Rapid in-process moisture determination on a fluid bed dryer

Importance of the topic

Real-time, in-process moisture determination is critical for optimizing drying operations in pharmaceutical solid dosage manufacturing. Fluid bed drying is widely used because it provides efficient heat transfer and uniform drying of individual particles, but failing to detect the drying endpoint accurately can cause costly over-drying or under-drying, reduced product quality, operational inefficiency and rejects. Inline near-infrared (NIR) spectroscopy with robust probe designs enables continuous monitoring and closed-loop control of dryers, reducing lab sampling, solvents and operator workload while improving process understanding and energy efficiency.

Objectives and overview of the study

The application note documents the development and validation of an inline moisture assay for a fluid bed dryer using a Thermo Scientific Antaris MX FT-NIR process analyzer equipped with a purgeable-tip diffuse-reflectance fiber-optic probe. Goals were to demonstrate accurate moisture quantification across a broad moisture range, to show how probe designs and sampling cycles mitigate common fluidized-bed sampling problems, and to compare the FT-NIR approach with traditional laboratory moisture methods.

Methodology

• Process sampling: A purgeable-tip diffuse-reflectance fiber-optic probe was installed into a fluid bed dryer; the dryer operated in a 120 s purge-dry / 10 s filter-shake cycle to produce a new data point every 2 minutes.
• Spectral acquisition: Antaris MX FT-NIR collected 12 scans per sample at 8 cm-1 resolution over 4800–10000 cm-1 (method region later specified as 4850–9980 cm-1).
• Reference method: Karl Fischer titration (primary wet chemistry) was used for calibration reference measurements; each reference sample run in duplicate (~10 min per KF result).
• Data preprocessing: Spectra were preprocessed with first derivative, Multiplicative Scatter Correction (MSC) and a Savitzky–Golay filter (7 data points, 3rd order polynomial) to correct baseline, scattering and pathlength variation due to changing particle distance and size.
• Calibration model: Partial Least Squares (PLS) regression with 5 latent factors was used to relate processed spectra to Karl Fischer moisture values. Calibration used samples spanning approximately 1–23% moisture.

Used instrumentation

• Thermo Scientific Antaris MX FT-NIR process analyzer (diffuse reflectance configuration).
• Purgeable-tip fiber-optic probe and discussion of retractable probe with internal cleaning chamber for high-moisture or sticky products.
• Process controller with Thermo Scientific RESULT software for automated probe purging, probe movement and synchronized data collection.
• Karl Fischer titrator used as the laboratory reference method.

Key results and discussion

• Spectral band selection: Method region targeted 4850–9980 cm-1 to include the water combination band near 5150 cm-1 and the 1st overtone near 7000 cm-1. The 7000 cm-1 region exhibited pronounced variation across the 1–23% moisture range.
• Model performance: RMSEC = 0.56% (calibration), correlation coefficient R = 0.995. RMSECV = 1.05% (cross-validation), Rcv = 0.981. These statistics indicate excellent linearity and predictive performance across the tested moisture range.
• Sampling and probe considerations: Fluidization causes variable pathlength and local concentration in front of the probe, and product crust or residues can contaminate measurement interfaces. Purgeable-tip probes and retractable probes with automated cleaning mitigate these issues, enabling representative continuous measurements.
• Throughput and timeliness: FT-NIR produced near real-time moisture data (new point every ~2 minutes) versus 10+ minutes per Karl Fischer sample, enabling much faster endpoint detection.

Benefits and practical applications

• Non-destructive, solvent-free, inline moisture monitoring reducing dependency on lab sampling and reagents.
• Faster detection of drying endpoint enables energy savings, reduced product over-drying and fewer rejected batches.
• Enables closed-loop process control (recipe-to-process transition) and improved process understanding across R&D and manufacturing using a common analyzer platform.
• Capable of multicomponent and at-line/online measurements for additional PAT applications (raw material ID, blend uniformity, content uniformity, API quantification).

Limitations and operational considerations

• Representative sampling in a fluidized bed is challenging; probe selection (purgeable vs retractable) depends on product stickiness and moisture range.
• Robust calibration requires appropriate reference measurements spanning expected process variation (particle size, formulation differences, temperature effects).
• Maintenance of probe purge/cleaning systems and validation of probe placement are necessary for long-term reliability.

Future trends and potential applications

Inline FT-NIR for fluid bed dryers aligns with broader PAT and Industry 4.0 initiatives. Expected developments include deeper integration with process controllers for automatic endpoint control, improved adaptive calibration strategies to handle formulation and process variability, enhanced probe designs for difficult-to-measure products, and extension of multivariate models to include blend and API quantification simultaneously with moisture. Combining NIR with data analytics and digital twins could further optimize dryer throughput and energy use.

Conclusion

The Antaris MX FT-NIR analyzer paired with a purgeable-tip or retractable probe accurately quantified moisture in a fluid bed dryer across 1–23% moisture with strong calibration statistics (RMSEC 0.56%, R = 0.995; RMSECV 1.05%, Rcv = 0.981). Inline FT-NIR delivers rapid, non-destructive moisture data that makes endpoint detection straightforward, reduces lab workload and solvents, and enables closed-loop control to improve energy efficiency and decrease the risk of product loss from over- or under-drying.

Reference

• Heil C., Thermo Fisher Scientific. Application note AN51543: Use of Antaris MX FT-NIR with purgeable/retractable probes for moisture monitoring in a fluid bed dryer.