Turning green coffee into roasted brown beans conjures myths, trends and desires. From the small gourmet roaster to the industrial plant, the growing consumer interest in quality and variety has increased the need to understand and influence the processes that take place when coffee is being roasted.
In complex chemical reactions during coffee roasting, precursor compounds convert not only into valuable flavor carriers but also many other substances, some of which are thought to have beneficial health effects. To be able to influence the process during roasting, for example to increase the concentration of health-promoting substances, powerful online measurement technology is needed. Photoionization mass spectrometry (PIMS) can play an important role to investigate and describe the complex roasting process online, in real time .
An analytical chemist’s view into the roasting drum
Fig. 1 Setup of the measurement system. Top left: Microprobe sampling as described in ; Bottom left: Illustration of the mass spectrometer set-up for 248 nm REMPI; Right: Photoionization mass spectrometer with variable PI source
Even before roasting, raw green coffee beans contain around 300 volatile compounds but lack both the right color and the characteristic aroma for which roast coffee is prized. The most important chemical reaction paths that occur during roasting are the Maillard reaction, the Strecker degradation and caramelization reactions. They give coffee its characteristic color and aroma.
In photoionization, photons of the UV and VUV range are used for ionization, in a way that – due to their higher ionization energy – the formation of ions from ambient gases such as nitrogen, oxygen, but also of water vapor is suppressed. Combined with the excellent time resolution of time-of-flight mass spectrometry (TOF-MS), PIMS can be seen as ideal for online monitoring and investigating processes in complex gas mixtures. PI can be performed as either single photon ionization (SPI) or resonance-enhanced multi-photon ionization (REMPI). While REMPI ionizes aromatic and, in the lower UV range, aliphatic nitrogen compounds highly selectively, SPI can generate an overview of the entire spectrum of contained organic components. Both SPI and REMPI are soft ionization methods, mainly generating molecular ions and thus simplifying the interpretation of the spectra. An appropriate choice of light source can further increase the selectivity for target analysis (see Fig. 1).
A rapid way of telling Arabica and Robusta apart
Fig. 2 Roasting can easily cleave a water molecule from the diterpenes cafestol and kahweol. Testing for dehydrokahweol is a rapid way to determine whether a coffee batch is of higher quality Arabica or only of the Robusta variety.
The pentacyclic diterpenes kahweol (m/z 314) and cafestol (m/z 316), as well as 16-O-methylcafestol (m/z 330), all of which are found in coffee, help to distinguish between the coffee varieties Arabica or Robusta. While cafestol occurs in both varieties, 16-O-methylcafestol can only be found in Robusta. In addition, the concentration of kahweol is significantly lower in Robusta than in Arabica beans. Although the volatility of these compounds is rather low, the dehydrated forms can be detected in the roasting gas (Fig. 2).
The chemical profiles of the individual roasting phases
The excellent time resolution achieved by photoionization allows the release profiles of individual compounds to be selectively investigated over the entire roasting process. One of the most interesting aspects of coffee roasting is the degradation or conversion of chlorogenic acids. In addition to a number of other compounds, thermal degradation products such as 4-vinylguaiacol, vanillin, phenols, vinylcatechol and ethylcatechol can be detected. The release profiles allow conclusions to be drawn about the reactions taking place and how they are affected by the roasting conditions.
Fig. 3 Top: Contour diagram of all SPI mass traces over the course of roasting. By systematically grouping similar mass traces using NMF it is possible to identify and quantify the relative contribution to the roasting process of the four roasting phases (white). Below: SPI mass spectra at the respective maximum of each calculated roasting phase
The observed release processes can be systematically evaluated by non-negative matrix factorization (NMF). Over the time it is roasted, coffee goes through four phases which can best be described as “evaporation”, “initial roasting”, “main roasting” and “over-roasting” . In addition to the roasting phases, the mass spectrum can also reveal the chemical representation and composition (Fig. 3).
Online prediction models as an aid to roasting
A way to make further use of what photoionization MS delivers is to feed its results back into process control . In a first step, the real-time roasting data are linked with certain offline measurements of, for example, color, phenol content or flavoring in order to create a predictive model by PLS regression. It is then possible to display the respective properties in real time (Fig. 4).
Fig. 4 Real-time prediction of the total phenol content in equivalents of gallic acid and the degree of roasting (“Colorette”). The grey area displays the predictive error (root of the mean square error) of a Monte Carlo cross validation.
A greater understanding of coffee roasting and the processes involved can form the basis for targeted interventions to create certain tastes or avoid or enhance specific properties such as the resulting polyphenol content. Real-time analysis in combination with modeling the coffee roasting process would not replace trained coffee roasters but to provide them with an additional tool to get the best out of the green bean.
Category: Process Analysis | Coffee Roasting
 Hanley, L., Zimmermann, R. (2009) Light and Molecular Ions: The Emergence of Vacuum UV Single-Photon Ionization in MS, Anal. Chem. 2009 May 29; 81, 4174-4182, DOI: 10.1021/ac8013675
 Czech, H., Schepler, C., Klingbeil, S., Ehlert, S., Howell, J., Zimmermann, R. (2016) 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. 2016 Jun 29, 64, 5223-5231. DOI: 10.1021/acs.jafc.6b01683
 Heide, J., Czech, H., Ehlert, S., Koziorowski, T., Zimmermann, R. (submitted) Towards smart on-line coffee roasting process control: Feasibility of real-time prediction of coffee roast degree and brew antioxidant capacity by single-photon ionization mass spectrometric (SPI-TOFMS) monitoring of roast gases, J Agric Food Chem
Header image: iStock.com | Nodar Chernishev, engyuzheng, xamtiw; Composing: LUMITOS
Date of publication: