Optimization of the Ultrasound-Assisted Extraction of Bioactive Compounds from Cannabis sativa L. Leaves and Inflorescences Using Response Surface Methodology

Abstract

This study investigated the effects of particle size and ultrasonic parameters on the yields of bioactive compounds extracted from the leaves and inflorescences of hemp. The total flavonoid and anthocyanin contents were determined using the spectrophotometric method. The response surface methodology (RMS) was employed to optimize the yield of bioactive substances. On the basis of the developed model, the highest flavonoid yield was obtained under the following extraction conditions: particle size, 0.59 mm; extraction time, 10.71 min; ultrasound intensity, 7.13 W∙cm⁻²; extraction yield, 9.28 mg QE∙g⁻¹; determination coefficient, R² = 0.97. The optimal conditions for extracting anthocyanins were as follows: particle size, 0.25 mm; extraction time, 15 min; ultrasound intensity, 8.60 W∙cm⁻²; extraction efficiency, 20.27 mg Cy-GE∙100 g⁻¹; determination coefficient, R² = 0.87. This study helped confirm the importance of pulsed ultrasound-assisted extraction in obtaining bioactive compounds from hemp.

1. Introduction

Hemp (Cannabis sativa L.) is a herbaceous, wind-pollinated plant that is widespread around the world and belongs to the Cannabaceae family. The species Cannabis sativa L. (true hemp) includes fiber (Cannabis sativa L. var. sativa) and Indian (narcotic) (Cannabis sativa L. var. indica) hemp. These differ in their levels of cannabinoids, mainly THC, the content of which in fiber hemp is below 0.2% [1,2,3,4].

Poland has a long tradition of growing cannabis. Hemp is cultivated in areas with valuable natural resources, including the Lublin region, Podlasie, and Greater Poland. The Lublin region currently ranks fourth in terms of its cultivation area, but it is first in terms of the number of farms.

C. saliva L. has recently received a lot of attention for its nutritional and pharmaceutical value, although in the past, it was grown mainly for the fibers from hemp stalk and for the oil from hemp seeds [5,6,7]. The medicinal use of cannabis has been known for over 5000 years, and the pharmacological properties of cannabinoids, dominant in cannabis, can be useful in the treatment of various diseases [7,8]. Indeed, cannabinoids (phytocannabinoids) are one of the most important bioactive compounds in cannabis [9]. The best-known cannabinoids found in Cannabis sativa L. are THC and CBD.

The panicle elements of the plant (inflorescences), along with the leaves, are used in the production of industrial hemp extracts. They contain volatile terpenes and phytocannabinoids, including cannabidiol, better known by the trade name CBD. The latter is usually presented to consumers in the form of dietary supplements, cosmetics, and even drugs, although its production in Poland is currently prohibited. The popularity of cannabidiol is likely due to its pro-health and healing properties, some of which have been scientifically proven and some of which are described within the so-called natural medicine movement [10]. CBD oils, ointments, and capsules have a beneficial effect on general health and can be used in the symptomatic treatment of eczema and other skin diseases, as well as various types of pain, especially rheumatoid arthritis, joint diseases, cancer, and even multiple sclerosis. The health properties of CBD also factor into the popularity of daily skin care cosmetics that have been derived from it, the popularity of which increases every year.

Hemp is also rich in natural antioxidants and other bioactive ingredients, such as bioactive peptides, phenolic compounds, tocopherols, carotenoids, and phytosterols. The content of these ingredients is mostly influenced by environmental and agronomic factors and, to a lesser extent, by genetic variability. Fibrous hemp inflorescences are a source of polyphenolic compounds with proven health-promoting properties [11]. Bioactive substances are biologically active compounds of a natural origin that can have a beneficial and multidirectional effect on the body. As food ingredients, they can modify, strengthen, or weaken various body functions, thus limiting the development of disease processes. These compounds are characterized by antioxidant, anti-inflammatory, neuroprotective, antihypertensive, antiproliferative, and hypocholesterolemic effects, which have mainly been assessed with in vitro studies [2,8,12,13,14,15].

