This is the submitted version of the journal article: Zhang, Wei; Han, Ning; Luo, Jiangshui; [et al.]. «Critical role of phospho- rus in hollow structures cobalt-based phosphides as bifunctional catalysts for water splitting». Small, Vol. 18, issue 4 (Jan. 2022), art. 2103561. DOI 10.1002/smll.202103561 This version is available at https://ddd.uab.cat/record/270835 under the terms of the license 1 High-performance Bifunctional hollow structures Cobalt-based 2 Phosphides Catalysts for Water Splitting 3 Wei Zhang1†, Ning Han1†, Jiangshui Luo*2, Wei Guo1, Sijie Xie1, Zhenyu Zhou1， 4 Palaniappan Subramanian1, Kai Wan1, Xu Han3, Jordi Arbiol3,4, Chi Zhang*5, 5 Maowen Xu6, Xuan Zhang*1 and Jan Fransaer*1 6 1 Department of Materials Engineering, KU Leuven, Leuven, 3001 Belgium. 7 2 College of Materials Science and Engineering, Sichuan University, Chengdu, 8 610065 P.R. China. 9 3 Catalan Institute of Nanoscience and Nanotechnology (ICN2), CSIC and BIST, 10 Campus UAB, Bellaterra, 08193 Barcelona, Catalonia, Spain 11 4 ICREA, Pg. Lluís Companys 23, 08010 Barcelona, Catalonia, Spain 12 5 School of Applied Physics and Materials, Wuyi University, Jiangmen, 529020 P.R. 13 China. 14 6 Key Laboratory of Luminescent and Real-Time Analytical Chemistry, Ministry of 15 Education, School of Materials and Energy, Southwest University, Chongqing 400715, 16 P. R. China. 17 † W. Z. and N.H. contributed equally to this work. 18 * Corresponding authors. 19 E-mail addresses: jiangshui.luo@scu.edu.cn (J. L); chizhang@wyu.edu.cn (C. Z); 20 xuan.zhang@kuleuven.be (X. Z); jan.fransaer@kuleuven.be (J. F) 1 21 Abstract 22 Cobalt phosphides electrocatalysts have great potential for water splitting, but 23 the unclear active sides hinder the further development of cobalt phosphides. Wherein, 24 cobalt phosphides hollow structure (CoP-HS, CoP2-HS and CoP3-HS) catalysts based 25 on same sacrificial template were prepared. Surprisingly, these cobalt phosphides 26 exhibited similar OER but quite different HER performances. The identical OER 27 performance of these CoPx-HS in alkaline solution were attributed to the similar 28 surface reconstruction to CoOOH. CoP-HS exhibited the best catalytic activity for 29 HER among these CoPx-HS, originating from the adjusted electronic density of 30 phosphorus to affect absorption-desorption process on H. Moreover, CoP-HS 31 displayed good performance (cell voltage of 1.47 V at a current density of 10 mA 32 cm−2) and high stability in 1 M KOH. This work could provide the guidance for the 33 future investigations on transition metal phosphides for water splitting from material 34 design to mechanism understanding. 35 2 36 1. Introduction 37 Hydrogen will almost certainly play a role in future renewable energy plans.1,2 38 Generation of hydrogen by sustainable electrochemical processes form water is an 39 appealing strategy.3,4 Electrochemical water splitting involves two reactions: the 40 cathodic hydrogen evolution reaction (HER) and the anodic oxygen evolution reaction 41 (OER). Presently, platinum (Pt) and noble metal oxides, including ruthenium dioxide 42 (RuO2) and iridium dioxide (IrO2), are the state-of-the-art electrocatalysts to drive 43 HER and OER, respectively.5 Nevertheless, the scarcity and associated high costs of 44 these noble metals severely restrict their large scale application.6 Hence, the lower 45 product cost, tolerance of catalysts working conditions for both reactions, the 46 development of bifunctional catalysts, free from Pt-group metals, has gained 47 increasing attention.7,8,9 48 Different transition metal-based catalysts have been explored as bifunctional 49 catalysts for electrochemical water splitting, such as oxides, hydroxides, borides, 50 selenides, nitrides and so on.10,11,12,13 Among these nonprecious metal electrocatalysts, 51 transition metal phosphides (TMPs) are considered as the promising alternatives 52 catalyst for both HER and OER. In particular, cobalt phosphides have attracted the 53 most attention due to their high catalytic activity and durability in alkaline solution.14 54 For instance, Yoo and co-workers reported that porous CoP nanoparticles exhibit good 55 HER and OER activities owing to the increase of the accessible catalytic active 56 sites.15 However, the poor conductivity, unclear active sides and low reaction kinetics 57 hinder the practical use of cobalt phosphides. 3 58 To further improve electrochemical performance of TMPs, two approaches have 59 been proposed: (1) Rational design of nanostructures to increase the number of 60 exposed active sites and promote ion transport, such as nanoneedle, nanosheet, 61 core-shell and hollow nanostructure.16,17,18,19 Among those, hollow nanostructured 62 catalysts materials have attracted a lot of research interests, because the high 63 surface-to-volume ratio, low density and small transmission lengths of mass and 64 charge.20 For example, compared with Ni–Co–P nanosheets, the Ni–Co–P hollow 65 nanobricks are demonstrated as the more efficient bifunctional electrocatalysts for 66 overall water splitting with a potential of 1.62 V to reach the current density of 10 mA 67 cm-2.21 (2) Understanding of active sites in cobalt phosphides. The mechanism of 68 water splitting on cobalt phosphides is still under debate due to the complicate phase 69 and structure change and unclear active sites. Some studies announced that the cobalt 70 phosphides are stable during OER cycling.22,23,24 Others believe that cobalt 71 phosphides transform to oxide/oxyhydroxide during OER.25 Besides, cobalt 72 phosphides have various stoichiometries, such as Co2P, CoP, CoP2 and CoP3. 73 In this work, based on the same ZIF-67 sacrificial template, a series of cobalt 74 phosphides with a similar, hollow morphology were synthesized. The resulting 75 catalyst exhibits similar nanocages with a porous structure. Interestingly, the different 76 cobalt phosphides showed similar OER performance but different HER performance. 77 In-depth structural characterizations and electrochemical studies show that the 78 different cobalt phosphides convert during OER to the active catalyst. After cycling in 79 alkaline solution, the active surface for all catalysts becomes oxyhydroxide which had 4 80 a similar activity with an overpotential of 270~280 mV at a current density of 10 mA 81 cm−2 for OER. On the contrary, all the cobalt phosphides with hollow structures 82 remained stability during HER test. Among these catalysts, CoP-HS nanocages 83 showed the best catalytic activity for HER with an overpotential of -116 mV which 84 could be the catalytically active sites for HER at a current density of -10 mA cm-2 in 1 85 M KOH. Density functional theory (DFT) computation revealed that the P-sites in 86 CoPx-HS are the primary catalytically active site for HER, and the calculated ΔGH* 87 based on P-sites of CoP-HS agrees well with the corresponding normalized HER 88 performances. When CoP-HS//CoP-HS were used for water splitting, a current 89 density of 10 mA cm-2 is achieved at a remarkably low cell voltage of 1.47 V, which is 90 due to the hollow structure of the catalysts and the high activity of the P-sites in CoP. 5 91 2. Results and discussion 92 93 Figure 1. (a) Schematic illustration of the synthesis process of hollow cobalt phosphides 94 nanocages. SEM images of pure ZIF-67 (b), Co(OH)2 (c), Co3O4 (d), CoP-HS (e), CoP2-HS (f) 95 and CoP3-HS (g). 