Producing chemicals and energy directly from biomass is an effective way to lessen the dependence on petroleum-based resources, which are on the verge of depletion. 5-Hydroxymethylfurfural (HMF), a representative furanic compound generated by acid-catalyzed dehydration of hexose sugars, is an important platform chemical linking raw biomass utilization and fine chemical production [1]. With two functionalities (−OH, C=O) attaching to a furan ring, HMF can be converted to several value-added chemicals via various redox processes [2]. One of its reduced derivatives, 2, 5-dihydroxymethylfuran (DHMF), is an important precursor for the production of polyesters and polyurethane foams [3]. Its main oxidation product, 2, 5-furandicarboxylic acid (FDCA), is a primary feedstock to produce polyethylene 2, 5-furandicarboxylate, a renewable polymer alternative to petroleum-derived polyethylene terephthalate plastics [4]. Heterogeneous catalytic hydrogenation and oxidation of the HMF are the main methods for the production of DHMF and FDCA, respectively. However, these methods usually involve high-pressure H₂ or O₂, high temperature, and noble metal-based catalysts (e.g., Au, Pt, Ru, and Pd) [3,4], which hinder their wide applications. Therefore, developing efficient methods for HMF conversion under mild conditions with cost-effective catalysts is highly desired.

The electrochemical method offers a promising way to replace the conventional heterogeneous catalytic system because of its advantages such as mild operating conditions, dispensing with noble-metal-based catalysts and easily controlled reaction process [5]. However, a typical electrochemical process usually involves two half-reactions occurring in pairs, yet many commercialized electrochemical production technologies utilize only one half-reaction to obtain desired products. For example, the cathodic HMF electrochemical reduction is usually paired with anodic water oxidation, which has sluggish kinetics and its main product oxygen gas is not valuable [6]. On the other hand, the anodic HMF oxidation is often paired with cathodic water reduction, in which less-value, flammable and combustible H₂ is a primary product [7].

Paired electrolysis, in which the anodic oxidation and cathodic reduction are paired in one single electrochemical cell, offers a robust way to obtain desired products in both electrodes and achieve complete electron economy [8]. Theoretically, the maximum electron efficiency of paired electrolysis could reach 200%. Moreover, the operation costs and equipment investments can be significantly reduced in the practical applications of the paired electrolysis technology by lowering the required reactor number and processing steps. For example, when the cathodic HMF reduction is paired with anodic HMF oxidation catalyzed by homogeneous 4-acetamido-2,2,6,6-tetramethylpiperidine-1-oxyl (4-ACT-TEMPO), DHMF and FDCA can be produced in the cathode and anode, respectively, with an overall electron efficiency of 187% [9]. However, the use of homogeneous 4-ACT-TEMPO will greatly increase the costs of downstream separation. Meanwhile, the cathodic HMF reduction is usually conducted under acidic conditions, in which acid-resistant catalysts are required [10], while the anodic HMF oxidation is usually performed under alkaline conditions with alkaline-resistant catalysts. Indeed, the Ag-based materials are reported as the most effective cathodic catalysts for HMF reduction [6,9], while the non-noble metal-based catalysts, such as Fe, Ni, and Co hydroxides have been exploited for anodic HMF oxidation [11]. Apart from the requirements of different electrocatalysts in anode and cathode, the chemical incompatibilities, mismatched optimal current densities for the two half-reactions, and crossover issues are also big challenges for the paired electrolysis of HMF.

Transition metal oxides with highly oxidized redox couples like Co^3+/4+, Ni^3+/4+, Fe^3+/4+, and Mn^3+/4+ have been known as active centers for many electro-oxidation reactions, such as oxygen evolution reaction (OER) and urea oxidation [12,13], while the metal oxides with reduced redox couples such as Co^2+/3+ and Cu^+/2+ are considered as the main active centers for some electro-reduction reactions, such as oxygen reduction reaction (ORR) and CO₂ reduction [14,15]. Among the numerous previously investigated transition metal oxides, Co₃O₄ should be a promising material for simultaneously catalyzing electro-oxidation and reduction reactions [16,17], as it can form a highly oxidized Co^3+/4+ redox couple under oxidation conditions, and a reduced Co^2+/3+ redox couple under reduction conditions.

