monitoring of cocoa volatiles produced during roasting by selected ion flow tube-mass spectrometry...

C:Fo

odCh

emist

ry

Monitoring of Cocoa Volatiles Produced duringRoasting by Selected Ion Flow Tube-MassSpectrometry (SIFT-MS)Yang Huang and Sheryl A. Barringer

Abstract: Selected ion flow tube-mass spectrometry (SIFT-MS) was used to measure the real-time concentrations ofcocoa volatiles in the headspace during roasting. Alkalized and unalkalized Don Homero and Arriba cocoa beans wereroasted at 120, 150, and 170 ◦C in a rotary roaster. The concentrations of total alcohols, acids, aldehydes, esters, ketones,and alkylpyrazines increased, peaked, and decreased within the timeframe used for typical roasting. The concentrations ofalkylpyrazines and Strecker aldehydes increased as the roasting temperature increased from 120 to 170 ◦C. For most of thevolatile compounds, there was no significant difference between Arriba and Don Homero beans, but Arriba beans showedhigher concentrations of 2-heptanone, acetone, ethyl acetate, methylbutanal, phenylacetaldehyde, and trimethylpyrazine.For unalkalized Don Homero beans (pH 5.7), the time to peak concentration decreased from 13.5 to 7.4 min forpyrazines, and from 12.7 to 7.4 min for aldehydes as the roasting temperature increased from 120 to 170 ◦C. Also, at150 ◦C roasting, the time to peak concentration was shortened from 9 to 5.1 min for pyrazines, and from 9.1 to 5 minfor aldehydes as the pH increased from 5.7 to 8.7.

Keywords: cocoa, Maillard reaction, roasting, volatiles

Practical Application: SIFT-MS allows for real-time monitoring of the key volatile compounds contributing to chocolateflavor, with minimal sample preparation, thus can be used to facilitate adjusting the roasting conditions, such as temperatureand time, to optimize chocolate flavor during roasting. Real-time monitoring during roasting can also be used to evaluatethe flavor quality of different types of beans by comparing the concentrations of key flavor compounds.

IntroductionThe most important step in the development of cocoa flavor is

the roasting step. The typical roasted, sweet odorants of cocoa areformed, and undesired compounds with low boiling points, suchas acetic acid, are removed. Thermal treatment initiates the Mail-lard reaction between amino acids and reducing sugars in cocoabeans, which generates most of the aromatic volatiles contribut-ing to the flavor of chocolate. To date, more than 600 volatilecompounds have been identified in cocoa, including pyrazines,aldehydes, ethers, thiazoles, phenols, ketones, alcohols, furans, andesters (Dimick and Hoskin 1981). Pyrazines and aldehydes are themajor compounds formed during roasting (Heinzler and Eich-ner 1992), and they are the most important contributors to thedesirable chocolate aroma (Counet and others 2002).

Roasting conditions affect cocoa aroma quality because theMaillard reaction is affected by temperature and the time re-taining at this temperature. Despite the fact that roasting is themost important process in generating the key odorants of cocoa,the continuous change of these volatile compounds during roast-ing is unclear. Thus, real-time monitoring of the concentrations

MS 20100943 Submitted 8/20/2010, Accepted 11/4/2010. Authors are withDept. of Food Science and Technology, The Ohio State Univ., 2015 Fyffe Road,Columbus, OH 43210, U.S.A. Direct inquiries to author Barringer (E-mail:[email protected]).

of important volatiles during roasting becomes important, whichprovides direct and rapid results of the concentrations of importantvolatile compounds and aids in determining the preferred roastingtemperature and time.

Many studies have measured cocoa and chocolate volatiles usinggas chromatography-mass spectrometry. Selected ion flow tube-mass spectrometry (SIFT-MS) is a relatively new technique, whichcan also be used to study levels of volatile compounds. SIFT-MSapplies judiciously-selected positive precursor ions for chemicalionization reactions coupled with mass spectrometric detection torapidly quantify targeted volatile compounds. The volatile com-pounds can be identified and quantified in real time based on theknown rate coefficients for reaction of the reagent ions with thetarget compounds. This allows for highly sensitive, real-time anal-ysis of complex mixtures of volatile compounds without trappingor preconcentration (Spanel and Smith 1999). Thus, the prepa-ration of sample is simplified and artifacts or changes in relativeproportions are not introduced by the preconcentration steps. Dis-advantages are that the reaction kinetics of each volatile, if not al-ready published in the literature, must be carefully measured beforethe volatile can be reliably quantified. In addition, many volatilesproduce the same masses after reaction with the precursors. Whenthis occurs, the mass can be discarded as unsuitable for analysis,the results at that mass reported as a mixture, or in a few cases theconflicting masses can be subtracted out. Some knowledge of thevolatiles expected to be in the sample is required, so that volatiles

C© 2011 Institute of Food Technologists R©doi: 10.1111/j.1750-3841.2010.01984.x Vol. 76, Nr. 2, 2011 � Journal of Food Science C279Further reproduction without permission is prohibited

C:FoodChemistry

Monitoring of cocoa volatiles. . .

