Flavour and Fragrance Journal

RESEARCH ARTICLE

Mechanistic Insight Into the Release Kinetics of

d-­Limonene From Electrosprayed Alyssum homolocarpum
Seed Gum Nanoparticles at Simulated Mouth Conditions
Khadije Khoshakhlagh1 | Mohebbat Mohebbi1 | Arash Koocheki1 | Alireza Allafchian2

1Departmentof Food Science and Technology, Ferdowsi University of Mashhad (FUM), Mashhad, Iran | 2Research Institute for Nanotechnology and
Advanced Materials, Isfahan University of Technology (IUT), Isfahan, Iran

Correspondence: Mohebbat Mohebbi ([email protected])

Received: 29 January 2024 | Revised: 29 July 2024 | Accepted: 31 August 2024

Keywords: diffusion | electrospraying | erosion | flavour release | mechanistic modelling | nanocapsules

ABSTRACT
The present study aimed at investigating and modelling the release behaviour of encapsulated d-­limonene from electrosprayed
Alyssum homolocarpum seed gum (AHSG) nanoparticles. For this purpose, release profiles of d-­limonene from 10% or 20% loaded
nanocapsules were obtained in deionised water and simulated mouth conditions by a spectrophotometric method. The experimental
results showed that complete release takes ranging between 7 and 15 min, depending on the d-­limonene loading and type of release
medium. A gradual increase in the flavour release rate over time without initial burst revealed there are several phenomena involved
in the release process due to the hydrophilic nature of AHSG nanocapsules. Based on these findings, a mechanistic approach mod-
elled flavour diffusion as a transport process of combined matrix swelling and erosion mechanisms. The model parameters were
obtained via best fitting procedure applying simulated annealing (SA) algorithm. The good correlation between the predicted and ex-
perimental results of d-­limonene release confirmed validation of the developed model. The simulation results showed that although
matrix erosion contributes more than diffusion process in d-­limonene release from AHSG nanocapsules, the model describing the
release mechanism as only governed by the erosion is not able to provide accurate predictions of flavour release as compared to the
coupled model.

1   |   Introduction carefully so that they provide the required functional attributes in

the final product [6].
Encapsulation is a well-­known technology within the pharmaceu-
tical, chemical, cosmetic, foods and printing industries [1, 2]. In The study of the release of encapsulated materials is one of the
the food industry, encapsulation can be used for different purposes most important aspects in evaluating the delivery systems. Several
including stabilising the sensitive ingredients during processing factors can affect the release behaviour of bioactive ingredient
and storage, enhancing health benefits of food products, masking from encapsulates. The most influential factors are properties of
the undesirable flavours and aroma and controlling the delivery bioactive substances (especially solubility) [7, 8], encapsulating
of the compound with a desired rate and at the specific target agent features (mechanical and physicochemical properties such
site [3–5]. Therefore, design of these systems must be performed as hydrophilicity/hydrophobicity) [7], initial loading, geometry

© 2024 John Wiley & Sons Ltd.

Flavour and Fragrance Journal, 2024; 0:1–12 1 of 12

https://doi.org/10.1002/ffj.3810
 10991026, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ffj.3810 by Gaziantep Universitesi, Wiley Online Library on [21/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
and morphology of the particles [9], the length scale of release (i.e., Briefly, the emulsions containing 0.5% (w/w) AHSG, 0.1% (w/w)
dimension of the matrix) and the conditions of release medium Tween 20 as emulsifier and 10% or 20% d-­limonene (based on
(such as pH, ionic strength and temperature) [10–12]. the gum weight) were prepared using a high-­speed homoge-
nizer (Heidolph Silent Crusher M Homogenizer, Germany) at
Based on the physical or chemical properties of encapsulating 12,000 rpm for 3 min. The fresh emulsions were electrosprayed
material, release mechanisms from a polymeric carrier can by applying a high voltage of 20 kV, at the constant flow rate of
be divided into three main categories: diffusion, swelling and 0.1 mL/h and 15 cm distance from the needle tip to collector in
erosion [13, 14]. Diffusional mass transport is almost always in- horizontal mode (MS-­2200, Medifusion, Korea).
volved in the controlled release systems, but its contribution on
the control of release is different depends on degradability of the
polymer matrix. In a nonbiodegradable system, diffusion is the 2.3   |   Morphological Characterisation
predominant mechanism [15]. For swelling-­controlled systems, of Nanocapsules
diffusion is enhanced due to water imbibition into the particles
which causes polymer disentanglement [16]. For biodegradable Particle size and morphology of the AHSG nanocapsules were
matrix, diffusion takes place in combination with polymer deg- characterised using a field-­emission scanning electron micros-
radation and erosion processes [17, 18]. copy. The deposited samples on the collector were sputter coated
with gold–palladium mixture under vacuum for 5 min and
Mathematical modelling plays an indispensable role in under- were evaluated by FE-­SEM (TESCAN MIRA 3 XMU, Czech),
standing of the delivery systems and relevant release mechanisms. operated at an accelerating voltage of 20 kV. The diameters of
In addition, with the help of modelling, it is possible to evaluate the the capsules on the photographs were measured by the ImageJ
effect of design parameters, such as the device size and geometry, software, Version 1.43. Average diameter was obtained from a
on the release behaviour of bioactive agent [17, 19]. The mathemat- minimum of 100 random measurements.
ical models describing the release profiles are classified as empiri-
cal models or mechanistic models [20]. Since physical mechanisms
are not considered for empirical modelling, these models give a 2.4   |   d-­Limonene Release Experiments
limited information on the mechanisms that control the release
process. On the contrary, mechanistic models which obtained by The release media used in this study were artificial saliva
decomposing the complex problem into its individual parts, pro- and deionised water. The artificial saliva was prepared by the
vide physical insight into the mechanisms involved [21, 22]. method of van Ruth, Roozen, and Cozijnsen [26]. The produced
saliva consisted of NaHCO3 (5.208 g), K 2 HPO 4 ·3H 2O (1.369 g),
Our previous studies have explored the production method and NaCl (0.877 g), NaN3 (0.500 g), KCl (0.477 g), CaCl 2·2H 2O
characterisation of electrosprayed Alyssum homolocarpum seed (0.441 g), mucin (2.160 g) and 200,000 U of α-­a mylase in 1 L
gum (AHSG) nanocapsules containing d-­limonene [23, 24]. The of distilled water. The pH of the produced saliva was adjusted
aim of the present study was an experimental and theoretical to pH 7.0 by using 0.1 mol/L NaOH or 0.1 mol/L HCl; then it
investigation to evaluate the release behaviour of encapsulated was centrifuged at 41g for 5 min to remove any undissolved
d-­limonene from AHSG nanocapsules. material.

