An Examination of the Solution Chemistry, Nucleation Kinetics, Crystal

May 18, 2009 - 43600, Bangi, Selangor, Malaysia, and Institute of Particle Science and .... 2854 Crystal Growth & Design, Vol. 9, No. 6, 2009. Anuar e...
1 downloads 0 Views 2MB Size
Published as part of a special issue of selected papers presented at the 8th International Workshop on the Crystal Growth of Organic Materials (CGOM8), Maastricht, Netherlands, September 15-17, 2008.

CRYSTAL GROWTH & DESIGN 2009 VOL. 9, NO. 6 2853–2862

An Examination of the Solution Chemistry, Nucleation Kinetics, Crystal Morphology, and Polymorphic Behavior of Aqueous phase Batch Crystallized L-Isoleucine at the 250 mL Scale Size Nornizar Anuar,*,§ Wan Ramli Wan Daud,‡ Kevin J. Roberts,† Siti Kartom Kamarudin,‡ and Siti Masrinda Tasirin‡ Faculty of Chemical Engineering, UniVersiti Teknologi MARA, 40450 Shah Alam, Selangor, Malaysia, Department of Chemical and Process Engineering, UniVersiti Kebangsaan Malaysia, 43600, Bangi, Selangor, Malaysia, and Institute of Particle Science and Engineering, School of Process, EnVironment and Material Engineering, UniVersity of Leeds, Leeds, LS2 9JT, United Kingdom ReceiVed February 5, 2009

ABSTRACT: An examination of the aqueous solution solubility and batch crystallization kinetics of L-isoleucine at the 250 mL scale size under a poly- and isothermal process condition is presented. Solubility data determined are consistent with the existence of two L-isoleucine polymorphic forms, in which both forms have different solubility and they are enantiotropically related. These polymorphs (A and B) can be recovered at different cooling rates of cooling crystallization. Crystal characterization using optical microscopy, differential scanning calorimetry, X-ray powder diffraction, and Fourier transform IR microscopy confirm this polymorphic behavior. Polythermal crystallization kinetic studies revealed the crystallization temperature increases with cooling rate and solute concentration, which results in a decrease of the metastable zone width (MSZW) with a decreasing cooling/heating rate. The study also revealed that cooling rates affect the polymorph formation, where at cooling rates of 0.25-0.75 °C/min, form B is formed, while a more stable polymorph A can be recovered at a cooling rate of 0.10 °C/min. Isothermal studies showed that the range of nucleation rate is between 1.79 × 10-5 and 7.53 × 10-4 kg/(m3 min), and the interfacial surface free energy at high and low supersaturation system is 1.74 and 0.576 mJ/m2, respectively. For a high supersaturation system, the critical cluster radius r* is between 5-17 Å, associated with between 3-121 molecules (N*), and for a low supersaturation system, r* is between 3 and 14 Å and N* is between 1 and 64. For a pH range of 5.1-6.3 and a temperature range between 10 and 80 °C, zwitterion species of L-isoleucine has remained as a dominating species in both solubility and crystallization studies. Thermodynamics properties generated from solubility data were also presented and discussed. Introduction Amino acids are high value added materials with significant importance in many applications in the pharmaceutical, food, and fine chemical industries. There are 20 naturally occurring R-amino acids whose structure is chiral, having a generic formula and R as the side chain, which determines the properties of individual amino acids such as structure, size, electric charge, and solubility in water. They are useful study materials because they are available in pure form, physically well characterized, moderately soluble in water, and comparatively easy to grow.1

Successful attempts at crystallizing amino acids ranging from industrial production to the growth of single crystals have been * Corresponding author. Telephone: +60 3 5543 6315. Fax: +60 3 5543 6300. E-mail: [email protected]. § Universiti Teknologi MARA. ‡ Universiti Kebangsaan Malaysia. † University of Leeds.

previously reported. Nevertheless, reports on crystallization conditions of amino acids in the absence of additives have received less attention, even though precipitation and crystal growth in pharmaceutical industries are frequently performed in the absence of impurities.2 One of the reasons for this is the growing interest in modifying the original shape of crystals for crystal shape perfection and obtaining desired polymorphic crystals. Studies on amino acids have been long carried out by many previous researchers. Among them are on glycine, the simplest form of amino acid,3-7 L-glutamic acid,8-10 Lhistidine,11 DL-methionine,12 and L-threonine.13 Determination of amino acids morphology, intermolecular interaction, and morphological prediction have also been carried out using molecular modeling, for example, DL-methionine,14 L-glutamic acids,15,16 glycine,17 and L-arginine salt.18 Nevertheless, there have not been many studies on L-isoleucine crystallization, for example, resolved molecular,23 solubility,24 effect of L-leucine and L-valine on L-isoleucine morphology,25 morphology prediction,26 and Raman spectra pattern determination of L-isoleucine.27 In this work, L-isoleucine was chosen as a model compound for this fundamental study reflecting its interesting crystal chemistry and its varied surface properties. It is part of a branch

10.1021/cg900133t CCC: $40.75  2009 American Chemical Society Published on Web 05/18/2009

2854

Crystal Growth & Design, Vol. 9, No. 6, 2009

Anuar et al.

