The conversion of renewable and abundant biomass feedstock available in nature into chemicals and fuels is considered a feasible technique to alleviate the current increasing environmental and resource problems [1-3]. Furfural (FA) is an important platform chemical, industrially produced from lignocellulosic biomass through acidic hydrolysis. It can be converted into a variety of important chemicals and fuels, such as 1, 2-pentanediol (1, 2-PeD), 1, 5-pentanediol (1, 5-PeD), furan, 2-methyl furan, cyclopentanone, and γ-valerolactone [4-7]. Particularly, the selective hydrogenolysis of FA and its derivatives, furfuryl alcohol (FFA) and tetrahydrofurfuryl alcohol (THFA), into useful diols, such as 1, 2-PeD and 1, 5-PeD, is noticeably attractive [8-12]. These PeDs are widely used for the production of microbicides, cosmetics, polyesters, plastics, etc. [6, 13]. Nowadays, PeDs are generally produced via a cost-intensive multistep route from petroleum-derived feedstocks involving selective oxidation and reduction reactions [9, 13]. The availability of FA from abundant renewable lignocellulose [3], in contrast to non-renewable C₅ petroleum feedstocks, makes the selective hydrogenolysis of FA and its derivatives of FFA/THFA a practical pathway for the sustainable production of PeDs with high energy efficiency.

Thus far, the selective hydrogenolysis of FA or its derivatives at the C–O bonds in the furan ring to produce PeDs mainly focused on using supported catalysts of group Ⅷ precious metals (Ru [13], Pt [12, 14, 15], Rh [16-21], and Ir [22-26]). Low-valency metallic oxides (such as ReO_x, MoO_x, VO_x or WO_x)-incorporated Rh [16, 18, 19, 21] and Ir [22-24, 26] catalysts attracted much attention for the hydrogenolysis of THFA to synthesize 1, 5-PeD. Tomishige's group [19-25] led pioneering work on such reactions and achieved a high 1, 5-PeD yield of 94.0% at 373 to 393 K and 8 MPa H₂. Further, basic supports, such as MnO_x, CeO₂, and hydrotalcite (HT)-supported noble metals of Ru [13] and Pt [12, 14] were applied to convert FA/FFA to 1, 2-PeD at 423–443 K and 1–2 MPa H₂; 1, 2-PeD with a yield of up to 80% was achieved over a Pt/HT catalyst at a high Pt to FFA molar ratio of 1/1. In addition, there are a few reports on the use of non-precious metals, such as Cu [8, 27-29], Co [10, 30], and Ni-based [11, 31] catalysts in the hydrogenolysis of FFA and THFA to PeDs. Generally, Cu-based catalysts exhibit higher selectivity toward 1, 2-PeD, while Co and Ni-based catalysts prefer the generation of 1, 5-PeD. For instance, Cu-Mg₃AlO_4.5 [8] with a basic support hydrogenolyzed FFA to 1, 2-PeD and 1, 5-PeD with yields of 51.2% and 28.8%, respectively, at 413 K and 6 MPa H₂, while a Ni-Y₂O₃ [11] composite catalyst selectively hydrogenolyzed FFA to 1, 5-PeD (41.9%) rather than 1, 2-PeD (1.2%) at 423 K and 2 MPa H₂. Clearly, studies on FA and its derivatives subjected to hydrogenolysis still encounter low activities or selectivities toward target PeDs, or the use of noble metal catalysts. It is thus urgent to develop effective and environmentally benign non-precious metal catalysts for the efficient conversion of FA and its derivatives into value-added PeDs. However, the development of effective methods for tuning the chemoselectivity of FA and its derivatives for hydrogenolysis is a great challenge at present.

Recently, owing to the flexibility of their electronic and crystal structure as well as their chemical versatility, perovskite-type oxides with a general ABO₃ structure have been studied for catalysis applications [32-37]. Using perovskite-type oxides as precursors to stabilize metal particles on mixed oxides is also an attractive option to produce active and stable catalysts [32, 33]. In addition, the specific surface acidity/basicity of perovskite-type oxides also contributes to the novel performance of the catalysts [36, 37]. Taking these advantages of perovskite-based catalysts and the high C–O hydrogenation/hydrogenolysis activity of the non-noble metal Cu [8, 38, 39] into consideration, a series of Cu-LaCoO₃ perovskite-type mixed oxides with different Cu loadings were synthesized in the current study and investigated for the selective hydrogenolysis of FFA to value-added PeDs. The Cu-LaCoO₃ catalysts were found to exhibit a higher selectivity for 1, 5-PeD over 1, 2-PeD, with a 1, 5-PeD/1, 2-PeD selectivity ratio of up to 3/1; this result is much higher than that obtained in previous studies on Cu-based catalysts, such as copper chromite (3/4) [29], Cu-Mg₃AlO_4.5 (3/5) [8], and Cu-Al₂O₃ (1/2.2) [27]. The possible reasons for the higher selectivity of Cu-LaCoO₃ catalysts towards 1, 5-PeD are discussed in terms of the effect of different reduction atmospheres and the catalysts characterized at different states are.

