AROMA IMPACT COMPOUNDS
*, A. SEN **, W.
The volatile fraction of roasted coffee is extremely complex, consisting of
more than 700 compounds  with a wide variety of functional groups.
During the last decade, efforts have been undertaken to evaluate those vo-
latile compounds which contribute significantly to the aroma of roasted
On the basis of the odor unit concept , Tressl  has suggested
that 2-furfurylthiol is the most important odorant. In addition, he repor-
ted that the other compounds listed in Table 1 are of significance for the
coffee flavor. Recently, Holscher et al. [4, 5], using gas chromatogra-
phy/olfactometry (GC/O) of serial dilutions of the volatile fraction (aroma
extract dilution analysis, AEDA ), confirmed that some of the odorants
suggested by Tressl  are indeed intensely involved in the coffee flavor
(Table 1). In addition, these authors identified further character impact
compounds which are summarized in Table 1.
It is well-known [review in 7] that the two varieties of coffee, Arabica
and Robusta, differ in their aromas. Several authors have compared the vo-
latile fractions of the two varieties, but they have not evaluated the con-
tribution of the identified compounds to the flavor differences of the two
Vitzthum et al.  have recently reported, that 2-methylisobor-
neol resembles the typical earthy aroma impression of the Robusta coffee.
Table 1: Important odorants of the coffee flavor according to Tressl 
and Holscher et al. [4, 5]
Compound Tressl  Holscher et al. [4, 5]
Methional - +
Ethyldimethylpyrazine + +
Acetylpyrazine + -
Trimethylpyrazine - +
Linalool - +
Guaiacol + +
Furaneo1 + +
(diacetyl) - +
acid - +
Acetic acid - +
The aim of the following study was to identify the potent odorants of Ara-
bica and Robusta coffee (powder and brew) and to show the differences of
these two varieties. The identification experiments were focussed on those
compounds which were evaluated by AEDA as important odorants. As AEDA is
only a screening method , some odorants, contributing to important notes
within the odor profile of the two varieties, were quantified and the odor
units were calculated on the basis of odor/taste threshold values in water.
The methods used for the analysis of the coffee aroma are summarized in
Table 2. The volatiles were isolated by solvent extraction from both, the
roasted ground powder and the brew. The aroma extract was separated from
the non-volatile compounds by high vacuum transfer and the volatile frac-
tion obtained was analysed by GC/O. The odorants were characterized by
their retention index, odor quality and relative odor activity
The odorants with high FD-factors were identified and their odor threshold
values were determined. Important odorants were quantified using stable
isotope dilution methods and then compared on the basis of odor activity
values calculated by dividing the concentration levels in coffee (powder
and brew) through the flavor thresholds in water.
Methods used for the analysis of coffee aroma
Solvent extraction, high vacuum
Aroma extract dilution analysis
(column chromatography, HPLC,
(Capillary GC, MS, NMR)
Sensory characterization 
Stable isotope dilution assay
odor units 
Retention index and odor of the volatiles
FD-Chromatogram (FD-factors: ranking of
the volatiles on the basis of their odor
units determined by GC/O)
Threshold values (in air, water)
Reference compounds, labeled compounds
Quantitative data of some important
their flavor significance
expressed as odor units
About 50-60 odorants were found in the GC-effluent of the aroma extracts of
Arabica and Robusta coffee. As examplified for the Arabica coffee (roasted
powder) the FD-chromatogram fFig. 1) indicated 38 odorants with FD-factors
> 16. Most of the 38 odorants were identified on the basis of GC and MS
data (footnote "d" in Table
Only the amounts of the compounds nos. 5,
21 and 37 were so low in the volatile fraction that no clear MS signals
were obtained. The identification of these compounds as 2-methyl-3-furan-
thiol (no. 5), 2,3-diethyl-5-methylpyrazine (no. 21) and bis(2-methyl-3-
furyl)disulphide (no. 37) was based only on comparison of GC and sensory
data with that of the corresponding reference substances. Their sensory im-
portance can be explained by the low threshold values (0.001-0.01 ng/1 air)
FD-Chromatogram of Arabica Coffee (roasted powder)
21 26 30 333/,
Figure 1. Flavor dilution chromatogram of the volatiles isolated from the
roasted powder of Arabica coffee. Numbering of the flavor com-
pounds as in Table
FD-factor: flavor dilution factor; RI: re-
Chimie I Biochimie
Three odorants (nos. 14, 17 and 35) appeared with the highest FD-factor (FD
= 2048) in the FD-chromatogram (Fig.
As shown in Table 3 they were
identified as 3-mercapto-3-methylbutylformate (no. 14), 3,5-dimethyl-2-
ethylpyrazine (no. 17) and (E)-ß-damascenone (no. 35). Their odor qualities
were described as "catty", "earthy-roasty" and "honey-like", respectively.
