Heat and Mass Transfer Operations-Crystallization Chapters/C06/E6-34-03-02.pdf · CHEMICAL...

UNESCO - EOLS

S

SAMPLE C

HAPTER

CHEMICAL ENGINEERING AND CHEMICAL PROCESS TECHNOLOGY – Heat and Mass Transfer Operations – Crystallization - J. Ulrich and M.J. Jones

HEAT AND MASS TRANSFER OPERATIONS - CRYSTALLIZATION J. Ulrich and M.J. Jones TVT, Martin-Luther-Univ. Halle-Wittenberg, Germany Keywords: Industrial Crystallization, Solubility, Phase Diagrams, Nucleation, Crystal Growth, Polymorphism, Hydrates and Solvates, Crystal Habit, Crystallization Technology, Melt Crystallization. Contents 1. Introduction 2. Solid-Liquid Equilibria 3. Kinetics 4. Properties of Crystals 5. Crystallization Technology 6. References Glossary Bibliography Biographical Sketches To cite this chapter Summary This chapter presents an overview of industrial crystallization both from a fundamental, scientific perspective and with an eye on the application of crystallization as a separation process in industrial processes. Crystallization as a separation process yields a solid product from a solution or a melt. The process itself, the product characteristics and the temporal behavior are determined by thermodynamics, i.e. solubility of the materials, and crystallization kinetics, respectively. The thermodynamics of solutions and melts are dependent on a range of physical parameters as discussed within. Crystallization kinetics can be divided into two separate processes, nucleation and crystal growth. Both play a significant role in the design of equipment for a given process and are treated in detail below. Additional factors such as the effect of crystallization on product properties as well as the effect of solid-state properties on the crystallization process are discussed in section 4. Section 5 provides details on commonly used crystallizer designs and highlights the distinction between solution and melt crystallization. 1. Introduction Crystallization is a thermal separation, and therefore a purification process that yields a solid product from a melt, from a solution or from a vapour. As for all thermal separations, non-equilibrium conditions are required as a driving force for the process. Here, evaporation of solvent or temperature reduction (cooling) are the most frequent

©Encyclopedia of Life Support Systems (EOLSS)

UNESCO - EOLS

S

SAMPLE C

HAPTER


means employed to establish the required non-equilibrium conditions. Pressure can, in principle, also be used to enforce the non-equilibrium state necessary for a phase change. However, in industrial applications this parameter is most frequently kept constant. As a consequence, processes involving pressure change will not be discussed here. Other technical routes to leading to crystalline products may involve chemical reactions (reactive crystallization), change of dielectric constant/ionic strength of a solution (salting-out) or crystallization induced by a change in solvent composition (drowning-out).Those type of crystallization processes often referred to as precipitation processes will not be covered here [Söhnel/Garside]. The main feature distinguishing crystallization from other thermal separation processes is the fact that it leads to a solid product. This is one of the key reasons why it lags behind separation techniques involving liquid-liquid or liquid-gaseous phase change processes in terms of research effort expended and knowledge available. In the past chemical engineering has tended to avoid the solid state as far as possible due to the problems inherent in solids handling. Encrustation and the concomitant heat transfer problems, seizing of moving parts, generation of hazardous dust and the general difficulty inherent in solid-liquid separation are just some of the issues involved. However, the attitude towards crystallization is changing and the positive aspects of the techniques, low energy consumption and the potential for high-purity products, are more widely appreciated. [Mullin], [Myerson], [Mersmann], [Tavare], [Jones], [Nyvlt/Ulrich], [Hartl], [Jancic/Grootscholten], [Ulrich 2002 (in Kirk-Otthmer)] Crystallization is a highly selective process and operates at lower temperatures when compared to a separation by distillation for the same material. Melt crystallization has the additional advantage of not requiring a solvent, although it is not a method suitable for all materials. Manifold reasons exist for the growing importance of crystallization as an industrial separation process. On the one hand, many products are solids under ambient conditions, in particular in the specialty chemicals sector, in the pharmaceutical industry and in the manufacture of foods. On the other hand, the ability to manipulate, at a minimum, the macroscopic properties of crystals such as shape and size by judicious choice of crystallization conditions makes the technique a very attractive option in those industry sectors already mentioned, where reproducible product quality with well defined properties (flow properties, color, dissolution characteristics, polymorphic form etc.) are of utmost importance. [Ulrich/Glade], [Ulrich (in Hofmann 2004)], [Arkenbout], [Jansens/van Rosmalen (in Hurle)], [Matsuoka (in Garside)], [Sloan/McGhie], [Toyokura/Hirasawa (in Mersmann)], [Ulrich 2002 (in Myerson)] In recent years a number of new monographs on industrial crystallization have appeared and these provide a good representation of the state of the art. A good overview of the developments in the field over the past 30 years can be gleaned from the proceedings of the International Symposia on Industrial Crystallization which take place at triennial intervals and are organised by the European Federation of Chemical Engineering (EFCE). [Mullin], [de Jong/Jancic], [Jancic/de Jong 1982], [Jancic/de Jong 1984], [Nývlt/Zacek], [Mersmann], [Rojkowski], [Biscans/Gabas], [Garside 1999], [Chianese], [Ulrich 2005]


UNESCO - EOLS

S

SAMPLE C

HAPTER


The key factors in the design of any thermal separation process are the thermodynamics of the system under consideration as well as its kinetics. The thermodynamics defines the limits of what can be achieved, the kinetics define the time-scale and therefore the size of the equipment required. This general statement also holds for crystallization. As a consequence, fundamental knowledge of phase diagrams and solubilities of the materials to be separated is required prior to the design of any industrial process. In many instances the literature only provides phase diagrams for pure materials. With increasing number of components phase diagrams become progressively more complex. Realistic systems generally contain many components and determining phase behavior under operating conditions is one of the greatest challenges in crystallization. Knowing the theoretical limits of a crystallization process as defined by the thermodynamics is only part of the picture, the other part is knowing what can be achieved in a finite and reasonable time scale and the information on this aspect of crystallization is contained in the process kinetics. Two processes are important in crystallization, both with their own characteristic kinetics. One of these processes is nucleation [Kashehiev], the other is crystal growth. Both of these phenomena are dependent on a large number of variables that in some cases may be ill defined. In addition, further factors that influence the number and nature of particles obtained have to be considered in the design of mass crystallization processes. Whenever suspensions of crystals in solution are involved, attrition and agglomeration have to be taken into account. In this chapter only the “real” kinetic parameters, nucleation and crystal growth, will be discussed. Finally crystals possess an internal structure, an external shape and consequently a finite size (or size distribution in the case of a quantity of crystals). These parameters determine many bulk properties of a given crystalline material, such as dissolution rate, bio-availability, color, flow properties etc. A general overview of these properties will also be provided. 2. Solid-Liquid-Equilibria This section provides an overview of solid-liquid equilibria and the type of information useful to crystallization that can be gleaned from their representation in the form of phase diagrams. Different types of – idealized – phase and solubility diagrams shall be discussed together with associated phenomena. 2.1. Solubilities and Phase Diagrams A solution is a homogeneous mixture of two or more chemical species. For a liquid solution saturation is reached when the liquid phase, in contact with the solid phase, no longer changes its composition. A saturated solution therefore has a constant composition that is not changed by the addition of further amount of the dissolved material. For a two component system, the solubility of one component in the other is dependent on temperature and pressure. For three and more component systems the


UNESCO - EOLS

S

SAMPLE C

HAPTER


solubility of one component also depends on the relative amounts of the other components present. For liquid solutions the pressure dependence is negligible if the pressure difference is small (which is the case in most applications) and will not be considered here. However, the temperature dependence (and also the pressure dependence providing the difference is sufficiently large, see [Moritoki]) of the solubility can be considerable and it is therefore important to state the temperature for which the solubility is reported even under ‘normal’ conditions. In the following the discussion will focus on two component systems only. It is common terminology for solutions of solids in liquids to denote the liquid component as the solvent and the solid component as the solute, even if the amount of solute in the solution exceeds the amount of solvent. Various measures of composition are in use to report solubilities and the use of a particular set of units should be carefully selected to suit the purpose the data are required for. The most common measures are

• Mass of solute per unit mass of solvent (kg/kg) • Mass of solute per unit mass of solution (kg/kg) • Mass of solute per unit volume of solvent (kg/m3 or g/L) • Molar amount of solute per unit volume of solvent (kmol/ m3 or mol/L) • Mole fractions (dimensionless number)

