You are on page 1of 8

Fuel 89 (2010) 20542061

Contents lists available at ScienceDirect

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

Kinetics of biodiesel synthesis from sunower oil over CaO heterogeneous catalyst
Dj. Vujicic a, D. Comic b, A. Zarubica a, R. Micic b, G. Boskovic a,*
a
Faculty of Technology, University of Novi Sad, Cara Lazara 1, Novi Sad, Serbia
b
NIS Petrol, Novi Sad Oil Renery, Put Sajkaskog Odreda 4, Novi Sad, Serbia

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

Article history: Calcium oxide as a heterogeneous catalyst was investigated for its effect on the biodiesel synthesis from
Received 3 April 2009 rened sunower oil. Experiments were carried out using a commercial bench stirred tank reactor of
Received in revised form 28 November 2009 2 dm3 volume, at 200 rpm, with a methanol to oil ratio 6 to 1 and 1 mas.% catalyst loading as constant
Accepted 30 November 2009
parameters. Ester yields were followed as a function of temperature (60120 C), pressure (115 bars)
Available online 4 January 2010
and reaction time (1.55.5 h). The temperature of 100 C was found to be optimal for the maximum
(91%) conversion to methyl esters, while pressure had a positive impact up to 10 bars at 80 C. The cat-
Keywords:
alyst activation in air leading to the formation of strong basic sites was found to occur at 900 C. Catalyst
Biodiesel
Heterogeneous catalysis
particle coalescence took place during the reaction, giving a gum-like structure, and resulted in a signif-
Calcium oxide icant catalyst deactivation. The pseudo-rst order reaction was established, with a knee at 80 C in the
Transesterication of sunower oil Arrhenius plot separating the kinetic and diffusion regimes. During the reaction progress, an activation
Diffusion restrictions energy decrease from 161 to 101 kJ/mol, and from 32 to (3) kJ/mol, was found for the kinetic and dif-
fusion regimes, respectively.
2009 Elsevier Ltd. All rights reserved.

1. Introduction rication process utilizing a heterogeneous catalyst. Therefore the


majority of contemporary investigations in the domain of biodiesel
Due to ecological, geo-political and economical reasons, orien- production are oriented towards the discovery of suitable solid cat-
tation to biodiesel usage is becoming an inevitable part of sustain- alysts. Benets that are looked-for are shorter contact times, lower
able economy. Removal of waste greases, net reduction of CO2 reaction temperatures, optimal oil/alcohol molar ratios and lower
emissions and the related Green house effect, biodegradability, catalyst loading, all resulting in higher end product yields of higher
reduction of the dependence on oil import, orientation to domestic purity. Hence, several conventional solid bases and acids (with and
resources, balancing of national energetic policies, are all reasons without promoters/dopants) have been investigated in this sense
for the fact that biodiesel has become an important element of thus far [1,2], as well as catalysts representing immobilized en-
the legislative policies in EU countries. A minimal obligatory frac- zymes (lipases) on suitable solid supports [3]. Thus, the corre-
tion of biodiesel in total diesel yearly consumption is dened in sponding activity of oxides, methoxides and hydroxides of
most EU countries. For non-member countries, the EU assumes alkaline-earth metals have been determined, but the reaction con-
the acknowledgment of various EU eco-standards, and among are ditions in these investigations were conned to atmospheric pres-
these connected to the use of biodiesel. sure and temperatures up to the boiling points of the applied
Technology of biodiesel production by transesterication reac- alcohols [4]. Among alkaline-earth metal oxides only MgO and
tion has reached its maturity during almost one century long usage CaO should be noteworthy, due to BaO solubility in methanol
of homogeneous catalysts. However, numerous advantages of het- and SrO tendency to react with CO2 and water [5]. In the case of
erogeneous over homogeneous catalysis are generally known. In Ca-based materials, only the oxide, but not the hydroxide and car-
the particular case these advantages can be classied as: (a) eco- bonate, has shown considerable catalytic activity in transesterica-
logical elimination of the washing section and huge amounts of tion of soybean oil with reuxing methanol [6].
waste water; easier disposal of spent solid catalyst; (b) economical In this work the ability of CaO as a potential heterogeneous cat-
cheaper reusable catalyst, production of glycerol and the end alyst for biodiesel synthesis from rened cooking sunower oil was
product of high purity; and (c) investment-type process simpli- investigated in a broad range of reaction temperatures and pres-
cation and elimination of entire sections from the process schemes. sures. Kinetic parameters including the activation energies were
At the moment, however, there is only one commercial transeste- calculated. A correlation of ester yields to the physicochemical
properties of differently activated catalyst samples was ascer-
* Corresponding author. Tel.: +381 21 485 3629. tained in order to determine those characteristics that make CaO
E-mail address: boskovic@uns.ac.rs (G. Boskovic). a suitable catalyst for biodiesel production.

