Glycine Open Dimers in Solution: New Insights into α-Glycine

Aug 28, 2012 - Gabriele C. Sosso , Ji Chen , Stephen J. Cox , Martin Fitzner , Philipp Pedevilla ... A salt or a co-crystal – when crystallization p...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/crystal

Glycine Open Dimers in Solution: New Insights into α‑Glycine Nucleation and Growth Yin Yani,*,† Pui Shan Chow,† and Reginald B. H. Tan†,‡ †

Institute of Chemical & Engineering Sciences, A*STAR (Agency for Science, Technology and Research), 1 Pesek Road, Jurong Island, Singapore 627833 ‡ Department of Chemical & Biomolecular Engineering, National University of Singapore, 4 Engineering Drive 4, Singapore 117576 ABSTRACT: The traditional view that α-glycine is preferentially nucleated from aqueous solution due to the presence of glycine cyclic dimers as precursors in the solution has been refuted in recent studies, which suggest that glycine molecules are present mainly as monomers in the solution. Our present work investigates the clustering behavior of glycine molecules in supersaturated aqueous solutions using molecular dynamics (MD) simulation, and our results show that glycine molecules exist predominantly as monomers at all concentrations studied for hydrogen-bond criteria of 2.2 Å/160° but that there are always at least 12% of glycine molecules existing as open dimers. Further, MD simulations are performed to study the effect of different building units on the growth of α-glycine crystal in supersaturated solutions. Our results suggest glycine open dimers to be the most favorable growth unit for α-glycine crystal. The presence of open dimers and the predominant monomers that are able to form open dimers in solution provides a convincing explanation for the preferential nucleation and growth of αglycine. 1a) packed in a centrosymmetric packing arrangement.21,22 The (011) faces of α-glycine expose protonated amine and

1. INTRODUCTION Polymorphism is an important phenomenon in pharmaceutical and chemical industries. It occurs when a molecule is able to form different crystal structures.3−6 In the pharmaceutical industry, it is desirable to be able to control the formation of a polymorph because different polymorphs may have different solubility and stability, which may affect the bioavailability of the drug. Factors that can lead to the formation of different polymorphs are temperature, presence of additives,7 types of solvent,2,8 pH of the solutions,9,10 etc. With the same type of solvent, it is still hard to predict whether one polymorph will nucleate or grow faster than the other polymorphs.1 Therefore, it is crucial to understand how all these factors affect the crystallization process to be able to identify the conditions that a desired crystal structure can be obtained. Glycine is a commonly used model compound for polymorphism studies due to its simple structure. It is also widely used in the pharmaceutical industry as the active pharmaceutical ingredient.11 Glycine has three polymorphs with the order of stability: γ > α > β.12−14 Under the ambient conditions, formation of α-glycine polymorph from aqueous solution is in favor compared to the other two polymorphs. The most stable form, γ-glycine, can be crystallized by changing the solution pH,9,12,15 addition of salts or additives,16,17 slow evaporation of microdroplets,18 evaporation of a thin film of neutral aqueous solution on a glass surface,19 application of polarized laser radiation,17 and application of a strong dc electric field.20 The most unstable, β polymorph, however, crystallizes from alcohol in water solution.8 All three polymorphs are formed from zwitterionic glycine molecule (+NH3CH2COO−). α-Glycine consists of zwitterionic glycine cyclic dimers (shown in Figure © XXXX American Chemical Society

Figure 1. (a) Double hydrogen-bonded (cyclic) dimer in the crystal structure of α-glycine polymorph, (b) Single hydrogen-bonded (open) dimer observed in solution.

carboxylate functionality, and they are normally the fastest growing faces resulting in the commonly observed prismatic morphology of a glycine crystal.9 γ-Glycine consists of a helical chain of glycine molecules connected by hydrogen bonds, and β-glcyine consists of hydrogen-bonded molecular sheets in polar organization.1 Chew et al.23 reported that α-glycine grows many times faster than the γ-glycine in aqueous solutions. The selective crystallization of α-glycine from aqueous solution occurs due to two explanations. First, in the solutions, glycine exists mainly as hydrogen-bonded cyclic dimers, which is similar to the Received: April 3, 2012 Revised: August 22, 2012

A

dx.doi.org/10.1021/cg300452n | Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design

