The photoluminescence spectra of the divalent Ge (Ge2+) center in GeO2-SiO2 glasses with different photosensitivities were investigated by means of excitation-emission energy mapping. The ultraviolet light induced photorefractivity has been correlated with the local structure around the Ge2+ centers. The glasses with a larger photorefractivity tended to exhibit a greater band broadening of the singlet-singlet transition on the higher excitation energy side accompanied by an increase in the Stokes shifts. This strongly suggests the existence of highly photosensitive Ge2+ centers with higher excitation energies. It is also found that the introduction of a hydroxyl group or boron species in GeO2-SiO2 glasses under appropriate conditions modifies the local environment of Ge2+ leading to an enhanced photorefractivity.
©2003 Optical Society of America
The ultraviolet (UV)-induced photochemical reactions in GeO2-SiO2 glasses have been widely utilized for functionalizing the optical fibers as fiber Bragg gratings , nonlinear optical devices, etc., which play an important role in DWDM telecommunication technology. The photo-induced refractive index change in the GeO2-SiO2 glasses is reported to be on the order of 10-5~10-4 , which is less than the practical requirement for the direct laser writing of a channel waveguide structure and a strong Bragg grating. Therefore, sensitization is indispensable for the laser modification of the glasses. High pressure H2 treatment, so-called H2 loading, is now practically used as a post-treatment for the photosensitization of the GeO2-SiO2 glasses . Recently, OH flooding (massive hydroxyl formation) has been found to be an effective sensitization technique . The addition of a sensitizing component such as B and Sn is also known to be effective for increasing the photorefractivity. In fact, GeO2-SiO2 fibers doped with B have a high photosensitivity exceeding 10-3. On the other hand, it was reported that GeO2-SiO2 glass films deposited by the CVD method have a large photosensitivity [6, 7].
Although it has been more than 20 years since the discovery of the photo-induced grating formation in a fiber core made from GeO2-SiO2 glasses by Hill et al. in 1978 , the reaction mechanisms have not yet been fully understood in spite of significant advances in the practical telecommunication field of fiber Bragg gratings. It has been considered that the photo-induced refractive index change is due to macroscopic densification and/or an absorption coefficient change associated with the structural changes from the intrinsic germanium oxygen deficient center (GeODC) to other defects such as the germanium electron trap center (GEC) or GeE’ center . The structure model of GeODC, the Ge-Ge homobond and divalent Ge center (Ge2+ or GLPC) observed at 5.06 eV and 5.16 eV in optical absorption, respectively, have already been proposed .
It has been reported from the latest studies on the photochemical reaction mechanism, the selective excitation of Ge2+ centers using a XeF excimer laser light  and MO calculations [12,13] that the Ge2+ center is likely to be the most significant GeODC responding to the dense UV photons. In addition, the 5eV absorption due to the Ge2+ center involves a large inhomogeneous line broadening .
In the present study, the photoactivity of the Ge2+ center in the GeO2 -SiO2 glasses was investigated by means of photoluminescence (PL) spectrum mapping in order to correlate the photoactivated chemical processes with the photosensitivity. It is well-known that the Ge2+ center in Ge-doped silica glass shows PL of around ~280 and ~400 nm under ~248nm (5 eV) excitation, referred to as the α and β bands, respectively . Differing from the absorption spectra, the PL spectrum mapping allows us to extract information on the structural distribution around the Ge2+ center because the PL spectrum is correlated only with the Ge2+ center [15,16]. Various preparation methods and post-treatments were applied to the Ge-doped silica glasses in order to obtain glasses with different photosensitivities.
2. Experimental procedure
The germanosilicate glass samples used were thin films prepared by the plasma-enhanced chemical vapor deposition (PECVD) method (PD-10C, Samco International, Japan) and commercial preforms of optical fibers prepared by the vapor phase axial deposition (VAD) method (Shin-Etsu Chemical Co. Ltd., Japan). GeO2-SiO2 PECVD glass films were deposited on a silica glass substrate with raw materials of Si(OC2H5)4, Ge(OCH3)4, and B(OC2H5)3. The thickness of the films was determined to be 5 µm. The fiber preform, of which the Ge content of the core was 10 mol%, was cut into 1-mm thick disks and polished to an optical finish. Some of the fiber preform disks were H2-loaded at 10.5 MPa for 10 days and then stored in liquid N2.3 For the OH-flooding treatment, the H2-loaded preform was heated at 1000 °C in an electric furnace for 10 s. All the samples used in the present study are listed in Table 1.