In recent years, the demand for bioactive substances has increased, which has also resulted in a search for new, more efficient methods for their extraction. To this end, various separation techniques are used, such as supercritical fluid extraction, microwave-assisted extraction, accelerated solvent extraction, ultrasound-assisted extraction, and pulsed electric-field-assisted extraction. The ultrasound-assisted extraction method enables the extraction of bioactive ingredients in a very short time, at a low temperature, and with lower energy and solvent requirements [16]. As a non-thermal extraction technique, it better preserves the functionality of bioactive compounds; however, process variables, such as frequency, power, duty cycle, temperature, time, type of solvent, and liquid–solid ratio, must be individually selected for each raw material [17]. The advantage of ultrasonic treatment is also the inactivation of microorganisms and enzymes, which extends the shelf life of the obtained products [18].

Most of the work on cannabis is devoted to ultrasound-assisted continuous extraction, which, as noted, enables the extraction of bioactive ingredients in a very short time, at a low temperature, and with lower energy and solvent requirements. Flores-Sanchez and Verpoorte [19], as well as Choi et al. [20], conducted continuous ultrasound-assisted extractions (for 10 min) in order to obtain cannabinoids and flavonoids from cannabis. Nagy et al. [8] performed an ultrasonically assisted extraction (for 10 min) on spontaneous C. sativa, demonstrating the presence of several flavonoid derivatives. The total flavonoid amounts in the leaves and the male and female inflorescences were 3.84, 6.09, and 7.79 mg∙g⁻¹, respectively.

Currently, the pulsed ultrasound field method is used more and more often in order to support the process of extraction of bioactive compounds [21]. The advantages of this solution are comparable or higher extraction yields, slower temperature increases during the extraction process, and significantly lower energy consumption [22,23]. Thus far, no research has been conducted on the effect of pulsed ultrasound-assisted extraction with respect to obtaining bioactive substances from cannabis inflorescences. The aim of this study was to determine the optimal conditions for the ultrasound-assisted extraction of bioactive compounds from Cannabis sativa L. with the help of the Box–Behnken experimental design.

2. Materials and Methods

2.1. Raw Material and Regents

Dried true hemp was obtained from an organic farm located in Biała Podlaska County, Lubelskie Voivodeship. The material was crushed in the Zelmer MM 1200 device, then sieved and divided into fractions. For further research, 3 fractions with the following mean particle weights were used: 0.25, 0.75, and 1.25 mm.

Potassium chloride, sodium acetate and hydrochloric acid (Stanlab, Lublin, Poland) were used for the analysis of anthocyanins. For determining flavonoids, aluminum chloride and quercetin (Sigma-Aldrich—Merck, Taufkirchen, Germany) were used. All reagents used for analytical procedures were of analytical grade.

2.2. Methods

The research was carried out according to the scheme presented in Figure 1.

2.2.1. Ultrasound-Assisted Extraction

Raw material in the amount of 1.5 g from each fraction was placed into a conical flask, and an aqueous solution of ethyl alcohol with a concentration of 60% was poured over it. The flask was sealed with a 25.4 mm (1 inch) diameter ultrasonic probe, then placed in the water bath to stabilize the temperature (30 °C). The experimental samples were sonicated with a VC750 Sonics processor (Sonics and Materials Inc., Newtown, CT, USA) operating at a frequency of 20 kHz. The sonication was performed at three amplitudes—30%, 50%, and 70%—corresponding to the ultrasound intensity—1.6, 5.1, and 8.6 W∙cm⁻², respectively. The samples were sonicated in the following processor arrangement: 5 s on–10 s off. The effective operation times were 5, 10, and 15 min, and the total extraction times were 15, 30, and 45 min, respectively. The extracts obtained in this way were stored in a refrigerator (2 °C) and collected for further chemical analysis.

