Reliable information about industrial crystallization can be difficult to find and to apply to new situations. The work process outlined here will help engineers approach crystallization challenges
Project and plant engineers frequently face issues with crystallization processes, but often possess little prior experience to address them. In probing for answers, engineers often discover that reliable prediction of equipment capacities from machine geometry and physical properties of liquids and solids is improbable, and that most of the literature describing crystallization and separation equipment is confusing. They are also likely to find that feasibility and scaleup correlations are based primarily on empirical know-how developed by crystallization equipment users and manufacturers. Given that, how should engineers proceed toward obtaining a reliable and economic solution to their current cystallization problem?
Using the work process outlined in this article can help guide engineers in approaching crystallization process problems. This includes developing a strong understanding of the system’s solubility data, conducting laboratory tests to gain hands-on experience with a material and setting up a pilot-scale system.
Investigate system basics
It is important to first understand why crystallization could be an appropriate separation method and which types of crystallization are available. For separation or purification needs, distillation is usually considered first. However, distillation may not be practical because of azeotrope formation, low relative volatilities of the components to be separated or other reasons. Sometimes, materials are unstable at high temperatures, which might prohibit using distillation. Crystallization is a good fit for some of these applications.
Crystallization is used industrially to create a particulate solid phase (product), which is recovered and purified by washing and drying. Sometimes the solid product is a “waste” product, but more often, the product is a valuable commodity. Crystallization falls into two main categories: solution crystallization and crystallization from the melt (also referred to as “fractional crystallization”). Solution crystallization implies that material is crystallized from a mixture containing a solvent. Crystals are formed by either concentrating the desired component by solvent evaporation or cooling or both, and the solid phase is typically formed at a temperature well below its freezing point. In melt crystallization, crystals are generated by cooling the solution, and are recovered at temperatures near the pure component freezing point. The melt crystallization product is recovered as a liquid, and particulate solids processing can largely be avoided. This can be helpful if the product is preferred in liquid form.
What can be learned in the laboratory?
When evaluating the feasibility of crystallization as a separation technique for a new process, small-scale testing can provide valuable insight. Significant development work can be accomplished in the laboratory at this early stage. Using solubility information, a basic process scheme can be proposed and tested. Often, through this work, the type of crystallization technique to be used, as well as operating conditions, are defined. Furthermore, the number of crystallization stages required and an appropriate solid-liquid separation technique can all be determined at laboratory scale. Conducting small-scale crystallization tests also allows a first look at crystal morphology (Figure 1) and particle size distribution, and provides a framework for estimating product purity and yield.
In addition, in existing commercial crystallization processes, sometimes troubleshooting is needed to resolve an issue, or re-design is needed due to changes to the process, such as increasing production capacity or switching to a new feedstock with different types or concentrations of impurities. Typically, for these scenarios, there becomes a need for either increased yield, quality or capacity for the crystallization operation.
For existing processes, it often makes sense to do off-line testing at smaller scale. This enables the investigators to gain a deeper understanding of problems, since it is usually simpler to change conditions as well as deal with potential solids handling issues in small-scale equipment. Moreover, good qualitative observations can be obtained by working with smaller equipment. For example, using glass crystallizers, which are readily available for laboratory-scale work, can provide a real advantage. Also at this scale, the cost of raw materials is much lower. Downtime can be minimized, since, in many cases, the commercial process can continue to operate while improvements are being made independently. Figure 2 shows a laboratory crystallizer unit.
Accurate phase-equilibrium data
Regardless of what may be the specific focus or objective of the development or troubleshooting work involving crystallization, reliable solubility data are essential. Obtaining accurate solubility curves or phase diagrams is a critical first step. Theoretical freezing curves can be generated using the Van’t Hoff equation, if the pure component melting point and heat of crystallization are known [Equation (1)].
Van’t Hoff Equation:
Where x2 is the solubility mass fraction, ∆Hf is the crystal heat of formation, R is the gas constant, T is saturation temperature, and TM is the melting point temperature.
