- Email: [email protected]
Crystallization of a polymorphic hydrate system
Crystallization of a Polymorphic Hydrate System F. TIAN,1 H. QU,2 M. LOUHI-KULTANEN,3 J. RANTANEN1 1
Faculty of Pharmaceutical Sciences, Department of Pharmaceutics and Analytical Chemistry, University of Copenhagen, Copenhagen, Denmark 2
Institute of Chemical Engineering, Biotechnology and Environmental Technology, University of Southern Denmark, Odense, Denmark 3
Department of Chemical Technology, Lappeenranta University of Technology, Lappeenranta, Finland
Received 18 March 2009; revised 11 May 2009; accepted 28 May 2009 Published online 30 June 2009 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.21865
ABSTRACT: Nitrofurantoin can form two monohydrates, which have the same chemical composition and molar ratio of water, but differ in the crystal arrangements. The two monohydrates (hydrates I and II) could be produced independently via evaporative crystallization, where supersaturation and solvent composition were both found to have an effect. Hydrate I showed much slower crystallization than hydrate II. During cooling crystallization, the nucleation and growth of hydrate II was again dominant, consuming all supersaturation and leading to no hydrate I formation. Seeding of hydrate I during cooling crystallization was also applied, but the hydrate I seeds were not able to initiate its nucleation rather than dissolving into crystallizing solution. Although solubility tests revealed that hydrate II is more stable than hydrate I due to its lower solubility (110 4 and 131 12 mg/mL for hydrates II and I, respectively), this difference is rather small. Therefore, the small free energy difference between the two hydrates, together with the slow crystallization of hydrate I, both lead to a hindrance of hydrate I formation. Furthermore, the crystal structure of hydrate II demonstrated a higher H-bonding extent than hydrate I, suggesting its more favorable crystallization. This is in good agreement with experimental results. ß 2009 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 99:753–763, 2010
Keywords: hydrate; crystallization; supersaturation; polymorphism; crystal structure; nitrofurantoin
INTRODUCTION Understanding the crystallization mechanism of organic molecules is of great scientific importance.1,2 Crystal structure and particle habit determine critical material properties.3–5 The occurrence of polymorphs during crystallization from solution quite often follows the Ostwald’s rule of stages, where the least stable form
Correspondence to: J. Rantanen (Telephone: 45-35-33-6585; Fax: 45-35-33-60-30; E-mail: [email protected]) Journal of Pharmaceutical Sciences, Vol. 99, 753–763 (2010) ß 2009 Wiley-Liss, Inc. and the American Pharmacists Association
crystallizes out first and then transforms to the more stable form.6 However, concomitant polymorphism is also frequently observed in the crystallization of organic compounds, where more than one polymorphic form will crystallize at the same time. The concomitant polymorphism can be caused by competing nucleation of the modifications,7,8 solvent-mediated transformation from one form to another,9 or cross-nucleation of another polymorph nucleating on the initially present one.10–14 Compared with the systems following Ostwald’s rule of stages, concomitant polymorphism represents a much more complicated issue and makes the polymorphism control extremely difficult.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
753
754
TIAN ET AL.
Hydrates with the same molar ratio of water can also crystallize into different crystalline packings and thus exhibit polymorphism. Examples of such systems including fluprednisolone monohydrate,15 succinyl sulfathiazole monohydrate,16 nedocromil sodium monohydrate,17 niclosamide monohydrate,18 and nitrofurantoin (NF) monohydrate.19 The relative thermodynamic stability of these hydrate polymorphs has not been thoroughly investigated, and the understanding of the crystallization behavior of polymorphs of hydrates is very limited. To understand and control crystallization of a polymorphic hydrate system, we used two polymorphs of NF monohydrate as a model compound. NF monohydrate, 1-[[(5-Nitro-2-furanyl)-methylene]amino]2,4-imidazolidinedione monohydrate, is known to crystallize into two crystalline structures: P21, ˚ , b ¼ 6.476 A ˚ , c ¼ 16.969 A ˚ , b ¼ 96.688 a ¼ 9.783 A ˚ ˚ ˚. and Pbca, a ¼ 12.642 A, b ¼ 9.857 A, c ¼ 17.383 A The monoclinic P21 structure is called hydrate I, and the orthorhombic Pbca is named hydrate II.19 NF hydrate has been intensely investigated.19–27 However, according to the authors’ knowledge, all of these publications have studied hydrate II, while only a few have characterized hydrate I. Pienaar and coworkers were the first who successfully prepared the hydrate I by slowly evaporating a saturated NF solution at 408C. The authors reported that NF hydrate I was observed within 7 days whereas they found NF hydrate II needles within 3 days.19,25 This suggests a slower crystallization rate for hydrate I relative to hydrate II. Alternatively, Otsuka and Matsuda26 reported that hydrate I could be produced via grinding hydrate II at humidity conditions for over 7 h. The authors did not offer any explanations as to why hydrate I could be converted through grinding of hydrate II. Because grinding is often used to induce high-energy intermediate forms, for instance the amorphous form, one explanation is that hydrate I is an unstable form in comparison to hydrate II. Recently, Kelly also performed the crystallization of two NF monohydrates. Pure NF hydrate I, however, could not be achieved since it always crystallized concomitantly with hydrate II when suspending anhydrous NF b-polymorph in supersaturated solutions.27 Although the relative stability of two NF hydrates was not clearly discussed in any of these publications described above, the relatively large difference in the number of publications between these two NF hydrates implies that NF JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
hydrate I is not so often encountered as NF hydrate II when crystallizing NF hydrate. The main aim of this research is therefore to explore the crystallization mechanism of two different polymorphs of NF monohydrate, in order to answer the questions of whether their crystallization follows the classic polymorphism theories and what roles typical crystallization parameters, that is, supersaturation and water activity, play in their crystallization? With the growing importance of hydrate in various pharmaceutical and chemistry fields, a particular purpose of this work is also to show the general fundamentals of thermodynamic and kinetic factors that govern the crystallization of various hydrate polymorphs which could lead us to a systematic understanding of these systems.
MATERIALS AND METHODS Materials Stable NF hydrate II was crystallized according to the method reported by Pienaar et al.,19 where it has also reported that NF anhydrate can exist as two polymorphs, metastable a-polymorph and stable b-polymorph. NF powder (obtained from Unikem A/S, Copenhagen, Denmark) was identified as b-polymorph and used as purchased. The phase purity of b-polymorph was confirmed by X-ray powder diffraction, Raman and IR spectroscopy. One gram of NF powder (b-polymorph) was dissolved in a 200 cm3 acetone–water solution (volume ratio 1:1) at 558C and then cooled to room temperature. Metastable NF hydrate I was prepared by a method slightly modified from that published by Pienaar et al.9 Polyethylene glycol aqueous solution (1%, w/v) was prepared by dissolving polyethylene glycol (500 mg) (PEG; Simon & Werner GmbH, Flo¨ rsheim am Main, Germany, molecular weight ¼ 20,000) in distilled water (50 mL) at room temperature. Seventy-five milligrams of NF powder (used as purchased from Unikem A/S) was then dissolved in a mixture of acetone (20 mL) and polyethylene glycol solution (1%, w/v, 10 mL) at 558C. The solution vessel was loosely covered and placed undisturbed at room temperature to allow slow evaporation.
Evaporative Crystallization of NF Monohydrates Evaporative crystallization of NF monohydrates was carried out in six different acetone–water DOI 10.1002/jps
CRYSTALLIZATION OF A POLYMORPHIC HYDRATE SYSTEM
mixtures containing 0.11, 0.21, 0.26, 0.44, 0.67, and 0.80 mol fraction of water. The initial NF solution was prepared by dissolving 0.075 g NF powder (used as purchased from Unikem A/S) separately in each solvent at 558C, and then slowly cooling to room temperature while the solution vessels were tightly sealed. After cooling, each clear solution (3 mL) was placed on a microscopic slide to allow solvent evaporation, and the crystallization process was recorded with optical and Raman microscopy. Aqueous Solubility Measurements of NF Monohydrates Excess amounts of NF hydrates I and II powder were suspended separately in distilled water in test tubes. The test tubes were tightly sealed with screw caps, and rotated in a thermostatically controlled water bath at 238C for 72 h. The absorbances of the NF solutions were measured at 370 nm by ultraviolet–visible (UV–Vis) spectrophotometer (Evolution 300, Thermo Fisher Scientific, Inc.), and the concentration of the solutions was calculated based on a standard curve. The standard NF solutions were prepared by dissolving NF in distilled water in the concentration spanning 1–20 mg/mL. The crystalline form of the undissolved powder from solubility tests was characterized by Raman microscopy (A Renishaw system 1000 micro-Raman spectrometer).
