Structure of alkali tungsten tellurite glasses by X-ray photoelectron spectroscopy

Structure of alkali tungsten tellurite glasses by X-ray photoelectron spectroscopy

Journal of Non-Crystalline Solids 349 (2004) 60–65 www.elsevier.com/locate/jnoncrysol Structure of alkali tungsten tellurite glasses by X-ray photoel...

431KB Sizes 0 Downloads 18 Views

Journal of Non-Crystalline Solids 349 (2004) 60–65 www.elsevier.com/locate/jnoncrysol

Structure of alkali tungsten tellurite glasses by X-ray photoelectron spectroscopy J.W. Lim a, H. Jain a,*, J. Toulouse a, S. Marjanovic a, J.S. Sanghera b, R. Miklos b, I.D. Aggarwal b a

Center for Optical Technology, Lehigh University, Bethlehem, PA 18015, USA b Naval Research Laboratory, Washington, DC 20375, USA Available online 2 November 2004

Abstract The electronic structure of two alkali tungsten tellurite glass series: (a) 10Li2O Æ xWO3 Æ (90  x)TeO2, and (b) xK2O Æ (10  x)Li2O Æ 25WO3 Æ 65TeO2 has been investigated by X-ray photoelectron spectroscopy. In contrast to alkali tellurite glasses, the core level spectra of the various elements appear unaffected when TeO2 is replaced by WO3, with the alkali oxide concentration remaining constant. The O1s spectra do not indicate a clear separation of the bridging and non-bridging oxygen contributions. Thus we conclude that WO3 behaves essentially as a network former in tellurite glasses. We have also obtained valence band spectra of the glasses and observed major changes with the addition of WO3. In particular, the intensity corresponding to the lowest energy part of the spectra, attributed to the O2p bonding orbital, decreases significantly with increasing WO3, indicating the appearance of a new bonding configuration.  2004 Elsevier B.V. All rights reserved.

1. Introduction Numerous glass compositions have been developed over the years for different applications in photonics. Among them, tellurite glass and glass-ceramics are promising choices due to their high refractive index (larger than 2.0), wide band infrared transmittance (extending up to 6 lm), and large third order non-linear optical susceptibility [1,2]. In addition, tellurite glasses combine the attributes of a short wavelength UV edge (about 350 nm), good glass stability, rare earth ion solubility, a slow corrosion rate, and relatively low phonon energy (600–850 cm1) among oxide glass formers [3]. Tellurite

*

Corresponding author. Tel.: +1 610 758 4217; fax: +1 610 758 4244. E-mail address: [email protected] (H. Jain). 0022-3093/$ - see front matter  2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jnoncrysol.2004.08.263

glasses are capable of providing a large stimulated emission cross section over a broad bandwidth. These fascinating properties of tellurite glasses are associated with their anomalous network and electronic structure. However, a definite picture of their network and electronic structure has not been obtained. Therefore, the purpose of this work is to examine the electronic structure of tungsten tellurite glasses, which will ultimately help us understand the fundamental origin of their optical properties. Several research groups have investigated the network and electronic structures of tellurite glasses by using different experimental methods. For example, binary tellurite glasses containing alkali oxides [4–8] have been studied by infrared [4–7], Raman [4,6–8], and X-ray photoelectron spectroscopies [4]. Our study has focused on the structural development of tellurite glasses within the Li2O–WO3–TeO2 ternary system. In order to investigate the electronic structure we have used X-ray photoelectron spectroscopy (XPS).

