Facile Solvent-Free Synthesis of Metal Thiophosphates and Their Examination as Hydrogen Evolution Electrocatalysts

The facile solvent-free synthesis of several known metal thiophosphates was accomplished by a chemical exchange reaction between anhydrous metal chlorides and elemental phosphorus with sulfur, or combinations of phosphorus with molecular P2S5 at moderate 500 °C temperatures. The crystalline products obtained from this synthetic approach include MPS3 (M = Fe, Co, Ni) and Cu3PS4. The successful reactions benefit from thermochemically favorable PCl3 elimination. This solvent-free route performed at moderate temperatures leads to mixed anion products with complex heteroatomic anions, such as P2S64−. The MPS3 phases are thermally metastable relative to the thermodynamically preferred separate MPx/ MSy and more metal-rich MPxSy phases. The micrometer-sized M-P-S products exhibit room-temperature optical and magnetic properties consistent with isolated metal ion structural arrangements and semiconducting band gaps. The MPS3 materials were examined as electrocatalysts in hydrogen evolution reactions (HER) under acidic conditions. In terms of HER activity at lower applied potentials, the MPS3 materials show the trend of Co > Ni >> Fe. Extended time constant potential HER experiments show reasonable HER stability of ionic and semiconducting MPS3 (M = Co, Ni) structures under acidic reducing conditions.


Introduction
Hydrogen is an important fossil fuel alternative with high energy density (~120 kJ/g) and potentially low environmental footprint [1,2]. There is strong interest in the electrochemical splitting of water into hydrogen and oxygen using renewable energy resources (green hydrogen). Sluggish kinetics for electrochemical water splitting, particularly the oxygen evolution reaction (OER), limits effective large-scale hydrogen production [3]. Expensive precious-metal catalysts, such as RuO 2 and IrO 2 for OER [4,5], and Pt for the hydrogen evolution reaction (HER) [6,7], show high activity for electrocatalytic water splitting, though in some cases they suffer from long-term stability issues. In addition to complex metal oxides, there is renewed interest in investigating non-oxide materials with surface structures that can catalyze water splitting electrochemistry. A diverse range of transition-metal phosphides, nitrides, sulfides, oxides, carbides, and borides reportedly show electrocatalytic activity for water splitting reactions [8][9][10][11][12][13][14].
Within the M-P and M-S families, several compositions and structures show electrocatalytic HER activity; these include Ni 2 P, CoP, MoS 2 , CoS 2 , and NiS 2 with appreciable electrocatalytic activity and stability in electrochemically reducing environments in acidic electrolytes [14][15][16][17]. Our recent synthetic and electrocatalytic work describes solvent-free direct MCl 2 /P reactions to form phosphorus-rich MP 2 and MP 3 materials that have moderate HER catalytic behavior and robust stability in acid [18]. The most HER active phosphides, i.e., those with the lowest applied potentials, were CoP 3 and NiP 2 . Anion-rich structures have lower metal content and potentially protective polyanion networks that may better shield the metals from degradation in corrosive environments versus metal-rich systems. While phosphides and sulfides are frequently grown as nanostructures on porous supports, well-defined syntheses of micrometer-sized products are desirable to examine bulk properties and minimize size-dependent effects on catalytic behavior. Theoretical studies on metal phosphides indicate that surface P-H bonding may be more favorable for HER kinetics and so non-metal anions on the solid surface may be as important or more important than surface transition metal cations [14,[19][20][21][22][23].
In addition to anion-rich metal phosphide or sulfide structures, there exist many non-metal-rich thiophosphate structures that contain P x S y anions in extended ionic solid networks. Solid-state materials in the layered MPS 3 family contain P 2 S 6 4− anions, are often semiconducting [24,25], and show isolated metal-ion paramagnetism [26,27]. In addition to MPS 3 (M= Mn, Fe, Co, Ni), other observed crystalline M-P-S phases include those closer to MX 2 compositions (MP x S 2-x such as CoP 0.5 S 1.5 ) [28]. Layered MPS 3 structures have been extensively studied for their electronic, magnetic, and intercalative properties [29][30][31][32][33]. These anion-rich layered structures have been examined as materials for the reversible intercalation of lithium and sodium ions in battery studies [34][35][36][37][38]. The MPS 3 materials have also been studied as photo-electrocatalysts and hydrogen storage materials [39][40][41]. There are recent studies of MPS 3 materials as heterogeneous electrocatalysts for hydrogen reduction and oxygen evolution reactions (HER and OER) [42][43][44][45]. Several recent comprehensive reviews compare and contrast MPS 3 structures, electronic and catalytic properties, and photochemical behavior, which serves to highlight resurgent interest in these compounds with catalytic activity and low metal content [46][47][48].
The syntheses of MPS 3 materials often require careful and extended heating to produce these P/S-rich structures without decomposing them into other more stable metal-rich structures or poorly crystalline products. For example, MPS 3 structures are typically produced by heating pure elements at~650 • C for 5-17 days [38,42], or at 750 • C for 7 days [45]. Partially crystalline MPS 3 materials are grown from layered metal oxides reacted with phosphorus and sulfur at 520 • C [37]. Crystalline MPS 3 is also produced from a reactive P 2 S 5 melt with elemental metal powders heated at 580-650 • C for 2-5 days [49]. The P 2 S 5 reaction study noted that 580 • C was the maximum CoPS 3 reaction temperature to avoid formation of CoPS impurities. Overheating the products of MPS 3 reactions may lead to the decomposition into multi-phase products, including binary phosphides, sulfides, and lower content thiophosphates. The single-step, solvent-free, moderate temperature synthesis of crystalline MPS 3 materials is challenging.
This report describes facile exchange reactions that are solvent-free and thermochemically driven to produce materials in the M-P-S family via reactions of anhydrous metal chlorides with phosphorus and either sulfur or pre-bonded molecular P 2 S 5 at 500 • C. The successful reactions are explained in the context of reaction thermochemistry. The structure, composition, and physical properties of the M-P-S products are reported. Crystalline singlephase MPS 3 (Fe, Co, Ni) materials synthesized by this solvent-free exchange reaction are examined as HER electrocatalysts in 0.5 M H 2 SO 4 and compared with prior electrocatalytic results for these thiophosphate materials.

