Easily Recycled CuMgFe Catalysts Derived from Layered Double Hydroxides for Hydrogenolysis of Glycerol

A series of CuMgFe catalysts with different (Cu + Mg)/Fe molar ratios derived from hydrotalcites were prepared by coprecipitation for the hydrogenolysis of glycerol to 1,2-propanediol (1,2-PDO). X-ray diffraction (XRD), scanning electron microscopy (SEM), transmission electron microscopy (TEM), X-ray photoelectron spectra (XPS), vibrating sample magnetometer (VSM), hydrogen temperature-programmed reduction (H2-TPR), CO2-TPD, and H2-TPD (temperature-programmed desorption of CO2 and H2) were used to investigate the physicochemical properties of the catalysts. The CuMgFe-layered double oxides (CuMgFe-4LDO) catalyst with (Cu + Mg)/Fe molar ratio of 4 exhibited superior activity and stability. The high glycerol conversion and 1,2-propanediol selectivity over CuMgFe-4LDO catalyst were attributed to its strong basicity, excellent H2 activation ability, and an increase in the surface Cu content. The CuMgFe catalysts could be easily recycled with the assistance of an external magnetic field due to their magnetism.

Layered double hydroxides (LDHs), also known as hydrotalcite-like compounds, are a class of anionic clay materials that allow the uniform mixing of different bivalent and trivalent cations. Thermal decomposition of LDHs leads to the formation of mixed oxides with small crystal size, basicity, high dispersion, and large specific surface area.
It has been reported that Cu-based catalysts derived from hydrotalcites are of basicity and dispersed copper particles [8,[24][25][26][27], which could improve the catalytic performance in the hydrogenolysis of glycerol. Geng et al. [28] disclosed that the Cu-Ca-Al catalyst derived from hydrotalcite is more active and selective for the formation of propanediols. Nonetheless, these catalysts suffer from serious difficulties in the recovery and reuse from the perspective of ecological and economical sustainability. Magnetic catalysts can be easily separated by an external magnetic field, reducing the consumption of auxiliary substances, saving energy and time in separation, and bringing significant economic and environmental benefits. However, to the best of our knowledge, there is no detailed understanding of magnetic Cu-based catalysts derived from hydrotalcites in the hydrogenolysis of glycerol.
The aim of the present work is to develop easily recycled Cu-containing catalysts in the hydrogenolysis of glycerol. Magnetic CuMgFe mixed-oxide catalysts derived from hydrotalcites were prepared by coprecipitation, and the effect of (Cu + Mg)/Fe molar ratio on the activity of the catalysts was discussed in detail.

Structure and Morphology of CuMgFe-xLDH and CuMgFe-xLDO
The XRD patterns of CuMgFe-xLDH are shown in Figure 1a. CuMgFe-2LDH, CuMgFe-3LDH, CuMgFe-4LDH, and CuMgFe-5LDH exhibited the characteristic reflections of hydrotalcite. Compared with CuMgFe-2LDH, the sharp diffraction peaks of CuMgFe-3LDH evidenced a better crystallization of the phase of hydrotalcite. Oppositely, CuMgFe-4LDH and CuMgFe-5LDH exhibited a lower crystallization with the increase of (Cu + Mg)/Fe molar ratio. Increasing (Cu + Mg)/Fe molar ratio might lead to the structural distortion and the orderliness decline of hydrotalcite. trivalent cations. Thermal decomposition of LDHs leads to the formation of mixed oxides with small crystal size, basicity, high dispersion, and large specific surface area. It has been reported that Cu-based catalysts derived from hydrotalcites are of basicity and dispersed copper particles [8,[24][25][26]27], which could improve the catalytic performance in the hydrogenolysis of glycerol. Geng et al. [28] disclosed that the Cu-Ca-Al catalyst derived from hydrotalcite is more active and selective for the formation of propanediols. Nonetheless, these catalysts suffer from serious difficulties in the recovery and reuse from the perspective of ecological and economical sustainability. Magnetic catalysts can be easily separated by an external magnetic field, reducing the consumption of auxiliary substances, saving energy and time in separation, and bringing significant economic and environmental benefits. However, to the best of our knowledge, there is no detailed understanding of magnetic Cu-based catalysts derived from hydrotalcites in the hydrogenolysis of glycerol.
The aim of the present work is to develop easily recycled Cu-containing catalysts in the hydrogenolysis of glycerol. Magnetic CuMgFe mixed-oxide catalysts derived from hydrotalcites were prepared by coprecipitation, and the effect of (Cu + Mg)/Fe molar ratio on the activity of the catalysts was discussed in detail.

