Highly Selective Alkaline Oxide Promoted Ni/silica for the Production of Cyclohexanol/‐one from Lignin‐Derived Guaiacol

Catalytic hydrodeoxygenation (HDO) is a promising route to upgrade bio‐oils from plants or pyrolysis to basic chemicals. KA‐oil (cyclohexanol/‐one) is an example of a highly valuable feedstock for the industrial synthesis of caprolactam and perlon, which can potentially be produced via a completely sustainable route. This work investigates the alkaline promoter effects on Ni‐based catalysts for the selective HDO of lignin surrogate guaiacol using mesoporous Ni/MgO and co‐impregnated Ni‐MgO/SiO2 catalysts. Due to preferred demethoxylation by alkaline oxides and their accumulation on nickel corners, responsible for dehydration, conversion, and selectivity for KA‐oil rise. Furthermore, influences of operating parameters are investigated and optimized by design of experiments (DoE), while long‐term stability (24 h) and water resistance remain high. In a final proof of concept, the application of Ni‐CaO/SiO2 was the most promising material (U=93 %, S=86 %), demonstrating a sustainable route from lignin‐based bio‐oil to produce a bulk commodity for caprolactam synthesis.


Introduction
The use of renewable raw materials and energy sources is becoming increasingly important, especially due to the rising demand for fuels and basic chemicals, the natural limitation of fossil raw materials, and increasing geopolitical tensions. [1]For decades now, catalytic hydrodeoxygenation (HDO) has been one of the most promising fields in research to upgrade bio-oils from wood pyrolysis to green fuels and basic chemicals.However, the enormous potential of sustainable bio-based carbon sources is currently still used for thermal purposes, such as 80 million tons/year of lignin from the paper industry, which could be exploited as raw material in the chemical industry. [2,3]oday, most research is conducted in cheaper batch experiments, which complicates knowledge transfer for process scaleup and, consequently, for an economic continuous-mode application in sustainable processes. [4,5]Furthermore, for very complex systems such as the HDO of bio-oils, the often-used "one factor at a time" approach (OFAT) rarely leads to optimal operating conditions.In addition, to understand the influences of the different operating parameters, it is increasingly necessary to use Design of Experiments (DoE) to describe effects and their interactions in a quantifiable way. [4,6]dditionally, most research focuses on obtaining fuels through the complete HDO over bifunctional metal-acid catalysts. [7]In contrast, producing basic chemicals still containing oxygen by selective HDO is rarely considered, even though they are currently predominantly produced from fossil raw materials. [5,8]For instance, more than one million tons of KA-oil (cyclohexanol and cyclohexanone) are produced annually from BTX stream as an educt for ɛ-caprolactam, which serves in the polymer industry as a starting material for the production of polyamide 6 (perlon). [9]Therefore, the alternative generation of KA-oil from renewable, locally available carbon sources, such as bio-oils or fractions of the 80 million tons of chemically unused black bleach from paper production, is of particular interest to the chemical industry.
As the commonly used bifunctional metal-acid catalysts are unsuitable for selective HDO to KA-oil because of the undesirable dehydration of cyclohexanol to cyclohexene at acid centers, we recently investigated systematically the influence of low-acid metal-silica catalyst systems. [10]Due to geometric (hexagonal crystal symmetry) and energetic correspondence (center of d-band) for selective ring hydrogenation as well as their moderate oxophilicity, nickel catalyst systems are ideally suited for converting guaiacol to the incompletely deoxygenated products cyclohexanol and cyclohexanone (KA-oil). [10]In follow-up work, we focused on the influence of nickel dispersity and operating parameters on the reaction network.Here, we clarified that the primary reaction proceeds as structuresensitive reaction with phenol as an intermediate.Meanwhile, direct ring hydrogenation to 2-methoxycyclohexanol (2MC) should be avoided to increase the catalyst's activity and selectivity for KA-oil. [11]] N. Kretzschmar In the literature, binary catalyst systems with a metal oxide component such as MoO x , WO x , or ReO x are frequently discussed, while strongly alkaline metal oxides such as MgO or CaO are out of focus.[12,13] However, from the HSAB concept (hard and soft acids and bases), stronger alkaline functionality should support initial demethoxylation towards phenol.The successful application of Ni on MgO carrier for selective HDO to cyclohexanol has already been demonstrated by Nakagawa et al. [13] and Long et al. [14] However, in contrast to an easily scalable and controllable continuous and solvent-free operation, these studies were carried out in batch mode in a threephase system with water [13] or decalin [14,15] as an additional solvent.Furthermore, clumping of the magnesium oxide and leaching of the impregnated metal from the support was observed during liquid phase hydrogenation.
Therefore, this research aims to study the more sustainable continuous-mode operating, solvent-free, and highly selective HDO of lignin towards cyclohexanol and cyclohexanone (KAoil), using 2-methoxyphenol (guaiacol) as a model feedstock for lignin-derived bio-oils, the most abundant renewable raw material worldwide. [3]In addition to using mesoporous MgO as a carrier material, the co-impregnation of Ni/SiO 2 catalysts with varying MgO contents is evaluated.Using the most active Ni-MgO/SiO 2 material as an example, a reduced experimental design reveals the influence of the operating parameters T, p, and time on stream (ToS).Long-term stability over 24 h and the stability towards water are checked.Ultimately, the reason for increased conversion and high selectivity for KA-oil is investigated by directly converting the intermediate 2-methoxycyclohexanol (2MC) and varying the metal oxide component with rising alkalinity (ZnO, MgO, CaO) as final proof of concept.

