Vous êtes sur la page 1sur 11

IrenTechnology

Journal of Chemical Tsibranska,and


Bartosz Tylkowski
Metallurgy, 51, 5, 2016, 489-499

Solid-liquid extraction of bioactive compounds:


effect of polydispersity and particle size evolution
Iren Tsibranska1, Bartosz Tylkowski2

1
Institute of Chemical Engineering Received 06 January 2016
Bulgarian Academy of Sciences Accepted 02 April 2016
1113 Sofia, Bulgaria
E-mail: tsibranska@ice.bas.bg
2
Universitat Rovira i Virgili
Departament de Enginyeria Química
Av. Països Catalans, 26 - 43007 Tarragona
Spain

ABSTRACT

Extraction of bioactive compounds from natural products is of growing research interest. The present study
focuses on the role of polydispersity in analyzing the kinetic curves of solid-liquid extraction and determining the ef-
fective diffusion coefficients in the solid. A numerical solution of the polydisperse particles diffusion model is applied.
Partial dissolution of the solid matrix is included taking into consideration the time evolution of the particle size. The
asymmetry of the particle size distribution and the necessity to apply the polydisperse model or the mean average
size as a modeling approach is thoroughly discussed. Examples referring to two experimental systems (Propolis and
Sideritis ssp. L) are given. Conventional batch extraction of polyphenols by an ethanolic solvent is considered there.
Keywords: solid-liquid extraction, polyphenols, particle size distribution, mathematical modeling.

INTRODUCTION the extract. Typically bioactive compounds accumulate


at higher concentrations in specific parts of the plant
A number of investigations are lately focusing on the (leaves, stems, roots, flowers, seeds) or are located in
physico-chemical and therapeutic properties of bioactive the surface or interior layers. The raw materials after
substances present in natural resources through develop- grinding contain particles of various sizes irregularly-
ment of techniques for their extraction and means for shaped, often with an effective amount of two spatial
quality control. Solid-liquid extraction under controlled coordinates, as well as particles of a critical size in the
process conditions is important for the reproducibility three-dimensional space. The solid phase polydispersity
of the end products and preservation of their bioactiv- and the presence of anisotropy have to be also taken
ity. Standard low pressure extraction, CO2 supercritical, into account in treating many raw materials. The repro-
ultrasound or microwave assisted extraction have been ducibility of the extract solutions obtained which may
studied in order to find the optimal extraction condi- contain greater or fewer amounts of active substances
tions and to compare the yield obtained and extract (by weight ratio) among other ballast ones extracted are
composition. An important part of such investigations other issues of interest. In fact the accurate description of
concerns the extraction kinetics and the respective rate the extraction mechanism requires the treatment of the
parameters because of their predictive capabilities in the possible interactions between the solvent and the solid
process evaluation. matrix (soaking the solvent and swelling of the material),
The process of extraction is quite difficult for de- the target component content (possible degradation) at
scription and modeling due to the diversity and specific- any stage of the process well as the effect of various
ity of the starting structure of the solid and its change factors on the extraction effectiveness. An important
throughout the process, the extraction of ballast com- criterion from practical point of view is the ability of the
pounds and the reproducibility of the initial content of model to predict correctly the approach to equilibrium, as
489
Journal of Chemical Technology and Metallurgy, 51, 5, 2016

the latter determines the contact time for the extraction authors apply a two-shape extraction model to describe
process. With regard to the correct description of the adequately the experimental data [5].
kinetics of extraction from plant materials, models of Different interpretations of the mass transfer pro-
different complexity are used, mostly based on the Fick’s cedure, affecting the initial course of the kinetic curves
law, the chemical rate law and empirical models, which are proposed in a number of simplified models related to
have proved to be applicable to most extraction curves. A the external mass transfer in the liquid film around the
comprehensive review of these mathematical approaches particles, the washing, the swelling of the solid particles,
for modeling batch extraction processes is given in [1]. etc. The description of the initial section of the extraction
The unsteady state models, providing estimation of the curve can be also improved by decrease of the particles
effective (constant or variable) intraparticle diffusion size, though the effect of this parameter should be also
coefficient (De) are of special interest [2], because solid accounted of in respect to the rate limiting mechanism.
phase diffusion is usually the controlling mass transfer Moreover, polydisperse systems will affect essentially
step. The determination of the rate of mass transfer, De, the course of the kinetic curve because of the asymmetry
is of importance for the equipment design. The analytical of the particle size distribution (PSD). In principle the
solution of the diffusion equation in case of negligible particle size is the parameter of primary significance
external mass transfer resistance among those affecting the conventional extraction kinet-
ics. In view of its importance the modeling of complex
kf R
( Bi
= > 100 ) polydisperse systems and the parameter identification
De of the kinetic coefficients is of certain interest. In this
is often used to determine the effective diffusivity: case models of better theoretical grounds are preferably
used as they are expected to possess greater predictive
C − Ce 4 ⋅ R ∞ 1 − (−1) n  n 2π 2 
=
C0 − Ce

