Fuel 203 (2017) 739–748

Contents lists available at ScienceDirect

Fuel
journal homepage: www.elsevier.com/locate/fuel

Full Length Article

Process optimization, kinetic and thermodynamic studies on biodiesel

production by supercritical methanol transesterification with CH3ONa
catalyst
Dan Zeng a, Liu Yang b, Tao Fang c,⇑
a
Department of Chemical Engineering and Applied Chemistry, University of Toronto, 200 College Street, Toronto, Ontario M5S 3E5, Canada
b
School of Energy and Power Engineering, Xi’an Jiaotong University, Shaanxi 710049, China
c
School of Chemical Engineering and Technology, Xi’an Jiaotong University, Shaanxi 710049, China

a r t i c l e i n f o a b s t r a c t

Article history: This study provided a novel supercritical methanol transesterification process with little amount of
Received 9 February 2017 CH3ONa catalyst to synthesize biodiesel from soybean oil by response surface technology. The maximum
Received in revised form 12 April 2017 biodiesel yield was 97.42% under the optimal conditions of 250 °C, methanol to oil molar ratio of 23:1,
Accepted 5 May 2017
1.0 wt% CH3ONa and 20 min. Temperature was proved to have the most significant effect on transester-
Available online 15 May 2017
ification, followed by methanol to oil molar ratio and catalyst amount. The kinetic model suggested a
1.5th order reaction with the activation energy of 27.06 kJ
mol1 and the pre-exponential factor of
Keywords:
102.71. The values of thermodynamic parameters including enthalpy, entropy and Gibbs free energy
Biodiesel
Transesterification
for this process were calculated as 23.15 kJ
mol1, 0.22 kJ
mol1
K1 and 137.43 kJ
mol1 at 250 °C,
Supercritical methanol respectively. Reduced reaction temperature, catalyst amount, and reaction time are among the advan-
Kinetics tages of the integrated process for biodiesel production over single supercritical process and conventional
Thermodynamics catalytic methods.
Ó 2017 Published by Elsevier Ltd.

1. Introduction duction of biodiesel [11,12]. Conversely, alkali-catalyzed transes-

terification has been widely applied in industry as they can
As the majority of global energy, the fossil sources are limited catalyze the reaction at low temperature and atmospheric pres-
and assumed to be exhausted by the explosive increment of energy sure, which leads to high conversion within a specified time [13].
demand. Biodiesel has gained interest around the world over the Rashid et al. [14] synthesized biodiesel by base-catalyzed transes-
last few decades because of its environmental benefits. As an alter- terification of rice bran oil and the highest biodiesel yield of 83.3%
natively renewable fuel, biodiesel is generally produced by transes- was achieved at the optimized conditions of 7.5:1 methanol to oil
terification of vegetable oils or animal fats. It is renewable, molar ratio, 0.88% catalyst concentration, 55 °C and 60 min. Amini-
biodegradable, non-toxic, sulfur-free, and can be used to reduce Niaki et al. [15] obtained the sunflower biodiesel similar yield of
most regulated exhaust emissions [1,2]. 83.4% at the conditions of molar ratio of oil to methanol at 1:5.5,
Four main approaches to synthesize biodiesel are direct use and 1.0% percentage of KOH catalyst at 51.7 °C in 65.5 min. Vicente
blending of vegetable oils, micro-emulsions, thermal cracking et al. compared four primary alkali catalysts (CH3ONa, CH3OK,
(pyrolysis) and transesterification [3]. Among them, transesterifi- NaOH, KOH) for methanolysis of vegetable oil. It demonstrated that
cation reaction of renewable sources is the most common method near 100% biodiesel yields were only obtained with methoxide cat-
for biodiesel production catalyzed by acids, alkalis, and lipases [4– alysts because the methoxide anion is strongly basic and has high
10]. Methanol is the most frequently used alcohol in transesterifi- catalytic activity in transesterification [16,17]. Among them, CH3-
cation because of its low cost and its physical and chemical advan- ONa is the most commonly used catalyst for chemical transesteri-
tages as the shortest chain alcohol. The relatively slow reaction fication [18]. Borugadda et al. [19] summarized biodiesel
rate of acid-catalyzed transesterification and the high cost of production from various feedstock under different conditions using
lipases limit these two kinds of methods in the commercial pro- homogeneous alkali catalysts. They found that the catalyst amount
is usually over 1.0%, even 6.0% and the reaction time takes several
⇑ Corresponding author. hours for conventional catalytic transesterification methods. More-
E-mail address: taofang@mail.xjtu.edu.cn (T. Fang). over, when the raw material has a high percentage of free fatty

http://dx.doi.org/10.1016/j.fuel.2017.05.019
0016-2361/Ó 2017 Published by Elsevier Ltd.
740 D. Zeng et al. / Fuel 203 (2017) 739–748

acids (FFA) or water, the alkali catalyst will be consumed by unde-

sired saponification reactions, and the process needs complicated
downstream separation and purification processes [13].
To overcome the limitations of conventional catalytic processes,
Saka and Kusdiana proposed a new supercritical methanol (SC-
MeOH) process for biodiesel production, which requires shorter
reaction time and simpler purification procedure. He et al. [20]
studied biodiesel synthesis via supercritical methanol transesterifi-
cation from soybean oil and found that the biodiesel yield was
more than 96% at 310 °C, methanol to oil weight ratio of 40:1 for
25 min with the gradual heating reaction process. Román-
Figueroa et al. [21] investigated a non-catalytic transesterification
of crude castor oil into fatty acid methyl esters (FAME) and
obtained 96.5% of biodiesel yield at the optimized conditions of
methanol to oil molar ratio of 43:1within 90 min at 300 °C. 1- high temperature furnace; 2-oscillator; 3-temperature controller;
García-Martínez et al. [22] presented the biodiesel produced from 4-batch reaction vessel
tobacco seed oil by supercritical methanolysis and achieved the
Fig. 1. Schematic diagram of the experimental reactor.
maximum yield of 91.1% at 303.4 °C and 90 min with a molar ratio
of methanol to oil 43:1. In summary, the biodiesel yield enhanced
prominently with the usage of supercritical technology; however, the following analysis. The yield data were obtained by calculating
optimum reaction conditions were found to be at higher tempera- the average yield of three reactors with the deviation lower than
ture (>300 °C) and longer reaction time (25–90 min). These draw- 5%.
backs lead to expensive equipment cost and high energy
consumption hindering the application of supercritical technology 2.2. Property of the raw material
in the biodiesel industry.
This research presents a novel integrated process for supercrit- The saponification value (SV) and the acid value (AV) of soybean
ical methanol and little amount of CH3ONa catalyst to produce bio- oil were measured according to the methods specified by American
diesel from soybean oil. The proposed method combines the Oil Chemists’ Society (AOCS) in our previous study [23]. They were
advantages of supercritical technology and base-catalyzed transes- 193.5 mg KOHg1 oil and 6.4 mg KOH
g1 oil, respectively. Thus,
terification method. It has minimal undesired reactions with a rel- the average molecular weight (Mr) was 841.8, calculated by Eq. (1).
atively low reaction temperature in contrast to supercritical
technology; and requires less catalyst amount, much shorter reac- 56:1  1000  3
Mr ¼ ð1Þ
tion time, and without the saponification reactions compared with ðSV  AVÞ
the conventional base-catalyzed transesterification process. Pro-
cess parameters including reaction temperature, molar ratio of
methanol to soybean oil, and catalyst amount (mass percent of 2.3. FAME analysis
CH3ONa to soybean oil) were discussed. Besides, the kinetic and
thermodynamic study of the integrated process were also The contents of FAME in the samples diluted by n-hexane were
conducted. analyzed by the gas chromatography-mass spectrometry (GC–MS)
5973-6890 (Agilent Technologies Inc., Palo Alto, California, USA)
equipped with an HP-5 capillary column (30 m  £
2. Experimental section 0.25 mm  0.25 lm). The analysis conditions were as follows: the
detector temperature of 300 °C, the injector temperature of
2.1. Materials and methods 250 °C, a carrier gas of helium with a flow rate of 1.0 mL/min, a
split ratio of 30 and the sampling volume of 1 lL. The oven temper-
Soybean oil was purchased from Xi’an Jiali Oil Industry Co., Ltd. ature was initially held at 200 °C for 10 min and then elevated to
(Xi’an, Shaanxi Province, China). Methanol (analytical grade purity) 230 °C at a ramp rate of 5 °C/min, finally held at 230 °C for 2 min.
was obtained from Tianjin Fuyu Fine Chemical Co., Ltd., CH3ONa The external standard was methyl oleate. The biodiesel yield (Y)
(analytical grade purity) from Tianjin Fuchen Chemical Reagents was determined by Eq. (2).
Factory, and methyl oleate (CAS number: 112-62-9) from Yake C S  AFAME
Chemical Reagent Co., Ltd. (Suzhou) with purities of 99.0%. The Y¼ ð2Þ
AS  C B
self-designed experiment reactor (Huaxing Petroleum Instrument
Co., Ltd., Nantong, Jiangsu Province, China) for biodiesel production where CS is the concentration of external standard (g/g), AFAME is the
is presented in Fig. 1. Briefly, it consists of a batch reaction vessel total peak area of FAME, AS is the peak area of external standard and
(with an inner volume of  10 ml, and the maximum pressure CB is the concentration of the biodiesel product (g/g).
and temperature of 50 MPa and 500 °C, respectively) and a heating
furnace. 2.4. Experimental design
Firstly, soybean oil and CH3ONa-dissolved methanol solution in
the specified proportions (the total amount was 6 mL) were Response surface methodology (RSM), which is originated by
charged into three batch reaction vessels and then the vessels were Box and Wilson in 1951 [24], has received noticeable attention in
placed in the high-temperature furnace that has been preheated to recent decades as an important branch of experimental design. It
the set temperature. After the transesterification with oscillating is an efficient way to find the relationship between the reaction
the reactor at the rate of 50–60 rpm, the vessels were taken out variables and the response variables within the region of interest
of the furnace and quenched by cold water for 5 min. Finally, the [25]. Many studies [14,26–28] have proved that the RSM is a suit-
product of each reactor was collected, and the oil phase was com- able technique for optimizing transesterification of biodiesel
pletely separated by washing with hot water (80 °C) to prepare for production.
 D. Zeng et al. / Fuel 203 (2017) 739–748 741

