Effect of the Shape of Ultrasonic Vessels on the Chemical Properties of Extracts from the Fruit of Sorbus aucuparia

Abstract

The goal of this study was to analyse the effect of sonoreactor dimensions on the effectiveness of the ultrasound-assisted extraction (UAE) of bioactive substances from rowan (Sorbus aucuparia L.). Sonication was carried out with a VC750 Sonics processor at the following amplitudes of ultrasound: 12, 24, and 36 µm. The frequency of the ultrasound was 20 kHz. Extraction was conducted in a 2 s on–4 s off pulse system. The total phenolic content and antioxidant activity were determined using a spectrophotometric method. The pH value of the extracts was measured using a combined pH metric electrode, type EPS-1 (Elmetron). Response surface methodology (RSM) was used to optimise the investigated variables. On the basis of the developed model, the following variable values were obtained: TPC—12.48 gallic acid equivalent (GAE)/g, 2,2${}^{\prime }$-azino-bis-(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS) values—126.54 $\mathsf{\mu }$mol Trolox (TE)/g, 2,2-diphenyl-1-picryl-hydrazyl-hydrate (DPPH) values—79.58 $\mathsf{\mu }$mol TE/g, ferric reducing antioxidant power assay (FRAP) values—120.65 $\mathsf{\mu }$mol TE/g for an extraction vessel with a diameter of 35 mm and TPC—11.68 mg GAE/g, ABTS values—120.49 $\mathsf{\mu }$mol TE/g, DPPH values—75.90 $\mathsf{\mu }$mol TE/g, FRAP values—155.54 $\mathsf{\mu }$mol TE/g for an extraction vessel with a diameter of 25 mm. Significant influences of time, ultrasound amplitude, and alcohol concentration on the course of the extraction process in the tested extraction vessels were found. The optimum extraction conditions for an extraction vessel with a diameter of 35 mm were obtained for an ethyl alcohol concentration of about 60%, and for an extraction vessel with a diameter of 25 mm the optimum conditions were for an alcohol concentration in the range of 30–50%. A very strong positive linear correlation was found between the concentration of ethyl alcohol and the pH value of the obtained extracts for both extraction reactors. The developed models of pulsed ultrasound-assisted extraction were characterised by a good predictive capacity (77.49–91.39%) and can be used for obtaining polyphenols from rowan.

1. Introduction

In recent years, there has been a revival of interest in raw materials with strong antioxidant properties due to their health potential. Plants that were used in folk medicine are experiencing a renaissance, and rowan is an example of one. Currently, about 250 species of rowan are known [1]. In Poland, five species grow wild, including Sorbus aucuparia (S. aucuparia), commonly known as rowan [2]. The chemical composition of rowan fruits (genus Sorbus L.) includes sugars, organic acids, phenolic acids, flavonoid compounds, anthocyanins, tannins, cyanogenic glycosides, vitamins, and bioelements [3,4,5].

Due to their rich chemical composition, fruits and other organs of rowan have numerous therapeutic effects: antioxidant, antidiabetic, antihyperlipidemic, anti-inflammatory, antibacterial, anticancer, antiviral, antifungal, vasoprotective, neuroprotective, cardioprotective, and hepatoprotective [6,7,8,9]. In unconventional medicine, rowan fruits are used in the treatment of diabetes, metabolic disorders, kidney stones, arthritis, rheumatism, gastritis, duodenal and small intestine catarrhs, irritable bowel syndrome, and diseases of the liver and gallbladder [10] as a laxative and a detoxifying agent [11].

The choice of the extraction technique plays a crucial role in obtaining bioactive compounds from plant matter. Currently, different techniques are used, such as Soxhlet extraction, maceration, supercritical fluid extraction, microwave-assisted extraction, pressurised solvent extraction, extraction assisted by a pulsed electric field [12,13], and UAE. In contrast to other methods, UAE offers several benefits: a short extraction time, minimal solvent usage and low-cost equipment. The mechanism principle behind the enhanced extraction process facilitated by ultrasound involves the disruption of plant cells, and releasing their contents into the solvent [14]. Process conditions should be selected in such a way as to avoid thermal and chemical degradation as well as biochemical changes in phenolic compounds. The effectiveness of the extraction process is influenced by many factors, including time, power, type and amount of solvent, pH, temperature, the degree of material fragmentation, and the operating mode of the ultrasonic processor (continuous or pulsed) [15].

The process of acoustic energy transmission to the extracted sample depends greatly on the type of ultrasonic system used. The shape of the reaction vessel is critical during utilising an ultrasonic bath. An important factor here is, in particular, the lack of homogeneity of the sound pressure in the entire volume of the sample due to the point arrangement of the emitters in the ultrasonic bath [16]. Progress in the construction of acoustic generators has allowed the use of transducers immersed directly in the solution, a so-called probe system. This technique allows a higher energy to be transferred to the solvent, but this system also has a sharp decrease in ultrasound intensity in both the longitudinal and transverse directions. Some authors have tried to optimise the extraction efficiency by changing the immersion of the ultrasonic probe in extraction flasks. Sun et al. [17] observed a substantial decrease in the extraction yield of $\beta$-carotene from citrus peels as the liquid height increased. Kulkarni et al. [18] investigated how the size and diameter of the extraction vessel impact the amount of mangiferin extracted from Mangiferaindica leaves. Their findings demonstrated that as the extraction vessel’s diameter grows, the yield of mangiferin increases until reaching a certain threshold, after which it stabilises.

The current state of knowledge prompts us to assume that the shape of the extraction vessels may have a significant impact on the extraction efficiency. Therefore, the aim of the study was to analyse the impact of the dimensions of the extraction vessels on the efficiency of extraction of rowan fruits assisted by a pulsed ultrasonic field.

2. Materials and Methods

2.1. Raw Material and Regents

Dried fruits of S. accapuria L. were purchased from Runo and have an ecological certificate PL-EKO-04 EU Agriculture. Rowan was harvested in 2022 in Podlasie (Poland). The fruits were stored in a climatic chamber for two months until analysis.

Gallic acid (98% pure), Folin–Ciocalteu reagent (FCR), sodium carbonate (powder, ≥99.5%), and methanol (anhydrous, 99.8%) were used for the analysis of polyphenols. For determining antioxidant activity, 2,2-diphenyl-1-picreylhydrazyl, 2,4,6-Tris(2-pyridyl)-s-triazine (TPTZ), Trolox, iron (III) chloride (anhydrous, powder, ≥99.99%), acetic acid, and hydrochloric acid were used, along with 2,2’-Azino-bis (3-ethylbenzothiazoline-6-sulfonic acid) diammonium salt and phosphate-buffered saline (PSB). All chemical reagents for analytical processes were of analytical quality and were procured from Sigma-Aldrich–Merck (Taufkirchen, Germany). A UV 1800 (Shimadzu; Japan) spectrophotometer was used to measure the absorbance.

2.2. Grinding of Raw Material and Sieve Analysis

The raw material was crushed using a Zelmer MM 1200 apparatus and was subsequently divided into fractions utilising a horizontal sieve shaker AS 400 Control. The fraction obtained after passing through sieves with a mesh diameter ranging from 1.0 to 2.0 mm was chosen for further analysis.

2.3. Ultrasound-Assisted Extraction

One and a half dried fruits were placed inside the extraction chamber and subsequently covered with 50 mL of an aqueous solution containing either 30%, 60%, or 90% ethyl alcohol. The ratio of raw material to solvent was 0.03 g/mL. Next, the extraction chamber was put in a cooling jacket that was connected to an ultra-thermostat in order to stabilise the temperature. The extraction chamber was covered using a 19 mm diameter ultrasonic probe at the top. The parameters of the extraction chambers were as follows: flask number 1: diameter 35 mm, aspect ratio (d/h = 1.75) and flask number 2: diameter 25 mm, aspect ratio (d/h = 0.33).

For the extraction of the samples, a VC750 Sonics processor operating at a frequency of 20 kHz was utilised. The samples were treated at three different ultrasound amplitudes: 12, 24, and 36 µm, corresponding to ultrasound intensities of 1.3, 7.5, and 14 W/cm${}^2$ [19]. The sonication process was carried out in the following processor arrangement: 2 s on and 4 s off. The effective sonication times were 5, 10, and 15 min, resulting in total extraction times of 15 min, 30 min, and 45 min, respectively. After ultrasonic treatment, the extracts were stored in a refrigerator at 2 °C for subsequent chemical analysis.

2.4. Box–Behnken Plan and Statistical Analyses

In this study, a three-level, three-factor Box–Behnken plan was implemented using Design-Expert v13 software from Stat-Ease in Minneapolis, MI, USA. The independent variables consisted of extraction time ($X_1$), ultrasound amplitude ($X_2$), and solvent concentration ($X_3$), while the dependent variables included total phenolic content (TPC), antioxidant activity measured using DPPH, FRAP, and ABTS assays, and pH. The experiment comprised a total of 15 randomly conducted combinations, including 3 center points utilised for estimating the pure error [20]. The levels of the three factors evaluated in the experiment are shown in

Table 1. The levels of the Box–Behnken plan.

Run	Times	Ultrasound Amplitude	Solvent Concentration
1.	5	12	60
2.	5	24	30
3.	5	24	90
4.	5	36	60
5.	10	12	30
6.	10	12	90
7.	10	24	60
8.	10	24	60
9.	10	24	60
10.	10	36	30
11.	10	36	90
12.	15	12	60
13.	15	24	30
14.	15	24	90
15.	15	36	60

.

A second-order polynomial model was employed to demonstrate the impact of the independent variables on the response variable using the following equation [20].

$$Y={\beta }_0+{\beta }_1{X}_1+{\beta }_2{X}_2+{\beta }_3{X}_3+{\beta }_{11}{X}_1^2+{\beta }_{22}{X}_2^2+{\beta }_{33}{X}_3^2+{\beta }_{12}{X}_1{X}_2+{\beta }_{13}{X}_1{X}_3+{\beta }_{23}{X}_2{X}_3$$

In the provided equation, Y represents the response variables, including TPC, ABTS value, DPPH value, FRAP value, and pH. The independent variables are denoted as $X_1$, $X_2$, and $X_3$. The equation includes the constant term ${\beta }_0$, as well as the linear (${\beta }_{1,2,3}$), quadratic (${\beta }_{11,22,33}$), and interactive (${\beta }_{12,13,23}$) coefficients [20].

The statistical analysis was performed in ANOVA. The significance of the regression coefficient was assessed using an F-test, considering p-values below 0.05 as statistically significant. The best conditions were determined using Derringer’s desirability function, aiming to maximise the response for each independent variable. The validity of the developed model was evaluated by comparing the predicted values with the actual experimental results [20].

2.5. TPC Assay

The total phenolic content was assessed following the methodology proposed by Singleton et al. [21] with a modification made by Kobus et al. [19]. The measurement was performed in three replicates for each sample. The results are expressed as mg gallic acid equivalent per 1 g of dry matter (mg GAE g${}^{-1}$ dry matter).

2.6. Antioxidant Activity

2.6.1. DPPH Assay

The determination of free radical scavenging activity with the DPPH assay was performed following the methodology worked out by Blois [22] with some adjustments made by Kobus et al. [23]. For this analysis, 60 $\mathsf{\mu }$L of the extract was mixed with an aliquot of 5 mL of freshly prepared $6·{10}^{-5}$ M DPPH radical in methanol. The measurement was performed in three replicates for each sample. Antioxidant activity was expressed as a Trolox equivalent in µg per g of dry matter.

2.6.2. FRAP Assay

The ferric-reducing antioxidant power assay was carried out according to the methodology described by Benzie and Strain [24] with some adjustments made by Kobus et al. [23]. In this study, the extracts were diluted 5 times with distilled water. Then, 60 $\mathsf{\mu }$L of the diluted extract was taken and 5 mL of FRAP was added. The absorbance was measured at 596 nm. Each sample was measured in triplicate. Antioxidant activity was expressed as a Trolox equivalent in µg per g of dry matter.

2.6.3. ABTS Assay

To prepare the reagent, 0.032 g of potassium persulfate was dissolved in 50 mL of PSB, then ABTS (0.192 g) was added. The mixture was left in the dark at room temperature for 16 h. Subsequently, ABTS solution was then diluted with distilled water to obtain an absorbance of 0.70 ± 0.05 at 729 nm. The extracts were diluted by a factor of 5 using distilled water. Then, 60 $\mathsf{\mu }$L of the extract was combined with 6 mL of the ABTS solution. After a 30 min incubation period, the absorbance at 729 nm was registered. The antioxidant activity was quantified as a Trolox equivalent in µg per g of dry matter. Each sample was measured in triplicate.

2.7. pH Measurement

The pH value of the extracts was measured using a combined pH-metric electrode of type EPS-1 (Elmetron). Three measurements were taken for each of the extracts and the average was calculated.

3. Results and Discussion

3.1. TPC

Figure 1 shows the influence of input variables on the TPC content for two extraction vessels with diameters of 35 and 25 mm.

The shape of the curves shown in Figure 1a indicates that the inflection points have been reached for the variables of ultrasound amplitude and solvent concentration, which means that further increases in the values of these variables will result in a decrease in the polyphenol extraction efficiency. There is a positive linear relationship between the extraction time and polyphenol content, which means that the extraction efficiency increases with increasing time. For the vessel with a diameter of 25 mm (Figure 1b), an inflection point was reached only in the case of solvent concentration. For the variables of ultrasound amplitude and time, we observed a non-linear growth in the TPC value with increasing time and a linear growth with increasing ultrasound amplitude.

In order to confirm the significance of the impact of the tested input variables on the TPC content, a statistical analysis was performed. The findings of this examination are displayed in

Table 2. Effect of the independent variables on the dependent variables and statistical significance for $TP{C}_1$ of extracts obtained in vessels with a diameter of 35 mm.

Source	Sum of Squares	DF	Mean Square	F-Value	p-Value
Model	85.93	5	17.19	17.86	0.0002	significant
$X_1$	6.95	1	6.95	7.22	0.0249
$X_2$	44.05	1	44.05	45.78	<0.0001
$X_3$	8.19	1	8.19	8.51	0.0171
$X_2^2$	12.43	1	12.43	12.92	0.0058
$X_3^2$	16.20	1	16.20	16.83	0.0027
Residual	8.66	9	0.9624
Lack of Fit	8.51	7	1.22	15.68	0.0612	not significant
Pure Error	0.1550	2	0.0775
Total	94.59	14
$R^2$ = 0. 9084, adj $R^2$ = 0. 8576, CV = 10.49, Adeq Precision 14.06

and

Table 3. Effect of the independent variables on the dependent variables and statistical significance for $TP{C}_2$ of extracts obtained in vessels with a diameter of 25 mm.

Source	Sum of Squares	DF	Mean Square	F-Value	p-Value
Model	78.55	5	15.71	30.14	<0.0001	significant
$X_1$	21.03	1	21.03	40.34	0.0001
$X_2$	2.68	1	2.68	5.14	0.0495
$X_3$	40.43	1	40.43	77.56	<0.0001
$X_1^2$	3.88	1	3.88	7.44	0.0233
$X_3^2$	9.58	1	9.58	18.38	0.0020
Residual	4.69	9	0.5213
Lack of Fit	4.43	7	0.6334	4.91	0.1797	not significant
Pure Error	0.2582	2	0.1291
Total	83.25	14
$R^2$ = 0. 9436, adj $R^2$ = 0. 9123, CV = 9.03, Adeq Precision 16.94

.

The results of the analysis of variance showed that for both models, the linear effects of all variables (time, ultrasound amplitude, and solvent concentration) and the square effects of the ultrasound amplitude and solvent concentration turned out to be statistically significant. Despite these similarities, it is easy to see that the impact of the input variables on the TPC content for individual models was different. For the vessel with a diameter of 35 mm, the linear effect of the ultrasound amplitude was the most important, and the linear effect of time was the least significant. For the vessel with a diameter of 25 mm, the linear effect of the solvent concentration was the most important, and the linear effect of the ultrasound amplitude was the least significant.

The Tukey test (p < 0.05) showed that in all cases of extraction with 60% solvent, the dimensions of the extraction vessels had no effect on the average TPC value. In the case of extraction of polyphenols using a 90% solvent, there were always differences in the average values depending on the extraction vessels used.

A regression analysis was employed to develop two models (Equations (1) and (2)) that describe the effect of input variables on TPC:

$$TP{C}_1=-8.762+0.186X_1+0.805X_2+0.245X_3-0.013X_2^2-0.002X_3^2$$

(1)

$$TP{C}_2=6.065-0.493X_1+0.048X_2+0.139X_3+0.041X_1^2-0.002X_3^2$$

(2)

Both models exhibit statistical significance (p = 0.0002; p < 0.0001), indicating their validity. The lack of fit for the model is statistically insignificant (p = 0.0612 and p = 0.1797), which further validates the accuracy of the model. The high values of the determinant coefficient (0.9084 and 0.9436) and the adjusted $R^2$ coefficient (0.8576 and 0.9123) demonstrate a strong correlation between the input variables and polyphenol content. The low values of the coefficient of variation (10.49% and 9.03%) indicate minimal differences between the predicted and experimental values, highlighting the precision and reliability of the conducted experiment.

