Tin disulfide (SnS 2 ) has been synthesized and studied by several researchers because of its potential suitability for applications
the basal planes of SnS 2 are perpendicular to the current col lector, photoexcited charges can be transported more efficiently because conductivity along the SnS 2 basal planes is ≈10 4 times that across them. [16] One method of achieving perpendicular orientation is by growing SnS 2 in the form of vertical nano flakes on the current collector. In addition to orientation, the height and packing density of nanoflakes need to be optimized to ensure high optical absorption and efficient charge transport. Finally, not only optimized morphology, but also high crystal linity and phase purity are needed to realize high PEC perfor mance from SnS 2 photoelectrodes.
Several methods of synthesizing vertically oriented SnS 2 nanoflakes on conductive substrates have been reported in the literature such as hydrothermal reaction, [11] chemical vapor deposition (CVD), [28,30] and close space sublimation (CSS). [20,31] However, only a few of these methods have been used to opti mize the nanoflakes for photoelectrochemical applications. In one case, [28] vertical SnS 2 nanoflakes were synthesized on a fluorinedoped tin oxide (FTO) current collector using CVD, and a photocurrent of ≈1.5 mA cm -2 was obtained at the ther modynamic potential for water oxidation (1.23 V RHE ) in 0.5 m Na 2 SO 4 . In spite of having strong light absorption, these SnS 2 nanoflakes yielded photocurrents much lower than the theoreti cally possible ≈11 mA cm -2 for 2.2 eV bandgap material. [32] In order to further understand and improve the performance of SnS 2 photoanodes, we synthesized vertical SnS 2 nanoflakes on FTO using CSS, and studied the dependence of photocurrent on composition and morphology.
In the present work, vertical SnS 2 nanoflakes were grown directly on FTO substrates (typical area 2 cm × 2.5 cm) using a custombuilt CSS system with moderate vacuum. The experi mental parameters were tuned to minimize phase impurities and maximize photocurrent in phosphate buffer with sodium sulfite (Na 2 SO 3 ) hole scavenger. The optimized samples demon strate excellent crystallinity and do not require postannealing. Furthermore, when measured as a photoanode in phosphate buffer with Na 2 SO 3 hole scavenger, they exhibit a photocurrent as high as 4.5 mA cm -2 at 1.23 V RHE under simulated AM1.5G illumination, which is the largest photocurrent reported for a SnS 2 photoanode under these conditions. We relate the high photocurrent of these SnS 2 photoanodes to vertical orienta tion, optimized height, stepped morphology, and high intrinsic electron mobility and photoexcited carrier lifetime. This work illustrates CSS as a facile method to obtain highquality ver tical SnS 2 nanoflakes, which are promising for PEC and other applications.
A custombuilt CSS system (Figure S1, Supporting Informa tion) was used to grow vertical SnS 2 nanoflakes directly on FTOcoated glass substrates. The system was designed to allow independent control of source temperature (T src ), substrate temperature (T sub ), and source-substrate separation (d ss ). Con trolled experiments were carried out to determine the optimal growth conditions (T src , T sub , d ss , and growth duration). Xray diffraction (XRD), scanning electron microscopy (SEM), and PEC measurements with Na 2 SO 3 hole scavenger (Figure S2, Supporting Information) were carried out to characterize the purity, morphology, and PEC performance of the nanoflakes at each growth condition. T src was fixed at 525 °C because lower temperatures (<450 °C) did not cause sublimation and higher temperatures (>600 °C) led to the deposition of black Sn 2 S 3 particles. The growth duration was fixed at 30 min because longer durations resulted in only small increases in SnS 2 nano flake height and performance due to conversion of the source material to SnS (Figures S3 and S4, Supporting Information). Increasing T sub from 424 to 453 °C caused the growth of taller and wider SnS 2 nanoflakes with higher PEC performance, but further increasing T sub to 471 °C caused the formation of Sn 2 S 3 nanorods, which produced extremely low photocur rent (Figure 1a-c; Figure S5, Supporting Information). Then, keeping T sub fixed at 450 °C, and decreasing d ss from 19 to 10 mm resulted in taller nanoflakes with increased opacity (Figure 1d-f; Figure S6, Supporting Information) as expected because the concentration of the SnS 2 vapor is higher closer to the source. The maximum photocurrent of 4.5 mA cm -2 was obtained from the tallest nanoflakes (≈1.5 µm height) grown with d ss = 10 mm. Positioning a boat containing sulfur powder (at ≈200 °C) close to the source boat containing SnS 2 powder during the growth suppressed the conversion of the SnS 2 source to SnS and resulted in increased sublimation and a further increase of nanoflake height to ≈2.3 µm. How ever, these nanoflakes produced much lower photocurrents (Figure S7, Supporting Information), thus placing the optimal nanoflake height in the 1.5-2.3 µm range, but likely closer to 1.5 µm.
Based on the preceding optimization experiments, the optimal nanoflake height was ≈1.5 µm, which could be obtained with T sub = 450 °C, t = 30 min, and d ss = 10 mm. Several SnS 2 nano flake photoanodes were prepared under these experimental conditions, and their PEC performance was measured by linear sweep voltammetry (LSV) in different electrolytes for com parison (Figure 2a). The same SnS 2 nanoflakes that produced a maximum photocurrent of 4.5 mA cm -2 at 1.23 V RHE under AM1.5G illumination in phosphate buffer + 1 m Na 2 SO 3 pro duced 2.6 mA cm -2 in 0.5 m Na 2 SO 4 (i.e., for water oxidation) without the Na 2 SO 3 hole scavenger. The aqueous solution of Na 2 SO 4 is commonly used to characterize water oxidation per formance of photoanodes because SO 4 2-ions require a high potential of 2.01 V RHE to be oxidized to S 2 O 8 2-while water can be oxidized to oxygen at 1.23 V RHE . [33] The photocurrent of 2.6 mA cm -2 obtained from our nanoflakes is about 1.7 times the photocurrent reported previously from CVDgrown vertical SnS 2 nanoflakes at 1.23 V RHE in 0.5 m Na 2 SO 4 under similar light illumination, and represents a new record for photocur rent, to the best of our knowledge. [28] However, since the photocurrent produced by the SnS 2 nanoflakes decreased rapidly within a few minutes in Na 2 SO 4 solution (also discussed in Section 2.4 and in Figure S12c in the Supporting Information), it is not meaningful to report an efficiency for photoelectrochemical water oxidation. Instead, an electrolyte containing 0.1 m KI and 1 m H 2 SO 4 , in which SnS 2 nanoflakes are very stable and produce a photocurrent of 2.7 mA cm -2 at 0.6 V RHE , was used to further characterize the performance and measure the wavelengthdependent inci dent photontocurrent efficiency (IPCE, also known as external quantum efficiency) for photoelectrochemical iodide oxidation. The excellent ability of the optimized SnS 2 nanoflakes to absorb and generate photocurrent is evident in the IPCE (Figure 2b) measured under both frontside (electrolyteside) and back side (glassside) illumination in 0.1 m KI + 1 m H 2 SO 4 at a fixed voltage of 0.6 V RHE . The layout of the front and back illumina tion is further illustrated in the schematic in Figure S8a (Sup porting Information). As shown in Figure 2b, the IPCE with front illumination reaches almost 100% at shorter wavelengths and that with back illumination peaks around 62% at 440 nm. Both front and back IPCE plots have an onset near 580 nm. At shorter wavelengths, IPCE with back illumination is lower than that with front illumination because the light is absorbed or scattered by the glass/FTO substrate, whereas this difference is not significant at longer wavelengths. We also validated the consistency of our LSV and IPCE measurements by integrating the IPCE over the measured spectral irradiance of the incident light from the Xe lamp solar simulator used during the PEC tests (Figure S8b, Supporting Information). This resulted in a photocurrent of 3.0 mA cm -2 for back illumination, which is very similar to the photocurrent of 2.7 mA cm -2 measured at 0.6 V RHE by LSV.
In order to investigate the source of high photocurrent and the reason behind the existence of an optimum height of the SnS 2 nanoflakes, we further analyzed the structural and opto electronic properties of the optimized nanoflakes. The high resolution transmission electron microscope (TEM) image of a nanoflake in Figure 3a along with the corresponding selected Adv. Energy Mater. 2019, 1901236 0 1 2 3 4 5 0.2 0 .7 1.2 ( y t i s n e D t n e r r u C m c A m -2 ) Voltage (V) vs RHE 0 20 40 60 80 100 300 4 00 500 6 00 IPCE (%) Wavelength (nm) front illumination back illumination b a Na 2 SO 3 KI + H 2 SO 4 Na 2 SO 4 Figure 2. a) Linear sweep voltammetry of optimized SnS 2 nanoflakes in 0.5 m Na 2 SO 4 , 1 m phosphate buffer + 1 m Na 2 SO 3 , and 0.1 m KI + 1 m H 2 SO 4 with back illumination. b) Incident photon-to-current conversion efficiency (also known as external quantum efficiency) of vertical SnS 2 nanoflakes measured with front and back illumination in 0.1 m KI + 1 m H 2 SO 4 at 0.6 V RHE .
