- Email: [email protected]
Mass transfer during pressurized low polarity water extraction of lignans from flaxseed meal
Journal of Food Engineering 89 (2008) 64–71
Contents lists available at ScienceDirect
Journal of Food Engineering journal homepage: www.elsevier.com/locate/jfoodeng
Mass transfer during pressurized low polarity water extraction of lignans from flaxseed meal Colin H.L. Ho a,b, Juan E. Cacace a, G. Mazza a,b,* a b
Pacific Agri-Food Research Centre, Agriculture and Agri-Food Canada, 4200 Hwy 97 Summerland, BC, Canada V0H 1Z0 Department of Food Science, University of Manitoba, Winnipeg, Manitoba, Canada R3T 2N2
a r t i c l e
i n f o
Article history: Received 20 July 2006 Received in revised form 3 April 2008 Accepted 8 April 2008 Available online 14 April 2008 Keywords: Extraction Packed bed Kinetic Diffusion Subcritical water Optimization Linum Lignans Phenolics Secoisolariciresinol diglucoside Functional foods Nutraceuticals
a b s t r a c t The feasibility of extracting lignans from flaxseed meal using pressurized low polarity water (PLPW) in a fixed bed extraction cell was evaluated. Optimization experiments were performed using surface response methodology for flow rate, bed depth and solvent to solid ratio (S/S). Maximum yield of lignan was obtained at flow rate of 0.6–2 mL/min, bed depth to internal diameter ratio (L/ID) of 20–25, and S/S ratio of 77–150 mL/g. The effect of flow rate and the mass transfer mechanisms governing the extraction of lignans from flaxseed meal was also evaluated. A diffusion model and a two site kinetic model were used to describe the extraction of lignans from flaxseed meal by PLPW. The two models gave similar quality of data fits. The diffusion model yielded diffusivities ranging from 2 1013 to 9 1013 m2 s1 for extraction of lignans at 180 °C, pH 9 and a 1:1.5 meal to co-packing material ratio. Ó 2008 Elsevier Ltd. All rights reserved.
1. Introduction Safe, environmentally acceptable and effective bioseparation has become a key research need for industrial and chemical engineering industries in the 21st century (Nobel and Agrawal, 2005). Water, an inorganic solvent, exerts high affinity for hydrophilic compounds. As a result, its application in extraction of low polarity compounds is limited (Wongkittipong et al., 2004). Pressurized low polarity water (PLPW) (also referred as superheated water; subcritical water; pressurized hot water), has been widely promoted as environmentally benign alternative to orthodox organic solvents (methanol, acetone, ethanol, hexane) due to its reduced polarity above normal boiling point while maintaining water in the liquid state (Hawthorne et al., 1999). PLPW has working temperatures ranging between 100 and 374 °C (Poliakoff and Licence, 2002; Cacace and Mazza, 2006). Advancement in PLPW extraction is necessary to improve the process efficiency and economic potential. Mathematical modelings involving mass transfer
* Corresponding author. Address: Pacific Agri-Food Research Centre, Agriculture and Agri-Food Canada, 4200 Hwy 97 Summerland, BC, Canada V0H 1Z0. Tel.: +1 250 494 6376; fax: +1 250 494 0755. E-mail address: [email protected] (G. Mazza). 0260-8774/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.jfoodeng.2008.04.003
parameters (flow rate, temperature, pressure, etc.) in food systems are gaining more attention (Welit-Chanes et al., 2005). It is important to develop models for the extraction process where different extraction operating parameters are optimized for process economics. However, such predictions require the understanding of mass transfer mechanisms. The PLPW extraction process usually begins by putting the biomass in a fixed/packed bed inside a cylindrical/tubular extraction cell. The packed sample is in contact with the flowing liquid and the separation process kinetic requires the collection and analysis of solute enriched extracts in a time dependent manner. PLPW is effective in altering sample matrices and displacing solutes from their original binding sites. However, few attempts have been documented to differentiate the relative influence of partitioning thermodynamics and desorption kinetics from the sample matrix on extraction rates and recoveries of phytochemicals from food samples (Kubatova et al., 2002). Despite the increased interest in PLPW, its application in the separation of bioactives from food and the transport mechanisms of PLPW extraction of plant components are not yet well understood. Although the diffusion of the dissolved solute within the solid is usually the rate limiting steps for most botanicals (Schwartzberg and Chao, 1982), partitioning of the solute between the solid
65
Colin H. L. Ho et al. / Journal of Food Engineering 89 (2008) 64–71
Nomenclature Bi C Ci De Do Ea F Ks L Mt M1
Biot number (dimensionless) solute concentration in the extractor at any time during the extraction process (mg/mL) initial solute concentration at the beginning of extraction (mg/mL) effective diffusion coefficient (m2/s) initial diffusivity (m2/s) activation energy for diffusion (kJ/mol) fraction of solute released quickly (dimensionless) mass transfer coefficient (m/s) bed depth (m) total amount of diffusing substance extracted after time t (mg/g) equilibrium solute concentration in solution, maximum amount of solute that can migrate (extracted) after infinite time (mg/g)
matrix and the solvent have been reported as the rate limiting mechanism for subcritical water extraction of essential oil from savory (Kubatova et al., 2002). Therefore, the prevailing mechanism during mass transfer depends on both the solute to be extracted and properties of the extracting solvent. Secoisolariciresinol diglucoside (SDG) is the major lignan present in flaxseed meal. The extraction of SDG and other bioactives from flaxseed meal using PLPW was previously studied with changing temperature, pH, solvent to solid ratio and co-packing materials (Ho et al., 2007). However, the limiting extraction mechanism was not determined. The purpose of this paper is to optimize the extraction of lignans from flaxseed meal in terms of flow rate, S/S ratio and bed depth and to elucidate the mechanisms controlling the mass transfer process of PLPW extraction in packed beds by applying two mass transfer models. 2. Materials and methods The extraction apparatus and analytical techniques used are as described by Cacace and Mazza (2006). Defatted flaxseed meal was obtained from Flora Manufacturing and Distribution Ltd. (Burnaby, BC). Particle size was approximately 0.22 mm in diameter as measured by Ho et al. (2007). A schematic of the extraction system, characteristic dimensions and geometry of the extraction vessel are shown in Fig. 1. The experiments were carried out in 10.6 mm ID (1/2 inch OD) columns packed with flaxseed meal and an appropriate amount of co-packing glass beads (3 mm in diameter) mixed with the meal to achieve a 1:1.5 ratio (w/w) of meal to glass beads as suggested by Ho et al. (2007). All extractions were carried out in one of three stainless steel cylindrical extraction cells of lengths 16, 27 and 30 cm. The cell length was chosen to minimize dead volume in the cell depending on the desired bed depth. The extraction cell was packed and mounted vertically in the oven with solvent flowing from bottom to top. Pressurized water was pumped at the desired constant flow rate, passing through the packed bed once, and extracting the solute in a continuous process. After passing out of the cell, the extracts were cooled, the pressure reduced and the extracts collected in containers. The collection vessel was changed periodically to provide a plurality of collection volumes, thereby separating and individually collecting multiple eluant fractions. Buffered deionized ultrapure water used for extraction was made using 0.25 M sodium carbonate and 0.25 M sodium bicarbonate and was adjusted to pH 9 as recommended by Ho et al. (2007). At the end of extractions, SDG contents
Rg Rr R2 T X Y dp k1 k2 r t u b
universal gas constant, J mol1 K1 (1.987 cal/K mol) radius of solid particle (m) coefficient of determination (dimensionless) temperature (°C, K) independent variables of polynomial regression model dependent variables of polynomial regression model diameter of solid particle (m) rate constant for fast extraction stage (min1) rate constant for slow extraction stage (min1) radial distance from centre of spherical particle (m) extraction time (min) superficial velocity (m/s) regression coefficients of polynomial model (dimensionless)
of the extracts were evaluated by HPLC (Eliasson et al., 2003; Cacace and Mazza, 2006). 3. Experimental design Optimizations of key PLPW extraction conditions were carried out using surface response methodology (Haaland, 1989). The selected experimental design was a three factor five level central composite design for flow rate, bed depth and solvent to solid ratio. The experiment consisted of 18 runs including four replicates at the centre point. Fixed conditions employed in the experiments included a uniform temperature of 180 °C, constant flaxseed meal to co-packing material ratio (1:1.5) and use of pH 9 buffered water (Ho et al., 2007). The actual and coded (inside bracket) values of the factors of the experimental design are given in Table 1. Data were analyzed using the response surface regression (RSREG) procedure of SAS (SAS Institute Inc.) to fit the following second-order quadratic polynomial regression model. Y ¼ b0 þ
3 X i¼1
bi X i þ
3 X i¼1
bii X 2i þ
3 XX
bij X i X j
i
where Y are dependent variables (lignan yields), b0, bi, bii, bij are constant and regression coefficients of the model. Xi and Xj are the independent variables in the model (flow rate, bed depth and solvent to solid ratio). RSREG provides the analysis of variance (ANOVA) and is able to estimate the coefficient parameters of the model, the contribution of each type of effect (linear, quadratic, and cross-product), and the shape of the curve. A goodness of fit test of the model was performed with the regression (REG) procedure by backward elimination to only keep variables significant at the 0.1% level. Plots were made using Sigma Plot software (SPSS Inc., Chicago, IL). To further investigate the effect of flow rate on extraction rate and to evaluate the extraction mechanism model, experimental runs using a ‘‘one-factor-at-a-time” method were carried out at different flow rates. For these experiments, flaxseed meal (3.6 g + 5.5 g glass beads at 14 cm bed depth) was sequentially extracted with pH 9 water at 180 °C at four different flow rates (0.6, 2, 4, 7.4 mL/min). Also, additional extractions of flaxseed meal with co-packing beads at 130, 160, and 190 °C constant temperatures were conducted with pH 9 buffered water at 1 mL/min. The natural logarithm of lignans diffusivities obtained at 130, 160, 180 and 190 °C were plotted as a function of the reciprocal of absolute temperature (1/T) and the energies of activation for lignans were calculated from the slope (Ea/Rg) of the straight lines.
66
Colin H. L. Ho et al. / Journal of Food Engineering 89 (2008) 64–71
Pressure Gauge Temperature Recorder Oven
Back Pressure Regulator Cooling Bath
Extract WaterHeating Coil
Liquid Pump
Extraction Cell
Thermocouple Pressure Relief Valve
Water Reservoir
dp
L
u Fig. 1. Characteristic dimensions and geometry of the packed bed PLPW extraction vessel.
4. Mass transfer models
tion within the diffusion volume changes with respect to time is known as Fick’s second law (Cussler, 1984; Mantell et al., 2002):
4.1. Diffusion model The general conservation equation for one-dimensional nonsteady state diffusion in a spherical particle when the concentra-
oC o2 C 2 oC ¼ De þ ot or2 r or
! ð1Þ
67
Colin H. L. Ho et al. / Journal of Food Engineering 89 (2008) 64–71 Table 1 Central composite experimental design with 3 variables for extraction of lignans at 180 °C with pH 9 buffered water and 1:1.5 meal to co-packing ratio in a 10.6 mm internal diameter cell Run
Flow rate (mL/min)
Bed depth (cm)
Solvent to solid ratio (mL/g)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
2 (1)a 2 (1) 2 (1) 2 (1) 6 (+1) 6 (+1) 6 (+1) 6 (+1) 0.6 (1.68) 7.4 (+1.68) 4 (0) 4 (0) 4 (0) 4 (0) 4 (0) 4 (0) 4 (0) 4 (0)
7 (1) 7 (1) 21 (+1) 21 (+1) 7 (1) 7 (1) 21 (+1) 21 (+1) 14 (0) 14 (0) 2.2 (1.68) 25.8 (+1.68) 14 (0) 14 (0) 14 (0) 14 (0) 14 (0) 14 (0)
39 (1) 115 (+1) 39 (1) 115 (+1) 39 (1) 115 (+1) 39 (1) 115 (+1) 77 (0) 77 (0) 77 (0) 77 (0) 12 (1.68) 142 (+1.68) 77 (0) 77 (0) 77 (0) 77 (0)
a Numbers in parentheses are coded values of variables in the experimental design.
The initial condition is C ðt¼0Þ ¼ C i If surface mass transfer resistance is negligible then the boundary conditions are oC ¼0 or ðr¼rc Þ C ðr¼rb Þ ¼ 0 where C is the solute concentrations (mg/mL) at any location in the particle at time t (s); Ci is the initial solute concentration (mg/mL); De is the mass effective diffusion coefficient (m2/s) assuming that De is constant with the concentration; t is extraction time (s); r is radial distance coordinate from centre of spherical particle (m); rc is the centre of the spherical particle (r = 0); rb is radius of spherical particle (m). The solution to Eq. (1) for a spherical particle assuming negligible external resistance to mass transfer is given by Crank (1975) and Cussler (1984): " # 1 Mt 6 X 1 De n2 p2 t ¼1 2 exp ð2Þ p n¼1 n2 M1 R2r where Mt is the total amount of solute (mg/g) removed from flaxseed meal after time t; M1 is maximum amount (mg/g) of solute extracted after infinite time; Mt/M1 is ratio of total migration to the maximum migration concentration and Rr is the average radius of flaxseed meal particle. When time becomes large, only one term in the series is significant and Eq. (2) becomes ! Mt 6 De p2 t ¼ 2 exp 1 ð3Þ M1 p R2r
fusivity De is estimated from Eq. (2) using Microsoft Excel Solver program (Tutuncu and Labuza, 1996). 4.2. Two site kinetic model The mathematical model proposed by So and MacDonald (1986) and Kubatova et al. (2002) uses two steps to define an extraction curve: a certain fraction (F) of the solutes desorb at a faster rate defined by k1, and the remaining fraction (1 F) desorbs at a slower rate defined by k2, thus:
Table 2 Surface response and ANOVA for lignan yields in extracts Run
Amounta (mg/g)
Yieldb (%)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Model Linear Quadratic Cross-product R2 Effects Flow rate Bed depth Solvent to solid ratio
6.6 17.4 19.1 20.9 5.4 8.7 4.8 17.1 20.7 7.7 2.8 19.4 2.5 17.5 13.4 12.4 8.1 10.9
31.6 83.0 91.2 100.0 25.9 41.5 22.8 81.7 99.2 37.0 13.5 93.0 12.0 86.9 64.2 59.4 38.6 52.3 ***c ***
** ** **
a
Lignan yields in mg per g of flaxmeal expressed as SDG equivalents. Compound yields in weight percentage of total content in flaxseed meal. c ***Significant at 0.01 level, **significant at 0.05 level, *significant at 0.1 level, NS non significant (p > 0.1). b
Table 3 Regression coefficients and analysis of variance of the second-order polynomial model for lignans of flaxseed meal extracts Variablesa
Coefficientc
Intercept X1b X2 X3 X 21 X 22 X 23 X1 X2 X1 X3 X2 X3
19.29NS, d 15.95* 3.99** 0.47 *** 1.52*
Model R2
***
0.35
NS
0.8867
P P P y^ i=l i=l i=l j=i+l Polynomial model Y ¼ b0 þ 3i¼1 bi X i þ 3i¼1 bii X 2i þ 2i¼1 P3 j¼iþ1 bij X i X j adjusted by backward elimination with the goodness-of-fit test at the level of 0.1%. b X1 = flow rate, X2 = bed depth, X3 = solvent to solid ratio. c Lignan yields in mg/g of flaxmeal as SDG equivalents determined by HPLC (Eliasson et al., 2003). d ***Significant at 0.01 level, **significant at 0.05 level, *significant at 0.1 level, NS non significant (p > 0.1). a
Two methods were used to determine the effective diffusion coefficient. The first method is the linear (graphical) solution from which 2 De can be determined from the slope Slope ¼ pRD2 e of the linear r portion of the ln (1 Mt/M1) vs. time plot (Dibert et al., 1989). The second method is non-linear regression of Eq. (2) with effective diffusivity (De) as the fitting parameter. The effective dif-
NS NS 0.8927
68
Mt ¼ 1 ½Fek1 t ½ð1 FÞek2 t M1
Colin H. L. Ho et al. / Journal of Food Engineering 89 (2008) 64–71
ð4Þ 100
5. Results and discussion The models developed by surface response analysis for yields of lignans (p < 0.01) were significant (Table 2). The coefficients of determination, R2, indicated the degree of fit of the second-order regression models were high for lignans (R2 = 0.89). Regression coefficients and analysis of variance of the adjusted polynomial second-order models for lignans are presented in Table 3. The coefficients of solvent to solid ratio and bed depth were positive for lignans implying that higher levels of solvent to solid ratio and bed depth would result in higher recovery of the target compounds in the extracts. The order of significance of process variables affecting flaxseed meal extraction can be ranked as follows: solvent to solid ratio > bed depth > flow rate.
80
SDG Yield %
where t is extraction time (min); F is the fraction of the solute released quickly; (1 F) is the fraction of the solute released slowly; k1 is the first-order rate constant describing the fast released fraction (min1); and k2 is the first-order rate constant describing the slow released fraction (min1). The Excel solver regression routine was used to fit data to Eq. (4). The fitted parameters are F, k1 and k2.
60
40
20 0 140 lv 120 en 100
So
tt
o
so 80 lid 60 ra tio 40 (m L/ 20 g)
7
6 5 4 3 2 1
ate ow r
)
/min
(mL
Fl
constant bed depth 14 cm
5.1. Effect of solvent to solid ratio
5.2. Effect of bed depth The effect of bed depth was significant for yields of lignan (p < 0.05) (Table 2). Yield percentages for SDG (Figs. 2B and 3) increased when raising the bed depth from 2 cm to 26 cm regardless of the S/S selected. The high bed depth studied (21–26 cm) gave higher yields for SDG (93–100%) extraction. The gain in recovery for bed depth was higher (steeper slope) (>50%) than the gain obtained by S/S (shallow slope) (<40%). This confirms that the use of a
120
100
SDG Yield (%)
Solvent to solid (S/S) ratio is the ratio of the amount of extraction fluid used to initial mass of starting material. Fig. 2A and B show that SDG extraction yield increased with S/S ratio regardless of flow rate and bed depth, respectively. The optimum S/S ratio to reach the maximum SDG yield was about 110 mL/g with 2 mL/min flow at a 14 cm bed depth or 120 mL/g for a bed depth of 20 cm at 4 mL/min. A larger S/S (>140 mL/g) ratio was required to improve recovery of SDG until equilibrium was reached when flow rate was higher than 4 mL/min at 14 cm bed depth (Fig. 2A). Similarly, at a uniform flow of 4 mL/min, a large S/S ratio was necessary to expedite recovery when the bed depth was less than 20 cm (Fig. 2B). However, the S/S ratio could be reduced with a lower flow rate and higher bed depth without affecting the yield. Thus a decrease of S/S from 110 mL/g to 40 mL/g did not affect SDG yield when flow rate of 2 mL/min was used on a bed depth of 21 cm (9 min residence time) (Fig. 3). A similar degree of S/S reduction resulted in a decrease of yield from 80% to about 40% in the two extraction conditions with a residence time of 3 min. Results in Table 3 shows that the linear effect of S/S was significant for SDG (p < 0.01). Neither quadratic nor cross-product interaction effects with S/S were detected. This was consistent with response surface plots of SDG which demonstrated yields increased almost linearly with S/S (Fig. 2). An increase in S/S would favor extraction by modifying the concentration gradient between the solution in the extraction cell and the surface of the sample matrix. Yields were also higher by increasing the solvent to powder ratio in the extraction of anthocyanins from sunflower husks (Pifferi and Vaccari, 1983). Use of high solvent to solid ratios, however, results in dilute solutions.
80 60 40 20 25 20
0 140
)
15
120
Sol
100
ven
t to
h
10
80
soli
d ra 60 tio ( 40 mL /g)
5
d
Be
t ep
(m
d
20
constant flow rate 4 mL/min Fig. 2. Effect of flow rate (A) and bed depth (B) and solvent to solid ratio on extractions of SDG from flaxseed meal at 180 °C using pH 9 buffered water with 1:1.5 meal to glass beads ratio.
larger bed depth can reduce the use of water due to the need for a lower S/S ratio. Lignan yields in the extracts increased with bed depth (Fig. 3). There were significant differences in yields observed for extractions with different residence times (Fig. 3). The differences in yields and rates of extraction can be attributed to the difference in residence times (t), which defines the contact time of water with meal particles. The residence time is calculated as a function of bed depth and superficial velocity (t = L/u) where L is bed depth and u is the superficial velocity. Fig. 3 shows two curves at a constant bed depth of 7 cm with different flow rates. The curve at 2 mL/min gave >60% more SDG than a flow of 6 mL/min at the same S/S ratio. Extraction at 2 mL/min, again, enabled higher SDG recovery than 6 mL/min at the same bed depth of 21 cm (Fig. 3). Therefore,
69
120
120
100
100
80
80
SDG yield (%)
SDG Yield (%)
Colin H. L. Ho et al. / Journal of Food Engineering 89 (2008) 64–71
60
40 7cm 6mL/min (residence time 1min) 7cm 2mL/min (residence time 3min) 21cm 6mL/min (residence time 3min) 21cm 2mL/min (residence time 9min)
20
60
40 0.6mL/min 2mL/min 4mL/min 7.4mL/min
20
0
0 0
20
40
60
80
100
120
140
0
100
changing the bed depth from 7 cm to 21 cm had a significant effect on both the extraction rate and total recovery of SDG from flaxseed meal. Raising the bed depth from 7 to 21 cm increased SDG yields by 20–40% and extraction time was reduced by about 50–80% at both flow rate conditions. The highest SDG yield (100%) was obtained at a residence time of 9 min while the lowest yield (40%) was observed at 1 min residence time (Fig. 3). In addition, the run at 21 cm and 2 mL/min (9 min residence time) required the least S/S ratio to reach extraction equilibrium; whereas runs with 3 min residence time obtained by combination of 2 bed depths and 2 velocities produced almost identical yields utilizing the same volume of S/S ratio (Fig. 3). Longer residence time allows the solvent sufficient time to penetrate through the solid pores. Liu and Wyman (2003) also proposed that longer residence time was responsible for enhanced removal of solutes from biomass using compressed hot water. 5.3. Effect of flow rate Fig. 2A shows enhanced yields of extracted solutes with declining solvent flow rate. Flow rate was significant for SDG (p < 0.05) (Table 2). SDG showed large percentage increase in yields (40%) with reduction of flow rate at any given S/S (Fig 2A). The combination of low flow rate (1–2 mL/min) and low S/S provides SDG yields higher than those at high flow (6–7 mL/min) rate and low S/S. High flow rate coupled with high S/S (>100) could produce similar yields to those at low S/S and flow rate; however, large S/S leads to huge solvent consumption (diluted extract) which is undesirable. Therefore, a flow rate between 1 and 2 mL/min may be preferred. Kinetics of extraction were similar for the four flow rates studied (0.6, 2, 4, 7.4 mL/min) as shown by the yield of lignan SDG against time (Fig. 4). Thus it is likely that mass transfer is controlled by the internal diffusion in the flaxseed meal and lower dependency on external mass transfer between the solid surface and bulk liquid phase which is consistent with a negligible effect of increased flow rate (Kubatova et al., 2002). More recently, Cacace and Mazza (2006) reported that permeability within flaxseed (diffusion) exhibited greater resistance than external mass transfer, and the optimal flow rate for the extraction of lignans from whole flaxseed with PLPW was 0.5 mL/min in the tested range of 0.3–4 mL/min. Shotipruk et al. (2004) found that extracted anthraquinone concentration was higher for lower flow rates. Thus, high flow rate does not always speed up mass transfer depending on internal structure of the food matrix and the extractive capacity
300
400
500
Fig. 4. Effect of flow rate on PLPW extraction of SDG from flaxseed meal at constant bed depth 14 cm (3.6 g meal + 5.5 g glass beads) with pH 9 buffered water at 180 °C.
of the solvent (Pinelo et al., 2006). In agreement with our findings from the effects of flow rate, calculations of the Biot number Bi ¼ KDser resulted in values from 11 to 47. These values are higher than 10 which indicate that diffusion within the solid controls the process. 5.4. Model fitting of experimental data Example plots of extraction data at flow rates 2 and 6 mL/min and fitted regression lines are displayed in Fig. 5. Effective diffusion coefficients were calculated from the plots. The increase in extraction rate with temperature reflects the increase of diffusivity values. The values of diffusivity (both linear and non-linear) ranged from 1.4 to 15.7 1013 m2/s for lignan (Table 4). Fig. 6 shows the fitted curves for extraction yields of lignan SDG at three different temperatures using the two site kinetic model. The effect of temperature on the extraction of lignan indicated the maximum recoveries was reached at around 160–190 °C. The plots show that the values calculated from the model adequately match the experimental values (R2 > 0.96). Elevated temperatures were found to enhance extraction rates and reduced extraction time from 420 min at 130 °C to 100 min at 190 °C. About 80–90% of lignans were extracted in less than 60 min extraction time at 190 °C (Fig. 6).
0.2 experimental data 6mL/min regression line experimental data 2mL/min regression line
0.0 -0.2
ln (1-Mt/M∞)
Fig. 3. SDG yield as a function of solvent to solid ratio for extraction at 180 °C, pH 9 for two bed depths and two flow rates. Bed depth 7 cm (1.8 g meal + 2.7 g glass beads); 21 cm (5.5 g meal + 8.2 g glass beads).
200
Time (min)
Solvent to Solid Ratio (mL/g)
-0.4 -0.6 -0.8 -1.0 -1.2 0
1000
2000
3000
4000
Time (s) Fig. 5. Application of linear solution using Fick’s law for SDG extraction at flow rate 2 and 6 mL/min, respectively with pH 9 PLPW at bed depth 7 cm at 180 °C.
70
Colin H. L. Ho et al. / Journal of Food Engineering 89 (2008) 64–71
Table 4 Effective diffusion coefficients for lignans extracted with PLPW at different temperature and pH with a fixed meal to co-packing glass beads ratio of 1:1.5 using 210 solvent to solid ratio at water flow rate of 1 mL/min at 10 cm bed depth Temp (°C)
pH
130 160 190 130 160 190
9 9 9 4 4 4
Linear
Non-linear
De 1013 (m2 S1)
De 1013 (m2 S1)
2 5.3 15.7 1.4 3.2 14.3
1.4 3.5 10.8 1.1 2.6 10.1
An increase in temperature causes a decrease in water viscosity and dielectric constant and therefore an increase in diffusion (Ong et al., 2006). The increase in diffusivity with increased temperature may also be caused by an increase of the internal energy of the molecules and thus their mobility. This is consistent with the findings of Yang et al. (1998) who showed that increasing temperature favored PLPW extraction by increasing the diffusion coefficient of organic solutes, which accelerated the rate of mass transfer and reduced extraction time. An increase in temperature from 160 to 190 °C decreased extraction time from 250 min to 100 min for SDG (Fig. 6). However, the problem with the use of very high temperature is that all reactions are accelerated, including all the unwanted side reactions and
A
120
SDG Yield %
100
80
60
130°C experimental 160°C experimental 190°C experimental 130°C two site kinetic model 160°C two site kinetic model 190°C two site kinetic model
40
20
0 0
100
200
300
400
500
Time (min) 120
B
SDG Yield %
100
80
60
2mL/min experimental 6mL/min experimental 6mL/min two site kinetic model 2mL/min two site kinetic model
40
20
0 0
100
200
300
Time (min) Fig. 6. Experimental fitting of two site kinetic model to SDG recovery data obtained at various temperatures (A) and at two different flow rates at fixed bed depth 21 cm (B) with meal to co-packing ratio 1:1.5 using pH 9 buffered water.
Table 5 Predicted equilibrium concentrations and kinetic coefficients obtained by fitting two site kinetic model to PLPW extraction of SDG data at 1 mL/min from 2 g flaxseed meal pH
Packing (g)
Equilibrium conc
Kinetic coefficient min1
R2
Stage 1
Stage 2
Stage 1
F*a
1 F*b
k1 104
Temp. 130 °C, 210 mL/gS/S 4 (1) 0 (1) 0.2 4 (1) 3.0 (+1) 0.3 9 (+1) 0 (1) 0.1 9 (+1) 3.0 (+1) 0.2
0.8 0.7 0.9 0.8
10 52 4 7
43 58 28 36
0.93 0.94 0.92 0.91
Temp. 160 °C, 210 mL/gS/S 4 (1) 0 (1) 0.3 4 (1) 3.0 (+1) 0.4 9 (+1) 0 (1) 0.4 9 (+1) 3.0 (+1) 0.5
0.7 0.6 0.6 0.5
34 114 114 118
61 84 84 133
0.94 0.96 0.96 0.98
Temp. 190 °C, 210 mL/gS/S 4(1) 0 (1) 0.9 4 (1) 3.0 (+1) 1.0 9 (+1) 0 (1) 1.0 9 (+1) 3.0 (+1) 0.9
0.1 0.0 0.0 0.1
505 564 503 498
165 822 737 145
0.98 0.97 0.94 0.96
a b c d
Stage 2 c
k2 104
d
F (dimensionless) fraction of solute released quickly. 1 F is the fraction of the solute released slowly. k1 (min1) is the first-order rate constant for the quickly released fraction. k2 (min1) is the first-order rate constant for the slowly released fraction.
degradations (Wongkittipong et al., 2004). Therefore, the optimum temperature selected for extracting lignan SDG from flaxseed meal was 180 °C in the present study. The augmentation effect of high pH on lignans extraction was discussed in Ho et al. (2007). Extraction time increased as temperature decreased and the rise was more noticeable at acidic pH 4. At the alkaline pH 9, the favorable effect of pH on extraction yield might have compensated for the reduction of extraction rate caused by lower temperature, preventing a larger increase in extraction time. The lowest diffusivity values were obtained for lignan at low temperatures and pH (Table 4), which further indicates that the times required for extractions at low temperature and pH were longer due to reduced diffusivities. At 190 °C, equilibrium fraction of the fast extracting stage F was higher compared to the equilibrium fraction of the slow stage (1 F) (Table 5). The pH and co-packing glass beads had no apparent effect on F values at 190 °C. In addition, both k1 and k2, increased sharply with temperature. At 160 °C, both fast and slow fractions were equally important. The kinetic constants at 160 °C demonstrated moderate extraction rates. Extraction rate was improved by either increasing pH or addition of packing glass beads. At low temperature (130 °C), the fast extracting stage was not important as the F values were small (Table 5). The effective diffusivity values were fitted to the Arrhenius type equation: Ea De ¼ D0 exp Rg T where De and Do are the effective diffusivity and initial diffusivity, respectively, Ea is the energy of activation (kJ/mol) for diffusion, Rg is the universal gas constant (kJ/mol K), and T is the absolute temperature (K). Fig. 7 shows the ln De vs. 1/T plot. Activation energies for SDG were 51 and 56 kJ/mol at pH 9 and pH 4, respectively. Cacace and Mazza (2003) reported activation energy values of 70– 97 kJ/mol for diffusion of phenolic from berries. Spiro and Selwood (1984) reported activation energy for caffeine diffusion through coffee beans and tea leaf to be 32 kJ/mol and 23 kJ/mol, respectively. Activation energy can be reduced due to the lower energy barrier to initiate diffusion since thermal energy can overcome solute–sol-
Colin H. L. Ho et al. / Journal of Food Engineering 89 (2008) 64–71
-27.0
-27.5 2
R =0.97
Ln D
-28.0
-28.5
-29.0
-29.5
-30.0 0.0021
0.0022
0.0023
0.0024
0.0025
1/T (1/K) Fig. 7. Arrhenius-type relationship between effective diffusivity and temperature for SDG using 1 mL/min pH 9 buffered water with 1: 1.5 meal to co-packing material ratio and S/S ratio 210 mL/g.
ute and solute–matrix interactions by decreasing activation energy in the extraction process (Richter et al., 1996). 6. Conclusion The experimental data from PLPW extraction of lignans from flaxseed meal subjected to various processing conditions (temperature, pH, solvent to solid ratio, co-packing materials, flow rate, bed depth) were accurately described using both a two site kinetic model and a diffusion model. High temperature (180°), high pH (9) with 1:1.5 meal to co-packing glass beads resulted in a more rapid extraction by increasing diffusivity and reducing activation energy. Diffusivities attained from the diffusion model ranged from 2 1013 to 9 1013 m2 s1, for rapid extraction of lignans under the mentioned conditions. Flow rate had little influence on extraction kinetics. PLPW is a promising technique that enables production of biobased products in an environmentally sustainable manner without the use of toxic organic solvents. Acknowledgment We thank Lana Fukumoto for her assistance with the HPLC analysis of lignans, and Rod Hocking for manufacturing the extraction cells. We are also grateful to CBIN, Natural Resources Canada for financial support of this research. References Cacace, J.E., Mazza, G., 2006. Pressurized low polarity water extraction of lignans from whole flaxseed. Journal of Food Engineering 77, 1087–1095.
71
Cacace, J.E., Mazza, G., 2003. Mass transfer process during extraction of phenolic compounds from milled berries. Journal of Food Engineering 59, 379–389. Crank, J., 1975. The Mathematics of Diffusion. Clarendon Press., Oxford, England. Cussler, E.L., 1984. Diffusion: Mass Transfer in Fluid System. Cambridge University Press, Cambridge, UK. pp. 146–177. Dibert, K., Cros, E., Andrieu, J., 1989. Solvent extraction of oil and chlorogenic acid from green coffee Part II: Kinetic data. Journal of Food Engineering 10, 199–214. Eliasson, C., Kamal-Eldin, A., Andersson, R., Aman, P., 2003. High-performance liquid chromatographic analysis of secoisolariciresinol diglucoside and hydroxycinnamic acid glucosides in flaxseed by alkaline extraction. Journal of Chromatography A 1012, 151–159. Haaland, P.D., 1989. Experimental Design in Biotechnology. Marcel Decker Inc, New York, NY. pp. 189–249. Hawthorne, S.B., Miller, D.J., Yang, Y., Lagadec, A.J.M. 1999. Method of manipulating the chemical properties of water to improve the effectiveness of a desired chemical process. US Patent# 6,001,256. Ho, C.H.L., Cacace, J.E., Mazza, G., 2007. Pressurized low polarity water extraction of lignans from flaxseed meal. Lebensmittel Wissenschaft und Technologie 40, 1637–1647. Kubatova, A., Jansen, B., Vaudoisot, J.F., Hawthorne, S.B., 2002. Thermodynamic and kinetic models for the extraction of essential oil from savory and polycyclic aromatic hydrocarbons from soil with hot (subcritical) water and supercritical CO2. Journal of Chromatography A 975, 175–188. Liu, C., Wyman, C.E., 2003. The effect of flow rate of compressed hot water on xylan, lignin, and total mass removal from corn stover. Industrial and Engineering Chemistry Research 42, 5409–5416. Mantell, C., Rodriguez, M., De La Ossa, E.M., 2002. Semi-batch extraction of anthocyanins from red grape pomace in packed beds: experimental results and process modelling. Chemical Engineering Science 57, 3831–3838. Nobel, R.D., Agrawal, R., 2005. Separations research needs for the 21st century. Industrial and Engineering Chemistry Research 44, 2887–2892. Ong, E.S., Cheong, J.S.H., Goh, D., 2006. Pressurized hot water extraction of bioactive or marker compounds in botanicals and medicinal plant materials. Journal of Chromatography A 1112, 92–102. Pifferi, P.G., Vaccari, A., 1983. The anthocyanins of sunflower. II. A study of the extraction process. Journal of Food Technology 18, 629–638. Pinelo, M., Sineiro, J., Nunez, M.J., 2006. Mass transfer during continuous solid– liquid extraction of antioxidants from grape byproducts. Journal of Food Engineering 77, 57–63. Poliakoff, M., Licence, P., 2002. Reinventing the chemical industry. Nature 419, 880. Richter, B.E., Jones, B.A., Ezzell, J.L., Porter, N.L., 1996. Accelerated solvent extraction: a technique for sample preparation. Analytical Chemistry 68, 1033–1039. Schwartzberg, H.G., Chao, R.Y., 1982. Solute diffusivities in leaching processes. Food Technology 36, 73–86. Shotipruk, A., Kiatsongserm, J., Pavasant, P., Goto, M., Sasaki, M., 2004. Pressurized hot water extraction of anthraquinones from the roots of Morinda citrifolia. Biotechnology Progress 20, 1872–1875. So, G.C., MacDonald, D.G., 1986. Kinetic of oil extraction from canola (rapeseed). Canadian Journal of Chemical Engineering 64, 80–86. Spiro, M., Selwood, R.M., 1984. The kinetics and mechanism of caffeine infusion from coffee: the effect of the particle size. Journal of the Science of Food and Agriculture 35, 915–924. Tutuncu, M.A., Labuza, T.P., 1996. Effect of geometry on the effective moisture transfer diffusion coefficient. Journal of Food Engineering 30, 433–447. Welit-Chanes, J., Vergara-Balderas, F., Bermudez-Aguirre, D., 2005. Transport phenomena in food engineering: basic concepts and advances. Journal of Food Engineering 67, 113–128. Wongkittipong, R., Prat, L., Damronglerd, S., Gourdon, C., 2004. Solid–liquid extraction of andrographolide from plants – experimental study, kinetic reaction and model. Separation and Purification Technology 40, 147–154. Yang, Y., Belghazi, M., Hawthorne, S.B., Miller, D.J., 1998. Elution of organic solutes from different polarity sorbents using subcritical water. Journal of Chromatography A 810, 149–159.
Contents lists available at ScienceDirect
Journal of Food Engineering journal homepage: www.elsevier.com/locate/jfoodeng
Mass transfer during pressurized low polarity water extraction of lignans from flaxseed meal Colin H.L. Ho a,b, Juan E. Cacace a, G. Mazza a,b,* a b
Pacific Agri-Food Research Centre, Agriculture and Agri-Food Canada, 4200 Hwy 97 Summerland, BC, Canada V0H 1Z0 Department of Food Science, University of Manitoba, Winnipeg, Manitoba, Canada R3T 2N2
a r t i c l e
i n f o
Article history: Received 20 July 2006 Received in revised form 3 April 2008 Accepted 8 April 2008 Available online 14 April 2008 Keywords: Extraction Packed bed Kinetic Diffusion Subcritical water Optimization Linum Lignans Phenolics Secoisolariciresinol diglucoside Functional foods Nutraceuticals
a b s t r a c t The feasibility of extracting lignans from flaxseed meal using pressurized low polarity water (PLPW) in a fixed bed extraction cell was evaluated. Optimization experiments were performed using surface response methodology for flow rate, bed depth and solvent to solid ratio (S/S). Maximum yield of lignan was obtained at flow rate of 0.6–2 mL/min, bed depth to internal diameter ratio (L/ID) of 20–25, and S/S ratio of 77–150 mL/g. The effect of flow rate and the mass transfer mechanisms governing the extraction of lignans from flaxseed meal was also evaluated. A diffusion model and a two site kinetic model were used to describe the extraction of lignans from flaxseed meal by PLPW. The two models gave similar quality of data fits. The diffusion model yielded diffusivities ranging from 2 1013 to 9 1013 m2 s1 for extraction of lignans at 180 °C, pH 9 and a 1:1.5 meal to co-packing material ratio. Ó 2008 Elsevier Ltd. All rights reserved.
1. Introduction Safe, environmentally acceptable and effective bioseparation has become a key research need for industrial and chemical engineering industries in the 21st century (Nobel and Agrawal, 2005). Water, an inorganic solvent, exerts high affinity for hydrophilic compounds. As a result, its application in extraction of low polarity compounds is limited (Wongkittipong et al., 2004). Pressurized low polarity water (PLPW) (also referred as superheated water; subcritical water; pressurized hot water), has been widely promoted as environmentally benign alternative to orthodox organic solvents (methanol, acetone, ethanol, hexane) due to its reduced polarity above normal boiling point while maintaining water in the liquid state (Hawthorne et al., 1999). PLPW has working temperatures ranging between 100 and 374 °C (Poliakoff and Licence, 2002; Cacace and Mazza, 2006). Advancement in PLPW extraction is necessary to improve the process efficiency and economic potential. Mathematical modelings involving mass transfer
* Corresponding author. Address: Pacific Agri-Food Research Centre, Agriculture and Agri-Food Canada, 4200 Hwy 97 Summerland, BC, Canada V0H 1Z0. Tel.: +1 250 494 6376; fax: +1 250 494 0755. E-mail address: [email protected] (G. Mazza). 0260-8774/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.jfoodeng.2008.04.003
parameters (flow rate, temperature, pressure, etc.) in food systems are gaining more attention (Welit-Chanes et al., 2005). It is important to develop models for the extraction process where different extraction operating parameters are optimized for process economics. However, such predictions require the understanding of mass transfer mechanisms. The PLPW extraction process usually begins by putting the biomass in a fixed/packed bed inside a cylindrical/tubular extraction cell. The packed sample is in contact with the flowing liquid and the separation process kinetic requires the collection and analysis of solute enriched extracts in a time dependent manner. PLPW is effective in altering sample matrices and displacing solutes from their original binding sites. However, few attempts have been documented to differentiate the relative influence of partitioning thermodynamics and desorption kinetics from the sample matrix on extraction rates and recoveries of phytochemicals from food samples (Kubatova et al., 2002). Despite the increased interest in PLPW, its application in the separation of bioactives from food and the transport mechanisms of PLPW extraction of plant components are not yet well understood. Although the diffusion of the dissolved solute within the solid is usually the rate limiting steps for most botanicals (Schwartzberg and Chao, 1982), partitioning of the solute between the solid
65
Colin H. L. Ho et al. / Journal of Food Engineering 89 (2008) 64–71
Nomenclature Bi C Ci De Do Ea F Ks L Mt M1
Biot number (dimensionless) solute concentration in the extractor at any time during the extraction process (mg/mL) initial solute concentration at the beginning of extraction (mg/mL) effective diffusion coefficient (m2/s) initial diffusivity (m2/s) activation energy for diffusion (kJ/mol) fraction of solute released quickly (dimensionless) mass transfer coefficient (m/s) bed depth (m) total amount of diffusing substance extracted after time t (mg/g) equilibrium solute concentration in solution, maximum amount of solute that can migrate (extracted) after infinite time (mg/g)
matrix and the solvent have been reported as the rate limiting mechanism for subcritical water extraction of essential oil from savory (Kubatova et al., 2002). Therefore, the prevailing mechanism during mass transfer depends on both the solute to be extracted and properties of the extracting solvent. Secoisolariciresinol diglucoside (SDG) is the major lignan present in flaxseed meal. The extraction of SDG and other bioactives from flaxseed meal using PLPW was previously studied with changing temperature, pH, solvent to solid ratio and co-packing materials (Ho et al., 2007). However, the limiting extraction mechanism was not determined. The purpose of this paper is to optimize the extraction of lignans from flaxseed meal in terms of flow rate, S/S ratio and bed depth and to elucidate the mechanisms controlling the mass transfer process of PLPW extraction in packed beds by applying two mass transfer models. 2. Materials and methods The extraction apparatus and analytical techniques used are as described by Cacace and Mazza (2006). Defatted flaxseed meal was obtained from Flora Manufacturing and Distribution Ltd. (Burnaby, BC). Particle size was approximately 0.22 mm in diameter as measured by Ho et al. (2007). A schematic of the extraction system, characteristic dimensions and geometry of the extraction vessel are shown in Fig. 1. The experiments were carried out in 10.6 mm ID (1/2 inch OD) columns packed with flaxseed meal and an appropriate amount of co-packing glass beads (3 mm in diameter) mixed with the meal to achieve a 1:1.5 ratio (w/w) of meal to glass beads as suggested by Ho et al. (2007). All extractions were carried out in one of three stainless steel cylindrical extraction cells of lengths 16, 27 and 30 cm. The cell length was chosen to minimize dead volume in the cell depending on the desired bed depth. The extraction cell was packed and mounted vertically in the oven with solvent flowing from bottom to top. Pressurized water was pumped at the desired constant flow rate, passing through the packed bed once, and extracting the solute in a continuous process. After passing out of the cell, the extracts were cooled, the pressure reduced and the extracts collected in containers. The collection vessel was changed periodically to provide a plurality of collection volumes, thereby separating and individually collecting multiple eluant fractions. Buffered deionized ultrapure water used for extraction was made using 0.25 M sodium carbonate and 0.25 M sodium bicarbonate and was adjusted to pH 9 as recommended by Ho et al. (2007). At the end of extractions, SDG contents
Rg Rr R2 T X Y dp k1 k2 r t u b
universal gas constant, J mol1 K1 (1.987 cal/K mol) radius of solid particle (m) coefficient of determination (dimensionless) temperature (°C, K) independent variables of polynomial regression model dependent variables of polynomial regression model diameter of solid particle (m) rate constant for fast extraction stage (min1) rate constant for slow extraction stage (min1) radial distance from centre of spherical particle (m) extraction time (min) superficial velocity (m/s) regression coefficients of polynomial model (dimensionless)
of the extracts were evaluated by HPLC (Eliasson et al., 2003; Cacace and Mazza, 2006). 3. Experimental design Optimizations of key PLPW extraction conditions were carried out using surface response methodology (Haaland, 1989). The selected experimental design was a three factor five level central composite design for flow rate, bed depth and solvent to solid ratio. The experiment consisted of 18 runs including four replicates at the centre point. Fixed conditions employed in the experiments included a uniform temperature of 180 °C, constant flaxseed meal to co-packing material ratio (1:1.5) and use of pH 9 buffered water (Ho et al., 2007). The actual and coded (inside bracket) values of the factors of the experimental design are given in Table 1. Data were analyzed using the response surface regression (RSREG) procedure of SAS (SAS Institute Inc.) to fit the following second-order quadratic polynomial regression model. Y ¼ b0 þ
3 X i¼1
bi X i þ
3 X i¼1
bii X 2i þ
3 XX
bij X i X j
i
where Y are dependent variables (lignan yields), b0, bi, bii, bij are constant and regression coefficients of the model. Xi and Xj are the independent variables in the model (flow rate, bed depth and solvent to solid ratio). RSREG provides the analysis of variance (ANOVA) and is able to estimate the coefficient parameters of the model, the contribution of each type of effect (linear, quadratic, and cross-product), and the shape of the curve. A goodness of fit test of the model was performed with the regression (REG) procedure by backward elimination to only keep variables significant at the 0.1% level. Plots were made using Sigma Plot software (SPSS Inc., Chicago, IL). To further investigate the effect of flow rate on extraction rate and to evaluate the extraction mechanism model, experimental runs using a ‘‘one-factor-at-a-time” method were carried out at different flow rates. For these experiments, flaxseed meal (3.6 g + 5.5 g glass beads at 14 cm bed depth) was sequentially extracted with pH 9 water at 180 °C at four different flow rates (0.6, 2, 4, 7.4 mL/min). Also, additional extractions of flaxseed meal with co-packing beads at 130, 160, and 190 °C constant temperatures were conducted with pH 9 buffered water at 1 mL/min. The natural logarithm of lignans diffusivities obtained at 130, 160, 180 and 190 °C were plotted as a function of the reciprocal of absolute temperature (1/T) and the energies of activation for lignans were calculated from the slope (Ea/Rg) of the straight lines.
66
Colin H. L. Ho et al. / Journal of Food Engineering 89 (2008) 64–71
Pressure Gauge Temperature Recorder Oven
Back Pressure Regulator Cooling Bath
Extract WaterHeating Coil
Liquid Pump
Extraction Cell
Thermocouple Pressure Relief Valve
Water Reservoir
dp
L
u Fig. 1. Characteristic dimensions and geometry of the packed bed PLPW extraction vessel.
4. Mass transfer models
tion within the diffusion volume changes with respect to time is known as Fick’s second law (Cussler, 1984; Mantell et al., 2002):
4.1. Diffusion model The general conservation equation for one-dimensional nonsteady state diffusion in a spherical particle when the concentra-
oC o2 C 2 oC ¼ De þ ot or2 r or
! ð1Þ
67
Colin H. L. Ho et al. / Journal of Food Engineering 89 (2008) 64–71 Table 1 Central composite experimental design with 3 variables for extraction of lignans at 180 °C with pH 9 buffered water and 1:1.5 meal to co-packing ratio in a 10.6 mm internal diameter cell Run
Flow rate (mL/min)
Bed depth (cm)
Solvent to solid ratio (mL/g)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
2 (1)a 2 (1) 2 (1) 2 (1) 6 (+1) 6 (+1) 6 (+1) 6 (+1) 0.6 (1.68) 7.4 (+1.68) 4 (0) 4 (0) 4 (0) 4 (0) 4 (0) 4 (0) 4 (0) 4 (0)
7 (1) 7 (1) 21 (+1) 21 (+1) 7 (1) 7 (1) 21 (+1) 21 (+1) 14 (0) 14 (0) 2.2 (1.68) 25.8 (+1.68) 14 (0) 14 (0) 14 (0) 14 (0) 14 (0) 14 (0)
39 (1) 115 (+1) 39 (1) 115 (+1) 39 (1) 115 (+1) 39 (1) 115 (+1) 77 (0) 77 (0) 77 (0) 77 (0) 12 (1.68) 142 (+1.68) 77 (0) 77 (0) 77 (0) 77 (0)
a Numbers in parentheses are coded values of variables in the experimental design.
The initial condition is C ðt¼0Þ ¼ C i If surface mass transfer resistance is negligible then the boundary conditions are oC ¼0 or ðr¼rc Þ C ðr¼rb Þ ¼ 0 where C is the solute concentrations (mg/mL) at any location in the particle at time t (s); Ci is the initial solute concentration (mg/mL); De is the mass effective diffusion coefficient (m2/s) assuming that De is constant with the concentration; t is extraction time (s); r is radial distance coordinate from centre of spherical particle (m); rc is the centre of the spherical particle (r = 0); rb is radius of spherical particle (m). The solution to Eq. (1) for a spherical particle assuming negligible external resistance to mass transfer is given by Crank (1975) and Cussler (1984): " # 1 Mt 6 X 1 De n2 p2 t ¼1 2 exp ð2Þ p n¼1 n2 M1 R2r where Mt is the total amount of solute (mg/g) removed from flaxseed meal after time t; M1 is maximum amount (mg/g) of solute extracted after infinite time; Mt/M1 is ratio of total migration to the maximum migration concentration and Rr is the average radius of flaxseed meal particle. When time becomes large, only one term in the series is significant and Eq. (2) becomes ! Mt 6 De p2 t ¼ 2 exp 1 ð3Þ M1 p R2r
fusivity De is estimated from Eq. (2) using Microsoft Excel Solver program (Tutuncu and Labuza, 1996). 4.2. Two site kinetic model The mathematical model proposed by So and MacDonald (1986) and Kubatova et al. (2002) uses two steps to define an extraction curve: a certain fraction (F) of the solutes desorb at a faster rate defined by k1, and the remaining fraction (1 F) desorbs at a slower rate defined by k2, thus:
Table 2 Surface response and ANOVA for lignan yields in extracts Run
Amounta (mg/g)
Yieldb (%)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Model Linear Quadratic Cross-product R2 Effects Flow rate Bed depth Solvent to solid ratio
6.6 17.4 19.1 20.9 5.4 8.7 4.8 17.1 20.7 7.7 2.8 19.4 2.5 17.5 13.4 12.4 8.1 10.9
31.6 83.0 91.2 100.0 25.9 41.5 22.8 81.7 99.2 37.0 13.5 93.0 12.0 86.9 64.2 59.4 38.6 52.3 ***c ***
** ** **
a
Lignan yields in mg per g of flaxmeal expressed as SDG equivalents. Compound yields in weight percentage of total content in flaxseed meal. c ***Significant at 0.01 level, **significant at 0.05 level, *significant at 0.1 level, NS non significant (p > 0.1). b
Table 3 Regression coefficients and analysis of variance of the second-order polynomial model for lignans of flaxseed meal extracts Variablesa
Coefficientc
Intercept X1b X2 X3 X 21 X 22 X 23 X1 X2 X1 X3 X2 X3
19.29NS, d 15.95* 3.99** 0.47 *** 1.52*
Model R2
***
0.35
NS
0.8867
P P P y^ i=l i=l i=l j=i+l Polynomial model Y ¼ b0 þ 3i¼1 bi X i þ 3i¼1 bii X 2i þ 2i¼1 P3 j¼iþ1 bij X i X j adjusted by backward elimination with the goodness-of-fit test at the level of 0.1%. b X1 = flow rate, X2 = bed depth, X3 = solvent to solid ratio. c Lignan yields in mg/g of flaxmeal as SDG equivalents determined by HPLC (Eliasson et al., 2003). d ***Significant at 0.01 level, **significant at 0.05 level, *significant at 0.1 level, NS non significant (p > 0.1). a
Two methods were used to determine the effective diffusion coefficient. The first method is the linear (graphical) solution from which 2 De can be determined from the slope Slope ¼ pRD2 e of the linear r portion of the ln (1 Mt/M1) vs. time plot (Dibert et al., 1989). The second method is non-linear regression of Eq. (2) with effective diffusivity (De) as the fitting parameter. The effective dif-
NS NS 0.8927
68
Mt ¼ 1 ½Fek1 t ½ð1 FÞek2 t M1
Colin H. L. Ho et al. / Journal of Food Engineering 89 (2008) 64–71
ð4Þ 100
5. Results and discussion The models developed by surface response analysis for yields of lignans (p < 0.01) were significant (Table 2). The coefficients of determination, R2, indicated the degree of fit of the second-order regression models were high for lignans (R2 = 0.89). Regression coefficients and analysis of variance of the adjusted polynomial second-order models for lignans are presented in Table 3. The coefficients of solvent to solid ratio and bed depth were positive for lignans implying that higher levels of solvent to solid ratio and bed depth would result in higher recovery of the target compounds in the extracts. The order of significance of process variables affecting flaxseed meal extraction can be ranked as follows: solvent to solid ratio > bed depth > flow rate.
80
SDG Yield %
where t is extraction time (min); F is the fraction of the solute released quickly; (1 F) is the fraction of the solute released slowly; k1 is the first-order rate constant describing the fast released fraction (min1); and k2 is the first-order rate constant describing the slow released fraction (min1). The Excel solver regression routine was used to fit data to Eq. (4). The fitted parameters are F, k1 and k2.
60
40
20 0 140 lv 120 en 100
So
tt
o
so 80 lid 60 ra tio 40 (m L/ 20 g)
7
6 5 4 3 2 1
ate ow r
)
/min
(mL
Fl
constant bed depth 14 cm
5.1. Effect of solvent to solid ratio
5.2. Effect of bed depth The effect of bed depth was significant for yields of lignan (p < 0.05) (Table 2). Yield percentages for SDG (Figs. 2B and 3) increased when raising the bed depth from 2 cm to 26 cm regardless of the S/S selected. The high bed depth studied (21–26 cm) gave higher yields for SDG (93–100%) extraction. The gain in recovery for bed depth was higher (steeper slope) (>50%) than the gain obtained by S/S (shallow slope) (<40%). This confirms that the use of a
120
100
SDG Yield (%)
Solvent to solid (S/S) ratio is the ratio of the amount of extraction fluid used to initial mass of starting material. Fig. 2A and B show that SDG extraction yield increased with S/S ratio regardless of flow rate and bed depth, respectively. The optimum S/S ratio to reach the maximum SDG yield was about 110 mL/g with 2 mL/min flow at a 14 cm bed depth or 120 mL/g for a bed depth of 20 cm at 4 mL/min. A larger S/S (>140 mL/g) ratio was required to improve recovery of SDG until equilibrium was reached when flow rate was higher than 4 mL/min at 14 cm bed depth (Fig. 2A). Similarly, at a uniform flow of 4 mL/min, a large S/S ratio was necessary to expedite recovery when the bed depth was less than 20 cm (Fig. 2B). However, the S/S ratio could be reduced with a lower flow rate and higher bed depth without affecting the yield. Thus a decrease of S/S from 110 mL/g to 40 mL/g did not affect SDG yield when flow rate of 2 mL/min was used on a bed depth of 21 cm (9 min residence time) (Fig. 3). A similar degree of S/S reduction resulted in a decrease of yield from 80% to about 40% in the two extraction conditions with a residence time of 3 min. Results in Table 3 shows that the linear effect of S/S was significant for SDG (p < 0.01). Neither quadratic nor cross-product interaction effects with S/S were detected. This was consistent with response surface plots of SDG which demonstrated yields increased almost linearly with S/S (Fig. 2). An increase in S/S would favor extraction by modifying the concentration gradient between the solution in the extraction cell and the surface of the sample matrix. Yields were also higher by increasing the solvent to powder ratio in the extraction of anthocyanins from sunflower husks (Pifferi and Vaccari, 1983). Use of high solvent to solid ratios, however, results in dilute solutions.
80 60 40 20 25 20
0 140
)
15
120
Sol
100
ven
t to
h
10
80
soli
d ra 60 tio ( 40 mL /g)
5
d
Be
t ep
(m
d
20
constant flow rate 4 mL/min Fig. 2. Effect of flow rate (A) and bed depth (B) and solvent to solid ratio on extractions of SDG from flaxseed meal at 180 °C using pH 9 buffered water with 1:1.5 meal to glass beads ratio.
larger bed depth can reduce the use of water due to the need for a lower S/S ratio. Lignan yields in the extracts increased with bed depth (Fig. 3). There were significant differences in yields observed for extractions with different residence times (Fig. 3). The differences in yields and rates of extraction can be attributed to the difference in residence times (t), which defines the contact time of water with meal particles. The residence time is calculated as a function of bed depth and superficial velocity (t = L/u) where L is bed depth and u is the superficial velocity. Fig. 3 shows two curves at a constant bed depth of 7 cm with different flow rates. The curve at 2 mL/min gave >60% more SDG than a flow of 6 mL/min at the same S/S ratio. Extraction at 2 mL/min, again, enabled higher SDG recovery than 6 mL/min at the same bed depth of 21 cm (Fig. 3). Therefore,
69
120
120
100
100
80
80
SDG yield (%)
SDG Yield (%)
Colin H. L. Ho et al. / Journal of Food Engineering 89 (2008) 64–71
60
40 7cm 6mL/min (residence time 1min) 7cm 2mL/min (residence time 3min) 21cm 6mL/min (residence time 3min) 21cm 2mL/min (residence time 9min)
20
60
40 0.6mL/min 2mL/min 4mL/min 7.4mL/min
20
0
0 0
20
40
60
80
100
120
140
0
100
changing the bed depth from 7 cm to 21 cm had a significant effect on both the extraction rate and total recovery of SDG from flaxseed meal. Raising the bed depth from 7 to 21 cm increased SDG yields by 20–40% and extraction time was reduced by about 50–80% at both flow rate conditions. The highest SDG yield (100%) was obtained at a residence time of 9 min while the lowest yield (40%) was observed at 1 min residence time (Fig. 3). In addition, the run at 21 cm and 2 mL/min (9 min residence time) required the least S/S ratio to reach extraction equilibrium; whereas runs with 3 min residence time obtained by combination of 2 bed depths and 2 velocities produced almost identical yields utilizing the same volume of S/S ratio (Fig. 3). Longer residence time allows the solvent sufficient time to penetrate through the solid pores. Liu and Wyman (2003) also proposed that longer residence time was responsible for enhanced removal of solutes from biomass using compressed hot water. 5.3. Effect of flow rate Fig. 2A shows enhanced yields of extracted solutes with declining solvent flow rate. Flow rate was significant for SDG (p < 0.05) (Table 2). SDG showed large percentage increase in yields (40%) with reduction of flow rate at any given S/S (Fig 2A). The combination of low flow rate (1–2 mL/min) and low S/S provides SDG yields higher than those at high flow (6–7 mL/min) rate and low S/S. High flow rate coupled with high S/S (>100) could produce similar yields to those at low S/S and flow rate; however, large S/S leads to huge solvent consumption (diluted extract) which is undesirable. Therefore, a flow rate between 1 and 2 mL/min may be preferred. Kinetics of extraction were similar for the four flow rates studied (0.6, 2, 4, 7.4 mL/min) as shown by the yield of lignan SDG against time (Fig. 4). Thus it is likely that mass transfer is controlled by the internal diffusion in the flaxseed meal and lower dependency on external mass transfer between the solid surface and bulk liquid phase which is consistent with a negligible effect of increased flow rate (Kubatova et al., 2002). More recently, Cacace and Mazza (2006) reported that permeability within flaxseed (diffusion) exhibited greater resistance than external mass transfer, and the optimal flow rate for the extraction of lignans from whole flaxseed with PLPW was 0.5 mL/min in the tested range of 0.3–4 mL/min. Shotipruk et al. (2004) found that extracted anthraquinone concentration was higher for lower flow rates. Thus, high flow rate does not always speed up mass transfer depending on internal structure of the food matrix and the extractive capacity
300
400
500
Fig. 4. Effect of flow rate on PLPW extraction of SDG from flaxseed meal at constant bed depth 14 cm (3.6 g meal + 5.5 g glass beads) with pH 9 buffered water at 180 °C.
of the solvent (Pinelo et al., 2006). In agreement with our findings from the effects of flow rate, calculations of the Biot number Bi ¼ KDser resulted in values from 11 to 47. These values are higher than 10 which indicate that diffusion within the solid controls the process. 5.4. Model fitting of experimental data Example plots of extraction data at flow rates 2 and 6 mL/min and fitted regression lines are displayed in Fig. 5. Effective diffusion coefficients were calculated from the plots. The increase in extraction rate with temperature reflects the increase of diffusivity values. The values of diffusivity (both linear and non-linear) ranged from 1.4 to 15.7 1013 m2/s for lignan (Table 4). Fig. 6 shows the fitted curves for extraction yields of lignan SDG at three different temperatures using the two site kinetic model. The effect of temperature on the extraction of lignan indicated the maximum recoveries was reached at around 160–190 °C. The plots show that the values calculated from the model adequately match the experimental values (R2 > 0.96). Elevated temperatures were found to enhance extraction rates and reduced extraction time from 420 min at 130 °C to 100 min at 190 °C. About 80–90% of lignans were extracted in less than 60 min extraction time at 190 °C (Fig. 6).
0.2 experimental data 6mL/min regression line experimental data 2mL/min regression line
0.0 -0.2
ln (1-Mt/M∞)
Fig. 3. SDG yield as a function of solvent to solid ratio for extraction at 180 °C, pH 9 for two bed depths and two flow rates. Bed depth 7 cm (1.8 g meal + 2.7 g glass beads); 21 cm (5.5 g meal + 8.2 g glass beads).
200
Time (min)
Solvent to Solid Ratio (mL/g)
-0.4 -0.6 -0.8 -1.0 -1.2 0
1000
2000
3000
4000
Time (s) Fig. 5. Application of linear solution using Fick’s law for SDG extraction at flow rate 2 and 6 mL/min, respectively with pH 9 PLPW at bed depth 7 cm at 180 °C.
70
Colin H. L. Ho et al. / Journal of Food Engineering 89 (2008) 64–71
Table 4 Effective diffusion coefficients for lignans extracted with PLPW at different temperature and pH with a fixed meal to co-packing glass beads ratio of 1:1.5 using 210 solvent to solid ratio at water flow rate of 1 mL/min at 10 cm bed depth Temp (°C)
pH
130 160 190 130 160 190
9 9 9 4 4 4
Linear
Non-linear
De 1013 (m2 S1)
De 1013 (m2 S1)
2 5.3 15.7 1.4 3.2 14.3
1.4 3.5 10.8 1.1 2.6 10.1
An increase in temperature causes a decrease in water viscosity and dielectric constant and therefore an increase in diffusion (Ong et al., 2006). The increase in diffusivity with increased temperature may also be caused by an increase of the internal energy of the molecules and thus their mobility. This is consistent with the findings of Yang et al. (1998) who showed that increasing temperature favored PLPW extraction by increasing the diffusion coefficient of organic solutes, which accelerated the rate of mass transfer and reduced extraction time. An increase in temperature from 160 to 190 °C decreased extraction time from 250 min to 100 min for SDG (Fig. 6). However, the problem with the use of very high temperature is that all reactions are accelerated, including all the unwanted side reactions and
A
120
SDG Yield %
100
80
60
130°C experimental 160°C experimental 190°C experimental 130°C two site kinetic model 160°C two site kinetic model 190°C two site kinetic model
40
20
0 0
100
200
300
400
500
Time (min) 120
B
SDG Yield %
100
80
60
2mL/min experimental 6mL/min experimental 6mL/min two site kinetic model 2mL/min two site kinetic model
40
20
0 0
100
200
300
Time (min) Fig. 6. Experimental fitting of two site kinetic model to SDG recovery data obtained at various temperatures (A) and at two different flow rates at fixed bed depth 21 cm (B) with meal to co-packing ratio 1:1.5 using pH 9 buffered water.
Table 5 Predicted equilibrium concentrations and kinetic coefficients obtained by fitting two site kinetic model to PLPW extraction of SDG data at 1 mL/min from 2 g flaxseed meal pH
Packing (g)
Equilibrium conc
Kinetic coefficient min1
R2
Stage 1
Stage 2
Stage 1
F*a
1 F*b
k1 104
Temp. 130 °C, 210 mL/gS/S 4 (1) 0 (1) 0.2 4 (1) 3.0 (+1) 0.3 9 (+1) 0 (1) 0.1 9 (+1) 3.0 (+1) 0.2
0.8 0.7 0.9 0.8
10 52 4 7
43 58 28 36
0.93 0.94 0.92 0.91
Temp. 160 °C, 210 mL/gS/S 4 (1) 0 (1) 0.3 4 (1) 3.0 (+1) 0.4 9 (+1) 0 (1) 0.4 9 (+1) 3.0 (+1) 0.5
0.7 0.6 0.6 0.5
34 114 114 118
61 84 84 133
0.94 0.96 0.96 0.98
Temp. 190 °C, 210 mL/gS/S 4(1) 0 (1) 0.9 4 (1) 3.0 (+1) 1.0 9 (+1) 0 (1) 1.0 9 (+1) 3.0 (+1) 0.9
0.1 0.0 0.0 0.1
505 564 503 498
165 822 737 145
0.98 0.97 0.94 0.96
a b c d
Stage 2 c
k2 104
d
F (dimensionless) fraction of solute released quickly. 1 F is the fraction of the solute released slowly. k1 (min1) is the first-order rate constant for the quickly released fraction. k2 (min1) is the first-order rate constant for the slowly released fraction.
degradations (Wongkittipong et al., 2004). Therefore, the optimum temperature selected for extracting lignan SDG from flaxseed meal was 180 °C in the present study. The augmentation effect of high pH on lignans extraction was discussed in Ho et al. (2007). Extraction time increased as temperature decreased and the rise was more noticeable at acidic pH 4. At the alkaline pH 9, the favorable effect of pH on extraction yield might have compensated for the reduction of extraction rate caused by lower temperature, preventing a larger increase in extraction time. The lowest diffusivity values were obtained for lignan at low temperatures and pH (Table 4), which further indicates that the times required for extractions at low temperature and pH were longer due to reduced diffusivities. At 190 °C, equilibrium fraction of the fast extracting stage F was higher compared to the equilibrium fraction of the slow stage (1 F) (Table 5). The pH and co-packing glass beads had no apparent effect on F values at 190 °C. In addition, both k1 and k2, increased sharply with temperature. At 160 °C, both fast and slow fractions were equally important. The kinetic constants at 160 °C demonstrated moderate extraction rates. Extraction rate was improved by either increasing pH or addition of packing glass beads. At low temperature (130 °C), the fast extracting stage was not important as the F values were small (Table 5). The effective diffusivity values were fitted to the Arrhenius type equation: Ea De ¼ D0 exp Rg T where De and Do are the effective diffusivity and initial diffusivity, respectively, Ea is the energy of activation (kJ/mol) for diffusion, Rg is the universal gas constant (kJ/mol K), and T is the absolute temperature (K). Fig. 7 shows the ln De vs. 1/T plot. Activation energies for SDG were 51 and 56 kJ/mol at pH 9 and pH 4, respectively. Cacace and Mazza (2003) reported activation energy values of 70– 97 kJ/mol for diffusion of phenolic from berries. Spiro and Selwood (1984) reported activation energy for caffeine diffusion through coffee beans and tea leaf to be 32 kJ/mol and 23 kJ/mol, respectively. Activation energy can be reduced due to the lower energy barrier to initiate diffusion since thermal energy can overcome solute–sol-
Colin H. L. Ho et al. / Journal of Food Engineering 89 (2008) 64–71
-27.0
-27.5 2
R =0.97
Ln D
-28.0
-28.5
-29.0
-29.5
-30.0 0.0021
0.0022
0.0023
0.0024
0.0025
1/T (1/K) Fig. 7. Arrhenius-type relationship between effective diffusivity and temperature for SDG using 1 mL/min pH 9 buffered water with 1: 1.5 meal to co-packing material ratio and S/S ratio 210 mL/g.
ute and solute–matrix interactions by decreasing activation energy in the extraction process (Richter et al., 1996). 6. Conclusion The experimental data from PLPW extraction of lignans from flaxseed meal subjected to various processing conditions (temperature, pH, solvent to solid ratio, co-packing materials, flow rate, bed depth) were accurately described using both a two site kinetic model and a diffusion model. High temperature (180°), high pH (9) with 1:1.5 meal to co-packing glass beads resulted in a more rapid extraction by increasing diffusivity and reducing activation energy. Diffusivities attained from the diffusion model ranged from 2 1013 to 9 1013 m2 s1, for rapid extraction of lignans under the mentioned conditions. Flow rate had little influence on extraction kinetics. PLPW is a promising technique that enables production of biobased products in an environmentally sustainable manner without the use of toxic organic solvents. Acknowledgment We thank Lana Fukumoto for her assistance with the HPLC analysis of lignans, and Rod Hocking for manufacturing the extraction cells. We are also grateful to CBIN, Natural Resources Canada for financial support of this research. References Cacace, J.E., Mazza, G., 2006. Pressurized low polarity water extraction of lignans from whole flaxseed. Journal of Food Engineering 77, 1087–1095.
71
Cacace, J.E., Mazza, G., 2003. Mass transfer process during extraction of phenolic compounds from milled berries. Journal of Food Engineering 59, 379–389. Crank, J., 1975. The Mathematics of Diffusion. Clarendon Press., Oxford, England. Cussler, E.L., 1984. Diffusion: Mass Transfer in Fluid System. Cambridge University Press, Cambridge, UK. pp. 146–177. Dibert, K., Cros, E., Andrieu, J., 1989. Solvent extraction of oil and chlorogenic acid from green coffee Part II: Kinetic data. Journal of Food Engineering 10, 199–214. Eliasson, C., Kamal-Eldin, A., Andersson, R., Aman, P., 2003. High-performance liquid chromatographic analysis of secoisolariciresinol diglucoside and hydroxycinnamic acid glucosides in flaxseed by alkaline extraction. Journal of Chromatography A 1012, 151–159. Haaland, P.D., 1989. Experimental Design in Biotechnology. Marcel Decker Inc, New York, NY. pp. 189–249. Hawthorne, S.B., Miller, D.J., Yang, Y., Lagadec, A.J.M. 1999. Method of manipulating the chemical properties of water to improve the effectiveness of a desired chemical process. US Patent# 6,001,256. Ho, C.H.L., Cacace, J.E., Mazza, G., 2007. Pressurized low polarity water extraction of lignans from flaxseed meal. Lebensmittel Wissenschaft und Technologie 40, 1637–1647. Kubatova, A., Jansen, B., Vaudoisot, J.F., Hawthorne, S.B., 2002. Thermodynamic and kinetic models for the extraction of essential oil from savory and polycyclic aromatic hydrocarbons from soil with hot (subcritical) water and supercritical CO2. Journal of Chromatography A 975, 175–188. Liu, C., Wyman, C.E., 2003. The effect of flow rate of compressed hot water on xylan, lignin, and total mass removal from corn stover. Industrial and Engineering Chemistry Research 42, 5409–5416. Mantell, C., Rodriguez, M., De La Ossa, E.M., 2002. Semi-batch extraction of anthocyanins from red grape pomace in packed beds: experimental results and process modelling. Chemical Engineering Science 57, 3831–3838. Nobel, R.D., Agrawal, R., 2005. Separations research needs for the 21st century. Industrial and Engineering Chemistry Research 44, 2887–2892. Ong, E.S., Cheong, J.S.H., Goh, D., 2006. Pressurized hot water extraction of bioactive or marker compounds in botanicals and medicinal plant materials. Journal of Chromatography A 1112, 92–102. Pifferi, P.G., Vaccari, A., 1983. The anthocyanins of sunflower. II. A study of the extraction process. Journal of Food Technology 18, 629–638. Pinelo, M., Sineiro, J., Nunez, M.J., 2006. Mass transfer during continuous solid– liquid extraction of antioxidants from grape byproducts. Journal of Food Engineering 77, 57–63. Poliakoff, M., Licence, P., 2002. Reinventing the chemical industry. Nature 419, 880. Richter, B.E., Jones, B.A., Ezzell, J.L., Porter, N.L., 1996. Accelerated solvent extraction: a technique for sample preparation. Analytical Chemistry 68, 1033–1039. Schwartzberg, H.G., Chao, R.Y., 1982. Solute diffusivities in leaching processes. Food Technology 36, 73–86. Shotipruk, A., Kiatsongserm, J., Pavasant, P., Goto, M., Sasaki, M., 2004. Pressurized hot water extraction of anthraquinones from the roots of Morinda citrifolia. Biotechnology Progress 20, 1872–1875. So, G.C., MacDonald, D.G., 1986. Kinetic of oil extraction from canola (rapeseed). Canadian Journal of Chemical Engineering 64, 80–86. Spiro, M., Selwood, R.M., 1984. The kinetics and mechanism of caffeine infusion from coffee: the effect of the particle size. Journal of the Science of Food and Agriculture 35, 915–924. Tutuncu, M.A., Labuza, T.P., 1996. Effect of geometry on the effective moisture transfer diffusion coefficient. Journal of Food Engineering 30, 433–447. Welit-Chanes, J., Vergara-Balderas, F., Bermudez-Aguirre, D., 2005. Transport phenomena in food engineering: basic concepts and advances. Journal of Food Engineering 67, 113–128. Wongkittipong, R., Prat, L., Damronglerd, S., Gourdon, C., 2004. Solid–liquid extraction of andrographolide from plants – experimental study, kinetic reaction and model. Separation and Purification Technology 40, 147–154. Yang, Y., Belghazi, M., Hawthorne, S.B., Miller, D.J., 1998. Elution of organic solutes from different polarity sorbents using subcritical water. Journal of Chromatography A 810, 149–159.