Food Chemistry 141 (2013) 459–465

Contents lists available at SciVerse ScienceDirect

Food Chemistry

journal homepage: www.elsevier .com/locate / foodchem

Determination of compounds responsible for tempeh aroma

0308-8146/$ - see front matter � 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.foodchem.2013.03.047

⇑ Corresponding author. Tel.: +48 61 8487273; fax: +48 61 8487314.E-mail address: henrykj@up.poznan.pl (H. Jelen).

Henryk Jelen ⇑, Małgorzata Majcher, Alexandra Ginja, Maciej KuligowskiFaculty of Food Science and Nutrition, Poznan University of Life Sciences, Wojska Polskiego 31, 60-624 Poznan, Poland

a r t i c l e i n f o a b s t r a c t

Article history:Received 19 November 2012Received in revised form 11 February 2013Accepted 13 March 2013Available online 20 March 2013

Keywords:TempehAEDAAroma analysisFlavourGas chromatography–olfactometry GC–OOdour activity values OAVGas chromatography/mass spectrometryGC/MS

Tempeh is a fermented food, popular mainly in south-east Asia, but also among vegetarians worldwide. Itis produced by fermenting soybean or other beans with Rhizopus strains and usually eaten deep-fried,steamed or roasted. The flavour of tempeh depends upon the fermentation time, beans used and the(eventual) frying process. Our goal was to identify compounds responsible for the unique aroma of fer-mented and fried soy tempeh. Gas chromatography–olfactometry (GC–O) with the aroma extract dilutionanalysis (AEDA) approach, was used to determine key odorants after 1 and 5 days of fermentation andsubsequent frying. Comprehensive gas chromatography–mass spectrometry (GC � GC–ToF-MS) was usedfor their quantitation using stable isotope dilution analysis (SIDA) or standard addition (SA) methods.Odour activity values (OAV) were calculated for 19 out of 21 key odorants. Tempeh was fermented for5 days and fried, and the main aroma compounds were found to be the following: 2-acetyl-1-pyrroline,(FD = 1024, OAV 1380), 2-ethyl-3,5-dimethylpyrazine (FD = 512, OAV 338), dimethyl trisulfide, (FD = 512,OAV 900), methional (FD = 512, OAV 930), 2-methylpropanal (FD = 512, OAV 311) and (E,E)-2,4-decadi-enal (FD = 512, OAV 455). The frying process induced the increase or appearance of the main key odorantsin tempeh.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Tempeh (also known as tempe) is a traditional Indonesian foodproduced by the fermentation of soybeans using Rhizopus species.Apart from soybeans, other substrates used for fermentation havebeen utilised: chickpeas (Angulo-Bejarano et al., 2008), barley(Feng, Eriksson, & Schnürer, 2005) and beans (Ashenafi & Busse,1991). The tempeh of a specific flavour is produced from pressedcopra (Hachmeister & Fung, 1993). Tempeh is normally consumedfried, boiled, steamed or roasted. During the fermentation process,the enzymatic digestion of substrates, leads to an increasedamount of free aminoacids, water-soluble nitrogen compounds,free fatty acids, and to the development of characteristic flavour.During the fermentation of tempeh, a decrease in the amount ofcrude lipids is observed, as lipids serve as the main source of en-ergy for the microorganisms during the fermentation process.The protein hydrolysis may amount to 25% of the initial soy protein(Sparringa & Owens, 1999).

In addition to being a source of proteins and lipids in a vegetar-ian diet, tempeh became of interest due to its interesting nutri-tional/functional properties. It has been observed that fermentedsoybean products, such as tempeh, miso and natto, are more resis-tant to lipid oxidation than unfermented soybeans. The isoflavonelevels in tempeh are relatively high compared to other soybean

products. Raw tempeh contains the highest levels of daidzein andgenestein, compared to tofu or soybean drinks. However, the pro-cess of tempeh deep frying significantly (up to 45%) reduces the to-tal isoflavones contents (Haron, Ismail, Azlan, Shahar, & Peng,2009). The potential use of tempeh as a functional food has in-creased. Methods of production of c-aminobutyric acid enrichedtempeh, which has antihipertensive effects (Aoki, Furuya, Endo, &Fujimoto, 2003a; Aoki et al., 2003b), as well as isoflavone-enrichedtempeh (Nakajima, Nozaki, Ishihara, Ishikawa, & Tsuji, 2005) havebeen reported.

The properties, described above, of tempeh are mainly a resultof the microbial/enzymatic activity of microorganisms, used forits production and biotransformation of soy constituents. The mainfungus used for the preparation of tempeh in Indonesia is Rhizopusoligosporus, which is considered a domesticated form of Rhiyopusmicrospores (Feng, 2006). More strains identified in tempeh, in-cluded Rhizopus formosaensis, R. and Rhizopus oryzae (Babu, Bhaky-araj, & Vidhyalakshmi, 2009). It has been postulated that thenatural habitat for Rhizopus could be fresh leaves of the Hibiscusspecies, which is used to wrap cooked soybeans (Ogawa, Toku-masu, & Tubaki, 2004). Apart from Rhizopus species, various micro-organisms, including filamentous fungi, yeasts and bacteria, arefound in the traditional tempeh. Tempeh co-inoculation with lacticacid bacteria (LAB) is performed to improve the safety of the prod-uct and this process contributes also to the fermentation process(Ashenafi & Busse, 1991). The growth of LAB (Lactobacillusplantarum), coexisting with R. oligosporus, especially with the

460 H. Jelen et al. / Food Chemistry 141 (2013) 459–465

acidification of the media during the tempeh production process,could inhibit the growth of pathogenic bacteria (Feng et al., 2005).

