Food Research International 119 (2019) 6–14

Contents lists available at ScienceDirect

Food Research International

journal homepage: www.elsevier.com/locate/foodres

Encapsulation and controlled release of hydrophobic flavors using T

biopolymer-based microgel delivery systems: Sustained release of garlic
flavor during simulated cooking
⁎
Minqi Wang, Takahiko Doi, David Julian McClements
Department of Food Science, University of Massachusetts, Amherst, MA 01003, United States

A R T I C LE I N FO A B S T R A C T

Keywords: The objective of the current study was to determine whether biopolymer microgels could be used to encapsulate
Microgels and control the release of allyl methyl disulfide (AMDS), a lipophilic compound in garlic, which has flavoring,
Encapsulation anticancer, antioxidant, and antimicrobial properties. AMDS is a volatile compound that is easily lost during
Flavor release food processing, storage, and preparation, which reduces its desirable functional attributes. In this study, AMDS
Retention
was loaded into oil-in-water emulsions that were then incorporated into biopolymer microgels. These microgels
Alginate
Nanoemulsions
were fabricated by injecting a mixture of AMDS-loaded lipid droplets and sodium alginate into a calcium ion
solution. The impact of microgel properties on the amount of AMDS retained during heating from room tem-
perature to boiling for 30 min was determined to simulate cooking conditions. Encapsulation of AMDS-loaded
lipid droplets in microgels delayed flavor release appreciably (3-fold longer). The microgels were also found to
remain intact throughout the cooking process. Moreover, flavor retention was improved when an optimum level
of lipid droplets (around 15%) were included in the microgels, which was attributed to their impact on flavor
partitioning and microgel stability. These results suggest that biopolymer microgels may be useful for controlling
flavor release during cooking.

1. Introduction stability, retention, and release of garlic flavors in foods.

Encapsulation technologies have been developed to protect flavors
Many of the volatile components in garlic have a characteristic from evaporation and degradation in the food industry (Desai & Park,
aroma, which makes it an important flavor component in various foods 2005; Gibbs, Kermasha, Alli, & Mulligan, 1999; Madene, Jacquot,
(Lee, Kim, & Lee, 2003). In addition, in vitro and in vivo studies have Scher, & Desobry, 2006; Reineccius, 1991). Many of these technologies
reported potential health benefits of certain garlic components, in- involve trapping the flavor molecules within colloidal particles speci-
cluding antioxidant (Andrada et al., 2008; Fanelli, Castro, de Toranzo, fically designed to inhibit their volatilization and degradation during
& Castro, 1998), anticancer (Sparnins, Barany, & Wattenberg, 1987), storage and cooking (Dordevic et al., 2015). A range of food-grade
antimutagenic (Rose, Whiteman, Moore, & Zhu, 2005), cholesterol- materials can be used to construct these colloidal particles, including
lowering (Lin, Salpietro, Deretey, & Csizmadia, 2000), and anti- polysaccharides (e.g., starches and gums), proteins (e.g., milk, egg,
microbial (Ross, O'Gara, Hill, Sleightholme, & Maslin, 2001; Tsai et al., meat, and plant proteins), and lipids (e.g., triacylglycerol, essential, and
1985) activities. Consequently, garlic components have potential for mineral oils). In addition, a range of fabrication technologies are
utilization as both flavor and nutraceutical ingredients in foods. How- available to assemble these materials into colloidal particles, including
ever, many of these components are easily lost during the manu- molecular complexation (Augustin & Hemar, 2009), homogenization,
facturing, storage, and preparation of foods because they have a rela- milling, injection, phase separation (Joye & McClements, 2014;
tively high volatility and are chemically unstable. The loss of these McClements, 2017), spray drying, spray chilling, extrusion (Castro
components occurs particularly quickly at elevated temperatures, and et al., 2016; Madene et al., 2006), and freeze drying (Buldur & Kok,
may lead to an undesirable change in the flavor profile of cooked foods, 2011) methods. The art and science of developing an effective delivery
as well as a change in biological activity. Consequently, there is con- system for a particular application is to identify the most appropriate
siderable interest in developing novel methods of controlling the wall materials and fabrication technology to create particles with the

⁎
Corresponding author at: Department of Food Science, University of Massachusetts Amherst, Amherst, MA 01003, USA.
E-mail address: [email protected] (D.J. McClements).

https://doi.org/10.1016/j.foodres.2019.01.042
Received 11 December 2018; Received in revised form 18 January 2019; Accepted 20 January 2019
Available online 22 January 2019
0963-9969/ © 2019 Elsevier Ltd. All rights reserved.
M. Wang et al. Food Research International 119 (2019) 6–14

desired encapsulation, protection, retention, and release properties. 2.3. Fabrication of alginate microgels
In this study, we focus on the utilization of biopolymer microgels
prepared from a natural polysaccharide (alginate) to encapsulate a Aqueous alginate (1% w/w) solutions were prepared by dissolving
model volatile lipophilic garlic component: allyl methyl disulfide powdered sodium alginate in distilled water and then stirring con-
(AMDS) (Yan, Wang, & Barlow, 1992). Alginate is a food-grade bio- tinuously at 60 °C for an hour. The solution was then mixed with the
polymer commonly used in the pharmaceutical industry to fabricate AMDS-loaded emulsion (1:1 mass ratio) for 2 h with continuous stirring
microgels designed to encapsulate and deliver drugs (Jain & Bar- to form a dispersion that contained 10% oil (w/w) and 0.5% alginate
Shalom, 2014; Lee & Mooney, 2012; Tonnesen & Karlsen, 2002). The (w/w). AMDS-loaded alginate microgels were then prepared using a
biopolymer network inside alginate microgels consists of linear anionic semi-automatic encapsulation unit (Encapsulator B-390, Buchi,
alginate chains held together by cationic calcium ions in an “egg-box” Switzerlands) with a nozzle size of 120 μm operated at a vibrating
structure (Ching, Bansal, & Bhandari, 2017; George & Abraham, 2006; frequency of 800 Hz, an electrode potential of 800 V, and a pressure of
Lee & Mooney, 2012; Sikorski, Mo, Skjak-Braek, & Stokke, 2007). The 250–300 mbar. The AMDS-loaded emulsion/alginate mixtures were
relative amounts of α-L-glucuronic acid (G) and β-D-mannuronic acid sprayed into 10 mL of 10% (w/w) calcium chloride solution with con-
(M) groups in the alginate molecules (i.e., the G/M ratio) determines tinuous stirring. The microgels formed were incubated in the calcium
the permeability and physicochemical properties of the microgels chloride solution for two hours at ambient temperature to promote
formed (Amsden, 1998; Smidsrod & Skjakbraek, 1990). The functional cross-linking of the alginate molecules. The microgels were then col-
properties of alginate microgels can also be varied by altering their lected by filtration and washed with distilled water and phosphate
particle size, shape, alginate concentration, calcium concentration, and buffer to remove any excess calcium ions from their surfaces.
additive contents (Ching et al., 2017). Consequently, there is con- Unfilled alginate microgels were fabricated using the same ap-
siderable scope for tailoring their properties to particular food appli- proach but without the addition of the emulsion. Briefly, 0.5% alginate
cations (Zhang, Zhang, Chen, Tong, & McClements, 2015). solution was extruded into 10% CaCl2 solution under continuous agi-
The main objective of the current study was to evaluate whether tation. The working parameters and washing steps were the same as
alginate microgels could be used to control flavor retention and release those for the preparation of the filled alginate microgels. Filled alginate
during simulated cooking. Consequently, lipophilic AMDS molecules microgels with oil levels of 2.5%, 5%, 15%, 20%, 30% were fabricated
were initially mixed with an oil phase and then converted into an oil-in- by incorporating different levels of emulsions (10:1 oil-to-emulsifier) in
water emulsion. The AMDS-loaded lipid droplets were then in- the alginate solutions prior to injection. All the other steps were the
corporated into biopolymer microgels fabricated from alginate using a same as already described.
simple injection method. We hypothesized that encapsulation of the
flavor molecules within biopolymer microgels would reduce the extent
of flavor loss during cooking. The impact of lipid droplet loading level 2.4. Simulated cooking conditions
on the physicochemical and retention properties of the microgels was
determined during simulated cooking. The information obtained in this Test samples (microgels or emulsions) were mixed with double
study may be useful for the development of novel encapsulation tech- distilled water with continuous stirring at 300 rpm (1:10 mass ratio) in
nologies to control the flavor profile of foods. a 400 mL beaker (height 110 mm, outer dimension 77 mm). The re-
sulting mixture was then heated using a hotplate (Isotemp digital stir-
2. Materials and methods ring hotplates, Fisher Scientific, Waltham, MA) from room temperature
to 100 °C with continuous stirring at 300 rpm until the end of the
2.1. Materials treatment. The change in temperature over time was measured using a
thermometer and recorded. Samples were collected throughout the
Corn oil was obtained from a regional supermarket and used as cooking process for analysis of flavor retention and particle properties.
received (Mazola, ACH Food Companies, Inc., Mississauga, ON The procedure was designed to simulate actual cooking conditions and
Canada). The following chemicals were purchased from the Sigma so some water evaporation will have occurred during this process.
Chemical Company (St. Louis, MO): alginic acid (sodium salt) (Lot#
SLBT1081, viscosity of 1% alginic acid in water is 4–12 cP mPa.s); Nile
Red (N3013-100MG); and, Fluorescein isothiocyanate isomer I. Allyl 2.5. Gas chromatography analysis
methyl disulfide was purchased from TCI America (Portland, OR).
Calcium chloride dehydrate was purchased from Fisher Science The headspace concentration of AMDS above the samples was de-
Education. Casein sodium salt was purchased from MP Biomedicals termined using gas chromatography (GC2010, Shimadzu, Columbia,
(Solon, OH). All chemicals used were of analytical grade. Double dis- MD) equipped with a headspace sampler (AOC-6000, Shimadzu,
tilled and deionized water was used to prepare all solutions. Columbia, MD). Samples (0.8 mL) were incubated at 50 °C for 10 min
before being exposed to a divinylbenzene/carboxen/poly-
2.2. Preparation of AMDS-loaded emulsions dimethylsiloxane (DVB/carboxen/PDMS) solid-phase microextraction
(SPME) fiber (50/30 μm, Supelco, Bellefonte, PA) for 1 min to adsorb
An oil-in-water emulsion was prepared using a method described volatile components. The volatile compounds collected were desorbed
previously (Zhang et al., 2016). Briefly, an aqueous phase was prepared for 2 min at 250 °C in the injector at a split ratio of 1:10, and then
by mixing 2% (w/w) sodium caseinate with 5 mM phosphate buffer separated on a fused-silica capillary column (30 m × 0.32 mm inner
(pH 7.0) and stirring for at least 2 h to ensure dissolution. An oil phase diameter × 1 μm) coated with 100% polydimethylsiloxane (Equity−1,
was prepared by dissolving AMDS (500 ppm) in corn oil. A coarse Sigma–Aldrich, Natick, MA). The GC column temperature program was
emulsion was then prepared by homogenizing 20% (w/w) oil phase as follows: initial temperature of 80 °C for 1 min, then a temperature
with 80% (w/w) aqueous phase using a high-shear mixer for 2 min ramp of 30 °C/min until a temperature of 100 °C was reached, followed
(M133/1281–0, Biospec Products, Inc., ESGC, Switzerland). The dro- by holding for 3 min. Helium was used as carrier gas at flow rate of
plet size in the coarse emulsion was then reduced by passing it through 26.6 mL/min. Concentrations were determined from peak areas using a
a high-pressure homogenizer (Microfluidizer, M110Y, Microfluidics, standard curve prepared using an AMDS standard. The retention of the
Newton, MA) with a 75-μm interaction chamber (F20Y) at an opera- flavors in the delivery systems was determined by the ratio of the re-
tional pressure of 12,000 psi (82.7 MPa) for 3 passes. The resulting maining to the initial quantity of flavors in the liquid phase during heat
emulsions were stored in a refrigerator at 4 °C prior to utilization. treatment as a function of time (Supplementary Material).

7
M. Wang et al. Food Research International 119 (2019) 6–14

2.6. Particle size analysis filled microgels and lipid droplets creamed to the top. This behavior is
due to the difference in densities of the colloidal particles in the various
The particle size distribution of the samples was measured using a delivery systems (Zhang et al., 2016). Corn oil (ρ = 920 kg m−3) has a
static light scattering device that measures the angular dependence of lower density than water (ρ = 1000 kg m−3), whereas alginate
the intensity of scattered light (Mastersizer 2000, Malvern Instruments (ρ = 1500 kg m−3) has a higher density (McClements, 2017). Conse-
Ltd., Malvern, Worcestershire, UK). Samples were diluted with phos- quently, the lipid droplets in the emulsions move upwards due to
phate buffer (5 mM, pH 7.0) prior to analysis to avoid multiple scat- gravity, whereas the unfilled alginate microgels move downwards.
tering effects. The refractive index of the particles used in the calcula- Presumably, the filled microgels moved upwards because the con-
tions was 1.507. The average particle sizes are reported as the volume- tribution of the lipid droplets to the overall density of the particles was
weighted mean diameter (d43). greater than the contribution of the alginate molecules (Matalanis &
McClements, 2013). There was also an appreciable difference in the
2.7. Microstructure analysis optical properties of the microgels depending on whether they con-
tained lipid droplets or not. The unfilled microgels appeared visibly
The microstructure of all systems was examined using optical and/ transparent whereas the filled microgels appeared opaque (white),
or confocal scanning laser microscopy with a 60× objective lens and which can be attributed to the influence of the lipid droplets on the
10 × eyepiece (Nikon D-Eclipse C1 80i, Nikon, Melville, NY, U.S.). A light scattering properties. The unfilled microgels have a refractive
small aliquot of sample was placed on a microscope slide and covered index fairly similar to that of water, and therefore do not scatter light
with a cover slip prior to analysis. For confocal microscopy, the oil strongly and appear relatively transparent. In contrast, the lipid dro-
phase in the microgels was dyed by adding about 0.1 mL of Nile red plets within the microgels have a relatively large refractive index
solution (1 mg/mL ethanol) to 2 mL of sample prior to analysis. contrast with the surrounding aqueous phase and have dimensions si-
Similarly, the protein phase in the microgels was dyed by adding about milar to the wavelength of light, and therefore scatter light strongly
0.1 mL of fluorescein thiocyanate isomer I (FITC) solution (1 mg/mL leading to an opaque whitish appearance.
dimethyl sulfoxide) to 2 mL of sample before analysis. The excitation The optical microscopy images showed that both the filled and
and emission wavelengths used for Nile red were 543 nm and 605 nm, unfilled microgels had spherical shapes with diameters ranging from
respectively while they were 488 nm and 515 nm for FITC, respectively. about 100 to 1000 μm (Fig. 1B). The filled microgels were slightly
The acquired microstructural images were analyzed using the image bigger than the unfilled ones, which suggests that the presence of the
analysis software associated with the microscope (NIS-Elements, Nikon, lipid droplets may have interfered with the formation of the biopolymer
Melville, NY). network inside the microgels.
Influence of simulated cooking on microstructure and particle size. After
2.8. Statistical analysis fabrication, the emulsions and microgels were subjected to simulated
cooking conditions, which involved heating them from room tem-
All experiments were carried out at least in triplicate using freshly perature to boiling, and then holding them for 30 min. The appearance
prepared samples and the results are expressed as the mean ± (Fig. 2A), microstructure (Fig. 3) and particle size distribution (Fig. 2B,
standard deviation. Statistical differences of the experimental results Fig. 3) of the emulsions and microgels were then measured after the
were determined by analysis of variance (ANOVA) using the SAS sta- samples had been exposed to the simulated cooking conditions.
tistical software package (SAS Inst. Inc., Cary, NC). The Duncan's Emulsions: After simulated cooking, large oil droplets and oiling-off
multiple-range test was used to determine differences between means, were visible at the top of the emulsion samples, suggesting that some
and p < .05 was considered to be statistically significant. droplet coalescence and phase separation occurred during heating.
Previous studies have also reported droplet coalescence and phase se-
3. Results and discussions paration in caseinate-stabilized oil-in-water emulsions after thermal
treatment (Liang et al., 2017). The particle size distribution was
3.1. Optimization of flavor analysis conditions monomodal before cooking, but became bimodal after cooking. More-
over, an increasing number of relatively large oil droplets
Initially, preliminary experiments were carried out to optimize the (d = 10–100 μm) was observed within the microscopy images as the
conditions used to quantify flavor release from the samples (emulsions boiling time increased, again indicating that droplet coalescence oc-
or filled microgels) during cooking. Head space analysis by gas chro- curred during heating. These results indicate that caseinate-coated lipid
matography was used to determine the concentrations of AMDS in the droplets are not stable to aggregation during long-term boiling.
emulsions and filled microgels. A DVB/CAR/PDMS fiber was selected to Microgels: After simulated cooking, the aqueous phase surrounding
detect this garlic flavor compound because of its higher extraction ef- the microgels became slightly cloudy, which suggested that some of the
ficiency of sulfur compounds compared to other fiber coatings(Kremr lipid droplets were released from them. Nevertheless, the microgels
et al., 2015). Chromatography analysis was carried out either directly themselves remained largely intact after exposure to boiling for 30 min,
on the samples or after the flavors had been extracted using a non-polar which suggested that they were relatively heat-stable. Moreover, the
solvent, i.e., ethanol, acetonitrile, hexane, or dimethyl sulfoxide. We particle size distribution of the microgel suspensions remained mono-
found no advantage to extracting the flavors from the test samples prior modal throughout the cooking process (Fig. 3B). However, we did ob-
to analysis, and so carried out all subsequent flavor analyses directly on serve a slight decrease in the mean diameter of the microgels after
the samples themselves. heating, which may have occurred because of shrinkage or surface
erosion during heating. A similar phenomenon has also been reported
3.2. Comparison of microgels to emulsions for glyceryl palmitostearate-loaded calcium alginate microgels during
heating (Pongjanyakul, Sungthongjeen, & Puttipipatkhachorn, 2006).
Initially, the physical properties, structure, and flavor retention Overall, these results suggest that the microgels have better heat-re-
profile of biopolymer microgels and oil-in-water emulsions were com- sistance during simulated cooking than the emulsions.
pared. Flavor retention. The retention of the flavor throughout the simu-
Initial properties: Preliminary experiments were carried out to de- lated cooking process was then measured using headspace analysis for
termine the properties of three colloidal delivery systems: emulsions; emulsions and microgels containing 10% oil phase (Fig. 4). The samples
unfilled microgels; and filled microgels (Fig. 1). After storage, the un- took about 10 min to increase from around 30 to 93 °C, and then re-
filled microgels sedimented to the bottom of the test tubes, whereas the mained at this temperature during the 30 min of cooking. The flavor

8
M. Wang et al. Food Research International 119 (2019) 6–14

Fig. 1. Appearances and particle size distributions of

different delivery systems prepared in buffer solu-
tions (5 mM phosphate buffer, pH 7). (A). From left to
right the photograph shows emulsions (10% oil),
unfilled alginate microgels, and emulsion-filled algi-
nate microgels (10% oil). (B). Particle size distribu-
tions and microstructures of different delivery sys-
tems at pH 7.

retention profiles were distinctly different for the emulsion and mi- which again slows down their release (Yven, Guichard, Giboreau, &
crogel samples. For the emulsions, there was only a slight reduction in Roberts, 1998). It should be noted that the lipid droplets coalesced and
flavor retention during the first 10 min as the sample reached boiling phase separated during heating, which would increase their effective
temperature, but then there was a steep decline, with almost no flavor size, and therefore may impact the release profile at longer boiling
remaining after 30 min. For the microgels, there was again only a slight times. Similarly, the microgel particles shrank during the simulated
reduction in flavor retention during the first 10 min, but then the de- cooking process, which may also alter their release profiles.
cline in flavor retention during boiling was much less steep than for the
emulsions. The time for 50% of the AMDS to be released from the 3.3. Influence of oil loading on microgel characteristics and flavor retention
particles (t50) in both colloidal delivery systems were calculated to
compare their flavor retention properties during cooking. The t50 value The model flavor used in this study (AMDS) is a relatively hydro-
for the microgels (12.6 min) was > 3-fold longer than that for the phobic molecule (Log P = 2.87), and therefore we postulated that its
emulsions (3.2 min), indicating that the alginate microgels were able to retention within the microgels would be increased at higher lipid
inhibit flavor loss during boiling. loading levels. For this reason, alginate microgels containing 2.5%, 5%,
There are a number of possible reasons for the delayed loss of the 10%, 15%, 20% and 30% lipid droplets were fabricated as described
flavor molecules from the microgels compared to the emulsions earlier, and then the impact of lipid loading level on their size, stability,
(McClements, 2017). First, the release rate of molecules from spherical and flavor retention during simulated cooking were assessed (Figs. 5 to
particles decreases as the particle size increases due to the longer dif- 7).
fusion path length and lower specific surface area (Li, Hu, Du, Xiao, & Impact on physical properties and structure. Visual observation of the
McClements, 2011). The diameter of the microgels (d ≈ 270 μm) was initial samples indicated that microgels containing 2.5% and 5% oil
initially over 1000-fold larger than the diameter of the lipid droplets sedimented to the bottom of the test tubes, whereas microgels con-
(d ≈ 0.26 μm) in the emulsions, which should therefore lead to a much taining 10% oil and higher floated to the top (Fig. 5A). These results
slower release rate. Second, the flavor molecules trapped in the lipid indicate that oil concentration played an important role in determining
droplets only have to travel through the oil phase before being released, the gravitational separation of this type of delivery system, which can
but in the filled microgels they must travel through the oil phase and be attributed to the fact that oil is lighter than water whereas alginate is
then through a biopolymer network, which may hinder their diffusion heavier. Interestingly, one would expect that there is a microgel com-
(Seuvre, Diaz, Cayot, & Voilley, 2004). Third, the flavor molecules may position (between 5% and 10% oil) where the particles have a similar
bind to the biopolymer network due to attractive physical interactions, density to that of the surrounding aqueous phase, which should inhibit

Fig. 2. (A). Appearance of AMDS-loaded emulsions and filled alginate microgels in phosphate buffer (pH 7) before and after heating. (B). The volume-weighted mean
particle diameters (d43) of emulsions and microgels before and after heating.

9
M. Wang et al. Food Research International 119 (2019) 6–14

Fig. 3. Influence of boiling time on particle size distribution and microstructures: (A). Emulsions (10% oil) prepared in buffer solution (5 mM phosphate buffer,
pH 7). (B). Filled microgels (10% oil) prepared in buffer solution (5 mM phosphate buffer, pH 7). For the confocal microscopy images: the scale bar is 100 μm, red
represents oil, and green represents protein. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this
article.)

gravitational separation. In future studies, we intend to determine if droplets acted as physical barriers that did not allow the alginate chains
there are particle compositions where density matching can be to get as close together as they would normally. These results are in
achieved, as stability to gravitational separation is important in many agreement with those reported previously for the influence of lipid
food applications. droplet levels on the properties of calcium alginate microgels formed by
All of the initial microgel samples had monomodal particle size injection methods (Chan, 2011). There was no difference in the visible
distributions (Fig. 5C) and mean particle diameters between about 270 appearance of microgels with different lipid levels before and after heat
and 410 μm (Fig. 5B). There was a general trend of an increase in mi- treatment – they all appeared as small white spheres (Fig. 5D). At all
crogel size with increasing lipid droplet level, which suggested that the lipid levels, there was an appreciable decrease in the mean particle
lipid droplets may have interfered with biopolymer network formation diameter of the microgels (from 6 to 35%) after simulated cooking
during the injection process. This may have happened because the lipid (Fig. 5B), which may have been due to some shrinkage or surface

Fig. 4. (A). Temperature profile and AMDS retention in oil-in-water emulsions (10% oil) and alginate microgels (10% oil) during heating in phosphate buffer (pH 7);
(B). Time for a 50% decrease in the AMDS level (t50) in the emulsions (10% oil) and microgels (10% oil) during heating in phosphate buffer (pH 7) (B).

10
M. Wang et al. Food Research International 119 (2019) 6–14

Fig. 5. Impact of lipid droplet level on: (A). Initial appearance; (B). Initial mean particle diameters (d43); (C). Initial particle size distributions; (D). Appearance before
and after heating; (E). Microstructure below and after heating.

erosion. Confocal microscopy indicated that the microgels remained which can be attributed to increased volatilization of the AMDS at
largely intact after simulated cooking (Fig. 5E), and that some of the elevated temperatures. Nevertheless, there were differences in the re-
lipid droplets remained trapped inside of them. However, there did tention-time profiles depending on the lipid level inside the microgels.
appear to be some irregularities on the surfaces of the microgels after At relatively low lipid levels (2.5% and 5%), the rate of flavor loss was
the heat treatment, especially at the higher lipid droplet levels used. much faster than at higher lipid levels (10% to 30%). The time for half
This may have occurred because the presence of the lipid droplets led to of the flavor to be lost from the microgels was estimated from the ex-
the formation of a weaker biopolymer network inside the microgels, perimental data of retention (R) versus time (t) by extrapolation. The
which was then more easily disrupted during heating. impact of lipid level on the retention half-time is shown in Fig. 6B. The
Impact on heat retention. The impact of the lipid droplet level inside retention half-time increased from around 15 min at the lowest lipid
the microgels on the flavor retention during the simulated cooking level (2.5%) to around 47 min at an intermediate lipid level (15%), but
process was measured (Fig. 6). In general, there was relatively little then decreased upon a further increase in lipid level. The initial in-
flavor loss during the initial cooking stage when the temperature of the crease in retention with lipid content may have been because there
samples was increased from ambient temperature to boiling, but then were more lipid droplets inside the microgels to solubilize the flavor
the amount of flavor retained by the microgels decreased appreciably, molecules, i.e., greater partitioning (see next section). Conversely, the

