Assessment of Scale-Up Parameters of Microwave-Assisted Extraction via the Extraction of Flavonoids from Cocoa Leaves

This is an author generated postprint of the article: Chan, C.-H., Yusoff, R. and Ngoh, G.-C. (2015), Assessment of Scale-Up Parameters of Microwave-Assisted Extraction via the Extraction of Flavonoids from Cocoa Leaves.

Chem. Eng. Technol. doi: 10.1002/ceat.201400459

*Corresponding author. Tel: +603 8769 4315; Fax: +603 8925 6197; Email address: [email protected] The published version is available on http://dx.doi.org/10.1016/j.seppur.2015.01.041.

Assessment of Scale-Up Parameter of Microwave-Assisted

Extraction via the Extraction of Flavonoids from Cocoa Leaves

Please email to [email protected] for any inquiries

ABSTRACT:

Microwave-assisted extraction (MAE) is a promising technique for the extraction of flavonoid

compounds from plants. However, it is difficult to be scaled up due to complex mass transfer

involved. This has prompted the study of parameters for scaling up the system by considering

energy-related parameters namely nominal power density and absorbed power density (APD).

Modeling of MAE of flavonoid compounds from cocoa (Theobroma cacao L.) leaves using

film theory model was performed for this purpose. Operating parameters such as particle size

of sample, solvent to feed ratio and microwave irradiation power were also included in the

kinetic study. APD exhibits its potentiality in scaling up as it can characterize both the

extraction kinetics and extraction yields of MAE. Furthermore, it can be used as a reference

for predicting optimum extraction time of MAE at various heating conditions.

Keywords: absorbed power density (APD), optimum extraction time, kinetic modeling, film

theory.

Introduction

Cocoa (Theobroma cacao L.) plants are cultivated globally for the production of cocoa

powders and chocolates. The leaves of the plants are concentrated with bioactive compounds

such as flavonoids [1, 2] and they are normally disposed of during pruning without further

processing to recover the sated valuable compounds. As reported, Cocoa leaves are good source

for active compounds such as quercetin and catechin [1, 3]. These naturally occurring

antioxidants are excellent for the prevention and treatment of diabetes [4-6].

The extraction of active compounds from plants prevails in recent years especially with the

emergence of non-conventional extraction techniques such as microwave-assisted extraction

(MAE). MAE of plant constituents has demonstrated its favorable extraction performance over

conventional techniques in terms of its moderate capital cost and operability under atmospheric

conditions [7-9]. On the other hand, modeling of MAE for scaling up purpose by studying the

effects of parameters on the extraction kinetics enables the prediction of the extraction

behaviour. As a result, various kinetic models derived from Fick’s law [10], chemical kinetic

equations [11] and other empirical models [12] have been developed. However, issues

pertaining to the scaling up of the MAE process have yet to be resolved. These issues are

associated to the reported optimum MAE conditions in literature whereby they are applicable

only for specific microwave systems and restricted scale of extraction. The shortcomings have

prompted the investigation of suitable scale up parameter to describe the kinetics of MAE

process consistently regardless of the scale of extraction. The probable parameters are energy

related such as energy density/nominal power density (NPD) and absorbed power density

(APD). The nominal power density specifies the irradiated microwave power per unit

extraction volume and it is more significant as compared to microwave power level in the

optimization of MAE [13, 14]. It has been applied in the scaling up of MAE process from lab

to pilot scales for the extraction of essential oil [15, 16]. Meanwhile, absorbed power density

(APD), which indicates the absorbed power per unit extraction volume, is capable of describing

the actual heating performance of MAE [17].

The present work investigates the effects of particle size of sample (0.1-0.6 mm), solvent to

feed ratio (20-80 ml g-1) and microwave irradiation power (100-600 W) on MAE of flavonoids

from cocoa (Theobroma cacao L.) leaves using film theory model. Structural analysis of

extracted sample was also performed to substantiate the modeling results. To further determine

the suitability of the scale up parameter of MAE, the effect of applying NPD and APD at larger

scale MAE (100-300 ml) were studied. Based on the evaluation, a method to predict the

optimum extraction time of MAE at various microwave powers and solvent loadings was

proposed.

Materials and Methods

Materials and reagents

Flavonoid standards for isoquercitrin, (-)-epicatechin and rutin were obtained from Sigma-

Aldrich co. (USA). The solvents for chromatography analysis, acetonitrile and ethanol were

purchased from Merck co. (Germany). Denatured alcohol (EtOH) was obtained from LGC

Scientific co. (Malaysia) as extraction solvent.

