International Journal of Mass Spectrometry 424 (2018) 49–57
Contents lists available at ScienceDirect
International Journal of Mass Spectrometry journal homepage: www.elsevier.com/locate/ijms
Full length article
On-line analysis of coffee roasting with ion mobility spectrometry–mass spectrometry (IMS–MS) A.N. Gloess a,∗ , C. Yeretzian a , R. Knochenmuss b,c , M. Groessl c,d a Zurich University of Applied Sciences, Institute of Chemistry and Biotechnology, Coffee Excellence Center, Einsiedlerstrasse 31, 8820 Wädenswil, Switzerland b University of Bern, Department of Chemistry and Biochemistry, Freiestr. 3, 3012 Bern, Switzerland c TOFWERK, Uttigenstr. 22, 3600 Thun, Switzerland d University of Bern, Inselspital, Bern University Hospital, Department of Nephrology and Hypertension and Department of BioMedical Research, 3010 Bern, Switzerland
a r t i c l e
i n f o
Article history: Received 10 September 2017 Received in revised form 14 November 2017 Accepted 19 November 2017 Available online 22 November 2017 Keywords: Ion mobility spectrometry–mass spectrometry (IMS–MS) On-line analysis Coffee roasting Pyrazines Fatty acids
a b s t r a c t On-line analysis of coffee roasting was performed using ion mobility spectrometry–mass spectrometry (IMS–MS) with corona discharge ionization. This is the first time that formation of volatile organic compounds (VOCs) during coffee roasting was monitored not only in positive but also in negative ion mode, and not only with mass spectrometry, but also with ion mobility spectrometry. The temporal evolution of more than 150 VOCs was monitored during the roasting of Brazilian Coffea arabica. Mass-selective ion mobility spectrometry allowed a separation of isobaric and isomeric compounds. In positive ion mode, isomers of alkyl pyrazines were found to exhibit distinct time-intensity profiles during roasting, providing a unique insight into the complex chemistry of this important class of aroma active compounds. Negative ion mode gave access to species poorly detectable by other on-line methods, such as acids. In this study, the release of fatty acids during coffee roasting was investigated in detail. These increase early on in the roasting process followed by a decrease at the same time as other VOCs start to be formed. © 2017 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction Human beings are provided by nature with five senses: sight, hearing, taste, touch and smell. These warn us about danger, and fill life with delightful moments. Smell, for example, prevents us from eating spoilt food, but on the other, what would a good meal be without the delicious aroma? Mankind is therefore constantly trying to improve the aroma of food, and to understand the chemistry behind it. From a chemist’s perspective, aroma consists of volatile organic compounds (VOCs), and smell is an interaction between these VOCs and receptors in the nose. A specific aroma may be evoked by just a few VOCs, as raspberry is dominated by the ketones (4-(4hydroxyphenyl)-butan-2-one) and ␣- and -ionone [1]. For other aromas, however, a complex interplay of up to 40 VOCs is needed to create the final sensory impression [2]. Above a cup of coffee, for example, almost 1000 VOCs have been identified, and at least 20 of them are needed to reconstitute coffee aroma [3–15].
∗ Corresponding author. Present address: 8805 Richterswil, Switzerland. E-mail address:
[email protected] (A.N. Gloess).
This variety of coffee aroma compounds is largely generated during the roasting of green coffee beans, in a complex interplay of various chemical reactions. One of these is the Maillard reaction, in which sugars react with amino acids to form aroma compounds such as alkyl pyrazines, which contribute a roasty note. Pyridines, in contrast, are formed by thermal degradation of trigonelline. Analysis of these reactions could provide a deeper understanding of how to treat coffee beans to elicit best flavour properties. These analyses have traditionally been performed with gas chromatography. Samples are taken at different steps of the coffee roasting process and analysed off-line [11,16–20]. Direct monitoring of roasting is, however, preferable to gain insight into the complex pathways of chemical reactions of VOC formation. On-line analysis of VOC formation additionally saves the time and cost of sample preparation, and avoids process interruption. This approach has already been applied successfully for coffee roasting, especially using mass spectrometry coupled with different ionization methods: resonance enhanced multi photon ionization time-of-flight mass-spectrometry (REMPI-TOF-MS), single photon ionization TOF-MS (SPI-TOF-MS), [21–27] as well as proton-transfer-reaction mass spectrometry (PTR-MS). The first on-line analyses of coffee roasting with PTR-MS were performed using a quadrupole mass spectrometer [28–30], which is limited in
https://doi.org/10.1016/j.ijms.2017.11.017 1387-3806/© 2017 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4. 0/).
50
A.N. Gloess et al. / International Journal of Mass Spectrometry 424 (2018) 49–57
Fig. 1. Schematic setup of the coupling of the TOFWERK IMS-TOF to the coffee roaster.
Fig. 2. Mass spectra summed over the entire roasting process in negative (top; scaled to the peak at m/z 265 as base peak) and positive (bottom; scaled to the peak at m/z 195 as base peak) ion modes. The largest peaks are truncated to better show the lower abundant species.
Fig. 3. Summed ion mobility spectra corresponding to the mass spectra of Fig. 2, in negative (top) and positive (bottom) ion modes.
sensitivity and mass resolution because it is a scanning instrument. Coupling the PTR source to a time-of-flight mass spectrometer (PTR-TOF-MS) was a step forward, as it delivers information about all ions simultaneously, and the higher mass resolution allows separation of some isobaric compounds. First PTR-TOF-MS studies of coffee roasting showed specific formation dynamics for different VOCs, and how these changed with different time-temperature roasting-profiles to the same roast degree [31–34]. Further studies shed light on how chemical reactions varied when roasting different coffees along the same time-temperature roasting profile [35]: Both the start of aroma formation varied as well as the dynamics of VOC formation. While TOF-MS provides far more information and better mass resolution than a quadrupole MS, it is still unable to resolve mul-
tiple compounds of a single composition, such as isomers of alkyl pyrazines or of chlorogenic acids (CGAs). A partial solution to this problem is to perform a low resolution flash GC prior to injecting into the mass spectrometer. Even though experimental time resolution is reduced, on-line analysis can still be performed. This approach has been reported by Romano et al. for wine analysis and Ruzsanyi et al. for the analysis of human breath and of VOCs emanated through skin [36,37]. Both laser-based and proton transfer techniques exhibit some selectivity in the ionization process. For REMPI and SPI, VOCs must absorb light of the respective wavelength. In a PTR source, only molecules with proton affinities higher than water become ionized by H+ transfer. In both methods only positive analyte ions are generated. A different ionization technique, allowing the analysis of a
A.N. Gloess et al. / International Journal of Mass Spectrometry 424 (2018) 49–57
51
gas depends on the collision cross sections of the ions. Ions with a larger collision cross section move slower through the gas than smaller ones. Ions can have the same mass but different shapes, and therefore different cross sections such as isomers, conformers or diastereomers. While these cannot be separated by MS alone, they are often of great interest. Because the cross sections are not identical, ion mobility often can separate them, without a preceding chromatographic step. In the following, we present first results of on-line monitoring of coffee roasting with IMS-MS, showing that time resolution is found to be well matched to fast processes like roasting, and that separation based on collision cross section reveals a wealth of new information on VOC isomers, both in positive and negative modes. Detectable VOCs were found to have a surprisingly large mass range with m/z values up to 500. 2. Materials and methods Fig. 4. IMS spectra of pyrazines reveal the presence of isomers. From bottom to top: pyrazine (m/z 81.05), methyl-pyrazine (m/z 95.06), dimethyl-pyrazine or ethylpyrazine (m/z 109.08), trimethyl-pyrazine or methyl-ethyl-pyrazine (m/z 123.09), diethyl-pyrazine or dimethyl-ethyl-pyrazine (m/z 137.09), diethyl-methyl-pyrazine (m/z 151.10).
wider range of both positively and negatively charged VOCs would be preferable. Here we report a new approach, using an ion mobility time-offlight mass spectrometer and a corona discharge ion source. This technique retains the full advantage of a TOF MS, is fast enough for on-line analysis of coffee roasting, and has the following major advantages: – separation of isobaric and isomeric compounds based on their collision cross section. – positive and negative ion modes. – molecules with higher molecular mass can be analysed compared to PTR-TOF-MS.
2.1. Ion mobility spectrometry–mass spectrometry All measurements were carried out on a TOFWERK IMS-TOF (Thun, Switzerland). The system comprised an ion source (corona discharge), a 10 cm desolvation tube, an atmospheric pressure 20 cm drift tube (both made from resistive glass) and a TOFWERK HTOF TOFMS with two-stage interface to the IMS. Desolvation and drift tubes were thermostatted at 150 ◦ C with nitrogen as the drift gas. Ion mobility separation was carried out at a field strength of ca. 400 V/cm (reduced electric field strength ca. 2 Td). Mass spectra were acquired from m/z 10–500 in both positive and negative ion modes. Raw IMS-TOF data was post-processed using IMSviewer and Tofware (both Tofwerk, Switzerland). The IMS instrument uses a Hadamard multiplexing method. In addition to increasing sensitivity, properties of the time domain data can be exploited to increase both signal to noise ratio and effective resolution. The latter can significantly exceed the diffusion limit which applies to conventional pulsed-mode drift tube IMS [38]. 2.2. Coffee roasting
Ion mobility separates ions of identical mass but different collisional cross section by pulling them with an electric field through a collision gas. The speed of ion movement through the neutral
Roasting was performed with a drum roaster (Gene Café). Each batch of 120 g of Brazilian Yellow Bourbon washed coffee was
Fig. 5. Time trends for IMS-MS peaks assigned to alkyl pyrazines. Peaks of identical m/z but different drift time show different temporal behaviours during roasting.
52
A.N. Gloess et al. / International Journal of Mass Spectrometry 424 (2018) 49–57
Table 1 Isomers of alkyl pyrazines: name, aroma, molecular formula, monoisotopic mass and protonated mass in Da. alkyl pyrazine
aroma
molecular formula
mass (monoisotopic)/Da
protonated mass/Da
pyrazine 2-methyl pyrazine 2,3-dimethyl-pyrazine 2,5-dimethyl-pyrazine 2,6-dimethyl-pyrazine 2-ethyl-pyrazine 2,3,5-trimethyl-pyrazine 2-ethyl-3-methyl-pyrazine 2-ethyl-5-methyl-pyrazine 2-ethyl-6-methyl-pyrazine 2,3-diethyl-pyrazine 2,5-diethyl-pyrazine 2,6-diethyl-pyrazine 2,5-dimethyl-3-ethyl-pyrazine 3,5-dimethyl-2-ethyl-pyrazine diethyl-methyl-pyrazine
sweet, corn, roasted hazelnuta nutty, cocoa, roasted, chocolate, peanut, greena , popcornb nutty, coffee, peanut butter, walnut, caramellic, leatherya cocoa, roasted, nuttya , cocoa, roasted nutb roasted nut, cocoab peanut butter, musty, nutty, woody, roasted, cocoaa , peanut butter, woodb nutty, musty, earthy, powdery, cocoa, peanut, roasted peanuta nutty, peanut, musty, corn, raw, earthy, breadya coffee, beany, nutty, grassy, roasted (coffee pyrazine)a , fruit, sweetb roasted, potatoa raw, nutty, pepper, bell peppera hazelnut, roasted hazelnut, almond, roasted almond, peanut, roasted peanuta burnt, hazelnut, nuttya potato, roastb sweet, nutty, caramellic, coffee, corn, cocoa, potatoa , potatob
C4 H4 N2 C5 H6 N2 C6 H8 N2 C6 H8 N2 C6 H8 N2 C6 H8 N2 C7 H10 N2 C7 H10 N2 C7 H10 N2 C7 H10 N2 C8 H12 N2 C8 H12 N2 C8 H12 N2 C8 H12 N2 C8 H12 N2 C9 H14 N2
80.04 94.05 108.07 108.07 108.07 108.07 122.08 122.08 122.08 122.08 136.10 136.10 136.10 136.10 136.10 150.12
81.04 95.06 109.08 109.08 109.08 109.08 123.09 123.09 123.09 123.09 137.11 137.11 137.11 137.11 137.11 151.12
a b
http://www.thegoodscentscompany.com. http://www.flavornet.org.
roasted to an espresso-type roast degree (weight loss 18% ± 3%). The green coffee beans were filled in the roasting drum at a temperature of 27 ◦ C. Over the next ten minutes, the temperature was ramped continuously up to 230 ◦ C (maximum temperature of the coffee roaster), then held constant until roasting ended at 17.5 min. 2.3. IMS-MS – roaster coupling The IMS-TOF-MS inlet was coupled to the exhaust gas outlet of the coffee roaster as schematically shown in Fig. 1. Roast gas was withdrawn from the exhaust and diluted with nitrogen to prevent condensation (1/4 teflon tubing). The resulting diluted gas flow (approximately 5 l/min) was injected perpendicular to the corona discharge. 3. Results and discussion 3.1. Online analysis of coffee roasting The formation of VOCs during coffee roasting was monitored online with IMS-MS. Typical mass spectra averaged over the whole roasting process are shown in Fig. 2 for both negative and positive ion modes. A large number of VOC mass peaks were observed, particularly in negative mode. There is in principle no upper limit to the mass range that can be monitored with a TOF mass analyzer. However, considering that VOCs generally are found below 500 Da, this was chosen as an upper mass limit for this investiga-
tion. Positive ion mode MS spectra are dominated by protonated pyridine and caffeine (m/z 80 and 195, respectively), while negative ion mode MS spectra clearly show the presence of fatty acids (peaks between m/z 250–300). Even though the dominant negative ions generated by the corona discharge are derived from air, such as NO2 − and NO3 − , detection of VOCs is not impeded due to the dynamic range of the instrument, which is about three orders of magnitude. The corresponding averaged IMS data are shown in Fig. 3. Figs. 2 and 3 represent information averaged over the entire roasting process. Besides integrated mass spectra, on-line analysis also allows resolving in detail the time intensity information of VOCs during a roasting process. A time-intensity profile can be plotted for each peak in the mass spectrum, giving insight into formation and release of the compounds during the roasting process. Because coffee aroma contains many molecules, including isomers and isobars, identification is not always unambiguous using MS alone. One peak in the mass spectrum often corresponds to a range of molecules with the same molecular formula (isomers) or same nominal mass (isobars). IMS adds a second, orthogonal separation dimension, which is sensitive to the shape differences of isomers and isobars. The IMS-MS used here offers high mass accuracy (low ppm range) and resolution (approximately 5000 m/m), combined with mobility resolution of >100. This performance combination greatly facilitates the identification of target compounds, often separating isomers and isobars, as will be shown below.