Microorganisms are also responsible for the unique flavourproperties of tempeh. Contrary to the data on the nutritional prop-erties, microbiological aspects and raw materials used for the tem-peh production, data on the volatile compounds and especially onthe tempeh aroma are very limited (Feng, Larsen, & Schnürer,2007). Data on key odorants in this product, especially regardingraw and fried tempeh is nonexistent. The goal of this work wasto determine key odorants in fermented (raw) and fried tempeh,prepared from soybean using the strain of R. oligosporus NRRL2710 often used for its production on a commercial scale.

2. Materials and methods

2.1. Tempeh preparation

Tempeh was prepared using fungus R. oligosporus, NRRL 2710strain originating from Northern Regional Research Laboratory,Peoria Ill, USA. The inoculum, used for soy seeds inoculation, wasprepared from suspension of R. oligosporus spores cultured onPDA medium for 72 h. Seeds of soy Noviko variety, after dehullingand cooking (40 min.), were cooled down and inoculated with thespore suspension, placed on Petri dishes with a 15 cm diameterand fermented for a period of up to 5 days at 37 �C. Once fer-mented, soy beans overgrown with R. oligosporus formed into acake of the size of a Petri dish, and were fried in rapeseed oil for10 min at 170 �C.

2.2. Volatile compounds isolation

Tempeh cakes from Petri dishes were cut into approximately1 cm squares, frozen in liquid nitrogen, and ground to obtain ahomogenous sample (200 g). In the next step, 100 g of ground tem-peh was transferred into an Erlenmeyer flask and extracted sepa-rately with two solvents of different polarities – diethyl ether(200 ml) and methylene chloride (200 ml), for 2 h each. Both frac-tions were filtered and combined prior to distillation, which wasperformed using a solvent-assisted flavour evaporation (SAFE)apparatus described by Engel, Bahr, and Schieberle (1999). Duringthis procedure, the temperature of the water bath was held at40 �C, at reduced (<300 m Torr) pressure, obtained using an Ed-wards RV5 rotary vane pump. The distillate of the aroma com-pounds was collected in a flask cooled with liquid nitrogen. After30 min of distillation, the solution was dried over anhydrous Na2-

SO4, and the fraction was concentrated to about 400 ll using aKuderna Danish concentrator.

2.3. Gas chromatography–olfactometry (GC–O)

GC–O was performed using a HP 5890 chromatograph and thefollowing capillary columns: SPB-5 (30 m � 0.32 mm � 0.5 lm)and Supelcowax 10 (30 m � 0.32 mm � 0.5 lm; both columnsfrom Supelco, Bellefonte, PA). The GC was equipped with a Y split-ter, dividing effluent between an olfactometry port with humidi-fied air as a makeup flow and a flame ionisation detector. Theoperating conditions were as follows for the SPB-5 column: initialoven temperature, 40 �C (1 min), raised at 6 �C/min to 180 �C andat 20 �C/min to 280 �C. Operating conditions for the Supelcowax-10 column were as follows: initial oven temperature, 40 �C(2 min), raised to 240 �C at 8 �C/min rate, held for 2 min isother-mally. For all peaks and flavour notes, retention indices (RI) werecalculated to compare results obtained by GC–O with that obtainedby GC–MS and with literature data. Retention indices were

calculated for each compound using a homologous series of C6-C24 n-alkanes.

2.4. Gas chromatography–mass spectrometry (GC–MS)

Compound identification was performed using two instru-ments. Agilent Technologies 7890A gas chromatograph was cou-pled to a 5975C TAD quadrupole mass spectrometer of the sameproducer. This instrument was equipped with a Supelcowax-10column (30 m � 0.25 mm � 0.25 lm). Operating conditions forGC–MS were as follows: helium flow 32.2 cm/s; oven conditionswere the same as for GC–O. Mass spectra were recorded in an elec-tron impact mode (70 eV) in a scan range of m/z 33–350. Sampleswere also run on GC � GC-ToF-MS (Pegasus IV, Leco) running inboth one- and two-dimension modes to quantify odorants. TheGC was equipped with a DB-5 column (25 m � 0.2 mm � 0.33 lm)and a Supelcowax 10 (1.3 m � 0.1 mm � 0.1 lm) as a secondaryone. For the one-dimensional analysis, the secondary oven waskept at a temperature that was 30 �C higher than the first oven,for which a temperature program was used from 40 �C (1 min) at5 �C/min to 220 �C and kept for 5 min. Mass spectra were collectedat a rate of 30 scans/s, and the detector voltage was 1750 V. For thetwo-dimensional analysis, the temperature of the second oven waskept 5 �C higher than the first oven. Modulation time was opti-mised and set at 7 s, and mass spectra were collected at a rate200 scans/s.

Identification of volatiles was performed in two ways, depend-ing on the availability of standard compounds: full identificationcomprising comparison of mass spectra, retention indices (RI),and odour notes on two columns of different polarities was per-formed when the reference standard of the investigated compoundwas available. In some cases, the MS signal of the analyte was tooweak to facilitate mass spectra comparison. In these cases, RI andodour notes of the compounds were compared to a reference stan-dard. When standards were not available, tentative identificationwas performed on the basis of the comparison of the mass spec-trum of a compound with a NIST 05 library match and comparisonof retention indices with those available in the literature. Also, theodour characteristics for an analysed compound was comparedwith the literature data and used in tentative identification.

