Synthesis and Catalytic Activity of Cu-Co/CeO₂ Catalysts in the Hydrogenation of Furfural to Pentanediols

Abstract

This study presents a comprehensive characterization of monometallic (Co or Cu) and bimetallic (Co-Cu) catalysts supported on cerium oxide (CeO₂). XRD and TEM analyses revealed that crystallinity decreases after reduction and that metal dispersion is highly dependent on composition, with cobalt exhibiting greater dispersion than copper. The results confirmed a strong interaction between the metals and CeO₂, which alters the ceria structure and facilitates the reduction of the metal oxides. H₂-TPR and XPS data indicated that monometallic and the bimetallic 15Cu15Co catalysts achieved nearly complete reduction, whereas other bimetallic catalysts did not. Furthermore, CO chemisorption and H₂-TPD demonstrated that the hydrogen activation capacity correlates with the degree of catalyst reduction. Notably, bimetallic catalysts did not show enhanced hydrogen activation compared to their monometallic counterparts. This suggests that the dispersion and metal–support interaction are more critical factors for catalytic activity in this system than the formation of metal alloys. Although the furfural conversion was complete, the selectivity depended greatly on the catalyst composition. The 30Co_R catalyst was most selective for 1,5-pentanediol (38.4%), the 30Cu_R catalyst for 1,2-pentanediol (22.1%), and the bimetallic catalysts for THFA. Reutilising the 30Co_R catalyst after five catalytic cycles resulted in a gradual reduction in the selectivity of 1,5-pentanediol.

1. Introduction

The current production of pentanediols (PDO) is limited by the scarce availability of C5 compounds in petrochemicals. In contrast, other α,ω diols, such as 1,4-butanediol or 1,6-hexanediol, have a market of megatons per year [1], with a global market value of USD 7 billion. The global market value of 1,5-pentanediol (1,5-PDO) was USD 37 million in 2023, with a growth to USD 47 million in 2029 [2]; 1,2-pentanediol (1,2-PDO) will grow from USD 103 million in 2023 to USD 150 million in 2032 [3]. The interest in these PDOs lies in their multiple applications. For example, 1,2-PDO is used in various sectors such as cosmetics, agriculture, textiles, and pharmaceuticals due to its properties as a solvent and a wetting and antimicrobial agent [4,5,6]. 1,5-PDO, like other α,ω diols, is used in the synthesis of polyurethanes and polyesters [7]. It also finds applications in printing ink formulations [8,9] and plasticisers [10], among others. The industrial synthesis of 1,2-PDO involves a multi-step process from the selective oxidation of pentene to epoxypentene and subsequent hydrolysis to afford 1,2-PDO [11], while the synthesis of 1,5-PDO is based on the hydrogenation of glutaric acid, which is a by-product of the adipic acid production industry [9].

Furfural (FF) produced from biomass is a compound with five carbon atoms plus two oxygens so that through hydrogenation and hydrogenolysis reactions, it can be effectively converted into PDOs. In order to synthesise 1,5-PDO, it is crucial to select the appropriate catalyst and initial compound. The 1,5-PDO can be prepared in several ways. Firstly, it can be prepared directly from FF. Alternatively, it can be prepared from one of the intermediate hydrogenation products, such as furfuryl alcohol (FFA). Finally, it can be prepared by opening the furanic ring of tetrahydrofurfuryl alcohol (THFA). Depending on the starting compound, different catalytic systems have been proposed. When the initial reactant is THFA, noble metal-based systems with an oxyphilic promoter have been found to be the most promising. The noble metals that were examined included Rh, Ir, and Pt as well as Re, W, and Mo as oxyphilic metals. Tomishige et al. [12,13,14] conducted extensive research on these catalysts, observing a high selectivity towards 1,5-PDO, even utilising high THFA concentrations of up to 20 wt% [12,14]. In these studies, a consistent finding was the crucial synergistic effect between noble metal sites and oxophilic promoters in fostering THFA hydrogenolysis [13,15]. In addition to these noble metal-based catalysts, catalysts based on Ni have been tested [10,16,17,18,19]. Thus, Ni-based catalysts have shown high efficiency in the conversion of THFA alcohol to 1,5-PDO, particularly when used in combination with basic lanthanum hydroxide, achieving a yield of 96.9% and a selectivity of 94.6% for 1,5-PDO at 170 °C and 40 bar H₂ for 24 h [10]. Zhang et al. [17] evaluated Ni-La₂O₃-based catalysts prepared from perovskite LaNiO₃ reduced at different temperatures. The structural analysis of these catalysts showed that the Ni-La₂O₃ interface was enriched in Ni²⁺, La³⁺, and oxygen vacancies, i.e., O_v, relating the presence of these sites to the hydrogenolysis of THFA. According to the authors, the role of Ni in the mechanism is to activate the hydrogen that by spillover generates La-H and La-OH sites in the interphase region, where the La-OH sites activate the -OH of the THFA, forming the corresponding alkoxide, while the neighbouring hydride breaks the C2-O bond, generating the 1,5-PDO. With these catalysts, the maximum conversion of THFA was 55.8% with a selectivity to 1,5-PDO of 91.7% at 180 °C, 30 bar H₂, and 8 h reaction time. Peng et al. [16] studied Ni supported on Y₂O₃. The catalyst with a Ni/Y molar ratio of 5 produced a THFA conversion of 84.2% and a selectivity to 1,5-PDO of 99.5% after 24 h of reaction at 150 °C and 30 bar H₂. The support provides the basic sites that activate the THFA, and H₂ is activated in the interphase region between Ni and Y₂O₃, attacking the C-O bond of the THFA ring and generating the 1,5-PDO. Compared to noble catalysts, Ni-based catalysts require higher temperature conditions but lower H₂ pressures and the use of an organic solvent, mainly isopropanol. One of the advantages of these non-noble catalysts is the fact that they are highly stable, with reaction time up to 400 h [10].

One of the advantages of obtaining PDOs from THFA is the high selectivity showed by the catalysts. However, it should be noted that, depending on the catalyst, either high H₂ pressures or high temperatures are required to achieve high yields. In such conditions, it has been demonstrated that a higher yield is achieved towards 1,5-PDO than 1,2-PDO. However, when starting from FFA, it is more challenging to achieve selective control of the production of one of the PDOs [20,21,22,23,24,25,26,27], and consequently, other hydrogenation products are formed, such as THFA [20,21,22,23,24,25,26,27]. In these cases, the catalytic systems are not able to hydrogenate THFA, becoming the final product of the reaction. Dai et al. [28,29] reported that a copper-based catalyst impregnated on MFI zeolite obtained a yield to 1,5-PDO close to 75%, suppressing the formation of THFA, at the temperature of 160 °C and 25 bar H₂. Zhu et al. [30] studied Cu nanoparticles supported on MgO-La₂O₃, reaching PDOs yields (1,2 and 1,5-PDO) close to 80%, with 60% corresponding to 1,2-PDO, at 140 °C and 60 bar H₂. Liu et al. [31] also obtained similar results when Cu nanoparticles were supported on MgAlO_x. When CuMg [32] was supported on mesoporous silica, a higher yield was also observed for 1,2-PDO than for 1,5-PDO, with 2-MF being the main by-product of the reaction. The effect of Mg was studied by Shao et al. [33], who observed that it affected the dispersion of Cu, facilitated its reduction, and prevented sintering, thus increasing its recyclability. When Co acts as the active phase [34,35], it has been observed that the pattern of selectivities to the PDOs changes, with the selectivity then being higher towards 1,5-PDO. Thus, Li et al. [35] prepared a CoAlO_x mixed oxide by calcining of the corresponding hydrotalcite, achieving a 1,5-PDO yield of 55% and a 1,2-PDO yield of 7.5%, at 160 °C and 30 bar of H₂. Lee et al. [34] designed a Co-based catalyst supported on TiO₂, establishing a strong interaction between the metal and the support. This stabilised the Co on TiO₂, preventing leaching and achieving catalyst stability for up to 105 h in a continuous reactor. Under these conditions, the yield at 1.5-PDO remained above 30% at a temperature of 140 °C and 23.4 bar H₂. Catalysts based on Ni as the active phase have also been studied; for example, Nimbalkar et al. [36] in a study with a catalyst based on a Ni-Sn alloy achieved a 1,2-PDO yield of 80% at 200 °C and 40 bar H₂, while Wijaya et al. [20] achieved a maximum 1,5-PDO yield of 47.8% and a 1,2-PDO yield below 2% by modulating the Ni/Y molar ratio. The authors observed that if the catalyst contained excess Ni, the reaction would evolve towards THFA due to the ability of Ni to hydrogenate the furanic ring. Starting from FAL, noble metal-based catalysts have not been evaluated due to the high hydrogenating capacity of these metals leading to over hydrogenation products [4,27,30,37,38], and on the other hand, they are more selective to 1,2-PDO than to 1,5-PDO [4,27,37,39,40,41].

