Hydrothermal Transformation of Glycerol to Lactic Acid in Alkaline Medium Using Cu Catalysts Obtained from Hydrotalcite-Like Precursors

The conversion of glycerol into lactic acid in alkaline medium using heterogeneous catalysts has been highlighted and has proved to be an efficient alternative to the conventional fermentative route. This work investigated the production of lactic acid from glycerol in alkaline medium using copper catalysts obtained from hydrotalcite-like precursors in a continuous flow reaction system and the effect of different copper loading on catalytic behaviors in terms of yield and selectivity to lactic acid. The catalysts were synthesized by the coprecipitation method and characterized by XRF, XRD, N2 adsorption–desorption, H2-TPR, CO2-TPD, and the copper dispersion (i.e., metal atoms exposed) was determined by N2O oxidation. The reaction was performed for 30 h at 240 °C, 35 atm, using space velocity (WHSV) of 2 h−1, solution of 10 vol% glycerol, and NaOH/glycerol molar ratio of 0.75. Although the 30CuHT catalyst presented higher BET surface areas, Cu dispersion, and basicity than the 20CuHT catalyst, the best results in terms of glycerol conversion (96.5%) and yield (64%) to lactic acid were obtained in the catalytic test performed with the 20CuHT catalyst. This result may be related to the reaction steps that occur in the liquid phase in the presence of the hydroxyl group (OH-\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\text{OH}}^{-}$$\end{document} Furthermore, the high content of Cu may favor the hydrogenation of the reaction intermediates (2-hydroxypropenal and pyruvaldehyde), increasing the formation of 1,2-propanediol and consequently reducing the yield of lactic acid. The catalytic activity of the reduced and non-reduced catalysts was investigated, and the results showed that copper oxide also has catalytic activity. However, the reduction of copper oxide provides better results.


Introduction
The increase in greenhouse gas emissions due to human activities on the planet has been a concern for researchers and authorities. Energy production is the main contributor to the release of greenhouse gases by humanity, particularly the emission of CO 2 into the atmosphere from the use of fossil fuels. Projections of how the global energy system will develop over the next century are critical for assessing future climate change caused by humanity [1]. Rising fuel costs, uncertainty about energy supplies, dependence on foreign energy sources, and concern about global climate change and air quality make renewable energy alternatives, such as that produced from biomass, more attractive [2].
Biodiesel is an alternative to current fossil fuels [3]. Thus, it contributes to reducing greenhouse gas emissions [4], as it significantly reduces toxic emissions and can be used in the diesel engine without any engine modification, indicating that it has physical and chemical properties comparable to conventional diesel. Biodiesel exhaust gases during combustion contain less carbon monoxide, hydrocarbons, particulates, and sulfur dioxide than diesel oil [5].
Biodiesel usually is produced through the transesterification process of vegetable oils or animal fats. This process generates glycerol as a byproduct, approximately 10% (w/w) [6]; therefore, with the increase in biodiesel production, large surpluses of glycerol are expected in the world market. Thus, it is necessary to find an appropriate destination for the surplus glycerol, minimizing environmental problems generated by excess glycerol and adding value to the productive chain of biodiesel because the cost of biodiesel production is still much higher than that of diesel fuel [7]. According to Cannilla et al. [8], the impact of glycerol on the global economy of biodiesel has become crucial, and today, more than a byproduct, it can be considered a co-product. The glycerol market is expected to exceed $3 billion by 2022, with an estimated gain of 7.9% from 2015 to 2022. According to Fan and Burton [9], the cost of biodiesel production could drop significantly if the value of crude glycerol increases. This is one of the main motives for the large-scale search for new ways to utilize crude glycerol [10].
Glycerol is a versatile raw material for producing various chemicals, polymers, and fuels. Among these products, we highlight lactic acid (LA), given its growing application in the food, pharmaceutical, and polymer industries. In the latter sector, polylactic acid (PLA) has emerged as a biodegradable plastic that could replace polyethylene terephthalate (PET) in many packaging, textiles, and healthcare products. The LA market size was estimated at 3.3 × 10 5 tons by 2015 and is estimated to expand by 10% annually until 2025 [11]. This growing estimate of LA demand follows the report by Global View Research (California, USA), which indicates that the production of PLA should drive the increase in the demand for LA in the coming years.