Table 2 Observed ions in negative ion mode. tentative assignment
behaviour with roasting time
deprotonated mass/Da
phosphoric acid malic acid 4-vinyl guaiacol 4-ethyl guaiacol caffeic acid citric acid quinic acid ferulic acid fatty acids, see Table 3
decrease late maximum late maximum late maximum late maximum decrease late maximum late maximum maximum in middle of roasting
kahweol chlorogenic acid feruoyl quinic acid
maximum in middle of roasting maximum in middle of roasting maximum in middle of roasting
96.95 133.01 149.06 151.08 179.04 191.02 191.06 193.05 255.23, 279.23, 281.25, 283.26, 293.25, 295.26, 297.28, 311.30, 325.31, 367.36, 339.33 313.18 353.09 367.10
mass (monoisotopic)/Da 134.02 150.07 152.08 180.04 192.03 192.06 194.06
314.19 354.10 368.11
A.N. Gloess et al. / International Journal of Mass Spectrometry 424 (2018) 49–57
3.1.1. Pyrazines Alkyl pyrazines are important coffee aroma compounds formed during roasting in the Maillard reaction between amino acids and reducing sugars, with ␣-diketones as intermediates for the transamination followed by ring condensation [39–41]. Depending on the alkyl chain and its position, they contribute different flavours to the coffee, as listed in Table 1. These isomers have the same molecular formula and are therefore not distinguishable by pure mass spectrometry (without e.g. pre-separation by chromatography), even with a very high resolution. Since their molecular shapes, and consequently their collision cross sections, usually differ, they can be separated by IMS [42]. As shown in Fig. 4, two peaks can be seen in the ion mobility spectra of a single MS peak at m/z 109, presumably ethyl pyrazine (peanut butter, woody aroma) and dimethyl pyrazine (aroma of cocoa, roasted nut, coffee). As can been seen in the IMS-MS time traces (Fig. 5), both these IMS peaks change their intensity with roasting time. The peak at m/z 81 might correspond to the non-alkylated pyrazine molecule. In the IMS-MS traces, one would expect one single peak. However, in Fig. 4, three peaks are observed; a small one at a drift time of 20.8 ms, one intense at 27.4 ms and a third small peak at 28.8 ms. By comparison to measurements of the reference compound, we assign m/z 81, DT = 20.8 to protonated pyrazine. In the reference measurement, strong adduct formation was observed, so we tentatively assign the peak at 27.44 ms to a pyrazine ion-molecule adduct which forms in the corona ion source, and dissociates in the interface just prior to entering the MS. The third peak at 28.8 ms might correspond to an adduct, as well. The time-intensity profile of the IMS peak at m/z 81, DT = 27.44 ms is shown in Fig. 5. The intensity of the peak at m/z 81, DT = 20.8 is too small for a time resolved plot of the IMS peak. The intensity profile of the peak at m/z 81, DT = 27.44 ms shows that the pyrazine adduct is not released until the middle of the roasting process (at 230 ◦ C) and reaches its maximum near the end of roasting, which is a typical time intensity profile of pyrazine, compared to what was observed within PTR-MS studies of coffee roasting. By analogy to pyrazine it is likely, but not certain, that the peaks at higher drift time in the m/z 109, 123 and 137 traces of Fig. 4 also correspond to adducts. 2,3-Diethyl-5-methylpyrazine reference measurements showed that the two peaks at 24.9 ms and 26.2 ms correspond to conformers of the free molecular ion. The third peak at 28.7 ms is not yet identified. The differences in the temporal profiles of alkyl-pyrazine isomer peaks resolved by IMS–MS are striking. Some peaks seem not to be strongly dependent on the roasting time (e.g. compounds at m/z 123.09, 31.01 ms; m/z 151.10, 26.24 ms), and therefore may be present already in the green beans and might not necessarily be alkyl pyrazines. Other VOCs are clearly formed only during roasting (e.g. m/z 95.06, 23.61 ms; m/z 81.05). Interestingly, some compounds are most intense at the beginning of the roasting process and then gradually decrease with time (m/z 109.08, 23.81 ms; m/z 151.10, 28.73 ms). These compounds might be released only initially upon heating of the green coffee beans, or they might be thermally degraded with ongoing roasting time. These compounds might not be alkyl pyrazines. The identification of these ions will be the aim of further studies.
3.1.2. Negatively charged ions IMS-MS was also used to monitor negative ions during coffee roasting. To the best of our knowledge, this is the first time that this has been reported. Negative mode allows sensitive detection of organic acids by deprotonation of the carboxyl group. A selection of negative ions whose intensity varied with the roasting time is given in Table 2. While some can be tentatively assigned to organic acids or guaiacols, others still remain to be assigned. VOCs which have
53
Fig. 6. Time-intensity profiles for free fatty acids and their esters, see also Table 3.
not yet been identified are listed in the supplementary information. A prominent series of ions are assigned to the fatty acids (Table 3). 3.1.3. Fatty acids Lipids are a major compound class in coffee beans, constituting up to 17% by weight in Arabica and 14% by weight in Robusta [43]. Their thermal oxidative degradation leads to smaller molecules like aldehydes which can further react to heterocyclic volatiles [43]. The content of free fatty acids (FFA) in green coffee is in the range of 1–2% in Arabica and 1–3% in Robusta [43] and decreases slightly until the end of roasting. Monitored in negative ion mode, the release of fatty acids during coffee roasting could be analysed on-line, as shown in Fig. 6. The detection of fatty acids might be surprising, as they are nonvolatile. However, during coffee roasting, the green coffee beans are both physically and chemically altered. The hard, green coffee beans are transformed into brownish, brittle coffee smelling beans, much larger in size than the green beans. During this transformation, mainly water and CO2 is released from the beans. Additionally, volatile as well as non-volatile organic compounds are released, among others caffeine. The mechanism of the release of the nonvolatile compounds might be different depending on where the substance is located within the bean. Fatty acids, for example, are present in the wax coating the green bean, as well as inside the bean [43]. The fatty acids on the surface might be released while the size and hence the surface of the bean is growing. It might be caused by an interplay of temperature, the rupture of the surface structure, maybe they are withdrawn by the evaporating CO2 or water, and/or maybe the mechanical contact with the other been in the roaster drum facilitates their release. Concerning the compounds located inside the bean, they first have to migrate through the bean. With the ongoing of the roasting process, this is increasingly facilitated. The bean’s structure turns more and more into a glassy state, and around the first crack, the bean structure is ruptured due to the high pressure inside the bean, opening channels for the release of compounds. Roasting of small batches or even of single beans has shown [21–25,29,30] that some compounds are released more or less continuously during roasting while others show sudden peaks in their release patterns, correlating with the popping sound of the first crack. This might be due to where the compounds are situated or formed during roasting as well as how they interact with the cell structure of the bean. Within batch roasting, these sudden peaks are no longer visible due to the huge number of beans popping at different timings, leading to a more or less continuous release of the compounds.
54
A.N. Gloess et al. / International Journal of Mass Spectrometry 424 (2018) 49–57
With respect to a possible delay between formation and release of compounds during roasting, former studies have shown that the on-line detection of compounds with PTR-MS is representative for the formation time. Beans were sampled at different states of the roasting process and their VOC content analysed with headspace gas chromatography coupled to mass spectrometry (HS GC/MS) above the ground coffee powder. The VOC intensities and their temporal behaviour over the roasting process were very similar to the corresponding VOC intensity profile in the on-line analysis with PTR-TOF-MS. The time-dependent profiles for free fatty acids and their esters all follow the same trend: they are released already at the lowest temperature, but reach a maximum at 5–9 min, assuming that especially at the beginning the observed fatty acids are predominantly released from the surface of the beans. Towards the end of the roasting, their intensities decrease below the starting values, probably due to thermal degradation. The position of the maximum seems to correlate with molecular weight, lighter molecules are released earlier than heavier ones. This corresponds to the expected general trend of volatility versus molecular weight. The most abundant FFA in coffee are linoleic (C18:2) and palmitic acids (C16:0), followed by stearic and oleic acids (C18:0 and C18:1) [43]. Arachidic and linolenic acids are minor constituents. This fits with the ion intensities observed within these studies: C18:2 ≈ C16:0 ≈ C18:0 > C18:1 > C20:0. Normalized ion mobility spectra of some fatty acids are shown in Fig. 7a. For fatty acids with the same number of carbon atoms, the IMS peaks shift to shorter drift times with increasing number of double bonds, indicating a smaller collision cross section for the unsaturated fatty acids. In contrast, prolonging the chain length of the fatty acid increases the drift time and correspondingly the collision cross section. This is in agreement with previous studies that employed IMS-MS for lipidomics [38,44]. Due to the low abundance of some of the compounds, IMSMS time series were relatively noisy (approximately 500 ions are required for a high-quality ion mobility spectrum). A larger time window reduces noise but reduces temporal resolution. With this limitation, no significant differences in the temporal evolution of different IMS–MS peaks were observed; consequently, only timeseries based on MS data are shown. In Fig. 7b, the use of two-dimensional IMS–MS plots is shown to quickly distinguish between compounds belonging to different chemical families. FFAs exhibit relatively high collision cross section values for their mass (as commonly observed for lipids) and consequently occupy a very specific space in 2D IMS-MS plots [45]. Therefore, assignments based on accurate mass are confirmed using 2D IMS–MS plots, and information on compounds belonging to the same chemical class can be quickly and efficiently extracted from complex data sets. 3.1.4. Other acids In addition to fatty acids, other acids were observed, as tentatively identified in Table 2. As can be seen in the time series in Fig. 8, most of the lower molecular weight acids such as caffeic acid (m/z 179) and ferulic acid (m/z 193) are formed during
Fig. 7. (a) IMS spectra of mass peaks corresponding to free fatty acids show multiple isomers/isobars. From bottom to top: c16:0 (m/z 255.22), c18:2 (m/z 279.22), c18:1 (m/z 281.24), c18:0 (m/z 283.25), c20:0 (m/z 311.28), c22:0 (m/z 339.32). (b) 2D IMSMS plot showing drift time differentiation of fatty acids (green dots) from other VOCs (red dots). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
the roasting. After passing a maximum their intensity decreases with further roasting. In contrast, heavier acids such as kahweol (m/z 314), chlorogenic acid (m/z 354) and feruoyl quinic acid (m/z 368) already reach their maximum gas phase concentrations at earlier times. The temporal evolution of the intensities of m/z 367 (feruoyl quinic acid) and m/z 193 (ferulic acid) are consisted with the degradation of feruoyl quinic acid to ferulic acid during roasting [23–25]. The here observed intensity increase of m/z 149 (4-vinyl guaiacol) however does not correlate directly to its proposed formation by decarboxylation of ferulic acid [23–25]. A further reaction of 4-vinyl guaiacol to 4-ethyl guaiacol (m/z 151) would yet be consistent with the observed intensities. Further stud-
Table 3 Fatty acids: name, molecular formula, monoisotopic mass and deprotonated mass in Dalton (Da). fatty acid
molecular formula
mass (monoisotopic)/Da
deprotonated mass/Da
c16:0 c18:0 c18:1 c18:2 c20:0 c21:0/methyl icosanoate/tetra methyl ester of c16:0 c22:0
C15 H31 COOH C17 H35 COOH C17 H33 COOH C17 H31 COOH C19 H39 COOH C21 H42 O2 C21 H43 COOH
256.24 284.27 282.26 280.24 312.30 326.32 341.34
255.23 283.26 281.25 279.23 311.30 325.31 339.33
A.N. Gloess et al. / International Journal of Mass Spectrometry 424 (2018) 49–57
55
Fig. 8. Temporal evolution of possible organic acids. The m/z values are indicated, see Table 2 for corresponding tentative assignments.
Fig. 9. Online monitoring of the isobaric compounds quinic acid (DT 26.78 ms) and citric acid (DT 25.88 ms) by IMS-MS.
ies, adding reference substances to the roasting gas, will provide more information. For the MS peak at m/z 191, two IMS peaks are detected, which were tentatively assigned as the isobaric quinic and citric acids (m/z 191.06 and m/z 191.02). The temporal evolutions of the intensities of these two mobility peaks are inverted, as seen in Fig. 9. One decreases during roasting while the other increases. As it is known that citric acid is degraded with ongoing roasting, we assign the peak with drift time = 25.88 ms to citric acid. The peak at 26.78 ms is therefore assigned to quinic acid, which is formed during roasting as a degradation product of chlorogenic acids [43]. 3.2. Comparison with other MS techniques (PTR-TOF-MS, REMPI/SPI TOF MS) In comparison with techniques as PTR-MS or REMPI/SPI-MS for online analysis of coffee roasting [21–27,31–35], IMS-MS has the clear advantage of an additional separation step. This allows analysing VOCs with the same molecular formula, as long as their collision cross sections differ. This was shown here for one particular class of molecules, the alkyl pyrazines, which each contribute distinct and different aroma notes to the aroma of coffee, depending on the length and position of the alkyl chain. Another great advantage of the IMS–MS instrument used here is the corona discharge ion source. In contrast to PTR it generates both positive and negative analyte ions. This enlarges the
number of VOCs which can be monitored during roasting, notably giving access to the fatty acids. Although some fatty acids were also observed in recent REMPI studies [27,46], this ionization technique has the disadvantage that only VOCs with appropriate chromophores can be analysed. This might be an advantage for the analysis of guaiacols, for example, which are only slightly visible using PTR. Using IMS-MS with corona discharge ionization, however, intense signals were obtained. Consequently, IMS-MS with corona discharge ionization seems to combine the advantages of both PTR and REMPI/SPI. While IMS-MS with corona discharge ionization was shown above to provide a comparatively wide range of ion species, not all species are detected with equally high sensitivity. In particular, PTR–MS detected the formation of furans during coffee roasting [31–35], but they were not observed in the present study. The difference in furan sensitivities between IMS–MS with corona discharge ionization and PTR-MS illustrates the fact that all secondary charge transfer ionization techniques are subject to numerous factors that affect relative and absolute sensitivities. Rates and extents of reactions with primary ions vary widely, and are dependent on source conditions. In addition, complex samples such as roasting gas often suffer from competitive reactions, which are not always evident but may have a large effect. We have concentrated here on exploring the chemical space made visible by IMS–MS, quantitation will follow in future studies.
4. Conclusions Within this study it was shown that corona discharge coupled to IMS–MS is a powerful tool for on-line analysis of the temporal evolution of volatile organic compounds during coffee roasting. The method exhibits two main advantages: First, the ion mobility dimension separates based on collision cross section, which often resolves isomers and isobars that cannot be separated by MS alone. This was demonstrated for alkyl pyrazines, which make different coffee aroma contributions depending on their alkyl chain lengths and positions. The ion mobility spectra clearly showed that multiple isomers often contribute to single mass peaks. This is a huge step ahead in comparison to prior on-line monitoring methods which are unable to separate isomers. Second, the corona ion source provides easy and straightforward access to both positive and negative ions, while prior on-line methods are restricted to positive ions. This allows the routine observation of a much larger range of chemical species. An important class investigated here in negative mode are the many fatty acids.
56
A.N. Gloess et al. / International Journal of Mass Spectrometry 424 (2018) 49–57
In addition, we observed remarkably high molecular weight species in negative mode, which are not found with the more commonly employed positive mode PTR–MS instruments. This might be partially due to the negative mode ionization, the ionization technique as such and the instrument settings. In the author’s earlier studies with PTR-TOF-MS, a slight enhancement of signal intensities of higher molecular weight molecules was possible, but only a few additionally peaks were observable, much less than here. At the same time, the intensities of the aroma relevant VOCs with lower molecular weight were reduced dramatically. A parallel observation of low and high molecular weight molecules with PTR-TOF-MS was therefore not possible during on-line monitoring of coffee roasting. In conclusion, the results presented here clearly demonstrate that IMS–MS opens new perspectives for on-line analysis of coffee roasting by greatly expanding the range of species detected in positive and negative ion mode, combining advantages of PTR–MS and REMPI/SPI-MS analysis, and adding an additional separation step based on the collision cross section of the VOCs. Acknowledgement The authors thank Dr. Sebastian E.W. Opitz for performing the pyrazine reference measurements. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j.ijms.2017.11.017. References [1] H.-D. Belitz, Werner Grosch, Peter Schieberle, Lehrbuch der Lebensmittelchemie, Springer-Verlag, 2008 (6. Auflage). [2] A. Dunkel, M. Steinhaus, M. Kotthoff, B. Nowak, D. Krautwurst, P. Schieberle, Th. Hofmann, Nature’s chemical signatures in human olfaction: a foodborne perspective for future biotechnology, Angew. Chem. Int. Ed. 53 (2014) 7124–7143. [3] I. Blank, A. Sen, W. Grosch, Potent odorants of the roasted powder and brew of arabica coffee, Zeitschrift fuer Lebensmittel -Untersuchung und -Forschung 195 (1992) 239–245. [4] I. Blank, A. Sen, W. Grosch, Aroma impact compounds of arabica and robusta coffees. Qualitative and quantitative investigations, in: ASIC-14ème Colloque Scientifique International Sur Le Café, Association Scientifique Internationale du Café (ASIC), Paris, 1992, pp. 117–129. [5] M. Czerny, F. Mayer, W. Grosch, Sensory study on the character impact odorants of roasted arabica coffee, J. Agric. Food Chem. 47 (2) (1999) 695–699. [6] M. Czerny, W. Grosch, Potent odorants of raw Arabica coffee. Their changes during roasting, J. Agric. Food Chem. 48 (3) (2000) 868–872. [7] W. Grosch, Key odorants of roasted coffee: evaluation, release, formation, in: ASIC-18ième Colloqium Scientific International Du Café, Association Scientifique Internationale du Café (ASIC), Paris, 1999, pp. 17–26. [8] W. Grosch, Chemistry III. Volatile compounds, in: R.J. Clarke, O.G. Vitzthum (Eds.), Coffee: Recent Developments, Blackwell Science, London, 2001, pp. 68–89. [9] W. Grosch, Key odorants of roasted coffee: evaluation, release, formation, in: ASIC-18ème Colloque Scientifique International Sur Le Café, Association Scientifique Internationale du Café (ASIC), Paris, 2000. [10] W. Grosch, Instrumental and sensory analysis of coffee volatiles, in: ASIC-16ème Colloque Scientifique International Sur Le Café, Association Scientifique Internationale du Café (ASIC), Paris, 1995, pp. 147–156. [11] F. Mayer, M. Czerny, W. Grosch, Influence of provenance and roast degree on the composition of potent odorants in Arabica coffees, Eur. Food Res. Technol. 209 (3-4) (1999) 242–250. [12] F. Mayer, M. Czerny, W. Grosch, Sensory study of the character impact aroma compounds of a coffee beverage, Eur. Food Res. Technol. 211 (4) (2000) 272–276. [13] F. Mayer, W. Grosch, Aroma simulation on the basis of the odourant composition of roasted coffee headspace, Flavour Fragrance J. 16 (3) (2001) 180–190. [14] P. Semmelroch, G. Laskawy, I. Blank, W. Grosch, Determination of potent odourants in roasted coffee by stable isotope dilution assays, Flavour Fragrance J. 10 (1995) 1–7. [15] P. Semmelroch, W. Grosch, Studies on character impact odorants of coffee brews, J. Agric. Food Chem. 44 (2) (1996) 537–543.
[16] J. Baggenstoss, L. Poisson, R. Kaegi, R. Perren, F. Escher, Coffee roasting and aroma formation: application of different time-temperature conditions, J. Agric. Food Chem. 56 (14) (2008) 5836–5846. [17] Juerg Baggenstoss, Coffee Roasting and Quenching Technology− Formation and Stability of Aroma Compounds, Dissertation, Diss ETH No. 17696, 2008. [18] Moon, H.S. Yoo, T. Shibamoto, Role of roasting conditions in the level of chlorogenic acid content in coffee beans: correlation with coffee acidity, J. Agric. Food Chem. 57 (12) (2009) 5365–5369. [19] S. Schenker, Investigations on the Hot Air Roasting of Coffee Beans, Dissertation, ETH Zürich; Thesis/Dissertation No. 13620, 2000. [20] S. Schenker, C. Heinemann, M. Huber, R. Pompizzi, R. Perren, F. Escher, Impact of roasting conditions on the formation of aroma compounds in coffee beans, J. Food Sci. 67 (1) (2002) 60–66. [21] R. Dorfner, R. Zimmermann, C. Yeretzian, A. Kettrup, Laser ionization mass spectrometry (REMPI-TOFMS) for on-line analysis in food science: coffee-roasting and headspace experiments, in resonance ionization spectroscopy, in: J.C. Vickerman, I. Lyon, N.P. Lockyer, J.E. Parks (Eds.), Band FOOD ING – 1st International Convention, The American Institute of Physics, Vienna (Austria), 1998, pp. 309–312. [22] R. Dorfner, R. Zimmermann, C. Yeretzian, A. Kettrup, On-line analysis of food processing gases by resonance laser mass spectrometry (REMPI-TOFMS): coffee roasting and related applications, in: ASIC-18ème Colloque Scientifique International Sur Le Café, Association Scientifique Internationale du Café (ASIC), Paris, 2000, pp. 136–142. [23] R. Dorfner, T. Ferge, C. Yeretzian, A. Kettrup, R. Zimmermann, Laser mass spectrometry as on-line sensor for industrial process analysis: process control of coffee roasting, Anal. Chem. 76 (5) (2004) 1386–1402. [24] Ralph Dorfner, Online-Techniken zur Echtzeit-Analyse von flüchtigen Verbindungen in Kaffee-Röstgasen, Dissertation, Technischen Universität München, 2004. [25] Romy Hertz-Schünemann, C. Dorfner, T. Yeretzian, R. Streibel, R. Zimmermann, 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) (2013) 1253–1265. [26] R. Zimmermann, H.J. Heger, C. Yeretzian, H. Nagel, U. Boesl, Application of laser ionization mass spectrometry for on-line monitoring of volatiles in the headspace of food products: roasting and brewing of coffee, Rapid Commun. Mass Spectrom. 10 (1996) 1975–1979. [27] Hendryk Czech, Claudia Schepler, Sophie Klingbeil, Sven Ehlert, Jessalin Howell, Ralf Zimmermann, Resolving coffee roasting-degree phases based on the analysis of volatile compounds in the roasting off-gas by photoionization time-of-flight mass spectrometry (PI-TOFMS) and statistical data analysis: toward a PI-TOFMS roasting model, J. Agric. Food Chem. 64 (25) (2016) 5223–5231. [28] Chahan Yeretzian, Hugues Brevard, Alfons Jordan, Armin Hansel, Werner Lindinger, On-line analysis of coffee roasting by proton-transfer-reaction mass-spectrometry (PTR/MS), in: 218th ACS National Meeting (National Meeting Acs, Ed.), ACS, National Meeting, 218th ACS National Meeting, New Orleans, 1999. [29] C. Yeretzian, A. Jordan, H. Brevard, W. Lindinger, On-line monitoring of coffee roasting by proton-transfer-reaction mass-spectrometry, in flavour release, in: D.D. Roberts, A.J. Taylor (Eds.), ACS Symposium Series 763, ACS, Washington, DC, 2000, pp. 112–123. [30] C. Yeretzian, A. Jordan, R. Badoud, W. Lindinger, From the green bean to the cup of coffee: investigating coffee roasting by on-line monitoring of volatiles, Eur. Food Res. Technol. 214 (2) (2002) 92–104. [31] N. Alexia, F. Gloess, Barbara Schönbächler, Anita Vietri, Sandra Bongers, Thomas Koziorowski, Chahan Yeretzian, Effect of roasting on the aroma of coffee. Online analysis of roasting using proton transfer reaction mass spectrometry, GIT Labor Fachz. 55 (10) (2011) 712–715. [32] Alexia N. Gloess, Anita Vietri, Sandra Bongers, Thomas Koziorowski, Chahan Yeretzian, On-line analysis of the coffee roasting process with PTR-ToF-MS: changes in flavor formation for different coffee varieties, in: ASIC-24ème Colloque Scientifique International Sur Le Café, Association Scientifique Internationale du Café (ASIC), Paris, 2012. [33] Wieland, A.N. Gloess, M. Keller, A. Wetzel, S. Schenker, C. Yeretzian, On-line process control of the roast degree of coffee, Chimia (Aarau) 66 (6) (2012) 443. [34] F. Wieland, A.N. Gloess, M. Keller, A. Wetzel, S. Schenker, C. Yeretzian, Online monitoring of coffee roasting by proton transfer reaction time-of-flight mass spectrometry (PTR-ToF-MS): towards a real-time process control for a consistent roast profile, Anal. Bioanal. Chem. 402 (8) (2012) 2531–2543. [35] Alexia N. Gloess, Anita Vietri, Flurin Wieland, Samo Smrke, Barbara Schönbächler, José A. Sánchez-López, Sergio Petrozzi, Sandra Bongers, Thomas Koziorowski, Chahan Yeretzian, Evidence of different flavour formation dynamics by roasting coffee from different origins: on-line analysis with PTR-ToF-MS, Int. J. Mass spectrom. 365–366 (2014) 324–337. [36] Andrea Romano, Lukas Fischer, Jens Herbig, Hugo Campbell-Sills, Joana Coulon, Patrick Lucas, Luca Cappellin, Franco Biasioli, Wine analysis by FastGC proton-transfer reaction-time-of-flight-mass spectrometry, Int. J. Mass spectrom. 369 (0) (2014) 81–86. [37] Veronika Ruzsanyi, Lukas Fischer, Jens Herbig, Clemens Ager, Anton Amann, Multi-capillary-column proton-transfer-reaction time-of-flight mass spectrometry, J. Chromatogr. A 1316 (0) (2013) 112–118.
A.N. Gloess et al. / International Journal of Mass Spectrometry 424 (2018) 49–57 [38] M. Groessl, S. Graf, R. Knochenmuss, High resolution ion mobility-mass spectrometry for separation and identification of isomeric lipids, Analyst 140 (2015) 6904–6911. [39] W. Baltes, Model reactions on roast aroma formation. IV. Mass spectrometric identification of pyrazines from the reaction of serine and threonine with sucrose under the conditions of coffee roasting, Zeitschrift fuer Lebensmittel -Untersuchung und -Forschung 184 (1987) 485–493. [40] R. Silwar, GC–MS investigation of aroma compounds from the reaction of cysteine and rhamnose under roasting conditions, Zeitschrift fuer Lebensmittel -Untersuchung und -Forschung 195 (2) (1992) 112–119. [41] R. Tressl, Formation of flavor compounds in roasted coffee, in thermal generation of aromas, in: T.H. Parliment, R.J. McGorrin, C.T. Ho (Eds.), Band ACS Symposium Series von 16eme Colloque Scientifique International sur le Café, American Chemical Society, Washington, DC, 1989, pp. 285–301. [42] B. Abu, I. Kanu, Prabha Dwivedi, Maggie Tam, Laura Matz, Herbert H. Hill, Ion mobility-mass spectrometry, J. Mass Spectrom. 43 (1) (2008) 1–22. [43] I. Flament, Coffee Flavor Chemistry, John Wiley and Sons, Ltd., 2002.
57
[44] Jennifer E. Kyle, Xing Zhang, Karl K. Weitz, Matthew E. Monroe, Yehia M. Ibrahim, Ronald J. Moore, Jeeyeon Cha, Xiaofei Sun, Erica S. Lovelace, Jessica Wagoner, Stephen J Polyak, Thomas O. Metz, Sudhansu K. Dey, Richard D. Smith, Kristin E. Burnum-Johnson, Erin S. Baker, Uncovering biologically significant lipid isomers with liquid chromatography, ion mobility spectrometry and mass spectrometry, Analyst 141 (2016) 1649–1659. [45] Jody C. May, Cody R. Goodwin, Nichole M. Lareau, Katrina L. Leaptrot, Caleb B. Morris, Ruwan T. Kurulugama, Alex Mordehai, Christian Klein, William Barry, Ed Darland, Gregor Overney, Kenneth Imatani, George C. Stafford, John C. Fjeldsted, John A. McLean, Conformational ordering of biomolecules in the gas phase: nitrogen collision cross sections measured on a prototype high resolution drift tube ion mobility-mass spectrometer, Anal. Chem. 86 (4) (2014) 2107–2116. [46] Jana Diab, Romy Hertz-Schünemann, Thorsten Streibel, Ralf Zimmermann, Online measurement of volatile organic compounds released during roasting of cocoa beans, Food Res. Int. 63 (2014) 344–352.