X-ray photoelectron spectroscopy, fourier transform infrared spectroscopy and electrical conductivity studies of copper phosphate glasses

IOUlltlA£ 0 ¥

ELSEVIER

Journal of Non-Crystalline Solids 185 (1995) 101-108

X-ray photoelectron spectroscopy, Fourier transform infrared spectroscopy and electrical conductivity studies of copper phosphate glasses M.A. Salim a, G.D. Khattak a, * M. Sakhawat Hussain b a Department o[Physics, King Fahd University of Petroleum and Minerals, Dhahran 31261, SaudiArabia b Department of Chemistry, King Fahd University of Petroleum and Minerals, Dhahran 31261, Saudi Arabia Received 9 February 1994; revised manuscript received 17 November 1994

Abstract Phosphate glasses containing CuO with composition (CuO)x(P2Os)a-x, where x = 0.1, 0.2, 0.3, 0.4 and 0.5, were studied by X-ray photoelectron spectroscopy (XPS), Fourier transform infrared spectroscopy (FT-IR) and room temperature electrical conductivity, o'. The spin orbit components Cu 2P3/z and Cu 2pl/2 show doublet structures which may be associated with Cu + and Cu 2+. The ratio C = Cu+/Cutotal as a function of x was determined using the Cu 2p3/2 spectra. It is observed that the ratio, C, initially decreases with x, becomes minimum at x = 0.3 and then increases. Although the variation of o- with C was very small, it is close to the uncertainty in the measurement of its value; nevertheless it was maximum at C = 0.5, in agreement with the previous prediction. The FT-IR spectra indicated a maximum shift in the P = O absorption at 30% CuO contents in the glasses, which essentially follows the same pattern as the shift for the P - O - P absorption. Further, the development of the high energy component peak in the O ls (associated with P - O - P oxygen) spectra follows a trend similar to the 760 c m - I (also associated with P - O - P ) absorption band in the Fr-IR spectra.

1. Introduction Oxide glasses containing transition metal ions were reported as early as 1954 to have semiconducting properties [1]. Since then, a great deal of work has been carded out on semiconducting glasses [2-5]. The general condition for this semiconducting behavior is that the transition-metal ion should exist in more than one oxidation state, so that conduction can take place by the transfer of electrons from the low

* Corresponding author. Tel: +966-3 860 2255. Telefax: + 966-3 860 2293. E-mail: [email protected].

to high valence states. According to this condition, the dependence of the electrical conductivity on the ratio of the ion concentration in the low valence state to the total concentration of transition metal ions, C = Cu+/Cutotal, should be such that the maximum conductivity is at C-= 0.5. In fact, among all the glasses which have been so far investigated, maximum conductivity at C = 0.5 occurs only in iron phosphate glass [2-5]. In the present work, phosphate glasses containing CuO with composition (CuO)x(P2Os) 1-x, where x = 0.1, 0.2, 0.3, 0.4 and 0.5 were studied using X-ray photoelectron spectroscopy (XPS), Fourier transform infrared spectroscopy (FT-IR) and electrical conductivity, tr. It was found that variation in tr with C

0022-3093/95/$09.50 © 1995 Elsevier Science B.V. All fights reserved

SSDI 0 0 2 2 - 3 0 9 3 ( 9 4 ) 0 0 6 8 3 - 0

102

M.A. Salim et aL / Journal of Non-Crystalline Solids 185 (1995) 101-108

was small, but in agreement with the above prediction.

