- Email: [email protected]
Role of impurities in the 5.16 eV optical absorption band of Ge-doped silica
1 0 f J R N A L OF
ELSEVIER
Journal of Non-Crystalline Solids 216 (1997) 26-29
Role of impurities in the 5.16 eV optical absorption band of Ge-doped silica M. Martini, F. Meinardi, A. Paleari *, L. Portinari, G. Spinolo Dipartimento di Fisica, lstituto Nazionale Fisica della Materia, Universith di Milano, via Celoria 16, 1-20133 Milan, Italy
Abstract The optical absorption band at 5.16 eV and the related photoluminescence were investigated in Ge-doped silica samples as a function of the Ge content to study the relation between impurities and oxygen deficient defects. The data, collected in a wide doping range, showed that the dependence of the intensity of the 5.16 eV absorption band on the Ge content depends on the thermochemical parameters of the preparation process. Further, two distinct ranges of Ge-doping were individuated showing different dependence of the density of optically active center on the impurity concentration. © 1997 Elsevier Science B.V. PACS: 78.50.Ec; 61.70.Vn
1. Introduction Garino Canina [1] first observed that an absorption peak at ~ 5 eV (the so called B 2 band) was created in silica grown in a reducing environment, but he observed also that the presence of Ge gave as a result the same absorption [2]. The first who noticed that two different peaks could be present at the same (5 eV) energy were Cohen [3] and Arnold [4]. They showed that one band was likely to be related to the presence of impurities, while the other was produced by irradiation or ion implantation. A confirmation and clarification of such a picture was given by Tohmon et al. [5], who separated two types of B 2 bands: B2~ and B2~, peaked at 5.0 and 5.16 eV, respectively. The two absorption bands were
* Corresponding author. Tel.: + 3 9 - 2 239 2352; fax: +39-2 239 2414; e-mail: [email protected].
related to two different photoluminescence spectra: emissions at 2.7 and 4.4 eV are present in samples with the B2~ absorption band, emissions at 3.1 and 4.2 eV are present in samples with the B2~ absorption band [6]. In reality, the overall picture is even more complex, as photoluminescence excitation studies have shown [7]. Moreover, other excitation peaks have been found [8-10]. The B2~ absorption band and its related photoluminescence bands were proposed to be due to oxygen deficient centers (ODC), either produced during silica growth in reducing environment [5], or by neutron irradiation [6]. There is still a lack of agreement on the model associated to the absorption center: Skuja et al. [11] proposed a twofold coordinated silicon, while Thomon et al. [5] proposed the neutral oxygen vacancy. This latter model is supported by the observation by Arai et al. [12] of the bleaching of the B2~ band by UV light, resulting in the generation of Si E' centers. Less clear is the
0022-3093/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved. PH S 0 0 2 2 - 3 0 9 3 ( 9 7 ) 0 0 2 3 7 - 8
M. Martini et al. /Journal of Non-Crystalline Solids 216 (1997) 26-29
picture of the B2I3 center. There is agreement [2,8,13,14] on an ODC as well and an impurity related origin is also widely agreed upon. A correlation with Ge content was put forth by Garino Canina [2] and a Ge-related center (a twofold coordinated Ge) was proposed by Skuja et al. [8] and by Hosono et al. [13]. However, Tsai et al. [14] reported data which rule out the Ge 2+ model and apparently suggest a square root dependence of the 5.16 eV absorption intensity on the Ge content. From this dependence they inferred a divacancy type model of ODC in a fourfold coordinated Ge site, responsible for the 5.16 eV band. The knowledge of the dependence of the 5.16 eV absorption upon the Ge content should give some indications about the role of germanium in the optically active ODCs in silica and would also have technological importance for the optical applications of Ge-doped silica as photorefractive material [15]. Nevertheless, the experimental definition of the role of the Ge concentration on the ODC density may be difficult or hopeless without a proper control of the material parameters related to the process conditions. In particular, the preparation temperature may critically affect the number of ODCs [16] at fixed Gecontent. This fact was often neglected despite its relevance in comparing data of differently doped samples. In fact, commercial silica preforms with different Ge-doping are usually prepared at different temperatures since the melting point increases with decreasing GeO 2 content. Indeed, by comparing the optical properties induced by Ge-doping and those observed in pure silica as a result of ~/ and neutron irradiations or reducing chemical treatments, several similarities may suggest that the role of Ge is essentially that of inducing ODCs in the lattice without a necessary involvement of the impurities in the defect sites. In order to analyze the relation between Ge-doping and ODC density in a more extended doping range than previously investigated, we carried out a detailed analysis of the 5 eV absorption in homogeneous sets of Ge-doped samples prepared in controlled manufacturing conditions. Indeed, our results clarify the ODC versus Ge relation showing a linear dependence of the concentration of optical active defects on the Ge content at low doping level and an effect of preparation parameters.
