Digital Design of a Robust Continuous Crystallization Process: Using Mechanistic Modelling Tools to Minimize Material Requirements at the R&D Stage

Crystallization is a key unit operation in the isolation of many active pharmaceutical ingredients. The crystallization process commonly has a significant impact on the quality of drugs and the efficiency of downstream processes such as filtration, milling, centrifugation, drying, granulation, and tableting (Eren et al., 2019). One of the key challenges in industry is the integration of Quality-by-Design methods when designing unit operations and processes. This would help to provide potential ways to improve quality while reducing variability and is being actively encouraged by pharmaceutical regulatory agencies (Allison et al., 2015). Continuous manufacturing methods, such as continuous crystallization, are seen as a means of achieving that by reducing process costs and maximizing operational efficiency (Brown et al., 2018).

Continuous crystallization has been effectively shown to work for several compounds using various manufacturing technologies and scales, such as MSMPR, PFR and MSMPR cascade (Zhao et al., 2014, Jolliffe et al., 2018). One of the challenges of applying continuous crystallization in the pharmaceutical industry is adhering to the associated limitations on time and materials. This is especially difficult when the effects of lesser-understood phenomena are considered, such as secondary nucleation, fouling, or agglomeration. The uncertainty around this can be mitigated by using mechanistic modelling and group contribution theory at different stages of the process development workflow to provide accurate estimations of key parameters and physical properties. The aim of this work is to enhance and re-define a continuous crystallization workflow using a mechanistic modelling approach at an earlier stage and smaller scale of observation in order to limit the amount of material used for experimental data gathering, yet allows for the development of continuous crystallization processes.

A recent study demonstrated the use of a systematic, science-based workflow that aims to counter the uncertainty in crystallization processes by reducing the risk from an early stage, which in turn allows for the development of robust and consistent continuous crystallization processes (Brown et al., 2018).

However, the material requirements still need to be addressed and is a concern where the active ingredient being manufactured is expensive or has limited availability. This work focuses on 3 different molecules – paracetamol, fenofibrate and lovastatin – and looks at ways to revise the model in the following ways:

1. Stages 2 and 3 will use Statistical Associating Fluid Theory (SAFT) via SAFT-γ-Mie Equation of State (EoS) to screen and select suitable solvents in which to perform crystallisation.
2. Stage 5 will be conducted on micro-scale (Crystal16 and Crystalline) to reduce material consumption and gather sufficient data for crystallization model development and validation.
3. Stage 6 will be revised to include a model enhancement and validation step to predict crystallization kinetic properties.
4. Global Systems Analysis and Optimization steps beyond Stage 7 will be included to allow for process improvements in-silico and predicting scale-up options, further reducing material usage for process development.

These changes allow for an even further reduction in material requirements, which will lead to an increased efficiency and productivity in pharmaceutical process development.

Key Results

Experimental data from a paracetamol and 3-methyl-1-butanol system was collected from a series of experiments at micro-scale (Crystalline from Technobis Crystallization Systems). The results were subsequently used to calibrate a mechanistic crystallizer model. With reference to the solution concentration and PSD data presented in Figure 2 collected at 1L scale, the resulting model was shown to provide a good prediction of the crystallization behaviour at the larger scale. Figure 2 also shows close agreement in model predictions for the crystallization mechanistic models calibrated using microscale data (red lines) and 1L scale data (yellow lines), suggesting that the microscale experiments are sufficient for validating crystallization kinetic parameters, thus reducing the material usage when compared to estimating parameters at a larger scale (1L).

References

• Allison, G. et al. (2015) 'Regulatory and quality considerations for continuous manufacturing. May 20-21, 2014 Continuous Manufacturing Symposium'. J Pharm Sci, 104 (3), pp. 803-812.
• Brown, C.J. et al. (2018) 'Enabling precision manufacturing of active pharmaceutical ingredients: workflow for seeded cooling continuous crystallisations'. Molecular Systems Design & Engineering, (3), pp. 518.
• Eren, A. et al. (2019) 'Development of a Model-Based Quality-by-Control Framework for Crystallization Design'. 29th European Symposium on Computer Aided Process Engineering, Pt a, 46 319-324.
• Jolliffe, H. et al. (2018) 'Process modelling, design and technoeconomic evaluation for continuous paracetamol crystallisation'. 28th European Symposium on Computer Aided Process Engineering, 43 1637-1642.
• Simon, L. et al. (2018) 'Crystallization process monitoring and control using process analytical technology'. Process Systems Engineering For Pharmaceutical Manufacturing, Vol 41, 41 215-242.
• Zhao, L. et al. (2014) 'From discovery to scale-up: alpha-lipoic acid : nicotinamide co-crystals in a continuous oscillatory baffled crystalliser'. Crystengcomm, 16 (26), pp. 5769-5780.