We report the kinetics of below band-gap light induced photodarkening in (80-x)GeS2-20Ga2S3-xAgI (x=0 and 20 mol %) bulk chalcogenide glasses by measuring the time evolution of transmission spectra at every 10 milliseconds. The results prove clearly the enhancement of photosensivity upon doping of AgI compound in glasses. It is interesting to find that PD observed in AgI-doped glass totally disappears two hours later after the laser exposing even at room temperature. In significant contrast to 80GeS2-20Ga2S3 glass that the metastable part of PD remains for a long time. We expect such a fast auto-recovery property in AgI-doped glass can be utilized for optical signal processing.
© 2008 Optical Society of America
The exposure of amorphous and glassy chalcogenides to light of proper wavelength and intensity can result in significant structural changes due to their unique physicochemical properties [1–2]. Compared with oxide counterparts, chalcogenides having smaller optical bandgap enables them being much more sensitive to visible light from widely available He-Ne, or Ar+ ion lasers. Generally, photoinduced effects (PE) in chalcogenides can be divided into two groups, namely vectoral and scalar effects. By vectoral effects are meant those determined by polarization state of the inducing light, i.e. photoinduced anisotropy (dichroism and birefringence) , whereas scalar effects do not depend on the light polarization (e.g. photobleaching (PB), photodarkening (PD) and photorefraction (PR) ). There phenomena have been found in Ag-doped chalcogenides as well [5–6]. Recently, we’ve found a new kind of PE, i.e. total recovery of PR (“self-healing”) at room temperature in AgI-doped chalcohalide glass of high glass transition temperature (Tg, 363 °C) . However, our measurement was carried out after the laser was switched off. In this paper, we report the transient characters of below band-gap light induced PD in AgI-doped chalcohalide glass. As a comparison, laser induced PD in 80GeS2-20Ga2S3 glass is also studied.
Glass samples of nominal compositions (80-x)GeS2-20Ga2S3-xAgI (denoted as GGS and GGS-AgI for x=0 and 20, respectively) were synthesized from high purity elements (Ge, Ga and S, 5N) and compound AgI (powder, 5 N) at 1000 °C in evacuated (10-3 Pa) fused silica ampoules for 24 hours, then quenched in the air from 900 °C to room temperature. The obtained glass samples were annealed at 300 °C for 6 hours. Samples with size of 10×10×(0.8~1) mm were well polished to good optical quality. The amouphous character of bulk samples was confirmed by XRD measurements.
The samples were illuminated with an Ar+ ion laser of photon energy (2.54 eV and 2.33 eV) below optical bandgaps (Eg) of samples (GGS of 2.74 eV and GGS-AgI of 2.71 eV). The size of laser spot was 5 mm in diameter. The diameter of the relatively weak probing white light spot from the spectrometer was 2 mm. The two beams were directed such that they crossed each other at the sample with Ar+ laser spot completely covering the light spot from the spectrometer . Simultaneously with the full spectrum, transmission signals were also recorded at fixed wavelength of 470, 480, 493, 500, 515 and 530 nm for GGS, and of 480, 500, 515, 530, 545 and 560 nm for GGS-AgI, using different channels of the spectrometer.
The structure of glass samples were analized by measuring the Raman spectra by a FT Raman spectrophotometer (Bruker IFS 55/FRA 106) with a backscattering method, using the YAG: Nd laser (1064 nm) as the excitation source. All the measurements were executed at room temperature at ambient atmosphere.
Figure 1 shows the full transmission spectra of GGS (a) and GGS-AgI (b) taken before (as indicated by fresh in the insets) and after the laser exposure. The intensities of inducing light (38.8 mW/cm2 and 22.5 mW/cm2 for GGS and GGS-AgI, respectively) are chosen such that the average number of absorbed photons per unit time (s) per unit absorption volume are the same, i.e.≈1018 photons/s·cm3. PD are obseved in both samples. It is compared in the enlarged parts of the full spectra of GGS (inset in Fig.1 (a)) with GGS-AgI (inset in Fig.1 (b)), that the GGS-AgI shows more pronounced PD (larger redshift of the absorption edge). The decrease of transmission in the transparent region (λ>600 nm) may be connected with a beam fanning effect . Compared to the maximum change of transmission spectrum obtained in GGS (as indicated by maximum change in the inset (a)), there is seldom change can be found 24 hours later when the laser is switched off (as indicated by 24 h. later in the inset (a)). Whereas, the optical absorption edge of GGS-AgI returns to its original position only 2 hours later after the laser exposure (as indicated by 2 h. later in the inset (b)). This is consistent with our previous observation that the photoinduced PR almost totally disappeared in 2 hours . Such a total recovery of PD and PR at room temperature in materials of high Tg are scarcely reported in literatures.
