Highly efficient TiO2-supported Co–Cu catalysts for conversion of glycerol to 1,2-propanediol

Glycerol is a low-cost byproduct of the biodiesel manufacturing process, which can be used to synthesize various value-added chemicals. Among them, 1,2-propanediol (1,2-PDO) is of great interest because it can be used as an intermediate and additive in many applications. This work investigated the hydrogenolysis of glycerol to 1,2-PDO over Co–Cu bimetallic catalysts supported on TiO2 (denoted as CoCu/TiO2) in aqueous media. The catalysts were prepared using the co-impregnation method and their physicochemical properties were characterized using several techniques. The addition of appropriate Cu increased the glycerol conversion and the 1,2-PDO yield. The highest 1,2-PDO yield was achieved over a 15Co0.5Cu/TiO2 catalyst at 69.5% (glycerol conversion of 95.2% and 1,2-PDO selectivity of 73.0%). In the study on the effects of operating conditions, increasing the reaction temperature, initial pressure, and reaction time increased the glycerol conversion but decreased the selectivity to 1,2-PDO due to the degradation of formed 1,2-PDO to lower alcohols (1-propanol and 2-propanol). The reaction conditions to obtain the maximum 1,2-PDO yield were a catalyst-to-glycerol ratio of 0.028, a reaction temperature of 250 °C, an initial H2 pressure of 4 MPa, and a reaction time of 4 h.


Materials and methods
Preparation of catalysts. The TiO 2 (21 nm primary particle size (TEM), ≥ 99.5% trace metals, Sigma-Aldrich) was used as a catalyst support. The CoCu/TiO 2 catalysts were prepared using the co-impregnation method. Aqueous solutions of Co(NO 3 ) 2 •6H 2 O (2 M, 98%, QReC) and Cu(NO 3 ) 2 •3H 2 O (1 M, 99%, Ajax) were used as the precursors for Co and Cu, respectively. The amount of Co was kept at 15 weight percent (wt%) but the amounts of Cu were varied at 0, 0.25, 0.5, 1.0, or 2.0 wt.%. The predetermined amount of the Co and Cu solutions was loaded on the TiO 2 support. Then, each mixture solution was continuously stirred at room temperature for 1 h. The mixture was then dried in a hot-air oven at 100 °C for 12 h and calcined in an air furnace at 500 °C and a heating rate of 5 °C min -1 for 4 h to obtain xCoyCu/TiO 2 catalysts (where x = 15 wt% of Co and y = 0, 0.25, 0.5, 1 or 2 wt% of Cu, respectively).
Catalyst activity studies. The activity of each prepared catalyst was evaluated in a teflon-lined stainlesssteel autoclave (Parr 4848, actual volume = 200 mL), which was equipped with an electromagnetic stirrer and a temperature controller unit. Before the reaction, each catalyst was reduced using H 2 at a flow rate of 50 mL min −1 at 350 °C for 1 h in a plug flow reactor. Then, the catalyst was transferred to the teflon-lined stainless-steel reactor. Subsequently, glycerol (10 mL, 99.5%, QReC) and deionized water (10 mL) were rapidly introduced into the reactor to prevent the reduced catalyst from contact with air. Afterward, the reactor was sealed and purged three times with pure H 2 (99.999%, Air liquide) to eliminate any air. The reactor was pressurized to the desired H 2 pressure (2, 4, or 6 MPa) and the stirring speed was set at 800 rounds per minute (rpm). Then, the reactor system was heated to the desired reaction temperature (210, 230, 250, or 270 °C). After the reaction, the reactor was cooled to room temperature and the pressure was released to ambient conditions. A small amount of the liquid products (5.0 mL) was then sampled using a syringe filter (nylon 0.45 µm, CNW Technologies) for product analysis using gas chromatography (Shimadzu, GC-14A), equipped with a DB-WAX capillary column (30 m long, 0.53 mm inner diameter, and 1 µm thickness) and a flame ionization detector. The detected liquid products were 1,2-propanediol ( www.nature.com/scientificreports/ ethylene glycol (EG). For the quantification of each product, the pure chemical was purchased and a standard calibration curve with four concentration points and a coefficient of determination (R 2 ) > 0.99 was made using GC. The activity of the catalysts was presented in terms of glycerol conversion (%), product selectivity (%), and product yield (%) as shown in Eqs. (1), (2), and (3), respectively: where n gly in is the molar amount of glycerol before the reaction (blank), n gly out is the molar amount of unreacted glycerol after the reaction, and n p is the molar amount of desired product. Z p and Z gly are the number of carbon atoms of the desired product and glycerol, respectively. Note that, for the catalytic activity data, "others" refers to 2-PO, HA, and other unquantified products including gaseous products (e.g. propane and ethane).
For the reusability test of the catalyst, after the first cycle of the reaction, the catalyst was separated from the liquid products using a centrifuge at 8000 rpm for 15 min. Then, the catalyst was washed three times with DI water, dried overnight in the hot-air oven, and reduced using the H 2 flow as described previously before the new cycle of the reaction.

Characterization of catalysts.
The crystalline phases and the average metal crystallite sizes of the catalysts were analyzed using powder x-ray diffraction (XRD, JEOL JDX-3530 and Philips X-Pert) with Cu-Kα radiation at 45 kV and 40 mA at an angle (2θ) range of 10-80°. A step size of 0.02° and a step time of 0.5 s were used for the measurements.
The particle size distribution and the elemental composition mapping of the samples were analyzed using transmission electron microscopy with energy-dispersive X-ray spectroscopy (TEM with EDX: JEM-2100). After reduction with H 2 at 350 °C, the samples were suspended in ethanol solvent, dropped on carbon film coated on Cu TEM grids, and dried in a chamber filled with N 2 at room temperature before the analysis.
The surface area, pore volume, and average pore size of the catalysts were determined using an N 2 -physisorption analyzer (Brunauer-Emmett-Teller (BET): 3Flex Physisorption Micromeritics) at − 196 °C. Each catalyst was pretreated at 300 °C for 24 h in the system of the N 2 -physisorption analyzer before measurement. For each catalyst, the BET surface area was determined in a relative pressure (P/P 0 ) range of 0.05-0.30; the total pore volume was determined at P/P 0 of 0.995, and the pore size was computed using the Barrett-Joyner-Halenda method.
The elemental composition of the catalysts was determined using inductively coupled plasma-optical emission spectrometry (ICP-OES). Before the ICP-OES measurements, the solid samples were dissolved in hydrochloric acid solution.
The electronic states of cobalt (Co 2p) and copper (Cu 2p) for the samples were analyzed using x-ray photoelectron spectrometry (XPS; Kratos Axis Ultra DLD), using Al K α for the x-ray source.
The reducibility of the catalysts was analyzed using H 2 -Temperature-Programmed Reduction (H 2 -TPR, Micromeritics AutoChem II). Before the H 2 -TPR experiments, each catalyst (0.1 g) was pretreated at 120 °C and a heating rate of 5 °C min −1 under an He flow (50 mL min −1 ). Then, it was cooled to 50 °C and then heated to 900 °C at a heating rate of 10 °C min −1 under a flow of 10% H 2 in Ar. The H 2 consumed during the reduction was continuously monitored using a thermal conductivity detector (TCD).
The acidity of the catalysts was analyzed using NH 3 temperature-programmed desorption (NH 3 -TPD, Micromeritics AutoChem II). Before the NH 3 -TPD experiments, each catalyst (0.1 g) was pretreated using the same method as the pretreatment for the H 2 -TPR experiments. After the catalyst had cooled to 50 °C, it was exposed to 0.2% NH 3 in He for 1 h followed by purging with He for 1 h. Finally, the NH 3 -TPD measurement was carried out from 50 to 900 °C at a heating rate of 10 °C min -1 . The NH 3 -desorption was continuously monitored using a TCD. Note that all gas flow rates were 25 mL min -1 for gas/gas mixture measurements.
The surface morphology of the catalysts was analyzed using scanning electron microscopy (SEM; JEOL, JSM7600 F). Samples were imaged at a working distance of 4.0 mm and an acceleration voltage of 1.0 kV.

