Multistep Sulfur Leaching for the Development of a Highly Efficient and Stable NiSx/Ni(OH)2/NiOOH Electrocatalyst for Anion Exchange Membrane Water Electrolysis.

Nickel (poly)sulfides have been widely studied as anodic catalysts for alkaline water electrolysis owing to their diverse morphologies, high catalytic activities in the oxygen evolution reaction (OER), and low cost. To utilize low-cost and high-efficiency polysulfides with industry-relevant cycling stability, we develop a Ni-rich NiSx/Ni(OH)2/NiOOH catalyst derived from NiS2/Ni3S4 nanocubes. Ni-rich NiSx/Ni(OH)2/NiOOH shows improved OER catalytic activity (η = 374 mV@50 mA cm-2) and stability (0.1% voltage increase) after 65 h of a galvanostatic test at 10 mA cm-2 compared with commercial Ni/NiO and hydrothermally synthesized Ni(OH)2 (both show η > 460 mV@50 mA cm-2 along with 4.40 and 1.92% voltage increase, respectively). A water-splitting electrolyzer based on Pt/C||AF1-HNN8-50||NiSx/Ni(OH)2/NiOOH exhibits a current density of 1800 mA cm-2 at 2.0 V and 500 h high-rate stability at 1000 mA cm-2 with negligible attenuation of only 0.12 mV h-1. This work provides an understanding of truly stable species, intrinsic active phases of Ni polysulfides, their high-rate stability in a real cell, and sheds light on the development of stable chalcogenide-based anodic electrocatalysts for anion exchange membrane water electrolysis (AEMWE).

Introduction

Water electrolysis is of great importance for mitigating the greenhouse gas effect and producing high-purity hydrogen.[1−7] However, proton exchange membrane water electrolysis (PEMWE) has been impeded by high stack cost and scarcity of platinum-group metal (PGM)-based catalysts,[2,8−11] while classic alkaline water electrolysis shows poor polarization performance due to low conductivity of porous diaphragms.[2] Therefore, an anion exchange membrane (AEM),[12−16] combined with non-PGM catalysts, has been proposed as a scalable and cost-effective route for large-scale applications, yet still hindered by poor stability and low operating current density.[17−20] In particular, the kinetically sluggish oxygen evolution reaction (OER) slows down the overall anion exchange membrane water electrolysis (AEMWE), as it involves four electrons and must first break O–H and then overcome the formation energy of the O–O bond.[21−25] Therefore, it is needed to develop highly active, stable, and low-cost PGM-free OER catalysts for AEMWE.

To date, tremendous efforts for OER catalysts have focused on oxides/hydroxides, chalcogenides, and pnictides based on transition metals on porous substrates such as nickel foam (NF), which have impressively low overpotentials.[24,26−37] For example, Kim et al. developed (i) graphene-nanoplatelets-supported NiFe-MOF,[34] (ii) ruthenium core–shell and Ni single atom-based Ni–Ru catalyst,[35] (iii) crystalline–amorphous Ni2P@ FePOH, and (iv) amorphous NiFe (oxy)hydroxides, all exhibiting a high OER performance of 170–220 mV@10 mA cm–2.[36,37] Zhou et al. developed Ni3S2 nanorods on NF via a simple one-step hydrothermal process that exhibited 187 mV at 10 mA cm–2 in 0.1 M KOH.[38] However, Ni3S2/NF was only tested for 10 h at 10 mA cm–2 without supporting post-test development of the microstructure and composition. Additionally, Shang et al. fabricated the NiS/NF catalyst by in situ growth, showing rapidly declined performance after only 1000 CVs.[39] Overall, in situ-grown catalysts on NF exhibit high activity (low overpotential) for OER, but have the following problems for single-cell tests of AEMWE: (i) unfulfillable repeatability (uneven distribution and uncontrollable mass loading), (ii) low stability underflow mode (without binder reinforcement, largely washed away), and (iii) complete microstructure destruction after tests.[30,31] Compared with uncontrollable self-supported structures, catalyst coating layers with more controllable ink dispersion and catalyst mass loading exhibit high reproducibility and stability due to binder strengthening.[17,40−43] However, metal (poly)sulfides are almost unstable under strong polarization in alkaline solution during OER processes, especially under an oxygen-filled atmosphere. It has been demonstrated that transition-metal chalcogenides (TMCs) would be irreversibly oxidized to corresponding oxides/(oxy)hydroxides (TMOs/TMHOs).[44−49] However, these studies have not yet pointed out the structural and morphological behaviors of (poly)sulfides: (i) what extent (completely, partially) of sulfur leaching, (ii) the effect of residual sulfur on the stability of the (oxy)hydroxides, (iii) and more importantly, tangible high-current stability in single-cell tests, with almost no reports of polysulfides at a current density of ≥1000 mA cm–2.