Table 1–Information on the volatile compounds analyzed by SIFT-MS.

Volatile compounds Precursor Ion k (10−9cm3s−1) Product m/z Product Ion Reference

Pyrazines2-methylpyrazine NO+ 2.8 94 C5H6N2

+ Syft (2009)DMP O2

+ 2.7 108 C6H8N2+ Syft (2009)

TrMP NO+ 2.5 122 C7H10N2+ Syft (2009)

TMP O2+ 2.5 136 C8H12N2

+ Syft (2009)2,3-diethyl-5-methylpyrazine (EMP) O2

+ 2.5 150 C9H14N2+ Syft (2009)

AldehydesAcetaldehyde O2

+ 2.3 44 C2H4O+ Spanel and others (1997)Methylbutanal NO+ 3.0 85 C5H9O+ Spanel and others (2002)2-methylpropanal O2

+ 3.0 72 C4H8O+ Spanel and others (2002)Benzaldehyde O2

+ 2.4 106 C7H6O+ Spanel and others (1997)Methional O2

+ 2.5 104 C4H8OS+ Syft (2009)Nonanal O2

+ 3.2 138 C10H18+ Syft (2009)

(E,E)-2,4-decadienal NO+ 4.2 151 C10H15O+ Spanel and others (2002)Vanillin NO+ 2.8 152 C8H8O3

+ Syft (2009)Heptanal NO+ 3.3 113 C7H13O+ Spanel and others (2002)(E)-2-nonenal NO+ 3.8 139 C9H15O+ Spanel and others (2002)(E)-2-octenal NO+ 4.1 125 C8H13O+ Spanel and others (2002)Phenylacetaldehyde O2

+ 2.5 120 C8H8O+ Syft (2009)

KetonesAcetone NO+ 1.2 88 C3H6O �NO+ Spanel and others (1997)Acetoin O2

+ 2.5 88 C4H8O2 Syft (2009)Acetophenone NO+ 3.6 150 C8H8O �NO+ Spanel and others (1997)2,3-butanedione NO+ 1.3 86 C4H6O2

+ Spanel and others (1997)2-heptanone NO+ 3.4 144 C7H14O �NO+ Smith and others (2003)1-octen-3-one NO+ 2.5 156 C8H14O �NO+ Syft (2009)

AlcoholsLinalool NO+ 2.6 136 C10H16

+ Syft (2009)Benzyl alcohol NO+ 2.3 107 C7H7O+ Wang and others (2004)Methanol H3O+ 2.7 33 CH5O+ Spanel and Smith (1997)Ethanol NO+ 1.2 45 C2H5O+ Spanel and Smith (1997)N-amyl alcohol NO+ 2.5 87 C5H11O+ Spanel and Smith (1997)2-phenylethanol O2

+ 2.4 92 C7H8+ Wang and others (2004)

2-heptanol NO+ 2.4 115 C7H15O+ Wang and others (2004)Maltol NO+ 2.5 156 C6H6O3

�NO+ Syft (2009)

Furans5-methylfurfural NO+ 3.1 110 C6H6O2

+ Wang and others (2004)Furaneol NO+ 2.5 128 C6H8O3

+ Syft (2009)Furfural NO+ 3.2 96 C5H4O2

+ Wang and others (2004)Sotolon NO+ 2.5 128 C6H8O3

+ Syft (2009)

Nitrogen compoundsAzine H3O+ 3.3 80 C5H5N �H+ Spanel and Smith (1998a)Indole NO+ 2.8 117 C8H7N+ Wang and others (2004)Benzonitrile NO+ 2.2 133 C6H5CN �NO+ Spanel and Smith (1998a)2-acetylpyrrole H3O+ 3.3 110 C6H7NO �H+ Syft (2009)

Sulfur compoundsDimethyl trisulfide O2

+ 2.2 126 C2H6S3+ Wang and others (2004)

Dimethyl disulphide NO+ 2.4 94 (CH3)2S2+ Spanel and Smith (1998a)

