通过g-C3N4担载MNi12 (Fe, Co, Cu, Zn)纳米团簇调节甲烷化反应性能

韩萌茹 ，周亚男 ，周旋 ，储伟 ,2,＊

1四川大学化学工程学院，成都 610065

2四川大学新能源与低碳技术研究院，成都 610065

1 Introduction

In recent years, increasing carbon dioxide (CO2) emissions have been produced due to the continuous utilization of fossil fuels, which caused severe energy shortage and environmental threats1,2. To relieve the issue, lots of advanced technologies have been proposed3-5. Among them, CO2methanation6is an attractive way to produce recycle synthetic natural gas (SNG).Extensive studies7-9have been reported to use different heterogeneous catalysts for improving the CO2conversion technology.

Among heterogeneous catalysts, noble metal catalysts generally show higher reactivity and stability10, but the high cost and scarce resource limit their applications. Thus, tremendous non-precious metal catalysts11,12have been investigated for the replacement of noble metal catalysts. Among them, nickel (Ni)-based catalysts are commonly used in industry. Which is earthabundant, but still inferior promising in CO2methanation reaction13,14. It has been reported that Ni nanoparticles (NPs)supported on substrate can further enhance the stability and activity of catalysts15,16due to the strong metal-support interaction (SMSI)17.

Two-dimensional (2D) materials are excellent substrates for their large surface area, outstanding electronic and physicochemical properties18,19. Graphitic carbon nitride (g-C3N4), a unique 2D material, has drawn extensive attentions due to its high chemical stability, low cost and appropriate electronic structure20. Compared with other 2D materials (graphene21and h-BN22), g-C3N4has an inherent triangular porous structure,which is formed by six edge nitrogen atoms. The big defect site can be used as an anchoring point for the growth of metal atoms/NPs, influencing the structural stability of the material,and potentially improving surface reactivity. Li et al.23concluded that suitable transition metal doping (Cu and Mo)loading on g-C3N4(001) surface can efficiently reduce the energy barrier and control reaction route along the CO2conversion process. Liu et al.24synthesized fibrous Ag NPs/g-C3N4aerogel, which exhibits selectively photocatalytic dehydrogenation of pure methanol without undesirable byproducts (CO, CO2, etc.) formation at room temperature.

Apart from monometallic nanoclusters, bimetallic alloy nanoparticles (NPs) tune the surface electronic properties due to the formation of the hetero-atom bonds, thus enhancing their catalytic reactivities25-27. Specially, core-shell alloy nanoparticles exhibit more superior catalytic performances due to the expected core-shell interaction. Yang et al.28have suggested that icosahedral 13- and 55-atom Fe-Ni core-shell bimetallic particles have higher stabilities than that of the monometallic Fe and Ni particles. Feng et al.29demonstrated that the synergistic effect between Au and Pd as well as the strong interactions of AuPd nanoclusters (NCs) with g-C3N4endow AuPd NCs/g-C3N4as a potential electrocatalyst in oxygen reduction and hydrogen evolution.

Motivated by these studies, Ni-based core-shell alloy supported on g-C3N4substrate are designed in this work, which is expected to tune the catalytic reactivity in CO2methanation.The reactivity of CO on the slab is a critical descriptor30scaling with the catalytic activity in CO2methanation. From the“volcano relationship”31, it can be seen that CO is bonding tightly on Ni-based materials, which hinders the CO dissociation, thus, the new designed catalysts are desired to reduce the higher adsorption ability of CO, enhancing the efficiency of methanation reaction. To our knowledge, there is lack of theoretical and experimental investigations about it.

The objectives of this study are the following: (1) to discuss the effect of substituted core atom on the interaction of core-shell NPs; (2) to systematically investigate the influence of g-C3N4substrate on the properties of the Ni-based core-shell structures;(3) to explore the change of catalytic reactivity of MNi12NPs/g-C3N4composites toward the adsorption of CO.