27
2. Experimental procedure The investigated samples were cut from preforms of Ge-doped SiO 2 optical fibers prepared by the modified chemical vapour deposition (MCVD) method with the Ge content ranging from 300 ppm to 27% (supplied by FOS, Fibre Ottiche Sud, Battipaglia, Italy). Two sets of preforms (prepared at different times and covering the two doping ranges 0.03 to 4 at.% and 4 to 27 at.% Ge) were processed in nominally identical conditions of temperature, annealing time and atmosphere (except Ge concentration) during the isothermal high temperature (1700°C) deposition step, as well as during the densification process carried out at the temperature of 1100°C. We remark that this fact is not usually encountered within a set of commercial samples in such a wide range of doping, because different temperatures and oxygen partial pressures are often employed during the second step of preparation as a function of the Ge-doping. A third set of heavy doped (4 to 27 at.% Ge) commercial samples was measured for comparison. The concentration of germanium was determined by refraction index measurements and by scanning-electron-microscopy (SEM) analysis. Optical absorption spectra were measured at room temperature in the visible and near UV spectral region with a double beam spectrophotometer (Varian Cary 2300). The absorption coefficient was calculated by fitting the spectra by two Gaussian components and an exponential background, taking into account the increase of absorption contributions in the high energy side of the spectrum. Corrections for the reflectivity as a function of the Ge-content were considered, but they resulted negligible throughout the doping range. The photoluminescence spectra were detected by exciting with a Hinteregger hydrogen discharge lamp (McPherson) and a grating monochromator (McPherson) operating in hydrogen atmosphere. The photoluminescence in the 2.5-5.0 eV spectral range was analyzed by a grating monochromator (Jarrel-Ash) and an photomultiplier EMI 9924QA (Bialkali photocathode).
3. Results The 5.16 eV absorption band constitutes the main feature of the collected spectra in all the analyzed
28
M. Martini et al. / Journal of Non-Crystalline Solids 216 (1997) 26-29
60 c ~10
A
7
50
~
•
-~5
+, 4-0
/~3.0
c
3.5
4.0
4.5
.2 "3 3 0 o rO
:.c 2 0 Q_
x3
<
.......
lO
0 4.0
4.5
5.0
5.5
6.0
E n e r g y (eV)
Fig. 1. Representative room temperature optical absorption spectrum of a Ge-doped silica sample (2 at.% Ge). In the inset, the photoluminescence spectrum excited at 5 eV is reported.
samples (see Fig. 1). Its intensity depends on the Ge content, while peak position and bandwidth (FWHM = 0.35 eV) are unaffected. The 5.16 eV absorption
E o
~100
O0
•
(.D
g
~
V
10
o_
in our Ge-doped samples consists of a single Gaussian component whose features match those of the BED band already observed by other authors. This assignment is confirmed by the observation of the photoluminescence emissions at 4.2 and 3.1 eV (shown in the inset of Fig. 1), already related to the B2D band [7], previously attributed to ODC [2,8,13,14]. The dependence of the absorption coefficient of the 5.16 eV band on the Ge content in the investigated Ge-doped silica samples is reported in Fig. 2. The data show the following facts: (i) at small Ge content the absorption coefficient at 5.16 eV linearly increases with the Ge doping, (ii) on the contrary, in the large doped samples the Ge concentration does not show a clear effect on the absorption coefficient and (iii) the absorption coefficient changes from a set of samples to another, at fixed Ge content, with changes in the processing parameters.
