- Email: [email protected]
Character, mechanism of formation and transformation of point defects in type IV silica glass
] O U R N A L OF
ELSEVIER
Journal of Non-Crystalline Solids 169 (1994) 15-28
Character, mechanism of formation and transformation of point defects in type IV silica glass V . K h . K h a l i l o v *, G . A . D o r f m a n , E . B . D a n i l o v , M . I . G u s k o v , V . E . E r m a k o v Steko Ltd., Piskarevsky pr. 63, 195273 St. Petersburg, Russian Federation
(Received 9 August 1991;revised manuscript received 30 September 1993)
Abstract
In the temperature range 600-650°C, transformations of defects in quartz glass are observed. The dependence of a band at 163 nm on annealing temperature is measured. Changes of structure in this temperature range are observed for the -Si-H and -Si-OH centers. These are caused by the appearence of a structural relaxating process at ~ 675°C, the character of which is not determined. It is shown that the band at 163 nm is caused by three-coordinated non-paramagnetic silicon Si~-(T~- center). The band at 248 nm is attributed to an oxygen vacancy -----~i-Si=. It is shown that at temperature < 650°C the formation of Si-H centers (absorption band at 2254 cm-1) is due to the interaction of H 2 with oxygen vacancies. At temperatures > 650°C, the interaction of H 2 with T~- centers was achieved. The maximum of the band of the -=Si-H centers thus formed is at 2264 cm- 1. It was assumed that, in the type IV glass, the formation of Si-OH is due to an interaction of H 2 with peroxy linkages. Changes of the fundamental absorption edge are correlated with the changes of concentration of OH groups from 1500 ppm to 2 ppm. In the region < 2 ppm, the linear dependence of these changes on OH concentration is violated.
I. Introduction
It is well known [1] that the properties of pure quartz glass, independently of the oxidation methods of SiCI4, which are used in most of the technological processes, are sensitively influenced by fluctuations of the fusion conditions. This influence results not only in inhomogeneous distribution of impurities but is also accompanied by an inhomogeneous distribution of the intrinsic defects [1]. Moreover, there is a fluctuation of the glass structure [1]. Information about the main
* Corresponding author. Tel: + 7-812 249 2990. Telefax: + 7812 249 3510.
structural factors which make inhomogeneities in these types of glass is very important for improving the process. On the other hand, the possibility of producing quartz glass with different properties by one method but with different properties has its own interest with respect to investigating intrinsic defects and their transformations and the effect of characteristic impurity centers such as SiOH, Sill, SiCI and dissolved gases such as O 2. This work presents the results of an investigation of samples fabricated by oxidation of SiCI 4 in a low temperature oxygen-nitrogen plasma (type IV). The concentration of O H groups in the glass ranged from 0.5 to 300 ppm and chlorine from 1000 to 50 ppm. The quantity of metallic
0022-3093/94/$07.00 © 1994 Elsevier Science B.V. All rights reserved SSDI 0022-3093(93)E0275-D
16
V.Kh. Khalilov et aL /Journal of Non-Crystalline Solids 169 (1994) 15-28
impurities was less than 1 ppm by weight. Some samples were fabricated with differing concentrations of hydroxyl molecules and chlorine ions. The content of defect centers (known as oxygendeficit centers with absorption bands at 248 nm and 163 nm) in fabricated glass differed. The amplitudes of these bands were observed in the samples with low concentration of OH group and were absent in samples with concentrations of hydroxyl near 300 ppm. The source of the 248 nm absorption band is still the subject of discussion. Among the models for the center which are discussed now in the literature, the most probable are a two-coordinated silicon [2] and an oxygen vacancy [3-5]. In our opinion, the center associated with the 163 nm band is a three-coordinated non-paramagnetic silicon (T~ center) [6]. In Refs. [3] and [7], the 163 nm band is connected with oxygen vacancies, but in Ref. [8] its correlation with the peroxy radical was shown. Information about the structure of oxygen-deficit centers in quartz glass has been developed by using hydrogen as a probe, the spectroscopic manifestation of which, in the form of Sill and SiOH, is well known. In this paper, an investigation of the temperature dependences of absorption spectra for some of the fabricated samples after treatment by H 2 in the temperature range (from 200 to 1000°C) is reported. Using the same samples, investigations were made of absorption and reflection spectra, including the long wavelength edge of interband fundamental absorption of the 150 nm band, the spectroscopic properties of intrinsic and impurity defects, the oscillation spectra of hydrogen centers, and the region of fundamental intrinsic oscillations. The spectroscopic investigations of some of the samples, fabricated by one method and in which there were different equilibrium concentrations of impurity-defect centers, provided an opportunity to obtain some data on transformations of these centers. In particular, experimental results show a change of the transformation process of the impurity-defect centers in the glass, which is connected with the relaxation processes in the temperature range 650-700°C. It should be mentioned that the investigation of the processes
of interaction of glass with H 2 in this temperature range, with the control of concentration of impurities such as C12 and H 2 has a significance, particularly for the production and use of optical fibers.
2. Experimental The dependence of the optical densities of samples in the form of platelets in the investigated spectral range was measured. In the spectral range 150-400 nm and IR range 4000-2000 cm -~ samples with 0.8-1.0 mm thickness were used. In the 200-2000 nm spectral range, preform rods with lengths 200-1000 mm and diameter 20 mm or samples of glass from the blocks with diameter 30 mm and with length up to 100 mm were used. To prevent the cylindrical surface from affecting the measurements of absorption spectra in the long preform rods, the surfaces were treated with HF and ~<0.3 mm were removed, after which the surfaces were firepolished. After these treatments, the influence of the side surface on the measurements of transmission in the sample was not greater than the spectrophotometric measurement error. In the case of investigation of preform rods, the samples for measurement in VUV and IR spectral range were made with a thickness 0.8-1.0 mm from the samples previously treated by HF and firepolished. The measurements of absorption spectra of the rods were made by the classical double-beam method. Special measures were undertaken to maintain the diameter and the position of the light spot on the photodetector in the measuring and comparison channels. As a result, the measuring error was ~< + 1% (_+0.04 abc) in VUV and UV-VIS regions. The absorption and reflection spectra in the IR range were measured by a spectrophotometer (Shimadzu IR-470). As the process of measurement consisted in comparison of the same sample in the initial state and after cycles of treatment, the main problem was the reproducibility of measurements. The reproducibility was ~ 2.0 cm-1 in the wavelength scale and photometric accuracy _+0.002 abc. The sample was placed in one place
I(.Kh. Khalilou et aL /Journal of Non-Crystalline Solids 169 (1994) 15-28
17
nel a sample was placed which, within errors of measurement, was identical to the sample in the measuring channel (the error in thickness was ~< 2 × 10 -2 mm). In Fig. 1, curve la is the spectrum compensated by this method. All the changes in the spectrum of the sample after thermal treatments are shown clearly (Fig. 1, curve 2a). The criterion of identity of the sample surfaces after fabrication and after thermal treat-
of the measuring channel after each treatment. In the wavenumber ranges of SiOH (3680 cm- 1) and Sill (2254 cm-1), the background of the sample absorption was subtracted. The optical density of the samples with 1 mm thickness in this spectral range changed practically from the transparent region at 3000 cm -1 to 1 at 2100 cm -~. The background was taken into account with the help of the method [9] in which in the reference chan0, O,fO.
0,078~
~
~ OE6.
!!!i!
~+~
O,OJo. ~O0.O
$~PO,O
lO##.O
~$00.0
Wclvenumber(cm-t ) b O,#if
ej~
O,4Jt
e,,~la
341
e,J~
....=.
/
~14k
0,tM
P, ilia
lll.~.o IJN.O
I~O,O
~IMO
Wclvenurnber(cm-z ) Fig. 1. Effect of the structural relaxation of vitreous silica, containing 300-0.5 ppm OH groups, after the annealing process. Curves la, lb: initial spectrum; curves 2a, 2b: spectrum after annealing at 1000°C, 6 h in air. Sample thickness 0.875 mm. Curves la, 2a: differential spectrum.
18
V.Kh. Khalilov et al. /Journal of Non-Crystalline Solids 169 (1994) 15-28
ment was the coefficient of reflection in the 3000-2200 cm -~ range. The absorption in this range is small ( < 0 . 1 c m - l ) ; the change in the reflection in our case was determined only by the quality of the polished surfaces and varied from sample to sample in the limit ~<0.5%. The absorption spectra in the 200-2500 nm spectral range in the glass samples from blocks were measured in a spectrophotometer (Hitachi U-3410) and in the case of rods the measurements were made on another spectrophotometer (SF-26). The absorption spectra in VUV-spectral range were investigated on a n o t h e r s p e c t r o p h o t o m e t e r (VMR-2). The concentration of chlorine in the samples was determined by a photocalorimetric method and the concentration of alkali by the method of flame photometry. The concentration of metal impurities was measured by spectral analysis. The samples were treated at different temperatures in the 250-1000°C range, in atmospheres of dry hydrogen, dry and wet (800 ppm of O H groups) air at pressures 1-1.1 atm. The time of annealing for each temperature was chosen in consideration of spectra saturation in the limits of sensitivity of the spectrophotometers. After the treatment at each temperature in a gas, the samples were cooled to room temperature and measurements were made. 3. The process of low temperature structural relaxation
During annealing cycles of preform rods in differing atmospheres, absorption bands at 2680, 2420 and 2258 cm-1 were observed to grow with increase of annealing temperature. The maximum of the 2258 cm -1 band grew in intensity and shifted to smaller wavelength Fig. 1 (curves la, lb). The shift of maximum and the growth of the intensity at the 2258 cm -~ band, seen in Fig. 1 (curves lb, 2b), are shown on the differential curve (Fig. 1, curve 2a) in the form of a complex contour with a change of the sign of the optical density at the point where there is a crossing of the contour lines. The existence of the negative and decreasing region near 2100 cm -1 on the same curve shows the appearance of absorption
near short-wave absorption edge of the fundamental oscillation. The changes of the absorption bands at 2680, 2420 and 2258 cm -~ did not depend on concentration of O H groups and chlorine in our samples. Moreover, in the absorption spectra of the pure quartz samples which were fabricated by different methods from synthetic SiO 2 or from SiCI4, these small absorption bands exist independently of concentration of hydroxyl and chlorine. Data on these bands are found in Ref. [10]. However, everything which was written previously allows one to make a proposal about the character of their origin, which we assert is connected with the combination frequencies of the fundamental oscillations of SiO 2. It should be mentioned that the most intense band at 2258 cm -1 had a position approximately twice the fundamental band at 1120 cm -1 in the reflection spectrum. The maximum of the absorption band at 2258 cm -1 coincided, within errors of measurement, with the 2254 cm-~ band, which is due to a S i - H center [11]. To resolve these bands, the width of which is one half that of the S i - H band, and to investigate their temperature kinetics, special efforts are needed. Nevertheless, below (Sections 4-6) we show that the detectable formation of S i - H centers during anneals in H 2 takes place only in the 'dry' samples, in which the concentration of O H groups ~< 10 ppm. During annealing in an air atmosphere (wet or dry), the S i - H centers do not appear and changes in the examined region are connected only with the change of the intensity and position of the 2258 cm -1 band. The growth of the 2680, 2420 and 2458 cm -1 bands occurs in the temperature range 600-650°C and continues up to 1000°C. We show that, at these temperatures, the kinetics of transformation of the intrinsic defects in the glass change. We proceed from the assumption that the observed effects are connected with structural relaxation process. It was found that, with the increase of the annealing temperature to 1000°C, the band at 1120 cm -1 shifted to 1124 cm -1 and the intensity increased 3% (Fig. 2.) The noted changes are not big, but they appear systematically and are reproducible. Reproduction of the results was ~< 0.5% during the measurements of the reflection spectra and ~< 2.0 cm -1 in the
l~.Kh. Khalilov et al. /Journal of Non-CrystaUine Solids 169 (1994) 15-28
wave-length scale. The maximum of the 439 cm- 1 band did not change in the limit of measurement error, but increased its intensity ~ 6% from the initial value (Fig. 2). Thus the question arises: are the changes in the reflection spectra during the annealing of the samples in different atmosphere connected with the relaxation processes in the volume of the samples or are they caused by surface effects?