96 The synthesis process of the three different cobalt phosphides nanocages are 97 shown in Figure 1a. In order to ensure the same morphology for all CoPx catalysts 98 were synthesized based on the same ZIF-67 sacrificial template. The rhombic 99 dodecahedral ZIF-67 particles are shown in Figure S1a. Next, the ZIF-67 particles 100 were converted into hollow Co(OH)2 structures by the reaction of with an aqueous 6 101 Na2MoO4 solution based on the Kirkendall effect.26,27 Co3O4 nanocages were 102 fabricated by annealing the hollow Co(OH)2 particles at 500 ºC in air. Subsequently, 103 the hollow Co(OH)2 and Co3O4 particles were annealed together with NaH2PO2 in N2 104 atmosphere together at 700 ºC, resulting in the formation of hollow CoP2 particles 105 (named CoP2-HS) and hollow CoP particles (named CoP-HS), respectively. Hollow 106 CoP3 particles (named CoP3-HS) was prepared by annealing Co3O4 particles with red 107 phosphorus in N2 at 700 ºC. As shown in Figure 1b, the prepared rhombic 108 dodecahedron shape ZIF-67 particles exhibited a uniform particle size of 109 approximately 800 nm and perfectly flat facets. The X-ray diffraction (XRD) pattern 110 of ZIF-67 matched well with the simulated pattern (Figure S1a), which demonstrates 111 the purity of the starting phase. It clearly shows that during these transformations, the 112 initial morphology and particle size of ZIF-67 is preserved (Figure 1c&1d). Because 113 the structure was transformed from solid to hollow, the surface of the Co(OH)2 and 114 Co3O4 nanocages becomes wrinkled and protuberance-like. The formation of hollow 115 structure of Co(OH)2 and Co3O4 can be identified from several broken samples. XRD 116 pattern presented in Figure S1b shows the presence of both α-Co(OH)2 and 117 β-Co(OH)2 in the prepared Co(OH)2. As shown in Figure S1c, the diffraction pattern 118 of the Co3O4 is in good agreement with the standard pattern of Co3O4 (JCPDS 1-1152) 119 powder. The uniform distribution of Co and P in the synthesized Co3O4 was proven by 120 the energy-dispersive X-ray (EDX) (Figure S2). After phosphidation, the three cobalt 121 phosphides have the original polyhedron shape of the ZIF-67 precursor. The diameters 122 of these particles are in the range of 600 to 800 nm. As shown in Figure S3, the 7 123 overlap of P and Co elements in SEM-EDS mapping confirms the homogeneous 124 formation of all three cobalt phosphides (CoP-HS, CoP2-HS, and CoP3-HS). 125 126 Figure 2. TEM images of CoP-HS (a), CoP2-HS (b) and CoP3-HS (c). EELS chemical 127 composition maps of CoP-HS (a), CoP2-HS (b) and CoP3-HS (c) obtained from area inside the red 128 square on the HAADF STEM micrograph. Individual Co L2,3-edges at 779 eV (red), P 129 L2,3-edges at 132 eV (green) and composite of Co-P. HRTEM images and corresponding indexed 130 power spectra (FFT) of CoP-HS showing a CoP orthorhombic crystal with Pnma phase along its 131 [12-2] axis (d), CoP2-HS showing a CoP2 monoclinic crystal with P21/C phase along its [10-1] 132 axis (e) and CoP3-HS showing a CoP3 cubic crystal with Im3 phase along its [-113] axis (f). (h) 133 BET surface area of CoP-HS, CoP2-HS and CoP3-HS based on cobalt content. (g) XRD patterns of 8 134 CoP-HS, CoP2-HS sand CoP3-HS. XPS spectra of Co 2p (i) and P 2p (j) for CoP-HS, CoP2-HS and 135 CoP3-HS. 136 The well-defined nanocage-shaped structure and particle sizes (range from 600 137 nm to 800 nm) of the three cobalt phosphides are also confirmed by transmission 138 electron microscopy (TEM) images (Figure 2). The EELS chemical composition 139 maps of the three cobalt phosphides were obtained from the region inside the red 140 square in the HAADF STEM micrograph, shown in Figure 2a-c, for CoP-HS, 141 CoP2-HS, and CoP3-HS, respectively. As can be seen in these maps, Co and P are 142 uniformly distributed in the selected nanostructure. But the P signal is stronger in 143 CoP2-HS and CoP3-HS. A HRTEM micrograph was taken from the nanoparticle 144 inside the orange square (Figure 2d). Details of this region and its corresponding 145 power spectrum reveals that this nanoparticle has a crystal phase that is in agreement 146 with the CoP-HS orthorhombic phase (space group = Pnma) with a=5.0760 Å, 147 b=3.2770 Å and c=5.5990 Å. From the crystalline domain in Figure 2d, the