Energy 89 (2015) 965e973

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Biodiesel synthesis over millimetric g-Al2O3/KI catalyst

Aminul Islam a, b, Yun Hin Tauq-Yap a, b, *, Pogaku Ravindra c, Siow Hwa Teo a, b,
S. Sivasangar a, b, Eng-Seng Chan d, **
a
Department of Chemistry, Faculty of Science, University Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysia
b
Catalysis Science and Technology Research Centre, Faculty of Science, University Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysia
c
Centre of Materials and Minerals, School of Engineering and IT, Universiti Malaysia Sabah, 88999 Kota Kinabalu, Malaysia
d
Chemical Engineering Discipline, School of Engineering, Monash University, Jalan Lagoon Selatan, 46150 Bandar Sunway, Selangor, Malaysia

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

Article history: The use of spherical millimetric gamma-alumina (g-Al2O3) as a catalyst support for the production of
Received 4 August 2014 biodiesel from palm oil was demonstrated. The catalyst support was produced using dripping method,
Received in revised form and KI catalyst was loaded on the support using impregnation method. The highest FAME (fatty acid
23 April 2015
methyl ester) yield of 98% was obtained when the reaction was carried out under the conditions of
Accepted 11 June 2015
catalyst to oil ratio of 0.6 g (4wt.%, gcat./goil) using 0:24 gKI =ggAl2 O3 of catalyst loading, reaction time of 4 h,
Available online 2 July 2015
temperature of 60
C and methanol to palm oil molar ratio of 14:1. XRD (X-ray diffraction) analysis
showed the formation of K2O and KAlO2 phases on the KI/g-Al2O3 catalyst which were possibly the active
Keywords:
Solegel
sites for the transesterication reaction. The highest number and strength of basic sites generated from
g-alumina the solid phase reaction of the KI/g-Al2O3 catalyst loaded with 0.24 g kF/g g-Al2O3 were conrmed by
Calcination temperature programmed desorption of CO2 (CO2-TPD) analysis. The nitrogen adsorptionedesorption
Transesterication isotherms was revealed a mesoporous structure of the catalysts. The leaching of potassium species in
Biodiesel reused catalysts was observed, as veried by XRF (X-ray uorescence). KI/g-Al2O3 had a long life time and
maintained sustained activity even after being repeatedly used for 11 cycles. The biodiesel yield was
comparable to that produced from smaller catalysts. Thus, the catalyst could be potential for industrial
use, as they can be handled safely, easily separated from the process uid and used in xed bed reactor to
minimize pressure drop.
2015 Elsevier Ltd. All rights reserved.

1. Introduction KI/g-Al2O3 [5], KOH/ZrO2 [6], KNO3/Al2O3 [7] and K/KOH/g-Al2O3

The most effective way to reduce CO2 (carbon dioxide) emis-
sions is to reduce fossil fuel consumption. Therefore, the biodiesel is
gaining more attention as a carbon neutral source of energy. Recent
research efforts have been directed towards the more economical
and environmental friendly way to produce biodiesel using solid
catalysts [1,2]. A number of studies have been focused on using
solid basic or acidic catalyst for biodiesel production. Different
types of powdered solid basic catalysts ordering on a micrometer
scale have been studied, including KI/Al2O3 [3], CaO/KI/g-Al2O3 [4]
* Corresponding author. Department of Chemistry, Faculty of Science, University
Putra Malaysia, 43400 UPM Serdang, Selangor, Malaysia. Tel.: 60 3 55145821;
fax: 60 3 55146207.
** Corresponding author.
E-mail addresses: tauq@upm.edu.my (Y.H. Tauq-Yap), chan.eng.seng@
monash.edu (E.-S. Chan).
[8]. Recently, the utilization of supported nano catalyst has been
shown to be efcient for biodiesel production [9e11].
Despite the high percentage of FAME (fatty acid methyl ester)
(>90%) achieved by powdered catalysts, many catalytic systems
have not been commercialized because of the difculties encoun-
tered when trying to separate such catalysts from the reaction
media [12]. In addition, the small particle size gives rise to several
problems such as high pressure drops, poor mass/heat transfer,
poor contact efciency and difculties in handling and separation
[13,14]. There are also possible health risks caused by inhalation of
small particles. Therefore, the fabrication of catalyst with macro-
scopic form could be useful for industrial applications, as reported
by Wang et al. [15].
One of the main reasons to consider heterogeneous catalysis for
industrial processes is the ease of catalyst separation after the re-
action. Separation processes represent more than half of the total
investment in equipment for the chemical and fuel industries [16].