Results and discussion
Catalyst characterization. The crystallization analysis of catalysts by XRD. The XRD patterns of 15Co/ TiO 2 , 15Co0.5Cu/TiO 2 , 15Co1Cu/TiO 2 , and 0.5Cu/TiO 2 in the 2θ range from 10° to 80° are shown in Fig. 1. Note that the XRD measurement of all catalysts was carried out after the H 2 reduction. As observed, the diffraction peaks of TiO 2 with both anatase (2θ = 25.3°, 37.9°, 48.0°, 53.9°, and 62.6°) 17 and rutile (2θ = 27.4°, 36.1°, 36.9°, 38.5°, and 41.2°) 16 structures were observed in every catalyst, and all the XRD peaks of TiO 2 were almost identical. This suggested that the loading of Co and Cu metal had not affected the crystallinity of the TiO 2 . In the XRD profiles of 15Co/TiO 2 , 15Co0.5Cu/TiO 2 , and 15Co1Cu/TiO 2 , only one clear diffraction peak of metallic Co at 2θ = 44.3°1 8 was observed; other diffraction peaks indicating the metallic Co (e.g. 2θ = 36.9° and 62.6°) were overlapped with the diffraction peaks of the TiO 2 . For the Cu-containing catalysts, no diffraction peak of  www.nature.com/scientificreports/ Cu species was observed, possibly because the amount of Cu loading was small (< 2 wt%) and the crystal size of the Cu species was smaller than the size limit detection of the instrument (< 2.5 nm).
The morphological analysis of catalysts by TEM. The structural morphologies of 15Co/TiO 2 , 15Co0.5Cu/TiO 2 , 15Co1Cu/TiO 2 , and 0.5Cu/TiO 2 were characterized using TEM, as shown in Fig. 2. Note that the catalysts were exposed to air before imaging using TEM because of this limitation during the sample preparation for the TEM experiment. As observed, the oxides of Co and/or Cu were distributed throughout the TiO 2 support. Every sample had a lattice d-spacing of 0.325 nm and 0.350 nm, specifying the crystal planes of the TiO 2 -rutile phase [1 1 0] and the TiO 2 -anatase phase [1 0 1], respectively 19 . Figure  Furthermore, the elemental distributions on the surface of 15Co/TiO 2 , 15Co0.5Cu/TiO 2 , 15Co1Cu/TiO 2 , and 0.5Cu/TiO 2 were analyzed using TEM with EDX-mapping, as shown in Fig. 3. The cobalt and copper species on the TiO 2 support correlated to the bright yellow and red spots on the EDX-mapping images, respectively. As seen, both metal species were uniformly distributed on the surface of the support. Based on the XRD results, the Cu species in 0.5Cu/TiO 2 , 15Co0.5Cu/ TiO 2 , and 15Co1Cu/TiO 2 , and the CuCO 2 O 4 phase in 15Co1Cu/TiO 2 were not observed. These results from TEM and TEM with EDX-mapping confirmed that the Cu and CuCo 2 O 4 species were present in the catalysts.
The textural property analysis by N 2 -physisorption. The textural properties (BET surface area, pore size diameter, and pore volume) of TiO 2 , 15Co/TiO 2 , 15Co0.5Cu/TiO 2 , 15Co1Cu/TiO 2 , and 0.5Cu/TiO 2 were determined using N 2 -sorption analysis, as shown in Table 1. Furthermore, to evaluate the N 2 adsorption-desorption isotherm of each catalyst, the plot of the amount of N 2 adsorbed and desorbed for each catalyst versus the relative  www.nature.com/scientificreports/ pressure (P/P 0 ) is shown in Fig. 4a. The plot of the pore size distribution of each catalyst is shown in Fig. 4b. Note that, before the BET measurement, the pure TiO 2 was treated using the same procedure as the catalyst preparation, except for adding the active metals. In Table 1, the pure TiO 2 support had the greatest BET surface area (50.5 m 2 g −1 ) with the largest pore volume (0.69 cm 3 g −1 ). After Co and/or Cu were impregnated onto the TiO 2 support, the surface areas became smaller than those of the pure TiO 2 , and the pore volumes and average pore size diameters were similarly in the ranges 0.29-0.41 cm 3 g −1 and 39.4-50.5 nm, respectively. Additionally, the www.nature.com/scientificreports/ greater the amount of loaded active metals, the lower the BET surface area, probably because the active metals filled the pores of the TiO 2 support during the catalyst preparation. The plots in Fig. 4a describe the adsorption-desorption behavior of N 2 onto the catalyst surface. According to the IUPAC classification, all the catalysts were classified into the type IV isotherm, identifying them as a mesopore material consisting of multilayer adsorption followed by pore condensation 22 . A hysteresis loop can be seen above P/P 0 values of 0.8 for every catalyst and it can be classified as type H3 hysteresis, revealing the capillary condensation of mesoporous materials 23 . For the pore size distribution curve shown in Fig. 4b, the peaks of all samples were very close to each other, approximately 50 nm.
