor Films, vol. 2, Elsevier Scientific Publishing Company, Amsterdam-Oxford-New York, 1976. W. F. H. Micklethwaite, Semiconductors and Semimetals, vol. 18, chapt. 3, Academic Press, New York and London, 1981. P. F. Kane and G. B. Larrabee, Characterization of Semiconductor Materials, chapt. 5, McGraw-Hill, 1970. E. Molva, J. P. Chamonal, G. Milchberg, K. Saminadayar, B. Pajot, and.G. Neu, Solid State Communications, vol. 44, no. 3, pp. 351-3559 1982. E. Molva, J. L. Pautrat, K. Saminadayar, G. Milchberg and N. Magnea, Physical Review B, vol. 30, No. 6, pp. 3344-3354, 15, Sept. 1984. E. Molva, J. P. Chamonial, and J. L. Pautrat, Phys. Stat. Sol. (6) 109, 635—644 (1982). J. BajaJ, private communication. S. H. Shin, J. BaJaj, L. A. Moudy, and D. T. Cheung, Appl. Phys. Lett. 4;, (1). Pp. 68-70, 1 July 1983. A. T. Hunter and T. C. McGill, J. Appl. Phys. 52 (9). PP. 5779- 5705, Sept. 1981. 17. 18. 19. 20. 21. 22. 23. D. 1.. Polla and C. E. Jones, J. Appl. Phys,. 52 (8). PD. 5118- 5131, August 1981. " D. I.. Polla, R. L. Aggarwal, J. A. Mroczkowski, J. F. Shanley, and M. B. Reine, Appl. Phys. Lett. 49 (4), pp. 338-340. T. W. James and R. E. Stoller, Appl. Phys. Lett. 44, (1). Pp. 56-58, 1, Jan. 1984. S. Wood, J. Greggi, Jr., and W. J. Takei, Appl. Phys. Lett. 46 (4). pp. 371-373. 15, Feb. 1985. ‘ N} .J. Zaluzee, Introduction to Analytical Electron Microscopy, Chapt. 4, Plenum Press, New York and London 1979. D. B. Williams and S. W. Edington, Norelco Reporter, Vol 28, No. 1, May 1981. N. J. Zaluzec, EMSA Bulletin 1984. 77 APPENDIX A CdTe Electron Microscope Sample Preparation: Step-By-Step Procedure 1) 2) Start with a bulk piece of CdTe approximately 7mm square and about 600 um thick. Place the CdTe piece, top surface up, in a small dish of 2% Br/MeOH solution for 4.5 minutes. Remove it and rinse it throughly in methanol. This etches away the top 25 pm to remove any mechanical defects, leaving a clean surface of CdTe. Place the CdTe piece on a filter paper in a covered petri dish to keep it clean. Be careful not to handle the CdTe with hands or metal objects. Any transferring of the sample should be done with clean plastic tweezers. To mount the CdTe on the 3000 rpm spinner, place and trim a piece of double sided tape on the spinner platform. Now center the CdTe piece, top-surface up, on the spinner platform and apply a small amount of pressure to the corners to secure to the tape. Be careful not to touch the center area of the sample. Turn off all lights and turn on a yellow safety light to protect the photoresist. Turn on the spinner and apply a drop of Shipley 1375 photoresist at the center of the CdTe piece. 3) 78 Let the spinner run for about 30 seconds and repeat the photoresist application until the surface of the CdTe takes on a reddish tint. This may take as many as 4 to 5 repetitions depending on temperature and humidity conditions. Using plastic tweezers, carefully remove the CdTe piece and place it on a teflon slide to dry. Once completely dry, clean any excess photoresist off the back side using acetone and a Q-tip. Be very careful not to let any acetone touch the front side. Return the sample to its petri dish. Again remember to do all undeveloped photoresist work in a yellow light environment to prevent premature development. Mount a 250 watt ultraviolet flood light on a ring stand about one foot off the base. Let the lamp warm up for 10 minutes. With the photoresist side of the CdTe piece up in an uncovered petri dish, place the photoresist mask, disc side down, on the sample and center one complete disc. It has been determined that the best results are obtained when only one disc is exposed per sample. Place the assembly under the flood lamp for 5 minutes. Remove the sample and develop it in a small beaker of Shipler Microposit 351 developer which has been diluted by 50% with distilled water. Use a slow agitating motion until the disc is clearly visible and the surrounding area is shiny (about 17 to 20 seconds). Rinse the disc throughly in distilled water. Examine the disc carefully to inspect for a clean, well shaped disc. If the disc is not satisfactory, strip it with acetone and return to step 2. When 4) 5) 79 a satisfactory photoresist disc is obtained, the room lights can be turned on and the CdTe piece can be returned to its petri dish. Obtain a small dish and fill it with a 4% Br/HBr solution. HBr is used to protect the photoresist mask which is soluble in methanol. Place the masked CdTe sample face up in the covered dish. Let