Acid-Catalyzed Transesterification of Canola Oil to Biodiesel under

Jun 30, 2007 - Biofuels from Vegetable Oils as Alternative Fuels: Advantages and Disadvantages. Nabel A. Negm , Maram T.H. ... Biodiesel and Green Die...
3 downloads 10 Views 109KB Size
2450

Energy & Fuels 2007, 21, 2450-2459

Acid-Catalyzed Transesterification of Canola Oil to Biodiesel under Single- and Two-Phase Reaction Conditions Fadi Ataya,† Marc A. Dube´,*,† and Marten Ternan†,‡ Department of Chemical Engineering, Centre for Catalysis Research and InnoVation, UniVersity of Ottawa, 161 Louis Pasteur Street, Ottawa, Ontario, K1N 6N5, Canada, and EnPross, Inc., 147 Banning Road, Ottawa, Ontario, K2L 1C5, Canada ReceiVed March 21, 2007. ReVised Manuscript ReceiVed May 22, 2007

Experiments were performed at ambient temperature to investigate the effects of mass transfer during the transesterification reaction of canola oil with methanol (MeOH) to form fatty acid methyl esters using a sulfuric acid (H2SO4) catalyst at a MeOH/oil molar ratio of 6:1. Experiments at ambient conditions resulted in reaction rates that were slow enough to permit the effects of mass transfer on the transesterification reaction to become more evident than at higher temperatures. For the two-phase experiments, it was postulated that the reaction occurred at the interface between the phases where the triglycerides (TG), MeOH, and H2SO4 were in contact with one another. The influence of mass transfer was investigated by (a) comparing a mixed versus quiescent two-phase reaction and (b) changing a two-phase reaction to a single-phase reaction through the addition of a solvent, tetrahydrofuran. The experiments revealed the presence of an induction period prior to the initiation of the reaction, and some of the factors influencing the induction period were identified.

Introduction Biodiesel is a renewable fuel produced from vegetable oils, animal fats, or grease. Its chemical structure is that of fatty acid alkyl esters (FAAEs). Because of diminishing petroleum reserves and the environmental consequences of exhaust gases from petroleum-derived fuels, biodiesel has attracted attention during the past decade as a renewable and environmentally friendly fuel. Because biodiesel is made entirely from vegetable oils or animal fats, it is renewable, environmentally benign, biodegradable, contains very little sulfur, and does not contain any polycyclic aromatic hydrocarbons or crude oil residues. Similar to petroleum diesel, biodiesel operates in compression-ignition engines, such as those used in farm equipment and private and commercial vehicles. Biodiesel is sometimes used in its pure form, B100, or as a 5% (B5) or 20% (B20) blend with petrodiesel. Minor engine modifications may be required in older diesel engines to avoid the dissolution of various seals when B100 biodiesel is used; however, B5 or B20 blends can be used in unmodified diesel engines. Because biodiesel is oxygenated, it is a better lubricant than low-sulfur petrodiesel fuel (15 ppm of S), increasing the life of engines, and is combusted more completely.1 The addition of 1-2% biodiesel to modern ultralow-sulfur diesel fuel can produce a blend having suitable lubricating properties.2 The dominant biodiesel production process, transesterification, involves the reaction of alkyl alcohol with vegetable or animal oils in the presence of a catalyst to yield mono-alkyl esters * To whom correspondence should be addressed. Telephone: (613) 5625920. E-mail: [email protected]. † University of Ottawa. ‡ EnPross, Inc. (1) Kiss, A. A.; Dimian, A. C.; Rothenberg, G. Solid acid catalysts for biodiesel productionsTowards sustainable energy. AdV. Synth. Catal. 2006, 348, 75. (2) Van Gerpen, J. Biodiesel processing and production. Fuel Process. Technol. 2005, 86, 1097.

(biodiesel) and glycerol. The transesterification reaction consists of a number of consecutive reversible steps. When methanol (MeOH) is the alcohol, 1 mol of triglyceride (TG) is converted sequentially to diglyceride (DG), monoglyceride (MG), and finally fatty acid methyl ester (FAME) and glycerol (GLY). The overall reaction sequence for the transesterification reaction is shown below. Overall Reaction:

1TG + 3MeOH T 1GLY + 3FAME

(1)

Stepwise Reactions:

1TG + 1MeOH T 1DG + 1FAME

(2)

1DG + 1MeOH T 1MG + 1FAME

(3)

1MG + 1MeOH T 1GLY + 1FAME

(4)

The reaction system is essentially heterogeneous because the nonpolar TG phase (TG is the major component) and the polar MeOH phase (MeOH is the major component) are immiscible.3 There are several variables that influence the reaction. Increasing the temperature increases the mutual solubility of the two phases.4 Increasing the stirring intensity favors conversion.5 Nevertheless, significant mass-transfer limitations have been reported even when the reaction mixture was stirred vigorously.6 An inhibition period at the beginning of the reaction has been (3) Ma, F.; Clements, L. D.; Hanna, M. A. Biodiesel fuel from animal fat. Ancillary studies on transesterification of beef tallow. Ind. Eng. Chem. Res. 1998, 37, 3768. (4) Muniyappa, P. R.; Brammer, S. C.; Noureddini, H. Improved converison of plant oils and animal fats into biodiesel and co-product. Bioresour. Technol. 1996, 56, 19. (5) Noureddini, H.; Harkey, D. W.; Gutsman, M. R. A continuous process for the glycerolysis of soybean oil. J. Am. Oil Chem. Soc. 2004, 81, 203. (6) Boocock, D. G. B.; Konar, S. K.; Mao, V.; Sidi, H. Fast one-phase oil-rich processes for the preparation of vegetable oil methyl esters. Biomass Bioenergy 1996, 11, 43.

10.1021/ef0701440 CCC: $37.00 © 2007 American Chemical Society Published on Web 06/30/2007