HydrocarbonsToluene NO+ 1.7 92 C7H8

+ Spanel and Smith (1998b)Myrcene NO+ 2.2 93 C7H9

+ Wang and others (2003)2-methylpentane H3O+ 2.2 85 C6H13

+ Syft (2009)Dodecane NO+ 1.5 169 C12H25

+ Spanel and Smith (1998b)Hexane O2

+ 1.7 86 C6H14+ Spanel and Smith (1998b)

PhenolsPhenol NO+ 2.0 94 C6H6O+ Spanel and Smith (1997)

EstersEthyl acetate NO+ 2.1 118 CH3COOC2H5

�NO+ Spanel and Smith (1998)Isobutyl acetate NO+ 2.2 56 C4H8

+ Francis and others (2007)2-phenethyl acetate O2

+ 2.5 104 C8H8+ Syft (2009)

Acids3-methylbutanoic acid NO+ 2.5 132 C5H10O2

�NO+ Syft (2009)Phenylacetic acid NO+ 2.5 136 C8H8O2

+ Syft (2009)Butanoic acid O2

+ 2.1 88 (CH3)2CHCOOH+ Spanel and Smith (1998)Isobutanoic acid O2

+ 2.5 88 (CH3)2CHCOOH+ Spanel and Smith (1998)Acetic acid NO+ 9.0 90 CH3COOH �NO+ Spanel and Smith (1998)

C280 Journal of Food Science � Vol. 76, Nr. 2, 2011

C:Fo

odCh

emist

ry


that produce a conflict at a given mass, but cannot possibly occurin the sample, can be eliminated from consideration.

SIFT-MS has been used in the analysis of volatiles in coffee,onion, banana (Smith and Spanel 2005), tomato (Xu and Bar-ringer 2009), garlic (Hansanugrum and Barringer 2010), jalapeno(Azcarate and Barringer 2010), and oxidation of olive oil (Davisand McEwan 2007). SIFT-MS has great potential for trace gasanalysis used to monitor food freshness, food preparation, and inthe brewing industry (Smith and Spanel 2005)

The purpose of this study was to apply SIFT-MS for real-timemonitoring of the important volatile compounds during roastinghence elucidating how roasting temperature and pH affect theformation of important cocoa volatile compounds.

Materials and Methods

Cocoa bean samplesTwo types of fermented and dried cocoa beans from Ecuador

were used in this study. One was Don Homero (ChocolateAlchemy Co., Eugene, Oreg., U.S.A.), and the other was Ar-riba (The Hershey Co., Hershey, Pa., U.S.A.). The cocoa beanswere stored at −18 ◦C in the freezer. Prior to roasting, they werethawed at room temperature for 2 h. To compare the volatiles ofcocoa beans with different pH values, the unalkalized cocoa beans(pH 5.7) were treated with alkali solution to make alkalized cocoabean samples with pH values at 7.2 and 8.7. The alkalized sampleswere prepared by soaking the thawed cocoa beans in 1 L potassiumcarbonate (Sigma-Aldrich Inc., St. Louis, Mo., U.S.A.) aqueoussolution for 3 h, then removing the alkali solution by passingthrough a screen, and drying at 70 ◦C for 3 h before roasting. Theconcentrations of the alkali solutions were 100 and 250 g/L. Thecocoa beans were weighed (300 g/batch), fed into a roaster (ModelCBR-101 Gene Cafe, Genesis, Hwasung-Si, Kyungki-Do, SouthKorea) and roasted at 120, 150, or 170 ◦C. The roasting timevaried from 15 to 60 min.

The pH values of the cocoa beans were determined by grindingthe cocoa beans into cocoa liquor. The cocoa liquor (20 g) wasthen placed into a 150-mL beaker, adding 90 mL boiling hot

distilled water with stirring, passing through filter paper, coolingto 20 ◦C, and measuring with a pH meter (Accumet model 10,Denver Instrument Co., Denver, Colo., U.S.A).

SIFT-MS analysisFor the cocoa bean samples, the sampling needle of the SIFT-MS

was inserted into the headspace of an aluminum vent pipe that wasconnected to the outlet of the roaster so that the instrument couldcontinuously detect the concentrations of the volatiles producedduring roasting. Glass wool was placed in the roaster outlet toprevent chaff from being sucked into the sampling needle. Theheating temperature for the capillary and the arm of the SIFT-MSinlet was adjusted to 180 ◦C so that the volatiles did not cool downwhen sampled into the SIFT-MS.

The in-jar cocoa bean sample was prepared by taking 50 groasted cocoa beans out of the roaster, putting them in a sealed jar,and equilibrating in a 50 ◦C waterbath for 90 min before testedby SIFT-MS.

A method for roasted cocoa volatiles was developed and im-ported into the SIFT-MS. There were 60 volatile compounds inthis method, including 5 pyrazines, 13 aldehydes, 8 ketones, 8alcohols, 5 acids, 4 furans, 3 esters, 5 nitrogen compounds, 2 phe-nols, 5 hydrocarbon, and 2 sulfur compounds. The concentrationsof volatile compounds were quantified from their reactions withthe precursor ions H3O+, NO+, or O2

+ based on known ki-netic parameters (Table 1) using the method described by Spaneland Smith (1999). The m/z values produced by reaction with 1of the 3 precursor ions were carefully chosen to avoid conflicts(different volatiles produce the same m/z value) based on pub-lished data (Table 1). The m/z values in Table 1 were chosenbecause they were not produced by other compounds in the sam-ple and, thus, uniquely measured the stated compound, with theexceptions of 2,3-, 2,5-, and 2,6-dimethylpyrazine, which werereported as dimethylpyrazine (DMP), the mixture of the 3 DMPs;as well as 2-methylbutanal and 3-methylbutanal, which were re-ported as methylbutanal, the mixture of the 2 methylbutanals.Also, compounds with irresolvable conflicts or concentrations

Figure 1–Total concentration of each volatilecategory in the headspace of unalkalized DonHomero beans for 60 min roasting at 150 ◦C.

Vol. 76, Nr. 2, 2011 � Journal of Food Science C281

C:FoodChemistry


below the detection threshold are not reported. The calibra-tion of the SIFT-MS was established by analyzing several or-ganic volatile compounds in dry air over a wide range ofpartial pressures using the method described by Smith and Spanel(2005).

Parameter settings of SIFT-MSThe parameters of the SIFT-MS was set as: inlet flow rate

120 cm3/min, scan time 3600 s for 60 min roasting, 900 s for15 min roasting, calculation delay time 5 s, product sample period100 ms, precursor sample period 20 ms, heated inlet temperature180 ◦C, carrier gas argon pressure 200 kPa, helium pressure 30psi. The flow tube vacuum pressure was 0.038 ± 0.003 Torr.

Statistical analysisTwo factors analysis of variance (ANOVA) was performed to

analyze for statistical differences. Significance was defined as P <

0.05.

Results and Discussion

Real-time concentration compared with roasting timeThe concentration of total alkylpyrazines, aldehydes, alcohols,

acids, esters, and ketones in the headspace of the roaster increasedas the roasting time increased, reached a peak concentration withinthe first 15 min then quickly decreased and leveled off (Figure 1).To validate that these results were not an artifact of the air flowin the roaster or the sampling method, cocoa beans were removed

0

20

40

60

80

100

120

140

160

180

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Con

cent

rati

on (

ppb)

Roasting time (min)

DMP-170C DMP-150C DMP-120C

0

20

40

60

80

100

120

140

160

180

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Con

cent

rati

on (

ppb)

Roasting time (min)

TrMP-170C TrMP-150CTrMP-120C

0

50

100

150

200

250

300

350

400

450

500

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Con

cent

rati

on (

ppb)

Roasting time (min)

acetaldehyde-170C acetaldehyde-150C acetaldehyde-120C

0

50

100

150

200

250

300

350

400

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Con

cent

rati

on (

ppb)

Roasting time (min)

methylbutanal-170C methylbutanal-150C methylbutanal-120C

Figure 2–DMP, TrMP, acetaldehyde, and methylbutanal in the roaster headspace at different roasting temperatures.


C:Fo

odCh

emist

ry


from the roaster every 3 min during roasting, placed in a jar, andthe headspace analyzed. In both roaster and jar headspace, thevolatiles followed the same pattern. Ramli and others (2006) alsofound a peak and fall during their roasting study.

Free amino acids and reducing sugars are the flavor precursorsthat interact during the roasting process to produce cocoa volatiles.These precursors are abundant in the fermented and dried co-coa beans at the very beginning of roasting, thus the generation

Table 2– Concentrations (ppb) of volatile compounds in theheadspace of roasted Arriba and Don Homero Ecuadorian co-coa beans at the end of 15 min roasting.

