Selective Hydrodeoxygenation of Lignin-Derived Phenolic Monomers and Dimers to Cycloalkanes on Pd/C and HZSM-5 Catalysts

Lignin is an abundant renewable resource with a high energy density, but it is considered to be difficult to process because of the high reactivity of its building blocks, that is, the substituted phenol units, which tend to react even at small concentrations and reaction temperatures.1 It has been shown recently that this reactivity can be overcome by combining metallic (Pd) and acidic functions (H₃PO₄, CH₃COOH, SO₃H, and OH) in the appropriate concentrations in an aqueous phase or ionic liquids.2 This has led to the successful hydrodeoxygenation of phenol derivatives and the synthesis of a pure cycloalkane product. Because of the different polarity of the substituted phenols and the alkanes, a second hydrocarbon phase is formed during the process, which can be easily separated. Noble and base metals have been found to be active for hydrogenation in the aqueous phase. Replacing liquid mineral acids by a solid acid (Nafion/SiO₂) allowed for an increase in efficiency.3

In addition to the hydrodeoxygenation of phenolic monomers, the selective cleavage of the aromatic carbon–oxygen (CO) bonds in aryl ethers is also challenging because of the strength and stability of these linkages.4 This cleavage is very important for facilitating the depolymerization of oxygenrich lignin by breaking down the COC linkages, and for the hydrodeoxygenation of lignin-derived phenolic dimer fragments to the deoxygenated biofuels. Here, we report on the use of a weaker solid acid, that is, a zeolite (HZSM-5), as a selective catalyst component for the quantitative hydrodeoxygenation of diversely substituted lignin-derived mono- and binuclear phenols to cycloalkanes in combination with a noble metal (Pd) in aqueous solutions at a mild temperature (473 K).

We have shown previously that phenol is converted to cyclohexane in water through the sequential hydrogenation of phenol to cyclohexanone and cyclohexanol on metal sites (Pd or Ni), dehydration of cyclohexanol on acid sites (H₃PO₄, CH₃COOH, or Nafion/SiO₂), and finally the hydrogenation of cyclohexene to cyclohexane on metal sites.1–3 To maximize the hydrodeoxygenation rate and selectivity under milder conditions (low reaction temperatures and pressures) in addition to the catalyst stability, various solid acids (acid-site densities and specific surface areas are listed in Table S1 in the Supporting Information) are explored in the presence of palladium Pd/C as hydrogenation catalyst. The characterization of the catalyst was achieved by determining the Brunauer–Emmett–Teller (BET) surface area and by using XRD, SEM, and TEM and is compiled in the Supporting Information. As a suitable solid acid should have a high acid-site density in combination with a sufficient stability in an aqueous phase above 473 K, the results (see Table S3 in the Supporting Information) for the conversion of 4-n-propylphenol show that solid Lewis acids, such as alumina, silica, and amorphous silica alumina, are not effective for oxygen removal (through dehydration of cycloalcohol). Such catalysts produce less than 3 % cycloalkanes and more than 90 % cycloalcohols in the presence of Pd/C at 473 K and 5 MPa H₂ for 0.5 h. This shows that the acidic hydroxyl groups and Lewis acid sites on oxide surfaces are ineffective for alcohol dehydration in the presence of water because of competing water molecules, which reduces the effective acid strength.5 Although alumina pillared clay leads to 50 % yields of cycloalkanes, the reaction rates are rather low because of diffusion limitations of the reactant.6 In contrast, solid Brønsted acids with a sufficiently high acid density, such as sulfated zirconia, Amberlyst 15, Nafion/SiO₂, and Cs_2.5H_0.5PW₁₂O₄₀, lead to 90 % yield of cycloalkanes. HZSM-5 with a Si/Al ratio of 45 and a Brønsted acid site (BAS) density of 0.278 mmol g⁻¹ produces yields of 93 % C₉ cycloalkanes, 2.5 % C₉ cycloalcohols, and 4.5 % ethers (formed by intermolecular dehydration of C₉ cycloalcohols). Hydrogenation of phenol on Pd/C leads to cyclohexanol as the initial product. Its dehydration in water is catalyzed by BAS in the pores, whereas ethers are formed on Lewis-acidic sites on the external surface.5 The dehydrated cycloalkene is immediately converted to the saturated cycloalkane by accessible Pd atoms.

Compared to other solid Brønsted acids, high yields of C₉ cycloalkanes can be produced from 4-n-propylphenol by using a combination of HZSM-5 and Pd/C at lower temperatures. Approximately 90 % cycloalkanes were produced in the presence of HZSM-5 and Pd/C at 433 K after 0.5 h, whereas less than 20 % cycloalkanes, but more than 80 % cycloalcohols, were formed in the presence of sulfated zirconia, Amberlyst 15, Nafion/SiO₂, and Cs_2.5H_0.5PW₁₂O₄₀ under identical conditions (Figure 1). This points to a lower apparent activation energy and a higher reaction rate for the cycloalcohol dehydration on HZSM-5 compared to the other solid Brønsted acids for the overall hydrodeoxygenation.

