Radial distribution of some defect-related optical absorption and PL bands in silica glasses

Radial distribution of some defect-related optical absorption and PL bands in silica glasses

Journal of Non-Crystalline Solids 277 (2000) 82±90 www.elsevier.com/locate/jnoncrysol Radial distribution of some defect-related optical absorption ...

274KB Sizes 0 Downloads 38 Views

Journal of Non-Crystalline Solids 277 (2000) 82±90

www.elsevier.com/locate/jnoncrysol

Radial distribution of some defect-related optical absorption and PL bands in silica glasses Yuryo Sakurai *, Kaya Nagasawa Department of Electrical Engineering, Shonan Institute of Technology, 1-1-25 Tujido-Nishikaigan, Fujisawa, Kanagawa 251-8511, Japan Received 20 June 2000

Abstract In order to investigate the origins of defect-related optical absorption and photoluminescence (PL) bands in silica glasses, we measured the radial distribution of paramagnetic defect centers, optical absorption band, and PL band in various silica glass rods. We studied the radial distribution of the following: the E0 center and the 5.8 eV absorption band, non-bridging oxygen hole center and the 1.9 eV PL band and 4.8 eV absorption band, and 5.0 eV absorption band and the 2.7 and 4.3 eV PL bands. As a result of this study, we con®rmed the correlation of both the radial distributions. These results suggest the following: the E0 center is the origin of the 5.8 eV absorption band, the nonbridging oxygen hole center is the origin of the 1.9 eV PL band and 4.8 eV absorption band, and the oxygen vacancy is the origin of the 5.0 eV absorption band, the 2.7 and 4.3 eV PL bands. Ó 2000 Elsevier Science B.V. All rights reserved.

1. Introduction As-grown high-purity silica glass rods exhibit radial distribution of impurities (hydroxyl groups and chlorine), defects …oxygen vacancy …O3 BSiA SiBO3 † and peroxy linkage …POL; O3 BSiAOA OASiBO3 †† [1]. These impurities and defects serve as precursors to other defects. Tohmon et al. [1] have suggested that it may be extremely important to consider these distributions when discussing defects in silica glass. The relation between defect centers and optical absorption or photoluminescence (PL) has been investigated by electron spin resonance (ESR) and optical methods (PL measurements and absorption properties) [2]. The existence of a radial correlation must be established * Corresponding author. Tel.: +81-466 344 111; fax: 81-466 358 897. E-mail address: [email protected] (Y. Sakurai).

by the observation of radial dependence. The prevailing view is that it is necessary to conduct some processing (for example, heat treatment) in order to observe this radial correlation. However, we considered that it might be possible to observe this radial correlation without doing such processing simply by using rod-shaped samples. With this point of view, we looked for the radial distribution using the rod-shaped samples. In our previous article, we reported the radial distribution of peroxy radial (POR; O3 BSiA OAO "; " denote an unpaired spin) and the 2.25 eV PL band (lifetime …s†: 300 ns, full width at half maximum (FWHM): 0.25 eV) [3], 3.8 eV absorbtion band and the 1.5 eV PL band (s: 3 ms, FWHM: 0.25 eV) [4,5] and 1.75 eV PL band (s: 200 ns, FWHM: 0.4 eV) and the extreme oxygen- de®cient states or Si clusters [6,7]. This present research deals with measurement of the radial distributions of paramagnetic defect

0022-3093/00/$ - see front matter Ó 2000 Elsevier Science B.V. All rights reserved. PII: S 0 0 2 2 - 3 0 9 3 ( 0 0 ) 0 0 3 2 9 - X

Y. Sakurai, K. Nagasawa / Journal of Non-Crystalline Solids 277 (2000) 82±90

centers, optical absorption bands, and PL bands in as-grown and high-dose (10 MGy) c-irradiated …60 Co† high-purity silica glass rods. We studied the radial distribution of the following: the E0 center …O3 BSi "† and the 5.8 eV absorption band [8±10], non-bridging oxygen hole centers (NBOHCs, O3 BSiAO ") and the 1.9 eV PL band (s: 15 ls, FWHM: 0.2 eV) [11±20] and 4.8 eV absorption band [11], and 5.0 eV absorption band and the 2.7 eV (s: 10 ms, FWHM: 0.4 eV) and 4.3 eV PL bands (s: few ns, FWHM: 0.4 eV) [21±23]. 2. Experimental The samples used in this study are listed in Table 1. All samples were bulk amorphous SiO2 prepared by the plasma (Ar; O2 and Ar ‡ O2 † chemical vapor deposition (CVD), ¯ame hydrolysis, and CVD soot remelting methods. Although samples A1±A4 were, as a whole, oxygen- de®cient, the edge of the samples had a surplus of oxygen and non-uniformity in the spatial distribution. The oxygen partial pressure …PO2 † used in the synthesis of sample A2 …PO2 ˆ 1:5%† is the same as that used in the synthesis of normal commercial grade silica glass. Samples A3 …PO2 ˆ 5%† and A4 …PO2 ˆ 15%† were synthe-

83

sized under excess oxygen pressure. Therefore, samples A3 and A4 were regarded as `oxygen-deficient ‡ O2 ' [1]. Samples B1 and B3 were also an `oxygen-deficient ‡ O2 ' type silica glass. Sample P was an oxygen-surplus-type silica glass with a 3.8 eV absorption band due to POL [24]. The AH and BH samples were prepared by adding hydroxyl groups during synthesis of the normal commercial grade glass. The OH1 sample likewise was prepared by adding hydroxyl groups, but the manufacturing method was O2 plasma. The samples were cut-out rod samples with a diameter 10±13 mm and a length of 40 mm. The samples were right cylindrical rods, with a diameter 10±13 mm and a length of 10 mm. The cylindrical surface was a half-circle, or fan-shape, with a diameter 10±13 mm and a length of 10 mm. All of the sample surfaces including the sides were mirror polished. We used both as-growth and c-irradiated samples. The c-irradiation was performed at room temperature in air using a 60 Co source with a dose rate of 10 kGy/h to achieve maximum doses of 10 MGy. PL measurements were made with a 1/4 meter-monochromator (320± 100 nm) equipped with a multichannel detector (200±1000 nm). The PL excitations were produced by illuminating the samples with an Ar‡ laser (488 nm (2.54 eV )), He±Ne laser (633 nm (1.96 eV)),

Table 1 Category, manufacturing methods, and impurities of the samples used for experimentsa

a

Sample name

Category

Manufacturing method

Impurity (ppm) Cl

OH

A1 A2 A3 A4 AH

Oxygen-de®cient PO2 ˆ 1:0% Oxygen-de®cient PO2 ˆ 1:5% Oxygen-de®cient PO2 ˆ 5:0% Oxygen-de®cient PO2 ˆ 15:0% High-OH …Oxygen-deficient ‡ OH†

Ar Ar Ar Ar Ar

plasma plasma plasma plasma plasma ‡ OH

12 000 3200 1000 400 340

free

B1 B3 BH

Oxygen-de®cient Mol.rate O2 : SiCl4 0.15:1 Oxygen-de®cient Mol.rate O2 : SiCl4 1:1 High-OH …Oxygen-deficient ‡ OH†

O2 plasma O2 plasma Ar plasma ‡ OH

1200 340 310

3.0 3.0 200

P

Oxygen surplus

Ar ‡ O2 plasma

370

0.6

D

High-OH

Flame hydrolysis

S1 S3 S4

Unknown (B2 b) Oxygen-de®cient Oxygen-de®cient

CVD soot remelting CVD soot remelting CVD soot remelting

PO2 : oxygen partial pressure during the synthesis; CVD: chemical vapor deposition.

free

free

0.8 3.0 3.3 500

1000 0.3 0.3

200 free 6.0

84

Y. Sakurai, K. Nagasawa / Journal of Non-Crystalline Solids 277 (2000) 82±90

He±Cd laser (325 nm (3.8 eV)) and a Nd:YAG laser (266 nm (4.66 eV)). The PL spectra were corrected for the non-linear response of the detection system. Measurements were carried out at room temperature. The absorption spectra in the visible±ultraviolet region were obtained with a double-beam UV±visible spectrometer. ESR spectra were taken of silica glass in the sample tube, which was immersed in liquid nitrogen in a Dewar at 77 K. The microwave frequency was near 9.4 GHz (X band), with a power of about 1 lW for measuring the E0 centers and about 5 mW for measuring the NBOHC and the POR. Each rod sample has a characteristic radial distribution of defects [1]. Investigating this distribution was very helpful in establishing a relationship between the observed PL band, absorption band, and ESR signal, because we would expect to ®nd some agreement between the

radial distribution of defects and the luminescence if a correlation does in fact exist. The radial distribution of the PL and absorption intensities were determined by moving the irradiation point 1±2 mm at a time from the center using an excitation beam focussed by a lens (see Fig. 1). At the end of the experiment, the radial distribution of the paramagnetic defect centers was measured by ESR. Samples were cut from the parent material with a width of 1.5±2 mm. The rod sample had to be cut so that the paramagnetic defect centers of these cut pieces could be measured with ESR. The high-dose c-irradiated samples are suitable for measuring the radial distributions of PL band, optical absorption band, and paramagnetic defect centers because the number of defect centers increases with increasing c-irradiation [25]. The experiment were repeated several times to obtain absorption, PL, and ESR intensity variance of less than 5%. 3. Results and discussion Figs. 2(a) and (b) show the absorption spectrum of various as-grown samples (2(a)) and c-irradiated sample P. Six absorption bands are observed at 2.0, 3.8, 4.8, 5.0, 5.17, and 5.8 eV. Fig. 3 shows the PL spectrum of various as-grown and c-irradiated samples. Eight PL bands are observed at 1.45, 1.75, 1.9. 2.25 (two types), 2.7, 3.08, and 5.0 eV. Fig. 4 shows ESR spectra for c-irradiated sample P. The three primary paramagnetic defect centers of silica detected by ESR, the E0 center, the POR and the NBOHC, were found in all samples. The main focus of our investigation is the cause and e€ect relationship between the E0 center and the 5.8 eV absorption band, NBOHC and the 1.9 eV PL band and 4.8 eV absorption band, and 5.0 eV absorption band and the 2.7 and 4.3 eV PL bands. 3.1. E0 center and the 5.8 eV absorption band

Fig. 1. Diagram of the experimental setup used to determine the radial distributions of (a) the PL intensity and (b) the ESR signal intensity in a silica glass sample.

Fig. 5 shows the radial distributions of the E0 center intensity (open square) and the 5.8 eV absorption band (closed circle) in c-irradiated sample A3. These intensities were independently normal-

Y. Sakurai, K. Nagasawa / Journal of Non-Crystalline Solids 277 (2000) 82±90

85

Fig. 3. Various PL bands in silica glass.

Fig. 2. Optical absorption spectra of various silica glass samples: (a) four di€erent as-grown silicas (A1: oxygen-de®cient (B2 a), P: oxygen surplus, D: high-OH, S1: unknown (B2 b)), (b) c-irradiated sample P.

ized to their maximum value. The data show a direct correlation between the E0 center intensity and the 5.8 eV absorption band. Fig. 6 shows the relationship between the intensity of the E0 center and the amplitude of the 5.8 eV absorption band. As shown in Fig. 6, the observed 5.8 eV absorption intensities are directly proportional to the E0 center intensity. This result establishes the correlation between E0 center and 5.8 eV absorption band. This result is consistent with previously reported results [9±11].

Fig. 4. ESR spectra of c-irradiated silica glass.

3.2. NBOHC and the 1.9 eV PL band and 4.8 eV absorption band Fig. 7 shows the radial distributions of the NBOHC intensity (closed circle) and the 1.9 eV PL band (The open circle is for 1.96 eV excitation, and the closed triangle if for 3.8 eV excitation) in c-irradiated sample P. The intensities were independently normalized to their maximum value. The radial distribution of the NBOHC intensity and the 1.9 eV PL absorption band track together.

86

Y. Sakurai, K. Nagasawa / Journal of Non-Crystalline Solids 277 (2000) 82±90

Fig. 5. Radial distribution of the E0 center intensity and 5.8 eV absorption band for the c-irradiated sample A3.



Fig. 7. Radial distribution of the NBOHC intensity … † and 1.9 eV PL intensity (( ): 1.96 eV excitation, ( ) BLACK: 3.8 eV excitation) for the c-irradiated sample P.

M

Fig. 6. Relationship between the E0 center intensity and 5.8 eV absorption band for various c-irradiated silica glass.

Fig. 8 shows the relationship between the NBOHC intensity and the 1.9 eV PL intensity (the open circle is for 1.96 eV excitation and the closed triangle is for 3.8 eV excitation). As shown in Fig. 8, the 1.9 eV PL intensities are in proportion to the NBOHC intensity. This result shows the correlation between NBOHC and 1.9 eV PL band. Fig. 9 shows the radial distribution of the NBOHC intensity (open circle) and the 4.8 eV absorption band (closed circle) in c-irradiated sample P. The intensities were independently normalized to their maximum value. The NBOHC intensity and the 4.8 eV absorption band track together. Fig. 10 shows the relationship between

Fig. 8. Ralationship between the NBOHC intensity and 1.9 eV PL intensity (( ): 1.96 eV excitation, ( ) BLACK: 3.8 eV excitation) for various c-irradiated silica glass.

M

the NBOHC intensity and the amplitude of the 4.8 eV absorption band. As shown in Fig. 10, 4.8 eV absorption intensities are in proportion to the NBOHC intensity. This result demonstrates the correlation between NBOHC and 4.8 eV absorption band. Figs. 8 and 10 show that the 1.9 eV PL intensity and the amplitudes of the 4.8 eV absorption band are correlated to NBOHC and may be caused by NBOHC, as proposed by Skuja et al. [11].

Y. Sakurai, K. Nagasawa / Journal of Non-Crystalline Solids 277 (2000) 82±90

Fig. 9. Radial distribution of the NBOHC intensity and 4.8 eV absorption band for the c-irradiated sample P.

87

Fig. 11. Radial distribution of 5.0 eV absorption band and two (2.7 and 4.3 eV) PL intensities for c-irradiated sample A2.

3.4. 2.25 eV PL band under 3.8 and 4.66 eV excitations

Fig. 10. Relationship between the NBOHC intensity and 4.8 eV absorption band for various c-irradiated silica glass.

3.3. 5.0 eV absorption band and the 2.7 and 4.3 eV PL bands under 4.66 eV excitation Fig. 11 shows the radial distribution of the 5.0 eV absorption band (closed triangle, B2 a band, due to the ground-to-triplet transition of oxygen vacancy) [21] and 2.7 (closed circle) and 4.3 eV (open square) PL bands in c-irradiated sample A2. The intensities were independently normalized to their maximum value. The data show a close correlation between the 5.0 eV absorption band …O3 BSiASiBO3 † and the 2.7 and 4.3 eV PL bands under 4.66 eV excitation.

Figs. 12(a) and (b) show the radial distribution of the 2.25 eV PL band (s: 25 ns, FWHM: 0.4 eV) [25] under 3.8 and 4.66 eV excitations in various c-irradiated samples ((a) AH and (b) A1). As shown in Fig. 12(a), in the AH sample, there is good correlation between the 3.8 and 4.66 eV excitations. On the other hand, in sample A1 (Fig. 12(b)), there is a distinct lack of correlation between the two excitations. Although we consider that the origin of this PL band under 3.8 eV excitation is the same as for 4.66 eV excitation, it is clear that the radial distributions of the 2.25 eV PL band for each excitation is di€erent (see Fig. 12(b)). This result suggests that the 2.25 eV PL band have a di€erent origin under 3.8 and 4.66 eV excitation. In order to understand this di€erence better, we measured the radial distributions of the 2.25 eV PL band under 3.8 eV excitation in A1±A4. samples. Fig. 12(c) shows the radial distributions of the 2.25 eV PL band under 3.8 eV excitation in c-irradiated A1±A4 samples. These samples were all prepared by the same method, the Ar plasma method, but under di€erent partial pressures of oxygen. The oxygen partial pressures went from A1 (1.0%) to A4 (15%). These samples, because they were synthesized under di€erent oxygen pressures, were used to investigate the e€ects of

88

Y. Sakurai, K. Nagasawa / Journal of Non-Crystalline Solids 277 (2000) 82±90

Fig. 12. Radial distribution of the 2.25 eV PL intensity under 3.8 and 4.66 eV excitations in c-irradiated (a) sample AH, (b) sample A1 and (c) samples A1±A4.

oxygen in silica glass. As shown in Fig. 12(c), the radial distribution of the 2.25 eV PL band depends on the sample. Although the PL intensity is greater at the outer edges than at the center for sample A1, this tendency is reversed with increasing PO2 . Samples A2, A3, and A4 show a decreased (and in this order) intensity at the outer edge in contrast with sample A1 (see Fig. 12(c)). We believe this observation re¯ects di€erences in the local structures of the PL centers associated with the oxygende®cient states introduced because the samples were prepared with di€erent oxygen partial pressures.

In a previous article [26], we suggested that this 2.25 PL band is associated with silicon clusters (E0d : oxygen-de®cient defects-clusters with ®ve silicon atoms) in amorphous SiO2 . As shown in Figs. 5 and 11, the oxygen-de®cient states decrease from the center to the outer edges of these samples. This suggests that during the synthesis of oxygen-de®cient-type silica glass, there is a higher probability that oxygen-de®cient states will form at the center of a sample rather than at the outer edges of the sample. A large number of E0 and oxygen-de®cient-associated defect centers (Si clusters or

Y. Sakurai, K. Nagasawa / Journal of Non-Crystalline Solids 277 (2000) 82±90

SiOx…x < 2† structure) may form near the center of a sample during its synthesis. In Sample A1, which was manufactured by the Ar plasma method in the presence of a very small amount of oxygen …PO2 ˆ 1:0%†, the center of the sample contained the highest concentration of oxygen vacancies (SiASi). If SiASi bonds are randomly distributed throughout the sample, SiO4 and O3 BSiASiBO3 units are the most probable units throughout the sample and have the highest probability of forming Si clusters. Therefore, if this PL is associated with only the E0d , PL intensities should be greater at the center than the outer edges of these samples. This was not the case in samples A1 and A2. This result indicates clearly the possibility of another PL mechanism, other than the E0d . In a previous article [3], we pointed out that the 2.25 eV PL band (s: 300 ns, FWHM: 0.2 eV) is, under ultra violet light, due to the POR. Although we considered that PL due to POR would be observed in the 2.25 eV PL band, this PL was not observed at room temperature in these samples. Therefore, the 2.25 eV PL observed here is not associated with the POR. Based on our results, we conclude that the observed 2.25 eV PL band (s: 25 ns, FWHM: 0.4 eV) is associated with E0d and other (unknown) oxygen-de®cient state. From the results reported above (plus our previous results, i.e., POR and the 2.25 eV PL band [3], 3.8 eV absorption band and the 1.5 eV PL band [4,5], and 1.75 eV PL band and the extreme oxygen-de®cient states or Si clusters [6,7]), we conclude that investigation of the radial distribution property is very useful as a method for establishing the origins of defect-related optical absorption and PL bands. These results may be important for the future study of the origins of defect-related optical absorption and PL bands. 4. Conclusion In order to investigate the model for the origin of defect-related optical absorption and PL bands in silica glasses, we studied the radial distribution of the following: the E0 center and the 5.8 eV absorption band, NBOHC and the 1.9 eV PL band and 4.8 eV absorption band, and 5.0 eV absorp-

89

tion band and the 2.7 and 4.3 eV PL bands in various silica glasses. As a result of our measurements, we con®rmed correlations of the radial distributions. These results suggest the following: the E0 center is the origin of the 5.8 eV absorption band, NBOHC is the origin of the 1.9 eV PL band and 4.8 eV absorption band, and oxygen vacancy is the origin of the 5.0 eV absorption band, the 2.7 and 4.3 eV PL bands. Examining the radial distribution was an e€ective means of verifying the correlation. Acknowledgements The authors acknowledge Professors Yoshimasa Hama and Yoshimichi Ohki of Waseda University and Dr Hiroyuki Nishikawa for helpful discussions. References [1] R. Tohmon, A. Ikeda, Y. Shimogaichi, S. Munekuni, Y. Ohki, K. Nagasawa, Y. Hama, J. Appl. Phys. 67 (1990) 1302. [2] D.L. Griscom, J Ceramic Soc. Jpn 99 (1991) 923. [3] Y. Sakurai, K. Nagasawa, J. Appl. Phys. 86 (1999) 1377. [4] Y. Sakurai, K. Nagasawa, J. Non-Cryst. Solids 261 (2000) 21. [5] Y. Sakurai, J. Non-Cryst. Solids 276 (2000) 159. [6] Y. Sakurai, K. Nagasawa, H. Nishikawa, Y. Ohki, J. Appl. Phys. 86 (1999) 370. [7] Y. Sakurai, Jpn. J. Appl. Phys. 39 (2000) 496. [8] C.M. Nelson, R.A. Weeks, J. Am. Ceram. Soc. 43 (1960) 396. [9] R.A. Weeks, E. Nonder, Paramagnetic Resonance, in: W. Low (Ed.), vol. 2, Academic Press, New York, 1963, p. 869. [10] E.P. O'Reilly, J. Robertson, Phys. Rev. B 27 (1983) 3780. [11] L.N. Skuja, A.R. Silin, Phys. Stat. Sol. A 56 (1979) K11. [12] G.H. Sigel, M.J. Marrone, J. Non-Cryst. Solids 45 (1981) 235. [13] J.H. Stathis, M.A. Kastner, Philos. Mag. B 49 (1984) 357. [14] J.H. Stathis, M.A. Kastner, Phys. Rev. B 35 (1987) 2972. [15] R.A.B. Devine, C. Fiori, J. Robertson, MRS Symposium Proceedings, vol. 61, Materials Research Society, Pittsburgh PA, 1986, p.177. [16] Y. Hibino, H. Hanafusa, J. Non-Cryst. Solids 107 (1988) 23. [17] R. Tohmon, Y. Shimogaichi, S. Munekuni, Y. Ohki, Y. Hama, K. Nagasawa, Appl. Phys. Lett. 54 (1989) 1650.

90

Y. Sakurai, K. Nagasawa / Journal of Non-Crystalline Solids 277 (2000) 82±90

[18] S. Munekuni, T. Yamanaka, Y. Shimogaichi, R. Tohmon, Y. Ohki, K. Nagasawa, Y. Hama, J. Appl. Phys. 68 (1990) 1212. [19] L. Skuja, Solid State Commun. 84 (1992) 613. [20] H. Nishikawa, T. Shiroyama, R. Nakamura, Y. Ohki, K. Nagasawa, Y. Hama, Phys. Rev. B 45 (1992) 586. [21] R. Tohmon, H. Mizuno, Y. Ohki, K. Sasagane, K. Nagasawa, Y. Hama, Phys. Rev. B 39 (1989) 1337. [22] R. Tohmon, Y. Shimogaichi, H. Mizuno, Y. Ohki, K. Nagasawa, Y. Hama, Phys. Rev. Lett. 62 (1989) 1388.

[23] H. Nishikawa, Y. Miyake, E. Watanabe, D. Ito, K.S. Seol, Y. Ohki, K. Ishii, Y. Sakurai, K. Nagasawa, J. Non-Cryst. Solids 222 (1997) 221. [24] H. Nishikawa, R. Tohmon, Y. Ohki, K. Nagasawa, Y. Hama, J. Appl. Phys. 65 (1989) 4672. [25] Y. Sakurai, K. Nagasawa, H. Nishikawa, Y. Ohki, J. Appl. Phys. 75 (1994) 1372. [26] H. Nishikawa, E. Watanabe, D. Ito, Y. Sakurai, K. Nagasawa, Y. Ohki, J. Appl. Phys. 80 (1996) 3513.