Finally, one of the routes for the synthesis of pentanediols would be to start from furfural itself. The hydrogenation and hydrogenolysis of the FF to one of the PDOs requires the formation of the FAL intermediate. Therefore, catalysts capable of evolving the FAL to PDO could also carry out the reaction directly from FF. Catalysts based on noble metals, such as Pt [39,42], Rh-Ir alloy [43], Rh [44], Ir-ReO_x-based catalysts doped with Pd [45], and Ru catalysts [40] have been tested in this reaction. The best results for the non-noble metals analysed are obtained when bimetallic or trimetallic catalysts are prepared versus monometallic ones since the synergy between the different metal phases, the presence of oxygen vacancies, and the modulation of the acid–base properties cause the PDO yields to increase. The main catalytic systems tested included those based on cobalt [7,46,47] and modified with Cu [48,49], with Ni [8], or with Cu and La [49]. As expected, with both noble metal and Co-based catalysts, selectivity control is difficult, and high yields towards one of the PDOs are not achieved. Kurniawan et al. [8] evaluated Co-Ni-Al based catalysts and observed that metal sites such as Ni⁰ and Co⁰ facilitate the dissociative adsorption of H₂, while the presence of oxygen vacancies in the metal oxide phases, such as CoO_x, enhances H₂ adsorption. The Co^δ+ sites in the O_v-CoO_x centres are the sites where FAL is adsorbed through the alcoholic group, where the subsequent cleavage of the C-O bond takes place, giving rise to 1,5-PDO. When the Ni⁰ particles were sufficiently small, they prevented parallel adsorption of FF and thus over hydrogenation of the ring, yielding THFA. By modulating the composition of the Co-Ni-Al catalyst, a yield of 47.5% was achieved at 1,5-PDO at 160 °C, at 3 h of reaction and 30 bar H₂. Wang et al. [48] synthesised catalysts based on Cu-Co supported on CeO₂, in which the metal particles acted as active sites for the activation of H₂, which by spill-over was transferred to the oxygen vacancies. Both the oxide phases and the basic sites were responsible for the tilted adsorption of FF, favouring the hydrogenation of FF to FAL and avoiding the over hydrogenation to THFA. The synergy between the Lewis sites and the oxygen vacancies was crucial in weakening the C-O bond of the alcoholic group, thereby facilitating the cleavage of the C-O bond and the subsequent release of 1,5-PDO. The role of Cu was double: it favoured a lower interaction of Co with CeO₂, promoting the reduction in CoO_x, as well as fostering the creation of more oxygen vacancies by promoting the reduction in CeO₂. Thus, the 5Cu-30Co/CeO₂ catalyst at 150 °C and 30 bar H₂ after 4 h produced 53.4% of 1,5-PDO with 12.8% of 1,2-PDO. Wang et al. [47] tested the activity of Fe-doped Co/CeO₂ with a CoO@Co core–shell structure, in which the Co⁰ was encapsulated by the Co²⁺. Co²⁺ was the centre for breaking the C2-O bond and Co⁰ for hydrogenating the double bond. The role of Fe was to maintain an optimal Co²⁺/Co⁰ ratio to increase substrate adsorption and inhibit saturation of the furanic ring, thus achieving a 50.2% yield at 1,5-PDO under the best reaction conditions. Barranca et al. [7] also tested Co-based catalysts, but in this case, Co was incorporated into the structure of a hydrotalcite, partially replacing the Mg²⁺ cation. The mixed oxides generated after calcination of the parent hydrotalcite were impregnated with Pt (2 wt%). The basic sites facilitated FF adsorption and subsequent hydrogenolysis. In this case, comparing the catalyst activity starting from FF or FAL, it was observed that in both cases, the production of PDOs (1,5 and 1,2-PDO) was 70% since the reaction from FF to FAL was faster than the rest of the reactions. For the direct hydrogenation of FF to PDOs, Liang et al. [49] designed a copper-, cobalt-, and lanthanum-based catalyst prepared by coprecipitation. This trimetallic catalyst exhibited high efficiency in the hydrogenation of FF to PDOs, achieving yields of up to 21.7% for 1,2-PeD and 50.1% for 1,5-PDO under optimum conditions (160 °C, 40 bar, 9 h). The synergistic interaction between Cu and Co, together with an appropriate number of basic sites provided by lanthanum, contributed to the high selectivity towards furan ring opening and the formation of PDOs. The addition of lanthanum to the catalyst increased its surface area in addition to providing numerous basic sites that favoured the adsorption of the C-O group of the furan ring, while Cu favours the presence of CoO_x species, which are the sites where the C-O bond breaking of the furan ring occurs and thus the formation of PDOs.

Tian et al. [46] introduced Co²⁺ into the ZIF-8 structure by reduction with NaBH₄ at room temperature and in aqueous medium, obtaining ZnCo-LDH nanosheets. By FTIR and DFT studies, the authors showed that a tilted FF adsorption was favoured, as this form of adsorption was the most energetically preferred and contributed to a higher furan ring opening. In addition, they showed that the Co sites were responsible for the adsorption and activation of H₂. With these catalysts, the authors reported a 46.9% yield at 1,5-PDO at 160 °C and 40 bar H₂ after 4 h of reaction.

Considering that the FF hydrogenation reaction requires the presence of metal sites for H₂ activation as well as partially oxidised metal species in conjunction with oxygen vacancies as FF/FAL activation sites, a family of Cu-Co-Ce-based catalysts was prepared, where both Cu and Co provide both the metallic and partially oxidised species after reduction in hydrogen, and Ce provides the oxygen vacancies due to the presence of the Ce⁴⁺/Ce³⁺ pair. A combined mass content of Cu and Co of 30 wt% was chosen, varying its concentration from 0 wt% to 22.5 wt%. The catalytic activity of these catalysts was evaluated in the liquid-phase hydrogenation of FF to obtain pentanediols. It was observed that, depending on the composition of the catalysts, the selectivity varies for the different reaction products, with the monometallic catalysts being more selective towards pentanediols, while the bimetallic catalysts are more selective towards THFA.

2. Results and Discussion

2.1. Catalysts Characterization

In the following discussion of the results, the solids are called monometallic when only Co or Cu are present and bimetallic when both metals are present in the composition of the solids.

