The Effect of Shear Mixing on the Blending of Cohesive Lubricants and Drugs

Almost all pharmaceutical manufacturing processes require handling and processing cohesive powders. The application of sufficient shear (i.e., the total deformation that the bulk of granular material undergoes under applied shear stress) is an essential factor in such processes. Sufficient shear is required to mill and de-lump materials, achieve sufficient flow, and homogenize cohesive ingredients. Shear mixing plays a critical role in the blending of dry powders, particularly for those that contain a minor cohesive component such as a solid lubricant or a drug. This mechanism is necessary to achieve a satisfactory homogeneity and disintegrate possible agglomerates. Excessive shear can be disadvantageous, however, and can lead to electrostatic buildup, attrition, and overlubrication.

Shear mixing is caused by a velocity gradient within the material. Because powder flows are poorly understood at the present time, the effect of variables such as blender size, fill level, or speed of rotation of the blender have on shear rate can be established only qualitatively at best. Present technology does not allow the measuring of shear rates in situ for granular processes, and the assessment of shear conditions is done through blend homogeneity. Nonetheless, the evolution of homogeneity is intimately associated with the constitutive characteristics (cohesion and density) of the powders being blended. Moreover, homogeneity is not the only important parameter (otherwise the results obtained in a screen mill would be the epitome of mixing performance). The exposure of a lubricant to high intensity of shear entails the risk of overlubrication, with resulting degradation of tablet hardness and dissolution. Processes for which shear is a critical variable present additional challenges during scale-up.

This article reviews the lubrication and de-agglomeration phenomena in which shear plays a critical role. The last section of this article briefly addresses scale and presents conclusions and recommendations for future work.

Effects of shear mixing in the lubrication of a blend

Magnesium stearate is one of the most widely used lubricants in the pharmaceutical industry, and its functionality is based on the formation of a film of lubricant on the carrier particles. Magnesium stearate exhibits various morphologies with distinctive shear strengths and abilities to form a film that reduces the friction between particles and the tablet press die. The excessive coating of carrier particles by hydrophobic magnesium stearate, however, reduces tablet solubility and prevents direct interaction among particles of other ingredients. As a result, the strength of the entire tablet is determined by the low shear strength of magnesium stearate, and the hardness of the tablets is proportionally reduced.

For many systems, the ideal lubrication operation provides the mildest mixing conditions that guarantee sufficient homogeneity of the magnesium stearate. The extent of shear applied during the mixing process is a critical variable. If the mixing process is too short, magnesium stearate might be inhomogeneously distributed; thus, some portions of the blend will contain excessive amounts of magnesium stearate, show an increased tendency to overlubrication, exhibit a decreased tablet mechanical strength, and have decreased dissolution. If excessive shear is applied, however, magnesium stearate particles become finely divided, flowability can be adversely affected, excessive coating of active pharmaceutical ingredients (APIs) and excipients by magnesium stearate might occur, and the entire blend might be overlubricated.

At the present time, shear rate cannot be assessed easily inside a blender, and therefore blend and tablet properties must be correlated with indirect variables such as mixing time (1), fill level, and the blender's rotational speed and scale (2). This article examines a case study, focusing on the blend's lubrication in a 30-L blender (L.B. Bohle, Warminster, PA). The flowability of granulated materials, which is a function of particle properties, is what mainly determines the formation of a film of magnesium stearate (3). Poor flow properties usually retard the formation of a lubricant film. This fact, expressed in terms of mixing mechanisms, indicates that shear rates in the bulk of the material determine the formation of the film. This article analyzes the effect of fill level (40, 60, and 85), blender rotation speed (6, 14, and 16 rpm), presence of internal baffles, and mixing time on the homogeneity of magnesium stearate for specific granulated materials. The blender was sampled using a groove sampler, and the content of magnesium stearate in samples was analyzed using near-infrared spectroscopy. The various shear conditions and mixing efficiencies for each speed or fill level result in different cohesive-powder homogeneities. The experimental results indicate how several variables affect the shear conditions.

Fill level. Fill level can lead to the most striking differences in mixing performance (see Figure 1). The mixing process is the slowest, and the homogeneity of magnesium stearate is the poorest, at the highest fill level (85%). Hindered mixing, often observed at a high fill level, produces a slow homogenization of magnesium stearate. Even though long mixing times can compensate for the slow mixing conditions of a high fill level, this step is often a poor solution: in a scaled-up operation, long mixing times will almost certainly result in overlubrication.

Figure 1: The effect of fill level on the outcome of the lubrication process is evidenced by the differences in the evolution for magnesium stearate homogeneity.

Rotation speed. Although vessel rotation speed has little effect on the mixing process (when measured as a function of vessel revolutions) of free-flowing granulated materials, the speed of the vessel is critical for cohesive materials such as magnesium stearate. The vessel rotation speed determines the shear rate, and thus has a direct effect on the outcome of the process (see Figure 2). The homogenization process is slower at 6 rpm than at 26 rpm. For long mixing times (320 revolutions), however, the same overall level of homogeneity is attained for both speeds. This effect suggests that the de-aggregation of the lubricant is a function both of shear rate and total shear.

Figure 2: The effect of rotation speed on the shear rates of the process.

Internal baffles. The presence of properly designed baffles can increase the axial mixing rate. In fact, a blender operating at 60% fill level achieves homogeneity slightly faster with the aid of baffles. Baffles do not affect shear rate substantially and, although they do not increase the risk of overlubrication, they do not improve the homogeneity of a cohesive-powder blend or promote the disintegration of agglomerates substantially.

Effects of shear mixing in the blending of drugs: avoiding agglomerates

In general, the blending of a cohesive API does not have problems associated with exposure to high shear rates or total shear such as those encountered in the blending of lubricants. To the contrary, most problems in homogenizing cohesive drugs are the consequence of low shear rates in blenders. Agglomerates containing a high proportion of API can form within a blender producing blends characterized by fine API particle size, hygroscopic material, or, in some instances, when the API tends to acquire an electrostatic charge. Such agglomerates can result in a small subpopulation of superpotent tablets that are observed only occasionally, but that can throw a manufacturing operation into disarray.

The case study presented here focuses on the blending of a cohesive drug using rotating bins of different sizes, followed by the passage of the blend through a high-shear device such as a conical mill at the discharge of the blender. The conical mill provides high shear rates and guarantees that the blend will be entirely and uniformly exposed to shear. Conical mills improve the distribution of the API and minimize drug agglomerates (4).

Shear rates increase as the scale of the blender increases. This effect is supported by the experimental results obtained for the blending of a cohesive drug and free-flowing excipients in a 56-L (relative standard deviation 57%) versus a 300-L bin blender (relative standard deviation 8.5%). The large-scale bin provides higher shear rates and renders more homogeneous mixtures.

All the material in a blender will not always be exposed to uniform shear, however. Agglomerates are found in a 300-L vessel after mixing for 256 revolutions at 9 rpm (see Figure 3). Though the blending process achieves a nearly satisfactory level of gross homogeneity, some samples show high active concentrations caused by the presence of drug agglomerates. These agglomerates are readily destroyed by the mill. Samples obtained at the discharge of the mill show a general improvement in the quality of the blend and the disappearance of drug agglomerates (see Figure 3).

Figure 3: Concentration profiles before and after milling.

The use of mills at the discharge should not be regarded as a universal solution, however. If the blend is highly inhomogeneous, the stream traversing the mill will exhibit differences in composition as a function of time. Lacking back-mixing capacity, a mill cannot eliminate such insufficiencies in homogenization. Typically, an additional mixing stage is required after milling to ensure sufficient homogeneity, although, in such cases, care should be taken to ensure that agglomerates do not re-form.

Shear rate and scale-up of the process

One of the most complex problems in powder blending is the scale-up of an operation that involves cohesive powders. The customary theoretical approach to this problem is to develop a dimensionless version of the equations that govern the flow in a specific geometry (5). As a result, one obtains several dimensionless groups, and experiments can help determine which of those groups are relevant for the specific conditions of operation. The behavior of free-flowing particles is determined mainly by gravitational and centrifugal forces, and the main criterion to scaling up such a process is the Froude number (6). Constitutive interparticle forces are involved in the blending process of cohesive powders, however. The scale-up for the latter process is much more complex, and a rigorous method for such a process has not been established.

Mixing in a tumbling vessel takes place in the flowing layer, and the rest of the material simply follows the motion of the container. Not surprisingly, a body of research is focused on understanding the behavior of this layer (7, 8). The study of the dynamics of avalanches has led some researchers to propose universal granular-flow properties applicable not only to blenders, but to numerous other systems (9). Correlations between the systems' subunits can be described either by an exponential correlation or by a power-law correlation. The exponential law implies the use of a characteristic length scale, whereas power-law systems are scale-free. Data on granular avalanche behavior of granular processes can be applied to fit both of these categories. For free-flowing materials, a theory successfully uses half of the length of the flowing layer as a scaling parameter (10).

These theories can be difficult to apply to cohesive systems under practical conditions. A substantial problem is how to account for the effect of cohesion on powder flow. The problem is extensive, and only a brief discussion is provided in this article. In simple terms, a cohesive powder can be defined as a material in which the adhesive forces between particles exceed the particle weight by at least an order of magnitude. In such systems, particles no longer flow independently; rather, they move in ''chunks'' for which characteristic size depends on the intensity of the cohesive stresses. The size of the chunks introduces an internal, material-dependent length scale that plays a role in flow scale-up (e.g., in determining whether a blend will flow easily into a tablet press die).

The effective magnitude of cohesive effects depends primarily on two factors: the intensity and nature of the cohesive forces and the packing density of the material (which determines the number of interparticle contacts per unit area). This dependence on density is the source of great complexity. Cohesive materials often display highly variable densities that depend greatly on the immediate processing history of the material. In spite of this complexity, a few guidelines can be offered within a fixed operational scale:

• Slightly cohesive powders mix faster than free-flowing materials.

• Strongly cohesive powders mix much more slowly than free-flowing materials.

• Strongly cohesive powders often require externally applied shear (e.g., in the form of an impeller, an intensifier bar, or a chopper).

• Baffles attached to vessels do not increase shear substantially.

Because there is no systematic means to measure cohesive forces under practical conditions, the effects of cohesion on scale-up have been studied rarely. The most important observation is that cohesive effects are much stronger in smaller vessels, and their effect tends to disappear in larger vessels. The reason is simple: although cohesive forces are surface effects, the gravitational forces that drive flow in tumbling blenders are volume effects. Thus, as the scale of the blender increases, gravitational forces grow faster, thus overwhelming the cohesive forces. This effect also can be explained by remarking that the characteristic chunk size of a cohesive powder flow is a property of the material, and thus, to a first approximation, it is independent of the blender size. As the blender grows larger, the chunk size–blender size ratio decreases. Both arguments can be expressed mathematically in terms of a dimensionless cohesion number, Π_c:

in which σ is the effective (surface averaged) cohesive stress (under actual flow conditions), ρ is the powder density under flow conditions, g is the acceleration of gravity, and R is the vessel size. The group σ = (σ ÷ ρg) is the aforementioned chunk size, which can be defined more rigorously as the internal length scale of the flow. Thus, as R increases, Π_c decreases, explaining why scale-up of powder-blending operations often succeeds in spite of limited understanding of cohesive powder-flow physics.

For processes in which shear entails effects as difficult to predict as overlubrication, overconfidence can lead to devastating process failures. For blenders with intensifier bars and other moving internal parts, the long-established practice has been to match the linear (tip) speed of the moving element, and this approach has been used successfully in numerous practical situations. For blenders without internal moving parts, experience clearly indicates that both shear rates and total shear per revolution increase as the blenders become larger. This effect is supported by ample data showing more frequent overlubrication and less frequent agglomeration in larger tumblers (2, 11). Unfortunately, the method outlined above has so far failed to produce reliable criteria for scaling of shear rates. Though a scaling approach leading to confident design would be greatly desirable, the safest approach at the moment is to seek an empirical correlation between tablet properties and operational variables.

Conclusion

Even though the effects of shear in blending processes are not understood completely, several reproducible observations provide guidance for process design and scale-up.

Tumbling blenders provide low shear intensity, especially in small-scale devices. Baffles do not substantially increase shear.

When blending cohesive APIs, the main process risk is the survival of particle agglomerates that can cause superpotent tablets to occur. Using de-lumping steps can greatly help alleviate this problem, provided that the agglomerates do not re-form during subsequent blending and lubrication.

During lubrication, process risk can be caused either by very short process times, which can lead to the inhomogeneous distribution of the lubricant, or the application of excessive shear, which might reduce lubricity, adversely affect powder flowability, and decrease tablet hardness and solubility. Practitioners must be careful to find an optimum process time. Maintaining such an optimum setting through process scale-up is not trivial because shear rates for tumbling blenders increase for larger systems.

It should be emphasized that a better understanding of the role of shear in process design and scale-up is long overdue. A fundamental limitation is the lack of well-defined experimental systems for which samples are sheared uniformly at a controlled rate to allow systematic quantification of the effects of shear on blend and tablet properties. The development of such a system is actively under way at various institutions, including Rutgers University. Promising results have been obtained and are currently being checked for reproducibility.

Marcos Llusá is a doctoral student in the department of chemical and biochemical engineering at Rutgers University, tel. 732.445.2588, llusa@eden.rutgers.edu and Fernando Muzzio, PhD, is a professor and the director of the NJ Particle Processing Research Center at Rutgers University, tel. 732.445.3357, muzzio@soemail.rutgers.edu Dr. Muzzio also is a member of Pharmaceutical Technology's editorial advisory board.

To whom all correspondence should be addressed.

References

1. G. Ragnarsson, A.W. Hölzer, and J. Sjögren, "The Influence of Mixing Time and Colloidal Silica on the Lubricating Properties of Magnesium Stearate," Int. J. Pharm. 3 (2–3), 127–131 (1979).

2. J.G. Van der Watt and M.M. de Villiers, "The Effect of V-Mixer Scale-Up on the Mixing of Magnesium Stearate with Direct Compression Microcrystalline Cellulose," Eur. J. Pharm. Biopharm. 43 (1), 91–94 (1997).

3. C.E. Bos, H. Vromans, and C.F. Lerk, "Lubricant Sensitivity in Relation to Bulk Density for Granulations Based on Starch or Cellulose," Int. J. Pharm. 67 (1), 39–49 (1991).

4. R.P. Poska, T.R. Hill, and J.W. van Schaik, "The Use of Statistical Indices to Gauge the Mixing Efficiency of a Conical Screening Mill," Pharm. Res. 10 (8), 1248–1251 (1993).

5. Y.L. Ding et al., "Scaling Relationships for Rotating Drums," Chem. Eng. Sci. 56 (12), 3737–3750 (2001).

6. C. Clump, "Mixing of Solids" in Mixing, Theory and Practice, Volume II, V. Uhl and J. Gray, Eds. (Academic Press, San Diego, CA, 1967), p. 284

7. L.P. Kadanoff et al., "Scaling and Universality in Avalanches," Phys. Rev. A: At., Mol., Opt., Phys. 39 (12), 6524–6537 (1989).

8. N. Sepúlveda, G. Krstulovic, and S. Rica, "Scaling Laws in Granular Continuous Avalanches in a Rotating Drum," Physica A 356 (1), 178–183 (2005).

9. H.E. Stanley et al., "Scaling and Universality in Animate and Inanimate Systems," Physica A 231 (1–3), 20–48 (1996).

10. G. Weir, D. Krouse, and P. McGavin, "The Maximum Thickness of Upper-Shear Layers of Granular Materials in Rotating Cylinders," Chem. Eng. Sci. 60 (7), 2027–2035 (2005).

11. A. Alexander et al., "Characterization of the Performance of Bin Blenders," Pharm. Technol. 28 (9), 54–74 (2004).