Journal of the Science of Food and Agriculture J Sci Food Agric 82:840±847 (online: 2002)

DOI: 10.1002/jsfa.1110

Profiles of volatile compounds and sensory

analysis of three blends of coffee: influence of
different proportions of Arabica and Robusta
and influence of roasting coffee with sugar
Cristina Sanz, Laura Maeztu, Ma Jose Zapelena, José Bello and
Concepción Cid*
Departamento de Bromatologı́a, Tecnologı́a de Alimentos y Toxicologı´a, Facultad de Farmacia, Universidad de Navarra, E-31080
Pamplona, Spain

Abstract: One hundred and forty-six volatile compounds were identi®ed and quanti®ed using a static
headspace sampler in three blends of coffee: Arabica/Robusta 80:20 (A80:R20) natural roasted coffee,
Arabica/Robusta 20:80 (A20:R80) natural roasted coffee and Arabica/Robusta 20:80 with 50% of
Robusta coffee roasted with sugar (A20:R80 50% Torrefacto). The different proportion of Arabica and
Robusta coffee in the blend A80:R20 versus A20:R80 in¯uenced the amounts of 20 chemical families of
volatile compounds. Aldehydes, ketones, alcohols, pyrroles, pyrazines, furans, thiazoles, thiophenes,
esters, oxazoles, lactones, sulphur compounds, pyridines, alkanes, alkenes, phenolic compounds,
benzenic compounds, acids, pyranones and terpenes were present in higher quantities in the sample
containing 80% of Arabica coffee, whereas sulphur compounds were more abundant in the coffee with
80% of Robusta. Sensory differences were also found between the two blends of coffee in the burnt,
caramel, nutty, earthy and roasty notes. Torrefacto coffee, widely consumed in Spain, is obtained by
roasting coffee with sugar. Higher quantities of ketones, alcohols, pyrazines, furans, pyridines,
alkanes, phenolic compounds, pyranones and terpenes were found in the blend A20:R80 50% Torre-
facto coffee versus A20:R80 natural roasted coffee. These differences in the volatile fraction were
perceived by our panellists in the intensities of the nutty, roasty, earthy, burnt and caramel notes.
# 2002 Society of Chemical Industry

Keywords: Torrefacto coffee; volatile compounds; static headspace; sensory analysis

INTRODUCTION and decreasing its aroma and ®neness. Usually, coffee

The two major species of coffee, Coffea arabica and roasted with sugar is of low quality and the defects of
Coffea canephora var robusta, differ considerably in the bean are masked by the sugar. This way of roasting
price, quality and consumer acceptance. Therefore coffee with sugar was invented to prevent coffee from
preferred ¯avour for coffee is obtained by blending the losing essential oils and to decrease staling of coffee by
varieties.1 Arabica is the coffee usually used in blends preventing oxygen from making contact with the bean.
for the aroma effect and Robusta is the one used for Originally it was popular as 100% Torrefacto, but now
taste and body.2 Owing to their low cost, Robusta a mix of pure Torrefacto and natural coffee has
coffees are used as ®llers in roasted and ground blends, become the standard. Blends can vary from 20±80%
adding earthy/phenolic notes to some dark-roasted (proportion of Torrefacto to natural coffee) to
coffee.1 30±70%, 40±60%, 50±50% and 60±40%.3 Torrefacto
In the Spanish market, Torrefacto coffee plays an coffee is widely used in Argentina, Costa Rica,
important role in consumption. To obtain Torrefacto Portugal and Spain, where its consumption represents
coffee, the coffee is roasted in the normal way and, at 83% of the consumption of coffee in the hotel trade.
the end of the process, sucrose (no more than 15% by The aroma fraction, amounting to only about 0.1%
weight) is added. At an approximate temperature of of the total ingredients,4 can be used to differentiate
roasting of 200 °C the sucrose changes to caramel via coffees from different geographic origins.5±8 Several
two non-oxidative reactions, caramelisation and Mail- research groups have studied the aroma pro®le of
lard reaction, forming a burnt ®lm round the coffee Arabica and Robusta coffees and their blends,9±15 as
bean, making it more bitter than natural roasted coffee well as their sensory differences.16±22 These investiga-

* Correspondence to: Concepción Cid, Departamento de Bromatologı́a, Tecnologı́a de Alimentos y Toxicologı́a, Facultad de Farmacia,
Universidad de Navarra, E-31080 Pamplona, Spain
E-mail: [email protected]
Contract/grant sponsor: PIUNA (Plan Investigacion Universidad de Navarra)
(Received 21 September 2001; revised version received 20 December 2001; accepted 30 January 2002)

