The non-bleached sunflower oils used in this study were obtained from a local refinery. Sunflower oil samples were placed in dark bottles and stored at -18 °C until bleaching. Acid-activated (Suprefast M1FF) bleaching clay was also procured from same refinery (Gaziantep, Turkey). All chemicals and reagents used in this study were analytical grade and purchased from Sigma-Aldrich.

A three-level, five factorial central composite rotatable design (CCRD) was employed to analyze the effects of microwave assisted factors on the bleaching and to optimize the bleaching conditions to obtain maximum bleaching quality. The independent variables were selected as microwave power (A), amount of clay (B) and bleaching time (C). The levels of the independent variables were defined as between A, 70-120 W; B, 0.01-0.5% (w/w); C, 2-10 min. Preliminary experiments were carried out to obtain information about the independent variables and their levels prior to the experimental design. The employed CCRD was composed of 20 experiments consisting of 6 center points. At the end of optimization, verification of experiments was carried out under the optimal conditions and the average value for the experiments was compared to the predicted value.

Microwave-assisted bleaching (MWB) treatments were performed using a microwave reactor (CEM, Discover SP). Bleaching was carried out in a 35-ml glass vessel. The vessel was not fully filled to avoid oil splash. 20 g sunflower oil were weighed into the vessels in all experiments. The bleaching earth was weighed for each vessel according to the amounts determined in the experimental design. After the addition of clay, the tube was sealed with a rubber lid (CEM) and placed in the microwave reactor. In all experiments, stirring speed mode was set on high. The mixing of clay and sunflower oil was carried out only in the reactor and during the bleaching period. The levels of microwave power and time were adjusted according to the experimental design. This microwave system consisted of an infrared sensor for temperature measurement. Therefore, temperature values varied depending on the microwave power reported during the experiment. At the end of the microwave treatment, the mixture of sunflower oil and clay was centrifuged (EBA 20, Hettich) at 6.000 rpm at 25 °C for 10 min. Sunflower oil was separated easily from the clay because of the clay sticking to the wall of the tube. The sunflower oils bleached by this way were stored in dark amber sample bottles at -18 °C until further analysis.

Peroxide value (PV) was determined according to AOCS standard method Cd 8-53 (AOCS, 1998). The p-anisidine value (AV) in sunflower oil was determined by the spectrophotometric (Optima, SP-3000nano, Japan) method as described in the AOCS Official Method Cd 18-90 (AOCS, 1998). PV and AV were converted to total oxidation value (Totox) according to Equation 1.

The color of the samples was determined by a HunterLab Colorflex colorimeter with standard illumination D₆₅, and colorimetric normal observer angle of 10°. Color values were evaluated as L*, a* and b*. Measured values L*, a* and b* were also converted to chroma and hue angle values given below in Equations 2 and 3, respectively.

The experimental design and statistical analysis were conducted using Design Expert version 7.0 software (State-Ease Inc., Minneapolis, Minn., U.S.A.). Statistical parameters such as lack of fit, the coefficient of determination (R²) and the F-test value obtained from the analysis of variance (ANOVA) were used to evaluate the best fitting polynomial model. Regression analysis and three-dimensional surface plots were generated to understand the effect of microwave-assisted conditions on response variables. All experiments were performed in three replicates and the results were given as mean ± standard deviation.

The experimental design and the experimental results for the microwave bleaching of sunflower oil are presented in Table 1. The bleaching efficiency and the oil quality have been analyzed in terms hue angle, chroma and totox value for MWB sunflower oil. The totox value the MWB sunflower oil obtained in this study ranged from 19.27 to 29.91; while the hue angle and chroma ranged from 92.78 to 99.43 and 30.57 to 55.48, respectively.

