2182 J. Chem. Eng.

Data 2009, 54, 2182–2188

Liquid-Liquid Equilibrium Data for Systems Containing Refined Rice Bran Oil,
Anhydrous Ethanol, Water, and Hexane

Irede Dalmolin, Marcelo Lanza, Antonio J. A. Meirelles, and Eduardo A. C. Batista*

ExTrAE, Laboratory of Extraction, Applied Thermodynamics and Equilibrium, Department of Food Engineering, Faculty of
Food Engineering, P.O. Box 6121, University of Campinas - UNICAMP, 13083-862, Campinas, SP, Brazil

The present work reports liquid-liquid equilibrium data for the systems containing refined rice bran oil +
anhydrous ethanol + water + hexane at (298.15 ( 0.2) K with mass fractions of (4.14, 9.84, and 14.79) %
of water in anhydrous ethanol and at (313.15 ( 0.2) K with a mass fraction of 4.40 % of water in anhydrous
ethanol. The instrumental technique of near-infrared spectroscopy allied to multivariate calibration (in this
work, termed “NIR method”) was used for determination of the phase compositions. To avoid phase separation
in sampling for the system at 313.15 K, the phase samples were diluted with tetrahydrofuran. The results
were evaluated by calculation of the global mass balance, resulting in relative deviations lower than 0.5 %
to all systems studied in this work. The experimental data were correlated using the NRTL model, and for
all systems, the global deviation was less than 1.0 %. These results indicate the good quality of the
experimental data and that the NIR method gives good results even when the components have a low mass
fraction in the mixture.

Introduction biological deacidification, reesterification, solvent extraction,

supercritical fluid extraction, and membrane technology.9
The vegetable oil sector has been one of the most dynamic An alternative refining process carried out at room temper-
of the world agriculture in the last few decades. The importance ature and atmospheric pressure that avoids the formation of
of Brazil as a producer and exporter country of oilseeds is waste products is liquid-liquid extraction, which consists of
evidenced by the numbers of the international market. Among using a solvent that has greater affinity with the free fatty acids
several species of oilseeds, the rice bran oil (RBO) has been (FFA) than the triacylglycerols (TAG), reducing the energy
used only to a small extent of its potential as edible oil because consumption and minimizing the losses of neutral oil. For the
of problems with the stability and storage of rice bran. The difference of the boiling point of each component, the posterior
extraction of oil from the rice bran aggregates value to the rice recovery of the solvent becomes possible. Several studies
production. This oil has nutritional and nutraceutical properties showed excellent results in the employment of short chain
and, therefore, the interest of industries in its use. Medical alcohols as solvent, such as anhydrous or hydrated ethanol,
studies indicate the hypocholesterolemic effect of rice bran oil which present low toxicity, easy recovery, good values of
in humans and animals. The majority of such studies suggests selectivity, and distribution coefficients of free fatty acid.10-14
that rice bran oil is more effective in decreasing serum and liver Moreover, environmental and energy issues have focused
cholesterol concentrations than oils with similar fatty acid attention on the use of vegetable oils in biodiesel production
composition, such as groundnut oil.1-3 by transesterification reactions.
The knowledge of the liquid-liquid equilibrium of such fatty
Crude vegetable oils are obtained by extraction from a solid
systems is an important stage in the optimization of the
matrix by pressing or using an organic solvent.4,5 In the
production of edible oil by liquid-liquid extraction or the
extraction process, the total/partial substitution of hexane
production of biodiesel by transesterification reactions. Further-
petroleum fractions becomes necessary because this usual
more, knowledge of the behavior of these systems involving
solvent is highly flammable and suspected to be dangerous to crude vegetable oil/solvents configures the principal topic in
human health. The full treatment to make the rice bran oil mass transfer operations and reactive mixtures. Thus, the aim
suitable for human consumption consists of preparation, extrac- of this work was to investigate the phase equilibrium of refined
tion from bran using solvent, solvent stripping, degumming, rice bran oil + anhydrous ethanol + water + hexane at 298.15
bleaching, deacidification, and deodorization.6 K, with mass fractions of (4.14, 9.84, and 14.79) % of water in
The vegetable oils can be deacidified by the traditional anhydrous ethanol and at 313.15 K with a mass fraction of 4.40
processes, such as chemical or physical refining; however, these % of water in anhydrous ethanol, using NIR spectroscopy
present several disadvantages, such as the high losses of neutral coupled with chemometric methods (multivariate calibration)
oil, large production of soapstock, and high energy consumption.7,8 based on statistical and mathematical procedures for the
Furthermore, a significant portion of the nutraceutical rice bran quantification of the components in each phase. The experi-
oil3 and other desirable compounds is lost during the refining mental data were correlated by the NRTL model.
process. Some new techniques that may be tried out are Experimental Section
* Corresponding author. Fax: + 55 19 3521 4027. E-mail: eacbat@ Materials. The refined rice bran oil was obtained com-
fea.unicamp.br. mercially from Irgovel S.A. (RS/Brazil) and stored in a place
10.1021/je900234u CCC: $40.75  2009 American Chemical Society
Published on Web 07/10/2009
 Journal of Chemical & Engineering Data, Vol. 54, No. 8, 2009 2183

Table 1. Fatty Acid Composition of the Refined Rice Bran Oil

fatty acids M composition
usual name proper name symbol Cx:ya g · mol-1 100 w 100 x
lauric dodecanoic L C12:0 200.32 0.01 0.01
myristic tetradecanoic M C14:0 228.37 0.24 0.29
palmitic hexadecanoic P C16:0 256.42 17.92 19.07
palmitoleic cis-9-hexadecenoic Po C16:1 254.41 0.15 0.16
stearic octadecanoic S C18:0 284.48 2.08 2.01
oleic cis-9-octadecenoic O C18:1 282.46 38.99 38.44
elaidicb trans-9-octadecenoic C18:1Tb 0.04 0.03
linoleic cis-9,cis-12-octadecadienoic Li C18:2 280.45 36.39 36.13
linolenic all-cis-9,12,15-octadecatrienoic Le C18:3 278.43 1.87 1.87
translinolenicb all-trans-9,12,15-octadecatrienoic C18:3Tb 0.42 0.38
arachidic eicosanoic A C20:0 312.53 0.71 0.63
gadoleic cis-9-eicosanoic Ga C20:1 310.51 0.51 0.46
behenic docosanoic Be C22:0 340.58 0.28 0.23
lignoceric tetracosanoic Lg C24:0 368.64 0.39 0.29
a
Cx:y, x ) number of carbons and y ) number of double bonds. b Trans isomers.

Table 2. Probable Triacylglycerol Composition of the Refined Rice

Bran Oil the fatty sample was prepared in the form of fatty acid methyl
M composition
esters according to the method of Hartman and Lago.19 The
chromatographic analyses were carried out using a capillary gas
main TAGa group g · mol-1 100 w 100 x
chromatography system under the same experimental conditions
POPb 50:1c 833.36 5.12 5.32 shown in Priamo et al.17
PLiP 50:2 831.34 5.11 5.32
The fatty acid composition of the vegetable oil is presented
POS 52:1 861.41 1.04 1.05
POO 52:2 859.40 11.18 11.27 in Table 1, and from this composition, it was possible to
POLi 52:3 857.38 19.53 19.73 determine the probable triacylglycerol composition of the refined
PLiLi 52:4 855.36 10.54 10.67 rice bran oil (Table 2) using the algorithm suggested by
PLiLe 52:5 853.35 1.23 1.25 Antoniosi Filho et al.20
SOO 54:2 887.45 1.50 1.46
OOO 54:3 885.43 7.24 7.08 To calculate the probable triacylglycerol composition, the
OOLi 54:4 883.42 15.63 15.32 quantities of trans isomers (see Table 1) were added with their
OLiLi 54:5 881.40 14.87 14.62 respective cis isomers. In Table 2, the main triacylglycerol
LiLiLi 54:6 879.38 6.13 6.04 represents the component with the greatest composition in the
LiLiLe 54:7 877.37 0.88 0.87
isomer set with x carbons and y double bonds. The results shown
a
Groups with a total triacylglycerol (TAG) composition lower than in Tables 1 and 2 allow us to calculate the average molar mass
0.5 % were ignored. b Each letter, whose name is given in Table 1, of the refined rice bran oil as 865.1 g · mol-1.
represents fatty acid esterified to glycerol. c x:y, x ) number of carbons Apparatus and Procedures. Measure of Density. The densi-
(except carbons of glycerol), y ) number of double bonds.
ties were determined experimentally at 298.15 K, using a digital
densimeter (Anton Paar model DMA-58, Austria) of oscillatory
with controlled temperature. The solvents used in this work were tube, with precision of ( 0.00001 g · cm-3, previously calibrated
anhydrous ethanol with purity of 99.9 % and hexane and with air and water (distilled and deionized) at the temperature
tetrahydrofuran, both with purity of 99.0 %, all from Merck of the analysis. A thermostat inside the instrument allows the
(Germany). Aqueous solvents with mass fractions of (4.14, 4.40, control of the temperature at ( 0.01 K.
9.84, and 14.79) % of water were prepared by the addition of Determination of Solubility CurWes. The solubility curves
deionized water to anhydrous ethanol. The reagent Karl Fischer were carried out by the cloud-point method17 and are necessary
(CombiTitrant 5, Merck, Germany) was also used. In this work, to obtain preliminary information about the system of interest
tetrahydrofuran was chosen as solvent to dilute the analytical and auxiliary in the selection of the validation and calibration
samples because it is an aprotic and moderately polar solvent standards used by the NIR method for the quantification of the
and can dissolve a wide range of nonpolar and polar chemical components in equilibrium. To determine the solubility curves,
compounds. There are many other solvents that could be used the same procedure of cloud-point method described by Priamo
to dilute the analytical samples, but tetrahydrofuran has high et al.17 was used in this work. The complete details can be seen
solvency power and could dissolve efficiently all samples of in that work.
studied systems in this work, which contain nonpolar (rice bran Determination of Liquid-Liquid Equilibrium Data. The
oil and hexane) and polar (ethanol and water) compounds. equilibrium data were determined using equilibrium glass cells
Solvent mixtures would not be appropriate in our case because (50 mL) similar to those used by Silva et al.21 Known quantities
they would increase the component quantity and would make of each component (rice bran oil, anhydrous ethanol, water, and
the quantification of the samples using the NIR method difficult. hexane) were put in contact inside the glass cells to reach
Quantifying systems containing more than five components leads equilibrium following the same procedure described by Priamo
to an increase of analytical error using the NIR method because et al.17 The complete details can be seen in that work. In the
of complexity in the selection of calibration and validation end of the experiment, the samples of the phases were carefully
standards and sensitivity in reading and interpretation of the collected for subsequent quantification of the components
near-infrared spectra.15-17 through the NIR method. For the systems at 313.15 K, a dilution
The refined rice bran oil used in this work was analyzed by technique was investigated. The phase samples of these systems
gas chromatography for the fatty acid methyl esters to determine were collected in duplicate. One of them was analyzed at the
the fatty acid composition, according to the official method temperature of the experiment, and the other one was diluted
(1-62) of the AOCS.18 Prior to the chromatographic analysis, with tetrahydrofuran to avoid the phase separation with the
2184 Journal of Chemical & Engineering Data, Vol. 54, No. 8, 2009

