A Material Science blog, written by Dr Robert Dennehy
I asked my team recently, “what makes a good crystallisation process?” but this is a decidedly complex question to answer, and one with varying correct answers.
The question arose when I was designing a crystallisation process for a particularly recalcitrant API (Active Pharmaceutical Ingredient). Once one problem had been solved, another was identified, as can often be the case within material sciences. For example, crystallisation from aqueous alcohols gave good (but not excellent) yields, but in some compositions, oiling occurred. Addressing the oiling required better control of supersaturation and slower cooling, but then the clock was ticking with respect to degradation of the API in the solvent. In the end, a process could be defined but it was like walking a tightrope and a high degree of control was needed to ensure the right quality of product was delivered in good yield. And then, of course, the inevitable happened: a new polymorph was found. It was then back to the beginning again!
However, I do think it’s useful to try and articulate what the characteristics of a good crystallisation process really are. So, I decided to try and tackle my own question. I would be interested to hear what you think makes a good crystallisation process, and whether you agree with me on my ideas below…
The first thing to say is that the crystallisation needs to deliver the API with the desired attributes. Some of these will be defined as Critical Quality Attributes (CQAs), whilst others may be needed from an API or formulation manufacturability perspective but may not find their way into the API specification.
Here’s a list of some common API attributes, it’s not exhaustive and there are many other API properties that might be measured and targeted for control, but these are some of the most frequent:
- Impurity profile
- Assay
- Solid state form
- Residual solvent content
- Colour
- Particle size
- Morphology
- Bulk density
- Powder flow
Besides the properties of the API, there are aspects of a ‘good’ process which broadly pertain to the selection of the solvent, such as:
- It is volume efficient
- It is high yielding
- Solvent can be recovered
- Environmental, Health & Safety impacts of solvent are less significant
- The API is sufficiently stable in the solvent
- The solvent can be removed on drying
The potential for API degradation is one aspect that doesn’t receive as much attention as it should. Degradation of a complex organic molecule in a solvent, particularly at temperature, should be expected. If required, more detailed kinetic studies of degradation might be conducted so the operating window for the process is well understood.
Finally, a ‘good’ crystallisation process will also have the following characteristics which are somewhat more difficult to place into a category. They are neither a function of the output properties or solely of the solvent selection. Nevertheless, they are still important:
- Cycle time is sufficiently short
- Reversibility: The solute can be taken back into solution, if needed, without isolation
- It’s simple: The fewer process steps the better
- Can be polish filtered to support GMP
- It Is tolerant of variable input
- Can be integrated back into the penultimate step of chemistry
- Provides a reproducible output
- Can be scaled easily
- A control strategy can be articulated
- Oiling-out is absent
- Filters with ease
The approach in designing a crystallisation process can and should be phase appropriate; some of these characteristics don’t need to be met for a Phase I supply. They may however, become more important for a process at multi-tonne scale in a commercial manufacturing plant. Also, it’s important to note that finding the perfect process is unlikely. Some characteristics are more important than others and might need to be traded-off so long as the API CQAs aren’t compromised.
This is my list of what I deem makes a good crystallisation process. Does it look the same as yours?
If you’re looking to identify and mitigate the risks that are halting your API success, you can take the SmartPath. SmartPath is CatSci’s digital API Risk Assessment that will highlight the risks standing in your way, with smart solutions to overcome these issues.