High‐Yield Production of Monolayer FePS3 Quantum Sheets via Chemical Exfoliation for Efficient Photocatalytic Hydrogen Evolution

2D layered transition metal phosphorus trichalcogenides (MPX3) possess higher in‐plane stiffness and lower cleavage energies than graphite. This allows them to be exfoliated down to the atomic thickness. However, a rational exfoliation route has to be sought to achieve surface‐active and uniform individual layers. Herein, monolayered FePS3 quantum sheets (QSs) are systematically obtained, whose diameters range from 4–8 nm, through exfoliation of the bulk in hydrazine solution. These QSs exhibit a widened bandgap of 2.18 eV as compared to the bulk (1.60 eV) FePS3. Benefitting from the monolayer feature, FePS3 QSs demonstrate a substantially accelerated photocatalytic H2 generation rate, which is up to three times higher than the bulk counterpart. This study presents a facile way, for the first time, of producing uniform monolayer FePS3 QSs and opens up new avenues for designing other low‐dimensional materials based on MPX3.


High-Yield Production of Monolayer FePS 3 Quantum Sheets via Chemical Exfoliation for Efficient Photocatalytic Hydrogen Evolution
Zhongzhou Cheng, Tofik Ahmed Shifa, Fengmei Wang, Yi Gao, Peng He, Kai Zhang,* Chao Jiang, Quanlin Liu, and Jun He* DOI: 10.1002/adma.201707433 monolayer and found a remarkably high value of 625.9 cm 2 V −1 s −1 at room tem perature, [9] which is even higher than other typical 2D materials such as mono layer MoS 2 (200 cm 2 V −1 s −1 ) [10] and WS 2 (214 cm 2 V −1 s −1 ). Although it is still under development, catalytic applica tions of nanostructured materials in this family have notable promising features. Theoretical speculations [9] reveal that the positions of their conduction and valance bands energy levels straddle the water redox potentials, making them appealing for photocatalyzing water splitting reac tions. Moreover, the calculated high car rier mobilities of the 2D MPX 3 materials such as MnPSe 3 indicate that the transfer of carriers to reactive sites would be easier in the photocatalytic process, reducing the possibility of elec tron-hole recombination. It is also important to note that the different kinds of elements in MPX 3 (particularly the calchogen and the metal atom) give rise to the variation of the bandgaps from 1.3 to 3.5 eV. [5] This rich and appropriate bandgaps bring the possibility to efficiently use visible light and design best performing photocatalysts. Our group has recently reported a systematic way of synthesizing ultrathin 2D NiPS 3 [11] and MnPX 3 (X = S and Se) [12] nanosheets. These experimental realizations communicated the promising hydrogen evolving activities of MPX 3 under illumination of simulated solar light without cocatalyst or sacrificial agents. In line with this, the 2D layered transition metal phosphorus trichalcogenides (MPX 3 ) possess higher in-plane stiffness and lower cleavage energies than graphite. This allows them to be exfoliated down to the atomic thickness. However, a rational exfoliation route has to be sought to achieve surface-active and uniform individual layers. Herein, monolayered FePS 3 quantum sheets (QSs) are systematically obtained, whose diameters range from 4-8 nm, through exfoliation of the bulk in hydrazine solution. These QSs exhibit a widened bandgap of 2.18 eV as compared to the bulk (1.60 eV) FePS 3 . Benefitting from the monolayer feature, FePS 3 QSs demonstrate a substantially accelerated photocatalytic H 2 generation rate, which is up to three times higher than the bulk counterpart. This study presents a facile way, for the first time, of producing uniform monolayer FePS 3 QSs and opens up new avenues for designing other low-dimensional materials based on MPX 3 .

Water Splitting
Owing to their interesting and useful properties, 2D mate rials, such as graphene, [1] MoS 2 , [2] and gC 3 N 4 , [3] have led to the increasing attention of scientific communities during the last decades. The size effects exhibited by fewlayered nanosheets of these materials result in fundamentally unique properties that substantially differ from their bulk counterparts. [4] Now adays, a new class of 2D layered metal phosphorus trichalcoge nides (MPX 3 , where M = Fe, Mn, Ni, etc. and X = S or Se) have received tremendous attentions. [5][6][7] It is gratifying that various important findings have been reported regarding the MPX 3 materials in different applications. [8] For instance, Zhang et el., have recently calculated the carrier mobility of 2D MnPSe 3 www.advmat.de www.advancedsciencenews.com issue of performance enhancement in photocatalysis is associ ated with strong light harvesting capability in the visible and nearinfrared region, multiple charge carrier generation, and large surface area to volume ratio. Quantum sheets (QSs) are the best candidates to virtually meet these requirements. [13,14] Recent studies show that the QSs demonstrate a particularly prominent multipleexciton generation effect, which is consid ered as a promising way to reduce heatrelated energy losses in solar system by splitting one highenergy photon into mul tiple lowenergy excitons thereby increasing energy conversion efficiency. [15,16] Given the very promising features of MPX 3 in catalysis and the attributes of QSs in photochemistry, the design of quantum confined materials in this family is of a great benefit to realize multiply advantageous photocatalyst.
