Surface oxidation of cadmium, indium, tin and antimony by photoelectron and Auger spectroscopy

Applications of Surface Science 10 (1982l 63-74 North-Holland Publishing Company

63

SURFACE OXIDATION OF CADMIUM, INDIUM, TIN AND A N T I M O N Y BY P H O T O E L E C T R O N A N D A U G E R S P E C T R O S C O P Y *

P. SEN, M.S. H E G D E and C.N.R. R A O ** Solid State and Structural Chemistry Unit. Indian Institute of Science. Bongal,re 560012, /mha

Received 26 June 1981; revised manuscript received I October 1981

Surface oxidation of Cd, In, Sn and Sb has been in,estigated by employing ',alcncc band~. metal 4d levels, and plasmon bands in X-ray pho!oelcctron spectra. O(KI_.L), metal M4N4sN..~, and plasrnon transitions in electron-induced Auger spectra as well as Auger transitiom, due to the metal (metal oxide) and plasmoi~sin X-ray-induced Auger spectra. The surface oxides are In ,O~, CdO and a mixture of SnO and SnO2 in the case of In, Cd and Sn respectively. The facility of surface oxidation is found to vary as In>Cd>Sn>Sb. Inter-atomic Auger transitiom, involxing oxygen valence bands have been identified on oxidized surfaces of Cd and In. I. Introduction

Surface oxidation of metals has been fruitfully studied by photoelectron spectroscopy, A u g e r electron spectroscopy a n d related techniques. In this laboratory, we have been investigating surface oxidation of transition metals for some time [1,_',. Special m e n t i o n m u s t be m a d e of a novel A u g e r method described by us recently, which employs relative intensities of metal Auger lines to s t u d y metal oxidation [I]. In this paper, we report results of a systematic investigation of the surface oxidation of c a d m i u m , indium, tin and a n t i m o n y by eraplc,ying photoelectron and A u g e r spectra. In photoelectron spectra, we have e~.amined the valence bands, core levels and the p l a s m o n bands. Auger spectra obtained both by electron a n d X - , a y excitations were studied. In order to quantify the m a g n i t u d e of surface oxidation, we have followed the variati,3n in intensity of the metal 4d, oxygen (Is), and p l a s m o n peaks in XPS, O ( K L L ) a n d metal M4N4sN45 lines in the electron-induced A u g e r spectra, a n d metal a n d metal oxide lines in the X-ray-induced Auger spectra. T h e s t u d y has enabled us to c o m p a r e the results from different m e t h o d s of evaluating surface oxidation of metals besides illustrating the use of p l a s m o n b a n d s in XPS a n d A u g e r spectra for the s t u d y of surface oxidation. In the case of Cd a n d In, it has been possible to show that A u g e r transitions involving oxygen valence b a n d s occur in oxidized surfaces. * Contribution No. 127 from the Solid State ar.d Structural Chemistry Unit. ** To whom all correspondence should be addressed. 0 3 7 8 - 5 9 6 3 / 8 2 / 0 0 0 0 - 0 0 0 0 / $ 0 2 . 7 5 © 1982 N o r t h - H o l l a n d

64

P. Sen et al, /

Surface "~xidation of Cd) hi, St) and Sb

2. Experimental Photoelectron spectra and Auger electron spectra were recorded using an ESCA-3 Mark-ll spectrometer of VG Scientific Ltd., UK, fitted with a sample preparation chamber. He I and AI K a radiations were used for obtaining the photoelectron spectra. Auger spectra obtained by bombardment with 4 k V electrons and also by X-ray radiation were recorded. High purity ( > 99.99%) vacuum deposited, polycrystailine films of cadmium, indium, tin and antimony were employed for the study. The metal surfe, ces were cleaned by etching with argon ions using a gun current of 100 #A through an accelerating voltage of 4 I:V. The samples were heated to 100°C after etching and cooled to laboratory temperature. Surface purity of the samples were checked using both XPS and AES. The samples were exposed to oxygen after each etch cleaning in the sample preparation chamber, monitoring the pressure by a nude ionization gauge. Spectra were recorded at a base pressure of 5 X 10-~° Torr in the spectrometer chamber. Oxygen exposures are expressed in langmuir~ (L) (1 L = 10 ~6Torrs),