Figure 1. Product distributions for the hydrodeoxygenation of 4-n-propylphenol with Pd/C and solid-acid catalysts at different temperatures [-one/ol: 4-n-propylcylohexanone/ol; ether: 4,4′-oxybis-(propylcyclohexane); -ane: alkyl cycloalkane]. Reaction conditions: 4-n-propylphenol (1.36 g, 0.010 mol), Pd/C (5 wt %, 0.040 g), solid acid (2.0 g), and H₂O (80 mL) were stirred at 680 rpm and 5 MPa H₂ (ambient temperature) for 0.5 h.

The dehydration rate of cycloalcohol (in situ produced by phenol hydrogenation) in water estimated from Figure 1 for the overall phenol hydrodeoxygenation reaction was approximately 41 mol mol h⁻¹ on HZSM-5 at 433 K, whereas these rates were approximately 0.3, 6.7, 11, and 4.3 mol mol h⁻¹ on sulfated zirconia, Amberlyst 15, Nafion/SiO₂, and Cs_2.5H_0.5PW₁₂O₄₀, respectively. In our previous study, aqueous H₃PO₄ also showed a low dehydration rate for cyclohexanol (turnover frequency: 15 mol mol h⁻¹) and a high activation energy (120 kJ mol⁻¹) in water,2b and, thus, the dehydration (the rate determining step in the sequence of reactions) had to be catalyzed at temperatures above 473 K. The apparent activation energy of dehydration on HZSM-5 was approximately 95 kJ mol⁻¹; for comparison, the activation energies for the other four solid acids were in the range 110–120 kJ mol⁻¹, thus allowing a lower reaction temperature when using the zeolite.

As this reaction exhibits a first-order dependence of the substrate concentration, we attribute the high dehydration rates on HZSM-5 to a higher concentration of the substrate in the zeolite pores. In situ IR spectroscopy shows that the cyclohexanol concentration is substantial in HZSM-5, whereas almost no cyclohexanol is adsorbed on other strong Brønsted solid acids at 313 K (Figure S2 in the Supporting Information). However, it has also to be considered that the micropore size of HZSM-5 allows only the smaller alcohol monomers to reach the BAS sites of the zeolite. Thus, the alcohol monomer–oligomer equilibrium is rapidly shifted towards monomers, which also accelerates the dehydration rate.7

The hydrodeoxygenation was explored at 433 K by varying hydrogen pressure (1.0–5.0 MPa) and concentrations of HZSM-5 (0.1–2.0 g, Table S4 in the Supporting Information). For the hydrogen pressures ranging from 3.0–5.0 MPa (ambient temperature), the hydrodeoxygenation activity remained nearly constant, with 90 % alkane formed after 0.5 h; but when the pressure was lowered to 2.0 MPa, the cycloketone yield increased to 10 %, and the alkane yield decreased to 76 %. As the pressure was continuously decreased to 1.0 MPa, the conversion was reduced to 80 %, with a selectivity of 75 % cycloketone and 12 % cycloalkane, which indicates that a low hydrogen pressure (1.0 MPa) stabilizes the ketone intermediate and suppresses the cycloalkane formation. The alkane yield increased in parallel to the decrease of the cycloalcohol yield (75–1.0 %), showing as expected that the higher acid-site concentration accelerates the alcohol dehydration.

To demonstrate the versatility of the approach, a variety of lignin-derived phenolic reactants, such as alkyl- or methoxy-substituted C₆–C₉ phenols, guaiacols, and syringols, were tested under optimized conditions in water at 473 K for 2 h (Table 1). The combination of Pd/C and HZSM-5 quantitatively converted all substituted phenolic monomers to their corresponding C₆–C₉ cycloalkanes and methanol. By comparison, the previously explored combined catalysts of Pd/C and liquid-acid H₃PO₄ required 523 K to efficiently hydrodeoxygenate methoxy-substituted phenolic monomers to cycloalkanes,2a, b again showing that zeolite is the more efficient catalyst. Some cyclopentanes were formed through isomerization of substituted cyclohexanes on Brønsted acid sites of HZSM-5. Isomers from cyclohexane were not detected in this system, implying that cyclohexane is more difficult to isomerize than aryl-substituted cyclohexanes, which is related to the fact that the tertiary carbocations are more stable than secondary carbocation intermediates during the isomerization.8 Aromatic molecules were not found when using Pd at 473 K.

