Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
Screening of four commercial catalysts (Pt/Al2O3, Pt/SiO2, Pd/Al2O3, and Pd/SiO2) and four acidic additives (hydrochloric, tungstic, phosphotungstic, and silicotungstic acids) shows that the combination of a platinum hydrogenation catalyst with tungsten containing acidic additives yields 1,3-propanediol from aqueous glycerol. The performance of the best catalytic system Pt/Al2O3 with silicotungstic acid as an additive was optimized by experimental design, capturing the influence of reaction time, glycerol concentration, acid concentration, pressure, and temperature on the formation of 1,3-propanediol from glycerol. High 1,3-propanediol yield in an aqueous batch system can be achieved (49 % conversion, 28 % selectivity) with excellent 1,3-propanediol to 1,2-propanediol ratios. A mechanistic interpretation is given for this bifunctional system, supported by the relative stability of 1,3-propanediol in comparison with 1,2-propanediol under the chosen reaction conditions.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Renewable glycerol is the starting material in a wide range of processes and can be converted into a plethora of products.1 A product of particular interest is 1,3-propanediol (13PD), which can be used to make polyester polypropylene terephthalate. 13PD can be produced by hydrogenolysis. Although its regioisomer 1,2-propanediol (12PD) can be produced at both high conversion and selectivity,2–4 this proves to be more challenging for 13PD (Table 1).5–16
Table 1. State of the art in selective hydrogenolysis of glycerol to 1,3-propanediol.
Hydrogenolysis is often mistakenly identified as the process in question, however, in most instances, the process in question is an elimination of water followed by hydrogenation of the formed double bond on a metallic surface.17, 18 The catalytic role of the acidic components in the catalytic systems listed in Table 1 is the initial dehydration of glycerol. This is the key step in defining the selectivity of the catalyst. If one of the two primary hydroxyl groups is eliminated, 12PD will be obtained. Only if the secondary alcohol is eliminated will 13PD be formed (Scheme 1).
The initial dehydration to form acetol is thermodynamically favored over the formation of 3-hydroxypropanal (Scheme 2).15 This exemplifies the difficulties in achieving high 13PD selectivities and it can be extrapolated that the formation of 13PD is kinetically controlled. The formation of a relatively stable secondary carbocation as an intermediate during the formation of 13PD, resulting from the acid catalyzed alcohol elimination, is at the heart of kinetic control.17, 19
Good conversions and 13PD selectivities have been obtained by using either water or 1,3-dimethyl-2-imidazolidinone (DMI) as a solvent (Table 1). However, from an environmental point of view, the use of water is clearly preferable.
Among the catalytic systems that display the highest 13PD selectivity, several tungsten containing species are present (Table 1). The tungsten compounds that have been tested include tungsten oxide, tungstic acid, and immobilized silicotungstic acid.7, 8, 12, 13, 15, 16 However, the role of the tungsten species has not yet been elucidated. Apparently, tungsten plays a role in 13PD selectivity that is not yet understood.
These two conclusions from Table 1 prompted this investigation to detail the role of several tungsten containing acids in the conversion of aqueous glycerol.
The catalysts used in this work were purchased from Acros Organics (5 % Pt/Al2O3, 5 % Pd/Al2O3), Sigma Aldrich (1 % Pt/SiO2), and Strem (5 % Pd/SiO2). 12PD 99 %, n-propanol 99.7 % (nPrOH), ethylene glycol 99.8 % (EG), silicotungstic acid ≥99.9 % (STA), phosphotungstic acid (PTA), and tungstic acid ≥99.0 % were purchased from Sigma Aldrich, glycerol 99+ % and 1,3-propanediol 98 % (13PD) were purchased from Acros Organics, ethanol 99.8 % (EtOH) was purchased from Merck, and methanol 99.8 % (MeOH) and hydrochloric acid 36–38 % were purchased from J.T. Baker. All the chemicals were used as received and the catalysts were not pretreated before reaction.
