Photocatalytic hydrogen evolution over nickel cobalt bimetallic phosphate anchored graphitic carbon nitrides by regulation of the d-band electronic structure

Yuanzheng Zhanga, Yunrong Daibc, Lifeng Yin*ac, Huihui Lia, Xiang Chena and Bin Chen*a
aState Key Laboratory of Water Environment Simulation, School of Environment, Beijing Normal University, Beijing 100875, PR China. E-mail:;
bSchool of Water Resources and Environment, China University of Geosciences (Beijing), Beijing, 100083, China
cDivision of Engineering and Applied Science, Linde-Robinson Laboratory, California Institute of Technology, Pasadena, California 91125, USA

Received 20th March 2020 , Accepted 29th April 2020

First published on 29th April 2020

Non-precious metal co-catalysts with high activity and stability are extremely desirable for economically viable photocatalytic molecular hydrogen (H2) evolution. Herein, nickel cobalt phosphate (NiCo–Pi) was introduced into graphitic carbon nitride layers (g-C3N4) via a sonication-assisted ion intercalation method as a substitute for noble metal co-catalysts. Under visible light irradiation, NiCo–Pi/g-C3N4 (Ni/Co molar ratio of 4[thin space (1/6-em)]:[thin space (1/6-em)]5) exhibited the highest photocatalytic activity (ca. 10[thin space (1/6-em)]184 μmol h−1 g−1) and stability for H2 evolution. Synchrotron radiation X-ray absorption spectroscopy (XAS) indicated that NiCo–Pi is closely bound to g-C3N4 via covalent binding, which accelerates electron transport. Moreover, the unoccupied d-orbital in NiCo–Pi causes the surface to strongly adsorb atomic hydrogen (*H). Theoretically, density functional theory (DFT) calculations demonstrated that the d-band center position of NiCo–Pi is relocated upon adjusting the Ni/Co molar ratio, which changes the adsorption energy of NiCo–Pi toward intermediate state *H. This work provides new insights for exploring the role of the bimetallic composition in non-noble co-catalysts for highly efficient H2 evolution.

1. Introduction

Increasingly serious energy and environmental crises have led to extensive research into renewable and clean energy. Molecular hydrogen (H2) evolution over semiconductor photocatalysts is considered an ideal approach for producing renewable energy.1,2 In the last few decades, studies of photocatalytic hydrogen evolution have made great progress.3–6 To reinforce the H2 yield, noble metal co-catalysts, which trap electrons, promote the separation of photo-induced charge carriers, and act as the reactive sites for proton reduction,5,7 are usually loaded onto a semiconductor. Since platinum-group metals (Pt, Pd, and Au) have advantages, such as higher work functions8 and sufficient free energy for hydrogen adsorption,9,10 they are considered ideal co-catalysts for H2 evolution. However, platinum-group metals are too expensive for extensive applications, since even a 0.1% dosage might triple the price of a photocatalyst.8 Therefore, alternative co-catalysts made from non-noble metals that retain the activities of noble metals are highly desirable.

Many researchers have focused on using low-cost metal compounds, including phosphides,11 hydroxides,12 nitrides,13 sulfides,14 and borides.15 However, current co-catalysts based on non-noble metal elements demonstrate unsatisfactory activities and stabilities (Table S1, ESI). Recently, transition metal phosphates have shown potential as electrodes in the field of electrocatalytic water splitting,16,17 mainly due to their open-structure frameworks that expose a wealth of reaction sites.18 Moreover, the phosphate group is favourable for the adsorption of water due to native atomic geometry distortion, which results in a lower over-potential for water splitting.19 In addition, transition metal phosphates possess inherent structural stability, which ensures that catalysts made from them remain stable under real reaction conditions.20 In light of these characteristics, transition metal phosphates are expected to benefit photocatalytic water-splitting processes as well. Some reports have suggested that specific electrocatalysts for water splitting could potentially be co-catalysts for photocatalytic H2 evolution.21,22 Very recently, Qin et al.23 applied a NiCoP@NiCo–Pi core/shell non-noble metal co-catalyst to photocatalytic hydrogen evolution for the first time. However, the relationship between the bimetallic components of cobalt and nickel has not been studied in detail. Some studies have implied that bimetallic phosphates are usually superior to monometallic phosphates when used to increase H2 yields in electrocatalytic processes,16,24 but the underlying mechanism related to the improved catalytic activity caused by bimetallic compounds has not been well explained.

Herein, we developed a simple strategy for preparing a NiCo–Pi/g-C3N4 visible-light responsive catalyst for H2 evolution. Then, we explored how the catalytic reactivity of bimetallic NiCo–Pi is affected by the Ni/Co molar ratio through experiments and DFT calculations. We expect to produce a promising and general strategy for developing non-noble metal photocatalysts aimed at high-efficiency photocatalytic H2 evolution.

2. Experimental

2.1 Materials

Urea, ethanol, isopropyl alcohol, and ammonium sulfite ((NH4)2SO3·H2O) were purchased from Sinopharm Chemical, China. Nickel acetate (NiC4H6O4), cobalt acetate (CoC4H6O4), chloroplatinic acid (H2PtCl6·6H2O), and triethanolamine (TEOA) were obtained from Sigma-Aldrich. Sodium pyrophosphate (Na4P2O7), Nafion membrane solution (5%), and polyvinylidene fluoride (PVDF) were purchased from Alfa (UK).

2.2 Methods

Synthesis of g-C3N4 and NiCo–Pi/g-C3N4 composite photocatalysts. A suitable amount of urea was added to a quartz crucible (filling volume: 50%) and then heated to 600 °C for 4 h at a rate of 5 °C min−1 to obtain g-C3N4 power. The NiCo–Pi/g-C3N4 hybrid was prepared through a sonication-assisted ion intercalation and hydrothermal route. Specifically, 0.25 g of g-C3N4 was added to 15 mL of water, which then underwent sonication for 0.5 h to obtain a homogeneous solution. Second, 5 mL of an aqueous solution containing 0.08 mmol of nickel acetate and 0.1 of mmol cobalt acetate (Ni/Co molar ratio of 4[thin space (1/6-em)]:[thin space (1/6-em)]5) was mixed with the above solution. After another 4 h of sonication, 2 mL of an aqueous solution containing 0.1 mmol of sodium pyrophosphate was added into the above solution dropwise, followed by stirring for 0.5 h. After that, the obtained mixture was transferred into a 25 mL hydrothermal reactor and held at 160 °C for 8 h. Finally, samples were obtained via vacuum filtration drying. Additionally, NiCo–Pi/g-C3N4 hybrids with other Ni/Co molar ratios (0[thin space (1/6-em)]:[thin space (1/6-em)]1, 1[thin space (1/6-em)]:[thin space (1/6-em)]2, 2[thin space (1/6-em)]:[thin space (1/6-em)]3, 1[thin space (1/6-em)]:[thin space (1/6-em)]1, 2[thin space (1/6-em)]:[thin space (1/6-em)]1, and 1[thin space (1/6-em)]:[thin space (1/6-em)]0) were also obtained using the same procedure.

