molecules
Article Supercritical Carbon Dioxide (scCO2) Extraction of Phenolic Compounds from Lavender (Lavandula angustifolia) Flowers: A Box-Behnken Experimental Optimization
Katarzyna Ty´skiewicz* , Marcin Konkol and Edward Rój Supercritical Extraction Department, Sie´cBadawcza Łukasiewicz - Instytut Nowych Syntez Chemicznych, Al. Tysi ˛acleciaPa´nstwaPolskiego 13A, 24-110 Puławy, Poland; [email protected] (M.K.); [email protected] (E.R.) * Correspondence: [email protected]; Tel.: +48-814-731-452
Academic Editors: Robert Shellie and Francesco Cacciola Received: 17 August 2019; Accepted: 13 September 2019; Published: 15 September 2019 Abstract: Due to their numerous health benefits associated with various diseases and anti-oxidation properties, the phenolic compounds collectively referred to as phytochemicals have attracted a lot of interest, however, a single extraction method for polyphenols has not been developed yet. Supercritical fluid extraction, a green extraction method, provides the final product without organic solvent residues. In this work the extraction of lavender was performed using supercritical carbon dioxide. A statistical experimental design based on the Box-Behnken (B-B) method was planned, and the extraction yields and total phenolic contents were measured for three different variables: pressure, temperature and extraction time. The ranges were 200–300 bar, 40–60 ◦C and 15–45 min. The extracts yields from scCO2 extraction were in the range of 4.3–9.2 wt.%. The highest yield (9.2 wt.%) was achieved at a temperature of 60 ◦C under the pressure of 250 bar after 45 min. It also corresponded to the highest total phenolic content (10.17 mg GAE/g extract). Based on the study, the statistically generated optimal extraction conditions to obtain the highest total phenolic compounds concentration from flowers of Lavandula angustifolia were a temperature of 54.5 ◦C, pressure of 297.9 bar, and the time of 45 min. Based on the scavenging activity percentage (AA%) of scCO2 extracts, it is concluded that the increase of extraction pressure had a positive influence on the increase of AA% values.
Keywords: lavender; optimization; phenolic compounds; supercritical fluid extraction; total phenolic content
1. Introduction Lavender (Lavandula angustifolia) is a perennial plant belonging to the family of Labiatae, widely used for essential oil production in Europe, Northern Africa, the Middle East and Asia. Due to its properties, lavender is commonly used by various industries, for instance, as a popular aromatic shrub for perfumes. The beneficial effects of lavender are attributed to the presence of a broad range of biologically active compounds (mainly phenolic and aromatic compounds) with antioxidant, antifungal and anti-carcinogenic activities. Besides this, lavender has been proven to exhibit antimicrobial, anti-inflammatory, antidepressive and other properties. Moreover, lavender has insecticidal activity attributed to the presence of aromatic and volatile compounds, including terpenes, linalool, pinenes, etc. [1,2]. According to Sytar et al. [3], the exact antioxidant properties of lavender are still unknown or only known to some extent. However, different studies are still being performed on lavender, especially in terms of its phenolic compounds content [4–6]. Figure1 shows the increasing tendency in the interest
Molecules 2019, 24, 3354; doi:10.3390/molecules24183354 www.mdpi.com/journal/molecules Molecules 2019, 24, x FOR PEER REVIEW 2 of 15
MoleculesAccording2019, 24, 3354 to Sytar et al. [3], the exact antioxidant properties of lavender are still unknown2 of or 15 only known to some extent. However, different studies are still being performed on lavender, especially in terms of its phenolic compounds content [4–6]. Figure 1 shows the increasing tendency inof Lavandulathe interest angustifolia of Lavandulabased angustifolia on bibliographic based search on bibliographic results with thesearch phrase results “Lavandula with angustifoliathe phrase” “andLavandula “Lavender” angustifolia in general.” and “Lavender” in general.
Figure 1.1.Evolution Evolution in in the the number number of papers of papers with the with keyword the keyword “Lavender” “Lavender” and “Lavandula and angustifolia“Lavandula” (Scienceangustifolia Direct,” (Science April Direct, 2019). April 2019).
Supercritical fluidfluid extraction (SFE) is a separation technique, the efficiency efficiency of which depends on several aspectsaspects such such as theas mobilethe mobile phase phase nature (purenature or (pure modified), or modified), the process the parameters process (temperature, parameters pressure(temperature, and time),pressure and and the typetime), of and raw the material, type of as raw well material, as its pretreatment as well as [ 7its]. Researcherspretreatment must [7]. Researcherschoose and optimizemust choose the extraction and optimize parameters the extraction properly parameters when it comes properly to the extractionwhen it comes of particular to the extractiongroups of compounds.of particular groups However, of compounds. in the case of However, lavender, in supercritical the case of lavender, fluid extraction supercritical is the mostfluid extractioncommon method is the most utilized common for extraction method utilized of the essential for extraction oil, which of the is essential a complex oil, matrix which consistingis a complex of matrixover one consisting hundred compoundsof over one [3 ,7hu–11ndred]. Jerkovi´cet compounds al. [ 11[3,7–11].] optimized Jerkovi supercriticalć et al. fluid[11] optimized extraction supercriticalof Lavandula angustifoliafluid extractionflowers of Lavandula in terms of angustifolia the content flowers of oxygenated in terms monoterpenes,of the content of coumarin oxygenated and monoterpenes,herniarin. The coumarin extraction and yield hernia was influencedrin. The extraction by the interaction yield was between influenced the by pressure the interaction and CO2 betweenflow rate, the as pressure well as between and CO2 the flow temperature rate, as well and as CObetween2 flow the rate. temperature The optimal and conditions CO2 flow basedrate. The on optimala Box-Behnken conditions design based were on w a temperatureBox-Behnken of design 41 ◦C, pressurewere w temperature of 100 bar and of a41 CO °C,2 flow pressure rate ofof 2.3100 kg bar/h. andSimilar a CO parameters2 flow rate (40 of◦ C,2.3 100 kg/h. bar) Similar were applied parameters by Nadalin (40 °C, et 100 al. [ 10bar)] for were the SFEapplied of lavender by Nadalin (Lavandula et al. [10]officinalis for the) flowers. SFE of Inlavender the study (Lavandula by Akgün officinalis et al. [8)], flowers. the SFE In was the conducted study by atAkgün relatively et al. low [8], pressurethe SFE (80–140was conducted bar) and at the relatively temperature low pressure used was (80–140 in the 35–50 bar) and◦C range. the temperature The application used ofwas supercritical in the 35–50 fluid °C range.extraction The to application phenolic compounds of supercritical has been fluid widely extraction developed to phenolic in a number compounds of studies has as been gathered widely in developeda recent review in a bynumber Ty´skiewiczet of studies al. [12as]. gathered Supercritical in a carbonrecent dioxidereview withby Ty a co-solventśkiewicz et was al. often[12]. Supercriticalused for the extractioncarbon dioxide of the phenolicwith a co-solvent compounds, was but often extraction used for with the pure extraction carbon dioxideof the phenolic was also compounds,employed [12 but]. In extraction some cases, with the pu SFEre providedcarbon dioxide higher was or similar also employed extraction [12]. efficiency In some values cases, compared the SFE providedto organic higher solvent or (methanol, similar extraction ethanol) extracts.efficiency Based values on compared the literature to reports,organic thesolvent methanolic (methanol, and ethanol)ethanolic extracts. extracts fromBasedBaccharis on the dracunculifolia literature reports,were characterizedthe methanolic by aand 2-fold ethanolic lower extraction extracts yieldfrom Baccharisthan the scCOdracunculifolia2 extract (60were◦C, characterized 400 bar) of this by plant. a 2-fold The lower analysis extraction of 3,5-diprenyl-4-hydroxycinnamic yield than the scCO2 extract (60acid °C, and 3-prenyl-4-hydroxycinnamic400 bar) of this plant. The acid analysis also indicated of 3,5-diprenyl-4-hydroxycinnamic a higher content in the scCO2 extractacid [and13]. 3-prenyl-4-hydroxycinnamicHowever, the literature data lacksacid also data indicated on the optimization a higher content of Lavandula in the scCO angustifolia2 extractextraction. [13]. However, the literatureThe aim ofdata the lacks current data study on the was optimization to investigate of the Lavandula extraction angustifolia of lavender extraction. flowers using supercritical CO2 (greenThe aim technology) of the current to obtain study polar was compounds to investigat and toe optimizethe extraction the main of extraction lavender parameters flowers using with supercriticalthe use of response CO2 (green surface technology) methodology to obtain (RSM) polar based compounds on the extraction and to yield optimize and total thephenolic main extraction content. parameters with the use of response surface methodology (RSM) based on the extraction yield and 2. Results and Discussion total phenolic content. 2.1. Supercritical Fluid Extraction of Lavender 2. Results and Discussion Supercritical fluid extraction has been applied to lavender extraction in a study by Av¸saret al. [14]. Lavandula officinalis flowers were extracted at 45 ◦C and 145 bar, resulting in an extraction yield of Molecules 2019, 24, 3354 3 of 15
4.68 wt.%. Similar yield values were observed in our study, although, different parameters were required. The yields of the Lavandula angustifolia flower extraction were 4.3 wt.% and 4.8 wt.%, respectively, for extraction at 50 ◦C/300 bar and 50 ◦C/200 bar. Costa et al. [15] performed SFE on Lavandula virdis extending the extraction with two separators. In comparison with the study by Av¸saret al. [14], lower extraction efficiency levels were observed (3.45 wt.% vs. 4.68 wt.%) for similar or not significantly different parameters (45 ◦C/145 bar vs. 40 ◦C/120 bar). The use of the second separator enabled achieving a higher yield of 9.27 wt.%. Increasing the pressure from 120 to 180 bar at the same extraction temperature (40 ◦C) did not influence the yield (3.45 wt.% vs. 3.41 wt.%). In the studies by Dahn et al. [16] volatile oils of Lavandula angustifolia were extracted in the range of 100–180 bar, 40–50 ◦C and 6.4–60 min. The highest extraction yield was obtained using the following parameters: 50 ◦C, 180 bar, and 60 min. The authors also succeeded in obtaining a yield of 9.0 wt.% at 53.4 ◦C and 140 bar. Our study resulted in obtaining the same yield (9.0 wt.%) at a higher temperature (60 ◦C) and over twice the pressure (300 bar). Lowering the pressure to 250 bar resulted in an even higher yield of up to 9.2 wt.%. Table1 presents the experimental results obtained in 15 runs of Lavandula angustifolia flower extraction.
Table 1. Responses of dependent variables using three levels-the factors Box-Behnken method to independent variables of the SFE method.
Total Phenolic Total Phenolic Temperature Antioxidant Activity Pressure (bar) Time (min) Observed Content (mg Content (mg ( C) (AA%) Exp ◦ Yield GAE/g Extract) GAE/g Dry Mass) (w/wt.%) C U C U C U Extracts Residues Extracts Residues 1 1 300 0 50 1 15 4.3 7.64 0.99 68.35 0.7 92.18 0.3 − ± ± 2 1 300 0 50 1 45 8.9 9.57 1.31 78.10 0.4 92.06 0.5 ± ± 3 0 250 0 50 0 30 7.3 8.84 1.28 61.27 1.1 92.09 0.4 ± ± 4 0 250 0 50 0 30 7.1 8.56 1.14 76.67 0.4 91.05 0.6 ± ± 5 0 250 1 60 1 15 6.5 8.25 1.20 52.39 0.5 89.88 0.9 − ± ± 6 1 200 1 40 0 30 7.1 8.75 1.32 50.55 0.7 93.44 1.1 − − ± ± 7 1 200 1 60 0 30 7.2 4.32 1.19 61.52 0.8 91.55 1.0 − ± ± 8 1 200 0 50 1 15 5.7 6.84 0.98 59.17 0.6 91.27 0.4 − − ± ± 9 0 250 1 40 1 15 6.2 9.81 1.19 78.83 1.3 91.39 0.6 − − ± ± 10 0 250 0 50 0 30 6.9 9.68 1.19 52.35 0.1 91.25 0.4 ± ± 11 0 250 1 40 1 45 6.9 6.91 1.08 63.89 0.5 90.97 1.2 − ± ± 12 1 200 0 50 1 45 4.8 5.9 1.08 60.65 0.3 91.36 0.4 − ± ± 13 1 300 1 60 0 30 9.0 9.95 1.23 65.86 0.7 91.18 0.8 ± ± 14 0 250 1 60 1 45 9.2 10.17 1.16 61.08 0.4 91.36 0.3 ± ± 15 1 300 1 40 0 30 7.4 5.82 1.30 57.00 0.5 91.35 0.4 − ± ± C—coded, U—uncoded.
2.2. Optimization of Operating Conditions—Extraction Yield A Box-Behnken statistical experimental design was utilized and the extraction yields were 2 measured for different variables such as pressure, temperature and time, coded as X1,X2,X3. The R adjusted of the extraction yield was 97.40% indicating that chosen models allowed one to fully predict the extraction yield, as presented in Figure2. The response surface model (RSM) obtained from the experimental design was evaluated using ANOVA and analysis of residuals. Multiple regression coefficients were determined by applying a least-squares technique in order to predict a second-order polynomial model. Both F-value and p-value were used to determine the significance of each coefficient. Corresponding p-values not higher than 0.1 suggest that, among the test variables used in this study X1 (pressure), X2 (temperature), X (time), X X (pressure temperature), X X (pressure time), X X (temperature time), (X )2 3 1 2 × 1 3 × 2 3 × 1 (pressure pressure), (X )2 (temperature temperature), (X )2 (time time) are all significant model × 2 × 3 × terms. The model F-value of 59.21 implies the model is significant. There is only a 0.02% chance that an F-value this large could occur due to noise. The results of variance and error for the response surface model analysis indicate that lack-of-fit is not significant relative to the pure error. There is a 49.87% chance that a lack-of-fit F-value this large could occur due to noise (Table2). Molecules 2019, 24, x FOR PEER REVIEW 4 of 15 Molecules 2019, 24, 3354 4 of 15
Predicted vs. Actual Residuals vs. Predicted
10 0.3
9 0.2
8 0.1
7 0.0
6 -0.1
5 -0.2
4 -0.3
45678910 4.0 5.0 6.0 7.0 8.0 9.0 10.0
X1: Actual X1: Predicted X2: Predicted X2: Residuals
Figure 2. Observed extraction yield versus predicted extraction yield and the analysis of residuals.
TableFigure 2. 2.Regression Observed coe extractionfficients ofyield predicated versus predicted second-order extr polynomialaction yield model and the for an thealysis response of residuals. variable.
TheSource response surface Sum ofmodel Squares (RSM) Degreeobtained of Freedomfrom the experimental Mean of Squares designF was-Value evaluatedp-Value using * ANOVAModel and analysis of residuals. 26.83 Multiple regressi 9on coefficients 2.98 were determined 59.21 by applying 0.0002 a least-squaresX1 (pressure) technique in 2.83order to predict a second-order 1 polynomial 2.83 model. 56.26 Both F-value 0.0007 and p-valueX2 (temperature) were used to determine 2.39 the significance 1 of each coefficient. 2.39 Corresponding 47.42 p-values 0.0010 not X3 (time) 6.00 1 6.00 119.26 0.0001 higher than 0.1 suggest that, among the test variables used in this study X1 (pressure), X2 X X 0.59 1 0.59 11.78 0.0186 1 × 2 (temperature),X X X3 (time), X17.29X2 (pressure × temperature), 1 X1X3 (pressure 7.29 × time), 144.82X2X3 (temperature<0.0001 × 1 × 3 time), (XX 1)2 X(pressure × pressure),1.01 (X2)2 (temperature 1 × temperature), 1.01 (X3)2 (time 20.06 × time) 0.0065 are all 2 × 3 X X 0.44 1 0.44 8.75 0.0316 significant1 × model1 terms. The model F-value of 59.21 implies the model is significant. There is only a X X 3.21 1 3.21 63.72 0.0005 0.02% chance2 × 2 that an F-value this large could occur due to noise. The results of variance and error for X X 2.56 1 2.56 50.89 0.0008 the response3 × 3surface model analysis indicate that lack-of-fit is not significant relative to the pure Residual 0.25 5 0.050 - - error. Lack-of-fitThere is a 49.87% chance 0.16 that a lack-of-fit F 3-value this large 0.053could occur due 1.14 to noise 0.4987(Table 2). Pure error 0.093 2 0.046 - - Table 2. Regression coefficients* Significant of predicated at p < 0.1. second-order “-” – not applicable. polynomial model for the response variable.
The plots of the quadraticSum of model withDegree three of variablesMean kept of at constant level and the other two Source F-value p-value * varying within the experimentalSquares ranges areFreedom shown in FigureSquares3. The final equation in terms of coded variablesModel is as follows: 26.83 9 2.98 59.21 0.0002 X1 (pressure) 2.83 1 2.83 56.26 0.0007 2 2 2 YX2 =(temperature)7.09 + 0.59X + 0.55X2.39+ 0.87X + 0.39X1 X + 1.35X2.39X + 0.5 0.35X47.42+ 0.93X 0.83X0.0010 (1) 1 2 3 1 2 1 3 − 1 2 − 3 X3 (time) 6.00 1 6.00 119.26 0.0001 TheX1final × X2 equation in terms0.59 of actual factors1 is as follows:0.59 11.78 0.0186 X1 × X3 7.29 1 7.29 144.82 <0.0001
X2 × X3 Y = 39.1395 1.010.0115X 1 1.1704X1 2 0.3376X31.01+ 0.0008X 1X2 +20.060.0018X 1X3 0.0065 X1 × X1 −0.44 − 1 − 2 0.442 2 8.75 0.0316 (2) + 0.0035X2X3 0.0001X1 + 0.0093X2 0.0037X3 X2 × X2 3.21 − 1 3.21 − 63.72 0.0005 X3 × X3 2.56 1 2.56 50.89 0.0008 Residual 0.25 5 0.050 - - Lack-of-fit 0.16 3 0.053 1.14 0.4987 Pure error 0.093 2 0.046 - - * Significant at p < 0.1. “-” – not applicable
Molecules 2019, 24, x FOR PEER REVIEW 5 of 15
The plots of the quadratic model with three variables kept at constant level and the other two varying within the experimental ranges are shown in Figure 3. The final equation in terms of coded variables is as follows: