Industrial Crops and Products 74 (2015) 6975

Contents lists available at ScienceDirect

Industrial Crops and Products

journal homepage: www.elsevier.com/locate/indcrop

Extraction of soybean oil using ethanol and mixtures with alkyl esters
(biodiesel) as co-solvent: Kinetics and thermodynamics
Joo Luiz Andreotti Dagostin, Danielle Carpin, Marcos L. Corazza
Department of Chemical Engineering, Federal University of Paran, 81531-990 Curitiba, PR, Brazil

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

Article history: The kinetics and thermodynamics of soybean extraction using ethanol and ethanol + alkyl esters mixtures
Received 7 January 2015 (biodiesel) as co-solvent were investigated. The extraction parameters investigated were temperature
Received in revised form 9 March 2015 (25, 40 and 55 C) and biodiesel to ethanol mass ratio. The extractions were carried out in batch mode
Accepted 27 April 2015
at a xed solvent to soybean mass ratio (4:1). Three different kinetic models were used to correlate the
Available online 20 May 2015
overall extraction curves, namely a rst-order, second-order, and a mass transfer model based on Ficks
Law. The mass transfer model presented a mean absolute relative deviation of 4.7%, while de rst and
Keywords:
second-order models showed values around 10.1% and 4.4%, respectively. From the experimental results
Soybean oil
Extraction
obtained, it was observed that higher temperatures affected positively the extraction yields. Furthermore,
Kinetic higher mass ratio of biodiesel to anhydrous ethanol provide higher extraction yields, at temperatures of
Ethanol 25 and 40 C. At 55 C, the extraction yields were not improved when adding the biodiesel as co-solvent to
Biodiesel the anhydrous ethanol. On the other hand, using hydrated ethanol as solvent resulted in lower extraction
yields, so higher amounts of biodiesel were necessary to reach the same levels of extractions performed
using anhydrous ethanol. The thermodynamic analyses showed that both H and S are positive for
this process, while negative values of G were found, that means the process evaluated is characterized
as endothermic, irreversible and occurs spontaneously.
2015 Elsevier B.V. All rights reserved.

1. Introduction both thermodynamic conditions: at room temperature and pres-

Soybean (Glycine max L.) is the most cultivated oilseed crop in
the world, mainly because it combines moderate oil content and
high amount of protein, which represents approximately 2040%
of its total mass, respectively (Shukla et al., 1992). Data from the
United States Department of Agriculture indicate an estimated
global soybean production in 2013/14 of 283.95 million metric tons.
This estimate shows US ranking as rst in world on soybean produc-
tion, followed by Brazil and Argentina, representing 31.52%, 30.82%
and 19.05% of global production, respectively (USDA, 2014).
The extraction is the most important process applied in soy-
bean oil production, occurring after the grains preparation steps,
as dehulling, cracking and aking. Most widely studied methods
for the oil extraction from plants involve either mechanical press-
ing (Subroto et al., 2015; Patil and Ali, 2006) or solvent extraction
(Kostic et al., 2014); the last may be accomplished using solvent at
Corresponding author at: Department of Chemical Engineering, Federal Univer-
sity of Paran, PO Box 19011, Polytechnic Center, Curitiba 81531-980, PR, Brazil.
Tel.: +55 41 33613190.
E-mail address: corazza@ufpr.br (M.L. Corazza).
sure or under subcritical (Liu et al., 2014; Ndlela et al., 2013)
and supercritical (Jokic et al., 2010) conditions. Commonly, soy-
beans are solvent-extracted using commercial hexane, a mixture
of hydrocarbons with a boiling point around 6569 C that con-
tains about 65% n-hexane and 35% of cyclopentane and hexane
isomers (Kemper, 2005). This solvent presents several advantages,
as low latent heat of vaporization, complete solubility with oil,
low corrosiveness, easy recovery and water immiscibility. How-
ever, there are also disadvantages associated, mainly concerning
environmental, safety and health issues. The US Environmental Pro-
tection Agency (EPA) has published online, through the Integrated
Risk Information System (IRIS), a complete toxicological review
about the n-hexane (Benedict et al., 2005). This review states that
individuals exposed to n-hexane vapor may exhibit injuries in the
nervous system, besides other symptoms as irritation of the respi-
ratory tract and eyes. Another problem associated with hexane is
that its vapor is heavier than air, forming a highly explosive mix-
ture with the last, even when small amounts are dispersed (Kemper,
2005).
Several alternative solvents have been proposed to replace
hexane in vegetable oil extraction including the use of trichlorethy-
lene, n-heptane, ethanol, isopropanol and n-propanol (Arnold and

