Full text of DOI:10.1111/j.1365-2621.2003.tb05740.x

J FS: Hypothesis Pa per
Rapid Analysis of Volatile Release
from Powders using Dynamic Vapor
Sorption Atmospheric Pressure
Chemical Ionization Mass Spectrometry
D.M. DRONEN AND G.A. REINECCIUS
ABSTRACT: It is hypothesized that a dynam ic vapor sorption instrum ent coupled with an atm ospheric pressure
chem ical ionization m ass spectrom eter (APCI-MS) m ay be useful for rapid analysis of volatile release from dry
food m aterials. Prelim in ary data from a related system (Vapor Gen erator In strum en t-APCI-MS) was used to
m on itor release of volatiles from spray-dried food polym ers as a fun ction of relative hum idity at 40 °C. The
system dem onstrated differences in volatile release as a function of volatile com pound, relative hum idity, and
food polym er. Com parison of these data with data collected using traditional shelf life study m ethodologies, and
concerns for the ability of the m ethod to accurately reflect tim e-dependent changes are discussed.
Keywords: DVS, VGI, APCI, en capsulation , volatile release
Dynam ic vapor sorption
Introduction
For applications pertaining to food m aterials, dynam ic vapor
sorption (DVS) has primarily been used to determine moisture sorp-
tion isotherms and related phenomena in dry food systems (Teoh
and others 2001; Levoguer and William s 2002). DVS instrum ents
allow for rapid equilibration of samples in a controlled RH and tem-
perature environm ent and have gravim etric capabilities to m oni-
tor sm all changes in sam ple weight (m oisture uptake). The short
time scale for moisture equilibration in DVS instruments presents
opportunities for rapid analysis of sam ples.
raditional m ethods for the study of volatile release from dry
T
food systems are extremely time- and labor-intensive, requir-
ing 6 to 12 mo of storage to obtain a full release profile for a system.
Additionally, it is well docum ented that volatile release from dry
m aterials is a function of both the storage relative hum idity (RH)
and temperature, thus making multiple storage conditions critical
to capturing a full picture of volatile release from a system . There
is, therefore, a need for rapid m ethods to characterize volatile re-
lease from food system s. Dynam ic vapor sorption m ethodologies
coupled with atmospheric pressure chemical ionization mass spec-
trometry (APCI-MS) has the potential to become a useful tool for the
rapid analysis of volatile release from dry food products and ingre-
dients.
Atm ospheric pressure chem ical ion ization m ass
spectrom etry (APCI-MS)
APCI-MS is com m only used for analysis of gas-phase sam ples
and is considered one of the best techniques available for real-time
analysis of volatiles (Linforth and others 1999). APCI-MS has been
used to directly m easure m outh- and nose-space volatiles during
eating, and it is capable of simultaneously monitoring 10 to 30 vol-
atiles with sensitivity on the order of ppb (Turner and others 2002).
APCI-MS can also be used in the presence of m oisture and air; it
produces single ions for m ost com pounds and operates at atm o-
spheric pressures to facilitate sam pling (Taylor and others 2000).
Using APCI-MS to monitor volatile release from dry systems would
provide unique, time-release profile information on the release of
individual com pounds.
Use of a vapor generator instrument (VGI; Surface Measurement
System s, London, U.K.), an instrum ent that provides a controlled
tem perature and hum idity environm ent without the gravim etric
capabilities of a DVS, in conjunction with APCI-MS to study real-
tim e volatile release from dry food polym ers, was recently devel-
oped by Firmenich SA (Geneva, Switzerland) and surface measure-
m ent system s (London, U.K.) (Blake and others 2003).
Moisture sorption and volatile release in dry food
system s
Moisture sorption properties of dry food products and ingredi-
ents are critical in their overall processing, storage stability, and
application perform ance (Levoguer and William s 2002). Early re-
search has shown that volatile diffusion in am orphous carbohy-
drates is strongly dependent upon the moisture content of the sys-
tem (Menting and others 1970a, 1970b; Chandrasekaran and King
1971). More recent research has dem onstrated that the com bina-
tion of moisture and temperature strongly influence the degree of
plasticization of the m aterial during storage, thereby governing
rates of volatile loss during storage (Reineccius 1988; Levi and Karel
1995; Goubet and others 1998; Gunning and others 1999; Whorton
2000).