Physicochemical characterization
Tab. 1 summarizes the surface areas determined by nitrogen physisorption and nickel fractions determined by ICP-OES analysis.The nickel dispersity and crystallite sizes determined by hydrogen pulse chemisorption are compared with the crystallite sizes estimated from P-XRD.The high specific surface area of the silica support material Aerosil380 (~320 m 2 /g) is maintained for all loaded catalyst materials.A specific surface area of ~65 m 2 /g is achieved for urea precipitation, comparable to the reported magnesium oxides produced by ammonia precipitation. [15]For surface-rich silica (Aerosil380) as carrier material, the surface area decreases with increasing MgO content due to the dilution effect.ICP-OES confirms the targeted Ni content of 10 wt.% in the reduced catalyst for all systems.Ammonia desorption (TPAD) shows no significant acidic properties for all materials except those with high MgO content of 10 wt.% and more.Due to the weak Lewis acidity of Si 4 + and Mg 2 + , the low signals can be attributed to the formation of magnesium silicates. [16](SI, Figure S12).
SEM images of the Ni-MgO/SiO 2 materials indicate no morphological changes due to co-impregnation, revealing typical agglomerates of small silica particles (SI, Figure S2).Platelet-like structures of more than 5 μm are formed for MgO by urea precipitation without Ni (SI, Figure S1).In contrast, the Ni/MgO forms fine sponge-like structures.Only Ni precipitated through urea (SI, Figure S3) reveals spherical particles with spiky, pointed growths that sinter with increasing calcination temperature, which explains the higher surface area of Ni/MgO compared to pure MgO.
The P-XRD analysis of all catalysts was performed before and after catalytic tests (SI, Figure S4 to S11).Based on the absence of a nickel oxide phase in P-XRD, it can be assumed that nickel oxide is completely reduced at pretreatment and reaction conditions.Meanwhile, consistent with literature reports, only MgO and no reduced Mg has been detected. [17]ue to partial reflex overlapping between MgO and Ni, particle size has not been estimated for MgO fractions � 10 wt.%. [14,15]As another consequence of their structural similarity, the formation of an intermetallic phase is likely, which is reported for Ni/ MgO. [14,15]Long et al. [14] varied nickel contents from 5 % to 20 % and detected a shoulder at ~62.5°2θ for Ni contents � 15 %.However, as a result of the lower Ni content in this study, no additional reflex is observed (Figure 1).The increasing reflex broadening with rising MgO content indicates a decreasing Ni crystallite size from ~20 nm for the Ni/SiO 2 material without MgO to ~16 nm for the Ni-MgO 2% /SiO 2 catalyst.In contrast, hydrogen pulse chemisorption measurements of the reduced systems indicate a decreasing hydrogen adsorption capacity with increasing MgO content and, thus, lower dispersity and larger Ni metal crystallite sizes (Table 1).[20] Temperature-programmed reduction, oxidation, and re-reduction (TPROR) up to 600 °C were recorded for all catalyst materials to mimic the conditions during pretreatment for catalytic tests (SI, Figure S13, and S14).For the Ni/SiO 2 material, a reduction max. at ~310 °C can be attributed to the reduction of NiO to Ni.The reduction of nickel-silica species causes an additional shoulder at higher temperatures. [21]he TPR of Ni/MgO prepared by urea precipitation has a small signal at ~230 °C and a larger signal at ~350 °C (SI Figure S13).Shifting the primary signal to higher reduction temperatures again indicates a stronger interaction (SMSI) and, thus, nickel(oxide) stabilization by a MgO phase. [22]For the Ni-MgO/ SiO 2 catalysts, a shift to higher reduction temperatures is also observed with increasing MgO content (see SI Figure S14).
In contrast to MgO as a support material, no additional lowtemperature TPR signal is detected for the systems on silica carrier.However, again there is a shoulder at a higher temperature due to SMSI. [21] deeper look into temperature-programmed desorption of CO 2 (TPCD) shows low intensity for all samples due to low amount of alkaline material, except sample Ni/MgO with MgO as pure carrier material (SI Figure S13).While unmodified silica support, sample Ni/SiO 2 and few additional amount of MgO (0.5 wt.%) show no significant TPCD signal, a rising amount of MgO (1.0-10 wt.%) or CaO and ZnO reveal a broad lowtemperature signal with desorption maximum at approx.200-250 °C.Ni-MgO (10 %) /SiO 2 and Ni-ZnO (1 %) /SiO 2 are special with a high-temperature peak at approx.300-400 °C.This can be attributed to less dispersed (bulk) metal oxide with stronger affinity to CO 2 and less to Ni.
X-ray photoelectron spectroscopy (XPS) after TPR and subsequent TPCD experiments reveals an elementary distribution with approx.3 wt.% Ni (1/3) and only fewer fraction of MgO and CaO below anticipated 1 wt.% on the surface (Table 2).It reveals a heterogeneous distribution of the metals on the surface.Additionally, a detailed analysis by means of high resolution XPS is done for the nickel and alkaline-metal components.Silicon and oxygen are neglected since their spectra are dominated by the unreacted SiO 2 matrix.The results for Ni2p are shown in Figure 2 and 3.They mainly show the expected oxidation of the metal Ni 2 + in a Ni(OH) 2 configuration according to the binding energy (854.8eV) and the satellite structure. [23]A slightly low energy shoulder at the Ni-metal position (852.7 eV) can be observed for all samples.An incomplete oxidation of the Ni centers could mean a NiÀ Si interaction which suppresses the oxidation process.The overall peak width and the height of this shoulder is increased for the samples treated by TPR.Non-homogenous chemical environment of Ni atoms due to further reduction could be the origin of this observation.These results in Figure 2 and 3  indicate an interaction between Si and Ni species with a broad spectrum due to sample inhomogeneity, while even after TPROR procedure not all surface Ni species remain completely reduced.From fitting, for all samples analyzed only 11-13 % remain as Ni 0 after reduction.Together with identification of pure Ni structures from TEM, a fast surface oxidation of highly reactive Ni particles prior to XPS analysis is plausible.The spectra of Mg and Ca of the alkaline metal oxides are presented in Figure 4 and 5.In each case only one species could be detected and the binding energy is in good agreement with the MgO and CaO literature values.The both samples after TPR treatment show again an increased peak width due to the inhomogeneity of the direct chemical environment and a slightly shift to higher binding energies.The presence of positive charges are normally the origin of such shifts.It is likely that the further reduction during TPR leads to a higher presence of Ni-cations which interact with the alkalinemetal oxides.
In summary, the physicochemical characterization reveals a strong chemical interaction of low amounts of MgO with partially reduced Ni x surface species on the nanoparticles.

