Molecular Speciation Controlling Stereoselectivity of Additives: Impact

Jan 18, 2007 - The stereoselective habit modification in α-glycine crystals in the presence of α-amino acid additives has been shown to be controlle...
0 downloads 0 Views 1MB Size
Download PDF
Molecular Speciation Controlling Stereoselectivity of Additives: Impact on the Habit Modification in r-Glycine Crystals Sendhil K.

Poornachary,†

Pui Shan

Chow,‡

Reginald B. H.

Tan,*,†,‡

and Roger J.

Davey§

Department of Chemical and Biomolecular Engineering, National UniVersity of Singapore, 4 Engineering DriVe 4, Singapore 117576, Institute of Chemical and Engineering Sciences, 1 Pesek Road, Jurong Island, Singapore 627833, and Molecular Materials Centre, School of Chemical Engineering and Analytical Sciences, UniVersity of Manchester, P.O. Box 88,Manchester M60 1QD, United Kingdom

CRYSTAL GROWTH & DESIGN 2007 VOL. 7, NO. 2 254-261

ReceiVed May 11, 2006; ReVised Manuscript ReceiVed NoVember 7, 2006

ABSTRACT: The demonstration of stereoselective habit modification in molecular crystals using “tailor-made” (or structurally related) additives, over the last two decades, has been a milestone in understanding the phenomenon of molecular recognition at crystal interfaces. The centrosymmetric R-glycine crystal has provided a classic example in earlier studies for elucidating the mechanisms of such stereoselective processes. In these previous studies, with chirally resolved (L or D-amino acids) and racemic R-amino acids (DL-amino acids) as tailor-made additives, habit modification was observed to be along the enantiopolar b-axis of R-glycine crystals. Revisiting this work has, however, revealed additional habit modification along the c-axis of the R-glycine crystal with certain R-amino acids as additives (viz. aspartic acid and glutamic acid). In the presence of L-Asp and L-Glu, the (01h1) and (01h1h)faces were of morphological importance and with the D-amino acids, the (011) and (011h) faces were well-developed. On the basis of the fact that these amino acids exist in two charged states (zwitterion and anion) and building on the stereoselectivity mechanism, it is surmised that the zwitterionic species interact along the b-axis of the R-glycine crystal and the anions alter their interaction from the crystallographic b-axis to the c-axis, due to electronic repulsive forces acting at the docking sites on the (010) and (01h0) faces. In this paper, we have used optical microscopy, molecular modeling, and IR spectroscopy to demonstrate and explain the newly observed habit modification. Introduction Interestingly, the origin of modern stereochemistry dates back to 1848, when Louis Pasteur observed that crystals consisting of molecules of opposite handedness (enantiomers) resolved from an optically inactive solution of sodium ammonium tartrate.1 Pasteur was able to distinguish visually between the enantiomeric crystals that had different crystal habits and hence could mechanically separate the conglomerate (physical mixture of enantiomeric crystals). Recent work by Leiserowitz and coworkers2 has achieved resolution of conglomerates by inducing stereoselective habit modification using chirally resolved additives. These additives are structurally similar to the solute molecules and hence are referred to as “tailor-made” auxiliary molecules. They are basically composed of two moieties: One, the “binder”, has a similar structure (and stereochemistry) to that of the substrate molecule on the crystal surface where it adsorbs. The second moiety, referred to as the “perturber”, is modified when compared to the substrate molecule and thus hinders the attachment of the oncoming molecular layers of the solute molecules to the crystal surface. Hence, the resolved additives (L- or D-amino acids) added to the solution in trace amounts stereoselectively adsorb on the surface of the crystal nuclei of the corresponding absolute configuration (viz. R and S). Subsequently, the additive inhibits the growth of either of the enantiomeric crystals, thereby enabling kinetic resolution (through difference in growth rates) of the conglomerates and inducing morphological differences. Leiserowitz and co-workers, in a subsequent classic work,3 have shown that habit modification in R-glycine crystals in the * To whom correspondence should be addressed. Tel: +65 6516-6360. Fax: +65 6779-1936. E-mail: [email protected]. † National University of Singapore. ‡ Institute of Chemical and Engineering Sciences. § University of Manchester.

presence of R-amino acids was always along the enantiopolar b-axis (discussed below). However, from our experimental studies, we observed that with L-Asp and L-Glu, growth inhibition was along the c-axis of R-glycine. Interestingly, on increasing the additive concentration, the action along the b-axis became dominant, validating the previous observation. The primary objective of this work is to explain the mechanism of this new habit modification in R-glycine crystals since it has not been reported previously. Furthermore, we explore the influence of solution speciation of additives and the role played by the perturber group, in particular, the partial delocalization of negative charge on the additive anion side chain moiety in controlling the stereoselective habit modification in R-glycine crystals. Habit Modification in r-Glycine: Stereoselectivity Mechanism3 Revisited Glycine (+H3NCH2CO2-), a simple amino acid, usually crystallizes as the R-polymorph from pure aqueous solution. In an R-glycine crystal structure,4 the zwitterionic molecules form centrosymmetric dimers through N-H‚‚‚O hydrogen bonds. Chains of these dimers are formed along the c-axis by additional N-H‚‚‚O hydrogen bonds. The dimer layers are stacked along the b-axis through C-H‚‚‚O interactions with a centrosymmetric space group P21/n (Figure 1a). The theoretical growth morphology of the R-glycine crystal is shown in Figure 1b. The glycine molecule is prochiral, containing two hydrogen atoms at the central carbon atom that are enantiotopic,1 namely, replacement of one of these hydrogen atoms by a different group yields a chiral molecule. From the packing arrangement in the R-glycine crystal structure, we observe that glycine molecules within a dimer layer (viz. “1” and “2” or “3” and “4”) are related by a center of inversion symmetry, and those in the alternate layers (viz. “1” and “3” or “2” and “4”) are related by a 2-fold screw

10.1021/cg060273r CCC: $37.00 © 2007 American Chemical Society Published on Web 01/18/2007

Habit Modification in R-Glycine Crystals

Crystal Growth & Design, Vol. 7, No. 2, 2007 255

Figure 2. Illustration of the stereoselective habit modification3 in R-glycine crystals: (a) pure glycine, (b) 0.5 wt % (w/w of glycine) of L-alanine, and (c) 0.5 wt % DL-alanine.

symmetry. The characteristic feature is that the enantiotopic C-HRe and C-HSi bonds of a glycine molecule are directed along the crystallographic +b- and -b-axes, respectively (cf. Figure 1a). As a result, the b-axis is enantiopolar and the (010) and (01h0) crystal faces are enantiotopic. The C-HRe bond of glycine molecule on the (010) surface could be enantioselectively substituted by the side chain group of a D-amino acid (of R configuration) and the C-HSi bond on the (01h0) surface with that of an L-amino acid (of S configuration). As a result, stereoselective habit modification is caused along the b-axis of R-glycine crystals in the presence of these additives.3 Typically, R-glycine crystallizes with prismatic bipyramidal morphology from pure aqueous solution (Figure 2a). In contrast, a prismatic pyramidal crystal habit with a (01h0) basal plane is obtained in the presence of L-alanine [+H3NCH(CH3)CO2-] due to growth inhibition along the -b-axis (Figure 2b). Similarly, D-Ala induces an enantiomorphous morphology because of growth inhibition along the +b-axis. With a racemic amino acid (DL-Ala), platelike crystals with dominant {010} faces are formed because of growth inhibition along (b-axes of R-glycine (Figure 2c). These observations are consistent with the previous work3 on habit modification in R-glycine crystals.

by dissolving 27.5 g of glycine in 100 g of water (3.67 M) at 40 °C, with appropriate amounts of additives added, and then cooled at 0.5 °C/min until visible crystals nucleated from the supersaturated solution (final crystallization temperatures were in the range of 1520 °C). For the unstirred crystallization, the glycine solution was prepared by dissolving 3.3 g of glycine in 10 g of water (4.4 M) at 50 °C. This solution was filtered through a 0.22 µm Millipore PVDF Durapore membrane filter and cooled to room temperature (22 °C) in a crystallization dish. Single crystals of R-glycine with distinguishable crystal faces (1-3 mm size) were obtained from the supersaturated solution in 12-24 h of time, allowing for detailed morphological analysis. The solution pH measurements were made at room temperature (22 °C) using a Mettler-Toledo (Seven Multi model) pH meter fitted with a glass calomel electrode and an ATC probe for temperature compensation. Microscopy, IR, and Diffraction. The crystals obtained were examined using an optical polarizing microscope (Olympus, BX51) connected to a CCD camera, and images were recorded using Soft Imaging System’s Analysis image capture software. Infrared spectra for the amino acids (L-Asp and L-Glu) and their salts in the solid state were collected with a Thermo Nicolet Avatar 360 Fourier Transform Infrared Spectrometer (FT-IR) using an attenuated total reflection (ATR) method. A single reflection germanium crystal was employed with a spectral range between 4000 and 675 cm-1. These solid-state spectra obtained were used as a reference for vibration assignments of the IR spectra obtained in solution. The IR spectra of the amino acids in water were collected with a Nicolet 4700 spectrometer (Thermo Electron Corp.), equipped with a Dipper-210 ATR-FTIR probe with a ZnSe crystal (Axiom Analytical Inc.). Deionized water was used as the background, and spectral data were collected in the range between 4000 and 650 cm-1. Solutions of L-Asp and L-Glu (2 wt % concentration) were prepared at 70 °C, and their spectra were collected at pH 3.0, 7.2, 8.9, and 11.7. For crystal structure confirmation, X-ray powder diffraction analysis was carried out using a Bruker D8 advance X-ray Diffractometer.