As shown in Figure 1, the crystalline phases present in the precipitated hydroxides and oxides as well as in the reduced catalysts are dominated by the fluorite structure of CeO₂ (PDF: 04-003-5454). Increasing the weight percentage of Cu or Co resulted in the observation of diffraction peaks corresponding to the nitrate hydroxide of copper (II), i.e., Cu₂(NO₃)(OH)₃ (PDF: 04-009-2766), and cobalt (II) hydroxide, i.e., Co(OH)₂ (PDF: 04-015-4721). Other minor phases detected were related to cobalt (III) oxyhydroxide, CoOOH, and copper (I) oxide. Moreover, Figure 1a shows that such detected phases are quite amorphous. The calcination of the hydroxides at 450 °C resulted in enhanced crystallinity of the CeO₂, accompanied by the identification of new oxide phases (Figure 1b). This transformation occurred due to the decomposition of the cobalt and copper hydroxides, leading to the formation of the respective oxides, i.e., cobalt spinel (PDF: 04-002-2531) and copper (II) oxide (PDF: 04-008-8216), respectively. It was observed that the presence of copper and cobalt oxides was clearly evident in monometallic solids. However, bimetallic solids exhibited no discernible signal attributable to these oxides; only 15Cu15Co_O showed minor peaks of Co and Cu oxides, showing the high dispersion of the oxides phases after calcination. Moreover, the diffraction peaks attributable to the CeO₂ phase were slightly shifted to high 2θ compared to the CeO₂ (Figure 1b), and this fact would indicate that the unit cell of CeO₂ was modified in some manner, likely by the incorporation of Co or Cu in the structure of CeO₂. In a similar manner, the reduction of the oxides resulted in the absence of diffraction peaks for both metals, Cu⁰ and Co⁰, for the bimetallic catalysts (Figure 1c). On the other hand, low-intensity diffraction peaks were detected for metallic copper and cobalt in the 30Cu_R and 30Co_R catalysts, respectively. On the other hand, the 7Cu22Co_R solid exhibited peaks indicative of the presence of Co₃O₄, thereby indicating that the complete reduction of the spinel had not been achieved. The decrease in the crystallinity of the catalysts after reduction at 400 °C is noteworthy. Moreover, a displacement in the CeO₂ diffraction peaks to lower 2θ (Figures S1–S5) after reduction at 400 °C is discernible. These changes in the positions of the diffraction peaks could be due to a partial reduction of the Ce⁴⁺ to Ce³⁺ (see below). The Ce³⁺ cation has an ionic radius larger than Ce⁴⁺, provoking an expansion of the unit cell and the concomitant shift of the diffraction peaks to lower 2θ.

The morphology of the particles (Figure S6) was analysed using transmission electron microscopy (TEM), and the distribution of the elements within the particles was analysed using energy-dispersive spectroscopy (EDS) mapping (see Figure 2). Firstly, the images (Figure S6) provide confirmation of the XRD analysis in the sense that the particle size was reduced when the oxides were reduced, and therefore, these solids showed lower crystallinity. Conversely, the presence of particles exhibiting both large polyhedral shapes and smaller, more rounded shapes can be observed irrespective of whether the oxides or reduced solids are analysed. It was not feasible to measure the spacing between the different crystallographic planes due to the scale of the images. This precludes the determination of the crystalline phase to which each particle belongs; however, based on the XRD analyses, it can be hypothesised that the observed polyhedral particles are likely CeO₂ since the most intense diffraction peaks correspond to the fluorite phase of CeO₂. In contrast, the smaller, rounder particles are likely to be copper or cobalt oxides. Upon reduction, these particles would become metallic Cu or Co particles.

EDS analysis of the bimetallic precursors, xCuyCo_O, demonstrates that the interaction between Co and Ce is greater than that between Ce and Cu, with Cu exhibiting a less homogeneous distribution than Ce and Co. As demonstrated by the element mapping images for the xCuyCo_O solids, there appears to be a greater propensity for Co to interact with Ce as the amount of Cu diminishes, with the 15Cu15Co_O catalyst exhibiting the most uniform distribution of the three elements. It is noteworthy that in the case of the solid 22Cu7Co_O, both Cu and Co are distributed in the same regions of the particles, while Ce is found in different regions. This observation indicates that Cu tends to form larger particles compared to Co, which appears to be more dispersed within the solid matrix. Consequently, the ratio between the metals has a substantial impact on their dispersion in the particles. A similar phenomenon is observed in the case of the solids 30Cu_O and 30Co_O, which exhibit agglomeration of CuO, forming larger particles compared to the dispersion observed in cobalt oxides. After the reduction, this behaviour is sustained, with Co exhibiting greater dispersion than Cu, which forms more substantial particle agglomerates.

The textural parameters are presented in

Table 1. Textural parameters and acidity of the solids xCuyCo_O.

Sample	SBET (m2/g)	Vp (cm3/g)	dav (nm)	Acidity 1 (mmol NH3/g)
CeO2_O	91.0	0.219	10.0	203
30Cu_O	113	0.222	8.10	30.3 (72)
30Co_O	104	0.137	5.50	52.7 (49)
7Cu22Co_O	145	0.162	4.50	9.17 (23.5)
15Cu15Co_O	121	0.206	7.40	20.6 (39.4)
22Cu7Co_O	133	0.150	4.40	10.7 (26.4)

¹ In brackets, the acidity of xCuyCo_R catalysts.

along with the acidity of the xCuyCo_O solids, and the N₂ isotherms and pore size distribution are shown in Figures S7a and S7b, respectively. All of them showed isotherm Type IV according to the IUPAC classification and a hysteresis cycle H3. The size and distribution of the pores corresponded to the mesoporous region, and the BET surface was very similar between them irrespective of Cu or Co loading. On the other hand, the acidity of these solids was very low, increasing after the reduction, and was lower than the CeO₂_O.

The reduction of the oxides was evaluated by means of H₂-TPR (Figure 3). The H₂-TPR profiles show a complex pattern enveloping several reduction peaks (

Table 2. Percentage of reduction of the catalysts determined from H₂-TPR, CO chemisorption results, and the percentage of Ce³⁺ before and after the reduction at 400 °C measured from XPS analysis.

	% of Reduction 1	Irreversible mmol CO/g 2	Dispersion (%) 2	Crystal Size (nm) 2	Ce3+ (%) 3
					O	R
30Cu_R *	107.2	55	1.2	74	2.7	16.1
30Co_R	92.5	164.7	3.2	26	4.0	18.7
7Cu22Co_R	42.5	104.7	2.1	40	7.7	9.8
15Cu15Co_R	93.0	157.4	3.2	27	5.2	18.0
22Cu7Co_R	40.1	161.6	3.4	26	5.6	13.7

* Reduced at 250 °C; ¹ determined from H₂-TPR; ² data obtained from CO chemisorption; ³ superficial Ce³⁺ atomic concentration measured from XPS analysis.

, Figures S8–S12) depending on the composition of the solids. From Table 1, it can be concluded that the monometallic oxides would be practically reduced, and the contribution of the reduction of CeO₂ to the total H₂ consumption would not be important.

The 30Co_O exhibited a profile of H₂ consumption consisting of two peaks at 307 and 372 °C as well as a broad band extending up to 550 °C. The occurrence of these peaks is attributed to the reduction of Co³⁺ to Co²⁺ (307 °C), followed by the subsequent reduction of Co²⁺ to Co⁰ at 372 °C [8,35,47]. Those temperatures of reduction are lower than those reported for the Co₃O₄ [50]. This suggests that there is an interaction between the cobalt oxides and the ceria phase, favouring the reduction of the cobalt spinel. There is also a broad band extending from 400 to 525 °C, which was ascribed to CoO_x species strongly interacting with the CeO₂ phase or even to the partial reduction of Ce⁴⁺ to Ce³⁺ [47]. The reduction of 30Cu_O took place at lower temperatures than did the 30Co_O solid. The CuO was completely reduced under temperatures as low as 250 °C. The H₂-TPR profile shows a peak centred at 194 °C, with shoulders both at lower and higher temperatures, and this reduction pattern points out that the copper oxides interact with the ceria phase with different degrees of intensity or show different particle sizes, as TEM images showed. For the bimetallic solids, the reduction of both copper and cobalt oxides was carried out at lower temperatures than for the monometallic cobalt solid. Nevertheless, the complete reduction of the oxides was not achieved since the actual H₂ consumption was lower than the theoretical one (Table 2). This is in accordance with the results of the XRD analysis, which revealed detectable peaks associated with Co₃O₄ for the 7Cu22Co_R catalyst. The 22Cu7Co_O solid showed an H₂ consumption profile with two reduction peaks at approximately the same temperature as 30Cu_O. When the Cu loading was decreased, the 7Cu22Co_O solid showed a profile very similar to that of 30Co_O, but the reduction peaks occurred at lower temperatures. Therefore, all those reduction peaks for the bimetallic catalysts include the reduction of both metal oxides, and a rough assignation of the reduction peaks could be that the reduction peaks at lower temperatures were mainly due to the copper oxides and at higher temperatures to the cobalt oxides. It is noticeable that the 15Cu15Co_R catalyst almost showed the complete reduction of the oxides, and this could be due to the close interaction between the three metals (see TEM images) that favour the reduction of the oxides phases. So, it can be concluded that for the bimetallic catalysts, the interaction between the three oxides is crucial to achieve a complete reduction of the oxides.

The surface composition of both mixed and reduced oxides was analysed by means of XPS technique. The atomic superficial concentrations of both kinds of solids are collected in Tables S1 and S2. The reduced solids showed a higher atomic concentration of Cu and Co than their oxide counterparts. The oxygen content was also decreased after reduction, and the Ce amount did not show a clear correlation with the Co and Cu metallic load. Analysis of the monometallic oxides showed a similar superficial composition, but this dramatically changed when both Co and Cu were jointly precipitated. Irrespective of the weight percentage of each metal, the surface was enriched in copper, and the surface metal content was lower than the monometallic ones. This could be due to the fact that the differences between the product of solubility of the corresponding hydroxides (K_ps (Ce(OH)₄) = 10⁻⁵⁴; K_ps (Co(OH)₂) = 5.92 × 10⁻¹⁵; K_ps (Cu(OH)₂) = 2.2 × 10⁻²⁰), considering that all the cationic metals precipitated as pure hydroxides, and as the K_ps of Ce(OH)₄ is the lowest, this hydroxide would precipitate at the lowest pH, then Cu(OH)₂ would start to precipitate, and finally Co(OH)₂. This sequence of hydroxide precipitation could potentially result in an enrichment of cobalt species on the surface, which was not experimentally observed. However, as demonstrated in the study on the precipitation of layered double-hydroxides (LDH) compounds [51], the pH at which individual hydroxides precipitate may not necessarily correspond with the pH of the precipitation process when hydroxides precipitate concurrently. Moreover, some of the octahedrally coordinated Cu²⁺ ions could be exchanged for Co²⁺ ions via the Diadochy reaction [52] since the ionic radii of both cations are very similar. This resulted in the substitution of Cu²⁺ ions for Co²⁺ ions in the laminar structure of the hydroxide nitrate of Cu²⁺ during its formation, thereby decreasing the surface concentration of Co²⁺. After reduction, it was observed that both monometallic and bimetallic catalysts maintained the superficial composition of the metals, with an increase in the copper and cobalt on the surface, improving the amount of both metals on the surface. Therefore, after the reduction, the metals ameliorated the dispersion on the surface of the catalysts, increasing the available metal centres for the reaction.

On the other hand, the photoelectron spectra of each element were analysed to ascertain their chemical state as well as the extent of the reduction of the oxides. In the case of the analysis of the Cu 2p and Ce 3d region, the X-ray exposure time was reduced with the aim to avoid photoreduction of Cu²⁺ [53] and Ce⁴⁺ [54]. As outlined in the relevant literature [54,55,56,57], the analysis of the photoelectron spectrum of the Ce 3d level is complicated by the differences in the occupancy of the 4f level in the final states and the coexistence of Ce³⁺ and Ce⁴⁺. This results in six peaks for Ce⁴⁺ assigned to the spin–orbit doublets corresponding to Ce 3d_5/2 and Ce 3d_3/2 and four for Ce³⁺, giving rise to ten peaks for the deconvolution of the Ce 3d level. In this analysis, the nomenclature found in the literature [54] was also followed. In this instance, due to the intricacy of the analysis, the peaks corresponding to v₀ and u₀ were excluded in the deconvolution process, as their ultimate contribution to the fit was negligible, and their energies were closely aligned with the v and u peaks when the Ce⁴⁺ reduction was minimal [56]. Moreover, the relative intensities of the peaks corresponding to the 3d_5/2 and 3d_3/2 doublets were adjusted to a ratio of 3:2. The deconvolution results for the Ce 3d level of the oxides as reduced catalysts are given in Tables S3 and S4, respectively. Although interpreting the deconvolution can be difficult, it is possible to detect the presence of Ce⁴⁺ and Ce³⁺ and observe the effect of reduction by tracking the variation in the intensities of the peaks labelled u’’ and v’, which are characteristic of Ce⁴⁺ and Ce³⁺, respectively (Figure S13 for an example). As can be seen, cerium is mainly found as Ce⁴⁺ in both the non-reduced and reduced catalysts. It was also observed that the contribution of Ce³⁺ increased after reduction. These results show that Ce³⁺ is already present in the oxides and that this contribution increases for all the catalysts after reduction at 400 °C. Table 2 shows the percentage of Ce³⁺ [57] present in each catalyst before and after reduction, demonstrating this increase. Although H₂-TPR analysis ruled out significant Ce⁴⁺ reduction, it appears that the catalyst surface is partially reduced, generating oxygen vacancies to counteract the diminution of positive charge. Furthermore, catalysts with a higher percentage of reduction, as determined by H₂-TPR, exhibit a higher surface Ce³⁺ content.

The photoelectronic Cu 2p core level of the oxides shows the typical shake-up satellite corresponding to Cu²⁺ (Figure S14a), whereas this shake-up is negligible for the reduced solids (Figure S14b), indicating that the Cu²⁺ has been reduced to a lower oxidation state. Moreover, the binding energy of Cu2p_3/2 peak (Table S5) was shifted to lower energies after the reduction treatment, corroborating that the reduction of Cu²⁺ to lower oxidation states was achieved. From the energy of the Cu 2p_3/2 of the reduced catalyst, it is difficult to assign this peak to Cu⁺ or Cu⁰. For this reason, the CuLMN band was recorded (Figure S15) for the reduced catalysts. The results of the deconvolution of the CuLMN signals are shown in Table S5. The CuLMN is the sum of three contributions, roughly centred at 919, 917, and 915 eV, respectively. Each peak corresponds to an oxidation state of Cu. Thus, the assignation of such peaks corresponds to Cu⁰, Cu⁺, and Cu²⁺, respectively [52,58]. Interestingly, the 30Cu_R and 22Cu7Co_R catalysts show the highest amount of Cu²⁺, and the catalyst with the highest cobalt loading shows the lower amount of Cu²⁺; therefore, it seems that interaction between copper and cobalt favoured the reduction of copper to lower oxidation states. Moreover, the XPS analysis corroborated the presence of Cu⁰ necessary for the H₂ activation.

Analysis of the Co 2p level is characterized by the presence of the spin–orbit doublets 2p_3/2 and 2p_1/2 as well as shake-up satellites. While CoO exhibits an intense satellite at approximately 5–6 eV from the 2p_3/2 photoelectron peak, Co₃O₄ exhibits a weak satellite at 10–11 eV, and metallic Co exhibits no satellites [59]. As Figure S16a,b show, the shapes of the spectra for the oxides and reduced catalysts are completely different. The spectra of the oxides are dominated by Co₃O₄, whereas the shape of the spectra of the reduced catalysts indicates that CoO is the dominant phase on the catalyst surface. The presence of metallic Co cannot be ruled out, as the 2p_3/2 peak shows a tail towards lower binding energies, which could indicate the existence of small metallic Co clusters.

The O 1s level of the oxides (Figure S17a) and reduced catalysts (Figure S17b) shows two contributions: one around 529 eV and the other around 531 eV (Table S6). The contribution of the lower-energy peak is higher than that of the higher-energy peak. The lower-energy peak would correspond to lattice oxygen [55,60], while the other peak could be due to oxygen associated with hydroxyl groups [26,61,62].