Figure 1. Molecular structure of L-isoleucine showing view from (a) the y-direction showing a hydrogen bond network between unit cell lattices; (b) view from the z-direction; and (c) view from the x-direction in a unit cell.

chain amino acid (BCAA) and plays an important role particularly during exercise and the maintenance and growth of skeletal muscle.19 It also shows clear polymorphism characteristics due to its potential conformational change due to aliphatic structural rotation, but until now, to our best knowledge, L-isoleucine polymorphism has yet been discovered. The first structure of 20 L-isoleucine suggested by Khawas was orthorhombic with space group P2221. Later, Torii and Iitaka21 presented more accurate atomic coordinates and suggested L-isoleucine crystallizes in P21 with monoclinic structure. However, both of these structures have identical powder diffraction data, which indicate that these two materials are of the same form. L-Isoleucine structure redetermination at 120 K by Go¨rbitz & Dalhus22 shows the same crystal parameters as reported by Torii and Iitaka,21 but with more accurate standard deviations (0.002-0.003 Å) for bond lengths between heavy atoms compared to the later with standard deviations in the range 0.011-0.018 Å. Factors controlling the crystallization process such as environment, nucleation, and growth kinetics28 can be controlled by monitoring solution supersaturation and metastable zone width (MSZW),29 which then results in improved crystal properties. Effective nucleation and growth can be achieved if a stable enough region for crystal growth is provided,30 and greater solution stability can be achieved for a growth solution having higher MSZW. However, too large of MSZW could result in a large growth barrier where nucleation may take place at high supersaturation resulting in high nucleation rates and concomitant small crystal size distribution.31 MSZW is also helpful to measure the onset of nucleation under isothermal conditions, that is, through measurement of an induction time, defined as the time elapsing between the creation of supersaturation and the formation of critical nuclei prior to bulk crystallization. Several important crystallization operating variables such as supersaturation, temperature, pH, agitation speed, and impurities are found to have an influence on the induction time.32-35 This paper reports on examination of the crystallization of L-isoleucine associated with measurement of the influence of solubility, solute concentration, and cooling rate on the nucleation kinetics (nucleation rate, interfacial surface tension) and resulting solid form. Materials and Methods Material. L-Isoleucine (C6H13O2N, MW of 131.2 (99.6% purity)) was obtained from Merck. Distilled water was used to make up the solution. Crystal Structure of L-Isoleucine. L-Isoleucine crystallizes in a monoclinic lattice with space group P21, Z ) 4 and cell parameters a

) 9.6810, b ) 5.3010, c ) 13.9560 Å, and β ) 96.1600 °. The molecules exist in two different conformations (molecules A and B) in a unit cell, hence having two molecules per asymmetric unit.21 Molecule A and molecule B rotate about C2 and C3 with gauche I and trans rotation, respectively, which results in different atomic R group positions in the unit cell, atomic bond lengths, and angle conformations. L-Isoleucine molecules are packed alternately between molecules A and B in the crystal lattice structure as shown in Figure 1. The different conformations for molecules A and B within the lattice structure result in different hydrogen bond motifs between the molecules as well as between the asymmetric pairs within the unit cell in the crystal packing structure.35 In a unit cell, molecules A and B in one of the asymmetric unit pairs is bonded together by a hydrogen bond between the amino hydrogen of molecule A and the carbonyl oxygen of molecule B. On the other hand, no hydrogen bonds exist between molecules A and B of the other asymmetric unit pair. In the x-axis direction, one of the amino hydrogens of molecule B forms two hydrogen bonds with carbonyl oxygen of molecule A from the neighboring unit cell forming a bifurcated H bond. On the other hand, each of the amino hydrogen atoms of molecule A forms hydrogen bonds with carbonyl oxygen atoms of molecule B forming a tetrahedral shaped hydrogen bond between them. The network of hydrogen bonds parallel to the plane of x- and y-axis results in a hydrophilic plane of symmetry. Molecular structures of L-isoleucine are arranged in the lattice unit cell creating hydrophilic and hydrophobic, where hydrophilic regions are involved in a hydrogen bond network, while the hydrophobic regions experience van der Waals interactions only.23,37 L-Isoleucine Solubility Determination. Solubility data of L-isoleucine in water was determined over the temperature range between 10 and 80 °C by placing the sample in a 250 mL jacketed reactor at the desired temperature. Excess L-isoleucine was added to a fixed amount of water and the sample was agitated at the desired temperature for 1 h. The temperature of the solution was held constant for 15 min, then raised to 15 °C above the desired equilibrium temperature, cooled to the desired temperature, and allowed to equilibrate for 4 h. Then, liquid samples were withdrawn from the reactor, filtered using a 0.20 µm filter, weighed, and dried in the oven at a temperature of 40 °C. The final weight of dried L-isoleucine was noted. Study of material solubility is important as it reflects molecular ordering in the precursor solution. Transformations from solution to solute molecules are dominated by intermolecular coordination with solvent molecules to the solid-solid state where strong solute-solute interactions lead to molecular self-assembly and a three-dimensional crystallographic structure. An ideal solubility can be expressed by

ln(x) )

[

∆Hf 1 1 R Tf T

]

(1)

where x is the mole fraction of solute in the solution, T is the solution temperature, Tf is the fusion temperature (melting point) of the solute, ∆Hf is the molal enthalpy of fusion, and R is the gas constant (8.314 J/mol · K). This ideal equation involves perfect mixing on the molecular scale of solute and solvent. In reality, solubility is solvent-dependent and can be described by the van’t Hoff’s equation; thus

Aqueous Phase Batch Crystallized L-Isoleucine

∆S ∆H R RTsat

ln(x) )

Crystal Growth & Design, Vol. 9, No. 6, 2009 2855

(2)

where ∆H and ∆S are the enthalpy of dissolution, respectively, Tsat is the saturation temperature, which is defined as the temperature of dissolution at an infinitely slow rate and determined from the y-intercept of the plot of dissolution temperature against the heating rate. The known Tsat then be used to calculate ∆H and ∆S from which the Gibbs free energy; ∆G for this process were calculated using eq 3.

∆G ) ∆H-T∆S

(3)

Deviation of this system from ideality can be measured using excess Gibbs free energy, ∆Gexcess and activity coefficient, γa.