Bean type

Arriba Don Homero

Roasting temperature (◦C) 120 150 170 120 150 170

(E)-2-nonenal 3 12 15 4 9 7(E)-2-octenal 2 7 7 3 7 17(E,e)-2,4-decadienal 0 25 20 1 7 311-octen-3-one 2 16 26 0 7 232,3-butanedione 38 73 62 26 62 622,3-diethyl-5-methylpyrazine 0 2 7 0 6 122-acetylpyrrole 3 30 48 6 19 572-heptanol 0 15 32 3 9 432-heptanonea 3 107 140 3 74 762-methylpentane 7 30 73 14 34 882-methylpropanal 8 24 29 6 17 232-methylpyrazine 2 9 5 2 3 122-phenethyl acetate 2 25 36 3 17 262-phenylethanol 33 46 63 14 56 463-methylbutanoic acid 17 93 86 16 92 825-methylfurfural 2 16 26 1 9 16Acetaldehyde 58 210 285 51 199 229Acetic acid 751 4295 3779 724 3709 3914Acetoin 17 70 14 23 13 40Acetonea 112 84 78 19 58 70Acetophenone 2 20 14 3 15 22Azine 18 67 70 8 56 86Benzaldehyde 13 32 28 3 22 24Benzonitrile 1 16 23 4 11 14Benzyl alcohol 21 133 104 22 106 83Butanoic acid 26 111 21 36 20 63Dimethyl disulphide 2 10 6 2 4 14Dimethyl trisulfide 9 128 413 6 53 106Dimethylpyrazine 2 17 32 7 25 31Dodecane 6 3 5 0 0 4Ethanol 59 84 124 46 73 110Ethyl acetatea 56 109 66 34 86 47Furaneol 8 39 76 10 20 48Furfural 1 21 26 2 10 14Heptanal 5 6 7 1 8 11Hexane 10 121 158 63 115 182Indole 2 14 16 1 12 13Isobutanoic acid 17 70 14 23 13 40Isobutyl acetate 10 0 0 0 6 11Linalool 0 12 15 2 16 6Maltol 3 33 52 0 15 46Methanol 681 5030 9068 576 4338 8528Methional 2 32 45 3 22 33Methylbutanala 76 193 173 46 160 148Myrcene 12 32 66 14 24 48N-amyl alcohol 3 3 11 2 4 8Nonanal 15 19 46 14 28 23Phenol 2 12 7 3 5 16Phenylacetaldehydea 28 49 154 13 34 38Phenylacetic acid 0 15 19 2 20 8Sotolone 9 43 85 11 23 53TMP 1 4 17 0 18 13Toluene 28 28 31 2 17 30TrMPa 11 84 79 9 42 70Vanillin 1 10 11 2 11 0aCompounds with significant difference between Arriba and Don Homero beans.

rate of these volatiles is very fast. The volatiles accumulate in theheadspace of the roaster and their concentrations continuously in-crease to a peak. As the roasting proceeds, both the free aminoacids and the reducing sugars are consumed. More than 85% ofthe total free amino acid is consumed by roasting at 135 ◦C for3 min (Bonvehı and Coll 2002), and about 70% of the glucoseand fructose is consumed at 140 ◦C in 15 min (Rohan and Stew-art 1966). The rate of generation of new volatiles slows downso the concentrations decrease until a balance between the gen-eration rate and loss rate is established where the concentrationsdo not change greatly. The roaster used was a convection roasterwith highly efficient heat-transfer, thus most volatiles peaked inconcentration within the first 15 min of roasting.

The effect of roasting temperature on the volatilesAlkylpyrazines and Strecker aldehydes are important com-

pounds in cocoa volatiles. Both categories are major contributorsto chocolate flavor (Dimick and Hoskin 1999; Bonvehı and Coll2002). The concentrations of these important volatiles, includ-ing dimethylpyrazine (DMP), trimethylpyrazine (TrMP), acetalde-hyde, and methylbutanal, increased in the roaster as the roastingtemperature increased from 120 to 170 ◦C (Figure 2).

Roasting below 120 ◦C is regarded as slightly roasted (Ziegleder1982). Only a low level of alkylpyrazines (below 20 ppb) and asmall amount of aldehydes (less than 100 ppb) were produced atthis temperature. Roasting between 120 and 140 ◦C is defined asnormally roasted, between 140 and 160 ◦C is strongly roasted, andabove 160 ◦C is over-roasted (Ziegleder 1982). At 150 ◦C, thepeak concentration of DMP and TrMP was increased to 60 ppb,and the acetaldehyde and methylbutanal to 250 ppb. At 170 ◦C,the peak concentrations of the alkylpyrazines further increased to180 ppb, and the aldehydes to 350 to 450 ppb. This temperatureeffect occurs because of different pathways in the Maillard reac-tion. In stage 2 of the Maillard reaction, the Amadori product isdegraded by 1 of 3 main pathways depending on the conditions(Figure 3). The pathways are through reductones, fission products,or Schiff base of hydroxymethylfurfural (HMF). High temperaturefavors the formation of fission products such as acetol, diacetyl, andpyruvaldehyde. These compounds are very reactive and they reactwith amino acids to form Strecker aldehydes and aminoketones,which are converted via dimerization to yield pyrazines at highertemperatures (Davies and Labuza 1997). Other pathways mayalso form aldehydes and pyrazines but this is the major pathway.Therefore, high temperature favors the formation of alkylpyrazinesand Strecker aldehydes. Hashim and Chaveron (1994) also

Table 3– Roasting time (min) to peak concentration in theheadspace of Don Homero beans.

