English

• 1.   Introduction

  The opportunities of photovoltaics exist for technologies that promise either significantly higher efficiency of energy conversion or lower production costs. A new generation of mixed organic-inorganic halide perovskites offers tantalizing prospects on both fronts [1-5]. Besides its outstanding optoelectronic properties such as benign defect physics and long carrier diffusion length [6], the halide perovskites bear a mixed ionic and covalent bonding nature. Within the formula ABX3, halide anions (X^-) and group Ⅳ cations (B²⁺) combine to form the BX₆^4- octahedrons through primarily covalent bonds, while small cations (for example, Cs⁺ or CH₃NH₃⁺) occupy the A sites and connect the BX₆^4- octahedrons through electrostatic attraction, forming perovskites [6-8]. To some extent, the ionic nature permits its dissolution in some polar solvents enabling low-cost solution processing, and the covalent backbone permits smooth carrier transport, thus achieving the unique combination of easy fabrication and high-performance device [2, 9, 10]. However, this near-perfect material still has some flaws. One negative aspect is that lead has been a major constitute of all high-performance perovskite cells to date, raising toxicity issues during device fabrication, deployment and disposal [1, 7]. The other is that perovskites generally undergo degradation when exposed to moisture and elevated temperature [11, 12].

  To seek new absorber materials for high-efficiency, low-cost solar cells, here, alkali metal pnicogen chalcogenides MPnQ₂ (M = Na, K, Rb, Cs, Pn = As, Sb, Bi, and Q = S, Se, Te) draw our attention. In addition to their earth-abundance and low-toxicity which is highly desirable for photovoltaic application, these materials also possess the mixed ionic and covalent bonding nature, in analogy to the halide perovskites [7]. Interactions between electropositive alkali metals and sulfide show a large degree of ionic character while the small electronegativity difference between VA elements and chalcogen render their bonds a large covalency [13, 14]. Here we take NaSbS₂, the focus of this paper, as an example. It is consisted of densely packed (1/∞) [SbS₂^-] polymeric chains and sodium ions. The alignment of the chains parallel to a axis is shown in Fig. 1a. The (1/∞) [SbS₂^-] chains are built from tetrahedral SbS₄ unit via corner bridging; Na⁺ ions are sandwiched between these chains to hold them together and achieve charge neutrality, as shown in Fig. 1b. The 3D crystal structure of the minimum unit of a single chain is also shown in Fig. 1c. Such crystal structure resembles the ionic-covalent bonding nature of halide perovskite. Similarly, the ionic band between Na⁺ and (1/∞) [SbS₂^-] polymeric chains makes it easy to dissolve this compound into aqueous solution, showing great potential of low-cost and environmentally benign manufacturing.

  图 1

  

  图 1  Crystal structure of NaSbS₂: (a) view along a-axis. (b) Na⁺ sandwiched between anionic polymeric chains; (c) a single chain.

  Figure 1.  Crystal structure of NaSbS₂: (a) view along a-axis. (b) Na⁺ sandwiched between anionic polymeric chains; (c) a single chain.

  下载: 全尺寸图片 幻灯片

  2.   Results and discussion

  2.1   Synthesis of NaSbS₂ precursor solution and fabrication of NaSbS₂ film

  Since NaSbS₂ possesses the combined nature of ionic bonding and covalent bonding, we naturally resorted to solution process to fabricate NaSbS₂ thin films. According to our previous report about generalized water-processed metal chalcogenide complexes [15], we applied aqueous solution processing to fabricate NaSbS₂ films since water is the cheapest and most environmentally benign solvent. Sb₂S₃ could be readily dissolved in aqueous (NH4)₂S solution under magnetic stirring at roomtemperatureand ambient pressure to prepare the Sb-S metal chalcogenide complexes (MCCs), followed by the addition of NaOH into the aqueous solution. To investigate the dissolution mechanism precisely, we applied Raman spectroscopy to characterize the NaSbS₂ precursor solutions. The typical Raman spectrum of the solution was shown in Fig. S1a in Supporting information. The distinct peak located at 2560cm^-1 indicated a S-H stretching vibrational mode of HS^- species formed upon dissolution of (NH₄)₂S in water and subsequent hydrolysis of S^2- [15]. And the peaks located at 363cm^-1 and 381cm^-1 were attributed to symmetric Sb-S stretching mode of Sb₄S₇^2-, the soluble product of Sb₂S₃ in water, similar to our previous report [16, 17]. None of Na chalcogenide complexes could be observed from the Raman spectrum, confirming that Na existed as free Na⁺ ions in the solution [18]. NaSbS₂ precursor solution may be processed through the following reaction: 2Sb₂S₃+(NH₄)₂S +2NaOH=Sb₄S₇^2-+2Na⁺+2NH₃·H₂O

  In our process, Na⁺ and Sb-S both completely dissolved into water and formed clear molecular-scale solutions. The ionic bond between Na⁺ and (1/∞) [SbS₂^-] chains contributed to the easily soluble characteristic without involving surfactant or other organic additives [19-21], confirming our previous analysis and highlighting the great potential of low-cost solution processing.

