Trace contaminant analysis in biodiesel with an Antaris II FT-NIR Analyzer

Trace contaminant analysis in biodiesel using an Antaris II FT-NIR Analyzer — concise summary

Significance of the topic

The control and rapid quantification of trace contaminants in biodiesel (FAME) are critical for process optimization, cost reduction and compliance with product quality standards (e.g., ASTM D6751). Trace levels of glycerol (free and bound), methanol, water and free fatty acids indicate incomplete reaction, separation or purification steps. A rapid, multi-component, in-line capable analytical technique reduces downtime, laboratory workload and enables fast process adjustments or closed-loop control strategies that improve yield and safety.

Study objectives and overview

This application study evaluated the capability of the Thermo Scientific Antaris II FT-NIR Analyzer to quantify multiple trace contaminants in biodiesel. Goals included demonstrating accuracy and speed versus conventional primary methods, developing multivariate calibration models for key analytes, and assessing the feasibility of transferring lab calibrations to on-line or in-line process monitoring (including multiplexing with Antaris MX).

Methodology

• Sample matrix: Biodiesel produced from soybean oil. Calibration standards were prepared by spiking a pure biodiesel base with controlled amounts of water, methanol, free fatty acids (FFA) and glycerides (tri-, di-, mono-), using pure reference reagents. A total of 68 standards covered wide concentration ranges (FAME 90.1–99.9% w/w; contaminants comprising the remainder).
• Spectral acquisition: Antaris II FT-NIR in transmission mode using glass cuvettes in a heated holder at 30 °C. Spectral range 10000–4000 cm⁻¹, resolution 4 cm⁻¹, 32 co-averaged scans, ~20 s collection time; background recorded between scans.
• Preprocessing and calibration: Spectra were mean-centered and converted to second derivative (Norris derivative, segment length = 5, gap = 5) to enhance peak definition and remove baseline/scattering effects. Partial Least Squares (PLS) regression (TQ Analyst software) was used to build quantitative models for multiple components. Spectral regions for each analyte were selected automatically/manually based on concentration-spectral information.
• Validation: Independent validation points were used to assess predictive performance across the low-concentration ranges of interest for contaminants.

Used instrumentation

• Thermo Scientific Antaris II FT-NIR Analyzer (laboratory bench instrument in this study).
• Transmission sample presentation: glass cuvettes with heated holder (30 °C).
• Software: Thermo Scientific TQ Analyst for preprocessing, PLS calibration and automatic region selection.

Main results and discussion

• Model performance: All calibrations produced high correlation coefficients (R > 0.93); many analytes had R ≈ 0.99. Reported calibration errors were small (Standard Error of Calibration < 0.2% for all components; specific RMSEC/RMSECV values reported per component in the study).
• Low-level quantification: The PLS models delivered accurate predictions in the critical low-concentration ranges typical for biodiesel contaminants. For example, total glycerol (free + bound) was predicted reliably across ~0–1% with excellent agreement between prediction and reference values; methanol was accurately predicted across ~0–2% with clear spectral signatures (notably O–H combination band changes near 4950 cm⁻¹ / 4900–5000 cm⁻¹).
• Model complexity: PRESS analysis indicated optimal PLS factor numbers were low (e.g., water model optimum at 4 factors), supporting robust, parsimonious models and reducing overfitting risk.
• Throughput advantage: The Antaris II FT-NIR provided quantitative results in ~30 seconds per sample versus ~40 minutes for gas chromatography–based assays for glycerol. Single-instrument, multi-analyte capability reduces the need for multiple ASTM methods and associated reagents/equipment.
• Spectral considerations: Many biodiesel components share functional groups, producing overlapping NIR features. Second-derivative preprocessing and multivariate PLS modeling were essential to resolve and quantify individual contaminants.

Practical benefits and applications

• Process monitoring: FT-NIR enables near real-time monitoring of conversion efficiency, separation and purification performance (glycerol carryover, residual methanol, water content, FFA presence), allowing faster corrective actions or automated closed-loop control.
• Operational efficiency: Replacing multiple primary laboratory tests with one FT-NIR method lowers analysis time, consumable costs and instrument footprint; multiplexing (Antaris MX) can further centralize measurement points for inline monitoring.
• Method transferability: Calibration models developed in the lab were demonstrated as transferable to production contexts, providing flexibility in instrument placement and sample presentation (lab-to-line potential).

Future trends and potential applications

• Inline and online deployment: Wider adoption of multiplexed FT-NIR systems (e.g., Antaris MX) for distributed, real-time measurement across production lines, enabling advanced process control and reduced product testing latency.
• Improved chemometrics: Integration of adaptive or hybrid modeling approaches and expanded spectral libraries to handle feedstock variability (different oil sources, fatty acid chain lengths) and matrix effects.
• Regulatory alignment: Development of standardized NIR-based methods and cross-validation with ASTM primary methods to support regulatory acceptance and quality assurance workflows.
• Edge analytics: Embedding model diagnostics and automated recalibration routines to maintain accuracy over time and across instrument platforms in industrial settings.

Conclusions

The Antaris II FT-NIR Analyzer combined with PLS chemometrics can rapidly and accurately quantify FAME and multiple low-level contaminants in soybean-based biodiesel. The approach delivers large time savings compared with traditional lab methods, supports in-line or at-line monitoring strategies, and can reduce operational costs and improve process control. The study demonstrates FT-NIR as a practical, multi-analyte monitoring tool for biodiesel production.

References

• S. Scherer, W. Kosman, C. Heil — Application note: Trace contaminant analysis in biodiesel with an Antaris II FT-NIR Analyzer. Thermo Fisher Scientific (application note AN51544_E, original authors and publisher as provided in the source document).