2. Experimental procedures The glasses were prepared by melting dry mixtures of CuO and P205 in alumina crucibles with composition (CuO)x(P2Os) l_x, where x = 0.1, 0.2, 0.3, 0.4 and 0.5. All chemicals used in this study were reagent grade. The crucible and contents were placed in a furnace, heated at 300°C for an hour prior to melting the mixture in order to minimize volatilization. The crucible was then transferred to a melting furnace maintained at a temperature of 1330° C. The melt was left for about 4 h under atmospheric conditions in the furnace. During this time the liquid was occasionally stirred with an alumina rod. The homogenized melt was then cast onto a stainless steel plate mold. The samples were disk-shaped with diameters of 1.5 cm and thicknesses of about 3 mm. The samples needed for electrical conductivity measurements were ground and polished with diamond paste down to a minimum grit size of 0.1 p,m. DC conductivity measurements at room temperature were made in a two-probe configuration using silver paint for electrical contacts. The specimens were circular disks of ~ 2 cm diameter and a thickness of 2-3 mm with silver paint electrodes (3 mm diameter) at the center of each face. Thus, the bulk path length for the current flow was shorter than the surface path lengths. Currents were measured using an electrometer Keithley 610C. The XPS measurements were carried out with a spectrometer (V.G. Scientific ESCALAB MKII) equipped with a dual aluminum-magnesium anode. Details of the system are given elsewhere [6]. The energy scale of the spectrometer was calibrated using Cu 2p3/2 = 932.4 eV and the energy separation between Cu 2p3/2 and Cu 2pl/2 was 19.8 eV. The charging of non-conducting glass samples was avoided by flooding the sample with a separate source of low-energy electrons. The energy and intensity of these external electrons were adjusted to obtain the best resolution as judged by the narrowing of the full width at half maximum (FWHM) of photoelectron peaks. It was found that, at the opti-

mum settings of the neutralizing gun (electron kinetic energy between 5 and 10 eV, electron emission current at the sample between 1 and 5 nA), the position of the adventitious C ls line was within + 0.5 eV of 284.6 + 0.2 eV. This peak arises due to hydrocarbon contamination and its binding energy is generally accepted as remaining constant irrespective of the chemical state of the sample. For the sake of consistency, all energies are reported with reference to the C ls transition at 284.6 + 0.2 eV. For XPS measurements, the standard oxide powder samples were embedded in substrates of indium foil supported by metallic backing. The samples were loaded through a fast entry airlock into a preparation chamber and finally into the analysis vessel. The base pressure in the analysis chamber during these measurements was less than 5 × 10 -11 mbar. Infrared spectra were recorded with a 5DXD spectrophotometer (Nicholet FT-IR or Perkin-Elmer FTIR) in the range 350-4600 cm -1. Sample pellets for various glasses were prepared for FT-IR measurements by mixing approximately 10-15 mg of the glass powder with about 150-160 mg of spectroscopic grade dry KBr powder and grinding the mixture to a very fine powder and then compressing the mixture in an evacuable die under 20 000 lb pressure. The KBr powder was dried at 110° C and allowed to cool in a vacuum desiccator prior to making the pellets, to avoid adsorption of the moisture.

3. Results 3.1. Core levels Cu 2p Core level spectra of the Cu 2p for CuO and Cu phosphate glasses of different CuO concentration are shown in Fig. 1. The Cu 2p spectrum for the Cu phosphate glasses exhibits the spin-orbit components Cu 2P3/2 and Cu 2pl/2 at binding energies of ~ 932.5 and 952.6 eV, respectively with satellites at about 10 eV higher binding energy. Further, the 2p peaks show doublet structure (peaks marked A and B in Fig. 1). The separation between the peaks A and B is about 2.4 eV. A similar doublet structure has been observed in the Cu 2p spectrum of high-Tc superconductors [7], where the authors have assigned peak A to Cu 2+ and peak B to Cu 3÷. However, we believe

M~A. Salim et al./ Journal of Non-Crystalline Solids 185 (1995) 101-108

H

Z D

E w H rr

5 IZ I11 t.Z H

025.

0315.

BINDING

9411.

ENERGY

006.

01115.

