There has been a surge in demand for carbon resources for production of transportation fuels, commodity chemicals, and polymeric materials [1]. However, traditional fossil resources are being increasingly exhausted, with associated problems including excessive emissions of greenhouse gases and other environmental pollution. As a renewable carbon source, biomass can be regenerated in a relatively short span of time and has great potential to partially replace fossil resources [2-7]. For example, γ-valerolactone (GVL) derived from inedible lignocellulose can serve as a flexible building block to manufacture various value-added chemicals and biofuels, including pentenoic acid, butane, 5-nonanone, 1, 4-pentanediol, and aromatics [8, 9]. Previous work has shown C8–C18 branched/straight-chain hydrocarbons fuels from GVL can be synthesized by sequential condensation and hydrodeoxygenation [10]. In addition, the application of GVL as a novel and green solvent in biomass conversion has recently received widespread attention [11-13].

A large number of studies have been devoted to GVL production by the hydrogenation of levulinic acid (LA) and its esters, which can be readily formed by the acid-catalyzed hydrolysis or alcoholysis of biomass-derived carbohydrates [14-16]. Several review articles have provided comprehensive overviews of the hydrogenation of LA and its esters to GVL in terms of the diversity of hydrogen sources, which mainly include molecular H₂, alcohols, formic acid, and water [17]. For instance, various supported noble metals (Ru, Pt, Ir, and Au) have proved to be effective catalysts for the hydrogenation of LA and its esters to GVL, using external molecular H₂ [18-20]. However, the high costs of noble catalysts set limits on their applications [17, 19, 21-24]. In this light, inexpensive non-noble catalysts such as Cu, Mo₂C, Ni, Co, and Zr are desirable alternatives [25-30]; nevertheless, relatively harsh reaction conditions are required to obtain GVL yields comparable to those in the case of noble catalysts. Recently, the use of alcohols, such as ethanol and isopropanol, as the solvent and hydrogen source for GVL production is gaining momentum, because the management and use of alcohols are convenient compared with those of H₂. For example, Meerwein-Ponndorf-Verley (MPV) reduction of levulinate esters to GVL can be achieved over various Zr-based catalysts by using ethanol or isopropanol as the in-situ hydrogen source [16, 31].

Methanol (MeOH) is also an appealing hydrogen source that can be catalytically reformed to H₂ and CO₂/CO [32]. For instance, Barta et al. [33, 34] reported a one-pot conversion of woody biomass to aromatics and/or liquid fuels in supercritical MeOH in the presence of a Cu-based catalyst. The advantages of employing Methanol (MeOH) as the hydrogen source for the GVL production include the fact that methyl levulinate (ML) can be easily produced by the methanolysis of carbohydrates in MeOH catalyzed by sulfuric acid [35]. Therefore, a facile two-step strategy for GVL production from biomass can be developed in MeOH by the integration of acid-catalyzed methanolysis of biomass to ML and the subsequent reduction of ML to GVL. In this respect, we previously reported that CuO could serve as a bi-functional catalyst for MeOH reforming and ML hydrogenation, giving a nearly complete ML conversion and a GVL selectivity up to 87.6% at 240 ℃ in 1 h. However, the CuO could be completely reduced to metallic Cu, losing its activity for MeOH reforming. Hence, it is highly desirable to prepare robust catalysts for GVL production using MeOH as the in-situ hydrogen source.

In this work, (n)CuO_x-CaCO₃ (where n is the molar ratio of Cu to Ca) compounds were prepared by an oxalate gel-coprecipitation method and were verified as effective and robust catalysts for H₂ production through MeOH reforming and ML hydrocyclization to GVL. Incorporation of the CaCO₃ with CuO resulted in the formation of Cu⁺ species in the CuO_x-CaCO₃ catalyst, which greatly promoted the hydrogenation of ML to GVL, even in the methanolysis products of cellulose. Hence, a facile two-step strategy for the production of GVL from cellulose could be developed over the effective and robust CuO_x-CaCO₃ catalyst, through the integration of acid-catalyzed methanolysis of cellulose, MeOH reforming, and ML hydrocyclization in MeOH medium.

