Analysis of green coffee beans by near-infrared spectroscopy

Before they are made suitable for consumption, raw (green) coffee beans must be roasted. For roasters, the continuous analysis of green coffee beans allows the improvement of warehouse management. This information helps roasters prioritize the order of lots to process and to optimize roasting settings for more consistency and energy efficiency. However, the time, effort, and somewhat complex workflows required when using traditional analytical methods (e.g., densimeters, water activity analyzers) can be an inconvenience.

Near-infrared spectroscopy (NIRS) is a fast, multiparameter analytical method suitable for the analysis of density, water activity, and moisture content of green coffee beans. Neither chemicals nor sample preparation are required, making NIRS easy to use in the warehouse, nearby the roaster, or in a quality control laboratory.

Up to 31 samples of green coffee beans were analyzed on a Metrohm DS2500 Solid Analyzer with the DS2500 Holder and NIRS mini sample cups (Figure 1). Green coffee beans were positioned into the NIRS mini sample cups for the analysis in diffuse reflection mode. Data acquisition and prediction model development were performed with the software package Vision Air Complete (Table 1).

Reference values for density, water activity, and moisture content were obtained with the respective primary methods. Water activity analysis followed the guideline of ISO 18787, moisture content was determined according to ISO 6673, and density was determined with a density determination set for a Precisa balance.

Table 1. Hardware and software equipment overview.

Equipment	Article number
DS2500 Solid Analyzer	2.922.0010
DS2500 Holder	6.7430.040
NIRS mini sample cups	6.7402.030
Vision Air 2.0 Complete	6.6072.208

Figure 1. A Metrohm DS2500 Solid Analyzer with green coffee beans held in a NIRS mini sample cup.

The obtained Vis-NIR spectra (Figure 2) were used to create prediction models for the different reference parameters. To verify the quality of the prediction models, the data sets for water activity and moisture were split into calibration and validation sets. A leave one out validation procedure was used for density. Correlation diagrams which display the relation between the Vis-NIR prediction and the reference values are shown in Figures 3–5 together with the respective figures of merit (FOM).

Figure 2. Selection of Vis-NIR spectra of green coffee bean samples. Data was obtained with a DS2500 Solid Analyzer. A spectra offset was applied for visualization purposes.

Figure 3. Correlation diagram and the respective FOMs for the prediction of the density of green coffee beans using a DS2500 Solid Analyzer. The lab values were determined using density determination set for a balance.

Figure 4. Correlation diagram and the respective FOMs for the prediction of water activity in green coffee beans using a DS2500 Solid Analyzer. The lab values were determined according to the guidelines in ISO 18787.

Figure 5. Correlation diagram and the respective FOMs for the prediction of moisture in green coffee beans using a DS2500 Solid Analyzer. The lab values were determined according to the guidelines in ISO 6673.

This Application Note shows the feasibility of near-infrared spectroscopy for the analysis of density, water activity, and moisture content in green coffee beans. Without using any chemicals, these quality parameters can be measured with results obtained in less than a minute. This easy-to-use method allows roasters to improve warehouse management by selecting raw beans to roast based on indicators of shelf life. Additionally, roast settings can be optimized for better product consistency and higher energy efficiency.

Contact

Metrohm Suisse SA

Industriestrasse 13
4800 Zofingen

Contact

Internal reference: AW NIR CH-0068-042023