area electron diffraction (SAED) pattern (inset) shows that the SnS 2 nanoflake is crystalline and possesses hexagonal crystal structure. The measured inplane dspacing of 3.16 Å is con sistent with that of the (100) planes of 2HSnS 2 (ICDD# 000230677). It is also evident from the SAED pattern with [001] zone axis that the flat surface of the SnS 2 nanoflakes is formed by the (001) basal planes of SnS 2 . In addition, although the SAED pattern is dominated by bright spots originating from a SnS 2 single crystal nanoflake, the faintly visible ring like pattern indicates a partially polycrystalline character of the nanoflakes. Furthermore, the lowmagnification TEM image in Figure 3b shows that the nanoflakes have surfaces formed by incom plete basal planes that contain small holes/patches (≈5 nm), which are stacked on top of fully formed single crystalline basal planes. The holes/patches could be the result of nonuniform growth of new basal planes on existing nanoflake surfaces lim ited by island nucleation and lateral motion of steps at a low supersaturation of SnS 2 vapor, as predicted by crystal nuclea tion theory, and as observed in the CVD growth of 2D mono layers. [34][35][36][37] However, such patches were not reported in the CVDgrown vertical SnS 2 nanoflakes, which could be due to higher supersaturation of SnS 2 vapor in their experiments. [6,28] An even lower magnification TEM image in Figure 3c shows the presence of discrete steps in our nanoflakes. We surmise that both the nanoscale holes/patches (Figure 3a,b) and the microscale steps (Figure 3c) on the surface of the nanoflakes provide additional edges that can act as electrochemically active sites and boost the rate of surface charge transfer, leading to enhanced photocurrent. The edges in MoS 2 electrocatalysts are reported to have a similar role. [38,39] While the edges of these steps, holes, and patches can also act as carriertrapping centers, the enhancement in surface charge transfer due to increased density of electrochemically active sites can outweigh the loss due to trapmediated recombination. The role of these edges on the PEC performance will be addressed experimentally and dis cussed in the latter sections.
The XRD pattern of the optimized nanoflakes on FTO (Figure 3d) shows the most dominant peaks from (001) and (100) crystal planes, and matches well with 2HSnS 2 standard. Similarly, the roomtemperature Raman characterization (Figure 3e) also suggests that our optimized nanoflakes are of high quality. A narrow band is observed at 313.5 cm -1 , con sistent with the outofplane (A 1g ) vibrational mode of 2HSnS 2 measured by several groups. [22,31,40,41] According to a previous report, [40] 2HSnS 2 also exhibits a single band at 205.5 cm -1 , unlike 4HSnS 2 or 18RSnS 2 that exhibit multiple bands near that wavenumber. Hence, a single band around 204.5 cm -1 shown in the 50 times enlarged spectrum in Figure 3e can be assigned to inplane (E g ) vibrational mode of 2HSnS 2 .
The optimized nanoflakes exhibit high lightharvesting effi ciency (LHE), which is essential for yielding high photocurrent.
The LHE of our optimized nanoflakes was measured using an integrating sphere, as described in our previous report. [42] As shown in Figure 4a, an LHE of about 90% was achieved at shorter wavelengths with absorption onset at around 580 nm. The Tauc plot in Figure 4b shows the indirect bandgap of SnS 2 to be about 2.1 eV, which is in agreement with other reports in the literature. [18][19][20]41] This indirect bandgap is also evident from the broad photoluminescence (PL) peak centered at 588 nm (2.11 eV) in Figure 4c.
Another important factor that impacts the performance of photoanodes is diffusion length (L D ) of the photoexcited charge carriers, which can be calculated from carrier lifetime (τ) and mobility (µ) using Equation (1), where k B is the Boltzmann con stant, T is absolute temperature, and e is elementary charge
The radiative charge carrier lifetime in our optimized SnS 2 nanoflakes was determined using timeresolved photolumi nescence (TRPL) decay of the 2.11 eV peak, corresponding to the bandgap emission, after exciting with 405 nm laser at near 1 sun peak excitation intensity (≈128 mW cm -2 ), as shown in Figure 4d. Immediately after the excitation, a fast response with a lifetime of 78 ps is observed, which is attributed to the response of the measurement system. After fitting a double exponential, the radiative carrier lifetime in SnS 2 nanoflakes is estimated to be ≈1.3 ns.