Acid-Catalyzed Transesterification to Biodiesel

reported in several studies.6-9 Although conversion generally increased with the reaction time, the relationship was nonlinear and complex.7 With basic catalysts, saponification of glycerol, one of the products, has been reported to enhance its reaction with the TG reactant.4 The acid-catalyzed reaction requires significantly longer reaction times than the base-catalyzed reaction to reach any specified conversion to biodiesel.8 The addition of a cosolvent enhances the miscibility of the phases, resulting in a homogeneous mixture and a faster reaction rate.6 Alternatively, ultrasonic irradiation can be used to produce emulsions from the immiscible liquids,10 or supercritical conditions can be used to promote the formation of a homogenized medium and drastically improve reaction rates.11 The key factor limiting the conversion of TG is the degree of solubility between the phases.4 Stirring enhanced the contact between the phases and was reported to facilitate the initiation of the reaction.12 Improvements to the reaction rate in two-phase biodiesel production systems can be attained by agitation, as shown in recent work with base catalysts.13 Currently, the industrial-scale production of biodiesel fuel entails the use of homogeneous base catalysts, leading to fast reaction rates and high yields. In addition to virgin vegetable oil feedstocks, there is interest in the use of low-grade or waste feedstocks containing a relatively high free fatty acid (FFA) content. These feedstocks present appreciable downstream processing problems when processed with base catalysts. The FFAs cause the formation of soaps that inhibit the separation of biodiesel from the glycerol.14 This is an undesirable complication of the process that adds fixed costs associated with the use of several separation units for the removal of the saponified products. The merits of faster conversion with the base-catalyzed process, compared to the acid-catalyzed process, are therefore offset by the higher costs associated with the required use of virgin vegetable oils having very low concentrations of FFAs. Sometimes, the cost of the virgin vegetable oil feedstock approaches that of the petrodiesel product. In contrast, soap is not formed with acid catalysts; therefore, lower cost feedstocks that contain large concentrations of FFA can be converted to biodiesel.15 The construction of a batch pilot plant using a two-step acid-catalyzed pretreatment process to produce biodiesel from a variety of feedstocks, including those with a high FFA content, was proven to be successful and resulted in high ester conversion.15 Zhang et al.16 conducted an economic assessment and sensitivity analysis of four continuous processes that produced biodiesel from waste cooking oil (7) Darnoko, D.; Cheryan, M. Kinetics of palm oil transesterification in a batch reactor. J. Am. Oil Chem. Soc. 2000, 77, 1269. (8) Freedman, B.; Pryde E. H.; Mounts, T. L. Variables affecting the yields of fatty esters from transesterified vegetable oils. J. Am. Oil Chem. Soc. 1984, 61, 1638. (9) Noureddini, H.; Zhu, D. Kinetics of transesterification of soybean oil. J. Am. Oil Chem. Soc. 1997, 74, 1457. (10) Stavarache, C.; Vinatoru, M.; Nishimura, R.; Maeda, Y. Conversion of vegetable oil to biodiesel using ultrasonic irradiation. Chem. Lett. 2003, 32, 716. (11) Saka, S.; Kusdiana, D. Biodiesel fuel from rapeseed oil as prepared in supercritical methanol. Fuel 2001, 80, 225. (12) Ma, F.; Clements, L. D.; Hanna, M. A. The effect of mixing on transesterification of beef tallow. Bioresour. Technol. 1999, 69, 289. (13) Ataya, F.; Dube´, M. A.; Ternan, M. Single-phase and two-phase base-catalyzed transesterification of canola oil to fatty acid methyl esters at ambient conditions. Ind. Eng. Chem. Res. 2006, 45, 5411. (14) Canakci, M.; Van Gerpen, J. Biodiesel production via acid catalysis. Trans. ASAE 1999, 42, 1203. (15) Canakci, M.; Van Gerpen, J. A pilot plant to produce biodiesel from high free fatty acid feedstocks. Trans. ASAE 2003, 46, 945. (16) Zhang, Y.; Dube´, M. A.; McLean, D. D.; and Kates, M. Biodiesel production from waste cooking oil: 2. Economic assessment and sensitivity analysis. Bioresour. Technol. 2003, 90, 229.

Energy & Fuels, Vol. 21, No. 4, 2007 2451

and from virgin vegetable oil using both acid and base catalysts. The results demonstrated that the acid-catalyzed process using waste cooking oil was potentially a competitive alternative to the common base-catalyzed process. Furthermore, the price of waste cooking oil, the price of biodiesel, and the plant capacity were evaluated as the major factors affecting the economic feasibility of the biodiesel production process. The influence of process variables on the acid-catalyzed transesterification reaction has been reported in several studies.8,14,17-19 Freedman et al.8 and Canakci and Van Gerpen14 examined the effect of the alcohol type on the acid-catalyzed transesterification of soybean oil at temperatures just below the boiling points of the alcohols. The results indicated that the effect of the reaction temperature, rather than the type of alcohol used, dominates the rate of the reaction and dictates the time required to achieve complete ester conversion. The effects of the reaction temperature, molar ratio of alcohol to oil, catalyst concentration, water content, and FFA content on the acidcatalyzed reaction were also studied. Canakci and Van Gerpen14 showed that the ester conversion increased with an increasing temperature, increasing molar ratio of alcohol to oil, and increasing acid-catalyst concentration. The addition of palmitic acid, a FFA, and water prompted significant reductions in the conversion of the ester. Moreover, the effects observed for the addition of palmitic acid and the addition of water were comparable. The reaction of FFAs with alcohol produced water, which inhibited the esterification of FFAs and the transesterification of the glycerides. Recently, Liu et al.20 reported that water strongly inhibited the catalytic activity of the H2SO4 catalyst. Goff et al.18 conducted a study to investigate the efficiency of different acid catalysts at elevated temperatures under different operating conditions and determined that H2SO4 was the most effective catalyst for the transesterification reaction. Experiments that produced ester yields greater than 99% were performed at a 9:1 molar ratio of MeOH/soybean oil, 120 °C, and 1 wt % H2SO4 catalyst (on the basis of the oil) for a reaction time of 19 h. Acid-catalyzed reaction kinetics have been reported in the literature.17,19 Freedman et al.19 investigated the acid-catalyzed butanolysis of soybean oil at an alcohol/oil molar ratio of 30:1 and a 1 wt % H2SO4 catalyst concentration at different temperatures in the range of 77-117 °C. The results indicated that essentially complete conversion was achieved in 20 h at 77 °C and 3 h at 117 °C. The forward reactions appeared to follow first-order kinetics because of the large excess alcohol requirements, while the reverse reactions followed second-order kinetics. Zheng et al.17 studied the acid-catalyzed transesterification reaction kinetics of waste frying oil using MeOH/oil molar ratios in the range of 50:1-250:1 and acid-catalyst concentrations ranging from 1.5 to 3.5 mol % (on the basis of the oil) at temperatures of 70 and 80 °C. The results demonstrated that the acid-catalyzed transesterification reaction of waste frying oil in MeOH effectively follows pseudo-first-order reaction kinetics, provided that the molar ratio of MeOH/oil is ap(17) Zheng, S.; Kates, M.; Dube´, M. A.; McLean, D. D. Acid-catalyzed production of biodiesel from waste frying oil. Biomass Bioenergy 2006, 30, 267. (18) Goff, M. J.; Bauer, N. S.; Lopes, S.; Sutterlin, W. R.; Suppes, G. J. Acid-catalyzed alcoholysis of soybean oil. J. Am. Oil Chem. Soc. 2004, 81, 415. (19) Freedman, B.; Butterfield, R. O.; Pryde, E. H. Transesterification kinetics of soybean oil. J. Am. Oil Chem. Soc. 1986, 63, 1375. (20) Liu, J.; Lotero, E.; Goodwin, J. G. Effect of water on sulfuric acidcatalyzed esterification. J. Mol. Catal. 2006, 245, 132.

2452 Energy & Fuels, Vol. 21, No. 4, 2007

Ataya et al.

proximately 250:1 at 70 °C or, at the reaction temperature of 80 °C, that the molar ratio is in the range of 74:1-250:1. In this paper, a series of acid-catalyzed transesterification reactions was performed at ambient conditions under varying conditions of homogenization. The goal was to investigate the effects of agitation and removal of the two-phase interface on the conversion of TG and yield of FAME. Ambient temperature conditions were used in an attempt to have reaction rate kinetics that were slow enough for mass-transfer effects to constitute a measurable fraction of the overall rate. Experimental Section Experiments were conducted to measure the conversions of TG and DG in canola oil and the corresponding yields of DG, MG, and FAME. All experiments were performed at ambient temperature (20 °C) and pressure (1 atm) over a time period of 24 h. The reaction mixture had a MeOH (Commercial Alcohols Incorporated) to canola oil ratio of 6:1. The mass of canola oil used in each batch was approximately 10 g. The feedstock canola oil was the “noname” brand marketed by Loeb grocery stores. Sulfuric acid catalyst, H2SO4 (BDH, Inc.), at 3 wt % by weight of the canola oil was used for all reactions. The reaction mixture was placed in a 120 mL vial. Vials were cylindrical in shape, with an inside diameter of ∼38 mm and an inside length of ∼105 mm. Our analysis showed that the reaction mixture initially contained TG, DG, and FFA. The mass ratios of DG/TG and FFA/TG in the feedstock were determined by high-performance liquid chromatography (HPLC) to be 0.0135 and 0.0152, respectively. Three experimental conditions were tested: the quiescent twophase reaction, the agitated two-phase reaction, and the agitated single-phase reaction. For the single-phase reactions, the homogeneous medium was obtained by adding 20 mL of HPLC-grade tetrahydrofuran (THF) (Sigma-Aldrich) to the reaction vials at a volumetric ratio of 1.5:1 THF/total sample. The quiescent twophase reaction vials were left standing for specified times, while the agitated two-phase reaction and agitated single-phase reaction vials were agitated using a Labline multiwrist shaker at its maximum frequency of 500 shakes/min for the times specified. The analysis of the reaction products included several steps. After the reaction, the samples were immediately washed with 20 mL of water, resulting in a lower polar phase, which was discarded, and an upper nonpolar phase. This upper phase was then filtered through 0.2 µm polytetrafluoroethylene syringe filters and analyzed using HPLC according to the method reported by Dube´ et al.21 HPLC vials containing 0.04 g of sample were weighed and diluted with THF to make up 20 mg/mL sample solutions for HPLC analysis. The HPLC (Waters Corporation) consisted of a pump, a flow and temperature controller, a differential refractive index detector, and two 300 × 7.5 mm Phenogel columns of 3 µm and 100 Å pore size (Phenomenex) connected in series. The system was operated using Waters Millennium 32 software. HPLC-grade THF was used as the mobile phase at a flow rate of 0.05 mL/min at 38 °C. The sample injection loop was 200 µL, and the injected sample volume was 20 µL. The running time required for product characterization was approximately 60 min. Calibration curves (Dube´ et al.)21 were generated for the standards (Sigma-Aldrich): triolein (TG), diolein (DG), monoolein (MG), methyl oleate (FAME), oleic acid (FFA), and glycerol (GLY). The areas under the peaks in the chromatograms for the product samples were used together with the calibration curves to determine the moles of the constituents (TG, DG, MG, FFA, and FAME) present in the nonpolar phase. The amounts of product constituents were subjected to further computations to provide the conversions and yields of the components used in the analysis. The moles of TG in the feedstock were calculated from the measured number of moles of the product (21) Dube´, M. A.; Zheng, S.; McLean, D. D.; Kates, M. A Comparison of attenuated total reflectance-FTIR spectroscopy and GPC for monitoring biodiesel production. J. Am. Oil Chem. Soc. 2004, 81, 599.

Figure 1. Fractional TG conversion to all products versus time, with 3 wt % H2SO4 (percent by weight of the canola oil). For all figures, unless otherwise indicated, [ represent a quiescent two-phase reaction, 0 represent an agitated two-phase reaction, and 2 represent an agitated single-phase reaction.

constituents plus the stoichiometry of the reactions. The moles of DG and FFA in the feed were calculated from the moles of TG in the feed and the DG/TG feed ratio and the FFA/TG ratio, respectively. The moles of MeOH in the feed was 6 × (moles of TG + moles of DG) in the feed. The moles of “MeOH in the product” and “glycerol in the product” were calculated using the moles of the product constituents plus stoichiometry of the reactions. With the above information on feed and product constituents, the conversions and yields were calculated. Standard International Union of Pure and Applied Chemistry (IUPAC) definitions were used for conversion (moles of feed converted/mol of feed) and yield (moles of feed converted to a product constituent/mol of feed). The Appendix presents all of the equations used to calculate the conversions and yields.

Results and Discussion The conversion of TG in the presence of 3 wt % H2SO4 catalyst was measured as a function of time for the transesterification of canola oil. Results are shown in Figure 1 for the three different experimental regimes: the quiescent two-phase reaction, the agitated two-phase reaction, and the agitated singlephase reaction. The effect of agitation can be seen by comparing the TG conversion profile for the quiescent two-phase reaction with that for the agitated two-phase reaction. After 24 h of the quiescent two-phase reaction, the fractional TG conversion was less than 0.01, whereas it was greater than 0.05 for the agitated two-phase reaction. Except for shaking, the reaction conditions were identical in both sets of experiments. This indicated that improved mixing had a beneficial effect on the rate of reaction. The shaking motion causes material in the bulk of both phases to move toward the interface, while simultaneously, material at the interface moves to the bulk, resulting in an enhanced interchange of material between the interface and the bulk within each individual phase. Agitation might also increase the number of droplets and decrease their dimensions, thereby increasing the interfacial surface area available for reaction. The change in conversion obtained by adding THF caused the two-phase reaction to become a single-phase reaction and can be seen in Figure 1. Fractional TG conversions after 24 h reached 0.4 in the agitated single-phase system. In a singlephase, there is no interface between phases and, therefore, there can be no limitations because of interphase mass transfer. These observations are consistent with the notion that mass transfer influences the rate of reaction when an interface is present.6

Acid-Catalyzed Transesterification to Biodiesel

Figure 2. Fractional DG conversion to all products versus time, with 3 wt % H2SO4. See the caption in Figure 1.

Figure 3. Agitated two-phase reaction. Conversion of TG (XTG) and yields of DG (YDG), MG (YMG), and FAME (YFAME) versus time at 3 wt % H2SO4.

