Ummul K. Sultana,
Tianwei He,
Aijun Du 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 4th December 2017
The fabrication of electrocatalysts that are active for more than one of the water splitting reactions has gained significant momentum. Here we demonstrate such a material produced via an electrochemical process that is based on amorphous cobalt sulfide films doped with oxygen which are active for the hydrogen evolution reaction (HER) and oxygen evolution reaction (OER) under alkaline conditions. The optimum electrochemical protocol was found to be a repetitive potential cycling approach rather than a constant potential to create an amorphous CoSx film containing oxygen. Samples with a Co:
S ratio of 1.56
:
1 were found to be active for the HER in 0.5 M H2SO4, phosphate buffer and 0.1 M NaOH. Significantly this activity is comparable to highly crystalline nanomaterials of cobalt sulfide. Density functional theory calculations indicated that a reduced S–Co coordination number, as encountered in amorphous materials, leads to an optimum binding energy for hydrogen adsorption on the material which facilitates good electron transfer kinetics. In addition, this material was also active for the OER in alkaline conditions with evidence of conversion to cobalt oxide which gave a low overpotential of 370 mV for an applied current density of 10 mA cm−2 with a Tafel slope of 67 mV dec−1. This simple approach shows promise for the fabrication of a dual action electrocatalyst for electrochemical water splitting under alkaline conditions.
In particular, cobalt based compounds such as sulfides,55,56 nanocomposites of CoS2 using reduced graphene oxide (RGO) and carbon nanotubes (CNT),37 amorphous CoSx composed of oxide and sulphide clusters,57 phosphides,49,58–60 phosphosulphides,61 and selenide complexes like CoS2xSe2(1−x) (ref. 62) have been identified as robust electrocatalysts for the HER. For synthesizing these HER catalysts most researchers have adopted wet chemical methods, vapour deposition,63 annealing64 or hydrothermal synthesis followed by temperature programmed reduction (TPR).65 However, electrochemically deposited nanomaterials have been shown to also perform as effective HER catalysts. The advantage of this approach is that it easy to perform, scalable, does not require expensive equipment and can be undertaken under ambient conditions.66 It also has a number of freedom parameters to tune the composition, morphology and density of active sites which is another major advantage.67–73 Recent theoretical work by Kornienko et al. suggested that amorphous CoSx where oxygen is incorporated into the material is also a candidate for the HER.57 Shanmugam et al. have reported that electrodeposited nickel iron sulphides on nickel foam74 and hydrothermally grown NiCo2S4 nanowires on nickel foam75 show bifunctional activity for the HER and the OER.
In addition there has been significant interest in developing materials that are not only active for the HER but also the oxygen evolution reaction (OER). In this area the majority of cobalt based materials reported are the oxides, hydroxides or oxy-hydroxides rather than sulphide based moieties. Cai et al. have demonstrated that oxygen incorporated into cobalt sulphide nanocubes are active for the OER, although the method of synthesis was quite involved using several chemical steps and high temperature annealing.76 However this material was not tested for the HER and to date no cobalt sulfide based materials have been reported that are active for both HER and OER. In this work we address this and use a simple and rapid electrochemical approach to deposit amorphous cobalt sulphide materials that contain oxygen, which are active for both HER and OER under alkaline conditions.
ΔG0H* = ΔEH + ΔEZPE − TΔSH | (1) |
ΔEH = EH* − E* − 1/2EH2 | (2) |
![]() | (3) |
is the entropy of H2 gas at the standard condition.84 The ΔSH can be obtained by:
![]() | (4) |
The calculated vibrational frequency for H2 gas is 4390 cm−1, the vibrational frequency of H adsorbed on CoS2 are 2523 cm−1, 662 cm−1, and 381 cm−1. Therefore the overall corrections are taken as:
ΔG0H* = ΔEH + 0.25 eV | (5) |
2H2O + 2e− → H2 + 2OH− | (6) |
2Co2+(aq) + 2e− → Co(OH)2(ads) | (7) |
Co(OH)2(ads) + 2e− → Co(s) + 2OH− | (8) |
![]() | ||
Fig. 1 Effect of TU on the cyclic voltammetric response recorded at a GC electrode in solutions containing 0.005 M Co(II) only and 0.005 M Co(II) + 0.005 M thiourea at different pH values. |
A small anodic process is observed on the reverse sweep which can be attributed to the reverse of eqn (7). The position of this peak varies slightly with the pH of the electrolyte. In the presence of thiourea (TU) a significant change occurs in the cyclic voltammetric response. The magnitude of the cathodic process decreases at all pH values and is shifted to more negative potentials. This can be attributed to the formation of CoSx via the following processes:88
CS(NH2)2 + 2OH− → S2− + OC(NH2)2 + H2O | (9) |
Co2+ + S2− → CoSx(ads) | (10) |
The electrochemically generated hydroxide ions from the reduction of water have been reported to react with thiourea to liberate sulphide ions which complex with Co2+ ions in solution to form CoSx which precipitates onto the electrode surface. However a recent study has concluded that in fact metal sulphides are synthesised via the formation of a (NH2)2CSs–M2+–OHh− complex which decomposes into the relevant metal sulphide via the following process:89
M2+ + CS(NH2)2 + 2OH− ⇄ M(OH)2CS(NH2)2 → MS + H2NCN + 2H2O | (11) |
On the reverse sweep a prominent oxidation peak can be seen in the presence of TU and is attributed to the oxidation of thiourea to formamide disulphide via:88
![]() | (12) |
After this process a smaller peak is observed towards the end of the sweep at all pH values and is attributed to the oxidation of CoSx on the electrode surface90 (this process is absent when TU is not present in the solution). Overall the behaviour over a pH range from 3–7 is quite similar showing distinct processes for TU oxidation to FD and the oxidation of CoSx at higher potential values. At pH 2 both of these processes are suppressed and indicates that thiourea is less susceptible to oxidation and also that the formation of CoSx is more inhibited.
It is well established that the potential waveform used to electrodeposit materials has a significant impact on morphology and electrocatalytic activity. Therefore chronoamperometry and repetitive potential cycling were chosen as methods to electrodeposit CoSx films. Fig. S1† shows current–time transients for the electrodeposition of CoSx at different pH values and Fig. S2† shows the subsequent anodic stripping voltammograms for the deposit on the electrode surface. This is consistent with the data shown in Fig. 1 where a significant oxidation process for CoSx oxidation occurred from 0.70 V over a pH range of 4–6. If the TU concentration was increased up to 0.5 M then the amount of CoSx formation increased, as evidence by the increased oxidation process at ca. 0.80 V at all pH values (Fig. S3†). The presence of CoSx was confirmed by EDS and XPS analysis as discussed later. It was also found that at pH values of 6–7 a precipitate of Co(OH)2 formed slowly in solution and therefore was not investigated further in electrodeposition reactions. Repetitive potential cycling has been used in many studies to create highly textured surfaces which have increased surface area and electrocatalytic activity for a variety of reactions. This was shown to be good method to synthesise nickel iron sulphides which were active for both HER and OER.74 Therefore a similar approach was undertaken here and shown in Fig. 2a are repetitive cyclic voltammograms recorded at an ITO electrode in the presence of 5 mM Co2+ and 5 mM TU where the pH was not adjusted and measured to be 5.6. The anodic potential limit was decreased to avoid any CoSx oxidation and allow the build-up of the deposit on the electrode surface. The cathodic potential limit was also curtailed to avoid the formation of any cobalt metal on the electrode surface. On the first cycle (initiated at positive potential) the same behaviour as seen in Fig. 1 (pH 5 or 6) is observed whereby a cathodic process is observed at ca. −0.40 V (C2) due to the formation of CoSx. On the anodic scan a large peak A1 is observed due to the oxidation of TU to FD. Upon cycling and taking the 3rd cycle as an example a new peak C1 appears which is due to the reduction of FD back to TU (note its absence of the 1st cycle as TU was not oxidised). The magnitude of process C2 also increases in the 3rd cycle illustrating the growth of the CoSx film. This growth in process C2 can also be observed in the 5th cycle. The effect of TU concentration is shown in Fig. 2b where the 15th cycle is shown. It can be seen that the magnitude of the C2 process increases with TU concentration indicating the formation of a greater amount of CoSx on the electrode surface.
The morphology, composition and their effect on electrocatalytic performance for the HER was investigated for samples electrodeposited on a GC electrode using chronoamperometry (10 min deposition time) and repetitive cycling (15 cycles at 15 mV s−1, Fig. S4†) at pH values of 3 and 5.6 at a constant Co2+ concentration of 5 mM and TU ranging from 5 mM to 1 M. With the CA technique the surface morphologies are quite different with the change of solution pH. At pH 3 for all TU concentrations the structures are more globular like (Fig. 3a1–d1) whereas they have a distinct layered surface when the electroplating solution pH is 5.6 (Fig. 3a2–d2) which is highly indicative of the formation of cobalt sulfide.91 The layered structure is observed for all concentrations of TU that were used, however the density of the film increases with TU concentration.
When a repetitive cycling protocol was used it can be seen that the formation of layered materials is favoured under nearly all conditions (Fig. 4). The only exception is when a low TU concentration of 5 mM is used at pH 3 where more globule like structures are seen like in the case of chronoamperometric deposition (Fig. 4a1). When the concentration of TU was increased to 0.05 M (Fig. 4b1) small layered deposits can be seen surrounding the main globule like deposits. At the higher TU concentrations the whole surface is dominated by layered structures. At pH 5.6 there is no evidence of globule like deposits and the thickness of the individual flake like deposits increases substantially with TU concentration until a thick and densely packed film is formed when 1 M TU was used (Fig. 4d2). Interestingly, it was found that all the samples synthesized through either chronoamperometry or repetitive cycling were XRD amorphous. A representative XRD pattern for the sample shown in Fig. 4c2 is presented in Fig. S5† where no diffraction peaks were observed. The presence of Co, S and O was confirmed by EDX analysis.
The freshly electrodeposited CoSx films were initially investigated as HER catalysts in acidic solution to identify the best material. Fig. 5a and b shows linear polarization curves recorded at the samples discussed in Fig. 3 and 4. It is immediately clear that the samples produced via constant potential deposition are not particularly active for the HER. From Fig. 5, it is evident that the samples prepared by the cyclic voltammetric protocol were much more active than those prepared by chronoamperometry and can be attributed to the significantly different deposit created on the electrode surface. The formation of this layered type material is highly critical to good HER performance. The best sample was prepared using 0.5 M TU at a pH of 5.6 (Fig. 5d). The onset potential is −0.22 V which is comparable to previous work reported for hollow cobalt sulphide nanomaterials.92 However a lower Tafel slope of 72 mV dec−1 was determined for our catalyst compared to 97 mV dec−1 in the aforementioned work. This suggests a Volmer–Tafel mechanism where the Volmer step is the rate limiting step.93 However, Tafel slopes in this case are only used as a guide to indicate the possible rate determining step. Care must be taken in interpreting such values as calculations are based on strict assumptions that do not always hold, hence the wide variation in Tafel slopes that are quoted for metal sulphide materials. The closest Tafel slope reported for cobalt sulphide materials that matches our work is by Sun et al. who reported a value of 72 mV dec−1 for highly crystalline cubic cattierite CoS2 produced via a hydrothermal method.94 It is therefore interesting to note that an amorphous material can demonstrate comparable behaviour to well defined crystalline materials for the HER. We also investigated this sample for the HER in alkaline solution (Fig. 5e) where it was found that it also maintained activity under these conditions. A Tafel slope of 142 mV dec−1 was determined which is also consistent with crystalline materials.94 In addition this material also performed well at neutral pH conditions and showed long term stability over a 24 h period (Fig. S6†) which is a significant improvement over previous work where amorphous CoSx films were unstable in neutral conditions and required annealing and electrochemical polarisation to ensure stability.56
To gain further insights as to why the samples produced by repetitive cycling are more active than those formed under constant potential conditions an XPS study was conducted. Illustrated in Fig. 6a are the Co 2p spectra for samples prepared with different amounts of TU in solution using the repetitive cycling method. Significantly, the Co 2p peaks in Fig. 6a indicate that the films produced with this method are not a pure CoSx film but also contain other oxygenated species like CoO or Co(OH)2 which appear over the range of 779 to 783 eV.95–101 The peak at around 786 eV can be characterized as the shake-up satellite peak of CoO or Co(OH)2.102 At TU concentrations of 0.5 and 1.0 M another new peak can be observed at 778 eV which has been attributed to the formation of Co–S bonds.76,103 Fig. 5b represents the S 2p core level spectra where the peaks from 160 to 165 eV can be attributed to the surface absorption of oxygen with sulfur and therefore doping of the CoSx film with oxygen.76 It can be seen from the relative intensities of the S 2p peaks that the intensity of the signal for the Co–S bands increases compared with the surfaced absorbed oxygen with sulfur peak as the TU concentration increases. This is consistent with the Co 2p spectra, which shows a clear increase in the Co–S peak at 778 eV. Though the high resolution spectrum of O 1s bands are often complex and difficult to interpret,104 in this study the peaks were fitted as shown in Fig. 6c. The peak at around 531 eV (from the as deposited film) in the spectrum indicates the presence of hydroxides and oxides of Co,104 with a slight shift to lower binding energy upon the incorporation of more sulfur in the films.
The composition of all samples was then determined by XPS and is shown in Table 1. A clear trend can be seen in that the concentration of sulfur increases when more TU is added to the electrolyte for both constant potential and repetitive cycling deposition processes, whereby more sulfur is incorporated for the latter approach and is consistent with the electrochemical data. The optimum HER performance was observed for the sample deposited from 5 mM Co2+ with 0.5 M TU using repetitive potential cycling (Fig. 5d). This sample has a Co:
S ratio of 1.56, increasing or decreasing this ratio did not improve performance. Interestingly, when a similar ratio was produced via the chronoamperometric approach the HER performance was quite poor and may be related to the morphology of the film. In Fig. 3c2 it is apparent that this film is very dense compared to the more open and porous structure created using the repetitive cycling protocol (Fig. 4c2) thereby facilitating access to the active sites of the catalyst.
Process | Bath composition | Atomic % | Ratios | |
---|---|---|---|---|
Co | S | Co/S | ||
CA | Co 0.005 M + TU 0.005 M | 100 | 0 | — |
Co 0.005 M + TU 0.05 M | 99.68 | 0.32 | 311.5 | |
Co 0.5 M + TU 0.5 M | 73.84 | 26.16 | 2.82 | |
Co 1.0 M + TU 1.0 M | 61.42 | 38.58 | 1.59 | |
CV | Co 0.005 M + TU 0.005 M | 98.72 | 1.28 | 77.12 |
Co 0.005 M + TU 0.05 M | 83.04 | 16.96 | 4.89 | |
Co 0.5 M + TU 0.5 M | 60.94 | 39.06 | 1.56 | |
Co 1.0 M + TU 1.0 M | 55.51 | 44.49 | 1.25 |
In order to further understand the experimentally-observed highly efficient HER performance of CoSx compounds, we calculated the hydrogen binding free energy on the CoSx catalyst as shown in Fig. 7 based on density function theory (DFT). The overall HER can be described as three steps, the initial state H+ + e−, the intermediate adsorbed H* and the final product 1/2H2.105 Hydrogen should not bind catalyst too strong and too weak and the ideal value for ΔGH is close to zero. Clearly the value of ΔGH for CoSx was quite negative (Fig. 7), which represent a very strong interaction between adsorbed H and CoSx. Therefore, it is expected that pristine CoSx exhibits poor HER reaction kinetics. However, when one S atom was removed from the Co atom, i.e. the S–Co coordination number is reduced, the hydrogen binding free energy (ΔGH) on nearly every S atom can be reduced to 0.03 eV, which is comparable to that of the state-of-the-art Pt106 (ΔGH = −0.03 eV) and MoS2 (ref. 63) (ΔGH = 0.08 eV) catalyst. Our calculations explain the experimentally-observed HER activity when the number of coordinated S-atoms around Co atom is reduced in an amorphous CoSx compound.
![]() | ||
Fig. 7 The calculated free-energy diagram of HER under standard conditions for CoSx cluster catalysts, and the referenced Pt and MoS2. |
The ability of an electrocatalyst to be functional for both the HER and the OER is highly advantageous and therefore we also tested our material for the OER under alkaline conditions. Fig. 8a shows the OER at each sample produced by cyclic voltammetry at pH 5.6 using different TU concentrations. It illustrates that the samples are quite comparable in the potential (1.60 V) that is required to generate a current density of 10 mA cm−2. For the sample prepared with 0.5 M TU that was active for HER in alkaline solution (Fig. 5d) a Tafel slope of 67 mV dec−1 was determined (Fig. 8b). This is consistent with previous work on oxygen containing CoSx nanocubes which gave the same Tafel slope in 1.0 M KOH and achieved the same current density at 1.52 V. The enhanced activity for the OER was predicted by DFT calculations to be due to the presence of oxygen and dangling Co–S bonds.76 The main advantage with the approach undertaken here compared to this previous study is that the synthesis can be done rapidly and in one step under ambient conditions where good adherence to the underlying support is achieved. In addition electrochemical deposition lends itself to scalability and the ability to be deposited with good adherence on large high surface area electrodes suitable for commercial electrolysis applications. These results also demonstrate that the samples containing higher sulfur content (using 0.5 M TU) perform equally well to the samples with little sulfur content (0.005 M TU) which are essentially cobalt oxide/hydroxide materials which have been documented previously as being active for the OER.107–112 Prior to the OER several redox peaks can be seen which are attributed to the oxidation of cobalt into +3 and +4 oxidation states where the latter has been postulated to be the active oxidation state of Co for the OER.108 The stability of these catalysts containing low and high S contents was then tested over a 24 h period where it can be seen that the sample with a higher sulfur content resulted in a very gradual loss in performance for 18 h after which it stabilised. This is contrary to the sample with a lower sulfur content which maintained good stability over 24 h.
SEM analysis of the sample prepared with 0.5 M TU after the OER is shown in Fig. S7† which shows two distinct regions which are different to the original starting material (Fig. 4c2). The flake like deposits have been replaced with more nodule like material as well as patches of interconnected plate like materials which are highly indicative of Co3O4.113 This is supported by XPS analysis (Fig. S8†) which shows the emergence of a peak at 529 eV for the O 1s core level spectrum and suppression of the shake up satellite peak at 788 eV in the S 2p spectrum. The presence of sulfur is still evident, however the intensity of the signal is lower indicating the loss of sulfur from the sample and conversion into a material dominated by surface oxygen as evidenced by the XPS data.
Due to these changes the sample was also investigated after the HER and XPS analysis of the sample after the reaction in alkaline solution shows that the composition of the material is unchanged which is also reflected in the SEM images (Fig. S9†) that indicate no change in the morphology. This is significantly different to the OER under oxidative conditions where there is conversion of the CoSx material into an oxide material such as Co3O4 during the course of the reaction which is also an effective catalyst for this reaction. This phenomenon has been highlighted recently by Jin114 who described that metal chalcogenides, nitrides and phosphides may in fact only be precursors to the active material that participates in the OER. Once these materials are oxidised they are converted into the respective metal oxides/hydroxide which is also evident in this study. Indeed, recent work has utilised metal chalcogenides/phosphides as a scaffold to produce oxide/hydroxide OER active catalysts on the surface of the original materials which have demonstrated good OER performance.115–117 Therefore it is important for studies on metal sulphides that the catalyst is characterised post reaction to see if the composition and morphology of the catalyst is maintained rather than changed into a different form that is also active for the reaction of interest.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c7ra10394h |
This journal is © The Royal Society of Chemistry 2017 |