The methods used for ammonia synthesis have significantly progressed during the last century. Haber-Bosh process [1, 2] proposed in the 20th century was a revolution in the synthetic ammonia industry. However, the limitations of the operating temperature of this process is overcome by the fast reaction rates and high equilibrium conversion [3]. Energy consumption and the adverse effects on equipment, caused by rigorous operating conditions, must be considered as well. In this context, electrocatalytic ammonia synthesis is an energy-saving and environmentally benign process, and is gaining increasing attention [4, 5]. The introduction of electrical energy effectively assists the activation of N₂ molecule and changes the reaction pathways. Development of efficient catalysts with good stability and high efficiency plays a very important role in reducing the applied voltage [6].

Electrocatalysts are categorized into several groups, such as those based on metals [7-13], metal oxides [14-16], polymers [17] and other hybrid materials [18, 19]. Recently, single atom catalysts [20-24] have been widely used in many electrocatalytic reactions. Li and coauthors [25] have proposed that a single transition metal (TM) atom supported on graphitic carbon nitride can result in a new class of low-cost, durable, and efficient oxygen evolution reaction catalysts. In addition, single TM atoms embedded into MoS₂ [26], carbon, and carbonitride material [27, 28] have been considered as promising catalysts in electrocatalytic ammonia synthesis.

The substrate is a key factor for improving the stability of these single-atom catalysts [20, 29]. Ideal substrates can distribute single atoms separately and evenly, and thus effectively prevent their aggregation. Pandey et al. [30] have successfully synthesized a new two-dimensional material (Ag/Ti₃C₂T_x MXene) that has strong interactions between the transition metal and MXenes. Thus, MXene can be a promising substrate. Based on density functional theory (DFT) calculations, Zhou et al. have reported a Ti single atom catalyst for CO oxidation [31] and CO₂ reduction [32]. In addition, Ag [33], Pt [34], and Pd [35]/MXene hybrids have been widely applied in electrochemistry. Recently, Wang's group [36] and Li's group [37] have successfully synthesized MXene-based for ammonia synthesis. However, there are a few systematic studies screening MXene-supported catalysts for electrocatalytic ammonia synthesis.

A systematic screening of TMs (Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Mo, Ru, Rh, Pd, Ag, Cd, and Au) has been carried out to understand the feasibility of MXene-supported single atom catalysts in ammonia synthesis and develop new electrocatalysts with low overpotential. The Gibbs free energy has been calculated to analyze the reaction paths and the overpotentials of Fe, Co, Ru, and Rh supported on MXene catalysts. The potential-determining step was found to be N₂ + H⁺ → NNH for the four TM/MXene catalysts. Further, we studied the relationship between the Gibbs free energy and overpotential for a series of TM-supported catalysts. The results reveal that different TM atoms have different overpotentials, ranging from 0.68 to 2.33 eV. There were two possible potential-limiting steps: (i) N₂ + H⁺ → NNH and (ii) NH₂ + H⁺ → NH₃. This work provides a new route for the rational designing and large-scale screening of these catalyst for electrocatalytic ammonia synthesis.

All the calculations were performed using the Vienna ab-initio Simulation Package (VASP), which is based on DFT [38, 39]. The exchange-correlation energy was calculated using the generalized gradient approximation and the Perdew, Burke, Ernzerhof [40, 41]. The effect of core electrons on the valence electron density was determined by The Projector Augmented Wave (PAW) method [42]. The cutoff energy for the plane wave basis sets was 400 eV. The convergence threshold of atomic position and cell parameters were set to be 10^-5 eV in energy and 10^-2 eV/Å in force. For geometric optimizations, the Brillouin zone [43] was sampled by 6×6×1 k-points mesh, while the same for electronic structures was sampled by 12×12×1 k-points mesh. The Van der Waals interaction was calculated by the DFT-D3 method [44]. Fully relaxed 3 × 3 supercells were used for all the structures. A vacuum space with thickness more than 20 Å was set to avoid the interactions of the periodic images. The charge density difference was analyzed using the VESTA code [45].