Preparation of reaction mixture
Aqueous reaction mixtures were prepared by dissolution of appropriate amounts of glycerol and acid in demineralized water in a volumetric flask.4 The hydrogenolysis of glycerol was performed in a HEL PolyBLOCK 8, a parallel autoclave reactor system consisting of eight 16 mL vessels. The appropriate amount of catalyst (25 μmol active metal, relative to glycerol) was added to 5 mL of aqueous reaction mixture. Reactors were purged 3 times with N2 (20 bar) and 3 times with H2 (40 bar). They were then pressurized to the appropriate pressure, stirred at 800 rpm and heated to the reaction temperature within 20 min. Stirring was terminated after the indicated reaction time and the reactors were allowed to cool down to RT. The reaction mixture was filtered over a nylon micro filter (Rotilabo, 0.2 μm). An HPLC sample was prepared by diluting filtered reaction mixture using H2SO4 (5 mm). The liquid products were analyzed by HPLC. The products that were detected by HPLC were 13PD, 12PD, ethylene glycol (EG), nPrOH, and ethanol (EtOH).4, 20
After a standard reaction procedure, the reaction mixture was centrifuged at 900 g using a Heraeus Megafuge 2.0R centrifuge. The supernatant was removed for sampling and 5.5 mL fresh reaction mixture was added to the centrifuged catalyst. Reaction mixture (0.5 mL) was taken for t=0 h HPLC sample and a second iteration was performed to the remaining reaction mixture, using the standard reaction procedure.
Analysis of the Catalysts
Porosity and surface area were determined by nitrogen physisorption measurements (Table 2) on a Quantachrome Autosorb-6B at 77 K. Prior to the measurements, the samples were degassed overnight under vacuum at 230 °C using a Quantachrome Autosorb degasser.
Table 2. Porosity and BET surface area of catalysts.
Metal loading [wt %]
Vpore [mL g−1]
SBET [m2 g−1]
Transmission electron microscopy (TEM) was performed using a FEI Tecnai TF20 electron microscope with a field emission gun as an electron source and was operated at 200 kV. Samples were mounted on Quantifoil carbon polymer supported on a copper grid by placing a few droplets of a suspension of ground sample in ethanol on the grid, followed by drying at ambient conditions.
Samples were analyzed on a CARBOSep COREGEL-87H3 column using a Waters 515 HPLC pump (0.8 mL min−1) equipped with a Shodex RI SE-61 detector, a Perkin–Elmer Series 200 Autosampler (10 μL injection) and a Chrompack HPLC column thermostat SpH 99 (70 °C). H2SO4 (5 mM , pH 1.5) in demineralized water was used as the eluent. This resulted in the following retention times: 9.9 (glycerol), 11.9 (EG), 12.7 (12PD), 13.0 (13PD), 14.2 (MeOH), 16.2 (EtOH), and 19.8 minutes (nPrOH) .4, 20 Calibration curves of these analytes were constructed to analyze the product concentrations. The products detected by HPLC were 12PD, 13PD, EG, nPrOH, and EtOH. Gaseous degradation and condensation products were not detected by HPLC. Conversion (C) and selectivity (Sanalyte) were calculated by Equations (1) and (2), respectively.
in which [glycerol]x and [analyte]x stand for glycerol and analyte concentration at reaction time x in hours. Degradation accounts for the formation of gaseous products and is calculated using Equation (3). In this case it is assumed that
Design of Experiment
Using Design Expert software (version 8.0.1), a five-level, five-factor, half fraction Central Composite Design (CCD) with 8 center points was designed to investigate the influence of the parameters on both glycerol conversion and 13PD selectivity. The selected factors were pressure, silicotungstic acid concentration, glycerol concentration, reaction time, and temperature. The experiments were performed in a randomized order.