J.W. Lim et al. / Journal of Non-Crystalline Solids 349 (2004) 60–65

2. Experimental method The glasses in this study were prepared in two series, namely (a) 10Li2O Æ xWO3 Æ (90  x)TeO2 and (b) xK2O Æ (10  x)Li2O Æ 25WO3 Æ 65TeO2 (see Table 1). All glasses in this work also contained 0.1 wt% Tm2O3, which was added for characterizing the laser properties but is not likely to affect any bulk structure discussed in this paper. In addition, a tungsten-free 20Li2O Æ 80TeO2 sample was used as a reference for the XPS spectrum. The glass series (a) emphasizes the substitution of tellurium oxide by tungsten oxide with constant alkali oxide content. The glasses were melted from reagent grade Li2CO3 (Alfa Aesar, Puratronic Grade, 99.999%), K2CO3 (Alfa Aesar, Puratronic Grade, 99.999%), WO3 (Alfa Aesar, Puratronic Grade, 99.998%), TeO2 (Alfa Aesar Puratronic Grade, 99.9995%) and Tm2O3 (Rhone-Poulenc, 99.90%). These compounds were mixed in the appropriate ratios to give 15 g batches. The mixture was first calcined and then homogenized by heating to 850 C in gold crucibles in air for 4 h. After removing from the furnace, the samples were cooled in air to room temperature. The glasses were then annealed at 350 C for 2 h. To make the 20Li2O Æ 80TeO2 glass, the melt was quickly quenched in water after removing from the furnace. The reason for the quick quench was that this sample partially crystallized under the same conditions used for the other glasses, presumably since it was on the edge of the glass-forming region. A high-resolution ESCA Scienta 300 spectrometer was used for the XPS experiments with monochromatic Al-Ka X-ray as the probe radiation. The glass samples were ground to 1 mm thickness to give a flat surface for proper clamping of the sample holder in the chamber. They were then fractured in situ in the ultra high vacuum preparation chamber where the base pressure was on the order of 109 Torr. This procedure minimized any contamination layer on the fractured glass surface. After fracturing of the samples, XPS data were Table 1 Glass composition and glass transition temperature Composition (mol%)

Glass transition temperature (Tg, C)

20Li2O Æ 80TeO2 xK2O Æ (10  x)Li2O Æ 25WO3 Æ 65TeO2

259

x=0 5 10 10Li2O Æ xWO3 Æ (90  x)TeO2

335 331 339

x=5 10 15 20 25

283 304 321 334 335

61

obtained immediately from the newly created surface. The photoelectrons were generated from the fracture surface of area about 3 · 1 mm2. The X-ray spot size was about 1.6 · 0.3 mm2. The ESCA instrument was operated for survey scans over the entire binding energy range as well as for the regional scans over the photoelectron peaks of interest. Our regions of interest were tungsten 4f (45–30 eV), oxygen 1s (540–525 eV), tellurium 3d5/2 (585–565 eV), carbon 1s (300–275 eV), and valence band (25–0 eV). An energy step size of 0.1 eV was used for the regional scans. The instrumental contribution to the line width was extremely small (<4%). Consequently, the experimental widths were primarily a combination of charging and natural line widths. The number of scans, which was 6–8 for our samples, was adjusted to give high signal to noise ratio. Since our samples are insulators, the glass surface exposed to the X-rays becomes positively charged due to emission of photoelectrons. In order to compensate for this charging effect, the sample surface was flooded with low energy electrons of about 4–8 eV, which gave a better resolution and reproducibility than a surface of neutral potential. The element core level spectra were analyzed using the Scienta software. A Shirley background was subtracted and a Voigt line shape, which is a mixture of Gaussian and Lorentzian, was used to analyze the peaks for each specific element in spectra. Measured binding energy of the peaks was corrected based on the calibration factor calculated from the difference between the measured binding energy of the C 1s peak and its reference value of 284.6 eV.