The hydrothermal transformation of glycerol to LA in an alkaline medium presents an alternative to conventional fermentation. The fermentative route produces pure isomers of LA (L(+) or D(−)) that could be obtained from the appropriate strain of LA bacteria [12], while chemical synthesis always results in a racemic mixture of LA [13]. The production of LA by hydrothermal conversion of glycerol in an alkaline medium was first described by Kishida et al. [14]. According to the authors, the first step consists of the dehydrogenation of glycerol to glyceraldehyde, and then the glyceraldehyde undergoes dehydration, forming 2-hydroxypropenal, which through a molecular rearrangement known as keto-enol tautomerization produces pyruvaldehyde. Subsequently, pyruvaldehyde is converted to LA by a mechanism similar to the rearrangement of benzylic acid [14]. This homogeneous catalysis process requires high temperatures, which could promote the degradation of LA and pyruvaldehyde, reducing the selectivity of LA [15]. This problem could be overcome by employing heterogeneous catalysts in the presence of a base.
Roy et al. [16] were one of the first to evaluate the hydrothermal transformation of glycerol in alkaline medium in the presence of heterogeneous catalysts at low temperatures. This study observed that using Cu catalysts and NaOH provides a promising one-pot, low-temperature route for producing LA from glycerol. Thus, numerous investigations were realized, focusing on studying the combined use of heterogeneous catalysts in alkaline medium [16][17][18][19]. In the study by Xu et al. [20], DFT (density functional theory) calculations showed that NaOH reduces the energy barrier of converting glyceraldehyde to LA. Additionally, the authors confirmed that glyceraldehyde is an intermediate of the reaction and inferred that NaOH mainly facilitates the dehydration of glyceraldehyde due to its deprotonation capacity.
Hydrotalcites, a family of anionic clays, are minerals with double-layer structure compounds of aluminum and magnesium hydroxides with interlayer spaces containing exchangeable anions, usually carbonates ( CO − 3 ). The total or partial substitution of divalent (Mg 2+ ) and/or trivalent (Al 3+ ) cations by other cations results in materials with isomorphous structures known as hydrotalcite-like compounds (HTLCs) [21]. When HTLCs are subjected to heat treatments, they form a mixture of oxides with excellent properties, such as high specific area, good thermal stability, high metal dispersion after reduction, and basic properties. These properties have led such materials to receive much attention and be applied in several areas [22].
Shen et al. [23] evaluated the effect of temperature and NaOH/glycerol molar ratio using a batch reactor and palladium catalysts supported on hydroxyapatite (HAP). The results showed that the successive increase of the reaction temperature from 190 to 230 °C presented successive increases in the selectivity to LA. However, when the reaction temperature increased to 250 °C, a slight reduction in the selectivity to LA occurred with a concomitant increase in the selectivity to formic acid, which was associated with parallel reactions of C-C splitting or LA oxidative cleavage. Likewise, favoring parallel reactions or degradation of LA were observed when the catalytic tests were performed with a NaOH/glycerol molar ratio greater than 1.1. Similar results were obtained by Qiu et al. [24], where they observed an increase in glycerol conversion with an increase in the NaOH/glycerol molar ratio from 1.0:1 to 1.3:1, associated with a loss of selectivity to LA. The extensive study performed by Qiu et al. [24] using nickel catalysts supported on HAP observed that increasing the temperature from 180 to 240 °C promotes an increase in the glycerol conversion but is accompanied by a reduction in the selectivity to LA. On the other hand, the study by Yin et al. [25] using nickel catalysts supported on graphite showed that increasing the NaOH/glycerol molar ratio from 1:1 to 1.5:1 and increasing the temperature from 180 to 230 °C favors the conversion of glycerol to LA. However, when the reaction temperature increases to 250 °C, a rapid reduction in selectivity to LA occurs. This reduction in selectivity may be related to the fact that nickel has good catalytic activity in the cleavage of the C-C, O-H, and C-H bonds of oxygenated organic compounds [26].
Most studies found in the literature on the transformation of glycerol to LA were carried out in batch reactors. Some exceptions are the research presented by Zhang et al. [27] and Shimanouchi et al. [28], which employed a continuous flow reactor system with alkaline homogeneous media without using catalysts. In previous studies of our group [29,30], employing a continuous flow reaction system, it was observed that reactions carried out with NaOH/glycerol molar ratios greater than 1 show a yield of close to 100% in LA. Therefore, for a more sensitive evaluation of the catalyst performance in the reaction, it was decided to use a NaOH/ glycerol molar ratio of 0.75. In addition, the temperature of 240 °C was selected to evaluate the different catalysts since this temperature presented the best results [29,30].
