Depolymerization and Hydrogenation of Organosolv Eucalyptus Lignin by Using Nickel Raney Catalyst

: The use of lignocellulosic biomass to obtain biofuels and chemicals produces a large amount of lignin as a byproduct. Lignin valorization into chemicals needs efﬁcient conversion processes to be developed. In this work, hydrocracking of organosolv lignin was performed by using nickel Raney catalyst. Organosolv lignin was obtained from the pretreatment of eucalyptus wood at 170 ◦ C for 1 h by using 1/100/100 ( w / v / v ) ratio of biomass/oxalic acid solution (0.4% w / w )/1-butanol. The resulting organic phase of lignin in 1-butanol was used in hydrogenation tests. The conversion of lignin was carried out with a batch reactor equipped with a 0.3 L vessel with adjustable internal stirrer and heat control. The reactor was pressurized at 5 bar with hydrogen at room temperature, and then the temperature was raised to 250 ◦ C and kept for 30 min. Operative conditions were optimized to achieve high conversion in monomers and to minimize the loss of solvent. At the best performance conditions, about 10 wt % of the lignin was solubilized into monomeric phenols. The need to ﬁnd a trade-off between lignin conversion and solvent side reaction was highlighted.


Introduction
The need to drastically reduce the exploitation of nonrenewable resources in favor of renewable ones is widely recognized. The acceleration of climate change due to the greenhouse gases released into the atmosphere requires the rules on which our economy is based to be rewritten. Lignocellulosic biomass is one of the most studied renewable sources for sustainably replacing petroleum in obtaining fuels and chemicals.
Lignin is a complex three-dimensional amorphous polymer composed of three main phenylpropane units (coniferyl, sinapyl, and p-coumaryl alcohols) connected by β-O-4 , β-5 , β-β , β-1 , 5-5 , and 5-O-4 linkages in addition to several different minor phenolic compounds, and its structure and composition is dependent on the plant. The complexity and degree of polymerization make it impossible to find two lignins with the same sequence of phenyl units; for this reason, it is more appropriate to refer to "lignins" [6,7]. Due to the recalcitrance of lignin to be degraded, it is typically a waste that is combusted to produce heat and power. On the other hand, lignin is the most abundant renewable source composed of aromatic units in nature and it is worth finding a sustainable way to use it as feedstock to produce intermediates for the chemical industry and high-value liquid fuels.
For this reason, in recent years, there has been an increased interest in lignin, evidenced by the exponential increase in scientific publications related to its valorization [8].
Metal catalysts and supported metal catalysts were employed in the presence of hydrogen to convert lignin into monomers in hydrogenolysis and hydrogenation processes. The most used catalytic systems are Pd/C, Ni/C, Pt/C, Pt/Al 2 O 3 , Cu-based porous oxides, supported NiW, NiMo, Ru-based materials, Ni-Ru, Ni-Rh, Ni-Pd, Ni-Au, WP, Ru x Ni1 −x /SBA-15, Al-SBA-15, MoO x /CNT, and S 2 O 8 2− -KNO 3 /TiO 2 [9,21,25,[31][32][33][34][35]. The Ni/Al alloy was utilized as starting material to produce a catalyst in novel hydrogenolysis: this was obtained exposing Ni, the active phase, by etching Al atoms in an alkaline aqueous solution [36]. Lignin depolymerization and product hydrogenation were tested using Ni/SiO 2 catalyst under mild condition [37]. Formic acid, as an in situ hydrogen source, and Ru/C as a catalyst in supercritical ethanol were tested to depolymerize lignin into a high-quality bio-oil; the combination of Ru/C and formic acid also resulted in a significant reduction of the oxygen content in the products [38]. New catalysts such as S 2 O 8 2− -KNO 3 /TiO 2 were synthesized and studied for the hydrocracking of lignin to improve the efficiency of the process and selectively target the formation of products [39]. The competition between hydrogenolysis and hydrodeoxygenation with the repolymerization was investigated in metallic catalysts of CrCl 3 and Pd/C [40]. The hydroconversion of a wheat straw soda lignin with a NiMoS/Al 2 O 3 catalyst was studied to resolve the reaction network [41].
A. McVeigh et al. investigated the depolymerization of lignin over a Pt/alumina catalyst, and they reported a clear solvent effect [42]. The measured high yield of monomers was attributed to the inhibition of repolymerization by methanol. The depolymerization reaction was also studied over Rh/alumina and Ir/alumina catalysts, indicating that there was considerable potential in optimizing the solvent system for lignin depolymerization.
Nickel Raney has been widely used as a catalyst for the hydrothermal conversion of lignin. Struven and Meier tested several nickel Raney catalysts in aqueous media with organosolv lignin at high temperature and pressure under H 2 atmosphere to produce selectively highly reactive phenol and its para-or ortho-alkylated derivatives in high yields [43].
The aim of this work was to use this catalyst to improve the economic sustainability of the lignin conversion into monomeric phenols. Thermal hydrocracking of organosolv lignin, originating directly from the pretreatment stream, was performed in alcoholic media under hydrogen atmosphere and mild conditions of temperature and pressure. A donor hydrogen solvent other than water was tested with this catalyst to explore its potential.

