Assembly of an actinide-uranium single atom catalyst on defective MXenes for efficient NO electroreduction

Abstract

The research on actinide-uranium single atom catalysts is particularly crucial in small pollutant molecule reduction due to the unique f orbitals of uranium with adjustable oxidation states. This work screens a series of actinide-uranium single atom catalyst embedded MXene monolayers with oxygen vacancies (UO₂@MXene, MXene = Ti₂CO₂, V₂CO₂, Cr₂CO₂, Zr₂CO₂, Nb₂CO₂ and Mo₂CO₂) for product selectivity in electrocatalytic NO reduction (ENOR) by well defined ab initio calculations. Our results indicate that the change from higher oxidation state U(VI) to sparingly soluble low valent U species anchored O-defective MXene surfaces when the model of uranyl adsorbate approached O vacancies of MXenes, which can overcome the limitation of catalytic performance of actinide metal centers. Furthermore, their efficiency for the electrochemical NO reduction reaction to NH₃ was evaluated. Among all actinide-uranium single atom catalysts, UO₂@Nb₂CO₂ exhibits the best activity and selectivity in NO reduction to NH₃, characterized by the lowest theoretical limiting potential of −0.14 V. Additionally, the constant-potential method (CPM) was performed to explore pH-dependent catalytic activity of UO₂@Nb₂CO₂. The findings indicate that the onset potential is −0.228 V vs. RHE at pH = 1, which is lower than 0.305 V vs. RHE at pH = 13, which suggests that the acid environment is relatively favorable for ENOR on UO₂@Nb₂CO₂. This work leverages environmental remediation and electro-catalytic reaction in an actinide-uranium complex, offering a novel approach to establish a theoretical framework for the most effective strategies of NH₃ synthesis.

---

1. Introduction

Anthropogenic nitrogen oxide (NO) emission is a significant source of air pollution, which mainly originates from the combustion of fuels.^1,2 One of the most attractive strategies for NO removal by developing efficient catalytic processes represents a critical technique in modern catalysis.³ Besides effectively mitigating emissions, conversion of NO pollutants to NH₃ products by electroreduction have garnered intense interest, and can be used in the field of agricultural production and other chemical syntheses.^4–6 Challenges arise, however, in the electrochemical NO reduction to NH₃. Specifically, the catalysts for the hydrogenation of NO to NH₃ still suffers from low activity of pure transition metals, high cost of noble metals and side reaction competition.^7,8 In this regard, many efforts should be made to identify alternative electro-catalysts for NO removal.⁹

Actinide-based complexes, a type of complex formed by actinide elements (such as uranium, plutonium, neptunium, etc.) and ligands, can afford the prospect of unique activity relative to s-, p- and d-block metals.^10–13 Owing to properties such as large atom radii, high kinetic lability, coordination numbers, one-electron redox processes and the involvement of f orbitals in bonding, catalytic reactivity in these actinide-based complexes has stimulated a lot of attention.^14–16 Uranium exhibits characteristics such as facile switching among oxidation states and kinetic stability of coordination structures, and can be exploited for heterogeneous molecular transformations and efficient chemical reactions due to its unique reactivity. Uranium exists mainly in the form of ²³⁸U with a half-life of 4.5 × 10⁹ in nature, which is considered a stable nuclide. Its natural radiation attenuation is very slow, which is supposed to a stable nuclide. However, uranium is a heavy metal ion, and its chemical toxicity can cause damage to human organs and environmental pollution. On the other hand, uranium also exists in other forms of ²³⁵U and ²³⁴U, which can cause radiological toxicity, and increase the risk of cancer for human.¹⁷ It is therefore extremely vital that uranium contaminants are effectively removed from ecosystems.

Besides extensive applications in nuclear power, the researchers proposed the exploration of latent application prospects of uranium-based catalysis, which are further beneficial functions of uranium.^18–21 In the early 20th century, the uranium can be effectively used as catalysis for the ammonia synthesis.¹⁸ Recently, various research groups were devoted to develop active catalysts based on a series of uranium containing compounds. For example, an electron-rich and arene-supported uranium complex has been proposed to improve hydrogen elevation reaction (HER) efficiency by the reversible transformation between trivalent and tetravalent uranium.

As another example, uranium(IV) hydride was synthetized and found to have a higher activity for CO or CO₂ conversion to methoxide.²¹ Currently, single-atom catalysts (SACs) with high atomic utilization and unique active sites have become hot topics in the field of chemistry and materials science. Furthermore, single actinide-uranium catalysts possess single atom characteristics, and were connected with homogeneous and heterogeneous catalysts widely used in electrocatalytic reactions.^22,23

Because the 5f orbital of uranium is close to its 6d orbital, a unique 6d–5f hybridization was formed, leading to high reactivity towards typically inert small molecules for single actinide-uranium catalysts. Chen et al. reported the dispersion of uranium (U) single atoms on TiO₂ monolayer by oxygen vacancy confinement as a highly efficient catalyst for N₂ reduction, and formation of U–O moieties in catalyst can lower the thermodynamic energy barrier of hydrogenation step.²²

In addition, U single atom catalysts on MoS₂ nanosheets from radioactive wastewater were synthesized by a pulse voltammetry method, and offered better alkaline HER activity.²³ However, many catalytic reactions for actinide complexes are unfortunately precluded due to the high oxophilicity of actinide metal centers and strong M–O bonds. Promoting catalytic performance of actinide metal centers involves transition from higher oxidation state U(VI) species to reduced sparingly soluble U(IV), which represents one of the most effective strategies to overcome such a limitation.²⁴ Therefore, it is imperative to design advanced active sites with reduction properties to anchor actinide metal centers and explore their catalytic activity for the atomically controlled heterogeneous uranium catalysts from theoretical perspectives.

Recently, graphene-like monolayer transition metal carbide and carbonitride materials (MXenes, M_n+1X_nT_x) were synthetized, offering excellent physicochemical properties, including favorable uranium adsorption.^25,26 Depending on the chemical etching treatment, the groups of T_x on MXene surfaces may change significantly. Comparted to O and OH terminated MXenes, F terminated MXenes are less stable. Moreover, the OH group can convert to an O group on MXene surfaces at high temperatures, and thus, O-terminated MXenes could be synthetized by an experimental method.²⁷

Furthermore, MXenes can be engineered with controllable coordination environments via atomic vacancies, providing anchoring sites for single metal atoms.^28,29 In general, defects on the surface and edges of MXenes can be fabricated by an invasive delamination method. Previous experimental studies indicate that the presence of Ti vacancies in Ti₃C₂T_x will transform the termination as well as the surface morphology and carbon vacancies in Ti₂CT_x and will afford better mechanical flexibility as well as electronic conductivity.³⁰ Compared to the TM and carbon vacancies in MXenes, oxygen (O) vacancies is very prevalent, which can also be fabricated by vacancy engineering, such as thermal annealing, chemical etching, etc.^31,32

Moreover, its concentration can be regulated in the range from 0 to 25%. Currently, a variety of MXene monolayers with O vacancies were synthetized experimentally, and it was reported that these substrates can capture single metal atoms at defect sites with no nucleation in the experimental process.^33,34 In addition, the reductive properties of exfoliated Ti₃C₂T_x were reported to have noteworthy functions for reduction from Cr(VI) to Cr(III) and U(VI) to U(IV).^35,36 In other words, vacancies would significantly improve the properties of MXenes and broaden its application in the field of catalysis. As a result, an indispensable assessment of the catalytic properties of actinide-uranium single atom catalysts on defective MXenes is very significant for exploring their practical applications, such as electrocatalytic NO reduction (ENOR).

Here, we reported the computational insights into potential conversion of NO molecules to valuable products using single actinide atom catalysts on defective MXenes. First, the adsorption behavior of UO₂(H₂O)₅²⁺ on six MXene monolayers with O vacancies was determined by AIMD calculations. When the uranyl adsorbate approaches O vacancies on the MXene surface, the uranyl ions lose one water group, which transforms into the UO₄(H₂O)₄ complex and was adsorbed on O-defective MXenes. Subsequently, theoretical investigations proved that the evolved UO₄ moieties on MXenes serve as important electron back-donation centers for NO adsorption, which promote NO activation and lower the thermodynamic energy barrier in the electroreduction steps. Furthermore, we present the evaluation of pH-dependent catalytic activity of the most promising UO₂@Nb₂CO₂, and find that the acid environment is more favorable for ENOR on its surface. This structure endows superior catalytic activity toward ENOR, providing potential avenues for the rational design of a novel actinide-uranium single atom catalyst.

2. Calculation details

This theoretical study employed the Vienna ab initio simulation package (VASP) combined with the projector augmented wave (PAW) method.^37,38 The generalized gradient approximation (GGA) method associated with the Perdew–Burke–Ernzerhof (PBE) was adopted to describe the exchange–correlation effects.³⁹ The convergence criteria in our spin-polarized calculations for energy and residual force on each atom were all convergent at 1 × 10⁻⁴ eV and 0.03 eV Å⁻¹ during structural optimization, respectively. By adopting the DFT-D3 method of Grimme, the van der Waals (vdW) interactions are described.⁴⁰ The plane wave cutoff energy was set as 450 eV and a gamma-centered k-point mesh of 3 × 3 × 1 was used in all the calculations.

To simulate the solution environment, the VASP-sol package is used and the dielectric constant (ε_r) is set to 80. The Hubbard U correction (only for U atom) was employed within the DFT + U approach and U–J was set to 4 eV.^41,42 Six 3 × 3 M₂CO₂ (M = Ti, V, Cr, Zr, Nb and Mo) MXene supercells were constructed, where a 20 Å-vacuum region in the z-direction is left alongside the MXene substrate to avoid steric hindrance and interaction due to periodicity. The COHP analysis was performed by the Lobster software to obtain the bonding and anti-bonding information in the p, d and f orbitals. The energy barriers were calculated to assess the stability of UO₂@MXenes via the climbing-image nudged elastic band (CI-NEB) method.

In this work, multiple explicit water layers were introduced to first simulate uranyl species adsorption on defective MXenes. We constructed explicit solvent models of Ti₂CO₂ (a = 10.48 Å and b = 9.08 Å), V₂CO₂ (a = 10.02 Å and b = 8.68 Å), Cr₂CO₂ (a = 9.28 Å and b = 8.04 Å), Zr₂CO₂ (a = 11.42 Å and b = 9.89 Å), Nb₂CO₂ (a = 10.80 Å and b = 9.36 Å) and Mo₂CO₂ (a = 9.94 Å and b = 8.61 Å) MXenes, where the thickness of the H₂O layer (35H₂O molecules) was set to 10 Å, and H₂O density is 1 g cm⁻³. The vacuum layers of six models are all set at 20 Å to screen influences of periodic structures. The ab initio molecular dynamics (AIMD) simulations based on the NVT ensemble are implemented in the Nose–Hoover thermostat. The AIMD simulations are sustained for 5000 fs at 298 K, and the step size is 1 fs.

The calculation formula of free energy change (ΔG) for each elemental reaction step was defined as in the following eqn (1) according to the CHE model proposed by Nørskov et al.⁴³

ΔG = ΔE + ΔE_ZPE − TΔS + ΔG_U + ΔG_pH (1)

where ΔE is the elemental reaction energy difference by DFT calculation, and ΔE_ZEP and ΔS denote the differences between the adsorbed intermediates without a substrate and the gas phase molecules in zero-point energy and entropy, respectively. The ΔG_U term is the correction of the applied electrode potential, which was related to the applied electrode potential and the number of electrons transferred, respectively. ΔG_pH was calculated by using ΔG_pH = K_BT × pH × ln 10 for the correction of pH, and the value of pH is set to 0 in our work.

