Selective electrocatalytic hydrogenation of bio-oil to oxygenated chemicals via suppression of deoxygenation

Catalytic hydrogenation of bio-oil provides an avenue to produce renewable chemicals. To this end, electrocatalytic hydrogenation is especially interesting when powered using low-carbon electricity; however, it has to date lacked the needed selectivity: when hydrogenating bio-oil to oxygenated hydrocarbons, for example, it reduces the desired oxygenated groups (-OH and -OCH 3 ). Here we report that Rh and Au modulate electronic structure of Pt and steer intermediate energetics to favor the hydrogenation while suppressing deoxygenation using computational studies and in-situ spectroscopies. PtRhAu catalysts achieve a record 47% faradaic eciency (FE) and a partial current density (J p ) of 28 mA·cm -2 toward oxygenated 2-methoxycyclohexanol from lignin-derived guaiacol under room temperature and ambient pressure, representing 1.5x FE and 3.5x J p increases compared to the best prior reports. We further demonstrate an integrated lignin biorenery where wood-derived lignin oils are selectively hydrogenated and funneled to the oxygenated 2-methoxy-4-propylcyclohexanol using PtRhAu catalysts.


Introduction
Biomass-derived bio-oil containing lignin-derived aromatic hydrocarbons is potential alternatives to fossil feedstocks in chemical production [1][2][3][4][5] . Unlocking this potential will bene t from progress in the selective hydrogenation of lignin-derived aromatic monomers to oxygen-functionalized chemical motifs -buildingblocks of pharmaceuticals, natural products, and agricultural chemicals -under gentle conditions, such as in scalable systems functioning at or near ambient pressures and at modest temperatures [6][7][8][9][10] .
Thermocatalytic hydrogenation has been widely studied for the valorization of lignin monomers 11,12 . Thermocatalytic hydrogenation is an energy-intensive process conducted at high reaction temperatures (~100-500 o C) and hydrogen pressures (~1-200 bar), necessary because of the stable pi bonds of aromatic rings 13 . Unfortunately, present-day thermocatalytic hydrogenation methods have so far led to the deoxygenation of lignin monomers (i.e. they reduce oxygenated functional groups, OFGs) producing partially or completely deoxygenated products (e.g. phenol, cyclohexanol, and cyclohexane) 12,[14][15][16][17] . For this reason, the production of OFG-rich hydrocarbons has required a multiple-step process of hydrogenation and oxidation 12,18 .
One-step hydrogenation of lignin monomers, if it selectively retains the needed OFGs characteristic of lignin precursors, would reduce the number of chemical steps and appeal to the chemical industry due to the reduced carbon footprint of lignin monomers compared to petroleum-derived feedstocks 4,19,20 .
Here we selectively hydrogenate lignin monomers: we are able to retain the original OFGs in a one-step aqueous electrocatalytic hydrogenation (ECH) process. The ECH of biomass bene ts from mild operating conditions (ambient temperature and pressure), avoids organic solvents and H 2 , and offers routes to tune product selectivity 21,22 . Moreover, Pt-based heterogeneous catalysts prefer the production of cis isomers 23 .
A key di culty in an electrocatalytic approach is that deoxygenation is thermodynamically favored in the cathodic, reducing, and electron-rich environment 10,16,24,25 . As a result, high FE and current density toward OFG-rich products are challenging to achieve -a factor further compounded by the competing hydrogen evolution reaction 10 .
We sought therefore a route to develop new electrocatalysis with high selectivity in mind. We selected metals that would modulate the electronic properties of Pt, addressing the energetics of key reaction intermediates to achieve these goals simultaneously. Using the resultant new electrocatalysts, we selectively hydrogenate lignin monomers (i.e. guaiacol and syringol with OFGs such as -OH and -OCH 3 ) to OFG-rich chemicals at room temperature and ambient pressure. The oxygenated 2-methoxycyclohexanol (2MC) from guaiacol ( Fig. 1a) and syringol is an OFG-rich intermediate for pharmaceutical compounds (i.e., β-lactam antibiotics for HIV) 26 and synthetic perfumes 27 .
We further demonstrate the versatility of PtRhAu catalysts in an integrated lignin biore nery to obtain oxygenated chemicals directly from lignocellulosic biomass. We rst converted birch wood to ligninderived oils containing 4-propylsyringol (4PS) and/or 4-propylguaiacol (4PG) using aldehyde-assisted fractionation (AAF) followed by hydrogenolysis 5,28 . These lignin-derived oils were then selectively hydrogenated and funneled to a single OFG-rich chemical 2-methoxy-4-propylcyclohexanol for synthetic perfumes using our novel electrocatalysts under ambient conditions.

Results
Density functional theory (DFT) calculationsWe began by screening a range of single-metal catalysts (Pt, Rh, Au, Ag, Ni, Pd, Cu, Ru, and Ir) (Fig. 1b, Figs. S1-S3, and Tables S1-S3). DFT calculations reveal that an increase in the guaiacol adsorption energy (E ad-G ) is closely linked to an increased partial current density toward 2MC (Fig. 1b). This preliminary study highlighted Pt (E ad-G = 2.12 eV) and Rh (E ad-G = 2.67 eV) as potential candidates due to their high adsorption energies for guaiacol (Fig. 1b). We then moved on to binary systems, modulating Pt with Rh (20% atomic concentration for optimized PtRh) to simultaneously increase the adsorption energy of guaiacol and current density toward 2MC.
To suppress the undesired loss of OFGs, we sought to avoid the reduction of the C-OCH 3 bond, which is more prone to scission than C-OH (Fig. S4). We reduced the C-OCH 3 bond length on the catalyst surface by modulating PtRh using a third transition metal. Our DFT calculations indicate that Au (with an adsorbed guaiacol C-OCH 3 bond length of 1.360 Å) has a stronger ability to localize the ơ-electron and reduce the C-OCH 3 bond length compared to Pt (bond length = 1.365 Å), Rh (bond length = 1.378 Å), and other selected metals (Table S1). We predicted, therefore, that it would kinetically suppress the detachment of the -OCH 3 group. Combining the above two descriptors (adsorption energy of guaiacol and C-OCH 3 bond length), we obtained a contour diagram (Fig. 1c) that suggests improved performance using ternary PtRhAu catalysts.
Next, with the goal of further understanding the mechanism, we calculated the surface reaction network and the energetics of the intermediates ( Fig. 2a and Fig. S4). The energy pro le (Fig. 2b) indicates that the rate-determining step along the ECH pathway is the addition of the rst pair of hydrogen atoms. Both PtRh (1.64 eV) and PtRhAu (1.59 eV) catalysts lead to a decrease in hydrogenation energy compared to Pt alone (1.79 eV), indicating a synergistic role of Rh in hydrogenation (Fig. 2b). As for the undesired demethoxylation (Fig. 2c) and dehydroxylation (Fig. 2d) pathways, PtRhAu catalysts are thermodynamically less favorable than pure Pt, indicating that Au inhibits deoxygenation. Amongst Pt, PtRh, and PtRhAu, only the ternary PtRhAu catalysts have an ECH pathway (1.64 eV) that is more favorable than the demethoxylation pathway (1.82 eV). By comparing ve elementary steps -some relating to the activity (adsorption, hydrogenation, desorption) and others to selectivity (demethoxylation and dehydrogenation) -we concluded that the ternary PtRhAu catalysts showed particular promise (  Fig. S13), we achieved a FE of 47 ± 1% and a partial current density of 28 ± 0.5 mA·cm -2 toward 2MC using the PtRhAu catalyst at -0.15 V (Fig. S14). 1 H NMR (Fig. S15) analyses con rm the production of 2MC from guaiacol by ECH at 60 mA·cm -2 after 1 hour. PtRhAu presents 1.3x and 4.7x increases in FE toward 2MC compared to PtRh (35 ± 1%) and Pt (10 ± 1%). The current density toward 2MC is 3.5x greater than the prior reports (7.95 mA·cm -2 , Table S4) that used a Ru catalyst 29 .
Tafel analysis of 2MC production from guaiacol by ECH gives the slopes of 100, 280, and 150 mV·dec -1 for PtRhAu, PtRh, and Pt respectively (Fig. 4c), indicating that ECH is improved with PtRhAu. Comparing the product distribution of PtRh and Pt ( Fig. 4d and Fig. S16) indicates that incorporation of Rh increases selectivity toward 2MC without increasing the production of organic by-products, suggesting that Rh promotes the selective hydrogenation of guaiacol toward 2MC. With the further incorporation of Au, the ternary PtRhAu catalysts enable higher FE toward 2MC and suppresses the formation of deoxygenated by-products (methoxy-cyclohexane, cyclohexanone, and cyclohexanol) (Fig. 4d). The optimal PtRhAu catalyst converts 91% of the guaiacol with a 74% selectivity giving a 68% yield of 2MC (Fig. S17).