2 Computational methods and Models

All the calculations were carried out in DMol3package32of Materials Studio based on density functional theory (DFT). The exchange-correlation interaction was treated with the generalized gradient approximation (GGA) using Perdew-Burke-Ernzerhof (PBE) functional33. With the aim to describe the van der Waals (vdW) interaction more accurately, DFT with the empirical dispersion correction (DFT-D) method was applied to the systems34. A Monkhorst pack of 3 × 3 × 1 k-points mesh was used for the Brillouin zone, which ensured the accuracy of geometry optimizations, as presented in Table S1 (Supporting Information). The double numerical atomic orbital was augmented by a polarization p-function (DNP), and orbital cutoff quality was set at fine. Furthermore, the core electrons were treated with DFT semi-core pseudopotentials (DSPPs)35.A 0.005 Ha of smearing was applied to improve the convergence,and we used a convergence criterion of 0.004 Ha·Å-1(1 Å = 0.1 nm) on force, 0.005 Å on displacement, and 2 × 10-5Ha on the total energy in geometry optimization. Spin-polarization was considered in all calculations.

The structure of single layer of g-C3N4was obtained by cleaving the unit cell of bulk g-C3N4along the (001) direction.One layer of 2 × 2 × 1 unit cell of g-C3N4(001) on the x-y plane was employed to achieve the periodicity, and a vacuum region of 1.5 nm was introduced along the z-direction for avoiding interactions between the slab and its repeated images, this vacuum is suitable for this work, as shown in Table S2(Supporting Information).

According to “magic numbers” of transition-metal clusters,the icosahedral (Ih) Ni13cluster was chosen for higher geometric stability36. The alloy MNi12NPs were formed by substituting the centeral Ni atom with other metals. To illustrate the influence of substituted M atom for isolated MNi12NPs in the gas phase,the cohesive energy (Ecoh) is calculated according to Eq. (1)

where, ENiand EMrefers to the total energy of isolated Ni atom and substituted atom, ENPsdenotes the total energy of MNi12NPs, and n defines the number of Ni atoms in the MNi12NPs.

In order to understand the influence of replacing central Ni atoms with other metal atoms, the change of cohesive energy(ΔEcoh) is computed using the following formula:

By this definition, a higher positive value of ΔEcohindicates a lower stability. On the contrary, a negative ΔEcohsuggests a higher stability.

To describe the interaction between MNi12NPs and g-C3N4nanosheet, the binding energy (Eb) is defined as following equation

where, E(total)is the total energy of MNi12NPs adsorbed on g-C3N4, E(g-C3N4)is the total energy of pure g-C3N4.

The adsorption energy of CO (ECO) on isolated MNi12NPs and MNi12NPs/g-C3N4composites is denoted according to Eq. (4)

where, Eslabcorresponds to the total energy of each stable slab,namely MNi12NPs and MNi12NPs/g-C3N4composites, ECOis the energy of isolated CO molecule and Eslab+COrefer to the total energy of CO adsorbed on the stable slab. By the definition, a negative value corresponds to an exothermic process, while the positive one means endothermic.

To further investigate the deformation of the cluster and its effect on the CO adsorption energy, the deformation energy(Edef) are discussed, the expression of Edefas follows

where Efixis the total energy of CO on the deformed NPs without g-C3N4(keep NPs fixed), and Efreeis the total energy of CO adsorbed on isolated MNi12NPs.

The deformation charge density Δρ(r) of CO on MNi12and MNi12/g-C3N4are computed, which is defined as37

where ρtotal(r), ρads(r) and ρslab(r) are electron density of the CO adsorbed on the stable slab (MNi12and MNi12/g-C3N4system),isolated CO molecule and the stable slab (MNi12and MNi12/g-C3N4system), respectively.

3 Results and discussion

3.1 g-C3N4 substrate and freestanding MNi12 NPs

It is reported that tri-s-triazine based g-C3N4is more energetically favorable than triazine one38. Hence, the tri-striazine based g-C3N4is chosen as the pristine model. After optimization, the lattice parameter of g-C3N4is 0.720 nm, which is agree well with previous theoretical calculation values of 0.715 nm39, 0.714 nm40and experiment value of 0.713 nm41.As Fig. 1a shown, three kinds of C―N bonds are included in this sheet: d1is in the middle of tri-s-triazine units, d2connects three tri-s-triazine units as bridges, and other C―N bonds are labelled as d3. The optimized bond lengths are in good consists with previous results42-44. To gain a deep insight of g-C3N4nanosheet, the partial density of states (PDOS) of N atoms are explored, as presented in Fig. 1b. It can be observed that N 2p orbital has a high intensity on the Fermi surface due to existence of sp2dangling bonds, which results in superior stability for trapping MNi12NPs.