Since measures to preserve the purity of surfaces of the samples during annealing were made and the results dependent on the atmosphere were reproducible, we assume that the changes in the reflection spectra are connected with the relaxation process in the volumes of the samples. The influence of the fused S i O 2 s t r u c t u r e on the vibrational spectra was investigated in Ref. [12]. The main result of this work was the defini-
Is.#.
"1
~,O,
78,0.
?e.o
F\
N C~
~f.o
//
t~O
Sit,
SV,O
:....: .¢0~0 IlOaO
19
i
\,o.o avenurnber(crn tl~.O"
i~tO
50.0
SOQO
)
Iff,#~
iC cJ c-
/\ P\
L_}
DC.
/
ST.P
~0
tteqo
\ rtJgo
tto~
t¢O~O
Wavenumber( crn - i ) Fig. 2. The reflection spectra in the region of the fundamental vibration bands of the vitreous silica (300-0.5 ppm OH-groups). (a) . . . . . . , initial sample; , annealing at 1000°C, 6 h. (b) . . . . . . , initial sample; ( C o n 0.5 ppm): ~ , initial sample (Coil 100 ppm). (c) . . . . . . , annealing 1000°C, ( C o n 0.5 ppm); - - , annealing at 1000°C (Coil 100 ppm).
20
V.Kh. Khalilov et al. /Journal of Non-Crystalline Solids 169 (1994) 15-28 80-
tion of the important role of fluctuations of the Si-O-Si angle on the growth of the width of the oscillating spectra during the transformation from crystal to glass. The topology of glass structure does not greatly influence the form of the oscillating bands. The variations of the Si-O-Si angle in the + 10° limit from nominal 138° showed that the position of maximum at the 1120 cm -~ band ought to be changed ~<75 cm-1, while the 439 cm-t band did not change. Based on the experimental result of a short-wave shift of the 1120 cm-1 band and the invariance of the 439 cmband, we propose that, at temperatures >~650°C, we can observe relaxations. The small increase of the intensity of these bands we attribute to a decrease in the fluctuation of the density of states. The observed relaxation process cannot be connected with a change of the fictive temperature of the sample during the annealing process because such a change shifts the 1120 cm-t band in the short-wave direction and the 439 cm-1 band in the longwave direction [13]. The additional argument that this relaxation process is not connected with the change of the fictive temperature is that this process is the same in the wet and in the dry preform samples. The annealing time of the preforms, independent of concentration of OH groups in them, was chosen from the conditions of saturation of the processes of H interaction in the samples (~< 30 h at the temperatures 600-650°C and 16 h at 1000°C). From the results of Refs. [14,15], it may be expected that the relaxation rate with the change of concentration of OH group in samples with 5-90 ppm will differ. As a result, during the annealing time, especially at low temperatures, we do not expect the dry and wet samples to be brought to the equilibrium condition.
4. The peroxy linkage in glass and preforms
During the successive annealing in a hydrogen atmosphere of type IV glass and preforms made from this glass, a growth of Si-OH centers was observed after anneals at temperatures ~<600°C (the temperature of the beginning of relaxation process) and the decrease of OH center concen-
"•600 .,.-I
40-
0
/_/,~
20"
o
o
[]
o 0 @. . . . . . . . i,, ~ J 200 400 0 i
i
i i
i
i
[
I
i
i
i
~ i
~ i
i
i
i
i
i
600
Temperature( CXXXDOH~t(SiOH ooooo Air(SiOH ^^^^Lx Air(SiOH 00000 H~.(Si0H
i
i i
i
i
i
i
,
i
,
i
800
T oc
i
i
i
i
i
[
1000
)
initial 5ppm) initial 5ppm) initial 90ppm) initial 8 0 p p m )
Fig. 3. Effect of temperature on the rate of formation of hydroxyl in vitreous silica with different initial content of O H groups, annealed in hydrogen and wet air atmosphere.
trations after anneals from 650 to 1000°C (Figs. 3 and 4). These dependencies and the magnitude of increase at each temperature do not depend on the initial concentration of OH groups in these samples. During annealing in an atmosphere of dry air and in vacuum ( ~ 10 -3 atm), the spectrum of Si-OH centers in the samples was not observed. An increase of the spectrum of Si-OH centers was observed during annealing in the atmosphere of wet air (Fig. 3). After annealing in H 2 an increase of Si-OH centers was 70 ppm, and in the wet air atmosphere the increase was < 25 ppm. From Fig. 3, it is seen that the increase of OH groups was observed until the temperature at which the relaxation process in the samples began. The beginning of the relaxation process is correlated with the decrease of the concentration of OH groups. The latter is connected obviously with degradation of centers which were precursors of Si-OH groups. The main indication of the formation of Si-OH centers is the absence of correlation with complex kinetics of formation and degradation of Si-H
V.I~.
_Q
aO00 ' ~
Khalilov et aL /Journal of Non-Crystalline Solids 169 (1994) 15-28
. . . . .
O,Ol~
i_-]
5
-•~
JXJ/ O.O/2
aota
o,~
.- j
gooo
" ~
21
O H groups in the samples from 0.5 ppm to 500 ppm, it can be stated that the concentrations of S i - O - O - S i centers in the investigated range of samples do not differ, within the errors measurement. The author of Ref. [9] did not find any evidence of formation of S i - H centers on the background of the growth of S i - O H groups during annealing of type IV samples in hydrogen and came to the same conclusion. The glass investigated in Ref. [9] was fabricated in an oxygen plasma, had a low concentration of O H groups and nearly 200 ppm of chlorine. The author of Ref. [9], did not compare the kinetics of formation of S i - H and S i - O H centers with spectroscopic evidence of intrinsic defects• However, in Ref. [3] the absorption spectra of similar types of glass in the range from 400 to 150 nm are presented. It follows from the data of this work [3] that, in the samples investigated in Ref. [9], the concentration of defects, which are responsible for absorption bands at 163 and 248 nm, is low, which, in our opinion, is a supporting argument for the conclusion of the author in Ref. [3].
o.ooo 5. The character of the absorption band at 248 am
000
•. r -
#OOO
JO00
2000
Wdvenumber(cm-I) Fig. 4. The formation of Sill, SiOH centers during the annealing in H 2 atmosphere of the vitreous silica with the concentration hydroxyl in the initial sample 0.5 ppm. 1, 400°C; 2, 450°C; 3, 500°C; 4, 600°C; 5, 700°C; 6, 1000°C. Thickness of the sample 0•875 ram.
centers in the dry and wet glass samples. Moreover, such a dependence cannot be discovered with all spectroscopic changes, which can be observed during annealing in different atmospheres• The only process which can explain the formation of S i - O H centers in such a condition is the interaction of peroxy linkages S i - O - O - S i with hydrogen. Since the increase of S i - O H groups is small with a change of the initial concentration of
The analysis of the temperature dependence of absorption spectra of 'dry' and 'wet' samples of type IV glass in different atmospheres showed that there is a simple correlation between the decrease of oxygen-deficit centers, which form the 248 nm band [3-5], and formation of S i - H centers, and the 2254 cm-1 band. The main factors which form such a conclusion are as follows. (a) The formation of S i - H centers with a 2254 cm-1 band was observed during annealing in H 2 in only 'dry' samples of glass and preforms. In these samples of glass the 248-nm band was observed, the intensity of which decreased to zero with increase in S i - H centers during annealing from 250 to 650°C (Figs. 4 and 5). In Figs. 6 and 7, the corresponding dependencies of the intensities at 248 nm and 2254 cm-1 (Sill center) are shown• From Fig. 6, it follows that in the lowtemperature region, to 280°C, the H 2 treatment produces an increase of the annealing rate of the
V.Kh. Khalilov et al. /Journal of Non-Crystalline Solids 169 (1994) 15-28
22
7
6
5
'
~ ~...~,~7_~,~: *"
90
7o
/:.~,.-,, ~,, , ,.g/ .~\.C ~ ! lti t ~'~
2
The annealing time for each temperature was sufficient for saturation in the case of H 2 which was not so in the case of oxygen or 'water'. The main point is that the processes involving hydrogen and chlorine do not affect the 248 nm band. (c) As mentioned in the previous Section, the correlation of temperature dependencies at 248 nm band with S i - O H centers was nil (Figs. 3 and
8). C~
5"0
~-- 50
/
,° / 5 0
200
250
300
350
Wavelength(nm)
(d) At temperatures >i 650°C, the interaction of centers with absorption at 248 nm band with hydrogen proceeds completely (Fig. 6). During this reaction, the 2254 era-1 ( S i - H center) shifts to 2264 cm-1, and its intensity increases (Fig. 4). These effects show the beginning of another reaction with H z, which are be discussed below. (e) During annealing of 'wet' samples, in whose spectra the 248 nm band is absent, absorption spectra for the formation of S i - H centers are not observed. (f) The 163 nm band, which is also observed in absorption spectra in 'dry' samples, does not
Fig. 5. The influence of the annealing of silica (Coil 0.5 ppm) in H on intensity of the band with maximum at 248 nm. 1, spectrum initial sample; 2, annealing at 280°C; 3, annealing at 400°C; 4, annealing at 500°C; 5, annealing at 600°C; 6, annealing at 700°C; 7, annealing at 1000°C. Thickness of the sample 50 cm.