All the reagents were of analytical grade and directly used without further pretreatment. H₂PtCl₆·6H₂O and RuCl₃·3H₂O were purchased from Shaanxi Rock New Materials Co., Ltd., China. PdCl₂ and RhCl₃·3H₂O were purchased from Beijing HWRK Chem Co., Ltd., China. CuCr₂O₄ catalyst was purchased from Yingkou Tianyuan Chemical Industry Research Institute Co., Ltd., China. FFA and THFA were purchased from Alfa Aesar. 5%H₂-95%Ar, 5%H₂-95%N₂, H₂ (99.999%), and He (99.999%) were obtained from Lanzhou Lanmei Cryogenic Products Co., Ltd., China. CO₂ (99.99%) was obtained from Lanzhou Hongli gas Co., Ltd., China.

In this study, a series of xCuO-LaCoO₃ (x/% = 0, 2, 5, 10, 15, and 20) perovskite-type mixed oxides were synthesized by the citric-complexing method [32]. The desired amounts of copper, cobalt, and lanthanum nitrates with a La/Co molar ratio of 1/1 were dissolved in deionized (DI) water at a concentration of Cu²⁺ = 0.1 mol/L. Later, citric acid (20% excess over the total molar content of metal cations) and polyethylene glycol 400 (24% molar amount of citric acid) were added to the solution. The solution was stirred at room temperature for 6 h and then stirred at 353 K until a foamy solid was formed. The foamy solid was dried at 383 K for 12 h and then calcined at 973 K for 2 h in static air with a temperature ramping rate of 3 K/min. The calcined samples were marked as xCuO-LaCoO₃, where x represents the nominal Cu loading. The corresponding reduced catalysts were labeled as xCu-LaCoO₃. LaCoO₃-supported Pt, Ru, Pd, and Rh catalysts with individual nominal loading of 5 wt% were also prepared by a similar method.

X-ray diffraction (XRD) experiments were performed on a PANalytical X'pert Pro Diffractometer with nickel-filtered Cu K_α radiation (λ = 1.05406 nm) at 40 kV and 40 mA. The patterns were collected in the 2θ range of 10°–80° at a scanning speed of 10°/min.

The Brunauer-Emmett-Teller (BET) surface area of the catalysts was determined using the N₂ adsorption-desorption method and the experiments were conducted on a Micromeritics Tristar Ⅱ 3020 instrument at liquid nitrogen temperatures (77 K). Prior to these measurements, the samples were pretreated with N₂ at 363 K for 2 h and then maintained for 4 h at 573 K.

The reducibility and surface acidity/basicity of the calcined samples were determined by H₂-temperature-programmed reduction (TPR) and NH₃/CO₂-Temperature programmed desorption (TPD) measurements, respectively, with a DAX-7000 instrument (Huasi Technology Co., Ltd, China). For H₂-TPR experiments, ~50 mg of the samples was placed in a quartz cell and pretreated at 473 K under He flow for 1 h. After cooling to 303 K, the samples were reduced in 5%H₂-95%Ar flow (40 mL/min) and the temperature was increased linearly to 1073 K at a ramping rate of 10 K/min. H₂ consumption was monitored by thermal conductivity detector (TCD). For CO₂-TPD experiments, ~0.2 g of the samples was placed in a quartz cell and pretreated at 473 K under He flow for 1 h and then reduced at 573 K in 5%H₂-95%Ar flow (40 mL/min) for 2 h. After cooling to 313 K, the samples were exposed to CO₂ flow (40 mL/min) and maintained for 1 h. Subsequently, the temperature was increased linearly to 1173 K at a ramping rate of 5 K/min under He flow; the desorbed CO₂ was monitored by TCD. NH₃-TPD of the 10 wt% Cu-LaCoO₃ catalyst was similar to the CO₂-TPD procedure above but with the saturation adsorption of NH₃ at 373 K for 1 h.

Transmission electron microscopy (TEM) tests were performed on a TECNAI G2 TF20 instrument at 200 kV. The TEM samples were prepared by ultrasonic dispersion in ethanol. After ultrasonic dispersion of the catalysts, the samples were deposited on copper grids with a porous carbon film support.

X-ray photoelectron spectra (XPS) measurements were carried out on an ESCALAB250xi spectrometer equipped with an Al K_α X-ray radiation source (hυ = 1486.6 eV). The sample binding energy can be calibrated with C 1s (E_b = 284.8 eV) peak as the internal standard, the error is about ± 0.2.

The chemical composition of the catalysts was determined by X-ray fluorescence (XRF) on a PANalytical MagixPW2403 instrument.

Carbon deposition and stability of the used catalysts were determined on a NETZSCH STA449F3 thermogravimetry-differential scanning calorimetry (TG-DSC) instrument. Initially, 10 mg of the samples were placed in an alumina crucible and heated to 1073 K at a heating rate of 10 K/min in a nitrogen atmosphere.