∆Gexcess ) ∆G-∆Gideal

(4)

( ∆SR - ∆H RT )

(5)

xγa ) exp

Solid-State Characterization. The product crystals were characterized using differential scanning calorimetry (DSC) (Metler Toledo DSC 820) by heating about 2 mg of sample at the rate of 10 °C/min and nitrogen purging. The X-ray powder diffraction (XRPD) were determined using a Bruker D8 advance diffractometer with Cu KR radiation at 40 mA, 40 kV, 2θ between 3° and 40° and rate of 0.01°/s. Meanwhile, the hydrogen bonds interactions between the molecules were determined using Fourier transform infrared (FTIR) microscopy (Thermo Nicolet) with a range of wavenumbers from 500 to 4000 cm-1. Crystal morphologies were observed by Meiji Techno 1559, using Zabeco software. Polythermal Nucleation. Metastable Zone Width Determination. In these crystallization experiments, the cooling and heating cycles of aqueous L-isoleucine were carried out repeatedly at known rates while visually observing the Tdiss and Tcryst. At the beginning of each crystallization cycle, the temperature was raised to approximately 15 °C above the saturation temperature and held for at least 30 min to ensure that all the embryos were fully dissolved. Cooling and heating rates used in this experiment were 0.10 °C/min, 0.25 °C/min, 0.50 °C/ min, and 0.70 °C/min. The concentrations of L-isoleucine used in these experiments were 40, 42, 44, 46, and 48 g/L. Isothermal Nucleation. L-Isoleucine in distilled water at a known concentration was predissolved and heated at temperature 15 °C higher than the saturated temperature for 60 min. The solution was then slowly cooled (at the rate of 0.2 °C/min) to the saturation temperature and maintained for 60 min. Then, the solution was drastically cooled (at the rate of 1.0-2.0 °C/min) to the desired temperature, T and kept constant at that temperature until the onset of nucleation. The induction time, τ was recorded as the time from the point at which the constant temperature, T was reached until the crystal formed as seen by visual observation. Solution pH Determination. The pH of a known concentration of L-isoleucine solution was measured during the crystallization process at a cooling rate of 0.10 °C/min. The solution was placed in a 250-mL reactor, which was fitted with a pH meter, a turbidity probe, and a thermocouple. For the turbidity probe, 100% transmittance signal corresponds to clear solution and 0% transmittance signal corresponds to a turbid solution. Sudden signal drops from 100 to 0% transmittance as the solution was cooled indicate the formation of crystals in the solution and was taken as point of crystallization. The background theory and details underpinning methods used in this work is defined in Supporting Information.

Results and Discussion Characterization of Solution State. The van’t Hoff plots of an ideal system and of L-isoleucine calculated using eqs 1 and 2 respectively are shown in Figure 2. The values of enthalpy of fusion, ∆Hf, and the melting temperature Tm used in eq 1 for an ideal system were found to be 89.78 kJ/mol and 288.6 °C, respectively. The L-isoleucine solubility was found to be close to the data reported previously24 and is found to be substantially higher than the ideal solubility plot, which indicates strong interactions between L-isoleucine and water molecules.38 The van’t Hoff

Figure 2. (a) van’t Hoff solubility data of L-isoleucine (ideal solubility, form A, form B), showing negative deviation of real system from ideality.

plot behavior is also consistent with the existence of a breaking line in the solubility-temperature plot at 45 °C (Tr), (shown in Figure 2) suggesting the existence of a second polymorphic form of L-isoleucine. Another branch chain amino acid, L-leucine, is suggested to have formed a second polymorph at 80 °C.39 The polymorphs were characterized using Raman spectroscopy, and it was suggested that the second polymorph is formed due to the conformational rearrangement of the structure. However, the molecular structure of the second polymorph was not resolved. Since L-isoleucine and L-leucine have almost similar molecular structures and configurations, then there is also some possibility the phase transformation of L-isoleucine could take place. Referring to Figure 2, the more stable polymorph (form A) has a solubility line at a temperature below 45 °C and the metastable polymorph (form B) has a solubility line higher than 45 °C. These two lines, which intersect at 45 °C, then generate two sets of enthalpy, ∆H and entropy, ∆S values, for both of the polymorphs (Table 1). The enthalpies of dissolution appear to be an endothermic process, while the entropies for both polymorphs show negative values. The contribution of positive values of overall ∆H come from the endothermic reaction of breaking solute hydrogen bonds to create a cavity for the solutes with water, and this reaction outweighs the exothermic reaction of hydrogen bond formation of carboxyl and amino group with water. This phenomenon is not uncommon for dissolution of nonpolar amino acids in water as determined by previous researchers.40,41 The negative ∆S and positive ∆G values are presumably due to the water ordering effect (clathrate) in the presence of a nonpolar material in polar water molecules.42 In a case like this, water molecules tend to form a cagelike structure around the solute molecules and are forced to form a more ordered arrangement around the hydrocarbons, which then results in an appreciable decrease in entropy and outweighs the increase in the process entropy of mixing. Lizhuang et al.43 in their work with solubility of tert-butyl alcohol in water speculated that the quantity of clathrate depends mainly on composition and temperature, proposing that the higher the temperature the more difficult is the formation of clathrate. This phenomenon probably is the explanation for the significant increase of ∆S value for form B, as shown in Table 1. Calculated activity coefficients for both forms are near to 1, which most of the time is an indication of an ideal system, but this assumption can be ruled out due to the negative values of ∆Gexcess. A system is regarded as an ideal system if it has both ∆Gexcess ) 0 and γa ) 1.44 Orella and Kirwan45 suggested that

2856

Crystal Growth & Design, Vol. 9, No. 6, 2009

Anuar et al.

Figure 3. (a) Plot of Gibbs free energy of forms A and B showing the transition temperature Tr and the crossing of lines suggesting they are enantiotropically related; (b) speciation diagram of L-isoleucine at 25 °C; the arrows show the pKa values of the species (pKa1 ) 2.36 and pKa2 ) 9.59).

Figure 4. Relationship between pH and variation of species concentration for (a) [RH°] and (b) [RH+] and [RH-].

apart from their zwitterionic behavior, low amino acid solubility in water causes the activity coefficient to be near unity, due to its insensitivity to the concentration of solute and strongly impacted by the solvent composition. They showed that glycine and L-alanine, which is known to have high solubility in water, only produce activity coefficients of slightly higher than unity. The negative values of mean ∆Gexcess support the fact suggested by Davey et al.38 that this system exhibits negative deviation from ideal behavior, indicating strong solute-solvent interaction. Here it is suggested that there are strong interactions between the hydrophilic head of the molecules (consisting of -NH3+ and -COO-) and the surrounding water molecules, this despite the solute’s sparingly soluble behavior in water. The plot of Gibbs free energy with temperature (Figure 3a) proves that the polymorphs are enantiotropically related,46 and the intersection point between the two lines represents the transition temperature, Tr, that is, T ≈ 45 °C. Referring to Figure 3b, L-isoleucine exists in three forms, namely, [RH+], [RH°], and [RH-] where they represent the total molecule’s charge carried in the solution, depending on the pH of the solution due to intramolecular proton transfer. The pKa values determined in this work are 2.36 and 9.59 at T ) 25 °C, corresponding to the pH interception point indicating control of dominating species in acidic and basic solution’s environment, respectively. The variation of L-isoleucine species with pH and temperature of solubility are shown in Figure 4a,b. As temperature of the solution increased, the solution’s pH was found to decrease from 6.3 to 5.6 due to the increase of solute concentration. As a dominating species at this

Table 1. Calculated Thermodynamic Properties from Solubility Data

Form A Form B ideal

∆H (J/mol)

∆S (J/molK)

∆Gmean (J/mol)

∆Gexcess, mean (J/mol)

activity coef, γa (-)

4071 10916 89783

-31 -9 160

13232 14009 39100

-29081 -21878 0

1.00 1.02 1.00

pH range, [RH°] concentration increases with temperature as at a higher temperature more solute is dissolved. The result also shows that there is an appreciable decrease of pH value of 0.25 with every 1 g/L of L-isoleucine dissolved in the solution. Increasing the solution temperature was found to lower the solution’s pH, producing more [H+] ions in the solution to cause dissociation of [RH-] to [RH°] and hence to [RH+]. It is expected that [RH-] should decrease and [RH+] increase with increasing [H+] as the temperature rises. However, the result shown in Figure 4b is unexpected, whereby [RH-] species increases and [RH+] decreases as temperature increases, indicating perhaps that for this pH range and concentration, the species dissociation is more temperature-dependent rather than pHdependent. The result suggests that increasing temperature promotes the dissociation of [RH+] to [RH°], and [RH°] to [RH-]. Comparison of ∆H and ∆S between the three ionic species for the two temperature ranges characterizing the two polymorphic forms (Table 2) shows that the properties for species [RH°] of both forms are almost identical to the properties of solute concentration (in Table 1) due to the fact that [RH°] species is

Aqueous Phase Batch Crystallized L-Isoleucine

Crystal Growth & Design, Vol. 9, No. 6, 2009 2857

Figure 5. (a) XRPD profiles for form A and B; inner box shows the similarities of (001) and (002) peaks for 2θ ) 6.37° and 12.68°, respectively, suggesting no substantial change in crystal structure and (b) DSC result showing the phase transformation temperature (267.7 °C) of form B to form A, and the melting point (288.6 °C) of form A.

the dominating species. It is also interesting to note that for both these temperature ranges, the [RH+] is an exothermic dissolution process while all the other species were endothermic. In the solid state, there is only [RH°] present and, upon dissolving in water, the breaking of lattice hydrogen bonds involves only these species. In solution, [RH°] dissociates to [RH+] by attaching [H+] water ions to carboxyl -COO- head forming a hydrogen bond, which is an exothermic process. Nevertheless, for the [RH+] species there is little difference in ∆H and ∆S between the two temperature ranges, and thus it can be concluded that the [RH+] contribution to the polymorphic transformation is not significant. In contrast, for the [RH°] and [RH-] species, there are significant differences. As temperature increases, [RH°] molecules dissociated to [RH-] as portrayed by the ∆H of [RH-]. However, the values are small as the [RH-] concentration is very much lower compared to [RH°]. Nonetheless, the negative values of ∆S is consistent with the formation of a clathrate structure within the solution, as discussed previously. Characterization of Polymorphs in Solid State. Crystal characterization using XRPD provides supporting evidence for the presence of the new crystal form (Figure 5a). Form A was found to have a profile similar to that simulated from the data in the Cambridge Crystallographic Data Centre (CCDC) (ref code: LISLEU) from Torii and Iitaka.21 Form B was characterized by three new peaks (2θ ) 19.55°, 19.68°, 25.56°), and six missing peaks (2θ ) 19.85°, 20.09°, 22.50°, 23.51°, 25.28°, 27.81°) compared to form A. The 2θ ) 6.37 ° and 12.68 ° peaks which correspond to (001) and (002) respectively in Form A (as shown in the inner box) were not found to significantly change suggesting that the two forms had quite similar c-axis lattice dimensions. Note also that the feed material used in this study prior to crystallization was form A. The XRPD is also supported by DSC results (Figure 5b), where form B shows two peaks while form A has only one (endothermic) peak with the first peak at T ) 267.7 °C corresponding to the enantiotropic transformation of form B to form A with both DSC peaks showing the common melting point of form (Tm ) 288.6 °C). DSC result of the feed material also shows the profile of the stable form and hence confirms the XRPD data. Generally, L-isoleucine has a flat elongated hexagonal shape, with angles of about 120° each (Figure 6a). As shown by Figure 6b,c, both of the polymorphs are not readily distinguishable by their morphology as they are related by 2-fold crystal symmetry. However, for the benefit of identifying the difference between the polymorphic forms, the angles were identified in which both of the side angles are named φ and δ, and the apex angle is ω.

Figure 6. (a) Sketch of the shape of L-isoleucine crystal showing the measured internal angles, (b) crystals of form A and (c) form B. Table 2. Thermodynamic Properties of Species at Two Temperature Ranges T < 45 °C +

[RH ] [RH°] [RH-]

T > 45 °C

∆H (J/mol)

∆S (J/mol K)

∆H (J/mol)

∆S (J/mol K)

-11400 4100 1200

-154 -31 -105

-12400 11000 10800

-157 -9 -75

Summary of their average angles measured from sample size ca. 60 are shown in Table 3. The result shows that form A has a wider apex angle (ω), compared to form B while the other two sides angles (φ and δ) for both forms are almost similar, and we believed the differences are due to experimental error. From the standard deviation for angles φ and δ, it suggests that form A has a wider distribution of angles compared to form B. Figure 7 further supports the above findings, where the histogram shows that both forms φ (and also δ) are almost the same, while apex angle ω of form A has a wider angle than form B. The crystals also were characterized using FTIR microscopy, and the result shows that form B has a new peak at wavelength 1038 cm-1 and a shifted peak at wavelength 968 cm-1 from the original peak of form A, supporting the presence of two different polymorphic forms. The crystals also were characterized using FTIR microscopy, and the result shows that form B has a new peak at wavelength 1038 cm-1 and a shifted peak at wavelength 968 cm-1 from the original peak of form A, supporting the presence of two different polymorphic forms. Polythermal Crystallization: Nucleation Mechanism. Examination of the polythermal data (Figure 8) reveals that for all heating rates, the temperatures of dissolution (Tdiss) were almost constant, while the temperatures of crystallization (Tcryst) increased with decreasing cooling rates. This shows that the dissociation behavior is not affected significantly by the solution heating rate in contrast to crystallization on-set where this material’s low nucleation rate affects a greater undercooling for the faster cooling rate. Figure 8 also shows that supersaturation ratio at the on-set of crystallization, Scrit increasing with

2858

Crystal Growth & Design, Vol. 9, No. 6, 2009

Anuar et al.

Table 3. Measured Internal Angles of Two Polymorphic Structuresa

Form A Form B a

φmean (°)

standard deviation, σ

ωmean (°)

standard deviation, σ

δmean (°)

standard deviation, σ

120.0 120.1

4.9 2.7

128.1 121.8

3.7 2.5

118.1 118.5

4.6 2.7

The angles are as defined by Figure 6a. Table 4. The Change of Polymorphic Forms, the Critical Supersaturation at the Onset of Nucleation and Critical Radius Prediction Based upon Homogeneous Nucleation Theory with Cooling Rates 44 g/L

Figure 7. Statistical analysis of morphological internal angles of the two polymorphic forms in the range 101 to 125°; the majority of φ angles of both forms lie in the same angle range (106-110°), while form A apex angle has a wider angle (range 116-120°) than form B (111-115°).

Figure 8. Crystallization and dissolution of L-isoleucine (Tdiss (48 g/L), Tdiss (40 g/L), Scrit values represent the supersaturation at nucleation and the vertical dotted line on the graph suggesting the intersection of formation of different polymorphic forms due to the cooling rate.

increasing cooling rate, which means that at the crystallization temperature, fewer nuclei formed for the lower cooling rates. Comparing the two extreme concentrations at a higher cooling rate, concentration 48 g/L crystallizes ata higher supersaturation and hence higher nucleation rate compared to a concentration of 40 g/L. On the other hand, at a rate of about 0.26 °C/min and below, the behavior reversed where at the point of crystallization, the concentration at 48 g/L crystallizes at a lower supersaturation. This is due to dependency of critical supersaturation; Scrit on cooling rates increases with solution concentration, and thus the differences in slopes of these two concentrations create the Scrit intersection point. This trend is expected as Scrit is directly related to crystallization temperature Tcryst, and the result also shows that Tcryst decreases at a higher rate with cooling rate for high solution concentration. Giron47 stated that at the beginning of the crystallization, depending on the supersaturation and the solubility curves of each polymorph, and due to thermodynamic reasons, the first crystal is formed

cooling rate (°C/min)

form

Tcryst

Scrit

r*crit

0.70 0.50 0.25 0.10

B B B A

41.4 44.2 48.7 51.4

1.15 1.13 1.08 1.05

4 5 7 11

as the most soluble form and then is converted to the less soluble form. Crystals recovered from a concentration of 44 g/L were analyzed using XRPD, and it was found that cooling rate does affect the formation of the polymorphic form of L-isoleucine. Crystallization with cooling rates between 0.25 to 0.70 °C/min produces form B polymorph, while form A polymorph was produced in cooling crystallization of 0.10 °C/min (Table 4). From Figure 8, there is a possibility that the Scrit slopes interception show an indication of different regions of polymorphic forms. However, the analysis of crystals recovered for all the solution cooling rates and concentrations were not carried out. It is interesting to note the dependence of crystallized form (i.e., A or B) on the cooling rate, with the metastable form B crystallizing at faster rates in accordance with the expectation of Ostwald’s rule of stages.48 Calculation of the critical supersaturation (Scrit) for nucleation reveals that form B crystallizes at a supersaturation >1.08. Drawing up calculation of interfacial surface energy, γT from induction time experiments (which will be discussed in this paper later), reveals cluster size r* < 7 Å results in form B. Reflecting on recent modeling studies,16 it is attractive to conclude that at those small cluster sizes, the metastable form is stabilized, perhaps reflecting the greater conformational freedom/disorder at those smaller sizes, but further work is obviously needed to clarify and support this suggestion. Figure 9a shows the dependency of metastable zone width (MSZW) on solution concentration. The MSZW values calculated range between 10 to 30 °C, where they represent the most minimum and maximum values for concentrations 48 and 40 g/L, respectively. These measured values are wider than other amino acids L-threonine13 and glycine salt49 but about the same range with L-arginine phosphate.50 In industrial crystallization, for a better control of product size wider MSZW are required. However, to keep optimum level of supersaturation, the process should ideally be operated at half way between the extremes of the MSZ to both achieve the required growth rate and to avoid the generation of fine crystals due to homogeneous nucleation at the MSZ boundary.51 The trend of relationship between MSZW and concentration are the same for all cooling rates, where MSZW decreases as cooling rates increases, regardless of polymorphs produced. Increasing the solution concentration also results in decreasing order of reaction, m. This shows that nucleation is easier to form at higher solution concentration and the contribution of cooling rates on the ability to nucleate becomes smaller. Nevertheless, at any concentrations, nucleation is easier to achieve if the solution is cooled at much slower rates. Good dependency of MSZW on cooling rates also

Aqueous Phase Batch Crystallized L-Isoleucine

Crystal Growth & Design, Vol. 9, No. 6, 2009 2859

Figure 9. (a) The dependency of MSZW at different cooling rates (key, a: 0.7 °C/min, b: 0.5 °C/min, c: 0.25 °C/min, d: 0.10 °C/min, e: at infinitely small rate ≈ 0 °C/min, f: order of reaction) on concentration. Higher MSZW indicates better control of nucleation, while lower order of reaction means easier nucleation formation. (b) Dependency of nucleation rate and nucleation rate constant with L-isoleucine solution concentration.

Figure 10. Turbidity and pH profile of L-isoleucine at concentration 44 g/L during cooling crystallization: (a) turbidity reading at 0.10 °C/ min, (b) pH of L-isoleucine solution at 0.10 °C/min, and (c) pH of water only; showing at higher temperature, the pH of the solutions exhibits the same behavior as water solution where the pH first decreases before it increases as the solutions were cooled down.

implied that the process was very much dependent on nucleation kinetics.52 Both nucleation rate J and rate constant kn were found to increase with concentration (Figure 9b), and this result supports the findings summarized in Figure 8. The results are expected as in high solution concentration there is an abundant fresh supply of material and a system with high supersaturation produces higher nucleation rate, since supersaturation is a driving force of phase transformation. Figure 10 shows the influence of the solution pH during crystallization for a concentration 44 g/L and a cooling rate of 0.10 °C/min. The result shows that the solution pH increases upon crystallization. However, at the region between 80 to 70 °C, the pH of the solution decreases as the temperature decreases, and this can be correlated with the behavior of water as demonstrated by experiments carried out by measuring the pH of water as it was cooled down from 80 to 20 °C (Figure 10c). Nevertheless, at T about 72 °C, the pH starts to increase (Figure 10b) and the crystals were noted start to form at T ) 51 °C (shown by arrow in Figure 10a). The formation of L-isoleucine species with pH during crystallization is shown in Figure 11a,b, whereby the amount of dominant [RH°] and [RH+] species reduces with increasing solution supersaturation and pH. However, the [RH-] shows a slight increase, but the amount of

the species is very small. L-Isoleucine crystals are packed from zwitterion molecules and reduction in the number of this species is expected. Meanwhile, the [RH+] and [RH-] species do not take part in the formation of crystals remaining in the solution and dissociate to other respective species as pH of the solution, as shown by the speciation diagram (Figure 3b). The dependency of [RH°] species on pH only took place from the point of crystallization onward, which shows that the influence of solution concentration on [RH°] is higher than pH as the concentration of [RH°] was not expected to change before the formation of crystals. Isothermal Crystallization: Nucleation Mechanism. Figure 12 shows the relationship between the induction time and both temperature and supersaturation, and it is obvious that the relationships for each concentration are made of two straight lines with different slopes. The two lines were distinguished based on the system nucleating at low and high supersaturations. The result shows that shorter induction time was recorded as supersaturation increases and this is probably due to higher solution supersaturation effectively reduces the size of critical nucleus and hence shortern the induction time. It was also found that at higher solution concentration, the onset of crystallization occurs at a shorter induction period. Higher solution concentrations ensure more supply of fresh solution and results in a prolonged lifetime of nuclei, which then enhances the crystal growth. Since this experiment is a fast cool crystallization and based on the conclusion made in polythermal experiment, it is most likely that the crystals formed in this work is of form B crystals, which then confirmed by the XRPD analysis of crystals recovered at S ) 52.4 °C) and 1.08 ) 1.04 (Tequilibrium (Tequilibrium ) 48.4 °C) from a concentration of 44 g/L. However, not all crystals recovered from all conditions in this work (i.e., concentrations and supersaturations) were tested. Treatment of data at different supersaturation systems results in two sets of interfacial surface energy and preexponential factor as shown in Figure 13. The result shows that the interfacial surface energy for this system is not affected by the change of concentration, but the value generated by high supersaturations system is higher than the low supersaturations system. The average value of interfacial tension for both high and low supersaturations system is 1.74 and 0.576 mJ/m2 respectively and is in order of magnitude expected for sparingly soluble salts,53 even though the γT

2860

Crystal Growth & Design, Vol. 9, No. 6, 2009

Anuar et al.

Figure 11. The plot of pH and L-isoleucine species profile during crystallization for concentration 44 g/L at a cooling rate of 0.10 °C/min, showing the formation of dominating [RH°] species over [RH+] and [RH-] as solution temperature and pH changed.