Three variables of transesterification including reaction temper- (46.636%) gave a high cetane number and decreased the viscosity
ature (150–250 °C), methanol to oil molar ratio (23:1–69:1) and of biodiesel [30].
CH3ONa amount (0.5–1.5 wt%) were employed to design the exper-
iments by RSM coupled with Box-Behnken design (BBD). As this 3.1.2. Development of mathematical model
model uses points outside the design space (1.00, 1.00 levels) Table 3 shows the ANOVA results to explore the model fitness.
which provides a reliable estimation over the entire design space, The F-value of 174.84 indicates the model is significant, and the P-
it is more efficient and economical than the corresponding full value is less than 0.0001, implying that only a 0.01% chance that a
three-level factorial design [29]. The central point was performed model with such large F-value could occur due to noise. For indi-
three times to consider reproducibility. The results of the experi- vidual terms, only those with Prob. > F less than 0.05 are deemed
mental design matrix and the biodiesel yield (response value) are as significant. Besides, the ‘‘Lack-of-fit F-value” of 3.23 implying
shown in Table 1, which were performed in a random order and the Lack of Fit is not significant relative to pure error. Therefore,
in triplicate to achieve repeatable results. The reaction time was the proposed model is valid and statistically significant. Subse-
fixed at 20 min in the optimization experiments. quently, the final equation regarding coded factors after eliminat-
ing the insignificant parameters is given in Eq. (4).
2.5. Statistical analysis
X ¼ 0:95 þ 0:13A  0:047AB  0:062BC  0:11A2  0:029B2
A quadratic polynomial model was developed to predict the  0:069C 2 ð4Þ
biodiesel yield as a function of the independent variables. The gen-
2
eral form of this second-order polynomial can be expressed as: A correlation coefficient R = 0.9968 which was found to be very
close to unity shows the excellent correlation between the exper-
X
3 X
3 X
2 X
3
imental data and the predicted data. It reveals that the regression
X ¼ b0 þ b i xi þ bii x2i þ bij xi xj ð3Þ
model can successfully predict the biodiesel yield by transesterifi-
i¼1 i¼1 i¼1 j¼iþ1
cation. The predicted R2 of 0.9568 was in reasonable agreement
where X is the predicted response; xi and xj are the experimental with the adjusted R2 of 0.9911 in the refined model. The signal
factor levels; b0 is the offset term; bi, bii and bij are the linear, quad- to noise ratio is measured with sufficient precision, and a ratio
ratic and interactive coefficients, respectively. A complete analysis greater than four is desirable. In this case, the ratio of 36.935 indi-
of variance (ANOVA) was performed by Design Expert Software Ver- cated an adequate signal. Thus, the model can be used to navigate
sion 8.0.6 (Stat-Ease, Inc. 2021 East Hennepin Ave., Suite 480 Min- the design space.
neapolis, MN 55413) at a significance level of 95%. Subsequently,
the quality of the proposed model was evaluated by the correlation 3.2. Effect of experimental factors
coefficient (R2) while indicating the consideration of a complete
quadratic model and eliminating terms that were not significant The surface plots of the biodiesel yield versus interactions of
as the analysis continued. two variables are presented in Figs. 3–5. In each graph, the remain-
ing variables are set constant at the center point.
3. Results and discussion
3.2.1. Effect of temperature
3.1. Process optimization Fig. 3 depicts the 3D response curve of the biodiesel yield
against reaction temperature and methanol to oil molar ratio.
3.1.1. Analysis of FAME with GC–MS The higher F-value of the model term represents a greater impact
A detailed quantification of the biodiesel composition was per- of temperature on biodiesel yield. Thus, it is apparently found that
formed by GC–MS, and the GC chromatogram is shown in Fig. 2. temperature significantly affects the conversion of soybean oil to
FAME profile of the experimental run conducted at the optimum FAME compared with the other two parameters (i.e. methanol to
conditions (250 °C, methanol to oil molar ratio of 23:1, 1.0 wt% oil molar ratio and catalyst amount). As temperature increases
CH3ONa and 20 min) is summarized in Table 2. 9-Octadecenoic from 150 to 250 °C, the yield increased from 63.47% to 97.42%.
acid, methyl ester, (9E)-(C18:1) and 9,12-Octadecadienoic acid, The variation tendency suggests that a higher temperature could
methyl ester (C18:2) are the two major constituents present in lead to a higher conversion yield. It is consistent with Demirbas’
FAME. The presence of these monounsaturated compounds conclusion that increasing reaction temperature, especially super-

Table 1
Experimental design matrix and the corresponding biodiesel yield in transesterification.