Article

structural unit of α-glycine crystal.24−27 Second, glycine cyclic dimers are interpreted to be the favorable growth unit for αglycine crystal.9 Towler et al.9 and Lee et al.28 showed that the nucleation of α-glycine is preferred in solutions at the isoelectric point (pH ≈ 6.0), whereas γ-glycine only nucleates from solutions at high and low pH. Erdemir et al.27 indicated that, in the solution at low pH values, the carboxyl group is protonated to give cations (+H3NCH2COOH), and at high pH values, the amino acid is deprotonated to give anions (H2NCH2COO−). These ions can be the growth units for γglycine and, together with zwitterionic monomers, prevent the formation of dimers and therefore inhibit the formation of αglycine.9 In his study, Erdemir et al.27 utilized small-angle X-ray scattering to examine the clustering behavior of glycine in supersaturated solutions. They found out that the majority of glycine molecules exist as dimers in aqueous solutions. However, recent experimental and simulation studies1,2,29 refuted this idea. Yu and co-workers1 suggested that glycine molecules are present mainly as monomers in the solution. They studied the existence of glycine dimers in supersaturated solutions by measuring the depression of the freezing point and the diffusivity. Their pulsed gradient spin−echo (PGSE) nuclear magnetic resonance (NMR) spectroscopy data showed that the diffusivity of glycine does not decrease with solution age, which contradicts the previous study as the evidence for dimerization of glycine. Nevertheless, it has been reported that the observation on the depression of the freezing point of water due to glycine led to suggestion that there is some degrees of aggregation between the molecules.2,30 Experimental investigation of molecular aggregation in solution is still very challenging. Therefore, molecular simulation techniques have been of great interest for such studies. Campo31 employed a molecular dynamics (MD) simulation to observe the interaction between zwitterionic glycine and water molecules in aqueous solution and the effect of the interaction after the addition of salts. In his study, he only considered very low concentration of glycine: 0.17 and 1.57 mol/dm3. The clustering behavior of glycine molecules was not observed because the lifetimes of glycine−glycine molecules were all shorter than 1 ps. Cheong et al.29 used MD simulation to study the crystal growth of α-glycine from solution. They determined the optimal force field for glycine crystal growth studies by comparing the ability of numerous force fields and charge sets to reproduce different properties of α-glycine bulk crystal and solutions. They then used the optimal force field to study how glycine molecules in solution interact and attach to the α-glycine crystal. However, in their study, they only observed one crystal face, which is (010). Their results showed significant crystal growth at the (010) face of αglycine crystal from a slightly supersaturated glycine solution. Gnanasambandam and Rajagopalan32 employed MD simulation together with the newly developed “extended interface structure analysis” to determine the relative crystal growth rates of (010) and (011) faces of α-glycine crystal and to predict the morphology of the crystal in solution environment. They observed that the growth rate of the (110) face is 2.88 times greater than that of the (010) face, which is consistent with the experimental observations. Banerjee et al. 33 used MD simulations to investigate hydrogen-bond formation and intermolecular interactions of glycine in aqueous solution using Gromos53a6 force field and SPC water. Their simulations included low and high concentrations of glycine. However, the simulations for high concentrations (3.6, 4.5, and 6.0 mol/dm3)

of glycine solutions were not observed in pure solution environment. Instead, glycine solution was placed on the (110) face of an α-glycine crystal to see how the solutions interact with the crystal face. They found that the interaction between atoms of glycine molecules provide interesting insight into the formation of hydrogen bonds, which in turn affect the diffusion coefficient at the interface. Hamad et al.2 used MD simulations to study the clustering behavior of glycine in aqueous solution over a range of temperatures and concentrations. Their results showed evidence of the formation of small hydrogen-bonded clusters of glycine; however, there was no evidence of the formation of double-hydrogen-bonded dimers in the solutions. Despite all the observations of glycine that have been done experimentally or with computer simulation, none of them observe the clustering behavior of glycine in highly supersaturated solution and none of them give a clear explanation of which growth unit is the most favorable one for the growth of α-glycine crystal. Therefore, the main objective for this work is to predict which growth units are more favorable for the growth of α-glycine crystal in highly supersaturated solutions. To achieve this, we analyze two different cases. First, we use MD simulation techniques to observe the clustering behavior of zwitterionic glycine molecules in different concentrations of supersaturated glycine solution. We aim to find out if monomers, cyclic or open dimers exist predominantly in the solutions. Second, we perform MD simulation of α-glycine crystal faces in contact with supersaturated solution to observe how glycine molecules in the solution interact with the various faces of the crystal. This study will give a better understanding on which growth units (monomers, cyclic, or open dimers of glycine) have stronger interaction to certain faces of α-glycine crystal.

