Structured fluids as microreactors for flavor formation by the maillard reaction

Article abstract of DOI:10.1021/jf991254a

Thermal reactions of cysteine/furfural and cysteine/ribose mixtures were studied in model systems to gain more insight into the influence of structured fluids such as L₂ microemulsions and cubic phases on the generation of aroma compounds. Formation of 2-furfurylthiol from cysteine/furfural was particularly efficient in L₂ microemulsions and cubic phases compared to aqueous systems. The reaction led to the formation of two new sulfur compounds, which were identified as 2-(2-furyl)thiazolidine and, tentatively, N-(2-mercaptovinyl)-2-(2-furyl)thiazolidine. Similarly, generation of 2-furfurylthiol and 2-methyl-3-furanthiol from cysteine/ribose mixtures was strongly enhanced in structured fluids. The cubic phase was shown to be even more efficient in flavor generation than the L₂ microemulsion. It was denoted 'cubic catalyst' or 'cubic selective microreactor'. The obtained results are interpreted in terms of a surface and curvature control of the reactions defined by the structural properties of the formed surfactant associates.

Full text of DOI:10.1021/jf991254a

4808
J. Agric. Food Chem. 2000, 48, 4808−4816
Str u ctu r ed F lu id s a s Micr or ea ctor s for F la vor F or m a tion by th e
Ma illa r d Rea ction
S. Vauthey, Ch. Milo, Ph. Frossard, N. Garti, M. E. Leser, and H. J . Watzke*
Nestle´ Research Centre, Nestec Ltd., Vers-chez-les-Blanc, P.O. Box 44, 1000 Lausanne 26, Switzerland
Thermal reactions of cysteine/furfural and cysteine/ribose mixtures were studied in model systems
to gain more insight into the influence of structured fluids such as L2 microemulsions and cubic
phases on the generation of aroma compounds. Formation of 2-furfurylthiol from cysteine/furfural
was particularly efficient in L2 microemulsions and cubic phases compared to aqueous systems.
The reaction led to the formation of two new sulfur compounds, which were identified as 2-(2-furyl)-
thiazolidine and, tentatively, N-(2-mercaptovinyl)-2-(2-furyl)thiazolidine. Similarly, generation of
2-furfurylthiol and 2-methyl-3-furanthiol from cysteine/ribose mixtures was strongly enhanced in
structured fluids. The cubic phase was shown to be even more efficient in flavor generation than
the L2 microemulsion. It was denoted “cubic catalyst” or “cubic selective microreactor”. The obtained
results are interpreted in terms of a surface and curvature control of the reactions defined by the
structural properties of the formed surfactant associates.
Keyw or d s: Microemulsion; cubic phase; structured fluids; Maillard reaction; formation of thiols;
flavor precursors; 2-furfurylthiol
INTRODUCTION
chosen in our study to gain more insight into the
influence of structured media on the generation of
aroma compounds. The use of model systems is the
primary tool to investigate the effect of structured fluids
on the Maillard reaction.
Structured fluids are generally composed of micro-
structures dispersed in a homogeneous phase. Much
attention was given in the past to association colloids,
that is, aggregates formed by amphiphilic molecules
(Hiemenz, 1986; Larsson, 1994; Laughlin, 1994) such
as micelles, vesicles, microemulsions, mesophases, and
macromolecule self-assemblies.
Monoglycerides were used in this work as surfactant
for the generation of association colloids used as mi-
croreactors for the Maillard reactions. Monoglycerides
consist of a single acyl chain (hydrophobic tail), which
is attached to a glycerol moiety (hydrophilic headgroup).
The headgroup is only weakly hydrated. The acyl chain
of the monoglycerides may differ in the degree of
unsaturation: It can be totally saturated or have either
one (e.g., oleic acid) or two (e.g., linoleic acid) double
bonds. When dispersed in water, monoglycerides are
known to exhibit a rich polymorphism; that is, they form
up to six distinct phases depending on the water content
and the temperature (Krog, 1997; Briggs et al., 1996;
Boyle and German, 1996). The degree of unsaturation
mainly determines at which temperature the different
phases are formed. For instance, with the saturated
monoglycerides the cubic phase is formed only at high
temperatures (>70-80 °C), whereas in the presence of
unsaturated monoglycerides, it is formed at room tem-
perature (see Figure 1). Moreover, the formation of a
microemulsion phase (isotropic fluid) at high tempera-
tures is best achieved with unsaturated monoglycerides.
The structures of the different phases have been inves-
tigated by a variety of techniques, such as small-angle
X-ray scattering (SAXS), small-angle neutron scattering
(SANS), and microscopy. Because they have both a
Maillard reactions, for example, thermally induced
reactions of amino acids and reducing sugars, have been
studied for a long time, also with respect to their
potential to generate flavors. It should be emphasized
that the thermal reaction between cysteine and carbo-
hydrates leads to the formation of >200 volatile com-
pounds (Mulders, 1973; Farmer et al., 1989; Mottram
and Whitfield, 1995). Of particular interest is the
generation of sulfur-containing aroma compounds such
as 2-furfurylthiol (FFT) and 2-methyl-3-furanthiol (MFT).
Both compounds are potent odorants with unique
sensory properties and were determined by aroma
extract dilution analysis (AEDA) as major contributors
to the aroma of a wide variety of food products including
coffee (Holscher et al., 1990), cooked beef (Gasser and
Grosch, 1988), and roasted sesame seeds (Schieberle,
1993). It was also shown that FFT and MFT belong to
the key impact compounds of heated cysteine/ribose
mixtures (Hofmann and Schieberle, 1995). Quantitation
of these compounds by stable isotope dilution assays in
dry-heated and aqueous reaction systems revealed a
strong influence of the reaction conditions on the
concentrations of these potent odorants (Schieberle and
Hofmann, 1998). Particularly, the formation of FFT was
strongly enhanced under the conditions of dry heating,
which reflected roasting conditions. Other parameters
affecting their generation, such as pH dependence, have
also been studied (Schieberle and Hofmann, 1996), as
well as the effectiveness of different sugar-derived
degradation products as precursors (Hofmann and
Schieberle, 1996). Furfural, a thermal pentose sugar
degradation product, was shown to be an intermediate
in the formation of FFT (Mu¨nch et al., 1997) and was
* Author to whom correspondence should be addressed
(telephone +21/785-8026; fax +21/785-8554; e-mail
heribert.watzke@rdls.nestle.ch).
10.1021/jf991254a CCC: $19.00 © 2000 American Chemical Society
Published on Web 09/07/2000

Structured Fluids as Microreactors for Flavor Formation
J. Agric. Food Chem., Vol. 48, No. 10, 2000 4809
F igu r e 1. Phase diagrams of distilled monoglyceride products from Danisco Ingredients: (a) from hydrogenated lard (Dimodan
P); (b) from sunflower oil (Dimodan LS); adapted from Danisco Ingredients, technical paper TP3-1e.
furanone (abhexon), and 3-acetyl-2,5-dimethylthiophene were
obtained from Fluka (Buchs, Switzerland). Dimodan LS (sun-
flower oil monoglyceride, >98%) and Dimodan PV (saturated
vegetable oil monoglyceride, >98%) were kindly provided by
Danisco Ingredients (Braband, Denmark). Water used in these
experiments was twice distilled.
hydrophobic site (tail region) and a hydrophilic site
(headgroup region including the solubilized water re-
gion), they are capable of dissolving both water-soluble
and oil-soluble compounds (Engstrom, 1990; Ericsson
et al., 1991).
Of particular interest is the cubic phase, which has a
bicontinuous structure. It is generally described as two
interpenetrating networks of water channels separated
by a curved bilayer, extending in three dimensions
[forming the so-called infinite periodic minimal surface
(Larsson, 1989)]. The water pore diameter has been
shown to be ∼5 nm for the fully swelled phase. The
curved bilayer structure gives rise to a very large
interfacial area (∼400 m2/g), which is an excellent
entrapment medium (Ericsson et al., 1991). The solu-
bilization capacity for organic compounds was found to
be governed by various factors but mostly by the
curvature of the interface and the degree of water
swelling (Ericsson et al., 1983, 1991; Razumas et al.,
1996; Caboi et al., 1997).
Mod el Rea ction s. L-Cysteine (2 mmol) and furfural (2
mmol) were dissolved in a phosphate buffer (10 mL; 0.5 M;
pH 5.0). The cubic phase was prepared by introducing 1 g of
this solution and 4 g of melted (60 °C) Dimodan PV (saturated
monoglyceride) into a Pyrex tube (i.d. ) 1.5 cm). After vigorous
stirring (Vortex), the tube was placed in an oil bath at 140 °C
for 1 min and stirred again to form a homogeneous cubic phase.
Finally, this system was thermally treated for 4 h at 100 °C.
To obtain a microemulsion (isotropic fluid) at this temperature,
Dimodan PV was replaced by the unsaturated monoglyceride
product Dimodan LS (C18:2 ) 67% and C18:1 ) 21%; Krog, 1998).
As reference, an identical thermal treatment was applied to 1
g of the buffer solution containing the reactants. After heating,
the mixture was cooled and 4 g of melted Dimodan PV was
added in order to have similar workup conditions for the
isolation of volatiles.
Bicontinuous cubic phases are optically isotropic and
completely transparent and have a consistency like gels,
exhibiting the highest viscosity of all of the mesophases
formed (Larsson, 1989; Fontell, 1990). Cubic phases can
be also in equilibrium with an excess of water (they do
not transform into other phases upon contact with
water, see Figure 1), which is particularly interesting
as a matrix system for a controlled release of drugs and
eventually for flavor compounds. In certain cases, even
stereoselectivity of organic reactions was achieved at
such interfaces (Holmberg, 1994).
Much has been written on the solubilization capacity
of cubic phases and on their potential as delivery
systems (Burrows et al., 1994; Osborne and Ward, 1995;
Chang and Bodmeier, 1997). However, their potential
as microreactors for the thermal generation of aromas
has not yet been studied. The aim of this work was to
compare, on a quantitative basis, the reaction products
of cysteine/furfural and cysteine/ribose model systems
performed in two different structured fluids, that is, a
cubic phase and a microemulsion phase, and in an
aqueous medium under the same conditions. The influ-
ence of the structured phases on the thermal reactions
is discussed.
In a second series of experiments, L-cysteine (6.67 mmol)
and D-ribose (20 mmol) were dissolved in a phosphate buffer
(10 mL; 0.5 M; pH 5.0). Cubic phase samples were prepared
by introducing 2 g of this solution and 8 g of melted (60 °C)
Dimodan PV into a Pyrex tube (i.d. ) 2 cm). Finally, this
system or 2 g of buffer solution alone (reference) was thermally
treated following the same procedure as mentioned above.
Isola tion of Vola tiles fr om th e Rea ction Mixtu r es. The
aqueous reference samples were first transformed into a cubic
phase by adding melted Dimodan PV (4 g) to the reaction
system to ensure maximal similarity in the reaction mixtures
before sample cleanup. After cooling, the mix was extracted
with diethyl ether (three times, total of 50 mL). The slurries
were combined and submitted to the vacuum transfer system
described by Sen et al. (1991) using only two cold traps
immersed in liquid nitrogen. Once the sample was entirely
frozen in liquid nitrogen, the vacuum was opened, the sample
thawed, and the solvent together with other volatiles trans-
ferred into the cold traps at 100 mPa during 1.5 h, thus
separating them from the monoglycerides. A high vacuum was
avoided to minimize carry-over of monoglycerides into the cold
traps. The distillate containing the aroma compounds was then
dried over anhydrous Na2SO4 and concentrated to 2 mL on a
Vigreux column (50 cm × 1 cm) and finally to 0.5 mL under a
gentle stream of nitrogen.
Qu a n tifica tion of Vola tiles. (i) Furfural/ Cysteine Reac-
tion. The diethyl ether used for the isolation of volatiles was
spiked with benzylmercaptan as internal standard (5 or 10
µg).
EXPERIMENTAL PROCEDURES
Ch em ica ls. L-Cysteine, D-ribose, furan-2-aldehyde (fur-
fural), 5-methylfurfural, 5-ethyl-4-methyl-3-hydroxy-2(5H)-
(ii) Ribose/ Cysteine Reaction. The diethyl ether was spiked
with the following amounts of standard compounds: 350 µg

4810 J. Agric. Food Chem., Vol. 48, No. 10, 2000
Vauthey et al.
of 5-methylfurfural for quantitation of MFT, FFT, and furfural;
350 µg of 5-ethyl-4-methyl-3-hydroxy-2(5H)-furanone (ab-
hexon) for quantitation of 4-hydroxy-5-methyl-3(2H)-furanone
(norfuraneol); and 4.7 µg of 3-acetyl-2,5-dimethylthiophene for
quantitation of bis(2-methyl-3-furyl) disulfide (MFT-MFT).
Quantitation in both reactions was based on the ratio of
peak areas using the FID trace and was not corrected for
possible differences in FID responses and recovery yields of
internal standard and analyte.
High -Resolu tion Ga s Ch r om a togr a p h y (HRGC) a n d
HRGC Ma ss Sp ectr om etr y (HRGC-MS). An HP5890 GC
equipped with a DB-1701 fused silica capillary column (J &W,
Folsom, CA; 30 m × 0.32 mm × 0.25 µm) was used. By means
of a Valco valve, the effluent (He used as carrier gas; 80 kPa)
was split to an FID and an FPD detector, both kept at 250 °C.
An aliquot (1 µL) of the concentrated samples was injected in
the splitless mode using an HP-7673A autosampler. Identifica-
tion of compounds was performed using an HP-5890 series 2
GC with a DB-5 capillary column (J &W; 30 m × 0.32 mm ×
0.25 µm) connected to a Finnigan MAT SSQ 7000 mass
spectrometer. Mass spectra in the electron impact mode were
generated at 70 eV and in the chemical ionization mode at
150 eV with isobutane as reagent gas. The GC oven temper-
ature was programmed from 20 °C (2 min) at 70 °C/min to 50
°C (2 min), at 6 °C/min to 180 °C, and finally at 20 °C/min to
240 °C (10 min). Linear retention indices (RI) were calculated
according to the method of van den Dool and Kratz (1963).
(Figure 4). The mass spectrum of the synthesized
compound was in good agreement with the spectrum of
the unknown compound at RI ) 1521.
Both compounds, N-(2-mercaptovinyl)-2-(2-furyl)thi-
azolidine and 2-(2-furyl)thiazolidine, as well as FFT
were quantified in the three systems investigated (Table
1). On the basis of the relative peak areas to the internal
standard benzylmercaptane, ∼2 µg of FFT was gener-
ated in the aqueous reaction system. Under the same
conditions ∼9 µg of FFT was generated in a microemul-
sion and ∼12 µg in the cubic structure, resulting in a
6-fold increase compared to the aqueous system.