Experimental Section

Results and Discussion

Materials. Glycine and other R-amino acids (used as additives) were obtained from Sigma-Aldrich with 99.0+% purity and were used as such. Deionized, 0.22 µm filtered water was used in preparing glycine solutions. Hydrochloric acid and sodium hydroxide were used for pH adjustment. Crystallization Experiments. Two experimental protocols were used as follows: stirred and unstirred crystallizations, depending on the suitability of the crystals for microscopic analysis. Under stirred conditions, crystallization experiments were performed in an agitated, 250 mL thermostated glass vessel. The glycine solution was prepared

New Habit Modification in r-Glycine. In the presence of and L-Glu (0.5 wt %, w/w glycine), R-glycine crystallized with an isometric crystal habit (Figure 3a,b). The dominant faces observed are {011} and {110} with the (01h0) face slightly developed (Figure 3c). This crystal habit is clearly distinguishable from that of the glycine crystal obtained from pure solution, which is usually elongated along the c-axis with less developed {011} faces (cf. Figure 2a). Besides, the crystal habit of R-glycine in the presence of these two additives was prismatic

Figure 1. (a) Packing arrangement in R-glycine viewed along the a-axis, delineated by dominant crystal faces. The hydrogen atoms on the (010) surface (C-HRe) and (01h0) surface (C-HSi) are enantiotopic.3 (b) Theoretical growth morphology of R-glycine showing the significant crystal faces (computed in Accelrys Materials Studio5 software by using Attachment Energy method;6 the DRIEDING7 forcefield was used with atomic charges calculated using the charge-equilibration method8).

L-Asp

256 Crystal Growth & Design, Vol. 7, No. 2, 2007

Poornachary et al.

Figure 3. New habit modification in R-glycine crystallized with the additives: (a) 0.5 wt % (w/w of glycine) of L-Asp, (b) 0.5 wt % of L-Glu, (c) 0.5 wt % of L-Asp (crystal faces labeled), and (d) 1.0 wt % of L-Asp.

bipyramidal, unlike with L-Ala, in which case habit modification had resulted in prismatic pyramidal crystal morphology (cf. Figure 2b). With a further increase in L-Asp concentration (between 0.75 and 1.0 wt %), again prismatic bipyramidal crystals were obtained, however, with partial growth inhibition along the -b-axis. At L-Asp concentrations between 1.0 and 2.0 wt %, the crystal habit was prismatic pyramidal with dominant (01h0) basal faces (Figure 3d), consistent with the proposed stereoselectivity mechanism.3 Likewise, in any of the typical experimental runs with L-Glu, we observed that the additive action was dominant along the (c-axes of R-glycine crystal, prior to the growth inhibition along the -b-axis. Additive Induced Molecular Speciation in Glycine Solution. In order to explain the newly observed habit modification, it is necessary to examine the chemistry of the additives in glycine solution. The species distribution of glycine and aspartic acid obtained from their pKa values9 as a function of pH is shown in Figure 4a. In the pH region 4-6, it is apparent that glycine exists predominantly as zwitterions, besides a small fraction of cations (ca. 0.02 mol fraction). In this region, aspartic acid exists in two charged states, as zwitterions and anions. Therefore, with the addition of L-Asp to glycine solution, the pH decreased from ca. 6.2 (pI of glycine 5.979) to ca. 4.2 (at 2 wt % concentration), accompanied by changes in molecular speciation (Figure 4b-1). Similarly, glutamic acid with a pKa value of 4.25 (for the side chain -COOH group) would undergo speciation in glycine solution and exists as zwitterions and anions in the pH domain 4-6 (Figure 4b-2). In contrast, alanine, with no carboxylic acid moiety in its side chain, would predominantly exist as zwitterions in this pH region (Figure 4b-3). Now, from the speciation data, the distribution coefficients for the zwitterionic and anionic species of the additives (viz. molar fraction of anions/molar fraction of zwitterions) could be calculated for the experimental conditions reported in the New Habit Modification in R-Glycine. At concentrations of 0.5 and 1.0 wt % of L-Asp, the distribution coefficients are ca. 18.25 and 9.53, respectively, and for L-Glu, the distribution coefficients at these concentrations are calculated as ca. 5.79 and 3.25, respectively. Hence, it is understood that the relative distribution

Figure 4. (a) Speciation of aqueous glycine (Gly) and aspartic acid (Asp) as a function of pH. (b) Ionic equilibrium of (1) Asp, (2) Glu, and (3) alanine (Ala), respectively, in Gly solution in the pH domain 4-6.

of the zwitterions increases significantly with an increasing concentration of the additives, as the solution pH changes simultaneously. Isolating the Effect of Gly+ on Habit Modification. From the species distribution, we identify three molecular species, which could potentially cause habit modification in R-glycine: L-Asp zwitterion, L-Asp anion, and Gly+ (glycine cation). In accordance with the stereoselectivity mechanism discussed before, the action of L-Asp zwitterions should be confined to the -b-axis of R-glycine. This was indeed observed experimentally, with partial to complete growth inhibition along the -b enantiomorphous half of the crystal (cf. Figure 3c,d). The dominant {011} faces (cf. Figure 3c) could, therefore, possibly be attributed to apparent growth inhibition along (c-axes, either by the Gly+ or by the L-Asp anions. In this context, we note that Davey and co-workers10 have conjectured the “selfpoisoning” mechanism, explaining the role of Gly+ on the growth inhibition of {011} faces of R-glycine at low pH values. In their work, Gly+ speciation on lowering the pH of pure glycine solution (using HCl acid) has been attributed to cause growth inhibition along fast-growing (c-axes of R-glycine crystal nuclei, thence resulting in the nucleation of γ-glycine at pH < 3.8. However, in their experimental studies, the authors could not observe habit modification in R-glycine in the transitional pH range (viz. pH 6.0 to pH 4.0). To this end, we conducted some additional crystallization experiments with pure glycine solution at pH 6.2 (the original pH) and pH 4.5 (by adjusting the pH using HCl acid) in a 250 mL crystallizer using the procedure as mentioned in Crystallization Experiments. We

Habit Modification in R-Glycine Crystals

Figure 5. Effect of Gly+ on habit modification in R-glycine crystals: (a) pH 6.2 (nil Gly+) and (b) pH 4.5 (ca. 0.7 wt % Gly+, w/w of glycine).

observed that R-glycine crystals obtained at pH 4.5 were more isometric with dominant {011} faces in comparison with those crystallized at pH 6.2, which were apparently elongated along the c-axis (Figure 5). These observations corroborate the proposed self-poisoning mechanism.10 Besides, given this observation, the implication is that very low concentrations of Gly+ species are required to produce an effect on the crystal habit of R-glycine. Therefore, in the case with L-Asp and L-Glu as additives, R-glycine crystals had to be obtained at pH 6.0 (near pI of glycine), in order to isolate the effect of Gly+ (produced on speciation of the additives) on habit modification. Confirming Additive Action along the c-Axis. R-Glycine crystals were obtained with the additives added in at the same concentration levels as before in the New Habit Modification in R-Glycine but at pH 6.0 (using NaOH to adjust pH). Now, in the absence of Gly+ species as a result of pH adjustment, habit modification was still observed along the (c-axes (Figure 6a). The dominant {011} faces indicated the action of L-Asp anions on these faces resulting in growth inhibition along the c-axis. Furthermore, from a closer observation, we could distinguish (see the crystal labeled “1”) that the {01h1} faces on the -b enantiomorphous part of R-glycine [viz. (01h1) and (01h1h) faces] are morphologically more significant than those on the +b enantiomorphous half [viz. (011) and (011h) faces]. Similar habit modifications were also observed with L-Glu (Figure 6b, see crystal labeled “1” for the difference in morphological importance between the {011} faces). Analogously, with D-Asp and D-Glu, we observed that the {011} faces on the +b enantiomorphous half of R-glycine were morphologically more significant as compared with the symmetrically related {01h1} faces on the -b enantiomorphous half. These subtle but significant observations on the habit modification will be explained using the stereoselectivity mechanism in the Mechanism of Molecula Differentiation. At higher additive concentrations, we observed prismatic pyramidal crystals with welldeveloped (01h0) basal faces indicating growth inhibition along the b-axis (Figure 6c,d). These results are again consistent with the observations made previously (New Habit Modification in R-Glycine). Here, we note that both from our experiments and from the previous study,10 the Na+ ions of NaOH did not have

