Comparison between ethanol and hexane for oil extraction from Quercus suber L . fruits By

El objetivo de este estudio fue (i) evaluar la posibilidad de reemplazar el n-hexano por el etanol para la extracción del aceite de los frutos de Quercus suber y (ii) optimizar, a escala de laboratorio, la preparación de la muestra y el tiempo de extracción para ambos solventes (n-hexano vs. etanol anhidro). Para ambos solventes, se ha evaluado el efecto del proceso de acondicionamiento en el rendimiento de la extracción. Para ello, se ha empleado un diseño factorial completo como función de cuatro variables: descascarillado (con vs sin cascarilla), y tratamiento térmico del material triturado (40C vs. 75C), a diferentes tiempos (5 min vs. 120 min) y presiones (10 kPa vs. 100 kPa). Los rendimientos más elevados de extracción de aceite se obtuvieron con n-hexano cuando los frutos descascarillados se acondicionaron a presión atmosférica. Cuando el material descascarillado fue tratado a 75C por 90 min, se obtuvieron mejores rendimientos con n-hexano. El etanol demostró no ser adecuado para la extracción del aceite de los frutos de Quercus, debido a la extracción conjunta de otros materiales a la vez del aceite.


INTRODUCTION
Quercus genus plants are characteristic of the Mediterranean Flora.For instance, Portugal has more than one million hectares with Quercus trees which represents an annual production of approximately 300,000 tons of acorns.This production is close to the annual production of wheat, higher than the production of other grains (barley, oats, rye and rice) and accounts for about half of the production of corn in this country.The extraction of the oil of Quercus fruits in a profitable way could reduce the importation of edible oils.The oil content of the fruits from Quercus rotundifolia Lam. may go up to about 14% (Ferrão and Ferrão, 1988).Until the early seventies, the extraction of the oil from Q. rotundifolia acorns was carried out in some oil extraction plants in Portugal (Ferrão and Ferrão, 1988).In spite of the low oil content of the fruits from Quercus suber (4.4%-9.1%),this oil has a fatty acid composition similar to olive oil (Ferrão and Ferrão, 1988) and, therefore, its extraction would increase revenues to the Quercus suber industry.The exploitation of this tree mainly lies on the use of its cork.Moreover, the extracted meal could be used as animal feedstuff (Ferrão and Ferrão, 1988).
Currently, hexane, a solvent obtained from petrochemical sources, is the solvent used for oil extraction.This solvent can be emitted during extraction and recovery and has been identified as an air pollutant since it can react with other pollutants to produce ozone and photochemical oxidants (Wan et al., 1995a;Hanmoungjai et al., 2000).Safety, environmental and health concerns have increased the interest in alternative solvents to hexane to reduce the emissions of volatile organic compounds to the atmosphere as well as potential traces of hexane in edible oils after refining.
As an alternative to organic solvent extraction, the extraction of oil from oil-containing materials with aqueous solutions (acidic or alkaline at 45 o C to 85 o C) has been investigated (Rhee et al., 1972;Hagenmaier et al., 1973;Hagenmaier, 1974;Lusas et al., 1982;Kim, 1989;Southwell and Harris, 1992;Rosenthal et al., 1996;Rosenthal et al., 1998;Hanmoungjai et al., 2000).However, due to the low selectivity of aqueous solutions for lipids, simultaneous extraction of proteins, carbohydrates and other compounds also occur.Also, aqueous extraction is limited by (i) the lower efficiency of oil extraction, (ii) the need of an additional de-emulsification step to recover the oil and (iii) the production of an aqueous effluent which requires further treatment (Hanmoungjai et al., 2000).
The most feasible alternative to hexane extraction seems to be the replacement of this solvent by other organic solvents recognized as environmentally safer.Several studies have been carried out, both at laboratory and pilot scales, aimed to replace hexane with other hydrocarbons (Wan et al., 1995a;Wan et al., 1995b;Conkerton, et al., 1995) or alcohols (Abraham et al., 1988;Rittner, 1992;Hron et al., 1994;Sineiro et al., 1998) as solvents for oil extraction.Among hydrocarbon solvents, heptane and isohexane were recommended as potential substitutes for hexane to extract oil from cottonseed (Wan et al., 1995a;Wan et al., 1995b;Conkerton et al., 1995).With respect to the use of alcohols, isopropanol and ethanol are the most promising solvents for the oil extraction from cottonseed (Abraham et al., 1988;Hron et al., 1994), sunflower seed (Sineiro et al., 1998) and soybean (Baker and Sullivan, 1983;Rittner, 1992).Ethanol is a worthy candidate to investigate as an alternative solvent since its cost is low and it may be produced from a large variety of biological materials using simple technology.In addition, although flammable (flash point= 8.9 o C; ignition temperature= 425 o C), this alcohol is recognized as non-toxic and has less handling risks than hexane (flash point = -23 o C; ignition temperature= 225 o C) (Rittner, 1992).It can also be obtained by fermentation and therefore labelled as ''natural''.The use of this alcohol as an extraction solvent also avoids eventual toxicity problems of meals for animal feedstuff.
The aim of this study was to (i) evaluate the feasibility of replacing n-hexane with ethanol for the extraction of the oil from Quercus suber fruits and (ii) to optimize the extraction conditions, namely preparation of samples (dehulling and conditioning), and extraction time for both solvents used (n-hexane v.s.ethanol), at laboratory scale.
For seeds with high oil content, a conditioning treatment (''cooking'') is currently carried out prior to mechanical pressing/expelling oil extraction.In general, a thermal treatment (conditioning technique) of seeds or other fatty materials with low oil content (lower than 20-25%), prior to solvent extraction, may also increase oil yield and quality (Hoffman, 1989).For instance, the best extraction conditions for cracked soybeans are achieved after heating at 55-75 o C for about 20-30 min to reach 11% moisture.For cottonseed, the best results are obtained with a treatment at 65 o C for 10-12 min (Hoffman, 1989).
In this study, the effect of conditioning on the yield of extracted oil was investigated, both for n-hexane and ethanol.Therefore, a full factorial design was used as a function of the following variables: dehulling, thermal treatment of the crushed material at different times and pressures.

