Liquid mixing and blending would seem to be among the most straightforward of pharmaceutical manufacturing unit operations. Mechanical process mixers have been on the market for more than 100 years, and not much separates mixers for drug making from those used in food and chemical industries.
“Sanitary features are the only distinguishing characteristic of pharmaceutical-grade mixers,” says Kevin McNamara of MixMor (Los Angeles). In business for more than half a century, MixMor serves the gamut of process industries but custom-designs most of its pharmaceutical-grade liquid mixer/blenders. Customers pay a premium — between $1,000 and $5,000 — for smooth vessel and impeller designs and all-stainless construction.
Many liquid mixing/blending operations are indeed straightforward — those involving two similar liquids, for example. They get complicated with very dissimilar fluids, such as oil and water, or whenever solids are involved. For very thick slurries, manufacturers begin to worry about uniformity, time-to-blend, and the horsepower of mechanical blending units. And for additives that may crystallize or combine with water, one must be on the lookout for the ultimate disaster of a blend: solidifying to rock-hard consistency, right in that brand-new, $100,000 stainless steel reactor.
In the GEA Diessel Continuous In-Line Blending System, ingredients are received directly from plastic totes provided by suppliers. Courtesy of GEA Diessel.
Like all pharmaceutical manufacturing operations, liquid blending is changing as companies warm to the idea of in-process monitoring and more formal process analytic technology. “Companies are now blending and mixing on the fly, unafraid of multivariant parameters,” says Tim Hoover, business development manager for GEA Liquid Processing (Columbia, Md.). “Instead of blending through a series of large stainless steel batch tanks, processors can take raw materials, blend them continuously, and send the result directly to the filler. We’re moving away from the stainless steel tank and batch mentality.”
The advantage of all this is lower capital costs for “tank farms,” less processing floor space, lower facility costs, and elimination of much of the cleaning and cleaning validation associated with large-scale processing.
Borrowing technology from parent company GEA Diessel, GEA developed continuous blending/mixing equipment that de-aerates liquid components and assures that meters are continuously calibrated to the process stream. GEA is working with instrument vendors to include downstream, on-the-fly monitoring of component concentrations. “It’s a question of marrying instrumentation technology with our mechanical metering systems,” Hoover notes.
Blending for consistency
Bioprocess chromatography is considered a high-risk operation because of variability in columns, chromatography resins and packing, and buffer flow and composition. TechniKrom, Inc. (Evanston, Ill.) has developed modular process skids which, through an innovative mix-and-measure scheme, minimize buffer variability. The idea: eliminating controllable variability (buffer composition) allows greater vigilance, and feedback, over more intractable sources of error.
Premixed bioprocess buffers prepared in large tanks may vary in composition by 3% to 10%. This variation principally results from mixing anomalies, temperature and concentration gradients within the tank, difficulty in measuring component feed, feedstock quality variability, mixing inefficiencies, and human error.
Observing that a great deal of variability disappears when mixing in small volumes, TechniKrom developed a buffer dilution/blending system that manages composition variables in a relatively small volume at a time, monitoring and controlling feedstocks at millisecond intervals, only delivering to the process when the blend is perfect. The sequence of mix, measure, control and deliver repeats, continuously and rapidly, until the desired volume is reached, thus nearly eliminating aggregate variability from all sources on the fly, before delivering the buffer to the next process step. TechniKrom claims blend accuracies as low as 0.1%, and has demonstrated its approach at flow rates ranging from 25 ml to 250 liters per minute, regardless of gradient type.
By contrast, conventional buffers are blended by addition of volumes of pre-mixed components. These systems use accurate flowmeter feedback but lack monitoring of the actual blend composition. Since premixed input components vary significantly by composition, the result is variable blends: Lack of quality in, lack of quality out.
“We learned very early to focus on accurate, low variability blending, since in LC applications such as protein separations a very small change in the mobile phase has a big impact on quality,” says John Walker, VP of Engineering. “Only when the controllable variability elements have been driven out of a process can a developer clearly determine the underlying effects of the remaining parameters, and then design the appropriate process controls to guarantee quality by design.”
Since the technology is modular, it may be deployed for almost any process requiring buffer, for example integration into a chromatography skid, as an add-on upgrade, or as a free-standing buffer dilution skid.
TechniKrom has specialized in process-scale liquid chromatography equipment for some time. The company’s offerings include custom-configured, modular, PLC- and HMI-controlled process skids featuring highly accurate delivery of liquid components for ultrafiltration and process/chromatography buffer solutions.
Many factors affect liquid blend uniformity of complex pharmaceutical formulations. Some, like feed composition, are obvious. Addition rate, blending/mixing power, chemical reactivity and kinetics (including competing reactions), crystallization, site of addition (above or below the surface), and vessel geometry all play a role. Moreover, what works at small scale rarely goes as smoothly in manufacturing suites.
More often than not, mixing problems can be resolved by supplying more power to the blending process, that is, by using a higher-horsepower blender. The difference in price between a 100 hp and 300 hp blender is significant, but the economic impact is often positive when time-to-blend is factored in.
“None of this stuff matters in beakers, when you’re mixing flour and water,” says Dr. Wojciech Wyczalkowski, director of technology and engineering at Philadelphia Mixing Solutions (Palmyra, Pa.)
Laser Doppler anemometer testing of a proprietary impeller design. Courtesy of Philadelphia Mixing Solutions.
The most common error in selecting liquid mixing/blending equipment is a lack of understanding of fluid rheology and properties. Customers tend to simplify blending scenarios, for example only describing them in terms of a single parameter, like viscosity or specific gravity. “It’s especially important with viscous blending to understand the complete fluid rheology, or at least the relation between shear rate and viscosity,” Wyczalkowski notes. “Most of the tests we do for customers are to determine fluid rheology. All too often when talking with them they don’t even understand what we’re asking for.”
When problems persist, Philadelphia Mixing will suggest that customers run a blend at the vendor’s site, which has a twofold benefit. “The customer learns about mixing, and we learn about the process,” says Wyczalkowski.
Philadelphia’s lab provides a controlled environment for testing and observing, through fully monitored glass mixing vessels, what is actually occurring during the blend. One of the company’s pet analytical techniques is conductive tomography, which uses process fluid conductivity to construct a three-dimensional picture of the blend in-process. The technique works for any pair of liquids, or any solid-liquid combination, that alters the electrical conductivity of the blend as a function of time.
Bill Scott, president of Scott Turbon Mixer (Adelanto, Calif.) is even more of a power-pusher than Wyczalkowski. Scott differentiates blending from mixing based on raw horsepower: the former is low-energy, the latter high-energy. “There’s a huge difference in effect between one horsepower per 10,000 gallons, which is stirring, to one horsepower per gallon, the kind of power used to mix dough and polymers,” he says. Just where the cutoff exists between blending and mixing is difficult to say. Scott defines blending at about one horsepower per 100 gallons or less — about the energy required to combine two light, dissimilar liquids, or water and an easily miscible powder (such as sugar or salt).
With powders, the minimum objective is preventing the solids from sinking to the bottom of the vessel. How much energy that takes depends on many factors: liquid viscosity and specific gravity, density of the powder, and vessel size, to name a few. Solids that tend to clump, like gums and starches, or oil-water emulsions, may require 20 to 50 hp per 1,000 gallons to disperse and hydrate. Adding a solid to the emulsion takes about 75 hp per 1,000 gallons. Higher viscosity and specific gravity for staring liquids increases power requirements even more. “And with pastes the sky is the limit,” says Scott. “Those could take one horsepower per gallon, which is huge.”
It might sound simple to anyone who uses kitchen-counter blenders, but industrial practitioners will tell you that liquid mixing/blending is not as straightforward as it appears. “Mixing is not what you learned in college,” says Scott. “You can take a class in hydrodynamics, but when it comes to liquid mixing, success is all about experience.”
