Flexibilization of Biorefineries: Tuning Lignin Hydrogenation by Hydrogen Partial Pressure

Abstract

The present study describes an interesting and practical catalytic system that allows flexible conversion of lignin into aromatic or aliphatic hydrocarbons, depending on the hydrogen partial pressure. A combination of experiment and theory shows that the product distribution between aromatics and aliphatics can be simply tuned by controlling the availability of hydrogen on the catalyst surface. Noticeably, these pathways lead to almost complete oxygen removal from lignin biomass, yielding high‐quality hydrocarbons. Thus, hydrogen–lignin co‐refining by using this catalytic system provides high flexibility in hydrogen storage/consumption towards meeting different regional and temporal demands.

Better together: A practical catalytic system is described that allows flexible conversion of lignin into aromatic or aliphatic hydrocarbons, depending on the hydrogen partial pressure. The hydrogen‐lignin co‐refining using this catalytic system provides high flexibility in hydrogen storage/consumption towards meeting different regional and temporal demands.

Introduction

Sustainable development requires the use of mainly renewable resources (REs), such as biomass and solar energy, for the production of chemicals and fuels with a low carbon footprint. ^[1] In an ideal scenario, chemicals or fuels are obtained through the combination of different REs without the input of unrenewable resources. For example, biomass or CO₂ could provide the carbon‐based raw material, while additional required energy in a refinery would be provided by solar or wind energy. ^[2] Lignin is an abundant and cheap biomass component with a high potential for the production of aromatics for which currently competitive conversion routes are missing.[ ³ , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ ] Although it can be effectively depolymerized via pyrolysis or “Lignin‐first” approaches, the key barrier is the very high amount of hydrogen required for further refining, mainly for oxygen removal.[ ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ²¹ , ²² , ²³ ] Recently, a self‐supported hydrogenolysis strategy has been proposed to further minimize the hydrogen amount used. ^[23] Current hydrogen production primarily relies on non‐RE resources, mostly natural gas. However, hydrogen can be produced from renewable solar/wind energy with considerable efficiencies, and with the cost of photovoltaic electricity falling to close to 2 Eurocent/kWh in solar‐rich regions, ^[24] cost‐competitive renewable hydrogen could come into reach. Then storage of the hydrogen in aromatic hydrocarbons becomes an interesting option to integrate the solar/wind energy into suitable energy carrier.[ ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ ] Co‐refining of hydrogen and lignin to mixtures of hydrocarbons with low or no oxygen content has the potential to directly deliver the chemicals or fuels, which are easy to transport and could directly be implemented in the current energy system.

Since both, the availability of hydrogen and lignin, can fluctuate substantially,[ ¹⁰ , ³⁰ , ³¹ ] co‐refineries should ideally be flexible, i. e. efficient oxygen removal (leading to valuable chemicals) with variable hydrogen amounts is required. Thus, catalytic systems operating efficiently under strongly fluctuating feed compositions (depending on hydrogen availability) are needed. ^[10] At low hydrogen availability, total oxygen removal with small amounts of hydrogen to aromatics is ideal. In contrast, at high hydrogen availability, full hydrogenation to aliphatic hydrocarbons is optimal, thus buffering varying hydrogen availability, and avoiding hydrogen liquefaction and transportation costs. ^[32]

Efficient hydrodeoxygenation of lignin‐mixtures at variable hydrogen supply with a single catalyst system is still a challenge, although many advanced catalytic systems show highly efficient oxygen removal from lignin feedstocks. Considering phenol as a simple model compound for the lignin fraction (Figure 1), its deoxygenation into cyclohexane or benzene occurs at high or low hydrogen consumption, respectively. The product distribution of phenol hydrodeoxygenation is probably determined by the different adsorption behavior of the catalyst. The formation of benzene or cyclohexane requires different sites and different binding of phenols on catalyst surfaces: vertical adsorption of phenol through oxygen on vacancy sites for direct oxygen removal leads to benzene formation, whereas the flat adsorption of phenol through the aromatic ring and oxygen for hydrogenation on multiple vacancies on catalyst surfaces would lead to cyclohexane formation.[ ³³ , ³⁴ , ³⁵ , ³⁶ , ³⁷ ] Thus, the product from a specific hydrodeoxygenation catalyst system is generally either aliphatic‐rich or aromatic rich, as illustrated by Figure 1.[ ³³ , ³⁸ , ³⁹ , ⁴⁰ , ⁴¹ , ⁴² , ⁴³ , ⁴⁴ , ⁴⁵ ] Although in a biomass hydrogenation process, the hydrogen can be produced internally by additional dehydrogenation of lignin‐derived cycloalkanes, one‐step tuning ability during hydrotreating is more attractive. ^[46] Recently, some of the authors and others described a holistic approach that merely uses all components of the raw biomass material to produce aromatic or aliphatic biofuels.[ ⁴⁷ , ⁴⁸ ]

Figure 1.

