Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

# Au–Pd separation enhances bimetallic catalysis of alcohol oxidation

## Abstract

In oxidation reactions catalysed by supported metal nanoparticles with oxygen as the terminal oxidant, the rate of the oxygen reduction can be a limiting factor. This is exemplified by the oxidative dehydrogenation of alcohols, an important class of reactions with modern commercial applications1,2,3. Supported gold nanoparticles are highly active for the dehydrogenation of the alcohol to an aldehyde4 but are less effective for oxygen reduction5,6. By contrast, supported palladium nanoparticles offer high efficacy for oxygen reduction5,6. This imbalance can be overcome by alloying gold with palladium, which gives enhanced activity to both reactions7,8,9; however, the electrochemical potential of the alloy is a compromise between that of the two metals, meaning that although the oxygen reduction can be improved in the alloy, the dehydrogenation activity is often limited. Here we show that by separating the gold and palladium components in bimetallic carbon-supported catalysts, we can almost double the reaction rate compared with that achieved with the corresponding alloy catalyst. We demonstrate this using physical mixtures of carbon-supported monometallic gold and palladium catalysts and a bimetallic catalyst comprising separated gold and palladium regions. Furthermore, we demonstrate electrochemically that this enhancement is attributable to the coupling of separate redox processes occurring at isolated gold and palladium sites. The discovery of this catalytic effect—a cooperative redox enhancement—offers an approach to the design of multicomponent heterogeneous catalysts.

## Main

Supported gold (Au)–palladium(Pd) nanoalloys are effective catalysts for redox reactions8, showing considerable rate enhancement compared with monometallic analogues, an effect referred to as synergy8,9,10,11,12,13,14,15. So far, there has been no definitive demonstration of the origin of this effect, which we have investigated for the oxidative dehydrogenation (ODH) of alcohols15,16, using the ODH of hydroxymethylfurfural (HMF) in aqueous alkaline solution as a model reaction to study the cooperative relation between Au and Pd nanoparticles.

Two monometallic Au and Pd catalysts and a bimetallic Au–Pd catalyst (Au:Pd atomic ratio, 4:1) were synthesized using sol immobilization (Extended Data Fig. 1) and supported on Vulcan XC-72R carbon (C), denoted Au/C, Pd/C and Au–Pd/C, respectively. These catalysts were investigated for HMF oxidation and reacted as a slurry (Extended Data Fig. 1e), as well as a physical mixture of Au/C and Pd/C (Fig. 1a, b, Extended Data Fig. 2) in aqueous sodium bicarbonate (NaHCO3; pH 8.7): the total moles of Au and Pd, and the mass of the carbon support were constant across experiments. The HMF conversion and product yields were monitored as a function of time, with the monometallic Pd/C catalyst showing limited activity over the duration of the experiment. Nonetheless, the initial reaction rate, which was determined after 5 min over a physical mixture of the two monometallic catalysts, was far greater (k = 4.9 × 10−5 M s−1) than that observed with the monometallic Au/C catalyst (k = 2.5 × 10−5 M s−1) and the monometallic Pd/C catalyst (k = 2.7 × 10−6 M s−1), which could therefore not be explained simply as an additive effect. Catalytic activity immediately increased after the addition of Pd/C to a reaction mixture containing Au/C (Extended Data Fig. 2f) and was not attributed to the addition of carbon alone (Extended Data Fig. 2g). Notably, the rate of HMF conversion in the presence of the Au/C and Pd/C physical mixture was greater than that observed over the Au–Pd/C alloy (k = 3.8 × 10−5 M s−1).

The differences in catalytic activity are further highlighted when the concentration of the terminal product formed (furan dicarboxylic acid (FDCA)) is considered (Fig. 1b). The FDCA produced over the physical mixture, after 2 h of reaction, was almost 9.5 times greater than that produced over the Au/C catalyst alone. In multistep processes, such as this one, rate enhancements become multiplicative over sequential reactions. Thus, enhancement occurs when other oxygen-containing functionalities are oxidized. The trend in reactivity of these catalysts is amplified further when the estimated turnover frequencies (TOFs) of the materials are considered (Extended Data Fig. 2j).

We next investigated the ODH of the intermediates in HMF oxidation, 5-hydroxymethyl-2-furan carboxylic acid (HMFCA) and 5-formyl-2-furan carboxylic acid (FFCA), as well as a range of other alcohols (Fig. 1c, Extended Data Fig. 3), and observed the same trends in  catalytic activity to that observed for HMF oxidation. Analysis of post-reaction solutions confirmed that these rate enhancements were not the result of metal leaching in any of the reactions, and scanning transmission electron microscopy X-ray energy dispersive spectroscopy (STEM-XEDS) confirmed that no detectable metal migration occurred in reactions over the Au/C and Pd/C physical mixture (Extended Data Fig. 4).

We subsequently compared the efficacy of the highly conducting XC72-R carbon support to a semiconducting (titanium dioxide (TiO2)) and insulating (boron nitride (BN)) support17,18 (Fig. 1d, Extended Data Fig. 3e). When using a physical mixture of Au/C and Pd/BN, the total yield of products formed was comparable to the sum generated in the reactions using Au/C and Pd/BN independently (Au/C + Pd/BN, 20.3%; Au/C, 17.9%; Pd/BN, 1.1%). When Pd/TiO2 was combined with Au/C, product yields notably greater than the sum of Au/C and Pd/TiO2 when used independently were observed (Au/C + Pd/TiO2, 29.2%; Au/C, 17.9%; and Pd/TiO2, 0.5%). Support effects in heterogeneous catalysis have many dimensions, but these experiments show that the rate enhancement correlates well with the ability of the support material to accept and conduct electrons to and from the surface reaction sites. Although TiO2 provides limited conductivity and a conduction band edge at approximately −0.37 V versus a reversible hydrogen electrode (RHE) for anatase structured TiO2 that can accept electrons from the HMF oxidation step, the higher bandgap of BN acts as a barrier to this transport.

Hence, we propose a reaction scheme (Fig. 2) that explains the cooperative role of Au and Pd in alcohol and (hydrated) formyl oxidation reactions. Building on current understanding19,20, we consider that the oxidation of HMF to HMFCA proceeds through a dehydrogenation reaction on the Au surface, which generates two protons in solution and two electrons that are transferred, via the support material, to Pd2+ where they are consumed in an oxygen reduction reaction (ORR); the catalytic cycle is completed by the rapid reaction in the liquid phase of the product hydroxide ions with protons. Increasing the pH of the reaction mixture affects the reaction rate by activating the substrate towards oxidation but not the electron-transfer mechanism. The oxidation of reaction intermediates HMFCA and FFCA are considered to proceed in the same way, terminating at FDCA. Alcohol oxidation over Au-supported catalysts is therefore proposed to be influenced by the rate of the coupled ORR occurring on Pd and the ease with which electrons can migrate about the system, which is dictated to a large extent by the conductivity of the support used. The lower activity observed over the Au–Pd/C catalyst can be explained by the change in the redox properties of Pd, when alloyed with Au21; the alloying of Au with Pd reduces the rate of the ORR on Pd sites. This hypothesis is further supported by additional testing data, which confirm that ODH on Au sites is limited by the ORR taking place on Pd sites (Extended Data Fig. 2h). Over a physical mixture of Au/C and Pd/C, as the mass of Pd/C is increased and the mass of Au/C is kept constant, a linear increase in ODH activity was observed, which could not be attributed to any direct contribution from the additional Pd or C present in the reaction. This also explains why the rate of reaction over both the monometallic Au/C catalyst and Au/C + Pd/C physical mixture is influenced by, but is not directly proportional to, oxygen (O2) pressure (Extended Data Fig. 2i). Oxygen pressure is proposed to influence the rate of the ORR, which, consequently, affects the observed rate of the ODH. Thus, the relationship between the O2 pressure and the ODH activity is indirect and limited by other factors, such as electron transport (in the physical mixture) or competitive adsorption of substrate and O2 (in the monometallic Au catalyst). It should, however, be noted that this enhancement, as O2 pressure is increased, could be attributed to gas–liquid mass transfer, and requires further exploration.

To confirm the critical role of the electron transfer, analogous electrocatalytic experiments were performed. The electrocatalytic oxidation of HMF is well established and the trend observed in peak current density from cyclic voltammetry (Fig. 3a) for the Au/C, Pd/C and Au–Pd/C catalysts aligns with that in previous studies22. The trend in electrocatalytic current density across the catalysts correlates strongly with the trend in thermal catalytic rate (Extended Data Fig. 6a).

Electron transfer between Au/C and Pd/C was demonstrated by externally short-circuiting Au/C and Pd/C electrodes together in a single-chamber cell. Under aerobic conditions (Fig. 3b, Extended Data Fig. 6c), the current was measured to flow from the Au/C electrode to the Pd/C electrode. The observed drop in current was attributed to the mass-transfer limitation of O2 from the solution to Pd electrode interface. The current density was near zero in the absence of HMF. A control reaction where Pd/C was replaced by C showed a marked decrease in the current density (from 715 µA cm−2 to 239 µA cm−2 at 30 min), demonstrating the critical role of Pd as a reaction partner in consuming electrons generated by HMF oxidation on Au/C. Repeated switching between a nitrogen (N2) atmosphere and an O2 atmosphere in the cell every 60 min led to concomitant switching of the current density between approximately 100 µA cm−2 and 600 µA cm−2 (Fig. 3b), demonstrating the essential role of O2 in the electrocatalytic reaction.

Linear sweep voltammetry of Au/C and Pd/C electrodes under O2 within the potential window reported for the ORR23 was performed in the absence and presence of HMF (Fig. 3c). In the absence of HMF, both Pd and Au showed clear reduction peaks at 0.72 V, attributable to the ORR. However, the ORR peak current of Au/C decreased substantially in the presence of HMF, but that of Pd/C was largely unaffected. This suggests that under the reaction conditions, the Au sites predominantly catalyse dehydrogenation, whereas Pd sites continue to catalyse oxygen reduction. The mechanistic role of electron transfer between the metals was further demonstrated using a dual-chamber electrochemical cell where the Au/C and Pd/C electrodes were physically separated by an anion exchange membrane, but electrically connected via an external circuit, with Au and Pd electrodes under an atmosphere of N2 and O2, respectively (Fig. 3d, Extended Data Fig. 6d). The short-circuit current between the electrodes was measured over time while monitoring the HMF conversion and selectivity in the Au/C chamber. The total molar electron transfer between the electrodes was then compared with that expected from HMF oxidation stoichiometry to determine the fraction of the ORR reaction occurring through a cooperative electron transfer from the Au/C to Pd/C. With identical weight loadings of Au and Pd on their respective electrodes, 65% of the ORR reaction occurred by electron transfer from Au/C to Pd/C, increasing to 82% on doubling the Pd loading. It has been demonstrated that: (1) no current was generated in the absence of HMF; (2) Au under N2 alone showed low activity towards HMF conversion; and (3) the yield of HMFCA was substantially decreased when the two electrodes were not electrically connected in the dual-chamber cell (Extended Data Fig. 6e). These electrochemical experiments identify the role of each metal and the nature of these cooperative redox enhancement (CORE) effects, supporting our hypothesis that the observed enhancement occurred through physical separation of the catalytic function, which was critically facilitated by electrical connectivity through the support.

The activity of a material possessing spatially separated Au and Pd nanoparticles on the same support grain for HMF ODH (denoted Au=Pd/C, with morphology confirmed by STEM-XEDS, Extended Data Fig. 7) was identical to that of the physical mixture, an effect we consider is owing to the highly conducting nature of the carbon support as demonstrated in the experiments where we have added additional carbon (Fig. 1e)

As we concluded that spatially separating the Au and Pd provides the origin of the rate enhancement, we reasoned that synthesizing nanoparticles comprising phase-separated Au and Pd regions within single nanoparticles (that is, not alloyed) would provide an even greater rate enhancement. We therefore synthesized small clusters of Pd situated on the surface of Au nanoparticles (Janus-like nanoparticles), denoted Au@Pd/C (Fig. 4j–l). This is evidenced through comparison with micrographs of the Au/C catalyst (Fig. 4a–c) and is in stark contrast to the particles present in the alloyed Au–Pd/C catalyst (Fig. 4g–i). The material was evaluated in the thermocatalytic reactions (Fig. 1a–c) and probed by cyclic voltammetry (Fig. 3a). In both cases, increased activity was observed, as predicted by our proposed model.

The validity of this concept was further evidenced when the performance of the Au@Pd/C catalyst and a physical mixture of Au/C and Pd/C catalysts was assessed over multiple uses. For the Au@Pd/C sample, we noted a slight loss in activity after use, which we consider was due to the formation of some surface Au–Pd alloy (Extended Data Fig. 8), which correlated with a decrease in catalytic activity over subsequent uses (Extended Data Fig. 9a, c). Analogous STEM-XEDS analysis of the used physical mixture of Au/C and Pd/C catalysts confirmed that there was no detectable metal migration or alloying (Extended Data Fig. 4). This correlated well with the reactivity data as the performance of the Au/C and Pd/C physical mixture remained stable over three uses (Extended Data Fig. 9b, d). The loss of catalytic activity with the formation of some alloy in the Au@Pd/C sample after several uses further strengthens our view of the role of Au in inhibiting the redox properties of Pd; as a higher proportion of Pd becomes alloyed with Au, the rate of the Pd redox cycle would also be diminished and hence its activity for the ORR.

It seems that by removing the need for electron transfer to occur through the carbon support, thus reducing any potential resistance, the reactivity of Pd towards ODH on the Au sites could be further increased. However, it should be noted that, in general, the Pd particles present in the Au@Pd/C catalyst are smaller than those present in the analogous Pd/C catalyst (Extended Data Fig. 8e). Although we suspect that these smaller Pd particles would provide a lesser contribution towards direct ODH, it is likely that the change in particle size could also influence the ORR rate, which we have confirmed to be rate limiting for ODH (Extended Data Fig. 2h).

Although we recognize that there are some mechanistic features that have not yet been fully addressed, we consider that for the oxidations described, the active sites on the Au and Pd nanoparticles work through redox cooperation. This CORE effect—controlled largely by electron transfer, as opposed to molecular transport—is, to our knowledge, a wholly new observation in heterogeneous catalysis and requires a conductive support; the electrochemical experiments are consistent with this view. We propose that the rate of the Pd redox cycle and ease with which electrons can migrate about the system ultimately dictates the intrinsic catalytic activity observed. For HMF oxidation, the primary model reaction studied, our Au@Pd/C catalyst is among the most active reported so far (Extended Data Table 1). We consider that the discovery of CORE effects will provide an additional dimension to research into bimetallic catalysts and provides a new foundation for the development of multicomponent heterogeneous catalysts.

## Methods

### Chemicals

Acetic acid (Sigma-Aldrich, purity >99.7%); boron nitride (Sigma-Aldrich, powder, 98%); chloroauric acid (HAuCl4·3H2O, Strem Chemicals, 99.8%); distilled water (Millipore, 18.2 MΩ cm at 25 °C); ethanol (Fisher Scientific, >99%); 5-formyl-2-furoic-acid (Tokyo Chemical Industry, >98%); glycerol (Sigma-Aldrich, >99.5%); graphene nanoplatelets (Alfa Aesar); hydroxymethylfurancarboxylic acid (Carbosynth, > 97%); 5-hydroxymethylfurfural (Sigma-Aldrich, >99%); 2,5-furandicarboxylic acid (Sigma-Aldrich, 97%); molecular oxygen (BOC, >99.95%); nafion (Sigma-Aldrich, 5 wt% in lower aliphatic alcohols and water, contains 15–20% water); palladium chloride (PdCl2, Sigma-Aldrich, >99.9%); polyvinyl alcohol (PVA, Sigma-Aldrich, 80% hydrolysed); 1-propanol (Sigma-Aldrich, anhydrous, 99.7%); 2-propanol (Sigma-Aldrich, anhydrous, 99.5%); 1,3-propanediol (Sigma-Aldrich, >99.6%); sodium bicarbonate (Fisher Scientific, >99.5%); sodium borohydride (NaBH4, Sigma-Aldrich, 99.99%); sodium hydrogen carbonate (Fisher Scientific, >99.5%); titanium dioxide P25 (Degussa, ≥99.5%); Vulcan XC72R (Cabot).

### Materials preparation

#### Synthesis of 1.75 wt% Au/C, 0.25 wt% Pd/C and 1 wt% Au–Pd/C catalysts by sol immobilization

Aqueous solutions of the metal precursors, PdCl2 (10 mg ml−1) or HAuCl4·3H2O (12.25 mg ml−1), were added to water (H2O; 140 ml) and stirred vigorously. PVA (mass of PVA/mass of Au, Pd or Au and Pd = 1/1) was subsequently added to the solution. A freshly prepared NaBH4 solution (0.15 M, NaBH4/total metal = 4/1, mole/mole) was added to the solution immediately, forming a sol. After this, the support material was added (0.5 g) was added to the colloidal solution with stirring facilitating the immobilization of the metal nanoparticles. After 30 min, the solid catalyst was recovered by filtration and washed with 500 ml distilled water to remove Na+, BH4 and BO2. Finally, the catalysts were dried at 110 °C for 16 h under static air. Schematics of these synthesis methods are given in Extended Data Fig. 1.

#### Synthesis of the 1 wt% Au@Pd/C catalyst by sol immobilization

The procedure used for the synthesis of this material is similar to that listed in the previous section. However, for the synthesis of this catalyst, Au and Pd colloids were prepared separately. The metal precursors were combined with analogous quantities of PVA and reduced with an analogous quantity of NaBH4, in separate beakers. Once the colloids had been synthesized, they were combined into one beaker and the desired quantity of support was immediately added. After 30 min of ageing, the solid catalyst was recovered by filtration and washed repeatedly with 500 ml distilled water to remove Na+, BH4 and BO2. Finally, the catalysts were dried at 110 °C for 16 h under static air. A schematic of this preparation is also shown in Extended Data Fig. 1.

#### Synthesis of the 1 wt% (Au=Pd)/C catalyst by sol immobilization

The procedure used for the synthesis of this material is similar to that listed in the above section, but with a two-step sequential reduction. First, Pd/C was synthesized by adding PdCl2 solution to 140 ml H2O, followed by the addition of analogous quantities of PVA and NaBH4, and 1 g carbon. This Pd/C catalyst was recovered by filtration, washed repeatedly with 500 ml distilled water, and then dried at 110 °C for 16 h under static air. Second, (Au=Pd)/C was synthesized by adding HAuCl4·3H2O solution to 140 ml H2O, followed by the addition of analogous quantities of PVA and NaBH4 to that used in the generation of the Pd/C material; subsequently the previously obtained Pd/C was added. The (Au=Pd)/C catalyst was recovered by filtration, washed repeatedly with 500 ml distilled water, and then dried at 30 °C for 16 h in a vacuum oven.

### Catalyst testing

#### Testing of catalysts in thermocatalytic experiments

The majority of the thermocatalytic testing in this work utilized HMF as the substrate. HMF oxidation was carried out in a glass Colaver reactor (50 ml). In a typical reaction, an aqueous solution (16 ml) consisting of HMF (0.1 M) and NaHCO3 (0.4 M) was heated to 80 °C under constant stirring (1,000 rpm). After temperature stabilization was achieved, the catalyst(s) (HMF/metal = 200/1, mol/mol) was added and the reactor was purged with O2 five times before the reaction was typically run for 2 h. The reaction was conducted under an atmosphere of 3-bar O2; this was continually fed in a semi-batch manner to maintain a constant head pressure in the reactor. Re-use experiments were also conducted for some of the catalysts. For this, the catalyst was removed and washed with deionized water (500 ml) and acetone (250 ml) after each experiment, and dried at 110 °C for 16 h before being re-used in the experiment. Analyses and quantification of post-reaction solutions was carried out using a high-performance liquid chromatography (HPLC) system (Agilent Technologies 1200 series) equipped with a diode array detector. A Hi-Plex H (300 mm × 7.7 mm) column was used to separate the products with 5 mM aqueous sulfuric acid (H2SO4) solution as the mobile phase, at a flow rate of 0.7 ml min−1. The conversion of HMF and the yield of each product were obtained directly using calibration curves of known concentrations.