M-P-S Synthesis from Elemental Phosphorus and Sulfur
Initial reactions between metal chlorides and combinations of elemental phosphorus and sulfur (P + S) showed that MPS 3 phases or Cu 3 PS 4 form in these heated solventfree exchange reactions, so these phases were stoichiometrically targeted along with a thermodynamically stable PCl 3 byproduct. The M-P-S reactions using P/S elemental reactants for either MPS 3 or Cu 3 P 4 products are shown in Equations (1) and (2). In each case, the P/S reactant is stoichiometrically balanced to match the target phase and provide excess phosphorus to sequester the chlorine as volatile PCl 3 (mp −94 • C, bp 76 • C).
3 MCl x + (3 + x) P +9 S → 3 MPS 3 + x PCl 3 (x = 2 for Ni, Co, x = 3 for Fe) The chosen metal dichlorides have melting points above the 500 • C reaction temperature (~700-1000 • C), while FeCl 3 melts and decomposes to FeCl 2 around 310 • C that melts at 667 • C. The non-metal reactants show low temperature phase transitions, specifically for molecular sulfur S 8 (mp 115 • C, bp 445 • C) and polymeric red phosphorus (sublimes as P 4 near 420 • C). At the end of the 500 • C reaction, the M-P-S products are isolated by transporting the PCl 3 byproduct and the unreacted non-metal intermediates (if present) to the empty end of the ampoule using a temperature gradient. While sulfur chlorides such as SCl 2 (∆H f = −50 kJ/mol) may potentially form as reaction byproducts, they have lower thermochemical stability than PCl 3 (∆H f = −320 kJ/mol) [50]. Thiophosphate product formation via this solvent-free exchange reaction follows our prior success with phosphorus-rich MP 2 and MP 3 materials grown using a thermochemically driven PCl 3 elimination strategy [18]. The thermochemical driving forces that lead to successful reactions between MCl x and different P and S reactants will be described later. The naming scheme for these products from reactions using P and S is M x P y S z [P + S], such as CoPS 3 [P + S].
The M-P-S products that resulted from reactions of iron, cobalt, and nickel chlorides with elemental P/S reactants are black solids, while the Cu-P-S product is a green solid. Powder X-ray diffraction (XRD) analysis of these metal thiophosphate products show that several of these products form crystalline MPS 3 structures ( Figure 1). The major phases that are observed include monoclinic FePS 3 (PDF #00-033-0672) and monoclinic NiPS 3 (PDF #01-078-0499) with several small peaks for cubic NiS 2 (PDF #04-003-4307). In the case of Co-P-S, two similar intensity crystalline phases are identified by XRD, monoclinic CoPS 3 (PDF #01-078-0498), and cubic-phase CoP 0.5 S 1.5 (Co 2 PS 3 , PDF #04-007-4518). This latter phase adopts the pyrite CoS 2 structure with phosphorus substituting some sulfur in a solid-solution formula of CoP x S 2-x where x =~0.50 [28]. The XRD pattern for the Cu-P-S shows only crystalline peaks for orthorhombic Cu 3 PS 4 (PDF #04-004-0447).

M-P-S Synthesis from Elemental P and P2S5
In a similar fashion to the elemental reactions described above, MPS3 and Cu3PS4 reactions were performed using a molecular P2S5 reactant (mp 288 °C, bp 514 °C) containing pre-bonded phosphorus and sulfur atoms. Additional red phosphorus was used to target

M-P-S Synthesis from Elemental P and P 2 S 5
In a similar fashion to the elemental reactions described above, MPS 3 and Cu 3 PS 4 reactions were performed using a molecular P 2 S 5 reactant (mp 288 • C, bp 514 • C) containing pre-bonded phosphorus and sulfur atoms. Additional red phosphorus was used to target the desired M-P-S product with PCl 3 byproduct formation (Equations (3) and (4)).
15 MCl x + 9 P 2 S 5 + (5x−3) P → 15 MPS 3 + 5x PCl 3 (x = 2 for Ni, Co, x = 3 for Fe) (3) 15 CuCl 2 + 4 P 2 S 5 + 7 P → 5 Cu 3 PS 4 + 10 PCl 3 (4) The XRD patterns of the products synthesized from the metal halides and P 2 S 5 /P are shown in Figure 2. Like the elemental reactions, the metal thiophosphates crystallized as monoclinic FePS 3 , monoclinic CoPS 3 , and monoclinic NiPS 3 , and orthorhombic Cu 3 PS 4 . The P 2 S 5 reaction with cobalt chloride produced single-phase CoPS 3 versus the elemental P/S reaction. No impurity phases were identified in the XRD patterns, though several peaks have different relative intensities as compared with their standard patterns, which may indicate preferred growth along a crystallographic direction or stoichiometry differences. The naming scheme for these products from reactions using P 2 S 5 and P is M x P y S z [P 2 S 5 + P], such as CoPS 3 [P 2 S 5 + P].  The unit cell structures of the MPS3 and Cu3PS4 products are shown in Figure 3. Extended structures are shown to illustrate the layering of the MPS3 materials with anionrich layers nearest to the interlayer region. The MPS3 structures contain 2 M 2+ cations with each P2S6 4-anion, which form PS3 pyramidal units that point into the interlayer spaces above/below the metal containing layer. The MPS3 form isostructural monoclinic structures with all three lattice parameters shrinking from Fe to Co to Ni in the MPS3 structure (see Supplementary Table S1 for details). In contrast, Cu3PS4 is an orthorhombic threedimensional structure that contains three Cu + cations and PS4 3-anions. The unit cell structures of the MPS 3 and Cu 3 PS 4 products are shown in Figure 3. Extended structures are shown to illustrate the layering of the MPS 3 materials with anionrich layers nearest to the interlayer region. The MPS 3 structures contain 2 M 2+ cations with each P 2 S 6 4− anion, which form PS 3 pyramidal units that point into the interlayer spaces above/below the metal containing layer. The MPS 3 form isostructural monoclinic structures with all three lattice parameters shrinking from Fe to Co to Ni in the MPS 3 structure (see Supplementary Table S1 for details). In contrast, Cu 3 PS 4 is an orthorhombic three-dimensional structure that contains three Cu + cations and PS 4 3− anions.