3. Results and discussion Valence bands ,of all the metals studied show the expected changes on exposure to oxygen. In fig. l we show the XPS valence band of In and the He I valence band of Sn at different oxygen exposures to illustrate the nature of changes occurring during oxidation. The valence band intensity of In does not show an appreciable change up to an expost~re of 100 L, and is accompanied by the emergence of a broad band around 4 eV due to the O(2p) of the ox,de at higher exposures. At the highest exposure of 106 L where In shows saturation coverage, the metal valence band almost completely disappears and the spectrum resembles that of In20.v The He I valence band of Sn shown in fig. l clearly shows the sharp Fermi edge of the metal. Exposure to oxygen is accorapanied by the appearance of two peaks at 2.6 and 3.7 eV due to oxide formation. The Fermi edge persists even at high oxygen exposures suggesting the presence of a significant proportion of the metal in agreement with the observations of Powell and Spicer [3]. It appears that a p:lre oxide lay.~r is not forni~d on the s-.~rface of Sn; the surface layers are likely to l:e composed of SnO, SnO 2 and Sn as pointed out by earlier workers [3-5]. Intensities of the metal 4d and 3d peaks decrease on exposure to oxygen; in addition, these peaks shift to higher binding energies due to oxidat:.on [6]. In the case of In, the shift is of - I. 1 eV; Cd and Sn show shifts of 0.2 and 0.4 eV respectively. The shift is neg'.!gible in the case of Sb. None of the core levels showed distinct peaks due to th e pure metal and the metal in the oxide upon oxidatior,, bu, they did show a significant increase in their half-widths. In

Surfaceoxida:ion of Cd, In. Sn ,~nd Sh

P. Sen et al. /

//

(b>

(o)

65

AI~3,40~In (~,d)

/ I03L

0"

~00L

poL

I~c''

~// i

0

~

iSSp)

0

5 BE (eV)

5 BE (eV)

J

10

Fig. I. (a) Effect of oxygen exposure on the He 1 valence band of fin, (b Effect of oxygen exposure on the XPS valence band of indium.

table 1, we have shown a few core level binding energies of the metals and their oxides. With progressive increase in exposure to oxygen, we notice the O(Is) peak intensity increases as expected. We could, therefore, follow the oxygen pickup by the surface by monitoring the O(ls) intensit),. Core level spectra (table 1) as well as vale.rice bands show that ln20~ and CdO are formed respectively on the surface (,f In and Cd at moderate to high oxygen exposures. This was confirmed by comparison of the spectra of the oxidized surfaces with those of the pure oxides. Formation of In203 ola the

Table I Metal core level binding energies (in cV) in X-ray photoele.'tmn spectra

Cd CdO In In 2C.~ Sn SnO2

4d

3d (5/2)

3d (3/2~

10.1 10.9 17.1 17.0 24,2 24.5

40a.6 40.~,9 44.~ ,8 445.0 485.0 489.0

411,5 412,9 451,4 452.7 493,5 494,7

66

P. Sen et al. /

Surface oxidation of Cd, hi. Su and Sb

surfaces of In was also suggested earlier from Auger studies [7]. In the case of Sn, t?!,emetal continued to be present even at high oxygen exposures; the oxide lay~:r,~ seem to consist of a mixture of SnO and SnO 2 as su~,gested by Lau a,~d Wertheim [4]. Sb showed no significant oxidation even at high exposures and we were unable to ascertain the nature of the oxide layer. Both the 4d and 3d levels of Cd, In. Sn and Sb show c,~aracteristic plasmon los.,; peaks in XPS. The bulk and surface plasmons of the 4d levels are shown in fig. 2. The position of the bulk plasmon peaks in the four metals shows a systematic variation with the increase in the atomic number. The energies of the surface plasmon peaks are generally related to those of bulk plasmon peaks by a factor of around 0.7. In going from Cd to Sn, w e notice that the intensity of the plasmon peaks increases, with Sn and In showing a larger number of plasmon peaks than Cd. The intensity ratios of the first bulk plasmon and 4d peaks for Cd, ln, Sn and Sb were found to be around 0.14, 0.20, 0,23 and 0.18 respectively. The 4d binding energies of Cd, In, L;n and Sb are 10.1, 17.1, 24.2 4d

81 s

"J

0

I

10

I,

20

]

30 (eV)

;~

60

l

5o

Fig. 2. ~asmon peaks appearing next Io ~he 4,:1 core levels of Cd, In. Sn and Sb. S and B refer Io surface and bulk plesmons respectively.

