File Name : fig. s1 delay vs. mz.png

Caption : fig. s 1 dependency of delay time (between laser pulse and extraction into the mass separator) and corresponding extracted m/z-ratio at highest intensity (transmission). diesel (din en 590) was used for providing a broader higher mass range. the red curve indicates a wide extrapolation for the shown dependency. 

File Name : fig s2.png

Caption : fig. s 2 time profiles for four different m/z-ratios and roasting times assigned to the most properly substances (58 m/z acetone, 98 m/z furfuryl alcohol, 126 m/z hydroxymethylfurfural, and 194 m/z caffeine). at each roasting time ending, the coffee in the roaster was emptied. the data shows the reproducibility of the system as each time profile shows an elongation of the previous. the data was collected with 2 hz and is not additionally smoothed or normalized, only time-shifted for matching the same starting point. the corresponding roasting degrees (colorette 4) for the depicted roasting times at 58 m/z are 167, 141, 124, 105, 92, 77, and 63, respectively (measured with colorette 3b). 

File Name : fig s3 coffee plant.png

Caption : fig. s 3: monitoring of coffee roast gas in an industrial plant. the spi/rempi-tof-ms system was coupled to a rotating drum roaster with a batch capacity of roughly 500 to 600 kg of coffee. the subplots a)- c) and b)- d) are related to each other. the subplots a) and b) show the mass traces for some major signals, whereas each line is the representation of three consecutive roasts. the measurements in a) and b) differ in roast condition and coffee composition. additionally, the roasts of b) are one hundred measurements apart from a). the subplots in c) and d) show corresponding mean mass spectra normalized to their highest signal.