(409e) Effect of Stirrer Design on the Homogeneity of a Pharmaceutical Powder Mixture

Pharmaceutical industry, as well as food, plastic or cement industries produces tablets, capsules or packets that are currently mixtures of 5 to 15 ingredients, and may contain several active ingredients. Standards have been developed to estimate the quality of the mixtures with respect to each of the actives and authorize the release of the products on the market. As recycling is generally not allowed in such processes, products that are failing at this stage are destroyed, which of course increases the prices of the products and indirectly induces a social cost. This makes mixture homogeneity assessment a key factor, also enhanced by the recent "Process Analytical Technologies" (PAT) initiative from the FDA. Because of their great adaptability to process qualification and control, continuous processes may be considered to form part of this small revolution. In addition, when referring to process qualification, continuous mixers are offering a key advantage with respect to batch mixers during the scale-up procedure. In fact, a full-scale pharmaceutical process is qualified if its performance has been verified at one-tenth of the industrial scale. This is a tedious problem as there is no guarantee that what has been proved with a 100 l mixer will still hold for a 1 m³ batch vessel. Conversely, continuous mixer qualification will just necessitate a full-scale validation of, for example, 1 hour if the industrial production time of a certain product is 10 hours.

When considering a continuous mixing process, the factors to taken into account can be listed as: the physical and surface properties of the particulate solids, the geometrical and dynamical properties of the mixer, or the accuracy of the dosage system. Basically, a continuous mixer aims to depress dosage fluctuations and to put the different products into contact. In particular, the choice of the geometry and size of the stirring device has to take into account the properties and use of the powders to mix. For "fragile" powders, one has to employ a stirrer that may not provoke particle attrition, while for cohesive powders, the mobile may enhance shear forces to counterbalance interparticle forces. From another hand, mobile configuration is responsible for the enhancement of radial mixing and/or axial mixing … and even so, mixture segregation. Anyway, all this is still very intuitive as there is no universal method for mixer calculation, design and scale-up. In this work, we focus on quantifying the influence of stirrer design on the continuous mixing process for a pilot-scale apparatus. We experimentally compare the effect of the mobile type in terms of mixture quality, but also on the agitation conditions that reigns inside a mixer under the form of correlations between the different variables of the problem. This will be presented for a real OTC drug containing 3 actives for a total of 9 ingredients, one active being present at a low dosage (0,5 %).

The pilot scale continuous mixer Gericke GCM 500® has been used in this study. The mixer itself is a hemi-cylindrical tank that can be classified in the convective mixer’s family as the motion of the particles is due to the action of a mobile that rotates inside the bulk. The stirrer can be of two very different types: paddles mounted on a frame with internal screw (mobile A or paddles mounted on a shaft (mobile B). Depending on the rotational speed, mobile B can induce a fluidised flow or a dense phase flow, while mobile A, probably most suitable for fragile products, only provokes dense flow. For instance, the two stirrers will be compared in the dense flow regime. The dosage system is made of three loss-in-weight feeders, which are in fact classical screw feeders, each being placed on a balance.

As stated above, only the data obtained in the dense phase flow regime can be used to compare the two different stirrer designs. It is found that the mass retained in the mixing chamber for stirrer B is higher than that with stirrer A. It can be said that A has a higher transport capacity than B, what can be explained by the inclination of the paddles, as well as by the presence of the internal screw. For the model mixture, an empirical linear relationship can be found to correlate the hold-ups, meaning that mobile B will allows residence times -at least- 20% higher than mobile A. From the viewpoint of mixture’s homogeneity, this may be a drawback for mobile B. In continuous mixing, a good mixer is the one that has a high capacity of dispersion in a small volume, or for a small mean residence time. Mobile B will have to compensate its highest hold-up by a highest dispersion capability. In addition, a general relationship is proposed that correlates hold-ups with operating variables (rotational speed, flow rate) : M = aN + kQ^b, for which a seems to be independent of the stirrer employed.

For continuous mixing processes, the assessment of mixture homogeneity must be performed at the outlet of the vessel either by considering the random sampling of n samples over the whole production time or by considering n consecutive samples for a given time period. In the second case, there is a chance to detect "local" defaults that cannot be detected by random sampling. However, there is also the risk to give a bad estimation because defaults can exist at a higher scale than that of the n consecutive samples. In this work, we consider the second strategy by placing a series of striated boxes on a conveyor belt at the outlet of the mixer, thus defining samples when the belt is operated. The conveyor also has adjustable speed, which in concordance to the flow rates used, can give rise to samples of desired sizes. The sample compositions are determined by HPLC for the through an industrially validated protocol.

All the mixtures have been analyzed in terms of pharmaceutical standards, based on the coefficient of variation, the comparison of the estimated mean with the real mean and the presence of important deviations in the compositions of the samples. All the mixtures produced are far from passing this severe test, which means that process optimization is not a trivial task. The main conclusion is that under given experimental conditions, mobile A always gives better results in terms of mixture homogeneity than mobile B, as soon as the flow regimes are the same. It is suggested that stirrer design may be improved to be more efficient from the viewpoint of dispersion and transport capacities.

Perspectives of this study will principally emphasize the issues concerned with the control of a continuous mixing process: sensor development, sensor model (chemiometrics) to convert the signal onto mixture homogeneity, mixing flow model to link homogeneity with agitation, sensibility to operating conditions to act on process variables in real time.