Pharmaceutical Crystallization - Crystal Growth & Design (ACS

Pharmaceutical Crystallization - Crystal Growth & Design (ACS...

0 downloads 77 Views 2MB Size


Pharmaceutical Crystallization Published as part of the Crystal Growth & Design 10th Anniversary Perspective Jie Chen,† Bipul Sarma,†,‡ James M. B. Evans,†,‡ and Allan S. Myerson*,†,‡ †

Department of Chemical Engineering, Massachusetts Institute of Technology, 77 Massachusetts Avenue 66-568, Cambridge, Massachusetts 02139, United States ‡ Novartis-MIT Center for Continuous Manufacturing, 77 Massachusetts Avenue, Building 66, Cambridge, Massachusetts 02139, United States ABSTRACT: Crystallization is crucial in the pharmaceutical industry as a separation process for intermediates and as the final step in the manufacture of active pharmaceutical ingredients (APIs). In this perspective article to celebrate 10 years of Crystal Growth & Design, we focus on three areas related to crystallization in the pharmaceutical industry: (1) advances in our understanding of the fundamentals of nucleation, (2) production and scale-up of novel solid forms, and (3) continuous processing. While the areas discussed are not new, they are areas, in our opinion, of significant current interest to the community engaged in crystallization in the pharmaceutical industry.

’ INTRODUCTION Crystallization is an important separation and purification process employed to produce a wide variety of materials in the fine chemical, food, and pharmaceutical industries. The control of crystal size, shape, and crystal form is crucial as they can influence downstream operations such as filtration, drying, and milling as well as influence the physical and chemical properties of the solid such as dissolution rate and solubility. Crystallization is important in the pharmaceutical industry as a separation process for intermediates and often serves as the final step in the manufacture of active pharmaceutical ingredients (APIs). In this perspective article to celebrate 10 years of Crystal Growth & Design, we focus on three areas related to crystallization in the pharmaceutical industry: (1) advances in our understanding of the fundamentals of nucleation, (2) production and scale-up of novel solid forms, and (3) continuous processing. While the areas discussed are not new, they are areas, in our opinion, of significant current interest to the community engaged in crystallization in the pharmaceutical industry.

nucleus ranges from less than a second to days, which makes it challenging for both experimental and computational techniques. A. Classical Understanding of the Nucleation Process. Classical nucleation theory (CNT), a simple and widely used theory, says that fluctuations give rise to the appearance of a small nucleus of a second phase. Although this second phase has a lower free energy than the initial phase, there is a free energy penalty associated with the creation of an interface. For a spherical nucleus, the free energy, ΔG, of transforming to the second phase is the sum of a negative volume term and a positive surface term, defined as ΔG ¼ -


where r is the radius of the nucleus, ΔGV is the bulk free energy difference per unit volume between the first and second phases, and γ is the surface free energy of the second phase per unit area. The positive surface free energy term dominates at small radii, which causes an increase in the total free energy change initially. As the cluster size increases, total free energy goes through a maximum at a critical size (rc), above which the total free energy decreases as the nucleus grows bigger and the growth becomes energetically favorable. The concept of classical nucleation theory is schematically shown in Figure 1. Given the known and highly accurate input parameters of CNT, the orders of magnitude difference between the predicted and measured nucleation rates for single component fluids, like the condensation of water, indicates the inadequacy of CNT.1 Several

’ ADVANCES IN UNDERSTANDING THE FUNDAMENTALS OF NUCLEATION Nucleation is the commencement of a new phase. It plays a decisive role in determining the crystal form obtained and the crystal size distribution (CSD). Thus, understanding the fundamentals of nucleation is crucial to the control of crystallization processes. The essential difficulty of studying nucleation and developing an accurate description of the process results from the fact that the size of the critical nucleus typically falls in the range of 10-1000 molecules and the time scale of forming a critical r 2011 American Chemical Society

4 3 πr ΔGV þ 4πr 2 γ 3

Received: November 23, 2010 Revised: January 31, 2011 Published: February 22, 2011 887 | Cryst. Growth Des. 2011, 11, 887–895

Crystal Growth & Design


Figure 2. The classical and the two-step nucleation models.