2.0 D " ~ 1.5
-F,-I
248 nm band. At temperatures > 280°C, the dependence is determined by interaction of H with centers which form the 248 nm band. In Fig. 8, we show the temperature dependence of the intensity at 248 nm band taking into account the annealing of this band in vacuum. From Fig. 8 it follows that the decrease of the intensity at 248 nm band is correlated with the increase of the concentration of S i - H centers (the 2254 cm -1 band). (b) In the case of annealing in 'wet' or 'dry' air, and in vacuum, a complex dependence of intensity of the 248 nm band on temperature was observed. The magnitude of the changes differs in these three mentioned cases (Figs. 6 and 9). It is connected mainly with the kinetics of the 248 nm band and was investigated with samples 14 mm in diameter and nearly 200 mm in length.
O.5 0 ox 0.0
.........
o
9,,'~,,9. . . .
, ..............
200
,oo
Temperature(
o ....
600
Q .........
800
T*C
r) ~ooo
)
CCCCO Ha o [ ~ Dry AirpV a ~ Fig. 6. The influence of the consequently annealing in different atmosphere of the vitreous silica (Coil 0.5 ppm) on intensity of the band at 248 nm. Thickness of the sample 50 em.
V..Kh. Khalilou et al. /Journal of Non-Crystalline Solids 169 (1994) 15-28
23
t00
30-
0.
__0
0
,_&."jt~ iV'*"~-~
~ "~-"25" O3 ,.Q
f8-
80.
/4
~
"#';,,'%\ / ";. i/I
20:
Z,:/
~'~ 60. (3) ~P
~)15
I
/
1:3.,
0
i&
/
~10 0 5
\.
C~
A
2O
/
n/ n /
0
200
0
600
800
Temperature(
ooo
0
T°C )
~00
Fig. 7. Effect of temperature on the rate of formation of Si-H centers in vitreous silica (initial content OH 0.5 ppm) annealed in H e atmosphere. Thickness of the sample 0.875 mm.
o - o \,*
u
0
0
200
4o0
600
~oo
Annealing T e m p e r a t u r e ( ° C CCXDOOhbsorption
~ooo
)
163 n m
cxXmno Concentration si0n ~,AmXA Absorption 248 n m
550
Fig. 9. The influence of the consequently annealing of the glass (Coil 0.5 ppm) in air on intensity of the band with maximum at 248 nm. Thickness of the sample 0.5 m. 1, spectrum initial sample; 2, annealing at 400°C; 3, annealing at 500°C; 4, annealing at 800°C; 5, annealing at 1000°C.
O
D
o
500
Wovelength(nm)
z~z~z~zxzxSIH(2254 c m - t ) 00000 SiH(2264 c m - 1 )
o
250
. . . . . Absorption Sill (2254 cm
~/
00000 Absorption SiH (2264 cm --
Fig. 8. Effect of temperature on the rate of formation (norrealized) of the intrinsic defects (163 and 248 n m absorption bands) and Si-H. S i - O H centers in vitreous silicaannealed in H 2 atmosphere.
change during annealing in H 2 to 600°C (Fig. 8). This invariance confirms the proposal that in the mentioned temperature range the formation of S i - H centers occurs only during reaction of centers at the 248 nm band with H2. The oxygen-deficit centers (248 nm band) interact with U 2 forming only S i - H centers, which have a band with a maximum at 2254 cm - ] [11]. This result is important, because it is consistent with the attribution of the oxygen deficit centers of the 248 nm band to oxygen vacancies - S i - S i - , which interact with H z according to the reaction -Si-Si- + Hz-Si=HH-Si-. In the case of the alternative model of two-coordinated silicon atoms [2], complex kinetics for the formation of centers containing hydrogen such ..H as =Si type centers are observed. "H
24
V.Kh. Khalilov et al. /Journal of Non-Crystalline Solids 169 (1994) 15-28 ~00
6. The character of the absorption band at 163 nm
,5
80.
The source of the absorption band at 163 nm is still being discussed [3,7,8]. In Ref. [8], a correlation with the peroxy radical is shown, while in Ref. [3] and Ref. [7] it is associated with an oxygen vacancy. In Ref. [5], this band which occurs after rapid cooling of Type IV samples, previously treated in H2, is attributed to a -Si~center (three-coordinated with oxygen, a nonparamagnetic Si atom). The 163 nm band and the 248 nm band are always present in samples with low concentrations of O H groups. The temperature dependence of the intensity of the 163 nm band differs from that of the 248 nm (Figs. 8 and 10). First, it may be noted that centers which form the 163 nm band are stable during annealing in vacuum and in air (wet or dry) in the temperature range 250-1000°C. Annealing in H 2 to 650°C changes the intensity of the 163 nm band. In the range 600-650°C, the intensity at this band decreases and by 700°C it is almost absent in the absorption spectrum (Figs. 8 and 10). The t e m p e r a t u r e dependence of the 163 nm band differs from the interaction of oxygen vacancies with H 2 (Section 5) and peroxy linkages (Section 4). This result excludes the possibility that these centers are related to the 163 nm. In the same temperature range (600°-650°C), growth of intensity of the 2254 c m - 1 band stops. In the temperature range 650-1000°C, the maximum of the S i - H band is shifted to 2264 cm -1 and its intensity increases to a maximum at 650°C (Figs. 4, 7 and 8). The shift of the maximum of the S i - H band to 2264 c m - 1 shows an interaction of hydrogen and a new type of center. The temperature dependence of the S i - H center at 2264 c m - 1 band and the 163 nm band are correlated (Fig. 8). It follows that the appearance of the 2264 cm -1 band is connected with interaction of hydrogen with centers which form the absorption band at 163 nm. In Fig. 4 we show absorption spectra relative to the initial spectrum. For this reason the increase of the intensity of S i - H band in the region 650°-1000°C is observed on the background of the increase and change of maximum position of
Ln
c--
/
60
i!'2
D
/
E O'3 C-CD I
/ / g
20
.}' .~
•
~o
150
150
170
180
igo
Wovelength(nm) Fig. 10. The influence of the annealing of the glass (Coil 0.5 ppm) in H z on intensity of the band with maximum at 163 nm. Thickness of the sample 0.875 mm. 1, spectrum initial sample: 2, annealing at 500°C; 3, annealing at 600°C; 4, annealing at 700°C; 5, annealing at 1000°C.
the intrinsic band at 2258 cm -1 (Fig. 1, Section 3). From comparison of Fig. l(curve 2a) and Fig. 4, it follows that the contribution of this band is not large compared with the contribution of the S i - H band. Nevertheless we have made an experiment which accounts for the 'background' and resolves the S i - H band in the temperature range 700 °1000°C. The results are shown in Fig. 11. Curve (a) in this figure corresponds to the spectrum for 'dry' glass. The samples were annealed at 1000°C in a dry air atmosphere before they were put in the comparison and measuring channels of spectrophotometer (see Section 2). It is seen that bands discussed in Section 3 are compensated on the curve. (The weak 3680 c m - 1 band shows the small difference of the initial concentration of O H groups in the reference sample and the sampie.) Then the sample from m e a s u r e m e n t channel was annealed in H a at 800°C. The S i - H center which appears during annealing in H 2 at temperatures > 700°C is shown by the band with
V.Kh. Khalilov et al. /Journal of Non-Crystalline Solids 169 (1994) 15-28
f.2 "7
25
ci iL
QO _. ~ , W,, L-
i -- ~,~ r,t,w
km m,, ~
tq m
,,2 ~
....
2
. . . . . . .
~t2-, o
Wavenurnber(cm- 1)
Wavenumber(crn-t) Fig. 11. The appearance and annealing S i - H center with maximum at 2264 cm - I in 'dry' and 'wet' glass without structural relaxation process (in channel of comparison of spectrophotometer; sample annealed at 1000°C, 9 h in air). (a) Coil ~ 8 ppm. (b) C o n ~ 100 ppm. (la), (lb): 1000°C, 9 h, air; (2a), (2b): 1000°C + 800°C, 16 h, HE; (3a) 1000°C + 1000°C, 6 h, air.
maximum at 2264 cm-1 (curve(2a), Fig. 11). After annealing the same sample in dry air at 1000°C for 6 h the band at 2264 cm-1 (Fig. 11 curve(3a)) was not detected. Simultaneously, the intensity at 163 nm band was restored to its initial value. For comparison, on the same figure spectra are shown
(curves (lb) and (2b)) for 'wet' samples, in which the band at 163 nm is absent. As mentioned above, in this case the spectrum of the S i - H center is not detected. All the results can be explained by the model, which we proposed in Ref. [6]. In this model, the
f
(3.) gO ¢-" ED
I -
" //~
/
~,
(J3 C-'E~ 4o
5 I
I---2O
2t]
tO
10 5 o gSO
aa0
35o
Wovelength(nm)
0
,so
~ 3~ Wovelength(nm)
Fig. 12. The influence of the annealing in H 2 atmosphere of the 'wet' - 'dry' vitreous silica. Thickness of the sample 50 cm. (a) Initial content of OH groups 100 ppm. 1, initial sample; 2, annealed in H 2 750°C, 16 h. (b) Initial content of OH groups 0.5 ppm. 1, initial sample: 2, annealed in H E 750°C, 16 h.