3. Results The differential thermal analysis of (a) 10Li2O Æ xWO3 Æ (90  x)TeO2 and (b) xK2O Æ (10  x)Li2O Æ 25WO3 Æ 65TeO2 glasses was conducted to determine the glass transition temperature (Tg). The results for 10Li2O Æ xWO3 Æ (90  x)TeO2 show that Tg increases monotonically with increasing WO3 concentration (Fig. 1). It increases from 283 C for a glass containing 5 mol% WO3 up to 335 C when the concentration of WO3 is increased to 25 mol%. Fig. 2 shows the O 1s photoelectron spectra for the 10Li2O Æ xWO3 Æ (90  x)TeO2 glass system. Each peak is symmetrical and there is no shoulder on the higher binding energy side of the main peak. The shape of the peak is nearly the same for all values of x. However, as shown in Fig. 3(a), the main O 1s peak shifts toward higher binding energy with increasing WO3 content except perhaps for the glass with 15%WO3. Heo et al. [4] reported a shoulder on the higher binding energy side of the O 1s photoelectron spectrum of alkali tellurite

62

J.W. Lim et al. / Journal of Non-Crystalline Solids 349 (2004) 60–65 340

o

Tg ( C)

320

300

280

260 5

10

15

20

25

WO 3 (mol.%) Fig. 1. Glass transition temperature (Tg) of 10Li2O Æ xWO3 Æ (90  x) TeO2 glasses.

Fig. 4. Te 3d X-ray photoelectron spectra for the 10Li2O Æ xWO3 Æ (90  x)TeO2 glass system.

glasses. They attributed this shoulder (538 eV) to the oxygen atom in TeO3 trigonal pyramids (tp) and the main peak at 536 eV to that in TeO4 trigonal bipyramids (tbp). Himei et al. [9] also observed a shoulder on the high binding energy side of the main O 1s peak in the photoelectron spectra for the xLi2O Æ (100  x)TeO2 glass surface that were exposed to air. However, they did not find the shoulder when the sample was fractured in the vacuum, thus confirming the high quality of our data. Fig. 4 shows Te 3d photoelectron spectra for the samples in the same system, where the higher binding energy peak is due to Te 3d3/2 and the lower binding energy peak is due to Te 3d5/2. Both peaks shift to higher binding energy with increasing WO3 as shown in Fig. 3(b) for Te 3d5/2. The W 4f photoelectron spectra for the samples are shown in Fig. 5. The shape of the peaks is nearly the same for all values of x, but the position shifts to higher binding energy with increasing W content. The observed binding energy and full width at half maximum (FWHM) for

(a)

B. E. (eV) for Te 3d 5/2

B. E. (eV) for O 1s

Fig. 2. O 1s X-ray photoelectron spectra for the 10Li2O Æ xWO3 Æ (90  x)TeO2 glass system.

531 530.8 530.6 530.4 530.2 530 x

= 510

15

x (mol %)

20

25

(b)

576.8 576.6 576.4 576.2 576 575.8 575.6 x=5

10

15

20

25

x (mol %)

Fig. 3. The main (a) O 1s and (b) Te 3d5/2 core level peak position for the 10Li2O Æ xWO3 Æ (90  x)TeO2 glass system.

J.W. Lim et al. / Journal of Non-Crystalline Solids 349 (2004) 60–65

63

Fig. 6. X-ray photoelectron spectra near the valence band of 10Li2O Æ xWO3 Æ (90  x)TeO2 glass system. The bottom curve is for a-TeO2 crystal [9] and indicated binding energies are calculated by Suehara et al. [14,15]. Fig. 5. W 4f X-ray photoelectron spectra for the 10Li2O Æ xWO3 Æ (90  x)TeO2 glass system.

toward smaller binding energy when lithium is replaced by potassium (Fig. 7). Each peak is symmetrical and there is no shoulder on the low or high binding energy side of the main peak. The shape of the peak is nearly the same for all values of x.

the core level O 1s, Te 3d5/2and W 4f7/2 peaks for our glass samples are summarized in Table 2. Experimental uncertainty of the binding energies is less than ±0.1 eV. Finally, Fig. 6 shows the valence band for the glasses in the 10Li2O Æ xWO3 Æ (90  x) TeO2 system. The O 1s and W 4f photoelectron spectra for the mixed alkali xK2O Æ (10  x)Li2O Æ 25WO3 Æ 65TeO2 glass system are shown in Fig. 7(a) and (b), respectively. The observed binding energy and FWHM are given in Table 2. Since the C 1s spectrum of the 5% and 10%K2O containing glasses showed an additional C peak from contamination, the Te 3d5/2 peak (576.05 eV) has been used as an internal reference. The O 1s and W 4f peaks shift

4. Discussion 4.1. Core level photoelectron spectra of lithium tungsten tellurite glasses When the O 1s spectra are analyzed, we find a single symmetric peak. The lack of a shoulder at high binding energy in Fig. 2, such as reported by Himei et al. [9], indicate that our data for the freshly fractured surface

Table 2 Binding energy of XPS O 1s, Te 3d5/2 and W 4f7/2 peaks for glass samples Composition (mol%)

Binding energy (eV) O 1s

20Li2O Æ 80TeO2 xK2O Æ (10  x)Li2O Æ 25WO3 Æ 65TeO2 x=0 5 10 10Li2O Æ xWO3 Æ (90  x)TeO2 x=5 10 15 20 25

Te 3d5/2 a

W 4f a

7/2

529.8 (1.27)

575.7 (1.13)



530.7 (1.31) 530.2 (1.31) 530.2 (1.28)

576.6 (1.25) 576.0 (1.26) 576.2 (1.18)

35.8 (0.98) 35.1 (0.94) 35.2 (0.92)

530.4 530.5 530.5 530.6 530.7

576.2 576.4 576.3 576.5 576.6

35.2 35.5 35.3 35.5 35.7

(1.34) (1.3) (1.27) (1.32) (1.31)

Experimental uncertainties of the binding energies are less than ±0.1 eV. a Full width at half maximum in parentheses.

(1.2) (1.18) (1.13) (1.26) (1.25)

(0.93) (0.92) (0.87) (1.02) (0.98)

64

J.W. Lim et al. / Journal of Non-Crystalline Solids 349 (2004) 60–65

Fig. 7. (a) O 1s (b) W 4f X-ray photoelectron spectra of xK2O Æ (10  x)Li2O Æ 25WO3 Æ 65TeO2 glass system (using Te 3d5/2 peak (576.05 eV) as internal reference).

are free from contamination. The C 1s spectrum of the samples (not shown) is a single component peak due to a hydrocarbon layer deposited from the vacuum system. We have used this C 1s peak as an internal reference at 284.6 eV. In Fig. 2, only a single, symmetric Voigt (Gaussian– Lorentzian) peak is observed in the O 1s photoelectron spectra and there is no variation of the shape with composition. For alkali silicate glasses, two components attributed to bridging oxygen (BO) and non-bridging oxygen (NBO) are generally observed [11,12]. The non-bridging oxygen atoms are detected as a shoulder at the lower binding energy side of the main O 1s peak due to bridging oxygen. This is the so-called chemical shift that depends directly on the electron density of the atoms; the higher electron density around an atom results in a lower binding energy of the ejected photoelectrons [4]. The O 1s binding energy for alkali tellurite glasses is smaller than the O 1s binding energy for the BO or NBO in alkali silicate glasses (e.g. 531.76 and 529.76 eV, respectively, for 20Na2O Æ 80SiO2 glass). This implies that the electron density on the oxide ions in tellurite glasses is higher than in silicates. In general, for alkali tellurite glass such as Li2O–TeO2, two components exist in the O 1s spectra. Sekiya et al. [7] proposed through Raman spectroscopic studies that NBOs, which are a part of the TeO 3þ1 polyhedra and TeO3 trigonal pyramid (tp) units, were formed in alkali tellurite glass