Tuning lignin hydrogenation by hydrogen partial pressure in one system. Right: selectivity and conversion of PhOH by hydrotreatment at different H₂ pressures. Reaction conditions: Phenol (3 mmol), pentane (2 mL), and hexadecane (20 mg) at various hydrogen pressures, t=180 min.

Results and Discussion

The present study describes a flexible catalytic pathway based on Ni₂P/SiO₂ catalysts developed to co‐refine hydrogen and lignin‐derived mixtures at different hydrogen availabilities, yielding hydrocarbons with different hydrogen content (aromatics and aliphatics). First, we discuss a flexible phenol hydrotreatment, leading to a mixture of products ranging from fully hydrogenated (cyclohexane) to unsaturated (benzene) products at different hydrogen pressures. A combined experimental and theoretical approach is used to understand how hydrogen pressure affects the reaction pathway and directs the conversion into different products. Second, we demonstrate that lignin‐derived mixture can be converted into either hydrogen‐rich or hydrogen‐deficient hydrocarbons by only adjusting the hydrogen pressure in the hydrotreatment stage of the products from lignin pyrolysis, using the same catalyst.

During the hydrotreatment of phenol over the Ni₂P/SiO₂ catalyst (46 wt % Ni), the hydrogen consumption can be controlled by merely adjusting the hydrogen pressure while maintaining efficient oxygen removal (Figure 1). Explicitly, partial pressure of hydrogen of only 3 bar favors benzene formation (with an aromatic/aliphatic ratio of 4 : 1), whereas the yield of cyclohexane increases with the H₂ pressure, with a 1 : 7 ratio at 15 bar.

To assess the effect of hydrogen pressure on hydrogen consumption, we performed the kinetic study of the reaction at H₂ pressures of 6 and 15 bar (see the Supporting Information, Figure S1c,d). Conversion at p ${}_{\mathrm{H}{}_2}$ =6 is 75 % after 180 min, and benzene selectivity reaches 72 % (see Table S1 for detailed product distribution), whereas conversion at p ${}_{\mathrm{H}{}_2}$ =15 bar is significantly higher (over 99 % at 40 min), and selectivity for cyclohexane is 88 %. At the lower partial hydrogen pressure, p ${}_{\mathrm{H}{}_2}$ =6 bar, phenol conversion proceeds quickly for approximately 40 min, but then the rate decreases appreciably, and no significant further conversion into cyclohexane is observed. The low cyclohexanol and cyclohexanone yields (Figure S5) indicate a small degree of phenol hydrogenation, corresponding to low hydrogen storage (50 % of consumed hydrogen is used for oxygen removal). At p ${}_{\mathrm{H}{}_2}$ =15 bar, phenol conversion is completed in 40 min, followed by slow benzene→cyclohexane conversion (only 27 % hydrogen is used for oxygen removal, whereas the rest is stored in hydrocarbons). The oxygen is mostly removed in the form of water, as is shown in Figure S6, accompanied by partial hydrogen storage in the products. At p ${}_{\mathrm{H}{}_2}$ =6 bar, approximately 1.9 mmol hydrogen is stored at 4 mmol consumption, whereas at p ${}_{\mathrm{H}{}_2}$ =15 bar, 7.9 of 10.9 mmol of consumed hydrogen is stored in the final product mixture. As the product distribution can easily be adjusted by controlling the reaction parameters (Figure 1), the system developed is capable of delivering hydrocarbon mixtures through hydrogen and lignin co‐refining at different hydrogen availabilities.

For further insight into this hydrogenation process, different reaction pathways were considered computationally. Possible reactions and intermediates for phenol conversion into benzene and cyclohexane are outlined in Figure 2a: i) hydrogenation of the aromatic ring (A→B→C→D) and ii) deoxygenation of intermediates B, C, and D into the corresponding products E, F, G, and I. The competition between hydrogenation and deoxygenation reactions determines product selectivity at different H₂ pressures. The reaction and activation energies of individual reaction steps on the Ni₂P catalyst were computationally investigated for various H₂ pressures by using the formalism of the density functional theory (DFT).

