Bleaching with the Mixed Adsorbents of Activated Earth and Activated Alumina to Reduce Color and Oxidation Products of Anchovy Oil

WANG Xiaohan, WEN Yunqi, BI Shijie, LI Zhaojie, XUE Yong,XUE Changhu, 3), and JIANG Xiaoming, 3), *

Bleaching with the Mixed Adsorbents of Activated Earth and Activated Alumina to Reduce Color and Oxidation Products of Anchovy Oil

WANG Xiaohan1), WEN Yunqi1), BI Shijie2), LI Zhaojie1), XUE Yong1),XUE Changhu1), 3), and JIANG Xiaoming1), 3), *

1),,266003,2),,071000,3),266109, China

Fish oil is a rich source of polyunsaturated fatty acids, and its refinement has drawn attention for years. An appropriate adsorbent can effectively remove the pigment impurities in the fish oil. This study evaluated the impact of different absorbents on the reduction of oxidation products and color of anchovy oil during the decolorization under high vacuum. Using the single factor design, four process parameters including adsorbents type, adsorbent amount, temperature and time were tested to determine the optimum decolorization parameter. The results showed the optimum decolorization conditions were that the fish oil was treated with 8% activated alumina at 80℃ for 40min. In the central group experiment, the addition amounts of mixed absorbents (activated earth and activated alumina), including the mass ratio of adsorbent in oil (5%–11%, w/w) and the mass ratio of activated earth in total absorbent (20%–80%, w/w) were optimized to remove the oxidation products. Under the optimum condition at 10.18% of adsorbent and 70% of activated earth, the total oxidation(TOTOX value) showed the minimum with the 44.4% of removal rate. Eight metal elements were analyzed in decolorized oil using inductively coupled plasma mass spectrometry (ICP-MS). The removal rates of Zn and Pb were 94.12% and 55.35%, respectively. The decolorization process using mixed absorbents under appropriate condition can significantly reduce the oxidation products and pigments in fish oil, which will benefit the industrial production of fish oil.

decolorization; anchovy oil; mixed adsorbent; TOTOX value; metal ions; activated alumina

1 Introduction

Fish oil contains high levels of polyunsaturated fatty acids (PUFA), as well as smaller molecular components, such as amino acid, minerals, carotenoids and vitamins (Za- kwan, 2017). However, the crude fish oil usually has the following three problems: 1) It is brownish yellow, which significantly reduces consumers’ acceptability of the oil, es-pecially as an edible products (Suseno, 2012); 2) When fish oil is stored at room temperature, omega-3 PUFA is susceptible to oxidative deterioration, resulting in the loss of fatty acids and the reduction of the nutritional value(Frieler, 2019); 3) In recent years, heavy metal- loaded industrial waste affects the general health of fish, fish products, and fish (product) consumers (Javed and Us- mani, 2017).

Decolorization of fish oil is a crucial step in fish oil re- finement, which can effectively improve the quality of fishoil by removing colored matter and natural pigments, yield- ing a light-colored oil (Monte, 2015). It can also re- duce the amounts of soap, trace metals, phospholipids, oxi- dation products and other harmful components (García- Moreno, 2013). Various adsorbents have been exten- sively used in edible oil industries for decolorization, acid neutralization, physical refining, wax separation, peroxi- des removal, flavor/aroma corrections, frying oil regenera- tion and similar purposes (Guner, 2019). One of the most common industrial adsorbents applied to the decolo- rization processes of edible oils is the activated earth. Ac- tivated earth is produced from mineral bentonite.It can ef- fectively adsorb pigments in fish oil with low cost (Icyer and Durak, 2018). In the decolorization process, activated earth is prevalently used to remove impurities such as oxi- dative products, phosphorus, trace metals and color pig- ments from crude oil.

Activated alumina has recently been developed as a new adsorbent. Because of its high surface area and mechanical strength, it has a good adsorbability of various metal ions, inorganic anions and organic ligands. The modification of the activated alumina with alkali ions increases its surface basicity and dynamic capacity (Sen and Sarzali, 2008). Me- tal-organic frameworks made by activated alumina can re- duce the yellowness of fish oil and enhance its sensory quality(Guner, 2019).

A great number of studies have emphasized the capa- city of traditional absorbents for pigment adsorption and oxidation products removal. The results show that some traditional adsorbents like activated carbon and diatoma- ceous earth can effectively extend the lifespan of used fry- ing oil. However, fewer studies have been conducted on the effect of decolorization processes on fish oil (Chutapa, 2013). When using traditional adsorbents such as activated carbon and activated earth, the content of caro- tenoids in fish oil can be greatly reduced. However, the change of oxidation products content in fish oil during the decolorization process by new adsorbents is not clear (Mon- te, 2015). The combination of four adsorbents, in- cluding single, binary and ternary mixtures, have been app- lied in the purification of used frying sunflower oil, how- ever, the methods have not been developed with scientific multi-factor data optimization methodology (Turan, 2019).

Response surface methodology (RSM) is an effective analysis method to predict optimum experimental factors and investigate relationships between independent variables and response results. The central composite design (CCD) is the most commonly used second-order design in re- sponse surface and its advantage is that it can better fit the response surface as compared to other experimental de- signs.

The purpose of this study was to evaluate the changes in color and oxidation products content during the decolo- rization process of fish oil. The optimization of experimen- tal parameters, including adsorbent type, adsorbent amount, temperature and contact time, were conducted through the single factor experiments. The optimization of significant- ly influencing mixed adsorbents in the decolorization pro- cess was conducted using RSM with facing CCD. Fur-thermore, comparisons of the absorption effect between single and mixed adsorbents in the decolorization process with respect to anchovy oil were especially presented. The TOTOX value (TV) and Lovibond color value of the oils were used to evaluate the removal efficiency of oxidation products and colored impurities. For simultaneously re- flecting the removal efficiency of two kinds of impurities, the results were provided in the form of synthesis scores that were set as 60% of the TV and 40% of the color va- lue based on the actual industrial demand. The optimized parameters, especially the amount of mixed adsorbents in the decolorization process, will provide useful informationfor the industrial refining processes of anchovy oil and other oils.

2 Materials and Methods

2.1 Materials and Reagents

The nitric acid (65%, v/v, GR grade) and the perchloric acid were obtained from Merck Chemicals Co., Ltd. (Shang- hai, China). Deionized water was produced by a Millipore Milli-Q water purification system (Burlington, MA, USA). Na2S2O3solution (0.10molL?1) and-anisidine were pur- chased from Tianjin Guangfu Research Institute (Tianjin, China). All other reagents of analytical grade were ob- tained from Sinopharm Chemical Reagent Co., Ltd. (Qing- dao, China). Crude anchovy oil was purchased from Zhou- shan Xinnuojia Bioengineering Co., Ltd. (Zhoushan, Chi- na).

2.2 Neutralization Process

To avoid non-predictable variables and to obtain the maximum adsorption potential of the tested mixed adsor- bents, the decolorization assays were performed using neu- tralized fish oil. Initially, 10kg of anchovy oil (stored at ?20℃) was thawed at room temperature for 3h and cen- trifuged at 5000rmin?1for 30min. Subsequently, the an- chovy oil was dealt with 3.65% sodium hydroxide solu- tion and 0.45% of excess alkali (wt%). The reaction was kept at temperatures between 60℃ and 65℃ for 40min. Then hot water about 100℃ was used to wash the oil un- til the eluent reached a pH of 7, subsequently obtain neu- tralized fish oil by the centrifugation at 5000rmin?1for 25min. The fish oil was isolated from air by filling nitro- gen during the whole experiment. The refinement of crude oil,including neutralization, washing and drying,follow- ed the previously described method (Crexi, 2009).

2.3 Decolorization Process

The anchovy oil was subjected to decolorization under partial vacuum (150Pa vacuum pressure). Approximately 40g of neutralized fish oil was introduced into a single- neck round-bottom flask loaded with a given amount of adsorbent.The mixture was stirred at 200rmin?1and kept at the desired temperature for a certain time. Four kinds of factors including the adsorbents type, adsorbent amount, temperature, and time were considered in this experiment. The adsorbents included activated earth, activated carbon, activated alumina, diatomite, chitosan, attapulgite and se- piolite. The other three factors were tested at different le- vels: adsorbent amounts were2%, 5%, 8%, 10%, and 15%of the whole weight; temperatureswere60, 70, 80, 90, and 100℃; and the experiment time was15, 20, 25, 30, 40, 50min. After the decolorization process, the fish oil was cooled to room temperature and the adsorbent was removed through two times of centrifugation.The first cen- trifugation was at 5000rmin?1for 15min and the second one was at 10000rmin?1for 10min. Finally, the anchovy oil samples were stored under nitrogen protection and at ?20℃.