2.2.2. The Total Flavonoid Content (TFC)

TFC was determined with spectrophotometry using quercetin as the reference standard. First, the sample extract (1.0 mL) was mixed with 1 mL of a 2% AlCl₃ × 6H₂O solution (in methanol), and the volume of the mixture was made up to 10 ml with distilled water. After incubating the mixture for 10 min at room temperature in the dark, absorbance was measured at 430 nm. The results were calculated as mg quercetin equivalent per 1 g dry weight (mg QE∙g⁻¹ dw).

2.2.3. The Total Anthocyanin Content (TAC)

The TAC was determined with the spectrophotometric method from absorbance measurements taken at different pH levels. First, the sample extract (0.5 mL) was mixed with 4 mL of potassium chloride and sodium acetate buffers at pH 1.0 and 4.5. After leaving the mixture for 15 min at room temperature, the absorbance was measured at 520 and 700 nm. The correct absorbance was determined from Formula (1):

$$\mathrm{A}={\left({\mathrm{A}}_{520}-{\mathrm{A}}_{700}\right)}_{\mathrm{pH}1.0}-{\left({\mathrm{A}}_{520}-{\mathrm{A}}_{700}\right)}_{\mathrm{pH}4.5}$$

(1)

The total anthocyanin content was expressed as cyanidin 3-glucoside equivalent (Cy-GE) in mg/g dry weight using Formula (2):

$$\mathrm{TAC}=\frac{\mathrm{A}}{\mathrm{L}\mathsf{\epsilon }}{\mathrm{M}}_{\mathrm{w}}·\mathrm{N}$$

(2)

A—correct absorbance;

L—cuvette thickness;

N—dilution factor;

M_w—molar mass of cyanidin 3-glucoside = 26,900;

ε—molecular absorbance of cyanidin 3-glucoside = 449.2.

2.3. Box–Behnken Experimental Design and Statistical Analyses

The experiment was performed on the basis of the Box–Behnken experimental design in the Design-Expert v13, Stat-Ease, Minneapolis, MI, USA). A three-level, three-factor BBD plan was used to determine the best combination of cannabis inflorescence extraction variables. Particle size (X₁), extraction time (X₂), and ultrasound intensity (X₃) were independent variables, whereas the dependent variables were the total flavonoid content and the total anthocyanin content.

The experiment comprised a total of 15 combinations, including 3 center points to estimate the pure error, and was carried out in randomized order. All experiments were performed in triplicate. The levels of the three factors evaluated in the design are listed in

Table 1. The Box–Behnken response surface design.

Run	X1	X2	X3
1.	0.75	5	8.6
2.	1.25	10	1.6
3.	0.25	10	8.6
4.	0.25	10	1.6
5.	0.75	10	5.1
6.	0.25	5	5.1
7.	0.75	10	5.1
8.	0.75	10	5.1
9.	0.25	15	5.1
10.	0.75	5	1.6
11.	0.75	15	1.6
12.	1.25	10	8.6
13.	1.25	10	5.1
14.	1.25	5	5.1
15.	0.75	15	8.6

.

A generalized, second-order polynomial model was used to explain the effect of the independent variables on each response of interest according to the following equation:

$$\mathrm{Y}={ \mathsf{\beta }}_0+{\mathsf{\beta }}_{1 }{\mathrm{X}}_1+{\mathsf{\beta }}_2{\mathrm{X}}_2+{\mathsf{\beta }}_3{\mathrm{X}}_3+{\mathsf{\beta }}_{11}{\mathrm{X}}_1^2+{\mathsf{\beta }}_{22}{\mathrm{X}}_2^2+{\mathsf{\beta }}_{33}{\mathrm{X}}_3^2+{\mathsf{\beta }}_{12}{\mathrm{X}}_1{\mathrm{X}}_2+{\mathsf{\beta }}_{13}{\mathrm{X}}_1{\mathrm{X}}_3+{\mathsf{\beta }}_{23}{\mathrm{X}}_2{\mathrm{X}}_3$$

(3)

where Y is the response variables (the total flavonoid content or the total anthocyanin content); X₁, X₂, and X₃ are the independent variables; β₀ represents the constant; and β_1,2,3. Β_11,22,33, and β_12,13,23 are the linear, quadratic, and interactive coefficients, respectively.

