Food Sci. Technol. Res., 9 (3), 264–270, 2003
Chemical and Sensory Changes in Flavor of Roux Prepared from Wheat Flour and Butter by Heating to Various Temperatures
Yukie KATO
Faculty of Education, Toyama University, 3190 Gofuku, Toyama-shi, Toyama 930-8555, Japan
Received November 25, 2002; Accepted April 30, 2003
The flavor of an unheated wheat flour and butter mixture (roux 0) and roux samples heated to between 100˚C and 180˚C at 20˚C intervals (roux I–V) were examined by chemical analyses of the aroma components with gas chro- matography (GC) and GC-MS and sensory evaluation of the roux samples. In the chemical analysis, it was deter- mined that large quantitative changes of aroma components with increased heating temperature occurred in the functional groups of ketones, carboxylic acids, furans and pyrazines. Furthermore, a cluster analysis showed that the difference in characteristic flavor between roux II (white roux) and roux IV–V (brown roux) depended on the quite different compositions of their components. On the other hand, sensory analysis of the roux showed that roux II–IV (heated to 120˚C–160˚C), namely white and brown roux, were highly evaluated for buttery and sweet attributes as well as odor preference. After the relationship between chemical and sensory changes in the roux flavor with increased heating temperature was surveyed by a principal component analysis, a correlation analysis clarified that the sensory odor preference correlated significantly with ketones (mainly methyl ketones) (r 0.900, p 0.05) or with cyclic ketoenols such as maltol (r 0.838, p 0.05). Those compounds also showed a high positive correlation with the evalua- tion for the sweet attribute. Therefore, methyl ketones or cyclic ketoenols are assumed to play an important role as indicators of the pleasant odor preference in roux heated to various temperatures.
Keywords: roux, flavor, cooking temperature, chemical composition, sensory evaluation, correlation analysis
Roux is prepared from wheat flour and butter, and is often the range of 100˚C–140˚C (Kato, 1992; Kato, 1996), several used in cooking to thicken a soup or sauce without heating, components remained to be identified or confirmed. Therefore, although the odor of flour remains to create an off flavor. When the changes in aroma composition in white and brown roux with the mixture of wheat flour and butter (or oil) is heated however, it increased heating temperature have not yet been clarified, and produces a characteristic pleasant flavor. The flavor profile varies there is still a need for a sensory evaluation of roux at various with the heating time and final temperature. When roux is pre- heating temperatures. pared at a temperature of 120–130˚C, its color does not change In this study, changes in the aroma components of unheated much, but it has a pleasant milk-like flavor, and is called white roux and roux heated in the range of 100–180˚C and changes in roux. However, when roux is prepared by heating to a higher the sensory evaluation of the roux samples were characterized. temperature like 160–180˚C, the color usually turns brown and a We also studied the correlation between the changes in aroma burnt flavor results, this is called brown roux. Both white and components and the sensory evaluation of the roux samples to brown roux are used as the base for various sauces and soups to determine on what the preferable odor in the white or brown provide the required viscosity and flavor. Although some studies roux depends. on the viscosity and sensory attributes of sauces have been reported (Hatae et al., 1979; Osawa & Nakahama, 1973; Kato, Materials and Methods 1995), studies on the flavor of sauces have been limited. Since Materials and chemicals Wheat flour (soft flour, Nisshin the mixture of wheat flour and butter contains carbohydrate, pro- Flour Co., Japan) and butter (Yukijirushi Dairy Co.) were pur- tein and lipid as the main constituents, it is reasonable to assume chased from a local market. Standard chemicals were obtained