Hydrogenolysis of Lignin and C–O Linkages Containing Lignin-Related Compounds over an Amorphous CoRuP/SiO₂ Catalyst

Abstract

Efficient depolymerization of C–O linkages is essential for converting lignin into fuels and higher value-added chemicals. In this work, CoRuP/SiO₂, an amorphous Ru-Co phosphide composite, was fabricated for the efficient hydrogenolysis of ether linkages. The 4–O–5 and α–O–4 linkages containing lignin-related compounds, such as diphenyl ether, benzyl phenyl ether, 3-methyl diphenyl ether, and dibenzyl ether, are selected as representatives of linkages in lignin. Under mild conditions, Ru-containing metallic phosphides have high-performance for the catalytic depolymerization of C–O linkages. Compared with other catalysts, CoRuP/SiO₂ shows an outstanding selectivity for benzene and excellent efficiency in depolymerizing diphenyl ethers, yielding only a small amount of by-products. Furthermore, the total acidity shows a linear relationship with the hydrogenolysis reactivity in cleaving aromatic ether bonds. The mechanisms for the catalytic hydrogenolysis of 4–O–5 and α–O–4 bonds over CoRuP/SiO₂ are proposed. Moreover, two-dimensional heteronuclear single quantum coherence nuclear magnetic resonance spectroscopic analysis demonstrates that CoRuP/SiO₂ could effectively depolymerize C–O bonds of lignin. These dominant hydrogenolysis products from lignin have excellent potential in the production of high value-added drugs or pharmaceutical intermediates. The hydrogenolysis of lignin can be a highly efficient alternative to the existing method of lignin utilization.

1. Introduction

As a result of the increasing depletion of fossil fuels, the environmental pollution and energy crisis are posing increasingly problems worldwide [1,2]. A promising approach to address these issues is to convert renewable and abundant biomass resources into high grade fuels and value-added chemicals [3,4,5]. Biopolymers rich in natural phenolic substances, i.e., lignin, meet these essential criteria [6,7,8].

Network polymer-like lignin consists of phenylpropane-related structural units connected by bridge bonds [9], and over 80% of the bonds are etheric. Thus, the efficient depolymerization of C–O bonds is essential for a higher value-added utilization of lignin [10].

Several strategies, including pyrolysis, hydrolysis, oxidation, and hydrogenolysis, have been proposed to depolymerize lignin for biodiesel and chemical production. Among these strategies, hydrogenolysis has received increasing interest due to its techno-economic feasibility [11,12].

Depolymerization of the 4–O–5 bond is a highly challenging issue because of its weak reactivity compared to β–O–4 and α–O–4 bonds [13]. Thus, diphenyl ether is extensively employed as a lignin-related probe compound to explore the depolymerization behaviour of the 4–O–5 bond.

For lignin conversion, transition metal oxides [14,15], phosphides [16,17,18], carbides [19,20], nitrides [21], and sulfides [22,23] are commonly used as catalysts. Specifically, transition metal phosphides could efficiently hydrodeoxygenate bio-oils due to their superior anti-poisoning properties [24]. In addition, the reactivity of monometal phosphides can be significantly improved by doping with precious metals [25,26]. Importantly, metal phosphides with both metal and acidic sites have excellent potential in C–O cleavage [27,28].

In this study, amorphous Ru-Co metallic phosphides supported on gas-phase SiO₂ are fabricated and their catalytic activity is evaluated using the 4–O–5 and α–O–4 linkages containing lignin-related compounds and lignin. The results show that Ru-Co metallic phosphide has an excellent selectivity to benzene. Most importantly, the total acidity correlates linearly with the hydrogenolysis yield (the sum of the phenol and benzene products). The mechanisms for the CoRuP/SiO₂ catalyzed hydrogenolysis of the 4–O–5 and α–O–4 bonds are proposed. Moreover, two-dimensional (2D) heteronuclear single quantum coherence (HSQC) nuclear magnetic resonance (NMR) spectroscopic analysis proves that CoRuP/SiO₂ could efficiently depolymerize etheric linkages in lignin.

2. Results and Discussion

2.1. Characterization of Catalysts

The broad diffraction hump located at ~22° observed for each catalyst is attributable to SiO₂ (Figure 1). However, no apparent diffraction peaks assignable to the metal phosphide were detected for all catalysts, indicating that the metal phosphide was highly dispersed or present in an amorphous form on SiO₂.