Figure 2 shows the time variations of the transmitted signals at fixed wavelength for GGS (a) and GGS-AgI (b). We repeated the on (exposure) and off (stopping of exposure) cycles twice in both samples. When the laser is switched off, the transmission of GGS (Fig.2 (a)) increases slightly and quickly to a constant value that is smaller than the original value. The recovery portion is the “transient” part of total PD, and the portion remaining after exposing to Ar+ ion laser is the metastable part . While in case of GGS-AgI (Fig.2 (b)), the transmission increases significantly within tens of second (fast recovery, indicated by rectangles) followed by a much slower recovery process (slow recovery, indicated by circles). Two hours later after the laser exposure, it reaches the original value. The results are reproducible.
Figure 3 shows the full transmission spectra of GGS-AgI obtained when being exposed by different photon energies (2.54 eV (488 nm) and 2.33 eV (532 nm)). The intensities of inducing lights (22.5 mW/cm2 and 198 mW/cm2 for 488 nm and 532 nm, respectively) are chosen such that the average number of absorbed photons per unit time (s) in unit absorption volume is the same, i.e.≈1018 photons/s·cm3. In both cases, the sample is irradiated for sufficient time so that the spectra obtained are steady state. More prononunced PD is observed when the sample being exposed by 488 nm laser (inset: amplified part of the full spectra). We have demonstrated that much larger PR was induced when the sample being exposed by 514 nm (2.41 eV) laser than by 488 nm laser , partly due to the longer penetration depth of 514 nm laser. However, in the present work, the 532 nm laser of even longer penetration depth does not bring about more pronounced PD. Note that, the absorption coefficient (α) at 532 nm lies in the weak-absorption tail (α≤100 cm-1), while those at 514nm and 488nm fall in the so-called Urbach edge (100≤α≤103 cm-1).
Figure 4 (a) shows the Raman shifts (Stokes) of non-exposing GGS (solid curve) and GGS-AgI (dashed curve). The data are scaled by the height of the GeS4 (343 cm-1) peak of GGS. The addition of AgI into Ge-Ga-S glass matrix leads to two distinct changes as maked by the cycles. First, enhancing of Raman scattering intensity was found in the range of 190 to 250 cm-1, i.e., two new bands 223 cm-1 and 244 cm-1 are formed. They can be ascribed to the presence of iodine-containing structural units e.g. GeS4-nIn . Secondly, weakening is found of Raman scattering in the range of 370 cm-1 to 400 cm-1, which can be assigned to the vibrations of edge-shared Ge(Ga)S4 tetrahedra . The detailed assignments of vibrational modes of GGS-AgI are shown in Fig. 4 (b). Based on these results the glass structure of GGS-AgI can be descibed as say, the tetrahedra e.g. Ge(Ga)S4 are corner bridged with iodine-containing units. The Ag+ ions exist as interstitial ions being hosted by iodide and/or mixed iodosulfide sites .
Figure 5 shows the Raman spectra of GGS-AgI taken before (as indicated by fresh) and after the laser exposure. The wavelength of the illumination for the Raman measurement is 1064 nm (YAG: Nd laser), which is far away from the absorbing region of glass sample. The spectra are presented with the background removed and the data scaled by the height of the GeS4 (343 cm-1) peak. The spectra are indistinguishable before and after expposing, i.e. the height, width and position of the peaks remain constant. Raman spectroscopy should be sensitive to bonding changes involving 1 % or more of the atoms. Not detecing any changes in the Raman spectra means that PD must result from changes in only a small portion of the total glass structure or from changes in the medium- and long-range glass structure. While the Raman data does not establish the structure of the PD state, it does exclude certain types of structural changes from being the source of the PD. First, the formation of crystals upon illumination can be omitted because of no sharp Raman peaks characteristic of crystalline lattice. Second, we see no evidence for the formation of homopolar bonds such as Ge-Ge (259 cm-1), Ga-Ga (208 cm-1) . Because formation of such bonds should be accompanied by the growth of Raman peak at corresponding wavelength.
