Md Abu Sayeed and
Anthony P. O'Mullane*
School of Chemistry, Physics and Mechanical Engineering, Queensland University of Technology (QUT), GPO Box 2434, Brisbane, QLD 4001, Australia. E-mail: anthony.omullane@qut.edu.au
First published on 5th September 2017
Recently there has been a noticeable shift towards developing amorphous bimetallic or trimetallic oxides for electrochemical water splitting. However, the fabrication of a homogeneous mixed metal oxide electrocatalyst suitable for water electrolysis is not a facile process. Here we introduce an electrochemical synthesis method that is rapid, simple and performed under ambient conditions. Using this approach it is possible to create a catalytically active FeCoNiOxHy amorphous material whose activity is dependent on the nature of the underlying support. The trimetallic oxide is significantly more active than any single or bimetallic oxide combination for the OER. This amorphous catalyst demonstrates not only excellent activity but also stability over extended time periods.
Interestingly, there has been a shift from utilising crystalline catalysts to producing amorphous materials for water splitting.22,25–27 A recent study has shown that a photochemical metal–organic deposition process can produce amorphous mixed metal oxides displaying activity for the OER.22 Crystallinity or lack thereof is therefore expected to be a critical factor in determining the activity of the catalyst as evidence is now emerging that a reversible crystalline to amorphous transition can occur during the OER, as reported for Co3O4.28 The formation of a thin amorphous layer at the surface of the crystalline metal oxide was found to be the active state for the OER, which reverted to the crystalline state once returned to non-catalytic conditions. Therefore, the formation of an amorphous homogeneous mixed metal oxide system that is active and robust is of significant interest for the OER.
Here we introduce an electrochemical protocol that results in the formation of a FeCoNiOxHy material under ambient conditions that is evenly distributed over the substrate electrode and is highly active and durable for the OER. An electrochemical approach was taken due to simplicity, cost-effectiveness and good adherence to the underlying electrode compared to other deposition techniques that may be more involved for OER electrode preparation such as hydrothermal, co-sputtering and thermal decomposition techniques.
For the equimolar solution containing all three metal salts the voltammogram is close to the addition of the three individual processes as evidenced by the large magnitude of current from −0.80 to −1.20 V. This result was further verified by the mathematical summation of the response from the individual reduction processes (Fig. S1†). It should be noted that there is a slight shift to more positive potentials (ca. 0.15 V) for all processes compared to the individual responses which may be due to the increased conductivity of the solution when three salts are present. Also in the CV it can be seen that there is current crossover between the forward and reverse scans which is highly indicative of nucleation growth phenomenon.30 This was analysed by performing current time transients (Fig. S2†) over a range of −0.70 to −1.00 V and analysing the data by the Hills–Scharifker method31 where it was found that at early times (<3 s) instantaneous nucleation and growth occurs.
Initially it was confirmed that the electrodeposition of all three components did in fact enhance the OER compared to either individually deposited materials or the bimetallic combinations. A potential of −0.95 V was chosen which is well within the water reduction process (Fig. 1) for a period of 90 s. It can be seen from Fig. 2 that the trimetallic system gave the highest current density at an overpotential of 0.34 V when compared to mono or bimetallic systems (the 5th cycle is presented which ensured a stable response). In addition this material showed the lowest Tafel slope of 32 mV dec−1 when compared to all other combinations (Table S1†). Interestingly the FeCo and FeNi combinations reduced the overpotential but could not facilitate as high current densities. This data is consistent with previous work where the presence of iron, although very inactive itself in the potential range of interest, promotes the activity of Co and Ni oxides.13,32,33
![]() | ||
Fig. 2 Cyclic voltammograms recorded at 10 mV s−1 in 0.1 M NaOH for mono-, bi- and trimetallic oxides showing the 5th cycle of the OER. |
It can be seen prior to the onset of oxygen evolution that there are significant faradaic processes occurring at the as-deposited materials which are illustrated in Fig. 3. For the individual M(OH)2 deposits characteristic redox processes can be seen for the Co(II)/Co(III) and Ni(II)/Ni(III) transitions with the latter process occurring at more positive potentials. For the case of iron hydroxide/oxide there are no oxidation processes in the potential range studied which is consistent with previous reports.32 For the CoNi hydroxide/oxide system it can be seen that the CV shows behaviour in a potential region between that of the individual components which indicates good electronic communication between the materials. For FeCo and FeNi hydroxide/oxide the inclusion of iron has a significant impact on the voltammetry which is evidenced by a dramatic reduction in the peak intensities and shift to less positive potentials for both the Co(II)/Co(III) and Ni(II)/Ni(III) transitions. The inclusion of iron also results in a dramatic increase in the OER current towards the end of the sweep which is much greater than that seen for the CoNi case. This is consistent with the recent work by Strasser who demonstrated that the incorporation of Fe into nickel oxide can increase the OER rate by 1–2 orders of magnitude.13 Finally for the trimetallic FeCoNi sample there is a broad oxidation process from 1.2 to 1.5 V prior to the onset of the OER. Although the onset for the OER is slightly more positive compared to the bimetallic samples containing iron the current density that can be achieved with the trimetallic sample at slightly higher potentials is significantly greater (Fig. 1). Therefore given the enhanced performance in terms of current density that can be achieved the trimetallic system was investigated in detail. The optimum OER activity was then determined via applying different potentials and deposition times as shown in Fig. S3† and found to be −0.95 V for 90 s. These conditions also resulted in the best long term performance of the electrocatalyst. A SEM image of the optimised sample is shown in Fig. 4. The formation of a layered sheet-like morphology with minimal cracks was observed which is indicative of M(OH)2 materials such as Ni(OH)2 and Co(OH)2 (ref. 34 and 35) and also consistent with iron oxyhydroxide films.36 SEM-EDX analysis revealed a composition of Fe25Co40Ni35Ox (Fig. S4†) for the as-deposited material which is comparable with the X-ray photoelectron spectroscopy (XPS) data as Fe20Co37Ni43OxHy. This indicates that the final composition is different to the ratio of metal salts used during the electrodeposition process which were equimolar. Therefore this indicates that the electrodeposition of nickel and cobalt hydroxide is preferred over iron hydroxide.
