The On-Line Monitoring of Powder Blending in a Bin Blender

Importance of the Topic

The uniformity of powder blends is a critical control point in the manufacture of solid oral dosage forms because blend homogeneity strongly influences final tablet content uniformity and potency. Traditional laboratory sampling (thieving) and off-line HPLC analysis are labor-intensive, time-consuming, potentially unsafe for potent APIs, and suffer from sampling reproducibility issues. On-line, real-time monitoring methods support PAT and QbD objectives by enabling faster endpoint determination, improved process understanding, and reduced risk of batch failures.

Objectives and Study Overview

This application note evaluates the use of a miniature MEMS-based near-infrared (NIR) analyzer (Antaris Target) as an on-line tool to determine blend endpoint and confirm blend uniformity directly on a rotating bin blender. The work demonstrates a calibration-free strategy based on a moving-window standard deviation of spectra to indicate convergence to homogeneity, and assesses operational factors (blender speed, fill level, component placement) that influence endpoint detection.

Methodology

The study blended formulations containing acetaminophen (APAP) and common excipients (lactose, microcrystalline cellulose—Avicel, and crospovidone) at various API loadings (2%, 15%, 60%, 70% w/w). The principal dataset discussed is a 70% APAP batch run at 15 RPM with approximately 251 rotations per blending cycle. Spectral acquisition and processing workflow:

• Spectral range captured: 1350–1800 nm, 0.5 nm data spacing (~901 points per spectrum).
• Each recorded spectrum was a co-average of 6 scans; one spectrum per rotation was acquired.
• Data collection was triggered by an on-board accelerometer when the material in the blender contacted the sapphire window located in a modified bin lid.
• Blend homogeneity was assessed by computing a moving-window standard deviation across whole spectra: a 5-rotation window was used, with pointwise averaging then summation of per-wavelength standard deviations to yield a single homogeneity metric per window.
• Spectral convergence (including second-derivative peak behavior) was visually and quantitatively monitored to identify endpoint.

Used Instrumentation

Instrumentation and key specifications reported:

• Antaris Target MEMS NIR analyzer (Thermo Fisher): spectral range 1350–1800 nm, semiconductor tunable laser source, 1 nm resolution Fabry–Pérot tunable filter, no moving parts, scan time ~100 ms, battery powered.
• Sapphire viewing window integrated into modified bin lid; effective optical spot diameter ~40 mm (chosen to represent material amount equivalent to a 600 mg dosage unit).
• Bohle bin blender employing counter-current blending; fill levels 60–90% and rotation rates 15–32 RPM were tested; both symmetric and asymmetric loading strategies were evaluated.
• Data handling: spectra exported as text and processed in Microsoft Excel for the moving-window standard deviation calculations; RESULT software (Antaris Target) can automate these calculations.

Main Results and Discussion

Key findings from the case study:

• The moving-window standard deviation metric decreases as blending proceeds and spectra converge; this behavior corresponds to increasing homogeneity and is visible both in raw and derivative spectra.
• The Antaris Target provided robust, vibration-insensitive on-line spectra via accelerometer-triggered acquisition without the need for mechanical limit switches.
• Blender size, rotation speed within the tested range, and initial active placement had minimal influence on the observed endpoint timing; blender fill level did affect the endpoint.
• The approach eliminates the need for application-specific chemometric concentration calibrations for blend-uniformity endpoint detection because the method monitors spectral variability rather than absolute concentration.
• Practical benefits include reduced labor, faster decision-making, avoidance of hazardous thieving operations for potent APIs, and potential for real-time process control.

Benefits and Practical Applications

The Antaris Target-based moving-window SD approach offers several practical advantages:

• Calibration-free indication of blend homogeneity simplifies deployment across formulations and reduces time-to-implementation.
• Real-time, on-line monitoring supports PAT initiatives and can shorten production cycles by identifying endpoints without off-line assays.
• Non-contact measurement through a window reduces contamination and operator exposure risks.
• Compact, battery-powered, accelerometer-triggered design enables retrofitting to existing bin blenders with minimal mechanical integration.

Future Trends and Potential Applications

Opportunities and directions to increase utility and adoption:

• Integration of automatic endpoint thresholds and closed-loop control to stop blending when a predefined spectral-variability metric is reached.
• Extension to broader wavelength coverage or incorporation of multivariate / chemometric methods for simultaneous quantitative monitoring of multiple critical components when needed.
• Combination with imaging or spatially resolved NIR techniques to detect segregation or local inhomogeneities not visible at a single spot.
• Further validation studies across formulations, blender geometries, and industrial-scale batches to support regulatory acceptance and standard operating procedures.
• Improved software automation for real-time data visualization, alarm generation, and electronic batch records integration.

Conclusion

The Antaris Target MEMS NIR analyzer, combined with a moving-window standard deviation metric, provides a practical, calibration-light method for on-line monitoring of powder blend homogeneity in bin blenders. The technique delivers rapid, reproducible indications of blend endpoint, reduces reliance on laborious thieving and off-line assays, and aligns with PAT/QbD objectives. Operational factors such as fill level require consideration when establishing endpoint criteria, and broader validation will facilitate wider industrial adoption.

Reference

Brush P., Hirsch J., The On-Line Monitoring of Powder Blending in a Bin Blender, Thermo Fisher Scientific Application Note 51115, 2006.