Unlocking Synergistic Catalysis in NiP: Dual Role of Electronic Structure and Lewis Acidity for Enhanced Oxygen Evolution Reaction

1. Introduction

The global surge in energy demand, coupled with the depletion of fossil fuel reserves and their severe environmental impact, has intensified the urgency for researchers to explore green energy alternatives and develop more efficient sustainable energy conversion technologies [1,2,3]. Hydrogen is widely recognized as a clean and renewable energy carrier due to its high energy density and zero carbon emissions. The most promising hydrogen production method thus far is water electrocatalytic decomposition [4,5,6,7]. In the water-splitting reaction, the oxygen evolution reaction (OER) is a critical bottleneck, as it entails a complex four-electron process that is often regarded as the rate-limiting step of the entire catalytic cycle. Consequently, the development of efficient OER electrocatalysts is of paramount importance [8,9,10].

Precious metals and their oxides, particularly Ru, Ir, RuO₂, and IrO₂, demonstrate superior OER electrocatalytic performance. However, their widespread adoption is severely limited by their prohibitive cost; the development of cost-effective yet highly efficient catalysts is crucial for commercial applications and large-scale deployment [11,12,13]. In recent years, earth-abundant transition-metal catalysts have garnered significant attention for their potential in electrocatalytic water splitting. Nickel phosphides, a representative transition-metal phosphide, have emerged as a notable contender. The favorable combination of their low preparation costs and high catalytic activity renders them promising candidates for OER catalysis [14,15,16]. It has been reported that the control of phosphorus content plays a crucial role in the OER activities of Ni_xP_y. It was proposed that the removal of phosphate increased the number of active sites for oxygen evolution in NiP films, thereby enhancing the OER performance[17]. Notably, NiP with a relatively high metal content exhibits superior activity towards OER[18,19,20,21,22,23]. Phosphorus within NiP functions as an excellent supporting layer beneath the surface of nickel oxide or oxyhydroxide during the OER, enhancing the stability of the catalyst [18,24]. In addition, under actual reaction conditions, solid surfaces undergo geometric, atomic, and electronic reconstructions to minimize surface free energy. The reconstructed atomic surface stoichiometry may either mirror the bulk termination stoichiometry or differ from it [25,26,27]. Under the actual electrochemical conditions, nickel phosphides have been proved to undergo various surface reconstructions, which significantly alter their catalytic performance compared to the bulk-terminated surfaces [28,29]. Exploring the OER mechanism of NiP, with stoichiometric ratio of 1:1, will augment our understanding of the relative and cooperative effects of phosphorus and nickel in Ni_xP_y catalysts, leading to optimization criteria for future nickel phosphide catalyst design. Therefore, it is essential to systematically investigate the NiP surface reconstructions and thoroughly explore the OER mechanism. Notably, transition-metal phosphides, including NiP, offer abundant metal and nonmetal adsorption sites for OER intermediates, introducing additional complexity beyond traditional metal-based systems. Most researchers believe that conventional descriptors, such as the d-band model related features, prove insufficient in providing comprehensive insights into these complex systems [30,31,32]. Therefore, a systematic analysis of the OER reaction mechanism at different active sites is crucial for the rational design of efficient OER catalysts.

In this work, we investigate the surface reconstruction and OER activity of NiP(100), (010), and (001) using first-principles density functional theory (DFT). We identified the most stable surfaces among the various surface reconstructions of those low index surfaces. During the investigation of OER activity, we found that the charges at the adsorption sites are linearly correlated with the oxygen adsorption intermediates and even the OER overpotential. The results show that the charge transfer at the site can regulate the adsorption energy of intermediate and impact the performance of the catalysts. We hope that our results can provide a reference for the design of base-metal catalysts.

2. Computational Details

For the DFT calculations, we use the Quantum ESPRESSO software package with a plane-wave basis set [33,34]. The exchange-correlation energy was calculated using the Perdew-Burke-Ernzerhof (PBE) formulation of the generalized gradient approximation (GGA) [35]. Non-local norm-conserving pseudopotentials constructed with the OPIUM code are used [36,37]. The van der Waals interaction (vdW) was approximated using the DFT-D3 method [38]. During NiP slab and the adsorption structure optimization, only the top half of the atoms are relaxed, keeping the lower half fixed. The convergence accuracy of the total energy of the system is 10^-5 eV, and the convergence accuracy of the forces is set to 10^-4 eV/Å. The k-point sampling in the first Brillouin zone for the supercells is shown in Table S1.