Intending to develop an efficient paired electrolysis system to convert HMF into DHMF and FDCA, we synthesize a self-supported ultrathin Co₃O₄ nanoarray material and use it as a bifunctional electrocatalyst for both HMF oxidation and reduction. The self-supported electrode is an electrocatalytic material in situ grown on conductive support without any additional binder, conductive agent, or current collector [18]. The Co₃O₄ catalyst is closely contacting with the current collector, which is favorable to electron conduction. The high surface area, good electron conductivity and variable valence states endow it with a good performance in the electrocatalytic conversion of HMF.

The self-supported Co₃O₄ NSA electrocatalyst was synthesized via a two-step template-free approach (Fig. 1). In the hydrothermal process, urea was used to provide a basic condition to form Co(OH)₂ precursor and also a mineralizer to promote the crystallization of hexagonal nanosheet arrays [19]. The NH₄F could play an important role in regulating the growth of the Co(OH)₂ nanosheets on the Ni foam substrate [19]. Thereafter, the formed Co(OH)₂ was thermally converted into spinel Co₃O₄ nanoarray supported on the Ni foam, as described by a simple oxidation reaction as follows: 3Co(OH)₂ + 1/2O₂ → Co₃O₄ + 3H₂O. This method is facile and cost-effective to synthesize large-scale self-supported Co₃O₄ NSA electrodes with high robustness and conductivity, which is promising for practical electrode manufacture.

Scanning electron microscopy (SEM) images indicate that the Co₃O₄ grew uniformly on the Ni foam substrate in the form of vertical hexagonal nanosheets (Figs. 2a-c). Three elements of Co, O, and Ni were distributed uniformly with atomic concentrations of 42.7%, 37.4%, and 19.9% respectively, as characterized by the energy distributive spectroscopy (EDS) mapping (Fig. 2d and Fig. S1 in Supporting information). Transmission electron microscopy (TEM) images reveal that the thin sheets were composed of smaller particles (Fig. 2e). The selected area electron diffraction (SAED) (inset of Fig. 2f) displays a polycrystalline structure with (311) planes observed, which was also confirmed by the lattice spacings in high resolution (HRTEM) image (Fig. 2f).

Fig. 3a displays the X-ray diffraction (XRD) pattern of the ultrathin Co₃O₄ nanoarrays supported on the Ni foam. As the results show, except for the two typical diffraction peaks from the Ni foam substrate, other several well-defined diffraction peaks are observed at 2θ values of 31, 37, 59, and 65 degrees. All these diffraction peaks could be successfully indexed to (220), (311), (422) and (511) plane reflections of the spinel Co₃O₄ crystalline structure ((JCPDF No. 42-1467; space group: Fd3m) with the standard peaks indicated by the green lines shown in Fig. 3a. These results indicate that the Co₃O₄ NSA has uniformly grown on the surface of Ni foam substrate.

The X-ray photoelectron spectroscopy (XPS) survey spectrum (Fig. S2 in Supporting information) confirms that Co, O, and Ni were the main components of the electrode catalyst. The peaks at 796.5 and 780.9 eV were assigned to Co 2p_1/2 and 2p_3/2 signal of Co²⁺, and the peaks at 794.6 eV and 779.6 eV belonged to Co 2p_1/2 and 2p_3/2 signal of Co³⁺ (Fig. 3b) [17]. The deconvoluted peaks of O 1s at 529.6 eV, 531.2 eV, and 532.5 eV indicate the existence of lattice O, adsorbed O, and O in H₂O (Fig. 3c) [20]. The Raman spectrum (Fig. S3 in Supporting information) further supports the existence of the Co₃O₄ phase, with the peaks at 654, 593, 504, 458, and 184 belonging to the A_1g, F_2g, F_2g, E_g, F_2g vibration modes of Co₃O₄ [21]. The EIS measurement was conducted to evaluate the charge transfer resistance at the interface of electrode and electrolyte. As the results showed (Fig. S4 in Supporting information), the semicircle reflecting the charge transfer resistance was small, no matter with or without HMF, indicating a fast charge transfer kinetics. This was probably related to the good electrical contact between the catalyst and the support. The surface area was characterized by N₂ adsorption-desorption tests and double-layer capacitance methods. The Brunauer-Emmett-Teller (BET) surface area was determined to be 6.84 m²/g with a pore surface area of 8.06 m²/g and a pore volume of 0.017 cm³/g (Fig. S5 in Supporting information). The double-layer capacitance determined from cyclic voltammetry (CV) was 83 mF/cm² for the Co₃O₄ NSA/Ni foam (Fig. 3d and Fig. S6 in Supporting information), which was higher than that of the raw Ni foam (23 mF/cm²), indicating a larger electrochemical active surface area (ECSA) could be obtained after growing Co₃O₄ on the Ni foam substrate.