(eV)

Fig. 1. Core level spectra of the Cu 2p in CuO powder and Cu phosphate glasses.

that peak A may be associated with Cu ÷ and peak B with Cu 2÷ for the reasons discussed below. Monovalent and divalent are the only two oxidation states in which Cu exists in various glasses [8]. Further, it is well known that copper compounds containing Cu 2+ are associated with strong satellites (also one can see in the present work in the CuO spectra, Fig. 1), while compounds with Cu ÷ have no satellites [9-12]. A decrease in the intensity of the satellite peak is observed upon going from CuO to Cu phosphate glass, which suggests that some of the Cu 2÷ present in the CuO are reduced to Cu ÷ in the glass. Moreover, the satellite structure (Fig. 1) clearly follows the trend of peak B, i.e., the intensity/peak height of the satellite increases with an increase in the height of peak B. Thus, it may be concluded that the peaks A and B are associated with Cu ÷ and Cu 2÷, respectively. It is known [13] that the presence of non-equivalent atoms of the same element in a solid gives rise to core-level peaks with measurably different binding energies. Non-equivalence of atoms can be a

103

result of: (a) difference in formal oxidation state, (b) difference in molecular environment, etc. The binding energy increases with an increase in the oxidation state of a metal atom. In situations where the formal oxidation state is the same, the general rule is that the binding energy of the core-level of the metal atom increases as the electronegativity (electron withdrawing power) of attached atoms or groups increases [13]. The oxidation states of copper either remain unchanged or reduced to lower valence states, while going from metal oxide (Cu 2÷ ) to glass (Cu ÷, Cu2÷). According to this, the energy of the Cu 2p level should either remain unchanged or decrease upon going from oxide to glass, contrary to the observed increase in energy (peak B). Therefore, the process (a) alone cannot account for the observed shifts. Perhaps formation of Cu phosphates (process (b)) could be the dominant process responsible for the higher energy shift of the Cu 2p levels (peak B). For example, the Cu 2p3/2 in CuO is reported to be at 933 eV, in CuSO 4 at 935.5 eV and in CuF 2 at 937 eV [14]. It may be mentioned that the group electronegativity of the (PO4) 3- ion is close to that of the (SO4) 2- ion [15]. Therefore, following from the above, the Cu 2p3/2 in copper phosphate is expected to have an energy close to that of the CuSO 4. Indeed this is the case as discussed above. Surface contamination, such as formation of carbonates and hydroxide over the glass surfaces, could also lead to a shift of the metal 2p level to a higher energy with respect to that of the standard oxide. However, the absence of the higher energy components of the C ls rules out any significant amount of carbonate presence over the surfaces of the samples. Mclntyre and Cook [16] have carried out an XPS study on the oxides and hydroxides of divalent Co, Ni and Cu. In going from an oxide to hydroxide, they observed shifts to higher energies for the metal 2p levels: Co (1.0 eV), Ni (1.6 eV) and Cu 0.6 eV). In the present work the shifts in the 2p levels for the divalent ions, in going from an oxide to glass are much larger (2.1 eV). Therefore, the observed shifts in glass may not be due to the formation of hydroxide, Further, the absence of component peaks due to the hydroxides in the 2p spectra rules out any significant amount of hydroxide that may have been present on the sample surfaces. In the light of the above discussion, the shift of peak B to higher energy

104

M.A. Salim et al. /Journal of Non-Crystalline Solids 185 (1995) 101-108 8.00

x10" 6.40

1

4.80

~ 3.20 1.60

g~l~

g34.00

93S.00 Ener~ {eV)

9&2.00

g&6.O0

Fig. 2. Cu 2p3/2 spectra for x = 0.3. The points are experimental data. Fitted components and their sums are represented by the continuous curves.