Metastable Zone Limit and Seeded Cooling Crystallization of NF from Acetone–Water Mixture Containing 0.56 mol Fraction of Water When supersaturation is generated by the cooling of a solution, nucleation is not spontaneous, until the supersaturation reaches a certain level. This level is called the metastable zone limit. Knowledge of the metastable zone limit is very important in crystallization, and therefore it is often measured in the laboratory and used to define a working zone for industrial crystallization processes. In the present work, the metastable zone limit of NF acetone water solution was measured with the polythermal method proposed by Nyvlt28 in a 250 mL jacketed glass crystallizer equipped with a magnetic mixer and thermostat. The mixing intensity was kept constant for all operations by maintaining the agitation speed of the magnetic bar at 400 rpm. In the present work, the DOI 10.1002/jps
755
acetone–water mixture containing 0.56 mol fraction of water was selected as the solvent. The selection of this solvent was based on the relative stability of anhydrate/hydrate NF in acetone– water mixtures, the NF hydrate II is more stable in the acetone–water mixture containing 0.56 mol fraction of water at the studied temperature range. The experiment was performed by cooling the NF solution saturated at 40, 30, and 258C, respectively. The original solution was prepared by dissolving 0.625, 0.749, and 0.997 g of NF crystals in 100 g solvent according to the solubility of NF. The detailed solubility measurements of NF in the acetone–water mixture containing 0.56 mol fraction of water at different temperatures were reported in an earlier publication from our group, where the solid residues recovered after solubility tests were identified as NF hydrate II.29 The solution was kept at 58C higher than the saturation temperature for 30 min to attain thorough dissolution of the solute, and after that the linear cooling was executed with a cooling rate of 0.1678C/min. The detection of the metastable zone limit was performed by visual observation. The temperature at which the sudden turbid of the solution occurred was recorded, and the metastable zone limit was obtained. The crystal product was filtered and analyzed with a Raman microscope to identify the form of the nucleated crystals. Cooling crystallization seeded with hydrate I was performed with two cooling schemes as shown in Figure 1. NF–acetone–water solution saturated at 408C was prepared as the same way as described in the metastable zone limit measurement. The solution was cooled to 37 and 368C for schemes I and II, respectively, and then it was seeded with 0.1 g of hydrate I crystals. The supersaturation level S at the seeding point can be calculated from the concentration and solubility of NF at the seeding temperature as follows: C C C where C is the NF concentration and C the solubility of NF at the seeding point. The supersaturation level for schemes I and II was 0.08 and 0.11, respectively. The experiments were ended when the nucleation occurred, which was indicated by a suddenly increased turbidity of the solution as mentioned earlier. The suspension was filtered immediately with a vacuum filter, and the filter cake was then analyzed with a Raman microscope and an optical microscope. S¼
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
756
TIAN ET AL.
Figure 1. Cooling schemes used in the seeded cooling crystallization.
Raman Microscopy A Renishaw system 1000 micro-Raman spectrometer was used to monitor the crystalline form produced by the evaporative crystallization. For each liquid sample, a thin layer of the sample (3 mL) was placed separately on a microscopy slide, and several sample areas were selected and measured. The spectrometer equipped with a diode laser was focused on the sample through a 20 objective with a spot area of 12 mm 89 mm. The exposure time for data collection was set at 1 s and two accumulations per sample with a laser power of 50 mW. Wire V.2.0 software was used for instrument control and data acquisition. A LabRam 300 Raman spectrometer from Horiba Jobin Yvon was used for mapping the composition of the crystal product obtained from cooling crystallization. The system employed an external cavity stabilized single mode diode laser at 785 nm operating at 100 mW. The Raman spectrometer was interfaced with an optical microscope equipped with a motorized x, y, z stage. The filtered crystal cake was placed under the microscope, and the Raman spectra were collected from a grid of points evenly distributed across the surface of the cake. The distance between the grid points was 60 mm along both the x and the y direction. The confocal aperture was set at 1000 mm and the slit-width was 100 mm. The exposure time was 2 s per point. The data analysis and chemical image production were carried out with Matlab software (Version 7.0.4 (R14SP2), The MathWorks, Inc., Natick, MA). Although preprocessing was often employed in spectral data analysis, it was found to have no noticeable influence on the final image JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
and thus was not used. For producing the chemical image, the region of strong and nonoverlapping Raman bands for the two components hydrates I and II, and univariate analysis were used. The percentage of hydrate I indicated by the color scale in the chemical image was calculated as: RPHI ¼
HI HI þ HII
where HI and HII are the characteristic peak height of hydrates I and II at 1173 and 1178 cm1, respectively. Light Microscopy The crystallization process of NF during evaporative crystallization was observed using a Zeiss Axiolab microscope (Carl Zeiss, Inc., Beograd, Austria), and recorded using a DeltaPix digital camera (Infinity with 1.3 Mega Pixels CMOS, Maalov, Denmark). DeltaPix software 1.6 was employed for data acquisition.
RESULTS AND DISCUSSION The NF crystals produced from acetone–water mixture containing 0.67 mol fraction of water are shown in Figure 2. Upon solvent evaporation, needle-like crystals were formed rapidly, becoming visible after about 2 min. In contrast, no plate crystals were observed until around 6 min. NF hydrate II has a typical needle-like morphology, whereas hydrate I exhibits plate shape. Thus, the relative nucleation and growth rates of two hydrates during evaporative crystallization are DOI 10.1002/jps
CRYSTALLIZATION OF A POLYMORPHIC HYDRATE SYSTEM
757
Figure 2. Crystallization of nitrofurantoin from an acetone–water mixture containing 0.67 mol fraction of water observed under light microscopy. The time point at which each picture was taken is indicated. Examples of hydrate I crystals are circled, and examples of hydrate II clusters are indicated with arrow. Picture of the plate crystals (hydrate I) with a higher magnification is shown below.
directly demonstrated in Figure 2. Phase identification of these two hydrates after 8 min was also carried out on-line using Raman microscopy (Fig. 3). Most bands in the spectral range of 1650–1100 cm1 are different in either position or pattern between the two monohydrates. One of the most prominent differences was observed at around 1615 cm1, which is associated with stretching of the C – N bond. Also, the band at 1345 cm1 for monohydrate II shifted to 1352 cm1 in monohydrate I. This band is originated from in-phase NO2 stretch, and is affected by different intermolecular distances between the O atoms in the NO2 group and H atoms in the furan and hydantoin moieties of neighboring molecules. The growth along the long-axis of hydrate I crystals is noticeably slower than growth along the long-axis of hydrate II crystals at all time points. Also, the nucleation rate of hydrate I was slower than that of hydrate II, based on the number of crystals counted on the slide. No sign from hydrate I could be detected until after about 6 min. The nucleation and growth of hydrate II, however, started readily following solvent evaporation. Therefore, it appeared to be that hydrate DOI 10.1002/jps
I has a slower crystallization kinetics than that of hydrate II. This is also inline with the observation from Pienaar and coworkers as mentioned in the Introduction Section, albeit different crystallization methods were employed in their study. NF was also dissolved in five other acetone– water mixtures containing 0.11, 0.21, 0.26, 0.44, and 0.80 mol fraction of water, respectively. For all the solutions, a mixture of hydrates I and II was produced following solvent evaporation, as revealed by optical and Raman microscopy (Fig. 4). Several areas with crystals having different morphology were measured for each solvent mixture. For a better comparison, only the representative spectra from each area are plotted together with the Raman spectra from pure hydrates I and II powder samples. With decreasing water activity, the fraction of hydrate I was observed to increase with decreased water activity. This observation suggested that the relative crystallization rates of hydrates I and II are dependent on both supersaturation and water activity in the solution. Furthermore, as described in the Introduction Section, concomitant crystallization of two NF hydrates has also been observed earlier by Kelly.27 The reason for the concomitant JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
758
TIAN ET AL.
in water at 238C for 72 h, all of the hydrate I crystals remained unchanged. Therefore, it is likely that the conversion from unstable hydrate I to stable hydrate II is very slow, due to their relatively small solubility difference. Further investigation focusing on discovering the stability relationship of these two hydrates is currently continued in our group.
Metastable Zone Limit and Seeded Cooling Crystallization
Figure 3. Composition of crystallized NF (from acetone-water mixture containing 0.67 mol fraction of water) identified by Raman microscopy: representative Raman spectra of selected crystals (top) and corresponding crystals under microscopy (bottom left). The blue spectrum is from the crystals indicated by blue arrows, and red signal is from those in red circles. Picture of the plate crystals (hydrate I) with a higher magnification is shown at the right side (bottom).
crystallization of two different polymorphs often resides in their similar free energy, and thus solubility and dissolution rate. The solubility of the two monohydrates in water at 238C revealed that they have rather close aqueous solubility values (131 12 mg/mL and 110 4 mg/mL for hydrates I and II, respectively). The slightly higher solubility of hydrate I than hydrate II indicated that the conversion of hydrate I to hydrate II in aqueous solution at 238C is thermodynamically spontaneous. A solvent-mediated transformation from hydrate I to hydrate II would occur driven by the solubility difference between the two forms. However, suspending the hydrate I JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
The metastable zone limit was measured by cooling the NF solutions saturated at 40, 30, and 258C. The nucleation of hydrate II was encountered in all experiments. The metastable zone limit of NF solution in current experimental conditions is shown in Figure 5. In principle, the NF molecules aggregate together to form clusters in the metastable zone limit, but the stable nuclei can only be formed and are only capable of growing when the solution is cooled to reach the metastable zone limit. Once the stable nuclei are formed, the nucleation proceeds with considerable rapidity. No hydrate I could be produced in any of the metastable zone limit measurement experiments. The reason could be that the growth of the hydrate II was much faster than that of the hydrate I, and thus the growth of hydrate II consumed the supersaturation rapidly and suppressed the crystallization of hydrate I. Seeding has been widely used in batch crystallization processes to control crystal size distribution18,19 and the polymorphic form of the crystals.20,21 The fundamental mechanism of seeding is that the seed crystals have a catalyzing effect on nucleation phenomena, and therefore seeding can initiate secondary nucleation at a moderate supersaturation level within the metastable zone limit, thus, eliminating the undesired primary nucleation, which is a process that is hard to control. The application of seeding to control polymorphic form requires that the secondary nucleation of the desired polymorph occurs with a considerable speed at a lower supersaturation than needed for spontaneous nucleation of the undesired polymorphs. Consequently, the supersaturation is consumed by the nucleation and growth of the seeded polymorph, and therefore the supersaturation will not reach the metastable zone limit of the undesired polymorph. It has been reported that seeding of the desired polymorph can eliminate the nucleation of the undesired DOI 10.1002/jps
CRYSTALLIZATION OF A POLYMORPHIC HYDRATE SYSTEM
759
Figure 4. Morphology of crystallized nitrofurantoin monohydrates from acetone– water mixtures (right), and the corresponding phase identification by Raman microscopy (left).
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
760
TIAN ET AL.
Figure 5. Schematic diagram of solubility and metastable zone limit of NF in acetone–water mixtures containing 0.56 mol fraction water (the cooling rate was 0.1678C/min, n ¼ 2). Solid residues recovered after solubility tests were identified as NF hydrate II.
modification of abecarnil, and thus the desired pure metastable form A can be produced.30 To further explore the nucleation mechanism of the two polymorphs of NF hydrate, two different seeded cooling crystallization experiments were performed. In scheme I, the solution was cooled down with a cooling rate of 0.1678C/min, and hydrate I seeds were added to the solution when the temperature reached 378C. The solution was further cooled down, and nucleation started at 30.88C, which was indicated by a sudden turbid of the solution. By analyzing the obtained product, it was found that the nucleated crystals were hydrate II. In scheme II, the solution was cooled to 368C and kept isothermal. Then hydrate I seeds were added to the solution. The solution became turbid 2 h after seeding, and the crystal product was harvested. It was observed that the produced crystals contained the hydrate I seeds and some newly nucleated hydrate II. For both schemes I and II, the seeding temperature was selected within the metastable zone (see Fig. 5), where the spontaneous nucleation is not expected to happen instantly. As shown in Figure 6A, the large dark-yellow particle is hydrate I and the surrounding small particles are hydrate II. The co-existence of the hydrate I seeds and the newly crystallized hydrate II was also confirmed by the corresponding Raman chemical image (Fig. 6B). Although seeding has been proven to be an effective way of controlling the polymorphic form of crystals,30,31 it failed to produce the target JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
Figure 6. A: Optical image of the hydrate I seed crystal with the nucleated hydrate II crystals recovered from the seeded cooling crystallization (the cooling scheme used was scheme II in Fig. 1, the scale bar is 1000 mm), and (B) the corresponding Raman chemical image (the color scale represents the relative peak height of hydrate I (RPHI), which is calculated as: RPHI ¼ HI =HI þ HII where HI and HII are the characteristic peak height of hydrates I and II at 1173 and 1178 cm1, respectively).
hydrate I in the present work. This observation proved that the secondary nucleation and growth kinetics of hydrate I is extremely slow within the metastable zone, and therefore the consumption rate of supersaturation was not significant compared with the generation rate of supersaturation by cooling. This led to a continuous increase of the supersaturation in the solution and eventually reached the metastable zone limit, where spontaneous nucleation of hydrate II occurred. Therefore, the production of hydrate I requires alternative polymorphic control techniques, such as tailor-made additives,32 ultrasoundinduced nucleation33 and the appropriate crystallization parameters, for instance supersaturation level and water activity in the solution. This study revealed that the crystallization of two polymorphs of NF monohydrate follows the polymorphism theories of independent and DOI 10.1002/jps
CRYSTALLIZATION OF A POLYMORPHIC HYDRATE SYSTEM
competitive nucleation. The independent nucleation was demonstrated most clearly by evaporative crystallization, where both of the two hydrates were observed in all solvent systems. Nucleation of the two hydrates is also competitive, as the cooling crystallization shows. The small solubility difference between the two hydrates retarded the formation of the unstable hydrate (hydrate I), where also the stable hydrate (hydrate II) had faster nucleation and growth rates than hydrate I. The crystallization of hydrate II consumed all the supersaturation in the solution, which thus impeded the crystallization of hydrate I under all cooling conditions. According to the interpretation of the classical crystal growth theory, the growth of crystal faces consists of two steps: first the growth units are transported to the crystal surface from bulk solution by diffusion and convection, and then the growth units are incorporated into the crystal lattice through an integration reaction. In most cases, the second step is the rate-controlling, and thus determines the relative crystallization rate of the polymorphs. The surface integration rate is affected by not only the thermodynamic driving force of crystallization (the free energy difference between the crystal and liquid phases of the substance or the difference in chemical potential of the substance in saturated and supersaturated solution for crystallization from solution), but also other factors, such as the surface diffusion coefficient
761
of the solute molecules and the properties of the crystal surface. The classical nucleation and crystal growth theories are derived from anhydrate crystallization, and the free enthalpy change of solid phase formation is estimated for only solute molecules.34 The growth mechanism of hydrate can be expected to be more complicated than anhydrate, due to the incorporation of water molecules into the crystal lattice, where the interaction between the water and the solute molecules will have to be taken into account with estimation of the free enthalpy of solid phase formation. The chemical structure of NF anhydrate (top), and crystal structures of NF hydrates I and II characterized by hydrogen bonding (A and B, respectively) are shown in Figure 7. NF hydrate I is formed through hydrogen bond bonding between the N-H of NF molecule and the O atom (water). The H atoms of a water molecule are also hydrogen bonded with the O14 (carbonyl group) from two different NF molecules. These hydrogen bonding connections are indicated by the green dotted lines in Figure 1. Therefore, water in NF hydrate I forms three H bonds to neighboring molecules which is one of the two most common hydrogen bonding arrangements within hydrates according to Gillon et al.35 In NF hydrate II, the hydrate is also formed by the hydrogen bonding of N–H. . .O–H (water), where one H atom of a water molecule is linked to O15 (second NF hydrate).
Figure 7. Chemical structure of NF anhydrate (top); Crystal structure of nitrofurantoin polymorphic hydrate I (A) and II (B) characterized with H-bonding arrangements. DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
762
TIAN ET AL.
Therefore, each water molecule in hydrate II is hydrogen bonded with two NF molecules. Although this is an unusual H-bonding motif, there also exists a series of other hydrogen bonds in NF hydrate II as suggested by Pienaar et al.19 The distances O (water). . .O (NO2), O (water). . . O5, O (water). . .O14, and O (water). . . N8 are all in ˚ , indicating weak the range 2.961(4)–3.211(4) A and possibly bifurcate hydrogen bonds. Intermolecular contact C7–H7. . .O15 also showed a geometry meeting the criteria for a hydrogen bond. NF hydrate II thus has a more complex hydrogen bonding scheme than hydrate I. Such complex hydrogen bonding scheme of hydrate II might then contribute to its higher stability, and leads to its more favorable crystallization.