Catalytic properties of Ni/MgO and Ni-MgO X /SiO 2
Figure 6 illustrates the conversion of guaiacol with ToS for the systems Ni/SiO 2 , Ni/MgO and the Ni-MgO x /SiO 2 samples with different MgO content between 0.5 wt.% and 10 wt.%.The conversion for Ni-MgO x /SiO 2 (IIIÀ V) is higher compared to the Ni/SiO 2 (I) except the system with 10 wt.% MgO (VI).However, the conversion decreases with increasing MgO content.Therefore, the highest conversion is achieved by the Ni-MgO 0.5 /SiO 2 system (III) with 0.5 wt.%, i. e., the lowest magnesium oxide content.As reference, there is no significant conversion using   The dotted blue line marks the MgO binding energy according to literature. [24]gure 5. XPS results for Ca(2p) for sample Ni 10 % CaO 1 % /SiO 2 (211) after calcination (R) and after subsequent TPROR and TPCD analysis.The dotted blue line marks CaO according to literature. [25]nmodified commercial silica or unmodified MgO, i. e. minor residual acidity or alkalinity has no significant influence (see SI, Figure S17 and S18).
For further analysis, the selectivities for KA-oil, phenol, 2MC, and cyclohexane for the different systems are displayed in Figure 7.While there is only a negligible change in selectivity for phenol, adding MgO drastically changes the selectivities of the target product KA-oil and the by-products 2MC and cyclohexane.The ratio of the individual components cyclohexanol and cyclohexanone is discussed in more detail in the next section (see Figure 8).As discussed for the conversion, the selectivity for KA-oil increases for all Ni-MgO x /SiO 2 (IIIÀ V) materials except the system with 10 wt.% of MgO (VI) compared to the Ni/SiO 2 (I).However, the selectivity decreases again with increasing MgO content.With the addition of 0.5 wt.% MgO, the KA-oil selectivity rises from ~52 % (Ni/SiO 2 ) to ~78 % (Ni-MgO 0.5 % /SiO 2 ), thus exceeding that of the Ni/MgO system of ~75 %.
Furthermore, the coke content decreases from 7.1 wt.% (Ni/ SiO 2 , I) to 4.5 wt.% (Ni/MgO, II), whereas the coke content for the unloaded support materials increases from 11.9 wt.% (SiO 2 ) to 17.4 wt.% (MgO).Therefore, as observed from physicochemical characterization, a synergistic effect between Ni and MgO and a strong interaction between MgO and the feed guaiacol are concluded, although no significant hydrogenation reaction is observed.The slight increase in (Lewis) acidity by higher MgO contents does not promote dehydration of cyclohexanol, as cyclohexane accumulation is not observed.The reason for the increased activity and selectivity for Ni-MgO x /SiO 2 (IIIÀ V) will be discussed in more detail following the optimization of the operating parameters.
For comparison with literature reports, please find calculated turn-over frequencies in the supporting information, Table S1 and S2.However, reports about continuous operating conversion of guaiacol are rare and, furthermore, often use strongly differing test parameters (see Table S3).