2. COMPUTATIONAL DETAILS 2.1. Simulations of the Supersaturated Solution. Dreiding force field in conjunction with the Mulliken charge set was used to model the atomic interactions for glycine. This force field has been shown previously34 to be able to produce the lattice energy of αglycine crystal close to the experimental lattice energy. Water molecules were modeled using this force field. The Dreiding water model gives a density of 996 kg/m3 and diffusion coefficient of 1.7 × 10−9 m2/s for pure water. These values are in good agreement with the values for the experimental density and diffusion coefficient of 997 kg/ m3 and 2.27 × 10−9 m2/s, respectively, for water.35,36 Therefore, these findings have justified the selection of Dreiding force field for water molecules. MD simulations were carried out using the Accelrys Materials Studio (version 5.0)37 for supersaturated solutions of glycine at six different concentrations at the temperature of 298 K. The solutions included 1600 water molecules and 108, 121, 140, 160, 180, and 200 zwitterions glycine molecules (corresponding to concentrations of 3.74, 4.19, 4.85, 5.54, 6.24, and 6.93 mol/dm3, respectively). The velocity Verlet integrator was used to integrate the equations of motion. The integration time step used was 1 fs. Ewald summation was used to enable the long-range interactions. A cutoff radius of 12.5 Å was used for both nonbonded and electrostatic interactions. Simulation in the NPT (constant number of particle, constant pressure, and constant temperature) ensemble was conducted at 298 K and 1 bar for at least 4 ns to ensure that the supersaturated glycine solutions have equilibrated. Figure 2 shows the snapshot of equilibrated supersaturated glycine solution at concentration of 4.85 mol/dm3. Equilibration was determined by observing the change in the thermodynamic properties (energies, temperatures, and densities) as a function of time. Each system was concluded to have reached equilibration conditions if these properties showed sufficiently small variations over time. The required time was about 100 to 150 ps B

dx.doi.org/10.1021/cg300452n | Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design

Article

Å thick vacuum slab was built in above the crystal face. For (01̅1̅) crystal face, it was selected and cleaved out to a depth of four unit cells, and the surface was extended to 4 × 8 unit cells. A 42 Å thick vacuum slab was then built in above the crystal face. The (010) face was selected and cleaved out of the crystal to a depth of two unit cells, and the surface was extended to 8 × 8 unit cells. A 40 Å thick vacuum slab was built in above the crystal face. A supersaturated glycine solution containing glycine in the form of monomers, open dimers, or cyclic dimers at a concentration of 3.74 mol/dm3 was placed on the three different faces of the α-glycine crystal. For dimer cases, all of the two glycine molecules were constrained so that they translate and rotate together as one fragment. An energy minimization procedure was first done for all the cases. MD simulation in the NVT (constant number of particle, constant volume, and constant temperature) ensemble was then conducted at 298 K for at least 200 ps to ensure that each system has equilibrated. During the dynamics simulation runs, only the solutions were movable, and the crystal surfaces were fixed. Energy minimization procedure was again done for the equilibrated system; however, in this case, the glycine molecules on the first layer of crystal surfaces were relaxed. Total interaction energy between glycine molecules in the solution and glycine molecules found in the first layer of certain crystal faces was determined by

E interaction = Etotal − (Egly surface + Egly solution) Figure 2. Snapshot of supersaturated glycine solution at glycine concentration of 4.85 mol/dm3.

(1)

in which, Etotal is the total interaction energy of the system (includes all atoms of the first layer of crystal surface and the glycine molecules in the solution), Egly_surface is the interaction energy of the atoms on the surface, and Egly_solution is the interaction energy of atoms of the glycine molecules in the solution. In addition to the interaction energy, the radial distribution functions based on the atoms of zwitterionic glycine in the solution to the atoms of glycine on the first layer of crystal faces were also demonstrated at 298 K.

depending on the system. The Nose/Hoover38 thermostat and Berendsen barostat39 were used to control the temperature and pressure, respectively. Relaxation time of 1 ps was used for the temperature and pressure coupling. 2.2. Simulations of the Crystal-Supersaturated Solution Interface. Following the observations of supersaturated glycine solution, molecular dynamics simulations of the interface between the crystal face and the supersaturated glycine solution were also carried out. This analysis would enable us to observe whether glycine monomers or dimers are more favaroble to attach to the α-glycine crystal surfaces. The crystal structures were obtained from the Cambridge Structural Database, version 5.26. The space group of the crystal is P21/n (monoclinic with the unit cell dimensions of a = 5.10 Å, b = 11.97 Å, c = 5.46 Å, and β = 111.8°). The crystal habit is typically prismatic bipyramidal as shown in Figure 3, with three faces