The conversion of DG during the transesterification of canola oil using 3 wt % H2SO4 catalyst is shown as a function of time in Figure 2. The negative conversions indicate that DG formation resulting from the conversion of TG (see eq 2) exceeded DG conversion to FAME (see eq 3). The negative fractional DG conversion values are much less than “-1” because there is a relatively small amount of DG initially present in the feedstock compared to the amount of TG. DG consumption results in a positive value for conversion, while DG production results in a negative value. Thus, the large negative DG conversion indicates that DG produced from the conversion of TG greatly exceeds the DG consumed in the reaction at that point in time. Figure 3 shows the same information, but the results are reported as positive values because DG is reported as a fractional yield and there is a net formation of DG from the conversion of TG. The fractional DG formation at early reaction times in the single-phase reaction exceeded those in the two-phase reaction (see Figure 2). This is an indication that the elimination of an interface and thereby the elimination of interfacial mass transfer limitations enhanced the reaction rate. A comparison of the quiescent two-phase reaction and the agitated two-phase reaction shows that DG formation with agitation exceeded that of the quiescent condition, where almost 0 DG conversion or formation took place. This demonstrates the beneficial effect of mixing on the rate of reaction. An induction period was observed prior to the initiation of TG conversion in the agitated two-phase reaction (Figure 1). The induction period is also seen in the DG conversion data for the agitated two-phase reaction (Figure 2). However, the

Energy & Fuels, Vol. 21, No. 4, 2007 2453

DG induction period is shorter than that for TG, and the DG conversion reaction initiates prior to that for TG (see Figures 1 and 2). The existence of a TG induction period in biodiesel production has been reported previously.5,6,13,19 Boocock et al.6 suggested that the TG induction period was caused by the twophase nature of the medium. His suggestion was substantiated by the absence of the TG induction period in the single-phase reaction as shown in Figure 1. The end of the TG induction period for the agitated two-phase reaction occurred at the end of the transitions from DG formation to DG conversion and back to DG formation (see Figure 2). The end of the TG induction period is seen in Figure 1, and the formationconversion-formation transition in DG is shown in Figure 2 at the same reaction time (∼8 h). There were two factors that were observed to influence the TG induction period. The first factor was agitation. The duration of the induction period in the quiescent reaction mixture was reduced when the reaction mixture was agitated (see Figure 1). This is one indication that mixing or enhanced mass transfer can affect the induction period. The second factor was the presence of an interface between the two phases. The removal of the interface by the addition of THF for the single-phase reaction resulted in no induction period (see Figure 1). Similar indications can be made with respect to the DG reaction induction period (see Figure 2). For the agitated two-phase reaction, there was a sudden increase in TG conversion of almost 4% at the end of the induction period (see Figure 1). This increase occurred simultaneously with the DG formation-conversion-formation transition, specifically, toward the end of the DG conversion and the beginning of DG formation. This conversion increase accounted for the majority of the TG conversion observed during the 24 h reaction period. A similar sudden increase in conversion was observed during some of our other experiments. There have been several explanations for the reaction mechanism. Boocock et al.6 stated that “the catalyst is located only in the methanol phase and therefore the reaction is limited by the oil concentration in that phase”. In contrast, Dasari et al.22 performed noncatalytic reaction studies and reported that “the reaction is kinetically controlled because methanol is able to diffuse into oil faster than it reacts to form methyl esters”. Both of these explanations are based on the reaction occurring in a bulk phase, the polar MeOH phase in one case and the nonpolar TG phase in the other. In contrast, the beef tallow experiments reported by Ma et al.12,23 led to observations that “without mixing, the reaction occurred only at the interface”.12 Several additional experiments were carried out to evaluate these three explanations. First, visual observations were made to further characterize the nature of the two phases. With a MeOH/canola oil molar ratio of 6:1, the canola oil nonpolar phase (TG-rich) occupied a larger fraction of the total liquid volume than the polar phase (MeOH-rich). The molecular weight of TG (885.4 g/mol) is much greater than that of MeOH (32.04 g/mol), and their densities are 910 and 790 kg/m3, respectively. The nonpolar TG phase had the largest volume; therefore, it was the continuous phase during agitation, and the droplets of dispersed phase were composed of the polar MeOH phase. When reaction mixtures were allowed to settle in vials, the frothy mixture that (22) Dasari, M. A.; Goff, M. J.; Suppes, G. J. Noncatalytic alcoholoysis kinetics of soybean oil. J. Am. Oil Chem. Soc. 2003, 80, 189. (23) Ma, F.; Clements, L. D.; Hanna, M. A. The effects of catalyst, free fatty acids, and water on transesterification of beef tallow. Trans. ASAE 1998, 41, 1261.

2454 Energy & Fuels, Vol. 21, No. 4, 2007

Ataya et al.

existed during agitation immediately began to separate into two layers. The top layer occupied the smaller volume (polar MeOH phase), while the bottom layer (nonpolar TG phase) occupied the larger volume. Initially, both layers were emulsions. The top layer became clear in less than 2 h, whereas the bottom layer took more than 2 days to clarify. Second, some of the individual feedstock components were mixed together in an attempt to understand their degree of immiscibility. Feedstock MeOH that contained water was mixed with H2SO4 in the feedstock proportions, and the MeOH, H2O, and H2SO4 components were completely miscible. Feedstock MeOH that contained water was mixed with FFA (oleic acid) in the feedstock proportions, and the MeOH, H2O, and FFA components were completely miscible. Feedstock MeOH that contained water was mixed with FFA and H2SO4 in the feedstock proportions, and in this case, the MeOH, H2O, FFA, and H2SO4 mixture formed an emulsion. Liu et al.20 investigated the impact of water on the acidcatalyzed esterification of acetic acid with methanol and suggested that the decreased catalytic activity of the protons is a result of the preferential solvation of the acidic species by water over methanol. Moreover, Gileadi and Kirowa-Eisner reported that water molecules, in the presence of methanol, act as scavengers for protons to form H3O+ cations.24 Therefore, in the case where MeOH, H2O, FFA, and H2SO4 were mixed in feedstock proportions and the emulsion was formed, a core of H3O+ cations and their sulfate anions may have formed micelles that have FFA molecules on their exterior. The OH end of the FFA would be adjacent to the H3O+ core and the hydrocarbon end in the polar MeOH phase. The emulsion would include FFA-H3O+ micelles having acid cores inside the polar MeOH phase that are shielded from contact with MeOH. Third, the analysis of the canola oil feedstock showed that it contained TG, DG, and FFA. Because the feedstock would have had a long time for reactions between its components to come to equilibrium, the following reaction would ensure that the canola feedstock contained at least a small amount of water:

TG + H2O T DG + FFA

(5)

Lotero et al.25 stated that the water molecules affect the catalyst accessibility to the TG molecules and inhibit the reaction. The presence of acid in both the dispersed and continuous phases was confirmed by experiment. After the reaction, the polar MeOH phase and the nonpolar TG phase were allowed to separate. Samples of each of the phases were removed, and titration with calcium hydroxide, Ca(OH)2, showed that both the polar MeOH phase and the nonpolar TG phase contained acidic species. Therefore, when H2SO4 contacts the canola oil feedstock, micelles would be expected to form. They would have H3O+ cores and FFA molecules at the interface, thereby shielding them from contact with TG molecules. FFA has an OH group on one end of a hydrocarbon chain. This would explain how the acid would be inside the nonpolar TG phase but be shielded from contact with the TG molecules by the FFA-H3O+ micelles. From the above, it is likely that, during settling, the bottom emulsion may have consisted of FFA-H3O+ micelles in the (24) Gileadi, E.; Kirowa-Eisner, E. Electrolytic conductivitysThe hopping mechanism of the proton and beyond. Electrochim. Acta 2006, 51, 6003. (25) Lotero, E.; Liu, Y.; Lopez, D. E.; Suwannakarn, K.; Bruce, D. A.; Goodwin, J. G. Synthesis of biodiesel via acid catalysis. Ind. Eng. Chem. Res. 2005, 44, 5353.