The PL spectrum mapping was obtained by measuring the PL spectrum under different excitation energies using a Hitachi 850 fluorescence spectrometer. Each PL spectrum was obtained by scanning the wavelength (λem) from 200 nm (6.20 eV) to 800 nm (1.55 eV) at 0.5nm intervals with a bandpass of 2.0 nm. The excitation wavelengths (λex) ranged from 200 nm (6.20 eV) to 500 nm (2.48 eV) at 5 nm intervals with a bandpass of 2.0 nm.
The optical absorption spectra were measured in the UV region using an Hitachi U-3500 spectrometer (Hitachi, Japan). A KrF excimer laser (λ=248 nm, E=5.0 eV, Lambda Physik, Germany) was used as the excitation source. The refractive index at 633 nm of the PECVD film was measured using a Metricon model 2010 prismcoupler.
3. Results and discussion
3.1. Photoluminescence spectrum mapping of the GeO2-SiO2 glasses
It is reported that the PECVD method is likely to produce a large amount of photoactive defects in the film, and the resultant sample shows a large photoinduced refractive index change exceeding 1.0×10-3 without any extra treatments . Figure 1 shows a bird’s-eye view of the PL spectrum of the as-deposited GeO2-SiO2 CVD film (sample A) at room temperature. One can observe three emission bands assigned to Ge2+ as indicated in the figure. The emission band centered at ~4.3 eV under ~5.0 eV excitation is referred to as the α band, which is due to the radiative transition between the first singlet state S1 and the ground state S0 as shown in Fig. 2. The emission band centered at ~3.1 eV under the ~5.0 eV excitation, i.e., the β band, is due to the radiative triplet-singlet transition via an intersystem crossing process between the S1 and T1 states. Both bands can be excited at ~5.0 eV (corresponding to the so-called B band in the absorption spectrum) irradiation. The small band at ~3.1 eV with ~3.7 eV excitation is also due to the same radiative transition between the triplet-singlet states as the β band, but an excitation process of this band is due to the S0→T1 forbidden transition. Therefore, the emission intensity of this band is very weak compared to the α and β bands. The peak intensity of the α and β bands are about 50 and 500 times, respectively, larger than that of this band.
Figure 3 shows the top view of the contour plots of the α-band in (a) the as-deposited CVD film and (b) the CVD film heated at 600°C for 1 hour in air (samples A and B). An interesting feature of the α band observed for these samples is that the emission energy shifts to lower energies (<4.3 eV) with increasing excitation energy (>5 eV). In other words, the Ge2+ center for the higher energy excitation exhibits a larger Stokes shift in its emission. In addition, it is observed that the as-deposited CVD film exhibited a larger distribution on the larger excited energy - smaller emission energy side than did the heat-treated one (Fig. 3(a)). The maximum refractive index change induced by the KrF excimer laser is 2.0×10-3 for the as-deposited CVD film (sample A) and 1.5×10-4 for the heat-treated CVD film (sample B). The peak top of the as-deposited film is located at a larger excitation energy than the heat treated one irrespective of the fact that the emission energy is almost the same. Therefore, the Ge2+ center shows a larger Stokes shift in a highly photosensitive glass than in a less photosensitive glass. The optical absorption spectra of samples A and B are shown in Fig. 4. The total absorption coefficient of the 5-eV band is smaller in sample B than in sample A, indicating that the α band component with a larger Stokes shift in sample A was in part thermally bleached.
