Solventless synthesis of solketal with commercially available sulfonic acid based ion exchange resins and their catalytic performance

2.1 Materials

2.2 Catalyst characterization

Brunauer-Emmett-Teller (BET) surface area of the resins was determined using the N2 adsorption-desorption technique at 196 K performed by a SA3100 analyzer (Beckman Coulter Ltd., High Wycombe, UK). Prior to the measurements, samples were degassed at 373 K during 120 min.

Porosimetry by mercury extrusion was completed for an approximate sample mass of 0.25 g of each resin in an Electron Pascal 440 series device (Thermo Scientific Ltd., Loughborough, UK) to quantify for mesopores and macropores. The calculation of the pore diameter was made using the Washburn cylindrical pore model, where the surface tension of mercury and the mercury contact angle were taken as 0.484 N/cm and 141°, respectively.

Thermogravimetric analyses were made in a 851e TGA/SDTA instrument (Mettler Toledo, Leicester, UK). Temperature was increased from 333 K to 833 K at a heating rate of 10 K/min under a nitrogen flow of 20 ml/min so as to keep an inert atmosphere. Each sample weighed approximately 10 mg. The curves obtained for the thermal decomposition of each catalyst were the average of three experiments.

Solid state ¹³C-NMR spectroscopy of the catalysts was performed at 400 scans in a DSX300MHz BACS60 device (Bruker Ltd., Coventry, UK) with an automatic sampler.

Elemental microanalysis was performed in a CHNS-932 equipment (LECO, Saint Joseph, MI, USA). Prior to the analysis of the resins, samples of 3 mg were dried overnight at 373 K to avoid errors due to the residual moisture.

For the potentiometric titration of the catalysts, a suspension of 25 mg of the resin in 40 ml of acetonitrile was prepared and kept under stirring for 24 h prior to the addition of a solution of N-butylamine in acetonitrile (0.005 N) at a constant flow rate of 0.4 ml/min. Potential was measured with an Eutech Instruments pH 700 pH-meter.

Additionally, Bronsted acidity of the resins was suspending overnight 10 mg of the resins in a solution of 7.5 g/l of KCl in water. The acid capacity was determined from measuring the pH value.

Table 1 compiles the physical and chemical properties of the resins determined from the aforementioned methods or as specified by the suppliers.

2.3 Apparatus and methodology

The experimental setup consisted of a glass reactor heated by means of a thermal resistor placed surrounding its outer walls with temperature being controlled by an OMRON E5CN PID controller linked to the resistor. Stirring of the system was regulated by a flat six-blade impeller on an IKA RW20 motor (250–2500 rpm) (IKA, Oxford, UK). Sample withdrawal was performed with a syringe with a wide-bore needle piercing a Teflon lid tightly fitted to the upper part of the reactor.

The operational procedure followed in all the experiments started by loading the reactants glycerol and acetone into the glass reactor. The catalyst was added when the set temperature was reached and samples were withdrawn thenceforth. The standard reaction conditions chosen for the study were a temperature of 313 K, a molar excess of acetone of 4.5 and a catalyst load of 0.5% of weight of catalyst per unit weight of the reactants (indicated as w/w onwards). In addition, the mode of operation of this reaction implies no removal of water as a reaction by-product to shift the equilibrium towards the products, considering the fact that it is known to be a major source of sulfonic acid-based ion exchange resins deactivation [39].

For the catalyst recycling experiments, up to four repetitions of the assay under the same conditions were completed. From experiment to experiment, the catalyst was filtered out of the reaction mixture, washed with methanol and dried at 373 K overnight. For catalyst regeneration, treatment with 10 ml of a solution containing hyperchloric acid (1 N) per gram of resin was completed.

2.4 Analysis of samples

Analysis of samples was performed by ¹H NMR spectroscopy with a BRUKER DPX 300MHz BACS60 (Bruker Ltd., Coventry, UK) device using methanol as an internal standard for quantification purposes and deuterium oxide as solvent for the samples.

