Selective Hydrogenolysis of Glycerol and Crude Glycerol ( a By-Product or Waste Stream from the Biodiesel Industry ) to 1 , 2-Propanediol over B 2 O 3 Promoted Cu / Al 2 O 3 Catalysts

The performance of boron oxide (B2O3)-promoted Cu/Al2O3 catalyst in the selective hydrogenolysis of glycerol and crude glycerol (a by-product or waste stream from the biodiesel industry) to produce 1,2-propanediol (1,2-PDO) was investigated. The catalysts were characterized using N2-adsorption-desorption isotherm, Inductively coupled plasma atomic emission spectroscopy (ICP-AES), X-ray diffraction (XRD), ammonia temperature programmed desorption (NH3-TPD), thermogravimetric analysis (TGA), temperature programmed reduction (TPR), and transmission electron microscopy (TEM). Incorporation of B2O3 to Cu/Al2O3 was found to enhance the catalytic activity. At the optimum condition (250 ◦C, 6 MPa H2 pressure, 0.1 h−1 WHSV (weight hourly space velocity), and 5Cu-B/Al2O3 catalyst), 10 wt% aqueous solution of glycerol was converted into 1,2-PDO at 98 ± 2% glycerol conversion and 98 ± 2% selectivity. The effects of temperature, pressure, boron addition amount, and liquid hourly space velocity were studied. Different grades of glycerol (pharmaceutical, technical, or crude glycerol) were used in the process to investigate the stability and resistance to deactivation of the selected 5Cu-B/Al2O3 catalyst.


Introduction
Owing to its environmental and sustainability benefits, biodiesel, produced mainly by transesterification of vegetable oils or animal fats, has been regarded as a promising substitute to the fossil-based transportation fuels.The biodiesel industry has, however, generated very large amounts of crude glycerol as its primary by-product or waste stream that needs to be disposed.The current global market of glycerol is likely to be saturated because of its limited utilization in different fields [1].Therefore, there is an urgent need of finding new applications for glycerol or crude glycerol for the sustainability of the biodiesel industry.Chemical valorization is one of the pathways in which glycerol could be converted to high-value chemicals for various applications, which could burst the demand for glycerol.
Hydrogenolysis of glycerol to 1,2-PDO over metal-based catalysts, such as Pt, Ru, Ir, Rh, Pd, Ni, and Cu, has been extensively reported in the literature [20][21][22][23][24][25][26][27][28].While noble-metal and Ni-based catalysts have demonstrated excellent catalytic activity, these catalysts often promote excessive C-C cleavage, resulting in the formation of degraded, lower-carbon compounds, such as ethylene glycol, ethanol, methanol, and methane [29].Cu-based catalysts demonstrated to be superior catalysts with excellent catalytic activity [27,28].Sato, et al. [27], found that copper nanoparticle catalyst prepared from Cu-Al hydrotalcite gave an excellent yield of 1,2-propanediol in the hydrogenolysis of glycerol, achieving 100% conversion of glycerol with 93% 1,2-PDO selectivity at ambient hydrogen pressure through dehydration-hydrogenation via hydroxyacetone (HA).Hydrogenolysis of glycerol to different chemical compounds is given in Scheme 1.The conversion of glycerol to 1,2-PDO involves the selective cleavage of a C-O bond at one of the primary carbon atoms without breaking the C-C bonds of glycerol to eliminate a water molecule, forming acetol as an intermediate compound or side-product, and subsequently absorb a molecule of hydrogen through hydrogenation, thus, requiring a catalyst with dehydration sites which are usually acid sites and hydrogenation sites which usually need a transition metal.Cu is known to be active for C-O bond hydro-dehydration and less active for C-C bond cleavage [30].The catalytic activity of Cu-based catalyst on different supports such as SiO 2 [31], ZnO [32,33], Al 2 O 3 [34,35], Cr 2 O 3 [14], zeolite [36], MgO [2], etc., have been investigated.Most of these studies were carried out in a batch reactor, while a continuous-flow process would be more desirable due to the ease of the process scale-up and the potential for commercialization of the process.
Promoters are usually incorporated in a catalyst to enhance its activity and stability.A suitable promoter increases the catalyst surface area and dispersion of the catalyst particles by preventing the agglomeration and sintering of the metals and improves the mechanical strength of the catalyst.Rh, Pd, and silicotungstic acid (H 4 SiW 12 O 40 ) can be effective promoters for Cu-based catalysts for hydrogenolysis of glycerol to 1,2-PDO, however, the use of these expensive promoters in this process would limit its commercialization potential [17,34,37].The use of inexpensive promoters, such as boric acid, has been reported in Ni/SiO 2 [38] and Cu/SiO 2 [27] catalyst systems with excellent interaction with the metal atoms, resulting a high metal dispersion with a suitable acidity and an outstanding catalytic activity.Hence, it would be interesting to investigate the catalytic behavior of boric acid-doped Cu/Al 2 O 3 catalyst for the hydrogenolysis of glycerol.
The selective hydrogenolysis of glycerol to 1,2-PDO has been investigated either in the presence or absence of any solvent [39,40].For instance, Gandaris et al. demonstrated a novel catalytic conversion process by employing formic acid as both a solvent and a source of hydrogen [21,41].Chaminand et al. examined the influence of solvent (aqueous and organic) on the hydrogenolysis of glycerol over a Rh/C catalyst in a batch reactor at 180 • C, 80 bar H 2 , and for 168 h, and reported a higher glycerol conversion (32%) in the organic solvent (sulfolane) than in water (21%) [24].As an inexpensive green solvent, water is certainly more desirable than any organic solvents; however, it is challenging to carry out the selective glycerol hydrogenolysis reaction in aqueous medium since water is formed as a by-product during the reaction, which could create a thermodynamic barrier to shift the reaction in the forward direction.Considering the environmental impact of organic solvents and the green/low-cost nature of water, water has been commonly used as a solvent for the selective hydrogenolysis of glycerol to 1,2-PDO [29,42,43].
The use of crude glycerol as a feedstock for the synthesis of propylene glycol is a valuable attempt for the economical production of propylene glycol and the sustainability of the biodiesel industry.However, as mentioned earlier, crude glycerol contains various impurities derived from the biodiesel production processes, including water, sodium or potassium hydroxides, esters, fatty acids, and alcohols.When crude glycerol is used as a feedstock for the conversion reaction, the impurities would cause operating problems by either deactivating the catalyst or plugging the reactors [11].There is not much research carried so far on the hydrogenolysis of crude glycerol in a flow reactor.The present work is aimed to investigate on conversion of both pure glycerol and crude glycerol into 1,2-PDO in aqueous medium over inexpensive Cu/Al 2 O 3 catalysts promoted by boric acid in a continuous-flow reactor.Therefore, the results of this work will contribute to new knowledge in the literature and open a new window to utilize the crude glycerol for high-value products.
acids, and alcohols.When crude glycerol is used as a feedstock for the conversion reaction, the impurities would cause operating problems by either deactivating the catalyst or plugging the reactors [11].There is not much research carried so far on the hydrogenolysis of crude glycerol in a flow reactor.The present work is aimed to investigate on conversion of both pure glycerol and crude glycerol into 1,2-PDO in aqueous medium over inexpensive Cu/Al2O3 catalysts promoted by boric acid in a continuous-flow reactor.Therefore, the results of this work will contribute to new knowledge in the literature and open a new window to utilize the crude glycerol for high-value products.The scope of the present work is to study the performance of Cu-based catalysts loaded on highly-dispersed B2O3 (which would partially transfer to boric acid in the presence of water) on alumina supports for the glycerol hydrogenolysis reaction.The effects of various process parameters (Cu loading, B addition amount, temperature, H2 pressure, weight hourly space velocity, purity of the glycerol feedstock, etc.) on the reaction were also investigated.Moreover, the stability of the selected B2O3-loaded Cu-based catalyst was tested.