Generally, the photoinduced changes are much less significant in bulk samples than in thin film counterparts. The results shown in Fig.1–3, though very small, are reliable due to 1) they are reproducible. 2) The resolution of spectrometer is 0.038 nm (FWHM), whereas the maximum shifts of absorption edge of GGS and GGS-AgI are, respectively, 1.6 nm and 4 nm (Fig. 1 and 3). 3) The real error of transparency is no more than 2 %, whereas the maximum changes of transparency around absorption edge of GGS and GGS-AgI are, respectively, 11 % and 27 % (Fig. 2).
PD observed in both materials is due to photoeffects, because the low intensity of the inducing light used would result in negligible temperature rise . Our quantitative calculation shows that the two-photon absorption (TPA) can be neglected as well. We use the two photon absorption coefficient (β) of Ge/Ga-Sb-S glasses (~0.1 cm/GW, ) as a first approximation. Our samples should have similar or smaller values because of larger Eg (β is proportional to Eg-2 ). Since the one-photon absorption coefficients (α) at 488 nm are 6 cm-1 and 14 cm-1 for GGS and GGS-AgI respectively, The TPA process becomes dominant (βI≥α) when the light intensity (I) is greater than 1011 W/cm2. It is much higher than the maximum level of the laser used.
We observed PD in 80GeS2-20Ga2S3 (also written as Ge23.5Ga11.8S64.7), whereas a significant PB was found in Ge25Ga10S65 chalcogenides . The main difference between these two samples lies in the substitution of Ga for Ge. As a result, the glass network of GGS becomes more open. A plausible corollary is that the larger structural flexibility the more likely inducing PD. Such a prediction can be supported by the fact that it is more often PD induced in As-based or As-enriched chalcogenides, whereas PB is more frequently reported in Ge-based or Ge-enriched chalcogenides .
PD observed GGS can be explained using the idea of a percolative growth of photodarkened sites upon illumination by photon-assisted site switching (PASS) . The increase of absorption coefficient Δαr is defined as Δαr=αexp-α0, where α0 and αexp are the original absorption coefficient and the one obtained during the laser exposure. For bulk glass samples, α can be calculated by equation :
where, t is time, Δαsr is the saturated change in Δα, τr is the effective rise time, β is the dispersion parameter (0≤β≤1) to be used in the fitting procedure. Although the physical meaning of β is still not clear, it is generally regarded as light intensity and temperature dependent . The decrease of absorption coefficient Δαd is defined as Δαd=-α0-αoff, where αoff is the absorption coefficient obtained when the laser is switched off. Note that, the penetration depth at 488 nm calculated for GGS-AgI (~0.085 cm) and GGS (~0.12 cm) are close to and larger than the thickness of glass samples (0.088 cm), respectively. The PD effects imparted in glasses are considered as bulk effects. Thus, acceptable are the absorption coefficients calculated by Eq. (1). The errors of absorption coefficient calculated from Eq. (1) are shown in Table I.
Figure 6 shows the time dependence of the increase of Δαr for GGS ((a), left part) and GGS-AgI ((b), left part), respectively. The theoretical curves based on Eq. (2) are indicated by the red lines. It should be noted that quite different characters of PD are observed in GGS-AgI, nonetheless, we can still use Eq. (2) to fit its Δαr. Because very often is that the slow processes of relaxation in glasses exhibit a markedly non-exponential time dependence . In fact, we obtained fairly good fitting with experimental data. We have fitted the data for the entire wavelengths studied but only showed selected wavelength for GGS (470 and 500 nm) and GGS-AgI (500 and 530 nm). The fitting of Eq. (2) to Δαr at different wavelength yields the effective rise time (τr) and the dispersion parameter (β) as listed in Table I. We find in both materials that βr is wavelength independent, whereas τr increases with increasing of wavelength. The variation of τr with wavelength contradicts the usually assumed parallel shift of the absorption edge upon the illumination. It tallies with that reported in As2S3 chalcogenides . Δαr in GGS-AgI increases more sharply and it will reach the saturated value earlier than that of GGS. This can be confirmed by the smaller values of τr (Table I). The time dependence of the decrease of Δαd for GGS ((a), right part) and GGS-AgI ((b), right part) are also shown in Fig. 6. When the laser is switched off, the decrease of the absorption coefficient is the signature of the dark recovery. However, only sight recovery is detected in GGS. That is, the absorption coefficients are not return to the original values. While in case of GGS-AgI, the two-steps dark recovery is observed. That is, a fast sharp recovery (indicated by rectangle) followed by a much slower one (indicated by circle). It conforms to the results shown in Fig. 2. The ultimate absorption coefficients are close to the original values.