π n =1 n ⋅ π
exp  − 2 De t  (1) capabilities and applicability to process optimization,
 R  design and control.
The present study focuses on the role of polydisper-
where C0 and Ce refer to the initial and equilibrium sity in analyzing the kinetic curves of solid-liquid extrac-
concentration of the solid phase, respectively. tion and determining the effective diffusion coefficients
Truncated to the first term Eq.(1) is easily applied in the solid. Two experimental systems are considered,
to experimental data, including polyphenols extraction including constant PSD or particle size evolution (due
from white distilled grape pomace [3], S. aromaticum to partial dissolution of the solid matrix). In both cases
and C. cassia [4]: conventional batch extraction of polyphenols with etha-
nolic solvent is applied (Propolis and Sideritis ssp. L).
d ln((C − Ce ) /(C0 − Ce )) π2
= − 2 De (2)
dt R Empirical models
When conventional extraction from plant materials Because of a similarity between extraction and
is considered, the observed values of De are in the range sorption kinetics, a number of simplified models, known
of 10-11 m2/s down to 10-14 m2/s [1]. Particles of a size from modeling the sorption kinetics, have been used,
less than 1mm are usually applied [1] in order to enhance including examples referring to polyphenols extraction
the mass transfer in the diffusion step and to reduce the from plant materials. A comparison of different simpli-
extraction time. The initial plant material, consisting fied kinetic models, mostly used for extraction of bioac-
of stems, leaves flowers, is either sorted in individual tive compounds from plant material, is given in [6-9]
groups and sieved (when the valuable substances are including examples of total polyphenols extraction from
differently distributed in the respective parts of the grape marc and yerba mate. The results are in favor of
plant) or grinded and subjected to extraction as a whole the pseudo-first order kinetics:
(when the bioactive compounds are found everywhere
C∞
in the plant in comparative quantities). The latter gives C
=l C∞ −
exp ( kt + a )
particles of a size of the micron order, yet various in
shape (cylindrical, plate) and characteristic length. Some or in logarithmic form:

490
Iren Tsibranska, Bartosz Tylkowski

 C∞  where C is dimensionless quantity, while and the expo-


ln   vs t (3) nent n=0.5 for Fick’s diffusion in case of flat geometry.
 C∞ − Cl  Additional intercept term is usually inserted to model the
and the standard function technique [10]: fast initial release from the particles and to take account
of the washing, or external mass transfer. This form of
Cl =
A − Bexp(−kt) (4) the equation can be obtained also as a partial case of the
where Cl refers to the current concentration in the liquid parabolic diffusion model [7] in the form:
phase, C∞ is C∞ the equilibrium concentration, while k
(s-1) is the rate constant. A similar expression, resulting
C A 0 + Bt1/ 2
= (10)
from the Fick´ s law solution for great time and infinite The search of a better description of the extraction
flat geometry is used in modeling yerba mate aqueous kinetics, which is never as simple as the referring to
extraction kinetics [11]. one-dimensional diffusion mechanism, has lead to com-
The second-order kinetic model: bined empirical equations, where two additive terms are
ascribed to different mass transfer mechanisms. Exam-
dCl
= k(C∞ − Cl ) 2 ples of such equations modeling the extraction kinetics
dt of polyphenols from plant materials are given below:
is transformed to after integration and subsequent lin- - Washing/swelling followed by intraparticle dif-
earization to: fusion [11]:

t 1 t C= C∞w t + Cd 1 − exp(−kt)
= + ∞( ) (11)
Cl kC∞2 C∞ Tw + t
(5)
This model has been found successfull in case of Here the concentration C (g/l) stands for the sum of
polyphenols extraction from peel and seeds of pome- Cw and Cd referring to the concentrations achieved by
granate marc [9], as well as in their ultrasound-assisted washing/swelling and diffusion, correspondingly. This
extraction [12]. Another three empirical models are equation gives satisfactory description of the bioactive
equally suited for modelling the extraction of total components extraction from yerba mate [11].
polyphenols from soybeans [13]: - Two-rates exponential equation [7, 16]:
- The hyperbolic model, known as Peleg’s model:
C =a (1 − exp(−bt)) + c(1 − exp(−dt)) (12)
t
C= ; (6)
K1 + K 2 t
which provides the evaluation of the solute concentra-
- The Weibull type model, known also as Page model, tion C (mg GAE/g) when the extraction proceeds with
two different rates.
C exp(−kt n )
= (7) Effect of polydispersity and particle size evolution
All empirical models pointed above are more or
- The logarithmic model: C = a log t + b (8) less able to correlate the experimental results, but the
C is evaluated as mg galic acid equivalent per g dry beans physical meaning of the parameters involved is not al-
in Eqs.(6) to (8). The describtion of the solid–liquid ex- ways clear. Two-stage models like the ones developed
traction kinetics of total polyphenols from grape seeds by Sovova [17] for supercritical fluid extraction fitted to
[14], as well as the β-carotene batch extraction from rose the experimental data, may lead to somewhat unrealistic
hip (Rosa canina) by the Peleg equation is characterized overestimation of a given part of the plant material con-
by a high correlation coefficient [15]. nected with a given release mechanism. The two-stage
The extracted amount under diffusion controlled character of the kinetic curve accepted in a number of
conditions can be generally described by a power law cases can be described by taking into account the actual
expression [7, 11]: polydispersity. Such a model is developed in [18] and
applied to extraction of oil from white lupin seeds. The
C = Bt n (9) modification includes more or less adjusting parameters.

491
Journal of Chemical Technology and Metallurgy, 51, 5, 2016

For example, an additional dust fraction is formally proach to equilibrium and the correct determination of the
considered aiming better approximation. The mechanism contact time required. In cases of pronounced asymmetry
corresponding to the initial fast release observed in the the description of the kinetic data by an effective diffusion
kinetic curve is not fully understood, but polydispersity coefficient based on a representative mean diameter may
is regarded as a possible reason for that [19], even not turn to be unsatisfactory [19, 25, 26], i.e. it may lead to
the only one [20]. the identification of different or unprobable value of the
The formal introduction of a time dependent diffu- diffusion coefficient.
sion coefficient can lead to a better description of the Polydispersity may prevail in scaling-up [25], and
kinetic curve in absence of additional supporting infor- hence the mathematical description which takes it into
mation. Thus the applicability of the simple diffusion account seems reasonable. The correct analysis of the
models, where the effective diffusion coefficient De is calculations requires the effect of polydispersity to be
the only adaptation parameter, can be increased. The greater than the experimental error, i.e. the deviation
initial value of De as well as the type and mathematical between the two calculated curves – that of a constant
formulation of the introduced function can be regarded (mean average) size and the other one of PSD, to be
as adjustable parameters, unless additional physical greater than the scattering of the experimental points
evidence is available. An example of a decreasing De(t) determined by the reproducibility of the raw material
relation, which can be satisfactorily modeled through and the analytical method. Furthermore, the calculated
the concentration dependence De (C ) is observed during curve has to describe the kinetic data in the entire range
extraction from mangrove barks [21,22]. Through analy- of time, including the approach to equilibrium, which
sis of the kinetic curve, an increasing two-parametric for practical purposes is defined as that required for 90
De (C ) function is found to provide a better match of % extraction of the raw material.
the experiment results. However, a pronounced concen-
tration dependence of the diffusivity is not expected for Model description
highly diluted plant extracts, which is often the case. A model considering the effects of polydispersity
An example of diffusivity De(t) increasing with time and particle size evolution is given in Table 1. The as-
is found when additional degradation of the solid occurs sumptions of the model are as follows:
[20] (ex. release through polymeric microcapsules). It is - One-dimensional unsteady state Fick’s diffusion
expressed in the form of a D(Mw) polynomial relation within the particle with constant coefficient of internal
(Mw being the molecular weight of the solid polymer). diffusion De (m2/s) (eq.15.1); a regular form is sug-
Solid-liquid extraction with particles of broad (and gested for each class of particles; it is denoted by the
typically asymmetric [19, 23] size distribution should corresponding value of the form factor a (= 0 plate, 1
be analyzed using polydisperse models [18, 24] to avoid for cylinder and 2 for sphere).
unrealistic values of the inner mass transfer coefficient - The regular geometry supposes symmetry, as
[25]. Such results may be obtained because of a faster shown by boundary condition (15.2);
R2 - The surface of all particles is surrounded at any
extraction (or lower diffusion time , respectively)
De time by fluid of concentration Cl (mg/ml), i.e. perfect
on behalf of the smallest particles and hence a delayed, mixing conditions exist in the extraction vessel, so that
slower approach to equilibrium for the larger particles Cl is only time dependent. The flux through the fluid film
in the tail of the distribution, whose characteristic dif- around each class of particles is different because of the
fusion time is much longer. Moreover, the inner mass different size, the external mass transfer coefficient and
transfer is differently influenced by the different parts of the surface concentration C R (see boundary condition
the particle size distribution, PSD. As the characteristic 15.3a). The latter assumes implicitly a solid/liquid distri-
diffusion time increases proportionally to the square of bution coefficient equal to 1, which is reasonable for low
the particle size, small particles have a stronger effect on concentrations of the solutes observed in plant extracts.
the steep initial part of the kinetic curve. The influence When intensive stirring is applied and the external mass
of the right part of the PSD becomes important at larger transfer resistance is reduced to be neglected, the result-
extraction times, which in turn is essential for the ap- ing Eq.(15.3b) is used;
492
Iren Tsibranska, Bartosz Tylkowski

Table 1. Model equations.

Each class of ∂C 1 ∂ a ∂C Eq.


particles in the PSD = De a (r ⋅ ),
∂t r ∂r ∂ r (15.1)

Boundary conditions Center, r=0 Surface, r=Rj (15.2)


∂C ∂C (15.3a)
=0 D = k f (t )(C R − C l )
∂r ∂r R

C R = Cl (15.3b)
Or

1 d (Vs C )
Mass balance in the (15.4a)
extraction vessel dCl 1 dR
=− = − ρ s 4πR 2 N
dt Vl dt Vl dt
Where N, number of particles, Rj(t)
dCl V d C (15.4b)
Or =− s (Rj=const)
dt Vl dt
Average solid (15.5)
concentration

(a + 1) R j a 
C = ∫ a +1 ϕ ( R j )  ∫ r ⋅ C (r ) ⋅ dr dR j
0
R  0 
ϕ ( R j ) - distribution function
Particle radius dR ∂C (15.6)
change R(t) ρs
dt
(
= −k f (t ) C R
)
− Cl = D
∂R R
External mass Sh=(4+1.2Pe2/3)1/2 (15.7)
transfer coefficient ρ 2R ⋅ v ν k 2R [36]
Pe=Re.Sc; Re= l ; Sc = l ; Sh = f
µl D AB D AB

- The volume averaged solid phase concentration is K


4
calculated in accordance with Eq. (5), where C (mg/mL) Vs ∑ϕ ( R ) πR 3j
j
3
= K1 (14)
is the local concentration inside the particle; Vs 0 4
- This flux is related to the particle size evolution
∑1 ϕ ( R j 0 ) 3πR j 0
3

(Eq.15.6), if we observe a partial dissolution of the solid


(as in the case of Propolis [23]). In both cases when We assume a constant volumetric part of the par-
polydispersity is present, the value of the external mass ticles in the PSD ϕ ( R j ) = ϕ ( R j 0 ) = const , which means
transfer coefficient has to be recalculated for every a similar particles size evolution with no change in the
particle size according to correlations known (Eq.7). distribution form. This is a rough simplification of the
- The solid volume Vs (see eq.4) is determined at real picture. The observations on Propolis show that the
any time on the ground of the particle size change, the particle sizes distribution curve is shifted to the left and
initial value being: straitened during extraction [23], see Fig. 1.

Ms Results and discussion


Vs 0 = (13)
ρs
● PSD characterization
where Ms and ρs are the solid mass and density, corre- The analysis of the particles by size is carried out by
spondingly. Vs is related to its initial value Vs0 through: the Image-Pro plus 5® software. Size evolution of the

493
Journal of Chemical Technology and Metallurgy, 51, 5, 2016

Fig. 1. Propolis particle size distribution measured at different extraction times.