0016-2361/$ - see front matter 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.fuel.2009.11.043
Dj. Vujicic et al. / Fuel 89 (2010) 20542061 2055

2. Experimental part ues, kinematic viscosity, acid number, cetane index and CFPP (bio-
diesel), were determined according to the appropriate standards:
2.1. Raw materials characterization and catalyst preparation JUS EN ISO 3675:1988, EN 14111:2003, JUS E K8.028, JUS ISO
3104:2003, EN 14104:2003, EN ISO 5165: 1998 and EN 116,
Fatty acids prole and other characteristics of the rened sun- respectively. Unpuried methanol (Metanolno-sircetni kombinat,
ower oil of a local producer employed for the transesterication Kikinda, Serbia) was used in the experiments with the following
reactions are given in Table 1. A typical gas chromatogram showing characteristics declared by the producer (in mas.%): water 0.05,
the fatty acid prole of the applied sunower oil is presented in free acids 0.003, formic acid 0.0023, free alkali 0.0003 and sum of
Fig. 1. The oil selection was based on the fact that sunower repre- aldehydes and ketones 0.003. Calcium oxide (Technical grade, Cen-
sents a typical oleaceous plant from the region. Besides this fact, trohem, Serbia) was used as the catalyst precursor, following its
the rened oil was of interest due to its low fraction of free fatty calcination in the static air conditions. The powder was calcined
acids that is known to limit base catalyzed transesterication. at 500 and 900 C, and then pressed into the form of pallets. Finally,
The physical and chemical properties of the feedstock as well as the pallets were crashed to smaller particles and sieved to obtain
of the obtained biodiesel density, iodine and saponication val- grains in the range of sizes 11.5 mm. This dimension range was
selected as a compromise taking into account the catalyst size
promising no internal extrudate diffusion restrictions and guaran-
tying safe maintenance of the catalyst grains within the perforated
Table 1
Characteristics of the feedstock and methyl esters in the product. reactor basket.

Property Sunower oil Methyl esters


of commercial obtained after 2.2. Apparatus and reaction procedure
quality 2.5 h at 100 C
Density (at 15 C), kg/m3 922.2 888.4 Biodiesel synthesis (fatty acids methyl esters, FAME, synthesis)
Refraction index 1.467 experiments were performed in a high pressure commercial labo-
Iodine number, g I2/100 g 132
ratory stirred reactor (Parr 4520) of 2 l in volume, coupled with a
Saponication value (mg KOH/g) 194 200
Kinematic viscosity (at 40 C) mm2/s 33.4 4.4 unit for control of pressure, temperature (thermocouple J type)
Acid number (mg KOH/g) 0.5 0.3 and impeller rotation. An integrated part of the reactor is a metallic
Cetane index (CI) 44 basket made out of two perforated concentric cylinders with the
Blur point, C 1 catalyst being loaded in between, and a strong mixer positioned
CFPP, C 3
Fatty acids composition, mass%
in the inner cylinder. Such a construction allows intensive convec-
Palmitic, C16:0 7.1 6.8 tion of reactants through the catalyst loading minimizing mass
Stearic, C18:0 4.8 3.7 transfer restrictions. The methanol: oil = 6:1 M ratio and the cata-
Oleic, C18:1 22.6 25.9 lyst loading of 1 mas.% (relative to total mass of reactants) of CaO
Linoleic, C18:2 65.5 63.2
were kept constant.
Linolenic, C18:3 0.1
Gadoleic, C22:1 0.2 Layout of a typical experiment was as follows: the required vol-
Behenic, C22:0 0.1 umes of methanol and oil were poured into the vessel, and after
Saturated 11.9 10.6 the basket with the preloaded catalyst was positioned into the
Monounsaturated 22.6 26.1 reactor its cover was closed. The reactor was then purged with
Polyunsaturated 65.5 63.3
DU 152.7
the stream of N2 in order to extrude the remaining air, and follow-
ing that the reaction vessel was heated to the desired temperature

Fig. 1. Gas chromatogram of the fatty acid prole of the used sunower oil.
2056 Dj. Vujicic et al. / Fuel 89 (2010) 20542061

