biomass and bioenergy 33 (2009) 271276

Available at www.sciencedirect.com

http://www.elsevier.com/locate/biombioe

Biodiesel production from palm oil via heterogeneous

transesterification

Jibrail Kansedo, Keat Teong Lee*, Subhash Bhatia

School of Chemical Engineering, Universiti Sains Malaysia, Engineering Campus, Seri Ampangan,
14300 Nibong Tebal, Pulau Pinang, Malaysia

article info abstract

Article history: This paper presents the study of the transesterification of palm oil via heterogeneous
Received 1 November 2007 process using montmorillonite KSF as heterogeneous catalyst. This study was carried
Received in revised form out using a design of experiment (DOE), specifically response surface methodology
16 May 2008 (RSM) based on four-variable central composite design (CCD) with a (alpha) 2. The
Accepted 21 May 2008 transesterification process variables were reaction temperature, x1 (50190  C), reaction
Published online 16 August 2008 period, x2 (60300 min), methanol/oil ratio, x3 (412 mol mol1) and amount of catalyst,
x4 (15 wt%). It was found that the yield of palm oil fatty acid methyl esters (FAME) could
Keywords: reach up to 79.6% using the following reaction conditions: reaction temperature of 190  C,
Biodiesel reaction period at 180 min, ratio of methanol/oil at 8:1 mol mol1 and amount of catalyst
Transesterification at 3%.
Heterogeneous catalysts 2008 Elsevier Ltd. All rights reserved.
Montmorillonite KSF

1. Introduction liquidliquid separation process. The recovery of the homoge-

The production of biodiesel or more commonly fatty acid
methyl esters (FAME) has attracted significant attention lately
due to the increasing demand for a cleaner, safer and renew-
able energy. FAME is produced mainly from vegetable oil,
which is a renewable and sustainable source. The production
of FAME can be categorized into homogenous, heterogeneous
and non-catalytic method depending on the type of catalyst
being used in the process. Traditionally, homogenous method
is used in many commercial production of FAME. However,
this method has many disadvantages.
As reported in the literature, heterogeneous transesterifi-
cation method proved to be more superior as compared to
the homogenous transesterification method especially on
the separation and purification of the product (FAME) [14].
In the homogenous method, the reactants, catalyst and
FAME are all in the liquid phase, thus resulting in a complex
nous catalyst is also hard, thus resulting in loss of useful
material. On the other hand, heterogeneous method, which
uses solid catalyst, does not face the same limitations.
Solidliquid separation process is relatively easy as compared
to liquidliquid separation process, making the recovery of
solid catalyst a lot easier. On the other hand, heterogeneous
method eliminates the formation of soap, thus omitting the
need of wash water. By eliminating the formation of soap,
heterogeneous method further prevents formation of emul-
sion in the mixture which could complicate the separation
and purification processes.
Basically, there are three types of heterogeneous catalyst
being used in the transesterification process; i.e. acid, base
(alkali) and enzyme [14]. The particular method, in which
these heterogeneous catalysts are being used, is generally
named after the type of the catalyst itself. For instance, acidic
process refers to heterogeneous transesterification using solid

* Corresponding author. Tel.: 60 4 5996467; fax: 60 4 5941013.

E-mail address: chktlee@eng.usm.my (K.T. Lee).
0961-9534/$ see front matter 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.biombioe.2008.05.011
272 biomass and bioenergy 33 (2009) 271276