Results and Discussion
Hydrogenolysis of glycerol can be considered as a two-step process, in which an acid promotes glycerol dehydration, and the resultant double bonds are hydrogenated over a heterogeneous catalyst (Scheme 1). Selectivity towards 13PD can be enhanced by selecting an appropriate acid and hydrogenation catalyst. The acid plays a decisive role in eliminating a hydroxyl group, and a swift consecutive hydrogenation prevents further dehydration.17, 18 Three types of tungsten additives (tungstic acid, phosphotungstic acid, and silicotungstic acid) were tested in combination with four commercially available hydrogenation catalysts (Pd/SiO2, Pd/Al2O3, Pt/SiO2, and Pt/Al2O3) to investigate their influence on aqueous glycerol conversion and 13PD selectivity. The platinum and palladium catalysts were chosen for their hydrogenation capabilities, whereas tungsten components have previously been shown to result in 13PD selectivity.7, 8, 12, 13, 15, 16 The tungsten heteropolyacids are known for their relative thermal instability. However, this type of catalyst has been successfully applied in similar conditions,21 even on an industrial scale.22
The performance of the additives tungstic acid, phosphotungstic acid and silicotungstic acid was tested in combination with four hydrogenation catalysts. To be able to make a proper comparison, the four catalysts were also tested in the absence of any additional acid, and in the presence of a non-tungsten containing acid, namely hydrochloric acid. In the absence of acid, glycerol was selectively converted into 12PD, and considerable quantities of degradation products were observed as well (Figure 2 a). The activities and selectivities of the four catalysts were comparable if hydrochloric acid pH 1.5 (similar to the pH of the silicotungstic and phosphotungstic acid) was used (Figure 2 b). Small amounts of 13PD were observed if hydrochloric acid was used. However, other degradation products were formed, which were mainly nPrOH. Apparently, dehydration is so favorable in these conditions that either the intermediate 3-hydroxypropanal cannot be hydrogenated in time, or that the initially formed propanediols are not stable under reaction conditions and are subsequently dehydrated and degraded. This suggests a stabilizing effect of the tungsten in the different tungsten containing acids. Possibly, a glycerol tungstic acid ester is formed as an intermediate, which, in combination with tungsten’s favorable acidic properties, leads to 13PD selectivity.23–25
Interestingly, 20 to 40 % 13PD formation is observed if a tungsten containing acidic additive is used (Figure 2 c–e) in combination with a Pt based catalyst. Moreover, only very small amounts of 12PD were observed. If heterogeneous tungstic acid is added (Figure 2 c), the platinum catalysts selectively produce 13PD, whereas the palladium catalysts yield 12PD. If phosphotungstic acid is used as a homogeneous additive, 13PD is formed on the platinum catalysts. Notably, very little to no 13PD or 12PD is formed on the palladium catalysts. Instead, higher amounts of nPrOH and degradation products were observed (Figure 2 d).
Silicotungstic acid, another homogeneous tungsten containing acid, gives similar results as phosphotungstic acid, although higher amounts of 13PD are formed with palladium catalysts (Figure 2 e).
Pt versus Pd
The four tested heterogeneous catalysts contain either palladium or platinum metal on either silica or alumina. The conversion of glycerol on these four catalysts is similar if hydrochloric acid is used (Figure 2 b), whereas in the other examples the Pt catalysts shows both higher conversion of glycerol and higher 13PD selectivity. It appears that platinum is a more active agent to hydrogenate the intermediate 3-hydroxypropanal before further dehydroxylation. This observation is supported by the fact that the 13PD selectivity over the platinum catalysts is higher than that of the palladium catalysts (Figure 2 c–e). The 12PD selectivity observed with Pd and tungstic acid (Figure 2 c) indicates that it is not only the acid that directs the elimination. Although a tungsten additive was used, no 13PD selectivity is observed. This probably owes to the fact that the solid and heterogeneous tungsten acid has fewer possibilities for glycerol coordination relative to the homogeneous silico- and phosphotungstic acid. This, in combination with the lower hydrogenation activity of palladium, results in 12PD formation.