Std. order Run order Temperature Methanol to oil molar ratio Catalyst amount (wt.%) Observed yield (%) Predicted yield (%)
(°C)
1 1 150 23 1.0 63.47 62.81
4 2 250 69 1.0 90.01 90.67
13 3 200 46 1.0 94.09 94.90
12 4 200 69 1.5 78.91 79.39
7 5 150 46 1.5 64.78 65.14
8 6 250 46 1.5 90.16 89.03
3 7 150 69 1.0 74.79 73.96
15 8 200 46 1.0 94.98 94.90
10 9 200 69 0.5 92.89 92.59
9 10 200 23 0.5 78.84 78.36
14 11 200 46 1.0 95.62 94.90
6 12 250 46 0.5 92.32 91.96
5 13 150 46 0.5 62.57 63.70
2 14 250 23 1.0 97.42 98.25
11 15 200 23 1.5 89.76 90.06
742 D. Zeng et al. / Fuel 203 (2017) 739–748

Fig. 2. A typical GC chromatogram of biodiesel.

critical temperatures has a favorable influence on the biodiesel previous article [32]. Since the enthalpy for the transesterification
yield [31]. The thermodynamic reason has been explained in our has become positive in the SC-MeOH temperature domain, the
reaction equilibrium is displaced in the direction of higher FAME
concentration. Considering that in Fig. 3 biodiesel yield does not
Table 2
Fatty acid profile of biodiesel (as % of methyl esters). considerably increase by elevating reaction temperature beyond
250 °C and higher reaction conditions cause higher energy con-
Residence Type of FAME % Area
sumption, the reaction temperature was chosen at 250 °C.
time
5.323 9-Hexadecenoic acid, methyl ester (C16:1) 0.198
3.2.2. Effect of methanol to oil molar ratio
5.611 Hexadecanoic acid, methyl ester (C16:0) 8.691
9.442 9,12-Octadecadienoic acid, methyl ester (C18:2) 38.452 The experiment results indicated that the methanol to oil molar
9.649 9-Octadecenoic acid, methyl ester, (9E)-(C18:1) 42.686 ratio has a gradually increasing impact on the conversion into
9.724 11-Octadecenoic acid, methyl ester (C18:1) 2.915 FAME, as illustrated in Fig. 4. A high molar ratio of methanol
9.954 9,12,15-Octadecatrienoic acid, methyl ester, (Z,Z,Z)- 0.861 pushes the chemical equilibrium forward. Furthermore, an excess
(C18:3)
10.250 Octadecanoic acid, methyl ester (C18:0) 3.898
amount of methanol increases the contact between methanol
12.502 Docosanoic acid, methyl ester (C22:0) 0.836 and soybean oil, improving transesterification reaction [33]. How-
14.948 cis-11-Eicosenoic acid, methyl ester (C20:1) 0.837 ever, with the increase of methanol ratio, the solubility of bypro-
15.558 Eicosanoic acid, methyl ester (C20:0) 0.625 duct glycerol in biodiesel increases that shifts reaction
Saturated fatty acid 14.050 equilibrium backsides. Considering the interaction between fac-
Monounsaturated fatty acid 46.636 tors, temperature gives the most prominent effect on the biodiesel
Polyunsaturated fatty acid 39.313
yield. As shown in Table 1, the highest value of 97.42% was
 D. Zeng et al. / Fuel 203 (2017) 739–748 743

Table 3
Result of ANOVA for response surface quadratic model.

Source Sum of square Degree of freedom Mean square F-value Prob.>F Significant
Model 0.22 9 0.024 174.84 <0.0001 Yes
A-A 0.14 1 0.14 984.29 <0.0001 Yes
B-B 6.301E004 1 6.301E004 4.56 0.0858 No
C-C 1.133E004 1 1.133E004 0.82 0.4067 No
AB 8.780E003 1 8.780E003 63.56 0.0005 Yes
AC 4.774E004 1 4.774E004 3.46 0.1221 No
BC 0.016 1 0.016 112.22 0.0001 Yes
A2 0.041 1 0.041 298.07 <0.0001 Yes
B2 3.142E003 1 3.142E003 22.75 0.0050 Yes
C2 0.017 1 0.017 126.52 <0.0001 Yes
Residual 6.906E004 5 1.381E004
Lack of fit 5.725E004 3 1.908E004 3.23 0.2452 No
Pure error 1.181E004 2 5.904E005
Cor. Total 0.22 14

R2 = 0.9968, R-Sq(adj) = 0.9911, R-Sq(pred) = 0.9568.

Fig. 3. Effects of temperature and methanol to oil molar ratio on biodiesel yield (catalyst amount = 1.0 wt%, reaction time = 20 min).

Fig. 4. Effects of methanol to oil molar ratio and catalyst amount on biodiesel yield (reaction temperature = 250 °C, reaction time = 20 min).

obtained at the reaction conditions of the methanol to oil molar tical for a multi-factor influence experiment. Therefore, the opti-
ratio at 23:1, so single factor analysis is not so accurate and prac- mum molar ratio of methanol to oil was 23:1.
744 D. Zeng et al. / Fuel 203 (2017) 739–748

Fig. 5. Effects of temperature and catalyst amount on biodiesel yield (methanol to oil molar ratio = 23:1, reaction time = 20 min).