decrease of temperature to ambient temperature. This last sample bran oil, (0.03 to 0.58) % for anhydrous ethanol, (0.01 to 0.02)
was analyzed at 298.15 K. % for water, and (0.06 to 0.29) % for hexane, being the lowest
NIR Method. The instrumental technique of near-infrared figures obtained for the lowest compositions.
spectroscopy was used in combination with the method of
multivariate regression for partial least-squares (PLS) and the Theoretical Calculations
chemometric method (multivariate calibration) based on statisti- Calculation of the DeWiations in the Mass Balance of the
cal and mathematical procedures for the analytical determination Phases. The quality and accuracy of the results of the
of the mass fractions of the components in each phase of the liquid-liquid equilibrium experiments were tested according
system in equilibrium. In fact, for the utilization of the NIR to the procedure developed by Marcilla et al.25 and already
spectroscopy, it is necessary to make an equipment calibration, applied to fatty systems by several works.12,15-17,26 The balance
through multivariate calibration relating the components of the equations and calculations for relative deviation for the global
studied system for then explorer to the quantitative determina- mass balance and for mass balance of each component used in
tions conjunctly with the infrared spectroscopy. To develop this work are presented by Priamo et al.17
calibration models appropriate for each system, sets of mixtures The procedure to determine the relative deviation for global
with known compositions close to the solubility curve, previ- mass balance consists of calculating the masses for both liquid
ously determined, were selected. phases and comparing their sum with the actual value for total
The solubility curve was divided into two parts, the oil phase mass used in the experiment, thus obtaining a relative deviation
and the solvent phase, and for each part, 30 mixtures were for each point of the overall mixture. The relative deviation for
prepared by gravimetry and used as calibration standards. A mass balance of each compound in each tie line could be
further set of about 10 mixtures for each part of the solubility calculated comparing the actual mass of the component put in
curve was also prepared by gravimetry and used as an external the overall composition by the sum of the calculated masses of
validation set to check the accuracy of the calibration models.15-17 the component in each phase (solvent and oil phases). These
It is important to point out that in systems with water its calculated masses for each component are obtained by the
composition was checked through Karl Fischer titration, ac- multiplication of the experimental mass fractions, determined
cording to the official method Ca 23-55 of the AOCS.22 by the NIR method, and the calculated masses for both liquid
The near-infrared spectra were measured using a Thermo phases (solvent and oil phases).
Nicolet infrared spectrometer (Nexus, model 670 FT-IR Esp, Thermodynamic Modeling. The NRTL parameters were
Madison, USA) connected to an air purging system, with a adjusted to the experimental data determined for the systems.
heated cuvette holder (Ventacon, model HNIR-1, Winchester, Although the RBO is a mixture of many components, the several
United Kingdom) assisted by temperature controller (Ventacon, triacylglycerols present in the vegetable oil behave in a very similar
model D-2). Thereby, the NIR spectra were able to be measured way in the liquid-liquid system under analysis, and in this case,
at temperatures of (298.15 ( 0.5 and 313.15 ( 0.5) K. Complete as an approach, such compounds can be adequately replaced by a
details about the procedures to measure the NIR spectra and pseudocomponent having the corresponding average physical-
the description of the PLS regression method can be seen in chemical properties. This approach was already evaluated by Lanza
the work of Priamo et al.17 et al.,16 which proved the reliability of this hypothesis.
In the external validation, the accuracy of the models was The mass fraction was used as a composition unit due to the
evaluated by the relative error (ER) obtained from the difference large difference in molar mass of the components in the
between the predicted value and the actual value (gravimetry) systems.13-17,26-29 Lanza et al.15 showed the activity coefficient
for the external validation standards. In the internal validation, equations and the isoactivity criterion, expressed in terms of
the performance of the calibration models was evaluated by the mass fraction and for multicomponent mixtures.
root-mean-square error of cross-validation (rmsecv). These Estimation of the NRTL parameters was based on minimiza-
values express the accuracy of the models and the proximity tion of the objective composition function following the
between the values predicted by the model (wi,pred) and the actual procedure developed by Stragevitch and d’Avila.30 The objective
values (wi,actual) obtained using the reference method. The errors function of composition can be seen in the works of Lanza et
were defined as al.15,16 and Priamo et al.17
The parameter estimation procedure involves flash calcula-
n
|wi,pred - wi,actual |
ER )
1
∑
n i)1 wi,actual
(1) tions31 for the middle point compositions of the experimental
tie lines and minimization of the objective function of
composition.

n The average deviations between the experimental and cal-
∑ (wi,pred - wi,actual)2 culated compositions in both phases were calculated according
rmsecv )
i)1
(2) to Lanza et al.15,16 and Priamo et al.17
n
where n is the total number of validation (external or internal) Results and Discussion
standards and the subscript i is the number of the standard. The components studied in this work received the following
The rmsecv (root-mean-square error of cross-validation) is notation: refined rice bran oil (1), anhydrous ethanol (2), water
calculated using the cross-validation standards (internal valida- (3), hexane (4) and tetrahydrofuran (5).
tion), according to the cross-validation procedure known as The content of water in refined rice bran oil was determined
“leave one out”.23,24 The cross-validation and the corresponding by Karl Fischer titration. The value obtained was 0.05 %, and
rmsecv values together with the values of the relative errors this was always considered in the overall experimental composi-
were used for selecting the number of PLS factors considered tion of the mixtures and the calculation of calibration and
in the construction of the calibration models for the systems. validation standards.
The uncertainties of the compositions for the NIR method varied Table 3 shows the results of the calibration models generated
within the following ranges: (0.02 to 0.70) % for refined rice through the NIR method for the systems containing refined rice
 Journal of Chemical & Engineering Data, Vol. 54, No. 8, 2009 2185

