By Ole G. Kjaergaard, M.Sc.(Chem.Eng.) Process Design Manager, Food Projects Division, GEA-Niro A/S R1 Aug. 2000
"Prilling" appears to be a somewhat vague but commonly used technical word, at least in the industry of fertiliser production. If we want to use it within the area of particle formation in general and specifically as a term describing a certain type of microencapsulated, particulate material, we obviously need a definition.
The word is, by the way, very hard to find in any dictionary or encyclopaedia, but Hawley's Condensed Chemical Dictionary says that prills are: "small round or acicular (means needle-shaped) aggregates of a material, usually a fertiliser, that are artificially prepared".
We will accept "small round" only, as we associate with droplet formation and want to avoid anything but spheres, - and "small" will have to become something even bigger than "normal powder" particles within the drying business.
A brief overview is found in (1), where the history of spray cooling of Ammonium Nitrate is outlined, starting with the first commercial application described in a German patent in 1918. The first uses were actually not for the fertiliser version but rather product forms useful in formulation of explosives.
For this purpose 2-fluid- and pressure-nozzles were applied as atomisers, taking the Particle Size Distribution (PSD hereafter) to as coarse as "95% between 400 and 2000 microns" equal to an average particle size (d50 hereafter) of 1200 microns or 1.2 mm. The term "prills" really does not originate until 1946 with the first true fertiliser-types, where particles in the range of 1-4 mm are needed for direct dosing onto the fields.
The prilling of the molten nitrate is done with slowly rotating discs or "Buckets" (with holes), - or by means of relatively simple "showerheads" at the top of tall "prilling towers", 30-70 meters high. The cooling, which leads to solidification, is accomplished by a counter-current air stream.
Special metal powders, sodium hydroxide, stearic acid, fats and certain wax powders are also for various applications produced this way. The wax has even been seen to carry a 50/50 payload of proteolytic enzymes for detergent use (2), - and thus we are already at micro-encapsulation by "prilling"! The PSD, however, is as "fine" as only the 200-500 micron (0.2-0.5 mm) range.
As already strongly indicated, prilling is also closely associated with cooling, chilling and congealing. "Big", prilled drops are most conveniently solidified by pure solidification of melts. Drying of big particles is a much slower and much more complicated process, and even small prills would require long processing times and more sophisticated equipment - or would have to be made using agglomeration techniques.
"Congealing" is therefore the normally selected process for the making of prills.
According to Websters Dictionary congealing is "the transition from a soft or fluid state to a rigid or solid as by cooling or freezing".
For fertilisers and in-organic products prilling can be limited to the creation of products consisting of spherical particles with d50 in the 1000 - 4000 micron (1 - 4 mm) range by congealing of coarse droplets from melts only.
For food and pharma products prilling can be confined to "smaller" sizes with d50 in the 200 - 1500 micron (0.2 - 1.5 mm) range, - the sizes up to maybe even 1mm average being obtainable by drying also - at least in combinations with fluid bed drying and special drying aids.
The creation of a spray of droplets is traditionally called "atomisation", but in connection with prilIing, this term is obviously an exaggeration.
Figure 1 lists and compares the normal ways of atomisation from very fast, highly turbulent and disordered conditions to slow, laminar and ordered, - arranged by
The stationary "Shower Heads" are primarily used for the big particle congealing of fertilisers and in-organic materials in troublesome and complicated multiple head configurations in order to achieve sufficient dispersion and contact to the process gas stream.
For the range of sizes interesting for prilling of microencapsulated food ingredients and pharmaceutical products, the best choices are:
In general, the faster types operate in the turbulent range, - have one or up to 40 high capacity openings that are far from filled with liquid. Their droplet size is primarily governed by the high exit speed of the liquid film, which shatters, into a wide droplet distribution.
The slower types operate under laminar conditions, - have hundreds or thousands of low capacity openings that may become filled completely. They generate strings of liquid that break up in a more well controlled way. The droplet size is closely linked to the actual size of the openings and the distributions become close to monosperse.
The best, most practical and complete guide to droplet formation is (3) (in German language).
Figure 1a shows a newly developed and patented atomiser wheel (4) of the medium-slow type with hundreds of big, not filled holes operating in the laminar range and producing the typical narrow droplet size distribution.
This specific surface area varies inversely with (size)1 , and runs in the orders of 1-100 m2/kg. A more realistic and dramatic impression is obtained by calculating the amount of surface coating (also at density 1000 kg/m3) that would be required to make a 10 micron thin protective coating, - expressed as % material to be added. This number ranges from less than 1% for big mm particles and increases to above 10% as we pass 0.3 mm (300 microns)
This specific surface "explosion" is one of the main reasons for selecting 0.2 mm (200 micron) as the minimum range of interest in prilling.
Prilling and other spray processes have the advantage of being able to create ideal spherical particles with narrow size distributions and minimal external specific surface area in one single process step, scaleable to basically any capacity range needed.
The particle surfaces will be smooth, without edges or sharp dents and can be well sealed. All narrow-sized, spherical particle powders with a span less of than 2 and a d50 bigger than 200 microns are often recognised as dustfree and very free-flowing.
As some shrinkage is inevitable during any type of solidification, the spheres are in fact never truly ideal. Even the small changes in volume during congealing (freezing of water being an interesting exception) will result in slightly hollow spheres or spheres with a one-sided indentation due to the fact, that the droplet surface will solidify first. Turbulent, intense mixing conditions will favour hollow spheres, whereas slower and ordered flow patterns will give indentations on the back-side of the droplets, where the material stays fluid.
In the case of solidification by drying, the shrinkage can be very pronounced - and also inflation by vapours trying to escape through a solidified outer film could occur. The typical result is clearly hollow spheres (even ruptured balloons) or particles with a "shower-cap" structure in the case of one-sided drying. The "shower-caps", when viewed as a single layer under the microscope, will roll over, stand on the circular edge and pretend to be perfect spheres ! Such structures could have almost twice the surface area, - which might be interesting in other applications, but would normally be undesired for encapsulation.
These phenomena are the main reasons for preferring solidification by congealing. If drying is involved, the drying rate will normally have to be slowed down and/or the feed formulation must be tailored to secure/minimise shrinkage.
Multiple core matrix systems is the type naturally chosen in connection with prilling, and reference is made to preceding chapters and sections of this publication and to (5), which deals with this subject from a pharmaceutical point of view (and considers 100-500 microns the prilling range with emphasis on drying !).
It is essential, that the matrix pay-load mixture must be a low viscosity fluid (less than 500 mPas, preferably less than 250 mPas) in order not to complicate the droplet formation. The payloads are therefore normally limited to below 50% by volume. The carrier or multi-component carrier system (and any materials used as aids during the process) must be chosen according to a number of other criteria, including approval for food use by relevant authorities.
Figure 3 highlights the phases of this relatively simple process.
The feed-matrix and the mechanical parts of the feed- and prilling-system must be heated a safe 5-30 0 C above the melting point of the matrix. It is also evident, that the congealing should be as spontaneous and complete as at all possible. This could be checked by Differential Scanning Calorimetry, which would also provide true values for specific heats and heats of fusion / congealing for the design calculations. The choice of melting point level is a balance between thermal stability of the payload and processing costs of providing the cooling gas, which typically is simply chilled atmospheric air.
Counter-current or multistage process layouts are obviously optimal.
Even with the previous remarks on minimising the specific surface area in mind, the contact area between the particles and the cooling gas is still very high from a heat exchange point of view. And even at the minimal velocity differences corresponding to the particles free-falling relative to the gas, the heat transfer rate at the surface of the prilled particles is high enough to ensure close to thermal equilibrium between the surrounding gas and at least the surface. Internal heat conduction could result in a difference in bulk temperatures of 5-15 0 C, highest for very big particles and/or poor heat conductors.
Any not yet congealed part of the droplets will stay at the melting point or super-cool. This will cause mechanical stability problems, troublesome later release of heat and further solidification or post-crystallisation, which in turn will lead to lumping.
In the worst case this could result in formation of blocks the size of sacks, containers or silos!
Drying requires as indicated more elaborate feed systems. Drying of big, prilled particles is a slow process and the drying usually even have to be further retarded to allow excessive shrinkage from the outside inward. The same general remarks apply to completion of solidification, stickiness and potential post-processing problems.
As drying also involves mass transfer, and as heats of evaporation are typically a factor 10-20 higher than the typical heats of congealing, the droplets can maintain a "wet bulb temperature" substantially lower than the temperature of the surrounding drying gas, as long as the surface is still sufficiently wet.
Prilling by drying often requires (at least a temporary) anti-sticking surface component of very finely divided "talcum powder" of an acceptable material.
Figure 4 is the overture to the questions and problems concerning selection and design of the best functioning and most economical process equipment for the simpler case of congealing.
Beyond the atomisation device discussed previously, the most important part is a process chamber to confine the droplets and the process gas. This chamber must have sufficient - but minimal - dimensions and still envelope the flight paths of the droplets, - and even the biggest and furthest flying particles of the distribution must not hit the walls before it is safe.
The process chamber dimensions needed for prilling of big particles are big and expensive!
Figure 4 shows the calculated minimum flying time, in seconds, for various droplet sizes from the different atomising devices, all characterised by their typical starting velocities. The flight starts as the droplet of a certain size leave the atomising device at the exit velocity in the given direction. The droplet flies subject to the forces of gravity and air-friction only. Heat transfer is calculated stepwise as function of velocity, size and particle temperature already obtained in the previous step.
The conditions selected are:
Material density: 1000 kg/m3 (as solid)
- feed temp. : 80 oC
- specific heat : 2.1 kJ / kg oC
- latent heat : 105 kJ/kg at 60 oC
The critical time end-point chosen is when the outer 1/3 of the mass has congealed and the particles have been slowed down to their low terminal free falling velocity relative to the cooling air. At this stage contact with the walls can be allowed.
This critical time of flight for the big, prilled particles is almost proportional to size, and depends only slightly on the starting velocity. The time and flight path must, however, be considered for the biggest, significant size of the distribution like the d99-size, which could be 1.5 to 2.2 times the average d50 as already indicated on fig.1.
Figure 5 shows the length of the critical flight path for prilling by congealing, at the same conditions as for fig.4 and up to the point of 1/3 solidification.
For the pressure nozzle (applicable up to approx. 500 micron average size with 1000 micron as significant d99 top size) and for the shower head (up to 4000 micron average size with 6000 micron as significant top size) , the most important dimension is height, which increases with almost (size)2, whereas starting velocity is less important.
For the smallest prills with a d50 of only 250 microns by pressure nozzle, a chamber with a free spray height of 5 meters is needed to accommodate the 550 micron d99 top size (and a not shown diameter of 3.5 m is needed at a 60 o spray angle).
For the coarsest prills with a d50 of 1500 microns produced with a shower head, the top size 2250 micron particles require a 40 meter tall tower ! A diameter of at least 3 meter is needed.
For the two rotary types, the most critical dimension is certainly the diameter.
For the fast rotary at nominal 30 m/s, the diameter develops with droplet size similar to the height for the pressure nozzle. If we take into account that size actually depends strongly on speed for this situation (size being almost inversely proportional to speed), we find that the peripheral speed of the wheel would typically vary from 70 m/s at 250 microns to 15 m/s at 1000 microns.
For a typical 500 micron average product, a chamber diameter of 7.5 meter is needed for the significant 1000 micron d99 size. A height of 6 meters is also required.
For the slow rotary at 3 m/s, size is controlled mainly by the hole-size. Coarse products with 1500 micron average particles can be obtained in a "only" 7.5 meter diameter chamber. The necessary height is however almost 50 meters for the d99 particles of size 2700 microns.
As a general approximation,
-the smaller prills up to say 500 microns average can be produced by pressure nozzle or the "30 m/s" type rotary in chambers going up to 10 meter dimensions, and
-the bigger prills up to 1500 microns require the "3 m/s" type rotary or the showerhead type in chambers going up to 7.5 meter diameter and 45 meter in height.
For drying, the times involved will be longer and the necessary dimensions even bigger.
It is in fact unusual to see size distributions coarser than 300 micron average being spray dried, and that would only be feasible with pressure nozzles atomisation in a tower at least 15 meter high. Big particles achieved by drying are typically agglomerates of smaller primary droplets and anything coarser than 400 micron average is virtually impossible by normal direct spray drying. Subsequent agglomeration, granulation, compaction or extrusion are the only alternatives.
Figure 6 shows a normal spray congealing plant with a "30 m/s" rotary atomiser wheel and designed for 250 to 500 micron average prilled particles. The chamber diameter and height are both 5 to 8 meters. The cooling gas is introduced around the atomiser in a downward swirling motion in co-current with the droplets.
In order to save building height, the chamber often has an almost flat bottom instead of the product collection cone normally seen on dryers (operating at less than maximum 150 micron average for a similar configuration).
The product lands as a thin layer on the bottom, hopefully sufficiently solidified. A rotating broom with cold air blows the particles towards the periphery and the product outlet.
Normally a second, final cooling stage is needed e.g. in the form of a gently vibrated fluid bed.
A small amount of fines that leave with the used process gas is collected separately.
In many cases, the gas is recycled to the gas cooler for
Figure 7 shows a typical layout for large prills with 750 - 1500 micron average with a slow "3 m/s" rotary cup- or Bucket-type wheel with hundreds or thousands of holes.
The capacity of each hole is rather small, in the order of only 0.1 - 3 kg/h, proportional to the peripheral speed and proportional to hole-diameter (d-hole) raised to a power between 2 and 3 as the exit velocity increases with the hole-diameter.
Speed is limited by considerations for the necessary chamber diameter.
The average droplet size depends very little on the speed, but mostly on hole size:
The marginally sufficiently solidified product is received in a fluid bed cooler, which also functions as the cooling gas disperser for the counter-current up-flow layout.
The other possible features are as described in the previous case.
The gas velocities have to be balanced via the selected chamber diameters: the product must fluidise properly in the bed and even the biggest particles must be transported to the product outlet - and not too many of the smallest particles should be carried overhead.
Figure 8 shows a picture of a typical 1 mm average prilled Stearic Acid. The shrinkage and low turbulence has produced small indentations in the spheres only visible on one particle in the lower left part of the sample. The other particles have rolled over and are standing on their indentations!
Figure 9 shows the special design needed to make even small prills by drying, the limitations and difficulties having been discussed previously!
The approx. 250 - 350 micron maximum average size is created with pressure nozzle or a "30 m/s" type rotary multi-holed wheel.
A carefully selected amount of drying gas at moderate temperature can be added on top in co-current to get some drying done, but the particles will arrive in the bottom part still far from dry and most likely in a more than critical state of stickiness.
In order to make it work at all without massive product build-up on the walls, a surplus of a fine dry powdering aid is supplied to the chamber via a pneumatic conveying system.
These powder particles will coat all sticky droplet surfaces. A substantial part of the drying is done in a fluid bed forming the bottom part. Typically even more drying and the final cooling takes place in an external fluid.
As the particles dry, most of the coating powder will be set free again and will recycle via the exhaust air cleaning system, together with the excess and any very fine particles. Only the amount of fine powder aid permanently bound to the final product has to be supplied from the fine powder reservoir.
Figure 10 lastly shows such a product. The layer of very fine sized powdering aid is clearly visible as are the typical, rather large indentations created by the massive shrinkage normally associated with the significant volume changes during drying.
The process described in (6) is of this type, run however as a batch operation with prilling and fluid bed drying run at separate times for ease of operation and control of product uniformity.
Finally, an interesting attempt to make the prilling and drying process work for bigger particle sizes in the range 400 - 1000 microns is described in (7).
Particle Size Distribution.
Average or mean particle size, the 50% volume or weight fractile.
d90, d10, d99
the 90, 10 and 99 % volume or weight fractiles of the PSD
(d90 - d10) / d50
quantity raised to power n