The objective of this study is to investigate the production of LA from glycerol in alkaline medium, using Cu catalysts derived from hydrotalcite-like compounds, in continuous flow reaction system for 30 h. Furthermore, we intend to evaluate the effect of different Cu loading on the catalytic performance and the effect of using the non-reduced catalyst in the reaction.

Characterization
Chemical composition analysis of the catalysts was performed by X-ray fluorescence (XFR) using a Rigaku Primini 1 3 spectrometer equipped with a palladium tube as a radiation source.
Textural properties were obtained from the N 2 adsorption/desorption isotherms at -196 °C using a Micromeritics TriStar 3000 instrument. Before the analysis, the sample was outgassed under a high vacuum at 250 °C for 24 h. The specific surface area was calculated using the Brunauer-Emmett-Teller (BET) method. Pore volume and pore diameter by the Barrett-Joyner-Halenda (BJH) method. Adsorption data were used to calculate the specific area of the samples at a relative pressure range of 0.06-0.20. The pore volume was estimated by BJH adsorption cumulative volume of pores between 20 Å and 600 Å diameter.
X-ray powder diffraction (XRD) patterns were collected in a Rigaku Miniflex using CuK α as a radiation source (30 kV and 15 mA). The angular interval of 5 to 90° was varied in 0.05° steps, using a 1 s counting time per step in continuous scan mode. Diffractograms obtained were compared with the Joint Committee on Powder Diffraction Standards (JCPDS) database. XRD analyses of the ex situ reduced catalysts were performed, and spent catalysts were analyzed after drying at 110 °C overnight and without further treatment after the reaction.
The analysis of temperature-programmed reduction with hydrogen (H 2 -TPR) was employed to evaluate the reducibility of the catalysts. This analysis was carried out in a quartz reactor at atmospheric pressure using a traditional apparatus equipped with TCD (thermal conductivity detector). Before the reduction, the samples were dehydrated at 150 °C under flowing Ar for 30 min. After reaching room temperature, the samples were reduced up to 1000 °C (10 °C min −1 ) under a flow of 30 mL min −1 of a mixture with 2% H 2 /Ar.
The catalyst basicity was evaluated by temperature-programmed desorption of carbon dioxide (CO 2 -TPD) employing a system equipped with a mass spectrometer QMG-220 (Pfeiffer). Primarily, the catalyst was reduced using a mixture with 2% H 2 /Ar up to 450 °C (10 °C min −1 ) with an isothermal at 450 °C for 30 min, then cooled to room temperature. The CO 2 adsorption was carried out at room temperature with a mixture of 10% CO 2 /He (30 mL min −1 ) for 30 min, followed by purging with He for 60 min. The desorption of chemisorbed CO 2 was performed by heating up to 1000 °C ( 20 °C min −1 ) under pure He flow (30 mL min −1 ). The ratio m/z = 44 was used to determine the amount of desorbed CO 2 .
The dispersion of metallic copper ( D Cu ) was obtained following similar procedures as shown by Gervasini and Bennici [32] and López-Suárez et al. [33]. Firstly, the catalysts were reduced using the same conditions employed for CO 2 -TPD analysis. Subsequently, the selective oxidation of the copper surface to Cu 2 O was performed using a 9.9% N 2 O/He mixture with a flow rate of 30 mL min −1 at 50 °C for 1 h. Finally, the surface copper oxide (Cu 2 O) was reduced by heating up to 1000 °C (10 °C min −1 ) under flowing 2% H 2 /Ar (30 mL min −1 ). Using a TCD, it was possible to quantify the H 2 consumed for the reduction of Cu 2 O and thus determine the amount of surface Cu°. The dispersion is calculated as a ratio between the amount of surface copper and total copper obtained by XRF analysis (Eq. 1).
From dispersion data, which considers that N 2 O oxidizes only the copper surface, the Cu metal surface area per unit weight of the catalyst S where D Cu , A V , L Cu , W Cu , and N A , are dispersion (%), Avogadro's number, Cu content (wt%) in the catalyst, atomic weight of Cu (63.5 g mol −1 ), and number of surface Cu atoms in unit surface area (1.47 × 10 19 atoms m −2 [34]), respectively [35].