2.4 Analytical Method

2.4.1 Peroxide value () ,-anisidine value () and

Theof the oil samples was measured according to the AOCS method Cd 8b-90 (Firestone, 1994) with some modification. About 1.0g of anchovy oil was diluted in 30mL of acetic acid-isooctane (3:2, v/v) mixed solvent. Then 1mL of saturated potassium iodide solution was add- ed, following by the blending and storage in the dark for 10min. Then 100mL distilled water was added before ti- trationwith 0.01molL?1Na2S2O3solution. When the so- lution turned into yellowish, 3mL of 0.2% starch solutionwas added to visualize the end-point that the solution turn- ed colorless (Wen, 2019).

was determined by recording the absorbance values using a Shimadzu UV-1700 spectrophotometer (Kyoto,Japan) as described in the AOCS method Cd 8b-90 (Fire- stone, 1994). This method utilized the chromogenic reac- tion of-anisidine with α- and β-unsaturated aldehydes (pri-mary 2-alkenes) in oil under acetic acid conditions. The ab- sorbance of this solution was determined at 350nm.

Theof fish oil can reflect the content of oxidation products including primary and secondary oxidation pro- ducts of fish oil. It is a synthetically oxidation index, cal- culated byandvalues throughthe following Eq. (1).

The removal rate can be calculated as the following Eq. (2) (Chew, 2017):

where0isthe TOTOX value of anchovy oil before de- colorization process;is the TOTOX value of anchovy oil after decolorization process.

2.4.2 Lovibond color

Lovibond tintometer (PLV 300 fully automatic tintome- ter, Perkone Scientific, Hangzhou, China) was used to de- termine the color of fish oil. The color was measured by matching with a set of standard colored, numbering glass- es, ranging in the scales from 0 to 70 for red (R), 0 to 70 for yellow (Y), 0.1 to 40 for blue (B), and 0.1 to 3.9 for neutral (N). In order to compare the color difference, the yellow values of different samples were selected as the in-dex,considering other color values of different samples showed little difference.

2.4.3 Metal ion detection

Inductively Coupled Plasma Mass Spectrometry (ICP- MS) offers a powerful and efficient mean of detecting trace metal elements at the level of parts-per-billion (ppb). While microwave digestion technology can quickly ionize the samples (Wang, 2019), which is also convenient for follow-up operations. Metal ion contents (includingCr, Mn, Fe, Cu, Zn, Cd, As, and Pb) were calculated with an ex- ternal calibration method by ICP-MS. Blank tests for the procedure were also performed (Wang, 2019). The operating conditions for ICP-MS in this study were sum- marized in Table 1.

Table 1 Equipment and operating conditions

2.4.4 Fatty acid composition

The fatty acid composition analysis was carried out on the neutralized fish oil and the decolorized fish oils. About20mg of fish oil was dissolved in 2mL-hexane and sha- ken well. Subsequently, 200μL of this mixturewas taken and dried with nitrogen, and then 2mL of methyl esteri- fication reagent (Vhydrochloricacid:Vmethanol=1:5) was added. This mixed solution was heated in a water bath at 90℃ after nitrogen filling and sealing. After keeping the reac- tion for 2h, 1.5mL of-hexane was added for extraction. After the mixture was shaken, rested and delamination, the upper liquid (about 1mL) was passed through a 0.22μm organic filter membrane and stored in liquid phase vial for GC detection (Zhang, 2019).

This analysis was performed by GC (Agilent, 7820) equipped with capillary column SUPELCOWAXR 10 (30m×0.32mm, 0.25μm) and flame ionization detector. The ana- lysis of FAMEs was performed by injecting 1.0μL with no shunt ratio. The column oven conditions were as fol- lows: Injection temperature was 220℃; the oven origin temperature was held at 150℃ originally, then increased to 220℃ at 3℃min?1and maintained at this temperature for 20min. The standard reference was used or qualitative analysis and the area normalization method was used for quantitative analysis (Kowalski, 2019).

2.5 Statistical Analysis

The CCD design was used to analyze the important fac-tors in the decolorization process. The rotational type CCDcontaining two factors was carried out with 12 experiments.The factor levels were composed of three points (±1 and 0), two axial points (±1.21), and four central points (0, 0, 0, 0). The experimental design matrix of rotational type CCD was shown in Table 2.

