Online measurement of volatile organic compounds released during roasting of cocoa beans

    On-line measurement of volatile organic compounds released during roasting of cocoa beans Jana Diab, Romy Hertz–Sch¨unemann, Thorsten Streibel, Ralf Zimmermann PII: DOI: Reference:

S0963-9969(14)00296-8 doi: 10.1016/j.foodres.2014.04.047 FRIN 5233

To appear in:

Food Research International

Received date: Revised date: Accepted date:

8 January 2014 16 April 2014 19 April 2014

Please cite this article as: Diab, J., Hertz–Sch¨ unemann, R., Streibel, T. & Zimmermann, R., On-line measurement of volatile organic compounds released during roasting of cocoa beans, Food Research International (2014), doi: 10.1016/j.foodres.2014.04.047

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT On-line

measurement

of

volatile

organic

compounds released during roasting of cocoa

IP

T

beans

SC R

Authors

MA

JMSC –Joint mass spectrometry centre (UR) University of Rostock Chair of analytical chemistry Dr.-Lorenz Weg 1 18059 Rostock

NU

Jana Diab (UR), Romy Hertz–Schünemann (UR) Thorsten Streibel* (UR), Ralf Zimmermann (UR, HMGU)

TE

D

JMSC –Joint mass spectrometry centre (HMGU) Helmholtz Zentrum München-German Research Center for Environmental Health (GmbH) CMA – Comprehensive Molecular Analytics Postfach 1129 D-85758 Neuherberg *corresponding author: [email protected]

AC

CE P

Keywords: cocoa; roasting gases; photo ionization; mass spectrometry; in- and outside the bean; volatiles

ACCEPTED MANUSCRIPT Abstract

AC

CE P

TE

D

MA

NU

SC R

IP

T

The gaseous species evolving during the roasting of cocoa beans were analyzed by a newly developed technique, a micro-probe – photo-ionization-Time of flight mass spectrometer system. A lot of volatile roasting products and typical aroma compounds, such as methylbutanal, phenylacetaldehyde, and pyrazines could be on-line detected. The differences between the roast gas composition in- and outside the bean were pointed out. Two additional signals at m/z 266 and 294 with Resonance-enhanced multi-photon ionization (REMPI), which argues for an aromatic character, and a higher methanethiol signal with Single photon ionization (SPI) could have been identified at first glance for the measurements inside the bean. Principle component analysis (PCA) and difference spectra revealed and confirmed these findings. Higher concentrations of theobromine and fatty acids are present outside and more caffeine inside the bean. With both ionization techniques the changes in the PCA are more prominent at the beginning of the roast courses which are mostly explained by caffeine and theobromine. Following the courses of these characteristic compounds showed the same results as obtained with the PCA and revealed that caffeine is always released before theobromine. The new analytical technique proved to be a fast, reliable and non-labor intensive method which could therefore be a practical technique for the on-line analysis and monitoring of the cocoa roasting process.

ACCEPTED MANUSCRIPT Introduction

AC

CE P

TE

D

MA

NU

SC R

IP

T

The composition of cocoa beans is highly variable, depending on growing conditions, genetics, postharvest fermentation, drying, handling during shipment and storage (Afoakwa, Paterson, Fowler, & Ryan, 2008; Humston, Knowles, McShea, & Synovec, 2010). The quality and the flavor of the end product (chocolate bars, cocoa-based beverages) depends on one hand on this composition of the raw material, but on the other hand also on post-harvest treatment and on the processing conditions of these treatments (Cambrai et al., 2010; Humston et al., 2010; Owusu, Petersen, & Heimdal, 2012; Torres-Moreno, Tarrega, Costell, & Blanch, 2012). Main compound classes in raw beans are polyphenols, specifically catechins (flavan-3-ols), proanthocyanidins and anthocyanins which provide precursors for flavor formation. They are already present in the fresh bean and are also formed during subsequent treatment by enzymatic reactions (Afoakwa et al., 2008; Krysiak, 2006; Oliviero, Capuano, Cämmerer, & Fogliano, 2009). Post-harvest treatment includes fermentation, drying and roasting, in which fermentation and roasting are the most important steps of the manufacturing treatment (Ramli, Hassan, Said, Samsudin, & Idris, 2006). Flavor compounds are formed during the roasting step from precursors formed during fermentation and drying such as free amino acids, short-chain peptides (provided by proteolysis of enzymes) and reducing sugars, revealing a direct relationship between these two steps (Afoakwa et al., 2008; Bailey, 1962; Frauendorfer & Schieberle, 2008). Important factors are the content of polyphenols, polysaccharides and proteins in the raw material. During roasting nonenzymatic browning (Maillard reaction) takes place. Thereby, the reaction between reducing sugars and amino acids plays the major role. Typical Maillard reaction products include dicarbonyls (e.g. diacetyl), heterocyclic compounds (e.g. pyrazines, pyroles, pyridines, furans, thiazoles), aldehydes formed by Strecker-degradation (e.g. phenylacetaldehyde, benzaldehyde), ketones, esters, alcohols and phenolic compounds (Bailey, 1962; Baltes, 1980; Cambrai et al., 2010; Frauendorfer & Schieberle, 2008). Furthermore, acidity is reduced during roasting by release of volatile acids (such as acetic acid) (Afoakwa et al., 2008). The beans get their final texture and color due to oxidation, condensation and complexation of polyphenolic compounds (Krysiak, 2006; Oliviero et al., 2009; Vitzthum, Werkhoff, & Hubert, 1975). Wrong handling during all manufacturing processes leads to off-flavors, for example moisture damage leads to growth of bacteria and mold. (Humston et al., 2010). Most studies comprise the analysis of key flavor compounds or constituents of roasted cocoa beans or the end products, such as chocolate, by using steam- or high vacuum distillation or solvent extraction or a combination of both, to excavate the aroma- and Maillard reaction compounds. The extraction or distillation is mostly followed by a Gas Chromatography - Mass Spectrometry (GC-MS) analysis or an Aroma Extract Dilution Analysis (AEDA) (Cambrai et al., 2010; Counet, Callemien, Ouwerx, & Collin, 2002; Frauendorfer & Schieberle, 2008; Gill, Macleod, & Moreau, 1984; Ramli et al., 2006; Schieberle, 1999; Schnermann & Schieberle, 1997; Van Praag, Stein, & Tibbetts, 1968; Vitzthum et al., 1975; Ziegleder & Stojacic, 1988). Since the identification of Linalool and fatty acid esters as constituents of cocoa by Bainbridge et al., 1912 (Bainbridge & Davies, 1912), a lot of studies dealt with the identification of volatile compounds of cocoa and their products (Gill et al., 1984; Humston, Zhang, Brabeck, McShea, & Synovec, 2009; Ramli et al., 2006; Van Der Wal, Kettenes, Stoffelsma, Sipma, & Semper, 1971; Van Praag et al., 1968; Vitzthum et al., 1975). Most of these compounds, released by unroasted and roasted beans, have been identified by using solvent extraction techniques followed by gas chromatography, solid phase micro extraction followed by comprehensive 2D gas chromatography coupled with time-of-flight mass spectrometry (SPME-GCxGC-TOFMS) and solvent extraction followed by gas chromatography coupled to mass spectrometry (SE-GC-MS) (Gill et al., 1984; Humston et al., 2009; Ramli et al., 2006). These techniques therefore helped to reveal the complexity of the aroma of roasted cocoa beans. Major compound classes are pyrazines, alcohols, aldehydes, esters, amines, acids, ketones and hydrocarbons (Afoakwa et al., 2008; Gill et al., 1984), whereby aldehydes and pyrazines are the major compounds formed during roasting and the most important contributors to the chocolate aroma (Vitzthum et al., 1975). Key flavor-active compounds are 3-methylbutanal, 2-methylbutanal, 2- and 3methylbutanoic acid, phenylacetaldehyde, 2-ethyl-3,5-dimethylpyrazine, 1-octen-3-one, 2,3-diethyl-5methylpyrazine, 2-nonenal, 2-methyl-3-(methyldithio)furan and 2-phenylethanol, with 2-