Figure 2.

a) Reaction intermediates on the phenol→benzene and phenol→cyclohexane pathways; the hydrogenation of intermediate B has been identified as the critical step that determines product selectivity (see Figure S2 for more details). b and c) Calculated activation barriers (in eV) of elementary reaction steps for hydrogenation of B at low and high pressure regimes, respectively. d) Simplified proposed phenol deuteration reaction pathways at 573 K under low and high pressure (see Figure S8 for more details). f) The benzene/cyclohexane D _n numbers at 40 min for different phenol deuteration reaction pressures.

DFT based calculations were performed for the Ni₂P (110) surface (previously determined as the thermodynamically most stable surface) for two different hydrogen coverages: low (1H^*) and high (2H^*) coverage models with one and two H atoms per surface Ni₃, respectively (Figure S3).[ ⁴⁹ , ⁵⁰ , ⁵¹ , ⁵² ] Details of the calculations can be found in the Supporting Information. The reaction steps [e. g., B→E(F), C→F(G), and D→G(I)] shown in Figure 2a consist of two elementary steps each, corresponding to the transfer of just one H atom between the surface and the reactant. Only the higher activation barrier of each reaction step is reported (Figure S7). The kinetically most favorable path is shown in green and blue for lower and higher surface hydrogen coverages, respectively. At higher surface hydrogen coverage (Figure 2c, blue), the reaction proceeds through the A→B→C→D→I path, along which the aromatic ring is hydrogenated first, and cyclohexanol (formed as an intermediate in higher concentrations than at low coverage, Fig. S8) is then deoxygenated to cyclohexane.

Conversely, at lower surface hydrogen coverage (Figure 2b), the kinetically favored reaction proceeds along the A→B→E path towards benzene. Product selectivity is thus determined by the relative activation barriers of the hydrogenation and deoxygenation steps of cyclohexadienol (B; Figure 2a–c; details in Figures S3 and S4). The calculations confirm the experimental observation that hydrogen coverage (corresponding to p ${}_{\mathrm{H}{}_2}$ ) plays a crucial role in product selectivity: (i) the relative activation barriers of the deoxygenation and hydrogenation steps shown in Figures 2b and 2c (B→E and B→C, respectively) as well as (ii) the relative stabilities of hydrogenation intermediates (B, C, and D) depend on the surface hydrogen coverage. At higher surface hydrogen coverage, the hydrogenation steps B→C→D leading to cyclohexane should be the dominant pathway (Figure 2c and Figure S7b). In contrast, at lower surface hydrogen coverage, the direct formation of E (benzene) from B is favored (Figure 2b and Figure S7a). The calculations show that the activation barriers strongly depend on the surface hydrogen coverage; when the coverage increases, all activation barriers decrease, and their relative values change. Thus, hydrogen surface coverage affects, both quantitatively and qualitatively, the barriers of the individual reaction steps. p ${}_{\mathrm{H}{}_2}$ determines the surface hydrogen coverage (see Supporting Information), which in turn directs the catalyzed reaction to different products, involving different amounts of unreacted hydrogen.

The unreacted hydrogen amounts, at different reaction times, were experimentally assessed at the H₂ pressures tested, 6 and 15 bar (Figure S6). The consumption rate of free hydrogen (x ${}_{\mathrm{H}{}_2}$ ) at low p ${}_{\mathrm{H}{}_2}$ is lower than that at high p ${}_{\mathrm{H}{}_2}$ , and it becomes very slow after 40 min (Figure S6a). The decrease in reaction rate may be accompanied by the decline in surface hydrogen coverage: hydrogen consumption lowers the p ${}_{\mathrm{H}{}_2}$ , which in turn reduces the hydrogen surface coverage, thereby decreasing the reaction rate and increasing the selectivity for benzene. The highest selectivity for benzene was observed at the lowest H₂ pressure (Figure 1), and the A→B→E pathway was identified as the most likely route towards benzene based on the DFT calculations. Nevertheless, as the difference in the activation barrier between each step is relatively small, other pathways, such as A→B→C→F→E (via cyclohexanone/cyclohexanol), cannot be totally ruled out.

We conclude that hydrogen consumption and hydrogen storage in phenol depend on the competition between hydrogenation and deoxygenation reactions. Low hydrogen pressures suppress the hydrogenation of the aromatic ring and promote the deoxygenation of the intermediate. High hydrogen pressures promote the hydrogenation of the intermediates and enhance hydrogen storage in the final products (aliphatic hydrocarbons). Thus, this catalytic system allows to simply change the selectivity of ring hydrogenation by changing hydrogen partial pressure, without negatively affecting deoxygenation. Accordingly, hydrotreating reactions were performed on a selection of additional model compounds of varying complexity (Table S5). In general, the catalytic system was able to tune the hydrogenation degree of the final product regardless of the presence of methoxy groups or alkyl chains.