Further odorants summarized in Table 3 were sotolon (no.
4-methoxybenzaldehyde (no. 32) and bis(2-methyl-3-furyl)disulphide
37). These compounds were identified for the first time in the coffee
Sotolon and abhexon smelled "seasoning-like" and, in higher dilu-
"caramel-like". Other volatiles contributing with "caramel-like,
odor qualities to the aroma of coffee were furaneol (no. 24), 3,4-
dimethyl-2-cyclopentenol-l-one (no. 22) and an unknown compound (no. 29).
The identification of such polar enoloxo compounds in low concentrations is
difficult, since they are more or less adsorbed at the capillary during the
GC-procedure . It was observed that the lower the amount injected on
the capillary the higher the proportion which was adsorbed. These effects
were the smallest on the FFAP stationary phase which was, therefore, used
for the AEDA of the enoloxo compounds.
Odorants with the impression "sweet" in combination with an additional odor
quality were diacetyl (no. 1,
2,3-pentandione (no. 3,
methional (no. 8,
p-anisaldehyde (no. 32,
sulphide (no. 37, meaty) and vanillin (no. 38,
About one third of the potent odorants were described as "roasty" in combi-
nation with an additional odor quality. 2-Furfurylthiol (no. 6, coffee-
3-mercapto-3-methylbutylformate (no. 14,
propylpyrazine (no. 15,
3,5-dimethyl-2-ethylpyrazine (no. 17,
2,3-diethyl-5-methylpyrazine (no. 21, earthy) and two unknown com-
pounds (nos. 19 and 26, earthy) belonged to this group.
Important odorants with "honey-like" or "spicy" aroma qualities were
damascenone (no. 35), phenylacetaldehyde (no. 18) and guaiacol (no. 23), 4-
and 4-vinylguaiacol (nos. 31 and 34), respectively.
The compounds found in this study as important for the flavor of roasted
coffee powder are in good agreement with those reported by Holscher et al.
The presence of 2-furfurylthiol, ethyldimethylpyrazine,
2-methoxy-3-isobutylpyrazine, 4-vinylguaiacol, furaneol and 2,3-pentandione
Important odorants of the roasted powder of Arabica coffee
(FD > 16)
Footnotes of Table 3
a Numbering as in Fia. 1.
RI: Retention index on the capillary .
Odor quality perceived at the sniffing-port.
The compound was identified by comparing it with the reference substance
on the basis of the following criteria:
RI data (on OV-
SE-54 and FFAP) as well as of the odor quality and threshold, which
was perceived at the sniffing-port.
The MS signals of the substance were too weak for an interpretation; the
compound was only identified by comparing it with the reference substance
on the basis of the resting criteria reported in footnote d.
also agrees with the suggestion of Tressl , that these odorants belong
to the key compounds of the coffee flavor. In the capillary gas chromato-
grams of the coffee volatiles, Holscher et al.  have localized two
tional odorants of unknown structure with high FD-factors. In the present
study, they were identified as sotolon and abhexon.
Differences between the powders
roasted Arabica and Robusta coffee
The odorants of roasted Arabica and Robusta coffee powders were compared on
the basis of their FD-factors. All of the odorants identified with FD-fac-
tors > 16 contribute to the aroma of both coffee varieties. Linalool was an
exception occurring only in Arabica coffee. According to the data summari-
zed in Table 4 3,5-dimethyl-2-ethylpyrazine appeared with the highest FD-
factor in both coffee varieties. However, differences were found in the
concentration levels of some important coffee odorants.
2,3-Diethyl-5-methylpyrazine and 4-ethylguaiacol were predominant in the
Robusta coffee and 3-mercapto-3-methylbutylformate, sotolon and abhexon in
the Arabica coffee. Further significant differences were found for 2-me-
thyl-3-furanthiol, phenylacetaldehyde, 3,4-dimethyl-2-cyclopentenol-l-one,
2-/3-methylbutanoic acid and linalool, all predominating in the Arabica
and for 3-methyl-2-buten-l-thiol, which prevailed in the Robusta
The comparison of the coffee varieties indicated in addition that compounds
causing "caramel-like, sweet-roasty" odor qualities were high in Arabica
while those having "spicy" and "earthy-roasty" qualities con-
tributed more significantly to those of the Robusta species. These
Table 4: Important odorants of roasted Arabica and Robusta coffee (powder
7 2-/3-Methylbutanoic acid
as in Fig. 1 and
each compound (Table
compound with highest FD-factor which
was set to 100.
Both coffee varieties were roasted
coffee powder: 300-500
54 g of the
1 L of hot
Chimie I Biochimie
differences in the composition of the important odorants corresponded to
the differences in the overall aromas of the two varieties.
Differences between the brews
Arabica and Robusta coffee
A comparison of the odorants isolated from the brews of Arabica and Robusta
coffee (Table 4) revealed a shift in the predominating flavor compounds.
Sotolon, abhexon, furaneol and 3,4-dimethyl-2-cyclopentenol-l-one showed
significant higher FD-factors in the Arabica than in the Robusta coffee.
This difference suggested that these odorants were mainly responsible for
the "sweet, mild" aroma of the Arabica coffee.
During the extraction with hot water the water-soluble enoloxo compounds
were enriched in the brew and,
enhanced the caramel-like flavor
in particular of the Arabica coffee. Hodge  has reported that a
planar enoloxo structural element in a volatile compound is responsible for
the caramel-like odor impression.
The aroma of the Robusta coffee brew was mainly influenced by compounds ha-
ving "roasty-earthy" and "spicy" odor qualities like 2,3-diethyl-5-me-
thylpyrazine, 3,5-dimethyl-2-ethylpyrazine, 4-ethylguaiacol, 4-vinylguaia-
col and the odorant no. 19. These odorants, in combination with the lower
amounts of compounds having caramel-like aromas, were responsible for the
earthy, less pleasant" flavor notes of Robusta coffee.
Compared to the powders, vanillin, methional, furaneol and sotolon in-
creased in the brews, in particular of the Arabica coffee. On the other
hand (E)-ß-damascenone, (E)-2-nonenal, the temperature labile thiols 3-mer-
capto-3-methylbutylformate, 2-furfurylthiol, 2-methyl-3-furanthiol, 3-me-
thyl-2-buten-l-thiol, and also linalool decreased strongly. This effect was
also observed for 2,3-diethyl-5-methylpyrazine and 2-methoxy-3-isobu-
tylpyrazine, which also decreased especially in the Arabica coffee.
The quantitative analysis of the compounds were performed by means of a
stable isotope dilution analysis (SIDA) in order to compensate for losses
during the isolation procedure . In the SIDA the odorant labeled with a
stable isotope is used as internal standard. Until now a SIDA was developed
for the quantification of furaneol, diacetyl , 2,3-pentandione, 4-
some important odorants
and Robusta coffees
stable isotope dilution assay.
internal standard with
Concentration: mg/1 brew prepared from
roasted coffee powder.
Sotolon was determined using d-j-abhexon
determined using d
Odor activity values (OAV)
important flavor compounds
Odor threshold (nasal
Flavor threshold (retronasal perception)
ethylguaiacol, abhexon, (E)-ß-damascenone and 3-mercapto-3-methylbutylfor-
mate in the brews of both, Arabica and Robusta coffees.
As shown in Table 5. the amounts of diacetyl and 2,3-pentandione were hig-
her in Arabica than in Robusta coffee brew indicating the importance of the
buttery top-notes for the mild aroma of Arabica coffee. This becomes more
clear when comparing the odor activity values (OAV, Table
OAV of diacetyl indicated that this dione contributed more significantly to
the aroma of the brews than 2,3-pentandione.
Quantitative measurements of furaneol which was used as indicator substance
for the caramel-like odorants revealed one reason for the odor difference
between Arabica and Robusta. The concentration of furaneol was 4.5 fold
higher in the brew of Arabica coffee than in the corresponding sample of
the Robusta species (Table
Calculation of OAV (Table 6) confirmed the
stronger effect of furaneol on the flavor of Arabica coffee compared to the
Robusta variety. The concentrations of sotolon and abhexon were lower than
those of furaneol, but also these enoloxo compounds prevailed in the Ara-
bica coffee (Table 5).
The OAV of diacetyl in Arabica coffee brew was
higher than the OAV
of furaneol. In contrast, the FD-factor of furaneol was
difference indicates the great losses of diacetyl during the isolation pro-
cedure compared to the higher boiling compounds like furaneol.
importance of high volatile compounds was underestimated by the AEDA.
The results in Table 5 show, furthermore, the significantly higher concen-
tration of 4-ethylguaiacol in the Robusta coffee compared to the Arabica
This agreed with the sensory data obtained by the AEDA indicating
the importance of phenol-derivatives for the aroma of Robusta coffee.
The predominance of 3-mercapto-3-methylbutylformate in the Arabica coffee
was also in agreement with the results of the AEDA.