Solketal was the chemical species monitored, the signal of which was quantified and related to the concentration present in the reaction sample using methanol as an internal standard. Figure 2 shows a representative spectrum of a reaction sample, with the peaks corresponding to the protons of every compound present in the sample properly identified.

Figure 2:

Representative ¹H-NMR spectrum 300 MHz spectra of a sample of the reaction with all the chemical components and shifts for the corresponding protons identified.

3.1 Characterization of ion exchange resins

Figure 3 plots the mercury intrusion curves (A) and its differentiated curve (B) performed for the determination of the pore size distribution. It can be observed that CT275DR and CT276 behave very similarly in the intrusion curve, with a maximum slope of the decrease of mercury volume intruded at a pore diameter between 65 nm and 80 nm, with a very narrow distribution and both can be said to be macroporous resins. A35dry and A36dry, by contrast, show their drop at much lower pore diameters, reaching the maxima at approximately 35 nm and 15 nm, respectively, although with much wider pore size distributions than those observed for both Purolite resins. In the particular case of A35dry, there are an appreciable amount of pores above 50 nm, which means that this resin can be regarded as mesoporous and macroporous; on the contrary, the pore size distribution size of A36dry barely stretches beyond 43 nm, so it can be said that it is mainly a mesoporous resin. Finally, GF101 shows a somewhat different trend, with a bimodal pore size distribution with a relative maximum at 12 nm and an absolute maximum at 190 nm. Nevertheless, the preponderant fraction of pores ranges from slightly below 100 nm to 1000 nm; therefore, it can mostly be considered macroporous.

Figure 3:

Mercury intrusion curves for the ion exchange resins (A) and pore size distribution derived from the intrusion curves (B).

Thermograms of the ion exchange resins are plotted in Figure 4. Despite the similar behavior of the presented curves for all of the resins, the weight drop intervals vary from resin to resin. At temperatures up to 350 K, there is a typical slight weight loss due to the removal of moisture within the matrix of the samples. However, in this case, due to the dryness of these particular resins, these curves practically do not show this removal (not higher than 1–2% regardless of the resin considered). Neglecting this residual moisture removal interval, the curves exhibit a two-stage decay behavior with a middle step in which the weight remains almost constant. The first stage corresponds to a weight loss of 20–25%, ascribable to the degradation of sulfonic acids, the resin GF101 being the one with a lower quantity of sulfonic acids and similar groups containing sulfur. Maximum degradation rate in this temperature interval is attained at 355 K in all cases. The second stage weight drop can be attributed to the degradation of resin polymeric matrices, as described in several references [37], [38]. In resin GF101, as depicted in Figure 4, there are two weight loss zones within the 500–800 K temperature interval in GF101, indicating two different polymeric matrices in this resin, one of them slightly less stable than the other. In Figure 4, maximal degradation rates of the polymeric supports are observed at 555 K for Purolite resins, at 266 K for Amberlyst resins, and at 266 K and 677 K for the Lewatit GF101. Judging from these temperatures and the maximal degradation rates of the polymers, whose order is A36dry≈A35dry>CT276>CT275DR>>GF101, the thermal stability of the polymer in GF101 appears to be somewhat higher than the rest. At temperatures higher than 750 K, a low and constant degradation rate, similar for all resins, is observed.

Figure 4:

Thermal decomposition curves of the resins from 333 K to 823 K.

Solid ¹³C-NMR was performed to acquire a deeper knowledge on the presence of sulfonic groups within the structure of the resins. Figure 5 compiles the spectra obtained for the solid samples, where a series of representative signals can be seen. First, secondary and tertiary aliphatic carbons can be identified at a shift of about 42.5 ppm, these being carbons accountable for the constitution of the backbone of the divinylbenzene structure [40]. Also, a signal, the maximum of which is at 127 ppm, is observed, which corresponds to the overlapping of the signals of aromatic carbon atoms at ortho- and meta-positions with respect to the carbon to which the sulfonic group is attached, which ranges between 123.0 ppm and 129.4 ppm as calculated in Table 2, considering the different carbons to which the sulfonic group can be attached [40], [41]. Finally, the peak at 140.7–145.5 ppm can be ascribed to directly alkylated or sulfonated carbon atoms, as estimated for the atoms a₁, b₁, a₂, c₂ and a₃ in Table 2; nevertheless, in the cases of CT275DR and GF101, a shoulder at 148 ppm appears. Detections at this value correspond to directly alkylated aromatic carbons at a para-position with respect to the sulfonated carbon atom, i.e. carbon atoms labeled as d₃ in Table 2. These signals at shifts very similar to the one reported herein corresponding to such type of carbon atoms in sulfonated solids have previously been reported [42], [43]. The greater the presence of these atoms, which appears to be more predominant in the mentioned resins, the more available the sulfonic group is owing to lower steric effects.

Figure 5:

¹³C-NMR spectra of the sulfonic resins with identifying guidelines (detailed in the text and Table 2).

Table 2:

Estimated ¹³C-NMR shifts for the carbon atoms of vinylbenzene sulfonated at ortho-, meta- and para- positions [40], [41].

Sulfonated vinylbenzene	Carbon atom	zia(alkyl chain)	δ (ppm)	zia(SO3H)	δ (ppm)	δtotal (ppm)a
	a1 b1 c1 d1 e1 f1	z2 z1 z2 z3 z4 z3	–0.6 15.7 –0.6 –0.1 –2.8 –0.1	z1 z2 z3 z4 z3 z2	15.0 –2.2 1.3 3.8 1.3 –2.2	142.9 142.0 129.2 132.2 127.0 126.3
	a2 b2 c2 d2 e2 f2	z3 z2 z1 z2 z3 z4	–0.1 –0.6 15.7 –0.6 1.3 –2.8	z1 z2 z3 z4 z3 z2	15.0 –2.2 1.3 3.8 –0.1 –2.2	143.4 125.7 145.5 131.7 129.7 123.5
	a3 b3 c3 d3	z4 z3 z2 z1	–2.8 –0.1 –0.6 15.7	z1 z2 z3 z4	15.0 –2.2 1.3 3.8	140.7 126.2 129.4 148

^aTotal shift was calculated according to δtotal=128.5+∑izi, where 128.5 is the shift corresponding to any carbon atom in the benzene ring and z_i is the shift due to the presence of a substituent in the position _i.

The elemental analysis of solids can also shed light on the presence of sulfonic groups. Figure 6 shows the elemental microanalysis performed for the resin samples, the oxygen weight fraction of which has been determined as the difference of the summation of the weight fraction of carbon, hydrogen, nitrogen and sulfur to 100%. The presence of sulfur ranges between 7% and 15%, CT275DR being the solid containing most sulfur, closely followed by GF101; A35dry and A36dry have similar sulfur content of 11.5% and 10.5%, respectively; finally, CT276 possesses the lowest content.

Figure 6:

Elemental microanalysis of the ion exchange resins.

Finally, Figure 7 depicts the potentiometric titration of the five resins. According to Pizzio’s classification, all of the resins have very strong sites given that the initial potential values are above 100 mV [44] and the order of strength would be as follows: CT275DR>A35dry>GF101>A36dry>CT276. However, it is worthwhile mentioning the fact that the curves exhibit different behaviors. CT276, CT275DR and A35dry show a relatively slight reduction of the potential at the beginning of the titration and then a steep decrease; A36dry presents a similar curve, although the drop is not as pronounced as in the previous three cases. By contrast, GF101 shows a marked decline of the potential practically from the beginning. This can lead one to think that the acidity of the latter catalyst becomes available more quickly than the others and, thus, can be associated to the highest strength and catalytic activity.

Figure 7:

Potentiometric titration of the sulfonic acid-based ion exchange resins.

