The Dark Crystal
Though crystallization is considered a “black art” in some circles, it’s still a crucial part of the API manufacturing process
Andrew Blythe-Dickens | | 7 min read | Practical
Crystallization may not be a new technique for isolating molecules, but it remains a critical part of the manufacturing process for many APIs. The technique is both effective in purifying the material and producing desired attributes for further downstream processing, including drug product formulation.
Crystallization should be seen as more than simply isolating a compound; it is an excellent way of improving quality at any stage; especially given that impurities can cause problems with subsequent onward processing. Controlled crystallization offers the opportunity to control both the polymorph and the particle morphology (the shape and size) of the resulting crystals.
Controlling particle size is most important in the final step of API manufacturing from a specification point of view. However, it is still important to consider physical properties in earlier stages of the synthesis because size can detrimentally impact processability of subsequent steps; for example, very fine particles may be difficult to filter efficiently, while suspensions of large particles may be difficult to line transfer between vessels during production.
As with particle size, controlling polymorphs is most important for the final step, but may also have implications earlier on; for instance, high levels of residual solvents or water may cause the formation of a solvate or hydrate form of an intermediate molecule that might interfere with the next chemical step.
The crystallization process
Various solid form chemistry studies are required to inform the development of an effective crystallization process. As well as a solvent screen and solubility evaluation, it is important to carry out a polymorphism study to identify whether there are any different crystal forms that might be made during crystallization, depending on the conditions. If the crystallization is of an intermediate, this study need not be too intense. However, if it is the final API, it will need to be more comprehensive because of the implications for the formulation of the drug product.
For a successful crystallization, the molecule typically needs to be soluble in one or more of the standard process solvents and be chemically robust. Other output requirements for an API, such as particle size distribution, will very much depend on what is required for a successful drug product formulation.
If a successful crystallization and suitable crystal form can be found, this may remove the need for a range of additional work. Of course, if the material is an intermediate, only solubility, purity, and some chemical stability data will be required to aid the design of the process. If the API falls into Class 1 or 2 on the Biopharmaceutical Classification System (BCS), where solubility of the drug substance is less of an issue, then a comprehensive preformulation evaluation may not be required. However, if it falls into Class 3 or 4 with low solubility or permeability, particle size control may be essential to achieve efficacy.
On some occasions, a suitable crystal form may not be achievable. Investigation into an alternative stable polymorph may be an option, which could offer alternative habits that might be suitable instead. If the molecule has one or more ionizable moieties in its structure, it is very possible that a salt version of the material could be successful; whereas if the molecule is neutral, then an alternative would be to create a cocrystal, where the API forms crystals that also incorporate a second biocompatible molecule.
Be the early bird
There are often significant advantages in determining a suitable crystallization early in the development process. Taking this approach will involve early collection of data on the material, which can be used throughout development and scale-up. With this work starting earlier – and being conducted in parallel to chemical development – it should be possible to reduce the overall development timeline. If the decision is taken to wait for “process typical” material to be manufactured instead, this work will inevitably become lengthier.
Development priorities may sometimes dictate that speed of material delivery is paramount, which can cause delays in crystallization development. In this instance, if data were collected on an ongoing basis to assist the chemical development team, it can help overcome issues and identify suitable solvents for reactions and clean-out processes. A poorly optimized process may be tolerated on a small scale, but the cost of inefficiency increases exponentially as the process is scaled up.
If the synthetic route needs to be modified, some of the crystallization development work might have to be repeated. For example, the impurity profile of the modified process may have changed and become more challenging to control or perhaps the crystallization performance has been affected by the nature and level of residual solvents.
Equally, the original isolation and crystallization process may remain appropriate – and a performance assessment will inform the decision on whether it is still effective at removing different impurities. The platform to build upon may already be there; for example, solubility data may have already been collected for alternative suitable crystallizations that perform adequately.
In addition to solvent choice, various other factors need to be considered during crystallization studies, including the equipment available. For example, a process may have been developed with vessel configurations that differ in capacity, dimensions, and agitator, and will certainly differ to those available at plant-scale. Modeling and simulation tools are available to aid in designing or translating a process to the appropriate scale that is required – and to ensure that the process still achieves the desired outcome.
Fundamentally, crystallization should always be seeded to achieve the greatest control. Adding an aliquot of seed crystals at an appropriate point will help control the final polymorph and avoid uncontrolled nucleation. The seed point is determined by solubility data and the metastable zone width curve, which is obtained via unseeded crystallization induction experiments. Importantly, the particle size of the crystals can be determined by the seeding process, which can affect whether the API meets any final particle size specification. Seeding comes with its own set of parameters to allow the efficient growth and control of crystals, including its loading concentration, the temperature it is carried out at, the shape and size of the crystal, as well as the seed surface area.
If there is only limited material available to develop the optimal crystallization, then in silico screening tools (such as COSMO-RS) can be used to predict outcomes. Automated screening instruments can also reduce the amount of material and operator time required in physical studies.
In an ideal world, development chemists will take a “right first time” approach and work alongside plant engineers, solid state chemists, formulation experts, and commercial teams to determine the optimal plan to achieve the desired results.
Alternatives and keys to success
Crystallization is not the only way in which the solid form of an API (or intermediate) can be optimized. Jet milling or micronization are other efficient ways of creating smaller particles of a homogeneous size within parameters that are suitable for making effective solid dose forms, particularly if these smaller particles improve dissolution.
Milling can have other benefits because it could remove the need to develop a crystallization process. Some form of milling is frequently the most effective way of achieving the necessary particle size, particularly if a size of 20 µm (or even lower) is required. However, a milling process will still require development and, unlike crystallization, cannot improve a product’s purity. Control of the final polymorph formation may also be lacking – and performance can be affected if there is variable input into the milling process.
When making and isolating intermediates, an alternative to a crystallization might be to telescope the process directly into the next stage without actually isolating the material. This may save time by removing the processing or crystallization step, and reducing the number of inter-batch cleaning processes that will be required; however, it increases the probability that an impurity may be carried forward into the next step, which could either interfere with that reaction, or even ones that follow later. It might also be more difficult to remove the impurity at a later stage once the desired molecule has different functional groups and properties.
Even if an alternative process can be substituted or used in addition to create smaller particle, crystallization remains the workhorse for making solid forms of APIs that are ready for formulation – and I don’t expect this to change any time soon (even if it is considered a “dark art”). What is the key to success? Addressing the crystallization process early in the development cycle offers real benefits for developers in terms of time and costs later during scale-up. Early work can also help in determining an API’s chance of success by generating consistent product throughout scale-up, which not only meets purity specifications, but also has the physical properties suitable for formulation into finished drug products. Ensuring a product has the physical properties to make handling and processing on plant is also vital to efficient operations during manufacturing. The upshot? Innovators save time and money.