Investigation of the influence of the operating parameters
The following studies were carried out using the most promising system, Ni 10 % -MgO 0.5 % /SiO 2 .The examination of longterm stability is performed for 24 h with hourly sampling (SI, Figure S19).The conversion decreases almost linearly with ToS, from ~93 % to ~84 % (0.45 % per h).This is still much slower compared to Ni 10 % /SiO 2 , for which the conversion decreases from ~80 % to ~60 % (0.83 % per h).The selectivity for hydrogenated ring products remains at a high level of > 90 % over 24 h ToS, with the selectivity for phenol remaining low � 1 %.The low selectivity for cyclohexane of ~5 % (2 h ToS) continues to decrease and remains < 2 % from 7 h ToS.In contrast, the selectivity for 2MC increases with ToS from ~12 % (2 h ToS) to ~23 % (24 h ToS) due to the selective coking of MgO, responsible for the increase in activity and selectivity for KA-oil.The strong coking of the unmodified MgO carrier (17.4 wt.% after 5 h ToS) supports this hypothesis.
Everything considered the Ni-MgO 0.5 % /SiO 2 remains more active and selective for KA-oil after 24 h ToS compared to Ni/ SiO 2, even at the beginning of the reaction (2 h ToS).
In addition to long-term catalytic tests, stability against cofeeding of water was investigated due to the high water content of real bio-oils (SI, Figure S20). [26]As a result, the coke formation of the catalyst after 5 h ToS decreases significantly from ~7 wt.% to ~4 wt.%.The overall result is a much more stable system.Meanwhile, the conversion drops from ~93 % (without H 2 O) to ~87 % (additional H 2 O, 50 : 50 (Vol.-%)H 2 O: guaiacol), while the selectivity for ring-hydrogenated products remains high.KA-oil, 2MC, and cyclohexane selectivities change slightly: The selectivity for phenol increases from < 1 % (without H 2 O) to ~4 % (with H 2 O), which can be attributed to the competitive adsorption of water and hydrogen on the catalyst surface, reducing the availability of hydrogen and, thus, the rate of hydrogenation.Concerning the target product, the selectivity for cyclohexanone increases from ~5 % (without H 2 O) to ~10 % (with H 2 O), which can be attributed again to the competing adsorption of H 2 O and H 2 .Moreover, DFT calculations indicate a stabilization of the keto form bound to the metal surface [27] by H 2 O and a destabilization of the C Ar À OMe bond, [28] which favors cyclohexanone formation.
Due to the complex system, the commonly used "one factor at a time" (OFAT) approach rarely leads to optimal operating parameters for bio-oil conversion, which is why the following optimization is based on DoE. [4]A determinant (D)-optimized design was chosen, whereby the five factors (temperature, ToS, gauge pressure, _ V H 2 , _ V N 2 ) and their six two-way interactions are determined by maximizing the determinant of the information matrix with the smallest possible number of 11 individual experiments and p-values below 0.005 (SI, Table S8 and S9). [29]he results of the developed prediction models are illustrated in Figure 8 for the conversion and selectivity of KA-oil and its individual components, cyclohexanol, and cyclohexanone.Based on the predictive model, a maximum conversion and selectivity for cyclohexanol at a temperature of 250 °C, a ToS of 2 h, and a carrier gas flow of hydrogen of 10 L/h (no nitrogen) at a pressure of 15 bar(g) is predicted to be ~95 % and 74 %; experiments confirm (93 � 2 % and 73 � 1 %).The maximum conversion and cyclohexanol selectivity can be attributed to the promotion of the hydrogenation activity, which finds its optimum at lower temperatures due to the exothermic nature of the hydrogenation reaction and the competing hydrogen and nitrogen adsorption. [30]From this trend the only mass transport limitation observed is due to the availability of hydrogen.
The decreasing conversion observed with increasing ToS results from the deactivation of nickel particles due to sintering and coking. [4,31]Furthermore, a negative interaction between temperature and ToS is observed, i. e., higher temperature leads to faster and stronger deactivation caused by inhibited hydrogenation and increasing accumulation of aromatic intermediates on the catalyst surface.For the carrier gas flows, a strong interaction can be observed due to the competing adsorption of hydrogen and nitrogen, favoring (H 2 "; N 2 #) or inhibiting (N 2 "; H 2 #) the hydrogenation.
In contrast, the maximum selectivity for KA-oil of ~80 % is predicted by reducing the pressure to 10 bar(g) and experimentally confirmed at 81 � 1 %.The slight increase in selectivity for KA-oil from 78 % to 81 % by reducing the pressure is a consequence of lowering the hydrogenolysis activity of Ni, i. e., methane formation from ~7 % to ~5 %.
Contrary, maximum selectivity for cyclohexanone of 16 % is predicted and experimentally confirmed at 300 °C, a ToS of 2 h, a carrier gas flow of 10 L/h hydrogen and 5 L/h nitrogen, and at 15 bar(g).The higher temperature of 300 °C and the dilution of the carrier gas stream with 5 L/h nitrogen reduce the availability of hydrogen on the catalyst surface.As a result, the hydrogenation activity decreases, and cyclohexanone, formed as an intermediate, is converted more slowly to cyclohexanol.However, an additional reduction of the hydrogen availability by reducing the hydrogen flow to 5 L/h results in a too substantial decrease of the hydrogenation activity, whereby phenol is preferentially formed with a selectivity of ~50 % (SI, Figure S21).This is why the interactions in the description model cannot be neglected, and the DoE is superior to the OFAT approach.