3.2.3. Effect of catalyst amount amount. However, a weaker interaction between reaction temper-
In our previous study, the biodiesel yield was only 94.80% by ature and catalyst amount was illustrated in Fig. 4.
the transesterification from soybean oil in supercritical methanol
without catalysts. The optimum reaction conditions were metha- 3.2.5. Evaluation of the integrated process
nol to soybean oil molar ratio of 46:1 at 300 °C for 45 min [23]. An additional experiment was performed under the optimum
In this work, little amount of CH3ONa (0.5–1.5 wt%) can lead to rel- reaction conditions to evaluate the accuracy of the model. The
atively mild reaction conditions. Fig. 5 shows the interaction effect experimental value of the biodiesel yield was 97.83%, which only
between temperature and catalyst amount on the biodiesel yield. has a percentage error of 4.27% with the predicted yield of
In the beginning, the biodiesel yield raised with the increase of 98.25%. Since the verification value well agreed with the estimated
CH3ONa amount. CH3ONa dissociation in methanol produces the value of the fitted model equation, it offers a reasonable accuracy
necessary methoxide anion to begin the reaction. Methoxide anion of the regression model to predict the optimum conditions for
(CH3O-) is strongly basic and has high catalytic activity in transes- achieving a maximum biodiesel yield.
terification reactions [17]. The reaction mechanism for methoxide-
catalyzed oil transesterification is formulated as three steps. As 3.3. Kinetic study of the transesterification process
expressed by Fig. 6, transesterification is a nucleophilic substitu-
tion reaction in the presence of methanol, and methoxide functions 3.3.1. Experiment
as a nucleophile [18]. The methoxide anion first attaches to the car- The kinetics of the transesterification process was investigated,
bonyl carbon atom of the triglycerides (TG) to form a tetrahedral and the variation curves of the biodiesel yield from 150 °C to
intermediate. Then, the tetrahedral intermediate reacts with 250 °C with respect to time are shown in Fig. 7. The other condi-
methanol to generate methoxide anion. At last, diglycerides (DG) tions are settled at the methanol to oil molar ratio of 23:1 and
and alkyl ester are formed by the rearrangement of the tetrahedral 1.0 wt% CH3ONa with the reactor oscillating velocity of 50–
intermediate. DG and monoglycerides (MG) are converted by the 60 rpm. From Fig. 7, it is evident that with an increase in time,
same mechanism to produce alkyl ester and the by-product glyc- the biodiesel yield increases gradually and tails at 20 min.
erol, where R1, R2, and R3 represent long chain alkyl groups, respec-
tively. The maximum yield of 97.42% shown in Table 1 was
3.3.2. Kinetic parameters of the process
obtained at the CH3ONa amount of 1.0 wt%.
The overall transesterification reaction of TG to form FAME and
glycerol (GL) in the presence of methanol (MeOH) is shown in Eq.
(5).
3.2.4. Effect of interaction process variables
The results of ANOVA in Table 3 revealed that A, AB, BC and all TG þ 3MeOH ! 3FAME þ GL ð5Þ
the quadratic terms are significant model terms having large effect
on the biodiesel yield because of their low P-values. The higher F The rate of formation of product has been found to be well cor-
value for a parameter demonstrate its higher significance on the related by the following rate equation
biodiesel yield. Thus, the order of influences of these parameters r A ¼ k0 cnA cm ð6Þ
B
was as follows: reaction temperature (A) > methanol to oil molar
ratio (B) > catalyst amount (C). Namely, the reaction temperature As the amount of methanol is excess, we may assume that the con-
is the most important variable for the catalytic-supercritical trans- B :Thus,
centration of methanol is constant. Besides, set k ¼ k0 cm
esterification reaction. Figs. 3–5 show the 3D response curves
dcA n
related to the effect of two variables on the biodiesel yield. The r A ¼  ¼ kcA ð7Þ
dt
curved nature of 3D response in Fig. 3 and 5 indicated relatively
significant interactions of reaction temperature with methanol to Because the concentration of TG at any moment of the reaction can
oil molar ratio and methanol to oil molar ratio with catalyst be expressed as cA ¼ c0 ð1  xÞ, so
 D. Zeng et al. / Fuel 203 (2017) 739–748 745

Fig. 6. Mechanism of methoxide-catalyzed transesterification of soybean oil.