2.5. Aroma extract dilution analysis (AEDA)

The flavour dilution factor (FD) of each of the odorants wasdetermined by an AEDA method (Grosch, 1993). The flavour ex-tract (2 ll) was injected into a GC column. Odour-active regionswere detected by GC-effluent sniffing (GC–O), and three panelists(two of them with >5 year experience in GC–O analyses) deter-mined the sensory description of the volatiles. The extract wasthen diluted stepwise by the addition of diethyl ether, and eachsample of the dilution series was re-analysed until no odour wasperceivable at the sniffing port. Retention data of the compoundswere expressed as RI on both columns.

2.6. Quantification of aroma compounds

For quantification, stock solutions of internal standards of therespective isotopically labelled compounds, were prepared indiethyl ether and added to the tempeh samples in a concentrationsimilar to that of the relevant analyte present (estimated in a pre-liminary experiments by peak area comparison). The suspensionwas stirred, and volatiles were isolated as described before usingSAFE. Distillates were analysed by GC � GC-ToF-MS, monitoringthe intensities of the respective ions listed in Table 1. For SIDA ana-lysed compounds, response factors were calculated in the standardmixture of labelled and unlabeled compound in a concentration of

Table 1Quantification method and ions used for the determination of main odorants intempeh samples.

Compound Quant methoda Quant ions UL/Lb

1. Dimethyl sulfide 2H6 62/682. 2-methylpropanal SA 723. 2-and 3-methyl butanal SA 864. 2,3-butanedione 13C4 86/905. 2-butanol SA 596. hexanal SA 727. 3-methyl-1-butanol SA 558. (Z)-4-heptenal SA 849. 1-octen-3-one 2H3 70/7310. 2-acetyl-1-pyrroline 13C2 111/11311. dimethyl trisulfide 2H6 126/13212. 2-ethyl-3,5-dimethylpyrazine 2H6 135/14113. 3-(methylothio)propanal 2H3 104/10714. 2,3-diethyl-5-methylpyrazine 2H7 150/15715. (E)-2-nonenal 2H2 141/14316. (E,Z)-2,6-nonandienal SA 7017. phenylacetaldehyde 13C2 91/9318. (E,E)-2,4-decadienal SA 8119. 2-methoxy phenol 2H3 124/127

a Quantification method: Stable Isotopes Dilution Analysis (SIDA) or SA – stan-dard addition (r2 = from 0.971 to 0.994).

b Ions used for quantification, UL – unlabeled, L – labelled

H. Jelen et al. / Food Chemistry 141 (2013) 459–465 461

500 ppb. The concentrations in the sample were calculated fromthe peak area of the analyte extracted ion and its correspondinginternal labelled standard. For some compounds, where isotopi-cally labelled analogues were not available, the standard additionmethod was used for their quantitation (compounds marked SAin Table 1). The calculation was done using the Chroma TOF soft-ware (version 3.34).

3. Results and discussion

The volatile compounds isolated from tempeh represented dif-ferent chemical classes, mainly aldehydes and ketones, hydrocar-bons, mono and sesquiterpenes, sulfur containing compounds,

Fig. 1. Total ion chromatogram (A) and extracted ion chromatogram (B) of volatile compbeen shown. In chromatogram (B) the following m/z values were selected for EIC: 80, 94,(2) – methylpyrazine; (3) – 2,5-dimethylpyrazine; (4) – 2,6-dimethylpyrazine; (5) – eth3,5-dimethylpyrazine; (9) – 2-ethenyl-6-methyl pyrazine; (10) – 2,3-diethyl-5-methylp

nitrogen containing compounds, alcohols and furans to name themain fractions. As many compounds on the GC/MS chromatogramswere coeluting, for their identification and quantification purposes,and also to inspect olfactory important regions for possible coelu-tions, comprehensive gas chromatography (GC � GC-ToFMS) wasused. The vast peak capacity of the GC � GC system, increase insensitivity due to modulation, and deconvolution offers a powerfultool for the identification of compounds. All these features, helpfulin target analysis, in compounds profiling result in a huge data setsof compounds from which key odorants have to be fished out forthe data to have any value from a flavour/sensory point of view.When restrictions were applied (i.e. selection of compounds withsignal to noise ratio >500 and quantity of NIST library match>700), a total 291 compounds were identified in 1 day fried tem-peh sample, which had the least compounds. Lowering the signalto noise limit, increased the number of peaks to several hundredfor each sample. As an example, a region of GC � GC with elutingpyrazines is shown on Fig. 1. Therefore, to determine sensory ac-tive compounds in such complex mixture, gas chromatographywith olfactometry (GC–O) had to be used.

Main odorants identified on the basis of GC–O were estimatedfor 1 and 5-day old tempeh, after fermentation and subsequent fry-ing. When the aroma of fermented and fried tempeh samples wasevaluated, the following descriptors were predominant in sensoryprofile analysis of 1 and 5-day old samples: beany (soybean), oily,fried (french fries), mushroom-like, moldy, boiled potatoes andnutty. Before frying, beany was the predominant note, followedby boiled potatoes, nutty, mushroom-like and moldy. In fried tem-peh, beany along with oily and fried odours prevailed, followed bynutty, boiled potatoes, mushroom-like and moldy. Tempeh fer-mented for one day had milder aroma, whereas tempeh after5 days fermentation had a more pronounced flavour. Usually inthe process of commercial tempeh production, fermentation lastsup to two days and tempeh fermented for a longer time can beused as a flavour enhancer.