In vitro release studies were performed in two different release

2   |   Materials and Methods media (artificial saliva or deionised water) to investigate the
effects of the type of release medium on the resulting flavour
2.1   |   Materials release kinetic. Twenty milligram of loaded nanocapsules were
dispersed in 5 mL of release media and continuously stirred
d-­Limonene (97%), hexane (HPLC grade, ≥ 95% purity), sodium at 100 s−1 with a magnetic stirrer at 37 ± 0.5°C for a specified
bicarbonate (NaHCO3, ≥ 99.5%), sodium chloride (NaCl, ≥ 99.0%), immersion period (ranging between 0.5 and 15 min). In order
sodium azide (NaN3, ≥ 99.5%), potassium chloride (KCl, ≥ 99.0%), to minimise the experimental error and to maintain the sink
calcium chloride dihydrate (CaCl2·2H2O, ≥ 99.0%), α-­ amylase condition, separate beakers containing the mixture of nanocap-
(from porcine pancreas, ≥ 5 units/mg solid) and mucin (from sules and release medium were considered for each sampling
porcine stomach, Type III: partially purified, bound sialic acid time. At the prescribed immersion time (0.5, 1, 1.5, 2, 2.5, 3, 5,
0.5%–1.5%) were purchased from Sigma-­Aldrich (USA). Tween 7, 9, 12 and 15 min), 2 mL of solution was withdrawn and mixed
20 (HLB = 16.7) was obtained from Merck (Darmstadt, Germany). with 1 mL of hexane. The withdrawn aliquot was centrifuged
A. homolocarpum seeds were purchased from a local medicinal at 1200 g for 20 min (EBA 270, Hettich Centrifuge, Germany),
plant market, Isfahan, Iran. The gum extraction was performed and, then, the supernatant which was hexane containing lim-
according to the optimised procedure described by Koocheki et al. onene, was subjected to measurement of absorbance at 252 nm
[25]. Freeze-­dried gum was milled, sieved, packed and kept in cool by using UV–Visible spectrophotometer (UNICO-­2100, USA).
and dry condition. The amount of released d-­limonene was quantified based on
the standard calibration curve of d-­limonene in n-­hexane.

2.2   |   Nanocapsules Preparation The cumulative percentage of d-­limonene released was calcu-

lated by dividing the cumulative amount of flavour released at
The production method of electrosprayed AHSG nanocap- time t (Mt) to the total weight of flavour loaded in the nanopar-
sules has been described in detail in our previous study [23]. ticles (M0):

2 of 12 Flavour and Fragrance Journal, 2024

 10991026, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ffj.3810 by Gaziantep Universitesi, Wiley Online Library on [21/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
∑t
Mt C is the time-­and position-­dependent flavour concentration in
Cumulative release percentage =
M
× 100 (1) the nanocapsules, t is time, Dt is the flavour diffusion coefficient
0
t=0
and R is the nanoparticle radius.

The initial flavour content in the nanocapsules (M0) was Initial and boundary conditions were as follows:
measured using the spectrophotometric method as described
above, but slight changes were made to measure the total C = C0 t=0 0<r<R (3)
amount of encapsulated limonene. About 30 mg of loaded
nanocapsules (10% or 20%) were mixed with 2 mL of water 𝜕C
=0 t>0 r=0 (4)
and 1 mL of hexane on a vortex shaker for 1 min. This blend 𝜕r
was heated at 45°C in a water bath with magnetic stirrer
C=0 t>0 r=R (5)
for 15 min. Cooled mixture was centrifuged at 4000 rpm for
20 min. The weight of encapsulated limonene was determined
Equation (3) indicates an initial homogeneous distribution of en-
with the same spectrophotometric method as that described
capsulated flavour over the entire volume of the nanocapsules.
above for the release assessments. The release tests were car-
Equation (4) indicates that symmetry conditions exist at the cen-
ried out with six replications.
tre of the capsules. Equation (5) describes that the concentration
of d-­limonene can be considered negligible on the surface of the
nanoparticles for times larger than zero (perfect sink conditions),
2.5   |   Mathematical Modelling of d-­Limonene because the lower initial flavour concentration than its solubility
Release limit [27].

2.5.1   |   Model Assumptions
In these conditions, Fick's second equation (Equation 2) was solved
analytically using separation of variable technique, that was:
As mentioned before, release of encapsulated compounds from
biodegradable particles is a complex process that involves many ∞
C − C0 D n2 𝜋 2 t
( )
physicochemical phenomena. To simplify the release problem − t 2
∑
= 1 + 2 ( − 1)n exp (6)
from AHSG nanocapsules, the following assumptions were Cs − C 0 n=1
R
made in this work:
where C0 is initial d-­limonene loaded and Cs is its concentration
• d-­Limonene-­loaded nanocapsules were assumed to be ini- on the surface of the nanocapsules.
tially completely spherical and radially symmetric.
To calculate the cumulative percentage of flavour released,
• Based on evidence from FTIR spectroscopy and DSC analy-
Equation (6) was written as follows:
sis results of AHSG nanocapsules, no interactions between
the polymer matrix and flavour were assumed [23]. ∞
Mt Dt n 2 𝜋 2 t
( )
6 ∑ 1
• It was considered that the flavour is initially homogeneously =1− 2 exp − (7)
M∞ 𝜋 n=1 n2 R2
distributed throughout the nanocapsules due to applying
emulsion electrospraying technique.
where Mt and M∞ are the total amount of d-­limonene released
• The flavour diffusion in the nanocapsules was assumed to from AHSG nanocapsules at time t and at infinite time, respec-
be Fickian. tively [29].
• The diffusion coefficient of d-­limonene within the AHSG
2.5.2.2   |   Swelling Phenomena. Due to hydrophilicity
nanocapsules depends on time, since polymer swelling and
nature of AHSG, upon contact with the release medium, water
disentanglement enhance the flavour mobility.
penetrates into the nanocapsules, which leads to polymer dis-
• The AHSG nanocapsules were considered to undergo the entanglement. The disentanglement of polymer chains will
bulk erosion process, due to the high hydrophilicity of cause the swelling of AHSG nanocapsules so that it allows an
AHSG and nano-­sized carrier systems [27, 28]. enhanced flavour diffusion [17].
• Perfect sink conditions were assumed throughout the release
To account for the effect of matrix swelling, a time-­dependent
experiment. This means that the flavour concentration in the
diffusion coefficient of d-­limonene was defined as Equation (8)
surrounding release medium can be considered negligible [17].
and replacing it in Equation (7):

