Xin Liuab,
Jinmei Dongab,
Bo Youa and
Yujie Sun*a
aDepartment of Chemistry and Biochemistry, Utah State University, Logan, Utah 84322, USA. E-mail: yujie.sun@usu.edu; Fax: +1-435-797-3390; Tel: +1-435-797-7608
bKey Laboratory of Comprehensive and Highly Efficient Utilization of Salt Lake Resources, Chinese Academy of Sciences, 18 Xinning Road, Xi'ning 810008, China
First published on 26th July 2016
The design of nonprecious, bifunctional, and highly competent electrocatalysts for both H2 and O2 evolution reactions (HER and OER) has attracted increasing interest recently. Herein, we report a cobalt-based electrocatalyst derived from ZIF-67 grown on carbon cloth (Co–P/NC/CC) for overall water splitting electrocatalysis. The as-prepared Co–P/NC/CC catalyst exhibited remarkable catalytic performance in 1 M KOH with Tafel slopes of 52 and 61 mV dec−1 for HER and OER, respectively. When serving as catalysts for both the cathode and anode, our Co–P/NC/CC demonstrated high efficiency and strong robustness. A thorough comparison with other control samples and detailed characterization results revealed that the superior activity and excellent stability of Co–N–C likely originated from the highly porous, self-supported, and binder-free nature of the electrocatalyst, as well as the high conductivity of carbon cloth. Hence, direct decoration of metal organic frameworks on conductive substrates represents an effective approach for the development of electrocatalysts not only promising for water splitting but also for many other applications.
Taking advantage of the large specific surface area and high porosity of metal organic frameworks (MOF), our group recently reported several MOF-derived electrocatalysts for various reactions, such as water splitting and oxygen reduction reaction.27,28 We demonstrated that a cobalt imidazolate-based MOF, ZIF-67, was able to act as a bifunctional electrocatalyst for overall water splitting after pyrolysis and phosphidation treatments.27 In order to load the ZIF-67-derived catalyst on the working electrode, a conductive polymer, Nafion, was utilized to immobilize the catalyst on the electrode surface. Such a viscous binder of Nafion will inevitably result in buried active sites, decreased specific surface area, and increased interfacial resistance, which would result in inferior performance of the catalyst and complicate the electrode preparation step.29–31 Therefore, we seek to develop MOF-derived electrocatalysts directly grown on conductive substrates with intimate physical and electric contact.32 To the best of our knowledge, direct growing of MOF-derived electrocatalysts on conductive substrates for overall water splitting has received scarce attention.
Herein, we report a facile and effective strategy to anchor ZIF-67 on the surface of carbon cloth (CC, Fig. S1a†) to yield ZIF-67/CC. Carbon cloth is a type of cheap and commercially available textile with high conductivity, robust corrosion resistance, excellent flexibility, and strong mechanical strength, representing an ideal 3D substrate to integrate catalysts for electrocatalytic applications.33–36 Pyrolysis of ZIF-67/CC followed by phosphidation resulted in cobalt phosphide nanoparticles imbedded in nitrogen-containing carbon matrices anchored on carbon cloth tightly (Co–P/NC/CC). Electrochemical studies of Co–P/NC/CC demonstrate its excellent activities for electrocatalytic HER, OER, and overall water splitting. A myriad of materials characterization techniques were also employed to study the composition and morphology of Co–P/NC/CC prior to and post electrocatalytic reactions. The excellent electrocatalytic performance of Co–P/NC/CC compared to control samples, such as ZIF-67-derived catalyst particles drop casted on carbon cloth, is primarily attributed to the intimate contact between Co–P/NC catalyst and carbon cloth, as well as the large surface area and high conductivity of the latter.