that it would generate a characteristic flavor from the Maillard from commercial sources. reaction during cooking. There are many reports on heated fla- Preparation of the roux Wheat flour (30 g) and butter (30 vor, including model systems of the Maillard reaction (Shieberle g) were mixed in an aluminum pan and heated in an electric & Hofmann, 1996; Bruechert et al., 1988; Cambero et al., 1992), cooker (300 W) until the final temperature reached 100, 120, but there has been little work on samples prepared by these ac- 140, 160 or 180˚C (Kato, 1992; Kato, 1996). Heating was tual cooking conditions, apart from work by the authors. Previ- stopped after 7–8 min when the mixture reached at 100˚C and ous reports have described that many aroma components have many tiny bubbles started rising (roux I). When heated for 10–11 been identified in brown roux heated to 160–180˚C by GC and min to about 120˚C, the mixture became more viscous and emit- GC-MS (Kato, 1993; Kato, 2001), but in roux samples heated to ted a pleasant aroma (roux II). The color changed little and it remained a white roux. When heated for 13–14 min to about E-mail: [email protected] 140˚C, the mixture took on a slightly brown color and a stronger Chemical and Sensory Changes in Flavor of Roux 265 aroma was emitted (roux III). Heating for another 16–19 min Table 1. Yields of aroma concentrates from the roux samples heated to from about 160˚C to 180˚C turned the roux dark brown and a various temperatures. Final heating Yield burnt odor was recognized (roux IV and V). The unheated mix- Sample Odor character temperature ( g/10 g of materialsa)) ture of materials was used as the control (roux 0) to examine the Roux 0 Unheatedb) 140.0 buttery change in aroma components and in odor by heating. Roux I 100˚C 160.0 buttery, milk-like Chemical analysis A 1-L round flask containing each Roux II 120˚C 315.0 milk-like, sweet roux sample prepared from 60 g of materials, 440 ml of purified Roux III 140˚C 310.0 milk-like, sweet Roux IV 160˚C 364.0 burnt, caramel-like water and 0.5 ml aqueous solution of 0.25 mg/ml internal stan- Roux V 180˚C 702.0 burnt dard (tridecane or 2,4,6-trimethyl pyridine) was attached to a a) The roux ingredients were wheat flour and butter (weight ratio, 1 : 1). b rotary evaporator (Tanchotikul & Hsieh, 1991). The volatile ) Control. components from each roux sample (roux 0–V) were collected by steam distillation under reduced pressure (20–24 mm of Hg) at 50˚C for 100 min (Kato, 1992). This procedure was conducted five times for use as the analytical sample. Analytical samples mean value and S.D. (%) of each of ten functional groups calcu- were then prepared with five replications to obtain samples 1–5 lated from the amounts of the components with five replicates for each roux (roux 0–V). A total of 30 samples (roux 0–V 5 were also calculated as done for each component, and the differ- replicates) were therefore prepared for the chemical analyses. ence between the mean values was analyzed by Student’s t-test. GC (gas chromatography) analysis was carried out with a Shi- In the sensory test, the evaluation values by 28 panelists were madzu model 12 gas chromatograph equipped with a flame ion- also estimated using the same statistical procedure as the chemi- ization detector (FID). The column was a CBP 20M capillary cal analysis. Relationship between the chemical and the sensory type 50 m in length 0.25 mm i.d. used previously (Kato, 2001). analysis was examined by a principal component analysis and GC-MS was a HP 5890 series II gas chromatograph coupled to a correlation analysis. The statistical analyses were conducted HP model 5972-mass spectrometer. The column was a 60 m using multi-statistics (Social Survey Research Information Co., 0.25 mm i.d. (0.25 m df) DB wax fused silica type (J&W Scien- Tokyo). tific) run under the same conditions as those for the GC analysis (Kato, 1996; Kato, 2001). Each component was identified by Results and Discussion comparing it with the GC retention index (Kovats index (KI)) Odor