There is a wide variety of equipment used to carry out the crystallization process, called crystallizers. Such equipment can be classified into four broad types:
Bulk solution crystallizers. Crystals are suspended in solution for a significant time while nucleation and growth occurs.
Precipitation vessels. Feed streams entering the vessel generate high supersaturation levels (by chemical reaction, drowning or salting out), very rapidly forming large numbers of small crystals.
Melt crystallizers forming multiple crystals. The bulk (typically > 90%) of the solution or melt forms crystals either in suspension or on a cooled surface. Impurities remain in the small amount of uncrystallized mother liquor.
Melt crystallizers forming large high-purity single crystals. Crystals form very slowly from high-purity melts, yielding large, pure and defect-free crystals. These are typically used for semiconductor manufacture.
All these type of equipment have aspects in common:
A region where supersaturation is generated to drive the crystallization.
A region where crystals are in contact with supersaturated solution for crystal growth. In some cases, crystals are present throughout the vessel, suspended by some form of agitation; in other cases, the crystals occupy only part of the vessel, typically as a fluidized bed.
The selection of the appropriate crystallizer for a particular task will depend upon the feed material available, the properties of the system and the product requirements of the customer. A typical design sequence involves:
Basic data collection.
Selection of supersaturation generation method.
Choice of batch or continuous operation.
Choice of specific equipment type.
Bench and pilot scale tests.
Full scale design.
A full procedure is too complex for this article; however some key aspects are outlined below. For further details, refer to any of the books given in the reference list.
Five main methods can be used to generate supersaturation:
Cooling, using the vessel walls, internal coils, or by pumping mother liquor through an external heat exchanger. This is used when solubility changes significantly with temperature and when the feed stream is near saturation at a high temperature.
Evaporation, by heating the mother liquor or reducing the pressure to form a boiling zone at the top of the vessel. This can be used for a wide range of systems, although it is more energy-intensive than cooling.
Reaction, where feed streams enter and mix resulting in a chemical reaction generating the product, usually at high levels of supersaturation.
Drowning out, where a miscible solvent is added resulting in a mixture in which the product is less soluble. This has similar characteristics to reaction crystallization.
Salting out, where a salt with a common ion is added to precipitate the product from solution. Again, this has similar characteristics to reaction crystallization.
Other techniques, e.g., pressure change, are occasionally found.
Crystallizers can be designed to operate in either batch or continuous mode (and, rarely, combinations of the two).
Batch crystallization is generally easier to control and is more flexible. It can operate over a wide range of conditions.
Continuous crystallizers produce a consistent product and are generally smaller and more energy efficient than batch equipment for the same production rate. Thus, continuous crystallizers are favored for high-production rate systems. However, they operate over only a narrow range of conditions, so more process knowledge is generally required to make sure they produce the required product specification.
For solution crystallizers, the simplest equipment is an agitated, cooled vessel. Although simple, it is far from optimal in terms of hydrodynamics, with poor crystal suspension. A draft tube and baffles are often added to improve suspension characteristics, and this leads to designs such as the Swenson Draft Tube Baffled (DTB) and Oslo-Krystal crystallizers.
The Swenson DTB has the main recirculation provided by a propeller inside a draft tube, with a settling zone to allow fines to be removed and dissolved, and a product elutriation leg, where large crystals are extracted against an upward flow carrying small crystals back into the vessel. Both evaporative and cooling versions are found.
The Oslo-Krystal unit has a fluidized bed of crystals, suspension and agitation being provided by an external circulation loop of either crystal magma or relatively crystal-free solution.
For further information on industrial solution crystallizers, refer to Myerson (1992), Mullin (1993) or the SPS Crystallization Manual.
Melt crystallizers fall into three general categories. The first has crystals in suspension within the vessel; crystals form on a cooled wall and are removed by a scraper. A melting zone can be added to increase purification of the crystals. The second has crystals forming on the outside of a cooled rotating drum or belt, with a scraper removing crystals directly as product. The third has crystals building up on a cooled wall; when crystallization is complete, residual mother liquor and impurities are drained off and the temperature raised to melt and recover the product. Product purity can be increased by "sweating" impurities out of the solidified crystals before recovery.
For further information on melt crystallizers, refer to Myerson (1992) or the SPS Crystallization Manual.
Precipitation processes are often carried out in agitated vessels of some kind, or by techniques such as impinging jets. Equipment selection is dependent upon system kinetics and product requirements. For further information, consult Söhnel and Garside (1992), Myerson (1992) or the SPS Crystallization Manual.
Scale up is a very complex procedure for crystallizers, and several points should be noted. Growth rates, provided they are measured using liquor containing the correct impurities, are simple to scale-up from bench measurements. Nucleation rates, however, provide a balancing complexity. For continuous crystallizers, secondary nucleation rates dominate and these are a combination of several different mechanisms (e.g., crystal-crystal and crystal-wall collisions); each mechanism will scale differently. For batch crystallizers, primary nucleation rates are dependent upon hydrodynamic and thermal conditions which are easier to determine. However, when coupled with the inaccuracies in nucleation rate measurement, scale-up of nucleation rate is always difficult. If successful full-scale operation is highly dependent upon accurate scale-up, then several intermediate-sized tests may be needed. Alternatively, features such as fines or ultrasonic treatment may be used to provide some additional control over effective nucleation rate on the full scale plant.
Batch crystallizers have the most control options available. Generally, the key parameter is the rate of generation of supersaturation expressed in terms of the cooling, evaporation or addition rate. The discussion below uses cooling rate as an example.
For the production of large crystals, the overall strategy is to form relatively few nuclei and then to grow these to product size under conditions which minimize further nuclei formation. Typically, this means cooling the solution (A) until it enters the metastable zone (B). Seeds can be added (G), or the solution cooled slowly until the labile zone is reached (C) and primary nucleation occurs (D). As the crystals grow (E) and the surface area increases, solute will be depleted increasingly rapidly, and the cooling rate can be increased to maintain the supersaturation level somewhere in the middle of the metastable zone (F).
Continuous crystallizers are more simple since they operate at steady-state. The only adjustable parameters are usually the feed rate (and hence residence time) and the supersaturation generation rate. If fines treatment or product classification is possible, then these are additional controls. As a result, continuous crystallizers tend to be easier to control, but can only be operated in a relative narrow range of conditions. It is also possible for them to operate in an oscillatory mode due to interactions between growth and nucleation rates, but this effect generally only appears in classifying crystallizers.
The size distribution from an ideal mixed-suspension, mixed-product removal (MSMPR) crystallizer, with size-independent growth and no growth dispersion, can be predicted by:
where n0 is the population density of nuclei; L is the crystal size; G, the growth rate; and τ, the residence time. Thus plotting log (population density) against size for an MSMPR should yield a straight-line graph. The mass-mean size from such a distribution is 4Gτ. Note that in the above equation, G is also a function of τ; increasing residence time does not proportionally increase the mean size since the growth rate would decrease.
A separate issue is the start-up of continuous crystallizers. It is generally recommended that at least 10 residence times are required for a continuous system to reach steady-state, though in practice some systems can take significantly longer. Thus, frequent startups and shutdowns of these types of equipment should be avoided.
Operating crystallizers do not always work perfectly. Some potential problems are listed below, but this is not exhaustive.
Crystal mean size is too small or too many fines are present. Nucleation rate is too high.
Crystal mean size is too large due to insufficient nucleation.
Crystal size distribution is too wide.
Variable size distribution from continuous crystallizer. Usually related to the use of fines treatment and/or product classification equipment.
Formation of unwanted agglomerates.
Formation of wrong polymorph or solvate; can also lead to problems with product caking or breaking down during storage.
Encrustation on vessel surfaces due to local high supersaturation levels or low crystal concentrations.
High impurity levels; either due to impurities becoming incorporated within the crystal or poor washing of residual mother liquor.
The cure for any given problem is too system-specific for a brief outline in this article; however, some confirmatory tests and remedies are obvious from the above descriptions.
G Crystal growth rate (m/s)
L Crystal size (m)
nL Population density, size L
τ Residence time (s)
Garside, J. (1992) Precipitation: Basic Principles and Industrial Applications, Butterworth-Heinemann, ISBN 0-7506-1107-3.
Mullin, J. W. (1993) Crystallization 3rd Edition, Butterworth-Heinemann, ISBN 0-7506-1129-4.
Myerson, A. S. (Editor) (1992) Handbook of Industrial Crystallization, Butterworth-Heinemann, ISBN 0-7506-9155-7.
Nývlt, J. (1992) Design of Crystallizers, CRC Press, ISBN 0-8493-5072-7.
Söhnel, O. and Garside, J. (1992) Precipitation: Basic Principles and Industrial Applications, Butterworth-Heinemann, ISBN 0-7506-1107-3.
SPS Crystallization Manual, Separation Processes Service, Harwell Laboratory, Didcot, Oxon, UK.