Selective Glycine Polymorph Crystallization by Using Microporous

Gianluca Di Profio , Efrem Curcio , Serena Ferraro , Carmen Stabile and Enrico Drioli ... Aamer Ali , Ramato Ashu Tufa , Francesca Macedonio , Efrem C...
0 downloads 0 Views 123KB Size
CRYSTAL GROWTH & DESIGN

Selective Glycine Polymorph Crystallization by Using Microporous Membranes Gianluca Di

Profio,*,†,‡

Selene

Tucci,†

Efrem

Curcio,‡

and Enrico

2007 VOL. 7, NO. 3 526-530

Drioli†,‡

Institute on Membrane Technology (ITM-CNR), c/o UniVersity of Calabria, Cubo 17/C, 87030, Rende, Italy, and Department of Chemical Engineering and Materials, UniVersity of Calabria (UNICAL), Cubo 41/A 87030, Rende, Italy ReceiVed September 11, 2006; ReVised Manuscript ReceiVed December 22, 2006

ABSTRACT: The selective crystallization of either R or γ polymorph of glycine was obtained by both static and dynamic membrane crystallization. The membrane matrix acts as a selective medium for solvent evaporation by modulating the rate for the achievement of supersaturation. In static membrane system this was obtained by changing the concentration of the stripping solution inside the membrane fibers, whereas in dynamic configuration it was obtained by varying the recirculation solution velocity. The increase in solvent evaporation rate induced an increase in the metastable zone width with effects on the heat and mass transfer during the nucleation kinetics. The consequent switching between a thermodynamically and a kinetically controlled nucleation stage resulted in the production of either a stable or metastable form of glycine. As control of polymorphism is critical in many pharmaceutical, electronic, and food applications, the possibility of inducing polymorph selection during a crystallization process by using membranebased techniques might represent a remarkable improvement. Introduction Control of crystals’ growth from solutions has received significant interest as it represents the primary purification/ separation technique for a wide range of substances in many industrial fields.1 For instance, crystallization is the most important unit operation as a means to form active pharmaceutical ingredients.2 Unfortunately, the fundamentals of the crystallization process are still not entirely understood. As a consequence, a significant number of productive cycles are designed on an empirical basis in order to control some features of the products, such as purity, shape, habit, size, and size distribution. Crystallization is also complicated by the fact that chemical compounds often show polymorphism.3 Different polymorphs have different structures and hence each of them is a unique material with its own physical and chemical properties (e.g., melting point, solubility, dissolution rate, bioavailability, optical properties, etc.). In the past few years membrane science and technology contributed to the development of a new concept of crystallizers in which hydrophobic microporous membranes act as physical supports contacting two isothermal subsystems subjected to mass interexchange in the vapor phase.4 In these systems, depending on the chemical and physical features of the membranes, the control of the rate of mass exchange by the crystallizing solution can be well behaved.5 The polymeric membrane surface can also act as the medium for heterogeneous nucleation, inducing a significant speeding up of the crystallization kinetics, even for biomacromolecules.6 The possibility to modulate the morphology (size and shape) and the high uniformity of the crystalline product obtained, for both inorganic7 and organic8,9 materials, characterizes the membrane-based method. In the present paper, we report on the crystallization of glycine by using membrane crystallizers. Depending on the operative conditions, the selective production of either R or γ polymorph of the amino acid was observed. * Corresponding author tel: +39 0984 492010; fax: +39 0984 402103; e-mail: [email protected]; web site: www.itm.cnr.it. † Institute on Membrane Technology. ‡ Department of Chemical Engineering and Materials.