CONCLUSION In this article, we have presented a phenomenon, namely, independent and competitive nucleation for the crystallization of two polymorphs of NF monohydrate. The mixtures of hydrates I and II produced from the evaporative crystallization suggested that the nucleation of hydrates I and II depends on both supersaturation level and the solvent composition in the solution (e.g., water activity). Although NF hydrate I is less stable than hydrate II indicated by its slightly higher aqueous solubility, seeding of the hydrate I failed to produce pure hydrate I in cooling crystallization experiments. This is due to the slow secondary nucleation rate of hydrate I. The production of the hydrate I of NF represents a challenging issue in polymorphism control, and it requires comprehensive understanding of the crystallization mechanisms combined with more robust polymorphic control techniques. It is expected that this crystallization mechanism can be utilized in various fields of chemical industry. With the growing interest in hydrate systems, we hope this work will stimulate more research into the understanding of the competing thermodynamic and kinetic factors that govern the crystallization of various hydrate polymorphs.
ACKNOWLEDGMENTS The authors thank Anne Zimmermann (Faculty of Pharmaceutical Sciences, University of Copenhagen, Denmark) for the collection of X-ray data, and Dr. Erik Skibsted (Novo Nordisk A/S, JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
Denmark) for help with Raman measurements. Access to Raman microscopy from Novo Nordisk A/S, Denmark, and to X-ray powder diffraction provided by Lundbeck A/S, Denmark are also acknowledged. The Academy of Finland (project No. 122828) is thanked for financial support. The anonymous reviewers are thanked for their constructive criticism.
REFERENCES 1. Chow K, Tong HHY, Lum S, Chow AHL. 2008. Engineering of pharmaceutical materials: An industrial perspective. J Pharm Sci 97:2855–2877. 2. Gavezzotti A. 2007. A solid-state chemist’s view of the crystal polymorphism of organic compounds. J Pharm Sci 96:2232–2241. 3. Byrn SR, Pfeiffer RR, Stowell JG. 1999. Solid state chemistry of drugs. 2nd edition. West Lafayette, IN: SSCI Inc. 4. Brittain HG. 2009. Polymorphism and solvatomorphism 2007. J Pharm Sci 98: 1617–1642. 5. York P. 1983. Solid-state properties of powders in the formulation and processing of solid dosage forms. Int J Pharm 14:1–28. 6. Ostwald WZ. 1897. Studies on formation and transformation of solid materials. Z Phys Chem 22:289– 330. 7. ter Horst JH, Kramer HJM, Jansens PJ. 2002. A new molecular modeling approach to predict concomitant nucleation of polymorphs. Cryst Growth Des 2:351–356. 8. Bernstein J, Davey RJ, Henck J-O. 1999. Concomitant polymorphs. Angew Chem Int Ed 38: 3440–3461. 9. Blagden N, Davey RJ. 2003. Polymorph selection: Challenges for the future? Cryst Growth Des 3:873– 885. 10. Stoica C, Tinnemans P, Meekes H, Vlieg E, van Hoof PJCM, Kaspersen FM. 2005. Epitaxial 2D nucleation of metastable polymorphs: A 2D Version of Ostwald’s Rule of Stages. Cryst Growth Des 5: 833–1308. 11. Chen S, Xi H, Yu L. 2005. Cross-nucleation between ROY polymorphs. J Am Chem Soc 127:17439– 17444. 12. Desgranges C, Delhommelle J. 2006. Molecular mechanism for the cross-nucleation between polymorphs. J Am Chem Soc 128:10368–10369. 13. Mitchell C, Yu L, Ward MD. 2001. Selective nucleation and discovery of organic polymorphs through epitaxy with single crystal substrates. J Am Chem Soc 123:10830–10839. 14. Yu L. 2003. Nucleation of one polymorph by another. J Am Chem Soc 125:6380–6381. DOI 10.1002/jps
CRYSTALLIZATION OF A POLYMORPHIC HYDRATE SYSTEM
15. Haleblian JK, Koda RT, Biles JA. 1971. Isolation and characterization of some solid phases of fluprednisolone. J Pharm Sci 60:1485–1488. 16. Shefter E, Higuchi T. 1963. Dissolution behavior of crystalline solvated and nonsolvated forms of some pharmaceuticals. J Pharm Sci 52:781–791. 17. Khankari RK. 1993. Physicochemical characterization and thermodynamic properties of nedocromil salt hydrates. PhD Thesis: University of Minnesota, Minneapolis, MN. 18. van Tonder EC, Maleka TSP, Liebenberg W, Song M, Wurster DE, de Villiers MM. 2004. Preparation and physicochemical properties of niclosamide anhydrate and two monohydrates. Int J Pharm 269:417–432. 19. Pienaar EW, Caira MR, Lotter AP. 1993. Polymorphs of nitrofurantoin. I. Preparation and X-ray crystal structures of two monohydrated forms of nitrofurantoin. J Crystallogr Spectrosc Res 23: 739–744. 20. Aaltonen J, Strachan CJ, Pollanen K, Yliruusi J, Rantanen J. 2007. Hyphenated spectroscopy as a polymorph screening tool. J Pharm Biomed Anal 44:477–483. 21. Otsuka M, Teraoka R, Matsuda Y. 1992. Rotatingdisk dissolution kinetics of nitrofurantoin anhydrate and monohydrate at various temperatures. Pharm Res 9:307–311. 22. Kojima T, Yamauchi Y, Onoue S, Tsuda Y. 2008. Evaluation of hydrate formation of a pharmaceutical solid by using diffuse reflectance infrared Fourier-transform spectroscopy. J Pharm Biomed Anal 46:788–791. 23. Otsuka M, Teraoka R, Matsuda Y. 1990. Characterization of nitrofurantoin anhydrate and monohydrate, and their dissolution behaviors. Chem Pharm Bull 38:2667–2670. 24. Wikstro¨ m H, Rantanen J, Gift AD, Taylor LS. 2008. Toward an understanding of the factors influencing anhydrate-to-hydrate transformation kinetics in aqueous environments. Cryst Growth Des 8: 2684–2693.
DOI 10.1002/jps
763
25. Caira MR, Pienaar EW, Loetter AP. 1996. Polymorphism and pseudopolymorphism of the antibacterial nitrofurantoin. Mol Cryst Liq Cryst Sci Technol Section A Mol Cryst Liq Cryst 279:241– 264. 26. Otsuka M, Matsuda Y. 1993. Effect of environmental humidity on the transformation pathway of nitrofurantoin modifications during grinding and the physicochemical properties of ground products. J Pharm Pharmacol 45:406–413. 27. Kelly RC. 2003. A molecular approach to understanding the directed nucleation and phase transformation of carbamazepine and nitrofurantoin in aqueous and organic solutions. Ph.D Thesis, University of Michigan; MI, USA. 28. Nyvlt J. 1968. Kinetics of nucleation in solution. J Cryst Growth 3–4:377–383. 29. Tian F, Qu H, Louhi-Kultanen M, Rantanen J. 2009. Mechanistic insight into the evaporative crystallization of two polymorphs of nitrofurantoin monohydrate. J Cryst Growth 311: 2580–2589. 30. Beckmann W, Nickisch K, Budde U. 1998. Development of a seeding technique for the crystallization of the metastable a modification of abecarnil. Org Process Res Dev 2:298–304. 31. Beckmann W. 2000. Seeding the desired polymorph: Background, possibilities, limitations, and case studies. Org Process Res Dev 4:372–383. 32. Davey RJ, Williams-Seton L, Lieberman HF, Blagden N. 1999. Stabilizing a solid-solid interface with a molecular-scale adhesive. Nature 402:797– 799. ˚ C. 2005. 33. Gracin S, Uusi-Penttila¨ M, Rasmuson A Influence of ultrasound on the nucleation of polymorphs of p-aminobenzoic acid. Cryst Growth Des 5:1787–1794. 34. Mersmann A. 2001. Crystallization Technology Handbook. 2nd edition. New York: Marcel Dekker. 35. Gillon AL, Feeder N, Davey RJ, Storey R. 2003. Hydration in molecular crystals—A Cambridge structural database analysis. Cryst Growth Des 3:663–673.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
Faculty of Pharmaceutical Sciences, Department of Pharmaceutics and Analytical Chemistry, University of Copenhagen, Copenhagen, Denmark 2
Institute of Chemical Engineering, Biotechnology and Environmental Technology, University of Southern Denmark, Odense, Denmark 3
Department of Chemical Technology, Lappeenranta University of Technology, Lappeenranta, Finland
Received 18 March 2009; revised 11 May 2009; accepted 28 May 2009 Published online 30 June 2009 in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jps.21865
ABSTRACT: Nitrofurantoin can form two monohydrates, which have the same chemical composition and molar ratio of water, but differ in the crystal arrangements. The two monohydrates (hydrates I and II) could be produced independently via evaporative crystallization, where supersaturation and solvent composition were both found to have an effect. Hydrate I showed much slower crystallization than hydrate II. During cooling crystallization, the nucleation and growth of hydrate II was again dominant, consuming all supersaturation and leading to no hydrate I formation. Seeding of hydrate I during cooling crystallization was also applied, but the hydrate I seeds were not able to initiate its nucleation rather than dissolving into crystallizing solution. Although solubility tests revealed that hydrate II is more stable than hydrate I due to its lower solubility (110 4 and 131 12 mg/mL for hydrates II and I, respectively), this difference is rather small. Therefore, the small free energy difference between the two hydrates, together with the slow crystallization of hydrate I, both lead to a hindrance of hydrate I formation. Furthermore, the crystal structure of hydrate II demonstrated a higher H-bonding extent than hydrate I, suggesting its more favorable crystallization. This is in good agreement with experimental results. ß 2009 Wiley-Liss, Inc. and the American Pharmacists Association J Pharm Sci 99:753–763, 2010
Keywords: hydrate; crystallization; supersaturation; polymorphism; crystal structure; nitrofurantoin
INTRODUCTION Understanding the crystallization mechanism of organic molecules is of great scientific importance.1,2 Crystal structure and particle habit determine critical material properties.3–5 The occurrence of polymorphs during crystallization from solution quite often follows the Ostwald’s rule of stages, where the least stable form
Correspondence to: J. Rantanen (Telephone: 45-35-33-6585; Fax: 45-35-33-60-30; E-mail: [email protected]) Journal of Pharmaceutical Sciences, Vol. 99, 753–763 (2010) ß 2009 Wiley-Liss, Inc. and the American Pharmacists Association
crystallizes out first and then transforms to the more stable form.6 However, concomitant polymorphism is also frequently observed in the crystallization of organic compounds, where more than one polymorphic form will crystallize at the same time. The concomitant polymorphism can be caused by competing nucleation of the modifications,7,8 solvent-mediated transformation from one form to another,9 or cross-nucleation of another polymorph nucleating on the initially present one.10–14 Compared with the systems following Ostwald’s rule of stages, concomitant polymorphism represents a much more complicated issue and makes the polymorphism control extremely difficult.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
753
754
TIAN ET AL.
Hydrates with the same molar ratio of water can also crystallize into different crystalline packings and thus exhibit polymorphism. Examples of such systems including fluprednisolone monohydrate,15 succinyl sulfathiazole monohydrate,16 nedocromil sodium monohydrate,17 niclosamide monohydrate,18 and nitrofurantoin (NF) monohydrate.19 The relative thermodynamic stability of these hydrate polymorphs has not been thoroughly investigated, and the understanding of the crystallization behavior of polymorphs of hydrates is very limited. To understand and control crystallization of a polymorphic hydrate system, we used two polymorphs of NF monohydrate as a model compound. NF monohydrate, 1-[[(5-Nitro-2-furanyl)-methylene]amino]2,4-imidazolidinedione monohydrate, is known to crystallize into two crystalline structures: P21, ˚ , b ¼ 6.476 A ˚ , c ¼ 16.969 A ˚ , b ¼ 96.688 a ¼ 9.783 A ˚ ˚ ˚. and Pbca, a ¼ 12.642 A, b ¼ 9.857 A, c ¼ 17.383 A The monoclinic P21 structure is called hydrate I, and the orthorhombic Pbca is named hydrate II.19 NF hydrate has been intensely investigated.19–27 However, according to the authors’ knowledge, all of these publications have studied hydrate II, while only a few have characterized hydrate I. Pienaar and coworkers were the first who successfully prepared the hydrate I by slowly evaporating a saturated NF solution at 408C. The authors reported that NF hydrate I was observed within 7 days whereas they found NF hydrate II needles within 3 days.19,25 This suggests a slower crystallization rate for hydrate I relative to hydrate II. Alternatively, Otsuka and Matsuda26 reported that hydrate I could be produced via grinding hydrate II at humidity conditions for over 7 h. The authors did not offer any explanations as to why hydrate I could be converted through grinding of hydrate II. Because grinding is often used to induce high-energy intermediate forms, for instance the amorphous form, one explanation is that hydrate I is an unstable form in comparison to hydrate II. Recently, Kelly also performed the crystallization of two NF monohydrates. Pure NF hydrate I, however, could not be achieved since it always crystallized concomitantly with hydrate II when suspending anhydrous NF b-polymorph in supersaturated solutions.27 Although the relative stability of two NF hydrates was not clearly discussed in any of these publications described above, the relatively large difference in the number of publications between these two NF hydrates implies that NF JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
hydrate I is not so often encountered as NF hydrate II when crystallizing NF hydrate. The main aim of this research is therefore to explore the crystallization mechanism of two different polymorphs of NF monohydrate, in order to answer the questions of whether their crystallization follows the classic polymorphism theories and what roles typical crystallization parameters, that is, supersaturation and water activity, play in their crystallization? With the growing importance of hydrate in various pharmaceutical and chemistry fields, a particular purpose of this work is also to show the general fundamentals of thermodynamic and kinetic factors that govern the crystallization of various hydrate polymorphs which could lead us to a systematic understanding of these systems.
MATERIALS AND METHODS Materials Stable NF hydrate II was crystallized according to the method reported by Pienaar et al.,19 where it has also reported that NF anhydrate can exist as two polymorphs, metastable a-polymorph and stable b-polymorph. NF powder (obtained from Unikem A/S, Copenhagen, Denmark) was identified as b-polymorph and used as purchased. The phase purity of b-polymorph was confirmed by X-ray powder diffraction, Raman and IR spectroscopy. One gram of NF powder (b-polymorph) was dissolved in a 200 cm3 acetone–water solution (volume ratio 1:1) at 558C and then cooled to room temperature. Metastable NF hydrate I was prepared by a method slightly modified from that published by Pienaar et al.9 Polyethylene glycol aqueous solution (1%, w/v) was prepared by dissolving polyethylene glycol (500 mg) (PEG; Simon & Werner GmbH, Flo¨ rsheim am Main, Germany, molecular weight ¼ 20,000) in distilled water (50 mL) at room temperature. Seventy-five milligrams of NF powder (used as purchased from Unikem A/S) was then dissolved in a mixture of acetone (20 mL) and polyethylene glycol solution (1%, w/v, 10 mL) at 558C. The solution vessel was loosely covered and placed undisturbed at room temperature to allow slow evaporation.
Evaporative Crystallization of NF Monohydrates Evaporative crystallization of NF monohydrates was carried out in six different acetone–water DOI 10.1002/jps
CRYSTALLIZATION OF A POLYMORPHIC HYDRATE SYSTEM
mixtures containing 0.11, 0.21, 0.26, 0.44, 0.67, and 0.80 mol fraction of water. The initial NF solution was prepared by dissolving 0.075 g NF powder (used as purchased from Unikem A/S) separately in each solvent at 558C, and then slowly cooling to room temperature while the solution vessels were tightly sealed. After cooling, each clear solution (3 mL) was placed on a microscopic slide to allow solvent evaporation, and the crystallization process was recorded with optical and Raman microscopy. Aqueous Solubility Measurements of NF Monohydrates Excess amounts of NF hydrates I and II powder were suspended separately in distilled water in test tubes. The test tubes were tightly sealed with screw caps, and rotated in a thermostatically controlled water bath at 238C for 72 h. The absorbances of the NF solutions were measured at 370 nm by ultraviolet–visible (UV–Vis) spectrophotometer (Evolution 300, Thermo Fisher Scientific, Inc.), and the concentration of the solutions was calculated based on a standard curve. The standard NF solutions were prepared by dissolving NF in distilled water in the concentration spanning 1–20 mg/mL. The crystalline form of the undissolved powder from solubility tests was characterized by Raman microscopy (A Renishaw system 1000 micro-Raman spectrometer).