Fig. 2 shows FD values of the analysed samples. Graph A showsFD values for compounds in one day tempeh, whereas graph B

ounds of tempeh fermented for one day and fried. Region where pyrazines elute has107, 108, 112, 119, 120, 121, 133, 135, 136, 149. Pyrazines identified: (1) – pyrazine;ylpyrazine; (6) – 2-ethyl-5-methylpyrazine; (7) – trimethylpyrazine; (8) – 2-ethyl-yrazine; (11) – tetramethylpyrazine.

A

B

Fig. 2. Flavor dilution factors (FD) vs. peak numbers of main odoriferous fractions of tempeh fermented and then fried after fermentation; (A) – tempeh fermented for oneday; (B) – tempeh fermented for 5 days. Bar numbers correspond to peak numbers listed in Table 2; (1) dimethyl sulfide; (2) 2-methylpropanal; (3) 2 and 3-methyl butanal;(4) 2,3-butanedione; (5) butanol; (6) hexanal; (7) 3-methyl butanol; (8) (Z)-4-heptenal; (9) 1-octene-3-one; (10) 2-acetyl-1-pyrroline; (11) (Z)-1,5-octadienone; (12)dimethyl trisulfide; (13) 2-ethyl-3,5-dimethylpyrazine; (14) 3-(methylthio)propanal; (15) 2,3-diethyl-5-methylpyrazine; (16) (E)-2-nonenal; (17) (E,Z)-2,6-nonadienal; (18)phenylacetaldehyde; (19) unknown; (20) (E,E)-2,4-decadienal; (21) 2-methoxy phenol.

462 H. Jelen et al. / Food Chemistry 141 (2013) 459–465

shows comparison of FDs in tempeh fermented for 5 days, andfried after fermentation. Based on the AEDA method of key odor-ants identification, 12 compounds have been identified with FDvalues ranging from 8 to 512 in one day fermented tempeh. Thecompound of the highest FD factor was 2-acetyl-1-pyrroline(FD = 512), followed by dimethyl trisulfide, (Z)-4-heptenal and 1-octene-3-one (all FD = 256). High FD (128) factors were registeredalso for 2-methylpropanal and 3-(methylthio)propanal (methion-al). These compounds contributed to the popcorn-like, cabbage,rancid, mushroom, rancid and boiled potato notes, respectively.Frying the one day fermented tempeh resulted in an increase ofsome FD values, especially for 2,3-butanedione, 2-butanol, hex-anal, 2-ethyl-3,5-dimethylpyrazine, (E,Z)-2,6-nonadienal and phe-nylacetaldehyde. When the one day tempeh was fried, five moreodorants were also identified: 3-methyl-1-butanol, (Z)-1,5-octa-dienone, 2-ethyl-3,5-dimethylpyrazine, (E,Z)-2,6-nonadienal andphenylacetaldehyde. In fried tempeh, the main odorants FDs were

similar to the fermented one, except an increase in hexanal was de-tected (FD = 128 vs. 32). The contribution of the remaining com-pounds specific for fried samples was of less importance,characterised with FD values of 32 and 64. Graph B shows the FDvalues for 5 day fermented and fried tempeh: 15 odorants were de-tected after fermentation and 21 compounds after frying, andmajority of the compounds was the same as for one day tempeh,although their intensities were higher, especially noted for 2-methylpropanal, 2-acetyl-1-pyroline, dimethyltrisulfide, 2-ethyl-3,5-dimethylpyrazine and 3-methylthiopropanal. When the odourcharacteristic of compounds identified by GC–O is compared withthe sensory characteristic of tempeh being analysed, described ear-lier in this chapter, some similarities can be observed. Among thecompounds with the highest FDs influencing the aroma of tempeh,were those associated with oily, rancid, fried oil/fatty (2-methylpropanal, (Z)-4-heptenal, (E,E)-2,4-decadienal, respectively),fried/popcorn, roasted (2-acetylpyrroline and pyrazines),

Table 2Main aroma fractions of tempeh prepared from soy fermented for 1 and 5 days and analysed after fermentation and subsequent frying of fermented samples. Bolded are OAVvalues which increased significantly in a result of frying.

Compound Odour RIa Wax/DB5

OTb [lg/L]

Fermented one day Fermented 5 days

Fermented Fried Fermented Fried

Concc [lg/kg]

OAVd Concc [lg/kg]

OAVd Concc [lg/kg]

OAVd Concc [lg/kg]