Catalyst Characterization
The textural properties of Cu/Al2O3 with B2O3 promoter measured by Brunauer-Emmett-Teller (BET) theory with N2 adsorption-desorption isotherms are presented in Table 1.In the Table, an initial improvement in the pore volume and the surface area of the catalyst can be observed by the incorporation of 0.25% B to Cu/Al2O3 sample indicating that B2O3 helps preventing the agglomeration Cu catalyst.However, the excess of B2O3 loading reduced the surface area and pore volume, which could be due to the coverage of the sample surface and blocking of some pores by B2O3.The scope of the present work is to study the performance of Cu-based catalysts loaded on highly-dispersed B 2 O 3 (which would partially transfer to boric acid in the presence of water) on alumina supports for the glycerol hydrogenolysis reaction.The effects of various process parameters (Cu loading, B addition amount, temperature, H 2 pressure, weight hourly space velocity, purity of the glycerol feedstock, etc.) on the reaction were also investigated.Moreover, the stability of the selected B 2 O 3 -loaded Cu-based catalyst was tested.

Catalyst Characterization
The textural properties of Cu/Al 2 O 3 with B 2 O 3 promoter measured by Brunauer-Emmett-Teller (BET) theory with N 2 adsorption-desorption isotherms are presented in Table 1.In the Table , an initial improvement in the pore volume and the surface area of the catalyst can be observed by the incorporation of 0.25% B to Cu/Al 2 O 3 sample indicating that B 2 O 3 helps preventing the agglomeration Cu catalyst.However, the excess of B 2 O 3 loading reduced the surface area and pore volume, which could be due to the coverage of the sample surface and blocking of some pores by B 2 O 3 .The XRD patterns of the reduced catalyst samples of Cu/Al 2 O 3 (containing 5 wt% Cu and 0-3 wt% B) are displayed in Figure 1.In this figure, all the catalysts have similar XRD patterns and the XRD peaks at 2θ = 36.3,45.5, 60.6, and 66.5 are ascribed to the X-ray diffraction of γ-Al 2 O 3 in these catalysts.From Figure 1, one can observe that there is a small peak at 44 The XRD patterns of the reduced catalyst samples of Cu/Al2O3 (containing 5 wt% Cu and 0-3 wt% B) are displayed in Figure 1.In this figure, all the catalysts have similar XRD patterns and the XRD peaks at 2θ = 36.3,45.5, 60.6, and 66.5 are ascribed to the X-ray diffraction of γ-Al2O3 in these catalysts.From Figure 1, one can observe that there is a small peak at 44 °C in 5Cu/Al2O3, typical of copper metal, disappeared after incorporation of B2O3.No X-ray diffraction lines of either Cu or B species were detected (although all catalysts contain 5 wt% Cu, above the detection limit of XRD) in 5Cu xB/Al2O3 catalysts, suggesting a high dispersion of the corresponding Cu particles in these catalysts.The reducibility of the catalysts was investigated using temperature-programmed reduction (TPR).Figure 2 illustrates the hydrogen TPR profiles of all the catalyst samples used in this study.Except 5Cu-0.25B/Al2O3,all other samples have a well-resolved single peak in the temperature range of 160-280 °C.The symmetric H2-TPR profile of the reduction peak indicates the homogeneous nature of the reduced samples and the formation of small, monodispersed metallic Cu particles [34], although in 5Cu-0.25B/Al2O3there exists a main reduction peak and a weak-shoulder peak in the temperature range of 180-220 °C , indicating the presence of two different Cu valence states (Cu +1 and Cu 0 ) [44].In the H2-TPR profiles (Figure 2), a shift in the reduction peak towards higher temperature with the increase in B content was observed.Similar observations have been reported in the literature where the authors ascribed the peak-shift to the strong interaction between CuO and B2O3 [27].
The catalyst acidity has an important role in the bifunctional mechanism (dehydration and hydrogenation) of selective hydrogenolysis of glycerol to 1,2-PDO [20].Therefore, NH3-TPD was used to investigate the strength of surface acid sites.The NH3-TPD profiles of the catalysts are presented in Figure 3. Ammonia is a basic molecule and is adsorbed by the acid sites of the catalyst.The higher is the ammonia desorption temperature, the stronger is acidity of the sites.The larger desorption area, the greater the acid sites.In the figure, a peak between 150-250 °C was observed for all the catalysts indicating that weak-intermediate acid sites were present on the catalyst surface [42].As clearly shown by the NH3-TPD profiles of the catalysts (presented in Figure 3), the intensity and area of desorption peaks gradually increased, reached maximum for 5Cu-3B/Al2O3 with the increase of boron content, proving the increase in acidity of the catalyst surface [38].Based on the peak areas, the number of acid sites in 5Cu-3B/Al2O3 is almost double that of the 5Cu/Al2O3 catalyst.The reducibility of the catalysts was investigated using temperature-programmed reduction (TPR).Figure 2 illustrates the hydrogen TPR profiles of all the catalyst samples used in this study.Except 5Cu-0.25B/Al 2 O 3 , all other samples have a well-resolved single peak in the temperature range of 160-280 • C. The symmetric H 2 -TPR profile of the reduction peak indicates the homogeneous nature of the reduced samples and the formation of small, monodispersed metallic Cu particles [34], although in 5Cu-0.25B/Al 2 O 3 there exists a main reduction peak and a weak-shoulder peak in the temperature range of 180-220 • C, indicating the presence of two different Cu valence states (Cu +1 and Cu 0 ) [44].
In the H 2 -TPR profiles (Figure 2), a shift in the reduction peak towards higher temperature with the increase in B content was observed.Similar observations have been reported in the literature where the authors ascribed the peak-shift to the strong interaction between CuO and B 2 O 3 [27].
The catalyst acidity has an important role in the bifunctional mechanism (dehydration and hydrogenation) of selective hydrogenolysis of glycerol to 1,2-PDO [20].Therefore, NH 3 -TPD was used to investigate the strength of surface acid sites.The NH 3 -TPD profiles of the catalysts are presented in Figure 3. Ammonia is a basic molecule and is adsorbed by the acid sites of the catalyst.The higher is the ammonia desorption temperature, the stronger is acidity of the sites.The larger desorption area, the greater the acid sites.In the figure, a peak between 150 and 250 • C was observed for all the catalysts indicating that weak-intermediate acid sites were present on the catalyst surface [42].As clearly shown by the NH 3 -TPD profiles of the catalysts (presented in Figure 3), the intensity and area of desorption peaks gradually increased, reached maximum for 5Cu-3B/Al 2 O 3 with the increase of boron content, proving the increase in acidity of the catalyst surface [38].Based on the peak areas, the number of acid sites in 5Cu-3B/Al 2 O 3 is almost double that of the 5Cu/Al 2 O 3 catalyst.
Based on the above characterization results, the roles of addition of B 2 O 3 may be summarized here: the addition of B 2 O 3 leads to higher reducibility of copper oxides and increased dispersion of Cu particles in the 5Cu-xB/Al 2 O 3 catalysts, which would account for the increased 1,2-PDO selectivity of the catalysts.On the other hand, the addition of B 2 O 3 results in increased acidity of the 5Cu-xB/Al 2 O 3 catalysts, which explains the enhanced glycerol conversion, as discussed previously.
Catalysts 2017, 7,196 5 of 16 Based on the above characterization results, the roles of addition of B2O3 may be summarized here: the addition of B2O3 leads to higher reducibility of copper oxides and increased dispersion of Cu particles in the 5Cu-xB/Al2O3 catalysts, which would account for the increased 1,2-PDO selectivity of the catalysts.On the other hand, the addition of B2O3 results in increased acidity of the 5Cu-xB/Al2O3 catalysts, which explains the enhanced glycerol conversion, as discussed previously.Based on the above characterization results, the roles of addition of B2O3 may be summarized here: the addition of B2O3 leads to higher reducibility of copper oxides and increased dispersion of Cu particles in the 5Cu-xB/Al2O3 catalysts, which would account for the increased 1,2-PDO selectivity of the catalysts.On the other hand, the addition of B2O3 results in increased acidity of the 5Cu-xB/Al2O3 catalysts, which explains the enhanced glycerol conversion, as discussed previously.

Influence of Cu Loading
The effects of Cu loading on the activity of Cu/Al 2 O 3 catalysts for glycerol hydrogenolysis were investigated at 230 • C, 5 MPa H 2 and 2 h −1 WHSV and the results are summarized in Table 2.It can be seen that with the increase of Cu loading, the glycerol conversion first increased and reached a maximum of 71% at a Cu loading of 5 wt%.This is attributed to the presence of extra active sites produced by the incorporation of Cu which accelerated the reaction process.When further increasing the Cu loading from 5 wt% to 15 wt%, the glycerol conversion and 1,2-PDO selectivity remained almost unchanged, likely due to the reduced dispersion of Cu particles and the blocked pores of the catalyst caused by agglomeration of excess Cu particles in the catalysts with too high Cu loadings.However, the selectivity of 1,2-PDO remains almost unaffected by catalyst loading and was found to be in the range of 85-87% in all the cases.Since 5 wt% loading of Cu metal over alumina demonstrated a promising catalytic performance, it was selected for all the further experiments.From the Table, the main by-products from Cu/Al 2 O 3 catalyzed glycerol hydrogenolysis are acetol, ethylene glycol (EG), plus relatively much smaller amounts of compounds denoted as "others" in Table 2.

