(367d) Automated Pseudo-Seeding for the Estimation of Crystallisation Kinetics in Small-Scale Experiments: Comparison of Model-Free and Population Balance Modelling Approaches

Crystallisation kinetics, including primary nucleation, secondary nucleation and crystal growth, can be estimated from small scale unseeded experiments. These experiments can be performed in small agitated vials and typically rely on primary nucleation at relatively high supersaturations to generate crystal suspensions which are then characterised via image analysis [1,2]. This approach is suitable for compounds that nucleate easily but less suitable for compounds that do not nucleate readily. Furthermore, as the approach relies on primary nucleation it requires higher supersaturations meaning it is difficult to estimate crystallisation kinetics at lower supersaturations, such as those found in continuous crystallisation processes. Our recent studies have instead utilised an external seeding approach to estimate crystallisation kinetics at low to moderate supersaturations [3,4]. However, external seeding has the drawback of requiring the preparation of seed crystals with sufficient size which are free from surface artefacts, such as initial breeding, and it is less amenable to automation.

We propose a novel pseudo-seeding methodology which avoids the need to prepare seed crystals externally and instead generates seed crystals in-situ by a carefully controlled partial dissolution of suspended solids present in the system. This methodology has the benefit of minimising experimental time and automating controlled dissolution to achieve the required seed loading. The resulting seed suspensions are then brought to a target temperature (and corresponding supersaturation), where ensuing crystallisation can be monitored using in situ imaging where the time evolution of crystal suspension properties can be estimated from image analysis.

Parameter estimation from small scale crystallisation experiments can be achieved with either model-based [1] or model-free [2] approaches. Both approached require tracking key crystal suspension properties (e.g., number, size, area, volume, shape) which are obtained from in situ imaging. Model-free approaches are usually based on applying simple statistical fits to single moments of the crystal size distribution to obtain the corresponding rates at the initial stages of crystallisation but can be less accurate with noisier data. Model-based approaches, usually in the form of population balance modelling, have the advantage of capturing multiple crystallisation mechanisms taking place simultaneously over longer periods of time but have the disadvantage of requiring a priori assumptions about specific functional forms for respective kinetic terms.

The pseudo-seeding methodology for the estimation of crystallisation kinetics in small-scale (< 5 mL) agitated vials has been demonstrated for crystallisation of two model compounds (mefenamic acid in 2-butanol/heptane and ibuprofen in ethanol/water solvent mixtures). We have directly compared and contrasted model-free and population balance modelling approaches using the same experimental data from both systems. For the conditions explored in this work, there was a satisfactory agreement between secondary nucleation and growth kinetics estimated from model-free and population balance modelling approaches, showing that both approaches are suitable for initial estimations of crystallisation kinetics under relatively low supersaturations where primary nucleation is insignificant.

The pseudo-seeding methodology proposed here allows for secondary nucleation and growth rates to be estimated for compounds that do not spontaneously nucleate, in a manner which is rapid, material-sparing, and amenable to automation. Therefore, this methodology would be well suited to obtain first estimates of crystallisation kinetics to inform early stages of process development in drug substance pharmaceutical manufacturing and provide initial data for crystallisation process modelling and scale-up.

References:

[1] Arruda et al., Automated and material-sparing workflow for the measurement of crystal nucleation and growth kinetics, Crystal Growth & Design, 23, 3845 (2023).

[2] Cashmore et al., Rapid assessment of crystal nucleation and growth kinetics: comparison of seeded and unseeded experiments, Crystal Growth & Design, 23, 4779 (2023).

[3] Briuglia et al., Measuring secondary nucleation through single crystal seeding, Crystal Growth & Design, 19, 421 (2019).

[4] Cashmore et al., Secondary nucleation of alpha-glycine induced by fluid shear investigated using a Couette flow cell, Crystal Growth & Design, 24, 4975 (2024).