intense nanosecond pulses of near-infrared laser light. This result is difficult to explain without relying on the two-step nucleation mechanism. If the nuclei would form by successive aggregation of molecules in an ordered manner as proposed by CNT, induced alignment of the molecules by laser would not cause a significant change in the structure of already ordered clusters and hence lead to the crystallization of different polymorphs. Additional support for the two-step nucleation mechanism has been provided by computer simulations. One of the first computational studies supporting this theory was reported by ten Wolde and Frenkel.9 They studied the homogeneous nucleation of a Lennard-Jones system using a Monte Carlo method and confirmed that, around the critical point, highly disordered liquid-like droplets formed first and then were followed by the restructuring of molecules inside the droplets to form crystalline nucleus beyond a certain critical size. With the development of more advanced and efficient sampling methods for studying rare events like nucleation, Bolhuis and co-workers were able to determine the size and structure of the critical nucleus in a Lennard-Jones system using the transition path sampling method.10 Their results showed that critical nuclei can be either bigger/less ordered or smaller/more ordered clusters. The existence of those bigger/less ordered critical nuclei again confirmed that density fluctuation and structure fluctuation do not necessarily need to happen simultaneously. C. Future Directions. A number studies published in the past decade support the two-step nucleation theory. However, mechanistic understanding of the nucleation pathway, especially the second step, is still limited. Very little information about how molecules restructure in the dense liquid-like droplets is available. The organization step was proposed as the rate limiting step, which is consistent with the observation that the nucleation from solution takes a longer time as the complexity of molecules increases since it would be more difficult for more complex molecules to arrange themselves in the appropriate lattice structures due to their high degree of conformational flexibility. Future studies are needed to further validate this idea. Molecular simulations offer a promising platform to gather microscopic information of the restructuring process of molecules. However, simple order parameters, such as the number of molecules and the bond order parameter Q6 to characterize the size and structure of aggregates employed by Bolhuis et al. in their study of LJ systems, are not applicable in studying the nucleation of molecular crystals from solution. More sophisticated order parameters to differentiate the solution and the crystal are needed, as well as an efficient sampling method to accelerate the evolution from a supersaturated solution to the appearance of critical nuclei and to the formation of crystals. Progress in developing these order parameters for studying the nucleation from solution has been reported by Santiso and co-workers, and

Figure 1. Free energy diagram for nucleation.

fundamental limitations exist in CNT. The first is that the critical nucleus is treated as a spherical droplet and the surface is modeled as an infinite plane with the curvature dependence of the surface tension neglected. These assumptions seemingly do not hold if the critical nucleus is on the order of a few nanometers. CNT also assumes the growth of clusters takes place by addition of one monomer at a time and the molecules in the nucleus are in an ordered array with the same structure as the resulting crystal. Consequently, it uses only the size criteria to decide whether aggregates become critical or not. This becomes insufficient for crystallization from solution, where at least two order parameters, density and periodic structure, are necessary to distinguish between old and new phases. Moreover, the clusters/aggregates on the pathway can be organized in a different manner from the resulting crystal. The fluctuations in density and structure do not need to happen simultaneously; one can dominate and serve as a prerequisite for the other one. In the past decade, a line of simulations, theories, and experimental studies, including those reported by our group, suggested an alternative mechanism of crystal nucleation where the structure fluctuation follows and is superimposed by density fluctuation, called the two-step nucleation mechanism (as shown in Figure 2), which will be discussed in detail. B. Two-Step Nucleation. Two-step nucleation theory was initially proposed for protein crystallization with some of the most direct evidence provided by Vekilov and co-workers.2 They directly monitored the liquid-liquid separation and nucleation of hemoglobin solutions using a microscope with differential interference contrast (DIC) imaging and observed the existence of a dense liquid phase of high hemoglobin concentration where the nucleation of deoxy-hemoglobin S polymers occurred. Other experimental studies, such as dynamic and static light scattering,3,4 differential scanning calorimetry,5 and small-angle X-ray scattering,6,7 have also suggested that this mechanism holds in the nucleation of proteins,3-5 colloidal particles,6 and even small organic molecules.7 These studies all demonstrate that solute molecules rapidly aggregate in the diffusion-limited aggregation regime to form mass fractal clusters in the initial stages of nucleation and then progressively restructure into compact structures at the later aggregation stage, thus supporting a two-step nucleation mechanism where a structure fluctuation occurs within a region of higher density of molecules existing for a limited time due to a density fluctuation. The nonphotochemical laser-induced nucleation (NPLIN) studies of glycine aqueous solutions performed by Garetz and co-workers8 showed that different polymorphs (γ and R) of glycine could be obtained by switching between linear and circular polarization in the irradiation of supersaturated glycine aqueous solutions with 888 |Cryst. Growth Des. 2011, 11, 887–895

