Abstract An overall sensory evaluation of a freshly pre-pared filtered coffee brew (FC) and an instant coffeebeverage (IC) using a sensory panel revealed a clear dif-ference in the aromas of both beverages, when smelled at60 °C. Application of the aroma extract dilution analysis(AEDA) on an extract of FC indicated 40 odour-activecompounds, of which (E)-β-damascenone, methional, 3-mercapto-3-methylbutyl formate, 5-ethyl-3-hydroxy-4-methyl-2(5H)-furanone, 4-hydroxy-2,5-dimethyl-3(2H)-furanone and several phenols were detected as the mostintense odorants. In IC, the same odorants were identi-fied based on an evaluation by AEDA, but the flavourdilution (FD) factors of nearly every compound weremuch lower. The largest differences were determined forsome sulfur-containing odorants (e.g., 2-methyl-3-furan-thiol, 2-furfurylthiol, 3-mercapto-3-methylbutyl formate)as well as 2-methoxyphenol, 4-ethyl-2-methoxyphenol,4-vinyl-2-methoxyphenol, vanillin and an unknownmeaty smelling compound, which were much higher inFC. The lack of these, in combination with the un-changed odour activities of, in particular, the caramel-like, seasoning-like smelling odorants is undoubtedly in-volved in the flavour differences between FC and IC.

Keywords Coffee beverage · Odorants · Aroma extract dilution analysis · Instant coffee · Sensory evaluation

Introduction

Investigations performed within the last decade haveshown that only a certain number of the nearly 1000 vo-

latiles present in roasted coffee powder really contributeto its characteristic overall aroma [1, 2, 3, 4] and studiesattempting to mimic the overall roasty, coffee-like aromabased on quantitative data and the use of reference com-pounds have recently been successful [4].

Coffee powder is, however, not used as such and sev-eral techniques are used to prepare an aqueous coffeebeverage thereof. The most common way is by brewing(extraction during filtration) or by infusion, but also di-rect boiling of the powder with hot water is quite popular[5]. The hot water treatment leads to the extraction ofcaffeine, the brown coloured melanoidins, acids and nu-merous further non-volatile compounds. Furthermore,the aroma volatiles are extracted. Recent investigationshave, however, shown that the extraction procedure re-sults in different yields [6, 7], depending on the chemicalstructure of the respective odorants.

About 50 years ago, the technological developmentled to the introduction of soluble coffee. The fast andsimple preparation of just mixing the instant powderwith hot water, led to a good consumer acceptance, inparticular, in England and Japan [10]. However, althoughthe technology of instant coffee production has been im-proved, it is common experience that the aroma of ICstill significantly differs from that of FC.

The key aroma compounds of a filtered coffee brew(FC) have recently been elucidated [7, 8] and the overallaroma has successfully been mimicked using 25 odorantsin the same concentrations as occurring in the brew [9].

Semmelroch and Grosch [6] have compared theodour-active compounds in a filtered coffee brew and aninstant coffee beverage using gas chromatography-olfac-tometry of headspace samples (GCO-H). The qualitativecomposition of both beverages was found to be verysimilar, but the odour intensities of most of the very vol-atile compounds considered were stronger in FC. Pollienet al. [11], who also used volatile isolation by headspacetechniques, partially confirmed this result but they de-tected higher intensities of 3-methylbutanal, 2-methoxy-phenol and some further, unidentified compounds in in-stant coffee.

C. Sanz · C. CidDepartamento de Bromatología, Tecnología de Alimentos y Toxicología, Facultad de Farmacia, Universidad de Navarra, 31080 Pamplona, Spain

M. Czerny · P. Schieberle (✉ )Deutsche Forschungsanstalt für Lebensmittelchemie, Lichtenbergstrasse 4, 85748 Garching, Germanye-mail: peter.schieberle@lrz.tum.deTel.: +49-89-289-13264, Fax: +49-89-289-14183

Eur Food Res Technol (2002) 214:299–302DOI 10.1007/s00217-001-0459-9

O R I G I N A L PA P E R

Cristina Sanz · Michael Czerny · Concepción CidPeter Schieberle

Comparison of potent odorants in a filtered coffee brew and in an instant coffee beverage by aroma extract dilution analysis (AEDA)

Received: 16 August 2001 / Revised version: 29 October 2001 / Published online: 7 March 2002© Springer-Verlag 2002

Up to now, the odour-active compounds in instantcoffee have only been screened using GC/olfactometryof headspace samples (GCO-H). This technique, howev-er, discriminates higher boiling odorants which areknown as further important contributors to coffee aroma,such as 4-hydroxy-2,5-dimethyl-3(2H)furanone or (E)-β-damascenone.