Table 1. Hydrodeoxygenation of phenolic monomers on Pd/C and HZSM-5 catalysts in water at 473 K.^[a]

Phenolic monomers	Conversion [%]	Selectivity [C %]
		Cycloalkanes	Methanol
[a] Reaction conditions: phenolic reactant (0.010 mol), Pd/C (5 wt %, 0.080 g), HZSM-5 (1.0 g), H2O (80 mL), 473 K, 5 MPa H2 (ambient temperature), 2 h, stirred at 680 rpm.
	100	100	–
	100	85	13
	100	88	11
	100	89	10
	100	98	–
	97	90	9.9
	97	90	9.7
	95	80	17

The reaction of the aqueous phenolic mixture (nine different phenol, guaiacol, and syringol derivatives, which made up 14.2 wt % of the phenols in the water solution) over the combination of Pd/C and HZSM-5 at 473 K and 5 MPa H₂ for 4 h led to an upper organic layer consisting of 87 % C₆–C₉ cycloalkanes and an aqueous layer containing 11 % methanol at a conversion of 85 % (Scheme 1), suggesting that this catalyst system can efficiently suppress the polymerization of the phenolic mixture even at high reactant concentrations.

Scheme 1. Hydrodeoxygenation of the phenolic mixture with Pd/C and HZSM-5 catalysts in water at 473 K. Conversion (%): 85 %, cycloalkane selectivity (C %): 87 %, methanol selectivity (C %): 10 %. Reaction conditions: the phenolic mixture (13.27 g), which included each phenolic monomer (0.010 mol), Pd/C (5 wt %, 0.40 g), HZSM-5 (4.0 g), and H₂O (80 mL) were stirred at 680 rpm, 473 K, and 5 MPa H₂ (ambient temperature) for 4 h.

In addition to the conversion of phenolic monomers, we also explored this catalyst combination for converting phenolic dimers to cycloalkanes. This is important, on the one side, because a significant concentration of phenolic dimers is formed in the thermal deconstruction of lignin,9 and, on the other side, as this novel reaction route for cleaving CO bond linkages could provide a fundamental chemical insight useful for the investigation of lignin depolymerization. The most representative linkages in phenolic dimers, for example, in spruce lignin, are β-O-4 (55 %), α-O-4 (8 %), 5-5 (5 %), 4-O-5 (5 %), β-1 (15 %), β-β (7 %), and β-5 (10 %). Different types of phenolic-dimer model compounds including both COC and CC linkages were, therefore, explored (see Table 2).

Table 2. Hydrodeoxygenation of phenolic dimers on Pd/C and HZSM-5 catalysts in water at 473 K.^[a]

Phenolic dimers	Conversion [%]	Selectivity [C %]
[a] Reaction conditions: phenolic dimer reactant (0.010 mol), Pd/C (5 wt %, 0.080 g), HZSM-5 (1.0 g), H2O (80 mL), 473 K, 5 MPa H2 (ambient temperature), 2 h, stirred at 680 rpm. [b] Isomer 10: (cyclopentylmethyl)cyclohexane. [c] Isomer 5: (2-cyclopentylethyl)cyclohexane. [d] Isomer 2: (5-cyclopentylpentyl)cyclohexane.
	100
(β-O-4)		46	54
	100
(α-O-4)		48	52
	100
(α-O-4)		48	52
	100
(α-O-4)		44	56
	100
(4-O-5)		100
	100
(4-O-5)		100
	100
(5-5)		85, isomer 10^[b]
	100
(β-1)		95, isomer 5^[c]
	100
(β-β)		98, isomer 2^[d]

The most abundant linkage in lignin (β-O-4, alkylaryl ether) was quantitatively converted to 46 % C₆ cyclohexane and 54 % C₈ ethylcyclohexane at 473 K and 5 MPa H₂ during 2 h reaction time. In a previous study, the flash vacuum pyrolysis of alkylaryl ethers, such as the β-O-4 linkage, produced styrene and phenol as primary products and led to uncontrollable free-radical reactions at 723 K and a wide variety of by-products with less than 30 % hydrocarbon selectivity at 10 % conversion.10 In the hydrothermal conversion, the primary products were phenol and ethylbenzene at a low conversion of the β-O-4 dimer, which implies that this COC linkage may be opened by hydrogenolysis or pyrolysis with hydrogen addition. By comparison, the CH₃Oaryl and HOaryl bonds in phenols were cleaved by hydrolysis and hydrogenation–dehydration using Pd and H₃PO₄, respectively.2a, b

