The actual concentration of nickel and aluminum in the prepared spinel-based catalysts was determined by ICP-OES, and the actual Ni/Al ratio was accordingly calculated (Table 1). All the prepared catalysts displayed the same Ni/Al molar ratio which is equal to the theoretical Ni/Al ratio. This indicated that no loss of Al or Ni ions occurred during the preparation process regardless of calcination temperature. This also led to the assumption that the different physiochemical properties of prepared nickel aluminate samples might be attributed to the variation in the distribution of Ni_octa and Ni_tetra with the variation of calcination temperature. The Ni/Al molar ratio (0.73) of the prepared nickel aluminate samples was higher than that of typical nickel aluminate (NiAl₂O₄, Ni/Al molar ratio was 0.5), indicating that there were additional Ni²⁺ ions in the present spinel structures. This could be due to the inversion structure of the nickel aluminate spinel structure, in which the additional Ni²⁺ ions were occupied in the octahedral sites to provide more desirable active sites upon the reduction.

N₂ physisorption was carried out to determine the S_BET, S_micro, and S_meso of synthesized catalysts. All the analyzed samples displayed the hysteresis loops corresponding to the mesoporous structure with no significant deviation in the pore diameter values. While increasing the calcination temperature of samples, the S_BET value was decreased from 185.86 to 156.02 m²/g, whereas the S_meso value was increased with a loss of S_micro and became stagnant after a certain calcination temperature (650^°C) (Table 1). Compared to the reported nickel aluminate, the obtained higher surface area and porosity of the samples could be attributed to the evolution of CO₂ from citric acid (Rogers et al., 2016; Miryam et al., 2017; Jing et al., 2018). Considering the textural properties of reduced catalysts, Ni/NiAlO_X-650 exhibited the highest mesoporous surface area (177.95 m²/ g) by which Ni/NiAlO_X-650 might possess a higher performance in the hydrogenation of PHE with a better mass transfer.

In order to confirm the Ni occupation in the as-prepared catalyst, we conducted XRD characterization. Figure 1 displayed the powder XRD patterns of the samples before and after reduction. Compared with the standard peaks of NiO (PDF#78-0643) and NiAl₂O₄ (PDF#78-1601), the peaks of the samples calcined at the lower temperature (<750^oC) were shifted (Figure 1a), which was related to the expansion in the nickel aluminate spinel structure. The expansion was derived from the replacement of smaller Al³⁺ (0.54 Å) by larger Ni²⁺ (0.69 Å) ions, especially at the octahedral sites. This phenomenon indicated that major Ni²⁺ ions occupied at the octahedral sites in the spinel structure. The occupancy of Ni²⁺ ions predominantly at the octahedral sites in the spinel structure can be evidenced by the larger intensity of the peak (45.0°) corresponding to the presence of Ni_octa sites as well as no intensity of the peak (31.4°) corresponding to the presence of Ni_tetra sites (Han et al., 2004; Tirsoaga et al., 2011; Miryam et al., 2017). On the other hand, for the sample prepared by calcining at 750^oC, the appearance of characteristic powder XRD peaks of NiAl₂O₄ and NiO species is apparent. Especially, the diffraction peak at 31.4°, which belonged to Ni_tetra, was obvious. It indicated that the higher calcination temperature promoted the occupancy of tetrahedral sites by Ni²⁺ ions (O'Neill et al., 1991). Meanwhile, the separation of NiO and Ni_tetra structures occurred at the higher calcination temperature. DR UV-Vis spectra of the calcined samples (Supplementary Figure S1) further proved this conclusion. The absorption bands of the stoichiometric NiAl₂O₄ spinel structure at 642 and 597 nm were not seen for the samples calcined less than 700^oC. The intensity of these peaks was weak for the sample calcined at 700^oC and became relatively strong in the case of the sample calcined at 750^oC. This was in accordance with the XRD results and further indicated the formation of the stoichiometric NiAl₂O₄ spinel structure with Ni²⁺ occupying Ni_tetra sites in the samples calcined at temperature (≥700°C) according to the ion-exchange reaction N i octa + A l t e t r a ⇌ N i tetra + A l octa (Elias et al., 2019).

After the reduction of calcined samples at 520^oC, the reduced samples were characterized by the powder XRD analysis (Figure 1b). The characteristic peak of metallic Ni species (PDF#65-2865) appeared with the obvious decrease in the intensity of the characteristic peak corresponding to Ni_octa sites. However, though the obvious decrease in the peak intensity was noticed, there were still the characteristic peaks of the nickel aluminate structure seen in Figure 1b. This suggested that those unreduced Ni²⁺ ions were present in the bulk nickel aluminate phase and served as the support. The peak corresponding to Ni(200) surface was considered for measuring the crystallite size (3 nm) of metallic Ni species by the Scherrer formula. This small crystallite size can be assigned to the strong MSI experienced by the predominant Ni_octa sites in the calcined samples, and these Ni²⁺ ions are supposedly reduced to small Ni⁰ sites with a higher dispersion.

From the H₂-TPR (Figure 2) analysis of calcined samples, the quantification of different Ni²⁺ was accomplished on the basis of Ni–support interaction (reducibility of different Ni²⁺ species). The H₂-TPR data of bare Al₂O₃ prepared by the same modified sol–gel method was also given to make sure the observed TPR peaks belong to the reduction of Ni²⁺ species. In general, for the nickel aluminate samples, the reducible Ni²⁺ ions were categorized into four types by reduction temperatures, such as the bulk or free NiO, NiO bonded to Al₂O₃, and Ni²⁺ ions occupying the crystalline spinel structure of NiAl₂O₄ (Ni_octa or Ni_tetra sites) (Zhou et al., 2015; Li et al., 2019). According to the qualitative analysis of XRD and DR UV-Vis data discussed above, the different Ni²⁺ ions found in the prepared nickel aluminate samples could be free NiO or NiO bonded to nickel aluminate, Ni_octa and Ni_tetra of the NiAl₂O₄ spinel structure. As compared to the literature, the reduction temperature for NiO seemed higher in Figure 2, suggesting the interaction of formed NiO species with nickel aluminate (Yu et al., 2019). In comparison to samples calcined at other temperatures, the reduction peak corresponding to NiO was negligible for the sample calcined at 650^oC. Furthermore, this particular sample calcined at 650^oC contains more Ni_octa sites (98%) of the spinel structure according to the intensity of reduction peak observed at 658°C. For the samples calcined at the temperature above 650^oC, the reduction peak corresponding to Ni_tetra appeared. Moreover, the increase in the intensity of reduction peak corresponding to NiO bonded to nickel aluminate was noticed. Thus, H₂-TPR analysis proved the decrease of the inversion parameter in the spinel structure of nickel aluminate with the increase of calcination temperature above 650^oC (O'Quinn et al., 2017).

X-ray photoelectron spectroscopy analyses were carried out to identify the electronic state of Ni on the surface of reduced catalysts (Figure 3). The characteristic Ni⁰ 2p_3/2, Ni²⁺ 2p_3/2, and the satellite peak of Ni²⁺ 2p_3/2 usually appeared at the binding energy of 852.5, 856.0, and 861.8 eV, respectively (Yancheshmeh et al., 2020). These three peaks were observed in the XPS analysis of the present catalysts. Compared to the sample prepared by mechanical mixing (852.32 eV), the distinctive peak of Ni⁰ 2p_3/2 appeared at the higher binding energy for Ni/NiAlO_x catalysts due to their electron-deficient Ni sites. As was discussed above, the electron-deficient state of Ni sites in the prepared catalysts could be advantageous for promoting the activation of aromatics through adsorption.