Crystal Growth & Design


This phenomenon has great importance in the pharmaceutical industry as different polymorphs can exhibit different physical and chemical properties such as solubility and dissolution rate, which in turn affects the bioavailability and stability of the drug substance. As different polymorphs have different physical and chemical properties, polymorph screening which attempts to find and characterize as many unique forms as possible is normally employed. In addition, intellectual property (IP) opportunities have led to significant efforts by both innovator and generic companies to find and patent unique crystal forms. The stability relationship of polymorphs of a molecule can be established by measuring their enantiotropic or monotropic relationship.13-16 Polymorphs are defined as enantiotropic when the transition point between the two phases is found at a temperature below the melting point of either of them, as shown in Figure 4a. When there is no transition point below the melting points of the two polymorphs, the two forms are monotropically related, as shown in Figure 4b. The heat-of-fusion rule states that in an enantiotropic system higher melting polymorph will have the lower heat of fusion. Chemical engineers are concerned with industrial implementation of processes developed in the laboratory. But all processes are scale dependent, for which reason the materials including APIs behave differently in small scale (laboratory scale) and large scale (plant and production). This is why scale-up of an API from milligram quantities in a laboratory to large scale (kilogram or ton scale) in a plant without changing its optimized properties and reproducibility and at low cost is a major challenge in the pharmaceutical industry. Scale-up in the pharmaceutical industry employs data taken in the laboratory to develop processes which will produce products of consistent quality (purity, size distribution, shape, and crystal form) as the demand for material increases during clinical trials and finally at full scale commercial production. This is a particularly demanding process in the pharmaceutical industry since specialized equipment is rarely employed and processes are “fit” into exisiting equipment. The scale-up of an API whether it is a stable or metastable form (single component or multicomponent systems) is one of the vital concerns that can be determined and needs to consider the following factors. (a) Time: Scale-up operations normally require more time than laboratory scale experiments because of larger volume of materials. This often means that crystallized material spends more time in a slurry prior to solid-liquid separation, thus increasing the probability of metstable forms transforming to more stable forms through solution-mediated transformation driven by solubility differences. (b) Reactant addition: The magnitude of heat released or absorbed during addition of reactants must be monitored and determined in the laboratory scale so that can be controlled in process. In addition, the thermal stability of raw materials and intermediates should be determined using thermogravimetric analysis (TGA) and differential scanning calorimetry (DSC) to detect endothermic or exothermic behavior. (c) Mixing: Mixing of solutions and slurries are a crucial part of scale-up operations. Mixing issues for cooling crystallizers generally involve issues relating to keeping the solids suspended and minimizing secondary nucleation, crystal breakage, and growth on surfaces and while important are less likely to cause problems during scale-up. Mixing issues in antisolvent or reactive crystallization are much more complex and can cause significant problems in process scale-up. Antisolvent (or reactant) addition rate and addition location are two crucial parameters which must be evaluated along with impeller type and agitation rate. Recent

Figure 3. Hypothetical transitions from solution to thermodynamic and kinetic crystals. Difference between ΔG‡thermodynamic and ΔG‡kinetic determines the ease of formation of kinetic crystals.

these recently developed order parameters are being tested on molecular crystals.11 The development of robust crystallization processes in which the crystal size distribution (CSD) and polymorphic outcome can be controlled and/or predicted requires a clear mechanisitic understanding of nucleation. While we are still far from this goal, significant progress has been made in the last 10 years and the number of research groups working on this problem has increased. Hopefully by the 20th anniversary of CGD in 2020 we will be able to report significant further progress in this area.