relative to the same for CuO is most likely due to the formation of copper phosphate. Phosphate may also be formed by Cu ÷ but in the case of monovalent Cu the shifts due to a change of molecular environment are known to be small ( < 0.7 eV) compared with as high as 4 eV in divalent copper [14,17]. A deconvolution method in which the Cu 2p spectrum is assumed to be composed of two overlapping peaks was undertaken. Each component peak in the spectrum was fitted to a sum of weighted Lorentzian-Gaussian peaks with a linear sloping background by means of a least-squares fitting program [18]. A best fit to the experimental data was found by varying the peak position, width and intensity of each of the two component peaks. This is shown only for x = 0.3 in Fig. 2 and, using these peaks areas, the ratio of Cu ÷ ions was calculated as follows: Cu + at (area under peak A) =AA; Cu 2÷ ct (area under peak B + area under satellite peak) = A B; Then, C ~ Cu+/(Cu+ + Cu 2+) = A A / ( A A +AB). The ratio C is shown in Fig. 3, as a function of glass composition, x. The validity of the above relation was checked by taking XPS spectra for the Cu20 and CuO powders, and then a mixture of the two. For a mixture which contained equal concentration of Cu + and Cu 2+, we obtained, using the above procedure, a value of C = 0.45, instead of 0.5. The commercial Cu20 (reagent grade) used in the present work was sup-

plied by BDH Laboratory and the manufacturer claims that the powder contains minimum 94% C%O, and some CuO (which is not more than 5%). Also, in the XPS spectrum of the Cu20, a weak satellite of core level Cu 2p was observed, confirming the presence of CuO. It may be noted that in pure Cu20 no satellite has been observed [9-12]. Therefore, the lower value of C observed, may partly be due to the CuO content in the Cu20 powder and partly due to uncertainty in the XPS measurements. However, it seems that the above procedure may yield values of C with an uncertainty of less than 10%. It is observed (Fig. 3) that the ratio C initially decreases with x, becomes minimum at x = 0.3 and then increases. The room temperature conductivity, or, is maximum at C = 0.5, in agreement with the previous prediction.

i

J

i

i

~o

~o

~o

66 62 6O

~e u

S8 54

50

o EI,O -r "~0.5

!

~0.0

"uLO m N ~0.5

o z

~o

~o

C~O Coneen~'afl~ x(°/d Fig. 3. Plots o£ the ratio C ~Cu+/Cuto~al, FWHM of the O Is peak (normalized) and dc electrical conductivity at room temperature, cr (normalized), versus CuO concentration, x. The lines are drawn as guides for the eye.

M.A. Salim et al. /Journal of Non-Crystalline Solids 185 (1995) 101-108

105

3.2. Core levels 0 Is In Fig. 4, spectra for the core level O ls for the glasses are compared with that of P205 powder. The O ls spectrum for the P205 consists of two overlapping peaks. It is well known that the orthorhombic P205, being based on an assembly of discrete molecules, consists of P O 4 tetrahedra shared by three neighboring tetrahedra; the fourth oxygen atom is considered to be double bonded to the phosphorus atom [19]. Therefore, in Fig. 4 the lower binding energy peak may be associated with non-bridging oxygen (P = O, bonding), and the high energy peak with the bridging oxygen (P-O-P). The ratio of the area of the P = O and the P - O - P peaks, if our peak assignment is correct, should be 1:1.5. For every phosphorus atom there should be one non-bridging oxygen atom and 1.5 bridging oxygen atoms. As seen from Fig. 4, the ratio of the two peaks seems to be not too far from the expected value. The above assignment of the peaks can be justified from the well-known research results on sodium silicate glasses in which the bridging oxygen atoms are covalently bonded to two Si atoms, while the nonbridging oxygen atoms are covalently bonded to one Si atom and ionically bonded to one Na ion. It is known that the charge density on the non-bridging oxygen atom is about - 0 . 9 6 and the charge density on the bridging oxygen atom is -0.65 [20]. This charge difference is reflected in a shift of the non-

528

530

532

534

536

538

BINNING EI~RGY leVI

Fig. 4. Core level spectra of the O ls for glasses and P205 powder.

bridging atom peak (O ls) towards a lower binding energy with respect to the bridging oxygen atom peak (O ls) of about 2.3 eV [20]. In the present case, the corresponding shift in the P205 is about 1.8 eV (Fig. 4).