Glycine represents a model polymorphic system. Three polymorphs are known: R, β, and γ.10-12 R-Glycine can be grown from almost neutral aqueous solutions (4 < pH < 8), where it exists as cyclic dimers of the zwitterionic molecule.13 β-Glycine, the least stable modification, crystallizes from water/ alcohol mixtures14 and it is monotropically related to the other two phases. The γ-glycine consists of molecules linked into a three-dimensional network,13 and can be formed from acidic or basic aqueous solutions (pH < 4; pH > 8), because in these conditions the charged glycine species inhibit the growth of the R form.15 A number of additives that inhibit the growth of R-glycine,16 the exposure of supersaturated glycine solutions to intense pulses of plane-polarized laser light,17-19 the confinement on patterned metallic substrates,20 and the crystallization in emulsions and lamellar phases21 are some methodologies addressed to obtain the γ-form. Recently, two additional polymorphs, δ- and -glycine, were obtained under highpressure.13 As the metastable R-glycine forms spontaneously from almost neutral aqueous solutions,13 in the past this induced consideration of the R form as the most stable modification at room conditions. However, by solubility,22 thermal,23 and calorimetric24,25 analyses, the relative stability at ambient temperature of the three forms was found to be γ > R > β. This is an indication that the relative stability of the polymorphs does not correlate directly with the easiness of their crystallization and that the spontaneous nucleation of R glycine in aqueous solutions is kinetically rather than thermodynamically controlled. In our experiments we observed an important effect of the rate of achievement of the supersaturation on the crystallization processes. In a membrane crystallizer the trans-membrane flux (J) is strictly related to the structural properties of the membrane, including porosity (), tortuosity (τ), pore size (r), and thickness (δ), as shown in the following equation:26

J)

2r 1 8RT 1/2 ∆p 3τ RT πM δ

( )

(1)

where R is the gas constant, T is the temperature, M is the molar mass, and ∆p is the gradient of partial pressure.

10.1021/cg0605990 CCC: $37.00 © 2007 American Chemical Society Published on Web 02/03/2007

Selective Glycine Polymorph Crystallization

Crystal Growth & Design, Vol. 7, No. 3, 2007 527

Control of the solvent evaporation rate, by modulating the gradient of partial pressure (which is the driving force of the evaporation process), allows control of the rate at which the supersaturation is reached; in this way, the switching from a thermodynamic to a kinetic control of the nucleation stage turns into the selective nucleation of either the stable (γ) or the metastable (R) polymorph of glycine. Experimental Section Glycine (from Sigma-Aldrich) was dissolved in bi-distilled water at a final concentration of 190 mg/mL. The pH of the stock solutions was around 6.2 ( 0.1. The solutions were filtered by syringe filters with nominal pore size of 0.22 µm in order to remove dust particles, and then fed to both static6 and dynamic8 membrane crystallization apparatus.6,7 In these systems the crystallizing and the stripping solution are “contacted” by the hydrophobic microporous membranes. The hydrophobic nature of the membranes (at the operative pressure used) prevents the passage of solutions in the fluid phase, but allows the establishment of a double liquid/vapor interface at the mouth of each pore on both sides of the membrane. A gradient of chemical potential between these two interfaces is the driving force of the solvent evaporation-migration-recondensation mechanism, which induces supersaturation in the crystallizing solution. In a static crystallizer, the solutions are quiescent, while in a dynamic configuration both the crystallizing and the stripping solutions are recirculated in a condition of axial flow in laminar regime. In static conditions tests were performed by changing the flow rate of solvent extraction, from the crystallizing solutions toward the stripping side, by varying the concentration of the stripping solution inside the membrane fibers. In the dynamic configuration, the variable parameter was the recirculation solutions’ velocity, which spanned the range 200-2000 µm/sec. Hydrophobic, polypropylene hollow fibers membranes (Accurel PP, nominal pore diameter 0.22 µm; thickness 530 ( 50 µm; inner diameter 1550 ( 150 µm, from Membrana GmbH, Germany) were used in all the experiments. Aqueous solutions of calcium chloride with a concentration ranging between 10 and 25 wt %/v, were used as stripping agents. All the trials were carried out at 20 °C. The solvent evaporation rate was evaluated by measuring the reduction of the volume of the crystallizing solutions as a function of time. The crystal morphology was assessed by an optical microscope (ZEISS-Axiovert 25) equipped with a video camera. Each sample was observed regularly every 30 min from the achievement of the supersaturation up to the appearance of the first visible crystals. It has been assumed that the time necessary for the crystal nucleus to become visible is negligible with respect to the formation of the nucleus itself from the mother solutions. Fourier transformed infrared analysis with an total attenuated reflectance method (ATR-FTIR) was performed on ground freshly prepared crystals, using a Perkin-Elmer Spectrum One. Weighed amounts (∼ 8 mg) of dry crystals were sealed in aluminum pans and analyzed by differential scanning calorimeter (DSC, PerkinElmer Ltd., Beaconsfield, Bucks), scanning from 100 to 200 °C at a heating rate of 15 °C/min.

