Real-Time Measurement of Granule Densification and Size in High Shear Wet Granulation: Combined Use of Focused Beam Reflectance Measurement with Drag Force Sensor Ajit S. Narang1, Brian Breza1, Kevin Macias1, Tim Stevens1, Divyakant Desai1, Sherif Badawy1, Dilbir Bindra1, 1Bristol-Myers Squibb, Co., New Brunswick, NJ Vadim Stepaniuk2, Valery Sheverev2 2Lenterra, Inc., Newark, NJ AAPS 2013 1 Purpose • Process analytical technologies (PAT) for real time monitoring and control of high shear wet granulation (HSWG) have achieved significant success in granule size distribution using focused beam reflectance measurement (FBRM). • However, granule densification is an important quality attribute that often correlates with granule porosity and drug product dissolution. • PAT tool to quantify granule densification, in parallel with size distribution, can provide complete attribute-control for the granulation processes, enabling building quality-by-design in the HSWG unit operation. • In this study, the resolution and sensitivity of a drag force flow (DFF) sensor in delineating granulation densification used concurrently with FBRM C35 probe was investigated. Methods • A placebo formulation consisting of microcrystalline cellulose, lactose monohydrate, croscarmellose sodium, and hydroxypropyl cellulose (HPC) was granulated with 40% w/w water in a 30 liter Pharma Connect granulator at impeller tip speed of 4.8 m/s and chopper speed of 1000 rpm. • Rate of granule size growth and densification were measured using in-line FBRM C35 probe and DFF sensor at different concentrations of HPC (1%, 3%, and 5% w/w). Shear Sensor • Product of Lenterra Inc. • Drag force on thin cylinder shear force • Minute deflections of the hollow pillar are detected by two optical strain gauges Drag Force Flow (DFF) Sensor (Fiber Bragg Gratings) attached on the Base inner surface of the pillar • Force and temperature measured Optical strain gauges • No moving parts, no gaps where particles Hollow pillar could be trapped • Measurement speed 500 Hz • Force as low as 1 mN can be detected Optical fibers Placement of Sensors in the High Shear Granulator DFF Sensor C35 Probe DFF Sensor • Focused beam reflectance measurement (FBRM) C35 probe for in-line measurement of chord length distribution (CLD). • DFF sensor for shear measurement. Experimental Conditions • Batches: • Test 1- HPC 1%; Test 2- HPC 3% ; Test 2- HPC 5%. • Blade RPM: 210 (4.8 m/s), chopper RPM: 1000 • Timing: • Test 1: Impeller starts – 9 s, water on- 259 s, water off- 439 s, impeller stops- 1370 s. • Test 2: Impeller starts – 10 s, water on – 250 s, water off – 432 s, impeller stops – 1333 s • Test 3: Impeller starts – 24 s, water on – 267 s, water off – 447 s, impeller stops – 1368 s • DFF Sensor • Position: 1” above the blade. • Acquisition rate: 500 Samples per second • Color convention on the plots: • Test #1 – red curve • Test #2 – green curve • Test #3 - blue curve • Light blue area – duration of water addition DFF Sensor Raw Data with Zero Correction 1% HPC batch • Increase in DFF shear during water addition and wet massing phase evident. DFF Sensor Raw Data with Zero Correction 3% HPC batch • Increase in DFF shear during water addition and wet massing phase evident. • 3% HPC provides signal differentiation from 1% HPC batch DFF Sensor Raw Data with Zero Correction 5% HPC batch • Increase in DFF shear during water addition and wet massing phase evident. • 5% HPC batch has signal different than 1% and 3% HPC DFF Sensor Time Resolved Signal Peaks due to consolidated granule impacts Continuous signal due to wet mass flow (sine fit) • Peak amplitude is proportional to the mass of the granule • Sine fit amplitude is proportional to the density of wet mass 10 Fast Fourier Transformation DC component Fundamental 10.56Hz Second harmonics Third harmonics Impeller frequency Figure 1 • High resolution data collection allows processing options such as FF transformation Amplitude of the Fundamental Harmonic 0.20 Water off 0.15 Amplitude, N Water on 0.10 0.05 Test 1 0.00 Test 2 Test 3 -0.05 0 100 200 300 400 500 600 700 800 900 1,000 1,100 1,200 1,300 1,400 1,500 Time, s • DFF sensor is ability to differentiate batches made with different HPC % w/w content as well as different stages of processing. Highest Peak Magnitude 1.4 Water off Test 1 Highest peak magnitude, N 1.2 Test 2 1.0 Test 3 Water on 0.8 0.6 0.4 0.2 0.0 0 100 200 300 400 500 600 700 800 900 1,000 1,100 1,200 1,300 1,400 1,500 Time, s • DFF sensor is able to differentiate batches made with different HPC % w/w content as well as different stages of processing. Time Dependent Histogram of Peak Amplitude Distribution: 1% HPC Time Dependent Histogram of Peak Amplitude Distribution: 3% HPC Time Dependent Histogram of Peak Amplitude Distribution: 5% HPC Sine Function Amplitude After Distribution Fitting 0.20 Water off Amplitude, N 0.15 0.10 0.05 0.00 Water on -0.05 0 100 200 300 400 500 600 700 800 900 1,000 1,100 1,200 1,300 1,400 Time, s • DFF sensor is able to differentiate batches made with different HPC % w/w content as well as different stages of processing. 1,500 Particle Size Distribution: Sieve Analysis 0.45 0.4 0.35 0.3 Normalized Amount 0.25 0.2 0.15 0.1 0.05 0 5% HPC 1500 855 3% HPC 568 303 1% HPC 165 113 Part Size (Microns) 38 • No significant difference in the particle size distribution of batches manufactured with different % w/w HPC levels. • Indicates the ability of DFF shear sensor to quantitate a binder-level related in-process attribute that is not necessarily PSD dependent. FBRM C35 Chord Length Distribution: 1% HPC FBRM C35 Chord Length Distribution: 3% HPC FBRM C35 Chord Length Distribution: 5% HPC Results Particle Size: • Differences in the rate of granule growth with different concentrations of HPC were evident in the FBRM measurement. Shear: • A high acquisition rate sensor that measures drag force on a thin cylindrical pillar provided high resolution unipolar signal, i.e., the pillar did not oscillate but deflect under an applied force and then quickly relaxed back into the equilibrium position. • Signal consisted of separate peaks, and their frequency generally synchronized in time with blades passing below the sensor. • The time-dependent periodic signal was clearly synchronized with the frequency of blades passing the sensor, and included a number of peaks of variable magnitude that may be interpreted as particle or granule impacts. Conclusions • The peak amplitudes were a function of the concentration of HPC used in the batch. • Basic statistical analysis of peak magnitudes suggested potential the development of a procedure to quantitatively characterize such parameters of the wet mass as densification, tackiness, and particle growth. • The DFF sensor was able to capture anticipated differences in wet mass consistency with different concentrations of binder.