olecules 2022, 27, x FOR PEER REVIEW 6 of 21  Table 1 summarizes key XRD and product yield information for the elemental P/S and P2S5/P reaction products targeting MPS3 (M = Fe, Co, Ni). The product yields for samples range from the ~60-95% range with higher yields for phase-pure products from P2S reactions. Energy dispersive spectroscopy (EDS) analysis performed on the M-P-S prod ucts shows low chlorine residues for FePS3 from the elemental P/S reaction to nearly undetectable chlorine (<2%) for other samples, indicating effective PCl3 elimination, consistent with our prior work on crystalline MP2/MP3 synthesis [18]. The P/S content from EDS data is slightly higher than the ideal 1:3 ratio for MPS3 and the overall higher nonmetal content is detected, which could indicate some excess P and/or S polymeric components on the particle surfaces. The Cu3PS4 reactions using P/S or P2S5/P reactants had high product yields (>90%) and undetectable chlorine by EDS. Semiquantitative EDS analysis translated to the relative product compositions for the (P + S) reaction of Cu3P1.38S3.75 and the (P2S5 + P) reaction of Cu3P1.20S3.60, which are near the Cu3PS4 target composition. Since EDS provides an estimate of the relative composition, it may be biased towards particle surface coatings. A more quantitative bulk analysis of the entire MPS3 samples was performed by ICP-OES analysis ( Table 1). The bulk ICP compositional data provide good agreement with MPS3 formulations and the (P + S) mixed phase cobalt product has a lower P/S content consistent with its detected Co0.5P0.5S1.5 composition.   Table 1 summarizes key XRD and product yield information for the elemental P/S and P 2 S 5 /P reaction products targeting MPS 3 (M = Fe, Co, Ni). The product yields for samples range from the~60-95% range with higher yields for phase-pure products from P 2 S 5 reactions. Energy dispersive spectroscopy (EDS) analysis performed on the M-P-S products shows low chlorine residues for FePS 3 from the elemental P/S reaction to nearly undetectable chlorine (<2%) for other samples, indicating effective PCl 3 elimination, consistent with our prior work on crystalline MP 2 /MP 3 synthesis [18]. The P/S content from EDS data is slightly higher than the ideal 1:3 ratio for MPS 3 and the overall higher non-metal content is detected, which could indicate some excess P and/or S polymeric components on the particle surfaces. The Cu 3 PS 4 reactions using P/S or P 2 S 5 /P reactants had high product yields (>90%) and undetectable chlorine by EDS. Semiquantitative EDS analysis translated to the relative product compositions for the (P + S) reaction of Cu 3 P 1.38 S 3.75 and the (P 2 S 5 + P) reaction of Cu 3 P 1.20 S 3.60 , which are near the Cu 3 PS 4 target composition. Since EDS provides an estimate of the relative composition, it may be biased towards particle surface coatings. A more quantitative bulk analysis of the entire MPS 3 samples was performed by ICP-OES analysis ( Table 1). The bulk ICP compositional data provide good agreement with MPS 3 formulations and the (P + S) mixed phase cobalt product has a lower P/S content consistent with its detected Co 0.5 P 0.5 S 1.5 composition.

Particulate Morphologies of M-P-S Materials
Representative SEM images of the M-P-S products from the direct solvent-free exchange reaction of metal chlorides with P/S or P 2 S 5 /P reactants are shown in Figures 4 and 5, respectively. The FePS 3 samples grow as flat plate-like structures (~3-15 µm wide by 1-2 µm thick) that are mixed with microparticle aggregates. Similarly, the CoPS 3 products form as large aggregates comprised of faceted blocky particles~1-5 µm in size. In some cases, large monoliths consist of smaller fused octahedral particles that are several hundred nm in size. The NiPS 3 samples consist of large aggregates of small irregular fused particles that are in the~1-10 µm range, but with a wider range of sizes and less well-formed faceted particles than the Fe and Co samples. The Cu 3 PS 4 samples consisted of small particle aggregates (~1-5 µm) and larger faceted monoliths (~10-20 µm). Overall, the products from P/S or P 2 S 5 /P reactions are similar in morphology, but there are indications that the P 2 S 5 reactions may produce larger faceted particles, in the CoPS 3 case ( Figure 5). Several higher magnification TEM images of the P 2 S 5 /P MPS 3 products are shown in Figure 6 for smaller suspended portions of these catalyst samples. The morphologies are generally small few micrometer-sized faceted particles that are fused into larger aggregates. Some of the smallest particles are near 0.5 µm in size.

Particulate Morphologies of M-P-S Materials
Representative SEM images of the M-P-S products from the direct solvent-free exchange reaction of metal chlorides with P/S or P2S5/P reactants are shown in Figures 4 and 5, respectively. The FePS3 samples grow as flat plate-like structures (~3-15 µm wide by 1-2 µm thick) that are mixed with microparticle aggregates. Similarly, the CoPS3 products form as large aggregates comprised of faceted blocky particles ~1-5 µm in size. In some cases, large monoliths consist of smaller fused octahedral particles that are several hundred nm in size. The NiPS3 samples consist of large aggregates of small irregular fused particles that are in the ~1-10 µm range, but with a wider range of sizes and less wellformed faceted particles than the Fe and Co samples. The Cu3PS4 samples consisted of small particle aggregates (~1-5 µm) and larger faceted monoliths (~10-20 µm). Overall, the products from P/S or P2S5/P reactions are similar in morphology, but there are indications that the P2S5 reactions may produce larger faceted particles, in the CoPS3 case ( Figure  5). Several higher magnification TEM images of the P2S5/P MPS3 products are shown in Figure 6 for smaller suspended portions of these catalyst samples. The morphologies are generally small few micrometer-sized faceted particles that are fused into larger aggregates. Some of the smallest particles are near 0.5 µm in size.