Materials
Ripened fruits from a Quercus suber L. tree were obtained in Lisbon, Portugal.Anhydrous ethanol p.a. and n-hexane p.a. were used.

Size reduction
The fruits (with or without husks) were crushed in a coffee mill (knife cutter type) for about 20 sec.

Moisture content of the fruits
Samples (5g) of both whole and dehulled crushed fruits were dried at 103 o C to attain a constant mass.Moisture content was expressed on a dry basis.Analyses were run in triplicate.

Optimization of fruit conditioning
The effect of dehulling the fruits, conditioning temperature (40 o C vs.75 o C), time (5 min vs.120 min) and pressure (10 kPa vs.100 kPa) on the yield in oil was investigated, both when n-hexane or ethanol were used as extraction solvents.A full factorial design 2 4 (2 levels and 4 factors) was followed and Table I contains the decoded experimental design (2 4 = 16 experiments).
With this experimental design, several variables (factors) are tested simultaneously with a minimum number of trials.In addition, variables and interactions with significant effect on the extraction yield will be identified (Gacula and Singh, 1984;Haaland, 1989;Montgomery, 1991).
For these experiments, the extracts were obtained in a Soxhlet apparatus for 5 h at the boiling points of the solvents tested (78 o C and 67-69 o C, for ethanol and n-hexane, respectively).A ratio of crushed fruits to solvent of 1:6 (m/v) was used.Subsequently, the solvent was evaporated under reduced pressure and the amount of extracted material evaluated by weighting.The yields of extract (%) were expressed on a dry basis, i.e., mass of extract per mass of dry matter.
After full factorial experiments, additional experiments were carried out in an attempt to determine the near-optimum extraction time with n-hexane.In these experiments, dehulled Quercus acorns were used and conditioning occurred at 75 o C under normal pressure.

Water activity assay
The water activity (aw) of the crushed fruits after conditioning was measured at 25 o C in a Rotronic Hygroskop DT with a lithium chloride humidity sensor (DMS-100H).

Optimization of extraction time
After defining the best conditions for sample preparation for oil extraction, the extraction time was optimized.The oil was obtained from the fruits by solvent (n-hexane p.a.) extraction in a Soxhlet apparatus.To optimize the extraction time, the solution was recovered in a flask, at different extraction times, the solvent evaporated under reduced pressure and the extracted oil quantified.Experiments were carried out in duplicate and the results were expressed on a dry basis.

