Highly crystalline Ni-doped FeP/carbon hollow nanorods as all-pH efficient and durable hydrogen evolving electrocatalysts.

Herein, we report the synthesis of uniform hollow nanorods of Ni-doped FeP nanocrystals hybridized with carbon as electrocataysts for the electrocatalytic hydrogen evolution reaction (HER). These hollow nanorods are prepared based on the etching and coordination reaction between metal-organic frameworks and phytic acid, followed by a pyrolysis process. Benefiting from the abundant active sites, the improved mass and charge transport capability, the optimized Ni-doped FeP/C hollow nanorods exhibit excellent HER activities for achieving a current density of 10 mA cm-2 at an overpotential of 72, 117, and 95 mV in acidic, neutral, and alkaline media, respectively, as well as superior stability. X-ray photoelectron spectroscopy and basic density functional theory calculations suggest that the improved HER activity originates from the synergistic modulation of the active components, structural and electronic properties. This protocol provides a general and friendly strategy to construct hollow phosphides for energy-related applications.

INTRODUCTION

Hydrogenpan> generation through water splitting is considered as a green and efficient way to meet the increasing demands for renewable and clean energy sources (, ). As two important components of electrocatalytic water splitting, catalysts and electrolytes determine the energy transfer efficiency of hydrogen evolution reaction (HER) (, ). Platinum-based HER electrocatalysts are the best candidates because of their high activity and low Tafel slope, but prohibitive cost and scarcity greatly restrict their wide application (). In the past decade, intensive research efforts have been devoted to developing nonprecious electrocatalysts, including metal alloys (), oxides (, ), carbides (), sulfides (), nitrides (), selenides (), and nonmetallic electrocatalysts (). However, very few of them show both high activity and stability for HER at all pH values. Therefore, it is of high importance to develop earth-abundant electrocatalysts for HER with high activity and stability that can function efficiently over the full pH range.

Recently, trapan class="Chemical">nsition metal phosphides (TMPs) have attracted extensive interest in view of their low cost, nontoxicity, good chemical stability, and catalytic activity toward HER in electrolytes with a wide range of pH values (–). Iron phosphide (FeP) is a good conductor of heat and electricity and also has high thermal and chemical stability (). Thus, it is better to access the active corner and edge sites on the crystallite surface, making it uniquely advantageous as an electrocatalytic material. Recently, FeP has been reported much for HER in acidic media with different morphologies, such as nanoparticles (), nanosheets (), microcubes (), nanowires (), and nanorods (). However, few of them show comparable activity to Pt-based electrocatalysts and, even fewer show pH-universal activity and stability. One important factor hindering further improvement of catalytic activity and stability for FeP-based electrocatalysts is their low-porosity structures or the high adhesion between the surface of the catalyst and the as-formed hydrogen bubbles ().

One popular strategy for efficiepan class="Chemical">nt HER activity is to design and construct hollow micro/nanostructured hybrid electrocatalysts with a large npan>umber of exposed reactive active sites and reduced diffusion length for both mass and charge transport (, ). Recently, metal-organic frameworks (MOFs) have been regarded as very promising platforms to obtain various micro/nanostructures, especially hollow nanostructures, because of their inherently high specific surface areas, abundant metal/organic species, and extraordinary tunability of structures and compositions (–). Besides, regulating the surface electronic structure and optimizing the hydrogen adsorption energy by doping of cations or anions are also important and effective strategies to improve the intrinsic activity of catalysts (, ). With its strong coordination ability and high phosphorus content, phytic acid with a molecular formula of C6H6(H2PO4)6 is chosen recently as the phosphorus source to prepare phosphides (). Motivated by the above considerations, we set out to design and synthesize Ni-doped FeP/C hollow nanorods using MIL-88A as the precursor and phytic acid as the etchant and phosphorus source (Fig. 1). After subsequent annealing treatment, the derived Ni-doped FeP/C material perfectly retains the structure of hollow nanorods. As a pH-universal HER electrocatalyst, the as-prepared Ni-doped FeP/C hollow nanorods show high activity and superior stability in solutions with a full pH range.

Fig. 1

Schematic illustration of the formation of Ni-doped FeP/C hollow nanorods.

RESULTS

Materials synthesis and characterization

The morphology and structure of the as-prepared papan class="Chemical">n class="Chemical">Ni-doped npan> class="Chemical">MIL-88A are characterized by field emission scanning electron microscopy (FESEM) and transmission electron microscopy (TEM). The FESEM and TEM images show that the Ni-doped MIL-88A samples with different Ni contents (denoted as NM-1, NM-2, and NM-3 for Ni/Fe atomic ratios of 1:3, 2:2, and 3:1 in the reactants, respectively) have similar highly uniform hexagonal rod-like solid nanostructures (Fig. 2, A to H). With the increase of doping amount, the average length of these hexagonal nanorods decreases from 3.2 to 1.6 μm, while the average width grows from 400 to 600 nm (fig. S1). X-ray powder diffraction (XRD) patterns (Fig. 2I) and Fourier transform infrared (FTIR) spectra (Fig. 2J) show that the doping process does not bring too much change in the crystal structure and composition of MIL-88A (). With the increasing amount of nickel doping, most peaks disappear and only weak (101) and (002) crystallographic facets are reserved, implying the change of growth rates and directions (). The results of the nitrogen sorption isotherms (Fig. 2K and fig. S2) show that the specific surface area increases with the increasing usage amount of nickel nitrate until the Ni/Fe atomic ratio in the reactant reaches 1:1. However, the specific surface area of NM-3 with a higher atomic ratio of Ni/Fe is smaller than that of NM-2. This may be due to the structural distortion of MOFs in the weakly acidic environment produced by the hydrolysis of remaining nickel ions. Besides, iron concentration and pH value also influence the particle size (). After being treated in phytic acid solution (0.1 M) at 90°C, solid inners of the MIL-88A and Ni-doped MIL-88A samples are etched out to generate hollow nanostructures (Fig. 3). These phytic acid–treated MIL-88A (PTM) and Ni-doped MIL-88A (PTNM-1, PTNM-2, and PTNM-3 derived from NM-1, NM-2, and NM-3, respectively) hollow samples generally retain the overall rod-like appearance, but their aspect ratios become smaller after this etching and coordination reaction. When the reaction occurs at different temperatures, the nanorods could be transformed into well-dispersed hollow particles with core-shell/yolk-shell configurations or agglomeration of small hollow nanoparticles (fig. S3). XRD patterns indicate the phase transformation during the coordination reaction between iron ions and phytic acid (fig. S4A). The changes in FTIR peaks provide the evidence of the formation of metal-phytic compounds (fig. S4B) ().

Fig. 2

The influence of doping amount on the morphology and structure of MIL-88A.

(A to D) FESEM images and (E to H) TEM images of as-prepared (A and E) MIL-88A, (B and F) NM-1, (C and G) NM-2, and (D and H) NM-3. (I) XRD patterns, (J) FTIR spectra, and (K) nitrogen sorption isotherms of MIL-88A, NM-1, NM-2, and NM-3. a.u., arbitrary units.

Fig. 3

Morphology and structure characterizations of MIL-88A treated with phytic acid.

(A to D) FESEM images and (E to H) TEM images of (A and E) PTM, (B and F) PTNM-1, (C and G) PTNM-2, and (D and H) PTNM-3 hollow nanorods.

(A to D) FEpan>SEM images and (E to H) TEM images of as-prepared (A and E) MIL-88A, (B and F) NM-1, (C and G) NM-2, and (D and H) NM-3. (I) XRD patterns, (J) FTIR spectra, and (K) nitrogen sorption isotherms of MIL-88A, NM-1, NM-2, and NM-3. a.u., arbitrary units.

(A to D) n class="Chemical">FEpan>SEM images and (E to H) TEM images of (A and E) PTM, (B and F) PTNM-1, (C and G) n class="Chemical">PTNM-2, and (D and H) PTNM-3 hollow nanorods.

After a subsequent papan class="Chemical">n class="Chemical">carbonizationpan> treatment, all the diffraction peaks in the XRD pattern (Fig. 4A) of the representative Ni-doped FeP/C sample derived from PTNM-3 (denoted as NFP/C-3) is consistent with an FeP standard pattern (JCPDS card no. 65-2595). No extraneous characteristic peaks of Ni- or Fe-based materials are observed, indicating the high purity of products and the successful doping of Ni species into the lattice of FeP (). FESEM and TEM images show that the resultant NFP/C-3 sample (Fig. 4, B and C) preserves the uniform rod-like morphology with a well-defined hollow interior. Similarly, the FeP/C sample derived from PTM and other Ni-doped FeP/C samples (NFP/C-1 derived from PTNM-1 and NFP/C-2 derived from PTNM-2) also inherit the structural features from their hollow precursors (fig. S5). A high-resolution TEM (HRTEM) image (Fig. 4D) shows that the Ni-doped FeP nanocrystals are surrounded by continuous carbon species, which is beneficial for the stability of catalysts (). A magnified TEM image (Fig. 4E) and a corresponding line scan marked with a blue rectangle (Fig. 4F) show the lattice fringe with a spacing distance of 0.29 nm, which corresponds to the (002) planes of FeP. Elemental mapping images (Fig. 4G) show the homogenous distribution of C, P, Fe, and Ni elements in the hollow nanorod. The corresponding energy-dispersive x-ray (EDX) spectrum also confirms the existence of C, P, Fe, and Ni elements with low atomic percentage of Ni species (fig. S6). The atomic ratio of Ni/Fe for NFP/C-3 from the EDX is consistent with the results of inductively coupled plasma atomic emission spectroscopy with a value of 4/235. More information about carbon species in the resultant phosphides is provided by the Raman spectra (fig. S7) (, ). The analogical locations and peak intensity ratio of D and G bands (located at 1335 and 1596 cm−1) for FeP/C and three NFP/C samples indicate that the Ni ions should have been dominantly incorporated into the FeP lattice instead of carbon. Nitrogen adsorption/desorption isotherms (fig. S8) show that NFP/C-3 has the largest specific surface area of 72 m2 g−1 with an average pore size of about 3 nm. The pore size distributions of the four phosphide samples appear to be very similar, which may also suggest that nickel atoms are doped into the lattice of FeP. The mesoporous structure facilitates to the electroactive surface, which is beneficial for enhancing HER performance ().

Fig. 4

Morphology and structure characterizations of NFP/C-3 hollow nanorods.

(A) XRD pattern, (B) FESEM image, (C) TEM image, (D and E) HRTEM images, (F) line scan of the HRTEM image indicated by the blue rectangle in (E), and (G) elemental mapping.

(A) XRD pattern, (B) papan class="Chemical">n class="Chemical">FESEM image, (C) TEM image, (D anpan>d E) HRTEM images, (F) line scan of the HRTEM image indicated by the blue rectangle in (E), and (G) elemental mapping.

Electrochemical performances

To investigate the dopipan class="Chemical">ng effect for HER, we first compare the catalytic performances of bare FeP/C and Ni-doped FeP/C hollow nanorods in acidic medium (0.5 M H2SO4 solution, pH ≈ 0). The used reference electrode is calibrated before tests (fig. S9). As shown in linear sweep voltammetry (LSV) curves (Fig. 5A), with the increase of doping amount, the overpotential required to reach the current density of 10 mA cm−2 becomes lower. The specific comparison of electrocatalytic performance of the four catalysts in 0.5 M H2SO4 solution is listed in table S1. The overpotential at 10 mA cm−2 of NFP/C-3 hollow nanorods is only 72 mV, which is not only much smaller than that of the FeP/C sample (157 mV) but also comparable to those of many other transition metal–based phosphides (table S2), implying superior electrocatalytic activity. Moreover, the NFP/C-3 hollow nanorods show the smallest onset overpotential of ~22 mV versus reversible hydrogen electrode (RHE) (fig. S10). Considering the negligible activity contribution from a carbon fiber paper (CFP) substrate (fig. S11), we attribute the improvement in electrocatalytic performance to the doping effect. Besides, we have also prepared NFP/C-3 with low crystallinity and NFP-3 without carbon as control samples. NFP/C-3 with low crystallinity can only be obtained by annealing PTNM-3 at 800°C for a shorter period (fig. S12). All the electrochemical data including the overpotential at 10 mA cm−2, Tafel slope, stability, and charge transfer resistance (Rct) show the superiority of NFP/C-3 with high crystallinity (fig. S13). For comparison, a sample named NFP-3 without carbon is prepared by annealing the phytic acid–treated oxide sample derived from NM-3. Structural analyses (fig. S14) show that the structure of NFP-3 without carbon is destroyed severely. Degradations of electrocatalytic activity and stability (fig. S15) highlight the importance of carbon for structural integrity and catalytic performance.

Fig. 5

Electrochemical performances of electrocatalysts measured in acidic, neutral, and alkaline solutions.

(A) LSV curves, (B) corresponding Tafel slopes, and (C) the capacitive current density ΔJ0.25 V as a function of scan rate in the range of 0.2 to 0.3 V versus RHE in 0.5 M H2SO4 of FeP/C, NFP/C-1, NFP/C-2, and NFP/C-3. LSV curves of FeP/C and NFP/C-3 in (D) 1.0 M PBS and (E) 1.0 M KOH solutions (insets: Tafel slopes). (F) Chronopotentiometry curves at a constant current density of 10 mA cm−2 for 12 hours of NFP/C-3 in 0.5 M H2SO4, 1.0 M PBS, and 1.0 M KOH solutions.

(A) LSVpan> curves, (B) corresponding Tafel slopes, and (C) the capacitive current density ΔJ0.25 V as a function of scan rate in the range of 0.2 to 0.3 V versus RHE in 0.5 M H2SO4 of FeP/C, NFP/C-1, NFP/C-2, and NFP/C-3. LSV curves of FeP/C and NFP/C-3 in (D) 1.0 M PBS and (E) 1.0 M KOH solutions (insets: Tafel slopes). (F) Chronopotentiometry curves at a constant current density of 10 mA cm−2 for 12 hours of NFP/C-3 in 0.5 M H2SO4, 1.0 M PBS, and 1.0 M KOH solutions.

To further study the underlyipan class="Chemical">ng electrocatalytic mechanism for HER, the corresponding Tanpan> class="Chemical">fel plots based on the LSV curves are presented in Fig. 5B. These linear regions of Tafel plots are fitted to the Tafel equation (η = b log j + a, where b is the Tafel slope), yielding Tafel slopes of 127, 87, 81, and 54 mV decade−1 for FeP/C, NFP/C-1, NFP/C-2, and NFP/C-3 hollow nanorods, respectively. The Tafel slope of NFP/C-3 hollow nanorods indicates that HER occurs via the Volmer-Heyrovsky mechanism and the electrochemical recombination with an additional proton is the rate-limiting step (, ). The exchange current density (J0) for each sample is obtained by extrapolation of the corresponding Tafel plot (fig. S16). The J0 of NFP/C-3 hollow nanorods is ~0.723 mA cm−2, which outperforms that of the FeP/C sample (0.594 mA cm−2) and those of many transition metal–based phosphides (table S2). The electrochemically active surface area (ECSA) of catalysts is further evaluated through electrochemical double-layer capacitance (fig. S17). Herein, NFP/C-3 hollow nanorods exhibit a capacitance of 80.2 mF cm−2, which is much larger than that of FeP/C hollow nanorods (Fig. 5C), indicating more exposed electrochemically active sites. Electrochemical impedance spectroscopy (EIS) data (fig. S18) also show the reduced solution resistance (Rs) and Rct for the Ni-doped samples due to the improvement in hydrophilicity and conductivity caused by the dopant (). The results of the hydrophilic test further indicate the improvement in hydrophilicity with a smaller contact angle (fig. S19).

It is noteworthy that pan class="Chemical">numerous microbial and strongly alkaline electrolysis cells need active and stable electrocatalysts in neutral and basic conditions (). These Ni-doped FeP/C hollow nanorods show high HER activities with an onset overpotential of ~55 mV (fig. S20A) and a small overpotential of 117 mV at 10 mA cm−2 (Fig. 5D) in a neutral solution [1.0 M phosphate-buffered saline (PBS) solution, pH ≈ 7]. Compared with the FeP/C sample, the corresponding smaller Tafel slope (inset in Fig. 5D) and higher exchange current density of the doped hollow nanorods (fig. S20B) indicate the favorable dynamic process in neutral environment. Moreover, NFP/C-3 hollow nanorods reveal high electrocatalytic performance in a strongly alkaline solution (1.0 M KOH solution, pH ≈ 14). Once again, the doped hollow nanorods demonstrate better HER activities over the FeP/C sample with smaller overpotential and Tafel slope (Fig. 5E). In addition, these Ni-doped FeP/C hollow nanorods show enhanced properties with smaller onset overpotential and higher exchange current density (fig. S20, C and D). The activities of NFP/C-3 are also comparable to those of many reported nonprecious metal–based HER electrocatalysts in neutral and basic conditions (tables S3). Chronopotentiometric response measurement shows that the NFP/C-3 hollow nanorods stably run over 12 hours at a current density of 10 mA cm−2 without obvious amplification in overpotential in acidic, neutral, and alkaline solutions (Fig. 5F). The LSV curves for NFP/C-3 hollow nanorods before and after the stability test also indicate superior stability. We further examine its ECSA and structural and compositional changes after the stability test in an acidic solution as an example. The subtle variations further confirm its high structural and chemical stability (fig. S22). The above observations confirm that these Ni-doped FeP/C hollow nanorods are highly active and stable HER electrocatalysts over the full pH range.

DISCUSSION

To validate the generality of this sypan class="Chemical">nthesis strategy for TMPs, Co-glycerate nanospheres, Ni-glycerate nanospheres, Cu-MOF ([Cu3(btc)2]) polyhedrons, and ZIF-67 dodecahedrons are prepared as the templating precursors for synthesis of corresponding TMPs. TEM images show the similar morphology evolution process from solid precursors to hollow phosphides (fig. S23). XRD patterns (fig. S24) confirm the phase purity for each phosphide, including CoP (JCPDS card no. 29-0497), Ni2P (JCPDS card no. 03-0953), and Cu3P (JCPDS card no. 02-1263). Despite the dissimilarity in shape and composition, all these precursors can react with phytic acid and finally evolve into corresponding phosphides.

To provide deeper insight ipan class="Chemical">nto the catalytic contributionpan> from nickel doping at the atomic level, x-ray photoelectron spectroscopy (XPS) and spin-polarized density functional theory (DFT) calculations are performed. XPS analyses of the Fe 2p (Fig. 6, A and B) of FeP/C and NFP/C-3 show two different chemical states with two additional satellite peaks. One is associated with FeP at lower binding energy and the other is Fe2+ or Fe3+ chemical environment at higher binding energy (, ). The ratio between the two iron chemical states (Fe2+/3+/FeP) is 4.69 for FeP/C and 1.65 for NFP/C-3. For P 2p (Fig. 6, C and D), a pair of peaks at lower binding energy matches well with that of metal phosphide. A second pair of peaks at higher binding energy is assigned to a phosphate environment (). The ratio between the two phosphorus chemical states (P-O/Fe-P) is 5.26 for FeP/C and 1.51 for NFP/C-3. Moreover, the weak signal of Ni 2p implies the presence of trace amount of nickel species (fig. S25). From the XPS analysis, it could be concluded that (i) the outermost surface of catalysts is oxidized to phosphates, (ii) the negative shift for Fe 2p and the positive shift for P 2p indicate the electron interactions with the transfer from P atoms to Fe atoms, and (iii) nickel doping can significantly increase the relative content of electrocatalytically active phosphides.

Fig. 6

XPS and DFT calculation results about the promotion of dopant on the catalytic performances.

High-resolution XPS spectra of (A and B) Fe 2p and (C and D) P 2p for FeP/C and NFP/C-3; (E) ΔGH* as a function of θH for Pt, FeP, and Ni-doped FeP. (F to H) DFT-calculated electronic structures. (F) the charge density difference on Ni-doped FeP: ρ = ρ[Ni-FeP] + ρ(Fe) − ρ[Ni] − ρ[FeP]; yellow and cyan isosurfaces represent electron accumulation and electron depletion, respectively. (G) Section line and (H) isosurface two-dimensional view.

High-resolution XPS spectra of (A apan class="Chemical">nd B) Fe 2p and (C anpan>d D) P 2p for FeP/C and NFP/C-3; (E) ΔGH* as a function of θH for Pt, FeP, and Ni-doped FeP. (F to H) DFT-calculated electronic structures. (F) the charge density difference on Ni-doped FeP: ρ = ρ[Ni-FeP] + ρ(Fe) − ρ[Ni] − ρ[FeP]; yellow and cyan isosurfaces represent electron accumulation and electron depletion, respectively. (G) Section line and (H) isosurface two-dimensional view.

We find that the trepan class="Chemical">nds in binding strengths are generally preserved for coverages between the lowest hydrogen coverage rate (θH) and the coverage where HER occurs (junction point between solid lines and dotted lines in Fig. 6E) (). For Pt, the best-known HER electrocatalyst, its hydrogen adsorption free energy (ΔGH*) changes little within a moderate range of −0.13 to 0.02 eV regardless of the alternation of θH. While for FeP, when the θH is below 50%, the ΔGH* is very low, implying strong bonding between adsorbed hydrogen and the catalyst. As a result, active sites are blocked to reduce HER efficiency. When the θH exceeds 50%, its ΔGH* becomes much lower again to suppress HER performance, which is consistent with the phenomenon reported by Jaramillo and co-workers (). After nickel doping, the ΔGH* becomes higher, especially at a low θH of 25%, from −0.47 to −0.2 eV, implying the transfer of electrons from P atoms to Fe atoms, which is consistent with the XPS results (Fig. 6, A to D). Besides, when the θH is up to 75%, the value of ΔGH* for Ni-doped FeP becomes closer to zero instead of lower for pure FeP, suggesting the more favorable HER process. This improvement of the kinetic process is also reflected by the decrease of the Tafel slope from 127 mV decade−1 for FeP/C to 54 mV decade−1 for NFP/C-3. The optimized structures for the above calculations can be seen in figs. S26 to S33. To further explore the principle of improving the material properties by nickel doping, the differential charge distribution calculation is performed. A large number of electron holes are generated around the Ni atom, indicating the occurrence of charge transfer (Fig. 6, F to H). The transfer of electrons from Ni atoms to P atoms increases the electron density around P atoms, resulting in weakened P–H bonding. Compared with pure FeP, the transfer of electrons leads to the lower binding energy of Fe 2p in Ni-doped FeP and stronger antioxidant capability. Meanwhile, the negative electrification of Fe atoms prevents the huge decrease of ΔGH* with the increase of θH. The weakened hydrogen adsorption and P–H bond both make H atoms more easily desorbed and more favorable for HER ().

In summary, we presepan class="Chemical">nt a facile route to synthesize Ni-doped FeP/C hollow nanorods with tunable aspect ratios based on the etching and coordination reaction between MOFs and phytic acid. Owing to the synergistic modulation of the component and the structural and electronic properties by the cationic dopant, the as-prepared hollow nanorods of Ni-doped FeP nanocrystals hybridized with carbon exhibit high electrocatalytic activities and superior stability for hydrogen evolution over the full pH range. The present strategy provides a general and effective route to design highly efficient pH-universal phosphide-based electrocatalysts with tunable chemical compositions and structures.

MATERIALS AND METHODS

Synthesis

Synthesis of MIL-88A hexagonal nanorods and Ni-doped MIL-88A hexagonal nanorods

Accopan>rding to the literature method (), 1.2 mmol of fumaric acid was first dissolved in 25 ml of deionized (DI) water and stirred at 400 rpm at 70°C for 10 min. Then, 1.3 mmol of Fe(NO3)3·9H2O was added into the solution and stirred for another 10 min. Last, the solution was transferred to a Teflon-lined stainless steel autoclave (40 ml in total capacity) and kept at 110°C for 6 hours. After cooling down to room temperature, the obtained MIL-88A hexagonal nanorods were separated by centrifugation, washed with DI water and ethanol, and finally dried in a vacuum drying oven. Ni-doped MIL-88A hexagonal nanorods were synthesized by the same method used for MIL-88A except that additional Ni(NO3)2·6H2O was added into the solution. Ni-doped MIL-88A samples with different Ni contents are denoted as NM-1, NM-2, and NM-3 for Ni/Fe atomic ratios of 1:3, 2:2, and 3:1 in the reactants, respectively.

Synthesis of M-glycerate (M = Co, Ni) sphere precursors

Accopan>rding to our previously reported literature with slight modification (), 0.108 g of cobalt nitrate hexahydrate and 8 ml of glycerol were dissolved into 40 ml of isopropanol, and after stirring for 40 min at 800 rpm, the as-prepared solution was then transferred to a Teflon-lined stainless steel autoclave and kept at 180°C for 6 hours. For Ni-glycerate, 0.145 g of nickel nitrate hexahydrate and 7.5 ml of glycerol were dissolved into 52.5 ml of isopropanol, and then 1 ml of water was dropped into the solution. After stirring for 40 min at 800 rpm, the as-prepared solution was then transferred to a Teflon-lined stainless steel autoclave and kept at 200°C for 10 hours. The product was isolated using centrifugation, washed five times with ethanol, and dried at 80°C.

Synthesis of [Cu3(btc)2]n polyhedron precursors

Accopan>rding to a previously reported literature with slight modification (), 41.0 mg of copper nitrate trihydrate and 1.43 g of lauric acid were dissolved in 10 ml of butanol. The mixed solution was stirred at 400 rpm at 70°C until a transparent solution was obtained. Then, 20 mg of benzene-1,3,5-tricarboxylic acid was added and continuously stirred for another 10 min. The as-prepared solution was then transferred to a Teflon-lined stainless steel autoclave and kept at 140°C for 4 hours. The resulting blue powder was isolated using centrifugation, washed five times with ethanol, and dried at 80°C.

Synthesis of ZIF-67 polyhedron precursors

Accopan>rding to a previously reported literature with slight modification (), a solution of cobalt acetate tetrahydrate (0.6 g in 5 ml of water) was added into a solution of 2-methylimidazole (2.24 g in 5 ml of water). The resulting mixture was aged at room temperature for 5 hours. Last, the purple precipitate was collected by centrifugation, washed three times with methanol, and dried for further use.

Synthesis of PTM and Ni-doped MIL-88A hollow samples

One hupan class="Chemical">ndred milligrams of as-prepared MIL-88A or Ni-doped MIL-88A was dispersed in 10 ml of DI water. After stirring for 10 min, 15 ml of phytic acid solution (0.1 M) was added into the above solutions; the solution was kept at 90°C for 3 hours. The acid-treated products were separated by centrifugation, washed with DI water and ethanol, and dried in a vacuum drying oven. The phytic acid–treated MIL-88A is denoted as PTM. The phytic acid–treated NM-1, NM-2, and NM-3 are denoted as PTNM-1, PTNM-2, and PTNM-3, respectively.

Synthesis of phytic acid–treated Co-glycerate, Ni-glycerate, Cu-MOF, and ZIF-67 precursors

One hupan class="Chemical">ndred milligrams of as-prepared Co-glycerate, Ni-glycerate, Cu-MOF, and ZIF-67 were dispersed in 15 ml of ethanol separately. After stirring for 10 min, 10 ml of phytic acid ethanol solution (0.1 M) was added into the above solutions and kept at room temperature for different durations (3 hours for Co-glycerate and Ni-glycerate and 30 min for Cu-MOF). Then, the acid-treated products were separated by centrifugation, washed with ethanol, and dried in a vacuum drying oven.

Synthesis of FeP/C and Ni-doped FeP/C hollow nanorods

The as-synthesized papan class="Chemical">n class="Chemical">PTM anpan>d PTNM nanorods were transferred into a tube furnace and carbonized in a flowing H2 (5%)/Ar atmosphere. Specifically, the materials were first heated from room temperature to 200°C with a ramp rate of 1°C min−1 and kept for 2 hours. Then, the temperature was further increased to 800°C with the same rate and kept for another 2 hours.

Synthesis of hollow CoP nanospheres, core-shell Ni2P nanospheres, shattered Cu3P/C nanoparticles, and hollow CoP/C polyhedrons

The as-synthesized papan class="Chemical">n class="Chemical">phytic acid–treated npan> class="Chemical">Co-glycerate, Ni-glycerate, Cu-MOF, and ZIF-67 were transferred into a tube furnace and carbonized in a flowing H2 (5%)/Ar atmosphere with the same calcination procedure and time for NFP/C-3.

Synthesis of NFP/C-3 with low crystallinity and without carbon

The Npan>FP/C-3 with low crystallinity was prepared by the same method for NFP/C-3 except the annealing time, which was reduced from 2 hours to 0.5 and 1.0 hour. To prepare NFP-3 without carbon, PTNM-3 (100 mg) was calcined in air at 600°C for 3 hours with a ramp rate of 1°C min−1 to consume carbon completely. Then, the obtained sample was immersed into 10 ml of phytic acid ethanol solution (0.1 M) and kept for 3 hours at room temperature. Afterward, phytic acid–treated oxides were annealed in a flowing H2 (5%)/Ar atmosphere with the same annealing process for NFP/C-3.

Materials characterization

The morphology and structure of the products were characterized usipan class="Chemical">ng FESEM (JEOL, JSM-6700F) and TEM (JEOL, JEM-1400). Elemenpan>tal mapping images were collected using a TEM (JEOL, JEM-2100F) equipped with EDX spectroscopy. XRD patterns were collected on a Bruker D2 Phaser with Ni-filtered Cu Kα radiation (λ = 1.5406 Å) at a voltage of 30 kV and a current of 10 mA. XPS (ESCALAB 250) was performed by an ESCALAB 250 x-ray photoelectron spectrometer to evaluate the chemical state of the elements on the surface of the materials. All XPS spectra were corrected using the C 1s line at 284.6 eV, and curve fitting and background subtraction were accomplished. Raman spectra were collected on a Renishaw Invia Reflex Raman microscope equipped with a 758-nm excitation laser. N2 adsorption measurements were performed in an ASAP 2010 Micromeritics apparatus. The samples of MIL-88A, NM-1, NM-2, and NM-3 were evacuated at 423 K for 20 hours before analysis. The samples of FeP/C, NFP/C-1, NFP/C-2, and NFP/C-3 were evacuated at 473 K for 20 hours before analysis. The pore volume and pore size distribution were calculated from the N2 adsorption curve by the Barrett-Joyner-Halenda (BJH) method. FTIR spectra were collected using a PerkinElmer FTIR spectrometer. Inductively coupled plasma atomic emission spectroscopy (ICP-AES) using a ThermoFisher iCAP6500Duo was used to analyze the concentration of nickel and iron.

Electrochemical measurements

All measurements were performed ipan class="Chemical">n a standard three-electrode system with a CHI 760E electrochemical workstation at room temperature. Hydrogen-saturated 0.5 M H2SO4, 1.0 M PBS, and 1.0 M KOH solutions were chosen as electrolytes for HER. A carbon rod and an Ag/AgCl electrode (KCl-saturated) were used as the counter electrode and the reference electrode, respectively, in all measurements except for the reference calibration. The preparation of the working electrodes containing the investigated catalysts can be found as follows. A suspension was prepared by dispersing catalysts (10 mg) in a solution of water and ethanol (2 ml, Vwater/Vethanol = 1/4), followed by the addition of 5 weight % Dupont Nafion 117 solution (50 μl). The mixed solution was sonicated for 5 min to obtain a homogeneous catalyst ink. The dispersion (20 μl) was pipetted onto a piece of clean CFP (0.5 cm by 0.5 cm), which was subjected to overnight solvent evaporation in air. The mass loading of catalysts was 0.4 mg cm−2. Before the electrochemical measurements, the reference electrode was calibrated by performing cyclic voltammetry (CV) scans with a scan rate of 2.0 mV s−1 in hydrogen-saturated electrolytes with a Pt wire, a Pt nanosheet, and an Ag/AgCl electrode as the counter electrode, the working electrode, and the reference electrode, respectively. The average of the two potentials where the current crossed zero was taken to be the thermodynamic potential for the hydrogen electrode reactions. All the LSV curves were carried out at 5.0 mV s−1. The ohmic drop correction was performed by the method of positive feedback, and the compensation level is 95%. All LSV curves were corrected with iR-compensation. EIS measurements were conducted in the range of 0.01 to 105 Hz with an amplitude of 0.01 V.

Computational methods

Spin-polarized DFT calculatiopan class="Chemical">ns were performed in the plane wave and ultrasoft pseudopotentials (USPP) as implemented in Quantum ESPRESSO. Adsorption energies were calculated using the Grimme-D2 vdw-correction with Perdew-Burke-Ernzerhof exchange functional correction. The plane-wave cutoff and density cutoff were 25 and 225 Ry, respectively. FeP bulk lattice constants (relative error) were determined to be a = 5.155 Å (−0.73%), b = 5.760 Å (−0.55%), and c = 3.054 Å (−1.45%), in good agreement with experimentally measured values (). All the atomic structures for the models were fully relaxed with self-consistency accuracy of 10−5 Ry, and the residual forces were within 10−3 Ry/bohr for geometry optimizations. The occupancy of the one-electron states was calculated using an electronic temperature of κBT = 0.01 Ry for surfaces and 0.001 Ry for molecules in vacuum. All energies were extrapolated to T = 0 K. A vacuum slab of 10 Å was used for surface isolation to prevent interaction between two surfaces (15 Å only change less than 1‰ energy), and the top atomic layers were relaxed with all other layers fixed to simulate bulk structure. Brillouin zone was simply using k-points with 5 × 5 × 1 on slab (increase the Brillouin zone make little change to final energy) and 5 × 5 × 5 on bulk structures.

Since θH is extremely importapan class="Chemical">nt for the calculation of HER, especially for TMPs and sulfides, the exploration of θH is necessary (). The θH is defined as the fraction of a monolayer with respect to the number of available sites on the basal planewhere ΔEH is the hydrogen chemisorption energy, ΔEZPE is the difference in zero-point energy between the adsorbed and the gas phase, and ΔSH is the gas phase entropy change of H adsorption. ΔEH is computed by ΔEH = Eslab+ − Eslab + ( − 1/2 EH2, where n is the number of H atoms in the calculation. Moreover, TΔSH and ΔEZPE were obtained by following the scheme proposed by Nørskov et al. (): The entropy of adsorption of 0.5H2 is , where is the entropy of H2 in the gas phase at standard conditions. The (002) plane observed by HRTEM was selected for calculation, and no other planes were discussed in this paper.