3. RESULTS AND DISCUSSION 3.1. Density and Self-Diffusivity. To justify the use of Dreiding force field for the solution in our simulation, its density was observed at different concentration and compared to the experimental data.40 Our simulation underestimated the density values when compared to the experimental densities. However, the trend for both results shown in Figure 5 agrees with each other. Figure 5 also shows that our simulated density values are in good agreement with Cheong et al.29 simulated data obtained using Gromos force field. The diffusion coefficients of glycine molecules in solution at different concentrations were also investigated. This property is important for observing how crystal growth takes place as it determines the mobility of glycine molecules. The selfdiffusivities (D) are calculated from the Einstein relationship

Figure 3. α-Glycine crystal morphology predicted with Dreiding force field in Accelrys Materials Studio.37 ({110}, {010}, and {01̅1̅}) (Figure 4) that are usually well developed.9,21 Only these three faces were considered in our study. The (110) face was selected and cleaved out of the crystal to a depth of four unit cells, and the surface was extended to 8 × 4 unit cells. A 52

D = lim

t →∞

⟨|ri(t ) − ri(0)|2 ⟩ 6t

(2)

Figure 4. Different faces of α-glycine crystal: (a) (010), (b) (110), and (c) (01̅1̅) (red = O, white = H, gray = C, blue = N). C

dx.doi.org/10.1021/cg300452n | Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design

Article

MD simulations is lower than the experimental diffusion coefficient reported. However, the values are all in the same order of magnitude. Therefore, the agreement is considered acceptable. We have also compared our results to the simulation data obtained by Banerjee and Briesen33 using Gromos force field. While our simulated diffusion coefficients are lower than the experimental values, Banerjee and Briesen33 simulated results are overestimated. Quantitative accuracy of diffusion coefficient is difficult to obtain from MD simulation due to the limitation of the number of molecules that can be employed in the simulation. Since the main aim of this study is not to obtain accurate quantitative values of certain properties but to assess and understand the interaction behavior of glycine molecules, the use of Dreiding force field for this work is justified. 3.2. Clustering of Glycine Molecules in Supersaturated Solution. There are two possible types of hydrogen bonds that can be formed between zwiterrionic glycine molecules in solution, i.e., NH···O and CH···O bonds. To analyze how glycine molecules interact with each other in a supersaturated solution, our MD simulations determine the radial distribution functions (RDFs), g(r), based on the hydrogen atoms to the oxygen atoms. RDF describes how the particle density of a system varies with a distance, measured from a reference particle. Figure 7a shows two distinct peaks appearing on the gNH···O plot. This indicates that the strongest interaction that occurred between zwitterionic glycine molecules is the NH···O hydrogen bond, which is the hydrogen bond found in α-glycine crystal structure. These bonds play a major role in determining the glycine crystal structure.41 However, the CH···O interaction shown in Figure 7b is much weaker compared to the NH···O interaction. The highest peak of the gNH···O plot is at the distance of ∼2 Å. The distance of the NH···O hydrogen bond found in the structure of α-glycine crystal is 2.2 Å. To obtain the quantitative insight on the RDF plots, the coordination number was determined by

Figure 5. Density of glycine supersaturated solution at 298 K compared to the experimental data obtained from Dalton et al.40

in which ⟨|ri(t) − ri(0)|2⟩ is the mean square displacement averaging over all the molecules, ri(0) is the position of the molecule at the time origin, ri(t) is the position of the molecule at time t. Figure 6 shows that, in general, the self-diffusivity of glycine decreases with increasing glycine concentration. This is

Figure 6. Diffusion coefficient of glycine in a supersaturated solution at 298 K compared to the experimental data obtained from Huang et al.1 (T = 298 K) and simulation data obtained from Banerjee and Briesen33 (T = 293 K).

expected because the mobility of glycine molecules is hampered by the larger number of molecules present in the solution. In addition to that, the close proximity of the molecules has increased the number of hydrogen bonds formed, which can further hinder the mobility of the molecules. The trend agrees well to the experimental1 and simulation29 results that also show a decrease in self-diffusivity of glycine as its concentration increases. The diffusion coefficient of glycine obtained from

r1

N = 4π

∫r 0

r 2g (r )ρdr

(3)

The coordination number represents the number of nearest oxygen atoms of a glycine molecule that are hydrogen bonded to the hydrogen atoms of another glycine molecule. The first peak of gNH···O (between r0 ≈ 1.7 Å and r1 ≈ 2.7 Å) gives

Figure 7. Comparison of different types of interaction of glycine molecules in the supersaturated solutions at different concentrations: (a) NH···O hydrogen bond and (b) CH···O hydrogen bond. D

dx.doi.org/10.1021/cg300452n | Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design

Article

Figure 8. Percentages of glycine molecules present in monomers, dimers, and bigger clusters, as a function of glycine concentration, in the MD simulation at 298 K, for hydrogen-bond criteria of (a) 2.2 Å/140° and (b) 2.2 Å/160°.