3. Results and discussion

3.1 Adsorption mechanism of uranyl species on O-defective MXenes

Previous investigations indicate that point defects or vacancies including metal, C and O defects usually emerge on the MXenes, which have an important influence on the properties of MXenes. As discussed in previous work, oxygen vacancies are introduced onto the MXene surface, leading to a change in its interaction properties with the adsorbates. To clearly address the thermodynamic feasibility of oxygen vacancies on the MXene surface, the formation energies of O-vacancies were calculated by means of H₂-assisted and direct heat treatment (more details in Table S1†). It was found that the formation energies for six M₂CO₂ MXenes (M = Ti, V, Cr, Zr, Nb and Mo) are 1.74, 0.86, 0.80, 2.02, 1.49 and 1.17 eV by H₂-assisted heat treatment, respectively, smaller than those obtained by direct heat treatment. Thus, formation of O-vacancies in MXenes by H₂-assisted heat treatment is feasible for achieving precise control.⁴⁴

The first task in our calculation is to investigate the binding properties of uranyl species on the MXenes with oxygen vacancies. Although uranyl adsorbates can bind with various ligands (e.g., OH⁻, H₂O, and CO₃²⁻), the work focuses on the simple penta-coordinated UO₂(H₂O)₅²⁺ species, as it is commonly found in nuclear waste water.⁴⁵ Subsequently, adsorption behaviors of UO₂(H₂O)₅²⁺ on six M₂CO₂ MXenes (M = Ti, V, Cr, Zr, Nb and Mo) with O vacancies were determined to find out adsorption performance of different water layers by AIMD calculations. According to experimental studies, uranyl ions are partially reduced to low valence uranium species after electro-adsorption, and fixed on the surface of electrode materials.⁴⁶

For O-defective MXenes, the initial adsorption configuration for uranyl adsorbate can be roughly divided into two categories, including the model of uranyl adsorbate approaching O vacancies on the MXene surface and uranyl species inserted in the middle of the water layer. Fig. 1(a)–(f) show dynamic evolution when uranyl adsorbate approaches M₂CO₂ MXene surfaces with O vacancies after 5 ps AIMD simulation at 298 K. After 5 ps simulations, we find that all initial adsorption configurations are not stable, and the axial O atom of uranyl species occupied O vacancies of MXenes, and they eventually stabilize in the form of UO₄(H₂O)₄ by replacement of its water molecules (Fig. 1).

Fig. 1 Dynamic evolution when uranyl adsorbate approaches (a)–(f) M₂CO₂ MXene (M = Ti, V, Cr, Zr, Nb and Mo) structures with O vacancies after 5 ps AIMD simulation at 298 K. (g) Interfacial charge density distributions and Bader charge analysis of UO₂@MXenes. The yellow and cyan areas indicate accumulation and depletion of electrons, respectively.

Furthermore, the total energy oscillated and the geometric structures of UO₂@MXenes are preserved well after 1.5 ps at 298 K, which confirmed the kinetic stability of UO₂@MXenes. Taking Fig. 1e for example, the uranyl ions lose one water group, which transforms into a UO₄(H₂O)₄ complex and was adsorbed on O-defective Nb₂CO₂ (namely, UO₂@Nb₂CO₂). The left one axial U–O bond length in uranyl species was 1.891 Å, larger than that of UO₂(H₂O)₅²⁺, indicating that the oxidation reduction reaction occurs between uranyl adsorbate and Nb₂CO₂ MXene with O vacancies during the adsorption process. Thus, the existence of vacancies on MXene surface affords valuable opportunities for tailoring the adsorption behaviors.

Furthermore, the adsorption behavior of uranyl species inserted in the middle of the water layer was also analyzed (Fig. S1†). As shown in Fig. S1,† UO₂(H₂O)₅²⁺ species was maintained and the hydrogen bonds were formed between the water layer and terminal O atom of MXenes. Additionally, we also observed that H₂O molecules do not interact with O defects, which was consistent with the previous case. For comparison, the O vacancies directly act as active sites, and play an important role in behaviors of uranyl ions between uranium dioxide and O vacancies, leading to the redox reactions with the substrate.

To gain further insights into the internal structural changes induced by the anchored uranyl adsorbate, the implicit solvent model was used for simplicity in subsequent calculations. The valence state of the U atom and the interfacial charge density distributions of the six systems are shown in Fig. 1g. The valence state of U atoms on UO₂@MXenes was defined as the number of electrons transferred from the U atom to the surrounding O ligands through Bader charge analysis. For the six UO₂@MXene systems (Ti, V, Cr, Zr, Nb, Mo substrates), the U atoms carried positive charges ranging from +2.35e to +2.69e. Notably, the U valence state increased from Ti to Cr and from Zr to Mo, a trend consistent with single-atom catalysts (SACs) on two-dimensional materials.

Generally, the valence state of TMs is associated with d and f orbitals of U, which can influence the binding strength with the adsorbed intermediates. To better understand the interaction between UO₂ and MXenes with O vacancies, the binding energy was explored as shown in Table S2.† It can be found that the binding energies of UO₂@MXenes are all positive, suggesting that the synthesis of UO₂@MXenes is by exothermic reactions. The corresponding interfacial charge density distributions are obtained to elucidate the charge transfer and binding properties between U atoms and O atoms of the substrate. The U atom is denoted by cyan areas, which shows a single U atom to be positively charged. Moreover, charge transfer between the anchored U atom and MXenes can be observed for six UO₂@MXenes. Additionally, the electron localization function (ELF) for UO₂@MXenes was also investigated to verify the interaction between U and MXenes with O vacancies (Fig. S2†). The results show that there were significant charge localization phenomena between the O groups of MXenes and the U atom, indicating a strong ionic bond between U and O atoms.

Furthermore, the partial density of states (PDOS) and crystal orbital Hamilton populations (COHP) of UO₂@MXenes were calculated as shown in Fig. S3.† The PDOS shows that the orbital of a single U will hybridize with the O orbital of the substrate to generate new orbitals near the Fermi level on UO₂@MXenes, which were confirmed by COHP analysis and the bonding orbitals and antibonding orbitals resulting from the hybridization of the f orbital of the U atom and p orbital of the O atom. Therefore, the empty and occupied f orbitals of U on UO₂@MXenes catalysts play a crucial role in the activation of the adsorbate.

In addition, we also found that the f orbitals of U in UO₂@Ti₂CO₂ and UO₂@Nb₂CO₂ exhibit significant splitting near the Fermi level. In the meantime, there is obvious hybridization between the p orbitals of O and the split f orbitals of U in UO₂@Nb₂CO₂. The PDOS for the formation of three U–O bonds by occupying oxygen vacancies of MXenes is shown in Fig. S4.† It is evident that the p orbitals of the three O atoms are nearly identical on UO₂@Zr₂CO₂ and UO₂@Mo₂CO₂. However, the p orbitals of the three O atoms for three U–O bonds on UO₂@Nb₂CO₂ show significant differences below −4 eV compared to those of UO₂@Zr₂CO₂ and UO₂@Mo₂CO₂, which indicate that the three O atoms have different orbital distributions. Thus, UO₂@Nb₂CO₂ may have better catalytic activity among the six UO₂@MXenes catalysts.