We found that PtRhAu catalyst provides a high stereoselectivity toward cis-2-methoxycyclohexanol with 99.5% selectivity ( Fig. 4e and Figs. S18-S20). The preferential production of the thermodynamically unfavored cis isomer of 2MC on Pt-based electrocatalysts is an interesting result and distinct from the observations from thermocatalysis 23,30 .
We explored the stability of the ternary catalysts by performing ECH on guaiacol at a constant current density of 60 mA·cm -2 . The PtRhAu catalyst maintains a FE of over 40% toward 2MC at -0.15 V for over 12 hours of continuous operation (Fig. 4f). We also selectively hydrogenated the lignin monomer syringol (2,6-dimethoxyphenol) and retain one methoxy group (-OCH 3 ) to obtain the same product 2MC with a FE of 36% at a total current density of 60 mA·cm -2 (Fig. S21). This selective retention of one methoxy group provides potential to funnel wood-derived lignin monomer mixtures into a single product, which is demonstrated in later section using wood-derived lignin monomer mixtures.
Mechanistic investigations. By tracking changes in X-ray absorption near edge structure (XANES) and extended X-ray absorption ne structure (EXAFS) spectra (Figs. 5a-d), we were able to examine the local electronic and coordination structure of PtRhAu catalysts. XANES plots for Pt, Au, and Rh (Figs. 5a-c) reveal the metallic Pt, Au, and Rh phase in all samples, consistent with the results of X-ray photoelectron spectroscopy (XPS) (Fig. S22). We also investigated the catalyst stability during the ECH of guaiacol at a current density of 60 mA·cm -2 using in-situ XAS (Fig. S23). Pt/Au L 3 -edge and Rh K-edge XANES of all the catalysts show no obvious change in the valence state of Pt, Au, and Rh during the ECH of guaiacol.
EXAFS spectra of the Rh K-edge for PtRhAu (Fig. 5d) reveal a signi cant change in atomic bonding of the Rh atom compared to pure Rh. PtRhAu and PtRh catalysts exhibit a decreased Rh-M (M = Pt, Au, and Rh) coordination number and an increased Rh-M interatomic distance (PtRhAu > PtRh > Rh) compared to pure Rh, suggesting the formation of a PtRhAu solid solution alloy 31 . This increased Rh-M interatomic distance is consistent with the lattice expansion observed using STEM and XRD (Fig. 3). Using in-situ EXAFS to investigate atomic bonding near the Rh atom of PtRh and PtRhAu catalysts (Fig. S24), we found no obvious changes in the Rh-M coordination number and interatomic distance, indicating that PtRhAu alloy structure is stable during the ECH of guaiacol.
We investigated the ECH mechanism by evaluating reaction intermediates using electrode potentialdependent in-situ Raman spectra (Fig. 5e) and in-situ infrared re ection-absorption spectroscopy (IRRAS) (Fig. S25). We associate the characteristic peaks from 0 to -0.37 V versus RHE with the ECH of guaiacol. PtRhAu has a wider electrochemical potential window (0 to -0.37 V) compared to Pt (0 to -0.14 V) (Fig.  S26). The in-situ Raman spectra of PtRhAu show intensive peaks associated with the adsorption (CCH wag, in-plan C-H bending, and C=C of aromatic ring) 32,33 and ECH (hydrogenation) 34 of guaiacol on catalyst surface, indicating a favorable ECH of guaiacol. The PtRhAu catalyst shows decreased intensity of peaks associated with C-O cleavage (C-O stretch and C-OCH 3 stretch) 33 in a potential range of 0 and -0.18 V compared to Pt (Fig. S26), indicating the suppression of deoxygenation. In contrast, Pt shows a strong peak located at 1161 cm -1 and evident deoxygenation at -0.09 V (Fig. S26). In-situ IRRAS of PtRhAu (Fig. S25) shows intense peaks related to the adsorption of guaiacol and 2MC product, along with suppressed peak related to C-O cleavage, con rming suppressed deoxygenation on PtRhAu catalyst surface.
