Process implementation manager of Helsinn Chemicals Ireland, Colm Campbell, highlights how crystallisation processes can be designed to provide actives with specific bulk and physical characteristics
A core expertise developed at Helsinn Chemicals Ireland (HCI), part of Helsinn Chemical Operations, pivots around crystallisation and powder technology. This capability, which is embedded in the process development group, focuses not only on recognising the molecular and supramolecular aspects of drug substances, but also their bulk behaviour.
Bulk powder performance, is often neglected in API manufacture and development programmes. Its importance is not confined only to flow and handling in the primary and secondary manufacturing processes, but also impinges upon the performance of the active in the drug product itself. This can have important implications for process and operating costs.
Helsinn has, through direct investment or appropriate partners, assembled a suite of analytical tools for studying pharma solids at these fundamental levels,
The company’s focus involves ‘traditional process development’, where scale-up and optimisation activities are executed on pre-clinical and phase 1 synthetic methods.
However, the powder technology perspective offers an appreciation of practical powder handling issues. More importantly, it places the company in a position to supply APIs tailored to the direct needs of the customer.
This bridging of the drug substance-drug product gap can add value for speciality pharma organisations, particularly companies applying innovative formulation technology to generic actives. Helsinn’s expertise sits perfectly with this type of strategy; the company uses crystallisation and powder characterisation to tailor actives, to interface with the specific formulation technology in question.
Projects already carried out include modification of crystallisation processes to provide easier to filter active substances1 and use of sound crystallisation practice to modify an established process, affording a more granular habit that makes it easier to filter, de-liquor, dry and handle.2 The company has also published a powder characterisation study that showed bulk differences markedly affecting performance, based on data derived from splitting a process stream into two apparently identical crystallisers.2
Further literature3,4 presents more general discussion of the company’s processing and technology capabilities and illustrates how this type of analysis can improve processes, resulting in cost savings from energy reductions, yield improvements and labour savings.
This case study, which involves crystallisation process development and understanding, describes how variables in a crystallisation process likely to affect product characteristics were identified and studied. It also highlights the techniques used to predict tabletting performance for physically different materials. Table 1 summarises the key supporting technologies used.
An extensive array of powder characterisation equipment is now available and any crystallisation study should be preceded by brainstorming to determine which techniques are relevant to the particular study in question.
In this case, the project owner had manufactured several batches of the API candidate on a small scale and preliminary tabletting experience suggested that each of the batches behaved very well during processing. Helsinn’s job was to study the material, to understand why it behaved well, but also to determine what factors in the crystallisation are likely to affect the performance. This should help us with the main task in hand – the large-scale manufacture of material for clinical trials, with conservation of tabletting performance.
In this study, the company believed that dry powder rheology, particle size analysis and surface area (BET) would be the most suitable performance indicating techniques. Some rheological data is provided in table 2 for three batches of test material, while table 3 shows key particle size statistics and a representative distribution.
The rheological data provides insight into why the formulator found these batches so easy to work with and some of the key points are addressed below:
Feed rate from the hopper: Powders can behave poorly in hoppers for several reasons. If the powder is fine and co-hesive it will have high unconfined yield strength and bridge easily. It may also adhere to the hopper walls, causing funnel flow or turbulent flow. This can then lead to flooding and/or segregation.
For coarse and angular powders, like this material, there is a possibility that bridging in the hopper could occur due to the mechanical interlocking of the particles. However, as a rule coarse non-cohesive powders, such as this one, flow quite readily from a hopper.
Flow into the dies: The efficiency of filling the die is a critical part of the whole process, as it has a huge effect on subsequent process steps. Inefficient filling can lead to poor compressibility, uneven compression, excess entrained air and poor tablet strength. Non-cohesive powders would flow most consistently into the dies and thus yield the most efficient filling.
As can be seen from the aeration data, fig 1, the material aerates readily and becomes fluidised, an indication of very low cohesion. Note that in this case, the high basic flow energies are a factor of the high particle density, high bulk
density, low compressibility and high level of mechanical interlocking, and not because the material is cohesive.
Compressibility in the die: The compression process is conducted at very high stress levels, where the compression caused is a function of both removing the air from the powder as well as physically breaking up the particles and causing plastic deformation. This compression stage will depend on the efficiency of the packing in the die and the compressibility, permeability and strength of the particles. This material would be expected to exhibit good compressibility.
Tablet strength and capping: The strength of the tablet is a factor of the cohesion between particles and the success of the compression stage. A successful compression stage is made possible by removal of void space in the tablet. This will be determined by both the bulk properties and particle properties. Lower permeability of the powder could lead to air entrapment and thus capping. The aeration data, in conjunction with the highly angular crystal morphology, suggest that this material would be very permeable.
The crystallisation process involves dissolving a metal hydroxide in water at above 90oC, hot filtering to remove any of the metal carbonate and adding to an aqueous solution of an organic acid at above 90oC. The solution is then stirred for a period and cooled to below 40oC, whereupon the metal salt product is isolated by filtration and washed.
Clearly, there are many variables in this process that may, potentially, affect crystal and ultimately, bulk and tabletting properties; these include actual temperatures of the hydroxide and acid solutions, addition rate of the hydroxide solution, cooling profile, mixing and final isolation temperature. The initial experiments involved studying these key variables in isolation and then, if appropriate, investigate a Design of Experiment (DOE) to gauge multi-variable responses. Key experiments were carried out using Lasentec FBRM technology to gain insight into in situ crystallisation behaviour. However, the main response studied, from routine experiments anyway, was the d(90) – 90% of the particles are smaller than size x – in the particle size distribution.
As expected, the study suggested that some parameters were more important than others in influencing crystal and bulk properties. The addition rate of the hydroxide to the acid solution was shown to be of little importance, while low mixing regimes were capable of producing larger crystals. Cooling rates also exerted negligible influence on the particle size distribution, an observation that was predicted by the FBRM measurements, which showed that most of the particle growth finishes very early in the cooling profile. Stressing the process, for example, by adding the hydroxide solution at low temperatures, gave more profound effects.
The key outcome was that the excellent tabletting properties were maintained across a whole range of crystal shapes and sizes. This leads to confidence that, while the process showed certain sensitivities to some crystallisation parameters, these were of little relevance to rheological and more importantly, tabletting performance.
This apparent robustness was borne out by subsequent manufacturing experience, where the scale-up in the production plant furnished product that matched the early formulation performance.
We see development of crystallisation processes, tailored to the production of crystal forms and morphologies for particular formulation technologies, as a key area for adding value to off-the-shelf, generic actives, providing tailored actives, and for optimising processes.