A Theoretical Study of Fe Adsorbed on Pure and Nonmetal (N, F, P, S, Cl)-Doped Ti3C2O2 for Electrocatalytic Nitrogen Reduction

The possibility of using transition metal (TM)/MXene as a catalyst for the nitrogen reduction reaction (NRR) was studied by density functional theory, in which TM is an Fe atom, and MXene is pure Ti3C2O2 or Ti3C2O2−x doped with N/F/P/S/Cl. The adsorption energy and Gibbs free energy were calculated to describe the limiting potentials of N2 activation and reduction, respectively. N2 activation was spontaneous, and the reduction potential-limiting step may be the hydrogenation of N2 to *NNH and the desorption of *NH3 to NH3. The charge transfer of the adsorbed Fe atoms to N2 molecules weakened the interaction of N≡N, which indicates that Fe/MXene is a potential catalytic material for the NRR. In particular, doping with nonmetals F and S reduced the limiting potential of the two potential-limiting steps in the reduction reaction, compared with the undoped pure structure. Thus, Fe/MXenes doped with these nonmetals are the best candidates among these structures.


Introduction
Ammonia is a raw material for the production of various fertilizers and is a potential energy source that is easy to store and transport, environmentally friendly, and relatively safe. Ammonia synthesis is important in agricultural production and energy development. However, most ammonia synthesis still relies on the Hubble-Bosch method proposed in the 20th century, which requires harsh reaction conditions (400-600 • C and 20-40 MPa) [1][2][3]. This method consumes a large amount of energy and causes significant greenhouse gas emissions [4]. In addition, other negative effects, such as adverse effects on the equipment under high-temperature and high-pressure conditions, need to be considered. Therefore, the development of environmentally friendly and less energy-demanding methodologies for NH 3 synthesis is urgently needed. Electrocatalytic ammonia synthesis has attracted increasing attention owing to its high efficiency and environmental friendliness. The introduction of electrical energy has a remarkable influence on N 2 activation and changes the reaction pathways [5], which is beneficial for the development of new stable and efficient catalysts.

Computational Methods
Density functional theory (DFT) calculations were performed using the Vienna ab initio simulation package v. 5.4.4. (University of Vienna, Vienna, Austria) [35,36]. The generalized gradient approximation with Perdew-Burke-Ernzerhof was used as an exchangecorrelation function [37]. The projector-augmented wave method was adopted to describe the effect of the core electrons on the valence electron density [38]. The cut-off energy was set to 600 eV. The convergence criteria for the energy and force were 10 −5 eV and 10 −2 eV/Å, respectively. The thickness of the vacuum layer was more than 20 Å to avoid interactions in the z-direction, and the x-and y-directions were set as periodic boundary conditions. A 3 × 3 × 1 supercell was used for all the structures. For geometric optimization, the Brillouin zones were sampled with 4 × 4 × 1 Monkhorst-Pack meshes [39], and DFT-D3 was used to accurately describe Van der Waals interactions [40]. Charge transfer was computed by Bader charge population analysis [41,42] and the electron localization function (ELF) was analyzed using the VESTA code [43].
The substitution energies (∆E sub ) of doping different nonmetallic elements (N/F/P/S/ Cl) on the surface of Ti 3 C 2 O 2 can be expressed as where E O and E NM represent the energies of a single O atom and nonmetallic elements (N, F, P, S, Cl), respectively, and were calculated using H 2 [44], H 2 O [45], NH 3 [46], HF [47], H 3 PO 4 [48], H 2 S [49], and HCl [50] from the Open Quantum Materials Database (OQMD) [51,52]. The adsorption energy (∆E ads ) of Fe anchored on NM-Ti 3 C 2 O 2−x (NM represents the surface nonmetals, O, N, F, P, S, and Cl) was calculated using the following formula: ∆G was calculated as described by Nørskov et al. [53]. Under standard reaction conditions, the chemical potential of a proton and electron pair (µ[H + + e − ]) is equal to half that of gaseous hydrogen (µ[H 2 ]). ∆G was calculated using the following formula: where ∆E is the potential energy change calculated by DFT, ∆ZPE is the zero-point energy correction, and it is calculated by calculating the frequency of the adsorbed species. T∆S is the entropy correction, which is usually available from some database, where T = 298 K; ∆G pH and neU are the contributions from the pH and electrode potential (U), respectively; n is the number of electrons transferred; U is the applied bias. ∆G pH is defined as where k B is Boltzmann's constant. For all the calculations, the pH was set to zero. The ∆E ads values of different adsorbates were calculated as follows: where ∆E cat-mol is the energy of the entire adsorption structure, E cat is the energy of the catalyst, and E mol is the energy of the adsorbate molecules such as N 2 and N x H y .

Geometric Structure
Bare Ti 3 C 2 is a hexagonal lattice with P3m1 group symmetry, five atomic layers of Ti-C-Ti-C-Ti, two exposed Ti layers, and an experimental lattice constant of 3.057 Å [54]. After structural optimization, a = b = 3.020 Å, which was in good agreement with the experimental values. Bare MXenes are unstable under relevant NRR operating conditions [55], and they are always functionalized by electronegative functional groups [56], as they are chemically exfoliated from the bulk MAX phase by HF [17,57]. O-terminated Ti 3 C 2 was used for further experiments. There are different possibilities for the adsorption of O on Ti 3 C 2 . According to previous studies [5], the most stable structure is O adsorbed at the hollow sites of the contralateral surface Ti atoms, as shown in Figure 1a,b. Nonmetallic elements (N/F/P/S/Cl) were used to modify the Ti 3 C 2 O 2 surface. ∆E sub indicates the stability of a surface before and after doping with nonmetallic elements. The ∆E sub values for N, F, P, S, and Cl were 1.79, −1.04, 0.81, −0.27, and −1.01 eV, respectively. The structure became more stable after doping with F, S, and Cl when ∆E sub < 0 and became more unstable after doping with N and P when ∆E sub > 0. Among these doping situations, doping with F had the best stability, compared with doping with other nonmetallic elements.
Pure Ti 3 C 2 O 2 and Ti 3 C 2 O 2 modified with nonmetallic elements ( Figure S1) were used to support single Fe atoms. Two different hollow sites (H1 and H2) and an O-top site on the surface were considered, as shown in Figure 1a. The O-top was unstable, and the E ads values of Fe adsorbed on H1 and H2 are listed in Table 1. Except for the F-doped structures, the Fe atoms preferred to adsorb on the H1 site, as the E ads was smaller. Notably, in the F-doped structure, the Fe atom was adsorbed on the next-nearest H1 site ( Figure 1e). As shown in Table 1, the doping of N, F, P, and S facilitates the adsorption of Fe, while it is more difficult for Fe to adsorb on the Cl-doped structure. Figure 1c-h show the most stable adsorption positions for the different catalysts.  Table 1. Adsorption energies of Fe adsorbed on different sites and N2 adsorbed on differen lysts, the charge on N2, and the charge transferred after N2 adsorption.

Species
Eads of Fe (eV) Eads of N2 (eV) Charge Transferred on N H1 H2

N2 Adsorption
Based on the Fe/NM-Ti3C2O2−x structure, N2 adsorption was calculated using There are two different positions for N2 adsorption, and advanced research has sh that N2 adsorption is closer end to end than side to side [5]. Figure 2a-f show the stable structure of N2 adsorbed on different catalysts from end to end, and Figure show the ELF of these structures. Eads ranged from −0.55 eV to −0.92 eV, which ind that the N2 adsorption has strong spontaneity, and the absolute value of Eads from to large was in the order: Fe/P-Ti3C2O2−x < Fe/S-Ti3C2O2−x < Fe/N-Ti3C2O2−x < Fe/F-Ti3C < Fe/Cl-Ti3C2O2−x < Fe/Ti3C2O2 (Table 1). After N2 adsorption, the N≡N bond lengt Fe/Ti3C2O2, Fe/N-Ti3C2O2−x, Fe/F-Ti3C2O2−x, Fe/P-Ti3C2O2−x, Fe/S-Ti3C2O2−x, and F  Table 1. Adsorption energies of Fe adsorbed on different sites and N 2 adsorbed on different catalysts, the charge on N 2 , and the charge transferred after N 2 adsorption.

Species
E ads of Fe (eV) E ads of N 2 (eV) Charge Transferred on N 2 (e) H1 H2

N 2 Adsorption
Based on the Fe/NM-Ti 3 C 2 O 2−x structure, N 2 adsorption was calculated using E ads . There are two different positions for N 2 adsorption, and advanced research has shown that N 2 adsorption is closer end to end than side to side [5]. Figure 2a-f show the most stable structure of N 2 adsorbed on different catalysts from end to end, and Figure 2g-l show the ELF of these structures. E ads ranged from −0.55 eV to −0.92 eV, which indicates that the N 2 adsorption has strong spontaneity, and the absolute value of E ads from small to large was in the order: Fe/P-   Table 1. The results show that N 2 gains electrons in all these catalysts and the translated charges increase with an increase in the number of valence electrons from N to O or from P to S and Cl in the same period. However, doping with F did not obey this rule, which may be due to the special adsorption site of Fe. Fe was adsorbed on the first nearest H1 site and followed the trend from N to O and F. These findings were consistent with those of Wang et al. [58]. A strong positive correlation exists between the electron gains of N 2 and the change in bond length: N 2 on Fe/Cl-Ti 3 C 2 O 2−x gained the most electrons and had the largest increase in bond length relative to the gas phase, whereas N 2 on Fe/P-Ti 3 C 2 O 2−x gained the least electrons and had the smallest increment in bond length relative to the gas phase.
electrons from N to O or from P to S and Cl in the same period. However, doping with F did not obey this rule, which may be due to the special adsorption site of Fe. Fe was adsorbed on the first nearest H1 site and followed the trend from N to O and F. These findings were consistent with those of Wang et al. [58]. A strong positive correlation exists between the electron gains of N2 and the change in bond length: N2 on Fe/Cl-Ti3C2O2−x gained the most electrons and had the largest increase in bond length relative to the gas phase, whereas N2 on Fe/P-Ti3C2O2−x gained the least electrons and had the smallest increment in bond length relative to the gas phase.
The partial density of states of N2 adsorbed on Fe/Ti3C2O2 or Fe/NM-Ti3C2O2−x ( Figure  3) shows spin-up and spin-down of the d orbital of the Fe atom and the p orbital of the N atom. At the Fermi level, almost no spin-up was observed, whereas the spin-down was more obvious, and the d orbital of Fe effectively overlapped with the P orbital of N near the Fermi level. The electrons in the occupied d orbital of Fe/NM-Ti3C2O2−x transferred to the antibonding orbitals of N2, as shown in Table 1, and the adsorbed N2 on different catalysts gained electrons from 0.13 e to 0.21 e, thus lowering the bond energy of N2.

N2 Reduction Mechanism
The overall e-NRR reaction on the cathode is N (g) + 6(H + e ) → 2NH (g) (6) and the anode reactions provide protons and electrons. Liu et al. [59] summarized the mechanism of the e-NRR. The e-NRR is divided into dissociation and association mechanisms by different hydrogenation (protonation and reduction) sequences and the breaking of the N≡N triple bond. In the dissociation mechanism, the N≡N bond is broken during the adsorption process (* denotes the adsorption site).

* + N → * N
Then, two separated N atoms on the surface of the catalysts receive protons and electrons, and ammonia is formed in the last hydrogenation step: * N + H + e → * NH (8) * NH + H + e → * NH (9) * NH + H + e → * NH (10) * NH → NH (11) In the association mechanism, the N≡N bond breaks at a certain hydrogenation step. According to the hydrogenation sequence, it can be further classified into distal, alternating, and enzymatic pathways. The hydrogenation step in the enzymatic pathway is similar to that in the alternating pathway; the difference is that N2 adsorbs side to side in the enzymatic pathway, but ends in the distal and alternating pathways. For the distal and alternating pathways, the first two steps are

N 2 Reduction Mechanism
The overall e-NRR reaction on the cathode is N 2 (g) + 6 H + + e − → 2NH 3 (g) (6) and the anode reactions provide protons and electrons. Liu et al. [59] summarized the mechanism of the e-NRR. The e-NRR is divided into dissociation and association mechanisms by different hydrogenation (protonation and reduction) sequences and the breaking of the N≡N triple bond. In the dissociation mechanism, the N≡N bond is broken during the adsorption process (* denotes the adsorption site).
Then, two separated N atoms on the surface of the catalysts receive protons and electrons, and ammonia is formed in the last hydrogenation step: * N + H + + e − → * NH (8) * NH + H + + e − → * NH 2 (9) * NH 2 + H + + e − → * NH 3 (10) * NH 3 → NH 3 (11) In the association mechanism, the N≡N bond breaks at a certain hydrogenation step. According to the hydrogenation sequence, it can be further classified into distal, alternating, and enzymatic pathways. The hydrogenation step in the enzymatic pathway is similar to that in the alternating pathway; the difference is that N 2 adsorbs side to side in the enzymatic pathway, but ends in the distal and alternating pathways. For the distal and alternating pathways, the first two steps are * + N 2 → * N 2 (12) In the distal pathway, the N atom moves away from the catalytically gained protons and electrons, releasing the first NH 3 molecule, as follows: * N 2 H + H + + e − → * NNH 2 (14) * NNH 2 + H + + e − → * N + NH 3 (15) Hydrogenation then occurs on the remaining N atom and releases the second NH 3 molecule according to Reactions (8)- (11). In the alternating pathway, hydrogenation occurs on two newton atoms alternatively, and NH 3 is formed until the N≡N bond is completely broken.
* N 2 H + H + + e − → * NHNH (16) After the first NH 3 is released, the remaining *NH 2 obtains protons and electrons and releases the second ammonia according to Reactions (10) and (11). Figure 4 shows the other mixed pathways that follow neither the distal nor alternating pathways but a combination of two paths. Optimized structures of all the possible elementary steps in NRR is showed in Figure S2.
In the distal pathway, the N atom moves away from the catalytically gained protons and electrons, releasing the first NH3 molecule, as follows: * N H + H + e → * NNH (14) * NNH + H + e → * N + NH (15) Hydrogenation then occurs on the remaining N atom and releases the second NH3 molecule according to Reactions (8)- (11). In the alternating pathway, hydrogenation occurs on two newton atoms alternatively, and NH3 is formed until the N≡N bond is completely broken. * N H + H + e → * NHNH After the first NH3 is released, the remaining *NH2 obtains protons and electrons and releases the second ammonia according to Reactions (10) and (11). Figure 4 shows the other mixed pathways that follow neither the distal nor alternating pathways but a combination of two paths. Optimized structures of all the possible elementary steps in NRR is showed in Figure S2. The ΔG values calculated by DFT calculations considered all correction terms, including the zero-point energy, temperature, and entropy corrections. Table 2 illustrates the EZPE and entropy corrections (TS) of different reaction intermediates on Fe/Ti3C2O2 using the TS values obtained from the National Institute of Standards and Technology [60] at T = 298 K. The catalyst as a substrate is immobilized, although the surface is different, we compared the zero-point energy with the study of Ling [61]; the difference is marginal, as N2 reduction also occurred on the transition metal atoms in Ling's research, and only the EZPE of NH3 was significantly different. NH3 is a gas phase, not an adsorbent, so other research was also compared [5]. The calculated EZPE and TS of H2 are 0.27 and 0.4 eV [60], respectively. The ∆G values calculated by DFT calculations considered all correction terms, including the zero-point energy, temperature, and entropy corrections. Table 2 illustrates the E ZPE and entropy corrections (TS) of different reaction intermediates on Fe/Ti 3 C 2 O 2 using the TS values obtained from the National Institute of Standards and Technology [60] at T = 298 K. The catalyst as a substrate is immobilized, although the surface is different, we compared the zero-point energy with the study of Ling [61]; the difference is marginal, as N 2 reduction also occurred on the transition metal atoms in Ling's research, and only the E ZPE of NH 3 was significantly different. NH 3 is a gas phase, not an adsorbent, so other research was also compared [5]. The calculated E ZPE and TS of H 2 are 0.27 and 0.4 eV [60], respectively. As shown in Figure 5a-f, for all structures, the first protonation was likely to generate *NNH species; the ∆G values for Fe/Ti 3 C 2 O 2 , Fe/N-Ti 3 C 2 O 2−x , Fe/F-Ti 3 C 2 O 2−x , Fe/P-Ti 3 C 2 O 2−x , Fe/S-Ti 3 C 2 O 2−x , and Fe/Cl-Ti 3 C 2 O 2−x increased to 0.90, 1.04, 0.85, 0.99, 0.88, and 1.01 eV, respectively. The second step is more likely to form *NNH 2 instead of the *NHNH species in the alternate path, as the energy requirements are higher, and the increments in ∆G for Fe/Ti 3 C 2 O 2 , Fe/N-Ti 3 C 2 O 2−x , Fe/F-Ti 3 C 2 O 2−x , Fe/P-Ti 3 C 2 O 2−x , Fe/S-Ti 3 C 2 O 2−x , and Fe/Cl-Ti 3 C 2 O 2−x were 0.1, 0.06, 0.12, −0.05, 0.12, and 0.07 eV to form *NNH 2 , respectively. In the subsequent hydrogenation steps, the intermediate configuration in the alternating pathway was easier to form than the first NH 3 molecule desorption in the distal pathway. The first NH 3 is not desorbed until the fifth proton is added, and adsorptive *NH 3 is formed when the sixth proton is added. The reaction *NNH 2 → *NHNH 2 → *NH 2 NH 2 → *NH 2 → *NH 3 is exothermic, and larger energy input is required until the adsorptive *NH 3 is desorbed to form the second NH 3 molecule. The ∆G values of Fe/Ti 3 C 2 O 2 , Fe/N-Ti 3 C 2 O 2−x , Fe/F-Ti 3 C 2 O 2−x , Fe/P-Ti 3 C 2 O 2−x , Fe/S-Ti 3 C 2 O 2−x , and Fe/Cl-Ti 3 C 2 O 2−x were 1.95, 1.11, 0.97, 1.07, 1.09, 0.99 eV, respectively. However, it was reported that the use of an acidic electrolyte can promote NH 3 desorption, as the protonation of adsorbed NH 3 to form NH4 + can easily proceed [62,63], so the actual energy barrier is even smaller. For all these structures, the two potential limiting steps were the first hydrogenation of N 2 to form the *NNH species and the last process of NH 3 desorption to form the second NH 3 molecule. Compared with the original structure, nonmetallic doping was beneficial for the desorption of the last NH 3 molecule, but only the doping of F and S was beneficial for the formation of *NNH and NH 3 . Figure 6 shows the most possible reaction pathway for different catalysts. All these structures are likely to follow the mixed pathway: N 2 → *N 2 → *NNH → *NNH 2 → *NHNH 2 → *NH 2 NH 2 → *NH 2 → *NH 3 → NH 3 . In addition, the doping of nonmetals has a remarkable effect on NRR. For N 2 adsorption, E ads is reduced, compared with the nondoped structure, which may be the reason why NH 3 desorption is easier in the last step. In the hydrogenation process, the doping of different nonmetals also makes each step of the hydrogenation easier or harder. The doping of N, P, and Cl makes it difficult for *N 2 to form *NNH, whereas F and S facilitate the formation of *NNH from *N 2 . From *NNH to *NNH 2 , only the doping of P shows an obvious impact and makes the transformation occur spontaneously. In comparison, the other doped nonmetals do not show a great effect. The doping of nonmetal also does not have much influence on *NNH 2 → *NHNH 2 → *NH 2 NH 2 → *NH 2 → *NH 3 , as these reactions are exothermic for all structures. Considering the stability of nonmetal doping, the best catalysts may be Fe/F-Ti 3 C 2 O 2−x and Fe/S-Ti 3 C 2 O 2−x .   Figure 6 shows the most possible reaction pathway for different catalysts. All these structures are likely to follow the mixed pathway: N2 → *N2 → *NNH → *NNH2 → *NHNH2 → *NH2NH2 → *NH2 → *NH3 → NH3. In addition, the doping of nonmetals has occur spontaneously. In comparison, the other doped nonmetals do not show a great fect. The doping of nonmetal also does not have much influence on *NNH2 → *NHN → *NH2NH2 → *NH2 → *NH3, as these reactions are exothermic for all structures. C sidering the stability of nonmetal doping, the best catalysts may be Fe/F-Ti3C2O2−x a Fe/S-Ti3C2O2−x.

Conclusions
The reaction pathway of the TM atom, Fe, adsorbed on pure Ti3C2O2 and surface n metal (N/F/P/S/Cl)-doped Ti3C2O2 as the N2 reduction reaction catalyst was calculated ing DFT. The main limiting steps of the reaction are *N2 + H → *NNH and *NH3 → N and the limiting potentials of the two steps can reach 0.85-1.01 and 0.97-1.95 eV, resp tively. Compared with pure Ti3C2O2, nonmetal doping has an impact on catalytic perf mance. The doped nonmetal (N/F/P/S/Cl) reduces the energy barrier to form NH3 in last step, and only the doping of F and S is beneficial to the formation of *NNH in the f step and the desorption of *NH3 in the last step. Therefore, the materials doped wit and S are considered better candidate materials for NRR among the tested catalysts. O research demonstrates a feasible way to search for new NRR catalysts by modifying surface of MXenes and loading TM atoms as new catalysts.
Supplementary Materials: The following are available online at www.mdpi.com/xxx/s1, Figure  S1: Top and side views of Ti3C2O2 and its nonmetal doped structure, Figure S2: Optimized struc tures of all the possible elementary steps in NRR, taking Ti3C2O2 as an example. Other nonmeta doped Ti3C2O2 show similar geometric structure.

Conclusions
The reaction pathway of the TM atom, Fe, adsorbed on pure Ti 3 C 2 O 2 and surface nonmetal (N/F/P/S/Cl)-doped Ti 3 C 2 O 2 as the N 2 reduction reaction catalyst was calculated using DFT. The main limiting steps of the reaction are *N 2 + H → *NNH and *NH 3 → NH 3 , and the limiting potentials of the two steps can reach 0.85-1.01 and 0.97-1.95 eV, respectively. Compared with pure Ti 3 C 2 O 2 , nonmetal doping has an impact on catalytic performance. The doped nonmetal (N/F/P/S/Cl) reduces the energy barrier to form NH 3 in the last step, and only the doping of F and S is beneficial to the formation of *NNH in the first step and the desorption of *NH 3 in the last step. Therefore, the materials doped with F and S are considered better candidate materials for NRR among the tested catalysts. Our research demonstrates a feasible way to search for new NRR catalysts by modifying the surface of MXenes and loading TM atoms as new catalysts.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/nano12071081/s1, Figure S1: Top and side views of Ti 3 C 2 O 2 and its nonmetal doped structure, Figure S2: Optimized structures of all the possible elementary steps in NRR, taking Ti 3 C 2 O 2 as an example. Other nonmetal-doped Ti 3 C 2 O 2 show similar geometric structure.

Author Contributions:
The study was planned and designed by H.L., X.W., C.W. and P.Q. DFT calculations were performed by H.L. and X.W. The manuscript was prepared by H.L., L.X. and M.S. H.L., X.W., C.W., L.X., M.S. and P.Q. discussed the results and commented on the manuscript together. All authors have read and agreed to the published version of the manuscript. Data Availability Statement: The datasets generated during and/or analyzed during the current study are available from the corresponding author.