http://dx.doi.org/10.1016/j.energy.2015.06.036
0360-5442/ 2015 Elsevier Ltd. All rights reserved.
966 A. Islam et al. / Energy 89 (2015) 965e973
Thus, the utilization of millimetric spherical catalyst support and
their application in transesterication reactions will be a practical
alternative to catalyst in powder form in view of following advan-
tages; i) easy of separation of the catalyst by simple ltration, ii) the
catalyst is easy to handle and reusable for several times. In addition,
in the design of a catalyst, the shape should also be taken into ac-
count, as the shape preferably, spherical could minimize the abra-
sion of catalyst in the reaction medium [17]. In fact, the life time of
the catalyst could be increased. Until now, the macroscopic (up to
1.0 mm in diameter) magnesia-rich magnesium aluminate spinel
catalyst, such as MgO$MgAl2O4 and MgO/MgAl2O4/g-Al2O3 for
biodiesel production have been reported by Wang et al. [15].
However, the biodiesel yield was still insufcient for practical ap-
plications. For instant, biodiesel yield obtained with the catalyst
MgO$MgAl2O4 and MgO/MgAl2O4/g-Al2O3 were substantially low
at 57% and 36% respectively in 10 h.
The aim of this work was to improve the biodiesel yield from
transesterication of palm oil using a spherical millimetric gamma-
alumina (g-Al2O3) catalyst support. g-Al2O3 was chosen because it
can easily be synthesized, it possesses good support properties, and
it is commercially available. KI (potassium iodide) was used as a
model catalyst. The catalysts were prepared at different loading
contents, and the effects of these catalysts on the transesterication
reaction were evaluated in terms of the FAMEs yield. In addition,
the characteristics of the synthesized catalysts, such as the
morphological, structural, textural and base properties, were car-
ried out using stereozoom microscopy, XRD (X-ray diffraction), N2
adsorption/desorption isotherms and CO2-TPD (temperature-pro-
grammed reduction) techniques. A screening of the reaction con-
ditions has been carried out by examining the effects of reaction
time, methanol to oil molar ratio and the calcination temperature of
catalyst. Besides, the relationship between the basicity of the
catalyst and their catalytic activity in the transesterication of palm
oil was also discussed. Finally, the reusability of the catalyst was
determined and the leachate of catalyst into the reaction product
was veried by XRF (X-ray uorescence).
2. Materials and methods
2.1. Materials
Boehmite (AlOOH) in a powder form was provided by BASF
Catalysis LLC, USA. Reagents used during catalyst synthesis were:
hydrochloric acid, HCl (JT Baker, Mexico), ammonia solution (Sys-
tem ChemAR, Malaysia), parafn oil (870 g/mL density at 20
C,
Ajax chemicals, Australia) and potassium iodide, KI (Sigma-Aldrich,
GmbH, Germany). Commercial edible grade palm oil was purchased
from a local store (Lam soon edible oil, Malaysia). Methanol (Lab-
scan Asia Co. Ltd. Thailand) was used for the transesterication
reactions and hexane was purchased from Fisher Scientic, UK.
Methyl heptadecanoate (internal standard for gas chromatography)
and standard references for fatty acid methyl ester analysis were
supplied by Fluka Analytical (Buchs, Switzerland).
2.2. Preparation of catalyst
Spherical millimetric gamma-alumina (g-Al2O3) support was
prepared using oil drop granulation method described in previous
publications [18e20]. Spherical millimetric gamma-alumina (g-
Al2O3) support was prepared using a solegel process previously
described by Wang et al. [18]. The boehmite powder (300 g/L) was
suspended in deionized water and dispersed ultrasonically by a
sonicator. The sonication process took place for 3 min at ampli-
tude of 70%. The pH of the resulting boehmite suspension prior to
drop formation was then adjusted to 1.0 by addition of an
appropriate amount of 1N HCl, where the sol turned to a gel. The
gel was then transferred dropwise by a pump at a constant ow
rate (0.5 mL/min) into a liquid column consisting of parafn oil at
the upper layer and ammonia solution at the bottom layer (Fig. 1).
Droplets of boehmite gel were formed due to the surface tension
effect during transiting through the oil layer and the gel droplets
were further aged in the ammonia solution for 1 h. The parafn
solution column was 25 cm in height, which provided enough
time for the gel drops to form spherical particles [19,20]. During
aging, ammonia would neutralize the acid in the wet-gel beads
and thus, the wet-gel droplets became rigid. Then, the beads were
then dried in the air at room temperature (25
C) for 12 h and
were later calcined at 800
C for 3 h. The calcined beads were then
impregnated with an aqueous solution of KI (potassium iodide), as
shown in Table 1. Finally, the bead-supported catalysts were
calcined at 500
C for 3 h to make the catalyst active for trans-
esterication reaction.
2.3. Catalyst characterization
The shape and size analysis of the support beads were carried
out by analyzing the digital images captured by a digital camera
(Moticam-350, version 2.0 ML, China) installed on a stereozoom
microscope (Stemi DV4, Carl Zeiss, Germany) using image analyzer
(SigmaScan Pro 5.0, SPSS Inc). In order to measure the shape of the
bead quantitatively, SF (sphericity factor) was used. It express the
divergence of a particle shape from spherical [21], where the value
zero indicates a perfect sphere and higher values indicate a greater
degree of shape distortion. SF was calculated according to eq. (1):