The Gibbs free energy calculation was performed following the work of Nrskov et al. [46]. Under the standard reaction condition, the chemical potential of a proton and electron pair (μ(H⁺+e^-)) is equal to half that of gaseous hydrogen (μ(H₂)). For each elemental step, the Gibbs free energies (△G) were calculated using the following equation:

Here, E is the total energy calculated by VASP, ZPE is the zero-point energy, H and S are the heat capacity and entropy, respectively, and T is equal to 298 K. G_pH and G_U are the contributions from the pH and electrode potential (U), respectively. G_U is defined as

where n is the number of electrons transferred. G_pH is defined as

where k_B is the Boltzmann constant. For all the calculations presented here, pH is set to zero.

The adsorption energies (E_ad) of the different adsorbates were calculated using the expression

where E_total is the total energy of the slab after adsorption, E_catal is the energy of the bare catalyst, and E_mol is the total energy of the adsorbate molecules, i.e., N₂ and N_xH_y.

A quintuple-layered conventional bare Ti₃C₂ MXene consisting of a carbon layer sandwiched between two Ti layers (Ti-C-Ti-C-Ti) was used. The lattice constants of the optimized structure are a = b = 3.110 Å, which are well consistent with the experimentally obtained value of 3.057 Å [47]. Because Ti₃C₂ MXenes are chemically exfoliated from the bulk MAX phase by HF [48, 49], they are always functionalized by the surface group O. Fig. 1(a) depicts the four different highly symmetric adsorption sites that have been studied. This includes the top site above the Ti atom (T), the bridge site of the Ti–Ti bond (B), and the two different hollow sites (H1 and H2). Calculations show that the hollow site with Ti center (Fig. 1(b)) is the most energetically favorable. Subsequently, the highly symmetrical single atom adsorption sites are considered to be the most active. All the adsorption energies are listed in Table S1 (Supporting Information). Here, we take Ru, Rh, Fe, and Co atoms as example, all of which are placed at the hollow site with C center, as shown in Fig. 1(c).

N₂ adsorption has been investigated on Fe/MXene, Co/MXene, Ru/MXene, and Rh/MXene. Both end-on and side-on modes of N₂ adsorption on the top of a single atom have been examined. Fig. 2 shows that all the N₂ molecules prefer to adopt an end-on configuration. N₂ chemical adsorption has high E_ad values (-0.65 to -1.30 eV), thereby indicating a strong and energetically favorable N₂ adsorption (Table S2). The order of E_ad obeys the following trend: Fe/MXene > Co/MXene > Ru/MXene > Rh/MXene. This suggests that Fe has a strong N₂-philicity. The N≡N bond lengths in these four catalysts are 1.134, 1.133, 1.136, and 1.128 Å, respectively, and are slightly longer than the bond length in gas phase N₂ (1.11 Å). Similarly, N≡N bond length in the TM/BN catalyst is around 1.12 Å [27].

ELF analysis [50, 51] was carried out to estimate the bond strengths. The value of ELF is set in the range of 0 to 1, where 1, 0.5, and 0 represent covalent, metallic, and nonbonding characters, respectively. Strong ELF, ∼0.8, is observed around C and O atoms, while the ELF around Ti atoms was ∼0.5. The localization of electron pairs between Ti and C atoms, O and TM atoms, and C and TM atoms indicates a metallic-ionic character (ELF = ∼0.3). In addition, there is a weak localization of electrons near the hollow site (ELF = ~0.3) on the surface of Ti₃C₂O₂ MXene.

Further, the charge density differences and Bader charge analysis [52] were performed. Green represents the gain of charge, while yellow represents the loss of charge. It can be seen that N₂ acts as the electron acceptor, with the electrons being transferred from the TM atom in all the structure. The charge transfer for all the catalysts is very small. For Fe atom, the amount of charge transfer is 0.18 e, which is the largest among the four catalysts (Table S3). It can be inferred that greater the charge transfer, the more active is the metal. Consequently, higher the number of electrons gained by N₂, better is the activation of the N–N bond. This is consistent with the previous calculation.