MeOH, ethanol, oxalic acid, and all metallic nitrates, including Cu(NO₃)₂·3H₂O, Ca(NO₃)₂·4H₂O, Mg(NO₃)₂·6H₂O, Ba(NO₃)₂, Ni(NO₃)₂·6H₂O, Zn(NO₃)₂·6H₂O, and Al(NO₃)₃·9H₂O, were purchased from Sinopharm Chemical Reagent Co. Ltd. GVL (98%) and ML (99%) were supplied by Aladdin Reagent Co. Ltd. (Shanghai, China) and TCI (Shanghai) Development Co. Ltd. (Shanghai, China), respectively. All the chemicals were analytical reagents and used without further purification.

CuO_x-CaCO₃ with varied Cu/Ca molar ratios and other Cu-based bimetallic catalysts were prepared by an oxalate gel-coprecipitation method. For preparation of the (3/2)CuO_x-CaCO₃, the desired amounts of Cu(NO₃)₂·3H₂O and Ca(NO₃)₂·4H₂O (Cu and Ca precursors) were dissolved into absolute ethanol (0.1 M of the total nitrates) under vigorous magnetic stirring (1000 rpm) at room temperature. Then, the saturated oxalic acid/ethanol solution (oxalic acid/(Cu²⁺ + Ca²⁺) molar ratio = 1.5) was rapidly added into the prepared metal precursor solution. The resultant gel-like precipitates were formed within several seconds. The gel-like precipitates were aged at room temperature for 30 min and separated by centrifugation to give the catalyst precursor, which was dried at 110 ℃ overnight. The dried gel was ground to pass through 100 mesh and calcined at 500 ℃ for 5 h in air atmosphere to give the target (3/2)CuO_x-CaCO₃ catalyst [36-38]. CuO and CaCO₃ were also prepared by the same oxalate gel-coprecipitation method. CaO was obtained by the calcination of CaCO₃ at 900 ℃ for 5 h.

Chemical compositions of the fresh and spent catalysts and possible elemental leaching in the reaction filtrates were measured using inductively coupled plasma optical emission spectrometry (ICP-OES, Agilent Technologies 7700X) after digestion in concentrated HNO₃. The X-ray diffraction (XRD) patterns of the catalysts were obtained using a Rigaku Ultima Ⅳ diffractometer with a Cu K_α radiation source and the following parameters: 40 kV, 30 mA, 2θ from 10° to 80°, at a scanning speed of 10°/min. The specific surface areas of the catalysts were determined by N₂ adsorption-desorption analysis at –196 ℃ on a Micromeritics ASAP 2020 HD88. Degassing was performed before sorption measurements for all samples. Scanning electron microscopy (SEM) images were obtained using a Hitachi S-4800 microscope with an accelerating voltage of 15 kV. Energy dispersive X-ray spectrometry (EDS) and energy-dispersive X-ray (EDX) element mapping images were obtained using a ZEISS SUPRA 55 SAPPHIRE field emission scanning electron microscope. X-ray photoelectron spectroscopy (XPS) spectra were recorded with a Thermo Scientific ESCALAB 250Xi electron spectrometer under a base pressure < 2 × 10^-9 mbar. A monochromatic Al K_α (hν = 1486.6 eV) excitation energy source was used. NH₃ temperature-programmed desorption (NH₃-TPD) and H₂ temperature-programmed reduction (H₂-TPR) were performed on a Micromeritics AutoChem Ⅰ 2920, equipped with a thermal conductivity detector (TCD). For NH₃-TPD, after loading into the quartz tube, a sample (40 mg) was first pretreated at a rate of 20 ℃/min up to 550 ℃ in He gas flow for 1 h to remove adsorbed species on its surface. Then, the temperature was reduced to 120 ℃ at a rate of 50 ℃·min^-1 and the sample was treated with 10% NH₃-He (vol%) mixture for 0.5 h. After NH₃ adsorption, the remaining or weakly adsorbed NH₃ was purged by high-purity He for 1 h at 120 ℃. TPD was performed in He flow by raising the temperature to 800 ℃ at a rate of 20 ℃/min and then holding at 800 ℃ for 0.5 h. H₂-TPR was performed using a similar procedure under 10% H₂/Ar. Elemental analysis for C and H was conducted with an Elementar Vario EL Ⅲ (Elementar Analysensysteme GmbH, Germany), and all samples were measured twice for averaging.