PD observed in GGS-AgI can be explained by the “photo-electron-ionic” model . The band structure of GGS-AgI is shown in Fig. 7. The top of valence band (VB) is formed by sulfur p-lone-pair electrons (P-LP) band, closely followed by Ag+d-I-p hybridized band (Ag+d-I-p HB) . The bottom of conduction band (CB) is formed by e.g., S-S wrong bonds (S-S WBs). Upon the laser exposure, electrons would be excited from P-LP band to CB and trapped by defects (self-traped excitons thus formed, STEs). The holes left behind would diffuse into Ag+d-I-p HB. Under the Coulombic replusive interaction with holes, the Ag+ ions start to diffuse. The movement of Ag+ ions would be further accelerated via the Coulombic attractive interaction with trapped electrons or STEs. Thus a local Ag+-rich region is formed, consequently the optical bandgap decreases (PD) due ot the strong coupling of Ag+ d-electron orbitals with S2- P-LP obitals .
One of the possible explanaitons of the “self-healing” observed in GGS-AgI has been suggested in terms of photochemical reactions . Here we shall discuss it from the viewpoint of photoinduced STEs. In glasses, STEs are of Wannier-Mott type . As shown in Fig. 7, the energy levels of STEs lie below the CB. No STEs would exist if the thermal fluctuation at room temperature (~25 meV) were strong enough to overcome the binding energy of STEs. Thus the Ag+ ions gathering around STEs would diffuse back to the unilluminated region because of the strong cation-cation replusive interaction. PD, thereby, would disappear at room temperature. This assumption would explain the two-steps recovery of PD as well. The fast and sharp recovery may relate to the extinguishing of STEs and the much slower recovery may arise from the diffusion of the Ag+ ions back to the unilluminated region.
In conclusion, we have studied the transient characters of below band-gap light induced PD in AgI-doped chalcohalide glass (GGS-AgI). Compared to the AgI-free sample (GGS), two features are worth reiterating here: first, enhanced photosensitivity; second, “self-healing”. The “self-healing” takes place following two steps, i.e. the fast recovery within tens of second and the much slower recovery within few hours. The former may connect with the vanishing of STEs at room temperature. As a result, the latter occurs because of the diffusion of Ag+ ions back to the unilluminated region.
I’m much indebted to Prof. Shimakawa from Gifu University, Gifu 501-1193, Japan for the fruiful discussion. The Author also thank to Ministry of Education, Youth and Sports of Czech Republic for financial support to Research Centre grant LC 523 and grant MSM 0021627501, GA203/06/1368 and to project of the Academy of Sciences of the Czech Republic AVOZ 40500505..