Metastable Zone Limit and Seeded Cooling Crystallization of NF from Acetone–Water Mixture Containing 0.56 mol Fraction of Water When supersaturation is generated by the cooling of a solution, nucleation is not spontaneous, until the supersaturation reaches a certain level. This level is called the metastable zone limit. Knowledge of the metastable zone limit is very important in crystallization, and therefore it is often measured in the laboratory and used to define a working zone for industrial crystallization processes. In the present work, the metastable zone limit of NF acetone water solution was measured with the polythermal method proposed by Nyvlt28 in a 250 mL jacketed glass crystallizer equipped with a magnetic mixer and thermostat. The mixing intensity was kept constant for all operations by maintaining the agitation speed of the magnetic bar at 400 rpm. In the present work, the DOI 10.1002/jps
755
acetone–water mixture containing 0.56 mol fraction of water was selected as the solvent. The selection of this solvent was based on the relative stability of anhydrate/hydrate NF in acetone– water mixtures, the NF hydrate II is more stable in the acetone–water mixture containing 0.56 mol fraction of water at the studied temperature range. The experiment was performed by cooling the NF solution saturated at 40, 30, and 258C, respectively. The original solution was prepared by dissolving 0.625, 0.749, and 0.997 g of NF crystals in 100 g solvent according to the solubility of NF. The detailed solubility measurements of NF in the acetone–water mixture containing 0.56 mol fraction of water at different temperatures were reported in an earlier publication from our group, where the solid residues recovered after solubility tests were identified as NF hydrate II.29 The solution was kept at 58C higher than the saturation temperature for 30 min to attain thorough dissolution of the solute, and after that the linear cooling was executed with a cooling rate of 0.1678C/min. The detection of the metastable zone limit was performed by visual observation. The temperature at which the sudden turbid of the solution occurred was recorded, and the metastable zone limit was obtained. The crystal product was filtered and analyzed with a Raman microscope to identify the form of the nucleated crystals. Cooling crystallization seeded with hydrate I was performed with two cooling schemes as shown in Figure 1. NF–acetone–water solution saturated at 408C was prepared as the same way as described in the metastable zone limit measurement. The solution was cooled to 37 and 368C for schemes I and II, respectively, and then it was seeded with 0.1 g of hydrate I crystals. The supersaturation level S at the seeding point can be calculated from the concentration and solubility of NF at the seeding temperature as follows: C C C where C is the NF concentration and C the solubility of NF at the seeding point. The supersaturation level for schemes I and II was 0.08 and 0.11, respectively. The experiments were ended when the nucleation occurred, which was indicated by a suddenly increased turbidity of the solution as mentioned earlier. The suspension was filtered immediately with a vacuum filter, and the filter cake was then analyzed with a Raman microscope and an optical microscope. S¼
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
756
TIAN ET AL.
Figure 1. Cooling schemes used in the seeded cooling crystallization.
Raman Microscopy A Renishaw system 1000 micro-Raman spectrometer was used to monitor the crystalline form produced by the evaporative crystallization. For each liquid sample, a thin layer of the sample (3 mL) was placed separately on a microscopy slide, and several sample areas were selected and measured. The spectrometer equipped with a diode laser was focused on the sample through a 20 objective with a spot area of 12 mm 89 mm. The exposure time for data collection was set at 1 s and two accumulations per sample with a laser power of 50 mW. Wire V.2.0 software was used for instrument control and data acquisition. A LabRam 300 Raman spectrometer from Horiba Jobin Yvon was used for mapping the composition of the crystal product obtained from cooling crystallization. The system employed an external cavity stabilized single mode diode laser at 785 nm operating at 100 mW. The Raman spectrometer was interfaced with an optical microscope equipped with a motorized x, y, z stage. The filtered crystal cake was placed under the microscope, and the Raman spectra were collected from a grid of points evenly distributed across the surface of the cake. The distance between the grid points was 60 mm along both the x and the y direction. The confocal aperture was set at 1000 mm and the slit-width was 100 mm. The exposure time was 2 s per point. The data analysis and chemical image production were carried out with Matlab software (Version 7.0.4 (R14SP2), The MathWorks, Inc., Natick, MA). Although preprocessing was often employed in spectral data analysis, it was found to have no noticeable influence on the final image JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
and thus was not used. For producing the chemical image, the region of strong and nonoverlapping Raman bands for the two components hydrates I and II, and univariate analysis were used. The percentage of hydrate I indicated by the color scale in the chemical image was calculated as: RPHI ¼
HI HI þ HII
where HI and HII are the characteristic peak height of hydrates I and II at 1173 and 1178 cm1, respectively. Light Microscopy The crystallization process of NF during evaporative crystallization was observed using a Zeiss Axiolab microscope (Carl Zeiss, Inc., Beograd, Austria), and recorded using a DeltaPix digital camera (Infinity with 1.3 Mega Pixels CMOS, Maalov, Denmark). DeltaPix software 1.6 was employed for data acquisition.
RESULTS AND DISCUSSION The NF crystals produced from acetone–water mixture containing 0.67 mol fraction of water are shown in Figure 2. Upon solvent evaporation, needle-like crystals were formed rapidly, becoming visible after about 2 min. In contrast, no plate crystals were observed until around 6 min. NF hydrate II has a typical needle-like morphology, whereas hydrate I exhibits plate shape. Thus, the relative nucleation and growth rates of two hydrates during evaporative crystallization are DOI 10.1002/jps
CRYSTALLIZATION OF A POLYMORPHIC HYDRATE SYSTEM
757
Figure 2. Crystallization of nitrofurantoin from an acetone–water mixture containing 0.67 mol fraction of water observed under light microscopy. The time point at which each picture was taken is indicated. Examples of hydrate I crystals are circled, and examples of hydrate II clusters are indicated with arrow. Picture of the plate crystals (hydrate I) with a higher magnification is shown below.
directly demonstrated in Figure 2. Phase identification of these two hydrates after 8 min was also carried out on-line using Raman microscopy (Fig. 3). Most bands in the spectral range of 1650–1100 cm1 are different in either position or pattern between the two monohydrates. One of the most prominent differences was observed at around 1615 cm1, which is associated with stretching of the C – N bond. Also, the band at 1345 cm1 for monohydrate II shifted to 1352 cm1 in monohydrate I. This band is originated from in-phase NO2 stretch, and is affected by different intermolecular distances between the O atoms in the NO2 group and H atoms in the furan and hydantoin moieties of neighboring molecules. The growth along the long-axis of hydrate I crystals is noticeably slower than growth along the long-axis of hydrate II crystals at all time points. Also, the nucleation rate of hydrate I was slower than that of hydrate II, based on the number of crystals counted on the slide. No sign from hydrate I could be detected until after about 6 min. The nucleation and growth of hydrate II, however, started readily following solvent evaporation. Therefore, it appeared to be that hydrate DOI 10.1002/jps
I has a slower crystallization kinetics than that of hydrate II. This is also inline with the observation from Pienaar and coworkers as mentioned in the Introduction Section, albeit different crystallization methods were employed in their study. NF was also dissolved in five other acetone– water mixtures containing 0.11, 0.21, 0.26, 0.44, and 0.80 mol fraction of water, respectively. For all the solutions, a mixture of hydrates I and II was produced following solvent evaporation, as revealed by optical and Raman microscopy (Fig. 4). Several areas with crystals having different morphology were measured for each solvent mixture. For a better comparison, only the representative spectra from each area are plotted together with the Raman spectra from pure hydrates I and II powder samples. With decreasing water activity, the fraction of hydrate I was observed to increase with decreased water activity. This observation suggested that the relative crystallization rates of hydrates I and II are dependent on both supersaturation and water activity in the solution. Furthermore, as described in the Introduction Section, concomitant crystallization of two NF hydrates has also been observed earlier by Kelly.27 The reason for the concomitant JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
758
TIAN ET AL.
in water at 238C for 72 h, all of the hydrate I crystals remained unchanged. Therefore, it is likely that the conversion from unstable hydrate I to stable hydrate II is very slow, due to their relatively small solubility difference. Further investigation focusing on discovering the stability relationship of these two hydrates is currently continued in our group.
Metastable Zone Limit and Seeded Cooling Crystallization
Figure 3. Composition of crystallized NF (from acetone-water mixture containing 0.67 mol fraction of water) identified by Raman microscopy: representative Raman spectra of selected crystals (top) and corresponding crystals under microscopy (bottom left). The blue spectrum is from the crystals indicated by blue arrows, and red signal is from those in red circles. Picture of the plate crystals (hydrate I) with a higher magnification is shown at the right side (bottom).
crystallization of two different polymorphs often resides in their similar free energy, and thus solubility and dissolution rate. The solubility of the two monohydrates in water at 238C revealed that they have rather close aqueous solubility values (131 12 mg/mL and 110 4 mg/mL for hydrates I and II, respectively). The slightly higher solubility of hydrate I than hydrate II indicated that the conversion of hydrate I to hydrate II in aqueous solution at 238C is thermodynamically spontaneous. A solvent-mediated transformation from hydrate I to hydrate II would occur driven by the solubility difference between the two forms. However, suspending the hydrate I JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
The metastable zone limit was measured by cooling the NF solutions saturated at 40, 30, and 258C. The nucleation of hydrate II was encountered in all experiments. The metastable zone limit of NF solution in current experimental conditions is shown in Figure 5. In principle, the NF molecules aggregate together to form clusters in the metastable zone limit, but the stable nuclei can only be formed and are only capable of growing when the solution is cooled to reach the metastable zone limit. Once the stable nuclei are formed, the nucleation proceeds with considerable rapidity. No hydrate I could be produced in any of the metastable zone limit measurement experiments. The reason could be that the growth of the hydrate II was much faster than that of the hydrate I, and thus the growth of hydrate II consumed the supersaturation rapidly and suppressed the crystallization of hydrate I. Seeding has been widely used in batch crystallization processes to control crystal size distribution18,19 and the polymorphic form of the crystals.20,21 The fundamental mechanism of seeding is that the seed crystals have a catalyzing effect on nucleation phenomena, and therefore seeding can initiate secondary nucleation at a moderate supersaturation level within the metastable zone limit, thus, eliminating the undesired primary nucleation, which is a process that is hard to control. The application of seeding to control polymorphic form requires that the secondary nucleation of the desired polymorph occurs with a considerable speed at a lower supersaturation than needed for spontaneous nucleation of the undesired polymorphs. Consequently, the supersaturation is consumed by the nucleation and growth of the seeded polymorph, and therefore the supersaturation will not reach the metastable zone limit of the undesired polymorph. It has been reported that seeding of the desired polymorph can eliminate the nucleation of the undesired DOI 10.1002/jps
CRYSTALLIZATION OF A POLYMORPHIC HYDRATE SYSTEM
759
Figure 4. Morphology of crystallized nitrofurantoin monohydrates from acetone– water mixtures (right), and the corresponding phase identification by Raman microscopy (left).
DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
760
TIAN ET AL.
Figure 5. Schematic diagram of solubility and metastable zone limit of NF in acetone–water mixtures containing 0.56 mol fraction water (the cooling rate was 0.1678C/min, n ¼ 2). Solid residues recovered after solubility tests were identified as NF hydrate II.
modification of abecarnil, and thus the desired pure metastable form A can be produced.30 To further explore the nucleation mechanism of the two polymorphs of NF hydrate, two different seeded cooling crystallization experiments were performed. In scheme I, the solution was cooled down with a cooling rate of 0.1678C/min, and hydrate I seeds were added to the solution when the temperature reached 378C. The solution was further cooled down, and nucleation started at 30.88C, which was indicated by a sudden turbid of the solution. By analyzing the obtained product, it was found that the nucleated crystals were hydrate II. In scheme II, the solution was cooled to 368C and kept isothermal. Then hydrate I seeds were added to the solution. The solution became turbid 2 h after seeding, and the crystal product was harvested. It was observed that the produced crystals contained the hydrate I seeds and some newly nucleated hydrate II. For both schemes I and II, the seeding temperature was selected within the metastable zone (see Fig. 5), where the spontaneous nucleation is not expected to happen instantly. As shown in Figure 6A, the large dark-yellow particle is hydrate I and the surrounding small particles are hydrate II. The co-existence of the hydrate I seeds and the newly crystallized hydrate II was also confirmed by the corresponding Raman chemical image (Fig. 6B). Although seeding has been proven to be an effective way of controlling the polymorphic form of crystals,30,31 it failed to produce the target JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
Figure 6. A: Optical image of the hydrate I seed crystal with the nucleated hydrate II crystals recovered from the seeded cooling crystallization (the cooling scheme used was scheme II in Fig. 1, the scale bar is 1000 mm), and (B) the corresponding Raman chemical image (the color scale represents the relative peak height of hydrate I (RPHI), which is calculated as: RPHI ¼ HI =HI þ HII where HI and HII are the characteristic peak height of hydrates I and II at 1173 and 1178 cm1, respectively).
hydrate I in the present work. This observation proved that the secondary nucleation and growth kinetics of hydrate I is extremely slow within the metastable zone, and therefore the consumption rate of supersaturation was not significant compared with the generation rate of supersaturation by cooling. This led to a continuous increase of the supersaturation in the solution and eventually reached the metastable zone limit, where spontaneous nucleation of hydrate II occurred. Therefore, the production of hydrate I requires alternative polymorphic control techniques, such as tailor-made additives,32 ultrasoundinduced nucleation33 and the appropriate crystallization parameters, for instance supersaturation level and water activity in the solution. This study revealed that the crystallization of two polymorphs of NF monohydrate follows the polymorphism theories of independent and DOI 10.1002/jps
CRYSTALLIZATION OF A POLYMORPHIC HYDRATE SYSTEM
competitive nucleation. The independent nucleation was demonstrated most clearly by evaporative crystallization, where both of the two hydrates were observed in all solvent systems. Nucleation of the two hydrates is also competitive, as the cooling crystallization shows. The small solubility difference between the two hydrates retarded the formation of the unstable hydrate (hydrate I), where also the stable hydrate (hydrate II) had faster nucleation and growth rates than hydrate I. The crystallization of hydrate II consumed all the supersaturation in the solution, which thus impeded the crystallization of hydrate I under all cooling conditions. According to the interpretation of the classical crystal growth theory, the growth of crystal faces consists of two steps: first the growth units are transported to the crystal surface from bulk solution by diffusion and convection, and then the growth units are incorporated into the crystal lattice through an integration reaction. In most cases, the second step is the rate-controlling, and thus determines the relative crystallization rate of the polymorphs. The surface integration rate is affected by not only the thermodynamic driving force of crystallization (the free energy difference between the crystal and liquid phases of the substance or the difference in chemical potential of the substance in saturated and supersaturated solution for crystallization from solution), but also other factors, such as the surface diffusion coefficient
761
of the solute molecules and the properties of the crystal surface. The classical nucleation and crystal growth theories are derived from anhydrate crystallization, and the free enthalpy change of solid phase formation is estimated for only solute molecules.34 The growth mechanism of hydrate can be expected to be more complicated than anhydrate, due to the incorporation of water molecules into the crystal lattice, where the interaction between the water and the solute molecules will have to be taken into account with estimation of the free enthalpy of solid phase formation. The chemical structure of NF anhydrate (top), and crystal structures of NF hydrates I and II characterized by hydrogen bonding (A and B, respectively) are shown in Figure 7. NF hydrate I is formed through hydrogen bond bonding between the N-H of NF molecule and the O atom (water). The H atoms of a water molecule are also hydrogen bonded with the O14 (carbonyl group) from two different NF molecules. These hydrogen bonding connections are indicated by the green dotted lines in Figure 1. Therefore, water in NF hydrate I forms three H bonds to neighboring molecules which is one of the two most common hydrogen bonding arrangements within hydrates according to Gillon et al.35 In NF hydrate II, the hydrate is also formed by the hydrogen bonding of N–H. . .O–H (water), where one H atom of a water molecule is linked to O15 (second NF hydrate).
Figure 7. Chemical structure of NF anhydrate (top); Crystal structure of nitrofurantoin polymorphic hydrate I (A) and II (B) characterized with H-bonding arrangements. DOI 10.1002/jps
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
762
TIAN ET AL.
Therefore, each water molecule in hydrate II is hydrogen bonded with two NF molecules. Although this is an unusual H-bonding motif, there also exists a series of other hydrogen bonds in NF hydrate II as suggested by Pienaar et al.19 The distances O (water). . .O (NO2), O (water). . . O5, O (water). . .O14, and O (water). . . N8 are all in ˚ , indicating weak the range 2.961(4)–3.211(4) A and possibly bifurcate hydrogen bonds. Intermolecular contact C7–H7. . .O15 also showed a geometry meeting the criteria for a hydrogen bond. NF hydrate II thus has a more complex hydrogen bonding scheme than hydrate I. Such complex hydrogen bonding scheme of hydrate II might then contribute to its higher stability, and leads to its more favorable crystallization.