Temperature (◦C)

120 150 170

pH 5.8 7.6 8.6 5.7 7.2 8.7 5.6 7.1 8.5

Dimethylpyrazine — 11.5 — 9.6 8.0 4.8 7.9 6.9 3.1TrMP 11.8 11.1 5.9 9.8 8.3 5.7 8.2 7.9 4.1TMP 12.5 — 7.2 9.7 — 5.8 8.2 6.5 3.2Total pyrazines 13.5 11.3 5.9 9.0 8.5 5.1 7.4 7.3 3.3Acetaldehyde 12.8 9.5 3.8 9.5 7.8 5.4 7.7 7.5 7.2Methylbutanal 12.1 9.3 3.5 9.6 7.2 5.2 7.6 6.5 5.5Total aldehydes 12.7 9.0 4.1 9.1 7.7 5.0 7.4 7.5 4.0Ketones 12.6 11.2 4.1 11.4 8.6 4.9 8.0 7.6 4.4Acids 13.2 12.5 5.9 9.0 11.3 5.1 7.2 8.9 3.9Alcohols 14.8 10.5 4.9 14.0 7.8 4.1 14.0 5.7 2.5

“—” indicates concentrations are relatively constant with no clear peak.


C:FoodChemistry


demonstrated that the concentrations of all methylpyrazines, ex-cept for tetramethylpyrazine (TMP), increased linearly in relation-ship with the roasting temperature.

During roasting, DMP and TrMP had similar peak concen-trations (170 to 180 ppb) at 170 ◦C (Figure 2), and both werehigher than TMP, which was only 40 ppb at its peak concentra-

tion. DMP has 8 amino acid precursors, including alanine, va-line, leucine, phenylalanine, threonine, lysine, asparagines, andglutamate. TrMP has 6, including alanine, valine, leucine, pheny-lalanine, threonine, and glycine (Arnoldi and others 1988;Amrani-Hemaimi and others 1995), while only leucine is identi-fied as the precursor for TMP (Arnoldi and others 1988). A total

Figure 3–Maillard reaction pathways (adapted from Martins and others 2001).

Figure 4–DMP in the headspace duringreal-time roasting of Don Homero cocoa beanswith different pH values at 150 ◦C.


C:Fo

odCh

emist

ry


of 17 free amino acids, including all of these pyrazine precursors,are present in cocoa beans (Bonvehı and Coll 2002). Among them,the total content of the precursors for DMP and TrMP were 283and 191 mg/100 g separately, but the leucine content was only 29mg/100 g (Bonvehı and Coll 2002). The precursor amino acids toform DMP and TrMP are abundant while the precursor to formTMP is limited. Thus, more DMP and TrMP were formed duringroasting than TMP.

The peak concentrations of Strecker aldehydes at 170 ◦C,400 ppb for acetaldehyde and 350 ppb for methylbutanal (Fig-ure 2), were higher than that of alkylpyrazines (180 ppb), in-dicating the generation of these aldehydes is greater than thealkylpyrazines, unless the volatility of these aldehydes is higherthan the alkylpyrazines. However, alkylpyrazines have a higherHenry’s Law Constant than these aldehydes, so the concentra-tion of alkylpyrazines should be higher if both volatile compoundswere generated in equal volume. Since the alkylpyrazine concen-tration in the headspace is lower than the aldehyde concentration,the generation of these aldehydes must be larger than that of thealkylpyrazines. Alkylpyrazines are the advanced Maillard reactionproducts, while Strecker aldehydes are intermediates. These inter-mediates are generated during roasting and only a portion of themcontinue reacting with amino acids to form the alkylpyrazines oraldimines.