The multiple regressions and backward elimination were employed to get the best fitting statistical model for each response. Table 2 summarizes the results from the ANOVA and regression analyses for the hue angle, chroma and totox values. Non-significant lack of fit and high R² value indicate that the statistical model fits well with data (Içyer and Durak, 2018; Islam et al., 2018). The quadratic model was determined as the best fitting model according to non-significant lack of fit and R² values for all responses (Table 2). Adequate precisions for totox, hue angle and chroma were found as 18.37, 36.56 and 22.02. These values were higher than 4. Therefore, it can be said that there is a good correlation between the estimated values and experimental data (Gasemloo et al., 2019). Similarly, Chew et al. (2017) and Garcia-Moreno et al. (2013) reported that the quadratic model was found suitable for the bleaching studies based on totox and color parameters. The reduced model created with the terms that are only statistically significant are expressed in Equations (4), (5), (6) for totox (Y1), hue angle (Y2) and chroma values (Y3), respectively:

The magnitude of coefficients for bleaching time and amount of clay are higher than the coefficients for current microwave power and other interactions. On the other hand, the magnitude of coefficients for microwave power and amount of clay are higher than the coefficients for current hue angle, chroma and other interactions, indicating that these parameters have a more significant effect on hue angle, chroma and totox values. The other parameters seem to have a quadratic effect on the hue angle, chroma and totox values.

The oxidation of fats and oils during bleaching is one of the important deteriorative reactions. Totox value is one of the important ways to express the total oxidation degree of fats and oils. Hence the totox value was taken as oxidation indicator in this study. Among the independent variables, microwave power, time and their interaction (AC) were found to significantly affect (p < 0.05) the Totox value. The second-order term of time (C2) also had significant (p < 0.05) effects on the totox value. On the other hand, neither the first-order term of amount of bleaching clay nor its interactions with other responses (AB, BC) had a significant (p > 0.05) effect on the totox value of bleached sunflower oil. The 3D response surface plots for the totox value of bleached sunflower oil samples as a function of independent variables are shown in Figure 1a. The totox value for sunflower oil increased with the rise in microwave power and longer irradiation time. The exposure of material to high microwave power increased the temperature of the material. Therefore, the increase in temperature due to high microwave power negatively affected the oxidation stability of the oil. Marrakchi et al. (2015) stated that the bleaching temperature significantly affects oxidation stability. In another study in which the optimization of industrial bleaching conditions was studied, it was emphasized that temperature and bleaching earth had significant effects on totox values (Chew et al., 2017). Contrary to the findings of Chew et al. (2017) the amount of clay in MWB was not found to be more effective on the totox value than the commercial bleaching process. It may be due to using a small amount of clay and less bleaching time in MWB conditions. İçyer and Durak (2018) emphasized that especially p-anisidine and totox value increased as bleaching time increased in both conventional and ultrasonic bleaching methods. The maximum totox value (29.91) obtained from MWB was found to be comparable to the totox value (31.07) reported by İçyer and Durak (2018) for 30 min conventional bleaching. In general, increasing the amount of clay can facilitate the removal of unwanted compounds in the oil during bleaching. On the other hand, the catalytic properties of acid-activated bleaching earth convert hydroperoxides into secondary oxidation products. Consequently, this situation affects the totox value (Zschau, 2000; Bonveh et al., 2001). It is quite clear from our findings that the low microwave power (80 W) and short processing time (8 min) has no effect on oil quality and only a small effect on peroxide and p-anisidine values of the oil processed by microwave. Even if the totox value increased slightly with increasing peroxide and p-anisidine values, they are similar to the findings obtained by the conventional bleaching process (Chew et al., 2017; Zschau, 2000; Bonveh et al., 2001). In addition, it should be kept in mind that the following deodorization process removes the formed oxidation products which are commonly produced in conventional technology, which has been emphasized in the literature in many sources (Shahidi et al., 1997; Zschau, 2001; Skevin et al., 2012).