Paired electrolysis is advantageous for maximizing electron efficiency and producing useful products in both chambers. We integrated the paired electrolysis of HMF in an electrolyzer using the Co₃O₄ NSA/Ni foam as electrodes on both sides. KOH and PBS were selected as electrolytes for anode and cathode reactions, respectively (Scheme 1 and Fig. S7 in Supporting information). The paired electrolysis was conducted by setting the cathodic potential at a constant value. Under −0.489 V vs. RHE, the current was as high as over 12 mA/cm² at the initial 20 min and declined from 20 min to 60 min. Later, the current kept at around 2 mA/cm² from 60 min to 120 min (Fig. 4a). Accordingly, the conversion of HMF reached about 80% in the initial 60 min and slowed down after 60 min. Overall, 95.7% yield of FDCA and 78.8% yield of DHMF were obtained in the anode and cathode, respectively (Figs. 4b and c). To examine the impact of voltage, the paired electrolysis was conducted under different potentials. The HMF conversion ratio increased with the increasing potential applied in both electrodes, with a maximum value of 99.4% and 94.5% for the anode and cathode, respectively (Fig. 4d). The anodic selectivity increased with the increasing voltage and peaked at 97.3% at –0.589 V vs. RHE, while the cathodic selectivity decreased with the increasing voltage and peaked at 95.3% at −0.289 V vs. RHE (Fig. 4e). Since the electrons could be used in both anode and cathode, the total electron utilization efficiency for the paired electrolysis could be more than 100%. As shown in Fig. 4f, the total electron efficiency for the paired electrolysis from −0.289 V to −0.489 V was higher than 120%, while when the potential was shifted to −0.589 V, the Faradaic efficiency dropped down due to the enhancement of HER competition. Therefore the best performance was obtained at −0.489 V.

Five successive cycles of chronoamperometric tests were carried out to evaluate the durability of the Co₃O₄ NSA/Ni foam electrode in the paired HMF electrolysis. In the cathode (Fig. 5a), the HMF conversion ratio slightly decreased from 93% to about 89% after 5 cycles, while the corresponding DHMF selectivity kept almost over 80%. Meanwhile, in the anode (Fig. 5b), the HMF conversion and its selectivity to FDCA were maintained at over 95% in the five cycles.

The cycled Co₃O₄ NSA electrodes were then characterized with SEM and HRTEM. As shown in Fig. S8 (Supporting information), the anodic and cathodic Co₃O₄ NSA still kept their nanosheet morphology after five cycles, but some small particles were observed in the cathodic Co₃O₄ NSA. The changes in crystal phase and electronic structure before and after use were further analyzed with XRD and XPS. For cathodic catalyst, the XRD result (Fig. S9c in Supporting information) exhibits a weakened crystallinity of Co₃O₄, and XPS analysis (Figs. S9a and b in Supporting information) shows a reduction of surface cobalt state, together with a decrease in lattice oxygen and an increase in adsorbed oxygen. These results lead to the conclusion that Co₃O₄ experienced a reduction during reaction in the cathode. As for the anodic catalyst, the morphology was kept unchanged after five cycles (Fig. S8), and its XRD (Fig. S9c) and XPS results (Figs. S9a and b) also suggest no prominent change compared with the original Co₃O₄, indicating that Co₃O₄ was stable under the oxidative and alkaline conditions, which is beneficial for the long-term run. The above-paired electrolysis system was compared with previously reported work involving paired electrolysis of HMF (Table S1 in Supporting information). Our work is advantageous in the development of a bifunctional non-noble catalyst, a mild reaction condition and a satisfactory product yield.