Spectroscopic and Magnetic Analysis of M-P-S Products
The MPS 3 (M = Fe, Co, Ni) and Cu 3 PS 4 samples were analyzed by IR spectroscopy to identify P-S bond vibrations for the thiolate anions. The IR spectra for several reaction products are shown in Supplementary Figure S1. Each sample shows a clear peak around 570-580 cm −1 , which is characteristic of the PS 3 asymmetric stretching vibration [41]. The P-S thiolate vibrations for FePS 3 , CoPS 3 , and NiPS 3 from P/S reactants are at 572 cm −1 , 580 cm −1 , and 572 cm −1 , respectively. The Cu 3 PS 4 sample shows a similar, but less intense peak near 500 cm −1 due to the structural differences between the MPS 3 and Cu 3 PS 4 . The IR data for the samples synthesized from P/P 2 S 5 reactants displayed similar IR results.
The diffuse reflectance optical absorption results for the M-P-S materials are shown in Supplementary Table S2. All MPS 3 samples are black, visibly reflective, and thus show broad absorption across the visible light region and do not show band gap onsets until the edge of the visible/IR region (E g~1 .3-1.7 eV, 740-940 nm). The Cu 3 PS 4 samples are brownish-green and have slightly higher band gaps and visible absorption onsets (E g = 2.36 eV,~525 nm). These optical absorptions and band gaps of single-phase products are consistent with literature reports for these MPS 3 and Cu 3 PS 4 materials. In contrast to observed optical properties, theoretical band structures predict a small band gap for FePS 3 and zero band gap for NiPS 3 and CoPS 3 [51].
Room temperature magnetic susceptibility measurements on the M-P-S materials from either synthetic method result in magnetic moments that are generally consistent with paramagnetic spin-only magnetic moments for isolated metal ions observed for metal thiophosphates (Supplementary Table S3) [42]. Specifically, the MPS 3

Thermochemical Comparison of MPS 3 Materials to MP x and MS y Counterparts
The direct reactions of elements often require relatively high temperatures, multi-day reaction times, and produce thermodynamically stable products. In the case of elemental reactions with 3d metals and phosphorus or sulfur, the MP x and MS y or MX 2 , where X is a combination of P/S, products are generally more thermodynamically stable, and therefore they are more easily produced compared to the mixed anion-layered MPS 3 structures with P 2 S 6 4− dumbbell anions. A comparison of standard heats of formation [50,52], and the calculated energy/atom values [51], for several related MP 2 /MP 3 , MS 2 , MPS, and MPS 3 products are shown in Table 2. Since the heats of formation for the mixed M-P-S phases have not been experimentally determined, comparisons are made using the calculated eV/atom formation energies. The eV/atom energy stability comparisons show that the MS 2 structures (energies in bold) are stable relative to both the MP x and MPS 3 structures. In the cobalt case, the CoPS structure is near CoS 2 in stability. Overall, the per atom formation energies indicate that all three MPS 3 materials are predicted to be unstable versus MPS and M x S y structures and P x S y products [51].
The thermochemical stability differences are directly applicable to the elemental syntheses, which are the most common routes to synthesize bulk phosphide, sulfide, and MPS 3 materials. Typically, complex M-P-S materials, such as MPS 3 (M = Fe, Ni, Co) are produced via elemental reactions at temperatures of 650 • C or higher that take several days to over a week to complete reactions [42,48,[53][54][55]. It has been noted that some elemental syntheses are complicated by the formation of more stable metal sulfides, such as CoS 2 . Tuning reaction thermochemistry through judicious choices of reactants and thermodynamically favored byproducts can enable reactions to proceed in a more facile manner than observed in solid-solid or solid-gas elemental reactions. In our previous work on the solvent-free synthesis of MP 2 and MP 3 materials from direct metal halide reactions with elemental red P at 500 • C, the thermochemical driving force was primarily the formation of a stable PCl 3 byproduct (∆H f = −320 kJ/mol) [18]. This led to moderately exothermic reaction enthalpies (∆H rxn from~−20 to −100 kJ/mol); endothermic MP x reactions (e.g., MnP 4 ) were unsuccessful. Relevant to this current synthetic work, is whether reactions of metal halides with elemental sulfur would yield S 2 Cl 2 (∆H f = −58 kJ/mol) and stable metal sulfides (∆H f~1 50-300 kJ/mol). All such reactions for Fe, Ni, and Co are predicted to be highly endothermic (>+140 kJ/mol). These control reactions were performed, and no metal sulfides were produced, instead only metal halides and transported sulfur were detected (Supplementary Table S4).
In contrast, when both P and S (or P 2 S 5 and P) reactants were used, the PCl 3 elimination pathway is again thermochemically available (as shown in Equations (1)-(4)), and metal thiophosphates form in moderate to high yields. It is possible that initial M-P formation occurs between MCl x and P 4 vapor, followed by sulfur incorporation from S 8 vapor. It is also likely that in the P/S and P 2 S 5 /P reactions with metal halides, the formation of volatile P x S y molecular intermediates favor the production of P 2 S 6 4− anions (and MPS 3 formation) during solid-gas exchange reactions at the solid MCl x surface. Molecular P 2 S 5 and other P x S y products are formed from heating phosphorus and sulfur [56].
All non-metal reactants used in these solvent-free M-P-S syntheses should be in the liquid or vapor state prior to the 500 • C reaction temperature being reached: sulfur S 8 (mp 115 • C, 445 • C), polymeric red phosphorus (sublimes as P 4 near 420 • C), and molecular P 2 S 5 (mp 288 • C, bp 514 • C). In practice, some surface reactions are observed by~350 • C, thus reactions on the ground MCl x powder surface may initially occur with liquid sulfur or P 2 S 5 and vaporized P 4 . Since the strategy for these thermochemically-driven chemical exchange reactions are successful for the synthesis of MPS 3 (M = Fe, Ni, Co) and Cu 3 PS 4 structures, it may be a synthetically useful strategy to access other metal thiophosphate and mixed metal thiophosphate structures.

