Article pubs.acs.org/IECR
Comparative Study of Reactive Flash Distillation vs Semibatch Reactor Technologies for the Glycerol Hydrochlorination with Gaseous HCl Cesar A. de Araujo Filho,*,† Kari Eran̈ en,† Jyri-Pekka Mikkola,†,‡ and Tapio Salmi† †
Department of Chemical Engineering, Johan Gadolin Process Chemistry Centre, Laboratory of Industrial Chemistry and Reaction Engineering, Åbo Akademi University, FI-20500 Turku/Åbo, Finland ‡ Department of Chemistry, Chemical-Biochemical Centre, Technical Chemistry, Umeå University, SE-90187 Umeå, Sweden S Supporting Information *
ABSTRACT: The present work provides a systematic comparison of solvent-free glycerol hydrochlorination with semibatch and reactive flash distillation technologies. All the experiments were performed at atmospheric pressure and constant flow rate of gaseous HCl in the temperature range of 70 to 120 °C. Both acetic acid and adipic acid were used as homogeneous catalysts, separately, at a concentration of 12% by moles of each. In addition, a series of noncatalytic experiments was investigated. A comparative analysis between reactive flash distillation and semibatch operation suggested that reactive flash distillation only increases the production rate of the desired product 1,3-dichloropropan-2-ol (αγ-DCP) for the highest temperature, i.e. 120 °C. Many aspects of the HCl liquid uptake were also exposed once water was allowed to leave the liquid phase, revealing that water also has a positive influence on the reaction rate because it promotes HCl solubility and hydrolysis. Such an important effect was not predicted by previous investigations, and it is hereby described for the first time. Additional semibatch experiments were conducted, in which different amounts of water and αγ-DCP were initially added. It was found that, water exerts competing effects in the glycerol hydrochlorination; addition of αγ-DCP showed an improvement of the reaction kinetics and decrease of HCl consumption. It is suggested that instead of using reactive distillation, a wiser choice to perform glycerol hydrochlorination would be to recycle large amounts of the product stream to achieve high conversion levels at milder temperatures and consuming less HCl gas.
1. INTRODUCTION The world supply of glycerol is strongly influenced by the production of first generation biodiesel,1 because glycerol is a stoichiometric coproduct of this process, constituting approximately 10% of the total weight of the final product.2 Even though oil prices have decreased significantly during the past year, several countries have committed to increasingly blend their fossil diesel with biodiesel;3 therefore, one may still expect a considerable production of crude and low-quality glycerol in the upcoming years. In order to wisely make use of this inexpensive and abundant raw material, the production of fine chemicals by glycerol upgrading has been the subject of intensive investigations.1,2,4−11 One of the alternatives investigated is the hydrochlorination of glycerol to dichlorohydrins for production of epichlorohydrin,12−14 which can be further used to e.g. make epoxy resins. This market has increased significantly in the recent years, especially in the Asia-Pacific area, reflecting in the start-up of commercial plants in Thailand, China, and South Korea.15 Glycerol hydrochlorination is a parallel-consecutive reaction, commonly performed in the presence of moderately weak organic acids as homogeneous catalysts, e.g. acetic, propionic, and succinic acids,16,17 in a temperature range of 70−120 °C. © 2016 American Chemical Society
Recently, it has been proven by our group that the noncatalytic reaction pathway of glycerol hydrochlorination can also be considerable at elevated temperatures, e.g. 105 and 120 °C.18 Hydrogen chloride (HCl) can be added to the system both in aqueous form,19,20 i.e. hydrochloric acid, or as gas.16,18,21 An overview of the glycerol hydrochlorination reaction is depicted in Figure 1. The overall hydrochlorination process is irreversible, even though there are reversible steps within its mechanism.16,18,22 The presence of water has shown to retard the glycerol hydrochlorination kinetics, the reason why this reaction is preferably conducted using HCl gas instead of aqueous solutions of hydrochloric acid.16,18,23 A more detailed analysis of the reaction mechanism illustrated in Figure 2 is thoroughly discussed by de Araujo Filho et al. (2014). The first hydrochlorination can take place in any hydroxyl group of the glycerol molecule; however, the 3-chloro-1,2propanediol (α-MCP) is the largely preferred product. The second chlorination can only happen from the monochlorohyReceived: Revised: Accepted: Published: 5500
August 30, 2015 April 16, 2016 April 27, 2016 April 27, 2016 DOI: 10.1021/acs.iecr.5b03195 Ind. Eng. Chem. Res. 2016, 55, 5500−5513
Article
Industrial & Engineering Chemistry Research
Figure 1. Overview of the reactions in glycerol hydrochlorination.
Figure 2. Reaction mechanism of glycerol hydrochlorination.
drin α-MCP, and the preferred product is αγ-DCP. In fact, 1,2dichloro-3-propanol (αβ-DCP) is formed in trace amounts only. A number of papers have been published on this topic; most of them have focused on the batch or semibatch operation and product distribution in the presence of aqueous or gaseous hydrogen chloride16−21,24 and weak organic acids as catalyst. Despite few disagreements on the detailed reaction mechanism reported in different sources,16,18,21 it is generally accepted that water exerts a negative effect on the glycerol hydrochlorination with hydrochloric acid solutions.15,20,25 A plausible explanation
for this observation can be drawn by analyzing the works of Hughes and Ingold,26,27 who studied the solvent influence in many substitution and elimination reactions. It was found that, for reactions that depend upon the interaction between ions of different charges, an increase in solvent polarity would slow down the reaction rate. Such an effect is even more pronounced if the interaction between the ions would promote the destruction of the charge. Indeed, the glycerol hydrochlorination reaction relies on the interaction of two pairs of opposite charged ions (vide steps 3, 4, 7, and 8 in Figure 2), in which there is charge destruction. Such polarity phenomena is well 5501
DOI: 10.1021/acs.iecr.5b03195 Ind. Eng. Chem. Res. 2016, 55, 5500−5513
Article
Industrial & Engineering Chemistry Research
Figure 3. Solvent-free glycerol hydrochlorination under semibatch and reactive flash distillation regimes. 8% malonic acid was used as catalyst for both experiments. The f igure is based on tables published in refs 16 and 17.