The selective hydrogenolysis of FFA was carried out in a 100 mL stainless steel autoclave at a stirring speed of 800 r/min. All the calcined samples were used in the powder form. Prior to each test, the calcined samples with a granule size of 60–80 mesh were pre-reduced in 5%H₂-95%N₂ flow (40 mL/min) at 573 K for 3 h. In a typical run, 30 g of 5 wt% FFA in ethanol solution together with the pre-reduced catalyst were introduced into the autoclave. After purging thrice with H₂, the reactor was pressurized to 6 MPa and heated to 413 K to start the reaction. For comparison, the hydrogenolysis of THFA was also studied over the 10Cu-LaCoO₃ catalyst using similar conditions.

After centrifugation, the products were identified using an Agilent 7890A/5975C gas chromatograph-mass spectrometer (GC-MS) with an HP-5MS column. The reactant and liquid products were analyzed by gas chromatography (Agilent 7890A GC) with a PONA capillary column (50 m × 0.20 mm × 0.50 μm) and a flame ionization detector (FID). Conversion and product selectivity were determined by an internal standard method and calculated as follows:

Fig. 1 shows the XRD patterns of calcined xCuO-LaCoO₃ samples with different Cu loadings. Strong peaks due to the formation of LaCoO₃ perovskite-type oxides were seen in the pattern of the LaCoO₃ support. Additionally, segregated phases due to the formation of La₂O₂CO₃ (diffraction peaks at 22.3°, 25.8°, 30.4°, 44.4°, and 47.4°) and Co₃O₄ (at 36.8 °) were also observed. La₂O₂CO₃ was probably formed by CO₂-induced decomposition of LaCoO₃ to La₂O₃ and Co₃O₄ and the further reaction of La₂O₃ and CO₂ in air during the calcination process [33]. The incorporation of Cu, even at an amount as small as 2 wt%, resulted in the decomposition of the LaCoO₃ perovskite structure to La₂O₂CO₃ and Co₃O₄ and almost no diffraction peaks associated with the LaCoO₃ perovskite could be observed at 5 wt% Cu. Furthermore, the diffraction peaks of Co₃O₄ further intensified in the pattern of 5CuO-LaCoO₃. No clear diffraction peaks corresponding to CuO could be seen in the samples with Cu loading below 5 wt%, which is probably due to the high dispersion of Cu in these samples. Increasing the Cu loading to 10 wt% led to the appearance of diffraction peaks corresponding not only to the LaCoO₃ perovskite, but also CuO, with a simultaneous decrease in the Co₃O₄ diffraction peaks. The peaks of the LaCoO₃ perovskite and CuO further intensified with an increase in the Cu loading to 20 wt%, while the peaks assignable to Co₃O₄ almost disappeared in the pattern of 20CuO-LaCoO₃. It seems that the incorporation of small amounts ((2–10) wt%) of Cu caused the segregation of LaCoO₃, which is similar to the effect observed when Zn was incorporated in LaCoO₃ [40]. The gradual increase in the intensity of CuO peaks with an increase in the Cu loading could be ascribed to the aggregation of CuO particles.

Fig. 2 presents the XRD patterns of 10Cu-LaCoO₃ samples at different states. After reduction in pure H₂ at 573 K, the diffraction peaks of CuO disappeared; meanwhile, the diffraction peaks corresponding to cubic Cu⁰ (PDF#04-0836) appeared (Fig. 2(2)), suggesting that CuO in the sample was reduced to Cu⁰. The disappearance of Co₃O₄ diffraction peaks suggests that Co₃O₄ was reduced to Co species with low valencies, i.e., Co⁰ and/or CoO. Nonetheless, no diffraction peaks associated with Co⁰ or CoO could be observed, which is probably due to the high dispersion of these species. After reduction in 5%H₂-95%N₂ at the same temperature, however, almost no diffraction peaks associated with Cu⁰ and/or Cu₂O were observed (Fig. 2(3)), while the diffraction peaks of CuO (PDF#80-1268) and Co₃O₄ (PDF#74-2120) still remained, showing that the catalyst was hardly reduced under these conditions. These findings also suggest that the catalyst exhibited a high reducibility in pure H₂. As for the catalyst reduced in 5%H₂-95%N₂ and used for one time, the diffraction peaks of Cu⁰ appeared in its XRD pattern; no CuO could be observed and the intensity of the diffraction peaks of Co₃O₄ decreased (Fig. 2(4)), indicating that CuO was reduced to Cu⁰ and Co₃O₄ was largely reduced to Co⁰ and/or CoO with high dispersion. This finding also suggests that the reaction media of FFA in alcohol has a stronger reducibility than 5%H₂-95%N₂ at 573 K. The diffraction peaks of Co₃O₄ almost disappeared after the catalyst was used for 4 cycles (Fig. 2(5)), indicating that Co₃O₄ was continuously reduced to Co⁰ and/or CoO during the repeat reactions. The reduction of CoO, which is more difficult to be reduced than Co₃O₄ [10, 41], is also observed during glycerol hydrogenolysis [41]. The intensity of the diffraction peaks of Cu⁰ as well as those of La₂O₂CO₃ (Fig. 2(5)) decreased after the catalyst was used four times. Such obvious structural changes in the catalyst during repeated runs may be related to the changes in its catalytic performance, as will be discussed later.

Fig. 3 shows the XPS survey of the Cu and Co species. As can be seen in Fig. 3(a), the binding energy of Cu 2p_3/2 at 933.5 eV along with a satellite peak in the range of 940–945 eV indicate the existence of Cu²⁺ in the calcined sample [42]. The decrease in the binding energy of Cu 2p_3/2 to 932.5 eV and the weakening or even disappearance of the satellite peak suggests that Cu species were reduced to Cu with low valencies (Cu⁰ or Cu₂O) in both reduced samples as well as the samples used for one time. The simulation of Cu 2p (Fig. 3(a)–(2)) for the reduced catalyst confirmed the coexistence of large amounts of Cu species with low valencies. The decrease in the intensity of the Cu 2p spectra with a high binding energy corresponding to the used catalyst indicated a decrease in the Cu²⁺ amount [42]. As for the Co 2p spectrum of the calcined sample, the asymmetric peak at 780.5 eV corresponding to Co 2p_3/2 and the presence of a weak peak of Co 2p_1/2 at about 796.6 eV are characteristic of Co³⁺ [43]. The peaks of Co 2p_3/2 at 782.3 eV and the weak peak at 798.5 eV were associated with Co²⁺. The simulation of Co 2p ((b)–(2)) for the reduced catalyst confirmed the coexistence of Co³⁺ and Co²⁺ species [43]. From Fig. 3(b), we can see that after being used once, the Co 2p signal weakened (Fig. 3(b)–(3)) and exhibited a tendency to shift towards lower binding energies, indicating an increase in low-state Co species. These findings are in line with the results of XRD characterization shown in Fig. 2.

Fig. 4 shows the TEM images of calcined CuO-LaCoO₃ samples with Cu loadings of (2, 10, and 20) wt% as well as those of reduced and used 10Cu-LaCoO₃ catalysts. The calcined samples generally exhibited an amorphous structure, which is in line with previous findings by Predoana et al. [44]. It is difficult to discriminate between CuO nanoparticles and the LaCoO₃ support when the Cu loading was low (i.e., 2 wt%), which indicates the uniform dispersion of CuO. With further increase in Cu loading, a small amount of aggregated particles with sizes around 5–10 nm were observed in 10CuO-LaCoO₃ and the aggregated particles became more obvious when the Cu loading was 20 wt%. Such aggregation could be ascribed to CuO, as evidenced from XRD characterization (Fig. 1). There was no obvious change in the morphology of the 10Cu-LaCoO₃ catalyst after reduction in 5%H₂-95%N₂ at 573 K (Fig. 4(d)). High resolution transmission electron microscope (HRTEM) images and Energy dispersive spectrdmeter (EDS) mapping analysis of the reduced 10Cu-LaCoO₃ catalyst indicate that Cu, Co, and La were distributed quite evenly and La₂O₂CO₃, LaCoO₃ and Co₃O₄ coexisted in the catalyst; these observations are consistent with the XRD results. However, a number of dark particles could be clearly seen when the catalyst was used four times. These dark particles may be associated with the continuous sintering of Cu⁰ particles during the reaction process. Note that the deposited carbon due to coke formation can also cause the particles in the used catalyst to appear a little bit darker. The formation of coke in the used catalyst was supported by a clear mass loss in the TG profile in the range of 550–855 K, as will be shown later.

The reducibility of the calcined samples was studied by TPR (Fig. 5). It can be seen that the LaCoO₃ support was reduced in two major regions. The first one in the range of 644–890 K and centered at about 763 K might be associated with the reduction of Co³⁺ to Co²⁺ and the second one at temperatures above 890 K might be attributed to the reduction of Co²⁺ owing to interactions with La³⁺ species to Co⁰ [45]. An obvious shift in the reduction peak towards lower temperature regions was observed upon the incorporation of 2 wt% Cu. The onset reduction temperature decreased noticeably from 644 K (LaCoO₃ support) to 527 K (2CuO-LaCoO₃) and the peak temperature also decreased to 653 K for the latter sample. The major peak in the low-temperature region for the catalysts with (5 and 10) wt% Cu broadened and the peak shifted to around 713 K. Further increase in Cu loading to 15% and above led to an obvious broadening and shifting of the reduction peak towards higher temperatures. Because CuO can be reduced at lower temperatures when compared to CoO_x, the obvious shift in the reduction temperature for low-Cu loading ((2–10) wt%) catalysts might be ascribed to the spillover of the adsorbed hydrogen from Cu⁰ to cobalt oxides [10, 46]. Thus, it is likely that the collective reduction of Cu²⁺ to Cu⁰ and Co³⁺ to Co²⁺ as well as a small amount of Co²⁺ to Co⁰ contributes to the major peak at temperatures below 890 K for samples with Cu loading below 10 wt%. The shifting of the reduction temperature to higher regions for samples with high Cu loadings (> 10%) could be attributed to the formation of CuO and perovskite-type LaCoO₃ with large crystallite sizes, as inferred from the XRD characterization results (Fig. 1).