  To determine the optimum spraying and annealing conditions, thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) of dried NaSbS₂ precursor solution were first measured. Water as the only solvent can be easily evaporated and (NH₄)₂S can be readily eliminated completely via H₂S and NH₃ within 100 ℃ [22]. Therefore, only one apparent weight loss occurred at low temperature (100 ℃ < T < 220 ℃) corresponding to the leaving of excess sulfur held within the film, as shown in Fig. S1b. The reaction of Na₂S with Sb₂S₃ into NaSbS₂ probably led to the corresponding exothermic peak at this temperature (red line). At T>220 ℃, the weight loss started to cease and turned into flat but an exothermic peak existed at ≈ 305 ℃ which probably indicated the phase transition of NaSbS₂, as will be discussed later. And linear fitting showed the phase transition begin at approximately 281 ℃. To confirm our supposition of phase transition and the as-deposited film was NaSbS₂, we characterized the structure of as-deposited films by temperature-dependent X-ray diffraction (XRD). As shown in Fig. 2a, at low temperature (T < 260 ℃), sharp peaks at 2θ =26.75°, 31.15°, 44.4° could be attributed to the diffraction of the (111), (200), (220) planes of cubic NaSbS₂ (JCPDS 29-1169), respectively. At high temperature (T>300 ℃), sharp peaks at 2u 16.9°, 30.68°, 31.69° corresponded to the planes of (100), (221), (002) of monoclinic NaSbS₂ (JCPDS 32-1039), respectively. When the temperature was between 260 ℃ and 300 ℃, two phases (cubic structure and monoclinic structure) coexisted, echoing previous DSC observation that phase transition occurred at ≈280 ℃. Subsequently, the cooling temperaturedependent XRD patterns were shown in Fig. 2b. It indicated that as temperature decreased even to 100 ℃, the phase structure maintained monoclinic structure. The phase transition was irreversible, at least in our case, which was different from that of the traditional phase-change materials like VO₂ [23] and ZrO₂ [24]. Based on TGA-DSC and XRD results, we applied spray pyrolysis processed at 230 ℃, the minimum temperature to eliminate residual solvent and excess sulfur, to get a pure NaSbS₂ precursor film. And then annealing the precursor film at various temperatures of 300 ℃, 350 ℃, 400 ℃, 450 ℃, 500 ℃ and it turned out that 500 ℃ annealing yielded best results in terms of crystallinity and phase purity. XRD pattern of 500 ℃ annealed device matched perfectly with monoclinic NaSbS₂ (JCPDS 32-1039), indicating phase-pure NaSbS₂ without any impurities or secondary phase was obtained. The monoclinic lattice parameters for the NaSbS₂ film, calculated from the full diffraction pattern, were a = 8.245 Å, b = 8.283 Å, c = 6.845 Å, agreeing well with the published values [25].

  图 2

  

  图 2  (a) Temperature-dependent XRD patterns of NaSbS₂ from 25 to 500 ℃, indicating the phase transition from cubic to monoclinic structure. (b) Temperature-dependent XRD of NaSbS₂ from 400 to 100 ℃. The bottom in panel a and b shows the standard diffraction pattern of NaSbS₂ (JCPDS 29-1169 and JCPDS 32-1039). Insets are the crystal structure of cubic (alpha phase) and monoclinic (beta phase) of NaSbS₂.

  Figure 2.  (a) Temperature-dependent XRD patterns of NaSbS₂ from 25 to 500 ℃, indicating the phase transition from cubic to monoclinic structure. (b) Temperature-dependent XRD of NaSbS₂ from 400 to 100 ℃. The bottom in panel a and b shows the standard diffraction pattern of NaSbS₂ (JCPDS 29-1169 and JCPDS 32-1039). Insets are the crystal structure of cubic (alpha phase) and monoclinic (beta phase) of NaSbS₂.