Statistical analysis
The analysis of the results of the full factorial experiments was performed by using the software "Statistica TM ", version 5, from Statsoft, USA.
The linear effects of each of the 4 factors under study (dehulling, temperature, pressure and conditioning time), as well as their linear interactions, on the yield in oil were calculated, when both ethanol and n-hexane were used.Their significance was evaluated by analysis of variance.The results of the full factorial experiments were used to establish first-order models.First-order coefficients were generated by regression analysis.The goodness of fit of the models was evaluated by the determination

Moisture content of the fruits
The average moisture content (dry basis) of whole and dehulled Quercus suber fruits were 57.3% (σ=0.04) and 59.4% (σ=0.03),respectively.These results were used to express the amount of extracts obtained with ethanol or n-hexane on a dry basis.
Optimization of the conditioning operations of the fruits: Both the decoded matrix and the obtained results (i.e., aw values of crushed fruits, after conditioning and before extraction, and the amount of extracted material by n-hexane or ethanol) are shown in Table I.These data were used to calculate the significant linear effects of each variable and their interactions on the experimental responses to select the best operation conditions (Table II).
With respect to aw values of the crushed fruits, after conditioning and before extraction, only conditioning time had a significant and negative effect on them.As expected, the water activity decreased when longer drying times were used.The other factors tested had no significant effect on aw values.
When hexane was used for oil extraction, the temperature and conditioning time, as well as the interaction (Pressure) x (Time), had a significant and a positive effect on the yield of oil (Table II).The interaction (Husks) x (Temperature) showed to have a negative significant effect on the extraction yield.Therefore, the fruits should be dehulled, crushed and conditioned under atmospheric pressure, at higher temperatures and times, prior to hexane extraction.
Both for water activity values and oil extracted by n-hexane, high values of R 2 and R 2 adj were obtained for the models fitted to the experimental data points.However, the goodness of fit and therefore the selection of the most suitable model should not be done on the basis of R 2 and R 2 adj values alone.The R 2 and R 2 adj indicators must be complemented by For the extraction of oil with n-hexane from dehulled Quercus fruits, the following first order polynomial equation can be used to estimate the yield in oil: Oil Yield = 4.06 + 0.89 T + 0.46 t + 0.56 (P x t) where the oil is expressed in mass percentage on a dry basis, T is the temperature ( o C); t is time (min) and P, pressure (kPa) of conditioning of the acorns prior to solvent extraction.
In addition, complementary experiments were carried out in an attempt to determine the near-optimum conditioning time, for the oil extraction with n-hexane, when the highest temperature tested (75 o C) was used (Table III).The best yield in oil was obtained after a treatment at 75 o C for 90 min.
When ethanol was used, the amounts of extracts obtained were from 4 to 12 times higher than the extracts achieved with n-hexane (reference solvent) (Table I).However, in spite of the observed variation, no significant effects of the conditioning factors tested were observed on ethanol extraction (Table II).This may be ascribed to a lower selectivity of ethanol for the oil with a consequent extraction of other compounds such as phosphatides, polyphenols, pigments and soluble sugars (Hron et al., 1982;Hron et al., 1994;Pomeranz and Meloan, 1994;Sineiro et al., 1996).In fact, in acorns from Q. rotundifolia, contents of total phenolic compounds up to 6.8%, in acorn pulp, and 9.3%, in acorn hulls, were quantified Extraction Time (h) Extracted Oil (%, w/w) (Bruno-Soares et al., 2000).Ethanol showed not to be an adequate solvent for the direct extraction of the oil from Quercus acorns.

Optimization of the oil extraction time
Due to the low selectivity of ethanol for oil extraction from Quercus fruits, the optimization of the extraction time was only carried out for n-hexane extraction.When this solvent was used in a Sohxlet equipment, after about a 8-hour extraction time, no significant increase in extracted oil was observed (Fig. 3).
Alternative organic solvents to hexane for oil extraction from Quercus fruits have to be investigated.
Design, 2 4 : the effect of different conditioning conditions (i) on the water activity (aw) values of crushed fruits, after conditioning and before extraction, and (ii) on the amount of extracted material (dry basis), when n-hexane or ethanol were used.(Factors: with (1) or without husks (0), temperature, T, time, t, and pressure, P) different conditioning factors and respective significance levels (α) on (i) the water activity (aw) of crushed fruits, after conditioning and before extraction, and (ii) on extract yield when either hexane or ethanol were used (* at α<0.05; ** at α<0.01; (a) not significant effects).Determination coefficients (R 2 ) and adjusted R 2 (R 2 adj) of the fitted models to experimental data other tools, such as the graphic plots of theoretical values predicted by the model vs. observed experimental values, to decide on the goodness of fitof the models(Doran, 1995).In fact, for hexane extraction, a linear relationship between the predicted values and the experimental values was observed (Fig.1), confirming the adequacy of the first order model.However, for the water activity values of the samples before extraction, experimental points are scattered around the theoretical line ''observed values vs predicted values'' (Fig.2) indicating a certain lack of fit of the model.

Figure 1
Figure 1 Relationship between the experimental values of the extracts obtained with n-hexane and the corresponding values estimated by a first-order model (line).