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

Significance of the topic

Rapid, accurate quantification of trace contaminants in biodiesel is crucial for process optimization, product quality control, and regulatory compliance (e.g., ASTM D6751). Trace levels of glycerol (free and bound), water, methanol and free fatty acids indicate incomplete reaction, separation, neutralization or drying steps. Implementing fast, robust analytical methods at multiple process points reduces turnaround time compared with traditional laboratory assays, enables real-time process control, and lowers chemical and operational costs.

Objectives and study overview

This application study evaluated the Thermo Scientific Antaris II FT-NIR Analyzer for rapid quantification of multiple trace contaminants in soybean-derived biodiesel (FAME). Goals were to develop chemometric calibrations able to predict concentrations of FAME, water, free fatty acids (FFA), methanol, and individual glycerides (tri-, di-, mono-) and to demonstrate analytical performance, speed advantages, and suitability for inline/online process monitoring.

Methodology

• Sample set and standards: A base biodiesel sample (soybean-based FAME) and enriched samples were combined to produce 68 standards spanning FAME content 90.1–99.9% by weight; contaminants comprised the remainder. Pure standards of FFA, methanol, tri/di/monoglycerides and glycerol were used to prepare mixtures with controlled mass additions (mass uncertainty ±0.0002).
• Spectral acquisition: FT-NIR transmission spectra collected with Antaris II over 10000–4000 cm-1, resolution 4 cm-1, 32 co-averaged scans, collection time ≈20 s. Samples run in glass cuvettes in a heated holder at 30°C; background collected between scans.
• Preprocessing and chemometrics: Spectra were mean-centered and transformed using a second derivative (Norris derivative, segment length 5, gap 5) to enhance peak definition and remove baseline/scattering. Partial Least Squares (PLS) regression models were built using TQ Analyst software. Spectral regions were selected based on concentration and spectral information; model complexity (number of PLS factors) was optimized via PRESS analysis.

Used instrumentation

• Thermo Scientific Antaris II FT-NIR Analyzer (laboratory FT-NIR configured for transmission).
• Glass transmission cuvettes and heated cuvette holder maintained at 30°C.
• TQ Analyst chemometric software (for preprocessing, PLS calibration, and validation).

Main results and discussion

• Model performance: Robust PLS models were obtained for all target components. Standard error of calibration (SEC) for components was <0.2% (weight %), and correlation coefficients exceeded 0.93 for every analyte. Representative calibration performance: total glycerol R ≈ 0.9985 (PLS factors = 8); methanol R ≈ 0.9817 (PLS factors = 4); water R ≈ 0.9926 (PLS factors = 4); FAME R ≈ 0.9992 (PLS factors = 8). RMSEC and RMSECV values were low, indicating good fit and cross-validated robustness.
• Spectral markers: Methanol variations were observable in the OH combination band around 4900–5000 cm-1 (notably a feature near 4950 cm-1), with clear separation between high and low methanol standards after second-derivative processing. Many contaminant species share functional groups, making second-derivative preprocessing essential to resolve subtle spectral differences.
• Model optimization: PRESS plots indicated that only a few PLS factors were necessary to reach the PRESS minimum (e.g., water optimized at 4 factors), demonstrating compact and stable models.
• Analysis speed and comparative advantage: The Antaris II FT-NIR provided results in about 30 seconds per sample. In contrast, GC-based methods for glycerol required roughly 40 minutes. Multiple ASTM primary methods (different instruments and consumables) would otherwise be necessary to quantify the same suite of analytes; FT-NIR can consolidate these into a single, reagent-free measurement.

Benefits and practical applications

• Real-time or near-real-time measurement: Fast turnaround enables corrective actions during production (e.g., adjusting reaction stoichiometry, improving separation or drying steps) and supports closed-loop process control.
• Single-instrument multi-analyte capability: FT-NIR replaces multiple laborious primary tests (GC, titration, centrifugation-based water tests), reducing lab workload, consumable costs, and complexity of QA workflows.
• Flexible deployment: Methods developed on the Antaris II laboratory instrument can be transferred to inline/online Antaris analyzers (including multiplexed Antaris MX configurations) using fiber-optic probes, enabling centralized monitoring of several process points with one instrument.
• Cost and throughput advantages: Faster analysis and the potential for inline multiplexing reduce sampling/transport delays and overall processing costs, improving production efficiency and product compliance.

Future trends and possibilities

• Wider adoption of inline/multiplexed FT-NIR for continuous biodiesel process monitoring, minimizing off-line sampling.
• Improved model transfer and calibration maintenance workflows to ensure robustness across plants, feedstocks, and instrument variants.
• Integration with process control systems and advanced multivariate monitoring for automated closed-loop optimization.
• Extension of calibrations to additional feedstocks (e.g., waste oils, animal fats) and to a broader set of impurities and degradation products.
• Advances in chemometric methods (adaptive models, transfer learning) and instrumentation (compact spectrometers, embedded analytics) to lower entry barriers and increase onsite deployment.
• Regulatory and standardization progress to further accept FT-NIR as a validated alternative or screening tool to primary ASTM methods for routine QA/QC.

Conclusion

The Antaris II FT-NIR Analyzer, coupled with targeted chemometric models (second-derivative preprocessing and PLS regression), demonstrated rapid, precise quantification of FAME and multiple trace contaminants in biodiesel across relevant concentration ranges. Models achieved high correlation and low error, allowing prediction in seconds rather than tens of minutes required for laboratory primary methods. The approach supports inline/online monitoring, process optimization, cost savings, and simplified testing protocols, making FT-NIR a valuable tool for biodiesel production monitoring.

Reference

• Scherer S., Kosman W., Heil C. Trace contaminant analysis in biodiesel with an Antaris II FT-NIR Analyzer. Thermo Fisher Scientific application note. Authors affiliated with Valparaiso University and Thermo Fisher Scientific. Document and copyright information as provided by Thermo Fisher Scientific (AN51544).