http://dx.doi.org/10.1016/j.indcrop.2015.04.054
0926-6690/ 2015 Elsevier B.V. All rights reserved.
70 J.L.A. Dagostin et al. / Industrial Crops and Products 74 (2015) 6975
Choudhury, 1962; Seth et al., 2007; Gandhi et al., 2003). Among
these, the ethanol seems to be a promising solvent for soybean oil
extraction in the US and Brazilian industries. Both countries are
pioneers in the production of ethanol and soybean, ranking rst
and second for both commodities, respectively. According to the
Renewable Fuels Association (RFA), the US and Brazil produced
13,300 and 6267 millions of gallons of ethanol in 2013, respec-
tively, corresponding to 83.52% of the total ethanol produced in the
world. Despite the lower total production, Brazilian ethanol has the
advantage of lower cost of production and higher productivity per
acre due to the use of sugarcane as feedstock (Budny, 2007). With
regard to health and safety, ethanol presents lower risk than hydro-
carbons. There is no evidence of toxicity by inhalation of ethanol
(Nadeau et al., 2003; Clayton and Clayton, 1994), and it presents
lower ammability compared to hexane (Anderson, 2005).
Due to its polar nature, however, ethanol is a weaker oil solubi-
lizing agent than the hexane. Ethanol presents a limited solubility
with triacylglycerols and is capable of solubilizing not only the oil
present in the matrix but also water and other polar components.
If hydrated ethanol is applied as solvent, its ability of solubilizing
triglycerides is further decreased, gradually limiting the process
efciency (Dagostin et al., 2015). In order to increase the solubil-
ity of a component into another some thermodynamic artices can
be used, as increasing the pressure and temperature or adding a
co-solvent to the mixture. By denition, a co-solvent is a substance
that presents complete solubility in both solute and solvent and
improves the overall solubility of the system.
Studies have shown that biodiesel is able to solubilize oils and
alcohols at adequate proportions and operational temperatures
(Dagostin et al., 2015; Silva et al., 2013). Thus, biodiesel could be a
feasible co-solvent to be used together with ethanol in oil extraction
processes.
Biodiesel is a mixture of alkyl esters commonly produced by a
transesterication reaction between triglycerides (such as soybean
oil) and an alcohol (usually short chain alcohols as methanol or
ethanol) in the presence of a catalyst yielding the alkyl esters and
glycerol as by-product (Fukuda et al., 2001). Taking into account
the integration between the oil extraction plant and biodiesel
production unit, the biodiesel (ethyl esters) application as co-
solvent for soybean oil extraction is a promising strategy. In this
context, the main subject of this study was to evaluate the kinet-
ics and thermodynamics aspects of soybean oil extraction using
(ethanol + biodiesel) mixtures in batch systems. The experimen-
tal variables studied were time, temperature and solvent-to-solids
ratio for the soybean oil extraction.
2. Experimental procedure
2.1. Material
Soybean akes (G. max) were gently supplied by IMCOPA
(Araucria, Brazil). These were vacuum-packed in high-density
polyethylene bags, which were stored under refrigeration (12 C)
before analyses. Fatty acid ethyl esters (ethylic biodiesel) con-
taining 97.6% esters content (wt%) were produced as previously
reported (Dagostin et al., 2015). The anhydrous (>99.8% v/v) and
hydrated ethanol (96% v/v) of analytical grade were purchased from
PANREAC (Barcelona, Spain). The chemicals were used as received,
without further purication.
2.2. Methods
2.2.1. Extraction apparatus and experimental procedure
Batch extractions were performed with weighed portions of
soybean and pre-determined amounts of solvent in Erlenmeyer
asks (50 mL). The solvents used were anhydrous ethanol or
hydrated ethanol 96% (v/v) and their mixtures with biodiesel, at
different solvent to biodiesel mass ratios. Sample and solvent were
previously heated/cooled to the extraction temperature prior to
extraction. Just after mixing the solvent with the soybean, these
were immediately sealed with Paralm MTM and covered with PVC
lm to avoid solvent loss. The temperature of extraction was con-
trolled by a Dubnoff reciprocal shaker (Ethik Technology, model
304-TPA) at 25, 40 or 55 C applying vigorous shaking. At the end
of a predetermined extraction time (1, 2, 5, 10, 15, 30, 60, 90, 120
and 180 min), the bulk phase was immediately collected using a
syringe tted with a paper membrane (14 m) and transferred to
tubes. Each experiment was carried out at least in duplicate. The
oily extract was determined gravimetrically by evaporating the col-
lected samples (60 C, 24 h). The volatile fraction was considered as
ethanol, while the heavier fraction was composed by soybean oil
and fatty acid ethyl esters. Mass balance was made to calculate the
amount of oil extracted. Extraction yields are expressed as amount
of extract (mass) per 100 g of soybean.
Total oily extract was obtained by two Soxhlet exhaustive types
of extraction following the AOCS Ac 344 method (AOCS, 2001).
The rst exhaustive extraction used petroleum ether (5 h) and the
last anhydrous ethanol (7 h).
2.2.2. Kinetic modeling
Although power law models are used extensively in adsorption
processes, they are also able to describe well solidliquid extraction
processes (Man et al., 2012; Sayyar et al., 2009). An equation in the
form of power law is a relationship of Y = k Xn , where X and Y are the
variables of interest, n is called the exponent of the power law and
k is a constant. For an extraction kinetic, it becomes as described by
Eq. (1).
dC
= k(C)n = k(C C)n (1)
dt
where k is the constant of extraction rate (min1 ), C is the solute
concentration at equilibrium (g solute g1 soybean), C is the con-
centration of solute extracted at a given time (also in terms of g
solute g1 soybean) and n is the order of the equation.
2.3. First-order model
If at the beginning of the process t = 0, the concentration of solute
in the solvent is C = 0; therefore, Eq. (1) takes the form:
C = C (1 expkt ) (2)
2.3.1. Second-order model
With the initial conditions at t = 0 with C = 0, the following equa-
tion is obtained from Eq. (1) after integration and rearranging:
t
C= 1 t
(3)
2 + C
kC
2.3.2. Mass transfer kinetic model (MTKM)
The mass transfer kinetic model used in this work was derived
from the Ficks law of diffusion for an innite slab in a well stirred
solvent of innite volume, according to Chan et al. (2014) and Crank
(1975). The basic assumptions used in this work to simplify the
problems of mass transfer were
(a) The soybean akes are symmetrical, in the form of innite
plates measuring 2L.
(b) The solid particles are considered a porous pseudo-
homogeneous medium. The concentration of solute in the
solid particle depends on time and thickness.
 J.L.A. Dagostin et al. / Industrial Crops and Products 74 (2015) 6975 71