P. Sen et al. /

Sur[ace oxidation of

Cd. In. Sn and Sb

67

and 31.8eV. In the case of Cd, In and Sn it appears that the plasmon loss intensity increases as the kinetic energy of the ejected photoelectron from the 4d core hole decreases; this, however, does not hold for Sb. The surfa,.~e plasmon peaks are extremely sensitive to the su,-face cleanliness. Expost;re of a clean surface to oxygen readily causes the disappearance of the surface plasmon peaks. The bulk plasmon peaks are also affe,'ted by exposure to oxygen. Although both the surface and the bulk pla:;mon peaks disappear on exposure to oxygen, their rates differ as expected. Surface plasraons disappear faster than the bulk plasmons, as can be seen from figs. 3 and 4, From fig. 4 ~.e see that the plasmon intensities, lp, decrease very fast with oxygen exposure; in the case of In, both the surface and the bulk plasmon intensities go to zero at ~ 104 L. Plasmons intensities in Cd are also sensitive to oxygen exl:osure, but, in Sn the intensity does not go to zero even at l06 L. This is in agreement with the results from our valence band and core level studies. The decrease in plasmon intensity is negligible in Sb. From the intensity variation of the plasmon peaks we conclude that ff,e facility of surface oxidation in the metals studied varies as In ; - C d > Sn > Sb. From fig. 4, we also see that the variation in plasmon intensities is comparable at 300 and 400 K, the variation being only slightly more rapid at the higher temperalure.

4d

I n + O 2 AT 300K

B

j r

\.~..._ ~ I 20

~ I 25

I 30 BE

~

_

1000 L " 105 L I 35

(eV)

Fig. 3, Effect of oxygen exposure on the plasmon pt.aks of In. Variation in the intensity and position of the 4d peaks are also seen.

68

P. Sen ez al. /

Surface oxidation of ('d. In. S,t and Sb

':,0 ',..,

"..'"~-.B

~",, s "~'\, "a'--a..~._-?cC _.~

1).6

3.2

~ "~---~ :~-~_

0.2

zI

Log L

[ 1.0~---,~7~" ' ' ~ ~

/ -- 0,6['~

|

~2[

~',

\,

ce

&"&

"'---8

'a......~

z.I

Log L

..~l"°I"~'-'L~-~-"~- ~ 0.61

Sb

!

012~ l

Log L

Sn

I

I

2

t,

I 6

Log k

Fig. 4. Variation of the bulk (B) and surface (SI plasmon intensities. Ip. on oxygen exposure: ( A ! 400 K; (C)) 300 K. Intensities are rela "~e to zero-exposure intensity.

Bulk plasmon peaks are also seen in the electro:~-mduced Auger spectra of the metals. Fig. 5 shows the electron-in~luced Augei spectra of Cd and In with the plasmon peaks appea~ing around 362 and 386 eV respectively. On exposure to o×yge., these plasmon peaks disappea- accompanied by lhe emergeace of the O(KLL) peak, X-ray-induc,.d Auger spectra of the metals also show bulk plasmon peak.~ as shown in the case of Cd in fig. 6. Plasmon peaks in the Auger spectra, of In, Cd and Sn, have earlier been reported in the literature I7-9]. Cd, In, Sn and Sb show characteristic MNN and MNV transitions in the electron-induced Auger spectra (fig. 5). Weak transitions which appear at higher kinetic energies than M45N45N4s transitions are due to the M4(M)N(M)V(M ) and Ms(M)N(M)V(M ) transitions. In Cd, these peaks are located at 404 and 397 eV respectively, while in In they appear at 425 and 416 eV r~peetively. The difference in energy between these transitions equals the difference in the M 4 and M 5 core levels of the metal. We observe the expected decrease in the intensity of the metal MNN and MNV Auger lines accompanied by the progressive development of the O(KLL) line. On large

KE(eV)

45O

cc

400

/1~'~

II

l

Plosrnon

104L

I

~o~

--

ff

°~_11

(b;

I

Fig. 5. Effect of oxygen exposure on the electron.induced Auger spectra of (a) In, (b) Cd.

500

i KL23L23)

(a)

380 KE (eV;

i

360

Plosmon

=.

>

70

p. Sen et al. /

,,

Surface oxidation of Cd. In. Sn and Sb

/m.._.,(/./A\\

385

380

375

370

365

KE(eV)

Fi&.6. F.ffectof oxygen expesure on d-.e X-ray-indu.~,Jspectra of Cd.

exposure to oxygen we find the appearante ol new peaks in place of the MNV transitions which we ascribe to an inter, atomic Auger process involving the oxygen va~,ence levels. The transition probability of the inter-atomic process is very low compared to the intra-atomic process. However, on oxidation, the probability of occurrence of inter-atomic transitions increases as the metal valence band gets depleted. Such peaks appear at 427 and 396 eV in the case of In and Cd respectively (fig. 5). Occurrence of inter-atomic Auger transitions in transition metal oxides has been discussed by us in another paper from this Laboratory [ 10l. In the X-ray-induced Auger spectrum ,,of In (fig. 7), the energy difference between the two mai.n Auger lines at 402.6 and 410.2eV corresponds to the spin-orbit sp!i',ting of the 3d level; similar spin-orbit splittings are seen in the Auger spectra of thor other metals as well. Besides the spin-orbit splittings, Auger spectra of the four metals show f i e structure. Thus, in the case of In, the peaks of 412.8 and 404.8eV are associated with a 2-hole 4d s final state. The X-ray induced Auger spectra of Cd, In and Sr~ recorded by us are identical to those reported in the literature [7,9,11]. On oxposure to oxygen, the intensities of the metal Auger lines in the X-ray induced spectra decrease accompanied by the appearance of characteristic lines due to the oxides (at lower kinetic energy). The chemical shift values for the M4N4sN45 transitfons are 0.8 3.0, 3.4 and 0.2 eV respectively for Cd, In, Sn and Sb; the eorre,nponding values for the MsN4~N,,s transitions are 1.3, 3.8, 5.5 and 0.5 eV respectively. It must be noted that the Auger chemical shifts are

P. Sen et al. /

Surface oxidation of Cd, In. Sn and Sh

71

~04 C/s

3~104CI s I

,~10

,

,

,

,

I

405

,

I

L

400

395

.

.

.

.

.

I

-

390

K E (eV)

Fig. 7. Effecl cf oxygen exposurc on the X-ray-inducedspectra of In.

very much larger than the chemical shifts of the 3d cole levels in the X-ray photoelectron spectra. The large Auger shitt between the metals and their oxides enables us to employ them for following the oxidat;on process, q-he chemical shifts between the metals and their oxides are generally negligible in transition metals like Fe. Co, Ni and Cu, and this necessitates tile use of metal Auger intensities for studying the oxidation of these -netals [1]. In the case of zinc, however, the chemical shift between the oxide and the metal was found to be qui'e appreciable [12]. Among the four metals studied, the Auger chemical shift is maximum -n tile case of Sn and least in the case of Sb. We hav,~ i'ollowed the oxidation of Cd, In, Sn and Sb by following the variation 2D the following spectral featmes with oxygen exposure; (i) area under the metal 4d peak in the X-ray photoelectron spectra; (ii) intensity of the bulk plasmon; (iii) relative intensities of the Auger lines due to the metal and the oxide in the X-ray-in3uced Auger spectra; (iv) intensity of the metal Auger lines in the X-ray-induced Auger spectra; and (v) ratio of metal and O(KLL) Auger line intensities in electron-induced Auger spectra. The intensity of the bulk plasmon peak was fitted into the formula 1 = .toe-x/x, where I 0 is the intensity when the surface is devoid of oxygen; 1 is a function of x, the thickness of the oxide layer surface, and ~, the mean escape depth. We have estimated [13] ~ to be 19.2 tk after correcting for the geometric

72

Surface oxidation of Cd, In, Sn and Sb

P. 'Jen et a L /

factor arising from the deviation of the angle of collection of electrons from 90 °. Each monolayer has been taken to be 2 .~ thick and the off-normal correction to be 1.25. A similar treatment was employed for the intensity of the ntetal line in the X-ray-induced Auger spectra. The area under the metal 4d peak (in XPS) has contributi, .,s from the sabstrate metal as well as from the layer of oxide in top. If the i:~..-nsity from the oxide layer is ! I and the intensity from the metal is 12, then the total intensity under the peak is 1 = I, + 1:~. where It = l o ( l - - e - x / ; ~ ) ,

12 = iO e - x / x ,

a n d !~ is the intensity of the peak d u e to the m,.~tal in the bulk metal oxide. In order to obtain I °, we have employed density data as follows: 10 --- c l ° (c < 1),

Cd

~5

~s o o

3

In

~'10 ,,,

]

;

.~',' ,';'

/

-~L-_-_-~~:'-:~" 2

;(;"¢' .._.-..o l(;" o ' : I - - a

#i i

,

z,

6

Log L

Sr~

2

ts / /s ,/.,o

3

4-

Sb

0) /,

~2 o

//

(iP

Y.! . I~.

s

L- -".,-7 ,

,o""" o r -C]

,jr"

//

•

6

Log C

1

d"

/

2

~

Log L

t3

Z

4

Log L

Fig. g. Variation in the numb.~r of oxide layers with oxygen exposure obtained from d i f f e r e n t ~..:thods: (()) from bulk pla~mon intensity: (L~) from 4d peak intensity in XPS; (×) relative intensity of metal and metal oxide lines in X-ray-induced Auger spectra: squares, from t h e inteasitie~ of the MdN45N45 line in the X-ray-induced Auger spectra: (0) ratio of O(KL2~L,.0 a ~ M4N45:N,sstransitions in electron-induced Auger spectra.

P. Sen et aL /

73

Surface oxidation of Cd. In. Sn and Sh

where c is given by, M/ 1 M MO : .M/.I~!0 1M /"M/YM ~ ¢X pM/,OMo.

Here, 1 ~ is the intensity from pure metal a n d IM M° the intensity of the metal from the oxide. T h e value of c, calculated from the crystal structure data, was f o u n d to agree with the value obtained from density data (0.816,0.676 an,-I 0.752 for In, Cd and Sn respectively). W e obtain the intensity variation with exposure, L, as I(L):I,

+I 2:lP(l-e-s/x)+I

oe-.~/x ::l~)[c+(l_c)e

,:x],

or

X

--In

I ,. l l(t.)/lO2_

c

.

In the X-ray-ioduced Auger spectra the ratios of the intensities of the metal a n d oxide lines were employed:

12 - Y -

1o

e-,,/x

or

~ = In

.

This equation was also used to analyse O ( K L L ) / M 4 N 4 s N n s A u g e r intensity ratios from electron-induced A u g e r spectra. T h e n u m b e r of oxide layers, x, calculated e m p l o y i n g the above methods, is plotled against oxygen exposure ill fig. 8. W e see that the different m e t h o d s give similar results. This quantitative analysis of the oxidation shows that the m a g n i t u d e of growth of the oxide layers as well as the facility of oxidation vary in order In > Cd > Sn > Sb

Acknowledgement T h e a u t h o r s thank the D e p a r t m e n t of Science a n d Technology and the University G r a n t s C o m m i s s i o n for s u p p o r t of this research.

References [I] C.N.R. Rao, D.D. Sarma and M.S. Hegde, Proc. Roy. SOc. (London) A370 (1980) 269. [2] A. Srinivasan, K. Jagannathen, M.S. Hegde and C.N.R. Rao, Indian J. Chem. IRA (1979) 463 [3] R.A. Powell and W.E. Spicer, Surface Sci. 55 (1976) 681. [4] C.L. Lau and G.K. Wertheium, J. Vacuum Sci. Technol. 15 (I ( I ) 622. [51 M. Negasaka, H. Fuse and T. Yamashina, Thin Solid Films 29 (1975) L29. [6] C.N.R. Rao, D.D. Sarma, M.S. Hegde and S. Vasudevan, Proc. Roy. SOc. (London) A367 (1979) 239. [7] S.M. Rossnagel, H.F. Dylla and S.A. Cohen, J. Vacuum Sci. Technol. 16 (1978) 558.

74 [8] [9] [10] [I II [12] [13]

P. Sen et al. /

Sur/ace oxidation of Cd. In, Spa and Sb

R.A. PowelL Appl. Surface Sci. 2 (1979) 397. W.M. Mulari¢ and W.T. Peria, Surface Sci. 26 (1971) 125. C.N.R. Ran and D.D. Sarma, Phys. Rcv. B, to be published. S,M. Barlow, P. Bz.yat-Mokhtari and T.E. Gallon, J. Phys. CI2 (1979) 5574. D.D. Sating, M.S. Hegde and C.N.R. Rao, Chem. Phys. Letters 74 (1980) 443. C.C. Chang. Surface Sci. 48 (1975) 9.