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

SC R

IP

T

methylbutanal and 3-methylbutanal contributing most to the chocolate character (Afoakwa et al., 2008; Frauendorfer & Schieberle, 2008; Schieberle, 1999; Schnermann & Schieberle, 1997). Since most techniques use extraction methods, such as solvent extraction, high vacuum distillation or steam distillation to extract the roast products from the bean (Vitzthum et al., 1975), the extractable products of roasted cocoa beans are well analyzed. However, a continuous real-time on-line monitoring of the process of cocoa roasting is not possible with such extractive off-line techniques, but require a fast analytical method with high time resolution. Such an approach could be very helpful to get a better understanding of the dynamics of flavor generation and could form the basis for future quality control measures. Photo Ionization Time of Flight Mass spectrometry (PI-TOFMS) is a potential tool for the investigation of cocoa roasting. On the one hand, photo ionization is a very soft (almost fragment free) technique for generating ions in mass spectrometry, i.e. only molecular ions are obtained and bulk gases such as oxygen or nitrogen are not ionized, which leads to an increase in sensitivity and avoids interferences or saturation effects at the detector. In addition, depending on the actual realization of photo ionization a certain selectivity of analyzed compounds is achievable. Resonance Enhanced Multi-Photon ionization (REMPI), where molecules are ionized by absorption of at least two UV-photons, is very selective for aromatic and poly-aromatic compounds. Single photon ionization (SPI) employing VUV-photons on the other hand (is more universal and suitable for a wide range of organic compounds, but excluding all species with higher ionization energy than the applied wavelength. Time-of-flight mass spectrometry enables fast detection of ionized components of complex gaseous mixtures, yielding a complete mass spectrum with every ionization event. In previous studies, PI-TOFMS has been combined with a novel sampling approach that applied a microprobe connected to a transfer capillary leading to the ion source of the mass spectrometer (R. Hertz-Schünemann, Dorfner, R., Yeretzian, C., Streibel T., Zimmermann R., 2013; R. HertzSchünemann, Streibel, Ehlert, & Zimmermann, 2013; Hertz, Streibel, Liu, McAdam, & Zimmermann, 2012). The tip of the microprobe could be inserted in a single coffee bean, thus sampling the evolved compounds generated during the roasting process directly at the place of formation. The aim of this work was to apply this experimental approach to determine the volatile compounds released during the roasting of cocoa beans. Furthermore the roast products formed inside the cocoa bean are compared to the volatile roast products measured outside the cocoa bean.

ACCEPTED MANUSCRIPT Material and Methods

T

Roasting and Sampling

AC

CE P

TE

D

MA

NU

SC R

IP

The investigated cocoa beans are organic raw beans from the Dominican Republic. Figure 1 shows a scheme of the experimental setup of the measurements. It consists of a newly designed µ-probe sampling device and a PI-TOFMS system. The µ-probe was originally designed for the analysis of cigarette smoke inside a burning cigarette (Hertz et al., 2012). It consists of a stainless steel capillary (ID 0.2 mm, OD 0.4 mm), One end sticks out (4 mm) and can be inserted into the cocoa bean and the other end is located inside a heated aluminum base body and is connected to the transfer capillary via a capillary union. The µ-probe was now applied to the analysis of the processes inside and outside a cocoa bean during roasting. Thereby, compounds formed during the cocoa roasting process are analyzed in two different ways by means of the µ-probe sampling device. In both cases the roasting process was simulated with a heating gun at 250 °C. For the measurements outside the bean, the bean was roasted inside a glass to avoid the dilution of the roast gases (Figure 1a). For the measurements inside the bean, the micro probe was inserted directly into the bean via a 5 mm deep whole, drilled with a 1mm drill (see Figure 1b). The hole in the sample was sealed not vacuum tight with inorganic glue based on zirconium oxide. The temperature was controlled via a thermocouple placed on the surface of the bean. At least 5 replicates were conducted for every distinct experimental setup. The µ-probe is heated continuously via heating cartridges to avoid condensation of volatile compounds. The released compounds were transferred to the ion source of the mass spectrometer via a continuously heated transfer line (deactivated stainless steel, OD 530 µm, ID 280 µm) to avoid condensation and plugging of the capillary. To guarantee the continuous heating, the capillary is located inside a heating hose (2 m, 250°C). The techniques of SPI and REMPI have already been described in detail elsewhere (Boesl, 2000; Butcher, 1999; Dorfner, 2004; F. Mühlberger, Zimmermann, R., Kettrup, A., 2001), therefore only a brief description is given in the following. The PI-TOFMS instrument is equipped with a 10 Hz Nd:YAG-laser (Continuum Inc., Santa Clara, CA, USA) which provides 335 nm and 226 nm pulses via harmonic generation for the photo ionization. The 226 nm beam is directly used for REMPI while the 335 pulse is used to generate the 118 nm for SPI via a third harmonic generation (THG) gas cell filled with xenon. The mass spectrometer is a reflectron TOFMS (Kaesdorf Instrumente, Munich, Germany). The PI-TOFMS system has also been described in detail in the literature (R. Hertz-Schünemann et al., 2013; Hertz et al., 2012; F. Mühlberger, Hafner, Kaesdorf, Ferge, & Zimmermann, 2004; F. Mühlberger, Zimmermann, R., Kettrup, A., 2001).

Principle component analysis (PCA) PCA was applied to reveal differences between the measurements inside and outside the cocoa bean and to analyze the course of roasting. It was performed by the “Unscrambler” program (CAMO Software AS). The datasets were baseline corrected and normalized to the total ion current before importing them to the software. For the comparison of the measurements inside and outside the bean, all spectra of 15 minutes measurement time were averaged for each replicate. For tracing of the roasting courses the data of the mean values of all measurements in- or outside the bean were averaged every 90 seconds.

ACCEPTED MANUSCRIPT Results

T

Analysis of roast gases outside the bean

AC

CE P

TE

D

MA

NU

SC R

IP

A great variety of compounds released during the roasting process is depicted in Figure 2. Two of the main compounds released during the roasting of cocoa are theobromine (m/z 180) and caffeine (m/z 194). Besides these two alkaloids, further typical cocoa products can be identified with SPI-TOFMS. Key aroma compounds, such as methylbutanal (m/z 86), octenone (m/z 126), nonenal (m/z 140) and 2methyl-3(methyldithio)furan (m/z 160) can be identified (Frauendorfer & Schieberle, 2008; Schieberle, 1999; Schnermann & Schieberle, 1997). A summary of the major m/z ratios and the corresponding assumptions of contributing structures, measured with SPI, can be found in Table1. One has to note that signals in the spectrum cannot be assigned exclusively to one compound due to the mass-resolution of the time-of-flight mass spectrometer being too low. The m/z ratio of 86, for example could also be diacetyl, a typical Maillard reaction product and known component of cocoa roast gases and chocolate (Counet et al., 2002; Huang & Barringer, 2011; Van Praag et al., 1968), instead of methylbutanal or it could be a mixture of both. Huang et al., 2011 found slightly higher amounts for methylbutanal (46-193 ppb) than for diacetyl (26-73 ppb) which could mean that methylbutanal dominates the mixture at m/z 86. Therefore, all assignments of compounds in Table 1 were carried out in accordance with the literature and the characteristics of the ionization technique used and all possible compounds are listed. For the assignment in the text also quantities from the literature are included. For example the m/z ratio of 136 is referred to as tetramethylpyrazine, because of its higher amounts (9-84 ppb) compared to phenylacetic acid (2-20 ppb) (Huang & Barringer, 2011). Some pyrazines that contribute to the chocolate aroma, such as ethyl-dimethylpyrazine (m/z 136) and diethyl-methylpyrazine (m/z 150) can be found. Pyrazines have been identified as aroma compounds in many heated food, such as roasted coffee, broiled meat, roasted peanuts and of course roasted cocoa beans (Baltes, 1980; Mason, Johnson, & Hamming, 1966; Ramli et al., 2006; Ziegleder, 1982). Beside pyrazines, aldehydes are the main compound class in roasted cocoa beans(Gill et al., 1984). Methylbutanal, propanal, butanal and many other adehydes were found with single photon ionization. As mentioned in the introduction most of the aroma compounds are formed via the Maillard reaction. Therefore a lot of other typical Maillard reaction products can be identified either. The m/z ratios 56 (propenal or butene), 58 (propanal or acetone), 72 (butanone or butanal), 84 (pentenone or methylbutanal), 86 (pentanal or methylbutanal), 114 (heptanone or others), 126 (octenal or octenone) and 128 (octanal or others) can be assigned to aldehydes and/or ketones. A lot of heterocyclic compounds and their derivatives can be detected as well: Pyrrole (m/z 67) (methylpyrrole (m/z 81), ethyl-/dimethylpyrrole (m/z 95), acetylpyrrole (m/z 109)), Furane (m/z 68) (methylfurane (m/z 82), ethyl-/dimethylfurane (m/z 96), acetylfurane (m/z 110)) and pyrrolidine (m/z 91) (methylpyrrolidine (m/z 85), ethyl-dimethylpyrrolidine (m/z 99)). These heterocyclic compounds are also Maillard reaction-products. For example pyrazines are formed via the Strecker degradation, a by-reaction of the Maillard reaction. A degradation of the chain of the sugar molecule, because of dehydration and enolisation leads to the formation of α-carbonyls, which can react with amino acids to form besides aldehydes, better known as Strecker-aldehydes, α-aminoketones (Rainer Cremer & Eichner, 2000; Zhang, 1992; Ziegleder, 1982). These α-aminoketones dimerise to form pyrazines. Furans, another compound class of heated food are predominately formed via the caramelization of sugars. Fatty acids are known components of cocoa beans (Belitz, Grosch, & Schieberle, 2001), but their presence in the roast gas was surprising. Palmitic acid (m/z 256), linoleic acid (m/z 280), oleic acid (m/z 282) and stearic acid (m/z 284) were present in the gases evolving during the cocoa roast process. Table 1: Assigned compounds (main contribution) for the m/z ratios observed by Single photon ionization, according to literature and previous studies using this ionization technique (Afoakwa et al., 2008; Dorfner, 2004; Gill et al., 1984; Huang & Barringer, 2011; Milder, 2005; Ramli et al., 2006; Van Der Wal et al., 1971; Van Praag et al., 1968; Vitzthum et al., 1975)

