Results and discussion
Comparison of in-line SFV and off-line LD particle-size measurements.
The GSD of the end products measured in-line using SFV was compared with the off-line determined LD granule sizes for all
DOE batches. Although SFV and LD are based on different measurement principles (LD assumes spherical particles, whereas SFV
does not), similar GSDs were expected. Figure 1 displays the average D50 values measured with SFV (i.e., the average of the
last granulation minute, or six data points) and LD for the end granules of all 19 DOE batches. Similar D50 differences between
the 19 experiments were obtained by the two particle-sizing techniques. However, D50 sizes measured with LD were always lower
than those obtained by SFV. The same observations were made for the D10 and D90 values.
Figure 1: Average D50 results obtained with spatial-filter velocimetry (SFV) and laser diffraction (LD) for all design-of-experiments
batches. (ALL FIGURES ARE COURTESY OF THE AUTHORS)
The authors believe that this difference in GSD is caused by the LD measurement technique. The quantified particles experience
rapid accelerations as the air stream passes through a venturi. The high shear applied during this process and the subsequent
collisions with the wall of the apparatus may cause granules to break or crumble. Pressurized air also disperses the granules
during the SFV measurements, but the particles pass directly through the measurement zone. No collisions occur because high
shear is not applied. The authors' hypothesis was confirmed by the GSD measurement of low-friability spherical granules (i.e.,
Cellets 100, 200, and 350, Pharmatrans Sanaq Pharmaceuticals) using SFV and LD under software and experimental settings identical
to those for the DOE granules. The LD measurements did not systematically underestimate the GSD, in contrast to observations
obtained for the breakable granules. An additional explanation for the discrepancy between SFV and LD values might be found
in the assumption of a spherical shape during LD measurements.
Figure 2a: Differences between average D50 values measured with spatial-filter velocimetry (SFV) and laser diffraction (LD),
arranged according to increasing SFV particle size.
Size segregation during fluidization should also be addressed because it influences in-line SFV measurements (23). Inappropriate
fluidization can cause a high amount of larger granules to be present in the lower part of the chamber and a high amount of
smaller granules to appear in the upper part of the chamber. The SFV probe is placed in the upper part of the chamber, which
the largest particles cannot reach under low inlet airflow rates, potentially resulting in an underestimation of granule size.
Figures 2a and 2b display the difference between SFV and LD D50 results for the 19 DOE batches, arranged according to increasing
SFV end-granule size and LD end-granule size in the x axes, respectively. Because these differences did not increase as a function of increasing end granule size, the authors
concluded that no size segregation had occurred.
Figure 2b: Differences between average D50 values measured with spatial-filter velocimetry (SFV) and laser diffraction (LD),
arranged according to increasing LD particle size.
These primary results suggest that although a systematic difference exists between LD and SFV data, the SFV technique can
successfully measure the actual particle-size distribution during granulation.