Hydrogenation of PHE was performed under optimal reaction conditions (300°C, 5 MPa, a H₂/oil volume ratio of 600, and WHSV of 52 h⁻¹) to find out the best catalyst among the as-synthesized catalysts. The change in selectivity of PHP with reaction time during the hydrogenation of PHE over the attempted catalysts is shown in Figure 4, indicating PHP as the major product without the formation of any gas and ring opening products. All the Ni/NiAlO_X catalysts presented remarkable initial activity both in terms of PHE conversion (∼99%) and PHP selectivity (∼97%) even under high WHSV (52 h⁻¹) due to the improved adsorption of aromatics that was promoted by the electron-deficient Ni sites. Ni/NiAlO_X-650 reveals a comparatively higher hydrogenation activity than the other Ni/NiAlO_X-T catalysts due to its metallic Ni sites predominantly produced from Ni_octa sites of the calcined form which expose the strong MSI. As evidenced by the CO-pulse adsorption data (Table 2), the strong MSI advantageously provided the higher Ni dispersion (1.99%) in Ni/NiAlO_x-650 catalysts. The relatively higher S_meso of Ni/NiAlO_X-650 (Table 1) could also contribute to the higher performance Ni/NiAlO_X-650 in the hydrogenation of PHE.

To explain the gradual decrease of PHP selectivity (Figure 4) with an increase of s-OHP selectivity during the PHE hydrogenation, the characterization of spent catalyst was important. The comparative of XPS analysis of fresh and spent Ni/NiAlO_X-650 was shown in Figure 5. The characteristic binding energy (Ni⁰ 2p_3/2) corresponding to metallic Ni species has been shifted to higher binding energy for the spent Ni/NiAlO_X-650 catalyst, which ranges from 852.4 to 852.7 eV, indicating the increase of electron deficiency of Ni⁰ sites during the PHE hydrogenation. For further insights about the role of the electronic structure of Ni⁰ species in PHE hydrogenation, the change in the electronic structure of surface Ni⁰ species after the PHE hydrogenation was studied by FT-IR analysis using CO as a probe molecule (Figure 6). In comparison, the upward shift, that is, 2050 cm⁻¹ to 2064 cm⁻¹ (spent catalyst at 4 h) and 2067 cm⁻¹ (spent catalyst at 8 h), is observed in the characteristic stretch of linear CO absorption after PHE hydrogenation due to the weakening of interaction between surface Ni⁰ sites and CO (Yang et al., 2017). This can be attributed to the enhancing trend of electron deficiency of Ni⁰ sites during the PHE hydrogenation progress, which hampered the π-back bonding from Ni⁰ to CO (Lee and Oyama 2006), as supported by XPS results. Contrary to linear CO adsorption, no obvious shift was noticed in the peaks corresponding to the bridged CO adsorption (1936 cm⁻¹ and 1874 cm⁻¹) but the intensity of such peaks is apparently decreased. In particular, the bridged CO adsorption peak completely disappeared in the CO FT-IR spectrum of spent Ni/NiAlO_X-650 catalysts for 8 h. This suggested that the Ni⁰ sites in the spent Ni/NiAlO_X-650 catalysts became completely unsuitable to establish the CO absorption through bridging mode.

To ensure that the decrease in activity during the progress of PHE hydrogenation occurred primarily by the change in the electronic structure, TEM analysis of the spent catalysts have been performed (Figure 7; Yang et al., 2017). According to TEM results, the fresh and spent Ni/NiAlO_X-650 catalysts displayed the same pattern of lattice fringes with 0.204 nm lattice spacing that corresponds to Ni(111) plane, indicating that the geometric structure had no impact in inhibiting the PHE hydrogenation with time. Expecting the same adsorption mechanism for aromatics like CO over Ni⁰ sites (Lee and Oyama 2006), the charge transfer between surface Ni and aromatic rings could be reduced by the enhanced electron-deficiency of Ni⁰ sites, which was unfavorable for aromatic activation.

To correlate the electronic properties of Ni⁰ sites with the PHP selectivity at different reaction times, a plot is made between the binding energy of Ni⁰ 2p_3/2 in different catalysts (Figure 8). To support this study, a catalyst prepared from the reduction of a mechanical mixture containing NiO and Al₂O₃ (NiO + Al₂O₃-mixed) was taken as reference, whose binding energy (852.38 eV) was considered as a null point for the comparison to derive the shift in binding energy of Ni⁰ 2p_3/2. As expected, the PHP selectivity was zero while using the NiO + Al₂O₃-mixed catalyst evaluated at the same hydrogenation conditions. For the comparison, the instant selectivity (1 h) as well as the end selectivity (8 h) of PHP while using different catalysts have been considered. According to Figure 8, the binding energy of Ni⁰ 2p_3/2 in the range of 852.4–852.6 eV, that is, 0.1–0.3 eV shift from the actual binding energy of NiO + Al₂O₃-mixed, was essential, and that provided the favorable electron deficiency of Ni⁰ sites to enable better PHP selectivity (>90%). This big difference in activity caused by small changes in the electron state could be understood to seek the match of orbital energy levels. The electron-deficient state could improve the electronic donation from aromatics to Ni atom for σ bonding. However, an excessive deficient degree would limit the filling of electrons in antibonding orbitals to form π-back bonding.