Lignin Extraction and Reaction System
The organosolv lignin was obtained from ground eucalyptus treated with 1-butanol and oxalic acid at temperature of 170 • C for 60 min. The extraction was carried out in 300 mL Parr reactor loaded with 5 g of eucalyptus dry matter, 50 mL of 1-butanol, and 50 mL of oxalic acid solution 0.167 M, under stirring condition at 400 rpm. The obtained slurry was filtered and the cellulosic fraction was washed with water. The two liquid phases were separated by a separator funnel and the butanol phase (lighter) was used for the Processes 2021, 9, 1093 3 of 18 hydrocracking tests. The dry matter in the organic fraction was determined by keeping the sample at 60 • C overnight; it was 2.1% w/w. All used reagents were of pure grade.
Hydrocracking tests were performed using 300 mL Parr reactor (from Büchi, model Limbo-li, Essen, Germany) equipped with a 0.3 L vessel. About 30 g of organosolv lignin solution was added with 10 g 1-butanol containing a nickel Raney catalyst as a slurry 10% w/w of solid in alcoholic solution (purchased from Sigma, St. Louis, MO, USA). After swiping the freeboard volume with H 2 to remove air, the reactor was pressurized at 5.5 bar with H 2 at room temperature. In the experimental runs, two parameters were changed: temperature, in the range 260-300 • C, and time, in the range 30-90 min. At the end of the runs, the reactor was cooled down, and solution and gases were recovered and analyzed. The full procedure is schematized in Figure 1, while Figure 2 shows the Parr reactor and the experimental setup.

Analytical Methods
Lignin solution was analyzed by HPLC (Agilent 1100 series) with the GPC columns SUPELCO TSKgel-G4000HHR, TSKgel-G3000HHR, and TSKgel-G2500HHR to determine molecular weights and GC-MS (Agilent 6890N equipped with MSD5975B) with Agilent 5MS 30 m to analyze the molecules produced after hydrocracking. Guaiacol was used as internal standard for quantitative analysis. Produced gases were collected in bags at the end of the experimental run and analyzed by Micro GC Agilent 3000 series with PLOTQ and Molsieve columns.

Experimental Design
The hydrocracking tests were set up following an experimental design (Design expert 10) where 2 parameters were varied: By using these options, the software algorithm produced 10 runs with 2 center points ( Figure 3).

Results and Discussion
The extracted lignin, dissolved in butanol, constituted 27.5 wt % of the starting dry eucalyptus, with a purity of up to 70% estimated by thermogravimetric analysis (TGA).
Thermal hydrocracking of organosolv lignin, catalyzed by nickel Raney, converts lignin and part of the solvent into low-weight molecules and gases. The 1-butanol, used as solvent, has a critical pressure of 44 ± 1 bar and a critical temperature of 289 ± 2 • C [54]; all tests at 260 and 280 • C were performed below critical temperature and above critical pressure, whereas tests at 300 • C were carried out under supercritical conditions. Among the monomers derived from the lignin and determined by CG, the main components that were detected as responses to different testing conditions are reported in Table 1. These constituted the input data for the analysis of the experimental design and the following process optimization. The design was analyzed for each response; surfaces that correlated the predicted response to the variables were shown, allowing the determination of trends and the best conditions to use in order to obtain the desired values.

Response 1: Diphenylmethane 4-ethyl
In Figure 4a the correlation between the actual values (Table 1) and those predicted by the analysis is shown. In this case, the correlation factor is R 2 = 0.7120; therefore, any conclusion about the trend is qualitative. In Figure 5, the surface graph shows the variation of the response with the examined variables. The analysis of the diphenylmethane shows that the maximum production is predicted at 280 • C and at a lower time (30-35 min). At these conditions, 3% of lignin is expected to be converted into this component.

Response 2: 2,4-Dimethyl-3-(methoxycarbonyl)-5-ethylfuran
In Figure 4b, the correlation between the actual values (Table 1) and those predicted by the analysis is shown. Similarly to above, R 2 = 0.7604; therefore, any conclusions about the trend are qualitative. In Figure 6, the surface graph shows the variation of the response with the examined variables. The analysis shows that max production is predicted at 284 • C and at the highest time (90 min). At these conditions, about 2.2% of this component is predicted to be produced.