Examination of Electrocatalytic HER Activity for MPS 3 Products
Our prior work on MP 2 and MP 3 compounds, showed that the FeP 2 , NiP 2 , and CoP 3 exhibit HER activity despite their high non-metal phosphorus content, with the order of increasing activity of CoP 3 > NiP 2 >> FeP 2 [18]. These results indicate that non-metal rich materials can be useful HER catalysts, despite a low metal content. Carbon wax electrodes developed in our prior work were embedded with M-P-S powders via direct adhesion to the conducting wax surface and they were examined in 0.5 M H 2 SO 4 using linear sweep voltammetry (LSV) and constant potential time base measurements (chronoamperometry, CA). The Cu 3 PS 4 samples showed rapid degradation or dissolution during initial HER experiments and were not studied further. The approximate thickness of the embedded MPS 3 powders is about 40 µm, assuming 1 mg amounts and~3 g/cm 3 density. This estimation is likely an ideal upper limit that assumes a uniform, dense coating, but in reality, these micrometer-sized powders are loosely packed and partly embedded powders in the conducting wax surface. The single phase MPS 3 products from P 2 S 5 reactions, show parallels to the P-rich metal phosphides, with CoPS 3 showing higher activity than NiPS 3 , and FePS 3 showing low to negligible activity relative to the blank C wax tip (Figure 7). A 10%Pt/C powder was also used for comparison purposes and was analyzed in the same manner as the MPS 3 samples. The data in Table 3 show a summary of the average LSV results. The individual MPS 3 LSV data are shown with and without an applied 85% iR correction in Supplementary Figures S2-S5 and in Table 3. Both results are shown as several studies caution that care should be taken in making iR corrections on catalytic materials as they can mask catalyst charge transfer differences [57][58][59]. As expected, iR correction lowers the potentials necessary to achieve a current density of 10 mA/cm 2 by~50 mV and FePS 3 shows the highest cell resistance (~350 Ω vs.~65 Ω for NiPS 3 and CoPS 3 ). Tafel slopes for initial current flow are tabulated with representative graphs in Supplementary Figure S6. Electrochemically active surface area (ECSA) data show that these large MPS 3 crystallites have relatively low surface charge accumulation relative to the Pt/C standard powder (Table 3, Supplementary Figure S7).
MPS3 powders is about 40 µm, assuming 1 mg amounts and ~3 g/cm 3 density. This estimation is likely an ideal upper limit that assumes a uniform, dense coating, but in reality, these micrometer-sized powders are loosely packed and partly embedded powders in the conducting wax surface. The single phase MPS3 products from P2S5 reactions, show parallels to the P-rich metal phosphides, with CoPS3 showing higher activity than NiPS3, and FePS3 showing low to negligible activity relative to the blank Cwax tip (Figure 7). A 10%Pt/C powder was also used for comparison purposes and was analyzed in the same manner as the MPS3 samples. The data in Table 3 show a summary of the average LSV results. The individual MPS3 LSV data are shown with and without an applied 85% iR correction in Supplementary Figures S2-S5 and in Table 3. Both results are shown as several studies caution that care should be taken in making iR corrections on catalytic materials as they can mask catalyst charge transfer differences [57][58][59]. As expected, iR correction lowers the potentials necessary to achieve a current density of 10 mA/cm 2 by ~50 mV and FePS3 shows the highest cell resistance (~350 Ω vs. ~65 Ω for NiPS3 and CoPS3). Tafel slopes for initial current flow are tabulated with representative graphs in Supplementary Figure S6. Electrochemically active surface area (ECSA) data show that these large MPS3 crystallites have relatively low surface charge accumulation relative to the Pt/C standard powder (Table 3, Supplementary Figure S7).  Table 3 and LSV data are in Supplementary  Figures S2-S5. n/a n/a n/a 2/1 29% @ −749 mV Data with 85% iR compensation applied are listed in Table 3 and LSV data are in Supplementary  Figures S2-S5.
The LSV experiments to determine the HER activity of the MPS 3 samples were initially performed with a platinum counter electrode (CE) and the resulting data were comparable to that shown in Table 3 (and Supplementary Figures S2-S5) using a graphite counter electrode with stable activity from CoPS 3 and slow degradation in activity for NiPS 3 and FePS 3 after cycling to −750 mV RHE. These HER LSV results showed no observed Pt interference, similar to our prior MP 2 /MP 3 studies [18]. In contrast, extended time 18 h constant potential CA measurements conducted to examine MPS 3 stability do show an apparent Pt effect in certain cases. The 18 h CA experiments with CoPS 3 showed little or no difference when a platinum or graphite CE was used. The higher applied potential for the CA experiments performed using a Pt CE with NiPS 3 (−420 mV) and FePS 3 (−750 mV) catalysts displayed a significant increase in catalytic activity of about 300% over the 18 h constant potential period (Supplementary Figure S8). In contrast, when the graphite CE was used, NiPS 3 and FePS 3 showed nearly constant activity or slowly decreasing activity (Supplementary Figure S9 and Table 3), which are better representations of their HER activity and stability over time, without possible platinum deposition issues. Microprobe analysis of the NiPS 3 and FePS 3 electrode tips after CA experiments using the platinum CE shows M/P/S elements, but Pt, if present, is below detection limits (Supplementary Figure S10).