Table 1. Octanol-Water Partition Coefficients (Kow) component

log Kow

glycerol water methanol oleic acid (FFA) MG FAME DG TG

-1.8 -1.5 -0.74 7.64 6.04 8.02 14.64 23.29

nonpolar TG phase. Similarly, it is likely that, during settling, the top emulsion may have consisted of FFA-H3O+ micelles in the polar MeOH phase. The acid-catalyzed esterification reaction involves contact between the protonated FFA molecules and MeOH (see eq 6). The polar carboxylic group in the FFA molecules interacts with water through hydrogen bonds, thereby facilitating the FFAcatalyst interaction and the esterification reaction.23 Therefore, the presence of the FFA-H3O+ micelles in both phases allows for the protonation of the carbonyl group and the formation of the protonated carboxylic acid. This is the first step for the acidcatalyzed esterification reaction. The second step involves the reaction with the weak nucleophile MeOH.

FFA + MeOH T FAME + H2O

(6)

The octanol-water coefficient, Kow, is the ratio of the equilibrium concentration of a species in octanol to that in water at a specified temperature. Kow represents the tendency of a chemical species to partition itself between an organic and aqueous phase. This implies that species having similar Kow values will be miscible and vice versa. In accordance with the Kow values for the present study (see Table 1), most of the MeOH is expected to be in the polar MeOH phase, thereby allowing the esterification reaction to proceed readily in that phase, whereas the acid-catalyzed esterification reaction in the nonpolar TG phase is limited by the concentration of MeOH in that phase. The reaction of FFA molecules in the polar methanol phase would probably lead to a more pronounced destabilization or disintegration of the FFA-H3O+ micelles in the polar MeOH phase as compared to the nonpolar TG phase, and that may result in the difference in clarity (attributed to the removal of FFA-H3O+ micelles), upon settling, as observed in our experiments. The reaction medium is composed of polar MeOH phase droplets dispersed in the continuous nonpolar TG phase. On the basis of the octanol-water partition coefficients listed in Table 1, it is useful to speculate on the composition of each of the phases.26 At equilibrium, the chemical potential of each component in one of the phases will be equal to its chemical potential in the other phase. Nevertheless, the concentration in one phase could be much different than that in the other phase. On the basis of the values in Table 1, it might be expected that (a) the polar MeOH phase would contain most of the MeOH, H2O, and GLY, (b) the nonpolar TG phase would contain most of the TG and DG, and (c) the FFA, MG, and FAME would be distributed between the two phases. The feed components in the reaction (TG, DG, FFA, MeOH, H2O, and H2SO4) are expected to be distributed accordingly. TG and DG would be expected to be present in the nonpolar TG phase, whereas MeOH and H2O would be expected to be present in the polar (26) Allen, D.; Shonnard, D. Green Engineering: EnVironmentally Conscious Design of Chemical Processes; Prentice Hall PTR: Upper Saddle River, NJ, 2001.

Acid-Catalyzed Transesterification to Biodiesel

MeOH phase. FFA would be expected to present in both phases, and the titration experiments performed confirmed the presence of the acidic species in both phases. Surfactant molecules can be dispersed as single molecules, form aggregates or micelles, or adsorb as a film at the interface, and a dynamic equilibrium exists between theses states, resulting in an equal rate of adsorption and desorption.27 DG and FFA are surface-active agents and, as such, are expected also to be present at the interface as well as in the bulk polar and nonpolar phases. It is expected that the FFA molecules, owing to their chemical structure, are present at the interface because the FFA molecules contain both a polar and nonpolar group. The DG molecules contain one hydroxyl group each, rendering them slightly polar, while their hydrocarbon part can keep them in contact with the nonpolar phase. This allows the DG molecules to also be present at the interface, in addition to the FFA molecules. In contrast, the TG molecules are essentially nonpolar and thereby preferentially remain in the bulk of the nonpolar phase. Esterification reactions are faster than transesterification reactions. Sendzikiene et al.28 studied the kinetics of the acidcatalyzed esterification of FFA with MeOH and reported that the apparent reaction rate constants for the experiments carried out at 20, 34, and 60 °C were 1.27 × 10-4, 1.63 × 10-4, and 2.33 × 10-4 s-1, respectively, whereas studies for the acidcatalyzed transesterification reaction indicated values that were an order of magnitude smaller.19 For example, the acidcatalyzed transesterification of soybean oil with 1-butanol using 1% H2SO4 at 77 °C resulted in a forward reaction rate constant of 5.00 × 10-5 s-1 for the TG-DG reaction.19 The TG and DG conversion profiles for the two-phase reactions show the presence of an induction period prior to the initiation of the reaction, and the DG induction period is shorter than the TG induction period (see Figures 1 and 2). FFA conversion results have been reported elsewhere18 for the conversion of TG using an acidic catalyst. No FFA lag or induction period has been documented in the open literature; in fact, FFA reaction profiles indicate an almost instantaneous reaction of FFA.20 The acid-catalyzed transesterification reaction is a two-step procedure that involves contact between the protonated TG molecules and MeOH. The first step in the reaction mechanism, similar to that of the esterification of FFA, involves protonation of the carbonyl group of the ester by the acid catalyst, forming a resonance-stabilized complex and converting it into a strong electrophile. The second step involves the reaction with the weak nucleophile, MeOH.29 It is the protonated TG molecules that need to interact with MeOH for a reaction to occur. Contact between the TG molecules and the FFA-H3O+ micelles may occur in the nonpolar TG phase during agitation; however, as a result of the limited concentration of MeOH in that phase, the interaction between the protonated TG molecules and MeOH would be limited in that phase and any contact between theses species would probably occur at the interface. The combination of experiments described above strongly supports the “reaction at the interface” hypothesis rather than the reac(27) Patist, A.; Kanicky, J. A.; Shukla, P. K.; Shah, D. O. Importance of micellar kinetics in relation to technological processes. J. Colloid Interface Sci. 2002, 245, 1. (28) Sendzikiene, E.; Makareviciene, V.; Janulis, P.; Kitrys, S. Kinetics of free fatty acids esterification with methanol in the production of biodiesel fuel. Eur. J. Lipid Sci. Technol. 2004, 106, 831. (29) McMurry, J. Organic Chemistry, 4th ed.; Brooks/Cole Publishing Co.: New York, 1996.

Energy & Fuels, Vol. 21, No. 4, 2007 2455

tion in either of the bulk phases. The TG and DG conversion results are consistent with the “reaction at the interface” hypothesis. In the polar MeOH phase, FFA molecules (RCOOH) and the catalyst will react with MeOH in an esterification reaction to form FAME and water, according to eq 6. MeOH is the major component in the polar MeOH phase and is present in excess as a reactant. The acid-catalyzed esterification reaction is a comparatively fast reaction. The reaction will proceed until equilibrium is achieved. The reaction of FFA molecules within the polar MeOH phase would result in a net decrease in the number of FFA molecules, the number of FFA moles, and the concentration of the FFA species in the bulk polar phase. The difference in concentration between FFA at the interface and FFA in the bulk is a driving force that causes some of the FFA at the interface to transfer into the bulk polar phase. The flux of some of the FFA molecules from the interface to the bulk may permit the number of DG molecules at the interface to increase. The FFA and DG species are both surfactants, and a decrease in the interface concentration of FFA should result in a net movement of DG molecules from the bulk nonpolar TG phase to the interface, whereby the DG molecules will occupy sites at the MeOH-TG interface previously occupied by the FFA molecules that transferred in the bulk polar MeOH phase. Both the reaction equilibrium and phase equilibrium (masstransfer equilibrium) should be satisfied. FFA esterification is a faster reaction than DG transesterification.19,28 FFA molecules are expected to be present in the polar MeOH phase and the protonation of the FFA species in that phase is expected to be sterically promoted. Therefore, the FFA esterification reaction in the polar MeOH phase is both kinetically and sterically promoted. The octanol-water partition coefficients (see Table 1) show that FFA is expected to be more soluble in the polar MeOH phase than DG; i.e., it is expected that the number of FFA molecules in the polar MeOH phase is greater than the number of DG molecules in that phase. Therefore, the DG transesterification reaction in the polar MeOH phase is both kinetically and sterically hindered and less likely to occur. In the nonpolar TG phase, during agitation, the FFA-H3O+ micelles are likely to collide frequently with the TG molecules of the continuous phase. During these collisions, an acidic molecule may contact and associate with a TG molecule at the FFA-H3O+ micelle-TG interface, forming a protonated TG molecule. The TG transesterification reaction is influenced by the concentration of MeOH in that phase. It is likely that the reaction in eq 2 (TG + MeOH T DG + FAME) did not occur in the first 5 h, because the data in Figure 3 show that FAME was only formed at reaction times greater than 5 h. Rather, the reaction in eq 5 (TG + H2O T DG + FFA) occurred (see Figure 3). Some evidence to support the occurrence of eq 5 is the fact that the TG conversion was identical to the DG yield during the first 5 h, as can be seen in Figure 3. In principle, experimental FFA yield data should provide additional insight. In the experiments reported here, our HPLC analyses for FFA (retention time ) 28.1 - 28.4 min) were completely normal in the absence of FAME (retention time ) 28.6 - 28.9 min). However, after 5 h of the reaction, the FAME concentrations were much greater than the FFA concentrations. As a result, the FFA peak was not detected when a much larger FAME peak was present. The FFA analyses during the first 5 h of the reaction time showed a net FFA conversion. As a result, the

2456 Energy & Fuels, Vol. 21, No. 4, 2007

Figure 4. Ln(1/(1 - XTG)) versus time, with 3 wt % H2SO4, where XTG is the TG fractional conversion. See the caption in Figure 1.

FFA analyses of the experiments reported here did not provide any insight about the validity of eq 5. The DG transesterification reaction in the polar MeOH phase is limited by the number of DG molecules in that phase. DG molecules at the MeOH-TG interface need to be protonated prior to reaction with MeOH. Experimental analysis of the nonpolar TG phase showed that the FAME yield increased simultaneously with the increase in conversion of DG or the decrease in the yield of DG (see Figures 2 and 3). Moreover, the FAME yield (see Figure 3) showed that there was an apparent induction period in the production of FAME. The results also show that the decrease in the DG yield was followed by a net increase in the yield of DG. An additional experiment was performed to provide insight on the time required for the FFA formation reaction, eq 6, to come to equilibrium. To circumvent the problem of the FFA peak being lost in the FAME shoulder, as described above, 5 wt % FFA was added to the reaction mixture used in the experiments reported here. The FFA conversions from those experiments are shown in Figure 5. The FFA conversion increased sharply with time and reached its equilibrium conversion after 5 h. This observation is the basis for our suggestion that FFA equilibrium preceded DG conversion. Specifically, Figure 3 shows that the FAME yield and a net DG yield both began at reaction times in excess of 5 h. The TG induction period, followed by the sudden increase in TG conversion, is a phenomenon that we have observed in several experiments (see Figures 1 and 3). The transesterification reaction between the protonated TG molecules and MeOH occurs at the MeOH-TG interface. The protonation of the TG molecules in the nonpolar TG phase, as a result of the FFAH3O+ micelle-TG collisions during agitation in the TG induction period, would increase the number of protonated TG molecules in that phase. However, the contact between the protonated TG molecules and MeOH at the MeOH-TG interface is hindered because of the presence of surface-active species at the interface. During the inhibition period, the FFA equilibrium, both reaction and phase, is achieved (see Figure 5) and some of the DG molecules at the interface react (see Figure 2). This would increase the number of contact sites for the MeOH and protonated TG species at the interface, and the reaction would occur. The increased number of activated TG molecules formed during the TG induction period resulted in the sudden increase in TG conversion (see Figures 1 and 3).

Ataya et al.

Figure 5. FFA fractional conversion versus the reaction time (in hours). Reaction conditions were identical to those of the agitated two-phase experiments in Figure 1, except that 5 wt % FFA was added to the reaction mixture.

For a first-order reaction, a plot of ln(1/(1 - X)) as a function of time will be linear, with a slope equal to the reaction rate constant, k. Figure 4 presents such a plot for the acid-catalyzed biodiesel reaction with a 3 wt % H2SO4 catalyst. Kinetic reaction rate constants were determined for the following irreversible single-phase reactions in the presence of THF: k1

TG 98

k2

k3

DG 98 MG 98 GLY + FAME + FAME + FAME

(7)

The experimental conversion/yield data were represented satisfactorily using statistical regressions of the kinetic equations (eqs 12-15) that were based on the irreversible reactions shown as eq 7. As a result, we did not introduce the additional parameters (rate constants) that would have been necessary to include the possibility of reversible reactions. Nevertheless, it was recognized that, at larger conversions than those obtained in this study, other workers, such as Vicente et al.,30 have invoked a combination of both reversible and irreversible reactions. The expressions for the reaction rates that were integrated with their respective initial conditions are

rTG ) rDG )

dTG ) -k1CTG dt

t ) 0 CTG ) CTG0

dDG ) k1CTG - k2CDG dt

rMG )

dMG ) k2CDG - k3CMG dt rGLY ) k3CMG

(8)

t ) 0 CDG ) CDG0

(9)

t ) 0 CMG ) 0

(10)

t ) 0 CMG ) 0

(11)

The following expressions for the conversion of TG, XTG, and the yields of DG and MG, YDG and YMG, were derived using the method employed by Levenspiel.31 The resulting equations (eqs 12-15) are similar to those developed by Levenspiel.31 The difference is that a nonzero initial concentration of DG was included in the development of eqs 12-15. As a result, the (30) Vicente, G.; Martinez, M.; Aracil, J.; Esteban, A. Kinetics of sunflower oil methanolysis. Ind. Eng. Chem. Res. 2005, 44, 5447. (31) Levenspiel, O. Chemical Reaction Engineering, 3rd ed.; Wiley: New York, 1999; pp 42-55.

