Glycerol carbonylation with urea requires both acid and base sites; hence, the primary aim of this work was to understand the role of support and Zn loading in this reaction. For this purpose, a systematic characterization of the catalysts was carried out as discussed below. Table 2 shows the surface areas of different supports for a constant loading of Zn (5%). As seen from Table 2, Zn supported on MCM-41 catalyst showed the maximum surface area of 692 m2 g−1 which varied with the supports in the descending order of MCM-41 > SBA-15 > SiO2 > γ-Al2O3 > ZrO2 > s-ZrO2 in accordance with the nature of the support material (discussed later). MCM-41 having the maximum surface area was used for further study on Zn loading variation in the range of 2–10%. Figure 2 shows Zn loading vs. surface area along with GC yield profiles. As expected, surface area decreased from 692 to 498 m2 g−1 with increase in Zn loading from 2 to 10%. Although the trend is not linear, it indicates the dispersion of Zn within the mesoporous structure of silicalite MCM-41 and simultaneously affecting the uniform ordering of the mesoporous structure as can be seen from the XRD analysis. The increased dispersion of Zn with increase in loading has directly reflected in the enhancement of GC yield from 57 to 83%. This suggests that the catalytic activity cannot be simply correlated to the surface area of the catalysts. Further interpretation of the activity results was deduced from various surface acid–base properties and structural ordering of the support material.
Figure 3A displays small-angle XRD patterns of MCM-41 and SBA-15 supported Zn catalysts which showed good ordering of 2D-hexagonal structure indexed to (100), (110) and (200) reflections [31, 32]. This suggests the retention of hexagonal mesoporous ordering of the MCM-41 and SBA-15 even after Zn loading. Wide angle XRD patterns (Fig. 3B) of 5% Zn on MCM-41, SBA-15 and SiO2 show the amorphous nature of the samples while a single broad peak at 2θ = 23.1o (411) could be attributed to orthorhombic silica. There was no any reflection corresponding to ZnO phase, suggesting the well dispersed nature of ZnO particles on the support and/or inside the framework of mesoporous supports like MCM-41 and SBA-15 which is in accordance with the previous reports [33, 34]. In case of alumina support (Fig. 3B), peaks observed at 2θ = 37.2o (110), 42.0o (006), 45.3o (202) and 67.1o (214) were due to Al2O3 (JCPDF No. 81-1667) with a small peak at 2θ = 31.7o (100) of ZnO phase. The crystalline nature of 5% Zn/ZrO2 and s-ZrO2 samples was evident by the sharp peaks (Fig. 3B) at 2θ = 24.4o (110), 28.2o (ī11), 41.1o (102), 49.3o (022), 50.6o (ī22) and 59.9o (131) which were attributed to ZrO2 (JCPDF No. 83-0944). The sharp peaks observed at 2θ = 31.7o (100), 34.4o (002), 36.2o (101) and 66.4o (200) could be assigned to ZnO phase (JCPDF No. 80-0075). This confirmed that the formation of crystalline and/or large size ZnO particles was possible on ZrO2 and s-ZrO2 supports compared to MCM-41, SBA-15, SiO2 and Al2O3 supports. Nature of the support obviously influenced the surface areas of the respective catalysts, as discussed above.
As 5% Zn/MCM-41 showed the highest activity, effect of Zn loading on MCM-41 on structure and activity was also investigated. The low and wide angle XRD patterns of the parent MCM-41 and samples with Zn loading varying in the range of 2-10% are shown in Fig. 4A, B, respectively. Typical small-angle XRD of MCM-41 showed the reflections of (100), (110) and (200) planes corresponding to the ordered hexagonal mesoporous structure. In all the Zn-loaded samples, the presence of main (100) peak confirmed the unchanged hexagonal frame structure. However, loading of Zn varied from 2 to 10%, caused shifting of the main peak (100) to higher 2θ values with a peak broadening. The other two peaks (110) and (200) with low intensity confirmed the presence of ZnO within the pores of MCM-41 . In case of highest Zn loading of 10%, the reflection of (100) plane weakened considerably while reflections of peaks due to (110) and (200) planes almost completely vanished. Thus, higher Zn loading dramatically influenced the structural ordering of MCM-41 . The wide angle XRD patterns for various Zn loadings (Fig. 4B) did not show any peak corresponding to ZnO confirming excellent dispersion of ZnO inside the pores of MCM-41. The presence of Zn (ZnO) inside the framework of the MCM-41 was also confirmed by TEM analysis coupled with EDX (Fig. 5A, B).
