Optimized microemulsion production of biodiesel over lipase-catalyzed transesterification of soybean oil by response surface methodology

2.1 Materials

Lipase from porcine pancreas was purchased from Sigma Aldrich (Shanghai, PR China). Soybean oil was purchased from Fulinmeng (Shanghai, PR China). Methanol (≥99.5%), isooctane, n-octane, cyclohexane and n-hexane were obtained from Qiangsheng (Changshu, Jiangsu Province, PR China). Dodecyl Benzene Sulphonic Acid (DBSA) were purchased from Zanyu (Hangzhou, Zhejiang Province, PR China). All the chemicals used in the study were of analytical grade and were used without further purification. Magnetic stirrer was obtained from Haimen Kylin-Bell Lab Instrument company (Haimeng, Jiangsu Province, PR China). Design Expert 8.0.6 (USA) was used.

2.2 Preparation of microemulsion and enzymatic catalysis

The microemulsion system was formulated by mixing isooctane (6.4 g) and DBSA (1.6 g) with water (w₀: 2.5–4.3) at room temperature in a 50 ml round-bottom flask for 5 min until it became optically transparent and homogeneous. Then, a proportion of methanol (0.003–0.015 mol), oil (0.001 mol) and lipase (0–5.5%) was also added to the flask. Subsequently, the mixture was stirred for a certain time. The yield of biodiesel was analyzed by gas chromatography (Agilent 6890N GC), equipped with a hydrogen flame ionization detector (FID) and HP-5 capillary column (30.0 m×320 nm×0.25 μm). Reactants and products were identified by comparison with authentic samples. Methyl laureate was used as an internal standard.

2.3 Experimental design and statistical analysis

RSM was employed to analyze the operating conditions of transesterification to obtain a high conversion ratio. The software program Stat-Ease Design Expert (Version 8.0.6, Stat-Ease Inc., USA) was used in the statistical experimental design. The independent factors (X_i), levels and experimental design in terms of actual and coded variables are shown in Table 1. A total of 29 samples with different compositions were prepared in random order according to the Box-Behnken design, which included 29 experiments of four variables at three levels (-1, 0, 1) to evaluate the effect of those factors on biodiesel yield (Table 2) [19, 20]. The design aimed at getting the optimal conditions of the reaction.

The experimental data were fitted to the quadratic polynomial model. The interaction between the variables was elucidated and the second-order polynomial equation for predicting the optimal point was given as follows:

where Y (%) is the response value of biodiesel yield, X_i represents independent factors, and β₀, β_i, β_ii,β_ij are intercept, linear, quadratic and interaction constant coefficients, respectively. The accuracy and general ability of the quadratic polynomial equation could be evaluated by the coefficient of determination (R²) and its regression coefficient significance was checked by the F test [21]. The connection between the response and experimental levels of each factor could be expressed visually as response surface curves and contour plots, by which the optimal point for each independent variable was deduced. The significance of the second-order model was evaluated by analysis of variance (ANOVA).

3.1 Effect of water content w₀

Various w₀ (defined as the molar ratio of water to surfactant) over a range from 2.5 to 4.3 were performed to prepare the microemulsion for catalyzing biodiesel production [22, 23]. As can be seen (Figure 1), w₀ had a great impact on the catalytic activity of the lipase corresponding to the biodiesel yield. The biodiesel yield increased gradually from 40.2% to 93.9% when the w₀ increased from 2.5 to 4.3. The result showed a maximum enzymatic activity occurred at w₀=3.5. Subsequently, enzymatic activity decreased with increase of water. The reason was that reverse micelles with relatively low water content could not provide sufficient space to completely encapsulate enzyme molecules in the water microdroplets, presumably forcing enzymes into a less active conformation [14]. With the increase of w₀, the amount of water available for oil to form oil-water droplets increased, thereby increasing the available interfacial area and activating lipases which were beneficial for the biodiesel production.

Figure 1

Effect of w₀ (molar ratio of water to surfactant) on biodiesel yield [reaction conditions: lipase 4.4%, molar ratio (methanol/oil) 9, time 16 h, temperature 40°C, stirrer speed 250 rpm].

However, when the w₀ was above 3.5, coalescence of the emulsion water droplets would occur decreasing the interfacial area of the organic and aqueous layers. This had an inhibitory effect on the enzyme activity and disturbed the esterification by hydrolysis [24].

3.2 Effect of lipase concentration

The concentration of enzyme played a vital role in the transesterification reaction. The reaction rate increased with increasing concentration of enzyme. However, superfluous lipase might also have a negative effect and its industrial production cost is high. Thus, the optimum concentration of lipase was studied and results are listed in Figure 2. From Figure 2, it is seen that the biodiesel yield increased with increasing lipase concentration. The maximal point (93.9%) was obtained at a lipase concentration of 4.4%. When the lipase concentration was >4.4%, the biodiesel yield decreased slightly. This may be due to the fact that excess lipase would unite and the active site of the lipase cannot be exposed to the substrates, which hindered the full contact of individual lipase macromolecules with reactants [19, 25, 26].

