Alkaline water-splitting reactions over Pd/Co-MOF-derived carbon obtained via microwave-assisted synthesis

Cobalt-based metal–organic framework-derived carbon (MOFDC) has been studied as a new carbon-based support for a Pd catalyst for electrochemical water-splitting; i.e., the hydrogen evolution reaction (HER) and oxygen evolution reaction (OER) in alkaline medium. The study shows a high increase in the HER activity, in terms of low onset overpotential (onset η = 35 mV vs. RHE), high exchange current density (jo,s ≈ 0.22 mA cm−2), high mass activity (jo,m ≈ 59 mA mg−1), high kinetic current (jK ≈ 5–8 mA cm−2) and heterogeneous rate constant (k0 ≈ 4 × 10−4 cm s−1), which are attributed to the high porosity of MOFDC and contribution from residual Co, while the large Tafel slope (bc = 261 mV dec−1) is ascribed to the high degree of hydrogen adsorption onto polycrystalline Pd as a supplementary reaction step to the suggested Volmer–Heyrovsky mechanism. These values for the catalyst are comparable to or better than many recent reports that adopted nano-carbon materials and/or use bi- or ternary Pd-based electrocatalysts for the HER. The improved HER activity of Pd/MOFDC is associated with the positive impact of MOFDC and residual Co on the Pd catalyst (i.e., low activation energy, EA ≈ 12 kJ mol−1) which allows for easy desorption of the Hads to generate hydrogen. Moreover, Pd/MOFDC displays better OER activity than its analogue, with lower onset η (1.29 V vs. RHE) and ba (≈78 mV dec−1), and higher current response (ca. 18 mA cm−2). Indeed, this study provides a new strategy of designing and synthesizing MOFDC with physico-chemical features for Pd-based electrocatalysts that will allow for efficient electrochemical water-splitting processes.


Introduction
Hydrogen is an excellent fuel for the realization of clean and renewable energy. Hydrogen evolution reaction (HER) via electrochemical water-splitting process has become important in a world that strives for a clean environment and urgent mitigation of the negative consequences of climate change brought about by greenhouse gases. Efficient HER is generally dependent on the nature of electrocatalysts that drive the reaction.
In alkaline media, the electrocatalyst requires energy to break the H-O-H bonding before chemisorption and desorption of hydrogen intermediate (H ads ) to produce H 2 , as illustrated in eqn (1)- (3). 1 The Volmer is the rate-determining step (RDS), which is initial H-O-H cleavage, followed by chemisorption of the H ads onto active sites (M) of noble metal electrocatalyst, while the HER subsequently occur either by eqn (2) (Tafel) or (3) (Heyrovsky) depending on the nature of the electrocatalyst and values of Tafel slope for H 2 production, as the H ads recombine to yield H 2 or the catalyst continuously cleave water molecule and desorbed the H ads to produced H 2 , respectively. The latter (eqn (3)) may mostly be preferred for efficient HER electrocatalysis in alkaline conditions because of a limited amount of H ads which could only be generated by splitting water. 2 M þ H 2 O þ e À /MH ads þ OH À Volmer; b c ¼ 2:3RT aF ¼ 118:2 mV dec À1 (1) HER is extremely sluggish in alkaline compared to the acidic media due to the initial water dissociation, instability of adsorbed hydrogen (H ads ) and ultimately large overpotential, 2 hence explaining the preference of platinum (Pt)-based electrocatalysts for efficient catalysis of HER in alkaline electrolyte. According to density functional theory (DFT), the theoretical H 2 binding energy (E ads ) of palladium (Pd(111), E ads ¼ À0.48 eV) is close to that of Pt (Pt(111), E ads ¼ À0.43 eV), with excellent HER activities. 3 Moreover, Pd-based electrocatalysts have proved to exhibit improved electrocatalytic performance in alkaline conditions. 