Introduction

The increasing concerns on fossil fuel consumption and environmental problems call for clean and renewable energy sources [1, 2]. Hydrogen energy has received a broad attention and been identified as a desirable alternative to non-fossil fuel resource due to the advantages of high energy density, zero-emission, and abundant resource [3]. Recently, water splitting with high energy conversion and storage efficiency is used as an efficient and environmental technology to produce high-pure hydrogen [4,5,6,7,8]. However, its high overpotential cannot meet the requirement of anodic hydrogen evolution reaction (HER). Traditional noble metal-based catalysts exhibit negligible overpotential and outstanding durability but suffer from high cost and limited abundance, thereby hindering their large-scale industrial application [9,10,11,12,13]. This evokes a widespread search for non-noble metal-based alternatives, such as Co- [14], Ni- [15, 16], and Mo-based catalysts [17,18,19,20,21]. Particularly, Co-based catalysts serve as excellent examples of novel electrocatalysts for HER [3, 22].

Recent years have witnessed the rapid development of Co-based catalysts in HER [23, 24]. Originally, cobaloxime and diimine-dioxime compounds were reported by Artero et al., which showed a low overpotential for proton reduction when evaluated as molecular electrocatalysts for hydrogen evolution [25,26,27]. However, harsh synthesis condition significantly hinders their practical application. To address this issue, various Co-based catalysts, such as CoO [28], CoSe2 [29,30,31,32], CoP [33,34,35,36], CoPS [37, 38], Co2B [39], CuCo [40], and FeCo [41], have sprung up owing to simple preparation process, low cost, and excellent HER activity as well as durability relative to those Co-based complexes [42]. Nonetheless, most of them exhibit inferior electrocatalytic activities to platinum-based catalysts in overpotential and durability. Therefore, Co-based catalysts are still unable to substitute Pt-based catalysts in the short term.

Bimetallic catalysts combining platinum group elements with Co commonly deliver superior catalytic performance via cooperative interaction, which are not observable by their individual monometallic counterparts [3, 43,44,45,46]. Theoretical and experimental results have proved that alloying platinum group elements with Co could change the average energy of surface d-band and the width of d-band originated from cumulative strain and ligand effect. This can result in modification of the surface chemical properties and remarkably promote the HER activity [3, 43, 47, 48]. Among them, palladium has been considered to form alloy due to its higher affinity for hydrogen and relatively cheaper price than platinum [2]. In addition, integrating alloy with carbon substrates, especially encapsulating them in heteroatom-doped carbon, can enhance HER performance [43]. Despite of great advantages, the complex preparation procedure for most materials makes it urgent to search for relatively simple and effective approaches to realize the encapsulation of alloy in carbon layers.

Metal–organic frameworks (MOFs), a class of novel crystalline material assembled through coordination interaction between metal ions/clusters and organic linkers, have proved to be potential precursors to synthesize heteroatom-doped porous carbon stabilized metal/metal compounds for HER. The as-formed materials preserve metal source, abundant carbon and heteroatom species, high surface area, tunable porosity, and well-organized nanostructure [39, 49,50,51,52,53,54,55]. To date, RuCo-based catalysts [3] and PtCo-based catalysts [45] have been synthesized by using noble metal-doped MOFs as precursors. However, choosing Pd-doped MOFs as precursors to synthesize PdCo alloys wrapped in N-doped porous carbon was just reported by Chen et al. [43]. The as-prepared catalyst delivers remarkable HER activity with low overpotential and Tafel slope as well as excellent cycle stability in 0.5 M H2SO4 media than that of Co wrapped N-doped porous carbon. Therefore, it can be estimated that MOFs can be pyrolyzed to obtain the composites consisting of heteroatom-doped porous carbon and nanosized alloys through introducing noble metals into MOFs.