# 2002 Society of Chemical Industry. J Sci Food Agric 0022±5142/2002/$30.00 840

 Aroma pro®le and sensory analysis of three blends of coffee

tions were carried out using different techniques for (70 eV) with a scan range of 33±300 amu. The ion
extraction of the volatile compounds, but none used a source temperature was set at 230 °C.
static headspace sampler, which allows extraction of
the very volatile compounds of coffee and reduction of Identification of volatile compounds
the errors associated with manual handling. Volatile compounds were identi®ed by comparing
The aim of this work was to identify the differences their spectra with those of the Wiley library and also by
in the volatile fraction and sensory characteristics of comparison of their GC Kovats index and retention
two blends of coffee with different proportions of time with those of standard compounds and data from
each variety: Arabica/Robusta 80:20 (A80:R20) and the literature. The Kovats indices were calculated
Arabica/Robusta 20:80 (A20:R80), both natural according to the method of Tranchant23 and were
roasted. Another objective of the present study was compared with available literature data.24 Only known
to analyse Arabica/Robusta 20:80 (A20:R80) natural peaks are shown (see Table 1).
roasted and Arabica/Robusta 20:80 with 50% of
Robusta coffee roasted with sugar (A20:R80 50% Quantitative measurements
Torrefacto) in order to determine the changes in the Areas of peaks were measured by integration of the
volatile composition and sensory characteristics of a total ion current of the spectra or by calculation of the
coffee when roasting it with or without sugar. In both total area based on integration of a single ion.
cases the volatile fraction was analysed with a static
headspace sampler according to a method previously Sensory analysis
described by Sanz et al. 15 Eight ¯avour characteristics (malty, roasty, buttery,
fruity, caramel, nutty, burnt and earthy/musty) were
evaluated in the three blends of roasted coffee by
EXPERIMENTAL selected and trained judges. The selection and training
Materials processes were carried out according to UNE 87024-
Different blends of roasted coffee beans were pur- 1: 1995.25 Sensory ¯avour evaluation of the ground
chased from a local market: Arabica/Robusta 80:20 coffee samples was carried out in triplicate over six
(A80:R20) natural roasted, Arabica/Robusta 20:80 sessions, and three samples were analysed per session.
(A20:R80) natural roasted and Arabica/Robusta 20:80 Each sample was ground immediately before each
with 50% of Robusta coffee roasted with sugar session and served in a white porcelain dish labelled
(A20:R80 50% Torrefacto). Just before each analysis monadically with a three-digit code. The order of
the coffee was ground in a Retsch mill to a particle presentation was randomised among judges and
size 40.75 mm. sessions. All evaluations were conducted in isolated
Pure samples of some volatile compounds were sensory booths illuminated with white light in the
obtained from the sources reported earlier.15 In sensory laboratory under standarised conditions ac-
addition, other reference compounds were analysed: cording to UNE 87004: 1979.25
furan from Aldrich (Steinheim, Germany), pyrrole
from Sigma (Steinheim, Germany) and acetaldehyde, Data analysis
2-methylbutanal, 3-diethyl-2,5-methylpyrazine and Data analysis was carried out using SPSS 9.0 for
2-propanone from ACROS (New Jersey, USA). Windows. Values of areas for each volatile compound
are the mean of three determinations. Student's t-test
Static headspace–gas chromatography–mass was used to determine whether there were differences
spectrometry (SH–GC–MS) in the areas of each compound and each chemical class
The extraction was performed with a Hewlett Packard between A80:R20 and A20:R80 and between
(Wilmington, USA) model 7694 static headspace A20:R80 natural roasted and A20:R80 50% Torre-
sampler. Vials (10 ml) containing 2 g of roasted ground facto.
coffee were immediately sealed with silicone rubber
Te¯on caps and equilibrated in the static headspace
sampler at 90 °C for 60 min.
The GC±MS method used has been described in a RESULTS AND DISCUSSION
previous publication.15 The injector temperature was One hundred and forty-six volatile compounds were
set at 180 °C and helium (10 ml min 1 linear speed) identi®ed using a static headspace sampler in three
was used as the carrier gas. The oven temperature was blends of coffee: Arabica/Robusta 80:20 natural
maintained at 40 °C for 6 min and then programmed to roasted (A80:R20), Arabica/Robusta 20:80 natural
190 °C at 3 °C min 1. A 3 ml coffee headspace sample roasted (A20:R80) and Arabica/Robusta 20:80 with
was injected into an HP-Wax capillary column (60 m  50% of Robusta coffee roasted with sugar (A20:R80
0.25 mm  0.5 mm ®lm thickness; Hewlett Packard). 50% Torrefacto) (Table 1). Most of them have been
Mass spectrometry analysis was carried out using a previously identi®ed15 and, in addition, 39 new
Hewlett Packard model 5973 mass-selective detector compounds have been identi®ed. Table 1 shows the
coupled to the gas chromatograph. The mass spectro- areas of the 146 compounds identi®ed in the three
meter operated in the electron impact ionisation mode samples of coffee.

J Sci Food Agric 82:840±847 (online: 2002) 841

C Sanz et al

Table 1. Area (10 3) of each volatile compound identified in headspace gas of roasted ground coffee samplesa

KI b Rel c Compound A80:R20 LS d A20:R80 LS d A20:R80 Torrefacto

Aldehydes
645 A Acetaldehyde 199657 (5990) *** 114373 (1144) * 112895 (1129)
712 A Propanal 201294 (8052) *** 117410 (2348) * 113325 (3400)
747 A 2-Methylpropanal 388483 (23309) * 250722 (2507) * 261586 (2616)
839 C Butanal 10417 (938) * 7772 (1088) * 6009 (781)
880 A 2-Methylbutanal 168530 (1685) *** 104217 (4169) * 110692 (2214)
884 A 3-Methylbutanal 141026 (1410) *** 91529 (5492) * 84885 (849)
1042 A 2-Butenal 3996 (679) ** 2409 (434) ns 2471 (395)
1084 A Hexanal 5060 (810) * 3607 (577) * 4299 (602)
1102 C 2-Methyl-2-butenal 4710 (706) * 3466 (520) * 2863 (487)
1143 C 2-Pentenal 952 (67) ** 648 (78) * 562 (84)
Total aldehydes 1124125 *** 696153 ns 699588
Ketones
753 A 2-Propanone 935896 (9359) ** 773519 (38676) *** 600863 (6009)
866 A 2-Butanone 74974 (2249) *** 63360 (1267) *** 48878 (1955)
962 A 2,3-Butanedione 189711 (15177) *** 113268 (1133) ** 120463 (2409)
1048 C 2,4-Dimethyl-3-pentanone 1438 (288) * 932 (28) * 1000 (210)
1053 C 3-Hexanone 3504 (526) ns 3318 (498) * 2544 (356)
1058 A 2,3-Pentanedione 155353 (1554) *** 64225 (3211) *** 95267 (953)
1138 C 2,3-Hexanodione 5828 (699) ** 3267 (457) * 3629 (581)
1138 A 3-Penten-2-one 5976 (657) ** 4034 (686) * 3684 (626)
1143 C 3,4-Hexanedione 5683 (739) ** 2935 (470) * 3847 (577)
1202 C Cyclopentanone 4506 (541) * 3871 (116) ** 2807 (477)
1307 C 3-Hydroxy-2-butanone 70321 (2813) *** 48584 (1943) ** 57083 (571)
1323 C 1-Hydroxy-2-propanone 126597 (2532) *** 84181 (2525) *** 102560 (1026)
1366 C 3-Hydroxy-2-pentanone 21898 (1752) *** 11968 (359) *** 16294 (2118)
1385 C 2-Cyclopenten-1-one 7687 (769) ** 5557 (500) * 5342 (481)
1397 C 2-Methyl-2-cyclopenten-1-one 2806 (505) * 1969 (315) ns 1872 (300)
1399 C 1-Hydroxy-2-butanone 35009 (1400) *** 17857 (357) *** 30092 (1204)
1473 C 3-Methyl-2-cyclohexenone 391 (47) ** 251 (28) ** 313 (3)
1562 C 3-Methyl-2-cyclopenten-1-one 942 (141) ** 516 (77) * 731 (132)
1582 C 2,3-Dimethylcyclopent-2-en-1-one 1191 (155) ** 741 (148) * 876 (131)
Total ketones 1649711 *** 959530 * 1098146
Alcohols
913 B Ethanol 52826 (1057) *** 25296 (2024) * 27279 (273)
1051 B 2-Methyl-3-buten-2-ol 6992 (979) ** 2591 (440) ns 2657 (398)
1103 A 2-Methyl-1-propanol 4372 (743) * 3210 (546) ns 3092 (464)
1124 A 2-Propenol 1292 (271) * 876 (175) * 888 (178)
1130 A 2-Pentanol 258 (67) * 315 (82) ns 325 (91)
1220 C 2-Methyl-1-butanol 4516 (677) * 3536 (566) ns 3320 (631)
1220 C 3-Methylbutan-1-ol 2422 (533) * 1651 (330) ns 1624 (325)
1264 C 3-Methyl-3-buten-1-ol 7029 (633) ** 4571 (503) ns 4401 (484)
1332 C 2-Heptanol 759 (152) * 1187 (237) ** 2319 (186)
1337 C 3-Methyl-2-buten-1-ol 4168 (458) ** 1759 (387) * 1503 (150)
1509 C 2-Ethyl-1-hexanol 1077 (151) ** 547 (115) ** 1804 (343)
Total alcohols 85710 ** 45538 * 49212
Pyrroles
1149 B 1-Methylpyrrole 48638 (2432) ** 35179 (2111) *** 23287 (23451)
1194 C 1-Ethyl-1H-pyrrole 3825 (230) * 3512 (211) ** 2555 (102)
1225 D Dimethylpyrrole 3776 (378) ** 2198 (198) ** 1229 (147)
1542 A 1H-Pyrrole 33202 (4980) ns 32692 (1308) *** 14742 (1179)
1580 D Methylpyrrole 2984 (149) ** 2277 (114) *** 1094 (67)
1601 D Methylpyrrole 2777 (222) *** 174 (7) *** 1392 (84)
1661 B 2-Formyl-1-methylpyrrole 8696 (609) *** 4927 (443) * 5782 (405)
C N-Furfurylpyrrole 3660 (512) * 2632 (421) ** 3865 (309)
Total pyrroles 107558 ** 83591 *** 53945
Pyrazines
1231 B Pyrazine 47341 (1894) *** 24188 (1209) *** 52738 (1582)
1288 B 2-Methylpyrazine 303459 (3035) *** 180887 (3618) *** 277104 (5542)
1347 B 2,5-Dimethylpyrazine 47991 (1920) *** 29717 (1189) *** 21372 (1069)
1353 B 2,6-Dimethylpyrazine 99436 (3977) *** 46427 (1857) *** 85502 (3420)
1359 B 2-Ethylpyrazine 71619 (5013) *** 28405 (2272) *** 49337 (2960)
1372 C 2,3-Dimethylpyrazine 15095 (1057) ** 11918 (953) ** 17098 (1368)
1411 B 2-Ethyl-6-methylpyrazine 23682 (947) *** 11424 (800) *** 25247 (1515)
1419 B 2-Ethyl-5-methylpyrazine 14137 (848) ** 11816 (709) ** 9444 (283)
1432 C 2-Ethyl-3-methylpyrazine 7872 (1023) * 7158 (931) ** 5131 (257)
1433 B 2,3,5-Trimethylpyrazine 9114 (1002) ** 6182 (866) * 4979 (846)
1447 C N-Propylpyrazine 1458 (233) * 987 (168) * 1334 (253)
1463 C 2,6-Diethylpyrazine 1639 (229) ** 781 (125) ** 1805 (343)
1467 C 2-Vinylpyrazine 2526 (429) ** 1157 (162) ** 2237 (201)

842 J Sci Food Agric 82:840±847 (online: 2002)

 Aroma pro®le and sensory analysis of three blends of coffee

Table 1. continued

KI b Rel c Compound A80:R20 LS d A20:R80 LS d A20:R80 Torrefacto

1474 A 3-Ethyl-2,5-dimethylpyrazine 12834 (1027) * 9784 (1565) * 8286 (249)
1492 C 2,5-Diethylpyrazine 9150 (1556) * 8287 (1077) * 6618 (728)
1494 C 2,6-Diethylpyrazine 2706 (379) * 2417 (218) ** 1862 (149)
1520 C 2-Methyl-5-vinylpyrazine 1418 (270) ** 603 (97) ** 1447 (275)
1521 C 2-Methyl-6-vinylpyrazine 898 (180) * 628 (94) * 578 (110)
1665 C 2-Acetylpyrazine 948 (171) ** 555 (111) * 839 (101)
1667 C 5-Methyl-6,7-dihydro-5H-cyclopentapyrazine 1001 (140) ** 531 (101) ** 1085 (87)
1723 C 1-(5-Methyl-2-pyranizyl)-1-ethanone 511 (97) * 295 (59) ** 616 (99)
1730 C 1-(6-Methyl-2-pyranizyl)-1-ethanone 1484 (208) ns 1494 (149) * 1397 (84)
Total pyrazines 676321 *** 385641 *** 576055

Furans
716 A Furan 75175 (1504) ** 66184 (1986) *** 48220 (964)
832 B 2-Methylfuran 310409 (21729) ** 254047 (2540) *** 192092 (1921)
858 A 3-Methylfuran 15859 (1110) ** 11432 (1029) * 9202 (1472)
930 B 2,5-Dimethylfuran 23104 (1848) ** 15825 (1108) * 14472 (579)
975 B 2-Ethylfuran 1030 (196) * 717 (201) ns 650 (149)
1056 B 2,3,5-Trimethylfuran 1409 (113) ** 889 (178) * 1050 (221)
1075 B 2-Vinylfuran 5315 (585) ** 3065 (429) ns 3108 (466)
1160 B 2-Vinyl-5-methylfuran 10284 (1131) ** 5644 (734) * 6187 (495)
1181 C 2-(2-Propenyl)furan 2353 (471) * 1674 (254) ns 1628 (277)
1241 C 2-Pentylfuran 741 (163) ** 1629 (358) * 2137 (342)
1252 B Furfuryl methyl ether 7052 (564) ** 4772 (525) ns 4864 (486)
1283 B 2-Methyltetrahydrofuran-3-one 63835 (3192) *** 31864 (1275) *** 42784 (2139)
1304 C trans-2-Methyl-5H-propenylfuran 1456 (335) * 950 (190) ns 932 (149)
1461 C Furfuryl mercaptan 523 (110) ** 1105 (166) *** 242 (48)
1490 C 2-Furancarboxaldehyde 129037 (6452) *** 51627 (2581) *** 149865 (1499)
1516 C 2-Furfuryl methyl sulphide 3602 (504) * 3250 (455) * 2723 (245)
1519 C Furfuryl formate 24273 (1214) *** 13492 (1079) * 15396 (308)
1536 B 2-Acetylfuran 1470 (279) *** 13455 (1076) ** 16420 (985)
1548 C 1-(2-Furyl)-2-propanone 4707 (659) ** 2767 (166) * 3079 (431)
1559 C Furfuryl acetate 75172 (1503) *** 52560 (526) * 49688 (994)
1605 B 5-Methylfurfural 77789 (1556) *** 31263 (1251) ** 25901 (777)
1611 C 2-Propionylfuran 2356 (165) ** 1579 (237) * 1768 (195)
1620 B Furfuryl propionate 1905 (286) * 1511 (196) ns 1501 (180)
1626 B 2,2'-Bifuran 559 (123) ** 281 (28) ** 599 (144)
1636 C 2-Furfurylfuran 3175 (445) * 3554 (498) * 4434 (89)
1650 C 2-Acetyl-5-methylfuran 1276 (204) ** 733 (51) * 963 (116)
1686 B Furfuryl alcohol 620310 (6203) *** 332370 (3324) *** 537143 (5371)
1706 C 5-Methyl-2-furfurylfuran 1413 (212) * 1114 (145) * 1022 (82)
C Furfuryl methyl disulphide 286 (63) ns 284 (57) ** 723 (130)
Total furans 1466776 *** 909957 *** 1139485

Thiazoles
1260 C 2-Methylthiazole 236 (57) * 308 (62) * 239 (24)
1270 C 1,3-Thiazole 2078 (145) ** 1298 (260) * 1844 (387)
1304 C 4-Methylthiazole 2835 (595) ns 2797 (168) *** 1419 (241)
Total thiazoles 5149 * 4403 * 3502

Thiophenes
1021 A Thiophene 7057 (776) * 6105 (855) * 4181 (544)
1097 D Methylthiophene 4753 (713) * 3431 (343) * 3079 (493)
1127 D Methylthiophene 1430 (286) * 1158 (139) * 793 (198)
1565 C 2-Methyltetrahydrothiophen-3-one 2015 (121) * 1568 (204) * 1982 (258)
1565 C Tetrahydrothiophen-3-one 688 (151) * 454 (100) ns 425 (85)
1734 C 2-Thiophencarboxaldehyde 970 (165) ** 553 (105) ** 910 (109)
Total thiophenes 16914 * 13269 * 11370