The content of polyphenols in the obtained extracts ranged from 3.72 to 11.71 mg GAE/g for the vessel with a diameter of 35 mm and from 3.44 to 11.68 mg GAE/g for the vessel with the smaller diameter, depending on the other conditions of the experiment. The gained results are consistent with those of Šavikin et al. [25], who showed that the average value of TPC in S. aucuparia fruits is 8.62 mg GAE/g dry weight (d.w.). The results obtained in our experiment show the possibility of obtaining a higher extraction efficiency of polyphenols using a pulsed ultrasonic field compared to extraction in an ultrasonic water bath conducted for 20 min, where 10 g of the raw material was extracted in 50% ethyl alcohol (EtOH), which was used by Šavikin et al. [25]. Single reports by other authors are also available in the literature. For example, Denev et al. [26] reported that the content of polyphenols in S. aucuparia extracts was 2148.2 mg GAE/liter of extract. Olszewska et al. [27] showed that in S. aucuparia fruits, the TPC is 2.68 ± 0.08% GAE. Bozhuyuk et al. [28] reported that the TPC value for the 17 tested genotypes of S. aucuparia L. ranged from 123 mg to 189 mg GAE/100 g fresh weight (f.w.). Comparing the results of these authors, however, is difficult due to the expression of the TPC content in other units.

3.2. DPPH Assay

Figure 2 shows the dependence of the DPPH assay value on time, ultrasound amplitude, and solvent concentration for extracts obtained in extraction vessels with diameters of 35 and 25 mm.

The shapes of the curves shown in Figure 2a,b indicate that the solvent concentration has reached an inflection point, which means that further increases in this variable will result in a decrease in the DPPH value. For both vessels, the value of DPPH increases with time. For the vessel with a diameter of 35 mm, increasing the ultrasound amplitude translates into an increase in the DPPH value up to a certain point, and then a decrease in the value (Figure 2a). In the case of the vessel with a diameter of 25 mm, a non-linear increase in the value of the DPPH assay occurs with increasing ultrasound amplitude. Increasing the time translates into a linear increase in the DPPH value for both extraction vessels.

In order to confirm the significance of the effect of the examined input variables on the value of the DPPH assay, a statistical analysis was performed. The findings of this analysis are displayed in

Table 4. Effect of the independent variables on the dependent variables and statistical significance for $DPP{H}_1$ of extracts obtained in vessels with a diameter of 35 mm.

Source	Sum of Squares	DF	Mean Square	F-Value	p-Value
Model	2738.10	6	456.35	20.28	0.0002	significant
$X_1$	306.46	1	306.46	13.62	0.0061
$X_2$	1109.99	1	1109.99	49.33	0.0001
$X_3$	529.14	1	529.14	23.52	0.0013
$X_2·X_3$	142.11	1	142.11	6.32	0.0362
$X_2^2$	242.88	1	242.88	10.79	0.0111
$X_3^2$	451.51	1	451.51	20.07	0.0021
Residual	180.00	8	22.50
Lack of Fit	150.38	6	25.06	1.69	0.4170	not significant
Pure Error	29.63	2	14.81
Total	2918.10	14
$R^2$ = 0.9383, adj $R^2$ = 0.8921, CV = 7.97, Adeq Precision 16.93

and

Table 5. Effect of the independent variables on the dependent variables and statistical significance for $DPP{H}_2$ of extracts obtained in vessels with a diameter of 25 mm.

Source	Sum of Squares	DF	Mean Square	F-Value	p-Value
Model	3474.30	5	694.86	59.81	<0.0001	significant
$X_1$	134.59	1	134.59	11.58	0.0078
$X_2$	640.90	1	640.90	55.16	<0.0001
$X_3$	1991.79	1	1991.79	171.44	<0.0001
$X_2^2$	129.68	1	129.68	11.16	0.0086
$X_3^2$	536.07	1	536.07	46.14	<0.0001
Residual	104.56	9	11.62
Lack of Fit	99.08	7	14.15	5.17	0.1717	not significant
Pure Error	5.48	2	2.74
Total	3578.86	14
$R^2$ = 0.9708, adj $R^2$ = 0.9546, CV = 6.58, Adeq Precision 23.59

.

The results of the analysis of variance showed that for both models, the linear effects of all variables and the quadratic amplitudes of ultrasound and solvent concentration turned out to be statistically significant. In the case of extraction in the vessel with a diameter of 35 mm, the greatest influence on the value of the DPPH assay was the ultrasound amplitude, and the smallest influence was the time. In the case of the vessel with a diameter of 25 mm, the concentration of the solvent had the greatest influence, and the time had the least.

Both equations showed statistical significance (p < 0.0001). Additionally, the lack of fit of the models is statistically insignificant (p = 0.4170 and p = 0.1717), which indicates the correct validation of the model. The high values of the determination coefficient (0.9383 and 0.9708) and the adjusted coefficient (0.8921 and 0.9546) indicate a strong correlation between the input variables and the value of the DPPH assay. Moreover, the low CV values (7.97% and 6.58%) signify that the deviations between the experimental and predicted values are minimal, reflecting the high precision and reliability of the conducted experiment.

A regression analysis was employed to develop two models (Equations (3) and (4)) that describe the effect of the input variables on the DPPH values:

$$DPP{H}_1=-2.583+1.238X_1+2.684X_2+0.802X_3+0.017X_2{X}_3-0.056X_2^2-0.012X_3^2$$

(3)

$$DPP{H}_2=36.081+0.820X_1-1.224X_2+1.076X_3+0.041X_2^2-0.013X_3^2$$

(4)

The DPPH values range from 23.63 to 73.28 $\mathsf{\mu }$mol TE/g for the 35 mm diameter vessel and from 24.48 to 74.92 $\mathsf{\mu }$mol TE/g for the 25 mm diameter vessel depending on other experimental conditions. The obtained results are slightly lower than those reported in the literature. Olszewska et al. [27] showed that in the fruits of S. aucuparia the value of the DPPH assay is 105.5 $\mathsf{\mu }$mol TE/g d.w. Bobinaitė et al. [29] obtained a DPPH value of 103 $\mathsf{\mu }$mol TE/g of extract for ethanol extracts (S. aucuparia L.), and 309 $\mathsf{\mu }$mol TE/g of extract for aqueous extracts.