OAVd

1. Dimethyl sulfide Cabbage 718/502 7,6 120 16 121 16 134 17 145 192. 2-methylpropanal Rancid 870/562 0,7 187 267 192 274 224 320 218 3113. 2 and 3-methyl butanal Malty 920/651 0,4 nd nd nd nd nd Nd 224 5604. 2,3-butanedione Buttery 980/593 15 26 2 28 2 74 5 118 85. 2-butanol Spoilage 1012/552 17000 310 <1 250 <1 3112 <1 3150 <16. Hexanal Grass 1079/801 10 212 21 3465 346 1954 195 3386 3397. 3-methyl butanol Rancid 1215/732 29930 Nd Nd 5193 <1 135 <1 13488 <18. (Z)-4-heptenal Rancid 1267/901 0,06 9 150 15 250 9.1 152 18 3009. 1-octen-3-one Mushroom 1296/980 0,04 8.7 218 12 300 8.9 223 21 52510. 2-acetyl-1-pyrroline Popcorn 1325/931 0,1 14 140 124 1240 16 160 138 138011. (Z)-1,5-octadienone Geranium 1353/983 0,0012 Nd Nd Nd Nd Nd Nd Nd Nd12. Dimethyl trisulfide Cabbage 1369/963 0,01 1.5 150 8.2 820 1.2 120 9 90013. 2-ethyl-3,5-

dimethylpyrazineRoasted 1445/1081 0,16 Nd Nd 12 75 18 90 54 338

14. 3-(methylothio)propanal Boiled potato 1449/908 0,2 12 60 68 340 136 680 186 93015. 2,3-diethyl-5methyl

pyrazineEarthy,roasted

1490/1087 0.09 Nd Nd Nd Nd 1.7 19 4,4 49

16. (E)-2-nonenal Soap, fatty 1531/1156 0,69 16 23 23 33 Nd Nd 26 3817. (E,Z)-2,6-nonandienal Cucumber 1585/1135 0,03 Nd Nd Nd Nd Nd Nd Nd Nd18. Phenylacetaldehyde Honeyowy 1641/1038 6,3 Nd Nd 215 34 618 98 624 9919. Unknown Spicy 1688/unkn – Nd Nd Nd Nd Nd Nd Nd Nd20. (E,E)-2,4-decadienal Fatty, fried 1798/1321 0,2 Nd Nd Nd Nd Nd Nd 91 45521. 2-methoxy phenol Smoke 1859/1089 1 3.8 4 4.7 5 5.1 5 5 5

a Retention Indices determined on Supelcowax-10 and DB-5 columns (see Methods section).b Odor thresholds in water (Rychlik, Schieberle, & Grosh, 1998).c Mean values based on three replicates.d Odor activity values (OAV) calculated by dividing the concentration of analyte by its odor threshold value.

H. Jelen et al. / Food Chemistry 141 (2013) 459–465 463

mushroom-like (1-octene-3-one) or boiled potatoes (3-(methyl-thio)propanal – which correspond with descriptors observed inthe sensory assessment of tempeh samples.

Table 2 shows a different approach to the determination of keyodorants. Compounds selection for odour Acivity Values (OAV) wasbased on AEDA. These compounds were quantified using SIDA orSA methods. In Table 2 odour thresholds of analysed compoundshave also been listed. The importance of analysed compounds foraroma formation, was estimated using OAV – ratios of compoundsconcentration to their odour thresholds. For 1 day fermented tem-peh, the predominant odorants were 2-methylpropanal and 1-oc-tene-3-one (OAV > 200). The frying process resulted in asignificant increase of OAV: the compound with the highest OAVwas 2-acetyl-1-pyroline (OAV 1240), followed by dimethyltrisul-fide (820), hexanal (346) and 3-(methylthio)propanal (340). Thedata in Table 2 indicates that the frying process significantly in-creased the number and amounts of sensory important com-pounds. A similar trend was observed in 5 day fermentedtempeh. The dominant compounds in fermented tempeh were also2-methylpropanal, 1-octene-3-one and 3-(methylthio)propanal,which was the compound with the highest OAV (680). After friedanalysis the tempeh, which was fermented for 5 days, revealedan increase in the majority of key odorants and as a result theirOAV values. In Table 2, bolded OAVs represent compounds that sig-nificantly increased their concentrations after the process of frying.The compound with the highest OAV in fried tempeh was 2-acetyl-1-pyroline. Its concentration increased almost nine times as a re-sult of frying, both in tempeh fermented for one and five days.Fig. 3 shows the peak of 2-acetyl-1-pyrroline together with itsmass spectrum acquired on a GC � GC-ToFMS system. Its unequiv-ocal identification, in single dimensional GC, would be more diffi-cult due to a coelution with dimethyl pyrazine. The ion m/z = 111characteristic for 2-acetyl-1-pyrroline was magnified 50 times tovisualise this compound tending to be overlapped by the dimeth-

ylpyrazine tail. The crucial importance of 2-acetyl-1-pyrroline inthe formation of fried tempeh aroma was proven using both ap-proaches – FD and OAV. Increase in the 2-acetyl-1-pyrroline con-centration after frying, definitely indicates its origin from heatgenerated reactions in fried tempeh, as suggested by Hofmannand Schieberle (1998a). Also, its presence at similarly low levels,after 1 and 5 days fermentation, shows that this process does notcontribute to the increase of its contents, and the initial low levelmay be the result of soybean boiling in preparation of tempeh,rather than in result of microbial activity. Formation of 2-acetyl-1-pyrroline in Bacillus cereus surface cultures (strain isolated fromcocoa fermentation) has been reported (Romanczyk, McClelland,Post, & Aitken, 1995). Adams and De Kimpe suggested that the for-mation of 2-acetyl-1-pyrroline was catalysed by Bacillus cereusstrains from 4-aminobutanal diethyl acetal, as a precursor by per-forming an experiment with culture medium supplementationwith this compound. However, as the supplementation was doneprior to sterilization, this reaction can be considered to have a ther-mal origin. They also suggested that 2-acetyl-1-pyrroline can beproduced via enzymatic acetylation of 1-pyrroline, originatingfrom the degradation of ornithine and proline (Adams & De Kimpe,2007).