Esters
682 C Formic acid, methyl ester 388878 (7778) *** 219169 (1320) * 224299 (2508)
782 C Acetic acid, methyl ester 203606 (16289) * 194645 (1946) *** 134613 (1346)
850 B Acetic acid, ethyl ester 1866 (373) * 2507 (401) *** 6973 (976)
872 C Propanoic acid, methyl ester 7578 (1061) * 6400 (704) * 5129 (872)
1060 C Ethanethioic acid, methyl ester 473 (118) ns 445 (111) * 271 (81)
1119 B 2-Butenoic acid, methyl ester 274 (66) ns 284 (74) * 446 (116)
1180 C 2-Butenoic acid, 3-methyl, methyl ester 268 (70) * 311 (78) ns 372 (33)
1335 C Propanoic acid, 2-hydroxy, methyl ester 3111 (187) * 2704 (162) * 2601 (156)
1403 C Acetic acid, hydroxy, methyl ester 658 (158) * 464 (116) ns 515 (124)
1484 C 1-Hydroxy-2-propanone acetate 155882 (6235) *** 117100 (2342) *** 47388 (3317)
1556 C 1-Hydroxy-2-butanone acetate 14128 (565) *** 8521 (596) * 9245 (462)
C Benzoic acid, 2-hydroxy, methyl ester 223 (45) * 160 (35) ** 697 (202)
Total esters 776944 *** 552710 *** 432549

J Sci Food Agric 82:840±847 (online: 2002) 843

C Sanz et al

Table 1. continued

KI b Rel c Compound A80:R20 LS d A20:R80 LS d A20:R80 Torrefacto

Oxazoles
1163 C 4,5-Dimethyloxazole 2133 (107) * 1822 (128) ** 1372 (14)
1214 C Trimethyloxazole 2346 (141) * 2165 (130) ** 1533 (230)
Total oxazoles 4479 * 3987 ** 2905
Lactones
1673 C g-Butyrolactone 37756 (1510) ** 33181 (1327) ** 26986 (810)
Sulphur compounds
635 C Methanethiol 26159 (1308) ** 32367 (1295) *** 15117 (756)
660 C Carbon disulphide 13335 (1067) ** 19549 (1368) ** 15743 (945)
671 A Dimethyl sulphide 10326 (929) * 9247 (1480) ** 5054 (51)
1077 B Dimethyl disulphide 8530 (1109) * 6590 (857) * 5397 (594)
Total sulphur compounds 58350 * 67753 *** 41312
Pyridines
1203 B Pyridine 412785 (4128) *** 223880 (2239) *** 434771 (4348)
1239 D Methylpyridine 3663 (366) * 3221 (193) ** 2244 (135)
1239 C 2-Methylpyridine 1769 (354) ** 873 (175) ** 1854 (260)
1323 C 3-Methylpyridine 2171 (347) ** 839 (159) *** 2472 (148)
1644 C 2-Acetylpyridine 402 (88) ** 180 (54) ** 371 (52)
Total pyridines 420789 *** 228993 *** 441711
Alkanes
915 C 2,2,4,6,6-Pentamethylheptane 4752 (855) *** tre *** 19586 (392)
Alkenes
624 C 1,3-Pentadiene 25159 (1006) *** 11086 (554) ** 8862 (709)
Phenolic compounds
C 2-Methoxyphenol 4540 (590) * 3957 (554) *** 7445 (521)
Benzenic compounds
1040 B Toluene 2300 (368) * 1836 (367) * 1567 (313)
1273 C Ethenylbenzene 296 (80) ns 291 (70) * 213 (64)
1146 B p-Xylene 87 (24) * 124 (25) ** 59 (9)
1149 B m-Xylene 477 (115) * 408 (98) * 272 (19)
1240 C Styrene 585 (123) * 646 (149) ns 673 (47)
1248 B p-Cymene 153 (29) ** 59 (17) ** 256 (56)
1561 C Benzaldehyde 3421 (171) ** 2429 (170) ns 2376 (190)
Total benzenic compounds 7318 * 5793 ns 5415
Acids
1480 B Acetic acid 383916 (11517) ** 357242 (3572) *** 201613 (2016)
C Propanoic acid 6898 (207) *** 9566 (287) *** 6905 (483)
Total acids 390813 ** 366808 *** 208518
Pyranones
1439 C Tetrahydropyran-3-one 1510 (226) *** 318 (64) * 399 (80)
Terpenes
1169 D Terpene 992 (79) *** 289 (32) * 357 (43)
a
Mean values with their standard deviation in parentheses.
b
Kovats index calculated for the HP-Wax capillary column.
c
The reliability of the identi®cation proposal is indicated by the following: A, mass spectrum, retention time and Kovats index agreed with standards; B, mass
spectrum and Kovats index agreed with literature data; C, mass spectrum compared with the Wiley mass spectral database; D, tentative identi®cation according
to mass spectra.
d
Level of signi®cance for the differences between A80:R20 and A20:R80 and between A20:R80 natural roasted and A20:R80 50% Torrefacto: ns, not signi®cant
(p > 0.05).
* p 4 0.05.
** p 4 0.01.
*** p 4 0.001.
e
Trace compound.