Extraction procedures

Fresh cocoa leaves were obtained from local cocoa plantation in Pahang, Malaysia. The

collected leaves were dried in an air-drying oven at 40 oC for a day to reduce its moisture

content to 5-6%. The dried leaves were then powdered to various particle sizes. The desized

sample was stored at 4 oC in a storage container prior to extraction.

MAE was performed by using domestic microwave oven (nominal power 100-800 W)

equipped with fiber optic Luxtron I652 thermometer. Cocoa leaves powder (1-6 g) was mixed

with 85% (v/v) aqueous ethanol (optimum solvent) at varying volumes of 100-300 ml in a 500

ml closed Duran bottle. The mixture was placed in the microwave oven and heated up at

predetermined power without stirring. Upon completion of the extraction, the mixture was

transferred to a water bath to be cooled down to room temperature. The extract was filtered

using fine cloth followed by 0.2 µm regenerated cellulose filter prior to HPLC analysis. To

construct an extraction curve of MAE, extractions at predetermined operating conditions with

varying extraction times using fresh samples were conducted. The extraction curve enables the

investigation of the effects of extraction parameters on the overall extraction trend of MAE i.e.

Y/Ysat ; where Y is the extraction yield at certain extraction time and Ysat is the equilibrium

extraction yield.

For comparison purpose, dried cocoa leaves sample was extracted using 200 ml pure ethanol

for 6 hours via Soxhlet extraction. The same clean up procedure applied in MAE was adopted

for chromatography analysis in this extraction.

HPLC analysis

The flavonoid compounds were quantified using Agilent 1200 Series HPLC device with

Agilent ZORBAX Eclipse Plus C18 column, 5 µm (4.6 mm × 150 mm). The analytical method

proposed by Bonaccorsi et al. [18] was adopted. The mobile phase is acetonitrile and water in

linear gradient: 5–20% (0–15 min), 20–30% (15–20 min), 30–50% (20–30 min), 50–100%

(30–35 min), 100% (35–40 min), and 100–5% (40–50 min) at 1.0 ml min-1. The flavonoid was

analysed at 350 nm and 280 nm respectively for isoquercitrin and rutin; and (-)-epicatechin

using UV-DAD detector. The total extraction yield (mg g-1) is expressed as the mass of all

three extracted active compounds per mass of sample used.

SEM examination

To investigate the morphological structure of the sample after MAE and Soxhlet extraction,

the samples after extraction were dried before subject to SEM analysis. The samples were

examined using Field Emission Scanning Electron Microscope (FE-SEM AURIGA, ZEISS)

under high vacuum condition and an accelerating voltage of 1 kV at magnification of 500

and1000.

Determination of absorbed power density (APD)

Absorbed power density (APD) indicates the absorbed microwave power in a unit volume of

solvent (W ml-1). The details in determining the APD can be found in the previous work [17].

The APD values of blank extraction solvent under various microwave irradiation power and

solvent loading can be determined experimentally based on Eq. (1).

H

QAPDV t

=⋅

Eq. (1)

where Q is the amount of energy absorbed by solvent during heating (J), V is the solvent loading

(ml) and tH is the heating time (min). The total heat absorbed (Q), can be determined from the

temperature profile of the solvent using Eq. (2).

L p V vapQ m C T m H= ∆ + Eq. (2)

where mL is the mass of solvent, Cp is the heat capacity of solvent, ΔT is the temperature

difference after heating for specific time tH, mv is the mass of vaporized solvent, Hvap is the heat

of vaporization of solvent. The first term on the right hand side of Eq. (2) indicates the heat

required to heat up the solvent to its boiling point while the second term denotes the latent heat

of vaporization. In this study, a representative value of APD of blank solvent under various

extraction conditions, i.e. microwave irradiation power (100-600 W) and solvent loading (100-

300 W) was obtained by averaging the APD values calculated for each conditions at different

heating time (tH) using Eq. (1). The averaged APD values are tabulated in Tab. 1.

Table 1: APD values at different microwave irradiation power and solvent loading Solvent loading, V (ml)

Microwave irradiation power, P (W)

Absorbed power density, APD (W ml-1)

100

100 0.15 ± 0.02 200 0.43 ± 0.03 300 0.93 ± 0.06 600 2.24 ± 0.03

200 200 0.25 ± 0.02 250 250 0.37 ± 0.07 300 300 0.42 ± 0.08 Example of calculation of APD by calorimetric method

MAE operating conditions

Heating time, tH

Total heat absorbed, Q

Absorbed power density, APD

Average APD

(min) (J) (W ml-1) (W ml-1)

100 ml, 100 W 5.00 5266 0.18

0.15 13.00 10796 0.14 27.00 23649 0.15

Kinetic modeling of MAE

The extraction kinetics of MAE of antioxidant compounds from cocoa leaves was investigated

by modeling its extraction trend (Y/Ysat vs. time) using film theory model [19, 20]. The model

describes MAE process in washing and diffusion steps. The washing of active compounds from

the broken cells takes place in a constant and fast extraction rate. In the diffusion step, the

active compounds diffuse from the intact cells (or microwave-ruptured cells) into the solvent.

The expression of the film theory model is as follows:

1 (1 )exp( )sat

Y b k tY

= − − − ⋅ Eq. (3)

where b characterizes the washing step while k characterizes the diffusion step (min-1). The

curve fitting toolbox (version 2.1) in Matlab (Mathworks Inc., USA) was employed to

determine the constants of kinetic models through curve fitting with experimental extraction

curves. The goodness of fit of the model was evaluated based on sum-square error, root mean-

square error and adjusted R-square value.

Experimental design

The experimental design for the kinetic work is tabulated in Tab. 2 in which the equilibrium

extraction yields of MAE (Ysat) were determined by conducting the extraction in triplicate. The

effect of particle size of sample below and above the average thickness of the leaves sample

was first investigated through kinetic modeling (Tab. 2, no. 1 and 3). Following that, the

modeling of MAE was performed to study the effects of solvent to feed ratio of 20-80 ml g-1

(Tab. 2, no. 2-4) and microwave irradiation power of 100-600 W (Tab. 2, no. 3, 5-7).

Subsequently, the feasibility of the proposed parameters as scale-up parameter and the

extraction kinetics were evaluated by performing MAE at larger scales of 150-300 ml (Tab. 2,

no. 3, 8-10) based on nominal power density of 1 W ml-1. In conjunction with that, APD of

these conditions were determined and their effects on MAE performance were too assessed.

Table 2: Experimental design for extraction curves of MAE

a total extraction time.

No. Extraction conditions Total equilibrium extraction yields, Ysat (mg g-1)

1 2 g sample below 0.25 mm size, 100 ml, 100 W, 30 min a 12.60 ± 0.24

2 1.25 g sample at 0.25-0.60 size, 100 ml, 100 W, 30 min a 11.97 ± 0.17

3 2 g sample at 0.25-0.60 size, 100 ml, 100 W, 35 min a 11.24 ± 0.11

4 5 g sample at 0.25-0.60 size, 100 ml, 100 W, 30 min a 10.17 ± 0.07

5 2 g sample at 0.25-0.60 size, 100 ml, 200 W, 19 min a 11.64 ± 0.26

6 2 g sample at 0.25-0.60 size, 100 ml, 300 W, 9 min a 10.96 ± 0.16

7 2 g sample at 0.25-0.60 size, 100 ml, 600 W, 4 min a 10.76 ± 0.06

8 4 g sample at 0.25-0.60 size, 200 ml, 200 W, 35 min a 12.22 ± 0.17

9 5 g sample at 0.25-0.60 size, 250 ml, 250 W, 8 min a 11.97 ± 0.15

10 6 g sample at 0.25-0.60 size, 300 ml, 300 W, 10 min a 11.29 ± 0.21

Results and discussion

Kinetic study of MAE

Based on the film theory model, extraction kinetics of MAE showed that the washing step of

MAE is extremely fast and the associated period is difficult to be determined experimentally

[12, 21]. Therefore, the coefficient b, which indicates the extraction kinetics of washing step,

can only be approximated to the real condition and it was determined by minimizing the fitting

sum square error of the curve fitting of experimental data. As for the coefficient k for the

diffusion step, it is important for the construction of diffusive extraction curve of MAE. In this

study, all the extraction curves of MAE are fitted well with film theory model with adjusted R-

square value greater than 0.91.

The effect of sample particle sizes on the extraction yield of MAE is as detailed in Fig. 1. The

particle size larger than leaf thickness has no significant effect for sample with plate geometry

(e.g. leaf) as the crucial dimension for the diffusion of active compounds is the thickness of the

leaves [22]. This effect might turn significant only if the sample particle size is reduced below

the thickness of the leaf as demonstrated in the figure whereby additional 10% of equilibrium

extraction yields can be achieved. As a result of reducing particle size, more broken cells were

generated and this has promoted the washing of active compounds by the solvent at the

beginning of the extraction as indicated by high coefficient b. With regards to the extraction

kinetics, smaller size of sample improves the extraction rate as indicated by the coefficient kd.