4. Discussion
The first consideration (point (i)) confirms that the formation of optically active sites during the manufacturing process is related to the Ge content, at least at small Ge-doping. The nearly linear behaviour suggests that each Ge impurity may induce one ODC. These data are inconsistent with the proposal that the defects responsible for the 5.16 eV absorption are created in the lattice as divacancy sites [14]. A square root law would be expected in that case if the Ge-content is the only variable parameter. In fact, it was already suggested [14] that [ODC] and [Ge] concentrations should satisfy a thermochemical equilibrium relation of the following type: [ODC]" K = [Ge------]--[Oa] n/2,
,
0.01
,
,
.....
1
0.1
. . . . . . . .
I
. . . . . . . .
1 Ge C o n t e n t
I
10
. . . . . . .
100
(%)
Fig. 2. Filled symbols: 5.16 eV absorption coefficient versus Ge content in Ge-doped silica samples produced in nominally identical manufacturing conditions (different symbols correspond to sample sets supplied at different time). Open symbols: set of samples from standard commercial production lines, processed in different temperature and atmosphere conditions. Full line: unitary slope corresponding to a linear dependence.
(1)
where K is the thermochemical equilibrium constant (dependent on the temperature and on the specific lattice) and n is the number of ODCs arising from the presence of each Ge-substituted site in the specific thermochemical conditions. In this approach, the [ODC] + [Ge] 1/" relation (with n >_ 1) may indicate that some kind of equilibrium takes place involving the creation of at least one oxygen vacancy per substitutional Ge atom. The different ionic radii
ll4. Martini et al. / Journal of Non-Crystalline Solids 216 (1997) 26-29
and electronegativities of Ge (0.053 nm and 2.0, respectively) and Si (0.041 nm and 1.7, respectively) make indeed likely that the local perturbation constituted by isolated and diluted Ge impurities may induce oxygen coordination defects in the silica structure. In fact, the systematic observation of oxygen vacancy defects in Ge-doped silica is well established in the literature [14]. So, we may suppose that the Ge-substituted cationic sites are accompanied by coordination defects in the nearby environment which may favour the formation of neutral oxygen vacancies in the structure during the high temperature treatment. The linear dependence at small doping in Fig. 2, consistent with a single vacancy formation process, may suggest the attribution of optically active ODC to single vacancy defects. However it should be considered that single vacancies could be produced which subsequently diffuse and form divacancies. Concerning the larger doping range (point (ii)), the number of optically active ODC is unaffected by the Ge concentration. In this case, the ODC concentration is probably limited by the domain of non-stoichiometry of the structure. In fact, at doping levels > 2 at.% we cannot consider Ge as an impurity and the doped materials are rather to be treated as mixed oxides. So, a further increase of the number of Ge atoms does not imply an increase of ODC precursor sites. Other parameters, as the processing temperature and the oxygen partial pressure, become more relevant in determining the ODC density (as remarked in point (iii)).
5. Conclusions In summary, our data show that only at low doping levels the Ge density drives the ODC concentration, while at higher doping levels the manufacturing parameters are the relevant factors in the ODC formation. So, GeO 2 clustering, non-stoichiometry, porosity, densification and other material features should affect the dependence of the density of ODCs
29
on the impurity concentration at large Ge-doping. As a consequence, a single relation between Ge-content and 5.16 eV absorption band cannot be given throughout these doping ranges, more than one ODC formation mechanism being involved. These results also mean that, in different kinds of SiO 2, similar but quantitatively different relations between optical centers and impurity content are expected.
Acknowledgements We are pleased to thank Dr Anelli, Pirelli Cavi, Italy and Ing. Galasso, Fibre Ottiche Sud (FOS), for kindly supplying the Ge-doped samples and for their valuable assistance during the manufacturing process and the characterization of the samples.
