Prediction of choline chloride concentration on silage

Importance of the topic

Measuring choline chloride (ChCl) on silage is critical for the livestock industry because ChCl is an essential feed additive that supports fat transport, metabolism and overall animal health. Accurate, rapid quantification of ChCl ensures correct dosing, optimizes animal growth, reduces waste, and supports quality control in feed production. Traditional wet-chemistry methods are slow, resource-intensive and destructive; therefore, faster nondestructive techniques directly applicable in production environments have high practical value.

Objectives and study overview

This application note evaluated the ability of Fourier-transform near-infrared (FT-NIR) spectroscopy, using the Thermo Scientific Antaris FT-NIR Analyzer, to predict choline chloride concentration in ground corncob silage. Key aims were to demonstrate: rapid, through-plastic-bag measurement capability; accuracy and precision comparable to conventional titration; and suitability of chemometric calibration for routine quality control.

Methodology

• Sample set: 25 production silage bags treated with choline chloride were analyzed by diffuse reflectance FT-NIR.
• Reference analysis: choline chloride concentrations were determined by a silver chloride titration (used as the reference method for calibration).
• Spectral acquisition: Antaris FT-NIR Analyzer, measurements taken at 4 cm-1 spectral resolution with 8 co-averaged scans per sample; scans were collected directly through the samples’ plastic bags (non-destructive, no unpacking or extraction required).
• Preprocessing and chemometrics: spectra were pretreated using first-derivative transformation, multiplicative scatter correction (MSC), and a Norris smooth (segment length 11, gap 3). Partial least squares (PLS) regression models were built in TQ Analyst software to relate spectral data to reference concentrations. Selected spectral regions were used to focus the model on ChCl-related variance.

Used instrumentation

• Thermo Scientific Antaris FT-NIR Analyzer (original Antaris model used for data collection).
• TQ Analyst chemometric software for preprocessing and PLS model development.

Main results and discussion

• Calibration performance: The PLS calibration produced a strong linear relationship between predicted and reference ChCl concentrations with a correlation coefficient (R) of 0.98354 and an RMSEC reported at 0.425 (units consistent with reference concentrations used).
• Prediction accuracy: The model’s residual (prediction) error corresponded to approximately ±1.6% relative prediction error, indicating high quantitative accuracy for production monitoring.
• Speed and workflow advantage: FT-NIR measurements were completed in under 6 seconds per sample, compared with 6–12 hours per sample required for conventional extraction and titration methods. Scanning through the bag eliminated sample unpacking, extraction solvents, and destructive sample preparation.
• Spectral processing: First-derivative and scatter-correction pretreatments improved the ability of the chemometric model to isolate ChCl-related spectral variance from broad, overlapping near-IR bands and physical sample variability.
• Multicomponent potential: A single FT-NIR spectrum can be used to predict multiple properties (e.g., ChCl concentration, particle-size indicators or other additives) using additional calibrated models without increasing measurement time.
• Practical considerations: The study used a relatively small calibration set (25 samples); while results are encouraging, larger and more diverse sample sets strengthen robustness across batches, packaging types and matrix variability.

Benefits and practical applications

• Rapid, at-line or near-line quality control of choline chloride dosing during feed production.
• Non-destructive analysis that preserves samples for use as feed and reduces laboratory waste and reagent costs.
• Substantial reduction in turnaround time, enabling real-time decision making and process control.
• Reduced need for highly trained personnel and hazardous reagents used in traditional titrations.
• Flexibility to expand monitoring to additional components using the same spectral acquisition and chemometric framework.

Future trends and opportunities

• Model transfer and scalability: Developing robust calibration transfer procedures (standardization or domain-adaptation techniques) to deploy models across multiple instruments and production sites.
• Expanded calibrations: Increasing training-set size and matrix diversity (different silage types, moisture, packaging materials) to improve prediction robustness and reduce susceptibility to confounding factors.
• Integration with process analytics: Combining FT-NIR predictions with process control systems for automated dose adjustments and traceability in feed manufacturing lines.
• Advanced algorithms: Applying modern machine-learning approaches (e.g., ensemble methods or nonlinear regressions) and variable-selection strategies to further enhance accuracy and interpretability.
• Portable and inline sensing: Adoption of next-generation FT-NIR or miniaturized spectrometers for true inline monitoring and higher-throughput screening directly on conveyors or filling lines.

Conclusion

FT-NIR spectroscopy with chemometric modeling provides a rapid, accurate and nondestructive alternative to traditional wet-chemistry methods for quantifying choline chloride on silage. The Antaris-based calibration achieved high correlation (R≈0.9835) and a prediction error near ±1.6%, while enabling measurements through plastic packaging in under 6 seconds per sample. This approach offers clear time, cost, safety and workflow advantages for feed producers and quality laboratories, with additional opportunities for multi-analyte monitoring and inline process implementation.

References

• Hirsch J. Prediction of choline chloride concentration on silage. Thermo Fisher Scientific application note AN50942_E (2022). Author: Jeffrey Hirsch, Thermo Fisher Scientific, Madison, WI, USA.