Process development for recovery of crystals using DoE

Ojo, Ebenezer and Rayat, Andrea and Johnston, Andrea and Price, Chris John and Price, Chris and Florence, Alastair and Hoare, Mike (2017) Process development for recovery of crystals using DoE. In: British Association of Crystal Growth 50th Annual Conference, 2019-07-09 - 2019-07-11, Sussex Place.

[thumbnail of Ojo-etal-BACG-2017-Process-development-for-recovery-of-crystals-using]
Text. Filename: Ojo_etal_BACG_2017_Process_development_for_recovery_of_crystals_using.pdf
Final Published Version

Download (474kB)| Preview


Process design of a dead end filtration and resulting performance at industrial scale relies on small-scale data acquired with a few mL of suspended particles in a filter with a diameter of a few mm. and milliliter scale filtration systems. For rapid process development, it is important to gain early information on the filterability of the process stream regarding suspension flowability and cake compaction under different process conditions such as pressure difference, and filter pore size as a function of particle size distribution (PSD) and crystal concentration. The effect of these process variables was therefore investigated on process performance and product yield. A design of experiment (DoE) tool was also employed to identify optimum process conditions and to create a predictive model. In this work, a case study is presented on the characterisation of an active pharmaceutical ingredient (API) - paracetamol using laser diffraction and microscopic imaging techniques. The effect of changes in process parameters including pressure difference, PSD and concentration on filterability were investigated experimentally.The filtration rate was investigated at pressure driving forces of 100 to 700 mbar and for mean particle sizes ranging from micronised to granulated. The experimental design allowed the utilisation of filter capacity to be optimised and also enables predictive assessment of other process conditions. Validation of predictive model using parity plots shows acceptable agreement with regression >90% between the predicted and experimental data. The findings enhance understanding of the filtration step resulting in robust process development at laboratory scale.