All experiments were carried out in a 100-mL stainless steel high-pressure reactor (PARR, USA). After loading of ML (4 mmol), MeOH (19.5 g), and catalyst (0.25 g), the reactor was weighed, and then purged five times with N₂ before being sealed and heated to the target temperature with stirring (500 rpm). The reactor was naturally cooled to room temperature after the reaction, and the pressure was recorded before the collection of gaseous products. Gaseous products were analyzed by gas chromatography (GC) methods. The reactor was weighed again upon opening, and the weight loss compared with before the reaction was calculated. The weight loss in the present study was ascribed to the consumption of MeOH by reforming. The used catalyst was recovered by vacuum filtration, and washed with MeOH and ethanol before being dried at 60 ℃ for 2 h. Liquid products were centrifuged at 8000 rpm for 6 min before analysis by GC-mass spectrometry (GC-MS) and GC methods, respectively.

About 16 g microcrystalline cellulose (MCC) (90 μm) was added into a 400 mL Hastelloy-C high pressure reactor (Dalian-controlled plant), followed by 200 g MeOH solution with a H₂SO₄ concentration of 0.01 mol/L. The autoclave was heated to 200 ℃ and maintained at that temperature for 4 h under vigorous stirring (500 rpm). The reactor was cooled to room temperature after the reaction, and only a small quantity of insoluble humins was observed. Excess CaCO₃ was added to neutralize H₂SO₄, and the resulting CaSO₄ was separated by vacuum filtration. The resultant filtrate was also centrifuged and analyzed by the GC method. Based on the glucose units in cellulose, the weight yield of ML was 32.4% (44.8% molar yield), and the resulting liquid products contained around 3.7 wt% ML.

Next, 13.9 g of the methanolysis product (containing around 4 mmol ML) and 0.25 g catalyst were placed in the Parr reactor as described above. The subsequent procedure for the GVL production was the same as that detailed in Section 2.4.1.

The analysis for gaseous products was performed on a Shimadzu GC-2010 Plus analyzer equipped with an Agilent CP-7429 column and a TCD. GC analysis for liquid samples was conducted on an Agilent 7890 series equipped with a DB-Waxetr column (30.0 m × 0.25 mm × 0.25 μm) and a flame ionization detector operated at 270 ℃. The carrier gas (N₂) had a flow rate of 1.0 mL/min. Temperatures were programmed as follows: 40 ℃ (4 min), 15 ℃/min, 250 ℃ (5 min). The qualitative analysis of products after reaction was conducted using a Thermofisher Trace 1300 & ISQ LT GC-MS instrument with a TR-5MS column (15.0 m × 0.25 mm × 0.25 μm). Temperatures were programmed as follows: 40 ℃ (2 min), 150 ℃/min, 280 ℃ (2 min). The flow rate was 1.2 mL/min and the split ratio was 1:50 for the carrier gas (He). The mass spectra were obtained by electron impact ionization (EI) at an electron energy of 70 eV and with a 25 μA emission current. The operating parameters for the GC-MS analysis were consistent with those of the GC analysis. Note that the quantitative analysis of the by-products (hydrocarboxylation products and ring-opening products) was performed according to the standard curve of GVL, because these by-products were not commercially available. ML conversion (X_ML, %), GVL selectivity (S_GVL, %), and MeOH loss (L_MeOH, wt%) before and after the reaction were calculated by the following equations:

(1)

(2)

(3)

In this study, CuO, CaCO₃, and CuO_x-CaCO₃ catalysts were prepared by an oxalate gel-coprecipitation method, followed by calcination at 500 ℃ (Figs. 1 and S1). When the calcination temperature increased to 900 ℃, CaCO₃ was further degraded to afford CaO (Figure S1). As shown in Fig. 1, all the CuO_x-CaCO₃ catalysts displayed characteristic peaks of CuO (2θ= 35.5° and 38.7°) and CaCO₃ (2θ= 29.3°). The Cu/Ca molar ratio had a significant influence on the degree of ordering and crystallite size of CuO in these catalysts. As shown in Fig. 1 and Table 1, the intensity of diffraction peaks and crystallite size of CuO decreased with the increase of CaCO₃ content. Notably, all the catalysts (except for (4)CuO_x-CaCO₃) gave an apparent diffraction peak of Cu₂O at around 2θ= 42.1°, which became more visible with increasing CaCO₃ content (Fig. 1), indicating that the incorporation of CaCO₃ is conducive to the formation of Cu₂O even after the calcination of catalyst precursor at 500 ℃. Moreover, XPS analysis of (3/2)CuO_x-CaCO₃ further consolidated the above observation, as will be discussed later (Fig. 4).