Table 1 Characteristic infrared absorptions peaks (cm-1) for phosphate glasses doped with different amounts of CuO along with the spectrum of p:05 a Sample

P = O ReL

PO-

P-O-P

PO43-

P205

1246 (st) [211 1205 (st)

1115-925 (st)

760 (sh)

490 (sp)

1112-915 (st)

735 (st)

490 (sp)

1129-915 (st)

735 (st)

502 (sp)

1118-926 (st)

762 (st)

512 (sp)

1100-928 (st)

766 (st)

495 (sp)

1085-925 (st)

761 (st)

523 (sp)

(CuO)o.l(P205)o.9

b

(CuO)o.2(P205)o.8

1197 (st) b

(CuO)o.a(P205)o. 7

1276 (st) b

(CuO)o.4(P205)o. 6

1265 (st) b

(CuO)o.5(P205)o. 5

1174 (st) b

asp, sharp; st, strong; sh, shoulder; db, doublet. b This work.

106

M.A. Salim et al. /Journal of Non-Crystalline Solids 185 (1995) 101-108

IO0

1--

20

°+(oo

iloo

looo

moo

ioo

&oo

W A V E R E R (Cm-t )

Fig. 5. FT-IR spectra of P205 and two representative CuO-doped glass samples in the range 400-1400 cm-1, showing shifts in the P = O, P - O - and P-O-P vibrations.

A careful examination of Fig. 4 indicates some differences in the O ls spectra in different composition and in particular the variation in the high energy component (marked by an arrow in Fig. 4). The FWHM of the O ls peak (Fig. 3) increases with concentration, becomes maximum for x = 0.3 and then starts decreasing. 3.3. FT-IR spectra The FT-IR spectra of KBr pellets of glasses doped with different amounts of CuO are shown in Fig. 5 and the corresponding peak positions are listed in Table 1 along with those for the P205. It is well known that orthorhombic P205, rather being based on an assembly of discrete molecules, consists of PO 4 tetrahedra shared by three neighboring tetrahedra; the fourth oxygen atom is considered to be double bonded to phosphorus atom [19]. On this basis, the absorption spectra of P205 can be approximately assigned as follows on the basis of the assignments reported earlier [21]: PO43- group, 490-505 cm -1, strong-broad band; P - O - P , ring frequency, 760-765 cm -1, weakbroad absorption; P - O - , stretching vibrations, 910-1030 cm -~, strong doublet; P = O, double stretching vibrations, 1245 cm -~, broad-band. The absorption band observed at 760-765 cmis quite broad in PzO5 whereas it was not at all observed in the IR spectra of the crystalline phases

of other phosphates such as VPO 5 [22], hydrated Fe3(PO4)2, Ni3(PO4)2, Cu3(PO4) 2 [23] and anhydrous holmium phosphates and lutetium phosphate [24]. We repeated the spectrum of anhydrous P205 and also re-examined the spectra of NaH2PO 4, NasP3Ol0 and Ca3(PO4)2; in agreement with the earlier reports [21-24], we observed no bands in the 760-765 cm-1 region in these phosphates. Hence, it could be concluded that 760-765 cm -1 band is associated with P - O - P vibrations in PzO5 only and phosphates in general do not show such an absorption band. All CuO-doped P205 glasses revealed an enhanced absorption band around 760 cm-1 (Table 1) as compared with a weak and broad band in anhydrous P205. This essentially indicates that CuO has some interactions with the P - O - P vibrations in the copper phosphate glasses. A downward shift of about 30 cm-a for 10-20% CuO glasses was observed in the P - O - P vibrations in the CuO-doped glasses. This absorption was observed at 735 cm-1 as compared with 760 cm -1 in P205. When CuO contents are increased to 30% or above, the energy for the P - O - P absorption reached the value observed in anhydrous P205. Similar spectral changes were observed by Sayer and Mansingh [2] for phosphate glasses containing iron, cobalt, nickel and copper. The P - O - and P = O absorptions were broad and shifted downward in the spectra of glasses as compared with that of P205. This shift in the P = O vibrations is highest at 30% CuO contents in the glasses and essentially follows the same pattern as the shift for the P - O - P band. The P - O - absorption in the glasses is rather broader than in P205. This is expected because the presence of the metal ions in the glass matrix will affect the P = O as well as P - O - frequencies as a result of weak interactions of the metal with the available bonding sites in the P205 phosphate. The resulting shifts in the IR absorptions in the P205 should be a function of the amount of the metal ions present. The strong and sharp PO43- group frequency observed at 500 cm -1 in P205 was changed to a much broader band in the case of CuO glasses. However, no significant shifts in the frequency of the PO43- group were observed, suggesting very little effects, if any, of CuO on the PO43- group frequency.

