Density Functional Theory Study of Nitrogen Reduction to Ammonia on Bilayer Borophene

Abstract

The N₂ reduction reaction (NRR) under ambient conditions is highly desirable because of its potential to replace the energy-consuming Haber-Bosch process for ammonia production. In recent years, much attention has been devoted to transition metal-based catalysts. However, the development of metal-free electrocatalysts remains a great challenge. Here, the electrocatalytic performance of bilayer borophene is systematically studied based on first-principles calculations. It was found that bilayer borophene has high activity with an overpotential of 0.21 V via the enzymatic mechanism. Bond elongations of nitrogen bond are observed in end-on and side-on patterns, where the bond lengths are stretched to 1.13 and 1.21 Å, respectively. Around 0.36 e⁻ is transferred to the adsorbed N₂ with the contribution of bottom boron atoms. Our results propose bilayer borophene as a novel metal-free catalyst for nitrogen reduction, thus providing an avenue to explore highly efficient electrocatalysts for ammonia production under ambient conditions.

1. Introduction

The nitrogen, whose cycle consists of one essential part of the ecosystem, is important to the components of living creatures such as animals and plants. Around 78% of atmospheric composition is nitrogen gas, which is abundant and may be utilized as nitrogen resource. Ammonia plays an important role in the fertilizer production and food yield, where the main process (Haber–Bosch) consumes a lot of energy and produces around 1.6% of annual CO₂ emissions [1,2,3,4,5]. During this process, the nitrogen reduction to ammonia experiences six protonations based on the nitrogen and hydrogen gases (N₂ + 6H⁺ + 6e⁻ → 2NH₃), which relies heavily on the performance of the catalysts. A good catalyst should possess high activity, selectivity and stability, which should boost the production rates. Thus, it is of urgency to explore the efficient catalyst for ammonia production.

Electrochemical nitrogen reduction has been the center of attention due to its ambient conditions as well as environmental friendliness, which can produce value-added ammonia with low cost [6,7,8]. Due to the inert bonds of nitrogen with a high dissociation energy of 941 KJ mol⁻¹, the ammonia production rate is far from satisfactory, which has motivated many researchers to enhance the performance of the electrocatalysts [9,10,11,12,13]. Noble metals (such as Ru, Au, Pt and Pd) and their composites were widely studied due to their promoting effects on NRR process [14,15,16,17,18,19]. Liu et al. have synthesized a high-performance Ru@Ti₃C₂ MXene catalyst, whose Faraday efficiency was 13.33% with an ammonia production yield of 2.3µ mol h⁻¹ cm⁻² under ambient conditions [20]. A well-defined system with Au sub-nanoclusters dispersed were used to achieve a Faradaic efficiency of 8.11% [21]. Ran et al. have utilized isolated Pt atoms for the reduction of N₂, where the formed Pt-3O structure can serve as active sites for nitrogen chemisorption and activation [17]. Apart from the precious noble metals, the alternative transition metal (TM) counterparts were also screened out for better NRR performance [22,23,24]. Zhang et al. have developed a VN nanowire array with a high ammonia yield [25]. The transition metal based electrocatalysts were also proven effective for ammonia synthesis in the NRR process [7,26,27,28]. However, these metal-based electrocatalysts may possess some disadvantages such as high cost, poor stability and toxicity.

The metal-free electrocatalysts have attracted much attention because of their high activity, good conductivity, non-toxicity, abundance and durability [29,30]. Many metal-free candidates were screened out for potential use in NRR process, including carbon-based [31,32,33], phosphorus-based [34,35] and boron-based catalysts [36,37,38]. Among these well-known candidates, boron-based catalysts have been proven effective in the nitrogen fixation due to the “acceptance–donation” mechanism as observed in the TM electrocatalysts [39,40,41]. The empty d orbital of transition metals could accept the lone pair electron from the adsorbed N₂, while the partly filled d orbital could offer additional electrons to the antibonding orbital of N₂, thus activating the adsorbed species. In a similar way, boron atoms with three valence electrons and one empty orbital could capture and activate the adsorbed N₂, thus enhancing the activity of the catalysts [36,42]. Theoretically, various forms of boron-based catalysts, from single atom catalyst to boron sheets, were predicted to exhibit high activity towards nitrogen reduction to ammonia [43,44]. And a lot of boron-based electrocatalysts were proven effective for ammonia production [45,46,47,48]. Thus, boron element is one of the promising electrocatalytic candidates for ammonia production under ambient conditions.

