Quality Control: TABLETING
Simulate to Improve Tableting Processes
Move from "black art" to increased understanding
The quality by design (QbD) initiative requires each manufacturer to demonstrate understanding of its manufacturing process, but the fact that tableting is often described as a "black art" or a "black box operation" shows that there are some technical issues that still need to be resolved for some manufacturers. Even the powder compaction simulators that can be used to measure and understand the tableting process do not completely prevent inaccuracies that can lead to failure.
Sophisticated powder compaction simulators are used to study the tableting process in most major pharmaceutical companies. They allow the process parameters that affect tableting to be controlled individually so that each parameter's effect on the process can be measured. But what are the important process parameters that comprise an accurate simulation of production-scale tableting? And what parameters should have a defined range for controlling production?
The answers to these essential questions define the "design space" for this part of the manufacturing process. A selection of key considerations might include powder physical properties, powder feed method, punch draw down speed, pre-compression profile, resting time, main compression profile, resting time, ejection profile, push-off speed, die temperature, punch-to-die clearances, and humidity. It is also important to consider the parts of the process in which nothing seems to be happening; relaxation, de-gassing, and temperature transfer are important.
Different simulators allow varying levels of control of the process parameters, from full control of each parameter to only very basic speed or load control of the whole operation. Practically speaking, the momentum of a large production press turret and the stiffness of a large machine are difficult to simulate in the laboratory. Tableting forces can be high and small deflections significant. To closely control the process, the compaction simulator must be, in its effect, more powerful than the production machine to be simulated.
Visualizing the forces and speeds involved is difficult. A typical load of 15 kiloNewtons-the weight of a large car-on a 9.5 mm diameter round tablet generates a pressure of about 2,000 atmospheres, or 60,000 feet of water. Trapped air compressed to this pressure would heat to over 1,100oC adiabatically. On a large production press running at maximum speed, the main load applied to the tablet in the last 1 mm of compression happens in less than 5 milliseconds, and the dwell period for a punch head with a 10 mm diameter flat on top is only 3 milliseconds.
Understanding the measurement is the first step to understanding the process, and it requires some expertise. Production tableting is a dynamic process, and dynamic measurements are more involved than static measurements. For a rapidly changing value, the measurement is affected by the type of transducer, its mounting, the electronic conditioning, the data acquisition system, and any pre or post filtering of the signal. Loads are affected by inertia, temperatures by conduction. Any derived material properties are affected by the synchronization of the transducer signals. Unfortunately, standard calibrations will not reveal any of these potential errors.
Figure 1 (see p. 42) shows data recorded on an ESH compaction simulator. The relaxation of the load, which is considerable during the dwell periods at both the pre- and main compression stages, clearly has different phases and is time-dependent. Relaxation, or compaction, also occurs during the load buildup, and the strain rate determines how the compaction progresses, as well as the peak temperature and load that are achieved. These relaxation processes also highlight the importance of correctly simulating both the dwell timing and the timing between pre- and main compression, along with the delay before ejection.
The pressures, movements, and friction forces can only be measured on the outside of the tablet, but the materials data generated can be used to populate mathematical models, which in turn can be used to predict conditions inside the tablet, as well as to detail the pressure, temperature, and stress distributions in and around the tablet.1-2
Figure 2 (below) shows a theoretically generated result for Young's modulus relative to compaction pressure. The result is then corrected for errors in position that occur due to deformation of the punches. Then, to demonstrate data synchronization errors, a filter is applied to the position signal, causing a time shift equivalent to 250 microseconds during a fast test.
Together with punch load and punch position, die wall pressure measurements reveal a great deal of information about compaction behavior. The transfer of axial pressure into radial pressure measures the friction within the powder at the various stages of compaction, and the recovery behavior and residual radial pressure indicate the elasticity and expected ejection friction. The pressure distribution along the height of the tablet, or around a non-round shape, indicates the level of compaction variation to be expected.