Crystal Growth & Design, Vol. 7, No. 2, 2007 257

a pronounced effect on the {011} faces at the concentration levels used to adjust the solution pH. Following this, R-glycine crystals were obtained in the presence of racemic aspartic acid (DL-Asp or an equimolal mixture of L- and D-Asp). They exhibited equally developed {011} faces on both the (b enantiomorphous halves, confirming the action of additives along the c-axis (Figure 6e). If, on the other hand, the additive action had been confined to the enantiotopic crystal faces [viz. (010) and (01h0) faces], platelike crystals should be obtained due to growth inhibition along the (b-axes. However, with racemic glutamic acid, both isometric (Figure 6f) and platelike crystals (Figure 6g) were obtained either simultaneously or from different batches. From these experimental observations, we conclude that in the case of aspartic acid, the action along the c-axis was dominating over the action along the b-axis, and with glutamic acid, the action along either of the axes was kinetically competitive at the concentration levels studied. The reason behind this disparity in additive action is explained in the ensuing section using molecular modeling, speciation, and spectroscopic data. Mechanism of Molecular Differentiation. The experimental habit modification along the c-axis of R-glycine crystals was observed to be determined by the absolute configuration of the additive molecules and, therefore, should be stereospecific in nature. From the known solution chemistry of the additives and the observed habit modification along two different crystallographic axes, we surmise that molecular speciation controls the stereoselectivity of the additives on R-glycine crystal faces. We represent our proposition in Scheme 1, wherein the zwitterions make a stereospecific interaction on the enantiotopic faces3 (b-axis) and the anions interact along the c-axis. However, from the stereoselectivity mechanism, we understand that the molecular recognition process along the enantiopolar b-axis in R-glycine should be the same for both the zwitterionic and the anionic additive species. Therefore, our objective herein is to understand the intermolecular interactions of the additive species, especially the anions as compared to the zwitterions, at the R-glycine crystal interfaces so as to explain the new habit modification. (a) Interaction of Additive Molecular Species with r-Glycine Crystal. The interaction of the additive molecular species with the crystal faces of R-glycine was modeled and visualized in Accelrys Materials Studio.5 The L-Asp zwitterion was modeled in accordance with the stereoselectivity mechanism by replacing the C-HSi bond of glycine molecule on the (01h0) face with the side chain moiety of aspartic acid (-CH2COOH) (Figure 7a). On building the additive molecule, the molecular conformation of the side chain was kept as found within the crystal structure of L-aspartic acid.14 Subsequently, the anion was modeled by deprotonating the side chain carboxylic acid group. We can observe that this introduces a lone pair electron on the carboxylate moiety (Figure 7b), therefore resulting in electronic repulsion with the carboxylate group of the glycine molecule at the neighboring site (the O‚‚‚O contact length was measured to be ca. 2 Å). At distances less than 3.5 Å, repulsive forces due to the lone pair electrons of the oxygen atoms can be significantly stronger (ca. 1-2 kcal/ mol15). Such unfavorable interactions could prevent the anion from incorporating into the crystal lattice along the b-axis. Following this, the interaction of Asp anion along the c-axis was then envisaged by making a stereospecific docking on the (01h1) and (01h1h) faces (Figure 7c). At these crystallographic sites, we observe that the anions have no such repulsive interactions as on the (01h0) face, on account

258 Crystal Growth & Design, Vol. 7, No. 2, 2007

Poornachary et al.

Figure 6. R-Glycine crystals obtained at pH 6.0: (a, b) 0.5 wt % (w/w of glycine) of L-Asp and L-Glu, respectively; (c, d) 1.5 wt % of L-Asp and L-Glu, respectively; (e, f) 0.5 wt % of D + L-Asp and D + L-Glu, respectively; and (g) 2 wt % D + L-Glu.

Scheme 1.

Interaction of Additive Molecular Species with r-Glycine Crystala

a The dimer growth unit of R-glycine11-13 is shown as a pair of centrosymmetric glycine zwitterions.

of the oblique nature of the crystal faces delineating the glycine dimer motifs along the c-axis. An analogous explanation holds for the interaction of the D-Asp anions on the (011) and (011h) faces of R-glycine. This interaction mechanism is akin to that of racemic hexaflurovaline [(CF3)2-CH-CH(NH3+)CO2-], a tailor-made auxiliary molecule reported in an earlier work,16 designed to preferentially interact with the {011} faces of R-glycine crystal. This additive molecule was prevented from adsorbing along the b-axis because of steric and electronic repulsion of the hexafluroisopropyl moiety on the {010} faces.

However, it is also possible for the Asp anion molecule to impinge preferentially at a (010) step instead of the flat crystal surfaces of R-glycine, thereby circumventing the repulsive effects. Such an interaction mechanism for the adsorption of auxiliary molecules on the (010) steps of glycine crystals has been demonstrated in previous studies by applying second harmonic measurements.17 Subsequently, the anion molecule can block an adjacent site and hence cause growth inhibition along the c-axis. A recent work by Torbeev et al.18 has also explained the role played by the perturber moiety in controlling the stereoselective interaction with the (010) step planes of R-glycine. Such interactions had resulted in nucleation inhibition of the R-form and thereafter inducing the nucleation of either βor γ-forms. It is also significant to note that an L-Asp anion may not incorporate on the (011) and (011h) surfaces, which could be understood from the packing arrangement of the host molecules within the crystal structure (cf. Figure 1). In principle, both types of host prochiral molecule (viz. of pro-R and pro-S absolute configurations) are expressed at all of the four {011} faces in contrast to the (010) and (01h0) surfaces that individually express only one of the prochiral forms. However, the side chain of the L-Asp anion would have steric hindrance on incorporation on the (011) or (011h) faces at symmetrically related sites. Although a surface vacancy could be created to allow incorporation of L-Asp on these faces, in that case, it will be required to remove a host molecule on the {010} step below the current growth layer. Therefore, it is reasonable to expect that docking on these surfaces would be less significant and hence growth inhibition would be less as supported by the experimental evidence presented (cf. Confirming AdditiVe Action along the c-Axis).

Habit Modification in R-Glycine Crystals

Crystal Growth & Design, Vol. 7, No. 2, 2007 259

Figure 8. Molecular modeling of the interaction of L-Glu anion with the (01h0) face of R-glycine crystal.

Figure 7. Molecular modeling of the interaction of the additive species with R-glycine crystal: (a) L-Asp zwitterion and (b) L-Asp anion on the (01h0) surface; (c) L-Asp anion on the (01h1) and (01h1h) surfaces.

Hence, considering the growth inhibition mechanism for the {011} faces, it is one of blocking of an adjacent site in the current growth layer on the {010} step plane, whereas in the case of {010} faces, the additive blocks the subsequent growth layer and not the current one. Likewise, the interaction of L-Glu anion on the (01h0) surface was modeled by stereospecifically docking an L-Glu zwitterion molecule taken from its crystal structure19 and subsequently deprotonating the distal -COOH group (Figure 8). This methodology was preferred over the construction of the side chain moiety, since in this case, the substrate moieties (+H3NCRH-CO2-) of the glycine molecule in the host crystal lattice and the L-Glu zwitterion were stereospecifically superimposable. Besides, by this method, the side chain conformation of L-Glu was kept intact as found within its crystal structure. As in the case of L-Asp anion, it could be expected that L-Glu anion on occupying the site on the (01h0) surface is likely to experience electronic lone pair repulsions between its side chain COO-

group and the carboxylate group of a glycine molecule at the adjacent site. As a result of this, the L-Glu anion should occupy either the crystallographic site “1” on the (01h1) face or the site “2” on a (01h0) step of R-glycine and subsequently result in the c-axis habit modification (cf. Figure 6b). (b) Factors Controlling Additive Stereoselectivity: Inferences from IR Spectroscopy, CSD, and Molecular Modeling. From the molecular models shown earlier (cf. Figures 7b and 8), it is understood that partial delocalization of negative charge on the side chain carboxylate group of the additive anions caused the repulsive effects described therein. Primarily, this factor could result an increase in the energy penalty associated with the incorporation of the additive species on the {010} faces of R-glycine. Currently, we are working toward obtaining these energy penalty values by calculating the attachment and binding energies15 for the additives on the various surfaces of R-glycine crystals. Theoretically, the partial delocalization of negative charge is reflected by the pKa values9 of the distal -COOH group of the two acids, with Asp anion having a higher ionization constant (Ka ) 2.2 × 10-4) as compared to the Glu anion (Ka ) 5.6 × 10-5). Therefore, as per our proposed model, the difference in the pKa values could result in greater repulsive effects at the {010} faces for the former anionic species as compared to the latter. This was indeed reflected in the extent of habit modification along the c-axis of R-glycine in the presence of these two additives (cf. Confirming AdditiVe Action along the c-Axis) with aspartic acid having a greater impact on the {011} faces. In addition to the above, the partial delocalization factor could also be influential in causing intramolecular interactions within the molecular structure of the additive anions (viz. Asp and Glu). In this context, previous work20 on IR spectroscopic studies of these two amino acids in solution (water and D2O) has shown intramolecular charge transfer interactions21 between the CRNH3+ group and the side chain (or distal) COO- group in the anions. To this end, in order to provide evidence for such interactions, we have obtained IR spectra for L-Asp and L-Glu in water at different pH conditions: pH 3.0 (mostly zwitterions), pH 7.2 (anions), pH 8.9 (anions and dianions), and pH 11.7 (dianions only). The IR spectra showed a shift in the antisymmetric stretching frequency of the COO- groups on increasing the solution pH (Figure 9). A similar shift in the carbonyl antisymmetric stretching frequency was observed in the solidstate spectra of these two amino acids and their salts. It can be observed in Figure 9 that at pH 7.2, when Asp and Glu exist predominantly as anions, the antisymmetric stretching frequency of the COO- groups for the two amino acids differs by 36 cm-1 (viz. 1595 cm-1 for Asp and 1559 cm-1 for Glu). In contrast,

260 Crystal Growth & Design, Vol. 7, No. 2, 2007

Poornachary et al.