Subsequently, ZIF-67/CC was carburized under argon at 900 °C for 3 h. During this process, ZIF-67 nano-polyhedrons transformed to metallic cobalt nanoparticles imbedded in nitrogen-containing carbon matrices, which was denoted as Co/NC/CC. The SEM image of Co/NC/CC (Fig. S1b†) indicated that the original regular and smooth nano-polyhedrons of ZIF-67 shrank to rough porous composites. Nevertheless, the discrete nanopolyhedrons could still be identified. Next, Co/NC/CC was phosphatized under argon at 300 °C for 2 h, utilizing sodium hypophosphite as the phosphorous source and resulting in cobalt phosphide nanoparticles decorated within nitrogen-containing carbon matrices. This final product was named as Co–P/NC/CC. As shown in Fig. 1d, the SEM image of Co–P/NC/CC proved that the integrity of each shrunk nano-polyhedron was preserved, although a close inspection (Fig. 1e) implied some agglomeration. It should be noted that these new nano-polyhedrons were still tightly anchored on the surface of carbon cloth. The large enrichment of phosphorous in these nano-polyhedrons was well manifested in the elemental mapping images shown in Fig. 1f, along with the homogeneous distribution of Co, P, N, and C.
The XRD pattern of ZIF-67/CC (Fig. S2†) confirmed the crystallinity and identity of ZIF-67 grown on carbon cloth.37,38 After pyrolysis, the weak and broad XRD peaks of Co/NC/CC (Fig. S3†) at 44.2° and 51.5° suggested the formation of metallic cobalt; while the relatively intense peak around 25–30° was due to the substrate of carbon cloth. After phosphidation, despite the lack of well-resolved XRD peaks for Co–P/NC/CC, a comparison with standard XRD patterns of CoP and Co2P suggested the formation of a mixture of CoP and Co2P in Co–P/NC/CC (Fig. S4†). Elemental analysis data listed in Table S1† presented a Co/P atomic ratio is close to 4.5 (much larger than 2), implying that only a layer of CoP and Co2P mixture was formed on the surface while the inner composite was still metallic cobalt in Co–P/NC/CC.
In order to further shed light on the composition and valence states of the elements in each catalyst sample, X-ray photoelectron spectroscopy (XPS) was collected. Fig. S5a† presents the survey spectrum of Co–P/NC/CC which showed all the anticipated elements of Co, P, C, and N, consistent with the elemental mapping results (Fig. 1f). The high-resolution Co 2p XPS spectrum of Co–P/NC/CC is plotted in Fig. 2a. Deconvolution of this Co 2p spectrum resulted in primary Co 2p3/2 peaks at 778.5 and 782.1 eV, corresponding to cobalt phosphides and cobalt oxides, respectively. The presence of cobalt oxides was probably due to oxidation in air prior to the XPS measurement.33,39 The weak feature around 793.6 eV could be assigned to the Co 2p1/2 binding energy. The deconvolution of the high-resolution P 2p XPS spectrum led to a prominent peak at 134.0 eV, attributable to oxidized phosphorous species (e.g., phosphate). In the meantime, typical phosphide signal could still be observed at 129.7 and 131.1 eV, which were attributed to the P 2p3/2 and 2p1/2 features of phosphorous in cobalt phosphide. Overall, the binding energies at 778.5 and 129.7 eV were quite close to those of CoP.40 In addition, the high-resolution N 2p XPS spectrum was also collected and deconvoluted in Fig. S5b,† displaying binding energies at approximately 398.9, 399.4, and 401.7 eV, which could be attributed to pyridinic N, Co–N, and graphitic N, respectively.