In this work, we take a facile MOF-assisted strategy to synthesize a new kind of composite consisting of PdCo alloy nanoparticles encapsulated with N-doped porous carbon and reduced graphene oxide substrates (denoted as PdCo@N–C/rGO) by pyrolyzing Pd-incorporated ZIF-67/GO precursors under N2 atmosphere. The carbon shell effectively stabilizes PdCo alloys and prevents them from aggregation and corrosion during HER process. Meanwhile, rGO sheets provide ideal substrates for electrocatalysts due to their large surface area, good conductivity, and high stability. Benefiting from the synergistic effect of the PdCo alloy, N-doped porous carbon, and rGO components, the as-synthesized PdCo@N–C/rGO material exhibits outstanding HER performance in acidic, alkaline, and neutral conditions.

Experimental

Chemicals

All the reagents and solvents, including graphene oxide (GO, XFNANO Materials Tech Co., Ltd, Nanjing), sodium chloropalladite (AR, Aladdin), L-ascorbic acid (AR, > 99.0%, Aladdin), polyvinylpyrrolidone (PVP, M = 24,000, Aladdin), potassium chloride (KCl, Bodi Chemicals, Tianjin), potassium bromide (KBr, Regent Chemical, Tianjin), cobalt nitrate hexahydrate (Co(NO3)2·6H2O, AR, Aladdin), 2-methylimidazole (AR, Aladdin), methanol (CH3OH, 99.5%, Hengshan Chemicals, Tianjin), and distilled water were commercially obtained and used without further purification.

Preparation of Pd nanoparticles

Pd nanoparticles were prepared by following the reported procedure [56]. Briefly, an aqueous solution (8 mL) of PVP (105 mg), L-ascorbic acid (60 mg), KCl (185 mg), and KBr (5 mg) was added into an aqueous solution (3 mL) of sodium chloropalladite (57 mg). The mixture was stirred for 10 min and then agitated vigorously at 80 °C for 3 h. After cooling down to room temperature, the resultant black precipitate was separated by diethyl ether and then was dispersed in methanol (8 mL) for further use.

Preparation of Pd-ZIF-67/GO

In a typical procedure, 2 ml of methanol solution of Pd was firstly mixed with 30 mL of methanol solution of GO (1 mg mL−1). Then, 50 mL of methanol solution of 2-methylimidazole (0.5820 g) was added into the above mixture with continuous stirring for 1 h followed by quick pour of a 50 mL of methanol solution of Co(NO3)2·6H2O (0.6562 g). Finally, the mixture was kept undisturbed for 24 h. The resultant composite was collected by centrifugation, washed with methanol for at least five times, and then dried at 60 °C for 12 h. The as-obtained sample was denoted as Pd-ZIF-67/GO.

Preparation of PdCo@N–C/rGO

The as-synthesized Pd-ZIF-67/GO precursors were transferred into a tube furnace and maintained 350 °C for 1.5 h. Following, the pyrolysis temperature was further raised to 900 °C at a ramp rate of 5 °C min−1 and kept for 3.5 h under a nitrogen atmosphere. After cooling down to room temperature, the PdCo@N–C/rGO sample was collected and stored. By comparison, Co@N–C/rGO and PdCo@N–C samples were also prepared by pyrolysis of ZIF-67/GO and Pd-ZIF-67 precursors under the same conditions, respectively.

Material characterization

Power X-ray diffraction (PXRD) patterns were recorded in the range of 3–80° by step scanning with a Rigaku Mini Flex600 X-ray diffraction meter at 40 kV, 15 mA with a Cu-target tube. The morphology of samples was characterized by field emission scanning electron microscopy (FESEM, JEOL JSM-7500F), transmission electron microscope (TEM), high-resolution transmission electron microscopy (HRTEM), and energy-dispersive X-ray (EDX) (JEOL, JEM-1200EX, 20–120 kV). X-ray photoelectron spectroscopy (XPS) was operated on a Thermo Scientific ESCALAB 250Xi Versaprobe system. Inductively coupled plasma (ICP, Thermo Electron Corporation, X7) and elemental analysis (Vario EL cube) were used to quantitatively confirm the content of each element. Raman spectra were collected on a RTS-HiR-AM with Raman Laser source of 532 nm. Nitrogen adsorption–desorption measurement was performed on an ASAP 2020 (Micromeritics) at 77 K, and the BET method was utilized to calculate the specific surface areas (SBET). Pore size distribution was calculated based on Density Functional Theory (DFT) using the data from the adsorption branch.