Investigation of reasons for increased activity and selectivity
Based on our recent studies on Ni/SiO 2 systems and the control of the reaction network, an increase in conversion is possible by (i) promoting the conversion of 2MC, (ii) inhibiting the reaction pathway with 2MC as intermediate, and (iii) promoting the initial demethoxylation reaction of guaiacol to phenol, since 2MC reacts very slow and thus blocks the active sites.Furthermore, the target product KA-oil's selectivity can also increase by (iv) inhibiting the dehydration of cyclohexanol to cyclohexene (Scheme 1).
In the first step, Figure 9 displays the direct conversion of 2MC over Ni-MgO 0.5 /SiO 2 compared with our recent results for Ni/SiO 2 .The conversion decreases from ~30 % (Ni/SiO 2 ) to ~21 % (Ni-MgO 0.5 /SiO 2 ).In contrast, the selectivities for KA-oil increase from 36 % (Ni/SiO 2 ) to 44 % (Ni-MgO 0.5 % /SiO 2 ) and for methane from ~26 % (Ni/SiO 2 ) to ~33 % (Ni-MgO 0.5 % /SiO 2 ).The selectivity for cyclohexane drops significantly from ~13 % (Ni/ SiO 2 ) to ~3 % (Ni-MgO 0.5 % /SiO 2 ), which explains the overall increase in selectivity for KA-oil.Due to the decreasing tendency of irreversible dehydration of cyclohexanol to cyclohexene, which is rapidly hydrogenated to cyclohexane, more cyclohexanol accumulates, which is a component of KA-oil. [32]Based on the lower conversion for the reaction with 2MC, the increase in guaiacol conversion by adding MgO to the Ni/SiO 2 system is not due to (i) the promotion of the conversion of 2MC.The increased selectivity for KA-oil and lower selectivity for cyclohexane indicate (iv) the inhibition of the dehydration of cyclohexanol.
Consequently, the higher activity of Ni-MgO 0.5 % /SiO 2 in the conversion of guaiacol is based on either the (ii) inhibition of the reaction pathway with 2MC as an intermediate or (iii) promoting the initial demethoxylation reaction of guaiacol to phenol.If (ii) the ring hydrogenation had been inhibited, this would have also led to an inhibition of the ring hydrogenation of phenol and, therefore, to an accumulation, which is not observed (Figure 7).
As a result, (iii) promoting the initial demethoxylation reaction of guaiacol to phenol is the main cause of the increased activity.Since the formation of 2MC and phenol from guaiacol compete, the promotion of reaction (iii, demeth-oxylation) to phenol also leads to a lower dominance of reaction (ii, ring hydrogenation) to 2MC, which accounts for the lower yield for 2MC.The high activity of Ni/MgO (without silica) has also been attributed by Long et al. [14] and Li et al. [15] to the interaction of the phenolic hydroxyl group with MgO, weakening the C Ar -OCH 3 bond and thereby favoring demethoxylation (iii) and, thus, even coke formation (compare 17 wt.% for MgO). [13,33]The desorption of cyclohexanol from the catalyst surface is easier compared to phenol because of a decreasing acidity of the hydroxyl group due to the ring hydrogenation.
This hypothesis is tested by varying the metal oxide component in a final proof of concept.A metal oxide content of 1 wt.% was chosen to exclude possible inhomogeneities due to lower content.For this purpose, the influence of CaO 1 % as a stronger base and ZnO 1 % as a weaker base and a dehydrogenating component are investigated.TEM analysis and EDS mapping for samples with 1 wt.% MgO and 1 wt.% CaO confirm close proximity between reduced, crystalline Ni 0 particles and alkaline oxide promoters (see Figure 10, 11 and supporting Figure S22, S23, S24 for HRTEM images and EDS spectra).No core-shell particle formation was observed, but rather an agglomeration of Ni and MgO nanoparticles on the nm-scale.The characteristic lattice fringes of Ni and MgO are typically seen in HRTEM images of the Ni 10 % MgO (1 %) /SiO 2 sample (see Figure 10), whereas lattice fringes with d-spacings of 0.24 (MgO), 0.203 (Ni) and 0.17 nm have been observed for overlapping particles in the Ni 10 % CaO (1 %) /SiO 2 sample (Figure S25).Nanoparticles are stable against beam damage from HRTEM (Figure S26, S27).