CoP-HS 148 lattice fringe distances are 0.286 nm, 0.237 nm, and 0.199 nm, at 78.95º and 122.43º 149 which correspond to the orthorhombic CoP phase, visualized along its [12-2] zone 150 axis. For the CoP2-HS sample, a HRTEM micrograph was taken from the nanoparticle 151 inside the orange square (Figure 2e). Details of this region and its corresponding 152 power spectrum, reveals that this nanoparticle has a crystal phase that is in agreement 153 with the CoP2-HS monoclinic phase (space group = P21/C) with a=5.5510 Å, 154 b=5.5490 Å and c=5.6140 Å. From the crystalline domain in Figure 2e, the CoP2-HS 155 lattice fringe distances are 0.277 nm, 0.272 nm, and 0.281 nm, at 60.83º and 121.60º 9 156 which correspond to the monoclinic CoP2-HS phase, visualized along its [10-1] zone 157 axis. As illustrated in Figure 2f, the CoP3-HS cubic phase (space group = Im3) with 158 a=b=c=7.7112 Å can be proven from the corresponding power spectrum from a 159 HRTEM micrograph taken from the nanoparticle inside the orange square. The 160 CoP3-HS lattice fringe distances (Figure 2f) are 0.553 nm, 0.243 nm, 0.313 nm and 161 0.313 nm, at 46.97º, 72.21º and 106.50º which could be interpreted as the cubic 162 CoP3-HS phase, visualized along its [-113] zone axis. 163 The crystalline structures and successful phosphidation process of these samples 164 are also confirmed by the XRD (Figures 2g). The experimental XRD patterns match 165 well with corresponding standard three cobalt phosphide crystal structures (ICDD 166 PDF: 029-0497 for CoP-HS, ICDD PDF: 01-077-0263 for CoP2-HS and ICDD PDF: 167 029-0496 for CoP3-HS) without unidentified peaks. These results indicate that the 168 Co(OH)2 and Co3O4 were completely converted into the intended cobalt phosphides. 169 The Brunauer-Emmett-Teller (BET) specific surface area and pore size distributions 170 of the as-synthesized three cobalt phosphides were determined by N2 sorption 171 experiments. The N2 adsorption-desorption isotherms and Barrett-Joyner-Halenda 172 (BJH) pore-size distribution of CoP-HS, CoP2-HS and CoP3-HS nanocages are shown 173 in Figure S4. All of the cobalt phosphides show a type-IV isotherm loop, which 174 indicates the mesoporous characteristics of the materials. For the three cobalt 175 phosphides, the major pore size distributions are displayed at a region of 1.5 to 3.0 nm. 176 Moreover, BET surface areas of CoP-HS, CoP2-HS and CoP3-HS are 24.1, 19.2 and 177 15.1 m2 g-1 (Figure 2h), respectively. With increasing phosphorus content, the 10 178 specific surface area decreased. Interestingly, after normalizing the specific surface 179 areas to the mass of cobalt, the normalized specific surface areas of all three cobalt 180 phosphides are similar (Figure 2h). This phenomenon can be explained by the same 181 self-sacrificing template was used. All these results prove that the successful synthesis 182 of three different cobalt phosphides (CoP-HS, CoP2-HS and CoP3-HS) with similar 183 structure and morphology by the presented top-down strategy. 184 X-ray photoelectron spectroscopy (XPS) is used for the analysis of chemical 185 state and the overall electronic structure of these catalysts. As shown in Figure S5, the 186 XPS survey spectrums of three cobalt phosphides reveal the existence of Co and P. 187 The Co 2p spectra of the CoP-HS (Figure 2i) is deconvoluted into six peaks, which 188 binding energies located at 778.9, 793.8, 787.6 eV that can be assigned to Co 2p3/2, 189 and binding energies located at 781.7, 798.5, 803.8 eV that can be assigned to Co 2p1/2. 190 Among them, the peaks centered at 787.6 and 803.8 eV are the satellite peaks of 2p2/3 191 and 2p1/2, respectively. The binding energy at 778.9 and 793.8 eV can be attributed to 192 Co3+, and those located at 781.7 and 798.5 eV belong to Co2+. Compared with CoP, 193 the Co 2p peaks in CoP2-HS and CoP3-HS shifted to a higher binding energy. The 194 main reason of this shift is due to the higher P content in CoP2-HS and CoP3-HS. A 195 higher concentration of electronegative phosphorus shifts the peaks of Co3+ to higher 196 binding energies. In addition, compared to the Co2+ peaks in the three cobalt 197 phosphides, there is a gradual decrease in the peak intensities from CoP-HS to 198 CoP3-HS. In the P 2p spectrum of CoP-HS (Figure 2j), the peaks located at 129.5, 199 130.3 and 134.2 eV are indexed to P 2p3/2, P 2p1/2 in cobalt phosphide and surface 11 200 oxidation of P, respectively.28 Compared with CoP3-HS, the peaks of CoP-HS and 201 CoP2-HS in P 2p3/2, P 2p1/2 region shifted to lower binding energy. The negative shift 202 of P 2p binding energy, induced by changing cobalt-phosphorus composition, shows 203 the enhanced electron occupation, which can give rise to the improvement of the 204 electron-donating ability of the catalyst.29 205 206 Figure 3. (a) The LSV curves of CoP-HS, CoP2-HS, CoP3-HS, IrO2 and carbon paper measured in 207 1.0 M KOH solution toward OER at a scan rate 10 mV/s after activation by 50 CV cycles between 208 0.0 V and 0.85 V (vs. Hg/HgO) at a scan rate 50 mV/s. (b) The data of CoP-HS, CoP2-HS and 209 CoP3-HS at different current density. (c) The corresponding Tafel plots for the samples. (d) The 210 CV curves obtained at the 1st and 10th cycles at a scan rate of 10 mV/s in a 1.0 M KOH solution. 211 (e) The potential of CoP-HS, CoP2-HS and CoP3-HS versus time at a current density of 20 mA 12 212 cm-2 in 1.0 MKOH. (All the tests were taken on carbon paper). (f) The schematic of surface 213 composition investigation via XPS in different depths after ion beam etching. (g) High-resolution 214 P 2p XPS spectra of the fresh CoP-HS, fresh CoP2-HS, fresh CoP3-HS and post-OER samples. (h) 215 High-resolution Co 2p XPS spectra of fresh CoP-HS, fresh CoP2-HS, fresh CoP3-HS and 216 post-OER sample. (i) High-resolution O 1s XPS spectra of the post-OER CoP-HS, post-OER 217 CoP2-HS and post-OER CoP3-HS samples. 218 Cobalt based phosphides are considered as promising bifunctional catalysts for 219 both HER and OER. Firstly, the electrocatalytic OER performance of three cobalt 220 phosphides were examined in 1.0 M KOH using a three-electrode system and 221 different electrodes were prepared by coating carbon paper with approximately 1 mg 222 cm-2 of catalyst. The active site of cobalt based phosphides for OER is still on debate. 223 Even though, cobalt based phosphides have been reported as catalysts for OER,23,24,30 224 the structure and phase changes occurring during OER tests cannot be ignored. To 225 investigate the phase changes of three cobalt phosphides, the CV curves of three 226 cobalt phosphides at the 1st, 3rd, 5th, and 10th cycle were measured. As illustrated in 227 Figure 3d&S6, at the 1st cycle, the OER performance of CoP-HS seems better than 228 CoP2-HS and CoP3-HS, which could be explained by unstable of CoPx-HS system. 229 Furthermore, as the number of cycles increases, the OER performance of CoP2-HS 230 and CoP3-HS improves, while the OER performance of CoP-HS decreases. At the 231 10th cycle the CV curves of CoP-HS, CoP2-HS and CoP3-HS almost overlap. 232 Therefore, all catalysts were first activated by 50 CV cycles at a scan rate 50 mV/s 233 between 0.0 V and 0.85 V (vs. Hg/HgO). LSV with iR-correction are used for 234 evaluated the OER performances of these catalysts (Figure 3a). For the purpose of 13 235 comparison, The OER performances of commercial IrO2 and bare carbon paper were 236 also investigated. As shown in Figure 3a, the CoP-HS requires an overpotential of 237 281 mV at a current density of 10 mA cm-2. Similar performance of the CoP2-HS and 238 CoP3-HS catalysts can be achieved, which overpotentials of resp. 268 mV and 269 239 mV at a current density of 10 mA cm-2. All these catalysts show better performance 240 than commercial IrO2 (375 mV). Meanwhile, the bare carbon paper hardly shows any 241 electrocatalytic activity for OER even at high overpotential (Figure 3a). The 242 corresponding Tafel slopes are derived from plotting overpotential against log (j) 243 (Figure 3c). CoP2-HS and CoP3-HS exhibit a similar Tafel slope of 57~62 mV dec-1. 