Ru Catalysts Supported on Commercial and Biomass-Derived Activated Carbons for the Transformation of Levulinic Acid into γ -Valerolactone under Mild Conditions

: Ru catalysts (1 wt.%) supported on commercial and biomass-derived activated carbons (AC) have been prepared, thoroughly characterized, and used in the hydrogenation of levulinic acid to produce gamma-valerolactone (GVL). This is an important platform compound that plays a key role in the production of liquid fuels and that can also be used, for example, as a food ﬂavoring agent, antifreeze, and solvent. The study focuses on the inﬂuence of the carbon support characteristics, such as porous texture and acidity, on the properties and performance (LA conversion and selectivity to GVL) of the catalysts. Catalytic activity tests have been carried out at 170 ◦ C and also in noticeably milder conditions (70 ◦ C) to implement a less energy-demanding process. All the catalysts show high LA conversion and GVL yield at 170 ◦ C, while at 70 ◦ C, important differences between them, related to the support properties, have been found. The catalysts prepared with more acidic supports show better catalytic properties: very good catalytic performance (98% LA conversion and 77% selectivity to GVL) has been obtained in mild temperature conditions.


Introduction
The massive use of fossil fuels has led to their depletion and to important environmental pollution problems. To face this situation, it is necessary to promote the use of renewable energy sources and to develop both new strategies for energy production and sustainable methods to obtain chemical products. In this line, lignocellulosic biomass residues are renewable raw materials with the potential to produce high added-value products [1,2], such as fuels, platform molecules for the chemical industry, and functional materials [3]. The main sources of lignocellulosic biomass are agricultural and forestry residues.
Lignocellulosic biomass is composed of hemicellulose (15-30 wt.%), cellulose (30-50 wt.%), and lignin (15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30) wt.%), in proportions that depend on the biomass nature and with different reactivity and transformation possibilities ( Figure 1). The hydrolysis of cellulose produces glucose, which can be further hydrolyzed to 5-hydroxymethylfurfural (HMF), an interesting precursor for the synthesis of other high added-value chemicals, and leads to the formation of levulinic acid (LA) and formic acid (FA) [4,5] by rehydration. Levulinic acid can be also obtained from xylose, the hydrolysis product of hemicellulose, after dehydration to furfural and hydrogenation to furfuryl alcohol ( Figure 1). LA is one of the best chemical platform compounds, leading to suitable intermediates for the preparation of a variety of pharmaceutical and industrial products with applications, such as food flavoring agents, antifreeze, solvents, etc. [6][7][8][9][10]. Among the interesting products that can be obtained from LA, gamma-valerolactone (GVL) stands out because it is a relevant compound in a "cascade" of processes for the production of liquid fuels and also because it has important direct applications, for example, as a fragrance, food ingredient, fuel   [14,16].
Hydrogenation of LA to obtain GVL has been carried out with both homogenous and heterogeneous catalysts [12]. In reference to homogenous systems, Starodubtseva et al. [17] reported the effectivity of a Ru II -BINAP-HCl catalytic system, with 95% GVL yield in The process requires the use of a catalyst and severe pressure and temperature conditions to achieve a high GVL yield [15]. Regarding the catalytic system, it must be active for the hydrogenation of carbonyl groups, but some acidity is also required for the dehydration step.
Catalysts 2021, 11, x FOR PEER REVIEW 2 of 16 that can be obtained from LA, gamma-valerolactone (GVL) stands out because it is a relevant compound in a "cascade" of processes for the production of liquid fuels and also because it has important direct applications, for example, as a fragrance, food ingredient, fuel additive, and solvent. Some of these applications are related to interesting properties, such as limited reactivity with water or oxygen, high boiling point, low vapor pressure, and low toxicity [11][12][13]. GVL can be obtained from levulinic acid by any of the two routes shown in Figure 2 [14]: i) dehydration to produce the intermediate angelica lactone (AL), which is further hydrogenated, or ii) hydrogenation, to produce 4-hydroxypentanoic acid (HPA) by reduction of the carbonyl group, followed by dehydration. The process requires the use of a catalyst and severe pressure and temperature conditions to achieve a high GVL yield [15]. Regarding the catalytic system, it must be active for the hydrogenation of carbonyl groups, but some acidity is also required for the dehydration step.   [14,16].
Hydrogenation of LA to obtain GVL has been carried out with both homogenous and heterogeneous catalysts [12]. In reference to homogenous systems, Starodubtseva et al. [17] reported the effectivity of a Ru II -BINAP-HCl catalytic system, with 95% GVL yield in Figure 2. Scheme of potential reactions taking place in the hydrogenation of levulinic acid to obtain γ-valerolactone, based on schemes in [14,16].