This is due to enhanced surface contact area between the sample and solvent which has

shortened the diffusion path of active compounds [23-25]. Despite the improvement on MAE

kinetics and yields, sample with too small a particle size makes the separation of the extract

from the residue difficult and may incur additional clean up steps [9]. Hence, plant sample with

particle size of 0.25-0.60 mm was selected for subsequent kinetic studies.

Figure 1: Effect of particle size on the extraction kinetics of MAE; MAE conditions: 2 g sample, 100 W and 100 ml of 85% (v/v) EtOH; ● particle sizes < 0.25 mm; ○ particle sizes > 0.25 mm; — fitted curve for particle sizes < 0.25 mm; − − fitted curve for particle sizes > 0.25 mm.

The effect of solvent to feed ratio (20, 50 and 80 ml g-1) on the extraction kinetics of MAE was

investigated at the same solvent volume as shown in Fig. 2. The result shows the general trend

that extraction with higher solvent to feed ratio gives higher extraction yields due to decrease

in the mass transfer barrier [26, 27]. High solvent to feed ratio enhances the extraction yield of

the diffusion step but has no impact on the washing step of MAE. The result agrees with the

kinetic study of MAE of antioxidants from Balm (Melissa officinalis L.) leaves [24]. Also,

extraction with high solvent to feed ratio improves the rate of diffusion of the active compounds

with the greatest coefficient k at 80 ml g-1 as the mass transfer barrier decreased. However, the

coefficient k of 20 ml g-1 is higher than that in 50 ml g-1. This phenomenon is due to the

saturation of the active compounds in the solvent at 20 ml g-1 attributing to its much lower

equilibrium extraction yields as shown in Fig. 2. As a result, shorter extraction time give rise

to higher k value. Similar observation was also reported in the case of ultrasonic extraction of

Extraction time / min

0 5 10 15 20 25 30 35 40

Y/Y sa

t

0.5

0.6

0.7

0.8

0.9

1.0

b (1) k (min-1)

< 0.25 12.60 0.8239 0.1991 0.93 0.001688 0.01453> 0.25 11.24 0.5340 0.1335 0.99 0.002725 0.01846

Root mean-square

Particle size (mm) Ysat (mg g-1)

Extraction constants Adjusted R-square

Sum-square error

oil from tobacco seeds [19]. Considering the economical and feasibility aspects, kinetic studies

were based on solvent to feed ratio of 50 ml g-1.

Figure 2: Effect of solvent to feed ratio on the extraction kinetics of MAE; MAE conditions: sample with particle size of 0.25-0.6 mm, 100 W and 100 ml of 85% (v/v) EtOH; ● 20 ml g-1; ○ 50 ml g-1; ▼80 ml g-1; — fitted curve for 20 ml g-1; − − fitted curve for 50 ml g-1; −·− fitted curve for 80 ml g-1.

The effect of microwave irradiation power (100-600 W) on the extraction kinetics of MAE was

shown in Fig. 3. In general, microwave power has little impact on the washing step of MAE as

there is no significant change in the coefficient b for all the extraction curves. The figure also

shows that high microwave power strongly enhances the kinetics of the diffusion step and about

10 folds increase in the coefficient k is observed when the microwave power changed from 100

W to 600 W. This is because higher microwave power provides larger driving force for MAE

to disrupt the plant matrix so that the active compounds can be eluted to the bulk solvent. This

is further supported by the structure analysis of microwave-treated samples presented in Fig. 4

(a and b). The figure shows the ruptured plant cells and pits that formed on the surface of the

leaves sample attributed to the localized heating of microwave radiation that increased the

Extraction time / min

0 5 10 15 20 25 30 35 40

Y/Y sa

t

0.5

0.6

0.7

0.8

0.9

1.0

b (1) k (min-1)

20 10.17 0.5714 0.1620 0.96 0.0619 0.027850 11.24 0.5340 0.1335 0.99 0.0027 0.018580 11.97 0.5380 0.2067 0.98 0.0024 0.0173

Solvent to feed ratio

(ml g-1)Ysat (mg g-1)

Extraction constantsAdjusted R-square

Sum-square error

Root mean-square error

internal pressure for rupturing the cells [28]. The internal pressure might have been great that

it damaged the surface of the leave and formed channels for rapid dissolution of active

compounds into extraction solvent. Comparing the structure of the Soxhlet extracted sample in

