8F Interactions Between Wine Matrix Macro-Components and Aroma Compounds 419
Dynamic headspacemethods involving bubbling an inert gas through a dilute
binary solution, e.g. exponential dilution methods, have been used to determine the
activity coefficient at infinite dilution (∞). For example, this method has been used
by Langourieux and Crouzet (1997) to determine how wine polysaccharides mod-
ify the aroma vapour-liquid equilibrium and Dufour and Bayonove (1999b) have
used this method to study the influence of polyphenols on it. Lubbers et al. (1994b)
have used aheadspace stripping at equilibrium methodfor a similar purpose. This
approach requires an external calibration by injection of liquid standard solutions
into the gas chromatograph.
Most of thestatic headspace methodsdetermine the partition coefficient by quan-
tifying volatile concentration above a sample by gas-chromatography. Thevapour
phase calibration method (VPC)uses an external vapour standard for calibration.
One must assure that the pure component is completely vaporized before injection.
A widely employed alternative is theLiquid calibration static headspace (LC-SH)
method(Voilley et al. 1991; Nedjma 1997). A third approach uses HS-SPME. SPME
may be used to determine partition coefficients if short sampling times are applied:
the process must only samplethe headspace and not disrupt the equilibrium (Jung
and Ebeler 2003). This method has become very popular to study the effect of wine
macromolecules on the liquid-vapor equilibrium, (Whiton and Zoecklein 2000;
Escalona et al. 2002; Hartmann et al. 2002; Aronson and Ebeler 2004).
Some static headspace methods do not require an external calibration and are
based on measurements performed at thermodynamic equilibrium between liquid
and gas phase. In thephase ratio variation method (PRV)described by Ettre and
Collaborators (1993), the partition coefficient calculation is based on the fact that
the headspace concentration changes as a function of the phase volume ratio (gas
and liquid phases), while the partition coefficient remains constant. This method
has been recently applied to study the interactions between aroma compounds and
macromolecules in different food systems (Savary et al. 2006, 2007) but so far not
to the wine.
Ath`es et al. (2004) compared the data from three static headspace methodologies
(VPC, PRV and LC-SH) for determining gas/liquid partition coefficients of two
aroma compounds in hydroalcoholic, multicomponent solutions at “infinit” dilution.
They found that PRV was a simpler method compared to VPC and LC-SH and
that VPC and PRV were more accurate than LC-SH since errors due to gas leaks
and adsorption in gastight syringes are avoided. They suggested that these issues
could be responsible for significant bias (50% lower values) obtained when using
the LC-SH method. Nevertheless, all three methods were able to find an effect of
ethanol (up to 20%) on the release of aroma compounds from their model system
(Fig. 8F.1).
Aznar et al. (2004) developed a static headspace APCI-MS methodology, employ-
ing ethanol such as make up gas to act asthe proton transfer reagent ion. Using
ethanol as the reagent gas avoided the problem of high levels of ethanol in the
sample (wine, beer and other alcoholic beverages) relative to other volatile com-
pounds. Using this technique they found that ethanol decreased the partition coef-
ficients of most of the aroma compounds. The degree of reduction was related to