This lignin-derived oil was then selectively hydrogenated to the oxygenated 2-methoxy-4propylcyclohexanol (1, having -OH and -OCH 3 retained) with suppressed deoxygenation using PtRhAu catalyst. The PtRhAu catalyst shows superior performance on upgrading this lignin monomer mixture ( Fig. 6c and Fig. S30 and S31), achieving a 68% selectivity (up to 72% when using puri ed lignin bio-oil containing wood-derived 92 mol% 4-propylsyringol, Fig. S35) toward the target product 1 at a 96% conversion rate during a 3-hour continuous reaction. The FE toward product 1 reaches 19% during the initial rst hour of reaction with an applied current density of 20 mA·cm -2 (Fig. 6c).

Conclusion
By alloying Pt with Rh and Au, we realized ternary PtRhAu catalysts that selectively catalyze the hydrogenation of lignin-derived bio-oil (i.e. guaiacol), to the oxygenated chemicals (i.e. 2methoxycyclohexanol), while suppressing deoxygenation. Density functional theory calculations suggests that the Pt catalysts modulated with Rh and Au steer the intermediate energetics to increase guaiacol coverage, foster hydrogenation, and suppress deoxygenation. Structural characterization, X-ray studies, in-situ Raman/infrared re ection-absorption spectroscopy (IRRAS), and electrochemical measurements further validate the role of Rh and Au in improving catalytic hydrogenation performance and protecting oxygenated groups. We further achieved an integrated lignin biore nery from birch wood to oxygenated chemical, indicating that PtRhAu electrocatalysts can selectively hydrogenate and funnel wood-derived lignin monomers to an oxygen-functionalized chemical with suppressed deoxygenation reaction. The strategy suggests a means to valorize biomass to oxygen-functionalized chemicals.

Methods
Computational details. All calculations were carried out using the Vienna ab-initio simulation program (VASP) 36,37 . Detailed theoretical methods can be found in Supplementary Information. Catalyst synthesis. In a typical procedure, PtRhAu working electrode was synthesized by coelectrodepositing metals onto a Ti foam (1 x 1 cm 2 ) according to a slightly modi ed version of a previously reported method 38 . The electrodeposition was conducted in a three-electrode electrochemical cell using a potentiostat (Metrohm-Autolab, PGSTAT204). Carbon paper and an Ag/AgCl electrode (saturated with KCl) were used as the counter and reference electrodes, respectively. The substrate was subjected to 50 cycles of cyclic voltammetric scans in the range of potential between -0.5 to 1. according to the slightly modi ed version of a previously reported method 38 . The electrodeposition was conducted in a three-electrode electrochemical cell using a potentiostat (Metrohm-Autolab, PGSTAT204). Carbon paper and an Ag/AgCl electrode (saturated with KCl) were used as the counter and reference electrodes, respectively. The electrolyte consisted of an aqueous solution of 0.5 M Na 2 SO 4 containing 10 mM of metal precursor solution as needed. After the electrodeposition with 50 cycles of cyclic voltammetric scans in the range of potential between -0.5 and 1.7 V at a scan rate of 0.1 V·s -1 , the working electrode was washed with D.I. water and dried in ambient pressure and temperature.
Materials characterization. The morphology of the electrodes was characterized using transmission electron microscopy (TEM, Hitachi HF-3300) and scanning transmission electron microscopy (STEM, FEI Titan 80-300 HB) equipped with an electron energy loss spectroscopy (EELS) detector. The crystal structures were determined using synchrotron X-ray diffraction analysis (the incident X-ray wavelength of 0.6909 Å), equipped with a large Debye-Scherrer camera in the BL-12B2 beamline (Spring-8, National Synchrotron Radiation Research Center (NSRRC), Japan) in which the electron storage ring was operating at 8.0 GeV. The catalysts were deposited on carbon papers to collect the XRD patterns. The XRD patterns were calibrated using CeO 2 standard and modulated to that with a wavelength of 1.5413 Å using the software referred to as "Winplotr". In-situ X-ray absorption spectroscopy (XAS) were collected using the 9BM beamline of the Advanced Photon Source (Argonne National Laboratory, IL, United States). In-situ Raman spectroscopy was conducted using a Renishaw in Via Raman microscope, equipped with a 785 nm laser.