Hydrogenation of LA to obtain GVL has been carried out with both homogenous and heterogeneous catalysts [12]. In reference to homogenous systems, Starodubtseva et al. [17] reported the effectivity of a Ru II -BINAP-HCl catalytic system, with 95% GVL yield in ethanol, and Mehdi et al. [18] reported a 96% GLV yield using Ru(acac) 3 -TPPTS in a water solution. However, homogeneous systems have limitations, such as difficult recoverability and lower thermal stability, hindering their use on an industrial scale. Because of that, heterogeneous catalysts are the most promising and technically feasible option. A current challenge is to reduce the energy demand of the process by using more efficient (active and selective) catalysts and milder reaction conditions. Many authors have claimed that supported noble metal catalysts show good performance in the conversion of LA to GVL, among which ruthenium-based ones present the highest catalytic activity and selectivity [12,14,15,[19][20][21]. The ruthenium-supported catalysts have proved to be efficient in many other reactions, highlighting, as an example, their performance in hydrogenation reactions [14,22,23]. The support plays a relevant role, as it allows a good dispersion of the hydrogenation active species and regulates the degree of metal-support interaction, which has been reported to influence the catalytic activity [24,25]. Besides, it must also be responsible for the acidic character of the catalysts, being an active species as well.
Several materials have been reported as supports for Ru and other noble metals in the development of catalysts for the transformation of LA into GVL. Some examples are inorganic oxides like SiO 2 [26], ZrO 2 [27], and TiO 2 [28], polymeric materials as cross-linked sulfonated polyethersulfone (SPES) [29], materials of mineral origin like hydroxyapatite (HPA) [30], and carbon materials [21]. Among them, carbon materials can be considered an interesting option, due to their tunable surface area and porous structure, surface chemistry, and the possibility of being shaped in different morphologies [24]. Moreover, as they can be obtained from lignocellulosic biomass residues, the use of this kind of carbon materials to prepare catalysts for LA hydrogenation would contribute to an integral approach towards the valorization of the mentioned residues. From a broad perspective, they can be regarded as a suitable alternative to prepare green catalysts.
There are a number of publications dealing with Ru/C catalysts [19,24,25,31,32]. However, in most of them, the obtained catalytic activity results are not discussed in relation to the support physicochemical properties. Moreover, although interesting results have been already obtained with carbon-supported Ru catalysts, it is still a challenge to develop a catalytic system with high efficiency (high GVL yield) achieved in as mild as possible conditions (mainly, low temperature, low pressure, short reaction time, low metal content, and water as a solvent).
Thus, the purpose of this work is to develop carbon-supported Ru catalysts that fulfill the aforementioned requirements, optimizing as much as possible the resources. Catalysts with a low Ru content, 1 wt.%, have been prepared using different carbon materials as supports. They have been characterized in detail and tested in the LA hydrogenation to GVL both in regular and mild reaction conditions. The study focuses on the influence of the supports' characteristics (porous structure, surface area, and surface chemistry), on the catalysts' properties and catalytic behavior. Figure 3 shows the N 2 adsorption-desorption isotherms obtained for the supports and the reduced Ru R /C catalysts. The isotherms are type IV according to IUPAC classification [33], typical of materials that contain both micro and mesopores.  The calculated values of specific surface area and pore volumes for supports and catalysts (as prepared and reduced) are presented in Table 1.  Table 1 show that the carbon materials used in this work have significantly different textural properties, regarding both surface area and pore size distribution.

Textural Properties of Supports and Catalysts
After impregnation with the RuCl3 aqueous solution and drying, the samples show, in general, a small decrease in their adsorption capacity with respect to the corresponding carbon supports. However, after the reduction treatment there is a noticeable increase in the surface area of sample Ru R /LC (a very slight increase is also found for Ru R /WV). Data in Table 1 show that the surface area increase in sample Ru R /LC is related to a significant increase in the micro and supermicropore volumes. It seems that part of this porosity of the LC carbon was blocked and it has been cleared upon the catalyst reduction heat treatment. Such treatment, in pure H2 at 250 °C for 4 h, probably removes oxygen functional groups (OFG), leaving some pores accessible to gas adsorption. The process of OFG removal could be catalyzed by the reduced Ru species. Thus, as it will be shown in the next The calculated values of specific surface area and pore volumes for supports and catalysts (as prepared and reduced) are presented in Table 1. Data in Table 1 show that the carbon materials used in this work have significantly different textural properties, regarding both surface area and pore size distribution.