The experimental data were assessed via analysis of variance (ANOVA). The statistical significances of the regression coefficients were checked with an F-test, and p-values less than 0.05 were considered significant.

The optimal extraction conditions were estimated through Derringer’s desirability prediction tool, aiming at a maximum attainable response for each independent factor. The validity of the developed model was assessed by comparing the experimental values and the predicted values. Two additional independent experiments were conducted using the optimal conditions estimated with the models for each dependent variable (separately), as well as one experiment for both variables.

3. Results and Discussion

3.1. Total Flavonoid Content

The total flavonoid content extracted from the hemp ranged from 3.02 to 9.35 mg QE ∙g⁻¹, varying according to the experimental conditions (Figure 2). Similar results for the total flavonoid content, ranging from 1.83 to 11.20 mg QE∙g⁻¹ dw, were obtained in aerial studies on hemp parts from the Cannabis sativa L. variety, as carried out by Drinić et al. [15]. In the extracts obtained with 50% ethanol, the highest flavonoid content was 11.20 mg QE∙g⁻¹ dw for young plants and 5.21 mg QE∙g⁻¹ dw mature plants. Studies conducted by Izzo et al. [24] showed that the average flavonoid content in the inflorescences of cannabis cv. Carmagnola was about 0.62 mg∙g⁻¹ for samples with moisture content ranging from 8% to 12%.

Based on the obtained results, the effects of the independent variables on the dependent variables are presented in

Table 2. Effects of the independent variables on the dependent variables and their statistical significance with respect to the extraction of flavonoid compounds from hemp.

Variables	p Value	Coefficient
X1	0.0054	−1.15
X2	0.0153	+0.8872
X3	0.0016	+1.51
X1X2	0.4980	−0.2534
X1X3	0.5001	−0.2521
X2X3	0.0202	−1.17
${\mathrm{X}}_1^2$	0.0002	−3.58
${\mathrm{X}}_2^2$	0.0457	−0.9557
${\mathrm{X}}_3^2$	0.0445	−0.8214

.

It was observed that the linear and quadratic terms of the particle size, extraction time, and ultrasound intensity significantly affected (p < 0.05) the extraction of flavonoid compounds. However, the linear and quadratic term of the particle size, and the quadratic terms of the extraction time and ultrasound intensity demonstrated negative correlations, indicating that an increase in the magnitude of these variables may favor the extraction of flavonoid compounds only up to a certain value. This effect is very visible in the case of the disintegration degree, as the total flavonoid content initially increased with the rising particle size, from 0.25 to 0.75 mm, and then decreased as the particle size increased from 0.75 to 1.25 mm. The degree of fragmentation was the variable that most significantly influenced the extraction of flavonoid compounds.

In general, the yield of flavonoids increased with an increased time of extraction. Above 14 min, a slight decrease in the total flavonoid content was visible, so a further extension of the extraction time may reduce the extraction of the target compound. Being exposed to ultrasound for too long causes structural damage in the solute and reduces the extraction efficiency, which was confirmed during the extraction of phenolic compounds from waste coffee grounds [25], phenolic compounds from black chokeberry waste [26], and flavonoids from hawthorn seeds [27]. The extraction time had the least influence on the total flavonoid content in the obtained extracts. The total flavonoid content increased with the intensity of the ultrasound. The growth in the intensity of the ultrasound from 1.6 to 8.6 W∙m⁻² increased the total flavonoid content by 47.7%. However, applying ultrasound intensity above the test range may reduce the extraction yield of the target compound. The different effects of ultrasound power on the content of phenolic compounds were observed by Al-Dhabi et al. [25]. They demonstrated an increase in the efficiency of phenolic compound extraction from coffee-ground waste when the ultrasound power increased from 100 to 244 W, as well as a decrease in efficiency when the ultrasound power exceeded 250 W. In the work of Al-Dhabi et al. [25], a statistically significant negative interaction was found between the extraction time and the intensity of the ultrasound, which means that the effect of the combined action of the two predictors is less than the sum of the individual effects.

