A simple procedure to evaluate the performance of fats and oils at frying temperatures

Se propone un procedimiento estándar para evaluar el comportamiento de aceites y grasas a temperaturas de fritura. En este procedimiento se utilizan las ventajas del aparato Rancimat, que permite el uso de tubos estándar, la corrección de la temperatura, en su caso, y la igualdad de temperatura en todos los tubos dadas las características del bloque de calentamiento. De los resultados obtenidos en muestras de 8 g de aceite calentadas a 180° C durante 10 h, analizadas por triplicado, se obtuvieron coeficientes de variación inferiores al 6% para la determinación de compuestos polares y polímeros. En caso de limitación en la cantidad de aceite, se propone utilizar 2 g de muestra, manteniendo similares valores para la relación superficie a volumen de aceite, lo que permite obtener valores de alteración y coeficientes de variación del mismo orden. Se analizan finalmente las ventajas globales del procedimiento y sus distintas posibilidades en la evaluación de grasas de fritura. Como ejemplo, se aplica el procedimiento a la evaluación del efecto de los antioxidantes naturales de los aceites de girasol.


INTRODUCTION
The frying process is influenced by a number of variables among which the type of process, i.e., continuous or discontinuous, surface-to-oil volume ratio, temperature, oil unsaturation degree, and presence of naturally occurring or added minor compounds, are of particular relevance (Gere, 1982 and1983;Boskou, 1988;Dobarganes and Márquez-Ruiz, 1996).Due to the difficulties encountered to define and/or control such variables and the additional strong interactions existing between them (Jorge etal., 1996a), it is not easy to replicate results from different laboratories and moreover, the conclusions drawn from data obtained under apparently similar conditions might differ greatly.
From the literature, it is generally reported that the main groups of compounds formed during frying are primarily due to the action of temperature and oxygen while diglycerides and free fatty acids, otherwise coming from triglyceride hydrolysis, are the least representative compounds even when frying foods of high moisture content (Sebedio et al., 1990;Pérez-Camino et ai., 1991;Arroyo etal., 1992;Cuesta etal., 1993;Dobarganes et ai, 1993;Arroyo et al., 1995;Jorge etal., 1996b).
A number of studies on thermoxidation and frying have been recently carried out in our laboratory to evaluate the performance of low-and high-oleic sunflower oils as substitutes for more saturated fats in continuous and discontinuous frying (Jorge et al., 1996a and1996b).Particularly, in laboratory frying experiments, results for total polar compounds were 20.4 and 17.2%, for conventional and high-oleic sunflower oils, respectively, after 6 hours under discontinuous frying conditions, and 11.7 and 7.0% under simulated continuous frying conditions (Jorge et al., 1996b).Nevertheless, highly variable values were obtained in thermoxidation experiments, depending on the conditions established (Jorge etal., 1996a).
In addition to the extensive information provided by these studies, and based on the results obtained, some complementary experiments have been undertaken in our laboratory with the aim of designing a simple test to evaluate performance of fats and oils under strictly controlled conditions.The procedure proposed simulates the usual conditions of the main variables of the discontinuous frying process, namely, temperature, surface-to-oil volume ratio, presence of air, and hence can be of great utility to compare results between laboratories.Additionally, changes in the variables controlled could be assayed in order to study their action in the process.
Conditions applied for HPSEC in both methods were as follows: A Waters 510 HPLC pump and a Rheodyne 7725i injector with a 10 |im sample loop (Waters Associates, Milford, MA, USA), a refractive index detector (Hewlett Packard, Pittsburg, PA, USA) and two 100 and 500 Â Ultrastyragel columns (Waters Associates, Milford, MA, USA) connected in series, were used.The columns were 25 cm x 0.77 cm inner diameter, packed with a porous, highly cross-linked styrenedivinylbenzene copolymer (<10 |im).High-performance liquid chromatography grade tetrahydrofuran served as the mobile phase with a flow of 1 mL/min.Sample solutions of around 50 mg of oil/mL and 15 mg of polar compounds/mL in tetrahydrofuran were used for analyses 1 and 2, respectively.

Heating procedure
8 ± 0.01 g of oil were weighed out in a Rancimat vessel and inserted in the heating block previously heated at 180 ± 1° C.After 10 hours-heating, samples were taken out and kept at -30° C until analyses.
For limited amounts of samples, standard glass tubes of 10 cm X 10 mm i.d. were used. 2 ± 0.01 g of sample were weighed directly into the tube, in turn introduced into the Rancimat reaction vessel containing 6 g of glycerol to facilitate heat transfer, and inserted in the heating block.Conditions applied were the same described above.
Rancimat instructions were carefully observed for glassware cleaning and temperature correction.No bubbling of air was applied during heating and the reaction vessels were left open.
2. Polar compounds and distribution of minor glyceridic compounds were determined by combination of adsorption chromatography and HPSEC (Dobarganes et al., 1988).Briefly, non-polar and polar fractions of oil samples were separated by silica column chromatography, eluting with a mixture 90:10 hexane: diethyl ether and diethyl ether, respectively.Polar fractions were further analyzed by HPSEC in order to quantitate triglyceride oligomers (TGO), triglyceride dimers (TGD), oxidized triglyceride