ACCEPTED MANUSCRIPT literature data with concentration information 51-285 ppb [3]

m/z

assigned compound (main contribution)

44

acetaldehyde*, propane

48

methanethiol

108

ethylpyrrolidine, dimethylpyrrolidine dimethylpyrazine*, cresol*, benzyl alcohol*, …

56

butene, propenal

109

acetylpyrrole*

58

acetone*, propanal, …

110

1-26 ppb [3]// 0.39 µg/g [4]

59

Propylamine

112

5-methyl-2-furfural*, acetylfuran methyl furfuryl alcohol, , …

67

Pyrrole

114

3-140 ppb [3]

68

furane, pentadiene, …

122

2-heptanone*, methylhexanone, … trimethylpyrazine*, methylbenzyl alcohol*, …

70

dihydrofuran, pentene, …

124

71

Pyrrolidine

126

72

butanone, methylpropanal*

74

methylpropanol, propanoic acid

79

pyridine*

81

methylpyrrole*

82

2-Methylfuran*

84

pentenone, methylbutenal, … piperidine, methylpyrrolidine, … methylbutanal*, diacetyl*, pentanal*, …

0.23 mg/kg [6] 2-17 ppb [3], 2-26 ppb [3]//1.5 µg/kg [6]

SC R

8-76 ppb [3]//1.2 mg/kg [6]

136

tetramethylpyrazine*, phenylacetic acid*, …

19.7 ppm [1]//9-84 ppb [3]//3.54 µg/g [4]//0.43-6.08 mg/kg [5], 220ppb [3]//5.7 mg/kg [6]

52 µg/g [4]

138

52 µg/g [4]

140

nonadienal, methoxyethylpyrazine, … nonenal*, …

3-15 ppb [3]

32 µg/g [4]

150

diethyl-methylpyrazine*, ethyl benzoate, …

2-12 ppb [3]//0.45 or 2.06 µg/g [4]//3.3 µg/kg [6]

152

1-11 ppb [3]

180

methoxy-isopropylpyrazine, vanillin*, … Theobromine

46-193 ppb [3]//33.9 mg/kg [6], 26-73ppb [3], 0.39 µg/g [4] methylpyrazine*, phenol*, … 2-12 ppb [3]//1.74 µg/g [4], 2-16 ppb [3]

194

Caffeine

256

tetradecanoic acid ethyl ester, hexadecanoic acid (palmitic acid)*

95

Formylpyrrole, ethylpyrrole*

32 µg/g [4]

280

96

furfural*, ethylfuran, furfurol

1-26 ppb [3]

282

98

furfuryl alcohol, hexenone, …

octadecadienoic acid (linoleic acid) octadecenoic acid (oleic acid) Hexadecanoic acid ethyl ester*, octadecanoic acid (stearic acid)

94

CE P

AC

86

D

furaneol*, octanal, …

TE

MA

guaiacol*, methyl acetylfuran, … octenal*, octenone*, hydroxymethylfurfural, …

10.6 ppm [1]//1-18 ppb [3]//4.06 µg/g [4]//0.53-1.74 mg/kg [5]//0.92 mg/kg [6], 14-63 ppb [3]//7.5 mg/kg [6]

128

85

6-29 ppb [3]

2-32 ppb [3]// 3.09 or 3.54 µg/g [4]// 2.51mg/kg [5], 9.9 µg/kg [6], 21133 ppb [3] 3-57 ppb [3]

IP

99

literature data with concentration information

T

assigned compound (main contribution)

NU

19-112 ppb [3]

m/z

284

52 µg/g [4]

0.26µg/g [4]

[1] Ziegleder 1982: SE-GC (Ziegleder, 1982), [2] Granvogel, 2006: SE-LC-MS-MS(Granvogl, Bugan, & Schieberle, 2006), [3] Huang, 2011: SIFTMS(Huang & Barringer, 2011), [4] Gill, 1984: High vakuum distillation GC-MS(Gill et al., 1984), [5] Ramli, 2006: SE-GC-MS(Ramli et al., 2006), [6] Frauendorfer, 2008: SE-GC-MS(Frauendorfer & Schieberle, 2008) *: Concentrations given in the second column belong to the Compounds marked with asterisks in the first column. Concentrations from different literature are separated with // and for different compounds with a comma.

The measurements were also conducted by using a different wavelength for ionization. The result for the REMPI-technique, which is selective for aromatics, is also shown in Figure 2 (right). With REMPI there are four dominant signals, which can be assigned to theobromine (m/z 180), caffeine (m/z 194), indole (m/z 117) and phenylacetaldehyde (m/z 120). Phenylacetaldehyd is a potent aroma compound of coffee, beer and cocoa (Adamiec, Rössner, Velsek, Cejpek, & Savel, 2001; Belitz et al., 2001). It is a typical product of the Strecker degradation of phenylalanine, a by-reaction of the Maillard reaction, as mentioned and one of the main aroma contributors in chocolate (Counet et al.,

ACCEPTED MANUSCRIPT

SC R

IP

T

2002; Schnermann & Schieberle, 1997). It was found to be the main volatile during cocoa roasting (Gill et al., 1984). The highest signal at m/z 59 can be assigned to propylamine. Small amounts of other amines, such as methyl-propylamine (m/z 73) and phenyl-ethylamine (m/z121) were also found. These amines were found in roasted cocoa beans and a new thermogenic Strecker-type formation pathway was proposed (Granvogl et al., 2006). The propylamine found here could therefore be formed by the same Strecker-type reaction or might be a degradation product of the other amines. One has to note that the high signal is not a sign of high amounts of propylamine, but could be due to a favored ionization with the wavelength used. The other signals are rather low compared to the dominant ones. A detailed compilation of the signals and proposed structures can be found in Table 2. In the higher mass region there are also some signals at which m/z 358, 360 and 374 could be Matairesinol, Lariciresinol and Hdroxymatairesinol, respectively. They have also been found by Milder et al., 2005, in coffee (Milder, 2005), but until now these compounds were not found during the roasting of cocoa. Furthermore as with SPI furan, pyrrole, pyrazine and their derivatives were found.

104

styrene, methional*, …

117

indole*, …

118

benzofuran, indane, methylstyrene, …

120

phenylacetaldehyde*, ethenyl-methylpyrazine*, acetophenone*, …

121

ethyl-methylpyridine, phenylethylamine*

131

literature data with concentration information 1988/2793 µg/kg [2]

D

m/z 73

assigned compounds (main contribution) Methylpropylamine*, dimethylformamide

MA

NU

Table 2: Assigned compounds (main contribution) for the additional m/z ratios observed by REMPI, according to literature and previous studies using the ionization technique (Afoakwa et al., 2008; Dorfner, 2004; Gill et al., 1984; Huang & Barringer, 2011; Milder, 2005; Ramli et al., 2006; Van Der Wal et al., 1971; Van Praag et al., 1968; Vitzthum et al., 1975)

butyl-dimethylpyrazine*, phenyl-ethyl acetate*, …

178

dimethyl(methylbutyl)pyrazine, phenanthrene, …

206

C2-alkylated lphenanthrene

2-22 ppb [3]//10.25 µg/g [4]//5.5 mg/kg [6], 0.48 mg/kg [5], 13-154 ppb [3]//2.57 µg/g [4]

208

9,10-anthracenedione, Methylferulic acid

7746/10216 µg/kg [2]

218

Benzonaphthofuran

methylindol (skatole)

220

C3-alkylated phenanthrene

132

Cinnamaldehyde, dimethylstyrene, …

232

Diphenylpyrazin

134

diethylbenzene, methylacetophenone, …

234

C4-alkylated phenanthrene (retene)

143

Methylchinoline

286

Retinol?

144

phenylfuran, methylquinoxaline

294

?

145

Phenyloxazol, …

358

Matairesinol ?

146

furyl-pyrazine, ...

360

Lariciresinol ?

148

ethyl-dimethyl-dihydrocyclopentapyrazine, phenylbutanone, …

374

Hydroxymatairesinol ?

TE

164

160

2-45 ppb [3]

m/z 162

Tentatively assigned compounds (main contribution) methyl-ethyl-dihydrocyclopentapyrazine, safrol, …

AC

CE P

1-16 ppb [3]

literature data with concentration information

26 µg/g [4], 2-36 ppb [3]//45 µg/g [4]//0.23-0.96 mg/kg [5]//0.93 mg/kg [6]

methyl-furylpyrazine, 0.59 µg/kg [6] methyl(methyldithio)furan*, … [1] Ziegleder 1982: SE-GC (Ziegleder, 1982), [2] Granvogel, 2006: SE-LC-MS-MS(Granvogl et al., 2006), [3] Huang, 2011: SIFT-MS(Huang & Barringer, 2011), [4] Gill, 1984: High vakuum distillation GC-MS(Gill et al., 1984), [5] Ramli, 2006: SE-GC-MS(Ramli et al., 2006), [6] Frauendorfer, 2008: SE-GC-MS(Frauendorfer & Schieberle, 2008)

ACCEPTED MANUSCRIPT

T

*: Concentrations given in the second column belong to the Compounds marked with asterisks in the first column. Concentrations from different literature are separated with // and for different compounds with a comma.

SC R

IP

Analysis of roast gases inside the bean

AC

CE P

TE

D

MA

NU

The study of the roast gases inside the cocoa bean performed with a newly designed µ-probe sampling system also showed a lot of volatile species. The results for both ionization techniques, SPI and REMPI, are presented in Figure 3. Besides the higher signal for methanthiole (m/z 48) compared to the acetaldehyde/propane-signal at m/z 44, no significant differences are obvious at first glance between the measurements inside and outside the cocoa bean. With REMPI two new signals in the higher mass region are striking. These are the m/z ratios of 266, which could be assigned to methylperylene or methylbenzopyrene and m/z 294, which cannot be assigned to a known product. Besides these two additional signals no obvious differences could have been noted for the comparison of the measurements inside and outside the cocoa bean. For this reason the data of both measurements, inside and outside the bean were normalized and subtracted from each other to obtain difference plots, shown in Figure 4. Furthermore a statistical method was used, the principle component analysis (PCA), to separate the measurements and extract the responsible masses for the differences (results are not shown). With both ionization techniques used, theobromine (m/z 180) is more dominant outside and caffeine (m/z 194) is more prominent inside the cocoa bean during roasting. Furthermore the fatty acids (m/z 256, 280, 282, 284) and the resinol derivatives (m/z 358, 360, 374) are more present outside the bean. There are two possible explanations for higher amounts of a substance outside the bean. Firstly the properties of a substance, such as solubility, sterical structure, bonding to other compounds and so on, can promote or constrain its transport through the bean and secondly the substance is formed or released predominantly from the outer layers or surface of the bean. The latter may be due to the higher temperature at the surface of the bean due to exposure to the heating gun. The first explanation is more likely for theobromine and caffeine. One reason for the difference could be that the additional alkyl group hinders the caffeine to diffuse through and leave the bean as easy as theobromine or that theobromine is released more easy, because it is only slightly bonded to tannins (Belitz et al., 2001). For the fatty acids a diffusion through the cocoa bean is unlikely, therefore a possible explanation is that they are formed or released more effective at the surface or from the outer layers of the cocoa bean maybe because of the higher temperature at the surface due to the heating process via a heating gun or due to different composition of the outer layer compared to inner layers of the cocoa bean. An advantage of the PCA is the possibility of tracing the course of the roasting as shown in Figure 5. In Figure 5 the roasting courses and the corresponding m/z ratios of the measurements with SPI and REMPI inside the bean are depicted. Every step corresponds to 90 seconds. It was observed that at the beginning of the roast process, circa during the first 10 minutes (circa until step number 6), most changes take place The course of roasting in both cases can be followed by some characteristic masses, which are also responsible for the changes at the beginning of the roasting process. With SPI acetaldehyde (m/z 44) and caffeine (m/z 194) are released very early in the roasting process, followed by theobromine (m/z 180) and formyl-and/or ethylpyrrole (m/z 95). With REMPI the m/z ratio 120 (phenylacetaldehyde), a typical Maillard reaction product and key aroma compound for cocoa, was released before caffeine, indole (m/z 117), theobromine and propylamine (m/z 59). Phenylacetaldehyde is formed in a very early stage of the roasting which seems to underline its importance in the aroma of roasted cocoa. In both cases caffeine is released before theobromine. This could maybe due to the lower point of sublimation of caffeine compared to theobromine (caffeine: 178 °C, theobromine 290°C (Belitz et al., 2001)). At the end point of the two courses a lot of compounds are formed and released, which can be connected to the Maillard reaction and aroma of cocoa, such as aldehydes, ketones and heterocycles. This indicates the initial time for the main part of the Maillard reaction. Aroma active compounds, except phenylacetaldehyde, are not released in this case until ca.

ACCEPTED MANUSCRIPT

SC R

IP

T

10 minutes in the roasting. With REMPI the resinol derivatives are formed at the end of the roasting process. They could therefore maybe act as indicators of the end of roasting or even over-roasting. With SPI propene is released clearly after the other compounds shown. It could be a pyrolysis product of primary formed substances and could therefore also act as an indicator for over-roasting. The courses of the characteristic masses for the roast process obtained by the PCA are shown in Figure 6. In both cases the same observations can be made as with the PCA. With SPI acetaldehyde is formed at a very early point of the roasting, directly followed by caffeine. Theobromine is released after an initial phase and propene only at a very late time in the roasting process, which could therefore be used to indicate the end of the roasting. With REMPI the courses of the selected masses can also be followed. Again, as with the PCA, phenylacetaldehyde is released at the beginning of the roasting process, followed by indole, theobromine and propylamine.

NU

Conclusion

AC

CE P

TE

D

MA

The PI-TOFMS technique with a newly constructed µ-probe sampling system was successfully used to analyze the roast gases of cocoa and a lot of volatile roast products could have been found. The found products are in good agreement with the literature, so the new analytical technique seems to be suitable for the analysis of cocoa roast gases. Compared to previous analytical methods using for example vacuum distillation and steam distillation to extract the analytes from the cocoa, the presented technique enables faster and on-line measurements , since no sample preparation is required. Furthermore, evolving roast gases inside the cocoa bean were measured. . At first glance the product pattern of the roast gases inside and outside the whole bean seemed to be very similar, except the two additional signals at m/z 266 and 194 with REMPI and the higher methanethiole signal with SPI. With the help of difference plots and statistical methods it was possible to clearly distinguish the two measurements and identify the differences. With both wavelengths used theobromine has a higher presence outside the bean, whereas caffeine is more dominant inside the bean. This could be explained by the slightly different chemical properties of the two alkaloids and therefore different diffusion behavior. Furthermore, the dominance of fatty acids outside the bean could be explaned by a higher presence in the outer layers or a more effective release because of the higher temperature at the surface due to the heating gun. The most changes take place within the first 10 minutes, which are mostly explained by caffeine, theobromine and the key aroma compound phenylacetaldehyde. In the later part of the roasting, only slightly changes can be observed and most of the Maillard and aroma products are formed. The plotting of the roasting courses showed that caffeine is always released before theobromine, which can be explained by the different sublimating points. For future measurements it would be interesting to repeat the measurements using a small scaled roast oven with a whole batch of cocoa beans or even to go into the field and measure the roast gases of an industrial roast oven under real conditions. This helps to analyze the behavior of the cocoa beans during the roasting under real conditions. With such an assembling, experiments can be conducted where an over-roasting can be provoked and therefore typical marker can be confirmed. Furthermore roast gas compositions can be analyzed in different stages of the roasting process and can be connected to an aroma of the end product. This could help to monitor the roasting process on-line to find the optimal flavor composition and to avoid over-roasting. An analysis of different cocoa bean types (criollo and forastero) is possible to determine if there are differences. Also the influence of different growing locations or conditions or fermentation techniques is an interesting point. Furthermore the temperature of roasting can be varied to find the optimal roasting temperature. In addition, it would be interesting to confirm some of the structures proposed for the found signals to combine the assembling to a mass spectrometer with higher mass resolution or a different analytical technique. Since the presented technique was developed and successfully applied for the analysis of cigarette smoke {Hertz, 2012 #258} other applications are imaginable besides cocoa roasting gases. In principle any roasting or combustion process is possible. The technique has also already been applied to the analysis of coffee roast gases {Hertz-Schünemann, 2013 #300; Hertz-Schünemann, 2013 #287}.

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

SC R

IP

T

In summary, the presented analytical technique seems to be well suited for the analysis of cocoa roast gases. It provides a fast, reliable and non-labor intensive method, which could therefore be a practical technique for the on-line analysis and monitoring of the cocoa roasting process.

ACCEPTED MANUSCRIPT References

AC

CE P

TE

D

MA

NU

SC R

IP

T

Adamiec, J., Rössner, J., Velsek, J., Cejpek, K., & Savel, J. (2001). Minor Strecker degradation products of phenylalanine and phenylglycine. European Food Research and Technology, 212 (2), 135140. Afoakwa, E. O., Paterson, A., Fowler, M., & Ryan, A. (2008). Flavor formation and character in cocoa and chocolate: A critical review. Critical Reviews in Food Science and Nutrition, 48 (9), 840857. Bailey, S. D. M., D.G.; Bazinet, M.L.; Weurman, C. (1962). studies on the volatile components of different varieties of cocoa beans. Journal of Food Science, 27, 165-170. Bainbridge, J. S., & Davies, S. H. (1912). CCXXXI. - The essential oil of cocoa. Journal of the Chemical Society, Transactions, 101, 2209-2221. Baltes, W. (1980). Die Bedeutung der Maillardreaktion für die Aromabildung in Lebensmitteln. Lebensmittelchem. Gerichtl. Chem., 34, 39-47. Belitz, H.-D., Grosch, W., & Schieberle, P. (2001). Lehrbuch der Lebensmittelchemie (5., vollst. überarb. Aufl. ed.). Berlin: Springer. Boesl, U. (2000). Laser mass spectrometry for environmental and industrial trace analysis. J. Mass Spectrom., 35, 289-304. Butcher, D. J. (1999). Vacuum Ultraviolet Radiation for Single-Photoionization Mass Spectrometry: A Review. Microchemical Journal, 62, 354–362. Cambrai, A., Marcic, C., Morville, S., Houer, P. S., Bindler, F., & Marchioni, E. (2010). Differentiation of chocolates according to the cocoa's geographical origin using chemometrics. Journal of Agricultural and Food Chemistry, 58 (3), 1478-1483. Counet, C., Callemien, D., Ouwerx, C., & Collin, S. (2002). Use of gas chromatography-olfactometry to identify key odorant compounds in dark chocolate. Comparison of samples before and after conching. Journal of Agricultural and Food Chemistry, 50 (8), 2385-2391. Dorfner, R., Ferge, T., Yeretzian, C., Kettrup, A., Zimmermann, R. (2004). Laser Mass Spectrometry as On-Line Sensor for Industrial Process Analysis: Process Control of Coffee Roasting. Analytical Chemistry, 76, 1386-1402. Frauendorfer, F., & Schieberle, P. (2008). Changes in key aroma compounds of Criollo cocoa beans during roasting. Journal of Agricultural and Food Chemistry, 56 (21), 10244-10251. Gill, M. S., Macleod, A. J., & Moreau, M. (1984). Volatile components of cocoa with particular reference to glucosinolate products. Phytochemistry, 23 (9), 1937-1942. Granvogl, M., Bugan, S., & Schieberle, P. (2006). Formation of amines and aldehydes from parent amino acids during thermal processing of cocoa and model systems: New insights into pathways of the strecker reaction. Journal of Agricultural and Food Chemistry, 54 (5), 17301739. Hertz-Schünemann, R., Dorfner, R., Yeretzian, C., Streibel T., Zimmermann R. (2013). On-line process monitoring of coffee roasting by resonant laser ionisation time-of-flight mass spectrometry: bridging the gap from industrial batch roasting to flavour formation inside an individual coffee bean. J. Mass Spectrom., 48 (12), 1253-1265. Hertz-Schünemann, R., Streibel, T., Ehlert, S., & Zimmermann, R. (2013). Looking into individual coffee beans during the roasting process: direct micro-probe sampling on-line photoionisation mass spectrometric analysis of coffee roasting gases. Analytical and Bioanalytical Chemistry, 1-14. Hertz, R., Streibel, T., Liu, C., McAdam, K., & Zimmermann, R. (2012). Microprobe sampling-Photo ionization-time-of-flight mass spectrometry for in situ chemical analysis of pyrolysis and combustion gases: Examination of the thermo-chemical processes within a burning cigarette. Analytica Chimica Acta, 714, 104-113.

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

SC R

IP

T

Huang, Y., & Barringer, S. A. (2011). Monitoring of Cocoa Volatiles Produced during Roasting by Selected Ion Flow Tube-Mass Spectrometry (SIFT-MS). Journal of Food Science, 76 (2), C279C286. Humston, E. M., Knowles, J. D., McShea, A., & Synovec, R. E. (2010). Quantitative assessment of moisture damage for cacao bean quality using two-dimensional gas chromatography combined with time-of-flight mass spectrometry and chemometrics. Journal of Chromatography A, 1217 (12), 1963-1970. Humston, E. M., Zhang, Y., Brabeck, G. F., McShea, A., & Synovec, R. E. (2009). Development of a GC x GC-TOFMS method using SPME to determine volatile compounds in cacao beans. Journal of Separation Science, 32 (13), 2289-2295. Krysiak, W. (2006). Influence of roasting conditions on coloration of roasted cocoa beans. Journal of Food Engineering, 77 (3), 449-453. Mason, M. E., Johnson, B., & Hamming, M. (1966). Peanut compounds: Flavor components of roasted peanuts. Some low molecular weight pyrazines and a pyrrole. Journal of Agricultural and Food Chemistry, 14 (5), 454-460. Milder, I. E. J. A., I.C.W.; van de Putte, B.; Venema, D.P.; Hollmann, P.C.H. (2005). Coffee. British Journal of Nutrition, 93, 393-402. Mühlberger, F., Hafner, K., Kaesdorf, S., Ferge, T., & Zimmermann, R. (2004). Comprehensive on-line characterization of complex gas mixtures by quasi-simultaneous resonance-enhanced multiphoton ionization, vacuum-UV single-photon ionization, and electron impact ionization in a time-of-flight mass spectrometer: Setup and instrument characterization. Analytical Chemistry, 76 (22), 6753-6764. Mühlberger, F., Zimmermann, R., Kettrup, A. (2001). A Mobile Mass Spectrometer for Comprehensive On-Line Analysis of Trace and Bulk Components of Complex Gas Mixtures: Parallel Application of the Laser-Based Ionization Methods VUV Single-Photon Ionization, Resonant Multphoton Ionization, and Laser-Induced Electron Impact Ionization. Analytical Chemistry, 73 (3590-3604). Oliviero, T., Capuano, E., Cämmerer, B., & Fogliano, V. (2009). Influence of roasting on the antioxidant activity and HMF formation of a cocoa bean model systems. Journal of Agricultural and Food Chemistry, 57 (1), 147-152. Owusu, M., Petersen, M. A., & Heimdal, H. (2012). Effect of fermentation method, roasting and conching conditions on the aroma volatiles of dark chocolate. Journal of Food Processing and Preservation, 36 (5), 446-456. Rainer Cremer, D., & Eichner, K. (2000). The reaction kinetics for the formation of Strecker aldehydes in low moisture model systems and in plant powders. Food Chemistry, 71 (1), 37-43. Ramli, N., Hassan, O., Said, M., Samsudin, W., & Idris, N. A. (2006). Influence of roasting conditions on volatile flavor of roasted Malaysian cocoa beans. Journal of Food Processing and Preservation, 30 (3), 280-298. Schieberle, P., Pfnuer, P. (1999). Characterization of key odorants in chocolate. Flavor Chemistry: thirty years of progress, New York, 147-153. Schnermann, P., & Schieberle, P. (1997). Evaluation of Key Odorants in Milk Chocolate and Cocoa Mass by Aroma Extract Dilution Analyses. Journal of Agricultural and Food Chemistry, 45 (3), 867-872. Torres-Moreno, M., Tarrega, A., Costell, E., & Blanch, C. (2012). Dark chocolate acceptability: Influence of cocoa origin and processing conditions. Journal of the Science of Food and Agriculture, 92 (2), 404-411. Van Der Wal, B., Kettenes, D. K., Stoffelsma, J., Sipma, G., & Semper, A. T. J. (1971). New volatile components of roasted cocoa. Journal of Agricultural and Food Chemistry, 19 (2), 276-280. Van Praag, M., Stein, H. S., & Tibbetts, M. S. (1968). Steam volatile aroma constituents of roasted cocoa beans. Journal of Agricultural and Food Chemistry, 16 (6), 1005-1008. Vitzthum, O. P., Werkhoff, P., & Hubert, P. (1975). Volatile components of roasted cocoa: basic fraction. Journal of Food Science, 40, 911-915.

ACCEPTED MANUSCRIPT

SC R

IP

T

Zhang, Y. (1992). Contribution of peptides to volatile formation in the maillard reaction of casein hydrolysate with glucose. Journal of Agricultural and Food Chemistry, 40 (12), 2467-2471. Ziegleder, G. (1982). Gaschromatographische Röstgradbestimmung von Kakao über methylierte Pyrazine. Deutsche Lebensmittel-Rundschau, 78. Jahrg.; Heft 3, 77 - 81. Ziegleder, G., & Stojacic, E. (1988). Changes in the flavour of milk chocolate during storage. Lagerungsbedingte Veänderungen im Aroma von Milchschokoladen, 186 (2), 134-138.

Figures

NU

Figure 1: design of the different sampling techniques (a-microprobe outside the whole bean, b-microprobe inside the whole bean) and the PI-TOF-MS system

MA

Figure 2: mass spectra of the roasting of cocoa beans sampled outside the bean and obtained with SPITOFMS (left) and REMPI-TOFMS (right) Figure 3: mass spectra of the measurements inside the cocoa bean using a µ-probe and different ionization techniques, SPI-TOFMS (left) and REMPI-TOFMS (right)

D

Figure 4: difference spectra for SPI (left) and REMPI (right)

TE

Figure 5: PCA results for the measurements inside the cocoa bean: a) courses of roasting (SPI) b) responsible masses for the courses in a), c) course of roasting (REMPI) d) responsible masses for the course in c)

AC

CE P

Figure 6: courses of selected compounds measured with SPI (left) and REMPI (right)

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

Figure 1

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

Figure 2

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

Figure 3

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Figure 4

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

Figure 5

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

Figure 6

ACCEPTED MANUSCRIPT Highlights of the manuscript

T

IP

SC R

NU MA D TE



CE P

 

a micro-probe – photo-ionization-Time of flight mass spectrometer was successfully applied for the analysis of cocoa roast gases a lot of aroma and Maillard reaction products were identified slight differences between the product pattern of roasting gases evolving inside and outside a single cocoa bean were observed The most changes, within the first 10 minutes, can be explained by a small number of compounds and in the later part of the roasting a lot of typical Maillard reaction products are formed

AC