starting from ambient temperature, with a constant temperature desorption isotherms. Related data were obtained by dynamic
ramp of 2 C/min. At zero reaction time, to = 0, when the corre- low-temperature adsorption/desorption of N2 (LTNA) using He as
sponding temperature was reached, the impeller was started. In carrier gas on a Micromeritics ASAP 2010. Scanning Electron
such a way there was no reaction progression before the zero reac- Microscopy (SEM) (JEOL JSM-6460LV, accelerating voltage of
tion time was reached. Due to the high viscosity of the reaction 25 kV) was used for the investigation of surface morphology of cat-
mixture, as well as the mechanical restrictions of the catalyst, alysts samples. Before being submitted to SEM characterization,
the mixing rate was limited to 200 rpm. A blank run preceding the solid samples were coated with gold in order to achieve suf-
the catalyst testing, done at harsh reaction conditions (120 C, cient conductivity. Acidbase character of the catalyst samples
5 h reaction time), proved that the transesterication reaction in were examined by Fourier Transformed InfraRed spectroscopy
question does not occur in the absence of the CaO catalyst. (FTIR) using Nicolet 5700. Prior to the IR analyses the samples were
Two sets of experiments were performed providing information degassed and exposed to phenol vapors and then to vacuum in or-
on the correlation of the biodiesel yield as a function of tempera- der to remove the physically adsorbed phenol.
ture and pressure, separately. The temperature dependence ori-
ented experiments (isothermally conducted at the given
temperature from the 60 to 120 C temperature range) consisted 3. Results and discussion
of ve independent runs differing in reaction times: 1.5; 2.5; 3.5;
4.5 and 5.5 h. Reaction aliquots were taken at the end of every In order to investigate the inuence of temperature of the cata-
run. During these experiments pressure was kept deliberately con- lyst activation on biodiesel yield two batches of catalysts were pre-
stant, although at temperatures above the boiling point of metha- pared differing only in the calcination temperature: 500 and
nol, a pressure increase was evident. Pressure inuence oriented 900 C. Unexpectedly low activity observed in transesterication
experiments were performed in the range 515 bar, by utilizing reaction at 100 C and at atmospheric pressure for the sample acti-
nitrogen gas as the ballast up to the required total pressure. Con- vated at lower temperature (Fig. 2) may be related to thermal
stant reaction temperature of 80 C was maintained in the course properties of these materials. Namely, in the thermograph of the
of these runs as the optimal one based on results of the previously fresh CaO sample given in Fig. 3 two endothermal effects exist.
mentioned set of experiments. Sampling was done continuously The rst one at 480 C indicates the decomposition of Ca(OH)2
during the reaction, and due to the pressurized reaction conditions [7], known to perform less efciently in the transesterication
a special probe tube positioned at the very bottom of the equip- reaction in comparison to its counterpart oxide [4,6,8]. A much less
ment was used. That is to say, samples were taken at 1 h time pronounced endothermic effect at 740 C can probably be attrib-
intervals (after 1.5; 2.5; 3.5; 4.5 and 5.5 h) during a single run, uted to the remaining CaCO3 decomposition [7]. Both effects are
assuming that the volume of the aliquot was such as not to alter followed by the corresponding changes in TGA prole shown in
the catalyst/oil/methanol ratio. the same gure (Fig. 3), accounting for the loss of about 10.5 and
1.5 mas.%, respectively. Based on these thermogravimetric proper-
2.3. Product characterization and yield calculation ties it seems that the applied lower activation temperature lays
close to the line delimiting the true-highly active catalyst from
In order to determine the reaction yields products form a single the less active, or inactive-contaminated form, i.e. the fraction of
run were cooled to 55 C, ltrated and transferred to a funnel for hydrates that have been previously shown as being much less ac-
separation. Separated layers of biodiesel (esters and unreacted tri- tive [4,6] and carbonates with completely no activity, both in
glycerides) and glycerol with the catalyst traces were visible yet transesterication of soybean oil [6]. Although results of the ther-
after 15 min, however full separation was achieved after 16 h. Ef- mal analysis presented in Fig. 3 suggest that 500 C should be the
ciency of the catalyst was ranked as yield calculated according to temperature high enough for the complete dehydration of the
following formula: hydroxide, it is obvious that in reality this process requires higher
temperatures.
Mass of ester layer after separation In contrast to the substantial differences in the catalytic activity
Yield%  Purity%
Total mass of reactants at reaction start and sample compositions, textural properties remain almost
unaffected by the calcination temperature under investigation
where Purity is fraction of esters in the biodiesel layer obtained by (Table 2). There is an exception when mean pore diameters are
GC analysis. The 0.6 ll aliquots were diluted with chloroform and
obtained solutions were analyzed by gas chromatography (PE Auto- 100
system XL) according to the method SRPS EN 14103. The conditions 500 C
of GC analysis were as follows: 30 m long capillary column (0.3 lm 900 C
layer of Polyethylene Glycol) and 1.8 ml/min of He as carrier gas 80
were used for components separation at 210 C for 10 min, and
Total ester fraction, %

ame ionization detector was used for their detection. A standard


mixture of methyl esters was used for qualitative analysis and 60
methyl heptadecanoate (above 99%, Fluka) was used as the internal
standard for quantication purposes.
40

2.4. Catalyst characterization


20
Detailed physicochemical characterization of catalyst samples
included the study of thermal, textural, morphological and surface
0
properties. Thermal (TA) and thermo gravimetric analyses (TGA)
1.5 2.5 3.5 4.5 5.5
were performed in the static air conditions 201000 C, using Der-
Reaction time, h
ivatograph MOM, M-1000. Textural characteristics were investi-
gated by means of surface area determined by BET, as well as by Fig. 2. Fraction of esters in biodiesel obtained at 100 C and atmospheric pressure
mean pore diameter and pore volume, both determined using as a function of catalyst activation temperature (500 and 900 C) and reaction time.
Dj. Vujicic et al. / Fuel 89 (2010) 20542061 2057

Fig. 3. TA and TGA proles of a fresh CaO sample in temperature range 201000 C.