system due to the addition of alkali oxides. Thus the Tg decreases with the addition of Li2O in the Li2O– TeO2 system due to cleavage of the network formed by TeO4 tbp units and the consequent increase of NBOs. However, there is no non-bridging oxygen indicated in our O 1s peak, indicating a more homogeneous electron distribution on the addition of WO3. The Te 3d photoelectron spectra in Fig. 4 also consist of a single and symmetric Voigt peak. The shape does not change with composition, but the binding energy shifts to higher value. Sekiya et al. [7] proposed that TeO4 trigonal bipyramids (tbp) are deformed into lower-symmetry TeO 3þ1 polyhedra with the addition of alkali oxide. Then TeO3 trigonal pyramids (tp) start to appear at higher alkali content. Moawad et al. [13] also have observed a variety of Te–O units in vanadium tellurite glasses. However, there is no indication of any corresponding minor components in the Te 3d photoelectron spectra of our samples. The binding energy of the W 4f7/2 peak in Fig. 5 also shifts in a similar manner to the O 1s and Te 3d core level peaks. Table 2 shows the FWHM for the O 1s, Te 3d5/2, and W 4f 7/2 core level peaks for the glass samples. There is no significant change with increasing WO3 content and there is no other peak on the lower binding energy side in the W 4f spectra. Thus, it appears that the alkali concentration is too small to affect the core level binding energy, and WO3 does not affect the structure significantly. Previous structural investigations by Raman and neutron scattering have considered the existence of both WO6 and WO4 units, but the former units appear to be present predominantly [10,19–21]. Since we do not observe measurable changes in the W 4f7/2 peak, the fraction of tungsten in the two oxidation states is not affected by the W/Te ratio; primarily it remains in the +6 oxidation state. Overall, there is little change in the core level spectra of the 10Li2O Æ xWO3(90  x)TeO2 glass series with varying x. 4.2. Core level photoelectron spectra of mixed alkali (Li, K) tungsten tellurite glass series The core level peaks of the xK2O Æ (10  x)Li2O Æ 25 WO3 Æ 65TeO2 glass system shift toward smaller binding energy with the replacement of lithium by potassium, as shown in Fig. 7. As the electron affinity of K is lower than that of Li, more charge is transferred from the former to the oxygen in the network. However mixed alkali glass (5%K2O Æ 5%Li2O) does not follow the linear trend of peak shift. 4.3. Valence band spectra The valence band spectra for the 10Li2O Æ xWO3 Æ (90  x) TeO2 glasses are shown in Fig. 6. Relatively large changes are observed between 1 and 15 eV with

J.W. Lim et al. / Journal of Non-Crystalline Solids 349 (2004) 60–65

increasing WO3 content. The peaks at a binding energy of 13 eV and 3.5 eV become smaller with increasing x. A complex change in bonding is observed at 6.8 eV. For the 15% WO3 sample, a peak appeared and it became broader. Then the intensity of the peak at 6.8 eV increased as the composition reached 25%WO3. According to Suehara et al. [14,15], who calculated the valence band of an a-TeO2 single crystal (Fig. 6), the peak at 22 eV is mainly due to O 2s orbital with a small contribution from the Te 5s and Te 5p orbitals. The band at 13 eV represents the Te 5s orbital admixed with O 2s antibonding states. The band at 8 eV refers to the bonding orbitals between Te 5p and O 2p containing a small amount of O 2s antibonding character. The band at 3 eV mainly originates from the O 2p states. In the XPS study of WO3 powder [16,17] and thin film [18], the valence band spectra show a peak at 6.8 eV, which has its origin in the O 2p electron orbital. These peak assignments are compared with the valence band spectra of our compositions wherein hybridization of bonding orbitals with the addition of tungsten oxide is likely. First of all, there is no change at 22.5 eV, suggesting that O 2s does not participate in bonding with W. The intensity of the peak at 13 eV in Fig. 6 decreases on replacing TeO2 by WO3. Therefore, this band should be primarily due to Te 5s electrons. The band at 6.8 eV, which becomes stronger with increasing WO3, is most likely due to O 2p associated with W, which is consistent with the literature data for the valence band of WO3[16– 18]. The peak at 3.5 eV decreases sharply for x > 10% and becomes a part of the background for the 25%WO3 sample. This band should be related to O 2p level associated with Te because it decreases with the addition of WO3. The patterns at 3.5 eV and 6.8 eV are reversed for glasses with more than 15%WO3. From these data, it is quite apparent that tungsten affects the structure of the alkali tellurite glass system. 4.4. Glass transition temperature As determined from the thermal analysis of 10Li2O Æ xWO3 Æ (90  x)TeO2 glass samples, the glass transition temperature increased monotonically with increasing WO3 concentration as shown in Fig. 1. The glass structure consists of a continuous network mainly of TeO4 tbps at the low percent of WO3. The increase of Tg is due to the formation of W–O–Te linkages with the increase in WO3 content [10]. 5. Conclusions The shape of core-level XPS spectra for all the elements show little change as TeO2 is replaced by WO3