Furthermore, the partial density of states (PDOS) of N₂ adsorption configuration exhibits the d states of TM atoms and p states of N atom both in the spin-down and spin-up states, as shown in Fig. 3. There is spin polarization in the TM/Ti₃C₂O₂ MXene. The d orbitals of TM atoms effectively overlap with N 2p orbitals near the Fermi level. Consequently, the interaction between Fe and N₂ molecules is the strongest. The electrons in the occupied d orbitals of Fe/Ti₃C₂O₂ MXene were transferred into the antibonding orbitals of N₂, thereby resulting in the decrease of the bond order of N₂. Thus, Fe exhibits the best N₂ activation performance.

In general, ammonia synthesis may take place via two reaction mechanisms. Based on the N₂ bond breaking step, it can be defined as either associative or dissociative mechanism. Considering the probable configurations of N_xH_y species, five possible reaction pathways were investigated.

In Fig. 4, red and purple lines represent the two associative pathways (association A and association B, respectively), while blue, yellow, and black lines represent the dissociative pathways (dissociation A, dissociation B, and dissociation C, respectively). In the association A pathway, protonation occurs on one side of N₂ until the first ammonia is released. The second ammonia is generated by three other hydrogenation steps wherein the ^*N species is hydrogenated into ^*NH, ^*NH₂, and NH₃. In the association B pathway, the second H⁺ + e^- pair attacks the NNH species to form ^*NHNH species. The following protonation steps are equally probable and generate ^*NHNH₂ and NH₂NH₂. The last two steps yield two ammonia molecules, one after the other. In the dissociative pathway, N₂H separates into N atom and NH species after the hydrogenation. The dissociation A pathway shows that the protonation occurs first on ^*NH to form ammonia, followed by the synthesis of the second ammonia, as in association A. The N atoms of N₂ in the dissociation B pathway are oriented more symmetrically. Protonation preferably yields symmetrical products, such as ^*NH and ^*NH, and ^*NH₂ and ^*NH₂. Dissociation C pathway generates ^*NHNH₂ species during ^*NNH₂ hydrogenation. The next step is the dissociation into two ^*NH₂ species.

DFT calculations were also carried out to obtain a reasonable approximation of the Gibbs free energy. The correction terms (zero-point energy, enthalpy, temperature, and entropy corrections) of adsorbates and gas-phase molecules are listed in Tables S1 and S2.

The free energy profiles along the most probable pathways are summarized in Fig. 5, and the corresponding atomic configurations along the reaction paths are displayed under the profiles. It can be seen that Fe, Co, and Rh have the same reaction pathway. In Fig. 5(a), the first protonation is likely to generate ^*NNH species, with the Gibbs free energy increasing to 0.88 eV. The generation of ^*NNH species is also endothermic with 0.56 eV (Fig. 5(b)). The Gibbs free energy for the first step in the presence of Rh/MXene is 0.80 eV (Fig. 5(d)). The △G value for the second step involving the formation of ^*NNH₂ is 0.19 eV, which is larger than that on Rh/MXenes (0.07 eV). For Co/MXene catalyst, the Gibbs free energy for the formation of ^*NNH₂ was slightly smaller (-0.02 eV). All the three steps following this are exothermic, as indicated by the negative Gibbs free energies. The Gibbs free energies of these three steps are -0.62, -0.45, and -0.55 eV for Fe/MXene. The Gibbs free energies for the corresponding steps in Rh/MXene is -0.60, -0.66, and -0.04 eV, while those for Co/MXene are -0.27, -1.03, and -0.64 eV. The Gibbs free energy for the step involving ammonia synthesis is positive (0.53 eV) in the presence of Fe/MXene catalyst, while it is negative in the presence of Rh/MXene (△G = -0.20 eV) and Co/MXene (DG = -0.14 eV). The Ru/MXene has a similar pathway. The △G value for the formation of ^*NNH is 1.12 eV. The second step is the addition of H⁺ into the –NH terminal to form ^*NNH₂ species (△G = -0.67 eV). In the third step, the triple bond of N₂ breaks, and one ammonia is released with a free energy of -0.83 eV. There are three other protonation steps before the second ammonia is formed. The Gibbs free energy of these three steps are 0.34, -0.67, and -0.44 eV, respectively.