An aroma extract dilution analysis (AEDA) of solventextracts can be applied to measure the sensory impor-tance of also the medium and high boiling odorants. Theaim of this work was, therefore, to establish the sensorydifferences between FC and IC using AEDA and to com-pare the intensities of potent odorants in both beveragesbased on flavour dilution (FD) factors.

Materials and methods

Materials

Ground roasted coffee and instant coffee powder were supplied by the same manufacturer and were made from the variety Coffeaarabica, but from different batches. After manufacturing andtransport, the samples were stored at –30 °C until use. For thepreparation of the filtered coffee brew (FC), hot water (1.1 L,95 °C) was poured onto the coffee powder (30 g) in a filter (No. 4,Plus Warenhandelsgesellschaft, Hamm, Germany) using a domes-tic coffee machine. For instant coffee beverage, a commerciallyavailable instant coffee powder (3.2 g) was dissolved in 200 mL ofhot water (95 °C).

Chemicals

Pure samples of reference compounds nos. 1–3, 7–12, 14, 19, 21,25, 29, 31, 33, 36, 38 and 40 given in Table 2 below were pur-chased from Aldrich (Steinheim, Germany), odorants nos. 28 and34 were obtained from Lancaster (Mühlheim/Main, Germany).Components nos. 22 and 30 were gifts from Haarmann & Reimer(Holzminden, Germany) and Givaudan-Roure (Dübendorf, Swit-zerland), respectively. The following odorants were synthesizedaccording to the literature cited: no. 13 [4], nos. 15 and 16 [3].

High-resolution gas chromatography-olfactometry (HRGC-O) and high-resolution gas chromatography-mass spectrometry(HRGC-MS)

HRGC-O and HRGC-MS were performed as recently reported[12].

Isolation of volatiles

After preparation, FC and IC (100 mL each) were cooled to roomtemperature and then stirred with dichloromethane (25 mL) for5 min. The emulsions formed were distilled using the solvent as-sisted flavour evaporation (SAFE) technique recently developedby us [13]. The distillates obtained were separated into the neutral-basic and the acidic fractions [14] and the fractions were concen-trated to 0.1 mL by distilling off the solvent on a Vigreux column(40×1 cm) followed by microdistillation [15]. Aroma extract dilu-tion analyses (AEDA) were performed as detailed in [15].

Triangular test

A triangular test was performed in a sensory panel room as de-scribed by Schieberle and Hofmann [16]. The panel consisted of

10 experienced assessors. Freshly prepared samples (15 mL) werefilled into covered glass beakers (diameter 40 mm, capacity45 mL) and maintained at 60 °C. One sample of IC and two sam-ples of FC were presented to the panel. In a first session the asses-sors had to evaluate by sniffing the sample which was different.The number of correct answers in the triangular test was summedup and the significance of the result (p<0.001) was determined ac-cording to Jellinek [17]. The confidence limit of 99.9% wasreached, if 10 answers out of 10 were correct.

Assessors, who found out the right sample, were asked in asecond session to decide in which of the differing samples theodour qualities “sweetish/caramel”, “earthy”, “roasty/sulfury” and“smoky” were detectable with a higher intensity (two-tailed test).The numbers of answers were summed up and the results wereevaluated according to Jellinek [17].

Results and discussion

In a sensory experiment, 10 panelists were asked to eval-uate the differences in the intensities of four odour notes,namely sweet-caramel, earthy, roasty-sulfury and smoky,in either a freshly prepared filtered coffee brew (FC) orin an instant coffee beverage (IC). The results showed(Table 1) that eight panelist clearly judged the odourquality sweetish-caramel to be higher in the instant cof-fee brew. On the other hand, the roasty-sulfury note wasevaluated with a higher intensity in the filtered coffee by7 of the ten panelists. Although also earthy and smokyattributes were rated higher in the FC by some panelists,these results were out of the significance range (Table 1).

AEDA of an extract containing the volatiles isolatedfrom the freshly filtered coffee brew revealed 40 odour-active compounds in the FD-factor range of 32 to 4096.(Table 2). The highest flavour dilution factors werefound for 5-ethyl-3-hydroxy-4-methyl-2(5H)-furanone(abhexone; seasoning-like), 4-vinyl-2-methoxyphenol (phe-nolic), 2-methoxyphenol (phenolic) and (E)-β-damasce-none (boiled apple-like). Seven other compounds, name-ly 4-ethyl-2-methoxyphenol, 4-hydroxy-2,5-dimethyl-3(2H)furanone, methional, 3-mercapto-3-methylbutylformate, 3-hydroxy-4,5-dimethyl-2(5H)furanone, vanil-lin and 2(5)-ethyl-4-hydroxy-5(2)-methyl-3(2H)-furanonewere identified as further contributors to the aroma ofthe freshly prepared coffee brew, thereby confirming ear-lier data by Blank et al. [8].

