Coffee flavour: An overview

Abstract

This paper is a concise review of the research on coffee flavour to serve as a rapid reference on the subject. It covers the process of roasting coffee beans, the volatile and non-volatile components generated by the process and the chemical reactions responsible for their formation. Volatile compounds significant on the determin-ation of coffee aroma are given according to the most recent research. Finally, the paper discusses the chemical indexes used over the years to characterize coffee flavour deterioration and estimation of shelf-life.

Full text

COFFEE FLAVOUR: AN OVERVIEW 99
Copyright © 2004 John Wiley & Sons, Ltd. Flavour Fragr. J. 2004; 19: 99–104
FLAVOUR AND FRAGRANCE JOURNAL
Flavour Fragr. J. 2004; 19: 99–104
Published online 4 February 2004 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/ffj.1325
* Correspondence to: R. A. Buffo, Department of Food Science, Faculty of
Agricultural and Food Sciences, University of Manitoba, 250 Ellis Building,
Winnipeg, MB, Canada R3T 2N2.
E-mail: rabuffo@hotmail.com
Coffee flavour: an overview
Roberto A. Buffo1,* and Claudio Cardelli-Freire2
1Department of Food Science and Nutrition, University of Minnesota, St. Paul, MN, USA
2Meals and Sides Division, ConAgra, Los Angeles, CA, USA
Received 21 June 2001
Accepted 28 October 2003
ABSTRACT: This paper is a concise review of the research on coffee flavour to serve as a rapid reference on the
subject. It covers the process of roasting coffee beans, the volatile and non-volatile components generated by the
process and the chemical reactions responsible for their formation. Volatile compounds significant on the determin-
ation of coffee aroma are given according to the most recent research. Finally, the paper discusses the chemical
indexes used over the years to characterize coffee flavour deterioration and estimation of shelf-life. Copyright © 2004
John Wiley & Sons, Ltd.
KEY WORDS: coffee flavour; volatiles; non-volatiles; chemical indexes
Introduction
Coffee is a tropical plant that grows at 600–1800 m
above sea level.1 It is native of Ethiopia, from where
it spread first to India and then to Indonesia, Brazil,
Colombia and Central America.2 What we commonly
know as ‘coffee’ is a beverage prepared by brewing
roasted, ground beans. The plant produces red cherry-like
fruits containing two seeds, which, after being separated
from the fruit pulp, are known as ‘green coffee’. They
are packed in sacks and transported to consuming
countries. When received, they are blended with green
beans from other origins and roasted to produce the
characteristic flavour and colour associated with coffee
beverage.3
Roasting
Green coffee beans lack the colour and characteristic
aroma of roasted coffee; both are formed during the
roasting process. Coffee oil, which comprises about 10%
of the roasted beans, carries most of the coffee aroma.
The aroma is made up of a complex mixture of volatile
compounds, whereas non-volatiles determine sourness,
bitterness and astringency.4
The roasting process can be roughly divided into three
phases:2
1. An initial drying phase, evidently endothermic, during
which moisture is eliminated. The smell of the beans
changes from green to peasy to bread-like, and the
colour turns yellowish.
2. The actual roasting phase, during which a number of
complex pyrolytic reactions take place. The chemical
composition of the beans is drastically modified, with
release of large amounts of carbon dioxide and the
formation of the many hundreds of substances asso-
ciated with coffee aroma and taste. The beans change
colour to a dark brown. Initially the process is
exothermic. Pyrolytic reactions reach a maximum
of 190–210 °C, when the process becomes endo-
thermic with the release of volatile compounds. The
overall reaction becomes exothermic again at about
210 °C.
3. A final rapid cooling phase to stop the final exother-
mic part of the roasting operation, using air or water
as the cooling agent.
