Influence of common juniper berries pretreatment on the essential oil yield, chemical composition and extraction kinetics of classical and microwave-assisted hydrodistillation

Industrial Crops & Products 122 (2018) 402–413

Contents lists available at ScienceDirect

Industrial Crops & Products journal homepage: www.elsevier.com/locate/indcrop

Influence of common juniper berries pretreatment on the essential oil yield, chemical composition and extraction kinetics of classical and microwaveassisted hydrodistillation

T

⁎

Miljana S. Markovića, , Dragana B. Radosavljevića, Vladimir P. Pavićevićb, Mihailo S. Ristićc, Svetomir Ž. Milojevića, Nevenka M. Bošković-Vragolovićb, Vlada B. Veljkovićd a

Faculty of Technical Sciences, University of Priština, Kneza Miloša 7, 38220 Kosovska Mitrovica, Serbia Faculty of Technology and Metallurgy, University of Belgrade, Karnegijeva 4, 11000 Belgrade, Serbia Institute for Medicinal Plants Research Dr. Josif Pančić, Tadeuša Košćuška 1, 11000 Belgrade, Serbia d University of Niš, Faculty of Technology, Bulevar oslobodjenja 124, 16000 Leskovac, Serbia b c

A R T I C LE I N FO

A B S T R A C T

Keywords: Juniper berries Juniperus communis L. Essential oil Kinetics Microwave-assisted hydrodistillation Optimization

The present paper dealt with the influence of the common juniper berries pretreatment on the yield, chemical composition and extraction kinetics of juniper essential oil (JEO) obtained by classical (HD) and microwaveassisted hydrodistillation (MAHD). The highest JEO yield was obtained by HD from one-minute dry-ground juniper berries (2.23 ± 0.00 g/100 g). No statistically significant influence of swelling and distillation technique on JEO yield was observed. Therefore, the optimal pretreatment process involved no swelling and one-minute grinding. However, no significant difference in the chemical composition of the JEOs obtained by the two techniques was observed. A new phenomenological kinetic model was developed on the basis of the mechanism of JEO extraction by both HD MAHD, which assumed three simultaneously-occurring stages: washing, unhindered diffusion and hindered diffusion. The main advantage of developed model was its ability to describe the variations of JEO yield and distillation rates with time. Furthermore, it had the smallest mean relative percentage deviation compared to the well-known kinetics models and the parameters that all were statistically significant, so it was recommended for modeling the kinetics of JEO extraction by HD and especially MAHD.

1. Introduction The common juniper (Juniperus communis L.) is a coniferous evergreen perennial tree or shrub from the Cupressaceae family mostly widespread in the mountains of Europe, Asia and North America (Ložienė and Venskutonis, 2016). Berry-like fruits of common juniper, known as juniper berries (Juniperi fructus), are commercially the most essential part due to many medicinal and food-ingredient uses (EMA, 2011; EMEA, 1999; Veljković and Stanković, 2003). Nowadays, the juniper berries and essential oil are pharmaceutical raw materials recognized by the European Pharmacopoeia. Juniper berries are reported to have powerful diuretic, antiseptic, stomachic, antirheumatic, antiviral and anti-inflammatory activities that are primarily associated with the juniper essential oil (JEO) (EMA, 2011). The commonly predominant constituent of JEO is α-pinene, although sabinene, myrcene, limonene, and terpinen-4-ol are also important constituents (Ložienė and Venskutonis, 2016). Therefore, a large number of studies have been dealing with the yield and chemical composition of JEOs obtained by ⁎

Corresponding author. E-mail address: [email protected] (M.S. Marković).

https://doi.org/10.1016/j.indcrop.2018.06.018 Received 15 February 2018; Received in revised form 7 May 2018; Accepted 6 June 2018 0926-6690/ © 2018 Elsevier B.V. All rights reserved.

various techniques all over the world (Chatzopoulou and Katsiotis, 1995; Damjanović-Vratnica et al., 2003, 2006; Marongiu et al., 2006; Pavićević et al., 2016; Marković et al., 2017). These studies show that the yield and chemical composition of the JEO are influenced by the soil and climate at which the plants were grown (Butkienë et al., 2004; Tasić et al., 1993), the fruit ripeness (Ložienė and Labokas, 2012), the grinding process (Chatzopoulou and Katsiotis, 1995) and the used extraction techniques (Damjanović-Vratnica et al., 2003, 2006; Pavićević et al., 2016). The majority of essential oils are commonly obtained by hydrodistillation (HD). During the last decade, typical hydrodistillation have been advanced using microwave radiation for heating the aqueous suspension of the crushed plant material. This technique, known as microwave-assisted hydrodistillation (MAHD), has widely been used to obtain essential oils from various plant materials (Amaresh et al., 2017; Dong et al., 2017; Golmakani and Rezaei, 2008a,b; Kapás et al., 2011; Karakaya et al., 2014; Kusuma and Mahfud, 2017a,b,c; Mohammadhosseini, 2017; Pavićević et al., 2016; Phutdhawong et al.,

Industrial Crops & Products 122 (2018) 402–413

M.S. Marković et al.

100 mL of the water collected from the swelling were added to the disintegrator, which was then switched on/off each 20 s during 1, 2 or 3 min. Besides that, for one-minute grinding, 50 mL of the same water was added after 20 and 40 s of grinding (i.e. 100 mL in total). For twominute grinding, the same water (2 × 50 mL) was added after 40 and 60 s of grinding; for three-minute grinding, the water (2 × 50 mL) was added after 80 and 120 s of grinding. The rest of the swelled berries was wet ground in the same way in the two additional batches. Finally, the three suspensions of the ground swelled berries (300 g in 600 mL of water) were mixed and 600 mL of distilled water was added to the resulted suspension to achieve the juniper berries-to-water mass ratio (hydromodul) of 1:4. The hydromodul is commonly used in the range between 1:3 and 1:10 (Milojević et al., 2008). The former ratio is frequently applied in the industrial production of essential oil from juniper berries (Stanković et al., 1994) while the latter ratio is recommended by official pharmacopoeias (Pharmacopoeia Jugoslavica, 1984). In the present case, the hydromodul of 1:3 was avoided to prevent overheating of the plant material which could occur due to the excessive soaking of water by the ground berries.

2007; Sourmaghi et al., 2014) due to a number of advantages, such as: higher heating speed, no direct contact of plant material with heat source, easier process control and so on (Nitthiyah et al., 2017). Despite its advantages, MAHD is rarely employed for recovery of the essential oil from juniper berries. So far, only Pavićević et al. (2016) have studied the kinetics of the microwave-assisted separation and chemical composition of the JEO. Besides that, Dahmane et al. (2015) have reported the chemical composition of the essential oil obtained from common juniper needles by MAHD. The kinetics of essential oil HD from various plant materials has been extensively investigated (Busato et al., 2014; Milojević et al., 2013; Rezazi et al., 2016; Sovova and Aleksovski, 2006) while the MAHD kinetics is less considered (Kusuma and Mahfud, 2017a,b,c; Pavićević et al., 2016). The kinetics of JEO extraction by HD and MAHD are usually described by the two- and three-parameter models (Milojević et al., 2008; Pavićević et al., 2016). Only Pavićević et al. (2016) compared the kinetic models for HD and MAHD of the JEO. These models assume two simultaneous processes: (a) rapid distillation of the essential oil from external surfaces of the plant particles (so called washing) and (b) slow diffusion of the essential oil through the plant particles. The present paper deals with the effects of the pretreatment of ripen common juniper berries on the yield, chemical composition and extraction kinetics of essential oil obtained by HD and MAHD. The first goal was to estimate the influence of grinding and swelling on JEO yield and to optimize these two pretreatment techniques by the response surface methodology (RSM) in order to get the best JEO yield. Second, a new phenomenological kinetic model was developed, which assumed that JEO distillation by both HD and MAHD occurs via three stages: washing, unhindered diffusion and hindered diffusion appearing simultaneously. The unhindered diffusion involves the JEO mass transfer from ruptured organs without any limitation while the hindered diffusion is the JEO mass transfer through membranes of intact plant organs. So far, these two diffusions occurring within plant particles have been considered as a united mass transfer (diffusion) process. Besides that, the model assumed that the maximum rates of the three extraction mechanisms occurred after certain period of distillation time. This novel kinetic model was compared with the phenomenological model involving simultaneous washing and diffusion (Pavićević et al., 2016; Sovova and Aleksovski, 2006), the model involving instantaneous washing and diffusion (Milojević et al., 2008), the exponential model involving only diffusion (Morin et al., 1985) and the second-order model (Muhammad Hazwan et al., 2012).

