The effect of roasting method on headspace composition of robusta coffee bean aroma

Eur Food Res Technol (2007) 225:9 19 DOI 10.1007/s00217-006-0375-0 ORIGINAL PAPER The effect of roasting method on headspace composition of robusta coffee bean aroma Ewa Nebesny Grażyna Budryn Józef Kula Teresa Majda Received: 10 January 2006 / Revised: 14 April 2006 / Accepted: 11 May 2006 / Published online: 2 August 2006 C -Verlag 2006 Abstract Robusta coffee beans were roasted by three methods, i.e. convectively at 230 C, by microwaves at 700 W, and by the coupled convective microwave (CMR) method (the simultaneous convective heating at 230 C and microwaving at 700 W) for 590, 670, and 370 s, respectively. The ultimate temperature of roasted beans was 238, 207, and 228 C, respectively. Volatile compounds were determined in the headspace by GC-SPME both in samples of roasted coffee and in green beans to find effects of roasting methods on their formation and retention. Eighty-two and 148 odorants were identified in green and roasted coffee, respectively. The highest contents of the latter were found in coffee roasted by the coupled method because both the relatively short time of roasting and moderately high final temperature of beans favored retention of volatile aroma compounds. Because of these reasons, the contents of odorants were the lowest in convectively roasted coffee. Keywords Coffee. Volatile substances. Microwaves E. Nebesny G. Budryn ( ) Institute of Chemical Technology of Food, Faculty of Biotechnology and Food Sciences, Technical University of Lodz, Stefanowskiego 4/10, 91-479 Lodz, Poland e-mail: gbudryn@snack.p.lodz.pl J. Kula T. Majda Institute of Fundamentals of Food Chemistry, Faculty of Biotechnology and Food Sciences, Technical University of Lodz, Stefanowskiego 4/10, 91-479 Lodz, Poland Introduction Only several dozens of almost 1000 volatile compounds, which were hitherto identified in roasted coffee, determine the aroma of beans and brews. They form an exceptionally complex mixture and identification of individual compounds, which provide the pleasant aroma of coffee is extremely difficult. It is obvious that a single compound cannot be regarded as a marker of its intensity [1]. 2-Furfurylthiol, 4-vinylguaiacol, β-damascenone, methanethiol, several alkylpyrazines, furanones, acetaldehyde, propanal, methylpropanal, 2- and 3-methylbutanal rank among substances responsible for the smashing coffee aroma [2, 3]. Selection of good quality coffee beans, roasting technology, and conditions of brews preparing have been improved for ages. Therefore, consumption of coffee infusions is rather a source of pleasant sensory impressions and their stimulating effect became less important (caffeine-free coffee). The highest quality brews are usually derived from arabica variety (Coffea arabica), which is more noble and tasteful than robusta (Coffea canephora). However, cultivation of the latter is easier. Therefore, it is continuously grown despite explicitly worse sensory attributes relative to arabica. There are many publications, particularly patents, presenting new methods of modification of preliminary post-harvest steps of processing of robusta coffee beans and their roasting conditions, leading to improvement of sensory properties of this variety. They include such methods as selection of the appropriate roast degree, preliminary saturation with steam, ultra-fast roasting, roasting in fluidized bed, and roasting using microwaves [4 8]. The latter method is believed to yield the extremely uniformly roasted beans, without burnt surface and with the full roasty aroma. It is particularly recommended for roasting of worse quality robusta coffee beans, which require relatively long time of roasting, leading in case

10 Eur Food Res Technol (2007) 225:9 19 of application of traditional methods to local over-roasting [9]. Our earlier studies, focused on comparison of effects of traditional convective roasting and roasting by microwaves on sensory attributes and selected physicochemical properties of roasted coffee revealed that microwaved coffee displayed better sensory properties [10] and higher antioxidative activity. However, the contents of free radicals were also higher [11, 12]. This work aimed at the analysis of odorants in headspace of coffee beans roasted by microwaving, by the coupled convective microwave method and for comparison by convective heating. Materials and methods Material Green beans of Coffea canephora (robusta), cultivar Kouilou, type Superieur, purchased from the Ivory Coast and derived from coffee cherries by dry method, were delivered by Agros S.A. (Poland). The number of defective beans and impurities in 300 g samples of raw material exceeded 60. Thus, according to the norm ISO 10470:1993 [13], it was higher than the permissible number and therefore the examined coffee beans were qualified to the worst (V) quality class. The concomitant visual and olfactory evaluation, carried out according to the norm ISO 4116:1980 [14] revealed that aroma, form, and color of these coffee beans were typical of robusta variety. Methods Roasting Green coffee beans were roasted in convective microwave Gourmet 8601 oven (Bosch, Germany), providing convective heating at temperatures up to 270 C and microwave heating with gradual regulation of power up to 700 W. Convective and microwave heating can be also conducted simultaneously in this roaster. The sample remains immobile in the oven, which is equipped with mixer of microwaves to provide their homogeneity. Maximum loading of coffee beans to the oven is 1 kg. Samples of green coffee beans (100 g) with an initial humidity of 7.5% were either convectively heated at 250 C or microwaved at 700 W, or heated using simultaneously both these methods. Each sample of beans was spread on a tray in the form of an equalized layer with an initial thickness of 10 mm. The latter gradually rose during roasting because the beans swelled. Roasting was completed when optimal sensory attributes were achieved. Sensory properties were evaluated by QDA method on the basis of certain quality markers of taste, aroma, and texture, described elsewhere [10]. Values of one of these quality markers, i.e. the overall aroma of brews, were 5.0, 5.8, and 4.8 for coffee beans roasted by convective, microwave, and CMR method, respectively. For each of the applied roasting methods, the optimum sensory attributes corresponded to a drop in solid substance of approximately 9.5% which was equivalent to the medium roast. Roasting was carried out for 590, 670, and 370 s, respectively, and the ultimate temperatures of roasted coffee beans were 238, 207, and 228 C, respectively (measured by a thermocouple Therm-2285-2B, Poland). The temperature of beans was measured on the surface of coffee beans immediately on completion of roasting [33]. Humidity of coffee beans roasted by the examined methods was 2.0, 1.3, and 2.4%, respectively. The color of roasted coffee beans was determined by CIE L a b method using a Specord M-40 spectrophotometer (Carl Zeiss, Germany). Values of L were 42.4, 41.0, and 39.0, values of a were 8.4, 9.1, and 10.9, and values of b were 25.8, 26.9, and 29.0 for convective, microwave and the coupled method, respectively. Headspace aroma isolation Coffee beans were ground in a laboratory mill WZ-1 (ZBPP, Poland) and sieved through 400 µm 400 µm sieve. Samples of ground coffee (5 g) were placed in a cylindrical vial (5 cm 10 cm). The latter was spiked with a syringe (Supelco, USA) equipped with 2 cm 5 /30 µm DVB/Cartoxem/PDMS fibers to achieve condensation of volatile compounds on the fibers, leading to their solid phase microextraction (SPME) from headspace. Time of achievement of equilibrium was 30 min at 60 C. This temperature was chosen because it provided more efficient extraction of less volatile compounds than at room temperature, at which estimation of aroma of roasted coffee is usually conducted. Furthermore, it may allow further comparison of the headspace of roasted coffee beans with that of brews, for which the temperature of contact with olfactory receptors is close to 60 C. Headspace analysis was also carried out at 20 C (results not presented in this publication). Evaluation of roasted coffee aroma is frequently conducted at this temperature as imply numerous published data. Concentrations of volatile compounds in the headspace considerably increase with a rise in desorption temperature as showed studies of Sanz et al. [15]. But at elevated temperature, the composition of headspace is markedly different from that achieved at room temperature (the difference increases with a rise in temperature), and some of headspace components undergo degradation or oxidation. Therefore, temperatures exceeding 60 C were not applied for presented studies. Time of approaching the equilibrium between the headspace and roasted coffee beans was selected on the basis of already mentioned studies of Sanz et al. [15]. They found small

Eur Food Res Technol (2007) 225:9 19 11 differences in percentage content of individual groups of volatile compounds caused by increasing this time from 30 to 150 min. Additionally, certain compounds, e.g. that containing sulfur, which are very labile, can be oxidized during longer incubation. All analyses were carried out under constant and reproducible conditions (the loading of ground coffee inside the vial, maintenance of temperature, position of the fibers in the syringe etc.). GC analysis Analysis was carried out using GC Carlo-Erba Instruments HRGC-530 Mega Series gas chromatograph (Italy), equipped with flame ionization detector (FID) and splitless sampler (SSL). Time of desorption of compounds in a sampler was 10 min at 250 C. Compounds were separated on Quandrex (30 m 0.32 mm) capillary column lined with FFAP stationary phase (0.5 µm). Temperature of desorption was increased from 35 to 320 C, with 5 min isotherm at 35 C and 45 min isotherm between the examined samples. Temperature gradient was 4 C/min. Temperature of sampler and detector was 250 C. The rate of carrier gas flow (N 2 )was1cm 3 /min. Percentage contents of individual compounds were determined on the basis of surface area under all peaks in chromatograms. The total surface area of all peaks was the quantity equivalent to the overall amount of volatile headspace components emitted from examined coffee samples. Volatile compounds were identified on the basis of their spectra determined by gas chromatography/mass spectrometry GC/MS (coupled with the NIST mass spectral database) and comparison of their GC Kovats index (KI) and retention time (RT) with literature data. Statistical analysis Analyses were carried out in triplicate and their results were subjected to statistical analysis, which included determination of average surface area of each peak (Table 1) and standard deviation of the latter as well as monodirectional analysis of variation (ANOVA) at the significance level p 0.05. Results and discussion Roasting was terminated when optimum sensory properties (determined according to description in Methods section) were approached. The latter corresponded to the same decrease in a solid substance content for three applied roasting methods. Thus, the optimum roast level was the same in all cases. The time of convective and microwave heating was similar (slightly longer for the latter method, i.e. 590 and 670 s, respectively) but application of microwaves significantly reduced the ultimate temperature of coffee beans, from 238 C for convective heating to 207 C for microwaving. It results from intensive heating of all parts of beans by microwaves versus heating of their surface in convective method. Besides, the microwaves can cause changes in chemical reactions occurring during roasting. For instance, degradation of carbohydrates and chlorogenic acids can be more intensive [16, 17]. The simultaneous application of both heating methods (unaltered parameters) resulted in intensification of heating and chemical reactions what significantly reduced the time of roasting to 370 s. Also the ultimate temperature of roasted coffee (228 C) was lower relative to convective heating. Green coffee beans have a corny, slightly nutty note, which during roasting changes into the characteristic, roasty, intensive, and pleasant flavor. Usually, concentration of volatile compounds increases 50-fold during roasting, 1 kg of roasted coffee contains 5 g of volatile compounds and concentrations of components of coffee aroma do not exceed ppm level [18]. We tried to compare composition of volatile fraction of coffee roasted by three different methods, i.e. convective, microwave, and coupled convective microwave. Headspace analysis was also conducted for green coffee. A rise in total volatile compounds contents (caused by roasting) ranged from 6.5 to 9.2 (calculated on the basis of the total surface area below the peaks) (Fig. 1). It was lower than that reported by Clarke, who probably used other analytical methods, leading to more effective extraction of volatile substances. Mayer and Grosch [19], who determined the degree of releasing of different groups of odorants from roasted coffee to the headspace, assayed the contents of 22 volatile compounds in the latter. These substances were emitted for 30 min at 80 C (the level of their desorption was 12 65%) and the authors did not employ SPME to enrich this phase on the fibers. Thirty-one furans, 25 pyrazines, 31 carbonyl compounds, 18 benzene derivatives, 11 sulfur-containing compounds, 5 pyridines, 6 pyrrols, 11 alcohols, 3 hydrocarbons, 4 esters and single pyranes, oxazols and lactones were detected in the examined coffee headspaces (Table 1). The majority of these substances were detected earlier by other researchers who analyzed coffee aroma [15, 20 23]. All roasting methods employed for our studies yielded the same headspace components and the chromatograms differed only in the surface area below the peaks and percentage contents of individual peaks in the total surface area. However, odorant profiles of foods containing the same volatile components can be completely different because of different proportions of the latter [24]. Furans constituted the most abundant group with respect of a number of identified compounds and their participation in the coffee headspace. They accounted for only 2% of green coffee s headspace (6 derivatives), but the headspace of

12 Eur Food Res Technol (2007) 225:9 19 Table 1 Volatile components of headspace of green (G) and roasted robusta coffee beans; convectively roasted (CR), microwaved (MR), and roasted by the coupled convective microwave method (CMR) Surface area of GC peak (%) Peak no. Compound KI ID RT G CR MR CMR 1 1,3-Pentadiene 505 A 2.14 nd 0.51 0.35 0.20 2 Methanethiol 511 A 2.24 0.22 1.10 a 0.68 1.15 a 3 Ethanol 543 A 2.47 0.29 0.05 a,b 0.07 a 0.03 b 4 Dimethylsulphide 562 A 2.60 0.03 0.41 a 0.33 b 0.35 a,b 5 Carbonoxidesulphide 573 B 2.69 nd 0.05 a,b 0.07 a 0.03 b 6 Propanal 598 A 2.92 0.07 0.10 0.17 a 0.14 a 7 2-Methylpropanal 629 A 3.98 0.24 tr tr tr 8 Propanone 648 A 3.00 0.63 0.43 0.58 a 0.52 a 9 N,N,N -trimethylhydrazine 684 B 3.09 nd 1.02 1.85 0.79 10 3-Methylfuran 748 A 3.60 nd tr tr tr 11 Butanal 750 A 3.63 nd tr tr tr 12 2-Methylfuran 792 A 4.27 nd 0.02 a 0.03 a 0.02 a 13 2-Buten-1-ol 805 B 4.38 nd 0.40 a 0.35 0.30 a 14 Ethylacetate 807 A 4.43 0.04 nd nd nd 15 2-Butanone 809 A 4.45 0.10 0.58 a 0.54 a 0.59 a 16 2-Methyl-2-buten-1-ol 821 B 4.62 0.94 0.69 1.06 a 1.29 a 17 3-Methyl-3-buten-2-ol 830 A 4.74 1.14 0.31 0.44 0.68 18 2,4-Dimethyltetrahydrofuran 865 B 5.18 0.17 nd nd nd 19 2,3-Dihydro-4-methylfuran 869 B 5.24 nd tr 0.02 a 0.03 a 20 Ethanol 883 A 5.42 0.11 0.03 a 0.05 a 0.04 a 21 2,5-Dimethylfuran 909 A 5.74 nd 0.05 0.11 a 0.09 a 22 2,4-Dimethylfuran 945 B 6.17 nd 0.01 a 0.01 a 0.02 a 23 Pentanal 953 B 6.27 0.04 nd nd nd 24 2,2-Dimethylpropanal 974 B 6.52 nd 0.06 0.12 a 0.09 a 25 2,3-Butanedion 999 A 6.82 nd 0.27 a 0.35 a 0.46 26 2-Pentanone 1001 B 6.85 0.17 nd nd nd 27 α-pinene 1017 B 7.50 0.39 nd nd nd 28 Tiophene 1024 A 7.77 nd 0.05 a 0.04 a 0.03 a 29 2-Ethyl-5-methylfuran 1034 B 8.17 nd 0.02 a 0.03 a 0.03 a 30 Toluene 1047 A 8.67 0.67 0.03 a 0.04 a 0.03 a 31 2-Butenal 1056 A 8.99 0.21 0.05 0.10 a 0.12 a 32 3-Hexanone 1060 A 9.22 nd 0.04 a 0.06 a 0.04 a 33 2,3,5-Trimethylfuran 1065 A 9.42 nd tr tr 0.01 34 3-Methylbutanoic acid ethylester 1074 B 9.83 0.47 nd nd nd 35 2,3-Pentadiene 1077 B 9.89 0.22 0.05 0.12 a 0.17 a 36 Dimethyldisulphide 1086 A 10.12 nd 0.01 a 0.02 a 0.02 a 37 Phenol 1090 A 10.37 nd 0.06 0.11 a 0.10 a 38 Hexanal 1092 A 10.49 4.33 0.02 0.05 a 0.05 a 39 2-Methylthiophene 1103 A 10.92 nd 0.02 a 0.03 a 0.03 a 40 3-Methyl-2-butenal 1107 A 11.07 nd 0.04 a 0.07 b 0.05 a,b 41 Xylene 1116 A 11.45 0.35 nd nd nd 42 2-Butenoic acid methyl ester 1121 A 11.65 nd 0.01 a 0.03 0.01 a 43 4-Hydroxy-2-butanone acetate 1127 B 11.90 0.64 nd nd nd 44 3-Pentene-2-one 1141 A 12.49 nd tr 0.03 0.06 45 3-Methyl-3-hexanone 1148 B 12.75 nd 0.11 a 0.13 a 0.20 46 1-Methylpyrrole 1157 A 13.12 nd 0.88 0.44 a 0.48 a 47 3,4-Hexanedione 1164 A 13.44 0.33 0.01 a 0.02 a 0.01 a 48 3-Methylphenol 1168 B 13.59 nd 0.17 0.10 a 0.11 a 49 β-myrcene 1192 B 14.58 1.95 nd nd nd 50 Pyridine 1199 A 14.87 3.47 7.57 a 8.49 a 5.32 51 2,4,5-Trimethyloxazol 1209 A 15.27 nd 0.05 a 0.03 a 0.04 a 52 3-Methyl-1-butanol 1224 A 15.88 2.95 nd nd nd 53 2-(2-Propenyl)-furan 1226 A 15.95 nd 0.08 0.23 0.16 54 1,2-Dimethylpyrrole 1229 A 16.07 nd 0.03 tr 0.07 55 Pyrazine 1232 A 16.20 nd 0.76 a 0.89 a 0.88 a 56 2-Methyl-1-butanol 1237 A 16.38 1.83 nd nd nd