Table 2 exposed to phenol vapors. This molecule showing weakly acidic


Textural properties of CaO catalyst activated at different temperatures. properties is known as a good probe for the identication of basic
Calcination BET surface Mean pore Total pore active sites on the catalyst surface [11]. In the case of MnO2, for
temperature (C) area (m2/g) diameter (nm) volume (cm3/g) example, dissociative phenol adsorption is assumed, most proba-
500 4.6 12.6 0.01 bly as H+ and C6H5O on OH and Mn, respectively [12]. As shown
900 4.3 14.6 0.02 in Fig. 4, there is a substantial difference in the IR spectra of phenol
adsorbed on CaO surfaces of the samples with different thermal
history. Namely, bands in the region 12601180 cm1, attributed
concerned, being slightly lower in value for the sample activated at to OH-groups directly bonded to the phenol aromatic ring [13],
lower temperature. However, pores in both activated samples be- are exclusively present in the lower calcination temperature pre-
long to the mesopores class with their dimensions being one order treated sample. The lack of these bands in the sample pretreated
of magnitude larger than the triglyceride dimensions [9], thus any at 900 C indicates no phenol adsorption, but instead probably a
signicant diffusion restriction should not be expected. Therefore dissociative adsorption of H+ and C6H5O type occurred. Therefore
reasons for the tremendous variation in the catalytic efciency of we suggest that only weak basic sites are present on the surface of
the samples activated at different temperatures have to be looked the low-temperature activated sample in the form of Ca2+ adsorb-
elsewhere (other CaO properties). ing phenol, while on the surface of the well dehydrated CaO (cal-
As already stated the catalyst samples with various thermal his- cined at 900 C) strong basic sites consist of O2 reacting with
tories perform rather differently in the transesterication reaction. the H+ originating from the dissociated phenol. This is similar to
It is obvious forasmuch as the total ester yields are concerned, but the earlier ndings on the basic character of CaO [11]. It was also
also comparing these at different reaction times (Fig. 2). Thus for previously claimed that the required calcination temperature to
example, the ester yield of a 1.5 h long run is higher by a factor generate maximum amounts of electron donating sites on CaO sur-
of 15 for the sample activated at 900 C, however, this decreases face was 700 C, while heating to 900 C was necessary when
to a factor of 5 for the experiment which lasted for 5.5 h. This indi- CaCO3 was to be used as a CaO precursor [14].
cates that the reaction proceeding over the lower temperature cat- The absence of the band at 1647 cm1, usually assigned to the
alyst attains the steady-state rather slowly, or at least much slower bending of OH-groups of physisorbed water molecules, on both
compared to the higher temperature catalyst. One may speculate samples regardless of their calcination temperature, indicates that
that the reason lays in the slow formation of an active phase of the catalyst deactivation is not caused by water [9]. The former is
the catalyst once it contacted with the reaction mixture. Namely, in line with TG results showing less than 2 mas.% of water content,
if the formation of a (coordinated) methoxide ion and of a proton- the amount reported earlier as the limiting for catalyst deactiva-
ated catalyst are required as the rst step of the reaction, as was tion [5]. The symmetric and asymmetric stretching vibrations of
suggested before [6,10], an induction period may be also expected, OCO bonds of unidentate carbonate at the surface of CaO cal-
followed by the reaction rate increase as reactants merge more in- cined at lower temperature are conrmed by IR bands at about
tensely in the course of the biodiesel production. Anyhow the re- 1475, 1074 and 864 cm1 [9], while the same are absent in the case
sults do not conclusively indicate that the active species in the of the sample activated at higher temperature (Fig. 4). This is also
two catalyst samples differ qualitatively, but in their quantity, in accordance with the results of TG analysis (Fig. 3) and catalyst
which may be further elucidated based on FTIR results. activity results observed in this and some previous work [6]. Based
Knowing that the acid/basic character of the catalyst is the cru- on the discussed FTIR results and literature data advocating higher
cial one for the transesterication reaction, the samples calcined at catalyst calcination temperature further kinetic investigations
different temperatures were submitted to FTIR analysis after being were performed only on CaO catalyst sample activated at 900 C.
2058 Dj. Vujicic et al. / Fuel 89 (2010) 20542061

(a) 60
55

50

45

40

2847,67

702,38
2917,14
%T

1294,89
3060,16
3019,30
35

763,68
1021,11

820,89
865,83
1074,23

992,51
1168,22
3517,82

30
3403,41

1597,28

1262,20
25

555,28
20

518,50
465,38
1417,48
1478,77
3640,41

15

10
4000 3500 3000 2500 2000 1500 1000 500
-1
Wavenumbers (cm )

(b) 50
48
46
44
42
40
38 1270,37

1119,18
36

1000,68

874,01
2847,67
2917,14

34
32
%T

1593,19

30
3432,01

28
26
1462,43

1413,39

24
22
20
18
3640,41

16
14
12
10
4000 3500 3000 2500 2000 1500 1000 500
-1
Wavenumbers (cm )

Fig. 4. FTIR spectra of fenol adsorbed CaO catalysts samples precalcined at different temperatures: (a) at 500 C, (b) at 900 C.