  下载: 全尺寸图片 幻灯片

  2.2   Material characterization of NaSbS₂ film

  To achieve thin film with good morphology, the conditions (annealing temperature and duration) for the soft and final bake are crucial. Here soft bake denotes the annealing during the spray pyrolysis in air ambient, and final bake designates the final annealing of the NaSbS₂ film inside an inert atmosphere. The optimal fabrication procedure from the precursor solution to the objective thin film follows the rules: (ⅰ) A pure film should be formed at the minimum temperature to exclude the residual solvent and the excessive elements and generate crystalline gains as few as possible. (ⅱ) During the final bake, the temperature should high enough to promote atom movement, thus crystal boundary converges and large and high-quality crystalline gains generate. We designed experiments to comply to this empirical theory utilizing scanning electron microscopy (SEM) to characterize the microstructural evolution as a feed back. According to the aforementioned analysis of TGA-DSC and XRD results, we selected 230 ℃, the minimum spray temperature to produce a pure NaSbS₂ film, as the soft bake condition. SEM spectrum (Fig. 3a) showed that the 230 ℃ sprayed film was porous and individual grains were not clearly distinguishable. Despite annealed at 260 ℃ for 30 min, the highest temperature to maintain the cubic-phase a-NaSbS₂, the size of the gains was still less than 100 nm, as shown in Fig. S2 in Supporting information, which were undesirable for efficient solar cells. We thus increased the annealing temperature and five final bake temperatures were investigated: 300 ℃, 350 ℃, 400 ℃, 450 ℃, 500 ℃. From the top-view SEM (Fig. 3a-f), it revealed that as the annealing temperature increased, the films were gradually densified, and the gain size increased. When annealed at over 500 ℃, the gain size was over 1mm because that higher temperature resulted in faster ions' thermal movement, promoting diffusion and consequently promoting grain growth and sintering [26]. But there were still some occasional pinholes in the film because of the weight loss, as evidenced in the TGA data. The occasional pinholes may lead to current loss and affect the photovoltaic device performance. From the cross-sectional SEM (Fig. 3g-h), the films were about 0.5mm thick and comprised of a monolayer of micrometer size grains, a highly preferable microstructure for solar cells. After high temperature annealing, the monoclinic NaSbS₂ had better crystallization and superior morphology. Therefore, according to the analysis of TGS-DSC, XRD and SEM, most of the subsequent explorations were based on the 500 ℃ annealed β-NaSbS₂ film. Afterwards, energy-dispersive spectroscopy (EDS) was employed to determine the surface composition of the NaSbS₂ film. As shown in Fig. S3 in Supporting information, it confirmed that Na, Sb and S homogenously distributed among the crystals, excluding the compositional gradients within particles or multiphase coexistence, a consequence of the mixing of elements at molecular level and the optimized film fabrication procedure. Moreover, quantitative analysis of EDS data revealed the average Na/Sb/S ratio to be approximately 1:1:2 (Table S2 in Supporting information), which corresponded well with the elemental ratio of NaSbS₂.

  图 3

  

  图 3  The top-view (a–f) and cross-sectional (g–i) SEM images of NaSbS₂ film on FTO after annealing with different temperatures of 230 ℃ (a), 300 ℃ (b), 350 ℃ (c), 400 ℃ (d & g), (e) 450 ℃ (e & h), (f) 500 ℃ (f & i).

  Figure 3.  The top-view (a–f) and cross-sectional (g–i) SEM images of NaSbS₂ film on FTO after annealing with different temperatures of 230 ℃ (a), 300 ℃ (b), 350 ℃ (c), 400 ℃ (d & g), (e) 450 ℃ (e & h), (f) 500 ℃ (f & i).

  下载: 全尺寸图片 幻灯片

  X-ray photoelectron spectroscopy (XPS) was utilized to investigate the chemical nature of the final-baked NaSbS₂ film. Before measurement, the film was etched by argon ion sputtering for 600 s (~10 nm thickness) to eliminate the interference of surface contamination originated from ambient exposure. As shown in Fig. S4a in Supporting information, only elements Na, Sb and S existed and no detectable carbon and oxygen presented in the film within detection limits. Magnified XPS spectrum of Na, Sb and S were shown in Fig. S4b-d. The binding energy of Na 1 s was 1070.15 eV, in good agreement with Na⁺ [27]. Antimony core-level spectrum showed the binding energy of Sb 4d_5/2 and 4d_3/2 was 31.78 eV and 32.98 eV with a separation of 1.2 eV, consistent with Sb³⁺ [28, 29]. Similarly, the binding energy of S 2p_3/2 and 2p_1/2 was 159.97 eV and 161.17 eV, respectively [17]. Both peaks were characteristic of S^2-. The perfect Gaussian-Lorentzian peak fitting for all these three peaks excluded the presence of Sb⁵⁺ and O in the film within XPS detection limit. This is surprising at first glance, yet is fully reasonable after some analysis. First of all, all the reactants and solvents were carbon free thus eliminating the C element from the source. This is a direct embodiment of the advantage of easy dissolution of NaSbS₂ due to their ionic-covalent bonding nature. As a reference, copper zinc tin sulfoselenide (CZTSSe) films often contained detrimental carbon layers at the Mo back contacts due to the incomplete removal of organic additives which were compulsory for the preparation of CZTSSe nanocrystal inks [30, 31]. Second, our NaSbS₂ thin film excluded oxygen impurities for the following reasons: (ⅰ) water as one of the two oxygen sources could be easily evaporated. (ⅱ) OH^-, the other oxygen sources in the solution, could pair with excess NH₄⁺ and evaporated away. (ⅲ) The decomposition products of (NH₄)₂S contain H2S, which could provide a strongly reductive atmosphere. According to our experiences about the water-based solution-processed CZTSSe and Sb₂S₃ film, even if some oxide generated, H2S could convert them into sulfides efficiently [15, 32]. This is probably the most important reason that our soft bake was carried out in air ambient yet our NaSbS₂ film was oxygen-free. Collectively, XPS analysis confirmed the pure phase of NaSbS₂ and the normal valence state of Na⁺Sb³⁺S₂^2-.