M.A. Salim et al. /Journal of Non-Crystalline Solids 185 (1995) 101-108

4. Discussion It is observed (Fig. 3) that the ratio C initially decreases with x (the concentration of Cu in glass), becomes minimum at x---0.3 and then increases. Although the variation in room temperature conductivity, o-, with C was very small, it is close to the uncertainty in the measurement of its value; nevertheless it was maximum at C = 0.5, in agreement with the previous prediction. As discussed below, the XPS and FT-IR investigations suggest some compositional dependent changes in the bondings of oxygen in the glass. These changes may arise due to some structural changes in the glass. The O ls peaks for the glasses may be composed of more than one component peak, which may correspond to different bonding modes of the oxygen such as P - O - , P - O - P and P = O ~ Cu [25]. As mentioned above, the energy separation between the O Is components associated with P = 0 and P - O - P is about 1.8 eV for P205. The O Is component associated with P = O--, Cu will probably be located in between the locations of P = 0 and P - O - P components [25]. In fact, Goldman's [26] study of sodiumiron silicate glasses has concluded that the O ls component associated with Si-O ~ Fe 2÷ and Si-O --* Fe 3+ is nearly coincident with non-bridging (Si O Na) peak. Therefore, it may be possible that in the present case the O ls components of P - O ~ Cu and P - O - are close to each other. The O ls components of P - O - P , P - O - and P - O ~ AI have been observed at binding energies of 534, 532.3 and 531.8 eV, respectively, in sodium phosphate glasses doped with Al203 [25]. The build-up of higher energy component in the spectra suggests the possible bonding of the oxygen to be P - O - , P - O - C u and P - O - P . Since the binding energies of the core level O ls for the oxygen in the P - O - and the P - O ~ Cu are close to each other, the XPS fails to make a clear distinction between the two modes of bonding. On the other hand, FT-IR can easily distinguish between these two bonding modes because the absorption peaks corresponding to the two modes are well separated (Table 1). The IR results suggest the presence of all three oxygen bonding modes in this type of glass. The FT-IR observations (Fig. 5 and Table 1) are also in agreement with the XPS results, especially

107

with the O ls spectra (Fig. 4). The magnitude of the high energy component peak (associated with P - O - P oxygen) in the O ls spectra follows a trend similar to the 760 cm-~ absorption band (also associated with the P - O - P ) in the IR spectra. For the O ls spectra (Fig. 4), unambiguous fitting of three component peaks, corresponding to the P - O - , P - O - C u and P - O - P , may not be easy. However, the FWHM of the O ls spectra (treated as a single peak) may provide a rough guide to the variations in the individual component peaks. The variation of the FWHM of the O ls spectra (Fig. 3) with the composition of the glass may possibly be due to some compositiondepende.nt structural changes in the glass. These structural changes may also have some effect on the electrical conductivity of the glass. In the present work, no satisfactory explanation for the physical relationship between the electrical conductivity and the FWHM of the O ls peak was found.