SiO2 versus Al2O3
A comparison of the activity and selectivity between SiO2 and Al2O3 based catalysts shows that the support has limited influence on the selectivity. However, the Al2O3 supported catalysts generally have a higher activity than their SiO2 counterpart. The only exception is Pt/Al2O3 versus Pt/SiO2 if phosphotungstic acid is used as an additive (Figure 2 d). In this case, the activity of Pt/SiO2 exceeds that of Pt/Al2O3, but with poor selectivity.
The influence of the support on the activity might be explained by the basicity of the Al2O3 support. The basic sites on the surface can provide additional dehydration and condensation opportunities, whereas the weak acid sites of the SiO2 surface are irrelevant if comparing the acidity of the additive.
Furthermore, the higher metal dispersion on the alumina support compared with silica (Figure 1), explains the higher conversion relative to the silica supports.
Central Composite Design
The catalytic performance tests of the previous section show that the Pt/Al2O3 catalyst gives highest 13PD selectivity in combination with high catalytic activity. The combination of Pt/Al2O3 with silicotungstic acid as an additive was chosen as the catalytic system for further optimization. Silicotungstic acid is preferable over phosphotungstic acid because of its superior activity, whereas the homogeneous silicotungstic acid is considered to be a better defined system than the heterogeneous tungstic acid (Figure 2 f).
The effects of reaction time (3.5–23.5 h), aqueous glycerol concentration (10–1380 mM), silicotungstic acid concentration (0–12.5 mM), pressure (20–50 bar), and temperature (150–220 °C) on the conversion of glycerol and the selectivity towards 13PD over the Pt/Al2O3 catalyst were investigated through a five-level, five-factor, half fraction Central Composite Design (CCD), consisting of 34 experiments (16 factorial points, 10 star points, and 8 center points). The center point was reproduced eight times to obtain a good estimate for the standard deviation of the system and to warrant reproducibility of the reaction over time.
In this design procedure, the experimental results of the CCD (Supplementary Information, Table S1) were correlated by a second order polynomial [Eq. (4)], capturing both the linear and nonlinear effects of the individual variables, and the effect of the interaction between variables.
Here, y is the response (in this case glycerol conversion or 13PD selectivity), xi is variable i, β0 is the intercept, βni is the coefficient of variable i, βij is the coefficient of the interaction between variable xi and xj„ and ε includes the experimental error and the effects of any uncontrolled variable present. The polynomial is represented in Response Surface Model (RSM) plots (Figure 3), which readily visualize the influence of the variables.
The correlation model for the conversion of glycerol indicates that temperature and reaction time have the most significant effect (Figure 3 a and b). Other variables such as pressure and silicotungstic acid concentration have an effect on glycerol conversion as well, but to a lesser extent (Figure 3 a, c, d). Notably, the glycerol concentration does not influence the glycerol conversion. To clarify, Pt/Al2O3 is most active at high glycerol concentration, as the concentration of the platinum in the samples remains constant at 25 μmol.
The model for the 13PD selectivity shows that temperature and silicotungstic acid concentration are the most important variables influencing the selectivity (Figure 3 e and f).
The RSM plots analysis reveals that highest glycerol conversions are obtained at long reaction time and high temperature, whereas the highest 13PD selectivity is obtained at lower temperatures and low silicotungstic acid concentration. Resultantly, the activity and selectivity cannot be optimized simultaneously. This result can be rationalized by considering the reaction mechanism of 13PD formation (Scheme 1). The high temperatures positively influence the initial endothermal dehydration of glycerol, thereby increasing the glycerol conversion. However, it has an opposite effect on the subsequent hydrogenation of the formed double bond because this is an exothermal process. The high temperatures also promote degradation of the formed products.
Silicotungstic acid concentration
Although it is established that silicotungstic acid has a positive effect on 13PD selectivity, high silicotungstic acid concentrations influence the 13PD selectivity negatively. Apparently, a minimal silicotungstic acid concentration is required to achieve 13PD selectivity at all, whereas too much acid will promote further dehydration, thereby decreasing the 13PD selectivity.