Fig. 4 (c and d) with the dried sample in Fig. 4 (e and f), both structures appeared rather similar

with shrunk cells. The structural change in the extracted sample explains the positive effects

imparted by microwave heating. Besides rupturing of plant cells, increase in microwave power

also enhances the diffusivity of the active compounds as the diffusivity increases with

temperature [23, 29]. Similar finding was shown in the MAE of oil from olive cake [12]. In

this study, though the extraction time of MAE can be shortened to a few minutes at high

microwave power of 300-600 W, less than 10% decrease in the total equilibrium extraction

yields was witnessed. This could be due to thermal degradation of active compounds at high

microwave power as reported by Biesaga [30]. The extraction temperature of MAE at high

microwave power is higher than that at low microwave power when reaching the equilibrium

extraction yield as shown in Fig. 3 in which the temperature exceeded the boiling point of

solvent at about 70oC. At high microwave power heating of 300-600 W, superheating

phenomena [31] might have caused thermal degradation of active compounds and thus the

suitable range of microwave power for the MAE is preferred between 100-200 W.

In this study, film theory model is able to describe the extraction kinetics of MAE under the

effects of various operating parameters. The feasibility of the model is strongly dependent on

the acquisition of experimental data and it is impractical when large extraction scale is

involved. Hence, the extraction behaviors of MAE at larger scales can alternatively be

predicted based on scale-up parameter. The selection of suitable scale up parameter and its

application are elaborated in the following section.

Figure 3: Effect of microwave irradiation power on extraction kinetics of MAE; MAE conditions: 2 g sample with particle size of 0.25-0.6 mm, 100 ml of 85% (v/v) EtOH and 50 ml g-1; ● 100 W; ○ 200 W; ▼ 300 W; △ 600 W; —temperature profile and fitted curve for 100 W; − − temperature profile and fitted curve for 200 W; −·− temperature profile and fitted curve for 300 W; ··· temperature profile and fitted curve for 600 W.

Extraction time / min

0 5 10 15 20 25 30 35 40

Y/Y sa

t

0.5

0.6

0.7

0.8

0.9

1.0

b (1) k (min-1)

100 11.24 0.5340 0.1335 0.99 0.0027 0.0185200 11.64 0.6168 0.2686 0.91 0.0128 0.0400300 10.96 0.5459 0.7800 0.97 0.0024 0.0185600 10.76 0.5011 1.8280 0.94 0.0090 0.0359

Root mean-square error

Microwave power (W)

Ysat (mg g-1)

Extraction constantsAdjusted R-square

Sum-square error

Solv

ent t

empe

ratu

re /

o C

30

40

50

60

70

80

100 W200 W

300 W600 W

Figure 4: Scanning electron micrographs of microwave-treated sample (a and b), sample after Soxhlet extraction (c and d) and dried sample (e and f); Magnification of 1000 for micrographs a, c and e; Magnification of 500 for micrographs b, d and f.

Evaluation of scale-up parameters of MAE

Considering a larger scale MAE i.e. larger solvent loading without external mass transfer

limitation in a well-mixed system, its operating parameters such as particle size of sample and

solvent to feed ratio remain unaffected except for the optimum microwave power. Logically,

as the extraction scale increases, more power should be applied to the system to provide

(a) (b)

(c) (d)

(e)

(f)

sufficient heating energy for the extraction process. Thus, the amount of microwave power for

larger scale MAE of flavonoids from cocoa leaves using nominal power density and absorbed

power density (APD) was studied.

In the investigation, MAE was conducted at larger scales (100-300 ml) at nominal power

density of 1 W ml-1. The modeling result in Fig. 5 shows that MAE at high solvent loading

requires shorter extraction time as compared to that at low solvent loading even though the

same nominal power density was applied as indicated by the change in the coefficient of

diffusion step (k). Theoretically, MAE conducted at larger scales under the same solvent to

feed ratio and the same heating power density would give constant rate of extraction and

extraction yields. However, the change in the extraction kinetics observed suggesting that the

nominal power density might not be a reliable scale-up parameter for MAE. In view of that,

the real heating power of extraction system which is the absorbed power density (APD) of the

large scale MAE and its effect on the extraction kinetics was investigated.

Figure 5: Extraction kinetics of MAE at different solvent loading under nominal power density of 1 W ml-1; MAE conditions: 2 g sample with particle size of 0.25-0.6 mm, solvent of 85% (v/v) EtOH and 50 ml g-1; ● `100 ml (100 W); ○ 200 ml (200 W); ▼ 250 ml (250 W); △ 300 ml (300 W).

