RD-0144 887 INVESTIGATION OF THE EFFECT OF ANNEALING ON THE /i.

DISLOCATION RND SUB-GRAIN..(U) ROYAL SIGNALS AND RADARESTABLISHMENT MALVERN (ENGLAND) N TROUGHTON MAR 84

UNCLASSIFIED RSRE-HEMO-3698 DRIC-BR-922i8 F/G 29/12

Eu'.

~~3211111UI.0-~8 I~I.

IIIIII 32 .2

Th -6

MiP1 1111 ,[W N 1

I r'(i 0.01ifl it A I ,( \'fitkr1P r ,iwi-

" UNLWH11

~RSRE

MEMORANDUM No. 3680

ROYAL SIGNALS & RADARESTABLISHMENT

iNVESI IGATION OF THE EFFECT OF ANNEALING ON THEDiSLOCATION AND SUB-GRAIN BOUNDARY CONTENT OF

* SOLVENT-EVAPORATION -GROWN CdTe

Aut-or N 1, ughtorl

0 PROCUREMENT EXECUTIVE,> MINISTRY OF DEFENCE,

CD rRSRE MALVERN,WORCS.

-

:: UNLIMITED

............ ......................

rI Pinfl r r A'F I ,C)Vr PNM NT I ',I, II"I

RC'AL S IL NALS AND H.,L)AR LSIABLISHMtNI

M.~rnor aid urn ) l

TI TI[ INVESTIG;ATION OF THE EF.FECT OF ANNEAI.1N(, ON TRLn 1)1;.oc(ATlAIND) SUB-GRAIN BOUN:ARY CONTENI OF S LV ,NI -VkAt'k\AIlO: (;.(W: '1l

AIL'THO,: N. Trought on

DATI: March 19S-4

SUMMARY

Solvent-evaporation grown Cdte wafers have been examint-d using opticalmicroscopy and etching techniques to determine the subgrain boundarystructure and density. The results indicate that the material contaii::subgrains of the order of 100-300 ;jm in diameter, the boundaries of whichare formed by polygonised dislocations in densities exceeding 105 cm- 1 .Annealing for periods of more than 24 hours at temperatures above 800 Cleads to a reduction in the sub-grain boundary dislocation density and,in the best case, to complete removal of some of these boundaries.

V~cce~ onFor _

i*i

Copvr ,gh:,

CControllei HMSC' London

1984

1 :0I t )' liCF I) I l I)NNt,' ill tN I! vPI Nl~i

INVES'I IGATI ON S 0 Till. EFFICT O1' ANNLALINt; ON Till) ID Slt;A'l ION ANI)

SUB-(;RAIN BOUNDARY CONTENT OF SOLVENT EVAPORATION CROWN C"dTte

Nicholas Troughton

I4

INTRODUCT ION

Cadmium Telluride has theoretical potential to be a good material flodevice substrates. However, for best performance good quality single crystalmaterial is essential. It is difficult to get large single crystals due toproblems of growth - twinning 1I1 and other defects are easily grown intocrystals. Epitaxial layers grown by MOCVD show up these faults as they -a1reasily "grow through". Even if a large grain is chosen, the surface layer isstill not good. This can be due to low angle grain boundaries or sulgrainboundaries.

V iThe examination of these sub-grain boundaries, and attempts to remove themfrom slices is the scope of this project. Three main techniques were used tk.exarmine the problem; etching,mechanical deformation of samples and annealing.Etching along with optical and infra red microscopy was used to find theeffects of mechanical deformation and annealing.

ETCHING

Etching is a method commonly used in materials work to probe the structureof a material. The material is polished, mechanically and chemically to ensurea mirror polish. All the samples used were lapped with 60 urm carborundum andpolished with 0.3 um alumina/ethanediol, and finally chemically polished in 2%Br 2/MeOH. The etch will preferentially attack high energy areas and soreadily shows up grain boundaries and twins. 2Z Bromine-Methanol attacks thesurface indiscriminately, as it reacts too fast to etch, and hence is used fora chemical polish.

Many etch mixtures have been used on cadmium Telluride 12] but the twomost useful ones for sub-grain boundary examination are a weak bromine/methanolsclution and a hydrogen peroxide/hydrofluoric acid solution.