After impregnation with the RuCl 3 aqueous solution and drying, the samples show, in general, a small decrease in their adsorption capacity with respect to the corresponding carbon supports. However, after the reduction treatment there is a noticeable increase in the surface area of sample Ru R /LC (a very slight increase is also found for Ru R /WV). Data in Table 1 show that the surface area increase in sample Ru R /LC is related to a significant increase in the micro and supermicropore volumes. It seems that part of this porosity of the LC carbon was blocked and it has been cleared upon the catalyst reduction heat treatment. Such treatment, in pure H 2 at 250 • C for 4 h, probably removes oxygen functional groups (OFG), leaving some pores accessible to gas adsorption. The process of OFG removal could be catalyzed by the reduced Ru species. Thus, as it will be shown in the next paragraph, the TPD profiles of the reduced catalysts are different from those of the pristine supports (Table S1 and Figure S1). These differences can be related to modifications associated to the impregnation step, but also to a catalytic effect of metallic Ru to decompose OFG. Data of Table S1 show that the amount of CO 2 and CO evolved increases for all the reduced catalysts compared to the supports, but such an increase is particularly high in the case of sample Ru R /LC. This means that in this case, the surface chemistry (and likely also the textural properties) of the LC carbon is more prone to thermally induced modifications, in particular, those catalyzed by the Ru nanoparticles.
A thorough analysis comparing the different pore volumes (Table 1) in supports and reduced catalysts shows that, in general, and due to the low Ru loading, the supported Ru nanoparticles have a relatively slight effect on the porosity. However, such an effect is not equal for the different catalysts and allows to draw an approximate image of the location of the Ru particles within the porosity. Figure S2 shows a graphic of such a comparison. Thus, for sample Ru R /SA-30, there is only a small decrease of the micropore volume (V DR,N2 , and V DR,CO2 ); for Ru R /WV-1100, the pore volume does not decrease at all; in the case of Ru R /ASC, there is a significant decrease of the micropore volume (V DR, N2, V DR, CO2, and V super micro ) and in the case of Ru R /LC, the decrease is noticeable in the mesopore volume, although, as commented, the micropore volume also increases, due to OFG removal. Figure 4 shows the CO 2 and CO evolution TPD curves obtained for the carbon supports. CO 2 and CO evolve as a consequence of the decomposition of oxygen functional groups (OFG) with different thermal stability and different chemical nature [34][35][36][37]. CO 2 is mainly generated by the decomposition of functional groups with acidic character like carboxylic acids, anhydrides, and lactones, that decompose in the following temperature intervals:~230-380 • C,~520-550 • C, and~650-670 • C, respectively. On the other hand, CO emission is due to the decomposition of weakly acidic phenol type groups, and carbonyls and quinones, with basic character, that decompose at higher temperatures: at~650-700 • C and~750-943 • C, respectively. paragraph, the TPD profiles of the reduced catalysts are different from those of the pristine supports (Table S1 and Figure S1). These differences can be related to modifications associated to the impregnation step, but also to a catalytic effect of metallic Ru to decompose OFG. Data of Table S1 show that the amount of CO2 and CO evolved increases for all the reduced catalysts compared to the supports, but such an increase is particularly high in the case of sample Ru R /LC. This means that in this case, the surface chemistry (and likely also the textural properties) of the LC carbon is more prone to thermally induced modifications, in particular, those catalyzed by the Ru nanoparticles. A thorough analysis comparing the different pore volumes (Table 1) in supports and reduced catalysts shows that, in general, and due to the low Ru loading, the supported Ru nanoparticles have a relatively slight effect on the porosity. However, such an effect is not equal for the different catalysts and allows to draw an approximate image of the location of the Ru particles within the porosity. Figure S2 shows a graphic of such a comparison. Thus, for sample Ru R /SA-30, there is only a small decrease of the micropore volume (V DR,N2, and V DR,CO2); for Ru R /WV-1100, the pore volume does not decrease at all; in the case of Ru R /ASC, there is a significant decrease of the micropore volume (VDR, N2, VDR, CO2, and Vsuper micro) and in the case of Ru R /LC, the decrease is noticeable in the mesopore volume, although, as commented, the micropore volume also increases, due to OFG removal. Figure 4 shows the CO2 and CO evolution TPD curves obtained for the carbon supports. CO2 and CO evolve as a consequence of the decomposition of oxygen functional groups (OFG) with different thermal stability and different chemical nature [34][35][36][37]. CO2 is mainly generated by the decomposition of functional groups with acidic character like carboxylic acids, anhydrides, and lactones, that decompose in the following temperature intervals: ~ 230-380 °C, ~ 520-550 °C, and ~ 650-670 °C, respectively. On the other hand, CO emission is due to the decomposition of weakly acidic phenol type groups, and carbonyls and quinones, with basic character, that decompose at higher temperatures: at ~ 650-700 °C and ~ 750-943 °C, respectively. Besides, CO and, especially, CO2 desorption at high temperature, above ~ 860 °C (Figure 4) is not common and can be explained by the decomposition of OFG related to the presence of residual phosphorus species remaining from the H3PO4 activation treatment. As previously reported, C-O-P bonds present on the surface are thermally unstable at temperatures above 700 °C [38,39].