It should be pointed out that vanillin and the four fu-ranones had not been detected in previous studies report-ed when using GC/O of headspace samples [7, 11], how-

300

Table 1 Number of panelists (out of 10) ranking a certain odourquality to be higher in a filter coffee brew (FC) or in an instantcoffee beverage (IC)a

Odour quality FC IC

sweetish/caramel-like 1 8roasty/sulfurous 7 0earthy 5 0smoky 3 1

a All panelists were able to clearly differentiate the IC from the FCsamples in a triangle test.

ever, these odorants were very recently proven as impor-tant contributors to the aroma of the coffee beverage [9].Thus, the AEDA of solvent extracts/distillates is con-firmed as a necessary further step to reveal the entiregroup of odorants contributing to a food flavour.

In an extract prepared from instant coffee, the three seasoning-like and caramel-like smelling furanones3-hydroxy-5-ethyl-4-methyl-, 3-hydroxy-4,5-dimethyl-2(5H)furanone and 4-hydroxy-2,5-dimethyl-3(2H)fura-none clearly showed the highest FD factors (Table 2).Compared to the FC, these were only slightly lower oreven equal.

Most of the other odorants identified in the FC werealso present in IC. Exceptions are 2-acetyl-2-thiazoline,

3-methylindole, dimethyl trisulfide and 2 unknown com-pounds with clove-like (RI 2364FFAP) or sweet odournotes (RI 1373FFAP). However, all these odorants showedcomparatively lower FD factors. It is, however, interest-ing to note that no further odorant was identified in IC,which was not present in the FC and, also nearly noodorant showed a higher FD factor in the IC compared toFC.

Using GC/O of headspace samples, Pollien et al. [11]had reported dimethyl trisulfide, 2-furfurylthiol, methio-nal and 2-methoxyphenol to be higher or equal in IC.However, this could not be confirmed in our studies, butmay be explained by a different type of instant coffeeused in their studies.

301

Table 2 Most important odorants (FD≥32) in a freshly filtered coffee brew (FC) and an instant coffee beverage (IC)

No. Compounda Odour qualityb RIc FD factor in

DB-FFAP DB-5 FC IC

1 2,3-butanedionee buttery 982 590 32 42 1-octen-3-onee mushroom-like 1290 980 32 43 2-methyl-3-furanthiole meaty 1309 869 32 14 unknown roasty, meaty 1318 – 32 325 dimethyl trisulfidee sulfurous 1350 970 32 <16 unknown sweet 1373 – 32 <17 2-furfurylthiole roasty 1411 911 64 48 3-isopropyl-2-methoxypyrazinee earthy 1420 1097 64 649 3-ethyl-2,5-dimethylpyrazined earthy 1433 1079 32 1

10 methionale cooked potato 1446 906 1024 25611 2-ethyl-3,5-dimethylpyrazined earthy 1444 1080 256 3212 2,3-diethyl-5-methylpyrazinee earthy 1480 1158 256 3213 3-mercapto-3-methylbutyl formatee catty 1507 1020 1024 6414 3-isobutyl-2-methoxypyrazinee earthy 1514 1183 32 1615 2-ethenyl-3,5-dimethylpyrazinee earthy 1545 1109 128 6416 2-ethenyl-3-ethyl-5-methylpyrazinee earthy 1577 1172 256 417 unknown animal-like 1610 – 64 1618 unknown meaty 1615 1022 1024 119 2-/3-methylbutanoic acidd sweaty 1654 – 256 3220 unknown earthy 1717 – 64 121 2-acetyl-2-thiazolinee roasty 1747 1106 32 <122 (E)-β-damascenonee boiled apple-like 1807 1395 ≥4096 25623 unknown smoky 1818 – 32 6424 unknown spicy 1826 – 64 6425 2-methoxyphenold phenolic, burnt 1857 1092 ≥4096 25626 unknown phenolic 1876 – 128 127 unknown spicy 1935 – 256 3228 4-ethyl-2-methoxyphenold phenolic 2020 1283 1024 6429 4-hydroxy-2,5-dimethyl-3(2H)-furanone (furaneol)d caramel-like 2029 1067 1024 102430 2(5)-ethyl-4-hydroxy-5(2)-methyl-3(2H)-furanone (ethylfuraneol)e caramel-like 2064 1145 1024 25631 4-methoxyphenold phenolic 2074 1074 64 132 unknown spicy 2171 – 64 133 3-hydroxy-4,5-dimethyl-2(5H)-furanone (sotolon)e spicy 2191 1127 1024 102434 4-vinyl-2-methoxyphenold clove-like 2200 1320 ≥4096 25635 unknown phenolic 2219 – 1024 25636 5-ethyl-3-hydroxy-4-methyl-2(5H)-furanone (abhexone)d spicy 2260 1196 ≥4096 102437 unknown clove-like 2364 – 128 <138 3-methylindoled mothball-like 2490 1410 32 <139 unknown coconut-like 2561 – 128 6440 vanillind vanilla-like 2581 1425 1024 128