Figure 9. Molecular vibration (-COO- antisymmetric stretch) frequencies of the additives at different solution pH. The data points corresponding to pH 2.9 are of the R-COO- group. Lines are drawn to aid the eye.

in the zwitterionic form (pH ∼3.0), little difference is observed in the antisymmetric stretch of the R-COO- groups. Again, the difference of the antisymmetric stretching frequencies of COOgroups between Asp and Glu diminishes at higher pH values in the absence of the influence of NH3+sAt pH 8.9, the difference is only 11 cm-1, and at pH 11.7, it is further decreased to 3 cm-1. This is consistent with the observation made previously20 for the anions of these two monoamino dicarboxylic acids in D2O at neutral pD and hence corroborates the intramolecular charge-transfer interaction. As it was reasoned earlier,20 the nearer the charged NH3+ is to the COO- group, the greater the wavenumber at which the antisymmetric COOabsorption occurs, viz. 1595 cm-1 for Asp anion and 1559 cm-1 for Glu anion. Now, under the influence of such an intramolecular interaction between the charged moieties, it was expected that the anion side chain would tend to orient toward the CR-NH3+ moiety. Indeed, this was reflected by a conformational change of ca. 10° torsion in the side chain fragment of Asp anion within its crystal structures as compared to the zwitterion (Figure 10a,b; molecular structures were extracted from Cambridge Structural DatabasesCSD Ver. 5.26). Besides this conformational change, the other distinguishing feature observed was the formation of an intramolecular H-bridge in the Asp anion molecular structures (in ca. 50% of the hits) between the CR-NH3+ group and the side chain (distal) COO- group, absent in the L-Asp zwitterion (cf. Figure 10a,b). This, in turn, could relate to the intramolecular charge transfer interaction in the L-Asp anion as shown by the IR spectroscopic studies. According to the proposed model for additive interaction with the host crystal (cf. Figure 7b), it could be envisaged that an L-Asp anion with the modified conformation is most likely to have higher repulsive effects along the b-axis, when attempting to dock on the (01h0) face of R-glycine. Furthermore, continuing with this line of thought, a conformational analysis23 was made for the L-Glu anion using the torsions τ1 and τ2 as defined in Figure 10c. It resulted in minimum energy conformers that showed the expected orientation in the anion side chain (in comparison with the zwitterion), due to favorable Coulombic interactions between the CR-NH3+ group and the side chain -COOH group (Figure 10d). These calculations were carried out at molecular mechanics level using DREIDING force field7 and Mulliken charges24 in Materials Studio. However, in a solution environment, it is known that the solvation effect could considerably stabilize the charged moieties through intermolecular H-bonding with water mol-

Figure 10. Molecular structures illustrating conformational changes in the side chain: (a) L-Asp zwitterion in its crystal structure;14 (b) D-Asp anion in L-ornithine D-aspartate monohydrate crystal structure;22 (c) L-Glu zwitterion in its crystal structure;19 and (d) minimum energy conformer obtained from the conformational analysis of L-Glu anion. The intramolecular H-bridge between the CR-NH3+ and the distal COOgroup in the anion molecular structures is represented by the dash lines. The torsions τ1 and τ2 are defined by the carbons C1, C2, C3, and C4 and C2, C3, C4, and C5, respectively, of the glutamic acid molecule.

ecules. Nevertheless, it is reasonable that any considerable amount of conformational change in the side chain of L-Glu anion (as a result of the partial delocalization of negative charge and the subsequent intramolecular charge transfer interaction), could increase the repulsive effects on the (01h0) face of R-glycine. Finally, we investigate the possibilities for relieving the repulsive effects for anion incorporation on the {010} faces of R-glycine. From the conformation analysis made earlier, it was calculated that the energies associated with the changes in the molecular conformation of the anion side chain are of the order of 1.0 kcal/mol for L-Asp and 6.0 kcal/mol for L-Glu. If these energy penalties were to be imparted by relatively subtle but cooperative changes in the torsion angles that describe the orientation of the anions side chain, then the repulsive effects due to close O‚‚‚O contacts could be relieved to a certain extent. However, two of the factors discussed earlier may restrain changes in the molecular conformation of the anion side chain at the crystal interfaces. First, because of highly stereospecific intermolecular interactions between the adsorbed additive and the neighboring molecules at the host crystal surface, the side chain fragment is less flexible.6 Second, the favorable intramolecular charge transfer interactions in the additive anions may also limit changes in their conformation. Summing up, it is gratifying to note that the experimental habit modification in R-glycine crystals correlates well with the speciation data as well as the conformational analysis of the additives. The discussions presented herein highlight that the pKa values (or the partial delocalization factor), the side chain length, and the conformation of the additive anions are significant, in that order, in determining the Coloumbic interactions at the crystal interfaces.