244 The Tafel slope results suggest that CoP-HS, CoP2-HS and CoP3-HS have similar 245 charge transfer kinetics during OER. The electrochemical durability of the catalyst is 246 a key parameter for practical application. The electrocatalytic stability of different 247 cobalt phosphides electrodes were investigated by chronopotentiometry at a constant 248 current density of 20 mA cm-2 (Figure 3e). The chronopotentiometry curves show that, 249 after 100 h, the catalytic activity of CoP-HS, CoP2-HS and CoP3-HS dropped by 250 6.5 %, 6.0% and 7.7%, respectively. This result suggests a good electrochemical 251 stability of all three cobalt phosphides for the OER in alkaline solution. At the cycling 252 tests, the three cobalt phosphides prepared in this paper show almost identical results 253 (1.563 V for CoP-HS, 1.560 V for CoP2-HS and 1.563 V for CoP3-HS, vs. RHE), 254 which indicates that the P composition does not influence the OER performance of 255 cobalt based phosphides in an alkaline medium. This could be caused by the 256 oxidization of the similar structure of CoP-HS, CoP2-HS and CoP3-HS surface during 14 257 the OER test. After the OER electrochemical durability test, SEM and EDS 258 measurements were conducted to characterize the morphologies and surface 259 compositions of the three cobalt phosphides. The SEM images show that the 260 morphology and structure of three cobalt phosphides are reasonably retained after 261 OER durability test (Figure S7a-c). But the surface composition of three cobalt 262 phosphides changes showing a loss of phosphorous after long-term OER durability 263 test in an alkaline medium (Figure S7d). This result is consistent with previous 264 papers. Metal-based materials, such as metal chalcogenides, metal pnictides and metal 265 carbides, are partly or completely transformed to oxides or (oxy) hydroxides under 266 sufficiently extended periods during alkaline OER testing22,31,32. To further investigate 267 the post-OER electrodes, the XPS was used to analyze the electrodes. Combining a 268 sequence of ion gun etch cycles, more quantified vertical information and information 269 evolution with thicknesses was obtained by depth-profiling X-ray photoelectron 270 spectroscopy (Figure 3f). As shown in Figure 3g, the characteristic signals for 271 phosphorous almost completely disappears from the surface. As for cobalt, only two 272 broad peaks corresponding to oxidized cobalt species were observed on the surface 273 for all three cobalt phosphides after OER (Figure 3h), which is consistent with a 274 previous report on CoOOH.33 To further validate the formation of CoOOH, the O/Co 275 peak area ratio of was calculated based on the XPS detection at the surface and the 276 result is about 2:1 (Figure 3h&i). Argon ion sputtering was applied to remove the 277 oxidized surface (2.5 nm/per step) of the three cobalt phosphides after OER and 278 measured again to observe the XPS signals of P 2p (Figure 3g), Co 2p (Figure 3h), 15 279 and O 1s (Figure 3i). However, XPS results remain unchanged, which indicate the 280 complete transformation of the near surface region of all three cobalt phosphides (the 281 active region) during OER durability test. This explains the similar OER 282 performance after durability test. According to these results, the near surface region of 283 all three cobalt phosphides transform via self-reconstruction into CoOOH. Therefore, 284 P composition will not affect the OER performance in an alkaline medium. After the 285 conversion into CoOOH, these catalysts have comparable or better OER activity than 286 reported metal-based phosphides electrocatalysts as shown in Table S1. 287 288 Figure 4. (a) The LSV curves of CoP-HS, CoP2-HS, CoP3-HS and Pt/C measured in 1.0 M KOH 289 toward HER at scan rate 10 mV/s. (b) The corresponding Tafel plots for the samples in 1.0 M 290 KOH. (c) Nyquist plots of CoP-HS, CoP2-HS and CoP3-HS in 1.0 M KOH. (All the tests were 291 taken on carbon paper). (d) The CV curves of CoP-HS, CoP2-HS and CoP3-HS measured in 1.0 M 292 KOH solution for 1st and 10th cycles at a scan rate 10 mV/s. (e) The chronopotentiometry curve 16 293 of CoP-HS at the current density of -20 mA cm-2 for 100 h in 1 M KOH. The inset in (e) is the 294 SEM image of CoP-HS after 100 h HER stability measurement in 1 M KOH. (f) Overall water 295 splitting performance based on the CoP||CoP and Pt/C||IrO2 electrodes in 1 M KOH. The inset in (f) 296 is the potential of CoP||CoP and Pt/C||IrO2 at a current density of -10 mA cm-2. (g) Long term 297 durability test of CoP in the two-electrode electrolyzes at 20 mA cm-2 for overall water splitting. 298 The inset in (g) is optical image during the overall water splitting. (h) Amount of H2 and O2 as a 299 function of time. 300 On the contrary, the catalytic process of cobalt based phosphides for HER is 301 different from OER. Based on previous studies, the P sites in cobalt phosphides could 302 be the active site for HER. To investigate the HER activity of three cobalt phosphides, 303 LSV curves after CV scans in 1.0 M KOH were recorded at a scan rate of 10 mV/s, 304 and all the curves were IR corrected (Figure 4a). As expected, the HER activity of 305 bare carbon paper is negligible (Figure S8a), and the Pt/C exhibits the best activity 306 with an overpotential of 25 mV at the current density of -10 mA cm-2. 116 mV is 307 required for the CoP-HS to achieve the current density of -10 mA cm-2, which is much 308 lower than that of CoP2-HS (159 mV) and CoP3-HS (170 mV). The Tafel plots 309 derived from the LSV polarization curves are displayed to estimate the reaction 310 kinetics. As expected, the Tafel slope of commercial Pt/C is 34 mV dec-1 (Figure 4b), 311 which is consistent with the reported value.34 The Tafel slope of CoP-HS, CoP2-HS 312 and CoP3-HS is 89, 90, and 115 mV dec-1, respectively. According to literature, the 313 HER occurs through three different individual steps, which are named Volmer, 314 Heyrovsky, and Tafel reaction.35 The Tafel values of three cobalt phosphides indicate 315 that a Volmer-Heyrovsky reaction, which happens on the surface of catalysts, is the 17 316 rate-determining step. Moreover, electrochemical impedance spectroscopy was used 317 to investigate the electron transfer kinetics during the HER process. As shown in 318 Figure 4c, the Nyquist plots of different catalysts are measured at -0.19 V. The inset 319 in Figure 4c displays the equivalent circuit model. The intersection of the plots on the 320 real axis represents the solution resistance (Rs) and the diameter of the semi-circle 321 represents the charge transfer resistance (Rct) across the electrode-solution interface, 322 repectively.36 CoP-HS possesses the smallest Rct value.37 The phase and 323 electrochemical stability of the three cobalt phosphides during HER are investigated 324 in the same way as in the OER experiments. Firstly, the CV curves of three cobalt 325 phosphides at the 1st, 3rd, 5th, and 10th cycles were conducted. As illustrated in 326 Figure 4d&S9, after 3 CV cycles, the CV curves of the three cobalt phosphides 327 display only minor changes, which indicates that three cobalt phosphides reach a 328 stable electrochemical state in a short period of time. Overall, CoP-HS is always 329 displaying the best HER performance in three cobalt phosphides. It could be 330 explained by XPS that the bonding energy of P 2p negatively shifted with the decrease 331 of P concentration in cobalt phosphides (CoP-HS