Schottky barriers at metal-CdTe interfaces

Vacuum/volume 31/numbers 10-12/pages 639 to 643/1981

0042-207X/81/100639-05502 00/0 @ 1981 Pergamon Press Ltd

Printed m Great Britain

Schottky barriers at m e t a I - C d T e interfaces M H Patterson and R H W i l l i a m s , Ulster, Colerame, Northern Ireland

Department of Phystcs, School of Phystcal Scmnces, New Untvers~ty of

We have adopted the multltechntque approach to mvestlgate the Schottky bamer formatton for a range of metals on clean cleaved (1 10) surfaces of cadmtum tellurtde We have probed the growth of tm, mckel and copper hires on these surfaces by XPS, UPS, LEED and AES Thtck him values of Schottky bamer hetght were estabhshed by C-V and I-V techmques, and conhrmed by momtormg the degree of band bendmg observed by UPS The results are dmcussed together wtth those already reported for Au, Ag and AI on CdTe The results indicate that m many cases the/nterfaces formed are not abrupt and that mtermtxmg between the metal and the CdTe occurs The observattons are dtscussed m terms of the vartous theortes of Schottky bamer formatton In particular the role of the metal atoms as posstble dopants, the role of the heats of reactton of the metal with the CdTe surface, the relevance of the metal work functton and the apphcab/hty of the defect model to the metaI-CdTe system are dtscussed

Introduction The m e c h a m s m responsible for the f o r m a u o n of Schottky barriers at the interface between metals a n d s e m i c o n d u c t o r s has been the subject of m u c h research over the past years Several elegant theories I 6 have been put forward to account for the Schottky b a m e r formation, but to date there has been httle agreement as to the most i m p o r t a n t processes revolved These theories have considered the relevance of intrinsic surface states on the semiconductor ~, the t u n n e l h n g of metal wave functions into the semiconductor 2 s, m a n y body effects 4, a n d metal reduced gap s t a t e s / s In general, though, all these theories m a k e the assumption t h a t the interface formed between the metal a n d the clean semiconductor is b o t h ordered a n d atomically a b r u p t Recently however it has been shown that this a s s u m p h o n is not universally correct 6 s, and that m m a n y cases considerable intermixing of the metal and s e m i c o n d u c t o r occurs even for interfaces fabricated at r o o m t e m p e r a t u r e It has recently been estabhshed t h a t defects, such as cation or a m o n vacancies, caused by this intermixing at the interface can d o m i n a t e the formatxon of the Schottky barrier 7 9,10 The interfaces formed between metals and the III V c o m p o u n d semiconductors have been extensively investigated a n d it has been found that some metals, such as alumlnlum, can disrupt the semiconductor surface by chemically reacting with it 1~ O t h e r workers have s h o w n that considerable intermixing of the metal a n d semiconductor for the Au G a S b system occurs even at r o o m t e m p e r a t u r e 9 It has also been s h o w n that exposure of a clean semiconductor surface to an atmosphere, such as 0 2, prior to metal evaporation can cause a change m the transport propertms of the metal semiconductor interface, m some cases as a result of doping the surface layers of the semiconductor 12 13 As a result of these lnvesugatlons st is n o w b e c o m i n g generally accepted that the f o r m a t i o n of Schottky barriers on the III V

c o m p o u n d semiconductor ~s often d o m i n a t e d by these defects at or near the surface of the semiconductor 7 9 lO T o find out how apphcable the 'defect' model ~s to other metal semiconductor systems we have extended our investigations to a metal II VI semiconductor system In th=s paper we report on the fabrlcaUon of metal contacts to c a d m m m tellunde I n m a l studies of Au, Ag and AI film growth on vacuum cleaved CdTe have already been described elsewhere 8

