FT-NIR Analysis of Wine

Importance of the topic

Rapid, multi-parameter analysis of wine composition is critical for quality control across winemaking stages (harvest, fermentation, blending, bottling) and for regulatory/taxation purposes. Near-infrared (NIR) spectroscopy combined with robust chemometrics enables simultaneous, non-destructive quantification of both chemical and physical variables (ethanol, sugars, acids, density, pH, etc.) with minimal sample preparation and high throughput, making it attractive for production and QA/QC laboratories.

Objectives and study overview

This application note evaluates the Thermo Scientific Antaris FT-NIR analyzer for quantitative analysis of multiple wine parameters. Goals included demonstrating simultaneous prediction of key analytes and physical properties, developing robust calibrations using Partial Least Squares (PLS) regression, and comparing analytical performance (accuracy, precision, stability) to traditional methods.

Methodology and instrumentation

• Sample handling: wine samples were degassed when necessary and equilibrated to 40 °C; measured in 1 mm glass cuvettes mounted in a three-position heated holder.
• Spectrometer settings: Antaris FT-NIR liquid analyzer, spectral range 4000–10000 cm⁻¹, InGaAs transmission detector with a C attenuation screen, 100 co‑averaged scans at 4 cm⁻¹ resolution (maximum instrument resolution 2 cm⁻¹), 30 s pre-collection delay.
• Data acquisition and workflows: RESULT software was used for validated spectrum acquisition with simplified collection workflows for routine operation.
• Chemometrics: TQ Analyst software with PLS regression was used to build calibrations. Spectral pretreatments included second-derivative processing and baseline corrections; automatic and manual region selection targeted spectral windows with highest correlation to reference values.

Used instrumentation

• Thermo Scientific Antaris FT‑NIR analyzer (Fourier-transform NIR)
• InGaAs transmission detector
• Heated cuvette holder (1 mm pathlength glass cells)
• RESULT software for data collection and operation
• TQ Analyst for chemometric model development

Main results and discussion

Calibrations were developed for a broad set of wine parameters (ethanol, total and volatile acids, total sugars, sugar-free extract, density, pH, °Brix refraction). Key outcomes:

• Ethanol: 124 standards (4.23–27.63 % v/v). Optimal spectral regions included ~4100–4600 and 5700–6000 cm⁻¹ with second-derivative preprocessing. Final PLS model used 4 factors (chosen to avoid overfitting). Correlation coefficient R ≈ 0.9984, RMSECV ≈ 0.26 % v/v.
• Density: 133 standards (0.987–1.076 g·cm⁻3). Regions around 4502–4829 and 6038–6205 cm⁻¹ were selected with a one-point baseline. Final model used 6 PLS factors. Correlation coefficient R ≈ 0.9993, RMSECV ≈ 0.0008 g·cm⁻3.
• Other analytes: Very strong calibrations were achieved across many quality variables. Representative performance: °Brix R ≈ 0.9998 (RMSECV ≈ 0.08 °Brix); total sugars and sugar fractions R > 0.996; total and volatile acids R ≈ 0.987–0.979 (RMSECV values consistent with practical QC needs); pH showed lower correlation (R ≈ 0.95) reflecting the known challenge of predicting pH spectroscopically.

Model diagnostics (PRESS plots and cross‑validation) indicated well-defined minima in prediction error and limited overfitting when appropriate numbers of PLS factors were selected. Spectral windows with strong ethanol and water combination/overtone absorption (e.g., ethanol features near ~4400 cm⁻¹) were key contributors. Water-dominated regions with complete absorption were avoided.

Benefits and practical applications

• Throughput: analyses that previously required multiple time-consuming wet-chemical or chromatographic methods can be performed in seconds to minutes with a single spectrum.
• Multiparameter capability: one measurement predicts many compositional and physical parameters simultaneously, simplifying workflows and reducing sample consumption.
• Non-destructive and flexible sampling: long NIR pathlengths and transmission measurement enable analysis through glass/plastic containers and rapid at-line or QC-lab testing.
• Operational advantages: validated RESULT workflows reduce the need for spectroscopy expertise at the bench; robust FT instrumentation provides high scan-to-scan repeatability and method transferability.

Future trends and potential applications

• Process integration: expanding at-line or in-line FT‑NIR monitoring for fermentation control and real‑time process analytical technology (PAT).
• Calibration transfer and networked models: wider deployment across sites using standardized transfer protocols to maintain model performance between instruments.
• Advanced chemometrics and machine learning: incorporation of nonlinear or ensemble methods to improve predictions for complex or minor constituents and to better handle matrix variability.
• Miniaturization and handheld NIR: enabling field-based grape and must screening to inform harvest and blending decisions.
• Expanded analyte scope: extending calibrations to secondary metabolites, phenolic profiles, or markers of spoilage and authenticity.

Conclusion

The Antaris FT‑NIR analyzer, combined with RESULT acquisition workflows and TQ Analyst chemometrics, provides a rapid, accurate, and robust solution for simultaneous determination of multiple wine quality parameters. Calibrations demonstrated excellent statistical performance for key attributes (ethanol, density, °Brix, sugars, acids), offering a practical alternative to many traditional chemical and physical assays in quality control and process monitoring contexts.

References

1. Application Note 50813: FT‑NIR Analysis of Wine, J. Hirsch & L. Tenkl (Thermo Fisher Scientific), 2007. Thermo Fisher Scientific Antaris FT‑NIR application report.