Influence of B Incorporation
The catalytic effect of B was studied on 5Cu/Al 2 O 3 at the reaction conditions of 230 • C, 6 MPa H 2 , 2 h −1 (WHSV), and at a hydrogen flow of 100 mL/min, and the results are summarized in Figure 4.The glycerol conversion and 1,2-PDO selectivity for the catalyst 5Cu/Al 2 O 3 without B was 73% and 87% , respectively.As B was introduced into the 5Cu/Al 2 O 3 catalyst, both catalytic activity and 1,2-PDO selectivity were improved significantly.At these experimental conditions, 5Cu-1B/Al 2 O 3 demonstrated the superior performance among other B 2 O 3 incorporated catalysts and achieved 80% glycerol conversion with 98% selectivity towards 1,2-PDO.Similar results have been reported by Zhu et al. for B 2 O 3 -doped Cu/SiO 2 catalysts for the synthesis of propylene glycol [27].The authors attributed the enhancement in the glycerol conversion to the synergistic effect caused by Cu-B surface interaction which accelerated the surface activity of Cu metal.The improvement in the selectivity towards 1,2-PDO was directly associated with the enhanced hydrogenation activity of Cu towards acetol.A further increase in boron content reduced the glycerol conversion and 1, 2-PDO selectivity, which might be due to the masking effect of boron over the Cu catalyst surface and the pores, as evidenced by the substantial decreases in both BET surface area and total pore volume (Table 1).
There are two possible explanations for the enhancement effects of B 2 O 3 .First, B 2 O 3 can act as a Lewis acid, promoting the dehydration of glycerol to acetol.Secondly, in the calcinations stage, Cu(NO 3 ) 2 would first decompose to form CuO, then CuO reacts with B 2 O 3 to from copper borate, which will convert to well-dispersed nano Cu particles after H 2 reduction.B 2 O 3 is usually added to impart thermal stability of a catalyst [45].Tan et al. used CoB/Al 2 O 3 in Fischer-Tropsch synthesis (under conditions comparatively harsher than ours) and reported enhanced thermal stability of the catalyst owing to the presence of boron oxide at the surface sites of the catalyst [45].However, the mechanism of thermal stabilization of boron incorporated catalysts is not clear.One possible explanation is owing to the strong interaction between Ni and B, and Co and B in Ni-B/SiO 2 and Co-B/MCM-41, respectively, leading to suppression of metal sintering and improvement in the dispersion of active species [38,45].Yin et al. reported that the incorporation of B 2 O 3 to Cu-based catalysts generated more active sites of Cu in dimethyl oxalate hydrogenation [46].He et al. [30] proposed that the acidity and electron affinity of B 2 O 3 are higher than that of SiO 2 support, which is favorable for the formation of strong interactions between Cu and boron species in Cu-B/SiO 2 catalysts [47,48].Thus, in the present study, the stabilizing effect of B 2 O 3 could also be due to the strong interaction between Cu and B species which helped retard the aggregation of Cu particles during calcination, reduction, and reaction, as evidenced by H 2 -TPR profiles (Figure 2) clearly indicating a shift in the reduction peak towards higher temperature with an increase in B content.
In addition, as evidenced by the NH 3 -TPD profiles of the non-reduced catalysts (presented in the Figure 3), the addition of B 2 O 3 enhanced the acidity of the Cu/Al 2 O 3 catalysts, as similarly reported in the literature [38].The improved acidity would also play a role in the enhancement of catalytic activities of the catalysts.proposed that the acidity and electron affinity of B2O3 are higher than that of SiO2 support, which is favorable for the formation of strong interactions between Cu and boron species in Cu-B/SiO2 catalysts [47,48].Thus, in the present study, the stabilizing effect of B2O3 could also be due to the strong interaction between Cu and B species which helped retard the aggregation of Cu particles during calcination, reduction, and reaction, as evidenced by H2-TPR profiles (Figure 2) clearly indicating a shift in the reduction peak towards higher temperature with an increase in B content.
In addition, as evidenced by the NH3-TPD profiles of the non-reduced catalysts (presented in the Figure 3), the addition of B2O3 enhanced the acidity of the Cu/Al2O3 catalysts, as similarly reported in the literature [38].The improved acidity would also play a role in the enhancement of catalytic activities of the catalysts.

Influence of Temperature
The reaction temperature dependence of glycerol hydrogenolysis over 5Cu-B/Al2O3 is illustrated in Figure 5.As expected, the glycerol conversion improved dramatically from 15% (170 °C) to 99% (270 °C) as the temperature elevated.While, an initial insignificant decrease in the 1,2-PDO selectivity from 98% (180 °C) to 96% (250 °C) followed by a remarkable decrease to 85% at 270 °C was observed.It is assumed that the low selectivity at high temperatures is related to the formation of large amounts of undesired by-products, such as the over-hydrogenolysis products 1-propanol, 2-propanol, and the degradation products methanol, ethanol, and ethylene glycol.The selectivity to 1,2-PDO remains high in the moderate range of temperature (200-250 °C), indicating that it is favorable to promote the hydrogenolysis of glycerol to 1,2-PDO in this range of temperatures, as described previously [43].
The reaction mechanisms of glycerol hydrogenolysis to 1,2-propanediol has been reported in the literature [49,50].Based on the products formed in the present study it is believed that the reaction involves a dehydration and hydrogenation mechanism in which dehydration of glycerol forms acetol which undergoes hydrogenation in the next step to form 1,2-propanediol.The dehydration mainly controlled by the acidity of the catalyst and the hydrogenation occurs on the active metal sites [18].The other steps involved in the process include the excess hydrogenolysis of 1,2-PDO to form 1propanol/2-propanol, cleavage of the C-C bond in glycerol to give ethylene glycol and methanol.

Influence of Temperature
The reaction temperature dependence of glycerol hydrogenolysis over 5Cu-B/Al 2 O 3 is illustrated in Figure 5.As expected, the glycerol conversion improved dramatically from 15% (170 • C) to 99% (270 • C) as the temperature elevated.While, an initial insignificant decrease in the 1,2-PDO selectivity from 98% (180 • C) to 96% (250 • C) followed by a remarkable decrease to 85% at 270 • C was observed.It is assumed that the low selectivity at high temperatures is related to the formation of large amounts of undesired by-products, such as the over-hydrogenolysis products 1-propanol, 2-propanol, and the degradation products methanol, ethanol, and ethylene glycol.The selectivity to 1,2-PDO remains high in the moderate range of temperature (200-250 • C), indicating that it is favorable to promote the hydrogenolysis of glycerol to 1,2-PDO in this range of temperatures, as described previously [43].
The reaction mechanisms of glycerol hydrogenolysis to 1,2-propanediol has been reported in the literature [49,50].Based on the products formed in the present study it is believed that the reaction involves a dehydration and hydrogenation mechanism in which dehydration of glycerol forms acetol which undergoes hydrogenation in the next step to form 1,2-propanediol.The dehydration mainly controlled by the acidity of the catalyst and the hydrogenation occurs on the active metal sites [18].The other steps involved in the process include the excess hydrogenolysis of 1,2-PDO to form 1-propanol/2-propanol, cleavage of the C-C bond in glycerol to give ethylene glycol and methanol.

