Selective Hydrogenation of Naphthalene over γ -Al 2 O 3 -Supported NiCu and NiZn Bimetal Catalysts

: A series of Cu and Zn modiﬁed Ni / Al 2 O 3 catalysts were prepared using an incipient impregnation method for the selective hydrogenation of naphthalene into tetralin. X-ray di ﬀ raction (XRD), H 2 -Temperature programmed reduction (H 2 -TPR), transmission electron microscopy (TEM) and X-ray photoelectron spectroscopy (XPS) were applied to reveal the structure regulation, and density functional theory (DFT) calculations were performed to investigate the electronic e ﬀ ect and reactant adsorptions on the active sites. The results showed that the addition of CuO promoted the hydrogenation of naphthalene with an inhibited tetraline selectivity. However, a simultaneously increasing naphthalene conversion and tetraline selectivity were achieved over the Zn modiﬁed Ni / Al 2 O 3 catalysts. The characterization and calculation results revealed that the doping of CuO improved the hydrogenation activity with a low tetralin selectivity due to the H spillover from the Cu. The addition of ZnO decreased the interaction between NiO x and Al 2 O 3 in NiZn / Al 2 O 3 catalysts, which e ﬃ ciently increased the reduction ability of NiO x species and, thus, improved the naphthalene hydrogenation activity. The electron transfer from ZnO to NiO x weakened the adsorption of tetraline and resulted in increased tetraline selectivity. This work provides insight into developing e ﬃ cient catalysts for heavy aromatics conversions via rational surface engineering.


Introduction
With the rapid expansion of oil refining, and the aromatics and ethylene industry, the production of heavy aromatics byproducts also increased dramatically, producing millions of tons of heavy aromatics in China [1]. The composition of these heavy aromatics is complicated, and typically contains the condensed ring compounds, which makes them not suitable for the direct blending of oil products or chemical utilization. Hence, it is urgent to develop new methods to utilize the large quantity of heavy aromatics. On the other hand, the light aromatics are an important category of basic chemical raw materials and additives for the high octane oils. The market demand of light aromatics, such as benzene, toluene, xylene (BTX), etc., is flourishing [2][3][4][5][6][7]. Hence, efficiently converting the heavy aromatics into light aromatics can realize the diversification of the raw materials for light aromatics producing, which is of significant economic value.
For the efficient catalytic conversion of heavy aromatics into BTX, the corresponding catalysts must possess appropriate hydrocracking activity [8][9][10][11][12]. Namely, in this process, the condensed ring compound should be precisely hydrogenated into mononuclear aromatics and then cracked into BTX. Hence, selective catalytic hydrogenation is the most important key technology in the catalyst design, which ensures the catalytic efficiency as well as the selectivity. Naphthalene is the simplest condensed ring aromatic, which is typically used as the probe molecule for the investigation of selective hydrocracking of heavy aromatics [4,[13][14][15]. The selective hydrocracking process includes the hydrogenation of naphthalene into tetralin and the following cracking of tetralin into BTX. The deep hydrogenation of tetralin to decalin should be prevented [11,16].
Ni/Al 2 O 3 catalysts used in the hydrogenation of naphthalene have been reported recently, and the studies were mainly focused on the preparation of catalysts and the operating conditions of the reaction [17][18][19][20][21][22][23][24]. The adsorption modes of naphthalene on the active sites of the Ni/Al 2 O 3 were also investigated via theoretical calculations, such as the π and π/σ adsorption modes of the aromatic ring on the Ni surface as well as the σ desorption mode of tetralin from the Ni surface, which play important roles in the catalytic reactions [25]. However, the electronic structure adjustment of the Ni active sites and the resulting changes of the adsorption modes and selective hydrogenation activity are still rarely reported, making the mechanism and regulation rules for the selective hydrogenation of naphthalene over the Ni-based catalysts still unclear.