(c) The solute is uniformly distributed in the matrix before the 2.3.4. Thermodynamic assessment
extraction process. The distribution coefcient (K) for the liquidsolid systems com-
(d) The mixture of solvent and solute is homogeneous. The con- prising ethanol or (ethanol + biodiesel) and soybean was calculated
centration of the solute in the solvent phase is only a time using the following equation:
dependent variable.
C C
(e) The extraction process is a phenomenon of diffusion in which K= = (8)
Cs C0 C
the diffusion coefcient is a constant.
(f) The diffusion of compounds occurs in parallel and do not inter- where Cs is the concentration of non-removed extract from the
act with each other. soybean at equilibrium (in g oil/g of soybean). Values of C0 were
(g) External mass transfer resistance is negligible. The concen- determined experimentally, while values of C were obtained by
tration of solute in the bulk solvent is the same as the model tting (MTKM).
concentration of solute in the solvent at the interior of the solid The thermodynamic parameters of enthalpy change H (kJ
matrix. mol1 ) and entropy change S (J mol1 K1 ) from the soybean oil
(h) The bulk solvent was considered a liquid of innite volume. The extraction were calculated by a linear t of the Vant Hoff equation:
slab is washed by a liquid constantly kept pure. This assumption
resulted in an adjustable parameter model and was considered H S
lnK = + (9)
in this case to simplify the calculations. RT R
where R is the universal gas constant (J mol1 K1 ). The Gibbs free
So, assuming the solids are immersed in a liquid of innite vol- energy change, G (kJ mol1 ), was determined according to the
ume, the resistance to external mass transfer negligible, with the expression:
particles to be extracted having a slab geometry with half of their
G = H T S (9)
thickness equal to L (mm), and considering the diffusion only in the
direction of the thickness of the plate (Crank, 1975; Treybal, 1981),
the respective initial and boundary conditions can be written as 3. Results and discussion
follows:Initial t = 0, C = C0 , L < x < L
Boundary t > 0, (C/x) = 0, x = 0 3.1. Kinetics and modeling
t > 0, C = Ci , x = Lwhere C0 and Ci are the initial and interfa-
cial concentrations of solute in the particle, and the corresponding The soybean was characterized after packaging and storage. The
model is expressed by Eq. (4). moisture content measured in the akes was 9.33 0.11% and their
  thickness was 0.220 0.019 mm. The extraction yield (Soxhlet)
8 

M 1 (2n + 1)2 2 Dt using petroleum ether was 23.88 0.13%, while for the extraction
=1 2 exp (4) using ethanol reached 25.58 0.25%. The fact that the extraction
M  (2n + 1)2 4L2
n=0 yield with ethanol was higher remains in the ethanol capability of
where M and M are the amount of solute transferred in a given removing other solutes than just the lipid or nonpolar components.
time and innite time, respectively (g), D is the diffusivity of the Preliminary tests were conducted to identify the time needed
solute inside the particle (mm2 min1 ), t is the extraction time for the extraction processes to reach the equilibrium condition.
(min). In terms of solute concentration, the equation can be rewrit- It was identied that within the temperature range and types of
ten as: solvents used, 2 h was sufcient to achieve the equilibrium. In
  these tests, higher temperatures and the addition of biodiesel to
C C  8 

1 (2n + 1)2 2 Dt the ethanol favored the oil extraction. Thus, to study the effect of