The elemental composition analysis by ICP-OES. The elemental compositions of 15Co/TiO 2 , 15Co0.5Cu/TiO 2 , 15Co1Cu/TiO 2 , and 0.5Cu/TiO 2 were determined using ICP-OES, as shown in Table 1. The results indicated that the actual amount of cobalt and/or copper in each catalyst was close to the theoretical value.
The oxidation states analysis of catalysts by XPS. The surface chemical states of the catalysts were determined using XPS, as shown in Fig. 5. For the Co spectra (Fig. 5a.), the deconvoluted peaks at 775.3, 777.4, and 779.4 eV in the orbital 2p 3/2 position corresponded to the metallic Co 0 , Co 2+ , and satellites, respectively 24 . The binding energy (B.E.) value between each spin-orbit separation of Co 2p 3/2 and Co 2p 1/2 was consistent at 15 eV, which is characteristic of Co 3 O 4 spinel 25 . The greatest area of the deconvoluted peak at 777.4 eV indicated that the Co 2+ species represented the greatest amount among all the cobalt species 26 . For the Cu spectra (Fig. 5b), the deconvoluted peaks at 934.6 eV and 954.6 eV corresponded to Cu 2p 3/2 and Cu 2p 1/2 , respectively. The splitting energy of 20 eV indicated the formation of Cu 2+27 . Furthermore, the deconvoluted peak at 932.5 eV in the orbital Cu 2p 3/2 position could be assigned to Cu 0 or Cu + because the range of the B.E.s for Cu 0 and Cu + are overlapped 26 . After adding the Cu to Co/TiO 2 , a positive shift for the B.E. of Cu on CoCu/TiO 2 was observed. This phenomenon was similarly observed in other Co-Cu catalysts for methanol decomposition to hydrogen production 26 and direct synthesis of ethanol and higher alcohols from syngas 28 . This indicates that copper loses electrons and variation of copper chemical states in the bimetallic Co-Cu catalysts 26 . In other words, an electronic interaction occurs between the Co and Cu species over the TiO 2 support. www.nature.com/scientificreports/ The reducibility analysis of catalysts by H 2 -TPR. The reducibility of 15Co/TiO 2 , 15Co0.5Cu/TiO 2 , 15Co1Cu/ TiO 2 , and 0.5Cu/TiO 2 was characterized using H 2 -TPR, as shown in Fig. 6. The H 2 -TPR profile of the catalysts can be used to identify the species of metal on the surface and the nature of the active site of the catalysts. For the monometallic catalysts, the 15Co/TiO 2 had two reduction peaks, with one sharp peak at 200 °C and one broad reduction peak between 240 to 500 °C, assigned to the reductions of Co 3 O 4 to CoO and CoO to Co, respectively 25 . The TPR profile of 0.5Cu/TiO 2 had a reduction peak at 115 °C with a shoulder peak at 160 °C, corresponding to the reduction of Cu 2 O and CuO to Cu, respectively 29 . For the bimetallic catalysts, the TPR profile of 15Co0.5Cu/ TiO 2 had an overlapped peak at a lower temperature (205 °C) involved with the reduction of Co 3 O 4 to CoO, CuO to Cu, and Cu x Co 3-x O 4 to Cu and Co 30 , and a relatively large broad reduction peak at higher temperature (approximately 280-500 °C), assigned to the reduction of CoO to Co 31 . Interestingly, 15Co1Cu/TiO 2 had an overlapped reduction peak around 100-250 °C with a relatively small broad peak around 250-300 °C. This can be explained by 15Co1Cu/TiO 2 , with the main peak and a shoulder peak around 100-250 °C being attributed to a complicated reduction of the oxides of Cu and Co to Cu and Co, suggesting that the Cu-and Co-species particles were in close contact, which led to an occurrence of a spillover mechanism during the reduction of the Co-Cu catalysts 32 . In other words, because of the close contact of the Cu-and Co-species particles, the adsorbed atomic hydrogen could transfer from Cu 0 to Co oxide species and encouraged reduction to Co 0 .
The acidity analysis of the catalysts by NH 3 -TPD. The acidity of 15Co/TiO 2 , 15Co0.5Cu/TiO 2 , and 0.5Cu/TiO 2 was determined using NH 3 -TPD, as shown in Table 2. The acid sites were calculated from the peak areas of NH 3 desorption signals in different temperature ranges, which can be defined as weak acid sites (< 300 °C), medium acid sites (300-500 °C), and strong acid sites (> 500 °C) 32 . From Table 2, the TiO 2 support had the highest number of total acid sites. After the metals were impregnated, the total acid sites of catalysts decreased, which was attributed to the occupation of acid sites by metal species 11 . For the bimetallic catalyst, the addition of Cu to Co/ TiO 2 enhanced the formation of acid sites due to the formation of CuCo 2 O 4 species 32 (as deduced from TEM) and the synergistic interaction of these metals with the TiO 2 support 11 .
