(598f) Application of a Newly Developed Blendability Model in Determining Stages of Mixing | AIChE

(598f) Application of a Newly Developed Blendability Model in Determining Stages of Mixing


Shi, W. - Presenter, Bristol-Myers Squibb
Sprockel, O. L., Bristol-Myers Squibb

The current study presents a new technique of determining blendability and differentiating various stages of mixing using a previously developed powder mixing kinetic model [Powder Tech. 235 (2013), pp 400-404]. The model defines the resistance to mixing as the cohesive energy of a key ingredient to be mixed and the driving force of mixing as the shear energy of the mixing environment.  Using iron powder as the key ingredient and a magnetic field to manipulate the cohesive energy, it was found that the rate constant follows an Arrhenius relationship between the cohesive energy and the shear energy. After the shear energy at a given operational condition is quantified by using iron powder, iron powder is substituted with active pharmaceutical ingredients (APIs). The cohesive energy of the API is then derived with the assumption that the shear energy is maintained in the mixture when the key ingredient is at a low concentration. With the application of process analytical tools (PAT), the cohesive energy of API is further used to characterize the shear energy at various operational conditions, including fill fraction and size of the mixing equipment.  Additionally, the model shows that the mixing of powder does not always follow a single-stage process, but rather a multi-staged process. The multi-staged process is likely related to the physical change in aggregation status of powder during mixing, i.e., from the aggregated status to the non-aggregated status.  Such differentiation of these stages is dependent on the strength of aggregates and is not captured using other mixing indices. Further fine-tuning of the model is required to understand the transitional regions between these stages and to reveal other possible stages during mixing, such as demixing.  In summary, the study demonstrates that the model is capable of differentiating blendability of APIs and projecting mixing performance upon scale-up.