Figure 2

Effect of lipase concentration on biodiesel yield [reaction conditions: w₀ (molar ratio of water to surfactant) 3.5, molar ratio (methanol/oil) 9, time 16 h, temperature 40°C, stirrer speed 250 rpm].

3.3 Effect of molar ratio

In order to research the optimal molar ratio of the alcohol to oil, ratios from 3:1 to 15:1 were tested. Transesterification was a reversible reaction; increase of methanol drove the transesterification equilibrium towards the formation of biodiesel and improved the yield of biodiesel. When alcohol was in excess, the conversion decreased instead. This is observed in Figure 3, where an increase of biodiesel yield is noticed before the yield subsequently decreased for this reaction. A maximum of 93.9% was acquired at a molar ratio of 9:1. Therefore, with increase in methanol, methyl ester first increased and then decreased as a result of the decrease of enzyme activity caused by excessive methanol [27, 28]. Furthermore, the increased methanol diluted the active centers in the entire volume of the solution, which may also have an adverse effect on the biodiesel yield [29].

Figure 3

Effect of molar ratio on biodiesel yield [reaction conditions: w₀ (molar ratio of water to surfactant) 3.5, lipase 4.4%, time 16 h, temperature 40°C, stirrer speed 250 rpm].

3.4 Effect of time

The influence of reaction time in the range of 8–24 h was examined, in addition to comparing DBSA microemulsion with and without lipase as a catalyst. Reactions were carried out and the yield of biodiesel increased to the maximal value at 16 h (Figure 4). Then, the biodiesel yield decreased slightly as the reaction time was prolonged. This is due to the water in the reaction system inducing the hydrolyzation of biodiesel [28, 30]. Therefore, 16 h was a correct parameter for the reaction.

Figure 4

Effect of reaction time on biodiesel yield [reaction conditions: w₀ (molar ratio of water to surfactant) 3.5, lipase 4.4%, molar ratio (methanol/oil) 9, temperature 40°C, stirrer speed 250 rpm].

3.5 Study on different reaction system

In general, DBSA microemulsion systems contain surfactant, water and organic solvent mixtures and they can be divided into several systems based on different organic solvents, such as isooctane, n-octane, cyclohexane and n-hexane. It was found that a DBSA/isooctane/water microemulsion system achieved the best result. The biodiesel yields in different reaction systems are presented in Figure 5; isooctane showed a good yield. A possible explanation for this result was the difference in the molecular structure of the organic media used. N-octane and n-hexane are straight, short-chain alkanes, which can embed in the interfacial septum formed with DBSA molecules. The hydrocarbons can then form an additional layer in the interfacial membrane [31]. Penetration of the mostly saturated hydrocarbons into the surfactant layer of the microemulsion impedes the contact and/or interaction between lipase and its substrates resulting in a smaller product yield [32]. In addition, the log p value of the solvent also has a little effect on the yield, as the log p value of the solvent increases the extent of esterification yield [25]. Cyclohexane has a unique ring structure that does not have a proper structure to penetrate the DBSA interfacial membrane. Thus, lipase activities in these media were low compared with the lipase activity in isooctane. The biocatalysis in microemulsion had an organic-solvent dependency [31]. Therefore, isooctane was the preferable organic phase in the DBSA microemulsion system in further research.

Figure 5

Effect of different organic solvents on biodiesel yield [reaction conditions: w₀ (molar ratio of water to surfactant) 3.5, lipase 4.4%, molar ratio (methanol/oil) 9, time 16 h, temperature 40°C, stirrer speed 250 rpm].

3.6 Optimization of the reaction conditions

Based on the results of single factor experiments, four process variables were used to develop the experimental design. The experimental design and results are presented in Tables 1 and 2. As can be seen, the biodiesel yield ranged from 42.4% to 96.7% and the design points of number 8 and number 29 gave the minimum and maximum yields, respectively.

The experimental data were well qualified for the model equation, which could be expressed as follows:

where Y is the biodiesel yield and X₁, X₂, X₃ and X₄ are the coded values of the test variables w₀, lipase concentration, reaction time and substrate mole ratio, respectively; Values of Prob>F<0.05 indicate model terms are significant. In this case X₁, X₃, X₄, X₁X₂, X₂X₄, X₃X₄, X12,X22,X32,X42 are significant model terms to affect the biodiesel yield (Table 3).