Herein, we introduced Cu and Zn as two typical electronic structure promoters to Ni/Al 2 O 3 catalysts, and investigated their catalytic performances for naphthalene selective hydrogenation thoroughly. The influences of the above promoters on the structure and hydrogenation properties of Ni/Al 2 O 3 catalysts were studied via the necessary characterizations, and the density functional theory (DFT) calculations were then performed to reveal the electron transfer effect between Ni and promoters as well as the adsorption behaviors of reactant molecules. Based on these results, the role of the promoters in regulating the selective hydrogenation activity will be discussed in detail. Figure 1 shows the XRD patterns of γ-Al 2 O 3 -and γ-Al 2 O 3 -supported Ni, NiCu, and NiZn catalysts after calcination at 550 • C for 3 h. For the Al 2 O 3 sample, the typical peaks of γ-Al 2 O 3 were observed at 2θ = 37.7, 39.3, 45.9, and 66.9 • . As the standard XRD patterns of γ-Al 2 O 3 (green bars in Figure 1) and NiAl 2 O 4 (orange bars in Figure 1) are similar and difficult to identify, the relative intensity of the characteristic peaks was used to differentiate the two phases. The intensity of the peak located at 45.9 • is clearly higher than that of the peak located at 37.7 • , which is in good accordance with standard γ-Al 2 O 3 . After the impregnation of NiO, the intensity of the peak located at around 37 • increased and exhibited almost the same intensity as the peak located at around 45 • . This phenomenon is ascribed to the introduction of NiAl 2 O 4 in C0, C1, and C2-2. According to the characteristic patterns of NiAl 2 O 4 , the intensity of the peak at 37.0 • was higher than that of the peak at 45.0 • , which led to an increasing intensity on samples with the impregnation of NiO. The characteristic peaks of NiO, CuO, and ZnO were not observed, indicating that the transition metals were highly dispersed on the alumina surface and strongly interacted with Al 2 O 3 . The TEM and STEM results in Figure 2 show that the metal oxides were not uniformly dispersed on the Al 2 O 3 surface. The size of the metal oxides particles ranged from 5 nm to about 15 nm. To further investigate the effect of CuO and ZnO on the Ni/Al 2 O 3 catalyst, HRTEM images ( Figure 3) of NiCu/Al 2 O 3 and NiZn/Al 2 O 3 were taken. We found that CuO and NiO particles were detected and separately loaded on the Al 2 O 3 surface for NiCu/Al 2 O 3 , and both the NiO and CuO particles were anchored on Al 2 O 3 . In addition, although the NiO and CuO were located nearby, NiCu alloy was not observed. However, on NiZn/Al 2 O 3 , we found that some NiO and ZnO particles were closely joined together, which directly shows the interaction between ZnO and NiO. When the NiO disperses on ZnO/Al 2 O 3 surface, the interaction between NiO and Al 2 O 3 may decrease accordingly.  We observed that the reduction peaks of NiO x in Ni/Al 2 O 3 were in the wide temperature range of 550-800 • C and a reduction of NiO crystals appearing at 200-300 • C was not observed. The results further indicate the strong interaction between NiO x species and the alumina support, and the Ni species in the form of Ni-Al spinel structure were more difficult to reduce than the NiO crystal [26][27][28]. For the NiCu/Al 2 O 3 sample, the reduction peak of CuO emerged at approximately 250 • C, which is consistent with Cu/Al 2 O 3 , indicating a similar state of CuO for Cu/Al 2 O 3 and NiCu/Al 2 O 3 . The peak temperatures of NiO x shifted from 676 • C for Ni/Al 2 O 3 to 643 • C for NiCu/Al 2 O 3 . This shift indicates that the doping of CuO in Ni/Al 2 O 3 can efficiently accelerate the reduction of NiO x . The result can be ascribed to the spillover of H atoms, which facilitates the reduction of NiO x . The pre-reduced Cu promotes the dissociation of absorbed H 2 to form active H species and react with the surface NiOx through surface transport [29,30]. For the NiZn/Al 2 O 3 sample, the reduction peak of NiO x split into two peaks, located at 650 and 770 • C, separately. This indicates that the NiO x species may exist in different forms.