The independent variables studied were the percentage of adsorbent related to the fish oil mass1(5.00%, 5.52%, 8.00%, 10.48%, and 11.00%), and the percentage of activated earth related to the total adsorbent mass2(20.00%, 25.21%, 50.00%, 74.79%, and 80.00%). The level values of these independent variables were referred to preliminary tests and literature (Icyer and Durak, 2018). The re- sponses of these experiments were analyzed by using a re- gression analysis and the optimum conditions for the de- colorization process was determined through the response surface of the CCD.

The statistical model was obtained from the second- order regression analysis (in actual form) using the least squares method, considering their effects and their inter- actions on the responses ofand Lovibond color, as re- presented in Eq. (3).

Table 2 Experimental matrix of the CCD (rotational type) and results for TOTOX value and color

where a0is the intercept; aand aarethe regression coef- ficients of each factor, of each quadratic term and of the interaction term between them, respectively;1and2are the percentage of adsorbent in relation to oil mass and per- centage of activated earth to the total adsorbent mass, re- spectively; andYis the actual responses of the synthesis score ofand Lovibond color. Meanwhile, the Lovibondvalue accounted for 40% and theaccounted for 60% of the synthesis score.

These experiments were carried out in triplicate and the data were expressed as mean±standard deviation (=3). The statistical analysis of the experimental data was performed by SPSS 20.0 software (SPSS Inc, Chicago, IL, USA). The differences in the averages were determined using a one-way analysis of variance (ANOVA). The difference was significant at the level of 95% (<0.05).

3 Results and Discussion

3.1 Analysis of the Crude and Decolorized Oil

Referring to previous studies (Suseno, 2012), the factors affecting the decolorization of fish oil included ad- sorbent type, adsorbent content, temperature and reaction time. As shown in Fig.1A, different types of adsorbents re-duced the oxidation products of decolorized fish oil. Amongthem, activated alumina had the best removal effect, as thereduced to 32.44, and the removal rate reached to69.38%, followed by activated earth (removal rate,49.16%), attapulgite (removal rate, 37.25%), activated car- bon (removal rate, 29.56%), sepiolite (removal rate, 14.52%) and diatomite (removal rate, 6.13%). The lowest removal effect was fromchitosan (removal rate, 2.88%). Activated aluminahas large specific surface area and more activity center, in which the protonation and deprotonation of these surface hydroxyl groups cause the oxide surface to deve- lop an electrical charge, leading the promotion of adsorp- tion (Osman, 2018). As a result, activated alumina had obvious effect on the removal of oxidation products and colored compounds. The adsorption ability of chito- san for fish oil was not significant as thewas almost unchanged. The main functional group that is responsible for adsorbing impurities in chitosan is the N atom, and chi- tosan is typically used to adsorb pollutants from biodiesel wastewater (Pitakpoolsil and Hunsom, 2013). Finally, ac- tivated alumina was selected to determine the optimum concentration of adsorbent for the decolorization of fish oil.

Thecorresponding to various activated alumina con- centrations were presented in Fig.1B. It can be found that the oxidation products content in fish oil decreased with the increase of the adsorbent mass. As reported, the ad- sorption capacity was expected to increase as the specific area increases(Plata, 2020). When the adsorbent con-tent was 20%, the lowestwas 8.15. However, theincreased when the adsorbent amount reached to 25%.The increasing amount of the adsorbent doesn’talways benefit the dispersion of the adsorbent in fish oil,because of the limited activated area of adsorbent contacting with oxida- tion products. In a system with higher solid content, these interactions are perhaps physically blocking some adsorp- tion sites on the adsorbing solutes and thus causing de- creased adsorption (Wang, 2007). However, in the actual industrial production process, the increase of ad- sorbent content will lead the increase of cost. Moreover, it is not conducive to subsequent filtration. Therefore, con- sidering the additive amount of industrial production, 8% of activated alumina was finally selected for the subse- quent optimum experiments.

As shown in Fig.1C, the effects of different tempera- tures on the removal of oxidation products and color by activated alumina were noticeably different. When the tem- perature was 80℃, thehad the minimum of 19.79. Therefore, the optimum temperature for activated alumina was approximately 80℃. If the temperature was too high, thevalues showed an increase, which might be caused by the new oxidation productions of fish oil and the limit- ed adsorption capacity of activated alumina. On the other hand, lower temperature suppressed the activity of acti- vated alumina and influenced the adsorbent ability (Su- seno, 2012). Therefore, the appropriate temperature at 80℃ was selected as the optimum temperature for fur- ther experiments on the decolorization process of fish oil.