The mechanism for the electro-oxidation and hydrogenation of HMF over Co₃O₄ NSA was investigated. As for the anodic part, there were two possible mechanisms for the electrocatalytic oxidation of HMF, namely, direct oxidation and indirect oxidation [11]. In direct oxidation, the valence state of the catalyst keeps unchanged throughout the reaction, and the potential is used to directly drive the oxidation of substrates. In indirect oxidation, the potential does not directly drive the oxidation of the substrate but changes the chemical state of the catalyst, and the catalyst serves as a redox mediator to oxidize the substrate [11]. In this work, an indirect oxidation mechanism was preferred due to the following phenomenon: First, in the CV tests, the addition of HMF enhanced the oxidation peak of Co³⁺, which is possibly due to the generated Co⁴⁺ being quickly reduced by HMF, thus accelerating the oxidation process from Co³⁺ to Co⁴⁺ (Fig. S10a in Supporting information). The electron transfer between the electrode, Co₃O₄, and HMF was fast enough thus making the HMF diffusion become the rate-limiting step, which is the reason for the linear relationship between HMF concentration and peak current (Figs. S10b and c in Supporting information). We further conducted multi-potential measurements to verify our assumption. As shown in Fig. S10d (Supporting information), the Co₃O₄ was first oxidized at 1.42 V to enrich the high-valence Co⁴⁺. When the potential was changed to an open circuit potential (OCP), a reduction current appeared, indicating a discharge from Co⁴⁺ to Co³⁺. However, when 50 mmol/L HMF was injected into the electrolyte before switching to OCP, the reduction current became not evident, suggesting the Co⁴⁺ was consumed by the injected HMF [12]. These results indicate that the oxidation of HMF over Co₃O₄ at 1.42 V is an indirect oxidation process [22].

From the above results, we can infer that a valence state circulation of cobalt in Co₃O₄ existed in the catalytic oxidation process. (Fig. 6a) In detail, Co³⁺ was first oxidized into Co⁴⁺ under the positive potential (reaction 1), and HMF was hydrated in the presence of OH⁻ (reaction 2). The high-valence Co⁴⁺ could directly seize proton and electron from the hydrated aldehyde group, which led to the generation of HMFCA and Co³⁺ (reaction 3). Co⁴⁺ could continue to oxidize the hydroxy group into aldehyde group, which generated FFCA (reaction 4). Once the aldehyde group was hydrated (reaction 5), FFCA could be further oxidized into FDCA by Co⁴⁺ (reaction 6). The circulation of Co³⁺ and Co⁴⁺ continued in the oxidation process. OH⁻ promoted the oxidation of Co₃O₄ into CoO₂ and the hydration of HMF, which largely accelerated the reaction kinetics.

As for the cathodic HMF reduction, in CV results (Fig. S11a in Supporting information), the reduction of Co³⁺ to Co²⁺ seems quenched in the presence of HMF. However, the valence state of cobalt transformed into +2 after the reaction, as shown in Fig. S9a. Thus, we attribute the reduction peak at −0.15 V to the reduction of Co³⁺ to Co²⁺, and the peak at −0.4 V should be the one-electron reduction of HMF. Besides, the ECH process only proceeded in PBS electrolyte rather than the non-buffered Na₂SO₄ electrolyte (Fig. S11b in Supporting information), which indicates that PBS rather than water was the proton donor. Considering that HER reaction was accompanied by the ECH process, which suggests adsorbed hydrogen (H*) must exist on the electrode [23], we propose a two-step reduction mechanism (Fig. 6b). That is, Co₃O₄ was first reduced to a lower-valence Co(OH)₂ (reaction 7), and H* was formed at the surface of Co(OH)₂ (reaction 8). The adsorbed HMF could be reduced to a ketyl radical through proton-coupled electron transfer (PCET, reaction 9), which could be further reduced to DHMF by H* or another (PCET) process (reaction 10).

In summary, this work demonstrates that HMF could be efficiently converted into two important biobased polymer precursors, DHMF and FDCA, in a paired electrolyzer with a single non-noble metal-based catalyst. The Co₃O₄ NSA/Ni foam could work as a bifunctional catalyst for simultaneous electrocatalytic oxidation and hydrogenation of HMF. Under the optimized conditions, 78.8% yield of DHMF and 95.7% of FDCA were obtained along with an overall Faradaic efficiency of 127.0%. This work shows the potential advantages of paired electrolysis for achieving efficient utilization of electrons for producing valuable chemicals.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

The authors gratefully acknowledge the support from the National Natural Science Foundation of China (Nos. 22122608, 21976170, U20A20325 and 51821006), the Fundamental Research Funds for the Central Universities (Nos. WK3530000009, YD2400002001), and the Program for Changjiang Scholars and Innovative Research Team in University of the Ministry of Education of China.

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.cclet.2022.108034.