From the dispersion data, it is possible to estimate the average crystallite sizes of the Cu according to Anderson [36] where d Cu is the crystallite size (in nm), V m is the Cu atomic volume (0.0118 nm 3 ), and A m is the surface area of a single Cu atom (0.068 nm 2 ) [37].

Catalytic Tests
The catalytic tests were performed in a continuous system using a fixed-bed reactor of Inconel 625, with an internal diameter of 0.5 cm. The catalysts in the form of pellets were reduced in situ at a temperature determined by TPR analysis (i.e., 450 °C with an isothermal at 450 °C for 30 min) using a mixture of 30 vol% H 2 /N 2 (90 mL min −1 ). The 20CuHT catalyst was also evaluated without reduction. The catalytic tests were performed for 30 h at 240 °C with 35 atm of pressure. The pressure in the system was monitored through manometers and controlled by a Back Pressure Regulator (BP-6, GO Regulator). The initial 2 h were discarded due to the purge of the reaction system before taking the 30 h reaction. The period between 7 and 24 h of reaction was not analyzed (overnight). Although the first 2 h of reaction were discarded, the results were much more stable after the overnight period, allowing a consistent analysis of the results.
The catalytic evaluations were performed using an aqueous solution of glycerol 10 vol% and NaOH/glycerol molar ratio equal to 0.75. The reagents were injected into the reactor with an HPLC pump (Eldex 1SAM) under the feed rate of 0.041 mL min −1 . The catalyst mass employed was 1.25 g resulting in a weight hourly space velocity (WHSV) of 2 h −1 .
The catalyst bed is distributed approximately 6 cm in the reactor. Thus, the residence time of glycerol is 29 min. The liquid phase was collected hourly and subsequently analyzed by a Shimadzu Prominence high-performance liquid chromatography (HPLC) equipped with a Bio-Rad Aminex HPX-87H column, using 0.01 M H 2 SO 4 as eluent at 0.6 mL min −1 , and 60 °C, and both UV and refractive index detectors. The samples for analysis were diluted using an aqueous solution of 0.01 M H 2 SO 4 . The glycerol conversion, selectivity, and yield to products were calculated according to the equations presented in our previous study [30].

Catalytic Tests
According to the literature, the alkaline medium is a factor of great importance in the hydrothermal reactions of conversion of glycerol into lactic acid. Sharninghausen et al. [38] verified in a test without catalyst carried out in a batch reactor using a glycerol/KOH molar ratio of 1:1.1 at 115 °C that there was no product nor byproduct formation, and the glycerol conversion was practically null. A catalytic test without a catalyst was performed by our group in a fixed-bed reactor using a NaOH/glycerol molar ratio equal to 1 and 240 °C/35 atm. The results showed no formation of products or byproducts, and the conversion of glycerol was also negligible [29]. On the other hand, in the reaction carried out with a copper-based catalyst (Cu/ZnO) without the presence of the base (NaOH), there was no formation of LA, but a yield of 8% was obtained for 1,2-propanediol (1,2-PDO), and the glycerol conversion was 12% [39]. Thus, it is observed that the concomitant use of an alkaline medium and catalyst is necessary to obtain the conversion of glycerol to LA under moderate temperature and pressure conditions since the catalyst promotes the conversion of glycerol due to the metallic sites decreasing the use of high temperatures in the reaction [40]. Figure 1 shows the glycerol conversion obtained during the 30 h of reaction. It is observed that after an initial period, the results were relatively stable, especially after the overnight period. The glycerol conversion showed relatively high values independently of the Cu concentration in the catalysts. After the overnight period, all reduced catalysts showed average conversions in the 89 to 97% range, while the non-reduced catalyst (20CuHTS) showed the lowest average conversion (78.1%). Thus, it is demonstrated that the copper oxide phase has catalytic activity. However, reducing the copper oxide to metallic Cu is necessary to obtain higher glycerol conversion. Figures 2A, B shows the selectivities to LA and 1,2-PDO, respectively, obtained during the 30 h of reaction. The catalytic tests performed with the 20CuHT and 20CuHTS catalysts showed the highest selectivities to LA during the reaction (~ 70%). On the other hand, the catalytic tests performed with the 10CuHT catalyst showed the lowest average values (25-30 h) of selectivity to LA (27.4%). The catalytic test carried out with 30CuHT catalyst presented values of selectivity to LA similar to those obtained using the 20CuHT and 20CuHTS catalysts after the overnight period. Thus, the increase in the Cu loading in the catalyst (30CuHT) did not increase the selectivity for LA. However, progressive increases in the Cu loading favor the formation of 1,2-PDO, which is the only identified byproduct of the reaction, Fig. 2B. The average selectivity (25-30 h) to 1,2-PDO obtained by 10CuHT, 20CuHT, and 30CuHT catalysts was 4.3%, 19.1%, and 22.0%, respectively. The catalytic test performed with the non-reduced catalyst (20CuHTS) showed a lower selectivity to 1,2-PDO than the 20CuHT catalyst. This result indicates that the secondary reaction of the formation of 1,2-PDO presents a greater dependence on metallic sites. The literature widely accepts that 1,2-PDO is the main byproduct of the glycerol transformation into LA in an alkaline medium when performed concomitantly with heterogeneous catalysts. The 1,2-PDO could be produced from the hydrogenation on the metallic surface of 2-hydroxypropenal and pyruvaldehyde, which are two intermediates of the glycerol conversion into LA [41,42].