’ PRODUCTION AND SCALE-UP OF NOVEL SOLID FORMS Pharmaceutical research involving an active pharmaceutical ingredient (API) often deals with various solid forms. These forms include polymorphs, host-guest complexes, network solids, salts, solvates (including hydrates), and co-crystals. In an effort to formulate poorly soluble compounds, solid forms such as co-crystals and metastable polymorphs are being developed and processes for their production are being developed and scaled up as well, which makes this a topic of current interest. Crystals are ordered three-dimensional structures with the structure dictated by forces acting at the molecular level. The crystallization process consists of two steps, nucleation and crystal growth. The nuclei are stable only when they reach a critical size and the critical size is dictated by parameters such as supersaturation and temperature. Since different solid forms have different solubilities, the number of potential solid forms which are thermodynamically accessible depends on the level of supersaturation. According to the Ostwald rule of stages,12 the structure that crystallizes first is the one which has the lowest energy barrier (highest energy, kinetically metastable). A hypothetical free energy landscape of solution, thermodynamic crystal and kinetic crystal is shown in Figure 3. A. Metastable Polymorphs. Polymorphism is often defined as the ability of a compound to exist as two or more crystalline phases. Polymorphism can occur because of the differences in the packing of molecules in the crystal lattice, conformational flexibilities, and supramolecular synthon competitions and are called packing, conformational, and synthon polymorphism, respectively.13,14 889 |Cryst. Growth Des. 2011, 11, 887–895

Crystal Growth & Design


Figure 4. The stability relationship of polymorphs of a compound established by measuring their enantiotropic or monotropic relationship, known as heat-of-transition rule. (a) Fundamental energy vs temperature (E/T) diagram for dimorphic enantiotropic system shows that form I is stable below transition point. Above the transition point form II is stable. (b) Fundamental E/T diagram for dimorphic monotropic system. Form I is more stable than form II at all temperatures below the melting point.

work has employed computational fluid dynamics (CFD) to examine mixing during antisolvent crystallization.17-19 (d) Stability: A complete profile of the stability data of the metastable forms of the critical raw materials and intermediates should be generated in the laboratory under conditions similar to those in the plant. On the basis of these data, packaging and storage conditions for critical raw materials and intermediates can be established. For example, Rotigotine (Neupro)20 developed by Aderis pharmaceuticals and marketed by Schwartz underwent quality and storage issues from Committee for Medicinal Products for Human Use due to the unexpected appearance of snowflake-like crystals of a second polymorph during storage. (e) Filtration: the compatibility of a filter cloth with a process fluid depends on parameters such as time flow, temperature, pressure, weight, pore size, and shape. (f) Centrifugation. (g) Drying. (h) Maintenance of temperature, pH, water content, and other related parameters. Apart from these important factors, the product economics based on projected market size and allowable manufacturing cost must be defined. While the norm within the pharmaceutical industry is to develop the most stable polymorphic form, there are occasional situations in which the development of a metastable form is attempted. The justifications for developing a metastable form normally involve the need for a faster dissolution rate/higher concentration to achieve rapid absorption and efficacy or a higher solubility to achieve acceptable systemic exposure for a low-solubility drug. In addition, the manufacture of metastable forms of APIs is often attempted by generic manufacturers who wish to come to market after composition of matter patents have expired but before separate patents on crystal forms have expired. The risks associated with development of a metastable form must be mitigated by laboratory work and work related to scale-up. This work should include understanding of the rates of solution-mediated transformation among crystalline forms as a function of time and temperature. In addition, laboratory work should also determine whether form changes in the drug product will have an effect on product quality or bioavailability. Analytical methodology and sampling procedures should be in place which ensure that a problem will be detected before dosage forms which have compromised quality or bioavailability can reach patients.21 Because of better separation of impurities in the

Scheme 1. Optimized Solid-State Drug Development Procedure for the Scale-up of a Desired Polymorph

crystallization step, faster dissolution rate, ease for semisolid formulation, and shorter processing time, the metastable form I of Ritonavir was considered for the formulation.22 Initially, the metastable form B which was eventually switched to another metastable form A of Abecarnil23 was chosen in early drug development because of its ease of crystallization. Abecarnil formulations with the thermodynamically stable form C were abandoned because of the large difference in the physicochemical properties as compared with the metasble form. Process development for both stable and metastable forms should generally follow the sequence shown in Scheme 1. Solid form selection is crucial in determining the type of conditions of crystallization to be used. In addition, the role of impurities on polymorph formation/transformation should be examined closely. Small levels of impurities can inhibit the formation of a particular form and thus the transformation of a less stable to more stable form.24 The selective nucleation25,26 of a desired polymorph can be achieved by control of supersaturation and nucleation temperature, choice of solvent, or use of additives. Advanced methods such the use of external fields or surface templating can also be 890 |Cryst. Growth Des. 2011, 11, 887–895

Crystal Growth & Design


considered but are more useful in polymorph screening. A rapid quenching of a supersaturated solution by addition of an antisolvent might be an indication for the robust manufacture of the metastable polymorph. Even though there is a high risk of phase transformation, desired metastable polymorphs can be generated by seeding experiments with great care. It is important to be sure that there are no detectable nuclei of the stable polymorph during metastable polymorph crystallization. The phase transformation of a mixture of polymorphs and solvates into the stable polymorph can be monitored by the use of in situ Raman spectroscopy. The use of tailor-made additives is possible, for example, to support the emergence of a metastable polymorph which sometimes inhibits the nucleation of the stable polymorph; however, this is difficult in the pharmaceutical industry. In terms of development of robust processes for isolating desired polymorphic materials, although limited, a structural approach has been demonstrated.27-30 The tailor-made materials generally suggest pharmacologically and toxicologically inactive components of pharmaceuticals. It is assumed that tailor-made additives act by the same mode of action and have the same molecular target and thus might exert effects in an additive manner. However, the potential development of impurity by these additives in the drug development process is practically achievable. The possibility of designing additive molecules to selectively inhibit the crystallization of the more stable polymorph on the basis of conformational recognition, allowing kinetics to dominate the crystallization process and leading to the stabilization of a metastable phase, was reported. The solid liquid separation, solubility, compaction, particle flow, and formulation characteristics will all be polymorph dependent. Thus, polymorphism can be exploited to a certain extent such that the structure with desired properties appropriate for a particular formulation is produced. However, the isolation of a new polymorph can threaten product specifications and radically change the status quo in the patent arena. Zantac and Norvir are two classic examples. The focus on polymorphism in recent years has led to a significant improvement in screening methods, laboratory automation, and analytical methods. Polymorph predictions methods31 continue to improve, however, are not yet at the point of being of general use to the industry. Improvements in PAT methods have improved the ability to develop and control processes in which metastable polymorphs are produced. However, the challenge still remains to understand and control the nucleation of desired forms and form conversion through the use solvents and additives. B. Pharmaceutical Co-Crystals. Drugs with multicomponent crystalline phases such as salts, solvates, and hydrates also carry desired drug properties similar to single-component polymorphs. A thorough screening of 245 compounds by Aptuit solid-state chemistry division reported that overall 50% compounds are polymorphic.32 Another screening showed formation of about 60% co-crystals, 34% hydrates, and 33% solvates. In general, if a compound is insoluble, hygroscopic, or difficult to crystallize, a search for salts or co-crystals will be conducted. The idea of using co-crystals as APIs is relatively new, and thus we are not aware of any approved drug which is a co-crystal; however, most pharmaceutical companies are quite active in this area, and it is likely that formulations which employ co-crystals are being developed. This is reflected by the growing number of patent applications for co-crystals in recent years.

