1. Introduction

Photoinduced phenomena in chalcogenide glassy semiconductors, especially photoinduced structural transformations [1] and photoinduced optical anisotropy [2], continue to attract the attention of scientists and engineers [1–9]. Photoinduced structural transformations are manifested in reversible effects of photodarkening, which is the change in the optical band gap, film transparency and also in the change of some other film properties such as the change of solubility in different organic and inorganic solvents [10]. Photoinduced optical anisotropy is the emergence of dichroism and/or birefringence in an initially optically isotropic chalcogenide film under the action of linearly polarized light. This effect is explained by the orientation of interatomic bonds or specific defects of the glass, leading to the appearance of some optical axis with a direction determined by the polarization vector of the exciting light [2]. Both effects, photodarkening and photoinduced anisotropy in chalcogenide glassy films, are widely studied and applied in electro-optics [1,2,9]. However most of the investigations have been carried out with films thicker than 0.5 – l µm. Study of photodarkening in the nano-dimensional (ND) glassy chalcogenide films revealed contradictory results. Tanaka et al. have found that the reversible photodarkening disappears in glassy As₂S₃ films thinner than ~50 nm, when illuminated at room temperature [11]. Hayashi and Mitsuishi [12] observed the photodarkening disappearance at room temperature when the thickness of the glassy As₂Se₃ film was less than 50 nm. At the same time, Eguchi et al. observed the photoinduced changes in the transparency of the ND (65 – 200 nm) As₂S₃ films at ~80 K [13]. At last, Indutnyi and Shepeljavi have shown that even very thin (0.7–2.5 nm) As₂S₃ films, which were embedded in transparent SiO layers, had essential photodarkening at 80 K [14]. K. Tanaka, who analyzed the question of photodarkening of very thin chalcogenide films, concluded that further studies remain to be done [15]. As to the photoinduced optical anisotropy, this phenomenon in the ND glassy chalcogenide films, was not studied at all to the best of our knowledge. Its existence widens the range of applications of chalcogenide glassy films such as higher capacity of memory storage and photoalignment of liquid crystals.

Few years ago photoalignment of liquid crystals on thick As₂S₃ glassy films was reported [16] and very recently it was also shown to be possible in ND films and even more efficient than the use of thick films [17]. Understanding the importance of search and development of new efficient photoalignment materials for a practical application [18], we decided to study the peculiarities of photoinduced optical anisotropy and other photoinduced phenomena in the ND glassy chalcogenide films and to compare them with such phenomena in thick chalcogenide films.

2. Experimental

Chalcogenide glassy As₂S₃ films were fabricated by thermal evaporation of commercially available As₂S₃ glass (Amorphous Materials, Garland, TX) from quartz crucibles onto the clean glass substrates in vacuum ~2-5 10⁻⁶ Torr. Deposition occurred with the source to substrate distance of 30 cm at a rate typically of 0.2–0.3 nm/s. Different film thicknesses were investigated: 10, 20, 50, and 100 nm.

For the measurements we applied the two-beams method used in our previous study of photoinduced processes in chalcogenide glassy films [2]. We used the apparatus shown schematically in Fig. 1 . The beam of Ar⁺ laser (488 nm) (1) is divided using semi-transparent mirror (2) and mirror (5) into two beams. The first is the more powerful 1.24 W/cm² pumping beam used to produce photodarkening and photoinduced anisotropy in the sample - ND As₂S₃ film (9). The linear polarization state of this beam could be changed to the orthogonal one with a half wave plate (3). This beam can be interrupted with the aid of a shutter (4).

 

Fig. 1 Experimental setup for the investigation of the photodarkening and photoinduced anisotropy. 1 – Ar⁺ laser, 2 – Semitransparent mirror, 3 – Half wave plate, 4 – Shutter, 5 – Mirror, 6 – Attenuator, 7 – Electro-optical modulator, 8 – Polarizer, 9 – Experimental sample, 10 – Photodiode, 11 - Lock-in amplifier, 12 - Computer.

Download Full Size | PDF

The second beam weakened by an attenuator (6), down to 0.4 mW/cm², functions as a probing beam. This attenuated light beam passes through an electro-optical modulator (7), which modulates the light polarization alternately between two orthogonal states at a frequency of 2 kHz. Then this probing beam passes through the sample (9) and incident on the Si photodiode (10), permitting to measure the photoinduced transmission anisotropy (dichroism) D = 2 (I_y - I_x) / (I_y + I_x), where I_y and I_x are the intensities of the beams with two orthogonal polarization vectors. For the measurement of the difference signal (I_y – I_x) we used the method of synchronous detection. The current of the photodiode is amplified with a lock-in amplifier (11) and fed into a computer (12).

