Predicting Blending Performance

Many powder processors share the need to blend multi-component mixtures to homogeneity. In the pharmaceutical industry, for example, the consistent dispersion of an active ingredient, often in very small quantities, within a larger mass of excipient is an essential precursor to uniform dosing, whether the delivery vehicle is a tablet or a dry powder inhaler. When blending dry foodstuffs, the inclusion of small quantities of flavor enhancers or color presents a similar challenge.
    On the other hand, in many industries the challenge is to blend large quantities of bulk materials which may or may not have similar properties. Blending metal powders destined for molding, or the amalgamation of the multiple components of cement, exemplify this type of application. In both cases the homogeneity of the resulting blend has a direct impact on product quality and consistency.
    Despite its widespread use, blending is a unit operation that tends to rely heavily on trial and error to achieve success. Understanding of the factors that influence blending kinetics is still in its infancy, and there is little practical advice available on how to characterize individual powders or mixtures so as to predict blending performance. In this article we assess the potential value of dynamic powder characterization in this context. Experimental data suggest that the flow energy measurements produced by dynamic testing directly correlate with how powders perform in the blender, providing easily accessible information to support the optimization of blending processes.

Information Requirements for Optimized Blending
Those designing and operating blending processes need analytical techniques that allow them to:
• predict blending performance from the properties of the constituents
• assess the success of the blending process in the laboratory and also if possible, on-line.
    Unfortunately neither of these things is easy to do. The progression of discrete populations of particles to a homogeneous whole is a complex process impacted by the array of parameters that influence inter-particular movement. There is currently only limited understanding of the powder characteristics that influence how quickly and easily a multi-component mixture will blend to homogeneity and performance cannot yet be dependably predicted from fundamental particle variables such as size and shape. Individual powders can be difficult to characterize reliably, and blends even more so, subject as they are to the added complication of segregation. The difficulties of engineering a robust sampling and measurement system compound the challenge of developing real-time monitoring solutions.
    Current industrial practice for off-line process monitoring is typically based on thief sampling: taking between 10 and 30 samples from different points within the blender. Subsequent laboratory analysis shows how the mixture uniformity varies in space and time, quantifying homogeneity. However, this approach is time consuming and, perhaps more importantly, disturbs the mixture being measured, potentially compromising data integrity.
    The search for reliable on-line measurement has focused attention on NIR and Raman spectroscopy, laser induced fluorescence (LIF), and thermal effusivity. These proven on-line technologies for compositional analysis are quick and reliable, but are normally limited to single point measurement. This makes it crucial to ensure that the single point measured is truly representative of the overall batch. Furthermore validation of these methods usually relies on thief sampling with all its attendant issues. Laboratory-based QC is generally and necessarily retained as a backup when on-line techniques are applied.

The Importance of Powder Flowability
When evaluating new techniques that may be helpful in predicting and monitoring blending performance, it is useful to consider what it is about a powder that encourages or inhibits blending. Such considerations must focus on powder flowability since the ease with which powders blend depends directly on how they flow, both as individual components, and perhaps more importantly relative to one another.
    In order for a powder to flow, or for blending to progress, particles must move relative to one another, a process that is influenced by the strength of inter-particular interactions which in turn are governed by:
• friction
• mechanical interlocking
• liquid bridging
• inter-particle forces of attraction such as electrostatic charge or van der Waals forces
• the effect of gravity

    Considering first the frictional forces, these can inhibit movement between particles and the movement of particles relative to the surfaces of a constraining vessel. Generally speaking, smoother particles, or materials with low surface roughness, will exhibit freer particle movement. In contrast, mechanical interlocking is where particles slot together like pieces of a jigsaw, and is a function of overall particle shape, as oppose to roughness. Interlocking can produce significant resistance to flow but particle orientation is important here as even particles with the potential to interlock my simply glance past each other (see figure 1).
    The presence of liquid in a powder, often, but by no means always, reduces flowability due to the formation of liquid bridges that can cause particles to adhere more strongly to each other. Conversely, some particles exhibit relatively strong inter-particular forces even in the absence of moisture, as they pick-up and retain electrostatic charge, for example, or because the van der Waals forces are large.
    In any system it is the interaction of these mechanisms that gives rise to the specific flowability profile of the powder and these can be used to rationalize the impact of specific changes. So, for example, consider what would happen if the particle size of a specific material were reduced. Assuming all other factors were kept identical reducing particle size would have a negligible effect on the strength of frictional forces or the extent of mechanical interlocking. However, it would increase van der Waals forces, since these rise exponentially with decreasing particle size, and diminish the effect of gravity, which is a function of particle mass. The expectation would therefore be that inter-particular movement would become less easy, decreasing flowability, which is exactly what tends to happen with finer particles.
    Such rationalizations help to construct an explanation of powder behavior that is relevant to blending performance but they also illustrate why inter-particular movement is not easily predicted. There are a vast number of influential parameters and the relative effect of different mechanisms varies uniquely from sample to sample. Measuring flowability is the pragmatic alternative to a theoretical analysis and it is reasonable to suggest from the preceding discussion that the resulting data should provide some insight into how easily two powders might blend. The following case study details work carried out to test this hypothesis.

