Dr Michael Caves, India Business Development Manager, Malvern Instruments, in this article looks at how milling can be automated using real-time particle size measurement, and examines the potential benefits
Traditionally, pharmaceutical mills have been operated manually with reference to off-line particle size measurement. However, with the goal of greater manufacturing efficiency now at the top of the agenda, this practice is beginning to change. The FDA’s PAT initiative invites the industry to use innovative technology and better control strategies to develop processes that are more reliable and efficient — the goal being reduced risk and lower production costs. For milling, the technology available to achieve these goals is commercially proven and reduces the barriers to uptake. The specific example considered is spiral jet milling, a technology widely used in the pharma industry, but the approach is equally applicable to almost all milling processes.
Spiral jet milling
Easy to clean, and with no moving parts, spiral jet mills deliver fine materials with a narrow particle size distribution. They are highly suitable for temperature-sensitive materials and enable milling in an inert atmosphere for materials that are susceptible to oxidation. However, the principle of operation is relatively complex because such mills combine comminution and classification. Potentially this makes process optimisation more difficult. Figure 1 illustrates how a spiral jet mill works. Feed is accelerated into the flat cylindrical milling chamber using pressurised air. Air is also fed tangentially into the chamber at very high velocity via a number of injection nozzles around the circumference. The circulating air picks up the feed, promoting particle particle collisions, especially at the outer edges of the chamber where particle concentration is at its highest. At the vortex finder, in the centre of the chamber, air exits, carrying with it particles of a certain size. This size is determined by the drag and the centrifugal forces acting on individual particles. Centrifugal forces throw larger particles back out to the edges of the chamber while finer particles are entrained in the exiting air stream. At the design stage, particle size cut off can be controlled by varying a number of factors: the geometry of the milling chamber; the distance of the vortex finder from the base; the diameter of the vortex finder; and the number, diameter, and angle of the air nozzles. However, once the mill is installed there are far fewer variables that can be used to control performance. Within the manufacturing environment particle size typically is controlled by varying milling and injector pressure and/or material feed rate. Optimising these parameters within the constraint of meeting the particle size specification is the key to efficient milling.
Why real-time particle size analysis?
Batch or campaigned processing is extremely common in the pharma industry, and many mills are operated manually, in this way, with reference to off-line analysis. When faced with a new batch of material, a typical approach is to establish conditions that produce the required particle size distribution by milling small samples and then fix operating parameters. Where a process is controlled using off-line analysis, the information on which to base operating decisions is relatively limited. There may be a considerable delay between taking a sample and the return of results, and, during routine operation, pragmatism may dictate that samples are taken only once or at most twice per hour.
When establishing conditions for a new batch, delays make iteration towards a successful conclusion a lengthy process. And on an ongoing basis the lack of information makes it difficult to efficiently change operating parameters during a run. This explains why the decision is often taken simply to fix operating conditions once an in-specification sample has been produced. The problem with a fixed process is that product quality then depends on the consistency of the feed. Where batches of material are being processed this should, arguably, be a reasonable assumption, but this may not always be the case.
For instance, the sample used to set the operating conditions may not be representative of the bulk, or segregation may occur during processing. Furthermore, with this approach every new batch is a new, time and material consuming, optimisation problem. Product quality and yield are often low. In contrast, using a continuous, real-time particle size analyser to monitor the milled product greatly increases the amount of information available for control. It is possible to confirm that the specification is met all of the time, and also to instantly observe the effect of any operational changes made. In addition, the data such analysers produce can be used to automate control. So, for example, in the case of spiral jet milling, it becomes possible to automatically vary feed rate to maintain the particle size specification.
Automated milling solutions
Example data from a fully automated milling solution is shown in Figure 2. The top dark blue line shows the feeder output, the light blue line shows Dv90 (particle size below which 90 per cent of the volume of particles exists), and the orange line Dv50. The set point for the system is defined in terms of a Dv50 and shown in green. At the start of this period of operation the mill is running in steady state at a set point of three microns. The measured Dv50 is within +/- 0.3um of the set point. When the set point is changed to four microns the system quickly increases the feed rate until the particle size distribution reaches the new target. In the next part of the trial the set point is returned to three microns and in the last part raised once more to six microns. Each new set point is reached extremely quickly, with the fast response of the PID control loop bringing the Dv50 to within 10 per cent of set point in ~10 sec (in each case, as quickly as the dynamics of the system will allow). At each set point the mill runs extremely steadily and produces a very consistent product. The mill is under extremely tight control and runs at optimum efficiency.
Automating mill operation optimises process efficiency, an increasingly important goal in pharma manufacture. It ensures consistently high product quality and maximum mill throughput, simultaneously minimising the requirement for manual analysis for control purposes and for QC, and moving operation closer to the goal of real-time release. The feed rate to a spiral jet mill is automatically varied on the basis of the particle size data to give very tight process control. The success of the solution is evident from trial data, which shows rapid transition between different set points, the speed of change limited only by the fundamental dynamics of the system. Automating control in this way results in an integrated, highly efficient solution for micronisation and milling applications that is beneficial at both pilot and commercial scale.