SF Dmax  Dper Dmax Dper (1)
where Dmax is the maximum diameter passing through a bead
centroid (mm), and Dper is the diameter perpendicular to Dmax
passing through the bead centroid (mm).
The crystal structure of the catalyst was determined by Shi-
madzu diffractometer model XRD-6000 using Cu-Ka radiation. The
samples were scanned from 10 to 70
with a speed of 0.05 s1. The
peaks were analyzed with the DiffracPlus software and the phases
were identied by the PDF (powder diffraction le) database
(JCPDS, International Centre for Diffraction Data).
The basicity of the catalyst was determined by temperature
programmed desorption of CO2, as described by Islam et al. [17].
The experiments were performed by loading the samples (100 mg)
in the Thermo Finnigan TPD/R/O 1100 series catalytic surface
analyzer connected to a thermal conductivity detector. The samples
were preheated to 800
C under a owing stream of helium for
30 min (10
C min1, 30 mL min1). Then, the temperature was
decreased to 30
C under a CO2 ow for 1 h (30 mL min1) at which
the adsorption of CO2 will be carried out. Subsequently, the sample
was ushed with helium at 30
C for 30 min in order to remove any
physisorbed molecules from the samples prior to the CO2 desorp-
tion analysis. The desorption of CO2 was carried out by heating the
sample up to 800
C under a helium ow (10
C min1,
30 mL min1) and the amount of desorbed CO2 was detected and
determined using a thermal conductivity detector.
The surface area and pore structure of catalysts were deter-
mined at liquid nitrogen temperature (196
C) with a Sorptomatic
1990 (Thermo Finnigan Italia S.P.A) instrument. Prior to the mea-
surements, the samples were degassed at 150
C under vacuum
conditions until static vacuum was reached. The BET (Brunauer,
Emmett and Teller) method was used to measure the surface area of
the catalysts. The BJH (Barrett, Joyner and Halenda) method was
employed to measure the pore distribution from the desorption
isotherm. The pore volume was calculated according to the
 A. Islam et al. / Energy 89 (2015) 965e973 967

Fig. 1. Apparatus for production of millimetric catalyst.

Gurvitch method. The analysis was automatically performed by the
Thermo Finnigan Sorptomatic 1900 series nitrogen adsorption/
desorption analysis software.
Elemental analysis on the reused catalyst was carried out by X-
ray uorescence in a Shimadzu EDX 720 spectrometer with a
rhodium X-ray source tube. The sample was placed on powders
placed inside the sample holder using a polypropylene lm. The
analysis were performed under vacuum conditions (<45 Pa) using
2-channels where the X-ray source was set at 50 and 15 kV for TieU
(Titanium eUranium) and NaeSc (SodiumeScandium) ranges,
respectively and the metal oxide content of the catalysts was
measured automatically by the instrument software.
2.4. Transesterication reaction conditions
In a batch process, all the reactions were performed in a 50 ml
bafed conical ask (Corning, USA) placed in an incubator orbital
shaker equipped with a temperature controller. In a typical assay,
palm oil 15 g, catalyst 0.6 g (4 wt.%, gcat/goil) were placed together
inside the conical ask. The molar ratio of oil and methanol was
adjusted as 1:14 according to the literature [22]. The reactions were
performed at 60
C over a period of 4 h. During the optimization
studies, the effect of time (1e6 h), molar ratio of methanol to oil
(8e18) and catalyst loading (0.6e0:33 gKI =ggAl2 O3 ) on the FAME
yield were investigated. The reaction mixture was stirred at an
agitation speed of 250 rpm to allow good contact between the
catalyst and reactants. After completion of the reaction, the catalyst
was recovered from the reaction mixture by simple ltration. The
ltrate was then centrifuged at a relative centrifugal force of 2500 g
for 10 min, and then excessive amount of methanol was evaporated
before analysis of biodiesel yield.
The fatty acid composition was determined by gas chromatog-
raphy using an internal standard. The samples were analyzed by a
Shimadzu GC-14B gas chromatograph equipped with a FID (ame
ionization detector) and a capillary column Rtx-65
(30 m  0.5 mm  0.25 mm). The injector and detector tempera-
tures were 240 and 280
C, respectively. Methyl heptadecanoate
(0.1 g) was dissolved in 100 mL of hexane to prepare the standard
solution. In all, 0.2 g of biodiesel sample was dissolved in 10 mL of
standard solution for GC analysis and 0.5 mL sample was injected
into the GC which uses helium as a carrier gas. The chromato-
graphic prole of the mixture of the standards and the biodiesel
product was shown in Fig. 2. The content in FAME (fatty acid methyl
ester) yield was determined in accordance with European regulated
procedure EN 14103 [17]. The experiments were duplicated and
data shown were the average values within 2.5% errors.
To calculate the biodiesel yield, the response factors, Rf for each
compound have been calculated using the correspondent standard
compound according to the Eq. (2).
   
Ais Crs
Rf  (2)
Ars Cis

Table 1
Impregnation of catalyst.

Step-1 Step-2 Step-3 Step-4 Step-5

Catalyst (g) for Catalyst (g) for Total weight of Volume of aqueous solution Calcination Total The calcinated catalyst,
1 g support 3 g of support catalyst and (ml) with catalyst and 3 g weight (g) g (catalyst support) used
support (g) of support was immersed for biodiesel production
(4 wt%, g cat./g oil)