The hydrofluoricperoxide etch used is a 3:2:2 mixture of hydrofluoricacid: Hydrogen peroxide: water. The hydrofluoric acid used was 40%, Aristargrade, the hydrogen peroxide 100 vols Arialar, and the water was deionised.The samples were etched for - 30 secs, washed in water then methanol. Toremove the inevitable black oxide film, a dip in dilute Bromine Methanol isessential. When the oxide film is removed to leave a shiny film; the sample isremoved, washed in methanol and dried. The etch pits formed are easilyvisible under an optical microscope. Unfortunately, the etch will only attackthe (111) A face of the crystal, so if the slice is not orientated correctly,it will not etch. The photograph (I) shows a grain which does have the rightorientation, but has a large twin running through it, which is not affected

* by the etch. A dilute solution of Bromine-Methanol (about J%) in the presenceof light, will show up subgrain boundaries(2 This can be seen by the photo-graph (2). The photograph shows a grain boundarythree twins and a network ofsubgrain boundaries. The microscope has to be used with Nomarski interferenceattachment for maximum resolution of very small variations in surface topo-graphy. These variations are slight, and occasionally very difficult to see.

0

_ ._ , , - " .

I!! POn 0) Ij 1) A I (iA\'t lI"JMtN r I II N

Thi makts thm appear dubious evideuce. They can be shown to be real cffvctt;I thei th-t artifacts of bromilie methanol light etching, by ,po li shin g an.d ?v.-

etching. This etching method i!; not as ea:;y to achievu as tht, hydrot 1,t iIacid-peuxide etch also uscd, but it is avai lable oni any oritentat ion faci.

lNs'rRON LOADIN(;

The Instron is a machine to mchani-cally deform samples. It works byhaving a crossbar move downwards onto the sample. The crossbar is cornectedto a gearing, arrangement to allow a variety of downward speeds. The samplc isconnected to a tension load cell which measures the load on the sample. Ahigh load measurement shows that the sample can stand up to high stress.Cadmiun. teiluride can not stand very large strain rates at room temperatare.A small furnace can be fitted around the sample so as to heat the sample up.

At - 2000 C, Cadmium Telluride can be deformed plastically for 3% deformationa: about 3 x 10 - 5 s - 1 strain rate. After deformation, the sample has to berepolished, and etched to find out the defects put into the sample. (gu-)

ANNFALIN(.

Annealing of samples was used to attempt to remove the defects. It isnecessary to heat up the sample to high enough temperature to enable the defectsto move in the sample. The annealing time has to be long enough for the defectsto reach the edge. Then, once the defects are removed, the sample has to belowered in temperature without fresh defects coming back in. Twins are verydifficult to move grain boundaries move slowly, but nobody has attempted tostudy sub-grain boundary movement before.

Samples were placed in silica glass ampoules, with an excess of eithercadmium or cadmium telluride and were pumped down to 10- 7 torr. (The samplewas ncr.,illy left overnight even if this value is naturally achievable aftera couple of hours). They were warmed to drive off any spare surface waterwhen the vacuun. was only about 10- 4 torr. The samples were then sealed intoa tube by heating the silica glass adjacent to the thimble until it melted,and caused a good seal. The sample tube was then put in the furnace, andleft for varying periods of time. When the samples were removed, this couldbe done either slowly, allowing the sample to rest at an intermediatetemperature, or dropped to room temperature. Sometimes the bulb was quenchedto ensure that no cadmium could be deposited on the sample.

PISULTS

The Instron loading experiments failed to prove conclusively the removal

or lack of removal of slip. This was due to

1) lack of sensitivity of etch towards slip bands

2) or lack of slip bands.

The lack of slip bands could have been due to the Instron load cell beingentirely non functional. Sub-grain boundaries were neither moved on the

surface or in the bulk.

Thin slices were annealed under the conditions laid out on the followinotable: the results column briefly relates how effective the annealing was.The CT 283 SE pieces did not possess a 011) A face so they could net be etchedusing the hydrofluoric etch which allows better definition. However, thesepieces did not show subgrain boundaries clearing.