The LA yield (Fig. 3A) showed a behavior similar to LA selectivity but at lower levels since the yield depends on glycerol conversion. Thus, it is observed that the catalytic test performed with the 20CuHT catalyst was the one that presented the highest yields because of the good selectivity and the high values of glycerol conversion. Although the 20CuHTS catalyst showed LA selectivities similar to the 20CuHT catalyst, the glycerol conversion was low, resulting in a lower LA yield. Thus, it is necessary to reduce the catalyst to obtain higher yields of LA. The catalytic test performed with the 10CuHT catalyst showed lower performance than the others during the entire reaction time, with a maximum LA yield of 25%, approximately. The maximum 1,2-PDO yield (Fig. 3B) was obtained with 30CuHT (20%), while the catalytic test performed with the 10CuHT catalyst showed the lowest yields among all catalysts. Thus, globally, the catalytic performance of the catalysts on the reaction of transformation of glycerol to LA can be classified in the following order: 10CuHT < 20CuHTS < 30CuHT < 20CuHT.
Some studies have evaluated the effect of different metallic contents on catalysts. Shen et al. [43] evaluated CuPd x catalysts (x = 0, 0.5, 1, 2, 3, 4, 5, and 100), where x equals mol Pd in 100 mol Cu; in this study, the highest yield at LA (93.3%) was obtained with the CuPd 2 catalyst. Qiu et al. [24] observed a similar behavior using catalysts based on nickel supported on HAP. Different Ni x /HAP catalysts were prepared with x = 0.1, 0.2, 0.3, and 0.4 (x, mole of Ni to the 100 g HAP), and the highest LA yield (87.2%) was obtained in the catalytic test performed with Ni 0.2 /HAP catalyst. The study by Yin et al. [25] obtained the same behavior, the selectivity to LA first increases and then decreased with increasing nickel loading on graphite-supported catalysts. Four different Ni x /graphite catalysts were prepared with x = 0.1, 0.2, 0.3, and 0.4 (x, mole of Ni to 100 g of graphite). The highest yield to LA (87.6%) was obtained using Ni 0.3 / graphite catalyst.