The TEM image of spent Ni/NiAlO_X-650 (Figure 7) indicated the nondisruptive dispersion of Ni⁰ active sites during the PHE hydrogenation, indicating the absence of sintering of Ni⁰ active sites. Meanwhile, the Raman spectrum of spent Ni/NiAlO_X-650 catalysts for 8 h (Supplementary Figure S2) rules out the formation of carbon deposition on the catalyst surface during PHE hydrogenation. Therefore, it was confirmed that the decrease in PHE hydrogenation capacity of studied catalysts occurred primarily due to the unfavorable increase of electron deficiency in Ni⁰ active sites.

The steric hindrance and competitive adsorption associated with the hydrogenation of s-OHP made it a major product in the main product of PHE hydrogenation Korre et al. (1995), Beltramone et al. (2008), Fu et al. (2015), Wang M et al. (2019), and thus, s-OHP conversion to PHP is regarded as the rate-determining step. Using s-OHP as the model compound, the real benefit of adequate electron deficiency in Ni⁰ sites in the hydrogenation reactions was highlighted. For the hydrogenation of s-OHP to PHP, the employed reaction conditions were as follows: 300°C, 5 MPa (H₂), and H₂/oil ratio of 600. The effects of internal and external diffusion were excluded by changing the mass flow rate and grain size, respectively. A carbon balance is attained over 96%. Figure 9 exhibited the change of s-OHP conversion and product selectivity with WHSV over Ni/NiAlO_X-650 catalysts, representatively. Three simultaneous reaction processes of s-OHP over the Ni/NiAlO_X-650 catalysts were predicted: saturation to PHP, isomerization to as-OHP, and dehydrogenation to THP and DHP. With the decrease of s-OHP, the content of dehydrogenation and isomerization yields in the product mixture was basically unchanged, but the content of PHP was gradually increased, which was almost equal to the conversion rate of s-OHP. These results proved that the conversion of s-OHP to PHP occurred through the hydrogenation route over Ni/NiAlO_X-650 catalyst.

The obvious reaction rate (r_obs) and the turnover frequency (TOF) in the performed s-OHP were calculated at the lower conversion (<20%) and given in Table 3. Ni/NiAlO_X-650 showed a higher r_obs (0.64 × 10⁻³ mol kg⁻¹ s⁻¹) and TOF (4.65 × 10⁻³ s⁻¹) than the other as-synthesized catalysts, and the activity trend of different catalysts was similar to that observed in the PHE hydrogenation. The higher efficiency of Ni/NiAlO_X-650 could be assigned to the abundance of Ni_octa sites after its calcination, which promoted the dispersion of Ni⁰ sites and provided the more suitable electron deficiency.

The TOF of Ni/NiAlO_x-650 in PHE hydrogenation was calculated and compared with that of the catalysts already reported in the literature (Table 3). Comparatively, Ni/NiAlO_X-650 (WHSV of 52 h⁻¹) accomplished a much higher average PHP selectivity (79.6%) than the other catalysts. For the r_obs and TOF calculation, WHSV was increased to 520 h⁻¹ to obtain the lower PHE conversion (<20%). At this condition, the calculated value of r_obs and TOF is 1.53 × 10⁻³ mol kg⁻¹ s⁻¹ and 14.64 × 10⁻ s⁻¹, respectively. These values were relatively higher than that reported for various catalysts, indicating the significance of Ni/NiAlO_X-650 possessing the suitable electron-deficient Ni⁰ sites.