Surface Chemistry of Supports and Catalysts
The quantification of the TPD profiles of the carbon supports is presented in Table 2. These data reveal important differences in the surface chemistry of the supports: biomassderived ACs have a lower amount of surface oxygen groups, and lower CO2/CO ratio, than commercial ones, which can be linked to a higher basic character of the former ones.  Besides, CO and, especially, CO 2 desorption at high temperature, above~860 • C ( Figure 4) is not common and can be explained by the decomposition of OFG related to the presence of residual phosphorus species remaining from the H 3 PO 4 activation treatment. As previously reported, C-O-P bonds present on the surface are thermally unstable at temperatures above 700 • C [38,39].
The quantification of the TPD profiles of the carbon supports is presented in Table 2. These data reveal important differences in the surface chemistry of the supports: biomassderived ACs have a lower amount of surface oxygen groups, and lower CO 2 /CO ratio, than commercial ones, which can be linked to a higher basic character of the former ones. The reduced catalysts have been also characterized by TPD experiments. As mentioned above, the TPD profiles of the reduced catalysts are significantly different from those of the pristine supports (Table S1 and Figure S1) as a consequence of modifications of the OFG associated with the impregnation step and the influence of the presence of metallic Ru in their thermal decomposition [40,41]. The amounts of generated CO 2 and CO, products of the OFG decomposition, are higher for the reduced catalysts than for the supports and reflects that with the aid of the Ru nanoparticles more OFG can be decomposed in the temperature interval used in the TPD experiments. The mentioned increment of decomposed OFG depends on the nature of the carbon material and has shown to be the highest for sample Ru R /LC. Figure 5 shows some representative TEM images of the Ru R /C catalysts. It can be observed that the four samples contain small Ru nanoparticles. The particle size distribution determined from the measurement of more than 100 particles (see Figure S3 and Table S2) reveals that the average particle size is around 2 nm in the four catalysts. The observed particles are very well distributed in the Ru R /ASC catalyst, whereas in the other three catalysts they seem to be concentrated in particular regions. Besides, for the Ru R /SA-30 catalyst some larger particles have also been observed (see inset of Figure S3). These differences imply that the support properties affect the interaction of the Ru precursor with the support's surface and, likely, also the modifications occurring during the reduction process.  The reduced catalysts have been also characterized by TPD experiments. As mentioned above, the TPD profiles of the reduced catalysts are significantly different from those of the pristine supports (Table S1 and Figure S1) as a consequence of modifications of the OFG associated with the impregnation step and the influence of the presence of metallic Ru in their thermal decomposition [40,41]. The amounts of generated CO2 and CO, products of the OFG decomposition, are higher for the reduced catalysts than for the supports and reflects that with the aid of the Ru nanoparticles more OFG can be decomposed in the temperature interval used in the TPD experiments. The mentioned increment of decomposed OFG depends on the nature of the carbon material and has shown to be the highest for sample Ru R /LC. Figure 5 shows some representative TEM images of the Ru R /C catalysts. It can be observed that the four samples contain small Ru nanoparticles. The particle size distribution determined from the measurement of more than 100 particles (see Figure S3 and Table  S2) reveals that the average particle size is around 2 nm in the four catalysts. The observed particles are very well distributed in the Ru R /ASC catalyst, whereas in the other three catalysts they seem to be concentrated in particular regions. Besides, for the Ru R /SA-30 catalyst some larger particles have also been observed (see inset of Figure S3). These differences imply that the support properties affect the interaction of the Ru precursor with the support's surface and, likely, also the modifications occurring during the reduction process.  Figure 6 shows the Ru 3p3/2 XPS spectra of the Ru R /C catalysts. Two Ru species with different 3p3/2 binding energy (B.E.) are present in these samples: Ru 0 (B.E. of about 462 eV) [42] and RuOx/Ru 0 (Ru oxide on metallic Ru, with B.E. of about 465 eV) [43]. The presence of RuCl3 can be discarded because the reported B.E. for this species is 464.1 eV [44]. The B.E. values, the atomic percentage of the two mentioned species, and the atomic Ru/C, Cl/C, and Cl/Ru ratios are presented in Table 3. Ru 0 is the predominant species in the four catalysts (more than 60 at.%). The presence of Ru as RuOx/Ru 0 means that partial surface oxidation of the Ru particles takes place upon exposure to the environment. It can be pointed out that the proportion of Ru 0 is the largest (83 at.%) in the Ru R /WV-1100 catalyst, meaning that the reduced Ru particles are more stable in this sample. However, there is not a clear relationship between this observation and the carbon support properties.   [42] and RuO x /Ru 0 (Ru oxide on metallic Ru, with B.E. of about 465 eV) [43]. The presence of RuCl 3 can be discarded because the reported B.E. for this species is 464.1 eV [44]. The B.E. values, the atomic percentage of the two mentioned species, and the atomic Ru/C, Cl/C, and Cl/Ru ratios are presented in Table 3. Ru 0 is the predominant species in the four catalysts (more than 60 at.%). The presence of Ru as RuO x /Ru 0 means that partial surface oxidation of the Ru particles takes place upon exposure to the environment. It can be pointed out that the proportion of Ru 0 is the largest (83 at.%) in the Ru R /WV-1100 catalyst, meaning that the reduced Ru particles are more stable in this sample. However, there is not a clear relationship between this observation and the carbon support properties.  Figure 6 shows the Ru 3p3/2 XPS spectra of the Ru R /C catalysts. Two Ru species with different 3p3/2 binding energy (B.E.) are present in these samples: Ru 0 (B.E. of about 462 eV) [42] and RuOx/Ru 0 (Ru oxide on metallic Ru, with B.E. of about 465 eV) [43]. The presence of RuCl3 can be discarded because the reported B.E. for this species is 464.1 eV [44]. The B.E. values, the atomic percentage of the two mentioned species, and the atomic Ru/C, Cl/C, and Cl/Ru ratios are presented in Table 3. Ru 0 is the predominant species in the four catalysts (more than 60 at.%). The presence of Ru as RuOx/Ru 0 means that partial surface oxidation of the Ru particles takes place upon exposure to the environment. It can be pointed out that the proportion of Ru 0 is the largest (83 at.%) in the Ru R /WV-1100 catalyst, meaning that the reduced Ru particles are more stable in this sample. However, there is not a clear relationship between this observation and the carbon support properties.   The slight differences in the Ru 3p 3/2 B.E. values corresponding to Ru 0 and RuO x /Ru 0 in the different catalysts are not significant. The standard deviation is 0.25 and 0.35, respectively-values which are well in the reported intervals for these data [44]-and can be associated with fitting difficulties, due to the lower intensity signal arising from the low Ru loaded catalysts.

TEM and XPS Analysis of Ru Catalysts
The Ru/C atomic ratio is very similar for the four catalysts, meaning that the Ru dispersion is similar, in agreement with the similitude of the Ru particle size determined from the TEM analysis. Small amounts of chlorine (~0.1-0.9 wt.%) and phosphor (~1.5-2 wt.%) have been detected to be present in the catalysts. According to the Cl/Ru ratio, the Ru R /WV-1100 and Ru R /ASC catalysts keep about 1/3 of the chlorine from the impregnation process. Some chloride commonly remains in catalysts prepared by a similar preparation method and conditions, as evidenced in other works as [24,30,45,46].

Catalytic Activity Tests
Blank experiments (without catalyst) do not show LA conversion. However, the tests carried out with the carbon supports show some LA conversion and GVL yield (at 170 • C, LA conversion is about 40% with 5% GVL yield, and at 70 • C, about 11% LA is converted, without GVL formation). These results can be considered in agreement with those reported by Sudhakar et al. [30], who proved that, at 70 • C and 5 bar H 2 , 2% LA conversion, with 99% selectivity to GVL, was achieved using hydroxyapatite support. Table 4 shows the catalytic activity results obtained with the four tested catalysts in tests performed at 170 • C. No other products apart from GVL have been detected in this study. Intermediates such as 4-hydroxypentanoic acid (HPA) or angelica lactone (AL) (Figure 2), are not present in the resulting solution. However, as the chromatographic column used for HPLC analysis, well suited for the analysis of LA and GVL, is not very appropriate for 1,4-pentanediol (PND) detection, the formation of this compound cannot be discarded. This would explain the fact that the selectivity to GVL did not reach 100%.
Recycling tests have been performed with three of the catalysts, showing that they maintain their high LA conversion and GVL yield in a second catalytic run. Table 5 shows the obtained results. The four catalysts show a very good catalytic behavior at 170 • C, with LA conversion close to 100% and GVL yield above 80%. Hence, operating at this reaction temperature, it is not possible to assess the relative influence of the properties of the catalysts (surface area, surface chemistry, and/or state of active species) in the catalytic behavior. Note that the mean nanoparticle size is similar in all the catalysts (~2 nm), so it is not a decisive parameter to explain the differences in the catalytic behavior of the studied catalysts.
The obtained catalytic results are interesting, but to put them in context it is necessary to compare them with analogous results obtained with different Ru/C catalysts reported in the literature. Such a comparison (Table 6) shows that the carbon supports used in the present work play a relevant role in the high performance of the prepared catalysts. Although reaction conditions differ in most of the reported studies, a general comparative analysis can be drawn from such a comparison. As shown in Table 6, most of the reported studies deal with 4-5 wt.% Ru catalysts, while the present work leads to comparable conversion and selectivity results with a much lower Ru loading, 1 wt.%.