n/a n/a n/a 2/1 29% @ −749 mV 10% Pt on C 47 ± 9 (122 ± 2) The results of the platinum CE track well with other reports describing the possibility of platinum ion migration from platinum counter electrodes that can impact metal catalyst stability in fuel cells and apparent catalyst activity in HER, particularly under acidic conditions and in the presence of oxygen-rich environments [60][61][62][63]. Graphite working electrodes show a significant increase in HER activity over time when using a platinum CE [64]. Platinum deposits forming at the cathode arise from reduction of platinum ions in the solution produced when a platinum anode is oxidized. The time lag for the deposition of platinum on the cathode is set by the slow rate of platinum anode oxidation and low concentrations of platinum ions in solution. Close proximity of the cathode and anode with rapid stirring of the electrolyte favor platinum deposition on the cathode. While Pt 2+ and Pt 4+ reduction to metal occurs near −1 V vs. NHE (similar to RHE at pH~0 conditions), there are indications that these values are substantially decreased depending on the nature of bound ligands on dissolved ions; for example, the standard potential for the reduction of ligated PtX 4 2− (aq) + 2e Pt (0) + 4X-(aq) varies with the halide as 0.758 V, 0.698 V, and 0.400 V vs NHE for X-of chloride, bromide, and iodide [65]. For Pt (X) 4 2− metal reduction, potentials move to lower negative potentials as the ligand X becomes larger and softer (e.g., I − or SCN − ). While the applied potentials here are below the~1 V needed for platinum reduction, the MPS 3 surfaces display significant bonded sulfur atoms that may serve to bind platinum ions analogous to a soft donor ligand and enhance their reduction. It is possible that dissolved P x S y ions from the catalyst could also enhance solution transport of platinum ions. In brief, the CoPS 3 is likely stable during HER because CoPS 3 catalyzes H 2 evolution at potentials positive for platinum deposition. Because NiPS 3 and FePS 3 reduce hydrogen at more negative applied potentials, their CA conditions require holding the catalyst at a higher applied potential that may be sufficient to more readily reduce platinum ions on the catalyst surface over long 18 h time periods when a Pt CE is used. Also, note that the standard Pt/C powder sample shows relatively low stability in acid under 18 h CA conditions (Table 3).
In addition to positioning crystalline catalyst particles in direct contact with the electrolyte solution, the carbon wax working electrode design allows for post-electrochemical analysis of the bulk catalyst by XRD and SEM in ways that are often difficult to achieve with a Nafion-embedded catalyst. Slices were taken off of the end of the wax electrode, i.e., the wax with embedded catalyst on the surface, after CA experiments and the slices were examined by powder XRD, similar to that reported in our recent metal boride paper [66]. These XRD results show clear retention of the original MPS 3 catalyst structures in the bulk material on the electrode surface (Figure 8). While the CA experiments suggest that NiPS 3 and FePS 3 exhibit some loss in catalytic activity with extended hydrogen reduction activity, their bulk structures remain intact. Surface decomposition reactions or their semiconductor band gaps may impact HER activities. SEM images of the MPS 3 solids on the wax tip after the CA experiments show generally smaller particle morphologies to those in Figure 5 (Supplementary Figure S11). The CoPS 3 and NiPS 3 are small <1 µm sized aggregates while the FePS 3 still shows some larger multi-micrometer sized faceted particles. Qualitative EDS elemental maps show that M-P-S elements are present in nearly MPS 3 composition on the surface along with oxygen (Supplementary Figure S12). While these MPS 3 samples show a range of SEM and TEM particle sizes and shapes and aggregation, their ECSA values are fairly similar ( Table 3). The CoPS 3 sample shows some surface roughening during HER, but its overall activity remains stable over the 18 h CA experiment period. Despite CoPS 3 having large crystallite sizes, its relatively larger ECSA (3x greater than NiPS 3 ) may reflect its better ability to transfer charge to bound H + versus NiPS 3 or FePS 3 , which both have smaller ECSA values. It is possible that the edge facets of these MPS 3 crystallites provide favorable sites for proton bonding.
In addition to positioning crystalline catalyst particles in direct contact with the electrolyte solution, the carbon wax working electrode design allows for post-electrochemical analysis of the bulk catalyst by XRD and SEM in ways that are often difficult to achieve with a Nafion-embedded catalyst. Slices were taken off of the end of the wax electrode, i.e., the wax with embedded catalyst on the surface, after CA experiments and the slices were examined by powder XRD, similar to that reported in our recent metal boride paper [66]. These XRD results show clear retention of the original MPS3 catalyst structures in the bulk material on the electrode surface (Figure 8). While the CA experiments suggest that NiPS3 and FePS3 exhibit some loss in catalytic activity with extended hydrogen reduction activity, their bulk structures remain intact. Surface decomposition reactions or their semiconductor band gaps may impact HER activities. SEM images of the MPS3 solids on the wax tip after the CA experiments show generally smaller particle morphologies to those in Figure 5 (Supplementary Figure S11). The CoPS3 and NiPS3 are small <1 µm sized aggregates while the FePS3 still shows some larger multi-micrometer sized faceted particles. Qualitative EDS elemental maps show that M-P-S elements are present in nearly MPS3 composition on the surface along with oxygen (Supplementary Figure S12). While these MPS3 samples show a range of SEM and TEM particle sizes and shapes and aggregation, their ECSA values are fairly similar ( Table 3). The CoPS3 sample shows some surface roughening during HER, but its overall activity remains stable over the 18 h CA experiment period. Despite CoPS3 having large crystallite sizes, its relatively larger ECSA (3x greater than NiPS3) may reflect its better ability to transfer charge to bound H + versus NiPS3 or FePS3, which both have smaller ECSA values. It is possible that the edge facets of these MPS3 crystallites provide favorable sites for proton bonding.  Previous literature shows a variety of HER activities for MPS 3 samples depending on preparation and level of catalyst support. Our HER results for crystalline unsupported MPS 3 powders embedded on a sticky conducting carbon wax electrode in acidic electrolyte are comparable with some prior studies on bulk and some nanostructured and supported MPS 3 materials. In basic solution, NiPS 3 requires near −350 mV applied potentials for 10 mA/cm 2 current densities, which is similar to our acid results [67]. Carbon nanosheet supported NiPS 3 and CoPS 3 show 10 mA/cm 2 current densities in basic electrolyte near −400 and −200 mV, respectively, reflecting the higher activity for CoPS 3 , analogous to our results [68]. In contrast, other studies report CoPS 3 and NiPS 3 with similar HER 10 mA/cm 2 current densities near −600 mV applied potentials in acidic electrolytes and show that CoPS 3 is stable upon extended cycling, while NiPS 3 is not [42]. A study of NiPS 3 supported on graphene achieved 10 mA/cm 2 activity near −100 mV applied potentials in acidic or basic electrolytes [69]. This study also reports −179 mV applied potential needed for unsupported NiPS 3 to achieve a 10 mA/cm 2 current density, similar to other studies [70]. NiPS 3 was also shown as stable in acidic environments when cycling was limited to −400 mV. FePS 3 materials reportedly show much lower activity than NiPS 3 materials, consistent with our results [70]. There are several studies showing that cobalt doping into NiPS 3 and FePS 3 structures greatly improves HER activity [71,72], and exfoliation of layered FePS 3 similarly improves activity [73]. Overall, the HER catalytic behavior demonstrated by the crystalline free-standing MPS 3 materials in this work is comparable to prior studies on unsupported and supported MPS 3 catalysts. Given that all three MPS 3 materials have near IR band gaps (~1.5 eV) and very non-metal rich compositions, it is not surprising that their HER activity requires substantial applied potentials. It is impressive that CoPS 3 converts H + to H 2 at relatively low applied potentials near −200 mV and exhibits good extended electrocatalytic stability towards HER. Photo-assisted electrocatalysis may further improve this material's HER activity.

Synthesis of M-P-S Materials in Sealed Ampoules
Transition-metal thiophosphate synthesis was performed using anhydrous metal halides and elemental non-metal reactant mixtures (P + S) or element/molecule combinations (P 2 S 5 + P). This work builds on our prior MP 2 /MP 3 synthesis using direct thermal reactions between metal halides and red phosphorus [18]. Both experimental routes described in this paper were balanced to target MPS 3 products where M = Fe, Co, Ni, except in the case of Cu-P-S reactions with CuCl 2 that were balanced to produce Cu 3 PS 4 as this phase was consistently formed during initial survey reactions. Typically, 0.500 g of the anhydrous metal halides FeCl 3 (3.08 mmol), CoCl 2 (3.85 mmol), NiCl 2 (3.86 mmol), or CuCl 2 (3.72 mmol) were ground with stoichiometric amounts of red phosphorus and elemental sulfur (or P 2 S 5 ) using an agate mortar and pestle in the argon-filled glove box. All reactions were designed to yield the respective M-P-S product and a PCl 3 byproduct, consistent with our prior MP 2 /MP 3 work [18,74]. The powders were loaded in a medium-wall Pyrex glass ampoule (9 mm OD) that was then closed with a valve and Cajon fitting, and removed from the glove box. The ampoule was evacuated on a Schlenk line for~15 min, and then flame-sealed under dynamic vacuum. All reactions were heated to 500 • C using a ramp rate of 100 • C/hr and held at 500 • C for~18-24 h. After heating, the ampoule end without the solid was pulled out of the furnace and cooled to room temperature. A colorless liquid and a yellowish-white solid was transported to the cooler end of the ampoule. When no further volatile transport was observed, the furnace was turned off and the tube was allowed to cool to room temperature. The glass ampoule was then opened in air to separate the product from transport. The isolated solid M-P-S product was weighed and stored in an argon-filled glove box. Little or no visible reaction was observed from the transport (e.g., smoking upon reaction with air), suggesting that PCl 3 or unreacted P 4 was sequestered in a more air-stable form with other P x S y transports. In air, phosphorus oxides can form from pure liquid PCl 3 visible as fuming white cloud and from P 4 visible as flame formation.

Sample Characterization
Powder X-ray diffraction (XRD) was performed on ground powders affixed to glass slides using a Bruker D8 Advance DaVinci diffractometer with nickel-filtered Cu Kα X-ray irradiation (40 kV, 40 mA, 10-80 degrees 2θ, 0.05 • /step). Reference XRD patterns were generated from crystal structure data using the PowderCell program [75]. The morphology and elemental analysis of M-P-S samples were investigated by scanning electron microscopy (SEM) and energy dispersive spectroscopy (EDS) on a Hitachi S3400N system. Samples were prepared by pressing ground powders onto carbon tape on aluminum stubs. EDS samples were prepared in a similar fashion but using thin pellets made with a KBr hand press. TEM was performed using a Hitachi S-7800 transmission electron microscope with an accelerating voltage of 100 kV on sonicated methanol suspended particles. Select electron microprobe analysis was performed on powders on C wax pieces using a JEOL JXA-8230 Electron Probe Microanalyzer. Selected powders on Cwax electrode tips were also examined by EDS elemental mapping after electrocatalysis. Inductively coupled plasma optical emission spectroscopy (ICP-OES, Perkin Elmer Optima DV 7000) was used to obtain quantitative compositional analysis of M-P-S materials dissolved in nitric acid (with heating as needed) versus commercial standards. The vibrational properties of the metal thiophosphate materials were analyzed using KBr pellets on a Nicolet Nexus 760 FT-IR spectrometer. Optical absorption measurements were performed at room temperature using a LabSphere RSA solid-state diffuse reflectance attachment on an HP 8453 UV-Vis spectrometer. The powder samples were embedded on filter paper and placed between two glass microscope slides. The absorption data were converted to Kubelka-Munk units, and the approximate energy band gap and onsets were determined. The band gaps were calculated using the Kubelka-Munk (KM) function [F (R) = (1−R) 2 /2R], where R is the diffuse reflectance of the sample. Extrapolation of the onset absorption events from a plot of the KM function versus energy yields estimated band gap (E g ) energies. The magnetic susceptibility of the M-P-S materials was analyzed at room temperature (298 K) using ground powders with a Johnson-Matthey MSB magnetic susceptibility balance. Molar magnetic susceptibility (χ m ) and spin only magnetic moment (µ B ) values were calculated from mass susceptibilities assuming paramagnetic behavior with correction for core diamagnetism.

Electrochemical Analysis
Working electrodes for electrocatalytic measurements were prepared using graphite/paraffin wax mixture (50% graphite in wax) inside a PTFE tube (C wax electrode) similar to that previously reported by our group [18,66]. This conducting carbon with an adherent (sticky) surface has shown utility in several prior electrochemical studies [76,77]. Working electrode tips were 1.4 cm long, 3.2 mm ID, and 6.4 mm OD, with a 0.080 cm 2 geometrical surface area and tip ends that were coned to reduce gas bubble adhesion. Prior to catalyst loading, blank C wax electrode tips were connected to brass current collectors and were submerged in a pre-heated water bath (55 • C) for~15 min. Homogeneous catalyst suspensions were prepared by brief sonication of~1-2 mg of catalyst with 20 µL of methanol and~5-10 µL aliquots were placed in an aluminum weigh boat to air dry. The air-dried catalyst-containing aluminum weigh boat was tared in a microbalance and was then placed in a preheated hot plate (55 • C). The softened C wax blank electrode tips were gently pressed onto the catalyst and excess catalyst powder on the PTFE tip lining was carefully returned to the aluminum weigh boat. The electrode tip was re-pressed several times at a clear space in the weigh boat to assure sample powders were firmly embedded on the wax. After the catalyst loading, the aluminum weigh boat was weighed on the previously tared microbalance and the mass of loaded catalyst was recorded; typical catalyst mass loadings ranged from about 0.5 to 1.0 mg.
Electrochemical measurements were performed in a Bioanalytical Systems (BASi) 100b potentiostat using a three-electrode cell with a C wax working electrode, 0.5 M H 2 SO 4 electrolyte, Hg/Hg 2 Cl 2 (SCE) reference electrode, and a platinum wire counter electrode; experiments using a graphite rod counter electrode (Alfa Aesar, 6.2 mm diam., SPK grade, 99.9995%) were conducted for comparison. A magnetic cross stir bar was placed directly under the working electrode (~6 mm away) to remove any gas bubbles and to minimize their adherence to the electrode surface. A schematic of the assembled cells and images of the C wax electrode are shown in Supplementary Figure S13. The SCE electrode potential values were converted to standard hydrogen electrode potentials for different pH values using the E RHE = E SCE + 0.059pH + E 0 SCE , with pH = 0.3 (0.5 M H 2 SO 4 ) and E 0 SCE = 0.241 V. All reported potentials are referenced to RHE values, and all current densities are calculated using the geometric surface area of the C wax electrode (0.08 cm 2 ). The electrolyte solutions were purged with H 2 gas (ultra-high purity 99.999%, Praxair) that were pre-humidified to minimize electrolyte evaporation by passing it through a water bubbler. Gas purging began 30 min before electrochemical measurements and continued throughout the experiment. HER activities and stability were evaluated using 50 linear sweep voltammograms (LSVs) taken without instrument iR correction. 30 LSV scans were taken with 85% instrument iR correction for comparison (typical R cell values for MPS 3 materials on C wax tips were M = Co, Ni~60 Ω, and Fe~300 Ω). Care was taken to consider the influence of iR corrections on these catalytic materials as this may mask catalyst charge transfer differences [57][58][59]. The electrochemical surface areas (ECSA) were determined from double-layer capacitance (C dl ) in the non-Faradaic region using cyclic voltammetry (CV) data at scan rates of 5, 10, 25, 50, and 75 mV/s with H 2 gas purge [18,58]. The capacitance values were converted to approximate areas using a 35 µF/cm 2 relationship [78]. The long-term HER stability of MPS 3 was investigated using 18 h time base chronoamperometry studies (CA) at constant potentials targeting~10 mA/cm 2 current density. XRD samples were prepared from the electrodes, post-electrochemical analysis, by cutting~1-2 mm slices from the end of the C wax electrode containing a thin surface coating of embedded MPS 3 powders and placing them in the well of an XRD sample holder for XRD analysis and used for EDS elemental mapping as described in our prior work (Supplementary Figure S13) [66]. The PTFE lining of the electrode tips was removed, and elemental mapping was conducted using electron microprobe analysis.

Conclusions
The direct solvent-free exchange reaction of anhydrous metal halides with either element P or S, P/S mixtures, or P 2 S 5 /P at 500 • C produces crystalline metal thiophosphates, with P 2 S 5 reactions generally leading to the phase-pure products, MPS 3 (M = Fe, Co, Ni) and Cu 3 PS 4 . The MPS 3 materials contain M 2+ cations and P 2 S 6 4− anions in a non-metal rich layered structure that form as micrometer-sized and sometimes faceted particles. While these three thiophosphates have similar low energy band gaps (~1.5 eV), they show a great variation in electrocatalytic HER activity when embedded on a conducting wax electrode. Whereas CoPS 3 shows the highest and most stable HER activity in acidic electrolyte, NiPS 3 is less active and stable, and FePS 3 appears highly resistant to performing HER. The effect of potential platinum counter electrode migration is identified for the lower activity NiPS 3 and FePS 3 samples. Post-electrocatalytic analysis of particles embedded on the wax electrode show strong evidence of bulk retention of the crystalline MPS 3 structure.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/molecules27165053/s1, crystal structure information for MP x S y materials; FT-IR data; UV-vis absorption and magnetic susceptibility results; thermochemical data for MP x and MS y reactions; LSV data for HER of MPS 3 with and without iR compensation; representative Tafel and ECSA data; graphs comparing 18 h CA experiments with platinum and carbon CE; microprobe analysis of select products on C wax tips; SEM/EDS of catalysts embedded on C wax electrodes after CA experiments; images of electrochemical cell [18,24,53,55,74,[79][80][81][82][83].  Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Not applicable.
Data Availability Statement: Experimental data used for graphical results shown in this study are available upon request from the corresponding author. Supplementary Materials provided with this paper contains additional tabular and graphical data that supports the reported results.