The parameters presented in Table 2 were re-estimated considering only the significant terms (p < 0.05). From the regression analysis, the model was adjusted to the experimental data, as presented in Equation (4):

$$\mathrm{TFC}=-10.61+19.19{\mathrm{X}}_1+1.28{\mathrm{X}}_2+1.78{\mathrm{X}}_3-0.07{\mathrm{X}}_2{\mathrm{X}}_3-14.32{\mathrm{X}}_1^2-0.04{\mathrm{X}}_2^2-0.07{\mathrm{X}}_3^2$$

(4)

The predictive equation was verified with analysis of variance (ANOVA), as can be seen in

Table 3. ANOVA results for extraction efficiency and the total flavonoid content.

Source	Sum of Squares	DF	Mean Square	F-Value	p-Value
Yield (%)
Model	90.57	7	12.94	31.01	<0.0001	significant
X1	10.55	1	10.55	25.28	0.0015
X2	6.30	1	6.30	15.09	0.0060
X3	18.31	1	18.31	43.89	0.0003
X2X3	5.43	1	5.43	13.02	0.0086
${\mathrm{X}}_1^2$	47.34	1	47.34	113.48	<0.0001
${\mathrm{X}}_2^2$	3.37	1	3.37	8.08	0.0249
${\mathrm{X}}_3^2$	2.49	1	2.49	5.97	0.0445
Residual	2.92	7	0.4172
Lack of Fit	2.65	5	0.5298	3.91	0.2162	not significant
Pure Error	0.2711	2	0.1355
Total	93.49	14
R2 = 0.9688; adj-R2 = 0.9375; CV = 11.13; Adeq Precision = 15.004

, and it provided a satisfactory fit to the experimental data.

The model is statistically significant (p < 0.0001), and the lack of model fit is statistically insignificant (p > 0.2162), which indicates that the model has been validated correctly. The high value of the R² coefficient (0.9688) and the corrected R² (0.9375) indicates the existence of a large correlation between the input variables and the total flavonoid content. A low CV value (11.13%) means that the deviations between the experimental and predicted values are low, and the reliability of the experiment and its precision is high. Adequate precision greater than four is desirable, and the ratio was found to be 15.004, which indicates an adequate signal and confirms that this model is significant for this extraction process.

3.2. The Total Anthocyanin Content

The results of the extraction of anthocyanin using an ultrasonic extractor are presented in Figure 3.

The total anthocyanin content extracted from the hemp ranged from 10.66 to 17.16 mg Cy-GE∙100 g⁻¹, varying according to the experimental conditions. None of the available reports have analyzed TAC in cannabis. Based on these results, the effects of the independent variables on the dependent variables are presented in

Table 4. Effects of the independent variables on the dependent variables and their statistical significance with respect to the extraction of anthocyanins from hemp.

Variables	p Value	Coefficient
X1	0.0141	−1.39
X2	0.0060	+1.72
X3	0.0003	+3.25
X1X2	0.6191	+0.2818
X1X3	0.4789	−0.4070
X2X3	0.0700	+1.22
${\mathrm{X}}_1^2$	0.7883	−0.1570
${\mathrm{X}}_2^2$	0.0901	−1.16
${\mathrm{X}}_3^2$	0.7702	+0.1709

.

It was observed that only the linear terms of the particle size, extraction time, and ultrasound intensity significantly affected (p < 0.05) the extraction of anthocyanins. However, the linear term of the particle size demonstrated negative correlations. As the particle size increased from 0.25 to 1.25, there was a clear decrease in the total anthocyanin content in the obtained extracts, from 15.30 to 12.52 mg Cy-GE∙100 g⁻¹ (Figure 3). This is most likely due to the area in which anthocyanins are stored in the plant material. Anthocyanins occur in plant vacuoles [28]; therefore, a greater degree of fragmentation facilitates their washing out and contributes to a higher extraction efficiency. Wang et al. [29] showed that a material particle size of 0.9 mm was sufficient to achieve an optimal pectin yield. The smaller particle sizes of raw materials with larger specific surface areas have a tendency to increase the heat and mass transfer between solvents and matrices [29]. Particle size had the least influence on the total anthocyanin content in the obtained extracts, and its increase caused the anthocyanin value to drop by 22.2%.

The ultrasound intensity was the variable that most significantly influenced the extraction of anthocyanins. The total anthocyanin content increased with the growth of the ultrasound intensity. An incremental increase in the ultrasound intensity, from 1.6 to 8.6 W∙m⁻², raised the total anthocyanin content by 61%. As the extraction time rose from 5 to 15 min, the total anthocyanin content increased by 28.2%. The linear term of the extraction time has also been observed during the extraction of anthocyanins and phenolic compounds from jabuticaba skin [30], as well as phenolic compounds from grape marc [31] and grape seeds [32]. In a previous experiment on the extraction of anthocyanins from hawthorn berries, we showed a growth in TAC with a commensurate increase in time when using the pulse mode of the ultrasound [22]. However, in the case of the continuous mode, a slight but statistically insignificant decrease in the total anthocyanin content was observed at a time of 15 min and an amplitude of 36 µm [22]. Zou et al. [33] indicated that the anthocyanin yield of mulberry quickly increased with the time of extraction, reaching the highest value at 40 min. From 40 to 100 min, the yield was almost constant. Mane et al. [34] investigated the effect of ultrasound-assisted extraction on the amount of anthocyanin extracted from Purple Majesty potatoes and pointed out that shorter times lead to the growth of anthocyanins in the extracts, with 5 min being optimal. Longer extraction times have shown a linear decrease in TAC obtained over a period of 120 min [34]. Tiwari et al. [35] showed that higher levels of ultrasonic amplitude and time have an adverse effect on the total anthocyanin content in grape juice.

The parameters presented in Table 4 were re-estimated, considering only the significant terms (p < 0.05). From the regression analysis, the model was adjusted to the experimental data, as presented in Equation (5):

$$\mathrm{TAC}=7.82-2.78{\mathrm{X}}_1+0.34{\mathrm{X}}_2+0.93{\mathrm{X}}_3$$

(5)

The predictive equation was verified with analysis of variance (ANOVA), as can be seen in

Table 5. ANOVA results for extraction efficiency and total anthocyanin content.

Source	Sum of Squares	DF	Mean Square	F-Value	p-Value
Yield (%)
Model	123.55	3	41.18	25.33	<0.0001	significant
X1	15.47	1	15.47	9.52	0.0104
X2	23.68	1	23.68	14.56	0.0029
X3	84.40	1	84.40	51.90	<0.0001
Residual	17.89	11	1.63
Lack of Fit	17.26	9	1.92	6.15	0.1476	not significant
Pure Error	0.6236	2	0.3118
Total	141.44	14
R2 = 0.8735; adj-R2 = 0.8390; CV = 9.17; Adeq Precision = 15.09

, and it provides a satisfactory fit to the experimental data.

The model is statistically significant (p < 0.0001), and the lack of model fit is statistically insignificant (p > 0.1476), which indicates that the model has been validated correctly. High values of the R² coefficient (0.8735) and corrected R² (0.8390) indicate the existence of a large correlation between the input variables and the total anthocyanin content. A low CV value (9.17%) means that the deviations between the experimental and predicted values are low, and the reliability of the experiment and its precision is high. Adequate precision was found to be 15.09, which indicates an adequate signal and confirms that this model is significant for this extraction process.

3.3. Optimization of the Extraction Conditions

Derringer’s desirability prediction tool was used to calculate the maximum yield of flavonoids and anthocyanins. Based on this analysis, the optimal values of extraction for individual components were obtained (

Table 6. Comparison between the experimental yields and predicted yields of total flavonoid content (TFC), total anthocyanin content (TAC), and the simultaneous extraction of both bioactive compounds (TFC and TAC) determined in the optimized conditions.