Results and Discussion Depending on the operative parameters, namely the stripping solution concentration in quiescent conditions and the recirculation solution velocity in the dynamic configuration, either one of two crystal morphologies showed in Figure 1 was selectively observed. The two forms, named Form I and Form II, appeared in a time lag ranging from 2 to 10 days. The crystals were analyzed by ATR-FTIR; for the two forms I and II, two distinct types of vibrational spectra were obtained. An example of each of the two spectra is shown in Figure 2. Table 1 contains the characteristic peaks and their assignments.27,28 At almost neutral pH, glycine exists in solutions as cyclic dimers of the zwitterionic molecules; in the R crystals,

Figure 1. Optical images of the two crystal morphologies obtained during the membrane crystallization tests of glycine: (a) Form I; (b) Form II (the unit bar inside the figure corresponds to 50 µm).

Figure 2. ATR-FTIR spectra of the two crystal forms obtained in the experiments. Table 1. Peaks Observed (cm-1) in the ATR-FTIR Spectra from the Two Crystal Forms of Glycine and Their Assignments glycine crystals forms Form I

Form II

peak assignment

3152 3008 2972

3093 3008 2960 1664 1623 1574 1494 1436 1390 1334 1324 1154 1127 1043 928 890 685

NH asym stretch CH2 asym stretch CH2 sym stretch NH3+ asym def NH3+ asym def CO2- asym stretch NH3+ sym def CH2 bend CO2- sym stretch CH2 wagg CH2 twist NH3+ rock NH3+ rock CN asym stretch CH2 rock CC sym stretch CO2- bend

1607 1581 1504 1443 1408 1331 1132 1111 1033 909 892 695

these building blocks form a double-layers structure in which pairs of bi-layers are hydrogen bonded together. The γ-glycine, instead, consists of molecules linked by two hydrogen bonds (N-H‚‚‚‚O) to form helices around the 32 screw axes; a third lateral hydrogen bond connects the helices forming a threedimensional network. Differences in vibrational bands frequencies of each solid phase can be consequently attributed to variations of the intermolecular hydrogen bonding in each crystal structure. Hence, from the infrared analysis, the two forms I and II were recognized as R and γ glycine, respectively.

528 Crystal Growth & Design, Vol. 7, No. 3, 2007

Di Profio et al.

Figure 3. Representative DSC thermographs of the two forms of glycine obtained in membrane crystallization experiments: curve (a) Form I; curve (b) form II.

DSC characterizations of the two crystal forms confirmed these assignments (Figure 3). As reported by Boldyreva et al.,29 at room conditions the order of stability between the two polymorphs of glycine is γ > R. The γfR transition is thermodynamically forbidden at this temperature and the reverse transformation Rfγ, that should be thermodynamically allowed, was not observed (except in very few cases), probably due to kinetic hindrance. As the two polymorphs of glycine are enantiotropically related, at a certain higher (ranges of) temperature Ttransγ/R, the R form becomes more stable than the γ modification, thanks to an entropic contribution to the free Gibbs energy G (GR Gγ, HR > Hγ but SR > Sγ).24 Hence, at T g Ttransγ/R a phase change, in which the γ polymorph converts in the R modification, is observed. This is an endothermic transition, due to the difference ∆HR/γ between Hγ and HR, as shown in the first heating run in Figure 3b. The polymorphic conversion temperatures were shown to depend strongly on the size and shape of the crystals, on the thermal pretreatments, and on grinding and other mechanical treatment of the crystals.24 Generally, when a ground glycine sample is analyzed by DSC, as it is in the present case, a large signal is observed, because of the block-after-block transformation of the γ form into the R modification.29 In Figure 3a this transition is evidenced by a wide signal between 163 and 175 °C with two other weak signals around 150 and 185 °C. By cooling the R form to below the Ttransγ/R (first cooling run in Figure 3b), no reverse transformation Rfγ was observed (γfR has been observed to be irreversible), probably because of the high activation energy with respect to the energetic difference between the two structures. In the subsequent heating run (second heating run in Figure 3b), no transitions were observed, because of the complete conversion of the γ glycine to the R form. No such transformation were observed for crystals assigned to the R-structure by infrared spectra, as shown in Figure 3a. According to the literature,15 from almost neutral aqueous solutions (pH ∼ 6), the growth of the R modification should be favored; however, at pH 6.2, the preferential crystallization of γ glycine was achieved by careful control of the solvent evaporation rate through the pores of the membrane (Table 2). The action of the polymeric membrane as selective barrier for the vaporized water molecules allows precise regulation of the rate of the solvent evaporation. As shown in eq 1, several membrane properties can be changed to affect the gradient of partial pressure between the two solutions and, as a result, to