The quantity of heat transferred to the beans is a very
influential parameter of the roasting process. This can
be controlled by the roasting temperature and time. The
colour of the beans is directly correlated to the final
roasting temperature: the higher the temperature, the
darker the coffee, so that colour can be used to define the
end of the operation. The degree of roasting is usually
described as being ‘light’, ‘medium’ or ‘dark’. Roasting
time may be as long as 40 min, as is often the case in
Brazil, or as short as 90 s in fast roasting to produce
high-yield coffees. Roasting time influences the reactions
within the bean: longer roasting periods produce a bitter
coffee lacking a satisfactory aroma, whereas very short
roasting periods may be insufficient for the completion of

100 R. A. BUFFO ET AL.
Copyright © 2004 John Wiley & Sons, Ltd. Flavour Fragr. J. 2004; 19: 99–104
all pyrolytic reactions, resulting in coffee with under-
developed organoleptic characteristics.2
Chemical Composition of Roasted Coffee
The chemical reactions that take place during roasting
have not been completely elucidated. This results from
the extreme difficulty of reproducing or simulating in the
laboratory all the reactions that take place inside a bean,
since:
1. Not all the active precursors of aroma, colour and
taste present in green coffee have been identified.
2. The very large number of precursors give rise to com-
plex reactions, difficult to isolate, since intermediate
products react further.
3. Most reactions take place within intact bean cells,
which have very thick walls. These walls are com-
parable to autoclaves, where the pressure can only be
guessed.2
Roasting produces a net loss of dry matter, primarily
as gaseous carbon dioxide, water and volatiles products
of pyrolysis. There is a considerable degradation of poly-
saccharides, sugars, amino acids and chlorogenic acids,
and a moderate relative increase in organic acids and
lipid content. Roasting also produces high levels of
caramelization and condensation products. Caffeine and
trigonelline (N-methyl nicotinic acid) concentrations
remain almost unchanged.5
Non-volatile Components
As aforementioned, chemical components of roasted
coffee can be grouped into volatile and non-volatile sub-
stances, some of the former being responsible for the
aroma and the latter for the basic taste sensations of sour-
ness, bitterness and astringency. The non-volatile com-
pounds in roasted coffee are:2,5,11
1. Caffeine, which contributes to the strength, body and
bitterness of brewed coffee.
2. Trigonelline and its two non-volatile derivatives:
nicotinic acid (of nutritional value) and N-
methylnicotinamide.
3. Proteins and peptides that did not undergo Maillard
reaction.
4. Polysaccharides, namely cellulose, hemicellulose,
arabinogalactan and pectins, that play an import-
ant role in the retention of volatiles and contribute to
coffee brew viscosity.
5. Humic acids or melanoidins, as final products of the
Maillard reaction between amino acids and mono-
saccharides; they are the brown-coloured substances
that impart to roasted coffee its characteristic colour.
6. Carboxylic acids, mainly citric, malic and acetic, re-
sponsible for sourness.
7. Chlorogenic acids, mainly cinnamic, caffeic, ferulic,
isoferulic and sinapic, and their main degradation
product, quinic acid, all responsible for astringency.
8. Lipids, including triglycerides, terpenes, tocopherols
and sterols, that contribute to brew viscosity.
9. Minerals, with potassium being the most abundant
(around 40% w/w). Metals such as manganese (10–
50 ppm), iron (15– 40 ppm) and copper (2–5 ppm)
may catalyse reactions occurring during roasting and
storage.
Carbon Dioxide
Carbon dioxide is quantitatively the most important non-
aroma-contributing volatile compound in roasted coffee.
It is generated by pyrolysis and the Strecker degradation
reaction.6 The amount is dependent on the degree of roast
and can be up to 10 ml/g coffee.7 This gas is released
slowly in whole beans but rapidly after grinding. Because
it has a zero dielectric constant, carbon dioxide is very
soluble in the coffee matrix. Its loss from the bean is
most likely due to relationships between temperature and
the glass transition state in the system, although this later
physical effect has not been explored.
After coffee grinding, much of the carbon dioxide pro-
duced during roasting is released. A suitable degassing
time (2–8 h) must be applied before packaging using her-
metic containers. Efficient degassing of roasted ground
coffee is critical when it is packaged in flexible bags
without vacuum. If the time for degassing is too short,
the package can swell, losing appearance and even burst-
ing. To solve this, gas valves can be placed in the wall
of plastic bags to allow the release of carbon dioxide.8
It has been reported that 45% of the carbon dioxide is
released during the first 5 min after grinding.9 One work
in the literature states that 1.21 ml/g carbon dioxide and
0.002 ml oxygen/g roasted ground coffee (Arabica) were
released during the first hour after grinding.10 As particle
size decreases, the release of carbon dioxide is facilitated
because of greater surface-to-volume ratios. A decrease
of particle size from 1000 to 500 µm has been reported
to double the release of the gas.7 Pressurization with
nitrogen and carbon dioxide itself helps to prolong the
shelf-life of espresso coffee, because it maintains carbon
dioxide inside the coffee cells, avoiding cell wall damage
and further flavour release.11
Volatiles
The mechanisms of formation of coffee aroma are
extremely complex and there is clearly a wide range of

COFFEE FLAVOUR: AN OVERVIEW 101
Copyright © 2004 John Wiley & Sons, Ltd. Flavour Fragr. J. 2004; 19: 99–104
interactions between all the routes involved. The major
mechanisms include:2,12,13
1. Maillard reaction (non-enzymatic browning): a reac-
tion between nitrogen-containing substances on the
one hand (proteins, peptides, amino acids, seroto-
nine and trigonelline) and reducing carbohydrates,
hydroxy-acids and phenols on the other, to form
aminoaldoses and aminoketones by condensation.
2. Strecker degradation: a reaction between an amino
acid and an
α
-dicarbonyl with the formation of an
aminoketone that condenses to form nitrogen hetero-
cyclic compounds or reacts with formaldehyde to
form oxazoles.
3. Breakdown of sulphur amino acids, viz. cystine,
cysteine and methionine, that are transformed into
mercaptans, as well as thiophenes and thiazoles,
after reacting with reducing sugars or intermediate
products of the Maillard reaction.
4. Breakdown of hydroxy-amino acids, viz. serine and
threonine, able to react with sucrose to form mostly
alkylpyrazines.
5. Breakdown of proline and hydroxyproline, that react
with intermediate Maillard products; the former gives
pyridines, pyrroles and pyrrolyzines, whereas the
latter forms alkyl-, acyl- and furfurylpyrroles.
6. Degradation of trigonelline, forming alkyl-pyridines
and pyrroles.
7. Degradation of the quinic acid moiety, forming
phenols.
8. Degradation of pigments, mostly carotenoids.
9. Minor lipid degradation, primarily diterpenes.
10. Interaction between intermediate decomposition pro-
ducts (mostly unknown).
Aroma Volatiles in Roasted Ground Coffee
Active research on the volatile composition of coffee has
taken place in the last 25 years with the advent of gas
chromatography–mass spectrometry. The principal classes
of aroma compounds in roasted ground coffee are listed
in Table 1.
Significant Contributors to Aroma
The identification of new volatiles in roasted coffee may
well become an exercise in futility because the central
issue is to determine the chemicals effectively respons-
ible for coffee aroma.12 Gas chromatography–olfactometry
(GCO) techniques, such as AEDA (aroma dilution extrac-
tion analysis) and CharmAnalysis™ have allowed the
individualization of those key odorants by directly sniff-
ing the chromatographic eluent. From the list of around
80 compounds proposed by Holscher and Steinhart,26
Table 1. Classes of volatile compounds identified in
roasted coffee
Sulphur compounds
Thiols14
Hydrogen sulphide15
Thiophenes (esters, aldehydes, ketones)16
Thiazoles (alkyl, alcoxy and acetal derivatives)17
Pyrazines
Pyrazine itself18
Thiol and furfuryl derivatives19
Alkyl derivatives (primarily methyl and dimethyl)20
Pyridines
Methyl, ethyl, acetyl and vinyl derivatives14
Pyrroles
Alkyl, acyl and furfuryl derivatives14,21,22
Oxazoles23
Furans
Aldehydes, ketones, esters, alcohols, acids, thiols, sulfides and in
combination with pyrazines and pyrroles24
Aldehydes and ketones
Aliphatic and aromatic species25
Phenols26
Semmelroch et al.,13 using AEDA, selected only 14
compounds to reproduce coffee aroma based on odour
activity values (OAV), i.e. the ratio of concentration to
odour threshold (Table 2). Deibler, Acree and Lavin27
used CharmAnalysis™ to signal 30 potent odorants in
brewed coffee, of which 18 compounds were identified
by comparing their mass spectra, odour activity and
Kovát’s retention indices with those of authentic stand-
ards (Table 3).
Semmelroch and Grosch28 quantified these potent
odorants in ground roasted Arabica and Robusta coffee
and in the corresponding brews, finding that a change in
aroma takes place from roasted to brewed coffee due to
a shift in the concentration of the odorants. The polar
odorants, e.g. 4-hydroxy-2,5-dimethyl-3(2H)-furanone,
were preferentially extracted, whereas the yield of the
unpolar ones, e.g. 3-isobutyl-2-methoxy pyrazine, (E)-
β
-
damascenone and 2-furfurylthiol, was low. On the basis
of the quantitative data obtained, these authors formulated
synthetic mixtures (aroma models) for the Arabica and
Robusta brews. The models smelled clearly coffee-like
and reproduced the differences in the odour profile of the
original brews.
For roasted Arabica coffee, Czerny, Mayer and
Grosch29 carried out sensory assessments of model
systems of 27 odorants in an oil/water mixture. An
expert sensory panel evaluated the changes in the
overall odour after omission of one or more odorants.
These omission experiments indicated that 2-furfurylthiol,
4-vinylguaiacol, several alkylpyrazines, furanones,
acetaldehyde, propanal, methyl propanal and 2- and 3-
menthylbutanal had the greatest impact on the flavour
of ground coffee. In contrast, the absence of 2,3-
butanedione, 2,3-pentanedione,
β
-damascenone and vanil-
lin in the aroma model according to the corresponding

102 R. A. BUFFO ET AL.
Copyright © 2004 John Wiley & Sons, Ltd. Flavour Fragr. J. 2004; 19: 99–104
Table 2. Odour activity values (by AEDA) and formation mechanisms of volatiles in
the aroma of roasted ground Arabica coffee13
Compound Odour activity value Formation mechanism
(E)-
β
-Damascenone 2.7 × 105Carotene degradation
2-Furfurylthiol 1.7 × 105Maillard reaction
3-Mercapto-3-methylbutylformate 3.7 × 104Maillard reaction
5-Ethyl-4-hydroxy-2-methyl-3(2H)-furanone 1.5 × 104Maillard reaction
4-Hydroxy-2,5-dimethyl-3(2H)-furanone 1.1 × 104Maillard reaction
Guaiacol 1.7 × 103Phenol degradation
4-Vinylguaiacol 1.1 × 103Phenol degradation
Methional 1.2 × 103Maillard reaction
2-Ethyl-3-dimethylpyrazine 1.6 × 102Maillard reaction
2,3-Diethyl-5-methylpyrazine 95 Maillard reaction
3-Hydroxy-4,5-dimethyl-2(5H)-furanone 74 Maillard reaction
Vanillin 48 Phenol degradation
4-Ethylguaiacol 32 Phenol degradation
5-Ethyl-3-hydroxy-4-methyl-2(5H)-furanone 21 Maillard reaction
omissions was not noticed by the assessors. This study
confirmed for the most part the compounds previously
identified by Semmelroch et al.13 with the exception of
(E)-
β
-damascenone, the compound with the highest odour
activity value. Although the degradation of carotenoids
that generates
β
-damascenone has been regarded as
having a major importance on tea flavour and is even
suggested to play an important role in coffee aroma as
well,30 sensory studies did not validate this possibility.
In a follow-up study, Mayer, Czerny and Grosch31
quantified the most potent odorants in brewed medium-
roasted Arabica coffee. In accordance with previous re-
sults,28 the more polar compounds yielded much larger
amounts than the unpolar ones (>75% vs. <25%, respect-
ively). An aroma model was prepared based on these
results for the brew. In triangle tests, the model contain-
ing all 24 odorants was compared with a set of models
missing one or more compounds using the afore-
mentioned omission tests. These experiments indicated
that the aroma of the brew was mainly caused by
some alkylpyrazines, furanones and phenols, and by
2-furfurylthiol, methional, and 3-mercapto-3-methylbutyl
formate. The higher impact of methional and formate and
the lower aroma activity of 4-vinylguaiacol were in con-
trast to the results previously obtained for ground coffee
of the same provenance and roast degree.
Finally, Mayer and Grosch32 quantified 22 potent
odorants in the headspace of roasted Arabica coffee pow-
der, and a model mixture was prepared accordingly.
When evaporated, the similarity of the aroma of the
model to that of the roasted coffee headspace was scored
2.6 on a scale of 0 (no similarity) to 3.0 (identical). The
model included odorants such as acetadehyde, methyl-
propanal, 2- and 3-methylbutanal, 2,3-butadione, 2,3-
pentadione, 2-furfurylthiol, 2-ethyl-3,5-dimethylpyrazine
and 2,3-diethyl-5-methylpyrazine.
Chemical Indicators of Sensory
Deterioration
The onset of sensory staleness is used to determine the
end of shelf-life of roasted ground coffee.9 Staleness
in coffee is defined as ‘a sweet but unpleasant flavour
Table 3. Aroma occurrences resulting from CharmAnalysis27
Compound OAV (standardized) Sensorial description
Sotolon 81 Toast
β
-Damascenone 98 Fruit
2-Furfurylthiol 100 Toast
4-Vinylguaiacol 62 Cloves
2-Methyl-3-furanthiol 89 Nuts
Vanillin 71 Vanilla
Guaiacol 77 Plastic
Furaneol 58 Caramel
Methional 43 Potato
3-Methoxy-2 isobutyl pyrazine 38 Plants
2,4,5-Trimethylthiazole 38 Plastic
Abhexon 44 Honey
4-Ethyl guaiacol 31 Spice
5-Methyl-6,7-dihydrocyclopentapirazine 22 Cotton candy
2-Ethyl-3,5,-dimethylpyrazine 25 Burnt
cis-2-Nonenal 27 Toast
2-Isopropyl-3-methoxypyrazine 14 Green
2,3,5-Trimethylpyrazine 15 Toast

COFFEE FLAVOUR: AN OVERVIEW 103
Copyright © 2004 John Wiley & Sons, Ltd. Flavour Fragr. J. 2004; 19: 99–104
and aroma of roasted coffee which reflects the oxidiza-
tion of many of the pleasant volatiles and the loss of
others; a change in the flavour and the acid constituents
causing a partial bland tone’.33 Since the late 1950s
when the gas chromatographic work on coffee aroma was
initiated, a great evolution in analytic and data pro-
cessing techniques has occurred, and different staling
indexes have been developed and correlated to sensory
deterioration.
‘M/B’ Aroma Index
Raymond et al.34 defined the M/B aroma index as the
quotient of the concentrations of methylfuran and 2-
butanone. This ratio decreased from 2.6 to 0.1 within 4
days in roasted ground coffee dissolved in water at 23 °C
and 30 °C, which led to the suggestion that the M/B
index could be used as an indicator of staling. However,
no sensory results were presented and, in fact, this was a
storage study of a ‘coffee solution’ and not the roasted
ground coffee itself. Later Vitzthum and Werkhoff35
found a linear correlation (r2 = 0.96) between M / B
and a sensory index reported on a 1–5 scale (5 = best,
1 = worst) for the aroma of roasted ground coffee stored
at 20 °C in air (unstated moisture).
Further studies focused on the influence of the degree
of roasting and origin of coffee on the M/B index.
Kwasny and Werkhoff36 observed that the M/B index
determined in one kind of coffee might not be directly
applied to others. An empirical exponential function was
used to correlate the M/B index with degree of roasting:
M/B = a (degree of roasting)b
Robusta a = 31.3 b = 0.59 (r = −0.9999)
Salvador a = 20.5 b = 0.49 (r = −0.9898)
Kenya a = 21.9 b = 0.52 (r = −0.9900)
Colombia a = 28.4 b = 0.60 (r = −0.9951)
Furthermore, Spadone and Liardon37 reported that the
M/B index decreases continuously irrespective of storage
temperature or oxygen partial pressure in the package
(metal cans at 5% moisture, unstated basis). However,
correlations between M/B index and sensory evaluation
of brewed coffee prepared from the stored samples did
not show a linear trend, thus undermining the validity of
the parameter.
Flavour Quality Index (FQI)
Looking for a better index, Spadone and Liardon37
applied multiple linear regression to results of volatile
composition of roasted ground coffee vs. time of storage,
and defined a flavour quality index (FQI) based on five
key volatiles, viz. hexanal, vinylpyrazine, pyrrol, furfuryl-
methylketone and pyridine, with a good fit of their sen-
sory data (r2 = 0.87).
FQI = 6.53 + 0.027 [hexanal] − 0.08 [vinylpyrazine]
− 0.04 [pyrrol] − 0.022 [furfurylmethylketone]
− 0.001 [pyridine]
Nevertheless, the equation was only an experimental fit
and did not have a theoretical basis.
‘M/M’ Aroma Index
Vitzthum and Werkhoff35 defined the M/M aroma index
as being the ratio of the concentrations of methanol to 2-
methylfuran. This index showed an inverse linear relation
with the M/ B index and has been used to follow coffee
staling in industrial settings.
‘S’ Aroma Index
Radtke and Piringer38 defined the ‘S’ aroma index as the
sum of the concentrations of 2-methylpropanal, 3-
methylbutanal, diacetyl and 2-methylfuran, determined by
gas chromatographic analysis of the static headspace of
10 g coffee dissolved in 40 ml water at room temperature.
They also found a linear correlation between the index
and sensory evaluation by a panel of experts.
Sulphur Compounds as Aroma Index
Steinhart and Holscher39 applied headspace profile ana-
lysis in combination with a computer-aided discrimin-
ant analysis to evaluate aroma freshness of roasted
whole coffee beans in non-air-tight packs at 20 °C. Their
results indicated that aroma freshness of whole roasted
coffee is mainly determined by certain low-boiling com-
ponents, namely low-molecular Seight sulphur com-
pounds, Strecker aldehydes and
α
-dicarbonyls. The loss
of aroma freshness was due to the loss of certain aroma
impact volatiles, mainly methanethiol, which then could
be used as an indicator of freshness. The heavier com-
ponents, such as furfurylmercaptan, will remain in the
coffee and will cause an ‘ageing’ note. In addition, an
increase of dimethylsulphide, the oxidation product of
methanethiol, was observed, although no ‘true’ staling
compound was found.
Summary
Past research on coffee flavour has built a significant
knowledge base. About 900 volatile compounds have

104 R. A. BUFFO ET AL.
Copyright © 2004 John Wiley & Sons, Ltd. Flavour Fragr. J. 2004; 19: 99–104
been identified, although less than 20 of them have been
found to be relevant to coffee aroma. In addition, chem-
ical indicators of sensory deterioration have been pro-
posed and validated. Future research efforts will probably
focus on the elucidation of mechanistic and kinetic para-
meters for the formation and deterioration of key aroma
compounds. This would translate into practical guidelines
for producing and maintaining a good coffee aroma.
References
1. Clarke RJ. The Shelf Life of Coffee. Elsevier: London, 1993.
2. Illy A, Viani R. Espresso Coffee: The Chemistry of Quality.
Academic Press: London, 1995.
3. Vincent J. 1987. Coffee, vol 2: Technology. Elsevier: London.
4. Vitzthum OG. In Flavor Chemistry: 30 Years of Progress.
Teranishi R, Wick EL, Horstein I (eds). Kluwer: New York, 1999;
117–134.
5. Clarke RJ, Macrae R. Coffee Technology, vol 2. Elsevier: London,
1987.
6. Hodge JE. J. Ag. Food Chem. 1953; 1: 928.
7. Barbera CE. In Proceedings of the 3rd, ASIC Meeting, Trieste,
1967; 436– 442.
8. Labuza TP. Shelf-life Dating of Foods. Food & Nutrition Press:
Westport, CT, 1982.
9. Heiss R, Radtke R, Robinson L. In Proceedings of the 14th ASIC
Meeting, Colloquium on Coffee, Abidjan, 1977; 163–174.