In accordance with the XRD results, the selected region electron diffraction pattern (Figure 2a) validates CoRuP/SiO₂ exists in amorphous structure. Moreover, the elemental distribution showed that Co, Ru, and P were highly dispersed on SiO₂ (Figure 2b). Previous studies have proven that the amorphous catalysts possess excellent texture structures, high catalytic activity and better dispersion [29,30].

The pore size distributions and N₂ adsorption-desorption isotherms of the gas-phase SiO₂ and its supported catalysts are presented in Figure 3a,b, respectively. All catalysts displayed kind IV isotherms with a H1-type hysteresis loop, which suggested the presence of uniform sized and shaped mesopores in the catalysts. Moreover, a type IV isotherms was displayed by the gas-phase SiO₂, but with a H3 type hysteresis loop, corresponding to a material formed by aggregated particles. It is well known that the amorphous gas-phase SiO₂ is composed of grain-like agglomerates and shows no the especial morphology.

As shown in

Table 1. Structural characteristics of the support and catalysts.

Sample	Surface Area (m2/g)	Pore Volume (cm3/g)
SiO2	300.86	0.038
NiP/SiO2	97.67	0.036
CoP/SiO2	94.59	0.011
RuP/SiO2	42.47	0.0004
NiRuP/SiO2	60.07	0.0065
CoRuP/SiO2	65.06	0.0054

, the gas-phase SiO₂ has the highest BET area (300.86 m²/g) and pore volume (0.038 cm³/g). Loading metal phosphide, especially ruthenium phosphide, onto SiO₂ markedly reduced the specific surface area and pore volume because of partial blocking of the pores by metal phosphide species and the collapse of the pore structure during the fabrication procedures.

In the Co 2p_3/2 region (Figure 4a), three significant peaks at 785.7, 782.0, and 778.5 eV are, respectively, assigned to satellite, Co²⁺, and Co^δ+ species. The Co^δ+ species are assigned to the Co present in the Co₂P and CoP phases with δ ≈ 0, while the Co²⁺ species are correlated with the cobalt phosphate resulting from the partial passivation on the surface of CoRuP/SiO₂ [31,32]. By comparing the binding energy of RuO₂ (~463.7 eV), the peak at ~461.8 eV comes from Ru species (Ru^δ+) [33]. Previous studies have shown that Ru^δ+ species in metal phosphides are effective in absorption and activation of hydrogen [34]. Regardless of whether in metal phosphide or corresponding monometal phosphide, the Co and Ru species exhibit no apparent shift in binding energies. This reveals that there is no notable electron transfer between Co and Ru species.

Table 2. Acid sites analysis by NH₃-TPD.

Sample	CoP/SiO2	NiP/SiO2	RuP/SiO2	CoRuP/SiO2	NiRuP/SiO2
NH3 acidity/mmol·g−1	1.1	1.1	2.3	2.4	3.2

lists the acidic properties of the catalysts according to their NH₃-TPD results in Figure 5. With the introduction of Ru, the total acidity preeminently increases (Table 2) and the desorption peaks greatly shift towards higher temperature (Figure 5). The peak at ~250 °C is primarily assigned to Brønsted acid sites arising from PO–H groups [35], while another peak located at ~370 °C might be ascribed to Lewis acid sites originating from the unreduced metal species [24]. This indicates that the Lewis acid sites are stronger than the Brønsted acid sites.

2.2. Catalytic Performance

2.2.1. Screening of Catalysts

As presented in Figure 6, these Ru-containing catalysts with lower BET areas and pore volumes (Table 1) showed high reactivity for C–O cracking, which is consistent with the excellent deoxygenation performance of RuP₂/SiO₂ [36] and indicates that BET area and pore volume are not determining factors influencing the hydrogenolysis of 4–O–5 linkage. In addition, these catalysts display obviously distinct selectivities in the hydrogenolysis products. For example, NiRuP/SiO₂ yielded more phenol, while CoRuP/SiO₂ is shown to be an active and selective catalyst for benzene with only a slight amount of by-products. The introduction of ruthenium preeminently improved the hydrodeoxygenation performance of cobalt phosphide. Most importantly, the total acidity exhibits a strong linear correlation with the hydrogenolysis yield of diphenyl ether, as shown in Figure 7. Nevertheless, neither weak nor medium acidity displayed a linear correlation with the hydrogenolysis yield.