References [1] V. Garino Canina, C. R. Acad. Sci. 240 (1955) 1765. [2] V. Garino Canina, C. R. Acad. Sci. 242 (1956) 1765. [3] A.J. Cohen, Phys. Rev. 105 (1957) 1151. [4] G.W. Arnold, IEEE Trans. Nucl. Sci. NS-20 (1973) 220. [5] R. Tohmon, H. Mizuno, Y. Ohki, K. Sasagane, K. Nagasawa, Y. Hama, Phys. Rev. B39 (1989) 1337. [61 M. Guzzi, M. Martini, M. Mattaini, F. Pio, G. Spinolo, Phys. Rev. B35 (1987) 9407. [7] A. Corazza, B. Crivelli, M. Martini, G. Spinolo, J. Phys. 7 (1995) 6739. [8] L.N. Skuja, J. Non-Cryst. Solids 149 (1992) 77. [9] M. Bertino, A. Corazza. M. Martini, A. Mervic, G. Spinolo, J. Phys. 6 (1994) 6345. [10] B. Crivelli, M. Martini, F. Meinardi, A. Paleari, G. Spinolo, Solid State Commun. 100 (1996) 651. [11] L.N. Skuja, A.N. Streletsky, A.B. Pakovich, Solid State Commun. 50 (1984) 1069. [12] K. Arai, H. Irnai, H. Hosono, Y. Abe, H. Imagawa, Appl. Phys. Leu. 53 (1988) 1891. [13] H. Hosono, Y. Abe, D.L. Kinser, R.A. Weeks, K. Muta, H. Kawazoe, Phys. Rev. B46 (1992) 11445. [14] T.E. Tsai, E.J. Friebele, M. Rajaram, S. Mukhapadhyay, Appl. Phys. Lett. 64 (1994) 1481. [15] H. Hosono, K. Kawamura, N. Ueda, H. Kawazoe, S. Fujitsu. N. Matsunami, J. Phys. 7 (1995) L343. [16] J.M. Jackson, M.E. Wells, G. Kordas, D.L. Kinser, R.A. Weeks, J. Appl. Phys. 58 (1985) 2308.
ELSEVIER
Journal of Non-Crystalline Solids 216 (1997) 26-29
Role of impurities in the 5.16 eV optical absorption band of Ge-doped silica M. Martini, F. Meinardi, A. Paleari *, L. Portinari, G. Spinolo Dipartimento di Fisica, lstituto Nazionale Fisica della Materia, Universith di Milano, via Celoria 16, 1-20133 Milan, Italy
Abstract The optical absorption band at 5.16 eV and the related photoluminescence were investigated in Ge-doped silica samples as a function of the Ge content to study the relation between impurities and oxygen deficient defects. The data, collected in a wide doping range, showed that the dependence of the intensity of the 5.16 eV absorption band on the Ge content depends on the thermochemical parameters of the preparation process. Further, two distinct ranges of Ge-doping were individuated showing different dependence of the density of optically active center on the impurity concentration. © 1997 Elsevier Science B.V. PACS: 78.50.Ec; 61.70.Vn
1. Introduction Garino Canina [1] first observed that an absorption peak at ~ 5 eV (the so called B 2 band) was created in silica grown in a reducing environment, but he observed also that the presence of Ge gave as a result the same absorption [2]. The first who noticed that two different peaks could be present at the same (5 eV) energy were Cohen [3] and Arnold [4]. They showed that one band was likely to be related to the presence of impurities, while the other was produced by irradiation or ion implantation. A confirmation and clarification of such a picture was given by Tohmon et al. [5], who separated two types of B 2 bands: B2~ and B2~, peaked at 5.0 and 5.16 eV, respectively. The two absorption bands were
* Corresponding author. Tel.: + 3 9 - 2 239 2352; fax: +39-2 239 2414; e-mail: [email protected].
