Effects of Metal Combinations on the Electrocatalytic Properties of Transition-Metal-Based Layered Double Hydroxides for Water Oxidation: A Perspective with Insights.

Transition-metal-based layered double hydroxides (TM LDHs) have emerged as highly efficient water oxidation catalysts. They are promising and have the potential to replace the rare and expensive precious metal-based ones such as RuO2 and IrO2, which have been well established. In this perspective, we will summarize the current development of TM LDHs as oxygen evolution reaction (OER) catalysts toward electrochemical water splitting. Particular emphasis will be placed on the roles of the transition-metal cations and the effects of their combination on their catalytic performance for the OER. It is hoped that this perspective will provide fundamental guidelines for future researches in this booming area.

Introduction

The hydrogen and oxygen generation from water splitting is of great importance since it can simultaneously tackle the issues of energy crisis and environmental pollution problems. Indeed, it has aroused extensive research interest ever since the concept of water splitting was demonstrated, which has intensified in the recent few decades.[1] Electrochemical (EC) water splitting separates the whole process into two half-reactions, hydrogen evolution reactions (HERs) and oxygen evolution reactions (OERs). As is well known, compared to HER, OER is much more kinetically sluggish and requires a very high overpotential, thus limiting the water splitting efficiency and hindering the development of the hydrogen production industry based on water splitting.

The state-of-the-art catalysts of RuO2 and IrO2 show effective water oxidation activity.[2] However, the high cost of Ir ($16 181 kg–1) and Ru ($2000 kg–1) significantly limits their large-scale utilization.[3] Fortunately, the first-row transition-metal-based compounds with controlled chemical composition and microstructures have been found to have comparable water splitting performance, with much more bountiful resources and lower prices. Among these materials, transition-metal-based layered double hydroxides (TM LDHs) with the general formula of [M1–2+M3+(OH)2](A)·mH2O, consisting of edge-sharing hydroxyl coordinated octahedrons and intercalated anions between hydroxyl layers,[4] have been extensively studied, which surprisingly have exhibited even higher OER activities with overpotentials as low as ∼200 mV, which are smaller than that of RuO2 and IrO2 (∼250 mV). In spite of the tremendous research progress that has been made in developing TM LDH as efficient OER catalysts, the understanding of the underlying reasons for the high water oxidation performance of TM LDH is still limited. For instance, the synergistic effect between the divalent and trivalent cations and their effects on OER remain elusive. We notice that some excellent reviews have summarized the recent developments of TM LDH catalysts, but dedicated discussions on the effects of metal combinations in TM LDH on their catalytic performance are still missing despite the significance of such combinations.[2,5]

In this perspective, we survey the recent developments of TM LDHs with unary, binary, and ternary transition-metal ions and their OER catalytic performances, focusing on the effects of metal combinations that have strong influences on the OER activity. First, the basic criteria for evaluation of water oxidation catalysts including overpotential (η), Tafel slope (b), exchanged current density (i0), geometric current density (jg), and specific current density (j) will be introduced, followed by the discussion of the OER mechanism with which the TM LDH catalysts operate. Then, the specific TM LDHs with unary, binary, and ternary transition metals will be presented and the relationship between their chemical compositions and OER performance discussed. Explanations will be given on the active sites of NiFe LDH and the reactivity difference between different binary metal LDHs with binary transition metals. Finally, a summary and prospect will be provided encompassing the challenges and opportunities in this exciting research area.

Basic Criteria for OER Catalysts and Mechanisms of OER

The basic criteria to evaluate OER catalysis will be introduced in this section. Overpotential (η), calculated by the difference between the applied potential (E) to reach a certain current density and the equilibrium potential (Eeq) (as shown in eq ), is the critical and most frequently used descriptor to evaluate the performance of an OER catalyst. Eeq is the half reaction’s thermodynamically determined reduction potential, and E is the potential at which the redox event is experimentally observed. Generally, η is usually given at the onset of OER and at the current density of 10 mA cm–2, which corresponds to a 10% solar to hydrogen efficiency under 1 sun illumination. The existence of overpotential implies that the cell requires more energy than thermodynamically expected to drive a reaction. Therefore, a lower overpotential suggested a better OER activity.[2]Tafel slope (b), which describes the influence of potential/overpotential on steady-state current density, is another key descriptor to evaluate OER kinetics. The value of b could be calculated by eq , where R, T, and F are ideal gas constant, temperature, and Faradaic constant, respectively. α is the transfer coefficient that is highly related to Tafel slope. It has been reported that if b = 120 mV dec–1, the rate-determining step is dominated by the single-electron transfer step. If b = 60 mV dec–1, it hints that the chemical reaction after one-electron transfer reactions is the rate-determining step. If b = 30 mV dec–1, the rate-determining step is the third electron transfer step. Therefore, from the value of Tafel slope, we can roughly determine the rate-determining step of OER. Generally, small Tafel slope indicates fast reaction kinetics, and the rate-determining step is supposed to be at the ending part of the reaction.[2] Therefore, catalysts with small Tafel slopes often show good catalytic activity for OER. However, it should be noted that the Tafel slope is often overestimated if geometric current density is used because of the fact that the geometric current density is usually smaller than the specific current density. Moreover, Tafel slope is not accurate to describe the performance of OER catalysts because of its oversimplified assumptions.[6]Turnover frequencies (TOFs) refer to the turnover per unit time, representing the total number of moles transformed into the desired product by one mole of active site per time. Therefore, the number of TOFs determines the level of activity of the catalysts, which is given by eq , where j is the current density at a specified overpotential, A is the area of the electrode, and m is the number of moles of active materials deposited onto the electrodes.[7] Moreover, it has been suggested that TOFs at different overpotentials could be different; therefore, the applied overpotential should be provided when presenting TOF.[8]Exchange current density (i0) is defined as the current density at η = 0 (j0) divides surface area (A); the magnitude of i0 reflects the intrinsic charge transfer between reactant and catalyst (eq ). A higher i0 hints at better catalytic performance. i0 can be described by eq , in which k° is the rate constant, and ρ and ω are the reaction orders of Red and Ox, respectively.[9]Different from exchange current density, geometric current density (jg) is given by the current density normalized by geometric surface area at a certain overpotential. jg has practical meaning in developing water splitting devices; however, it usually overestimates the electrochemical performance of a catalyst due to the larger actual surface area than the geometric surface area. Figure shows the Tafel slopes of Co3O4 and IrO2 by using the current density calculated by Brunauer–Emmett–Teller (BET), electrochemical (EC), and disk surface area, respectively. From Figure , we can also see that the Tafel slope of Co3O4 is smaller than that of IrO2 with respect to the geometric surface area. One may rush to conclude that the OER performance of Co3O4 is better than IrO2. However, if the BET or EC surface area is used, the performance of Co3O4 is poorer than that of IrO2,[10] suggesting that the applied active surface area is quite important to determine the performance of a catalyst.[10] Generally, the more accurate the surface area used, the more accurate the current density that is obtained, leading to a more precise evaluation of the catalyst.