Fig. 1 (a) Optimized structure of g-C3N4 nanosheet;(b) the PDOS of N atoms in g-C3N4.

The geometrical structures of isolated alloy MNi12NPs are optimized, as presented in Fig. 2. Corresponding structural parameters and cohesive energy (Ecoh) of isolated MNi12NPs are calculated and shown in Table 1.

From Table 1, it can be seen that the Ecohof isolated alloy MNi12NPs range from -39.90 to -34.82 eV, which indicates that these alloy are all favorable formed in thermodynamics. Besides,it can be observed that the central M atoms with less filled dshell interact more strongly with surface Ni atoms. Compared with freestanding Ni13nanoparticles, the substitution of Fe and Co atom with ΔEcoh＜ 0 results in higher stability, but Cu and Zn replace the central Ni atoms with ΔEcoh＞ 0 leading to lower stability. According to results of Min/Max dM-Niand Min/Max dNi-Ni, it can be seen that the sizes of freestanding FeNi12NP and CoNi12NP are similar to isolated Ni13NP, but the size of MNi12NPs substituted with Cu and Zn are slightly larger than that of the Ni13nanocluster due to their bigger atomic radius. The difference between high spin states and low states are expressed using ΔE, the results demonstrate that high spin configurations are more stable than low spin state energetically (details are presented in the Supporting Information).

3.2 Interaction between MNi12 NPs and g-C3N4

According to tremendous of previous studies45-47, the center of big-hollow site is the most stable adsorption site among all the possible adsorption behaviours. Therefore, this adsorption site is chosen as the anchoring point for MNi12NPs in the current work.Considering different atomic numbers and locations of MNi12NPs interfacing with the g-C3N4surface, six initial adsorption models are studied, as shown in Fig. S1. The preferredoptimizing configurations of MNi12NPs on g-C3N4are presented in Fig. 3. Different from initial configuration, the g-C3N4plane distorts into corrugated structure after MNi12NPs loading, which is in good consistent with previous results48,49.The configurations of MNi12NPs have also changed because of different central atoms. In terms of CoNi12NP and ZnNi12NP deposited on g-C3N4, the Ih structures of nanoclusters still exist.Three shell Ni atoms form a plane, bonding with the neighboring six N atoms near the big defect site of g-C3N4substrate to saturate the sp2dangling bonds. But for the other MNi12NPs (Fe,Ni, Cu), the configurations of NPs have been reconstructed, and six interfacial Ni atoms have been displaced slightly. These atoms are not only attached to adjacent C atoms, but also bonded with neighboring N atoms near the big vacancy site of g-C3N4substrate.

Fig. 2 Optimzed geometrical structures of isolated alloy MNi12 NPs: (a) FeNi12 NPs, (b) CoNi12 NPs, (c) Ni13 NPs, (d) CuNi12 NPs,(e) ZnNi12 NPs.

Corresponding structural parameters, binding energy (Eb) and the Hirshfeld charge of MNi12NPs (Q) are presented in Table 2.When MNi12NPs supported on g-C3N4substrate, Ebvaries from-9.40 to -8.39 eV, indicating that the interaction between MNi12NPs and g-C3N4are strong. In order to further investigate the deformation of clusters, the deformation energies (Edef) are calculated, as presented in Table 2. The results reveal that the deformation of NPs further weakens the CO adsorption energy.Besides, Edefof CoNi12-C3N4and ZnNi12-C3N4are the smallest,which is in accordance the little change of CoNi12and ZnNi12NPs (Fig.3b, d). According to the results of Min/Max dM-Niand Min/Max dNi-Ni, the configurations of MNi12NPs have little transformation attributing to the introduction of g-C3N4nanosheet. Besides, the positive values of Hirshfeld charge transfer indicates that MNi12NPs donating their electrons to g-C3N4. MNi12NPs acts as Lewis acids, while g-C3N4serves as Lewis base. To further understand the interaction of MNi12NPsand g-C3N4substrate, the PDOS of isolated MNi12NPs and MNi12NPs on g-C3N4are analysed, as presented in Fig. 4. The dashed lines plot the electronic properties of isolated MNi12NPs.It can be seen that 3d states of other central atoms overlap with the surface Ni-3d orbital, indicating the strong hybridization between them. But for ZnNi12NP, the energy level of Zn-3d is less compatible with the shell Ni-3d orbital. Hence, the interaction between central Zn atom and shell Ni atoms is less effective, which is the most unstable configuration among all studied isolated MNi12NPs. The solid lines describe the electronic properties of MNi12NPs supported on g-C3N4. In terms of surface Ni atoms, the peak of 3d orbital only has slight transformation after deposited on g-C3N4substrate. However, as for 3d states of central atoms, the peaks become wider and lower in comparison with the narrow and sharp speaks of isolated MNi12NPs, revealing the overlap between central atoms and surface Ni atoms reduce, the interaction between them is thus weakened.

Table 1 Structural parameters, cohesive energy (Ecoh) of freestanding MNi12 NPs and Hirshfeld charge of central M atom.

Table 2 Structural parameters, binding energy (Eb) of MNi12 NPs/g-C3N4 composites and Hirshfeld charge transfer for MNi12 NPs on g-C3N4.

Fig. 4 PDOS of isolated MNi12 NPs (dashed line) and MNi12 NPs on g-C3N4 (solid line).

According to the PDOS, it can be seen the effect of support mainly attributing to the overlap region between C-2p, N-2p and the 3d states of MNi12NPs near the Fermi level, C-2s and N-2s states show little hybridization with MNi12NPs. Compared with C-2p state, the area and height of PDOS for N-2p orbital increase due to the sp2dangling bonds of N atoms near the big vacancy site, which is in good accordance with the PDOS analysis of pure g-C3N4.

3.3 Catalytic reactivity of MNi12 NPs/g-C3N4 composites

In order to explore the effect of g-C3N4on catalytic reactivity of MNi12NPs, the binding strength of CO on MNi12NPs and MNi12NPs/g-C3N4composites are investigated respectively.Three adsorption sites are considered: top (T), the top of Ni atom;bridge (B), the midpoint of the Ni―Ni bond and center (C), the center of the triangular plane formed by three Ni atoms, as described in Fig. 5. Through DFT calculations, when the CO molecule is parallelly adsorbed on MNi12NPs/g-C3N4composites, the adsorption orientation of CO will move to vertical. Therefore, only vertical orientation of CO adsorption is selected finally. Considering distinct atoms of CO molecule close to surfaces, two adsorption configurations of CO are studied for each adsorption site, including C atoms attach to catalysts (O-C) and O atom point to catalysts (C-O). Hence, each MNi12NPs or MNi12NPs/g-C3N4composites have six possible adsorption situations for CO molecule, which are systematically calculated in the present work.

Fig. 5 Different adsorption configurations and adsorption sites for CO adsorption on MNi12 NPs and MNi12 NPs/g-C3N4 composites.

Fig. 6 (a) Eads of CO (eV) and (b) C―O bond lengths (nm) on isolated MNi12 NPs (black) and MNi12 NPs/g-C3N4 composites (red).

Corresponding ECOand structural parameters of all the possible adsorption behaviors are concluded in Table S1. It can be observed that the ECOof C-O configuration ranging from-0.53 to -0.10 eV, indicating physical adsorption. While the ECOwith O-C model vary from -2.68 to -2.13 eV, the strong adsorption energies suggest chemical adsorption. It reveals that the CO gas adsorb on catalysts with O-C configuration is more stable. Fig. S2 shows the most stable configurations of CO adsorbed on isolated MNi12NPs and MNi12/g-C3N4composites.

Fig. 6a, b show the curve of ECOand dC―O(the average C―O bond lengths) with different catalysts. It is worth mentioning that the ECOof CO on MNi12/g-C3N4composites all reduce in comparison with that of CO on isolated MNi12NPs, and the dC―Oin CO/MNi12/g-C3N4are shorter than that in CO/MNi12NPs system. The decreasing ECOand dC―Osuggest the weakening adsorption capacity with MNi12/g-C3N4complexes. What’s more, the weakening ability of Ni13/g-C3N4is lowest among MNi12/g-C3N4catalysts, indicating the synergistic reaction in alloy NPs. The little difference of dCOin ZnNi12NPs andZnNi12/g-C3N4have also verified previous analysis of PDOS.

Table 3 Hirshfeld electron transfer of CO on isolated MNi12 NPs and MNi12 NPs-C3N4 composites.

To get a deep insight into the new designed catalysts,electronic properties of Hirshfeld charge, electrostatic potential(ESP) and deformation charge density are explored. The system of CO adsorbed on FeNi12/g-C3N4composite is taken as the illustrative example.

The detail charge distribution is presented in Table 3. Positive value denotes lose electrons, while negative value represents gain electrons. The results demonstrate that CO and g-C3N4gains electrons, acting as electron acceptors, while MNi12NPs loses electrons and acts as electron donator. Because the g-C3N4substrate gains electrons from MNi12NPs, the electron number of CO obtained from MNi12/g-C3N4composites are reduced in comparison with CO on isolated MNi12NPs, which results in decreased ECOand dC―O.

For ESP analysis, the isosurfaces were constructed at 0.05 a.u.Red lobes indicate electron loss, blue lobes electron excess. As shown in the Fig. 7, the negative charge is enriched in the region of C and O atoms, positive charge accumulates in the region MNi12NPs, indicating that the C, O and MNi12NPs are activated obviously. When g-C3N4substrate is introduced, some positive charge gathers in the big defect of g-C3N4. The electron accumulates in CO molecule decrease, indicating that the catalytic activity of CO is weakened. This phenomenon is in accordance with the trends manifested by ECO, dC―Oand the Hirshfeld charge analyses.

In order to further investigate the metal-support interaction,the deformation charge density is analysed, as shown in Fig.8. In both plot, around the C atom in CO is a charge depletion region,which is in consistent with Hirshfeld charge analyses for CO molecule. In CO-MNi12system, only charge accumulation is discerned around NPs. But for the system of CO-MNi12-C3N4,charge accumulation and depletion are observed obviously in the interface between NPs and g-C3N4.This charge redistribution suggests the strong metal-support interaction , which further reduce the CO adsorption energy.

Fig. 7 The electrostatic potential of CO adsorbed on (a) isolated MNi12 NPs, and (b) MNi12 NPs-C3N4 composites.

Fig. 8 The deformation charge density analysis of CO adsorbed on(a) isolated MNi12 NPs, and (b) MNi12 NPs-C3N4 composites.

In summary, MNi12NPs/g-C3N4composites reduce the adsorption potential towards CO, tuning the surface reactivity.

4 Conclusions

Based on DFT calculation, alloy MNi12NPs supported on g-C3N4substrate were studied. The values of ΔEcohindicate the central M atom with less filled d-shell interacts more strongly with surface Ni atoms. When the MNi12NPs is deposited on g-C3N4, strong binding energies verify the high stability of these composites, which originates from the strong hybridization between these NPs and the sp2dangling bonds of N atoms in g-C3N4substrate. Furthermore, the SMSI also changes the electronic properties of catalysts. When MNi12NPs is anchored on g-C3N4, the Hirshfeld charge and ESP analysis indicate that some electrons transfer from MNi12NPs to g-C3N4substrate,leading to the decreasing electrons from MNi12NPs to CO. Thus,these new catalysts of MNi12NPs/g-C3N4composites weaken the adsorption capacity for CO, which lower surface reactivity and enhance the efficiency of methanation. This work suggests that g-C3N4is a promising material for improving the stability of deposited MNi12NPs and tuning catalytic reactivity for CO2methanation.

Acknowledgment: The authors highly appreciate the useful discussions and helps of Colleagues: Wenjing Sun, Liqiong Huang and Huan Li. The Analytical & Testing Center Sichuan University is acknowledged for providing Dmol3module.

Supporting Information: available free of charge via the internet at http://www.whxb.pku.edu.cn.