MS 20030311 Submitted 6/3/03, Revised 7/3/03, Accepted 7/23/03. Authors
are with the Univ. of Minnesota, Dept. of Food Science and Nutrition, 1334
Eckles Ave., St. Paul, Minn. 55108 U.S.A. Direct inquiries to author Reineccius
(E-mail: greinecc@umn.edu).
Objectives an d hypothesis
The objectives of this paper are to report and comment on pre-
2158 J OURNAL OF FOOD SCIENCE—Vol. 68, Nr. 7, 2003
© 2003 Institute of Food Technologists
Further reproduction prohibited without permission

Vola tile relea se using DVS-APCI-MS . . .
Table 1—VGI 2000M humidity ramp program
liminary results obtained using VGI-APCI-MS for analysis of vola-
tile release from food polymers, and to hypothesize on the poten-
tial for further use of VGI-APCI-MS or DVS-APCI-MS for rapid anal- Segment Segment
Ramp
rate
(%RH/min) (%RH)
End of
ramp
Dwell
time
(min)
Dwell
(%RH)
nr
type
ysis of volatile release from dry foods, fragrances, or flavoring
m aterials. Prelim inary experim ents at constant tem perature with
sam ples exposed to a constant high-RH condition or an RH ram p
program show the potential of this technique. Based on the prelim-
inary work presented and a review of the literature, we hypothesize
that DVS or VGI coupled to APCI-MS m ay be a useful tool for the
rapid analysis and comparison of the effects of moisture and tem-
perature on volatile release from dry food ingredients and products.
1
2
3
4
5
6
7
8
Ramp
Dwell
Ramp
Dwell
Ramp
Dwell
Ramp
Dwell
Ramp
Dwell
Ramp
Dwell
20
—
20
—
20
—
20
—
20
—
20
—
30
—
40
—
50
—
60
—
70
—
80
—
—
30
—
40
—
50
—
60
—
70
—
80
—
6
—
6
—
6
—
6
—
6
—
6
9
10
11
12
13
Materials and Methods
Materials
End seg.
Return to 0% RH set point
Maltodextrin [10 dextrose equivalent (DE), Maltrin M100®;
Grain Processing Corp., Muscatine, Iowa, U.S.A.], corn syrup solids
(25 DE, Maltrin M250; Grain Processing Corp.), soy protein concen-
trate (Prom ine® HV; provided by Central Soya, Fort Wayne, Ind.,
U.S.A.), whey protein isolate (BiPRO®; provided by Davisco Foods
Intl., Inc., LeSueur, Minn., U.S.A.), Acacia gum (Spraygum TM IRX
61232; Colloides Naturales Intl., Rouen, France), and an n-octenyl
succinic anhydride-m odified starch blended with high-DE corn
syrup solids (HiCapTM 100; Natl. Starch and Chemical Co., Bridge-
water, N.J., U.S.A.) were used as food polym ers for this study.
ion with a dwell tim e of 0.02 s and inter-channel delay of 0.02 s.
Volatiles were ionized by a 3.78 kV corona discharge (cone voltage
20 V) in the source (90 °C) (probe tem perature 20 °C) and then
passed into the high-vacuum region of the m ass spectrom eter.
MassLynx software (version 3.4) was used to collect and process the
data.
Ch em icals
Results and Discussion
Acetaldehyde, butyraldehyde, ethyl alcohol, ethyl form ate,
ethanethiol, 2-furaldehyde, lim onene, m ethyl alcohol, 2,3-pen- Ability to m on itor volatile release usin g VGI-APCI-
tanedione, and pyrrole were purchased from Sigm a-Aldrich (Mil- MS
waukee, Wis., U.S.A.) and were used as the model volatiles at 1000
ppm on a solids basis.
With the VGI system , there are 2 experim ental param eters that
can be controlled: temperature and RH. Within an experiment, tem-
perature remains constant, whereas RH can either be programmed
to maintain a constant RH, or RH can be modulated using a ramp
program (Table 1).
Powder preparation
Dry food polymers were mixed with water and allowed to rehy-
drate overnight. Solids levels were determined by polymer viscos-
ities (10 DE maltodextrin 40%; 25 DE corn syrup solids 40%; Acacia
gum 35%; modified starch 45%; soy protein concentrate 15%; whey
protein isolate 30%). The viscosity had to be low enough to permit
atom ization during spray-drying (about 250 cp). Volatile com -
pounds were added to the aqueous food polymer slurries immedi-
ately prior to drying and emulsified using a bench-top high-sheer
m ixer (Greerco Corp., Hudson, N.H., U.S.A.). Powders were ob-
tained by drying the polym er-water-volatile system s using a Niro
Utility spray-dryer with a radial vane wheel atomizer. Drying con-
ditions were standardized at inlet and exit air tem peratures of
200 °C and 90 to 100 °C, respectively. After drying, powders were
cooled to ambient temperature in plastic bags and stored in sealed
glass containers at –29 °C until used.
Using the VGI-APCI-MS in the 2 operating m odes noted above,
it was possible to compare differences in volatile release profiles as
a function of volatiles, food matrix, RH, and temperature (no tem-
perature studies were performed for this study). It must be noted,
however, that as the instrument was used and the experiments were
designed, it was possible to compare only the relative release pro-
files of volatiles. We could not compare volatile release across sys-
tem s in an absolute sense. The VGI apparatus uses such a sm all
sam ple that it was not possible to accurately control sam ple size,
thus we could not separate m atrix or volatile com pound effects
from differences in sample size. This limitation is unique to the use
of a VGI instrument because a DVS instrument has a microanalyt-
ical balance that yields sample size and moisture sorption informa-
tion. Thus, the data presented in this paper were scaled (norm al-
ized) on the highest intensity of release of each individual volatile
for each system and were com pared on this basis.
VGI-APCI-MS an alysis
A VGI 2000M unit (Surface Measurem ent System s, London,
U.K.) was used to expose powder sam ples to different RH condi- Com parison of volatile release via VGI-APCI-MS and
tions (a constant RH of 80% or a prescribed ramp) (Table 1) at 40 °C. static equilibrium system s
To analyze a sample, about 5 to 10 mg of powder were placed on a
glass microscope slide in the VGI. The chamber was rapidly sealed
and the RH program was im m ediately started. An N2 stream with
known RH flowed over the sam ple and exited the VGI cham ber at
the opposite end. A sample of the headspace purged from the VGI
chamber was fed into a Micromass ZMD (Micromass UK Limited,
Manchester, U.K.) APCI-MS at flow rates of 3.8 L/ h via a heated,
insulated capillary transfer line (tem perature 85 °C). The m odel
volatiles were m onitored in the selected ion m ode using the MH+
In a traditional volatile release storage study, data are collected
at points in time from an equilibrium system over a period of up to
12 mo. Analysis of samples using VGI-APCI-MS or DVS-APCI-MS at
the same temperature and RH, as in a typical shelf life study, should
provide release data that theoretically are equivalent; however, the
DVS- or VGI-APCI-MS method has the advantage of shorter analy-
sis tim e. Additionally, data produced via DVS- or VGI-APCI-MS
analysis would provide a m ore com plete picture, nam ely, nearly
continuous data collection for each volatile as it is released over
URLs and E-mail addresses are active links at www.ift.org
Vol. 68, Nr. 7, 2003—J OURNAL OF FOOD SCIENCE 2159

Vola tile relea se using DVS-APCI-MS . . .
time; as opposed to having only 1 time point every several days (as instrument more quickly and be detected longer. Thus, our data on
in a traditional storage study). Another m ethod for the analysis of time to detection and duration of release must be considered with
volatile release from microcapsules was proposed by Cadwallader caution.
and others (2002). This m ethod couples DVS with online purge-
To further the data com parison, Dronen and Reineccius (2003)
and-trap gas chrom atography. APCI-MS provides faster analysis found that significant losses of certain volatiles occurred at lower
and perhaps a m ore com plete picture of volatile release as com - RH levels and temperatures (11% to 33% RH, 25 °C) than we found
pared with purge-and-trap GC, which provides only a periodic sam- using the VGI-APCI-MS methods. The VGI-APCI-MS data present-
pling of volatiles.
ed in Figure 2 indicate that m easurable m ethanol release did not
An advantage of equilibrium system s, as described by Dronen occur until approximately 50% RH for the 25 DE CSS (except for li-
and Reineccius (2003) is that time-dependent phenomena such as m onene which m ay have been a surface residue). This is a m uch
matrix collapse, glass transition, or crystallization are permitted to higher RH than was observed in the referenced traditional study.
occur. These tim e-dependent phenom ena have a significant ef- This lag in detection of losses may be the result of the VGI-APCI-MS
fect on the release of volatiles (Whorton 2000) and may not be ob- method lacking the sensitivity to measure very low levels of volatile
served or properly accounted for when using a rapid, dynam ic release. These slow rates of release may be very significant over long
methodology of analysis such as DVS- or VGI-APCI-MS. Thus, it was time periods, such as typical storage periods for shelf-stable foods,
considered important to compare some of the VGI-APCI-MS data, but may not be measured by the more rapid VGI-APCI-MS method-
albeit preliminary and incomplete, to more traditional storage data ology. We must point out that the sampling system and MS were not
for validation of the more rapid methodology.
optimized for this analysis. With optimization of parameters, we may
In a traditional study of the release of 18 volatile com pounds have been able to detect lower levels of volatiles and, thereby, min-
from the food polymers 10 DE maltodextrin (MD) and 25 DE corn imized this problem; the issue of instrument detection limits for real
syrup solids (CSS) over tim e as a function of storage tem perature time measurement of released volatiles can be a significant barrier
and RH, it was found that volatiles were released more rapidly from to its application.
25 DE CSS than from 10 DE MD (Dronen and Reineccius 2003).
Com parison of volatile release from Acacia gum , m odified
Continuation of the referenced work has provided data on the re- starch, 10 DE MD, and 25 DE CSS using the VGI-APCI-MS with a
lease of volatiles from Acacia gum and m odified starch using the ram ped RH program (Figure 2) shows that the 10 DE MD system
sam e m ethods as described by Dronen and Reineccius (2003). released volatiles at a lower RH than the 25 DE CSS (Figure 2). This
Their work found that m odified starch had the greatest overall appears to be in conflict with the earlier work of Dronen and Rei-
rates of volatile release, and the 10 DE MD and 25 DE CSS had the neccius (2003) using traditional methodology. As explained earlier,
slowest rates of volatile release of the 4 materials studied (unpub- this apparent conflict may be due to the larger sample size or higher
lished data). These data are not consistent with our VGI-APCI-MS level of volatiles in the 10 DE MD as compared with the 25 DE CSS.
data in that the 25 DE CSS had the m ost rapid tim es to m axim um Thus, this conflict between methods may be explainable and cor-
volatile release and the shortest times to complete volatile release rectable by using a DVS-APCI-MS system .
using the constant 80% RH method (Figure 1). Additionally, on av-
Effects of tim e-dependent phenom ena on volatile
release profile
The effects of Tg and matrix collapse on volatile release are time-
erage, the Acacia gum had the longest tim es to m axim um release
and com plete volatile release. This conflict in data m ay be the re-
sult of differences in sam ple size or volatile concentration in the
powders because both factors would influence detection levels of dependent phenom ena. It is im portant to consider that high RH
volatiles in the VGI-APCI-MS m ethod. A larger am ount of sam ple conditions result in rapid structural changes in particle morpholo-
(or an equal am ount of sam ple that had higher levels of volatiles) gy, and high levels of volatile release are expected up to the point
would result in the volatiles reaching the detection lim its of the
Figure 2—Methanol release from 10 DE maltodextrin (᭜),
Figure 1—Release of butanal from 10 DE maltodextrin (᭜),
25 DE corn syrup solids (᭹), Acacia gum (᭝), m odified
starch (–), w hey protein isolate (×), and soy protein con-
centrate (᭛) using constant 80% relative humidity.
25 DE corn syrup solids (᭹), Acacia gum (᭝), and modified
starch (–) using the humidity ramp program, with points of
glass transition noted for each material (determined by use
of DSC, unpublished data).
2160 J OURNAL OF FOOD SCIENCE—Vol. 68, Nr. 7, 2003
URLs and E-mail addresses are active links at www.ift.org

Vola tile relea se using DVS-APCI-MS . . .
of structural collapse or dissolution. At the point of structural col-
Ultimately it would be important to compare and correlate vol-
lapse, volatile release declines and any volatiles not released upon atile release data obtained from DVS- or VGI-APCI-MS experiments
collapse are re-entrapped in the matrix (Whorton 2000). For exam- with release data obtained from traditional equilibrium experi-
ple, in traditional systems, 25 DE CSS and modified starch collapse ments on volatile loss and moisture sorption. Only when these data
at a relatively low RH, and all (or most) of the entrapped volatiles are can be correlated will DVS- or VGI-APCI-MS instruments have the
released during collapse.
potential to rapidly predict release of volatiles from dry foods, fla-
To understand the effects of tim e-dependent phenom ena on vors, and fragrance m aterials, and to predict shelf life stability as
volatile release measurement using a VGI-APCI-MS, we will consid- affected by volatile losses.
er the release profiles obtained for m ethanol, a very sm all polar
In sum m ary, VGI-APCI-MS was used to obtain release profiles
molecule that has been found to diffuse readily through food poly- for 10 volatile com pounds from 6 spray-dried food polym ers. We
m ers (Dronen and Reineccius 2003). Food polym er effects on the found differences in volatile release profiles across volatiles and
release profile of m ethanol are shown in Figure 2. The m ethanol food polymers as a function of time and RH. Whereas we have found
release profiles were generated using a RH ram p program (Table discrepancies between our preliminary data and more traditional
1). The approximate times when the RH in the VGI sample chamber m ethodologies to study volatile release, these discrepancies are
would have reached the glass transition (Tg) conditions for each often explainable. In several cases, they are linked to the use of a
material are noted on Figure 2 (unpublished DSC data). It has been VGI instrument as opposed to a DVS. The VGI system was originally
demonstrated that the glass transition of materials plays some role designed for observing changes in particle m orphology as a func-
in volatile release (Whorton 2000); however, our VGI-APCI-MS data tion of time and RH, not for monitoring volatile release. The use of
did not show the anticipated effects of passing through Tg or sub- a DVS in place of a VGI would perm it control of sam ple size and
sequent collapse. Whereas our inability to accurately control sam- would likely thereby perm it quantitative com parisons across sys-
ple size lim its the com parative value of our data, it is noteworthy tems. Additionally, the use of a DVS in place of the VGI would also
that Tg appeared to be poorly related to the onset of volatile release ensure that our samples reached moisture equilibrium within the
for the modified starch and 25 DE CSS, but perhaps related for the tim e fram e of our experim ents. This would ensure that we were
Acacia gum and 10 DE MD. In essence, m ethanol release was not m easuring volatile release at a given m oisture content.
detected from the m odified starch and CSS until the RH had sub-
Conclusions
stantially exceeded that of the Tg (at constant temperature, Tg can
be crossed by changing environmental RH). Methanol release was
detected close to the anticipated RH (Tg) for the Acacia gum and
MD. In general, it appears that there is a questionable relationship
between the Tg of a material and the release of volatiles. The lack of
correlation of volatile release onset with Tg was also found by Blake
and others (2003).
n conclusion, there are still m any questions about the ability of
DVS- or VGI-APCI-MS to predict shelf life stability of dry foods,
I
flavors, and fragrance m aterials as affected by volatile losses. We
question the ability of DVS- and VGI-APCI-MS to accurately reflect
time-dependent processes such as matrix collapse, glass transition,
or dissolution. Additionally, the detection thresholds of the system
m ay lim it the ability to predict volatile release under conditions
where loss is slow but significant over long storage periods. Corre-
lation of DVS- or VGI-APCI-MS data with data from traditional mois-
ture sorption and volatile loss experiments is imperative to estab-
lish relationships between DVS- or VGI-APCI-MS data and shelf life
stability. Despite these potential limitations and the questions that
rem ain, we feel that this m ethod is worthy of additional research
because of its potential to evaluate and compare a greater number
of encapsulated system s or food system s for their ability to retain
(or release) volatiles during storage than is possible using tradi-
tional m ethodology.
Tim e scale of experim entation
The tim e needed for sam ple m oisture equilibration in the VGI
m ay have been longer than the tim e allowed in our experim ents.
Thus, the sam ples m ay not have reached m oisture equilibrium at
each RH ramp. While the small sample size used minimizes the time
required for sample equilibration, we cannot ensure that the sam-
ples were at the RH we expected. Because the VGI can change RH
rapidly, the relationship we observed between RH and volatile re-
lease may have been shifted in time relative to a static storage sys-
tem. In a static volatile loss method, samples are allowed to come to
the desired test aw prior to starting the volatile loss study. In the VGI-
APCI-MS method, differences in the rates of moisture uptake and/
or amount of moisture needed to change aw between food polymers
would influence the aw of a test material in the system at any given
tim e. Using a DVS (with a gravim etric balance) instead of a VGI
would allow one to determine when the system has reached mois-
ture sorption equilibrium and thereby m inim ize this issue.
Acknowledgments
The authors thank Dr. Anthony Blake and Firmenich SA for mak-
ing their VGI-APCI-MS available to us for this feasibility study.
References
Blake A, Can tergian i E, Harvey B. 2003. Direct m on itorin g of flavor release
from encapsulated flavours. In: LeQuere JL, Etievant PX, editors. Flavour re-
search at the dawn of the twenty first century. Cachan, France: Lavoisier (Forth-
com in g).