If, as suspected, the reason for the increased activity is (iii) an interaction of the phenolic hydroxyl group with the alkaline center, CaO (pK B = 1.37) would lead to higher conversion and high selectivity for KA-oil compared to MgO (pK B = 1.8). [34]owever, if the reason is (ii) the inhibition of ring hydrogenation, higher conversion and high selectivity for KA-oil would be obtained with ZnO (pK B = 5.1) as the dehydrogenating component. [34]From Tab. 1, the Ni-MgO 1 % /SiO 2 system shows similar physicochemical properties compared to Ni-CaO 1 % /SiO 2 and Ni-ZnO 1 % /SiO 2 .Furthermore (Figure 12), a high conversion of ~93 % is achieved for Ni-CaO 1 % /SiO 2, comparable to Ni-MgO 1 % /SiO 2 .In contrast, the hydrogenation with Ni-ZnO 1 % /SiO 2 is inhibited, resulting in only ~23 % conversion.
CaO increases the selectivity for KA-oil from 36 % (Ni/SiO 2 ) to 79 % (Ni-MgO 1 % /SiO 2 ) up to 86 % (Ni-CaO 1 % /SiO 2 ), leading to the highest selectivity for KA-oil in this study (Figure 12).There is a significant increase in yield of + 40 %.In contrast, Ni-ZnO 1 % / SiO 2 leads to a low selectivity for KA-oil of 16 % but a muchincreased selectivity for phenol of ~16 % due to the dehydrating zinc oxide component.Compared to the Ni/SiO 2, all systems resulted in a lower selectivity for 2MC because of the promotion of the demethoxylation reaction for MgO and CaO and the inhibition of ring hydrogenation for ZnO.
Again, this proves (iii) that promoting the initial demethoxylation reaction of guaiacol by adding a strong (Lewis) base metal oxide is the crux of enhancing the yield of KA-oil.
Furthermore, compared to Ni/SiO 2, a lower selectivity for cyclohexane is observed.For ZnO, this is caused by the  decreased formation of the intermediate cyclohexanol on the route to cyclohexane.For MgO and CaO, despite the increased appearance of cyclohexanol, the selectivity for cyclohexane is reduced from 22 % (Ni/SiO 2 ) to 11 % (Ni-MgO 1 % /SiO 2 ) and 3 % (Ni-CaO 1 % /SiO 2 ).Due to the minor acidity of the chosen support material Aerosil380 (SI, Figure S12), the inhibition of dehydra-tion cannot be justified only by the neutralization of the acidic centers by the Lewis alkaline metal oxide, as observed in literature for Pt/TiO 2 and Pt-MgO/TiO 2 . [14,17,35]As recently demonstrated, the dehydration of cyclohexanol to cyclohexene and the hydrogenolysis to methane proceeds preferentially at particularly active centers of Ni such as corners, edges, and  defects in the metal crystal, i. e., it is favored by smaller metal particles. [27,36]However, based on the P-XRD results, nickel particle sizes even decrease (Figure 1, Table 1).Therefore, the unexpected decrease in dehydration indicates the enrichment of the MgO and CaO at these particularly active sites, neutralizing them and inhibiting the dehydration of cyclohexanol (Scheme 2).This hypothesis is confirmed by the lowered hydrogen adsorption capacity in the hydrogen chemisorption measurements, based on a strong interaction analogous to SMSI, [37] thus underestimating the dispersity (Table 1).Furthermore, the interaction of metal oxides with metal particles on their surface as an extension of strong metal support interaction (SMSI) phenomena has recently been observed by Lunkenbein et al. [18,19] The conclusions of the investigations are finally visualized with pictograms in Scheme 2. Selectivity by use of alkaline oxide promoters has been achieved by process parameter adaption and modifying the reactant-catalyst interaction with respect to the three crucial steps: (i) the preferred demethoxylation of guaiacol to phenol, (ii) the selective ring hydrogenation from phenol to cyclohexanone and -ol, and (iii) the inhibited subsequent dehydration of cyclohexanol to cyclohexene, formerly promoted by acid groups or uncovered Ni metal edges.

Conclusions
This research aimed to investigate the alkaline promoter effect on Ni-based catalysts, enabling the selective HDO of guaiacol as a model compound for lignin, one of the most abundant sustainable raw materials worldwide.We, therefore, targeted the sustainable production of KA-oil, an essential feedstock in the polymer industry for polyamide 6 (perlon).
We compared Ni-MgO x /SiO 2 with varying MgO contents (0.5 wt.% to 10 wt.%) with Ni/SiO 2 and Ni/MgO.Using the most promising material (Ni-MgO 0.5 % /SiO 2 ), the stability to water and ToS (24 h) was confirmed, and the cause of the promoter effect was investigated based on the direct conversion of 2MC and verified in a final proof of concept by varying the metal oxide component (ZnO, MgO, CaO).
Based on our investigations on alkaline modified Ni/SiO 2 with proven close proximity between Ni and alkaline oxides, the improved conversion and selectivity for KA-oil can be explained by promoting the initial base-catalyzed demethoxylation of guaiacol to phenol rather than by the promoted conversion of 2MC or the inhibition of the 2MC route.Furthermore, the increased selectivity can be explained by the prevention of the dehydration of cyclohexanol at particularly active nickel centers, such as corners, edges, and defects, by close proximity of Ni defects and MgO.These have been proven by variation of the alkaline component (ZnO, MgO, CaO), where the strongest base (CaO) resulted in the highest conversion (93 %) and selectivity for KA-oil (86 %), which represents a relative improvement in yield of KA-oil by + 40 % compared to unpromoted Ni/SiO 2 with additionally improved long-term stability.
Compared to the MgO support, silica support revealed that a sensitive adjustment of alkaline component and nickel ratio is crucial for improved catalytic activity, selectivity, and durability by preventing an overload of the active nickel surface with alkaline oxide.We further quantified the influence of the reaction conditions and optimized them by DoE, which enables a selective production of KA-oil based on sustainable biomass (lignin) utilization.
In contrast to the frequently investigated use of bifunctional acid-metal catalysts for fuel production, we have focused on the rarely investigated generation of basic chemicals using strongly alkaline metal oxides. [12,13]To the best of our knowledge, the superior Ni-MgO/SiO 2 system and the general application of CaO as a stronger alkaline component have not been described in the literature for HDO so far.Furthermore, in contrast to the frequently used batch experiments and OFAT approaches, we have focused on the continuous operation mode and the usage of DoE to enable the development of a sustainable and economically applicable process. [4]