The benzyloxybenzene (α-O-4) and the o- or p-hydroxyl-substituted α-O-4 model compounds were also quantitatively converted with an approximately 50 % yield of C₆ cyclohexane and 50 % yield of C₇ methylcyclohexane under these conditions. Hydrothermal conversion of the α-O-4 dimer in superheated water also followed ionic and radical pathways, yielding a broad product distribution with phenols and many secondary recombination products.11 Recently, it was reported that the combination of [Ni(cod)₂] (COD=cyclooctadiene), 1,3-bis(2,6-diisopropylphenyl)imidazolinium chloride (SIPr⋅HCl), and sodium tert-butoxide (NaOtBu) could selectively hydrogenolyze the CO bonds in aryl ethers to substituted phenols and arenes in m-xylene as solvent in the presence of hydrogen.12 The proposed mechanism includes that the nickel component cleaving the CO bond could be nickel hydride, a neutral nickel(0) complex, or an anionic nickel species. In the conversion of α-O-4 without Pd, HZSM-5 led to 63 % yield of recombination products, such as 2-benzylphenol, and 37 % yield of phenol in water at 473 K for 2 h. If Pd/C and HZSM-5 were jointly used, the alkylaryl ether was quantitatively cleaved into smaller deoxygenated cycloalkanes. Therefore, the mechanism for the hydrodeoxygenation of β-O-4 and α-O-4 model compounds is assumed to start with the Pd⁰-catalyzed hydrogenolysis of the ether to substituted phenols and arenes, followed by the hydrodeoxygenation (hydrogenation–dehydration) of the phenols or the hydrogenation of the produced arenes on a metal/acid dual functional catalyst, which leads to the formation of the target cycloalkanes.

The 4-O-5 dimers diphenyl ether (DPE) and 4-hydroxy-substituted DPE were quantitatively converted to C₆ cyclohexane at 473 K and 5 MPa H₂ by using the combination of Pd/C and HZSM-5. It has been reported previously that hydrolysis of the arylaryl ether bond requires severe conditions and high temperatures. Siskin et al. reported that DPE at 588 K gave a 92 % conversion to phenol in 15 % phosphoric acid and a 6.6 % conversion to phenol in 15 % aqueous sodium formate after three days.13 Although DPE could be quantitatively converted to phenol with a 100 % selectivity in a 15 % K₂CO₃ solution at very high vapor pressures,14 DPE was found to remain unchanged using only HZSM-5 in a separate experiment (not described) at such hydrothermal conditions, suggesting that the combination of Pd and HZSM-5 is essential for the cleavage of the arylaryl-ether bond.

The CC linkages in 5-5, β-1, and β-β were preserved, whereas the substituted hydroxyl and ketone groups were selectively removed at 473 K and 5 MPa H₂, leading to more than 95 % yields of hydrodeoxygenated C₁₂, C₁₄, and C₁₆ bicycloalkanes, respectively (Table 2). The 5–10 % isomers of cycloalkanes were produced by the Brønsted acid sites of the H-form of the zeolite. For 2,2′-biphenol, a 5 % yield of the ketone intermediate was also detected after 2 h reaction time, suggesting that the hydrogenation rate of 2,2′-biphenol is considerably lower than those of other dimers. This may be attributed to the fact that the intramolecular hydrogen bonds in 2,2′-biphenol enhance the stability of the aromatic ring and influence the spatial configuration of the two benzene rings, which could hinder the attack of the catalyst.15

In summary, the combination of Pd/C and HZSM-5 showed an extremely high selectivity in removing oxygen-containing groups (hydroxyl, methoxy, ketone, alkylOaryl, and arylOaryl) in lignin-derived substituted phenolic monomers and dimers through a cascade metal-acid-catalyzed cleavage of CO bonds in phenolic dimers and integrated hydrogenation and dehydration reactions in water at 473 K. This approach opens an efficient route for upgrading lignin-derived phenolic oil to transportation biofuels.

Experimental Section

The typical experimental procedure for the hydrodeoxygenation proceeded as follows. The reactant (0.0106 mol), H₂O (80 mL), Pd/C (5 wt %, 0.080 g), and HZSM-5 (Si/Al=45, 1.0 g,) were added into a Parr autoclave reactor. After flushing the reactor with H₂ three times, the reactions were conducted at 473 K for 2 h while stirring at a speed of 680 rpm. After cooling to ambient temperature, the organic products were extracted by using ethyl acetate. The organic and aqueous phases were gathered and analyzed by using GC and GC–MS. The calculation of conversions and selectivities was based on carbon: Conversion=(change of raw material/total amount of raw phenolic compounds)×100 %; selectivity=(carbon atoms in each product/total carbon atoms in the products)×100 %. The carbon balance in the liquid phase reached 95±3 %.

Acknowledgements

This research was performed in the framework of the European Graduate School for Sustainable Energy.