Response 3: 1,2,3-Trimethoxy Benzene
In Figure 4c, the correlation between the actual values (Table 1) and those predicted by the analysis is shown. In Figure 7, the surface graph shows the variation of the response with the examined variables. The analysis shows that the max production is predicted at 296 • C and at the highest time (90 min). At these conditions, about 1.7% is expected to be converted into this component.  Table 1).

Response 2: 2,4-Dimethyl-3-(methoxycarbonyl)-5-ethylfuran
In Figure 4b, the correlation between the actual values (Table 1) and by the analysis is shown. Similarly to above, R 2 = 0.7604; therefore, any co the trend are qualitative. In Figure 6, the surface graph shows the variation with the examined variables. The analysis shows that max production is p °C and at the highest time (90 min). At these conditions, about 2.2% of th predicted to be produced.

Response 2: 2,4-Dimethyl-3-(methoxycarbonyl)-5-ethylfuran
In Figure 4b, the correlation between the actual values (Table 1) and those predicted by the analysis is shown. Similarly to above, R 2 = 0.7604; therefore, any conclusions about the trend are qualitative. In Figure 6, the surface graph shows the variation of the response with the examined variables. The analysis shows that max production is predicted at 284 °C and at the highest time (90 min). At these conditions, about 2.2% of this component is predicted to be produced.  In Figure 4c, the correlation between the actual values (Table 1) and those predicted by the analysis is shown. In Figure 7, the surface graph shows the variation of the response with the examined variables. The analysis shows that the max production is predicted at 296 °C and at the highest time (90 min). At these conditions, about 1.7% is expected to be converted into this component.

Response 4: Total of Main Monomeric Compounds
In this analysis, we considered the sum of the identified monomers by GC-MS database (over 80% matching), including the first three responses plus guaiacol, syringol, pethylguaiacol, 2-methoxy, and 4-propyl phenol (the last four compounds were produced with an average contribution of 0.05%, 0.1%, 0.3%, and 0.3%, respectively). In Figure 4d, the correlation between the actual values (Table 1) and those predicted by the analysis is shown. In this case, the correlation factor R 2 is 0.7145; therefore, any conclusions about the trend are qualitative. In Figure 8, the surface graph shows the variation of the response with the examined variables. The analysis shows that the max production is predicted at 296 °C and at the highest time (90 min). At these conditions, the production of the main monomers is predicted to be about 9% (as wt % with respect to the solubilized lignin).

Response 4: Total of Main Monomeric Compounds
In this analysis, we considered the sum of the identified monomers by GC-MS database (over 80% matching), including the first three responses plus guaiacol, syringol, p-ethylguaiacol, 2-methoxy, and 4-propyl phenol (the last four compounds were produced with an average contribution of 0.05%, 0.1%, 0.3%, and 0.3%, respectively). In Figure 4d, the correlation between the actual values (Table 1) and those predicted by the analysis is shown. In this case, the correlation factor R 2 is 0.7145; therefore, any conclusions about the trend are qualitative. In Figure 8, the surface graph shows the variation of the response with the examined variables. The analysis shows that the max production is predicted at 296 • C and at the highest time (90 min). At these conditions, the production of the main monomers is predicted to be about 9% (as wt % with respect to the solubilized lignin).

Response 5: Molecules Derived from Solvent
In this response, we considered the sum of the identified monomers by GC-MS database derived from solvent: butyl butanoate, 2-ethyl-2-hexenal, 2-ethyl-1-hexanol, and 7-methyl-4-octanone. The amount, reported in Table 1, is the percent mass yield of loaded solvent. In Figure 4e, the correlation between the actual values (Table 1) and those predicted by the analysis is shown. In this case, the correspondence factor R 2 is 0.9336, and therefore the considerations about the trend are quantitatively significant. In Figure 9, the surface graph shows the variation of the response with the examined variables. The analysis shows that the max production is predicted at 300 • C and at the highest time (90 min). At these conditions, about 0.5 wt % of the solvent is converted in these molecules.

Response 5: Molecules Derived from Solvent
In this response, we considered the sum of the identified monomers by GC-MS database derived from solvent: butyl butanoate, 2-ethyl-2-hexenal, 2-ethyl-1-hexanol, and 7methyl-4-octanone. The amount, reported in Table 1, is the percent mass yield of loaded solvent. In Figure 4e, the correlation between the actual values (Table 1) and those predicted by the analysis is shown. In this case, the correspondence factor R 2 is 0.9336, and therefore the considerations about the trend are quantitatively significant. In Figure 9, the surface graph shows the variation of the response with the examined variables. The analysis shows that the max production is predicted at 300 °C and at the highest time (90 min). At these conditions, about 0.5 wt % of the solvent is converted in these molecules.