Eur Food Res Technol (2007) 225:9 19 13 Table 1 Continued Surface area of GC peak (%) Peak no. Compound KI ID RT G CR MR CMR 57 2-Methylpyridine 1243 A 16.65 nd 0.02 a 0.03 a 0.02 a 58 2-Pentylfuran 1249 A 16.90 0.20 0.02 a 0.03 a 0.03 a 59 Furfurylmethylester 1260 A 17.34 nd 0.16 a 0.21 a 0.11 60 Trans-ocimen 1263 B 17.43 0.20 nd nd nd 61 Styrene 1268 A 17.65 1.18 0.04 a 0.02 a,b 0.01 b 62 Tetrahydro-3-methylfuran 1270 B 17.74 nd 0.15 a 0.19 a 0.08 63 1-Pentanol 1272 A 17.85 0.21 nd nd nd 64 Pyrazinamid 1275 B 17.87 nd 0.20 a 0.16 a 0.08 65 2-Methylpyrazine 1285 A 18.34 3.65 4.49 a 6.19 4.94 a 66 3-Octanone 1287 B 18.43 0.94 nd nd nd 67 Mesitylene 1290 B 18.53 0.51 nd nd nd 68 2,5-Dimethylpyrrole 1292 A 18.54 nd 0.05 a 0.03 a 0.04 a 69 4-Methylthiazole 1297 A 18.82 nd 0.06 0.15 0.01 70 3-Hydroxy-2-butanone 1299 A 18.92 0.61 0.01 0.14 0.04 71 5-Methylpyridine 1306 B 19.17 nd 0.54 0.02 0.06 72 2-Heptenol 1308 A 19.25 0.68 nd nd nd 73 1,2-Ethanodiol 1310 B 19.35 nd 0.48 a 0.35 0.58 a 74 2-Pyrrolidinemethanol 1312 B 19.42 0.13 nd nd nd 75 2,5-Dimethylpirazine 1330 A 20.12 nd 0.63 a 0.75 a 0.97 76 3-Methyl-2-heksanol 1331 B 20.18 2.05 nd nd nd 77 2,6-Dimethylpyrazine 1339 A 20.47 2.18 1.74 a 2.11 1.63 a 78 2-Ethylpyrazine 1345 A 20.70 1.82 1.87 a 2.03 a 1.75 a 79 1-Allyl-2-pyrazine 1348 B 20.83 0.98 nd nd nd 80 2-Hydroxy-3-pentanone 1351 A 20.94 nd 1.02 a 1.17 a 0.98 a 81 n-hexyl formate 1354 B 21.08 0.20 nd nd nd 82 2,3-Dimethylpyrazine 1363 A 21.40 0.88 a 0.87 a 0.71 a 0.68 a 83 Hexanol 1365 B 21.52 0.45 nd nd nd 84 2-Cyclopenten-1-one 1372 A 21.60 0.16 0.09 a 0.10 a 0.10 a 85 2-Ethylxylene 1378 B 21.78 0.11 nd nd nd 86 2-Methyl-3-hexanone 1380 B 22.00 nd 0.20 a 0.19 a 0.20 a 87 2-Methyl-2-cyclopenten-1-one 1381 A 22.14 nd 0.07 0.10 a 0.11 a 88 1-Hydroxy-2-butanone 1388 A 22.40 nd 0.13 a 0.15 a 0.13 a 89 2-Ethyl-6-methylpyrazine 1398 A 22.80 0.58 1.53 a 1.50 a 1.13 90 2-Ethyl-5-methylpyrazine 1405 A 23.04 0.98 0.76 a 0.78 a 0.54 91 2-Amino-4-methylthiazole 1414 B 23.39 nd 0.07 a 0.30 0.08 a 92 2-Ethyl-3-methylpyrazine 1418 A 23.52 0.96 1.48 1.15 a 1.07 a 93 Ethoxybenzonitrile 1423 B 23.68 0.02 nd nd nd 94 E-2-nonenal 1428 A 23.87 0.09 nd nd nd 95 5-Methyl-2(5H)-furanone 1433 A 24.05 nd 0.19 a 0.15 a 0.16 a 96 2,4-(1H,3H)-Pyramidinedion 1439 B 24.29 nd 0.02 a 0.07 0.01 a 97 2,5-Diethylpyrazine 1443 A 24.44 0.20 0.04 a 0.07 b 0.05,B 98 Vinylpyrazine 1447 A 24.55 0.23 0.19 a 0.20 a 0.08 99 1-Octen-3-ol 1450 B 24.67 0.10 nd nd nd 100 2-Methyl-5-propylpyrazine 1458 B 24.97 0.88 a 1.29 b 1.21 b 0.76 a 101 2,2-Dimethylhexanal 1469 B 25.35 0.54 nd nd nd 102 2,6-Diethylpyrazine 1475 A 25.57 nd 0.51 a 0.49 a 0.24 103 Acetic acid 1478 A 25.67 7.41 nd nd nd 104 2-Methyl-3-propylpyrazine 1479 B 25.72 nd 0.06 a 0.08 a 0.02 105 2-Furaldehyde 1483 A 25.85 nd 1.10 a 1.37 a 1.93 106 1-(1-Propenylthio)-propane 1494 B 26.25 nd 1.85 a 2.29 a 3.32 107 2-Furfuryl metyl sulphide 1506 A 26.67 nd 0.28 a 0.30 a 0.05 108 2-Methyl-3,5-diethylpyrazine 1509 A 26.77 nd 0.32 a 0.36 a 0.18 109 2-Methyl-5-vinylpyrazine 1516 A 27.00 nd 0.23 a 0.27 a 0.13 110 Furfural formate 1524 A 27.30 nd 0.52 a 0.51 0,60a 111 1H-pyrrole 1534 A 27.64 nd 0.49 a 0.61 a 0.48 a 112 3,5-Octadien-2-one 1536 B 27.70 0.57 nd nd nd

14 Eur Food Res Technol (2007) 225:9 19 Table 1 Continued Surface area of GC peak (%) Peak no. Compound KI ID RT G CR MR CMR 113 Benzaldehyde 1546 A 28.04 1.80 0.93 a 0.95 a 0.85 a 114 2,3-Dimethyl-2-cyclopenten -1-one 1550 A 28.20 nd 0.18 a 0.16 a 0.14 a 115 1-Acetyloxy-2-butanone 1553 A 28.27 nd 0.25 a 0.25 a 0.33 116 2-Furanmethanol acetate 1557 A 28.40 0.15 0.29 0.51 0.13 117 1,2-Butandiol 1563 B 28.62 6.98 nd nd nd 118 Linalool 1568 B 28.75 0.31 nd nd nd 119 2-Acetyl-1-methylpyrrole 1574 A 28.99 nd 0.14 a,b 0.11 a 0.16 b 120 5-Methyl-2-furaldehyde 1605 A 30.04 0.17 1.33 a 1.49 a,b 1.74 b 121 Isopropenylpyrazine 1620 B 30.52 nd 0.52 a 0.49 a 0.46 a 122 Furfuryl alkohol propanoate 1626 A 30.70 nd 0.09 a 0.09 a 0.08 a 123 Dipropanoate-1,2-ethandiol 1630 B 30.82 nd 0.13 a 0.13 a 0.12 a 124 2-Amino-5-methylphenol 1634 B 30.95 nd 0.33 a 0.38 a 0.31 a 125 2,2 -Methylidenedifuran 1637 B 31.07 nd 0.35 a 0.29 a 0.29 a 126 5-Methyl-6,7-dihydro-(5H)- 1643 A 31.24 nd 0.30 a 0.27 a 0.33 a cyklopentapyrazine 127 Dihydro-3-methyl-2(3H)-furanone 1646 B 31.35 0.48 nd nd nd 128 2-Isoamyl-6-methylpyrazine 1648 B 31.39 nd 0.13 a 0.26 0.11 a 129 2-Acetyl-5-methylfuran 1653 A 31.55 nd 0.60 a 0.52 a,b 0.43 b 130 2-Acetylpyrazine 1662 A 31.87 nd 0.19 a 0.18 a 0.17 a 131 γ -Butyrelakton 1671 A 32.15 2.66 4.38 a 3.64 a,b 3.37 b 132 1,2,4-Triazolo-1,5-A-pyrazine 1675 B 32.29 nd 0.02 a 0.12 0.03 a 133 Acetophenone 1680 B 32.42 0.39 nd nd nd 134 2,3-Dimethyl-6-isobutylpyrazine 1683 B 32.54 nd 0.40 a 0.37 a 0.38 a 135 1-Acetyl-1-methylpyrrole 1686 B 32.64 nd 0.17 a 0.14 a 0.15 a 136 2-Furanmethanol 1694 A 32.89 1.02 10.96 a 10.21 a 16.37 137 5-Methyl-2,2 -methylidenedifuran 1700 B 33.07 nd 1.41 