  2.3   Optical and electrical characterization of NaSbS₂ film

  In order to characterize optical properties, transmission spectrum of NaSbS₂ film on FTO substrate was measured in the wavelength ranging from 400 to 1400 nm at room temperature, as shown in Fig. S5a in Supporting information. Transmittance began to steeply drop at the wavelength of ~900 nm and declined to zero at the wavelength shorter than 600 nm. Subsequently, according to the absorption spectrum, the absorption coefficient was calculated by using the following simplified formula: α = d^-1ln (T^-1), where d is film thickness determined by cross-sectional scanning electron microscopy (SEM) and T is the measured film transmittance data [33]. As shown in Fig. S5b in Supporting information, the absorption coefficient for NaSbS₂ reached 6 ×10⁴ cm^-1 for photon energy ~2.1 eV (wavelength of 590 nm) which was desirable for solar cell absorber application. Further, the band gap (E_g) related to absorption coefficient was determined by using the following equation: (αhυ)ⁿ = A (hυ-E_g) where A is a constant, h is the Planck's constant, υ is the frequency of the incident photon and n equals to 2 for direct band gap semiconductors and to 1/2 for an indirect band gap semiconductor [34]. Fig. S5c in Supporting information illustrated the plot of (αhυ)^1/2 vs. hυ and a good linear fitting line with its extrapolation to the x-axis was at 1.61 eV while the plot of (αhυ)² vs. hυ could not generate decent fitting results. In order to better understand the band structure of β-NaSbS₂, we performed first-principles electronic band structure calculations for the ground and excited states using the plane wave code CASTEP with the local density approximation (LDA) considering the Ceperly-Alder-Perdew and Zunger (CA-PZ) functional [35-37]. The calculations indicated that the band gap of β-NaSbS₂ was indirect (Fig. S5d in Supporting information). The valence band maximum (VBM) occurred at B while the conduction band minimum (CBM) occurred at Z. The calculated band gap (E_cal = 1.695 eV) was in a good agreement with the band gap (E_g = 1.61 eV) measured by UV-vis spectroscopy.

  To research the band position of our NaSbS₂ film, ultraviolet photoelectron spectroscopy (UPS) was applied to determine the VBM and Fermi energy. In Fig. S6a in Supporting information, the Fermi energy of NaSbS₂ was first established as-4.57 eV by subtracting the intercept at binding energy of 16.64 eV with the calibrated ultraviolet photon energy (He I excitation, 21.2 eV). Using the linear fitting of the UPS spectrum in the long tail to generate an extrapolation, the distance between VBM and Fermi energy was derived as 0.91 eV (Fig. S6a, inset). According to the measured band gap of 1.61 eV, VBM and CBM of our NaSbS₂ film were calculated to be at-5.48 eV and-3.87 eV, respectively. The position of Fermi energy slightly closer to the conduction band demonstrated that our NaSbS₂ film was weakly n-type. The band diagrams of both NaSbS₂ and CdS were plotted in Fig. S6b. Clearly, a type Ⅱ staggered heterojuction could form, facilitate electron injection from NaSbS₂ to CdS. We thus chose CdS as the buffer layer for our NaSbS₂ thin film photovoltaics as shown later.

  The Hall effect was measured to determine the electrical properties of NaSbS₂ film by using the Van der Pauw method [38]. The negative Hall coefficient indicated that our NaSbS₂ film was ntype, matching well with results from UPS measurement. The resistivity, electron mobility and concentration were estimated to be 7.88 × 10³Ω cm, 14-22 cm²/(Vs), and 6 ×10¹³-6 ×10¹⁴ cm^-3, respectively. The estimated electron concentration was relatively low, possibly due to the low incorporation of external impurities from our solution process. Further experiments are ongoing to reveal the underlying mechanism.

  To further investigate the photosensitivity property of NaSbS₂ film, a NaSbS₂ photoconductive photodetector was built. The details of the photodetector fabrication were involved in the Experimental section and schematic configuration of NaSbS₂ based photodetector was shown in Fig. 4a. First, the I-V curve of the contacts between NaSbS₂ film and FTO substrate which was shown in Fig. 4b was tested. The I-V curve was linear under dark (dark line), proving the good Ohmic contact between NaSbS₂ and FTO substrate. Subsequently, upon illumination with 530 nm light generated by a functional generator-controlled light-emitting diode (LED) with an intensity of 430mW cm^-2, current-time (I-t) characteristic of the device was recorded. As shown in Fig. 4c, under an external bias of 40 V, the dark and photocurrent were about 210 and 525 nA respectively, indicating the strong photo responses of our NaSbS₂ film. The drift of the baseline was ascribed to Na⁺ diffusion upon the external field, similar to the well-known field driven I^- in halide perovskite [39]. In addition, as shown in Fig. 4d, to fit the I-t decay curves, two exponential functions expressed by the formula were used:

  图 4

  

  图 4  (a) Schematic configuration of NaSbS₂ based photodetector device. (b) The current-voltage curve of FTO-NaSbS₂-FTO confirming that FTO form ohmic contacts with NaSbS₂ film. (c) Photosensitivity of NaSbS₂ deposited on FTO glass. (d) Fitting curve of photocurrent decay.

  Figure 4.  (a) Schematic configuration of NaSbS₂ based photodetector device. (b) The current-voltage curve of FTO-NaSbS₂-FTO confirming that FTO form ohmic contacts with NaSbS₂ film. (c) Photosensitivity of NaSbS₂ deposited on FTO glass. (d) Fitting curve of photocurrent decay.

  下载: 全尺寸图片 幻灯片

  The bi-exponential decay and two time constants (τ₁ = 122.2 ms, τ₂ = 1484.8 ms) implied that two types of defects with different depths existed in our NaSbS₂ film. The nature of these defects is not clear at this stage, and we assume their negative influence on the performance of our photovoltaic devices.