By taking the natural logarithm Eq. (9) can be written as Eq. (10)
dx
ln ¼ n ln½c0 ð1  xÞ þ ln k0 ð10Þ
dt
0
In Eq. (6), k ¼ k=c0
As shown in Eq. (10), the order of the reaction and the rate con-
stant are determined from the plot of ln dx
dt
vs. ln[c0(1  x)]. Here, we
use the differential method to identify the order of the reaction by
Exponential function in Origin software. The exponential function
expressed as

y ¼ y0 þ AeR0 x ð11Þ
Taken the obtained plot at 150 °C as an example, the fitting
function can be expressed as

xðtÞ ¼ 0:7586  0:7636 eð0:0396tÞ ð12Þ

The differential equation governing the transesterification reac-
tion of TG is expressed by
Fig. 7. Effect of temperature on biodiesel yield (experimental assessed).
dx
¼ 0:0303eð0:0396tÞ ð13Þ
dt
dcA d½c0 ð1  xÞ dx According to the experiment results, a well-fitted differential
r A ¼  ¼ ¼ c0 ¼ k½c0 ð1  xÞn ð8Þ
dt dt dt plot of ln dx vs. ln [c0(1  x)] can be obtained in Fig. 8. A straight line
dt
From Eq. (8), the rate equation can be expressed as Eq. (9) is observed as
dx k dx
¼ ½c0 ð1  xÞn ð9Þ ln ¼ 1:3965 ln½c0 ð1  xÞ  2:5979 ð14Þ
dt c0 dt
746 D. Zeng et al. / Fuel 203 (2017) 739–748

Fig. 8. Plot of ln[c0(1  x)] vs.ln dx at 150 °C. Fig. 9. Plot of 1/T vs. ln k.
dt

with the R2 value of 0.9884. From Eqs. (10) and (14), the reaction ature in Table 4 [34]. After taken the obtained parameters into Eqs.
order is obtained as 1.4, and the reaction rate constant is 0.0387. (15) and (7), the kinetic model of this integrated process can be
Similarly, the linear fitting plots at 175 °C, 200 °C, 225 °C and expressed as Eq. (17).
250 °C can be obtained, and the results are listed in Table 4.
dcA 27:06103
 ¼ 102:71e RT c1:5 A ð17Þ
3.3.3. Activation energy determination dt
The Arrhenius equation can be written as Eq. (15). By taking the Table 5 shows the comparison of kinetic parameters between
natural logarithm, Eq. (15) can be written as Eq. (16). this integrated process and other researchers’ work, the activation
Ea
energy for tannery waste [35], linseed oil [36], castor oil [36] and
k ¼ AeRT ð15Þ
rapeseed oil [33] is higher than the value obtained in this study.
It is because the different fatty acid composition of oil used and
Ea
ln k ¼  þ ln A ð16Þ the small amount of the involved catalyst, which lowered the acti-
RT
vation energy.
where A is the pre-exponential factor, and Ea is the activation
energy (kJ
mol1). R is the molar universal gas constant
(8.314 J
mol1
K1), and T represents the absolute temperature (K).
A linear correlation between ln k and T1 can be found in Eq.
(16). By using the rate constants listed in Table 4 and the corre-
sponding temperature from 150 °C to 250 °C, The straight line with
the R2 value of 0.9016 was obtained in Fig. 9. The activation energy
and pre-exponential factor can be calculated from the intercept
and the slope as 27.06 kJ
mol1 and 102.71, respectively. Besides,
the reaction order of SC-MeOH and CH3ONa-catalyzed transesteri-
fication is 1.5 which is averaged by the order value at each temper-

Table 4
Reaction order, reaction rate constant and relative value at different temperatures.

T/°C Rate constant k /min1 Order n R2

150 0.0387 1.4 0.9884
175 0.0858 1.9 0.9946
200 0.1207 1.8 0.9964
225 0.1578 1.4 0.9905
250 0.1730 1.0 0.9722
Fig. 10. Plot of 1/T vs. ln k/T.

Table 5
Comparison of the kinetic parameters calculated by different methods.

Method Feedstock Reaction order Pre-exponential factor Activation energy Ref.

n A Ea /kJ
mol1
Integrated process Soybean oil 1.5 102.71 27.06 This study
SC-MeOH Transesterification Tannery waste 1.0 1176.67 36.01 [35]
Linseed oil 4.68 46.50 [36]
Castor oil 32.40 35.00 [36]
Rapeseed oil 3.21  105 67.76 [33]
 D. Zeng et al. / Fuel 203 (2017) 739–748 747

Table 6
Thermodynamic parameters in the integrated transesterification process.

Feedstock DH (kJ
mol1) DS (kJ
mol1
K1) DG (kJ
mol1) Ref.

Soybean oil 23.15 0.22 137.43 This study
Tannery waste 31.37 0.23 153.64 [35]
Rapeseed oil 19.59 0.19 75.26–79.06 [37]
Soybean oil 28.33 0.18 83.30–87.69 [38]