5. Conclusions The XPS, FT-IR and room temperature electrical conductivity, tr, investigations were carried out on copper phosphate glasses. It is observed that the ratio C initially decreases with x, becomes minimum at x = 0.3 and then increases. It was found that the variation in tr with C was small but in agreement with the previous predictions. The FT-IR spectra support the findings of XPS investigations, as some changes in the bondings of oxygen in the glass with composition have been observed. These changes may also have some effect on the electrical conductivity of the glass. However, no satisfactory explanation for this was found. The authors wish to acknowledge the support of KFUPM Research Committee, Chemistry and Physics Departments. The assistance of Mr M.A. Khan for experimental work is appreciated.

References [1] E.P. Denton, H. Rawson and J.E. Stanworth, Nature 173 (1954) 1030.

108

M.A. Salim et al. / Journal of Non-Crystalline Solids 185 (1995) 101-108

[2] M. Sayer and A. Mansingh, Phys. Rev. B6 (1972) 4629. [3] D.L. Kinser and L.K. Wilson, in: Recent Advances in Science and Technology of Materials, Vol. 1, ed. A. Bishay (Plenum, New York, 1973) p. 77. [4] L. Murawski, C.H. Chung and J.D. Mackenzie, J. Non-Cryst. Solids 32 (1979) 91. [5] G.H. Chung and J.D. Mackenzie, J. Non-Cryst. Solids 42 (1980) 357. [6] E.E. Khawaja, Z. Hussain, M.S. Jazzar and O.B. Dabbousi, J. Non-Cryst. Solids 93 (1987) 45. [7] A. Balzarotti, M. DeCrescenzi, N. Motta, F. Patella and A. Sgarlata, Phys. B38 (1988) 6461. [8] C.R. Banford, Colour Generation and Control in Glass (Elsevier, Amsterdam, 1977). [9] G.A. Vernon, G. Stucky and T.A. Carlson, Inorg. Chem. 15 (1976) 278. [10] K.S. Kim, J. Electron Spectrosc. Rel. Phenomen. 3 (1974) 217. [11] Sven Larsson and M. Bragga, Chem. Phys. Lett. 28 (1977) 596. [12] G. Van der Laan, C. Westra, C. Haas and G.A. Swatzky, Phys. Rev. B23 (1981) 4369. [13] D. Briggs and J.C. Rivi~re, in: Practical Surface Analysis by Auger and X-ray photoelectron spectroscopy, ed. D. Briggs and M.P. Seah (Wiley, Chichester, 1988) p. 119.

[14] C.D. Wagner, in: Practical Surface Analysis by Auger and X-ray photoelectron spectroscopy, ed. D. Briggs and M.P. Seah (Wiley, Chichester, 1988) p. 477. [15] J.E. Huheey, J. Phys, Chem. 70 (1966) 2086. [16] N.S. Mclntyre and M.D. Cook, Anal. Chem. 47 (1975) 2208. [17] G.E. MuUenberg, ed., Handbook of X-ray Photoelectron Spectroscopy (Perkin-Elmer, Eden Prairie, MN, 1979). [18] A. Proctor and P.M.A. Sherwood, Anal. Chem. 52 (1980) 2315. [19] J. Wong and C.A. Angell, Glass Structure by Spectroscopy (Dekker, New York, 1976). [20] B.M.J. Smets and D.M. Krol, Phys. Chem. Glasses 25 (1984) 113. [21] J.M. Arzeian and C.A. Hogarth, J. Mater. Sci. 26 (1991) 5353. [22] R.N. Bhargava and R.A. Condrate, Appl. Spectrosc. 31 (1977) 230. [23] L. Den-Dot and I. Felner, Inorgan. Chimica Acta 4 (1970) 49. [24] S.D. Ross, Inoganic Infrared and Raman Spectra (McGrawHill, London, 1972) p. 207. [25] R.K. Brow, R.J. Kirkpatrick and G.L. Turner, J. Am. Ceram. Soc. 73 (1990) 2293. [26] D.S. Goldman, Phys. Chem. Glasses 27(1986) 128.