![]() | ||
Fig. 3 Cyclic voltammograms recorded at 10 mV s−1 in 0.1 M NaOH for mono-, bi- and trimetallic oxides showing the 1st cycle prior to the OER. |
![]() | ||
Fig. 4 SEM images of FeCoNiOxHy electrodeposited onto an Au electrode (a), (b) before and (c), (d) after OER. |
However this material is unlikely to be the active species involved in the OER as reported previously for electrodeposited Co(OH)2.35 Once Co(OH)2 is oxidised the composition changes to Co3O4 prior to the OER and then formation of Co(IV) occurs which is responsible for oxygen evolution.37 Therefore the sample was also imaged (Fig. 4c and d) after several potential cycles into the OER region over the range shown in Fig. 2. An interesting effect occurred whereby the surface loses its layered type structure and is replaced with a fractured surface containing isolated islands comprised of nodule like nanomaterials. EDX mapping of the reconstructed surface shows a highly even distribution of Fe, Co, Ni and O throughout the material without any evidence of phase segregation (Fig. 5 and S5†).
![]() | ||
Fig. 5 SEM-EDX maps of FeCoNiOxHy electrodeposited onto an Au electrode (a) before and (b) after OER. |
This is generally difficult to achieve with more conventional approaches such as thermal decomposition and coprecipitation.22 There is also clear exposure of the supporting gold electrode (Fig. 5) which in principle should be beneficial for oxygen evolution given the significant impact it has on the activity of cobalt oxide catalysts.7,35,37,38 The thickness of the as-deposited sample is 75 μm which decreased slightly after the OER to 67 μm (Fig. S6†).
The composition also changed compared to the as-deposited material and was determined to be Fe30Co30Ni40OxHy by SEM-EDX (Fig. S5†) and Fe30Co35Ni35OxHy by XPS (Fig. 6) indicating a slight enrichment of Fe and Ni compared to the as-deposited material. Taking the Co 2p core level spectrum the binding energy of Co 2p3/2 (Fig. 6b) before the OER conversion process is 780.1 eV with a clear satellite peak at higher energy which indicates the presence of Co(OH)2.35 After the OER reaction the satellite peak is diminished, which indicates the formation of Co3O4 at the surface. For Ni, the Ni 2p3/2 peak at a binding energy of 855.1 eV (Fig. 6c) is indicative of Ni2+ species and is consistent with the formation of NiO and Ni(OH)2.39 For Fe the Fe 2p3/2 peak at a binding energy of 710.8 eV (Fig. 6d) is indicative of Fe2O3 which is consistent before and after the OER reaction. It has been reported previously29 that electrochemically reduced iron nitrate initially forms Fe(OH)2 but then converts readily to Fe2O3 which is consistent here with the as deposited material.40 The O 1s spectrum (Fig. 6e) before the OER can be deconvoluted into two components indicating a mixture of hydroxide (higher binding energy) and oxide states (lower binding energy). However, after the OER only one broad component is observable indicating the majority of the sample comprises of metal hydroxides41 or oxygen defects. Also shown in Fig. 6f shows the Au 4f core level spectra showing binding energies of 84.9 eV and 88.5 eV for Au 4f5/2 and 4f7/2, respectively which can be assigned to metallic Au0 of the underlying electrode.42 The binding energy values shift by 2 eV indicating the formation of an oxidised surface43 after the OER reaction. The composition of the catalyst before and after the OER was also analysed by laser ablation inductively coupled plasma mass spectrometry and found to change from Fe24Co44Ni32OxHy to Fe28Co38Ni34OxHy which is consistent with the EDX and XPS results.
To gain further insights into the structure of the materials, FeCoNiOxHy was electrodeposited onto an Au coated (2–3 nm thick) TEM grid (Fig. 7a and b). EDS analysis with HRTEM shows that FeCoNiOxHy was electrodeposited on the gold particles and the underlying carbon film (Fig. S7†). From the FFT images in the carbon only region (Fig. 7b), it was observed that the materials were mainly amorphous, however it must be noted that there are some spots in the FFT pattern which indicates some degree of crystallinity at the localised nanoscale level. Interestingly, after the OER the material showed evidence of a transition into a more crystalline state (Fig. 7d). However, XRD analysis of the bulk material before and after the OER was dominated by the underlying gold substrate (Fig. 8) and suggests that this amorphous to crystalline transition does not extend through the bulk of the electrodeposited material. It is interesting that the gold peaks are suppressed after the OER which may indicate that the gold is being oxidised quite significantly during the process (Au film for this sample is 100 nm thick) which is expected at this applied potential.44 The two peaks around 2θ = 55° which shift after the OER were unable to be assigned and did not correspond to any mono-, bi- or tri-metallic oxide/hydroxide. Given the absence of any other major peaks at lower values does suggest the material is amorphous.
![]() | ||
Fig. 7 HR-TEM images for (a, b) electrodeposited mixed hydroxide and (c, d) post OER electrodeposited mixed hydroxide. The inset shows the FFT images. |
![]() | ||
Fig. 8 GIXRD patterns of (a) Au substrate only, (b) FeCoNiOxHy@Au after deposition and (c) FeCoNiOxHy@Au after 5 cycles of OER. |
It was also found that the underlying electrode influenced the OER whereby a gold electrode was found to show significantly better performance over glassy carbon (GC), Pd or Cu support electrodes (Fig. 9). Although the first cycle for FeCoNiOxHy deposited on GC showed similar behaviour to that on Au the current density quickly diminished after 5 cycles (Fig. 9b) and demonstrated less activity than both Pd and Cu support electrodes. The Tafel slope for the OER at FeCoNiOxHy on the different substrates (Au, GC, Pd and Cu) after 5 cycles was determined to be 32, 84, 52 and 73 mV dec−1 respectively, implying a significant dependence on the substrate for the OER (Fig. 9c). It should be noted that the morphology of FeCoNiOxHy deposited on GC is distinctly different compared to Au (Fig. S8†). The surface is fractured and does not contain the layered type structure seen for the case of Au (Fig. 4). However from SEM-EDX analysis (Fig. S9†) the composition is comparable (Fe30Co34Ni36OxHy) on a GC electrode. After the OER parts of the film were compromised but in general the morphology remained intact indicating that a significant restructuring process did not occur. Previous work on cobalt oxide catalysts postulated that underlying metals act as electron sinks to promote the formation of the Co(IV) oxidation state which is regarded as the active site for oxygen evolution.7 This may also play a role here given the high percentage of cobalt in the trimetallic oxide sample.
Determining the main active site for the OER however is a challenging issue that is difficult to address. Recently however, Strasser13 has shown for the FeNi system that the buildup of higher oxidation states of Ni(IV) and Fe(IV) is followed by O–O bond formation with the subsequent release of molecular oxygen. This process restores the metal site back to its reduced state. Once the Fe level is above 4% the rate of oxygen evolution is greater than the rate of metal oxidation and therefore lower valent metal centres are stabilised during catalysis and the active centre was determined to be Ni2+Fe3+OOH. Recent spectroscopic characterization of mixed Fe–Ni oxide electrocatalysts indicated that a NiFe2O4 phase was a contributing factor to enhanced OER activity as well as the presence of basic active sites.45 Bell et al.46 have also reported that Ni–Fe catalysts are active for the OER whereby the incorporation of iron into the film increases the potential at which the Ni(OH)2/NiOOH redox process occurs and decreases the average oxidation state of Ni in NiOOH which results in an increase in activity for the OER.46 Cao et al. have also investigated the Fe–Ni system in detail47–49 and described additional effects that promote OER activity such as the embedded Fe(III) increases charge mobility due to more oxygen vacancies that facilitate polaron hopping between neighbouring atoms as well as the creation of surface defects for the coordination of reactive species. As seen from Fig. 2 and 3 the incorporation of Fe into cobalt also enhances activity at low overpotentials and therefore an analogous process may be taking place. However the nature of the underlying electrode was not investigated and may have an influence on the processes taking place. The combination of electronic interaction between the catalyst layer and the support, the critical presence of Fe, and the surface morphology changes that occur, greatly influence OER activity.
The turnover frequency (TOF) is an excellent way to benchmark these materials as getting accurate surface areas for electrodeposited materials on this scale is not straightforward. As discussed by Lyons et al. the electrochemical equivalent of the TOF can be described by:50
TOF = 1NA/4FNatoms = J/4Q | (1) |
Footnote |
† Electronic supplementary information (ESI) available: XPS data for as-deposited films and films after OER, XRD data, chronoamperometry data recorded at Au and GC electrodes and SEM images showing the effect of deposition time and potential. See DOI: 10.1039/c7ra07995h |
This journal is © The Royal Society of Chemistry 2017 |