One of the most characteristic groups of compounds in tempehwere pyrazines. Fig. 1 shows the chromatogram of a region of vol-atile compounds where pyrazines were eluted. There were no pyr-azines detected after one day fermentation, but they appearedafter the frying process which suggests they form mainly in ther-mal reactions. However, they were detectable in tempeh after5 days of fermentation. Similarly, Dajanta found 10 pyrazines infermented soybeans, whereas, in cooked, non-fermented soybeans,only three pyrazines were formed. The increase in 2,5-dimethyl-pyrazine concentration was 15-fold (Dajanta, Apichartsrangkoon,& Chukeatirote, 2011). Considering these facts, pyrazines can alsobe produced in microbially catalysed reactions. Pyrazines are

Fig. 3. Chromatogram and mass spectrum of 2-acetyl-1-pyrroline isolated fromfried tempeh sample after 1 day of fermentation. Extracted chromatogram consistsof ions of m/z = 111 and m/z = 108 (�0.1). Ion m/z = 111 is characteristic for 2-acetyl-1-pyrroline, whereas m/z = 108 is characteristic for the coelutingdimethylpyrazine.

464 H. Jelen et al. / Food Chemistry 141 (2013) 459–465

known to contribute to the characteristic roasted flavour of heatedfood products (Maga, 1992). Of all the detected pyrazines, twoplayed a role in formation of the tempeh aroma: 2,3-diethyl-5-methylpyrazine and especially 2-ethyl-3,5-dimethylpyrazine(OAV 338 in 5 days tempeh fried). 2-ethyl-3,5-dimethylpyrazineand 2,3-diethyl-5-methylpyrazine were among 36 pyrazines,formed as a result of the reaction of ascorbic acid with various ami-noacids during heating for 20 min at 160 �C (Adams & De Kimpe,2009). 2-ethyl-3,5-dimethylpyrazine was identified as an odorantof an intense (earthy) smell, together with 2,3-diethyl-5-methylpyrazine (also of earthy smell) formed in a reaction of dry heatingof ribose/cysteine, glucose/cysteine and rhamnose/cysteine mix-tures (Hofmann & Schieberle, 1998b). As suggested, 2-ethyl-3,5-dimethylpyrazine formation is independent of the sugar moietyand it can be formed from alanine and 2-oxopropanal via the Strec-ker reaction. 2-ethyl-3,5-dimethylpyrazine and 2,3-diethyl-5-methyl pyrazine were identified as important key odorants in cof-fee (Blank, Sen, & Grosch, 1992) and roasted beef (Cerny & Grosch,1993). In fried soybeans, Wilkens and Lin (1970) identified pyra-zine, 2-methylpyrazine, 2,3-dimethylpyrazine, 2,5-dimethylpyr-azine, trimethylpyrazine, 2-ethyl-5-methylpyrazine and 2-ethyl-3,6-dimethylpyrazine. The occurrence of these pyrazines in aheated product would suggest they originate from the thermalreactions. However, there are reports on pyrazines occurence infermented products via microbial formation (Rizzi, 1988). Pyra-zines are also an important group of volatiles produced by Bacillus

subtilis growth on soybean (Dajanta et al., 2011; Owens, Allagheny,Kippling, & Ames, 1997).

Soy is rich in lipid fraction, which influences the formation ofcompounds during the fermentation process – oxidation afterlypolitic activity of R. oligosporus is one of the sources of aldehydesin fermented tempeh. Many of the detected aldehydes (hexanal, 3-methylbutanal, 2-methylbutanal, (Z)-4-heptenal, 2,6-nonadienal,2,4-decadienal, phenylacetaldehyde, 2-nonenal) have been identi-fied in whole meal and white wheat flour sourdoughs (Czerny &Schieberle, 2002). 2-methyl propanal and 3-methyl-1-butanal areformed from aminoacids degradation (valine and leucine, respec-tively). The amount of 2-methyl propanal did not vary much inall samples. On the contrary, the amount of 3-methyl-1-butanalwas high after frying 5 day fermented tempeh, which resulted inits high OAV (560). The high amount of 3-methyl-1-butanol (unsig-nificant from the OAV point of view) formed in the frying process,can be attributed to the degradation to aminoacids during the ther-mal reaction, with a subsequent reduction of the aldehyde.

The frying process contributed to the increase of aldehydes inthe analysed samples. These compounds also contributed to thefatty, rancid and fried notes of tempeh prepared in such a way.The increase in amount of aldehydes is also related to the contribu-tion of rapeseed oil, used for frying as a source of aldehydes formedin the autooxidation process. Hexanal and 2,4-decadienal – autox-idation products resulting from the frying process, contributed tothe characteristic aroma of fried tempeh. (E,E) 2,4-decadienal, 2-acetyl-1-pyrroline and 1-octene-3-one were potent odorants ofroasted peanuts. 3-(methylthio)-propanal, 3-methylbutanal and2-methylbutanal played less significant roles in the roasted pea-nuts flavour formation (Chetschik, Granvogl, & Schieberle, 2008).Methional (3-(methylthio)propanal) played a significant role inthe aroma of fried, but also fermented for 5 days samples. A Strec-ker aldehyde, is a product of methionine degradation and, asshown in Table 2, heating (frying process) also increased its con-tents in tempeh.