0.06 3  0.06 0.18 3.18 4.5 Calcination of z2.5 0.6

0.15 3  0.15 0.45 3.45 4.5 impregnated catalyst 0.6
0.24 3  0.24 0.72 3.72 4.5 0.6
0.30 3  0.30 0.90 3.90 4.5 0.6
0.33 3  0.33 0.99 3.99 4.5 0.6
968 A. Islam et al. / Energy 89 (2015) 965e973
Fig. 2. Gas chromatograms of (a) standard methyl ester and (b) biodiesel product.
where, Ais area of internal standard, Cis concentration of in-
ternal standard, Ars area of standard references,
Crs concentration of standard references.
The ME (methyl ester) was calculated using Eq. (3):
Ciss  Aif  Rf
ME
Aiss
where, Ciss concentration of internal standard in the sample,
Aiss area of internal standard in the sample, Aif area of in-
dividual FAMEs compound in the sample.
The biodiesel yield (%) was according to the Eq. (4):
total amount of methyl ester mol
Biodiesel yield %
3  charged amount of triglycerols mol
 100
2.5. Catalyst reusability
Catalyst reusability test were performed by repeating the
transesterication over an extended period of time with used cat-
alysts. When the transesterication reaction was completed, the
catalyst was recovered by ltration, washed with water, dried at
100
C for 1 h to remove the moisture content and subsequently
reused. The same amount of palm oil and methanol were added to
the recycled catalyst each time to react under the same conditions.
This procedure was repeated several reaction cycles to examine the
extent of stability of the catalyst.
3. Results and discussion
(3)
3.1. Shape and size of g-Al2O3 beads
The optical photograph of the millimetric g-Al2O3 support beads
are shown in Fig. 3. As evidenced by Fig. 3, the beads shape was
close to a sphere with a diameter of 2 mm. The SF (sphericity factor)
of the beads was 0.02 0.01 which can be considered as a spherical
bead, as reported by Chan et al. [23].
3.2. Structure and basicity of catalyst
(4)
The XRD patterns of support materials g-Al2O3 and synthesized
catalysts with various loading amount are shown in Fig. 4. As can be
seen, curve (a) had a typical XRD pattern of g-Al2O3 at 2q values of
31.9
, 37.6
, 39.5
, 45.8
and 66.7
(JCPDS File No. 00-029-0063). In
the sample with a low loading of 0:06 gKI =ggAl2 O3 (curve b) had
similar patterns, identical to that of the g-Al2O3 (31.9
, 37.6
, 39.5
,
45.8
and 66.7
). Considering the low potassium loading amounts,
the absence of diffraction lines of potassium compounds may be
assigned to their high degree of dispersion in the form of a
 A. Islam et al. / Energy 89 (2015) 965e973
Fig. 3. Optical photograph of KI/g-Al2O3 catalyst.
monolayer on the g-Al2O3 surface, or to their amounts being
smaller than those detectable via the XRD technique [24]. An
alternative reason for the absence of diffraction lines of potassium
compounds may be due to the higher scattering factor of the K
cations, which is comparable to that of the Al cations in the cat-
alysts [24]. When the catalyst loading was increased to 0.15 gKI/gg-
Al2O3 (curve c), the characteristic peaks of KI (JCPDS File No. 00-023-
0493) (2q 21.7
, 25.2
) was observed implying that the amounts
of salts within its detectable limit of XRD. The diffraction lines
(curve d) assigned to K2O (JCPDS File No. 00-023-0493) at
2q 23.8
and 27.6
and KAlO2 phase (JCPDS File No. 00-002-
0897)) at 2q 38.9
appeared with the KI catalyst loading of 0.24 g/
g g-Al2O3. A similar observation was obtained in the case of KNO3/
Al2O3, where it was found that the alkali metal oxides and alkali-
aluminate crystalline structure were the catalytically active sites
to catalyze the transesterication reactions [7]. The KI peak could
still be detected at a loading of 0.24 g/g g-Al2O3 although the in-
tensities of these peaks were not as high as those observed for
loadings above 0.24 g/g g-Al2O3. Based on these results, it is
reasonable to state that only a part of the KI decomposed under
Fig. 4. XRD patterns for samples (a) g-Al2O3, (b) 0:06 g gKI =ggAl2 O3 , (c) 0:15 gKI =ggAl2 O3 ,
(d) 0:24 gKI =ggAl2 O3 , (e) 0:30 gKI =ggAl2 O3 , (f) 0:33 gKI =ggAl2 O3 .
969
calcination conditions at loadings 0.24 g/g g-Al2O3. A conclusion
has been reached by Xie and Li [3] that the KI supported alumina
catalyst can be decomposed at 500
C to form metal oxide phases,
which was referred as active sites for transesterication reaction.
These results also agree well with the results of KI/Al2O3 reported
by Xie and Zhao [25]. When the loading amount achieved
0:30 gKI =ggAl2 O3 , the characteristics peaks of g-alumina (JCPDS File
No. 00-029-0063) at 2q 66.7
and potassium iodide (JCPDS File
No. 00-004-0471) at 2q 21.7
, 25.2
, 35.9
, 42.4
, 44.3
and 58.3

was appeared in the diffraction patterns (curve e) and more intense
when the KI loading was increased to 0.33 g/g-Al2O3 (curve f). This
observations could be associated with the over loading of KI on to
the g-Al2O3 support. The over loaded KI masked the metal oxide
phase and appeared in the diffraction pattern at the loading of
0.30e0.33 g/g-Al2O3.
The strength and number of basic sites of the catalysts were
analyzed by CO2-TPD. In general, the stronger the basic site of a
catalyst, the higher the temperature is needed to desorb the
adsorbed CO2. It has been reported that the basicity of supported
basic catalyst is dependent on the presence of oxide or/and hy-
droxide groups on the catalyst surface [26]. The TPD proles of
desorbed CO2 on KI/g-Al2O3 catalysts are shown in Fig. 5. The re-
sults showed complex desorption proles, which could be due to
the presence of a variety of basic sites of different strength. The
desorption peak at a temperature of 147e169
C can be attributed
to the interaction of CO2 with the sites of weak basic strength. It has
been proposed that these sites are assigned to the presence of OH
groups on the catalyst surface [26]. Moreover, strong and sharp
peaks appeared at 299e312
C, which are characteristics of CO2
desorption from the medium strength of basic sites. Basic sites of
high strength, which is related to free O2 anions located in a
particular position of the mixed oxide surface are characterized by
the presence of desorption peak at 613e634
C [26].
The total number of basic sites of the catalyst is shown in Table 2.
The total basicity was found to increase with KI loading and reached
a maximum of 867 mmol/g at KI loading of 0:24 gKI =ggAl2 O3 . It was
reported by Chorkendorff and Niemantsverdriet [27] that the oxide
phases provide sufcient adsorptive sites for the probe molecules
with acidic character (CO2). Thus, the generation of active oxide
phases such as, K2O, KAlO2 phases on KI/g-Al2O3 catalyst at 0.24 g/g
Fig. 5. CO2-TPD spectra for samples (a) g-Al2O3, (b) 0:06 g gKI =ggAl2 O3 , (c)
0:15 gKI =ggAl2 O3 , (d) 0:24 gKI =ggAl2 O3 , (e) 0:30 gKI =ggAl2 O3 , (f) 0:33 gKI =ggAl2 O3 .
970 A. Islam et al. / Energy 89 (2015) 965e973