26
V.Kh. Khalilov et aL /Journal of Non-Crystalline Solids 169 (1994) 15-28
three-coordinated non-paramagnetic silicon has a band at 163 nm. The charge of this center is compensated by interstitial chlorine ions: -Si ÷ CI(T~ center). The affinity between chlorine and electron (3.8 eV) causes the charge stability of the center. Low mobility of chlorine causes thermal stability. The interaction center with hydrogen is: - S i + C I - + H ° ~ - S i l l + C1°.
~w
/ ,
_Q
/']
g ._~
/
C~ t,/0u
7. Spectroscopic realization of HCI and C! 2 molecules in the glass, preforms and fibers
/
/
/ / ~000
Molecules of C12 in gas, in liquid and in solid matrix have been observed by a band near 330 nm. The amplitude and half-width of this band depends on its environment [16,17]. The presence of the band at 330 nm in the preforms made by VAD method was mentioned in Ref. [18]. The authors also attribute it to the C12 molecule. However, the position of maximum and nearly equal half-width is insufficient evidence for attributing the 330 nm band to molecular C12. The 330 nm band with a change of concentration of O H groups in the glass and preforms from 0.5 to 500 ppm is observed in the 'wet' samples (Fig. 12). The results of the chemical analysis show that the total concentration of C1 in the 'wet' preforms decreases and is 300-500 ppm. If the 330 nm band is formed by CI 2 molecules, in these samples it is located usually at interstitial positions. Samples of preforms with a 330 nm band were annealed in H 2 and dry air for 16 h at 750°C. The temperature was chosen so that structural relaxation was small. The band at 330 nm was not detected in the transmission spectrum after H treatment (Fig. 12). A treatment in dry air did not change the intensity of the 330 nm band at temperatures < 1000°C. In support of the thesis that the absence of the band at 330 nm is connected with the formation of HCI by an interaction between H 2 and C12, the IR absorption spectrum of samples with a length of 100 mm was measured in the region of the bands of HCI molecules. This region is in the range 2700-2900 cm-1 depending on the matrix and interstitial site of HC1 [19].
/
/
/
/ Wavenumher(cm - t)
/
~800
~ave number(cm -I )
Fig. 13. The absorption spectra in the region of vibration band of molecule HCI. Thickness of the sample 90 ram. (a) Samples annealed in H 750°C, 16 h. - . . . . . , initial content O H groups 100 ppm. initial content O H groups 0.5 ppm. (b) Initial content OI~I groups 0.5 ppm. - . . . . . , initial. annealed in H 2 750°C, 16 h.
It was found that in those samples with a band at 330 nm annealing produced a band at 2800 cm -1 (Fig. 13). In the case of annealing in the same conditions of 'dry' preforms without a band at 330 nm (Fig. 12), there were no detectable changes in the region 2700-2900 cm-~ (Fig. 13). The results show that, in the 'wet' samples, chlorine usually is in an interstitial site in the form of C12 or CI and after interaction with hydrogen forms HC1. With a higher content of chlorine in the 'dry' samples, in our opinion a structurally connected condition, -SiCI, is present. In our opinion, in the 'dry' samples the concentration of oxygen deficit defects is large (see Section 4) and they react with the dissolved chlorine. On the other hand, in the samples with O H concentration nearly 100 ppm and larger, the spectra of intrinsic defects are not detectable. In the region of long-wavelength edge of fundamental absorption of this glass, no absorption bands are observed. In these samples, after the HCI was formed during annealing in H 2, the appearance of a weak band at 165 nm was observed (Fig. 14). The condition of appearance of this band and the position of its maximum indicate that it differs from the 163 nm band. The 165 nm band is a characteristic of the HCI molecule [15] and the
V..Kh. Khalilov et al. /Journal of Non-Crystalline Solids 169 (1994) 15-28
TX
,f,,
f
f
//
"--d~
/
E o')
,r
¢
/ 0
~0
t70 ~o Wavelength(nm)
19o
~oo
Fig. 14. The transmission spectra of the glass (Con 0.5 ppm) before ( ) and after ( - - - - - - ) annealing in H 2 at 1000°C, 6 h.
n, the fundamental absorption edge is shifted to longer wavelength. It is shown [1] that, with a decrease of the concentration of O H groups in the glass, the refractive index in the visible region of the spectrum increases. At the same time, the edge of fundamental absorption shifts to the short-wave region of the spectrum [1,13]. So, for the case of very 'dry' glass in regions of optical heterogeneity, a long wavelength shift of the absorption edge with decrease of O H concentration is not expected. We have measured n in preforms with accuracy of + 10 -5 as a function of concentration of O H groups from 0.5 to 300 ppm. It was found that in the region of 2 ppm the linear dependence has a slope, which shows that in case of low concentration of O H groups the variation of n is determined not only by variation of O H concentration. This conclusion is the same as in Ref. [1]. Our samples of glass were fused by one method, with large differences in concentrations of hydroxyl and chlorine. These differences allowed us to obtain more accurate results and to determine
conditions for its appearance in the spectra is additional evidence of the formation of the HCI molecule.
8. The character of the complex heterogeneity in type IV quartz glass It is well known that, in the type IV glass in the direction perpendicular to the surface of growth, a heterogeneity of the refractive index, An, is observed [1,20]. In Refs. [1,20], it was shown that this phenomenon cannot be explained by the distribution of chlorine and hydroxyl groups nor by variations of the fictive temperature. The data in Refs. [1,13] show that the magnitude of the variation of the refractive index showing the optical heterogeneity is An ~ 10 -4. In regions of sample with increased n, the concentration of chlorine reaches 1000 ppm, and the concentration of O H groups is < 1 ppm (the average concentrations of chlorine is ~ 100 ppm and of O H groups ~ 40 ppm). In Refs. [1,13] it is also shown that, in the region of glass with increased
27
40
O
E 3o. (J
"~ 2o 0 (.._)
~ .2
/! tt
o 10. m
~,
0 t
t60
55
['70
Wavelength(nm) Fig. 15. The long wavelength edge of the fundamental absorption after annealing in H 2 at 800°C for 16 h. 1, COH 5 ppm; 2, COIl 100 ppm.
28
V.Kh. Khalilov et al. /Journal of Non-Crystalline Solids 169 (1994) 15-28
how the changes of refractive index in 'dry' and 'wet' samples are affected by differences in concentration of defects. Samples with high (100 ppm) and low (5 ppm) concentration of O H groups were annealed in H2. As mentioned above (Section 3), the structural refraction process in these samples is the same, and the position and intensity of fundamental infra-red bands do not differ. Simultaneously in the region of the UV-edge of the fundamental absorption, after annealing in H atmosphere, the bands at 163 nm and 248 nm are suppressed. In these samples there is a possibility to compare the long-wavelength edge of the fundamental absorption, not distorted, by defect absorption bands. The results of comparison confirm the results which were reported before in Ref. [1] about the long-wavelength shift of the edge of fundamental absorption in the dry samples (Fig. 15). However the analysis of the results which were shown in Ref. [1] with the same thermal prehistory and the identical spectra of fundamental infra-red bands brings us to the conclusion that the increase of n in 'dry' glass is due to an increase in the density of states in the 'tail' in the forbidden gap, the properties of which remain unknown.
9. Conclusion The temperature dependencies, which reflect the interaction between hydrogen and intrinsic defects in glass, indicate the existence of a temperature region 600-650°C which delineates two different effects of annealing. At these temperatures we observed significant changes in defect transformations by changes in bands at 248 and 163 nm. Significant changes of structure in this temperature range occur with S i - H and S i - O H centers. These differences are due to the appearance of structural relaxation processes at 600650°C for which mechanisms are not determined.
These processes are not correlated with a change of fictive temperature. The formation of the hydrogen- (Si-H) and oxygen-deficit centers also produced a new S i - H center and showed that the band at 248 nm is caused by the oxygen vacancy - S i - S i - ; the band at 163 nm is caused by threecoordinated non-paramagnetic silicon: -Si ÷ (T~center).
10. References [1] V.H. Khalilov, G.A. Dorphman, L.G. Karpov, I.V. Pevnitsky and V.V. Zahov, Phys. Chem. Glass (USSR) 13 (1987) 721. [2] L.N. Skuja, A.W. Streletsky and A.B. Pakovich, Solid State Commun. 5 (1984) 1069. [3] R. Tohmon, J. Jamasaka, K. Nagasawa, J. Ohki and J. Hama, J. Non-Cryst. Solids 95&96 (1987) 671. [4] G.W. Arnold, JEEE Trans. Nucl. Sci. NS-20 (1973) 220. [5] K. Nagasava, Y. Hoshi and Y. Ohki, Jpn. Appl. Phys. 26 (1987) L554. [6] V.H. Khalilov and I.V. Pevnitsky, Phys. Chem. Glass (USSR) 11 (1985) 504. [7] R.P. Gupta, Phys. Rev. B33 (1986) 7274. [8] M. Stepelbrock, D.L. Griscom and E.J. Frieble, J. NonCryst. Solids 32 (1979) 313. [9] J.E. Shelby, J. Appl. Phys. 51 (1980) 2589. [10] A.R. Silin, Phys. Chem. Glass (USSR) 4 (1978) 266. [11] G. Van Der Steen and H. Van Der Boom, J. Non-Cryst. Solids 23 (1977) 279. [12] R.B. Langhlin and J.D. Joannopoulos, Phys. Rev. B16 (1983) 2942. [13] A.E. Geissberger and F.L. Galeener, Phys. Rev. B28 (1983) 3266. [14] V.K. Leko and E.V. Mesherakova, Phys. Chem. Glass (USSR) 1 (1975) 447. [15] J.C. Mikkelsen and F.L. Galeener, J. Non-Cryst. Solids 37 (1980) 71. [16] H. Okabe, Photochemistry Small Molecule (World, Moskow, 1981). [17] D. Kalvert and D. Pitts, Photochemistry(World, Moskow, 1968). [18] A.D. Abramov, E.V. Anoikin, E.M. Dianov, et al., Highclean Substance (USSR) (1987) 12. [19] A.J. Barnes, S. Susuki, H.E. Hallam and G.F. Sorimshaw, Trans. Faraday Soc. 65 (1969) 3159. [20] I.V. Pevnitsky and V.H. Khalilov, Phys. Chem. Glass (USSR) 16 (1990) 668.
ELSEVIER
Journal of Non-Crystalline Solids 169 (1994) 15-28
Character, mechanism of formation and transformation of point defects in type IV silica glass V . K h . K h a l i l o v *, G . A . D o r f m a n , E . B . D a n i l o v , M . I . G u s k o v , V . E . E r m a k o v Steko Ltd., Piskarevsky pr. 63, 195273 St. Petersburg, Russian Federation
(Received 9 August 1991;revised manuscript received 30 September 1993)
Abstract
In the temperature range 600-650°C, transformations of defects in quartz glass are observed. The dependence of a band at 163 nm on annealing temperature is measured. Changes of structure in this temperature range are observed for the -Si-H and -Si-OH centers. These are caused by the appearence of a structural relaxating process at ~ 675°C, the character of which is not determined. It is shown that the band at 163 nm is caused by three-coordinated non-paramagnetic silicon Si~-(T~- center). The band at 248 nm is attributed to an oxygen vacancy -----~i-Si=. It is shown that at temperature < 650°C the formation of Si-H centers (absorption band at 2254 cm-1) is due to the interaction of H 2 with oxygen vacancies. At temperatures > 650°C, the interaction of H 2 with T~- centers was achieved. The maximum of the band of the -=Si-H centers thus formed is at 2264 cm- 1. It was assumed that, in the type IV glass, the formation of Si-OH is due to an interaction of H 2 with peroxy linkages. Changes of the fundamental absorption edge are correlated with the changes of concentration of OH groups from 1500 ppm to 2 ppm. In the region < 2 ppm, the linear dependence of these changes on OH concentration is violated.
I. Introduction
It is well known [1] that the properties of pure quartz glass, independently of the oxidation methods of SiCI4, which are used in most of the technological processes, are sensitively influenced by fluctuations of the fusion conditions. This influence results not only in inhomogeneous distribution of impurities but is also accompanied by an inhomogeneous distribution of the intrinsic defects [1]. Moreover, there is a fluctuation of the glass structure [1]. Information about the main
* Corresponding author. Tel: + 7-812 249 2990. Telefax: + 7812 249 3510.
structural factors which make inhomogeneities in these types of glass is very important for improving the process. On the other hand, the possibility of producing quartz glass with different properties by one method but with different properties has its own interest with respect to investigating intrinsic defects and their transformations and the effect of characteristic impurity centers such as SiOH, Sill, SiCI and dissolved gases such as O 2. This work presents the results of an investigation of samples fabricated by oxidation of SiCI 4 in a low temperature oxygen-nitrogen plasma (type IV). The concentration of O H groups in the glass ranged from 0.5 to 300 ppm and chlorine from 1000 to 50 ppm. The quantity of metallic
0022-3093/94/$07.00 © 1994 Elsevier Science B.V. All rights reserved SSDI 0022-3093(93)E0275-D
16
V.Kh. Khalilov et aL /Journal of Non-Crystalline Solids 169 (1994) 15-28
impurities was less than 1 ppm by weight. Some samples were fabricated with differing concentrations of hydroxyl molecules and chlorine ions. The content of defect centers (known as oxygendeficit centers with absorption bands at 248 nm and 163 nm) in fabricated glass differed. The amplitudes of these bands were observed in the samples with low concentration of OH group and were absent in samples with concentrations of hydroxyl near 300 ppm. The source of the 248 nm absorption band is still the subject of discussion. Among the models for the center which are discussed now in the literature, the most probable are a two-coordinated silicon [2] and an oxygen vacancy [3-5]. In our opinion, the center associated with the 163 nm band is a three-coordinated non-paramagnetic silicon (T~ center) [6]. In Refs. [3] and [7], the 163 nm band is connected with oxygen vacancies, but in Ref. [8] its correlation with the peroxy radical was shown. Information about the structure of oxygen-deficit centers in quartz glass has been developed by using hydrogen as a probe, the spectroscopic manifestation of which, in the form of Sill and SiOH, is well known. In this paper, an investigation of the temperature dependences of absorption spectra for some of the fabricated samples after treatment by H 2 in the temperature range (from 200 to 1000°C) is reported. Using the same samples, investigations were made of absorption and reflection spectra, including the long wavelength edge of interband fundamental absorption of the 150 nm band, the spectroscopic properties of intrinsic and impurity defects, the oscillation spectra of hydrogen centers, and the region of fundamental intrinsic oscillations. The spectroscopic investigations of some of the samples, fabricated by one method and in which there were different equilibrium concentrations of impurity-defect centers, provided an opportunity to obtain some data on transformations of these centers. In particular, experimental results show a change of the transformation process of the impurity-defect centers in the glass, which is connected with the relaxation processes in the temperature range 650-700°C. It should be mentioned that the investigation of the processes
of interaction of glass with H 2 in this temperature range, with the control of concentration of impurities such as C12 and H 2 has a significance, particularly for the production and use of optical fibers.
2. Experimental The dependence of the optical densities of samples in the form of platelets in the investigated spectral range was measured. In the spectral range 150-400 nm and IR range 4000-2000 cm -~ samples with 0.8-1.0 mm thickness were used. In the 200-2000 nm spectral range, preform rods with lengths 200-1000 mm and diameter 20 mm or samples of glass from the blocks with diameter 30 mm and with length up to 100 mm were used. To prevent the cylindrical surface from affecting the measurements of absorption spectra in the long preform rods, the surfaces were treated with HF and ~<0.3 mm were removed, after which the surfaces were firepolished. After these treatments, the influence of the side surface on the measurements of transmission in the sample was not greater than the spectrophotometric measurement error. In the case of investigation of preform rods, the samples for measurement in VUV and IR spectral range were made with a thickness 0.8-1.0 mm from the samples previously treated by HF and firepolished. The measurements of absorption spectra of the rods were made by the classical double-beam method. Special measures were undertaken to maintain the diameter and the position of the light spot on the photodetector in the measuring and comparison channels. As a result, the measuring error was ~< + 1% (_+0.04 abc) in VUV and UV-VIS regions. The absorption and reflection spectra in the IR range were measured by a spectrophotometer (Shimadzu IR-470). As the process of measurement consisted in comparison of the same sample in the initial state and after cycles of treatment, the main problem was the reproducibility of measurements. The reproducibility was ~ 2.0 cm-1 in the wavelength scale and photometric accuracy _+0.002 abc. The sample was placed in one place
I(.Kh. Khalilou et aL /Journal of Non-Crystalline Solids 169 (1994) 15-28
17
nel a sample was placed which, within errors of measurement, was identical to the sample in the measuring channel (the error in thickness was ~< 2 × 10 -2 mm). In Fig. 1, curve la is the spectrum compensated by this method. All the changes in the spectrum of the sample after thermal treatments are shown clearly (Fig. 1, curve 2a). The criterion of identity of the sample surfaces after fabrication and after thermal treat-
of the measuring channel after each treatment. In the wavenumber ranges of SiOH (3680 cm- 1) and Sill (2254 cm-1), the background of the sample absorption was subtracted. The optical density of the samples with 1 mm thickness in this spectral range changed practically from the transparent region at 3000 cm -1 to 1 at 2100 cm -~. The background was taken into account with the help of the method [9] in which in the reference chan0, O,fO.
0,078~
~
~ OE6.
!!!i!
~+~
O,OJo. ~O0.O
$~PO,O
lO##.O
~$00.0
Wclvenumber(cm-t ) b O,#if
ej~
O,4Jt
e,,~la
341
e,J~
....=.
/
~14k
0,tM
P, ilia
lll.~.o IJN.O
I~O,O
~IMO
Wclvenurnber(cm-z ) Fig. 1. Effect of the structural relaxation of vitreous silica, containing 300-0.5 ppm OH groups, after the annealing process. Curves la, lb: initial spectrum; curves 2a, 2b: spectrum after annealing at 1000°C, 6 h in air. Sample thickness 0.875 mm. Curves la, 2a: differential spectrum.
18
V.Kh. Khalilov et al. /Journal of Non-Crystalline Solids 169 (1994) 15-28
ment was the coefficient of reflection in the 3000-2200 cm -~ range. The absorption in this range is small ( < 0 . 1 c m - l ) ; the change in the reflection in our case was determined only by the quality of the polished surfaces and varied from sample to sample in the limit ~<0.5%. The absorption spectra in the 200-2500 nm spectral range in the glass samples from blocks were measured in a spectrophotometer (Hitachi U-3410) and in the case of rods the measurements were made on another spectrophotometer (SF-26). The absorption spectra in VUV-spectral range were investigated on a n o t h e r s p e c t r o p h o t o m e t e r (VMR-2). The concentration of chlorine in the samples was determined by a photocalorimetric method and the concentration of alkali by the method of flame photometry. The concentration of metal impurities was measured by spectral analysis. The samples were treated at different temperatures in the 250-1000°C range, in atmospheres of dry hydrogen, dry and wet (800 ppm of O H groups) air at pressures 1-1.1 atm. The time of annealing for each temperature was chosen in consideration of spectra saturation in the limits of sensitivity of the spectrophotometers. After the treatment at each temperature in a gas, the samples were cooled to room temperature and measurements were made. 3. The process of low temperature structural relaxation
During annealing cycles of preform rods in differing atmospheres, absorption bands at 2680, 2420 and 2258 cm-1 were observed to grow with increase of annealing temperature. The maximum of the 2258 cm -1 band grew in intensity and shifted to smaller wavelength Fig. 1 (curves la, lb). The shift of maximum and the growth of the intensity at the 2258 cm -~ band, seen in Fig. 1 (curves lb, 2b), are shown on the differential curve (Fig. 1, curve 2a) in the form of a complex contour with a change of the sign of the optical density at the point where there is a crossing of the contour lines. The existence of the negative and decreasing region near 2100 cm -1 on the same curve shows the appearance of absorption
near short-wave absorption edge of the fundamental oscillation. The changes of the absorption bands at 2680, 2420 and 2258 cm -~ did not depend on concentration of O H groups and chlorine in our samples. Moreover, in the absorption spectra of the pure quartz samples which were fabricated by different methods from synthetic SiO 2 or from SiCI4, these small absorption bands exist independently of concentration of hydroxyl and chlorine. Data on these bands are found in Ref. [10]. However, everything which was written previously allows one to make a proposal about the character of their origin, which we assert is connected with the combination frequencies of the fundamental oscillations of SiO 2. It should be mentioned that the most intense band at 2258 cm -1 had a position approximately twice the fundamental band at 1120 cm -1 in the reflection spectrum. The maximum of the absorption band at 2258 cm -1 coincided, within errors of measurement, with the 2254 cm-~ band, which is due to a S i - H center [11]. To resolve these bands, the width of which is one half that of the S i - H band, and to investigate their temperature kinetics, special efforts are needed. Nevertheless, below (Sections 4-6) we show that the detectable formation of S i - H centers during anneals in H 2 takes place only in the 'dry' samples, in which the concentration of O H groups ~< 10 ppm. During annealing in an air atmosphere (wet or dry), the S i - H centers do not appear and changes in the examined region are connected only with the change of the intensity and position of the 2258 cm -1 band. The growth of the 2680, 2420 and 2458 cm -1 bands occurs in the temperature range 600-650°C and continues up to 1000°C. We show that, at these temperatures, the kinetics of transformation of the intrinsic defects in the glass change. We proceed from the assumption that the observed effects are connected with structural relaxation process. It was found that, with the increase of the annealing temperature to 1000°C, the band at 1120 cm -1 shifted to 1124 cm -1 and the intensity increased 3% (Fig. 2.) The noted changes are not big, but they appear systematically and are reproducible. Reproduction of the results was ~< 0.5% during the measurements of the reflection spectra and ~< 2.0 cm -1 in the
l~.Kh. Khalilov et al. /Journal of Non-CrystaUine Solids 169 (1994) 15-28
wave-length scale. The maximum of the 439 cm- 1 band did not change in the limit of measurement error, but increased its intensity ~ 6% from the initial value (Fig. 2). Thus the question arises: are the changes in the reflection spectra during the annealing of the samples in different atmosphere connected with the relaxation processes in the volume of the samples or are they caused by surface effects?
Since measures to preserve the purity of surfaces of the samples during annealing were made and the results dependent on the atmosphere were reproducible, we assume that the changes in the reflection spectra are connected with the relaxation process in the volumes of the samples. The influence of the fused S i O 2 s t r u c t u r e on the vibrational spectra was investigated in Ref. [12]. The main result of this work was the defini-
Is.#.
"1
~,O,
78,0.
?e.o
F\
N C~
~f.o
//
t~O
Sit,
SV,O
:....: .¢0~0 IlOaO
19
i
\,o.o avenurnber(crn tl~.O"
i~tO
50.0
SOQO
)
Iff,#~
iC cJ c-
/\ P\
L_}
DC.
/
ST.P
~0
tteqo
\ rtJgo
tto~
t¢O~O
Wavenumber( crn - i ) Fig. 2. The reflection spectra in the region of the fundamental vibration bands of the vitreous silica (300-0.5 ppm OH-groups). (a) . . . . . . , initial sample; , annealing at 1000°C, 6 h. (b) . . . . . . , initial sample; ( C o n 0.5 ppm): ~ , initial sample (Coil 100 ppm). (c) . . . . . . , annealing 1000°C, ( C o n 0.5 ppm); - - , annealing at 1000°C (Coil 100 ppm).
20
V.Kh. Khalilov et al. /Journal of Non-Crystalline Solids 169 (1994) 15-28 80-
tion of the important role of fluctuations of the Si-O-Si angle on the growth of the width of the oscillating spectra during the transformation from crystal to glass. The topology of glass structure does not greatly influence the form of the oscillating bands. The variations of the Si-O-Si angle in the + 10° limit from nominal 138° showed that the position of maximum at the 1120 cm -~ band ought to be changed ~<75 cm-1, while the 439 cm-t band did not change. Based on the experimental result of a short-wave shift of the 1120 cm-1 band and the invariance of the 439 cmband, we propose that, at temperatures >~650°C, we can observe relaxations. The small increase of the intensity of these bands we attribute to a decrease in the fluctuation of the density of states. The observed relaxation process cannot be connected with a change of the fictive temperature of the sample during the annealing process because such a change shifts the 1120 cm-t band in the short-wave direction and the 439 cm-1 band in the longwave direction [13]. The additional argument that this relaxation process is not connected with the change of the fictive temperature is that this process is the same in the wet and in the dry preform samples. The annealing time of the preforms, independent of concentration of OH groups in them, was chosen from the conditions of saturation of the processes of H interaction in the samples (~< 30 h at the temperatures 600-650°C and 16 h at 1000°C). From the results of Refs. [14,15], it may be expected that the relaxation rate with the change of concentration of OH group in samples with 5-90 ppm will differ. As a result, during the annealing time, especially at low temperatures, we do not expect the dry and wet samples to be brought to the equilibrium condition.
4. The peroxy linkage in glass and preforms
During the successive annealing in a hydrogen atmosphere of type IV glass and preforms made from this glass, a growth of Si-OH centers was observed after anneals at temperatures ~<600°C (the temperature of the beginning of relaxation process) and the decrease of OH center concen-
"•600 .,.-I
40-
0
/_/,~
20"
o
o
[]
o 0 @. . . . . . . . i,, ~ J 200 400 0 i
i
i i
i
i
[
I
i
i
i
~ i
~ i
i
i
i
i
i
600
Temperature( CXXXDOH~t(SiOH ooooo Air(SiOH ^^^^Lx Air(SiOH 00000 H~.(Si0H
i
i i
i
i
i
i
,
i
,
i
800
T oc
i
i
i
i
i
[
1000
)
initial 5ppm) initial 5ppm) initial 90ppm) initial 8 0 p p m )
Fig. 3. Effect of temperature on the rate of formation of hydroxyl in vitreous silica with different initial content of O H groups, annealed in hydrogen and wet air atmosphere.
trations after anneals from 650 to 1000°C (Figs. 3 and 4). These dependencies and the magnitude of increase at each temperature do not depend on the initial concentration of OH groups in these samples. During annealing in an atmosphere of dry air and in vacuum ( ~ 10 -3 atm), the spectrum of Si-OH centers in the samples was not observed. An increase of the spectrum of Si-OH centers was observed during annealing in the atmosphere of wet air (Fig. 3). After annealing in H 2 an increase of Si-OH centers was 70 ppm, and in the wet air atmosphere the increase was < 25 ppm. From Fig. 3, it is seen that the increase of OH groups was observed until the temperature at which the relaxation process in the samples began. The beginning of the relaxation process is correlated with the decrease of the concentration of OH groups. The latter is connected obviously with degradation of centers which were precursors of Si-OH groups. The main indication of the formation of Si-OH centers is the absence of correlation with complex kinetics of formation and degradation of Si-H
V.I~.
_Q
aO00 ' ~
Khalilov et aL /Journal of Non-Crystalline Solids 169 (1994) 15-28
. . . . .
O,Ol~
i_-]
5
-•~
JXJ/ O.O/2
aota
o,~
.- j
gooo
" ~
21
O H groups in the samples from 0.5 ppm to 500 ppm, it can be stated that the concentrations of S i - O - O - S i centers in the investigated range of samples do not differ, within the errors measurement. The author of Ref. [9] did not find any evidence of formation of S i - H centers on the background of the growth of S i - O H groups during annealing of type IV samples in hydrogen and came to the same conclusion. The glass investigated in Ref. [9] was fabricated in an oxygen plasma, had a low concentration of O H groups and nearly 200 ppm of chlorine. The author of Ref. [9], did not compare the kinetics of formation of S i - H and S i - O H centers with spectroscopic evidence of intrinsic defects• However, in Ref. [3] the absorption spectra of similar types of glass in the range from 400 to 150 nm are presented. It follows from the data of this work [3] that, in the samples investigated in Ref. [9], the concentration of defects, which are responsible for absorption bands at 163 and 248 nm, is low, which, in our opinion, is a supporting argument for the conclusion of the author in Ref. [3].
o.ooo 5. The character of the absorption band at 248 am
000
•. r -
#OOO
JO00
2000
Wdvenumber(cm-I) Fig. 4. The formation of Sill, SiOH centers during the annealing in H 2 atmosphere of the vitreous silica with the concentration hydroxyl in the initial sample 0.5 ppm. 1, 400°C; 2, 450°C; 3, 500°C; 4, 600°C; 5, 700°C; 6, 1000°C. Thickness of the sample 0•875 ram.
centers in the dry and wet glass samples. Moreover, such a dependence cannot be discovered with all spectroscopic changes, which can be observed during annealing in different atmospheres• The only process which can explain the formation of S i - O H centers in such a condition is the interaction of peroxy linkages S i - O - O - S i with hydrogen. Since the increase of S i - O H groups is small with a change of the initial concentration of
The analysis of the temperature dependence of absorption spectra of 'dry' and 'wet' samples of type IV glass in different atmospheres showed that there is a simple correlation between the decrease of oxygen-deficit centers, which form the 248 nm band [3-5], and formation of S i - H centers, and the 2254 cm-1 band. The main factors which form such a conclusion are as follows. (a) The formation of S i - H centers with a 2254 cm-1 band was observed during annealing in H 2 in only 'dry' samples of glass and preforms. In these samples of glass the 248-nm band was observed, the intensity of which decreased to zero with increase in S i - H centers during annealing from 250 to 650°C (Figs. 4 and 5). In Figs. 6 and 7, the corresponding dependencies of the intensities at 248 nm and 2254 cm-1 (Sill center) are shown• From Fig. 6, it follows that in the lowtemperature region, to 280°C, the H 2 treatment produces an increase of the annealing rate of the
V.Kh. Khalilov et al. /Journal of Non-Crystalline Solids 169 (1994) 15-28
22
7
6
5
'
~ ~...~,~7_~,~: *"
90
7o
/:.~,.-,, ~,, , ,.g/ .~\.C ~ ! lti t ~'~
2
The annealing time for each temperature was sufficient for saturation in the case of H 2 which was not so in the case of oxygen or 'water'. The main point is that the processes involving hydrogen and chlorine do not affect the 248 nm band. (c) As mentioned in the previous Section, the correlation of temperature dependencies at 248 nm band with S i - O H centers was nil (Figs. 3 and
8). C~
5"0
~-- 50
/
,° / 5 0
200
250
300
350
Wavelength(nm)
(d) At temperatures >i 650°C, the interaction of centers with absorption at 248 nm band with hydrogen proceeds completely (Fig. 6). During this reaction, the 2254 era-1 ( S i - H center) shifts to 2264 cm-1, and its intensity increases (Fig. 4). These effects show the beginning of another reaction with H z, which are be discussed below. (e) During annealing of 'wet' samples, in whose spectra the 248 nm band is absent, absorption spectra for the formation of S i - H centers are not observed. (f) The 163 nm band, which is also observed in absorption spectra in 'dry' samples, does not
Fig. 5. The influence of the annealing of silica (Coil 0.5 ppm) in H on intensity of the band with maximum at 248 nm. 1, spectrum initial sample; 2, annealing at 280°C; 3, annealing at 400°C; 4, annealing at 500°C; 5, annealing at 600°C; 6, annealing at 700°C; 7, annealing at 1000°C. Thickness of the sample 50 cm.
2.0 D " ~ 1.5
-F,-I
248 nm band. At temperatures > 280°C, the dependence is determined by interaction of H with centers which form the 248 nm band. In Fig. 8, we show the temperature dependence of the intensity at 248 nm band taking into account the annealing of this band in vacuum. From Fig. 8 it follows that the decrease of the intensity at 248 nm band is correlated with the increase of the concentration of S i - H centers (the 2254 cm -1 band). (b) In the case of annealing in 'wet' or 'dry' air, and in vacuum, a complex dependence of intensity of the 248 nm band on temperature was observed. The magnitude of the changes differs in these three mentioned cases (Figs. 6 and 9). It is connected mainly with the kinetics of the 248 nm band and was investigated with samples 14 mm in diameter and nearly 200 mm in length.
O.5 0 ox 0.0
.........
o
9,,'~,,9. . . .
, ..............
200
,oo
Temperature(
o ....
600
Q .........
800
T*C
r) ~ooo
)
CCCCO Ha o [ ~ Dry AirpV a ~ Fig. 6. The influence of the consequently annealing in different atmosphere of the vitreous silica (Coil 0.5 ppm) on intensity of the band at 248 nm. Thickness of the sample 50 em.
V..Kh. Khalilou et al. /Journal of Non-Crystalline Solids 169 (1994) 15-28
23
t00
30-
0.
__0
0
,_&."jt~ iV'*"~-~
~ "~-"25" O3 ,.Q
f8-
80.
/4
~
"#';,,'%\ / ";. i/I
20:
Z,:/
~'~ 60. (3) ~P
~)15
I
/
1:3.,
0
i&
/
~10 0 5
\.
C~
A
2O
/
n/ n /
0
200
0
600
800
Temperature(
ooo
0
T°C )
~00
Fig. 7. Effect of temperature on the rate of formation of Si-H centers in vitreous silica (initial content OH 0.5 ppm) annealed in H e atmosphere. Thickness of the sample 0.875 mm.
o - o \,*
u
0
0
200
4o0
600
~oo
Annealing T e m p e r a t u r e ( ° C CCXDOOhbsorption
~ooo
)
163 n m
cxXmno Concentration si0n ~,AmXA Absorption 248 n m
550
Fig. 9. The influence of the consequently annealing of the glass (Coil 0.5 ppm) in air on intensity of the band with maximum at 248 nm. Thickness of the sample 0.5 m. 1, spectrum initial sample; 2, annealing at 400°C; 3, annealing at 500°C; 4, annealing at 800°C; 5, annealing at 1000°C.
O
D
o
500
Wovelength(nm)
z~z~z~zxzxSIH(2254 c m - t ) 00000 SiH(2264 c m - 1 )
o
250
. . . . . Absorption Sill (2254 cm
~/
00000 Absorption SiH (2264 cm --
Fig. 8. Effect of temperature on the rate of formation (norrealized) of the intrinsic defects (163 and 248 n m absorption bands) and Si-H. S i - O H centers in vitreous silicaannealed in H 2 atmosphere.
change during annealing in H 2 to 600°C (Fig. 8). This invariance confirms the proposal that in the mentioned temperature range the formation of S i - H centers occurs only during reaction of centers at the 248 nm band with H2. The oxygen-deficit centers (248 nm band) interact with U 2 forming only S i - H centers, which have a band with a maximum at 2254 cm - ] [11]. This result is important, because it is consistent with the attribution of the oxygen deficit centers of the 248 nm band to oxygen vacancies - S i - S i - , which interact with H z according to the reaction -Si-Si- + Hz-Si=HH-Si-. In the case of the alternative model of two-coordinated silicon atoms [2], complex kinetics for the formation of centers containing hydrogen such ..H as =Si type centers are observed. "H
24
V.Kh. Khalilov et al. /Journal of Non-Crystalline Solids 169 (1994) 15-28 ~00
6. The character of the absorption band at 163 nm
,5
80.
The source of the absorption band at 163 nm is still being discussed [3,7,8]. In Ref. [8], a correlation with the peroxy radical is shown, while in Ref. [3] and Ref. [7] it is associated with an oxygen vacancy. In Ref. [5], this band which occurs after rapid cooling of Type IV samples, previously treated in H2, is attributed to a -Si~center (three-coordinated with oxygen, a nonparamagnetic Si atom). The 163 nm band and the 248 nm band are always present in samples with low concentrations of O H groups. The temperature dependence of the intensity of the 163 nm band differs from that of the 248 nm (Figs. 8 and 10). First, it may be noted that centers which form the 163 nm band are stable during annealing in vacuum and in air (wet or dry) in the temperature range 250-1000°C. Annealing in H 2 to 650°C changes the intensity of the 163 nm band. In the range 600-650°C, the intensity at this band decreases and by 700°C it is almost absent in the absorption spectrum (Figs. 8 and 10). The t e m p e r a t u r e dependence of the 163 nm band differs from the interaction of oxygen vacancies with H 2 (Section 5) and peroxy linkages (Section 4). This result excludes the possibility that these centers are related to the 163 nm. In the same temperature range (600°-650°C), growth of intensity of the 2254 c m - 1 band stops. In the temperature range 650-1000°C, the maximum of the S i - H band is shifted to 2264 cm -1 and its intensity increases to a maximum at 650°C (Figs. 4, 7 and 8). The shift of the maximum of the S i - H band to 2264 c m - 1 shows an interaction of hydrogen and a new type of center. The temperature dependence of the S i - H center at 2264 c m - 1 band and the 163 nm band are correlated (Fig. 8). It follows that the appearance of the 2264 cm -1 band is connected with interaction of hydrogen with centers which form the absorption band at 163 nm. In Fig. 4 we show absorption spectra relative to the initial spectrum. For this reason the increase of the intensity of S i - H band in the region 650°-1000°C is observed on the background of the increase and change of maximum position of
Ln
c--
/
60
i!'2
D
/
E O'3 C-CD I
/ / g
20
.}' .~
•
~o
150
150
170
180
igo
Wovelength(nm) Fig. 10. The influence of the annealing of the glass (Coil 0.5 ppm) in H z on intensity of the band with maximum at 163 nm. Thickness of the sample 0.875 mm. 1, spectrum initial sample: 2, annealing at 500°C; 3, annealing at 600°C; 4, annealing at 700°C; 5, annealing at 1000°C.
the intrinsic band at 2258 cm -1 (Fig. 1, Section 3). From comparison of Fig. l(curve 2a) and Fig. 4, it follows that the contribution of this band is not large compared with the contribution of the S i - H band. Nevertheless we have made an experiment which accounts for the 'background' and resolves the S i - H band in the temperature range 700 °1000°C. The results are shown in Fig. 11. Curve (a) in this figure corresponds to the spectrum for 'dry' glass. The samples were annealed at 1000°C in a dry air atmosphere before they were put in the comparison and measuring channels of spectrophotometer (see Section 2). It is seen that bands discussed in Section 3 are compensated on the curve. (The weak 3680 c m - 1 band shows the small difference of the initial concentration of O H groups in the reference sample and the sampie.) Then the sample from m e a s u r e m e n t channel was annealed in H a at 800°C. The S i - H center which appears during annealing in H 2 at temperatures > 700°C is shown by the band with
V.Kh. Khalilov et al. /Journal of Non-Crystalline Solids 169 (1994) 15-28
f.2 "7
25
ci iL
QO _. ~ , W,, L-
i -- ~,~ r,t,w
km m,, ~
tq m
,,2 ~
....
2
. . . . . . .
~t2-, o
Wavenurnber(cm- 1)
Wavenumber(crn-t) Fig. 11. The appearance and annealing S i - H center with maximum at 2264 cm - I in 'dry' and 'wet' glass without structural relaxation process (in channel of comparison of spectrophotometer; sample annealed at 1000°C, 9 h in air). (a) Coil ~ 8 ppm. (b) C o n ~ 100 ppm. (la), (lb): 1000°C, 9 h, air; (2a), (2b): 1000°C + 800°C, 16 h, HE; (3a) 1000°C + 1000°C, 6 h, air.
maximum at 2264 cm-1 (curve(2a), Fig. 11). After annealing the same sample in dry air at 1000°C for 6 h the band at 2264 cm-1 (Fig. 11 curve(3a)) was not detected. Simultaneously, the intensity at 163 nm band was restored to its initial value. For comparison, on the same figure spectra are shown
(curves (lb) and (2b)) for 'wet' samples, in which the band at 163 nm is absent. As mentioned above, in this case the spectrum of the S i - H center is not detected. All the results can be explained by the model, which we proposed in Ref. [6]. In this model, the
f
(3.) gO ¢-" ED
I -
" //~
/
~,
(J3 C-'E~ 4o
5 I
I---2O
2t]
tO
10 5 o gSO
aa0
35o
Wovelength(nm)
0
,so
~ 3~ Wovelength(nm)
Fig. 12. The influence of the annealing in H 2 atmosphere of the 'wet' - 'dry' vitreous silica. Thickness of the sample 50 cm. (a) Initial content of OH groups 100 ppm. 1, initial sample; 2, annealed in H 2 750°C, 16 h. (b) Initial content of OH groups 0.5 ppm. 1, initial sample: 2, annealed in H E 750°C, 16 h.
26
V.Kh. Khalilov et aL /Journal of Non-Crystalline Solids 169 (1994) 15-28
three-coordinated non-paramagnetic silicon has a band at 163 nm. The charge of this center is compensated by interstitial chlorine ions: -Si ÷ CI(T~ center). The affinity between chlorine and electron (3.8 eV) causes the charge stability of the center. Low mobility of chlorine causes thermal stability. The interaction center with hydrogen is: - S i + C I - + H ° ~ - S i l l + C1°.
~w
/ ,
_Q
/']
g ._~
/
C~ t,/0u
7. Spectroscopic realization of HCI and C! 2 molecules in the glass, preforms and fibers
/
/
/ / ~000
Molecules of C12 in gas, in liquid and in solid matrix have been observed by a band near 330 nm. The amplitude and half-width of this band depends on its environment [16,17]. The presence of the band at 330 nm in the preforms made by VAD method was mentioned in Ref. [18]. The authors also attribute it to the C12 molecule. However, the position of maximum and nearly equal half-width is insufficient evidence for attributing the 330 nm band to molecular C12. The 330 nm band with a change of concentration of O H groups in the glass and preforms from 0.5 to 500 ppm is observed in the 'wet' samples (Fig. 12). The results of the chemical analysis show that the total concentration of C1 in the 'wet' preforms decreases and is 300-500 ppm. If the 330 nm band is formed by CI 2 molecules, in these samples it is located usually at interstitial positions. Samples of preforms with a 330 nm band were annealed in H 2 and dry air for 16 h at 750°C. The temperature was chosen so that structural relaxation was small. The band at 330 nm was not detected in the transmission spectrum after H treatment (Fig. 12). A treatment in dry air did not change the intensity of the 330 nm band at temperatures < 1000°C. In support of the thesis that the absence of the band at 330 nm is connected with the formation of HCI by an interaction between H 2 and C12, the IR absorption spectrum of samples with a length of 100 mm was measured in the region of the bands of HCI molecules. This region is in the range 2700-2900 cm-1 depending on the matrix and interstitial site of HC1 [19].
/
/
/
/ Wavenumher(cm - t)
/
~800
~ave number(cm -I )
Fig. 13. The absorption spectra in the region of vibration band of molecule HCI. Thickness of the sample 90 ram. (a) Samples annealed in H 750°C, 16 h. - . . . . . , initial content O H groups 100 ppm. initial content O H groups 0.5 ppm. (b) Initial content OI~I groups 0.5 ppm. - . . . . . , initial. annealed in H 2 750°C, 16 h.
It was found that in those samples with a band at 330 nm annealing produced a band at 2800 cm -1 (Fig. 13). In the case of annealing in the same conditions of 'dry' preforms without a band at 330 nm (Fig. 12), there were no detectable changes in the region 2700-2900 cm-~ (Fig. 13). The results show that, in the 'wet' samples, chlorine usually is in an interstitial site in the form of C12 or CI and after interaction with hydrogen forms HC1. With a higher content of chlorine in the 'dry' samples, in our opinion a structurally connected condition, -SiCI, is present. In our opinion, in the 'dry' samples the concentration of oxygen deficit defects is large (see Section 4) and they react with the dissolved chlorine. On the other hand, in the samples with O H concentration nearly 100 ppm and larger, the spectra of intrinsic defects are not detectable. In the region of long-wavelength edge of fundamental absorption of this glass, no absorption bands are observed. In these samples, after the HCI was formed during annealing in H 2, the appearance of a weak band at 165 nm was observed (Fig. 14). The condition of appearance of this band and the position of its maximum indicate that it differs from the 163 nm band. The 165 nm band is a characteristic of the HCI molecule [15] and the
V..Kh. Khalilov et al. /Journal of Non-Crystalline Solids 169 (1994) 15-28
TX
,f,,
f
f
//
"--d~
/
E o')
,r
¢
/ 0
~0
t70 ~o Wavelength(nm)
19o
~oo
Fig. 14. The transmission spectra of the glass (Con 0.5 ppm) before ( ) and after ( - - - - - - ) annealing in H 2 at 1000°C, 6 h.
n, the fundamental absorption edge is shifted to longer wavelength. It is shown [1] that, with a decrease of the concentration of O H groups in the glass, the refractive index in the visible region of the spectrum increases. At the same time, the edge of fundamental absorption shifts to the short-wave region of the spectrum [1,13]. So, for the case of very 'dry' glass in regions of optical heterogeneity, a long wavelength shift of the absorption edge with decrease of O H concentration is not expected. We have measured n in preforms with accuracy of + 10 -5 as a function of concentration of O H groups from 0.5 to 300 ppm. It was found that in the region of 2 ppm the linear dependence has a slope, which shows that in case of low concentration of O H groups the variation of n is determined not only by variation of O H concentration. This conclusion is the same as in Ref. [1]. Our samples of glass were fused by one method, with large differences in concentrations of hydroxyl and chlorine. These differences allowed us to obtain more accurate results and to determine
conditions for its appearance in the spectra is additional evidence of the formation of the HCI molecule.
8. The character of the complex heterogeneity in type IV quartz glass It is well known that, in the type IV glass in the direction perpendicular to the surface of growth, a heterogeneity of the refractive index, An, is observed [1,20]. In Refs. [1,20], it was shown that this phenomenon cannot be explained by the distribution of chlorine and hydroxyl groups nor by variations of the fictive temperature. The data in Refs. [1,13] show that the magnitude of the variation of the refractive index showing the optical heterogeneity is An ~ 10 -4. In regions of sample with increased n, the concentration of chlorine reaches 1000 ppm, and the concentration of O H groups is < 1 ppm (the average concentrations of chlorine is ~ 100 ppm and of O H groups ~ 40 ppm). In Refs. [1,13] it is also shown that, in the region of glass with increased
27
40
O
E 3o. (J
"~ 2o 0 (.._)
~ .2
/! tt
o 10. m
~,
0 t
t60
55
['70
Wavelength(nm) Fig. 15. The long wavelength edge of the fundamental absorption after annealing in H 2 at 800°C for 16 h. 1, COH 5 ppm; 2, COIl 100 ppm.
28
V.Kh. Khalilov et al. /Journal of Non-Crystalline Solids 169 (1994) 15-28
how the changes of refractive index in 'dry' and 'wet' samples are affected by differences in concentration of defects. Samples with high (100 ppm) and low (5 ppm) concentration of O H groups were annealed in H2. As mentioned above (Section 3), the structural refraction process in these samples is the same, and the position and intensity of fundamental infra-red bands do not differ. Simultaneously in the region of the UV-edge of the fundamental absorption, after annealing in H atmosphere, the bands at 163 nm and 248 nm are suppressed. In these samples there is a possibility to compare the long-wavelength edge of the fundamental absorption, not distorted, by defect absorption bands. The results of comparison confirm the results which were reported before in Ref. [1] about the long-wavelength shift of the edge of fundamental absorption in the dry samples (Fig. 15). However the analysis of the results which were shown in Ref. [1] with the same thermal prehistory and the identical spectra of fundamental infra-red bands brings us to the conclusion that the increase of n in 'dry' glass is due to an increase in the density of states in the 'tail' in the forbidden gap, the properties of which remain unknown.
9. Conclusion The temperature dependencies, which reflect the interaction between hydrogen and intrinsic defects in glass, indicate the existence of a temperature region 600-650°C which delineates two different effects of annealing. At these temperatures we observed significant changes in defect transformations by changes in bands at 248 and 163 nm. Significant changes of structure in this temperature range occur with S i - H and S i - O H centers. These differences are due to the appearance of structural relaxation processes at 600650°C for which mechanisms are not determined.
These processes are not correlated with a change of fictive temperature. The formation of the hydrogen- (Si-H) and oxygen-deficit centers also produced a new S i - H center and showed that the band at 248 nm is caused by the oxygen vacancy - S i - S i - ; the band at 163 nm is caused by threecoordinated non-paramagnetic silicon: -Si ÷ (T~center).
10. References [1] V.H. Khalilov, G.A. Dorphman, L.G. Karpov, I.V. Pevnitsky and V.V. Zahov, Phys. Chem. Glass (USSR) 13 (1987) 721. [2] L.N. Skuja, A.W. Streletsky and A.B. Pakovich, Solid State Commun. 5 (1984) 1069. [3] R. Tohmon, J. Jamasaka, K. Nagasawa, J. Ohki and J. Hama, J. Non-Cryst. Solids 95&96 (1987) 671. [4] G.W. Arnold, JEEE Trans. Nucl. Sci. NS-20 (1973) 220. [5] K. Nagasava, Y. Hoshi and Y. Ohki, Jpn. Appl. Phys. 26 (1987) L554. [6] V.H. Khalilov and I.V. Pevnitsky, Phys. Chem. Glass (USSR) 11 (1985) 504. [7] R.P. Gupta, Phys. Rev. B33 (1986) 7274. [8] M. Stepelbrock, D.L. Griscom and E.J. Frieble, J. NonCryst. Solids 32 (1979) 313. [9] J.E. Shelby, J. Appl. Phys. 51 (1980) 2589. [10] A.R. Silin, Phys. Chem. Glass (USSR) 4 (1978) 266. [11] G. Van Der Steen and H. Van Der Boom, J. Non-Cryst. Solids 23 (1977) 279. [12] R.B. Langhlin and J.D. Joannopoulos, Phys. Rev. B16 (1983) 2942. [13] A.E. Geissberger and F.L. Galeener, Phys. Rev. B28 (1983) 3266. [14] V.K. Leko and E.V. Mesherakova, Phys. Chem. Glass (USSR) 1 (1975) 447. [15] J.C. Mikkelsen and F.L. Galeener, J. Non-Cryst. Solids 37 (1980) 71. [16] H. Okabe, Photochemistry Small Molecule (World, Moskow, 1981). [17] D. Kalvert and D. Pitts, Photochemistry(World, Moskow, 1968). [18] A.D. Abramov, E.V. Anoikin, E.M. Dianov, et al., Highclean Substance (USSR) (1987) 12. [19] A.J. Barnes, S. Susuki, H.E. Hallam and G.F. Sorimshaw, Trans. Faraday Soc. 65 (1969) 3159. [20] I.V. Pevnitsky and V.H. Khalilov, Phys. Chem. Glass (USSR) 16 (1990) 668.