related to two different photoluminescence spectra: emissions at 2.7 and 4.4 eV are present in samples with the B2~ absorption band, emissions at 3.1 and 4.2 eV are present in samples with the B2~ absorption band [6]. In reality, the overall picture is even more complex, as photoluminescence excitation studies have shown [7]. Moreover, other excitation peaks have been found [8-10]. The B2~ absorption band and its related photoluminescence bands were proposed to be due to oxygen deficient centers (ODC), either produced during silica growth in reducing environment [5], or by neutron irradiation [6]. There is still a lack of agreement on the model associated to the absorption center: Skuja et al. [11] proposed a twofold coordinated silicon, while Thomon et al. [5] proposed the neutral oxygen vacancy. This latter model is supported by the observation by Arai et al. [12] of the bleaching of the B2~ band by UV light, resulting in the generation of Si E' centers. Less clear is the
0022-3093/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved. PH S 0 0 2 2 - 3 0 9 3 ( 9 7 ) 0 0 2 3 7 - 8
M. Martini et al. /Journal of Non-Crystalline Solids 216 (1997) 26-29
picture of the B2I3 center. There is agreement [2,8,13,14] on an ODC as well and an impurity related origin is also widely agreed upon. A correlation with Ge content was put forth by Garino Canina [2] and a Ge-related center (a twofold coordinated Ge) was proposed by Skuja et al. [8] and by Hosono et al. [13]. However, Tsai et al. [14] reported data which rule out the Ge 2+ model and apparently suggest a square root dependence of the 5.16 eV absorption intensity on the Ge content. From this dependence they inferred a divacancy type model of ODC in a fourfold coordinated Ge site, responsible for the 5.16 eV band. The knowledge of the dependence of the 5.16 eV absorption upon the Ge content should give some indications about the role of germanium in the optically active ODCs in silica and would also have technological importance for the optical applications of Ge-doped silica as photorefractive material [15]. Nevertheless, the experimental definition of the role of the Ge concentration on the ODC density may be difficult or hopeless without a proper control of the material parameters related to the process conditions. In particular, the preparation temperature may critically affect the number of ODCs [16] at fixed Gecontent. This fact was often neglected despite its relevance in comparing data of differently doped samples. In fact, commercial silica preforms with different Ge-doping are usually prepared at different temperatures since the melting point increases with decreasing GeO 2 content. Indeed, by comparing the optical properties induced by Ge-doping and those observed in pure silica as a result of ~/ and neutron irradiations or reducing chemical treatments, several similarities may suggest that the role of Ge is essentially that of inducing ODCs in the lattice without a necessary involvement of the impurities in the defect sites. In order to analyze the relation between Ge-doping and ODC density in a more extended doping range than previously investigated, we carried out a detailed analysis of the 5 eV absorption in homogeneous sets of Ge-doped samples prepared in controlled manufacturing conditions. Indeed, our results clarify the ODC versus Ge relation showing a linear dependence of the concentration of optical active defects on the Ge content at low doping level and an effect of preparation parameters.
27
2. Experimental procedure The investigated samples were cut from preforms of Ge-doped SiO 2 optical fibers prepared by the modified chemical vapour deposition (MCVD) method with the Ge content ranging from 300 ppm to 27% (supplied by FOS, Fibre Ottiche Sud, Battipaglia, Italy). Two sets of preforms (prepared at different times and covering the two doping ranges 0.03 to 4 at.% and 4 to 27 at.% Ge) were processed in nominally identical conditions of temperature, annealing time and atmosphere (except Ge concentration) during the isothermal high temperature (1700°C) deposition step, as well as during the densification process carried out at the temperature of 1100°C. We remark that this fact is not usually encountered within a set of commercial samples in such a wide range of doping, because different temperatures and oxygen partial pressures are often employed during the second step of preparation as a function of the Ge-doping. A third set of heavy doped (4 to 27 at.% Ge) commercial samples was measured for comparison. The concentration of germanium was determined by refraction index measurements and by scanning-electron-microscopy (SEM) analysis. Optical absorption spectra were measured at room temperature in the visible and near UV spectral region with a double beam spectrophotometer (Varian Cary 2300). The absorption coefficient was calculated by fitting the spectra by two Gaussian components and an exponential background, taking into account the increase of absorption contributions in the high energy side of the spectrum. Corrections for the reflectivity as a function of the Ge-content were considered, but they resulted negligible throughout the doping range. The photoluminescence spectra were detected by exciting with a Hinteregger hydrogen discharge lamp (McPherson) and a grating monochromator (McPherson) operating in hydrogen atmosphere. The photoluminescence in the 2.5-5.0 eV spectral range was analyzed by a grating monochromator (Jarrel-Ash) and an photomultiplier EMI 9924QA (Bialkali photocathode).