With the detailed characterization data in hand, we next assessed the electrocatalytic performance of these catalysts for H2 evolution. All the electrochemical experiments were conducted in 1.0 M KOH using a three-electrode configuration unless otherwise noted. The linear sweep voltammetry (LSV) curves of Co–P/NC/CC and control samples (carbon cloth, ZIF-67/CC, and Co/NC/CC) were plotted in Fig. 3a. It is apparent that Co–P/NC/CC catalyzed H2 evolution much earlier than all the control samples with a catalytic onset potential around −0.1 V vs. reversible hydrogen electrode (RHE). Without phosphidation, Co/NC/CC initiated HER catalytic current approximately 50 mV more negative than that of Co–P/NC/CC. Most likely due to the poor conductivity and lack of active sites in ZIF-67, ZIF-67/CC only acted slightly better than the background carbon cloth for HER. Specifically, Co–P/NC/CC only required overpotentials of −171 and −243 mV to reach catalytic current densities of 10 and 100 mA cm−2, respectively. These overpotential requirements are much smaller than those of Co/NC/CC and ZIF-67/CC to achieve the same current densities and also compare quite favourably with those of recently reported HER electrocatalysts in strongly alkaline electrolytes (see detailed comparison in Table S2†).41–44 Even though Pt/C exhibited a smaller onset potential, the current density of our Co–P/NC/CC surpassed that of Pt/C beyond −220 mV vs. RHE (Fig. S8a†). Fig. 3b presented the derived Tafel plots of each catalyst and apparently Co–P/NC/CC exhibited a much smaller Tafel slope (52 mV dec−1) compared to those of carbon cloth (81 mV dec−1), ZIF-67/CC (75 mV dec−1), and Co/NC/CC (100 mV dec−1), implying a faster HER kinetic rate of the former. In addition, electric impedance spectroscopy was collected to further probe the superior HER activity of Co–P/NC/CC. As shown in Fig. 3c, the Nyquist plot of Co–P/NC/CC showed a smaller semi-circular diameter than those of the control samples and thus demonstrated a lower resistance for electrocatalytic HER. In order to determine whether the high HER performance of Co–P/NC/CC is due to its increased electrochemically active surface area (ECSA), we next measured its double-layer capacitance via cyclic voltammetry in the non-faradaic region (0.123 to 0.223 V vs. RHE, Fig. S6†). It has been widely accepted that cyclic voltammetry at non-faradaic potentials is able to estimate ECSA, especially when the electrodes consist of similar composites. By plotting the difference in current density between the anodic and cathodic scans at 0.173 V vs. RHE, nearly linear plots could be obtained in Fig. 3d. The extracted slopes of the linear fitting resulted in similar ECSA for both Co–P/NC/CC and Co/NC/CC, which were substantially larger than those of ZIF-67/CC and carbon cloth. Therefore, it is clear that although large ECSA is beneficial towards HER, the higher HER activity of Co–P/NC/CC relative to Co/NC/CC is most likely owning to the phosphidation step which yielded more HER active cobalt phosphides on the cobalt nanoparticles.
The OER performances of these catalysts were also evaluated in the same electrolyte, 1.0 M KOH. Apparently, Co–P/NC/CC exhibited a much smaller onset potential of ∼1.5 V vs. RHE and greater catalytic current than those of Co/NC/CC, ZIF-67/CC, and CC (Fig. 4a). Noticeably, Co/NC/CC required an overpotential of 360 mV to reach a current density of 10 mA cm−2; while after phosphidation, Co–P/NC/CC only required an overpotential of 330 mV to arrive at the same current density. ZIF-67/CC and carbon cloth were not very active towards OER and they required potential more positive than 1.6 V to produce appreciable catalytic current. Even though these catalysts exhibited substantially different catalytic onsets, their derived Tafel slopes fell in a close range of 50–70 mV dec−1. Electric impedance spectra of these catalysts are shown in Fig. 4c, wherein Co–P/NC/CC presented the smallest resistance, in agreement with its best OER activity. Despite the similarity in the electrochemical surface areas of Co–P/NC/CC and Co/NC/CC for HER application, a rather obvious enhancement in ECSA was observed for Co–P/NC/CC relative to Co/NC/CC for O2 evolution (Fig. 4d and S7†). It implied that during electrocatalytic OER, the surface oxidation of Co–P nanoparticles would likely result in more porous and hence a larger number of electrochemically active sites than what occurred in the case of Co/NC/CC. This observation is consistent with recent reports showing that metal phosphide-based electrocatalysts actually demonstrated better OER performance the corresponding metal oxides. The estimated ECSA values of Co–P/NC/CC, Co/NC/CC, ZIF-67/CC, and carbon cloth follow the same trend of their OER activity, emphasizing the critical role that ECSA plays in electrocatalysis. A detailed comparison of Co–P/NC/CC with other reported electrocatalysts for OER under strongly alkaline conditions can be found in Table S3,† suggesting the excellent performance of Co–P/NC/CC for OER applications.45–49 Noticeably, our Co–P/NC/CC could also rival the state-of-the-art OER catalyst IrO2 (Fig. S8b†), showing similar onset potential but much faster catalytic current rise.
