Optimization of conventional solid-liquid extraction and microwave-assisted extraction of polyphenols and antioxidant compounds of blueberry ( Vaccinium corymbosum ) pomace through response surface methodology

2.1 Reagents and standards

Folin and Ciocalteu’s phenol reagent (2 N), sodium carbonate anhydrous, cyanidin-3-o-glucoside, sodium carbonate, potassium chloride, ethanol absolute were purchased from Merck (Darmstadt, Germany). 2,2-diphenyl-1-picrylhydrazyl (DPPH) and 6-hydroxy-2,5,7,8-tetramethylchroman-2-carboxylic acid (Trolox) were purchased from Sigma-Aldrich (Steinheim, Germany). Sodium acetate and glacial acetic acid (99.8%) were purchased from Winkler LTDA (Santiago, Chile). Hydrochloric acid (36.5–38.0%) was purchased from JT Baker (Deventer, Holland).

2.2 Raw material and sample preparation

Blueberries (Vaccinium Corymbosum) were acquired from blueberry plantations in Parral (Maule Region), Linares Province, Chile. Blueberries from the “Legacy” variety were used, which were collected and immediately frozen at –20°C until further processing. Subsequently, the frozen berries were peeled with a surgical knife to obtain the pomace and the parts that composed them, i.e., seeds and peel, were ground with a food processor (Moulinex AD6011, Ecully, France) until reaching a particle size smaller than 1.6 mm. Then, the samples obtained were deposited in small plastic bags and frozen at –20°C until use. Each sample was defrosted at 4°C for 12 h before use.

2.3 Conventional solid-liquid extraction (SLE)

For each extraction, 1.24 g of the ground blueberry pomace sample was added to 9.92 mL of dissolvent solution (ethanol-water) to the concentration specified in the experimental design (Table 1). Afterwards, using a pH-meter (HI 111, Hanna Instruments^®, Rumania), the pH of the marc-solvent mixtures was adjusted to the operating pH as specified in Table 1. Once pH was adjusted, the mixture was careful poured in a 15 mL test tube with a lid and protected from light. Subsequently, the test tube was placed horizontally and immobilized inside a shaking incubator (NB-205, N-BIOTEK Inc, Korea) previously heated to operating temperature as specified in the experimental design (Table 1). Then, the agitation of the incubator was set at 200 rpm allowing pomaces to move along the test tube to initiate the extraction process. After the extraction time in Table 1, all the extract was transferred to 1.5 mL Eppendorf tubes that were then centrifuged (IEC Centra MP4R Benchtop Centrifuge, Manasquan, New Jersey, USA) at 13000 rpm, for 12 min, at room temperature. Finally, the supernatant of each Eppendorf tube was deposited in an amber glass flask for storing the blueberry marc extract at 4°C until analysis. Each procedure was performed in triplicate.

The extractions were conducted in a household microwave oven (Fancy WT1700, Somela^®, Maipú, Chile). For each extraction, 1.24 g of ground blueberry pomace were weighed in a digital scale (Kern^®, PFB 3000-2, Balingen, Germany) and then deposited inside a glass bottle (Schott AG, Mainz, Germany) with a lid. Subsequently, 14.3 mL of each ethanol-water dissolvent solution were added as specified in Table 1 and adjusted to pH 2.0. Once pH is adjusted, sample temperature was measured and the sample was placed into a microwave, where extraction was carried out. The samples were irradiated with microwaves under the power conditions specified in the experimental design (Table 1). After the extraction time specified in Table 1, the sample was taken out from the microwave and the temperature of the extract was measured. Next, the volume of the extract obtained was measured and the extract was deposited in 1.5 mL Eppendorf tubes for centrifugation at 13000 rpm, for 12 min, at room temperature. Finally, the supernatant of each Eppendorf tube was deposited in an amber glass flask for the storage of the blueberry extract at 4°C for its subsequent analysis. Each procedure was performed in triplicate.

2.5 Determination of total phenolic content (TPC) and total anthocyanin content (TAC)

Total phenol content in the blackberry marc extract was measured through a spectrophotometric method that employed the Folin-Ciocalteau reagent [23]. An alicuota of marc extract (50μL) and 2 N Folin-Ciocalteu reagent (250μL) were added into a test tube. The solution was homogenized with a vortex mixer (Velp Scientifica^® ZX3, Usmate, Italy) and the mixture was incubated for 3 min, after which 20 %sodium (750μL) was added. The solution was then filled up with distilled water until a total volume of 5 mL and homogenized. After 2 h in the dark at room temperature, absorbance was read with a UV-Visible spectrophotometer (Shimadzu, UV Mini-1240, Japan) at 760 nm. Accurate serial dilutions of gallic acid (50–1000 mg/L) were prepared to construct a standard calibration curve (y = 0.00113× +0.0029; R² = 0.997). TPC was expressed in mg of gallic acid equivalent (GAE) per 100 g of fresh fruit (F.F.). All measurements were performed in triplicate.

Anthocyanin total content was calculated through the differential pH method [24] using a UV-Vis spectrophotometer (UV Mini-1240, Shimadzu, Japan). Absorbance (A) was read at 520 nm and 700 nm in 1.0 and 4.5 pH buffers, respectively, using A = (A₅₂₀ - A₇₀₀) pH 1.0 - (A₅₂₀ - A₇₀₀) pH 4.5 with a molar extinction coefficient of 26.900. The results are expressed in cyanidin-3-glucoside mg per 100 g of F.F. All measurements were performed in triplicate.

2.6 Determination of antioxidant capacity by DPPH assay

The antioxidant capacity of the extract was calculated using the 2.2-Dipheny-l-picrylhydrazyl (DPPH) free radical. DPPH free radical sweeping activity was measured using the method by Brand-Williams et al. [25]. The test was carried out by taking 0.1 mL of marc extract and react with 2.9 mL of the DPPH solution (63μM) for 30 min in the dark. Absorbance was read at 517 nm. A standard calibration curve was drawn using a Trolox Equivalent standard (6-hydroxy-2,5,7,8-tetramethylchroman-2-carboxylic acid) (y = 0.1119×–0.2700; R² = 0.999). The results were expressed as mg Trolox Equivalent (TE) per 100 g of F.F. All measurements were performed in triplicate.

2.7 Experimental design

This study dealt with two extraction techniques (SLE and MAE). An experimental design was created using the response surface methodology for each technique in order to find the optimal conditions for maximizing the extraction of phenols, anthocyanins and antioxidants from blueberry pomace. These designs were developed and implemented using the STATGRAPHICS Centurion XVI (16.2.04 version) program.

For conventional solid-liquid extraction (SLE), the design created studied four independent variables (ethanol concentration, pH, temperature and extraction time), which had three levels according to the conditions presented in Table 1, and through a Box-Behnken response surface design with 28 experimental runs, 4 central points per block, one single block, 13 degrees of freedom, and randomized, as shown in Table 2.

For microwave assisted extraction, 3 independent variables were studied (ethanol concentration, microwave power and irradiation time) according to the conditions presented in Table 1 and through a Box-Behnken response surface design with 16 experimental runs, 4 central points per block, a single block, 6 degrees of freedom and randomized, as shown in Table 3.

The study of the experimental factors above for each extraction method focused on measuring three response variables: total phenol concentration through the Folin-Ciocalteu method, anthocyanin concentration through the differential pH method and antioxidant capacity through the DPPH method.

Experimental data were adjusted to the quadratic polynomial model presented in the following general equation (Equation (1)): Y = b 0 + ∑ i = 1 k ( b i X i ) + ∑ i = 1 k ( b ii X i 2 ) + ∑ ij = 1 k ( b ij X i X j )(1)

Where, Y is the response value; X_i and X_j represent the values of independent variables; b₀ is a constant coefficient, and b_i¸ b_ii and b_ij are regression coefficients of linear, quadratic and interaction terms, respectively.