2.2.2. Influence of Initial H₂ Partial Pressure on Hydrogenolysis Performance

Initial H₂ partial pressure significant influenced the distribution of products, as displayed in Figure 8. The hydrogenolysis conversion of diphenyl ether was only 19.2% in the absence of H₂, but reached 97.5% at 0.2 MPa, while the benzene yield achieved a maximum value of 59.4%. It is noteworthy that water can act as an important cocatalyst and provide active hydrogen species for hydrogenolysis reaction in the absence of H₂ [37]. The hydrogenolysis conversion gradually approached 100% and the benzene yield significantly decreased to 0 with increasing H₂ partial pressure to 0.6 MPa. However, the cyclohexane yield climbed from 2.0% to 48.7%. These findings agreed with the previous studies where the hydrogenolysis reaction preferentially occurred with deficient hydrogen species levels [38].

2.2.3. Hydrogenolysis Mechanism

To gain preliminary insights into the hydrogenolysis mechanism of C–O linkages over CoRuP/SiO₂, 3-methyl diphenyl ether was further selected as a probe compound, with the results shown in Figure 9. As the reaction proceeded, the content of 3-methyl diphenyl ether dropped dramatically during the initial 60 min and further diminished gradually to 1.7% at 100 min. Concomitantly, the contents of benzene and m-cresol increased equivalently in the first 60 min, following the stoichiometric coefficient of the hydrogenolysis reaction. However, a further elongation of the reaction time resulted in a decrease in the m-cresol content, which finally dropped to 6.0% at 100 min. In contrast, the benzene content reached an almost maximum value of 32.3% at 80 min and remained nearly constant for the remaining time. A small amount of cyclohexane was detected at 100 min, which indicated that the hydrogenation of benzene occurred. The content of toluene increased gradually during the whole reaction process and finally reached 23.1% without hydrogenated derivatives.

Biatomic active hydrogen (H∙∙∙H), originating from the activated H₂ adsorption on precious metal Ru, can be heterolytically split into a H⁻ and a H⁺ on metallic phosphides [37,39]. Subsequently, H⁺ can be trapped by the negatively charged P and PO species to afford Brønsted acidity [24,35,40], while H⁻ can transfer towards and deposit on Lewis acid sites, i.e., unreduced Ru and Co metal species [24,32,41]. The higher hydrogenolysis activity of CoRuP/SiO₂ might originate from the facile splitting of H₂ into H⁻ and H⁺ by its stronger Lewis acidity. Moreover, previous studies found that the H⁻ species on Lewis acid sites present much more active than the H⁺ species for the hydrogenolysis and hydrodeoxygenation reaction on metallic phosphides [24,35]. In contrast, water can act as an important cocatalyst to facilitate direct deoxygenation to afford aromatics [37].

Accordingly, the possible initiation reactions and reaction steps for the catalytic depolymerization of 3-methyl diphenyl ether by CoRuP/SiO₂ are proposed as shown in Figure 10 and Scheme 1, respectively. H⁺ attaching to the electronegative oxygen atom in 3-methyl diphenyl ether does not initiate the C–O cleavage due to the thermodynamic instability of intermediates m11 and m12, which reveals that the proton-assisted initiation reaction, as displayed in Figure 10a, does not easily occur [42]. Actually, the 4–O–5 C–O bond is not depolymerized, even in thermal phosphoric or formic acid [43].

However, even without water and Lewis site assistance (as depicted in Figure 10c,d), the depolymerization of the 4–O–5 linkage initiated by H⁻ attacking substituted aromatic carbon atom (Figure 10b) is more thermodynamically favorable due to the stability of the resulting intermediates, m21 and m31 [44]. Additionally, previous research proved that the hydrogenolysis of 3-methyl diphenyl ether, an unsymmetrical diaryl ether, preferentially occurred at the electron-deficient benzene-ring side [45] via intermediate m2, resulting in the formation of benzene and intermediate m21, as shown in Scheme 1. The successive H⁺ abstraction from the Brønsted acid site or H₂O by intermediate m21 leads to the generation of m-cresol, which is a dominant product and can react with absorbed H⁻ on metal to finally afford toluene via the thermodynamically favorable intermediate m23.

Furthermore, benzyl phenyl ether and dibenzyl ether were also subjected to hydrogenolysis over CoRuP/SiO₂, with their corresponding product distribution shown in

Table 3. Hydrogenolysis of benzyl phenyl ether and dibenzyl ether over RuCoP/SiO₂.

