Biodiesel (FAME) Analysis by FT-IR

Significance of the topic

Biodiesel (fatty acid methyl esters, FAME) is an increasingly important renewable fuel and lubricant additive for compression-ignition engines. Rapid, robust and easily deployable analytical methods are required across the production chain: for process control in biodiesel plants (ensuring complete transesterification and removal of glycerol) and for regulatory and fuel-quality laboratories (verifying labeled blend ratios such as B2, B20, B100). Fourier-transform infrared (FT-IR) spectroscopy with attenuated total reflectance (ATR) offers a fast, low-volume, reagent-sparing approach suitable for both high-concentration FAME (B100) and blended fuels when combined with multivariate calibration.

Objectives and study overview

This application note demonstrates a FT-IR/ATR workflow to quantify FAME content in diesel/biodiesel blends. Goals include: establishing spectral markers for FAME quantification, optimizing ATR sampling geometry (path length) to remain within the linear absorbance range, developing partial least squares (PLS) chemometric calibrations consistent with ASTM guidance, and demonstrating robustness of calibrations to variability in petroleum diesel base and instrument-to-instrument transfer.

Used instrumentation

• Nicolet 380 FT-IR spectrometer with KBr beamsplitter and DTGS detector.
• Smart ARK attenuated total reflectance accessory with interchangeable ZnSe top plates (45° and 60° incidence options).
• OMNIC spectroscopy software for data acquisition.
• TQ Analyst software for chemometric (PLS) model development.
• ACLS algorithm (Thermo Scientific / Sandia) used for calibration transfer between instruments.

Methodology

• Samples covered a broad range of feedstocks and matrices (samples obtained from state agriculture office, NREL, international sources, an oil terminal and a home producer).
• Spectra collected in transmission-equivalent ATR mode: 32 scans at 4 cm-1 resolution (~40 s collection time); sample volume ≈0.4 mL; background collection performed after tri-solvent cleaning (acetone–toluene–methanol).
• Key diagnostic regions: strong carbonyl ester C=O band at ≈1750 cm-1 and C–O stretching region near 1170–1200 cm-1. The 1750 cm-1 feature is largely free of petroleum interferences; the 1170–1200 cm-1 band overlaps with variable petroleum absorptions and therefore benefits from multivariate treatment.
• Quantitation strategy: PLS regression following ASTM committee recommendations; calibration sets designed to represent matrix variability (different petroleum bases and biodiesel feedstocks).
• ATR path-length control: interchangeable top plates provide reproducible effective path lengths. Two plates were evaluated: 45° (higher sensitivity) and 60° (reduced effective path length and lower absorbance).

Main results and discussion

• Spectral signatures: FAME samples show prominent ester bands (1750 cm-1, 1170–1200 cm-1). The 1750 cm-1 band provides a robust primary marker because it is free from most petroleum interferences; the 1170–1200 cm-1 region requires modeling to compensate for variable petroleum contributions.
• Linear absorbance constraints: to maintain Beer–Lambert linearity the instrument absorbance should remain below ~1.2 AU. Using the Smart ARK plates, the 45° ZnSe plate produced stronger signals that exceeded 1.2 AU for FAME contents above about B30, while the 60° ZnSe plate kept FAME absorptions near 0.8 AU at high FAME concentrations. The 45° plate is preferable for low-concentration blends (greater sensitivity); the 60° plate is preferable for high FAME loadings to avoid saturation.
• PLS calibration performance: well-constructed PLS models with a small number of factors accurately predicted FAME content. Prediction errors reported were approximately ±0.07 (absolute percentage points) for the low-range configuration (45° plate) and ±0.10 for the high-range configuration (60° plate).
• Matrix variability and model inoculation: initial calibrations built from a single petroleum base produced poor predictions for blends made with different petroleum stocks. Including a small number of representative spectra (three samples) from the alternate petroleum base (“inoculation”) markedly improved predictions. Example improvements: samples with actual FAME of 18%, 20% and 2% were initially predicted as 15.95, 14.36 and 1.21% respectively; after inoculation predictions improved to 17.92, 20.01 and 2.03% respectively.
• Calibration transfer: inter-instrument offsets can be removed using the ACLS algorithm; a calibration built on one instrument can be transferred to another by inoculating the model with a limited set of spectra from the second instrument—an approach analogous to methods used in FT-NIR networks.
• Extended analyte potential: although the presented work focused on FAME content, the FT-IR/ATR approach can be expanded (with appropriate standards and PLS models) to quantify other constituents such as free fatty acids and residual glycerol.

Benefits and practical applications

• Speed and low sample volume: single ATR spectra in ~40 s using ≈0.4 mL make this approach suitable for high-throughput QC and terminal screening.
• Minimal sample preparation and simple cleaning procedure reduce operational burden compared with wet-chemical tests.
• Good accuracy for regulatory and blending verification when calibration sets capture petroleum matrix variability.
• Plate interchangeability provides a practical way to tailor sensitivity and avoid spectral saturation across the full range of blend concentrations without frequent instrument recalibration.
• Software tools (PLS and ACLS) support robust local calibration development and transfer across instruments and sites.

Future trends and applications

• Broader multivariate libraries: building calibration libraries that explicitly sample global petroleum variability and different biodiesel feedstocks will improve universal model performance for regulatory networks and fuel terminals.
• Expansion of analyte scope: validated models for free fatty acids, mono-/diglyceride residues, methanol contamination and glycerol would enable a single FT-IR workflow to support more aspects of production QA/QC.
• Integration with process analytics: inline or at-line ATR probes coupled to chemometrics could provide real-time monitoring of transesterification completion and downstream purification.
• Standardization and inter-laboratory transfer: wider adoption of algorithms such as ACLS and standardized inoculation procedures will simplify method transfer among laboratories and instrument models.
• Complementary techniques: combining FT-IR with orthogonal methods (e.g., GC for FAME profiling) can provide comprehensive compositional control for specialty feedstocks and regulatory compliance.

Conclusion

FT-IR spectroscopy with ATR sampling and chemometric PLS calibration is an effective, rapid and resource-efficient method for quantifying biodiesel (FAME) in petroleum diesel blends. Careful attention to ATR effective path length (choosing appropriate ARK plates to keep absorbance in the linear range), inclusion of representative petroleum-base variability in calibration sets, and use of transfer algorithms (ACLS) enable accurate predictions across sample types and instruments. With appropriate model expansion, the technique can support a wider range of quality attributes beyond FAME concentration.

References

• Bradley M., Biodiesel (FAME) Analysis by FT-IR, Thermo Fisher Scientific Application Note 51258, 2007.
• ASTM Committee D02 (work on biodiesel analysis; partial least squares recommended for blend quantitation).