Extraction time / min

0 5 10 15 20 25 30 35 40

Y/Y sa

t

0.5

0.6

0.7

0.8

0.9

1.0

b (1) k (min-1)

100 0.15 11.24 0.5340 0.1335 0.99 0.0027 0.0185200 0.25 12.22 0.5611 0.2323 0.99 0.0025 0.0175250 0.37 11.97 0.6333 0.2540 0.92 0.0098 0.0330300 0.42 11.29 0.6090 0.3587 0.97 0.0051 0.0216

Solvent loading

(ml)

Ysat

(mg g-1)

Extraction constantsAdjusted R-square

Sum-square error

Root mean-square error

APD (W ml-1)

Unlike the nominal power density which only serves as an indicator for the power setting of

the microwave extractor employed, absorbed power density (APD) accounts for the real

heating power of extraction system. When the MAE was conducted at larger scale (100-300

ml) based on nominal power density of 1 W ml-1, the APD values of these conditions changed

from 0.15-0.42 W ml-1 (Tab. 1). The increase in APD indicates that the heating power of the

extraction system is improved and as a result, it has enhanced the extraction kinetics of MAE

as previously showed in Fig. 5. In general, APD depends on the dielectric properties of the

extraction solvent. It increases exponentially with nominal microwave power but decreases

with increasing solvent loading as shown in Fig. 6. The effect of APD on the extraction kinetics

of MAE was further illustrated in Fig. 7 whereby the extraction curves of MAE at various

heating conditions were correlated with their respective APD values. Furthermore, APD is able

to characterize the extraction kinetics of MAE regardless of solvent loading and microwave

irradiation power. MAE at high APD values gives shorter extraction time and there is an

optimum extraction time of MAE at every APD values. On the other hand, MAE at low APD

(< 0.5 W ml-1) gives higher extraction yield than at high APD. Again, the decrease in the

extraction yields is likely due to the degradation of active compounds at high heating power as

previously discussed. Based on the APD findings observed in Fig. 7, conducting MAE at larger

scales at same APD value would give consistent extraction kinetics and yields.

From a holistic point of view, APD is a reliable scale-up parameter as it affects and

characterizes both the extraction yields and the optimum extraction time of MAE despite the

heating conditions (microwave power) and extraction scales (solvent loading). The intrinsic

properties of APD make it a useful reference in predicting of the optimum extraction time of

MAE at larger scales.

0.2

0.4

0.6

0.8

1.0

100

150

200

250

300

100150

200250

300

APD

/ W

ml-1

Nomina

l micr

owav

e pow

er / W

Solvent loading / ml

Figure 6: Absorbed power density (APD) of MAE system under the influence of solvent loading and microwave irradiation power.

6

7

8

9

10

11

12

13

0.0

0.5

1.0

1.5

2.02.5

05

1015

2025

Tota

l yie

ld /

mg

g-1

APD / W ml-

1

Extraction time / min

Figure 7: The extraction profiles of MAE under the effect APD. MAE conditions: ● 2 g sample, 100 W, 100 ml; ○ 2 g sample, 200 W, 100 ml; ▼ 2 g sample, 300 W, 100 ml; ■ 2 g sample, 600 W, 200 ml; ✕ 4 g sample, 200 W, 200 ml; ♢ 5 g sample, 250 W, 250 ml; ▲ 6 g sample, 300 W, 300 ml.

Prediction of optimum extraction time of MAE using APD

The correlation between optimum extraction time region (OETR) and APD was established in

accordance with the modeling results presented in section 3.1. Figure 8 depicts the contour of

the extraction yield of MAE. The OETR in this context corresponds to the extraction time that

can achieve 80-95% of the extraction yields during the diffusion step period and it is bounded

by t80% and t95%. This correlation can be used to facilitate the selection of optimum extraction

time of MAE at specific solvent to feed ratio (50 ml g-1 in this case) under the influence of

microwave power and solvent loading at their corresponding APD values. As illustrated in the

figure, OETR of MAE decreases as APD increases and MAE conducted beyond the OETR will

vaporize large amount of extraction solvent (more than 20% of solvent volume) without

enhancing the extraction yields. In this study, the optimum MAE conditions was around 0.25

W ml-1APD value and 15 min extraction time. With reference to the contour plot, MAE

conducted at optimum conditions gives extraction yields more than 12.00 mg g-1 which is

higher than the 11.64 mg g-1 obtained from that of the optimized Soxhlet extraction conducted

as a comparison study. The performance chart of MAE in Fig. 8 not only can be used for MAE

of flavonoids compounds from cocoa leaves sample at any extraction scales but also for

different instrumental setups of microwave system with known APD.