Table 2. Glycine Polymorphs Obtained in Different Conditions of Solvent Extraction Flow Rate (J) in the Static Configuration, and at Various Recirculation Solutions Velocities (W) in the Dynamic Membrane System J/(mL/h)

polymorph

V/µm‚sec-1

polymorph

6.9 × 10-3 13.8 × 10-3 18.4 × 10-3 20.7 × 10-3 23.0 × 10-3 27.6 × 10-3 34.5 × 10-3

γ γ R R R R R

349.6 539.9 709.1 899.9 989.9 1349.8 1845.5

γ γ R R R R R

influence the driving force of the evaporation. This provides an opportunity to systematically study the effects of the rate of variation of the supersaturation on the nucleation stage which, in turn, affects the structural outcome of the crystallization process. In static membrane crystallization experiments, the concentration of the stripping solution inside the membrane fibers was chosen as a variable to change the driving force. Gradual evaporation drives the glycine solution to the condition of supersaturation necessary for the phase transition. For J < 1.38 × 10-2 mL/h the only form obtained was invariably the γ-glycine, while for J > 1.84 × 10-2 mL/h the kinetic product, R-glycine, always appeared. By knowing the initial solute concentration and measuring the solvent trans-membrane flow rate, through a mass balance, the supersaturation of the solution (defined as β ) C/S where C is the actual concentration and S is the solubility of glycine at 20 °C22) has been calculated in correspondence of the nucleation point (Figure 4). The curve in Figure 4 clearly shows as the value of supersaturation in correspondence of which the nucleation occurs increases with the rate of solvent evaporation. Namely, the supersaturation at the nucleation point increases from 1.19 for J ) 6.9 × 10-3 mL/h to 2.41 for an evaporation rate of 3.45 × 10-2 mL/h. A similar behavior has been observed during the dynamic membrane crystallization tests. For recirculation solution velocities lower than 500 µm‚sec-1 the γ-polymorph was selectively crystallized, while for V higher than 700 µm‚sec-1, the metastable R-form was the only one obtained (Table 2). In dynamic membrane crystallizers the effect of the solution velocity reflects directly on the solute concentration profile near the membrane surface and therefore on the trans membrane flow

Selective Glycine Polymorph Crystallization

Crystal Growth & Design, Vol. 7, No. 3, 2007 529

Figure 5. Behavior of the induction time (tind) for the glycine crystals appearance as function of the recirculation solution velocity (V) in dynamic conditions.

favor precipitation, and the interfacial energies, which favor dissolution, is the driving force for precipitation. The magnitude of this driving force changes with the surface-area-to-volume ratio of the embryos and gives rise to a critical radius above which the nucleus grows and below which the precipitate redissolves into the solution. When the rate of variation of supersaturation is low, due to the low solvent evaporation rates, if the least stable structure or a mixture of critical nuclei forms at the same time, nuclei of the more stable polymorph have time to grow at the expense of the less stable form, via solventmediated transfer of solute. However, for higher solvent evaporation rates, the increase in the MZW induces nucleation at higher values of supersaturation and the sudden achievement of such supersaturation freezes the system in the production and growth of the metastable form, which is the first to appear in accordance with the Ostwald rule of stages. It is hence straightforward the strict relation between the MZW and the polymorphic yield of the system. In a membrane crystallizer, the membrane matrix acts as a selective barrier for solvent evaporation, allowing a progressive and finely controlled evaporation mechanism, which affects the MZW. Glycine polymorphs selection, dependent on the solvent evaporation rate, is therefore directly related to the rate of achievement of the supersaturation necessary for nucleation and, at the end, to the MZW. This seems to be responsible for the variation in the kinetics of the nucleation stage and, finally, of the switching from a thermodynamic to a kinetic control in the nucleation stage in the different conditions. The result is the selective production of either a stable or metastable polymorph of the amino acid. Lee et al.20 observed a similar effect on the possibility to control the structural outcome in the crystallization of glycine. In fact, they were able to impact the polymorph distribution of the metastable R and the less stable β-form in dependence of the solvent evaporation rate, by using hydrophilic metallic patterned surfaces. Here, by using hydrophobic membrane, we extended the possibility of polymorphic selection between the stable and metastable form of glycine by an evaporation-controlled mechanism. This experimental evidence demonstrates the driving force of the crystallization (supersaturation) and the rate of its variation can be an effective tool to promote polymorphic selection during crystallization.

