1. Introduction

Solar energy related materials and devices play an important role for creation of renewable and sustainable energy for future resources. The III-VI diindium trichalcogenides including In₂S₃ and In₂Se₃ are defect semiconductors that usually act as a buffer layer or an absorption layer consisted in a thin-film solar cell [1,2]. For the development of Cd-free solar cells (i.e. avoid CdS), the use of indium chalcogenides as a buffer layer or absorption layer was an advantage for green energy. For the diindium trichalcogenides, the most stable phases existed for the crystalline In₂X₃ (X = S, Se) at room temperature are β-phase In₂S₃ (β-In₂S₃) [3,4] and that of α-phase In₂Se₃ (α-In₂Se₃) [5,6]. There are generally three crystallographic phases of α, β, and γ existed in the In₂S₃ depending on different growth temperatures [3]. The β-In₂S₃ is a stable phase commonly with a tetragonal or cubic-like structure. Its crystalline state relates to a spinel lattice with the cation vacancies randomly located on either the octahedral sites only or on both types of octahedral and tetrahedral sites [7]. The ordered modification of the β-In₂S₃ phase was therefore interpreted as a quasi-ternary compound consisted of In atoms, S atoms, and a lot of vacancies in an unit cell [7]. For In₂Se₃, the alloy possesses at least five different crystalline phases denoted as α, β, γ, δ, and κ found in the literatures [8–10]. The α-In₂Se₃ is the most stable phase of indium selenide at room temperature with a hexagonal layer structure stacking in Se-In-Se = In = Se tetra-octahedral configuration along c-axis in one monolayer [6]. Because of the misvalency between III and VI atoms, a natural defect-like phase of ◘₁-III₂-VI₃ is usually found in the unit cell of general α-In₂Se₃ layers. The “◘” is a vacant site (vacancy) of Se or In atom. Owing to the In₂Se₃ originally show n-type semiconducting behavior [11], the indium selenide is inferred to have stronger Se-vacancy effect (donor) in building the unit cell. The Se vacancies may dominate the band edge and transport property of the layer compound. By the existence of dangling bonds of the S or Se vacancies in In₂X₃ (X = S, Se), a surface oxidation layer can easily form on the material interface of β-In₂S₃ and α-In₂Se₃ under ambient air [12]. Therefore, in additional to the intrinsic β-In₂S₃ and α-In₂Se₃ properties, the surface formation oxide and the natural defects in the diindium trichalcogenides may contribute significantly to the optoelectronic property and photo-electric conversion behavior of the compounds.

In this paper, we evaluate the effect of surface formation oxide and intrinsic defects on improving optoelectronic property of β-In₂S₃ and α-In₂Se₃, which will also enhance and extend the photoelectric conversion range of the chalcogenides. The single crystals of β-In₂S₃ and α-In₂Se₃ are grown by chemical vapor transport (CVT) method using ICl₃ as a transport agent. Different optical techniques of transmittance, photoluminescence (PL), thermoreflectance (TR), photoconductivity (PC), and surface photoconductive response (SPR) measurements were respectively carried out to investigate the band-edge structure and imperfection states of the β-In₂S₃ and α-In₂Se₃ crystals. The β-In₂S₃ and α-In₂Se₃ chalcogenides are confirmed to be a direct semiconductor with an energy gap of 1.935 eV for β-In₂S₃ and 1.453 eV for α-In₂Se₃, respectively. The PL results of the chalcogenides reveal a surface oxidation layer was formed on each of the β-In₂S₃ and α-In₂Se₃ crystals and which emits higher-energy luminescence above band gap. The PC and SPR measurements respectively verified the contributions of photoelectric conversion in the chalcogenides coming from bulk, surface oxide, or intrinsic defects below the band gap of β-In₂S₃ and α-In₂Se₃. On the basis of the experimental results a wide-energy-range and high-efficiency photodetector that combining PC- and SPR-mode operations for β-In₂S₃ and α-In₂Se₃ can be made. The testing results show well-behaved function of photoelectric conversion in near infrared to ultraviolet region via the auxiliary of forming surface oxide on the crystalline face of both defect-like β-In₂S₃ and α-In₂Se₃ chalcogenides.