profile and yield of the aroma concentrate The roux and MS data to those of authentic specimens. samples released characteristic odors during heating. The odor Sensory analysis In the sensory test, the roux samples character and yield of each aroma concentrate in roux 0–V are (roux 0–V) were prepared under the same heating conditions as shown in Table 1. Although roux I (heated to 100˚C) retained the in the chemical analysis. Approximately 10 g roux samples in fairly strong buttery odor like that of the unheated material, roux teacups of approximately 150 ml capacity were presented to II–III (heated to 120˚C–140˚C) had a milk-like and fairly strong members of a panel. The samples were covered with saran-rap sweet odor, and roux IV–V (heated to 160˚C–180˚C) had a burnt and aluminum foil to seal in the flavor and conceal the color of and sweet odor. roux, then warmed to a constant temperature of 50˚C in an oven. The amount of the aroma concentrate in roux I increased Twenty-eight semi-trained female students majoring in home slightly, but when heated to above 120˚C, the amount of this con- economics participated as panelists. They were presented roux centrate increased markedly compared with that in the unheated samples and asked about the intensity of four characteristic mixture. The aroma concentrate yields from roux II–IV (the attributes (buttery, sweet, aromatic and burnt odor) and the grade white roux and earlier brown roux) were about double or more of odor preference as an overall acceptance evaluation of roux (approximately 315–364 g/10 g), that of the unheated mixture flavor. These evaluation items were chosen based on a prelimi- (approximately 140 g/10 g), while that from roux V (the brown nary test and previous papers (Kato, 1992; Kato, 1993). The odor roux) increased dramatically and was about five times the intensity of the four attributes was evaluated on a scale of 1 (not amount of the unheated mixture. detectable) to 5 (extremely detectable), and the odor preference Chemical changes in the aroma components of the roux for roux samples was done on a scale from 1 (no preference) to 5 samples Many aroma compounds were detected in each roux (strong preference). After panelists evaluated the flavor of the sample and were identified by comparing their mass spectra and two roux samples, they were permitted to rest for approximately KI values with those of reference or authentic compounds. Sixty- 10 min. Then they evaluated the flavor of two other roux sam- three aroma components were selected from the compounds ples, and the flavor of the two remaining roux samples flavor identified at more than two heating temperature stages in the with sample order was approximately balanced between the ses- roux samples. The amount of each component is presented as the sions. mean and S.D. of the concentration ( g/10 g) of five replicates Statistical analysis The amount ( g/10 g) of each compo- and summarized in Table 2. These sixty-three components could nent was calculated using an internal standard, GC peak area % be classified into ten functional groups: eleven were hydrocar- and yield of the aroma concentrate, and the mean value and stan- bons, four were aliphatic alcohols, six were aldehydes, five were dard deviation ( g/10 g) of each component with five replicates ketones, eight were carboxylic acids, three were lactones, six in each roux were calculated. In addition, using the mean value were furans, four were cyclic ketoenols, eleven were pyrazines, of each component (relative GC peak area %) as a variable, a and five were other nitrogen-containing compounds. Further- cluster analysis was conducted to examine the degree of similar- more, the composition of the ten functional groups was calcu- ity among the 30 samples from roux 0 and roux I–V. Further, the lated from the total amount of the components in the roux, and 266 Y. K ATO