It’s the solids, stupid
In fact, one could say that the cutting edge of liquid mixing/blending is formulations involving solids. Contract manufacturer Patheon (Mississauga, Ont.) does a “fair amount” of semi-solid manufacturing, which Dr. Anil Kane, associate director for formulation development, defines as any multi-phase blending involving at least one liquid component. Examples include oil/water (for emulsions), solid/liquid dispersions, and the range of formulations that are the basis of gels, creams, and ointments.
Semi-solid mixing is especially challenging with very potent active ingredients. The key is choosing a vehicle in which the active exhibits the proper partition coefficient (assuming at least two liquid components), so the API is solubilized or dispersed uniformly throughout the final product. Patheon often uses combinations of hydrophilic and lipophilic vehicles to achieve the right partition balance — for example, mixtures of glycerol, propylene glycols, lipids, solubilizers and surfactants. “The choice of co-solvent blends is also critical,” Kane notes.
Sometimes the semi-solid blend needs a bit of help. Kane recalls one project where everything went well until the introduction of one vehicle component, which caused the API crystals to grow in size. Since the vehicle in question was essential, Patheon introduced wet milling to reduce crystal particle size. Similar problems may arise when API crystals form hydrate or semi-hydrate crystals.
Sonic mixing/blending, an interesting alternative to mechanical agitation, uses sound waves to create micro-sized vapor pockets in liquids. During this process, known as cavitation, tiny, highly energetic bubbles implode, causing a nearly infinite number of tiny shock waves within the fluid. Instantaneous temperatures reach the 5,000 K range, which can be exploited for chemical catalysis. Thanks to cooling rates approaching 100,000 K per second, sonication may also be used for liquid-liquid and liquid-solid processing.
Advanced Sonic Processing (Oxford, Conn.), develops sonic process equipment — mostly to accelerate chemical reactions, but for mixing and preparing crystals for milling before blending into liquids as well. “In mixing mode, there’s no degradation of ingredients,” assures president David Hunicke. One setup, for example, accelerates enzymatic reactions while sparing delicate proteins. “The trick is to use the precise ultrasonic dose,” he says.
Ultrasonics is very good at micro-mixing dry and wet additives. When clusters are acoustically activated in a liquid, they naturally de-agglomerate to uniformly wet particles, then uniformly disperse throughout the liquid. For liquid-liquid mixing, say for emulsions, the material must first be “macro-mixed” with a mechanical mixer. “Most processors think of ultrasound as a polishing operation,” says Hunicke. “Emulsion tightness is directly proportional to the uniformity of constituents entering the sonic reactor.”
LIQUID BLENDING: TIPS FROM THE PRO
Bill Scott, head of Scott Turbon Mixer, offers manufacturers a unique blend of advice and wisdom:
* Aim for tank diameter-to-height ratio of 1:1 to 1.25:1, especially for emulsions and dispersions. “Tough blends and short tanks don’t mix,” says Scott.
* Select dished (rounded-bottom) tanks for liquid mixing. Flat or sloped bottoms give dead spots.
* Know your materials of construction. Customers tend to overspend in this department, Scott says — for example, specifying 316 or 316L stainless steel. “Customers want it to be able to go to the moon and back, which is crazy,” Scott says. “304 stainless is fine. Biotech sometimes requires 316 or 316L, but that’s the exception.”
* Know the limitations of your blending equipment: a 1-hp mixer can only do one horsepower’s worth of work.
* Throw a lot of horsepower at difficult mixing jobs. The processing time saved, and the lower costs from reduced stopping, testing and re-starting, will more than pay for the extra investment.
* Add dry ingredients quickly, even simultaneously. Processors can cut batch times by 80% or more through rapid solids addition, provided there’s enough energy to do the job. “There’s no reason why the batch should not be ready for filling a short time after the last ingredient is added,” Scott says.
Figure 1 (top right) depicts the conventional flow in a tank with a standard impeller and baffles to promote mixing and blending.
Courtesy of Philadelphia Mixing Solutions.
Figure 2 shows flow in a tank with a standard impeller inside a draft tube.
Courtesy of Philadelphia Mixing Solutions.