Catalyst Characterization
The content of oxides in catalysts calcined at 500 °C is shown in Table 1. All catalysts presented real loading of CuO lower than the nominal contents (10, 20, and 30 wt%). The differences could mainly be attributed to losses during preparation due to metal salt precipitation problems and the highly hygroscopic character of metal nitrates. Although the CuO content did not reach the desired value, this does not represent a problem for the synthesis of HTLCs, since the fraction of trivalent cations (M 3+ /(M +3 + M 2+ )) is within the range of 0.20 and 0.33, where pure hydrotalcite-like compounds could be obtained [44][45][46]. Table 1 shows the BET surface areas, pore volumes, and pore diameter obtained by the BJH method of the catalysts before and after calcination at 500 °C for 3 h. The calcination of the HTLCs resulted in an expressive increase in the specific area. This increase in BET surface areas with calcination agrees with the literature since the thermal decomposition of HTLCs leads to high surface area Mg-Al oxides [47]. Therefore, BET surface areas of calcined HTLCs become significantly larger than their precursors. The increased area may be due to the removal of CO 2 during the decomposition of HTLCs, increasing the porosity of the mixture of oxides obtained. Although some works report a decrease in the BET surface area due to the increase in the Cu loading [48,49], it is observed in Table 1 that the catalyst composition does not significantly influence the textural properties of the calcined catalysts. Adsorption isotherms of N 2 for all samples exhibited the type IV pattern with H3-type hysteresis loops for the desorption isotherms ( Figure S1A, B), typical of mesoporous materials (20 to 500 Å) [50]. The H3-type hysteresis loop is observed in aggregates of platelike particles that give rise to slit-shaped pores [51]. Thus, this indicates the presence of narrow slit-like pore particles with irregularly shaped internal voids and wide size distribution [52] (Figure S2A, B). Figure 4A shows the XRD patterns of HTLCs with diffraction peaks at 2θ = 11.3°, 22.8°, 34.5°, 38.6°, 45.5°, 60.3°, and 61.5°, which can be attributed to the (003), (006), (009), (015), (018), (110) and (113) crystalline planes, respectively, and this corresponds to a well-crystallized HTLC (JCPDS 22-0700). The absence of other phases suggests that Cu 2+ has isomorphically replaced Mg 2+ cations in the brucitelike layers [53][54][55], and this may be related to the similarity between the ionic radii Cu 2+ (0.73 Å) and Mg 2+ (0.65 Å) [56]. However, comparing the three diffractograms, a slight loss of crystallinity is observed with the increase of the Cu  loading, as indicated by both the loss of intensity and sharpness of (110) and (113) reflections observed around 2θ = 60° and 62°. According to Kannan et al. [56], this is expected due to Jahn-Teller distortion at higher copper loading, leading to poor long-range ordering. With calcination at 500 °C for 3 h, it is observed that the lamellar structure of HTLCs was destroyed, leading to the formation of mixed oxides of MgO, Al 2 O 3, and CuO [57], Fig. 4B. The destruction of the lamellar structure of the HTLCs resulted in the formation of the MgO periclase phase, with diffraction peaks at 2θ = 36.9°, 42.9°, and 62.3° (JCPDS 45-0946). No diffraction peaks corresponding to copper and aluminum oxide species were observed. This result may indicate that these species are well dispersed on the MgO surface or inserted into the MgO structure to form a solid solution [53,56].
XRD analysis was also performed to characterize the catalysts after reduction at 450 °C for 30 min ( Figure S3). The diffractograms of the reduced catalysts are practically identical, except for a slight loss of crystallinity with increasing CuO content. Furthermore, a significant similarity is observed between the diffraction patterns of the calcined and reduced samples. The main diffraction peak of metallic Cu (JCPDS 04-0836) (2θ = 43.3°) overlaps the peak of MgO. However, the second peak of greater intensity (2θ = 50.4°) was not observed in any of the three catalysts, suggesting that the catalysts have high metallic dispersion. Similar results were observed by Nagaraja et al. [58]. In their study, the reduced Cu/MgO catalyst (16.04 wt.% Cu) prepared by coprecipitation did not show diffraction peaks of metallic Cu since they could be in the amorphous form or have a high dispersion resulting in undetectable crystallites size by XRD. However, higher copper loading in the catalyst may result in higher Cu° crystallites that may be detectable by XRD [58,59].
XRD analysis on the spent catalysts revealed changes in the crystalline phases of catalysts after 30 h of reaction ( Figure S4). Diffraction peaks associated with the metallic Cu phase with peaks at 2θ = 43.3°, 50.4°, and 74.1° (JCPDS 04-0836) can be observed in the 10CuHT and 30CuHT catalysts. These catalysts had their CuO reduced before the reaction, and only after the catalytic test was it possible to observe the crystalline structure of metallic Cu, suggesting the occurrence of sintering of Cu particles during the catalytic test. However, the 20CuHT catalyst did not show the characteristic peaks of metallic Cu after the reaction period, which may indicate that the sintering process of Cu particles was less significant for this catalyst. After the reaction, all catalysts presented the characteristic peaks of the CuO phase (JCPDS 48-1548), and only the catalysts reduced before the reaction exhibited the Cu 2 O phase (JCPDS 65-3288). Thus, a partial oxidation of metallic Cu to Cu 2 O occurs during the reaction. Although it is possible to form Cu 2 O from CuO reduction during the reaction [20], the absence of Cu 2 O peaks after the reaction for the non-reduced catalyst indicates that Cu 2 O is probably coming from Cu oxidation. The partial oxidation of metallic Cu to Cu 2 O during the glycerol hydrogenolysis reaction was also observed by Mendonça et al. [60]. In addition, other crystalline phases were identified as brucite Mg(OH) 2 (JCPDS 44-1482) and magnesite MgCO 3 (JCPDS 86-2345). Although partial oxidation of the metallic phase occurred during the reaction, this did not result in a loss of catalytic activity since the glycerol conversion and selectivity to LA remained relatively stable after the overnight period; this is in agreement with the literature, as shown by Roy et al. [16], Liu and Ye [42], and Shen et al. [61], the CuO and Cu 2 O phases have catalytic activity for converting glycerol to LA.