Optimization
The optimization process has taken into account that our goal was to maximize the production of low-molecular-weight molecules from the lignin and, at the same time, to save energy by operating at low temperature and shorter time. We also expected solvent consumption to be reduced to a minimum. For this reason, the criteria shown in Table 2 were ranked as input according to their importance in reaching the goal.

Constraints
Lower Upper Name Goal Limit Limit Importance Figure 9. Prediction of the production of the molecules derived from the solvent at different conditions of temperature and residence time.

Optimization
The optimization process has taken into account that our goal was to maximize the production of low-molecular-weight molecules from the lignin and, at the same time, to save energy by operating at low temperature and shorter time. We also expected solvent consumption to be reduced to a minimum. For this reason, the criteria shown in Table 2 were ranked as input according to their importance in reaching the goal. Following the constraints reported in Table 2, the software (Design Expert 10) elaborated the solution shown in Table 3, where the best parameters to adopt in the examined hydrocracking process are located near the center point of the test matrix (279 • C and 60 min). Figure 10 shows the position of the desired solution in the contour plot for each response. Under optimized conditions, we can observe that, compared to a 15% yield reduction (in the total number of monomers produced at the highest conditions), the temperature and time are reduced (respectively by 21 • C and 30 min). Moreover, the production of monomers derived from the solvent decreased by 57% wt.

Gas Analyses
The following gases were detected in the gas phase after the hydrogenation: carbon monoxide, carbon dioxide, methane, ethane, and propane. Generally, the most abundant was propane, followed by carbon dioxide and methane, with a common trend for all tests (Figure 11). The increase in temperature and reaction time increased the amount of produced gases, reported in Figure 11 as wt % of the total starting organic solution. Consumption of solvent was large at severe conditions as shown by the large amount converted into gases. Table 4 reports the calculated mass balance of the process at the different conditions and shows that the consumption of solvent transformed into gas became relevant under the most severe conditions; for instance, 53% was converted at 300 • C and 90 min of reaction. The wt % of final liquid and solid with respect to the starting solution. 2 The wt % of recovered gas with respect to the starting solution (sum of CO, CO 2 , CH 4 , C 2 H 6 , C 3 H 8 ). 3 ∑massout/∑massin. Figure 11. Amount of produced gases at different temperatures and residence times.

HPLC-GPC Analyses
The organosolv lignin and organic solution obtained after hydrocracking were analyzed by HPLC to determine the obtained fragments and their molecular weights. The chromatogram of organosolv lignin showed molecular weights under 5500 dalton ( Figure 12). The peak in the chromatogram at very low molecular weight was due to impurity in butanol, as deduced from the chromatogram obtained with a light scattering detector. After the reaction of hydrocracking, almost all fragments had molecular weights below the 1100 dalton, with the main peaks of the distribution below 350 dalton, corresponding to trimers, dimers, and monomers. It was inferred that more severe conditions increased the production of molecules having low molecular weight, as expected. Chromatograms at different wavelengths (220.4, 273.4, 254.4, and 320.4 nm) were obtained with similar results. Longer residence times of reaction led to the difference being amplified, especially at high temperature, with an increased production of molecules with low molecular weight. At low temperature, there was a negligible difference between chromatograms. There was a regular correlation between the increase in reaction severity, in particular with the temperatures, and the production of molecules having low molecular weight.
The results obtained are compared with other works reported in the literature in Table 5. A larger quantity of monomers is obtained with noncommercial catalysts and higher temperatures, reaching 80% of yield using bifunctional catalysts [45] or much more severe conditions such as 100 bar of H 2 , 400 • C, and 240 min [47]. A detailed economic analysis should be performed to select the most convenient catalyst and solvent, including for example the recycling of products, catalyst cost and life, and byproducts. This is evidently out of the scope of this paper, which is focused on highlighting the role of the solvent used in the butanol-based organosolv pretreatment and one of the cheapest catalysts on the market, the Ni Raney.