As reported by Yan et al. [25] (entry 1) a high LA conversion and GVL yield has been obtained with a 5 wt.% Ru/C catalyst using methanol as a solvent at 130 • C under 12 bar H 2 in 160 min reaction time. Although the authors do not give details about the catalyst support, they seem to have used a commercial one. In any case, information on the carbon properties is not provided.
Manzer [19] (entries 2 and 3) has studied the LA hydrogenation to GVL as the first step in the production of α-methylene-γ-valerolactone (MeMLB) at 150 • C in dioxane. The authors prepared the catalyst by incipient wetness impregnation with a metal salt solution, followed by reduction with H 2 (400 • C, 2 h), but neither details about the properties of the carbon used as support, nor the catalyst dosage are detailed. 100% LA conversion and 97% selectivity to GVL were obtained in 240 min using 34 bar H 2 (entry 3) .
In the work of Ruppert et al. [31] (entry 4), the catalyst was prepared with the commercial activated carbon C-DARCO, but the authors did not give information about surface area or surface chemistry (it is probably a support with around 1600 m 2 /g). LA hydrogenation was carried out in water at 190 • C and 10 bar H 2 for 60 min using a relatively low S/C ratio. In these conditions, although LA conversion was high, the specific reaction rate was low.
A recent work of Jędrzejczyk et al. [24] has focused on the nature of the carbon material used as support for Ru/C catalysts (entry 5). These authors have used four commercial carbon materials with surface area ranging from 649 to 973 m 2 /g and some derived samples with different grain size. Reaction conditions are 190 • C, 10 bar H 2 , 60 min, and S/C ratio of 58. The study has included a good characterization of supports and catalysts, and some differences in acidic properties are reported for them. However, the authors attribute the observed differences in catalytic activity to a different metal-support interaction strength related to defects on the surface of the carbon material and the type of Ru species. In their study, the most active catalyst is the one with the highest number of defects, which, according to the authors, stimulate the strong interaction of the metal with the support.
Finally, the work of Song et al. [32] deals also with Ru/C catalysts prepared with quite different carbon materials (an AC from olive stones, high surface graphite, carbon nanotubes, and reduced graphene oxide). Entries 6 and 7 show the data corresponding to the highest and lowest LA conversion achieved. With all the studied catalysts, excepting Ru/AC, selectivity to GVL is 80%. The differences in the catalysts' behavior are attributed to differences in the Ru particle size (2.9 nm in Ru/AC and 1.2 nm in Ru/rGO). The reported catalytic behavior is very good.
This comparison with literature results indicates that the catalysts reported in the present work show, with much lower Ru content, a comparable or in some cases even better catalytic behavior, with a high specific reaction rate and a very high selectivity to GVL (entry 8). This good behavior is probably due to the properties of the carbon materials used as support, but also to the selected reaction conditions.
To better analyze the effect of the carbon properties of these supported ruthenium catalysts, and also to perform LA hydrogenation in milder and less energy-demanding reaction conditions, a second set of catalytic activity tests has been carried out at 70 • C.
The obtained results (Figure 7) show that, at this lower temperature, the four catalysts present a significantly different behavior, which would be related to the carbon support properties. Thus, at high reaction temperature, the four catalysts give a high LA conversion, close to 100%, but as the reaction temperature decreases, the particular features of the catalysts show their influence in the process. Compared to the results obtained at 170 °C, LA conversion at 70 °C is only slightly lower in the case of Ru R /SA-30 and Ru R /WV-1100 catalysts, but it is about 15% and 30% lower for Ru R /LC and for Ru R /ASC catalysts, respectively (these changes are also observed in the rs values). As the LA conversion at 70 °C for two catalysts is only slightly lower than at 170 °C, it can be assumed that the effect of temperature on the hydrogen diffusion in the solvent is not very important [25]. The poorer LA conversion and selectivity to GVL shown by the biomass-derived carbon-supported catalysts suggests that the support acid- Compared to the results obtained at 170 • C, LA conversion at 70 • C is only slightly lower in the case of Ru R /SA-30 and Ru R /WV-1100 catalysts, but it is about 15% and 30% lower for Ru R /LC and for Ru R /ASC catalysts, respectively (these changes are also observed in the r s values). As the LA conversion at 70 • C for two catalysts is only slightly lower than at 170 • C, it can be assumed that the effect of temperature on the hydrogen diffusion in the solvent is not very important [25]. The poorer LA conversion and selectivity to GVL shown by the biomass-derived carbon-supported catalysts suggests that the support acidity plays a relevant role in the process, in agreement with [14,30,47,48]. It can be assumed that if the support acidity is lower, as in the case of Ru R /ASC and Ru R /LC, the dehydration step required to convert LA into GVL is hindered (see reaction scheme in Figure 2).
