Ramón
Arcas
*a,
Yuuki
Koshino
b,
Elena
Mas-Marzá
a,
Ryuki
Tsuji
b,
Hideaki
Masutani
b,
Eri
Miura-Fujiwara
b,
Yuichi
Haruyama
c,
Seiji
Nakashima
d,
Seigo
Ito
*b and
Francisco
Fabregat-Santiago
*a
aInstitute of Advanced Materials (INAM), Universitat Jaume I, 12006 Castelló, Spain. E-mail: rarcas@uji.es; itou@eng.u-hyogo.ac.jp; fabresan@uji.es
bDepartment of Materials and Synchrotron Radiation Engineering, Graduate School of Engineering, University of Hyogo, 2167 Shosha, Himeji, Hyogo 671-2280, Japan
cLaboratory of Advanced Science and Technology for Industry, University of Hyogo, 3-1-2 Kouto, Ako, Hyogo 678-1205, Japan
dDepartment of Electronics and Computer Science, Graduate School of Engineering, University of Hyogo, 2167 Shosha, Himeji, Hyogo 671-2280, Japan
First published on 29th June 2021
Society is demanding clean energy to substitute greatly polluting carbon-based fuels. As an alternative, the use of green hydrogen produced by electrocatalysis constitutes a nice strategy as its products and reactants are not toxic to the environment. However, the use of scarce materials and high overpotentials to accomplish the oxygen evolution reaction (OER) make electrocatalysis an uncompetitive process. To solve these challenges, a low-cost procedure for the preparation of earth-abundant Ni, Fe and NiFe decorated electrodes has been developed. For this purpose, pencil graphite rods have been selected as highly porous substrates. A reasonable performance is achieved when they are employed for the OER. Furthermore, for the first time, a detailed analysis of impedance spectroscopy allows the association of the Ni redox transitions Ni2+/Ni3+ and Ni3+/Ni4+ (including the identification of the hydrated α–γ and the non-hydrated β phases) with an electrochemical redox capacitance response. Additionally, the Ni3+/Ni4+ redox peak capacitance together with a quick decrease in the charge transfer resistance indicates the implication of Ni4+ in the OER. These results show the utility of impedance spectroscopy as a non-destructive and non-invasive technique to study these electrochemical systems in detail under operating conditions.
Electrocatalysis is regarded as an appropriate technique to produce pure hydrogen by water splitting.3 To do this, external electric power is employed which, ideally, may be obtained from natural sources, e.g. solar or wind power.4 For an efficient energy conversion, the overpotentials needed to drive the oxygen evolution reaction (OER) and the hydrogen evolution reaction (HER) must be minimized. From a thermodynamic and kinetic point of view, the OER is the limiting reaction as it involves four electrons and a larger overpotential.5,6 Traditionally, noble metal oxides such as RuOx and IrOx, have been employed as anodes due to their high performance for the OER.7,8 Recently, RuOx has been deposited on pencil graphite rods (PGR) to obtain high-performance porous electrodes.9 However, the cost of these noble metals is high as they are not earth-abundant, which makes water splitting an uncompetitive process if they are used.10 In addition, the RuOx catalyst shows low stability in alkaline solutions. In this study, high OER activity was achieved using only inexpensive metal-based materials without using expensive precious metal-based materials.
Earth-abundant Ni-based materials have drawn interest as an attractive alternative to noble metals for water oxidation.11 Among the different Ni electrocatalysts, nickel oxide NiOx and Ni oxide hydroxide (NiOOH) have attracted great attention mainly due to their efficiency and robustness.12 Since Corrigan discovered the higher activity and the decrease in the overpotential of the OER when Fe was incorporated onto a NiOOH layer,13 a noteworthy approach has been developed to understand and design new electrodes with better OER performances. This made Ni1−xFexOOH layered double hydroxides (LDH) the best catalyst based on earth-abundant materials for water oxidation in alkaline media today.14 However, the performance of this material is strongly dependent on different factors, such as the chemical and electronic structure of the electrode, the electrochemical environment, the electrode preparation method, etc.15–21 A graphene-nanoplatelet-supported (Ni, Fe) metal–organic framework (MOF) with outstanding performance for water splitting in alkaline media and high stability has been reported.22 It is well-known that NiFe catalysts are more stable in alkaline aqueous solutions than other noble metal catalysts such as RuOx. It has been shown that the combination of Ru and Ni, in a compressed metallic Ru-core and oxidized Ru-shell with Ni single atoms (SAs), led to low overpotentials and high current densities in strong acidic media for water oxidation.23
In this paper, we study the electrochemical response of pencil graphite rods decorated with nickel and nickel-iron alloys to perform the oxygen evolution reaction. Techniques such as flame annealing (FA) or electrodeposition were used to decorate the PGRs obtaining reasonable results for the OER. Structural analysis was used to identify the oxidation states and contents of nickel and iron in the samples while electrochemical measurements allowed us to associate the differences in the electrical response of Ni2+/Ni3+ transition with the presence of hydrated and non-hydrated phases and showed the activation of the OER after Ni4+ formation.
The CV and IS measurements of the ED@Ni/PGR sample were performed using Fe free KOH electrolyte. The KOH employed in this experiment was treated by following the procedure reported by Boettcher et al.24 Ni(NO3)2·6H2O salt was dissolved in high purity KOH 1 M solution in order to precipitate Ni(OH)2. Once the salt was precipitated, the solution was centrifuged and the supernatant was decanted. This procedure was repeated three times until Ni(OH)2 was completely pure. Then, the pure Ni(OH)2 was redissolved in high purity 8 M KOH and stirred for 1 h. Finally, the purified electrolyte was centrifuged and decanted into a polypropylene bottle for its electrochemical use.
(1) |
Fig. 1 Schematic structure of (a) PGR, (b) graphite branches, and (c) details of graphite branches decorated with the catalyst for the OER. |
We tested FA@PGRs for the OER; however, they were not good catalysts for this reaction. For this reason, in order to improve the performance of PGRs, we developed new electrodes by the deposition of Ni and NiFe on the PGR (Fig. 1c). Ni was deposited following two different methodologies, namely flame annealing (FA) and electrodeposition (ED), obtaining the corresponding electrodes FA@Ni/PGR ED@Ni/PGR (see the Experimental section for experimental details). To prepare the PGR decorated with NiFe we followed only the FA procedure (FA@NiFe/PGR).
The morphology and the composition of the decorated PGRs were studied by SEM and EDX techniques (see Fig. S1–S3 in the ESI†). The SEM image of the ED@Ni/PGR sample showed an ordered and homogeneous surface composed of Ni particles of variable size, around 200 nm on top of the PGR. Compared to the ED sample, FA samples present smaller particles on the top of a disordered and non-homogeneous PGR surface. From the microanalysis made by SEM, we could conclude that the amount of Ni deposited by ED is higher than that by FA (see Fig. S1 and S2 in ESI†). Moreover, we could also determine the Ni/Fe ratio of each sample. For the FA@NiFe/PGR, the Ni:Fe ratio was ∼3:2. In the case of FA@Ni/PGR and ED@Ni/PGR iron was also detected (Ni:Fe ratio 9:1 and 24:1, respectively). The presence of iron in these two samples was due to contamination of iron of the bare PGR (∼0.2%).9 The presence of iron, even in small amounts, has implications for the performance of the electrodes which will be explained later.
XPS measurements were performed to investigate the chemical environment in the electrodes at the surface level (see Fig. S4 in the ESI†). The analysis of the high resolution XPS spectra and the data fitting of the Ni (2p) peak in each sample confirms the non-homogeneity environment of these samples. The Ni (2p) peak in the FA@Ni/PGR and FA@NiFe/PGR electrodes shows the presence of Ni in two different oxidation states, Ni2+ and Ni3+.26,27 The proportion of each oxidation state was quantified as a function of the peak area (Table S2, ESI†) and no differences were observed between these two samples. By contrast, the Ni (2p) peak in the ED@Ni/PGR sample points to the presence of Ni in the electrode as a mixed phase between metallic and oxidized Ni. The quantification of each state indicates the same concentration of Ni2+ species, indicating that ED led to Ni0 species instead of Ni3+.
The presence of Fe could also be observed by XPS as a Fe (2p) peak. The presence of Fe in the samples was detected as oxidation state species Fe2+ and Fe3+. This peak was already found for the bare PGR treated by FA (FA@PGR), confirming that impurities of Fe came from the PGR. Additionally, Fig. S4c† shows the high resolution XPS spectra of O (1s) for the electrodes. The presence of C–O interactions either as a double or singlet bond dominates in the sample response. The differences in the O (1s) spectra in the samples have been attributed to Ni–O and Fe–O interactions.28,29
The analysis of the FA@PGR sample in Fig. 2a shows a hysteretic behaviour in the flat area of the CV. This capacitive behaviour could be associated with the microporous structure of the rods and the ability of carbon to absorb small cations.30 In the case of the catalyst decorated samples, hysteresis cannot be clearly observed mainly due to the presence of the Ni(OH)2/NiOOH redox peak, which appears preceding the region corresponding to the OER.
The potential at which the Ni2+/Ni3+ redox peak appears, depends on both the Fe content in the catalyst and Ni phase. Specifically, the presence of more than 10% of Fe yields to the a smaller voltage difference between the Ni2+/Ni3+ redox peak and the onset of the OER. By contrast, lower amount of Fe in the Ni electrode separates both phenomena. Thereby, the Ni redox peak is displaced to lower overpotentials while, at the same time, the OER overpotential is higher.13,24 In our case, as can be seen in Fig. 2a, we observe the same trend. Thus, for ED@Ni/PGR with a Ni:Fe ratio of 24:1 (estimated by EDX) the wide redox peak for Ni2+/Ni3+ appears at an overpotential of 0.12 V vs. RHE in the forward direction (0.08 V vs. RHE in the reverse direction) with the onset overpotential for the OER occurring at 0.26 V. However, for FA samples, FA@Ni/PGR and FA@NiFe/PGR with Ni:Fe ratios 9:1 and 3:2, respectively (see Fig. S2 and S3 in the ESI†), both the redox peak and the OER onset are so close such that the two phenomena approach each other and overlap in the CV. From the steady-state J–V curves of the electrodes in Fig. 5d (and Fig. S6d in the ESI†), we could determine with better accuracy both the onset of the OER and, after subtracting the faradaic contribution, the peaks of the overpotentials of the redox states. Thus, for the two FA samples that have Ni:Fe ratios above 9:1, the same values are obtained for the Ni2+/Ni3+ redox peak, ∼0.20 V vs. RHE in the forward direction (and 0.09 V in the reverse direction) and for the onset overpotential for the OER, 0.22 V vs. RHE.
Focusing now on the performance of the electrodes, the interfacial overpotential needed to deliver 10 mA cm2 (of the geometrical area) is ∼270 mV for ED@Ni/PGR and ∼240 mV for FA@NiFe/PG, see Tables 1 and S1.† These values are in line with the ones reported for NiFe deposited onto supporting electrodes made of graphene and exfoliated graphite as reported in Table 1. Furthermore, they improve some of the published values obtained for planar metallic supporting electrodes shown also in Table 1.31,32 Therefore, these materials provide top performance electrodes for the OER while keeping an easy and low-cost processing method. We believe that the performance of our electrodes is due to their high surface area and the interaction between Ni and Fe. As can be seen from the data in Table 1, there is still room to improve the performance of the electrodes, for example minimizing series resistance by including carbon doping to improve its conductivity.
Catalyst | Electrode support | Electrolyte | H@10 mA cm−2 (mV) | References |
---|---|---|---|---|
NiFe | PGR | 8 M KOH | 252 (240) | This work |
Ni | PGR | 8 M KOH | 290 (270) | This work |
RuOx | PGR | 1 M KOH | 312 | 9 |
Ni0.8Fe0.2–AHNA | NiFe NW | 1 M KOH | 180 | 14 |
NiFe | EG | 1 M KOH | 214 | 31 |
NiFe | DG | 1 M KOH | 310 | 32 |
NiFe | Ni foam | 1 M KOH | 240 | 27 |
NiFe | Pt | 1 M NaOH | 340 | 33 |
NiFe | GNP | 1 M KOH | 280 | 22 |
Ni–Ru | RuOx | 0.5 M H2SO4 | 184 | 23 |
Ru | Pt | 1 M NaOH | 290 | 33 |
Tafel plots in Fig. 2b also showed the increase in the OER catalytic activities of PGR by the addition of the catalysts. In addition to the decrease of the onset potential observed for the decorated samples vs. the bare one, the Tafel slope is reduced ∼60 mV dec−1 when the catalyst is deposited on top of the PGR, which has a slope 120 mV dec−1 when uncoated. Tafel slopes of 60 mV dec−1 have been associated with the proton-coupled electron transfer mechanism for the OER.34
For a better understanding of the performance of Ni and NiFe on PGR electrodes, IS analysis in the area of the Ni2+/Ni3+ redox peak and the OER onset was performed. Fig. 3 shows the Nyquist and Bode plot data of FA@PGR and FA@NiFe/PGR obtained at overpotentials before and after the OER onset (−0.05 and 0.35 V vs. RHE). The plots obtained for ED@Ni/PGR are omitted here as they have minimum differences with the FA@NiFe/PGR sample. IS shapes are characterized by two features. At high frequencies (Fig. 3b), a 45° line appears which is followed by an arc at medium-low frequencies. This behaviour is described with a transmission line in two different situations: (i) when transport resistance (Rt) of electrons in the porous material is smaller than charge transfer resistance (Rct) towards the solution, the transmission line presents the shape obtained at low potentials (Fig. 3, line black and blue); (ii) in the opposite case, when Rt > Rct, the transmission line has the shape shown at high voltages (Fig. 3, line gray and cyan) which is given by a Gerischer element.35–38
The first situation is observed at low potentials when there is no current flow to the electrolyte. In this case, Rct is very large (≫Rtr) and the two samples show very similar impedance. When the current flow is activated at higher overpotentials (0.35 V vs. RHE), Rct becomes smaller than Rtr and the arc width (=[RtrRct]1/2) becomes smaller. The incorporation of the catalysts on the PGRs favours the OER performance; this effect is also observed on the IS measurements by the smaller arc in the NiFe case, indicating a better charge transfer (smaller Rct) for the decorated PGRs.
Detailed analysis of IS measurements at overpotentials between −0.2 and 0.4 V vs. RHE was performed by fitting the experimental data with the equivalent circuits (EC) proposed in Fig. 4. The EC suggested for these porous electrodes consists of a transmission line, where electrons are regularly distributed around the PGR (Fig. 4a). However, when the metal catalyst is incorporated into the PGR, the electron conduction occurs along the PGR catalyst line, i.e. where the catalyst is in contact with the PGR. In this new EC (Fig. 4b), a new capacitance is incorporated to describe the OER in the not catalyst covered area.39 The elements employed in these two EC are described as:
• Rs is the series resistance, which includes the resistance of the carbon rod out of the electrolyte together with the resistances of the bulk of the electrolyte and at the contacts.
• rtr is the electron transport resistance per unit length of the PGR electrode immersed in the electrolyte, which yields a total transport resistance Rtr = rtrL.
• rct,G is the charge transfer resistance at the graphite/electrolyte interface. In the case of the ED@Ni/PGR and FA@NiFe/PGR samples, it accounts for the charge transfer at the surface of the PGR which is not coated by the catalyst. At the macroscopic level, the total charge transfer at this interface is given by Rct,G = rct,G/L.
• cG is the capacitance given by the graphite/electrolyte interface. In the case of the decorated samples, it accounts for the uncoated PGR/electrolyte surface. The total capacitance of the graphite is given by CG = cGL.
• rct,cat is the charge transfer resistance associated with Ni or NiFe catalysts. The total contribution to the total charge transfer resistance of the catalyst is given by Rct,cat = rct,cat/L.
• ccat is the capacitance associated with the catalysts (Ni and NiFe) deposited on the PGRs and includes the contribution of their redox states. The total contribution of the catalyst capacitance is given by Ccat = ccatL.
• ZD is the impedance diffusion associated with the reactive species at the diffusion layer in the solution.
With these definitions, the charge transfer resistance and the capacitance of the PGR are given by the parallel combination of the graphite and the catalyst contributions, Rct−1 = Rct,cat−1 + Rct,G−1 and CPGR = Ccat + CG, which are the effective values we measure.
The parameters obtained from the IS fitting are represented as a function of the overpotential (Fig. S6†) and the interfacial overpotential (Fig. 5). Focusing on the capacitance data of the FA@PGR sample (Fig. 5a, black line), a nearly constant capacitance of ∼180 mF cm−2 corresponding to the PGR is observed, with a small peak at an η of 0.35 V, the overvoltage at which charge transfer activates in this sample. We associate this peak with the iron contents of FA@PGR, as a low amount of Fe was detected in this electrode by SEM microanalysis. When we purposely deposited Fe on PGR, the same peak with higher intensity was observed (Fig. S7†). The baseline capacitance observed for ED@Ni/PGR is ∼150 mF cm−2, and for FA@NiFe/PGR is ∼180 mF cm−2, both being attributed to the PGR. The higher Ni coverage of PGR in the ED sample causes a decrease in the PGR capacitance, as indicated in ECSA measurements.
When the Ni and NiFe are deposited on the PGR, three capacitance peaks are observed, suggesting a more complex explanation for the peaks observed in the cyclic voltammetry of Fig. 2a. In fact, and according to the literature, the two peaks observed at low overpotentials are attributed to two different Ni2+/Ni3+ redox transitions.40,41 As commented before, the energy of these transitions depends on the Fe content but also to the phase segregation in the sample, which is a function of the deposition environment and the flame annealing treatment.42 There are two possible phases for Ni(OH)2 molecules present at the Ni surface: β-Ni(OH)2, which is the normal and stable phase and the α-phase, a hydrated form of the nickel hydroxide, 3Ni(OH)2·2H2O. The formation of the α-phase is related to the ability of incorporating water between the layers during the nickel deposition on the PGR. We expect that the thermal annealing will yield to a larger proportion of the de-hydrated phase while electrodeposition, which is not thermally treated, will provide a larger amount of the hydrated phase. In any case, α to β conversion phase could happen as a consequence of aging and temperature.43
The oxidation of Ni2+ to Ni3+ by the application of a potential lead to the formation of γ- and β-NiOOH (hydrated and de-hydrated), respectively. According to the literature, α-Ni(OH)2 oxidizes to γ-NiOOH (α/γ) at η = 0.12 V (E0 = 1.35 V vs. RHE) and β-Ni(OH)2 oxidizes to β-NiOOH (β/β) at η = 0.20 V (E0 = 1.43 V vs. RHE),43–46 which match very well with the peaks found in the capacitance of the ED@Ni/PGR electrode in Fig. 5a. This result confirms that the small amount of Fe still present in the ED sample has a minimal effect in the Ni oxidation states, as discussed above. The larger height on the first peak indicates that α/γ transition is the most important one in this sample and dominates the peak in the CV. The β/β transition also occurs, causing the wide peak measured in the CV curve.
The third peak, that appears in Fig. 5a at an overpotential of ∼0.26 V (E0 = 1.49 vs. RHE), corresponds to the redox transformation between Ni3+/Ni4+,43,47,48 which some studies have associated with the formation of NiOO− after deprotonation of NiOOH.12 This peak is strongly affected by the iron concentration of the electrode and marks the starting of the quick decrease of Rct observed in Fig. 5c. Rct is related to the activation of the charge transfer mechanism that yields to the onset of the OER in the CV and the J–V curve in Fig. 2a and 5d, respectively.
For the FA@NiFe/PGR, a clear peak at an overpotential of 0.19 V vs. RHE is observed. We associated this peak to the β/β transitions of Ni2+/Ni3+ which is much larger than in the case of the ED and slightly displaced to more negative potentials. In this case, the data suggest that the α/γ transition contribution to the capacitance is much smaller than in the ED@Ni/PGR case, and only produces a distortion of the Ni2+/Ni3+ peak. Consequently, the peak observed in the CV of Fig. 2 is displaced towards positive values. This result matches well with results obtained in many previous reports.49–51 Therefore, these data suggest that the effect of Fe rather than displacing the Ni3+/Ni2+ peaks is favouring the presence of the β phases of Nickel rather than the α/γ.
For the third peak of the capacitance associated with Ni3+/Ni4+, now we see a reduction in the overpotential needed to make the redox transition to the Ni3+/Ni4+ which now occurs at 0.23 V, see Fig. 5d. Consequently, the OER is activated at an overpotential 30 mV smaller, in good agreement with literature data. Associated with this effect, we can see that the abrupt drop of Rct matches very well with the OER activation in the CV (Fig. 5d). The analysis of Rct shows even more clearly how the onset decrease in Rct is displaced towards smaller overpotentials as we move from the FA@PGR sample to the ED@Ni/PGR and then to the FA@NiFe/PGR, in perfect agreement with the J–V response observed in Fig. 5d.
Non-electrochemical techniques have been used to correlate the detection of the different Ni phase transitions for Ni(OH)2/NiOOH (α/γ and β/β) and the Ni3+/Ni4+ transition with electrochemical techniques.41,45 To the best of our knowledge, the IS analysis performed here has successfully demonstrated for the first time the suitability of this technique for that purpose, showing that IS is a low-cost, easy and useful tool to characterize these kind of samples.
Finally, the transport resistances of FA@NiFe/PGR and FA@PGR present similar values at low voltages, while the ED@Ni/PGR sample presents slightly larger values, see Fig. 5b. This result suggests that larger Fe concentrations produce larger conductivity of the Ni/PGR. Deeper analysis is needed to fully understand this behaviour, which is outside the scope of this paper. At the potentials of the Ni2+/Ni3+ redox transition, Rtr diminishes until a valley is formed, which we associate with the contribution of the redox species to the overall conductivity of the film.
Footnote |
† Electronic supplementary information (ESI) available: ECSA data, SEM images, XPS spectra and complementary impedance spectroscopy plots. See DOI: 10.1039/d1se00351h |
This journal is © The Royal Society of Chemistry 2021 |