= exp (5) solvent to soybean ratio, the experiments of extraction at equilib-
C 2 (2n + 1)2 4L2
n=0 rium were performed using those which presented the best and the
where C and C are the solute concentration in the solid parti- worst operational conditions, it means, A = {55 C, 90% anhydrous
cle (w/w) at a given time and after innite time, respectively. The ethanol with 10% biodiesel (mass to mass)} and B = {25 C, anhy-
kinetic parameters were obtained by minimizing the mean square drous ethanol}, respectively. To ensure that the extraction would
differences between experiments and calculated extraction yields reach the equilibrium, the contact time between solvent and soy-
(Cexp and Ccal , respectively), with the LevenbergMarquardt opti- bean was raised to 3 h. The extraction yield as a function of the
mization method. solvent to soybean ratio is shown in Fig. 1.
According to Fig. 1, in the operational condition (B) the oil
extraction increases linearly as the solvent to soybean mass ratio
2.3.3. Models evaluation
increase, until the (7:1) ratio is reached. In condition (B), the extrac-
The calculations of the average absolute relative deviation
tion yield was sharply lower when compared to condition (A) up
(AARD) according to Eq. (6), and the root mean square error (RMSE)
to (7:1), and from this solvent to soybean mass ratio the overall
according to Eq. (7), were performed to assess the goodness of
extraction capacity was the same for both processing conditions.
ttings of those kinetic models in to correlate the experimental
In condition (A), the extraction yield at a ratio equal or higher
data.
than (3:1) (solvent to soybean mass) resulted in greater yields when
1  Cexp Ccal
n compared to the conventional Soxhlet process (23.88%). This indi-
AARD = | | 100 (6) cates that the extraction with solvent A (90% anhydrous ethanol and
n Cexp
i=1 i
10% biodiesel) presented lower selectivity than petroleum ether, in
 other words, it permits the extraction of other components from the
n (Cexp Ccal )2 matrix than not the oil (nonpolar acylglycerols). By means of the
RMSE = (7) extraction kinetics, as follows, it is possible to check whether this
i=1 n
reduced selectivity is an effect of the biodiesel usage as co-solvent
where n is the number of experimental data available, and Cexp and or of ethanol itself.
Ccal are the concentrations of solute obtained experimentally and According to the results obtained for the best condition (A)
estimated by the model, respectively. at equilibrium a xed solvent to soybean mass ratio of (4:1) was
72 J.L.A. Dagostin et al. / Industrial Crops and Products 74 (2015) 6975

26
20

Extraction yield (g oil/100 g soybean)

24
Extraction yield (g oil/100 g soybean)

22
15
20

18
10
16

14
5
12

10
0
2 4 6 8 10
0 20 40 60 80 100 120
Solvent ratio (g solvent/g soybean) Time (min)
Fig. 1. Extraction yield as a function of the solvent to soybean mass ratio used: (A) Fig. 2. Overall extraction curves as a function of time. Kinetic experiments at
55 C, (90% anhydrous ethanol + 10% biodiesel, mass basis) (); (B) 25 C, anhydrous 25 C and solvent to soybean mass ratio (4:1), with solvent conditions: anhydrous
ethanol (). ethanol, EtOH (); 95% anhydrous ethanol + 5% biodiesel, EtOH5% (); 90% anhy-
drous ethanol + 10% biodiesel, EtOH10% (). The solid line corresponds to the mass
transfer model, Eq. (8), truncated at the rst term.
used for all kinetic experiments. This ratio was chosen in order to
reduce solvent consumption and due to the characteristics of this
operational process, in which large-scale extractions typically use 25
continuous steps that ensure higher process efciency.
Extraction yield (g oil/100 g soybean)

In this study, the kinetics data of extraction were measured at

different temperatures (25, 40 and 55 C) and using different mass
fractions of biodiesel with anhydrous ethanol (0, 5 and 10%). To
describe these extraction processes, rst-order and second-order
models as well as the mass transfer kinetic model (MTKM) were
tted to the experimental data. The parameters obtained for each
model and the statistical results are presented in Table 1. The mass
transfer model was truncated at the rst term (n = 0). As shown in
Table 1, this order used was enough to get a good description of the
experimental data.
The models that best represented the experimental data of
extractions with anhydrous ethanol were the second-order and the
mass transfer based models. The difference between both models
was considered not signicant since their coefcients of determi-
nation (R2 ) were 0.9747 and 0.9731, and the absolute deviation
(AARD) were 5.20% and 5.00% for the second-order and mass trans-
fer models, respectively. These two models are showing that the
extraction rate (k) and the equilibrium constant (C ) increase as
the temperature increases. However, at 25 C, when the co-solvent
is added the extraction rate decrease, showing that the mass trans-
fer rate during the extraction is decreased due to the addition of
a larger molecule (fatty acid methyl esters). At 40 and 55 C, the
extraction rate is less affect by the addition of the co-solvent. As
the mass transfer model is based on theoretical diffusion phenom-
ena and provided lower average deviations (AARD and RMSE, from
Table 1), it was chosen to be set graphically as shown in Figs. 24.
Fig. 2 shows the kinetic curves at 25 C for the extractions with
anhydrous ethanol (EtOH), and its mixtures with 5% biodiesel (95%
anhydrous ethanol + 5% biodiesel, mass basis, named EtOH5%) and
10% biodiesel (90% anhydrous ethanol + 10% biodiesel, mass basis,
named EtOH10%). The extraction capability of the three solvents is
very similar up to 15 min of extraction. After this time, the solubility
of oil in ethanol is increased using EtOH10%, followed by EtOH5%
and nally EtOH, which were able to extract about 19.9%, 17.1% and
15.1% of oil from the soybeans.
Figs. 3 and 4 present the kinetics of extraction using anhydrous
ethanol and its mixtures with biodiesel at 40 and 55 C, together
with the representation of the mass transfer model tting. It can
be seen that the extraction yield at equilibrium increased along the
20
15
10
5
0
0 20 40 60 80 100 120 140 160 180
Time (min)
Fig. 3. Overall extraction curves as a function of time. Kinetic experiments at
40 C and solvent to soybean mass ratio (4:1), with solvent conditions: anhydrous
ethanol, EtOH (); 95% anhydrous ethanol + 5% biodiesel, EtOH5% (); 90% anhy-
drous ethanol + 10% biodiesel, EtOH10% (). The solid line corresponds to the mass
transfer model, Eq. (8), truncated at the rst term.
increase in the temperature of extraction for the same solvents.
When biodiesel was used as co-solvent, it was possible to see (Fig. 3)
the same behavior observed for the kinetic curves at 25 C (Fig. 2),
i.e., similar extraction up to 15 min and the increase in the extrac-
tion yield regarding solvents with higher biodiesel content when
the equilibrium is reached. However, it can be noticed a minor
difference in the kinetics of extraction among the solvents when
comparing to those previously observed at 25 C. At 55 C (Fig. 4),
there were no signicant differences observed between using anhy-
drous ethanol and adding biodiesel as a co-solvent (anhydrous
ethanol + biodiesel mixtures) either in all the kinetics of extraction
of soybean oil or for the equilibrium yield reached. Furthermore, all
extraction conditions at 55 C were able to extract completely the
oil present in the soybean seeds whereas for the extraction at 40 C
just when using ethanol with 10% of biodiesel the total extraction
was reached.
 J.L.A. Dagostin et al. / Industrial Crops and Products 74 (2015) 6975 73