The standard analysis of ANOVA results (Table 3) indicated that the model F value of 34.38 and a low probability p (<0.0001) implied that the model was significantly suitable. The lack of lit F value of 2.38 demonstrated that the lack of fit was not significant relative to the pure error. The coefficient of determination (R²) of the model was 0.9570, which indicated a good accuracy and a general fitness of the polynomial model. Figure 6 shows that the responses predicted from the model were in agreement with the observed values in the range of the operating variables. The adjusted R² value (0.9291) also indicated the model’s goodness of fit. The predicted R² of 0.8530 was in reasonable agreement with the adjusted R² of 0.9291. Thus, this model can be used to navigate the design space. Regression coefficients of a predicted quadratic polynomial model are shown in Table 3.

Figure 6

The predicted biodiesel yield versus actual yield.

The response surface three-dimensional (3D) and contour plots described by the regression model were drawn to display the effects of the independent variables on biodiesel yield (Figure 7). The influence of the different variables in the conversion of the systems can be clearly seen.

Figure 7

Response surface three-dimensional (3D) and contour plot for biodiesel yield [A and D, which present the interaction of w₀ (molar ratio of water to surfactant) with lipase concentration, respectively; B and E, which present the interaction of lipase concentration with molar ratio, respectively; C and F, which present the interaction of reaction time with molar ratio, respectively].

Figure 7A and D show the effect of w₀, lipase concentration and their mutual interaction on the biodiesel yield at fixed values of w₀ ratio and lipase concentration in their center values. At the lowest w₀ (3.2) with the lowest lipase concentration (3.3%), the biodiesel yield was only 66.4%. When the w₀ increased to 3.5, a maximum conversion of >89.4% in biodiesel yield was observed with the increase of lipase concentration (4.4%). Then, the trend reversed when w₀ and lipase concentration surpassed values of 3.8 and 5.5%, respectively. It could be interpreted that, under a certain value, the w₀ and lipase amount can speed the reaction rate of transesterification together. In addition, the trend reversed when the w₀ and lipase amount exceeded that value, because a higher w₀ decreased the stability of the lipase, and superfluous lipase [>5.5% (w/w)] would unite which resulted in the catalytic activity reduction. The contour plot and 3D-plot of Figure 7A and D present a slantwise elliptical pattern. This indicated that the biodiesel yield was greatly affected by w₀ (X₁) and lipase concentration (X₂) and was in good agreement with the analysis outcome of joint test for different variables utilized in this study (see Table 3).

The effect of varying the lipase amount and substrate molar ratio on the biodiesel yield is presented in Figure 7B and E. The interactive effect of lipase amount and substrate molar ratio was also highly significant. The slantwise elliptical pattern represented that the biodiesel yield increased at first, and was followed by a slight decline with increase of lipase loading. The result indicated that the number of active sites increased with increase in the lipase amount and drove more substrate molecules towards products [27]. The biodiesel yield increased gradually with increasing methanol concentration, but excess methanol caused enzyme activity loss, which decreased the yield. When the ratio of methanol to oil was 9:1 and lipase concentration was 5.5%, the production yield reached a peak (94.1%). Excess enzyme can compensate for the activated center inhibited by excess methanol. Therefore, there is a slight decline of the yield with excess methanol and lipase. A high conversion rate could be obtained at the optimum ratio of substrate for the reaction of oil and methanol. The effect of the molar ratio of substrates on the conversion yield could be attributed to the thermodynamic shift of the equilibrium to the synthesis of fatty acid methyl ester due to the methanol excess under a certain value.

The interaction of reaction time and substrate molar ratio on methyl ester synthesis is presented in Figure 7C and F. The reaction time and molar ratio were fixed at their center points. The slantwise elliptical pattern represented that the conversion of biodiesel increased with increasing molar ratio and reaction time, which reflected a general effect of ascending reaction time on the reaction. Subsequently, the biodiesel yield emerged a peak with a maximum value and then the biodiesel yield declined. Excess methanol can damage the activated center of lipase in the reaction, but it can dissolve in oil well under a certain amount (molar ratio of methanol to palmitate acid=9:1) and accelerate the reaction rate. With a reaction time beyond 16 h, the reaction equilibrium was reached and the reaction thus became unfavorable. From the analysis of the response surface (Figure 7C and F), reaction time, substrate molar ratio was extremely significant.

3.7 Validation of the model

According to the reaction result, the regression model showed a perfect fitness for the biodiesel yield. The optimal reaction parameters evaluated from the regression model [Eq. (2)] were as follows: 3.6 w₀, 4.3% lipase concentration, 10.5:1 molar ratio (methanol/oil) and 17 h reaction time. Three parallel experiments were conducted under the optimal conditions. The average methyl ester yield was 96.8%, which was consistent with the predicted value (97.5%). Thus, the regression model was considered to be effective and accurate in predicting the biodiesel yield.