rate; J increases with V, as the transport coefficients rise.26 This effect is showed in Figure 5, where the induction time (tind) for the crystals’ appearance as function of the feed solution velocity has been reported. tind decreases with V as the trans-membrane flow rate rises. Therefore, as the flow rate for solvent extraction, and consequently the velocity at which the supersaturation is reached is high, the kinetically favored form appears, while for low rate of solvent extraction the growth of the more stable form is again favored. These results were confirmed by repetitive experiments. A possible explanation at the base of the observed experimental results relies on the effect of the solvent evaporation rate on the width of the metastable zone (MZW). Several studies already demonstrated the MZW increases with the rate of achievement of supersaturations.30-32 This affects the rate of both heat and mass transfer during crystallization and, finally, the properties of the crystalline product such as size, size distribution, shape,32 and the overall crystal quality.33 According to classic nucleation theory, for a given polymorphic system, the crystallization of one form is associated with the overcoming of its nucleation barrier (that is related to the MZW). During nucleation, when solute molecules aggregate to form a cluster, the competition between the volumetric free energies, which

Conclusions The growth of the metastable R polymorph of glycine should be favored from almost neutral aqueous solutions. However, at pH 6.2, variation in the solvent evaporation rate in a membrane crystallizer can impact the polymorph yield of glycine, inducing the selective production of the γ form (the thermodynamic product). In this sense, the membrane structure acts as a selective medium for solvent evaporation by modulating the rate for the achievement of supersaturation and therefore the metastable zone width. The trans-membrane flow rate depends, among other parameters, on the partial pressure gradient between the two sides of the membrane. In static conditions the ∆p has been raised by increasing the concentration of the stripping solution, whereas in the dynamic system it has been raised by acting on the recirculation solution velocities. In both cases, this resulted in the increase in the solvent evaporation rate which is associated with the increase in the width of the metastable zone and therefore the kinetics of nucleation. The switching from a thermodynamic to a kinetic control of the nucleation stage and, at the end, the selective production of either a stable or metastable form of glycine was consequently obtained. As the control of polymorphism is critical in many pharmaceutical, electronic, and food applications, the possibility of

Figure 4. Relation between the solvent evaporation rate J and the supersaturation β at the nucleation point for the static membrane crystallization experiments.

530 Crystal Growth & Design, Vol. 7, No. 3, 2007

inducing polymorph selection during a crystallization process by membrane-based techniques might represent a remarkable improvement. Acknowledgment. The MIUR, Centre of Excellence CEMIF.CAL at the University of Calabria (UNICAL) is acknowledged for funding. We are also grateful to Dr. A. Crispini (Chemistry Department, UNICAL) and to Dr. A. Gugliuzza (ITM-CNR) for analytical support. References (1) Myerson, A. S. Handbook of Industrial Crystallization, 2nd ed.; Butterworth-Heinemann: London, 2002. (2) Paul, E. L.; Tung, H. H.; Midler, M. Powder Technol. 2005, 150, 133-143. (3) Bernstein, J.; Davey, R. J.; Henck, J. O. Angew. Chem., Int. Ed. 1999, 38, 3440-3461. (4) Curcio, E.; Criscuoli, A.; Drioli, E. Ind. Eng. Chem. Res. 2001, 40, 2679-2684. (5) Curcio, E.; Di Profio, G.; Drioli, E. J. Cryst. Growth 2003, 247, 166-176. (6) a. Di Profio, G.; Curcio, E.; Cassetta, A.; Lamba, D.; Drioli, E. J. Cryst. Growth 2003, 257, 359-369. b. Curcio, E.; Fontananova, E.; Di Profio, G.; Drioli, E. J. Phys. Chem. B 2006, 110, 12438-12445. (7) Drioli, E.; Curcio, E.; Criscuoli, A.; Di Profio, G. J. Membr. Sci. 2004, 239, 27-38. (8) Di Profio, G.; Curcio, E.; Drioli, E. J. Struct. Biol. 2005, 150, 4149. (9) Di Profio, G.; Perrone, G.; Curcio, E.; Cassetta, A.; Lamba, D.; Drioli, E. Ind. Eng. Chem. Res. 2005, 44, 10005-10012. (10) Marsh, R. E. Acta Crystallogr. 1958, 11, 654-663. (11) Iitaka, Y. Acta Crystallogr. 1960, 13, 35-45. (12) Iitaka, Y. Acta Crystallogr. 1961, 14, 1-10. (13) Dawson, A.; Allan, D. R.; Belmonte, S. A.; Clark, S. J.; David, W. I. F.; McGregor, P. A.; Parsons, S.; Pulham, C. R.; Sawyer, L. Cryst. Growth Des. 2005, 5, 1415-1427. (14) Weissbuch, I.; Torbeev, V. Yu.; Leiserowitz, L.; Lahav, M. Angew. Chem., Int. Ed. 2005, 44, 3226-3229. (15) Towler, C. S.; Davey, R.; Lancaster, R. W.; Price, C. J. J. Am. Chem. Soc. 2004, 126, 13347-13353.

Di Profio et al. (16) Weissbuch, I.; Leiserowitz, L.; Lahav, M. AdV. Mater. 1994, 6, 953966. (17) Zaccaro, J.; Matic, J.; Myerson, A. S.; Garetz, B. A. Cryst. Growth Des. 2001, 1, 5-8. (18) Garetz, B. A.; Matic, J.; Myerson, A. S. Phys. ReV. Lett. 2002, 89, 175501. (19) Sun, X.; Garetz, B. A.; Myerson, A. S. Cryst. Growth Des. 2006, 6, 684-689. (20) a. Lee, A. Y.; Lee, I. S.; Dette, S. S.; Boerner, J.; Myerson, A. S. J. Am. Chem. Soc. 2005, 127, 14982-14983. b. Lee, A. Y.; Lee, I. S.; Myerson, A. S. Chem. Eng. Technol. 2006, 29, 281-285. (21) Allen, K.; Davey, R. J.; Ferrari, E.; Towler, C.; Tiddy, G. J. Cryst. Growth Des. 2002, 2, 523-527. (22) Park, K.; Evans, J. M. B.; Myerson, A. S. Cryst. Growth Des. 2003, 3, 991-995. (23) Sakai, H.; Hosogai, H.; Kawakita, T.; Onuma, K.; Tsukamoto, K. J. Cryst. Growth 1992, 116, 421-426. (24) Perlovich, G. L.; Hansen, L. K.; Brauer-Brandl, A. J. Therm. Anal. Calorim. 2001, 66, 699-715. (25) Boldyreva, E. V.; Drebushchak, V. A.; Drebushchak, T. N.; Paukov, I. E.; Kovalevskaya, Y. A.; Shutova, E. S. J. Therm. Anal. Calorim. 2003, 73, 409-418. (26) Datta, R.; Dechapanichkul, S.; Kim, J. S.; Fang, L. Y.; Uehara, H. J. Membr. Sci. 1992, 75, 245-263. (27) Ambujam, K.; Selvakumar, S.; Prem Anand, D.; Mohamed, G.; Sagayaraj, P. Cryst. Res. Technol. 2006, 41, 671-677. (28) Ferrari, E. S.; Davey, R. J.; Cross, W. I.; Gillon, A. L.; Towler, C. S. Cryst. Growth Des. 2003, 3, 53-60. (29) Boldyreva, E. V.; Drebushchak, V. A.; Drebushchak, T. N.; Paukov, I. E.; Kovalevskaya, Y. A.; Shutova, E. S. J. Therm. Anal. Calorim. 2003, 73, 419-428. (30) He, G.; Bhamidi, V.; Tan, R. B. H.; Kenis, P. J. A.; Zukoski, C. F. Cryst. Growth Des. 2006, 6, 1175-1180. (31) Mersmann, A.; Bartosch, K. J. Cryst. Growth 1998, 183, 240-250. (32) Kim, K.-J.; Mersmann, A. Chem. Eng. Sci. 2001, 56, 2315-2324. (33) Talreja, S.; Kim, D. Y.; Mirarefi, A. Y.; Zukoski, C. F.; Kenis, P. J. A. J. Appl. Crystallogr. 2005, 38, 988-995.

CG0605990