Strain Rate Simulation
Strain rate for a tablet is determined by dividing the speed that the punch "tips" are closing by the separation of the punch tips. As the tablet is compressed, the punches follow an approximately sinusoidal profile. When the punches reach the dwell period at the limit of travel, the strain rate is reduced almost to zero. Compaction still occurs during the dwell period, so the load reduces and the elastic strain within the press continues to push the punch tips together. At the end of the dwell period, as the punches start to move apart, any remaining elastic energy within the press is released; this energy release keeps pushing the punches together, effectively increasing the dwell time. Then the upper punch moves away, and after a short delay as the tablet relaxes, the ejection movement starts.
At higher loads, the theoretical punch tip movement is reduced by the elasticity of the punches, the cam wheels, their bearings, and the press structure. In a small lab-scale press, the punch speed will also be changed by the dynamic reaction of the drive system to the increasing load. A "spring back" in the drive system will cause an over-speed during the following dwell period, which may hide the error in any average speed indication.
Figure 3 (see p. 45) shows the effect of drive system wind-up-assuming a transient 10% drop-off in speed-calculated from the work done. The spring back is shown in approximately 15 milliseconds. The assumed figures are mathematically derived.
The strength, power, and momentum of a large production press tend to generate higher strain rates than cam-driven laboratory presses when operated at the same theoretical punch speed. Unfortunately, this tendency generally makes production tablets more liable to capping or de-lamination than development tablets. Conversely, hydraulic simulators, controlling to calculated motion profiles, tend to generate higher strain rates than a production press, and defects are more likely to be seen on the development tablets than on production.
Figure 3 shows theoretical strain rate profiles for different types of tablet presses compared to a profile that assumes a perfectly rigid machine. Strain rate and load profiles were generated by using a simple exponential loading relationship and solving the equations for load, position, and machine compliance. The effect of compliance in the drive system of a cam-operated lab-scale press is simulated by assuming that the lost speed is due only to elastic effects. It is unlikely that any speed control system can react quickly enough to correct this error at production speeds.
The data is plotted for a typical tablet peak load of 15 kiloNewtons and estimated stiffness values for the different types of machines. To predict the strain rate errors for any particular combination of compaction simulator and production machine, the stiffness of the punches and the compacting roller assemblies should be measured or obtained from the manufacturers. The dynamic speed variation of a cam-operated simulator has a larger effect on strain rate and is the hardest to measure. A clean, unfiltered, accurate punch position signal will allow comparison with a theoretical position profile; otherwise, a high frequency velocity measurement of the turret or cam (de-pending on the design) is required.
Figure 4 (below) shows results from the production of lubricated microcrystalline cellulose tablets in a temperature-controlled punch-and-die set. The result at -3oC may be affected by condensation, but the results at +50oC and +75oC also show large differences in tablet hardness, particularly at low peak compaction pressures. Because compaction at different strain rates causes different heating effects in the powder, temperature testing and strain rate testing should be combined.
A sophisticated compaction simulator will also have the facility to replicate the die filling, ejection, and takeoff parts of the tableting process. The condition of the powder entering the die affects the compaction properties, including the spread of lubricants and the density and homogeneity of the powder stack. The ejection timing and profile affect the stress state of the tablet on exit, along with its tendency for capping, chipping, or delamination. Takeoff speed and angle, as well as the associated takeoff force, can be tested over a range of values. A range of speeds and timings can be tested to set the "knowledge space" and then the design space for these parameters.
The tableting process has been regarded as a "black art" for good reasons. It is technically challenging to test the process at production conditions before commercial quantities of powder exist, and physical differences between the development and production presses can significantly affect the process.
Implementation of QbD for tableting should force an analysis of the development processes currently used and drive some improvements forward. But if the initiative becomes a paperwork exercise made to fit around the existing process and equipment, the cost will be high and little will be gained. If the quality of the testing and knowledge of the process are increased, there will be significant savings from improved quality and reliability in production, and the QbD boxes can be checked off. �
1. Zavaliangos A, Galen S, Cunningham J, et al. Temperature evolution during compaction of pharmaceutical powders. J Pharm Sci. 2008; 97(8):3291-3304.2. Han LH, Elliot JA, Bentham AC, et al. A modified Drucker-Prager Cap model for die compaction simulation of pharmaceutical powders. Int J Solid Struct. 2008;45(10):3088-3106