Although the Lovibond tintometer is widely accepted as a standard method for measuring the color of oils, Hunterlab has been used in many studies to measure the color of oils (Abedi et al., 2015; Abedi et al., 2016; Chew et al., 2017). In addition, Abedi et al. (2015) reported a correlation between b* and a* values and carotenoid and chlorophyll concentrations, respectively. Hue angle and chroma were monitored for the prediction of color change during the MWB of sunflower oil. While hue angle represents color appearance (0°: red, 90°: yellow and 180°: blue), chroma is defined as a measurement of whiteness and vividness of color (Mínguez-Mosquera et al., 1991). All independent variables (A, B and C) were found to be significantly effective on the hue angle and chroma value. In addition, some interactions such as microwave power/amount of clay (AB), amount of clay/time (BC) and second-order term of time (C²) were found to be significantly effective (p < 0.05) on the hue angle. García-Moreno et al. (2013) found a similar correlation between hue angle and chroma. Response surface plots for chroma and hue angle values for bleached sunflower oil samples as a function of independent variables are shown in Figures 1b, 2a and 2b, respectively. Based on Figure 2a, microwave power and amount of clay had significantly positive effects on hue angle. Moreover, interactions between microwave power and amount of clay had a synergistic effect on hue angle. The interaction of microwave power and amount of clay caused an increase in hue angle. This means that the redness was removed from the sunflower oil. Bleaching time was found as another important parameter for hue angle. In particular, interaction of time and amount of clay caused a positive effect on the hue angle (Figure 2b). Furthermore, the second order effects of these two variables on the hue angle had a negative coefficient which means that these variables had reached an optimum value in the studied range. In other words, the increase in time and amount of clay did not cause further rising in hue angle after a certain point. Interactions of BC and second-order term of amount of clay (B²) were found significant on the chroma. Chroma value decreased with increasing amount of clay and time and vice versa (Figure 1b). A negative correlation was determined between the hue angle and chroma value. Although the hue angle increased, the chroma values decreased. Marrakchi et al. (2015) reported a similar trend in their study.

The optimization of MWB was carried out by minimizing totox and chroma, maximizing hue angle in sunflower oil samples. Under optimal conditions, the temperature started at ambient temperature and reached 100 °C at the end of bleaching. In other words, the temperature increased during the treatment period but remained below 100 °C without applying vacuum. Optimal conditions for MWB were found as 80 W, 8.0 min and 0.4% for microwave power, time and amount of clay, respectively. Compared to studies in the literature on reducing clay and bleaching time, it was shown that optimized MWB conditions provided more effective decreases. For example, Abedi et al. (2015) reported that two optimized bleached oils were produced by using 1.22 and 1.35% clay amount in 28 minutes. In a different study on the bleaching of sunflower and soybean oil by using a high voltage electric field, the same researchers (2016) found optimal conditions at 1% clay amount and 20 minutes; 0.5% clay amount and 20 minutes for soybean and sunflower oil, respectively. In another study Asgari et al. (2017) studied the bleaching of olive oil and they obtained 13 minutes and 1.21% for bleaching time and amount of clay. The bleaching time could not be decreased below 10 minutes in any of these studies. It ranged from 15 min to 30 min. In these studies, the amount of clay used was at its minimum at 0.8% as determined by İçyer and Durak (2018). Temperature is one of the important parameters for effective bleaching. The temperature values used in the ultrasonic bleaching process can be a limiting factor because with the rapid increase in the liquid temperature, a dense bubble formation begins in the liquid and continues increasingly. But the tendency of the bubbles to collapse becomes harder due to the high pressure inside of them acting as a bed (Thompson and Doraiswamy, 1999). Therefore, in the ultrasound-assisted bleaching process, the bleaching effect at high temperatures needed for obtaining the desired oil quality cannot be due to sonication (Abedi et al., 2015). In addition, in cases where high temperatures are inevitable, unwanted taste and odors have been observed in the ultrasonic method (Jahouach-Rabai et al., 2008). But due to having lower specific heat, vegetable oils can be heated to high bleaching temperatures easily in a short time by microwave, which is an important advantage compared to the ultrasound method. Abedi et al. (2016) used 1.5% bleaching earth and 30 minutes bleaching time to imitate the industrial process as a control system. The microwave bleaching results showed that MWB was more advantageous in terms of clay usage (0.4%) and bleaching time (8 min) than those used in industrial processes.

For verification of the response values estimated as a result of the optimization, three independent MWB bleaching studies were carried out under optimal conditions and experimental response values were determined. Experimental and estimated values for each response under optimal conditions are given in Table 3. The adequate precision values obtained from Design Expert for totox, hue angle and chroma were found to be 23.31, 96.91 and 37.66, respectively. One-way t-test showed that there was no statistically significant difference between the experimental data and the predicted values for all three responses (p > 0.05).