11
M. Wang et al. Food Research International 119 (2019) 6–14

Fig. 6. (A). Impact of lipid level on flavor retention versus time profiles for microgels during simulated cooking; (B). Impact of lipid level on retention half-time.

decrease in retention observed at higher lipid contents may have been 4.8Dπ 2t ⎤
R (t ) = 100 × exp ⎡−
because the lipid droplets interfered with the structural integrity of the ⎢
⎣ KDC d ⎥
2
⎦ (1)
alginate microgels, thereby allowing faster diffusion of the flavor mo-
lecules out of the microgels. These results suggest that there is an op- Here, R(t) is the percentage of flavor retained inside the spherical
timum lipid content that should be utilized to obtain a sustained release particle at time t, D is the translational diffusion coefficient of the flavor
profile. Interestingly, more rapid flavor release was observed for those molecules through the interior of the particle, d is the particle diameter,
microgels that underwent sedimentation (2.5 and 5% oil), than those and KDC is the equilibrium partition coefficient between the dispersed
that underwent creaming (≥ 10% oil). This may have been because phase (particles) and continuous phase (surrounding liquid). This
microgels at the bottom of the containers were subjected to higher equation assumes that the particles are introduced into a well-stirred
temperatures and more vigorous agitation than those at the top of the continuous phase that initially contains no flavor. A convenient para-
containers during cooking, which led to faster flavor release. meter to compare the ability of different particles to retain flavor mo-
lecules is the retention half-time (t1/2), which is the time for half of the
flavor molecules to diffuse out of the particles:
3.4. Theoretical predictions of flavor retention
0.0146d 2KDC
t1/2 =
D (2)
Theoretical models of molecular release from colloidal particles are
useful for designing and predicting the behavior of flavor delivery These equations can be used to provide some valuable insights into
systems. To a first approximation the retention of flavor molecules in- the main factors expected to impact the release of flavor molecules from
side a spherical particle can be modelled using the Crank model (Lian, biopolymer microgels. The equations show that the retention of the
Malone, Homan, & Norton, 2004). flavor molecules mainly depends on their partition coefficients,

Fig. 7. (A). Theoretical prediction of the impact of microgel size (10% oil droplets) on flavor retention; (B). Theoretical prediction of the impact of oil loading level
(d = 270 μm) on flavor release.

12
M. Wang et al. Food Research International 119 (2019) 6–14

diffusion rate through the particles, and the particle size. In general, the could be used to control the release of flavors in desserts, sauces, or
values of KDC and D depend on the composition and structure of the meat products. However, further work is still required to determine the
microgels. Expressions for these quantities have been derived for oil impact of the microgels on the sensory attributes (texture, appearance,
droplet-loaded microgels (Lian et al., 2004): stability, and mouthfeel) of foods, and to determine whether the mi-
crogels can be produced economically on a large scale.
2DG (1 − ∅O ) + K OW DO (1 + 2∅O )
D = DG
DG (2 + ∅O ) + K OW DO (1 − ∅O ) (3)
Acknowledgements
KDC = 1 − ∅O + ∅O K OW (4)
This material was partly based upon work supported by the National
Here, DG is the diffusion coefficient through the gel network, DO is Institute of Food and Agriculture, USDA, Massachusetts Agricultural
the diffusion coefficient through the oil phase, ϕO is the volume fraction Experiment Station (MAS00491).
of oil droplets inside the microgels, and KOW is the oil-water partition
coefficient of the flavor molecules. The above equations were used to Appendix A. Supplementary data
calculate the impact of microgel size and oil loading on the release rate.
In these calculations, it was assumed that the gel network primarily Supplementary data to this article can be found online at https://
consisted of water since the level of alginate present was only 0.5%, i.e., doi.org/10.1016/j.foodres.2019.01.042.
DG = DW. In these predictions, it was assumed that the viscosities of the
aqueous and oil phases were 1 and 50 mPas, respectively, and that the References
diffusion coefficients in these two phases were 2.19 × 10−10 m2 s−1
and 4.37 × 10−12 m2 s−1, respectively. The diffusion coefficients were Amsden, B. (1998). Solute diffusion within hydrogels. Mechanisms and models.
calculated using the Stokes-Einstein equation assuming an effective Macromolecules, 31(23), 8382–8395.
molecular radius of 1 nm for the flavor molecules. A KOW value of 741 Andrada, M. F., Martinez, J. C. G., Szori, M., Zamarbide, G. N., Vert, F. T., Viskolcz, B., ...
Csizmadia, I. G. (2008). Thermodynamics of competing oxidation reactions of allyl
was used in the predictions based on the reported logP value for ADMS methyl disulfide by hydrogen peroxide: A first principle molecular computational
of 2.87. study on the conformations of allyl methyl disulfide and its oxidized products. Journal
The theoretical predictions show that flavor retention should be- of Physical Organic Chemistry, 21(12), 1048–1058.
Augustin, M. A., & Hemar, Y. (2009). Nano- and micro-structured assemblies for en-
come more prolonged as the dimensions of the microgels increases capsulation of food ingredients. Chemical Society Reviews, 38(4), 902–912.
(Fig. 7A), and as the oil loading inside the microgels increases (Fig. 7B). Buldur, P. M., & Kok, F. N. (2011). Encapsulation of food flavors via coacervation method.
These predictions partly account for the experimental observation that Current Opinion in Biotechnology, 22, S96.
Castro, N., Durrieu, V., Raynaud, C., Rouilly, A., Rigal, L., & Quellet, C. (2016). Melt
flavor retention was more extended for the relatively large ADMS- Extrusion Encapsulation of Flavors: A Review. Polymer Reviews, 56(1), 137–186.
loaded microgels (d ≈ 270 μm) than for the ADMS-loaded lipid droplets Chan, E. S. (2011). Preparation of Ca-alginate beads containing high oil content: Influence
(d ≈ 0.26 μm) (Fig. 4A). Nevertheless, the retention-half time was not of process variables on encapsulation efficiency and bead properties. Carbohydrate
Polymers, 84(4), 1267–1275.
106-fold different between the two samples, as would be predicted form
Ching, S. H., Bansal, N., & Bhandari, B. (2017). Alginate gel particles-a review of pro-
Eq. 2 based on these differences in particle size. This may have been duction techniques and physical properties. Critical Reviews in Food Science and
because after being released from the colloidal particles (oil droplets or Nutrition, 57(6), 1133–1152.
Desai, K. G. H., & Park, H. J. (2005). Recent developments in microencapsulation of food
microgels), the flavor molecules still had to diffuse through the aqueous
ingredients. Drying Technology, 23(7), 1361–1394.
phase before reaching the top of the samples and evaporating. The Dordevic, V., Balanc, B., Belscak-Cvitanovic, A., Levic, S., Trifkovic, K., Kalusevic, A., ...
theoretical predictions suggest that the flavor retention should become Nedovic, V. (2015). Trends in Encapsulation Technologies for Delivery of Food
increasingly more extended as the oil droplet level inside the microgels Bioactive Compounds. Food Engineering Reviews, 7(4), 452–490.
Fanelli, S. L., Castro, G. D., de Toranzo, E. G. D., & Castro, J. A. (1998). Mechanisms of the
increases (Fig. 6B). However, our experiments showed that there was preventive properties of some garlic components in the carbon tetrachloride pro-
initial an increase in the retention time with increasing oil content, but moted oxidative stress. Diallyl sulfide, diallyl disulfide, allyl mercaptan and allyl
then a decrease (Fig. 6B). As mentioned earlier, this may have been due methyl sulfide. Research Communications in Molecular Pathology and Pharmacology,
102(2), 163–174.
to some change in the structure of the microgels at higher lipid droplet George, M., & Abraham, T. E. (2006). Polyionic hydrocolloids for the intestinal delivery of
levels, due to the ability of the droplets to interfere with the formation protein drugs: Alginate and chitosan - a review. Journal of Controlled Release, 114(1),
of the biopolymer network. It should also be noted that the microgels 1–14.
Gibbs, B. F., Kermasha, S., Alli, I., & Mulligan, C. N. (1999). Encapsulation in the food
used in the experiments were polydisperse, which may lead to a industry: A review. International Journal of Food Sciences and Nutrition, 50(3),
broadening of the retention-time profiles compared to the theoretical 213–224.
predictions (which assumed that the microgels were monodisperse). Jain, D., & Bar-Shalom, D. (2014). Alginate drug delivery systems: Application in context
of pharmaceutical and biomedical research. Drug Development and Industrial
Pharmacy, 40(12), 1576–1584.
4. Conclusions Joye, I. J., & McClements, D. J. (2014). Biopolymer-based nanoparticles and micro-
particles: Fabrication, characterization, and application. Current Opinion in Colloid &
Interface Science, 19(5), 417–427.
In conclusion, this study has shown that the retention of a model Kremr, D., Bajerova, P., Bajer, T., Eisner, A., Adam, M., & Ventura, K. (2015). Using
garlic flavor can be increased by dissolving it within lipid droplets that headspace solid-phase microextraction for comparison of volatile Sulphur com-
are then encapsulated inside filled alginate microgels. The microgels pounds of fresh plants belonging to families Alliaceae and Brassicaceae. Journal of
Food Science and Technology, 52(9), 5727–5735.
appeared to be physically stable to boiling, i.e., they remained intact Lee, K. Y., & Mooney, D. J. (2012). Alginate: Properties and biomedical applications.
after boiling for 30 min. Moreover, encapsulation delayed the release of Progress in Polymer Science, 37(1), 106–126.
the garlic flavor during boiling, which was attributed to the increase in Lee, S. N., Kim, N. S., & Lee, D. S. (2003). Comparative study of extraction techniques for
determination of garlic flavor components by gas chromatography-mass spectro-
the distance the flavor molecules had to diffuse out of the microgels. metry. Analytical and Bioanalytical Chemistry, 377(4), 749–756.
Theoretical analysis of flavor release suggested that flavor retention Li, Y., Hu, M., Du, Y. M., Xiao, H., & McClements, D. J. (2011). Control of lipase di-
could be improved by increasing the microgel size or the oil droplet gestibility of emulsified lipids by encapsulation within calcium alginate beads. Food
Hydrocolloids, 25(1), 122–130.
concentration. In practice, however, there appeared to be an optimum Lian, G. P., Malone, M. E., Homan, J. E., & Norton, I. T. (2004). A mathematical model of
droplet concentration at intermediate levels, which may have been volatile release in mouth from the dispersion of gelled emulsion particles. Journal of
because the fat droplets altered the structural integrity of the microgels Controlled Release, 98(1), 139–155.
Liang, Y. C., Gillie, G., Matia-Merino, L., Ye, A. G., Patel, H., & Golding, M. (2017).
at high levels.
Structure and stability of sodium-caseinate-stabilized oil-in-water emulsions as in-
The microgels developed in our study may be useful for controlling fluenced by heat treatment. Food Hydrocolloids, 66, 307–317.
flavor release in commercial food and beverage products during Lin, A. C., Salpietro, S. J., Deretey, E., & Csizmadia, I. G. (2000). Multidimensional
cooking processes that involve high temperatures. For instance, they conformational analysis of allyl methyl disulfide: A key component of garlic.

13
M. Wang et al. Food Research International 119 (2019) 6–14

Canadian Journal of Chemistry-Revue Canadienne De Chimie, 78(3), 362–382. compatible interactions in calcium-alginate gels from fiber X-ray diffraction.
Madene, A., Jacquot, M., Scher, J., & Desobry, S. (2006). Flavour encapsulation and Biomacromolecules, 8(7), 2098–2103.
controlled release - a review. International Journal of Food Science and Technology, Smidsrod, O., & Skjakbraek, G. (1990). Alginate as Immobilization Matrix for Cells.
41(1), 1–21. Trends in Biotechnology, 8(3), 71–78.
Matalanis, A., & McClements, D. J. (2013). Hydrogel microspheres for encapsulation of Sparnins, V. L., Barany, G., & Wattenberg, L. W. (1987). Effects of Allyl Methyl Disulfide
lipophilic components: Optimization of fabrication & performance. Food (Amd), Allyl Methyl Trisulfide (Amt), Propyl Methyl Disulfide (Pmd), and Propyl Methyl
Hydrocolloids, 31(1), 15–25. Trisulfide (Pmt) on Glutathione-S-Transferase (Gst) activity and Benzo(a)Pyrene (Bp)
McClements, D. J. (2017). Designing biopolymer microgels to encapsulate, protect and Carcinogenesis in mice. Vol. 28, Proceedings of the American Association for Cancer
deliver bioactive components: Physicochemical aspects. Advances in Colloid and Research142.
Interface Science, 240, 31–59. Tonnesen, H. H., & Karlsen, J. (2002). Alginate in drug delivery systems. Drug
Pongjanyakul, T., Sungthongjeen, S., & Puttipipatkhachorn, S. (2006). Modulation of Development and Industrial Pharmacy, 28(6), 621–630.
drug release from glyceryl palmitostearate-alginate beads via heat treatment. Tsai, Y., Cole, L. L., Davis, L. E., Lockwood, S. J., Simmons, V., & Wild, G. C. (1985).
International Journal of Pharmaceutics, 319(1–2), 20–28. Antiviral Properties of Garlic - Invitro Effects on Influenza B, Herpes-Simplex and
Reineccius, G. A. (1991). Carbohydrates for Flavor Encapsulation. Food Technology, 45(3), Coxsackie-Viruses. Planta Medica, 51(5), 460–461.
144. Yan, X. J., Wang, Z. B., & Barlow, P. (1992). Quantitative estimation of garlic oil content
Rose, P., Whiteman, M., Moore, P. K., & Zhu, Y. Z. (2005). Bioactive S-alk(en)yl cysteine in garlic oil based health products. Food Chemistry, 45(2), 135–139.
sulfoxide metabolites in the genus Allium: The chemistry of potential therapeutic Yven, C., Guichard, E., Giboreau, A., & Roberts, D. D. (1998). Assessment of interactions
agents. Natural Product Reports, 22(3), 351–368. between hydrocolloids and flavor compounds by sensory, headspace, and binding
Ross, Z. M., O'Gara, E. A., Hill, D. J., Sleightholme, H. V., & Maslin, D. J. (2001). methodologies. Journal of Agricultural and Food Chemistry, 46(4), 1510–1514.
Antimicrobial properties of garlic oil against human enteric bacteria: Evaluation of Zhang, Z. P., Zhang, R. J., Chen, L., Tong, Q. Y., & McClements, D. J. (2015). Designing
methodologies and comparisons with garlic oil sulfides and garlic powder. Applied hydrogel particles for controlled or targeted release of lipophilic bioactive agents in
and Environmental Microbiology, 67(1), 475–480. the gastrointestinal tract. European Polymer Journal, 72, 698–716.
Seuvre, A. M., Diaz, M. D. L. A., Cayot, P., & Voilley, A. (2004). Influence of the com- Zhang, Z. P., Zhang, R. J., Zou, L. Q., Chen, L., Ahmed, Y., Al Bishri, W., ... McClements, D.
position and the structure of different media on the release of aroma compounds. Le J. (2016). Encapsulation of curcumin in polysaccharide-based hydrogel beads: Impact
Lait, 84(3), 305–316. of bead type on lipid digestion and curcumin bioaccessibility. Food Hydrocolloids, 58,
Sikorski, P., Mo, F., Skjak-Braek, G., & Stokke, B. T. (2007). Evidence for egg-box- 160–170.

14