2. Experiment

The crystals of In₂S₃ and In₂Se₃ were grown by chemical vapor transport (CVT) using ICl₃ as a transport agent [13]. The powdered compounds of the crystals were prepared from the elements (In: 99.9999%, S: 99.999%, and Se: 99.999% in purity) by reaction at about 800 °C for 2 days in evacuated quartz ampoules. To improve the stoichiometry, sulfur (selenium) was added to the stoichiometric mixture of the constituent elements. About 10 g of the synthesized elements together with an appropriate amount of transport agent (10 mg/cm³ ICl₃) were introduced into a quartz ampoule (22mm OD, 17mm ID, 20 cm in length), which was then cooled with liquid nitrogen, evacuated to 10⁻⁶ Torr and sealed. A growth temperature of 820 °C (heating zone) → 680 °C (growth zone) with a gradient of −7 °C/cm was set for In₂S₃ and 950 °C (heating zone) → 800 °C (growth zone) with a gradient of −7.5 °C/cm was set for In₂Se₃. The reaction was maintained for 288 hrs to produce large single crystals. At the end of the growth process, synthetic In₂S₃ single crystals of a red color and a maximum size of 8.0 × 5.0 × 1.5 mm³ with a smooth crystalline surface were obtained. X-ray diffraction measurements confirmed β-In₂S₃ crystalline phase of the chalcogenide, and lattice constants were determined to be a = 7.610 Å and c = 32.331 Å. Energy dispersive X-ray spectroscopy (EDS) indicated a little oxygen is existed in the surface of the as-grown β-In₂S₃ crystal. For In₂Se₃, the synthetic In₂Se₃ plates have a hexagonal shape. The size areas of the as-grown layers are from ten to hundreds μm² and some of them are higher to several mm². The as-grown In₂Se₃ plates show black and shiny surface. The thickness of the hexagonal plates is ranging from 10s to 100s μm. The weak van der Waals bonding between the layers means that the In₂Se₃ layered crystal can be easily separated to thin out from the c plane by using a razor blade or a Scotch tape. The results of X-ay diffraction and TEM confirm α crystalline phase of the as-grown In₂Se₃. The lattice constants of the as-grown α-In₂Se₃ are determined to be a = 4.018 Å and c = 19.228 Å, respectively. X-ray photoelectron spectroscopy (XPS) of the as-grown α-In₂Se₃ surface shows an oxidation layer was formed on the crystal plane of the hexagonal plates. The stoichiometric content of the surface oxidation layer is approximately In₂Se_3-3xO_3x with inner layer content approaching In₂Se₃ and outer layer content close to In₂O₃.

TR experiments were implemented using indirect heating manner with a gold-evaporated quartz plate as the heating element [14]. The thin sample of the chalcogenide was closely attached on the heating element by silicone grease. The on-off heating disturbance uniformly modulates the α-In₂Se₃ and β-In₂S₃ periodically. An 150 W tungsten halogen lamp (or an 150 W xenon-arc lamp) filtered by a PTI 0.2-m monochromator provided the monochromatic light. The incident light was focused onto the sample with a spot size less than hundred μm². An EG&G type HUV-2000B Si photodetector acted as the detection unit and TR signal was measured and recorded via an EG&G model 7265 lock-in amplifier. PL experiments were carried out using a QE65000 spectrometer. A Q-switched diode-pumped solid-state laser of 266 nm was employed as the pumping light source. A closed-cycle cryogenic refrigerator with a thermometer controller facilitates the low-temperature measurements.