Similarly, a contactfree timeresolved terahertz spectroscopy (TRTS) technique was used to study the dynamics of photo excited free charge carriers on picosecond time scales, and estimate intrinsic mobility of SnS 2 nanoflakes. [43][44][45][46][47][48] CSSgrown SnS 2 vertical nanoflakes were optically excited with 400 nm laser pulses with 100 fs duration and with 28 µJ cm -2 fluence, and transient photoconductivity (Figure 5a) was measured using a terahertz probe in the 0.25-1.70 THz frequency range, which corresponds to 1-7 meV energy range. With energy in the meV range, terahertz pulses are absorbed by the free car riers, and transient change in transmission of the terahertz probe pulse peak is a measure of photoconductivity (Figure 5a). For comparison, chemical vapor transport (CVT)grown SnS 2 single crystals were also measured in the same manner. Photo conductivity in both single crystal SnS 2 and vertical SnS 2 nano flakes exhibits a multiexponential decay. The fastest component (2-3 ps) is more pronounced at higher excitation fluence values (Figure S9, Supporting Information), and can be thus ascribed to carrier-carrier scattering. The slower 10-15 ps decay likely represents carrier trapping at defect and edge states. Finally, a much slower component that we attribute to free carrier recom bination decays over >250 ps. The corresponding free carrier lifetime is difficult to determine accurately from TRTS meas urement as it is limited by the pump-probe delay that is avail able to us. However, while ≈250 ps is the lower limit of the photoexcited carrier lifetime, the radiative lifetime of ≈1.3 ns determined by TRPL is the upper limit. Thus, we can place the lifetime of the photoexcited carriers in the optimized ver tical nanoflake array in (0.25-1.30 ns) range. The lower end of this range is similar to the radiative lifetime of 0.25 ns meas ured for exfoliated single crystals, [22] while the higher end of the range is closer to the values of 2.4 and 3.0 ns measured for CVDgrown thin (20 nm) and thick (32 nm) SnS 2 nanoflakes, respectively. [41] Adv. Energy Mater. 2019, 1901236 500 600 700 8 00 ) . u . a ( y t i s n e t n i L P Wavelength (nm) 10 60 110 160 210 260 310 1.7 2 .2 2.7 (hαν) 0.5 (eV/cm) 0.5 hν (eV) -1 0 1 2 3 4 5 6 7 8 PL intensity (a.u.) time (ns)
In addition to providing insights into free carrier lifetimes after photoexcitation, TRTS allows us to measure intrinsic mobility. Photoexcitationinduced changes in the amplitude and the phase of the THz pulse waveform transmitted through the sample [43,[49][50][51] were used to extract complex photoinduced change in conductivity 20 ps after excitation with 28 µJ cm -2 , 400 nm pulses. At 20 ps, carrier-carrier scattering is no longer appreciable, and thus intrinsic carrier scattering time and mobility can be determined. Real (Δσ 1 ) and imaginary (Δσ 2 ) photoconductivity components are plotted in Figure 5b,c for single crystals and vertical nanoflakes, respectively, as a func tion of THz frequency (ω). In general, the complex conductivity (σ ˆ) of both of these samples can be described by Equation ( 2), (also known as the Drude-Smith model), where N is the charge carrier density, m* is the effective carrier mass, τ DS is the effective scattering time, and the cparameter character izes the degree of carrier localization due to the presence of boundaries [50,[52][53][54][55][56][57][58][59][60] σ ω τ ωτ ωτ
The cparameter and τ DS were first determined by fitting the measured frequencyresolved complex conductivity to this model. Then, intrinsic and longrange mobilities of carriers in our samples were estimated using Equations ( 3) and ( 4), respectively, where a previously reported value of effective mass m* = 0.375m e was used [61] µ τ = e m * intrinsic D S (3) µ τ ( ) = + -e m c * 1 long range D S
We find that for single crystal SnS 2 , Equation (2) reduces to the Drude model (with c = 0) and the mobility is estimated to be 800 cm 2 V -1 s -1 , which is similar to a theo retically predicted value of electron mobility in monolayer SnS 2 . [23] However, for ver tical nanoflakes, the localization parameter is nonzero (c = -0.72) due to confinement of carriers within individual nanoflakes, and the intrinsic and longrange mobilities are estimated to be 330 and 90 cm 2 V -1 s -1 , respectively. This intrinsic mobility is slightly higher than that reported in previous studies and approaches the theoretically predicted one, which is a testament to the high crys tallinity of the nanoflakes and low defect concentration. [22,23,25,62] Since the impor tant charge transport in PEC applications occurs within individual nanoflakes, the high intrinsic mobility can improve the overall photocurrent in SnS 2 photoanodes.
Our finding of an optimum height of SnS 2 nanoflakes is expected to be due to a balance between two competing requirements-the generation of photoexcited charges by optical absorption and the diffusion of those photoexcited charges. In terms of optical absorption, Figure S10 (Supporting Informa tion) shows that ≈2 µm thick SnS 2 is required to absorb most of the photon energies. Similarly, in terms of photoexcited car rier diffusion, combining the carrier lifetime (0.25-1.30 ns) and TRTS mobility using Equation (1) yields the diffusion length of ≈(0.4-1.0 µm) in our nanoflakes. As expected, the optimized nanoflake height of ≈1.5 µm falls in between the absorption depth and the diffusion length.