  2.4   Fabrication and performance of NaSbS₂ thin film photovoltaic device

  Finally, the potential of our NaSbS₂ thin films in photovoltaic device was tested. Solar cells with the substrate structure of glass/ FTO/NaSbS₂/CdS/i-ZnO/AZO/Au were fabricated as shown in Fig. 5a. We chose FTO as the substrate because it could form ohmic contact with NaSbS₂ as evidenced in the photodetector device. As the UPS analysis revealed, CdS could form a type-Ⅱ staggered heterojunction with NaSbS₂ film and thus worked as a buffer partner. In Fig. 5b, we present the J-V characteristics for the first-ever NaSbS₂ thin film solar cell with an efficiency of 0.13%, an open-current voltage (V_OC) of 0.28 V, a short-current density (J_SC) of 1.53 mA cm^-2, and a fill factor (FF) of 29.3%, which was measured under 100 mW cm^-2 simulated AM 1.5 G irradiation. The total area of every individual cell defined by mechanical scribing was 0.108 cm². Considering the very limited optimization work done so far, the result of our device is still very encouraging although the preliminary device efficiency was quite low. What's more, the standard device structure that has been reported compatible with CIGS and CZTSSe may not be compatible with NaSbS₂ due to its weakly n-type nature. Very recently, IBM researchers reported that for weakly n-type Ag₂ZnSnSe₄ absorber, an alternative stack with a SnO:F (FTO) back contact and MoO₃ buffer achieved a promising performance of above 5% [40]. Inspired by them, the following optimization processes are underway: seeking for suitable buffer layer (NiO, MoO₃ etc.), enhancing the quality of NaSbS₂ film, optimizing the band alignment and interfacial defects and so on.

  图 5

  

  图 5  (a) Schematic configuration of photovoltaic device. (b) J-V curves of NaSbS₂ solar cell performance in the dark and under 100 mW cm^-2 simulated AM 1.5 G irradiation, respectively.

  Figure 5.  (a) Schematic configuration of photovoltaic device. (b) J-V curves of NaSbS₂ solar cell performance in the dark and under 100 mW cm^-2 simulated AM 1.5 G irradiation, respectively.

  下载: 全尺寸图片 幻灯片

  3.   Conclusion

  In summary, we systematically investigated most of the fundamental properties of NaSbS₂ including material, optical and electrical properties and fabricated high quality thin films with a low-cost, sustainable spray-pyrolysis method as well as a thinfilm photovoltaic device for the first time. In analogy to the halide perovskites, NaSbS₂ possess the mixed ionic and covalent bonding nature, promising great potential in easy fabrication and competitive photovoltaic characteristics. NaSbS₂ had an irreversible phase transition from cubic phase (α-NaSbS₂) to monoclinic phase (β-NaSbS₂) as temperature exceeding 280 ℃. After high-temperature annealing, the phase-pure, crack-free NaSbS₂ thin films were formed consisted of large gains (> 1 μm). As-deposited film had an indirect band gap of 1.61 eV with CBM and VBM located at-3.87 eV and-5.48 eV, respectively. Moreover, NaSbS₂ was shown as a weakly n-type semiconductor with a carrier density of 6 ×10¹³-6 ×10¹⁴ cm^-3 and an electron mobility of 14-22 cm²/(V s). Finally, a conventional solar cell based on the structure of glass/FTO/NaSbS₂/ CdS/i-ZnO/AZO/Au was preliminarily established, achieving a firstever efficiency of 0.13%. Due to the n-type nature of NaSbS₂, new device structure (back contact, buffer, etc.) was under exploring. Our already accumulating investigation paves the way for previously-inaccessible NaSbS₂ photovoltaics.

  4.   Experimental

  4.1   Chemicals

  Antimony sulfide (Sb₂S₃, powder, 99.999%) was purchased from Alfa Assar. Ammonium sulfide (40-48 wt% in water) and sodium hydroxide (NaOH, pellet, 98%) were purchased from Aladdin. All chemicals were used as received without any further purification.

  4.2   Preparation of NaSbS₂ precursor solution

  All syntheses were performed at room temperature in a wellventilated fume hood. To prepare the Sb-S precursor solution, 0.1699 g of Sb₂S₃ was dissolved in 25 mL ammonium sulfide solution at room temperature, which formed a clear yellow solution after about two days of magnetic stirring. We noticed that the age of the ammonium sulfide solution influenced the dissolution speed, and adding a tiny amount of elemental S generally speeded up the dissolution process. The preparation of NaSbS₂ solution was completed by further adding 0.04 g of NaOH pellet into the Sb-S precursor solution.