(8)
( )
Dt = D0 exp kd t
2.5.2   |   Governing Equations
where D0 is the initial flavour diffusion coefficient, and kd is
2.5.2.1   |   Fickian Diffusion. For the mathematical descrip-
polymer disentanglement rate constant in the boundary condi-
tion of diffusional flavour release, Fick's second law under
tion at the interface with the surrounding medium [27].
nonsteady-­state condition for spherical geometry was applied:
( 2 ) 2.5.2.3   |   Erosion Mechanism. Erosion is a complex phe-
𝜕C 𝜕 C 2 𝜕C
= Dt + 0<r<R (2) nomenon that involves several physical and chemical processes,
𝜕t 𝜕r 2 r 𝜕r
such as the cleavage of the polymer chains, bond hydrolysis

3 of 12
 10991026, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ffj.3810 by Gaziantep Universitesi, Wiley Online Library on [21/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
and polymer mass loss. Based on the nature of the polymer 2.5.3   |   Solving and Validation of Model
matrix and its degradation rate, erosion mechanism is classified
as bulk (homogeneous) or surface (heterogeneous) erosion [30]. The resulting model given in Equation (11) contains six un-
Bulk erosion takes place throughout the eroding particle, while known parameters (i.e., D 0, kd, ke, tmax , φ d and φ e), which must
the surface erosion occurs at the external matrix boundary. have been estimated before model validation. These param-
When the rate of polymer degradation is slower than the rate eters were obtained for four release treatments by adjusting
of solvent penetration into the polymeric matrix, polymer will the model to the three replications of d-­limonene release ex-
undergo bulk erosion, whereas, surface erosion is expected, perimental data, using the least squares method (LSM). This
when the polymer tends to degrade faster compared to solvent model was a complex nonlinear problem with time-­dependent
penetration rate [17, 27]. In this study, due to the high hydro- terms and multiple variables, which necessitated the use of
philicity of AHSG and low diameter of the nanoparticles, a faster an efficient and effective optimisation method for data fitting.
rate of solvent penetration into the capsules was considered For this purpose, the simulated annealing (SA) algorithm
than the rate of polymer degradation and, thus, bulk-­eroding was used for the least squares function minimisation. SA is
behaviour was expected. one of the simplest and best-­k nown metaheuristic method for
solving the highly complicated combinatorial optimisation
In general, modelling of release profiles from surface-­eroding problems [34, 35]. SA algorithm is able to converge towards
devices is easier than bulk-­eroding types. Because in the first the global minimum successfully without the requirement to
case, erosion proceeds at constant rate at any time during ero- knowledge of the initial values and bounds of the model pa-
sion and the encapsulated substance is released simultaneously rameters. The developed mathematical model was written in
with the layer-­by-­layer matrix erosion from the outermost sur- MATLAB software (version 2017b, MathWorks, USA), then it
face. In contrast, bulk-­eroding systems have no constant erosion is optimised by hybridisation of two optimisation algorithms,
rate and thus, release from these matrices are more complicated that is, simulated annealing solver (simulannealbnd func-
to model [27, 31]. tion) and pattern search approach (options.HybridFcn = @
patternsearch).
To explain the d-­limonene release from bulk erosion AHSG
nanoparticle, a mechanistic model called the Prout–Tompkins To validation of the mathematical model, theoretical predictions
equation was used [32, 33], derived from the model with optimised parameters were com-
pared with the three independent experimental data sets of fla-
x
( )
ln = ke t − ke tmax (9) vour release.
1−x
The determination coefficient (R 2) and the squared root mean
here x is the fractional mass released at time t that is equivalent
square error (RMSE) were selected as the criteria to evaluate the
to the fraction of flavour released Mt/M∞, ke is the erosion rate
quality of the regressions and model validation.
constant, and tmax is the time required to reach maximum matrix
erosion rate.
2.5.4   |   Statistical Analysis
After substituting x in Equation (9) with Mt/M∞, the final equa-
tion to describe flavour release due to erosion was written as:
Statistical analysis of data was performed through one-­ way
( ( )) analysis of variance (ANOVA) using SPSS software. The signif-
Mt exp ke t − tmax
= (10) icant differences for each parameter were determined by LSD
M∞
( ( ))
1 + exp ke t − tmax test at p value < 0.05.

The overall fraction of flavour release at a given time t was

obtained by a linear combination of the equations given by 3   |   Results and Discussion
Equations (7) and (10) as follows,
3.1   |   Morphology of Nanocapsules
[ ∞ )]
Mt D n2 𝜋 2 t
(
6 ∑ 1 Figure 1 shows the FE-­SEM images and corresponding size
= 𝜑d 1 − 2 2
exp − t 2
M∞ 𝜋 n=1 n R distributions of the AHSG electrosprayed nanocapsules con-
(11) taining 10% or 20% d-­limonene. It is clear that the electrospray-
[ ( ( )) ]
exp ke t − tmax
+ 𝜑e ( ( )) ing of both d-­limonene/gum emulsions led to the formation of
1 + exp ke t − tmax
spherical nano-­sized capsules with narrow size distribution.
However, little differences were recognised in the particle size
where φ d and φ e are the diffusion and erosion mechanisms and capsule morphology. 10% d-­limonene emulsion produced
contribution fractions, respectively. In the analysis, to intro- capsules in the range of 35–70 nm with an average diameter
duce mathematical consistency it is necessary to consider the of 54.13 ± 2.08 nm. Furthermore, 10% d-­limonene capsules
relation, φ d + φ e = 1. This coupled model (Equation 11) is able showed a slight particles aggregation (Figure 1a). When the
to describe the three different mechanisms of d-­l imonene concentration of d-­limonene in emulsion increased to 20%, in-
release from AHSG nanocapsules occurring simultaneously dividual particles without the aggregation were obtained. In
that comprise flavour diffusion, matrix swelling and polymer addition, particle size distribution curve shifted to the right
erosion. (50–90 nm), and the mean diameter of capsules increased

4 of 12 Flavour and Fragrance Journal, 2024

 10991026, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ffj.3810 by Gaziantep Universitesi, Wiley Online Library on [21/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
FIGURE 1    |    FE-­SEM images and size distribution of AHSG nanocapsules with 10% (a) and 20% (b) d-­limonene loading.

