Salt content in pickled vegetables by near-infrared spectroscopy (NIRS)

Salt content in pickled vegetables by near-infrared spectroscopy (NIRS) — Expert summary

Significance of the topic

The salt (sodium chloride) concentration in pickled vegetables is a critical quality and safety parameter affecting taste, shelf life, microbial stability and regulatory compliance. Rapid, reagent-free methods for routine monitoring of salt are highly desirable in production and incoming/outgoing quality control because conventional wet-chemical assays are time-consuming, require reagents and generate waste. Near-infrared spectroscopy (NIRS) offers a fast, non-destructive alternative suitable for high-throughput screening and potential inline monitoring when robust calibration models are available.

Objectives and study overview

This application study evaluated the feasibility of using a bench-top NIR spectrometer to quantify salt content in ready-to-eat pickled vegetables (olives and peppers). The main goals were to: develop chemometric prediction models linking NIR spectra to reference salt values; assess model performance using an independent validation set; and demonstrate the operational workflow for rapid routine analysis in a production context.

Methodology

• Samples: 107 pickled olive samples and 45 pickled pepper samples covering a range of salt concentrations.
• Sample preparation: Samples were ground to homogenize the matrix prior to spectral acquisition.
• Spectral acquisition: Diffuse reflectance NIR spectra recorded from 1000 to 2250 nm. To address sample heterogeneity, multipoint and rotating measurements were applied for each sample.
• Reference method: Salt (NaCl) reference concentrations obtained by potentiometric titration according to Metrohm Application Note AN-T-003.
• Calibration strategy: Data split into calibration (75%) and external validation (25%) sets. Chemometric models built and assessed by standard figures of merit (R2P, SEC, SECV, SEP).

Instrumentation used

• OMNIS NIR Analyzer Solid (diffuse reflectance NIR spectrometer for solids and viscous samples), spectral range 1000–2250 nm.
• Small cup OMNIS NIR, 60 mm sample vessel and Small holder OMNIS NIR, 60 mm to ensure reproducible positioning and rotation.
• Automated multiposition measurement capability to improve reproducibility for inhomogeneous samples.
• OMNIS Software (stand-alone license) with Quant Development module for building and editing quantification models.

Main results and discussion

• Spectral information: Although NaCl lacks distinct NIR absorptions, changes in salt concentration alter water absorption features (changes in band height, width and position), enabling indirect quantification of salt by chemometric analysis of the NIR spectra.
• Model performance — pickled olives (n=107): R2P = 0.940, SEC = 0.26% (salt), SECV = 0.32%, SEP (external validation) = 0.42%. This indicates excellent predictive performance and low prediction error for olives.
• Model performance — pickled peppers (n=45): R2P = 0.885, SEC = 0.59% (salt), SECV = 0.83%, SEP = 0.86%. Performance is good but less accurate than for olives, likely reflecting greater sample heterogeneity, smaller sample set, or more complex matrix effects in peppers.
• Validation approach: A true external validation set (25% of samples) was used to estimate real-world prediction errors (SEP), enhancing confidence in model transfer to routine use.
• Operational performance: Measurements can be completed in seconds (typically <10 s per measurement), enabling high sample throughput without chemical reagents.

Benefits and practical applications

• Fast, non-destructive screening of salt in pickled vegetables, enabling near real-time decision making on the production line.
• Elimination or substantial reduction of reagents and hazardous waste compared to titration methods.
• Automated sample rotation and multipoint measurement reduce variability from heterogeneous samples, improving routine reproducibility.
• Potential integration into process control systems or laboratory automation for continuous quality control and batch release testing.
• Suitable for routine QC, incoming raw material inspection and finished product checks where a validated calibration covering the relevant concentration and matrix variability is available.

Limitations and considerations

• Indirect measurement: Because NaCl has no specific NIR bands, the method relies on correlations with water absorption and matrix-mediated spectral signatures. Robust calibrations that capture the relevant variability (matrix composition, temperature, sample physical state) are essential.
• Matrix dependence: Models developed for specific product types (e.g., olives) may not transfer well to different products (e.g., peppers) without additional calibration data.
• Sample homogeneity and presentation: Grinding and multipoint rotation were necessary here; for whole or highly heterogeneous samples additional sampling strategies are required.
• Calibration maintenance: Regular validation, drift checks and periodic model updates are recommended to maintain accuracy over time and across production lots.

Future trends and potential applications

• Calibration transfer and standardization methods to enable model sharing across instruments and sites, reducing the need for extensive local re-calibration.
• Integration of NIR measurements with process analytical technology (PAT) for inline or at-line process control and dynamic adjustment of brining operations.
• Advanced chemometrics and machine learning to improve robustness against matrix variability and to enable multi-component prediction (e.g., combined moisture, salt and acidity monitoring).
• Miniaturized and handheld NIR devices or hyperspectral imaging for spatially resolved salt mapping of heterogeneous products.
• Data fusion with orthogonal analytical methods (titration, ion chromatography) to build hybrid models that extend accuracy and traceability for regulatory or validation purposes.

Conclusion

This application demonstrates that NIRS, when combined with appropriate sample handling and chemometric modeling, provides a fast, reliable and reagent-free method to quantify salt in pickled vegetables. Models achieved strong performance for olives (R2P = 0.940, SEP = 0.42%) and good performance for peppers (R2P = 0.885, SEP = 0.86%), supporting NIR as a practical tool for routine quality control. Careful calibration design, validation and ongoing maintenance are essential to ensure consistent results across product variations and production conditions.

Reference

• Reference salt determination: Potentiometric titration according to Metrohm Application Note AN-T-003 (used as the reference method for calibration and validation).
• Instrumentation and software described in this study: OMNIS NIR Analyzer Solid, Small cup OMNIS NIR 60 mm, Small holder OMNIS NIR 60 mm; OMNIS Software with Quant Development and Stand-Alone license.