Electrochemical performance and product analysis. Electrocatalytic measurements were conducted in a three-electrode system equipped with Pt as counter electrodes using an electrochemical station (Metrohm-Autolab, PGSTAT204). All potentials were measured against an Ag/AgCl reference electrode (saturated KCl, BASi) and converted to the RHE reference scale using the following equation: Linear sweep voltammetry (LSV) measurements were performed by scanning the potential in the negative direction at a scan rate of 5 mV·s -1 , while the electrolyte was magnetically stirred at 1000 RPM. The ECH of the lignin monomer was conducted using a two-chambered H-cell with Na on 117 membrane as the separator. All working electrodes had a geometric surface area of 1 cm 2 . 20 mL of 0.2 M HClO 4 containing 120 mM guaiacol was used as the electrolyte in cathode. We also performed bulk ECH of guaiacol using PtRhAu catalyst in 1 L of electrolyte for a continuous reaction of 12 hours.
The liquid products were quanti ed using a gas chromatography-mass spectra (GC-MS) (PerkinElmer Clarus 680) equipped with a Stabilwax column (fused silica, Restek) and a 700 MHz Agilent DD2 nuclear magnetic resonance (NMR) spectrometer. 4-propyl-cyclohexanone was used as the internal standard. Prior to the product analysis using GC-MS, 900 uL of electrolyte solution was collected from the reaction solution and extracted with 300 uL of dichloromethane. For NMR characterizations, 3 mL of electrolyte solution was extracted with 1 mL of chloroform-D. The faradaic e ciency (FE) toward 2MC was calculated using the following equation: where n 2MC is the total amount of 2MC (in moles), F is the faradaic constant, I (in amperes) is the current, and t (in seconds) is the time for the constant current.
Upgrading birch wood to lignin monomer oils. Detailed experimental methods are in Supplementary Information.
Electrocatalytic hydrogenation of wood-derived lignin monomers. The ECH of the wood-derived lignin monomers was conducted using a two-chambered H-cell with Na on 117 membrane as the separator. All working electrodes (PtRhAu deposited on carbon felt) had a geometric surface area of 1 cm 2 . 20 mL of 0.2 M HClO 4 containing 6 mM lignin monomers mixture (4.8 mM 4PS and 1.2 mM 4PG) was used as the catholyte. The liquid products were quanti ed using the gas GC-MS (PerkinElmer Clarus 680) by following the similar procedure described previously. The liquid products were further characterized using 700 MHz Agilent DD2 nuclear magnetic resonance (NMR) spectrometer. The faradaic e ciency (FE) toward 2M4PC was calculated using the following equation: where n 2M4PC is the total amount of 2M4PC that is 2-methoxy-4-propylcyclohexanol (in moles), I (in amperes) is the current, t (in seconds) is the duration for the constant current applied, C 4PS is the conversion of 4PS, and C 4PG is the conversion of 4PG.
The overall conversion rate of lignin monomer oil was calculated using the following equation:     Integrated lignin valorization from birch wood to an oxygenated product 2-methoxy-4-propylcyclohexanol.

Declarations
a, Integrated lignin valorization process for high-value 2-methoxy-4-propylcyclohexanol from birch wood. 2-methoxy-4-propylcyclohexanol (1, red) is the target structure in bold. b, Mass balances during the propionaldehyde assisted fractionation of lignocellulosic biomass, as performed on birch wood. All the numbers are provided as weight percentages. The provided weight percentages of the sugars, stabilized sugars, furfural, hydroxymethylfurfural (5-HMF), stabilized lignin, and lignin monomers have been corrected for the mass of the stabilizing group, hydration, dehydration, or hydrogenation, to match their initial structure in the native biomass. Abbreviations: hydroxymethylfufural (HMF), acid soluble lignin (ASL). c, Concentration evolution of lignin monomer mixture containing 4-propylguaiacol (2, blue) and 4propylguaiacol (3, black) and target product 2-methoxy-4proplycyclohexanol (1, red) over 3 hours. (Inset) Electrocatalytic hydrogenation on PtRhAu catalyst at a current density of 20 mA·cm-2, showing the faradaic e ciency (FE) at 1 hour, conversion rate (C%), and product selectivity (S%) after 3 hours of reaction.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.