Structure and Morphology of CuMgFe-xLDH and CuMgFe-xLDO
The XRD patterns of CuMgFe-xLDH are shown in Figure 1a. CuMgFe-2LDH, CuMgFe-3LDH, CuMgFe-4LDH, and CuMgFe-5LDH exhibited the characteristic reflections of hydrotalcite. Compared with CuMgFe-2LDH, the sharp diffraction peaks of CuMgFe-3LDH evidenced a better crystallization of the phase of hydrotalcite. Oppositely, CuMgFe-4LDH and CuMgFe-5LDH exhibited a lower crystallization with the increase of (Cu + Mg)/Fe molar ratio. Increasing (Cu + Mg)/Fe molar ratio might lead to the structural distortion and the orderliness decline of hydrotalcite.  Additionally, the diffraction peaks at 42.9 • and 62.3 • were also observed, which could be associated with the (200) and (220) planes of the MgO phase (JCPDS card no. 04-0829).
The morphology of CuMgFe-3LDH and CuMgFe-4LDH revealed by SEM is shown in Figure 2. CuMgFe-3LDH displayed a layered structure of solid lamellar (Figure 2A), whereas some solid lamellar structures accumulated obviously in CuMgFe-4LDH ( Figure 2B). Figure 2C,D shows the images of CuMgFe-3LDH and CuMgFe-4LDH after calcination, which maintained the plate-like morphology of the original precursor. High resolution transmission electron microscope (HRTEM)was also used to reveal the structure of CuMgFe-4LDO sample. A typical HRTEM image of CuMgFe-4LDO showed two identified reflection patterns with interplanar distances of 0.253 nm and 0.205 nm ( Figure 2E), corresponding to the (002) plane of CuO phase and (220) plane of CuFe 2 O 4 phase [29,30], respectively, which was in line with the XRD results.
planes of CuFe2O4 spinel phase (JCPDS card no. 77-0010), respectively. Meanwhile, the CuO phase (at around 35.5° (002) and 58.3° (202)) also might be present, but its diffraction peaks were partially overlapped with those of CuFe2O4. Consequently, it was difficult to distinguish in the XRD results. The diffraction peaks at 30.3° and 43.3° could correspond to the (220) and (400) planes of the Fe2O3 phase (JCPDS card no. 39-1346), respectively. Additionally, the diffraction peaks at 42.9° and 62.3° were also observed, which could be associated with the (200) and (220) planes of the MgO phase (JCPDS card no. 04-0829).
The morphology of CuMgFe-3LDH and CuMgFe-4LDH revealed by SEM is shown in Figure 2. CuMgFe-3LDH displayed a layered structure of solid lamellar (Figure 2A), whereas some solid lamellar structures accumulated obviously in CuMgFe-4LDH ( Figure  2B). Figure 2C,D shows the images of CuMgFe-3LDH and CuMgFe-4LDH after calcination, which maintained the plate-like morphology of the original precursor. High resolution transmission electron microscope (HRTEM)was also used to reveal the structure of CuMgFe-4LDO sample. A typical HRTEM image of CuMgFe-4LDO showed two identified reflection patterns with interplanar distances of 0.253 nm and 0.205 nm ( Figure 2E), corresponding to the (002) plane of CuO phase and (220) plane of CuFe2O4 phase [29,30], respectively, which was in line with the XRD results.

Magnetic Behavior of CuMgFe-xLDO Catalysts
The magnetic behavior of the catalysts was analyzed using VSM. All catalysts showed narrow S-shape type loops in Figure 3, indicating that all catalysts were ferromagnetic. Magnetic saturation (Ms), remanence (Mr), and coercivity (Hc) calculated from magnetically recorded data are listed in Table 1. The lower saturation magnetization and coercivity values might be due to the presence of CuO and MgO in the catalysts. The ratio of the remanence to the saturation magnetization (Mr/Ms) decreased in the following order: CuMgFe-5LDO > CuMgFe-3LDO > CuMgFe-2LDO > CuMgFe-4LDO, which was related to the inter-and intragrain exchange interactions, sub-lattice magnetization, magnetic anisotropy, and morphology of the tested sample [31]. The lower ratio corroborated its significant superparamagnetic behavior [32].

Magnetic Behavior of CuMgFe-xLDO Catalysts
The magnetic behavior of the catalysts was analyzed using VSM. All catalysts showed narrow S-shape type loops in Figure 3, indicating that all catalysts were ferromagnetic. Magnetic saturation (Ms), remanence (Mr), and coercivity (Hc) calculated from magnetically recorded data are listed in Table 1. The lower saturation magnetization and coercivity values might be due to the presence of CuO and MgO in the catalysts. The ratio of the remanence to the saturation magnetization (Mr/Ms) decreased in the following order: CuMgFe-5LDO > CuMgFe-3LDO > CuMgFe-2LDO > CuMgFe-4LDO, which was related to the inter-and intragrain exchange interactions, sub-lattice magnetization, magnetic anisotropy, and morphology of the tested sample [31]. The lower ratio corroborated its significant superparamagnetic behavior [32].

H2-TPR of CuMgFe-xLDO Catalysts
The reducibility of catalysts was investigated by hydrogen temperature-programmed reduction (H2-TPR). As shown in Figure 4, the first shoulder peak (at 205 °C, 209 °C, 191 °C, and 210 °C) in the low-temperature range for all the samples was ascribed to the reduction of CuO to Cu. The subsequent peak (at 235 °C, 242 °C, 225 °C, and 246 °C) was attributed to the reduction of CuFe2O4 to metallic Cu and Fe2O3 [33,34]. The third peak in the range of 300-550 °C might correspond to the transformation of Fe2O3 to Fe3O4 [35]. The peak in the high-temperature range of 550-900 °C for all the samples could be attributed to the continuous reduction of Fe3O4 to metallic Fe via FeO [33,36]. It should be noted that the reduction temperature of CuO in the CuMgFe-xLDO samples shifted to lower temperature compared with that of pristine CuO (around 290 °C) [33], and the reduction peak of CuFe2O4 in the CuMgFe-xLDO samples decreased significantly in comparison with that of pristine CuFe2O4 (around 340 °C) [33]. The findings indicated that the strong synergistic effect between copper and iron occurred in the CuMgFe-xLDO samples. Furthermore, the reduction peaks of CuO and CuFe2O4 in the CuMgFe-4LDO catalyst were lower than those of other catalysts, suggesting that the stronger interaction between copper and iron might occur in the CuMgFe-4LDO catalyst. Cu species could be well dispersed on the CuMgFe-4LDO catalyst surface.

H 2 -TPR of CuMgFe-xLDO Catalysts
The reducibility of catalysts was investigated by hydrogen temperature-programmed reduction (H 2 -TPR). As shown in Figure 4, the first shoulder peak (at 205 • C, 209 • C, 191 • C, and 210 • C) in the low-temperature range for all the samples was ascribed to the reduction of CuO to Cu. The subsequent peak (at 235 • C, 242 • C, 225 • C, and 246 • C) was attributed to the reduction of CuFe 2 O 4 to metallic Cu and Fe 2 O 3 [33,34]. The third peak in the range of 300-550 • C might correspond to the transformation of Fe 2 O 3 to Fe 3 O 4 [35]. The peak in the high-temperature range of 550-900 • C for all the samples could be attributed to the continuous reduction of Fe 3 O 4 to metallic Fe via FeO [33,36]. It should be noted that the reduction temperature of CuO in the CuMgFe-xLDO samples shifted to lower temperature compared with that of pristine CuO (around 290 • C) [33], and the reduction peak of CuFe 2 O 4 in the CuMgFe-xLDO samples decreased significantly in comparison with that of pristine CuFe 2 O 4 (around 340 • C) [33]. The findings indicated that the strong synergistic effect between copper and iron occurred in the CuMgFe-xLDO samples. Furthermore, the reduction peaks of CuO and CuFe 2 O 4 in the CuMgFe-4LDO catalyst were lower than those of other catalysts, suggesting that the stronger interaction between copper and iron might occur in the CuMgFe-4LDO catalyst. Cu species could be well dispersed on the CuMgFe-4LDO catalyst surface.