Experimental C a d m m m tellurlde crystals, doped with m d m m , with carrier concentratxons m the range 1016 10 iv cm -3 were grown at our l a b o r a t o r y by a verucal B r l d g m a n t e c h m q u e O u r m e t h o d of e x p e r i m e n t a t i o n revolves carrying out c o m p l e m e n t a r y surface investigations sequentially on the same sample m the same ultra high v a c u u m c h a m b e r without breaking the v a c u u m To enable us to do this we have three u h v c h a m b e r s at our disposal O n e contains the techmques of X-ray photoelectron spectroscopy (XPS), Auger electron Spectroscopy (AES) and Low energy electron diffraction ( L E E D ) The second u h v c h a m b e r contains faclhtles for XPS, L E E D and uv photoelectron spectroscopy (UPS) The third uhv c h a m b e r contains AES, L E E D and the faclhty to carry out I V and C V m e a s u r e m e n t s After m o u n t i n g the crystals m statable holders, the crystals could be cleaved to produce atomically clean (110) surfaces m a v a c u u m of ~ 10 -1° torr O u r a p p r o a c h is first of all to characterize t h o r o u g h l y the v a c u u m cleaved (110) surface of C d T e by the techmques available to us Controlled a m o u n t s of metal, from fractions o f a m o n o l a y e r upwards are e v a p o r a t e d o n t o the clean surface of CdTe The early stages of Schottky b a m e r f o r m a t i o n are then m o m t o r e d by XPS, 639

M H Patterson and R H Wtlhams Schottky barners at metal-CdTe interfaces

U P S and AES Once a thick overlayer of metal had been established the Schottky barrier heights were measured, m uhv, by conventional I V and C V techniques At this stage L E E D was carried out to determine the degree of order present m the metal overlayer The results presented are for experiments carried out at r o o m temperatures

(b)

Results In this paper we summarize the results obtained for the evaporation of various metals on vacuum cleaved CdTe The general a p p r o a c h is typified by the results for Cu Figure l(a) shows the U P S spectra measured for vacuum cleaved CdTe with progressive deposition of copper onto the surface Spectra are measured in the angle resolved m o d e along the surface normal The various peaks observed m the clean spectrum may be understood, to a first approximation, in terms of bulk electron states 14 The two close peaks prominent at a binding energy of 10 9 and 11 5 eV originate an the spin orbit split Cd4d levels As copper is deposited on the clean surface all levels In the photoemlsslon spectrum appear to shift to lower binding energies This ts due to the fact that the Fermi level at the surface is shifted towards the valence band by 0 2 eV, as dlustrated in Figure l(a) (Inset) This interpretation assumes that the bands near the atommally clean surface are fiat and that the electron escape depth (10 20 A) IS short c o m p a r e d to the band b e n d m g (several hundred A) following metal deposition The Schottky barrier measured by I V m e t h o d s following the d e p o s m o n of a thick Cu film was 0 4 eV which is entirely consistent with the Fermi level shift shown m Figure l(a) A n o t h e r noticeable feature in the spectra is the gradual appearance of a small peak, at a binding energy of ~ 10 3 eV, 1 e at an energy of ~ 0 6 eV less than that of the Cd4d levels in CdTe, as Cu is progresswely deposited This peak is due to cadmium atoms which are in a different e n w r o n m e n t to those cadmium atoms m the CdTe, l e due to c a d m m m incorporated in the copper overlayer Unfortunately the possible out-diffusion of tellurium is difficult to assess by U P S since no core levels are accessible However XPS studies lead to the conclusion that both Cd and Te are incorporated in the Cu electrode and that the interface IS very far from being atommally abrupt and extends over many tens of angstroms Finally L E E D observations of the thick metal overlayer indicate polycrystalhne film growth for all metals studied Nmkel behaved in a very similar m a n n e r again leading to a non abrupt interface and a Schottky barrier height of 0 45 eV (see Table 1) Gold led to a m u c h higher Schottky barrier, with U P S behavlour

Ef

2

4

6

8 t0 t2 14 16 Energy,

Ef 2

4

eV

6

8 io [2 14 16 Energy,

eV

Figure I (a) Angle resolved photoelectron spectra for vacuum cleaved CdTe with controlled evaporation of copper Spectra shown are for normal emission, hv = 212 eV, angle of modence of hght = 55" The insets show the posu~on of the valence and conductmn bands w~th respect to the Fermi level (b) Angle resolved photoelectron spectra for vacuum cleaved CdTe with controlled evaporation of indium Spectra shown are for normal emlssmn, hv = 212 eV, angle of incidence of light = 55c The resets show the position of the valence and conducuon bands with respect to the Fermi level similar to Cu and NI, but no evidence of Cd In a different chemical e n v i r o n m e n t to that In CdTe was found F o r silver contacts on