the sample etch for 1 hour and 15 minutes. Remove the sample and rinse it in distilled water. This etching step produces a plateau which should be ~150 um in depth as was shown in Figure 17. Measure the plateau with an appropriate depth gauge and re-etch if necessary to obtain the proper dimension. Once etching is complete, strip off the photoresist with acetone and return the CdTe piece to its petri dish. Remount the CdTe sample, plateau side up, onto the spinner. Plug the spinner into a variac and set on a slow speed. Under yellow light, apply a thick coat of photoresist to completely cover the CdTe piece. Remove the sample and let it dry on a teflon.slide. This may take several hours as the photoresist coat is thick. Once the sample is dry, clean the back side using acetone and a Q-tip. Now paint a very narrow bead of photoresist onto the sides and back edge of the sample in order to protect it while etching off the back side (see Figure 47). If this step is not taken, the disc will ruM: have the chance to emerge with a well formed shape. This should all 80 Backside V Photoresist protective layer Figure 47. CdTe piece with photoresist protective layer prior to semistatic backside etching. 6) 7) 81 be performed on the teflon slide. The photoresist will bleed under, to the top side. Once the edges are dry, turn the sample over and allow the top to dry. After all the photoresist is completely dry, develop in a 50-50 solution of developer and distilled water and return to its petri dish. Fill a medium sized dish with a 4% Br/HBr solution and place it on a magnetic stirring plate. Plug the plate into a variac and turn it on. Put a small stirring "peanut" in the dish and set to the slowest possible stable speed. Now center a teflon basket (see Figure 48) in the dish and place the sample face down in the center of the basket. Cover the dish with a watch glass and let the back side etch until the CdTe disc can be retrieved. This will take 6 to 9 hours. Remove the sample and clean it throughly in distilled water. Place the sample on a filter paper and strip off the photoresist with acetone. Using vacuum tweezers, put the disc on a clean filter paper and return it to the petri dish. Follow the chemical thinning procedure in the manual to set up and tune the South Bay model 550C Jet thinner. Use a 10% glycerol, 0.5% Br in methanol etching solution. Remove the sample holder and secure a mylar disc to it with a thin ring of Apiezon M grease. Take care not to block the light sensor hole of the holder. The vacuum tweezers can be used to manipulate the mylar disc into place. Now place the CdTe disc, back side up in the center of the holder on top of the mylar disc and Figure 48. Teflon basket used to hold CdTe piece during semistatic backside etching. 8) 9) 83 cover the sides with grease. Leave the center portion of the sample uncovered as shown in Figure 49. It is desirable to cover all but the center 1mm of the CdTe disc in order to seal the sample from the etchant and define the thinning region to insure a well formed dimple. Mount the holder assembly into position on the jet thinner and activate the automatic sensor mode. The trigger level is quite sensitive; however, very nice results can be obtained once a feel for the apparatus is obtained. When the sensor turns off the jet stream, remove the sample holder assembly and rinse it in methanol. Examine the assembly intact under a light microscope for a perforation in the CdTe. Repeat the jet thinning until a microscopic perforation is obtained. Now dissolve the grease with toluene. Submerge the top of the sample holder in toluene. Remove the CdTe and mylar discs from the holder by poking a needle through the back of the holder. Slide the mylar disc off the CdTe disc with care and retrieve the sample from the toluene bath using vacuum tweezers. Put the sample on a clean, small filter paper and submerge it in a warm bath of methanol for 30 minutes. Note that it is wise to have a large filter paper in the bottom of the bath. Be very careful with the CdTe sample as the edge of the perforation is only 10003 thick and very 84 CdTe Apiezon M grease LED light sensor 7 hole Holder ————qr—— ——-——-———————-—————' .l I I I I l I I I I I I I I I I I l ! I I I I I Figure 49. CdTe disc mounted with Apiezon M grease on holder for South Bay model 5500 jet thinner. 85 fragile. Remove the CdTe disc and transfer it to a clean .filter paper. The sample is now ready tc>be mounted in the microscope. "I7'11IIIILIIIIIIIIIIIIIIIIIII