From the above list it is clear that, for the first two cases, it has to be stated explicitly whether composition reported refers to the solvent or the solute in order to avoid confusion. Moreover, it is often necessary to mention the initial state of the solute, as many substances can exist as solvated or unsolvated solids (vide infra). For example, two solutions of sodium carbonate in water will have different compositions if in one case the decahydrate is used as starting material and in the other the same mass of monohydrate is employed. Ambiguity arises if this is not accounted for. The solubility of most materials increases with temperature. However, a number of examples exist, where the solubility shows the reverse trend (sodium sulfate, calcium carbonate, iron sulfate dihydrate, to name but a few). The dependence of the solubility upon temperature is best represented graphically in the form of a solubility curve, which maps the composition of a solution at the solubility limit onto the temperature. A generalized solubility curve is shown in Figure 1 below. The solubility is a continuous function of temperature for a given form of a given material. Polymorphs of the same material, solvates and hydrates (vide infra) generally have different solubilities. A change in solid form is normally evidenced by a discontinuity in the solubility curve, which results from different slope, that is, temperature dependence, of the solubility of the respective forms. In Figure 1 the solubility of a fictitious substance A that forms three phases, two hydrates and one anhydrous phase, is shown. Two discontinuities can be seen marking the transition points between the stable regions for the trihydrate and monohydrate (left) and the monohydrate and anhydrous phase (right). The solubilities of the respective stable phases are indicated by the solid line. In addition, the solubilities of the different metastable forms are indicated by the dotted line. Solubilities are determined experimentally and any convenient analytical technique that provides a quantitative measure of composition can be employed. Not every technique


UNESCO - EOLS

S

SAMPLE C

HAPTER


is suitable for every situation and the decision as to which method is used has to be made on a case by case basis.

Figure 1: Typical solubility diagram showing the dependence of solution phase concentration in equilibrium versus temperature. Transition points between different

phases are highlighted. (from [Nordhoff]). The full line represents the solubility of the stable phase in a given region, the dashed line represents the solubility of the respective

metastable phases. In the simplest case of a two component system containing a solute that is non-volatile and stable with respect to temperature the composition can be determined by taking a given mass of the solution, evaporating to dryness and weighing the remaining solid (solubility measurements should always be based upon a mass of solution due to the higher accuracy compared to volume measurements). The more components present and the more complex the system, the more difficult it becomes to accurately determine the composition by a single analytical technique. A number of factors have to be accounted for when determining solubilities. First and foremost, it is important to ensure that the solution is indeed in equilibrium. This is particularly important in the case of highly concentrated, viscous solutions where dissolution rates may become vanishingly small the closer one approaches the equilibrium composition. Here it is important to provide good agitation and to allow for sufficient time before taking a measurement. One method to assess whether equilibrium has been reached is to take measurements from two solutions, one where saturation is approached from an undersaturated state, the other where saturation is approached from


UNESCO - EOLS

S

SAMPLE C

HAPTER


a supersaturated state. Both compositions should be identical within experimental error limits if equilibrium has been achieved. At this point it is important to stress that crystallization is a non-equilibrium process. The value of knowing the equilibrium properties of solutions lies in the fact that they dictate the operating conditions for the crystallization process. The driving force required for nucleation and crystal growth is the level of supersaturation in the solution (supercooling in melts). This means, that crystallization can only occur at solution compositions where the amount of solute exceeds the solubility limit. Such solutions are called supersaturated solutions and the region of phase space where supersaturated solutions exist is known as the metastable zone. The driving force for crystallization (both nucleation and crystal growth, see section on kinetics below) increases with increasing supersaturation. Supersaturation can be achieved in a number of ways, most commonly by cooling a saturated solution (for solutes where solubility increases with temperature, solutes with reverse temperature dependence must, of course, be heated) or by evaporation of solvent. As is the case for solubility, several measures exist to express supersaturation. These are the difference between the equilibrium composition at the temperature of the supersaturated solution and the actual composition in the supersaturated solution c

*c

- *c c cΔ = , (1)

the ratio of these two concentrations

/ *S c c= , (2) and, relating these two, the relative supersaturation

/ * -1 c c Sσ = Δ = (3) For melts, the driving force is usually expressed in terms of the supercooling

= * - θ θΔ θ , (4) which represents the difference between the equilibrium temperature of the melt *θ and the actual temperatureθ . Occasionally supercooling also finds use in solution crystallization. Quantifying the supersaturation is important for two reasons. As mentioned above the supersaturation is a measure of the driving force for the crystallization process. The second reason concerns the theoretical yield of the process. The theoretical yield of any crystallization is easy to calculate as it is given by the difference between the actual composition of the feed solution at the starting point of the process and the equilibrium composition at the end-point of the process. This definition already accounts for any temperature gradients employed in the process. Of course the real yield of the process may differ from the theoretical yield. As the supersaturation decreases, crystal growth rates also decrease and the theoretical yield will only be reached in the limit of infinite time. The consequence of this is a lower yield than possible. On the other hand, yields


UNESCO - EOLS

S

SAMPLE C

HAPTER


may be higher than theoretically possible if growth is fast and if the crystals obtained contain a significant amount of liquid and/or impurity inclusions. 2.2. The Metastable Zone Another important non-equilibrium property of a supersaturated solution is the width of the metastable zone. The metastable zone limit denotes the composition of the solution at a given set of conditions where the solution becomes labile and at which point spontaneous nucleation must occur (vide infra). Since nucleation is a kinetically governed process and can be induced by different mechanisms, the metastable zone width is an ill-defined quantity and depends upon time, purity of the solution, cooling or evaporation rate, agitation etc.. The maximum width of the metastable zone is given by those concentrations where instantaneous nucleation occurs once the corresponding temperature has been reached. In other words it is defined by those points where the induction time for nucleation vanishes. All points with finite induction times lie within the metastable zone. The metastable zone width can be measured by determining the point at which first crystal nuclei are detectable. This can be achieved by optical methods such as turbidity measurements or ATR-FTIR (attenuated total reflectance Fourier transform infrared spectroscopy), by measurement of acoustic properties of the solution (ultrasound), or by measuring any other easily measurable physical parameter such as density or conductivity. Unfortunately there is no single analytical technique that is suitable for all substances; the best method has to be determined after careful consideration of the individual system.

Figure 2: Stages of stability in solution crystallization (from [Strege])


UNESCO - EOLS

S

SAMPLE C

HAPTER


Figure 2 shows a generalized solubility diagram which includes the metastable zone width for different induction times (cf [Mersmann 2001]). The line at defines the limit of the metastable zone, at higher concentrations of solute at constant temperature the solution is labile.

ind 0t =

2.3. Phase Diagrams for Melt Crystallization A melt is normally understood to be a liquid close to its solidification point. Usage of the term in terms of industrial crystallization is not quite as rigid and encompasses any homogeneous liquid mixture where all major components as individual entities would solidify above ambient temperature. The only distinction between a solution and a melt is then the fact that, for the solution, at least one major component (the solvent) is liquid under ambient conditions. In melt crystallization it is not particularly helpful to talk about solubilities. Here, solid-liquid phase diagrams are the more appropriate point of reference. In principle, solubility curves are part of the solid-liquid phase diagram and therefore contain the same information. They are useful for determining equilibrium temperatures, theoretical yields, stability regimes of particular phases and the magnitude of the driving force for a given system under given conditions [Ulrich 2003]. Phase diagrams can be represented graphically using any combination of physical variables. For a binary system only three parameters are required to define the state of the system, pressure, temperature and composition. Pressure-temperature diagrams are useful for determining regions of existence of stable modifications. In terms of understanding the “possibilities” of a melt crystallization process, a temperature-composition (T-x) diagram is more useful, given the fact that the pressure dependence of the solid-liquid equilibrium is usually negligible under ambient conditions. However, at high pressures pressure can be used to direct a crystallization process in much the same manner as temperature at lower pressures (see [Moritoki]). In the case of the Tx diagram, the equilibrium temperature is plotted against the composition of the solid and the liquid phase, usually expressed in terms of mole fractions, respectively. There are a number of types of typical solid-liquid phase diagram representing ideal and different non-ideal behaviors. The discussion here will be restricted to an ideal binary system in order to illustrate pertinent points. For other types of solid-liquid phase diagrams (for example eutectic, peritectic or compound forming systems) the authors refer to the books by [Walas], [König], for example. Figure 3 shows a generalized ideal temperature composition diagram. For an ideal binary system where solid and liquid are in equilibrium, the composition of the solid phase at a given temperature is different from the composition of the liquid phase and the T-x diagram therefore consists of two lines separating three regions in phase space. One of these lines represents the composition of the solid phase (solidus), the other the composition of the liquid phase (liquidus). Indeed, it is only this difference of composition that allows the separation of the components from the melt. The region below the solidus line is the region where the solid phase is stable and no liquid phase exists. The region above the liquidus line is the region of stability of the liquid phase. Both phases coexist only in the region circumscribed by the solidus and liquidus lines.