The NiP cell structure in the international crystal structure database (ICSD; Collection Code 27159, space group Pbca) is selected as the bulk structure for NiP calculation [39]. The lattice parameters of NiP completely relaxed are a = 6.103 Å, b = 4.929 Å, and c = 6.905 Å, which are within 1% of the experimental values (a = 6.050 Å, b = 4.881 Å, and c = 6.890 Å). We construct a slab model to simulate the surface structures. The (2×1) supercell with two formula units of NiP along the z-direction ensures the decoupling of the continuous plates. Adsorption was allowed on one side of the slab, and in order to ensure the decoupling of the consecutive slabs, a 15 Å thick vacuum region is employed, with an artificial dipole correction inserted in the center of the vacuum layer [40]. This work studies three surface directions of NiP: (001), (010), and (100) and uses a fully relaxed unit cell to construct the supercell structure to construct the slab model. Under actual electrochemical conditions, NiP undergoes various surface reconstructions, which affect the catalytic activity of the reaction. Interface tailoring can effectively improve the activity and stability of the catalyst [41,42,43]. Judging by the pristine surface slab geometry directly generated from the bulk, the slab layer geometry repeats itself within two layers. Therefore, our search for low-energy reconstructions starts with one or the other layer of the original crystal structure to ensure that the search covers the possible reconstructions (see the Supporting Information (SI) for more details).

The catalytic activity of OER catalysts depends on the Gibbs free energies of intermediates (∆G_HO*, ∆G_O*, and ∆G_HOO*) [44].For an ideal OER cycle, the Gibbs free energy change between two adjacent intermediates should be 1.23 eV (U = 0 V). By calculating the free energy of each OER reaction step, we can use the calculated OER overpotential to evaluate the catalyst OER performance. More computational details are provided in the Supporting Information.

3. Results and Discussions

3.1. Stability of NiP Surfaces

To identify the most stable surface, we computed the surface energies of various NiP slab models with different surface terminations. Here, we have studied the surface stability of (100)-Pris:P₂, (100)-C1:Ni₄P₂, (100)-C2:Ni₂P₂, (001)-Pris:Ni₂P₂, (001)-C1:Ni₂P₂, (010)-Pris:Ni₂P₂, (010)-C1:P₄, (010)-C2:P₂, where the symbol is defined as (facet)-Reconstructions: termination(Pris represents the pristine bulk structure termination, C1 represents the cutting of one layer of the top surface atoms, C2 represents the cutting of two layers of the top surface atoms).

Based on our calculations, we have found that P-rich surface terminations are the most energetically favorable surface terminations for NiP among the low-index surfaces. The surface energy diagram of the slabs of all three lattice planes showed that the surfaces with P atom terminations have low surface energies indicating that they are thermodynamically stable. Before the formation of bulk NiP region (−1.2 eV < ∆µ_P < 0 eV), four stable surface terminations with lower surface energy ((010)-C1:P₄, (010)-C2:P₂, (001)-C1:Ni₂P₂, (100)-C1:Ni₄P₂) are shown in Figure 1b. Among these surfaces, NiP(010) exhibits the lowest surface energy, indicating the most stable surface termination.

3.2. Stability of the Adsorption Structures

To investigate the relationship between the catalytic activity and surface active sites, we constructed multiple adsorption systems of the OER intermediates on the various adsorption sites of the surfaces. Judging by the surface density of states (DOS), we found that many surface sites are similar in terms of geometric position and electronic structure. Taking (100)-C1:Ni₄P₂ as an example, the surface layer consists of four Ni atoms and two P atoms (Figure 1c). Ni¹ and Ni³ are completely equivalent due to the full overlap of the DOS (see Figure 1d). The same applies to Ni² and Ni⁴, as well as P¹ and P². For the NiP(001)-C1:Ni₂P₂ surface, both the two Ni and P sites are equivalent, leading to one set of Ni-P sites. For (010)-C1:P₄ and (010)-C2:P₂ surfaces, the topmost layer only contains P atoms, and they are all equivalent within the surfaces correspondingly (Figure S5).

Given the equivalency of the surface sites, we considered one monolayer (1 ML) adsorption condition of the OER cycle species. Due to the simple structure of the adsorbates, such as O*, HO*, and HOO*, we chose the on-top adsorption site for each case. To determine stable adsorption structures, we tested four adsorption distances: 1.3 Å, 1.5 Å, 2 Å, and 2.5 Å. Our findings reveal that the diverse adsorption sites on the surfaces create a complicated atomic interaction for adsorbed OER intermediates. Notably, the available adsorption sites encompass Ni₂P-hollow, Ni on-top, P on-top, as well as the bridge sites between Ni-Ni and Ni-P atoms (see the adsorption energy (E_ads) for various sites in Table S2). The stable oxygen intermediate adsorption sites are Ni₂P-hollow, Ni on-top, P on-top, and the Ni-Ni and Ni-P bridge sites. For the HO* group, the stable sites are P on-top and the Ni-Ni bridge sites. Similar to the O* group, the HOO* can be stably adsorbed on Ni on-top, P on-top, and the Ni-Ni bridge sites.

3.3. OER Pathways on NiP Surfaces

To understand the mechanism of OER on the NiP surface, we investigated the OER process on the four stable NiP surfaces as determined in the previous section (Figure 1b). We found that on the four stable surfaces, there are two reaction pathways for the OER, which are described as the single-site pathway and the dual-site pathway. The single-site refers to all the intermediate species adsorbing on the same site, whereas the dual-site pathway introduces a second adsorption site of the oxygen intermediate during the reaction cycle (Figure 2). The single-site pathway exists on all four stable surfaces, whereas the dual-site reaction can happen only on Ni₄P₂ and P₂ surfaces. For each intermediate, the preferred adsorption location can be determined by comparing the adsorption energy at various adsorption sites [45]. For Ni₄P₂ surface, the preferred adsorption site for HO* and HOO* is the Ni-Ni bridge. The energetically favored adsorption site for O* is Ni₂P-hollow instead of Ni-Ni bridge, which converts the pathway from single-site to dual-site (Figure 2a). The adsorption energy difference between O* adsorption on Ni-Ni bridge and Ni₂P-hollow is 1.84 eV, according to our calculation. The dual-site pathway usually involves sublayer atoms and includes both P and Ni atoms. For the P₂ surface, both HO* and HOO* are stably adsorbed on the P atom top site. O* is preferentially adsorbed on a Ni-P bridge site forming another dual-site pathway with the adsorption energy 0.42 eV lower than the O* adsorbed on P atom (Figure 2b). For the P₄ and the Ni₂P₂ termination, all the OER intermediates are adsorbed more strongly on the on-top P site than other sites, which can be described as a single-site pathway (Figure S6). We subsequently evaluated the OER catalytic activity of various sites based on the established reaction paths.

3.4. Investigation of Electrocatalytic OER Activities

To investigate the relationship between the nature of surface active sites and catalytic activity, we calculated the Gibbs free energy of the OER intermediates at U = 0 V and U = 1.23 V according to the proposed reaction pathways (Figure 3a–c). One important parameter used to evaluate catalytic activity is the OER overpotential, which is determined by the free energy of the rate-determining step (RDS). For the single-site OER pathway of P₄ and Ni₂P₂ surfaces, the RDS is O* oxidation to HOO* with overpotentials as large as 1.54 V and 1.57 V (U = 0 V) (Figure 3c). Moreover, the adsorption energies of the O* on P₄ and Ni₂P₂ surfaces are very similar (Figure 3d), which leads to the overpotential being pretty similar to each other. For the P₂ surface, the RDS is the oxidation of O* to HOO* for both the single-site and dual-site reaction pathways (Figure 3b). Notably, the overpotential for the single-site pathway is 1.68 V, whereas the overpotential is significantly lower at 1.07 V for the dual-site pathway. Therefore, the OER prefers the dual-site pathway on the P₂ surface. Furthermore, the involvement of the Ni atom enhances the adsorption of O* on the P₂ surface (E_ads(O*)=−4.56eV) compared to the single-site pathway with the P atom-only interaction. However, the joint interaction between Ni and P can cause too strong adsorption of the O*, hindering the OER cycle. In the case of the Ni₄P₂ dual-site pathway, O* adsorbs on the Ni₂P-hollow site with a strong adsorption energy of -5.16 eV and an overpotential of 1.30 V, hindering the participation of O* in subsequent reactions. By contrast, Ni₄P₂ with the single-site pathway shows the lowest overpotential of 0.78 V amongst all the surfaces, featuring the oxidation of HOO* to O₂ as the RDS (Figure 3a) rather than the more common RDS of O* to HOO*. The metal atom site (Ni-Ni bridge site) provides moderate adsorption to the intermediates and activates the reaction cycle (Figure 3d). According to the Sabatier principle, the adsorption energy of the intermediate needs to fall within an appropriate range to ensure the equilibrium between the adsorption and desorption of the intermediate [46]. Given that O* exhibits a strong adsorption affinity to the Ni₂P-hollow site, more energy is required to desorb O*, resulting in a larger overpotential and shifting the RDS to the oxidation of O* to HOO*; therefore, it hinders subsequent reactions. Consequently, for the Ni₄P₂ surface, the OER prefers a single-site pathway (Figure 2a) rather than a dual-site one.