Figure 8: Contour plot of extraction yield of MAE under the effect of APD and extraction time; − − Extraction time required to achieve 50%, 80% and 95% of the extraction yields during the diffusion step. The extraction yield of MAE is presented in mg g-1.

Industrialization of MAE using APD parameter

MAE is known to be efficient in terms of solvent consumption and extraction time as compared

to conventional techniques [32], thus is very promising and can potentially save millions of

operating cost upon commercialization. In this work, the unique intrinsic properties of APD

parameter in terms of characterization of extraction kinetic and prediction of the optimum

extraction time are useful in facilitating the commercialization and industrialization of MAE.

For instances, APD is reliable in indicating the extraction kinetics of MAE hence it can be

served as reference for scaling up and reproducing the extraction using different microwave

setup. It can also be used as performance indicator to deduce the extraction efficiency of a

MAE at any extraction scale and microwave setup by knowing the optimum APD value of the

specific plant extraction. Most importantly, APD is a potential equipment design parameter for

industrial microwave extractor especially in the determination of extraction vessel

configuration and microwave power output as it addresses the important criteria of the design

specifications such as microwave absorption capability and the required power for optimum

extraction. In short, MAE operated based on APD is in accordance with green extraction

principles [33] whereby it offers great operational flexibility in achieving certain extraction

efficiency without engaging excessive heating power, and adaptability for different type of

plant extraction due to the ease of control of the parameter.

Conclusion

Film theory model is able to describe the extraction kinetics of MAE of flavonoid compounds

from cocoa leaves under the effects of various operating parameters. Both the washing step and

diffusion step of MAE can be enhanced by particle size reduction of sample. Higher microwave

heating power enhanced diffusivity and employment of higher solvent to feed ratio ensuring

lower mass transfer resistance improved the equilibrium yield of MAE. From a scaling up

prospective, absorbed power density (APD) is more feasible than nominal power density as it

characterizes the extraction profile MAE. APD describes the heating performance of MAE

system intrinsically and with this unique properties, APD is a good predictive tool for larger

scale extraction and can be applied to accelerate the commercialization of MAE.

Acknowledgement

This work was carried out under the Centre for Separation Science and Technology (CSST),

University of Malaya and financially supported through University of Malaya Research Grant

(UMRG:RP002A-13AET).

References

[1] H. Osman, R. Nasarudin, S.L. Lee, Food Chem. 2004, 86 (1), 41-46. DOI:

10.1016/j.foodchem.2003.08.026

[2] B. Sultana, F. Anwar, Food Chem. 2008, 108 (3), 879-884. DOI:

10.1016/j.foodchem.2007.11.053

[3] H. Osman, S.F. Lam, LWT - Food Sci. Technol. 2005, 38 (4), 315-321. DOI:

10.1016/j.lwt.2004.06.013

[4] M. Vessal, M. Hemmati, M. Vasei, Comp. Biochem. Physiol. C Toxicol. Pharmacol.

2003, 135 (3), DOI: 357-364.10.1016/s1532-0456(03)00140-6

[5] O. Kamiyama, F. Sanae, K. Ikeda, Y. Higashi, Y. Minami, N. Asano, I. Adachi, A.

Kato, Food Chem. 2010, 122 (4), 1061-1066. DOI: 10.1016/j.foodchem.2010.03.075

[6] P. Daisy, K. Balasubramanian, M. Rajalakshmi, J. Eliza, J. Selvaraj, Phytomedic. 2010,

17 (1), 28-36. DOI: 10.1016/j.phymed.2009.10.018

[7] B. Howard, Am Lab 1995, 27 (8) 24-26.

[8] C. Sparr Eskilsson, E. Björklund, J.Chromatogr. A 2000, 902 (1), 227-250. Doi:

10.1016/s0021-9673(00)00921-3

[9] C.-H. Chan, R. Yusoff, G.-C. Ngoh, F.W.-L. Kung, J.Chromatogr. A 2011, 1218 (37),

6213-6225. DOI: 10.1016/j.chroma.2011.07.040

[10] J.G. Gujar, S.J. Wagh, V.G. Gaikar, Sep. Purif. Technol. 2010, 70 (3), 257-264. DOI:

10.1016/j.seppur.2009.08.005

[11] G. Spigno, D.M. De Faveri, J. Food Eng. 2009, 93 (2), 210-217. DOI:

10.1016/j.jfoodeng.2009.01.006

[12] F. Amarni, H. Kadi, Innov. Food Sci. Emerg. Technol. 2010, 11 (2), 322-327. DOI:

10.1016/j.ifset.2010.01.002

[13] M.J. Alfaro, J.M.R. Belanger, F.C. Padilla, J.R.J. Pare, Food Res. Int. 2003, 36 (5), 499-

504. DOI: 10.1016/S0963-9969(02)00198-9

[14] Y. Li, L. Han, R. Ma, X. Xu, C. Zhao, Z. Wang, F. Chen, X. Hu, J. Food Eng. 2012,

109 (2), 274-280. DOI: 10.1016/j.jfoodeng.2011.09.021

[15] A. Filly, X. Fernandez, M. Minuti, F. Visinoni, G. Cravotto, F. Chemat, Food Chem.

2014, 150 (0), 193-198. DOI: 10.1016/j.foodchem.2013.10.139

[16] L. Petigny, S. Périno, M. Minuti, F. Visinoni, J. Wajsman, F. Chemat, Int. J. Mol. Sci.

2014, 15 (5), 7183-7198. DOI: 10.3390/ijms15057183

[17] C.-H. Chan, R. Yusoff, G.-C. Ngoh, Food Chem. 2013, 140 (1–2), 147-153. DOI:

10.1016/j.foodchem.2013.02.057

[18] P. Bonaccorsi, C. Caristi, C. Gargiulli, U. Leuzzi, Food Chem. 2008, 107 (4), 1668-

1673. DOI: 10.1016/j.foodchem.2007.09.053

[19] I.T. Stanisavljević, M.L. Lazić, V.B. Veljković, Ultrason. Sonochem. 2007, 14 (5), 646-

652. DOI: 10.1016/j.ultsonch.2006.10.003

[20] D.T. Velickovic, D.M. Milenovic, M.S. Ristic, V.B. Veljkovic, Biochem. Eng. J. 2008,

42 (1), 97-104. DOI: 10.1016/j.bej.2008.06.003

[21] D. Franco, J. Sineiro, M. Pinelo, M.J. Núñez, J. Food Eng. 2007, 79 (1), 150-157. DOI:

10.1016/j.jfoodeng.2006.01.047

[22] R. Wongkittipong, L. Prat, S. Damronglerd, C. Gourdon, Sep. Purif. Technol. 2004, 40

(2), 147-154. DOI: 10.1016/j.seppur.2004.02.002

[23] M. Cissé, P. Bohuon, F. Sambe, C. Kane, M. Sakho, M. Dornier, J. Food Eng. 2012,

109 (1), 16-21. DOI: 10.1016/j.jfoodeng.2011.10.012

[24] Š.S. Herodež, M. Hadolin, M. Škerget, Ž. Knez, Food Chem. 2003, 80 (2), 275-282.

DOI: 10.1016/s0308-8146(02)00382-5

[25] M. Hojnik, M. Skerget, Z. Knez, LWT - Food Sci. Technol. 2008, 41 (10), 2008-2016.

DOI: 10.1016/j.lwt.2007.11.017

[26] D. Franco, M. Pinelo, J. Sineiro, M.J. Núñez, Bioresour. Technol. 2007, 98 (18), 3506-

3512. DOI: 10.1016/j.biortech.2006.11.012

[27] W. Qu, Z. Pan, H. Ma, J. Food Eng. 2010, 99 (1), 16-23. DOI:

10.1016/j.jfoodeng.2010.01.020

[28] Y. Kong, Y.-G. Zu, Y.-J. Fu, W. Liu, F.-R. Chang, J. Li, Y.-H. Chen, S. Zhang, C.-B.

Gu, J. Food Compos. Anal. 2010, 23 (4), 382-388. DOI: 10.1016/j.jfca.2009.12.009

[29] L. Rakotondramasy-Rabesiaka, J.-L. Havet, C. Porte, H. Fauduet, Sep. Purif. Technol.

2007, 54 (2), 253-261. DOI: 10.1016/j.seppur.2006.09.015

[30] M. Biesaga, J. Chromatogr. A 2011, 1218 (18), 2505-2512. DOI:

10.1016/j.chroma.2011.02.059

[31] F. Chemat, E. Esveld, Chem. Eng. Technol. 2001, 24 (7), 735-744. DOI: 10.1002/1521-

4125(200107)24:7<735::AID-CEAT735>3.0.CO;2-H

[32] T. J. Mason, F. Chemat, M. Vinatoru, Curr. Org. Chem. 2011, 15 (2), 237-247. DOI:

http://dx.doi.org/10.2174/138527211793979871

[33] F. Chemat, M.A. Vian, G. Cravotto, Int. J. Mol. Sci. 2012, 13 (7), 8615-8627. DOI:

10.3390/ijms13078615