Fig.1 Effects of different factors on physicochemical indexes.

Fig.1D showed the effect of contact time on the oxida- tion products content of fish oil with 8% of activated alu- mina and 80℃. The results presented the lowest(15.14) for 40min of conact time. It also can be found that the amount of oxidation products decreased with increasing contact time at initial reaction stage and finally reached equilibrium after 40min. The effect on thewith 20 and 25min showed little difference. The prolongation of reac- tion time was beneficial for the full contact of the adsor- bent (aluminum oxide) and fish oil. However, when the re-action time was too longin high temperature, more secon- dary oxidation of fish oil occurred, resulting in more oxi- dation products which led to an increased(Asgari, 2017). According to the result, 40min was selected as the optimum reaction time in the decolorization process of fish oil.

3.2 The Optimization of the Adsorbent Amount

Table 2 showed the results of the CCD matrix. In rela- tion to Lovibond color, all values of the experimental de- sign matrix were lower than the commercial value (30 yel-low) (Monte, 2015). Therefore, the decolorization pro- cess was effective for all the experimental results.

The statistical analysis of the synthesis score revealed the main effect of the percentage of adsorbent on the re- moval of oxidation products and color in the fish oil, in- cluding the effect of the percentage of activated earth in the adsorbent mixture and the percentage of adsorbent in fish oil, which were all significant at the level of 95% (<0.05) (Table 3). Though the quadratic effects of the percen-tage of activated earth in the total adsorbent mass were not significant at the 95% level, these factors were remain-ed in the analysis because the simple effect was significant.As activated earth with specific surface areas and pore vo-lumes can improve the adsorption capacity of metal impu- rities, phosphatide and color bodies, the percentage of ac- tivated earth in relation to the total adsorbent affect the de-colorization process significantly (Hussin, 2011). Thus,the treatment with the high concentration of activated earth had the low synthesis score.

Table 3 Analysis of the effects of the adsorbent content and the activated earth in adsorbent content on the responses of the experimental design matrix

The fitted model was verified by analysis of variance, and SPSS was used to evaluate the significance of the re- gression model. The synthesis scores of the statistical mo- dels were predictive. The conclusions from the predicted model were that the total optimal adsorbent amount in fish oil was 10.18%, the proportion of activated earth in rela- tion to the total adsorbent was 70%, and the comprehen- sive synthesis score of prediction was 20.99. Compared to the actual value of 22.37, the relative error of the predic- tive data to the experimental data was only 6.16%. It was considered that the response surface model was effective. These values stated a satisfactory adjustment of the qua- dratic models to the experimental data, indicating that 95% of credibility interval in the responses could be explained by the statistical model for synthesis score as Eq. (4). The positive coefficients of12,22and12indicated that the synthesis score increased by the increment of these fac- tors. On the other hand, a negative value indicated an in- verse relationship between the factor and the synthesis score as the response (Daraei, 2019).

, (4)

whereis the synthesis score;1is the percentage of adsorbent in relation to fish oil, and2is the percentage of activated earth in relation to total adsorbent respective- ly (in actual form), respectively.

Fig.2 indicated the response surface for the synthesis score obtained by Eq. (4), and showed that the use of lar- ger amounts of activated earth and higher concentrations of adsorbent resulted in a lower synthesis score. However, the synthesis score did not show an evident decline in a certain range of activated earth or adsorbent addition. Thus, the best working condition for the decolorization process of fish oil was using 10.18% of adsorbent in relation to the fish oil and 70% of activated earth in relation to the total adsorbent mass. The results indicated a 75.89% in theand 88.57% of yellow reduction in Lovibond color value. Additionally, the yellow in Lovibond color was the low- est in this condition.

Fig.2 Response surface for the synthesis score (TV:Color value=6:4) of the fish oil after the decolorization.

3.3 Metal Ion Detection

The neutralized and decolorized fish oils obtained by the optimal bleaching condition were digested in an acidic so- lution and sent to ICP-MS for metal ion detection. The results were shown in Table 4. The Relative Standard De- viation (RSD) of these eight metal elements in the sample was all less than 10%, indicating that the method had good accuracy and precision. Three metal elements showed sig- nificant downward trend. The Cr element was reduced from 92.67μgkg?1to 45.98μgkg?1, with a removal rate of 50.38%. The removal rate of Pb was 55.35% as it was reduced from 60.50μgkg?1to 27.01μgkg?1. Cd was not detected after the decolorization process. The content of heavy metals afterthe decolorization process was far lower than the standard for heavy metals which was regulated by the Maximum Levels of Contaminants in Foods (General Administrationof Quality Supervision, 2017). The heavy metals in the neu- tralized fish oil could be adsorbed by the mixed adsorbents and removed to a certain extent after multiple centrifugal du-ring the decolorization process. With respect to the four transition metal elements of Mn, Fe, Cu and Zn, the con- centration of Zn was significantly reduced from 3975.80μgkg?1to 233.69μgkg?1, with a removal rate of 94.12%, whereas the removal effect of other three transition ele- ments was not noticeable. Because transition metal ions are an important factor affecting the oxidation of fish oil (Gao and Guo, 2017), the decrease in Zn content could ef- fectively decrease the PUFA loss caused by fatty acid oxi-dation of fish oil. However, the content of Cu increased by 47.90%, possibly due to impurities in the adsorbents. In conclusion, the decolorization process had a certain effect on the removal of metal elements in fish oil, especially Cr, Pb, Zn, and Mn. However, some metal ions in fish oil were not removedeffectively. Other refining processes must be used to further reduce the contents of these metals.

Table 4 The composition and content of metal in neutralized and decolorized fish oil obtained by the optimal bleaching condition

Notes:*Values were significantly different (<0.05) when decolo- rized fish oil is compared to neutralized fish oil. ‘–’, not mentioned.

3.4 Fatty Acid Composition Analysis

To obtain a more detailed assessment of fatty acid com- position differences between the decolorized fish oil (in the best conditions) and neutralized fish oil, the percen- tage of free fatty acids was analyzed. Table 5 showed the fatty acids profiles of the neutralized fish oil and the de- colorized oil under the optimal conditions. As shown in Table 5, the fatty acids of anchovy oil in this study were mainly composed of C14–C22 fatty acids, including four kinds of saturated fatty acids (SFA), five kinds of mono- unsaturated fatty acids (MUFA) and eight kinds of poly- unsaturated fatty acid (PUFA). SFA accounted for 37.91%, MUFA for 37.96% and PUFA accounted for 24.13%. The content of unsaturated fatty acids was significantly higher than that of SFA. In SFA, the content of C16:0 was 24.84%, which was significantly higher than C14:0 (6.67%) and C18:0 (4.47%). Among MUFA, C18:1 had the highest con- tent of 28.61%, while the content of C16:1 was 8.15%. In PUFA, docosahexaenoic acid (DHA) showed the highest content of 12.59%, which was generally higher than that in squids, mackerels, tunas, and other fishes(Subrama-niam, 2003), followed by eicosapentenoic acid (EPA)(6.31%), C18:2 (1.50%), C18:3 (1.33%), C22:2 (1.13%), and C20:4 (1.03%).

Table 5 Fatty acid profiles of neutralized and decolorized fish oil obtained in the optimal bleaching condition

Note:*Values were significantly different (<0.05) when decolo- rized fish oil is compared to neutralized fish oil.

When compared with the neutralized fish oil with re- spect to C20:5 and C22:6, the content change of these two fatty acids in decolorized fish oil was not significant (>0.05). The EPA content increased from 5.41% to 6.31% and the DHA content decreased from 13.86% to 12.59%. SFA accounted for 37.91% of the neutralized fish oil, which increased by 5.62% in the decolorized oil; MUFA decreased by 8.42% and PUFA increased by 5.64% in de- colorized oil as compared to the neutralized fish oil. This result showed that the decolorization process had little ef- fect on the fatty acid composition of fish oil. The little loss of PUFA produce few oxidative products, which is bene- fit for the increase of the oxidation stability (Zhong., 2007). So this process in this study could meet the re- quirements of industrial production.

4 Conclusions

A new mixed adsorbent composed of activated earth and activated alumina was used to remove the oxidation pro- ducts and color from fish oil. Single factor experiments were performed to firstly investigate the effectiveness of adsorbent type, adsorbent amount, treatment temperature and time on the removal efficiency. From the results, ac- tivated alumina with 8% additive amount at 80℃ for 40min showed the most efficient removal of oxidation pro- ducts and color. The CCD experiment was applied to opti- mize the added amount of mixed adsorbent. The condi- tions for the decolorization process of the fish oil refine- ment (10.18% of adsorbent in relation to the oil mass and 70% of activated earth in relation to the total adsorbent mass) and their interactions were identified. The model was significant and reliable, and could effectively predict the changes ofunder decolorization conditions. Un- der the optimum conditions, the comprehensive synthesis score was 22.37. Meanwhile, Zn and Pb elements in the fish oil were reduced by 94.12% and 55.35%, respective- ly. The PUFA content in decolorized fish oil increase by 5.64%. This study determined various parameters for the removal of oxidation products from fish oil during deco- lorization refining, and provided a reference condition for the industrial production of fish oil.

Acknowledgement

This work was supported financially by the Ocean Uni-versity of China, under the Classification of Project Num- ber of 2018YFC0311201.

Asgari, S., Sahari, M. A., and Barzegar, M., 2017. Practical mo- deling and optimization of ultrasound-assisted bleaching of olive oil using hybrid artificial neural network-genetic algo- rithm technique., 140: 422-432, https://doi.org/10.1016/j.compag.2017.06.025.

Chew, S., Tan, C., and Nyam, K., 2017. Application of response surface methodology for optimizing the deodorization para- meters in chemical refining of kenaf seed oil., 184: 144-151, https://doi.org/10.1016/ j.seppur.2017.04.044.

Chutapa, S., Bhundit, I., and Chanin, T., 2013. Cytotoxicity of used frying oil recovered by different adsorbents.–, 47 (6): 874-884.

Crexi, V. T., Souza-Soares, L. A., and Pinto, L. A. A., 2009. Crap () oils obtained by fishmeal and ensilage pro- cesses: Characteristics and lipid profiles., 44 (2): 1642-1648, https://doi.org/10.1111/j.1365-26 21.2009.01982.x.

Daraei, P., Zereshki, S., and Shokri, A., 2019. Application of non- toxic green emulsion liquid membrane prepared by sunflower oil for water decolorization: Process optimization by response surface methodology., 77: 215-222, https://doi.org/10.1016/j.jiec.2019.04. 039.

Firestone, D., 1994.’. American Oil Chemists’ Society.

Frieler, L., Ho, T. M., Anthony, A., Hidefumi, Y., Yago, A. J. E., and Bhandari, B. R., 2019. Crystallisation properties of amor- phous cyclodextrin powders and their complexation with fish oil., 56 (3): 1519-1529, https:// doi.org/10.1007/s13197-019-03643-7.

Gao, X., and Guo, Z., 2017. Biomimetic superhydrophobic sur- faces with transition metals and their oxides: A review., 14 (3): 401-439, https://doi.org/10. 1016/s1672-6529(16)60408-0.

García-Moreno, P. J., Guadix, A., Gómez-Robledo, L., Melgosa, M., and Guadix, E. M., 2013. Optimization of bleaching con- ditions for sardine oil., 116 (2): 606-612, https://doi.org/10.1016/j.jfoodeng.2012.12.040.

General Administration of Quality Supervision, Inspection and Quarantine of the People’s Republic of China, 2017. Maxi- mum Levels of Contaminants in Foods. GB 2762-2017. Stan- dards Press of China, Beijing (in Chinese).

Guner, M., Yilmaz, E., and Yuceer, Y., 2019. Off-odor removal from fish oil by adsorbent treatment with selected metal-or- ganic frameworks., 34 (3): 163-174, https://doi.org/10.1002/ffj.3489.

Hussin, F., Aroua, M. K., and Daud, W. M. A. W., 2011. Tex- tural characteristics, surface chemistry and activation of bleach- ing earth: A review., 170 (1): 90-106, https://doi.org/10.1016/j.cej.2011.03.065.

Icyer, N. C., and Durak, M. Z., 2018. Ultrasound-assisted bleach- ing of canola oil: Improve the bleaching process by central composite design., 97 (3): 640- 647, https://doi.org/10.1016/j.lwt.2018.07.030.

Javed, M., and Usmani, N., 2017. An overview of the adverse effects of heavy metal contamination on fish health., 89 (2): 389-403, https://doi.org/10.1007/ s40011-017-0875-7.