CONCLUSION In this article, we have presented a phenomenon, namely, independent and competitive nucleation for the crystallization of two polymorphs of NF monohydrate. The mixtures of hydrates I and II produced from the evaporative crystallization suggested that the nucleation of hydrates I and II depends on both supersaturation level and the solvent composition in the solution (e.g., water activity). Although NF hydrate I is less stable than hydrate II indicated by its slightly higher aqueous solubility, seeding of the hydrate I failed to produce pure hydrate I in cooling crystallization experiments. This is due to the slow secondary nucleation rate of hydrate I. The production of the hydrate I of NF represents a challenging issue in polymorphism control, and it requires comprehensive understanding of the crystallization mechanisms combined with more robust polymorphic control techniques. It is expected that this crystallization mechanism can be utilized in various fields of chemical industry. With the growing interest in hydrate systems, we hope this work will stimulate more research into the understanding of the competing thermodynamic and kinetic factors that govern the crystallization of various hydrate polymorphs.
ACKNOWLEDGMENTS The authors thank Anne Zimmermann (Faculty of Pharmaceutical Sciences, University of Copenhagen, Denmark) for the collection of X-ray data, and Dr. Erik Skibsted (Novo Nordisk A/S, JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010
Denmark) for help with Raman measurements. Access to Raman microscopy from Novo Nordisk A/S, Denmark, and to X-ray powder diffraction provided by Lundbeck A/S, Denmark are also acknowledged. The Academy of Finland (project No. 122828) is thanked for financial support. The anonymous reviewers are thanked for their constructive criticism.
REFERENCES 1. Chow K, Tong HHY, Lum S, Chow AHL. 2008. Engineering of pharmaceutical materials: An industrial perspective. J Pharm Sci 97:2855–2877. 2. Gavezzotti A. 2007. A solid-state chemist’s view of the crystal polymorphism of organic compounds. J Pharm Sci 96:2232–2241. 3. Byrn SR, Pfeiffer RR, Stowell JG. 1999. Solid state chemistry of drugs. 2nd edition. West Lafayette, IN: SSCI Inc. 4. Brittain HG. 2009. Polymorphism and solvatomorphism 2007. J Pharm Sci 98: 1617–1642. 5. York P. 1983. Solid-state properties of powders in the formulation and processing of solid dosage forms. Int J Pharm 14:1–28. 6. Ostwald WZ. 1897. Studies on formation and transformation of solid materials. Z Phys Chem 22:289– 330. 7. ter Horst JH, Kramer HJM, Jansens PJ. 2002. A new molecular modeling approach to predict concomitant nucleation of polymorphs. Cryst Growth Des 2:351–356. 8. Bernstein J, Davey RJ, Henck J-O. 1999. Concomitant polymorphs. Angew Chem Int Ed 38: 3440–3461. 9. Blagden N, Davey RJ. 2003. Polymorph selection: Challenges for the future? Cryst Growth Des 3:873– 885. 10. Stoica C, Tinnemans P, Meekes H, Vlieg E, van Hoof PJCM, Kaspersen FM. 2005. Epitaxial 2D nucleation of metastable polymorphs: A 2D Version of Ostwald’s Rule of Stages. Cryst Growth Des 5: 833–1308. 11. Chen S, Xi H, Yu L. 2005. Cross-nucleation between ROY polymorphs. J Am Chem Soc 127:17439– 17444. 12. Desgranges C, Delhommelle J. 2006. Molecular mechanism for the cross-nucleation between polymorphs. J Am Chem Soc 128:10368–10369. 13. Mitchell C, Yu L, Ward MD. 2001. Selective nucleation and discovery of organic polymorphs through epitaxy with single crystal substrates. J Am Chem Soc 123:10830–10839. 14. Yu L. 2003. Nucleation of one polymorph by another. J Am Chem Soc 125:6380–6381. DOI 10.1002/jps
CRYSTALLIZATION OF A POLYMORPHIC HYDRATE SYSTEM
15. Haleblian JK, Koda RT, Biles JA. 1971. Isolation and characterization of some solid phases of fluprednisolone. J Pharm Sci 60:1485–1488. 16. Shefter E, Higuchi T. 1963. Dissolution behavior of crystalline solvated and nonsolvated forms of some pharmaceuticals. J Pharm Sci 52:781–791. 17. Khankari RK. 1993. Physicochemical characterization and thermodynamic properties of nedocromil salt hydrates. PhD Thesis: University of Minnesota, Minneapolis, MN. 18. van Tonder EC, Maleka TSP, Liebenberg W, Song M, Wurster DE, de Villiers MM. 2004. Preparation and physicochemical properties of niclosamide anhydrate and two monohydrates. Int J Pharm 269:417–432. 19. Pienaar EW, Caira MR, Lotter AP. 1993. Polymorphs of nitrofurantoin. I. Preparation and X-ray crystal structures of two monohydrated forms of nitrofurantoin. J Crystallogr Spectrosc Res 23: 739–744. 20. Aaltonen J, Strachan CJ, Pollanen K, Yliruusi J, Rantanen J. 2007. Hyphenated spectroscopy as a polymorph screening tool. J Pharm Biomed Anal 44:477–483. 21. Otsuka M, Teraoka R, Matsuda Y. 1992. Rotatingdisk dissolution kinetics of nitrofurantoin anhydrate and monohydrate at various temperatures. Pharm Res 9:307–311. 22. Kojima T, Yamauchi Y, Onoue S, Tsuda Y. 2008. Evaluation of hydrate formation of a pharmaceutical solid by using diffuse reflectance infrared Fourier-transform spectroscopy. J Pharm Biomed Anal 46:788–791. 23. Otsuka M, Teraoka R, Matsuda Y. 1990. Characterization of nitrofurantoin anhydrate and monohydrate, and their dissolution behaviors. Chem Pharm Bull 38:2667–2670. 24. Wikstro¨ m H, Rantanen J, Gift AD, Taylor LS. 2008. Toward an understanding of the factors influencing anhydrate-to-hydrate transformation kinetics in aqueous environments. Cryst Growth Des 8: 2684–2693.
DOI 10.1002/jps
763
25. Caira MR, Pienaar EW, Loetter AP. 1996. Polymorphism and pseudopolymorphism of the antibacterial nitrofurantoin. Mol Cryst Liq Cryst Sci Technol Section A Mol Cryst Liq Cryst 279:241– 264. 26. Otsuka M, Matsuda Y. 1993. Effect of environmental humidity on the transformation pathway of nitrofurantoin modifications during grinding and the physicochemical properties of ground products. J Pharm Pharmacol 45:406–413. 27. Kelly RC. 2003. A molecular approach to understanding the directed nucleation and phase transformation of carbamazepine and nitrofurantoin in aqueous and organic solutions. Ph.D Thesis, University of Michigan; MI, USA. 28. Nyvlt J. 1968. Kinetics of nucleation in solution. J Cryst Growth 3–4:377–383. 29. Tian F, Qu H, Louhi-Kultanen M, Rantanen J. 2009. Mechanistic insight into the evaporative crystallization of two polymorphs of nitrofurantoin monohydrate. J Cryst Growth 311: 2580–2589. 30. Beckmann W, Nickisch K, Budde U. 1998. Development of a seeding technique for the crystallization of the metastable a modification of abecarnil. Org Process Res Dev 2:298–304. 31. Beckmann W. 2000. Seeding the desired polymorph: Background, possibilities, limitations, and case studies. Org Process Res Dev 4:372–383. 32. Davey RJ, Williams-Seton L, Lieberman HF, Blagden N. 1999. Stabilizing a solid-solid interface with a molecular-scale adhesive. Nature 402:797– 799. ˚ C. 2005. 33. Gracin S, Uusi-Penttila¨ M, Rasmuson A Influence of ultrasound on the nucleation of polymorphs of p-aminobenzoic acid. Cryst Growth Des 5:1787–1794. 34. Mersmann A. 2001. Crystallization Technology Handbook. 2nd edition. New York: Marcel Dekker. 35. Gillon AL, Feeder N, Davey RJ, Storey R. 2003. Hydration in molecular crystals—A Cambridge structural database analysis. Cryst Growth Des 3:663–673.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 99, NO. 2, FEBUARY 2010