Then, these results highlight the important influence of the surface chemistry of the carbon supports on the catalytic performance of the Ru/C catalysts under mild conditions, being the acidity of the catalysts that favors the LA hydrogenation and selectivity to GVL. It can be then concluded that the reaction temperature can be lowered, maintaining a high activity if a carbon support with a suitable acidity is chosen.
It must also be pointed out that at 70 • C the Ru R /WV-1100 catalyst keeps a GVL yield relatively close to the one obtained at 170 • C (75% vs. 87%), showing good catalytic performance at this relatively low reaction temperature.
Reported studies using such a low temperature for LA conversion to GVL are scarce. For example, Yao et al. [29] performed the reaction at 70 • C, but at 30 bar H 2 for 120 min with a commercial 2 wt.% Ru/C catalysts (1850 m 2 /g). Using an S/C ratio of 348, they obtained an r s value of 95 h −1 (much lower than the one obtained for most of the catalysts in this work (Figure 7)). In the work of Galletti et al. [49], the reaction was also carried out at 70 • C and 30 bar H 2 , but acid agents were also used to increase the activity. They found that the combination of Amberlyst A70, or niobium phosphate, with a commercial 5 wt.% Ru/C catalyst (880 m 2 /g) led to a catalytic system that produced a high GVL yield (99%) in 3 h reaction time and r s value of 558 h −1 .
The use of a reaction temperature as low as 70 • C is a promising alternative to increase the energetic efficiency of the process. The optimization of the operation conditions, together with the modification of the catalyst acidity to obtain more efficient bifunctional catalysts, could improve the good catalytic results already achieved in the present study for this reaction.

Materials
The carbon materials used as supports to prepare the Ru/C catalysts are two commercial activated carbons from MeadWestvaco (now Ingevity Corporation, North Charleston, SC, USA), named SA-30 and WV-1100, and two activated carbons prepared from biomass residues by activation with phosphoric acid (H 3 PO 4 ). ASC was prepared by hydrothermal carbonization from almond shells in conditions, such as those reported in [50]. Milled almond shells were mixed with an aqueous H 3 PO 4 solution (25 wt.%) in a stainless-steel autoclave (H 3 PO 4 /almond shell wt. ratio = 1) and heated at 200 • C for 24 h. The resulting hydrochar was then heated (450 • C, 2 h, N 2 50 mL min −1 ). Afterward, the sample was washed with distilled water at 65 • C and dried at 110 • C overnight. LC was prepared from Luffa Cylindrica, a lignocellulosic plant original from the region of Metidja in northern Algeria (commonly called loofah sponge, by chemical activation using 85% phosphoric acid, in a H 3 PO 4 /LAC wt. ratio of 5/1, at 550 • C, 1 h, N 2 60 mL min −1 . The obtained carbon was washed many times with distilled water, using a Soxhlet, until a neutral pH and dried (105 • C, 24 h).

Catalysts Preparation
Carbon supported Ru catalysts were prepared by impregnation using an aqueous solution of RuCl 3 (18 mL solution/g of support) of the appropriate concentration (0.56 mmol/L) to get 1 wt.% Ru (nominal content). The mixture was mechanically stirred for 24 h and then ultrasonicated for 3 h. Finally, the solvent was removed in a stove at 115 • C.
The as-prepared catalysts are named Ru/Sup, where Sup is the name of the carbon material used as support as indicated before. If they have been submitted to a reduction treatment (250 • C, H 2 (75 mL/min), 4 h) the name includes the letter "R" as a superscript (Ru R /Sup).

Characterization of Supports and Catalysts
The surface area and porosity of supports and catalysts were characterized by gas adsorption, N 2 at −196 • C and CO 2 at 0 • C [51,52], using a volumetric Autosorb-6B apparatus from Quantachrome. Before the analysis, the samples were degassed at 250 • C for 4 h. The apparent surface area (S BET ) was calculated by application of the BET equation to the nitrogen adsorption data [53]. The Dubinin-Radushkevich equation was applied to the N 2 and CO 2 adsorption isotherms to determine the total micropore volume (V DR, N2 , micropores smaller than 2 nm) and the narrow micropore volume (V DR, CO2 , micropores smaller than 0.7 nm), respectively [54]. The difference between V DR, N2 and V DR, CO2 is known as V super micro [51,53] . The mesopore volume (V meso, pores of size between 2 and 20 nm) was estimated as the difference of the volume of N 2 adsorbed as a liquid at P/P 0 = 0.9 and at P/P 0 = 0.2 [53], while the total pore volume (V T ) corresponds to the volume of N 2 adsorbed as a liquid at P/P 0 = 0.99.
The surface chemistry of the supports was studied by temperature-programmed desorption (TPD) in the following conditions: about 10 mg of the sample were heated at 20 • C /min in He flow (100 mL/min) up to 950 • C. The equipment used consists of a thermobalance (TA-SDT Q600) coupled to a mass spectrometer (Thermostar, Balzers, Pfeiffer Vacuum, Asslar, Germany) allowing the simultaneous record of weight loss and analysis of evolved gases (CO 2 , CO, and H 2 O).