Table 1
The parameters of the rst-order, second-order and mass transfer kinetic models and its standard deviations tted for anhydrous ethanol and its mixtures with biodiesel.

T ( C) Solvent C a,b (gslt gs 1 ) kc (min1 ) R2 AARD RMSE

First-order model
25 EtOH 0.1458 0.0022 0.2849 0.0205 0.9450 7.73 0.04
EtOH5% 0.1635 0.0029 0.2284 0.0183 0.9374 9.48 0.05
EtOH10% 0.1905 0.0044 0.1271 0.0107 0.9224 14.99 0.09
40 EtOH 0.2072 0.0039 0.1998 0.0162 0.9148 11.33 0.10
EtOH5% 0.2148 0.0056 0.1848 0.0178 0.9030 12.80 0.10
EtOH10% 0.2362 0.0044 0.1603 0.0122 0.9264 12.79 0.11
55 EtOH 0.2437 0.0039 0.2570 0.0202 0.9038 10.24 0.12
EtOH5% 0.2439 0.0056 0.2661 0.0253 0.8572 11.34 0.14
EtOH10% 0.2462 0.0048 0.3505 0.0358 0.8814 9.71 0.09
Second-order model
25 EtOH 0.1563 0.0013 2.6693 0.1340 0.9879 3.27 0.02
EtOH5% 0.1762 0.0013 1.8910 0.0804 0.9922 3.38 0.02
EtOH10% 0.2085 0.0033 0.8617 0.0692 0.9741 8.14 0.05
40 EtOH 0.2236 0.0021 1.3221 0.0728 0.9831 4.80 0.04
EtOH5% 0.2375 0.0045 1.0967 0.1044 0.9640 7.39 0.06
EtOH10% 0.2524 0.0027 0.9610 0.0571 0.9819 5.86 0.05
55 EtOH 0.2594 0.0021 1.4890 0.0779 0.9710 3.73 0.05
EtOH5% 0.2603 0.0037 1.5638 0.1244 0.9539 5.40 0.08
EtOH10% 0.2622 0.0033 2.0242 0.1738 0.9639 4.85 0.05
Mass transfer kinetic model (MTKM)
25 EtOH 0.1491 0.0014 0.0745 0.0037 0.9827 3.13 0.02
EtOH5% 0.1682 0.0015 0.0579 0.0027 0.9855 3.20 0.03
EtOH10% 0.1971 0.0029 0.0320 0.0020 0.9724 7.18 0.05
40 EtOH 0.2142 0.0021 0.0495 0.0023 0.9801 4.50 0.05
EtOH5% 0.2255 0.0034 0.0450 0.0028 0.9746 6.85 0.05
EtOH10% 0.2424 0.0020 0.0417 0.0016 0.9866 3.86 0.05
55 EtOH 0.2505 0.0025 0.0633 0.0035 0.9659 4.63 0.07
EtOH5% 0.2527 0.0036 0.0636 0.0043 0.9491 6.02 0.08
EtOH10% 0.2545 0.0032 0.0878 0.0065 0.9606 5.580 0.05

The values are the parameters (C and k) standard deviation.

a
C is the equilibrium concentration.
b
gslv and gs correspond to the units (grams) of solvent and soybean, respectively.
c
Parameter k for MTKM corresponds to (D/L2 ) from Eq. (5), for L = 0.110 mm. Parameter k for second-order model corresponds to (gs gslv 1 min1 ).

Table 2
The parameters of the rst-order, second-order and mass transfer kinetic models and its standard deviations tted for hydrated ethanol (96%) and its mixtures with biodiesel:
conditions C (25 C, ethanol 96%) and D (55 C, ethanol 96% + 20% biodiesel).