Here, we fabricated the NiS2/Ni3S4 catalyst by a one-step, template-free method, the initial composition, phase, and microstructure of which were proved as S-rich NiS2/Ni3S4 composite nanocubes (Figure A). Then, a multistep electrochemical leaching method was applied to leach sulfur from S-rich NiS2/Ni3S4 nanocubes. NiS2/Ni3S4 was first partially converted to Ni(OH)2 (Figure B) and then to NiOOH, forming a highly stabilized Ni-rich NiS/Ni(OH)2/NiOOH catalyst after long-term tests (Figure C). It exhibited higher activity and stability than commercial Ni/NiO and hydrothermally synthesized Ni(OH)2 under both 100 mV s–1 cyclic voltammetry for 10 000 cycles and a constant current density of 10 mA cm–2 for 65 h. Moreover, this catalyst, coupled with Pt/C, was tested in single cells and exhibited higher performance (1800 mA cm–2 at 2.0 V) and higher stability (>500 h at 1000 mA cm–2) than Ni/NiO (1067 mA cm–2, <50 h) due to the refined and redistributed NiS/Ni(OH)2/NiOOH structure, suppressing bubble-induced voltage increase and catalyst shedding.

Figure 1

Schematic illustration of (A) NiS2/Ni3S4 composite nanocubes as a “precatalyst” and corresponding (B) activated NiS/Ni(OH)2/NiOOH covered with NiS2/Ni3S4 residues and (C) fully stabilized NiS/Ni(OH)2/NiOOH heterostructure.

Results and Discussion

Structural Characterization

NiS2/Ni3S4 was synthesized by a 4 h solvothermal method and explained in detail in Section . X-ray diffraction (XRD) was used to study the crystal structure of NiS2/Ni3S4 (Figure A) and shows diffraction peaks at 2θ values corresponding to the planes of NiS2 (JCPDS: 11-0099)[50−54] and Ni3S4 (JCPDS: 00-047-1739).[55,56] In addition, peaks of α-S8 in a small angle range (20–25°) are attributed to the byproduct of thioacetamide (TAA, S precursor).[57] X-ray photoelectron spectroscopy (XPS) was used to investigate the surface chemical state of NiS2/Ni3S4 (Figure B–D). The Ni 2p spectrum of NiS2/Ni3S4 was fitted into two peaks at 857.8 and 853.7 eV corresponding to Ni2+ 2p3/2 (sulfate) and Ni2+ 2p3/2 (sulfide)(Figure B),[50,52] indicating that pristine NiS2/Ni3S4 almost fully consists of Ni2+, which could result in low initial OER performance due to poor electrophilicity of adsorbed oxygen.[55,58,59] The S 2p spectrum (Figure C) was fitted into three peaks: the peaks at 161.5 and 162.6 eV correspond to 2p3/2 and 2p1/2 of S2–.[50,52,53] The peaks at 162.5 and 163.7 eV are attributed to 2p3/2 and 2p1/2 of S22–.[51,52,60] Moreover, the peaks at 163.5 and 164.7 eV are assigned to the spin–orbit of 2p3/2 and 2p1/2 in α-S8, indicating the remnants of S from TAA during the sulfurization process, which was produced by the reaction between Sn2– and H+ due to decreased pH.[50,61] The binding energies of O 1s at ∼529 and ∼531 eV, corresponding to NiO and Ni(OH)2, were not detected (Figure D),[62,63] indicating no oxides and hydroxides on the surface of the pristine NiS2/Ni3S4 catalyst. Therefore, the surface of NiS2/Ni3S4 is mainly composed of Ni2+, S2–, and S22–.

Figure 2

(A) XRD pattern of NiS2/Ni3S4. (B) Ni 2p peaks and the fitting results, (C) S 2p peaks and the fitting results, and (D) O 1s peaks and the fitting results of the NiS2/Ni3S4 catalyst. TEM, high-resolution TEM (HRTEM), HAADF-STEM images and corresponding elemental mappings of NiS2/Ni3S4: (E) low-magnification TEM, (F) geometric size of single nanocube, (G) HRTEM images of the NiS2 nanocube, (H) NiS2/Ni3S4 composite nanocube, (I) HAADF-STEM image of NiS2/Ni3S4 nanocube, and (J, K) distribution of Ni (green) and S (yellow) in EDX mappings.

The morphology of NiS2/Ni3S4 was observed by transmission electron microscopy (TEM). Figure E–G shows that NiS2/Ni3S4 inherits the cube-shaped morphology with a size of 80 ± 20 nm and the thickness of the nanocube is greater than 25 nm. This is achieved by moderate pH for producing elemental sulfur that would further react with nickel sulfides to form NiS2/Ni3S4 polysulfides (Figure S1) and can be controlled by the reaction time. Sulfur species suffered from precipitation with Ni2+ (after 2 h of reaction, leading to the formation of NiS/Ni3S4), polymerization with elemental S to form polysulfides (after 4 h, leading to the formation of NiS2/Ni3S4), and their dissolution (after 6 h, leading to the formation of NiS/NiS2) with the release of H2S gas.[61]Figure H shows the high-resolution TEM image with lattice fringes with an interplanar spacing of 0.28 and 0.54 nm, corresponding to the (200) lattice planes of NiS2 and the (111) lattice planes of Ni3S4, further confirming the crystal structures of NiS2 and Ni3S4 shown by XRD measurements.[51,64] The high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) image and the corresponding energy-dispersive X-ray (EDX) mapping of NiS2/Ni3S4 shows the distribution of Ni and S throughout the composite structure (Figure I–K). Furthermore, it is worth noting that the quantitative EDX analysis (Ni38S62) is consistent with the XPS and XRD results analyzed above.