particles during extraction is continuously followed by


microscope Axiovert 4OC for transmitted-light bright- V   d  b 
= 1 − exp−   
field and phase contrast with condenser 0.4 and object
Vtot  a  (16)
 
traverser M. The optical micrographs of the particles are
taken by digital camera DeltaPix Invenio 3S connected where V denotes the total volume contained by micropar-
to the microscope [23]. Two examples of PSD from ticles of diameter less than d, Vtot is the total volume of
milled raw material are considered: the population, while a and b are parameters referring
A) Extraction of polyphenols from Sideritis ssp. to the main mean diameter and the spread of the mic-
L, details of the experimental procedure being given in roparticles sizes.
[19]; the particle size of the milled plant ranges from In accord with the analysis reported in ref. [20], the
10µm to 200µm, while the mean number-based particle polydispersity affects the kinetics and has to be account-
size is 40.8 µm. ed for by the model in order to provide an adequate fit of
B) Extraction of polyphenols from Propolis; the the experimental data for distributions of b ≤ 3 (eq,16).
initial PSD shows that 90% of the particles size are in The following values for the Rosin-Rammler pa-
the range of (15-52) ±2 μm with a mean number-based rameters a and b are obtained (see Fig. 2a and Fig. 2b)
diameter of 32 μm [23, 27]. Partial dissolution of the par- on the ground of the experimental data:
ticles occurs during extraction so that the initial of 36μm a=142 µm and b=3.03 (R2=0.925) for the PSD of case A);
changes to a final volume averaged size of 15μm [23]. a=46.7 µm and b=4.43 (R2=0.961) for the PSD of case B);
The PSD observed are shifted to the left and straitened In case A) the regression coefficient is lower, which
during the extraction, see Fig. 1. The partial dissolution can be explained by the cylindrical shape of the parti-
of the solid matrix, as well as the destruction of mechani- cles (of predominantly fibrous structure) as well as the
cally unstable bigger agglomerates of particles in contact distribution curve tendency to binomial one because of
with the solvent contributes most probably to it. fragmentation of other parts of the plant material (see
To quantify these observations the experimental [19]). In such cases the Rosin–Rammler function cannot
PSD data is modeled by the cumulative Rosin–Rammler perform well [28]. The pronounced spread of the PSD
distribution function: is expected to influence the kinetics as suggested also

494
Iren Tsibranska, Bartosz Tylkowski

Fig. 2. Cumulative Size Distribution of particles and Rosin-Rammler curve fit referring to: A) Sideritis; B) Propolis.

by the value of b (=3.03), eq. (16). for PSD (1.12 10-12 vs 1.52 10-12) is obtained in case
In case B) where the particles shape is close to the of considering only the first 60 min of the extraction,
spherical one and the PSD is more symmetrical a better where the effect of the initial part of the kinetic curve
correspondence between calculated and experimental is stronger. A diffusion coefficient decreasing with time
data is observed, the value of b being 4.4. During extrac- is to be expected if it is attributed to the asymmetric
tion this parameter tends to increase, i.e. polydispersity (extended to the right) experimental PSD.
decreases, as the experimental PSD is straitened and The correspondence between the calculated (accord-
shifted to the left (Fig.1). According to the results ing to Eq. 1) and the experimental solid phase extraction
reported in ref. [20] the deviation between the model curve is shown in Fig. 5. The observed deviation is due
accounting for PSD and the mean diameter-based model to the fact that the simplified model fails to describe the
becomes negligible in this case, i.e. the latter provides very different parts of the kinetic curve. The experimen-
a satisfactory description of the experimental kinetics. tal data is based on the maximum extractable quantity
of total polyphenols (17.5 mg/(g solid) [19]), while the
Effect of polydispersity final concentration in the solid (Cequil) is calculated
A. No time evolution of the PSD (Rj=const): according to the mass balance for a solid/liquid ratio of
Example with polyphenols extraction from Sideritis ssp. L). 1:15. The same figure shows the kinetic curve calculated
The effective diffusion coefficient is searched by (R² = 0.932) by the power model (Eq.9). The intercept
fitting the experimental kinetic curves of the liquid is set equal to the experimentally observed value of 17.5
phase concentration with the solution of Eqs. (15): 1, mg/g solid. The obtained value of n = 0.433 corresponds
2, 3a, 4b, and 5 in Table 1. A value of 1.5 10-12 m2/min
is obtained (see the calculated and experimental kinetic
shown in Fig. 3).
When simplified models like that described by Eq.
(2) are used to determine the effective solid phase dif-
fusion coefficient, the calculation is based on the mean
averaged size of the dried and powdered plant material
(40.8 µm). The value De = 0.56 10-12 m2/min is obtained
from the slope of the plot (C − Ce ) /(C0 − Ce ) vs time
shown in Fig. 4. It is lower than that obtained by the
numerical solution accounting for polydispersity, and
reflects the influence of the right part of the kinetic curve
where diffusion is slower due to the presence of bigger Fig. 3. Liquid phase kinetic curve for Sideritis calculated by
particles. A better correlation (R2=0.989) and diffusivity the model accounting for PSD [19] (solid/liquid ratio 1:15;
values closer to that obtained by the model accounting C t =0
=17.5 mg/g solid [17]; Ce (Cl final) = 116.7 mg/ml).