Table 2. Amounts of the aroma components in the roux heated to various temperatures (n 5). No. Componenta) KIb) 0c) IIIIII IV V Hydrocarbons 4 Dodecane 1249 0.4 0.3d) 1.0 0.6 0.7 0.9 0.6 0.3 0.8 0.6 1.0 0.7 17 Tetradecane 1406 0.5 0.4 0.3 0.2 11.5 9.5 0.8 0.6 0.6 0.6 0.6 0.4 e) 19 Hydrocarbon (C15) 1453 0.4 0.1 0.4 0.3 3.7 2.2 0.4 0.3 0.7 0.7 0.7 0.6 22 Pentadecane 1496 1.1 0.9 1.7 0.8 11.7 4.2 1.6 1.3 1.9 0.6 2.3 1.3 31 Hexadecane 1625 0.7 0.4 1.3 1.6 0.5 0.8 0.2 0.2 0.4 0.4 0.9 0.6 34 Heptadecane 1664 1.3 1.9 0.5 0.4 0.2 0.3 0.9 0.6 1.8 1.2 0.3 0.4 40 Octadecane 1802 0.6 0.3 1.5 0.4 2.0 1.8 1.7 1.7 2.1 2.8 2.9 1.5 44 Nonadecane 1893 3.8 2.2 0.7 0.1 1.9 1.8 2.4 3.3 1.9 2.9 0.2 0.5 e) 45 Hydrocarbon (C20) 1923 4.2 0.9 3.5 1.3 13.8 4.3 7.8 3.8 8.0 3.2 3.7 2.3 51 Eicosane 1998 0.2 0.2 0.1 0.2 2.0 4.2 0.6 1.1 0.1 0.2 0.7 1.3 59 Tricosane 2328 0.9 0.9 2.1 1.2 1.0 0.9 1.5 1.1 1.7 1.6 0.4 0.5 Alcohols 11 1-Hexanol 1362 2.0 1.0 2.2 1.1 5.8 2.5 1.5 0.2 1.1 1.5 0.4 0.5 27 1-Octanol 1565 0.4 0.4 1.0 0.4 1.0 0.3 1.3 0.3 0.8 0.5 2.6 0.5 38 1-Decanol 1770 0.1 0.1 0.7 0.7 0.8 1.6 0.9 1.0 0.2 0.3 0.5 0.6 46 2-Phenyl ethanol 1963 1.5 0.9 0.4 0.4 1.1 1.3 3.0 1.6 1.2 0.2 2.7 2.4 Aldehydes 11-Hexanal 1086 0.7 0.3 1.4 1.3 1.2 1.2 0.9 0.6 1.6 0.7 1.9 0.5 31-Heptanal 1190 0.2 0.4 0.2 0.5 0.6 0.7 1.2 0.9 1.7 0.8 1.6 1.0 61-Octanal 1288 0.1 0.2 0.2 0.1 1.8 2.6 0.8 0.5 1.1 0.3 2.7 2.5 14 1-Nonanal 1395 0.7 0.6 0.7 0.4 4.5 1.9 2.5 1.3 1.4 1.0 2.9 1.3 26 Benzaldehyde 1523 0.9 0.6 1.2 0.5 1.4 1.2 1.8 0.9 5.3 1.5 5.7 3.0 37 1-Dodecanal 1711 3.7 0.8 5.2 1.9 8.3 2.7 6.6 1.5 8.6 3.7 6.8 1.9 Ketones 22-Heptanone 1187 0.3 0.4 6.2 0.9 9.9 3.2 25.6 6.4 24.0 6.2 2.4 0.1 13 2-Nonanone 1392 0.8 0.5 8.4 2.0 17.2 3.1 46.8 5.4 35.3 5.6 5.4 2.2 29 2-Undecanone 1597 1.7 0.4 7.5 0.9 13.8 2.1 41.7 3.2 31.9 4.7 7.6 3.0 41 2-Tridecanone 1812 1.1 0.6 1.9 0.4 4.1 1.9 14.3 2.4 12.2 4.3 6.3 3.9 53 2-Pentadecanone 2024 0.3 0.1 0.3 0.3 1.1 0.5 4.6 0.7 3.0 2.1 0.9 1.0 Carboxylic acids 20 Acetic acid 1477 0.9 0.4 1.1 0.9 2.0 0.8 1.7 0.9 2.6 0.7 12.1 2.3 33 Butanoic acid 1654 1.0 0.7 1.2 0.3 1.3 0.7 1.5 1.2 3.8 0.8 5.6 0.7 43 Hexanoic acid 1870 4.1 1.5 6.9 1.2 4.0 1.1 4.1 3.3 5.7 3.2 7.7 4.7 47 2-Ethylhexanoic acid 1970 1.8 1.0 2.2 0.5 2.2 1.4 2.5 0.8 3.2 2.4 1.6 1.4 55 Octanoic acid 2082 36.2 5.1 35.9 4.6 38.2 4.5 24.9 5.9 23.3 7.3 10.4 3.6 56 Nonanoic acid 2189 1.5 0.8 4.2 4.9 6.8 1.8 5.1 0.3 2.3 5.1 5.6 3.8 58 Decanoic acid 2297 22.5 3.0 17.0 1.6 16.9 2.9 16.5 1.8 11.9 2.8 5.2 1.7 62 Dodecanoic acid 2503 4.4 1.0 4.5 1.8 5.0 1.5 6.3 3.2 1.5 1.5 7.5 5.5 Lactones 49 -Octalactone 1983 0.9 0.6 0.8 0.4 1.6 0.9 2.0 1.5 3.6 1.7 7.9 2.0 57 -Decalactone 2207 7.0 1.1 6.8 1.6 7.3 2.6 9.7 1.6 7.5 2.2 3.8 1.8 61 -Dodecalactone 2433 2.8 0.6 2.4 0.7 2.2 0.9 2.7 1.2 2.5 1.4 1.6 1.1 Furans 21 Furfural 1489 0.6 0.2 1.7 0.4 3.5 1.4 1.6 1.1 1.9 1.4 13.0 6.8 24 2-Furfuryl formatee) 1505 — — — d 2.9 5.1 1.8 1.0 25 2-Acetylfran 1515 — — 0.9 0.8 0.7 0.5 1.8 1.9 3.2 0.2 28 5-Methyl-2-furfural 1581 — — 0.4 0.5 1.1 0.3 1.2 0.5 6.4 2.2 35 Furfuryl alcohol 1675 — — 2.0 0.5 9.4 0.9 61.1 6.9 320.2 25.9 63 5-Hydroxymethyl-2-furfural 2515 — — d 3.5 1.0 2.5 2.0 0.6 0.7 Cyclic ketoenols 39 2(5H)Furanone 1767 — — 1.9 1.4 1.0 1.3 0.5 0.4 0.8 0.6 42 Cyclotenef) 1843 — — — 0.8 0.6 0.4 0.1 0.7 0.3 48 Maltolg) 1980 — — 0.8 1.0 2.1 0.5 6.0 2.5 5.8 2.6 60 4,5-Dihydro-5-propyl-2(3H)- furanonee) 2400 — — d 1.5 2.0 0.7 0.7 1.0 1.0 Pyrazines 5 Methylpyrazine 1272 — — 0.8 0.6 0.9 0.7 7.9 2.5 31.6 3.4 72,5-Dimethylpyrazine 1326 — — 1.0 1.0 0.9 0.7 1.3 0.3 3.5 1.5 82,6-Dimethylpyrazine 1333 — — 1.1 0.8 1.0 1.3 1.4 0.7 7.9 2.5 92-Ethylpyrazine 1340 — — 0.4 0.6 0.4 0.3 2.8 0.6 11.0 1.7 10 2,3-Dimethylpyrazine 1350 — — 1.1 1.4 1.0 0.1 1.4 0.4 8.9 0.8 12 2-Ethyl-6-methylpyrazine 1386 — — 0.1 0.3 0.2 0.1 0.5 0.4 4.5 0.6 15 2-Ethyl-5-methylpyrazine 1397 — — 1.0 1.2 0.9 0.7 1.8 1.3 1.7 1.5 16 2-Ethyl-3-methylpyrazine 1403 — — 0.3 0.3 0.3 0.4 0.7 0.4 3.1 1.5 18 Trimetylpyrazine 1442 — — 0.4 0.4 0.3 0.2 1.5 0.5 4.1 1.0 23 2-Ethenyl-6-methylpyrazine 1498 — — — d 1.7 1.1 0.9 0.9 32 2-Acetylpyrazine 1644 — — — d 0.9 0.4 2.5 0.3 Other nitrogen-containing compounds 30 2-Acetylpyridine 1604 — — 2.2 1.4 1.9 0.9 1.7 1.0 0.7 0.4 36 1-Acetyl-2-methylpyrrolee) 