XPS Analysis of CuMgFe-xLDO Catalysts
Cu 2p and Fe 2p spectra of CuMgFe-xLDO catalysts are shown in Figure 5a,b. A Cu 2p 3/2 main peak at 933.5-933.9 eV was accompanied by a satellite peak at 942.2-942.8 eV, which was related to CuFe 2 O 4 . Additionally, all catalysts exhibited a Cu 2p 3/2 peak at 931.8-931.9 eV together with a satellite peak at 940.0-940.2 eV, which could be associated with the presence of CuO. A Cu 2p 1/2 main peak at 952.2-952.8 eV along with a satellite peak at 961.6-962.0 eV was also observed, in accordance with those of Cu 2+ [37,38]. All Fe 2p spectra in Figure 5b showed two main peaks at 710.9-711.1 eV and 724.2-724.5 eV, which belonged to Fe 2p 3/2 and Fe 2p 1/2 , respectively. Two accompanying satellite peaks at the binding energies of 718.5-718.8 eV and 732.3-732.8 eV were characteristics of Fe 3+ cations [38][39][40]. Mg 1s spectra of CuMgFe-xLDO catalysts are also presented in Figure 5c. All catalysts showed a peak at 1303.5-1304.1 eV, which could be due to the presence of MgO [41].

XPS Analysis of CuMgFe-xLDO Catalysts
Cu 2p and Fe 2p spectra of CuMgFe-xLDO catalysts are shown in Figure 5a,b. A Cu 2p3/2 main peak at 933.5-933.9 eV was accompanied by a satellite peak at 942.2-942.8 eV, which was related to CuFe2O4. Additionally, all catalysts exhibited a Cu 2p3/2 peak at 931.8-931.9 eV together with a satellite peak at 940.0-940.2 eV, which could be associated with the presence of CuO. A Cu 2p1/2 main peak at 952.2-952.8 eV along with a satellite peak at 961.6-962.0 eV was also observed, in accordance with those of Cu 2+ [37,38]. All Fe 2p spectra in Figure 5b showed two main peaks at 710.9-711.1 eV and 724.2-724.5 eV, which belonged to Fe 2p3/2 and Fe 2p1/2, respectively. Two accompanying satellite peaks at the binding energies of 718.5-718.8 eV and 732.3-732.8 eV were characteristics of Fe 3+ cations [38][39][40]. Mg 1s spectra of CuMgFe-xLDO catalysts are also presented in Figure 5c. All catalysts showed a peak at 1303.5-1304.1 eV, which could be due to the presence of MgO [41].

XPS Analysis of CuMgFe-xLDO Catalysts
Cu 2p and Fe 2p spectra of CuMgFe-xLDO catalysts are shown in Figure 5a,b. A Cu 2p3/2 main peak at 933.5-933.9 eV was accompanied by a satellite peak at 942.2-942.8 eV, which was related to CuFe2O4. Additionally, all catalysts exhibited a Cu 2p3/2 peak at 931.8-931.9 eV together with a satellite peak at 940.0-940.2 eV, which could be associated with the presence of CuO. A Cu 2p1/2 main peak at 952.2-952.8 eV along with a satellite peak at 961.6-962.0 eV was also observed, in accordance with those of Cu 2+ [37,38]. All Fe 2p spectra in Figure 5b showed two main peaks at 710.9-711.1 eV and 724.2-724.5 eV, which belonged to Fe 2p3/2 and Fe 2p1/2, respectively. Two accompanying satellite peaks at the binding energies of 718.5-718.8 eV and 732.3-732.8 eV were characteristics of Fe 3+ cations [38][39][40]. Mg 1s spectra of CuMgFe-xLDO catalysts are also presented in Figure 5c. All catalysts showed a peak at 1303.5-1304.1 eV, which could be due to the presence of MgO [41]. The surface element composition of CuMgFe-xLDO catalysts is summarized in Table  2. The surface Cu content and surface Cu/(Mg + Fe) atomic ratio of the CuMgFe-4LDO catalyst were slightly higher than those of the other catalysts, suggesting that Cu species could be well dispersed on the CuMgFe-4LDO catalyst. This was in line with the H2-TPR results.  The surface element composition of CuMgFe-xLDO catalysts is summarized in Table 2. The surface Cu content and surface Cu/(Mg + Fe) atomic ratio of the CuMgFe-4LDO catalyst were slightly higher than those of the other catalysts, suggesting that Cu species could be well dispersed on the CuMgFe-4LDO catalyst. This was in line with the H 2 -TPR results.

Basicity of Reduced Catalysts
The base properties of the CuMgFe-LDO catalysts were evaluated by the adsorption of CO 2 on the basic sites. The CO 2 -TPD profiles of CuMgFe-LDO catalysts are presented in Figure 6. There were two obvious desorption domains occurring at 50-400 • C and 400-700 • C for all the catalysts. The stronger the basic sites of a catalyst were, the higher the desorption temperature of CO 2 was [42]. The low-temperature peaks below 200 • C could be attributed to the desorption of CO 2 on weak basic sites, while the desorption peaks in the range of 200-400 • C could be ascribed to the desorption of CO 2 on medium basic sites. The peaks at 549-599 • C were associated with the desorption of CO 2 on strong basic sites. The weak and medium basic sites corresponded to surface OH − and Lewis acid-base pairings, respectively, and the strong basic sites were related to the contribution of low-coordination surface O 2− [43][44][45]. Notably, the desorption temperature on strong basic sites of the CuMgFe-4LDO catalyst was significantly higher than those of the other catalysts. It has been reported that dehydrogenation of glycerol leads to glyceraldehyde and it subsequently dehydrates to hydroxyacrolein on basic sites of the catalyst [46]. The strongest basic sites on the CuMgFe-4LDO catalyst might enhance its catalytic performance. As shown in Table 2, the total amount of basic sites decreased in the following order: CuMgFe-4LDO > CuMgFe-3LDO > CuMgFe-5LDO > CuMgFe-2LDO, indicating that the total amount of basic sites of the catalyst increased with the moderate increase of (Cu + Mg)/Fe molar ratio. Conversely, excessive (Cu + Mg)/Fe molar ratio decreased significantly the contribution of medium and strong basic sites , and thereby reducing the total amount of basic sites of the CuMgFe-5LDO catalyst. This could be associated with the morphology, crystal plane, and crystal size of MgO in the CuMgFe-5LDO catalyst. Li et al. found that MgO (100) surface primarily ended with alternant Mg 2+ /O 2− ions providing medium basic sites, while MgO (111) surface primarily ended with O 2− ions providing strong basic sites [47]. It was reported by Marianou et al. that the basicity of MgOs was strongly affected by the morphology, texture, and chemical composition of the materials [48]. Samples with smaller crystal size and higher surface area exhibited a higher total number of basic sites [48].  [47]. It was reported by Marianou et al. that the basicity of MgOs was strongly affected by the morphology, texture, and chemical composition of the materials [48]. Samples with smaller crystal size and higher surface area exhibited a higher total number of basic sites [48].  Figure 7. In the range of 50-700 °C, the desorption of H2 might be assigned to three different H species. The desorption peak at low temperature (50-350 °C, Hα) might be attributed to hydrogen desorption from Cu sites. The desorption peak at high temperature (350-700 °C, Hβ, Hγ)  Figure 7. In the range of 50-700 • C, the desorption of H 2 might be assigned to three different H species. The desorption peak at low temperature (50-350 • C, H α ) might be attributed to hydrogen desorption from Cu sites. The desorption peak at high temperature (350-700 • C, H β , H γ ) could correspond to the hydrogen desorption from two different Fe sites. Apparently, the CuMgFe-4LDO catalyst exhibited the highest H α desorption peak temperature, indicating that a stronger metal-hydrogen interaction occurred on the surface of the CuMgFe-4LDO catalyst. This could be due to the high Cu dispersion in the CuMgFe-4LDO catalyst, which afforded more unsaturated coordination centers for the hydrogen adsorption [33]. It has been reported that enhanced H 2 activation ability could improve the activity for glycerol hydrogenolysis [24].