Optimized height is likely not the only reason for the enhanced photocurrent in our SnS 2 nanoflakes compared to the previously reported CVDgrown SnS 2 nanoflakes, which also had high packing density and heights in the range of few micrometers, yet achieved significantly lower photo current. [28] Upon closer examination of the nanostructure of our SnS 2 nanoflakes using highmagnification SEM (Figure 6a), we observe that our nanoflakes consist of discrete steps on the sides (which were also observed by TEM in Figure 3c). We believe that the edges are the sites of the electrochemical reac tion, and that the photocurrent could be improved by spreading these reaction sites over the entire nanoflake, similar to the case of stepped MoS 2 . [63] These steps in CSSgrown SnS 2 nano flakes are formed either due to nucleation and growth of new basal planes onto the sides of the existing basal planes at high vapor concentrations or due to interaction of neighboring nano flakes during the prolonged growth. The steps on the side of the nanoflakes are also visible in the atomic force microscopy (AFM) image in Figure 6b, which was taken from a horizontally laid piece of a nanoflake obtained by separating the asgrown SnS 2 nanoflakes from the growth substrate by sonicating in ethanol, followed by dropcasting onto a Si/SiO 2 substrate. The height profile in Figure 6c shows that the height of the dis crete steps is about 4-5 nm, while the overall thickness of the nanoflake is about 30 nm, which is similar to the average nano flake thickness measured using SEM (Figure S11, Supporting Information).
In order to support our claim that CSSgrown SnS 2 nano flakes contain multiple edge sites which are electrochemi cally active, we deposited Cu particles by electrochemical reduction and PbO x particles by photoelectrochemical oxida tion. As shown in Figure 6d (Cu) and Figure 6e (PbO x ), the nanoparticles are preferentially deposited on the top edges of the nanoflakes as well as at the steps on the sides of the nanoflakes, indicating that multiple edge sites are avail able in every nanoflake for the redox reaction. This not only demon strates the presence of these steps but also that the steps and edges in SnS 2 are electrochemically more active than the other regions, which has been previously reported for MoS 2 . [63] The selective deposition of nanoparticles has also been used by many other researchers to visualize and iden tify active sites. [64,65] Since each of our optimized SnS 2 nano flakes contains additional active sites on their faces due to the presence of steps, the overall photo current can be expected to be higher than CVDgrown nanoflakes that lack the stepped structure.
SnS 2 is known to be stable in ambient conditions and is free of surface oxides. However, it has relatively poor sta bility in aqueous solutions. [12,17] This is also evident in the chronoamperometric (CA) tests of our optimized SnS 2 nano flake photoanodes (Figure S12a,b, Supporting Information). Photocurrents from these photoanodes decrease significantly during the 10 min test in 1 m phosphate buffer, 1 m phosphate buffer + 1 m Na 2 SO 3, and 1 m H 2 SO 4 at a constant applied potential of 0.6 V RHE. The photoanodes are also unstable in 0.5 m Na 2 SO 4 , as shown in CA tests performed at 1.23 V RHE in Figure S12c (Supporting Information).
Controlled stability tests performed on CVTgrown single crystal SnS 2 (Figure S13a-e, Supporting Information) suggest that the edges of SnS 2 are not only the active sites responsible for improving photocurrent but also the main sites of degra dation. As shown in Figure S13a-c (Supporting Information), while the pristine single crystal SnS 2 electrode (with minimal edges) shows no signs of degradation, the same SnS 2 trans ferred to indium tin oxide (ITO) (with significantly more edges formed during mechanical exfoliation) shows significant loss of performance over time, implying that SnS 2 edges are respon sible for degradation. This result is similar to reports on MoS 2 , for which it has been proven that edges are preferentially oxi dized in water. [66] Due to the difference in area coverage of our single crystal and exfoliated SnS 2 samples, the values of the photocurrents could not be compared directly. Hence, another single crystal SnS 2 electrode was tested in KI + H 2 SO 4 (in which SnS 2 was shown to be stable) with and without mild abrasion as shown in Figure S13d,e
single crystals used here are not the same as our CSSgrown SnS 2 nanoflakes, the nature of the edges can be expected to be similar.
