Simple eco-friendly synthesis of the surfactant free SnS nanocrystal toward the photoelectrochemical cell application

A simple, low cost, non-toxic and eco-friendly pathway for synthesizing efficient sunlight-driven tin sulfide photocatalyst was studied. SnS nanocrystals were prepared by using mechanical method. The bulk SnS was obtained by evaporation of SnS nanocrystal solution. The synthesized samples were characterized by using XRD, SEM, TEM, UV-vis, and Raman analyses. Well crystallized SnS nanocrystals were verified and the electrochemical characterization was also performed under visible light irradiation. The SnS nanocrystals have shown remarkable photocurrent density of 7.6 mA cm−2 under 100 mW cm−2 which is about 10 times larger than that of the bulk SnS under notably stable operation conditions. Furthermore, the SnS nanocrystals presented higher stability than the bulk form. The IPCE(Incident photon to current conversion efficiency) of 9.3% at 420 nm was obtained for SnS nanocrystal photoanode which is strikingly higher than that of bulk SnS, 0.78%. This work suggests that the enhancement of reacting area by using SnS nanocrystal absorbers could give rise to the improvement of photoelectrochemical cell efficiency.

the device based on the TiO 2 or ZnO could harvest lesser than 5% of the sunlight. On the other hands, WO 3 14,15 , BiVO 4 16 and α-Fe 2 O 3 17,18 are very attractive materials which can serve as oxygen evolving anodes for absorbing energy from the ultraviolet light zone of solar spectrum because of their band gaps of 2.2, 2.45 and 2.6 eV, respectively. However, there are still some limitations on their performance, such as low absorbance in visible light zone, poor charge-carrier transport, poor collection of photogenerated electrons, and limited chemical stability in an electrolyte under illumination. Recently, tin (II) sulfide (SnS) has emerged as an attractive candidate for low cost catalytic materials of photovoltaic absorbers as well as photoelectrodes for solar water splitting because of their promising high electron (2.37 × 10 4 cm 2 V −1 s −1 ) and holes (7.35 × 10 4 cm 2 V −1 s −1 ) mobilities. Genernally, it is well knwon that the high carrier mobility of semiconductor is preferable for photovoltaic device applications since it prevents photogenerated excitons from recombination. Therefore, semiconductor with low carrier mobility greatly hinders its practical applications 19 . For this reason, high carrier mobility of SnS at the photoanode surface is preferable because it can promote fast photocatalytic process 20,21 . To take advantages of this kind of SnS nanocrystals, an ideal architecture of nanostructured films could be conceived, consists of SnS nanocrystals with high crystallinity and good electrical property on the substrate (Fig. 1).
SnS is known as a potential candidate for solar energy application due to its excellent optical and fascinating electrical properties 22,23 . The optical anisotropy of SnS showed an antireflection properties which results in high photoactivity. There is only a weak interaction in the 2D nature of SnS. This interaction allows easy separation and fabrication of layered composite structures. The layered composite structures create a wide range of Van der Waals heterostructures. Therefore, the processability of SnS makes it a perfect system in exploring new 2D feature 24 . Up to date, SnS nano sheets have been synthesized using various methods, for example: chemical bath deposition 25 , chemical vapor transport 26 , chemical vapor deposition (CVD) 27 , colloidal nanoparticle synthesis [28][29][30][31] , electrochemical deposition 32 , vacuum evaporation 33 , spray pyrolysis 34 , and dip deposition 35 . Among these SnS layers, the mono-layer nanostructured SnS photoanode has exhibited the high photocurrent density of 7 mA cm −2 , which was synthesized by chemical spray pyrolysis 36 . However, this method used thiourea, which is environmentally hazardous, and is a toxic material as well. SnS layer fabricated with nanocrystalized SnS could be a promising alternative to traditional solution-based methods, which does not require high vacuum and high process temperature 37 . However, conventional nanocrystals contain surface ligands which act as electrical insulator in PEC device. Therefore, surfactant free stable nanocrystals are highly desirable. Here, we present a mechanical alloying (MA) process which is a simple, low cost, and eco-friendly pathway to produce surface clean semiconductor nanocrystals of surfactant free tin sulfide (SnS). To the best of our knowledge, the photocurrents in this work are higher than the any of previously reported SnS systems.

Results and Discussion
Structural and Morphological Characterization. Powder X-ray diffraction (XRD) data was used to confirm the SnS phase purity. Figure 2a represents the XRD pattern of the SnS nanocrystals particles. Every peak in Fig. 2a was well indexed with values in the standard card 38 of the orthorhombic SnS phase (JCPDS No. 00039-0354), the structure with lattice constants (a 1⁄4 4.329 Å, b 1⁄4 11.190 Å, c 1⁄4 3.983 Å). In addition, the diffraction peaks were strong and sharp, which suggest that the SnS were well crystallized. And the result also confirmed the presence of highly pure SnS nanocrystals phase without any impurity.
The use of Raman spectroscopy to probe the detailed structure of materials is reviewed comprehensively. Figure 2b shows the Raman spectra of SnS nanocrystal. The Raman modes are observed at 70, 95, 160, 195, 219 cm −1 corresponding to one B 1g or B 2g mode, one A g , one B 3g (LO) and two A g (LO), modes respectively. They are in good agreement with what we can find from the literatures. It confirms the phase purity of SnS nanocrystal, while no impurities, such as SnS 2 (315 cm −1 ), Sn 2 S 3 (153 cm −1 ) and SnO 2 (472 cm −1 ) were observed.
We observed the thickness of SnS nanocrystal thin film about 180 nm in cross-section view of SnS nanocrystal thin film (Fig. S1). The morphology of SnS nanocrystals were shown in Fig. 2c. It can be observed that the SnS nanocrystals were distributed homogeneously. TEM analysis (Fig. 2d) roughly showed isolated SnS nanocrystals. Lattice fringes are displayed in the image, indicating that the SnS nanocrystals were well crystallized. The statistical results of size distribution for the SnS nanocrystals are presented in Fig. 2e. Columns in the Fig. 2e represents the statistical count ratio corresponding to the grain size. The grain size of the individual nanocrystals was found from TEM images. The summarized datum of the individual nanocrystal size could be used to calculate statistical size distribution. It can be roughly estimated that the grain size is between 1.5-2.5 nm, suggesting that the SnS nanocrystals are homogeneous.
Since a true optical band gap is an indication of a pure phase of a material, we performed UV-Vis absorption measurements using an integrating sphere and determined the band gap of our synthesized SnS nanocrystals. The results are shown in Fig. 3. The spectrum covered almost the entire visible range, suggesting SnS nanocrystals can absorb energy from the ultraviolet light zone of solar spectrum. A direct band gap value of 1.20 eV was characterized from the extrapolated intercept with the energy (hv) axis, which is well in accordance with literature values and with that of standard commercial bulk SnS as well. This indicates that our synthesized SnS nanocrystals are in pure crystal phase. However, we could not observe the possible blue shift in the band gap due to the larger size of the particles.  The SnS nanocrystals Photoanode Performance. The thickness dependence of the SnS films on the catalytic activity was then investigated (Fig. S2). We first describe the results obtained from the SnS nanocrystal samples. Low current density was observed for the thin (50 ± 20 nm) films while thicker films (100 ± 30, 200 ± 50 nm) displayed better catalytic currents as their thickness increased. However, the current density decreased when the films thickness reached 300 ± 100 nm. Meanwhile, the films with thicknesses of 200 nm have shown good stability during electrolysis. The similar behavior was observed for the bulk sample.
The linear sweep voltammogram spectrum of PEC cell recorded at 50 mV s −1 in 0.1 M K 4 Fe(CN) 6 and 0.01 M K 3 Fe(CN) 6 with chopped visible light is shown in Fig. 4. A cathodic photocurrent could be possible in the n-type semiconductor because of the accumulation of majority carriers on the electrolyte side ascribed to the part below the flat band. The above part as the band starts to align in response to the applied potential is the depletion part. The cell onset potential (V on ) was found to be 0.22 V versus RHE which was 90 mV more negative than the bulk one. The lower photovoltage may be attributed to the difference in the density of photoanode material, which changes the hole injection barrier. A photocurrent is about 7.6 mA cm −2 at 1.05 V versus RHE and it reaches 8.2 mA cm −2 at more cathodic potentials. The dark currents of the SnS nanocrystals and the bulk SnS (Fig. 4b) remained at a level of less than 0.8 mA cm −2 between 0.4 V and 1.1 V versus RHE. In contrast to the visible-light irradiation, the SnS nanocrystals presented a much enhanced photocurrent density of 7.6 mA cm −2 at 1.05 V versus RHE, approximately, 9.3 times larger than that of the bulk material. Increased reaction area with water due to the larger surface area of SnS nanocrystal layer compared to the bulk SnS layer could be the reason of this incredibly improved photocurrent density. It can be noticed that at −0.2 V vs. RHE the spikes flatten out, suggesting that the kinetic barriers have been solved 39 . High photocurrent of the cells is obtained that is comparable with other previously reported SnS systems 30,32,36,37,[40][41][42][43][44][45][46] in the literature listed in Table 1.
In a number of reports in Table 1 present a smart strategy to increase the short circuit current for the further improvement of the photoelectrochemical cell. Ray's group obtained short-circuit current density of 0.3 mA cm −2 by optimizing molar concentration ratio of S 2− /Sn 2+30 . They found 1.2 mA cm −2 of short-circuit current with device based on Cu doped sprayed SnS thin films 36 . After then, nanostructured SnS photoanodes were adopted and the device exhibited a 0.42 mA cm −2 of short-circuit current 40 . On the other hands, Xi's group also reported high photocurrent density 3 mA cm −2 by using SnS atomic layer 37 . The cells fabricated in this work have shown a short circuit current density of 0.45 mA cm −2 which means higher short-circuit current density is in urgent need for the improvements of mass production and practical applications. And applying atomic layer material in 0.5 M Na 2 SO 4 would be a good way to improve short-circuit current density.
The maximum theoretical photocurrent density (J ph, max ) for the material with band gap of 1.2 eV was estimated to be 36.7 mA cm −2 by calculating from eq. 1 51 .
Therefore, the saturation current density, J ph (7.6 mA cm −2 ) in our case reaches only about 20% of the theoretical maximum photocurrent J ph,max (36.7 mA cm −2 ). In other words, it still remains about 85% of possible increase in the photocurrent density that can be achieved by developing pure SnS. One of the origins of this loss could be the nature of surfactant free SnS nanocrystal. Even though surfactant free SnS nanocrystal could provide better conductivity in the device, reaction kinetics at the surface of the SnS nanocrystals could be sluggish due to the surface trap state at the surface. The photocurrent spikes observed in the Fig. 5 indicate the presence of theses surface traps 36 . The reproducibility of the photocurrent response can be switched from the "ON" state to the "OFF" state by chopped visible light. Stability measurements were then carried out with chopped light at 1.23 V versus RHE. The results are shown in Fig. 5.
It takes around 0.5 s for response and decay in Fig. 5a, suggesting rapid response characteristics. A photocurrent density of 7.5 mA cm −2 was decreased by only 24% after 50 min. At the same time, the low dark current ensured the excellent stability of the cell. It is worth noting that the photocurrent densities of the SnS nanocrystals showed high stability even after 3,000 s of irradiation (24% decrease), while the bulk material displayed 87% decrease after 900 s (Fig. 5b). This is a clear evidence for the enhanced stability of the SnS nanocrystals.
SEM images taken before and after the chronoamperometric measurements of nanocrystal SnS are shown in Fig. S1. After the measurements, SEM showed that pieces of SnS have been detached from the surface, probably because of the H 2 bubbles pulling the catalyst. Figure 6 described the position of the conduction and valence bands of the SnS nanocrystals edges and the redox levels of the electrolyte. The photogenerated holes were extracted from the band edge of the SnS nanocrystals to the redox levels of the electrolyte. The electrolyte accepted the photo-generated holes from the SnS nanocrystals valence band, and then the oxygen produced at the photoanode surface. Meanwhile, the Pt-counter electrode received photogenerated electrons from the conduction band through the external wire and the hydrogen came out at the Pt surface. The visual phenomena of the oxygen and hydrogen evolution reactions are displayed in Fig. 6b.
To quantitatively evaluate its photo-conversion efficiency, incident photon-to-current conversion efficiency (IPCE) measurements were carried out and the results are shown in Fig. S3. The SnS nanocrystal photoanode possessed an IPCE of 9.3% at 420 nm, strikingly higher than the 0.78% efficiency of bulk sample.
The ratio metric power-saved figure-of-merit Ф saved, NPAC (NPAC = non-photoactive, identical catalyst) (eq. 2) can estimate the ability of a photoanode to achieve H 2 evolution. Ф saved, NPAC is obtained at the maximum power, it is less catalyst-dependent 51 .
(2)  Table 1. Photocurrent density of SnS photoelectrodes in different systems. Where the faradaic efficacy (η F ) assume to be 100%, P in represents the power of the incident illumination, and J photo,m and V photo,m stand for the photocurrent and photovoltage at the maximum power point respectively. Character "m" represents maximum. J photo is the photocurrent density which is obtained by the current density under illumination (J light ) minus the current density of the corresponding catalyst (J dark ). The photovoltage V photo is given by the difference between the potential applied to the photoanode under illumination (V light ) and the potential applied to the catalyst (V dark ). As expected, the Φ saved,NPAC values of the SnS nanocrystal photoanode (5%) were significantly higher than that of the bulk SnS system (0.158%).

Conclusion
In conclusion, the SnS nanocrystals as a H 2 -evolving catalyst yields a novel nanostructure type of photoelectrode in water splitting. This system is based on earth-abundant elements and can be easily processed using low cost and low temperature spray-casting method. A simple, low cost, non-toxic and eco-friendly pathway was used to synthesize sunlight-driven tin sulfide photocatalyst. The SnS nanocrystals were well crystallized and their grain size is between 1.5-2.5 nm. The SnS nanocrystals exhibited a direct optical band gap of 1.20 eV. The linear sweep voltammogram showed that the SnS nanocrystals presented photocurrent density of 7.6 mA cm −2 which is dramatically larger than that of bulk SnS and is higher than the any of previously reported SnS systems as well. The stability experiment confirmed that the SnS nanocrystals were more stable than the bulk SnS. The SnS nanocrystal photoanode possessed an IPCE of 9.3% at 420 nm, strikingly higher than the 0.78% efficiency of bulk sample. Moreover, figure-of-merit, Ф saved, NPAC was evaluated and discussed. Their Ф saved, NPAC values of the SnS nanocrystals and the bulk SnS reached 5% and 0.158%, respectively. Based on low cost, low fabrication temperature and other various advantages, this work suggests that the SnS nanocrystal absorbers hold great promise for bringing a wealth of eco-friendly environment.

Experimental Section
Measurement characterization. The XRD of SnS nanocrystals were performed by D8 Advance X-ray diffractometer (Rigaku Dmax-RB with Cu Kα X-ray source, Germany). UV-vis absorption spectra were characterized by using a Lambda 750 S UV-vis-NIR spectrophotometer. The morphology of the SnS nanocrystals were recorded with a field-emission transmission electron microscope (TEM, JEM-2100F, Japan) operated at 200 kV. Electrochemical measurements were investigated using a CHI 660E electrochemical workstation (Chenhua, Shanghai, China). A visible-light source performed by a 350 W xenon lamp equipped with a UV-cut-off filter (providing visible-light with >420 nm). A power meter (model FZ-A) used to detected the incident light intensity (~100 mW cm −2 ). IPCE spectra were operated on a QE/IPCE Measurement Kit (Newport, USA). All the kit components automatically were in control of Oriel Tracq Basic V5.0 software. A 300 W Xe lamp was used as the visible-light source, through a monochromator (74125 Oriel Cornerstone 260 1/4 m) onto the cell, and the monochrometer generated a photocurrent action spectrum with a sampling interval of 10 nm through the spectral range (420-920 nm) and a current sampling time of 2 s, where the light intensity and the generated photocurrent were detected with a 2931-C dual channel power/current meter and a 71675 calibrated UV silicon photodetector. IPCE can be calculated concretely as follows: where h is the Planck's constant, c is the speed of light, I is the measured photocurrent density at a specific wavelength, λ is the incident light wavelength, and J light is the recorded irradiance intensity at a specific wavelength.

Synthetic Methodology
Synthesis of the SnS nanocrystals. In a typical procedure, the 15 g mixtures of high purity tin and sulfur powders (molar ratio of 1:1) as the starting materials were sealed with stainless steel balls (2-12 mm) in a stainless steel jar. The mass ratio of ball and powder turn to 10:1. The jar was then loaded on a SFM-3 Mixer/mill machine to start mechanical alloying process. After 30 hours, the synthesis done, the particles were taken out the jar.

Synthesis of the bulk SnS.
A typical dissolution experiment as follow: 325 mg the SnS nanocrystals powder into a three-neck round-bottom flask. Distilled en (5 mL) and edt (0.45 mL) were then added. The mixture was lightly heated under stirring and was sonicated for a 10 min. A heating mixture was then used to further heat the mixture to 50 °C for 15 h with stirring.