3.4. Thermodynamic analysis Acknowledgments

Thermodynamic parameters mainly including the enthalpy The authors would like to acknowledge financial support for
(DH), entropy (DS), and the Gibbs free energy (DG) are important this research from the National Natural Science Foundation of
features to evaluate the behavior of transesterification reactions. China (No. 21376186).
As the value of Gibbs free energy can be calculated from the
Eyring-Polanyi equation, which can be written as Eq. (18).
  References
kB T DG
k¼j exp  ð18Þ
h RT [1] Hoekman SK, Broch A, Robbins C, Ceniceros E, Natarajan M. Review of biodiesel
composition, properties, and specifications. Renewable Sustainable Energy Rev
where j is the transmission coefficient and usually taken as unity, 2012;16:143–69.
kB is the Boltzmann constant (1.38  1023 J
K1), and h is the [2] Sajjadi B, Raman AAA, Arandiyan H. A comprehensive review on properties of
Planck’s constant (6.63  1034 J
s). Taking the natural logarithm edible and non-edible vegetable oil-based biodiesel: composition,
specifications and prediction models. Renewable Sustainable Energy Rev
of Eq. (18) and substituting DG ¼ DH  T DS gives 2016;63:62–92.
        [3] Sharma YC, Singh B, Upadhyay SN. Advancements in development and
k  DH 1 kB DS
ln ¼ þ ln j þ ln þ ð19Þ characterization of biodiesel: a review. Fuel 2008;87:2355–73.
[4] Ma Z, Shang ZY, Wang EJ, Xu JC, Xu QQ, Yin JZ. Biodiesel production via
T R T h R
transesterification of soybean oil using acid catalyst in CO2 expanded methanol
Eq. (19) describes the relationship between enthalpy and liquids. Ind Eng Chem Res 2012;51:12199–204.
[5] Wu L, Huang KL, Wei TY, Lin ZJ, Zou Y, Tong ZF. Process intensification of
entropy with the connection of rate constant. The values of these NaOH-catalyzed transesterification for biodiesel production by the use of
two thermodynamic parameters can be obtained from the slope bentonite and co-solvent (diethyl ether). Fuel 2016;186:597–604.
and the intercept of the linear plot of 1/T vs. ln (k/T). From [6] Kansedo J, Lee KT. Process optimization and kinetic study for biodiesel
production from non-edible sea mango (Cerbera odollam) oil using response
Fig. 10, values of enthalpy and entropy were found to be
surface methodology. Chem Eng J 2013;214:157–64.
23.15 kJ
mol1 and 0.22 kJ
mol1
K1, respectively. Thus, the [7] Veny H, Aroua MK, Sulaiman NMN. Kinetic study of lipase catalyzed
value of Gibbs free energy can be calculated as 137.43 kJ
mol1 transesterification of jatropha oil in circulated batch packed bed reactor.
at 523 K. The positive value of enthalpy indicates that the external Chem Eng J 2014;237:123–30.
[8] Sendzikiene E, Sinkuniene D, Kazanceva I, Kazancev K. Optimization of low
heat input is required to raise the energy level for the reactants to quality rapeseed oil transesterification with butanol by applying the response
transform to their transition state. The negative value of entropy surface methodology. Renewable Energy 2016;87:266–72.
means the reactants joined to form transition state, which is more [9] Guldhe A, Singh P, Ansari FA, Singh B, Bux F. Biodiesel synthesis from
microalgal lipids using tungstated zirconia as a heterogeneous acid catalyst
structured to the ground state. The positive Gibbs free energy and its comparison with homogeneous acid and enzyme catalysts. Fuel
shows that the SC-MeOH and CH3ONa-catalyzed transesterifica- 2017;187:180–8.
tion is unspontaneous and endergonic. Similar trends for the ther- [10] Saydut A, Erdogan S, Kafadar AB, Kaya C, Aydin F, Hamamci C. Process
optimization for production of biodiesel from hazelnut oil, sunflower oil and
modynamic parameters of the transesterification reaction have their hybrid feedstock. Fuel 2016;183:512–7.
been reported in many previous works. Some examples are pre- [11] Aguieiras ECG, Cavalcanti-Oliveira ED, Freire DMG. Current status and new
sented in Table 6. Various feedstock and different transesterifica- developments of biodiesel production using fungal lipases. Fuel
2015;159:52–67.
tion methods lead to some differences in values of [12] Pinnarat T, Savage PE. Assessment of noncatalytic biodiesel synthesis using
thermodynamic parameters for biodiesel production. supercritical reaction conditions. Ind Eng Chem Res 2008;47:6801–8.
[13] Leung DYC, Wu X, Leung MKH. A review on biodiesel production using
catalyzed transesterification. Appl Energy 2010;87:1083–95.
4. Conclusions [14] Rashid U, Anwar F, Ansari TM, Arif M, Ahmad M. Optimization of alkaline
transesterification of rice bran oil for biodiesel production using response
surface methodology. J Chem Technol Biotechnol 2009;84:1364–70.
The process combining supercritical technology and alkali- [15] Amini-Niaki SR, Ghazanfari A. Comparison of fuel and emission properties of
catalyzed transesterification possesses the advantages of signifi- petro diesel and sunflower biodiesel prepared by optimized production
cantly decreased temperature, relatively small catalyst consump- variables. Fuel 2013;109:384–8.
[16] Vicente G, Martı´nez M, Aracil J. Integrated biodiesel production: a comparison
tion, and enhanced reaction rate compared with either single
of different homogeneous catalysts systems. Bioresour Technol
method. It was successfully used to synthesize biodiesel under 2004;92:297–305.
the optimum conditions of 250 °C, methanol to oil molar ratio of [17] Ma F, Hanna MA. Biodiesel production: a review. Bioresour Technol
23:1, 1.0 wt% CH3ONa and 20 min, the biodiesel yield exceeded 1999;70:1–15.
[18] Liu L. How is chemical interesterification initiated: nucleophilic substitution or
97%. The temperature was found to be the most significant param- a-proton abstraction? J Am Oil Chem Soc 2004;81:331–7.
eter based on RSM analysis. The second-order polynomial regres- [19] Borugadda VB, Goud VV. Biodiesel production from renewable feedstocks:
sion model was validated to be appropriate in designing the status and opportunities. Renewable Sustainable Energy Rev
2012;16:4763–84.
similar reaction system. The kinetic model of this integrated pro- [20] He H, Wang T, Zhu S. Continuous production of biodiesel fuel from vegetable
27:06103
cess was found to be  dcdtA ¼ 102:71e RT c1:5
A . Further, the ther-
oil using supercritical methanol process. Fuel 2007;86:442–7.
[21] Roman-Figueroa C, Olivares-Carrillo P, Paneque M, Palacios-Nereo FJ, Quesada-
modynamic parameters of the transesterification process were Medina J. High-yield production of biodiesel by non-catalytic supercritical
evaluated. The values of DH, DS and DG were found to be methanol transesterification of crude castor oil (Ricinus communis). Energy
23.15 kJ
mol1 and 0.22 kJ
mol1
K1 and 137.43 kJ
mol1 at 2016;107:165–71.
[22] Garcia-Martinez N, Andreo-Martinez P, Quesada-Medina J, de los Rios AP,
523 K, respectively, which show that the reaction process was Chica A, Beneito-Ruiz R, Carratala-Abril J. Optimization of non-catalytic
endothermic, endergonic and unspontaneous. transesterification of tobacco (Nicotiana tabacum) seed oil using
748 D. Zeng et al. / Fuel 203 (2017) 739–748