The PC and SPR experiments are the optical techniques do not need any optical sensor. The photodetector is the sample itself. For SPR measurement, a thin sample was attached on a copper sample holder by silver paste, and the holder was the bottom electrode of the measurement. The top surface of the thin sample can be micropatterened and coated with one gold or copper mesh on the sample plane. This is the top electrode of the SPR measurement. The incident monochromatic light source of SPR is similar to TR. A load resistor connected in series with the sample was used for sensing the photocurrent under incident light’s illumination. A DC voltage was supplied to the circuit. For PC measurement, the β-In₂S₃ or α-In₂Se₃ sample was cut into a rectangular shape with indium coating two ends acted as the ohmic contact electrodes. The main difference between the SPR and traditional PC measurements is the direction of applying electric field to the sample. For α-In₂Se₃, the SPR is normal to the c plane (ε||c-axis) while PC is the in-plane photoconductivity of the c plane (ε⊥c-axis).

3. Results and discussion

Displayed in the left hand side of Fig. 1 is the crystal morphology of as-grown β-In₂S₃ crystals. The crystals reveal red colored and transparent type. Most of them present favorite crystalline faces of {100} and {111} such as a cubic like structure. According the formation of the outline shapes of the as-grown crystals, the crystallographic system of In₂S₃ tends to be that of a cubic or tetragonal phase. The crystallography of In₂S₃ is naturally a defect semiconductor. Its crystalline state relates to a spinel lattice with the cation vacancies randomly located on either octahedral sites or on both octahedral and tetrahedral sites [7]. In general, the β-phase In₂S₃ (tetragonal) [15] possesses the same structure as that of α-In₂S₃ (cubic) except that the cation vacancies of α-In₂S₃ are disordered in its lattice [15]. For α-In₂Se₃, the easily forming face of the layered α-In₂Se₃ can be the c plane. It means that the slowest growth rate of the hexagonal α-In₂Se₃ is along c axis and an approximately equal-fast growth rate will occur on the individual edge plane of the hex-octahedron to appear in a nearly identical edge length. With this rule, the α-In₂Se₃ microcrystal can easily form a perfect hexagon plate. As shown in the right hand side of Fig. 1, the scanning electron microscopy (SEM) image displays different sizes of hexagonal microplates which crowd together to form some well-defined hexagonal micro lotuses arranged at one edge of the basal plane for α-In₂Se₃. The hexagonal α-In₂Se₃ is essentially a layered type crystal with a relevant c plane appeared in its outline.

 

Fig. 1 Crystal morphology of β-In₂S₃ and α-In₂Se₃ crystals. The left hand side shows the crystal picture of β-In₂S₃ which seems to be a cubic- or tetragonal-like structure. The right hand side displays the SEM image of layered α-In₂Se₃ which presents a beautiful cluster of hexagonal micro plates at the basal plane edge of α-In₂Se₃.

Download Full Size | PDF

Figure 2 displays the room-temperature TR and transmittance spectra of (a) β-In₂S₃ and (b) α-In₂Se₃ in the energy range between 1.25 to 2.4 eV near band edge. The dashed lines are the experimental TR spectra and the open-circle lines are the least-square fits to a derivative Lorentzian line-shape function expressed as ΔR/R = Re[${\displaystyle \sum A}\cdot e^{j\text{φ}}{({\text{E-E}}_{\text{g}}{}^{\text{d}}+j\Gamma )}^{-m}$] [16], where $A$ and ${\text{φ}}_{}^{}$ are the amplitude and phase of the line shape, and ${\text{E}}_{\text{g}}{}^{\text{d}}$ and ${\Gamma }_{}^{}$ are the energy and broadening parameter of the interband transition. The value of m = 0.5 is used for the first derivative line shape analysis of critical-point transition of direct band gap for β-In₂S₃ and α-In₂Se₃ [16]. The obtained direct band gaps at 300 K (indicated by arrow) are E_g^d = 1.935 eV for β-In₂S₃ and E_g^d = 1.453 eV for α-In₂Se₃, respectively. The values of E_g^d for the chalcogenides are approximately in accordance with the center location of each transmittance spectrum for matching the absorption edges of β-In₂S₃ and α-In₂Se₃ as indicated in Fig. 2. It lends clear evidence that both β-In₂S₃ and α-In₂Se₃ are direct semiconductors with the gap values dominated in near infrared to red color portion. The α-In₂Se₃ has a lower band gap, and therefore it possesses a wider energy range for optical absorption. The direct band gap of In₂X₃ is originated from a chalcogen (X = S, Se) p state to the indium 5s transition [17,18].