Thermocatalytic experiments were also conducted on other substrates. The conditions used for these additional experiments varied and were selected through consideration of the literature. The conditions used for testing of the additional substrates (glycerol, ethanol, 1-propanol, 2-propanol, 1,3-propanediol, HMFCA and FFCA) are listed below.

#### 1-Propanol oxidation

1-Propanol (0.4 M); Na2CO3 (1.2 M); H2O (16 ml); 70 °C; pressure of O2 $$({p}_{{{\rm{O}}}_{2}})$$ = 3 bar; 2 h; catalyst amounts for Au@Pd/C, Au/C + Pd/C and Au–Pd/C, 143.1 mg; Au/C, 72.1 mg; Pd/C, 71 mg.

#### 2-Propanol oxidation

2-Propanol (0.4 M); Na2CO3 (1.2 M); H2O (16 ml); 70 °C; $${p}_{{{\rm{O}}}_{2}}$$ = 3 bar; 2 h; catalyst amounts for Au@Pd/C, Au/C + Pd/C and Au–Pd/C, 143.1 mg; Au/C, 72.1 mg; Pd/C, 71 mg.

#### 1,3-Propanediol oxidation

1,3-Propanediol (0.4 M); Na2CO3 (1.2 M); H2O (16 ml); 70 °C; $${p}_{{{\rm{O}}}_{2}}$$ = 3 bar; 2 h; catalyst amounts for Au@Pd/C, Au/C + Pd/C  and Au–Pd/C, 143.1 mg; Au/C, 72.1 mg; Pd/C, 71 mg.

#### HMFCA oxidation

HMFCA (0.1 M); NaHCO3 (0.4 M); H2O (16 ml); 80 °C; $${p}_{{{\rm{O}}}_{2}}$$ = 3 bar; 1 h; catalyst amounts for Au@Pd/C, Au/C + Pd/C and Au–Pd/C, 143.1 mg; Au/C, 72.1 mg; Pd/C, 71 mg.

#### FFCA oxidation

FFCA (0.1 M); NaHCO3 (0.4 M); H2O (16 ml); 80 °C; $${p}_{{{\rm{O}}}_{2}}$$ = 3 bar; 15 min; catalyst amounts for Au@Pd/C, Au/C + Pd/C and Au–Pd/C, 143.1 mg; Au/C, 72.1 mg; Pd/C, 71 mg.

#### HMF oxidation

HMF (0.1 M); NaHCO3 (0.4 M); H2O (16 ml); 80 °C; $${p}_{{{\rm{O}}}_{2}}$$ = 3 bar; 30 min; Au/C (72.1 mg, when employed); Pd/C (mass varied from 0 to 260 mg); carbon (mass varied from 140 to 260 mg). See Extended Data Fig. 2h.

#### HMF oxidation

HMF (0.1 M); NaHCO3 (0.4 M); H2O (16 ml); 80 °C; $${p}_{{{\rm{O}}}_{2}}$$= 0.6–3 bar; 30 min; catalyst amounts for Au/C + Pd/C (Au/C, 72.1 mg;   Pd/C, 71 mg) and Au/C (72.1 mg). See Extended Data Fig. 2i.

Given that in some of the reactions where initial rates were monitored high conversion rates were readily achieved, a different activity calculation for the substrate turned over (ActivitySTO) was used. As HMF oxidation to FDCA is a sequential reaction whereby the substrate undergoes sequential dehydrogenation reactions, it accounts for multiple turnovers. The equations used for the calculation are:

ActivitySTO = (moles of HMFCA × 1 + moles of FFCA × 2 + moles of FDCA × 3)/time (s)

ActivitySTO is used to present the catalyst performance in Extended Data Fig. 2h, i.

Productivity = (moles of HMFCA × 1 + moles of FFCA × 2 + moles of FDCA × 3)/(mass of catalyst (kg) × time (h))

Productivity is used to present the catalyst performance in Fig. 1e.

#### Testing of catalysts by electrochemical experiments

The experimental procedure for depositing the catalysts onto the working electrode was as follows. (1) A catalyst ink was made by adding 0.007 g of the monometallic catalysts or 0.014 g of the bimetallic catalysts into 1 ml distilled water mixed with 0.1 ml nafion solution. (2) The prepared ink solution was then sonicated for 150 s. (3) Then 0.02 ml of the ink was dropped onto the surface of the glassy carbon electrode (surface area 0.07065 cm2), which was then left to dry at room temperature for 16 h. (4) The final amount of catalyst on the surface of the glassy carbon electrode was 1.27 × 10−4 g for the monometallic systems and 2.55 × 10−4 g for the bimetallic systems.

Cyclic voltammetry experiments were performed under a N2 atmosphere, with N2 bubbled continuously (150 ml min−1) through the base electrolyte (0.11 M NaOH in 45 ml deionized water) solution for 20 min before recording measurements. A platinum wire (7.5 cm long and 0.5 mm diameter, 99.95% purity, from BASinc) was used as the counter electrode and a saturated calomel electrode (RE-2BP, ALS) was used as the reference electrode. The working electrode was first reduced at a fixed negative potential (−1.0 V) for 10 s to clean the surface. A background was then recorded with NaOH electrolyte for three cycles (−0.8 V to 0.6 V) under N2. Then HMF solution (5 ml, 0.2 M) was subsequently added into the base electrolyte solution and the cyclic voltammetry traces were recorded for an additional three cycles (−0.8 V to 0.6 V). All the experiments were conducted at a scan rate of 50 mV s−1 at room temperature with stirring and the third and final cycle was used for comparison.

Linear sweep voltammetry experiments were carried out by bubbling N2 through the electrolyte solution for 5 min before again conducting an electrode surface cleaning step; achieved through holding the electrode at −1.0 V for 10 s. Oxygen was subsequently bubbled through the electrolyte solution (150 ml min−1) for 30 min, before running the linear sweep voltammetry experiments. All the experiments were conducted from −1.1 V to 0 V at a scan rate of 10 mV s−1 at room temperature, while stirring. All reported current densities are normalized by the electrode surface area of 0.07 cm2.

A series of additional experiments were also conducted to assess electron transfer between the Au and Pd components. For this, Au/C and Pd/C catalysts were deposited separately onto the surface of two glassy carbon electrodes following the same procedure described previously. Two kinds of cell were used for these experiments; a single-chamber cell and a dual-chamber cell. For the single-chamber experiments, O2 was bubbled through the electrolyte solution for 20 min at a flow rate of 150 ml min−1, before the start of the experiment. The two electrodes were introduced through the cap of the reactor into the reaction solution while the solution O2 was still flowing and a blank run was performed with a base electrolyte (0.1 M NaOH in 50 ml deionized water), by connecting the electrodes through a shorted multi-meter. Oxygen was then bubbled into the solution at 50 ml min−1, while the current was recorded by a digital multi-meter (25XT, Wavetek Meterman, 0.5% accuracy). All the reactions with HMF were carried out under the same conditions as the blank, but with the addition 0.1 M HMF to the electrolyte solution. At the end of each experiment a 0.4-ml aliquot of the electrolyte solution was collected and analysed by HPLC. Under thesereaction conditions, only HMFCA was observed in the solution. The dual-chamber cell consists of two compartments with a volume of 30 ml each, separated by an anionic membrane (HMED-0510-2, HUAMOTECH) (Extended Data Fig. 6d). In this case, O2 was bubbled in one compartment and N2 in the other compartment for 20 min with a flow rate of 150 ml min−1. While the O2 was bubbling, the Pd/C electrode was rapidly introduced into the O2 compartment and the Au/C electrode into the N2 compartment. The flow rate of each gas was reduced to 50 m min−1 before connection of the electrodes and the start of the reaction. An identical concentration of HMF (0.02 M) in the 0.1 M NaOH electrolyte was introduced into the two compartments to avoid any diffusion from one compartment to the other before the start of the reaction. At the end of each experiment, a 0.2-ml aliquot of the electrolyte solution was collected from each compartment and analysed by HPLC. Under these reaction conditions, only HMFCA was produced in the solution. The protocol used for calculating the electron transfer efficiency in the short-circuit-current experiments is given below.

### Electron transfer efficiency (charge transfer) for single and dual cells

The moles of electrons, that is, n electrons, generated by the system was calculated using the equation below24, where Q (C) is the quantity of electric charge I × t (I, current in amperes; t, time) generated during the time (s) of the reaction and F is the Faraday constant (96,485 C mol−1):

$$n({\rm{electrons}})=Q/F$$
(1)

Q is obtained by integrating the curve (current versus time) recorded during the reaction.

The stoichiometric moles of electrons generated during the reaction is twice the number of moles of HMFCA detected by HPLC, using the following equations:

$${{\rm{C}}}_{6}{{\rm{H}}}_{6}{{\rm{O}}}_{3}({\rm{HMF}})+2{{\rm{OH}}}^{-}\to {{\rm{C}}}_{6}{{\rm{H}}}_{6}{{\rm{O}}}_{4}({\rm{HMFCA}})+{{\rm{H}}}_{2}{\rm{O}}+2{{\rm{e}}}^{-}$$
(2)
$$n({\rm{electrons}})=2n({\rm{HMFCA}})$$
(3)

The efficiency of the electron transfer is the ratio between the measured moles of electrons flowing from the Au/C electrode to the Pd/C electrode (equation (1)) divided by the theoretical moles of electrons, which should be generated from reaction stoichiometry based on the HPLC analysis of the products in the Au/C electrode compartment after reaction (2).

$${\rm{Electron}}\,{\rm{transfer}}\,{\rm{efficiency}}( \% )=(n({\rm{electrons}}){\rm{fromequation}}(1)/n({\rm{electrons}}){\rm{fromequation}}(3))\times 100$$

### Catalyst characterization

#### Scanning transmission electron microscopy

Samples for examination by STEM were prepared by dry dispersing the catalyst powder onto a holey carbon film supported by a 300-mesh copper TEM grid. Bright-field and high-angle annular dark field (HAADF) STEM images were taken using an aberration-corrected JEOL JEM ARM-200CF microscope at Lehigh University, operating at 200 kV, equipped with a single JEOL Centurio silicon drift detector for XEDS. The Au@Pd/C catalyst was analysed using a JEM ARM-200CF microscope equipped with dual Centurio silicon drift XEDS detectors at the electron physical science imagining centre at Diamond Light Source (UK). The Au=Pd/C catalyst was analysed using a JEM ARM-200CF microscope equipped with Oxford Instruments X-MAXN 100LTE XEDS detectors at the National University of Singapore. Particle size distribution histograms were generated by analysis of representative HAADF electron micrographs using ImageJ (version 1.53f51).

The geometric Mackay icosahedral model25 was used to estimate the number of Au atoms associated with the entire particle volume and exposed particle surface, as a function of Au particle size. The particle diameter d was estimated using the equation d= (2n + 1) × 0.288 (nm), where n is the number of shells in the Mackay model; the total atoms per particle (Ntotal) was calculated using the equation Ntotal = 10/3 × n3 + 5n2 + 11/3 × n + 1; the total number of exposed surface atoms (Nsurface) was calculated using the equation Ntotal = 10 × n2 + 2.The Mackay icosahedron chosen to best represent each binning interval was that which had the closest value to the median value of the binning interval.

#### Microwave plasma atomic emission spectroscopy

Reaction solutions were first filtered to remove any heterogeneous catalyst from the sample. Further filtration was carried out using polytetrafluoroethylene (PTFE) syringe filters (0.456 µm). The samples were then analysed using an Agilent MP-AES 4100 MP-AES spectrometer. Samples were investigated for the presence of precious metals (Au and Pd) using multiple wavelength calibrations for each individual element.

#### Inductively coupled plasma mass spectrometry

Reaction solutions were first filtered to remove any heterogeneous catalyst from the sample. Further filtration was carried out using PTFE syringe filters (0.456 µm). The samples were then analysed using an Agilent 7900 inductively coupled plasma mass spectrometer with an I-AS autosampler. A five-point calibration was conducted using certified reference material from PerkinElmer and an internal standard also certified from Agilent. All samples were diluted 10 times; 500 μl of sample was added into a 5-ml class A volumetric flask with deionized H2O (1% nitric acid (HNO3) and 0.5% hydrochloric acid (HCl) matrix); the calibrants were matrix matched. Nickel sampling and skimmer cones were used for the analysis and the helium mode on the ORS4 Octopol was used to help with interference reduction.

### Temperature programmed reduction

TPR was performed on an Anton Paar, ChemBET. Samples (about 20 mg) were pre-treated in helium (30 ml min−1) at 110 °C for 2 h. The TPR analysis was performed in 10% hydrogen/argon (30 ml min−1) from 30 °C to 350 °C at 10 °C min−1.

### X-ray photoelectron spectroscopy

XPS was performed on a Thermo Fisher Scientific K-alpha+ photoelectron spectrometer. Samples were mounted in small recesses within the Thermo Scientific copper powder plate. Samples were analysed using a micro-focused monochromatic aluminium X-ray source operating at 72 W (6 mA × 12 kV) using the 400-μm spot option, which is an elliptical area of approximately 400 μm × 600 μm. Data were recorded at pass energies of 150 eV for survey scans and 40 eV for high-resolution scans with 1-eV and 0.1-eV step sizes, respectively. Charge neutralization of the sample was achieved using a combination of both low-energy electrons and argon ions, which results in a C (1s) energy for the carbon support of 284.5 eV, which is typical of graphitic-like carbons. Data analysis was performed in CasaXPS (v.2.3.24) using Scofield cross-sections with an energy dependence of −0.6 after removal of a Shirley background.

## Data availability

All data that led us to understand the results presented here are available with the Article or from the corresponding author upon reasonable request. Source data are provided with this paper.

## References

1. Mallat, T. & Baiker, A. Oxidation of alcohols with molecular oxygen on solid catalysts. Chem. Rev. 104, 3037–3058 (2004).

2. Ciriminna, R., Pandarus, V., Béland, F., Xu, Y. J. & Pagliaro, M. Heterogeneously catalyzed alcohol oxidation for the fine chemical industry. Org. Process Res. Dev. 19, 1554–1558 (2015).

3. Sheldon, R. A., Arends, I. W. C. E., Brink, G. J., Ten, & Dijksman, A. Green, catalytic oxidations of alcohols. Acc. Chem. Res. 35, 774–781 (2002).

4. Abad, A., Concepción, P., Corma, A. & García, H. A collaborative effect between gold and a support induces the selective oxidation of alcohols. Angew. Chem. Int. Ed. 44, 4066–4069 (2005).

5. Kulkarni, A., Siahrostami, S., Patel, A. & Nørskov, J. K. Understanding catalytic activity trends in the oxygen reduction reaction. Chem. Rev. 118, 2302–2312 (2018).

6. Nørskov, J. K. et al. Origin of the overpotential for oxygen reduction at a fuel-cell cathode. J. Phys. Chem. B 108, 17886–17892 (2004).

7. Zhu, X. et al. Optimising surface d charge of AuPd nanoalloy catalysts for enhanced catalytic activity. Nat. Commun. 10, 1428 (2019).

8. Enache, D. I. et al. Solvent-free oxidation of primary alcohols to aldehydes using Au–Pd/TiO2 catalyst. Science 311, 362–365 (2006).

9. Trindell, J. A., Duan, Z., Henkelman, G. & Crooks, R. M. AuxPd(300-x) alloy nanoparticles for the oxygen reduction reaction in alkaline media. ChemElectroChem 7, 3824–3831 (2020).

10. Hutchings, G. J. & Kiely, C. J. Strategies for the synthesis of supported gold palladium nanoparticles with controlled morphology and composition. Acc. Chem. Res. 46, 1759–1772 (2013).

11. Wang, A., Liu, X. Y., Mou, C. Y. & Zhang, T. Understanding the synergistic effects of gold bimetallic catalysts. J. Catal. 308, 258–271 (2013).

12. Wang, D., Villa, A., Porta, F., Prati, L. & Su, D. Bimetallic gold/palladium catalysts: correlation between nanostructure and synergistic effects. J. Phys. Chem. C 112, 8617–8622 (2008).

13. Gao, F. & Goodman, D. W. Pd–Au bimetallic catalysts: understanding alloy effects from planar models and (supported) nanoparticles. Chem. Soc. Rev. 41, 8009–8020 (2012).

14. Han, S. & Mullins, C. B. Catalytic reactions on Pd–Au bimetallic model catalysts. Acc. Chem. Res. 54, 379–387 (2020).

15. Davis, S. E., Ide, M. S. & Davis, R. J. Selective oxidation of alcohols and aldehydes over supported metal nanoparticles. Green Chem. 15, 17–45 (2013).

16. Parmeggiani, C., Matassini, C. & Cardona, F. A step forward towards sustainable aerobic alcohol oxidation: new and revised catalysts based on transition metals on solid supports. Green Chem. 19, 2030–2050 (2017).

17. Pelaez, M. et al. A review on the visible light active titanium dioxide photocatalysts for environmental applications. Appl. Catal. B 125, 331–349 (2012).

18. Golberg, D. et al. Boron nitride nanotubes and nanosheets. ACS Nano 4, 2979–2993 (2010).

19. Suo, Y., Zhuang, L. & Lu, J. First-principles considerations in the design of Pd-alloy catalysts for oxygen reduction. Angew. Chem. Int. Ed. 46, 2862–2864 (2007).

20. Zope, B. N., Hibbitts, D. D., Neurock, M. & Davis, R. J. Reactivity of the gold/water interface during selective oxidation catalysis. Science 330, 74–78 (2010).

21. Han, Y. F. et al. Au promotional effects on the synthesis of H2O2 directly from H2 and O2 on supported Pd–Au alloy catalysts. J. Phys. Chem. C 111, 8410–8413 (2007).

22. Chadderdon, D. J. et al. Electrocatalytic oxidation of 5-hydroxymethylfurfural to 2,5-furandicarboxylic acid on supported Au and Pd bimetallic nanoparticles. Green Chem. 16, 3778–3786 (2014).

23. Zhu, Y. et al. Boosting oxygen reduction reaction activity of palladium by stabilizing its unusual oxidation states in perovskite. Chem. Mater. 27, 3048–3054 (2015).

24. Ringeisen, B. R. et al. High power density from a miniature microbial fuel cell using Shewanella oneidensis DSP10. Environ. Sci. Technol. 40, 2629–2634 (2006).

25. Mackay, A. L. A dense non-crystallographic packing of equal spheres. Acta Cryst. 15, 916–918 (1962).

26. Siyo, B. et al. Influence of support on the aerobic oxidation of HMF into FDCA over preformed Pd nanoparticle based materials. Appl. Catal. A 478, 107–116 (2014).

27. Villa, A., Schiavoni, M., Campisi, S., Veith, G. M. & Prati, L. Pd-modified Au on carbon as an effective and durable catalyst for the direct oxidation of HMF to 2,5-furandicarboxylic acid. ChemSusChem 6, 609–612 (2013).

28. Han, X. et al. Base-free aerobic oxidation of 5-hydroxymethylfurfural to 2,5-furandicarboxylic acid over a Pt/C–O–Mg catalyst. Green Chem. 18, 1597–1604 (2016).

29. Zhou, C. et al. Functionalized carbon nanotubes for biomass conversion: the base‐free aerobic oxidation of 5‐hydroxymethylfurfural to 2,5‐furandicarboxylic acid over platinum supported on a carbon nanotube catalyst. ChemCatChem 7, 2853–2863 (2015).

30. Artz, J. & Palkovits, R. Base‐free aqueous‐phase oxidation of 5‐hydroxymethylfurfural over ruthenium catalysts supported on covalent triazine frameworks. ChemSusChem 8, 3832–3838 (2015).

31. Mishra, D. K. et al. MnCo2O4 spinel supported ruthenium catalyst for air-oxidation of HMF to FDCA under aqueous phase and base-free conditions. Green Chem. 19, 1619–1623 (2017).

32. Schade, O. et al. The influence of the gold particle size on the catalytic oxidation of 5-(hydroxymethyl)furfural. Catalysts 10, 342 (2020).

33. Megías‐Sayago, C. et al. Effect of gold particles size over Au/C catalyst selectivity in HMF oxidation reaction. ChemCatChem 12, 1177–1183 (2020).

34. Ferraz, C. P. et al. 5-Hydroxymethylfurfural and furfural base-free oxidation over AuPd embedded bimetallic nanoparticles. Catalysts 10, 75 (2020).

35. Yang, J. et al. Effect of the oxygen coordination environment of CaMn oxides on the catalytic performance of Pd supported catalysts for aerobic oxidation of 5-hydroxymethyl-2-furfural. Catal. Sci. Technol. 9, 6659–6668 (2019).

36. Liu, Y. et al. Active oxygen species promoted catalytic oxidation of 5-hydroxymethyl-2-furfural on facet-specific Pt nanocrystals. ACS Catal. 9, 8306–8315 (2019).

37. Megias-Sayago, C. et al. Understanding the role of the acid sites in 5-hydroxymethylfurfural oxidation to 2,5-furandicarboxylic acid reaction over gold catalysts: surface investigation on CexZr1−xO2 compounds. ACS Catal. 8, 11154–11164 (2018).

38. Yu, K. et al. The role of Bi-doping in promoting electron transfer and catalytic performance of Pt/3DOM-Ce1−xBixO2−δ. J. Catal. 365, 292–302 (2018).

39. Schade, O. R., Kalz, K. F., Neukum, D., Kleist, W. & Grunwaldt, J. D. Supported gold- and silver-based catalysts for the selective aerobic oxidation of 5-(hydroxymethyl)furfural to 2,5-furandicarboxylic acid and 5-hydroxymethyl-2-furancarboxylic acid. Green Chem. 20, 3530–3541 (2018).

40. Masoud, N., Donoeva, B. & de Jongh, P. E. Stability of gold nanocatalysts supported on mesoporous silica for the oxidation of 5-hydroxymethyl furfural to furan-2,5-dicarboxylic acid. Appl. Catal. A 561, 150–157 (2018).

41. Wan, X. et al. Base-free aerobic oxidation of 5-hydroxymethyl-furfural to 2,5-furandicarboxylic acid in water catalyzed by functionalized carbon nanotube-supported Au–Pd alloy nanoparticles. ACS Catal. 4, 2175–2185 (2014).

42. Bonincontro, D. et al. AuPd-nNiO as an effective catalyst for the base-free oxidation of HMF under mild reaction conditions. Green Chem. 21, 4090–4099 (2019).

43. Gui, Z. et al. Efficient aerobic oxidation of 5‐hydroxymethylfurfural in aqueous media with Au–Pd supported on zinc hydroxycarbonate. ChemCatChem 8, 3636–3643 (2016).

44. Kerdi, F. et al. Evaluation of surface properties and pore structure of carbon on the activity of supported Ru catalysts in the aqueous-phase aerobic oxidation of HMF to FDCA. Appl. Catal. A 506, 206–219 (2015).

45. Ait Rass, H., Essayem, N. & Besson, M. Selective aerobic oxidation of 5‐HMF into 2,5‐furandicarboxylic acid with Pt catalysts supported on TiO2‐ and ZrO2‐based supports. ChemSusChem 8, 1206–1217 (2015).

46. Chen, H. et al. Atomic layer deposition of Pt nanoparticles on low surface area zirconium oxide for the efficient base-free oxidation of 5-hydroxymethylfurfural to 2,5-furandicarboxylic acid. Appl. Catal. A 555, 98–107 (2018).

## Acknowledgements

We thank L. Kang and R. Wang from University College London and Cardiff University for access and assistance with the electron microscopy; and the Diamond Light Source for access to beamline E01 (proposal number EM18909). C.J.K. acknowledges funding from the National Science Foundation Major Research Instrumentation programme (GR# MRI/DMR-1040229). S.M.A. thanks the Saudi Arabian government for his PhD scholarship. X.H. and Q.H. thank Cardiff University School of Chemistry for financial support. Q.H. also acknowledges the support by National Research Foundation (NRF) Singapore, under its NRF Fellowship (NRF-NRFF11-2019-0002). K.W. and L.Z. thank the Chinese Scholarship Council (CSC) for financial support. XPS data collection was performed at the EPSRC National Facility for XPS (‘HarwellXPS’), operated by Cardiff University and UCL, under contract number PR16195. We thank Cardiff University and the Max Planck Centre for Fundamental Heterogeneous Catalysis (FUNCAT) for financial support.

## Author information

Authors

### Contributions

X.H., O.A., M.D., R.J.L., S.P., P.J.M. and G.J.H. contributed to the design of the study. X.H., O.A., L.Z., I.T.D., R.J.L., K.W., J.F. and F.W. conducted experiments and data analysis. O.A. and X.H. conceived the mechanism. X.H., O.A., M.D., R.J.L., S.P., P.J.M., G.S., D.B., S.M., C.J.K. and G.J.H. provided technical support, conceptual advice and result interpretation. X.H., O.A., S.P., G.S., D.J.M., S.M.A., T.E.D., Q.H. and C.J.K. conducted catalyst characterization and corresponding data processing. M.D., O.A., D.B. and G.J.H. wrote the manuscript. X.H. wrote the Extended Data figures and tables. X.H., O.A., M.D., R.J.L., S.P., G.S., D.B., S.M., C.J.K. and G.J.H commented on and amended both documents. All authors discussed and contributed to the work.

### Corresponding author

Correspondence to Graham J. Hutchings.

## Ethics declarations

### Competing interests

The authors declare no competing interests.

## Peer review

### Peer review information

Nature thanks Bin Wang and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

## Extended data figures and tables

### Extended Data Fig. 1 Diagrammatic representation of the sol-immobilisation method used for catalyst preparation.

a, Monometallic Au/C and Pd/C. b, Au=Pd/C. c, Au@Pd/C catalysts. d, alloyed Au–Pd/C. e, Schematic representation of our reactor set-up for the thermocatalytic experiments.

### Extended Data Fig. 2 Time-on-line data of aqueous HMF oxidation over series of Au/Pd catalysts and their TOF.

a, Au/C. b, Pd/C. c, Au–Pd/C alloy. d, Physical mixture of Au/C + Pd/C. e, Au@Pd/C. f, Au/C followed by the addition of Pd/C after 30 min. g, Au/C followed by the addition of C after 30 min. Reaction conditions: HMF (0.1 M); NaHCO3 (0.4 M); H2O (16 ml); Au/C: 72.1 mg; Pd/C: 71 mg; Au–Pd/C alloy: 143.1 mg; Au@Pd/C: 143.1 mg; C: 71 mg; 80 °C; pO2 = 3 bar. Key: FDCA yield (■), FFCA yield (♦), HMFCA yield (▲), HMF conversion (), mass balance (*). Associated error bars correspond to mean ± s.d. (n = 5). h, The influence on ODH activity when various quantities of Pd/C (▲) and C () are added to Au/C (72.1 mg); ODH activity exhibited by various quantities of Pd/C, in the absence of Au/C, is also displayed (◂). i, The influence of oxygen pressure (0.6–3.0 bar) on ODH activity over a physical mixture of Au/C + Pd/C (▲) and Au/C (♦) is displayed. The reaction conditions used for h and i are given in Methods. j, Summary of each catalyst in terms of HMF conversion, initial rate and TOF at a 5-min reaction time. The total active sites available in each catalyst was estimated using the Mackay model, based on the presented particle size distributions25. Further information relating to how the TOFs were calculated, can be found in Methods.

### Extended Data Fig. 3 Catalytic performance over Au/C, Pd/C, Au–Pd/C alloy, Au/C + Pd/C physical mixture and Au@Pd/C catalysts in series of alcohol oxidation reactions.

a, Glycerol oxidation. b, Ethanol oxidation. c, 5-Formyl-2-furancarboxylic acid (FFCA) oxidation. d, 5-Hydroxymethylfuroicacid (HMFCA) oxidation. Reaction conditions were all listed in Methods. e, Conversion values for the HMF oxidation reaction after 5 and 15 min time-on-line for the various catalysts studied in this work. Reaction conditions: HMF (0.1 M); NaHCO3 (0.4 M); H2O (16 ml); Au/(C/TiO2/BN): 72.1 mg, Pd/(C/TiO2/BN): 71 mg, Au@Pd/C, Au–Pd/(C/TiO2/BN) and Au/(C/TiO2/BN) + Pd/(C/TiO2/BN): 143.1 mg; 80 °C; pO2 = 3 bar; reaction time: 5 and 15 min. * presents the test on bare supports in HMF oxidation, HMF (0.1 M); NaHCO3 (0.4 M); H2O (16 ml); C/TiO2/BN: 60 mg; 80 °C; pO2 = 3 bar; reaction time: 30 min.

### Extended Data Fig. 4 Electron microscopy analysis of Au/C + Pd/C (physical mixture) catalyst after one cycle of use in the oxidation of HMF.

a, b, Representative complementary BF- and HAADF-STEM micrographs showing metal nanoparticle size and spatial distribution. c, Atomic resolution HAADF-STEM micrograph of a C grain supporting Au particles. d, A C grain supporting Pd particles confirming that the Au and Pd remain separated under our reaction conditions. e, f, Representative XEDS spectra of individual Au particles and Pd particles in the catalyst, respectively. No evidence of Au or Pd migration or intermixing after the catalytic reaction was observed.

### Extended Data Fig. 5 XPS data.

a, b, Au 4f and c, d, Pd 3d /Au 4d regions for Au/C and Pd/C monometallic catalysts before and after a typical HMF oxidation reaction as a physical mixture. Among which, a, fresh Au/C; b, used Au/C; c, fresh Pd/C; and d, used Pd/C. TPR data for the physically mixed Au/C + Pd/C catalyst and the Au@Pd/C catalyst e, before and f, after HMF oxidation.

### Extended Data Fig. 6 Electrochemical and thermal catalytic oxidation of aqueous HMF over Au/Pd catalysts.

a, Correlation between the thermo- and electro-catalytic HMF oxidation over the series of catalysts. For thermocatalytic experiments, the initial rates were from a 5-min reaction. The current densities were from the maxima observed in the corresponding CV experiments (Fig. 3a). Associated error bars correspond to mean ± s.d. (n = 3). b, Aqueous HMF oxidation over the mono- and bi-metallic Au–Pd catalysts. Reaction conditions: HMF (0.1 M); NaOH (0.4 M); H2O (16 ml); 25 °C; pO2 = 3 bar; 30 min; catalyst amounts for Au@Pd/C and Au–Pd/C:143.1 mg, Au/C: 72.1 mg, Pd/C: 71 mg, carbon balance: ca 92%. c, Catalytic performance in short circuit with current density (normalized by an electrode surface area of 0.07 cm2) generated as a function of time in the single cell. Reaction conditions: 0.1 M NaOH and 0.02 M HMF in 50 ml H2O; Au (working electrode) and Pd or C (counter electrode); 25 °C; O2 flow: 50 ml min−1. d, H-type dual cell consists of Au as the anode in an N2 flow, Pd as cathode in an O2 flow. The two cells connect via an anion exchange membrane. Reaction conditions: each cell contains 0.1 M NaOH and 0.02 M HMF in 35 ml H2O; 25 °C; gas flow O2/N2: 50 ml min-1. e, Reaction conditions: i: 0.1 M NaOH and 0.02 M HMF in 50 ml H2O, 25 °C, N2 flow: 50 ml min−1; ii: same as i, except for the O2 flow: 50 ml min−1; iii: each cell contains 0.1 M NaOH and 0.02 M HMF in 35 ml H2O, 25 °C, O2/N2 flow: 50 ml min−1; iv: same as iii, except for the disconnection of Au and Pd electrodes; v- same as iii, except the mass of Pd/C is doubled.

### Extended Data Fig. 7 Representative STEM-HAADF images and X-ED spectra of nanoparticles in the Au=Pd/C catalysts and its corresponding activities.

a, Lower magnification STEM-HAADF image of the Au = Pd/C catalyst. b, c, X-ED spectra obtained from individual nanoparticles, showing a Au-only and a Pd-only nanoparticle. d, STEM-HAADF image and the corresponding X-ED spectrum (inlet) of a Janus-like particle occasionally found in this Au=Pd/C catalyst. e, Activity comparison to the physical mixture f. Reaction conditions: HMF (0.1 M); NaHCO3 (0.4 M); H2O (16 ml); Au/C: 72.1 mg; Pd/C: 71 mg; Au=Pd/C: 143.1 mg; 80 °C; pO2 = 3 bar; reaction time: 30 min. Associated error bars correspond to mean ± s.d. (n = 3).

### Extended Data Fig. 8 Electron microscopy analysis of Au@Pd/C catalyst after one use in the oxidation of HMF.

a, b, Representative complementary pair of BF- and HAADF-STEM micrographs showing metal nanoparticle size and spatial distribution. ce, Atomic resolution HAADF-STEM micrographs of particles. The yellow arrows in e highlight certain atomic columns that appear lower in contrast, indicating some alloying of Pd with the Au matrix. f, A representative XEDS spectrum obtained from a typical nanoparticle, showing the presence of both Au and Pd.

### Extended Data Fig. 9 Reusability data for the prepared Au/Pd catalysts in HMF oxidation reaction.

a, c, The Au@Pd/C catalyst. b, d, the physical mixture Au/C + Pd/C catalyst. Reaction conditions: HMF (0.1 M); NaHCO3 (0.4 M); H2O (16 ml); Au/C: 72.1 mg; Pd/C: 71 mg; Au@Pd/C: 143.1 mg; 80 °C; pO2 = 3 bar; reaction time: 60 min. Key: FDCA yield (■), FFCA yield (♦), HMFCA yield (▲), HMF conversion (), mass balance (*).

## Rights and permissions

Reprints and Permissions

Huang, X., Akdim, O., Douthwaite, M. et al. Au–Pd separation enhances bimetallic catalysis of alcohol oxidation. Nature 603, 271–275 (2022). https://doi.org/10.1038/s41586-022-04397-7

• Accepted:

• Published:

• Issue Date:

• DOI: https://doi.org/10.1038/s41586-022-04397-7