Transesterication reactions which are mostly performed at peratures of 80 and 100 C are signicantly affected by the dura-
atmospheric pressure are limited by the reaction temperature tion of reaction run, and high steady-state conversions were
which must not surpass the methanol boiling point. Otherwise reached after 5.5 and 2.5 h, respectively. Finally, the results at
extensive methanol evaporation occurs, followed by a reaction 120 C are noteworthy since the maximal conversion is reached
mixtures deprivation of methanol and considerable decrease of even at, or earlier than the minimal reaction time applied in the
the biodiesel yield. Therefore the usage of both higher reaction investigation; however, the yield achieved is signicantly lower
temperature and pressure are a real advantage and are feasible than the one accomplished at both 80 and 100 C. This ca. 6% lower
when transesterication reactions are performed in an autoclave. conversion of triglycerides indicates unfavorable conditions for the
In Fig. 5 FAME yields with time-on-stream at different reaction catalyst operation, probably being considerably limited by internal
temperatures are presented. The lowest applied temperature of or/and external diffusion, or a competitive reaction occurring at
60 C, the only one which is below the boiling point of methanol, higher temperatures.
turned out to be insufcient (enough) for a signicant conversion Fig. 6 depicts the inuence of the reaction pressure on the bio-
of the triglycerides in the presence of CaO regardless of the applied diesel yield at optimal reaction temperature (80 C). Under the
reaction time. In contrast, the achieved activities at reaction tem- investigated pressures, coupled with the applied reaction temper-
Dj. Vujicic et al. / Fuel 89 (2010) 20542061 2059

100 oil [16], exceeding, however, the iodine number limit imposed by
EU (UNE-EN 14214). Viscosity is one of the most important param-
80
eters determining the quality of diesel fuels in terms of their ability
to be atomized and injected into the engine, particularly at lower
Ester fraction, %

60C temperatures. The viscosity values shown in Table 1 agree with the
60 80C fact that the value of biodiesel fuel viscosity is about an order of
100C magnitude lower than that value of the parent oil. In this particular
40 120C case, however, a more signicant viscosity reduction did not occur
upon transesterication due to slightly increased content of mono-
unsaturated (C18:1) and decreased content of unsaturated (C18:2)
20
acids in the FAME relatively to the feedstock [15,17].
The cetane number (CN) is also an important indicator of the
0 ignition quality of diesel fuels and the obtained relatively low value
0 1 2 3 4 5 6
Duration of reaction, h is in the usual range for methyl esters originating from this partic-
ular starting material [17,18]. Since predictions of CN are usually
Fig. 5. Efciency of transeterication reaction with CaO activated at 900 C as a not applicable to non-conventional diesel fuels such as biodiesel
function of reaction temperature and reaction time. [19], in the particular case cetane index (CI), based on saponica-
tion and iodine values [20], is reported here as an alternative
non-engine value characterizing combustion properties of bio-
100 diesel. Thus, the CI value lower than 46 (Table 1), which has been
established as the minimal standard, could be associated with a
considerable fraction of unsaturated fatty acids, i.e. particularly
80 high fraction of highly unsaturated components such as linolenic
acids. On the other hand, good low-temperature properties pre-
Ester fraction, %

60 sented as CFPP (Table 1) are due to a signicant fraction of mono-


1 bar unsaturated components among total unsaturated compounds.
5 bar Namely, oleic acid acts as solvent in which saturated esters are dis-
40 solved [21].
10 bar
15 bar 3.1. Kinetic approach
20