Table 3. Relative Errors (ER) and Root-Mean-Square Error of Cross-Validation (rmsecv) in the Quantification of the Validation Standards
(External, ER, and Internal, rmsecv) Using the NIR Method for the Systems Containing Refined Rice Bran Oil (1) + Anhydrous Ethanol (2) +
Water (3) + Hexane (4) at (298.15 ( 0.2) K with Mass Fractions of (4.14, 9.84, and 14.79) % of Water in Anhydrous Ethanol (100 w3′) in
Overall Composition
solvent phase oil phase
100 w3′ oil ethanol water hexane oil ethanol water hexane ER/%
4.14 ER/% 12.50 0.45 0.20 1.13 0.25 0.40 3.48 1.14 2.44
rmsecv/% 0.64 0.58 0.01 0.09 0.70 0.45 0.01 0.29
9.84 ER/% 66.43 0.13 0.56 3.37 0.07 0.36 0.83 2.10 9.23
rmsecv/% 0.04 0.04 0.02 0.07 0.06 0.04 0.01 0.06
14.79 ER/% 9.15 0.06 0.02 1.19 0.13 0.59 1.89 1.68 1.84
rmsecv/% 0.02 0.09 0.01 0.09 0.08 0.03 0.01 0.08

bran oil (1) + anhydrous ethanol (2) + water (3) + hexane (4) at water, inconvenient for obtaining a monophasic system desired
(298.15 ( 0.2) K with mass fractions of (4.14, 9.84, and 14.79) % in the production of biodiesel.
of water in anhydrous ethanol using the NIR method: the errors Table 5 shows the relative deviation for the global mass
between the compositions calculated by the calibration models and balance of the pseudoquaternary systems and for mass balance
those calculated by gravimetry for the external validation standards of each compound. The deviations for global mass balance for
(ER) and for the cross-validation standards (rmsecv). the systems evaluated according to the procedure suggested by
In Table 3, we can observe that the errors for the oil in the Marcilla et al.25 were (0.14, 0.41, and 0.40) % for the systems
solvent phase and for water in the oil phase are higher than the with mass fractions of (4.14, 9.84, and 14.79) % of water in
other ones due to the small mass fraction of these components anhydrous ethanol, respectively. This indicates the good quality
in those phases. The oil mass fraction in validation standards of the experimental data. Minor relative deviations were found
varied in the range of 0.0009 to 0.0011, and any small variations for the refined rice bran oil and anhydrous ethanol. Water is
in the results lead to greater errors. However, the deviations of the minor component, and due to that, it presented higher
the calibration models present low values for the root-mean- deviations of (5.37, 6.97, and 9.52) %. The system with 14.79
square errors of cross-validation (rmsecv), for both phases, % in mass of water in anhydrous ethanol presented undesirable
evident for points lined up in the tie line, shown in Figure 1. effects in two tie lines with higher mass fraction in hexane, as
Table 4 presents the overall experimental composition of the formations of small coalesced drops of the oil phase distributed
mixtures and the corresponding tie lines for the system in the solvent phase, and these remained until the end of the
containing refined rice bran oil (1) + anhydrous ethanol (2) + rest time (24 h). It can be explained due to the similarity in
water (3) + hexane (4) at (298.15 ( 0.2) K with mass fractions phase density (Table 6). Also in the last one, which corresponds
of (4.14, 9.84, and 14.79) % of water in anhydrous ethanol. to the tie line with 19.99 % of hexane in the global composition,
Figure 1 shows a comparative diagram of the experimental phase inversion occurred.
points of the liquid-liquid equilibrium for the three pseudoqua- Tables 7 and 8 present the comparison between content of
ternary systems at (298.15 ( 0.2) K with different mass fractions water determined by Karl Fischer titration and by the NIR
of water in anhydrous ethanol. Through Figure 1, the effect of method for the pseudoquaternary systems determined at (298.15
water content in the hexane distribution coefficient can be seen. and 313.15) K, respectively, with different contents in mass of
Whereas the amount of water (highly polar) in the system water in anhydrous ethanol. It is possible to observe a good
increases, the hexane (apolar) tends to dislocate to the oil phase agreement between the analyses for both phases, but as water
(apolar). The increase of the water in the system promotes an is presented in a very small mass fraction in the systems and
increase of the biphasic region, due to the decrease of the mutual any fluctuations on the measurement give high deviations.
solubility between refined rice bran oil and solvent, ethanol + To avoid phase separation in the samples of liquid-liquid
equilibrium experiments at higher temperatures due to the
decrease of temperature, a dilution technique was tested. This
technique consists of diluting the samples with an inert solvent
to keep them homogeneous.
First, to test this dilution technique, a well-known pseudot-
ernary system, rice bran oil + anhydrous ethanol + hexane at
298.15 K,17 was selected, and two samples of each phase were
collected. One of these samples followed the usual NIR method.
The other one was diluted with tetrahydrofuran in approximately
50 % in mass, before measurement by the NIR method. The
composition of the diluted samples was calculated in free basis
of tetrahydrofuran. The results showed a good agreement
between both samples.
A second test was done with a well-known system at higher
temperature, rice bran oil + anhydrous ethanol + hexane at
313.15 K,17 to evaluate the effect of temperature in the dilution
technique. Here again, two samples of each phase were
Figure 1. Liquid-liquid equilibrium for the pseudoquaternary systems collected. The nondiluted sample was kept at the temperature
containing refined rice bran oil (1) + anhydrous ethanol (2) + water (3) +
hexane (4) at (298.15 ( 0.2) K: b, mass fraction of 4.14 % of water in
of the experiment to avoid phase separation in the quartz cell
anhydrous ethanol; s, tie lines; 9, mass fraction of 9.84 % of water of the NIR spectrometer. The diluted sample was analyzed at
in anhydrous ethanol; ----, tie lines; 2, mass fraction of 14.79 % of water in room temperature. The results again showed a good agreement
anhydrous ethanol; · · · · , tie lines. between both samples, and then the dilution technique could
2186 Journal of Chemical & Engineering Data, Vol. 54, No. 8, 2009