described by the books of Reichard and Welton28 and Amis.29 In addition, from a chemical kinetic viewpoint, water should exert a negative effect on glycerol hydrochlorination, because it appears on the right-hand side of equilibrium steps 1, 2B, and 5 of the reaction mechanism. As a matter of fact, the kinetic models developed18,23 suggest that water removal from the reaction medium during the solvent-free glycerol hydrochlorination shall increase the reaction rate. It has been proposed16,23,25 that a reactive flash distillation regime, where water would be continuously stripped from the reaction mixture, would provide a combination of an enhanced reaction rate and a preseparation unit, facilitating the downstream processing of the liquid. The same concept has been widely applied on the polyesterification reaction under reactive distillation regime.30,31 Notwithstanding, only a limited amount of data is available in the literature presenting glycerol hydrochlorination under reactive flash distillation conditions. Santacesaria and coworkers have published two separate papers16,17 on the solvent-free glycerol hydrochlorination where they explicitly report the mole fractions of semibatch and what appears to be a reactive flash distillation experiments using 8% of malonic acid as a homogeneous catalyst. Figure 3 illustrates the product distribution published on two separate articles.16,17 Interestingly, the two graphs presented in Figure 3 suggest that the temperature increase approximates the hydrochlorination kinetics in reactive distillation and semibatch operation. Granted that the boiling points of glycerol, α-MCP, and αγDCP (vide Table 2) are much higher than the reaction temperatures in question (100 and 110 °C), it is improbable that they would escape from the reaction mixture. Nevertheless, no definite conclusions can be drawn by examining the limited data available in open literature. A more systematic comparison between semibatch and reactive flash distillation experiments is needed. Luo et al.32 conducts a steady-state analysis of pilot-scale experiments and simulations of glycerol hydrochlorination under reactive distillation conditions. However, the paper was more focused on the process design aspects of glycerol hydrochlorination, and no discussion was presented on the performance of the reactive distillation concept compared to semibatch technology. The present work provides a systematic comparison between reactive flash distillation and semibatch experiments, aiming to evaluate which concept results in a better performance for solvent-free glycerol hydrochlorination. Due to the nature of this process, the temperature was the main parameter
investigated. Experiments were conducted in a temperature range from 70 to 120 °C, for both catalytic and noncatalytic reactions (Figure 2). Acetic acid was used for most catalyzed experiments, but adipic acid was investigated too, due to its low volatility at the temperature range studied. The performance of reactive flash distillation and semibatch experiments is directly compared, and many novel phenomena are revealed from this analysis. In order to evaluate the role of water, extra semibatch experiments were carried out in which water or αγ-DCP was added to the initial reaction mixture. As it turned out, water clearly showed to promote competing effects in the overall process of solvent-free glycerol hydrochlorination. In addition to that, the use of αγ-DCP in the initial reaction mixture showed positively surprising results, bringing new ideas for an optimized way to produce dichlorohydrins from glycerol.
2. MATERIALS AND METHODS 2.1. Experimental Apparatus. The reaction vessel was a 250 mL glass jacketed reactor, equipped with a PTFE coated radial flow impeller (Bohlender GmBH) providing vigorous stirring of the liquid phase. All the experiments were conducted at atmospheric pressure and with hydrogen chloride gas (AGA, 99.8%). A glass sinter provided a fine dispersion of the gas phase at the inlet. A manometer was coupled to one of the reactor necks to check at all times that the inner gauge pressure was zero. The HCl volumetric flow was measured by means of a rotameter. Both reactive flash distillation and semibatch experiments were conducted in the same reactor container. However, in the semibatch experiments, the condenser was connected to the reactor through one of its necks, preventing volatile components to leave the system. In the reactive flash distillation experiments, the condenser was detached from the reactor, and a pipe led the gas from the reactor to the condenser, in such a way that the condensed liquid was collected in a vessel provided with a sampling valve. The condenser temperature was kept constant at −4 °C. Figures 4 and 5 depict a schematic view of the apparatus for both operation modes. 2.2. Experimental Procedure. For the experiments performed under reactive flash distillation conditions, glycerol (Sigma-Aldrich, > 99%) was heated under vigorous stirring inside the reactor until it reached the desired temperature. Meanwhile, acetic acid (J. T. Baker, > 99%) was heated separately in a heating plate, under stirring, to the desired temperature. Once the desired temperature was attained, acetic acid was promptly weighted and transferred to the reactor. In 5502
DOI: 10.1021/acs.iecr.5b03195 Ind. Eng. Chem. Res. 2016, 55, 5500−5513
Article
Industrial & Engineering Chemistry Research
ionization detector (FID). Helium was used as a carrier gas. The injection volume was 1 μL. The column temperature program was the following: first, the initial temperature was 50 °C, and then it was increased to 90 °C at a rate of 5 °C/min; second, the temperature was increased to 280 °C at a rate of 10 °C/min and held at this temperature for 10 min. The injector temperature was 270 °C, and the detector temperature was 300 °C. This method provided the mass fractions of glycerol and its derivatives, disregarding the presence of other compounds. The mass fraction of HCl in the liquid phase was analyzed with an automatic titrator (751 GPD Titrino, Metrohm) using a standard solution of sodium hydroxide (0.2M, Merck S.A.). Two titrations were carried out for each sample, and an average was used as a final value. Table 1 depicts a statistical report concerning all the titration results. Table 1. Statistical Report of HCl Mass Fraction Doubled Analysis
Figure 4. Overview of the semibatch setup.
parameter
value
minimum standard deviation maximum standard deviation average standard deviation
6.0 × 10−4% 8.27% 0.91%
Given the nature of the analytical methods used in the present paper, the authors deem useful to define two sets of components present in our system: the bulk liquid phase and GCD (glycerol and chlorinated derivatives) compounds. The bulk liquid is the set that comprises all species in the liquid phase, and the GCD compounds are a subset that comprises only glycerol and its chlorinated derivatives. This concept was successfully applied by us when modeling solvent-free glycerol hydrochlorination in semibatch mode.18 An illustration of this concept is provided by Figure 6.
