- Email: [email protected]
Cadmium oxidation in different environments: an XPS study
Journal of Electron Spectroscopy and Rekzted Phenomena. 60 (1992) 37&383 Elsevier Science Publishers B.V.,. Amsterdam
Cadmium oxidation an XPS study
375
in different environments:
S. Ciampi, V. Di Castro, C. Furlani and G. Polzonetti Department of Chemistry. University “La Sapienza”, P.k A. MOFO5, 00185 Rome (Italy) (First received
17 February
1992; in final form 30 April 1992)
Abstract The behaviour of polycrystalline cadmium in different oxidative atmospheres was investigated. The interactions between cadmium and 0,, CO,, water, and mixtures of these were studied. The cadmium sample was only reactive towards oxygen. A progressive growth of the 01s signal as a function of 0, exposure was observed, together with a shift in the Cd3d and Cd MNN peaks towards the positions reported for CdO. The 01s spectra show the presence of several components in the range of exposures analysed and the best fit of the experimental signals is obtained for the components at 529.3eV and 531.3 eV. The signal at lower binding energy is characteristic of 02- in the oxide state, while the component at higher binding energy could be due to either chemisorbed oxygen or oxygen in a non-equivalent site in the oxide structure.
INTRODUCTION
The mechanism of oxidation reactions has been studied in detail for many transition metals by using X-ray photoelectron spectroscopy (XPS) and other surface techniques, but little work has been done on cadmium, although this element is of great technological interest. The most usual application of cadmium is in the field of corrosion: cadmium is easily passivated and is thus used in thin layers on iron and other substrates to prevent corrosion processes. Some results on cadmium oxides and on oxygen chemisorption on cadmium surfaces using different electron spectroscopic techniques such as XPS, UPS, AES, EELS have been reported [l--5]. All the results show the growth of more than one oxygen species on the cadmium surface, a behaviour similar to that observed for the Zn-0 system [4,6-g], and several hypotheses have been proposed regarding the chemical nature of these oxygen species. Correspondence to.- V. Di Castro, Universit$. Chim., P.le A. Moro, I 00185, Roma, Italy.
degli Studi di Roma “La Sapienza”,
0368~2048/92/$05.00
Publishers
0
1992 Elsevier
Science
B.V. All rights reserved.
Dipto.
di
376
S. Ciampi et al./J. Electron
Spectrosc.
Relat. Phsnom.
60 (1992) 375-383
Analysis of the valence-region photoemission of the Cd-O system shows a regime (isolated oxygen incorporated distinction between a low-exposure
in the surface) and a high-exposure regime (oxide formation) [l]. Exposure of a cadmium sample to low amounts of oxygen gives rise to a 01s signal at 530.9 eV; further oxygen exposures result in a second peak at 528.6.eV, which is the predominant signal at high exposure [5] and characteristic of an oxide state. Analysis of the 01s signal, originated by cadmium exposure to air, shows the progressive growth of two 01s signals at 529.4 and 531.6 eV [2]. The first of these signals has been assigned to oxide, while the second has been proposed as due to a different species, probably hydroxide. The present paper reports the results of an XPS and AES investigation on the reactivity of polycrystalline cadmium exposed to controlled amounts of pure 0,, CO,, water, and mixtures of these. These reagents were chosen with the aim of clarifying whether species other than 0, can react with the cadmium surface. In fact in a relatively narrow binding-energy region it is possible to find 01s signals arising from carbonate, hydroxide and other species [2,4,5]. The results obtained in this work are compared with those obtained previously in our laboratory for the oxidation of cadmium in air [Z]. EXPERIMENTAL
The cadmium polycrystalline sample (Ventron 99.99%) was scraped and polished before introduction into the spectrometer. Whilst in the instrument preparation chamber, the sample was cleaned by means of argon ion sputtering (2 keV, 10 PA) for 90 min before each experiment. The cleanliness was checked by means of XPS and the absence of signals due to oxygen or carbon was confirmed. The gases were then admitted in controlled amounts into the preparation chamber through a stainless steel leak valve. The 0, gas (Mattheson research purity) flowed over a liquid nitrogen trap before introduction into the preparation chamber. The water used was twice distilled and was degassed. The exposures were made at a pressure of 1 x 10m6Torr and reported in langmuirs (1 L = 1 x 10m6Torr x 1 s). XPS measurements were made using a VG ESCA3 MKl spectrometer equipped with a non-monochromatised Al Ka,% source (hv = 1486.6 The pressure in the analysis chamber was below 7 x lo-’ Torr. The energy values reported herein are averages of at least two runs the tabulated values are accurate to f 0.2eV. The curve fitting was formed using an IBM AT computer and a combination of Gaussian Lorentzian curves. RESULTS