Table 4. Liquid-Liquid Equilibrium Data for the Pseudoquaternary Systems Containing Refined Rice Bran Oil (1) + Anhydrous Ethanol (2)
+ Water (3) + Hexane (4) at (298.15 ( 0.2) K with Mass Fractions of (4.14, 9.84, and 14.79) % of Water in Anhydrous Ethanol (100 w3′) in
Overall Composition
overall composition solvent phase oil phase
100 w3′ 100 w1 100 w2 100 w3 100 w4 100 w1 100 w2 100 w3 100 w4 100 w1 100 w2 100 w3 100 w4
4.14 49.97 47.93 2.10 0.00 0.68 95.07 4.25 0.00 90.90 8.94 0.16 0.00
47.46 47.93 2.10 2.51 0.82 92.91 4.23 2.04 87.54 9.39 0.19 2.88
44.99 47.92 2.09 5.00 1.00 91.22 4.22 3.56 83.65 9.51 0.19 6.65
40.00 47.93 2.09 9.98 1.40 86.91 4.19 7.50 76.83 10.40 0.25 12.52
35.00 47.92 2.09 14.99 1.81 82.85 4.15 11.19 69.43 11.38 0.29 18.90
30.09 47.93 2.08 19.90 2.22 78.52 4.10 15.16 62.07 12.61 0.35 24.97
9.84 49.97 45.08 4.95 0.00 0.84 88.55 10.61 0.00 93.77 5.99 0.24 0.00
47.47 45.09 4.94 2.50 0.63 86.79 10.48 2.10 90.60 6.10 0.25 3.05
44.97 45.09 4.94 5.00 0.62 85.53 10.42 3.43 86.74 6.32 0.25 6.69
40.01 45.08 4.94 9.97 0.64 83.06 10.32 5.98 78.78 6.80 0.26 14.16
35.00 45.08 4.94 14.98 0.56 80.39 10.22 8.83 70.51 7.29 0.26 21.94
30.09 45.08 4.93 19.90 0.46 78.12 10.11 11.31 62.32 7.84 0.27 29.57
14.79 49.98 42.60 7.42 0.00 0.05 83.57 16.38 0.00 94.87 4.89 0.24 0.00
47.47 42.61 7.42 2.50 0.05 81.96 16.24 1.75 91.44 4.97 0.24 3.35
44.97 42.60 7.42 5.01 0.05 81.07 16.22 2.66 87.36 5.17 0.25 7.22
39.99 42.60 7.42 9.99 0.05 79.97 16.18 3.80 79.02 5.46 0.24 15.28
34.99 42.60 7.41 15.00 0.06 78.75 16.15 5.04 70.35 5.73 0.24 23.68
30.05 42.56 7.40 19.99 0.06 78.18 16.09 5.67 61.79 5.84 0.24 32.13

Table 5. Relative Deviations for the Global Mass Balance and for Table 7. Comparison between Mass Fractions of Water
Mass Balance of Each Compound in Equilibrium of the Determined by Karl Fischer Titration and by the NIR Method for
Pseudoquaternary Systems Containing Refined Rice Bran Oil (1) + the Pseudoquaternary Systems Containing Refined Rice Bran Oil (1)
Anhydrous Ethanol (2) + Water (3) + Hexane (4) at (298.15 ( 0.2) + Anhydrous Ethanol (2) + Water (3) + Hexane (4) at (298.15 (
K with Mass Fractions of (4.14, 9.84, and 14.79) % of Water in 0.2) K with Mass Fractions of (4.14, 9.84, and 14.79) % of Water in
Anhydrous Ethanol (100 w3′) and at (313.15 ( 0.2) K with Mass Anhydrous Ethanol (100 w3′) in Overall Composition
Fraction of 4.40 % of Water in Anhydrous Ethanol (100 w3′) in
100 w3′ ) 4.14 % 100 w3′ ) 9.84 % 100 w3′ ) 14.79 %
Overall Composition
tie lines Karl Karl Karl
average relative deviation/% and phases Fischer ER/% Fischer ER/% Fischer ER/%
100 oil ethanol water hexane global TL 1a SPb 4.27 0.47 10.25 3.51 17.02 3.76
T/K w3′ (1) (2) (3) (4) deviation/% OPb 0.19 15.79 0.26 7.69 0.32 25.0
298.15 ( 0.2 4.14 0.04 0.02 5.37 1.13 0.14 TL 2 SP 4.23 0.01 10.11 3.66 16.94 4.13
9.84 0.10 0.12 6.97 1.88 0.41 OP 0.2 5.00 0.27 7.41 0.33 27.27
14.79 0.49 0.28 9.52 2.89 0.40 TL 3 SP 4.16 1.44 10.0 4.20 15.81 2.59
313.15 ( 0.2a 4.40 0.25 0.08 12.33 3.62 0.12 OP 0.21 9.52 0.27 7.41 0.30 16.67
TL 4 SP 4.86 13.79 9.81 5.20 16.86 4.03
a OP 0.28 10.71 0.29 10.34 0.34 29.41
System determined with dilution of analytical samples with tetra-
hydrofuran (5). TL 5 SP 4.63 10.37 9.53 7.24 15.02 7.52
OP 0.32 9.38 0.32 18.75 0.33 27.27
Table 6. Density of the Phases in Equilibrium for the Systems TL 6 SP 4.24 3.30 9.43 7.21 14.91 7.91
OP 0.37 5.41 0.34 20.59 0.30 20.0
Containing Refined Rice Bran Oil (1) + Anhydrous Ethanol (2) +
Water (3) + Hexane (4) at (298.15 ( 0.2) K with Mass Fraction of a
TL is tie line. b SP is solvent phase and OP is oil phase.
14.79 % of Water in Anhydrous Ethanol (100 w3′) in Overall
Composition Table 8. Comparison between Mass Fractions of Water Determined
F/g · cm-3 by Karl Fischer Titration and by the NIR Method for the
Pseudoquaternary Systems Containing Refined Rice Bran Oil (1) +
overall composition in hexane (100 w4) solvent phase oil phase Anhydrous Ethanol (2) + Water (3) + Hexane (4) at (313.15 ( 0.2)
15.00 0.81716 0.83285 K with Mass Fraction of 4.40 % of Water in Anhydrous Ethanol
19.99 0.81344 0.80862 (100 w3′) in Overall Composition
tie line and phases Karl Fischer ER/%
TL 1a SPb 4.61 4.56
be used in a new liquid-liquid equilibrium experiment at high OPb 0.34 32.35
temperature. The relative deviations for mass balance of TL 2 SP 4.49 6.68
components for the system using the dilution technique were OP 0.36 22.22
TL 3 SP 4.39 8.43
lower than those found by Priamo et al.17 OP 0.39 15.38
Table 9 presents the results of the calibration models TL 4 SP 4.56 4.61
generated by the NIR method for the system containing refined OP 0.40 12.5
TL 5 SP 3.99 13.78
rice bran oil (1) + anhydrous ethanol (2) + water (3) + hexane OP 0.46 2.17
(4) at (313.15 ( 0.2) K, with a mass fraction of 4.40 % of TL 6 SP 3.95 13.67
water in anhydrous ethanol, using the dilution technique of OP 0.51 11.76
analytical samples with tetrahydrofuran (5): the low errors in a
TL is tie line. b SP is solvent phase and OP is oil phase.
the quantification for the external validation standards (ER) and
for the cross-validation standards (rmsecv).
The calibration and external validation standards generated bran oil (1) in solvent phase presented high value error, due to
for the system containing refined rice bran oil (1) + anhydrous small fraction when this is submitted to the dilution.
ethanol (2) + water (3) + hexane (4) with mass fraction of In Table 10 are shown the overall experimental composition
4.14 % of water in anhydrous ethanol at (298.15 ( 0.2) K were of the mixture and the corresponding tie lines for the system
diluted with tetrahydrofuran (5) at dilution factor well-known containing refined rice bran oil (1) + anhydrous ethanol (2) +
of approximately 50 % (in mass fraction). Only the refined rice water (3) + hexane (4) at (315.15 ( 0.2) K, with a mass fraction
 Journal of Chemical & Engineering Data, Vol. 54, No. 8, 2009 2187