UNESCO - EOLS

S

SAMPLE C

HAPTER


Here the composition of the respective phases in equilibrium with each other at any given temperature is determined by the end points of the tie line (Figure 3a). The tie line is a horizontal line (that is a line at constant temperature) that connects the liquidus and the solidus lines. For a crystallization process, the composition of the product at a given final temperature of the process is independent of the starting point. In this example, any melt with a composition in the range between

Asx and Alx will lead to a solid product enriched in component A with composition

Asx and a liquid enriched with component B with the composition Alx . However, depending upon the exact starting composition, the yield of the solid phase will differ and can be determined by the lever-rule. Figure 3b illustrates this point: starting from a composition xF the temperature is lowered to Tα corresponding to point of solidification at the initial composition of the melt. The corresponding solid composition is determined by the intersection of the tie-line at Tα with the solidus line and is represented by Sx α . However, employing the lever rule shows that the solid yield at this point is zero. At the final temperatureTω

, the composition of the solid is given by Sx ω and the composition of the liquid phase is

lx ω.

The relative proportions of solid and liquid is given by the ratio of the lengths of the lines R and C.

Figure 3: Solid-liquid temperature-composition (Tx) phase diagram for an ideal binary system.

On the left the three regions of phase space separated by the solidus and liquidus lines can be seen. Within the two-phase region the composition of the respective phases is determined by the locus of the end-points on the respective equilibrium lines. On the right the progress of a crystallization process beginning at a composition F is shown (see [König (in Ulrich/Glade)]). Figure 4 shows a simple eutectic phase diagram, which represents the most common phase behavior of binary organic systems (about 85% of all known organic binary systems are eutectic or pseudo-eutectic). The eutectic phase diagram is characterized by four regions, the homogeneous liquid, two regions of coexistence, and an


UNESCO - EOLS

S

SAMPLE C

HAPTER


inhomogeneous two-phase solid consisting of pure component A and pure component B. The two regions of coexistence are separated by the eutectic point, which is defined by the minimum in the liquidus curve and where the solid and the liquid have the same composition. The coexistence region to the left of the eutectic point describes a system consisting of pure solid A (S1) and the liquid phase, the region to the right of the eutectic point, conversely, describes a system consisting of pure solid B (S2) and the liquid phase. Eutectic behavior has important implications for crystallization. Whereas in the case of an ideal solution described above any degree of purity can be achieved given sufficient separation stages (the number of stages required to obtain a pure substance is infinite), in this case a pure component can be obtained, in an ideal case, in a single separation step. In contrast to the ideal solution, the yield is no longer a function of the initial composition of the melt alone, it also depends upon the position of the eutectic point in the phase diagram. Starting with a melt composition to the left of the eutectic point (F1) yields only pure component A, pure B cannot be crystallized. Moreover, the yield decreases, the closer the initial melt composition is to the eutectic point. The same holds for the situation where the initial composition is to the right of the eutectic point (F3), where only pure component B is obtained. For a melt with eutectic composition, no separation is possible as the solid composition is the same as the composition of the liquid. In principle all other types of phase diagram (compound forming and peritectic systems) can be derived from the ideal solution and the simple eutectic system. For a more detailed discussion of phase diagrams we refer to [König (in Ulrich/Glade)], [Walas].

Figure 4: Simple eutectic phase diagram. On the left the four regions are shown with their respective phases. The right-hand diagram indicates the different outcomes of crystallization from different starting points. For a liquid of composition F1 one obtains pure A and a solid with eutectic composition, for a liquid of composition F3 pure B is obtained together with the eutectic solid. In the case of a liquid with composition F2, no purification is achieved as the liquid and the solid have the same composition [König (in Ulrich/Glade)].


UNESCO - EOLS

S

SAMPLE C

HAPTER


3. Kinetics 3.1. Nucleation Given a sufficient driving force, that is supersaturation in the case of solutions or supercooling in the case of a melt, a liquid-to-solid phase transformation commences with the initial formation of clusters, ordered collections of the crystallizing species. These clusters, or nuclei, are precursors to the crystals eventually formed. In order to grow into a macroscopically detectable crystal, these nuclei have to reach a certain, critical, size. The critical nucleus size is governed by the excess free energy ΔG of the nucleus which, in turn, is given by the sum of the surface excess free energy ΔGS and the volume excess free energy ΔGV of the particle (Figure 5). The surface excess free energy is proportional to the square of the radius of the particle and is a positive quantity, the volume excess free energy is a negative quantity and is proportional to r3. Figure 5 shows a qualitative plot of these different quantities. The critical radius is a measure of the critical nucleus size and is determined by the maximum in the excess free energy curve. Below the critical radius redissolution of the nucleus is energetically favorable as evidenced by the slope of ΔG, above the critical radius growth is more favorable as this leads to a reduction of the excess free energy.

Figure 5: Gibbs free energy versus radius of the nucleus


UNESCO - EOLS

S

SAMPLE C

HAPTER


However, different nucleation mechanisms exist [Strickland-Constable] and the discussion above applies to primary homogeneous nucleation, which will occur if the system is free of impurities, mechanically undisturbed and when the thermodynamic non-equilibrium is at or beyond the metastable limit. In real life this situation is only rarely achieved. Under normal circumstances there are always impurities in the solution or melt, be they other chemical species such as by-products from synthesis, or particulate impurities such as dust or particles resulting from abrasion from the equipment. Mechanical disturbances result from agitation of the solution or vibrations from ancillary equipment. Surface roughness of the equipment also falls into this classification. The more realistic situation is therefore the case of primary heterogeneous nucleation which occurs at a much lower supersaturation and where impurities or rough vessel walls function as nuclei. The most frequently observed nucleation mechanism is called secondary nucleation. Secondary nucleation requires the presence of crystals of the material to be crystallized and occurs at much lower supersaturation than primary nucleation. As a rule, secondary nuclei are formed by the removal of structured assemblies from the surface of the crystals. There are different mechanisms which lead to those secondary nuclei, these are: • initial breeding: nuclei result from simply placing crystals into a supersaturated

solution or supercooled melt via the washing off of dust particles from the surface of the crystals

• collision breeding: nuclei result from fragments of existing crystals which are broken off due to mechanical impact on crystal faces due to crystal-crystal, crystal-wall, or crystal-stirrer (-pump) collisions

• fluid shear: nuclei result from clusters or outgrowths being forced from the solid-liquid boundary layer due to shear forces resulting from liquid motion. A prerequisite for this behavior is that the growing crystal already has a size larger than the critical nucleus

Collision breeding is the most frequently observed and dominant secondary nucleation mechanism, at least in the majority of industrial, mass production, crystallization processes. There are two aspects to secondary nucleation. The positive aspect results from the fact that without secondary nucleation as a permanent source of new crystal nuclei, a continuous crystallizer with continuous crystal withdrawal would rapidly experience a lack of growing crystals. The negative aspect is the fact that many more secondary nuclei are produced in an uncontrolled process than are required. As a result a very fine crystalline product is produced, unless measures are taken to reduce the power input into the crystallizer. More important than the magnitude of the power input itself is the means by which the power is brought into the equipment. Power is required to homogenize the suspension (temperature and concentration, dispersion and circulation of solids) and to transport the suspension. The key power input sources are pumps and impellers, and these are where most secondary nuclei are produced. Secondary nucleation rates can be controlled via diameter and tip speed of the impeller blade. Lower tip speeds and larger diameters result in lower secondary nucleation rates.