Since the discovery of novel graphene material, few-layer 2D materials have raised tremendous enthusiasm due to their unique geometry, large specific area and tuneable electronic structures, which possess great potential in various application fields such as superconductors, sensors and batteries [49,50,51]. As a counterpart of graphene, 2D boron materials also emerged with intriguing semimetallic or metallic properties [52,53]. However, 2D boron nanosheets are thermodynamically unfavorable compared to that of its counterparts (e.g., MoS₂, graphene and g-C₃N₄), where the continuous deposition of boron atoms on monolayer tends to form clusters.

In recent years, 2D bilayer borophene was synthesized experimentally through interlayer covalent bonding [54], which may be applied in many areas such as electronics, photonic devices and batteries. Inspired by this work, the catalytic behavior of bilayer borophene in NRR process is extensively investigated. From a thermodynamical perspective, the possible reduction pathways are studied. Adsorption behavior and charge density are analyzed to evaluate the interaction between N₂ and the reactive site. The calculation results exhibit the potential of bilayer borophene as an efficient NRR catalyst with a low overpotential of 0.21 V. Therefore, our simulation reveals the excellent NRR performance of bilayer borophene, thus providing a new type of 2D material for ammonia production under mild conditions.

2. Results and Discussion

Due to the high cost and emission challenges of the Haber–Bosch process, the search for high-performance catalysts is increasing. Multiple transition metals have been found effective in activating N₂ molecules because of their partially filled orbitals, as shown in Figure 1a.

The empty d orbitals of TM could accept the lone pair electrons from the N₂ molecules, while the electrons in the occupied orbitals could be transferred to the empty π* orbital of N₂. The “acceptance–donation” mechanism can enhance the bonding between N₂ and TM catalysts as well as weakening the inert bonds of adsorbed N₂, thus facilitating the subsequent elementary steps. The electron-deficient boron also possesses the partially occupied orbitals in the metal-free elements. As shown in Figure 1b, the single boron atom has one fully occupied 2s orbital and one half-filled 2p orbital, while the remaining two orbitals are empty. After the sp³ hybridization process, three valence electrons are distributed in three sp³ orbitals, leaving one orbital empty. The newly formed orbitals may have the potential to activate the adsorbed N₂ molecules through “acceptance–donation” mechanism. In Figure 1c, when two half-filled sp³ orbitals interact with suitable substrates, the empty sp³ orbital can obtain the electrons while the occupied one can donate an electron to the antibonding π* orbital of N₂ molecules. The structural configuration of boron sheet is shown in Figure 1d. The four-coordinated boron atoms form interlayer bonds with a length of 1.88 Å, which is in agreement with the previous report [55].

The nitrogen reduction to ammonia is a complicated process with multiple intermediates through different pathways. To evaluate the potential performance of catalysts, it is important to screen the thermodynamically favorable pathways by identifying the potential-determining step. For NRR process, the PDS usually lies in the hydrogenation of adsorbed N₂ and conversion of *NH₂ to *NH₃, where * denotes the adsorption site. The possible NRR pathways are classified as distal, alternating and enzymatic pathways as depicted in Figure 2. In the distal and alternating pathways, N₂ adsorbs on the reactive site in an end-on manner. For the distal pathway, the N atom, farthest from the reactive site, would be continuously attacked by the proton-electron pairs until the first NH₃ release. The other N atom would form the second NH₃ molecule through step-by-step hydrogenation. For the alternating pathway, the two N atoms would be alternatively attacked until the last step, where two NH₃ molecules are released. Similar to the alternating pathway, the enzymatic one also involves alternating hydrogenation steps with the final desorption of two NH₃ molecules. However, the difference lies in the initial adsorption configurations of N₂, where both N atoms are adsorbed on the surface in a side-on manner.