4 Hence, Pd-based electrocatalysts are viable substitute electrocatalysts for HER in an alkaline medium because of their high selectivity for H 2 binding, rapid desorption of H ads , quick water splitting and transformable formation of H ads to produce highly puried H 2 gas. 5 Presently, several attempts are being made to reduce the cost of HER by using low-Pd loaded electrocatalyst on various carbon supports including carbon bre, 6 carbon paper substrate, 7 Vulcan XC-72, 8 graphitic nanober, 9 and reduced graphene oxide, 10 to mention a few. Although reports on these carbonbased supports show good results, a lot still needs to be done to enhance the electrocatalytic performance in terms of current density, energetics and kinetics. To our knowledge, there are little reports on the use of metal-organic framework-derived carbons (MOFDC) for HER and OER in alkaline electrolyte. 11 The choice of Co-MOF sacricial template to synthesize novel carbon support (MOFDC) is because of the interesting inherent properties of MOF such as tunable pore size, enhanced surface area, electron density, electrical conductivity and thermal stability, and the ease of Co removal aer carbonization by etching with acid. 12,13 This work describes the rst report on the effect of MOFDC as a support for Pd electrocatalyst for HER activities. It is clearly shown that MOFDC signicantly enhances both the energetic and kinetics of HER when compared to its analogue and many literature reports using other carbon-based supports.

Experimental
Nanostructured palladium electrocatalyst supported on MOFDC (Pd/MOFDC) was synthesized using previously reported microwave-assisted synthesis (summarized in Fig. 1). 14 In brief, cobalt nitrate hexahydrate (Co(NO 3 ) 6 $6H 2 O, 0.4366 g), 4,4 0biphenyldicarboxylic acid (BPDC, 0.3634 g) and triethylamine (TEA, 1.5 mL) were mixed in a ask containing dimethylformamide (DMF, 50 mL) and stirred for 15 min to obtain a homogeneous mixture. The mixture was then heated in a microwave (600 W, 150 C, 30 min) forming precipitates, which were washed and dried in vacuum (60 C) to obtain solid cobalt-based metal-organic frameworks (Co-MOF). The Co-MOF was then annealed from room temperature to 600 C at 5 C min À1 and then to 800 C and maintained for 5 h. The resulting solution formed was cobalt nanoparticles on metalorganic frameworks-derived carbon (Co/MOFDC). The Co/ MOFDC was then chemically etched with hydrochloric acid (3.0 M, 10 mL) to remove the cobalt nanoparticles, resulting in MOFDC. Pd nanoparticles (20 wt%) were dispersed onto the MOFDC by following this procedure: potassium tetrachloropalladate crystals (K 2 PdCl 4 , 61.35 mg) were dissolved in EG (50 mL) and stirred for 30 min and pH of the solution was raised to 12 using NaOH (1.0 M). Then, the MOFDC (80 mg) was added and further stirred for 30 min. The mixture was irradiated with MW (600 W, 140 C, 1 h) to disperse Pd nanoparticles on the support, aer which the pH of the resulting product (Pd/ MOFDC) was lowered to 3, using nitric acid (HNO 3 , 0.1 M). The nal product was washed and dried (80 C, 4 h) in a vacuum oven. A similar procedure was followed for making Pd/carbon black (CB). The physical properties of the materials and electrocatalysts were thoroughly probed using Raman spectroscopy (Horiba LabRam HR Raman spectrometer equipped with Olympus BX41 microscope), powdered X-ray diffraction (Bruker D2 XRD), transmission electron microscope (FEI Tecnai T12 Sprint TEM), Brunauer-Emmett-Teller (BET, Micrometric Tristar 3000) and thermal gravimetric analysis (PerkinElmer TGA 4000). All electrochemical measurements were carried out using a rotating disk electrode (RDE) with AUTO LAB PGSTAT204. The HER kinetics on the Pd/MOFDC and Pd/CB were investigated utilizing a conventional three-electrode system: the working electrode is a glassy carbon electrode (GCE, 5 mm diameter) modied with the Pd/MOFDC (GCE-Pd/MOFDC) and Pd/CB (GCE-Pd/CB) while platinum wire and Ag|AgCl, 3.0 M KCl were employed as counter and reference electrodes, respectively. Before the measurements, electrocatalyst inks were prepared and modied on the GCE as follow: 2 mg of each electrocatalyst powder was dispersed in a mixture of