Comparison between A80:R20 and A20:R80 compounds, pyridines, alkanes, alkenes, phenolic
The gas chromatographic pro®les of the coffee blends compounds, benzenic compounds, acids, pyranones
A80:R20 and A20:R80 referring to the area of each and terpenes.
compound and chemical class were signi®cantly A higher quantity of sulphur compounds was found
different. All the chemical families showed signi®cant in the blend containing 80% of Robusta (Table 1).
differences between the two blends: aldehydes, Other authors have also found higher concentrations
ketones, alcohols, pyrroles, pyrazines, furans, thia- of these compounds in Robustas than in Arabicas
zoles, thiophenes, esters, oxazoles, lactones, sulphur using the headspace technique.26,27 In the present

844 J Sci Food Agric 82:840±847 (online: 2002)

 Aroma pro®le and sensory analysis of three blends of coffee

study, four sulphur compounds were identi®ed using a Arabica. However, Leino et al 18 reported that the
static headspace sampler and a mass-selective detector proportions of 3-methylbutanal and 2-methylpropanal
(MS): methanethiol and carbon disulphide were more were higher in Arabica than in Robusta and that
abundant in A20:R80, and dimethyl sulphide and 2-methylfuran was about the same in both species. In
dimethyl disulphide were more abundant in A80:R20. our work, 2-methylpropanal, 3-methylbutanal and
These results are partly in agreement with Holscher 2-methylbutanal were present in higher quantities in
and Steinhart,28 who concluded that methanethiol was A80:R20 than in A20:R80 (Table 1). The Strecker
two times higher in Robusta than in Arabica coffees derivate 3-methylbutanal has been reported by Blank
and that dimethyl sulphide was present in higher et al 16 to be the key odorant of roasted Arabica ground
proportion in Arabica than in Robusta. According to coffee and to have malty notes. Semmelroch and
Leino et al,18 dimethyl sulphide was present in higher Grosch19 also reported the Strecker aldehydes
proportion in Arabica than in Robusta, and carbon 2-methylpropanal, 2-methylbutanal and 3-methyl-
disulphide and dimethyl disulphide were about the butanal to have malty and fruity notes, but they were
same in both coffees. These differences in comparison not useful either for differentiating between coffees of
with our results could be due to the different time and different origins. In our study, no signi®cant differ-
temperature programmed in the headspace, par- ences were found in the malty note detected between
ameters that affect the extraction of volatile com- A80:R20 and A20:R80 by our panellists (Fig 1).
pounds when using this method.15 Total amounts of the a-dicarbonyls 2,3-butano-
There are numerous furan compounds in coffee, dione and 2,3-pentanodione were higher in A80:R20
responsible for the burnt sugar, burnt and caramel than in A20:R80 (Table 1), which was in agreement
aroma.18 Leino et al 18 did not ®nd any differences in with the results reported by Holscher and Steinhart.28
furan compounds in the samples they analysed, except Blank et al 17 indicated that compounds causing
for 2-methylfuran. We identi®ed the same furan com- `buttery odour' were higher in Arabica than in Robusta
pounds in both samples (A80:R20 and A20:R80), but species. Semmelroch and Grosch19 and Blank et al 16
they differed signi®cantly in their abundance in these showed these volatiles as potent odorants of Arabica
samples. and Robusta powders, but they were not useful in
Among the aldehydes identi®ed, many of them are differentiating between the two coffees. In our work
important key odorants of coffee. Acetaldehyde and the analytical results showed these compounds to be
propanal have been reported to have fruity notes19 and more abundant in the blend with a higher proportion
in our work have been useful, according to our panel- of Arabica coffee. These differences were not detect-
lists, to differentiate between A80:R20 and A20:R80 able by our panellists, since they did not report any
(Fig 1). Different results have been shown by different differences between the two types of coffee in the
authors for the Strecker aldehydes. Liardon et al 29 and buttery note (Fig 1).
Holscher and Steinhart28 noted that 2-methylpropa- Some studies have indicated that pyrazines are
nal, 2-methylbutanal and 3-methylbutanal were more predominant in Robustas,16,17,31 but, in our study,
abundant in Robusta than in Arabica. Piringer30 pyrazines were more abundant in A80:R20 than in
showed that 2-methylpropanal, 3-methylbutanal and A20:R80 (Table 1). However, the earthy odour quality
2-methylfuran contents were higher in Robusta than in previously connected with pyrazines18,32,33 was more
intense in the sample containing a higher proportion of
Robusta (Fig 1). This higher earthy note in Robusta
coffees has also been pointed out by Blank et al 16 and
Leino et al. 18
The roasty note, more intense in the coffee contain-
ing a higher proportion of Arabica, was the sensory
characteristic for which the highest differences were
found by our panellists (Fig 1). These differences
could not be related to any particular compound
identi®ed in our work. According to Blank et al,16 the
roasty note is due to compounds such as 2-furfuryl-
thiol, 3-mercapto-3-methylbutylformate, 2-methoxy-
3-isopropylpyrazine, 3,5-dimethyl-2-ethylpyrazine
and 2,3-diethyl-5-methylpyrazine, compounds that
were not identi®ed in our work, probably because
they were in such low quantity that they could not be
extracted with the static headspace.