Figure 1

Influence of surface area on evaluation of a catalyst. Tafel plots of OER on IrO2[11] and Co3O4 made at 300 °C, in which the OER currents are normalized by the disk surface area, BET surface area, and EC surface area, separately. Reproduced with permission from ref (10). Copyright 2018 Elsevier Inc.

Understanding the mechanism of OER has a fundamental importance in designing new OER catalysts; therefore, let us first discuss the general mechanisms of OERs before presenting the detailed reactivity about LDH. The catalytic cycle is shown in Figure ; association mechanism and oxy–oxy coupling mechanism are generally proposed. There are four elementary steps for the association mechanism (eqs –9): association of hydroxide anions to form absorbed OH* accompanied by losing one electron, generation of reactive oxy intermediate O* from OH* with loss of one electron and generation one molecule of water, nucleophilic attack of absorbed oxy O* by the hydroxide anion with release of one electron to form O–O bond giving OOH*, and formation of one molecule of oxygen with release of an electron and one molecule of water to regenerate the catalyst and complete the catalytic cycle. For the oxy–oxy coupling mechanism (eqs , 7, and 10), one molecule of oxygen will be generated accompanied by the regeneration of catalyst after generating the oxy intermediate O*. For the association mechanism, the formation of OOH* is generally regarded as the rate-determining step due to the large energy barrier according to density functional theory (DFT) calculations.[12] For the oxy–oxy coupling mechanism (given by eqs , 7, and 10), the coupling between two oxy is supposed to withstand a very high kinetic barrier and thus is the rate-determining step.[13]

Figure 2

Catalytic cycle for the OER on transition-metal-based catalysts in alkaline conditions.

In the above-mentioned processes, formation of OOH* involves oxidation of oxygen from O* to OOH*, which is usually regarded as the rate-determining step. Therefore, LDH with high oxidation ability would facilitate the formation of OOH*. In addition, OER involves formation and cleavage of metal–oxygen bonds; in principle, catalysts with superior OER activity should possess a suitable oxygen bonding strength, neither too strong nor too weak. As oxidation ability and oxygen binding energy of LDH vary with the change of transition metals, they have a critical influence on the OER activity of LDH. Therefore, in this perspective, we will focus on the effects of chemical composition on the oxygen binding energy and oxidation ability of LDH. Unary, binary, and ternary transition-metal-based LDHs toward OER will be discussed.

Unary Metal-Based LDH

Unary metal-based LDHs exhibited limited OER activity, but they provide an ideal platform for us to understand the intrinsic OER activity of LDHs due to their structural simplicity. In this part, we will first introduce Ni-based LDH, followed by Fe-based LDH, Co-based LDH, and the recently reported V-based LDH. As transition-metal hydroxide and transition-metal-oxy/hydroxide can interconvert via Bode’s Diagram,[14] we also treated transition-metal oxyhydroxide as LDH for simplicity. The performance evaluation factors such as overpotential at 10 mA cm–2, Tafel slope, and TOF of unary metal LDHs are given in Table .

Table 1

Overpotential at 10 mA cm–2, Tafel Slope, and TOF in 1 M KOH of Unary-Transition-Metal-Based LDHs

catalysts	overpotential at 10 mA cm–2 (mV)	Tafel slope (mV dec–1)	TOF (s–1)	references
Ni(OH)2/NiOOH	297a	29	0.17	Boettcher[16]
γ-NiOOH	660b			Friebel and Bell[17]
γ-FeOOH	550b			Friebel and Bell[17]
α-Co(OH)2	400b	44	0.070	Wang[19]
β-Co(OH)2	463b	39	0.021	Wang[19]
β-CoOOH	426b	36	0.042	Wang[19]
Co LDH	393	59	0.001	Hu[25]
Co LDH	340	56	0.801	Kang and Yao[21]
monolayer Co LDH	350	45	0.003	Hu[25]
VOOH	270	68		Liang and Wang[24]

Overpotential at 1 mA cm–2.

In 0.1 M KOH solution.

VIII Group Single Transition-Metal Hydroxides/Oxyhydroxides

Ni-based compounds are the most widely used OER catalysts; actually, NiO was employed for OER early in the 1980s.[15] However, it did not arouse research interests until 2012 when Boettcher in situ generated nickel layered hydroxide/oxyhydroxide from NiO through an electrochemical conditioning process (Figure ).[16] The as-in situ-generated Ni hydroxide/oxyhydroxide exhibited an outstanding OER performance, with a low overpotential of 297 mV at 1 mA cm–2, an extremely small Tafel slope of 29 mv dec–1, and a considerably large TOF of 0.17 s–1 at η = 300 mV in 1 mol L–1 (M) KOH, better than that of the state-of-the-art catalyst of IrO catalysts (η = 378 mV at 1 mA cm–2, b = 49 mv dec–1, TOF = 0.0089 s–1 at η = 300 mV in 1 M KOH).

Figure 3

Proposed in situ transformation from the thermally prepared oxides to the layered hydroxide/oxyhydroxide structure from NiO. Reproduced with permission from ref (16). Copyright 2012 American Chemical Society.