The Gibbs free energy variation of intermediates can be utilized to assess the catalytic performance of the bilayer borophene for the NRR process, which can be complicated due to various intermediates through different pathways. Three possible pathways, namely distal, alternating and enzymatic pathways, are illustrated in Figure 3, where the relative Gibbs free energy change is revealed. It can be seen that the side-on adsorption configuration is more favorable than that of end-on configurations. For the distal pathway, the PDS lies in the first hydrogenation of *NN intermediate, where the Gibbs free energy increases by 1.24 eV. However, the second hydrogenation on the same nitrogen atom is an exothermic process with a free energy change of −0.47 eV. The generation of *N intermediate is also spontaneous with an energy drop of 1.88 eV. Similarly, a slight energy decrease is observed (0.11 eV) in the conversion of *N to *NH. However, the further protonation of *NH is endothermic, requiring around 0.01 eV. The formation of *NH₃ and final product is spontaneous, with a downhill Gibbs free energy change for the easy desorption of NH₃. The alternating pathway is similar to the distal one, both of which share the same PDS with the transformation of *NN to *NNH intermediate. The generation of *NHNH, *NHNH₂ and *NH₂NH₂ is spontaneous, where the Gibbs free energy changes are −0.53 eV, −0.16 eV and −1.77 eV, respectively. The negligible energy increase (0.02 eV) for the conversion of *NH₂NH₂ to *NH₂ is observed, during which the first NH₃ molecule is released. The final two steps for the desorption of the second NH₃ molecule are also thermodynamically favorable.

The mentioned distal and alternating pathways are hard to proceed due to the large overpotential of 1.08 V. However, the enzymatic pathway seems more promising despite the same PDS occurring in the first hydrogenation step. It is worth noting that N₂ tends to adsorb on the surface vertically, while the *NNH intermediate would be stable in a horizontal way with two nitrogen atoms adsorbed on the surface. For the enzymatic pathway, the energy change in the PDS is around 0.37 eV, whose overpotential (0.21 V) is much less than that of distal and alternating pathways. Besides, the subsequent hydrogenation steps are thermodynamically favorable except that an energy increase of 0.28 eV is observed in the formation of the *NH₃ intermediate. The solvation effect was also studied, which shows that the catalytic performance of the electrocatalysts remains strong. It should be noted that the study was conducted considering a low-coverage regime to reduce the complexity of the simulations and computational cost. Besides, the free energy is solely considered to evaluate catalyst performance, because it is widely reported that the Brønsted–Evans–Polanyi (BEP) relationship exists between the calculated activation energy and reaction energy [56], which is also assumed here to reduce the computation cost.

To gain further insight into the adsorption behavior and activity of the bilayer borophene system, the adsorption pattern, bond length change, electronic structures and charge transfer are analyzed as shown in Figure 4. As illustrated in Figure 4a, the adsorption of N₂ on the catalytical surface can occur through end-on or side-on configurations, which is a prerequisite for the subsequent hydrogenation process. The N₂ chemisorption is accompanied by bond length change, which is an important indicator for the activation of the triple bond in N₂. As shown in Figure 4b, different bond variations are observed for end-on and side-on patterns, where the bond stretches to 1.13 and 1.21 Å, respectively, compared to the N₂ molecule (1.11 Å). The bond elongation indicates that the adsorbed intermediate is activated for further hydrogenation. The charge density difference reveals the charge transfer and interaction when N₂ is adsorbed on the B atoms. According to the “acceptance and donation” mechanism, the B atoms can accept lone pair electrons from N₂ and donate electrons to the antibonding orbitals of N₂, enhancing the interaction between adsorbed N₂ and the bilayer borophene catalyst and triggering the NRR process. The charge transfer for end-on adsorption is shown in Figure 4c, where the accumulated charge density between N₂ and B atoms indicates that electrons are transferred from the nitrogen, while the increased charge density on the two nitrogen atoms suggests that excess electrons are donated into the antibonding orbitals of N₂. Bader charge analysis shows that around 0.36 e⁻ is transferred from the bilayer borophene. It should be noted that the bottom layer also contributes to the accumulated charge density through covalent bonding, which can modulate the surface electronic structure and improve electrocatalytic performance. As shown in Figure 4d, the external potentials ranging from −0.5 V to 0.5 V can have a dramatic effect on the adsorption behavior of reactant molecules, where a decreased potential increases N₂ adsorption in both horizontal and enzymatic configurations. It can be expected that lower potentials (<0.5 V) could enhance N₂ adsorption for further reduction.

In addition, the selectivity of the catalysts is also important for assessing performance. A good catalyst should produce the desired products without many by-products. Thus, we evaluate the HER performance on the bilayer borophene. As shown in Figure 5, the adsorption energies of protons are far from zero, indicating poor HER performance. Therefore, the bilayer borophene also possesses high selectivity.