Entry	Substrate	Conv./%	Yield/%
1		98.8	8.7	0.6	21.7	34.8
2		99.0	23.5	0.1	-	52.8

and possible reaction pathways exhibited in Scheme 2 and Scheme 3, respectively. Similarly, both the principal steps for the depolymerization of dibenzyl ether and benzyl phenyl ether, α–O–4 linkage containing compounds, involve H⁻ attaching to α–C in the initial step, followed by the subsequent aliphatic C–O bond cleavage. It is generally accepted that an acid possessing a lower pK_a value facilitates the formation of the more stable corresponding conjugated base. Therefore, it allows us to deduce that the major pathway for benzyl phenyl ether hydrogenolysis involves the formation of intermediate b21 rather than b31 and b41, while that for the hydrogenolysis of dibenzyl ether prefers b31 to d31 as a dominant intermediate. In the current work, we centered our endeavors on reaction pathways that were in reasonable agreement with experimental results, while not refuting the existence of other complex reaction pathways.

2.2.4. Hydrogenolysis of Lignin over CoRuP/SiO₂

To evaluate the changes of etheric bonds in lignin and gain insights into the mechanisms of C–O depolymerization, 2D-HSQC-NMR characterizations of fresh dealkaline lignin and the corresponding residue were performed [46,47,48,49]. The lignin sample in Figure 11 displays the aliphatic ester region. As shown in Figure 11a, obvious cross-signals of C_α–H_α, C_β–H_β, and C_γ–H_γ (δC/δH = 67.2/4.5, 76.7/4.4 and 59.0/3.8 ppm, respectively) reveal that A is the predominant structural unit of lignin rather than B, C, and I. Notably, the signals of the A, B, C and I units in residue disappeared completely (Figure 11b). In comparison with our previous work using NiRuP/SiO₂ as a catalyst [50], CoRuP/SiO₂ can even efficiently catalyze the complete elimination of the methoxy group in lignin under the same reaction conditions, which confirms that CoRuP/SiO₂ possesses higher activity in cleaving the C–O bond or deoxygenation.

The organics obtained from the hydrogenolysis of lignin were detected by liquid chromatography-mass spectrometry, and the results are shown in

Table 4. The predominant products from hydrogenolysis of lignin.

Name	Yield (μg)
	Organic Phase	Aqueous phase
2,4-Dimethoxybenzaldehyde	1.0	-
2,3-Dihydroxytoluene	35.3	17.1
2,6-Dimethoxyphenol	29.6	15.9
2-Methoxy-4-vinylphenol	116.2	45.8
3-(3,4-Dimethoxyphenyl) propionic acid	5.1	3.3
4-Hydroxy-3-methoxyphenylacetone	116.7	57.1
4-Methylcatechol	488.5	194.7
Acetosyringone	2.9	2.4
Acetovanillone	93.6	42.5
Isoeugenol	4.3	-
Phenyl acetate	22.6	5.3

. The detected organics all contain hydroxyl groups except for 2,4-dimethoxybenzaldehyde and phenyl acetate. As shown in Table 4, the major hydrogenolysis products from dealkaline lignin are acetovanillone, 4-methylcatechol, 4-hydroxy-3-methoxyphenylacetone, and 2-methoxy-4-vinylphenol with yields of 136.1, 683.2, 173.8, and 162.0 μg, respectively. These main products have significant potential in the production of high value-added drugs and pharmaceutical intermediates. It is known that chronic pain and depression-like symptoms can be alleviated using 4-methylcatechol, by which a brain-derived neurotrophic factor is produced [51]. 4-Hydroxy-3-methoxyphenylacetone is a key intermediate for synthesizing carbidopa, which is an important therapeutic drug for Parkinson’s syndrome. Moreover, 4-hydroxy-3-methoxyphenylacetone can also be used as feedstock to produce compounds with biological activities for treating osteoporosis and inhibiting cell mitosis [52]. In addition to anti-inflammatory properties, 2-methoxy-4-vinylphenol is capable of causing cell cycle arrest, which gives it the ability to treat pancreatic cancer [53]. Acetovanillin can be used to treat inflammatory diseases and has been used to treat tracheal asthma [54].

3. Materials and Methods

3.1. Chemicals and Reagents

Ni(NO₃)₂·6H₂O (AR) was obtained from Oukai Chemical Co., Ltd., Xiangyang, China. Co(NO₃)₂·6H₂O (≥99.99%), diphenyl ether (≥99.9%), 3-methyl diphenyl ether (≥98%), benzyl phenyl ether (≥98%), dibenzyl ether (95%), ethyl acetate (99.5% GR), and n-dodecane (>99.0% GC) were purchased from Aladdin Biochemical Technology Co., Ltd., Shanghai, China and other chemicals were obtained from Macklin Biochemical Co., Ltd., Shanghai, China.