The kinetics of photoinduced dichroism was studied by exposing the studied film to the pump beam either persistently or for intervals of ~500 seconds separated by ~500 seconds periods of rest. Alternating periods of pumping and rest were produced using a shutter. When the photodarkening effect was studied, the polarizer (8) additionally was placed into the weakened laser beam, permitting to irradiate the sample (9) with the pulses of the attenuated light beam and to measure photoinduced transmission changes as a function of time. All experiments were performed at room temperature.

3. Experimental results

In Fig. 2 the photodarkening effect is demonstrated for two of the studied ND As₂S₃ films of different thickness. All these data are obtained using maximum value of pumping beam intensity 1.24 W/cm². The relative changes in transmittance T/T₀ in all studied ND films (12% – 20%) are essentially much smaller (due to their much smaller thickness) than those usually observed in thick films (100%-500%), but the existence of photodarkening in these ND films is determined very reliably. In Fig. 3 we compare for one of these films the photodarkening process under the action of the exciting beam at three different intensities: 1.24 W/cm², 0.24 W/cm² and 0.07 W/cm². We can conclude that in the studied range of light intensity the final value of darkening weakly depends on the light intensity but the time required for reaching the final value increases essentially as the intensity decreases.

 

Fig. 2 Transparency changes in As₂S₃ films with different thicknesses: (1) 20 nm (blue) (2) 100 nm (red), induced by the laser beam with intensity 1.24 W/cm². Pump and probe beams polarization vectors are parallel along y.

Download Full Size | PDF

 

Fig. 3 Transparency changes in 100 nm thick As₂S₃ film induced by the laser beam with intensities: (1) 1.24 W/cm², (blue) (2) 0.24 W/cm², (red) and (3) 0.07 W/cm², (green). Pump and probe beams polarization vectors are parallel along y.

Download Full Size | PDF

Then, we decided to check whether the photodarkening effect in the nano-dimensional glassy As₂S₃ films is accompanied by the change of solubility in different organic and inorganic solvents, just as it proceeds in thick As₂S₃ films [10]. For this aim, we irradiated the nano-dimensional films by the Ar⁺ laser beam of intensity 1.24 W/cm² during 300 sec and measured the time of complete dissolution of irradiated and non-irradiated areas in the solution containing 10% of Morpholine and 90% of Dimethylsulfoxide, which was shown to be efficient selective developer for thick As₂S₃ films [10]. Results of this control are shown in the following Table 1 , where the contrast of dissolution (ratio of the full dissolution times for irradiated and non-irradiated parts) in the nano-dimensional films is compared with that for the 2 µm As₂S₃ films, irradiated in the same regime.

Table 1. Selective Dissolution of As₂S₃ Films

View Table | View all tables in this article

It is seen that the effect of photoinduced change of dissolution in the nano-dimensional films is comparable (or even stronger) to that for the 2 µm As₂S₃ films. These data allow us to conclude that the ND As₂S₃ films can function as efficient photoresists similar to thick As₂S₃ and As₂Se₃ films [10,19].

Figure 4 illustrates the photoinduced dichroism in the ND As₂S₃ films of different thickness. All characteristics are obtained with maximum value of pumping beam intensity 1.24 W/cm². It is seen that in the films with thickness 20 – 100 nm the value of saturated dichroism feebly depends on the thickness but in the 10 nm film this value is significantly lower. This Fig. 4 represents not only dichroism generation but also dichroism reorientation. Dichroism can be reversed many times practically without fatigue by switching the pump light polarization to the orthogonal one. After stopping the irradiation, the level of the dichroism decreases slowly, similar to the same relaxation process observed in thick chalcogenide films [20], but the relaxation in the ND films proceeds quicker than in the thicker films. In the case of 10 nm film after stopping the irradiation, the dichroism quickly diminishes to zero (approximately within 5 minutes).

 

Fig. 4 Kinetics of linear dichroism generation and reorientation in As₂S₃ films with thicknesses: (1) 100 nm, (blue) (2) 50 nm, (red) (3) 20 nm, (green) and (4) 10 nm, (dark blue) under the action of linearly polarized laser beam with intensity of 1.24 W/cm² having horizontal (I_x) and vertical (I_y) directions of the polarization vector. The probe beam polarization is modulated between x and y directions at 2 kHz frequency and the anisotropy is calculated every 1sec.

Download Full Size | PDF

We do not see here essential difference in the kinetics of anisotropy growth for initial dichroism generation and its reorientation, which was clearly recorded in the case of thick films [2], but such difference appears when we decreased the intensity of the pumping beam. As it is seen from Fig. 5(a) , obtained for the beam intensity 0.07 W/cm², the initial dichroism growth in the non-treated film is rather slow (20 – 30 min) while the dichroism reorientation occurs much faster (< 3 min). In the case of long irradiation with the non-polarized light, the following irradiation with linearly-polarized light results in the rapid appearance of dichroism as it is shown in Fig. 5(b). Our experiments showed also that by varying the light intensity the values of saturated dichroism change very weakly, while the time of the dichroism growth strongly depends on the light intensity.

 

Fig. 5 (a) Kinetics of linear dichroism generation and reorientation in 100 nm As₂S₃ film under the action of linearly polarized laser beams with pump intensity 0.07 W/cm² having horizontal (I_x) and vertical (I_y) directions of the polarization vector. (b) The same when polarized light irradiation starts after long irradiation with non-polarized light. The probe beam polarization is modulated between x and y directions at 2 kHz frequency and the anisotropy is calculated every 1sec.

Download Full Size | PDF

In Fig. 6 we demonstrate that under irradiation with circularly-polarized light (at the moment designated by _{$I_0$}), the photoinduced dichroism in the ND films decreases up to zero, just as it occurs in thick chalcogenide glassy films [2]. In addition, this decrease proceeds much quicker than in the thick films case. This effect was observed in all studied ND films of different thickness.

 

Fig. 6 Kinetics of linear dichroism generation and reorientation in 100 nm As₂S₃ film under the action of linearly polarized laser beams with pump intensity 0.07 W/cm² and also the dichroism decay under irradiation with circularly-polarized light of the same intensity: $I_0=I_x=I_y$.

Download Full Size | PDF

4. Discussion

The reliable establishment of the existence of the photodarkening effect in very thin chalcogenide films, presented in this work, is important for constructing different electro-optical devices based on such films. Contradictory results were reported earlier on the study of photodarkening in such nanoscale-thick glassy films, where some investigators observed and some others did not observe photodarkening in the ND glassy As₂S₃ and As₂Se₃ films. In our opinion this is due to three reasons: (i) the low signal from such thin films which needs more careful measurement, (ii) the low light intensity that was used for excitation in previous works, and (iii) due to the measurement of the transparency spectrum shift, which was very small. All early investigators applied Xenon or Mercury-Xenon lamps having a power in the range of 5 – 20 mW/cm² [11–13]. Therefore, in our experiments we decided to apply for excitation a strong laser light with a power ~1.24 W/cm² and to measure not the photoinduced shift of transmission spectrum but direct photoinduced change of the studied film transmission, using the reliable two-beam method of investigation. In consequence, we observed the photodarkening effect in all studied ND As₂S₃ films. In order to confirm our conclusion, we compared the results of irradiation of ND and thick As₂S₃ films with strong Ar⁺ laser light (1.24 W/cm², 488 nm) and weak (~15 mW/cm²) Mercury light. We measured the values of photodarkening (T/T_o) after irradiation during 10 min in all these cases. These results are shown in the Table 2 .

Table 2. Photodarkening in the As₂S₃ Films Using Different Light Sources

View Table | View all tables in this article

The data in Table 2 clearly demonstrate that in the ND films (20 nm and 50 nm) only very weak photodarkening (0.98 and 0.96) was observed under Mercury light irradiation and much stronger photodarkening (0.17 and 0.12) had place with the laser light irradiation. At the same time, for the 1000 nm film with both irradiation sources essentially larger values of photodarkening were measured. These results confirm our assumption that the reason of the photodarkening absence in the very thin films, reported in [11,12], is the application of the measurement method with small sensitivity.

Existence of the photodarkening effect in the nano-structured films correlates well with the other effects of photoinduced structural transformations – photoinduced change of dissolution rate, which is also shown in this work. The observation reported here of photoinduced structural transformations in ND As₂S₃ films closes existing assumptions about special structure of very thin chalcogenide films and about special effects of the substrate on the property of such films [11]. However substrate effects are not excluded as they can have effects on the kinetics of the photoinduced phenomena particularly when the film thickness becomes very small.

In this research we demonstrated that the properties of photoinduced anisotropy in the ND As₂S₃ films, such as dichroism generation and reorientation, dark relaxation, weak dependence of the saturated dichroism values on the light intensity, diminishing of dichroism till zero under irradiation with circularly-polarized light are practically similar to those in the thick chalcogenide glassy films. Decreasing the ND film thickness in the range 100 – 20 nm leads to the gradual decrease of the saturated dichroism value, but the next decrease of the thickness down to 10 nm is accompanied by the very essential dichroism reduction. Simultaneously, in the 10 nm films a very quick dark relaxation of dichroism was discovered.

The observed phenomena in the ND chalcogenide films allow us to conclude that just as it was assumed for the thick chalcogenide films [2], irradiation with both polarized and non-polarized above-band-gap light creates some centers in the non-irradiated film that can be oriented quickly by linearly polarized light. The constant dichroism saturation value at different light intensities strongly suggests that the number of centers that can be created and/or oriented is limited to a certain value. Such centers can be regarded as the valence-alteration pairs that are characteristic defects in chalcogenide glasses [20–22]. Photoinduced structural transformations of chalcogenide films are also known to occur as monitored by Raman scattering [23,24]. Many of the observations reported here have some similarity to the kinetics of other photoinduced effects in glasses such as that in germanosilicate fibers, although the type of defects generated is different and therefore a more quantitative model can be established based on the kinetic model developed previously by Abdulhalim [25]. The model is based on the existence of transient large energy fluctuations which produce transient traps for generated photocarriers and their energy release to enhance defects generation and subsequent photostructural transformation. The anisotropy is manifested by the driven orientation of the defects along the polarization vector of the pumping light. Once a defect is created it can be in a metastable state and needs to be stabilized. Hence according to this model the kinetics is dominated by two time constants, the first which is the generation time constant, _{${\tau }_G$}, relatively fast (on the order of tens of seconds) and a 2nd time constant ${\tau }_S$ which is very slow (of the order of hours or days) that depends on the stabilization process of the defects. Preliminary results of fitting the temporal behavior are presented in Fig. 7 showing that indeed a best fit is obtained using two time constants (correlation better than 97%). The generation time constant decreases with the light intensity as _{${\tau }_G\propto 1/I$} for a single photon absorption process and as ${\tau }_G\propto 1/I^2$ for a two photons absorption process. This behavior explains the intensity dependence of the relaxation rate observed in Fig. 3 and the initial fast rise and decay of the photoinduced signals presented in Figs. 4 and 5. The steady state value of the photoinduced dichroism according to the same model [25] depends on the total number of defects that can be created, which is certainly a function of the film thickness particularly when the film thickness becomes few tens of nm. It also depends on the ratio between the defects generation and stabilization rates which can be influenced by the substrate effects when the films are less than 20nm thick. More quantitative and in depth theoretical investigation of the effects observed here is on going and is planned to be published in a separate paper.

 

Fig. 7 Experimental (dots) photoinduced dichroism signal versus time and fitting curves using a two exponentials expression (solid curves) for the three samples of thickness 20nm, 50nm and 100nm under the conditions of Fig. 4. The data in the table are the two exponential decay constants and coefficients (_{$A_G,A_S$}) for the generation and stabilization processes.

Download Full Size | PDF

As it was shown in our previous research of thick chalcogenide films [19], the phenomenon of photoinduced anisotropy is characteristic of many different chalcogenide glasses which have different optical, electrical, thermal and physico-chemical properties. The photoinduced anisotropy is expected to be characteristic also for ND films of various chalcogenide glasses possessing different properties. It should be mentioned that we have already observed similar behavior on the chalcogenide glass Ge₃₂As₈S₆₀. Therefore, using chalcogenide glassy films for photoalignment of various liquid crystals allows selecting films possessing necessary properties in each specific case. It is important to address that the ND chalcogenide films weakly absorb visible light and are practically completely transparent in the infrared range. This property is very prospective in the future use of such films for photoalignment of liquid crystals devices to be used in the IR range.

5. Conclusions

Our experiments demonstrate that photoinduced structural transformations, expressed in photodarkening and in the change of the dissolution rate in some selective solvents, are characteristic for the ND glassy As₂S₃ films just as they exist in thick chalcogenide films. Linearly polarized light was shown to excite also the linear dichroism in the ND glassy As₂S₃ films. This dichroism can be multiply reoriented when changing the orientation of the polarization vector of the pump beam. Dark relaxation of photoinduced dichroism in the ND films proceeds quicker than in the thicker films. Preliminary model based on two exponential rate processes seems to explain the main results, however further analysis will be performed both experimentally and theoretically to better understand this phenomenon. Low light absorption and possibility of changing the physico-chemical properties when selecting the chalcogenide films for the photoalignment of liquid crystals are certainly very important factors. Having photoinduced anisotropy in nanoscale thick films triggers another important application, and that is for large capacity optical data storage devices based on multilayered structures of ND chalcogenide glass films.