Condition Kinetic model C a,b (gslt gs 1 ) kd (min1 ) R2 AARD RMSE

C 1st order 0.0595 0.0021 0.3583 0.0621 0.8858 8.96 0.01

2nd order 0.0639 0.0009 7.8954 0.6946 0.9865 2.81 0.00
MTKMc 0.0612 0.0016 0.0880 0.0128 0.9461 6.04 0.01
D 1st order 0.2417 0.0078 0.2487 0.0368 0.9294 9.13 0.05
2nd order 0.2587 0.0048 1.4600 0.1638 0.9819 4.30 0.02
MTKMc 0.2478 0.0030 0.0649 0.0040 0.9915 2.96 0.02

The values are the parameters (C and k) standard deviation.

a
C is the equilibrium concentration.
b
gslv and gs correspond to the units (grams) of solvent and soybean, respectively.
c
MTKM stands for mass transfer kinetic model, parameter k for second-order model corresponds to (gs gslv 1 min1 ).
d
Parameter k for MTKM corresponds to D/L2 from Eq. (5), for L = 0.110 mm.

At 55 C, the extraction yield was about 2% higher for all extrac-
tions when compared to the total oil obtained by the conventional
Soxhlet (petroleum ether). This conrms that the use of ethanol
causes other components to be extracted, besides those soluble
in petroleum ether or n-hexane. Similar results were obtained by
Arnold and Choudhury (1962) in the extraction of soybean oil with
ethanol at different levels of hydration. These authors found that
for higher levels of ethanol hydration fewer amounts of oil were
extracted and larger amounts of non-lipid components as well.
Kwiatkowski and Cheryan (2002) performed the extraction of corn
oil with anhydrous ethanol in three stages. As the extraction took
place, the water was straightly removed from the corn and mixed
with ethanol in the successive stages. Therefore, the oil extraction
capability of the solvent decreased, while the extraction of non-
lipid compounds increased.
The analysis from the kinetic data indicates that the extraction
process was limited by the ethanol to oil solubility, and not by
the mass transfer phenomena. After 2030 min, the extraction rate
decreased sharply and soon reached the equilibrium. All extraction
processes needed 4060 min to achieve equilibrium.
In our previous work (Dagostin et al., 2015), the phase equilib-
rium results showed that the solubility of soybean oil is reduced in
hydrated ethanol systems. Higher water contents in ethanol leads
to lower solubility of oil. A reduced mutual solubility between
oil and the solvent phase implies on a lower extraction capacity
because the saturation of the solvent phase is reached with lower
contents of solute.
In order to verify the feasibility of using hydrated ethanol in
soybean extraction, trials were made using ethanol 96% at 55 C and
a solvent to soybean mass ratio of (4:1). In these experiments, the
extraction time applied was 3 h using different biodiesel amounts
added to the ethanol 96%, as (co-solvent + solvent), respectively.
These results are presented in Fig. 5.
74 J.L.A. Dagostin et al. / Industrial Crops and Products 74 (2015) 6975

30
Extraction yield (g oil/100 g soybean)

25

Extraction yield (g oil/100 g soybean)

25

20 20

15
15

10

10
5

5
0 20 40 60 80 100 120 140 160 180 0
0 20 40 60 80 100 120
Time (min) Time (min)
Fig. 4. Overall extraction curves as a function of time. Kinetic experiments at
Fig. 6. Kinetics of extraction using different conditions. Ethanol 96%, 25 C, (4:1)
55 C and solvent to soybean mass ratio (4:1), with solvent conditions: anhydrous
solvent to soybean mass ratio (); 80% (ethanol 96%) + 20% biodiesel, 55 C, (4:1)
ethanol, EtOH (); 95% anhydrous ethanol + 5% biodiesel, EtOH5% (); 90% anhy-
solvent to soybean mass ratio (). The dashed line corresponds to the second-order
drous ethanol + 10% biodiesel, EtOH10% (). The solid line corresponds to the mass
model. The solid line corresponds to the mass transfer model, Eq. (8), truncated at
transfer model, Eq. (8), truncated at the rst term.
the rst term.

Regarding the kinetics of extractions with anhydrous ethanol

at 55 C and the solvent to soybean mass ratio of (4:1), as pre- sented in Fig. 4, the total amount of extracted oil was the same
for the three different solvent/co-solvent conditions (EtOH, EtOH5%
and EtOH10%). It was observed no inuence of adding biodiesel to
ethanol over the soybean oil extraction. However, different results
were obtained when hydrated ethanol (ethanol 96%, with water)
25
Extraction yield (g oil/100 g soybean)

was used as solvent at these same conditions, as it can be observed

from Fig. 5. For the extraction with hydrated ethanol, about 12%
yield was achieved and the extraction yield was increased following
the biodiesel addition into this solvent. In addition, the extraction
20 behavior showed to follow a sigmoidal function (Fig. 5). Above 20%
biodiesel in hydrated ethanol there was no signicant difference in
the total amount of oil extracted.
To check the kinetics of extraction using hydrated ethanol, two
experimental conditions were chosen: C (25 C, ethanol 96%) and D
15 (55 C, ethanol 96% + 20% biodiesel, in mass). Those were the worst
and best conditions considered within the limits studied in this
work. Table 2 shows the parameters and errors for conditions C
and D of extractions using hydrated ethanol, while Fig. 6 shows
their means and the tted second-order and mass transfer models,
10
0 5 10 15 20 25 represented by the dashed and solid lines, respectively. In these
kinetics, condition C showed a high correlation with the second-
Biodiesel in ethanol 96% vol (w%)
order kinetic model (R2 = 0.9865), while condition D showed a high
Fig. 5. Extraction yield as a function of the biodiesel content used with ethanol 96% correlation with the MTKM (R2 = 0.9915). The maximum amount of
as the solvent phase at 55 C and a solvent to soybean mass ratio of (4:1). solute extracted at condition C was slightly higher than 6%, which

Table 3
Distribution constants (K) and the thermodynamic parameters (H , S and G ) for the extraction of soybean akes using anhydrous ethanol and (anhydrous
ethanol + biodiesel) mixtures.

Solvent Temperature K H S G R2
(K) (kJ mol1 ) (J mol1 K1 ) (kJ mol1 )

EtOH 298.15 1.40 95.01 320.02 0.41 0.969

313.15 5.15 5.21
328.15 47.26 11.61
EtOH5% 298.15 1.92 101.11 342.95 1.14 0.966
313.15 7.44 6.28
328.15 81.52 13.14
EtOH10% 298.15 3.36 109.86 377.36 2.65 0.984
313.15 18.09 8.31
328.15 195.77 15.86

EtOH: Anhydrous ethanol.

EtOH5%: 95% anhydrous ethanol + 5% biodiesel.
EtOH10%: 90% anhydrous ethanol + 10% biodiesel.
 J.L.A. Dagostin et al. / Industrial Crops and Products 74 (2015) 6975
represents approximately 40% of the maximum amount extracted
using anhydrous ethanol (15.1%) at the same conditions. On the
other hand, the results obtained at condition D (hydrated ethanol
with 20% of biodiesel added as a co-solvent, Fig. 6) were very sim-
ilar to that using anhydrous ethanol (EtOH) and its mixtures with
biodiesel (EtOH5% and EtOH10%) at 55 C, where it was possible
to perform the complete extraction of the soybean oil. However,
if there were used contents of biodiesel lower than 20% in condi-
tion D, lower extraction rates and lower extraction yields might be
achieved.
3.2. Thermodynamic analysis
The mass transfer model (MTKM) tted to the experimental
kinetic was used to calculate the distribution coefcient (K). Table 3
presents the results of the thermodynamic analysis in terms of K,
H , S and G , as well as the coefcient of determination found
for the Vant Hoff equation (Eq. (9)) correlation.
All enthalpy changes found were positive, indicating that the
extraction process is endothermic and requires an energy input
during the process. The positive values of entropy indicate the irre-
versibility of the process and the values of the Gibbs free energy
changes for the soybean oil extraction were negative, showing the
spontaneity of this process. Decreasing values of G were found
for higher biodiesel additions in the solvent and also at higher
temperatures. This observation states that the use of biodiesel as co-
solvent of the ethanol, as well as the temperature increase, favors
the oil extraction from soybean akes.
4. Conclusions
Ethanol with alkyl esters (biodiesel) mixtures as co-solvent was
successfully applied to extract oil from soybean akes. The effect of
operating temperature, solvent to soybean mass ratio and mass of
biodiesel in ethanol at different concentrations was evaluated on
the kinetics and the extraction yield. High extraction rates and over-
all extraction yield were obtained at higher temperatures and for
all solvents mixtures evaluated. At lower temperatures, the effect
of biodiesel as a co-solvent was more pronounced. Mixtures of
(ethanol + biodiesel) showed, in general, a greater ability to extract
the solute when compared to ethanol as the solvent. The highest
extraction conditions were those using absolute ethanol at 55 C,
with or without biodiesel, and also those using a mixture of (20%
of biodiesel added to hydrated ethanol (96%), in mass) as solvent.
The second-order kinetic model and the mass transfer model
for an innite slab were the kinetic models that better correlated
the overall extraction yields at different experimental conditions.
Moreover, for the thermodynamics of equilibrium, it was found for
the extraction processes that H and S are positive, and G
is negative indicating that these are endothermic, irreversible and
spontaneous processes.
The use of biodiesel as co-solvent of ethanol for obtaining soy-
bean oil seems promising in the sense of lowering the energy
required to the process, wherein its effect becomes higher at lower
temperatures. Furthermore, our results showed the technical via-
bility of soybean oil extraction using ethanol with biodiesel a
co-solvent.
Acknowledgements
This work was supported nancially by Conselho Nacional
de Desenvolvimento Cientco e Tecnolgico (CNPq), a
75
Brazilian government agency for the advancement of science.
Research scholarships were granted to Joo Luiz Andreotti Dagostin
and Danielle Carpin by Coordenacao de Aperfeicoamento de
Pessoal de Nvel Superior (CAPES), a Brazilian government agency
for the improvement of higher education personnel, and to Marcos
L. Corazza by CNPq. The authors are also grateful to IMCOPA for
supplying the soybean akes.
References
Anderson, D., 2005. A primer on oils processing technology. In: Shahidi, F. (Ed.),
Baileys Industrial Oil and Fat Products. , sixtieth ed. Wiley-Interscience, New