Kowalski, R., Kowalska, G., Pankiewicz, U., Mazurek, A., W?o- darczyk-Stasiak, M., Sujka, M.,., 2019. The effect of an addition of marjoram oil on stabilization fatty acids profile of rapeseed oil., 109: 225- 232, https://doi.org/10.1016/j.lwt.2019.04.016.

Monte, M. L., Monte, M. L., Pohndorf, R. S., Crexi, V. T., and Pinto, L. A. A., 2015. Bleaching with blends of bleaching earth and activated carbon reduces color and oxidation products of carp oil., 117 (6): 829-836, https://doi.org/10.1002/ejlt.201400223.

Osman, D. I., Attia, S. K., and Taman, A. R., 2018. Recycling of used engine oil by different solvent., 27 (2): 221-225, https://doi.org/10.1016/j.ejpe.2017.05. 010.

Pitakpoolsil, W., and Hunsom, M., 2013. Adsorption of pollu- tants from biodiesel wastewater using chitosan flakes., 44 (6): 963- 971, https://doi.org/10.1016/j.jtice.2013.02.009.

Plata, V., Rojas, ó., and Gauthier-Maradei, P., 2020. Improve- ment of palm oil biodiesel filterability by treatment with re- activated spent bleaching earths., 260: 116-125, https:// doi.org/10.1016/j.fuel.2019.116198.

Sen, T. K., and Sarzali, M. V., 2008. Removal of cadmium me- tal ion (Cd2+) from its aqueous solution by aluminium oxide (Al2O3): A kinetic and equilibrium study., 142 (3): 256-262, https://doi.org/10.1016/j.cej. 2007.12.001.

Subramaniam, S., Witoon, P., Joan, M. K., Casey, C., and Lloyd, S., 2003. Oil production from catfish viscera., 80 (4): 377-382, https://doi. org/10.1007/s11746-003-0707-z.

Suseno, S. H., Tajul, A. Y., Nadiah, W. A., and Noor, A. F., 2012. Improved of color properties onoil during adsorbent refining using magnesol xl., 19 (4): 1383-1386.

Turan, S., Yal?uk, A., ?i?lio?lu, K., and Ramadan, M. F., 2019. Purification of used sunflower frying oil with adsorbent mix- tures using an active filtration method., 96 (3): 171-181.

Wang, W., Chen, H., and Wang, A., 2007. Adsorption characte- ristics of Cd(II) from aqueous solution onto activated paly- gorskite., 55 (2): 157- 164, https://doi.org/10.1016/j.seppur.2006.11.015.

Wang, Z., Xu, J., Liu, Y., Li, Z., Xue, Y., Wang, Y.,., 2019. Arsenic speciation of edible shrimp by high-performance li- quid chromatography-inductively coupled plasmamass spec- trometry (HPLC-ICP-MS): Method development and health assessment., 52 (14): 2266-2282, https:// doi.org/10.1080/00032719.2019.1608224.

Wen, Y. Q., Xue, C. H., Xu, L. L., Wang, X. H., Bi, S. J., Xue, Q. Q.,., 2019. Application of plackett-burman design in screening of natural antioxidants suitable for anchovy oil., 8 (12): https://doi.org/10.3390/antiox8120 627.

Zakwan, J., E., and Lubis, Z., 2017. Production mono-diglyce- ride (MDG) from refined deodorized palm oil (RBDPO) by enzymatic process., 24 (1): 56-59.

Zhang, L. Y., Ding, L., Shi, H. H., Xu, J., Xue, C. H., Zhang, T. T.,., 2019. Eicosapentaenoic acid in the form of phos- pholipids exerts superior anti-atherosclerosis effects to its tri- glyceride form in ApoE-/-mice., 10 (7): 4177- 4188, https://doi.org/10.1039/C9FO00868C.

Zhong, Y., Madhujith, T., Mahfouz, N., and Shahidi, F., 2007. Compositional characteristics of muscle and visceral oil from steelhead trout and their oxidative stability., 104 (2): 602-608, https://doi.org/10.1016/j.foodchem.2006.12. 036.

.0086-532-82032597 E-mail: jxm@ouc.edu.cn

August 19, 2020;

November 3, 2020;

January 27, 2021

? Ocean University of China, Science Press and Springer-Verlag GmbH Germany 2021

(Edited by Qiu Yantao)