In order to investigate the modes of degradation in our SnS 2 nanoflakes, Xray photoelectron spectroscopy (XPS) was carried out before and after PEC tests in phosphate buffer and sulfuric acid without any sacrificial agents. Com pared to the asgrown SnS 2 nanoflakes (Figure 7a), those tested in sulfuric acid show significant anodic dissolution (Figure 7b), evident from SEM images and XPS peak for elemental sulfur, which is one of the products of the disso lution reaction. [17] This observation is further corroborated by inductively coupled plasma optical emission spectrom etry (ICPOES) analysis (Table S1, Supporting Information) that shows high concentration of Sn dissolved in the elec trolyte. Similarly, the nanoflakes tested in phosphate buffer undergo surface oxidation (Figure 7c), which seems to slow anodic dissolution, as the nanoflakes appear more intact in the SEM images. All of these findings corroborate oxida tion and dissolution of SnS 2 photoanodes observed by other researchers. [12,17] In contrast, the SnS 2 nanoflakes showed substantially improved stability in the presence of iodide ions in the acidic electrolyte, which is consistent with a previous report. [17] During both CA and LSV tests in 1 m H 2 SO 4 + 0.1 m KI (Figure S14, Supporting Information), our optimized SnS 2 nanoflakes showed no signs of degradation. Furthermore, the maximum stable photocurrent of 2.7 mA cm -2 was obtained at 0.6 V RHE . In order to test longerterm stability in KI+H 2 SO 4 , a nonoptimized SnS 2 nanoflake sample was used. As shown in Figure S15a-c (Supporting Information), the SnS 2 nanoflakes maintain their photocurrent beyond 2 h (after which the test was terminated), and their nanostructure is not affected by the test. This suggests that both dissolution and oxidation of SnS 2 are effectively suppressed by the fast kinetics of the iodide oxidation reaction. Due to relatively high photocurrent and photoelectrochemical stability, these SnS 2 nanoflakes can be utilized directly in applications such as iodine evolution, and dyesensitized solar cells with iodide-triiodide redox couples. As for the photoelectrochemical water oxidation, suitable strategies need to be developed to protect the edges during prolonged use.
Adv. Energy Mater. 2019, 1901236 1901236 (9 of 11)
In summary, we have optimized the growth of vertically aligned SnS 2 nanoflakes directly on FTO substrates using a custom designed close space sublimation system, and obtained nano flakes with excellent optoelectronic and photoelectrochemical properties. The optimized SnS 2 nanoflakes exhibit high intrinsic carrier mobility (330 cm 2 V -1 s -1 ) and long photo excited carrier lifetimes (1.3 ns), which result in large diffusion lengths of ≈1 µm. Photoanodes made of these nanoflakes yield photocur rents as high as 4.5 mA cm -2 in aqueous phosphate buffer with 1 m Na 2 SO 3 and 2.6 mA cm -2 in aqueous 0.5 m Na 2 SO 4 , both at 1.23 V RHE under simulated sunlight. They also produce stable photocurrents up to 2.7 mA cm -2 at 0.6 V RHE in the electrolyte containing 1 m H 2 SO 4 and 0.1 m KI. Using various characteri zations and nanoparticle deposition, the origin of this high photocurrent is shown to be the combination of excellent opto electronic properties, unique stepped morphology that exposes multiple edge sites in every nanoflake, and the optimized nanoflake height (≈1.5 µm) that balances light absorption and charge transport. We recognize that the thickness and packing density of the nanoflakes also need to be optimized, which can be addressed in the future. Through this work, we highlight a unique path to controlling the orientation and morphology of 2D SnS 2 nanoflakes for efficient solar energy conversion.
Hydrothermal Synthesis of SnS 2 Powder: SnS 2 powder to be used as an evaporation source for close space sublimation was synthesized by modifying a method reported elsewhere. [67] At first, 0.3 m thiourea (Sigma Aldrich, >99%) and 0.1 m tin(IV) chloride pentahydrate (Sigma Aldrich, >98%) were dissolved in 40 mL deionized (DI) water and stirred vigorously for 30 min. Then, the solution was poured into a steel autoclave lined with Teflon and heated at 180 °C for 24 h. Afterward, the SnS 2 precipitate was collected, and rinsed with ethanol and DI water several times. This precipitate was initially dried in air at 50 °C to allow most of ethanol and water to evaporate. Then it was dried further at 200 °C in vacuum for 8 h to remove any remaining moisture and unreacted sulfur.
Synthesis of Vertical SnS 2 Nanoflakes by CSS: First, FTO glass (Hartford Glass, IN) substrates of typical size 2 cm × 2.5 cm were sonicated and cleaned three times in a 1:1:1 mixture of isopropanol, acetone, and deionized water. Then they were dried with compressed air and used to grow SnS 2 nanoflakes by CSS.