The same trend was observed for 2-(2-furyl)thiazoli-
dine and N-(2-mercaptovinyl)-2-(2-furyl)thiazolidine, for
which the cubic phase as matrix gave the highest
amounts of 240 and 123 µg, respectively. In the aqueous
solution, 2-(2-furyl)thiazolidine was detectable in only
one sample, whereas N-(2-mercaptovinyl)-2-(2-furyl)-
thiazolidine could not be detected under the experimen-
tal conditions. These results indicate that the cubic
structure is the most efficient matrix for the generation
of these three volatile compounds under the chosen
reaction conditions. Moreover, the ratio of the amount
for each compound formed in the cubic phase and in the
microemulsion differed significantly. Compared to the
microemulsion, the amount of N-(2-mercaptovinyl)-2-
(2-furyl)thiazolidine was 10 times higher in the cubic
phase, whereas ratios of about 1.3 and 6 were observed
for FFT and 2-(2-furyl)thiazolidine, respectively. Pos-
sible reasons for the different reaction yields will be
discussed later.
A hypothetical pathway for the generation of 2-(2-
furyl)thiazolidine from furfural (I) and cysteine (II) is
displayed in Figure 5. The first step in the reaction is
the addition of cysteine to the carbonyl group of furfural.
The subsequent elimination of a molecule water gives
rise to a Schiff base (III). This compound may undergo
decarboxylation, resulting in product (IV). This inter-
mediate can finally form a five-membered ring by
addition of the nucleophilic sulfur to the imine, which
leads to the formation of 2-(2-furyl)thiazolidine. Com-
pound III may also first cyclize to compound IIIb and
subsequently lose CO2. The 2-(2-furyl)thiazolidine may
then react with mercaptoacetaldehyde, released from
cysteine upon Strecker degradation, to finally give rise
to the tentatively identified N-(2-mercaptovinyl)-2-(2-
furyl)thiazolidine.
Ribose/Cystein e Rea ction . This system was chosen
because many parameters influencing flavor generation
such as pH, water activity, and temperature have
already been studied (Hofmann and Schieberle, 1995)
and because of the importance of C5 sugars in the
formation of meatlike and savory flavors. In the follow-
ing, we selectively determined potent odorants that
could be identified under the analytical conditions
rather than giving an entire view of volatiles generated.
RESULTS
F u r fu r a l a n d Cystein e a s P r ecu r sor s of F F T. In
the first series of experiments, cysteine was heated with
furfural. Maillard reactions are often studied at elevated
temperatures of 140-180 °C. In our study 100 °C was
chosen due to the necessity to stay in the cubic phase
region of the Dimodan PV (80-120 °C) and to avoid the
use of pressurizable reaction vessels.
The cubic phase of the Dimodan PV/water system (80:
20), used in this study, reverses to an L2 microemulsion
at temperatures >120 °C (Krog, 1998). It is also known
that cubic to hexagonal and hexagonal to L2 phase
transition temperatures decrease for unsaturated mono-
glycerides, such as Dimodan LS. Thus, by keeping the
monoglyceride/water at a constant ratio (80:20) and
maintaining the temperature at 100 °C, we could obtain
a reversed microemulsion phase by replacing Dimodan
PV with Dimodan LS (Figure 1).
Rea ction in Wa ter . Figure 2 shows GC/FPD chro-
matograms of volatile compounds obtained from cys-
teine/furfural mixtures reacted in different media (wa-
ter, microemulsion, and cubic phase).
In the aqueous phase (Figure 2A), besides the stan-
dard benzylmercaptane, only two sulfur compounds
were detected in trace amounts. One compound was
identified as FFT on the basis of its the retention index
(RI) and MS spectra. It is the expected product from
the reaction of furfural with cysteine.
Rea ction in Str u ctu r ed F lu id s. The yield of FFT
was significantly increased in both the microemulsion
and the cubic phase. In addition, two new compounds
were detected in relatively large amounts (Figure 2B,C)
with RI values of 1521 and 1997 on a DB-1701 gas
capillary column.
On the basis of MS/EI (Figure 3) and MS/CI, these
two molecules were identified as 2-(2-furyl)thiazolidine
(RI ) 1521) and, tentatively, N-(2-mercaptovinyl)-2-(2-
furyl)thiazolidine (RI ) 1997). To confirm these struc-
tures, 2-(2-furyl)thiazolidine was prepared from furfural
and cysteamine in aqueous solution by heating an
equimolar mixture (20 mmol each) at 100 °C for 1 h
The sensory properties of the reaction flavor obtained
in the aqueous system were compared with those of the
cubic phase sample by using a laboratory panel. The
reaction carried out in the cubic phase was found to
have a more intense overall flavor with a strong rubber,
eggy, roast chicken-like aroma. The aroma of the aque-
ous reaction was evaluated after transformation of the
solution into a cubic phase system in order to have
similar volatile-matrix interactions. This mixture was
weak in aroma, reminiscent of meat and lard with some
burnt character.

Structured Fluids as Microreactors for Flavor Formation
J. Agric. Food Chem., Vol. 48, No. 10, 2000 4811
F igu r e 2. GC/FPD chromatogram (DB-1701 capillary column) of volatile compounds generated from the reaction of furfural/
cysteine in (A) water, (B) microemulsion, and (C) cubic phase. The internal standard benzylmercaptane is indicated by f.
The volatiles formed in the cubic phase and in water
were monitored by GC/FID (Figure 6) and GC/FPD
(Figure 7). The FPD chromatograms evidenced the
differences between the two samples indicating genera-
tion of higher amounts of sulfur-containing compounds
in the cubic system.
A multiple standard addition method was used to
quantify six compounds in the reaction mixtures that
were identified by comparison with the corresponding
reference compounds based on RI on two capillary
columns of different polarities (DB-5 and DB-1701) and
their MS/EI spectra. Three compounds, covering the
range of chemical and physical properties of the volatiles
generated in the Maillard reaction, were chosen as
internal standards (see Experimental Procedures). Pos-
sible differences in recovery yield of the analyte and the
standard were not taken into account for the calculation
of concentrations.
Similarly to the reaction of cysteine/furfural, there
was a significant increase in the concentration of FFT

4812 J. Agric. Food Chem., Vol. 48, No. 10, 2000
Vauthey et al.
F igu r e 3. Mass spectra (MS/EI) of 2-(2-furyl)thiazolidine and
tentatively identified N-(2-mercaptovinyl)-2-(2-furyl)thiazoli-
dine.
F igu r e 5. Hypothetical reaction pathway leading from cys-
teine and furfural to 2-(2-furyl)thiazolidine and N-(2-mercap-
tovinyl)-2-(2-furyl)thiazolidine.
F igu r e 4. Synthesis of 2-(2-furyl)thiazolidine from cysteamine
and furfural.
(1995) could not be detected, although they may have
been formed.
Ta ble 1. Am ou n ts of Vola tile Rea ction P r od u cts
Gen er a ted fr om F u r fu r a l a n d Cystein e Solu bilized in
Differ en t Ma tr ices
Besides MFT and FFT, the formation of their precur-
sors norfuraneol and furfural was favored in the cubic
phase, being formed in ∼2.5-fold higher amounts. These
two compounds were recently found to be the key
intermediates in the formation of thiols from C5 sugars
(Hofmann and Schieberle, 1998).
The occurrence of 8.3 µg of MFT-MFT indicates that
∼50% of the MFT was dimerized. The dimer could have
been formed either during the Maillard reaction itself
or afterward during the isolation procedure. Because
MFT is very unstable in water and some organic
solvents (Hofmann et al., 1996), we assume that the
presence of the dimer is at least partly due to artifacts.
Partitioning of MFT into the hydrophobic chain region
of the cubic phase could be the reason for its enhanced
generation.
amounta (µg)
2-furfuryl- 2-(2-furyl)thi- N-(2-mercaptovinyl)-
matrix
water
microemulsion
cubic phase
thiol
azolidine
2-(2-furyl)thiazolidine
1.8
8.7
ndb
42
ndb
11.7
123
11.6
240
a
b
Mean values of duplicates (SD < 20%). Not detected.
in the cysteine/ribose reaction mixture carried out in
the cubic phase compared to the reaction in water, in
which only trace amounts were detected (see Table 2).
The detection limit under the given conditions was ∼0.5
µg/5 g of cubic phase sample. The potent odorant MFT
was detected only in the cubic phase system. The
absence of this compound in the reference is surprising,
but not contradictory to the findings in the literature
(Hofmann and Schieberle, 1995). Indeed, this compound
was detectable in the reference when the reaction
system was directly extracted with diethyl ether, that
is, no cubic phase was formed before extraction, and
when the isolation of volatiles was done by vacuum
distillation. The losses and poor recovery during isola-
tion from the cubic phase also explain that some of the
other odorants reported by Hofmann and Schieberle
DISCUSSION
This work clearly demonstrates that structured fluids,
such as association colloids, strongly influence the
thermal generation of volatiles from cysteine/furfural
and cysteine/ribose mixtures. Various hypotheses will
be discussed in an attempt to gain a first understanding
of these phenomena.
Wa ter Activity. The Maillard reaction is known to
be dependent on water activity (Bell and Labuza, 1994).

Structured Fluids as Microreactors for Flavor Formation
J. Agric. Food Chem., Vol. 48, No. 10, 2000 4813
F igu r e 6. GC/FID chromatogram (DB-1701 capillary column) of volatile compounds generated from the reaction of ribose/cysteine
in (A) water and (B) cubic phase.
Likewise, thermal flavor generation is influenced by
water activity, as shown in dry-heated model systems
of cysteine with different monosaccharides (Hofmann
and Schieberle, 1998). In our study, however, both
structured fluids (microemulsion and cubic phase) ex-
hibited a water activity of ∼1 and, consequently, the
differences in the flavor formation cannot be explained
by differences in macroscopic water activity.
Reaction pathways may, however, depend on water
and reactant mobility and their availability to each
other. In aqueous solutions, the reactants are totally free
to interact, whereas in the investigated structured fluids
the water is confined within cubic phase channels or
the internal core of the microemulsion, respectively.
Often, it has been shown that rate enhancement is
primarily due to an increase of reactant concentration
in the water droplet (Holmberg, 1994; Tascioglu, 1996).
In this study, furfural and cysteine were entirely
solubilized in water before the different systems (cubic
phase and microemulsion) were prepared, and their
concentrations were based on the water content being
identical in the systems. Furthermore, the partitioning
of the reactants between the lipophilic and hydrophilic
regions in the structured fluid phases lowers the actual
concentration of the reactants in the water region.
Nevertheless, we observed higher yields for FFT and
MFT.
Role of th e Su r fa cta n t)Wa ter In ter fa ce. One of
the possible mechanisms, which lead to the remarkable
rate enhancement and to different product distribution,
is the partitioning (compartmentalization) of the reac-
tants at the surfactant-water interface or close to it,
thus inducing a concentration gradient. Locally high
concentrations at the interface are thought to strongly
accelerate bimolecular reactions. Moreover, the close
proximity of the reactants as well as their restrained
movement within the interface will increase the fre-
quency of molecular collisions (cage effect).
This hypothesis is quite coherent with previous
results demonstrating that aspartame, a dimeric amino
acid, is clearly solubilized at the interface of microemul-
sions (Furedi-Milhofer et al., 1999). Moreover, it was

4814 J. Agric. Food Chem., Vol. 48, No. 10, 2000
Vauthey et al.
F igu r e 7. GC/FPD chromatogram (DB-1701 capillary column) of volatile compounds generated from the reaction of ribose/
cysteine in (A) water and (B) cubic phase.
Ta ble 2. Am ou n t of Vola tile Rea ction P r od u cts
Gen er a ted fr om Ribose a n d Cystein e Solu bilized in
Differ en t Ma tr ices
It has been shown in the literature that the release
of H2S from cysteine solution at pH 3-9 shows a first-
order reaction enhanced by alkaline pH (Zheng and Ho,
amounta (µg)
1994). The activation energies are 31.3 and 29.4 kcal/
mol for pH 3.0 and 9.0, respectively. On the other hand,
H2S release activation energies from glutathione were
18.8 and 19.9 kcal/mol at pH 3 and 9, respectively.
Glutathione evolves H2S more rapidly than cysteine. In
this case, the negative charge of the carboxyl group is
further away from the R-carbon than it is in cysteine
(Zheng and Ho, 1994). Therefore, the hydrogen atom of
the R-carbon is more readily abstracted. If cysteine is
attracted to the surface, its negative charge will be
embedded in the bilayer and will appear to be more
distant to the R-carbon than it is in free cysteine. As a
result, the pathway leading to the formation of FFT will
become more favorable, as observed in our experiments.
Therefore, besides a concentration effect, the specific
microenvironment at the surfactant-water interface
may influence the transition state by lowering the
activation energy (Tascioglu, 1996) and lead to a rate
enhancement in the generation of volatiles.
compoundb
cubic phase
water
2-methyl-3-furanthiol (MFT)
2-methyl-3(2H)-furanone
furfural
2-furfurylthiol (FFT)
3-mercapto-2-pentanone
norfuraneol
18.4
36.7
874
12.0
tr
698
8.3
tr
ndc
13.1
351
trd
ndc
291
tr
MFT-MFT
MFT-FFT
tr
a
b
Mean values of duplicates (SD < 20%). Compounds were
identified by comparison with the reference compound based on
retention indices on DB-5, DB-1701, and MS/EI spectra. c Not
detected.
shown that even water-soluble molecules such as bu-
tanedione can induce a phase transition of the cubic
phase into a lamellar phase due to interactions with the
monoglyceride surfactant headgroup (Vauthey, 1998).
Therfore, one may expect that cysteine, like aspartame
or butanedione, will equally penetrate into the interface
and will be localized or immobilized at the interface.
R ole of t h e In t er fa cia l Ar ea . Rate enhancement
occurring in microemulsions is generally referred to as
microemulsion catalysis (Rathman, 1996; Romsted et al.,

Structured Fluids as Microreactors for Flavor Formation
J. Agric. Food Chem., Vol. 48, No. 10, 2000 4815
interface and its area. The present work highlights the
potential of structured fluids in the thermal generation
of flavors and opens up new avenues for enhanced flavor
generation.
LITERATURE CITED
Bell, L. N.; Labuza, T. P. Influenec of the low-moisture state
on pH and its implication for reaction kinetics. J . Food Eng.
1994, 22, 291-312.
Boyle, E.; German, J . B. Monoglycerides in membrane systems.
Crit. Rev. Food Sci. Nutr. 1996, 36, 785-805.
Briggs, J .; Chung, H.; Caffrey, M. The temperature-composi-
tion phase diagram and mesophase structure characteriza-
tion of the monoolein/water system. J . Phys. II Fr. 1996, 6,
723-751.
F igu r e 8. Structure of the bicontinuous cubic phase [adapted
from Seddon (1990)].
Burrows, R.; Collet, J . H.; Attwood, D. The release of drugs
from monoglyceride-water liquid crystalline phases. Int. J .
Pharm. 1994, 111, 283-293.
1997). Tables 1 and 2 show the strong effect on the
generation of volatile products in Maillard model reac-
tions. Nevertheless, it has been shown in this work that
flavor formation in cubic phases is even more efficient
and yields higher amounts of key aroma compounds. We
denominate this effect cubic catalysis; it can mainly be
understood in terms of differences in the surfactant-
water interface properties, such as interface area or
curvature, assuming that the chemical reactions occur
preferably at the interfacial layer.
It is well established that the three-dimensional
structure of the cubic phase consists of water channels
separated by a bilayer arranged in such a way that it
forms a so-called infinite periodic minimal surface (see
Figure 8). These result in a large interfacial area of
∼400 m2/g of cubic phase (Ericsson et al., 1991). On the
other hand, assuming the formation of spherical drop-
lets in the reversed microemulsion (L2), a much smaller
interfacial area is formed, which could explain the
differences in flavor compounds generated between the
two structured fluids. It is, however, not possible at this
stage to predict how an increase in the interfacial area
will influence the final production of the different
volatile compounds. Moreover, the influence of the
interface curvature is difficult to estimate. Indeed, very
different yield ratios were obtained for FFT, 2-(2-furyl)-
thiazolidine, and N-(2-mercaptovinyl)-2-(2-furyl)thiazo-
lidine generated in the cubic phase compared to the
microemulsion (Table 1).
Caboi, F.; Nylander, T.; Razumas, V.; Talaikyte, Z.; Monduzzi,
M.; Larsson, K. Structural effects, mobility, and redox
behavior of vitamin K1 hosted in the monoolein/water liquid
crystalline phases. Langmuir 1997, 13, 5476-5483.
Chang, C.-M.; Bodmeier, R. Effect of dissolution media and
additives on the drug release from cubic phase delivery
systems. J . Controlled Release 1997, 46, 215-222.
Engstrom, S. Drug delivery from cubic and other lipid-water
phases. Lipid Technol. 1990, 2 (2), 42-45.
Ericsson, B.; Larsson, K.; Fontell, K. A cubic protein-mo-
noolein-water phase. Biochim. Biophys. Acta 1983, 729, 23-
27.
Ericsson, B.; Eriksson, P. O.; Lofroth, J . E.; Engstrom, S. Cubic
phases as delivery systems for peptide drugs. Polymeric
Drugs and Drug Delivery Systems; ACS Symposium Series
469; American Chemical Society: Washington, DC, 1991;
pp 251-265.
Farmer, L. J .; Mottram, D. S.; Whitfield, F. B. Volatile
compounds produced in Maillard reactions involving cys-
teine, ribose, and phospholipids. J . Sci. Food Agric. 1989,
49, 347-368.
Fontell, K. Cubic phases in surfactant and surfactant-like lipid
systems. Colloid Polym. Sci. 1990, 268, 264-285.
Furedi-Milhofer, H.; Garti, N.; Kamyshny, A. Crystallization
from microemulsionssa novel method for the preparation
of new crystal forms of aspartame. J . Cryst. Growth 1999,
198/ 199, 1365-1370.
Gasser, U.; Grosch, W. Identification of volatile flavour
compounds with high aroma values from cooked beef. Z.
Lebensm. Unters. Forsch. 1988, 186, 489-494.
Hiemenz, P. C. Principles of Colloid and Surface Chemistry;
Dekker: New York, 1986.
Hofmann, T.; Schieberle, P. Evaluation of the key odorants in
a thermally treated solution of ribose and cysteine by aroma
extract dilution techniques. J . Agric. Food Chem. 1995, 43,
2187-2194.
Hofmann, T.; Schieberle, P. Studies on intermediates generat-
ing the flavor compounds 2-methyl-3-furanthiol, 2-acety-2-
thiazoline and sotolone by Maillard-type reactions. In
Flavour SciencesRecent Developments; Taylor, A. J ., Mot-
tram, D. S., Eds.; The Royal Society of Chemistry: Cam-
bridge, U.K., 1996; pp 182-187.
Hofmann, T.; Schieberle, P. Quantitative model studies on the
effectiveness of different precursor systems in the formation
of the intense odorants 2-furfurylthiol and 2-methyl-3-
furanthiol. J . Agric. Food Chem. 1998, 46, 235-241.
Hofmann, T.; Schieberle, P.; Grosch, W. Model studies on the
oxidative stability of odour-active thiols occurring in food
flavors. J . Agric. Food Chem. 1996, 44, 251-255.
Holmberg, K. Organic and bioorganic reactions in microemul-
sions. Adv. Colloid Interface Sci. 1994, 51, 137-174.
Holscher, W.; Vitzthum, O. G.; Steinhart, H. Identification and
sensorial evaluation of aroma-impact-compounds in roasted
Colombian coffee. Cafe Cacao The 1990, 34, 205-212.
Additional information on the binding constants of the
reactants and the products is required to gain a more
precise insight into the distribution of compounds
between the aqueous phase and the interface.
CONCLUSIONS
Monoglycerides/water systems forming bicontinuous
cubic phases as well as microemulsions have been used
as “microreactors” for the generation of aroma com-
pounds by the Maillard reaction.