2.3. JEO distillation: HD and MAHD For both HD and MAHD of JEO, the Clevenger apparatus was employed, as in the previous study (Pavićević et al., 2016). For HD, the 2 L distillation round-bottom flask was placed in an electric heater. For MAHD, the Clevenger apparatus with the distillation flask was placed in a laboratory microwave oven (maximum power: 900 W; frequency: 50 Hz). The intensity of heating (700 W) in both HD and MAHD ensured the same aromatic water flow rate (8.5 ± 0.5 mL/min). The prepared suspension of ground berries was added to the distillation flask and heating was started. The appearance of the first drop of the JEO designated the beginning of the distillation process. From that moment, the time and the volume of the collected JEO were recorded during 4 h of distillation. Occasionally, the JEO collected in the graduated tube was discharged into a 10 mL graduated cylinder having 0.1 mL grading divisions. The JEO collected during the distillation was dried over anhydrous sodium sulfate, stored in glass bottles and held in a refrigerator at 4 °C until its analysis. 2.4. Analytical procedures The oil was analyzed by analytical GC/FID and GC/MS, using the normalization procedure, based upon the integration of chromatograms obtained by GC/FID.

2. Material and methods 2.4.1. Gas chromatography (GC) GC analysis was carried out using a 7890A Agilent gas chromatograph (Agilent Technologies Co. Ltd, Shanghai Branch Company, Shanghai, China) equipped with a split-splitless injector, a flame ionization detector (FID) and a 30 m × 0.25 mm HP-5 (cross-linked PhenylMethyl Siloxane) column with 0.25 μm film thickness (Agilent). Hydrogen was used as carrier gas at 210 °C (under constant pressure; 1 mL/min). The temperature of the injector and the detector were 220 °C and 240 °C, respectively. The column was initially at 60 °C, increased linearly to 240 °C at the rate of 3 °C/min and held at 240 °C for 10 min. The solution of juniper essential oil in ethanol (∼1%) was injected using ALS (1 μL, split mode, 1:30). Statistics has been covered by FID specification (results with a range of deviation for the level 1%).

2.1. Juniper berries Ripe juniper berries were collected from the southern hillsides of the Kopaonik mountain (1000 m above the sea level, 43°14′6″N, 20°49′18″E), Kosovo and Metohija, Serbia, in September 2016. The juniper berries were dried in the shade and packed in the multilayer paper bags. 2.2. Pretreatment of juniper berries The juniper berries were ground both dry and wet by a disintegrator (Bosh, 500 W; medium intensity) prior to essential oil distillation. For the dry grinding, juniper berries (100 g) were added into the disintegrator, which was switched on/off each 20 s during 1, 2 or 3 min to avoid overheating of the berries that would cause the loss of a part of JEO. Then, the ground berries (300 g) were suspended into distilled water (1200 mL) and the resulted suspension was subjected to distillation. For the wet grinding, juniper berries (300 g) were first kept in distilled water (1000 mL) for 24, 48 or 72 h for swelling and then separated from the water. One third of the swelled juniper berries and

2.4.2. Gas chromatography/mass spectrometry (GC/MS) The same chromatographic analytical conditions as those mentioned for GC/FID were employed for GC/MS analysis, along with capillary column HP-5MS (30 m × 0.25 mm, 0.25 μm film thickness), using HP G 1800C GCD Series II Electron Ionization Detector (EID) system (Hewlett-Packard, Palo Alto, CA, USA). Instead of hydrogen, helium was used as carrier gas. Transfer line was heated at 240 °C. Mass 403

Industrial Crops & Products 122 (2018) 402–413

M.S. Marković et al.

distance between the model and the experiment (Akaike, 1973). This model is expected to have a good fit to the experimental data but fewer parameters, thus balancing between their complexity and goodness of fit. AIC is defined as follows:

spectra were acquired in EI mode (70 eV), in m/z range 40-400. Sample solutions in ethanol (∼1%) were injected by ALS (1 μL, split mode: 1:30). The constituents were identified by comparison of their mass spectra to those from Wiley275 and NIST/NBS libraries, using different search engines, Probability Merge Search (PBM) included in instruments G1701DA. ver. D.00.00.38 data station software and NIST 2.0 search program. The experimental values for retention indices were determined by the use of calibrated Automated Mass Spectral Deconvolution and Identification System software (AMDIS ver.2.64), compared to those from available literature (Adams, 2007), and used as additional tool to approve MS findings.

where L is the maximum value of the likelihood function and K is the number of parameters. The first part of Eq. (4) estimates the fit between the model and the experimental data while the second part is a penalty for including extra parameters in the model. However, AIC requires a bias correction for small data samples (Hurvich and Tsai, 1989; Sugiura, 1978):

2.5. Statistical evaluation of pretreatment

AICc = AIC +

The influence of pretreatment (grinding and swelling time) of the juniper berries and the distillation technique (i.e. electric or microwave heating) on JEO yield was statistically evaluated. JEO yield achieved by HD was assessed by the RSM combined with an experimental design with two factors (grinding time and swelling time at 3 and 4 levels with replication, respectively; 24 experiments in total). The used software suggested the use of a linear model extended with a two-factor interaction for the statistical evaluation of the influence of the process factors on JEO yield (called here 2FI model):

where n is the data sample size. If n is large with respect to K, the AIC correction term is negligible. In cases where there are relatively few data per estimated parameter, AICc should be used rather than AIC. According to a rule of thumb, this correction is needed if n K < 40 (Burnham and Anderson, 2002). The preferred model with respect to relative quality will be the model with the minimum AICc value.

Y = b0 + b1 X1 + b2 X2 + b12 X1⋅X2

AIC = −2⋅log (L) + 2K

3.1.1. Influence of juniper berries pretreatment on JEO yield The influence of the juniper berries pretreatment by grinding and swelling on JEO yield obtained by the 4 h HD and MAHD was investigated first by the Duncan’s multiple range test at the 95% confidence interval. Table 1 shows that grinding and swelling times affect JEO yield. With increasing of the grinding time, JEO yield decreases regardless of the grinding technique (dry or wet grinding). The best JEO yield of 2.23 ± 0.00 g/100 g was obtained from the one-minute dryground juniper berries. On the basis of the Duncan’s multiple range test at the 95% confidence interval, this JEO yield was higher than all other ones obtained by HD at the 95% confidence interval, except the two achieved by HD with the one- and two-minute wet-ground 24 h-swelled juniper berries (2.19 ± 0.01 g/100 g and 2.14 ± 0.01 g/100 g) and one obtained by MAHD with one-minute wet-ground 24 h-swelled juniper berries (2.20 ± 0.00 g/100 g). So far, contradictory observances were reported on the effect of the heating method on essential oil yield under the same other distillation conditions. While Sourmaghi et al. (2014) observed a higher essential oil yield by HD than by MAHD from Coriandrum sativum, Golmakani and Rezaei (2008a,b) reported a higher Table 1 The effect of pretreating juniper berries on the JEO yield.

The parameters of the used kinetic models were calculated by the nonlinear regression on the basis of the experimental JEO yields using the R-Project software. The mean relative percentage deviation MRPD (%), and the coefficient of determination, R2 , were used to assess the goodness of fit of the used kinetic models on the basis of the experimental and model values of JEO yield, qi and qm, i , respectively:

100 n

∑ i=1

Technique

Grinding time, min

Swelling time, h

Final JEO yield,a g/ 100 g

Standard deviation, g/ 100 g

Duncan’s test

HD

1

0 24 48 72 0 24 48 72 0 24 48 72 0 24

2.23 2.19 2.14 2.10 1.95 2.14 2.13 2.15 1.85 1.89 2.04 2.14 2.14 2.20

0.00 0.02 0.05 0.00 0.03 0.05 0.00 0.03 0.07 0.02 0.05 0.06 0.02 0.00

a a, b, c b, c, d d, e f a, b, c c, d b, c, d g g e d, e b, c, d a, b

|qi − qm, i | qi

(2) 2

n

R2 = 1 −

∑i = 1 (qi − qm, i )2 n

∑i = 1 (qi − q )2

(5)

3.1. Influence of juniper berries pretreatment on JEO yield and chemical composition

(1)

2.6. Model quality assessment

MRPD =

2K (K + 1) n−K−1

3. Results and discussion

where Y is the JEO yield, b0 is the constant regression coefficient, b1 and b2 are the linear regression coefficients and b12 is the two-factor interaction regression coefficient, X1 is the grinding time and X2 is the swelling time. Multiple nonlinear regression was used to determine the parameters of the developed model’s Eq. (1) while its fit was evaluated by the analysis of variance (ANOVA) revealing the statistically significant process factors and their interactions with the 95% confidence level (p-value < 0.05). As the analyzed experimental data on JEO were unbalanced, the ANOVA type III was applied. A number of optimization points, where the maximum JEO yield was achieved for a set of the pretreatment conditions, were also found by the same computer program. The Duncan’s multiple range test were used for comparing a pair of independent means and a set of means with the 95% confidence interval, respectively. The Kolmogorov-Smirnov normality test was used to assess if JEO yield was normally distributed. The R-Project software (open source, http://cran.us.r-project.org) was used for the statistical assessment.

n

(4)

(3) 3

where q is the mean of the analyzed set of experimental values. Furthermore, the Akaike Information Criterion (AIC) was used for selecting the best model among the six models with different number of parameters that were developed on the basis of the same dataset. It provides information about the relative quality of these models (not in an absolute sense). The selected model minimizes the Kullback-Leibler

MAHD

a

404

1

Mean value from two experiments.

Industrial Crops & Products 122 (2018) 402–413

M.S. Marković et al.

model. However, the significant lack of fit ( p = 0.0015 < 0.050 ) indicated that the reduced 2FI model might not accurately represent the observed JEO yields in the applied experimental region. This significant lack of fit was attributed to the small variance of replicated runs, proved by a small mean relative error between the replicates ( ± 0.8%), causing the small pure error. Thus, the significant lack of fit did not invalidate the developed, reduced 2FI model (Veljković, 2014). Moreover, since the small MRPD-value ( ± 2.0% based on 24 data, Fig. S1c, Supplementary material) and other above-discussed statistical criteria characterized the reduced 2FI model as a good one (Table 2), it was adopted as adequate for calculating and optimizing JEO yield. Fig. 1 shows the response surface 3D plot along with contour plot for JEO yield achieved by HD as a function of grinding time and swelling time resulted from Eq. (6b). These plots were suitable for visualizing not only the effects of the two factors and their interaction on the JEO yield but also the optimal process conditions. The two-factor interaction allowed for twisting the plane resulted from the linear part of Eq. (6b). These plots showed clearly that JEO yield increased with the increasing swelling time and decreasing grinding time. Both plots showed the highest JEO yield for the one-minute 24 h-swelled juniper berries. The optimal levels of the pretreatment process factors were determined with the aim of achieving the maximum JEO yield using the developed 2FI model, Eq. (6b), within the applied experimental ranges. The optimal pretreatment process, determined by the used software, involved no swelling and one-minute grinding. The predicted best JEO yield of 2.22 g/100 g agreed with the experimentally obtained JEO yield of 2.23 g/100 g. On the basis of the results of the statistical evaluation, the further investigation was focused on the kinetics of HD and MAHD of the oneminute dry- and wet 24 h-swelled juniper berries.

final essential oil yield from fresh aerial parts of thyme in a shorter time by MAHD than by HD. On the other side, the same equilibrium essential oil yield by the two heating methods were found from fennel seeds (Kapás et al., 2011), rosemary leaves (Karakaya et al., 2014), aerial parts of summer and winter savory (Rezvanpanah et al., 2008) and leaves of Cinnamomum iners (Phutdhawong et al., 2007). 3.1.2. Evaluation of pretreatment process factors by RSM The JEO yield dataset from Table 1 was first assessed by the Kolmogorov-Smirnov normality test and for the presence of outliers. The analyzed dataset was significantly drawn from a normally distributed population at the 0.05 level (t-statistic = 0.243, p = 0.0988) (Fig. S1a, Supplementary material) and with no outlier value (Fig. S1b, Supplementary material). Then, the parameters of Eq. (1) were determined by the nonlinear regression and the influential process factors and their interactions were evaluated through the ANOVA. The developed regression models in terms of coded and uncoded factors are as follows, respectively: (6a)

Y = 2.079 − 0.099⋅X1 + 0.050⋅X2 + 0.101⋅X1⋅X2 and

(6b)

Y = 2.430 − 0.200⋅X1 − 0.004⋅X2 + 0.003⋅X1⋅X2

For this model, the ANOVA showed that all terms were statistically influential on the JEO yield at the 95% confidence level, as it can be seen in Table 2. According to its F-and p-values (27.5 and < 0.0001, respectively), the developed 2FI model is significant. The p-value (< 0.0001) of the 2FI model indicated less than only 0.01% chance of that this F-value was large because of the noise in the experiment. Having a higher F-value, grinding time ( X1) had a more significant effect on JEO yield than swelling time ( X2 ). Besides that, grinding time negatively influenced JEO yield while the swelling time had a positive effect, as it can be concluded from the linear regression coefficients (b1 < 0 and b2 > 0 , respectively), and can be seen in Eq. (6a). The increase of the grinding time resulted in a smaller JEO yield, which might be attributed to a greater loss of the JEO during grinding due to its evaporation throughout the prolonged heating. The increase of the JEO loss during grinding was prevented by the juniper berries immersion in water. The positive effect of swelling might be due to a more effective rupturing of juniper berries, which increased the JEO availability, and the heat absorption by water, which prevented overheating and the JEO evaporation during the grinding process. It was interesting to observe in Eq. (6a) that the X1 − X2 interaction (b12 > 0 ) had the positive and largest effect on JEO yield. Several statistical criteria were analyzed to estimate how the developed 2FI model fitted the experimental data of JEO yield. Value of the coefficient of correlation (R = 0.897 ) showed a strong dependence between JEO yield and the two pretreatment process factors. The R2value of 0.805 implied a fairly well goodness of fit of the developed 2FI 2 2 model. The Rpred -value (0.775) was in the agreement with Radj (0.745), indicating a good prediction of JEO yield by the developed 2FI model while the C . V . -value (2.8%) indicated a good reproducibility of the

3.1.3. Influence of juniper berries pretreatment on JEO chemical composition Table 3 reports the mass fractions (%) of the selected constituents of the JEOs obtained by HD and MAHD from the juniper berries prepared by different pretreatment methods (dry or wet grinding, grinding time and swelling time). These constituents had mass fractions larger than 0.5% and they all made about 95–96% of the essential oils mass. In the case of the essential oils obtained by HD, wet grinding (72 h – swelling time) provided higher mass fractions of high-volatile constituents (the mean value of 2.47% for all grinding times) and medium-volatile constituents (the mean value of 1.10% for all grinding times), consequently lower mass fractions of low-volatile constituents (the mean value of 3.35% for all grinding times) than a dry grinding, leading to the more valuable final product. As far as MAHD was concerned, wet grinding (one-minute grinding, 24 h – swelling time) gave higher mass fraction of high-volatile constituents (1.30%), approximately the same of medium-volatile constituents, consequently lower mass fraction (1.07%) of low-volatile constituents than a dry grinding. 3.2. Mechanism of JEO distillation

Table 2 The results of the ANOVA (type III) for JEO yield achieved by HD from juniper berries ground and swelled for different times. Source

Sum of squares

df

Mean square

F-value

p-value

Model X1 (grinding time) X2 (swelling time) X1⋅X2 Residual Lack of Fit Pure Error Cor Total

0.281 0.158 0.033 0.090 0.068 0.056 0.012 0.349

3 1 1 1 20 8 12 23

0.094 0.158 0.033 0.090 0.003 0.007 0.001

27.5 46.3 9.6 26.5

< 0.0001 < 0.0001 0.0006 < 0.0001

7.1

0.0015

Fig. 2 illustrates variations of JEO yield during HD and MAHD from the one-minute dry- and wet-ground 24 h-swelled juniper berries from the beginning of the heating up to the achievement of the equilibrium when no further distillation of JEO occurs. The MAHD started earlier than HD regardless of the juniper berries pretreatment, which agreed with earlier observances (Golmakani and Rezaei, 2008a,b; Kapás et al., 2011). The preheating time was about four times less for the former distillation technique, which might be attributed to the faster transfer of the dissipated microwave irradiation energy. The curves for both distillation techniques differ to each other according to their shape. While the HD curves are hyperbolic, the MAHD curves are slightly sigmoidal with a transition point in the washing stage. The sigmoidal shape of the MAHD curves has already been observed for MAHD of essential oils

2 2 R2 = 0.805, Radj = 0.775; Rpred = 0.745 and C . V . = 2.8%.