The adsorbed amount of CO was determined by the difference between two CO adsorption isotherms: the first one represents the strongly adsorbed CO and the physiosorbed CO, which after evacuation of the weakly adsorbed CO; the second isotherm was obtained, and by the difference between the two isotherms extrapolated to zero P, the irreversible adsorption of CO was determined. For the analysis of the metal dispersion, a CO:M (Cu and/or Co) ratio of 1:1 and a cubic shape of the particles, which is more appropriate than the shape observed in the TEM images, are considered. The amount of irreversible CO chemisorbed (Table 2) shows that 30Cu_R adsorbed the lowest amount of CO among all the catalysts, while 30Co_R is the catalyst that chemisorbs the highest amount of CO. However, for the xCuyCo_R catalysts, the catalyst with the lowest Cu content showed the lowest CO chemisorption. These results indicate that the metal dispersion follows the order 30Co_R ≈ 22Cu7Co_R > 15Cu15Co_R > 7Cu22Co_R > 30Cu_R. The analysis reveals that the particle sizes are larger than those detected by XRD, for which only Cu⁰ particles were detectable in the case of 30Cu_R, Co₃O₄ in the case of the 7Cu22Co_R catalyst, and Co⁰ for 30Co_R. It would be expected that the 22Cu7Co_R and 15Cu15Co_R catalysts would show the lowest capacity for CO chemisorption due to their high Cu concentration. However, the CO chemisorption data for the 30Co_R and 30Cu_R catalysts suggests that metallic Co would be primarily responsible for CO chemisorption in bimetallic catalysts. Therefore, those catalysts with greater Co dispersion could chemisorb greater amounts of CO. The EDS figures show that the 7Cu22Co_R catalyst had the lowest Co dispersion compared to the other two bimetallic catalysts and therefore the lowest capacity to chemisorb CO. The other three catalysts containing Co have similar Co dispersions and therefore similar CO chemisorption data. Furthermore, it should be noted that the 7Cu22Co_R catalyst had a Co spinel phase, which could contribute to lower CO chemisorption. However, it is important to note that Pope et al. [63] determined the dispersion of Co₃O₄ by CO adsorption at the temperature of −78 °C. The disparities observed between TEM images, XRD analysis, and data obtained from CO chemisorption may be attributable to the varying stoichiometries of CO adsorption. It is important to note that these stoichiometries can vary, for example, from linear to bridging (1 CO molecule per two atoms) or subcarbonyl formation (two or more CO molecules per atom), when adsorbed on cobalt [64]. Furthermore, as demonstrated in Table 2, a percentage of Ce³⁺ is present, generating the oxygen vacancy that can also chemisorb CO [65].

The adsorption of CO has also been studied by DRIFT spectroscopy to obtain information about the oxidation state of metals. The process of adsorption was conducted at a low temperature of −150 °C to prevent a reaction between the CO and the catalyst surface, which could cause an electronic effect on the cationic centres. This temperature allows the weak CO adsorption on metal cations to be more clearly observable. Ceria exhibits an intense band (Figure 4) at 2156 cm⁻¹ with two shoulders: one at a higher wave number (2163 cm⁻¹) and one at a lower wave number (2146 cm⁻¹). The other catalysts show this band but with significantly reduced intensity, except for catalyst 30Cu_R. The band at 2156 cm⁻¹ was associated with the interaction of CO with weak surface OH acid centres, and the band at 2163 cm⁻¹ was associated with the adsorption of CO on Ce⁴⁺ [66]. The bands at 2127 cm⁻¹ (very weak) and 2107 cm⁻¹ have been described as the interaction of CO with Ce³⁺ [66,67]. Therefore, given their absence, it can be concluded that their presence on the exposed surface is negligible.

The 30Co_R catalyst exhibited a broad band in which there are two main contributions at 2138 cm⁻¹ and 2119 cm⁻¹, attributed to the interaction of CO with Co³⁺ and Co²⁺, respectively [67,68]. Within the same broad band, CO adsorption with ceria is observed at 2156 cm⁻¹. For this catalyst, even though it has been reduced to 450 °C, CO adsorption bands with metallic Co are not observed. This suggests that the surface of the metallic Co particle is covered with unreduced cobalt oxides. Additionally, XRD analysis revealed the presence of cobalt spinel, while XPS analysis confirmed the presence of cobalt oxides on the catalyst surface. The presence of metallic Co is reflected in the catalyst’s magnetic properties when a neodymium magnet is brought close to it (Figure S18). In the case of the 30Cu_R catalyst, a band centred at 2096 cm⁻¹ with a shoulder at higher wave numbers (2104 cm⁻¹) is observed. Considering the arrangement of these two bands, the shoulder can be attributed to linear CO interacting with Cu⁺, while the band at 2096 cm⁻¹ is due to interaction with metallic Cu [69,70,71].

Bimetallic catalysts exhibit CO adsorption bands on metallic Cu species, and the band corresponding to Cu⁺ disappears. Conversely, the CO adsorption bands observed for Co species appear significantly less intense, even for the bimetallic catalyst with the highest Co content (7Cu22Co_R). A shift of the band at 2096 cm⁻¹ towards lower wavelengths (2090 cm⁻¹) is observed in the bands associated with Cu species. This observation suggests an electronic transfer from Co to Cu species

As the hydrogenation reaction of FF requires the activation of the H₂, it was studied by H₂-TPD following activation of H₂ at 400 °C. Depending on the amount adsorbed, it is possible to determine the number of active centres [72,73]. Through the desorption temperature, it is possible to ascertain how H₂ has been chemisorbed and in which types of centres it has been adsorbed [74]. The H₂-TPD plots of the catalysts are displayed in Figure 5, and the results of the H₂-TPD analysis are presented in

Table 3. Analysis of H₂-TPD plots.

Catalyst	H2 Activation (μmol/g)
30Cu_R	632.4
30Co_R	442.3
7Cu22Co_R	214.1
15Cu15Co_R	496.0
22Cu7Co_R	118.2

. The findings presented in Table 3 demonstrate a direct correlation between the extent of catalyst reduction and the enhanced H₂ activation capacity. Thus, monometallic catalysts and the 15Cu15Co_R catalyst exhibit a higher H₂ adsorption capacity than bimetallic catalysts, which may be because these catalysts are completely reduced, and the metallic phases are more prone to activate H₂.

In all cases, the catalysts demonstrate a complex hydrogen desorption pattern, indicating the presence of different hydrogen activation sites. This complexity is more pronounced in the case of bimetallic catalysts. This phenomenon can be attributed to the presence of different phases in the reduced catalysts. XPS analysis revealed the coexistence of metal and oxide phases on the surface of the reduced catalysts in addition to an increase in Ce³⁺. This indicates the presence of oxygen vacancies, which serve to compensate for the decrease in positive charge. XRD analysis has also corroborated the presence of Ce³⁺, as the presence of such cations results in a shift of the fluorite diffraction peaks to lower 2θ, indicating an expansion of the unit cell. The H₂ consumption analysis of the H₂-TPR revealed that only the monometallic catalysts and 15Cu15Co_R were almost completely reduced. This finding suggests the formation of metallic phases. The experimental evidence indicates the presence of metallic phases, oxide phases in interaction with metallic phases, M-MO_x and M^δ+-CeO₂ interfaces, and oxygen vacancies within these reduced catalysts. This complicates the assignment of potential hydrogen desorption bands to a specific phase, rendering the process cumbersome and non-straightforward.

The curves of the monometallic catalysts show a band with a maximum at a temperature of 115 °C for the 30Cu_R catalyst and at 174 °C for the 30Co_R catalyst, while 30Cu_R shows an intense and broad band extending from 180 to 480 °C and a less intense band at higher temperatures. The 30Co_R catalyst shows a more complex profile with different peaks (246, 331, 440, and 559 °C). Considering the aforementioned points, we can tentatively associate the bands at 115 and 174 °C with the dissociation of H₂ on Cu and Co metal centres [12]. The band appearing between 180 and 480 °C is for 30Cu_R, and the bands between 200 and 490 °C, in the case of 30Co_R, are for the interaction of H₂ with the M-MO_x and/or M^δ+-CeO₂ interfaces [12,75,76], and the bands above 500 °C correspond to spillover processes [74,77].

It is evident that the bimetallic catalysts with 22% of Cu or Co exhibited the lowest percentage of reduction, consequently resulting in a significant reduction in the capacity to activate H₂. The 22Cu7Co_R catalyst did not show the broad band presented by the monometallic Cu catalyst between 180 and 480 °C and exhibited a band centred at 124 °C, displaced with respect to that of the 30Cu_R catalyst towards higher temperatures and extending up to 200 °C. Subsequently, a series of bands of reduced intensity are exhibited, extending up to 600 °C. The 7Cu22Co_R catalyst also showed the band at lower temperature but with the maximum at 156 °C, with this temperature being lower than that of the 30Co_R and extending up to 300 °C. This band seems to encompass the two bands presented by the 30Co_R catalyst at 174 and 246 °C. It also shows a complex desorption band between 300 and 500 °C, which is present in the other bimetallic catalysts, and bands at high temperatures like those shown by the 30Co_R catalyst. Finally, the 15Cu15Co_R catalyst was identified as the bimetallic catalyst that adsorbed the most H₂. It is important to note that it exhibited the highest percentage of reduction, and its desorption profile was determined to be a mixture of those observed for the monometallic and bimetallic catalysts. The desorption profile exhibited bands at temperatures below 200 °C, a complex profile between 200 and 500 °C similar to that of the bimetallic catalysts, and desorption bands at temperatures above 500 °C. Contrary to the findings of other authors, who reported that the formation of bimetallic alloys favoured H₂ activation, these bimetallic catalysts did not have a higher H₂ activation capacity. Chen et al. [78] demonstrated that, for Ni-Co catalysts, increasing the nickel content of the alloy enhances the adsorption strength of H₂, shifting the desorption of H₂ towards higher temperatures and increasing the adsorption capacity. On the other hand, Ren et al. [79] studied the promoter effect of rare earth elements (Ce, Y, and La) on Cu/SiO₂ catalysts and postulated that the presence of these elements caused the Cu⁰ electrons to flow into the d orbitals of the dopant element and thus leave empty d orbitals that could bind H₂ molecules. The presence of oxophilic elements such as ReO_x [80] was also found to favour hydrogen dissociation on Ni catalysts. While Liu et al. [81] did not observe any effect on the H₂ adsorption capacity when Ni was doped with La, and Zhao et al. [82] observed a similar phenomenon when Ni was doped with iron. Therefore, the formation of a CuCo alloy could not be favoured in the xCuyCo_R catalysts.

2.2. Catalytic Activity

The catalysts were tested in the hydrogenation of furfural in the liquid phase. For this purpose, the reaction was carried out in a batch hydrothermal reactor that allows for precise control of temperature and pressure. For catalyst screening, the reaction was carried out with 1 mmol of FF in 5 mL of isopropanol (IPA) at a temperature of 170 °C and a pressure of 40 bar of H₂. IPA was chosen as the solvent due to the higher solubility of H₂ in secondary alcohols [33,83]. The reaction time was set at 22 h because at this reaction time the highest PDO yields were obtained, and there were no reproducibility problems, which were observed at shorter reaction times. Prior to the reaction, the catalysts were reduced to 250 °C for the 30Cu_R catalyst and to 400 °C for the other catalysts in a flow of H₂. Subsequently, they were passivated in a stream of O₂/He (1 vol% O₂) for 30 min due to the pyrophoric nature of the catalysts.

Figure 6 shows the catalytic activity of the xCuyCoCe_R catalysts. In all cases, 100% conversion of FF is observed after 22 h. The carbon balance ranges from 50 to 75%, depending on the catalyst. In this reaction, 2-methylfuran (2-MF) and isopropyl furfuryl ether (IFE) were detected as minor compounds, resulting in a carbon balance of less than 100%. Selectivity for the different products varies depending on the catalyst: the 30Co_R catalyst is most selective for 1,5-PDO (38.4%), while the 30Cu_R catalyst is most selective for 1,2-PDO (22.1%). Furthermore, the 1,2-PDO exhibited a tendency whereby an increase in Cu loading was accompanied by an increase in 1,2-PDO. A similar relationship was observed between 1,5-PDO and Co loading, wherein an increase in Co loading was associated with an increase in 1,5-PDO. Bimetallic catalysts generally exhibit greater selectivity towards THFA, ranging from 23.1% for the 15Cu15Co_R catalyst to 31.3% for the 22Cu7Co_R catalyst. From these findings, it seems that copper species were involved in the 1,2-PDO synthesis and cobalt ones in the 1,5-PDO. Regarding the selectivity towards 1,5-PDO exhibited by bimetallic catalysts, an increase is observed as the Co content increases, reaching a maximum for the 15Cu15Co_R catalyst and then decreasing for the 7Cu22Co_R catalyst. It is reasonable to expect that the selectivity of this catalyst will increase in line with the Co content. However, another factor that must be considered when assessing the selectivity of the reaction is the presence of oxygen vacancies. As demonstrated in the literature [21,74], a synergistic relationship between partially reduced Co species, primarily Co²⁺, and oxygen vacancies is imperative for the activation of the furan ring and the hydroxyl group. This enables them to be attacked by the hydrogen activated in the metallic Co. In this work, we were not able to determine the oxygen vacancies because we do not have the EPR equipment to measure them. However, the presence of Ce³⁺ can serve as an indicator of the presence of oxygen vacancies in the catalysts. As demonstrated in Table 1, the surface concentration of Ce³⁺ increases for bimetallic catalysts and decreases with the 7Cu22Co_R catalyst. This finding is consistent with the observation that the selectivity to 1,5-PDO decreases for this catalyst rather than increasing and approaching that of 30Co_R. Therefore, we can conclude that the presence of oxygen vacancies is necessary to produce 1,5-PDO in Co-based catalysts. Table S7 shows the various catalytic systems reported in the bibliography for both cobalt- and copper-based monometallic catalysts compared with 30Co_R and 30Cu_R. In general, the behaviour of the catalysts synthesized in this work showed a similar trend to others reported in the literature, although 30Co_R and 30Cu_R needed longer reaction times and temperatures. Table S7 shows that alumina-based catalysts were the least selective for PDOs, while catalysts supported on ceria or basic supports (mainly lanthanide oxides) were the most selective for 1,5-PDO. This was due to their basic properties activating the hydroxyl group of FAL and the oxygen vacancies favouring the adsorption of both FF and FAL.

The literature has already shown that copper-based catalysts are more selective in opening the ring between C5 and O6 [24,32,33,84], while Co catalysts are more selective at 1.5-PDO by opening the ring between C2 and O6 [35,47,83] (Figure 7). Under the reaction conditions tested, these catalysts have not shown activity in opening the furan ring once THFA has been formed, as has been shown by other authors [7,33,47,48,49]. On the other hand, the formation of 1-pentanol is the product of the hydrogenolysis reaction of 1,5-PDO, while 2-MF is formed from the breaking of the C1-O1 bond of the alcohol group of FFA (Figure 7). The formation of FFA was not observed, which is the intermediate product of the reaction from which pentanediols can be formed by ring opening and subsequent hydrogenation of the diene formed [8,22], 2-MF by hydrogenolysis of the C1-O1 bond, or THFA by hydrogenation of the furan ring. This finding suggests that the formation of FFA must be very rapid and that its conversion to other products must also be greatly favoured by these catalysts. As is shown later, the possibilities for producing different compounds depend largely on the catalyst used as well as on the reaction conditions. The formation of THFA over Co-based catalysts was also associated with the Co particle sizes since bigger particles were prone to adsorb the furan ring by a flat adsorption configuration, giving rise the formation of THFA [25].

The selection of a 22 h reaction time is founded upon the examination of the catalytic activity of the 30Co_R catalyst over time (Figure 8). It is noteworthy that the yield of reaction products after one hour does not reach 10%, even though the FF conversion was 100%. This would indicate that both FF and FFA could be adsorbed on the surface of the catalyst and thus not detected in the solution. As the reaction progresses over time, they were converted in the detected products. Adsorption studies of FF [47] and FAL [33] followed by FTIR have shown that both compounds can be strongly retained by the surface of the catalysts. Shao et al. [33] observed the presence of bands that are indicative of FAL adsorption on CuMg catalysts. This observation established a correlation between FAL adsorption and the strong interaction of FAL or reaction compounds with the basic centres of the catalysts. This strong adsorption subsequently allows the compound to evolve into other reaction products, whereas if the adsorption were weak, it would not favour the evolution of FAL into other products. Similarly, Tian et al. [46] and Wang et al. [47] observed a marked shift in the band associated with the carbonyl group of FF after adsorption onto the surface of the catalysts. This experimental finding is associated with a weakening of the C=O group, allowing its activation for subsequent transformation into reaction products. It can thus be concluded that the observed phenomena must be occurring on the 30Co_R catalyst. This is evidenced by the fact that compounds were not detected at low reaction times, and the C balance declined to less than 20% after one hour of reaction. As the reaction time progresses, the reaction products are detected, thereby substantially improving the C balance in comparison with a one-hour reaction time. Moreover, through DFT studies, several authors have demonstrated the high adsorption energy of both FF and FFA on Co catalysts [84], which could explain this observed experimental fact. Subsequently, between 3 and 9 h, no significant differences in the selectivities of the reaction products are observed, and after 22 h, the selectivity towards 1,5-PDO improves, accompanied by the presence of 1-pentanol.

Continuing with the study of how reaction conditions affect the outcome, the effect of temperature was studied (Figure 9), keeping the rest of the parameters constant. The yield at 1.5-PDO follows a volcano-like trend, reaching its maximum at 170 °C, at the expense of THFA [47,83]. If the reaction temperature is increased to 180 °C, then the hydrogenolysis of FFA to 2-MF is favoured over the opening or hydrogenation of the furan ring [77]. At lower reaction temperatures, THFA is the major reaction product, as shown by other authors [35,47,49], because lower temperatures favoured a parallel orientation of the furan ring in relation to the metal surface, facilitating the hydrogenation of the furan ring [35].

Another parameter that has a significant impact on this reaction is the initial H₂ pressure. To this end, the evolution of the reaction products was studied from 1 bar of H₂ to 40 bar, while all other parameters were kept constant (Figure 10). Increasing the pressure has been shown to improve product yield by reducing the amount of other unquantifiable products. Furthermore, it was observed that a minimum of 30 bar of H₂ is required to achieve yields of 1.5-PDO exceeding 20%. In contrast, the 30Co_R catalyst was shown to reduce FF to THFA at a pressure of 1 bar, indicating its capacity to activate sufficient H₂ to hydrogenate the furan ring. However, higher pressures are required to open this ring, as demonstrated by Wang et al. [84] with CuCe-based catalysts. It is only at a pressure of 10 bar that a residual presence of FFA (approximately 1%) is observed.

Finally, the recyclability of the 30Co_R catalyst was evaluated after five reaction cycles (Figure 11). After each cycle, the dissolution was removed, and a new fresh solution was poured into the reactor without reactivating the catalyst. FF conversion was complete in all cycles, although, as demonstrated in Figure 11, the carbon balance declined from the first cycle onwards. The yield at 1.5-PDO exhibits a decline from one cycle to the next, reaching 10% after the fifth cycle, whereas THFA remains stable during the first three cycles and then also declines in the fourth and fifth cycles. It was observed that, while 2-MF was a minor compound in the initial cycle (<1%), its production increased from the second cycle onwards, remaining stable from cycles 2 to 5.

In order to understand the alteration in selectivity, the catalyst was analysed following the initial three reuse cycles. Elemental analysis (Figure S19) indicates that the catalyst contains 2% carbon and 0.77% hydrogen, corresponding to a carbon-to-hydrogen molar ratio of 0.29. This molar ratio is comparable to that observed in alcohols such as 1-pentanol or PDOs, yet it differs from that of FF, FFA, and THFA. Thermogravimetric analysis indicates an approximate weight loss of 5%, concomitant with an intense exothermic peak. The presence of multiple shoulders on this peak indicates that the C deposits must be of varied nature. Regarding the surface composition, the catalyst was analysed using XPS (Table S2). On the one hand, the surface C content increases, accompanied by a decrease in the superficial Ce content but not in the Co content. The slight increase in the surface amount of Co indicates that the C would be deposited on the Ce sites. In relation to Co2p, after three reaction cycles, it is evident that, in congruence with the observations concerning the reduced catalyst 30Co_R, Co is predominantly present as CoO. However, a marginal shift in the binding energy of the 2p_3/2 peak towards higher energies (~0.8 eV) was detected. This indicates an occurrence of charge transfer from Co²⁺ to the oxygen in its immediate environment. This phenomenon can be discerned during the recording of the O1s level signal (Figure S20, Table S6), where a band emerges at 528.1 eV, along with the bands at 529.7 eV and 531.8 eV. The first two are typical of lattice oxygen and the latter of -OH groups. This indicates a change in both Co²⁺ and oxygen. It could be hypothesised that the band at 528.1 eV would be associated with Co²⁺-O and that the band at 529.7 eV would be associated with Ce-O, in view of the observed changes. Furthermore, the Ce 3d signal shows no significant changes with respect to the band in the catalyst before the reaction (Figure S21, Table S4), and furthermore, the percentage of Ce³⁺ in the used catalyst barely decreases after the two reaction cycles, decreasing from 18.7% to 18.1%, which would indicate that the number of oxygen vacancies generated when forming Ce³⁺ would not be severely affected. The C1s level (Figure S22) shows a band at 284 eV, characteristic of sp2 C. This C is found in the FF and FAL compounds. In addition, another band appears at 285.5 eV, more intense than the previous one. This band, shifted by 0.7 eV from aliphatic C (284.8 eV), indicates that the C has a more electronegative element in its environment, such as oxygen, which removes charge from the C. Considering the position of this band, we can propose that it is due to C-OH. Finally, the catalyst was studied using XRD; we observed in this case that the crystallinity of the phases present increases considerably after the reaction, with the diffraction peaks due to CoO clearly distinguishable (Figure S23). After these experimental observations, we can conclude that the reaction causes changes in the cobalt phase without affecting the ceria. It was hypothesized [22] that oxophilic cobalt CoO_x species are responsible for the adsorption of both FF and FFA in an inclined configuration, thereby favouring the opening of the FFA ring. This phenomenon can be attributed to the high electron density exhibited by these species, which impedes the adsorption of the C=C bond of the furan ring on the surface. Subsequent to the reaction, partial oxidation of Co²⁺ is observed, indicating that the inclined adsorption of FFA is no longer favoured and can be accommodated in a flat configuration, thereby reducing the furan ring, accompanied by a decrease in 1,5-PDO. Furthermore, the regrowth of particles as observed by XRD must be considered, as this would have a detrimental effect by decreasing the active centres and thus the activity [30]. The decline in the sum of yields could be attributed to the heightened propensity of the alcohols involved in the reaction to adsorb on the surface. This phenomenon can be attributed to the increase in Lewis acidity of the surface that occurs with the partial oxidation of the Co²⁺.

3. Materials and Methods

3.1. Synthesis of the Catalysts

The synthesis of the xCuyCoCe catalysts was carried out with the utilization of the coprecipitation method, selected for its simplicity and cost effectiveness. The predetermined quantity of the ceria precursor (Ce(NO₃)₃·6H₂O (Sigma-Aldrich, St. Louis, MO, USA, 99.99%)) was dissolved in 100 mL of distilled water, followed by the addition of the Cu (Cu(NO₃)₂·3H₂O (Merck KGaA, Darmstadt, Germany, 99.5%)) and Co (Co(NO₃)₂·6 H₂O (Sigma-Aldrich, St. Louis, MO, USA, 98%)) salts (x and y represent the weight percentages of each metal, which were 7.5%, 15%, and 22.5%.). The pH of the solution was subsequently adjusted to 9 using a solution 0.5 M of NaOH, and the mixture was aged for 90 min under vigorous stirring. Then, the solid was filtrated using a Büchner funnel and a vacuum pump, and it was thoroughly washed with distilled water to eliminate any potential sodium cation contamination. The solid was then left to dry overnight at temperatures beginning at 80 °C, followed by calcination at 450 °C for two hours, with a heating ramp of 5 °C/min. Before reaction, the oxides were reduced ex situ at the selected temperature under a H₂ stream for 1 h.

The hydroxides precursors were labelled xCuyCoCe_OH, where x and y account for the mass ratio of Cu and Co used in the synthesis. The hydroxides were converted in the corresponding oxides by calcination and were named xCuyCoCe_O. Finally, the reduced catalysts were labelled as xCuyCoCe_R

3.2. Characterization Techniques

The textural characteristics of the xCuyCo_O materials were evaluated through nitrogen adsorption–desorption isotherms at –196 °C using an ASAP 2020 instrument (Micromeritics, Norcross, GA, USA). Prior to analysis, the samples were degassed under vacuum at 150 °C for 12 h (10⁻⁴ mbar). The specific surface area was calculated using the Brunauer–Emmett–Teller (BET) method, assuming a nitrogen molecular cross-sectional area of 16.2 Ų. The pore size distribution was derived from the adsorption branch of the isotherm using the BJH method. The total pore volume was determined from the volume of nitrogen adsorbed at a relative pressure (P/P₀) of 0.996.

Powder X-ray diffraction (PXRD) analysis was employed to investigate the crystalline phases of the solids. Measurements were performed using a PANalytical X’Pert PRO diffractometer (Bruker, Rheinstetten, Germany) equipped with a germanium monochromator and Cu Kα radiation (λ = 1.5406 Å). This technique provided insights into the degree of crystallinity and phase composition of the metal-loaded ceria supports.

The TG-DSC data were recorded with a Mettler-Toledo (TGA/DSC-1) instrument (Columbus, OH, USA) equipped with a MX5 microbalance.

The morphology and dispersion of Ce, Cu, and Co were examined by transmission electron microscopy (TEM) combined with energy-dispersive X-ray spectroscopy (EDS), using a FEI Talos F200X microscope (Thermo Fisher Scientific, Waltham, MA, USA). This instrument integrates high-resolution STEM and conventional TEM imaging modes with EDS-based elemental mapping, enabling spatially resolved 3D chemical characterization. Prior to analysis, samples were ultrasonically dispersed in ethanol, and a drop of the suspension was deposited on a carbon-coated copper grid (300 mesh).

X-ray photoelectron spectroscopy (XPS) was used to study the surface elemental composition and oxidation states of the different elements. Spectra were collected using a Physical Electronics PHI 5700 spectrometer (Physical Electronics, Eden Prairie, MN, USA) equipped with non-monochromatic Mg Kα radiation (15 kV, 300 W, 1253.6 eV) and a multichannel detector. The analysis was conducted in constant pass energy mode (29.35 eV) with a 720 μm diameter analysis area. The C 1s signal at 284.8 eV was used for charge referencing. Data processing and peak fitting were performed with the PHI ACCESS ESCA-V6.0F software, using a Shirley-type background and Gaussian–Lorentzian functions for deconvolution.

CO chemisorption was performed under static volumetric conditions by using a Micromeritics ASAP 2020 (Micromeritics, Norcross, GA, USA) apparatus. Approximately 180 mg of the sample was reduced in situ with a flow of 10vol%H₂/Ar (30 mL/min) at a fixed temperature for 1 h. Then, they were cooled down at 50 °C and degassed with Ar (30 mL/min) at 25 °C for 10 h. The chemisorption isotherms were obtained by measuring the amount of CO adsorbed between 50 and 500 mmHg at 25 °C. After completing the initial analysis, the CO physiosorbed was evacuated, and a second isotherm was measured to determine the CO irreversibly chemisorbed.

A Nicolet 5700 spectrometer (GMI, Ramsey, MN, USA), equipped with a high-sensitivity Hg-Cd-Te cryodetector, was utilized to collect diffuse reflectance infrared Fourier-transform (DRIFT) spectra. The diffuse reflectance accessory Playing Mantis manufactured by Harrick Co. (CHC-CHA-3 Harrick Scientific Products, Pleasantville, NY, USA) was used as an optical mirror accessory. Approximately 30 mg of the previously ground sample was placed in a low-temperature catalytic reaction chamber (CHC-CHA-3 Harrick Scientific Products, Pleasantville, NY, USA) that allows treatment in situ between 123 and 873 K. To identify the state of oxidation and different metals (Ce, Co, and Cu) in the surface, carbon monoxide (CO) was used as a probe molecule at low temperature. Prior to the CO adsorption, reduced and passivated catalysts were exposed to a flow of 10% H₂/Ar (50 mL min⁻¹) at a temperature of 523 K for one hour. Subsequently, H₂ was substituted with Ar (50 mL min⁻¹), and the sample was cooled to 123 K. Thereafter, 5% CO/He flow (30 mL min⁻¹) was introduced for adsorption. Physiosorbed CO was removed by flushing with Ar at this temperature for 30 min. All spectra were recorded with 128 scans and a resolution of 4 cm⁻¹.

The H₂ activation was determined by temperature-programmed desorption of hydrogen (H₂-TPD) using an AutoChem II 2920 analyser (Micromeritics, Norcross, GA, USA). Approximately, 150 mg of sample was initially reduced, following the same conditions explained above, and cooled down in the H₂/Ar flow up to 50 °C. Next, the sample was degassed in Ar flow during 30 min. The TPD analysis was carried out in an Ar flow, increasing the temperature from 50 °C up to 800 °C with a heating rate of 5 °C/min; the evolved hydrogen was monitored with a thermal conductivity detector (TCD). The same equipment was used for the determination of the total acidity of the catalysts by means of temperature-programmed desorption of ammonia (NH₃-TPD). Nearly 50 mg of sample was initially pretreated in a helium flow (50 mL/min) with a heating ramp of 30 °C/min up to 400 °C, where it was held for 15 min. After cooling to 80 °C under He, the sample was exposed to a 10% NH₃/He mixture for 20 min at the same temperature. Physiosorbed ammonia was removed by purging with helium for 90 min. The desorption step was carried out by heating the sample from 80 to 450 °C while monitoring ammonia release with a thermal conductivity detector (TCD).

3.3. Catalytic Analysis

Catalytic tests were conducted using a 10 mL batch reactor. In the Teflon vessel, 5 mL of isopropanol, 1 mmol of FF, and the catalyst previously reduced at 400 °C for 1 h with a heating rate of 10 °C/min under a H₂ flow of 60 mL/min were added. (For tests involving 30Cu_R catalyst, reduction was performed at 250 °C.)

The reactions were carried out in an aluminium block where the reactor was placed to ensure uniform heat distribution at different temperatures (120–180 °C), pressures (1–40 bar H₂), and times (1–22 h). Before pressurizing the reactor with hydrogen, any existing air was purged by passing a nitrogen flow through the system.

The aliquots were analysed by gas chromatography using a flame ionization detector and a CP-Wax 52 CB capillary column.

The FF conversion and products yield were calculated as follows:

$$Conversion\left(\%\right)={\displaystyle \frac{moloffurfuralconverted}{moloffurfuralfed}}\times 100$$

(1)

$$Yield\left(\%\right)={\displaystyle \frac{molofproduct}{moloffurfuralfed}}\times 100$$

(2)

4. Conclusions

The study demonstrates that Cu- and Co-based monometallic and bimetallic catalysts supported on CeO₂ exhibit high conversion of furfural to products of interest, such as 1,5-pentanediol and tetrahydrofurfuryl alcohol. Monometallic catalysts show a greater capacity for H₂ activation, while bimetallic catalysts exhibit a complex interaction between the metal phases and the oxides, which influences product selectivity. Selectivity towards 1,5-PDO is higher in catalysts with higher Co content, while catalysts with higher Cu content favour the formation of 1,2-PDO. Furthermore, it was observed that the complete reduction of oxides and the generation of oxygen vacancies on the surface are key factors in improving catalytic activity. Although bimetallic catalysts do not outperform monometallic catalysts in H₂ activation, their design allows for the modulation of selectivity and the optimisation of the production of desired compounds under specific reaction conditions.