1. Introduction

The chalcogenide glasses – amorphous semiconductors that contain as a major constituent one or more of the “chalcogen” elements (sulfur, selenium and tellurium), covalently bonded to network formers, such as As, Ge, Sb, Ga, etc. – are of interest for a variety of applications. Their low phonon energy results in wide infrared transmission windows that, depending on composition, can include parts of the short-wave infrared (SWIR), mid-wave infrared (MWIR), and long-wave infrared (LWIR). They have thus been used for bulk optics, optical fiber, and planar photonics applications in these bands [1,2]. Their high optical nonlinearity, with values of n₂ up to >1000 × greater than that of silica, make them useful for a variety of nonlinear applications [3].

Many compositions of chalcogenide glass are readily deposited via thermal evaporation. Furthermore, they often deposit congruently, meaning the composition of the deposited film matches that of the source material. For these reasons, thin films of chalcogenide glasses are of significant interest for integrated optic waveguides for both linear and nonlinear applications [2]. Recently, they have also been proposed for use in dielectric metasurface applications [4,5].

To date, arsenic trisulfide, As₂S₃, has been the most widely used and studied of the chalcogenide glasses for thin film applications. As₂Se₃, however, is of interest as an alternative for several reasons: 1) It has a high linear refractive index, n = 2.8, compared to that of As₂S₃ (n = 2.4), enabling more design flexibility for both waveguides and metasurface applications; 2) Its nonlinearity (n₂>900 × that of silica) is significantly higher than that of As₂S₃ (n₂>200 × that of silica); and 3) Its band gap of 1.8 eV [6] and wide transmission window, spanning from approximately 1.5-14 µm, make it useful for applications in the SWIR, MWIR, and into the LWIR. Passive integrated optic waveguides with As₂Se₃ used as the core material have been demonstrated [6–9]. Recently, As₂Se₃ films have been used for applications including supercontinuum generation [10–12] and high-Q MWIR resonators [13] and proposed for use in slow light generation [14].

Despite these promising characteristics some material challenges related to As₂Se₃ remain. When the authors fabricated initial samples and left them under ambient conditions, the films degraded rapidly. Photo-induced damage in As₂Se₃ has been observed previously in thin films under ambient conditions [15] and under laser irradiation [16]. It has also been observed that UV light leads to photo-induced damage in As₂Se₃ optical fiber [17], significantly reducing the mechanical strength of the fiber [18].

Berkes et al. proposed a mechanism for the photo-induced degradation [15], described by the reaction

This previous work assumes that this degradation process results in a Se-rich (in comparison to As₂Se₃) phase of arsenic selenide glass. Here, we report a series of experiments on As₂Se₃ thin films showing that exposure to oxygen and moisture together with below-band gap light do indeed cause defect formation but that these defects consist of both crystalline selenium and arsenic oxide. We investigated techniques to mitigate these effects and found that a passivation layer (e.g. as thin as 10 nm) of Al₂O₃ deposited via atomic layer deposition (ALD) combined with storing samples in the dark can almost completely prevent these degradation effects.

2. Experiment

Bulk As₂Se₃ samples were batched from purified precursors in an N₂ purged glovebox. A melt was formed by heating the precursors in a sealed quartz ampoule at 750 °C for 10 hours in a rocking furnace. The glass melt was quenched in water and annealed. The glass boule was removed, and small pieces were used as deposition sources. 1.2 µm thick films of As₂Se₃ were deposited via thermal evaporation onto sapphire and BaF₂ substrates. The resulting films were spatially uniform with low surface roughness (root mean squared roughness <10 Å) and with a composition similar to those of the deposition material. The samples on sapphire substrates were kept at a temperature of approximately 27 °C under six different sets of conditions for 344 days, with details shown in Table 1. Samples were kept under ambient conditions in both light and dark; coated with a thin Al₂O₃ passivation layer and kept in both light and dark; stored under N₂ in the dark (no sample was stored under N₂ in the light due to the challenge of establishing a properly calibrated light source into the glove box); and kept under vacuum in the dark as a control. The samples that were exposed to light were illuminated with a halogen bulb, and irradiance was measured with a calibrated optical power meter.

Table 1. Names and storage conditions of As₂Se₃ film samples

View Table

Samples were removed periodically for characterization. Images were captured of the films’ surfaces using an optical microscope with a 20 × objective. XRD was performed using CuKα radiation from a sealed X-ray tube and a Rigaku SmartLab x-ray diffractometer in grazing incidence with a 2θ geometry with ϕ = 1°. The sample on a BaF₂ substrate was kept under the same conditions as the Ambient/light sample in Table 1, and this sample was used for transmittance measurements in an Analect Diamond-20 FTIR spectrometer.

3. Results and discussion

We monitored the degradation of films stored under the conditions detailed in Table 1. Figure 1 shows microscope images of each film at different points in time. The Vacuum sample, which was not exposed to either light or atmosphere, showed no defects and remained pristine after 344 days. The Ambient/light sample showed the greatest degree of defect formation. By Day 133, significant dendrimer growth was evident, and by Day 344, the film was almost completely covered with crystallites. For the Ambient/dark, Al₂O₃ passivation/ambient/light, and N₂/dark samples, defects are visible as early as Day 16, but the number of defects does not increase significantly during the observation period, and the growth of these defects is limited.

 

Fig. 1 Optical microscope images of films at different times.

Download Full Size | PDF

These results indicate that exposure to atmosphere together with light result in the formation of defects. It is important to note that it is below-band gap light that is activating defect formation; while no filter was applied in this case, absorption of the As₂Se₃ is low above the optical band gap, and the photo-induced crystallization can be primarily attributed to the absorption of UV light [17]. The fact that the N₂/dark sample shows defect formation indicates that even the small amount of moisture and O₂ present in the glove box can aid in defect growth. The Al₂O₃ passivation/ambient/light sample does show some defect formation by the end of the experiment, possibly indicating that the surface was not perfectly passivated or that the passivation layer was damaged during handling. The Al₂O₃ passivation/ambient/dark sample shows no significant defect growth and appears to be pristine at the end of the observation period, indicating that As₂Se₃ film degradation is inhibited by eliminating exposure to moisture, O₂, and light.

In order to understand the nature of the defects, XRD data were collected for the films. Figure 2(a) shows XRD results for films kept under each condition at the end of the experiment (after 344 days). Only the Ambient/light sample exhibits sufficient crystallite growth for diffraction peaks to be observed. The other samples all show broad amorphous peaks associated with the As₂Se₃ glass structure. Figure 2(b) shows results for the Ambient/light film as a function of time with diffraction peaks identified. An As₂O₃ peak emerges (2θ = 14°) after only 16 days; Se peaks are observable by 133 days; and strong As₂O₃ and Se peaks are evident after 344 days. Based on Scherrer analysis, both the As₂O₃ and Se crystallites are approximately 250 nm in size after 344 days. The XRD results indicate that the defects, shown for the Ambient/light sample in Fig. 1, consist of As₂O₃ and Se crystallites.

 

Fig. 2 XRD results for (a) films kept under each condition after 344 days (insets: photographs of the Vacuum sample and the Ambient/light sample after 344 days), and (b) the ambient/light film as a function of time with peaks indexed.

Download Full Size | PDF

Given that these XRD results show evidence for crystalline Se as well as As₂O₃, it is clear that Eq. (1) does not provide a full accounting of the defect formation process. A more complete description is given by

It has been shown that a photo-induced crystallization process occurs for Se in which Se crystallites are induced to grow on amorphous films as a result of illumination [19]. The process is similar to that described above for As in which the absorption of photons produces electron-hole pairs, resulting in broken bonds and the potential for clustering. Because this process in known to occur for Se, the emergence of Se crystallites in the case of As₂Se₃ degradation is not entirely surprising. Unlike As₂O₃, the formation of selenium oxide appears to be less energetically favorable here, so the Se remains in crystalline form.

FTIR transmittance data were obtained in order to show the effects of defect formation on the films’ optical properties. The results are shown in Fig. 3, with only the Ambient/light sample showing a significant change during the observation period. Transmittance did not change significantly for the other samples and is thus not shown on the plot. For the Ambient/light sample, fringing is visible at shorter wavelengths due to thin film interference. For this sample, transmittance drops slightly through 31 days and then drops significantly by 344 days as the film becomes highly scattering. By 344 days, we observe absorption peaks at 9.6 µm and 12.5 µm associated with As-O.

 

Fig. 3 FTIR data for the Ambient/light sample as a function of time.

Download Full Size | PDF

Taken together, the microscopy, XRD, and spectroscopy results provide strong evidence that the formation of defects in As₂Se₃ films is due to photo-induced crystallization with the aid of moisture and oxygen, which produces As₂O₃ and Se crystallites. These crystallites increase in number and size to the point where they scatter a large fraction of incident light and have a major impact on the films’ optical properties.

4. Conclusions

Thin films of As₂Se₃ are promising for use in a variety of applications such as integrated optics or dielectric metasurfaces in the SWIR, MWIR, and LWIR. Degradation, caused by photo-induced crystallization in the presence of atmosphere was observed in thin film As₂Se₃ samples. We examined the mechanism of this degradation and found it to be due to the formation of As₂O₃ and Se crystallites. We describe a mechanism that accounts for the evolution of both species. These effects are minimized by passivating a film’s surface with a thin (~10 nm) Al₂O₃ layer and eliminating exposure to below-band gap light. In practice, a long pass filter can be used in order to block light below the band gap while transmitting light at the operation wavelength.