Sulfides were also tempeh key odorants, especially dimethyl tri-sulfide, which concentration increased 6–8-fold as a result of fryingand reached OAV 820 and 900 in fried tempeh samples after 1 and5 days fermentation, respectively. Sulfides are formed during thedegradation of methionine, where methional can be a startingpoint for methanethiol formation, which subsequently yields di-methyl disulfide, and also trimethyl- and tetramethyl disulfides.Increase of dimethyltrisulfide, as a result of fermentation of soyduring production of thua nao, was noted by Dajanta and cowork-ers (Dajanta et al., 2011). DMS is a volatile compound, however itsamount did not decrease during frying, which may be attributed toits formation from S-methylmetionine during heating, as observedin heated paprika powder (Cremer & Eichner, 2000). The last groupof compounds important for the tempeh aroma, are 8-carbon alco-hols and ketones. They form a specific group of compounds charac-teristic of fungal secondary metabolism (Jelen & Wasowicz, 1998),where triglycerides and fatty acids are transformed using fungallipoxygenase/hydroperoxylyase systems. Linoleic acid is convertedto the most characteristic, mushroom smelling compound – 1-oc-tene-3-ol and a non-volatile 10-oxo-trans-8-decenoic acid. Relatedcompounds (also 1-octene-3-one) usually accompany 1-octene-3-ol in fungal cultures. Similarly to other groups of compounds theiramount increased after frying. This can be caused by the fasterthermal degradation of formed intermediates (hydroperoxides) inthermal rather than enzymatic processes.

Apart from the above discussed groups of compounds, therewere also numerous alcohols, ketones, terpenes and other com-pounds detected by GC � GC-ToFMS, however, their importancefor the flavour of tempeh is of minute importance. A study onthe formation of volatile compounds by R. oligosporus strains iso-lated from tempeh was described by Feng and colleagues (Feng

H. Jelen et al. / Food Chemistry 141 (2013) 459–465 465

et al., 2007). In their detailed study, volatiles were monitored in 10strains inoculated on malt extract agar (MEA), as well as grown onsoybean in tempeh fermentation and co-cultivated with L. planta-rum, and the profile of the volatile compounds was similar. Etha-nol, acetone, 2-butanone, and 3-methyl-1-butanol weredominant compounds produced by NRRL 2710 on soybean. Theyalso detected 3-octanone and 1-octene-3-ol, which were mush-room odour compounds from tempeh, also in soybean control,indicating activity of the lipoxygenase pathway in soybean. Fengand coworkers suggest that characteristic beany flavour com-pounds (hexanol and hexanal) were removed in the tempeh fer-mentation process. Our data is in contrast with this, as hexanalwas an abundant compound in fermented for 5 days tempeh evenbefore frying.

4. Conclusions

Analysis of the key odorants of tempeh after different fermenta-tion time and after frying, revealed several groups of sensoryimportant compounds. These groups comprised of nitrogen con-taining compounds, 2-acetyl-1-pyrroline and pyrazines, however,out of the 12 detected pyrazines only two were of major signifi-cance. Another important group was aldehydes, formed in lipoxy-genase pathways and autooxidation, the amount of aldehydesincreased significantly during the frying process, methionine deg-radation of sulfur compounds (dimethyltrisulfide) and also 1-oc-tene-3-one. Characterization of key odorants using FD wasgenerally in good accordance with OAV method results. A vastamount of identified peaks by GC � GC-ToFMS, indicate a needfor sensory guided (GC–O) analysis, in detection of volatiles impor-tant for the flavour of tempeh and also other food products.

References

Adams, A., & De Kimpe, N. (2007). Formation of pyrazines and 2-acetyl-1-pyrrolineby Bacillus cereus. Food Chemistry, 101, 1230–1238.

Adams, A., & De Kimpe, N. (2009). Formation of pyrazines from ascorbic acid andamino acids under dry-roasting conditions. Food Chemistry, 115, 1417–1423.

Angulo-Bejarano, P. I., Verdugo-Montoya, N. M., Cuevas-Rodriquez, E. O., Milán-Carillo, J., Mora-Escobedo, R., Lopez-Valenzuela, J. A., et al. (2008). Tempeh flourfrom chickpea (Cicer arietinum L.) nutritional and physicochemical properties.Food Chemistry, 106, 106–112.

Aoki, H., Furuya, Y., Endo, Y., & Fujimoto, K. (2003a). Effect of c-aminobutyric acidenriched tempeh-like fermented soybean (GABA-tempeh) on the bloodpressure of spontaneously hypertensive rats. Bioscience, Biotechnology, andBiochemistry, 68, 1806–1808.

Aoki, H., Uda, I., Tagami, K., Furuya, Y., Endo, Y., & Fujimoto, K. (2003b). Theproduction of new tempeh-like fermentem soybean containing a high level ofc-aminobutyric acid by anaerobic incubation with Rhizopus. Bioscience,Biotechnology, and Biochemistry, 67, 1018–1023.

Ashenafi, M., & Busse, M. (1991). The microflora of soak water during tempehproduction from various beans. Journal of Applied Bacteriology, 70, 334–338.

Babu, P. D., Bhakyaraj, R., & Vidhyalakshmi, R. (2009). A low cost nutritious food‘‘Tempeh’’ – A review. World Journal of Dairy & Food Sciences, 4, 22–27.