eV). and perand
AND DISCUSSION
The oxidation
of polycrystalline
cadmium
in different
environments
was
S. Ciampi et d/J.
Electron Spectrosc. Relat. Phenom. 60 (1992) 375-383
377
h
e
d
(x3)
b
(X3)
a
535.0 525.0 530.0 BINDING ENERGY (eV) Fig. 1. 01s photoemission spectra at different oxygen exposures:(a) 6OL; (b) 1OOL; (c) 2OOL; (d) 400 L; (e) 600 L; (f) SO0 L, (g) 2000 L; (h) IOmin at atmospheric pressure.
investigated using XPS and AES. The interactions between cadmium and 0,, CO,, degassed water, mixtures of 0, and water, mixtures of CO, and water, and mixtures of 0, and CO, were studied. The results obtained for the oxidation of polycrystalline cadmium by 0, were compared with those obtained in previous research work in which the cadmium was exposed to air [2]. The changes in the 01s spectral region after exposure of the cadmium sample to increasing amounts of pure 0, are shown in Fig. 1. In order to obtain a better understanding of the oxidation mechanism occurring at the sample surface, the Cd3d and CdM,N,,,~iV,,~ signals were also analysed. The 01s spectra reported in Fig. 1 clearly show the presence of several components, due to different species of oxygen, in the exposure range analysed. The best fit of the oxygen signals was obtained for the components at 529.3 eV and 531.3eV on the binding energy (BE) scale. As an example, Fig. 2 shows the fitting performed for the 01s experimental signal at three different 0, exposures and it can be seen that there is no change
S. Ciampi et al-/J. Electron Spectrosc.
Relat. Phenom.
60 (1992) 375-383
(b)
525.0
530.0 BlNDfNG
535.0 ENERGY
t eV
1
Fig. 2. Curve fitting of the 01s photoemission peaks for cadmium exposed to the following amounts of oxygen: (a) 60 L, (b) 600 L, (c) 10 min at atmospheric pressure.
in the position of the two peaks during oxidation. The two oxygen signals were attributed to two distinct species of oxygen which arise together as shown in Fig. 3(a). In this figure the Ols/Cd3d,, area ratio is plotted against the oxygen exposure for the two 01s peaks, and the oxygen at lower BE is always the main component with an intensity of more than twice the second one. The oxidation reaction is fast up to an exposure of 400L 0,; it then slows down, probably because at this exposure a passivated layer is formed on top of the cadmium surface. However, in all the range of exposure analysed the growth of the two oxygen signals takes place simultaneously, with a higher rate for the peak at lower BE. The signal at 529.3eV is in the energy range characteristic for 02- in the oxide state [4] and, therefore, can be attributed to oxygen which has reacted with the metal to form CdO. The 01s component at 531.3 eV could be due to the presence of different oxygen species, because signals due to either hydroxyl groups, defective oxides, or oxygen chemisorbed on the surface have been reported at similar BE values [2,4,5]. The results obtained in the present work are similar to those obtained for the interaction between air and cadmium in a previous investigation [2], where a clean cadmium sample was exposed to air and composite 01s
S. Ciampi et d/J.
Electron
Spectrosc.
Rehzt. Phenom.
60 (1992) 375-383
379
(a) IO-
I 0
1000
I
I
2000 Oxygen
Exposure
2000 Air
I
3000
Exposure
4000
,I 11
I 1Atmx
10’
( Langmuir)
3000
( Langmuir)
Fig. 3. (a) 01s/Cd3&,2 area ratios as a function of oxygen exposure. (b) Ols/Cd3d,, area ratios as a function of air exposure. Key: A, oxygen component at lower BE; l, oxygen component at higher BE.
signals were detected. The best fit for the resulting 01s signal was obtained for the two peaks at 529.4 + 0.2 eV and 531.6 f 0.2eV, which is similar to the results obtained in the present work for a cadmium sample exposed to 0,. The area ratios of the two 01s components are very similar for exposure
380
S. Ciampi et al.jJ. Electron Spectrosc. Relat. Phenom. 60 (1992) 375-383
of the cadmium sample to air and to oxygen. In order to show this similarity, Fig. 3(b) shows the results for a clean cadmium sample exposed to air [2]. Comparing the trends reported in Fig. 3 and considering the dilution of 0, in air, the origins of the two components seem to be the same, although the species associated with the peak at lower BE seems to be favoured by exposure to pure oxygen. For cadmium exposed to air [Z], the peak at higher BE is most likely due to hydroxyl groups because of the presence of water in the air, but this is not likely to be the case for the sample oxidised by pure 0,. No 01s signals were observed at 533.4eV, the reported binding energy value for the 01s signal due to adsorbed molecular water [4], in either of the two experiments. In order to investigate further the presence of hydroxyl groups in the oxidised surface, we studied the behaviour of the cadmium sample towards water. The clean cadmium surface was exposed to increasing amounts (from 12 to 815OL) of degassed water in order to evaluate whether a reaction takes place under these conditions. All exposures were done at a pressure of 1 x 10-“Torr, except for the last exposure which was obtained by exposing the sample to H,O vapours at 760 Torr for 10min. No signal was detected in the 01s energy region, i.e. the sample resulted unreactive in such experimental conditions. As CO, adsorption often occurs on the surface of metal samples, the presence of this molecule was investigated by exposing the clean cadmium sample to increasing amounts (50 to 2000L) of CO, at pressures as high as 2 x 10-6Torr. Under the conditions used no signal due to CO, was observed in the 01s or Cls BE regions. Various metal oxide surfaces readily physically and/or chemically adsorb reactive molecules like water and carbon dioxide, and reactivity is sometimes a function of the degree of oxidation [lo]. In order to estimate the rectivity of the oxidised cadmium surface, four samples at different stages of oxidation (clean cadmium + 8OL, + 200 L, + 400 L and + 2000 L of 0,) were successively exposed to increasing amounts of water (50 to 5800 L). No variation in the cadmium spectra or the 01s signals was detected after these treatments. A similar experiment was done with CO, exposing each of the four oxidised cadmium samples to increasing amounts of CO, (50 to 1OOOL)and, as for the experiment with water, no change was seen in the spectra of the cadmium oxidised surface after CO, exposure. Although inactive alone, water and CO, could play a role in the oxidation process when combined. To determine the effect of these compounds we exposed the clean cadmium sample to the following mixtures: oxygen and water, oxygen and carbon dioxide, and carbon dioxide and water. The cadmium sample only reacted with the mixtures containing oxygen (O,/H,O and O,/CO,) and the same behaviour as that observed for pure 0, was seen. The above results indicate that, under the experimental conditions used,
S. Ciampiet al&J. Electron Spectrosc. Relat. Phenom.60 (1992)375-383
381
cadmium is reactive only towards 0,, and thus the formation of hydroxyl or carbonyl species on the cadmium surfaces can be excluded. In order to obtain further information and to make a better estimate of the type of oxidation occurring on the cadmium surface, the cadmium 3d signals were analysed. These peaks show an anomalous and characteristic negative BE shift from cadmium metal to CdO. It was found that as the exposure of the clean cadmium sample to oxygen is increased the Cd3d,,, peak shifts from 405.0 to 404.7eV BE and the full width at half maximum increases from 2.0 to 2.4eV, while the kinetic energy of the CdM,N&Y,,, Auger peak decreases from 383.7 to 382.4eV. Although the observed shifts appear anomalous for an oxidation process, this is the expected trend for the cadmium oxidation, and indicates a transition from cadmium metal to cadmium oxide, the negative chemical shift being characteristic of Cd0 formation [3]. Analogous values have been reported previously for cadmium exposed to air [2]. Thus the experimental data on the 01s and Cd3d,,, signals show that cadmium is progressively oxidised to Cd0 by oxygen, and during this reaction two oxygen containing species are formed. The 01s peak at lower BE is characteristic of oxygen in an oxide state, while the signal at higher BE could be due to oxygen bonded to cadmium in a different way. As we can exclude the formation of hydroxyl groups on the cadmium surface, the higher BE 01s peak could be due either to chemisorbed oxygen or to defective oxide. Previous UPS and EELS results 111have shown an initial chemisorption regime followed by progressive formation of a Cd0 structure, with spectra similar to the ones reported for bulk oxide. In a XPS study of cadmium exposed to different amounts .of oxygen the growth of two 01s signals has been reported and attributed to the formation of two distinct bulk oxide phases; the species associated with 01s at higher BE was assigned to an excess lattice oxygen anion [SJ. As cadmium and zinc (group IIb) show similar behaviour towards oxygen and their oxidised surfaces exhibit similar 01s signals, in order to elucidate the cadmium oxidation mechanism, it seems reasonable to make a comparison with the data reported in the literature for the oxidation of zinc. The oxidation of zinc surfaces involves the formation and growth of multilayer oxide islands, where oxygen incorporation at the perimeter of the islands is the rate-limiting step as observed by XPS [6,9] and Hz absorption techniques [9]. The growth of two 01s signals during the oxidation process has been reported [6-g]: a main peak at about 530.0eV with a partially resolved shoulder at about 532.0eV. Moreover, the relative intensity of the two components does not change noticeably on increasing the 0, exposure. The peak at 530.0eV has been assigned to oxygen species in the ZnO structure. The shoulder at higher BE has been attributed to
S. Ciampi et al./J. Electron Spectrosc. Relat. Phenom. 60 (1992) 375-383
382
oxygen chemisorbed on the patches of clean metal between the oxide islands [7,9] although alternative hypothesis such as chemisorption on both metal and oxide and the formation of a non-oxide-like oxygen within the ZnO layer have not been ruled out [6,7]. For cadmium and zinc it is reasonable to suppose a similar growth of the oxidised layer with the formation of multilayer oxide islands and adsorption of oxygen on the clean cadmium between the oxide patches. On the grounds of this hypothesis the two 01s peaks at 529.3 eV and 531.3 eV could be attributed to Cd0 and adsorbed oxygen respectively. In the present study, however, the BE positions of the two oxygen components and their corresponding area ratios were constant over the range of exposures analysed, showing that the two oxygen species are formed simultaneously. This trend seems more in agreement with the hypothesis that the oxygen enters non-equivalent positions in the Cd0 structure, perhaps forming a defective oxide where the excess oxygen in a Cd0 modified structure gives rise to the 01s signal at higher BE. CONCLUSIONS
A polycrystalline cadmium sample was exposed to different oxidative atmospheres, namely 02, H,O, CO, and mixtures of these, at room temperature. The cadmium only reacted with oxygen with no change being observed after exposure to other gases even at high exposure. Progressive growth of the 01s signal as a function of 0, exposure was observed, together with a shift in the Cd3d and CdMNN peaks towards the positions reported for Cd0 in the literature. In all the analysed range of exposures the 01s spectra showed several peaks and the best fit was obtained for the components at 529.3 and 531.3eV. The intensities of the two 01s components both increase on increasing exposure to 0,, and their area ratios remain constant. The oxygen signal at lower BE is characteristic for 02- in an oxide structure, while the component at higher BE can be due only to chemisorbed oxygen or oxygen in a non equivalent site in the oxide structure, because under the experimental conditions used the formation of hydroxyl groups can be ruled out. On the basis of the constancy of both the peak position and the area ratio of the two oxygen signals we suppose that a defective Cd0 structure, in which oxygen atoms are present in two non-equivalent. situations, is formed on top of the cadmium sample. Oxygen in the stoichiometric oxide is characterised by the peak at lower BE and the excess oxygen is characterised by the peak at higher BE. REFERENCES 1
L. Braicovich,
G. Rossi, R.A. Powell
and W.E.
Spicer,
Phys. Rev. B, 21 (1980) 3539.
S. Ciampi et al-/J. Electron Spectrosc. Relat. Phenom. 60 (1992) 37&383 2 3 4 5 6 7 8 9
10
383
S. Ciampi, V. Di Castro and G. Polzonetti, J. Electron Spectrosc. Relat. Phenom., 56 (1991) Rl. S.W. Gaarenstroom and N. Winograd, J. Chem. Phys., 67 (1977) 3500. K. Wandelt, Surf. Sci. Rep., 2 (1982) 1. J.S. Hammond, SW. Gaarenstroom and N. Winograd, Anal. Chem., 47 (1975) 2193. D. Briggs, J. Chem. Sot. Faraday Discuss., 60 (1975) 81. CR. Brundle and R.I. Bickley, J. Chem. Sot., Faraday Trans. 2, 75 (1979) 1030. G.E. Hammer and R.M. Shemenski, J. Vat. Sci. Technol. A, 1 (1983) 1026. L. Chan and G.L. Griffin, J. Vat. Sci. Technol. A, 3 (1985) 1613. C.T. Au, A.F. Carley and M.W. Roberts, Philos. Trans. R. Sot. London, Ser. A, 318 (1986) 61.
Cadmium oxidation an XPS study
375
in different environments:
S. Ciampi, V. Di Castro, C. Furlani and G. Polzonetti Department of Chemistry. University “La Sapienza”, P.k A. MOFO5, 00185 Rome (Italy) (First received
17 February
1992; in final form 30 April 1992)
Abstract The behaviour of polycrystalline cadmium in different oxidative atmospheres was investigated. The interactions between cadmium and 0,, CO,, water, and mixtures of these were studied. The cadmium sample was only reactive towards oxygen. A progressive growth of the 01s signal as a function of 0, exposure was observed, together with a shift in the Cd3d and Cd MNN peaks towards the positions reported for CdO. The 01s spectra show the presence of several components in the range of exposures analysed and the best fit of the experimental signals is obtained for the components at 529.3eV and 531.3 eV. The signal at lower binding energy is characteristic of 02- in the oxide state, while the component at higher binding energy could be due to either chemisorbed oxygen or oxygen in a non-equivalent site in the oxide structure.
INTRODUCTION
The mechanism of oxidation reactions has been studied in detail for many transition metals by using X-ray photoelectron spectroscopy (XPS) and other surface techniques, but little work has been done on cadmium, although this element is of great technological interest. The most usual application of cadmium is in the field of corrosion: cadmium is easily passivated and is thus used in thin layers on iron and other substrates to prevent corrosion processes. Some results on cadmium oxides and on oxygen chemisorption on cadmium surfaces using different electron spectroscopic techniques such as XPS, UPS, AES, EELS have been reported [l--5]. All the results show the growth of more than one oxygen species on the cadmium surface, a behaviour similar to that observed for the Zn-0 system [4,6-g], and several hypotheses have been proposed regarding the chemical nature of these oxygen species. Correspondence to.- V. Di Castro, Universit$. Chim., P.le A. Moro, I 00185, Roma, Italy.
degli Studi di Roma “La Sapienza”,
0368~2048/92/$05.00
Publishers
0
1992 Elsevier
Science
B.V. All rights reserved.
Dipto.
di
376
S. Ciampi et al./J. Electron
Spectrosc.
Relat. Phsnom.
60 (1992) 375-383
Analysis of the valence-region photoemission of the Cd-O system shows a regime (isolated oxygen incorporated distinction between a low-exposure
in the surface) and a high-exposure regime (oxide formation) [l]. Exposure of a cadmium sample to low amounts of oxygen gives rise to a 01s signal at 530.9 eV; further oxygen exposures result in a second peak at 528.6.eV, which is the predominant signal at high exposure [5] and characteristic of an oxide state. Analysis of the 01s signal, originated by cadmium exposure to air, shows the progressive growth of two 01s signals at 529.4 and 531.6 eV [2]. The first of these signals has been assigned to oxide, while the second has been proposed as due to a different species, probably hydroxide. The present paper reports the results of an XPS and AES investigation on the reactivity of polycrystalline cadmium exposed to controlled amounts of pure 0,, CO,, water, and mixtures of these. These reagents were chosen with the aim of clarifying whether species other than 0, can react with the cadmium surface. In fact in a relatively narrow binding-energy region it is possible to find 01s signals arising from carbonate, hydroxide and other species [2,4,5]. The results obtained in this work are compared with those obtained previously in our laboratory for the oxidation of cadmium in air [Z]. EXPERIMENTAL
The cadmium polycrystalline sample (Ventron 99.99%) was scraped and polished before introduction into the spectrometer. Whilst in the instrument preparation chamber, the sample was cleaned by means of argon ion sputtering (2 keV, 10 PA) for 90 min before each experiment. The cleanliness was checked by means of XPS and the absence of signals due to oxygen or carbon was confirmed. The gases were then admitted in controlled amounts into the preparation chamber through a stainless steel leak valve. The 0, gas (Mattheson research purity) flowed over a liquid nitrogen trap before introduction into the preparation chamber. The water used was twice distilled and was degassed. The exposures were made at a pressure of 1 x 10m6Torr and reported in langmuirs (1 L = 1 x 10m6Torr x 1 s). XPS measurements were made using a VG ESCA3 MKl spectrometer equipped with a non-monochromatised Al Ka,% source (hv = 1486.6 The pressure in the analysis chamber was below 7 x lo-’ Torr. The energy values reported herein are averages of at least two runs the tabulated values are accurate to f 0.2eV. The curve fitting was formed using an IBM AT computer and a combination of Gaussian Lorentzian curves. RESULTS
eV). and perand
AND DISCUSSION
The oxidation
of polycrystalline
cadmium
in different
environments
was
S. Ciampi et d/J.
Electron Spectrosc. Relat. Phenom. 60 (1992) 375-383
377
h
e
d
(x3)
b
(X3)
a
535.0 525.0 530.0 BINDING ENERGY (eV) Fig. 1. 01s photoemission spectra at different oxygen exposures:(a) 6OL; (b) 1OOL; (c) 2OOL; (d) 400 L; (e) 600 L; (f) SO0 L, (g) 2000 L; (h) IOmin at atmospheric pressure.
investigated using XPS and AES. The interactions between cadmium and 0,, CO,, degassed water, mixtures of 0, and water, mixtures of CO, and water, and mixtures of 0, and CO, were studied. The results obtained for the oxidation of polycrystalline cadmium by 0, were compared with those obtained in previous research work in which the cadmium was exposed to air [2]. The changes in the 01s spectral region after exposure of the cadmium sample to increasing amounts of pure 0, are shown in Fig. 1. In order to obtain a better understanding of the oxidation mechanism occurring at the sample surface, the Cd3d and CdM,N,,,~iV,,~ signals were also analysed. The 01s spectra reported in Fig. 1 clearly show the presence of several components, due to different species of oxygen, in the exposure range analysed. The best fit of the oxygen signals was obtained for the components at 529.3 eV and 531.3eV on the binding energy (BE) scale. As an example, Fig. 2 shows the fitting performed for the 01s experimental signal at three different 0, exposures and it can be seen that there is no change
S. Ciampi et al-/J. Electron Spectrosc.
Relat. Phenom.
60 (1992) 375-383
(b)
525.0
530.0 BlNDfNG
535.0 ENERGY
t eV
1
Fig. 2. Curve fitting of the 01s photoemission peaks for cadmium exposed to the following amounts of oxygen: (a) 60 L, (b) 600 L, (c) 10 min at atmospheric pressure.
in the position of the two peaks during oxidation. The two oxygen signals were attributed to two distinct species of oxygen which arise together as shown in Fig. 3(a). In this figure the Ols/Cd3d,, area ratio is plotted against the oxygen exposure for the two 01s peaks, and the oxygen at lower BE is always the main component with an intensity of more than twice the second one. The oxidation reaction is fast up to an exposure of 400L 0,; it then slows down, probably because at this exposure a passivated layer is formed on top of the cadmium surface. However, in all the range of exposure analysed the growth of the two oxygen signals takes place simultaneously, with a higher rate for the peak at lower BE. The signal at 529.3eV is in the energy range characteristic for 02- in the oxide state [4] and, therefore, can be attributed to oxygen which has reacted with the metal to form CdO. The 01s component at 531.3 eV could be due to the presence of different oxygen species, because signals due to either hydroxyl groups, defective oxides, or oxygen chemisorbed on the surface have been reported at similar BE values [2,4,5]. The results obtained in the present work are similar to those obtained for the interaction between air and cadmium in a previous investigation [2], where a clean cadmium sample was exposed to air and composite 01s
S. Ciampi et d/J.
Electron
Spectrosc.
Rehzt. Phenom.
60 (1992) 375-383
379
(a) IO-
I 0
1000
I
I
2000 Oxygen
Exposure
2000 Air
I
3000
Exposure
4000
,I 11
I 1Atmx
10’
( Langmuir)
3000
( Langmuir)
Fig. 3. (a) 01s/Cd3&,2 area ratios as a function of oxygen exposure. (b) Ols/Cd3d,, area ratios as a function of air exposure. Key: A, oxygen component at lower BE; l, oxygen component at higher BE.