Figure 9. Lifetime distribution of hydrogen bonds formed between zwitterionic glycine molecules in solutions for hydrogen-bond criteria of (a) 2.2 Å/140° and (b) 2.2 Å/160°.

glycine molecules existing as bigger clusters increases. Figure 8 shows that at all glycine concentrations studied, there are about 15−36% (hydrogen-bond criteria of 2.2 Å/140°) and about 20−34% (hydrogen-bond criteria of 2.2 Å/160°) of zwitterions glycine molecules existing as dimers. They are, however, predominantly open dimers (Figure 1b). The presence of the open dimers is justified by the coordination number of ∼1.0, obtained from the first peak of the gNH···O plot in Figure 7a. The lifetime of glycine−glycine hydrogen bonds was also investigated for all supersaturated solution cases. The calculation was averaged over the last 200 ps of simulation time, and the trajectories were stored every 1 ps. Figure 9 shows that the majority of hydrogen bonds formed have a lifetime of about 1− 3 ps. This indicates that the hydrogen bonds typically dissociate in less than 3 ps. However, about 2 ps later, new hydrogen bonds between the dissociate monomer to either the previous bonded glycine monomer or the other adjacent glycine monomers can form again. Because of this reason, therefore, open dimers always exist in solution. It is also shown that, in general, the number of hydrogen bonds formed increases as the glycine concentration increases. The average lifetime of hydrogen bonds (shown in Figure 10) formed between glycine molecules in supersaturated solution is about the same for both hydrogen-bond criteria at different concentrations. The average lifetime of the glycine−glycine hydrogen bonds is longer at all concentrations for the case of hydrogen-bond criteria of 2.2 Å/ 140° than that of hydrogen bond-criteria of 2.2 Å/160°. At the concentration of 4.85 mol/dm3 and above, there is about 0−3%

coordination numbers of 0.89, 0.90, 1.24, 1.06, 1.34, and 1.58 for systems with concentrations of 3.74, 4.19, 4.85, 5.54, 6.24, and 6.93 mol/dm3, respectively. In general, the coordination number increases as the glycine concentration increases. As discussed earlier, α-glycine crystal structure is preferentially nucleated from aqueous solution, and its crystal structure consists of zwitterionic glycine cyclic dimers packed in a centrosymmetric packing arrangement.21,22 Therefore, the preferential nucleation of α-glycine is traditionally thought to be due to the presence of glycine cyclic dimers as precursors in the solution. Our MD simulations determined clustering behavior of glycine molecules in supersaturated solutions. Two specific hydrogen-bond criteria,2 2.2 Å/140° (a maximum H···O distance of 2.2 Å and a minimum N−H···O angle of 140°) and 2.2 Å/160°, were considered in the analysis. The first criteria was chosen because the structure of all three polymoprhs of glycine fall within these limits, and the second criteria was chosen because it gave the best fits to the freezing point depression data obtained from experiment.2,30,42 Our simulation results show that glycine molecules exist predominantly as monomers at all concentrations studied for hydrogen-bond criteria of 2.2 Å/160°. This finding agrees with the experimental1 and simulation2,29 data in the literatures. However, for hydrogen-bond criteria of 2.2 Å/140°, glycine molecules exist predominantly as monomers only at glycine concentration of less than 4.19 mol/dm3. As the glycine concentration increases, the percentage of glycine molecules existing as monomers slightly decreases, and the percentage of E

dx.doi.org/10.1021/cg300452n | Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design

Article

solution, MD simulations were also performed to analyze the effect of different building units on the growth of α-glycine crystal. The interaction energies between glycine molecules that exist as cyclic dimers, open dimers, and monomers to the first layer of α-glycine crystal faces are shown in Table 2. The Table 2. Interaction Energy between Glycine Molecules in the Solution to the Glycine Molecules at the First Layer of αGlycine Crystal total interaction energy (kJ/mol)

Figure 10. Average lifetime of the hydrogen bonds formed between zwitterionic glycine molecules in supersaturated solution as a function of glycine concentration.