Indeed, no 13PD is formed in reactions that use 0.7 mM silicotungstic acid and only 12PD and degradation products are observed (Figure 4). Gradually increasing the acid concentration leads to an optimal 13PD selectivity around 2.1 to 2.8 mM, and suppresses the 12PD selectivity, thereby inverting the product selectivity. A further increase of the acid concentration results in lower propanediol selectivities in favor of degradation. Therefore, it can be concluded that 2.1 mM is the optimal silicotungstic acid concentration, with a glycerol conversion of 49 % and a 13PD selectivity of 28 %.
The differences in selectivity, as described above, can be explained if different mechanisms are considered, at different acid concentrations. The first is the reforming of glycerol, which results in the formation of degradation products and is catalyzed by platinum at low acid concentrations. This reaction pathway is inhibited by acid.26 If more acid is present, a dehydration—hydrogenation mechanism is at play, resulting in the formation of propanediols. Depending on the amount of acid present, either 12PD is formed (low acid concentration) or 13PD is formed (higher acid concentration). If too much acid is present, the propanediols are subsequently degraded, and propanol and degradation products are obtained.
High glycerol concentrations
The glycerol concentration does not have a significant influence on glycerol conversion. A concentration of 25 μmol platinum was used in all the experiments and, thus, the Pt/Al2O3 apparently converted more glycerol, so displays a higher activity at higher glycerol concentrations.
This effect was investigated in more detail by increasing glycerol concentrations from 8.3 wt % (1000 mM) up to 100 wt % glycerol (Figure 5), while maintaining the platinum loading at 25 μmol. If the relative conversions and selectivities are compared (Figure 5 a), it is evident that the conversion initially decreases, although at high glycerol concentrations, the conversion shows an upward trend. However, the selectivities towards 13PD, 12PD, and nPrOH show a downward trend in favor of degradation products. This is most apparent from the absolute conversion and production values (Figure 5 b). The amount of glycerol converted per mole of platinum and acid is increased tremendously, although glycerol is mostly converted into degradation products. Attention must be brought to the fact that three processes take place over the platinum catalyst: glycerol hydrogenolysis, glycerol degradation to gaseous products, and glycerol condensation (aldol condensation). The calculated total reduction product (TRP, 13PD+12PD+2×nPrOH) plateaus at approximately 10 mmol. This can be considered the maximum hydrogenation capacity of Pt/Al2O3 under these conditions. Any surplus of glycerol will therefore automatically be converted into gaseous and condensed product, which explains the high increase of conversion at high glycerol concentrations. The acid promotes this conversion, which is clearly not a limiting factor.
To obtain a better understanding of the degradation process of glycerol, the stability of its primary and secondary metabolites was studied using HPLC analysis. This was performed by exposing glycerol and its degradation products to the same reaction conditions, and following both conversion and the selectivities of all reactions. This results in the network in Scheme 3, in which both the conversion of the starting materials and the selectivity are shown. Glycerol is selectively converted to 13PD, rather than 12PD. This is attributed to the higher stability of 13PD (28 % conversion) relative to 12PD (88 % conversion) under these reaction conditions. This provides, in addition to the selectivity enhancing properties of silicotungstic acid, a supporting explanation for the high 13PD selectivities and high 13PD/12PD ratios obtained using the Pt/Al2O3 and silicotungstic acid catalytic system.
The consequences of the relative stabilities of the glycerol metabolites for the product formation can be visualized in a time dependent experiment. Figure 6 shows that over the course of 72 h glycerol is steadily converted. However, the amount of 13PD, 12PD, and nPrOH stabilizes after approximately 18 h reaction time, and the amount of degradation products builds up continuously during the 72 h. Clearly, part of the 13PD, 12PD, and nPrOH produced subsequently degrades. The rates of production and degradation of 13PD, 12PD, and nPrOH are in balance, thereby stabilizing their respective yields in the period from 18 to 72 h.
Two catalyst recycling studies were performed, using different glycerol concentrations (Figure 7). The activity of the catalyst remains the same if 1000 mM glycerol is used. However, the 13PD selectivity decreases in the following recycling run. Both activity and selectivity reduces in a subsequent run at 100 mM glycerol.