The performed ANOVA and Tukey’s test (p > 0.05) showed that the mean value of the DPPH assay obtained for both models differed significantly during the extraction using an ultrasound amplitude of 24 $\mathsf{\mu }$m for 10 min and an ultrasound amplitude of 36 $\mathsf{\mu }$m at a solvent concentration of 90%. The largest difference between the average values of the DPPH assay was 55% and concerned the extracts obtained with the following parameters: time, 15 min; ultrasound amplitude, 24 µm; and solvent concentration 90%.

3.3. FRAP Assay

Figure 3 shows the influence of the input variables on the value of the FRAP assay for extraction vessels with diameters of 35 and 25 mm.

The shape of the curve shown in Figure 3a indicates that the solvent concentration variable has reached an inflection point, which means that further increases in the value of this variable will result in a decrease in the FRAP value. For the vessel with a diameter of 25 mm, a linear decrease in the FRAP value was observed with increasing solvent concentration. For the vessel with a diameter of 35 mm, a linear increase in the FRAP value with increasing time and a non-linear increase with increasing ultrasound amplitude were observed. For the vessel with a diameter of 25 mm, increases in the ultrasound time and amplitude result in a linear increase in the value of the FRAP assay (Figure 3b).

In order to confirm the significance of the influence of the examined input variables on the value of the FRAP assay, a statistical analysis was performed. The findings of this analysis are displayed in

Table 6. Effect of the independent variables on the dependent variables and statistical significance for $FRA{P}_1$ of extracts obtained in vessels with a diameter of 35 mm.

Source	Sum of Squares	DF	Mean Square	F-Value	p-Value
Model	5130.54	6	855.09	33.13	<0.0001	significant
$X_1$	186.40	1	186.40	7.22	0.0276
$X_2$	3136.07	1	3136.07	121.50	<0.0001
$X_3$	1153.80	1	1153.80	44.70	0.0002
$X_1·X_3$	174.19	1	174.19	6.75	0.0317
$X_2^2$	138.85	1	138.85	5.38	0.0490
$X_3^2$	371.21	1	371.21	14.38	0.0053
Residual	206.48	8	25.81
Lack of Fit	197.48	6	32.91	7.31	0.1252	not significant
Pure Error	9.00	2	4.50
Total	5337.02	14
$R^2$ = 0. 9613, adj $R^2$ = 0. 9323, CV = 5.20, Adeq Precision 19.14

and

Table 7. Effect of the independent variables on the dependent variables and statistical significance for $FRA{P}_2$ of extracts obtained in vessels with a diameter of 25 mm.

Source	Sum of Squares	DF	Mean Square	F-Value	p-Value
Model	8936.26	3	2978.75	13.16	0.0006	significant
X1	1226.55	1	1226.55	5.42	0.0400
X2	2774.67	1	2774.67	12.26	0.0050
X3	4935.04	1	4935.04	21.80	0.0007
Residual	2490.42	11	226.40
Lack of Fit	2440.31	9	271.15	10.82	0.0874	not significant
Pure Error	50.11	2	25.06
Total	11,426.68	14
$R^2$ = 0.7821, adj $R^2$ = 0.7226, CV = 15.05, Adeq Precision 11.19

.

It can be observed that in the case of both analysed models, the linear effects of all variables had a significant impact on the value of the FRAP assay. In the case of extraction in the vessel with a diameter of 35 mm, the square coefficients of solvent concentration and ultrasound amplitude are also statistically significant.

In the case of extraction in the vessel with a diameter of 35 mm, the ultrasound amplitude had the greatest impact on the value of the FRAP assay, and time had the least impact. In the case of extraction in the vessel with a diameter of 25 mm, the solvent concentration had the greatest impact on the FRAP value, and time had the least effect.

The influence of individual variables on the FRAP values for both models is the same as in the case of the DPPH values.

The p-values of models are statistically significant (p < 0.0001; p = 0.0006), and the lack of fit of the models are statistically insignificant (p = 0.1252 and p = 0.0874), which indicates the correct validation of the equations. In the case of the first model ($FRA{P}_1$), the determination coefficient (0.9613) and the adjusted coefficient (0.9323) are high which shows the existence of a high correlation between the input variables and the FRAP value. In the case of second equation ($FRA{P}_2$), the determination and adjusted coefficients are lower (0.7821 and 0.7226) which signifies that the scatter of the data around the regression lines is higher. Furthermore, the coefficient of variation for the first model (5.20%) is much lower than for the second (15.05%), which means that the deviations between the experimental and predicted values are lower.

A regression analysis was employed to develop two models (Equations (5) and (6)) that describe the effect of input variables on the FRAP value:

$$FRA{P}_1=43.073-1.674X_1+3.688X_2+0.493X_3+0.044X_1{X}_3-0.042X_2^2-0.011X_3^2$$

(5)

$$FRA{P}_2=87.629+2.476X_1+1.552X_2-0.828X_3$$

(6)

The FRAP values range from 59.46 to 119.00 $\mathsf{\mu }$mol TE/g for the vessel with a diameter of 35 mm and from 52.98 to 154.27 $\mathsf{\mu }$mol TE/g for the vessel with a diameter of 25 mm depending on other experimental conditions. There are significant differences in the FRAP values in the available literature. Olszewska et al. [27] reported that the FRAP value is 347.0 $\mathsf{\mu }$mol TE/g d.w. Bobinaitė et al. [29] reported that the value of the FRAP assay is 118 ± 1.6 $\mathsf{\mu }$mol TE/g of extract. Bozhuyuk et al. [28] indicated genotype differences, obtaining FRAP values of 3.36 and 6.92 mM TE/100 g f.w. Mikulic-Petkovsek et al. [30] calculated values from 3.4 to 4.9 mM TE/100 g f.w. for rowan varieties growing in the Czech Republic.