Type of beansArriba cocoa is from the variety Nacional, which has a full

cocoa aroma with additional floral, spicy, and green flavors, andis a premium Ecuador cocoa often used in fine dark chocolate(Fowler 2009). Don Homero cocoa is a hybrid variety (CCN51),which is not classified as a fine cocoa and is said to have acidicand harsh flavors (Afoakwa and others 2008). During roasting ofboth types of beans, the concentration of all volatile compoundsincreased as temperature increased from 120 to 170 ◦C, and thetime to peak concentration was quite similar. For most of thevolatile compounds compared, there was no significant differencebetween the 2 types of beans, but Arriba beans did have sig-nificantly higher concentrations of 2-heptanone, acetone, ethylacetate, methylbutanal, phenylacetaldehyde, and TrMP than theDon Homero beans (Table 2). 2-Heptanone has a spicy aroma(Ansorena and others 2001), acetone has a faintly aromatic, sweet-ish aroma; ethyl acetate has a fruity and sweet aroma; methylbutanalhas the aroma of chocolate, phenylacetaldehyde has a flowery andhoney aroma; and TrMP has the aroma of cocoa, roasted, andgreen (Counet and others 2002). The flowery, spicy, and greenaromas represent the typical Arriba aroma. The higher concentra-tions of 2-heptanone, acetone, ethyl acetate, phenylacetaldehyde,methylbutanal, and TrMP in the headspace above Arriba com-pared to Don Homero may suggest the flavor difference betweenthe two varieties.

Time to peak concentration: temperature and pHBoth temperature and pH affect the time to peak concentration

during roasting (Table 3). For samples with the same pH, theones roasted at higher temperatures showed a shorter time toreach the peak concentration. The time to peak concentrationof unalkalized Don Homero beans decreased from 13.5 to 7.4min for pyrazines, and from 12.7 to 7.4 min for aldehydes as theroasting temperature increased from 120 to 170 ◦C. The effectof temperature is best defined by the temperature dependence ofthe rate constant in the Arrhenius equation (Davies and Labuza1997), in which the relationship between the rate constant and

the temperature is exponential. An elevated roasting temperaturepromotes the reactions between the amino acids and the reducingsugars.

The pH of cocoa beans is another important factor affecting thetime to peak concentration during roasting because pH influencesthe reaction velocity. At the same roasting temperature, the time topeak concentration was shortened as the pH increased (Table 3).For example, at 150 ◦C roasting, the time to peak concentrationdecreased from 9 to 5.1 min for pyrazines, and from 9.1 to 5 minfor aldehydes as the pH increased from 5.7 to 8.7.

Looking at DMP, the time to peak at pH 8.7 (4.8 min) oc-curred earlier than samples at pH 7.2 (8 min) and pH 5.7 (9.6min) when roasted at 150 ◦C (Figure 4). Also, the cocoa beans atpH 8.7 showed the highest concentration followed by cocoa beansat pH 7.2 and 5.7 (Figure 4), which matches the cocoa liquor re-sults found in a previous study (Huang and Barringer 2010). Basicconditions (pH > 7) favor the formation of reducing sugars in theopen-chain configuration. It also facilitates the deprotonation ofthe amino group on the protein into the −NH2 form (Davies andLabuza 1997). Reducing sugars in the open-chain configurationand amino acids in the −NH2 form are active reactants for the firststage of the Maillard reaction (Martins and others 2001). Sincemore active reactants are available at basic conditions than acidconditions, the Maillard reaction occurred quickly and the reac-tion products, aldehydes, and pyrazines, reach peak concentrationquickly shortening the time to peak concentration. Temperatureand pH are crucial factors affecting the Maillard reaction. Hightemperature and basic conditions accelerate the Maillard reaction,thus these volatiles are accumulated in a shorter time.

ConclusionsMost volatiles, including pyrazines, aldehydes, alcohols, acids,

esters, and ketones were generated and reached maximum con-centrations within the first 15 min of roasting. The real-time con-centrations of alkylpyrazines and Strecker aldehydes increased asthe roasting temperature increased. Arriba cocoa showed higherconcentrations of some important alkylpyrazines and aldehydesthan Don Homero cocoa. The peak concentration occurs earlieras the roasting temperature or the pH of the cocoa increases.

ReferencesAfoakwa EO, Peterson A, Fowler M, Ryan A. 2008. Flavor formation and character in cocoa

and chocolate: a critical review. Crit Rev Food Sci Nutr 48:840–57.Amrani-Hemaimi M, Cerny C, Fay LB. 1995. Mechanisms of formation of alkylpyrazines in

the Maillard Reaction. J Agric Food Chem 43:2818–22.Ansorena D, Gimeno O, Astiasaran, Bello J. 2001. Analysis of volatile compound by GC-MS of

a fermented sausage: chorizo de Pamplona. Food Res Int 34:67–75.Arnoldi A, Arnoldi C, Baldi O, Griffini A. 1988. Flavor components in the Maillard reaction of

different amino acids with fructose in cocoa butter-water. Qualitative and quantitative analysisof pyrazines. J Agric Food Chem 36:988–92.

Azcarate C, Barringer SA. 2010. Effect of enzyme activity and frozen storage on jalapeno peppervolatiles by selected ion flow tube-mass spectrometry. J Food Sci 75(9):C710–21.

Bonvehı JS, Coll V. 2002. Factors affecting the formation of alkylpyrazines during roastingtreatment in natural and alkalinized cocoa powder. J Agric Food Chem 50(13): 3743–50.

Counet C, Callemien D, Ouwerx C, Collin S. 2002. Use of gas chromatography-olfactometryto identify key odorant compounds in dark chocolate: comparison of samples before and afterconching. J Agric Food Chem 50:2385–91.

Davis BM, McEwan MJ. 2007. Determination of olive oil oxidative status by selected ion flowtube mass spectrometry. J Agric Food Chem 55:3334–8.

Davies K, Labuza TP. 1997. The Maillard reaction: application to confectionery products. In:Zeigler G, editor. Confectionery science. Pennsylvania, Pa.: Penn State Univ. Press. p 35–66.

Dimick PS, Hoskin JM. 1981. Chemico-physical aspects of chocolate processing: a review. CanInst Food Sci Technol J 4:269–82.

Dimick PS, Hoskin JC. 1999. The chemistry of flavour development in chocolate. In: BeckettST, editor. Industrial chocolate manufacture and use. 3rd ed. Oxford: Blackwell Science.p 137–52.

Fowler MS. 2009. Cocoa beans: from trees to factory. In: Beckett ST, editor. Industrial chocolatemanufacture and use. 4th ed. York, U.K.: Wiley-Blackwell. p 44.

Hansanugrum A, Barringer SA. 2010. The effect of milk on the deodorization of malodorousbreath after garlic ingestion. J Food Sci 75(6):C549–58.


C:FoodChemistry


Hashim L, Chaveron H. 1994. Extraction and determination of methylpyrazines in cocoa beansusing coupled steam distillation-microdistillator. Food Res Int 17:537–44.

Heinzler M, Eichner K. 1992. The role of Amadori compounds during cocoa processing-formation of aroma compounds under roasting conditions. Z Lebensm Unters Forsch 21:445–50.

Huang Y, Barringer SA. 2010. Alkylpyrazines and other volatiles in cocoa liquors at pH 5 to 8,by selected ion flow tube-mass spectrometry (SIFT-MS). J Food Sci 75(1):121–7.

Martins SIFS, Jongen WMF, van Boekel MAJS. 2001. A review of Maillard reaction in food andimplications to kinetic modeling. Trends Food Sci Technol 11:364–73.

Ramli N, Hassan O, Said M, Samsudin W, Idris NA. 2006. Influence of roasting conditions onvolatile flavor of roasted Malaysian cocoa beans. J Food Process Pres 30:280–98.

Rohan TA, Stewart T. 1966. The precursors of chocolate aroma: changes in the sugars duringthe roasting of cocoa beans. J Food Sci 31: 206–9.

Smith D, Spanel P. 2005. Selected ion flow tube mass spectrometry (SIFT-MS) for on-line tracegas analysis. Mass Spectrom Rev 24:661–700.

Spanel P, Smith D.1998. SIFT studies of the reactions of H3O+ , NO+ and O2+ with several

ethers. Int J Mass Spectrom Ion Processes 172:239–47.

Spanel P, Smith D. 1999. Selected ion flow tube-mass spectrometry: detection and real-timemonitoring of flavours released by food products. Rapid Commun Mass Spectrom 13:585–96.

Spanel P, Ji YF, Smith D. 1997. SIFT studies of the reactions of H3O+, NO+ and O2+ with a

series of aldehydes and ketones. Int J Mass Spectrom Ion Processes 165(166):25–37.Spanel P, Van Doren J, Smith, D. 2002. A selected ion flow tube study of the reaction of H3O+,

NO+ and O2+ with saturated and usaturated aldehydes and subsequent hydration of the

productions. Int J Mass Spectrom 213:163–76.Syft Technologies Inc. 2009. Kinetics library database. Syft Technologies Inc., Christchurch,

New Zealand.Wang TS, Spanel P, Smith D. 2004. A selected ion flow tube study of the reactions of H3O+,

NO+ and O2+ with some phenols, phenyl alcohols and other carboxy compounds in support

of SIFT-MS and PTR-MS. Int J Mass Spectrom 239(2–3):139–46.Xu Y, Barringer SA. 2009. Effect of temperature on lipid-related volatile production in tomato

puree. J Agric Food Chem 57(19):9108–13.Ziegleder G. 1982. Gas chromatographic determination of roasting degree for cocoa via methy-

lated pyrazines. Dtsch Lebensm Rundsch 78:77–81.


monitoring of cocoa volatiles produced during roasting by selected ion flow tube-mass spectrometry...

Documents