Rate law of the transesterication reaction can formally be ex-


0 pressed as:
0 2 4 6
Time-on-stream, h dTG
r k  TG  ROH3
Fig. 6. Esters yield as a function of reaction pressure and time-on-stream.
dt
based on previously reported mechanistic consideration [21]. Three
mole-equivalents of the alcohol (ROH, MeOH in this case) react ini-
atures, methanol would not come to a boil. Results summarized in tially with a triglyceride (TG), and through three consecutive re-
Fig. 6 show substantial differences in the biodiesel yield (smaller verse (equilibrium) reactions convert it in a stepwise manner to a
triglyceride conversion) which were achieved in short runs, i.e. diglyceride, monoglyceride, and nely to free glycerol. One mole-
those that lasted from 1.5 to 3.5 h. After 4 h of time-on-stream, equivalent of the corresponding ester is produced in every step of
however, the biodiesel yield was not any longer a function of pres- these reversible reactions. Strictly speaking this overall conversion
sure in the investigated interval. It seems that the pressure inu- should follow a forth order reaction rate law, however due to a huge
ence upon the studied set-up in these investigations was not surplus of the transesterifying alcohol in the reaction mixture the
one-sided. That is to say, the triglyceride conversion as a function reaction could be safely considered as obeying the pseudo-rst or-
of pressure reaches a maximum at 10 bar, while further pressure der kinetics [22,23].
increase has a negative impact on it. Based on the values biodiesel yield and regarding them a as the
The physical and chemical properties of a typical batch of the function of reaction temperature and time (Fig. 5), the correspond-
produced biodiesel are given in Table 1, in parallel to the same of ing rate constants were calculated (Table 3). Zero reaction time and
the feedstock. Similar fatty acids compositions of the feedstock the particular duration of a single reaction (run) (1.55.5 h) were
and product are in line with the well known fact that biodiesel taken as the integration limits. In addition, partial rate constants
quality depends only on the feedstock nature. In particular, both at particular temperatures, assuming the consecutive increments
are characterized with high fraction of unsaturated acids, resulting of reaction time as integration limits, were calculated and are pre-
in high degree of unsaturation (DU) value calculated for the prod- sented in Table 4. In both cases the (pseudo) rst order reaction
uct (Table 1). In addition, similar properties among the transesteri-
cation outcomes of the performed runs regardless of the
conditions applied are due to fact that neither the catalyst nor Table 3
the reaction settings favor the production of any specic type of es- Reaction kinetic constants as function of reaction temperature and of experiment
duration, (103 min1).
ters. That is to say catalyst activity, and not its selectivity, is the
parameter which has to be considered. As is known, the fatty acid Total reaction time (h) k (60 C) k (80 C) k (100 C) k (120 C)
composition of the feedstock plays an important role in some crit- 1.5 2.67 73.71 175.40 220.86
ical biodiesel parameters, such as the viscosity, CN and cold ow 2.5 3.08 88.09 140.77 131.96
properties [15]. Obviously, some characteristics of the biodiesel 3.5 3.26 81.81 100.41 94.25
4.5 5.32 77.68 76.94 72.79
product are questionable; e.g. iodine number which is on the bor-
5.5 8.30 65.63 61.00 59.22
der value characteristic for biodiesels originating from sunower
2060 Dj. Vujicic et al. / Fuel 89 (2010) 20542061

Table 4 0
Reaction kinetic constants as the function of reaction temperature and time-on- 0.0025 0.0026 0.0027 0.0028 0.0029 0.003 0.0031
stream (103 min1).
-1
Time-on-stream (h) k (60 C) k (80 C) k (100 C) k (120 C)
1. h
1.5 2.67 73.71 175.40 220.86 2. h
-2
2 3.69 109.67 88.83 1.39 3. h
3 3.70 66.12 0.47 0 4. h
4 12.55 63.21 5.20 2.31 -3 5. h
5 21.74 11.41 10.78 1.85

ln k
-4

rate and the stoichiometry requiring one third mole of a triglycer-


-5
ide for one mole of esters was assumed. As can be seen from Table
3 the maximal reaction rate constant was achieved at different
-6
reaction times depending on the applied reaction temperature.
That is to say at 60 C the reaction constant had the greatest value
when the reaction was allowed to run for 5.5 h, while at 100 and -7
120 C the maximal reaction rate constant was achieved during 1/T, K
the shortest applied reaction period. The results imply that the
Fig. 8. Arrhenius graph as function of time-on-stream.
transesterication reaction occurs stepwise, with the rst period
possibly representing the time needed for the formation of an ac-
tive phase of the catalyst. According to Kouzu et al. an induction time-on-stream assuming time limits of 1 h increments, which
period prior to the transesterication reaction may be expected conrm the previous ndings. Thus, the highest reaction rates
due to catalytic active sites formation [6]. Accordingly, the forma- are achieved at the fth and second hour for reactions performed
tion of these catalyst active sites was not completed at 60 C not at 60 and 80 C, respectively, and during the rst hour for the reac-
even for the longest performed reaction time. Thus, the results pre- tion runs at 100 and 120 C.
sented in Fig. 5 indicate the obtained conversion of less than 5% at In Figs. 7 and 8 Arrhenius plots for the above calculated rate
60 C for 5.5 h of the reaction time. In contrast, at higher tempera- constants at the investigated temperature interval are presented.
tures worked at in this work, the maximal rate constants corre- Families of curves with characteristic slopes dene activation ener-
spond to those runs which lasted for the shortest time period, i.e. gies as a function of the reaction progress. As seen from Fig. 7 there
only 1.5 h. Data in Table 4 show rate constants as a function of is a knee indicating a change of reaction regime at the steady
temperature point regardless of the reaction time. In the kinetic re-
gime, which dominates over the temperature interval 6080 C,
0 the activity function of temperature is lower as the reaction time
0.0025 0.0026 0.0027 0.0028 0.0029 0.003 0.0031 is prolonged. This is accompanied by a drop of activation energy
-1 from 162.1 kJ/mol to 101.0 kJ/mol following the increase of the
1.5h reaction time from 1.5 to 5.5 h. These activation energy values
2.5h are in line with the similar ones found previously for the rapeseed
-2
3.5h oil [24,25]. The same trend is valid for the diffusion controlled re-
4.5h gime starting at 80 C upwards, accompanied by a drop in the acti-
-3 5.5h
vation energy from 31.9 kJ/mol to actually 3.0 kJ/mol.
ln k