Figure 5. Overview of the reactive flash distillation setup.
the reactive flash distillation experiment with adipic acid, the catalyst and glycerol were heated together inside the reactor until it reached the reaction temperature. At this point, the HCl gas flow was switched on. This moment was set to be “time zero”. For the experiments carried out under semibatch conditions, the glycerol, catalyst, and water (in some cases) were consecutively loaded to the reactor which was previously set at the reaction temperature. The mixture was subjected to vigorous stirring until it reached the reaction temperature. Finally, the HCl valve was opened, and this moment was set to be “time zero”. The condenser was placed in a vertical position, and it was long enough so that it condensed all volatile components, except gaseous HCl. Ultimately, the gas leaving the condenser was bubbled into neutralization bottles containing a concentrated solution of sodium hydroxide, in order to be neutralized before being led to the fume hood outlet. All experiments were carried out for 3 h and with a constant HCl flow of 1.6 L/min. Samples were withdrawn (ca. 1 mL each) from the reactor vessel by means of a plastic syringe at designated times and immediately quenched to −4 °C. During the reactive flash distillation experiments, distillate samples were withdrawn each time the collector vessel had accumulated a significant amount of liquid. 2.3. Analytical Methods. The mass fraction of glycerol and its derivatives were analyzed by using a gas chromatograph (Hewlett-Packard 6890 Serie) equipped with a capillary column (J&W Scientific, HP-5, 30 m × 0.32 mm 0.25 mm) and a flame
Figure 6. Bulk liquid phase comprises all components present in the liquid phase, while the GCD compounds are a subset which constitutes glycerol and chlorohydrins.
This differentiation between bulk liquid and GCD compounds arises from the fact that the gas-chromatographic analysis is only able to detect the mass fractions of GCD compounds. Therefore, the molar fractions reported for glycerol, α-MCP, β-MCP, αγ-DCP, and αβ-DCP (GCD compounds) are calculated only with respect to each other, discarding the presence of catalyst, HCl, and water. On the other hand, the mass fractions of HCl, determined by the titration analysis, take into account all compounds present in the liquid phase (bulk liquid). 5503
DOI: 10.1021/acs.iecr.5b03195 Ind. Eng. Chem. Res. 2016, 55, 5500−5513
Article
Industrial & Engineering Chemistry Research
3. RESULTS AND DISCUSSION 3.1. Overview of Empirical Properties. The solvent-free glycerol hydrochlorination is characterized by a considerable change in the viscosity of the reaction mixture. Apart from the HCl absorption and water formation, an important cause of such an effect may be explained by the fact that the chlorinated products present significantly lower viscosity, due to the substitution of the hydroxyl by the chlorine group. Indeed, at 25 °C, the glycerol viscosity is 749.3 cP,33 meanwhile the αγDCP viscosity is 12.1 cP.34 Furthermore, it has been observed that solvents which have an OH group substituted by a Cl group presented a significant decrease of the HCl solubility. Therefore, one can infer that HCl shall experience significantly different interactions with glycerol and chlorinated compounds in the reaction mixture. In fact, it is reasonable to assume that although all compounds are subjected to the same reaction conditions, glycerol hydrochlorination may have the influence of different parameters than α-MCP hydrochlorination, due to their significantly different physicochemical properties. Such differences are more pronounced when the hydrochlorination is performed under solvent-free conditions, rather than in aqueous solutions, because the concentration of GCD compounds is much higher in the first case. In fact, under noncatalytic conditions, the conversion of glycerol may achieve significant levels, leading to the production of considerable amounts of α-MCP, which, however, remained unreacted throughout the experiment.18 Data concerning HCl dissociation in glycerol, α-MCP, and αγ-DCP are not available in the literature. However, it has been found that the equivalent conductance of HCl in aliphatic alcohols decreases dramatically with an increase of the carbon chain; as HCl being a strong electrolyte in methanol, moderate electrolyte in ethanol, and weaker electrolyte in n-propanol,35 even when HCl solubility can be found to be relatively high in these compounds.22,36 Therefore, it can be argued that glycerol may be able to absorb considerable amounts HCl, but it may not be able to dissociate it to a high extent. It is important to remark that, for hydrochlorination, it is necessary to have HCl in its dissociated form, H+ and Cl−.18,23 Finally, it is reasonable to assume that, in the reaction mixture, glycerol, acetic acid, and water shall promote the HCl solubility; however, mainly water should be responsible for its dissociation. Acetic acid was used as the main homogeneous catalyst due to its large industrial applicability and also for the sake of comparison of the reactive flash distillation with semibatch data published previously.18 Adipic acid has a much higher boiling point than acetic acid (Table 2); therefore, its use aimed to investigate the influence of a nonvolatile catalyst under semibatch and reactive flash distillation conditions. 3.2. Reactive Flash Distillation Experiments. 3.2.1. Distillate Analysis. For the experiments performed under reactive flash distillation conditions, the analysis of the distillate showed that glycerol and its chlorinated derivatives were present in negligible amounts, so that they could be reasonably assumed to be in batch mode despite the presence of the outside condenser. Luo et al.32 also confirmed this observation experimentally when studying the steady-state glycerol hydrochlorination on a reactive distillation column, where they report negligible amounts of α-MCP and αγ-DCP in the overhead distillate. In fact, the boiling points of GCD compounds are high given the studied temperature range (70−120 °C), and it is unlikely to consider that they would evaporate in significant
amounts. Table 2 presents the boiling points of the main compounds in the reaction mixture. Table 2. Atmospheric Boiling Point of the Most Abundant Species in the Reaction System33 T (°C) glycerol α-MCP αγ-DCP adipic acid acetic acid water HCl
290 213 174 338 117 100 −85
It can thus be assumed for our system that only acetic acid, water, and HCl are stripped under the reactive flash distillation regime. Figure 7 depicts the cumulative mass of distillate for acetic acid catalyzed and noncatalyzed experiments. It is observed that the mass of distillate increases with an increase of temperature for both catalyzed and noncatalyzed experiments. Nevertheless, the amount of distillate collected is considerably higher for catalyzed experiments. This is mainly due to the presence of acetic acid, which catalyzes hydrochlorination, producing considerable amounts of water which is stripped. The HCl analysis of the distillate samples is displayed in Figure 8. From Figure 8, a high concentration of HCl in the distillate samples for both sets of experiments is observed, even though HCl is not supposed to condense at −4 °C. This happens because gaseous HCl is transferred to the distillate not through condensation but, instead, by absorption. The liquid film formed on the surface of the condensing tube readily absorbs the unreacted HCl gas passing in the vicinity, dragging it to the condensed phase. This provides an evidence, for future discussion, that gaseous HCl solubility is very much dependent on the presence of liquid water. It is also evidenced from Figure 8 that the concentration of HCl in the distillate is lower for catalyzed compared to noncatalyzed experiments. This can be attributed to the fact that catalyzed experiments, having considerably higher conversions, produce more water that is able to evaporate, consequently diluting HCl concentration in the distillate. In addition to that, catalyzed experiments consume more HCl that could eventually be absorbed in the condenser. 3.2.2. Reactor Analysis. Figure 9 depicts the glycerol mole fraction inside the reactor for catalyzed and noncatalyzed experiments at different reaction temperatures. From Figure 9 it is observed that the glycerol consumption increases with an increase of temperature for both sets of experiments, much similar to the behavior described by de Araujo Filho et al. (2014) for semibatch experiments. Figure 10 depicts the HCl liquid uptake for acetic acid catalyzed and noncatalyzed experiments performed at different temperatures. Interestingly, for the experiments performed at temperatures exceeding or close to the atmospheric boiling point of water, the HCl uptake reaches a maximum and monotonically decreases thereafter. Such an effect suggests that water plays a major role on the solubility of HCl in the reaction mixture. It is rational to assume that, in the very beginning of the reaction, 5504
DOI: 10.1021/acs.iecr.5b03195 Ind. Eng. Chem. Res. 2016, 55, 5500−5513
Article
Industrial & Engineering Chemistry Research
Figure 7. Cumulative mass of distillate in noncatalyzed and catalyzed experiments. Initial amount of glycerol loaded was 200 g.
Figure 8. HCl mass fraction in distillates from catalyzed and noncatalyzed experiments.
Figure 9. Glycerol mole fraction for catalyzed and noncatalyzed experiments, at different temperatures and 200 g of initial glycerol load. *Mole fraction of GCD compounds calculated only with respect to each other.
glycerol is the seed that initiates the HCl dissolution and dissociation. However, as the reaction proceeds, mono- and dichlorohydrins are formed, and the HCl solubility declines. Therefore, water is the component responsible to absorb HCl after the reaction has proceeded for some time. However, if the water is allowed to leave the liquid phase, the HCl solubility should consequently decrease after a certain degree of glycerol conversion. Figure 11 compares the HCl liquid uptake and product distribution of the experiments performed using 12% adipic acid and 12% acetic acid at reactive flash distillation regime. The compounds β-MCP and α,β-DCP are excluded from this and all future product distribution analysis due to their very low
yields (maximum molar fractions ca. 0.05 and 0.003, respectively) and static behavior.16,18,32 From Figure 11 it is observed that the HCl uptake presents a similar pattern for both experiments, although the one performed with acetic acid shows a higher HCl uptake. On the other hand, by analyzing the product distribution depicted in Figure 11, one can see that the production of αγ-DCP is higher for the experiment using adipic acid, evidencing that it has more rapid kinetics. This result is in agreement with the results presented by Santacesaria et al. (2010), which demonstrated that adipic acid does enhance the hydrochlorination reaction rate more than acetic acid for semibatch experiments. 5505
DOI: 10.1021/acs.iecr.5b03195 Ind. Eng. Chem. Res. 2016, 55, 5500−5513
Article
Industrial & Engineering Chemistry Research
Figure 10. HCl liquid uptake at different temperatures for noncatalyzed (left) and catalyzed (right) experiments. Initial glycerol loaded = 200 g.
Figure 11. Comparison of experiments performed at 105 °C under reactive flash distillation regime. Both experiments had 200 g of initial load of glycerol and 12% by moles of catalyst (acetic or adipic acid). *Mole fraction of GCD compounds calculated only with respect to each other.
catalyst, both in semibatch and reactive flash distillation regimes. Indeed, it is observed from Figure 13 that the semibatch experiment is slightly faster than the reactive flash distillation due to its similar glycerol consumption but higher production of αγ-DCP. The HCl liquid uptake also reaches a maximum and considerably decreases for the experiment at reactive flash distillation conditions. Analogous to the previous discussion concerning the noncatalytic experiments, water removal might be responsible for these changes observed. It may be inferred that the lack of water in the reactive flash distillation system compromises both the HCl absorption capacity and the HCl dissociation, which have a negative effect on the hydrochlorination reaction rate. At this point, the data presented no clear evidence that the removal of water would enhance the hydrochlorination reaction rate. Figure 14 depicts the product distribution and HCl liquid uptake for experiments performed with 12% of the initial acetic acid concentration, both in semibatch and reactive flash distillation regimes. Figure 14 provides an overview of the catalytic semibatch and reactive flash distillation processes in a wide range of temperatures (70−120 °C) and, therefore, exposes different kinetic and mass transfer phenomena in action. As a matter of fact, Figure 7 already revealed that, for temperatures exceeding the boiling point of water, the distillate cumulative increases dramatically, evidencing that there should be a turning point around this temperature. It can be noticed from Figure 14 that, for the experiments conducted under the atmospheric boiling point of water (100 °C), the first chlorination is disfavored in the reactive
3.3. Comparison between Reactive Flash Distillation and Semibatch Experiments. As previously reported in the literature,18,20 noncatalytic glycerol hydrochlorination is unable to produce significant amounts of dichlorohydrins, e.g. αγ-DCP and αβ-DCP. The most significant product formed both in semibatch and reactive flash distillation mode is the monochlorohydrin α-MCP, although small quantities of βMCP are also detected. Figure 12 depicts the product distribution and HCl liquid uptake for noncatalytic experiments performed at different temperatures and reaction regimes. It is observed from Figure 12 that, at the same temperature, the glycerol consumption is faster for semibatch rather than for reactive flash distillation experiments; furthermore, the corresponding HCl liquid uptake is higher for semibatch as well. Interestingly, such differences increase with temperature. This can be attributed to the fact that, at the reactive flash distillation mode, water is continuously stripped from the liquid phase, causing significant decrease in the HCl dissociation and, hence, diminishing the reaction rate. Therefore, it is reasonable to suggest that water plays a positive role on the solvent-free glycerol hydrochlorination, because it promotes the dissociation of HCl. Though logical, such an observation had not yet been experimentally demonstrated and has often been neglected in the literature when proposing reactive distillation as an improved alternative over semibatch operation for this reaction system. The first catalytic experiments to be compared are the ones conducted with adipic acid as the catalyst, which is considered to be a nonvolatile compound under the reaction conditions. Figure 13 depicts the product distribution and the HCl liquid uptake for experiments performed using 12% of adipic acid as 5506
DOI: 10.1021/acs.iecr.5b03195 Ind. Eng. Chem. Res. 2016, 55, 5500−5513
Article
Industrial & Engineering Chemistry Research
Figure 12. continued
5507
DOI: 10.1021/acs.iecr.5b03195 Ind. Eng. Chem. Res. 2016, 55, 5500−5513
Article
Industrial & Engineering Chemistry Research
Figure 12. Glycerol and α-MCP mole fractions and HCl liquid uptake for noncatalytic experiments under semibatch and reactive flash distillation conditions. All experiments were performed with an initial glycerol load of 200 g and an HCl volumetric flow of 1.6 L/min. *Mole fraction of GCD compounds calculated only with respect to each other.
Figure 13. Product distribution and HCl liquid uptake for semibatch and reactive flash distillation experiments performed using 12% adipic acid as catalyst and at 105 °C. *Mole fraction of GCD compounds calculated only with respect to each other.
distillation regime, which might be attributed to the previously discussed promoting effect that water exerts on the HCl dissociation. Nevertheless, this effect diminishes gradually with the increase in temperature, even though more water is removed in those conditions. At this point, the positive effect of water removal, previously described in the Introduction, is revealed. At temperatures of 105 and 120 °C the first chlorination reaction rate equals in both semibatch and reactive flash distillation mode, similar to the behavior observed in Figure 13, despite the decrease in HCl liquid uptake. The overall effect of water removal is not precisely clear in the analysis of formation of the desired product, i.e. αγ-DCP, because this reaction is coupled with the first chlorination step, which is already subject of competing effects. Nevertheless, it is observed that the only temperature in which reactive flash distillation shows an improvement over semibatch technology for the production of αγ-DCP is at 120 °C. Interestingly, the effect of water removal, better observed in the reactive flash distillation experiment at 120 °C in Figure 14, seems to be considerably important at temperatures well above water boiling point. Even though the HCl liquid uptake suffers from a considerable diminishment due to the stripping of water, the observed reaction rates are faster than under semibatch conditions. 3.4. Semibatch Experiments with Initially Added Water and αγ-DCP. In order to experimentally demonstrate that water has a positive effect on the HCl liquid uptake profiles (as strongly suggested in the reactive flash distillation
experiments), a series of semibatch experiments was carried out in which water was initially added to the liquid phase. The experimental matrix of this series is provided by Table 3. Figure 15 depicts the product distribution and HCl liquid uptake for the s 1, 2, 3, and 4. It is observed from Figure 15 that the glycerol consumption and αγ-DCP production are slowed down by the addition of water. That can be attributed both to the increased initial dilution and by the water retarding effects. Nevertheless, in the present discussion, the focus is not necessarily on the hydrochlorination kinetics but, instead, on the HCl uptake. It is clear from Figure 15 that the HCl liquid uptake is positively influenced by the presence of water in the system. Interestingly, a perfect linear relationship between the initial amount of water added and the HCl absorption capacity was noticed, and it is depicted in Figure 16. In Figure 16, the degree of explanation (R2) is equal to 0.999. It shows clearly that, when more water is initially present in the system, more HCl is absorbed. Finally, we can state conclusively that water plays a crucial role on the hydrogen chloride fate in the solvent-free glycerol hydrochlorination: it enhances the HCl uptake and dissociation. On the other hand, with the aim to ally demonstrate the negative effect of water on the hydrochlorination kinetics more clearly, one more was performed in semibatch conditions. Instead of adding water to the initial reaction mixture, pure αγDCP was added. The idea was to reproduce the same initial mole fractions as 3 in Table 3. Reminding that, as αγ-DCP is 5508
DOI: 10.1021/acs.iecr.5b03195 Ind. Eng. Chem. Res. 2016, 55, 5500−5513
Article
Industrial & Engineering Chemistry Research
Figure 14. continued
5509
DOI: 10.1021/acs.iecr.5b03195 Ind. Eng. Chem. Res. 2016, 55, 5500−5513
Article
Industrial & Engineering Chemistry Research
Figure 14. Glycerol and αγ-DCP mole fractions for experiments performed at different temperatures at reactive distillation and semibatch conditions. 12% by moles of acetic acid was used as catalyst, and 200 g of glycerol was the initial load. *Mole fraction of GCD compounds calculated only with respect to each other.