2.3 Characterization

The crystal structures of the products were studied via X-ray diffraction (XRD, X'pert pro MPD). Scanning electron microscopy (SEM) data were obtaining using a Hitachi S-4800 field emission secondary electron microscope and a PGT EDS ImixrImagist analyser. Images were collected with a beam potential of 10 kV. Transmission electron microscopy (TEM, JEM-2100F) studies were carried out to observe the microstructures, sizes and element distributions of the samples. X-ray photoelectron spectroscopy (XPS, ESCSLAB 250Xi) was used to analyse the chemical states (calibrated based on the C 1s peak at 284.6 eV). Fourier transform infrared spectroscopy (FT-IR, Nicolet iS 50) was used to analyse the chemical bond compositions of samples. The light absorption properties were analysed via ultraviolet visible (UV-vis) spectrophotometer (Shimadzu UV 2600). Electron paramagnetic resonance (EPR) spectra of hydrogen radicals were investigated using the X band with a Bruker E500 spectrometer. The photogenerated carrier separation performances of samples were characterized using a photoluminescence (PL) spectrometer (FluoroMax-4). Ultraviolet photoemission spectroscopy (UPS, ESCSLAB 250Xi) with a He I light source (21.2 eV) was carried out to measure the work functions of samples. Photoelectrochemical measurements were made using a PGSTAT204 workstation (Autolab, Metrohm) with a typical three-electrode system, with a platinum plate as the counter electrode, a Ag/AgCl electrode as the reference electrode, and a glassy carbon (GC) electrode (5 mm in diameter) coated with sample as the working electrode. The working electrode preparation was as follows. 6 mg of catalyst was dispersed in 1 mL of a mixed solution of ultrapure water (670 μL), isopropanol (300 μL), Nafion solution (5%, 20 μL), and PVDF solution (5%, 10 μL), followed by ultrasonic treatment for 30 minutes to obtain a uniform slurry solution. Then, 5.5 μL of slurry was dripped onto the surface of the GC and dried at 60 °C. Mott–Schottky data were measured at frequencies of 1.0, 1.5 and 2.0 kHz, respectively. Electrochemical impedance spectroscopy (EIS) data were measured from 105–0.01 Hz. All tests were conducted in 0.2 M Na2SO4 aqueous solution (pH = 6.8).

2.4 Photocatalytic H2 evolution activity and stability

Photocatalytic H2 generation at room temperature was carried out in a quartz glass reactor (150 ml) sealed with a silica gel gasket. Low-power light-emitting diodes (LEDs) (410, 420, 440, 460 and 520 nm) as light sources were placed at points around the reactor at distances of 3 cm. Specifically, 10 mg of sample was suspended in 5 mL of deionized water and treated with ultrasound for 10 min to form a uniformly dispersed solution. The above solution was then poured into 120 mL of aqueous solution containing 10 vol% triethanolamine (TEOA) scavenger. Before irradiation, N2 was bubbled through the suspension for 0.5 h to remove dissolved oxygen. The suspension was stirred continuously throughout the experiment. During irradiation, 50 μL of gas was taken from the reactor every 1 h. The quantitative analysis of H2 was carried out via gas chromatography (GC 7890A, Agilent) with a thermal conductivity detector (TCD) according to the established hydrogen standard curve (Fig. S1). For comparison, the H2 production performances of samples were further tested with the assistance of Pt (ca. 1 wt%) co-catalyst. For photocatalytic stability testing, the photocatalyst was collected from the reaction solution via vacuum filtration with a 0.1 μm PTFE membrane (Amicrom, UK), washed with water and dried in an oven at 60 °C. The collected photocatalyst showed no apparent weight loss (for example, the dry weights of samples reused 0, 1, 2, 3, and 4 times are 10, 9.7, 9.6, 9.4, and 9.5 mg, respectively) and was used for repeat hydrogen evolution experiments.

The apparent quantum efficiency (AQE) was calculated as follows:25

image file: d0cy00556h-t1.tif
where Ne represents the number of electrons; Np represents the number of photons; n represents the number of molecules generating H2; and NA, S, P, λ, t, h, and c are the Avogadro constant, irradiation area, incident light intensity, monochromatic light wavelength, time, Planck constant, and vacuum speed of light, respectively. The light intensities of LED lights with different wavelengths at the same irradiation distance were measured via a visible-light radiometer (FZ-A, BNU, CHN) and are as follows: 410 nm, 4.7 mW cm−2; 420 nm, 5.6 mW cm−2; 440 nm, 4.9 mW cm−2; 460 nm, 6.0 mW cm−2; and 520 nm, 7.8 mW cm−2.

2.5 XAS tests and analysis

X-ray absorption spectroscopy (XAS) tests were conducted using the BL14W1 beamline of the Shanghai Synchrotron Radiation Facility. Fluorescence mode was used to collect data. The Si(111) double crystal monochromator was calibrated with Co and Ni foil. Samples were made into discs with a diameter of 13 mm using a tablet press with LiF powder as the binder. The acquired EXAFS data were obtained through ATHENA and ARTEMIS software analysis.26,27

2.6 Calculation methods

First-principles calculations were implemented using the Vienna ab initio simulation package (VASP).28 The projector-augmented wave (PAW) potential method was adopted to describe the electron-ion interactions, and the generalized gradient approximation (GGA) of Perdew–Burke–Ernzerhof (PBE) was used as the exchange-correlation function.29 The cutoff energy was set to 400 eV. For geometry optimization, the energy criterion of ion relaxation was set to 10−5 eV, and the force convergence criterion is 0.01 eV Å−1. The vacuum thickness was set to 20 Å. The Brillouin zone sampling for NiCo–Pi used a 5 × 4 × 1 k-point grid. The hydrogen adsorption energies on NiCo–Pi with different Ni/Co molar ratios were calculated via ΔE(H*) = E(M+H)EM − 1/2E(H2), where E(M+H) represents the energy of NiCo–Pi with an adsorbed H atom, EM is the energy of NiCo–Pi with a clean surface, and E(H2) is the energy of a gaseous H2 molecule. The Gibbs free-energies of the adsorbed states were obtained via ΔGH* = ΔE(H*) − ΔEZPETΔS, where ΔEZPE and ΔS represent the zero-point energy difference and the entropy change before and after the reaction, respectively.