Optimized Condition	Extraction Variables			Response	Yield of Extraction
	X1	X2	X3		Predicted	Experimental a	Predictive Capacity (%)
TFC	0.59	10.71	7.13	TFC	9.28	8.98 ± 0.21	96.77 ± 2.27
TAC	0.25	15	8.60	TAC	20.27	17.83 ± 1.1	87.96 ± 5.43
TFC and TAC	0.60	12.51	8.60	TFC	8.99	8.52 ± 0.2	94.77 ± 2.22
				TAC	18.43	16.74 ± 0.98	90.83 ± 5.32

X₁—particle size (mm); X₂—extraction time (min); X₃—ultrasound intensity (W∙cm⁻²); ^a—mean ± standard deviation of the three experiments.

).

The predictive capacity of the models was evaluated by comparing the predicted and experimental values that were obtained from the tests, applying the optimized conditions for each response. In all cases, the extraction yield of bioactive substances was slightly lower than the values calculated on the basis of the developed models. Better agreement between the predicted and experimental responses was obtained for the flavonoids, possibly due to the greater number of parameters included in the model describing flavonoid extraction. However, considering the high complexity of the matrix, the proposed models showed a satisfactory predictive capacity for all evaluated compounds.

3.4. Energy Consumption during Pulsed Ultrasound-Assisted Extraction

Energy consumption during sonication is mainly dependent on the time and amplitude of the ultrasonic vibrations and, to a lesser extent, on factors related to the extraction environment, such as the viscosity and temperature of the solvent and the size and concentration of the immersed solids. Some of these parameters can be changed during treatment. Because of this, it is important to determine the total quantity of energy emitted into the solid–liquid system [21]. The energy emitted by the ultrasonic device during extraction is presented in

Table 7. The energy emitted by the ultrasonic device during extraction.

Ultrasound Intensity (W∙cm−2)	Time (min)	Energy Consumption (kJ)
1.6	5	2720
1.6	10	4843
1.6	15	7041
5.1	5	7631
5.1	10	15,410
5.1	15	21,821
8.6	5	14,011
8.6	10	22,945
8.6	15	32,012

.

The analysis of the data contained in Table 7 shows a directly proportional relationship between the time of ultrasonic treatment and energy consumption; thus, it can be assumed that, during sonication, the physical properties of the solvent were constant.

4. Conclusions

This study investigated the effects of extraction conditions on the yield of flavonoids and anthocyanins using the Box–Behnken response surface methodology. The influence of process variables (particle size, extraction time, and ultrasound intensity) on the extraction efficiency depended on the tested bioactive substance.

For flavonoids, in the entire tested range, a significant positive effect was found with respect to extraction time and ultrasound intensity on the efficiency of the process, whereas in the case of particle size, the highest efficiency was obtained for a particle size of 0.75 mm. The optimal conditions (on the basis of the model) for the extraction of flavonoids from cannabis were as follows: particle size, 0.59 mm; extraction time, 10.71 min; and ultrasound intensity, 7.13 W∙cm⁻².

For anthocyanins, we observed a negative influence from the particle size distribution on the extraction efficiency and a positive effect from the ultrasound intensity and time. The optimal conditions (on the basis of the model) for the extraction of anthocyanins were as follows: particle size, 0.25 mm; extraction time, 15 min; and ultrasound intensity, 8.60 W∙cm⁻².

The statistical data showed that the developed models were precise and adequate compared with the experimental data. For both models (anthocyanins and flavonoids), high values for the coefficient of determination (0.87–0.97) and the corrected coefficient of determination (0.84–0.94) were obtained. The developed extraction procedure, with the application of a pulsed ultrasound field, proved to be efficient for obtaining an enriched fraction of bioactive compounds with very high flavonoid and anthocyanin contents from Cannabis sativa L.