Table 2 Table 3
Effect of KI loading on basicity of KI/g-Al2O3 beads. The surface area and pore structure of KI/g-Al2O3 catalyst.

Catalyst KI loading on g-Al2O3 The amount desorbed CO2 (mmol/g catalyst) Catalyst KI loading on Surface area Pore volume Pore diameter
g-Al2O3 (m2/g) (cm3/g) (nm)
g-Al2O3 _ 19
KI/g-Al2O3 0.06 198 g-Al2O3 _ 349 1.38 11.2
0.15 318 KI/g-Al2O3 0.06 326 1.09 9.0
0.24 867 0.15 299 0.84 7.3
0.30 466 0.24 277 0.63 7.0
0.33 457 0.30 244 0.49 3.9
0.33 133 0.43 3.5

gAl2O3, as evident from XRD (Fig. 4; curve d) may desorb higher

CO2 and thus, increase the basicity. decreased when the KI loading was increased from 0.06 g to 0.33 g/
However, the decrement in basicity for both catalysts at loadings g g-Al2O3. This decrement of surface area has been previously
above 0.24 g/g g-Al2O3 was observed. These observations could be observed and attributed to the effect of catalyst deposition on the
due to the overloading of KI that saturated the g-Al2O3 surface. As a support, which resulted in partial blocking of the porous network
result, the dispersion of excess KI could mask the actives K2O, KAlO2 [29].
sites and incapacitate the function of active base site in the struc-
ture of catalyst at higher loadings. Therefore, the basicity of the
catalyst might be reduced. Similar observations have been found by 3.4. Catalytic activity
Tauq-Yap et al. [28] and reported that a decrease in basicity at high
catalyst loading was due to reduction of active bases in the struc- The activity of catalysts was evaluated in the transesterication
ture of catalyst. A mechanism has also been proposed by Xie et al. of palm oil with methanol in terms of FAME (fatty acid methyl
[7] to explain the decrease of basic sites with increasing catalyst ester) yield at different catalyst loadings, as shown in Fig. 7. When
loading. It has been suggested that the vacancies formed in the
alumina structure during the thermal treatment favours the
migration of metal K ions of KI salt to the vacant sites that sub-
sequently decompose to form basic sites. When the amount of salt
loaded on g-Al2O3 is more than the saturation uptake, it could not
likely be decomposed to create basic oxide sites on the catalyst [7].

3.3. Surface area and pore structure of catalyst

The nitrogen adsorptionedesorption isotherms and pore size

distributions of the g-Al2O3 support are shown in Fig. 6. The
isotherm corresponded to type IV according to the IUPAC classi-
cation and showed a characteristic hysteresis loop in the relative
pressure (P/P0) range 0.8e1.0 associated with capillary condensa-
tion taking place in mesopores (Fig. 6a). The pore size distributions
of the g-Al2O3 support are unimodal, with pore diameters centred
in the range 5.8e6.1 nm (Fig. 6b).
Table 3 shows the results of surface area the surface area, pore Fig. 7. Effect of catalyst loading on FAME yield. Reaction conditions: methanol/oil
volume and pore diameter of the catalysts at different loadings. As molar ratio 14:1, reaction time 4 h, reaction temperature 60
C, catalyst amount 0.6 g
evidenced by the data in Table 3, the surface area and pore volume (4wt.%, gCat./goil) and catalyst calcined at 500
C.

Fig. 6. N2 adsorptionedesorption isotherm (a) and BJH pore size distribution (b) of the millimetric g-Al2O3 beads.
 A. Islam et al. / Energy 89 (2015) 965e973 971

Fig. 8. Proposed mechanism of basic catalyst [27].

the catalyst loading increased from 0.06 to 0.24, the FAME yield was
Fig. 9. Effect of reaction time on FAME yield. Reaction conditions: methanol/oil molar
increased dramatically from 51 to 98 % due to a concomitantly ratio 14:1, catalyst loading 0:24 gKI =ggAl2 O3 , reaction temperature 60
C, catalyst
remarkable increase in surface basicity in the heterogeneous amount 0.6 g (4 wt.%, gCat./goil) and catalyst calcined at 500
C.
contribution of lower concentration of KI. Beyond the loading of
0.24, a signicant decrease in basicity to 466 mmol/g catalyst (0.3 g
KI/g Al2O3), although the catalyst maintained the higher FAME yield strength, stronger bases are in general more effective to initiate the
of 90%. This phenomenon indicate that the KI compound in the transesterication of triglycerides [34]. As per the TPD-CO2 result,
catalyst can dissolved readily into the polar phase (methanol) to the higher level of base strength (curve c in Fig. 5) at a loading of
form potassium methoxide as the homogeneous contribution of 0.24 g/g g-Al2O3 might result in higher biodiesel yield compared
catalyst, thus promising biodiesel yield due to its superior mass with the other loadings.
transfer characteristics. The porous properties were shown in Table 3. The mesoporous
The trend of the FAME yield can be related to the basic sites of characteristic (i.e. 7 nm in diameter) of the support beads at
metal oxides where the maximum basicity was achieved at 0:24 gKI =ggAl2 O3 loading might also permit access of bulky mole-
0:24 gKI =ggAl2 O3 (see Table 1). The basic sites of metal oxide can cules (i.e. triglycerides) to the active sites resulting substantial in
work synergetically with Lewis acid sites constituting acid-base increase catalytic activity [35]. However, pores with diameter of
pair sites [30]. It was reported by Chorkendorff and Niemantsver- less than 5 nm were occupied in the catalysts volume in the case of
driet [27] that metal oxide provides sufcient adsorptive sites for higher loading of KI (more than 0:24 gKI =ggAl2 O3 ). If, as some
alcohol in transesterication reaction and concluded that the high workers [36] have suggested, the triglyceride molecule can be
transesterication activity of catalyst might be due to the mani- diffused into catalyst having pore diameter of more than 5 nm then,
festation of the dissociation of alcohol to RO and H on basic sites diffusion of the triglyceride molecule could be restricted when the
of metal oxide catalyst surface (Fig. 8). Thus, the high yield of bio- diameter of the catalyst was less than 5 nm and this would in turn,
diesel could be attributed to the generation of oxides phases such be decreased the catalytic activity.
as, K2O, KAlO2 phases, as evident from XRD (curve d in Fig. 4) The effect of reaction time on the FAME yield was investigated,
formed at catalyst loading of 0:24 gKI =ggAl2 O3 , which can be as shown in Fig. 9. The FAME yield increased from 11% to 98% as the
considered as active sites for transesterication reactions, as pro- reaction time was increased from 1 to 4 h and remained constant
posed by Xie and Zhao [25]. A similar observation was obtained in thereafter, indicating an equilibrium FAME yield. Fig. 10 shows the
the case of KNO3/Al2O3, where it was found that the alkali metal effect of methanol/oil molar ratio on the FAME yield. According to
oxides and alkali-aluminate crystalline structure were the catalyt-
ically active sites to catalyze the transesterication reactions [7].
Other workers [31] reported that the formation of alkali metal
oxides and alkali-aluminate crystalline structure were the catalyt-
ically active sites for transesterication reaction, although Xie et al.
[32] suggested from their work that the formation of active oxide
phase largely depends on the proper metal loading during prepa-
ration of the supported catalyst.
However, the loadings may be reached above its saturation
uptake at higher than 0:24 g=ggAl2 O3 and thus, the decomposition
of catalyst might be precluded, as evident from XRD. Therefore, the
basicity as well as the catalyst activity might be reduced. This
observation is in agreement with the results reported by Boz and
Kara [33] who concluded from their work that the high basic
strength and basicity of the catalysts could improve trans-
esterication process. It can be inferred from the foregoing dis-
cussion that the formation of the oxide phase at the loading of
0.24 g/g g-Al2O3 may contribute to higher basicity (Table 2) as well
as catalytic activity towards the transesterication reaction (Fig. 7).
In addition, the biodiesel yield could also be affected by the level Fig. 10. Effect of methanol/oil molar ratio on FAME yield. Reaction conditions: reaction
of strongly basic sites present in the catalysts. As the ability of bases time 4 h, catalyst loading 0:24 gKI =ggAl2 O3 , reaction temperature 60
C, catalyst amount
to abstract a proton from a alcohol is directly connected to the base 0.6 g (4 wt.%, gCat./goil) and catalyst calcined at 500
C.
972 A. Islam et al. / Energy 89 (2015) 965e973

Table 4
Comparison of biodiesel yield.
g h
Catalyst Reaction condition BY (%) BP Reuse catalyst References

Oil type a
RT (
C) b
Cat. (g) Oil (g) c
M/O d
Rt (h) e
BD (g) Calcination
Cycle g
BY (%)

CT (
C)
f
t(h)

KI/Al2O3 Soybean MR 0.4 16 15:1 8 15.36 500 3 96 0.12 _ _ [25]

C4H4O6HK/ZrO2 soybean 60 1 18.22 16:1 2 17.86 512 5 98.03 0.49 5 89 [37]
KOH/Al2O3 palm 60 3 100 15:1 2 91 500 3 91 0.45 _ _ [38]
Ca(NO3)2/Al2O3 palm 60 1 10 65:1 3 9.4 450 4 94 0.31 _ _ [39]
KNO3/Al2O3 palm 60 1 10 65:1 3 9.47 550 4 94.7 0.31 _ _ [39]
KF/Al2O3 sunower 75 0.25 28 4:1 2 27.44 450 2 98 0.47 4 88 [40]
Egg shell Soybean 65 0.69 23 9:1 3 21.85 800 2 95.5 0.32 13 95 [41]
CaO/Al2O3 palm 65 3 50 12:1 5 49.32 718 5 98.64 0.19 2 91 [42]
KOH/iAC palm 64 5 50 24:1 1 49 __ __ 98 0.98 3 87 [43]
KI/gAl2O3 Palm 60 0.6 15 14:1 4 14.5 500 3 98 0.24 11 79 Present study

All of the reactions were performed under atmospheric pressure.

a
Reaction temperature.
b
Catalyst used for reaction.
c
Molar ratio of methanol to oil.
d
Reaction time.
e
Biodiesel.
f
Calcination temperature of the catalyst.
g
Biodiesel yield (%) biodiesel(g)/oil (g)  100.
h
Biodiesel productivity biodiesel (g)/oil(g) time.
i
Activated carbon.

reaction stoichiometry, 3 mol of methanol are required to react
with each mole of triglyceride, but in practice a higher molar ratio is
employed in order to drive the reaction towards completion and
produce more methyl esters as product. The yield was found to
increase considerably and reached a maximum when the meth-
anol/oil molar ratio was increased to 14, beyond which no further
increment in yield could be observed. The observations are in
agreement with previous reports [37].
The comparison of the FAME yield was compiled in Table 4. The
results showed that the biodiesel yield produced using the spher-
ical millimetric g-Al2O3 supported catalyst is comparable to that
reported in the previous studies. One of the main reasons to
consider heterogeneous catalysis for industrial processes is the ease
of catalyst separation after the reaction. Therefore, the easy sepa-
ration of catalyst from process uid and the safe handling are the
two of the main rationale to select the millimetric spherical catalyst
for biodiesel production.
3.5. Catalyst reusability
The stability of KI/g-Al2O3 catalyst was evaluated by performing
the reusability test on the catalyst. Fig. 11 presents the effect of
catalyst reuse on the FAME yield. The amount of the biodiesel yields
after the 11th cycle obtained in this work was 79%, which is com-
parable in biodiesel yield to that of the results obtained in previous
studies (see Table 4). The XRF results (Table 5) show that the
amount of K2O in the catalyst, after the rst and eleventh cycles,
was 24% and 19%, respectively. This could be due to the leaching of
K2O from the g-Al2O3 support. It was reported by several re-
searchers [7,25,39,44] that the activity of the alumina supported
catalyst such as KI, KF, HNO3, K2CO3, NaNO3, or KOH was reduced
after successive reuse due to the leaching of active species.
Although the catalyst was reused for 11 times, it showed good
mechanical stability as the support remained spherical with no sign
of disintegration (see Fig. 12). This indicates its potential for use in
industrial environment. However, the problem with metal leaching
from the solid supported catalyst is still a major consideration due
to the cost of recovery or replacement of the catalyst. A method of
controlling metal leaching in supported solid catalysis has been
proposed by Hunter et al. [45] who concluded from their work that
stronger bond with the metal complex formed by the inclusion of
legands could be controlled leaching of the active phase from a
solid catalyst. Other workers [46] have also discussed a similar
phenomenon and have suggested that the incorporation of

Table 5
Elemental analysis of reused KI/g-Al2O3.

Component Elemental composition (%)

1st Reuse 7th Reuse 11th Reuse

Al2O3 45.964 67.162 73.62

K2O 53.319 32.296 26.14
SO3 0.58 0.435 0.16
CuO 0.087 0.067 0.04
ZnO 0.05 0.04 0.04
Fig. 11. Reusability of KI/g-Al2O3 catalyst in transesterication of palm oil. Reaction
conditions: methanol/oil molar ratio 14:1, catalyst loading 0:24 gKI =ggAl2 O3 , reaction Reaction conditions: methanol/oil molar ratio 14:1, catalyst loading
temperature 60
C, reaction time 4 h, catalyst amount 0.6 g (4 wt.%, gCat./goil) and 0:24 gKI =ggAl2 O3 , reaction temperature 60
C, reaction time 4 h, catalyst amount
catalyst calcined at 500
C. 0.6 g (4 wt.%, gCat./goil) and catalyst calcined at 500
C.
 A. Islam et al. / Energy 89 (2015) 965e973
Fig. 12. Optical photograph of KI/g-Al2O3 catalyst after 11 cycle of successive reuse.
dendrimers into the solid supported catalyst could prevent active
phase leaching from the catalyst. Thus, issues related to the catalyst
stability and its leaching properties should be taken into consid-
eration for further study.
4. Conclusion
A mesoporous millimetric gamma-alumina bead supported KI
catalyst has been synthesized for transesterication of palm oil
with methanol. The highest FAME yield (i.e. 98%) was obtained at
the catalyst loading of 0:24 gKI =ggAl2 O3 where, the highest catalyst
basicity and mesoporous structure were observed. In fact, the re-
covery and recyclability of catalyst have been improved by using
millimetric gamma-alumina bead supported KI catalyst. In addi-
tion, the catalyst showed good reusability and mechanical stability
after being reused for 11 times. However, the rate of catalyst
deactivation is proportional to the concentration of metal oxide
that leached away from the support material in the course of re-
action. Thus, the research related to the controlling of leaching of
active species of the catalyst needs to address in future, as it may
affect the performance of catalyst. The further extension of this
approach could be the effect of catalyst size on the pressure drops,
poor mass/heat transfer, contact efciency properties together with
the design of reactor.
References
[1] Islam A, Tauq-Yap YH, Chu CM, Chan ES, Ravindra P. Process Saf Environ Prot
2013;91:131e44.
973
[2] Alhassan FH, Yunus R, Rashid U, Sirat K, Islam A, Lee HV, et al. Appl Catal A
2013;456:182e7.
[3] Xie W, Li H. J Mol Catal A Chem 2006;255:1e9.
[4] Asri NP, Machmudah S, Budikarjono K, Roesyadi A, Goto M. Chem Eng Process
Process Intensif 2013;72:63e7.
[5] Ganguly M, Sarma M. Res J Chem Environ 2011;15:386e8.
[6] Takase M, Zhang M, Feng W, Chen Y, Zhao T, Cobbina SJ, et al. Energy Convers
Manage 2014;80:117e25.
[7] Xie W, Peng H, Chen L. Appl Catal A 2006;300:67e74.
[8] Ma H, Li S, Wang B, Wang R, Tian S. J Am Oil Chem Soc 2008;85:263e70.
[9] Li H, Govind KS, Kotni R, Shunmugavel S, Riisager A, Yang S. Energy Convers
Manage 2014;88:1245e51.
[10] Feyzi M, Hassankhani A, Raee HR. Energy Convers Manage 2013;71:62e8.
[11] Islam A, Tauq-Yap YH, Chan ES, Moniruzzaman M, Islam S, Nabi MN. Energy
Convers Manage 2014;88:1200e18.
[12]] Cole-Hamilton DJ. Science 2003;299:1702e6.
[13] Centi G, Perathoner S. Catal Today 2003;79e80:3e13.
[14] Matatov-Meytal Y, Sheintuch M. Appl Catal A 2002;231:1e16.
[15] Wang Y, Zhang F, Xu S, Yang L, Li D, Evans DG, et al. Chem Eng Sci 2008;63:
4306e12.
[16] King CJ. Weinheim, electronic version 2007.
[17] Islam A, Tauq-Yap YH, Chu CM, Ravindra P, Chan ES. Renew Energy 2013;59:
23e9.
[18] Wang ZM, Lin YS. J Catal 1998;174:43e51.
[19] Islam A, Tauq-Yap YH, Chu CM, Chan ES, Ravindra P. J Porous Mater 2012;19:
807e17.
[20] Yang Z, Lin YS. Ind Eng Chem Res 2000;39:4944e8.
[21] Islam A, Tauq-Yap YH, Ravindra P, Moniruzzaman M, Chan ES. Adv Powder
Technol 2013;24:1119e25.
[22] Akbar E, Binith N, Yaako Kamarudin ZSK, Salimon J. Green Chem 2009;11:
1862e6.
[23] Chan ES, Lee BB, Ravindra P, Poncelet D. J Colloid Interface Sci 2009;338:
63e72.
[24] Xie W, Yang Z, Chun H. Ind Eng Chem Res 2007;46:7942e9.
[25] Xie W, Zhao L. Energy Convers Manage 2013;76:55e62.
[26] Cantrell DG, Gillie LJ, Lee AF, Wilson K. Appl Catal A 2005;287:183e90.
[27] Chorkendorff I, Niemantsverdriet JW. Weinheim, Germany: Wiley-VCH; 2003.
[28] Tauq-Yap YH, Lee HV, Yunus R, Juan JC. Chem Eng J 2011;178:342e7.
[29] Melero JA, Stucky GD, Van Grieken R, Morales G. J Mater Chem 2002;12:
1664e70.
[30] Zecchina A, Lamberti C, Bordiga S. Catal Today 1998;41:169e77.
[31] Evangelista JP, Chellappa T, Coriolano AC, Fernandes Jr VJ, Souza LD, Araujo AS.
Fuel Process Technol 2012;104:90e5.
[32] Xie W, Wang H, Li H. Ind Eng Chem. Res. 2012;51:225e31.
[33] Boz N, Kara M. Chem. Eng. Commun 2009;196:80e92.
[34] Sree R, Seshubabu N, Saiprasad P, Lingaiah N. Fuel Process Technol 2009;90:
152e7.
[35] Cejka J, Mintova S. Catal Rev 2007;49:457e509.
[36] Granados ML, Poves MDZ, Alonso DM, Mariscal R, Galisteo FC, Tost RM, et al.
Appl Catal B Environ 2007;73:327e36.
[37] Qiu F, Li Y, Yang D, Li XX, Sun P. Bioresour Technol 2011;102:4150e6.
[38] Noiroj K, Intarapong PP, Luengnaruemitchai AA, Jai-In SS. Renew Energy
2009;34:1145e50.
[39] Benjapornkulaphong S, Ngamcharussrivichai C, Bunyakiat K. Chem Eng J
2009;145:468e74.
[40] Verziu M, Flore M, Simon S, Simon V, Filip P, Parvulescu VI, et al. J Catal
2009;263:56e66.
[41] Wei Z, Xu C, Li B. Bioresour Technol 2009;100:2883e5.
[42] Zabeti M, Wan-Daud WMA, Aroua MK. Fuel Process Technol 2009;90:770e7.
[43] Baroutian S, Aroua MK, Abdul Raman AA, Sulaiman NMN. Fuel Process
Technol 2010;91:1378e85.
[44] Alonso DM, Mariscal R, Tost RM, Zafra Poves MD, Granados ML. Catal Commun
2007;8:2074e80.
[45] Hunter DL, Moore SE, Willis GG. U.S. Patent No. 4,647,708. Washington, DC:
U.S; 1987.
[46] King AS, Twyman LJ. J Chem Soc Perkin Trans 2002;1(20):2209e18.