1. M. Frumar, B. Frumarova, T. Wagner, and P. Nemec, “Photo-Induced phenomena in Amorphous and Glassy chalcogenides” in Photo-Induced Metastability in Amorphous SemiconductorsA. V. Kolobov ed (Wiley, Weinheim, 2003) chap 2 p 23. [CrossRef]
2. K. Tanaka, “Sub-Gap Photo-Induced Phenomena in Chalcogenide Glasses” in Photo-Induced Metastability in Amorphous SemiconductorsA. V. Kolobov ed (Wiley, Weinheim, 2003) chap 5 p 69 [CrossRef]
3. P. Krecmer, A. M. Moulin, R. J. Stephenson, T. Rayment, M. E. Welland, and S. R. Elliott, “Reversible Nanocontraction and Dilatation in a Solid Induced by Polarized Light,” Science , 277, 1799–1802 (1997) [CrossRef]
4. A. C. van Popta, R. G. DeCorby, C. J. Haugen, T. Robinson, J. N. McMullin, D. Tonchev, and S. O. Kasap, “Photoinduced refractive index change in As2Se3 by 633 nm illumination,” Opt. Express. 10, 639–644 (2002). [PubMed]
5. K. Tanaka, T. Gotoh, and H. Hayakawa, “Anisotropic patterns formed in Ag-As-S ion-conducting amorphous semiconductor films by polarized light,” Appl. Phys. Lett. 75, 2256–2258 (1999). [CrossRef]
6. M. Frumar and T. Wagner, “Ag doped chalcogenide glasses and their applications,” Curr. Opin. Solid. St. M. 7, 117–126 (2003). [CrossRef]
7. J. Ren, T. Wagner, J. Orava, M. Vlcek, B. Frumarova, and M. Frumar, “Reversible photoinduced change of refractive index in ion-conducting chalcohalide glass,” Appl. Phys. Lett. 92, (2008) (to be published) [CrossRef]
8. A. Ganjoo and H. Jain, “Millisecond kinetics of photoinduced changes in the optical parameters of a-As2S3 films,” Phys. Rev. B. 74, 024201: 1–6 (2006) [CrossRef]
10. K. Shimakawa and Y. Ikeda, “Transient response of photodarkening and photoinduced volume change in amorphous chalcogenide films,” J. Optoelectron. Adv. Mat. 8, 2097–2100 (2006)
11. J. Heo and J. D. Mackenzie, “CHALCOHALIDE GLASSES ⎕. Vibrational spectra of Ge-S-I glasses,” J. Non-Cryst. Solids. 113, 246–252 (1989) [CrossRef]
12. C. Julien, S. Barnier, M. Massot, N. Chbani, X. Cai, A. M. Loireau-Lozac’h, and M. Guittard, “Raman and infrared spectroscopic studies of Ge-Ga-Ag sulphide glasses,” Mat. Sci. Eng. B-Solid. 22, 191–200 (1994) [CrossRef]
13. M. Mitkova, Y. Wang, and P. Boolchand, “Dual Chemical Role of Ag as an Additive in Chalcogenide Glasses,” Phys. Rev. Lett. 83, 3848–3851 (1999) [CrossRef]
14. V. Nazabal, P. Nemec, J. Jedelsky, C. Duverger, J. Leperson, J. L. Adam, and M. Frumar, “Dysprosium doped amorphous chalcogenide films prepared by pulsed laser deposition,” Opt. Mater. 29, 273–278 (2006) [CrossRef]
15. L. Petit, N. Carlie, K. Richardson, A. Humeau, S. Cherukulappurath, and G. Boudebs, “Nonlinear optical properties of glasses in the system Ge/Ga-Sb-S/Se,” Opt. Lett. 31, 1495–1497 (2006) [CrossRef]
16. K. Tanaka, “Optical nonlinearity in photonic glasses,” J. Mater. Sci. 16, 633–643 (2005)
17. S. H. Messaddeq, V. K. Tikhomirov, Y. Messaddeq, D. Lezal, and M. Siu Li, “Light-induced relief grating and a mechanism of matastable light-induced expansion in chalcogenide glasses,” Phys. Rev. B 63, 224203: 1–5 (2001) [CrossRef]
18. G. Yang, H. Jain, A. Ganjoo, D. Zhao, Y. Xu, H. Zeng, and G. Chen, “A photo-stable chalcogenide glass,” (submitted to Optics Letters)
19. J. Schwarz, H. Ticha, and L. Tichy, “Temperature shift of the optical gap in some PbO-ZnO-P2O5 glasses,” Mater. Lett. 61, 520–522 (2007). [CrossRef]
20. M. Krbal, T. Wagner, M. Frumar, M. Vlcek, and B. Frumarova, “Kinetics of optically-and thermally-induced diffusion and dissolution of silver in evaporated As33S33.5Se33.5 amorphous films: their properties and structure,” Phys. Chem. Glasses. 47, 193–197 (2006)
21. A. Laref, W. Sekkal, A. Zaoui, M. Certier, and H. Aourag, “Tight-binding calculations of electronic properties of AgI,” J. Appl. Phys. 86, 4435–4439 (1999) [CrossRef]
22. M. Kastner, “Bonding Bands, Lone-Pair Bands, and Impurity States in Chalcogenide Semiconductors,” Phys. Rev. Lett. 28, 355–357 (1972) [CrossRef]
23. J. Singh and H. E. Ruda, “Concept of Excitons” in Optical Properties of Condensed Matter and ApplicationsJ. Singh ed (John Wiley & Sons, Ltd, 2006) chap 4 p 63 [CrossRef]