Electrochemical HER measurement

All the electrochemical HER measurements were performed in a standard three-electrode system on a CHI760E electrochemical workstation (Shanghai Chenhua Co., China) in 0.5 M H2SO4, 1.0 M KOH, and 1.0 M KHCO3 solutions. The working electrodes were prepared as follows: 5 mg of catalyst and 20 µl of 5 wt% Nafion solution were ultrasonically dispersed in 280 μL of ethanol/water mixed solvent (9:5 v/v) for at least 1 h to form a homogeneous ink. Then, 5 μL of catalyst ink (catalyst loading: ca. 0.66 mg cm−2) was loaded onto the glassy carbon electrode surface and then dried for measurement. Prior to HER measurements, the electrolyte was bubbled with N2. All of the potentials were calibrated to a RHE. The HER polarization curves were measured with a scan rate of 5.0 mV s−1. Tafel slope was obtained by plotting overpotential against the logarithm of current density (log |J|). Chronoamperometry measurement was performed to evaluate the long-term durability. To estimate the electro-chemical active surface area of catalyst, the HER cyclic voltammograms were obtained by ranging potential from −0.25 to −0.15 V (vs. SCE, in 0.5 M H2SO4) with various scan rates of 10, 20, 40, 60, 80, 100, 120, 140, 160, 180, and 200 mV s−1. By plotting the Δj = ja−jc against the CV scan rate, the linear slope that is twice of the double layer capacitance (Cdl) is used to represent electrochemical active surface area (ECSA).

Results and discussion

The general synthesis process of PdCo alloy-encapsulated N-doped porous carbon supported on rGO substrate is presented in Scheme 1. Firstly, Pd-ZIF-67/GO precursor was prepared by using Pd nanoparticles, Co(NO3)2·6H2O, methylimidazole (MeIM), and graphene oxide (GO) as starting materials. The power X-ray diffraction (PXRD) patterns unambiguously reveal the successful preparation of Pd-ZIF-67/GO. All peaks are in good agreement with simulated ZIF-67 and Pd (card: JCPDS no. 65–6174), as shown in Figure S1 (Supporting Information). The field-emission scanning electron microscopy (FESEM) images show that Pd-incorporated ZIF-67 microcrystal is nearly rhombic and supported onto GO substrate, as shown in Figure S2 (Supporting Information). Subsequently, Pd-ZIF-67/GO precursors were pyrolyzed at 350 °C followed by 900 °C in N2 atmosphere, resulting in the formation of PdCo@N–C/rGO material. At a low annealing temperature, the solvent molecules can escape from the precursor to form abundant micropores and mesopores. During the high carbonization temperature, PdCo alloy could be encapsulated by N-doped carbon derived from ligands, and GO sheets were simultaneously reduced to rGO. The rGO not only ensures good electrical conductivity but also acts as a substrate to make ZIF-67-derived N-doped carbon uniformly loaded [57].

Scheme 1
scheme 1

Schematic illustration for the synthetic process of PdCo@N–C/rGO material

Full size image

The morphology and phase characteristics of PdCo@N–C/rGO composite were primarily examined by FESEM and transmission electron microscopy (TEM). As shown in Fig. 1a, b, the general morphology of PdCo@N–C/rGO is different from the original ones after thermal treatment. The rGO sheets show a shrunk and crinkled structure, while Pd-incorporated ZIF-67 octahedrons become rough and are dispersed on rGO sheets uniformly (Fig. 1c, d). At this point, PdCo alloy nanoparticles are also formed. The size distribution of PdCo nanoparticles mainly ranges from 20 to 80 nm. The high-resolution TEM (HRTEM) images demonstrate that PdCo nanoparticles are encapsulated by ZIF-derived carbon shell (Fig. 1e, f). The surface element mapping result unambiguously verifies that PdCo@N–C/rGO is composed of Pd, Co, N, and C elements and they are homogeneously distributed (Fig. 1g). The BET surface area and pore size distribution characteristics of as-prepared PdCo@N–C/rGO and Co@N–C/rGO samples were examined by nitrogen adsorption/desorption measurement. The isotherm of PdCo@N–C/rGO exhibits typical microporous and mesoporous characteristics with a type-IV curve. The Brunauer–Emmett–Teller (BET) surface area is 245.9 m2 g−1, which is near to that of Co@N–C/rGO (224.9 m2 g−1), indicating that the introduction of Pd has a little impact on the surface area (Figure S3, Supporting Information). According to inductively coupled plasma optical emission spectrometry (ICP-OES) analysis, the weight contents of Pd and Co for PdCo@N–C/rGO were determined to be 2.0% and 29.8%, respectively.

Figure 1
figure 1

(a, b) SEM, (c, d, e) TEM, (f) HRTEM images, and (g) TEM image with corresponding elemental mapping of C, N, Co, and Pd of as-synthesized PdCo@N–C/rGO composite

Full size image

The PXRD pattern reveals the phase component of PdCo@N–C/rGO. As shown in Fig. 2a, a broad peak at 26.5° (2θ) is assigned to a characteristic of graphite carbon (002) peak with good graphitization. In addition, the PdCo@N–C/rGO presents three peaks at 43.6°, 50.5°, and 74.7° (2θ), which are typically assigned to (111), (200), and (220) of cobalt palladium in Fm-3 m cubic system (JCPDS: no. 65–6174). Compared with simulated PXRD patterns of cobalt (JCPDS: no. 88–2325) and Palladium (JCPDS: no. 65–6174), the peak locations of PdCo@N–C/rGO shift, indicating that Pd is successfully incorporated into the Co cubic structure to form an alloy phase [58, 59]. The structural feature of the carbonized samples was also characterized by Raman spectroscopy. As shown in Fig. 2b, PdCo@N–C/rGO shows a slight peak-shift at ca. 670 cm−1 compared with Pd or Co@N–C/rGO, indicating the existence of strong interaction between Pd and Co, which is in accordance with PXRD results. Meanwhile, two characteristic peaks at 1357 cm−1 and 1578 cm−1 correspond to typical D and G bands of PdCo@N–C/rGO, which represent characteristic features of disordered and graphitic carbon, respectively [57]. The ID/IG ratio is ca. 0.86, indicating a high graphitization degree. This is ascribed to the introduction of rGO component. X-ray photoelectron spectroscopy (XPS) was employed to investigate the chemical state and effect of Pd on electron structure of Co. As shown in Fig. 2c, the full XPS spectrum confirms the existence of Pd, Co, C, and N, which is in good accordance with energy-dispersive X-ray (EDX) elemental mapping result. In high-resolution N1s spectrum of PdCo@N–C/rGO, two obvious peaks are deconvoluted into four individual peaks with binding energies at 398.8, 399.8, 401.1, and 402.2 eV, which are assigned to pyridinic-N, pyrrolic-N-metal, pyrrolic-N, and quaternary-N, respectively [3, 60]. In high-resolution Pd 3d spectra (Fig. 2e), both Pd 3d5/2 and 3d3/2 peaks of PdCo@N–C/rGO positively shift to higher binding energies compared with Pd, indicating the existence of chemical interaction between Pd and Co in PdCo@N–C/rGO [61]. The binding energy in PdCo@N–C/rGO could originate from the d-band hybridization between metal atoms upon alloying owing to the fact that stronger bonds can be formed in the d-orbital, which thus reduces the ability to form strong bonds with adsorbed reactants [62]. Similarly, the high-resolution Co2p spectra of PdCo@N–C/rGO and Co@N–C/rGO are presented in Fig. 2f. Each Co2p peak can be resolved into two pairs of doublets, and two obvious Co0 and CoII species peaks, indicating the existed CoII comes from surface oxidation of metallic Co [3]. In addition, both Co 2p3/2 and 2p1/2 peaks of PdCo@N–C/rGO shift to higher binding energies relative to the Co 2p3/2 and 2p1/2 peaks of Co@N–C/rGO, further demonstrating the strong electron interaction between Pd and Co, which probably facilitates the electrocatalytic activity and durability for HER.

Figure 2
figure 2

a XRD pattern of as-prepared PdCo@N–C/rGO; b Raman spectra of PdCo@N–C/rGO, Co@N–C/rGO and Pd samples; c XPS spectrum; and d high resolution N 1 s spectrum of PdCo@N–C/rGO sample; e high-resolution Pd 3d and f Co 2p spectra of PdCo@N–C/rGO, Co@N–C/rGO and Pd samples

Full size image

Intrigued by the structural and component features, PdCo@N–C/rGO composite was evaluated as an electrocatalyst for HER in 0.5 M H2SO4 electrolyte by using a typical three-electrode electrochemical system. By comparison of HER performance, Pd-ZIF-67 and ZIF-67/GO were prepared and went through the same treatment. The final products were denoted as PdCo@N–C and Co@N–C/rGO (the structures and morphologies were confirmed by XRD and TEM technique as shown in Figures S4 and S5, Supporting Information). The electrocatalytic performance was firstly evaluated by an overpotential value versus reversible hydrogen electrode (RHE) at 10 mA cm−2, which is considered to reach 12.3% efficient solar water-splitting device at this current density [3, 63]. Figure 3a presents the polarization curves of samples with a scan rate of 5 mV s−1. Observably, PdCo@N–C/rGO exhibits more positive onset potential than Co@N–C/rGO, which is near to the thermodynamic potential of H2 evolution. In addition, the overpotential value of PdCo@N–C/rGO is measured to be 87 mV at a current density of 10 mA cm−2, which is close to the value of commercial Pt/C catalyst (49 mV) among these samples. Obviously, the overpotential of PdCo@N–C/rGO is lower than that of PdCo@N–C (122 mV) and Co@N–C/rGO (252 mV). It indicates that PdCo alloy, N-doped porous carbon, and rGO play indispensable roles in the improvement in electrochemical performance for HER.

Figure 3
figure 3

(a) Polarization curves and (b) Tafel plots of PdCo@N–C/rGO, PdCo@N–C, Co@N–C/rGO, and commercial Pt/C in 0.5 M H2SO4; (c) estimation of Cdl by plotting the current density variation (160 mV vs. RHE); (d) chronoamperometric curve (I vs. t) of PdCo@N–C/rGO at an applied potential of −0.158 V (vs. RHE)

Full size image

Subsequently, the corresponding kinetics of the above-mentioned catalysts were evaluated by Tafel plots measured in a 0.5 M H2SO4 electrolyte. Typically, the hydrogen evolution mechanism in acidic electrolyte mainly contains three principal steps including the Volmer (Eq. 1), the Heyrovsky (Eq. 2), and the Tafel step (Eq. 3) [56], as listed as follows:

$${{\text{H}}_3}{{\text{O}}^+ } + {{\text{e}}^- } + * \to {{\text{H}}^* } + {{\text{H}}_2}{\text{O}}$$
(1)
$${{\text{H}}_3}{{\text{O}}^+ } + {{\text{e}}^- } + {{\text{H}}^* } \to {{\text{H}}_2} + {{\text{H}}_2}{\text{O}}$$
(2)
$${{\text{H}}^* } + {{\text{H}}^* } \to {{\text{H}}_2}$$
(3)

As shown in Fig. 3b, the Tafel slope of commercial Pt/C (32 mV dec−1) demonstrates that the typical Volmer–Tafel mechanism happened. Under the same conditions, the PdCo@N–C/rGO catalyst presents a Tafel slope value of 66 mV dec−1, which demonstrates a Volmer–Heyrovsky mechanism occurred in hydrogen evolution process of PdCo@N–C/rGO [1, 58]. Moreover, the Tafel slope of PdCo@N–C/rGO is smaller than that of PdCo@N–C and Co@N–C/rGO (91 and 118 mV dec−1), and this value is better than most of those reported Co-based catalysts (Table S1, Supporting Information). The reason could be ascribed to the smaller electron transport barrier for PdCo@N–C/rGO, in which PdCo alloy core can transfer more number of electrons to the nitrogen-doped carbon shell and graphene than pure metal core in order to enhance C-H bond, which will decline free energy ΔGH* and thereby enhance the electrocatalytic activity [3].

The electrochemical active surface areas of PdCo@N–C/rGO, PdCo@N–C, and Co@N–C/rGO were evaluated by electrochemical double-layer capacitance (Cdl) (Fig. 3c). The Cdl was simulated by using a cyclic voltammetry (CV) method (Figure S6, Supporting Information). The results show that the Cdl of PdCo@N–C/rGO is much higher than that of Co@N–C/rGO and PdCo@N–C, implying the abundant electrochemical active surface area on PdCo@N–C/rGO catalyst. Such an increasement in Cdl endows PdCo@N–C/rGO with excellent HER activity. Electrochemical impedance spectroscopy (EIS) was conducted to further explain the enhanced performance of PdCo@N–C/rGO, and corresponding Nyquist plots of three catalysts are shown in Figure S7. Commonly, the semicircle in low frequency region is related to the charge-transfer resistance (Rct). As can be seen, PdCo@N–C/rGO exhibits smaller diameter than that of Co@N–C/rGO and PdCo@N–C, indicating smaller Rct and easier electron transfer in PdCo@N–C/rGO electrode, which thus contributes to highly electrochemical activity. In addition, the stability of PdCo@N–C/rGO catalyst was evaluated by Chronoamperometry (CA) test and TEM characterization after cycling. As shown in Fig. 3d, the catalytic activity shows a slight reduction over 40 h at a catalytic current of nearly 10 mA cm−2 with an applied potential of −0.158 V (vs. RHE), demonstrating its good durability in acidic electrolyte. The morphology of PdCo@N–C/rGO after cycling was well preserved with the initial morphology, and the EDX mapping presents the uniform distribution of C, N, Co, and Pd elements (Figure S8, Supporting Information), indicating its structural stability.

Combining the low overpotential, small Tafel slope, and good durability of PdCo@N–C/rGO, the improved catalytic performance could be attributed to the synergistic effect among PdCo alloys, N-doped porous carbon shell, and rGO substrate. Firstly, rGO substrate offers good conductivity and large surface area to facilitate mass diffusion and electron transport on PdCo@N–C/rGO catalyst. Secondly, PdCo alloy nanoparticles provide more active sites to enhance quick adsorption and desorption of H2 in acidic electrolyte. Thirdly, ZIFs-derived N-doped porous carbon shells make PdCo nanoparticles disperse uniformly and prevent them from corrosion, which are beneficial for enhanced stability and durability of PdCo@N–C/rGO in acidic system [1].

In addition, the electrocatalytic performance of obtained catalysts was also investigated in alkaline and neutral media. Similarly, the PdCo@N–C/rGO catalyst exhibits the lowest overpotential and smallest Tafel slope than that of PdCo@N–C and Co@N–C/rGO with the values of 251 mV, 149 mV dec−1 in 1.0 M KOH (Figure S9a, S9b) and 115 mV, 41 mV dec−1 in 1.0 M KHCO3 (Figure S9d, S9e). In addition, EIS analysis also verifies that PdCo@N–C/rGO possesses the lowest charge transfer resistance than Co@N–C/rGO and PdCo@N–C (Figure S9c, S9f), corroborating the high HER performance of PdCo@N–C/rGO.

Conclusion

In summary, a PdCo@N–C/rGO composite was successfully prepared through one-step pyrolysis of Pd-incorporated ZIF-67/GO precursor and applied as an electrocatalyst in HER. The synergetic effect among PdCo alloy, N-doped porous carbon shell, and electrochemically conductive rGO support endows the as-made PdCo@N–C/rGO catalyst with excellent HER performance in acidic, alkaline, and neutral electrolytes, including low overpotential (87, 251, 115 mV at 10 mA cm−2), small Tafel slope (66, 150, 41 mV dec−1), and long-term durability. Meanwhile, the PdCo@N–C/rGO catalyst shows advantages over PdCo@N–C and Co@N–C/rGO, respectively. This work might bring new opportunities to develop alloy-based HER catalysts via MOF pyrolysis strategy.