The decreased 13PD selectivity after recycling is attributed to the sintering of the Pt particles, visible in the TEM pictures in Figure 8 c, d. Increased Pt particle sizes result in slower hydrogenation, which increases the risk of dehydroxylation of the intermediate 3-hydroxypropanal into acrolein. This ultimately results in increased amounts of nPrOH and degradation products.
The observation that degradation only increases if 1000 mM glycerol is used and the activity of the catalyst remained the same, could be explained by the higher probability of condensation products at elevated glycerol concentrations because acid catalyzed condensation does not require small Pt particles, i.e., the competing reaction is more likely to occur.
Pt/Al2O3 in combination with silicotungstic acid additive is an excellent system to selectively convert glycerol into 13PD. Both the use of tungsten containing additives and platinum are critical for high 13PD selectivity. A more detailed study of the reaction parameters shows that silicotungstic acid concentration and temperature have the strongest influence on 13PD selectivity. A minimal amount of silicotungstic acid is required to obtain 13PD selectivity, and high concentrations result in degradation. Temperature is the parameter that positively influences glycerol conversion, which makes optimization of both glycerol conversion and 13PD selectivity difficult.
These results regarding both glycerol conversion and 13PD selectivity can be explained if hydrogenolysis is considered as an acid catalyzed dehydration followed by a hydrogenation over a metallic surface (Scheme 1).17
The initial acid catalyzed dehydration is the selectivity controlling step. The formation of a relatively stable secondary carbocation intermediate results in the formation of 13PD. The reaction is therefore kinetically controlled. Following this mechanism, it is the acid that controls the final selectivity. Indeed, there is a clear distinction in selectivity using hydrochloric acid or a tungsten containing acid (Figure 2). Apparently, the tungsten containing acids possess favorable 13PD selectivity inducing properties. This may be through coordination of tungsten to glycerol, as described for ReOx and MoOx, which have a similar 13PD favoring coordination as that envisaged for tungsten compounds.17, 18, 27, 28
However, the subsequent hydrogenation plays a major role for the yield of 13PD as well. A swift hydrogenation of 3-hydroxypropanal to 13PD prevents the further dehydration of 3-hydroxypropanal. This explains the higher 13PD selectivities over the more active platinum catalysts relative to the palladium catalysts.
Owing to the contradictory temperature requirement for the endothermic dehydration versus the exothermic hydrogenation and increased elimination products at higher temperatures it is difficult to obtain both high glycerol conversion and 13PD selectivity. Moreover, 13PD is relatively stable relative to 12PD, resulting in excellent 13PD/12PD ratios. Nevertheless, it is still prone to degradation. Therefore, it would be beneficial to merge the tungsten and hydrogenation metal in a bifunctional catalyst so both functionalities are in close proximity.
The catalytic system Pt/Al2O3 with silicotungstic acid additive converted glycerol selectively into 13PD (49 % conversion and 28 % 13PD selectivity) with an excellent 13PD/12PD ratio, making this commercially available catalytic system one of the best performing reactions in batch operated processes to date. This was obtained using water instead of the environmentally and technically unattractive 1,3-dimethyl-2-imidazolidinone. The use of both tungsten containing additives and platinum was critical for high 13PD selectivity. 13PD is more stable than 12PD under reaction conditions with this combination, explaining the excellent 13PD/12PD ratio.
The effect of hydrogenation metal, acid concentration, temperature and reaction time on glycerol hydrogenolysis and 13PD selectivity can be rationalized if glycerol hydrogenolysis is regarded as a two-step dehydration-hydrogenation reaction. The optimal reaction conditions for dehydration and hydrogenation are not compatible, which makes it extremely difficult to obtain both a high conversion of glycerol and high 13PD selectivity. However, this could be improved if the acidic sites and hydrogenation metal are in close proximity as in a bifunctional catalyst.
J.t.D. gratefully acknowledges financial support from NWO ASPECT (053.62.020).