Influence of Hydrogen Pressure
Figure 6 shows effects of hydrogen pressure (2-8 MPa) on the performance of 5Cu-1B/Al2O3 catalyst for glycerol hydrogenolysis (240 °C, 10 wt% aq.glycerol and WHSV 0.4 h −1 ).Generally, both the glycerol conversion and 1,2-PDO selectivity increased by increasing the hydrogen pressure from 2 MPa to 6 MPa, as expected.A further increase in hydrogen pressure did not result in any significant increase in the glycerol conversion and 1,2-PDO selectivity.At 6 MPa H2, the glycerol conversion and 1,2-PDO selectivity attained 45% and 95%, respectively.The low hydrogen pressure (2 MPa) caused incomplete hydrogenation of acetol to 1,2-PDO.

Influence of Hydrogen Pressure
Figure 6 shows effects of hydrogen pressure (2-8 MPa) on the performance of 5Cu-1B/Al 2 O 3 catalyst for glycerol hydrogenolysis (240 • C, 10 wt% aq.glycerol and WHSV 0.4 h −1 ).Generally, both the glycerol conversion and 1,2-PDO selectivity increased by increasing the hydrogen pressure from 2 MPa to 6 MPa, as expected.A further increase in hydrogen pressure did not result in any significant increase in the glycerol conversion and 1,2-PDO selectivity.At 6 MPa H 2 , the glycerol conversion and 1,2-PDO selectivity attained 45% and 95%, respectively.The low hydrogen pressure (2 MPa) caused incomplete hydrogenation of acetol to 1,2-PDO.

Influence of Hydrogen Pressure
Figure 6 shows effects of hydrogen pressure (2-8 MPa) on the performance of 5Cu-1B/Al2O3 catalyst for glycerol hydrogenolysis (240 °C, 10 wt% aq.glycerol and WHSV 0.4 h −1 ).Generally, both the glycerol conversion and 1,2-PDO selectivity increased by increasing the hydrogen pressure from 2 MPa to 6 MPa, as expected.A further increase in hydrogen pressure did not result in any significant increase in the glycerol conversion and 1,2-PDO selectivity.At 6 MPa H2, the glycerol conversion and 1,2-PDO selectivity attained 45% and 95%, respectively.The low hydrogen pressure (2 MPa) caused incomplete hydrogenation of acetol to 1,2-PDO.

Influence of Weight Hourly Space Velocity (WHSV)
Effects of WHSV (0.05-0.8 h −1 ) on the performance of 5Cu-1B/Al 2 O 3 catalyst for glycerol hydrogenolysis (250 • C, 6 MPa H 2 , 10 wt% aq.glycerol) were studied by changing the flow rate of the feedstock and the results are shown in Figure 7.It is evident that the glycerol conversion drops with increasing WHSV because of the shortened residence time.However, the selectivity of 1,2-PDO remains almost unchanged (96-98%) when the WHSV varies between 0.1 and 0.8 h −1 , but it was as low as 78% when the WHSV was reduced to 0.05 h −1 , which is likely caused by the excessive hydrogenolysis reaction converting 1,2-PDO to ethylene glycol and other lower alcohols like ethanol and methanol at too long a residence time.Hence, to get a good conversion of glycerol with high selectivity to 1,2-PDO, the optimal WHSV is likely 0.1 h −1 .

Influence of Weight Hourly Space Velocity (WHSV)
Effects of WHSV (0.05-0.8 h −1 ) on the performance of 5Cu-1B/Al2O3 catalyst for glycerol hydrogenolysis (250 °C, 6 MPa H2, 10 wt% aq.glycerol) were studied by changing the flow rate of the feedstock and the results are shown in Figure 7.It is evident that the glycerol conversion drops with increasing WHSV because of the shortened residence time.However, the selectivity of 1,2-PDO remains almost unchanged (96-98%) when the WHSV varies between 0.1 and 0.8 h −1 , but it was as low as 78% when the WHSV was reduced to 0.05 h −1 , which is likely caused by the excessive hydrogenolysis reaction converting 1,2-PDO to ethylene glycol and other lower alcohols like ethanol and methanol at too long a residence time.Hence, to get a good conversion of glycerol with high selectivity to 1,2-PDO, the optimal WHSV is likely 0.1 h −1 .

Influence of Glycerol Feedstock Purity
One of the objectives of the present work was to evaluate the possibility of using low-grade glycerol for hydrogenolysis to 1,2-PDO.Specifically, crude glycerol (54.7% purity) and technical grade glycerol (91.6% purity) were tested, in comparison to pharmaceutical grade glycerol (99.9% purity).The mass composition of different grades of glycerol used in this work is given in Table 3.It was assumed that the presence of impurities would adversely affect the performance of the catalyst, as discussed previously.For instance, the presence of water would impose a thermodynamic barrier, limiting the reaction.The salt impurities could deactivate the catalyst surface and other organic impurities present in crude glycerol could compete with glycerol in the adsorption process on the catalyst surface, hence, reducing the glycerol conversion and 1,2-PDO selectivity.

Influence of Glycerol Feedstock Purity
One of the objectives of the present work was to evaluate the possibility of using low-grade glycerol for hydrogenolysis to 1,2-PDO.Specifically, crude glycerol (54.7% purity) and technical grade glycerol (91.6% purity) were tested, in comparison to pharmaceutical grade glycerol (99.9% purity).The mass composition of different grades of glycerol used in this work is given in Table 3.It was assumed that the presence of impurities would adversely affect the performance of the catalyst, as discussed previously.For instance, the presence of water would impose a thermodynamic barrier, limiting the reaction.The salt impurities could deactivate the catalyst surface and other organic impurities present in crude glycerol could compete with glycerol in the adsorption process on the catalyst surface, hence, reducing the glycerol conversion and 1,2-PDO selectivity.Figure 8 shows the glycerol conversions and selectivity of different products achieved with different grades of glycerol with 5Cu-1B/Al 2 O 3 catalyst at the reaction conditions of 250 • C, 10 wt% aq.solution, 6 MPa, WHSV 0.1 h −1 .As expected, reactions performed with technical grade and crude glycerol resulted in substantially reduced glycerol conversion and 1,2-PDO selectivity, which clearly indicates the negative impact of the impurities in the glycerol feedstock on the hydrogenolysis of glycerol due to the deactivation of the catalyst [11].
Catalysts 2017, 7, 196 10 of 16 aq.solution, 6 MPa, WHSV 0.1 h −1 .As expected, reactions performed with technical grade and crude glycerol resulted in substantially reduced glycerol conversion and 1,2-PDO selectivity, which clearly indicates the negative impact of the impurities in the glycerol feedstock on the hydrogenolysis of glycerol due to the deactivation of the catalyst [11].

Long-Term Stability and Catalyst Deactivation
The long-term performance of hydrogenolysis of pure glycerol over 5Cu-1B/Al2O3 catalyst was tested at 250 °C, 6 MPa H2 flow, and 0.1 h −1 , and the results are given in Figure 9.No sign of any decline in the catalyst activity (>95% glycerol conversion and >97% 1,2-PDO selectivity) was observed up to 60 h, despite the harsh reaction conditions, which suggest the promise of the 5Cu-1B/Al2O3 catalyst for industrial applications.After this time, the glycerol conversion gradually decreased.Meanwhile, the product distribution did not show any appreciable change during this period.These results are in good-agreement with those reported in the literature [34].The list of different products with their selectivity after the first hour of reaction is given in Table 4.
Deactivation of the catalyst was observed after 60 h on stream, as shown in Figure 9. Usually, in a heterogeneous system, the catalyst deactivation occurs due to destruction of the support structure, sintering, coking, fouling, or leaching of the catalyst [51].Comparing the surface area and the pore volume of the fresh and spent catalyst (Table 1), it is revealed that both the BET surface area and total pore volume for the spent catalyst were reduced, suggesting the destruction of the support, sintering of the Cu metal, or deposition of fouling materials inside the pores of the 5Cu-1B/Al2O3 catalyst might occur during the long-term test.To prove this hypothesis, TEM, TGA, and ICP-AES analysis were performed on the fresh and spent (after 70 h on stream) catalysts of 5Cu-1B/Al2O3.TEM micrographs of fresh and spent catalyst are illustrated in Figure 10a,b, where the presence of Cu particles was confirmed by Energy Dispersive X-Ray Analysis (EDX).Limited by the magnification of the TEM instrument, however, only clusters of the Cu particles were observable in the fresh and spent catalysts.
Figure 11 shows the TG thermogram of fresh and spent catalysts of 5Cu-1B/Al2O3 after 70 h on stream.The TGA measurements were performed at 10 °C/min over a temperature range of 50 °C to 800 °C under a constant flow of air of 20 mL/min.From the TG thermograms of the fresh and spent catalysts, a total of 30 wt% weight loss was observed over the range of 50 °C to 800 °C for the spent catalyst, compared with only <8 wt% weight loss for the fresh catalyst.This result may evidence the

Long-Term Stability and Catalyst Deactivation
The long-term performance of hydrogenolysis of pure glycerol over 5Cu-1B/Al 2 O 3 catalyst was tested at 250 • C, 6 MPa H 2 flow, and 0.1 h −1 , and the results are given in Figure 9.No sign of any decline in the catalyst activity (>95% glycerol conversion and >97% 1,2-PDO selectivity) was observed up to 60 h, despite the harsh reaction conditions, which suggest the promise of the 5Cu-1B/Al 2 O 3 catalyst for industrial applications.After this time, the glycerol conversion gradually decreased.Meanwhile, the product distribution did not show any appreciable change during this period.These results are in good-agreement with those reported in the literature [34].The list of different products with their selectivity after the first hour of reaction is given in Table 4.
Deactivation of the catalyst was observed after 60 h on stream, as shown in Figure 9. Usually, in a heterogeneous system, the catalyst deactivation occurs due to destruction of the support structure, sintering, coking, fouling, or leaching of the catalyst [51].Comparing the surface area and the pore volume of the fresh and spent catalyst (Table 1), it is revealed that both the BET surface area and total pore volume for the spent catalyst were reduced, suggesting the destruction of the support, sintering of the Cu metal, or deposition of fouling materials inside the pores of the 5Cu-1B/Al 2 O 3 catalyst might occur during the long-term test.To prove this hypothesis, TEM, TGA, and ICP-AES analysis were performed on the fresh and spent (after 70 h on stream) catalysts of 5Cu-1B/Al 2 O 3 .TEM micrographs of fresh and spent catalyst are illustrated in Figure 10a,b, where the presence of Cu particles was confirmed by Energy Dispersive X-Ray Analysis (EDX).Limited by the magnification of the TEM instrument, however, only clusters of the Cu particles were observable in the fresh and spent catalysts.
Figure 11 shows the TG thermogram of fresh and spent catalysts of 5Cu-1B/Al 2 O 3 after 70 h on stream.The TGA measurements were performed at 10 • C/min over a temperature range of 50 • C to 800 • C under a constant flow of air of 20 mL/min.From the TG thermograms of the fresh and spent catalysts, a total of 30 wt% weight loss was observed over the range of 50 • C to 800 • C for the spent catalyst, compared with only <8 wt% weight loss for the fresh catalyst.This result may evidence the deposition of fouling materials due to polymerization of glycerol on the spent catalyst, which could contribute to the deactivation of the catalyst by blocking the catalyst active sites.
Moreover, the concentration of Cu, B and Al in the fresh and the spent catalyst was measured by ICP-AES and given in Table 1.A negligible change in the concentration of Cu between the fresh and spent catalyst was observed (4.79% to 4.46%) indicating a trivial role of leaching on the catalyst deactivation.However, a significant change in the concentration of B and Al was observed in the spent catalyst indicating the destruction of the support structure in a long run that facilitates the deactivation of the catalyst.
Catalysts 2017, 7, 196 11 of 16 deposition of fouling materials due to polymerization of glycerol on the spent catalyst, which could contribute to the deactivation of the catalyst by blocking the catalyst active sites.Moreover, the concentration of Cu, B and Al in the fresh and the spent catalyst was measured by ICP-AES and given in Table 1.A negligible change in the concentration of Cu between the fresh and spent catalyst was observed (4.79% to 4.46%) indicating a trivial role of leaching on the catalyst deactivation.However, a significant change in the concentration of B and Al was observed in the spent catalyst indicating the destruction of the support structure in a long run that facilitates the deactivation of the catalyst.deposition of fouling materials due to polymerization of glycerol on the spent catalyst, which could contribute to the deactivation of the catalyst by blocking the catalyst active sites.Moreover, the concentration of Cu, B and Al in the fresh and the spent catalyst was measured by ICP-AES and given in Table 1.A negligible change in the concentration of Cu between the fresh and spent catalyst was observed (4.79% to 4.46%) indicating a trivial role of leaching on the catalyst deactivation.However, a significant change in the concentration of B and Al was observed in the spent catalyst indicating the destruction of the support structure in a long run that facilitates the deactivation of the catalyst.