3. Results The 5.16 eV absorption band constitutes the main feature of the collected spectra in all the analyzed
28
M. Martini et al. / Journal of Non-Crystalline Solids 216 (1997) 26-29
60 c ~10
A
7
50
~
•
-~5
+, 4-0
/~3.0
c
3.5
4.0
4.5
.2 "3 3 0 o rO
:.c 2 0 Q_
x3
<
.......
lO
0 4.0
4.5
5.0
5.5
6.0
E n e r g y (eV)
Fig. 1. Representative room temperature optical absorption spectrum of a Ge-doped silica sample (2 at.% Ge). In the inset, the photoluminescence spectrum excited at 5 eV is reported.
samples (see Fig. 1). Its intensity depends on the Ge content, while peak position and bandwidth (FWHM = 0.35 eV) are unaffected. The 5.16 eV absorption
E o
~100
O0
•
(.D
g
~
V
10
o_
in our Ge-doped samples consists of a single Gaussian component whose features match those of the BED band already observed by other authors. This assignment is confirmed by the observation of the photoluminescence emissions at 4.2 and 3.1 eV (shown in the inset of Fig. 1), already related to the B2D band [7], previously attributed to ODC [2,8,13,14]. The dependence of the absorption coefficient of the 5.16 eV band on the Ge content in the investigated Ge-doped silica samples is reported in Fig. 2. The data show the following facts: (i) at small Ge content the absorption coefficient at 5.16 eV linearly increases with the Ge doping, (ii) on the contrary, in the large doped samples the Ge concentration does not show a clear effect on the absorption coefficient and (iii) the absorption coefficient changes from a set of samples to another, at fixed Ge content, with changes in the processing parameters.
4. Discussion
The first consideration (point (i)) confirms that the formation of optically active sites during the manufacturing process is related to the Ge content, at least at small Ge-doping. The nearly linear behaviour suggests that each Ge impurity may induce one ODC. These data are inconsistent with the proposal that the defects responsible for the 5.16 eV absorption are created in the lattice as divacancy sites [14]. A square root law would be expected in that case if the Ge-content is the only variable parameter. In fact, it was already suggested [14] that [ODC] and [Ge] concentrations should satisfy a thermochemical equilibrium relation of the following type: [ODC]" K = [Ge------]--[Oa] n/2,
,
0.01
,
,
.....
1
0.1
. . . . . . . .
I
. . . . . . . .
1 Ge C o n t e n t
I
10
. . . . . . .
100
(%)
Fig. 2. Filled symbols: 5.16 eV absorption coefficient versus Ge content in Ge-doped silica samples produced in nominally identical manufacturing conditions (different symbols correspond to sample sets supplied at different time). Open symbols: set of samples from standard commercial production lines, processed in different temperature and atmosphere conditions. Full line: unitary slope corresponding to a linear dependence.