Table 3. Initial Amounts of Compounds Loaded to the Reactora experiment
glycerol (g)
acetic acid (g)
water (g)
1 2 3 4
200 200 200 200
17.8 17.8 17.8 17.8
0 10 25 40
All s performed in semibatch mode, at 105 °C and under a constant gas flow of HCl of 1.6 L/min.
a
the last chlorinated product, it should remain inert throughout the reaction. The al matrix representing this is shown in Table 4. In experiment 5, the presence of initial amounts of αγ-DCP in the reaction mixture requires a correction of the mole fractions detected by gas chromatograph, so that it may be comparable to other experiments carried out without excess of αγ-DCP. Therefore, the normalized mole fractions of glycerol and αγ-DCP are calculated as follows xgly norm xgly = 0 xgly (1) norm xαγ =
Figure 16. Correlation between the HCl absorption capacity and the initial amount of water added to the system.
Table 4. Initial Amounts of Compounds Loaded to the Reactor in Experiment 5a glycerol (g)
acetic acid (g)
αγ-DCP (g)
5
126.1
11.3
113.0
Experiment performed in semibatch mode, at 105 °C and under a constant HCl gas flow of 1.6 L/min.
a
0 xαγ − xαγ 0 xgly
experiment
Equation 1 intends to normalize the reactant concentration, which has been diluted with the product (αγ-DCP), while eq 2 expresses the mole fraction of αγ-DCP produced by the
(2)
Figure 15. Product distribution and HCl liquid uptake for s 1−4, performed with different initial amounts of water. *Mole fraction of GCD compounds calculated only with respect to each other. 5510
DOI: 10.1021/acs.iecr.5b03195 Ind. Eng. Chem. Res. 2016, 55, 5500−5513
Article
Industrial & Engineering Chemistry Research
Figure 17. Comparison between the product distribution and HCl liquid uptake for experiments 3 and 5. *Normalized mole fraction of GCD compounds calculated only with respect to each other.
Figure 18. Product distribution and HCl liquid uptake of experiments 1 and 5.
provide a reaction mixture which is able to achieve a more rapid conversion spending less amounts of HCl gas. αγ-DCP would work as an ideal solvent for the reaction, being generated by the process in significant amounts, requiring no additional separation efforts. Finally, in light of the results presented in this paper, we may conclude that the removal of water by reactive flash distillation does not bring such a significant improvement over semibatch technology in terms of reaction rates and product distribution, due to unforeseen negative effects brought by the removal of water from the system. Nevertheless, it is now possible to suggest that an improved technology to perform the glycerol hydrochlorination would be by means of a tailored gas−liquid reactor (bubble column or series of tank reactors), in which a part of the product stream containing αγ-DCP would be recycled to the reactant input stream.
reaction, taking into account the dilution of the glycerol. Figure 17 compares the product distribution and HCl liquid uptake for experiments 3 and 5. It can be observed from Figure 17 that the consumption of glycerol and production of αγ-DCP are considerably faster when αγ-DCP is added rather than an equivalent molar amount of water. Thus, it becomes clear that water also exerts a negative effect on the glycerol hydrochlorination kinetics. It is also noticed that the HCl liquid uptake is much lower for experiment 5, which is explained by the low solubility of HCl in αγ-DCP. Finally, let us compare the performance of experiment 1 (in which neither water nor αγ-DCP are added) and experiment 5 (where αγ-DCP is added). Figure 18 depicts the product distribution and HCl liquid uptake for the aforementioned experiments. From Figure 18 it is noticed that experiment 5 shows a faster glycerol conversion and a relatively faster αγ-DCP production, being slightly surpassed by experiment 1 only at 120 min. It shows that the presence of αγ-DCP in the beginning of the reaction enhances the hydrochlorination kinetics for semibatch systems. That is because the liquid phase in experiment 1 has a much higher polarity than the liquid phase in experiment 5, because the concentrations of water and HCl are significantly higher in the latter. Since the glycerol hydrochlorination reaction depends upon the interaction between two ions of different charges, leading to a charge destruction (as discussed in the Introduction), the diminishment of the liquid phase polarity due to addition of αγ-DCP promotes a significant enhancement of the reaction rate.28 This result is of great value. It implies that in a series of semibatch operating reactors, the recycle of αγ-DCP would
4. CONCLUSIONS A qualitative analysis of selected physical and chemical properties, such as density, viscosity, and HCl solubility, was provided for glycerol, α-MCP, and αγ-DCP molecules. It was possible to conclude that these compounds present significantly different physical-chemical properties which, in the solvent-free hydrochlorination, influence the overall kinetic and mass transfer behavior of the reaction mixture. The overall product distribution for reactive flash distillation experiments followed a similar pattern to the semibatch experiments presented previously.18 On the other hand, the HCl in liquid phase showed an unforeseen behavior. For temperatures exceeding or approaching the boiling point of water (100 °C, 1 atm), the HCl uptake reached a maximum; thereafter it decreased monotonically. Such an effect is 5511
DOI: 10.1021/acs.iecr.5b03195 Ind. Eng. Chem. Res. 2016, 55, 5500−5513
Article
Industrial & Engineering Chemistry Research attributed to the stripping of water from the liquid phase, causing a consequent decline of the HCl solubility and dissociation in the reaction mixture, which reduces the rate of hydrochlorination. Previous literature sources have considered the role of water from the viewpoint of reaction kinetics, discarding the influence of water on the HCl solubility and dissociation.16,18,23 In fact, according to the kinetic equations derived previously, water removal would enhance the overall hydrochlorination rate, because it would accelerate certain steps in the reaction mechanism. For noncatalytic experiments, the comparison between semibatch and reactive flash distillation revealed that the removal of water played a significant role on the hydrochlorination kinetics. The semibatch experiments displayed a higher conversion for all temperatures studied (70−120 °C). This observation can be explained by the removal of water in reactive flash distillation experiments, which consequently suppressed the HCl dissociation in the liquid. It is the first time that the presence of water is proven to have a positive effect on the solvent-free glycerol hydrochlorination. For adipic acid catalyzed experiments, the comparison between semibatch and reactive distillation at 105 °C demonstrated that the semibatch experiment had more rapid kinetics. Such a result is important because adipic acid is nonvolatile at the temperature studied; therefore, it eliminates the hypothesis that the reactive flash distillation experiment is slower due to catalyst evaporation. For the acetic acid catalyzed experiments, the comparison between semibatch and reactive flash distillation at different temperatures (70−120 °C) showed that only at 120 °C the performance of reactive flash distillation surpassed semibatch operation. It was the only condition, within the reactive flash distillation regime, that the kinetic effect of water removal was observed. Additional semibatch experiments, in which different quantities of water were initially added, exposed clearly that water enhances the HCl uptake in the reaction mixture. Following the same principle, a semibatch experiment was performed, where αγ-DCP was initially added. The experiment revealed that at certain conditions water plays a negative kinetic effect on the solvent-free glycerol hydrochlorination as predicted in the literature. The experiment performed with additional αγ-DCP revealed rapid hydrochlorination kinetics and a lower HCl uptake, bringing up new ideas for the industrial application of this reaction system, because αγ-DCP is produced in the hydrochlorination process and can be easily recycled to the reactant stream without any principal change on the separation units downstream. Finally, it could be proposed that, instead of performing glycerol hydrochlorination in reactive distillation mode, it could be preferably carried out in a series of tank reactors or a bubble column, where a part of the product stream containing αγ-DCP could be recycled to the reactor inlet.