The activity of catalysts for hydrogenolysis of glycerol to 1,2-PDO. The    www.nature.com/scientificreports/ at 51.8% with 60.2% glycerol conversion and 86.0% of 1,2-PDO selectivity when the Cu loading was 0.5 wt%. This was because i) there was a synergistic catalytic effect between the Co and Cu that occurred between the interfaces of these metal particles; ii) there was an increase in acid sites 33 , especially the weak acid site when the Cu was added to Co/TiO 2 (as indicated in the NH 3 -TPD results in Table 2) that can favor the activation of the C-O bond from glycerol to HA during the dehydration step 34 so that the glycerol conversion increased; and iii) the presence of multiple Co sites (as indicated in the results of XPS in Fig. 5 and H2-TPR in Fig. 6) promoted the overall hydrogenolysis of glycerol to 1,2-PDO 21,32 .
Increasing the Cu loading on Co/TiO 2 from 0.5 to 2 wt% decreased the performance of the catalysts, probably because the excess amount of Cu over the surface (as indicated by the H 2 -TPR results in Fig. 6) could reduce the number of interfaces between Co and Cu. In addition, the catalyst's pores could have been blocked by the high metal loadings 35 (as indicated by the BET results in Table 1).
Effect of operating conditions. Catalyst-to-glycerol ratio. The effect of the glycerol-to-catalyst ratio on the hydrogenolysis of glycerol over 15Co0.5Cu/TiO 2 is presented in Fig. 7. The catalyst-to-glycerol ratio (on a weight basis) was varied from 0.011 to 0.033. As observed, increasing the catalyst-to-glycerol ratio from 0.011 to 0.033 increased the glycerol conversion and the 1,2-PDO yield from 29.4% to 69.8% and 24.5% to 59.3%, respectively, because the number of active sites increased with the increasing catalyst-to-glycerol ratio 33,36 . It was also observed that the selectivity of 1,2-PDO did not change much in this range for the catalyst-to-glycerol ratio; the selectivity of 1,2-PDO was approximately 83-86%, implying that over hydrogenolysis of 1,2-PDO did not occur.
Reaction temperature. The effect was investigated of the reaction temperature on the activity of the hydrogenolysis of glycerol over 15Co0.5Cu/TiO 2 and the results are shown in Fig. 8. The increase in reaction temperature from 210 to 270 °C led to a large increase in the glycerol conversion from 21.9 to 99.6%. The selectivity toward 1,2-PDO gradually decreased from 90.0 to 51.5% and the maximum 1,2-PDO yield at 69.5% was obtained at 250 °C. The results indicated that the higher reaction temperature favored glycerol conversion but led to lower selectivity of 1,2-PDO due to the formation of 1-PO and 2-PO 11,13 . www.nature.com/scientificreports/ Reaction pressure. The effect was examined of initial hydrogen pressure in the pressure range 2-6 MPa on the glycerol hydrogenolysis over 15Co0.5Cu/TiO 2 at 250 °C, as presented in Fig. 9. Increasing the initial pressure from 2 to 6 MPa resulted in an increased glycerol conversion from 90.2 to 99.8%, because the solubility of hydrogen into an aqueous solution was enhanced by the increased pressure 13 . The selectivity toward 1,2-PDO slightly increased from 72.3 to 73.0% and further sharply decreased to 63.0% at 6 MPa hydrogen pressure due to the degradation of 1,2-PDO to 1-PO and 2-PO 14 . Therefore, the optimal initial hydrogen pressure to obtain the maximum 1,2-PDO yield was 4 MPa for this catalyst.
Reaction time. The effect of reaction time on the glycerol hydrogenolysis over 15Co0.5Cu/TiO 2 was investigated at a catalyst-to-glycerol ratio of 0.028, a reaction temperature of 250 °C, and an initial hydrogen pressure of 4 MPa, as shown in Fig. 10. As seen, the glycerol conversion increased from 90.2 to 99.8% when the reaction time increased from 2 to 8 h. However, the selectivity of 1,2-PDO decreased from 72.3 to 63.0% with increasing reaction time, which also contributed to the degradation of formed 1,2-PDO to 1-PO 7 . Overall, the local optimal reactions to obtain the maximum 1,2-PDO yield (69.5%) was a catalyst-to-glycerol ratio of 0.028, a reaction temperature of 250 °C, an initial hydrogen pressure of 4 MPa, and a reaction time of 4 h for this catalyst.
Reusability test of 15Co0.5Cu/TiO 2 . The reusability of the catalyst is an important factor that should be considered for commercial purposes. The reusability of 15Co0.5Cu/TiO 2 for the hydrogenolysis of glycerol over five cycles is shown in Fig. 11. As observed, the glycerol conversion and the 1,2-PDO yield decreased after each hydrogenolysis of glycerol reaction. After five cycles, the glycerol conversion and 1,2-PDO yields decreased from 95.2% and 69.5% to 52.5% and 29.3%, respectively, while 1,2-PDO selectivity gradually decreased from 73.0%     www.nature.com/scientificreports/ to 55.8%. The catalyst used for five cycles was further analyzed for its physicochemical properties using SEM, XRD, and N 2 -physisorption, as shown in Figs. 12 and 13, and Table 4, respectively. Comparing the SEM images, N 2 -physisorption results, and XRD pattern of the fresh and used catalysts, the surface morphology, the textural properties, and the crystalline properties were similar. In contrast, the analysis of the fresh catalyst and the used catalyst after five cycles using ICP-OES (see Table 5) showed that the amount of Co had substantially decreased in the used catalyst. Furthermore, the crude product of the hydrogenolysis of the glycerol reaction was analyzed using ICP-OES; it was found that the amounts of Co and Cu in the crude product were 374.5 and 0.2 ppm, respectively. These results indicated that gradual leaching, especially of Co, occurred during the reaction and the recycling process, leading to the gradual deactivation of the catalyst, similar to the results reported by Feng et al. 7 A further study in the prevention of the leaching is, therefore, needed.
Comparative activity of the catalysts for hydrogenolysis of glycerol to 1,2-PDO. From previous research, most catalysts were bifunctional, consisting of acidic sites and metallic sites, which are involved in the dehydration and hydrogenation steps, respectively. Figure 14 illustrates the comparative activity of bifunctional catalysts for the hydrogenolysis of glycerol to 1,2-PDO, with the detail of each catalyst shown in Table S1. High reaction temperatures around 170-250 °C are commonly used since the hydrogenolysis of glycerol is an endothermic reaction. In addition, high initial H 2 pressures around 1-6 MPa are preferred because they provide an extensive massive dispersion of H 2 and more H 2 is dissolved in the liquid phase, which is advantageous for the hydrogenolysis of glycerol. Comparing our catalyst performance with others, CoCu/TiO 2 was highly effective in the hydrogenolysis of glycerol to 1,2-PDO, achieving a maximum 1,2-PDO yield at 69.5% with 95.2% glycerol conversion in the current study. Catalysts providing a 1,2-PDO yield above 90% are shown in Fig. 14  . Nevertheless, our catalyst can be easily prepared, the Cu and Co precursors are relatively   www.nature.com/scientificreports/ inexpensive, and the activity of CoCu/TiO 2 can be further improved by fine-tuning the Co-to-Cu-to-TiO 2 ratio and optimizing the product formation using design of experiments.

Conclusion
The 15Co0.5Cu/TiO 2 catalyst was highly active for the hydrogenolysis of glycerol to 1,2-PDO and has potential for industrial use. The maximum 1,2-PDO yield was achieved at 69.5% with 95.2% glycerol conversion and 73.0% 1,2-PDO selectivity under the maximized conditions of a catalyst-to-glycerol ratio of 0.028 and a reaction temperature of 250 °C at 4 MPa H 2 for 4 h. The addition of Cu to Co/TiO 2 caused a synergistic catalytic effect between the Co and Cu, providing much higher activity for 1,2-PDO formation than from the monometallic catalysts. The NH 3 -TPD and H 2 -TPR analyses suggested that an increase in the number of acid sites (especially weak acid sites) and the presence of multiple Co sites, respectively, can favor the hydrogenolysis of glycerol to 1,2-PDO. In the study on the effects of operating conditions, increasing the reaction temperature, initial pressure, and reaction time increased the glycerol conversion but decreased the selectivity to 1,2-PDO due to the degradation of formed 1,2-PDO to lower alcohols (1-propanol and 2-propanol).