3. Computational Details

All density functional theory (DFT) calculations were carried out using the Vienna ab initio simulation package VASP (version 5.4) [57,58]. The projector augmented wave (PAW) method was used to describe the electron-ion interactions [59]. The exchange-correlation potential was described by the adoption of Perdew–Burke–Ernzerhof (PBE) functional approximation [60,61]. A cutoff energy of 400 eV was employed for the plane-wave basis set. The slab was constructed to represent the catalytic surface, totaling 182 atoms. More than 15 Å of vacuum was added along the z direction to reduce interactions between periodic images. Monkhorst–Pack k-point mesh was employed over the Brillouin zone for all calculations. The sheet model was fully relaxed until the residual energy and force were below 10⁻⁵ eV and 0.01 eV Å⁻¹, respectively. Grimme’s DFT-D3 method was introduced to consider the van der Waals correction during all calculations [62]. The Gibbs free energy change in every elementary step was calculated based on the computational hydrogen electrode (CHE) model. The Gibbs free energy change is obtained as follows:

$$∆G=∆E+{∆E}_{ZPE}-T∆S+∆G_U+∆G_{PH}$$

(1)

where ∆E is the computed reaction energy change of two reaction elementary steps, ∆E_ZPE is the change of zero-point energy. T represents the temperature which is set to be 298.15 K. ∆S is the entropy change. Zero-point energy and entropy change were obtained by VASPKIT package from the vibrational frequency analysis, where the entropy change is assumed to be zero in consideration of absent translational and rotational motions on intermediate configurations [63]. For the gas molecules, the corresponding zero-point energy and entropy were calculated based on NIST-JANAF thermochemical tables [64,65]. ∆G_U is the free energy resulting from the applied potential. ∆G_pH is the pH correction and can be calculated by the following equation,

$$∆G_{PH}=T{k}_B\times pH\times In10$$

(2)

It should be noted that the pH influence is set to be zero since this factor does not affect the simulation results. The onset potential is determined by the most positive free energy change, which is defined as the potential-determining step (PDS). The overpotential η is obtained by

$$\eta =\left|U_{limiting}-U_{equilibrium}\right|$$

(3)

where U_equilibrium is the equilibrium potential for the N₂ reduction process (−0.16 V). The limiting potential (U_L) can be obtained through the equation U_L = −ΔG_max/e, where ΔG_max is the maximum free energy change screened out from all steps. The Bader charge program was utilized to analyze the charge transfer between the adsorbent and electrocatalyst [66,67,68,69]. The solvation effect was studied using an implicit solvation model [70] with a dielectric constant of 78.4 as employed in other works [71]. The geometries were fully relaxed in the presence of solvent.

To consider the potential effects, the chemical-potential-dependent thermodynamic study was conducted within the framework of JDFTx [72]. The grand free energy was obtained with the constant-potential method (CPM) ranging from −0.5 V to 0.5 V. The linear polarizable continuum model (PCM) was adopted, where the charge-asymmetry-corrected, local-response, nonlocal-cavity solvation model (CANDLE) was utilized to simulate the aqueous environment composed of 1 M KF [73]. All calculations were carried out with plane-wave cutoff and charge density cutoff of 20 Ha and 100 Ha, respectively [74]. The PBEsol pseudopotentials were used with the default exchange functional.

4. Conclusions

In summary, based on first-principles calculations, we investigated the electrocatalytic performance of bilayer borophene for the NRR process. Three pathways, including distal, alternating, and enzymatic pathways, were identified and evaluated with varied hydrogenation steps. The simulations show that the NRR process through distal and alternating pathways is not thermodynamically favorable due to the high overpotential of 1.08 V, while a small overpotential of 0.21 V is required through the enzymatic mechanism. Bond elongation of the adsorbed *NN intermediate is an indicator of nitrogen activation. The charge density and Bader charge transfer verify the “acceptance–donation” mechanism, where the bottom B layer acts as an electron reservoir to donate excess electrons for improved activity. Our theoretical analysis provides preliminary evidence that bilayer borophene could serve as an efficient NRR catalyst under specific electrochemical conditions, potentially enabling a cost-effective process for ammonia production.