Table 1. Barrier height measurements (q~h),Schottky theory predmtlons (~bm--/CdTe), heats of reaction of various metals wuh CdTe (the heat of reaction value AH* includes an allowance for the heat of condensation of the metal z°)

640

Bamer hmghts (eV) UPS

Metal

I V

Au AI Ag In Sn Cu Nl

0 92 Ohmic Ohmic Ohmic ~04 ~04 ~045

0 85 Ohmic Ohmm Ohmic 03 02 04

~m TM

(~bm - Yfd re)

AHR

(eV)

(eV)

(eV metal atom l) (eV metal atom -1)

5I 4 26 4 28 4 12 442 465 515

0 82 <0 0 _<0 0 14 037 087

- 0 25 + 0 29 - 0 04 +033 +059 +040

AH*

-- 3 67 2 67 -256 -281 -291 --41

M H Patterson and R H Wtlhams Schottky barriers at metal-CdTe interfaces

clean CdTe (110) surfaces no Fermi level m o v e m e n t was observed a n d indeed the I V m e a s u r e m e n t s showed largely ohmic b e h a v l o u r consistent with a very low barrier Again, like Au, n o evidence of Cd in a different chemical e n v i r o n m e n t to that in CdTe was found The results o b t a i n e d for the three metals In, AI a n d Sn showed noticeable and i m p o r t a n t differences in b e h a v l o u r from those illustrated by Figure l(a) Following the deposition of small fractions of a m o n o l a y e r of In, for example, the whole p h o t o emission spectrum, originating in the bulk CdTe, displays a shift to higher binding energy, Figure l(b), indicative of a shift of the Fermi level towards the c o n d u c t i o n b a n d by ~ 0 2 eV [-see inset of Figure 1 (b)] As the metal overlayer grows into a c o n t i n u o u s film the spectra experience a shift back to the original binding energies Similar effects were observed for A1 Tin also showed similar b e h a v l o u r except that the final Fermi level position IS a b o u t 0 2 eV closer to the valence b a n d at the surface as c o m p a r e d to the flat b a n d condition (see Table 1) In all three cases electrical m e a s u r e m e n t s showed barrier b e h a v l o u r consistent with these interface Fermi level positions F o r these three interfaces XPS studies showed evidence of considerable intermixing The instability of the v a c u u m cleaved (110) surface of CdTe was d e m o n s t r a t e d when the same investigation of the interface chemistry was carried out using the technique of Auger electron spectroscopy The results proved to be inconsistent a n d occasionally contradicted the b e h a v l o u r s h o w n up by X P S An example of this is s h o w n in Figure 2 where the f o r m a t i o n of an a l u m l n l u m electrode is m o n i t o r e d by b o t h AES and XPS It is obvious that the interface mixing process IS being influenced by the incident electron b e a m The disorder introduced by the electron b e a m may be due, at least partially, to localized heating of the surface as a result of the very low thermal conductivity of the CdTe (0 058 W c m - ~ deg ~ at 32 7 ' C 15) This disorder can lead to a considerable shift in the Fermi level at the surface of the crystal pinning it in the lower part of the b a n d gap Using U P S we have noted 8 that this shift can be as m u c h as 0 6 eV It must therefore be stressed that great care must be exercised In the

(a]

//=

8

(b)

8

Evaporahon hme F~gure 2. Deposluon profiles of alummlum deposmng on the vacuum cleaved (110) face of CdTe as measured by (at XPS, (b) AES Key • = AI, O=Te, × =Cd

interpretation of data resulting from experimental techniques where electron beams are incident on the surface of the CdTe