2

•0 i _ i _ i _, :i i , . . ., _ J -. , . -.- , -. - . ."- "

ljL Q) G)

OUV) '4".

L' M..- Lb :11 . " , V, ' -, -41 C(U' *O i- u v) )-, *. :C V" w) -

61Z J "XU r.~U -'- -) *: I" .f. a f, ;on :1) I . WC :j D ) 03 C) U) C, 0 .U -~ cc

D > 46J &JO CO 41IC IC 4-2 C' .Z

D 4.6l t-C~jt E-.C.ErC> U- U C~ U) w -UM

:3 1-." Cl U ) CU C.'U *" n .) , >- C.

CCC) ~ ~ ~ C m .fl .. j C: 2 2 *- .- C 2 -,l C. in' tr ~ )

CL- -C/ cn C- - 2 C: :- CL S., CCC -. m, c) cc "IC - )

tc Cl L M) W. Cl C ; Ul 2, CC, -4 Q) Q) 0) U) r2~ C U .~ l . t.;'C

C,

J,- En 0 (1) C Q U .C. U, CcI . cc XL . L

U) U) sr) 9. C. U) V ) U) V)' fl VI T V

S CX:

a) 0 ) a) (If c U) ) (L) C) U ) U3-. ~ ~ ~ -3-4 F-'3. C . ~ C - ) C. 3 - )

C C Cl Cl l 7)C -Q l Cl ul C l C

CC C

0 ) ;0%,C OD ID In C) 10 %) U) U) 00CC cc a

H ~ ~ ~ t It H H o -v"' r-I a, a, Ln ':C U 0 .

IN CU IN C1 U4 U 1- U

w 3 L; C1C) j l

L) L

Hi

DLiSCUSS ION

The main parameters investigated were: length of anneal time, how rapidly

the sample was cooled, and the temperature of the anneal.

Generally, the longer the sample was left annealing, the better the

result. Sample CT 282 SE 21a was annealed for 24 hours (Photographs 5 & 6)

whereas sample CT 282 SE 27 was annealed for 168 hours (Photographs 7 & 8).

This shows the annealing is not completed at the shorter time, rather than the

cooling is putting these faults back into the sample. For the high

temperature annealing, the longer annealing times gave better results and for

the 9b hour long annealing gave very good results in some areas of the sample.

Co-qarison of CT 282 SE 21b and 22a suggests that rapid cooling will put

mere stress into the samples and will give poorer results.

Phe higher temperature anneals generally gave better results. A short(24 hrs) anneal of CT 282 SE 19a gave a much better result than CT 282 SE 27(ccmpare photograph 9 & 10 with 5 & 6). Keeping the sample in even longeriaroved the surfaces. On CT 282 SE 19b half of the material exhibits stwingsubgrain structure whilst half seems 'clean' - see photographs 11 & 12. This

suggests that the problem is not one of edges forming stresses in the coolingprocess, but incomplete removal of defects. Why the defects should be

clusttred on one side of sample is a mystery. They are not so clustered inthe identically run sample CT 282 SE 15c; Photograph 13.

CONCLUS ION

From this work I conclude that, it is possible to reduce the amount ofsubgrain boundaries in cadmium telluride by annealing at 800 0 C. Unfortunately,it may be difficult to anneal for the time necessary to do so withoutcontaminating the sample. The subgrain structure may be further reduced bycooling the sample at a lower rate. This might be achieved by having alarger distance between the sample and the cadmium in the bulb of the ampoulewhen quenching.

REFERENCES

1. A. W. Vere, S. Cole, D. J. Williams,J. Electronic Materials Vol 12. No 3 May 1983

D. J. Williams - private communication

4

0,n H O L: , f " ; Ml

. . . . . . .

P$lFfirfim) Ar w'V 4NMpW1 4-W4I.Ncf

FI_ITu CApTIONS

Fig I CT 282S1/4 Cdle slice as grown

Fig 2 CT 282SE/4 (Normarski Contrast)

Fig 3 CT 282SE/4 Showing sub-grain structure

Fig 4 Slip lines introduced by plastic deformation i.: compressit i

Fig 5 CT 282F/21a before annealing

Fig 6 CT 282SE/21a after annealing (24 hours at 600 C)

Fig 7 CT 282SE/27 before annealing

Fig 8 CT 282SE/27 after annealing (168 hours at b00''.

Fig 9 CT 282SE!19a before annealing

Fig 10 CT 282SE/19a after annealing (24 hours at 80&0)

Fig 11 CT 282SE/19b after annealing (96 hours at 800,). Specimen cd4t

Fig 12 CT 282SE/196 after annealing (96 hours at 800 0C). Specim - :etr L

F hig 13 CT 282SE/15c afttr annealing (96 hours at 80OUC)

a -' -" - , - , -.-" " " " -, -.- -. ..

ii(' .:~:v1 (AIt I i cQ o 5 ;is rowji

C I

AOi

by1 oMe

11 1 it'od

I II~, (251SI/I1:1bef ore ;mn(.l'i ing

0c

;1 0

4~ 1411 49)4 C I, I '~.'i h'j!.~ I I" I

0

6

p 1 -) >~~I ~; I: / -~ 7 - I 'i i'! .1 I n'b

0

a

I

4c -r

.4

AS

- I

* t .*t.

*II

~., M XI *S. I'. *lIITilI ~ui.' i IiH i'ii ()(14Y ()

0

I,

S

~W R - .. AL - . .. ~

-, ,,..F~t

,

(II

~.jj

~ ~. 7 1 1 ~i~t: I JL~;

S

a

.6

4

S

U

S

Pig IQ I if ter ~une.ii in ~: (7 hnni--~ ;iL ~I)() ()

S

S

itt PIIC (f (II At r;rIktf f(Nkil Nj I

ii

0

Ci; II (>1 282SEI/19b after annealinig (96 hours at 800 0 C) Specilnen edge

6I

. . ..0" " . . . ' m l 'lm " i ' . .. . . " . . . .

0I

' r ~ rj

0

~ .

I p

* ~2 CT> I 9~ after anneal jug (96 i.ours at ~p(cjmvfl celare

a . / .7).1~.

-1.

* :1/

///

0

//

0

'4

C'

0

0

D(OLUMIl %I L;OgTI ." I I

r I I ' a iicat on of ,heet . UN(LASS 1 . ..ii

(A far a% po', te ihi , sheet should contain only uncla ,,ified information, If f i'. nece, ary te entpr

c asfl'ied ini:rmafion, the box concerned must be marked to indicate the ca'',icf alion eqj (P) (C)

)Rl( ReferenLe (t known) 2. Originafor's Reference 3. Aqen(y Reference Report Seru-ityMEMORANT)Vrl 3680 1 1/ ( f ia "" t :

2" € .. ,, Cde (i b. Originator (Corporate Author) Name and location

known)ROYAL SIGNALS AND RADAR JESTALISi:

3. -nTcrnQ Aqenc ' fa. ornnsoring Agenc v (Contract Au ' ' y Nipe a'd toc ....

Cede (!f known)

V " 4le Investigation of the effect of annealing on the disiccation, aicd sub-

grain boundary content of solvent-evaporation-growi CdTe.

a Title in foreign Language (it, the case of translations)

*t?t. Preseneej at (for con 4erence 7acer ) Title, vlace and date of conference

f.. Author ' Surrname, initials 9(a) Author 2 9(b) Authors 3,4... 10. Date Dc. re'.

TROUGHTON, N.

Coe4ra~t Number 12. Period 13. Pro)ect 14. Other Reference

Dstrib tien statement UNLLMITED

Dfscr:+cr (or keywords)

continue on seaarate Cece of vacer

At tract

Solvent-evaporation grown CdTe wafers have been examined using optical

• microscopy and etching techniques to determine the sub-grain boundary structure

and density. The results indicate that the material contains sub-grains of the

order of 100-300 um in diameter, the boundaries of which are formed by p.]v-

gonised dislocations in densities exceedingl05 cm-l. Annealing for periods of

more than 24 hours at tt mperatures above 800C' leads to a reduction in the slb-

grain boundary dislocation density and, in tle best case, to complete remr,val

, of soe of these bound-i ies.

,'. . . . p,.

0"

I