Habit Modification in R-Glycine Crystals

Conclusions In this paper, we have reported new experimental observations demonstrating the stereoselective habit modification in R-glycine crystals with R-amino acids as tailor-made additives. While the previously reported stereoselectivity mechanism in an R-glycine crystal3 confines the additive action to be along the enantiopolar b-axis, the new habit modification with aspartic and glutamic acid indicated additive action along both the b- and the c-axes at different concentration levels. The habit modification was explained based on the solution speciation of the additives and the subsequent interaction of the molecular species (zwitterion and anion) with the crystal faces of R-glycine. The change in additive action from the crystallographic b-axis to the c-axis of R-glycine was proposed to be the result of electrostatic repulsive interactions between the anionic species and the glycine molecule at the enantiotopic faces. Our proposition was successfully corroborated by combining speciation and IR spectroscopic data and molecular modeling. This observation provides a new perspective to our understanding of the molecular recognition process at crystal interfaces, which is crucial from the point of a stereochemical control23,25 of nucleation and growth in molecular crystals. In addition, the phenomenon of molecular speciation of additives and the resulting alteration of its stereoselectivity, as demonstrated here, may provide a plausible explanation to some of the unforeseen effects of impurities on crystal growth in industrial crystallization systems ranging from pharmaceuticals through specialty chemicals. Acknowledgment. We thank Professor Brian Cox and Dr. Simon N. Black of AstraZeneca (Macclesfield, United Kingdom) for helpful discussions on solution speciation and molecular modeling. We thank Geoffrey Dent of the University of Manchester for his help in the spectroscopic studies. We gratefully acknowledge Merck, Sharp, and Dohme (Singapore) for partial financial support to S.K.P. References (1) Morrison, R. T.; Boyd, R. N. Organic Chemistry, 6th ed.; Prentice Hall: Englewood Cliffs, NJ, 1992; 130 pp.

Crystal Growth & Design, Vol. 7, No. 2, 2007 261 (2) Addadi, L.; Berkovitch-Yellin, Z; Domb, N.; Gati, E.; Lahav, M.; Leiserowitz, L. Nature 1982, 296, 21-26. (3) Weissbuch, I.; Addadi, L.; Berkovitch-Yellin, Z.; Gati, E.; Weinstein, S.; Lahav, M.; Leiserowitz, L. J. Am. Chem. Soc. 1983, 105, 66156621. (4) Jonsson, P. G.; Kvick, A. Acta Crystallogr. Sect. B: Struct. Crystallogr. Cryst. Chem. 1972, 28, 1827. (5) Materials Studio Modeling, Version 3.2.0; Accelrys Software Inc. USA. (6) Berkovitch-Yellin, Z. J. Am. Chem. Soc. 1985, 107, 8239-8253. (7) Mayo, S. L.; Olafson, B. D.; Goddard, W. A., III. J. Phys. Chem. 1990, 94, 8897-8909. (8) Rappe, A. K.; Goddard, W. A. J. Phys. Chem. 1991, 95, 3358. (9) CRC Handbook of Chemistry and Physics, 85th ed.; CRC Press: Boca Raton, FL, 2004; p 7-1. (10) Towler, C. S.; Davey, R. J.; Lancaster, R. W.; Price, C. J. J. Am. Chem. Soc. 2004, 126, 13347-13353. (11) Ginde, M.; Myerson, A. S. J. Cryst. Growth 1992, 116, 41-47. (12) Gidalevitz, G.; Feidenhans’l, R.; Matlin, S.; Smilgies, D. M.; Christensen, M. J.; Leiserowitz, L. Angew. Chem., Int. Ed. Engl. 1997, 36, 955-959. (13) Carter, P. W.; Hiller, A. C.; Ward, M. D. J. Am. Chem. Soc. 1994, 116, 944-953. (14) Derissen, J. L.; Endeman, H. J.; Peerdeman, A. F. Acta Crystallogr. Sect. B: Struct. Crystallogr. Cryst. Chem. 1968, 24, 1349. (15) Berkovitch-Yellin, Z.; van Mil, J.; Addadi, L.; Idelson, M.; Lahav, M.; Leiserowitz, L. J. Am. Chem. Soc. 1985, 107, 3111-3122. (16) Weissbuch, I.; Leiserowitz, L.; Lahav, M. AdV. Mater. 1994, 6, 952956. (17) Weissbuch, I.; Lahav, M.; Leiserowitz, L.; Meredith, G. R.; Vanherzeele, H. Chem. Mater. 1989, 1, 114-118. (18) Torbeev, V. Y.; Shavit, E.; Weissbuch, I.; Leiserowitz, L.; Lahav, M. Cryst. Growth Des. 2005, 5 (6), 2190-2196. (19) Lehmann, M. S.; Koetzle, T. F.; Hamilton, W. C. J. Cryst. Mol. Struct. 1972, 2, 225-233. (20) Pearson, J. F.; Slifkin, M. A. Spectrochim. Acta 1972, 28A, 24032417. (21) Mulliken, R. S. J. Am. Chem. Soc. 1952, 74, 811-824. (22) Soman, J; Vijayan, M. Acta Crystallogr. Sect. C: Cryst. Struct. Commun. 1988, 44, 1794. (23) Davey, R. J.; Blagden, N.; Potts, G. D.; Docherty, R. J. Am. Chem. Soc. 1997, 119, 1767-1772. (24) Leach, A. R. Molecular Modeling: Principles and Applications; Addison Wesley Longman: Essex, England, 1996. (25) Weissbuch, I.; Lahav, M.; Leiserowitz, L. Cryst. Growth Des. 2003, 3 (2), 125-150.

CG060273R