H2-TPD of Reduced Catalysts
H2-TPD profiles of the CuMgFe-xLDO catalysts are shown in Figure 7. In the range of 50-700 °C, the desorption of H2 might be assigned to three different H species. The desorption peak at low temperature (50-350 °C, Hα) might be attributed to hydrogen desorption from Cu sites. The desorption peak at high temperature (350-700 °C, Hβ, Hγ) could correspond to the hydrogen desorption from two different Fe sites. Apparently, the CuMgFe-4LDO catalyst exhibited the highest Hα desorption peak temperature, indicating that a stronger metal-hydrogen interaction occurred on the surface of the CuMgFe-4LDO catalyst. This could be due to the high Cu dispersion in the CuMgFe-4LDO catalyst, which afforded more unsaturated coordination centers for the hydrogen adsorption [33]. It has been reported that enhanced H2 activation ability could improve the activity for glycerol hydrogenolysis [24].   Table 3. With the increase in (Cu + Mg)/Fe molar ratios from 2 to 4, the conversion of glycerol increased from 38.0% to 47.8%. On the contrary, the conversion of glycerol over the CuMgFe-5LDO catalyst decreased. The CuMgFe-4LDO catalyst exhibited the highest glycerol conversion and 1,2-PDO selectivity among all the catalysts. According to the reaction mechanism from glycerol to 1,2-PDO proposed by Montassier [49], first, dehydrogenation of glycerol on copper would form glyceraldehyde in equilibrium with its enolic tautomer. Then, a nucleophilic reaction of water or adsorbed OH species led to a dehydroxylation reaction. Subsequently, 1,2-PDO was formed by hydrogenation of the intermediate unsaturated aldehyde (2-hydroxy acrolein). Therefore, it can be concluded that the hydrogenolysis of glycerol to 1,2-PDO needs both metal sites for activation of hydrogen and base sites for dehydration. The higher hydrogenolysis activity over CuMgFe-4LDO catalyst might be due to the following factors. Primarily, the increase in the surface Cu content might be a factor in improving the hydrogenolysis activity of glycerol (Table 2). Furthermore, compared with the other catalysts, the stronger basic sites and the higher amount of basicity ( Figure 6 and Table 2) on the CuMgFe-4LDO catalyst favored dehydration reaction of glyceraldehyde and its enolic tautomer. Consequently, the intermediate unsaturated aldehyde (2-hydroxy acrolein) was formed. Finally, on the basis of the H 2 -TPD results (Figure 7), enhancing H 2 activation on Cu metal sites accelerated hydrogenation of intermediate unsaturated aldehyde (2-hydroxy acrolein), thereby improving the selectivity to 1,2-PDO.

Hydrogenolysis of Glycerol on Reduced CuMgFe-4LDO Catalyst at Different Temperatures
The activity of CuMgFe-4LDO catalyst for hydrogenolysis of glycerol at different temperatures is summarized in Table 4. The conversion of glycerol increased significantly from 47.8% (at 180 • C) to 75.3% (at 200 • C), suggesting that glycerol hydrogenolysis accelerated with increasing reaction temperature. Nevertheless, the selectivity to 1,2-PDO declined slightly from 97.5% (at 180 • C) to 96.5% (at 200 • C), indicating that no obvious cleavage of C-C bonds over CuMgFe-4LDO catalyst occurred even at higher temperatures.

Recycled Usage of Reduced CuMgFe-4LDO Catalyst
The recycling procedure of the CuMgFe-4LDO catalyst was performed for examining the stability of the catalyst. The spent catalysts were separated by an external magnetic field. Due to the magnetism for CuMgFe-4LDO catalysts, the catalysts could be easily recycled, as shown in Figure 8. The activity of recycled CuMgFe-4LDO catalyst is summarized in Table 5. The conversion of glycerol decreased slightly from 47.8% (of the fresh catalyst) to 46.9% (in the second recycle), and then it remained stable (in the third and the fourth recycles). After five times of recycling, the conversion of glycerol over the CuMgFe-4LDO catalyst decreased by 3.5%. No apparent weight loss of catalysts was observed after five times recycles.

Recycled Usage of Reduced CuMgFe-4LDO Catalyst
The recycling procedure of the CuMgFe-4LDO catalyst was performed for examining the stability of the catalyst. The spent catalysts were separated by an external magnetic field. Due to the magnetism for CuMgFe-4LDO catalysts, the catalysts could be easily recycled, as shown in Figure 8. The activity of recycled CuMgFe-4LDO catalyst is summarized in Table 5. The conversion of glycerol decreased slightly from 47.8% (of the fresh catalyst) to 46.9% (in the second recycle), and then it remained stable (in the third and the fourth recycles). After five times of recycling, the conversion of glycerol over the CuMgFe-4LDO catalyst decreased by 3.5%. No apparent weight loss of catalysts was observed after five times recycles.  To explore the reasons for activity loss, the compositions of the reduced catalyst and five-times-recycled catalyst were determined by induced coupled plasma-optical emission spectroscopy (ICP-OES). No obvious leaching Cu was observed after five recycles (Table 5). Meanwhile, to confirm further no Cu leaching, the catalyst was filtered off after  To explore the reasons for activity loss, the compositions of the reduced catalyst and five-times-recycled catalyst were determined by induced coupled plasma-optical emission spectroscopy (ICP-OES). No obvious leaching Cu was observed after five recycles (Table 5). Meanwhile, to confirm further no Cu leaching, the catalyst was filtered off after 5 h reaction (halfway through the reaction) in the fifth recycle. At this moment, glycerol conversion and the selectivity to 1,2-propanediol were 28.5% and 97.0%, respectively. Then, the filtrate was transferred to a 100 mL stainless steel autoclave. Moreover, hydrogenolysis of glycerol was still performed under no catalyst conditions. Once more, after 5 h reaction, glycerol conversion and the selectivity to 1,2-propanediol did not further increase, indicating that no Cu leached into the filtrate. Furthermore, the XRD results of the reduced catalyst and five-times-recycled catalyst were also analyzed ( Figure S1), Cu metal sizes were calculated from the diffraction peaks (220) according to the Scherrer equation. Cu metal size increased from 9.2 nm to 10.8 nm after five recycles, suggesting that the decreased activity of CuMgFe-4LDO catalyst could be due to slight sintering of copper metal.

Preparation of CuMgFe-Mixed Oxides Catalysts
Four CuMgFe-LDH precursors with different (Cu 2+ + Mg 2+ )/Fe 3+ molar ratios were prepared by coprecipitation. The A solution was a 0.2 M aqueous solution containing the nitrates of Cu 2+ , Mg 2+, and Fe 3+ . The B solution was an aqueous solution of NaOH and Na 2 CO 3 with a concentration of 0.25 M and 0.8 M, respectively. Solution A and B were added drop-wise into the deionized water with vigorous stirring. During the coprecipitation, the slurry was kept at pH 10.0 ± 0.1 by adjusting dropping rates. The resulting suspension was aged at 60 • C for 18 h. The final precipitate was filtered, washed, and dried at 110 • C for 12 h. Layered double hydroxides were gained, and they are donated as CuMgFe-xLDH (x = 2, 3, 4 or 5 according to the (Cu 2+ + Mg 2+ )/Fe 3+ molar ratios of 2, 3, 4, or 5). Subsequently, the hydrotalcites were calcined at 600 • C for 5 h in air, and the products are designated as CuMgFe-xLDO (x = 2, 3, 4, or 5). The nominal Cu content is 9 wt% in the CuMgFe-xLDO samples.
The morphology of the CuFeMg-LDH was investigated using a Quanta 400 FEG scanning electron microscope (FEI, Hillsboro, Oregon, OR, USA) with an accelerating voltage of 20 kV.
Transmission electron microscopy (TEM) images were obtained using a Tecnai G 2 F20 transmission electron microscope (FEI, Hillsboro, Oregon, OR, USA). The samples were ultrasonically dispersed in ethanol.
The reducibility of the catalysts was studied by hydrogen temperature-programmed reduction (H 2 -TPR) using a fixed-bed reactor. The catalysts were reduced under a 10% H 2 /Ar mixed gas (30 mL/min) from 50 • C to 900 • C at a rate of 10 • C/min. The hydro-gen consumption was analyzed on-line using an SC-200 gas chromatograph (Chuanyi, Chongqing, China) equipped with a thermal conductivity detector (TCD).
The X-ray photoelectron spectra (XPS) were obtained using an XSAM800 spectrometer (Kratos, Manchester, UK) with an Al Ka (hm = 1486.6 eV) X-ray source, and the binding energies were corrected using C (1 s) at 284.6 eV.
The basicity and the H 2 -activation ability of the reduced catalysts were determined by temperature-programmed desorption of CO 2 and H 2 (CO 2 -TPD and H 2 -TPD). The catalysts were reduced at 400 • C in 10% H 2 /Ar for 2 h and then cooled to 50 • C in a He flow. Subsequently, CO 2 was fed into the reactor for 0.5 h. Then the catalysts were purged at 50 • C with He for 3 h. Finally, the samples were heated linearly to 750 • C at a rate of 10 • C/min in a He flow. While the desorbed CO 2 was recorded continuously by a TCD detector. H 2 -TPD was carried out by the same procedure. Only CO 2 was supplanted by the 10% H 2 /Ar. The magnetization was characterized by a superconducting quantum interference SQUID magnetometer (Quantum Design, San Diego, CA, USA) with a maximum field of 20 kOe at room temperature. The saturation magnetization (Ms), coercive force (Hc), and residual magnetization (Mr) were measured.

Catalytic Experiments
Hydrogenolysis of glycerol was performed in a 100 mL stainless steel autoclave with a mechanical stirrer and an electric temperature controller, operated under H 2 pressure of 2.0 MPa. Prior to reaction, the catalysts were reduced by 10% H 2 /Ar stream at 400 • C for 2 h in a fixed-bed flowed reactor. A total of 8.0 g aqueous solution of 75 wt% glycerol and 10 wt% (based on glycerol) of the catalysts were charged into the autoclave. The liquid products were analyzed by using a Scion 456C GC gas chromatograph (Techcomp, Shanghai, China) equipped with a flame ionization detector (PEG-20M column: 30 m × 0.25 mm × 0.5 µm). The gas products were analyzed by using a Scion 456C GC gas chromatograph equipped with a thermal conductivity detector (TDX-01 column: 3 m × 3 mm).

Conclusions
The CuMgFe-xLDO catalysts derived from different (Cu + Mg)/Fe metal ratios were prepared by coprecipitation. The activity of the CuMgFe-4LDO catalyst was higher than those of other CuMgFe-xLDO catalysts, and the conversion of glycerol and the selectivity to 1,2-PDO reached 47.8% and 97.5% at 180 • C, respectively. The superior catalytic performance of CuMgFe-4LDO was associated with its strong basicity, excellent H 2 activation ability, and an increase in the surface Cu content. The CuMgFe-4LDO catalyst also exhibited good stability. Furthermore, the CuMgFe-xLDO catalysts could be easily recycled with the assistance of an external magnetic field due to their magnetism.
Author Contributions: Conceptualization, X.Y.; methodology, X.Y. and F.Z.; software, X.Y.; validation, X.Y.; formal analysis, X.Y.; investigation, X.Y. and F.Z.; resources, X.Y. and Y.W.; data curation, X.Y.; writing-original draft preparation, X.Y.; writing-review and editing; X.Y., F.Z., and Y.W.; supervision, X.Y.; project administration, Y.W. and D.C.; funding acquisition, X.Y. All authors have read and agreed to the published version of the manuscript. Data Availability Statement: Data presented in this study are available on request from the corresponding author.