■
■
in different chlorinated and nonchlorinated solvents (PDF)
AUTHOR INFORMATION
Corresponding Author
*Phone: +358 2 2154983. Fax: +358 2 2154479. E-mail: cesar. araujo@abo.fi. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work is a part of the activities of the Johan Gadolin Process Chemistry Centre (PCC), centre of excellence financed by Åbo Akademi University. The financial support given by the Graduate School of Chemical Engineering (C.A.) and from Academy of Finland (T.S.) is gratefully acknowledged. In Sweden, the Bio4Energy program, Kempe Foundation, and Knut&Alice Wallenberg Foundation through the Wallenberg Wood Science Center are acknowledged (J.-P.M.).
■
NOMENCLATURE wt % = weight percentage of HCl with respect to bulk liquid x = mole fraction of GCD compound
Subscripts and Superscripts
0 = initial quantity norm = normalized GCD mole fraction Abbreviations
■
GCD = glycerol chlorinated derivatives, formed by glycerol, α, β, αγ, and αβ chlorohydrins α-MCP = 3-chloropropane-1,2-diol β-MCP = 2-chloropropane-1,3-diol αγ-DCP = 1,3-dichloropropane-2-ol αβ-DCP = 1,2-dichloropropane-3-ol
REFERENCES
(1) Tan, H. W.; Aziz, A. R. A.; Aroua, M. K. Glycerol production and its applications as a raw material: A review. Renewable Sustainable Energy Rev. 2013, 27, 118−127. (2) Karinen, R.; Krause, A. New biocomponents from glycerol. Appl. Catal., A 2006, 306, 128−133. (3) Sorda, G.; Banse, M.; Kemfert, C. An overview of biofuel policies across the world. Energy Policy 2010, 38 (11), 6977−6988. (4) Guerrero-Pérez, M. O.; Rosas, J. M.; Bedia, J.; Rodríguez-Mirasol, J.; Cordero, T. Recent inventions in glycerol transformations and processing. Recent Pat. Chem. Eng. 2009, 2 (1), 11−21. (5) Liang, Y.; Sarkany, N.; Cui, Y.; Blackburn, J. W. Batch stage study of lipid production from crude glycerol derived from yellow grease or animal fats through microalgal fermentation. Bioresour. Technol. 2010, 101 (17), 6745−6750. (6) Villorbina, G.; Tomàs, A.; Escribà, M.; Oromí-Farrús, M.; Eras, J.; Balcells, M.; Canela, R. Combining AlCl3.6H20 and an ionic liquid to prepare chlorohydrins esters from glycerol. Tetrahedron Lett. 2009, 50 (23), 2828−2830. (7) Adhikari, S.; Fernando, S. D.; Haryanto, A. Hydrogen production from glycerol: An update. Energy Convers. Manage. 2009, 50 (10), 2600−2604. (8) Abad, S.; Turon, X. Valorization of biodiesel derived glycerol as a carbon source to obtain added-value metabolites: Focus on polyunsaturated fatty acids. Biotechnol. Adv. 2012, 30 (3), 733−741. (9) Hirschmann, S.; Baganz, K.; Koschik, I.; Vorlop, K. Development of an integrated bioconversion process for the production of 1,3propanediol from raw glycerol waters. Landbauforsch Volk. 2005, 55 (4), 261−267.
ASSOCIATED CONTENT
S Supporting Information *
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.iecr.5b03195. Table containing the changes in density and viscosity of chlorinated compounds and their parent alcohol was provided together with a summary of HCl solubility data 5512
DOI: 10.1021/acs.iecr.5b03195 Ind. Eng. Chem. Res. 2016, 55, 5500−5513
Article
Industrial & Engineering Chemistry Research
distillation in a semibatch reactor system. Chem. Eng. Sci. 1998, 53 (1), 113−121. (31) Salmi, T.; Paatero, E.; Lehtonen, J.; Nyholm, P.; Harju, T.; Immonen, K.; Haario, H. Polyesterification kinetics of complex mixtures in semibatch reactors. Chem. Eng. Sci. 2001, 56 (4), 1293− 1298. (32) Luo, Z.-H.; You, X.-Z.; Zhong, J. Design of a reactive distillation column for the direct preparation of dichloropropanol from glycerol. Ind. Eng. Chem. Res. 2009, 48 (24), 10779−10787. (33) Green, D.; Perry, R. Perry’s Chemical Engineers’ Handbook, 8th ed.; McGraw-Hill Education: New York, 2007. (34) Oshin, L. A.; Treger, Y. A.; Motsarev, G. V.; Sonin, E. V.; Sergeev, E. V.; Skibinskaya, M. B.; Golfand, E. A. Organochlorine Industrial Products (in Russian); Chemistry: Moscow, 1978. (35) Janz, G. J.; Danyluk, S. S. Conductances of Hydrogen Halides in Anhydrous Polar Organic Solvents. Chem. Rev. 1960, 60 (2), 209−234. (36) Fogg, P. G.; Gerrard, W. Solubility of Gases in Liquids: A Critical Evaluation of Gas/Liquid Systems in Theory and Practice; John Wiley & Sons: Chichester, 1991.
(10) Yazdani, S. S.; Gonzalez, R. Anaerobic fermentation of glycerol: a path to economic viability for the biofuels industry. Curr. Opin. Biotechnol. 2007, 18 (3), 213−219 Energy biotechnology/Environmental biotechnology.. (11) Zakaria, Z. Y.; Linnekoski, J.; Amin, N. Catalyst screening for conversion of glycerol to light olefins. Chem. Eng. J. 2012, 207-208, 803−813. (12) Carrà, S.; Santacesaria, E.; Morbidelli, M.; Schwarz, P.; Divo, C. Synthesis of epichlorohydrin by elimination of hydrogen chloride from chlorohydrins. 1. Kinetic aspects of the process. Ind. Eng. Chem. Process Des. Dev. 1979, 18 (3), 424−427. (13) Carrà, S.; Santacesaria, E.; Morbidelli, M.; Schwarz, P.; Divo, C. Synthesis of epichlorohydrin by elimination of hydrogen chloride from chlorohydrins. 2. Simulation of the reaction unit. Ind. Eng. Chem. Process Des. Dev. 1979, 18 (3), 428−433. (14) Tuck, C. O.; Pérez, E.; Horváth, I. T.; Sheldon, R. A.; Poliakoff, M. Valorization of Biomass: Deriving More Value from Waste. Science 2012, 337 (6095), 695−699. (15) Santacesaria, E.; Vitiello, R.; Tesser, R.; Russo, V.; Turco, R.; Di Serio, M. Chemical and technical aspects of the synthesis of chlorohydrins from glycerol. Ind. Eng. Chem. Res. 2014, 53 (22), 8939−8962. (16) Tesser, R.; Santacesaria, E.; Di Serio, M.; Di Nuzzi, G.; Fiandra, V. Kinetics of Glycerol Chlorination with Hydrochloric Acid: A New Route to alfa,gama-Dichlorohydrin. Ind. Eng. Chem. Res. 2007, 46 (20), 6456−6465. (17) Santacesaria, E.; Tesser, R.; Di Serio, M.; Casale, L.; Verde, D. New Process for Producing Epichlorohydrin via Glycerol Chlorination. Ind. Eng. Chem. Res. 2010, 49 (3), 964−970. (18) de Araujo Filho, C. A.; Eränen, K.; Mikkola, J.-P.; Salmi, T. A comprehensive study on the kinetics, mass transfer and reaction engineering aspects of solvent-free glycerol hydrochlorination. Chem. Eng. Sci. 2014, 120, 88−104. (19) Luo, Z.-H.; You, X.-Z.; Li, H.-R. Direct Preparation Kinetics of 1,3-Dichloro-2-propanol from Glycerol Using Acetic Acid Catalyst. Ind. Eng. Chem. Res. 2009, 48 (1), 446−452. (20) Dmitriev, S.; Zanaveskin, N. Synthesis of epichlorohydrin from glycerol. Hydrochlorination of glycerol. Chem. Eng. Trans. 2011, 24, 43−48. (21) Bell, B. M.; Briggs, J. R.; Campbell, R. M.; Chambers, S. M.; Gaarenstroom, P. D.; Hippler, J. G.; Hook, B. D.; Kearns, K.; Kenney, J. M.; Kruper, W. J.; et al. Glycerin as a renewable feedstock for epichlorohydrin production. The GTE process. Clean: Soil, Air, Water 2008, 36 (8), 657−661. (22) Gerrard, W. Solubility of Gases and Liquids: A Graphic Approach; Plenum Press: New York, 1976. (23) de Araujo Filho, C. A.; Salmi, T.; Bernas, A.; Mikkola, J.-P. Kinetic model for homogeneously catalyzed halogenation of glycerol. Ind. Eng. Chem. Res. 2013, 52 (4), 1523−1530. (24) Lee, S. H.; Song, S. H.; Park, D. R.; Jung, J. C.; Song, J. H.; Woo, S. Y.; Song, W. S.; Kwon, M. S.; Song, I. K. Solvent-free direct preparation of dichloropropanol from glycerol and hydrochloric acid gas in the presence of H3PMo12-xWxO40 catalyst and/or water absorbent. Catal. Commun. 2008, 10 (2), 160−164. (25) Britton, E. C.; Heindel, R. L. Preparation of glycerol dichlorohydrin. 2,144,162, January 24, 1939. (26) Hughes, E. D. Mechanism and kinetics of substitution at a saturated carbon atom. Trans. Faraday Soc. 1941, 37, 603−631. (27) Hughes, E. D.; Ingold, C. K. Mechanism of substitution at a saturated carbon atom. Part IV. A discussion of constitutional and solvent effects on the mechanism, kinetics, velocity, and orientation of substitution. J. Chem. Soc. 1935, 244−255. (28) Reichardt, C.; Welton, T. Solvents and Solvent Effects in Organic Chemistry; Wiley-VCH: Weinheim, 2006. (29) Amis, E. S. Solvent effects on reaction rates and mechanisms; Academic Press: New York, 1966. (30) Lehtonen, J.; Salmi, T.; Harju, T.; Immonen, K.; Paatero, E.; Nyholm, P. Dynamic modelling of simultaneous reaction and 5513
DOI: 10.1021/acs.iecr.5b03195 Ind. Eng. Chem. Res. 2016, 55, 5500−5513