With this approach, freezing temperature versus concentration is plotted to construct a phase diagram. While theoretical solubility curves can be calculated, it is preferred to ascertain the solubility of the system experimentally. At the least, it is recommended that several data points from the calculated curve be verified in the laboratory, since reality often differs from what is calculated theoretically, or even from what is reported in the scientific literature.
A good approximation of solubility can be determined quite easily in the laboratory. It is essential to have a reliable temperature measurement device and an accurate method for quantifying the composition of the solution. First, a slurry of the solids suspended in solution is prepared and held at a stable temperature for a length of time sufficient for it to equilibrate. Then, a sample of the saturated mother liquor is taken while either filtering out solids, or after allowing them to settle. The saturated solution is analyzed to determine its equilibrium composition. A curve can be generated by repeating the procedure over a range of temperatures.
The phase diagram depicted in Figure 3 shows solid-liquid equilibrium data for a binary (two-component) eutectic system. The freeze curve indicates the temperature at which a solid begins to crystallize from a solution with the corresponding composition. It shows the range of conditions at which a pure sample of component A or B will crystallize. At the eutectic point, both components will crystallize together at the given composition, such that separation (or purification) is not possible. This point represents the conditions at which the maximum yield can be achieved.
Consider a scenario where the solution is cooled. As the solution is cooled to 0˚C, solids begin to form, and the solution composition shifts accordingly to become less concentrated in component A. Thus, further cooling is necessary to continue crystallization. As the temperature is lowered, pure crystals of component A will theoretically form until the solution is cooled to its eutectic point, shown at –40˚C in the figure. From this curve, given the starting composition and the final temperature, the theoretical product yield and the remaining solution (known as mother liquor) composition can be calculated.
Using the same example, say the final crystallizer temperature is –35˚C. After solids removal is complete, the remaining mother liquor still contains about 40 wt.% of component A, and 60% component B. The theoretical product yield for this case is found by doing a material balance, which simplifies to the following Equation (2), where x1 is the feed mass fraction of component A, y1 is the feed mass fraction of component B, x2 is the final mass fraction of component A, and y2 is the final mass fraction of component B.
% yield = [(x 1 – (x 2 y 1)/y 2)]/x 1 × 100 (2)
For the given example, the resulting yield of component A is calculated to be 83.3%.
For this type of two-component system, the data are used to compute a chart that combines product yield and total solids in suspension (TSS) as a function of temperature and initial feed concentration. The chart organizes the overall material balance for the system and serves as a basis for preliminary evaluation of product yield as a function of process variables. Development and use of this chart narrows the choices available for zones of probable operation and also provides guidance regarding necessary staging of the process. A typical yield chart is presented in Figure 4.
This has proven to be a valuable tool for quickly estimating the yield that can be obtained with a given feed concentration and operating temperature. The value for TSS is also included for consideration, since there is a practical upper limit that should be observed in order to end up with a slurry stream that is manageable and can be pumped and transferred without plugging. The operating TSS is usually the highest TSS with which the equipment can comfortably operate. This level often ranges from 30 to 40% TSS.
For the example above, wherein the feed contains 80% of the desired product, component A, and 20% of the undesired component B, a maximum of ~87% yield can be obtained. To reach this maximum, cooling to –40˚C is necessary. However, if cooling were completed in only one step, the slurry density would be exceedingly high. At greater than 60% TSS, it is unlikely that this slurry could be processed. Thus, splitting the process into multiple steps is probably required.
To reach the maximum yield, but limit slurry density to less than or equal to 40%, it will be necessary to have three stages of crystallization, as follows:
1. In the first stage, 80% feed is cooled from its initial temperature to roughly –10˚C. Crystals begin to form at 0˚C and continue to build up until they reach 40% TSS as the solution is cooled to –10˚C. Nearly 50% yield is achieved in the first stage. The crystals are removed via an appropriate solid-liquid separation step and the mother liquor continues to the next stage.
2. In stage two, the mother liquor from the previous stage contains roughly 65% solute (desired component A). This becomes the feed for stage two. Since its saturation point is –10˚C, cooling below this point will trigger the formation of solids. The final temperature will be –30˚C, which corresponds to the maximum 40% TSS. Approximately 60% of the product contained in the stage-two feed is recovered in this step (that is, 60% of the remaining 50% from step one is recovered), bringing the overall yield to about 80%.
3. The third stage is necessary for obtaining the maximum yield, which will be reached at just above the eutectic point (–40˚C). Often, it is not worth approaching conditions too near the eutectic, where impurities are more likely to solidify.
It is important to note that, often, systems do not behave according to theory. For instance, the solid-liquid equilibrium state may not be easily achieved by simply cooling to equilibrium temperature. The region of conditions where a solution is stable below its equilibrium line is referred to as the metastable zone. When operating in the metastable region, the solution is supersaturated, but crystals do not form. For a system with a large metastable zone, considerable sub-cooling or seeding (the addition of fine crystals as nucleation points) will likely be necessary to trigger nucleation of the dissolved solute.
Crystallization scheme selection
Selecting the operating mode for crystallization is dictated by laboratory evaluation and desired product form. However, determining whether to use evaporation, cooling, or some combination of both can often be accomplished even before testing begins. The phase diagram and yield charts can prove quite useful in selecting a preliminary crystallization scheme. Studying the phase diagram is a good place to start. When solubility is not significantly affected by changes in temperature, evaporative crystallization is a likely candidate. Most commonly, commercial crystallization processes incorporate evaporative crystallization, and often, multi-effect evaporation or vapor recompression is utilized to reduce energy demand.
Conversely, cooling crystallization can be effective when the solid-liquid equilibrium is strongly dependent on temperature. Although cooling crystallization is simpler, it is associated with more fouling problems due to formation of encrustations on cold surfaces and is often avoided. Yet, sometimes cooling crystallization is necessary, such as for processing materials with very high boiling points. In other cases, cooling crystallization is used when a very low operating temperature is required in order to achieve good product yields or due to reactivity or degradation at higher temperatures.
In addition, evaporative cooling (also known as vacuum cooling or vacuum crystallization), a hybrid process that achieves cooling by evaporation at sub-atmospheric pressure, can be considered when solubility has a moderate dependence on temperature. It offers the advantage of low operating temperatures relative to evaporative crystallization, while avoiding fouling on cold surfaces.
Melt crystallization is a less common and more specialized technique that is capable of producing extremely high product purity, and it can be useful when a liquid product is acceptable. When melt crystallization is possible, it offers the benefit that solids handling operations can be avoided, and the associated separation and drying equipment is not necessary. One limitation is that the product must be able to be handled at its melting point. If a material is either unstable at its melting point or its melting point temperature is very high or very low, such that processing at that temperature is impractical, melt crystallization is not a possible solution. An example of a laboratory melt crystallizer is presented in Figure 5.
Laboratory crystallization tests
Once a recovery chart has been assessed and a preliminary crystallization scheme has been selected, actual crystallization tests can be carried out in the laboratory using authentic or synthetic process solution. These tests are used to verify that the scheme and conditions that were selected are appropriate and will provide valuable practical experience regarding processing of the given feed stream.
During crystallization tests, both ease of nucleation and crystal growth are observed, which helps to determine a practical cooling or evaporation rate. Cooling or evaporation must be approached so that solidification occurs slowly enough to grow crystals of adequate size and purity. Either cooling or evaporating too aggressively can lead to fine crystals crashing out of solution or encrustation on cold or hot surfaces due to localized high supersaturation. Suitable crystallizer residence times can be estimated based on the time needed for sufficient crystal growth.
Solid product is recovered by some means of solid-liquid separation. Laboratory-scale pressure filters and basket centrifuges are available, and are convenient for these early studies. Product purity is measured by an appropriate analytical method, such as gas chromatography (GC) or high-pressure liquid chromatography (HPLC), and yield can be calculated. Particle shape (habit) can be determined using microscopy, and a number of analytical methods are available for performing particle size analysis. One simple method for measuring particle size distribution uses a sonic sifter to separate product crystals based on their ability to pass through a stack of sieves with gradually decreasing screen size (known as sieve analysis). The average size and mass for each group of particles is determined, and the data are plotted for a visual representation of crystal size.
Typically, it is premature to do product-purity studies at this stage, but conducting product recovery and washing tests are useful for understanding how easily desired yield and purity can be obtained.
In practice, crystals that are recovered from the mother liquor are rarely 100% pure. Impurities are expected to be present due to occlusions (when impurities are physically trapped inside the crystal lattice), inclusions (when impurities are incorporated into the crystal structure), and due to imperfect solid-liquid separation. Any solid-liquid separation technique will leave behind some amount of impurity-containing mother liquor. In some cases, adjusting the solid-liquid separation technique can result in better removal of impurities, giving a better outcome. Another common option is to perform washing of the crystals. Washing is a standard method for removing residual mother liquor from a solid crystalline product. By incorporating a few sequential wash steps, the bulk of the mother liquor can be effectively removed. Nonetheless, washing must be optimized to improve purity with a minimal loss of yield, since some dissolution of crystals will certainly occur.
During this step in testing, recommendations can also be made regarding seeding. Sometimes nucleation for solid formation can be achieved solely by controlling temperature. In other situations, seeding may be required, and even agitation can play a role. Seeding is often necessary when the metastable zone is large, such that the solution endures significant supersaturation before nucleation. An appropriate seeding technique can prevent large amounts of fine crystals from crashing out of solution, versus growing fewer larger crystals that are easier to filter and transport. Often this must be done carefully so that crystals begin to form and grow in such a way as to produce the desired crystal size.
Crystal morphology is an aspect of crystallization that can have a major impact on processing. Moreover, it is not a variable that is easily manipulated. Usually for industrial processes employing crystallization, there is little concern with specific morphology as long as the particles filter well (without excessive plugging) and can be transferred easily. However, when certain crystal shapes, such as needles, are present, serious handling problems can occur. In these instances, filters can blind off quickly, and settling of particles may be slow. Needle-shaped crystals may also be more prone to breakage. When crystal morphology hinders operation, production capacity may be significantly affected, and downstream processing often must be modified to compensate for slower and less efficient solid-liquid separation.
Crystal size is also a concern, but offers more possibilities for control. With an appropriate seeding technique, a reasonable cooling rate, and adequate residence time, a manageable crystal-size distribution can often be attained.
Attrition of crystals is known to occur, and can be a particular challenge for some fragile solid materials. When crystal breakage is severe, particle size can decrease to such an extent that solid-liquid separation is impacted. In this case, measures to improve handling procedures are necessary. Attrition can be mitigated by using a moderate flowrate and a pump design that is suited for gently handling slurries. When these controls are unable to produce the desired size, specialized crystallizer designs intended to grow large crystals can be tested by collaborating with crystallizer equipment vendors.
A simple light microscope is useful for studying crystal shape and size. An example of a photomicrograph of a crystal product sample taken with a light microscope is shown in Figure 6. With this tool, engineers can track the appearance of crystals with response to changing process variables. Observation with a microscope can reveal changes in habit that result from the presence of various impurities. It allows comparison of crystals before and after washing to see any notable improvements. In addition, it provides a look at crystal breakage (attrition) that could result from processing.
Pilot plant testing
Pilot-scale operation also serves an important purpose in the development of a new process, and is a good tool for troubleshooting. A crystallization pilot plant minimizes risk of scaleup by using equipment that is geometrically similar to commercial equipment, and is used to obtain reliable engineering design data. Process recycles are often incorporated at pilot scale, and this can be extremely important in refining the material balance and verifying the crystalline product purity that can be achieved. Furthermore, pilot-scale crystallizers are able to accommodate production of larger quantities of product material to use for market testing and development. Testing of scalable solid-liquid separation and drying equipment is often conveniently conducted in the pilot unit.
There are three primary objectives in commercial crystallization processes: achieving product purity, yield and capacity. While it is desirable to maximize each of these variables, in reality, when one of these is pushed too far, the other variables will be negatively impacted. Thus, prioritization is necessary to develop a crystallization process with an optimum balance among these elements.
In cases where higher purity (better quality) is desired, multiple methods can be investigated as potential routes for improvement. Sometimes, (as mentioned above), a better solid-liquid separation is needed, to separate the impurity-containing mother liquor away from a solid product. Other times, this can be accomplished by changing the solid-liquid separation technique. A new type of centrifuge or filter may be beneficial. Adjusting the washing procedure might give rise to incremental improvements as well. In still other cases, the crystals are small, and growing larger crystals can help improve solid-liquid separation. Yet another factor influencing purity is the rate of crystallization. In general, slower crystallization yields a better quality product. When all else fails to produce the target purity, recrystallization of a first stage product could be a solution.
When a higher yield is needed, the approach might also include growing larger crystals or selecting another solid-liquid separation device to decrease the loss of solids. Incorporating recycle streams might also be a way to increase overall yields, since the product contained in the saturated mother liquor has another opportunity to undergo crystallization. For certain systems, concentrating the crystallizer feed or cooling to a lower temperature might be an option to improve yields, but when slurry density (solids in suspension) becomes too high, solids handling issues can be a challenge. It may be necessary to add an additional crystallizer stage to manage the higher mass of solids.
For boosting production capacity, more effective heat transfer is usually the answer. This allows the crystallizer to accommodate a higher feedrate, which in turn, decreases the residence time. As a result, particle size may decrease, and purity follows suit. Thus, increasing capacity must be approached with regard to the proper balance of capacity versus yield and purity. However, engineers can attempt to compensate by improving washing or solid-liquid separation, or both.
It is more efficient to make the effort to understand an issue and solve problems early in process development than after a poor design is in place. Nonetheless, in many situations, issues must be solved at a later stage. In either case, following the work process outlined in this article is an effective way to approach the matter.
When presented with any crystallization problem, it is important to begin with a strong understanding of the system’s solubility data. Careful assessment of solubility curves and yield charts, as covered in this article, can provide valuable insight into the root cause of an issue as well as potential solutions before proceeding to experimental testing. Laboratory tests provide the basis to work out the fundamental design and operating conditions of the system. There is no substitute for gaining hands-on experience in working with a specific material and directly observing the crystallization process. This often reveals characteristics that can impact processing, but are subtle and might not be appreciated otherwise.
Once the basic process design has been established and operation is reasonably well-understood, a pilot plant is recommended to define recycle schemes and to gather scaleup data before finally implementing at commercial scale.
Edited by Scott Jenkins
Brooke Albin is a project manager at MATRIC (Mid-Atlantic Technology, Research and Innovative Center; 1740 Union Carbide Drive, South Charleston, WV 25303; Email: firstname.lastname@example.org; Website: www.matricinnovates.com). In this role, she is responsible for the coordination and execution of research and development (R&D) activities for numerous clients. She has 10 years of experience in development of new processes to produce chemicals from biofeedstocks, and has specialized skills in the area of solids handling and crystallization. She received her B.A. in chemistry and B.S. in biology from Alderson-Broaddus College, and earned her M.S. in chemical engineering at West Virginia University.
Charles Moyers is a senior engineering scientist at MATRIC (Mid-Atlantic Technology, Research and Innovative Center; same addresss as above; Email: email@example.com; Website: www.matricinnovates.com). He is a fellow of AIChE and has co-taught several AIChE education courses, including Industrial Crystallization, Liquid-Solid Separation, and Drying. He holds a B.S. from Virginia Tech, an M.S. from West Virginia University and a Ph.D. from the University of Delaware, all in chemical engineering.
This overview presents the traditional and emerging types of continuous crystallizers Crystallization is an important technique for the purification of…
The following are recently published titles relevant to the chemical process industries (CPI) Sustainable Water Management and Technologies, Vol. 2:…
This company now offers a wider range of process air stages by introducing the large blower series known as Alpha…
Chemists and engineers from the Performance Materials Segment of Evonik Industries AG (Essen, Germany; www.evonik.com) have developed a new process…
For petroleum refineries, the cost of converting hydrofluoric-acid-based alkylation units to the relatively safer sulfuric-acid-based alkylation process (diagram) has been…