supercritical methanol to biodiesel production. Energy Convers Manage [31] Demirbas A. Biodiesel from vegetable oils via transesterification in
2017;131:99–108. supercritical methanol. Energy Convers Manage 2002;43:2349–56.
[23] Zeng D, Li Y, Fang T. Production of biodiesel in supercritical methanol and [32] Zeng D, Li RS, Jin TW, Fang T. Calculating the thermodynamic characteristics
related correlation on phase equilibrium. Adv Mat Res 2012;560–561:555–62. and chemical equilibrium of the stepwise transesterification of triolein using
[24] Box GEP, Wilson KB. On the experimental attainment of optimum conditions. J supercritical lower alcohols. Ind Eng Chem Res 2014;53:7209–16.
Roy Stat Soc Ser B (Stat Method) 1951;13:1–45. [33] Kusdiana D, Saka S. Kinetics of transesterification in rapeseed oil to biodiesel
[25] Myers RH, Montgomery DC, Anderson-Cook CM. Response surface fuel as treated in supercritical methanol. Fuel 2001;80:693–8.
methodology: process and product optimization using designed [34] O’Brien P, Ross SD. A new differential method for determining orders and rates
experiments. 3rd ed. John Wiley & Sons; 2009. of reactions and its application to solid state reactions. Oart II. Some empirical
[26] Noshadi I, Amin NAS, Parnas RS. Continuous production of biodiesel from observations on the isothermal dehydration of calcium oxalate monohydrate.
waste cooking oil in a reactive distillation column catalyzed by solid Thermochim Acta 1982;53:195–202.
heteropolyacid: optimization using response surface methodology (RSM). [35] Ong LK, Kurniawan A, Suwandi AC, Lin CX, Zhao XS, Ismadji S.
Fuel 2012;94:156–64. Transesterification of leather tanning waste to biodiesel at supercritical
[27] Mohamad M, Ngadi N, Wong SL, Jusoh M, Yahya NY. Prediction of biodiesel condition: kinetics and thermodynamics studies. J Supercrit Fluids
yield during transesterification process using response surface methodology. 2013;75:11–20.
Fuel 2017;190:104–12. [36] Varma MN, Madras G. Synthesis of biodiesel from castor oil and linseed oil in
[28] Chakraborty R, Sahu H. Intensification of biodiesel production from waste goat supercritical fluids. Ind Eng Chem Res 2007;46:1–6.
tallow using infrared radiation: process evaluation through response surface [37] Encinar JM, Pardal A, Sánchez N. An improvement to the transesterification
methodology and artificial neural network. Appl Energy 2014;114:827–36. process by the use of co-solvents to produce biodiesel. Fuel 2016;166:51–8.
[29] Bezerra MA, Santelli RE, Oliveira EP, Villar LS, Escaleira LA. Response surface [38] Wu L, Wei T, Lin Z, Zou Y, Tong Z, Sun J. Bentonite-enhanced biodiesel
methodology (RSM) as a tool for optimization in analytical chemistry. Talanta production by NaOH-catalyzed transesterification: process optimization and
2008;76:965–77. kinetics and thermodynamic analysis. Fuel 2016;182:920–7.
[30] Ramos MJ, Fernandez CM, Casas A, Rodriguez L, Perez A. Influence of fatty acid
composition of raw materials on biodiesel properties. Bioresour Technol
2009;100:261–8.