3. Results and discussion

The synthesis of NiCo–Pi/g-C3N4 via a three-step strategy is schematically depicted in Fig. 1. Co2+/Ni2+ ions in cobalt acetate tetrahydrate and nickel acetate tetrahydrate, respectively, were simultaneously intercalated into the interlayer space of g-C3N4 via ultrasonic dispersion. The inserted Co2+/Ni2+ ions can be bound on the surface of the g-C3N4 layer due to the absorption of the rings.30 With the addition of Na4P2O7 solution, the P2O74− ions act as an ideal ligand,31 strongly combining with the Co2+/Ni2+ ions via coordination interactions to form a pyrophosphate complex. After the formation of metal-chelated structures on the g-C3N4 layers, NiCo–Pi/g-C3N4 with a layer-by-layer stacking structure was finally obtained through a phosphorylation reaction following hydrothermal treatment at 160 °C for 8 h.32
image file: d0cy00556h-f1.tif
Fig. 1 A schematic diagram showing the NiCo–Pi/g-C3N4 synthesis strategy.

The X-ray diffraction (XRD) patterns (Fig. S2) of the original g-C3N4 and NiCo–Pi/g-C3N4 reveal the structural changes before and after ion intercalation. It can be seen that g-C3N4 exhibits two typical diffraction peaks at 13.1° and 27.4°. The peak at 13.1° belongs to the (100) diffraction plane, which is derived from the in-plane structural packing of tri-s-triazine units. The peak centered at 27.4° derives from the interplanar stacking of conjugated aromatic systems along the (002) plane.33 Compared to the original g-C3N4, no diffraction shift of the (002) crystal surface peak was observed, which indicates that NiCo–Pi contributes little to the lattice distortion of g-C3N4. However, different from the original g-C3N4, the (002) diffraction peak intensity of NiCo–Pi/g-C3N4 observed decreased, which indicates that the long-distance ordered nature of g-C3N4 is lost during the growth of NiCo–Pi and the stacking of interlayer structures becomes loose, resulting in some layers being exfoliated. In addition, there are some new diffraction peaks (blue color) in the NiCo–Pi/g-C3N4 spectrum, which can be mainly assigned to CoNi(PO4)2, indicating the formation of NiCo–Pi solid solution. The Fourier transform infrared (FT-IR) spectra (Fig. S3) show the tri-s-triazine characteristic peaks of NiCo–Pi/g-C3N4 and g-C3N4. The bands within the 1242–1636 cm−1 range correspond to the typical stretching modes of C and N heterocycles, and the characteristic bands at 808 cm−1 correspond to the vibrational mode of triazine units,34,35 indicating that the structure of g-C3N4 was integrally preserved after being coupled with NiCo–Pi. Moreover, the detailed pore structures of samples were studied via N2 adsorption (Fig. S4 and Table S2).

SEM imaging of pure g-C3N4 (Fig. 2a) shows that the sample has a layered structure with a size of 3–5 μm. TEM imaging of g-C3N4 (Fig. S5) shows a wrinkled sheet structure, which is caused by surface defects in the sample, minimizing the total surface energy. The SEM image of the NiCo–Pi/g-C3N4 hybrid illustrated in Fig. 2b shows that the composite exhibits a more obvious lamellar structure, smaller size and rougher surface than g-C3N4, indicating that after ultrasonic and hydrothermal treatment, NiCo–Pi grew between the layers of g-C3N4. As shown in Fig. 2c and d, platelet-like NiCo–Pi with a size of 50–100 nm tightly attached to g-C3N4, suggesting the successful synthesis of ultrathin NiCo–Pi/g-C3N4 hybrids. The HRTEM image in Fig. 2e clearly presents lattice fringes of d = 0.33 nm, corresponding to the (202) crystal plane of NiCo–Pi.32 Elemental digital mapping images (Fig. 2f) reveal that Ni, Co, P, and O elements are uniformly distributed in the support structure, providing clear evidence that the NiCo–Pi platelets were homogeneously inserted into the gaps between adjacent g-C3N4 sheets to form a layer-by-layer structure of g-C3N4 and NiCo–Pi.

image file: d0cy00556h-f2.tif
Fig. 2 SEM images of pure g-C3N4 (a) and NiCo–Pi/g-C3N4 (b); TEM images of NiCo–Pi/g-C3N4 (c and d) and a HRTEM image of NiCo–Pi/g-C3N4 (e); and the corresponding elemental mapping images of NiCo–Pi/g-C3N4 (f).

The photocatalytic performances of various g-C3N4-based samples were evaluated for H2 evolution. Fig. 3a shows that the H2 yield is linear with respect to time, suggesting continuous and stable H2 evolution. The original g-C3N4 exhibited negligible activity toward H2 evolution. After hybridizing with NiCo–Pi, the H2 evolution activity was impressively increased, depending on the Ni/Co molar ratio. Interestingly, there existed a “golden ratio” (Ni/Co molar ratio of 4[thin space (1/6-em)]:[thin space (1/6-em)]5) at which the H2 yield reached as high as 10[thin space (1/6-em)]184 μmol h−1 g−1 (Fig. 3b), 1.2 times greater even than that of Pt (1.0 wt%)/g-C3N4 (the optimal value of the Pt/g-C3N4 photocatalyst: 8469 μmol h−1 g−1). In addition, the NiCo–Pi content in the NiCo–Pi/g-C3N4 hybrids dramatically affected the photocatalytic H2 evolution (Fig. S6). The H2 yield increased from 2.0 wt% to 6.0 wt% and decreased thereafter. However, under the same conditions, NiCo–Pi alone did not show any H2 evolution activity, suggesting that the excellent H2 evolution yield is closely related to interactions between NiCo–Pi and g-C3N4, or to new structures created between them. In addition, EPR spectra showed that cobalt and nickel significantly promoted the proton reduction reaction (Fig. S7).

image file: d0cy00556h-f3.tif
Fig. 3 (a) The H2 evolution yields over g-C3N4-based samples at 410 nm. (b) A comparison of the H2 evolution rates for all samples at 410 nm. (c) Cycling tests of photocatalytic H2 generation over NiCo–Pi/g-C3N4 at 410 nm. (d) The external quantum efficiency of NiCo–Pi/g-C3N4 under irradiation at different wavelengths.

NiCo–Pi/g-C3N4 was not only capable of continuously and stably catalyzing the release of H2 but it was also sufficiently stable to resist light etching: no noticeable changes in the yield rate were detected over 5 cycles and 20 h of the H2 evolution process (Fig. 3c). Fig. 3d shows the dependency between the wavelength of the incident photons and the H2 evolution AQE using NiCo–Pi/g-C3N4. The AQE values of NiCo–Pi/g-C3N4 at 410 and 420 nm were 18.4% and 14.3%, respectively, which were extensively enhanced relative to previously reported yields from Pt/g-C3N4,25,28 and higher than those previously reported from g-C3N4 loaded with a NiCoP co-catalyst;36 this may be due to the fact that phosphate functional groups are more favorable for water adsorption.19 The AQE values are 3.9%, 0.7% and 0.1% at 440, 460, 520 nm, respectively, indicating that the irradiation light wavelength when using NiCo–Pi/g-C3N4 should be less than 460 nm.

To look into the surface chemical environment, the NiCo–Pi/g-C3N4 hybrids were characterized via XPS. Fig. S8 shows that the C 1s spectra has two peaks centered at 284.7 and 288.2 eV, corresponding to impurity carbon (C–C) and sp2-hybridized carbon (N–C[double bond, length as m-dash]N),37 respectively. Fig. 4a shows that the N 1s spectrum of g-C3N4 possesses three peaks. The peak at 398.6 eV is attributed to sp2-hybridized nitrogen (C–N[double bond, length as m-dash]C), the peak at 399.8 eV corresponds to tertiary nitrogen (N–(C)3), and the characteristic peak at 401.0 eV corresponds to amino groups (C–N–H).38 Note that the characteristic peak from N–(C)3 in the NiCo–Pi/g-C3N4 hybrid shows a 0.4 eV blue-shift, indicating that NiCo–Pi undergoes chemical interactions with g-C3N4 through Co–N and Ni–N bonds.39 The P 2p XPS spectrum (Fig. S9) displays three peaks at 133.0, 134.1 and 135.1 eV, which are attributed to the characteristic P 2p3/2 absorption peaks of P(V), indicating the formation of phosphate species.40 Fig. 4b shows the XPS spectrum of Co 2p; the peaks centered at 782.0 and 798.1 eV correspond to the characteristic peaks of Co2+.41 As shown in Fig. 4c, Ni 2p3/2 and Ni 2p1/2 peaks can be observed at 856.1 and 873.9 eV, and the peaks at 861.8 and 873.9 eV correspond to satellite peaks of Ni 2p, verifying the existence of Ni2+. Compared with the characteristic Co 2p3/2 peak in Co3(PO4)2 (780.7 eV) and Ni 2p3/2 peak in Ni3(PO4)2 (855.5 eV),42 the Co 2p3/2 and Ni 2p3/2 peaks in the NiCo–Pi/g-C3N4 hybrids show significant red-shifting, which further proves that Co–N and Ni–N bonds were formed.43

image file: d0cy00556h-f4.tif
Fig. 4 (a)–(c) N 1s, Co 2p, and Ni 2p XPS spectra of NiCo–Pi/g-C3N4. (d) Co K-edge XANES experimental spectra of NiCo–Pi/g-C3N4, Co3O4, Co(NO3)2, and Co foil. (e) Ni K-edge XANES experimental spectra of NiCo–Pi/g-C3N4, NiO, and Ni foil. (f) FT spectra of the Co K-edge and (g) Ni K-edge in R space for NiCo–Pi/g-C3N4. (h) WT spectra for the k2-weighted Co K-edge and (i) Ni K-edge.

XAS testing was applied to explore the detailed local electronic environments of the NiCo–Pi/g-C3N4 hybrids. Fig. 4d and e shows X-ray absorption near-edge structure (XANES) spectra, in which the Co white line peak of NiCo–Pi/g-C3N4 is located close to the Co(NO3)2 reference and, similarly, the Ni white line peak is located close to the NiO reference, implying that the valence states of Co and Ni are both bivalent. The small front edge feature peaks at about 7709 eV and 8335 eV for the Co/Ni K-edges can be attributed to electronic transitions from the 1s core level to the unoccupied 3d level,44,45 implying the existence of holes in the d-orbital, which are particularly active during metal (M)–H bond formation.46

The Fourier transform (FT) of the Co K-edge extended X-ray fine structure (EXAFS) of NiCo–Pi/g-C3N4 exhibited two peaks at approximately 1.50 and 2.32 Å (non-phase-correction, Fig. 4f), corresponding to Co–N/O and Co–Ni/Co coordination, respectively. Similarly, the FT of the Ni K-edge also showed two peaks located at 1.50 and 2.41 Å (non-phase-corrected, Fig. 4g), corresponding to Ni–N/O and Ni–Co/Ni coordination, respectively.39,44 EXAFS fitting data from the samples are shown in Fig. S10 and Tables S3 and S4. However, EXAFS cannot distinguish the coordinated N and O atoms due to their adjacent atomic numbers.47 To provide evidence for Co–N and Ni–N coordination, the wavelet transform (WT) of EXAFS analysis was conducted due to its powerful ability to separate backscattering atoms in k and R space.48 Fig. 4h and i shows that the WT contour plots of Co foil and Ni foil both present one intensity maximum at approximately 7.0 Å, which can be assigned to Co–Co and Ni–Ni coordination. NiCo–Pi exhibits an intensity maximum at 4.1 Å, which can be associated with Co–O or Ni–O coordination. By comparison, the WT of NiCo–Pi/g-C3N4 at the K-edges of Co and Ni exhibits maxima at 3.5 and 3.9 Å, which are lower than that of NiCo–Pi, indicating that Co and Ni are bonded with an element lighter than O. Therefore, it is proved that Co–N and Ni–N bonds exist between NiCo–Pi and g-C3N4. In addition, the electronegativity of N is much stronger than those of Ni and Co atoms. Therefore, this unique bonding state could provide a resistance-free path for enhancing electron transfer.39

To obtain insights into the significantly improved H2 evolution performance, we compared the optical properties of the original and modified photocatalysts in detail using UV-vis absorption spectroscopy. Fig. S11a shows that the adsorption edge of g-C3N4 was located at approximately 465 nm. However, the light absorption of NiCo–Pi/g-C3N4 did not show any significant changes, indicating that the band gap was maintained after NiCo–Pi was integrated with g-C3N4. On the basis of the Kubelka–Munk formula, the derived electronic band gap of the as-prepared sample was calculated to be 2.85 eV (Fig. S11b).