The stabilities of UO₂@MXenes are key issues for their actual use. Based on the above investigations, we found that the UO₂ moiety was still located on the six catalyst surfaces, which shows basic structural stability. Additionally, the calculated energy barriers are around 0.79–1.03 eV for UO₂@MXenes from S1 to S2 sites, and it was difficult for metal clusters to form due to the high diffusion energy barrier (Fig. S5†). Thus, we believe that the UO₂@MXenes can exhibit good stabilities over a long duration.

3.2 NO adsorption and reduction on the UO₂@MXene surface

A variety of products can be generated by electro-reduction of a NO molecule. However, the chemisorption of NO on the catalyst surface, as the first step in ENOR processes, is the key step for the subsequent reaction. Herein, we considered different NO adsorption modes on the U atom of the catalyst surface. Three adsorption modes (namely, N-end, O-end and side-end) were used in our calculations due to the large radius of the actinide metal atom. The adsorption energies for NO on the UO₂@MXenes was calculated as follows,

ΔE(*NO) = E(total) − E(substrate) − E(NO) (2)

where E(total), E(substrate) and E(NO) are the total energy of NO adsorption on UO₂@MXenes, UO₂@MXenes and a free NO molecule, respectively.

As shown in Fig. 2a and S6,† the N-end configuration of *NO adsorption was more stable than the O-end and side configurations for UO₂@M₂CO₂ (M = Ti, V, Cr and Mo), respectively. However, the side configuration of *NO adsorption was preferable for UO₂@Zr₂CO₂ and UO₂@Nb₂CO₂. The adsorption energies (E_ads-*NO) are −0.82, −1.17, −1.31, −1.99, −0.95 and −1.28 eV for UO₂@M₂CO₂ (M = Ti, V, Cr Zr, Nb and Mo), respectively. After *NO adsorption on UO₂@M₂CO₂ surfaces (M = Ti, V, Cr, Zr, Nb, Mo), the N–O bond lengths exhibit significant elongation with calculated values of d_NO = 1.198 Å (Ti), 1.188 Å (V), 1.181 Å (Cr), 1.340 Å (Zr), 1.229 Å (Nb), and 1.188 Å (Mo) respectively, indicating the NO molecule has been extensively activated on these catalysts.

Fig. 2 (a) Calculated NO adsorption energies with different adsorption modes on UO₂@MXenes, (b) the relationship between NO bond lengths and the valence state of U, and (c) the Bader charge and charge density differences induced by NO adsorption on UO₂@MXenes. Cyan and yellow represent electron charge depletion and accumulation, respectively.

Interestingly, a scaling relationship was found between NO bond lengths and the valence state of the U atom (Fig. 2b). This result implied that the low valence state of the U atom results in the higher *NO adsorption energy. In addition, it can be quantitatively supported by the Bader charge analyses that 0.41, 0.29, 0.24, 0.87, 0.41 and 0.24|e| are transferred to *NO from UO₂@M₂CO₂ (M = Ti, V, Cr Zr, Nb and Mo), respectively (Fig. 2c). From Fig. 2c, we can observe that charge accumulates (yellow areas) for the NO molecule, leading to the formation of U–N or U–O bonds. On the other hand, the f orbital electron of U atom was donated into the antibonding orbital, which weakens the N–O bond of *NO, causing charge depletion between U and N or O atoms.

In addition to analysis of interfacial charge density difference and Bader charge, the NO activation on six catalysts was also investigated with PDOS and COHP (Fig. S7†). Taking UO₂@Nb₂CO₂ for example, further observations on NO adsorption on UO₂@Nb₂CO₂ were disclosed in the PDOS analysis on the NO p orbital and f and d orbitals of the U atom. Once NO adsorbed on UO₂@Nb₂CO₂, the PDOS plot indicates that the electrons are transferred from the f and d orbital of the U atom to NO π^* orbitals to form a covalent interaction between the adsorbate and catalyst.

However, it was found that the NO-p states remain comparatively localized, which is evidenced by the narrower states in Fig. S7.† This feature suggests an ionic bonding formation between NO molecules and the U atom of UO₂@MXenes because additional electrons were transferred from the U atom to the p_z orbital of NO, which corresponds to the Bader charge of the U atom on UO₂@MXenes (Fig. 2c). The calculated COHP between U and N indicated that new orbitals are generated near the Fermi level via the p orbital of NO and f orbital of the U atom. In addition, the calculated ICOHP value for the U–N bond was −1.66 eV, which is weaker than U–O (O denoted an axial O atom, −6.60 eV). To sum up, NO can be activated on all the investigated UO₂@MXene monolayers by using end-on or side-on chemisorption configurations.

The NO-to-NH₃ conversion by ENOR was chosen to elucidate the catalytic performances of UO₂@M₂CO₂ because NH₃ is a potential hydrogen energy carrier and a significant crude material for fertilizers. In a typical electrocatalytic NO reduction reaction at low NO coverage, the possible reaction paths and intermediates involved are multifarious, and a flow chart is shown in Fig. 3a.

Fig. 3 (a) Schematic illustration of NO reduction pathways denoted by different color arrows, (b) screening of UO₂@MXenes for ENOR based on the free-energy changes in the key reduction steps and the NH₃ desorption at the zero electrode potential (U = 0 V). (c)–(h) The calculated free-energy change of each elementary step in ENOR on UO₂@MXenes at U = 0 V.

As a prerequisite for the reaction to proceed, NO should be effectively captured and activated on the catalyst surface first. Next, the activated NO molecule should be selectively reduced to *NHO or *NOH through coupling with one proton–electron (H⁺/e⁻) pair. Then, the reduced *NHO species couples with other H⁺/e⁻ pairs to produce the precursor *NH₂O or *NHOH intermediate, which can be further reduced to NH₃ by adding three H⁺/e⁻ pairs. On the other hand, *NOH was reduced to the NH₃ product by *N, *NH and *NH₂ intermediates, with the release of one water molecule in the meantime.⁴⁷

As mentioned above, the pathways of NO conversion to NH₃ are complicated by gaining H⁺/e⁻ pairs. First, the first *NO + H⁺/e⁻ → *NHO reduction step on the six UO₂@MXene surfaces was investigated at low NO coverage. It was worth noting that this step had commonly been considered to be the rate-determining step in ENOR investigations.

However, the computed free-energy changes (ΔG_*NO–*NHO) were exothermic for all catalysts. This is attributed to the existence of d and f orbitals in the U atom. According to previous work and our computational cost, we focus on the third (*NH₂O + H⁺/e⁻ → *NH₂OH) and the fifth (*NH₂ + H⁺/e⁻ → *NH₃) hydrogenation steps, and the calculated free-energy changes (ΔG_{*NH2O–*NH2OH} and ΔG_*NH2–*NH3) for the two key reduction steps at zero electrode potential are shown in Fig. 3b.⁴⁸

More data about ΔG_*NO–*NOH, ΔG_*NHO–*NH2O and ΔG_*NHO–*NHOH steps are presented in Table S3.† From the free-energy changes of the two key steps, we found that the more difficult the third H⁺/e⁻ gaining step, the easier the fifth H⁺/e⁻ gaining step. Additionally, the catalytic rates of NH₃ separation cannot be ignored because higher NH₃ adsorption energy leads to more difficult desorption. The NH₃ desorption free energies (ΔG_des-NH3) were also calculated as illustrated in Fig. 3b, to screen out suitable U-based MXene catalysts which facilitate rapid NH₃ release from the UO₂@MXenes surface.

For the UO₂@MXenes considered here, the calculated NH₃ desorption free energy on the UO₂@Nb₂CO₂ surface (−0.19 eV) was exothermic compared to that on the other five catalysts. Based on the above correlations, UO₂@Nb₂CO₂ exhibits highly active performance for efficient ENOR as indicated by lower values of ΔG_{*NH2O–*NH2OH} and ΔG_des-NH3 than those of other actinide-uranium single atom catalysts. Hence, reduction to NH₃ and desorption are facilitated at UO₂@Nb₂CO₂. These findings suggest that emerging SAC design strategies should prioritize uranium-based actinide metal atoms as primary active sites, establishing a new paradigm for targeted catalyst development.

The ΔG of all elementary steps at U = 0 V along the minimum energy path is depicted in Fig. 3c–h. Fig. S8 and S9† show the adsorption structures to generate NH₃via gaining five H⁺/e⁻ on the UO₂@MXenes catalysts. Regarding the UO₂@Nb₂CO₂ catalyst, the NO to NH₃ generation path is *NO → *NHO → *NH₂O → *NH₂OH → *NH₂ → *NH₃, and the speed-limiting step was *NH₂O → *NH₂OH, corresponding to a low limiting potential of −0.14 V.

To check the origin of the low U_L, the charge density differences and PDOS caused by NH₂O adsorption are shown in Fig. S10 and S11.† More electrons are transferred between the catalyst and NH₂O among the six catalysts, and U–N as well as U–O bonds were formed around the Fermi level. The impact of the U–J value on UO₂@MXenes cannot be overlooked. Fig. S12† shows the Gibbs free energy distribution diagram of the NO reduction pathway toward NH₃ on the UO₂@Nb₂CO₂ catalyst (U–J value = 0 or 3 eV). For comparison, the UO₂@Nb₂CO₂ catalyst exhibits a limiting potential of NH₃ generation of −1.25 and −0.34 V when the U–J value was set to 0 or 3 eV, respectively. The analysis suggests that ENOR catalyzed by UO₂@Nb₂CO₂ was associated with the U–J value. However, the limiting potential of the step is still *NH₂O → *NH₂OH; thus, the U–J value = 4 eV is feasible in our calculations with the optimal catalyst for achieving efficient ENOR performance.⁴²

In order to further explore the effect of interfacial water molecules on the reduction products of NO, multiple explicit water layers were introduced for NO, NHO, NH₂O, NH₂OH, NH₂ and NH₃ adsorption on UO₂@Nb₂CO₂ based on AIMD investigations. As shown in Fig. S13,† we found that the reduced intermediates can be stabilized on the U atom of the catalyst after the introduction of an explicit solvent model, and the energy and temperature oscillate over 5 ps for six systems. These results exhibit the basic stability of the interface for UO₂@Nb₂CO₂.

In addition to UO₂@Nb₂CO₂, UO₂@Ti₂CO₂ exhibits similar catalytic activity with a limiting potential of −0.54 V and low ΔG_des-NH3 (0.17 eV), which is superior to those of bulk metal catalysts (ΔG_PDS < 1.0 eV for ENOR). This is, therefore, an indication that the above discussion verified that NH₃ synthesis together with NO removal could be realized on the UO₂@Ti₂CO₂ catalyst. For UO₂@Cr₂CO₂, the NO to NH₃ generation path is similar to that on UO₂@Nb₂CO₂. The variation of the N–O bond length from *NO to *NH₂OH was 1.181, 1.356, 1.399, and 1.439 Å, respectively, which indicates no N–O bond breakage among these intermediates.

Generally, the *NH₂OH product desorption is feasible by controlling suitable reaction conditions. However, it is very difficult to formation of *NH₂OH as the free energy change (ΔG_{*NH2O–*NH2OH}) was very higher (1.04 eV), which means the ENOR process on UO₂@Cr₂CO₂ is thermodynamically difficult. In summary, these results not only demonstrated the high efficiency of the minimum free energy path, but also indicated that UO₂@Nb₂CO₂ is the best catalyst for the formation of NH₃ with a more positive limiting potential at low NO concentration.

3.3 Activity and selectivity of UO₂@Nb₂CO₂ toward NH₃ production

In this section, we focus on the origin of activity and selectivity of UO₂@Nb₂CO₂. According to the previous experimental work, the HER is the major competitive reaction against ENOR, which dramatically leads to a decrease in the current density and faradaic efficiency during ENOR. The selectivity in our calculation was determined by comparing the ΔG value of *NO and *H adsorption. As an excellent electrocatalytic material, the ΔG(*NO) should be more negative than ΔG(*H). Otherwise, the *H adsorption on the catalyst surface will dominate to incommode the active *NO. Fig. 4a depicts the comparison of ΔG(*NO) and ΔG(*H) for all of the UO₂@MXenes candidates, and the corresponding optimized configurations are shown in Fig. S14.† All candidates are located in the *NO dominant region, indicating that these six UO₂@MXenes exhibit good selectivity toward ENOR under experimental conditions.

Fig. 4 (a) The calculated ΔG(*NO) and ΔG(*H) on UO₂@MXenes; the dashed line indicates ΔG(*NO) = ΔG(*H), (b) free-energy diagrams of the (NO)₂ dimer reduction to N₂O or N₂ pathway on UO₂@Nb₂CO₂ and UO₂@Ti₂CO₂ at U = 0 V, (c) volcano illustration for limiting potentials of the UO₂@MXenes for ENOR as a function of the descriptors of ΔG(*NO), and (d) charge fluctuations of the three moieties for UO₂@Nb₂CO₂ with reaction intermediate adsorption along the consecutive reduced pathway. The three moieties included moiety 1 (the U atom combined with an axial O atom), moiety 2 (the adsorbed N_xH_yO_z intermediates), and moiety 3 (the catalysts without moiety 2).

The other predominant competitive reaction is generation of N₂O and N₂ by the NO dimer (NO)₂ at a high NO concentration. The latest experiments suggest that the adsorbed *NO is likely coupled with a solvated NO molecule, leading to the formation of a (NO)₂ dimer.⁴⁷ Therefore, NO reduction performance was controlled by regulating the NO partial pressure at different NO concentrations, and the selectivity toward the designated product strongly depends on NO coverage.

Subsequently, we examined the scenarios involving (NO)₂ dimer reduction. The (NO)₂ dimer reduction to N₂O or N₂ by continuous H⁺/e⁻ pair gaining on UO₂@Ti₂CO₂ and UO₂@Nb₂CO₂ catalyst were discussed as illustrated in Fig. 4b, and the corresponding optimized configurations are depicted in Fig. S15.† As displayed in Fig. 4b, the pathway for byproduct N₂ formation was *2NO(2N-end) → *NONOH → *NNO → *NNOH → N₂, and the calculated U_L is −0.63 and −1.29 V on UO₂@Nb₂CO₂ and UO₂@Ti₂CO₂ catalysts, respectively.

Regarding the N₂O byproduct of the Gibbs free-energy diagrams, the pathway was *2NO(2O-end) → *O + N₂O → *OH → H₂O, the determined-limiting step is *OH → H₂O and the maximum Gibbs free energy variation is 0.93 and 0.78 eV on UO₂@Nb₂CO₂ and UO₂@Ti₂CO₂ surfaces, respectively. Compared to the U_L of NH₃ on the same catalysts, we found that the whole reaction of ENOR should be easier than N₂O or N₂ byproduct formation to ensure selectivity to NH₃, indicating the superior selectivity of the actinide-uranium single atom catalyst under experimental conditions for NH₃ synthesis. Therefore, the HER and N₂ and N₂O formation can be effectively inhibited on UO₂@Nb₂CO₂ and UO₂@Ti₂CO₂ surfaces.

Across all of calculations for the whole ENOR process, the NO adsorption Gibbs free energy has a key effect on its activation, which can further affect the subsequent reduction step. Then, we explored the factors that can quantitatively show that ΔG(*NO) is vital for NO reduction. It can be found that the rate-determining step for UO₂@Zr₂CO₂ was totally different from that of the other five systems (Fig. 3), which was ascribed to the energies of adsorbed intermediates during ENOR. Therefore, as shown in Fig. 4c, the relationship between G_ad(*NO) and U_L is illustrated and a reverse volcano plot is achieved without UO₂@Zr₂CO₂. The vertex of the volcano plot represents the moderate adsorption energy of intermediates in nature, which have superior catalytic properties for the catalyst. It can be identified from the square of correlation coefficients (R² = 0.981 for the left line and R² = 0.985 for the right line) that the linear fitting results are matchable and the constructed descriptor G_ad(*NO) is feasible.

Besides, to further determine the role of the support in ENOR on UO₂@Nb₂CO₂, we calculated the Bader charge of the catalyst. For charge calculation, we divided each adsorbate into three moieties, including moiety 1 (the U atom combined with an axial O atom), moiety 2 (the adsorbed N_xH_yO_z species), and moiety 3 (the catalysts without moiety 2). As shown in Fig. 4d, both moiety 2 and moiety 3 on UO₂@Nb₂CO₂ possess a negative charge change trend with a clear fluctuation, while moiety 1 suffered loss of charges and its values fluctuate in a narrow range.

Obviously, moiety 2 and moiety 3 of the catalyst have a synergic influence on NO adsorption and activation, and their assemblies form an eligible catalyst with excellent performances. Hence, the MXene supports can be considered as an electron reservoir to promote carrier mobility, and actinide-uranium single atoms can act as an electron donor as well as a transmitter.

A variety of SACs have been extensively investigated theoretically for ENOR in previous studies. For example, TM/g-CN (U_L(NH₃) = −0.39 V) and Co–N₄/graphene (U_L(NH₃) = −0.12 V) are expected to possess high ENOR activity and selectivity.^49,50 More studies are presented in Table S4.† Except for 2P@C₂N and Cu@Ti₃C₂O₂-V₀, our UO₂@Nb₂CO₂ actinide-uranium single atom catalyst exhibited superior or considerable catalytic performance based on the smaller limiting potential. Moreover, the referred carbon substrates, such as graphene, g-CN and g-C₂N, are always soft and have poor resistance under oxidation conditions.⁵¹ However, MXenes are inorganic substrates with high surface activity. This suggests that these MXene-based catalysts were effective supports for ENOR.

3.4 pH-dependent catalytic activity of UO₂@Nb₂CO₂

Although the observed pattern of free energy disposal for the most promising UO₂@Nb₂CO₂ has been predicted with the computational hydrogen electrode (CHE) model, activity originating from d and f orbitals of the U atom, the electrode potential and neglecting corrections of pH restricted the practical applicability. To answer these questions, we used the constant-potential method (CPM) to evaluate pH-dependent catalytic activity of the most promising UO₂@Nb₂CO₂ mentioned above.⁵² The detailed calculated method is presented in the ESI.†

First, the current density associated with the *NH₂O formation step was calculated to represent the reactivity of the NO electro-catalytic reduction reaction as it is considered as the rate-determining step. The relative current density is defined as follows in eqn (S3) in the ESI.† As shown in Fig. S16,† the onset potential for NO electroreduction at pH = 13 is 0.305 V/RHE, whereas it is −0.228 V/RHE at pH = 1 from our microkinetic model. Based on the above analysis, the pH-dependent free energy plots (referenced to the RHE) are further acquired, as shown in Fig. 5a and b. For the UO₂@Nb₂CO₂ catalyst, the onset potential is −0.228 V vs. RHE at pH = 1, which is lower than 0.305 V vs. RHE at pH = 13. This result demonstrates that the acid environment is more favorable for ENOR on UO₂@Nb₂CO₂. Therefore, the acid environment and low applied potential would make *H₂NO formation even more favorable, which could address the pH-dependent activity.

Fig. 5 Free energy plots of ENOR on UO₂@Nb₂CO₂ (a) at pH = 1 and different potentials and (b) at pH = 13 and different potentials.

Combined with previous experimental analysis, uranyl ions are more easily reduced under acidic conditions. Under acidic conditions, the concentration of H⁺ can inhibit the hydrolysis of uranyl ions and enable them to exist in the solution at a higher concentration, thereby facilitating reactions with our selected substrate. Furthermore, the activation energy of the reaction may be reduced, leading to an accelerated reduction rate of uranyl ions because the presence of acid may promote the collision frequency and effective collision probability between uranyl ions and the substrate. In addition, U(VI) species can be reduced and immobilized on Ti₂CT_x MXene materials by a pH-dependent reduction mechanism, which shows the great application prospects of MXene materials in the elimination of other oxidized contaminants.³⁶ We expect that further improvement indicates a necessity for actinide-uranium single atom catalysts with vacancy modified MXenes in experiments.

4. Conclusion

In summary, ENOR using actinide-uranium single atom catalysts has been elucidated for sustainable NH₃ synthesis. The AIMD results indicate that strong adsorption was found between uranyl species and MXenes, and the axial O atom of uranyl species occupied the O vacancy site of the MXene surface when the model of the uranyl adsorbate approached O vacancies of MXenes, which transforms into the UO₄(H₂O)₄ complex and was adsorbed on O-defective MXenes.

Subsequently, NO conversion was carried out during the formation of the (UO)₄ active center. The calculated E_ads-*NO was −0.82, −1.17, −1.31, −1.99, −0.95 and −1.28 eV for UO₂@M₂CO₂ (M = Ti, V, Cr Zr, Nb and Mo), respectively, indicating the end-on or side-on chemisorption configurations of all the investigated UO₂@MXene catalysts. By free energy calculations on six catalysts, the introduction of the UO₄ moiety dramatically boosted the ENOR activity of UO₂@Nb₂CO₂. Notably, UO₂@Nb₂CO₂ exhibited high activity and selectivity with a relatively low limiting potential of −0.14 V.

Additionally, a CPM calculation for pH-dependent catalytic activity reveals that the onset potential is −0.228 V vs. RHE at pH = 1 and 0.305 V vs. RHE at pH = 13, which suggests that the acid environment is relatively favorable for ENOR on UO₂@Nb₂CO₂. The unique f orbital of U atom and flexible MXene supports were responsible for the promoted ENOR kinetics. Our investigation indicated that Nb₂CO₂ was used as an uranium adsorption material in the case of nuclear fuel enrichment and reprocessing, and also broadened the catalytic applications of actinide elements.

Data availability

The data are available from the corresponding author on reasonable request.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (52162024), Doctoral Science Foundation of East China University of Technology (DHBK2021007), the Guangdong Basic and Applied Basic Research Foundation (No. 2025A1515010442), and the Basic Research Program of Shenzhen (No. JCYJ20240813103559008).

References

1. J. Long, S. Chen, Y. Zhang, C. Guo, X. Fu, D. Deng and J. Xiao, Angew Chem. Int. Ed. Engl., 2020, 59, 9711–9718.
2. V. I. Parvulescu, P. Grange and B. Delmon, Catal. Today, 1998, 46, 233–316.
3. B. H. Ko, B. Hasa, H. Shin, Y. R. Zhao and F. Jiao, J. Am. Chem. Soc., 2022, 144, 1258–1266.
4. H. M. Liu, L. C. Bai, A. Bergmann, B. R. Cuenya and J. S. Luo, Chem, 2024, 10, 2963–2986.
5. H. M. Liu, Y. D. Zhang and J. S. Luo, J. Energy Chem., 2020, 49, 51–58.
6. T. Mou, J. Liang, Z. Y. Ma, L. C. Zhang, Y. T. Lin, T. S. Li, Q. Liu, Y. L. Luo, Y. Liu, S. Y. Gao, H. T. Zhao, A. M. Asiri, D. W. Ma and X. P. Sun, J. Mater. Chem. A, 2021, 9, 24268–24275.
7. J. John, D. R. MacFarlane and A. N. Simonov, Nat. Catal., 2023, 6, 1125–1130.
8. J. Long, S. M. Chen, Y. L. Zhang, C. X. Guo, X. Y. Fu, D. H. Deng and J. P. Xiao, Angew. Chem., Int. Ed., 2020, 59, 9711–9718.
9. Y. M. Zang, Q. Wu, S. Wang, B. B. Huang, Y. Dai, T. Heine and Y. Ma, Adv. Energy Mater., 2024, 15, 2303953.
10. W. G. Chen, Y. W. Xiang, Z. P. Wu, M. J. Wang, Y. M. Fang, Z. Z. Zhu, S. Q. Wu and X. R. Cao, Appl. Surf. Sci., 2023, 637, 157887.
11. J. Lim, C. Y. Liu, J. Park, Y. H. Liu, T. P. Senftle, S. W. Lee and M. C. Hatzell, ACS Catal., 2021, 11, 7568–7577.
12. P. L. Arnold and Z. R. Turner, Nat. Rev. Chem, 2017, 1, 0002.
13. D. C. Sonnenberger and J. G. Gaudiello, Inorg. Chem., 1988, 27, 2747–2748.
14. F. Q. Yu, G. Y. Zhang, M. X. Shu and H. M. Wang, Angew. Chem., Int. Ed., 2024, 64, e202416467.
15. M. Falcone, L. Chatelain, R. Scopelliti, I. Ivkovi and M. Mazzanti, Nature, 2017, 547, 332–335.
16. A. R. Fox, S. C. Bart, K. Meyer and C. C. Cummins, Nature, 2008, 455, 341–349.
17. A. Y. Romanchuk, I. E. Vlasova and S. N. Kalmykov, Front. Chem., 2020, 8, 630.
18. M. J. Monreal, S. I. Khan, J. L. Kiplinger and P. L. Diaconescu, Chem. Commun., 2011, 47, 9119–9121.
19. N. Tsoureas, L. Maron, A. F. Kilpatrick, R. A. Layfield and F. G. N. Cloke, J. Am. Chem. Soc., 2020, 142, 89–92.
20. G. Feng, M. Zhang, D. Shao, X. Wang, S. Wang, L. Maron and C. Zhu, Nat. Chem., 2019, 11, 248–253.
21. D. P. Halter, F. W. Heinemann, L. Maron and K. Meyer, Nat. Chem., 2017, 10, 259–267.
22. T. Chen, T. Liu, B. B. Pang, T. Ding, W. Zhang, X. Y. Shen, D. Wu, L. Wang, X. K. Liu, Q. Q. Luo, W. K. Zhu and T. Yao, Sci. Bull., 2022, 67, 2001–2012.
23. M. Liu, H. M. Chen, X. R. Tang, H. H. Liu, B. Y. Tu, W. C. Guo, Y. M. Zheng, Y. M. Liu, Y. J. Tang, R. He and W. K. Zhu, Small, 2022, 18, e2107444.
24. J. Leduc, M. Frank, L. Jürgensen, D. Graf, A. Raauf and S. Mathur, ACS Catal., 2019, 9, 4719–4741.
25. Z., H. Fu, N. Wang, D. Legut, C. Si and R. F. Zhang, Chem. Rev., 2019, 119, 11980–12031.
26. Y. J. Zhang, J. H. Lan, W. Lin, Q. Y. Wu and W. Q. Shi, J. Hazard. Mater., 2016, 308, 402–410.
27. I. Persson, L. A. Nas̈lund, J. Halim, M. W. Barsoum, V. Darakchieva, J. Palisaitis, J. Rosen and P. O. A. Persson, 2D Materials, 2018, 5, 015002.
28. Q. X. Zhang, J. A. Wang, Q. H. Yu, Q. Z. Li, R. Z. Fan, C. Li, Y. Y. Fan, C. Zhao, W. H. Cheng, P. Y. Ji, J. Sheng, C. H. Zhang, S. H. Xie, G. Henkelman and H. Li, Nat. Synth., 2025, 4, 252–261.
29. Y. Y. Sun, J. Lee, N. H. Kwon, J. Lim, X. Y. Jin, Y. Gogotsi and S. J. Hwang, ACS Nano, 2024, 18, 6243–6255.
30. Y. X. Yu, J. Phys. Chem., 2016, 120, 5288–5296.
31. L. H. Karlsson, J. Birch, J. Halim, M. W. Barsoum and P. O. A. Persson, Nano Lett., 2015, 15, 4955–4960.
32. M. Khazaei, A. Ranjbar, M. Arai, T. Sasaki and S. J. Yunoki, Mater. Chem., 2017, 5, 2488–2503.
33. D. Zhao, Z. Chen, W. Yang, S. Liu, X. Zhang, Y. Yu, W. C. Cheong, L. Zheng, F. Ren and G. Ying, J. Am. Chem. Soc., 2019, 141, 4086–4093.
34. S. Q. Zhou, Y. X. Zhao, R. Shi, Y. C. Wang, A. Ashok, F. Héraly, T. R. Zhang and J. Y. Yuan, Adv. Mater., 2022, 34, 2204388.
35. Y. L. Ying, Y. Liu, X. Y. Wang, Y. Y. Mao, W. Cao, P. Hu and X. X. Peng, ACS Appl. Mater. Interfaces, 2015, 7, 1795–1803.
36. W. Lin, H. Song, L. Y. Yuan, Z. Li, Y. J. Zhang, J. K. Gibson, L. Zheng, Z. F. Chai and W. Q. Shi, Environ. Sci. Technol., 2018, 52, 10748–10756.
37. G. Kresse and J. Furthmuller, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54, 11169–11186.
38. G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 1758–1775.
39. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868.
40. G. Lars, J. Chem. Theory Comput., 2014, 10, 968–980.
41. H. L. Shi, P. Zhang, S. S. Li, B. Sun and B. T. Wang, Phys. Lett. A, 2009, 373, 3577–3581.
42. Y. H. Wang, J. M. Xue, G. Nie and X. Guo, Chem. Phys. Lett., 2020, 750, 137444.
43. J. K. Nørskov, J. Rossmeisl, A. Logadottir, L. Lindqvist, J. R. Kitchin, T. Bligaard and H. Jónsson, J. Phys. Chem. B, 2004, 108, 17886–17892.
44. X. Qian, L. Li, Y. L. Li, Z. Y. Liu, Z. Q. Tian, C. Zhan and L. Chen, Phys. Chem. Chem. Phys., 2021, 23, 12431–12438.
45. Y. J. Zhang, Z. J. Zhou, J. H. Lan, C. C. Ge, Z. F. Chai, P. H. Zhang and W. Q. Shi, Appl. Surf. Sci., 2017, 426, 572–578.
46. Y. Liao, R. l. Lei, X. F. Weng, C. Yan, J. X. Fu, G. X. Wei, C. Zhang, M. Wang and H. Q. Wang, J. Hazard. Mater., 2023, 442, 130054.
47. Y. Wang and M. H. Shao, ACS Catal., 2022, 12, 5407–5415.
48. S. Z. Liu, G. R. Xing and J. Y. Liu, Appl. Surf. Sci., 2023, 611, 155764.
49. H. Niu, Z. F. Zhang, X. T. Wang, X. H. Wan, C. Shao and Y. Z. Guo, Adv. Funct. Mater., 2021, 31, 2008533.
50. Z. X. Wang, J. X. Zhao, J. Y. Wang, C. R. Cabrera and Z. F. Chen, J. Mater. Chem. A, 2018, 6, 7547–7556.
51. H. Yan, C. L. Su, J. He and W. Chen, J. Mater. Chem. A, 2018, 6, 8793–8814.
52. Z. Y. Ma, Y. Luo, P. Wu, J. W. Zhong, C. Y. Ling, Y. H. Yu, X. Q. Xia, B. W. Song, L. X. Ning and Y. C. Huang, Adv. Funct. Mater., 2023, 33, 2302475.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5ta01121c

---