UNESCO - EOLS

S

SAMPLE C

HAPTER


Should control of tip speed and enlargement of impeller diameter not be sufficient to reduce the amount of fines to below a value of the order of 1010 in terms of number per cm3 (while still less than 1% in terms of mass percentage), other measures may have to be taken to control the product. In general not all nuclei will grow fast enough to become product size particles due to the fact that each crystal has its own, individual growth rate (this phenomenon is called growth rate dispersion). In some cases small particles will agglomerate. In all the other cases a fines trap has to be introduced to the crystallizer, either as an internal or an external design feature. A fines trap consists of an additional loop within the crystallizer in which the conditions are such, that all fines smaller than a given particle size passing though this loop are redissolved. Nucleation is also a problem concerning the start-up of a crystallizer. Primary nucleation is difficult to control and unreliable, as it will not always occur at precisely the same supersaturation: primary heterogeneous nucleation depends on the number and the nature of the impurities and is therefore, within certain limits, a random event. As a consequence reproducibility of product quality cannot be guaranteed and the performance of the crystallizer will vary. Assuming that nuclei start to grow at high supersaturation, liquid inclusions or dendritic growth (vide infra) are likely, as well as massive formation of small particles that tend to agglomerate. All of these phenomena lead to poor product purity and quality. In addition, this may lead to strong tendency for caking in storage. If nucleation commences at a too low a supersaturation, the ensuing crystal growth may be slow and will pose a problem with respect to production time or crystallizer size, respectively. In order to produce high quality crystals in a reproducibly manner, secondary nucleation by means of seeding is the preferred method of inducing the crystallization process. Here it is important always to introduce the seed crystals at the same supersaturation. Keeping the driving force (the supersaturation) constant subsequent to seeding normally results in high quality crystals, providing the supersaturation selected coincides with the optimum growth rate for the system under consideration. The optimum growth rate is either determined from laboratory tests guided by the rule that a crystallizer should run under such conditions that the concentration in solution remains at the centre of the metastable zone [Hofmann in Hofmann]. In order to maintain constant growth rates requires good control of the supersaturation and has to take into account the constantly increasing crystal surface area, which is ultimately responsible for the reduction in supersaturation. In order to conduct the seeding correctly it is necessary to have a deeper understanding of the system than merely knowing the right supersaturation (which differs for each system). In addition, relevant factors are the seed mass, seed size, seed surface quality (perfect or irregular), the seed state (dry or in suspension) and finally where in the crystallizer the seeds should be added. Some general rules can be stated for seeding [Beckmann], [Heffels/Kind], [Doki], [Warstat/Ulrich]. The surface area of the seeds added has to be large enough to avoid any additional nucleation at the moment the process is started or while it is in its early stages, otherwise a bimodal crystal size distribution will result. On the other hand the seeds should be small enough and low enough in number to produce the desired amount


UNESCO - EOLS

S

SAMPLE C

HAPTER


of product in the process. Choosing too large a seed crystal size leads to a low yield, too many seed crystals lead to a small product crystal size. A properly carried out seeding procedure together with a good control strategy for the supersaturation and an optimised, gentle agitation provides the basis for a reproducible and controlled crystallization process in terms of number and size (size distribution) of crystals in the product. Many open questions remain in this area and until a complete model or theory regarding seeding is available this will remain an active field of investigation. 3.2. Crystal growth As for nucleation, crystal growth only occurs if there is a driving force as a result of non-equilibrium thermodynamic conditions. The crystal growth rate is a physical property of a given material. However growth rates depend not only on the temperature, pressure and composition of the mother liquor but also on parameters such as supersaturation (under cooling), fluid flow conditions, history of the crystals, the nature of the surfaces of crystals and the presence or absence of additives (impurities) in the mother liquor. Crystal growth rates are of key importance to the engineer since they determine the retention time and therefore the size of the crystallizer, which consequently means the investment costs. Linear growth rates for crystals grown from solutions (suspensions) are of the order of 10–7 to 10 –9 m s-1. For melt layer growth the processes can feature growth rates in the range of 10 –5 to 10 –7 m s-1 where forced circulation of the homogeneous melt is employed. Different measures of growth rates are in use other than the increase of a characteristic length denoted by G. Growth rates are also measured in terms of mass increase per surface area, R, or in terms of the displacement of a single crystal face, v. Each of these growth rates can be converted to the others, providing the shape of the crystals remains constant during crystal growth. In that case the following equations hold:

s v2 / /(3 ( ) )( / )G v dL dt f f A dm dt Rρ= = = (5) Here L is the characteristic length, t the time, A the surface area, ρ the density and ( sf ) and ( vf ) the surface and the volume shape factor, respectively. Since crystallization is a heat and mass transfer process, three steps always have to be considered. These are: • transport of material to be crystallized from the bulk solution to the vicinity of

the crystal surface • transfer of material from the solution boundary layer to the solid state (generally

referred to as surface integration) • dissipation of the heat of crystallization liberated at the point of growth


UNESCO - EOLS

S

SAMPLE C

HAPTER


Although these three physical phenomena always occur, one or the other may dominate the crystallization process. As a rule-of-thumb, the dissipation of the heat of crystallization can be neglected. In solution crystallization the processes occurring for the remaining two cases are referred to either as “surface integration controlled” or “diffusion controlled” growth if one or the other step is rate determining. In melt crystallization the heat control is generally viewed as the rate determining step. The same mechanisms that apply to crystallization also apply to dissolution processes. Crystal growth rates depend upon temperature [Ulrich 1993], supersaturation (supercooling) and the nature of the fluid flow in the vicinity of the crystal surface. As a rule, the growth velocity is greater the higher the temperature and the higher the supersaturation. The influence of the fluid flow is not quite as straightforward. However, there exists a limit above which an increase in the flow velocity has no effect on the growth velocity. This limit coincides with a minimum thickness of the solid-liquid boundary layer that determines the mass transfer. The nature of the crystal surface also plays a role in the growth rate and therefore the history of the crystals is an important factor. The perfection (or lack thereof) of the surface can influence the growth velocity in both directions, either accelerating or suppressing crystal growth. At this point it should be stressed that each crystal has its own, individual face growth rates due to the factors discussed above. These individual rates can range from zero to a maximum growth rate characteristic for the material under consideration. This phenomenon crystals displaying individual growth rates despite starting to grow under (nominally) identical conditions with respect to crystal size and growth environment is termed growth rate dispersion (GRD) [Ulrich 1993], [Ulrich 1989]. As a consequence and from the point of view of industrial crystallization it is important to collect relevant kinetic data. It is clear from the above discussion, that an average growth rate for an assembly of crystals is required. Although individual growth rates for single crystals are informative in terms of determining the development of crystal shape (habit) during the growth process, many crystals would have to be monitored in order to gather statistically significant information that incorporates, for example, the effect of GRD and is suitable for use in the design of industrial crystallization equipment. Therefore, methods providing access to mass growth rates utilizing either stirred tanks or fluidized beds are to be preferred for this purpose. Here fluidized beds operated with seed crystals of known size and quality are the better choice since all parameters of interest (vide supra) can be varied independently [Mohameed], [Kruse], [Al-Jabburi]. The growth rate is measured monitoring the mass increase over time for a constant number of seed crystals of known size. Attrition of the crystals can be neglected in most cases, due to the gentle operating conditions applied in of small fluidized beds. Particles resulting from attrition, should they be generated under the operating conditions selected, are normally dissolved in that part of the fluidized bed, where the solution is undersaturated. Classical crystal growth theories are of course important for an understanding of the growth process and a number have been developed and improved over the years.


UNESCO - EOLS

S

SAMPLE C

HAPTER


However, for the chemical engineer dealing with mass crystallization they are not of key importance. Growth theories are discussed in detail in all standard text books on industrial crystallization [Ohara/Reid], [Mullin 1992], [Jancic], [Myerson 2002], [Ulrich 2002]. Most crystal growth theories only consider the solid (crystal) and tend to neglect the influence of the conditions in the liquid. Nonetheless, there exists enough experimental evidence that demonstrates that the liquid and its composition are of great importance in the crystal growth process. The most important growth models are the spiral growth model incorporated in the Burton-Cabrera-Frank (BCF) theory of crystal growth and the so-called ‘birth and spread’ model. The BCF kinetic theory of growth considers the density dislocation leading to spiral growth on a crystal surface and the supersaturation. For low supersaturation a quadratic dependence of the growth rate as a function of the supersaturation results which changes to a linear dependence at high supersaturation. The ‘birth and spread’ model makes the assumption that growth occurs due to surface nucleation. Here the supersaturation dependence of the growth rate is proportional to

5/6σ . Further details can be found in the standard monographs cited above. The normal engineering approach to growth rates is to determine these experimentally as a function of supersaturation at a given temperature using the original mother liquor and well defined seed crystals. In the case of batch crystallization in particular, it is extremely important to verify that the laboratory scale crystallization leads to the same quality of product (purity and crystal growth rates), especially when the batch-time is nearing its end and the majority of impurities present have accumulated in the remaining mother liquor. In some cases the accumulation of impurities may not only affect growth rates but may also influence the resulting crystal habit. In order to maintain good yields and high product purity it is desirable to maintain a constant supersaturation throughout the crystallization process. Especially in the case of batch processes this requires permanent adjustment of the driving force. Industrial experience suggests that a solution crystallization process should best be operated at the centre of the metastable zone. If the process operates close to the upper limit of the metastable zone, a small fluctuation in the process variable is sufficient to induce nucleation. This would result in a reduction of the crystal size distribution and facilitate oscillations in the of the crystallization process with respect to the output of crystal size distributions. Conversely, if the process is operated close to the solubility limit, insufficient mass per time is produced. Neither natural or linear cooling/evaporation provide a constant driving force. As the available crystal surface area increases with growth of the crystals, the rate of reduction of supersaturation increases under linear or natural cooling (evaporation) conditions. In order to maintain constant supersaturation, an appropriate cooling or evaporation program must be employed. The previously mentioned increasing impurity concentration in the remaining mother liquor can not only change the thermodynamics of the system, but can also exert a strong influence on the kinetics of crystal growth. These effects have to be understood


UNESCO - EOLS

S

SAMPLE C

HAPTER


prior to the design of a crystallizer in order to introduce appropriate control measures if required. 4. Properties of Crystals 4.1. Crystal Structures In contrast to amorphous solids, liquids and gasses, crystals possess a well defined 3-dimensional order. Fortunately, symmetry considerations dictate that only a limited number of structures exist in which a space-filling packing of building blocks is possible [Hahn]. The crystal structure is important even in industrial mass crystallization as the internal arrangement of the building blocks defines the macroscopically observable shape of the crystal and thus influences post-crystallization processing, that is solid-liquid separation, drying and packaging. Energy considerations lead to the conclusion that for each substance there is only one minimum energy structure. However, it is not correct to draw the obverse conclusion that only one crystal structure is possible per chemical species, since the most stable crystal structure may be different under different conditions of temperature and pressure. Unfortunately, depending upon the crystallization conditions, a given material may crystallize in different modifications that are not necessarily minimum energy structures at the given conditions. Although these structures are thermodynamically metastable, the kinetics of transformation to the stable phase may be (infinitely) slow so that these additional crystal modifications must be treated as stable in the context of product properties and life-time. This phenomenon, the coexistence of chemically identical, thermodynamically metastable phases alongside the stable phase is known as Polymorphism. Polymorphism has to be taken into consideration in the design of any crystallization process since different polymorphs will exhibit different physical properties as a result of their different crystal structures. This is particularly critical in industries where certain product properties (such as density, solubility, dissolution rate, bioavailability, color, hardness) are important, for example in the pharmaceutical industry. Since it is not yet possible to predict polymorphs reliably, screening is an important step in the development of a drug substance and polymorph control is essential during the development of a crystallization process: providing a patient the wrong polymorph of a given drug may lead to inefficacy in the best case, or to a lethal dose in the worst case of too rapid dissolution. In addition to the polymorphs, there may also exist a number of so-called pseudopolymorphs for a given species. Pseudopolymorphs are solids that consist of two distinct chemical species, one of which is a solvent molecule that forms an integral part of the crystal structure, the other is common to the different materials. For this reason the abovementioned terminology, though widely used, is unfortunate and these crystalline materials are better referred to as solvates or, in the case of integration of water, hydrates. Hydrates are often observed in inorganic salts. Frequently one substance has several different hydrate forms, that is number of different materials exist that vary in the amount of water contained in the crystal lattice. The same behavior can be observed for organic crystals both with water incorporated into the lattice or with an integrated organic solvent. Furthermore, materials have been observed to incorporate more than one type of solvent molecule. The existence of solvates and hydrates


UNESCO - EOLS

S

SAMPLE C

HAPTER


therefore not only depends upon the temperature and pressure at which they are crystallized, their stability window also depends upon the composition of the solvent. One substance may crystallize in a number of polymorphic forms and also in the form of different solvates or hydrates. Even solvates and hydrates can exist in different polymorphic forms. It is therefore clear that the design and control of reliable and reproducible crystallization processes, in particular in the case of drug substances where reproducibility is of paramount importance, poses a difficult task. Despite the inherent difficulties, polymorphism can also be viewed as an opportunity for overcoming problems with perhaps undesirable physical properties of energetically stable structures: by judicious choice of crystallization conditions and by producing a particular polymorph with certain, desired properties, the product properties can be tailored to suit. The existence of polymorphs, solvates and hydrates leads to the potential for unexpected phase transitions during the production process which may result in unexpected products with unwanted properties. For example, a hydrate may be sensitive to ambient humidity and lose its water if humidity becomes too low. Contact with another solvent may also induce phase transformations. It has been shown, for example, that certain hydrates lose their water when brought into contact with organic solvents. The accompanying phase transition results in acicular (needle-like) crystals that form a dense, felt-like mass. This crystalline mass becomes difficult to process. On the one hand it no longer presents a freely flowing slurry, on the other hand it retains a large amount of fluid that is difficult to remove using standard solid-liquid separation techniques. The first case also poses a challenge in solid-liquid separation, in particular if a phase transition occurs during the filtration process. Here one may be left with a solid block of an unwanted phase that has to be painstakingly removed and discarded [Ulrich/Jones], [Malet]. 4.2. Crystal Shape (Habit) The external shape of a crystal is determined by the crystal structure, that is, the internal order of the individual building blocks as determined by the space group symmetry. The crystal faces expressed by the macroscopic crystal always map a lattice plane defined by the microscopic unit cell. Nonetheless, crystals of the same chemical species with the same crystal structure can exhibit different habits. This is due to the fact that growth rates of individual crystal faces, and therefore their relative morphological importance, are subject to supersaturation effects and the influence of solvent and impurity molecules. Crystals of the same material grown from different solvents, for example, may have different aspect ratios or may even display different crystal faces. Fast growth due to high supersaturation has a great effect on crystal habit and in extreme cases leads to dendritic growth resulting in highly branched and fragile crystals. Fast growth is desirable from an economic point of view since fast growth means short residence times and smaller equipment, i.e. lower investment costs. Apart from the possibility of dendritic growth, rapid growth rates have other disadvantages that result in detrimental product quality. Fast crystal growth encourages liquid inclusions in the crystals and often leads to rough surfaces. Liquid inclusions are critical in two respects: not only do increasing amounts of included liquid reduce the overall product quality, the accumulation of impurities in the mother liquor over the crystallization process also reduces product purity.


UNESCO - EOLS

S

SAMPLE C

HAPTER


The other factor influencing crystal habit mentioned above is the effect of impurities. No mother liquor is ever truly free of impurities and it has long been recognized that these impurities can lead to changes in the observed crystal shape. In terms of post-crystallization processing this can have a negative effect on filterability, drying, or on the handling of the dry solids, in particular if crystals with an acicular or tabular (plate-like) habit are produced. On the other hand, the effect of impurities can be of advantage to tailor the shape of the crystals. By exploiting the different interactions of different chemical species with the growing crystal faces, it is possible to manipulate the final macroscopic shape of the crystals by deliberately adding an impurity (additive) to the crystallization liquor. The fastest growing crystal faces grow out and are not observed in the final crystal habit. One strategy to influence the shape of the crystal is therefore to find an additive that reduces the growth rate of faster growing faces while leaving the growth rates of the slower growing faces unchanged. This can be achieved by identifying additives that fit into the lattice on the fast growing surfaces (but not the others) and therefore block the addition of the regular growth units. Several so-called ‘tailor-made’ additives have been reported in the literature [Weissbuch], [Chen], [Clydesdale]. 4.3. Crystal Size and Crystal Size Distribution In mass crystallization crystal sizes are normally discussed in terms of the mean diameter of a size distribution. A quick glance into any book on particle technology will reveal that there are a number of different definitions of “mean diameter” in use. For the purpose of this discussion it is important to point out that any researcher dealing with size distribution has to be careful to select the correct definition for the problem in hand. Details of these definitions are not the object of this section. In industry requirement are often stated quite generally such that a product has to meet a certain specification with regard to the proportion of particles above or below a certain size limit. One means of tightening the specification of a size distribution is to define a second characteristic value, usually the width of the desired distribution. This still leaves the shape of the distribution open. One particular crystal size distribution function that should be named at this point is the RRSB (Rosin, Rammler, Sperling, Bennett) distribution, since it provides a good fit in many cases. The size (which from this point onwards should be taken to mean the mean size) of a crystal is important for a variety reasons: flow properties, color, dissolution rates, agglomeration tendencies, mixing and demixing properties, filterability etc. can all depend upon the crystal size. Even though larger crystals are often desirable, there are cases where smaller crystals are of key interest. Nanotechnology as well as dyes and pigments are two areas where small crystals can exhibit properties that are lost above a certain upper limit. A significant body of work concerning the controlled production of uniformly small crystals has emanated from the photographic materials industry. These small crystals and their controlled production is, however, a typical domain of so-called precipitation crystallization which is not the object of this discussion.


UNESCO - EOLS

S

SAMPLE C

HAPTER


Apart from achieving a well defined and often narrow width of the crystal size distribution, it is often desirable to have a narrow, mono-modal distribution which can be viewed as single-sized. The term ‘narrow distribution’ alone will have different meanings in different situations. Crystals of table salt (sodium chloride), for example, must be small enough to fit through the openings of the salt cellar (generally smaller than 400 µm), yet large enough not to exhibit the typical properties of a powder such as caking. Powders are generally not free-flowing since the attractive forces between the crystals are stronger than the gravitational force. The critical lower size limit where powder properties become noticeable is dependent on the material. In suspension crystallization it is not possible to produce a narrow, single-sized distribution of crystals. Even when starting from seed crystals of the same size with the same feedstocks with highly homogeneous growth conditions for all crystals in all regions of the crystallizer, a distribution with finite width will always be the result. There are several reasons for this, primarily growth rate dispersion. Furthermore, inhomogeneities in the growth conditions can never be excluded (spatial variations in temperature, supersaturation, concentration of impurities or fluid flow patterns), however, methods exist to control, minimize or reduce excess nuclei and to produce a product close to the desired size distribution. 5. Crystallization Technology In this chapter, typical crystallization equipment frequently found in industrial applications is presented. Special cases are not discussed here, even though many applications require tailor-made solutions. The equipment discussed in the following is based upon “standard” designs frequently sold by the major manufacturers of crystallizers. 5.1. Solution Crystallization Figure 6 shows the three most frequently built, continuous evaporative solution crystallizer types, together with their heat exchangers. In common with all evaporative crystallizers, the evaporation zone lies towards the top of the crystallizer at the interface of the solution with the atmosphere. Inside the crystallizers the supersaturation must be kept within the metastable zone at all locations in order to avoid primary nucleation and to provide controlled conditions for crystal growth. A forced circulation (FC) crystallizer, as shown in Figure 6, left, is driven by an axial flow pump that circulates the entire suspension (solution and crystals) through the heat exchanger. As a result all crystals pass through the pump and are therefore liable to impact with the pump impeller, resulting in attrition and contributing to secondary nucleation. This limits the maximum achievable crystal size. Larger crystals can be obtained using the draft tube baffled crystallizer (DTB, Figure 6, centre), where only the clear solution, together with some fines, is circulated through the heat exchanger, while the crystals are circulated at low speed in the apparatus by means of a very large impeller pump (often exceeding 1 m in diameter) located within the draft


UNESCO - EOLS

S

SAMPLE C

HAPTER


tube. The low circulation rate results in less attrition of the crystals due to reduced mechanical impact with both pump components and other crystals. The largest crystals can be obtained with the “Oslo”-type crystallizer (Figure 6, right). Here crystals grow in a fluidized bed created by the supersaturated solution circulating from the evaporation zone down the central draft tube towards the bottom of the equipment and from there back upwards to create the fluidized bed that is located in the lower part of the crystallizer, between the tube and the inner wall of the crystallizer. Only cleared solution is circulated through the heat exchanger. Gentle flow conditions and longer retention times are required in order to obtain larger crystals and there is no size reduction due to pumps or impellers.

Figure 6: The three most important types of evaporative solution crystallizers. Left, the forced circulation (FC) crystallizer, centre, the draft tube baffled (DTB) crystallizer and right, the Oslo crystallizer. Images kindly provided by Messo Chemietechnik, Germany In all evaporative crystallizers the liquid layer just beneath the evaporation zone is the region with the highest supersaturation. As the liquid interface is subject to periodic variations in height due to the effect of motion in the equipment, this area is prone to encrustation due to variable wetting and resultant nucleation. Large crystals require long retention times, long retention times require large equipment and large equipment results in expensive processes. Whenever possible, the application


UNESCO - EOLS

S

SAMPLE C

HAPTER


of FC crystallizers is preferable to DTB or Oslo-type crystallizers due to their lower cost. Cooling crystallizers for suspension crystallization techniques can be treated in the same manner as melt crystallization equipment as discussed in the following. The key problem is to avoid nucleation at the cooled surfaces. Scraped surfaces are used where encrustation does occur (leading to ill- formed crystals). Alternatively, a process design avoiding too high a supercooling at the walls or too long a retention time at those locations that are prone to scaling is required. A typical process flow sheet for a solution crystallization process is provided in Figure 7. The flow sheet represents a four stage sodium chloride purification process. The intake is brine with a flow rate of 60 t/h. The first, Oslo-type crystallizer produces large crystals of the order of 2 mm side length at a rate of 2.5 t/h, the remainder of the plant (three successive FC crystallizers followed by post-crystallization processing) produces regular table salt at a rate of 6.5 t/h. The size of the equipment can be gauged via the total height of the building, which is 30 m, and the surface area required, which is 2000 m2 for such a process. Further information on evaporative solution crystallizers can be found in [Hofmann (in Hofmann)].

Figure 7: Process flow sheet for a sodium chloride crystallization plant. From the left: the first, Oslo-type crystallizer produces large sodium chloride crystals, the three forced circulation crystallizers produce standard table salt. Following the crystallizers, settlers

for solid-liquid separation, cyclones for fines removal and storage silos can be seen. Image kindly provided by Messo Chemietechnik, Germany.

5.2. Melt Crystallization Melt crystallization can be subdivided into two distinct methods. These are suspension based and solid layer based processes. The different techniques are described in detail in the literature cited, for example [Arkenbout] or [Ulrich/Glade]. The suspension based processes are frequently combined with so-called wash columns ([Scholz/Ruemekorf


UNESCO - EOLS

S

SAMPLE C

HAPTER


(in Ulrich/Glade)], in order to achieve a better solid-liquid separation. A flow sheet for a plant from Niro Process Technology B.V. used for p-Xylene purification is shown in Figure 8.

Figure 8: Flow sheet detailing a (generalized) melt crystallization process with wash column. Figure kindly provided by Niro Process Technology B.V., The Netherlands.

Solid layer melt crystallizers follow a different concept. Here the product is obtained by enforced encrustation. Heat is withdrawn from the melt through the crystalline layer by means of cooled walls within the crystallizer. The driving forces are much higher than in the case of suspension crystallization since the feed melt always remains above its melting point. Growth rates are about two orders of magnitude higher than in suspension crystallization (10-7 – 10-5 m/s compared to 10-9 – 10-7 m/s). However, in melt crystallization the surface area of the product is limited and about two orders of magnitude lower than in suspension crystallization. The photograph in Figure 9 illustrates the full scale of the process equipment. The crystallization vessels can be seen on the right hand side of the image, the wash columns (screw-type) are on the left behind the feed heat recovery exchangers. Solid layer techniques are almost exclusively discontinuous, but they can be staged. Due to the very fast growth rates single stages lead to rather impure products (both crystalline and melt fractions). These are normally collected and reused in subsequent stages. Further purification and higher yield can be achieved in this manner. Storage tanks are cheaper than crystallizers and for this reason all stages of a given process are usually carried out in a single crystallizer and the intermediate products are stored in tanks until such a time that enough material has accumulated for the next stage.


UNESCO - EOLS

S

SAMPLE C

HAPTER


Figure 9: Melt crystallization plant used for the purification of p-Xylene. The crystallization vessels can be seen to the right. Figure kindly provided by Niro Process

Technology B.V., The Netherlands. Figure 10 shows a flow sheet for such a design. Central to the plant is a falling film crystallizer. Detailed descriptions of melt layer crystallization can be found in [Stepanski/Schäfer (in Ulrich/Glade)].

Figure 10: Typical flow sheet for a staged solid layer crystallization plant. Central to the design is a falling film melt crystallizer. To the left of the crystallizer three storage


UNESCO - EOLS

S

SAMPLE C

HAPTER


tanks can be seen. These are used to collect residual melt and purified material, which can then be reused in the process. Flowsheet kindly provided by Sulzer Chemtech Ltd,

Switzerland. The falling film equipment shown above represents the so-called dynamic solid layer technique which requires forced circulation of the melt passing over the cooled surface. Figure 11 provides an example of such a crystallizer. (p-Xylene, CEPSA, Spain).

Figure 11: A falling film melt crystallization plant used for the purification of p-xylene (CEPSA, Spain). Photograph kindly provided by Sulzer Chemtech Ltd, Switzerland

Static solid layer techniques represent the other case encountered in solid layer melt crystallization equipment. Here, the feed melt is simply filled into the equipment and there is no movement of the feed melt apart from natural convection. Figure 12 shows an example of a static solid layer melt crystallizer employing cooled plates rather than cooled pipes. Static processes require slower crystallization rates and therefore higher volumes in order to achieve the same product purity and the same production rate as falling film equipment, but are cheaper to implement and run, as no additional equipment such as a circulation pump is required. The role of the wash column used in suspension crystallization is replaced by washing, rinsing or sweating in the case of solid layer crystallization. Once the crystallization process is sufficiently advanced, the residual melt is drained and replaced by a purer washing liquid in the case of washing or rinsing steps.


UNESCO - EOLS

S

SAMPLE C

HAPTER


Figure 12: Sketch of a simple static melt crystallizer. The cooled plates can be seen in the interior of the crystallizer, inlets and outlets for feed/product and cooling medium

are highlighted. Sketch kindly provided by Sulzer Chemtech Ltd, Switzerland. In the case of sweating, the crystalline layer is heated to close to the melting point of the pure substance in the absence of a washing liquid, in order to drain any impurities or inclusions. All of these post-crystallization techniques lead to purer products and are frequently employed. Further details on post processing steps can be found in the books on melt crystallization cited. The disadvantages that solid layer crystallization is a discontinuous process and provides only limited surface area are often outweighed by the fact that solid-liquid separation is easily achieved in such equipment, that there is no handling of solids and that the equipment is easy to scale-up. As a result the technology is competitive in terms of market value. For suspension crystallization techniques the advantages and disadvantages are reversed compared to solid layer techniques. General rules for the selection of the one or the other technique are still not available and a decision as to which technique is more suitable for a given problem must be made on a case-to-case basis. Glossary Critical cluster size:

A measure of the size of a crystal nucleus required to grow into an energetically stable crystal.

Growth: The process by which crystal nuclei increase in size and develop


UNESCO - EOLS

S

SAMPLE C

HAPTER


into fully formed crystals. Growth rate: The rate with which crystals grow. Different measures of growth

rate are in use. Metastable Zone: The region in phase space where the solution (melt) is beyond

equilibrium but cannot nucleate due to small the small nucleation rate. Once nuclei have formed, they grow in the metastable zone.

Nucleation: The formation of an initial crystalline particle (nucleus) from the supersaturated solution.

Nucleation rate: The rate at which crystal nuclei are formed. The nucleation rate depends upon supersaturation.

Polymorphism: The phenomenon where a chemical entity exhibits more than one crystal structure. Each distinct crystal structure is termed a polymorph and will display different physical properties.

Phase diagram: A graphical representation of the state of a system (existence of distinct phases) depending upon selected physical parameters.

Solubility: A measure of the amount of solute that can dissolve in a given solvent. Solubility is a physical property of a given system.

Supercooling: A measure of the deviation of the system temperature from the equilibrium temperature.

Supersaturation: A measure of the deviation of the solution concentration from the equilibrium concentration.

Bibliography Al Jibbouri S. and Ulrich J. (2004). Impurity Adsorption Mechanism of Borax for a Suspension Growth Condition: A Comparison of Models and Experimental Data, Crystal Research & Technology, 39, 540 – 547 [On the effect of impurities on a crystallization process]

Arkenbout G. F., (1995). Melt Crystallization Technology, Technomic Publishing Company, Inc., Lancaster, Pennsylvania, USA [An overview of melt crystallization technology]

Beckmann W., Nickisch K. and Budde U. (1998). Development of a Seeding Technique for the Crystallization of the Metastable A Modification of Abecarnil, Organic Process Research & Development, 2(5), 298 – 304 [An example of seeding techniques applied to a particular crystallization problem]

Biscans B. and Gabas N. (1996). 13th Symposium on Industrial Crystallization, Toulouse, France [An overview of progress in scientific research in industrial crystallization over the preceding three years]

Brittain H. G. (1999). Polymorphism in Pharmaceutical Solids, Marcel Dekker, Inc., New York, USA [An overview of the effect of polymorphism on the properties and manufacture of pharmaceuticals]

Chianese A. (2002). 15th Symposium on Industrial Crystallization, AIDIC, Milano, Italy [An overview of progress in scientific research in industrial crystallization over the preceding three years]

Chen B.D., Potts G.D., Davey R.J., Garside J., Bergmann D., Niehoerster S. and Ulrich J. (1994). Growth of Meta-Chloronitrobenzene Crystals in the Presence of “Tailor-Made” Additives: Assignment of the Polar Axes from Morphological Changes, Journal of Crystal Growth, 144 (3/4), 297 – 303 [An account of the effect of additives on crystal shape]

Clydesdale G., Roberts K.J. and Lewtas K. (1994). Computational Modelling Study of the growth Morphology of the Normal Alkane Docosane and its Mediation by “Tailor-Made” Additives, Molecular


UNESCO - EOLS

S

SAMPLE C

HAPTER


Crystals and Liquid Crystals Science and Technology, 248, 751 – 784 [An example of molecular simulations applied to crystallization]

de Jong E.J. and Jancic S.J. (1979). Industrial Crystallisation '78, North-Holland, Amsterdam, The Netherlands [An overview of progress in scientific research in industrial crystallization over the preceding three years]

Doki N., Kubota N., Sato A., Yokota M., Hamada O. and Masumi F. (1999), Scaleup Experiments on Seeded Batch Cooling Crystallization of Potassium Alum; AIChE Journal; 45 (12), 2527 – 2533 [An example of scaling up a seeded crystallization process]

Garside J., Davey R.J. and Jones A.G. (1991). Advances in Industrial Crystallization, Butterworth-

Heinemann, Oxford, UK [Selective overview lecture on the state-of-the-art in industrial crystallization in 1991]

Garside J.(1999). 14th Symposium on Industrial Crystallization, Institute of Chemical Engineering, Rugby, UK [An overview of progress in scientific research in industrial crystallization over the preceding three years]

Hahn T. (Ed.) (2002). International Tables for Crystallography, Kluwer Academic Publishers, Dordrecht, The Netherlands [A detailed account of fundamentals of crystal symmetry and space groups]

Heffels S.K. and Kind M. (1999). Seeding Technology: An Underestimated Critical Success Factor for Crystallization, 14th International Symposium Industrial Crystallization, IchemE, Rugby, UK [An account of the impact of seeding techniques in crystallization]

Hofmann G. (2004). Kristallisation in der industriellen Praxis, Wiley-VCH Verlag GmbH & Co. KgaA, Weinheim, Germany [“Crystallization in the industrial practise“, materials from a professional development course]

Hofmann, G. (2004). Übersicht über die behandelten Themen, in: Hofmann, G. (Ed.), Kristallisation in der industriellen Praxis, Wiley-VCH Verlag GmbH & Co.KgaA, Weinheim, Germany [An overview of topics discussed in the above book]

Hurle D.J.T. (1994). Handbook of Crystal Growth, Chap. 6, Vol. 2A, Elsevier Science Publisher, Amsterdam, The Netherlands [An overview of the science if crystal growth]

Jancic S.J. and de Jong E.J. (1982). Industrial Crystallization 81, North-Holland Publishing Co.,

Amstersam, The Netherlands [An overview of progress in scientific research in industrial crystallization over the preceding three years]

Jancic S.J. and de Jong E.J. (1984). Proceedings of the 9th Symposium on Industrial Crystallization,

Elsevier, Amsterdam, The Netherlands [An overview of progress in scientific research in industrial crystallization over the preceding three years]

Jancic S.J. and Grootscholten P.A.M. (1984). Industrial Crystallization, D. Reidel Publishing Co., Dordrecht, The Netherlands [A handbook on industrial crystallization]

Jansens P.J. and van Rosmalen G.M. (1993). Fractional Crystallisation in: Hurle D.J.T. (Ed.). Handbook of Crystal Growth, Chap. 6, Vol. 2A, Elsevier Science Publisher, Amsterdam, The Netherlands [Crystallization as a separation technique]


UNESCO - EOLS

S

SAMPLE C

HAPTER


Kashehiev D. (2000). Nucleation: Basic Theory with Application, Butterworth-Heinemann, Oxford, UK [Fundamentals of the initial formation of crystalline solid particles]

König, A. (2003). Phase Diagrams in Melt Crystallization, in: Ulrich J. and Glade H. (Eds.), Shaker Verlag, Aachen, Germany [An overview of typical phase behavior encountered in crystallization]

Kruse M. (1993). Zur Modellierung der Wachstumskinetik in der Lösungskristallisation, VDI Fortschrittberichte Nr. 209, VDI-Verlag, Düsseldorf, Germany [“Modelling growth rates in solution crystallization“]

Mallet F., Petit S. and Coquerel G. (2003). New Data and Interpretations on the Formation of Whiskers of Molecular Crystals, in: 10. BIWIC, Coquerel G. (Ed.), Verlag Mainz, Aachen, 165 – 171 [Tentative growth model of small, needle-shaped crystals]

Matsuoka M. (1991). Developments in Melt Crystallization, in: Advances in Industrial Crystallization, Garside J., Davey R.J., and Jones A.G. (Eds.), Butterworth-Heinemann, Oxford

Mersmann A. (1990). 11th Symposium on Industrial Crystallization, Garmisch-Partenkirchen, Germany [An overview of progress in scientific research in industrial crystallization over the preceding three years]

Mersmann A. (2001). Crystallization Technology Handbook, Marcel Dekker, Inc., New York, USA [A handbook on industrial crystallization and the technology of crystallization]

Mohameed H.-A. (1996) Wachstumskinetik in der Lösungskristallisation mit Fremdstoffen, Shaker Verlag, Aachen, Germany [“Growth kinetics in the presence of additives in solution crystallization”]

Moritoki, M., Wakabayashi M and Fujikawa, T. (1979). Fractional Crystallization Applying Very High Pressure, in Industrial Crystallization 78, de Jong E.J. and Jancic S.J (Eds.), North-Holland Publishing Company, Amsterdam, The Netherlands [An account of the effect of pressure on a crystallization process]

Mullin J.W. (1976). Industrial Crystallisation, Plenum Press, New York, USA [A handbook on industrial crystallization]

Mullin J.W., (1992). Crystallization, 3 ed., Butterworth-Heinemann, Oxford, UK [A handbook on industrial crystallization]

Myerson A.S. (1999). Molecular Modelling Applications in Crystallization, Cambridge University Press, Cambridge, UK [An overview of molecular simulation techniques applied to crystallization phenomena]

Myerson A.S. (2002). Handbook of Industrial Crystallisation, 2nd ed., Butterworth-Heinemann, New York, USA [A handbook on industrial crystallization]

Nordhoff, S.(1999). Zur Pseudopolymorphie unter Lösungsmitteleinfluß, Shaker Verlag, Aachen, Germany [“On the influence of solvents on pseudopolymorphs“]

Nyvlt J. and Ulrich J. (1995). Admixtures in Crystallization, VCH, Weinheim, Germany [On the effect of additional components on crystallization processes]

Nyvlt J. and Zacek S. (1989). Proceedings of the 10th Symposium on Industrial Crystallization, Elsevier, New York, USA [An overview of progress in scientific research in industrial crystallization over the preceding three years]


UNESCO - EOLS

S

SAMPLE C

HAPTER


Ohara M. and Reid R.C. (1973). Modelling Crystal Growth Rates from Solution, Prentice-Hall Press, Englewood Cliffs, NJ, USA [A treatise on the fundamental physics of crystal growth]

Randolph A.D. and Larson M.A. (1988). Theory of Particulate Processes, 2nd ed., Academic Press, New York, USA

Rojkowski Z.H. (1993). Proceedings of the 12th Symposium on Industrial Crystallization, Warsaw, Poland [An overview of progress in scientific research in industrial crystallization over the preceding three years]

Scholz, R. and Ruemekorf, R. (2003). Separate Organics by Suspension Crystallization, in: Ulrich J. and Glade H. (Eds.), Shaker Verlag, Aachen, Germany [An account of suspension crystallization and its application to organic solids]

Söhnel O. and Garside J. (1992). Precipitation: Basic Principles and Industrial Applications, Butterworth-Heinemann, Oxford, UK [A discussion of precipitation techniques]

Sloan G.J. and McGhie A.R. (1988). Techniques of Melt Crystallization, John Wiley & Sons, New York, USA [An overview on melt crystallization techniques]

Stepanski, M. and Schäfer, E. (2003). Separate Organics by Melt Crystallization: A Guide to When and How to Use this Technique, in: Ulrich J. and Glade H. (Eds.), Shaker Verlag, Aachen, Germany [An account of suspension crystallization and its application to organic solids]

Strege, C. (2004). On (Pseudo-) Polymorphic Phase Transformations, PhD Thesis, Martin-Luther-Universität Halle-Wittenberg, Germany [A research thesis on phase transitions and their impact on crystallization]

Strickland-Constable R.F. (1968). Kinetics and Mechanism of Crystallization, Academic Press, New York, USA [On the kinetics of crystallization]

Tavare N.S. (1994). Industrial Crystallization: process simulation analysis and design, Plenum Press New York, USA [On process simulation as a tool in designing crystallization processes]

Toyokura K. and Hirasawa I. (2001). Suspension Crystallization from the Melt, in Mersmann A. (Ed.), Crystallization Technology Handbook, Marcel Dekker, Inc., New York, USA [An account of suspension crystallization from melts]

Ulrich J. (1993). Kristallwachstumsgeschwindigkeiten bei der Kornkristallisation – Einflussgrößen und Messtechniken, Shaker Verlag, Reihe Verfahrenstechnik, Aachen, Germany [“Crystal growth rates in suspension crystallization – physical factors and measurement techniques“]

Ulrich J. (1989). Growth Rate Dispersion – A Review, Crystal Research & Technology, 24, 249 – 257 [An overview of the phenomenon of crystals growing at different rates under otherwise identical conditions]

Ulrich J. (2002). Crystallization, in Kirk-Othmer Encyclopedia of Chemical Technology (Online), John Wiley & Sons, Inc. [An overview of industrial crystallization]

Ulrich, J. (2004). Solid Layer Crystallization, in: Ulrich J. and Glade H. (Eds.), Shaker Verlag, Aachen, Germany [Fundamentals of melt crystallization]

Ulrich J. and Glade H. (2003). Melt Crystallization, Shaker Verlag Aachen, Aachen, Germany [A detailed account of the fundamentals and applications of melt crystallization]


UNESCO - EOLS

S

SAMPLE C

HAPTER


Ulrich J. (2004). Fremdstoffbeeinflussung in der Kristallisation, in Hofmann G. (Ed.), Kristallisation in der Industriellen Praxis, Wiley-VCH, Weinheim, Germany [“The effect of additives on crystallization”]

Ulrich J. and Jones M.J. (2005). Kristalle mit integriertem Lösungsmittel, Nachrichten aus der Chemie, 53, 19 – 23 [“Crystals with integrated solvents”]

Ulrich J. (2005). 16th Symposium on Industrial Crystallization Dresden, VDI-Verlag, Germany [An overview of progress in scientific research in industrial crystallization over the preceding three years]

van der Eerden J.P. and Bruinsma O.S.L. (1995). Science and Technology of Crystal Growth, Kluwer Academic Publishers, Dordrecht, The Netherlands

Walas S.M. (1985).Phase Equilibria in Chemical Engineering, Butterworth-Heinemann, Boston, London, Oxford, UK [An account of the impact of thermodynamic properties on engineering]

Warstat A., Ulrich J. (2005). Seeding During Batch Cooling Crystallization – An Initial Approach to Heuristic Rules, in: Ulrich J. (Ed.), 16th Symposium on Industrial Crystallization Dresden, VDI-Verlag, Germany [An report on research to develop rules for seeding techniques]

Weissbuch I., Lahav M. and Leiserowiz L. (2001). Crystal Morphology Control with Tailor-Made Additives: A Stereochemical Approach, Advances in Crystal Growth Research, 381 – 400 [An account of the effect of additives on the solid phase crystallized] Biographical Sketches Joachim Ulrich received the Diplom-Ingenieur degree from the Technical University Clausthal-Zellerfeld in Germany, the Dr.-Ing. from the Rheinisch-Westfälische Technische Hochschule Aachen and the Habilitation from the Universität Bremen. He has held a post-doctoral position at the Waseda University, Tokio, Japan and in 1999 was appointed chair of thermal process engineering at the Martin-Luther-Universität Halle-Wittenberg in Halle (Saale). He is currently chair of the European Federation of Chemical Engineering. His interests include all aspects of industrial crystallization, especially melt crystallization and batch crystallization. Matthew Jonathan Jones received the Diplom-Chemiker degree from the Westfälische Wilhelms-Universität Münster in Germany and the PhD from the University of London (University College), UK. After post-doctoral positions at the University of St. Andrews and Heriot-Watt University, Edinburgh and a teaching fellowship at the University of Leeds, he took up the position of senior scientist in the thermal process engineering group at the Martin-Luther-Universität Halle-Wittenberg, Halle (Saale), Germany in 2003. His research interests include the industrial crystallization of proteins and the application of molecular modeling to crystallization and materials in the solid state. To cite this chapter J. Ulrich and M.J. Jones, (2006), HEAT AND MASS TRANSFER OPERATIONS - CRYSTALLIZATION, in Chemical Engineering, [Eds. John Bridgwater,Martin Molzahn,Ryszard Pohorecki], in Encyclopedia of Life Support Systems (EOLSS), Developed under the Auspices of the UNESCO, Eolss Publishers, Oxford ,UK, [http://www.eolss.net]


Date post:	07-Feb-2018
Category:	Documents
Upload:	duongtruc
View:	215 times
Download:	0 times

Heat and Mass Transfer Operations-Crystallization Chapters/C06/E6-34-03-02.pdf · CHEMICAL...

Documents