Seeded growth rates of ritonavir in copovidone at 75% relative humidity (RH) and 50 °C were evaluated by single-particle tracking second harmonic generation (SHG) microscopy and found to be ∼3-fold slower for crystallites at the surface compared to the bulk. The shelf lives of final dosage forms containing amorphous solid dispersions (ASDs) are often dictated by the rates of active pharmaceutical ingredient crystallization. Upon exposure to elevated RH, the higher anticipated water content near the surfaces of ASDs has the potential to substantially impact nucleation and growth kinetics relative to the bulk. However, quantitative assessment of these differences in growth rates is complicated by challenges associated with discrimination of the two contributions (supersaturation and molecular mobility) in ensemble-averaged measurements. In the present study, "sandwich" materials were prepared, in which sparse populations of ritonavir single-crystalline seeds were pressed between two similar ASD films to assess bulk crystallization rates. These sandwich materials were compared and contrasted with analogously prepared "open-faced" samples, without the capping film, to assess the surface crystallization rates. Single-particle analysis by SHG microscopy time-series during in situ crystallization produced average growth rates of 3.8 μm/h for bulk columnar crystals with a particle-to-particle standard deviation of 0.9 μm/h. In addition, columnar crystal growth rates for surface particles were measured to be 1.3 μm/h and radiating crystal growth rates for surface particles were measured to be 1.0 μm/h, both with a particle-to-particle deviation of 0.4 μm/h. The observed appearance of radiating crystals upon surface seeding is attributed to reduced ritonavir solubility upon water adsorption at the interface, leading to higher defect densities in crystal growth. Despite substantial differences in crystal habit, correction of the surface growth rates by a factor of 4 from geometric effects resulted in relatively minor but statistically significant differences in the growth kinetics for the two local environments. These results are consistent, with viscosity being a relatively weak function of water absorption coupled with primarily diffusion-limited growth kinetics.
Keywords: active pharmaceutical ingredients; amorphous solid dispersions; crystallization kinetics; nonlinear optics; second harmonic generation.