Electrochemically Assisted Sulfur Leaching

The electrocatalytic behaviors of NiS2/Ni3S4 were investigated for the OER in 1 mol L–1 KOH on a glassy carbon electrode or carbon paper (CP). Surface reconstruction of the polysulfides takes place in a highly polarized oxidation environment and sulfur could be leached out into the electrolyte. The (oxy)hydroxide/polysulfide catalyst formed by the in situ electrochemical tests (while sulfur was leached out) exhibited higher electrochemical performance than their “initial form” and the hydrothermally synthesized oxides/hydroxides.[45−47] Therefore, before recording linear sweep voltammetry (LSV), the focus was on studying the sulfur leaching process at an ultralow current density of 0.1 mA cm–2 (Figure A). Two platforms were observed during sulfur leaching: at 1.160 V for 2 min and 1.400 V for 21 min, indicating structural evolution from initial NiS2/Ni3S4 to NiOOH via Ni(OH)2 as an intermediate species.[49] As shown in Figure B, the suppression of overpotential from 365 (initial NiS2/Ni3S4) to 339 mV leads to a 7.12% reduction. After sulfur leaching, only the oxidation peak of Ni(OH)2 remains at 1.381 V, indicating that surface sulfur in S-rich NiS was leached out and exhibits no oxidation peaks at 1.313 V. To support this observation, XPS, XRD, Fourier transform infrared (FTIR) spectrometer, and STEM-EDX were used to investigate the structure after sulfur leaching. XRD results (Figure C) show that the sulfur impurity peaks at 20–25° and the peaks of NiS2/Ni3S4 at 35.3 and 45.3° disappeared, indicating phase transformation of polysulfides. In addition, FTIR of NiS2/Ni3S4 after electrochemical oxidation at 1.35 V for 6 h exhibits an emerged peak at 3640 cm–1, attributed to nonhydrogen bonded hydroxide (OH–) in Ni(OH)2, while the peaks around 800–1000 cm–1 for NiS (NiS2, Ni3S4) are weakened, indicating that NiS2/Ni3S4 was partially transformed to Ni(OH)2 at the first oxidation stage (∼1.3 V).[65] Meanwhile, FTIR spectrometry of NiS2/Ni3S4 (Figure S4A) shows an attenuated peak at 3640 cm–1 after 6 h at 1.70 V, which agrees well with the result from Yavuz et al. who held Ni(OH)2 under different voltages and concluded that the decreased peak intensity of Ni(OH)2 was due to its oxidation to NiOOH at 1.4–1.5 V.[66] To detect the product of S leaching in the electrolyte, the anions in the electrolyte were precipitated by barium (Ba2+) ions after the electrochemical test and excluded by hydrochloric and nitric acids for carbonate (CO32–) and then further analyzed by FTIR. The peaks are largely identical to those of the reported barium sulfate (Figure S4B).[67] The XPS results after the sulfur leaching process support this observation with the disappeared peaks at 160–165 eV in the S 2p region (Figure D) and the emerged peak at ∼531.0 eV in the O 1s region of NiS2/Ni3S4, indicating sulfur leaching and the formation of nickel(oxy)hydroxide. In addition, the HAADF images, as well as the corresponding elemental mappings present higher atomic ratios for nickel in the structure and lower ones for sulfur: Ni93S7 (Figure S4C–E), further supporting the leaching of sulfur from NiS2/Ni3S4. Two possible pathways of sulfur leaching have been proposed and presented in Figure E. The S–S bond in the α-S8 impurity (XRD, Figure C) can be oxidized to the S–O bond and then to sulfate ions that are soluble in the electrolyte, as shown in the XPS results (Figure D, S 2p, ∼169.0 eV). The sulfur in NiS2/Ni3S4 can be oxidized to the S impurity (S–S), and then follow the leaching path of the S–S bond to sulfate ions.

Figure 3

(A) Chronopotentiometry curve of NiS2/Ni3S4 recorded at an ultralow current density of 0.1 mA cm–2 for the sulfur leaching process. (B) LSV curves of NiS2/Ni3S4 before and after sulfur leaching in 1 M KOH recorded at 5 mV s–1. (C, D) XRD patterns (C represents the peaks of the carbon substrate), XPS spectra of NiS2/Ni3S4 after sulfur leaching. (E) Schematic illustration of sulfur leaching from NiS2/Ni3S4 and the impurity.

The activity enhancement of NiS2/Ni3S4 obtained by the reaction time of 4 h after sulfur leaching prompted us to investigate whether further improvement in activity was possible when changing the reaction time to 2 or 6 h. The Ni–S-2 h and Ni–S-6 h show an increase in performance after sulfur leaching but are less active than the 4 h reaction (Figure S2). These differences in performance can be explained by both morphologies and phase compositions. The TEM and HRTEM images show that Ni–S-2 h consisted of nanoparticles with a size of 20 ± 5 nm, whereas large aggregates were observed for Ni–S-6 h (Figure S3). Moreover, the phase composition of the catalysts changed with the reaction time: for Ni–S-2 h, the hexagonal NiS (JCPDS: 75-0613)[54] and a small number of cubic NiS2 (JCPDS: 1-0099)[50−54] and Ni3S4 (JCPDS: 00-047-1739) were detected (Figure S5). For Ni–S-6 h, there was a mixture of α-S8 and NiS/NiS2/Ni3S4. These results agree well with the literature, which explains that after the initial coprecipitation of Ni2+ and S2– (2 h), more complicated S-rich polysulfides were formed with longer vulcanization time (4 h), and then some super-rich polysulfides were dissolved in low pH solutions (6 h).[61] Therefore, NiS2/Ni3S4, which consists of more conductive and active polysulfides with the microstructure of small-sized nanocubes, shows better OER performance than Ni–S-2 h with more NiS phase, and Ni–S-6 h with the NiS phase and large particle size.