ethanol (2 mL) and Naon (5 wt%, 50 mL) and ultrasonically mixed for uniform distribution. Each electrocatalyst ink (20 mL) was coated on the polished surface of the GCE and allowed to dry in the air to afford Pd catalyst loading (0.0199 mg cm À2 ). The HER experiment was carried out in freshly prepared KOH electrolyte (0.1 M) at a controlled and varied temperature in water-bath thermostat. Moreover, electrochemical impedance spectroscopy (EIS) was used to conrm the HER kinetics at room temperature, 1600 rpm, overpotential 0.2 V vs. RHE and frequency range (100 kHz to 0.1 Hz). All potentials are referred to reversible hydrogen electrode (RHE) with conversion formula: RHE ¼ E Ag|AgCl + 0.197 + 0.059 pH, where pH ¼ 12.57.

Physical characterization: Raman, XRD, BET and TEM measurements
The chemical bond vibrations of the Co-MOF, Co/MOFDC, MOFDC and Pd/MOFDC were studied using Raman spectroscopy (Fig. 2a). The spectra show the Co-O linkage and the organic frameworks conrmed the synthesis of the Co-MOF. 14 which was broken down aer carbonization at high temperature, conrmed with TGA ( Fig. 2b shows the XRD patterns of the Co-MOF, calcined product (Co/MOFDC) and derived carbon (MOFDC) obtained aer chemical etching. The XRD pattern of the Co-MOF displays a wellcrystalline material, which is fairly like the Co-MOF synthesized using benzene-1,3,5-tricarboxylic acid linkage. 15 The slight difference in the crystallinity is ascribed to the difference in the architectural structure of chelating agents. Aer pyrolysis, the original peaks are replaced by series of peaks at 25.8 for hexagonal graphitic carbon (002), while the peaks at 36.5 , 42.6 , 44.0 , 51.7 , and 61.6 are attributed to a mixture of Co and Co 3 O 4 nanoparticles with miller indices Co 3 O 4 (311), Co 3 O 4 (100), Co (111), Co (200) and Co 3 O 4 (440) respectively. 16 The chemically etched Co/MOFDC conrm that most Co and Co 3 O 4 nanoparticles had been removed, giving rise to a pronounced carbon peak at 25.8 for the MOFDC (002) with little trace of peaks of Co/Co 3 O 4 remaining. These results agree with the Raman spectral (Fig. 2a). The XRD patterns of the Pd/MOFDC and Pd/CB are described in Fig. 2c. As observed, the hexagonal carbon (002) at 25.6 remains for the Pd/MOFDC and Pd/CB. The electrocatalysts exhibit well-dened crystalline peaks for the face-centered cubic structure of Pd nanoparticles at around 40.0 , 46.5 , 68.0 for Pd(111), Pd(200) and Pd(220), respectively. 17, 18 Fig . 3 illustrates the TGA of the Co-MOF, Co/MOFDC and MOFDC. The TGA prole of the as-synthesized Co-MOF in air displays two decomposition steps. The rst step indicates weight loss from room temperature to 134 C, showing loss of water molecules from the pores of the Co-MOF because the Co-MOF has strong adsorption for water molecules. 16 The second weight loss is observed from 372 C to 541 C, ascribable to the decomposition of the BPDC frameworks of the Co-MOF and formation of Co/MOFDC. These results are consistent with the previous report for Co-based MOF using 2,5-dihydroterephthalic acid (DHTP) 16 and corroborate the Raman (Fig. 2a) and XRD (Fig. 2b) analysis for the decomposition of the Co-MOF frameworks at high temperature. The formation of Co nanoparticles in the second phase of the Co-MOF was conrmed from the TGA curve of the Co/MOFDC, where some of the metallic Co nanoparticles are oxidized between 400 C and 541 C to Co 3 O 4 nanoparticles, as the analysis was done in the air, and Co residue (86.6 wt%) remains aer 900 C. The TGA prole of the MOFDC shows stability to 420 C, where weight loss gradually begins to 800 C and Co metal residue (6.5 wt%) remains aer 900 C. This observation is like the report obtained for carbon microsphere. 19 Fig . 4 compares the TEM micrographs of the MOFDC and Pd/MOFDC. The morphology of the MOFDC with no traces of metal nanoparticles, as seen in Fig. 4a. Pd nanoparticles are revealed in Fig. 4b. The average particle size of Pd in the Pd/ MOFDC is estimated to be 6.8 nm (Fig. 4c).
The specic surface area (S BET ) of the Pd/MOFDC, MOFDC and Co-MOF are determined using BET analysis, summarized in Table  1. The Co-MOF has the lowest surface area and pore volume, but the highest pore size. Aer carbonization and chemically etched, the surface area and pore volume increased for the MOFDC, but the pore size reduced. Deposition of Pd nanoparticles on the   MOFDC increased slightly the surface area and pore size while the pore volume decreased. The increased surface area and pore size in the Pd/MOFDC may be traced to microwave irradiation during its synthesis, while the decreased pore volume is expected as Pd nanoparticles ll some of the pores. 20 3.2 Electrochemical measurements 3.2.1 Hydrogen evolution reaction (HER) study. Electrochemical activities of the Pd/MOFDC and Pd/CB electrocatalysts toward HER were obtained using LSV at various rotation speeds (400, 900, 1600, 2500, 3600 and 4900 rpm) at a scan rate of 5 mV s À1 in N 2 -deaerated 0.1 M KOH electrolyte. Cathodic (HER) polarization curves of bare GCE and the electrocatalysts at 1600 rpm and other rotations for the electrocatalysts are illustrated in Fig. 5a and b, respectively. Current densities of the electrocatalysts increased for the HER in alkaline electrolyte, as the rotation speeds increased. The response of the Pd/MOFDC electrocatalyst is more than the Pd/CB at 1600 rpm (Fig. 5a and all other rotations (Fig. 5b). These observations could be traced to the modication of electron properties of the Pd by the increased defect of the MOFDC (Fig. 2a) and contributions from the Co metal residue TGA (Fig. 3). The electrochemical active surface area (ECSA) of the Pd/MOFDC and Pd/CB electrocatalysts was determined by running CV in KOH solution only, then PdO reduction for the catalysts was integrated, evaluated and are previously recorded as 263.4 cm 2 mg À1 and 235.3 cm 2 mg À1 , respectively. 14 To understand the HER kinetics, the polarization curves were modelled to the typical Tafel equation, as given in eqn (4) and (5).
Taking the log of eqn (4), re-arranging and making the overpotential (h/V vs. RHE) as the subject lead to the cathodic and anodic Tafel equations (eqn (5) and (6), respectively): (6) where j c and j a represent the cathodic and anodic current densities (mA cm À2 ), respectively; j o,s is the exchange current density (mA cm À2 ), R is the universal gas constant, T is absolute temperature, F is Faraday constant, a a and a c are the anodic and cathodic electron transfer coefficients, z is the number of electrons involves in electrode reaction (z ¼ 1). The Tafel plots of the electrocatalysts are presented in Fig. 5c and d. The kinetic parameters of the electrocatalysts for HER in N 2 -deaerated 0.1 M KOH are summarized in Table 2.