Baskaran et al. [62] stated that the memory effect is one of the most well-known properties of hydrotalcite. According to the authors, when the calcination temperature used on the hydrotalcite is mild enough to prevent the formation of spinels, hydrotalcite can regain its original layered structure in an aqueous solution when thermally labile anions are present. This memory effect of hydrotalcite can be observed in the XRD of the unreduced spent catalyst (20CuHTS). However, the CuO phase (JCPDS 48-1548) was segregated from the hydrotalcite structure. This is also in agreement with other works in the literature, which state that from a mixture of Mg-Al oxides derived from a hydrotalcite, it is possible to return the hydrotalcite structure with the rehydration of the mixture [63].
Finally, it was observed that for the spent catalysts reduced before the reaction, two new peaks are formed at 2θ = 8.9º and 17.9º, which increase in intensity with increased Cu loading. Such peaks may originate from the displacement of the first peaks of the hydrotalcite phase. According to Kwon et al. [63], the first three planes ((003), (006), (009)) of hydrotalcite are generated by a lattice formed from interlayer materials. The disappearance or shifting of these planes may indicate that the interlayer content has been modified [63][64][65]. Thus, the peaks at 2θ = 8.9º and 17.9º may be related to hydrotalcite restructuring from the mixture of oxides with a possible change in the composition of interlamellar anions.
In order to determine the reduction behavior of the catalysts, H 2 -TPR analysis was performed (Fig. 5), and reduction degrees are shown in Table 2. CuO has a reduction temperature in the range of 200 to 400 °C [54], which agrees with the reduction profile observed in the catalysts. It is noted that the increase in the Cu loading on the catalysts caused a decrease in the temperature of the maximum reduction peak. According to the literature, peak reduction at lower temperatures could be attributed to the reduction of CuO dispersed on the surface of the catalyst, which has a lower interaction with the support. In addition, dispersed copper offers a larger 1 3 reactive area and a higher concentration of defects. In this way, the reduction of CuO could start at lower temperatures. On the other hand, peak reduction at higher temperatures could be attributed to the reduction of bulk CuO [66][67][68]. Therefore, these results suggest that the increase of copper loading on the catalysts evidences an increase in the metallic dispersion. However, this behavior is the opposite of the expected since the increase in CuO loading tends to promote particle agglomeration [69,70]. The reduction degree obtained during TPR experiments has been estimated from H 2 consumption, considering the reduction of Cu 2+ to Cu°. The degree of reduction of the catalysts was around 100% for all catalysts. Experimental errors may have been responsible for a reduction degree value slightly greater than 100% (10CuHT).