Mechan ism of volatile loss
If a volatile compound is no longer detectable in an encapsulat-
ed m aterial, one m ay not necessarily know if the volatile was lost
because of evaporation (diffusion) from the particle or if it has re-
acted with the encapsulation matrix, oxygen, or other volatiles. As
a rapid methodology, DVS- or VGI-APCI-MS may provide informa-
tion on the proportion of volatile losses that occur due to diffusive
losses as com pared with losses due to chem ical reactions. This is
proposed because, in theory, the shorter analysis tim e in the VGI-
APCI-MS m ay m inim ize the tim e fram e for these chem ical reac-
tions. Unfortunately, we have no data to validate or repudiate this
hypothesis at this point.
Cadwallader KR, Zhou Q, Huda MS. 2002. Measurin g flavor release from m i-
croen capsulan ts un der con trolled en viron m en tal con dition s. In : ACS 223rd
Natl. Meeting Book of Abstracts. AGFD-0024.
Chandrasekaran SK, Kin g CJ. 1971. Reten tion of volatile flavors during dryin g
of fruit juices. AIChE Sym posium Series. 67:122–30.
Dron en DM, Rein eccius GA. 2003. Mass tran sfer param eters of volatile com -
pounds in dry carbohydrate m atrices. In: LeQuere JL, Etievant PX, editors. Fla-
vour research at the dawn of the twenty-first century. Cachan, France: Lavoisi-
er (Forthcom ing).
Goubet I, Le Quere JL, Voilley AJ. 1998. Retention of arom a com pounds by carbo-
hydrates: influence of their physicochem ical characteristics and of their phys-
ical state: a review. J Agric Food Chem 46:1981–90.
Gunning YM, Gunning PA, Kem sley EK, Parker R, Ring SG, Wilson RH, Blake A.
1999. Factors affectin g th e release of flavor en cap su lated in carboh ydrate
m atrices. J Agric Food Chem 47:5198–205.
URLs and E-mail addresses are active links at www.ift.org
Vol. 68, Nr. 7, 2003—J OURNAL OF FOOD SCIENCE 2161

Vola tile relea se using DVS-APCI-MS . . .
Levi G, Karel M. 1995. The effect of phase transitions on release of n-propanol Taylor AJ, Linforth RST, Harvey BA, Blake A. 2000. Atm ospheric pressure chem -
entrapped in carbohydrate glasses. J Food Eng 24:1–13.
Levoguer CL, William s DR. 2002. Application Note 102: Moisture sorption prop-
ical ion ization m ass spectrom etry for in vivo an alysis of volatile flavor re-
lease. Food Chem 71:327–38.
erties of food products and packaging m aterials studied by DVS. Available at: Teoh HM, Schm idt SJ, Day GA, Faller JF. 2001. Investigation of cornm eal com po-
<http:/ / www.sm suk.co.uk>. Accessed June 2002.
nents using dynam ic vapor sorption and differential scanning calorim etry. J
Food Sci 66:434–40.
Turner JA, Linforth RST, Taylor AJ. 2002. Real-tim e m onitoring of therm al flavor
generation in skim m ilk powder using atm ospheric pressure chem ical ioniza-
tion m ass spectrom etry. J Agric Food Chem 50:5400–5.
Whorton BC. 2000. Effect of polym eric phase transitions on the controlled re-
lease an d oxidative stability of flavor m odel system s en capsulated in tradi-
tional carbohydrate carriers. [PhD dissertation]. St. Paul, Minn.: Univ. of Min-
nesota. 237 p. Available from : Dissertation Express <http:/ / www.lib.um i.com /
dxweb>. Accessed 6 June 2003.
Linforth RST, Baek I, Taylor AJT. 1999. Sim ultaneous instrum ental and sensory
analysis of volatile release from gelatin and pectin/ gelatin gels. Food Chem
65:77–83.
Menting LC, Hoogstad B, Thijssen HAC. 1970a. Diffusion coefficients of water
and organic volatiles in carbohydrate-water system s. J Food Tech 5:111–26.
Menting LC, Hoogstad B, Thijssen HAC. 1970b. Arom a retention during the dry-
ing of liquid foods. J Food Tech 5:127–39.
Rein eccius GA. 1988. Spray dryin g of food flavors. In : Risch S, Rein eccius G,
editors. Flavor encapsulation. ACS Sym posium Series 370. Washington, D.C.:
Am erican Chem ical Society. p 55–66.
2162 J OURNAL OF FOOD SCIENCE—Vol. 68, Nr. 7, 2003
URLs and E-mail addresses are active links at www.ift.org