 

Fig. 2 Transmittance and TR spectra of (a) β-In₂S₃ and (b) α-In₂Se₃ semiconductors. The TR transition features (transition energies) matched well with the center locations of the transmittance spectra for the two chalcogenides. The experimental evidence indicated that β-In₂S₃ and α-In₂Se₃ are direct semiconductors.

Download Full Size | PDF

To verify the existence of surface formation oxide in the crystal plane of β-In₂S₃ and α-In₂Se₃, PL measurements of the two chalcogenides were respectively carried out. Figure 3(a) shows the PL spectra of a surface polished and a surface unpolished β-In₂S₃ sample at 300 K. With the energies below 1.7 eV, two defect luminescences by sulfur vacancy (V_S) to indium vacancy (V_In) recombination are observed. The emissions of E_g^d and E^ox are respectively the recombination coming from direct band gap and surface formation oxide β-In₂S_3-3xO_3x [19]. From Fig. 3(a), the thinner oxide layer (see the surface-polished sample) enhances the pumping intensity of the 266-nm laser beam on β-In₂S₃ to achieve stronger PL emission for the intrinsic defect emissions. For the unpolished β-In₂S₃, the intensity of the surface-state emission (E^ox) is stronger than that of the surface-polished β-In₂S₃. It shows the existence of surface oxidation layer in β-In₂S₃. As shown in Fig. 3(b) are the PL spectra of the c-plane α-In₂Se₃. It is clearly that only one broadened peak centered at ~2.63 eV at 300 K is found. The emission is contributed by the In₂Se_3-3xO_3x (0≤x≤1) surface oxide [20]. When the temperature lowered down to 20 K, the PL peak of Fig. 3(b) shows an energy increase and a line-width (L_W) narrowing effect such as the general semiconductor behavior. The energy of the PL peak at 20 K is about 2.72 eV. The PL results of Fig. 3(b) reveal that the adsorption of oxygen by Se vacancies in α-In₂Se₃ will form an oxidation layer In₂Se_3-3xO_3x (0≤x≤1). Because the formation of surface oxide layer with larger thickness, the PL signal from the inner α-In₂Se₃ bulk [i.e. E_g^d, see Fig. 3(b)] cannot strongly emit out. Also owing to lubricant property between the individual layers, the surface of layered α-In₂Se₃ is hard to be polished. The insets shown below Figs. 3(a) and 3(b) respectively depicted the representative scheme of surface formation oxide for the tetragonal β-In₂S₃ and hexagonal α-In₂Se₃. Because of the misvalency between the III and VI atoms, a defect-like phase and a surface oxidation layer can easily form in the chalcogenide compounds.

 

Fig. 3 (a) Surface polished and surface unpolished PL spectra of β-In₂S₃ at 300 K. (b) The PL spectra of hexagonal α-In₂Se₃ semiconductor at 20 and 300 K. The lower insets respectively depict the surface formation oxide on the sample plane.

Download Full Size | PDF

Figure 4(a) shows the SPR and PC spectra of β-In₂S₃ crystal of 1.25 to 4 eV at room temperature. The measurement configuration of the operation mode for the SPR and PC experiments is also included in the inset for comparison. The load resistor is selected as R3 (i.e. 1MΩ) and the photoresponse is detected in Volt unit by AC lock-in detection. The main difference between the PC and SPR measurements is the dissimilarity of the electric-field direction ε. The SPR is ε// k and PC is ε⊥ k, where k is the optical vector of the incident light. As shown in Fig. 4(a) there are two main peak responses correlated with the direct band gap E_g^d and that of the surface formation oxide simultaneously detected in both the PC and SPR spectra of β-In₂S₃. Especially the SPR spectrum show much higher photoresponses than those of the PC spectrum owing the assistance of surface band bending effect by surface states and surface oxides existed on the crystal plane of β-In₂S₃. It leaves relevant evidence that a surface formation oxide was formed on the indium sulfide, and which assists photoelectric conversion of β-In₂S₃ in the higher energy portion above band gap significantly. The lower energy portion of Fig. 4(a) can also display a defect-related absorption (E_D) below E_g^d in the β-In₂S₃. For α-In₂Se₃, the testing of photoresponsivity of the PC- and SPR-mode operations was also implemented. Each of the spectra was normalized to its peak intensity (i.e. normalized photoresponse) for comparison of their energy positions. The light source is a tungsten halogen lamp dispersed by a monochromator. The PC mode operation shows a high signal-to-noise ratio response with a peak photosensitivity near 1.5 eV while when we select the SPR mode, a well-behaved photoresponse ranging from ~1.5 to ~3.8 eV can be detected. The results show wide energy range photodetection from near infrared to ultraviolet can be fabricated by using only one material (i.e. α-In₂Se₃) with different mode operations. As shown in the PC mode spectrum of Fig. 4(b), the optical absorption below direct band gap E_g^d still shows higher photoresponse. It lends evidence that a defect-related optical absorption still occurred inside the α-In₂Se₃ material. Figure 4(c) shows the testing functional performance of the PC mode of α-In₂Se₃ at lower energy range. The light source is a xenon-arc lamp which originally contains many plasma line features below 2 eV [see Fig. 4(b)]. For comparison purpose, a commercialized silicon photodetector is also tested and compared. The value of band gap of Si is 1.12 eV, lower than that of 1.453 eV for α-In₂Se₃. However, a lot of sharp line features (E_D) below 1.453 eV can still be detected by the PC mode photodetector owing to the assistance of defect transitions inside the α-In₂Se₃. The defect transitions (E_D) in both β-In₂S₃ and α-In₂Se₃ facilitate the photoelectric conversion range extension to much longer wavelength region. The inset in Fig. 4(c) depicts the defect and band-edge transitions of In₂S₃ and In₂Se₃ obtained the optical measurements. The defect transitions E_D are mainly coming from chalcogen vacancies (V_S and V_Se) and indium vacancies (V_In), which contribute most of the photoelectric conversion responses below the band gap.

 

Fig. 4 (a) The PC and SPR spectra of β-In₂S₃ at 300 K. The operation modes for the PC and SPR measurements are also included. (b) The PC and SPR spectra of hexagonal α-In₂Se₃. (c) Lower-energy test of the PC spectrum of α-In₂Se₃. The inset depicts a representative scheme of defect and band-edge transitions for the indium chalcogenides.

Download Full Size | PDF

4. Conclusions

In conclusion, the β-In₂S₃ and α-In₂Se₃ crystals have been successfully grown by chemical vapor transport method using ICl₃ as the transport agent. The β-In₂S₃ is shown to be tetragonal structure and that of α-In₂Se₃ is a hexagonal layer structure. The transmittance and TR measurements determine and verify the direct band gaps of the In₂X₃ (X = S, Se). PL, PC, and SPR measurements of the β-In₂S₃ and α-In₂Se₃ identify the existence of intrinsic defects and surface formation oxide of the chalcogenides. The surface oxidation layer on the β-In₂S₃ and α-In₂Se₃ assists the photoelectric conversion in visible to ultraviolet region while the intrinsic defects (V_S, V_Se, and V_In) inside the crystals facilitate the optical absorption of near infrared region. By simultaneously using the PC- and SPR-mode operations, a wide energy range and high sensitivity photodetector can be made by the indium chalcogenides. These chalcogenides are native solar energy materials with well-behaved function operating under sunlight from near infrared to ultraviolet.