Figure 12. Dependency of induction time on supersaturation, the two sets of lines for each concentration distinguishing system nucleating at low and high supersaturations.

value for low supersaturations determined in this work falls at the lower end of the range. Meanwhile, for both supersaturations, the pre-exponential factor A was found to increase with supersaturation (Figure 13b). Nevertheless, the A value for concentration 42 g/L of high supersaturation systems shows a deviation from this trend. The pre-exponential factor indicates the probabability of molecular collision in the mother liquor during the nucleation process. Higher solution

concentration results in more molecules present in the solution and hence produces higher probability of molecular collisions. Higher number of nuclei formed as supersaturation increases is associated with increasing nucleation rate.54 At higher supersaturation, the nuclei can be expected to become stable and grow faster due to the reduction of free energy barrier for the formation of critical radius size. Figure 14a,b shows the increase of cluster critical radius r* and its number of molecules N* with temperature, while both of the functions decrease with increasing supersaturations (for both low and high supersaturations). This result is consistent with the classical theory of homogeneous nucleation stating that the size of a critical nucleus increases with temperature.55 For solute concentration of 48 g/L (Figure 14) for high supersaturations system, the size critical cluster radius r* is between 14-17 Å, associated with between 67 to 121 molecules, while for low supersaturations system the values of r* lie between 8 and 14 Å with number of molecules associated to it being 14-64. However, for the working concentration range in this work, the overall calculated values of r* is between 5-17 Å for high supersaturations system, associated with between 3 to 121 molecules while for low supersaturations system, the values of r* lies between 3-14 Å with the number of molecules associated to it being 1-64. The lower values of both r* and N* were contributed by low supersaturations system, and it is obvious that they were underestimated, which suggests the possibility of a hetero-

Figure 13. The plot of variation of (a) the interfacial surface energy; and (b) pre-exponential factor with concentration. Both the high and low supersaturation systems result in two sets of interfacial surface energy and pre-exponential factor.

Aqueous Phase Batch Crystallized L-Isoleucine

Crystal Growth & Design, Vol. 9, No. 6, 2009 2861

Figure 14. The dependency of (a) r* and (b) N* on temperature and supersaturation ratio at concentration 48 g/L, showing the plots separating high and low supersaturations.

geneous rather than that of a homogeneous nucleation process.35 Heterogeneous nucleation can take place with a lower energy barrier compared to the homogeneous case with crystallization carried out under heterogeneous nucleation mechanism producing a narrower MSZW in comparison to the homogeneous nucleation case,56 and the significant reduction of MSZW indicates the former needs lower energy barrier to nucleate. Liu57 suggested that at low supersaturations, heterogeneous nucleation will be dominant and at high supersaturations, homogeneous nucleation will occur. In this, the high energy barrier at low supersaturations promotes heterogeneous nucleation in order to affect a lower nucleation barrier, in comparison to the higher supersaturation the energy barrier is readily low and thus homogeneous nucleation is favored. However, he also suggested that most of the time even at high supersaturations, heterogeneous nucleation can take place, which has been often falsely identified as homogeneous nucleation. Conclusions The van’t Hoff plot of solubility reveals a breaking point at 45 °C, suggesting existence of the second polymorph of L-isoleucine due to the different solubility of the two enantiotropically related polymorphic forms, with form A more stable at lower temperatures than the new form B and vice versa. This finding is consistent with XRPD, FTIR microscopy, DSC, and morphological analysis. The solubility study also suggests that the system experiences negative deviation from ideal solution behavior implying enhanced solute/solvent interaction. Polythermal crystallization studies of the mixture of these polymorphic forms reveal that as the concentration of L-isoleucine increases, the rate of nucleation increased while the MSZW and order of reaction decreased. Crystallizing L-isoleucine in a low cooling rate results in formation of form A and high cooling rate produces form B. The zwitterion [RH°] species remain as dominating species in solution regardless of increasing solute concentration and temperatures. From isothermal nucleation studies, it was found that the interfacial surface energy for this system is not affected by the change of concentration while the preexponential factor increased as solute concentration increased. The critical cluster radius r* and the associated number of molecules per cluster N* were found to increase with increasing temperature and decreasing supersaturation ratio. The behavior of low supersaturation crystallization is more

consistent with a heterogeneous nucleation mechanism in contrast to higher supersaturation where homogeneous nucleation is more likely. Acknowledgment. The authors would like to express their gratitude to the Ministry of Science and Technology, Malaysia, for funding of this work which was carried out in collaboration with the Institute of Particle Science and Engineering, at the University of Leeds. One of us (N.A.) would like to thank Universiti Teknologi MARA for the sponsorship of this Ph.D. study. This work is part of Grant 02-02-02-0001 PR0023/1106 and Grant 100-IRDC/SF 16/6/2(66/2007). Supporting Information Available: This material is available free of charge via the Internet at http://pubs.acs.org.

References (1) Black, S. N.; Davey, R. J. J. Cryst. Growth 1988, 90, 136. (2) Cashell, C.; Corcoran, D.; Hodnet, B. K. Cryst. Growth Des. 2005, 5, 593. (3) Kunihisa, K. S. J. Cryst. Growth 1974, 23, 351. (4) Sakai, H.; Hosogai, H.; Tsukamoto, T. J. Cryst. Growth 1992, 116, 421. (5) Li, L.; Rodrı´guez-Hornedo, N. J. Cryst. Growth 1992, 121, 33. ´ .; Hofland, G. W.; Crommelin, (6) Bouchard, A.; Jovanovı´, N.; Martı´n, A D. J. A.; Jiskoot, W.; Witkamp., G.-J. J. Supercrit. Fluids 2008, 44, 409. (7) Liu, Z.; Zhong, L.; Ying, P.; Feng, Z.; Li, C. Biophys. Chem. 2008, 132, 18. (8) Kitamura, M. J. Cryst. Growth 2002, 237-239, 2205. (9) Cashell, C.; Corcoran, D.; Hodnett, B. K. J. Cryst. Growth 2004, 273, 258. (10) Roelands, C. P. M.; ter Horst, J. H.; Kramer, H. J. M.; Jansens, P. J. J. Cryst. Growth 2005, 275, e1389–e1395. (11) Kitamura, M.; Furukawa, H.; Asaeda, M. J. Cryst. Growth 1994, 141, 193. (12) Ramachandran, E.; Natarajan, S. Cryst. Res. Technol. 2006, 41, 411. (13) Ramesh Kumar, G.; Gokul Raj., S.; Sankar, R.; Mohan, R.; Pandi, S.; Jayavel, R. J. Cryst. Growth 2004, 267, 213. (14) Matsuoka, M.; Yamanobe, M.; Tezuka, N.; Takiyama, H.; Ishii, H. J. Cryst. Growth 1999, 198/199, 1299. (15) Dressler, D. H.; Hod, I.; Mastai, Y. J. Cryst. Growth 2008, 310, 1718. (16) Hammond, R. B.; Pencheva, K.; Roberts, K. J. J. Phys. Chem. B 2005, 109, 19550. (17) Lin, C. H.; Gabas, N.; Canselier, J. P.; Pepe, G. J. Cryst. Growth 1998, 191, 791. (18) Dhanaraj, G.; Shripathi, T.; Bhat, H. L. J. Cryst. Growth 1991, 113, 456. (19) “Amino Acids of the 21st century” (6) - Branched chain amino acid (BCAAs) in Amino Acids Link News, October 2005, 12, 1. (20) Khawas, B. Acta Crystallogr. 1970, B26, 1385. (21) Torii, K.; Iitaka, Y. Acta Crystallogr. 1971, B27, 2237.

2862 (22) (23) (24) (25) (26) (27) (28) (29) (30) (31) (32) (33) (34) (35)

(36) (37) (38) (39)

Crystal Growth & Design, Vol. 9, No. 6, 2009

Go¨rbitz, C. H.; Dalhus, B. Acta Crystallogr. C 1996, C52, 1464. Dalhus, B.; Go¨rbitz, C. H. Acta Crystallogr. B 2000, B56, 720. Zumstein, R. C.; Rousseau, R. W. Ind. Eng. Chem. Res. 1989, 28, 1226. Furuta, S.; Rousseau, R. W.; Teja, A. S. J. Cryst. Growth 1995, 148, 197. Givand, J. C.; Rousseau, R. W.; Ludovice, P. J. J. Cryst. Growth 1998, 194, 228. Almeida, F. M.; Freira, P. T. C.; Lima, R. J. C.; Reme’ dios, C. M. R.; Mendes Filho, J.; Melo, F. E. A. J. Raman Spectrosc. 2006, 37, 1296. Paul, E. L.; Tung, H.-H.; Midler, M. Powder Technol. 2005, 150, 133. Roberts, K. J. 15th International Symposium in Industrial Crystallization. 2002. Carpenter, K. J.; Wood, W. M. L. The 2nd Asian Particle Technology Symposium 2003; 1-55. Groen, H. C.; Roberts, K. J. Phys. Chem. 2001, 105, 10723. So¨hnel, O.; Mullin, J. W. J. Cryst. Growth 1978, 44, 377. So¨hnel, O.; Mullin, J. W. J. Cryst. Growth 1982, 60, 239. Go´mez-Morales, J.; Torrent-Burgue´s, J.; Rodrı´guez-Clemente, R. J. Cryst. Growth 1996, 169, 331. Liang, J. K. Ph.D. Thesis, Process Scale Dependence of L-Glutamic Acid Batch Crystallized from Aqueous Solution in Relation to Reactor Internals, Reactant Mixing and Process Conditions, University of Heriot-Watt, United Kingdom, 2002. Aitipamula, S.; Desiraju, G. R.; Jaskolski, M.; Nangia, A.; Thaimattam, R. Cryst. Eng. Comm. 2003, 5, 447. Dalhus, B.; Go¨rbitz, C. H. Acta Crystallogr. 1999, B55, 424. Davey, R. J.; Mullin, J. W.; Whiting, M. J. L. J. Cryst. Growth 1982, 58, 304. Bougeard, D. Ber. Bunsen-Ges. Phys. Chem. 1983, 87, 279.

Anuar et al. (40) Korolev, V. P.; Batov, D. V.; Smirnova, N. L.; Kustov, A. V. Russ. Chem. Bull. (Int. Ed.) 2007, 56, 739. (41) Park, B. H.; Yoo, K.-P.; Lee, C. S. Fluid Phase Equilib. 2003, 212, 175. (42) Privalov, P. L.; Gill, S. J. Pure Appl. Chem. 1989, 61, 1097. (43) Lizhuang, Z.; Xiaoling, W.; Shuquan, Z. J. Chem. Thermodyn. 2002, 34, 1481. (44) Acree, W. E. Thermodynamic Properties of Nonelectrolyte Solution; Academic Press Inc.: London, England, 1984. (45) Orellat, C. J.; Kirwan, D. J. Ind. Eng. Chem. Res. 1991, 30, 1040. (46) Burger, A.; Ramberger, R. Mikrochim. Acta [Wien] 1979, II, 259. (47) Giron, D. Thermochim. Acta 1995, 248, 1. (48) Ostwald, W. Z. Phys. Chem. 1897, 22, 289. (49) Balakrishnan, T.; Ramamurthi, K. Cryst. Res. Technol. 2006, 41, 1184. (50) Arunmozhi, G.; Jayavel, R.; Subramanian, C. J. Cryst. Growth 1997, 178, 387. (51) Omar, W.; Ulrich, J. Cryst. Res. Technol. 2006, 41, 431. (52) Smith, L. A.; Duncan, A.; Thomson, G. B.; Roberts, K. J.; Machin, D.; McLeod, G. J. Cryst. Growth 2004, 263, 480. (53) Liu, X.; Wang, Z.; Duan, A.; Zhang, G.; Wang, X.; Sun, Z.; Zhu, L.; Yu, G.; Sun, G.; Xu, D. J. Cryst. Growth 2008, 310, 2590. (54) El-Shall, H.; Jeon, J.-H.; Abdel-Aal, E. A.; Khan, S.; Gower, L.; Rabinovich, Y. Cryst. Res. Technol. 2004, 39, 214. (55) Mullin, J. W. Crystallization, 4th ed.; Elsevier Butterworth-Heinemann: Oxford, England, 2001. (56) Taggart, A. M.; Voogt, F.; Clydesdale, G.; Roberts, K. J. Langmuir 1996, 12, 5722. (57) Liu, X. Y. J. Chem. Phys. 2008, 112, 9949.

CG900133T