A custom-built CSS system (Figure S1d, Supporting Information) was used to grow vertical SnS 2 nanoflakes directly on FTO-coated glass substrates. The system consisted of a 2 in. diameter quartz tube that enclosed a stainless steel (SS) fixture for holding the FTO substrate, and a custom-made stainless steel source boat (L × W × H = 25 mm × 7 mm × 5 mm) containing SnS 2 powder. A rotary vane pump (KJLC-RV206) was used to pump down the tube. A 60 mm diameter co-flow premixed laminar flat-flame porous-plug burner equipped with a co-flow shroud (manufactured by Holthuis and Associates, Sebastopol, CA) was used to provide heat externally. The source temperature was controlled by the CH 4 -to-air ratio (typical flow rates: 3.5 lpm CH 4 and 35 lpm air) in the flat flame burner while the substrate temperature was controlled by the flow rate of water (typical flow rate: 0.5 lpm) through the substrate holder. Source and substrate temperatures were measured using K-type thermocouples that were in contact with the source and substrate, respectively, and the total pressure in the quartz tube was monitored by a convection enhanced Pirani gauge.
In a typical growth, a clean FTO substrate was mounted onto the underside of the water-cooled substrate holder using stainless steel wires such that it was directly above the flat flame. The source boat containing 0.1 g of SnS 2 powder was placed off to one side of the substrate, and the tube was pumped down to about 10 mTorr. Then, the burner was started in order to preheat the system. To ensure that the substrate had the same temperature throughout the growth process, the source boat was pushed under the substrate only after the substrate temperature had stabilized. This was marked as the beginning of growth. When the desired growth time had elapsed, the flat flame was extinguished and the system was allowed to cool naturally to room temperature while maintaining vacuum. The quartz tube was then vented and the sample was removed.
Synthesis of Single Crystal SnS 2 : CVT yielded the single crystal SnS 2 material. Quartz tubes (19 mm O.D., 17 mm I.D., G.M. Associates, Oakland, CA) were sealed at one end. Cleaning included a 1 week submersion in a 10 w/v% aqueous KOH solution, rinsing in 18 MΩ water, and oven drying at >100 °C. Loading under an air ambient included 1.3 g of tin (99.999% purity, 1-3 mm pieces, Strem, Newburyport, MA, used as received), 0.75 g of sulfur (≥99.5%, Sigma-Aldrich, used as received), and 0.20 g of iodine (99.999% trace metal basis, Sigma-Aldrich, used as received). The iodine mass was chosen to yield an ≈4 mg cm -3 charge for a 23 cm long tube that, in concert with the temperature values below, yields the 18R polymorph with minimal contributions from other polytypes. [68] Connection to a diffusion-pumpequipped Schlenk line with a base pressure below 1 × 10 -3 Torr afforded tube evacuation. Prior to evacuation, the tube contents were cooled in a water ice bath for 5 min to minimize the loss of iodine vapor during evacuation, followed by expeditious sealing under vacuum. Heating in a two-zone tube furnace utilized rapid heating to 800 °C at the newly sealed deposition side, and 2 °C min -1 heating to 700 °C in the "source" zone. Following a 6 h period of 700 °C in the source zone and 800 °C in the deposition zone, the temperature in the deposition zone decreased directly to and remained at 575 °C for 240 h while the source zone remained at 700 °C. Following this growth period, two successive cooling steps included one 6 h step of decreasing the temperature in the source zone at 2 °C min -1 to 300 °C while raising the temperature in the deposition zone at 2 °C min -1 to 600 °C, and a second cooling step to room temperature in both zones at a maximum rate of 2 °C min -1 .
Characterization: The morphologies and crystal structures of SnS 2 nanoflakes on FTO were characterized using SEM (JEOL 7000F, 10 kV), TEM (JEOL 2100, 200 kV), AFM (NaioAFM, Nanosurf, static force mode), and parallel beam XRD (PANalytical Empyrean, Cu kα, 45 kV, 40 mA). ICDD# 00-023-0677, ICDD# 04-03-3325, ICDD# 00-014-0619, and ICDD#00-039-0354 were used as standard patterns for 2HSnS 2 , 4HSnS 2 , Sn 2 S 3 , and SnS, respectively.
Wavelength-dependent LHE of the SnS 2 photoanode was measured using an integrating sphere with white-light illumination from the Xe lamp (model 6258, Oriel) and calculated from the measured reflection (R) and transmission (T) as LHE (λ) (%) = 100% -R(λ) (%) -T(λ) (%).
Raman spectra of the samples were obtained using an XploRa Raman microscope (Horiba Scientific, USA) operating with a 532 nm laser. Laser light was focused on the sample using a 100× magnification lens from Olympus, and a 2400 line grating was used along with accumulation time of 2 s and 25 repetitions to collect the signal.