Acid-Catalyzed Transesterification to Biodiesel

Energy & Fuels, Vol. 21, No. 4, 2007 2457

Table 2. Rate Constants (s-1) 5.55 × 10-6 8.35 × 10-6 1.80 × 10-6

k1 k2 k3

equations used here include an additional factor, containing the ratio of the initial feedstock DG/TG molar ratio

ln

YDG )

(

YMG )

)

1 ) k1t 1 - XTG

CDG ) CTG0

(() ( ) )

1

1-

(

k2 k1

(12)

)

k2 CDG0 1+ 1 exp(-k2t) - exp(-k1t) (13) CTG0 k1

CMG ) CTG0

( )( 1

k2 1k1

1 - exp(-k1t) k1 1k3

( ( ) )

the TG molecules.13 The methoxy ion concentration is related both to the ionization of methanol,32

)

k2 CDG0 (1 - exp(-k2t)) 1+1 (14) k2 CTG0 k1 1k3

YGLY )

CGLY ) CTG0

( )( 1

k1t - 1 + exp(-k1t) k1 k1 1k3 k2

( ) ( ( ) ) ( ) 1-

k2 k1

Figure 6. Agitated single-phase reaction. Conversion of TG (XTG) and yields of DG (YDG), MG (YMG), and FAME (YFAME) versus time at 3 wt % H2SO4.

CH3OH ) H+ + CH3OKMeOH ) 3 × 10-16 )

[H+ ][CH3O-] [CH3OH]

(16)

and to the ionization of water

)

k2 CDG (k2t - 1 + exp(-k2t)) 1+1 (15) k2 k2 CTG0 k1 1k3 k3

When regressions were performed, the rate constants shown in Table 2 were obtained for the single-phase reaction. Specifically, k1 was obtained using the TG conversion data in Figure 4 and eq 12; k2 was obtained using the DG yield data in Figure 6 and eq 13; and k3 was obtained using the MG yield data in Figure 6 and eq 14. Finally, the values for k1, k2, and k3 were used in eq 15 to predict the GLY yield. There were no adjustable parameters in eq 15. Nonetheless, the experimental GLY values and the predictions from eq 15 are in good agreement, as shown in Figure 6. This agreement supports our suggestion that these rate constants have physical meaning and are not just correlation parameters. The experimental data and the regression equation, eq 12, for TG conversion are compared in Figure 4 for the singlephase reaction. The experimental DG and MG yield data for the single-phase reactions are compared with their respective regression equations, in Figure 6. As indicated above, the use of irreversible reactions provided satisfactory agreement. The single-phase reaction with an acid catalyst and with a basic catalyst can also be compared. A similar analysis of reaction rates was applied to our data13 for the single-phase reaction with a basic (NaOH) catalyst. It resulted in rate constants that were 3 orders of magnitude larger than those in Table 2. In both cases, the contact between TG and MeOH would be equally intimate, because in both cases, the reaction occurred in a single phase. Therefore, the only difference could be the difference in the two catalysts. The reaction that occurs with the basic catalyst is considered to be between the methoxy ions (a reaction intermediate) and

H2O ) H+ + OH- KH2O ) 1 × 10-14 ) [H+][OH-]

(17)

Equations 16 and 17 can be combined to obtain

[CH3O-] [OH-]

)

3 × 10-16[CH3OH] 1 × 10-14

) constant

(18)

This indicates that, for a constant methanol concentration, the concentration of methoxy ions will be a constant that is approximately 2 orders of magnitude less than the concentration of the OH- ions that are provided by the basic catalyst. The transesterification reaction under acid-catalyzed conditions occurs between methanol and a protonated TG molecule. According to the conventional mechanism,29 a proton is added to one of the carbonyl oxygen atoms in the TG molecule to form a TG cation. The concentration of protonated TG cations is expected to be similar to that of carbenium ions, which have concentrations33 that are generally less than the methoxy ions calculated above. Moreover, the fact that the protonated TG molecules are more bulky than their methoxy ion counterparts would result in a difference in the rates of diffusion of these reaction intermediates from the FFA-H3O+ micelle interface through the nonpolar TG phase to the polar MeOH phase to the interface, of course, in favor of the methoxy ion species. The concentration of the protonated TG reaction intermediates formed from the catalysts may also contribute to the transesterification reaction with acid catalysts being slower than the transesterification reaction with basic catalysts. TG conversion during the two-phase reaction was substantially less than during the single-phase reaction, as shown in Figure 1. A difference in the mass-transfer rate is one plausible (32) Ballinger, P.; Long, F. A. Acid ionization constants of alcohols: Acidities of some substituted methanols and related compounds. J. Am. Chem. Soc. 1960, 81, 795. (33) Bywater, S.; Worsfold, D. J. The reaction of diphenylethylenes with boron trifluoride and water. Can. J. Chem. 1977, 55, 85.

2458 Energy & Fuels, Vol. 21, No. 4, 2007

Ataya et al.

explanation for the difference in conversions. For the agitated single-phase reaction, the line had a steeper slope (see Figure 4). The steeper slope in the single-phase reaction was attributed to a kinetically controlled regime. The smaller value of the slope in the two-phase reaction suggests that mass transfer influenced the rate of reaction for the two-phase agitated reactions (see Figure 4). The equations for the mass transfer of TG and the flow rate of TG from the bulk of the nonpolar TG phase to the interface are

NTG ) kL(CTG-B - CTG-S)

(19)

FTG ) NTGaVRX ) kL(CTG-B - CTG-S)aVRX

(20)

and the equations for the reaction and the flow rate of TG at the interface are

-rTG ) kRXCTG-S

(21)

FTG ) -rTGVRX ) kRXCTG-SVRX

(22)

Equations 20 and 22 were combined as follows:

CTG-B ) (CTG-B - CTG-S)eq 20 + (CTG-S)eq 22 ) FTG FTG 1 FTG 1 + ) (23) + kLaVRX kRXVRX VRX kLa kRX

(

to obtain

FTG ) -rTG ) VRX

)

( )

1 CTG-B ) kOVCTG-B (24) 1 1 + kLa kRX

Although the nomenclature is slightly different, eq 19 is the conventional form of the equation that is used to describe interfacial mass transport.34 The value of the overall rate constant, kOV, was obtained from the slope of the bold dashed line in Figure 4. This line was based on the expression: XTG ) 1 - exp(-kOVt). Then, when the value of kRX ) k1 ) 5.55 × 10-6 s-1, the rate constant for the single-phase reaction, the value of kLa was determined to be kLa ) 7.28 × 10-7 s-1. A comparison of the kLa value with the kRX value suggests that the mass-transfer phenomenon is an order of magnitude slower than the reaction kinetic phenomenon. It is appropriate to compare (a) the straight dot-dashed bold line (in Figure 4) for two-phase TG conversion obtained from eq 24 that included one mass-transfer process and one kinetic rate process and (b) the nonlinear pointed line that represents the experimental data for two-phase TG conversion. Because the linear equation, eq 24, does not match the nonlinear experimental data, it is possible that the phenomena occurring are more complex than a simple combination of one mass-transfer process and one kinetic rate process. Diffusion during two-phase TG conversion using an acid catalyst can be compared with that using a base catalyst. There are three phenomena that are pertinent. (1) We have already noted above that during single-phase TG conversion the rate constants with an acid catalyst are 3 orders of magnitude slower than that with a base catalyst. (2) Similarly, the rate of twophase acid-catalyzed TG conversion is also much slower than that of two-phase base-catalyzed TG conversion. (3) Both acidand base-catalyzed two-phase TG conversion are controlled by (34) Bird, R. B.; Stewart, W. E.; Lightfoot, E. N. Transport Phenomena, 2nd ed.; Wiley: New York, 2002; p 545.