acid catalyst. Currently there are many feasible heteroge-
neous catalysts used in the transesterification process such
as metal oxides [59], metal complexes [10], active metals
loaded on supports [11,12], zeolite [13], resins [14,15],
membranes [16,17], and lipases [18,19]. Some of these hetero-
geneous catalysts are already patented and used in commer-
cial production of FAME, such as those invented by Axens
Technology [20]. These catalysts have been proven to have
high activity towards transesterification process. However,
most of these catalysts are used on oils other than palm oil,
such as soybean oil and rapeseed oil.
Thus, the main objective of this study to produce fatty acid
methyl esters (FAME) from palm oil using montmorillonite
KSF as heterogeneous catalyst. This study is aided with
a statistical design of experiment using response surface
methodology (RSM) to accumulate and analyze information
on the effect of four process variables on the yield of transes-
terification in a rapid and efficient manner using minimum
number of experiments.
palm oil FAME (response). The four transesterification process
variables studied are reaction temperature, reaction period,
ratio of oil to methanol and amount of catalyst. Table 2 lists
the range and levels of the four independent variables studied.
The CCD comprises a two-level small factorial design (12
 24 8), eight axial or star points and five center points. The
value of a (alpha) for this CCD is fixed at 2. The complete
design matrix of the experiments employed and results are
given in Table 3. All variables at zero level constitute to the
center points and the combination of each of the variables
at either its lowest (2.0) level or highest (2.0) level with
the other variables at zero level constitutes the axial points.
Each response of the transesterification process was used
to develop a mathematical model that correlates the yield of
palm oil FAME to the transesterification process variables
studied through first order, second order and interaction
terms, according to the following second order polynomial
equation,
X
4 X
4 X
4
Y bo bj xj bij xi xj bjj x2j (1)
j1 ij1 j1

2. Materials and methods
2.1. Materials
Purified palm oil was purchased from Yee Lee Edible Oils Sdn
Bhd, Malaysia whereas methanol from R&M Chemicals, UK.
Methyl heptadecanoate (used as internal standard) and stan-
dard references for FAME analysis: methyl myristate, methyl
palmitate, methyl stearate, methyl oleate, methyl linoleate
were obtained from Fluka Chemie, Germany. The catalyst,
montmorillonite KSF (in powder form), also purchased from
Fluka Chemie, Germany, was used as received without any
pretreatment. The given properties of montmorillonite KSF
were humidity of 12 %, free acid of 812%, and has its surface
coated with sulfuric acid. Montmorillonite KSF has a limited
self-life, therefore it was stored in a freezer (below 18  C) to
maintain its properties prior to use. The properties of palm
oil are given in the Table 1.
2.2. Design of experiments
The experimental design selected for this study is a central
composite design (CCD) that helps in investigating linear,
quadratic, cubic and cross-product effects of the four transes-
terification process variables (independent) on the yield of
where Y is the predicted yield of palm oil FAME, mol mol1, xi
and xj represent the variables or parameters, bo is the offset
term, bj is the linear effect, bij is the first order interaction
effect and bjj is the squared effect.
2.3. Model fitting and statistical analysis
Design Expert software version 6.0.6 (STAT-EASE Inc., Min-
neapolis, USA) was used for regression analysis of the exper-
imental data to fit the second order polynomial equation
and also for evaluation of the statistical significance of the
equation developed.
2.4. Activity study
Transesterification reactions were carried out in a batch
reactor with a magnetic stirrer, as shown in Fig. 1. Mixture
of palm oil, catalyst and methanol was charged into the
reactor. The mixing intensity of the magnetic stirrer was set
at 190200 rpm. The reaction temperature, duration, ratio of
oil/methanol and amount of catalyst (montmorillonite KSF)
were set according to the values proposed in the DOE as
shown in Table 3. The experiments were carried out in
a random manner to minimize the effects of the uncontrolled
factors. The excess methanol from the samples was removed
using rotary evaporator. The upper layer of the sample was

Table 1 Properties of palm oil.

Fatty acids Percentage (%) Table 2 Levels of the transesterification process
variables chosen for this study.
Lauric, C-12:0 0.1
Myristic, C-14:0 1.0 Variable Coding Unit Levels
Palmitic, C-16:0 42.8
a 1 0 1 a
Stearic, C-18:0 4.5

Oleic, C-18:1 40.5 Reaction temperature x1 C 50 80 120 155 190
Linoleic, C-18:2 10.1 Reaction period x2 min 60 120 180 240 300
Linolenic, C-18:3 0.2 Ratio of oil/methanol x3 mol mol1 1:4 1:6 1:8 1:10 1:12
Other/unknown 0.8 Amount of catalyst x4 wt% 1 2 3 4 5
 biomass and bioenergy 33 (2009) 271276 273

Table 3 Experimental design matrix and results (sorted out according to the order of catalyst code).
Catalyst code Experimental variables Yield
(%)
Reaction Reaction Ratio of oil/ Amount of
temperature ( C) period (min) methanol (mol mol1) catalyst (% g)