1702 — — — d 0.2 0.2 0.9 0.6 50 2-Acetylpyrrole 1998 — — 0.1 0.2 0.1 0.2 0.9 1.1 5.3 2.6 52 6-Qinazolinole) 2002 — — — — 2.3 1.0 2.1 1.1 54 1H-Pyrrole-2-carboxaldehyde 2046 — — 0.2 0.5 0.1 0.2 0.6 0.4 5.3 1.1 d, detected; —, not detected. a) The components are listed in order of their retention times according to the functional groups. The aroma components in ketones, carboxylic acids and furans are also showed in Fig. 1–3. b) Kovats index of samples. c) These show samples: 0, unheated materials; I, roux of 100˚C; II, roux of 120˚C; III, roux of 140˚C; IV, roux of 160˚C; V, roux of 180˚C. d) Values in the table are mean standard deviation ( g/10 g of materials) from five rep- licates of each roux. e) Tentatively identified. f ) 2-Hydroxy-3-methyl-2-cyclopenten-1-one. g) 3-Hydroxy-2-methyl-4H-pyran-4-one. Chemical and Sensory Changes in Flavor of Roux 267 the mean value and S.D. with five replicates were calculated and Five aliphatic aldehydes (C6–C12) and benzaldehyde were shown in Table 3. Each roux had a different composition in terms detected. Although the influence of the heating temperature was of functional groups. The contents of carboxylic acids in Table 3 low for these aldehydes as shown in Table 2, they had a strong accounted for the highest proportion in roux 0–II and ketones in odor; benzaldehyde is especially known to have a pleasant roux II–IV, while high proportions of furans were predominant in almond, nutty aroma (Chung & Cadwallader, 1993). roux IV and V heated to a high temperature. The quantitative The carboxylic acids also seem to have mainly originated changes of the components in ketones, carboxylic acids and from butter, as reported in previous paper (Kato, 1996). The furans are presented in Fig. 1, Fig. 2 and Fig. 3, respectively. amounts of octanoic acid and decanoic acid were very high, even Straight-chain hydrocarbons, alcohols, aldehydes and ketones in the unheated roux, at 36.2 and 22.5 g/10 g, respectively are known to be common thermal oxidation products of lipids (Table 2 and Fig. 2). However, these two acids have been gradu- (Scanlan et al., 1968). As shown in Fig. 1, the identified aliphatic ally degraded. In contrast, the low molecular weight acids acetic ketones were almost all of the methyl type with a carbon number acid and butanoic acid seem to have increased with increased from 7 to 15, which have a fruity and floral odor as described in higher temperature of the roux, the increase in acetic acid volatile of strawberry jam (Guichard et al., 1991). They marked- amount being statistically significant between roux IV and roux ly increased with increase in heating temperature, reaching a V. Acetic acid seemed to give the roux a stimulating flavor. The maximum at 140˚C, and constituting more than 40% of all com- flavor of the brown roux samples (roux IV and V) thus seemed to ponents shown in Table 3, where 2-nonanone and 2-undecanone be more stimulating than that of the white roux samples (roux II). showed high amounts of approximately 46.8 and 41.7 g/10 g, Three lactones, -octalactone, -decalactone, -dodecalactone respectively. It has been reported that methyl ketones in heated were identified. As shown in Table 2 and Table 3, their quantity milk were generated from -keto-acids contained in milk, so it is was not high and the total amounts did not change much with reasonable that those in the roux samples were generated from heating temperature. Lactones exhibit a characteristic buttery butter (Wong & Patton, 1963). odor, so those with carboxylic acids are believed to be involved in the basic odor of the roux (Kato, 1996).
Fig. 1. Change in the amount of each component in ketones. Different let- Fig. 2. Change in the amount of each component in carboxylic acids. Dif- ters in each component show a significant difference at the level of p 0.05 ferent letters in each component show a significant difference at the level of between the two roux with increased heating temperature. p 0.05 between the two roux with increased heating temperature.