a 1.28 a,b 0.90 b 138 2,6-Dimethyl-p-benzoquinone 1711 B 33.39 nd 0.74 a 0.74 a 0.56 140 2-Pyrrolidinone 1714 B 33.42 0.17 nd nd nd 141 2-Acetyl-3-methylpyrazine 1721 B 33.69 nd 0.75 a 0.72 a 0.55 142 p-ment-8-en-3-one 1728 B 33.89 nd 0.17 a 0.21 a 0.09 143 2-Allyl-3-methylpyrazine 1739 B 34.24 nd 0.75 0.56 0.31 144 4-(5-Methyl-2-furanyl)-2-butanone 1745 B 34.40 nd 0.12 a 0.14 a 0.06 145 3-Methyl-2(5H)-furanone 1750 B 34.54 nd 0.07 a 0.06 a 0,02 146 Pyrazine carboxamide 1755 B 34.69 nd 0.59 a 0.56 a 0.47 a 147 1-Methylene-1H-indene 1763 B 34.92 1.00 nd nd nd 148 3,4-Dimethyl-2,5-furandione 1764 B 34.95 nd 0.64 a 0.55 a 0.27 149 3-Octen-2-ol 1769 B 35.12 nd 0.14 a 0.17 a 0.07 150 3-Ethyl-4-methyl-2,5-furandione 1777 B 35.34 nd 0.28 a 0.31 a 0.26 a 151 1,2-Dimetoxybenzene 1780 B 35.43 0.22 nd nd nd 152 Methyl salicylate 1801 B 35.89 1.03 0.46 0.16 a 0.12 a 153 4-Pyridinecarboxylic acid methyl ester 1804 B 36.15 nd 0.19 a 0.20 a 0.13 154 2-Hydroxy-3-methyl-2-cyclopenten-1-1829 B 36.92 nd 1.19 a 1.08 a 1.52 one 155 Ethyl salicylate 1832 B 37.00 2.05 0.61 a 0.53 a 0.49 a 156 3-Methyl-2-butenoic acid 1839 B 37.12 0.70 nd nd nd 157 3-Octen-2-one 1851 B 37.57 nd 0.44 a 0.43 a 0.21 158 1-(2-Furanylmethyl)-1H-pyrrole 1858 B 37.79 nd 0.21 a 0.22 a 0.35 159 1-Ethylidene-1H-indene 1867 B 38.07 0.61 nd nd nd 160 3-Phenylfuran 1880 B 38.49 nd 0.13 a 0.15 a 0.09 161 2-Methoxyphenol 1892 B 38.85 0.02 2.70 1.94 a 1.68 a 162 3-Acetoxy-4-methylpyridine 1897 B 39.00 nd 0.14 0.08 0.24 163 α-ethylbenzyl alcohol 1908 B 39.30 0.35 0.11 a 0.11 a 0.18 164 1-(2-Phenyl)-1-propanol 1920 B 39.64 nd 0.11 a,b 0.08 a 0.13 b 165 2-Hydroxy-2,3-dimethyl-2-cyclopenten- 1926 B 39.82 nd 0.45 a 0.31 0.49 a 1-one 166 Phenylethyl alcohol 1937 B 40.12 1.66 nd nd nd

Eur Food Res Technol (2007) 225:9 19 15 Table 1 Continued Surface area of GC peak (%) Peak no. Compound KI ID RT G CR MR CMR nd: not detected; tr: traces. KI: Kovats index; ID: identification method; A: mass spectrum consistent with the NIST mass spectral database; B: mass spectrum and Kovats index consistent with literature data. Values in each line bearing the same letters are not significantly different (p>0.05) from one another 167 Furfurylmethyl disulphide 1938 B 40.14 nd 0.13 a 0.06 0.12 a 168 Trans-bicyclodecane 1963 B 40.84 nd 0.30 0.05 0.10 169 Benzthiazole 1984 B 41.43 0.26 nd nd nd 170 Maltol 2006 B 42.02 nd 0.83 a 0.94 a 0.94 a 171 3-Formyl-1-methylpyrrole 2012 B 42.19 0.25 1.25 a 1.11 a 1.17 a 172 2,2 -(Oxydimethylene)-difuran 2028 B 42.60 nd 0.43 a 0.29 b 0.37 a,b 173 Benzimidazole 2039 B 42.90 nd 0.38 a 0.34 a 0.37 a 174 p-hydroxybenzenesulphonic acid 2048 B 43.12 nd 1.00 0.68 a 0.64 a 176 Phenyl carbamate 2064 B 43.34 nd 0.23 a 0.17 b 0.20 a,b 177 p-ethyl-2-methoxyphenol 2065 B 43.55 nd 2.28 a 1.55 2.04 a 178 2,5-Dimethyl-4(1H)-hydroxy-3(2H)- 2071 B 43.74 nd 0.41 a 0.31 0.51 a furanone 179 2,3-Dimethoxytoluene 2078 B 43.90 0.05 nd nd nd 180 2-Methyl-4-quinazolinone 2085 B 44.09 nd 0.33 a 0.23 0.35 a 181 4-Methyl-2(1H)-quinolinone 2114 B 44.85 nd 0.21 a,b 0.18 a 0.26 b 182 3,6-Dimethyl-4H-pyrid-[1,2-a]- 2144 B 45.62 nd 0.08 0.13 a 0.14 a pyrimidin-4-one 183 4-Methylpyrrol-1,2-A-pyrazine 2167 B 46.22 nd 0.03 a 0.03 a 0.07 184 4-Acetoxy-3-methoxystyrene 2235 B 47.97 nd 3.69 2.87 5.51 185 1-(2-Furanyl)-2-hydroxyethanone 2256 B 48.49 nd 0.03 a 0.01 a 0.03 a 186 2,4-Decadienal 2269 A 48.82 nd 0.02 0.08 0.26 187 4,4,5,5-Tetramethyl-2,7-octandione 2289 B 49.32 nd 0.11 a 0.09 a 0.13 a 188 5-Methyl-5-furfuryl-2(5H)-furanone 2311 B 49.84 nd 0.07 a 0.07 a 0.09 a 189 Benzoic acid 2465 B 53.52 0.37 nd nd nd 190 Indole 2476 B 53.77 nd 0.29 a 0.20 a 0.41 191 Bis-2-furfuryldisulphide 2520 B 54.84 nd tr 0.01 a 0.01 a 192 n-hexadecanoic acid 2971 B 66.35 0.85 nd nd nd 193 Caffeine 3163 B 82.94 2.01 0.23 0.30 a 0.37 a roasted coffee contained 31 derivatives, which accounted for 21% (convective heating), 20% (microwave roasting), and 26% (CMR method) of all odorants. According to Clarck [18] furans made up from 38 to 45% of total volatile compounds of roasted coffee so because of polar structure, reducing the volatility, their content in the headspace was lower. Furans are typical ingredients of roasted coffee. They impart the caramel, burnt aroma. Concentration of 2-furanmethanol was the highest, particularly in the headspace of coffee derived by CMR method(p<0.05, Table 1, no. 136). High contents of 2-furaldehyde acetate, 2-furanmethanol, and 5- methyl-2-furaldehyde were also observed. The characteristic furan odorant of coffee aroma is 2,5-dimethyl-4-hydroxy- 3(2H)-furanone, which has sweet, caramel note and low threshold level (1 ng/dm 3 air). Its lowest concentration was detected in coffee roasted by microwaving (p<0.05, Table 1, no. 178). The total content of furans was considerably affected by concentration of 2-furanmethanol, which has a relatively high boiling temperature. Short time of roasting resulted in accumulation of larger amounts of this compound in coffee roasted by the coupled method what increased the content of furans in the respective headspace. Pyrazines (12 identified derivatives) made up 12% of green coffee aroma. Hence, they are typical components of the latter. Their content in roasted coffee achieved 20% (convective roasting), 22% (microwave roasting), and 17% (CMR method), and the number of pyrazine derivatives rose to 25. According to Clarck [18], they constitute up to 25 30% of roasted coffee aroma what exceeds their content in headspace, like in case of furans. 