Evaluation of near infrared spectroscopy for prediction of quality attributes and authentication of green coffee beans
von Adnan Adnan
Datum der mündl. Prüfung:2017-11-23
Erschienen:2020-10-14
Betreuer:Prof. Dr. Elke Pawelzik
Gutachter:Dr. Christian Möllers
Gutachter:Prof. Dr. Armin Schmitt
Zum Verlinken/Zitieren:
http://dx.doi.org/10.53846/goediss-8216
Dateien
Name:Adnan_dissertation_finalrevision - SUB.pdf
Size:1.20Mb
Format:PDF
Lizenzbestimmungen:
Zusammenfassung
Englisch
Coffee is one of the most popular beverages in the world as well as an important commodity for several exporting and importing countries, including Indonesia. There are several quality parameters of the green beans that are generally used for trading, e.g., moisture content (MC), species, origin, and defect beans. There are no general agreements on the definitions and methods for the quality measurement. However, there is a variety of analytical methods for the determination of quality parameters—for example, physical, chemical, and biological approaches. Among these approaches, near infrared (NIR) spectroscopy has the potential to serve as an alternative method for the determination of green coffee beans quality because it is fast, reliable, and accurate. Therefore, the main focus of the present study lay on the evaluation of NIR spectroscopy for its capacity to predict quality attributes and authenticity of green coffee beans. The study was divided into three parts: prediction of MC, discriminating among species, and identifying the origins of intact green coffee beans using NIR spectroscopy. The green coffee bean samples used for prediction of MC were taken from different islands in Indonesia, while the samples used for discriminating among species and identifying origins were taken from Java Island to consider the variety of environmental factors, agricultural practices, and postharvest treatments. The results of the first study showed that a three-component partial least squares regression (PLSR) model using raw spectra can fairly accurately predict MC in intact green coffee beans. It furthermore demonstrated that a simplified model based on only seven selected wavelengths opens the possibility of creating a more affordable NIR instrumentation. The second study showed that UV-Vis spectroscopy-based determination of two important compounds—i.e. caffeine and chlorogenic acid and NIR spectroscopy using 7 selected wavelengths and LDA are applicable to discriminate reliably among species. The third study showed that neither NIR spectroscopy nor 18O and 2H values were suitable for origin determination. The Sr value in the green beans, however, can potentially be used as a tracer. It may, hence, be concluded that a combination of NIR spectroscopy and multivariate analysis can predict the moisture content in intact green beans and discriminate among coffee species but is not suitable for the identification of Java coffee origins in the examined samples.
Keywords: green coffee beans; near infrared spectroscopy; moisture content; stable isotope; species determination