Blank, I., Sen, A., & Grosch, W. (1992). Potent odorants of the roasted powder andbrew of Arabica coffee. Zeitschrift für Lebensmittel-Untersuchung und -ForschungA, 195, 239–245.

Cerny, C., & Grosch, W. (1993). Quantification of character impact odour compoundsof roasted beef. Zeitschrift für Lebensmittel-Untersuchung und -Forschung A, 196,417–422.

Chetschik, I., Granvogl, M., & Schieberle, P. (2008). Comparison of the key aromacompounds in organically grown, raw West-African peanuts (Arachis hypogaea)and in ground pan roasted meal produced thereof. Journal of Agricultural andFood Chemistry, 56, 10237–10242.

Cremer, D. R., & Eichner, K. (2000). Formation of volatile compounds during heatingof spice paprika (Capsicum annuum) powder. Journal of Agricultural and FoodChemistry, 48, 2454–2460.

Czerny, M., & Schieberle, P. (2002). Important aroma compounds in freshly groundwholemeal and white wheat flour – identification and qualitative changesduring sourdough fermentation. Journal of Agricultural and Food Chemistry, 50,6835–6840.

Dajanta, K., Apichartsrangkoon, A., & Chukeatirote, E. (2011). Volatile profiles ofthua nao, a Thai fermented soy product. Food Chemistry, 125, 464–470.

Engel, W., Bahr, W., & Schieberle, P. (1999). Solvent assisted flavour evaporation – anew versatile technique for the careful and direct isolation of aromacompounds from complex food matrices. European Food Research andTechnology, 209, 237–241.

Feng, X. M., Eriksson, R. B. A., & Schnürer, J. (2005). Growth of lactic acid bacteriaand Rhizopus oligosporus during barley tempeh fermentation. InternationalJournal of Food Microbiology, 104, 249–256.

Feng X. M. (2006). Microbial dynamics during barley tempeh fermentation. DoctoralThesis. Swedish University of Agricultural Sciences, Uppsala, ISBN:91-576-7108-7.

Feng, X. M., Larsen, T. O., & Schnürer, J. (2007). Production of volatile compounds byRhizopus oligosporus during soybean and barley tempeh fermentation.International Journal of Food Microbiology, 113, 133–141.

Grosch, W. (1993). Detection of potent odorants in foods by aroma extract dilutionanalysis. Trends in Food Science & Technology, 4, 68–73.

Hachmeister, K. A., & Fung, D. Y. C. (1993). Tempeh: A mold-modified indigenousfermented food made from soybeans and cereal grains. Critical Reviews inMicrobiology, 19, 137–188.

Haron, H., Ismail, A., Azlan, A., Shahar, S., & Peng, L. S. (2009). Daidzein andgenestein contents in tempeh and selected soy products. Food Chemistry, 115,1350–1356.

Hofmann, T., & Schieberle, P. (1998a). 2-oxopropanal, hydroxyl-2-propanone, and1-pyrroline – important intermediates in the generation of the roast-smellingfood flavor compounds 2-acetyl-1-pyrroline and 2-acetyltetrahydropyridine.Journal of Agricultural and Food Chemistry, 46, 2270–2277.

Hofmann, T., & Schieberle, P. (1998b). Identification of key aroma compoundsgenerated from cysteine and carbohydrates under roasting conditions.Zeitschrift für Lebensmittel-Untersuchung und -Forschung A, 207, 229–236.

Jelen, H., & Wasowicz, E. (1998). Volatile fungal metabolites and their relation to thespoilage of agricultural commodities. Food Reviews International, 14, 391–426.

Maga, J. A. (1992). Pyrazine update. Food Reviews International, 8, 479–558.Nakajima, N., Nozaki, N., Ishihara, K., Ishikawa, A., & Tsuji, H. (2005). Analysis of

isoflavone content in tempeh, a fermented soybean, and preparation of a newisoflavone-enriched tempeh. Journal of Bioscience and Bioengineering, 100,685–687.

Ogawa, Y., Tokumasu, S., & Tubaki, K. (2004). An original habitat of tempeh molds.Mycoscience, 45, 271–276.

Owens, J. D., Allagheny, N., Kippling, G., & Ames, J. M. (1997). Formation of volatilecompounds during Bacillus subtilis fermentation of soya beans. Journal of theScience of Food and Agriculture, 74, 132–140.

Rizzi, G. P. (1988). The biogenesis of food related pyrazines. Food ReviewsInternational, 5, 375–400.

Romanczyk, L. J., McClelland, C. A., Post, L. S., & Aitken, W. M. (1995). Formation of 2-acetyl-1-pyrroline by several Bacillus cereus strains isolated from cocoafermentation boxes. Journal of Agricultural and Food Chemistry, 43, 469–475.

Rychlik, M., Schieberle, P., & Grosh, W. (1998). Compilation of odor thresholds, odorqualities and retention indices of key food odorants. Garching, Germany: DeutscheForschungsanstalt für Lebensmittelchemie.

Sparringa, R. A., & Owens, J. D. (1999). Protein utilization during soybean tempehfermentation. Journal of Agricultural and Food Chemistry, 47, 4375–4378.

Wilkens, W. F., & Lin, F. M. (1970). Volatile compounds of deep fat-fried soybeans.Journal of Agricultural and Food Chemistry, 18, 337–339.