To investigate photogenerated electron transfer at the interface of the NiCo–Pi/g-C3N4 hybrids, the band edge potentials of the materials were studied via Mott–Schottky measurements (Fig. S12a and b). The derived flat band potentials were approximately −1.26 and −1.18 V (vs. Ag/AgCl) for g-C3N4 and NiCo–Pi/g-C3N4, corresponding to −1.06 and −0.98 eV versus a normal hydrogen electrode (NHE) (ENHE = EAg/AgCl + 0.197 V).49 The decrease in flat band potential (∼0.08 eV) indicates the existence of a heterojunction effect. Fig. S12c shows the UPS spectra of the samples, for analyzing changes in work function (eΦ) before and after contact. It can be seen that the cutoff energy (Ecutoff) values of g-C3N4 and NiCo–Pi/g-C3N4 are 18.41 eV and 18.15 eV, respectively, and their Fermi energy levels (EF) are both 0 eV. So eΦ can be calculated to be 2.79 eV and 3.05 eV according to the equation: eΦ = hv − |EcutoffEf|.50 The increase (0.26 eV) in the work function between g-C3N4 and NiCo–Pi/g-C3N4 confirms the formation of a Schottky junction.51 In addition, NiCo–Pi has been proven to be a metallic material,52 which is further supported by the fact that electrons generated by g-C3N4 can be trapped by NiCo–Pi. The band gap of g-C3N4 is 2.85 eV, so the VB potential should be 1.79 V versus NHE, which is in accordance with the VB-XPS observations. As shown in Fig. S12d, the apparent VB value for g-C3N4 is 1.98 eV. Based on the above analysis, a possible photocatalytic mechanism over the NiCo–Pi/g-C3N4 Schottky junction is proposed in Fig. S13. As described above, after NiCo–Pi contacts with g-C3N4, the work function of NiCo–Pi/g-C3N4 increased by 0.26 eV compared with that of g-C3N4, which can be attributed to the migration of free electrons from g-C3N4 to NiCo–Pi, making the Fermi level reach an equilibrium state between g-C3N4 and NiCo–Pi. Moreover, the migration of free electrons induces the formation of a built-in electric field from g-C3N4 to NiCo–Pi. The alignment of electrons induces a space charge layer on the g-C3N4 side, leading to the upward bending of the energy band, where the Schottky barrier is formed. As a result, under visible light irradiation, photoinduced electrons in g-C3N4 will be transferred to NiCo–Pi, driven by the built-in electric field, reducing protons to H2. Meanwhile, the holes left in g-C3N4 are simultaneously consumed by TEOA.

A series of photoelectrochemical characterization experiments were further carried out. Fig. 5a shows that all the samples exhibit good photoresponsivity and stability during on/off light illumination cycles. It can be seen that the photocurrent densities of NiCo–Pi/g-C3N4 samples with different Ni/Co molar ratios are much higher than that of g-C3N4, indicating that the introduced NiCo–Pi could improve the electron–hole separation and transportation efficiencies.53 Moreover, PL emission spectroscopy was employed to examine the electron–hole separation efficiencies. Generally, a lower PL emission intensity indicates a lower charge carrier recombination rate.54 As shown in Fig. 5b, the intensities of the NiCo–Pi/g-C3N4 hybrids are remarkably lower than that of g-C3N4, which implies that the electron–hole recombination rate is apparently inhibited in the NiCo–Pi/g-C3N4 hybrids, suggesting that photoelectrons were transferred to NiCo–Pi via Co/Ni–N bonds.39,55 However, the NiCo–Pi/g-C3N4 sample with the “golden ratio” (Ni/Co molar ratio of 4[thin space (1/6-em)]:[thin space (1/6-em)]5) did not exhibit the highest charge separation rate, meaning that the H2 evolution activity is not mainly determined by the charge separation rate. The same phenomenon is also observed from electrochemical impedance spectroscopy (EIS). After integrating NiCo–Pi, the semicircles in the Nyquist plots became much smaller, indicating that the charge transfer resistance had decreased (Fig. 5c). Besides, the fitted equivalent circuit (inset, Fig. 5c) indicated that NiCo–Pi could decrease the Rct impedance at the electrode/electrolyte interface,56 and NiCo–Pi/g-C3N4 with a Ni/Co molar ratio of 1[thin space (1/6-em)]:[thin space (1/6-em)]1 exhibited the smallest Rct value, about 21.6 times lower than that of pure g-C3N4 (Table S5). However, the resistance of NiCo–Pi/g-C3N4 samples with Ni/Co molar ratios of 4[thin space (1/6-em)]:[thin space (1/6-em)]5, 1[thin space (1/6-em)]:[thin space (1/6-em)]1, and 2[thin space (1/6-em)]:[thin space (1/6-em)]3 did not change noticeably; however, there was a substantial difference in the H2 evolution performance depending on the ratio. Therefore, the enhancement in the photocatalytic activity arising from the addition of NiCo–Pi is not primarily caused by the effective separation and migration of photogenerated carriers. LSV curves were obtained to explore the H2 evolution potentials of the NiCo–Pi/g-C3N4 hybrids (Fig. 5d). Impressively, the overpotential value of NiCo–Pi/g-C3N4 (Ni/Co molar ratio of 4[thin space (1/6-em)]:[thin space (1/6-em)]5) was the lowest at the same current density. A lower overpotential means that less energy is needed for the reaction: H+ + e → 1/2H2.57 Therefore, a considerable decrease in the overpotential indicates that NiCo–Pi (Ni/Co molar ratio of 4[thin space (1/6-em)]:[thin space (1/6-em)]5) can provide effective sites for proton reduction, thus improving the electron utilization efficiency.15 We believe that the disparities in H2 evolution activities at different Ni/Co molar ratios are mainly due to different potential barriers to the H2 evolution reaction.

image file: d0cy00556h-f5.tif
Fig. 5 (a) Transient photocurrents, (b) PL spectra (excitation at 320 nm), (c) EIS Nyquist plots, and (d) LSV curves from g-C3N4 and NiCo–Pi/g-C3N4 hybrids with different Ni/Co molar ratios.