(1)
where K is the thermochemical equilibrium constant (dependent on the temperature and on the specific lattice) and n is the number of ODCs arising from the presence of each Ge-substituted site in the specific thermochemical conditions. In this approach, the [ODC] + [Ge] 1/" relation (with n >_ 1) may indicate that some kind of equilibrium takes place involving the creation of at least one oxygen vacancy per substitutional Ge atom. The different ionic radii
ll4. Martini et al. / Journal of Non-Crystalline Solids 216 (1997) 26-29
and electronegativities of Ge (0.053 nm and 2.0, respectively) and Si (0.041 nm and 1.7, respectively) make indeed likely that the local perturbation constituted by isolated and diluted Ge impurities may induce oxygen coordination defects in the silica structure. In fact, the systematic observation of oxygen vacancy defects in Ge-doped silica is well established in the literature [14]. So, we may suppose that the Ge-substituted cationic sites are accompanied by coordination defects in the nearby environment which may favour the formation of neutral oxygen vacancies in the structure during the high temperature treatment. The linear dependence at small doping in Fig. 2, consistent with a single vacancy formation process, may suggest the attribution of optically active ODC to single vacancy defects. However it should be considered that single vacancies could be produced which subsequently diffuse and form divacancies. Concerning the larger doping range (point (ii)), the number of optically active ODC is unaffected by the Ge concentration. In this case, the ODC concentration is probably limited by the domain of non-stoichiometry of the structure. In fact, at doping levels > 2 at.% we cannot consider Ge as an impurity and the doped materials are rather to be treated as mixed oxides. So, a further increase of the number of Ge atoms does not imply an increase of ODC precursor sites. Other parameters, as the processing temperature and the oxygen partial pressure, become more relevant in determining the ODC density (as remarked in point (iii)).
5. Conclusions In summary, our data show that only at low doping levels the Ge density drives the ODC concentration, while at higher doping levels the manufacturing parameters are the relevant factors in the ODC formation. So, GeO 2 clustering, non-stoichiometry, porosity, densification and other material features should affect the dependence of the density of ODCs
29
on the impurity concentration at large Ge-doping. As a consequence, a single relation between Ge-content and 5.16 eV absorption band cannot be given throughout these doping ranges, more than one ODC formation mechanism being involved. These results also mean that, in different kinds of SiO 2, similar but quantitatively different relations between optical centers and impurity content are expected.
Acknowledgements We are pleased to thank Dr Anelli, Pirelli Cavi, Italy and Ing. Galasso, Fibre Ottiche Sud (FOS), for kindly supplying the Ge-doped samples and for their valuable assistance during the manufacturing process and the characterization of the samples.
References [1] V. Garino Canina, C. R. Acad. Sci. 240 (1955) 1765. [2] V. Garino Canina, C. R. Acad. Sci. 242 (1956) 1765. [3] A.J. Cohen, Phys. Rev. 105 (1957) 1151. [4] G.W. Arnold, IEEE Trans. Nucl. Sci. NS-20 (1973) 220. [5] R. Tohmon, H. Mizuno, Y. Ohki, K. Sasagane, K. Nagasawa, Y. Hama, Phys. Rev. B39 (1989) 1337. [61 M. Guzzi, M. Martini, M. Mattaini, F. Pio, G. Spinolo, Phys. Rev. B35 (1987) 9407. [7] A. Corazza, B. Crivelli, M. Martini, G. Spinolo, J. Phys. 7 (1995) 6739. [8] L.N. Skuja, J. Non-Cryst. Solids 149 (1992) 77. [9] M. Bertino, A. Corazza. M. Martini, A. Mervic, G. Spinolo, J. Phys. 6 (1994) 6345. [10] B. Crivelli, M. Martini, F. Meinardi, A. Paleari, G. Spinolo, Solid State Commun. 100 (1996) 651. [11] L.N. Skuja, A.N. Streletsky, A.B. Pakovich, Solid State Commun. 50 (1984) 1069. [12] K. Arai, H. Irnai, H. Hosono, Y. Abe, H. Imagawa, Appl. Phys. Leu. 53 (1988) 1891. [13] H. Hosono, Y. Abe, D.L. Kinser, R.A. Weeks, K. Muta, H. Kawazoe, Phys. Rev. B46 (1992) 11445. [14] T.E. Tsai, E.J. Friebele, M. Rajaram, S. Mukhapadhyay, Appl. Phys. Lett. 64 (1994) 1481. [15] H. Hosono, K. Kawamura, N. Ueda, H. Kawazoe, S. Fujitsu. N. Matsunami, J. Phys. 7 (1995) L343. [16] J.M. Jackson, M.E. Wells, G. Kordas, D.L. Kinser, R.A. Weeks, J. Appl. Phys. 58 (1985) 2308.