Discussion O u r results show that for C d T e there is very good general agreement between the Schottky barrier characteristics measured by C V/I V techniques and by U P S O n e of the main differences between the b e h a v l o u r of CdTe a n d other materials IS in the rate of development of the Schottky barrier If we c o m p a r e the f o r m a t i o n of gold contacts to CdTe and the III V c o m p o u n d G a A s the difference in b e h a v l o u r is illustrated F o r G a A s the Schottky barrier is largely formed at a coverage of ~ 20°/. of a m o n o l a y e r 16 F o r C d T e this is certainly not the case a n d m u c h h~gher metal coverages are needed to produce the m a x i m u m shift in the Fermi level at the surface F r o m previous discussion it is clear that the metal CdTe interfaces are not atomically a b r u p t In fact of all the III V and II VI c o m p o u n d s studied by us CdTe has by far the most unstable surface It is clear that electron beams disrupt the surface, as does the metal deposition itself O n e problem that contributes to this effect IS the low thermal conductivity of CdTe The formation of metal CdS and metal CdSe interfaces, b o t h related II VI matermls, has been described by Brucker a n d Brlllson ~v It was noted that. like CdTe, interfaces are not a b r u p t a n d that the interface widths can be considerable Earlier Brlllson ~ noticed a relationship between the heat of reaction (AHR) of the metal with the semiconductor material a n d the Schottky barrier height (~bb)for a range of systems These heats of reaction are determined, for the reaction

M+-

1 x

C d T e ~ -1 ( M x T e ) + 1 Cd x x

from heats of formation (AH/) values of the c o m p o u n d semiconductor CdTe and the most stable metal tellurlde product TM These are normalized per metal atom, analogous to the AH R calculations of Andrews and Phllhps 19 The heats of reaction of various metals with C d T e are tabulated in Table 1 F r o m Brlllson's results 11 it appears that there exists two definite regions of Schottky barrier b e h a v l o u r F o r unreactlve metals one generally obtained high Schottky barriers, whereas reactive metals produced lower Schottky barrier b e h a v l o u r A s h a r p transition between these two types of b e h a v l o u r occurred at a critical heat of reaction F o r CdS a n d CdSe the critical heats of reaction are nearly identical at ~ + 0 5 eV metal a t o m -1 In Figure 3 we plot AH R against 4~hfor various metals on CdTe The value o f A H R for Au is not known, as the heat of formation of the stable gold tellurtde (AuTe2) is not d o c u m e n t e d F r o m the results it does appear that those metals with high heats of reaction (greater than 0 30 eV metal a t o m 1) lead to Schottky barriers Those metals with heats of reaction less t h a n this figure give very low or ohmic b e h a v l o u r If, however, one takes into account the heat of c o n d e n s a t i o n 2° of the metal then the relahonshlp between the heat of reaction and Schottky barrier height disappears However it is not really clear how meaningful the heat of c o n d e n s a t i o n is in these systems as the metal a t o m s are not simply forming metallic overlayers with a b r u p t Interfaces on impinging on the CdTe surface Brlllson 2~ has s h o w n that the most reactive metals, those with negative heats of reaction, form the most a b r u p t interfaces As most of the metals investigated by us, on CdTe, have positive heats of reaction, the exceptions being In and AI which have slightly negative heats of reaction, then the fact that the interfaces between the metals and

641

M H Patterson

and R H W//hams

NI Sn,

I

In. Ag

at metal-CdTe

I.

\

I Ocz I I I/ I

I

I

1 1

I

I

I

,

-5

barrws

x

cu

I I

Al.

Schottky

-4

Heat of reactlon

(AH,)

I

I

__I,

-I

0

/eV

metal

I I

2

atom-’

Fqyre 3. Bdrrler heights (&) correlated to heats of red&Ion (AH,) for various metals on CdTe 0 = Heats of redctlon of metal + CdTe substrate x =Heats of reactlon of metal +CdTe substrate, with hedt of metdl condensation Included

the CdTe are not abrupt Is consistent with the observdtlons of Brlllson on other IILVI compounds It has also been suggested by Brdlson that surface fields may lead to enhanced dlffuslon across these Interfaces In the metalLCdTe systems fields can be very high (up to 10’ V cm-‘) and It 1s clear that these may have a slgmficant effect on the dlffuslon These ideas ~111 be pursued m d later pubhcatlon In many of the theories which attempt to describe the basic physlcal processes responsible for Schottky barrier formation workers attempt to mterpret the way m which the index of mterface behavlour, S, varies from one metalLsemlconductor system to another and m particular the sharp transition m values of S which seems to be near to zero for metals on covalent semiconductors and clove to unity for metals on lomc \emlconductors22 The quantity, S, IS obtained from the dependence of & on the metal work function, $,,,, usmg a relatlonshlp of the form22