The H2 evolution activity of NiCo–Pi/g-C3N4 was effectively regulated by the Ni/Co molar ratio. Hence, DFT calculations were used to obtain fundamental insights into the role of the Ni/Co composition in determining the hydrogen conversion behaviour over NiCo–Pi/g-C3N4. Generally, H2 evolution can be divided into three stages: the initial state (H+ + e) involving the reactants: an intermediate transition state (adsorbed H*); and the final state (1/2H2).58,59 The Gibbs free-energy of H* adsorption (ΔGH*) has been proven to be a key value for characterizing the sorption behaviour of hydrogen over a solid surface. An ideal catalyst for H2 evolution should have a ΔGH* value close to zero to make the bond strength appropriate.60 NiCo–Pi is a monoclinic crystal bimetallic phosphate and belongs to the space group P21/a (Fig. S14). DFT calculations were employed to examine the ΔGH* values of NiCo–Pi(100) and Pt(111) (Fig. 6a). The ΔGH* values for NiCo–Pi with Ni/Co molar ratios of 0[thin space (1/6-em)]:[thin space (1/6-em)]1, 1[thin space (1/6-em)]:[thin space (1/6-em)]3, 3[thin space (1/6-em)]:[thin space (1/6-em)]1, and 1[thin space (1/6-em)]:[thin space (1/6-em)]0 were largely positive, except for 1[thin space (1/6-em)]:[thin space (1/6-em)]1, indicating low H* adsorption efficiency. Most importantly, NiCo–Pi samples with Ni/Co molar ratios of 2[thin space (1/6-em)]:[thin space (1/6-em)]3 and 4[thin space (1/6-em)]:[thin space (1/6-em)]5 show near-zero values of ΔGH*, even closer to zero than the ΔGH* value of Pt, suggesting that the remarkable catalytic activity for H2 evolution might be closely associated with hydrogen adsorption over the surface.61 Calculated d-band density of states (DOS) plots for NiCo–Pi are illustrated in Fig. S15. The d-band DOS near the Fermi level for NiCo–Pi samples with Ni/Co molar ratios of 1[thin space (1/6-em)]:[thin space (1/6-em)]0 and 0[thin space (1/6-em)]:[thin space (1/6-em)]1 were much lower than those of NiCo–Pi samples with other Ni/Co molar ratios, resulting in lower carrier densities and conditions that are not conducive to the charge transfer process,62,63 which agrees with the transient photocurrent and EIS measurements. On the other hand, the d-band central position could affect the H* adsorption strength of the catalyst, with the binding energy increasing as the d-band centre (εd) becomes closer to the Fermi level.60,64 Therefore, we studied the relationship between the centre of the d-band and the adsorption strength of the hydrogen atom on the NiCo–Pi surface (Fig. 6b). Our results are consistent with the d-band theory.65 The d-band centre of NiCo–Pi with a Ni/Co molar ratio of 4[thin space (1/6-em)]:[thin space (1/6-em)]5 (−1.58 eV) is closest to the Fermi level and has the strongest adsorption capacity for hydrogen atoms. This agrees with our calculated ΔGH* values. Fig. S16 shows adsorption models of H on the NiCo–Pi surface. Note that the bond length of H–Ni/Co varies regularly with the Ni/Co molar ratio. The H–Ni/Co bond length increased from 1.457 Å to 1.492 Å as the Ni/Co molar ratio changed from 0[thin space (1/6-em)]:[thin space (1/6-em)]1 to 4[thin space (1/6-em)]:[thin space (1/6-em)]5, and decreased to 1.442 Å as the Ni/Co molar ratio changed from 4[thin space (1/6-em)]:[thin space (1/6-em)]5 to 1[thin space (1/6-em)]:[thin space (1/6-em)]0. The appropriate bond length of H–Ni/Co on NiCo–Pi with a molar ratio of 4[thin space (1/6-em)]:[thin space (1/6-em)]5 creates a situation in which hydrogen atoms can more readily move to the active site to form H2 molecules.66,67 Our findings suggest that bimetallic Ni/Co can significantly enhance the d-band density of states near the Fermi level to promote charge transfer. More importantly, the d-band centre shifts with the Ni/Co molar ratio, thus affecting the value of ΔGH*.

image file: d0cy00556h-f6.tif
Fig. 6 (a) H2 evolution free-energy diagrams of Pt and NiCo–Pi with different Ni/Co molar ratios. (b) H adsorption energy trends on NiCo–Pi with different d-band center positions.


In summary, NiCo–Pi was directly hybridized into g-C3N4 layers via a facile ion intercalation method. With an optimized Ni/Co molar ratio and tight interfacial contact with adjacent g-C3N4 layers, NiCo–Pi/g-C3N4 can release H2 continuously and stably under visible light. The yield of H2 evolution was as high as 10[thin space (1/6-em)]184 μmol h−1 g−1, which is significantly better than the catalytic activity of Pt/g-C3N4 (8469 μmol h−1 g−1). XAFS analysis and theoretical calculations revealed that the photocatalytic activity is closely related to the electronic structure of NiCo–Pi. Our research reveals the pivotal ability to tune the hydrogen atom adsorption behaviour by adjusting the d-band centre of NiCo–Pi and enhancing the charge transfer. We expect that this strategy will contribute to the development of other non-noble metal co-catalyst materials and provide a route for improving photocatalytic hydrogen evolution performance.

Conflicts of interest

There are no conflicts to declare.


This work was financially supported by the National Natural Science Foundation of China (Grant 21777009), the Bill and Melinda Gates Foundation (BMGF RTTC Grants OPP1111246 and OPP1149755), the Beijing Natural Science Foundation (Grant 8182031), and the Major Science and Technology Program for Water Pollution Control and Treatment (Grant 2018ZX07109).