Here, we made use of an important concept to enable us design QSs. That is, most of the MPX 3 compounds have higher inplane stiffness (60-120 N m −1 ) and lower cleavage energies (0.29-0.54 J m −2 ) [9] as compared to graphite(0.37 J m −2 ). [17] This suggests that the surfaceactive MPX 3 nanosheet can be practically exfoliated from the bulk. [18,19] Following this, we employ a facile way to exfoliate layered FePS 3 into monolayer QSs with a uniform lateral size of 4-8 nm. Benefitting from our facile exfoliation, the inplane surface is greatly altered toward exposing more active sites, which is prominent feature of efficient catalysis. Accordingly, photocatalytic test shows that the hydrogen generation rate of the FePS 3 QSs is up to three times (290 µmol g −1 h −1 ) higher than that of bulk FePS 3 (94 µmol g −1 h −1 ) under the same illumination conditions. We believe that, this work opens up a clear avenue for investi gation of other members in this family at the level of QSs.
The bulk FePS 3 crystal was synthesized by chemical vapor transport (CVT) method in a sealed quartz tube as can be seen from the schematic diagram in Figure 1a. The scanning elec tron microscopy (SEM) image of the as prepared hexagonal FePS 3 sheet, with a size about 200 µm, and the corresponding energy dispersive Xray spectroscopy (EDS) elemental map ping images are clearly shown in Figure 1b. More crystal struc tures of the bulk FePS 3 are displayed in Figures S1 and S2 in the Supporting Information. To obtain the desired FePS 3 QSs, a twostep exfoliation method was implemented. First, the prepared bulk FePS 3 was intercalated by reacting with hydra zine (N 2 H 4 ) in hydrothermal condition ( Figure 1c). Second, the intercalated FePS 3 crystals were exfoliated by sonication at a high power of 100 W for 4 h to form the QSs, as shown in Figure 1d. It can be seen that the QSs are evenly distributed on the substrate. The statistical distribution of the diameter (inset) illustrates that the isolated QSs tend to be 4-8 nm in size. This may provide better homogeneity for further charac terization and testing. More transmission electron microscopy (TEM) images of the QSs are depicted in Figure S3 of the Sup porting Information. It is obvious that the FePS 3 QSs, with the size below 10 nm, are well dispersed on the ultrathin carbon film with a homogeneous thickness ( Figure S3b,c, Supporting Information). The mechanism of the exfoliation processes can reasonably be explained by a redox rearrangement model. [20] At the first glance, part of the N 2 H 4 were oxidized to N 2 H 5  www.advmat.de www.advancedsciencenews.com during intercalation. Up on heating, the intercalated N 2 H 5 + was decomposed to different gaseous species such as N 2 , NH 3 , and H 2 owing to its poor thermal stability. This brought the forma tion of expanded crystals in FePS 3 , which after were treated at high power sonication. The ultrasonic waves in solvent would generate cavitation bubbles and then collapse into highenergy jets, breaking up the expanded crystals and producing QSs. In addition, modeling has shown that if the surface energy of the solvent is similar to that of the layered material, the energy difference between the exfoliated and reaggregated states will be very small, removing the driving force for reaggrega tion. [21][22][23] For example, graphene, hBN, transition metal dichalcogenides materials, and some transition metal oxide materials have been exfoliated by using suitable solvents such as Nmethylpyrrolidone. [24] In our system, FePS 3 material was estimated to have a surface energy of ≈100 mJ m −2 . [9,25] On the other hand, the solvent surface energy is related to the surface tension by [23] E T S sur sol sur where γ is the solvent surface tension (mN m −1 ), E sur sol is the sol vent surface energy (mJ m −2 ), T is the temperature (≈300 K), and S sur sol is the solvent surface entropy (≈0.1 mJ m −2 K −1 ). Given the surface tension of water (γ = 72 mN m −1 at 300 K) and hydrazine hydrate (γ = 74 mN m −1 at 300 K), their surface energy can be calculated as 102 and 104 mJ m −2 , respectively, which are very close to that of FePS 3 . This result means that water and hyadrazine hydrate are suitable solvents to realize the successful exfoliation into monolayer FePS 3 QSs.
In order to have a detailed examination of the samples, we  for the case of QSs. This can be attributed to the diminishing of (001) plane as the bulk FePS 3 crystals is transformed into monolayer. We also characterized the surface chemical states by Xray photoelectron spectroscopy (XPS), shown in Figure S7 in the Supporting Information. Accordingly, the highresolution XPS spectra of Fe 2p region indicate that the oxidation state of Fe in FePS 3 is +2. A critical comparison of XPS spectra (bulk vs QSs) reveals that there is no obvious change on the surface chemical states, further corroborating the safety and stability of our exfoliation method.