Table 3. Composition of the functional groups in the roux heated to various temperatures (n 5). Roux Functional groups 0 (Unheated) I (100˚C) II (120˚C) III (140˚C) IV (160˚C) V (180˚C) Hydrocarbons 10.03 3.43a 8.27 2.09a 15.48 6.25a 5.96 2.27b 5.54 1.95b 1.96 0.84c Alcohols 2.87 0.35a 2.74 1.11a 3.76 0.73a 2.17 0.57b 0.89 0.29c 0.89 0.37c Aldehydes 4.51 1.17a 5.53 0.89a 5.65 1.95a 4.45 0.78a 5.40 0.68a 3.08 1.00b Ketones 2.93 1.12a 15.20 2.35b 14.53 3.19b 42.91 4.85c 29.25 5.81d 3.29 1.18e Carboxylic acids 51.65 4.94a 45.64 2.49a 24.16 2.79b 20.10 4.18b 14.90 3.67b 7.37 1.60c Lactones 7.61 1.53a 6.22 1.16a 3.54 0.99b 4.63 1.25b 3.69 0.91b 1.89 0.27c Furans 0.46 0.17a 1.03 0.24b 2.17 0.45c 5.25 0.43d 19.50 2.34e 49.15 4.49f Cyclic ketoenols 0.00 0.00 0.00 0.00 0.86 0.62a 1.75 0.72a 2.07 1.04a 1.18 0.85a Pyrazines 0.00 0.00 0.00 0.00 2.66 1.21a 2.50 1.04a 6.47 1.62b 11.46 1.50c N-compounds 0.00 0.00 0.00 0.00 0.09 0.14a 0.34 0.08a 1.09 0.55b 1.92 0.81c Others 19.92 2.08 15.34 2.71 28.12 1.49 10.23 3.04 13.23 2.71 17.80 4.34 a) The composition of the functional groups was calculated from the amount ( g/10 g) of each component in the roux. The value in each functional group is mean standard deviation calculated from five replicates of each roux. b) Different letters on each roux in each functional group show a significant difference at the level of p 0.05 between the two roux with increased heating temperature. 268 Y. K ATO
Furans, cyclic ketoenols and nitrogen-containing compounds supposition that they were generated during the heating process. like pyrazines and pyrroles were only detected in roux heated to Almost all the alkylpyrazines here have a fairly strong burnt odor above 120˚C, except for furfural, as shown in Table 2. These and characteristic nutty and roasted aroma as reported by Maga compounds are well-known Maillard reaction products found in and Sizer (1973), so these nitrogen compounds would have processed foods. The furans of 5-methyl-2-furfural, furfuryl alco- played important roles as potent odorants in the heated roux, hol and 5-hydroxymethyl-2-furfural (Fig. 3) are known to be despite only being present in small quantities. derived from Amadori rearrangement products of the Maillard To summarize, the aroma of the roux samples was changed reaction (Shigematsu et al., 1977). markedly depending on the final heating temperature. The flavor In the furans, furfuryl alcohol was increased markedly with of roux II heated to 120˚C (white roux) seemed to be character- heating temperature to reach 61.1 g/10 g in roux IV (heated to ized by overlaying the slight roasted odor on the basic buttery 160˚C) and 320.2 g/10 g in roux V (heated to 180˚C) as shown odor of the aldehydes, carboxylic acids and lactones, while the in Table 2 and Fig. 3. A significant difference in mean value was flavor of roux IV–V heated to 160˚C–180˚C (brown roux) shown between all combinations of roux pairs among II, III, IV seemed to have been strongly influenced by the Amadori rear- and V (Fig. 3). This seems to have contributed to the odor of V, rangement products: furans, cyclic ketoenols and nitrogen-con- because it constituted the largest quantity in roux V as shown in taining compounds. Table 2. Similarity among the roux samples by a cluster analysis Cyclotene and maltol in cyclic ketoenoles were identified in The degree of resemblance in the composition of the aroma com- the roux samples (Table 2). They are known to be Amadori rear- ponents among the roux samples (roux 0–V), was determined by rangement products as well as the furans already described a cluster analysis using the GC peak area % of each of the thirty- (Blank et al., 1996). As they have often been noted for their very five components; selection was based on a preliminary principal characteristic sweet odor in cooked food (Baltes et al., 1989), component analysis from the sixty-three components described they seem to have contributed greatly to the heated roux. in the experimental section and information on the contribution Various kinds of pyrazines were identified in roux II–V. They of each component to the roux flavor. The tree-shaped figure in were not detected inunheated sample or in roux I (heated to Fig. 4 shows items 1–5, 6–10, 11–15, 16–20, 20–25 and 26–30 100˚C), but increased heating temperature raised their amounts representing roux 0, roux I, roux II, roux III, roux IV and roux V, as shown in Table 2. They were markedly higher in roux IV respectively. The items of roux samples could be clearly sepa- (heated to 160˚C) and roux V (heated to 180˚C), confirming the rated into two clusters of 1–20 and 21–30. This first cluster includes roux 0 and roux I–III and constitutes the white roux.
Fig. 4. Cluster analysis based on content of characteristic each of thirty- Fig. 3. Change in the amount of each component in furans. Different let- five components in the roux samples heated to various temperatures. Items ters in each component show a significant difference at the level of p 0.05 1–5, 6–10, 11–15, 16–20, 21–25 and 26–30 represent roux 0, roux I, roux II, between the two roux with increased heating temperature. roux III, roux IV and roux V, respectively.
Table 4. Sensory evaluation for characteristic odor attributes and odor preference of the roux (n 28). Rouxa) Item 0 (Unheated) I (100˚C) II (120˚C) III (140˚C) IV (160˚C) V (180˚C) Butteryb) 3.21 1.34a 3.46 1.35a 3.71 1.01a 4.11 0.79a 3.79 0.69a 2.46 1.20b Sweetb) 2.07 1.15a 2.32 1.06a 3.43 1.00b 4.04 0.84c 3.71 1.05c 2.11 1.17d Aromaticb) 1.29 0.46a 1.71 0.94b 2.36 1.16c 3.50 1.00d 3.57 1.07d 3.57 1.29d Burntb) 1.04 0.19a 1.29 0.66a 1.21 0.42a 1.68 0.82b 1.86 0.78b 3.96 1.17c Preferencec) 1.82 0.95a 2.29 0.98a 3.36 1.13b 4.14 1.08c 3.89 0.88c 2.32 1.06d a) Roux samples were the same roux as chemically analyzed. b) Rating scale for odor attributes: 1, not detectable; 3, somewhat detectable; 5, extremely detectable. c) Rating scale for odor preference: 1, no preference; 3, some preference; 5, strong preference. d) Different letters on each roux in each item show a significant difference at the level of p 0.05 between the two roux with increased heating temperature. Chemical and Sensory Changes in Flavor of Roux 269
Fig. 6. Correlation between the content of aroma component group in the Fig. 5. Principal component analysis of the functional groups in the roux roux samples and the sensory evaluation for sweet attribute. The content of the aroma components is the mean value of the functional group shown in aroma and the sensory evaluation for the roux flavor. Figure characters: 1, Hydrocarbons; 2, Alcohols; 3, Aldehydes; 4, Ketones; 5, Carboxylic acids; Table 3. *p 0.05. 6, Lactones; 7, Furans; 8, Cyclic ketoenols; 9, Pyrazines; 10, Other nitrogen- containing compounds; A, Buttery; B,.Sweet; C, Aromatic; D, Burnt; E, Odor preference.
The second cluster includes roux IV and roux V and constitutes brown roux. The first cluster showed separation into two clusters, roux 0–II (items 1–5, 6–10 and 11–15) and roux III (items 16– 20). Roux III (heated to 140˚C), which has a thin yellow color, might have differed a little in the composition of the aroma com- ponents from roux 0–II, which had no coloring. On the other hand, the brown roux samples IV and V are clearly separated, with little variation in the five replicates of the samples. It is pre- sumed that the large proportion of furfuryl alcohol (about 50%) masked other minor differences in the replicates. These results have shown how the flavor in each roux is affected by a small difference of only 20˚C in the final heating temperature. Sensory changes in the odor of the roux samples The results of sensory evaluations for unheated materials (roux 0) and Fig. 7. Correlation between the content of aroma component group in the roux I–V are shown in Table 4. The buttery odor was the stron- roux samples and the sensory evaluation for odor preference. The content of gest of all attributes in roux 0–roux IV, but not in roux V; there- the aroma components is the mean value of the functional group shown in fore, this odor seems to be the basic flavor in not only unheated Table 3. *p 0.05. materials but also heated roux. The sweet odor was evaluated highly in roux II–IV (heated to 120˚C–160˚C), and roux III (heated to 140˚C) was given the highest evaluation for this at- tribute of all the samples. The aromatic odor was evaluated high- 4 was determined by a principal component analysis and a corre- ly in roux III–V when heated to a temperature higher than 140˚C, lation analysis. and the burnt odor was evaluated significantly higher in roux V. In Fig. 5, the factor scores and the loadings of the chemical As an overall acceptance evaluation for the roux, roux II–IV and sensory items for the first principal component (PC 1) and (heated to 120˚C–160˚C) were found to be preferable to the other the second principal component (PC 2) are illustrated. PC 1, roux in odor (roux 0–1 and roux V). Specifically, roux III (heated which accounted for 57.0% of the variation, seemed to separate to 140˚C) with a thin yellow color was found the most desirable the chemical and sensory items by the final heating temperature, of all roux samples. and PC 2, accounting for 33.2%, seemed to separate those items Relationship between chemical and sensory changes in the by preference of roux flavor. As B (sweet attribute), E (odor pref- roux samples Studies on the correlation between chemical and erence) and 4 (ketones) were not greatly different and a short dis- sensory data have been reported (Guichard et al., 1991; Refs- tance from 8 (cyclic ketoenols), those items were found to gaard et al., 1996). In this paper, the relationship between the correlate highly. Based on the result of the principal component content (%) of each functional group (1–10) as shown in Table 3 analysis, a correlation analysis was conducted. In Fig. 6, the sen- and sensory evaluation of each attribute (A–E) as shown in Table sory evaluation for the sweet attribute correlated with the