405

Industrial Crops & Products 122 (2018) 402–413

M.S. Marković et al.

Fig. 1. Response surface and contour plots for JEO yield (Y ) as a function of grinding time ( X1) and swelling time ( X2 ) (hydromodul: 1:4 and distillation rate: about 8.5 ± 0.5 mL/min).

essential oil removal rate reduced gradually, which was attributed to the limitations of unhindered and hindered diffusions. This mechanism was tightly associated with the anatomy of juniper berries and their comminution. Elongated tubercles of juniper berries, containing essential oil, could be ruptured by comminution, releasing a part of the essential oil at the external surface of the plant particles, from which it was rapidly “washed”. Simultaneously, the essential oil from the ruptured reservoirs diffused without any limitation toward the external surface of the plant particles (unhindered diffusion) while the essential oil from the intact reservoirs diffused through membranes and other barriers present in the plant particles (hindered diffusion). As a consequence of this mechanism, essential oil distilled off slower and slower

from cherry laurel (Prunus laurocerasus) leaves (Karabegović, 2011). This phenomenon might be attributed to certain inertial effects caused by non-uniform heating of the aqueous suspension. Nonpolar essential oil was heated indirectly at its own rate, which differed from the heating rate of polar water. If the plant material was water-swelled, this effect was less noticeable (Fig. 2). On the other hand, the general shape of the curves confirmed the existence of fast and deaccelerated essential oil distillation stages in the initial and later periods (washing and diffusion, respectively). The initial rapid increase of essential oil yield was due to the fast “washing” of the essential oil from the outer surface of the juniper particles. During the intermediary transition stage up to the equilibrium stage, the

Table 3 The effect of pretreating juniper berries on the juniper essential oil composition (in %).a. Technics

HD

Swelling time, h Grinding time, min

0 1

24 1

48 1

72 1

0 2

24 2

48 2

72 2

0 3

24 3

48 3

72 3

0 1

24 1

1.82 32.58 18.35 1.74 14.76 69.25 0.76 0.43 3.74 1.32 1.23 2.85 10.33 0.54 1.41 1.64 2.06 0.53 1.19 5.78 1 1.13 1.45 16.71

1.94 34.09 17.88 1.81 14.65 70.38 0.81 0.5 3.72 1.38 1.16 3.09 10.66 0.51 1.25 1.53 1.84 0.47 1.12 4.97 0.89 1.06 1.32 14.95

1.91 33.61 17.75 1.79 15.05 70.11 0.83 0.48 3.83 1.4 1.19 3.09 10.83 0.52 1.27 1.49 1.88 0.49 1.09 5.34 0.91 1.06 1.35 15.39

1.97 33.97 17.88 1.64 14.96 70.42 0.92 0.54 3.97 1.58 1.24 3.54 11.78 0.51 1.16 1.43 1.72 0.48 1.04 4.55 0.8 0.99 1.21 13.89

1.75 30.65 17.22 1.77 14.54 65.94 0.89 0.49 3.94 1.54 1.34 3.35 11.55 0.6 1.56 1.84 2.39 0.58 1.35 6.32 0.97 1.27 1.67 18.55

1.97 33.24 17.96 1.76 14.38 69.31 0.87 0.53 3.79 1.5 1.24 3.28 11.21 0.54 1.31 1.61 1.95 0.48 1.17 5.15 0.88 1.13 1.34 15.55

1.96 33.4 17.35 1.82 14.46 68.98 0.89 0.55 3.8 1.52 1.21 3.24 11.21 0.53 1.31 1.59 2.02 0.48 1.17 5.27 0.96 1.19 1.44 15.96

2.03 34.03 17.35 1.89 14.49 69.8 0.99 0.54 3.84 1.69 1.27 3.69 12.03 0.5 1.19 1.43 1.74 0.45 1.07 4.73 0.79 1.1 1.3 14.3

1.86 31.28 17.29 1.89 13.99 66.31 0.6 1.35 4.11 1.09 0.99 3.8 11.94 0.63 1.53 1.82 2.16 0.55 1.42 5 0.86 1.36 1.46 16.81

2.08 34.25 15.89 1.94 13.92 68.08 1.15 0.66 4.03 1.93 1.26 4.24 13.27 0.52 1.15 1.47 1.72 0.42 1.12 4.4 0.82 1.14 1.31 14.09

2.11 35.15 16.09 1.96 13.94 69.24 1.11 0.63 4.1 1.89 1.26 4.07 13.07 0.51 1.08 1.35 1.72 0.41 1.03 4.16 0.73 1.07 1.31 13.37

2.14 35.19 15.76 1.97 13.62 68.69 1.19 0.63 4.04 2.01 1.29 4.15 13.32 0.51 1.12 1.44 1.73 0.41 1.09 4.24 0.86 1.11 1.31 13.82

1.75 31.79 18.72 1.83 14.21 68.3 0.55 0.65 3.73 1.02 1.02 2.75 9.71 0.56 1.51 1.79 2.21 0.52 1.35 5.79 1.04 1.21 1.64 17.61

1.84 32.58 19.05 1.75 14.39 69.6 0.57 0.55 3.71 1.06 1.02 2.73 9.64 0.54 1.39 1.63 2.04 0.5 1.23 5.62 0.96 1.11 1.51 16.54

1 α-Thujene 2 α-Pinene 3 Sabinene 4 β-Pinene 5 Myrcene High-volatile components 6 α-Terpinene 7 p-Cymene 8 Limonene 9 γ-Terpinene 0 Terpinolene 11 Terpinen-4-ol Medium-volatile components 12 α-Terpinyl acetate 13 Sibirene 14 γ-Elemene 15 α-Humulene 16 trans-b-Farnesene 17 γ-Muurolene 18 Germacrene D 19 σ-Cadinene 20 Germacrene B 21 Germacrene D-4-ol Low-volatile components a

KIE

KIL

924.2 933.2 972.9 974.2 993.6

924 932 969 974 988

1014.2 1022.5 1026.8 1056.3 1086 1175.8

1014 1020 1024 1054 1086 1174

1346.6 1395.7 1425.8 1447.5 1454 1476.8 1485.8 1518.9 1551.7 1572.1

1346 1400 1434 1452 1454 1478 1484 1522 1559 1574

MAHD

Mean value from two experiments. Hydromodul: 1:4 and distillation rate: about 8.5 ± 0.5 mL/min. 406

Industrial Crops & Products 122 (2018) 402–413

M.S. Marković et al.

were adopted: i) Plant particles were isotropic and equal in size, shape and initial JEO content; ii) The JEO was considered a pseudo-single component; iii) The amount of the JEO that could be extracted corresponded to the amount of the JEO distilled off until the system reached equilibrium when no further distillation of JEO occurred; iv) The total amount of the extracted JEO at any time during the distillation, q , equaled to the sum of the JEO amounts extracted from the plant particles (g/100 g) by washing, unhindered diffusion and hindered diffusion, qw , qd1 and qd2 , respectively:

q = qw + qd1 + qd2

(7)

v) The JEO distillation rates for washing, unhindered diffusion and hindered diffusion were exponential function of time, respectively:

Fig. 2. The progress of common JEO distillation by HD and MAHD from the one-minute dry- and wet-ground 24 h-swelled juniper berries (HD, dry grinding: solid line; HD, wet drying: dash line; MAHD, dry grinding: dot line; and MAHD, wet drying: dash dot line) (hydromodul: 1:4 and distillation rate: about 8.5 ± 0.5 mL/min).

d q = α w e−kw | t − τw | dt w

(8a)

d q = αd1 e−kd1 | t − τd1| dt d1

(8b)

and with the progress of distillation. Furthermore, the essential oil yield was higher with MAHD than with HD, which had already been observed for the initial stage of essential oil distillation from juniper berries (Pavićević et al., 2016), rosemary (Karakaya et al., 2014), fresh aerial parts of thyme (Golmakani and Rezaei, 2008b) and black zira (Mazidi et al., 2012) seeds. This was explained by the more effective absorption of microwave irradiation energy than electric thermal energy by water.

d q = αd2 e−kd2 | t − τd2| dt d2

(8c)

where k w , kd1 and kd2 were the distillation rate constants, α w , αd1 and αd2 were maximum distillation rates for washing, unhindered diffusion and hindered diffusion occurring at t = τw, t = τd1 and t = τd2, respectively. Preliminary nonlinear regression determination showed that for HD, the parameter τw was very small, i.e. close to zero. Thus, for HD, it was assumed that τw = 0 . vi) There was no resistance to the mass transfer of JEO from the external surfaces of the plant particles; vii) The water phase and the JEO were completely immiscible; viii) The amount of JEO collected in the separator, divided by the amount of the plant material, was equal to the JEO yield from the plant material in the distillation vessel with a time delay.

3.3. Kinetics of JEO distillation by HD and MAHD 3.3.1. Model development The kinetic model was derived for a batch distillation vessel, where the plant material was immersed in water. The suspension was heated by either conventional electric heater or microwave irradiation, water vapor was produced and carries JEO vapor from the distillation vessel into a condenser. Then, the resulting condensate was gravitationally separated into the JEO and the floral water. The JEO extraction from the ground plant material was assumed to occur through three simultaneous processes. This approach was based on the fact that a major part of the JEO, originated from the destroyed oil organs, was located at external surfaces of the plant particles while the rest of the JEO was uniformly distributed within the plant particles. The essential oil, located at the plant particle surface, was recovered by rapid distillation, called “washing”. This was a high rate of JEO transfer process from the plant particle surface into the liquid phase during the very initial period of the hydrodistillation process. The remaining JEO was extracted through two types of diffusional processes occurring within the plant particles. The first diffusional process, called “unhindered diffusion”, involved the transfer of JEO through ruptured organs, unhindered by membranes or other barriers present in the plant particles. The second diffusional process, called “hindered diffusion”, was the transfer of JEO through membranes of intact, unruptured organs. These diffusional processes became significant after the completion of the washing stage and both were much slower than the washing. Of the two diffusional processes, unhindered diffusion was considered faster. Each JEO distillation mechanism occurred by its own rate, which reached the maximum value after a certain period of time since the beginning of the distillation process. To develop the mathematical model of this process, the following assumptions, some of them already used for modeling the kinetics of HD (Milojević et al., 2008, 2013) and MAHD (Pavićević et al., 2016),

From Eqs. (8a)–(8c), respectively it follows αw

qw =

⎧ kw ⋅e

αd1

qd1 =

−kw τw

⎨ αw ⋅[2 ⎩ kw ⎧ kd1 ⋅e

−

(e kw t − 1), t < τw

e−kw τw

− e−kw (t − τw ) ], t ≥ τw

−k d1 τd1 (e k d1 t

⎨ αd1 ⋅[2 ⎩ kd1

−

e−kd1 τd1

(9a)

− 1), t < τd1 − e−kd1 (t − τd1) ], t ≥ τd1

(9b)

and α d2

qd2 =

⎧ k d 2 ⋅e

−k d2 τd2

⎨ αd2 ⋅[2 ⎩ k d2

−

(e kd2 t − 1), t < τd2

e−kd2 τd2

− e−kd2 (t − τd2) ], t ≥ τd2

(9c)

while the total JEO amounts extracted by washing, unhindered diffusion and hindered diffusion, qw, ∞, qd1, ∞ and qd2, ∞, respectively were as follows:

qw, ∞ =

αw (2 − e−kw τw ) kw

(10a)

qd1, ∞ =

αd1 (2 − e−kd1 τd1) kd1

(10b)

αd2 (2 − e−kd2 τd2) k d2

(10c)

and

qd2, ∞ =

407

Industrial Crops & Products 122 (2018) 402–413

M.S. Marković et al.

Fig. 3. Model I: Contributions and extraction rates of three mass transfer processes occurring simultaneously with the progress of the JEO HD and MAHD from the one-minute dry- (HD – a and b; MAHD – e and f) and wet-ground 24 h-swelled (HD – c and d; MAHD – g and h) juniper berries (washing stage – dash dotted lines, unhindered diffusion – dashed lines, hindered diffusion – dotted lines and total – solid line).

408

Industrial Crops & Products 122 (2018) 402–413

M.S. Marković et al.

Table 4 Values of the kinetic models’ parameters with corresponding standard errors, calculated total JEO amounts (for models I and Ia), R2 and MRPD.a. Model

Parameter

Distillation and grinding method HD

MAHD

Dry grinding

Wet grinding

Dry grinding

Wet grinding

αw , g/100 g/min αd1·102, g/100 g/min αd2 ·103, g/100 g/min kw , min−1 kd1·102, min−1 kd2 ·102, min−1 τw τd1 τd2 qw, ∞

0.199 ± 0.02 2.15 ± 0.87 3.42 ± 1.43 0.167 ± 0.04 5.33 ± 2.03 1.60 ± 1.55b 0 20.0 ± 6.10 132.5 ± 27.32 1.19

0.253 ± 0.008 1.21 ± 0.18 1.80 ± 0.69 0.178 ± 0.01 3.22 ± 1.27 1.70 ± 2.36b 0 25.0 ± 5.36 149.7 ± 52.2 1.43

0.156 ± 0.005 1.12 ± 0.12 1.64 ± 0.59 0.196 ± 0.014 4.86 ± 1.88 1.00 ± 0.80b 9.9 ± 0.10 35.1 ± 4.31 118.5 ± 37.10 1.48

0.228 ± 0.006 0.91 ± 0.11 2.21 ± 0.66 0.257 ± 0.012 4.72 ± 1.66 1.96 ± 1.18b 7.64 ± 0.10 30.0 ± 3.44 119.7 ± 14.39 1.65

qd1, ∞

0.67

0.58

0.42

0.34

qd2, ∞

0.40

0.20

0.29

0.21

(q∞) cal, g/100 g R2 MRPD, %

2.26

2.21

2.18

2.21

0.998 ± 2.1

0.997 ± 0.9

0.999 ± 0.7

0.999 ± 0.8

αw , g/100 g/min αd1·102, g/100 g/min kw , min−1 kd1·102, min−1 τw τd1 qw, ∞

0.169 ± 0.005 4.93 ± 0.61 0.105 ± 0.006 1.07 ± 0.29 0 60 ± 17.42 1.60

0.228 ± 0.008 6.49 ± 0.72 0.143 ± 0.007 1.75 ± 0.37 0 50 ± 9.35 1.60

0.138 ± 0.006 6.37 ± 0.80 0.153 ± 0.013 1.94 ± 0.36 10.03 ± 0.16 50 ± 9.2 1.59

0.210 ± 0.008 5.05 ± 0.54 0.221 ± 0.017 1.69 ± 0.36 7.74 ± 0.16 50 ± 9.33 1.73

qd1, ∞

0.68

0.59

0.53

0.47

(q∞)cal, g/100 g R2 MRPD, %

2.28 0.998 ± 1.8

2.19 0.997 ± 1.5

2.12 0.999 ± 2.3

2.20 0.999 ± 1.6

kw , min−1 kd1·102, min−1 fw q∞, g/100 g R2 MRPD, %

0.118 ± 0.006 1.05 ± 0.17 0.628 ± 0.02

0.175 ± 0.009 1.73 ± 0.15 0.619 ± 0.02

0.059 ± 0.011 0.5 ± 2.82b 0.856 ± 0.21

0.087 ± 0.015 0.85 ± 2.21b 0.860 ± 0.07

2.29 ± 0.04 0.998 ± 1.8

2.19 ± 0.02 0.999 ± 1.2

2.20 ± 0.86 0.978 ± 14.6

2.20 ± 0.27 0.972 ± 10.7

kd1·102,min−1 fw q∞, g/100 g R2 MRPD, %

5.84 ± 0.48 0.061 ± 0.03b

7.46 ± 0.66 0.068 ± 0.04b

5.23 ± 0.38 0b

7.53 ± 0.51 0b

2.02 ± 0.04 0.965 ± 5.8

2.00 ± 0.03 0.960 ± 6.2

2.04 ± 0.04 0.976 ± 14.6

2.08 ± 0.04 0.967 ± 11.5

Model IV: Eq. (15)

kd1·102, min−1 q∞, g/100 g R2 MRPD, %

6.48 ± 0.44 2.00 ± 0.04 0.961 ± 6.8

8.33 ± 0.61 1.99 ± 0.03 0.956 ± 7.0

5.23 ± 0.30 2.04 ± 0.03 0.976 ± 14.6

7.53 ± 0.51 2.08 ± 0.04 0.967 ± 11.5

Model V: Eq. (17)

kd1·102, L g−1 min−1 q∞, g/100 g R2 MRPD, %

3.94 ± 0.18 2.24 ± 0.02 0.994 ± 2.7

5.56 ± 0.23 2.20 ± 0.02 0.995 ± 2.0

3.43 ± 0.65 2.33 ± 0.07 0.944 ± 23.2

4.33 ± 0.61 2.32 ± 0.06 0.954 ± 13.5

Model I: washing + unhindered diffusion + hindered diffusion; Eqs. (7)–(11)

Model Ia: washing + diffusion; Eqs. (7)–(11), τd2 = 0 and kd2 = 0

Model II: Eqs. (12) and (13)

Model III: Eq. (14)

a b

Pretreatment conditions: one-minute grinding and 24 h swelling. Hydromodul: 1:4. Distillation rate: about 8.5 ± 0.5 mL/min. Statistically insignificant.

fw + fd1 = 1

On the basis of Eq. (7), the total JEO amount extracted at the end of distillation, q∞, was defined as follows:

q∞ = qw, ∞ + qd1, ∞ + qd2, ∞

Eqs. (12) and (13) were designated as Model II. Milojević et al. (2013) developed this model for conventional HD. Also, Sovova and Aleksovski (2006) derived the same kinetic expression from a phenomenological model for HD of essential oil from the seeds. This model was verified for the MAHD of JEO from juniper berries (Pavićević et al., 2016). Another two simpler kinetic models, called Model III and Model IV, could be derived from Eq. (12) by assuming instantaneous washing followed by diffusion (k w → ∞) and diffusion with no washing (k w → ∞ and fw = 0), which were presented by the following equations, respectively:

(11)

Eqs. (7)–(11) defined a kinetic model designated as Model I. If τw = τd1 = τd2 = 0 , which meant that the rates of the three extraction mechanisms were maximum at t = 0, hindered diffusion was negligible, i.e. kd2 = 0 , and total diffusion was characterized by the rate constant kd1, then on the basis of Eqs. (7), (8a), (8b), (8c), (10a), (10b), and (10c), the amount of JEO extracted until time t, q , was:

q = q∞ [1 − fw ⋅e−kw ⋅ t − fd1 ⋅e−kd1⋅ t ]

(13)

(12)

q = q∞ [1 − (1 − fw )⋅e−kd1 t ]

where fw = qw, ∞ q∞ and fd1 = qd1, ∞ q∞. It was obvious that 409

(14)

Industrial Crops & Products 122 (2018) 402–413

M.S. Marković et al.

Fig. 4. Model Ia: Contributions and extraction rates of two mass transfer processes occurring simultaneously with the progress of the common JEO HD and MAHD from the one-minute dry- (HD – a and b; MAHD – e and f) and wet-ground 24 h-swelled (HD – c and d; MAHD – g and h) juniper berries (washing stage – dash dotted lines, diffusion – dashed lines and total – solid line).

410

Industrial Crops & Products 122 (2018) 402–413

M.S. Marković et al.

techniques exhibited a maximum rate in the early process stage (up to about 20–25 min and 30–35 min for HD and MAHD, respectively). As it can be seen in Table 4, the rate constants decreased in the following order: k w > kd1 > kd2 , indicating that washing was the fastest. However, the hindered diffusion rate constant was statistically insignificant and hence, this mechanism might be neglected. The model including only washing followed by diffusion (actually, combined unhindered and hindered diffusions), Model Ia, assuming kd2 = 0 and the maximum washing and diffusion rates at t = τw and t = τd1, was applied to describe the variations of JEO yield and distillation rate with time, which can be seen in Fig. 4. All kinetic parameters of this model, presented in Table 4, are statistically significant. A good agreement between the experiment and the Model Ia is illustrated by Fig. 4. Values of the models parameters and the statistical criteria used for assessing the goodness of models fitting (MRPD and R2 ) are presented in Table 4. On the basis of the high coefficient of determination (R2 > 0.94) and the acceptable MRPD (< 24%), all six models could be used for modeling the kinetics of HD and MAHD in the case of both dryand wet-ground juniper berries. Among these six models, the Model Ia involving washing and combined diffusion, Eqs. (7)–(11) with kd2 = 0, appeared to be the best one for both HD and especially MAHD of the juniper berries pretreated in the same way as it had the largest coefficient of determination (R2 > 0.99), statistically significant parameters and low MRPD (less than ± 2.5%). The second best model was the model of simultaneous washing and diffusion (Model II). The worst in this group of models was the simplest exponential model (Model IV) while the second-order model was as well as the other two-parameter Model III. The same was observed for the other four models (Figs. S2–S5, Supplementary material). A good agreement between the experiment and the Models II and III has already been reported for JEO extraction from dry-ground juniper berries by HD (Milojević et al., 2013) and MAHD (Pavićević et al., 2016). Since the Model Ia was based on the physical mechanism of JEO extraction, had all statistically significant parameters and provided small MRPD -value (Table 4), it can be recommended for modeling the kinetics of JEO extraction by both HD and MAHD. For both HD and MAHD, the washing rate constant, k w , and the maximum distillation rate, α w , of the Model Ia were larger for swelled, wet-ground juniper berries than for dry-ground ones. Besides that, k w was larger for MAHD, regardless of the plant material pretreatment method, while α w was less for MAHD than for HD when juniper berries were pretreated by the same method. The parameter τw for MAHD was less for wet-ground than dry-ground juniper berries, which was attributed to less noticeable non-uniform heating in the case of wetground plant material. In the case of the Model II, both washing and diffusion rate constants, k w and kd1, respectively are larger for HD than for MAHD; for the former technique, they have higher values for swelled, wet-ground juniper while for the latter technique, for dry-ground juniper berries. The JEO fraction due to washing, fw , is larger for MAHD than for HD and approximately the same for dry- and swelled, wet-ground juniper berries. Logically, the JEO fraction due to diffusion, fd1, is larger for HD. The values of these three parameters are smaller than those previously reported by Pavićević et al. (2016) for MAHD of dry-ground juniper berries at similar distillation rates. This could be attributed to different origin of juniper berries and hydromodul applied in the present (1:4) and previous (1:3) study. The equilibrium JEO yield calculated by all six models agreed quite well with the experimentally determined ones (MRPD less than about ± 8%) but the overall best agreement was observed for the Model Ia.

and

q = q∞ (1 −

e−kd1 t )

(15)

Eq. (14) is the same as the kinetic expression developed by Milojević et al. (2008) for conventional HD, where fw is the washing coefficient, corresponding to the washable part of the JEO that can be extracted, and kd1 is the coefficient of slow JEO distillation. The model has been verified for the extraction of the JEO by HD (Milojević et al., 2008; Pornpunyapat et al., 2011; Stanisavljević et al., 2010) and MAHD (Kapás et al., 2011; Kusuma and Mahfud, 2017a). Eq. (15) is actually based on the assumption of the pseudo first-order kinetics with respect to the essential oil remaining in the plant material. This model was verified for the kinetics of HD (Morin et al., 1985) and MAHD (Kapás et al., 2011). Besides the above-mentioned kinetic models, the second-order rate of JEO extraction from plant particles has been used sometimes for describing the process kinetics (Model V):

dq = k2 (q∞ − q)2 dt

(16)

where k2 is the second-order rate constant. After integration and transformation, the following expression was derived:

q = q∞

t 1 q∞ k2

+t

(17)

where 1 (q∞ k2) corresponded to the time when q = q∞ 2 . This was a hyperbole meaning that when t → ∞, then q → q∞. This model was verified for the kinetics of MAHD (Kusuma and Mahfud, 2017b,c). 3.3.2. Kinetic modeling of JEO distillation The kinetics of JEO distillation by HD and MAHD was described by five models: - Model of simultaneous washing, unhindered diffusion and hindered diffusion, Eqs. (7)–(11), Model I, assuming the maximum washing, unhindered diffusion and hindered diffusion rates at t = τw, t = τd1 and t = τd2, respectively; - Model of simultaneous washing and diffusion, Eqs. (12) and (13), Model II, assuming the maximum washing and diffusion rates at t = 0; - Model of instantaneous washing followed by diffusion, Eq. (14), Model III, assuming the maximum diffusion rate at t = 0; - Model of diffusion (with no washing), Eq. (15), Model IV, assuming the maximum diffusion rate at t = 0; and - Model of second-order rate law, Eq. (17), Model V, assuming the maximum distillation rate at t = 0. Fig. 3 shows the contributions of the three JEO distillation mechanisms to JEO yield, as well as the variations of distillation rate during HD and MAHD while the kinetic parameters are presented in Table 4. In both cases, all three mass transfer mechanisms could clearly be observed. Washing occurred in the early initial stage (up to 40 min), reaching a maximum JEO yield, which appeared to be higher for MAHD (dry grinding: 1.48 g/100 g and wet grinding: 1.65 g/100 g) than for HD (dry grinding: 1.91 g/100 g and wet grinding: 1.43 g/100 g). The contribution of unhindered diffusion was obviously higher than that of hindered diffusion. Unhindered diffusion was effective in JEO recovery in about first 2–2.5 h of the distillation for both techniques while hindered diffusion became important in the final stage of the distillation. Regarding the JEO washing rate, an important difference between HD and MAHD was observed. Namely, the HD washing rate was the highest at t = 0 while the maximum MAHD washing rate was observed in the very beginning of the distillation (for less than 10 min), which was attributed to the sigmoidal shape of the curve with the transition point in the washing stage. Unhindered diffusion for both distillation