Co-crystals can be defined as multiple-component crystal systems that coexist through hydrogen bonds or noncovalent interactions. The reactants are solids at ambient conditions. However, pharmaceutical co-crystals are the crystalline molecular complexes of an API with another pharmaceutically acceptable molecule or Generally Regarded As Safe (GRAS) chemicals. Food additives, preservatives, excipients, vitamins, minerals, amino acids, biomolecules, and other APIs can be selected as co-crystal formers (CCF). Salts and co-crystals are multicomponent crystals and a continuum exists linking cocrystals and salts based on the extent of proton transfer between the components. Despite their pharmaceutical applications, cocrystals are also useful in designing extended supramolecular architectures, preparing NLO materials, solid-state photodimerization reactions, and enantio separation of racemic compounds. Generally amorphous solids have useful properties such as higher solubility and higher dissolution rate because they have higher free energy and sometimes better compression characteristics than corresponding crystals. Unfortunately, amorphous solids are often physically and/or chemically less stable than corresponding crystals. Salts make APIs more bioavailable, but more than 30% of pharmaceutical compounds lack suitable functionality for salt formation. Pharmaceutical co-crystals are an alternate and interesting pathway to improve the physiochemical profile of an API. From a thermodynamic point of view, pharmaceutical co-crystals are stable and high energy forms. Therefore, they can have an impact on properties such as solubility, dissolution rate, stability, and hygroscopicity. The strategy to improve bioavailability of poorly soluble drugs involves investigation drug conformer combinations that have the potential of forming energetically and structurally robust interactions. The literature reveals that there have been relatively few pharmaceutical co-crystal case studies with pharmacokinetic details.33-35 Selection of proper coformers enables and improves product or impurity removal by co-crystallization. The principle is based on the coformer pure component and product solubility. Solubility is a thermodynamic property while dissolution is a kinetic process. Drug bioavailability depends on adsorption that can be assessed by Fick’s First law, J = PC, where the flux (J) of a drug through the gastrointestinal wall depends on the permeability coefficient (P) of the gastrointestinal barrier for the drug and the drug concentration (C). Generation and maintenance of the metastable supersaturated state is a strategy to improve intestinal absorption of poorly watersoluble drugs. Two essential steps need to be considered and they are termed the “spring and parachute” approach (as shown in Figure 5). A thermodynamically unstable, supersaturated solution of a drug can only be generated starting from a high energy form of a drug which is known as the “spring”. An example might be an amorphous API which is much more soluble than the crystalline material. A combination of excipients such as co-solvents, lipids, or polymer-based formulations can deliver the drug in solution as high energy solid forms that can easily provide an accelerated dissolution or a higher apparent solubility and is known as the “parachute effect”. The apparent solubility is the apparent equilibrium between drug in solution and a solid whose structure is in the high energy state. A high energy form of the drug (the spring) provides the driving force to solubilize the drug at a concentration greater than its equilibrium solubility level and a similar effect resulted by the combination of excipients (the parachute) by inhibiting or retarding precipitation. 891 |Cryst. Growth Des. 2011, 11, 887–895

Crystal Growth & Design


Figure 5. “Spring and parachute” approach to promote and maintain supersaturation of poorly soluble drugs.

Co-crystals can exist in several forms as (i) neutral with a 1:1 ratio or varying stoichiometries (AB, AB2, A2B, etc.), (ii) cocrystal polymorphs, (iii) co-crystal solvates or hydrates, and (iv) neutral and ionic components (AB 3 HCl) together. To design co-crystals, the first step is to follow the hydrogen bond rules stated by Etter.36,37 Common hydrogen bond synthons are acid 3 3 3 acid, acid 3 3 3 amide, amide 3 3 3 amide, acid 3 3 3 pyridine, urea 3 3 3 ketone, halogen 3 3 3 halogen, etc. Itraconazole, caffeine, and theophylene are a few examples of drugs considered to synthesize pharmaceutical co-crystals based on acid 3 3 3 pyridine synthons. It is a fact that bioavailability is intimately related to whether the drug is present in neutral or ionized form. Similar to co-crystals, salts are also a multicomponent system that can have different physical and chemical properties. In salts, protons are transferred from acid to base and remain in the ionic state, however are neutral in co-crystals. The transfer from neutral to ionic hydrogen bond can be as a continuum of the intermediate X 3 3 3 H 3 3 3 Y bond that depends on the ΔpKa value (pKa of base - pKa of acid) in solution and the crystalline environment that determines the extent of proton transfer. For an example, in the carboxylic acid and pyridine system, the carbonyl stretch frequencies confirm whether the complex is co-crystal or salt. To predict the occurrence of a neural or ionic hydrogen bond between acid and base, the ΔpKa rule or “Rule of 3” has been postulated.38 The most common co-crystal screening methods39-45 are reaction crystallization and slurry crystallization. Melt crystallization, neat and solvent drop grinding, supercritical fluid crystallization (SCF), and twin screw extrusion have also been used in the preparation of co-crystals.46,47 The most effective design terminology is the phase solubility diagrams (PSD, as shown in Figure 6)or ternary phase diagrams (TPD) of a co-crystal system. PSD or TPD will ensure the eutectic points or transition concentrations, where two solid phases co-exist in equilibrium with a liquid phase and define thermodynamically stable regions of the co-crystal in relation to its pure components.48 Therefore, PSD is the best informative tool for co-crystal synthesis in solution-based methods as well to identify whether the co-crystal will be congruently or incongruently saturated in a particular solvent at a particular temperature. A congruently saturating co-crystal is thermodynamically stable during slurrying and can be readily formed by slurrying a stoichiometric ratio of co-crystal components and vice versa.