Half-Cell Performance

To further evaluate the activity of the designed NiS2/Ni3S4, the OER performance compared with commercial Ni/NiO is presented in Figure A,B. The corresponding onset potentials and overpotentials at 10 mA cm–2 (η10) are summarized in Table S1 and Figure C. Initially, the onset potential of NiS2/Ni3S4 (1.554 V) resembled that of Ni/NiO (1.529 V) but improved after sulfur leaching (1.492 V). After 3000 CV cycles, onset potentials of NiS2/Ni3S4 and Ni/NiO decrease to 1.478 and 1.497 V, respectively. Increasing the number of cycles to 8000 and 10 000, we found that the catalytic stability of NiS2/Ni3S4 is superior to that of commercial Ni/NiO with a lower onset potential increase of only 0.021 V (0.031 V for Ni/NiO) from 3000 to 10 000 CVs. Moreover, the η10 of NiS2/Ni3S4 changes from 370 mV (initial) to 305 mV after 3000 CVs and to 341 mV after 10 000 CVs, while for Ni/NiO the trend is from 377 mV (initial) to 339 mV (after 3000 CVs) and 428 mV (after 10 000 CVs). The difference lies in an extra improvement of OER catalytic activity for NiS2/Ni3S4 from initial to that after sulfur leaching, owing to the formation of more active NiS/Ni(OH)2/NiOOH heterostructure and the electrochemical tuning effect during the pretreatment process.[45,48,68−70] The electrochemically active surface area was calculated and shows that the double-layer capacitance (Cdl) value of NiS2/Ni3S4 is 0.422/0.451 mF cm–2 (initially/after sulfur leaching), which is almost as double as that of Ni/NiO for 0.235 mF cm–2 (Figure S6), indicating that NiS2/Ni3S4 has a larger ECSA and exposes more active sites and thus performs better. Electrochemical impedance spectroscopy (EIS) was further employed to evaluate the charge transfer process of catalysts. The charge transfer resistance (Rct) of the initial NiS2/Ni3S4 is higher than that of Ni/NiO and Ni(OH)2 due to the sulfur impurity (Figure S7). After sulfur leaching, the Rct of NiS2/Ni3S4 decreases and becomes lower than that of Ni/NiO and Ni(OH)2. This indicates Ni-rich NiS/Ni(OH)2/NiOOH produced by sulfur leaching can improve the charge transfer efficiency of pristine NiS2/Ni3S4. As can be concluded from XPS, XRD, ECSA, and EIS results, the Ni-rich NiS/Ni(OH)2/NiOOH are more active and conductive than pristine NiS2/Ni3S4.

Figure 4

(A, B) LSV curves recorded at 5 mV s–1 and (C, D) corresponding overpotential at 10 mA cm–2 and Tafel slopes of NiS2/Ni3S4 and commercial Ni/NiO before and after sulfur leaching, 3000, 8000, and 10 000 CVs in 1 M KOH. (E) OER stability of NiS2/Ni3S4 after sulfur leaching, Ni/NiO and Ni(OH)2 at a constant current density of 10 mA cm–2. (F) LSV curves of NiS2/Ni3S4 after sulfur leaching, 40 and 65 h, (G) Ni/NiO and (H) Ni(OH)2 before and after 40 and 65 h in 1 M KOH recorded at 5 mV s–1, and (I) corresponding Tafel slopes of NiS2/Ni3S4, Ni(OH)2, and Ni/NiO.

Tafel plots of NiS2/Ni3S4 and Ni/NiO are summarized in Figure D and show a similar tendency as η10: from 73.5 (initial), 46.0 (after 3000 CVs) to 62.3 mV dec–1 (after 10 000 CVs) for NiS2/Ni3S4 and from 69.0 (initial), 67.2 (after 3000 CVs) to 84.7 mV dec–1 (after 10 000 CVs) for Ni/NiO. Furthermore, there is an additional enhancement in the catalytic kinetics for NiS2/Ni3S4 from initial to post sulfur leaching due to instantaneous oxidation of the surface nickel sulfides to hydroxides, forming a hybrid structure of NiS/Ni(OH)2/NiOOH. Looking more closely at a long-term test from 3000 CV cycles to 10 000 CV cycles, the deterioration in performance with increasing overpotential could be explained by three possible reasons:

Continued sulfur leaching and phase transformation: as confirmed by the XRD, the peak shifts occur but are not fully consistent with the specific phase due to the partially amorphous structure after electrochemical oxidation (Figure S9).[63] Moreover, the XPS supports the formation of more Ni(OH)2/NiOOH species on the surface during the cycling process (Figure S10B), while SEM-EDX mapping of the electrodes confirms that sulfur was continuously leached to a large extent, from “initial (0%)”, “after S leaching (66.7%)”, “after 3000 CVs (80.6%)”, to “after 10 000 CVs (96.6%)” (Figure S10C), suggesting that ca. 20 wt % sulfur within the catalyst surface best promotes OER activity.

Microstructure changes: the microstructure changed with different cycles of cyclic voltammetry, which can be supported by HAADF images and corresponding elemental mappings after 3000 (Figure S11) and 10 000 CVs (Figure S12). Finally, NiS2/Ni3S4 was transformed into nanoparticles.

Catalyst shedding: part of NiS2/Ni3S4 fell off into the electrolyte and caused a direct decrease in catalytic activity, which can be inferred from weakened peaks of S 2p (Organic S from Nafion, 170–175 eV, Figure S10A).

In addition to the excellent electrocatalytic activity, the NiS2/Ni3S4 electrocatalyst also shows remarkable stability with a potential increase of only 11.0 mV after 65 h at 10 mA cm–2 (Figure E). In comparison, 59.9 and 23.8 mV potential increase has been recorded for Ni/NiO and Ni(OH)2, respectively. LSV curves before and after 40 and 65 h are shown in Figure F–H. First, η50 of NiS2/Ni3S4 increases by only ∼2 mV (∼0.53%) before and after 65 h, with an average η50 of 373.7 mV. However, the current density of Ni/NiO and Ni(OH)2 cannot reach 50 mA cm–2 at the beginning, and the η50 of Ni/NiO increases by ∼35 mV (∼9.09%) from “after 40 h” to “after 65 h”, with an average value of 402.5 mV, while with an average value of 400.5 mV for Ni(OH)2, indicating the best activity and stability of NiS2/Ni3S4.

Tafel slopes (Figure I, supported by Figure S13) show that NiS2/Ni3S4 almost kept at an average value of 53.6 mV dec–1, which is much lower and stable than that of Ni/NiO and Ni(OH)2 at average values of 70.5 and 77.5 mV dec–1, indicating faster and stable kinetics. The active sites of NiS2/Ni3S4, Ni/NiO, and Ni(OH)2 are mainly the Ni in NiOOH,[49,63] but the difference lies in the surface morphology and phase composition. With Ni-rich polysulfide, NiS2/Ni3S4 after long-term tests exhibits better stability and performance retention, which is promising as an anodic catalyst in AEMWE single cells.

Recently, the performance improvement of Ni-based catalysts after electrochemical tests was attributed to iron impurities.[71,72] In this sense, the OER performance of NiS2/Ni3S4 in 1 M NaOH was also studied and compared with 1 M KOH (Figure S8), and the corresponding overpotentials (η10) are summarized in Table S2. The NiS2/Ni3S4 catalyst in KOH and NaOH shows almost the same results after sulfur leaching (341 mV in NaOH, 340 mV in KOH) and after 3000 CVs (311 mV in NaOH, 305 mV in KOH), suggesting that the performance improvement is not related to Fe impurities in the electrolyte solution but sulfur leaching and phase transformation effects.

Single-Cell Performance

Inspired by the electrocatalytic activity and stability of NiS2/Ni3S4 toward OER, an alkaline electrolyzer was constructed to investigate its feasibility for practical water splitting. The adopted single-cell configuration is shown in Figure A and is as follows: End plate||current collector, electrolyte channel, and heater||PTFE||Pt/C@C paper||FAA-3-50||Ni–S or Ni/NiO. Two representative cells based on Ni/NiO and NiS2/Ni3S4 and a detailed test protocol are presented in Figures S14 and 15. To be consistent with the half-cell tests, the stabilization process of sulfur leaching (Figure S16A) was also maintained in the single-cell tests with the same potential scan rate of 100 mV s–1. After three times of sulfur leaching, the current density of the NiS2/Ni3S4-based cell (Figure B) increases from 1152 mA cm–2 (initial) to 1424 mA cm–2 (1st), 1539 mA cm–2 (2nd), and 1587 mA cm–2 (3rd) at 2.0 V. This suggests that the leaching of sulfur from S-rich to Ni-rich NiS with the formation of a NiS/Ni(OH)2/NiOOH heterostructure contributes to the improvement of cell performance, which is consistent with the results from the half-cell test. To provide a fair comparison, Ni/NiO-based cells also underwent the sulfur leaching process (Figure S16B). After sulfur leaching three times, the current density of Ni/NiO-based cells remains stable with specific values ranging from 1417 mA cm–2 (initial) to 1419 mA cm–2 (1st), 1421 mA cm–2 (2nd), and 1414 mA cm–2 (3rd) at 2.0 V (Figure S17A,B), indicating that sulfur leaching is the crucial factor for the performance improvement of NiS2/Ni3S4-based cells. Moreover, the EIS results of both cells (Figure S18) are similar to those of the half-cell tests (Figure S7), suggesting that sulfur leaching causes a drop in internal resistance (RΩ) and Rct.

Figure 5

(A). Illustration of the single-cell configuration. (B) Polarization curves of the cell, Pt/C||FAA-3-50||NiS2/Ni3S4, before and after three times of sulfur leaching by a dynamic potential scanning method at 5 mV s–1. (C) Polarization curves after conditioning at 1.7 V for 6 h by a galvanostatic method (5 min step–1), (D) stability at 1000 mA cm–2, (E) polarization curves before and after stability tests, and (F) degradation and stability analysis of Ni/NiO- and NiS2/Ni3S4-based cells.

After 2 h of system temperature stabilization and 0.5 h of open-circuit voltage, both Ni/NiO- and NiS2/Ni3S4-based cells were held at 1.7 V for 6 h for further conditioning. The current density of the Ni/NiO-based cell continuously decreases from 400 to 336 mA cm–2, while the current density of the NiS2/Ni3S4-based cell remains stable at 445 mA cm–2, except for a slight decrease in the first 0.5 h from 469 to 445 mA cm–2, while the current density (Figure S19A, 5 mV s–1) increases from 1587 (after the 3rd sulfur leaching) to 1738 mA cm–2 (after conditioning), indicating better stability under moderate polarization conditions and further S-leaching-induced performance improvement of the NiS2/Ni3S4-based cell. Meanwhile, the faradic efficiency of the initial cell (92.2% to 94.8%) is much lower than that after conditioning at 1.7 V for 6 h (97.4%), indicating that a small amount of current was used to oxidize sulfur to sulfate ions (Figure S19B,C). After conditioning, polarization performance was tested using a galvanostatic method, which is more accurate than dynamic scanning. As shown in Figure C, the NiS2/Ni3S4-based cell exhibits a much higher current density of 1550 mA cm–2 than the Ni/NiO-based cell with only 900 mA cm–2 and most FAA-3-50-based cells in the literature (Table S3), indicating that NiS2/Ni3S4 can also be activated for higher catalytic activity than Ni/NiO in full cells.

To verify the stability of NiS2/Ni3S4 under continuous operation, a more stable membrane “AF1-HNN8-50” was used as the ionic conductor. As shown in Figure E, NiS2/Ni3S4-based cells assembled with AF1-HNN8-50 membranes exhibit a further improved current density of 1800 mA cm–2 at 2.0 V, which is also much higher than that of Ni/NiO-based cells showing 1067 mA cm–2. Then, the long-term test (Figure D) shows that NiS2/Ni3S4-based cells are highly stable with a low voltage increase rate of 0.12 mV h–1, while that of Ni/NiO-based cells is as high as 1.7 mV h–1. Compared with the reported single-cell stability (Figure S20A–C and Table S4), the cells based on Pt/C||AF1-HNN8-50||NiS2/Ni3S4 exhibit one of the lowest “voltage increase rates” under the highest current density of 1000 mA cm–2 for a long duration of 500 h. The current density at 2.0 V (Figure S21A,B) decreases from 1800 mA cm–2 (initial), 1600 mA cm–2 (after 185 h), 1512 mA cm–2 (after 310 h), and 1455 mA cm–2 (after 400 h) to 1400 mA cm–2 (after 500 h), while that of Ni/NiO-based cells decreased from 1067 to 890 mA cm–2 after only 50 h. The EIS shows that the degradation is not caused by the Rct of the electrodes but by the continuous increase of the membrane resistance, as evidenced by the constant semicircle diameter and a slight right shift in EIS curves (Figure S21C) at 1000 mA cm–2 (membrane-resistance-controlled stage, which increases gradually). On the other hand, the degradation of the Ni/NiO-based cells after 50 h is due to an increased Rct, while the membrane resistance remains unchanged (Figure S21D). After opening the cells, it was found that NiS2/Ni3S4 remained on the substrates, and the membrane was brittle due to the high pressure, temperature, and current density (Figure S22A), while Ni/NiO was washed out entirely (Figure S22B). After membrane refreshing (all other conditions remained unchanged), the first intercept with the X-axis and the size of the semicircle in the EIS (Figure S22C) were found to be highly consistent with the initial condition, indicating a restored membrane-dominated internal resistance and a nearly unchanged Rct. After that, the polarization curves after 500 h were retested and compared with the initial curve in Figure E. The current density of NiS2/Ni3S4-based cells at 2.0 V remains stable from 1800 to 1713 mA cm–2 with a high retention rate of 95.2%, indicating that the catalyst is still active and promising for longer-term performance, and only hindered by membrane stability. The current density of Ni/NiO-based cells decreases from 1067 to 890 mA cm–2, exhibiting much lower performance retention. As shown in Figure F, the high stability of NiS2/Ni3S4-based cells can be attributed to suppressed bubble issues due to the refined and redistributed NiS/Ni(OH)2/NiOOH structure with a higher surface area, supported by the SEM images before and after 500 h. The surface morphology of the initial NiS2/Ni3S4@nickel fiber consists of large catalyst/ionomer clusters with a diameter of 1–5 μm (Figure S23A,C), while after 500 h the surface morphology is mainly composed of NiS/Ni(OH)2/NiOOH-based nanosheets with the diameter of 300–500 nm (Figure S23B,D).

Meanwhile, Ni/NiO-based cells suffered from serious bubble issues (also in RDE tests, Figure E), with the cell voltage increasing from 2.00 to 2.14 V (Figure D), while it is only 2.04 V in the polarization curve after 50 h at 1000 mA cm–2 (Figure E). This indicates a reversible voltage increase of 0.1 V caused by bubbles, which would increase the interfacial resistance in the long term and lead to a continuous voltage increase, promoting catalyst aggregation, shedding, and printing onto the membrane (Figure S22B).

Conclusions

A highly stabilized Ni-rich NiS/Ni(OH)2/NiOOH heterostructure was electrochemically derived from S-rich NiS2/Ni3S4 composite nanocubes by a multistep sulfur leaching process, with higher ECSA and conductivity than commercial Ni/NiO and hydrothermally synthesized Ni(OH)2. The morphological, structural, and compositional behaviors of Ni (poly) sulfides before and after OER processes were clarified by STEM, XRD, and XPS. It was proved that the initial S-rich NiS2/Ni3S4 composite nanocubes would be converted to Ni-rich NiS and Ni(OH)2/NiOOH that exhibited higher ECSA and conductivity after sulfur leaching, acting as true intrinsic species for OER. Meanwhile, polysulfides exhibited better electrochemical behaviors during 10 000 CVs at 100 mV s–1 than commercial Ni/NiO and hydrothermally synthesized Ni(OH)2. More importantly, the NiS/Ni(OH)2/NiOOH catalyst exhibited stable thermodynamic (overpotential) and kinetic (Tafel slope) performance during 65 h@10 mA cm–2 in a half-cell and 500 h@1000 mA cm–2 in a flow-mode full cell with negligible degradation, which can be practically applicable as an anodic catalyst for AEMWE. The present work provides a fundamental understanding and a specific approach to better utilize S-rich Ni (poly)sulfides and promotes further development of AEMWE by highly stabilized, Ni-rich, and low-cost anodic electrocatalysts.

Experimental Section

Materials

Chemicals and materials: Nickel chloride (NiCl2), thioacetamide (C2H5NS), and sodium hydroxide (NaOH) were purchased from Sigma-Aldrich and used without further purification. Potassium hydroxide (KOH) was purchased from Merck KGaA (EMSURE). Ni/NiO nanopowder was purchased from Alfa-Aesar. Carbon paper and Ni fiber were purchased from Toray and Bekeart, respectively. FAA-3-50 and AF1-HNN8-50 membranes, and FAA-3-SOLUT-10 and AP1-HNN8-00-X ionomers were supplied by Fumatech and Ionomr Innovations Inc. (Aemion), respectively.

Ni–S Catalyst Synthesis

NiCl2 (3.5 mmol) was dissolved in 60 mL of deionized (DI) water and then mixed with 1 mL of KOH solution (10 mM) under vigorous stirring for 5 min. Thioacetamide was then added slowly with vigorous stirring for 30 min. The mixture was further stirred and kept at 160 ± 2 °C for 2, 4, and 6 h (Figure S1). During this period, evaporated water from the flask was renewed every 30 min. The resulting black precipitates denoted as Ni–S-2 h, Ni–S-4 h, and Ni–S-6 h were filtrated and washed with deionized (D.I.) water and ethanol several times until a clear supernatant was observed. All precipitates were dried in a vacuum oven at room temperature for 24 h.

Ni(OH)2 Catalyst Synthesis

Ni(OH)2 was prepared by direct precipitation of 3.5 mM NiCl2 and 7.0 mM KOH under strong stirring for 4 h at 160 ± 2 °C. Washing and separation processes were the same as that of Ni–S catalysts.

Characterization Studies

The crystal information of Ni–S catalysts was studied by X-ray powder diffraction (D8 DISCOVER, Bruker) with a Cu Kα target. The morphology, elemental distribution, and detailed structural information were studied by scanning transmission electron microscopy (STEM), and energy-dispersive X-ray (EDX) spectroscopy on a Titan 80-200 electron microscope (Thermo Fisher Scientific) with a probe corrector (CEOS) and a high-angle annular dark-field (HAADF) detector, and high-resolution transmission electron microscopy (HRTEM) on a Titan 80-300 electron microscope (Thermo Fisher Scientific).[73,74] The chemical valence states of Ni–S samples were studied by X-ray photoelectron spectroscopy (XPS, Phi5000 VersaProbeII, ULVAC-Phi Inc) with Al Kα as the monochromatic (1.486 keV) source.[10] The catalysts for XRD and XPS were coated on the carbon paper (ca. 2.0 mg cm–2) by the drop-by-drop method with the same ink for RDE. After electrochemical measurements, the catalyst was removed by ultrasonic separation from RDE for TEM/HRTEM/STEM. The detection of structural changes in NiS2/Ni3S4 before and after electrochemical tests, and the proof of sulfate ions in the electrolyte were achieved by Fourier transform infrared spectrometry (FTIR, Monolithic diamond GladiATR, PIKE Technologies). The samples were prepared as follows: (i) the NiS2/Ni3S4 catalyst was deposited on a graphite substrate with Nafion as a binder; (ii) two of the above electrodes were used as working electrodes and kept at 1.35 and 1.7 V for 6 h, respectively; (iii) the catalyst on the electrodes was scraped off after the OER tests and cleaned with ethanol and D.I. water. Analysis steps for sulfate ions in the electrolyte after electrochemical tests were conducted as follows: First, a NiS2/Ni3S4-coated nickel fiber with an area of 10 cm2 and mass loading of 20 mg cm–2 was used as the working electrode and kept at 1.7 V for 1 h in 150 mL of 1 M KOH. Then, excessive barium chloride (BaCl2) was added into the above electrolyte and the pH of the mixture was further adjusted to “strong acid pH (< 1)” using hydrochloric acid to remove BaCO3, etc. And then, the resulting precipitate was collected and washed repeatedly with dilute nitric acid three times to further remove other acid-soluble precipitates. After the above steps, the obtained white particles were only to be BaSO4 or AgCl. Finally, fully dried powder was analyzed by FTIR.

Electrochemical Measurements

A glassy carbon (GC) electrode with a geometric area of ca. 0.19625 cm2 was polished with an Al2O3 slurry. Eight milligrams of catalyst powder was dispersed with 0.5 mL of DI water, 1.5 mL of IPA, and 20 μL of 5 wt % Nafion solution to form a homogeneous ink suspension. Ink suspension (10 μL) was dropped onto the GC electrode with a mass loading of 0.2 mg cm–2. A rotating disk electrode (RDE, Pine Research Instrumentation) system with an electrochemical workstation (VSP-150, BioLogic Sciences Instruments) was used for the electrochemical tests. The three-electrode system was composed of the electrolyte (200 mL of 1 M KOH), working electrode (catalyst-coated GC), counter electrode (platinum wire), and reference electrode (Hg/HgO). The obtained potential was corrected by the gap between Hg/HgO and RHE (0.926 V) and 85% of the internal resistance (IR) loss compensation according to the equation: EiR-corrected = E + 0.926 – 0.85 × iRs (where Rs represents the solution resistance tested by high-frequency AC impedance from 1 to 106 Hz with an amplitude of 0.005 V).[75] OER catalytic activity was checked by the linear sweep voltammetry (LSV) method. For the results “before sulfur leaching”, Ni–S catalysts were directly tested by LSV (1.0–1.7 V, 5 mV s–1) without any pretreatment under oxygen-saturated conditions. For the results “after sulfur leaching”, the cyclic voltammetry (CV) method (1.0–1.7 V, 100 mV s–1) was utilized to activate Ni–S by sulfur leaching: 10 cycles first and LSV were compared with that of “before sulfur leaching”, and then the electrolyte was refreshed. After that, 3–4 times 10-cycle CV pretreatment was repeated until there was no obvious change of LSV curves. For the results “after CVs”, the conditions (1.0–1.7 V, 100 mV s–1) were the same as those of “CVs pretreatment”. For Ni/NiO and Ni(OH)2, there were also 3–4 times 10-cycle CV pretreatment (denoted as “initial”) and all conditions of CVs were the same as those of Ni–S catalysts. OER stability was checked by the chronopotentiometry method at 10 mA cm–2. The LSV curves were recorded before and after 40 and 65 h stability tests at 5 mV s–1.

Single-Cell Configuration

Commercial Ni/NiO and the prepared NiS2/Ni3S4 were used as anodic catalysts coated on the Ni fiber (area: 5 cm2, thickness: 500 μm) with the mass loading of 5 mg cm–2, while Pt (wt 60%)/C was used as the cathodic catalyst coated on carbon paper (area: 5 cm2, thickness: 300 μm) with a mass loading of 0.8 mg cm–2. The catalyst-coated substrate (CCS) structure was achieved by the Sono-Tek ultrasonic spraying system. FAA-3-SOLUT-10 and AP1-HNN8–00-X were used as the ionomers in the corresponding FAA-3-50 and AF1-HNN8-50-based cells, which accounted for 20 wt % in NiS2/Ni3S4 and 25 wt % in Pt/C inks. Two liters of 1 M KOH without further purification was used as the electrolyte and would be automatically compensated with D.I. water by a water-level sensor to keep the alkali concentration stable during continuous water electrolysis. FAA-3-50 and AF1-HNN8-50 membranes were used as the ionic conductors, which were immersed in 1 M KOH for 12 h before use. All of the tests were conducted at 60 ± 1 °C with an electrolyte feeding rate of 50 mL min–1. To ensure repeatability, parallel cells were tested simultaneously and the torque used for assembling the cell was first fixed at 5.0 N·m and then at 10.0 N·m.

Single-Cell Testing Protocol

Single-cell measurements were performed with a potentiostat/galvanostatic setup (BioLogic, BCS-815). The protocol of single-cell testing is illustrated in Figure S15. First, it took 2 h to stabilize the temperature and flow rate of the electrolyte to 60 ± 1 °C and 50 mL min–1, respectively. Second, sulfur leaching (10 CV cycles between 1.2 and 2.0 V, @100 mV s–1), EIS (@200 mA cm–2, from 10 kHz to 0.1 Hz, with an amplitude of 10 mA), and LSV (between 1.2 and 2.0 V, @5 mV s–1) were performed three times for ∼0.5 h, followed by 0.5 h open-circuit voltage (OCV) and 6 h of conditioning at 1.7 V. Then, polarization curves were recorded by a galvanostatic method (@5 min step–1, totally ∼2 h). After that, stability tests were launched at 1000 mA cm–2 for 500 h. Finally, the cells were disassembled and reassembled with a new pretreated AF1-HNN8-50 membrane and restarted with polarization curves.