2-Methylpyrazine was the dominating ingredient of this mixture, particularly in coffee roasted by microwaving (p<0.05, Table 1, no. 65). These results are consistent with that of Hashim et al. [23], who estimated the dependence of the content of this compound in roasted coffee on roasting temperature and found that the optimum temperature was 205 C. He noticed that this substance was not only intensively synthesized but also faster emitted at higher roasting temperatures. Methylpyrazine has a note of burnt grass but because of a relatively high odor threshold its significance for the overall coffee aroma is rather limited. Other derivatives found in larger amounts in the headspace of green coffee beans were dimethylpyrazines (2,3-, 2,5-, and 2,6-isomers) that are characterized by aroma of tanned skin, raw nuts, and maize, respectively. Their odor thresholds are lower as compared to methylpyrazine, which is formed by reaction of demethylation during thermal processing of dimethyl derivatives. According to Hashim [23], the aroma of over-roasted coffee results from an increase

16 Eur Food Res Technol (2007) 225:9 19 Fig. 1 The dependence of the overall surface area under peaks of headspace components of roasted robusta coffee beans separated by gas chromatography on roasting method. G, green coffee; CR, convectively roasted; MR, roasted by microwaves, CMR; roasted by the coupled convective microwave method sum of volatile compounds (total surface area of peaks) x 10 4 3000 2500 2000 1500 1000 500 0 G CR MR CMR roasting method in a ratio of methylpyrazine to dimethylpyrazines. This ratio was the most acceptable for convective roasting (p<0.05, Table 1, nos. 65, 75, 77 and 82). Ethylmethyl-derivatives such as 2-ethyl-3-, 2-ethyl-5-, and 2-ethyl-6-methylpyrazine also explicitly affect aroma of coffee. Their contents were the lowest in coffee roasted by the CMR method (Table 1, nos. 88, 90 and 91). They are ingredients of many roasted seeds (e.g. sesame [25]) and they are known to display smashing taste and aroma. Ethylalkilpyrazines have a note of soil and mushrooms, and are very active components of aroma because they have very low odor thresholds (roughly 0.01 ng/dm 3 air) [26]. According to Blank et al. [27], 5-methyl-6,7-dihydro- (5H)-cyclopentapyrazine is a pyrazine derivative, which affects coffee aroma. Its percentage contents in the examined samples of roasted coffee were similar (p>0.05, Table 1, no. 126). Pyrazines are medium-volatile compounds. Both the time of roasting and temperature of coffee beans affect their evaporation. Therefore, reducing of the latter in microwave roasting increased their concentration as compared to the other examined roasting methods. Both green and roasted coffee contained numerous carbonyl compounds. Eighteen of them were identified in the headspace of green coffee (9% of all volatile ingredients), whereas the headspace of roasted coffee contained 31 of them (they accounted for 4, 6, and 7% of headspace for samples roasted by CMR, microwave, and convective method, respectively). Roasted coffee usually contained other carbonyl derivatives than green beans, in which their presence results from self-oxidation of fatty acids [28]. According to Clark [18], participation of carbonyl compounds in total volatile substances of roasted coffee is lower (approximately 3%). Thus, the majority of them migrate to the headspace where their concentration is higher. Roasting leads to generation of such substances as Strecker aldehydes and α-diketones. Carbonyl compounds are important ingredients of coffee aroma. Some of them are very volatile. Carbonyl derivatives have diverse notes, e.g. pleasant scent of flowers or butter-like, rancid and rotten stench. Important components of coffee aroma are α-diketones, including 2,3-butanedion, with butter-like note, very characteristic of coffee. It is an ingredient of synthetic aroma of coffee [3]. Its odor threshold is very low due to relatively high volatility. Grinding of coffee halves its content in 15 min [19]. The highest concentration of this compound was found in headspace of coffee roasted by CMR method (p<0.05, Table 1, no. 25). Low-molecularweight aldehydes, i.e. ethanal, propanal, 2-methylpropanal, and butanal (Table 1, nos. 3, 6, 7 and 11) are important carbonyl ingredients of coffee. They are very volatile substances, which convey the typical of coffee, malt-like sensation. Their fast evaporation after grinding of coffee brings about the perceptible change in aroma, leading to the weaker malt note [3]. Headspace of roasted coffee, particularly obtained by CMR method, contained high amounts of cyclic ketones, such as 2-hydroxy-3-methyl-2-cyclopenten-1-one and 2-hydroxy-2,3-dimethyl-2-cyclopenten-1-one (p<0.05, Table 1, nos. 154 and 165). These substances, characterized by sweet and caramel-like hint, are also of importance for coffee aroma. The group of hydroxy ketones was also rich in 2-hydroxy-3-pentanone (similar contents in examined samples, p>0.05, Table 1, no. 80). Green coffee contained large amounts of hexanal with a note of fresh green plants (p<0.05, Table 1, no. 38). The similar concentration of this substance in green coffee was detected by Czerny et al. [3]. Another compound typical of green coffee is E-2-nonenal,

Eur Food Res Technol (2007) 225:9 19 17 with cucumber-like flavor. We found it only in green coffee (Table 1, no. 94), although other authors detected it also in the roasted material [29]. High contents of carbonyl derivatives in convectively heated coffee imply that roasting conditions have a strong impact on both retention and formation of these substances. During the roasting process, oxidation of wax deposited on the surface of coffee beans and lipids leached from their core (what leads to generation of numerous carbonyl compounds) is promoted by the relatively long time of roasting and intensive heating of the surface in convective method. Other important ingredients of coffee aroma are benzene derivatives, including phenols. They were detected in green (17 derivatives accounting for 11% of headspace) and in roasted coffee (18 derivatives making up 10, 11, and 8% of headspace for convective, microwave and CMR method, respectively). Their participation in volatile fraction of roasted coffee is lower than in the headspace, like in case of carbonyl compounds (only about 5%; [18]). One of the sources of benzene derivatives detected in roasted coffee aroma is thermal degradation of chlorogenic acids [30]. Their aroma is diverse, i.e. spicy, clove-like or smoky. Clarke [18] reported that they were typical ingredients of coffee aroma, unique among roasted plant materials. Therefore, aroma of substitutes of roasted coffee is explicitly different from aroma of natural coffee. The most characteristic phenol derivatives are 2-metoxyphenol (guaiacol) and 4-ethyl-2- methoxyphenol (p-ethylguaiacol). They were found in highest amounts in coffee roasted by convective method (p<0.05, Table 1, nos. 161 and 177). Other benzene derivatives, abundant in headspace of roasted coffee were benzaldehyde, 2,6-dimethyl-p-benzoquinone, methyl and ethyl salicylates, and p-hydroxybenzenesulfonic acid (Table 1, nos. 113, 138, 152, 155 and 174). Derivatives of benzene, particularly those abundant in examined samples, have relatively high boiling temperatures and presumably therefore long time of roasting and relatively high temperature of this process had a weaker impact on their evaporation. Long time of roasting favors their formation and therefore samples of coffee roasted by convective and microwave methods are rich in benzene derivatives (particularly the latter). Sulfur-containing compounds are also important components of coffee aroma because of low boiling temperatures and odor threshold levels. Green coffee contained only 2 of these substances (less than 1% of headspace), while 11 of them were identified in roasted coffee (about 2% of headspace for all roasting variants). According to Clarke [18], they make up only 1% of volatile fraction of roasted coffee (i.e. less than in its headspace) owing to relatively low boiling temperatures. The most characteristic of them are methanethiol, dimethylsulfide, methyldisulfide, thiophen, and methylthiophen, which display sulfur note, from rotten sensation to garlic-like and onion-like flavors. Thiophens have aroma of boiled meat and turnip. Sulfur-containing compounds are unique ingredients of coffee, absent in other beverages. They have exceptionally low threshold levels and boiling temperatures. Their contents in all examined samples were similar, but the lowest amounts of methanethiol and dimethylsulfide were found in microwaved coffee. Thus, longer roasting time can cause their more pronounced evaporation (p<0.05, Table 1, nos. 2, 4, 28, 36 and 39). Some of sulfur-containing compounds, particularly methylthiazols, are responsible for unpleasant aroma, resembling that of burnt rubber [6]. Headspace of roasted coffee contained 4-methylthiazol and 2-amino-4- methylthiazol. Particularly rich in them was the microwaved coffee (p<0.05, Table 1, nos. 69 and 91). Methylfurfuryl sulfide and disulphide rank among compounds with the characteristic roasty note. The lowest amounts of the former and the latter were achieved by roasting by CMR and microwave method, respectively (p<0.05, Table 1, nos. 107 and 167). Pyridines are considerably less desirable components of coffee aroma. Only two of them were identified in green coffee (4% of headspace) and five in roasted material (8, 9, and 6% of headspace for convective, microwave, and CMR methods, respectively). Pyridines have bitter, astringent aroma. Pyridine ranked the first of them by weight. Its lowest content was detected in coffee derived by CMR method (p<0.05, Table 1, no. 50). It has an unpleasant, sharp aroma. Its much less abundant derivatives, such as 2-methyland 5-methylpyridine (p<0.05, Table 1, nos. 57 and 71) have milder note. Pyridine is one of compounds unique of coffee aroma and therefore it is regarded as a marker of coffee brews, also after their consumption [31]. Pyrrols, like pyridines, also negatively affect aroma of coffee. Green coffee contained two compounds of this group (less than 1% of headspace), while six pyrrol derivatives were identified in roasted coffee (3% of headspace for convective heating and slightly above 2% for the other roasting methods). Their aroma resembles mushrooms and smoke. The most abundant of them was 3-formyl-1-methylpyrrol (similar concentration in examined samples, p>0.05, Table 1, no. 171). 1H-Pyrrol and 1-methylpyrrol ranked second in abundance. Concentration of the latter was the highest in convectively heated coffee (p<0.05, Table 1, nos. 46 and 111), which was particularly rich in pyrrols. Alcohols are volatile compounds more characteristic of green coffee, which contained 13 of them (18% of headspace), whereas only 7 were detected in roasted coffee (2% of headspace for coffee roasted by convective and microwave methods, 3% of headspace for CMR method). Green coffee was particularly rich in 1,2-butandiol, with lesser amounts of 2-methyl-2-buten-1-ol, 3-methyl-3-buten- 2-ol and 3-methyl-1-butanol (p<0.05, Table 1, nos. 117, 16, 17 and 52). Roasted coffee contained relatively high amounts of 2-methyl-2-buten-1-ol, 3-methyl-3-buten-2-ol and 1,2-ethandiol (p<0.05, Table 1, nos. 16, 17 and 73).

18 Eur Food Res Technol (2007) 225:9 19 Owing to high volatility and susceptibility to oxidation, these compounds are detectable in higher amounts in freshly roasted coffee. These physical features probably contributed to their highest content in headspace of coffee roasted by CMR method (the shortest time of processing). Hydrocarbons were also identified in larger amounts in green coffee (four compounds, 3% of headspace). Roasted coffee contained three hydrocarbons (less than 1% of headspace for all examined roasting methods). Characteristic of roasted coffee was 1,3-pentadien (p<0.05, Table 1, no. 1), while characteristic of green coffee were α-pinen and β-mircen (p<0.05, Table 1, nos. 27 and 49). 2,3-Pentadien was detected in both green and roasted coffee (p<0.05, Table 1, no. 35). Volatile acids were found only in green coffee (three compounds, 9% of headspace). Acetic acid accounted for almost 7.5% of them. Despite numerous reports on the presence of this compound in roasted coffee it was not detected in our studies (Table 1, no. 103). Both green and roasted coffee contained four esters (slightly above 1% in the headspace of green beans versus approximately 0.5% in that of roasted coffee). Roasted coffee additionally contained N,N,N -trimethylhydrazine. The latter accounted for more than 1% of headspace of coffee roasted by convective and CMR methods, and for almost 2% of that of microwaved coffee (p<0.05, Table 1, no. 9). It was accompanied by 2,4,5-trimethyloxazol (less than 0.1% of headspace for all examined methods of roasting, p<0.05, Table 1, no. 51) and maltol (Table 1, no. 170). The latter was typical of roasted coffee. Samples of green and roasted coffee were relatively rich in γ -butyrelacton (almost 3 and 3 4% of headspace of green and roasted coffee, respectively; the highest content in convectively heated coffee, Table 1, no. 131) and caffeine (high percentage content in the headspace of green coffee, p<0.05, Table 1, no. 193). Green coffee contained also two derivatives of indene (Table 1, nos. 147, 159). Summary Total amount of volatile compounds, which was determined as a sum of surface area under peaks of compounds separated by gas chromatography, was considerably affected by the method of coffee roasting. It resulted from the fact that both time and temperature of thermal processing have an impact not only on formation of individual components of aroma but also on their retention in roasted coffee. Therefore, application of microwaves, particularly coupled with convective heating increased total amounts of volatile substances of roasted coffee. Microwaving considerably reduced the ultimate temperature of roasted coffee beans, and simultaneous application of both heating methods markedly decreased the time of roasting (at moderately high temperature of beans). Because of these reasons, this variant of roasting conditions was found to be the most advantageous for formation and retention of volatile substances. However, results obtained by SPME-GC analysis and by evaluation of sensory attributes were contradictory because the brews from coffee roasted by CMR method received the worse scores (the most acceptable were brews from microwaved coffee). Although the retention of odorants regarded as beneficial for coffee aroma was the highest in coffee roasted by CMR method, their proportions in headspace of microwaved coffee were superior to that of other examined samples. Microwaving resulted in high amounts of pyrazines and phenol-derivatives, which presumably improved sensory properties (nutty aroma), while coffee roasted by CMR method was rich in furans, which contributed to stronger bitter and burnt hint. Thus, lowering of the ultimate temperature of roasted coffee beans was more beneficial for the overall aroma than the considerable intensification of roasting and reducing its time. The advantageous effect of processing by microwaves on aroma of other foodstuffs was observed earlier, e.g. for sliced banana and chopped carrots [32]. Future studies should focus on comparison of contents of volatile compounds in coffee brews and their headspace to find correlation between results of sensory analysis and profiles of odorants. References 1. Sarrazin C, Le Quéré J-L, Gretsch C, Liardon R (2000) Food Chem 70:99 106 2. Sanz C, Czerny M, Cid C, Schieberle P (2002) Eur Food Res Technol 214:299 302 3. Czerny M, Mayer F, Grosch W (1999) J Agric Food Chem 47:695 699 4. Meyer F, Czerny M, Grosch W (1999) Eur Food Res Technol 209:242 250 5. Schröder I, Stern G, Hojabr-Kalali K, Schliekelmann K, Maier HG (1999) Dutsch Leb-Rundsch 93:216 218 6. Jensen MR, Kirkpatrick SJ, Leppla JK (1994) US Patent, 5,333,703 7. Sivetz M (1991) ASIC 14th Colloque, San Francisco 313 317 8. Le Viet T, Truchement B (1988) CH Patent, 665 754 A5 9. Gerling JF (1984) US Patent, 4,326,114 10. Nebesny E, Budryn G (2006) Eur Food Res Technol, online publication http://dx.doi.org/10.1007/s00217-006-0308-y 11. Nebesny E, Budryn G (2003) Eur Food Res Technol 217:157 163 12. Nebesny E, Budryn G (2006) Deut Lebensm Rundsch, in press 13. ISO 10470 (1993) 14. ISO 4149 (1980) 15. Sanz C, Ansorena D, Bello J, Cid C (2001) J Agric Food Chem 49:1364 1369 16. Lewandowicz G, Jankowski G, Fornal J (2000) Carbohyd Polym 42:193 199 17. Friedman M, Dao L (1990) J Agric Food Chem 38:805 808 18. Clarke RJ (1990) Ital J Food Sci 2:79 88 19. Mayer F, Grosch W (2001) Flavour Fragr J 16:180 190 20. Lee K-G, Shibamoto T (2002) Flavour Fragr J 17:349 351 21. Ramos E, Valero E, Ibáňez E, Reglero G, Tabera J (1998) J Agric Food Chem 46:4011 4016

Eur Food Res Technol (2007) 225:9 19 19 22. Bicchi CP, Panero OM, Pellegrino GM, Vanni AC (1997) J Agric Food Chem 45:4680 4686 23. Hashim L, Chaveron H (1996) Food Res Int 28:619 623 24. Belitz HD, Grosch W (1999) Food chemistry. -Verlag, Berlin, Hilderberg, p 322 25. Jung MY, Bock JY, Back SO, Lee TK, Kim JH (1997) Food Chem 60:95 102 26. Czerny M, Wagner R, Grosch W (1996) J Agric Food Chem 44:3268 3272 27. Blank I, Sen A, Grosch W (1991) ASIC 14th Colloque, San Francisco 117 128 28. Boosfeld J, Vitzthum OG (1995) J Food Sci 60:1092 1096 29. Spadone JC, Takeoka G, Liardon R (1990) J Agric Food Chem 38:226 233 30. Belitz HD, Grosch W (1999) Food chemistry. -Verlag, Berlin, Hilderberg, pp 344 346 31. Hida Y, Matsumoto M, Kudo K, Imamura T, Ikeda N (1997) Int J Legal Med 111:13 16 32. Nijhuis HH, Torringa HM, Muresan S, Yuksel D, Leguit C, Kloek W (1998) Trends Food Sci Technol 9:13 20 33. Yoshida H, Kajimoto G (1989) J Food Sci 54:1596 1600