The preparation of a cup of coffee may vary between countries, cultures and individuals. Here, an analysis of nine different extraction methods is presented regarding analytical and sensory aspects for four espressi and five lunghi. This comprised espresso and lungo from a semi-automatic coffee machine, espresso and lungo from a fully automatic coffee machine, espresso from a singleserve capsule system, mocha made with a percolator, lungo prepared with French Press extraction, filter coffee and lungo extracted with a Bayreuth coffee machine. Analytical measurements included headspace analysis with HS SPME GC/MS, acidity (pH), titratable acidity, content of fatty acids, total solids, refractive indices (expressed in°B rix), caffeine and chlorogenic acids content with HPLC. Sensory analysis included visual, aroma, flavor and textural attributes as well as aftersensation. The technical differences in the extraction methods led to a higher concentration of the respective quantities in the espressi than in the lunghi. Regarding the contents per cup of coffee, the lunghi generally had a higher content than the espressi. The extraction efficiency of the respective compounds was mainly driven by their solubility in water. A higher amount of water, as in the extraction of a lungo, generally led to higher extraction efficiency. Comparing analytical data with sensory profiles, the following positive correlations were found total solids $ texture/body, headspace intensity $ aroma intensity, concentrations of caffeine/chlorogenic acids $ bitterness and astringency.

Volatile organic compounds (VOCs), emitted from green coffee beans, during coffee roasting and from a cup of coffee, were all analysed by proton-transferreaction mass spectrometry. Firstly, the headspace (HS) of green beans was investigated. Alcohols dominate the HS, but aldehydes, hydrocarbons and organic acids were also abundant. Secondly, we roasted coffee under two different conditions and monitored on-line the VOCs emitted during the process. In a first roasting series, a batch of beans was roasted. After an initial drying phase, dominated by evaporation of water and methanol, the HS concentrations of VOCs such as acetic acid, acetaldehyde, pyridine and methylbutanal rapidly increased and went through a maximum at medium roast level. In a second series, just six beans were roasted. We observed sporadic bursts of some volatiles (furans, butanal, 2,3-pentanedione), coinciding with popping sounds. Other VOCs showed smooth time-intensity profiles (pyridine, pyrazine). These experiments gave a real-time insight into the complex processes taking place during roasting. Finally, the HS of coffee extracts, prepared from beans roasted to different roast levels, were analysed. Most VOCs showed a maximum concentration at medium roast level (e.g. pentanedione, furfural, 5-methyl furfural), while others showed a gradual increase (e.g. pyrrol) or decrease (e.g. methanol).

The objective of the project is to develop on-line, real-time, and noninvasive process control tools of coffee roasting that help deliver a consistent and high-quality coffee aroma. The coffee roasting process was analyzed by direct injection of the roaster gas into a time-of-flight mass spectrometer and ionized either by resonance enhanced multiphoton ionization (REMPI) at 266 and 248 nm or vacuum ultraviolet single-photon ionization (VUV-SPI) at 118 nm. The VUV ionization scheme allows detecting mainly the most volatile and abundant compounds of molecular mass below 100 m/z, while REMPI ionizes mainly aromatic compounds of molecular mass larger than 100 m/z. Combining the compounds ionized by resonant and single-photon ionization, approximately 30 volatile organic compounds are monitored in real time. Time-intensity profiles of 10 important volatile coffee compounds were discussed in connection with their formation chemistry during roasting. Applying multivariate statistics (principle component analysis) on time-intensity traces of nine volatile coffee compounds, the roasting degree could be traced as a consistent path in the score plot of the two most significant principle components (including 68% of the total variance), for a range of roasting temperatures (200-250 degrees C).

No abstract

No abstract