Dynamic Powder Testing
Dynamic powder testing involves measurement of the rotational and axial forces acting on a blade as it rotates through a powder sample, to generate flow energy values that directly quantify powder flowability. Specific Energy (SE) is measured by rotating the blade up through the sample, imposing a gentle lifting action, while Basic Flowability Energy (BFE) measurement involves a downward traverse of the blade, and the application of a compacting flow pattern which forces the powder against the confining base of the vessel.
    Dynamic testing can be applied to consolidated, conditioned, aerated or even fluidized powders and generates highly reproducible data giving the technique the sensitivity required to differentiate even closely similar powders. Furthermore testing strategies can be modified to directly study the potential impact of processing related changes. For example, flow energy can be measured as a function of blade tip speed to see whether a powder flows more or less easily when induced to flow more quickly.

Experiments on Blending Behavior
A series of experiments were carried out to investigate whether dynamic test data could be used to predict the blending performance of microcrystalline cellulose (MCC) and sodium benzoate, two materials widely used as pharmaceutical excipients. Blending performance was assessed using the technique of positron emission tomography (PET; see box) which, although time-consuming and ill-suited to industrial application, can be used in the laboratory to gather much more precise and detailed information about blending behavior than is possible with thief sampling.
    A rotary blender was used to mix 0.1 l (2%) of radioactive MCC with 5 l of non-radioactive MCC. Tests were carried out at both 10 and 15 rpm with the homogeneity of the mixture assessed at intervals using PET. Figure 2 shows that at higher rotational speeds, a homogeneous blend was produced in less time with fewer revolutions of the blender.
    This is perhaps intuitive but is in sharp contrast to the results obtained in a duplicate experiment with sodium benzoate. Here, the data indicate that lower rotational speeds reduce the number of revolutions required, suggesting that in this case, slower blending is more efficient. (Figure 3).
    To see whether this observed behavior correlated with powder flow properties, both MCC and sodium benzoate were characterized. Figure 4 shows how flow energy for MCC and sodium benzoate changes as a function of the speed of rotation of the instrument blade.
    In general, powders flow more easily at higher flow rates. Therefore, in dynamic testing, flow energy tends to reduce with increasing blade tip speed, as it does here with MCC. For sodium benzoate, however, flow energy increases with increasing blade tip speed, indicating greater resistance to flow as flow rate increases. No other powder property measurement – angle of repose, shear cell data, flow through an orifice, tapped density methods, or particle size reveals this difference.
    The results of the dynamic testing reflect the trends seen in the blending experiments, and can be rationalized with reference to particle shape. MCC has spherical particles that, at higher blade speeds, tend to become more aerated and flow more easily. Sodium benzoate particles, on the other hand, are platelet shaped, and at higher flow rates simply lock together more enthusiastically, creating an increased resistance to movement which is not conducive to efficient blending.

Investigating Multi-Component Blends
To further assess the correlation between flow energy and blending performance the experiments were extended to a ternary system. 2% of radioactive MCC was loaded into the blender on top of a 2-l layer of MCC, which in turn had been layered onto 3 l of lactose. Figure 5 shows flow energy data for the lactose, MCC, and the binary mixture, alongside PET data recorded during blending.
       The traced MCC blends more easily into bulk MCC than it does into an identical volume of lactose, while the MCC/lactose mixture shows the worst blending performance of the three. This unexpected ranking could be attributed to preferential partitioning of the MCC into a like material, but the flow energy data suggests that the poor dispersion is instead directly linked to increased resistance to flow of the bulk MCC/lactose mixture.

Looking Ahead
The widespread use of blending and the difficulties inherent in current monitoring practice make it valuable to identify analytical techniques that support the prediction and optimization of blending performance.
       These experiments suggest that dynamic powder testing produces data that directly correlate with critical aspects of blender performance. By quantifying whether a powder flows more or less easily when induced to flow at higher rates, dynamic data support the optimization of blending strategies and provides valuable insight into how easily one component will blend with another. This application of dynamic testing is very much in its infancy but research is ongoing, and in the future it’s possible that such analysis can play an important role in reducing current reliance on expensive large-scale blending trials and a trial and error approach to optimization.

Introducing PET
Positron emission tomography (PET) is a nuclear imaging technique with its roots in medical scanning. It creates images by detecting gamma rays released by a positron-emitting tracer substance.
       Because it involves radioactivity and requires the process vessel to be surrounded by a special camera (Figure 6), PET is not well suited to use in production. It is typically used to track just one material at a time, and is generally time-consuming.
       In the laboratory, however, PET provides an unparalleled insight into the mixing process. The technique is non-invasive and can characterize an entire blended mass through a stainless steel vessel wall, in three dimensions and without disturbing the powder. Both resolution and sensitivity are high.

Jamie Clayton is operations manager, and Dr. Brian Armstrong is a powder technologist, Freeman Technology. For more information, call +44 (0) 1684 851551, email [email protected], or visit www.freemantech.co.uk.

Figure 1: Appropriately shaped particles that meet in a certain orientation can lock together like pieces of a jigsaw and then exert significant resistance to flow.

Figure 2: At higher blender speeds, fewer revolutions are required to produce a homogeneous MCC blend. The % RSD parameter is associated with greater homogeneity; below 5% RSD the blend can be considered effectively homogeneous.

Figure 3: With sodium benzoate, higher blender speeds mean more revolutions are required to achieve blend homogeneity

Figure 4: Dynamic test data show that the MCC flows more easily when induced to flow more quickly while the sodium benzoate exhibits the opposite trend

Figure 5: Dynamic flow energy data (a) for a mixture of lactose and MCC suggests that MCC will blend less easily into a lactose/MCC mixture than into either individual component. PET analysis (b) confirms that blending performance mimics this trend.

Figure 6: A mixing vessel entering the PET camera to enable measurement of blend homogeneity

Figure 7: PET images showing a blend proceeding towards homogeneity