KSF 01 1 1 1 1 71.9
KSF 02 1 1 1 1 68.0
KSF 03 1 1 1 1 79.0
KSF 04 1 1 1 1 5.85
KSF 05 1 1 1 1 63.4
KSF 06 1 1 1 1 3.92
KSF 07 1 1 1 1 5.45
KSF 08 1 1 1 1 3.32
KSF 09 2 0 0 0 0.85
KSF 10 2 0 0 0 79.6
KSF 11 0 2 0 0 15.4
KSF 12 0 2 0 0 64.4
KSF 13 0 0 2 0 45.3
KSF 14 0 0 2 0 48.4
KSF 15 0 0 0 2 34.6
KSF 16 0 0 0 2 66.4
KSF 17 0 0 0 0 59.0
KSF 18 0 0 0 0 54.6
KSF 19 0 0 0 0 59.7
KSF 20 0 0 0 0 59.0
KSF 21 0 0 0 0 59.7

separated from the bottom layer and immediately quenched respectively. A quantity of 1 ml from each sample was injected
with n-hexane (dilution) prior to analysis of FAME content. into the column. The yield of palm oil FAME from the transes-
terification processes was calculated in similar method used
2.5. Analysis by Jitputti et al. [21].

The analysis of palm oil FAME content in the samples was

carried out using Gas Chromatography (GC) by means of Inert 3. Results and discussion
Cap WAX capillary column (30 mm  0.25 mm, I.D. 0.25 mm).
Helium was used as the carrier gas. Oven temperature at 3.1. Development of regression model equation
120  C was initially hold for 1 min and then increased to
220  C (hold 15 min) at a rate of 4  C min1. The temperatures The final equation in terms of coded factors after excluding
of the injector and detector were set at 220  C and 250  C, the insignificant terms (identified using Fishers test) is

To PID
Controller
Pressure Gauge Release
Valve
Nut/Pin
Lock
Nut/Pin
Metal Cover
Thermocouple
Metal Cup Mixture of palm
oil with methanol
Teflon and catalyst
Heater
Magnetic Insulator
Stirrer

Fig. 1 Schematic diagram of the experimental set-up for the transesterification process (batch reactor).
274 biomass and bioenergy 33 (2009) 271276

Y 58:42 19:70x1 12:23x2 7:95x4  5:39x21

Table 4 Analysis of variance (ANOVA) for the regression
 5:48x22  3:74x23  2:82x24 7:13x1 x2 12:04x1 x4 (2) model equation and coefficients after eliminating
 13:26x2 x4 ; with R 0:9846 insignificant terms.

(Only the terms that satisfied F  4.96 (F0.05(1,10)) were retained.) Source Sum of Degrees of Mean of F-test
squares freedom squares
Positive sign in front of the terms indicates synergistic
effect, while negative sign indicates antagonistic effect. Eq. Model 14822.47 10 1482.25 43.97
(2) shows that the yield of palm oil FAME has a linear and x1 3103.93 1 3103.93 92.08
quadratic effect on the four transesterification process vari- x2 1197.07 1 1197.07 35.51
x3 505.62 1 505.62 15.00
ables studied. Fig. 2 shows the experimental values versus
x21 729.83 1 729.83 21.65
predicted values using the model equation developed. A line x22 753.70 1 753.70 22.36
of unit slope, i.e. the line of perfect fit with points correspond- x23 350.59 1 350.59 10.40
ing to zero error between experimental and predicted values is x24 199.77 1 199.77 5.93
also shown in Fig. 2. This plot therefore visualizes the perfor- x1x2 203.42 1 203.42 6.03
mance of the model in an obvious way. The results in Fig. 2 x1x4 579.49 1 579.49 17.19
x2x4 703.44 1 703.44 20.87
demonstrated that the regression model equation provided
Residual 337.10 10 33.71
a very accurate description of the experimental data, indi-
cating that it was successful in capturing the correlation
between the four transesterification process variables to the
yield of palm oil FAME. KSF to catalyze the transesterification reactions, at any rate, is
due to the acidic properties of the catalyst itself. As mentioned
3.2. Effects of transesterification process variables earlier, montmorillonite KSF has free acid of 812%, and has
its surface coated with sulfuric acid. Sulfuric acid is known
From Table 4, it was observed that among the four individual as a typical strong acid, and since the transesterification reac-
variables studied, reaction temperature (x1) has the largest tions can be catalyzed with acid or base, hence the strong
effect on the yield of palm oil FAME (due to the highest F value) acidity of this montmorillonite KSF will definitely contribute
followed by reaction period (x2). Ratio of oil/methanol (x3) has to the high yield of palm oil FAME.
fairly significant effect whereas amount of catalyst (x4) has However, since the maximum yield occurs at the border of
otherwise insignificant effect. The quadratic term of (x1) and the experimental variables selected for this study, optimum
(x2) was fairly significant on the yield of palm oil FAME experimental conditions to obtain the highest yield cannot
compared to that of (x3) and (x4). The other significant terms be derived (except for the experimental conditions at the
include the effect of interaction between variables, particu- border of the design variables). This finding was also sup-
larly between variables (x2) and (x4) and between variables ported by the model developed using the DOE software
(x1) and (x4). Longer reaction period allows sufficient time for whereby no maximum yield was located within the experi-
the transesterification process to occur completely, or at least, mental domain examined. Therefore, the model developed
until the maximum level. by the DOE software was used to facilitate a straight forward
The results in Table 3 illustrate that there is a great possi- examination of the effects of the variables and their interac-
bility in improving the yield of palm oil FAME with proper tion towards the yield of the transesterification process.
selection of transesterification process variables using mont- Among the four interaction terms that were found to have
morillonite KSF as the catalyst. The ability of montmorillonite significant effect on the yield of palm oil FAME, emphasis
was given to the interaction between (x2) and (x4) and between
variables (x1) and (x4).
Fig. 3 shows the changes in yield with varying reaction
temperature at 2% and 4% catalyst. The other two process
variables: reaction period and ratio of oil to methanol were
kept constant at 180 min and 1:8, respectively. It is generally
perceived that higher percentage of catalyst used will result
in higher yield of triglycerides to fatty acid methyl esters
because of higher availability of active sites. However, this
perception may not always be true. Referring to Fig. 3, the
yield of oil to biodiesel was found to be lower at 4% catalyst
as compared to 2% catalyst when the reaction temperature
was lower than 95  C. This may suggest the real interactions
between the reaction temperature and the active site on the
catalyst itself.
Fig. 4 shows the changes in yield of palm oil FAME with
varying reaction period at 2% and 4% catalyst. The other two
process variables: reaction temperature and ratio of oil to
Fig. 2 Predicted versus experimental yield of palm oil methanol were kept constant at 120  C and 1:8, respectively.
FAME. As shown in Fig. 4, at a shorter reaction period, 2% catalyst
 biomass and bioenergy 33 (2009) 271276
Fig. 3 Effect of reaction temperature and amount of
catalyst on the yield of palm oil to FAME: (a) response
surface plot and (b) two-dimensional drawing.
showed a significant lower yield of palm oil FAME compared to
that of 4% catalyst. However, the yield by 2% catalyst
continues to steadily increase until it subsequently exceeds
to that of 4% catalyst after 210 min of reaction. On the other
hand, the 4% catalyst showed a fairly constant yield for
most of the time and slightly dropped when the reaction
period is beyond 210 min. From this behavior, it can be said
that the equilibrium of the reactions is reached rapidly with
4% catalyst, and after reaching the equilibrium, the transes-
terification reactions may become stagnant or start to reverse.
The equilibrium of transesterification reactions by 2% catalyst
may only be reached after longer reaction time because of
smaller active site availability, thus defines the staging incre-
ment of conversion by 2% catalyst from 120 min to 240 min.
After all, the total rate of reactions, be it transesterification
275
Fig. 4 Effect of reaction period and amount of catalyst on
the yield of palm oil to FAME: (a) response surface plot and
(b) two-dimensional drawing.
or reversed reaction, may directly be influenced by the total
active sites of the catalyst, whereby higher active sites will
accelerate both reactions.
4. Conclusion
Based on the experimental result obtained, it can be
concluded that montmorillonite KSF could be used as an
effective catalyst for the conversion of palm oil to FAME.
However, future research on this catalyst is still required to
investigate wider range of the parameters, and also new
possible parameters which can be included in the reaction
process such as the mixing intensity, reaction pressure and
so forth to properly commercialize this technology.
276 biomass and bioenergy 33 (2009) 271276
Acknowledgement
The authors would like to thank Universiti Sains Malaysia for
providing fund for this project.
references
[1] Ma F, Hanna MA. Biodiesel production: a review. Bioresource
Technology 1999;70:115.
[2] Fukuda H, Kondo A, Noda H. Biodiesel fuel production by
transesterification of oils. Journal of Bioscience and
Bioengineering 2001;92(5):40516.
[3] Van Gerpen J. Biodiesel processing and production. Fuel
Processing Technology 2005;86:1097107.
[4] Demirbas A. Progress and recent trends in biofuels. Progress
in Energy and Combustion Science 2007;33:118.
[5] Xie W, Huang X. Synthesis of biodiesel from soybean oil
using heterogeneous KF/ZnO catalyst. Catalysis Letters 2006;
107(12).
[6] Liu X, He H, Wang Y, Zhu S. Transesterification of soybean oil
to biodiesel using SrO as a solid base catalyst. Catalysis
Communications 2007;8:110711.
[7] Yang Z, Xie W. Soybean oil transesterification over zinc oxide
modified with alkali earth metals. Fuel Processing
Technology 2007;88:6318.
[8] Lopez Granados M, Zafra Poves MD, Martin Alonso D,
Mariscal R, Cabello Galisteo F, Moreno-Tost R, et al. Biodiesel
from sunflower oil by using activated calcium oxide. Applied
Catalysis B: Environmental 2007;73:31726.
[9] Kim HJ, Kang BS, Park YM, Kim DK, Lee JS, Lee KY.
Transesterification of vegetable oil to biodiesel using
heterogeneous base catalyst. Catalysis Today 2004;9395:
31520.
[10] Ferreira DAC, Meneghetti MR, Meneghetti SMP, Wolf SR.
Methanolysis of soybean oil in the presence of tin (IV)
complexes. Applied Catalysis A: General 2007;317:5861.
[11] Xie W, Li H. Alumina-supported potassium iodide as
a heterogeneous catalyst for biodiesel production from
soybean oil. Journal of Molecular Catalysis A: Chemical 2006;
255:19.
[12] Xie W, Peng H, Chen L. Transesterification of soybean oil
catalyzed by potassium loaded on alumina as a solid-base
catalyst. Applied Catalysis A: General 2006;300:6774.
[13] Suppes GJ, Dasari MA, Doskocil EJ, Mankidy PJ, Goff MJ.
Transesterification of soybean oil with zeolite and metal
catalysts. Applied Catalysis A: General 2004;257(2):
21323.
[14] Shibasaki-Kitikawa N, Honda H, Kuribayashi H, Toda T,
Fukumura T, Yonemoto T. Biodiesel production using
anionic ion-exchange resin as heterogeneous catalyst.
Bioresource Technology 2007;98:41621.
[15] Lopez DE, Goodwin JJG, Bruce DA. Transesterification of
triacetin with methanol on Nafion acid resins. Journal of
Catalysis 2007;245:38191.
[16] Guerreiro L, Castanheiro JE, Fonseca IM, Martin-Aranda RM,
Ramos AM, Vital J. Transesterification of soybean oil over
sulfonic acid functionalised polymeric membranes. Catalysis
Today 2006;118:16671.
[17] Dube MA, Tremblay AY, Liu J. Biodiesel production using
a membrane reactor. Bioresource Technology 2007;98:
63947.
[18] Ranganathan SV, Narasimhan SL, Muthukumar K. An
overview of enzymatic production of biodiesel. Bioresource
Technology 2008;99(10):397581.
[19] Shieh CJ, Liao HF, Lee CC. Optimization of lipase-catalyzed
biodiesel by response surface methodology. Bioresource
Technology 2003;88:1036.
[20] Bournay L, Casanave D, Delfort B, Hillion G, Chodorge JA.
New heterogeneous process for biodiesel production: a way
to improve the quality and the value of the crude glycerin
produced by biodiesel plants. Catalysis Today 2005;106:
1902.
[21] Jitputti J, Kitiyanan B, Rangsunvigit P, Bunyakiat K,
Attanatho L, Jenvanitpanjakul P. Transesterification of crude
palm kernel oil and crude coconut oil by different solid
catalysts. Chemical Engineering Journal 2006;116:616.