where 4,, 1s the semiconductor work function dnd C 1s a constant Kurtm et alz2 and later Schluter23 report that the value of S for CdTe 1s - 0 2 which IS somewhat surprlsmg m view of its Ionic nature It 1s clear that the values of barrier heights for various metals on CdTe correspond to a wide range of values This 1s not consistent with a low value of S for CdTe as was previously suggested This ~111 be considered later It 1s of Interest to probe the influence of defects m the metal-CdTe system It has been suggested that for the metalL -V systems, amon and cation vacdncles play an important role and that the resultant Schottky barriers can be related to an excess of anions or catlons mcorporated m the metal electrode Clearly for CdTe there appear to be defects at the interfaces but there does not seem to be any such simple reldtionship between the Schottky barrier behavtour dnd the excess of Cd or Te m the metal electrode as determined by XPS It 1s known, however, that the defect structure m CdTe IS complex Daw and Smlth2“ have calculated the defect levels due to simple neutral amon and c&on vacancies m the bulk and near the surface of several 111-V and II VI compound semiconductors The calculated levels fall within the band gap for the IIILV mater& and there seems a clear correlation between the Schottky barrier formation and the existence of these defects For CdTe, however, it seems that the correspondmg levels fdll well outside the band gap and as such may be meffectlve m pmnmg the Fermi level near the Interface and thus m Influencing the Schottky 642

mterfaces

barrier formation Hence although tn CdTe there are many defects at metal CdTe Interfaces they do not have such an mfluence on the Schottky barrier behavlour as 15 seen m metal- IIILV systems In addltlon to out-diffusion ofCd and Te atoms there 1~also the posslblhty of m-dlffuslon of metal atoms and we must therefore consider the effect of these dtoms d? dopdnts of the surfdce layer Doping of CdTe with In and Al can produce highly n-type material In the bulk In and Al are known to act as shallow donor\ lying at 0 014 eV below the conductlon band2’ Au, Ag and Cu, however, produce deep acceptor levels, when mtroduced as dopdnts tnto bulk CdTe, dt 0 3 0 4 eV above the valence band Clearly if Au does form deep acceptors then this could explain the high Schottky barrier It would then be surprlsmg thdt Ag did not behave m a 5lmdar manner, if m-dlffuslon and dopmg were controllmg the barriers Hence it appears that there IS no general reldtionship between the Schottky barrier and dopmg by the mdiffused metal atoms, dlthough m mdlvldual cases, In, Al dnd Sn, doping effects may be notrceable These cases ~111be consldered later If one assumes that there are no mtrmslc surface states present m CdTe26, then m conslderatlon of the Schottky model, one can achieve, m general, good agreement between the metal work function (4,) and the Schottky barrier height (4,,) For metals of high 4, we generally see high barriers, an example of this 1s Au Metals with work functions less than the electron affinity for CdTe (XcdTr=4 28 eV2’) show very low barrier or ohmic behavlour Examples of this behavlour are In, Al and Ag If we discount the result for Nl, then all the other metals so far studied on CdTe adhere to the Schottky model, with d value of S, the index of Interface behavlour close to umty The values of 4, used here are dll for polycrystdlline metdls2* However it 1s not clear, m view of the amount of mtermlxmg that occurs at the met&CdTe mterfaces,Just how relevant the values of 4, are It 1qpossible that the metal deposttlon procedure may cause changes m these metal work functions In, Al and Sn are particularly mterestmg On deposltlon of very small amounts of metal, the Fermi levels shift up the band gap near to the conduction band edge This 1s consistent with the metal atoms acting as surface or bulk shallow donors If we assume m-dlffuslon of metal atoms to n distance of the order of the escape depth glvmg rise to shdllow donors then the results achieved for the three metals cdn be explained by the Fermi level at the surface being locdted very near to the conduction bdnd edge, d\ represented by the inset m Figure l(b) As the metal contact further grows on the surface charge transfer now occurs from these shallow donors to the surface metal resulting in the Fermi level movmg back down the band gap For In and Al the Fermi level moves close to Its orlgmal posltlon to form ohmic contact further grows on the surface, charge transfer now occurs band gap to form a low Schottky barrier, consistent with the Schottky model predlctlons The reason for Nl being mconslstent with the Schottky model IS, as yet, unknown There 1s d need to study further the interfaces formed between transltlon metals and the vacuum cleaved (110) face of CdTe, to further probe the vahdlty of the Schottky model with respect to CdTe

Conclusions 1 The interfaces

cleaved

formed between various metals and the vacuum (110) face of CdTe are non abrupt, with Interface

M H Patterson and R H Wtlhams Schottky bamers at metaI-CdTe interfaces w i d t h s m s o m e cases distributed

over s o m e h u n d r e d s of

angstroms 2 All metal overlayers on C d T e are p o l y c r y s t a l h n e m n a t u r e 3 Defects do not a p p e a r to play an i m p o r t a n t role in c o n t r o l l i n g Schottky barrier formatton 4 Despite the presence of n o n a b r u p t interfaces a n d the associated defects the m o d e l ~ h l c h best fits the e x p e r i m e n t a l results for metal C d T e s y s t e m s is the S c h o t t k y model, with a value of S, the index of interface b e h a v l o u r , close to untty

Acknowledgements M H P a t t e r s o n wishes to t h a n k the E u r o p e a n Research Office of the U n i t e d States A r m y for s u p p o r t

References 1 j Bardeen, Pins Rel, 71,717 (1947) 2 V Heine, Pin's Ret, AI38, 1689 {19651 C R Crowell, J Va¢ S~l Tecllnol, I1, 1951 119741 4 j Inkson, J Ph~s C, 6, 1350 (19731 s S G Lotue and M L Cohen, Pins Rel, BI3, 2461 119761 " P W Chye, I Lmdau P Pmncetta, C M Garner, C Y Su and W E Splcer, Pins Rel, BIB, 5545 119781 " R H Wdhams V Montgommy and R R Varma, J Vat 5~t lethnol, 16, 1418 (1979}

T P Humphreys M H Patterson and R H Williams, I t'a~ ";el lectlnol, 17, 886 (1980) o W E Splcer, P W C h y e , P R Skeath, C Y S u a n d 1 Llndau 3 l~a( 5~z Fechnol, 16, 1422 (1979) 1o R H Wdhams, 1981 (in press) ~1 L J Bnllson, R Z Bachrach, R S Bauer and J C McMenamln, Phl'~ Ret Left, 42, 397 (1979) 2 V Montgomery, A McKinley and R H Williams, 5url ~,~t, 89, 635 (1979) ~3 R R Varma, M H Patterson and R H Wdhams, J P h ~ D 12, L71 (1979) 14 N J Shexchik, J Teleda M Cardona and D W Langer, Phi ~ brat 5olldl, B59, 87 119731 is N Klausutls, J A Adamskl, C V Colhns M Hunt, H L~pson and J R Welner, J Ele~ttml ,~,lafel 4, 625 (19751 t~, W E Splcer 1 Llndau, P Skeath and C Y Su, J 1 a~ 5,~1 7e~ ttnol, 17, 1(/19 (19801 1- C F Brucker and L J Bnllson, J I a~ ~,~t lechllol (in press) is D D Wagmam W H E',ans, V B Parker I Halow, S M Bade2, and R H Schumm, NBS Technical Notes 270 3 (19681 and 270 4 (19691 t g j M Andre,as and J C Phflhps, Plns Rel Left 35, 56 11975} _,0 Heat,, of fo, marion of gaseous atoms from elements m thmr standard states, ( RC ttamlbool, ol Clleolt~tr~ aml Ph~ Sl~ ~ 581h edn p F 230 -'~ L J Bnllson (m press) e_, S Kurtm T C McGIll and C A Mead Pin s Rel Lelt, 22, 14z~3 (19691 2~ M Schhlter, Phys Rc~, BIT, 5044 {19781 24 M S Day, and D L Smith, 4ppl PtlI~ Lefts 36, 690 (19801 2S K Zanlo, '~etllt~omlu¢ tOl s and 5emmletal~ Vol 13 Cadmmm 1ellurlde Academic Press Nev, Yolk (19781 2~, E J Mele and J D Joannopoulos, Pins Rel B17, 1528 119781 2- R K Swank Plns Ret 153, 844 11967) 2s H B Mlchaelson, 3 4ppIPhts, 48, 4729119771

643