While energetic analysis is helpful for explaining the OER catalytic activity at varies active sites, we present an analysis grounded in electronic structure theory to elucidate the relationship between the catalytic activity and the adsorption site characteristics.

Focusing on the RDS investigations, we found that the interaction between O* and the sites encompassing Ni and P atoms is usually stronger than in other cases. In the case of P₂ surface, the differential charge density depicts that O* adsorption reveals a significant redistribution of charge between O* and various surface sites (Figure 4a–d). The adsorbed O* induces electron accumulation at both the Ni-P bridge site and the P site. Due to the high electronegativity of oxygen, O* tends to obtain electrons from Ni and P atoms within the adsorption sites. In the single-site pathway, O* adsorbs on the P site and obtains electrons from the P atoms in the adsorption site. In the dual-site pathway, O* interacts with the Ni-P bridge site, and the electronic interaction between the O* and the P atoms on Ni-P bridge site is stronger than that on the P site. Meanwhile, O* exhibits a larger charge accumulation at Ni-P bridge site compared to that at P site. All these phenomena can be attributed to the additional electron donation from the Ni atom, which alters the local charge distribution and enhances the interaction between O* and P atoms. The electronic interaction of O* on the Ni₄P₂ surface is similar to that on the P₂ surface. In the single-site pathway, strong charge accumulation and depletion can be observed between O* and Ni atoms at the Ni−Ni bridge site. In the dual-site pathway, O* acquires electrons from both the Ni atoms and the P atom residing within the Ni₂P-hollow site. Notably, on the Ni₂P-hollow site, a relatively stronger electron dissipation occurs between O* and P atoms as compared to that between Ni atoms. The strong electronic interaction between O* and P engenders a more negative adsorption energy of O* at the Ni₂P site (E_ads(O*)=−5.16eV) in comparison to that of Ni-Ni bridge site composed solely of Ni atoms (E_ads(O*)=−3.32eV), thereby conferring enhanced adsorption stability.

Thoroughly analyzing the density of states (DOS) of the various adsorption structures among different terminated surfaces, we found that P atoms tend to assist in stabilizing O* at the adsorption sites. When Ni and P combine at the site, such a P involvement can further activate the Ni sites in the catalytic cycle, driving Ni as an electron donor. The interaction of O* at P sites of various surfaces presents a similar behavior in that the major oxygen p orbital peak lies around -5 eV and overlaps with the phosphorus p orbital, indicating a strong and stable chemical bond formation between O* and P. The non-bonding oxygen p orbital peak can be found around -2.5 eV with an almost half-full occupation, making the O* hard to propagate the further reaction (Figure 5b and Figure 5d). This phenomenon is attributed to relatively low adsorption energy and high overpotential compared to other cases. When O* adsorbs on the Ni-Ni bridge site found on the Ni₄P₂ surface (Figure 5a), the oxygen p orbital shifts to the higher energy region in general compared to O*-P case (Figure 5b), and the previous non-bonding part hybridizes with the Ni d orbital crossing the Fermi level, forming an active state to proceed the further reaction along the OER cycle. When a P atom presents within the adsorption site, as is the case for the Ni-P bridge site (Figure 5c), the major p contribution of the O* falls back to around -5 eV, and the reduced overlap between the oxygen p orbital and Ni d orbital suggests a weakened O*-Ni interaction. However, the d band center of Ni shifts higher to be close to the Fermi level, which activates the Ni atom in Ni-P bridge site for the next reaction step. In this case, the Ni atom provides its excess d electrons to O* (Figure 4a), leaving the unsaturated site for the opportunity of a second water to break to form HOO* (SI Eq. 3). Hence, the dual-site pathway is preferred on the P₂ surface.