495
Journal of Chemical Technology and Metallurgy, 51, 5, 2016

Fig. 4. Plot of experimental data for Sideritis [19] in ac- Fig. 6. Propolis: Extracted polyphenols vs liquid/solid ratio.
cordance with Eq.(2).

Fig. 5. Experimental and calculated solid phase kinetic data Fig. 7. Propolis: Experimental and calculated particle size
for Sideritis in accordance with Eq. 1 and the power model. evolution.

to the theoretical one expected for Fick’s type diffusion follows its evolution in correspondence with Eq. (15.6)
in spherical geometry. In general, the power law model where the solid density ρs is equal to1180 kg/m3 [23].
applied to extraction from vegetal material reports a The initial solid phase concentration is set equal to the
wide range of values for the diffusion exponent less maximum extractable quantity of polyphenols. The latter
than 1 (n < 1) [8, 15]. is defined on the ground of the plateau in Fig. 6 where
B. Particle size evolution (Rj=f(t)): Example with experimental data for extracted polyphenols
polyphenols extraction from Propolis
Vl ⋅ C l
The particle size evolution observed in ethanolic (M = ) is plotted versus the liquid/solid ratio
extraction from Propolis [23] is another effect related Ms
to polydispersity Mass transfer of Propolis constitu- (LSR).
ents [23, 29] is not thoroughly studied despite the great The dependence in Fig. 6 is treated as a nonlinear
interest to its bioactivity and the appropriate extraction function of the type:
technique [30-34] as well as the solvent effect [35] - 1
⋅ SLR
ethanol, water, nonethanolic mixture [36]. The results M= A
B
reported in ref. [23] confirm that partial dissolution of 1 + SLR
A
the solid matrix proceeds in parallel with the solid-liquid which is linearized in coordinates 1/M vs 1/LSR
extraction.
1 1
The mathematical description of the process is given ( =A + B ).
M SLR
by Eqs.(15): 1, 2, 3a, 4a, 6, 7 in Table 1. The solution
starts with the mean particle size of the initial PSD and The value of 1/B corresponds to the plateau of the

496
Iren Tsibranska, Bartosz Tylkowski

the first case as a result from milling the whole aerial


part of the plant. The course of the kinetic curve is sat-
isfactorily described as one-step diffusion controlled
mass transfer with a constant diffusion coefficient by a
model accounting for the polydispersity. In the second
case of predominantly spherical particles of less PSD
asymmetry the kinetics of extraction and particle size
evolution are successfully modeled on the ground of the
mean average size of PSD. The latter is regarded as time
varying because partial dissolution of the solid matrix
proceeds. The application of the simplified empirical
models in case of polydispersity and/or time evolution
of the solid particles may lead to unrealistic values of
Fig. 8. Experimental and calculated kinetic curves referring
the intra-particle mass transfer coefficient as evidenced
to polyphenols extraction from Propolis.
by the two systems studied. Models of better theoretical
initial curve in Fig.6, i.e. the maximum extractable mass grounds have greater potential in such cases in view of
of polyphenols per gram solid. This value is estimated the physical meaning, the predictive capabilities and
equal to 680 mg/g (R2 = 0.89) and is also checked by applicability to process optimization.
the experimental observation that extracted polyphenols
constitute 67.6 % of the solid substances dissolved [23]. References
The comparison between the experimental data
and that calculated by the model particle size evolution 1. Ch.-H. Chan, R. Yusoff, G.-Ch. Ngoh, Modeling and
(Eq.15 in Table 1) and extraction kinetic curves is illus- kinetics study of conventional and assisted batch
trated in Figs. 7 and 8 for liquid/solid ratio of 19.8. The solvent extraction, Chemical Engineering Research
whole soluble part of the solid is practically dissolved and Design, 92, 2014, 1169-1186.
in view of the previous experimental observations at a 2. Ch. Chilev, V. Koleva, E. Simeonov, A New Em-
liquid-solid ratio of 20 (ml g-1). The experimental kinetic pirical Model for Calculation the Effective Diffusion
curves in Figs. 7 and 8 are successfully fitted by the Coefficient for Solid−Liquid Extraction from Plants,
model solved for an effective intra-particle diffusion Industrial Engineering Chemistry and Research, 53,
coefficient of 1⋅10-13 m2 s-1. The initial value of the 2014, 6288-6296.
external mass transfer coefficient in these calculations 3. M. S. Guerrero, J. S. Torres, M. J. Nunez, Extraction
is 1.4⋅10-6 m s-1 (Bi = 440) which is in agreement with of polyphenols from white distilled grape pomace:
previous experimental observation [23] of the effective Optimization and modeling, Bioresource Technology,
mass transfer coefficient being of the order of 10–6 m s-1. 99, 2008, 1311-1318.
This coefficient is recalculated following the change of 4. K. R.Krishnan, M. Sivarajan, S. Babuskin, G. Arch-
the particle size according to Eq. (15.7). ana, P.A.S. Babu, M. Sukumar, Kinetic modeling of
spice extraction from S. aromaticum and C. cassia,
CONCLUSIONS Journal of Food Engineering, 117, 2013, 326-332.
5. R. Wongkittipong, L. Prat, S. Damronglerd, C. Gourdon,
The present study reports the mathematical de- Solid-liquid extraction of andrographolide from plants-
scription of solid-liquid extraction curves taking into experimental study, kinetic reaction and model, Separa-
account the presence of polydisperse particles. The tion and Purification Technology, 40, 2, 2004, 147-154.
results obtained are illustrated by two experimental batch 6. V. Sant’Anna, A. Brandelli, L.D.F. Marczak, I.C. Tessaro,
extraction systems where extraction of polyphenols by Kinetic modeling of total polyphenol extraction from
an ethanolic solvent is carried out (Sideritis ssp. L and grape marc and characterization of the extracts, Separa-
Propolis). Broad and asymmetric dispersion of particles tion and Purification Technology, 100, 2012, 82-87.
of a different shape (cylinders, plates) is observed in 7. D. Amendola, D.M. De Faveri, G. Spigno, Grape
497
Journal of Chemical Technology and Metallurgy, 51, 5, 2016