Textural and Structural Properties of Catalysts
Based on the HRTEM results, it is reasonable to deduce that parts of the NiO x species can interact with ZnO and, hence, more weakly interact with Al 2 O 3 . These NiO x species are marked as NiO x (S1). The other NiO x species interacting strongly with Al 2 O 3 are marked as S2 sites. According to the different reduction temperatures in H 2 -TPR, the S1 sites were more easily reduced compared with the S2 sites. To confirm this assumption, we further investigated the H 2 -TPR of the samples with different Ni/Zn ratios, and the results are given in Figure 4b. This clearly shows that the peak temperature of the S1 site decreased with the ZnO content and the relative area of the lower temperature peak increased with the ZnO content, which indicates the formation of more S1 sites on the NiZn/Al 2 O 3 catalysts.

XPS Analysis
The chemical states of Ni on catalysts after calcination at 550 • C in the air were investigated using XPS. The samples after calcination were further reduced in 100% H 2 at 450 • C for 2 h. The reduced samples were also tested by XPS to study the change of the Ni chemical states. The Ni 2p XPS results of Ni/Al 2 O 3 , NiCu/Al 2 O 3 , and NiZn/Al 2 O 3 after calcination are shown in Figure 5. After spectral deconvolution, two groups of peaks were found for Ni/Al 2 O 3 with binding energies (BEs) at 875.3, 873.7, 857.9, and 856.3 eV, which are all characteristic peaks of Ni(II). This phenomenon indicates that Ni species mainly existed as Ni(II) on Ni/Al 2 O 3 with two different forms. Binding energies at 873.7 and 856.3 eV were assigned to Ni(II) 2p 1/2 and Ni(II) 2p 3/2 of NiO x . The two peaks located at 875.3 and 857.9 eV were indexed to Ni(II) 2p 1/2 and Ni(II) 2p 3/2 of NiAl 2 O 4 [31,32]. Satellite peaks around 880.7 and 863.3 eV were observed, which are the characteristic peaks of Ni(II). The Ni 2p XPS spectrum of NiCu/Al 2 O 3 is similar to that of Ni/Al 2 O 3 , which exists with two groups of peaks located at 875.2, 873.6, 857.8, and 856.2 eV. The binding energies of Ni(II) on NiCu/Al 2 O 3 exhibited no evident difference with Ni/Al 2 O 3 , indicating that the CuO species have no obvious electron effects with NiO x species. Combined with the H 2 -TPR results above, this further proved that the doping of CuO mainly influences NiOx by H spillover rather than a formation of alloy. Binding energies at 874.9, 873.1, 857.5, and 855.7 eV were ascribed to Ni(II) on NiZn/Al 2 O 3 . Comparing with Ni/Al 2 O 3 , the characteristic peaks of Ni(II) from NiAl 2 O 4 on NiZn/Al 2 O 3 were shifted to lower BEs (from 856.3 to 855.7 eV, and from 873.7 to 873.1 eV), and the relative peak area at 855.7 eV also became larger. This result further confirms that the introduction of Zn species can weaken the interaction between Ni and Al 2 O 3 by forming NiO x (S1) species.
