Crystallization is the formation of an ordered solid phase from a liquid phase. The key word here is ‘ordered’—a molecule needs to adopt a structured repeating unit to form a stable solid phase that can be considered crystalline.

There are two key steps to forming crystals: nucleation and growth. Nucleation is where the solute molecules cluster together to form a larger structure of a certain critical size, and marks the point where the first crystals have formed. This should occur through the bulk of the solution, providing the sites for further solute molecules to adhere to and grow into large, discrete particles. These particles are the products that need to be isolated from the crystallization processes. The molecules of the active pharmaceutical ingredient (API) become particles of API, which can then be isolated and taken forward into the drug’s formulation.

Nucleation and growth are both driven by supersaturation—the difference between the solubility and the concentration of the solution—which needs to be directed to enable control over the outcome of the crystallization process. Supersaturation relates to two of the main transfer processes that occur in the crystallization vessel: the mass transfer and the heat transfer.

The mass transfer dictates how the solute molecules are transferred from the liquid phase into the solid phase, and the skill of the process is in balancing the rate of mass transfer and the quality of the crystals obtained against the overall process time. Crystals may have different faces that can grow at different rates and if this growth can be controlled, the size of the particles can be regulated, as well as limiting any impurity inclusion from the crystals being grown.

The heat transfer is significant because the temperature of the solution is crucial to driving the supersaturation. Ideally, the molecules will have a high solubility at an elevated temperature and a low solubility at a low temperature. This solubility gradient can be used to generate and control the supersaturation in the solution and, as the crystals grow, to guide the nucleation and growth of these particles.

An API will be derived from a chemical reaction, usually performed in the presence of a solvent and possibly also involving a catalyst. However, the reaction process is also likely to create any number of impurities that can contaminate the target product. As the API will almost certainly be a very high-value commodity, the aim is to have that in the solid phase, with the impurities remaining in the liquid phase so that they can be filtered off. Effectively, crystallization is being used as a purification technique to separate the product from the impurities and process solvent.

Mixing during crystallization is vital in facilitating good heat and mass transfer. If the crystallization is designed appropriately, the growth and dissolution can be used advantageously to control the particle size distribution of the material. During mixing, particles experience shear forces: higher shear rates in the mixing vessel will result in too high an attrition rate; while a shear rate that is too low could cause agglomeration.

Agglomeration is often caused by a crystallization that is too rapid, and can lead to very high residual solvent content and also result in an unsatisfactory particle size distribution. Attrition results from the breakage of particles during the crystallization and can cause a mix of different particle sizes.

Generation of supersaturation
There are several methods for generating the supersaturation, which is one of the most significant considerations when devising a crystallization process. Perhaps the most favored approach is a cooling crystallization, where the starting solution is simply cooled towards the metastable limit. Applying a temperature gradient to the system avoids the need to modify any composition of the solvent matrix.
A second option is the use of an anti-solvent addition. Changing the solvent composition of the system drives down the solubility for a known solution concentration; it can be exploited similarly to cooling, but scaling it up presents more challenges.

A third option is an evaporative crystallization. Evaporating or distilling out some of the solvent increases the concentration of the solution, thereby driving the supersaturation. Although this can be laborious to undertake at scale, with careful control over the rate of solvent removal, some high-performance crystallization processes can be achieved in this way.

The final method is reactive crystallization. This is typically required in situations where the free form of the API is highly soluble, but the desired salt to be crystallized is less so. A counter ion charged to perform the salt formation is added, which increases supersaturation and drives particle growth on an established seed bed within the crystallization vessel.

Crystallization process development
The ability to control supersaturation and the rate at which it is generated in the crystallization vessel will have a direct impact on the nucleation and growth of the particles, allowing the targeting of specific particle sizes that facilitate good filtration. Two batches of the same API that have undergone two different crystallization processes can have two completely different outcomes with regards to filtration rate and drying time.

An imperfect crystallization process means the formulation does not perform as anticipated. With a well-developed crystallization process, filtration and drying work very well, resulting in a uniform particle that formulates exactly as intended.

It is important to develop a crystallization process that gives the API the optimal characteristics when the product is scaled up. The first step is to assess the solubility, explore the critical process parameters, and then develop a crystallization process that allows control over the parameters to obtain a suitable batch of particles of the API. If the solubility is not understood, then it is unlikely that the crystallization can be controlled. This initial batch of material can then be used to devise a suitable isolation strategy and then remodel the process for returning to the plant scale.

The aim is to deliver robust crystallization, which can produce particles that are consistent in their polymorphic form, with an optimal yield and impurity profile within the drug substance specification. Another important goal is to produce particles with a fully controlled size distribution and a filtration and drying protocol that can then be transferred to any larger-scale manufacturing apparatus. Any cooling or anti-solvent addition rate must also be feasible using large-scale equipment.

Solubility assessments are typically carried out at both a low and high temperature. Generally, 5 °C and 40 °C are adequate, but this can be easily modified to suit plant limitations or opportunities. ICH Class 3 solvents are usually used for obvious toxicity reasons, and a wide range of process-relevant solvents may be studied for early-stage projects. In some cases, Class 2 solvents may be particularly suitable for the isolation of a particular solid form, or for providing a particular benefit in terms of the crystal morphology.

Once the solubility of the molecule is understood, and the solvent system to use for further crystallization development has been chosen, the next step is to determine the metastable zone width (Figure 1).
This is the difference between the solubility curve and the metastable limit. Understanding the location of these two curves relative to one another allows an appropriate design space to be considered, according to whether the solvent gives a very wide or very narrow metastable zone. This helps avoid any unwanted or uncontrolled spontaneous nucleation that could cause problems when the process is scaled up.

The metastable zone width measurements are crucial to allow the development of an appropriate seeding protocol. The solution cannot be undersaturated when seeds are added, nor is it desirable to have an already established particle bed to be present as the seeds are charged. Understanding the metastable zone width also offers the appropriate strategy for a polish filtration.

Once the solubility and the metastability of the API in the chosen solvent system have been understood, the crystallization process development experiments can begin. The first series of experiments would typically be small-scale trials to identify the most appropriate crystallization pathways for the molecule in the chosen solvent or solvent mixtures. Typically, these are carried out at the scale of a few hundred milligrams to enable the rapid screening of crystallization conditions while minimizing the consumption of valuable API material.

Investigations would include, for example, initial solution concentrations, the starting solvent compositions, the final solvent compositions, and the appropriate seeding protocol. The studies focus on the conditions that are likely to succeed when the process is scaled up and give an understanding of the sort of changes that can be made at a larger scale to obtain meaningful and relevant data sets.

The crystallization pathway would also be considered; for example, whether anti-solvent should be added before or after cooling, and the most appropriate approach to produce particles with the desired properties. Once the critical parameters have been identified, scaling the process from a few hundred milligrams to the tens of grams scale can begin.

The critical parameters identified in the small-scale trials—such as the anti-solvent addition rate, the temperature at which it is added, the cooling rate, and the seeding protocol – will be varied to understand how this alters the control over the crystallization at a larger scale.

Once the crystallization has been completed, several filtration and drying options can be considered. The first is a standard Büchner filter, and a fast filtration rate using such a filter would suggest good performance upon scale up. Where the filtration could be more challenging, there are alternatives. A laboratory-scale agitated filter dryer can offer an insight into an appropriate cake deliquoring strategy, before going for a dynamic drying.

The next question is what drying regime should be employed to ensure a free-flowing powder at the end of the filtration and drying. With crystallization, it is common to find a mixture of a hydrated or solvated species along with the unbound residual solvent within the filter cake, so any residual solvent needs to be removed while retaining enough moisture to, for example, preserve a hydrated form of the API that may be targeted from the crystallization.

Every API brings its challenges to crystallization and scale-up. Although there are general steps to understanding the crystallization process—the solubility, the metastability, the chosen process solvent, and the critical process parameters—these will not be the same for all APIs. Innovative and bespoke crystallization development goes into establishing a good, robust process, and the lack of a one-size-fits-all solution means the experience of process scientists in this field is invaluable to identify the clues that screening gives towards a robust, commercial solution.