It is also necessary to evaluate the potential-determining step (PDS) and the corresponding minimum overpotential. The potential-limiting step of Fe, Co, Rh, and Ru supported on MXene is the hydrogenation of N₂ to form ^*NNH species. The corresponding overpotentials of these four catalysts are 0.88, 0.85, 0.80, and 1.12 eV, respectively.

Further, more 3d/4d TM supported on MXene catalysts have been synthesized to study the relationship between the overpotential and Gibbs free energy of the potential-limiting step. The most stable N₂ adsorption on the TM/Ti₃C₂O₂ MXenes has been studied by Gibbs free energy calculations. The G_N2 represents the Gibbs free energy change from the gas phase to ^*N₂. The results indicate that the side-on N₂ adsorption is less energetically favorable than the end-on adsorption (Table S2). Previous studies on TM/C₂N also reported similar results [27]. The negative △G values indicate favorable N₂ adsorption on all the studied 3d TM/Ti₃C₂O₂ MXenes, except Sc and Zn (Fig. 6(a)). The late 4d TM-supported MXene, except for Ag, also represents good N₂ activation (Fig. 6(c)). In contrast to several single-atom catalysts [53-55], most of the TM/Ti₃C₂O₂ MXenes, and particularly Fe, could sufficiently activate N₂ molecules, and thus facilitate their subsequent reduction steps. Fig. 6(b) shows that the overpotentials of 3d TM-supported MXene are between 0.68 and 1.96 eV, while those for the late 4d TM-supported MXene are between 0.84 and 2.33 eV (Fig. 6(d)). In addition, overpotential has an inseparable relationship with Gibbs free energy. Calculations show that there are two possible potential-limiting steps: (ⅰ) N₂ + H⁺ → NNH and (ⅱ) NH₂ + H⁺ → NH₃. In the 3d TM, Mn/MXene has the lowest overpotential and the potential-determining step is NNH generation with a △G value of 0.68 eV. For the 4d TM, Mo/MXene has the lowest overpotential, with the potential-limiting step being the formation of the NNH with a DG value of 0.84 eV. Further, we have also studied the overpotential of Au/Ti₃C₂O₂ by this relationship (Table S6)_. The results reveal that Au single atom catalyst has slightly reduced overpotential (1.74 eV) compared to other Au catalyst [56].

According to the previous studies, designing electrocatalysts that can reduce the overpotential of electrocatalytic ammonia synthesis is a challenge. Contrary to the traditional Haber-Bosch process, the N≡N bond does not break in the N₂ activation step on the TM/Ti₃C₂O₂ MXenes. It can be concluded that such catalysts can alter the reaction pathway by effectively avoiding the steps with high energy barrier, which would ultimately reduce the overpotential.

Reaction paths and overpotentials of Fe, Co, Ru, and Rh supported on MXene catalysts in the electrocatalytic ammonia synthesis have been studied using DFT. Gibbs free energy calculations reveal that the potential-determining step is N₂ + H⁺ → NNH for all the four TM/MXene catalyst. Further, we studied the relationship between the Gibbs free energy of the potential-determining steps and overpotential for a series of TM/MXene catalysts. Different TM atoms have different overpotentials, ranging from 0.68 to 2.33 eV. Calculations show that there are two possible potential-limiting steps: (ⅰ) N₂ + H⁺ → NNH and (ⅱ) NH₂ + H⁺ → NH₃. This work provides a new route for the rational designing and large-scale screening of these catalyst for electrocatalytic ammonia synthesis.