Once regression models that relate the experimental variables under study to total phenol concentration, total anthocyanin concentration and antioxidant capacity are established, the application of the desirability to conduct a multiple response optimization was possible. The desirability function is an optimization technique that allows analyzing the three responses studied simultaneously. The desirability value varies from 0 to 1, and represents the closeness of a response to its ideal value, where a desirability of 1 indicates that the response reaches the optimal or ideal value. If any of the responses analyzed do not reach its ideal value, the specific and total convenience for such a response will be <1 [26].

The following scenario was considered when maximizing the function d_i (for the extraction of phenols, anthocyanins and antioxidants from blueberry pomace): d i = { 0 [ y i - y i , min y i , max - y i , min ] r 1 y i ⩽ y i , min y i , min < y i < y i , max y i ⩾ y i , max for i = 1 , 2 , … … , k Consequently,

d _i  = 0 indicates a value of the response that is least desirable

Where, y_i,min and y_i,max are the minimum and maximum acceptable values of the response (y_i), respectively, and r is the weight factor that determines the shape of the desirability graph.

The total desirability is defined as a geometric mean of the individual desirability (Equation (2)): D = ( d 1 x d 2 x … d k ) 1 / k(2)

Where, D is the total desirability and d_i is the i^th desirability, i = 1, 2,..., k. If all of the quality characteristics reach their ideal values, the desirability d_i is 1 for all i.

Model parameters, optimal conditions and the layout of the response surface were estimated used the STATGRAPHICS Centurion XVI (16.2.04 version) software. To calculate the statistically significant differences between the values obtained, a multifactor variance analysis (ANOVA) was conducted, which yielded a significance level of α= 0.05. Furthermore, the effects and their interactions were analyzed in a Pareto diagram, reflecting the significance of the effects and how these contribute to a specific response. With both designs, model adjustment was assessed through the determination coefficient (R²), R²-adjusted, lack of fit and p-valor obtained from the variance analysis (ANOVA). All the results are presented as the means of the triplicates together with their standard deviation.

3.1 Optimization of SLE by Response Surface Methodology (RSM)

The effect of four independent variables (ethanol concentration (%(v/v), X₁), pH (X₂), temperature (°C, X₃) and extraction time (min, X₄)) on total phenol content, anthocyanin concentration and antioxidant capacity was studied. The results of the response variables (Y) are presented in Table 2.

Total phenol concentration in the obtained extracts varied between 75.0±6.7 and 371.3±10.2 (mg GAE/100 g F.F), a range below the one reported by Fredes et al. [27], who found ranges around 520–720 (mg GAE/100 g F.F) for the same variety. This can be due to the differences in the operating conditions, degree of ripeness or harvest conditions [28, 29]. In turn, anthocyanin concentration in the obtained extracts varied between 63.2±1.1 and 245.8±0.8 (mg cyn-3-glu/100 g F.F); these values are similar to those reported by Bunea et al. [30] and Connor et al. [31] for the same variety. The antioxidant capacity of the obtained polyphenolic extracts varied between 172.3±4.1 and 459.5±5.8 (mg TE/100 g F.F), which is similar to the results of Connor et al. [32] and Ruiz et al. [32] for the same variety. However, the antioxidant capacity of the blueberry extracts is influenced by diverse factors such as genotype, ripeness and post-harvest conditions, among others [33].