faces

of the glycine molecules that exist as cyclic dimers. However, these cyclic dimers are short-lived because the double hydrogen bonds are usually broken within less than 10 ps. Because of the computational limitation that restricts the simulations to time scales on the order of picoseconds,43 it is easier to dissolve a crystal than to nucleate a crystal. Therefore, to observe the clustering of glycine molecules in solution, we have also performed another approach. MD simulation was performed by dissolving α-glycine crystal in a solution at different concentrations of glycine (1.66 mol/dm3 and 2.49 mol/dm3, corresponding to undersaturated solutions, and 3.74 mol/dm3 corresponding to a supersaturated solution). Two different temperatures were considered for two concentrations of glycine (1.66 and 3.74 mol/dm3). The clustering behavior of glycine molecules was studied using hydrogen-bond criteria2 of 2.2 Å/160°. Table 1 shows that, at all concentrations

percentage of glycine molecules temperature (K)

monomers

open dimers

bigger clusters

1.66 1.66 2.49 3.74 3.74

298 303 298 298 303

72.92 66.67 70.83 54.63 54.63

12.50 20.83 16.67 24.07 16.67

14.58 12.50 12.50 21.30 28.70

(110)

(01̅1̅)

(010)

cyclic dimer open dimer monomer

485.44 123.01 481.47

2654.42 3509.10 2612.07

1373.44 262.17 423.05

calculated interaction energies show that the strongest energy for all growth units occurred on the (011̅ )̅ face, which indicates that all growth units have the tendency to attach on (01̅1̅). Therefore, our simulation predicts this face to be the fastest growing face of the α-glycine crystal resulting in a crystal habit that is elongated along the c-axis. This finding is in agreement with the experimental9,21 and simulation32 results reported in the literature. In addition, our simulation results show that on the (110) face, the interaction energy for cyclic dimers is close to the one for monomers. However, glycine open dimers tend to attach stronger to the (01̅1̅) face, followed by cyclic dimers and monomers. On the (010) face, glycine cyclic dimers tend to attach stronger, followed by monomers and open dimers. As discussed earlier, there is only a very small percentage of dimers that exist as cyclic dimers in solution, and the remaining are mostly open dimers at highly supersaturated concentration of glycine (4.85 mol/dm3 and above). All dimers are found to be short-lived, but as disscussed earlier, our simulations show that, when a hydrogen bond of an open dimer dissociates, it can reassociate quickly. The open dimers are also found to have the strongest interaction energy on the fastest growing face (011̅ )̅ of α-glycine crystal. In addition, at the wide range of concentrations studied, there are always at least 12% of glycine molecules that exist as open dimers in solution. Therefore, our simulations suggest glycine open dimers to be the most favorable growth unit of the α-glycine crystal. Even though monomers exist predominantly in solution, from the interaction energy results, they are more favorable to attach to the (110) and (010) faces, which are not the fast growing faces of αglycine crystal. These monomers have the ability to form open dimers which can then be the most favorable growth unit for αglycine crystal. The existence of cyclic dimer is very rare in supersaturated solution, therefore, it is unlikely to be the growth unit of α-glycine crystal from solution. To further analyze the behavior of different growth units on the crystal faces, radial distribution functions, g(r), based on all atoms of zwitterionic glycine molecules in the solution to all atoms of the molecules on the first layer of crystal faces were calculated. Figure 11 shows higher values of the g(r) for monomers on the (110) and (010) faces. This observation agrees with the interaction energy results that show stronger interactions between monomers and these crystal faces. However, there are slightly higher values of g(r) for the glycine open dimers on the (01̅1̅) face, compared to values for glycine

Table 1. Percentages of Glycine Molecules Present in Monomers, Open Dimers, and Bigger Clusters at Various Concentrations of Glycine for Hydrogen Bond of 2.2 Å/ 160° concentration (mol/ dm3)

growth units

considered, glycine molecules exist predominantly as monomers and that there are about 12−24% of open dimers found in the solutions. At the concentration of 2.49 mol/dm3, the percentage of glycine molecules existing as open dimers is close to 17%. This is in reasonable agreement to the data reported by Huang et al.1 that estimated 25% of glycine molecules exist as dimers in the solution at the concentration of 2.9 mol/dm3. However, Huang et al.1 did not provide any information on whether those dimers are cyclic or open dimers. From our results, the existence of cyclic dimers is negligible in the undersaturated solution of glycine. There are only slight differences in the amount of monomers or open dimers appeared in the solution for different temperatures at the same concentration. 3.3. Simulations of the α-Glycine Crystal−Supersaturated Solution Interface. Apart from the study of clustering behavior of glycine molecules in supersaturated F

dx.doi.org/10.1021/cg300452n | Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design

Article

Figure 11. Comparison of the radial distribution functions based on glycine molecules (monomers or open dimers) in solution to the first layer of crystal faces: (a) (110), (b) (011̅ )̅ , and (c) (010).



monomers. This finding again supports the analysis stated earlier, explaining that glycine open dimers act more favorable as the growth unit of α-glycine crystal on the (011̅ )̅ face.