The basicity of the catalysts was evaluated by CO 2 -TPD analysis, and the desorption profiles are shown in Fig. 6. According to the literature, the area under the curve was divided into three desorption regions: up to 400 °C, 400-600 °C, and above 600 °C, related to low, medium, and high strength basic sites, respectively [71]. A wide range of CO 2 desorption temperatures can be observed, from 50 to 1000 °C, indicating the presence of basic sites of different strengths. However, the catalysts demonstrate a predominance of low-strength basic sites, Table 2. The desorption profiles of 20CuHT and 30CuHT catalysts are similar to those found by Yuewen et al. [72] for hydrotalcite-derived copper catalysts. However, the CO 2 desorption profile exhibited by the 10CuHT catalyst is very similar to the CuMg catalyst synthesized by coprecipitation with about 30% Cu [72], which is probably associated with the highest MgO content among the catalysts. Although the proportion of lowstrength basic sites does not change between catalysts, with increasing the Cu loading on the catalyst, there is a tendency to promote an increase in the proportion of high-strength  The concentration of basic sites of the catalysts (μmol CO 2 g −1 cat ) and their density (μmol CO 2 m −2 ) were calculated by integrating the CO 2 -TPD profile, and the results are shown in Table 2. The 10CuHT catalyst had the highest amount and density of basic sites among the catalysts. This result may be associated with the basic properties of the MgO, which has a more electron-donating character since the Al 2 O 3 content remains relatively constant in the catalysts (Table 1). With the increase of the Cu loading in the catalysts, a reduction of the basic density occurs ( Table 2). This behavior may be related to the MgO content of the catalyst. Table 2 shows the values of Cu dispersion ( D Cu ), and metallic surface area ( S N 2 O Cu ) obtained from the analysis of N 2 O oxidation. With Cu dispersion data, the average size of Cu crystallites can be determined from the Anderson equation (Eq. 3). The 30CuHT catalyst showed the highest dispersion (41.9%), followed by the 20CuHT catalyst (33.5%) and the 10CuHT catalyst (19.9%). The metallic surface area followed the same trend. These results corroborate the data obtained by XRD because, although the 30CuHT catalyst has the highest Cu loading, no peaks were identified and associated with the Cu phases (CuO and/or Cu°) since high dispersion results in small particle size. On the other hand, the 10CuHT catalyst, despite having the lowest dispersion, has the lowest concentration of Cu. Thus, such factors may contribute to the non-detection of Cu by XRD. Multiple factors can alter Cu dispersion on the catalytic surface, including synthesis methodology, calcination temperature, and Cu content. According to Nagaraja et al. [58], the synthesis methodology and the Cu content influence the dispersion of Cuº and the metallic area. Their study showed an increase in dispersion, and the metallic area, with increased Cu loading up to 16.04 wt% in a Cu/MgO catalyst, and above that (37.29 wt%), there was a drastic reduction in metallic dispersion [58]. Reddy et al. [74] observed an increase in Cu dispersion up to 20 wt% Cu loading for Cu/MgO catalyst. Our results agree with those observed by these authors because the increased Cu content on the catalysts resulted in higher metallic dispersions.
The best results were obtained in the catalytic test performed with the 20CuHT catalyst. However, the characterization of the catalysts indicates that the 30CuHT catalyst has a higher BET surface area and basicity besides superior metallic dispersion than the 20CuHT catalyst. The reason why the 20CuHT catalyst showed superior performance might be related to the reaction steps. It is widely accepted that the conversion of glycerol to LA begins with the dehydrogenation reaction of glycerol forming glyceraldehyde, and this step occurs on the metallic sites, while the other steps that lead to the formation of LA occur in solution in the presence of hydroxyl group ( OH − ) [30,40]. Thus, it is observed that content greater than 15 wt% of CuO (20CuHT), approximately, does not present a beneficial effect since the limiting step of the reaction is no longer the dehydrogenation of glycerol. In addition, high Cu content favors the hydrogenation of the reaction intermediates (2-hydroxypropenal and pyruvaldehyde), increasing the formation of 1,2-PDO once copper-based catalysts are efficient in glycerol hydrogenolysis due to their high activity for cleavage of C-O bonds associated with the low activity for cleavage of C-C bonds [75,76].

Conclusions
XRD analysis showed that hydrotalcite-like structures were obtained and that calcination causes the collapse of these structures, generating material with low crystallinity. Furthermore, the non-detection of diffraction peaks referring to the copper phases after reduction was associated with its high dispersion. With the calcination of the hydrotalcite-like precursors, materials with high BET surface area are obtained, contributing to the active phase dispersion. H 2 -TPR analysis showed a reduction in the temperature of the maximum reduction peak with increasing CuO content, which was associated with increased CuO dispersion. This result was corroborated by N 2 O oxidation analysis, which showed an increase in metallic dispersion with increasing copper content on the catalysts.
The average conversion (25-30 h) of glycerol obtained in the catalytic tests was 90 to 96% for the reduced catalysts. In contrast, the non-reduced catalyst (20CuHTS) showed a lower conversion (78.1%), demonstrating that the reduced catalyst (20CuHT) presents superior performance to the nonreduced catalyst, which may be related to the influence of metallic copper on the first step (dehydrogenation) of the transformation of glycerol into LA. However, although the 20CuHTS catalyst does not have metallic sites, the copper oxide phases (CuO and Cu 2 O) have catalytic activity but less than metallic Cu. In the reaction conditions used in this study, in order to obtain good results, copper contents above 6.7 wt% are necessary since the 10CuHT catalyst presented low values of LA yields. However, high concentrations of copper in the catalyst have a negative effect once it favors parallel reactions of 1,2-PDO formation.