Catalyst Preparation
Firstly, a Cu/Al2O3 catalyst was prepared by wet impregnation method [34] using a calculated amount of water-soluble metal salt of copper (II) nitrate hydrate [Cu(NO3)2•3H2O] and γ-Al2O3 as the support material.The catalyst was then dried at 90 °C for 12 h to form the Cu/Al2O3 precursor.The B2O3-modified Cu/Al2O3 catalysts were prepared by incipient wetness impregnation of the Cu/Al2O3 precursor with aqueous solutions containing the desired amount of H3BO3.After impregnation, these samples were dried overnight at 90 °C and then calcined at 400 °C for 5 h under a N2 flow of 20 mL/min at a heating rate of 2 °C/min.The obtained catalysts are designated as xCu-yB/Al2O3, where x and y represent the mass loading of copper and boron, respectively.

Catalyst Characterization
The surface area, total pore volume, and average pore diameter of the selected catalysts were determined by nitrogen isothermal (at −196 °C) adsorption with a Micromeritics ASAP 2010 BET apparatus after degassing the samples at 300 °C for 8 h in vacuum.
The acidity of the catalysts were measured by an ammonia temperature programmed desorption (NH3-TPD) test using Micromeritics AutoChem II analyzer.Around 0.35 g of the non-reduced calcined catalyst was pretreated in He at 400 °C to remove moisture and other adsorbed gases on the surface for 1 h.After cooling to 100 °C, the catalyst was saturated with pure NH3 for 30 min, and then purged with He to remove the physiosorbed NH3 for 30 min.The sample was heated to 700 °C at a ramp rate of 5 °C/min and the NH3 desorbed was detected by a mass spectrometer.

Catalyst Preparation
Firstly, a Cu/Al 2 O 3 catalyst was prepared by wet impregnation method [34] using a calculated amount of water-soluble metal salt of copper (II) nitrate hydrate [Cu(NO 3 ) 2 •3H 2 O] and γ-Al 2 O 3 as the support material.The catalyst was then dried at 90 • C for 12 h to form the Cu/Al 2 O 3 precursor.The B 2 O 3 -modified Cu/Al 2 O 3 catalysts were prepared by incipient wetness impregnation of the Cu/Al 2 O 3 precursor with aqueous solutions containing the desired amount of H 3 BO 3 .After impregnation, these samples were dried overnight at 90 • C and then calcined at 400 • C for 5 h under a N 2 flow of 20 mL/min at a heating rate of 2 • C/min.The obtained catalysts are designated as xCu-yB/Al 2 O 3 , where x and y represent the mass loading of copper and boron, respectively.

Catalyst Characterization
The surface area, total pore volume, and average pore diameter of the selected catalysts were determined by nitrogen isothermal (at −196 • C) adsorption with a Micromeritics ASAP 2010 BET apparatus after degassing the samples at 300 • C for 8 h in vacuum.
The acidity of the catalysts were measured by an ammonia temperature programmed desorption (NH 3 -TPD) test using Micromeritics AutoChem II analyzer.Around 0.35 g of the non-reduced calcined catalyst was pretreated in He at 400 • C to remove moisture and other adsorbed gases on the surface for 1 h.After cooling to 100 • C, the catalyst was saturated with pure NH 3 for 30 min, and then purged with He to remove the physiosorbed NH 3 for 30 min.The sample was heated to 700 • C at a ramp rate of 5 • C/min and the NH 3 desorbed was detected by a mass spectrometer.
The crystal structures of selected catalysts were examined by powder X-ray diffraction (XRD) with a PANalytical X'Pert Pro diffractometer with Cu Kα as the radiation source.Step-scans were taken over the range of 2θ from 6 to 95 • .
Temperature-programmed reduction (TPR) profiles of the catalysts were determined using a Micromeritics Autochem 2920 equipped with a thermal conductivity detector (TCD).These catalysts were first heated from ambient temperature to 550 • C at 10 • C/min under a 5% O 2 /He mixture flow at 50 mL/min for pre-treatment and then exposed to a flowing gas composed of H 2 and Ar (v/v = 1:9) at 50 mL/min and were heated from room temperature to 700 • C at a heating rate of 10 • C/min.
The morphologies of the spent catalyst were analyzed using a JEOL 2100F transmission electron microscope equipped with an energy-dispersive X-ray spectroscopy (EDS-INCA system from Oxford Instrument).
The thermogravimetric analysis (TGA) of the fresh/spent catalysts was conducted on a thermogravimetric analysis unit (Perkin Elmer Pyris 1 TGA unit).The TGA measurements were performed at 10 • C/min over a temperature range of 50 • C to 800 • C under a constant flow of air of 20 mL/min.
The ICP-AES analysis of the selective fresh and spent catalysts was performed by a certified analytical lab-Lakehead University Instrumentation Lab-following its well-established protocol.

Catalytic Tests
The hydrogenolysis of glycerol was carried out in a bench scale continuous down-flow tubular reactor (Inconel 316 tubing, 9.55 mm OD, 6.34 mm ID and 600 mm length) heated with an electric furnace.Typically, around 2.0 g of the catalyst was loaded in the constant temperature section of the reactor and supported on a porous Inconel metal disc (pore size: 100 µm) and some quartz wool.Prior to each run, the catalyst was reduced in situ in flowing H 2 (100 cm 3 /min) at 300 • C for 3 h at atmospheric pressure.The feed-a 10 wt% aqueous solution of glycerol (unless otherwise mentioned)-was pumped using a HPLC pump (Eldex) at a predetermined flow rate into the reactor.This translates to a corresponding WHSV, which is the mass of the glycerol per mass of catalyst per hour.All of the experiments are performed at a desired temperature and pressure (controlled by a temperature controller and a back-pressure controller, respectively) along with co-feeding of H 2 gas.The liquid and gas products were cooled and collected in a gas-liquid separator immersed in an ice-water trap.

Product Analysis
All the liquid components in the reaction mixture were analyzed by GC-MS on a Varian 1200 Quadrupole MS (EI) and Varian CP-3800 GC with VF-5 MS column (5% phenyl/95% dimethyl-polysiloxane, 30 m × 0.25 mm × 0.25 µm) using helium as the carrier gas at a flow rate of 0.5 mL/s.The oven temperature was maintained at 70 • C for 1 min and then increased to 290 • C at 40 • C/min.Injector and detector temperature were 300 • C. The components were identified by the NIST 98 MS library.Quantification of the chemical composition was performed on a GC-FID (Shimadzu-2010), from Shimadzu Scientific Instruments (Guelph, ON, Canada), calibrated with 1,2-PDO (99.9%), ethylene glycol (99.9%), and acetol (99.8%).Dimethyl sulfoxide (DMSO) was used as the internal standard.The GC-FID analysis was carried out using the similar separation conditions as mentioned above for the GC-MS analysis.The gas samples were analyzed by a GC-TCD (Agilent 3000 Micro-GC).
The reported yield and conversion are values after 4 h on-stream unless otherwise specified.Herewith, the product yield, glycerol conversion, and product selectivity are defined as follows:

Conclusions
The addition of B 2 O 3 to Cu/Al 2 O 3 catalysts enhanced the catalytic activity for the hydrogenolysis of glycerol to 1,2-PDO.Among all catalysts prepared and tested, 5Cu-1B/Al 2 O 3 demonstrated the best catalytic performance with 98 ± 2% glycerol conversion and 98 ± 2% 1,2-PDO selectivity in hydrogenolysis of 10 wt% aqueous solution of glycerol at the optimum conditions (250 • C, 6 MPa H 2 pressure, and 0.1 h −1 WHSV).Process parameters, such as temperature, hydrogen pressure, and liquid hourly space velocity, significantly influenced the catalytic activity for the glycerol hydrogenolysis reaction.The use of different grades of glycerol, including pharmaceutical-grade glycerol, technical-grade glycerol, and crude glycerol (glycerol purity varying from 54.7 to 99.9%) showed that the presence of impurities could reduce the glycerol conversion and 1,2-PDO selectivity.The long-term stability test demonstrated that the 5Cu-1B/Al 2 O 3 catalyst could be used up to 60 h without any appreciable change in activity.Destruction of the support structure, sintering of Cu metal, and coke deposition on the catalyst might be the main factors that deactivated the catalyst after 60 h on stream.The promoting effects of boron for the activities of Cu/Al 2 O 3 catalysts might be attributed to the strong interaction between Cu and B species that helped retard the aggregation of Cu particles during calcination, reduction, and reaction, as evidenced by H 2 -TPR profiles.In addition, as evidenced by the NH 3 -TPD profiles of the catalysts, the addition of B 2 O 3 enhanced the acidity of the Cu/Al 2 O 3 catalysts, which would also play a role in the enhancement of catalytic activities of the catalysts.

Figure 1 .
Figure 1.XRD pattern of the fresh Cu/Al2O3 catalysts loaded with various amounts of B2O3.

3 Figure 1 .
Figure 1.XRD pattern of the fresh Cu/Al 2 O 3 catalysts loaded with various amounts of B 2 O 3 .

Figure 2 .
Figure 2. H2-TPR profiles of the fresh Cu/Al2O3 catalysts loaded with various amounts of B2O3.

3 Figure 2 .
Figure 2. H 2 -TPR profiles of the fresh Cu/Al 2 O 3 catalysts loaded with various amounts of B 2 O 3 .

Figure 2 .
Figure 2. H2-TPR profiles of the fresh Cu/Al2O3 catalysts loaded with various amounts of B2O3.

Figure 8 Figure 7 .
Figure8shows the glycerol conversions and selectivity of different products achieved with different grades of glycerol with 5Cu-1B/Al2O3 catalyst at the reaction conditions of 250 °C, 10 wt%

Figure 11 .
Figure 11.Thermogravimetric analysis of fresh and spent catalyst of 5Cu-1B/Al2O3 after 70 h on stream.

Figure 11 .
Figure 11.Thermogravimetric analysis of fresh and spent catalyst of 5Cu-1B/Al 2 O 3 after 70 h on stream.

Table 1 .
Textural properties of the fresh/spent Cu/Al2O3 catalysts loaded with various amounts of B2O3 determined by N2 adsorption-desorption.

Table 1 .
Textural properties of the fresh/spent Cu/Al 2 O 3 catalysts loaded with various amounts of B 2 O 3 determined by N 2 adsorption-desorption.

Table 3 .
Composition of different grades of glycerol.

Table 3 .
Composition of different grades of glycerol.

Table 4 .
List of different products with their selectivity.

Table 4 .
List of different products with their selectivity.

Table 4 .
List of different products with their selectivity.