The Pd/MOFDC electrocatalyst shows higher j o,s values at all the rotations speeds than the Pd/CB electrocatalyst and compared literature (see Table 3) because of ease of continuous H-O-H cleavages and more facile desorption of H ads from the active sites (M) of Pd and increased rates of H 2 production, with agitation during HER as the rotation speed increased, as the M-H ads bond is predominantly moderated by van der Waals forces. 21 The H ads present in this medium is continuously generated by electrocatalytically splitting water for H 2 production. Compared to most reports, the Pd/MOFDC also shows higher mass activity (j o,m ) of 58.63 mA mg À1 at 1600 rpm in alkaline condition, due to the ease of continuous water splitting, yielding MH ads , and subsequently desorption of the H ads from the active sites of the Pd catalyst to eventually produce H 2 . 22 Table 3 compares HER kinetics of the Pd/MOFDC with the Pd/CB and literature. The HER onset overpotential (onset h) values of 35 mV and 56.7 mV were obtained for the Pd/MOFDC and Pd/CB, with corresponding overpotential (h) values (at 2.5 mA cm À2 ) of 273.2 mV and 324.4 mV, respectively. The less h values of the Pd/MOFDC suggests lower energy is required for the evolution of hydrogen relative to other electrocatalysts, due to its ease of continuous water splitting for H 2 production. Tafel slope of the electrocatalysts for the HER is a parameter employed to explain that the catalysts continuously split water, even aer the initial water splitting (Volmer). The high Tafel slopes of the catalysts suggest that Heyrovsky reaction (eqn (3)) proceeds immediately aer the RDS (eqn (1)). 23 Thus, Volmer-Heyrovsky processes may predominate as the HER mechanism, as the process requires continuous H-O-H cleavage, adsorption of H (MH ads ) and desorption of the H ads . 24 This assumption agrees with the study of free energy diagram of HER in alkaline conditions, which proved that Volmer-Heyrovsky mechanism dominate. 25 However, high Tafel slopes (160-266 mV dec À1 ) obtained here than the theoretical values for Volmer and Heyrovsky processes, have previously been ascribed to high hydrogen adsorption onto the active sites of polycrystalline Pd electrocatalysts, as supplementary reaction step, 26,27 and are also reported in the compared literature (see Table 3).
Koutecky-Levich plots of the electrocatalysts are employed to separate kinetic current from the mass transfer as the rotation speed increased, shown in Fig. 5e and f at varied overpotentials 0.20, 0.25 and 0.30 V (vs. RHE). The kinetic currents (j K ) were extrapolated from the intercept (1/j K ), as it is potentialdependent. At the overpotentials studied, the Pd/MOFDC records j K values of 5.20, 7.36 and 8.13 mA cm À2 , while the Pd/CB records 0.85, 1.26 and 2.23 mA cm À2 , respectively. Then, heterogeneous rate constants (k 0 ) of the electrocatalysts were determined from the j K values and averaged, following the  Fig. 6a. To understand the effect of temperature on the HER, the polarization curves were also modelled to the conventional Tafel plots (eqn (5)), as shown in Fig. 6b and c. The kinetic parameters of the temperature-dependent HER in N 2 -deaerated 0.1 M KOH solution on the electrocatalysts, at 1600 rpm and 5 mV s À1 , are summarized in Table 4. The j o,s values for the HER increased with increasing temperature. At all Table 3 Comparison of electrochemical HER activity of the electrocatalysts with literature in the alkaline electrolytes at 1600 rpm and 5 mV s À1 in terms of onset overpotential (onset h), Tafel slope (b c ), exchange current density (j o,s ) and mass activity (j o,m )   Arrhenius equation was further employed to determine the minimum amount of energy required (activation energy (E A )) for the electrocatalysis of HER to occur. Fig. 6d indicates the Arrhenius plots (plot of the natural logarithm of exchange current (ln j o,s ) against the reciprocal of temperature (T À1 )) where the E A was extrapolated from the slope ÀE A R and calculated as 12.34 AE 1.82 kJ mol À1 and 22.10 AE 8.95 kJ mol À1 for the Pd/MOFDC and Pd/CB, respectively. The values of the E A for these Pd-based catalysts were less than those previously reported for Pt/C (z18 kJ mol À1 ). 