Scale Up of a Granulation Phenomenon

A drug product's formulation processes and dissolution rates can be sensitive to changes in the physical nature of the active pharmaceutical ingredient (API). Consequently, manufacturers must ensure that the development and scale-up of the final manufacturing step of an API maintain consistency in a product's physical properties. Consistency typically is achieved by controlling the final crystallization step and then milling the dry material. Nonetheless, during process development, manufacturers also should take into account the effect of the drying process that is intermediate to the final crystallization and milling steps.

At the initial laboratory scale of product development, the final drying step is performed without agitation other than an occasional manual blending of the wet cake as it dries. In the early stages of development, material quantities typically are too limited to use dryers with the agitative environments representative of manufacturing-scale equipment. Instead, drying is performed on trays in vacuum ovens or through the filters that isolate the solid API from the crystallization liquors in the laboratory. Incorporating tray dryers at larger scales, however, is often impractical and undesirable because of the extensive manual labor that is required and the associated safety issues related to product exposure. Agitation is necessary when the final drying process for an API is scaled-up to the pilot plant and factory in alternative equipment because of the decrease in the drying material's surface area exposed to heat-transfer surfaces per unit volume. Agitation during drying also promotes homogeneity of the dry API. On the other hand, incorporating agitation in the scale-up of a drying process can lead to changes in a material's physical properties. Agitation can break solid particles into smaller particles, agglomerate particles into larger clusters of the original particles, or granulate particles into new shapes and sizes.

Granulation is conducted during many formulation processes to generate larger particles after blending an API with excipients. Several researchers have reported challenges associated with the scale-up of this unit operation (1–4). These studies focused on the granulation process as a distinct step and did not investigate the phenomena in a drying operation. Granulation also can be performed while drying an API without excipients if the solid material is deformable and the process solvent being removed during drying acts as a binder liquid.

The evolution of granule properties during a typical granulation process is controlled by three distinct stages: wetting, growth, and attrition (5). Wetting the particles is not relevant for granulation conducted while drying a material formed in a liquid environment. Granule growth depends on the deformability of the solids, the rate of mixing, and the nature and amount of liquid binder. The growth rate generally is faster with increasing liquid content and may require an induction period to consolidate the granules in the granulation device (6). If the liquid content is increased, granule attrition may occur. Figure 1 shows a general curve representing granule growth for a granulation process in which the liquid binder is added at a constant rate during granulation. Most pharmaceutical formulation granulation processes do not add liquid beyond the point where granule sizes will get smaller, but granulation during drying can start at a moisture content above this critical value.

Figure 1: Granule size progress for a typical granulation process in which binder liquid is added at a constant rate.

In this article, we aim to identify equipment that could be used for the large-scale manufacture of an API that could undergo granulation during the drying stage. This evaluation was motivated by differences in the physical properties between granulated and nongranulated versions of the solid API. The granulated material demonstrated superior solids-flow properties, including higher bulk density and reduced interparticle static forces. Granulated API facilitated formulation development of a preferred direct-compression process that was not viable with the original solid API morphology. This article focuses on the effect of equipment type and size on granulation.

Experiment

We used an API crystallized from an aqueous solvent system containing 10% 1-propanol. The crystals have a needle-like shape, which was retained during drying at the laboratory scale without agitation (see Figures 2 and 3). The filtered material that was charged to the dryers typically contained 55–75% moisture.

Figure 2: Magnified picture of particles at the start of the drying process (5.8 μm per subdivision).

Three dryer styles that were available at large scales were studied during the development of the process: rotary, conical, and agitated filter dryers. The rotary dryers were V-shaped and did not contain intensifier bars. The rotary dryer was ~2 L in volume at the laboratory scale and 140 L at the pilot-plant scale. A 4-rpm rotation speed was used at both scales.

A 15-L conical dryer with a screw-shaped agitator attached to an orbit arm was used at the laboratory scale, and a 300-L conical dryer with a comparable agitator style was tested at the pilot-plant scale. The tip speed of the agitator screw of the laboratory-scale unit was fixed at ~2 ft/s. The pilot-plant scale screw had a variable speed range of 2–50 rpm. For this work, the screw was set at the lowest value (tip speed = 5 ft/s) to match the laboratory speed as closely as possible.

Figure 3: Scanning electron microscope image of starting material (white bar = 5 μm)

Several sizes of filter dryers were evaluated, ranging from a 10-cm diameter laboratory-scale unit to a 2.25-m diameter factory-scale dryer. The diameters of the single-blade agitators were ~98% of the filter diameter in each dryer. Intermediate pilot-plant filter dryers with diameters of 0.5 and 0.8 m were used for most of the development work. At each scale, the tip speed of the agitator was maintained at ~0.4 ft/s during drying. The cycle time of mixing during drying was ~30 h. This long cycle time was imposed by a restrictive drying temperature limit required to ensure chemical stability of the API.

Results

At the laboratory scale, granulation did not occur in the rotary dryer. The needle-like particle shape shown in Figures 2 and 3 were maintained throughout drying. In contrast, drying in the 140-L pilot-plant scale dryer produced material that partially granulated during drying. Figure 4 shows microscope images of the material as it progressed through the drying process in the pilot-plant rotary dryer, and Figure 5 shows scanning electron microscope (SEM) images of the final dry particles from this dryer. These figures show that features of the original needle shape are still present on the surface of the particles. The impact from falling a relatively short distance in the laboratory dryer (~10 in.) was not enough to cause the particles to deform, but a fall of ~3 ft in the 140-L dryer did provide enough shear and energy to affect the particle shape.

Figure 4: Progression of particle shapes while drying in a pilot plant–scale rotary dryer with the original needle shape particles remaining at a moisture content of 56% (a), the appearance of some granules at 42% moisture (b), and a significant presence of deformed granules at 25% moisture (c) (5.8 μm per subdivision in each image).

In the 15-L laboratory conical dryer, the crystals consistently broke into particles of 1–4 μm in diameter when the moisture content of the starting material was greater than 50%. As drying progressed and the moisture content of the wet cake fell below ~40%, the smaller particles granulated to form larger particles. Breaking the initial needle-shaped particles into smaller particles before the onset of granulation was crucial to achieve a consistent granulation. When the starting material had a moisture content less than 50% in the 15-L conical dryer, the needles did not break into more uniformly sized smaller particles and subsequently did not granulate consistently.

Figure 5: Scanning electron microscope images of the dry product from the rotary dryer. In (a) the white bar corresponds to 200 μm and in (b) the white bar corresponds to 20 μm.

A problematic feature of starting the drying process at higher moisture contents in the conical dryer was that a hard layer of product formed against the wall when the wet cake was agitated with a moisture content greater than 50%. The formation of an extremely hard layer at high moisture contents is not uncommon in conical dryers, so the starting moisture content for a pilot-plant trial had to be set below 50% to prevent making a hard coating of product on the wall that could damage the agitator. A drying experiment in a 150-L conical dryer with a starting moisture content of 44% was performed. Unlike the results of the experiments conducted in the smaller scale dryer, the particles broke down in the pilot-plant dryer at a moisture content of 44% and granulation occurred as drying progressed (see Figure 6). The differences in performance upon scale-up is likely attributed to the faster speed of the screw agitator in the pilot-plant dryer (5-ft/s versus 2-ft/s tip speeds).

Figure 6: Progression of material in a pilot-plant conical dryer from (a) starting material at 44% moisture, to (b) completely broken particles after 80 minutes of mixing at 44% moisture, to (c) granule growth at 30% moisture (5.8 μm per subdivision in each image).

The tip speed of an agitator in an agitated filter dryer was facile to hold constant at varying scales. At a tip speed of ~0.4 ft/s, the needle-shaped particles broke at every scale ranging from a 10-cm diameter laboratory dryer to a 2.25-m diameter factory dryer when the moisture content was greater than 45%. Extended agitation at moisture contents greater than 45% did not strongly affect particle size after the material had broken down into particles of 1-4 μm diameters.

Figure 7: Progression of particle shape during drying in agitated filter dryer from (a) 70% moisture to (b) 63% moisture to (c) 30% moisture (5.8 μm per subdivision in each image).

Our experiments showed that once the moisture content fell below 45% during drying, granule growth occurred fastest at higher moisture contents, which is consistent with the generic granulation curve shown in Figure 1. If the drying was suspended but agitation continued at an intermediate moisture content of 30–45% moisture, extremely large granule growth occurred (see Figure 7). SEM images show that the granulated material from an agitated filter dryer (see Figure 8) had smooth surfaces without any indication of the original needle shape, which is in contrast with the material from the rotary dryer (see Figure 5). By maintaining a constant agitator tip speed and agitation time during drying, granulation performance was consistent from the laboratory scale to the pilot-plant and factory scales in the filter dryers used for this API.

Figure 8: Scanning electron microscope images show (a) a particle dried in an agitated filter dryer (white bar = 100 μm) and (b) the relatively smooth surface of particles dried in an agitated filter dryer (white bar = 10 μm).

Conclusion

Consistent performance upon scale-up in a rotary dryer with material that could granulate from the impact of the tumbling action was not achievable. Because the height that material falls in a rotary dryer changes at each scale, the force behind the deformation causing granulation also changes, thereby resulting in material with various particle shapes and physical characteristics at each scale. The images in Figure 5 reveal that the material dried in a pilot-plant scale rotary dryer reached an intermediate state between the original needle-shaped crystals (see Figure 3) and material that had been completely granulated (see Figure 8). Development of a process to maintain such an intermediate state upon scale-up was considered prohibitively difficult. Using intensifier bars in rotary dryers may help normalize granulation performance, but this was not attempted in this work.

For the material used in this study, consistent granulation performance upon scale-up relied on first breaking down the needle-shaped particles into smaller and more uniform shapes. The agitation in conical dryers first broke down the original crystals and then caused the granules to form as drying progressed. The minimum moisture content required for breaking the crystals in conical dryers was not constant upon scale-up from the laboratory to the pilot plant in this study, which may be attributed to the differing screw speeds between the two scales. The pilot-plant scale dryer with the faster screw speed broke down the particles at a minimum moisture content that was lower than that of the laboratory conical dryer. Conical dryers with more closely matching agitation speeds across multiple scales were not available for this work.

For a fixed tip speed in a filter-dryer, the minimum moisture content required to break down the particles was consistent at each scale. As drying with agitation progressed and the material fell below this critical moisture content, granule growth in the filter dryers was consistent in the laboratory, pilot-plant, and factory-scale filter dryers. The observations of the granulation phenomenon for this product during drying are likely applicable to other substances that can granulate in the presence of residual liquid from the final crystallization solvent system.

Joseph Kukura, PhD,* is a research fellow at Merck & Company, PO Box 2000, Rahway NJ 07065, tel. 732.594.1394, joe_kukura@merck.comBill Izzo, PhD, is a senior research fellow at Merck & Company. Charles Orella, PhD, is a senior investigator at Merck & Company.

*To whom all correspondence should be addressed.

Submitted Feb. 15, 2005. Accepted May 18, 2005.

References

1. A. Royce et al., "Process Control and Scale-Up of High-Shear Wet Granulation," Proc. Anal. Technol. 2 (2), 8–16 (2005).

2. R. Kalaiselvan et al., "A Systematic Approach to Scale-Up the High-Shear Granulation Process for Mebendazole Tablets to Achieve Scale-Independent Drug Release," Acta-Pharm-Turc, 46 (1), 25–34 (2004).

3. S.M. Iveson et al., "Nucleation Growth and Breakage Phenomena in Agitated Wet Granulation Processes," Powder Technol. 117 (1), 3–39 (2001).

4. H. Leuenberger, "New Trends in the Production of Pharmaceutical Granules: The Classical Batch Concept and the Problem of Scale-Up," Eur. J. Pharm. BioPharm 52 (4), 279–288 (2001).

5. L.L. Ausburger and M.K. Vuppala, "Theory of Granulation" in Handbook of Pharmaceutical Granulation Technology, D.M. Parikh, Ed. (Marcel Dekker, New York, NY, 1997), pp. 21– 33.

6. S.M. Iveson and J.D. Litster, "Granule Growth Regime Map for Liquid-Bound Granules," AIChE J. 44 (5), 1510–1514 (1998).