AUTHOR INFORMATION

Corresponding Author

*Tel: (65) 6796 3852. E-mail: [email protected]. Notes

4. CONCLUSIONS We have performed molecular dynamics simulations to observe the clustering behavior of zwitterionic glycine molecules in supersaturated solution at different concentrations of glycine. Our results have shown that glycine molecules exist predominantly as monomers in solution at all glycine concentrations studied for hydrogen bond criteria of 2.2 Å/ 160°. This finding agrees with the experimental results obtained by Huang et al.1 and the simulation results by Cheong et al.29 and Hamad et al.2 Our results also show that at least 12% of zwitterionic glycine molecules exist as open dimers. As the glycine concentration increases, the percentage of glycine molecules existing as monomers slightly decreases, and the percentage of glycine molecules existing as bigger clusters increases. Further MD simulation of α-glycine crystal faces in contact with supersaturated solutions showed that all growth units considered (monomers, cyclic dimers, and open dimers) have the strongest interaction energy to the (011̅ )̅ face. Therefore, it predicts this face to be the fastest growing face of the α-glycine crystal, which results in a crystal habit that is elongated along the c-axis. This is in excellent agreement with the observation of experimental data reported by Towler et al.9 and Weissbuch et al.21 Our simulation results clearly indicate that glycine open dimers are the most favorable growth unit for the α-glycine crystal. The presence of open dimers and the predominant monomers that are able to form open dimers in solution could provide the key insight into the preferential nucleation and growth of α-glycine.

The authors declare no competing financial interest.



REFERENCES

(1) Huang, J.; Stringfellow, T. C.; Yu, L. Glycine Exists Mainly as Monomers, Not Dimers, in Supersaturated Aqueous Solutions: Implications for Understanding Its Crystallization and Polymorphism. J. Am. Chem. Soc. 2008, 130 (42), 13973−13980. (2) Hamad, S.; Hughes, C. E.; Catlow, C. R. A.; Harris, K. D. M. Clustering of Glycine Molecules in Aqueous Solution Studied by Molecular Dynamics Simulation. J. Phys. Chem. B 2008, 112 (24), 7280−7288. (3) Dunitz, J. D. Phase Transitions in Molecular Crystals from a Chemical Viewpoint. Pure Appl. Chem. 1991, 63 (2), 177−185. (4) Bernstein, J.; Davey, R. J.; Henck, J.-O. Concomitant Polymorphs. Angew. Chem., Int. Ed. 1999, 38 (23), 3440−3461. (5) Bernstein, J. Polymorphism in Molecular Crystals; Oxford University Press: Oxford, U.K., 2002. (6) Davey, R. J. Pizzas, Polymorphs and Pills. Chem. Commun. 2003, 13, 1463−1467. (7) Dowling, R.; Davey, R. J.; Curtis, R. A.; Han, G.; Poornachary, S. K.; Chow, P. S.; Tan, R. B. H. Acceleration of Crystal Growth Rates: An Unexpected Effect of Tailor-Made Additives. Chem. Commun. 2010, 46 (32), 5924−5926. (8) Weissbuch, I.; Torbeev, V. Y.; Leiserowitz, L.; Lahav, M. Solvent Effect on Crystal Polymorphism: Why Addition of Methanol or Ethanol to Aqueous Solutions Induces the Precipitation of the Least Stable β Form of Glycine. Angew. Chem., Int. Ed. 2005, 44 (21), 3226− 3229. (9) Towler, C. S.; Davey, R. J.; Lancaster, R. W.; Price, C. J. Impact of Molecular Speciation on Crystal Nucleation in Polymorphic Systems:The Conundrum of γ-Glycine and Molecular Self Poisoning. J. Am. Chem. Soc. 2004, 126 (41), 13347−13353.