signals were detected. The best fit for the resulting 01s signal was obtained for the two peaks at 529.4 + 0.2 eV and 531.6 f 0.2eV, which is similar to the results obtained in the present work for a cadmium sample exposed to 0,. The area ratios of the two 01s components are very similar for exposure
380
S. Ciampi et al.jJ. Electron Spectrosc. Relat. Phenom. 60 (1992) 375-383
of the cadmium sample to air and to oxygen. In order to show this similarity, Fig. 3(b) shows the results for a clean cadmium sample exposed to air [2]. Comparing the trends reported in Fig. 3 and considering the dilution of 0, in air, the origins of the two components seem to be the same, although the species associated with the peak at lower BE seems to be favoured by exposure to pure oxygen. For cadmium exposed to air [Z], the peak at higher BE is most likely due to hydroxyl groups because of the presence of water in the air, but this is not likely to be the case for the sample oxidised by pure 0,. No 01s signals were observed at 533.4eV, the reported binding energy value for the 01s signal due to adsorbed molecular water [4], in either of the two experiments. In order to investigate further the presence of hydroxyl groups in the oxidised surface, we studied the behaviour of the cadmium sample towards water. The clean cadmium surface was exposed to increasing amounts (from 12 to 815OL) of degassed water in order to evaluate whether a reaction takes place under these conditions. All exposures were done at a pressure of 1 x 10-“Torr, except for the last exposure which was obtained by exposing the sample to H,O vapours at 760 Torr for 10min. No signal was detected in the 01s energy region, i.e. the sample resulted unreactive in such experimental conditions. As CO, adsorption often occurs on the surface of metal samples, the presence of this molecule was investigated by exposing the clean cadmium sample to increasing amounts (50 to 2000L) of CO, at pressures as high as 2 x 10-6Torr. Under the conditions used no signal due to CO, was observed in the 01s or Cls BE regions. Various metal oxide surfaces readily physically and/or chemically adsorb reactive molecules like water and carbon dioxide, and reactivity is sometimes a function of the degree of oxidation [lo]. In order to estimate the rectivity of the oxidised cadmium surface, four samples at different stages of oxidation (clean cadmium + 8OL, + 200 L, + 400 L and + 2000 L of 0,) were successively exposed to increasing amounts of water (50 to 5800 L). No variation in the cadmium spectra or the 01s signals was detected after these treatments. A similar experiment was done with CO, exposing each of the four oxidised cadmium samples to increasing amounts of CO, (50 to 1OOOL)and, as for the experiment with water, no change was seen in the spectra of the cadmium oxidised surface after CO, exposure. Although inactive alone, water and CO, could play a role in the oxidation process when combined. To determine the effect of these compounds we exposed the clean cadmium sample to the following mixtures: oxygen and water, oxygen and carbon dioxide, and carbon dioxide and water. The cadmium sample only reacted with the mixtures containing oxygen (O,/H,O and O,/CO,) and the same behaviour as that observed for pure 0, was seen. The above results indicate that, under the experimental conditions used,
S. Ciampiet al&J. Electron Spectrosc. Relat. Phenom.60 (1992)375-383
381
cadmium is reactive only towards 0,, and thus the formation of hydroxyl or carbonyl species on the cadmium surfaces can be excluded. In order to obtain further information and to make a better estimate of the type of oxidation occurring on the cadmium surface, the cadmium 3d signals were analysed. These peaks show an anomalous and characteristic negative BE shift from cadmium metal to CdO. It was found that as the exposure of the clean cadmium sample to oxygen is increased the Cd3d,,, peak shifts from 405.0 to 404.7eV BE and the full width at half maximum increases from 2.0 to 2.4eV, while the kinetic energy of the CdM,N&Y,,, Auger peak decreases from 383.7 to 382.4eV. Although the observed shifts appear anomalous for an oxidation process, this is the expected trend for the cadmium oxidation, and indicates a transition from cadmium metal to cadmium oxide, the negative chemical shift being characteristic of Cd0 formation [3]. Analogous values have been reported previously for cadmium exposed to air [2]. Thus the experimental data on the 01s and Cd3d,,, signals show that cadmium is progressively oxidised to Cd0 by oxygen, and during this reaction two oxygen containing species are formed. The 01s peak at lower BE is characteristic of oxygen in an oxide state, while the signal at higher BE could be due to oxygen bonded to cadmium in a different way. As we can exclude the formation of hydroxyl groups on the cadmium surface, the higher BE 01s peak could be due either to chemisorbed oxygen or to defective oxide. Previous UPS and EELS results 111have shown an initial chemisorption regime followed by progressive formation of a Cd0 structure, with spectra similar to the ones reported for bulk oxide. In a XPS study of cadmium exposed to different amounts .of oxygen the growth of two 01s signals has been reported and attributed to the formation of two distinct bulk oxide phases; the species associated with 01s at higher BE was assigned to an excess lattice oxygen anion [SJ. As cadmium and zinc (group IIb) show similar behaviour towards oxygen and their oxidised surfaces exhibit similar 01s signals, in order to elucidate the cadmium oxidation mechanism, it seems reasonable to make a comparison with the data reported in the literature for the oxidation of zinc. The oxidation of zinc surfaces involves the formation and growth of multilayer oxide islands, where oxygen incorporation at the perimeter of the islands is the rate-limiting step as observed by XPS [6,9] and Hz absorption techniques [9]. The growth of two 01s signals during the oxidation process has been reported [6-g]: a main peak at about 530.0eV with a partially resolved shoulder at about 532.0eV. Moreover, the relative intensity of the two components does not change noticeably on increasing the 0, exposure. The peak at 530.0eV has been assigned to oxygen species in the ZnO structure. The shoulder at higher BE has been attributed to
S. Ciampi et al./J. Electron Spectrosc. Relat. Phenom. 60 (1992) 375-383
382
oxygen chemisorbed on the patches of clean metal between the oxide islands [7,9] although alternative hypothesis such as chemisorption on both metal and oxide and the formation of a non-oxide-like oxygen within the ZnO layer have not been ruled out [6,7]. For cadmium and zinc it is reasonable to suppose a similar growth of the oxidised layer with the formation of multilayer oxide islands and adsorption of oxygen on the clean cadmium between the oxide patches. On the grounds of this hypothesis the two 01s peaks at 529.3 eV and 531.3 eV could be attributed to Cd0 and adsorbed oxygen respectively. In the present study, however, the BE positions of the two oxygen components and their corresponding area ratios were constant over the range of exposures analysed, showing that the two oxygen species are formed simultaneously. This trend seems more in agreement with the hypothesis that the oxygen enters non-equivalent positions in the Cd0 structure, perhaps forming a defective oxide where the excess oxygen in a Cd0 modified structure gives rise to the 01s signal at higher BE. CONCLUSIONS
A polycrystalline cadmium sample was exposed to different oxidative atmospheres, namely 02, H,O, CO, and mixtures of these, at room temperature. The cadmium only reacted with oxygen with no change being observed after exposure to other gases even at high exposure. Progressive growth of the 01s signal as a function of 0, exposure was observed, together with a shift in the Cd3d and CdMNN peaks towards the positions reported for Cd0 in the literature. In all the analysed range of exposures the 01s spectra showed several peaks and the best fit was obtained for the components at 529.3 and 531.3eV. The intensities of the two 01s components both increase on increasing exposure to 0,, and their area ratios remain constant. The oxygen signal at lower BE is characteristic for 02- in an oxide structure, while the component at higher BE can be due only to chemisorbed oxygen or oxygen in a non equivalent site in the oxide structure, because under the experimental conditions used the formation of hydroxyl groups can be ruled out. On the basis of the constancy of both the peak position and the area ratio of the two oxygen signals we suppose that a defective Cd0 structure, in which oxygen atoms are present in two non-equivalent. situations, is formed on top of the cadmium sample. Oxygen in the stoichiometric oxide is characterised by the peak at lower BE and the excess oxygen is characterised by the peak at higher BE. REFERENCES 1
L. Braicovich,
G. Rossi, R.A. Powell
and W.E.
Spicer,
Phys. Rev. B, 21 (1980) 3539.
S. Ciampi et al-/J. Electron Spectrosc. Relat. Phenom. 60 (1992) 37&383 2 3 4 5 6 7 8 9
10
383
S. Ciampi, V. Di Castro and G. Polzonetti, J. Electron Spectrosc. Relat. Phenom., 56 (1991) Rl. S.W. Gaarenstroom and N. Winograd, J. Chem. Phys., 67 (1977) 3500. K. Wandelt, Surf. Sci. Rep., 2 (1982) 1. J.S. Hammond, SW. Gaarenstroom and N. Winograd, Anal. Chem., 47 (1975) 2193. D. Briggs, J. Chem. Sot. Faraday Discuss., 60 (1975) 81. CR. Brundle and R.I. Bickley, J. Chem. Sot., Faraday Trans. 2, 75 (1979) 1030. G.E. Hammer and R.M. Shemenski, J. Vat. Sci. Technol. A, 1 (1983) 1026. L. Chan and G.L. Griffin, J. Vat. Sci. Technol. A, 3 (1985) 1613. C.T. Au, A.F. Carley and M.W. Roberts, Philos. Trans. R. Sot. London, Ser. A, 318 (1986) 61.