Figure 5 shows the contour plot of the α band in (a) the pristine fiber preform and (b) the OH-flooded one (samples D and E). The distribution of the α band in the OH-flooded fiber preform widely spread to the larger Stokes shift regions. The OH-flooded optical fiber was reported to have a 500 times larger photosensitivity than the H2-loaded one . Here again, it is observed that the α band of the photosensitive GeO2-SiO2 glass exhibits a wider distribution on the larger Stokes shift side. The absorption coefficient of the 5-eV band shown in Figure 4(b) increases by a factor of 3 after the OH-flooding treatment. This indicates that the GeO2-SiO2 glass is reduced resulting in the formation of GeODC and that the α band broadening is induced by the OH-flooding treatment. From these results, the Ge2+ center with the larger Stokes shift is considered to play an important role in the photoactivated processes, which produces a large refractive index change. Hereafter, such a photoactive Ge2+ center is referred to as PADGe (Photo-Active Divalent Ge).
3.2. Correlation between the Stokes shift in α band emission, the photosensitivity and local structure around the Ge2+ center in GeO2-SiO2 glasses
Because the energy corresponding the Stokes shift is dissipated as heat to the surroundings via various electronic transition processes, the degree should depend on the structure surrounding the Ge2+ center. In solid materials, this dissipated energy mostly changes to a phonon or thermal vibration. The α-band emission with a large Stokes shift then suggests that the surrounding structure of a Ge2+ center has a large degree of freedom of structural rearrangement. The absorption spectra of the heat-treated CVD film relative to the as-deposited one are shown in Fig. 6. The as-deposited CVD film was heated at 600 °C in air for 1, 2, 5, 10, 20, 30, 40, 80, 100, 120, 150, 180, 240 and 300 s. It can be seen that within the first 10 s the absorption band around 5.7 eV (corresponding to the Ge(2)) relaxed to give a 5.16 eV band, indicating that the structure giving the 5.7 eV absorption band relaxed to become the Ge2+ center. Because the absorption around 5.6 eV bleached first, this component should have a lower activation energy of the thermal relaxation energy than the other defects.
Next we will interpret the present results with the help of the coordination ordinate. It is considered that the Ge2+ center in the photosensitive CVD film has a large distribution of S1 state potential energy curves as schematically shown in Fig. 7. The potential energy curves of S1 state will be displaced to greater equilibrium bond lengths than that of the S0 ground state, because the S1 states usually have more antibonding character. In addition, the equilibrium bond lengths in the S1 state mainly depend on the degree of freedom of the structural rearrangement around the Ge2+ ion, because this point becomes the turning point for the vibration of the Ge atom in the electronic transition process. Therefore, in the present samples, the distribution of the structure around the Ge2+ center causes the distribution of the S1 state potential energy curves. Under lower energy excitation, the Ge2+ undergoes electronic transition from the vibrational ground state (S0) to the vibrational excited state in the S1(a) potential energy curve. It then gives up energy as a phonon nonradiatively followed by emitting the remaining excess energy as luminescence. On the other hand, under higher energy excitation, electronic transition to the vibrational excited state in the S1(b) potential energy curve occurs, and the Ge2+ then emits a lower photon energy. As a result, the electronic transition between the S0 and S1 states is possible over a wide range of excitation energy. In other words, when the caging effect around the Ge2+ center is weak, a large Stokes shift will be observed. The large inhomogeneous line broadening around ~5 eV as mentioned before is considered to be mainly due to this effect. In addition, the distortion of the potential energy curves and the restraint undertaken by an electron in stepping down the ladder of the vibrational levels in an excited state also possibly occurs due to this effect.