The PL of SnS 2 nanoflakes was measured using a Horiba iHR550 spectrometer with 405 nm fiber-coupled laser and a Horiba Synapse CCD camera. Similarly, TRPL was measured using Becker and Hickl's time-correlated single photon counting system (SPC-150) and ID Quantique's single photon avalanche photodiode (ID-100-50-STD). The pump was a 20 ps pulsed laser with a frequency of 19.4 MHz, average power of 0.5 µW, and a spot size of about 1 mm 2 . Hence, the instantaneous peak intensity of the pulsed laser incident on the sample was about 128 mW cm -2 , which was similar to 1 sun condition.
Intrinsic mobility of photoexcited charge carriers in SnS 2 nanoflakes was estimated using TRTS. The vertical nanoflakes on quartz were excited by ≈100 fs duration, 400 nm pulses from a regeneratively amplified 1 kHz repetition rate Ti:sapphire laser at normal incidence as described in the previous reports. [69,70] The sample was placed behind a 1.5 mm aperture in the center of an ≈3 mm to 400 nm beam spot to 1901236 (10 of 11)
Adv. Energy Mater. 2019, 1901236 ensure uniform excitation of the studied portion. THz probe pulses were generated using optical rectification of the 800 nm pulses from the same laser source in a 1 mm thick [110] ZnTe crystal and focused onto the sample by a combination of off-axis parabolic mirrors. Transmitted THz pulses were focused onto a second [110] ZnTe crystal and detected using free space electro-optic sampling.
XPS compared surface compositions of the as-grown SnS 2 nanoflakes with those following tests in sulfuric acid and in phosphate buffer. Measurements were performed using a Phi 5600 instrument that utilized a monochromated Al Kα anode operating at 13.5 kV and 300 W with photoelectron collection in 25 meV steps at a 23.5 eV pass energy. [71,72] Similarly, ICP-OES (Agilent 5100) was also used to measure the concentration of dissolved tin ions in the electrolyte after degradation tests.
Electrode Preparation: The SnS 2 electrodes were made by bonding nickel-chromium alloy wire to the FTO using silver epoxy (Ted Pella #16043), which was then covered using nonconductive epoxy (Loctite 1C). For single crystal electrodes, millimeter-sized single crystals of SnS 2 were stuck directly onto silver epoxy, and any surface other than SnS 2 was covered using nonconductive epoxy. Similarly, the exfoliated SnS 2 electrode was prepared by mechanically exfoliating single crystal SnS 2 pieces and transferring to ITO using thermal release tape.
Photo-Electrochemical Measurements: All photo-electrochemical measurements were performed in a three-electrode configuration where SnS 2 nanoflakes on FTO were the working electrodes while saturated calomel electrode (SCE) was the reference electrode and platinum wire was the counter electrode. All voltages measured using SCE were converted to reversible hydrogen electrode (RHE) using the following equation
A magnetic stirrer was used together with nitrogen gas purging to eliminate any mass transfer limitations. In the LSV measurements, the sweep rate was set to 10 mV s -1 .
Simulated sunlight for these measurements was provided by a xenon lamp (model 6258, Oriel) equipped with an AM1.5G filter (model 81094, Oriel), and the spectral irradiance was measured by a spectrometer (model USB2000+Rad, Ocean Optics), which is shown in Figure S8b (Supporting Information). During the PEC tests, the incident light was chopped/blocked at regular intervals to establish the dark current level.
For IPCE measurements, monochromatic light at various wavelengths was produced with a monochromator (Cornerstone 130 1/8 m, Newport), and the following equation was used
λ λ [ ] ( )= × × × λ --J P IPCE 1240 mA cm mW cm nm 100% ph 2 2 (6)
Several electrolytes were used throughout this project: 1 m H 2 SO 4 (pH 0.5), 1 m H 2 SO 4 + 0.1 m KI (pH 0.5), 1 m phosphate buffer (pH 7), 1 m phosphate buffer + 1 m Na 2 SO 3 (pH 7), and 0.5 m Na 2 SO 4 (pH 7.6). The phosphate buffer was prepared by dissolving potassium phosphate monobasic (KH 2 PO 4 ) and potassium phosphate dibasic (K 2 HPO 4 ) in deionized water while sulfuric acid of unit molarity was prepared by diluting commercially available concentrated acid with deionized water.
Deposition of Copper (Cu) and Lead Oxide (PbO x ) Particles: Copper particles were electrodeposited onto SnS 2 nanoflakes in 1 m CuSO 4 •5H 2 O (pH 3.82) by applying cyclic voltammetry (20 times) from 0.74 to 0.57 V RHE . Similarly, PbO x was photo-electrochemically deposited in 0.1 m Pb(NO 3 ) 2 + 1 m HNO 3 by applying voltage from 1.28 to 2.08 V RHE over 2 min. Then, a constant voltage of 1.78 V RHE was maintained for 6 min to allow PbO x particle growth.
Adv. Energy Mater. 2019, 1901236
© 2019 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim