The Analysis of Beer Components Using FT-NIR Spectroscopy

Importance of the topic

Beer is a chemically complex, widely consumed beverage whose sensory and safety attributes depend on multiple compositional parameters such as alcohol, color, refractive index and density. Routine quality control in breweries requires rapid, reliable measurements to ensure product consistency. Traditional reference methods (distillation, pycnometry, chromatography, photometry, enzymatic assays) are accurate but time-consuming, costly and often require separate assays for each analyte. Near-infrared Fourier transform spectroscopy (FT-NIR) coupled with multivariate calibration offers a fast, multi-component, no-preparation alternative suitable for at-line or laboratory quality control.

Objectives and overview of the study

The application note demonstrates feasibility and performance of FT-NIR spectroscopy for simultaneous quantification of multiple beer parameters using a transflectance sampling accessory. The study aimed to build and validate chemometric (PLS) models correlating FT-NIR spectra with reference laboratory values for alcohol content, color (EBC scale), refractive index and specific density, and to evaluate speed, accuracy and robustness of the approach for brewery QC workflows.

Used instrumentation

• Thermo Scientific Antaris FT-NIR Method Development Sampling (MDS) System
• Thermo Scientific SabIR fiber optic probe with transflectance accessory
• Spectralon reference for background measurements
• Thermo Scientific RESULT software for spectral acquisition
• Thermo Scientific TQ Analyst software for chemometric model development (PLS)

Methodology

The study acquired FT-NIR spectra from 27 beer standard samples without any sample preparation. Acquisition parameters were: spectral range 10,000–4,000 cm-1, resolution 8 cm-1, 32 co-averaged scans and a Spectralon background. Measurement used the SabIR probe with a transflectance module; single-sample acquisition time for quantitative prediction was approximately 25 seconds.

Spectral preprocessing included mean-centering and multiplicative scatter correction (MSC) as pathlength correction. Partial Least Squares (PLS) regression models were developed in TQ Analyst. Different spectral sub-regions and derivative treatments were applied per analyte to optimize signal-to-analyte correlation:

• Alcohol: 5500–4000 cm-1, first derivative
• Color (EBC): 9900–4100 cm-1, no derivative
• Refractive index: 7000–4000 cm-1, first derivative
• Specific density: 7162–4099 cm-1, second derivative

For derivative smoothing the Norris derivative filter was used with specific segment and gap settings (examples: segment 5 gap 2; segment 3 gap 2) where appropriate. Model validation relied on cross-validation diagnostics (RMSECV and PRESS plots) and external prediction statistics (RMSEP) against reference measurements.

Main results and discussion

PLS models produced excellent correlations between FT-NIR predictions and reference laboratory values for all four parameters. Key performance metrics from the calibrations were:

• Alcohol: correlation coefficient 0.99886; RMSEC 0.0430; RMSEP 0.0247; RMSECV 0.383
• Color: correlation coefficient 0.99983; RMSEC 0.0315; RMSEP 0.0793; RMSECV 0.187
• Refractive index: correlation coefficient 0.99965; RMSEC 0.132; RMSEP 0.172; RMSECV 0.367
• Specific density: correlation coefficient 0.99619; RMSEC 0.188×10^-3; RMSEP 0.306×10^-3; RMSECV 0.341×10^-3

Calibration and cross-validation diagnostics (PRESS plots) showed the expected behavior for robust models (decrease to a minimum then plateau), indicating appropriate model complexity and predictive capability. The study highlighted close agreement between FT-NIR predictions and conventional methods, with low prediction errors and fast measurement time per sample (~25 s). This supports FT-NIR as an accurate, rapid alternative for routine monitoring of these key beer quality attributes.

Benefits and practical applications of the method

• Simultaneous multi-parameter analysis from a single, no-preparation spectrum reduces lab workload and turnaround time.
• Short measurement time (~25 s) enables at-line or high-throughput laboratory QC.
• Non-destructive transflectance sampling with fiber probe supports flexible sampling formats (bottles, process streams, or sampling vials).
• High correlation to reference methods demonstrates suitability for replacement or screening prior to reference analyses, improving process control and consistency.

Practical deployment can lower operational costs by reducing reagent usage and eliminating many individual assays, while providing rapid feedback for process adjustments in brewing operations.

Future trends and possibilities for application

• Integration with process analytical technology (PAT) for real-time, in-line monitoring of fermentation and blending operations.
• Expansion of calibration libraries to cover wider style and matrix variability (different malts, adjuncts, yeast strains) and use of robust transfer methods to apply models across instruments and sites.
• Application of advanced machine learning and domain-adaptation techniques to improve model generalization and reduce need for extensive local calibration samples.
• Miniaturized and ruggedized NIR probes for brewery floor deployment and automated sampling systems to enable continuous QC.
• Combining NIR with other sensor modalities (e.g., Raman, dielectric) for comprehensive multi-parameter process control and fault detection.

Conclusion

This application note demonstrates that FT-NIR spectroscopy with a transflectance probe and PLS chemometrics can accurately and rapidly quantify alcohol, color, refractive index and specific density in beer without sample preparation. The method offers strong agreement with conventional reference techniques while providing substantial gains in speed and operational efficiency. With appropriate calibration strategy and validation, FT-NIR is well suited for laboratory and at-line quality control in brewing.

References

• Narimoto KM, Modesto de Oliveira Á. The Analysis of Beer Components Using FT-NIR Spectroscopy. Thermo Scientific Application Note 51892, 2010.
• Instrumentation and software referenced: Thermo Scientific Antaris FT-NIR MDS System, SabIR fiber optic probe, RESULT acquisition software, TQ Analyst chemometric software.