marc phenolics: Extraction kinetics, quality and sta- nal Russian Text © A.G. Egorov, A.B. Mazo, R.N.
bility of extracts, Journal of Food Engineering, 97, Maksudov, 2010, published in Teoreticheskie Osnovy
2010, 384-392. Khimicheskoi Tekhnologii, 44, 5, 2010, 498-506.
8. S. Kitanovic, D. Milenovic, V.B. Veljkovic, Empirical 19. I. Tsibranska, B. Tylkowski, R. Kochanov, K. Ali-
kinetic models for the resinoid extraction from aerial pieva, Extraction of biologically active compounds
parts of St. John’s wort (Hypericum perforatum L.), from Sideritis ssp. L., Food and Bioproducts Pro-
Biochemical Engineering Journal, 41, 2008, 1-11. cessing, 89, 4, 2011, 273-280.
9. W. Qu, Z. Pan, H. Ma, Extraction modeling and activi- 20. N.S. Berchane, K.H. Carson, A.C. Rice-Ficht, M.J.
ties of antioxidants from pomegranate marc, Journal Andrews, Effect of mean diameter and polydisper-
of Food Engineering, 99, 2010, 16-23. sity of PLG microspheres on drug release: Experi-
10. A. Minchev, S. Minkov, A model for determination ment and theory, International Journal of Pharmacy,
of the effective diffusion coefficient by the standard 337, 1, 2007, 118-126.
function technique, Journal of Applied Chemistry, 21. I.Tzibranska, Tannins extraction: I.Kinetics and dif-
57, 3, 1984, 717-720. fusion coefficients, Comptes Rendus de l‘Académie
11. A.R. Linares, S.L. Hase, M. L.Vergara, S.L. Resnik, Bulgare des Sciences, 53, 6, 2000, 71-74.
Modeling yerba mate aqueous extraction kinetics: 22. I.Tsibranska, E. Kostadinova, Recherche de la com-
Influence of temperature, Journal of Food Engineer- plexite minimale et utile pour un modele d‘adsorp-
ing, 97, 2010, 471-477. tion, J. Univ. Chem. Technol. Met. (Sofia), 37, 5,
12. A.M. Goula, Ultrasound-assisted extraction of 2002, 169-180.
pomegranate seed oil – Kinetic modeling, Journal 23. I. Tsibranska, B. Tylkowski, G. Peev, M. Giamberini,
of Food Engineering, 117, 2013, 492-498. R.Garcia-Valls, Mass transfer kinetics of biologi-
13. S. Jokić, D. Velić, M. Bilić, A. Bucić-Kojić, M. cally active compounds from Propolis, Bulgarian
Planinić, S. Tomas, Modelling of the Process of Chemical Communications, 44, 1, 2012, 64-69.
Solid-Liquid Extraction of Total Polyphenols from 24. L. Fiori, D. Basso, P. Costa, Seed oil supercritical
Soybeans, Czech Journal of Food Science, 28, 3, extraction: Particle size distribution of the milled
2010, 206-212. seeds and modeling, Journal of Supercritical Fluids,
14. A.Bucic-Kojic, M. Planinic, S. Tomas, M. Bilic, 47, 2008, 174-181.
D. Veli, Study of solid–liquid extraction kinetics of 25. J. M. del Valle, Extraction of natural compounds
total polyphenols from grape seeds, Journal of Food using supercritical CO2: Going from the laboratory
Engineering, 81, 2007, 236-242. to the industrial application, Journal of Supercritical
15. A. Sturzoiu, M. Stroescu, A. S. Guzun, T. Dobre, Fluids, 96, 2015, 180-199.
Empirical Models Applied for Kinetics Extraction 26. H. Sovová, Steps of supercritical fluid extraction
of β-carotene from Rosa canina, Revista di Chimie of natural products and their characteristic times,
(Bucharest), 62, 3, 201, 344-348. Journal of Supercritical Fluids, 66, 2012, 73-79.
16. J.E. Cacace, G. Mazza, Mass transfer process dur- 27. B. Tylkowski, B. Trusheva, V. Bankova, M.
ing extraction of phenolic compounds from milled Giamberini, G. Peev, A. Nikolovaa, Extraction of
berries, Journal of Food Engineering, 59, 2003, biologically active compounds from Propolis and
379-389. concentration of extract by nanofiltration, Journal
17. H.Sovova, Mathematical Model for Supercritical of Membrane Science, 348, 2010, 124-130.
Fluid Extraction of Natural Products and Extraction 28. S.E. Allaire, L.E. Parent, Size guide number and
Curve Evaluation, Journal of Supercritical Fluids, Rosin–Rammler approaches to describe particle size
33(1), 2005, 35-52. distribution of granular organic-based fertilizers,
18. A. G. Egorov, A. B. Mazo, R. N. Maksudov, Extraction Biosystems Engineering, 86, 4, 2003, 503-509.
from a Polydisperse Granular Layer of Milled Oil- 29. E. Mascheroni, V. Guillard, F. Nalin, L. Mora,
seeds with Supercritical Carbon Dioxide, Theoretical L.Piergiovanni, Diffusivity of Propolis compounds
Foundations of Chemical Engineering, 44(5), 2010, in Polylactic acid polymer for the development of
642–650. © Pleiades Publishing, Ltd., 2010. Origi- anti-microbial packaging films, Journal of Food

498
Iren Tsibranska, Bartosz Tylkowski

Engineering, 98, 2010, 294-301. cal and Biomedical Analysis, 81-82, 2013, 126-132.
30. I. Margereth, D.F. Suniarti, E.Herda, Z.A. Masud, 33. J. Xi, Z. Shouqin, Antioxidant activity of ethanolic
Optimization and comparative study of different extracts of Propolis by high hydrostatic pressure
extraction methods of biologically active compo- extraction, International Journal of Food Science
nents of Indonesian Propolis Trigona spp., Journal &Technology, 42.11, 2007, 1350-1356.
of Natural Products, 5, 2012, 233-242. 34. B. Trusheva, D. Trunkova, V. Bankova, Different
31. M. Jug, M. Z. Koncic, I. Kosalec, Modulation of anti- extraction methods of biologically active compo-
oxidant, chelating and antimicrobial activity of poplar nents from Propolis: a preliminary study, Chemistry
chemo-type Propolis by extraction procures, LWT Central Journal, 1.1, 2007, 1-4.
- Food Science and Technology, 57, 2014, 530-537. 35. L. Kubilenë, A. Marudka, V. Gailiûnaitë, The in-
32. F. Pellati, F. P. Prencipe, D. Bertelli, S. Benvenuti, An fluence of technological processes on extraction of
efficient chemical analysis of phenolic acids and flavo- chemical compounds of Propolis, Sveikatos Moks-
noids in raw Propolis by microwave-assisted extraction lai, 21, 2, 2011, 105-108.
combined with high-performance liquid chromatography 36. T. K. Sherwood, R.L. Pigford, Ch.R. Wilke, Mass
using the fused-core technology, Journal of Pharmaceuti- Transfer, McGraw-Hill, New York, 1975.

499

Vous aimerez peut-être aussi