Table 9. Relative Errors (ER) and Root-Mean-Square Error of Cross-Validation (rmsecv) in the Quantification of the Validation Standards
(External, ER, and Internal, rmsecv) Using the NIR Method for the System Containing Refined Rice Bran Oil (1) + Anhydrous Ethanol (2) +
Water (3) + Hexane (4) at (313.15 ( 0.2) K with a Mass Fraction of 4.40 % of Water in Anhydrous Ethanol (100 w3′) in Overall Composition
solvent phase oil phase
100 w3′ oil ethanol water hexane oil ethanol water hexane ER/%
4.40 ER/% 7.25 0.04 0.33 1.27 0.28 0.39 2.11 0.64 1.54
rmsecv/% 0.10 0.06 0.01 0.13 0.13 0.03 0.01 0.10

Table 10. Liquid-Liquid Equilibrium Data for the Pseudoquaternary System Containing Refined Rice Bran Oil (1) + Anhydrous Ethanol (2)
+ Water (3) + Hexane (4) at (313.15 ( 0.2) K with Mass Fraction of 4.40 % of Water in Anhydrous Ethanol (100 w3′) in Overall Composition
overall composition solvent phase oil phase
100 w1 100 w2 100 w3 100 w4 100 w1 100 w2 100 w3 100 w4 100 w1 100 w2 100 w3 100 w4
49.97 47.81 2.22 0.00 4.54 90.64 4.82 0.00 87.20 12.35 0.45 0.00
47.46 47.80 2.22 2.52 4.15 89.59 4.79 1.47 83.33 13.15 0.44 3.08
44.97 47.80 2.22 5.01 4.67 87.14 4.76 3.43 79.80 13.77 0.45 5.98
39.93 47.88 2.22 9.97 4.74 83.08 4.77 7.41 72.61 14.97 0.45 11.97
34.99 47.80 2.22 14.99 6.39 77.98 4.54 11.09 64.89 17.33 0.45 17.33
30.08 47.80 2.21 19.91 7.05 72.84 4.49 15.62 57.34 18.85 0.45 23.36

of 4.40 % of water in anhydrous ethanol and using the dilution in the selection of calibration and validation standards and
technique of the analytical samples with tetrahydrofuran (5). sensitivity in reading and interpretation of the near-infrared
In this work, equilibrium data were not determined for spectra in samples containing high water content at a temperature
systems at 313.15 K containing mass fractions higher than higher than ambient temperature. Lanza et al.15 determined
4.40 % of water in anhydrous ethanol because of the complexity liquid-liquid equilibrium data in temperatures higher than
298.15 K for pseudoternary systems in temperatures of 313.15
K, 318.15 K, and 328.15 K and concluded that quantifying
systems containing more than four components using NIR
spectroscopy for systems at higher temperatures than ambient
temperature leads to an increase in analytical error, and therefore
the determination of good results would be affected.
The relative deviations for the global mass balance and for
the mass balance of each compound in the equilibrium of the
pseudoquaternary system at (313.15 ( 0.2) K, with a dilution
technique with tetrahydrofuran (5), are shown in the Table 5. It
is possible to observe high relative deviations for the components
water (3) and hexane (4), but these are in small mass fraction
in the systems and diluted to measurement. However, the other
components presented low relative deviations, substantiated the
viability of the use of the dilution technique, and also measure-
ment on NIR spectroscopy allied to chemometric in complex
Figure 2. Liquid-liquid equilibrium for the pseudoquaternary systems systems and temperature higher than 298.15 K.
containing refined rice bran oil (1) + anhydrous ethanol (2) + water (3) +
hexane (4) at (298.15 ( 0.2) K. b, mass fraction of 4.14 % of water in The experimental data sets for all pseudoquaternary systems
anhydrous ethanol: s, NRTL model; 9, mass fraction of 9.84 % of water were correlated by the NRTL model (Figures 2 and 3), and the
in anhydrous ethanol; ----, NRTL model; 2 mass fraction of 14.79 % of NRTL parameters are presented in Table 11.
water in anhydrous ethanol; · · · · , NRTL model. Figure 2 shows a comparative diagram of the experimental
points of the liquid-liquid equilibrium for three pseudoquater-
nary systems at (298.15 ( 0.2) K with mass fractions of (4.14,
9.84, and 14.79) % of water in anhydrous ethanol described by
the NRTL model. Figure 3 presents the calculated and experi-
mental data of the pseudoquaternary system containing refined
rice bran oil (1) + anhydrous ethanol (2) + water (3) + hexane
(4) at 313.15 K, with mass fraction of 4.40 % of water in
anhydrous ethanol.
The average deviations between the experimental and calculated
compositions in both phases were calculated according to Lanza
et al.15,16 and Priamo et al.17 and are shown in Table 12.
As can be seen in Figure 2, the NRTL model correctly
describes the experimental data for the pseudoquaternary
systems at 298.15 K, with an average overall deviation of
0.95 % (Table 12). For the system at 313.15 K, the NRTL model
Figure 3. Liquid-liquid equilibrium for the pseudoquaternary systems
was also representative according to the low average overall
containing refined rice bran oil (1) + anhydrous ethanol (2) + water (3) + deviation of 0.55 %. Thus, the very low values of relative
hexane (4) at (313.15 ( 0.2) K, with mass fraction of 4.40 % of water in deviations for the mass balance and low deviations found in
anhydrous ethanol: 2, oil phase; 9, solvent phase; b, initial mixture/overall; the thermodynamic modeling using the NRTL model confirm
· · · · , NRTL model. the excellent quality of the experimental data.
2188 Journal of Chemical & Engineering Data, Vol. 54, No. 8, 2009

Table 11. NRTL Parameters for the Binary Interaction between (8) Cvengroš, J. Physical Refining of Edible Oils. J. Am. Oil. Chem. Soc.
Refined Rice Bran Oil (1), Anhydrous Ethanol (2), Water (3), and 1995, 72, 1193–1196.
Hexane (4) at Temperatures of (298.15 and 313.15) K (9) Bhosle, B. M.; Subramanian, R. New Approaches in Deacidification
of Edible Oils - A Review. J. Food Eng. 2005, 69, 481–494.
T/K pair ij Aij/K Aji/K Rij (10) Gonçalves, C. B.; Pessôa Filho, P. A.; Meirelles, A. J. A. Partition of
298.15 12a 873.64 1416.8 0.49874 Nutraceutical Compounds in Deacidification of Palm Oil by Solvent
13a -26.977 4624.6 0.16580 Extraction. J. Food Eng. 2007, 81, 21–26.
14 -689.80 160.89 0.50542 (11) Rodrigues, C. E. C.; Onoyama, M. M.; Meirelles, A. J. A. Optimization
23a -10.984 -173.64 0.15018 of the Rice Bran Oil Deacidification Process by Liquid-Liquid
24b 313.48 37.076 0.27200 Extraction. J. Food Eng. 2006, 73, 370–378.
34 1701.2 1565.0 0.20000 (12) Rodrigues, C. E. C.; Silva, F. A.; Marsaioli, A., Jr.; Meirelles, A. J. A.
313.15 12 -738.33 2295.1 0.26732 Deacidification of Brazil Nut and Macadamia Nut Oils by Solvent
13 -134.28 1301.8 0.69985 Extraction: Liquid-Liquid Equilibrium Data at 298.2 K. J. Chem. Eng.
14 -695.22 762.81 0.28831 Data 2005, 50, 517–523.
23 2429.1 -546.04 0.70000 (13) Batista, E.; Monnerat, S.; Kato, K.; Stragevitch, L.; Meirelles, A. J. A.
24b 38.0 1148.1 0.27200 Liquid-Liquid Equilibrium for Systems of Canola Oil, Oleic Acid,
34 3000.0 1019.5 0.20347 and Short-Chain Alcohols. J. Chem. Eng. Data 1999, 44, 1360–1364.
a (14) Batista, E.; Monnerat, S.; Stragevitch, L.; Pina, C. G.; Gonçalves, C. B.;
Parameters taken from Rodrigues et al.28 b
Parameters taken from Meirelles, A. J. A. Prediction of Liquid-Liquid Equilibrium for Systems
Lanza et al.15 of Vegetable Oils, Fatty Acids and Ethanol. J. Chem. Eng. Data 1999,
44, 1365–1369.
Table 12. Average Deviations between the Experimental and (15) Lanza, M.; Sanaiotti, G.; Batista, E.; Poppi, R. J.; Meirelles, A. J. A.
Calculated by NRTL Phase Compositions Liquid-Liquid Equilibrium Data for Systems Containing Vegetable
system 100 ∆w Oils, Anhydrous Ethanol, and Hexane at (313.15, 318.15, and 328.15)
K. J. Chem. Eng. Data 2009, 54, 1850–1859.
refined rice bran oil (1) + anhydrous ethanol (2) + 1.22 (16) Lanza, M.; Borges, N. W.; Batista, E.; Poppi, R. J.; Meirelles, A. J. A.
water (3) + hexane (4) (mass fraction of 4.14 % of Liquid-Liquid Equilibrium Data for Reactional Systems of Ethanolysis
water in anhydrous ethanol) at (298.15 ( 0.2) K at 298.3 K. J. Chem. Eng. Data 2008, 53, 5–15.
refined rice bran oil (1) + anhydrous ethanol (2) + 0.70 (17) Priamo, W. L.; Lanza, M.; Meirelles, A. J. A.; Batista, E. A. C. Liquid-
water (3) + hexane (4) (mass fraction of 9.84 % of Liquid Equilibrium Data for Fatty Systems Containing Refined Rice
water in anhydrous ethanol) at (298.15 ( 0.2) K Bran Oil, Oleic Acid, Anhydrous Ethanol and Hexane. J. Chem. Eng.
refined rice bran oil (1) + anhydrous ethanol (2) + 0.83 Data 2009, DOI: 10.1021/je9002338.
water (3) + hexane (4) (mass fraction of 14.79 % of (18) AOCS. Official Methods and Recommended Practices of the American Oil
water in anhydrous ethanol) at (298.15 ( 0.2) K Chemists’ Society, 3rd ed.; AOCS Press: Champaign, 1988; Vols. 1-2.
average overall deviation for the systems at 298.15 K 0.95 (19) Hartman, L.; Lago, R. C. A. Rapid Preparation of Fatty Acid Methyl
refined rice bran oil (1) + anhydrous ethanol (2) + 0.55 Esters from Lipids. Lab. Pract. 1973, 22, 475–476.
water (3) + hexane (4) (mass fraction of 4.40 % of (20) Antoniosi Filho, N. R.; Mendes, O. L.; Lanças, F. M. Computer
water in anhydrous ethanol) at (313.15 ( 0.2) K Prediction of Triacylglycerol Composition of Vegetable Oils by
HRGC. Chromatographia 1995, 40, 557–562.
(21) Silva, L. H. M.; Coimbra, J. S.; Meirelles, A. J. A. Equilibrium Phase
Conclusions Behavior of Poly (Ethylene Glycol) + Potassium Phosphate + Water
Two Phase Systems at Various pH and Temperatures. J. Chem. Eng.
Phase equilibrium data with good accuracy for the system Data 1997, 42, 398–401.