  4.3   Film deposition and device fabrication

  The procedure of film preparation and device fabrication is schematically shown in Fig. S7 in Supporting information. NaSbS₂ films were deposited on FTO substrates for two steps: (ⅰ) spray pyrolysis of the NaSbS₂ precursor solution onto a hotplate preheated to 230 ℃ using a home-made automatic spraying equipment. The solution was sprayed through a pneumatically controlled air-atomizing spray nozzle with a rate of 2 mL/min for about 20 min to form the precursor film. After the spray, the film was immediately removed from the hotplate. (ⅱ) The cooled film was transferred into a N2 filled glovebox and annealed on a hotplate with pre-heated temperatures of 300 ℃, 350 ℃, 400 ℃, 450 ℃, 500 ℃. The film was covered with a quartz helmet. After annealed for 5 min, the hotplate was turned off and the NaSbS₂ film was transferred to a cold ceramic substrate in 3 min. The NaSbS₂ films were applied for photodetector and photovoltaic devices. For photodetector devices, NaSbS₂ film which was prepared according to the previously described procedure was deposited onto the strip area where the FTO was etched away by laser on the center of FTO glass. The NaSbS₂ based photovoltaic device was fabricated with a substrate structure of glass/FTO/NaSbS₂/CdS/i-ZnO/AZO/Au. First, a 50 nm thick CdS layer was deposited onto the as-prepared NaSbS₂ film using chemical bath deposition according to a previously reported method [32, 41]. Subsequently, the window layers of intrinsic ZnO (50 nm) and ZnO:Al (400 nm) were deposited by radio frequency sputtering. The conductivity and transparency of our ZnO:Al was approximately 100-150V cm and 80% within visible light range, respectively. Finally, thermal evaporation was used to evaporate gold contacts (≈ 50 nm thick) through a shadow mask on the top of devices. The active device area of each solar cell device was 0.108 cm² defined by mechanical scribing.

  4.4   Materials and device characterization

  Raman spectroscopy was performed on the NaSbS₂ solution in a confocal back scatting configuration at room temperature by using the spectrograph (Horiba JobinYvon, LabRAM HR800, 532 nm excitation). The dried NaSbS₂ solution was characterized by thermal gravimetric analysis (TGA) and differential scanning calorimetry (DSC, PerkinElmer Instruments, Diamond TG/ DSC6300, ramp rate 5 ℃/ min, N₂ flowing environment). The crystal structure of NaSbS₂ thin films prepared on FTO substrate was characterized by x-ray diffraction (XRD, Philips, X pert pro MRD, with Cu Kα radiation, λ = 1.54178 Å). The composition of the films was determined by energy-dispersive spectroscopy (EDS, FEI Quanta 600 scanning electron microscope, 20 kV) and x-ray photoelectron spectroscopy (XPS, EDAX Inc, Genesis, 300 W). The morphology of NaSbS₂ film was studied by scanning electron microscopy (SEM, FEI Nova NanoSEM450, without Pt coating). Electrical properties of NaSbS₂ film were determined by Hall effect (Ecopia HMS-5500, with aluminum electrodes) and the optical transmittance (T) was recorded by UV-vis-near IR spectrophotometer (Perkin-Elmer Instruments, Lambda 950 using integrating sphere). The Fermi level and valence band of NaSbS₂ film were detected by ultraviolet photoelectron spectrometer (UPS, Specs UVLS, He I excitation, 21.2 eV, referenced to the Fermi edge of argon etched gold). For photoresponse measurement, the current-voltage (I-V) and current-time (I-t) curves were performed in an electromagnetically and optically shielded box using Agilent B1500A under a 530 nm LED with a power density of 430mW/cm2. Photovoltaic device performance was measured by a Keithley 2420 source-meter under a solar simulator with an Xe light source (450 W, Oriel, model 9119) and an Air Mass 1.5 G filter, providing simulated 1 sun illumination.

  Acknowledgment

  This work was financially supported by the Major State Basic Research Development Program of China (No. 2016YFA0204000), the National Natural Science Foundation of China (Nos. 61322401, 51402115 and 21403078), the HUST Key Innovation Team for Interdisciplinary Promotion (No. 2016JCTD111), Hubei Provincial Natural Science Foundation of China (No. 2016CFB464), the Fundamental Research Funds for the Central Universities (WUT: 2016IVA089, 2016Ⅲ030) and the director fund of Wuhan National Laboratory for Optoelectronics. The authors thank the Analytical and Testing Center of HUST and the facility support of the Center for Nanoscale Characterization and Devices, WNLO.

• 1. [1]

     Green M.A., A.Ho-Baillie , Snaith H.J. The emergence of perovskite solar cells[J]. Nat.Photonics, 2014, 8:  506-514. doi: 10.1038/nphoton.2014.134

  2. [2]

     Snaith H.J. Perovskites The emergence of a new Era for low-cost, high-efficiency solar cells[J]. J.Phys.Chem.Lett., 2013, 4:  3623-3630. doi: 10.1021/jz4020162

  3. [3]

     G. Hodes, D. Cahen, Photovoltaics Perovskite cells roll forward, Nat. Photonics 8 (2014)87-88.

  4. [4]

     Service R.F. Perovskite solar cells keep on surging[J]. Science, 2014, 344:  458. doi: 10.1126/science.344.6183.458

  5. [5]

     Park N.G. Organometal perovskite light absorbers toward a 20% efficiency low-cost solid-state mesoscopic solar cell[J]. J.Phys.Chem.Lett., 2013, 4:  2423-2429. doi: 10.1021/jz400892a

  6. [6]

     H. S. Jung, N. G. Park, Perovskite solar cells: from materials to devices, Small. 11 (2015)10-25.

  7. [7]

     M. Grätzel, The light and shade of perovskite solar cells, Nat. Mater. 13(2014) 838-842.

  8. [8]

     Kim H.S., Sang H.I., Park N.G. Organolead halide perovskite:new horizons in solar cell research[J]. J.Phys.Chem.C, 2014, 118:  5615-5625.

  9. [9]

     Jeon N.J., Noh J.H., Kim Y.C.. Solvent engineering for high-performance inorganic-organic hybrid perovskite solar cells[J]. Nat.Mater., 2014, 13:  897-903. doi: 10.1038/nmat4014

  10. [10]

      Burschka J., Pellet N., Moon S.J.. Sequential deposition as a route to high-performance perovskite-sensitized solar cells[J]. Nature, 2013, 499:  316-319. doi: 10.1038/nature12340

  11. [11]

      J. Zhao, B. Cai, Z. Luo, et al. , Investigation of the hydrolysis of perovskite organometallic halide CH3NH3PbI3 in humidity environment, Sci. Rep. 6 (2016)21976.

  12. [12]

      Kaltenbrunner, Adam, E.D.Głowacki. Flexible high power-per-weight perovskite solar cells with chromium oxide-metal contacts for improved stability in air[J]. Nat.Mater., 2015, 14:  1032-1039. doi: 10.1038/nmat4388

  13. [13]

      Goodenough J.B. Localized-itinerant electronic transitions in oxides and sulfides[J]. J.Alloy.Compd., 1997, 262:  1-9.

  14. [14]

      Harrison M.R., Francesconi M.G. Mixed-metal one-dimensional sulfides-A class of materials with differences and similarities to oxides[J]. Coord.Chem.Rev., 2011, 255:  451-458. doi: 10.1016/j.ccr.2010.10.008

  15. [15]

      Z. Xia, J. Zhong, M. Leng, et al. , Generalized water-processed metal chalcogenide complexes: synthesis and applications, Chem. Mater. 27 (2015)8048-8057.

  16. [16]

      Yang B., Xue D.J., Leng M.. Hydrazine solution processed Sb₂S₃, Sb₂Se₃ and Sb₂(S_1-xSe_x)₃ film:molecular precursor identification, film fabrication and band gap tuning[J]. Sci.Rep., 2015, 5:  10978. doi: 10.1038/srep10978

  17. [17]

      Yang B., Wang L., Han J.. CuSbS₂ as a promising earth-abundant photovoltaic absorber material:a combined theoretical and experimental study[J]. Chem.Mater., 2014, 26:  3135-3143. doi: 10.1021/cm500516v

  18. [18]

      Janz G.J., Downey J.R., Roduner E.. Raman studies of sulfur-containing anions in inorganic polysulfides.Sodium polysulfides[J]. ChemInform., 1976, 7:  1755-1758.

  19. [19]

      Panthani M.G., Stolle C.J., Reid D.K.. CuInSe₂ quantum dot solar cells with high open-circuit voltage[J]. J.Phys.Chem.Lett., 2013, 4:  2030-2034. doi: 10.1021/jz4010015

  20. [20]

      Stolle C.J., Panthani M.G., Harvey T.B., Akhavan V.A., Korgel B.A. Comparison of the photovoltaic response of oleylamine and inorganic ligand-capped CuInSe₂ nanocrystals[J]. ACS.Appl.Mater.Inter., 2012, 4:  2757-2761. doi: 10.1021/am3003846

  21. [21]

      V. A. Akhavan, B. W. Goodfellow, M. G. Panthani, et al. , Colloidal CIGS and CZTS nanocrystals: A precursor route to printed photovoltaics, ChemInform 43 (2012)2-12.

  22. [22]

      Zhang H., Solomon L.V., Ha D.H.. (NH₄)₂S, a highly reactive molecular precursor for low temperature anion exchange reactions in nanoparticles[J]. Dalton Transaction., 2013, 42:  12596-12599. doi: 10.1039/c3dt50803j

  23. [23]

      Gao Y., Luo H., Zhang Z.. Nanoceramic VO₂ thermochromic smart glass:A review on progress in solution processing[J]. Nano.Energy, 2012, 1:  221-246. doi: 10.1016/j.nanoen.2011.12.002

  24. [24]

      Panda D., Tseng T.Y. Growth dielectric properties, and memory device applications of ZrO₂ thin films[J]. Thin.Solid.Films, 2013, 531:  1-20. doi: 10.1016/j.tsf.2013.01.004

  25. [25]

      Smincakova E., Raschman P. Leaching of natural stibnite using Na₂S and NaOH Solutions[J]. Int.J.Energy.Eng., 2011, 2:  85-89.

  26. [26]

      Johnson M., Baryshev S.V., Thimsen E.. Alkali-metal-enhanced grain growth in Cu₂ZnSnS₄ thin films[J]. Energy Environ.Sci., 2014, 7:  1931-1938. doi: 10.1039/C3EE44130J

  27. [27]

      Wagner C.D. Chemical shifts of auger lines, and the auger parameter[J]. Faraday. Discuss., 1975, 60:  291-300. doi: 10.1039/dc9756000291

  28. [28]

      Xue D.J., Yang B., Yuan Z.K.. CuSbSe₂ as a potential photovoltaic absorber material:studies from theory to experiment[J]. Adv.Energy Mater., 2015, 5:  1-9.