In order to demonstrate the fact that directly growing ZIF-67 on carbon cloth is crucial towards the excellent electrocatalytic performance of Co–P/NC/CC for both HER and OER, we conducted the following control experiments. ZIF-67 nano-polyhedrons were carbonized and phosphatized in a similar fashion as that of ZIF-67/CC. Subsequently, the resulting cobalt phosphide(s)-imbedded and nitrogen containing carbon matrices were loaded on a carbon cloth with Nafion as the polymer binder to produce a control sample Co–P/NC–CC. As shown in Fig. S9,† the HER and OER performances of Co–P/NC–CC are inferior to that of Co–P/NC/CC with same mass loading of cobalt. The catalytic current of Co–P/NC–CC did not take off until scanning to −0.3 V vs. RHE for HER and 1.6 V vs. RHE for OER, both of which required a larger overpotential compared to Co–P/NC/CC. In addition, the stability of our Co–P/NC/CC relative to CC for OER was also inspected. As shown in Fig. S10,† although CC was not stable for long-term OER electrolysis, our Co–P/NC/CC manifested its perfect stability in 1.0 M KOH for 50 h at 10 mA cm−2, showing unchanged chronopotentiometric curve over the entire course of OER electrolysis. These results unambiguously prove the effectiveness of anchoring ZIF-67 on carbon cloth during the preparation stage in producing more active and robust electrocatalysts.
With the excellent HER and OER performance of Co–P/NC/CC discussed above, we subsequently sought to fabricate a two-electrode electrolyzer for overall water splitting with same type catalyst-loaded carbon cloth electrodes serving as both cathode and anode in 1.0 M KOH. The LSV curves of these two-electrode systems were compared in Fig. 5a. The Co–P/NC/CC catalyst couple exhibited a catalytic onset potential less than 1.7 V. In fact, this catalyst couple approached current densities of 10 and 100 mV cm−2 at cell voltages of 1.77 and 1.93 V, respectively. In sharp contrast, the control samples of Co/NC/CC, Co–P/NC–CC, and ZIF-67/CC required a cell voltage of 1.89, 1.95, and 2.03 V, respectively, to arrive at 10 mA cm−2. As expected, carbon cloth was least active for water splitting. 20 h chronopotentiometry experiments were conducted to assess the stability of these catalyst couples for extended function at a catalytic current of 10 mA cm−2 (Fig. 5b). The Co–P/NC/CC catalyst couple required a voltage of ∼1.85 V and maintained this voltage throughout the entire course of electrolysis. All the other catalyst samples needed voltages at least 100 mV larger than that of Co–P/NC/CC. After the long-term chronoamperometry experiment, the composition and morphology of Co–P/NC/CC as cathode (Fig. S11†) and anode (Fig. S12†) investigated via SEM and elemental mapping analysis. Overall, the anchored cobalt phosphide(s)-based polyhedrons were still observed on the surface of carbon cloth with the anticipated distribution of Co, P, N, and C. A significant amount of oxygen was observed when Co–P/NC/CC acted as the anode, in agreement with the partial oxidation of cobalt phosphide(s)/cobalt during electrocatalytic OER.
Footnote |
† Electronic supplementary information (ESI) available: Additional electrochemistry and characterization data, Fig. S1–S12 and Tables S1–S3. See DOI: 10.1039/c6ra17030g |
This journal is © The Royal Society of Chemistry 2016 |