Because the surface basicity of catalysts plays an important role in the catalytic hydrogenolysis of FA and FFA [8, 13, 14], the basicity of xCu-LaCoO₃ catalysts was characterized by CO₂-TPD (Fig. 6). A very small peak at around 598 K was detected in each sample, which could be attributed to moderate basic sites. It can also be seen that all the samples showed obvious CO₂ desorption at temperatures above 850 K, which may be associated with the formation of carbonate species when the sample was exposed to CO₂ [37]. Such carbonate species would contribute to the strong basicity of the catalysts. Thus, the 10Cu-LaCoO₃ catalyst with the largest CO₂ desorption peak area at temperatures above 850 K would possess the largest amount of strong basic sites. Characterization of the acidity of 10Cu-LaCoO₃ catalyst by NH₃-TPD (Fig. S1 in Supporting Information) indicated a rather low acidity for this catalyst.

Table 1 shows the textural properties of the calcined samples. The BET surface areas of the calcined CuO-LaCoO₃ samples increased from 3.5 m²/g to a maximum of 21.7 m²/g as the Cu loading increased from 0 to 5 wt% and then decreased to 1.7 and 3.8 m²/g at Cu loadings of 15 wt% and 20 wt%, respectively. The pore volumes of the samples were below 0.06 cm³/g. Such low surface areas and pore volumes of the samples could be ascribed to the perovskite-type structure and high calcination temperatures [47, 48]. The rather high BET surface area of 5CuO-LaCoO₃ may be associated with the negligible formation of a perovskite structure in this sample, as revealed by XRD characterization (Fig. 1). The average pore diameters of the CuO-LaCoO₃ samples were in the mesopore range of 10–25 nm. The Cu and Co loadings of the samples, as measured by XRF, were close to their actual amounts (Table 1). Comparing the characterization results of BET and CO₂-TPD, it could be inferred that the catalytic performance of the catalyst is more profoundly affected by the reduction properties of the catalyst rather than the BET surface area and basicity.

Table 2 shows the conversions and selectivities of FFA hydrogenolysis over different catalysts at 413 K and 6 MPa H₂. The conversion of FFA for xCu-LaCoO₃ catalysts increased sharply from 56.2% (2Cu-LaCoO₃) to 94.6% (10Cu-LaCoO₃) and then decreased gradually to 51.6% with a further increase in the Cu content to 20 wt% (entries 2–6). The selectivities for 1, 5-PeD were in the range of 35.0%–38.1% for xCu-LaCoO₃ catalysts with low Cu loadings ((2–10) wt%), which further decreased to ~29.0% for high Cu-loading ((15–20) wt%) catalysts. Meanwhile, 1, 2-PeD selectivities dropped from ~14.0% at low Cu loadings ((2–10) wt%) to ~12.7% at high Cu loadings ((15–20) wt%). For the xCu-LaCoO₃ catalysts, THFA was determined to be the major byproduct with a selectivity in the range of 32.1%–36.5%. When the amount of the catalyst is increased (0.15 g Cu⁰), the yield of 1, 5-PeD (40.3%) is close to that obtained with recently reported Cu-Co-Al (44% yield of 1, 5-PeD) [10] and Ni-Y₂O₃ (41.9% yield of 1, 5-PeD) [11] catalysts.

For comparison, the catalytic performances of LaCoO₃-supported noble metals, Pt, Ru, Pd, and Rh, with 5 wt% loading prepared by the same citrate-complexing method and a commercial CuCr₂O₄ catalyst were also studied (entries 8–12). 5Pt-LaCoO₃ exhibited moderate but inferior FFA conversion and PeD selectivity compared to 5Cu-LaCoO₃ and 10Cu-LaCoO₃. The higher selectivity for 1, 5-PeD (26.8%) obtained over 5Pt-LaCoO₃ as compared to 1, 2-PeD (9.3%) is in line with a previous study on the hydrogenolysis of FFA over Pt-Co₂AlO₄ [15]. Although high FFA conversions (> 80%) were attained over LaCoO₃-supported Ru, Rh, and Pd catalysts, these catalysts showed a quite low selectivity towards PeDs (combined selectivity < 20%) with THFA as the predominant product (selectivity > 70%) (entries 9–11). The commercial CuCr₂O₄ catalyst exhibited a rather low FFA conversion (9.5%) and a higher 1, 2-PeD selectivity (27.3%) over 1, 5-PeD (14.2%) under similar reaction conditions (entry 12). Clearly, among the catalysts investigated, 10Cu-LaCoO₃ exhibited the best performance for the production of PeDs by FFA hydrogenolysis with both high FFA conversion and PeD selectivity. Thus, this catalyst was selected for further studies.

To investigate the role of Cu⁰ and LaCoO₃ in FFA hydrogenolysis, LaCoO₃, Cu⁰ + LaCoO₃ physical mixture, and Cu⁰ catalysts were also tested (entry 1, 13, and 14). The LaCoO₃ support showed an extremely low FFA conversion (2.2%) and almost no activity for the cleavage of the furan ring C–O bonds and resulted in THFA as the main product (entry 1). Pure Cu⁰ exhibited a low 1, 2-PeD selectivity (2.7%) at a slightly higher FFA conversion (4.4%) as compared to the LaCoO₃ support (entry 14). Interestingly, the Cu⁰ + LaCoO₃ physical mixture catalyst exhibited not only a higher FFA conversion (7.8%), but also a higher selectivity towards PeDs with 1, 5-PeD being the major product (entry 13). These findings indicate that Cu⁰ provides the majority of the active sites for the generation of 1, 2-PeD from FFA, while the coexistence of Cu⁰ and LaCoO₃ with strong basicity (Fig. 6) could steer the cleavage of the furan ring C–O bond to produce 1, 5-PeD. The remarkably high FFA conversion and PeD selectivity obtained with 10Cu-LaCoO₃ (entry 4) indicates the importance of cooperation between Cu and LaCoO₃ with high dispersion for the efficient hydrogenolysis of FFA into PeDs. It seems that the partially reduced Cu⁰-CoO boundary played an important role in the selective hydrogenolysis of FFA to 1, 5-PeD [10], while Cu⁰ was mainly responsible for the production of 1, 2-PeD.

Fig. 7 displays the effect of reaction temperature on FFA hydrogenolysis activity over 10Cu-LaCoO₃ at 6 MPa H₂ in the FFA conversion range of 15%–25%. The reaction rate of PeDs, defined as moles of PeDs produced by a mole of the active metal (Co+Cu) with respect to time, increased drastically from 0.1 h^–1 at 393 K to 5.4 h^–1 at 453 K (Fig. 7(a)), with an activation energy of 47.8 kJ/mol. The selectivity for 1, 2-PeD and 1, 5-PeD firstly increased from 8.8% and 21.0% at 393 K to the maximum values of 13.7% and 38.5% at 433 K, respectively, and then were almost constant as the temperature increased to 453 K. An increase in the temperature resulted in a sharp drop in THFA selectivity from 61.1% at 393 K to 19.8% at 453 K, which shows that the increase of the relative reaction rate for C=C bond hydrogenation is lower than that of C–O bond hydrogenolysis with the increase of temperature. The selectivity for n-pentanol and 2-pentanol (referred to as pentanols), which are dehydration products of PeDs, increased linearly from 1.1% at 393 K to 9.5% at 453 K. Similarly, the selectivity for the C–OH dehydration products of FFA, 2-methylfuran (2-MF) and 2-methyltetrahydrofuran (2-MTHF), increased uniformly from 1.2% at 393 K to 14.9% at 453 K.

Because the hydrogenolysis of cyclic ether C–O bonds produces PeDs, C–OH bond dehydration and furan ring C=C bond hydrogenation of FFA are competitive reactions; at the same time, further hydrogenolysis of PeDs to pentanols (dehydration) and hydrogenation of 2-MF (furan ring C=C bonds) to 2-MTHF are secondary reactions involved in FFA hydrogenolysis [6, 27]. Fig 7(b) depicts a clearer picture seeing the changes of the activity of cyclic ether C–O bonds hydrogenolysis, C–OH bond dehydration and furan ring C=C bonds hydrogenation of the catalyst with the change of reaction temperature. The selectivity for PeDs and pentanols (cyclic ether C–O bond hydrogenolysis products) increased monotonously with increasing temperature; meanwhile, the selectivity for 2-MF, 2-MTHF, and pentanols (C–OH dehydration products) also increased linearly, suggesting that higher temperatures favor not only the cleavage of cyclic ether C–O bonds, but also the hydrogenolysis of C–OH bonds. In contrast, the selectivity for furan ring C=C bond hydrogenation products, THFA and 2-MTHF, decreased sharply from 61.1% at 393 K to 21.7% at 453 K with increasing temperature. Clearly, higher temperatures are more favorable for the hydrogenolysis of C–O over 10Cu-LaCoO₃ than for the hydrogenation of furan ring C=C bonds. Thus, an appropriately high temperature (433 K) is found to benefit the hydrogenolysis of FFA to PeDs.

Fig. 8 shows the effect of H₂ pressure on FFA hydrogenolysis over 10Cu-LaCoO₃ at a controlled FFA conversion of 15%–25%. The reaction rate for PeD production rapidly increased from 0.1 h^–1 at 2 MPa to 1.7 h^–1 at 8 MPa (Fig. 8(a)). The selectivity of 1, 5-PeD firstly increased from 25.2% at 2 MPa to a maximum of 34.3% at 6 MPa and then remained almost constant as the H₂ pressure increased to 8 MPa; 1, 2-PeD selectivity remained stable at 12% with an increase in H₂ pressure (Fig. 8(a)). The selectivity of THFA increased steadily from 27.3% at 2 MPa to 44.1% at 8 MPa with increasing H₂ pressure. Simultaneously, the combined selectivity of 2-MF and 2-MTHF decreased from 14.6% to 3.7% during the same process. In addition, a slight decrease in pentanol selectivity from 5.5% to 3.5% was observed with an increase in pressure. The above findings suggest that a high H₂ pressure favors not only the hydrogenolysis of cyclic ether C–O bonds but also the hydrogenation of furan ring C=C bonds; the reaction rate of the latter increased more rapidly, leading to a slight decrease in the products of the former reaction at relatively high pressures of 8 MPa. The gradual decline of the combined selectivity of the C–OH bond dehydration products of pentanols, 2-MF, and 2-MTHF indicates that the C–OH bond dehydration reaction was suppressed with increasing H₂ pressure (Fig. 8(b)). Clearly, an appropriately high H₂ pressure (~6 MPa) favors the production of the target PeDs at high yields.

