(35b) Understanding Powder Transfer in Force Feeders of Tableting Machines: A Comparison

Solid dosage forms, in particular tablets constitute the most frequent application form of active pharmaceutical ingredients (API) to patients. During tablet production, the three different processing stages, namely die filling, powder compaction, and tablet ejection, influence the tablet quality attributes to a different extent. The first stage, the die filling, is mainly defining the dose and its uniformity as it comprises the powder transfer from the blending hopper through the feeding system of the tableting machine into the dies of the rotating turret. For decades tablet press manufacturers have established a unique experience in pharmaceutical materials processing and tablet machine design however their “gut feeling” lacks a scientific basis. Although each of them have their own features and design components, only two main mechanisms for powder transfer from the hopper into the dies exist: gravity and force feeding. The latter mechanism is more popular and consists of one or more rotating paddle wheels in the feeder, which force the powder into the dies.

Supported by the increase of computational power, numerical methods that simulate the behavior of granular matter have gained increased attention also in the pharmaceutical industry within the last ten years. The discrete element method (DEM) is a numerical model that tracks the positions and velocities of individual particles through Newton’s second law of motion thus providing a high resolution of particle based systems. Although some groups investigated the powder flow in force feeders through the DEM and improved process understanding, some limitations still exist. Those include the consideration of large particle sizes and material properties that are not resembling pharmaceutical powders and the neglect of the poly-disperse particle size nature of tableting blends. In this study, light is shed on the powder flow in force feeders of different tableting machines by means of the DEM. An iterative calibration procedure allowed the computation of a model direct compression formulation. First, the powder flow for two different sets of paddle wheel shapes in one tableting machine (FETTE) is studied. Afterwards the performance of this machine is compared with another one (KIKUSUI) to identify the setup that provides best product quality in terms of API content and tablet mass.

The computations were carried out using the open source DEM code known as LIGGGHTS. The material properties mimic a low dose strength (5% (w/w) API content) direct compression formulation including one API and six excipient particle sizes. The micro-mechanical particle properties were calibrated through an iterative process whereby the powder flowability of various commonly used direct compression diluents was compared with the numerical blend. Two different lab-scale tableting machines (“FETTE”: FETTE 1200i, Fette compacting GmbH, Schwarzenbek, Germany) and (“KIKUSUI”: KIKUSUI VIRGO, Kikusui Seisakusho Ltd. Kyoto, Japan) were compared. In the FETTE, the hopper is located in the middle of the feeder on top of the first paddle wheel, known as dosing wheel. The dosing wheel is transferring the powder to two paddle wheels below, namely the filling wheel which is the first filling station and which is rotating in opposite direction of the turret, and the reverse dosing wheel. The latter constitutes the second die filling station which is rotating in same direction as the turret with 24 punch stations. The reverse dosing wheel has angular shaped blades whereas for the dosing and filling wheel two different configurations are available. The first set includes rod like shaped blades that are alternating arranged and penetrating half of the powder bed height. The second set differs in two ways. First the blade shape is angular and second the blade is of the same height as the geometry demarcation. The KIKUSUI with 18 punch stations contains only two opposed rotating paddle wheels that are, in contrast to the FETTE paddle wheels, overlapping. The feeding hopper is located above the left paddle wheel and the feeding plate contains an additional volume above the paddle wheels.

The numerical setup was adapted according to experiments which included an iterative filling of the system. Afterwards the actual high-speed tableting started by increasing the turret speed to obtain a throughput of 57,600 tablet per hour (40 rpm for FETTE and 53.3 rpm for KIKUSUI) at a paddle wheel speed of 30 rpm. The computations were carried out until in total 240 dies have been filled.

First of all, the powder flow in the force feeder of FETTE will be compared for the two different sets of paddle wheel shapes followed by a tablet quality comparison of the tableting machines FETTE and KIKUSUI.

The results show that the filling of the FETTE system itself caused particle size segregation for both paddle wheel sets that is then also found during steady state tableting. During steady state the following descending order of API content was found in different locations: for the rod like shaped paddle wheel combination: reverse dosing wheel (11.3%) > hopper (4.8%) dosing wheel (4.2%) > filling wheel (3.4%), for the angular shaped paddle wheel combination: reverse dosing wheel (14.2%) > dosing wheel (4.9%) > filling wheel (4.2%) > hopper (3.4%). These API content differences were also transferred into the die as follows.

In case of the rod like shaped paddle wheels, 85.3 mg of API mass entered the die underneath the filling followed by an additional 125.6 mg during reverse dosing wheel zone. Afterwards 5.3 mg API were scraped during dosing cam in which the lower punch was moved upwards by 3 mm to obtain a more confined powder bed that would subsequently be subjected to the compaction process. In contrast, for the angular shaped dosing and filling wheel combination more than double the API amount (175.6 mg) entered the die from filling and 120.8 mg from reverse dosing wheel zone. In other words, in the reverse dosing wheel, which was the same for both paddle wheel shape sets, a similar amount of API was transferred into the die. After scrapping of 16.8 mg API mass during dosing cam, 36.1% more API mass (279.6 mg vs 205.5 mg) compared to the rod like shaped paddle wheel set were in the filled dies. These differences between API masses can be explained by two phenomena. First a different API content in the paddle wheel zones (confer above) was found and the paddle wheel shape influenced the extent of the force feeding effect. As mentioned in the shape description, the angular shaped filling wheel was transporting the complete height of the powder bed during rotation in contrast to the rod like shaped wheel in which the blades only penetrated half of the powder bed height. Thereby more mass (angular shape) could be transported from the force feeder into the dies. As a consequence the blend mass flow rates were 32.8 g/s and 25.7 g/s for the angular and rod like shaped filling wheels, respectively. Although the final tablet masses were only of minor difference (4.3%), the API contents difference of 21.7% exceeded this value unproportionally, indicating that incorrect API dosing might constitute a quality risk. However further analysis is necessary to verify this observation in long term production (> 240 dies).

In case of the KIKUSUI tableting machine, particle size segregation during steady state tableting revealed significant different API contents throughout the feeder with 5.0%, 1.2%, and 4.3% in left, and right paddle wheel zones, and hopper, respectively. The first particles entered the die underneath the right paddle wheel zone resulting in 133.3 mg API mass. As a significantly higher API content was found in left paddle wheel zone, a 1.6-fold more API amount (216.2 mg) compared to right paddle wheel zone was filled in the second die filling half. The API particles entered the die later thereby the majority of them were located in the top portion of the die thus they were scraped to a considerably high extent during dosing cam (67.4 mg or 19.3%). At the end, a final API mass of 282.1 mg was obtained which was higher compared to both paddle wheel sets in FETTE.

The comparison of all tableting machines gave the following order in terms of tablet mass: KIKUSUI 5.0 +/- 0.1 g, FETTE angular 4.9 +/- 0.1 g, and rod like shaped paddle wheel set 4.6 +/- 0.2 g. With respect to API content 5.5 +/- 1.0 %, 5.6 +/- 0.8%, and 4.6 +/- 1.1% were obtained for KIKUSUI, FETTE angular, and rod like shaped paddle wheels, respectively. Hence most significant differences between the tableting machines could be revealed for API dosing.

These distinctive differences in tablet mass and API content showed that the powder transfer from the hopper into the dies was significantly influenced by the force feeder design, in terms of number of paddle wheels and their shape in addition to the housing geometry. This study shows that the tablet quality is not only affected by the formulation and process conditions (e.g. operating speeds) but the tableting machine exerts a major impact. In next studies the powder flow mechanism will be analyzed more thoroughly by particles’ velocity and coloring, residence time, and size segregation comparison to provide a more detailed understanding.