methyl 270 Y. K ATO ketones significantly (r 0.898, p 0.05), and with the cyclic Changes in wheat flour proteins during the process of cooking. ketoenols cyclotene and maltol, positively (r 0.755) but, in con- Nihon Kasei Gakkaishi, 30, 441–445 (in Japanese). trast, negatively with the carboxylic acids (r 0.461). We then Kato, Y. (1992). Effect of cooking temperature on the flavor of roux (part 1): Sensory perception of roux flavor and aroma components examined whether the sensory odor preference correlated highly of roux heated to 120˚C. Nihon Kasei Gakkaishi, 43, 871–877 (in with the compounds generated by heating. The results showed Japanese). that the odor preference had a significantly positive correlation Kato, Y. (1993). Effect of cooking temperature on the flavor of roux with the methyl ketones (r 0.900, p 0.05) and cyclic ketoenols (part 2): Flavor of roux heated to 160˚C. Nihon Kasei Gakkaishi, 44, 787–792 (in Japanese). (r 0.838, p 0.05), each of which showed a high positive corre- Kato, Y. (1995). Effects of cooking temperature on the flow properties lation with the sweet attribute, but had a low correlation with the and sensory evaluation of roux and sauce. J. Cookery Sci. Jpn., 28, furans (r 0.089) and pyrazines (r 0.308), as shown in Fig. 7. 167–172 (in Japanese). Therefore, the ketones (methyl ketones) and the cyclic ketoenols Kato, Y. (1996). Flavor of a wheat flour and butter mixture, and of are assumed to play a role as an indicator of the pleasant odor roux heated to 100˚C. Nihon Kasei Gakkaishi, 47, 231–236 (in Jap- anese). preference for the roux heated to various temperatures. Kato, Y. (2001). Flavor of brown roux: Volatile components of roux heated to 180˚C. Nihon Kasei Gakkaishi, 52, 627–633 (in Japanese). Acknowledgment The author wishes to thank Dr. K. Kubota in Food Maga, J.A. and Sizer, C.E. (1973). Pyrazines in Foods. A Review. J. Chemistry at Ochanomizu University for technical assistance and support Agric. Food Chem., 21, 22–30. in using GC-MS. Osawa, H. and Nakahama, N. (1973). Physical properties of white sauce. Nihon Kasei Gakkaishi, 24, 359–366 (in Japanese). References Refsgaard, H.H.F., Srockhoff, P.M. and Jensen, B. (1996). Sensory and Baltes, W., Kunert-Kirchoff, J. and Reese, G. (1989). Model reactions chemical changes of salmon during frozen storage. In “Flavour Sci- on generation of thermal aroma compounds. In “Thermal Genera- ence Recent Developments: Section 6-Sensory Methods in Flavour,” tion of Aromas,” ed. by T.H. Parliment, R.J. McGorrin, C-T, Ho. ed. by A.J. Taylor, and D.S. Mottram. The Royal Society of Chem- The American Chemical Society, Washington, DC, 409, pp. 143– istry, Cambridge, UK, pp. 344–347. 155. Scanlan, R.A., Lindsay, R.C., Libbey, L.M. and Day, E.A. (1968). Blank, I., Devaud, S. and Fay, L.B. (1996). New aspects of the forma- Heat-induced volatile compounds in milk. J. Dairy Sci., 51, 1001– tion of 3(2H)-furanones through the Maillard reaction. In “ Flavour 1007. Science Recent Developments: Section 4-Thermally Generated Fla- Shieberle, P. and Hofmann, T. (1996). Identification of the key odor- vour,” ed. by A.J. Taylor, and, D.S. Mottram. The Royal Society of ants in processed ribose-cysteine Maillard mixtures by instrumental Chemistry, Cambridge, UK, pp. 188–193. analysis and sensory studies. In “Flavour Science Recent Develop- Bruechert, L.J., Zhang, Y., Haung, T.-C., Hartman, T.G., Rosen, R.T. ments: Section 4-Thermally Generated Flavour,” ed. by A.J. Taylor, and Ho, C.-T. (1988). Contribution of lipids to volatiles generation and D.S. Mottram. The Royal Society of Chemistry, Cambridge, in extruded corn-based model systems. J. Food Sci., 53, 1444–1447. UK, pp. 175–181. Cambero, M.I., Seuss, I. and Honikel, K.O. (1992). Flavor compounds Shigematsu, H., Shibata, S., Kurata, T., Kato, H. and Fujimaki, M. of beef broth as affected by cooking temperature. J. Food Sci., 57, (1977). Thermal degradation products of several Amadori com- 1285–1290. pounds. Agric. Biol. Chem., 41, 2377–2385 Chung, H.Y. and Cadwallader, K.R. (1993). Volatile components in Tanchotikul, U. and Hsieh, T. C.-Y. (1991). Analysis of volatile flavor blue crab (Callinectes sapidus) meat and processing by-product. J. components in steamed rangia clam by dynamic headspace sam- Food Sci., 58, 1203–1208. pling and simultaneous distillation and extraction. J. Food Sci., 56, Guichard, E., Issanchou, S., Descourvieres, A. and Etievant, P. (1991). 327–331. Pectin concentration, molecular weight and degree of esterification: Wong, N.P. and Patton, S. (1963). Identification of some volatile com- Influence on volatile composition and sensory characteristics of pounds related to the flavor of milk and cream. J. Dairy Sci., 45, strawberry jam. J. Food Sci., 56, 1621–1627. 724–728. Hatae, K., Shimada, A. and Yoshimatu, F. (1979). Studies on roux: