Chapter 12
Flavor Release from Emulsions and Complex Media 1
2
3
3
A.Voilley ,M.A.EspinosaDiaz ,C.Druaux ,andP.Landy 1
2
ENSBANA, Université de Bourgogne, 1 Esplanade Erasme, F-21000 Dijon, France Givaudan Roure, Ueberlandstrasse 138, CH-8600 Switzerland Unilever Research, Olivier van Noortlaan 120, 3133 AT Vlaardingen, the Netherlands 3
The food matrix plays an important role in controlling flavor release at each step of food product preparation and consumption. Flavor release depends on the ability of the aroma compounds to be in the vapor phase and therefore on their affinity for the product, which participates in their rate of transfer. This is the reason why many studies of physicochemical interactions between volatiles and other constituents of the matrix have been extensively reported in simple model systems. The aim of this paper is to clarify the thermodynamic and kinetic properties of volatiles not only in relation to the composition but also with regard to the microstructure in order to explain the phenomena involved in flavor release. The results obtained recently in our group in the case of emulsions and model cheeses will be discussed; especially the modeling of mass transfer at the interface between the lipid and the aqueous phases in the matrix.
Flavor release depends on the affinity of the odorants for the food product, and therefore on their availability for the vapor phase. This is the reason why studies of the physicochemical interactions that occur between volatile and other constituents of the matrix have been so thoroughly reviewed (7-3). Past research was focused on using sensory evaluation or instrumental measurements to gain a better understanding of the mechanisms that occur between aroma compounds and non-volatile substances. The systems considered were often very simple, consisting of an aroma compound and a single constituent, usually in an aqueous solution. In general, the presence of proteins, polysaccharides, and lipids reduces the volatility of aroma compounds with respect to that in pure water, whereas the presence of salts increases their volatilities. Few studies have been reported on the volatility of aroma compounds in emulsions (45).
142
© 2000 American Chemical Society
143
In physicochemical terms, key features influencing transfer and release are the presence of the interface between the aqueous and lipid phase, the surface area of the interface, and the nature of the surface active agent absorbed at this oil-water interface. Mathematical models describing flavor release from liquid emulsions have been developed by M c Nulty (6) and Harrison and Hills (7). The first model is based on mass balance and partition coefficients of aroma compounds in the emulsion, whereas the second one is based on the penetration theory of interfacial mass transfer where the transfer through the emulsion-gas interface is the rate-limiting step. The penetration theory takes into account that the boundary layers are often not completely stagnant and that there is also mass transport by eddy diffusion. Mass transfer between the phases takes place when a volume element from the bulk phase comes into contact with a phase boundary for a short fixed time . During this fixed time contact, mass transfer takes place by molecular diffusion. Subsequently, the volume element is remixed with the bulk phase and the whole process is repeated. The penetration model predicts that the mass transport coefficient varies with the square root of the diffusion coefficient. Under dynamic conditions, the square root is often nearer to the truth. The objective of this study is to show how various factors, especially composition and structure of the matrix, can influence flavor release from emulsion and complex media, and are also explained by modeling.
Materials and Methods Six aroma compounds have been selected and their physicochemical characteristics are given in Table I. They were provided by International Flavors and Fragrances (Longvic - France) and their purity was higher than 9 8 % .
Table I. Physicochemical Characteristics of Aroma Compounds Aroma compound
Formula
Molecular weight
Diacetyl
C H
2-Nonanone
C
Ethyl acetate
C H
Ethyl butyrate
CO Hi2 0
Ethyl hexanoate
C
4
9
H
4
8
H
0
6
1
8
1 6
exp.
calc*. 6266
2
86
7599.2
0
142
53.3
2
88 2
2
0
8
Saturated vapor pressure (Pa) 25°C
0
Solubility in Log Ρ water calc** (g/lOOmL) 25°C
25 ( 1 5 ° C )
-2.0
0.04
2.9
12265.5
8.6
0.6
116
1599.8
0.6
1.7
144
133.3
0.05
2.8
* Calculated from Lee-Kesler model (8) ** Calculated from Rekker method (9)
26.7
144
The chemicals and food ingredients (sodium caseinate, β-lactoglobulin, triolein, n-dodecane, miglyol) used in this study were of analytical grade and purchased from suitable suppliers in France. Processed cheese was prepared according to the technique described by Druaux (10) with essentially anhydrous milk fat, cheddar, and milk powder. Quantitative descriptive analysis on flavor attributes was performed on the cheeses, using a trained panel. The vapor-liquid partition coefficient or the volatility of aroma compounds was determined by equilibrium headspace analysis or by exponential dilution (77). This last method consists of exhausting the liquid phase of volatile compounds in equilibrium with the vapor phase. A n inert gas passed through the liquid phase and carried the volatile compound into the headspace. The system was thermostated at 25°C. A sample of the vapor phase was automatically injected into a gas chromatograph at regular intervals. The variation of the chromatographic peak of the solute is an exponential function of time, provided the detector response is linear. The diffusivity of aroma compounds was determined by using the Stokes cell (72). The cell with a porous diaphragm enables simple measurements and specifies the diffusivity within a low viscosity medium. This method is the most often used and consists of following the variation of concentration difference between solutions placed in two compartments separated by a fritted glass. "The rotating diffusion cell" technique was used to measure the mass transfer of solutes through diffusion layers and liquid-liquid interfaces (73). The rotating diffusion cell is designed hydrodynamically in such a way that stationary diffusion layers of known thickness are created on each side of the oil layer. The thickness of the stagnant aqueous layers Ζ (m) present at each side of the filter filled with oil is given by the Levich equation: 1 ι Ζ = 0.643 η D ïœ 6
_I 2
eq 1
aq
2
where
η = viscosity of the aqueous phase (m /s), D = solute diffusion coefficient in the aqueous phase (m /s), ω = rotation speed of filter (s ). The total or the overall resistance 1/k (m .s), or R of a solute diffusing from one aqueous phase to another through the oil layer is expressed as follows: 2
a q
1
_1
1 2Z 2 / R = -= +—+ k D ok oD ,P where
α ki 1 D Ρ
0
= = = = =
~ eq2
porosity of the filter (0.8), permeability coefficient (m/s), filter thickness (m), solute diffusion coefficient in oil (m /s), solute liquid-liquid partition coefficient 2
145
The significance of the three terms of eq. 2 is as follows: 2Z/D (denoted R ) describes the resistance to diffusion through the two stagnant aqueous diffusion layers of thickness Ζ that are established at each side of the filter, 2/ocki (denoted Ri) relates to the resistance due to the solute transfer across the two aqueous phase/oil interfaces, l / a D P (denoted R ^ ) is the contribution of the diffusion through the lipid in the filter. Eq.2 is then simplified: aq
aq
oil
eq 3 J~~