3.3.3. Comparison of the developed model on the basis of AICc AIC and AICc are powerful tools for comparing models with different numbers of parameters, which are developed on the basis of the 411

Industrial Crops & Products 122 (2018) 402–413

M.S. Marković et al.

oil obtained by HD and MAHD is investigated. The type of the applied distillation technique did not significantly influence JEO yield and chemical composition, but the preheating time was about four times less for MAHD than for HD. JEO yield increased with increasing the swelling time and with decreasing the grinding time. Greater contents of the high and medium volatile components were found in the JEOs obtained from the juniper berries swollen for 24 h (HD) or 72 h (MAHD), compared to the JEO from the dry-ground juniper berries. The highest JEO yield (2.23 ± 0.00 g/100 g) was obtained from the oneminute dry-ground juniper berries by HD. Therefore, the optimal pretreatment process involved no swelling and one-minute grinding. A new phenomenological kinetic model involving simultaneous washing, unhindered diffusion and hindered diffusion (Model I) was developed for JEO extraction by both HD and MAHD. On the basis of the corrected Akaike information criterion (AICc), the new developed models (Models I and Ia) were assessed as better than the four existing kinetic models for MAHD of wet- and dry-ground juniper berries, while for HD of wet- and dry-ground berries, the best were models I and II. Since the Model Ia was based on the physical mechanism of JEO extraction and had the largest coefficient of determination (R2 > 0.99), a low MRPD value (less than ± 2.5%) and the statistically significant parameters, it was recommended for modeling the kinetics of JEO extraction by HD and especially MAHD. In addition, this model was verified for extracting essential oil from fennel seeds and cherry laurel leaves by MAHD, indicating its general applicability for modeling the kinetics of essential oil distillation.

Fig. 5. Comparison of the developed kinetic models for the JEO distillation from dry- and wet-ground juniper berries by HD and MAHD on the basis of AICc. Table 5 Verification of Model Ia for the essential oil obtained by MAHD from various plant materials. Parameter

Cherry laurel leaves (Karabegović, 2011)

Fennel seeds, crushed (Kapás et al., 2011)

αw , g/100 g/min αd1, g/100 g/min kw , min−1 kd1, min−1 τw τd1 qw, ∞

0.098 ± 0.02 0.04 ± 0.01 1.00 ± 0.40 0.43 ± 0.08 2.2 ± 0.18 5.0 ± 0.53 0.19

0.269 ± 0.05 0.11 ± 0.02 0.93 ± 0.29 0.51 ± 0.17 1.9 ± 0.12 7.8 ± 0.48 0.53

qd1, ∞

0.18

0.43

(q∞)cal, g/100 g R2 MRPD, % AICc

0.37 0.999 3.45 −84.91

0.96 0.988 6.04 −97.72

Acknowledgment This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.indcrop.2018.06.018. References Adams, R.P., 2007. Identification of Essential Oil Components by Gas Chromatography/ Mass Spectrometry, 4th ed. Allured Publishing Corporation, Carol Stream, IL. Akaike, H., 1973. Information theory as an extension of the maximum likelihood principle. In: Petrov, B.N., Csaki, F. (Eds.), Second International Symposium on Information Theory. Akademiai Kiado, Budapest. pp. 267–281. Amaresh, A., Guha, P., Khan, S., Zari, S.R., 2017. Comparative study of microwave assisted hydro-distillation with conventional hydro-distillation for extraction of essential oil from Piper betle L. Biosci. Biotechnol. Res. Asia 14, 401–407. Burnham, K., Anderson, D., 2002. Model Selection and Multimodal Inference–A Practical Information-Theoretic Approach, 2nd ed. Springer, New York. Busato, N.V., Silveira, J.C., Souza da Costa, A.O., Ferreira da Costa Junior, E., 2014. Modeling strategies for essential oil extraction by hydrodistillation and steam distillation. Ciênc. Rural 44, 1574–1582. Butkienë, R., Nivinskienë, O., Mockutë, D., 2004. Chemical composition of unripe and ripe berry essential oils of Juniperus communis L. growing wild in Vilnius district. Chemija 15, 57–63. Chatzopoulou, P.S., Katsiotis, S.T., 1995. Procedures influencing the yield and the quality of the essential oil from Juniperus communis L. berries. Pharm. Acta Helv. 70, 247–253. Dahmane, D., Dob, T., Chelghoum, C., 2015. Chemical composition of essential oils of Juniperus communis L. obtained by hydrodistillation and microwave-assisted hydrodistillation. J. Mater. Environ. Sci. 6, 1253–1259. Damjanović-Vratnica, B., Skala, D., Baras, J., Petrović-Đakov, D., 2003. A comparison between the oil, hexane extract and supercritical carbon dioxide extract of Juniperus communis L. J. Essent. Oil Res. 15, 90–92. Damjanović-Vratnica, B., Skala, D., Baras, J., Petrović-Đakov, D., 2006. Isolation of essential oil and supercritical carbon dioxide extract of Juniperus communis L. fruits from Montenegro. Flavour Fragr. J. 21, 875–880. Dong, X., Jiang, Z.-T., Jiang, S., Li, R., 2017. Composition comparison of essential oils extracted by hydrodistillation and microwave-assisted hydrodistillation from Petroselinum crispum grown in China. J. Essent. Oil Bear. Plants 20, 368–374. EMA, 2011. Assessment Report on Juniperus communis L., pseudofructus. EMA/HMPC/ 441930/2008, 12 November 2009.. European Medicines Agency, Committee on Herbal Medicinal Products . (Accessed 18 July 2017). http://www.ema.europa.eu/ docs/en_GB/document_library/Herbal_-_HMPC_assessment_report/2011/02/

same dataset, and selecting the best model among the tested set of models. Measuring the relative quality of these models, these criteria, in fact, balances between their complexity and goodness of fit; besides that, AICc involves the correction for the dataset size. However, since n K < 40 (n = 28, K = 2–9), AICc is a more authoritative criterion for comparing the models than AIC. As it can be seen in Fig. 5, where AICcvalues for the two developed and four existing kinetic models are compared, the Models I and Ia were by far the best ones for MAHD of wet- and dry-ground juniper berries, respectively while for HD of wetand dry-ground juniper berries, the best were models I and II, respectively. 3.3.4. Verification of Model Ia The Model Ia was tested with the experimental data for essential oil yield achieved by MAHD of fennel seeds (Kapás et al., 2011) and cherry laurel leaves (Karabegović, 2011). High R2 – values (close to 1) and low MRPD-values (Table 5) proved the Model Ia as successful in describing the kinetics of essential oil distillation from these two plant materials by MAHD. Thus, the Model Ia could be recommended as a general model for describing the kinetics of essential oil distillation from plant materials by MAHD. 4. Conclusion The influence of the common juniper berries pretreatment method on the yield, chemical composition and extraction kinetics of essential 412

Industrial Crops & Products 122 (2018) 402–413

M.S. Marković et al.

hydrodistillation from aerial parts of eleven Artemisia species: chemical compositions and diversities in different geographical regions of Iran. Rec. Nat. Prod. 11, 114–129. Morin, P., Gunther, C., Peyron, L., Richard, H., 1985. Etude des phénomènes physicochimiques intervant lors du procédé d’hydrodillation. Bull. Soc. Chim. Fr. 5, 921–930. Muhammad Hazwan, H., Hasfalina, C.M., Hismamuddin, J., Zurina, Z.A., 2012. Optimization and kinetics of essential oil extraction from citronella grass by ohmic heated hydro distillation. Int. J. Chem. Eng. Appl. 3, 173–177. Nitthiyah, J., Nour, A.H., Kantasamy, R., Akindoyo, J.O., 2017. Microwave assisted hydrodistillation – an overview of mechanism and heating properties. Aust. J. Basic Appl. Sci. 11, 22–29. Pavićević, V.P., Marković, M.S., Milojević, S.Ž., Ristić, M.S., Povrenović, D.S., Veljković, V.B., 2016. Microwave-assisted hydrodistillation of juniper berry essential oil: kinetic modeling and chemical composition. J. Chem. Technol. Biotechnol. 91, 883–891. Pharmacopoeia Jugoslavica, Edition Quarta, 1984. Savezni zavod za zdravstvenu zaštitu, Belgrade, in Serbian. Phutdhawong, W., Kawaree, R., Sanjaiya, S., Sengpracha, W., Buddhasukh, D., 2007. Microwave-assisted isolation of essential oil of Cinnamomum iners Reinw. ex Bl.: comparison with conventional hydrodistillation. Molecules 12, 868–877. Pornpunyapat, J., Chetpattananondh, P., Tongurai, C., 2011. Mathematical modeling for extraction of essential oil from Aquilaria crassna by hydrodistillation and quality of agarwood oil. Bangladesh J. Pharmacol. 6, 18–24. Rezazi, S., Hanini, S., Si-Moussa, C., Abdelmalek, S., 2016. Modeling and optimization of the operating conditions of Marrubium vulgare L. essential oil extraction process: kinetic parameters estimation through genetic algorithms. J. Essent. Oil Bear. Plants 19, 843–853. Rezvanpanah, S., Rezaei, K., Razavi, S.H., Moini, S., 2008. Use of microwave-assisted hydrodistillation to extract the essential oils from Satureja hortensis and Satureja montana. Food Sci. Technol. Res. 14, 311–314. Sourmaghi, M.H.S., Kiaee, G., Golfakhrabadi, F., Jamalifar, H., Khanavi, M., 2014. Comparison of essential oil composition and antimicrobial activity of Coriandrum sativum L. extracted by hydrodistillation and microwave-assisted hydrodistillation. J. Food Sci. Technol. 52, 2452–2457. Sovova, H., Aleksovski, S., 2006. Mathematical model for hydrodistillation of essential oils. Flavour Fragr. J. 21, 881–889. Stanisavljević, I.T., Stojičević, S.S., Veličković, D.T., Ristić, M.S., Lazić, M.L., Veljković, V.B., 2010. Kinetics of hydrodistillation and chemical composition of essential oil from cherry laurel (Prunus laurocerasus var. serbica Pančić) leaves. J. Essent. Oil Res. 22, 564–567. Stanković, M.Z., Veljković, V.B., Lazić, M.L., 1994. Bioactive Products from Juniper Berries (Juniperus communis L.). University of Niš, Faculty of Technology, Leskovac in Serbian. Sugiura, N., 1978. Further analysis of the data by Akaike’s information criterion and the finite corrections. Commun. Stat. A 7, 13–26. Tasić, S., Kovačević, N., Menković, N., Samardžić, Z., Ristić, M., 1993. Comparative studies of juniper berry oils from different regions of central Balkan. Acta Hortic. 344, 574–577. Veljković, V.B., 2014. Comment on “Study of fuel properties of rubber seed oil based biodiesel” [Energy Convers. Manage. 2014;78, 266–275] by Ahmad et al. Energy Convers. Manage. 86, 1186–1188. Veljković, V., Stanković, M., 2003. Herbs used in alcoholic drinks. In: Caballero, B., Trugo, L., Finglas, P. (Eds.), The Encyclopedia of Food Science and Nutrition, 2nd ed. Academic Press, London, pp. 3098–3107.

WC500102144.pdf. EMEA, 1999. Juniperi fructus. Juniperi fructus. Summary Report. European Medicines Agency, Committee for Veterinary Medicinal Products EMEA/MRL/562/99-FINAL.. http://www.ema.europa.eu/docs/en_GB/document_library/Maximum_Residue_ Limits_-_Report/2009/11/WC500014515.pdf . (Accessed 18 July 2017). Golmakani, M.T., Rezaei, K., 2008a. Microwave-assisted hydrodistillation of essential oil from Zataria multiflora Boiss. Eur. J. Lipid Sci. Technol. 110, 448–454. Golmakani, M.T., Rezaei, K., 2008b. Comparison of microwave-assisted hydrodistillation with the traditional hydrodistillation method in the extraction of essential oils from Thymus vulgaris L. Food Chem. 109, 925–930. Hurvich, C., Tsai, C.-L., 1989. Regression and time series model selection in small samples. Biometrika 76, 297–307. Kapás, Á., András, C.D., Dobre, T.Gh., Vass, E., Székely, G., Stroescu, M., Lányi, S., Ábrahám, B., 2011. The kinetic of essential oil separation from fennel by microwave assisted hydrodistillation (MWHD). Univ. Polytech. Bucharest Sci. Bull. Ser. B 73, 113–120. Karabegović, I.T., 2011. The Kinetics of Microwave-Assisted Extraction and Characterization of Bioactive Components from Cherry laurel (Prunus laurocerasus L.). University of Niš, Faculty of Technology, Leskovac in Serbian. Karakaya, S., El, S.N., Karagozlu, N., Sahin, S., Sumnu, G., Bayramoglu, B., 2014. Microwave-assisted hydrodistillation of essential oil from rosemary. J. Food Sci. Technol. 51, 1056–1065. Kusuma, H.S., Mahfud, M., 2017a. Microwave-assisted hydrodistillation for extraction of essential oil from patchouli (Pogostemon cablin) leaves. Period. Polytech.-Chem. Eng. 61, 82–92. Kusuma, H.S., Mahfud, M., 2017b. The extraction of essential oils from patchouli leaves (Pogostemon cablin Benth) using a microwave air-hydrodistillation method as a new green technique. RSC Adv. 7, 1336–1347. Kusuma, H.S., Mahfud, M., 2017c. Kinetic studies on extraction of essential oil from sandalwood (Santalum album) by microwave air-hydrodistillation method. Alexandria Eng. J. http://dx.doi.org/10.1016/j.aej.2017.02.007. in press. Ložienė, K., Labokas, J., 2012. Effects of abiotic environmental conditions on amount and enantiomeric composition of α-pinene in Juniperus communis L. Biochem. Syst. Ecol. 44, 36–43. Ložienė, K., Venskutonis, P.R., 2016. Juniper (Juniperus communis L.) oils. In: Preedy, V.R. (Ed.), Essential Oils in Food Preservation, Flavor and Safety, 1st ed. Academic Press, London, pp. 495–500. Marković, M.S., Bošković-Vragolović, N.M., Ristić, M.S., Pavićević, V.P., Veljković, V.B., Milojević, S.Ž., 2017. Fractionation of the essential oil from juniper (Juniperus communis L.) berries by hydrodistillation and rectification. Hem. Ind. 71, 471–477. Marongiu, B., Porcedda, S., Piras, A., Sanna, G., Murreddu, M., Loddo, R., 2006. Extraction of Juniperus communis L. ssp. nana Willd. essential oil by supercritical carbon dioxide. Flavour Fragr. J. 21, 148–154. Mazidi, S., Rezaei, K., Golmakani, M.T., Sharifan, A., Rezazadeh, Sh., 2012. Antioxidant activity of essential oil from black zira (Bunium persicum Boiss.) obtained by microwave-assisted hydrodistillation. J. Agric. Sci. Technol. 14, 1013–1022. Milojević, S., Stanković, T., Palić, R., Lazić, M., Veljković, V., 2008. Kinetics of distillation of essential oil from comminuted ripe juniper (Juniperus communis L.) berries. Biochem. Eng. J. 3, 547–553. Milojević, S., Radosavljević, D., Pavićević, V., Pejanović, S., Veljković, V., 2013. Modeling the kinetics of essential oil hydrodistillation from plant materials. Hem. Ind. 67, 843–859. Mohammadhosseini, M., 2017. Essential oils extracted using microwave-assisted

413