Figure 8 shows the absorption spectra around 2800nm of (a) the as-deposited CVD film (sample A) and (b) the OH-flooded fiber preform (sample E) used in the present study. The absorption bands in the infrared region of the as-deposited CVD film and the OH-flooded fiber preform induced by the OH flooding are assigned to the Si-OH and Ge-OH vibrations. The mechanism of H2 loading and OH flooding is deduced as follows. That is, the H2 molecules infiltrating the glass matrix by H2 loading will break the Si(Ge)-O-Si(Ge) bonds in the SiO4 (GeO4) tetrahedral network by the rapid heating-cooling process of the OH-flooding treatment or under irradiation of dense photons in the case of H2 loading, resulting in the formation of Si-OH (Ge-OH) groups. However, the heating time is too short for the thermal relaxation of the GeO2-SiO2 glass structure to occur. In the case of the CVD method, the reactants are introduced into the O2 plasma, equivalent to several thousand degrees Celsius for oxidation, and rapidly quenched to the substrate temperature of 400 °C. This rapid cooling process in the CVD method introduces a large amount of PADGe without a relaxation and the OH groups compared with the VAD method as a result of the oxidation of organic functional groups in the raw materials. Therefore, we concluded that the OH groups introduced by the OH-flooding treatment and by the CVD process have a crucial role in the formation of PADGe. A suggested structure model of PADGe is shown in Fig. 9. The OH group adjacent to the Ge2+ center lowers the dimension of the glass network and increases the degree of freedom of the structure around the Ge2+ center. Such a Ge atom can then move freely along a longer distance between the equilibrium bond length in the ground state and that in the excited state (corresponding to the S1(b) state in Fig. 7). In other words, the stronger antibonding character of the S1 state contributes to a larger Stokes shift. Recently, the authors have proposed a mechanism for the UV-induced structural change in which a Ge2+ center interacts with an adjacent GeO4 unit to form a compressed structure. Taking this mechanism into consideration, the degree of freedom around the Ge2+ center is considered to significantly affect the photoinduced structural change. Consequently, the Ge2+ center that shows a large Stokes shift behaves photoactively during the structural rearrangement under dense photon irradiation.
Figure 10 shows the PL spectrum mapping of the GeO2-B2O3-SiO2 CVD films (sample C). Sample C shows a very large Stokes shift relative to the binary one. In addition, sample C shows an extraordinarily larger maximum photoinduced refractive index change of 2.47×10-3. Here again, we can confirm that the GeO2-SiO2 glass showing a large Stokes shift in the α band has a high photosensitivity. We can then say that B-doping of the GeO2-SiO2 glass induces PADGe as in the case of the OH-flooding treatment and the CVD method. B2O3 has a more negative standard Gibbs free energy of formation (ΔG°r) above 1200°C than does SiO2. On the other hand, the ΔG°r of GeO2 is much less negative than that of B2O3 and SiO2 above 1000°C. Therefore, GeO2 is much more likely to be reduced under the conditions of the co-existence of B, resulting in the formation of a greater amount of PADGe compared with the case of the GeO2-SiO2 binary system. In addition, the local structure of B in the silicate network is reported to be 3-fold coordinated, which reduces the network dimension of the GeO2-SiO2 glass. For these reasons, B-doping of the GeO2-SiO2 glass is considered to effectively induce PADGe without optical attenuation in the telecommunication window differing from the OH introduction.
It has been believed that the photosensitivity of GeO2-SiO2 glasses mainly depends on the concentration of GeODC, in other words, the absorption at ~5 eV. However, we have shown in the present study that more strictly speaking it depends on the concentration of the PADGe which exhibits a larger Stokes shift in the absorption spectrum (Fig. 11). In fact, although sample C exhibits the largest photorefractivity, the intensity of the 5 eV absorption band is not the largest. The present authors propose that absorption around 5.6 eV with the larger Stokes shift plays the most important role in the photoinduced chemical reaction.
In a number reports concerning the PL spectra of GeO2-SiO2 glasses, the α band is explained by assuming the presence of α1 and α2 components  However, from the present PL spectrum mapping, it is more reasonable to think that they are identical, because the distribution of the structure around the Ge2+ center affects the λem and λex in complicated ways.
Up to now, it has been long believed that the photosensitivity of the GeO2-SiO2 glasses depends on the concentration of GeODC. The present study has, however, revealed based on the correlation between the photosensitivity of the GeO2-SiO2 glass and the Stokes shift of the α band in the PL spectrum mapping that it mainly increases with the increasing concentration of PAGe. That is, the PADGe, the Ge2+ showing a large Stokes shift in the α band, plays a vital role in the photoinduced structural change. In addition, it is found that both the photosensitization method of an optical fiber like the OH flooding and the preparation methods of photosensitive glass such as the CVD method and B-doping can induce the PADGe. It is possible to obtain a more photosensitive GeO2-SiO2 glass by selectively inducing the PADGe.
We thank the Ministry of Education, Science, Sports and Culture, Japan, for the Grant-in-Aid for COE Research on Elements Science, No. 12CE2005. One of the authors (M.T.) acknowledges the financial support by the Industrial Technology Research Grant Program in ‘00 from the New Energy and Industrial Technology Development Organization (NEDO) of Japan.
References and Links
1. Andreas Orthonos and Kyriacos Kalli, Fiber Bragg Gratings: fundamentals and applications in telecommunications and sensing, (Artech House, 1999) Chap. 2.
2. T. Fujiwara, M. Takahashi, and A. J. Ikushima, “Second-harmonic generation in UV-poled glass,” Jpn J. Appl. Phys. Suppl. 37, 15–18 (1998).
3. P. J. Lemaire, R. M. Atkins, V. Mizrahi, and W. A. Reed, “High-pressure H2 loading as a technique for achieving ultrahigh uv photosensitivity and thermal sensitivity in GeO2 doped optical fibers,” Electron. Lett. 29, 1191 (1993). [CrossRef]
4. M. Fokine and W. Margulis, “Large increase in photosensitivity through massive hydroxyl formation,” Opt. Lett. 25, 302 (2000). [CrossRef]
5. D. L. Williams, B. J. Ainslie, J. R. Armitage, R. Kashyap, and R. Campbell, “Enhanced uv photosensitivity in boron codoped germanosilicate fibers,” Electron. Lett. 29, 45 (1993). [CrossRef]
6. M. Svalgraad, C.V. Poulsen, A. Bjarklev, and O. Poulsen, “Direct uv writing of buried singlemode channel wave-guides in Ge-doped silica films,” Electron. Lett. 30, 1401–1403 (1994). [CrossRef]
7. M. Takahashi, A. Sakoh, K. Ichii, Y. Tokuda, T. Yoko, and J. Nishii, “Photochemical process of divalent germanium responsible for photorefractive index change in GeO2-SiO2 glasses,” Appl. Opt. 42, 4594–4598 (2003). [CrossRef] [PubMed]
8. K. O. Hill, Y. Fujii, D. C. Johonson, and B. S. Kawasaki, “Photosensitivity in optical fiber waveguides: Application to reflection filter fabrication,” Appl. Phys. Lett. 32, 647 (1978). [CrossRef]
9. T. E. Tsai, M. A. Saifi, E. J. Frieble, D. L. Griscom, and U. Ötenberg, “Correlation of defect centers with 2nd-harmonic generation in Ge-doped and Ge-P-doped silica-core single-mode fibers,” Opt. Lett. 14, 1023 (1989). [CrossRef] [PubMed]
10. J. Nishii, “Permanent index changes in Ge-SiO2 glasses by excimer laser irradiation,” Materials Sci. Eng. B 54, 1 (1998). [CrossRef]
11. M. Takahashiet al, “Photochemical reaction of Ge2+ in germanosilicate glasses under intense near-UV laser excitation,” J. Appl. Phys. 92, 3442 (2002). [CrossRef]
13. T. Uchino, M. Takahashi, K. Ichii, and T. Yoko, “Microscopic model of photoinduced and pressure-induced UV spectral changes in germanosilicate glass,” Phys. Rev. B 65, 172202 (2002). [CrossRef]
14. R. Tohmonet al, “Triplet-state defect in high-purity silica glass,” Phys. Rev. B 41, 7258 (1992). [CrossRef]
15. M. Martini, F. Meinardi, A. Paleari, G. Spinolo, and A. Vedda, “SiO2:Ge photoluminescence: Detailed mapping of the excitation-emission UV pattern,” Phys. Rev. B 57, 3718 (1998). [CrossRef]
16. N. Chiodini, F. Meinardi, F. Morazzoni, A. Paleari, and R. Scotti, “Ultraviolet photoluminescence of porous silica,” Appl. Phys. Lett. 76, 3209 (2000). [CrossRef]
17. A. Anedda, C. M. Carbonaro, R. Corpino, and A. Serpi, “Low temperature time resolved photoluminescence of the 3.1 and 4.2 eV emission bands in Ge-doped silica,” J. Non-Cryst. Solids 216, 19 (1997). [CrossRef]