Activation of glycerol and urea requires both acid and base sites, respectively, to form glycerol carbonate. Hence, the catalysts prepared in this work were characterized for their acid and base strength by NH3 and CO2-TPD, respectively. In NH3-TPD plots of Zn catalysts with different supports (Fig. 6), the peaks below 200 °C correspond to physisorbed NH3 whereas the peaks in the temperature range of 200–400 °C represent moderate acidity while, the peaks in the range of 400–700 °C represent the strong acidity. Only in case of ZrO2 and s-ZrO2 supported Zn catalysts, a sharp peak observed in the range of 650–700 °C clearly suggests strong acidic strength of both the catalysts . Among all the prepared catalysts, SiO2 supported catalyst showed a broad peak below 200 °C with the lowest strength of total acidity of 0.1547 mmol g−1 in terms of NH3 desorbed. In case of Al2O3 supported catalyst, a broad distribution of acidity all over the temperature range of 100–700 °C was observed which clearly suggests the presence of weak, moderate as well as strong acidic sites . MCM-41 supported Zn catalyst showed only a single broad peak below 200 °C indicating the presence of weak acid sites.
Table 3 shows comparative data of total acidity, basicity and acid/base ratio of the various loadings of Zn on MCM-41. Acidity of the catalysts initially increased with increasing Zn content up to 5% beyond which it remained almost constant at about 0.52 mmol of NH3 g−1. The increased acidity after Zn incorporation into the silica framework was due to the formation of Zn2+ Lewis acid sites in the silica framework as well as the BrØnsted acidity due to surface silanol groups. Similar observation was also described by Silvestre-Albero et al. for Zn supported MCM-41 catalyst . Similar to the acidity trend, basicity generated due to the oxygen anion of the ZnO was less in case of 2% Zn loading which then increased for 5% loading and afterwards, it remained constant. The acidity/basicity ratio was found to be constant (~0.8) for the range of 5–10% Zn loading on MCM-41. It was observed that glycerol conversion increased substantially with increase in Zn loading irrespective of the acid/base ratio.
Carbonylation of glycerol with urea in a batch operation
Various supports were screened for constant loading of 5% Zn for the carbonylation of glycerol with urea and the results are shown in Table 4. Zn/MCM-41 showed highest conversion of 78% with maximum GC selectivity of 98% while, Zn/SiO2 showed the lowest conversion of 20% but 97% selectivity to GC. In case of Al2O3, ZrO2 and s-ZrO2 supported Zn catalysts, glycerol conversion was about 60% but the GC selectivity decreased substantially to 67%. Since, both Al2O3 and ZrO2 were having strong acidic sites (Fig. 6), formation of other side products such as urethanes was possible, reducing the GC selectivity . It was clear from these results that the support such as MCM-41 having highest surface area was responsible for the maximum dispersion of active ZnO phase although having weak acidity, led to the highest activity. However, the selectivity to GC was strongly dependant on the appropriate acidity of the catalyst. Among the screened catalysts, as the Zn/MCM-41 showed the maximum glycerol conversion and GC selectivity further studies of process optimization and catalysts stability were performed using Zn/MCM-41 catalyst and these results are discussed below.
Effect of metal loading
Effect of active metal loadings on MCM-41 support in the range of 2–10% on conversion and selectivity of GC is shown in Table 5. Surprisingly, glycerol carbonylation with urea was also observed in absence of catalyst as well as with bare MCM-41 (entries 1 and 2, Table 5). The role of Zn as a catalyst was evident from the enhanced glycerol conversion and GC selectivity (entries 3–5, Table 5). The conversion of glycerol increased linearly from 63 to 82% which might be because of increase in both Lewis acid and base strength (Table 3) with increase in active metal loading from 2 to 5% while, further increase in metal loading to 10% did not affect the conversion significantly. GC selectivity observed was 92% for minimum Zn loading of 2% while, almost complete (~98%) selectivity to GC was achieved for Zn loading of 5 and 10% (entries 4–6, Table 5).
Effect of catalyst loading
The effect of loading of 10% Zn/MCM-41 catalyst was also studied in the range of 0.06–0.23 g, on activity and GC selectivity. As seen from Fig. 7, the conversion of glycerol increased from 70 to 85% while, selectivity to GC reached almost completion (from 97 to 98%), with about a fourfold increase in catalyst loading (0.06–0.23 g). This is obvious as increase in catalyst loading caused an increase in the acidic and basic sites that facilitated the carbonylation of glycerol with urea to give GC.