The textural properties of catalysts prepared by an oxalate gel-coprecipitation method were measured by N₂ adsorption-desorption isotherms. As shown in Table 1, CuO had the highest specific surface area (10.3 m²/g) among all these catalysts, with a total pore volume of 120.3 × 10^-3 cm³/g and a mean pore radius of 44.6 nm (Table 1, entry 1). Both CaO and CaCO₃ had relatively lower specific surface areas, smaller pore volume, and mean pore radius compared with CuO (entries 2 and 7, Table 1). The specific surface area, pore volume, and mean pore radius of (n)CuO_x-CaCO₃ catalysts gradually increased with increasing CuO content. For example, (1/4)CuO_x-CaCO₃ had a specific surface area of only 2.6 m²/g with a total pore volume of 6.1 × 10^-3 cm/g, whereas (4)CuO_x-CaCO₃ had a specific surface area of only 7.0 m²/g with a total pore volume of 106.3 × 10^-3 cm³/g.

The reduction behavior of Cu-based catalysts and its counterparts was studied using H₂-TPR. All the CuO_x-CaCO₃ catalysts exhibited broad bands for H₂ consumption; CuO showed a single peak at about 220 ℃, which was attributed to one-step reduction of Cu²⁺ to Cu⁰ [39]. No obvious peaks were observed in the pure CaO, whereas a notable peak at around 740 ℃ appeared in the H₂-TPR profile of CaCO₃ (Fig. 2(f) and (g). It is known that Ca²⁺ species cannot be reduced in an H₂ atmosphere; thus, the peak at around 740 ℃ was attributed to CO₂ formed by the degradation of CaCO₃ at elevated temperatures. This was verified by TG analysis for the CaCO₃ and (3/2)CuO_x-CaCO₃ catalysts, as shown in Fig. S2. On the other hand, the reduction peak for Cu²⁺ species in CuO_x-CaCO₃ catalysts was slightly shifted to a lower-temperature region, probably owing to an interaction between Cu and Ca species [40]. Overlapping shoulders in (4)CuO_x-CaCO₃ were observed in the temperature range of 150–250 ℃, indicating the existence of two types of Cu species in the oxidation state. This finding was in line with the characterization of XRD (Fig. 1).

Acidic sites on the catalysts are conducive to the intramolecular cyclization of intermediates to form GVL during hydrogenation [41]. However, no peaks and an almost straight line were observed in the NH₃-TPD profile of CuO over the full range of test temperatures, indicating that there were no active acidic sites on the surface of CuO (Fig. 3(a)). Only a small peak representative of a moderate acidic site for pure CaO was observed. Notably, a broad peak in the range of 650–750 ℃ was observed in the CaCO₃. This peak was ascribed to the release of CO₂ resulting from the decomposition of CaCO₃ at high temperatures, according to the TPD test, which was conducted under the same conditions but without NH₃ during the adsorption process (Fig. S2(c)). TG analysis of CaCO₃ and (3/2)CuO_x-CaCO₃ gave the same results (Fig. S2(a) and (b)). Therefore, no acidic sites were detected on the surface of CuO_x-CaCO₃. Interestingly, the decomposition temperature of CaCO₃ was lower in the presence of Cu species (Fig. 3(b)–(f)), which could be regarded as evidence of an interaction between Cu species and CaCO₃.

As expected, Cu, Ca and O were detected in the survey scan of (3/2)CuO_x-CaCO₃ (Fig. 4(a)). Binding energy (BE) values centered at 934.2 and 954.0 eV were attributed to Cu 2p_3/2 and Cu 2p_1/2 of (3/2)CuO_x-CaCO₃, respectively (Fig. 4(b)). Each Cu 2p emission line could be fitted by using two Gaussian components, which were representative of Cu²⁺ and Cu⁰/Cu⁺ species, respectively. Compared with CuO (954.2 eV for Cu 2p_1/2), the Cu 2p_1/2 spectrum of (3/2)CuO_x-CaCO₃ showed a drift toward lower binding energies, confirming the interaction between Ca and Cu [42]. Specifically, partial electrons probably transferred from Ca to Cu species in CuO_x-CaCO₃, owing to the stronger electronegativity of Cu (1.9) compared with that of Ca (1.0) [43, 44]. A higher proportion of Cu⁰/Cu⁺ species was observed in the high-resolution spectrum of Cu 2p of (3/2)CuO_x-CaCO₃ (Fig. 4(b)), indicating that CaCO₃ has a crucial role in stabilizing Cu⁰/Cu⁺ species. There were known to be defects on the surface of metal oxides, which could result in local charge imbalance of the surface species; therefore, Cu⁺ species were detected on the surface of CuO by XPS. However, trace amounts of Cu⁺ species cannot be detected by XRD.

The SEM images showed that the CuO mainly consisted of secondary particles with diameter of around 50 nm, and the particles were relatively uniform (Fig. 5(a)). In the case of (3/2)CuO_x-CaCO₃, the distribution of the particle size was non-uniform, and particles with a diameter greater than 100 nm were clearly present (Fig. 5(b)). Similar phenomena were observed for other CuO_x-CaCO₃ catalysts (Fig. S3). To gain more insight into the elemental distribution on the surface of CuO_x-CaCO₃ catalysts, EDX elemental mapping for these catalysts was carried out. As shown in Figs. 6 and S4, both Cu and Ca, in all the CuO_x-CaCO₃, were evenly connected and dispersed with O, respectively, in all regions.

MeOH acts as both a solvent and an in-situ H-donor for the conversion of ML to GVL in all experiments. The reforming of MeOH to produce H₂ and the subsequent hydrocyclization of ML to GVL are both catalyzed by CuO_x-CaCO₃. A plausible reaction pathway can be elucidated as in Scheme 1, based on the identification of products via GC-MS. It can be seen that methyl 4-hydroxypentanoate (MHP) occurs as an intermediate, although it is undetectable by GC-MS, indicating that the hydrogenation of ML to MHP is the rate-limiting step, while the ring closing of MHP to GVL is a fast step under the conditions used in this study [45, 46]. As well as the main product GVL, small amounts of by-products are formed through the hydrocarboxylation/alkoxycarbonylation of GVL followed by hydrogenation (Scheme 1) [47].

In this study, both the gaseous and liquid products were collected for analysis. Gaseous products contained H₂, CO, and/or CO₂; no methyl ether was detected (Fig. S5). As shown in Table 2, the CuO could provide a ML conversion (X_ML) of 91.5% with a GVL selectivity (S_GVL) of 82.2% at 240 ℃ in 1 h (Table 2, entry 6). Some by-products (S_ROP of 7.3%, S_HCP of 6.3%) derived from the hydrocarboxylation and ring-opening of GVL were also detected. However, a low L_MeOH with negligible X_ML and Y_GVL was obtained over CaO or CaCO₃ (Table 2, entries 7 and 8), indicating the poor catalytic performance of CaO or CaCO₃ in H₂ production and ML hydrogenation. Interestingly, an X_ML of 96.8% with a S_GVL up to 88.9% was achieved over (3/2)CuO_x-CaCO₃ under the same reaction conditions (Table 2, entry 3); S_HCP decreased to only 3.2% and Y_GVL increased to 86.1% in this case. The above observation clearly indicates that (3/2)CuO_x-CaCO₃ is highly active for ML hydrogenation to GVL.

To determine the role of CaCO₃ in the catalytic performance of (3/2)CuO_x-CaCO₃, several blank experiments were carried out. When a physical mixture of CuO and CaCO₃ was used as the catalyst, X_ML was reduced to 71.7% while the hydrocarboxylation by-product increased dramatically to 16.2%, but selectivity towards ring-opening by-product declined compared with CuO (Table 2, entries 6 and 9). This indicates that a simple physical mixture of CaCO₃ with CuO not only harms the catalytic activity of CuO for ML conversion, but also favors the side reaction of GVL hydrocarboxylation. Therefore, there are probably strong interactions between Cu species and CaCO₃ in the (3/2)CuO_x-CaCO₃, which may be responsible for the outstanding catalytic performance. (3/2)CuO-CaO was obtained by the calcination of (3/2)CuO_x-CaCO₃ at 900 ℃ for 5 h; its catalytic performance was inferior to that of (3/2)CuO_x-CaCO₃, which confirmed the importance of CaCO₃ during the hydrogenation of ML to GVL with MeOH as the in-situ H-source.

In addition, X_ML and S_GVL gradually increased as the molar ratio of Cu to Ca increased (from 1/4 to 3/2 CuO_x-CaCO₃), and then decreased with a further increase of the molar ratio of Cu to Ca to 4 (Table 2, entries 1–4). Notably, the (3/2)CuO-CaCO₃ catalyst also outperformed other Cu-based bimetallic catalysts for ML hydrogenation in MeOH (Table S1).