The analysis of Raman spectra gives more insight under standing of the distinction between the bulk and monolayer QSs, displayed in Figure 2d. According to the report by Scagliotti et al., [26] the Raman modes of FePS 3 originate from two parts of the crystal structure (the vibrations of metal atom and the P 2 S 6 unit, which belongs to the D 3d symmetry group). Three A 1g type modes (polarized, A 1g (1)(2)(3) ) and three E g type modes (depolarized, E g (1)(2)(3) ) from the vibrations of the P 2 S 6 unit could be resolved in the spectra ( Figure S8, Supporting Information). As shown in Figure 2d, the peaks at 214 cm −1 (E g (1) ), 240 cm −1 (A 1g (1) ), 274 cm −1 (E g (2) ), 372 cm −1 (A 1g (2) ), and 586 cm −1 (E g (3) ) can be assigned to the FePS 3 QSs, and those at 153 cm −1 (E u ), 223 cm −1 (E g (1) ), 245 cm −1 (A 1g (1) ), 272 cm −1 (E g (2) ), and 382 cm −1 (A 1g (2) ), 486 cm −1 (A 1g (3) ), and ≈580 cm −1 (E g (3) ) can be associated to the bulk FePS 3 . [26][27][28] The A 1g modes represent the stretching vibration of the P-P band, which indicate the outof plane vibrations of the P 2 S 6 unit ( Figure S8, Supporting Information). The A 1g (1) is due to the opposite movement of the S 3 P-PS 3 unit, which has been found very sensitive to alkaliion inter calation and to the caxis expansion. Moreover, the A 1g (2) and A 1g (3) modes represent the symmetric stretching vibration of the P-S bonds and the relative movement of the S 3 P-PS 3 unit, respectively. Moreover, the E g modes are meant for the tan gential vibration of the P-P bond, which indicate the inplane vibrations of the P 2 S 6 unit ( Figure S8, Supporting Information). The E g (1) and E g (2) are active in two orthogonal scattering geom etries where as the E g (3) is sensitive to the lattice distortions. [26] The peak at about 153 cm −1 is the Raman counterpart of the strongly infrared active E u type mode of the P 2 S 6 unit observed in all members of the MPS 3 compound absorption spectra.
When the unit cell is doubled along the c axis, the outof plane vibration of two P 2 X 6 units in adjacent layers becomes Raman active. [26,29] From the inset of Figure 2d, the absence of E u peak in QSs sample suggests the achievement of monolayer FePS 3 QSs.
To explore the optical properties of the FePS 3 bulk and QSs, UV/vis/NIR diffusereflection spectra (DRS) and ultraviolet photoelectron spectroscopy (UPS) were used and the results are shown in Figure 3. According to the Kubelka-Munk theory, the absorption coefficient (α) could be obtained by the Kubelka− Munk function (F(R)) from the diffuse reflectance measure ment. [30] Also, the bandgap energy of the material is estimated by the intercept of the tangent to the Xaxis (hv) in the Tauc plot [31] (inset of Figure 3a). The FePS 3 is known to be an indirect bandgap semiconductor so that the (αhv) 1/2 instead of (αhv) 2 is set as the Yaxis fro contracting the Tauc plot. [32] Accordingly, the bandgap values of FePS 3 bulk and QSs are found to be 1.60 and 2.18 eV, respectively (Figure 3a, inset). The UPS was used to determine the valence band energy level (E v ) of the bulk and QSs, as shown in Figure 3b. Through subtracting the width of He I UPS spectra from the excitation energy (21.22 eV), the E v values were calculated to be −5.40 eV for the bulk and −5.57 eV for the QSs (vs E Vacuum ), respectively. Then the conduc tion band energy (E c ) of the material can be calculated by E v -E g (−3.80 eV for bulk and −3.39 eV for QSs). As a matter of fact, QSs exhibit a wider bandgap (2.18 eV), compared to the bulk (1.60 eV). No matter how the FePS 3 QSs display a more nega tive overpotential than H + /H 2 (−0.20 V for bulk and −0.61 V for QSs, vs NHE, PH = 7), the feature of QSs enables the accumula tion of abundant hot electrons at E c for catalyzing the hydrogen evolution reaction ( Figure S9, Supporting Information).
Having seen the wellexfoliated QSs of FePS 3 and the appro priately positioned E c level for sunlight driven catalysis of H 2 gas production, we conducted photocatalytic hydrogen evolution experiments in aqueous solution containing 10 vol% trietha nolamine (TEOA) as the hole scavenger. The performance of bulk FePS 3 was also measured under exactly the same condi tion for the sake of comparison. As can be seen in  www.advmat.de www.advancedsciencenews.com bulk FePS 3 . Furthermore, the stability of QSs for hydrogen pro duction was evaluated and the result is depicted in Figure 4b. The sample exhibits a feasible ability in the cycling measure ments of H 2 generation under light illumination for a contin uous 40 h test. This enhancement can be attributed to the great amount of exposed boundaries providing more active sites and efficient separation of photogenerated electrons in QSs. From such all dimensional quantum confinement, it can be inferred that there are significant differences from the bulk in photo catalytic reaction, as shown in Figure 4c. Under the light illu mination, the photogenerated electron and hole are located in the conduction band and valence band separately. The photo generated electrons are consumed to produce H 2 molecules while the photogenerated holes are used to oxide the sacrificial agent (TEOA). Exfoliating the bulk into monolayered FePS 3 QSs highly contributes to the increase in surface area/active sites, which is beneficial for the enhancement of photocatalytic activity.
In summary, we have developed a facile way to obtain mono layer FePS 3 QSs from the bulk FePS 3 , which was first prepared by CVT method. The QSs demonstrate a lateral size below 10 nm. The exfoliation method brought no change on the chemical composition and state of the elements in FePS 3 . Given the 3D confinement, there resulted an associated change in the electronic band structure of the obtained QSs as compared to the bulk. A three times performance enhancement (290 µmol h −1 g −1 for the QSs and 94 µmol h −1 g −1 for bulk) is observed for QSs in photocatalytic hydrogen evolution reaction that can be attributed to the monolayer effects. We believe that these facilely synthesized monolayer QSs can also find a very promising performance in other applications as well.

Experimental Section
Preparation of FePS 3 Powders: A stoichiometric amount of high-purity elements (Fe: 99.99%, P: 99.999%, S: 99.99%,) and iodine (30 mg) as a transport agent were sealed into a quartz tube under vacuum and heated at 700 °C for two weeks. After cooled down to room temperature, a product of black powder with metallic luster was obtained.
Preparation of FePS 3 QSs: The as-synthesized bulk FePS3 (30 mg) with 20 mL of hydrazine hydrate (80%) was sealed in an autoclave and heated at 80 °C for 4 h. The mixture was then sonicated in a high-power sonic bath (100 W) for 4 h to form a homogeneous suspension. Eventually, the desired FePS 3 QSs were obtained by a centrifugal separation at 8000 rpm for 15 min to remove unexfoliated material. For the further measurements, the QSs were spin-coated on the FTO and dried at 60 °C for 24 h.
Characterization of Materials: SEM and the corresponding EDS mapping images were obtained from a Hitachi S4800 field-emission scanning electron microanalyzer with EDS. TEM and HRTEM images were obtained via employing a Tecnai G2 F20 with beam energy of 200 keV. The thickness was analyzed by AFM (MFP-3D Infinity). XRD   patterns were collected from a D/MAX-TTRIII(CBO) system using a Cu-Kα radiation (λ = 1.5418 Å). Raman spectroscopy was measured at room temperature on an inVia Renishaw system at the excitation line of 532 nm. The Raman peak of Si at 520 cm −1 was used as a reference to calibrate the spectrometer. XPS was performed on the Thermo Scientific ESCALab 250Xi using 200 W monochromatic Al Kα radiation. The 500 µm X-ray spot was used for SAXPS analysis. The base pressure in the analysis chamber was about 3 × 10 −9 mbar. Typically, the hydrocarbon C1s line at 284.8 eV from adventitious carbon is used for energy referencing. UV/vis/NIR DRS were recorded on a Lambda 750 spectrophotometer equipped with an integrating sphere. The valence band energy of the samples was analyzed on Thermo Scientific ESCALab 250Xi using UPS. The gas discharge lamp was used for UPS, with helium gas admitted and the HeI (21.22 eV) emission line employed. The helium pressure in the analysis chamber during analysis was about 2 × 10 −8 mbar. The data were acquired with −10 V bias.
Measurement of Photocatalytic H 2 Evolution: Photocatalytic water splitting experiments were conducted in a 500 mL cylinder quartz reactor at ambient temperature. A 300 W xenon lamp used as a light source. In a typical H 2 evolution experiment, the prepared photocatalyst (on FTO, about 0.6 mg cm −2 ) was placed at the bottom of reactor containing 100 mL of aqueous solution with 10% TEOA. Before irradiation, the system was vacuumed for about 30 min to remove the air inside and to ensure that the system was under the anaerobic condition. A certain amount of gas was intermittently sampled and analyzed by gas chromatography (GC7900, Shimadzu, Japan, TCD, nitrogen as a carrier gas and 5 Å molecular sieve column) by using Ar as a carrier gas. A baseline was recorded for each test before exposure to xenon lamp.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.