3.4. ABTS Assay

Figure 4 shows the influence of input variables on the ABTS assay value for extraction vessels with diameters of 35 and 25 mm.

Based on the analysis of the graph, it can be concluded that in the case of the vessel with a diameter of 35 mm, the ABTS value decreases linearly with increasing solvent concentration (Figure 4a). The shape of the curve shown in Figure 4b indicates that the solvent concentration variable has reached an inflection point, which means that further increases in the value of this variable will result in a decrease in the ABTS value. Increases in the ultrasound time and amplitude resulted in a linear increase in the ABTS value for both extraction vessels.

In order to confirm the significance of the impact of the tested input variables on the value of the ABTS assay, a statistical analysis was performed. The findings of this analysis are presented in

Table 8. Effect of the independent variables on the dependent variables and statistical significance for $ABT{S}_1$ of extracts obtained in vessels with a diameter of 35 mm.

Source	Sum of Squares	DF	Mean Square	F-Value	p-Value
Model	8514.31	3	2838.10	10.41	0.0015	significant
$X_1$	1427.29	1	1427.29	5.24	0.0429
$X_2$	5631.75	1	5631.75	20.66	0.0008
$X_3$	1455.27	1	1455.27	5.34	0.0413
Residual	2998.84	11	272.62
Lack of Fit	2959.49	9	328.83	16.72	0.0577	not significant
Pure Error	39.34	2	19.67
Total	11,513.15	14
$R^2$ = 0. 7395, adj $R^2$ = 0. 6685, CV = 17.46, Adeq Precision 9.39

and

Table 9. Effect of the independent variables on the dependent variables and statistical significance for $ABT{S}_2$ of extracts obtained in vessels with a diameter of 25 mm.

Source	Sum of Squares	DF	Mean Square	F-Value	p-Value
Model	14,306.22	4	3576.55	56.82	<0.0001	significant
$X_1$	1616.18	1	1616.18	25.68	0.0005
$X_2$	3875.95	1	3875.95	61.58	<0.0001
$X_3$	5914.16	1	5914.16	93.96	<0.0001
$X_3^2$	2899.92	1	2899.92	46.07	<0.0001
Residual	629.45	10	62.94
Lack of Fit	593.49	8	74.19	4.13	0.2097	not significant
Pure Error	35.96	2	17.98
Total	14,935.67	14
$R^2$ = 0.9579, adj $R^2$ = 0.9410, CV = 10.51, Adeq Precision 24.73

.

It can be observed that in the case of both models, the linear effects of all variables had a significant impact on the value of the ABTS assay. In the case of extraction in the vessel with a diameter of 25 mm, the square effect of solvent concentration was also significant. For both models, time had the least impact on the ABTS values. In the case of extraction in the vessel with a diameter of 25 mm, the concentration of the solvent had the greatest influence on the ABTS value, and in the case of the vessel with a diameter of 35 mm, the ultrasound amplitude had the greatest impact.

The p-values of the models are statistically significant (p < 0.0015; p < 0.0001), and the lack of fit of the models are statistically insignificant (p = 0.0577 and p = 0.2097), which indicates the correct validation of the equations. In the case of the second model $ABT{S}_2$, the determination coefficient (0.9579) and the adjusted coefficient (0.9410) are high, which shows the existence of a high correlation between the input variables and the ABTS value. In the case of the first equation $ABT{S}_1$, the determination and adjusted coefficients are lower (0.7395 and 0.6685), which means a greater scattering of data around the regression line. The model developed for extraction in the vessel with a diameter of 25 mm ($ABT{S}_2$) is characterised by a much higher correlation between variables.

A regression analysis was employed to develop two models (Equations (7) and (8)) that describe the effect of input variables on the ABTS value:

$$ABT{S}_1=41.776+2.671X_1+2.211X_2-0.450X_3$$

(7)

$$ABT{S}_2=-39.218+2.843X_1+1.834X_2+2.810X_3-0.031X_3^2$$

(8)

The ABTS values range from 18.89 to 134.06 $\mathsf{\mu }$mol TE/g for the 35 mm diameter vessel and from 32.56 to 126.13 $\mathsf{\mu }$mol TE/g for the 25 mm diameter vessel depending on other experimental conditions. The obtained results coincide with the values given by Olszewska et al. [27] (91.6 $\mathsf{\mu }$mol TE/g d.w.). Jabłońska-Ryś et al. [31] showed that the ABTS value for S. aucuparia L. fruits is 5.94 ± 0.06 µM TE/g f.w.

3.5. pH

Figure 5 shows the influence of input variables on the pH value for two extraction vessels with diameters of 35 and 25 mm.

Based on the analysis of the graphs, it can be concluded that the time and amplitude of the ultrasound do not change the pH value. The alcohol concentration changes the acidity of the obtained extracts. After an increase in alcohol concentration, the obtained extracts were characterised by a lower level of acidity (higher pH value). This indicates that changes in the concentration of alcohol in the solvent affect the extraction efficiency of individual polyphenols. This is due to the different polarities of the phenolic compounds. For example, succinic acid found in mountain ash is more soluble in ethanol than in water. In turn, some flavonoids, such as quercetin or rutin, are considered almost insoluble in water; they dissolve better in ethanol [32].

Our assumptions were indirectly confirmed by the results obtained by other authors [33,34]. The effect of ethyl alcohol concentration on the content of individual polyphenols in extracts was observed by Paini et al. [33]. The gallic acid content during a high pressure and temperature extraction of polyphenols from grap marc was 69.57 mg/100 g using 100% ethyl alcohol and 105.25 mg/100 g using water as solvents, respectively. However, for ultrasound-assisted extraction, the gallic acid content was higher when ethyl alcohol was used as a solvent. The effect of the ethyl alcohol concentration on the percentage of individual polyphenols in propolis extracts was also observed by Sun et al. [34]. In extracts obtained using water, 15 phenolic compounds were found, and in extracts obtained using a 75% solution of ethyl alcohol, 28 phenolic compounds were found. In the case of water, 60% of phenolic compounds were phenolic acids, and in the case of the 75% alcohol solution, 80% were flavonoids and phenolic acid esters. A higher content of phenolic acids obtained with the use of an alcohol with a lower concentration will increase the acidity (lower the pH) of these extracts.

In order to confirm the significance of the influence of the tested input variables on the pH value, a statistical analysis was performed. The findings of this examination are displayed in

Table 10. Effect of the independent variables on the dependent variables and statistical significance for $p{H}_1$ of extracts obtained in vessels with a diameter of 35 mm.

Source	Sum of Squares	DF	Mean Square	F-Value	p-Value
Model	3.20	1	3.20	1837.44	<0.0001	significant
$X_3$	3.20	1	3.20	1837.44	<0.0001
Residual	0.0226	13	0.0017
Lack of Fit	0.0218	11	0.0020	4.96	0.1795	not significant
Pure Error	0.0008	2	0.0004
Total	3.22	14
$R^2$ = 0.9930, adj $R^2$ = 0.9924, CV = 0.95, Adeq Precision 83.00

and

Table 11. Effect of the independent variables on the dependent variables and statistical significance for $p{H}_2$ of extracts obtained in vessels with a diameter of 25 mm.

Source	Sum of Squares	DF	Mean Square	F-Value	p-Value
Model	2.70	1	2.70	1602.88	<0.0001	significant
$X_3$	2.70	1	2.70	1602.88	<0.0001
Residual	0.0219	13	0.0017
Lack of Fit	0.0211	11	0.0019	4.42	0.1990	not significant
Pure Error	0.0009	2	0.0004
Total	2.72	14
$R^2$ = 0.9920, adj $R^2$ = 0.9913, CV = 0.94, Adeq Precision 77.53

.

It can be seen that for both models, only the linear effect of the alcohol concentration had a significant effect on the pH value. Both the first and the second equation were validated correctly, which is evidenced by low probability coefficients (p < 0.0001; p < 0.0001) and the lack of fit of the model (p = 0.1795 and p = 0.1990). The high values of the determination coefficient $R^2$ (0.9930 and 0.9920) and the adjusted coefficient $R^2$ (0.9924 and 0.9913) indicate that the scatter of the data around the regression lines is very low. The low CV values (0.95% and 0.94%) suggest minimal deviations between the experimental and predicted values, indicating high experimental reliability and precision.

A regression analysis was employed to develop two models (Equations (9) and (10)) that describe the effect of input variables on the pH value.

$$p{H}_1=3.118+0.021X_3$$

(9)

$$p{H}_2=3.191+0.019X_3$$

(10)

The pH value of the obtained extracts ranges from 3.72 to 5.09 for the vessel with a diameter of 35 mm and up to 4.97 for the vessel with a diameter of 25 mm depending on the other conditions of the experiment. The available literature reports do not contain data on the pH value of rowan fruit extracts.

3.6. Optimisation of the Processing Parameters

A numerical method was applied to compute the maximum value of polyphenols and the antioxidant activity by the DPPH assay value, the FRAP assay value, and the ABTS assay value. The optimal conditions for each response were listed in (

Table 12. Experimental and predicted values of response variables under optimum conditions.

Optimised Condition	Extraction Variables			Yield of Extraction
	${\mathsf{X}}_1$	${\mathsf{X}}_2$	${\mathsf{X}}_3$	Predicted	Experimental	Predictive Capacity (%)
$TP{C}_1$	14.34	25.20	59.74	12.48	10.51	84.27
$TP{C}_2$	14.71	27.28	39.29	11.68	10.16	86.98
$DPP{H}_1$	15.00	36.00	60.00	79.58	68.34	85.87
$DPP{H}_2$	15.00	35.83	55.26	75.90	65.87	86.78
$FRA{P}_1$	11.02	33.07	53.53	120.65	100.45	83.27
$FRA{P}_2$	14.98	35.89	30.05	155.54	120.54	77.49
$ABT{S}_1$	14.20	34.73	66,05	126.54	104.49	82.52
$ABT{S}_2$	11.54	34.55	47.95	120.49	111.12	91.39

).

The highest total content of polyphenols determined on the basis of the developed models was 12.48 mg GAE/g for the vessel with a diameter of 35 mm and 11.68 mg GAE/g for the vessel with a diameter of 25 mm. The optimal extraction conditions for both models were very similar in terms of the ultrasound time and amplitude. The greatest discrepancies were obtained for the concentration of the solvent, amounting to about 34.23%.

The highest antioxidant activity estimated by the DPPH value was 79.58 $\mathsf{\mu }$mol TE/g for model 1 and 75.90 $\mathsf{\mu }$mol TE/g for model 2. The optimum extraction conditions for both models were very similar and amounted to 15 min for the time, 36 $\mathsf{\mu }$m for the ultrasound amplitude, and 60% for the solvent concentration.

The highest antioxidant activity estimated by the FRAP value was 120.65 $\mathsf{\mu }$mol TE/g for the vessel with a diameter of 35 mm and 155.54 $\mathsf{\mu }$mol TE/g for the vessel with a diameter of 25 mm. There were also significant differences regarding the optimal extraction conditions. The largest difference observed in the solvent concentration was over 43.86%.

The highest antioxidant activity estimated by the ABTS value was 126.54 $\mathsf{\mu }$mol TE/g for the vessel with a diameter of 35 mm and 120.49 $\mathsf{\mu }$mol TE/g for the vessel with a diameter of 25 mm. The optimal extraction conditions were similar for the ultrasound amplitude and different for time and solvent concentration. The difference between the optimal solvent concentration for the tested vessel was 27.40%.

Based on the calculated optimal parameters, experimental values were determined for each of the responses. The predictive capacity of the models was calculated as the ratio of the experimental values to the predicted values. For most models, the value of this indicator ranged from 80 to 90%, which is considered a good level. The lowest predictive capacity value was obtained for the FRAP${}_2$ model, which was 77.49%, and the highest was for the ABTS${}_2$ model, which was 91.39%. A better correspondence between the predicted and experimental responses can be partly explained by the number of parameters present in the analysed models.

An analysis of the results showed a significant influence of the diameter of the extraction vessel on the course of the extraction process. For the vessel with a diameter of 35 mm, there is a faster increase in the efficiency and antioxidant activity of the extracts at lower levels of ultrasound amplitude compared to the vessel with a diameter of 25 mm. This phenomenon can be explained by the distance between the ultrasonic probe and the bottom of the vessel. For the reactor with a diameter of 35 mm, this distance is 20 mm, and for the reactor with a diameter of 25 mm, this distance is 75 mm. As a result, the flux of cavitation bubbles in the case of the first vessel is closer to the extracted material, which results in more intensive disintegration of solid particles and higher kinetics of the polyphenol extraction. Further increases in the ultrasound amplitude cause only a slight increase in the concentration of the extracted substances due to its decreasing concentration in the raw material. In the case of the vessel with a diameter of 25 mm, the distance between the ultrasonic probe and the extraction bed is much greater, which results in a weaker impact of the cavitation flux and thus slower extraction kinetics. Increasing the ultrasound amplitude causes a higher cavitation flux and accelerates the extraction kinetics. As a result, the final values of polyphenol yield at the maximum time do not differ much between the individual diameters of extraction vessels.

Similar observations regarding the influence of the position of the ultrasonic probe on the extraction efficiency were made by Sun et al. [17]. The all-trans-$\beta$-carotene extraction efficiencies with a 2 cm and 12 cm distance between the ultrasonic probe and the raw material were, respectively, 84.5% and 12.3% higher than the extraction efficiency obtained during maceration.

Kulkarni et al. [18] analysed the effect of the vessel diameter and height on the ultrasonic extraction efficiency of mangiferin from Mangifera indica leaves. The experiment was carried out for four different vessel depths (inside the ultrasonic bath), i.e., 1.016, 2.54, 5.08, and 7.62 cm. The highest extraction efficiency was obtained at a depth of 2.54 cm. The next stage of the experiment was carried out for different vessel diameters: 1.3, 1.5, 2.5, 3, 3.5, 5, and 7 cm. The authors showed that with the increase in the diameter of the extraction vessel, the extraction efficiency increased up to a certain limit, after which it remained constant. In vessels with a diameter of 1.3 cm, the yield of mangiferin was 26 mg/g, while in vessels with diameters of 3.5, 5, and 7 cm, the yield was approximately 31 mg/g [18].

A very interesting finding in this work is also a different effect of alcohol concentration on the course of the extraction process for individual extraction vessels. In the case of the vessel with the larger diameter (35 mm), an initial increase in the extraction efficiency occurs with increasing alcohol concentration in the solvent, and then, after exceeding the concentration of the solvent, the extraction efficiency decreases by about 60%. In the case of the extraction vessel with the smaller diameter (25 mm), the highest values of efficiency and antioxidant activity of the extracts were observed in the range of 30 to 50% alcohol, and further increasing the alcohol concentration resulted in a rapid decrease in the measured values. This suggests that the increase in alcohol concentration with increasing distance from the extraction bed contributes to a decrease in the extraction efficiency.

Our observations indicate that the cavitation flux has a smaller range in ethyl alcohol than in water, so the greater the distance of the ultrasonic probe from the bottom of the extraction vessel, the more limited the impact of the cavitation flux on the raw material, and thus the extraction is less effective.

For both solvents, an increase in the diameter of the cavitation bubbles and the length of the cavitation flux is visible with the growth in the ultrasound amplitude, which partly explains the growth in extraction efficiency with the growth in ultrasound amplitude.

An important factor that may affect the extraction efficiency in individual extraction vessels is also the volume of solvent remaining directly under the ultrasonic probe. The volume of solvent below the probe for the 25 mm diameter vessel is approximately 36,800 mm${}^3$ and for the 35 mm diameter vessel it is 19,200 mm${}^3$.

Differences in solvent volume result in differences in the energy density of the ultrasonic field in individual extraction vessels. The higher ultrasonic energy density in the 35 mm diameter vessel contributes to higher polyphenol extraction yields.

Interactions between the diameter of the extraction vessel and the concentration of alcohol in the solvent may result not only in a change in the efficiency of the extraction process but also affect the percentage of individual phenolic compounds in the extracts.

3.7. Conclusions

In this study, the effects of pulsed ultrasound-assisted extraction parameters on the chemical properties of the extract from rowan were investigated applying the response surface method.

A statistical analysis showed that the shape of the extraction vessels has a significant impact on the course of the extraction process. It was shown that the optimal conditions for obtaining the maximum yield and antioxidant activity of the obtained extracts significantly differ depending on the diameter of the extraction vessel used. The greatest differences were observed for the solvent concentration variable. In the case of the extraction vessel with a diameter of 35 mm, the optimal extraction conditions occur at an ethyl alcohol concentration of 60%. For the extraction vessel with a diameter of 25 mm, the optimal extraction conditions are in the range of 30–50% alcohol, depending on the type of output variable analysed.

A very strong influence of the alcohol concentration on the pH of the obtained extracts was also demonstrated. An increase in the concentration of alcohol in the solvent caused a decrease in the acidity of the extracts (higher pH values), which indicates that alcohol more efficiently extracts substances with a lower acidity (higher pH values). This fact, combined with the optimal solvent concentration values for individual extraction vessels, suggests that the obtained extracts, despite the similar content of total polyphenols, differ in the percentage share of individual phenolic substances.

The statistical analysis proved that all ten developed models were accurate and exhibited a good fit to the empirical data with a predictive capacity in the range of 77.49 to 91.39%.