Figure 1 presents the Pareto diagram for TPC, TAC and antioxidant capacity. These diagrams summarize the effects of all experimental variables on the response studied and show the statistically significant variables (p < 0.05). To study phenol concentration (Fig. 1A), the statistically significant variables (p < 0.05) were temperature (p = 0.0065), pH (p = 0.0176), ethanol concentration (p = 0.0042), and extraction time (p = 0.0375), in addition to the ethanol temperature-concentration interaction (p = 0.0355) and the quadratic effect of ethanol concentration (p = 0.0391). It was experimentally confirmed that pH is the only experimental factor that has a negative effect. In the case of anthocyanin concentration (Fig. 1B), pH (p = 0.0191), ethanol concentration (p = 0.0068), ethanol temperature-concentration interaction (p = 0.0400) and the quadratic effect of ethanol concentration (p = 0.0217) were statistically significant. Figure 1B shows that most statistically significant effects have a negative effect on anthocyanin concentration when the variable under study goes from a low level to a high level. This confirms the great instability of anthocyanins under the operating conditions, where relatively low levels of these variables have a better extraction performance for such compounds. Finally, for the variable of antioxidant capacity response (Fig. 1C), temperature (p = 0.0160), pH (p = 0.0420) and ethanol concentration (p = 0.0124), as well as the ethanol temperature-concentration interaction (p = 0.0307), were statistically significant (p < 0.05).

OPEN IN VIEWER

Fig. 1

Pareto charts for (A) total phenolic content; (B) total anthocyanin content and (C) antioxidant capacity of conventional solid-liquid extraction (SLE). The vertical line indicates significance at 95 %confidence level.

To optimize the extraction processes, the obtained results were adjusted to a polynomial quadratic model (Equation 1). The results of coefficient of determination, lack of fit and p values is shown in Table 4. Additionally, the same table presents the linear, quadratic and interaction terms that significantly affected the three response variables. Non-significant coefficients are not presented.

3.1.1 Response surface analysis of TPC, TAC and antioxidant capacity

Regarding the response variable of total phenol concentration in the experimental model applied, the coefficient of determination (R²), value that assesses the capacity of the model to predict the behavior of the response variable, presented a high correlation valor (Table 4). Likewise, the model shows an adjusted R² of 0.6390 and a lack-of-fit of 0.1462, which is higher at the significance level. Therefore, with 95 %confidence, it may be concluded that the lack-of-fit of the model is not statistically significant and consequently the model describes the results obtained adequately. However, despite its adequacy to describe the behavior of results, the model is no table to explain most of the variance of results, achieving only 63.9 %. Nevertheless, it is considered acceptable since a first order model adjustment has a R² = 0.5322.

Figure 2 shows the response surface diagrams obtained for the different responses in conventional solid-liquid extraction. It is observed how the four experimental factors influence total phenol concentration (Fig. 2a and 2b) whose behavior can be extended to all operating ranges under study. The effect of temperature on response is practically linear, with the highest operating temperatures having the best extraction performance. This was expected due to the bigger range of motion presented by both solvent and solute molecules, at high temperatures, generating a larger number of molecular interactions that translates into an increase in the solubility of the solute in the solvent, a rise in a diffusivity and a reduction in liquid viscosity [34]. The effect of ethanol concentration on response is quadratic, presenting a maximum value within the levels studied for such an effect, and in which the extraction of phenolic compounds is smaller, and even negatively affected, at concentrations above 55–60 (%). The latter is explained by the polarity difference presented by phenolic compounds from berries [35]. Li et al. [36] found that solvent concentration plays a critical role in the extraction of antioxidants from grape skin, who indicated that 48.8 %ethanol was most efficient for the recovery of grape skin antioxidant polyphenols. Similarly, Vrancheva et al. [37] reported that solvent concentration was the most important factor contributing to the extraction of different Vaccinium species components when RSM was used.

OPEN IN VIEWER

Fig. 2

Response surfaces for (a, b) total phenolic content; (c, d) total anthocyanin content and (e, f) antioxidant capacity obtained from the optimization design of the conventional solid-liquid extraction (SLE). (a, c and e) Effect of temperature and ethanol concentration; (b, d and f) Effect of pH and time.

The effect of pH on response is practically linear and inversely proportional in the 2.0–3.2 range, and quadratic in the 3.2–5.0 range. Likewise, at lower pH levels, extraction was improved, influenced mainly by the strong stability of anthocyanins at acid pH due to the predominance of their flavylium cation structure. The effect of time on response is quadratic and presents a maximum value within the range studied, but its effect is practically insignificant within the operating conditions studied.

Regarding the variable of total anthocyanin concentration of the experimental model applied, the model presents the coefficient of determinations (R²) and adjusted R² with values equal to 0.8442 and 0.6764, respectively, and the lack-of-fit of the model was 0.3164 (Table 4). Therefore, with 95 %confidence, it can be concluded that the regression model describes the behavior of the results adequately and is an acceptable model for explaining the variance of results, despite only explaining 67.6 %of the same. The diagrams in Fig. 2c and 2d indicate that the effect of the 4 experimental variables studied for assessing anthocyanin concentration is similar to that of total phenol concentration, which is reflected in the shape of the diagrams. Nevertheless, the response surfaces for anthocyanins are more curve in their ends and slightly steeper. This is explained by the fact that Fig. 2c and 2d addresses only one phenolic compound present in blueberries, anthocyanins, compared to the diagrams in Fig. 2a and 2b, which deal with all the phenolic compounds present, such as flavonols (quercetin and rutin), flavanols (catechin), hydroxycinnamic acids derivatives (chlorogenic acid), condensed tannins, and anthocyanins, among others [38]. This causes that one change in some of the experimental variables studied does not affect all compounds equally, minimizing the effects of such a change and generating flatter surface responses. However, concerning anthocyanins alone, all of them tend to behave in the same way when there is a change in operating conditions, thereby maximizing the effect of such a change in the performance of anthocyanin extraction. The latter leads to the conclusion that the most adequate operating conditions for obtaining polyphenolic extracts from blueberries are those that imply that anthocyanins present higher stability, such as moderate temperatures and acid media, which minimize their degradation. This result was consistent with Li et al., [36], who reported that the joint effects of temperature and extraction time had a significant negative effect on total anthocyanin content. They found that the extraction yield of TAC increased evidently with increasing of extraction temperature from 40 to 51°C, but beyond 51°C, the extraction yield of TAC had a declining trend.

Regarding the response variable of antioxidant capacity of the experimental model applied, the model presents a R² = 0.8293, an adjusted R² = 0.6455 and a lack-of-fit = 0.3033 (Table 4). This indicates that such a model is only able to explain 64.5 %of the variance of results. However, the model is considered acceptable for the optimization of the response studied, as models of orders below or above 2 exhibit worse R² correlation coefficients or it is not possible to apply them.

Figure 2e and 2f show the response surface obtained for antioxidant capacity. When comparing the response surfaces in these figures to those in Fig. 2a-2d, they are practically the same in shape, presenting only slight difference in their steeps and curves. This implies that the antioxidant capacity is closely related to the phenolic content and anthocyanin content present in the extract of blueberry pomace. Several studies have confirmed that the correlation between the antioxidant capacity and total phenol content and anthocyanin content is R² = 0.9938 and R² = 0.9543, respectively [39], and R² = 0.845 and R² = 0.77, accordingly [40].

Thus, it may be concluded that extraction conditions that maximize the recovery of total phenolic content and anthocyanins will also maximize the recovery of extracts with higher antioxidant capacity.

3.1.2 Optimum condition and its validation

To optimize the three responses simultaneously, the desirability function methodology was used in seeking to maximize the three responses studied. This series of experiments seeks to maximize the three responses studied. Therefore, after applying desirability function, the operating conditions for optimizing the total phenol concentration, anthocyanin concentration and antioxidant capacity of blueberry marc extracts are the following and based on them a desirability function of 1.0 is obtained.

The operating conditions that maximize the analyzed responses obtaining optimal levels, for the extraction of the compounds in this first designed by variable are the following: temperature: 50°C; pH: 2; ethanol concentration: 45 %and time: 108 min (Fig. 3). Through these, a maximum total phenol concentration of 371.87 (mg GAE/100 g F.F), of 285.92 anthocyanins (mg cyn-3-glu/100 g F.F), and a maximum antioxidant capacity of 498.96 (mg TE/100 g F.F) were obtained from the blueberry marc extracts (Table 5). The optimal value was validated through three extractions under optimal operating conditions. Subsequently, calculating the mean of these results, total phenol content, total anthocyanin content and real antioxidant capacity of the extracts was obtained under optimal extraction conditions, as shown in Table 5.

OPEN IN VIEWER

Fig. 3

3D surface diagram representing the desirability function for the optimization of the effects of ethanol concentration. pH, temperature and time for the recovery of total phenol content, total anthocyanin content and antioxidant capacity after solid-liquid extraction (SLE) of blueberry pomace.

OPEN IN VIEWER

Table 5

Comparison bewteen SLE and MAE methods with the experimental and predicted values obtained from the modeling for the response variables

Extraction methods		TPC (mg GAE/100 g)	TAC (mg Cyn-3-glu/100 g)	Antioxidant Capacity (mg TE/100 g)
SLE	Model predicted valuea	371.87	258.92	498.96
	Obtained valueb	335.95±22.00A	272.69±6.71A	528.96±15.95A
	Relative error (%)	9.66	5.32	6.01
MAE	Model predicted valuea	442.98	406.40	694.51
	Obtained valueb	426.19±37.19B	389.64±23.85B	654.11±14.05B
	Relative error (%)	3.79	4.13	5.82

^aPredicted using ridge analysis of response surface quadratic model; ^bMean±standard deviation of triplicate determinations from different experiments. Values with different letters (A, B) in the same column are significantly different (p < 0.05).

Table 5 shows that the real optimal values of the response variables do not differ much from the optimal values predicted for the multiple response optimization applied through desirability function, with errors below 10 %. These errors can be attributed to the fact that the regression models used for each response variable study are only able to explain on average 65 %of result variance, which allows the regression model to under- or overestimate the values of such variables.

The effect of the three independent variables (Ethanol concentration (%(v/v), X₁), Irradiation power (W, X₂), Irradiation time (s, X₃)) on total phenol content, anthocyanin content and antioxidant capacity was studied. The results of the response variables (Y) are presented in Table 3.

Total phenol concentration in the extracts varied between 112.0±13.2 and 428.4±8.0 (mg GAE/100 g F.F). Likewise, total anthocyanin concentration ranged from 98.1±6.0 to 393.5±8.0 (mg cyn-3-glu/100 g F.F); and antioxidant capacity was within the range of 195.8±7.1 and 709.2±11.8 (mg TE/100 g F.F). Zhang et al. [41] reported that the total content of anthocyanins extracted from the marc of Rabbiteye blueberries was 208.53±20.38 (mg cyn-3-glu /100 g of sample powder) and when comparing the efficacy of the different extraction techniques (high hydrostatic pressure, ultrasound, microwaves and heat), the microwave technique showed positive effect in the extraction of blueberry pomace, including a better extraction performance and shorter extraction time than other techniques.

Figure 4 presents the Pareto diagram for TPC, TAC and antioxidant capacity, as well as showing the statistically significant variables (p < 0.05). For the study of phenol concentration (Fig. 4A) the statistically significant variables (p < 0.05) were ethanol concentration (p = 0.0194), time (p = 0.0002), power (p = 0.0034), in addition to the ethanol-power concentration interaction ((p = 0.0312), the time-power interaction (p = 0.0204) and the quadratic effect of ethanol concentration (p = 0.0416) and power quadratic effect (p = 0.0329). In the case of anthocyanin concentration (Fig. 4B), ethanol concentration (p = 0.0186), time (p = 0.0001), power (p = 0.0020), in addition to the ethanol-power concentration interaction (p = 0.0121), the quadratic effect of ethanol concentration (p = 0.0255) and the quadratic power effect (p = 0.0144) were statistically significant (p < 0.05). Finally, for the variable of antioxidant capacity response (Fig. 4C), the statistically significant variables were ethanol concentration (p = 0.0033), time (p = 0.0001), power (p = 0.0017), in addition to the ethanol-power concentration interaction (p = 0.00195), the time-power interaction (p = 0.0097), the quadratic effect of ethanol concentration (p = 0.0074) and power quadratic effect (p = 0.0292).

OPEN IN VIEWER

Fig. 4

Pareto charts for (A) total phenolic content; (B) total anthocyanin content and (C) antioxidant capacity of microwave-assisted extraction (MAE). The vertical line indicates significance at 95 %confidence level.

The Pareto diagram (Fig. 4) shows that the effects of ethanol concentration, extraction time and microwave power are statistically significant for the three response variables studied. Nevertheless, the p value leads to the conclusion that extraction time will be the most determining factor for the concentration and antioxidant capacity of the phenolic extracts obtained.

3.2.1 Response surface analysis of TPC, TAC and antioxidant capacity

After adjusting the results to a regression model that estimated the effect of the statistically significant experimental variables on each response studied, a 95 %confidence level indicated that the most adequate models to describe the variance of results are those in Table 4, in which model parameters are presented for each independent variable.

Figure 5 shows diagrams of the surface response obtained for the different responses to microwave-assisted extraction. The influence of the 3 experimental factors on the responses studied can be also observed.

OPEN IN VIEWER

Fig. 5

Response surfaces for (a, b) total phenolic content; (c, d) total anthocyanin content and (e, f) antioxidant capacity obtained from the optimization design of microwave-assisted extraction (MAE); (a, c and e) Effect of ethanol concentration and irradiation time; (b, d and f) Effect of ethanol concentration and irradiation power. Z-Axis code in b,d and e figure: 1 = 204 W; 2 = 576 W; 3 = 792 W; 4 = 1020 W and 5 = 1200 W.

From Table 4, all the regression models established are observed to be more complex than the model studied in SLE, because the former include more variables or effects on which the response variable depends. Nevertheless, such models can explain 92 %of variance, and therefore are considered valid to describe the behavior of response variables under the effects of the experimental factors studied.

The diagrams in Fig. 5 show that the effect of ethanol concentration on the 3 response variables studied is quadratic, having a maximum value within the levels studied for such an effect. Consequently, the/an increase in ethanol concentration in the extractant medium will allow a higher extraction of phenolic compounds until 30–35 %. However, when ethanol concentration is higher, polyphenol extraction performance decreases due to the polarity of both water and ethanol molecules. The polarity of ethanol-water mixtures decreases when the concentration of ethanol is above 40 %. This change will affect ethanol molecules to a larger extent, since they have a lower dipole moment than water molecules. Therefore, added to the fact that anthocyanins are polar molecules, the solubility of phenolic compounds will decrease as the alcohol concentration in the solvent increases [13].

Our results are in accordance with previously published reports, such as Drevelegka et al., [42], who used microwave-assisted extraction in grape pomace, varying the concentration of ethanol (0–100 %), solvent/solid ratio (8–24 mL/g) and irradiation power (100–600 W), finding that the optimum concentrations was obtained under an ethanol concentration of 42 %, a solvent/solid ratio of 24 mL/g and irradiation power of 408 W, concluding that 42 %ethanol was the best solvent system for polyphenol extraction when microwaves were used. But, in some cases, an opposite effect was reported. Zheng et al. [13] used microwave-assisted extraction in blueberry and varying the concentrations of ethanol (40–80 %) found that the optimum concentrations were obtained with an ethanol concentration of 55.5 %to achieve the highest anthocyanin extraction rate of 73.73 %in blueberry.

Likewise, the effect of both microwave power and extraction time on the studied response is practically linear, and the high levels of both variables allow obtaining more concentrated extracts with higher antioxidant capacity. The rationale of using microwave radiation in the extraction of bioactive compounds is that radiation causes an increase in the temperature of the extractant medium, promoting the diffusion of phenolic compounds from the vegetal matrix to the solvent. Furthermore, this radiation is absorbed by some substances inside vegetable cells, especially by polar molecules like water, which sharply increases the internal temperature of cells until the cell walls break, thereby facilitating the extraction of bioactive compounds of interest [43]. Thus, a stronger microwave power will be translated into a temperature increase of both the solvent and the solid, promoting the diffusion and extraction of phenolic compounds over time.

3.2.2 Optimum condition and its validation

Finally, based on the regression models presented in Table 4, a multiple response optimization was conducted by applying the desirability function. As a result, the operating conditions that maximize the total phenolic and anthocyanin contents of the blueberry marc extracts, as well as their antioxidant capacity, obtained a desirability function of a 0.990 are presented below.

The operating conditions that maximize the 3 responses analyzed, obtaining optimal levels for the extraction of compounds in this second design, are: ethanol concentration: 29.71%, power: 1200 W and time: 95 s (Fig. 6). Through these conditions, a maximum total phenol concentration of 442.98 (mg GAE/100 g F.F), a total anthocyanin concentration of 406.40 (mg cyn-3-glu/100 g F.F), and a highest antioxidant capacity of 649.51 (mg TE/100 g F.F) were obtained from the blueberry marc extracts (Table 5). This optimal value was validated through 4 extractions under optimal operating conditions. After calculating the means of these results, total phenol content, total anthocyanin content and real antioxidant capacity of the extracts under optimal operating conditions were obtained, as shown in Table 5.

OPEN IN VIEWER

Fig. 6

3D surface diagram representing the desirability function for the optimization of the effects of ethanol concentration. Power, time for the recovery of total phenol content, total anthocyanin content and antioxidant capacity after Microwave Assisted Extraction (MAE) of blueberry pomace. Z-Axis code in b,d and e figure: 1 = 204 W; 2 = 576 W; 3 = 792 W; 4 = 1020 W and 5 = 1200 W.

In Table 5, the real optimal values of the response variables are practically the same as the values predicted by multiple response optimization with errors below 6 %. In addition to validating the optimal value found for microwave assisted extraction, this confirms that the regression models established for total phenol concentration, total anthocyanin concentration and antioxidant capacity, are the most adequate for describing the behavior of each response under the effect of the experimental variables studied.

When comparing real optimal values of response variables for both conventional solid-liquid extraction and microwave-assisted extraction, the results are shown in Table 5.

The results of one-way variance analysis showed a significant difference between SLE and MAE. In Table 5, the SLE yielded 335.95±22.00 mg GAE/100 g; 272.69±6.71 mg Cyn-3-glu/100 g; 528.96±15.95 mg TE/100 g for TPC, TAC, AC, respectively. MAE yielded 442.19±37.19 mg GAE/100 g; 389.64±23.85 mg Cyn-3-glu/100 g and 654.11±14.05 mg TE/100 g, TPC, TAC and antioxidant capacity, respectively. Based on these results, it may be concluded that microwave- assisted extraction is a suitable technique for retrieving the phenolic compounds from blueberry pomace, as it achieves an extraction yield almost 20 %higher than conventional solid-liquid extraction for the total phenol content and antioxidant capacity of extracts, as well as yields 40 %higher for total anthocyanin content. This is consistent with the literature, in which yields of phenolic compound extraction from vegetal matrixes are better with MAE than with SLE [21, 44]. Likewise, microwave-assisted extraction presents higher levels of selectivity towards anthocyanin extraction compared to conventional solid-liquid extraction, due to the polarity presented by such compounds.

The above, added to other advantages such as shorter extraction times, use of smaller quantities of solvent and products of better quality, make microwave-assisted extraction an attractive technique to be implemented in the extraction of polyphenolic from blueberry pomace extracts. Although, its implementation is limited by investment and operating costs, which are justified by a better extraction yield observed.