The conversions and selectivities of 10Cu-LaCoO₃ during FFA hydrogenolysis as functions of reaction time at 413 K and 6 MPa H₂ are shown in Fig. 9. The conversion of FFA increased drastically from 25.3% at 0.2 h to 94.6% at 2 h and full conversion was achieved after 4 h. The selectivity of 1, 5-PeD increased from 25.9% at 0.2 h to a maximum of 40.6% at 4 h and then remained almost constant as the reaction time increased to 8 h. The selectivity of 1, 2-PeD increased slightly from 11.0% at 0.2 h to 14.8% after 8 h. Simultaneously, a slight decrease in the selectivity of THFA from around 36.0% at 0.2 h to 28.4% was observed with an increase in the reaction time to 8 h; this might be a result of the more favored reaction of FFA hydrogenolysis to PeDs, which were not further hydrogenolyzed to pentanols or 2-MTHF. A separate reaction on the direct hydrogenolysis of THFA under the same conditions as those employed for FFA hydrogenolysis resulted in no reaction products, supporting that both PeDs and 2-MTHF are generated from FFA but not THFA, which is in line with previous findings [8, 15, 27]. Although 2-MF could be further hydrogenated to 2-MTHF and pentanols [15, 27], the slight increase in pentanol selectivity (from 3.5% to 5.6%) and simultaneous decline in the combined selectivity of 2-MF and 2-MTHF (from 6.0% to 3.5%) after 8 h of reaction indicates that further hydrogenation of 2-MF to pentanols is not the main cause for the observed decrease in the combined selectivity of 2-MF and 2-MTHF. The more favored reaction of FFA hydrogenolysis to PeDs as compared to FFA hydrogenolysis to 2-MF probably accounts for the decline in the combined selectivity of 2-MF and 2-MTHF. Thus, it is likely that further hydrogenolysis of PeDs contributed to the increase in pentanol selectivity.

The 10Cu-LaCoO₃ catalyst was repeatedly used for FFA hydrogenolysis to study its reusability (Fig. 10). The conversion of FFA increased from 93.7% to 99.3% after one cycle and then was almost constant for three more cycles. Nonetheless, a continual decrease was observed in the selectivity of 1, 5-PeD (from 37.9% to 28.7%) and 1, 2-PeD (from 13.0% to 7.8%) during four cycle reactions. In the meantime, the selectivity of THFA increased slightly from 31.7% to 41.8%. The increase in THFA selectivity at the expense of PeD selectivity suggests that the hydrogenation activity of furan ring C=C bonds increased, while the hydrogenolysis activity of cyclic ether C–O bonds declined. Thus, the increase in FFA conversion after recycling can be mainly attributed to the enhanced hydrogenation activity of furan ring C=C bonds, leading to the generation of high amounts of THFA, which cannot be further converted to PeDs under the used reaction conditions. Similar trends for FFA conversion and product selectivities were observed when the recycling of the catalyst was studied at low initial FFA conversion (30.5%, see Fig. S2 in SI).

In order to evaluate the reasons for these changes in the performance of the catalyst, it was characterized by XRD (Fig. 2) and TEM (Fig. 4) at three different states. The appearance of a Cu⁰ diffraction peak along with a largely disappeared Co₃O₄ diffraction peak after catalyst reduction (in 5%H₂-95%N₂) and usage for one time (Fig. 2((3) and (4))) gives us to understand that the active sites for FFA hydrogenolysis are probably associated with the presence of Cu⁰ and Co with low valencies. Because Cu⁰ itself (Table 2, entry 14) and its combination with acidic and basic supports favor the generation of 1, 2-PeD over 1, 5-PeD during FFA hydrogenolysis [8, 27, 29], the higher 1, 5-PeD selectivity obtained over Cu-LaCoO₃ catalysts as well as the physical mixture of Cu⁰ + LaCoO₃ indicates that Co species played an important role in the selective cleavage of the secondary C–O bond of FFA. The presence of Co species preferably cleaved secondary C–O bonds but not the primary C–O bond of FFA, which is in line with previous findings [10, 15, 30]. It was reported that the presence of large amounts of Co²⁺ facilitates the adsorption of FFA in a tilted conformation, which enhances the cleavage of the secondary C–O bonds of FFA [10], leading to a high selectivity toward 1, 5-PeD. Because CoO is much harder to be reduced than Co₃O₄ [10, 46], the partial reduction of Co₃O₄ to highly dispersed CoO in a hydrogenation atmosphere as well as the presence of an alcoholic solvent might be responsible for the largely diminished Co₃O₄ diffraction peak after the 10Cu-LaCoO₃ catalyst was used one time (Fig. 2(4)). Further diminishing of the Co₃O₄ diffraction peak after the catalyst was repeatedly used for three more cycles indicates the sequential reduction of Co₃O₄ to CoO and later to Co⁰ (Fig. 2(5)). The decrease in PeD selectivity, especially 1, 5-PeD selectivity, with the continuous use of the catalyst implies a decrease in the CoO amount in the catalyst. Thus, an increase in FFA conversion and THFA selectivity may be associated with an increase in the amount of Co⁰. Note that the sintering of Cu particles, as revealed by XRD (Fig. 2((3) and (4))) and TEM (Fig. 4((d) and (f))), in the used catalyst would also cause a decline in the hydrogenolysis activity of cyclic ether C–O bonds [27]. The significant weakening of all the diffraction peaks of the 10Cu-LaCoO₃ catalyst after four cycles can be associated with carbon deposition, as evidenced by the obvious mass loss (7.2%) in between 550–855 K in the TG profile of the used 10Cu-LaCoO₃ catalyst (Fig. 11(2)) [49]. The mass loss at temperatures above 855 K for both the freshly reduced catalyst and used catalyst can be associated with the decomposition of La₂O₂CO₃ [33].