10. Cartwright LC, Snell CT. Spice Mill 1974; 70(3), 16–24.
11. Illy E. The seduction of espresso. Presented at the 9th Annual
SCCA Conference and Exhibition, New Orleans, 1997.
12. Reineccius GA. In Proceedings of the 16th ASIC Meeting, 1995,
Kyoto; 249–257.
13. Semmelroch P, Laskawy G, Blank L, Grosch W. Flavour Fragr.
J. 1995; 10: 1.
14. Sivetz M, Desrosier NW. Coffee Technology. AVI: Westport, CT,
1979.
15. Shankaranarayana ML, Raghvan B, Abraham KO, Natarajan CP.
In CRC Crit. Rev. Food Technol. 1974; 4(3): 395– 435.
16. Bochmann G, Baltes W. In Amino-carbonyl reactions in food
and biological systems, Fujimaki M, Namiki M, Kato H (eds),
Developments in Food Science Conference Proceedings, 1983;
245–255.
17. Maga JA. In CRC Crit. Rev. Food Sci. Nutr. 1975; 153 –176.
18. Schutte L. CRC Crit. Rev. Food Technol. 1974; 4(4): 457.
19. Maga JA. In CRC Crit. Rev. Food Sci. Nutr. 1982; 1– 48.
20. Koehler PE, Mason ME, Newell JA. J. Ag. Food Chem. 1969; 17:
393.
21. Tressl R, Grunewald KG, Silwar R. Chem. Mikrobiol. Technol.
Lebensm. 1981; 7(1): 28.
22. Mabrouk AF. In Food Taste Chemistry, Boudreau, JC (ed.). ACS:
Washington, DC, 1979; 205– 245.
23. Dart SK, Nursten HE. Dev. Food Sci. 1985; 10: 239– 252.
24. Hurrel RF. In Developments in Food Science, Morton ID,
McLeod AJ (eds). Elsevier: Amsterdam, 1982; 399– 457.
25. Hodge JE. In Chemistry and Physiology of Flavors, Schultz HW,
Day EA, Libbey LM (eds). AVI: Westport, CT, 1967; 465– 491.
26. Holscher W, Steinhart H. In Thermally Generated Flavors, Parlia-
ment TH, Morello MJ, McGorrin RJ (eds). ACS: Washington,
DC, 1994; 204 –210.
27. Deibler KD, Acree TE, Lavin EH. In Food Flavors: Formation,
Analysis and Packaging Influences, Contis ET, Ho C-T, Mussinan
CJ et al. (eds). Elsevier: Amsterdam, 1998; 69–78.
28. Semmelroch P, Grosch W. J. Ag. Food Chem. 1996; 44: 537.
29. Czerny M, Mayer F, Grosch W. J. Ag. Food Chem. 1999; 47:
695.
30. Grosch W. Flavour Fragr. J. 1994; 9: 147.
31. Mayer F, Czerny M, Grosch W. Eur. Food Res. Technol. 2000;
211: 272.
32. Mayer F, Grosch W. Flavour Fragr. J. 2001; 16(3): 180.
33. Sivetz M. Coffee Processing Technology, vol 2. AVI: Westport,
CT, 1963.
34. Reymond D, Chavan F, Egli RH. In Proceedings of the 1st
International Congress in Food Science and Technology. Food
Sci. Technol. 1962; 4: 595.
35. Vitzthum OG, Werkhoff P. In Analysis of Foods and Beverages:
Headspace Techniques, Charalambous G (ed.). Academic Press:
New York, 1978; 115–133.
36. Kwasny H, Werkhoff P. Chem. Mikrobiol. Technol. Lebensm.
1979; 6(1): 31.
37. Spadone JC, Liardon R. In Proceedings of the 13th ASIC Meeting,
Paipa, 1989; 145–158.
38. Radtke R, Piringer OG. Deutsch. Lebensm.-Rundsch. 1981; 77(6):
203.
39. Steinhart H, Holscher W. In Proceedings of the 14th ASIC Meet-
ing, San Francisco, 1991; 156–164.