  29. [29]

      Zhou Y., Leng M., Xia Z.. Solution-processed antimony selenide heterojunction solar cells[J]. Adv.Energy Mater., 2014, 4:  1079-1098.

  30. [30]

      Wang G., Wang S., Cui Y., Pan D. A novel and versatile strategy to prepare metal-organic molecular precursor solutions and its application in Cu (In, Ga) (S, Se)₂ solar cells[J]. Chem.Mater., 2012, 24:  3993-3997. doi: 10.1021/cm3027303

  31. [31]

      Q. Guo, G. M. Ford, W. C. Yang, et al. , Fabrication of 7. 2% efficient CZTSSe solar cells using CZTS nanocrystals, J. Am. Chem. Soc. 132(2010)17384-17386.

  32. [32]

      Zhong J., Xia Z., Zhang C.. One-pot synthesis of self-stabilized aqueous nanoinks for Cu₂ZnSn (S, Se)₄ solar cells[J]. Chem.Mater., 2014, 26:  3573-3578. doi: 10.1021/cm501270j

  33. [33]

      Abaab M., Kanzari M., Rezig B., Brunel M. Structural and optical properties of sulfur-annealed CuInS 2 thin films[J]. Sol.Energy Mat.Sol.C, 1999, 59:  299-307. doi: 10.1016/S0927-0248(99)00043-4

  34. [34]

      E. A. Davis, N. F. Mott, Conduction in non-crystalline systems V. Conductivity, optical absorption and photoconductivity in amorphous semiconductors, Philos. Mag. 22(1970)903-922.

  35. [35]

      Ceperley D.M., Alder B.J. Ground state of the electron gas by a stochastic method[J]. Phys.Rev.Lett., 1980, 45:  566-569. doi: 10.1103/PhysRevLett.45.566

  36. [36]

      Perdew J.P., Zunger A. Self-interaction correction to density-functional approximations for many-electron systems[J]. Physical.Review.B:Condensed. Matter., 1981, 23:  5048-5079. doi: 10.1103/PhysRevB.23.5048

  37. [37]

      Poglitsch A., Weber D. Dynamic disorder in methylammoniumtrihalogenoplumbates (â…¡) observed by millimeter-wave spectroscopy[J]. J.Chem.Phys., 1987, 87:  6373-6378. doi: 10.1063/1.453467

  38. [38]

      Zhang H., Solomon L.V., Ha D.H.. (NH₄)₂S, a highly reactive molecular precursor for low temperature anion exchange reactions in nanoparticles[J]. Dalton.Trans., 2013, 35:  12596-12599.

  39. [39]

      Zhao Y., Nardes A.M., Zhu K. Solid-state mesostructured perovskite CH₃NH₃PbI₃ solar cells:charge transport, recombination, and diffusion length[J]. J.Phys.Chem.Lett., 2014, 5:  490-494. doi: 10.1021/jz500003v

  40. [40]

      T. Gershon, K. Sardashti, O. Gunawan, et al. , Photovoltaic device with over 5% efficiency based on an n-type Ag₃ZnSnSe₄ absorber, Adv. Energy Mater. 6(22) (2016)1-7.

  41. [41]

      J. Zhong, Z. Xia, M. Luo, et al. , Sulfurization induced surface constitution and its correlation to the performance of solution-processed Cu₃ZnSn (S, Se)₄ solar cells, Sci, Rep. 4(2014)6288.

• 

  Figure 1  Crystal structure of NaSbS₂: (a) view along a-axis. (b) Na⁺ sandwiched between anionic polymeric chains; (c) a single chain.

  下载: 全尺寸图片 幻灯片
  

  Figure 2  (a) Temperature-dependent XRD patterns of NaSbS₂ from 25 to 500 ℃, indicating the phase transition from cubic to monoclinic structure. (b) Temperature-dependent XRD of NaSbS₂ from 400 to 100 ℃. The bottom in panel a and b shows the standard diffraction pattern of NaSbS₂ (JCPDS 29-1169 and JCPDS 32-1039). Insets are the crystal structure of cubic (alpha phase) and monoclinic (beta phase) of NaSbS₂.

  下载: 全尺寸图片 幻灯片
  

  Figure 3  The top-view (a–f) and cross-sectional (g–i) SEM images of NaSbS₂ film on FTO after annealing with different temperatures of 230 ℃ (a), 300 ℃ (b), 350 ℃ (c), 400 ℃ (d & g), (e) 450 ℃ (e & h), (f) 500 ℃ (f & i).

  下载: 全尺寸图片 幻灯片
  

  Figure 4  (a) Schematic configuration of NaSbS₂ based photodetector device. (b) The current-voltage curve of FTO-NaSbS₂-FTO confirming that FTO form ohmic contacts with NaSbS₂ film. (c) Photosensitivity of NaSbS₂ deposited on FTO glass. (d) Fitting curve of photocurrent decay.

  下载: 全尺寸图片 幻灯片
  

  Figure 5  (a) Schematic configuration of photovoltaic device. (b) J-V curves of NaSbS₂ solar cell performance in the dark and under 100 mW cm^-2 simulated AM 1.5 G irradiation, respectively.

  下载: 全尺寸图片 幻灯片
•