It was shown that structured fluids which are orga-
nized on a nanometer scale can lead to remarkably
higher yields of FFT and MFT as well as different aroma
profiles in the thermal generation of volatile compounds.
The flavor enhancement is partly due to an increase in
the reaction yield of odor-active compounds and partly
due to the wider variety of reaction products formed in
these structured media. Cubic phases showed a signifi-
cantly higher flavor enhancement than microemulsions.
However, both structured fluids were significantly more
efficient as a reaction medium than water.
It is postulated that the dominant factors affecting
the reaction yield are the nature of the water/surfactant

4816 J. Agric. Food Chem., Vol. 48, No. 10, 2000
Vauthey et al.
Krog, N. Food emulsifiers and their chemical and physical
properties. In Food Emulsions; Friberg, S., Larsson, K., Eds.;
Dekker: New York, 1997.
Larsson, K. Cubic lipid-water phases: Structures and biomem-
brane aspects. J . Phys. Chem. 1989, 93, 7304-7314.
Larsson, K. LipidssMolecular Organization, Physical Func-
tions and Technical Applications; The Oily Press: Dundee,
U.K., 1994.
Schieberle, P.; Hofmann, T. Identification of the key odorants
in processed ribose-cysteine Maillard mixtures by instru-
mental analysis and sensory studies. In Flavour Sciences
Recent Developments; Taylor, A. J ., Mottram, D. S., Eds.;
The Royal Society of Chemistry: Cambridge, U.K., 1996;
pp 175-181.
Schieberle, P.; Hofmann, T. Characterization of key odorants
in dry-heated cysteine-carbohydrate mixtures: Comparison
with aqueous reaction systems. In Flavor Analysiss
Developments in Isolation and Characterization; Mussinan,
C. J ., Morello, M. J ., Eds.; ACS Symposium Series 705;
American Chemical Society: Washington, DC, 1998; pp 320-
330.
Seddon, J . M. Structure of the inverted hexagonal phase, and
non-lamellar phase transitions of lipids. Biochim. Biophys.
Acta 1990, 1031, 1-69.
Sen, A.; Laskawy, G.; Schieberle, P.; Grosch, W. Quantitative
determination of â-damascenone in foods using a stable
isotope dilution assay. J . Agric. Food Chem. 1991, 39, 757-
759.
Tascioglu, S. Micellar solutions as reaction media. Tetrahedron
1996, 52, 11113-11152.
Van den Dool, H.; Kratz, P. A generalization of the retention
index system including linear temperature programmed
gas-liquid partition chromatography. J . Chromatogr. 1963,
24, 463-471.
Vauthey, S. Solubilization of flavour compounds in structured
fluids. Ph.D. Thesis, University of Lausanne, 1998.
Zheng, Y.; Ho, C.-T. Kinetics of the release of hydrogen sulfide
from cysteine and glutathione during thermal treatement.
In Sulfur Compounds in Foods; Mussinan, C. J ., Keelan,
M. E., Eds.; ACS Symposium Series 564; American Chemi-
cal Society: Washington, DC, 1994.
Laughlin, R. G. The Aqueous Phase Behavior of Surfactants;
Academic Press: San Diego, CA, 1994.
Mottram, D. S.; Whitfield, F. B. Volatile compounds from the
reaction of cysteine, ribose, and phospholipids in low-
moisture systems. J . Agric. Food Chem. 1995, 43, 984-988.
Mulders, E. J . Volatile components from the non-enzymatic
browning reaction of the cysteine/cystine-ribose system. Z.
Lebensm. Unters. Forsch. 1973, 152, 193-201.
Mu¨nch, P.; Hofmann, T.; Schieberle, P. Comparison of key
odorants generated by thermal treatment of commercial and
self-prepared yeast extracts: influence of the amino acid
composition on odorant formation. J . Agric. Food Chem.
1997, 45, 1338-1344.
Osborne, D. W.; Ward J . I. Lyotropic liquid crystals as topical
drug delivery vehicles. Int. J . Pharm. Adv. 1995, 1 (1), 38-
45.
Rathman, J . F. Micellar catalysis. Curr. Opin. Colloid Interface
Sci. 1996, 1, 514-518.
Razumas, V.; Larsson, K.; Miezis, Y.; Nylander T. A cubic
monoolein-cytochrome c-water phase: X-ray diffraction, FT-
IR, differential scanning calorimetric, and electrochemical
studies. J . Phys. Chem. 1996, 100, 11766-11774.
Romsted, L. S.; Bunton, C. A.; Yao, J . Micellar catalysis, a
useful misnomer. Curr. Opin. Colloid Interface Sci. 1997,
2, 622-628.
Schieberle, P. Studies on the flavour formation of roasted white
sesame seeds. In Progress in Flavour and Precursor Studies;
Schreier, P., Winterhalter, P., Eds.; Allured Publishers:
Wheaton, IL, 1993; pp 343-360.
Received for review November 19, 1999. Revised manuscript
received J une 5, 2000. Accepted J une 5, 2000.
J F991254A