Comparison between A20:R80 natural roasted and

A20:R80 50% Torrefacto
Figure 1. Percentage of judges that detected each note when comparing The differences between the types of roasting, with or
A80:R20 and A20:R80 natural roasted. without sugar, were estimated by comparing the areas

J Sci Food Agric 82:840±847 (online: 2002) 845

C Sanz et al

of the 146 compounds listed in Table 1 and the

sensory results. Nine chemical classes were present in
signi®cantly higher quantities in A20:R80 50% Tor-
refacto than in A20:R80 (Table 1): ketones, alcohols,
pyrazines, furans, pyridines, alkanes, phenolic com-
pounds, pyranones and terpenes. The increases in
pyrazines, furans and pyridines con®rm that these
compounds are formed during roasting in greater
quantities when sugar is added. This fact was con-
®rmed by our panellists, who reported higher roasty,
burnt and caramel notes in the Torrefacto coffee.
According to Barcarolo et al,34 Reineccius35 and Baltes
and Bochmann,36 the higher abundance of pyrazines
and pyridines in A20:R80 50% Torrefacto could be
related to an increase in the Maillard reaction when
roasting coffee with sugar. However, furans could be
formed during the caramelisation reaction.36,37
The areas of aldehydes, pyrroles, thiazoles, esters, Figure 2. Percentage of judges that detected each note when comparing

oxazoles, lactones, sulphur compounds and acids were A20:R80 natural roasted and A20:R80 50% Torrefacto.

higher in A20:R80 than in A20:R80 50% Torrefacto

(Table 1). The decreases in these eight chemical
classes in the latter could be partly due to the fact that, intense in the coffee containing a higher proportion of
in the 2 g of roasted ground Torrefacto coffee Arabica.
analysed, part of the coffee had been replaced with When roasting coffee with sugar (A20:R80 50%
sugar. Torrefacto), pyrazines, pyridines and furans were
According to Lingle,38 the contribution of carame- formed in greater quantity than in natural roasted
lised sugars and products of the Maillard reaction to coffee owing to increases in the Maillard and cara-
coffee aroma is dif®cult to assess because of the great melisation reactions. Roasty, burnt and caramel notes
number, complexity and stability of aromatic com- related to these compounds were useful to differentiate
pounds. In general, the caramelised sugar tends to between the Torrefacto and natural roasted coffees,
contribute to the aromatic sensation perceived as being more intense in the former.
nutty, caramel-like or chocolate. Ho et al 39 established
that the Maillard reaction was responsible for the
generation of roasted, toasted or caramel-like aromas ACKNOWLEDGEMENTS
as well as the development of brown colours in foods. We thank PIUNA (Plan Investigacion Universidad de
Clarke and Macrae37 concluded that the nutty roasted Navarra) for supporting the development of this work.
aroma could be related to the pyrazines, and some
furans could be responsible for the caramel ones.
In our study, as expressed before, the panellists REFERENCES
detected more intense nutty roasted and burnt notes in 1 Parliament TH and Stahl HD, What makes that coffee smell so
A20:R80 50% Torrefacto than in the same natural good? Chem Technol 8:38±47 (1995).
roasted blend, as well as a higher caramel aroma (Fig 2 Correia AMNG, A in¯ueÃncia da torra na evolucËao dos aÂcidos
clorogeÂnicos do cafeÂ. PhD Thesis, Instituto Superior de Agro-
2). nomia, Universidade TeÂcnica de Lisboa (1990).
3 Bell J, Spain's great fertile ground for new ideas in coffee. Tea
Coffee Trade J 12:31±35 (2000).
4 Vitzthum OG, Weisemann R, Becker R and KoÈhler HS, Iden-
ti®cation of an aroma key compound in Robusta coffees. CafeÂ,
CONCLUSIONS
Cacao, Th e 34:27±33 (1990).
Arabica/Robusta 80:20 and Arabica/Robusta 20:80 5 Tipler A, Spragg RA, Aries RE and Lidiard DP, The charac-
differed signi®cantly in their GC pro®les. A higher terization of coffee using headspace extraction, capillary gas
quantity of sulphur compounds was found in the blend chromatography, selective detection systems, pattern recogni-
A20:R80, while furans, Streker aldehydes, a-dicarbo- tion software and principal component analysis, Proceedings of
nyls and pyrazines were more abundant in A80:R20. the 13th International Symposium on Capillary Chromatography,
Ed by Pat S, HuÈthig, Heidelberg, pp 337±347 (1991).
These differences in GC pro®les were connected to the 6 Bicchi CP, Binello AE, Pellegrino GM and Vanni AC, Charac-
sensory differences between the two blends of coffee. terisation of roasted coffee by S-HSGC and HPLC-UV and
The fruity notes, related to the Strecker aldehydes, principal component analysis. J Agric Food Chem 41:2324±
acetaldehyde and propanal, and the earthy odour 2328 (1993).
quality, connected with pyrazines, were useful to 7 Hernandez CV and Rutledge DN, Multivariate statistical
analysis of gas chromatograms to differentiate cocoa masses
distinguish between the two coffee blends. The highest by geographical origin and roasting conditions. Analyst
sensory difference between blends A80:R20 and 119:1171±1176 (1994).
A20:R80 was in the roasty note, which was more 8 Mosca AI, ComposicËao quimica do aroma do cafe e sua

846 J Sci Food Agric 82:840±847 (online: 2002)

 Aroma pro®le and sensory analysis of three blends of coffee

correlaccËao com origem e controlo de qualidade. MSc Thesis, tion Indices for the Analysis of Aromatic Compounds. Laboratoire
Universidade de EÂvora (1994). Flaveur (INRA), Theix (1996).
9 Liardon R and Ott U, Application of multivariate statistics for the 25 AENOR, AnaÂlisis Sensorial Tomol. AlimentacioÂn. RecopilacioÂn de
classi®cation of coffee headspace pro®les. Lebensm Wiss Technol Normas UNE (1997).
17:32±38 (1984). 26 Gutmann W, Werkhoff P, Barthels M and Vitzthum OG,
10 Kallio H, Leino M, Koullias K, Kallio S and Kaitaranta J, Vergleich der Headspace±Aromapro®le von Arabusta-Kaffee
Headspace of roasted ground coffee as an indicator of storage mit Arabica- und Robusta sorten, in Proceedings of the 8th
time. Food Chem 36:135±148 (1990). International Scienti®c Colloquium on Coffee. ASIC, Paris, pp
11 Shimoda M and Shibamoto T, Isolation and identi®cation of 153±161 (1979).
headspace volatiles from brewed coffee with an on-column 27 Nurok D, Anderson JW and Zlatkis A, Pro®les of sulfur
GC±MS method. J Agric Food Chem 38:802±804 (1990). containing compounds obtained from Arabica and Robusta
12 Bicchi CP, Panero OM, Pellegrino GM and Vanni AC, Charac- coffees by capillary column gas chromatography. Chromato-
terization of roasted coffee and coffee beverages by solid phase graphia 11:188±192 (1978).
microextraction±gas chromatography and principal compo- 28 Holscher W and Steinhart H, Investigation of roasted coffee
nent analysis. J Agric Food Chem 41:2324±2328 (1997). freshness with an improved headspace technique. Z Lebensm
13 Pollien P, Ott A, Montigon F, Baumgartner M, MunÄoz-Box R Untersuch Forsch 195:33±38 (1992).
and Chaintreau A, Hyphenated headspace±gas chromatogra- 29 Liardon R, Ott U and Daget N, Analysis of coffee headspace
phy±snif®ng technique: screening of impact odorants and pro®les by multivariate statistics, in Analysis Volatiles: Methods
quantitative aromagram comparisons. J Agric Food Chem and Applications, Ed by Schreier P, Walter de Gruyter, Berlin,
45:2630±2637 (1997). pp 447±459 (1984).
14 Mayer F, Czerny M and Grosch W, Sensory study of the 30 Piringer O, Die Bedeutung organischer Spurenkomponenten fuÈr
character impact aroma compounds of a coffee beverage. Eur die QualitaÈtsicherung von Verpackung una Packgut. Verpack-
Food Res Technol 211:272±276 (2000). ungs Rundschau 34:39±45 (1983).
15 Sanz C, Ansorena D, Cid C and Bello J, Optimising headspace 31 Silwar R and LuÈllmann C, Investigation of aroma formation in
temperature and time sampling for identi®cation of volatile Robusta coffee during roasting. CafeÂ, Cacao, The 37:145±152
compounds in ground roasted Arabica coffee. J Agric Food (1993).
Chem 49:1364±1369 (2001). 32 Wada K, Sasaki H, Shimoda M and Osajima Y, Objective
16 Blank I, Sen A and Grosch W, Aroma impact compounds of evaluation of various trade varieties of coffee by coupling of
Arabica and Robusta coffee. Qualitative and quantitative analytical data and multivariate analyses. Agric Biol Chem
investigation, in Proceedings of the 14th International Scienti®c 51:1753±1760 (1987).
Colloquium on Coffee. ASIC, Paris, pp 117±129 (1991). 33 Tressl R, Formation of ¯avor components in roasted coffee. ACS
17 Blank I, Sen A and Grosch W, Potent odorants of the roasted Symp Ser 409:285±301 (1989).
powder and brew of Arabica coffee. Z Lebensm Untersuch Forsch 34 Barcarolo R, Tutta C and Casson P, Aroma compounds, in
195:239±245 (1992). Handbook of Food Analysis, Ed by Nollet L Marcel Dekker,
18 Leino M, LapvetelaÈinen A, Menchero P, Malm H, Kaitaranta J New York 1:1015±1049 (1996).
and Kallio H, Characterization of stored arabica and robusta 35 Reineccius GA, The Maillard reaction and coffee ¯avor, in
coffees by headspace±GC and sensory analysis. Food Qual Proceedings of the 16th International Scienti®c Colloquium on
Prefer 3:115±125 (1991/2). Coffee. ASIC, Paris, pp 249±257 (1995).
19 Semmelroch P and Grosch W, Analysis of roasted coffee powders 36 Baltes W and Bochmann G, Model reactions on roasted aroma
and brews by gas chromatography±olfactometry of headspace formation. 1. Reaction of serine and threonine with sucrose
samples. Lebensm Wiss Technol 28:310±313 (1995). under the conditions of coffee roasting and identi®cation of
20 Semmelroch P and Grosch W, Studies on character impact new coffee aroma compounds. J Agric Food Chem 35:340±346
odorants of coffee brews. J Agric Food Chem 44:537±543 (1987).
(1996). 37 Clarke RJ and Macrae R (Eds) Coffee, Vol 1. Elsevier Science,
21 Grosch W, Flavour of coffee. A review. Nahrung 42:344±350 Barking (1985).
(1998). 38 Lingle TR, Time, temperature and turbulence, in The Coffee
22 Mayer F, Czerny M and Grosch W, In¯uence of provenance and Brewing Handbook, Ed by Lingle TR. Long Beach, CA, pp 30±
roast degree on the composition of potent odorants in Arabica 37 (1996).
coffees. Eur Food Res Technol 209:242±250 (1999). 39 Ho CT, Hwang HI, Yu TH and Zhang J, An overview of the
23 Tranchant J, Manuel Pratique de Chromatographie en Phase Maillard reactions related to aroma generation in coffee, in
Gazeuse. Masson, Paris, pp 301±337 (1982). Proceedings of the 15th International Scienti®c Colloquium on
24 Kondjoyan N and Berdague JL, A Compilation of Relative Reten- Coffee. ASIC, Paris, pp 519±527 (1993).

J Sci Food Agric 82:840±847 (online: 2002) 847