York, pp. 156.
AOCS, 2001. Ofcial Methods and Recommended Practices of the American Oil
Chemists Society, ftieth ed. AOCS Press, Champaign.
Arnold, L.K., Choudhury, R.B.R., 1962. Ethanol extraction of soybean oil. J. Ame. Oil
Chem. Soc. 39 (8), 379380.
J. Benedict, G. Holdsworth, E.D. Ellis, A. Marcus, Toxicological review of n-hexane
(CAS No. 110-54-3), In Support of Summary Information on the Integrated Risk
Information System (IRIS) U.S. Environmental Protection Agency Washington,
DC 2005, Retrieved 03 January, 2015 from http://www.epa.gov/iris/
toxreviews/0486tr.pdf
Budny, D., 2007. The global dynamics of Biofuels, Brazil Institute Special Report,
17.
Chan, C., Yusoff, R., Ngoh, G., 2014. Modeling and kinetics study of conventional
and assisted batch solvent extraction. Chem. Eng. Res. Des. 92 (6), 11691186.
Clayton, G., Clayton, F., 1994. Pattys Industrial Hygiene and Toxicology, fourth ed.
John Wiley & Sons, New York.
Crank, J., 1975. The Mathematics of Diffusion, second ed. Oxford University Press,
London.
Dagostin, J.L.A., Mafra, M.R., Ramos, L.P., Corazza, M.L., 2015. Liquidliquid phase
equilibrium measurements and modeling for systems involving {soybean
oil + ethyl esters + (ethanoln + water)}. Fuel 141 (1), 164172.
Fukuda, H., Kondo, A., Noda, H., 2001. Biodiesel fuel production by
transesterication of oils. J. Biosci. Bioeng. 92 (5), 405416.
Gandhi, A.P., Joshi, K.C., Jha, K., Parihar, V.S., Srivastav, D.C., Raghunadh, P.,
Kawalkar, J., Jain, S.K., Tripathi, R.N., 2003. Studies on alternative solvents for
the extraction of oil soybean. Int. J. Food Sci. Tech. 38, 369375.
S., Zekovic,
Jokic, Z., Vidovic,
S., Sudar, R., Nemet, I., Bilic, M., Velic,
D., 2010.
Supercritical CO2 extraction of soybean oil: process optimisation and
triacylglycerol composition. Int. J. Food Sci. Tech. 45 (9), 19391946.
Kemper, T.G., 2005. Oil Extraction. In: Shahidi, F. (Ed.), Baileys Industrial Oil and
Fat Products. , sixtieth ed. Wiley-Interscience, New York, pp. 5798.
Kostic, M.D., Jokovic, N.M., Stamenkovic, O.S., Rajkovic, K.M., Milic, P.S., Veljkovic,
V.B., 2014. The kinetics and thermodynamics of hempseed oil extraction by
n-hexane. Ind. Crop. Prod 52, 679686.
Kwiatkowski, J.R., Cheryan, M., 2002. Extraction of oil from ground corn using
ethanol. J. Am. Oil Chem. Soc. 79 (8), 825830.
Liu, J., Chen, P., He, J., Deng, L., Wang, L., Lei, J., Rong, L., 2014. Extraction of oil from
Jatropha curcas seeds by subcritical uid extraction. Ind. Crop. Prod. 62,
235241.
Man, H.C., Hamzah, M.H., Jamaludin, H., Abidin, Z.Z., 2012. Preliminary study:
kinetics of oil extraction from citronella grass by ohmic heated hydro
distillation. APCBEE Proc. 3, 124128.
Ndlela, S.C., de Moura, J.M.L.N., Olson, N.K., Johnson, L.A., 2013. Aqueous extraction
of oil and protein from soybeans with subcritical water. J. Am. Oil Chem. Soc.
89 (6), 11451153.
Nadeau, V., Lamoureux, D., Beuter, A., Charbonneau, M., Tardif, R., 2003.
Neuromotor effects of acute ethanol inhalation exposure in humans: a
preliminary study. J. Occup. Health 45 (4), 215222.
Patil, R.T., Ali, N., 2006. Effect of pre-treatments on mechanical oil expression of
soybean using a commercial oil expeller. Int. J. Food Prop. 9 (2), 227236.
Sayyar, S., Abidin, Z.Z., Yunus, R., Muhammad, A., 2009. Extraction of oil from
jatropha seeds-optimization and kinetics. Am. J. Appl. Sci. 6 (7), 13901395.
Seth, S., Agrawal, Y.C., Ghosh, P.K., Jayas, D.S., Singh, B.P.N., 2007. Oil extraction
rates of soya bean using isopropyl alcohol as solvent. Biosys. Eng. 97 (2),
209217.
Silva, D.I.S., Mafra, M.R., Silva, M.R., Ndiaye, P.M., Ramos, L.P., Cardozo Filho, L.,
Corazza, M.L., 2013. Liquidliquid and vaporliquid equilibrium data for
biodiesel reactionseparation systems. Fuel 108, 269276.
Shukla, B.D., Srivastava, P.K., Gupta, R.K., 1992. Oilseeds Processing Technology.
Central Institute of Agricultural Engineering, ICAR, Bhopal, India.
Subroto, E., Manurung, R., Heeres, H.J., Broekhuis, A.A., 2015. Mechanical extraction
of oil from Jatropha curcas L. kernel: Effect of processing parameters. Ind. Crop.
Prod. 63, 303310.
Treybal, R.E., 1981. Mass-Transfer Operations, third ed. McGraw-Hill, London.
USDA United States Department of Agriculture. World Agricultural Supply and
Demand Estimates Report (WASDE). Retrieved 10 November, 2014 from:
http://www.usda.gov/oce/commodity/wasde/