65

in the 10Li2O Æ xWO3 Æ (90  x)TeO2 glass series. Thus WO3 behaves as a network former. The W–O–Te linkages are stronger than the Te–O–Te linkages so that the glass transition temperature increases with x. By comparison, the valence band spectra show significant systematic changes that are consistent with the variation of composition.

Acknowledgments The authors wish to thank Dr A.C. Miller for his help with the XPS experiments and the National Science Foundation for financial support under grant no. DMR-9974031.

References [1] H. Nasu, T. Uchigaki, K. Kamiya, H. Janbara, K. Kubodera, Jpn. J. Appl. Phys. 31 (1992) 3899. [2] S.H. Kim, T. Yoko, S. Sakka, J. Am. Cerm. Soc. 76 (1993) 2486. [3] J.S. Wang, E.M. Vogel, E. Snitzer, Opt. Mater. 3 (1996) 187. [4] J. Heo, D. Lam, G.H. Sigel Jr., E.A. Mendoza, D.A. Hensley, J. Am. Ceram. Soc. 75 (1992) 277. [5] K. Suzuki, J. Non-Cryst. Solids 95&96 (1978) 15. [6] M. Dimitrova, Y. Dimitriev, M. Arnaudov, V. Dimitrov, Phys. Chem. Galss 30 (1989) 260. [7] T. Sekiya, N. Mochida, A. Ohtsuka, M. Tonokawa, J. NonCryst. Solids 144 (1992) 128. [8] Y. Himei, A. Osaka, T. Nanba, Y. Miura, J. Non-Cryst. Solids 177 (1994) 164. [9] Y. Himei, Y. Miura, T. Nanba, A. Osaka, J. Non-Cryst. Solids 211 (1997) 64. [10] T. Sekiya, N. Mochida, S. Ogawa, J. Non-Cryst. Solids 176 (1994) 105. [11] R. Bruckner, H.U. Chun, H. Goretzki, Glasstech. Ber. 51 (1978) 1. [12] S. Matsumoto, Y. Miura, T. Nanba, A. Osaka, Proc. 17th Int. Congress on Glass, vol. 3, 1995, p. 72. [13] H.M. Moawad, J. Toulouse, H. Jain, O. Latinovic, A.R. Kortan, Ceram. Trans (Optoelectronic Materials Technology in the Information Age) 126 (2002) 45. [14] S. Suehara, K. Yamamoto, S. Hishita, A. Nukui, Phys. Rev. B 50 (1994) 7981. [15] S. Suehara, K. Yamamoto, S. Hishita, A. Nukui, Phys. Rev. B 51 (1995) 14919. [16] R.J. Colton, J.W. Rabalais, Inorg. Chem. 15 (1976) 236. [17] T.H. Fleisch, G.J. Mains, J. Chem. Phys. 76 (1982) 780. [18] R.J. Colton, A.M. Guzman, J.W. Rabalais, J. Appl. Phys. 49 (1978) 409. [19] V. Kozhukharov, S. Neov, I. Gerasimova, P. Mikula, J. Mater. Sci. 21 (1986) 1707. [20] T. Kosuge, Y. Benino, V. Dimitrov, R. Sato, T. Komatsu, J. Non-Cryst. Solids 242 (1998) 154. [21] V. Balraj, K. Vidyasagar, Inorg. Chem. 38 (1999) 5809.