a Compounds which appeared in FC with an FD-factor of at least32 are given.b Odour description assigned during AEDA.c Retention-Index (RI) on the capillaries DB-5 and DB-FFAP.d The compounds were identified by comparing the RI on capillary

DB-5 and DB-FFAP, the odour quality perceived during sniffingand MS data with data of reference compounds.e The compounds were identified by comparing the RI on capillaryDB-5 and DB-FFAP and the odour quality perceived during sniff-ing with data of reference compounds.

With respect to the overall aroma impression evokedby FC and IC beverages, the following conclusions canbe drawn from these results: First, among the sulfur-con-taining components the FD factors of 2-methyl-3-furan-thiol, dimethyl trisulfide, 2-furfurylthiol (FFT), 3-mer-capto-3-methylbutyl formate (MMBF) and 2-acetyl-2-thiazoline were distinctly lower in IC. Therefore, it canbe assumed that the stronger roasty/sulfury note in FCcompared to IC (Table 1) is caused by the higher contentof, in particular, the roasty smelling odorants FFT andMMBF. As very recently demonstrated [18], these thiolscan be irreversibly bound to the coffee melanoidins dur-ing warm-keeping or storage of a coffee brew. Obviouslysimilar reactions take place during manufacturing of theinstant coffee leading to a loss of these odorants contrib-uting to the characteristic flavour of a freshly brewedcoffee.

Second, the intensities of the two 3(2H)-furanones(furaneol and ethylfuraneol), which are mainly responsi-ble for the sweetish/caramel-like quality in filtered cof-fee [7], did not much differ between the two beverages.Because the odour quality elicited by these compoundswas stronger in IC than in FC (Table 1) it can be as-sumed that obviously the thiols, (E)-β-damacenone andother compounds being higher in FC, mask the caramel-sweet odour attribute in the fresh coffee brew. However,when these compounds decrease in their concentration,e.g., in instant coffee, the caramel-like note becomes pre-dominant.

References

1. Holscher W, Vitzthum OG, Steinhart H (1990) Cafe CacaoThe 34:205–212

2. Blank I, Grosch W (1991) J Food Sci 56:63–673. Czerny M, Wagner R, Grosch W (1996) J Agric Food Chem

44:3268–32724. Czerny M, Mayer F, Grosch W (1999) J Agric Food Chem

47:695–6995. Pictet G (1987) In: Coffee. Volume 2, (Clarke RJ, Macrae R,

eds), Elsevier Applied Science, London, UK, pp 221–2566. Semmelroch P, Grosch W (1995) Lebensm Wiss Technol

28:310–3137. Semmelroch, P.; Grosch, W (1996) Studies on character im-

pact odorants of coffee brews. J Agric Food Chem 44:537–5438. Blank I, Sen A, Grosch W (1991) 14th International Confer-

ence on Coffee Science. San Francisco, July 14–19, 1991, pp117–129

9. Mayer F, Czerny M, Grosch W (2000) Eur Food Res Technol211:272–276

10. Clarke RJ (1987) In: Coffee. Volume 2, (Clarke RJ, Macrae R,eds), Elsevier Applied Science, London, UK, pp 35–58 and pp147–149

11. Pollien P, Krebs Y, Chaintreau A (1997) ASIC, 17th Interna-tional Conference on Coffee Science. Nairobi

12. Czerny M, Grosch W (2000) J Agric Food Chem 48:868–87213. Engel W, Bahr W, Schieberle P (1999) Eur Food Res Technol

209:237–24114. Schieberle P (1991) J Agric Food Chem 39:1141–114415. Schieberle P (1995) In: Characterization of Food: Emerging

Methods. (Gaonkar AG, ed), Elsevier Science BV, pp 403–43116. Schieberle P, Hofmann T (1997) J Agric Food Chem

45:227–23217. Jellinek G (1985) Sensory Evaluation of Food; Ellis Horwood:

Chichester, England18. Hofmann T, Czerny M, Calligaris S, Schieberle P (2001) J Ag-

ric Food Chem 49:2382–2386

302