In order to further elucidate the active species in the Cu-LaCoO₃ catalyst used for FFA hydrogenolysis, the effect of reduction atmosphere on the structure and reaction properties of the 10Cu-LaCoO₃ catalyst was investigated. Prereduction of the catalyst with low concentrations of H₂ (e.g., 5%) led to high selectivity (> 36%) towards 1, 5-PeD but slightly lower FFA conversions (Table 3). The selectivity of THFA increased and it became the major product at the expense of 1, 5-PeD with increasing H₂ concentration; full conversion of FFA was obtained upon prereduction of the catalyst with pure H₂. From the H₂-TPR results (Fig. 5), it can be understood that it is hard to reduce the calcined CuO-LaCoO₃ catalyst at a low prereduction temperature of 573 K with 5% H₂; this is proved by the XRD characterization results (Fig. 2(3)). Prereduction of the catalyst with high concentrations of H₂ (i.e., 100%) enhanced the reducibility of the catalyst as revealed by the presence of the diffraction peak of Cu⁰ and simultaneous disappearance of the diffraction peak of Co₃O₄ (Fig. 2(2)). Although no diffraction peaks of Co⁰ and/or CoO were observed after the catalyst was reduced in pure H₂, Co⁰ and/or highly dispersed CoO were deduced to be present in the catalyst. It has been reported that Co⁰ exhibited a higher C=C hydrogenation reactivity than Cu⁰ [50]. Thus, a high amount of Co⁰ would promote the hydrogenation of C=C in the furan ring, leading to the generation of a large amount of a furan-ring saturated product (THFA); this observation was reported in previous studies as well [10, 27]. Clearly, the increase in FFA conversion and THFA selectivity with an increase in the concentration of H₂ in the prereduction gas (Table 3) is associated with the increase of Co⁰ in the catalyst. The decline in 1, 5-PeD selectivity with increasing H₂ gas concentration can be related to the decreasing amount of partially reduced CoO species, due to their further reduction at high H₂ concentrations. Therefore, studying the effect of the prereduction atmosphere supported the previously mentioned findings on the recycling of 10Cu-LaCoO₃; cooperative catalysis occurred between Cu⁰ and CoO, which promoted the hydrogenolysis of FFA to PeDs, especially to 1, 5-PeD, while Co⁰ promoted the hydrogenation of FFA to THFA. Taking the catalytic performances of Cu-LaCoO₃ catalysts into consideration (Table 2), it is suggested that the synergetic effect between balanced Cu⁰ and CoO sites plays a critical role in achieving a high yield of 1, 5-PeD from FFA hydrogenolysis.

In summary, Cu-LaCoO₃ with a perovskite structure was employed as an effective catalyst for the selective hydrogenolysis of biomass-derived FFA to 1, 5-PeD and 1, 2-PeD. The catalytic performances of the catalysts were found to depend on Cu loading and prereduction conditions, such as the H₂ concentration of the reducing gas. The reaction parameters, including temperature and H₂ pressure, also greatly affected the reaction activity and product selectivity. It was demonstrated that cooperation between partially reduced Co₃O₄ (probably as CoO) and fully reduced Cu species (Cu⁰) induces a synergetic effect for the selective adsorption and catalyzed opening of the furan ring, leading to 1, 5-PeD with a high yield. Increasing the reducibility of the catalyst to enhance the reduction of Co species to Co⁰ would promote the production of THFA. By optimizing the catalyst composition and prereduction and reaction conditions, a high FFA conversion of ~100% and selectivity of 55.5% for 1, 5-PeD (40.3%) and 1, 2-PeD (15.2%) were achieved at 413 K and 6 MPa H₂ over a catalyst containing 10 wt% Cu loading and prereduced in 5% H₂. The important findings of this work would shed light on the development of efficient non-noble metal catalysts for the production of valuable PeDs from biomass.