refined rice bran oil + anhydrous ethanol + water + hexane were (22) AOCS. Official and TentatiVe Methods of the American Oil Chemists’
experimentally obtained at (298.15 and 313.15) K using NIR Society, 3rd ed.; AOCS Press: Champaign, 1993; Vol. 1.
spectroscopy for quantification of the phase compositions. The use (23) Brereton, R. G. Introduction to Multivariate Calibration in Analytical
of dilution of the analytical samples with an inert solvent, mainly Chemistry. Analyst 2000, 125, 2125–2154.
in experiments at higher temperature than environment, demon- (24) Otto, M. Chemometrics: Statistics and Computer Application in
Analytical Chemistry; Wiley, 1999.
strates a viable alternative to get good results. Low deviations are (25) Marcilla, A.; Ruiz, F.; Garcı́a, A. N. Liquid-Liquid-Solid Equilibria
obtained in the global mass balance, and the effect of temperature of the Quaternary System Water-Ethanol-Acetone-Sodium Chloride
in the use of NIR spectroscopy is minimized. at 25 °C. Fluid Phase Equilib. 1995, 112, 273–289.
Despite of complexity of the studied systems, the estimated (26) Rodrigues, C. E. C.; Peixoto, E. C. D.; Meirelles, A. J. A. Phase
parameters for the NRTL model are representative since the Equilibrium for Systems Composed by Refined Soybean Oil +
Commercial Linoleic Acid + Ethanol + Water at 323.2 K. Fluid Phase
description of the liquid-liquid equilibrium for all the systems Equilib. 2007, 261, 121–128.
presented deviations lower than 1.22 % in relation to the (27) Gonçalves, C. B.; Meirelles, A. J. A. Liquid-Liquid Equilibrium Data
experimental data. These parameters are useful in the design for the System Palm Oil + Fatty Acids + Ethanol + Water at 318.2
and simulation of liquid-liquid extractors, in the refining K. Fluid Phase Equilib. 2004, 221, 139–150.
process of the rice bran oil, and in the production of biodiesel. (28) Rodrigues, C. E. C.; Antoniassi, R.; Meirelles, A. J. A. Equilibrium
Data for the System Rice Bran Oil + Fatty Acids + Ethanol + Water
at 298.2 K. J. Chem. Eng. Data 2003, 48, 367–373.
Literature Cited (29) Rodrigues, C. E. C.; Pessôa Filho, P. A.; Meirelles, A. J. A. Phase Equilibria
(1) Rukmini, C. Chemical, Nutritional and Toxicological Studies of Rice for the System Rice Bran Oil + Fatty Acids + Ethanol + Water +
Bran Oil. Food Chem. 1988, 30, 257–268. γ-Oryzanol + Tocols. Fluid Phase Equilib. 2004, 216, 271–283.
(2) Deckere, E. A. M.; Korver, O. Minor Constituents of Rice Bran Oil (30) Stragevitch, L.; d’Ávila, S. G. Application of a Generalized Maximum
as Functional Foods. Nutr. ReV. 1996, 54, 120S–126S. Likelihood Method in the Reduction of Multicomponent Liquid-Liquid
(3) Orthoefer, F. T. Rice Bran Oil: Healthy Lipid Source. Food Technol. Equilibrium Data. Braz. J. Chem. Eng. 1997, 14, 41–52.
1996, 50, 62–64. (31) Null, H. R. Phase Equilibrium in Process Design; Wiley-Interscience:
(4) Resa, J. M.; González, C.; Fanega, M. A.; Ortiz de Landaluce, S.; New York, 1970.
Lanz, J. Enthalpies of Mixing, Heat Capacities, and Viscosities of
Alcohol (C1-C4) + Olive Oil Mixtures at 298.K. J. Food Eng. 2002,
51, 113–118. Received for review March 4, 2009. Accepted May 28, 2009. The
(5) Liu, J.; Shi, M.; Diosady, L. L.; Rubin, L. J. Three-Phase Extraction of Chinese authors wish to acknowledge CAPES (Coordenação de Aperfeiçoamento
Rapeseed Using the Karr Column. J. Food Eng. 1995, 24, 35–45. de Pessoal de Nı́vel Superior), FAPESP (Fundação de Amparo à
(6) Rodrigues, C. E. C. Deacidification of Rice Bran Oil by Liquid-Liquid Pesquisa do Estado de São Paulo, Processes 08/56258-8, 06/01864-5,
Extraction. Doctoral Thesis (D.Sc.) (in Portuguese), University of 05/53095-2, and 04/09591-2), and CNPq (Conselho Nacional de
Campinas, Campinas, SP: Brazil, 2004. Desenvolvimento Cientı́fico e Tecnológico, Process 306250/2007-1) for
(7) Ceriani, R.; Costa, A. M.; Meirelles, A. J. A. Optimization of the their financial support.
Ph,ysical Refining of Sunflower Oil Concerning the Final Contents
of trans-Fatty Acids. Ind. Eng. Chem. Res. 2008, 47, 681–692. JE900234U