To further understand the reaction mechanism, phenol deuteration was performed at different D₂ pressures, and the degree of deuteration of the final products (benzene and cyclohexane) was assessed (Figure 2e, Table S3). The pathways leading to benzene can be analyzed concerning the average number of deuterium atoms in benzene. As shown in Figure 2d and Figure S7, if the reaction proceeds exclusively through the A→B→E pathway, the average deuteration number (D _n) should be 1.5; if it progresses only through the A→B→C→F→E or A→B→C→D→G→F→E pathways, the average D _n of benzene should be 2.5 and 3.5, respectively. Similarly, the path leading to cyclohexane (A→B→C→D→I) should have an average deuteration number D _n of 7.

The deuteration levels assessed at 3, 6 and 15 bar are shown in Figure 2e and Table S3, and the corresponding selectivity and conversion values are summarized in Table S2. The low value of D _n found at low p ${}_{\mathrm{D}{}_2}$ (3 bar) is close to 1.5, strongly indicating that, under low‐pressure conditions, the reaction proceeds almost exclusively through the A→B→E pathway. The benzene D _n increases with the p ${}_{\mathrm{D}{}_2}$ (6 and 15 bar), thus indicating that, under these conditions, the reaction path towards benzene A→B→E becomes increasingly less important, and deoxygenation occurs after partial (A→B→C→F→E) or full (A→B→C→D→G→F→E) aromatic ring hydrogenation. Under lower p ${}_{\mathrm{D}{}_2}$ , the average deuteration number (D _n) of the cyclohexane is lower than expected, possibly due to liquid phase H‐transfer, as reported in previous studies.[ ⁵³ , ⁵⁴ ] Nevertheless, these results corroborate the finding discussed above that the degree of hydrogenation of the final product can be easily controlled by adjusting the hydrogen pressure without affecting the oxygen removal efficiency.

Hydrotreating of a lignin‐derived mixture was performed to assess whether the results described above for phenol can be transferred to an original biomass‐based feedstock. The lignin‐derived mixture is produced by pyrolysis of organosolv lignin, which consists of various methoxy‐substituted phenols, as shown in gas chromatography‐mass spectrometry (GC‐MS) measurements (Supporting Information A). ^[55] hydrotreating was performed at different hydrogen pressures to vary the reaction path (aliphatic/aromatic ratio), and the corresponding degree of hydrogenation. The composition of lignin‐derived compounds and the final products are quantitatively analyzed by GC and qualitatively characterized by MS, as shown in SI Appendix A−D. Some products resulted from aromatic alkylation reactions during pyrolysis were also identified.[ ⁵⁶ , ⁵⁷ ] GC‐MS analysis of the initial lignin‐mixture and the final products (Figure S9) suggests high oxygen removal efficiency because almost no oxygen‐containing products were detected. As GC‐MS analysis may be unable to provide a more detailed analysis of heavier compounds, FT (Fourier Transform) mass spectrometry (e. g. FT‐ICR and Orbitrap) is applied to analyze the complex mixtures further.[ ⁵⁸ , ⁵⁹ , ⁶⁰ ] The high mass accuracy and sensitivity of this method allow us to determine the exact elemental formula, which may provide a complete overview of the products. ^[61] Accordingly, a more detailed analysis of the initial reaction mixture and final products by positive‐ion atmospheric pressure chemical ionization (APCI) Orbitrap mass spectrometry (Figure S10 further confirmed the high oxygen removal efficiency). On the one hand, the initial mixture resulting from lignin pyrolysis contained compounds with 1–10 oxygen atoms, and the O₃ class had the highest intensity. On the other hand, the products from hydrotreatment at different hydrogen pressures primarily consist of hydrocarbon species and some minor oxygen‐containing compounds, with 1–4 oxygen atoms. Indeed, the amounts of oxygen‐containing compounds are quite low, as indicated by the intensity. These results confirm the high efficiency of the oxygen removal from the real lignin‐derived mixture by this approach, despite the different hydrogen pressures applied during the hydrotreatment.