Given the outstanding performance of (3/2)CuO-CaCO₃, the effects of reaction temperature and time on the hydrogenation of ML to GVL were investigated. When the reaction was conducted at 200 ℃ for 1 h, an X_ML of only 76.1% with an S_GVL of 92.3% was obtained (Table 3, entry 1). The low X_ML was mainly ascribed to the poor catalytic activity of the catalyst for ML hydrogenation at a relatively low temperature [48]. Moreover, only a small number of by-products were detected in this case. Notably, an increase in the reaction temperature to 240 ℃ brought about a significant improvement of X_ML to 96.8%, with a small decline in S_GVL compared with the case at 200 ℃ (Table 3, entries 1–3). Further increasing the temperature to 260 ℃ resulted in the gradual decrease of S_GVL to 84.7%, although the X_ML reached 99.1% (Table 3, entry 4). It appears that elevated temperatures greatly favor side reactions, indicating that these side reactions have higher activation energies than that of ML hydrogenation. An excess of gaseous products containing H₂, CO, and CO₂ formed by MeOH reforming could further promote the hydrocarboxylation of GVL (Scheme 1). Elevation of the temperature to 280 ℃ resulted in complete reduction of CuO and Cu₂O to Cu in (3/2)CuO-CaCO₃ (Fig. S6), and the side reaction was greatly promoted because of the inadequate Cu⁺ species, revealing that Cu⁺ species are conducive to suppressing the side reaction. When pre-reduced (3/2)CuO-CaCO₃ was used as the catalyst, an ML conversion of only 49.6% was obtained, along with a poor GVL selectivity (43.3%), at 220 ℃ in 1 h (Table 2, entry 10); this is specific evidence for the suppression of side reactions by Cu⁺.

The influence of reaction time was investigated at 240 ℃. When the reaction time was prolonged from 1 to 4 h, X_ML gradually increased from 96.8% to 100%, whereas S_GVL continuously increased to 96.1% in 3 h and then decreased 89.4% in 4 h (Table 3, entry 3 and entries 6–8). Further prolonging the reaction time to 4 h would result in more ring-opening of GVL at the cost of S_GVL.

In view of the excellent catalytic performance of (3/2)CuO_x-CaCO₃ in neat ML/methanol solution, it was highly desirable to test the catalytic activity of (3/2)CuO_x-CaCO₃ in the methanolysis product of carbohydrates. A detailed procedure is given in Section 2.4.2. The biggest challenge for the hydrogenation of ML to GVL in methanolysis products is the presence of humins, which are inevitable and complicated by-products that are formed during acid-catalyzed degradation of carbohydrates [49]. Humins can be easily adsorbed onto solid catalysts, leading to their deactivation. As shown in Table 4, an X_ML of 72.9% was obtained over (3/2)CuO_x-CaCO₃ at 240 ℃ in 1 h in the methanolysis product (Table 4, entry 1), while S_HCP sharply increased to 17.6% at the cost of S_GVL, compared with the case of neat ML/MeOH feedstock (Table 3, entry 3). The above observations clearly reveal that the presence of humins facilitates side reactions over (3/2)CuO_x-CaCO₃. With extension of the reaction time to 4 h, both X_ML and S_GVL gradually increased to 86.5% and 82.4%, respectively (Table 4, entries 2–4). In the methanolysis of MCC, the formation of humins is inevitable. We assumed that the adsorption of humins on the catalysts led to the indirect adsorption of ML, as the humins coated the catalysts with a loose structure and improved their overall specific surface area. This would have resulted in a reduced amount of ML in the solution. The adsorbed ML would not have been converted into GVL; thus, the X_ML values (Table 4) for the first several hours were inevitably higher than the real ones, while the opposite was true for S_GVL. A desorption experiment was conducted to verify the above assumption. After the reaction had proceeded for 1 h in methanolysis product, the catalyst was separated by centrifugation and washed by methanol with ultrasonic cleaning. About 7.1 wt% ML was washed out (based on the initial ML dosage). The overall selectivity (S_GVL+S_HCP+S_ROP) did not reach 100%, and the hydrocarboxylation and alkoxycarbonylation of GVL were reversible. Some by-products/intermediates were eventually converted into GVL when the reaction time was prolonged.

Notably, the initial black methanolysis product changed to a canary yellow and transparent liquid product after the reaction in the presence of (3/2)CuO_x-CaCO₃ (Fig. S7), suggesting that most humins in the methanolysis product were adsorbed onto the catalyst during the reaction, along with the ML. The carbon content of the used (3/2)CuO_x-CaCO₃ in methanolysis product was measured as 6.5 wt% by elemental analysis after the reaction. According to NH₃-TPD analysis, the adsorption of humins conferred a number of acid sites on the spent (3/2)CuO_x-CaCO₃ catalyst (Fig. S8), which were thought to be responsible for the improvement in the selectivity of by-products compared with that in the neat ML/MeOH group [47].

Leaching tests were performed to investigate the stability of the CuO_x-CaCO₃ catalyst. Hydrogenation of ML in both ML/MeOH solution and the methanolysis solution were conducted at 240 ℃ in 1 h, after which the reaction was stopped and the solid catalyst was separated from the reaction mixture by centrifugation. The resulting filtrate was reloaded into the reactor and then reacted at 240 ℃ for another 3 h. As shown in Figs. 7 and S9, only negligible fluctuation was observed, indicating that no soluble catalytic active species leached into the resulting filtrates. Element leaching (Cu and/or Ca species) was also verified by ICP-OES. The Cu and Ca contents in the reaction mixture were measured as 3.92 and 6.01 ppm for neat ML/MeOH solution, while there was almost no leaching of Cu and Ca species in the methanolysis solution (0.48 and 0.85 ppm, respectively) (Table S2). Therefore, the leaching of the catalyst was negligible during the reaction. The possible leaching of active metals during the reusability tests was also investigated; the results are provided in Table S2. The Cu and Ca contents of the filtrate did not increase during the recycling test, which clearly indicates that CuO_x-CaCO₃ is a robust catalyst.

A recyclability test was also conducted to evaluate the catalytic stability of (3/2)CuO_x-CaCO₃ in both the neat ML/MeOH solution and the methanolysis solution. In a typical reusability test, the used catalyst was separated by vacuum filtration and then directly reused in the next run. Unexpectedly, only negligible decreases in X_ML and S_GVL were observed with the recycling tests in both neat ML methanol solution and methanolysis product of cellulose (Tables 5 and S3); this was ascribed to the limited carbon deposition on the surface of the catalysts. This insignificant reduction in the catalytic ability of (3/2)CuO_x-CaCO₃ could be easily recovered by calcination of the used catalyst at 500 ℃ (Table 5, entries 8; Table S3, entry 7).

In view of the catalytic performance of (3/2)CuO_x-CaCO₃, the spent catalysts were further characterized for neat ML/MeOH solution and methanolysis solution. As shown in Fig. 8, the characteristic diffraction peaks for Cu⁺ species of the spent (3/2)CuO_x-CaCO₃ in both neat ML/MeOH solution and methanolysis solution were markedly intensified after the reaction compared with those of the fresh catalyst (Figure 1). Note that the majority of Cu species in the used (3/2)CuO_x-CaCO₃ existed in the form of Cu⁺/Cu⁰ instead of Cu²⁺ (Fig. 8(b)–(d)). However, the intensity of diffraction peaks for Cu⁺ or Cu⁰ species remained almost unchanged even after the eighth recycling test (Fig. 8(c)). On the contrary, CuO was completely reduced to metallic Cu after the first run under the same conditions (Fig. 8(h)). Interestingly, the Ca proportions in the catalysts greatly influenced the state of copper species after the reaction. Based on XRD patterns, there was no Cu phase for (1/4)CuO_x-CaCO₃ and (2/3)CuO_x-CaCO₃ (Fig. 8(e)–(f)); however, apparent peaks for Cu⁺ around 2θ= 36.2° were observed for all the catalysts, with relatively moderate Ca percentages (Fig. 8(b)–(d), g), especially in the case of (3/2)CuO_x-CaCO₃, whose Cu⁺ proportion was the highest among the three catalysts. The above observations reveal that the incorporation of moderate Ca by an oxalate gel-coprecipitation method effectively prevents the reduction of CuO to Cu in a reducing atmosphere, and greatly favors the stabilization of Cu⁺ species. Note that Cu²⁺ species were observed in the XPS spectra for the spent (3/2)CuO_x-CaCO₃ catalysts, probably owing to the difference in limit of detection between XPS and XRD analysis. This result also indicates that there was a small quantity of Cu²⁺ species dispersed on the catalyst surface, and that the particle size of the CuO in the spent catalysts was too small to be detected by XRD.

As shown in Fig. 9(b), the deconvolution of Cu 2p emission lines clearly indicates that both metallic and oxidized Cu crystallite species were present in the spent (3/2)CuO_x-CaCO₃, similar to the fresh one (Fig. 4(b)). The proportion of Cu⁰/Cu⁺ species in Cu 2p_3/2 drastically increased in the spent (3/2)CuO_x-CaCO₃ (Fig. 9(b)), indicating that more Cu²⁺ species were reduced to Cu⁰/Cu⁺ species after the reaction. Notably, no satellite peak for Cu²⁺ species of spent CuO was observed alongside the Cu 2p_3/2 and 2p_1/2 peaks (Fig. 9(d)). These observations agree well with the XRD analysis of the spent catalysts (Fig. 8), according to which the oxidation state species of Cu in the spent CuO catalyst were almost entirely reduced into metallic copper. This confirms the promotion of Ca species in stabilizing Cu⁺ species in (3/2)CuO_x-CaCO₃. On the other hand, BE signals of Ca 2p in the spent (3/2)CuO_x-CaCO₃ obviously shifted to higher values compared with the fresh (3/2)CuO_x-CaCO₃ or CaO (Fig. S10), suggesting an elusory interaction between Ca species and Cu species.

To further probe the deactivation of CuO_x-CaCO₃ catalysts, the carbon content of the spent catalysts was measured by elemental analysis. The C content of the spent (3/2)CuO_x-CaCO₃ was only 0.514 wt% after the first cycle, and only slightly increased to 1.119 wt% after the 7th cycle in neat ML/MeOH solution. SEM images showed that the spent (3/2)CuO_x-CaCO₃ after the first cycle in ML/MeOH solution was composed of irregular particles with a similar diameter to those in the fresh catalyst (Figs. 5(b) and S11(a)). Cu and Ca in the spent (3/2)CuO_x-CaCO₃ after the fourth and eighth cycles were evenly connected and dispersed with O in all regions, as before (Fig. S12(a)–(h)). Agglomeration of the spent catalyst was negligible with the recycling of the catalyst, which was probably a major contributor to the excellent catalytic performance of (3/2)CuO_x-CaCO₃ in ML/MeOH solution. However, severe carbon deposition of the catalyst occurred in the methanolysis product of cellulose. In these cases, the C content of the catalyst was 6.455 wt% after the first cycle, dramatically increasing to 38.14 wt% after the fifth cycle (Table S4). The deactivation of solid catalysts is strongly associated with the carbon deposition of the catalyst in organic solvents. Fortunately, the carbon deposits could be readily removed by calcination at 500 ℃.

Leaching and reusability tests of the CuO prepared by the same method were also conducted, as described in the supporting information (experimental section). The leaching of Cu species was much more severe for CuO than for the (3/2)CuO_x-CaCO₃ catalyst. According to the XRD pattern and XPS spectra for the spent CuO (Figs. 8(h) and 9(f)), the CuO was completely reduced to metallic Cu after the first run (Fig. 8(h)), with significant drops for the both the X_ML and Y_GVL compared with the first cycle (Table S5). This was in marked contrast to the (3/2)CuO_x-CaCO₃, whose Cu⁺ species were retained for at least eight continuous recycling runs (Table 5, Figs. 8 and 9). These results reaffirm the superior stability and reusability of the (3/2)CuO_x-CaCO₃ catalyst, due to the incorporation of CaCO₃.

The incorporation of CaCO₃ by an oxalate gel-coprecipitation method resulted in a robust dual-functional (3/2)CuO_x-CaCO₃ catalyst. This dual-functional catalyst could effectively catalyze both the in-situ production of H₂ by MeOH reforming and the simultaneous hydrogenation of ML to GVL. (3/2)CuO_x-CaCO₃ achieved GVL yield up to 95.6% at 240 ℃ in 3 h. Incorporation of CaCO₃ with CuO resulted in the formation of Cu⁺ species in the CuO_x-CaCO₃ catalyst, which greatly facilitated the hydrogenation of ML. In addition, (3/2)CuO_x-CaCO₃ could be reused for at least eight consecutive reaction cycles without significant loss of its catalytic activity. Notably, the presence of humins in methanolysis products has an inconspicuous influence on its catalytic performance. Therefore, it is possible to develop a facile two-step strategy for GVL production from cellulose through the integration of ML production, by methanolysis of cellulose, H₂ formation by MeOH reforming, and ML hydrogenation in an MeOH system.