Figure 6. A typical phase solubility diagram (PSD) of a co-crystal system (C). A & B are starting components; S represents solvent.

The scale-up strategy for a pharmaceutical co-crystal includes three crucial steps: (i) proper selection of a solvent system, (ii) construction of phase diagram to determine the thermodynamic stable regions and solid liquid equilibrium of the system, and (iii) kinetics of the system. To achieve a high throughput and a pure co-crystal phase, the solvent must be selected so that the coformers have a higher solubility than the API and the critical concentration of the coformer should be different from the solubility of the coformer at the operating temperature. The phase diagram identifies saturated liquid curves and the stable solid regions. Kinetics will determine nucleation and crystal growth and thus will aid in the development of seeding strategies and other parameters that may affect the co-crystal formation process. A typical scale-up of a process to make co-crystals was recently demonstrated by Sheikh and co-workers.49 The scale-up of carbamazepine-nicotinamide and caffeine-glutaric acid co-crystals was developed based using phase solubility behavior and slurry-based crystallization methods. The appearance of nonstoichiometric combinations and pure co-crystalline materials presents additional challenges during scale-up. The modification of chemical and physical properties of APIs can lead to extended patent coverage and patent life extension of products.50 Recent articles emphasize the development and importance of pharmaceutical co-crystals.51-59 Novel co-crystalline compositions of drug Tenofovir disoproxil and fumaric acid in a 2:1 molar ratio were for antiviral applications. The carbamazepine: nicotinamide and saccharin co-crystals were studied in terms of scale-up, crystal polymorphism, physical stability like dissolution and oral bioavailability and compared with the marketed form of carbamazepine (Tegretol). Co-crystals have become a standard tool in pharmaceutical solids form development along with polymorphs, pseudopolymorphs, and salts. As co-crystals enter the development phase, more emphasis needs to be directed in understanding and characterizing co-crystals phase diagrams and the development of robust procedures to scale-up processes for their manufacture.

’ CONTINUOUS PROCESSING Crystallization is a key separation and purification unit, which influences the performance of downstream process operations 892 |Cryst. Growth Des. 2011, 11, 887–895

Crystal Growth & Design


Table 1. Comparison of Batch and Continuous Processes variable equipment

Continuous processing offers orders of magnitude reduction

footprint and in equipment footprint giving rise to þ20% reduction in the capital


expenditure process

Batch processes demonstrate significant batch-to-batch


variability with regard to physical properties compared to

Figure 7. An example of a two-stage antisolvent/cooling crystallization MSMPR cascade.

continuous processes. yield

In a once through system, a batch process as a rule of thumb has a higher yield; however, with the appropriate recycling strategy, it is feasible to achieve higher continuous yields.


Process for tracing material in batch process is well understood


both from an operational perspective as well as from a regulatory perspective, whereas for a continuous process there is still a gap between operational understanding and regulatory acceptance of this.