RESULTS AND DISCUSSION
Table I shows the results obtained for total polar compounds and polymers in quadruplicate samples of 2 and 8 g oils heated for 10 h at 180° C, under the conditions described in the experimental part.As can be observed, repeatability was very good, the coefficient of variation being lower than 6% under all circumstances (analytical method, oil unsaturation degree and amount of sample).Differences between the results obtained for 2 and 8 g samples can be attributed to small differences in the surface-to-oil volume ratio.This variable, showing a slightly higher value in 8 g samples, has been proved to exert an important influence on the oxidative degradation (Jorge etal., 1996a).
It is important to note that total variations found are due to both heating differences and deviations associated with the analytical method used.Considering that repeatability relative standard deviations in the range of the mean values obtained in this study have been reported to be on the order of 3% (lUPAC, 1987) and 2% (AOCS, 1994) for polymers and polar compounds determinations, respectively, it can be deduced that heating introduced minimal differences.
Table I i presents distribution of the main groups of compounds formed during heating for 10 hours.A similar distribution pattern was obtained as compared to those found for used frying fats with similar level of polar compounds (Jorge et al., 1996b).Thus, higher amounts of polar compounds and, in particular, polymers, were formed in the most unsaturated oil while hydrolytic products, i. e., DG and FA, did not significantly change, since similar levels were already present in the starting oils (data not shown).On the other hand, it is important to note that values for total polymers as determined directly in the entire oil samples (Table I) fitted very well with results obtained for dimers plus oligomers, as quantitated in total polar compound fractions (Table II) despite the substantial difference between both determinations, that is, the presence of the most abundant group of compounds constituted by non-polar triglycerides in the total sample, otherwise eliminated in fractions of isolated polar compounds.Therefore, such consistent results indirectly indicated that response factors for the different groups of compounds, including non-polar triglycerides, are of the same order and hence not only repeatability but also accuracy of both determinations were highly satisfactory.
It is essential to point out that the values used for the variables were carefully selected: a) Temperature established was that most representative of frying procedures.b) Length of heating was decided as to achieve levels of polar compounds close to those normally found in used frying fats.Limitations established in some European countries, around 16% of polymers and 25% of polar compounds (Firestone, 1996) were considered to set up the maximum alteration levels obtained.
c) Sample amount was selected as to fulfill two requirements: -A surface-to-oil volume ratio as low as possible in order to resemble that used in domestic discontinuous frying.Hence a value around 0.4 cm-i was established in the procedure proposed while it oscilates between 0.3 and 0.5cm-i in small domestic fryers (Jorge et ai., 1996b).
-Complete introduction of the sample in the heating block in order to guarantee homogeneous heating.
Table III reflects the outstanding influence of apparently small changes in sample amounts on the oil degradation level.Results included correspond to duplicate experiments.It is observed that, as the surface-to-oil volume ratio decreased from 5 to 9 g samples, alteration dropped as expected but then, for lOg-samples, levels of polymers increased significantly, considering the good repeatability data already commented.From this point fonA/ard, a further decrease in the surface-to-oil volume ratio exerted again the expected effect in lowering alteration.A likely explanation for these results could be the different and concomitant action of two variables: surface-to-oil volume ratio and occurrence of convection currents (Jorge et al., 1996a).In fact, when oil amount approaches around lOg, the effective heating zone of the aluminum block is overpassed and thus the convection currents due to the higher temperature gradient created would contribute to increase oxidation in a greater extent than the opposite effect resulting from a parallel decrease in surface-to-oil volume ratio.Interestingly, these findings do not bring about any drawback for the procedure but add further possibilities of applications since testing amounts over 10 g could be of great utility to check the action of compounds with potential surface protective properties.Samples of 50 mg were withdrawn after 2,4, 6, 8 and 10 h and polymerized compounds were quantitated directly in the oil samples (8 g) according to the lUPAC method.Repeatability was checked and data are also listed in Table IV.Results showed that, under the conditions used, the increase in polymers was linear from 0 to 10 hours.A highly significant correlation was obtained, the coefficients being 0.999 and 0.984 for HOSO and SO, respectively, and the slopes differed from 0.95 to 1.19, respectively, thus denotating the higher tendency to polymerization of the unsaturated oil.Overall, it is worth remarking that the main advantage of this heating procedure by using Rancimat apparatus is the elimination of the most important drawbacks found to set up a standard method to compare results from different laboratories.Given the characteristics of the apparatus, i. e., standard vessels, temperature correction, homogenity of temperature in tubes, excellent control of temperature, some important variables can be standardized, such as temperature, surface-to-oil volume ratio, heating time, type of heating, thus allowing the study of the influence of the different variables of the frying process.
Among the possible applications, this procedure can be useful to check the action of minor compounds.As an example.Table V shows the results obtained after heating 2 g of HOSO and SO with naturally occurring and devoid of tocopherols, essentially atocopherol, and other minor compounds, eliminated by aluminum oxide following the method of Yoshida etal., 1992.Highly significant differences were found between starting oils and their counterparts lacking tocopherols thus suggesting that the antioxidant action of a-tocopherol at high temperature cannot be underestimated.The procedure has been extensively used to know the action of natural antioxidants in oils of different unsaturation degree and tocopherol content, and results and conclusions will be shortly communicated.

*
Results are means of four experi ments ** Results are means of two experi ments ACKNOWLEDGEMENT This study was funded by CICYT (Project ALI-95-0736).The authors thank Mercedes Giménez for assistance.Daniel Barrera Arellano was supported by a postdoctoral fellowship (95/9304-2) from Fundaçao de Apoío à Pesquisa do Estado de Sao Paulo (FAPESP).

Table I Repeatability of the procedure for polymers (wt%) and total polar compounds (wt%) determination in conventional (SO) and high-oleic (HOSO) sunflower oils heated at 180°C for 10 h. Sample(g) Oil 2 HOSO SO g HOSO SO
* Results are means of four experiments.

Table IV Evolution of polymer (wt%) formation In conventional (SO) and high-oleic (HOSO) sunflower oils heated at 180° C.
* Results are means of four experiments.