X-ray photoelectron spectroscopy (XPS, VGMicrotech Multilab ESCA-3000 spectrometer, ThermoFisher Scientific, Waltham, MA, USA) was used to characterize the surface chemical composition of the catalysts and the electronic states of the supported Ru species. Ru 3p electrons were analyzed (instead of 3d ones) since Ru 3d and C 1s electrons have similar B.E. [42,55].
Transmission electron microscopy (TEM, JEM-2010 with digital camera GATAN ORIUS SC600, (JEOL, Tokyo, Japan)) was used to analyze the size and distribution of the Ru particles in the reduced catalysts. Particle size was measured with the ImageJ program.

Catalytic Activity Tests
In a typical LA hydrogenation catalytic test 0.445 mL LA, 130 mg catalyst, and 25 mL distilled water were mixed in the Teflon insert of a 100 mL stainless-steel batch reactor (Parr Instrument Company, Moline, IL, USA). After purging with He and under mechanical stirring (500 rpm), the H 2 pressure was increased up to 15 bar at the selected reaction temperature (170 • C or 70 • C) and then, the system was kept in these conditions for 1 h. Afterward, the reactor was cooled down to room temperature and depressurized, and then the solid and liquid phases were separated by sedimentation. The supernatant solution was collected and filtrated with a syringe (filter of 0.45 µm) to be analyzed by High-Performance Liquid Chromatography (HPLC, Agilent 1100 series, Santa Clara, CA, USA), equipped with UV-Visible Variable Wavelength Detector) coupled to a Mass Trap Spectrometer (MS) with Ion Trap Analyzer (Agilent 1100 Series LC/MSD Trap SL, Santa Clara, CA, USA). The column used was ZORBAX Eclipse XDB-C8 (Agilent, Santa Clara, CA, USA) (dimensions 4.6 × 150 m 2 ). More details can be found in Table S3.
Blank experiments, without catalyst and with the carbon supports, were carried out in the same conditions and following the same procedures.
After catalytic tests at 170 • C, the liquid and solid phases were separated by sedimentation, and then the solution was removed and filtered using a syringe. In each case, the catalyst was washed with three portions (25 mL) of distilled water and dried (115 • C, 24 h).
Then, a second catalytic run was performed with the recovered catalyst under the same reaction conditions.

Product Analysis
LA conversion, GVL yield, selectivity to GVL and specific reaction rate were calculated as follows: LA conversion (%) = n LA,initial − n LA,final n LA,initial × 100 (1) GVL yield (%) = n GVL n LA,initial × 100 (2) GVL selectivity (%) = Yield Conversion × 100 Specific reaction rate, r s h −1 = n LA,initial − n LA,final n Ru × t × 100 (4) where n LA,initial represents the number of moles of LA added to the reactor and n LA , final and n GVL represent the number of moles of LA and GVL in the liquid reaction product, respectively. n Ru represents the number of moles of the active phase in the catalyst and t is the reaction time. Calibration was carried out using several standard and independent solutions of LA and GVL.

Conclusions
Active and selective Ru/C catalysts for the hydrogenation of levulinic acid to gammavalerolactone have been prepared using as supports two commercial and two biomassderived activated carbons (ACs).
The four Ru/C tested catalysts present a very high LA conversion and GVL yield when the reaction is performed at 170 • C. Thus, in these conditions the different support properties do not lead to differences between the catalysts, although they seem to be appropriate to achieve efficient catalysts. The prepared catalysts have shown a catalytic behavior superior to that of many of the Ru/C catalysts reported in the literature.
At lower reaction temperature (70 • C), noticeable differences between the catalysts' behavior have been observed. The two catalysts prepared with the commercial ACs show better performance, which has been attributed to the higher acidic character of these materials. In particular, the catalyst prepared with the WV-1100 AC shows 98% LA conversion and 77% selectivity to GVL, with a specific reaction rate (r s ) value very close to that obtained at 170 • C. These results highlight the importance of the support's surface chemistry and allow us to conclude that the reaction temperature can be lowered, maintaining a high activity if a carbon support with a suitable acidity is chosen. This also leads to the next research step, focused on the functionalization of the biomass-derived ACs to increase their acidity.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/catal11050559/s1, Table S1. Amount of CO 2 and CO from TPD experiments and calculated total oxygen content in the Ru/C catalysts (as prepared and reduced); Table S2. Mean nanoparticle size and the error associated; Table S3. Details of the HPLC analysis; Figure S1. TPD spectra of carbon supports and reduced catalysts: CO 2 and CO evolution (be aware of the different scale in yaxis); Figure S2. Pore volume data comparison for supports and reduced catalysts.