Figure 8. Schematic process flow diagram of the continuous crystallization system with multistage antisolvent addition (taken from Myerson and Alvarez, ref 19. Copyright 2010 American Chemical Society).

such as filtration, drying, and milling. Physical-chemical properties of the final product also depend on crystal characteristics. In the pharmaceutical industry, more than 90% of all APIs are crystals of small organic molecules. At the moment, the vast majority of industrial scale pharmaceutical crystallization processes are done in batches either as a cooling, antisolvent, or reactive crystallization. Although batch is the most common method and the methodologies for developing these types of crystallization processes are reasonably well understood, there are still significant issues with batch-to-batch variability which can lead to substantial issues in the downstream processing of the isolated material. Continuous processing while more difficult to develop affords a number of significant advantages over the batch processing. Continuous processes by their very nature require substantially smaller process equipment; therefore the cost of this equipment is substantially lower. For example, if we assume that we need to make 16T per annum of API with a solvent to API ratio of 20:1 and a 4 h residence time, we could support this with a 250 L reactor, whereas a comparable batch system would typically require a 5000 L reactor. On the basis of the reduced footprint and equipment size, a substantial reduction in both capital and operating expenses for a crystallization process should be achieved and savings in the region of 20% are expected.60 Continuous processing also offers enhanced reproducibility/ control of physical characteristics of the crystalline material. Once steady state has been achieved, all material crystallizes under uniform conditions, which leads to greater reproducibility and control/tuning of key characteristics such as the crystal size distribution (CSD) and polymorphic form. The benefits of reproducably producing consistent product will enable the eliminatation of a number of downstream corrective actions, for example, granulation (wet or dry) which is used to simply provide a “uniform” particle for the creation of solid dosage forms. In addition to providing more consistent and better quality powder mixtures, the ability to control/tune key physical properties (e.g., the CSD) is an attractive benefit of continuous processing. In the past decade or so, a large proportion of molecules in development tend to have poor bioavailability. One of the routinely used solutions for “accepted” batch is to

isolate the material and mill/micronize the product to produce material with a sufficiently small particle size so that the solubility and hence bioavailability is increased. While this process clearly works, it is both inefficient and risky as it adds additional unit operations and hence costs to the cost of goods sold (COGS). Moreover, there is a risk of inducing solid state transformations and chemical degradations in the product. Continuous crystallization on the other hand has the potential to consistently produce small particles with a sharp particle size distribution, as depending on reactor configurations. It is fairly trivial to constrain the growth of the particles and thus eliminate these inefficiencies and risks that are routinely faced. While in the very simplified case where we consider a once through system, the batch process can achieve a higher yield since batch processes can go to an equilibrium state as opposed to continuous processes which operate at a steady state. However, with the development of an appropriate recycle, it is possible to drive a continuous process yield at least equal to that of a batch system. Detailed comparisons between the batch and continuous processes are listed in Table 1. A. Continuous Crystallizers. There are a wide variety of continuous crystallizer designs used in the chemical industry.61 For pharmaceutical applications, two main categories of applicable crystallizer designs are mixed suspension mixed product removal (MSMPR) in a single or multiple stages (Figure 7) and plug flow reactors (Figure 8). The comparisons between these two types of crystallizers are listed in Table 2. Using either of these two basic approaches, a skilled practitioner can develop the desired continuous crystallization process including advanced strategies such as recycle and fines destruction.62 The choice of whether to use a MSMPR or a PFR system is primarily driven by the kinetics of the process with MSMPR generally being preferred for low conversions and long residence times and the PFR being preffered for higher conversions with short residence times. A secondary factor favoring the utilization of the MSMPR approach is the fact that it is relatively simple to convert existing batch capacity to continuous capacity. B. Challenges and Future Directions. While there are numerous examples of continuous crystallization in the chemical 893 |Cryst. Growth Des. 2011, 11, 887–895

Crystal Growth & Design


Table 2. Comparisons between MSMPR and PFR advantages MSMPR


temperatures are easy to control

typically less efficient

lower maintenance costs simple to replace parts of the system easier maintenance PFR

PFR typically has a higher efficiency than a MSMPR of the same volume

temperatures are harder to control

run for long periods of time without maintenance

maintenance is more expensive more complicated maintenance settling of solids/plugging

industry which have demonstrated the benefits of continuous crystallization in terms of economics and quality, the application of continuous crystallization within the pharmaceutical manufacturing industry has been slow. Significant efforts, especially in the pharmaceutical industry, to introduce new technologies, including their regulatory impact, are required before continuous crystallization will be truly embedded within the pharmaceutical community. Small Product Volumes. Crystallization in the pharmaceutical industry with a few exceptions tends to be relatively low volume