To further investigate the effect of doping ZnO on the chemical states of Ni 2+ , the XPS spectra of NiZn/Al 2 O 3 with different ZnO contents are also shown in Figure 6. With the increasing amount of ZnO, the relative peak area of the lower binding energy increased while the peak of the higher binding energy decreased, which indicates the increase of weaker interaction NiO x (S1) species. In addition, the lower binding energy peak shifted slightly to a lower energy direction with the increase of ZnO, which illustrates that the d electrons may shift from ZnO to the NiO x . The Ni species clearly obtain more electrons from Zn species with the increase of Zn content, which will weaken the interaction between NiO x and molecules with Π electrons [33]. The XPS results of Ni supported catalysts after H 2 reduction (Figure 7) show that the spectra of Ni 2p were shifted to lower BEs at around 853, 855, 870, and 872 eV. The results show that parts of NiO and NiAl 2 O 4 were reduced into low valency under 450 • C. The peaks located at 853 and 870 eV are ascribed to the partly reduced NiO, denoted as Ni δ1+ species, and the peaks located at 855 and 872 eV are assigned to the partly reduced NiAl 2 O 4 , denoted as Ni δ2+ species (δ1+ < δ2+ < 2). However, the ratio of Ni δ1+ /Ni δ2+ is varied among the catalysts. The NiCu/Al 2 O 3 showed the highest ratio of Ni δ1+ followed by NiZn/Al 2 O 3 and Ni/Al 2 O 3 . As discussed before, the doping of Cu can accelerate the reduction of Ni 2+ by H spillover and the Ni 2+ in the form of NiO x (S1) tended to be easier to reduce compared with the NiAl 2 O 4 in Ni/Al 2 O 3 . We observed that Ni 2p of NiZn/Al 2 O 3 was slightly shifted to a lower binding energy, which also indicates an electron transfer from ZnO to Ni δ+ .  Table 1 presents the catalytic performance of the samples. We found that a naphthalene conversion of 79.6% was achieved over a Ni/Al 2 O 3 catalyst. The NiZn/Al 2 O 3 and NiCu/Al 2 O 3 catalysts showed a higher hydrogenation activity with naphthalene conversions of 86.1% and 84.9%, respectively. However, the selectivity for tetraline of NiCu/Al 2 O 3 was 58.3%, which is much lower than that of Ni/Al 2 O 3 (72.6%) and NiZn/Al 2 O 3 (82.0%). As demonstrated in part 3.1, the addition of Cu in NiCu/Al 2 O 3 can promote the reduction of NiO x species through H spillover. Hence, NiCu/Al 2 O 3 could form more active Ni δ+ (0 ≤ δ ≤ 1) species in an H 2 atmosphere, which can enhance the hydrogenation activity of the catalysts. In addition, due to the strong electron affinity of Ni δ+ (0 ≤ δ ≤ 1) species, the aromatic ring is strongly adsorbed, which results in an excessive hydrogenation of tetraline into decalin. On the other hand, as the NiO x species can be easier to reduce in a NiZn/Al 2 O 3 catalyst compared with Ni/Al 2 O 3 , more Ni δ+ species with low valence states can be formed in an H 2 atmosphere, which leads to improved hydrogenation activity. However, as Zn 2+ species tend to provide electrons to Ni δ+ , the NiZn/Al 2 O 3 catalyst had a lower adsorption ability for tetraline than Ni/Al 2 O 3 and NiCu/Al 2 O 3 . The excessive hydrogenation of tetraline was inhibited in NiZn/Al 2 O 3 and resulted in improved tetraline selectivity. Table 1 also shows the catalytic performance of NiZn/Al 2 O 3 catalysts with different Ni/Zn ratios. The naphthalene conversion decreased slightly from 86.7% to 84.5% while the tetraline selectivity increased from 75.4% to 87.4% when the Ni/Zn ratio decreased from 5 to 1.25. The significant improvement in selectivity was mainly determined by the naphthalene and tetraline adsorption ability. As more S1 centers are formed in the catalysts with the Zn content, the adsorption ability of Ni δ+ for naphthalene and tetraline become weaker due to the electron transfer, which results in the slightly lower naphthalene conversion but higher tetraline selectivity.

Gas Phase Clusters
In order to reveal the structure activity relationship of the catalysts, we performed density functional theory (DFT) calculations to investigate the adsorption of the reactants on the model catalyst surface. Figure 8 gives the structures and energies of the metal cluster and the adsorption of H 2 on the metal clusters. The results show that the Ni 4 cluster had the largest binding energy (−1168 kJ/mol), followed by Ni 2 Cu 2 , Ni 2 Zn 2 , Cu 4 , and Zn 4 , revealing the strongest binding of Ni, and the weakest binding of Zn 4 . In addition, the structures of Ni 4 and Zn 4 were regular tetrahedron, vs. the planar rhombic of Ni 2 Cu 2 , Ni 2 Zn 2 , and Cu 4 . The Ni 2 Cu 2 had the largest adsorption energy for H 2 (−166 kJ/mol), followed by Cu 4 (−151 kJ/mol), Ni 4 (−145 kJ/mol), and Ni 2 Zn 2 (−108 kJ/mol). This indicates that the Cu promoted the H 2 adsorption on the Ni cluster, vs. the Zn, which inhibited the adsorption of H 2 on the Ni cluster.  Figure 9 shows all the possible models of metal clusters on the alumina surface, as well as the adsorption of H 2 , naphthalene, and tetralin on the model catalysts. Accordingly, the Ni 4 maintained its gas phase tetrahedron structure, vs. the 'Y' shape structure of Cu 4 . However, the Zn 4 exists as the form of two isolated Zn 2 dimmer with an adsorption energy of −235 kJ/mol, indicating the weak interaction of Zn with the alumina surface. The adsorption energies of the metal clusters on the alumina surface follow the order: Zn 4 < Cu 4 < Ni 2 Zn 2 < Ni 2 Cu 2 < Ni 4 . The adsorption of H 2 on the supported model catalysts follow the order: Zn 4 < Ni 2 Zn 2 < Ni 2 Cu 2 < Ni 4 < Cu 4 , indicating the Cu 4 cluster had the highest activity for H 2 activation and was able to induce H 2 dissociative adsorption. Although the Ni cluster showed higher adsorption ability for naphthalene and H 2 than the NiZn cluster in theory, the conversion of Ni/Al 2 O 3 was lower than that of NiZn/Al 2 O 3 ; this is mainly ascribed to the lower quantity of the reduced NiO x cluster under the H 2 atmosphere in Ni/Al 2 O 3 compared with NiZn/Al 2 O 3 . In addition, the DFT results presented that the NiZn clusters in the NiZn/Al 2 O 3 catalyst had a lower tetraline adsorption energy compared with Ni and Cu, which is a benefit for the tetraline desorption during the reaction and, hence, increased the tetraline selectivity for the NiZn/Al 2 O 3 catalyst.

The Adsorption of Metal Clusters on the Alumina Surface and Molecules on the Model Catalysts
The Bader charge analysis showed that the ZnNiO x alloy did present an electron transfer effect (confirming the XPS and H 2 -TPR results), which led to a reasonable adsorption of naphthalene and tetraline on the catalysts [34]. The optimization of the product adsorption is the primary cause of the improved tetralin selectivity on the NiZn/Al 2 O 3 catalyst.

Conclusions
In summary, Cu and Zn were introduced into Ni/Al 2 O 3 catalysts, and the selective hydrogenation of naphthalene to tetraline was investigated thoroughly over these catalysts. On one hand, both Cu and Zn efficiently promoted the hydrogenation activity of Ni/Al 2 O 3 . On the other hand, Cu promoters decreased the tetraline selectivity; however, Zn modification improved the selectivity. In the NiCu/Al 2 O 3 catalysts, NiCuO x alloy was not formed, and the electrical transfer between Ni and Cu was not obvious. However, the Cu promoter induced H spillover and accelerated the reduction of NiO x species, as well as naphthalene hydrogenation, which promoted the hydrogenation activity, however, with a decrease in the tetraline selectivity.
We found that the doping of ZnO weakened the interaction between NiO x and Al 2 O 3 and promoted the formation of NiO x (S1) species in NiZn/Al 2 O 3 . The NiO x (S1) species have a good reducibility, resulting in an improvement of the naphthalene hydrogenation activity. The greatly increased tetraline selectivity is ascribed to the optimized adsorption of the tetraline product on the Ni δ+ active sites, which resulted from the electron transferring between ZnO and NiO x . This work elucidated the mechanism of the structure regulation of NiM/Al 2 O 3 (M = Cu, Zn) for selective hydrogenation of naphthalene, and this opens up a new opportunity to develop efficient industry catalysts for heavy aromatics conversion.

Catalysts Preparation
We calcined 20 g of γ-Al 2 O 3 at 400 • C for 4 h and cooled it down to room temperature in a drier, which was used as a support. Ni(NO 3 ) 2 ·6H 2 O (9.90 g) was dissolved in purified water (8 mL) to obtain the impregnating solutions. Then, the γ-Al 2 O 3 support was mixed with the above nitrate solution by an incipient impregnation, and the C0 catalyst was obtained after drying at 120 • C for 12 h and then calcining at 550 • C for 3 h. The designed mass loading of Ni atoms was of 10 wt.%, and the C0 catalyst was defined as 10Ni/Al 2 O 3 in Table 2. Similarly, the C1 catalyst was produced by the incipient impregnation of the γ-Al 2 O 3 support with Ni(NO 3 ) 2 ·6H 2 O (9.90 g) and Cu(NO 3 ) 2 ·3H 2 O (3.39 g) mixing solution. The designed mass loading of Ni and Cu atoms were 10 wt.% and 4 wt.%, respectively, which was named as 10Ni4Cu/Al 2 O 3 . In this manner, the C2-1, C2-2, and C2-3 catalysts were prepared with the same procedure as the C1 catalyst with the designed Ni mass loading of 10 wt.%, and the designed Zn mass loadings were 2 wt.%, 4 wt.%, and 8 wt.%, respectively. The actual mass loading of the metal atom in the catalysts was analyzed using inductively coupled plasma-atomic emission spectroscopy (ICP-AES), and the labels and corresponding compositions of the catalysts are listed in Table 2.

Catalysts Characterization
The powder X-ray diffraction (XRD) patterns were performed to characterize the crystal phase of the catalysts with the D/Max2550 VB/PC diffract meter instrument (Rigaku Corporation, Japan) operating at 40 kV and 40 mA with CuKα radiation (λ = 0.154056 nm), and the diffraction data (2θ) was collected between 10-90 • with a scanning rate of 8 • /min. The metal mass loading was determined by an inductively coupled plasma-atomic emission spectroscopy (ICP-AES, Thermo Fisher Scientific Inc., Waltham, MA, USA). The H 2 temperature-programmed reduction (H 2 -TPR) was operated on an automated chemisorption analyzer from Micromeritics Instrument Corporation (AutoChem 2920 II, Norcross, GA, USA) with a thermal conductivity detector (TCD). In detail, 30 mg of catalysts were used and pretreated at 400 • C for 1 h with a heating rate of 10 • C/min in an Ar flow (30 mL·min −1 ) to remove the absorbed impurities. After being cooled to room temperature under Ar, the catalyst was flushed with 10% H 2 /Ar at 30 mL·min −1 instead of pure Ar, and the TPR profiles were obtained with a temperature ramp of 10 • C/min. X-ray photoelectron spectra (XPS) were acquired on an ESCALAB250 X-ray photoelectron spectroscopy of Thermo Fisher Scientific Inc., (Massachusetts, Waltham, MA, USA) using C 1s (284.6 eV) as a reference. The reduction of the catalysts for XPS test in Figure 8 was conducted in a fixed-bed reactor with pure H 2 , and the detailed conditions are given in Chapter 4.1.4. In order to avoid air exposure, the reaction tube was displaced with high purity nitrogen after the reduction. Then, the two shut-off valves on the top and bottom of the reaction tube were closed with pressurized nitrogen filling the interior. The sealed reaction tube was removed and then transferred into the nitrogen filled glove box. The XPS sample preparation was finished in the glove box and then transferred into the XPS equipment to complete the following test. The obtained data were analyzed with the XPSpeak41 software. High-resolution transmission electron microscopy (HRTEM) and scanning transmission electron microscopy were performed using a Tecnai 20 S-TWIN at 200 kV (FEI Company, Hillsboro, OR, USA).