3.2. Catalyst Preparation

Co(NO₃)₂·6H₂O, RuCl₃·xH₂O, citric acid and ammonium hypophosphite were added to deionized water in turn to dissolve. The resulting solution was then impregnated on the gas-phase SiO₂. Subsequently, the mixture was aged at ambient temperature and dried at 383 K. The resulting solid was then heated to 927 K under vacuum and held at this temperature for 2 h. The obtained catalyst is labeled as CoRuP/SiO₂. The total load of metal is 20 wt.% and the molar ratios of nP/nM (M = Co and Ru) and nCo/nRu were 2/1 and 5/1, respectively. The NiRuP/SiO₂ and other catalysts were prepared in the same procedures as CoRuP/SiO₂.

3.3. Catalyst Characterization

Ammonia temperature-programmed desorption was conducted with an AutoChem™ II 2920 Chemisorption Analyzer (Micromeritics, Norcross, GA, USA). X-ray diffraction (XRD) patterns were recorded with a SmartLab SE instrument (Rigaku, Tokyo, Japan). X-ray photoelectron spectra were recorded using an ESCALAB instrument (Thermo Fisher, Waltham, MA, USA). Energy-dispersive X-ray spectroscopy and transmission electron microscopy (TEM) images were obtained on a Tecnai™ G2 F20 transmission electron microscope (Thermo Fisher, Waltham, MA, USA). Nitrogen adsorption/desorption examination was employed for calculating the special surface area, pore volume, and its distributions using a Micromeritics ASAP 2460 analyzer (Norcross, GA, USA).

3.4. Hydrogenolysis of Model Compounds

Typically, 5 mL of water, 0.2 mmol of reactant and 0.003 g of the catalyst were put into a reactor, which was then charged with 8 bar N₂ and 2 bar H₂ after exclusion of air with H₂. Reactions were conducted at 250 °C for 1 h while vigorously stirring. The reaction was terminated by cooling the reactor to ambient temperature using ice water. Ethyl acetate was used to retrieve the organic products from the reaction mixture. Quantitative analysis of the products was performed on a GC-FID employing n-dodecane as an internal standard. The hydrogenolysis conversion and product yield were calculated using the following formulas:

$$\mathrm{Hydrogenolysis} \mathrm{conversion}\%=\frac{\mathrm{moles} \mathrm{of} \mathrm{model} \mathrm{compound} \mathrm{reacted} }{\mathrm{moles} \mathrm{of} \mathrm{model} \mathrm{compound} \mathrm{supplied}}\times 100$$

(1)

$$\mathrm{Product} \mathrm{yield} \left(\mathrm{x}\right)\%=\frac{\mathrm{moles} \mathrm{of} \mathrm{C} \mathrm{atoms} \mathrm{in} \mathrm{product} \left(\mathrm{x}\right) }{\mathrm{moles} \mathrm{of} \mathrm{C} \mathrm{atoms} \mathrm{in} \mathrm{model} \mathrm{compound}}\times 100$$

(2)

3.5. Hydrogenolysis of Lignin

Due to the less hydrogenolysis reactivity of lignin than model compounds, rather harsh reaction conditions were employed. Typically, 6 mL of water, 0.1 g of dealkaline lignin, and 0.1 g of the catalyst were put into the reactor, which was pressurized in 1.0 MPa H₂ after exclusion of air with H₂. The temperature was then raised to 280 °C and maintained for 3 h under vigorous stirring. After termination of the reaction, the hydrogenolysis products were retrieved with ethyl acetate. The liquid products in the aqueous and organic phases were analyzed quantitatively by HPLC-TQMS with external standard method. The corresponding residue was characterized by two-dimensional nuclear magnetic resonance heteronuclear single quantum coherence (2D-NMR-HSQC) spectroscopy after drying.

4. Conclusions

In this study, an amorphous Co-Ru phosphide composite catalyst was fabricated and employed to hydrogenolyze lignin and its related model compounds. Under mild reactions, CoRuP/SiO₂ can efficiently hydrogenolyze C–O bonds and is highly selective for benzene with a small number of by-products. Importantly, the total acidity is a predominant factor for hydrogenolysis of the 4–O–5 linkage and shows a linear correlation with the hydrogenolysis yield. Regardless of water and Lewis site assistance, the depolymerization of 4–O–5 and α–O–4 bonds over CoRuP/SiO₂ might be preferentially initiated by H⁻ attacking the substituted aromatic carbon atom and the α–carbon atom, respectively.