3.2 Catalytic experiments

The ionic exchange resins were tested for activity in the synthesis of solketal under solventless operation at a temperature of 313 K and a molar excess of acetone to glycerol of 4.5 to 1 with a catalyst load of 0.5% of weight of each resin per unit weight of the reactants. It is worthwhile remarking on the fact that at this temperature and these composition conditions, the reacting system is immiscible to a high extent. Moreover, the reaction between acetone and glycerol does not take place in the absence of catalysts, as indicated in the liquid-liquid equilibria studies existing for the systems acetone+glycerol+solketal [23] and acetone+glycerol+water [45]. As the products of the reaction generate, solketal and water, the system turns into a single liquid phase. The reaction herein reported does not take place in the absence of catalysts, unless supercritical conditions are achieved [46].

Figure 8 shows the evolution of the yield to solketal obtained throughout assays with each of the resins, with yield being defined as the percentage of the ratio of the concentration of solketal at each moment to the initial concentration of glycerol. In addition to the five resins herein tested, Amberlyst 15 (A15) has been used as a reference. It is a well-known classic cation exchange resin existing for over 50 years which has been characterized [47] and used in the dehydration-hydrogenation of xylose [48] or the acetylation of glycerol, a relatively similar reaction [38]. From the plot, it can be seen that the reaction reaches its equilibrium position at approximately 5 h only when GF101 is used; however, after 6 h, equilibrium is still not reached in the cases of CT275DR, A35dry, A36dry and A15, which exhibited similar performance, and CT276.

Figure 8:

Yield to solketal obtained for each of the tested catalysts at 313 K, a molar excess of acetone to glycerol of 4.5 and a catalyst load of 0.5% w/w.

The inset graph of Figure 8 shows the progress of the reaction within the first hour of the reaction. The shapes of the curves are typical of those obtained using heterogeneous catalysts. It can be seen that, from the very beginning of the reaction, the performance observed for GF101 is better than the rest. This observation could be explained given the features described for this throughout Section 3.1. As shown by mercury porosimetry, this resin exhibits a bimodal pore distribution size curve, with an absolute maximum at 190 μm, yet with a distribution around this size that spreads well across the macroporous region. At short times of reaction, there is still a liquid-liquid biphasic system due to the limited miscibility between the two reactant species; therefore, the viscosity of glycerol plays an important role. GF101 possesses a much wider macroporous region, which in turn facilitates the access of glycerol into the pores. In addition, the solid ¹³C-NMR analysis showed that the presence of sulfonic groups in para-positions with respect to the vinyl matrix is significant in GF101 as well as in CT275DR, the resin with the second best performance. These results also agree with the results obtained from the elemental microanalysis, in which the presence of sulfur was the highest for these two resins and the lowest for CT276, the least active catalyst. Moreover, the potentiometric titration and the acid capacity of the resins also show that GF101 has a greater acidity. In summation, the availability of the active acid sites appears to be the key aspect to explain the performance of these catalysts, despite the fact that the two most active resins do not possess the highest pore volume or surface area within their structures (see Table 1).

Additionally, the effect that the presence of water generated as a byproduct may exert on the catalyst has to be taken into account. Resins based on sulfonic acids like Amberlyst 15, Amberlyst 16 and Dowex Wx8 have been tested for deactivation in the presence of water [49], [50], [51], [52], [53], reaching the conclusion that there is a high affinity of sulfonic acids for water. This could explain the deactivation of the catalyst with time as well as the fact that the equilibrium conversion is difficult to shift towards the products and only reaches approximately 47%.

The performance of this reaction using these catalysts appears to be lower than in other studies present in the literature. Nevertheless, some of these studies report on the production of solketal following strategies that remove water (by-product of the reaction) from the medium as a means to shift the reaction towards the products. For instance, conversions of glycerol as high as 81% have been attained using ZrO2-SiO2 as a catalyst operating under reactive distillation at 343 K [54]; 84% employing Ar-SBA15 also at 343 K in three consecutive two-step batches under reflux and subsequent evaporation of water under vacuum [55]; 82% using montmorillonite K10 withdrawing water with a zeolite-based membrane [56] or 90% with p-toluenesulfonic acid applying a reactive distillation strategy [20]. In other cases, the strategy followed was to include solvents as part of the reaction medium, which affects the activity coefficients of each component in the medium, leading to different equilibrium conversions. In this way, Nanda et al. [32] achieved a 65% conversion using Amberlyst 35 at 323 K using methanol as solvent and Menezes et al. [57] a 66% conversion with SnCl₂ at 333 K in the presence of acetonitrile under reflux operation. In the case of the present study, it must be acknowledged that these conditions under a mode of operation with no water removal and absence of solvents had not been tried previously to test the endurance of the catalysts.

3.3 Reutilization of the catalyst

A series of recycling assays was performed in order to assess the stability of the resins to deactivation throughout consecutive experiments. The activity of the resins was considered as the turnover frequency (TOF) observed for each resin at each cycle, established as:

where r_max is the maximum reaction rate observed (mol·l⁻¹·min⁻¹) and C_cat the equivalent concentration of protons (mol·l⁻¹), obtained from the acid capacity of the resins (Table 1).

After observing the evolution of the yield to solketal in the inset of Figure 8, the evolution of the concentration of solketal with respect to time was fitted making use of the Hoerl curve [58] to the kinetic data obtained in successive recycling experiments, as defined by Eq. (2):

where C_Sk is the concentration of solketal (mol·l⁻¹), t is time and A, B and C are the fitting parameters of the equation. From the differentiation of Eq. (2), the reaction rate as a function of time is obtained, the maximum value of which is herein considered as r_max to be used in Eq. (1).

Table 3 shows the evolution of the TOF of each resin vs. the number of recycling experiments. It can be observed that GF101 has the highest activity in practically all the cycles followed by CT275DR, as already addressed. Nonetheless, it is worthwhile mentioning that as the reutilization of the catalysts progresses, the activity of these two resins approaches similar values to those obtained by the rest of the catalysts.

Figure 9 emphasizes on these findings. This graph plots the remaining activities referred to as the first use of each resin as a function of the total operating time. Each cycle was performed for 2 h in these reutilization experiments. It can be seen that a regeneration of the catalyst with perchloric acid improves to some extent the remaining activity with respect to the non-regenerated catalyst. The remaining activities of the non-regenerated and regenerated resins have been fitted relatively well to an exponential decay model of the following type:

Figure 9:

Progress of the experimental and fitted remaining activity with total operating time and number of recycles of the catalysts. Solid markers correspond to regenerated catalysts, open markers correspond to non-regenerated.

in which a_r is the remaining activity, t refers to the total operating time of the resin and A and k_d are the amplitude and the decay constant, i.e. the fitting parameters of the curve.

Table 4 compiles the values of the mentioned parameters as well as the half-life period. This value corresponds to the number of cycles at which the model predicts that the remaining activity will be equal to 0.5.

Both in Figure 9 and from the half-life value in Table 4 it can be inferred that GF101 suffers the highest decrease in activity despite showing the best performance in all the cycles. CT275DR, A35dry, A36dry and A15 have all similar half-life values of around six cycles. Curiously, the least active species in this work, i.e. CT276, showed the lowest decay in activity with only a loss of 18% after the first four cycles when no regeneration was made and 24% when such a procedure was performed, reaching lower, although somewhat similar values to those obtained for silica-included heteropolyacids, where the activity drop was estimated to be up to 13% after the same number of reutilization cycles [59].

For the present study, high purity glycerol was used to ensure maximum activities and stabilities for the catalysts tested. However, results are expected to be almost identical in what regards stability if technical-grade glycerol is employed, expecting a rapid deactivation of the catalyst if crude glycerol is the reagent. The latter result can be explained by cation exchange between sodium and protons, as crude glycerol is rich in sodium salts [55].