Acknowledgements. The work was supported by the Forschungskreis der Ernäh-
rungsindustrie (Hannover) and the AIF
We are grateful to Miss
Kustermann, Miss Foschum and Miss Reinhard for skilful technical
The volatile components of roasted Arabica and Robusta coffees (powder and
brew) were analysed by gas chromatography-olfactometry (GC/O) which
revealed the odorants having the highest odor-activity values (ratio of
concentration to odor
This procedure resulted in 38 odorants of
which 32 were identified. The powders of the two coffee varieties differed
in the concentration levels of these compounds.
The results indicate that the flavor difference between Arabica and Robusta
coffee (powder and brew) are mainly due to the predominance of enoloxo com-
pounds (sotolon, abhexon, furaneol, 3,4-dimethylcyclopentenol-l-one) in the
former and of 3,5-dimethyl-2-ethylpyrazine, 2,3-diethyl-5-methylpyrazine,
4-ethylguaiacol and 4-vinylguaiacol in the latter. Preparation of brews en-
hanced the flavor difference, as the concentration levels of water-soluble
odorants (furaneol, sotolon, abhexon) responsible for the "sweet-caramel"
flavour note increased more in the Arabica than in the Robusta coffee. On
the other hand, the alkylpyrazines and guaiacols were responsible for the
"spicy, harsh-earthy" aroma of the Robusta coffee.
Quantification of selected odorants using a stable isotope dilution assay
confirmed the differences between the Arabica and Robusta coffees (brew)
found by GC/O.
Die flüchtige Fraktion von Arabica- und Robusta-Röstkaffee (Pulver und Ge-
tränk) wurde durch Gaschromatographie-Olfaktometrie (GC/O) untersucht. Die
Analyse ergab 38 aromaaktive Verbindungen, von denen 32 identifiziert wur-
Die beiden Kaffeesorten zeigten Unterschiede in der Konzentration die-
Das Aroma des Arabica-Kaffees wird hauptsächlich durch Enoloxo-Verbindungen
(Sotolon, Abhexon, Furaneol und 3,4-Dimethyl-2-cyclopentenol-l-on) geprägt,
während im Robusta-Kaffee 3,5-Dimethyl-2-ethylpyrazin, 2,3-Diethyl-5-me-
thylpyrazin, 4-Ethyl- und 4-Vinylguajacol überwiegen. Die Bedeutung der
Enoloxo-Verbindungen nimmt im Arabica-Kaffeegetränk wegen der guten Wasser-
löslichkeit zu, so daß sie für das süßlich-karamelartige Aroma verantwort-
lich sind. Die stechend-erdige Aromanote von Robusta-Kaffee wird dagegen
von Alkylpyrazinen und Guajacol-Derivaten geprägt.
Quantitative Daten bestätigten die durch GC/O erhaltenen Ergebnisse in be-
zug auf die Unterschiede von Arabica- und Robusta-Kaffee
 Maarse H, Visscher CA (1989) Volatile Compounds in Food, Volume II,
TNO-CIVO Food Analysis Institute, Zeist, NL
 Guadagni DG, Buttery RG, Harris J (1966) J Sei Food Agrie 17:142-144
 Tressl R (1989) In: Thermal Generation of Aromas. ACS Symposium Series
Parliment TH, McGorrin RJ, Ho C-T (Eds.) American Chemical So-
ciety, Washington, pp. 285-301
 Holscher W, Vitzthum OG, Steinhart H (1990) Lebensmittelchemie
 Holscher W (1991) Thesis, Univ Hamburg, FRG
 Ullrich F, Grosch W (1987) Z Lebensm Unters Forsch 184:277-282
 Clarke RJ, Macrae R (1985) In: Coffee (Vol. 1: Chemistry) Elsevier Ap-
plied Science Publishers, London, New York
 Vitzthum OG, Weisemann C, Becker R, Köhler HS (1990) Cafe Cacao The
 Grosch W, Schieberle P (1988) In: Proceedings of the 2nd Wartburg
Aroma Symposium. Characterization, production and application of food
Akademie-Verlag, Berlin, pp. 139-151
 Blank I, Fischer K-H, Grosch W (1989) Z Lebensm Unters Forsch 189:426-
 Blank I, Grosch W (1991) J Food Sei 56:63-67
 Schieberle P, Grosch W (1987) J Agrie Food Chem 35:252-257
 Gasser U, Grosch W (1990) Z Lebensm Unters Forsch 190:3-8
 Blank I, Grosch W (1991) Lebensmittelchemie, submitted
 Hodge JE (1967) In: The Chemistry and Physiology of Flavors, Schultz
Day EA, Libbey LM (Eds.) AVI Publishing Company, pp. 465-491
 l^Cg-diacetyl was synthetised by C. Gassenmeier and P. Schieberle,
Deutsche Forschungsanstalt für Lebensmittelchemie, Garching, Germany
 Sen A, Laskawy G, Schieberle P, Grosch W (1991) J Agrie Food Chem