Experimental Section
The silica catalysts were loaded using the incipient-wetness method.The synthesis of the mesoporous MgO was carried out via urea precipitation.All chemicals used, and a detailed description of the synthesis procedures are given in the SI.
Catalytic investigations were carried out in a continuously operating fixed-bed tubular reactor under integral conditions.The analysis of the product spectrum was carried out at a GC-FID using the method we have recently published. [38]A detailed description of the experimental procedure is given in the SI.

Figure 4 .
Figure 4. XPS results for the Auger line Mg(KLL) for sample Ni 10 % MgO 1 % /SiO 2 (211) after calcination (R) and after subsequent TPROR and TPCD analysis.The dotted blue line marks the MgO binding energy according to literature.[24]

Figure 7 .
Figure 7.Comparison of KA-oil, phenol, 2MC, and cyclohexane selectivities over Ni/SiO 2 , Ni/MgO, and Ni-MgO x /SiO 2 with varying amounts of MgO (x).The dashed smoothing is based on a spline with λ = 0.05 and the confidence interval.

Figure 8 .
Figure 8. Results of the experimental design (DoE).Influence of conversion and selectivities for KA-oil and its components cyclohexanol and cyclohexanone by the operating parameters with their optimum settings.

Scheme 1 .
Scheme 1. Overview of the possible origins of the activity (i to iii) and selectivity for KA-oil (i to vi) enhanced by MgO based on the reaction network of the conversion of guaiacol on modified nickel-silica catalyst systems.

Figure 10 .
Figure 10.A high-resolution transmission electron microscopy (HRTEM) image of Ni-MgO (1 %) /SiO 2 specimen.FFT 1 is taken from an individual Ni nanoparticle evidenced by Ni (111) lattice plane with the d-spacing of 0.203 nm.FFT 2 reveals two lattice fringes of d = 0.24 and 0.203 nm which can be explained as the (111) lattice plane of Ni nanoparticle overlapping with the (111) lattice plane of MgO.The HRTEM image confirms the suggested spatial proximity of Ni nanoparticles and MgO nanoparticles.

Figure 11 .
Figure 11.High-angle annular dark-field image of Ni-MgO (1 %) /SiO 2 specimen taken by (a) scanning transmission electron microscopy and elemental mapping images of (b) Ni, (c) Mg, (d) Si and (e) O reveals regions with higher Mg content located very close to Ni nanoparticles.No core-shell particle formation was observed, but rather a agglomeration of Ni and MgO nanoparticles on the nm-scale.

Table 1 .
clearly Summary of physicochemical characterization of nickel systems with Aerosil380 and magnesium oxide as support material.The synthesis was carried out by precipitation with urea.[b] Me = Mg, Ca or Zn.[c] Overestimation of crystallite size by H 2 pulse chemisorption likely due to accumulation of metal oxide on the nickel particle surface.[d] Due to the superposition of the Ni and MgO reflections, a calculation is not performed.

Table 2 .
Weight fraction table from XPS analysis, additional data fitting for Ni2p signal to calculate ratio of Ni 0 (one signal) to Ni 2 + (3 distinct signals).