Except for the Ni₄P₂ surface, the RDS of the OER reaction on P₄, P₂, and Ni₂P₂ consistently involves the oxidation of O* to HOO*. Therefore, we performed Bader charge analysis on adsorption sites and oxygen intermediates to investigate the charge transfer (Table S3). Previous studies have demonstrated that substantial charge transfer between active sites and adsorbates typically correlates with lower adsorption energy [47,48,49,50]. To investigate the potential mechanism of the RDS on various NiP surfaces, we calculated the adsorption energy of O* and the amount of charge transferred at the adsorption site. We observed a strong linear correlation between the amount of charge transferred at the adsorption site and the adsorption energy of O* (Figure 6a). Within the charge range of 1 to 2 |e| for the adsorption site, the adsorption energy of O* exhibits a decreasing trend with increasing charge, indicating that charge transfer occurring at the adsorption sites can effectively modulate the adsorption energy of O*, which provides insights for optimizing intermediate adsorption to regulate the performance of OER catalysts [51,52]. In addition, we calculated the bader charge of the sites before adsorption. We found that the charge of the adsorption site and the OER overpotential are linearly related. The overpotential decreases as the charge lost from the adsorption site increases (Figure 6b). We also observed that the adsorption sites containing Ni atoms tend to form a Lewis acid system (Figure 6b). Previous studies have shown that OER intermediates exhibit behaviors similar to Lewis bases with lone electron pairs [53,54,55]. When the adsorption site loses electronic charge, the electron cloud density around the site atoms decreases relatively, enhancing its electron-accepting ability, making the overall structure exhibit Lewis acid characteristics, which is conducive to the formation and adsorption of oxygen-containing intermediates, thereby accelerating the OER kinetics. Moreover, the charge of the adsorption sites can directly influence the performance of the NiP catalysts, indicating that even though NiP exhibits metallic properties, the local chemical environment can still determine the catalytic activity. In other words, the catalytic activity of bulk metallic surface material, such as NiP, can be optimized by tailoring the local chemical environment on the surface.

4. Conclusions

In summary, we systematically investigated the OER mechanism of NiP surfaces by DFT calculations. Our results show that the reconstructed NiP surface with various stoichiometric ratio of Ni and P have a significant impact on OER activity. The NiP surface offers numerous adsorption sites for OER intermediates. The existence of disparate adsorption sites for oxygen intermediates on the Ni₄P₂ and P₂ surfaces allows the establishment of two reaction pathways, namely the single-site pathway and the dual-site pathway. Differential charge density and DOS analysis reveal that the interaction between O* and the sites encompassing Ni and P atoms is usually stronger than in other cases. P atoms tend to assist in stabilizing O* at the adsorption sites. Such a P involvement can further activate the Ni sites in the catalytic cycle when right amount of Ni and P combined at the site to drive Ni as an electron donor. The additional electron donation from the Ni atom can alter the local charge distribution and enhance the interaction between O* and P atom. Moreover, our calculated results demonstrate that the adsorption energy of oxygen intermediate exhibits a linear correlation with the charge of the adsorption sites, and the OER overpotential also presents a linear correlation with the charge of the sites before adsorption. The charge transfer at the site can effectively regulate the adsorption energy of oxygen intermediate and directly impact the performance of NiP catalysts. When the site undergoes charge loss, its electron-accepting capability is enhanced. Consequently, the entire system demonstrates Lewis acid characteristics, making it favorable for the OER. This work elucidates the OER mechanism on the NiP surface, demonstrating that the catalytic activity of bulk metallic surface materials like NiP can be optimized by tailoring the local chemical environment on the surface, which provides a valuable framework for guiding base metal catalysts design.