the reaction at the interface, which is a diffusion-influenced phenomenon. If the same diffusion process controlled both acid and base two-phase TG conversion, then both two-phase processes would be expected to proceed at the same rate. An explanation can be suggested for both acid- and base-catalyzed reactions proceeding at the interface but at different rates. The diffusing reaction intermediates are protonated TG molecules in the case of acid-catalyzed two-phase TG conversion and methoxy ions in the case of base-catalyzed two-phase TG conversion. Because the protonated TG molecules are much larger than the methoxy species, they will diffuse very slowly compared to the smaller methoxy species. The result is that acid and base catalysts produce different reaction intermediates that diffuse at different rates. The compositions of the transesterification products as a function of time are shown in Figure 3 for the two-phase reaction and in Figure 6 for the single-phase reaction. The fraction of unreacted TG in the product is the difference between 1.0 and the fractional TG conversion shown in the figures. The fractional yields of DG and MG are expressed as the amount of product formed per the amount of feedstock fed. The fractional yield of FAME was calculated using FAME produced from the conversion of TG, DG, or MG (see eqs 2-4) and the conversion of FFA to FAME (see eq 6). The conversions and yields from the single-phase reaction mixtures were greater than from the two-phase agitated reaction mixtures. The TG conversion and DG, MG, and FAME yields from the single-phase agitated reaction at 2.5 h (see Figure 6) are similar to those obtained for the agitated two-phase reactions at 24 h (see Figure 3). Progress through the various stepwise reactions (eqs 2-4) can be visualized from Figures 3 and 6. For the two-phase agitated reaction in Figure 3, approximately the same values were observed for TG conversion, DG yield, and FAME yield after a 24 h reaction time. This suggests that only the first step of the series reaction, eq 2, i.e., TG reacting to form DG and FAME, has occurred. For the single-phase reaction in Figure 6, it is seen that, after 24 h of reaction time, the yield of FAME exceeds the yield of DG. If the first step, eq 2, was the only reaction occurring, the FAME and DG yields should have been identical. Contributions to the additional FAME yields could have been obtained from the following reactions: any conversion of DG via eq 3, MG via eq 4, or FFA via eq 5 would have produced additional FAME. These findings suggest that the same reaction mechanism may be operating for both the singlephase and two-phase reactions for the acid-catalyzed transesterification. Conclusions Canola oil transesterification experiments were performed as both two-phase and single-phase reactions. For the two-phase reaction medium, under both acid- and base-catalyzed conditions, the reaction appears to occur at the interface of the two phases. This is consistent with our experimental data. For the single-phase medium where no mass-transfer resistance occurred, reaction kinetic rate constants were determined. The measured kinetic rate constants with an acidic catalyst (H2SO4) were found to be 3 orders of magnitude slower than those with a basic catalyst (NaOH). This can be attributed to the difference in concentration of the methoxy ions and the protonated TG species. The differences in the kinetic rate constants of the acid- and base-catalyzed processes can be attributed to the difference in concentrations of the respective ionic reaction intermediates. For the two-phase agitated reaction, a mass-transfer coefficient was calculated that was found to be 1 order of magnitude

Acid-Catalyzed Transesterification to Biodiesel

Energy & Fuels, Vol. 21, No. 4, 2007 2459

smaller than the kinetic rate constant for the single-phase reaction. For these cases, the differences in rates of TG conversion can be attributed to differences in diffusion caused by the size of the respective ionic reaction intermediates. Acknowledgment. The authors acknowledge the BIOCAP Canada Foundation and the Natural Sciences and Engineering Research Council (NSERC) of Canada for financial support of this research.

YMG )

NMG(product) NTG(feed) + NDG(feed)

(A4)

The FAME yield was calculated as

((NFAME(product) - (NDG(product) - NDG(feed)) - 2NMG(product) (NFFA(product) - NFFA(feed)))/3)/NTG(feed) + NDG(feed) (A5) Nomenclature

Appendix: Calculation Methodology The fractional conversion of TG to all products was calculated from

XTG ) 1 -

NTG(product) NTG(feed)

(A1)

XTG is the fractional conversion of TG; NTG(product) is the measured number of moles of TG in the product from the calibration curve; and NTG(feed) is the estimated original number of moles of TG in the feed, calculated from

NTG(feed) ) NTG(product) + (NDG(product) - NDG(feed)) + NMG(product) + ((NFAME(product) - (NDG(product) - NDG(feed)) 2NMG(product) - (NFFA(product) - NFFA(feed)))/3) (A2) NDG(feed) and NFFA(feed) are the estimated original number of moles of DG and FFA in the feed, respectively, evaluated from the original ratios of DG/TG and FFA/TG in the feed. NDG(product), NMG(product), NFAME(product), and NFFA(product) are the measured number of moles of DG, MG, FAME, and FFA in the product from the calibration curves, respectively. The yield of DG was calculated from

YDG )

NDG(product) - NDG(feed) NTG(feed) + NDG(feed)

The yield of MG was calculated from

(A3)

Ca(OH)2 ) Calcium hydroxide DG ) Diglyceride FAME ) Fatty acid methyl ester FFA ) Free fatty acid GLY ) Glycerol H2SO4 ) Sulfuric acid MeOH ) Methanol MG ) Monoglyceride NaOH ) Sodium hydroxide TG ) Triglyceride THF ) Tetrahydrofuran a ) Interfacial surface area per unit volume (m2/m3) Ci ) Concentration of species (i) (mol/m3) Ci0 ) Nonzero initial concentration of species (i) (mol/m3) CTG-B ) Bulk concentration of TG in the nonpolar TG phase (mol/ m3) CTG-S ) Bulk concentration of TG in equilibrium with TG at the TG-MeOH interface (mol/m3) FTG ) Flow rate of TG to the interface (mol/s) ki ) Reaction rate constant of species (i) (s-1) Ki ) Equilibrium constant of species (i) kL ) TG side mass-transfer coefficient (m/s) kOV ) Overall reaction rate constant for the two-phase reaction (s-1) kRX ) TG reaction rate constant for the single-phase reaction (s-1) NTG ) Mass transfer of TG to the interface (mol m-2 s-1) ri ) Reaction rate of species (i) (mol m-3 s-1) t ) time (h) VRX ) Reaction volume (m3) Xi ) Conversion of species (i) (mol/mol) Yi ) Yield of species (i) (mol/mol) EF0701440