A more detailed kinetic analysis based on the reaction progress


-4 with time-on-stream (Fig. 8) conrms the negative inuence of the
reaction time on biodiesel yield. Namely, the absence of points in
-5 the Arrhenius plot corresponding to longer runs performed at high-
er temperatures, even with a negative slope already in the kinetic
-6 regime obtained for the nal period of the reaction (5th h), are due
to the decreasing values of the reaction rate constants accompany-
-7 ing the reaction progress. These were the result of an increase of
1/T, K the unreacted triglycerides concentration with time-on-stream in
case of longer runs carried out at higher temperatures. In principle,
Fig. 7. Arrhenius graph as function of different reaction time length. the reaction rate decrease may be of a temporary character, as the

Fig. 9. SEM micrographs of the fresh (a) and used (b) catalyst.
Dj. Vujicic et al. / Fuel 89 (2010) 20542061 2061

consequence of diffusion resistance due to poor mixing [26]. How- of evidence for the equilibrium shift from products to the reactants
ever, it can also be of a permanent character due to inappropriate side, lowering the steady-state esters yield.
reaction conditions applied, as well as due to extensive catalyst
deactivation [26]. As shown in Fig. 9 presenting SEM pictures of Acknowledgements
a fresh catalyst sample and of a used catalyst (recovered from a
reaction run that lasted for 5.5 h at 100 C), a signicant catalyst Financial supports of Serbian Ministry of Science and Techno-
deactivation does occur due to particle coalescence to gum-like logical Development (Project ON 142024) and Victoria Oil, Biodie-
chunks inapt of providing good reactant contact. It has to be sel Production Company Sid, Serbia, are highly appreciated.
emphasized, however, that the explanation of Arrhenius plot
trends (Fig. 8) by the catalyst deactivation scenario, although con- References
vincing at the rst sight, is valid only for the rst hour of time-on-
stream. In all other cases incomplete Arrhenius curves correspond- [1] Arzamendi G, Campoa I, Arguinarena E, Sanchez M, Montes M, Gandia LM.
Synthesis of biodiesel with heterogeneous NaOH/alumina catalysts:
ing to higher temperatures must have been caused by some other comparison with homogeneous NaOH. Chem Eng J 1997;134:12330.
reason(s). Namely, in theory, the appearance of a maximum in the [2] Sharma YC, Singh B, Upadhyay SN. Advancements in development and
Arrhenius plot, followed by negative values of activation energies characterization of biodiesel: a review. Fuel 2008;87:235573.
[3] Ranganathan SV, Narasimhan SL, Muthukumar K. An overview of enzymatic
in higher temperature region, may be due to the opposite temper- production of biodiesel. Bioresour Technol 2008;99:397581.
ature effect on the reaction rate and catalyst surface coverage by [4] Gryglewicz S. Rapeseed oil methyl esters preparation using heterogeneous
the reactants [27,28]. Even more, it can also be a consequence of catalysts. Bioresour Technol 1999;70:24953.
[5] Yan S, Lu H, Liang B. Supported CaO catalysts used in the transesterication of
a wrong starting assumption on the constant number and activity rapeseed oil for the purpose of biodiesel production. Energy Fuel
of the reaction sites in the LangmuirHinshelwood kinetic model 2008;22:64651.
[27]. In this particular investigation, however, the atypical Arrhe- [6] Kouzu M, Kasuno T, Tajika M, Sugimoto Y, Yamanaka S, Hidaka J. Calcium oxide
as a solid base catalyst for transesterication of soybean oil and its application
nius plot trend (Fig. 8) is due to the negative reaction rate incre-
to biodiesel production. Fuel 2008;87:2798806.
ments with time-on-stream, most probably as the consequence [7] Liptay G. Atlas thermoanalytical curves, book 2. Budapest: Akadmiai Kiad;
of the equilibrium shift from products to the reactants side. This 1976.
[8] Arzamendi G, Arguinarena E, Campoa I, Zabala S, Gandia LM. Alkaline and
clearly suggests the negative effect of too long reaction runs car-
alkaline- earth metals compounds as catalysts for the methanolysis of
ried out in the laboratory stirred tank reactor on the steady-state sunower oil. Catal Today 2007;133:30513.
biodiesel yield. [9] Lpez Granados M, Zafra Poves M, Martn Alonso D, Mariscal R, Cabello
Galisteo F, Moreno-Tost R, et al. Biodiesel from sunower oil by using activated
calcium oxide. Appl Catal B: Environ 2007;73:31726.
[10] Demirbas A. Comparison of transesterication methods for production of
4. Conclusions biodiesel from vegetable oils and fats. Energy Convers Manage
2008;49:12530.
[11] Tanabe K. Solid acid and bases. Tokyo: Kodansha; 1970. p. 53.
In the investigated reaction conditions span (60120 C; 1 [12] Miura H, Sugiyama K, Kawakami S, Aoyama T, Matsuda T. Selective hydration
15 bar; 200 rpm; methanol:oil = 6:1; 1 mas.% catalyst loading), of acrylonitrile on metal oxide catalysts. Chem Lett 1982:1836.
[13] Prepsch K, Seibl S. Tables for identication of the organic compounds by means
CaO is capable of catalyzing the transesterication reaction of re-
of spectroscopy methods. Chemistry in industry. Zagreb; 1982 [in Croatian].
ned sunower oil in a laboratory stirred tank reactor. The catalyst [14] Tanabe K. New solid acids and bases. Stud Surf Sci Catal 1989:27.
activation in static air conditions at 900 C preceding the reaction [15] Ramos M, Fernndez C, Casa A, Rodrguez C, Prez . Inuence of fatty acid
is necessary for the sake of strong basic sites formation. The high- composition of raw materials on biodiesel properties. Bioresour Technol
2009;100:2618.
est FAME yield of 91% was achieved at 80 C for the performed [16] JUS EN 14214:2004. Automotive fuels. Fatty acid methyl esters (FAME) for
5.5 h reaction run, while almost the same yield was found for the diesel engines-requirements and test methods. Belgrade, Serbia:
reaction proceeding at 100 C for less than half of that time. There Standardization Institute; 2004.
[17] Demirbas A. Relationships derived from physical properties of vegetable oil
is a positive pressure effect upon the reaction up to 10 bar at 80 C. and biodiesel fuels. Fuel 2008;87:17438.
The main physicochemical properties of the obtained product, as [18] Srivastava A, Prasad R. Triglycerides-based diesel fuels. Renew Sustain Energy
the iodine number, viscosity, cetane index and cold ow proper- Rev 2000;4:11133.
[19] Ladommatos N, Goacher J. Equations for predicting the cetane number of
ties, do not completely meet the required biodiesel fuel specica- diesel fuels from their physical properties. Fuel 1995;74:108393.
tions due to the inappropriate fatty acid composition of the [20] Krisnangkura K. A simple method for estimation of cetane index of vegetable
feedstock. Viscosity reduction occurred upon transesterication oil methyl esters. JAOCS 1986;3:5523.
[21] Lopes J, Boros L, Krhenbhl M, Meirelles A, Daridon J, Pauly J, et al. Prediction
due to structural factors of resulted fatty compounds; however, of cloud points of biodiesel. Energy Fuel 2008;22:74752.
slightly decreased content of highly unsaturated acids in FAME [22] Freedman B, Buttereld R, Pryde E. Transesterication kinetics of soybean oil.
comparing to the feedstock hinders the expected degree of vis- JAOCS 1986;63:137580.
[23] Singh AK, Fernando SD. Reaction kinetics of soybean oil transesterication
cosity reduction of the obtained fuel. Cetane index is on the lower
using heterogeneous metal oxide catalysts. Chem Eng Technol
limit due to low fraction of saturates. At the here examined condi- 2007;30:171620.
tions, transesterication of the rened sunower oil is a (pseudo)- [24] Schwab AW, Bagby MO, Freedman B. Preparation and properties of diesel fuels
rst order reaction, however with a knee at 80 C in Arrhenius from vegetable oils. Fuel 1987;66:13728.
[25] Freedman B, Pryde EH, Mounts TL. Variables affecting the yields of fatty esters
plot separating the corresponding reaction regimes. Signicant dif- from transesteried vegetable oils. JAOCS 1984;61:163843.
fusion restrictions, both external and internal, exist at higher ap- [26] Levenspiel O. Chemical reaction engineering. New York: John Wiley & Sons,
plied temperatures due to an inadequate mixing and catalyst Inc.; 1972.
[27] Boudart M, Diega-Mariadassou G. Kinetics of heterogeneous catalytic
deactivation. Following the increase of the reaction duration, there reactions. Princeton: Princeton University Press; 1984.
is a decrease in the values of reaction rate constants, attaining neg- [28] Thomas WJ. In: Thomas JM, Zamaraev KI, editors. Perspective in
ative ones at the high temperature region. This represents a piece catalysis. Oxford: Blackwel Sci. Publ.; 1982. p. 25187.

The author has requested enhancement of the downloaded file. All in-text references underlined in blue are linked to publications on ResearchGate.

You might also like