Catalysts Evaluation
Tridecane and naphthalene were mixed to prepare the feed for the reaction with 8.5 wt.% naphthalene. The reaction was performed with a fixed-bed reactor, loaded with 5 g of catalysts.
All catalysts were online reduced in H 2 flowing at 450 • C for 2 h before reaction, and then the reactions were carried out with the weight hourly space velocity (WHSV) at 3.0 h −1 , the molar ratio of H 2 /hydrocarbon at 3.0, the pressure at 3.0 MPa, and the reaction temperature at 375 • C. The products were analyzed with Agilent gas chromatography 7890B with a flame ionization detector (FID) detector and HP-FFAP column (Agilent, Santa Clara, CA, USA). The conversion of naphthalene (X) and the selectivity (S) and yields (Y) for tetralin were calculated as follows: in which n 0 and n 1 represent the weight content of naphthalene (wt.%) in the feed and liquid product, and n 2 represents the weight content of tetralin (wt.%) in the liquid products.

Computational Details
The non-spinel γ-Al 2 O 3 model proposed by Digne et al. was used to describe the γ-Al 2 O 3 surface, in which more than 70% of the γ-Al 2 O 3 (110) surface was exposed under the real catalytic conditions [35,36]. The p (1 × 1) super cell model with a dimension of 67.86 Å 2 was applied. Sixteen Al 2 O 3 unit cells were distributed in eight atomic layers, with a 15 Å vacuum layer separating the slabs in the z direction of the surface. The bottom four layers were frozen in their bulk positions, and the top four layers together with the adsorbed species were allowed to fully relax. Figure 10 shows the side and the top views of the γ-Al 2 O 3 (110) surface with the topmost atoms labeled with numbers. On the clean surface, tri-coordinated aluminum (Al 3c ), tetra-coordinated aluminum (Al 4c ), di-coordinated oxygen (O 2c ), and tri-coordinated oxygen (O 3c ) atoms were exposed, and were unsaturated in coordination. All computational works were calculated with the PBE-D method with plane wave basis setting as in the VASP code [37,38]. The D3 method was performed to calculate the dispersive interactions for the optimized adsorption structures. The exchange and correlation energies were described by the generalized gradient approximation (GGA) formulation with the PBE functional [39]. The Kohn-Sham one-electron states were extended in accordance with plane-wave basis sets with the cut off energy of 350 eV. The projector augmented wave (PAW) method was applied to describe the electron-ion interactions [40,41]. The Brillouin zone was generated by the Monkhorst-Pack algorithm, and sampled with 5 × 5 × 5 and 3 × 3 × 1 k-point meshes used for the unit cell and surface slabs, respectively. The gas molecules were calculated in a 15 × 15 × 15 Å 3 cubic lattice. The convergence criteria were set at 1.0 × 10 -5 eV for the SCF energy, 1 × 10 -4 eV and 0.03 eV/Å for the total energy and the atomic forces, respectively.
The adsorption energies (E ads ) for the adsorption of molecules in the surface were calculated as: where E(M/slab), E(M) ,and E(slab) were the total energies of the spinel surface slab with the adsorbed molecule, the isolated gas phase molecule, and the surface, respectively. The binding energies of the gas phase cluster were calculated as: The adsorption energy for single H 2 molecule on the metal cluster were calculated as: where E(M n ), E(M n H 2 ), E(M), and E(H 2 ) were the energies of the isolated gas phase M n , M n H 2 cluster, single M atom and H 2 molecule, respectively.