significantly (p < 0.05) to 65.68 ± 8.80 nm (Figure 1b). Many were no initial burst releases (Figure 2). This phenomenon
studies with different encapsulation methods have also re- might be associated with the low d-­limonene content on the
vealed a direct relationship between the particle size and en- surface of the nanocapsules, as was previously measured [23].
capsulation loading [36–39]. More details on these findings are In addition, glass transition temperature (Tg) of AHSG nano-
included in our previous paper [23]. capsules has been estimated 47°C, which is higher than the
temperature of release medium (37°C) [23]. Thus, the polymer
was initially in glassy state characterised by restricted mobil-
3.2   |   d-­Limonene Release Profiles ity of polymer chains such that low free volume is available
for flavour diffusion, resulting in an extremely small initial
The d-­l imonene release from AHSG nanocapsules was evalu- release rate. As water penetrates into the nanocapsules, Tg de-
ated in deionised water (DW) or artificial saliva (AS) at 37°C. creases gradually with increasing water content and finally, a
Our previous study has shown a loading efficiency of around phase transition of biopolymer from glassy to rubbery state oc-
93% and 87% for 10% and 20% d-­l imonene-­loaded nanocap- curs once the Tg reaches below the temperature of the system.
sules, respectively [23]. These values were used to calculate This change of state results in polymer chain relaxation with
the release percentage of d-­l imonene (Equation 1). The release free volume expansion (matrix swelling), where the flavour
profiles of flavour showed a fairly fast release for all four treat- release is easier and faster as compared to that in the glassy re-
ments and complete release was achieved within 7–15 min gion [17, 40]. The release rate was accelerated at the later stage
(Figure 2). This behaviour was probably due to AHSG hydro- due to the degradation and erosion of AHSG matrix.
philic character, which results in a rapid matrix dissolution
in aqueous surrounding media [37]. On the other hand, this The flavour release rate from 10% loaded nanocapsules was al-
observation could be related to our previous ATR-­F TIR spec- ways slightly slower compared to 20% loaded samples in both
troscopy investigation of AHSG nanocapsules, which revealed media. The total amount of encapsulated d-­limonene was re-
no molecular interactions between encapsulated flavour and leased from 10% loaded nanocapsules over 9 and 15 min, in DW
matrix materials [23]. All four release experiments demon- and AS media, respectively, while increasing the initial loading
strated a slow-­release phase at the very beginning and there to 20%, led to a decrease of complete release time to 7 and 12 min,

5 of 12
 10991026, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ffj.3810 by Gaziantep Universitesi, Wiley Online Library on [21/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
These results can suggest that the matrix swelling and erosion
processes take place comparably slower in SA compared with
the DW medium, indicating a lower penetration rate of saliva
into the nanocapsules, which could probably result from the
higher viscosity and thus lower mobility of saliva due to the
presence of mucin. Furthermore, this conclusion is supported
by results from Kavanagh and Corrigan [46] who examined
the effect of dissolution medium variables, such as medium
composition and ionic strength on the swelling and erosion
of hydroxyl propyl methylcellulose (HPMC) matrices. They
reported that the rate of release medium uptake and erosion
of polymeric matrix decreased linearly with increasing ionic
strength of the medium [46]. Accordingly, AHSG nanocap-
sules in DW medium are more prone to water penetration,
swelling and subsequently faster matrix erosion due to the
very low ionic strength of deionised water (1 × 10 −7 M). In con-
trast, the relatively slower release of d-­limonene in AS could
possibly result from the higher ionic strength of saliva that
contained several different salts.

3.3   |   Estimation of Model Parameters

The model parameters were obtained by fitting the flavour

release data to the coupled model (Equation 11) are shown in
FIGURE 2    |    d-­Limonene release profiles from AHSG nanocapsules Table 1. The high coefficients of determination (R 2 > 0.99) and
with 10% or 20% loading in deionised water (a) and artificial saliva (b) low values of error (RMSE < 0.035) for all of the treatments indi-
at 37°C. cated the adequacy of this model to describe the release data and
acceptability of identified parameters.

in DW and AS solutions, respectively (Figure 2). This finding As observed in Table 1, the initial diffusion coefficient (D0) was
was inconsistent with some studies [41–45] as were reported that very low that ranged from 8.99 × 10 −20 to 1.05 × 10 −19 m2/s, prob-
the rate of bioactive release decreases by increasing initial level ably due to high hydrophobicity of d-­limonene oil. In addition,
of encapsulated substance due to smaller size, higher surface-­ extremely small diffusion coefficient values are caused by the
to-­volume ratios and consequently, shorter diffusion paths for glassy state of AHSG matrices at the early times of release pro-
particles with lower initial encapsulation loading. cess [13, 17].

In this case, the possible explanation for slower release rate The initial diffusivity of limonene within the AHSG matrix
from 10% loaded nanocapsules was less flavour mobility in the was independent of immersion medium composition, but was
biopolymer matrix of these samples due to their more compact significantly affected by flavour loading. The higher values of
structures and higher AHSG concentration. On the other hand, D0 for 20% loaded nanocapsules may be attributed to the more
20% d-­limonene nanocapsules provide a higher concentration mobility of d-­limonene molecules in these particles compared to
gradient, means greater driving force of the diffusion, and also 10% loaded samples, because of the less compact capsule shell
a lower density matrix, which are expected to degrade faster, with increasing d-­limonene to AHSG ratio.
and ultimately lead to flavour release more quickly.
The disentanglement rate constant (kd) decreased significantly
Beside the initial flavour loading, the type of release medium (p < 0.05) with increasing initial d-­limonene loading (Table 1).
was also found to affect the release behaviour of d-­limonene Smaller size of capsules containing 10% flavour are expected
from the AHSG electrosprayed nanocapsules. The release pro- to provide shorter penetration pathway for solvent molecules
files indicated that generally, the release of d-­limonene from and also higher surface area-­to-­volume ratio of these nanocap-
nanocapsules in AS exhibited a relatively gradual and sustained sules compared with those containing 20% flavour. This leads
trend over the period of experiments, whereas in DW medium, to a faster solvent penetration allowing more rapid polymer
after the initial slow stage of release, flavour release proceeded relaxation and swelling and thus, the higher kd values of 10%
faster (with a steeper slope) as compared to that in AS medium. loaded nanocapsules in both media. This observation was in
In comparison, the time to achieve 80% of limonene release (t80) consistence with the results obtained by Mahalakshmi et al. [47]
in AS were about 7.5 and 6 min while in DW, t80 took around 5.5 who observed that an excellent dissolution behaviour for zein
and 4 min for 10% and 20% loaded nanocapsules, respectively. electrosprayed nanocapsules than microparticles due to reduced
On the contrary, there were no clear differences in t15 (time to particle size and larger surface area.
achieve the 15% of release) between both media, implying that
the initial release mechanism was less affected by type of release Additionally, the estimated values of kd for release profiles in
medium than the later stages of release. deionised water were higher than those obtained for artificial

6 of 12 Flavour and Fragrance Journal, 2024

 10991026, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ffj.3810 by Gaziantep Universitesi, Wiley Online Library on [21/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
saliva, confirming that solvent penetration into the AHSG From Table 1, it can be drawn that by increasing d-­limonene
nanocapsules has occurred more rapidly for DW compared to loading, the erosion rate constant (ke) increases. This was likely
AS due to its lower viscosity and ionic strength as stated above attributed to the increased particle size of 20% loaded nano-
(Section 3.2). capsules as also has revealed a direct relationship between the

TABLE 1    |    Estimated parameters by fitting the mechanistic model to experimental data of d-­limonene release.

Parameter 10% Limonene, DW 10% Limonene, AS 20% Limonene, DW 20% Limonene, AS

D0 (m2/s) 8.99 × 10 −20 ± 0.00a 9.08 × 10 −20 ± 0.00a 1.05 × 10 −19 ± 0.00b 1.01 × 10 −19 ± 0.00b
kd (s−1) 4.39 × 10 −3 ± 0.00a 1.06 × 10 −3 ± 0.00b 7.03 × 10 −4 ± 0.00bc 2.29 × 10 −4 ± 0.00c
ke (s−1) 0.037 ± 0.007b 0.024 ± 0.004c 0.046 ± 0.008a 0.031 ± 0.000bc
φd 0.321 ± 0.034a 0.464 ± 0.021b 0.374 ± 0.028a 0.453 ± 0.050b
φe 0.679 ± 0.034a 0.536 ± 0.021b 0.626 ± 0.028a 0.547 ± 0.050b
tmax (s) 149.99 ± 0.00a 148.85 ± 5.27a 117.37 ± 2.68b 125.33 ± 4.10b
R2 0.990 0.993 0.997 0.998
RMSE 0.034 0.023 0.015 0.014
Note: Different letters within the same row indicate significant differences (p < 0.05).

FIGURE 3    |    Comparison of coupled model predictions (—) with experimental fraction of d-­limonene released (○). (a) 10% Loaded capsules in DW,
(b) 10% loaded capsules in AS, (c) 20% loaded capsules in DW, (d) 20% loaded capsules in AS.

7 of 12
 10991026, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ffj.3810 by Gaziantep Universitesi, Wiley Online Library on [21/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
particle size and matrix erosion rate by some other researchers a significant effect on erosion rate constant such that DW pro-
[33, 48, 49]. It may be related to less dense structure of larger vided higher values of ke compared to AS treatment. The higher
particles that allows more solvent medium penetrating into disentanglement rate (kd) at DW medium and thus faster inter-
the particles, resulting in a more rapid erosion rate as com- action of coating matrix with the surrounding media could be
pared to smaller capsules. Type of dispersion medium had also the reason of this behaviour.

TABLE 2    |    Results of fitting the mechanistic models to experimental data of d-­limonene release.

Model Evaluation criterion 10%, DW 10%, AS 20%, DW 20%, AS

Coupled model (considering three R2 0.987 0.996 0.997 0.997
mechanisms)
RMSE 0.082 0.048 0.033 0.035
Diffusion model (φd = 1, with constant R2 0.795 0.820 0.774 0.807
diffusivity)
RMSE 1.530 1.140 1.846 1.924
Erosion model (φe = 1) R2 0.985 0.855 0.996 0.879
RMSE 0.070 0.185 0.063 0.170
Diffusion–erosion model (with constant R2 0.977 0.975 0.993 0.985
diffusivity)
RMSE 0.071 0.046 0.063 0.034

FIGURE 4    |    Comparison between experimental fraction of d-­limonene released (○) and simulation results of a simple Fickian diffusion model
(—). (a) 10% Loaded capsules in DW, (b) 10% loaded capsules in AS, (c) 20% loaded capsules in DW, (d) 20% loaded capsules in AS.

8 of 12 Flavour and Fragrance Journal, 2024

 10991026, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ffj.3810 by Gaziantep Universitesi, Wiley Online Library on [21/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
As shown in Table 1, in all release profiles, the contribution of Based on these results, the developed model that combined dif-
matrix erosion (φe) was greater than the contribution of flavour fusion, swelling and erosion mechanisms is an appropriate ap-
diffusion (φd), implying that d-­limonene was released domi- proach to describe the d-­limonene release profiles from AHSG
nantly by the erosion process. This was consistent with our the- nanocapsules in both artificial saliva and deionised water media
oretical expectation that a non-­Fickian process was considered over the entire period of the experiments. The model provided
to describe the flavour release from nanocapsules because of the more accurate predictions of release behaviour for 20% d-­
high hydrophilic nature of AHSG [15, 50]. limonene capsules compared to 10% loaded samples (Table 2).
The predicted fractions of flavour released for 10% loaded nano-
The time to achieve maximum matrix erosion rate (tmax) de- capsules showed a little deviation from the experimental results
creased significantly with increasing the initial flavour load- in the early times of release process (Figure 3a,b). These samples
ing. The least value of tmax was 117.37 s, which identified for had an initial slow-­release phase possibly caused by particles ag-
limonene release from 20% loaded capsules into the DW me- gregation (as earlier illustrated in Figure 1a) and also by higher
dium, corresponding to the highest erosion rate constant contribution of AHSG in the matrix capsules, which make the
(ke = 0.046 s−1). flavour diffusion rate slower at the beginning compared to 20%
loaded capsules.

3.4   |   Model Validation
3.5   |   Simulation Results
After parameter fitting, model predictions were compared with
the independent experimental release data. Figure 3 and Table 2 Finally, the contribution of each mechanism to the d-­l imonene
shows a good agreement between experimental and predicted release from AHSG nanocapsules was examined by simulation
values for all of the release profiles (R 2 > 0.98 and RMSE < 0.082). studies. Figure 4a–d represents simulated fraction of flavour

FIGURE 5    |    Comparison between experimental fraction of d-­limonene released (○) and simulation results of an only bulk erosion model (—). (a)
10% Loaded capsules in DW, (b) 10% loaded capsules in AS, (c) 20% loaded capsules in DW, (d) 20% loaded capsules in AS.

9 of 12
 10991026, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ffj.3810 by Gaziantep Universitesi, Wiley Online Library on [21/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
release from a simple Fickian diffusion model (Equation 7) higher estimated values of φ d for AS medium as compared to
with constant diffusion coefficient by replacing Dt with De those that were calculated for release profiles in DW medium
(effective diffusivity). It can be obviously observed from (Table 1). Therefore, ignoring the diffusion process leads to
Figure 4 and Table 2 that this process alone does not correlate further reductions in accuracy of the model to predict the re-
well with the release behaviour of d-­l imonene over the entire lease profile in AS medium (Table 2).
period of four experiments (R 2 < 0.82 and RMSE > 1.14). The
first 2 min of release profiles could be only estimated relatively The last simulation model considered both diffusion and AHSG
accurately, where after this point the non-­F ickian transport erosion mechanisms, but flavour diffusion coefficient was as-
becomes more important and dominant in the release process sumed to be constant. This model provided better prediction
due to the swelling and erosion of the nanocapsule matrix. than the other two models (Figure 6 and Table 2) indicating
that a combination of diffusion and polymer degradation ade-
Figure 5a–d shows the simulation results of flavour release quately describes release process from polysaccharide carriers
by the mechanistic model when it is only governed by the [15]. Some other researchers have also used successfully the
nanoparticle erosion (Equation 10). It can be observed that the same approach for modelling the release of food bioactives or
fitting between the experimental release curves and simula- drugs from biodegradable matrices [3, 51–53]. These findings
tion results of the bulk erosion model does not correlate ap- allow simplifying the coupled model by considering a con-
propriately with the release behaviour of d-­limonene in the stant diffusion coefficient, especially for d-­l imonene release
initial stages, especially for artificial saliva medium, confirm- profiles in water medium, which have relatively less φ d than
ing that the mechanism of d-­limonene release was dominantly for the AS medium. This suggests the very low diffusion coef-
diffusion within the first seconds (Figure 5). The weaker cor- ficients of d-­limonene in AHSG nanocapsules cause that the
relation between measured release profiles in AS medium variation of De with time has no remarkable effect on the pre-
with simulated data from erosion model was in line with the diction accuracy of the model. A large reduction in the drug

FIGURE 6    |    Comparison between experimental fraction of d-­limonene released (○) and simulation results of a coupled model with constant
diffusivity (—). (a) 10% Loaded capsules in DW, (b) 10% loaded capsules in AS, (c) 20% loaded capsules in DW, (d) 20% loaded capsules in AS.

10 of 12 Flavour and Fragrance Journal, 2024

 10991026, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ffj.3810 by Gaziantep Universitesi, Wiley Online Library on [21/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
diffusion coefficient values by decreasing particle size at the 5. S. Safaeian Laein, I. Katouzian, M. R. Mozafari, et al., “Biological
nanometre scale has been previously demonstrated by Arifin, and Thermodynamic Stabilization of Lipid-­ Based Delivery Systems
Lee, and Wang [17]. Through Natural Biopolymers; Controlled Release and Molecular Dy-
namics Simulations,” Critical Reviews in Food Science and Nutrition 64
(2023): 7728–7747, https://​doi.​org/​10.​1080/​10408​398.​2 023.​2191281.

4   |   Conclusion 6. A. Matalanis, O. G. Jones, and D. J. McClements, “Structured

Biopolymer-­B ased Delivery Systems for Encapsulation, Protection,
and Release of Lipophilic Compounds,” Food Hydrocolloids 25, no.
In this work, a mechanistic model was developed to investigate 8 (2011): 1865–1880, https://​doi.​org/​10.​1016/j.​foodh​yd. ​2 011.​0 4.​014.
the release behaviour of d-­limonene from electrosprayed A. ho-
7. H. Ades, E. Kesselman, Y. Ungar, and E. Shimoni, “Complexation
molocarpum seed gum nanoparticles in deionised water and
With Starch for Encapsulation and Controlled Release of Menthone and
simulated mouth conditions. The experimental results presented Menthol,” LWT—Food Science and Technology 45, no. 2 (2012): 277–288,
that the release profiles of d-­limonene could be divided into two https://​doi.​org/​10.​1016/j.​lwt.​2 011.​08.​0 08.
phases which were the initial slow-­release phase, probably due
8. N. J. Zuidam and V. Nedovic, eds., Encapsulation Technologies for
to glassy state of AHSG matrices at the beginning of release pro-
Active Food Ingredients and Food Processing, vol. 410 (New York, NY:
cess, and the later fast-­release phase because of the important Springer, 2010), 127–160.
roles of swelling and erosion mechanisms in the d-­limonene
9. R. Pérez-­Masiá, R. López-­Nicolás, M. J. Periago, G. Ros, J. M. Lagaron,
release. Release profiles showed an increase in the d-­limonene
and A. López-­Rubio, “Encapsulation of Folic Acid in Food Hydrocol-
release rate with increasing initial flavour loading. Higher re- loids Through Nanospray Drying and Electrospraying for Nutraceutical
lease rates were also observed in deionised water with a lower Applications,” Food Chemistry 168 (2015): 124–133, https://​doi.​org/​10.​
viscosity and lower ionic strength compared to artificial saliva 1016/j.​foodc​hem.​2 014.​07.​051.
medium. The developed model takes into account the simul- 10. C. Maderuelo, A. Zarzuelo, and J. M. Lanao, “Critical Factor in the
taneous effects of flavour diffusion, matrix swelling and bulk Release of Drugs From Sustained Release Hydrophilic Matrices,” Jour-
erosion mechanisms to reach the highest prediction precision of nal of Controlled Release 154 (2011): 2–19, https://​doi.​org/​10.​1016/j.​jconr​
d-­limonene release profiles. The simulation results showed that el.​2 011.​0 4.​0 02.
the models considering only diffusion or erosion process as the 11. M. R. I. Shishir, L. Xie, C. Sun, X. Zheng, and W. Chen, “Advances
dominant release mechanism were not able to correlate appro- in Micro and Nano-­E ncapsulation of Bioactive Compounds Using
priately with the entire experimental release profiles. However, Biopolymer and Lipid-­B ased Transporters,” Trends in Food Science
neglecting the polymer disentanglement and assuming a con- & Technology 78 (2018): 34–60, https://​doi.​org/​10.​1016/j.​t ifs.​2 018.​
stant diffusion coefficient led to simplicity of the coupled model 05.​018.
with a slight deviation from actual results. 12. M. Wei, X. Pan, L. Rong, et al., “Polymer Carriers for Controlled
Fragrance Release,” Materials Research Express 7, no. 8 (2020): 082001,
https://​doi.​org/​10.​1088/​2 053-­​1591/​aba90d.

Conflicts of Interest 13. J. Siepmann and N. A. Peppas, “Modeling of Drug Release From
Delivery Systems Based on Hydroxypropyl Methylcellulose (HPMC),”
The authors declare no conflicts of interest. Advanced Drug Delivery Reviews 64 (2012): 163–174, https://​doi.​org/​10.​
1016/j.​addr.​2 012.​0 9.​028.
Data Availability Statement 14. M. P. Silva and J. P. Fabi, “Food Biopolymers-­Derived Nanogels
The data that support the findings of this study are available from the for Encapsulation and Delivery of Biologically Active Compounds: A
corresponding author upon reasonable request. Perspective Review,” Food Hydrocolloids for Health 2 (2022): 100079,
https://​doi.​org/​10.​1016/j.​f hfh.​2 022.​100079.
15. N. Malekjani and S. M. Jafari, “Release Modeling of Nanoencap-
References sulated Food Ingredients by Mechanistic Models,” In Release and Bio-
1. A. Madene, M. Jacquot, J. Scher, and S. Desobry, “Flavour Encapsula- availability of Nanoencapsulated Food Ingredients, edited by S. M. Jafari
tion and Controlled Release—A Review,” International Journal of Food (Cambridge, MA: Academic Press, 2020), 247–271, https://​doi.​org/​10.​
Science and Technology 41 (2006): 1–21, https://​doi.​org/​10.​1111/j.​1365-­​ 1016/​B978-­​0 -­​12-­​81566​5 -­​0.​0 0007​-­​2 .
2621.​2 005.​0 0980.​x. 16. D. Sharma, D. Dev, D. N. Prasad, and M. Hans, “Sustained Release
2. Z. A. Raza, S. Khalil, A. Ayub, and I. M. Banat, “Recent Develop- Drug Delivery System With the Role of Natural Polymers: A Review,”
ments in Chitosan Encapsulation of Various Active Ingredients for Mul- Journal of Drug Delivery and Therapeutics 9, no. 3-­s (2019): 913–923,
tifunctional Applications,” Carbohydrate Research 492 (2020): 108004, https://​doi.​org/​10.​22270/​​jddt.​v 9i3-­​s.​2859.
https://​doi.​org/​10.​1016/j.​carres.​2 020.​108004. 17. D. Y. Arifin, L. Y. Lee, and C. H. Wang, “Mathematical Modeling
3. A. I. Bourbon, M. A. Cerqueira, and A. A. Vicente, “Encapsula- and Simulation of Drug Release From Microspheres: Implications to
tion and Controlled Release of Bioactive Compounds in Lactoferrin-­ Drug Delivery Systems,” Advanced Drug Delivery Reviews 58, no. 12–13
Glycomacropeptide Nanohydrogels: Curcumin and Caffeine as Model (2006): 1274–1325, https://​doi.​org/​10.​1016/j.​addr.​2 006.​0 9.​0 07.
Compounds,” Journal of Food Engineering 180 (2016): 110–119, https://​ 18. J. Siepmann and F. Siepmann, “Modeling of Diffusion Controlled
doi.​org/​10.​1016/j.​jfood​eng.​2 016.​02.​016. Drug Delivery,” Journal of Controlled Release 161, no. 2 (2012): 351–362,
4. J. Gómez-­ Estaca, M. P. Balaguer, R. Gavara, and P. Hernandez-­ https://​doi.​org/​10.​1016/j.​jconr​el.​2 011.​10.​0 06.
Munoz, “Formation of Zein Nanoparticles by Electrohydrodynamic 19. H. Wen and K. Park, “Introduction and Overview of Oral Con-
Atomization: Effect of the Main Processing Variables and Suitability trolled Release Formulation Design,” In Oral Controlled Release For-
for Encapsulating the Food Coloring and Active Ingredient Curcumin,” mulation Design and Drug Delivery: Theory to Practice (Hoboken, NJ:
Food Hydrocolloids 28, no. 1 (2012): 82–91, https://​doi.​org/​10.​1016/j.​ John Wiley & Sons Inc., 2010), 1–19, https://​doi.​org/​10.​1002/​9 7804​
foodh​yd.​2 011.​11.​013. 70640​4 87.​ch1.

11 of 12
 10991026, 0, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ffj.3810 by Gaziantep Universitesi, Wiley Online Library on [21/11/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
20. C. Mircioiu, V. Voicu, V. Anuta, et al., “Mathematical Modeling of 34. D. Delahaye, S. Chaimatanan, and M. Mongeau, “Simulated An-
Release Kinetics From Supramolecular Drug Delivery Systems,” Phar- nealing: From Basics to Applications,” in Handbook of Metaheuris-
maceutics 11, no. 3 (2019): 140, https://​doi.​org/​10.​3390/​pharm​aceut​ics11​ tics, eds. M. Gendreau and J.-­Y. Potvin (New York, NY: Springer,
030140. 2019), 1–35.
21. D. Caccavo, S. Cascone, G. Lamberti, A. A. Barba, and A. Larsson, 35. R. E. Marsh, T. A. Riauka, and S. A. McQuarrie, “Use of a Simulated
“Drug Delivery From Hydrogels: A General Framework for the Release Annealing Algorithm to Fit Compartmental Models With an Applica-
Modeling,” Current Drug Delivery 14, no. 2 (2017): 179–189. tion to Fractal Pharmacokinetics,” Organization 7 (2007): 8.
22. E. Kaunisto, M. Marucci, P. Borgquist, and A. Axelsson, “Mechanis- 36. A. Amiri, A. Mousakhani-­Ganjeh, Z. Amiri, Y. G. Guo, A. P. Singh,
tic Modelling of Drug Release From Polymer-­Coated and Swelling and and R. E. Kenari, “Fabrication of Cumin Loaded-­Chitosan Particles:
Dissolving Polymer Matrix Systems,” International Journal of Pharma- Characterized by Molecular, Morphological, Thermal, Antioxidant and
ceutics 418, no. 1 (2011): 54–77. Anticancer Properties As Well As Its Utilization in Food System,” Food
Chemistry 310 (2020): 125821, https://​doi.​org/​10.​1016/j.​foodc​hem.​2 019.​
23. K. Khoshakhlagh, A. Koocheki, M. Mohebbi, and A. Allafchian,
125821.
“Development and Characterization of Electrosprayed Alyssum homo-
locarpum Seed Gum Nanoparticles for Encapsulation of D-­Limonene,” 37. N. Malekjani and S. M. Jafari, “Modeling the Release of Food Bio-
Journal of Colloid and Interface Science 490 (2017): 562–575, https://​doi.​ active Ingredients From Carriers/Nanocarriers by the Empirical,
org/​10.​1016/j.​jcis.​2 016.​11.​0 67. Semiempirical, and Mechanistic Models,” Comprehensive Reviews in
Food Science and Food Safety 20, no. 1 (2021): 3–47, https://​doi.​org/​10.​
24. K. Khoshakhlagh, M. Mohebbi, A. Koocheki, and A. Allafchian,
1111/​1541-­​4337.​12660​.
“Encapsulation of D-­L imonene in Alyssum homolocarpum Seed Gum
Nanocapsules by Emulsion Electrospraying: Morphology Charac- 38. R. Shtay, J. K. Keppler, K. Schrader, and K. Schwarz, “Encapsulation
terization and Stability Assessment,” Bioactive Carbohydrates and of (─)-­Epigallocatechin-­3 -­Gallate (EGCG) in Solid Lipid Nanoparticles
Dietary Fibre 16 (2018): 43–52, https://​doi.​org/​10.​1016/j.​b cdf.​2 018.​ for Food Applications,” Journal of Food Engineering 244 (2019): 91–100,
03.​0 01. https://​doi.​org/​10.​1016/j.​jfood​eng.​2 018.​0 9.​0 08.
25. A. Koocheki, S. A. Mortazavi, F. Shahidi, S. M. Razavi, R. Kadk- 39. X. Yang, N. Gao, L. Hu, J. Li, and Y. Sun, “Development and Evalua-
hodaee, and J. M. Milani, “Optimization of Mucilage Extraction From tion of Novel Microcapsules Containing Poppy-­Seed Oil Using Complex
Qodume Shirazi Seed (Alyssum homolocarpum) Using Response Sur- Coacervation,” Journal of Food Engineering 161 (2015): 87–93, https://​
face Methodology,” Journal of Food Process Engineering 33, no. 5 (2010): doi.​org/​10.​1016/j.​jfood​eng.​2 015.​03.​027.
861–882, https://​doi.​org/​10.​1111/j.​1745-­​4530.​2 008.​0 0312.​x.
40. M. L. Bruschi, “Mathematical Models of Drug Release,” in Strate-
26. S. M. van Ruth, J. P. Roozen, and J. L. Cozijnsen, “Comparison of gies to Modify the Drug Release From Pharmaceutical Systems, ed. M. L.
Dynamic Headspace Mouth Model System for Flavour Release From Bruschi (Cambridge, UK: Woodhead Publishing, 2015), 63–86.
Rehydrated Bell Pepper Cuttings,” in Trends in Flavour Research, eds.
41. E. Dadkhodazade, A. Mohammadi, S. Shojaee-­ A liabadi, A. M.
H. Maarse and D. G. van der Heij (Amsterdam, the Netherlands: Else-
Mortazavian, L. Mirmoghtadaie, and S. M. Hosseini, “Yeast Cell Mi-
vier, 1994), 59–64.
crocapsules as a Novel Carrier for Cholecalciferol Encapsulation: Devel-
27. L. L. Lao, N. A. Peppas, F. Y. C. Boey, and S. S. Venkatraman, “Mod- opment, Characterization and Release Properties,” Food Biophysics 13
eling of Drug Release From Bulk-­Degrading Polymers,” International (2018): 404–411, https://​doi.​org/​10.​1007/​s1148​3 -­​018-­​9546-­​3.
Journal of Pharmaceutics 418, no. 1 (2011): 28–41, https://​doi.​org/​10.​
42. M. Eltayeb, E. Stride, and M. Edirisinghe, “Preparation Characteriza-
1016/j.​ijpha​rm.​2 010.​12.​020.
tion and Release Kinetics of Ethylcellulose Nanoparticles Encapsulating
28. C. C. Lin and A. T. Metters, “Hydrogels in Controlled Release For- Ethylvanillin as a Model Functional Component,” Journal of Functional
mulations: Network Design and Mathematical Modeling,” Advanced Foods 14 (2015): 726–735, https://​doi.​org/​10.​1016/j.​jff.​2015.​02.​036.
Drug Delivery Reviews 58, no. 12–13 (2006): 1379–1408, https://​doi.​org/​
43. M. Hamdi, R. Nasri, S. Li, and M. Nasri, “Design of Blue Crab
10.​1016/j.​addr.​2 006.​0 9.​0 04.
Chitosan Responsive Nanoparticles as Controlled-­ Release Nanocar-
29. J. He, C. Zhong, and J. Mi, “Modeling of Drug Release From Bioerod- rier: Physicochemical Features, Thermal Stability and In Vitro pH-­
ible Polymer Matrices,” Drug Delivery 12, no. 5 (2005): 251–259, https://​ Dependent Delivery Properties,” International Journal of Biological
doi.​org/​10.​1080/​10717​5 4050​0176043. Macromolecules 145 (2020): 1140–1154, https://​doi.​org/​10.​1016/j.​ijbio​
mac.​2 019.​10.​039.
30. R. Langer and N. Peppas, “Chemical and Physical Structure of Poly-
mers as Carriers for Controlled Release of Bioactive Agents: A Review,” 44. S. F. Hosseini, M. Zandi, M. Rezaei, and F. Farahmandghavi, “Two-­
Journal of Macromolecular Science—Reviews in Macromolecular Chem- Step Method for Encapsulation of Oregano Essential Oil in Chitosan
istry and Physics 23, no. 1 (1983): 61–126, https://​doi.​org/​10.​1080/​07366​ Nanoparticles: Preparation Characterization and In Vitro Release
57830​8079439. Study,” Carbohydrate Polymers 95 (2013): 50–56, https://​doi.​org/​10.​
1016/j.​carbp​ol.​2 013.​02.​031.
31. F. Von Burkersroda, L. Schedl, and A. Göpferich, “Why Degrad-
able Polymers Undergo Surface Erosion or Bulk Erosion,” Biomaterials 45. S. M. Hosseini, H. Hosseini, M. A. Mohammadifar, et al., “Prepa-
23, no. 21 (2002): 4221–4231, https://​doi.​org/​10.​1016/​S 0142​-­​9612(02)​ ration and Characterization of Alginate and Alginate-­Resistant Starch
00170​-­​9. Microparticles Containing Nisin,” Carbohydrate Polymers 103 (2014):
573–580, https://​doi.​org/​10.​1016/j.​carbp​ol.​2 013.​12.​078.
32. J. F. Fitzgerald and O. I. Corrigan, “Mechanisms Governing Drug
Release From Poly-­α-­Hydroxy Aliphatic Esters: Diltiazem Base Release 46. N. Kavanagh and O. I. Corrigan, “Swelling and Erosion Properties
From Poly-­Lactide-­Co-­Glycolide Delivery Systems,” in Polymeric Deliv- of Hydroxypropylmethylcellulose (Hypromellose) Matrices-­Influence
ery Systems, Properties and Applications, eds. M. A. El-­Nokaly, D. M. of Agitation Rate and Dissolution Medium Composition,” International
Piatt, and B. A. Charpentier (Washington, DC: American Chemical So- Journal of Pharmaceutics 279 (2004): 141–152, https://​doi.​org/​10.​1016/j.​
ciety, 1993), 311–326, https://​doi.​org/​10.​1021/​bk-­​1993-­​0520.​ch023​. ijpha​rm.​2 004.​0 4.​016.
33. A. Lucero-­ Acuña and R. Guzmán, “Nanoparticle Encapsulation 47. L. Mahalakshmi, M. M. Leena, J. A. Moses, and C. Anandhara-
and Controlled Release of a Hydrophobic Kinase Inhibitor: Three makrishnan, “Micro-­and Nano-­Encapsulation of β-­Carotene in Zein
Stage Mathematical Modeling and Parametric Analysis,” International Protein: Size-­Dependent Release and Absorption Behavior,” Food &
Journal of Pharmaceutics 494, no. 1 (2015): 249–257, https://​doi.​org/​10.​ Function 11, no. 2 (2020): 1647–1660, https://​doi.​org/​10.​1039/​C9FO0​
1016/j.​ijpha​rm.​2 015.​07.​0 49. 2088H​.

12 of 12 Flavour and Fragrance Journal, 2024