37 The small value of the E A for the Pd/MOFDC indicates reduced energy is required for continuous water splitting, perhaps due to lowered hydride phase formation for rapid production of H 2 , 37 which is consistent with its onset overpotential and overpotential at 2.5 mA cm À2 . These observations are expected as Density Functional Theory (DFT) previously proved that the binding energy of hydrogen onto Pd-based catalyst is lowered with carbon supports that have increased electrons density. 38,39 Further investigation of the electrocatalysts for HER kinetics at room temperature, 1600 rpm and overpotential of 0.2 V vs. RHE was done using the EIS studies. The Nyquist plots (Fig. 7) with the modelled electrochemical equivalent circuit (inset) employed for tting, revealed the HER kinetic of the electrocatalysts (listed in Table 5) during this process. The most important parameter is the charge transfer resistance (R ct ), which is used to determine the j o,s , following eqn (7) below.
The R ct values of 603.20 U and 1160.00 U, resulted in calculated j o,s values of 0.2172 and 0.1129 mA cm À2 for the Pd/ MOFDC and Pd/CB, respectively. These values are approximately close to the kinetic values obtained using the Tafel plots.
3.2.2 Oxygen evolution reaction (OER) study. Out of curiosity, we also checked the ability of the Pd/MOFDC and Pd/CB to perform the oxygen evolution reaction (OER). As shown in Fig. 8a, the OER polarization curves for the electrocatalysts were conducted at 1600 rpm, 5 mV s À1 and room temperature. This polarization curve was modelled to Tafel plots (eqn (6)), as shown in Fig. 8b. The data show a satisfactory current densities (j) of 18.40 mA cm À2 and 14.68 mA cm À2 with onset h of ca. 1.29 V and 1.49 V vs. RHE, which are higher than thermoneutral potential of water (1.23 V vs. RHE) for easy water splitting, 40,41 and lowered than previously reported onset h value for the OER in the same medium, 42 while Tafel slopes (b a ) of 77.68 AE 12.81 mV dec À1 and 112.30 AE 4.96 mV dec À1 for the Pd/MOFDC and Pd/CB, respectively, as the large differences of these values, could be traced to the Co metal residue. The low Tafel slope of the Pd/MOFDC is associated with better OER activity than Pd/   CB. 41,43 These results show that the Pd/MOFDC electrocatalyst is not only good for HER but also excellent for OER, explaining its bi-functional behaviour for water electrolysis.

Conclusions
This work describes the rst study of the use of MOFDC as efficient support for Pd electrocatalyst for HER and OER. The study reveals that the HER and OER activities on the Pd/MOFDC electrocatalyst are greatly improved, with extremely higher current densities compared to Pd/CB and literature. The enhanced HER activity of the Pd/MOFDC is attributed to the electronic properties of the Pd being modied by the increased electron density of the MOFDC and Co metal residue, resulting in the ease of continuously splitting of water to yield MH ads and subsequently desorption of the H ads to produce H 2 , evidenced from the increased mass transfer (j o,s and j o,m ) and kinetic (j K and k 0 ) parameters and the energy required to achieve this is very low. This observation was corroborated with the less E A value of the Pd/MOFDC (12.34 AE 1.82 kJ mol À1 ) than the Pd/CB (22.10 AE 8.95 kJ mol À1 ). Additionally, the Pd/MOFDC displays satisfactory OER activity than its analogue, with lower onset h and b a , and higher current response. This new study provides an opportunity for further research to rationally design more MOFDCs using different ligands and transition metals for the efficient electrochemical water-splitting process.

Conflicts of interest
There are no conicts to declare.