Effect of mol ratio
The effect of the mol ratio of glycerol to urea in the range of 1:0.5–1:2 on conversion and selectivity to GC was studied by varying the initial concentration of urea at a constant glycerol and catalyst concentration for 10% Zn/MCM-41 catalyst and the results are presented in Table 5. The conversion of glycerol was much lower (44%) for the highest mol ratio of glycerol to urea (entry 1, Table 6) obviously as urea was the limiting reactant. At higher urea concentration, glycerol conversion increased up to 92% (entries 2 and 3, Table 6); however, GC selectivity dropped down considerably from 98 to 58% (entry 3, Table 6). This is because at higher urea concentration, further carbonylation of GC with urea is possible forming urethanes . At 1:1 mol ratio, highest selectivity of 98% to GC was achieved with 84% conversion, suggesting that 1:1 was an optimum ratio of glycerol to urea to give the highest activity and selectivity to GC.
Effect of the temperature
With the glycerol to urea ratio of 1:1, study of effect of temperature was done by varying the temperature in a range of 120–160 °C. As shown in Fig. 8, glycerol conversion linearly increased from 56 to 95% with increase in temperature from 120 to 160 °C. However, at higher reaction temperature of 160 °C, slight decrease in GC selectivity was observed with simultaneous increase in byproduct selectivity. Since, the maximum selectivity of 98% to GC was achieved at 140 °C, it was considered as the optimum reaction temperature.
Effect of the N2 flow rate
The effect of N2 flow rate was also studied by its variation as 10, 20 and 40 mL/h at optimum reaction temperature of 140 °C (Table 7). It was observed that both glycerol conversion and GC selectivity increased from 60 to 84% and from 86 to 98%, respectively, with increase in N2 flow from 10 to 20 mL/h. However, further rise in N2 flow of 40 mL/h showed only marginal increase in glycerol conversion of 87% while, GC selectivity remaining constant at 98%. Therefore, the optimum N2 flow was considered to be 20 mL/h.
From the process point of view, it was thought appropriate to study the glycerol-urea carbonylation in a continuous mode of operation. From the activity study in a batch operation, it was observed that 10% Zn loading showed only marginal increase in conversion with the same selectivity to GC as that of 5% Zn/MCM-41catalyst (Table 5). Hence, further study was done with 5% Zn/MCM-41 catalyst in a continuous mode of operation.
Initially, time on stream activity of 5% Zn/MCM-41 catalyst was evaluated for continuous glycerol carbonylation in presence of 2-propanol (IPA) as a solvent. Figure 9 shows a consistent performance for 24 h of reaction time at 140 °C temperature and 20 mL h−1 N2 flow with an average glycerol conversion of ~25% and GC selectivity in the range of 70–75%. However, unlike the product distribution observed in the batch operation, selectivity to GC was restricted to only 75% with ~10% selective formation of GD and ~15% selectivity to other byproducts (Fig. 9). The possible byproducts included isopropyl carbonate and di-isopropyl carbonate obtained by transesterification of IPA with urea.
Figure 10 displays the time on stream (TOS) activity of the 5% Zn/MCM-41 catalyst in an aqueous medium for the carbonylation of glycerol to glycerol carbonate at 140 °C temperature, and 20 mL h−1 N2 flow. Interestingly, in contrast to IPA run, here the selectivity to GC enhanced to ~91% with ~9% formation of GD and other byproducts. The run started with the higher glycerol conversion of 35% which then stabilized to 25%. As the change from organic solvent (IPA) to aqueous, drastically enhanced GC selectivity from 75 to 91% further studies of process optimization were done in an aqueous medium.
Optimization of operating conditions
Glycerol conversion and GC selectivity are greatly influenced by process parameters such as residence time, reaction temperature and glycerol concentration; hence, the effect of these parameters on the catalyst performance was carried out systematically and the results of which are discussed below.
The effect of liquid flow rate in terms of contact time on glycerol conversion is displayed in Fig. 11. It was observed that as the contact time increased from 0.2 to 0.3 h, a sharp rise in the glycerol conversion from 20 to 45% was observed with almost complete selectivity to GC. Further increase in contact time from 0.3 to 0.6 h caused linear increase in activity in terms of conversion from 45 to 55% with complete selectivity to GC. This follows that increase in residence time directly enhanced the number of possible collisions of the reactant molecules on the surface of the catalyst for lower liquid feed rate. Thus, the highest productivity of GC was achieved at 10 mL h−1 flow rate of aqueous glycerol. The effect of liquid flow rate on the global rate of glycerol carbonylation was also studied (Fig. 12) keeping all the other parameters constant. The linear decline in global rate from 0.7 × 10−8 to 0.09 × 10−8 k mol m−3 s−1 was observed with increase in liquid flow rate from 5 mL to 15 mL h−1. This may be due to the fact that in the lower range of liquid flow rate of 5–7 mL h−1, catalyst particles get partially wetted due to capillary action which is responsible for the direct interaction of liquid phase and solid surface of the catalyst, ultimately resulting in higher rate of carbonylation. In contrast to this, at higher liquid flow rates of 7–15 mL h−1, channelizing would hinder the access of liquid reactants to the catalyst sites resulting in the linear decrease in the global rate of carbonylation from 0.4 × 10−8 to 0.09 × 10−8 k mol m−3 s−1.
Figure 13 exhibits the effect of temperature on the glycerol conversion and GC selectivity in the carbonylation of glycerol with urea. With increase in temperature from 140 to 200 °C, a gradual increase in the glycerol conversion was observed along with simultaneous increase in GC selectivity from 91 to 100%. The rise in temperature from 140 to 170 °C resulted in marginal increase in glycerol conversion from 25 to 30%. However, further increase in temperature up to 200 °C, resulted in considerable increase in glycerol conversion from 30 to 45% with complete selectivity to GC without formation of any byproduct. Therefore, the optimum reaction temperature for the glycerol carbonylation was found to be 200 °C. Activation energy estimated from the Arrhenius plot was found to be 39.82 kJ mol−1, suggesting glycerol carbonylation reaction was kinetically controlled, under the present set of reaction conditions (Fig. 14).
Effect of glycerol loading on the catalytic activity for the glycerol carbonylation is shown in Fig. 15. Increase in glycerol loading from 26 to 36% caused twofold decrease in glycerol conversion was observed, without affecting the GC selectivity of 100%. This indicates that increased glycerol loading resulted in increased viscosity of the solution due to which substrate get strongly adsorbed on the active sites of the catalyst leading to the deactivation of catalytic activity.
Figure 16 shows the time on stream activity of 5% Zn/MCM-41 catalyst of 100 h at 200 °C, 10 mL h−1 feed flow and 20 mL h−1 N2 flow. Glycerol conversion was 50% within first 2 h after which it decreased slightly to 45% and then remained constant up to ~97 h of reaction time whereas, the GC selectivity was not affected (100%). As we compared the GC productivity in batch and continuous operation, the GC productivity in batch was five times (6.9 h−1) higher than that of continuous operation (1.38 h−1). This was due to higher residence time in a batch operation.
To confirm the glycerol conversion determined by gas chromatography of liquid samples, the quantification of NH3 released during the reaction was also carried out for few experiments. For this purpose, NH3 evolved from the reaction was dissolved in water at the reactor outlet using N2 as a carrier gas. The total amount of ammonia released was analyzed by volumetric analysis of both the reactor outlet gas sample and liquid reaction crude against strong 0.5 M HCl. The total amount of NH3 released was found to be 2.71 g which was in good agreement with the theoretically obtained value for 45% of glycerol conversion.
A possible reaction mechanism for carbonylation of glycerol with urea over Zn/MCM-41 catalyst
It is important to note that the present reaction of glycerol with urea to form GC proceeds without a catalyst under standard reaction conditions but with only 40% conversion and 90% selectivity to glycerol carbonate (entry 1, Table 5). In presence of Zn-based catalyst, glycerol conversion enhanced by twofold (87%) with almost complete selectivity towards GC. It was also observed that the parent MCM-41 showed 51% conversion of glycerol, as the OH groups of surface silanol can activate the urea molecule as shown in Scheme 2 (step I) . In step II, Lewis basic sites of ZnO can activate glycerol molecule, which is then prone to the attack on carbonyl carbon of urea with simultaneous release of 1 mol of NH3. In the next step (III) secondary –OH of activated glycerol again attacks over carbonyl carbon of urea to form five membered cyclic glycerol carbonate with the loss of second molecule of NH3. In the last step IV of the reaction pathway, the catalyst surface was regenerated by the release of GC. This suggests that the ZnO and MCM-41 play cooperative role in glycerol carbonylation reaction. This was confirmed by control experiment with only bare ZnO which gave lower glycerol conversion of 68% compared to that of 5% Zn/MCM-41 (78%) with GC selectivity of 98% (entry 3, Table 5).