The aromatic/aliphatic ratios of some typical products assessed by GC‐MS are outlined in Table 1, illustrating the trend of an increase in the aliphatic fraction with the increase in p ${}_{\mathrm{H}{}_2}$ . The distribution of hydrocarbons can be shifted from aliphatic‐rich (Figure S9a) to aromatic‐rich (Figure S9c). At p ${}_{\mathrm{H}{}_2}$ =20 bar, only aliphatics are produced, indicating a high hydrogen storage efficiency. At p ${}_{\mathrm{H}{}_2}$ =5 bar, approximately 2/3 of the products are aromatics, thus consuming at least 40 % less hydrogen than at high p ${}_{\mathrm{H}{}_2}$ (20 bar) conditions. Taking pyrolysis lignin as an example C₈H_6.3–7.3O_0.6–1.1(OCH₃)_0.3–0.8(OH)_1–1.2, the conversion of the raw material into aliphatic or aromatic hydrocarbons consumes 42 and 22 mol hydrogen per kg pyrolysis lignin. ^[62] The method presented allows the co‐refinery of 1 kg raw lignin material and hydrogen with an amount ranging from 29 to 42 mol, which means that 7 to 20 mol hydrogen are stored. Although GC‐MS measurements allow a detailed analysis of the volatile part of the reaction mixture, non‐volatiles are not detected. The results from additional studies by FT‐Orbitrap MS are summarized in Figure 3 for each hydrotreating product. Double bond equivalent numbers (DBE) are plotted against carbon numbers (Figure 3), and the intensity is indicated in a heatmap. The main products obtained under different pressures primarily consist of hydrocarbons with a carbon number lower than 20, as noted in the intensities of the populations shown in Figure 3. The major compounds shown with the highest intensity at each carbon number and their corresponding DBE number are highlighted for each hydrotreating condition in Figure 3. A high DBE number indicates more double bonds and, therefore, lower hydrogen consumption during the hydrotreating process, and vice versa. As observed, the DBE of major products changes dramatically. The difference in DBE number is more noticed at higher carbon numbers. For example, at a carbon number in the 40–50 range, an undoubtedly significant difference is observed, the DBE obtained at p ${}_{\mathrm{H}{}_2}$ =20 bar and p ${}_{\mathrm{H}{}_2}$ =5 bar is 10 and 25, respectively. Thus, these data further confirm that the flexible reaction path proposed here is indeed a useful method to control product selectivity by simply adjusting hydrogen consumption.

Table 1.

The ratios between selected aromatic and aliphatic products at reaction hydrogen pressures of p ${}_{\mathrm{H}{}_2}$ =20, 8, and 5 bar, at 573 K (20 and 8 bar) and 613 K (5 bar).

Pressure	20 bar	8 bar	5 bar
	0/100	29/71	40/60
	0/100	29/71	67/33
	0/100	24/76	62/38

Figure 3.

Double bond equivalent numbers (DBE) versus carbon number heatmaps (scaled according to the absolute intensity contours) of different hydrotreating products obtained at 5 (a), 8 (b) and 20 bar (c). Accordingly, the DBE of the major compounds with the highest intensity is plotted as a function of the carbon numbers for the different pressures. For comparison, the curves for all pressures are shown, with the one given in the legend highlighted by the corresponding color. DBE expresses the degree of unsaturation, that is, DBE 1 is equal to either one double bond or one ring structure.

Conclusion

In summary, the single catalytic system described above allows us to develop a balanced approach to efficiently upgrade lignin‐derived aromatics at the fluctuating hydrogen supply from wind/solar or even fossil fuel. Deoxygenation is almost complete so that the products are hydrocarbons with variable aromatic/aliphatic ratios and hydrogen content. When the production of aromatics is desirable and/or when the amount of available H₂ is limited, the process can be operated at a low p ${}_{\mathrm{H}{}_2}$ . Conversely, p ${}_{\mathrm{H}{}_2}$ can be increased (without changes in catalyst or plant design) when the target products are aliphatics and/or when molecular H₂ is available in excess. The resulting products have low oxygen content, and they can be easily incorporated into the current industry systems.[ ⁶¹ , ⁶³ ] This pathway has the potential to mitigate the effects of changes in hydrogen supply in lignin refining. Therefore, the high flexibility of this approach may have important technological applications for solar/wind energy conversion. On one hand, hydrogen can be used as a storage intermediate for solar/wind energy conversion systems at significant efficiencies (52 %–82 %, HHV). ^[25] On the other hand, lignin‐derived aromatics can be used as hydrogen storage intermediates. Thus, this system allows us to integrate more solar/wind energy into the current energy systems by converting different amounts of primarily clean energy sources into suitable energy carriers – hydrocarbons.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

Z. Cao, Y. Xu, P. Lyu, M. Dierks, Á. Morales-García, W. Schrader, P. Nachtigall, F. Schüth, ChemSusChem 2021, 14, 373.