Moisture and protein content in corn starch with NIR spectroscopy

Moisture and Protein Content in Corn Starch — NIR Application Note (AN-NIR-151)

Significance of the topic

Corn starch is a major ingredient and processing aid in the food industry, used as a gelling agent, thickener and as a molding/desiccant material in confectionery production. Moisture and protein content are critical quality attributes: moisture affects shelf-life, handling, drying operations and the performance of starch in formulations, while protein content influences functional properties, labeling and downstream processing. Rapid, reagent-free methods for these analytes enable tighter process control and faster quality release compared with conventional wet-chemical approaches.

Objectives and study overview

The study aimed to demonstrate the feasibility of near-infrared spectroscopy (NIRS) to quantify moisture and protein in corn starch rapidly and without chemicals. A dataset of 210 corn starch samples was measured to build and validate predictive models. Reference analyses were performed by Karl Fischer titration for water and the Kjeldahl method for protein. The goal was a practical at-line or QC-laboratory workflow using the OMNIS NIR Analyzer Solid.

Used instrumentation

The following instrumentation and software were employed in the study:

• OMNIS NIR Analyzer Solid (Article 2.1071.0010) — reflection-mode near-infrared spectrometer, spectral range 1000–2250 nm.
• Large holder OMNIS NIR, 100 mm (6.07402.100) — ensures repeatable sample positioning and allows rotation for multiposition scans.
• Large cup OMNIS NIR, 100 mm (6.07402.110) — sample vessel for powders and granulates measured in reflection.
• OMNIS Stand-Alone license and OMNIS Quant Development software — data acquisition and model development.
• Reference methods: Karl Fischer titration for moisture; Kjeldahl digestion for protein.

Methodology

210 corn starch samples were scanned in reflection with the OMNIS NIR Analyzer Solid across 1000–2250 nm using the large cup and holder. Automated multi-position measurements (vehicle rotation) were applied to mitigate sample heterogeneity. Spectral data were processed and quantification models were developed in OMNIS software. Model training used the measured spectra paired with laboratory reference values. An external validation set was employed to assess predictive performance and to derive figures of merit (FOM): coefficient of determination (R2), standard error of calibration (SEC), standard error of cross-validation (SECV) and standard error of prediction (SEP). Measurement time per sample is less than 10 seconds, enabling rapid throughput for routine analysis.

Main results and discussion

Both moisture and protein in corn starch were predicted with high accuracy using NIRS:

• Moisture prediction: R2 = 0.977; SEC = 0.21% (absolute); SECV = 0.23%; SEP = 0.28%. The high R2 and low errors indicate excellent agreement with Karl Fischer reference data and suitability for routine moisture screening and control.
• Protein prediction: R2 = 0.915; SEC = 0.032% (absolute); SECV = 0.033%; SEP = 0.038%. Protein was predicted with good accuracy given its typically low concentration in starch; the model shows reliable quantitative performance versus Kjeldahl reference values.

Key practical observations:

• External validation supports model robustness for independent samples, but ongoing monitoring of calibration transfer and sample variability is recommended.
• Spectral preprocessing and appropriate calibration design (covering expected compositional and physical variability) are critical to maintain accuracy, particularly for low-level analytes such as protein.
• Automated multi-position acquisition reduces the influence of nonhomogeneous packing or particle-size effects common in powders.

Benefits and practical applications

NIRS offers several advantages for corn starch quality control:

• Speed: results in seconds versus minutes to hours for wet chemistry.
• Reagent-free: eliminates chemical consumables and hazardous waste (e.g., reagents for Kjeldahl or Karl Fischer), lowering operating costs and environmental impact.
• At-line or laboratory deployment: supports rapid decision-making in production, incoming raw material checks and final product release.
• High throughput and automation-friendly: integration with sample-handling systems and OMNIS software enables routine, reproducible operation.

Future trends and potential applications

Near-infrared spectroscopy for starch analysis is likely to expand in several directions:

• Calibration transfer and standardization workflows to deploy models across multiple instruments and production sites.
• Integration with process analytical technology (PAT) for continuous monitoring during starch production and drying steps.
• Extension to additional quality attributes (e.g., amylose/amylopectin ratio, fines, ash) by expanding calibration datasets and spectral preprocessing strategies.
• Use of advanced chemometrics and machine learning methods to improve robustness against physical sample variability and to enable multi-analyte models.

Conclusion

This application note demonstrates that NIRS with the OMNIS NIR Analyzer Solid provides fast, accurate, and reagent-free quantification of moisture and protein in corn starch. The developed models achieved high R2 and low prediction errors when validated externally, supporting implementation in QC labs and at-line process control. Adoption of NIRS reduces analysis time and costs while enabling high-throughput, environmentally friendlier workflows. Successful operational use requires representative calibrations, attention to sample presentation and periodic validation to ensure sustained model performance.

References

Metrohm Application Note AN-NIR-151: Moisture and protein content in corn starch with NIR spectroscopy. OMNIS NIR Analyzer Solid measurements and model development; reference analyses by Karl Fischer titration and Kjeldahl method.