  1. S. Chen, T. Takata and K. Domen, Nat. Rev. Mater., 2017, 2, 17050 CrossRef CAS.
  2. D. Voiry, H. S. Shin, K. P. Loh and M. Chhowalla, Nat. Rev. Chem., 2018, 2, 0105 CrossRef CAS.
  3. X.-B. Li, C.-H. Tung and L.-Z. Wu, Nat. Rev. Chem., 2018, 2, 160–173 CrossRef CAS.
  4. C. M. Wolff, P. D. Frischmann, M. Schulze, B. J. Bohn, R. Wein, P. Livadas, M. T. Carlson, F. Jäckel, J. Feldmann, F. Würthner and J. K. Stolarczyk, Nat. Energy, 2018, 3, 862–869 CrossRef CAS.
  5. S. Park, W. J. Chang, C. W. Lee, S. Park, H.-Y. Ahn and K. T. Nam, Nat. Energy, 2016, 2, 16185 CrossRef.
  6. X. Wang, L. Chen, S. Y. Chong, M. A. Little, Y. Wu, W. H. Zhu, R. Clowes, Y. Yan, M. A. Zwijnenburg, R. S. Sprick and A. I. Cooper, Nat. Chem., 2018, 10, 1180–1189 CrossRef CAS PubMed.
  7. J. Yang, D. Wang, H. Han and C. Li, Acc. Chem. Res., 2013, 46, 1900–1909 CrossRef CAS PubMed.
  8. K. Chang, X. Hai and J. Ye, Adv. Energy Mater., 2016, 6, 1502555 CrossRef.
  9. P. Liu and J. A. Rodriguez, J. Am. Chem. Soc., 2005, 127, 14871–14878 CrossRef CAS PubMed.
  10. M. G. Walter, E. L. Warren, J. R. Mckone, S. W. Boettcher, Q. Mi, E. A. Santori and N. S. Lewis, Chem. Rev., 2010, 110, 6446–6473 CrossRef CAS PubMed.
  11. S. Cao, C. J. Wang, W. F. Fu and Y. Chen, ChemSusChem, 2017, 10, 4306–4323 CrossRef CAS PubMed.
  12. Y. Jia, L. Zhang, G. Gao, H. Chen, B. Wang, J. Zhou, M. T. Soo, M. Hong, X. Yan, G. Qian, J. Zou, A. Du and X. Yao, Adv. Mater., 2017, 29, 1700017 CrossRef PubMed.
  13. Z. Sun, H. Chen, L. Zhang, D. Lu and P. Du, J. Mater. Chem. A, 2016, 4, 13289–13295 RSC.
  14. F. Ma, Y. Wu, Y. Shao, Y. Zhong, J. Lv and X. Hao, Nano Energy, 2016, 27, 466–474 CrossRef CAS.
  15. L. Li, Z. Deng, L. Yu, Z. Lin, W. Wang and G. Yang, Nano Energy, 2016, 27, 103–113 CrossRef CAS.
  16. L. Yang, Z. Guo, J. Huang, Y. Xi, R. Gao, G. Su, W. Wang, L. Cao and B. Dong, Adv. Mater., 2017, 29, 1704574 CrossRef PubMed.
  17. C. Costentin, T. R. Porter and J. M. Saveant, J. Am. Chem. Soc., 2016, 138, 5615–5622 CrossRef CAS PubMed.
  18. A. A. Mirghni, M. J. Madito, T. M. Masikhwa, K. O. Oyedotun, A. Bello and N. Manyala, J. Colloid Interface Sci., 2017, 494, 325–337 CrossRef CAS PubMed.
  19. R. Guo, X. Lai, J. Huang, X. Du, Y. Yan, Y. Sun, G. Zou and J. Xiong, ChemElectroChem, 2018, 5, 3822–3834 CrossRef CAS.
  20. H. Kim, J. Park, I. Park, K. Jin, S. E. Jerng, S. H. Kim, K. T. Nam and K. Kang, Nat. Commun., 2015, 6, 8253 CrossRef CAS PubMed.
  21. W. Bi, L. Zhang, Z. Sun, X. Li, T. Jin, X. Wu, Q. Zhang, Y. Luo, C. Wu and Y. Xie, ACS Catal., 2016, 6, 4253–4257 CrossRef CAS.
  22. X. Zou and Y. Zhang, Chem. Soc. Rev., 2015, 44, 5148–5180 RSC.
  23. Z. Qin, Y. Chen, Z. Huang, J. Su and L. Guo, J. Mater. Chem. A, 2017, 5, 19025–19035 RSC.
  24. J. Xing, H. Li, M. Ming-Cheng Cheng, S. M. Geyer and K. Y. S. Ng, J. Mater. Chem. A, 2016, 4, 13866–13873 RSC.
  25. H. Ou, L. Lin, Y. Zheng, P. Yang, Y. Fang and X. Wang, Adv. Mater., 2017, 29, 1700008 CrossRef PubMed.
  26. B. Ravel and M. Newville, J. Synchrotron Radiat., 2005, 12, 537–541 CrossRef CAS PubMed.
  27. J. J. Rehr and R. C. Albers, Rev. Mod. Phys., 2000, 72, 621–654 CrossRef CAS.
  28. G. Kresse and J. Furthmüller, Phys. Rev. B: Condens. Matter, 1996, 54, 11169–11186 CrossRef CAS PubMed.
  29. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868 CrossRef CAS PubMed.
  30. Z. Chen, P. Sun, B. Fan, Z. Zhang and X. Fang, J. Phys. Chem. C, 2014, 118, 7801–7807 CrossRef CAS.
  31. O. F. Ikotun, N. Marino, P. E. Kruger, M. Julve and R. P. Doyle, Coord. Chem. Rev., 2010, 254, 890–915 CrossRef CAS.
  32. B. Li, P. Gu, Y. Feng, G. Zhang, K. Huang, H. Xue and H. Pang, Adv. Funct. Mater., 2017, 27, 1605784 CrossRef.
  33. P. Yang, H. Ou, Y. Fang and X. Wang, Angew. Chem., Int. Ed., 2017, 56, 3992–3996 CrossRef CAS PubMed.
  34. Y. Wang, F. Silveri, M. K. Bayazit, Q. Ruan, Y. Li, J. Xie, C. R. A. Catlow and J. Tang, Adv. Energy Mater., 2018, 8, 1801084 CrossRef.
  35. K. Schwinghammer, M. B. Mesch, V. Duppel, C. Ziegler, J. Senker and B. V. Lotsch, J. Am. Chem. Soc., 2014, 136, 1730–1733 CrossRef CAS PubMed.
  36. K. Li, Y. Zhang, Y.-Z. Lin, K. Wang and F.-T. Liu, ACS Appl. Mater. Interfaces, 2019, 11, 28918–28927 CrossRef CAS PubMed.
  37. M. David James, Q. Kaipei, S. Stephen Andrew, H. Albertus Denny, C. Xiaowei, G. Zhengxiao and T. Junwang, Angew. Chem., Int. Ed., 2015, 53, 9240–9245 Search PubMed.
  38. S. Cao, J. Low, J. Yu and M. Jaroniec, Adv. Mater., 2015, 27, 2150–2176 CrossRef CAS PubMed.
  39. C. Li, Y. Du, D. Wang, S. Yin, W. Tu, Z. Chen, M. Kraft, G. Chen and R. Xu, Adv. Funct. Mater., 2017, 27, 1604328 CrossRef.
  40. M. Ledendecker, S. Krick Calderón, C. Papp, H.-P. Steinrück, M. Antonietti and M. Shalom, Angew. Chem., Int. Ed., 2015, 54, 12361–12365 CrossRef CAS PubMed.
  41. C. Liu, B. C. Colon, M. Ziesack, P. A. Silver and D. G. Nocera, Science, 2016, 352, 1210–1213 CrossRef CAS PubMed.
  42. Y. Li and C. Zhao, Chem. Mater., 2016, 28, 5659–5666 CrossRef CAS.
  43. Y. Tan, H. Wang, P. Liu, Y. Shen, C. Cheng, A. Hirata, T. Fujita, Z. Tang and M. Chen, Energy Environ. Sci., 2016, 9, 2257–2261 RSC.
  44. A. Anspoks and A. Kuzmin, J. Non-Cryst. Solids, 2011, 357, 2604–2610 CrossRef CAS.
  45. P. D. Bonnitcha, M. D. Hall, C. K. Underwood, G. J. Foran, M. Zhang, P. J. Beale and T. W. Hambley, J. Inorg. Biochem., 2006, 100, 963–971 CrossRef CAS PubMed.
  46. R. Subbaraman and N. M. Markovic, Science, 2012, 43, 1256–1260 Search PubMed.
  47. X. Li, W. Bi, L. Zhang, S. Tao, W. Chu, Q. Zhang, Y. Luo, C. Wu and Y. Xie, Adv. Mater., 2016, 28, 2427–2431 CrossRef CAS PubMed.
  48. H. Fei, J. Dong, M. J. Arellano-Jimenez, G. Ye, N. Dong Kim, E. L. Samuel, Z. Peng, Z. Zhu, F. Qin, J. Bao, M. J. Yacaman, P. M. Ajayan, D. Chen and J. M. Tour, Nat. Commun., 2015, 6, 8668 CrossRef CAS PubMed.
  49. G. Ai, H. Li, S. Liu, M. Rong and J. Zhong, Adv. Funct. Mater., 2015, 25, 5706–5713 CrossRef CAS.
  50. Z. Zhuang, Y. Li, Z. Li, F. Lv, Z. Lang, K. Zhao, L. Zhou, L. Moskaleva, S. Guo and L. Mai, Angew. Chem., Int. Ed., 2018, 57, 2697–2701 CrossRef PubMed.
  51. Y. Yang, Z. Zeng, G. Zeng, D. Huang, R. Xiao, C. Zhang, C. Zhou, W. Xiong, W. Wang, M. Cheng, W. Xue, H. Guo, X. Tang and D. He, Appl. Catal., B, 2019, 258, 117956 CrossRef CAS.
  52. A. Jain, S. P. Ong, G. Hautier, W. Chen, W. D. Richards, S. Dacek, S. Cholia, D. Gunter, D. Skinner, G. Ceder and K. A. Persson, APL Mater., 2013, 1, 011002 CrossRef.
  53. M. Buscema, J. O. Island, D. J. Groenendijk, S. I. Blanter, G. A. Steele, H. S. van der Zant and A. Castellanos-Gomez, Chem. Soc. Rev., 2015, 44, 3691–3718 RSC.
  54. R. Shivanna, S. Shoaee, S. Dimitrov, S. K. Kandappa, S. Rajaram, J. R. Durrant and K. S. Narayan, Energy Environ. Sci., 2014, 7, 435–441 RSC.
  55. Y. Cao, S. Chen, Q. Luo, H. Yan, Y. Lin, W. Liu, L. Cao, J. Lu, J. Yang and T. Yao, Angew. Chem., Int. Ed., 2017, 56, 12191 CrossRef CAS PubMed.
  56. K. Li, Y.-Z. Lin, K. Wang, Y. Wang, Y. Zhang, Y. Zhang and F.-T. Liu, Appl. Catal., B, 2020, 268, 118402 CrossRef.
  57. Y. Shi and B. Zhang, Chem. Soc. Rev., 2016, 45, 1529–1541 RSC.
  58. J. Greeley, T. F. Jaramillo, J. Bonde, I. B. Chorkendorff and J. K. Norskov, Nat. Mater., 2006, 5, 909–913 CrossRef CAS PubMed.
  59. J. K. Norskov, T. Bligaard, J. Rossmeisl and C. H. Christensen, Nat. Chem., 2009, 1, 37–46 CrossRef CAS PubMed.
  60. Y. Zheng, Y. Jiao, M. Jaroniec and S. Z. Qiao, Angew. Chem., Int. Ed., 2015, 54, 52–65 CrossRef CAS PubMed.
  61. J. Ran, G. Gao, F. T. Li, T. Y. Ma, A. Du and S. Z. Qiao, Nat. Commun., 2017, 8, 13907 CrossRef CAS PubMed.
  62. Q. Gong, Y. Wang, Q. Hu, J. Zhou, R. Feng, P. N. Duchesne, P. Zhang, F. Chen, N. Han, Y. Li, C. Jin, Y. Li and S. T. Lee, Nat. Commun., 2016, 7, 13216 CrossRef CAS PubMed.
  63. W. Chen, J. Pei, C. T. He, J. Wan, H. Ren, Y. Wang, J. Dong, K. Wu, W. C. Cheong, J. Mao, X. Zheng, W. Yan, Z. Zhuang, C. Chen, Q. Peng, D. Wang and Y. Li, Adv. Mater., 2018, 30, 1800396 CrossRef PubMed.
  64. B. N. Hammer and J. K. Nørskov, Adv. Catal., 2000, 45, 71–129 CAS.
  65. X. Tian, X. Zhao, Y.-Q. Su, L. Wang, H. Wang, D. Dang, B. Chi, H. Liu, E. J. M. Hensen, X. W. Lou and B. Y. Xia, Science, 2019, 366, 850–856 CrossRef CAS PubMed.
  66. D. Y. Wang, M. Gong, H. L. Chou, C. J. Pan, H. A. Chen, Y. Wu, M. C. Lin, M. Guan, J. Yang, C. W. Chen, Y. L. Wang, B. J. Hwang, C. C. Chen and H. Dai, J. Am. Chem. Soc., 2015, 137, 1587–1592 CrossRef CAS PubMed.
  67. D. Y. Chung, S. W. Jun, G. Yoon, H. Kim, J. M. Yoo, K. S. Lee, T. Kim, H. Shin, A. K. Sinha, S. G. Kwon, K. Kang, T. Hyeon and Y. E. Sung, J. Am. Chem. Soc., 2017, 139, 6669–6674 CrossRef CAS PubMed.


Electronic supplementary information (ESI) available. See DOI: 10.1039/d0cy00556h

This journal is © The Royal Society of Chemistry 2020