Kinetic Analysis
The production data of aromatic monomers and nongaseous degradation products of the solvent were processed to determine the activation energy of the reactions. The kinetic expression of a reaction catalyzed by solids can be very complex, especially considering the mass transfer between the phases, which in this case are (1) hydrogen gas in equilibrium with (2) the liquid solution and (3) the solid catalyst dispersed in the liquid phase. Furthermore, the diffusion in the pores of the reactants and products should also be considered according to the chemical theory [55]. An elementary kinetic model was assumed, based on first-order reactions to empirically describe the set of reactions that occur as in the scheme shown in Figure 13. From a mechanistic point of view, this assumption corresponds to identifying only one limiting step in the overall kinetic process, which we could identify in the breaking of the intermolecular bond β-O-4. The tests were designed and performed to minimize diffusion effects between liquid bulk and catalyst by operating at high stirring speed and breaking the diffusion film. Furthermore, the nickel Raney had negligible porosity. Under these conditions, a kinetic regime can reasonably be assumed.
As the conversion rate of the lignin, L, generically produces a monomer M, the reaction can be indicated as reported in Equation (1).
The instantaneous rate of global reaction of L, a reagent that gradually decreases, with hydrogen H 2 can be assumed proportional to the concentration of the reactants and to the surface of the catalyst, S, as shown in Equation (2).
The concentration of hydrogen in the liquid, [H 2 ], can be approximated as constant because it is in equilibrium with the gaseous phase loaded with excess pressure of hydrogen gas. The surface of the catalyst, S, can also be considered constant: Then, with the constant k of the Arrhenius type being calculated as in Equation (4), the logarithmic relationship is obtained (Equation (5)).
By plotting the logarithm of the specific reaction speed d[L]/L as a function of the inverse of the absolute temperature at which these reaction rates are measured, a straight line is obtained from whose angular coefficient the value of Ea can be obtained ( Figure 14, Table 6).  The data in Table 1 were elaborated to obtain the average conversion rates for three single monomer species and for the total of monomers. In order to minimize the effect of thermal transients and errors on analytical measurements (higher at lower concentrations), the data corresponding to the maximum reaction time, 90 min, were considered. The elaborations are shown in Figure 14, where the ordinates show the average specific rate of formation in the time range 0-90 min and the x-axis shows the inverse of the absolute temperatures at which the tests were performed.
The activation energy values obtained by linear regression are reported in Table 6 and vary from 21 kJ mol −1 (diphenyl-methane 4-ethyl) to 64.3 kJ mol −1 (2,4-dimethyl-3-(methoxy-carbonyl)-5-ethylfuran), while for the total monomers, an average value of 60.0 kJ mol -1 was obtained. Compared to the energy of the β-O-4 bond (150 kJ mol -1 ), the activation energy is considerably lower due to the action of the catalyst and in accordance with recent research concerning the hydrogenation and hydrodeoxygenation of lignin monomers with carbon-supported Ni catalysts [56].
In the Arrhenius diagram in Figure 14, the straight line relating to the decomposition of butanol to nongaseous compounds (butyl butanoate, 2-ethyl-2-hexenal, 2-ethyl-1-hexanol, 7-methyl-4-octanone) had a better correlation coefficient, probably due to the minor experimental error in the determination of molecules with higher concentration. The activation energy was 102.1 kJ mol −1 , in accordance with what was reported for the decomposition of butanol at 270 • C in thermal desorption studies in TGA [57] and with the kinetics of alkene dehydration detected in the pyrolysis of n-butanol in continuous reactors at 700 K [58]. The decomposition of butanol into more elementary molecules (H 2 , C n H m , CO x ) was, however, the predominant reaction, especially at high temperatures, as described in the literature, in the case of reforming in the presence of nickel Raney [59].

Conclusions
Thermal treatment of organosolv lignin with hydrogen and nickel Raney catalyst in subcritical and critical condition of 1-butanol solvent depolymerizes the lignin, and oxygenated trimers, dimers, and monomers can be obtained. The conversion in monomers reached 9 wt %, though this value is limited to the identified molecules ((diphenylmethane 4-ethyl, 2,4-dimethyl-3-(methoxycarbonyl)-5-ethylfuran, and trimethoxy benzene), and higher yields could be obtained. The hydrogenation and cracking reactions also involve the solvent, and its use increases with the temperature and time. When also taking into account the desired minimization of solvent consumption between 20 and 35 wt %, the optimized conditions were found to be 279 • C and 60 min, with a monomeric yield of 7.4 wt %. The role of the reaction medium strongly suggests the need for further research to find a suitable trade-off between solvent type and monomer yield making the whole process sustainable.
Compared to other catalytic systems, such as those based on bifunctional catalysts [45], the Ni Raney catalyst appears low-performing. However, a full process analysis should be performed to establish which would be the more convenient option to be used, taking into consideration factors such as the recycling options or the capital cost of operating at more severe conditions [47].