NiFe LDH is the most effective for OER (will be discussed in Section ), but NiOOH and Ni(OH)2 are not so effective toward OER. Therefore, studying the OER activity of Fe-based LDH is important to understand the superior activity of NiFe LDH and has aroused much attention. Friebel and Bell studied the intrinsic OER activity of γ-FeOOH and found that the overpotential at 10 mA cm–2 of γ-FeOOH is 550 mV in 0.1 M KOH, which is smaller than that of Fe-free γ-NiOOH (η = 660 mV at 10 mA cm–2 in 0.1 M KOH). However, it is much higher than that of (Ni,Fe)OOH (η = 360 mV at 10 mA cm–2 in 0.1 M KOH).[17] Moreover, the calculations also indicated that the overpotential of γ-FeOOH is 520 mV, in good agreement with experiments.[17] Boettcher also studied the OER activity of FeOOH and suggested that FeOOH had high OER activity, but was limited by its poor conductivity, which has a measurable conductivity of 2.2 × 10–2 mS cm–1 only when the overpotential is larger than 400 mV.[18]

Similar to Ni and Fe, the remaining first-row group VIII transition metal, cobalt, can also form a layered double hydroxide structure and, of course, has received much interest. Wang compared the OER activities of α-Co(OH)2, β-Co(OH)2, and β-CoOOH[19] and found that α-Co(OH)2 will transform to γ-CoOOH before the OER, and the resulting γ-CoOOH inherits a large basal distance of α-Co(OH)2. It has an overpotential of 400 mV at 10 mA cm–2 in 0.1 M KOH, interestingly, a Tafel slope of 44 mV dec–1 when η is smaller than 350 mV, and 130 mV dec–1 when η is larger than 350 mV. Moreover, α-Co(OH)2 is more active than β-Co(OH)2, which might be due to the large interlayer space in the α-Co(OH)2.

Materials with ultrathin structures usually have high specific and exposed surface area and are abundant in vacancies, leading to a higher number of active sites and thus higher activities. Pan and Wei synthesized an atomically thin γ-CoOOH, with a thickness of only 1.4 nm. Expectedly, the as-prepared γ-CoOOH has very high mass activities and abundant-active sites, thus leading to a sharp increase of OER activity with η = 300 mV at 10 mA cm–2 and a Tafel slope of 38 mV dec–1 in 1 M KOH. Interestingly, the as-prepared γ-CoOOH is half-metallic in contrast to its bulk, which was proposed to be related to the presence of dangling bonds in the CoO6– octahedron as supported by DFT calculations.[20] Kang and Yao also prepared Co-based LDH with atomic thickness for OER. Owing to its ultrathin structure, the OER activity of the Co-based LDH can have an overpotential of 340 mV at 10 mA cm–2, a Tafel slope of 56 mV dec–1, and a TOF of 0.801 s–1 at η = 350 mV in 1 M KOH.[21] Besides ultrathin structures, a larger interlayer spacing of LDH can also lead to a higher number of active sites. Sun and Chen reported benzoate anion interacted with CoOOH with an interlayer spacing as large as 14.72 Å, which allows the easy permeation of water and hydroxide, resulting in a higher number of active sites.[22] And the overpotential at 50 mA cm–2 is only 291 mV in 1 M KOH.

V-Hydroxides/Oxyhydroxides

In addition to the intensively studied VIII transition-metal-based LDH, V and Mn compounds have also been investigated. In 2012, Markovic studied the trends for OER on 3d transition-metal hydr(oxy)oxide catalysts (M2+δOδ(OH)2−δ/Pt(111)) and discovered that the reactivity toward OER is in the order Mn < Fe < Co < Ni, which is governed by the OH–M2+δ bond strength (Ni < Co < Fe < Mn).[23] According to the Sabatier principle, too weak or too strong M–OH bonds would retard the OER reactivity. Copper and zinc have too many d electrons in the d orbitals, which cause strong repulsions between the d electrons and 2p electrons of oxygen. Expectedly, Cu(OH)2 and Zn(OH)2 exhibited poor OER activities. Early transition metals, such as titanium, have few d electrons in the d orbitals, which lead to very strong OH–M bonds, and are supposed to be ineffective toward OER.[23]

Despite that the early transition-metal hydroxide has a strong M–OH bond, which is proposed to be unfavorable for OER, VOOH hollow nanospheres, structurally resembling lepidocrocite γ-FeOOH as evidenced by their similar X-ray diffraction (XRD) pattern (Figure a), have been employed as an efficient OER catalyst by Liang and Wang,[24] exhibiting an overpotential of 270 mV for OER at 10 mA cm–2 and a Tafel slope of 68 mV dec–1 in 1 M KOH. It has been well established that V is an early transition metal, favoring its high oxidation states, +5 and +4, whereas V has an oxidation state of +3 in VOOH. It may be of concern that VOOH is unstable during OER; however, the stability test given in Figure b shows that VOOH is quite stable during OER. And the reactivity of VOOH does not decrease even after 5000 cycles, as shown in the inset of Figure b. Moreover, the prepared VOOH can be used as HER catalyst, with an overpotential of 164 mV at 10 mA cm–2 and a Tafel slope of 104 mV dec–1. The advanced water splitting performance of VOOH in this work was contributed to the large surface area from the hollow sphere morphology.

Figure 4

XRD patterns of VOOH (a), and long-term stability test of VOOH at 10 mA cm–2 for 24 h (b). The inset is the polarization curves of the VOOH catalyst recorded before and after 5000 sweeps. Reproduced with permission from ref (24). Copyright 2016 John Wiley and Sons.

Binary Metal-Based LDH

Unary-transition-metal-based LDHs are limited by their low intrinsic OER activity or conductivity. Fortunately, by doping the second metal ions into the unary-TM LDH, the as-formed binary LDHs such as NiFe LDH, NiCo LDH, and CoFe LDH showed a much higher OER performance, which will be summarized and discussed in this section. The performance evaluation factors such as overpotential at 10 mA cm–2, Tafel slope, and TOF of binary metal LDHs are given in Table .

Table 2

Overpotential at 10 mA cm–2, Tafel Slope, and TOF in 1 M KOH of Binary Metal-Based LDHs

catalysts	overpotential at 10 mA cm–2 (mV)	Tafel slope (mV dec–1)	TOF (s–1)	references
(Ni,Fe)OOH	336	30		Boettcher[16]
NiFe LDH	208	48	0.028	Yang[7]
NiFe LDH	245	40.4		Kuang and Sun[54]
monolayer NiFe LDH	300	40	0.05	Hu[25]
NiFeCNTs LDH	290a	31	0.56	Dai[26a]
NiFe-rGO LDH	195	39	0.987	Yang[7]
NiFe-GO LDH	210	40	0.38	Yang[7]
NiFe LDH/Ti3C2-MXene	300	43	0.26	Wang and Qiu[28]
NiFe LDH/Ni3S2 nanorods	190	38		Zhang and Huang[27]
NiCo LDH	271	72		Qian and Li[36]
NiCo LDH	290	113		Jiang and Ai[37]
NiCo LDH	282	42.6		Huang[38]
NiCo LDH	385	65	0.0025	Hu[25]
monolayer NiCo LDH	338	41	0.01	Hu[25]
NiMn LDH	640a			Sun[39]
NiMn LDH	312			Huang[38]
NiV LDH	250	50	0.054	Sun[41]
NiTi LDH	420	290	0.009	Zhang[40]
CoFe LDH	321	57.05	1.12	Wang[46]
ultrathin CoFe LDH	266	37.85	4.78	Wang[46]
CoFe LDH	290	83.0		Li and Ge[45]
CoMn LDH	325	43	0.075	Hu[48]
CoMn LDH	255	38		Cheng and Liu[49]
CoCr LDH	240	81	0.195	Huang[50]
ZnCo LDH	340		0.061	Asefa[51]

Overpotential at 5 mA cm–2.

NiFe LDH

LDH suffers from low conductivity, which is one of the major challenges needed to be overcome for an advanced OER performance. Therefore, various methods have been applied to tackle this problem.[7,12b,16,17,26] In 2013, Dai deposited NiFe LDH on carbon nanotubes (CNTs) to obtain NiFe-CNT LDH based on the fact that the end tips of carbon nanotube (CNT) have many carboxylic groups that can coordinate to metal centers (Figure a).[26a] The as-prepared NiFe-CNT LDH exhibited a Tafel slope of 31 mV dec–1 at 1 M KOH and has an overpotential of 290 mV at 5 mA cm–2. The TOF is 0.56 s–1 at an η = 300 mV in 1 M KOH. Moreover, the NiFe-CNT LDH exhibited a better stability than Ir/C at a constant current density of 5 mA cm–2. In 2014, Yang reported the graphene oxide (GO) intercalated NiFe LDH by substitution of anions (CO32− or Cl−) with GO (Figure b).[7] The as-prepared NiFe-GO LDH has an overpotential as low as 210 mV at 10 mA cm–2 and a Tafel slope of only 40 mV dec–1 in 1 M KOH. Moreover, the TOF can reach 0.38 s–1 at η = 300 mV. Since the conductivity of GO can be enhanced by reduction, we further reduce NiFe-GO LDH by hydrazine to obtain reduced GO intercalated NiFe-reduced graphene oxide (rGO) LDH; the overpotential and Tafel slope can be further lowered to 195 mV at 10 mA cm–2 and 39 mV dec–1, respectively. Moreover, the TOF can reach 0.98 s–1 at η = 300 mV. The higher conductivity of FeNi-rGO LDH is evidenced by alternating current impedance spectra. It should be noted that the basal spacing of FeNi-GO LDH is 1.1 nm, which is much larger than the basal spacing of FeNi-CO3 LDH (0.75 nm), suggesting that GO has successfully intercalated into the NiFe LDH. And the enlarged basal distance of NiFe LDH allows efficient association of reactants and dissociation of products, and that is why a sharp increase of TOF is observed. Later, NiFe LDH-GO and NiFe LDH-rGO have also been reported through solvothermal and chemical reduction methods by Zhan and Hou.[26j]

Figure 5

Schematic layout of the structure of NiFe-CNT LDH (a),[26a] NiFe-GO LDH (b),[7] and NiFe LDH on conducting Ni3S2 nanorods (c).[27] Reproduced with permission. Copyright 2013 American Chemical Society, Copyright 2014 John Wiley and Sons, reproduced by permission of The Royal Society of Chemistry.

Besides combination with conductive carbon materials, other conducting materials were also applied to reduce the charge transfer resistance of the LDHs. Wang and Qiu prepared hierarchical NiFe LDH/Ti3C2-MXene for OER.[28] Owing to the stabilization by Ti3C2-MXene, better conductivity, and electronic interaction between Ti3C2-MXene and NiFe LDH, the as-prepared hierarchical NiFe LDH/Ti3C2-MXene exhibited a Tafel slope of 43 mV dec–1 and overpotential of ∼300 mV at 10 mA cm–2 in 1 M KOH. The TOF at η = 300 mV is 0.26 s–1. Zhang and Huang grew NiFe LDH on conducting Ni3S2 nanorods, as shown in Figure c.[27] As Ni3S2 nanorods are much more electron rich than NiFe LDH, electrons will flow to NiFe LDH, leading to the partial reduction of NiFe LDH. To maintain the charge neutrality, oxygen vacancies are generated in NiFe LDH as evidenced by X-ray photoelectron spectroscopy (XPS). As a result, the intrinsic activity of NiFe LDH was enhanced. Moreover, the number of active sites also increased after being loaded on the Ni3S2 nanorods. The overpotential at 10 mA cm–2 is as low as 190 mV, and the Tafel slope is also only 38 mV dec–1 in 1 M KOH. Wang reported NiFe LDH@Au hybridized nanoarrays on nickel foam. The as-prepared NiFe LDH@Au/Ni foams exhibited overpotentials of only 221, 235, and 270 mV at 50, 100, and 500 mA cm2–, respectively, in 1 M KOH. Moreover, the Tafel slope decreased to 48.4 mV dec–1, in comparison to NiFe LHD/Ni foam with a Tafel slope of 71.1 mV dec–1.[29] Huang reported single-crystalline NiFe LDH array on a Ni foam, and the as-prepared NiFe LDH array showed an excellent OER activity.[30] The overpotentials at 10, 50, and 100 mA cm–2 are only 210, 240, and 260 mV in 1 M KOH, which is smaller than the coated NiFe LDH film. Moreover, the Tafel slope is 31 mV dec–1 in the overpotential region of 240–260 mV, suggesting a faster OER kinetics and large current density. Xie, Zheng, and Sun grew an amorphous NiFe-borate layer on a NiFe LDH surface and found that the OER performance was greatly boosted owing to the higher surface roughness and increased number of active sites.[31] NiFe LDH is effective toward OER but suffers from lower conductivity. It has been reported that substitution of the very electronegative oxygen to other less electronegative elements, such as sulfur,[32] selenium,[33] phosphorus,[34] and nitrogen,[35] can push up the valence bond of NiFe LDH, thus leading to a higher conductivity and a better OER activity.

Other Ni-Based Binary Metal LDHs

Besides NiFe LDHs, other Ni-based bimetal LDHs, including NiCo, NiMn, NiCr, NiTi, NiV, NiGa, and NiAl LDHs, also have been well studied, which will be discussed in this section.

Qian and Li reported NiCo LDH nanosheet arrays on Ni foam for overall water splitting.[36] The as-prepared NiCo LDH has an overpotential of 271 mV at 10 mA cm–2 and a Tafel slope of 72 mV dec–1 in 1 M KOH. And it should be noted that it was used as a bifunctional catalyst for overall water splitting for the first time, which showed an overpotential of 162 mV for HER at a current density of 10 mA cm–2. Similarly, Jiang and Ai reported NiCo LDH nanosheets for OER that has an overpotential of 290 mV at 10 mA cm–2, with the Tafel slope being 113 mV dec–1 in 1 M KOH.[37] And XPS characterizations suggested Co3+ and Co2+ coexist in the prepared NiCo LDH. NiCo LDH nanosheet has also been reported by Huang, which shows an overpotential of 282 mV at 10 mA cm–2 in 1 M KOH, and the Tafel slope is 42.6 mV dec–1.[38] By postexfoliation treatment, Hu reported a single layered NiCo LDH, which showed a much increased OER activity than the bulk counterpart.[25] The overpotenial at 10 mA cm–2 and Tafel slope decreased from ∼390 mV and 65 mV dec–1 to ∼334 mV and 41 mV dec–1, respectively, whereas the TOF increased from 0.025 to 0.01 s–1.

The early or middle transition-metal-Ni LDHs have also been studied, but admittedly, they are relatively less studied compared with NiFe LDH and NiCo LDH because the formed M–OH and M–O bonds are too strong, which is unfavorable for OER. For example, Sun reported NiMn LDH for OER with an overpotential of 640 mV at 20 mA cm–2 in 1 M KOH; however, it was less effective than NiFe LDH (401 mV at 20 mA cm–2 in 1 M KOH).[39] NiMn LDH nanosheet has also been synthesized by Huang for overall water splitting.[38] However, in this work, the NiMn LDH with an overpotential of 312 mV at 10 mA cm–2 is less effective than NiFe LDH with an overpotential of 220 mV at 10 mA cm–2. NiTi LDH, which acts as a precursor to prepare NiO-TiO2 ultrafine nanosheet,[40] is ineffective for OER. The Tafel slope is as high as 290 mV dec–1, and the TOF at an overpotential of 500 mV is only 0.009 s–1. Surprisingly, monolayer NiV LDH, which shows a comparable OER activity with NiFe LDH, has been reported by Sun.[41] V exists in V3+, V4+, and V5+, as evidenced by XPS shown in Figure a. The presence of V4+ and V5+ is attributed to the oxidation of V3+ during the synthesis. And the as-prepared monolayer NiV LDH (Ni0.75V0.25-LDH) has an overpotential of 250 mV, a Tafel slope of 50 mv dec–1, and a TOF of 0.054 ± 0.003 s–1 in 1 M KOH, whereas the related monolayer NiFe LDH (Ni0.75Fe0.25-LDH) has a overpotential of 300 mV, Tafel slope of 0.064 mV dec–1, and a TOF of 0.021 ± 0.003 s–1. Electrochemical impedance spectroscopies (EISs) indicated that Ni0.75V0.25-LDH has a lower charge transfer resistance, thus bearing with a high conductivity. Moreover, the as-prepared NiV LDH exhibited considerable stability, as shown in Figure b. DFT calculations have been performed on the OER mechanism catalyzed by Ni0.75V0.25-LDH. Here, V is supposed to be the active site, where H2O*, OH*, O*, OOH*, and OO* bind on V, and the rate-determining step is formation of OOH* from O*, which has an overpotential of 620 mV.

Figure 6

Zoom on the V 2p core-level XPS measurements of NiV LDH power deposited on an fluorine-doped tin oxide (a); long-term stability test of NiV LDH and NiFe LDH (b). Reproduced with permission from ref (41). Copyright 2015 Nature Publishing Group.

In addition, some main-group element-Ni LDHs are also reported; as expected, their OER activity is very low. To synthesize porous β-Ni(OH)2, Wang and Jin first prepared NiGa LDH in which the Ga3+ will be removed through base etching.[42] And it was found that the prepared NiGa LDH exhibited a little better OER activity than the β-Ni(OH)2 nanosheet, but it was much poorer than the porous β-Ni(OH)2 prepared by etching NiGa LDH. Similarly, the NiAl LDHs are also prepared as a precursor to synthesize the porous LDH. For example, Zhang and Xie employed NiAl LDH as a precursor to prepare β-Ni(OH)2 ultrathin nanomesh.[43] The prepared β-Ni(OH)2 ultrathin nanomesh has an excellent OER activity, whereas the NiAl LDH only shares a similar OER activity with the β-Ni(OH)2 nanosheet.

Co-Based Binary Metal LDH

Cobalt, with one less d electron than nickel, also received much attention.[44] Li and Ge reported CoFe LDH fabricated by coprecipitation, in which the ratio of Co to Fe can be tuned from 0.5 to 7.4.[45] The Co2Fe LDH gave the best OER activity with an overpotential of 290 mV at 10 mA cm–2 and a Tafel slope of 83 mv dec–1 in 1 M KOH. Again, ultrathin LDHs with abundant defects are supposed to have higher OER activities. Wang prepared ultrathin CoFe LDHs through Ar plasma etching.[46] The Ar plasma etched has a thickness of 0.6 nm, compared to the bulk with a thickness of 20.6 nm. The ultrathin CoFe LDHs are abundant in Co, Fe, and O vacancies, as evidenced by the decreased coordination numbers of Co, Fe, and O. The ultrathin CoFe LDH-Ar has a TOF of 4.78 s–1, whereas the bulk CoFe LDH has a TOF of 1.12 s–1. And the Tafel slope and overpotential at 10 mA cm–2 of CoFe LDH-Ar is 37.85 mV dec–1 and 266 mV, respectively, which are much lower than those in bulk CoFe LDH (57.05 mV dec–1 and 321 mV at 10 mA cm–2). It should be noted that ultrathin CoFe LDH-Ar has a much smaller charge transfer resistance. Xiong and Sun reported an ultrathin CoFe-borate-layer-coated CoFe LDH nanosheet array supported on Ti mesh. The as-prepared catalysts showed good OER activity at near-neutral condition (0.1 M K2B4O7 solution, pH = 9.2), with an overpotential of 418 mV at 10 mA cm2–.[47]

Boettcher studied the roles of Co and Fe in CoFe LDH.[18] Compared with CoOOH (rigorously Fe free) films that exhibited a TOF of 0.007 ± 0.001 s–1 and FeOOH films that showed a TOF of 0.016 ± 0.003 s–1, CoFeOOH with x between 0.4 and 0.6 has a high TOF up to 0.61 ± 0.10 s–1. Moreover, the Tafel slope decreased to 26–39 when x was in the range from 0.33 to 0.79, from 62 mV dec–1 for CoOOH to 45 mV dec–1 for FeOOH (Figure ). Moreover, it has been found that the Co2+/3+ wave shifts anodically with the increase of Fe amount. Considering the low conductivity and unstable character of FeOOH under the OER process in alkaline condition, CoFe LDH with the incorporation of Co had a higher conductivity, hence showing a higher OER activity.

Figure 7

Tafel slopes from the second cyclic voltammetry cycle (10 mV s–1) taken before (solid circle) and after (open circle) a 2 h polarization at η = 350 mV. Dotted lines link the pre- and postpolarization values. Reproduced with permission from ref (18). Copyright 2015 American Chemical Society.

Hu reported ultrathin CoMn LDH (3.6 nm) for OER; the as-prepared CoMn LDH has an overpotential of 325 mV at 10 mA cm–2 and Tafel slope of 43 mV dec–1 in 1 M KOH, better than the sum of Co(OH)2 and Mn2O3.[48] Interestingly, after anodic conditioning, the overpotential can be further reduced to 293 mV, which is proposed to be related to the accumulation of Co(IV) species in the amorphous layers. Recently, Cheng and Liu reported strongly electrophilic Mn4+-doped CoOOH nanosheet (i.e., CoMn LDH) for OER.[49] Theoretical calculations indicated that incorporation of Mn4+ leads to higher occupancy at the Fermi level (mainly conduction band), thus facilitating electron transfer in the CoMn LDH. Moreover, it was found that incorporation of Mn4+ enhanced the binding of OH– to Co by 0.7 eV. Owing to the increased conductivity and stronger OH– binding energy, CoMn LDH exhibited a higher OER activity with η = 255 mV at 10 mA cm–2 and Tafel slope of 38 mV dec–1 in 1 M KOH.

Considering that Co2+ is the OER active site and Cr3+-based oxides always exhibit good conductivity, Huang synthesized CoCr LDH for OER aiming for a good conductivity and OER activity.[50] Indeed, the conductance of CoCr LDH is 4.5 times and 21.4 times higher than that of CoOOH and Co(OH)2, respectively. Different atomic ratios of Co and Cr were studied, and it was found that the Co2Cr LDH gave the best OER activity with an overpotential of 240 mV at 10 mA cm–2 and a Tafel slope of 81.0 mV dec–1, which are the most active among the best of Co-based candidates. And the high activity was proposed to be contributed by the modified electronic structure, improved surface areas, and better conductivity introduced by Cr3+. Asefa prepared ZnCo LDH for water and alcohol oxidation in which the ratio of Co3+ to Co2+ was 1 and the ratio of Co to Zn was also 1.[51] The overpotential of ZnCo LDH is 0.34 V in 0.1 M KOH solution, which is lower than that of Co3O4 and CoOOH. The TOF at an η of 700 mV can reach up to 0.88 s–1. It should be noted that ZnCo LDH has a much smaller Faradaic impedance than Co3O4. Although it was suggested that Zn is inactive for OER and Co is the active center, the Zn2+ was proposed to facilitate the formation of highly oxidized Co ions in ZnCo LDH, similar to the role of Ca2+ in [Mn3CaO4]6+ catalyst. Later on, Xiang and Yan first mixed ZnSO4 and CoSO4 and then added some H2O2 to oxidize Co2+ to Co3+, followed by electrodepositing it to a three-electrode configuration to prepare ZnCo LDH.[52] The prepared ZnCo LDH has an overpotential of 427 mV at 2 mA cm–2 in 1 M KOH (510 mV at 10 mA cm–2), and the Tafel slope is 83 mV dec–1. The TOF at η = 700 mV is 3.56 s–1, much higher than that of ZnCo LDH prepared by the coprecipitation method.[51] CoAl LDH has been reported as a precursor to prepare Al-doped CoP nanoarrays for overall water splitting.[53]

Ternary Metal-Based LDH

Although binary metal-based LDHs exhibited an improved OER activity than unary metal-based LDH, they often suffer from poor conductivity. One may think incorporation of a third metal could involve new states in the forbidden band of binary metal-based LDH, thus leading to a higher conductivity. Moreover, introduction of another metal might increase the number of active sites. The overpotential at 10 mA cm–2, Tafel slope, and TOF of ternary metal LDHs are given in Table . By the way, there are some examples of Ru-[55] and Ir-[56]doped NiFe LDHs but with the focus on the HER in alkaline condition, which will not be discussed here.

Table 3

Overpotential at 10 mA cm–2, Tafel Slope, and TOF in 1 M KOH of Ternary Metal-Based LDHs

catalysts	overpotential at 10 mA cm–2 (mV)	Tafel slope (mV dec–1)	TOF (s–1)	references
NiCoFe	265	47	0.106	Yan and Strongin[59]
NiCoFe	210	42	0.7	Yang[57]
NiFeMn LDH	289a			Sun[39]
NiFeV LDH	195a	42		Liu and Sun[61]
NiFeAl LDH (Al%: 6.1%)	310a	69.6		Kuang and Sun[54]
D-NiFeAl LDH (Al%: 3.9%)	270a	46.1		Kuang and Sun[54]
NiFeZn LDH (Zn%: 25.2%)	298a	63.8		Kuang and Sun[54]
D-NiFeZn LDH (Zn%: 3%)	200a	34.9		Kuang and Sun[54]

Overpotential at 20 mA cm–2.

In 2014, Yang reported ultrathin NiCoFe LDH for OER.[57] The prepared Ni8Co2Fe LDH has a low overpotential of 210 mV at 10 mA cm–2, a Tafel slope of 42 mV dec–1, and TOF of 0.7 s–1 in 1 M KOH at an overpotenital at 300 mV, outperforming the Ni10Fe LDH with η = 210 mV at 10 mA cm–2, Tafel slope of 55 mV dec–1, and TOF of 0.53 at η = 300 mV. The specific surface area of Ni8Co2Fe LDH is 80.44 m2 g–1, much larger than that of Ni10Fe LDH (46.05 m2 g–1), suggesting that Co incorporation might lead to more exposed active sites. In addition, charge transfer resistance also decreased a lot upon incorporation of Co. Inspired by the fact that Ni2+/Co2+ confinement in the interlayer space of birnessite enhanced the OER activity,[58] Yan and Strongin prepared the cobalt intercalated NiFe LDH aiming for a better OER activity.[59] As the Co–O band length is found to be between CoOOH and Co(OH)2 based on extended X-ray absorption fine structure fit results, Co2+ and Co3+ are coexisting. The performed DFT calculations indicated that substitution of Ni2+ by Co2+ can lower the overpotential of OER (780 mV) by decreasing the Gibbs free energy differences between O* and OOH*, whereas substitution of Fe3+ by Co3+ leads to even lower overpotential (680 mV) of OER. The higher OER activity is thought to be the modified binding strengths of O* and OOH* due to the hybridization of 3d orbitals of Co and 2p orbitals of O at the valence band maximum. NiCoFe LDH has also been reported by Reguera with an overpotential of 250 mV at 10 mA cm–2.[60] In this contribution, Co3+ in low-spin configuration was thought to act as a shield of iron effect over nickel, which modulates the necessary potential for nickel oxidation.

Further, Sun reported that ternary NiFeMn LDH for OER with an overpotential of 289 mV at 20 mA cm–2 in 1 M KOH is more efficient than NiMn LDH, which has an overpotential of 640 mV at 20 mA cm–2. Moreover, the ternary NiFeMn LDH is also more effective than NiFe LDH (401 mV at 20 mA cm–2).[39] The higher OER activity of NiFeMn LDH compared to NiFe LDH was attributed to the fact that Mn4+ modifies the electronic structures of NiFe LDH, hence leading to a higher conductivity, as evidenced by DFT calculations and sheet resistance.

Liu and Sun reported NiFeV LDH with an overpotential of only 195 mV at 20 mA cm–2 and a Tafel slope of 42 mV dec–1 in 1 M KOH solution,[61] much better than NiFe LDH (η = 249 mV at 20 mA cm–2, b = 49 mV dec–1 in 1 M KOH) and NiV LDH (η = 330 mV at 20 mA cm–2, b = 72 mV dec–1 in 1 M KOH). The higher OER activity of NiFeV LDH was ascribed to the increased conductivity modified by V doping, as evidenced by DFT calculations and EIS. In addition, the ECSA of NiFeV is also larger than NiFe LDH, suggesting a higher number of active sites of NiFeV LDH.

To prepare NiFe LDH with atomic-scale defects, Kuang and Sun first prepared NiFeZn and NiFeAl LDH precursors; then, the Al and Zn were partially removed by alkaline etching to obtain defects containing NiFeZn and NiFeAl LDH, denoted D-NiFeZn LDH and D-NiFeAl LDH, respectively.[54] D-NiFeAl LDH has a lower OER activity than NiFe LDH, but higher activity than NiFeAl LDH. Interestingly, D-NiFeZn LDH shows a much higher OER activity than NiFe LDH and NiFeAl LDH, with an overpotential of only 200 mV at a current density of 20 mA cm–2 in 1 M KOH, and the resulting D-NiFeZn LDHs are abundant it the Ni–O–Fe unit, which is regarded as the active site. DFT calculations suggested that formation of O* from OH* is the rate-determining step, and the D-NiFeZn LDH has a lower overpotential than D-NiFeAl LDH. Interestingly, Rezvani and Habibi reported that ternary NiFeZn has a superior OER activity than binary NiFe LDH.[62] Moreover, the OER can take place at neutral condition, and the Tafel slope can be reduced to as low as 16 mV dec–1, whereas the NiFe LDH has a Tafel slope of 29 mV dec–1. However, the current density is only 5.41 mA cm–2 at an overpotential of 300 mV. The better OER performance of ternary NiFeZn was attributed to its higher conductivity modified by Zn2+.

Mechanistic Studies

Although transition-metal-based OER catalysts have been studied experimentally, the theoretical investigation of their structure–composition–performance relationship is far from satisfactory because of the inaccurate models. Actually, only the structure of β-Ni(OH)2 has been accurately determined,[63] which belongs to the P3̅m1 (brucite) space group, and the lattice parameters are a = b = 3.12 Å and c = 4.66 Å. Carter performed DFT + U calculations and found that the pure β-NiOOH has a proton-staggered structure and is antiferromagnetic (Figure ).

Figure 8

Experimental structure of β-Ni(OH)2 and calculated structure of β-NiOOH. Reproduced with permission from ref (64). Copyright 2015 American Chemical Society.

Studying the surfaces of catalysts has fundamental importance because most of the OERs take place at the surfaces. Different surfaces of the catalyst could lead to a different reactivity. As β-NiOOH shares a similar backbone with LDH and can be modified to be a more efficient OER catalyst, Carter performed theoretical calculations on the stability and chemistry of β-NiOOH at the vacuum and aqueous conditions.[65] The calculations suggested the surface stability in the vacuum follows the order (0001) > {011̅N} ≫ {012̅N}, N = 0, 1 due to the presence of dangling bonds at {011̅N} and {012̅N} surfaces. However, the order of stability becomes (0001) > {011̅N} ≈ {012̅N} in aqueous condition, owing to the water dissociation and adsorption at the {011̅N} and {012̅N} surfaces, which decreased the numbers of dangling bonds. Here, only low-index surfaces are studied on the consideration that low-index surfaces are much more stable than high-index surfaces and that there is no experimental evidence of generation of high-index surfaces during reactions (Figure ).

Figure 9

Represented low-index surface of β-NiOOH in aqueous condition. Reproduced with permission from ref (65). Copyright 2018 American Chemical Society.

Owing to its simplicity and great importance, the OER mechanism catalyzed by NiOOH has been well studied.[14b,66] Considering that the (0001) surface is the most stable, OER takes place on the (0001) surface most likely. Carter calculated the single-site association mechanism, binuclear H2O–O mechanism, binuclear OH–OH mechanism, and binuclear H2O2 mechanism on the (0001) surface of β-NiOOH and found that the binuclear H2O2 mechanism is energetically highly unfavorable, whereas the binuclear H2O–O mechanism and the binuclear OH–OH mechanism have the lowest overpotentials (∼0.5 V), lower than the single-site association mechanism (∼0.6 V).[66] However, as mentioned in Section , the binuclear mechanism is kinetically unfeasible.[13]

Nørskov and Bell performed DFT calculations to study the OER mechanism catalyzed by β-CoOOH, and it was found that the (101̅4) surface (η = 480 mV) gives a lower overpotential than the (011̅2) surface (η = 800 mV). The (101̅4) surface is abundant in Co3+ ions, whereas the (101̅4) surface has more Co4+ ions. Association of OH– to generate OH* would oxidize Co2+ to Co3+ on the (101̅4) surface, whereas it oxidizes the Co3+ to Co4+ on the (011̅2) surface. Transformation of Co3+ to Co4+ leads to a too-weak OH* binding, and formation of OH* becomes the rate-determining step. However, for the (101̅4) surface, the formation energy of OH* has an optimal value of 1.23 eV.[67]

There is a debate on whether Ni or Fe is the active site in NiFe LDH, and most work supports that Fe is the active site.[26m,48] The debate has been well reviewed by Kundu;[5d] therefore, we will not review the debate on the active sites of NiFe LDH. However, recently, Goddard performed extensive calculations using grand canonical quantum mechanics to have a better understanding of the synergy between Fe and Ni in (Ni,Fe)OOH, which sheds light on the debate. They suggested that both Fe and Ni are active sites; high-spin d4 Fe(IV) can stabilize the active O radical intermediate, whereas the low-spin Ni(IV) catalyzes the subsequent O–O coupling. Therefore, it is the synergy between Fe and Ni that leads to the optimal OER activity of (Ni,Fe)OOH. Furthermore, overpotential and Tafel slope were calculated to be 420 mV (Figure ) and 23 mV dec–1, in good agreement with experiments, 300–400 mV and 30 mV dec–1, respectively.[12b]

Figure 10

Calculated OER mechanism catalyzed by NiFe LDH by Goddard. Reproduced with permission from ref (12b). Copyright (2018) National Academy of Sciences.

Having established the idea that Fe could stabilize the O radical thus facilitating OER, Goddard further performed DFT calculations to substitute Fe with other transition metals (groups 3–9). And it was found that Co-, Rh-, and Ir-doped NiOOH could better stabilize O radical and exhibited even lower overpotentials, 270, 150, and 20 mV, respectively.[68] Although NiCo LDH was proposed to have a better OER activity than NiFe LDH by Goddard, it actually showed a poorer performance than NiFe LDH, which probably is due to the smaller stabilization ability of Co with the formed oxy (O*) group because of the larger number of electrons in the antibonding orbitals of metal and oxygen π bonds, as shown in Figure . It can be expected that lower number of electrons in the d orbitals could lead to stronger metal–oxy bond; however, if the metal–oxy bond is too strong, it is also unfavorable for OER, and that could be the reason why NiMn LDH and NiTi LDH exhibited poor OER activities.

Figure 11

Molecular orbital analysis of Fe=O and Co=O containing complexes.

As Ni4+ is responsible for catalyzing O–O coupling due to its strong oxidizing ability, one can expect that if the oxidation ability of metal is decreased, the ability of metal to catalyze the O–O coupling should also decrease. In the first-row transition metal, oxidation ability of metal ions with the same positive charge would decrease because of its lower electronegativity. Indeed, changing Ni to Co, we can see a decrease of OER activity, and that is why CoFe LDH exhibited a lower OER activity than NiFe LDH.

Summary and Outlook

In this perspective, we summarized the recent developments on OER catalyzed by LDH. The basic criteria to evaluate the performance of OER catalyst have been presented. Unary metal-based LDH has been discussed to have a deep understanding of the intrinsic activity of LDH. Basically, unary metal-based LDH is not effective for OER due to its poor conductivity and absence of synergistic effect. On the other hand, binary metal-based LDH is much more OER active. In particular, NiFe LDH is the most effective binary metal-based LDH toward OER. It outperformed other Ni-based binary metal LDHs owing to its suitable M–OH bond strength, which is neither too strong nor too weak. And it exhibited better OER activity than non-Ni-based LDH because Ni4+ has a strong oxidation ability, which can facilitate the formation of O–O bond. Finally, ternary metal-based LDH shows better OER activity owing to its increased conductivity. Moreover, introducing divalent defects leads to exposure of the Ni–O–Fe unit, which would greatly increase the OER activity of LDH.

Although LDH as a promising OER catalyst has been well studied over the past few years, some critical issues should be tackled to realize the practical application of these catalysts on hydrogen production via electrochemical water splitting.

It is known that LDH with a high OER activity almost always has a very high conductivity, but the underlying mechanisms are still not understood. Moreover, it has been noted that the conductivity of LDH depends on the applied potential.[18] Therefore, the influence of conductivity on the OER activity of LDH catalysts should be addressed systematically, especially the intrinsic conductivity of LDH, the contact resistivity between LDH and substrate, and the applied potential-dependent conductivity.

The synergistic effect between transition metals should be studied further to obtain the fundamental details, instead of simply going by the observation that the LDHs with more different transition metals are more OER active. Nowadays, the synergistic effect between Ni and Fe in NiFe LDH seems to be clearer, but the synergistic effect between other binary transition-metal-based LDHs and ternary transition-metal-based LDHs remains to be elucidated.

Close interplay between theoretical and experimental studies should be encouraged to tackle not only the thermodynamics but also the kinetics and dynamics of the intermediate steps of OER catalysis. For example, the key transition states leading to the formation of OOH* and O* should be located to have a better understanding of the kinetics of OER catalysis. For most LDHs, transition metals will be peroxided before OER takes place. Thus, preoxidations should be taken into considerations in future studies to have a better understanding of the active phase of LDH, which is usually ignored in the current practice.

Most OER experiments are performed in highly basic condition (pH > 13), but the OER performed in acidic and neutral conditions with more practical utilization is less reported and thus should be studied further in the future.