G

dx.doi.org/10.1021/cg300452n | Cryst. Growth Des. XXXX, XXX, XXX−XXX

Crystal Growth & Design

Article

on the Polymorphic Outcome. Cryst. Growth Des. 2008, 8 (1), 108− 113. (29) Cheong, D. W.; Boon, Y. D. Comparative Study of Force Fields for Molecular Dynamics Simulations of α-Glycine Crystal Growth from Solution. Cryst. Growth Des. 2010, 10 (12), 5146−5158. (30) Anslow, G. A.; Foster, M. L.; Klingler, C. The Absorption Spectra of Glycine Solutions and Their Interpretation. J. Biol. Chem. 1933, 103 (1), 81−92. (31) Campo, M. G. Molecular Dynamics Simulation of Glycine Zwitterion in Aqueous Solution. J. Chem. Phys. 2006, 125 (11), 114511−8. (32) Gnanasambandam, S.; Rajagopalan, R. Growth Morphology of α-Glycine Crystals in Solution Environments: An Extended Interface Structure Analysis. CrystEngComm 2010, 12 (6), 1740−1749. (33) Banerjee, S.; Briesen, H. Molecular Dynamics Simulations of Glycine Crystal−Solution Interface. J. Chem. Phys. 2009, 131 (18), 184705−8. (34) Poornachary, S. K.; Chow, P. S.; Tan, R. B. H. Effect of Solution Speciation of Impurities on α-Glycine Crystal Habit: A Molecular Modeling Study. J. Cryst. Growth 2008, 310 (12), 3034−3041. (35) Franks, F. A Matrix of Life, 2nd ed.; Royal Society of Chemistry: Cambridge, U.K., 2000. (36) Eisenberg, D.; Kauzmann, W. The Structure and Properties of Water; Oxford University Press: Oxford, U.K., 1969. (37) Materials Studio Modeling, 5.02; Accelrys Software Inc.: San Diego, CA. (38) Frenkel, D.; Smit, B. Understanding Molecular Simulation, 2nd ed.; Academic Press: San Deigo, CA, 2002. (39) Berendsen, H. J. C.; Postma, J. P. M.; van Gunsteren, W. F.; DiNola, A.; Haak, J. R. Molecular Dynamics with Coupling to an External Bath. J. Chem. Phys. 1984, 81 (8), 3684−3690. (40) Dalton, J. B.; Schmidt, C. L. A. The Solubilities of Certain Amino Acids in Water, the Densities of Their Solutions at 25 Degrees, and the Calculated Heats of Solution and Partial Molal Volumes. J. Biol. Chem. 1933, 103, 549−578. (41) Bisker-Leib, V.; Doherty, M. F. Modeling Crystal Shape of Polar Organic Materials: Applications to Amino Acids. Cryst. Growth Des. 2002, 3 (2), 221−237. (42) Lewis, W. C. M. The Crystallization, Denaturation and Flocculation of Proteins with Special Reference to Albumin and Hemoglobin; Together with an Appendix on the Physicochemical Behavior of Glycine. Chem. Rev. 1931, 8 (1), 81−165. (43) Yani, Y.; Lamm, M. H. Molecular Dynamics Simulation of Mixed Matrix Nanocomposites Containing Polyimide and Polyhedral Oligomeric Silsesquioxane (POSS). Polymer 2009, 50 (5), 1324−1332.

(10) Jones, H. P.; Davey, R. J.; Cox, B. G. Crystallization of a Salt of a Weak Organic Acid and Base: Solubility Relations, Supersaturation Control and Polymorphic Behavior. J. Phys. Chem. B 2005, 109 (11), 5273−5278. (11) Ceni, D.; Martins, L.; Pereira, A.; Fröehlich, P.; Bergold, A. An LC Ion-Pairing Method for the Determination of Fe(II) in Ferrous Bisglycinate Pharmaceutical Formulation. Chromatographia 2009, 69 (2), 189−194. (12) Boldyreva, E.; Drebushchak, V.; Drebushchak, T.; Paukov, I.; Kovalevskaya, Y.; Shutova, E. Polymorphism of Glycine: Thermodynamic Aspects. Part I. Relative Stability of the Polymorphs. J. Therm. Anal. Calorim. 2003, 73 (2), 409−418. (13) Boldyreva, E.; Drebushchak, V.; Drebushchak, T.; Paukov, I.; Kovalevskaya, Y.; Shutova, E. Polymorphism of glycine: Thermodynamic aspects. Part II. Polymorphic Transitions. J. Therm. Anal. Calorim. 2003, 73 (2), 419−428. (14) Price, S. L.; Hamad, S.; Torrisi, A.; Karamertzanis, P. G.; Leslie, M.; Catlow, C. R. A. Applications Of Dl_poly And Dl_multi To Organic Molecular Crystals. Mol. Simul. 2006, 32 (12−13), 985−997. (15) Perlovich, G.; Hansen, L.; Bauer-Brandl, A. The Polymorphism of Glycine. Thermochemical and Structural Aspects. J. Therm. Anal. Calorim. 2001, 66 (3), 699−715. (16) Bhat, M. N.; Dharmaprakash, S. M. Effect of Solvents on the Growth Morphology and Physical Characteristics of Nonlinear Optical γ-Glycine Crystals. J. Cryst. Growth 2002, 242 (1−2), 245−252. (17) Garetz, B. A.; Matic, J.; Myerson, A. S. Polarization Switching of Crystal Structure in the Nonphotochemical Light-Induced Nucleation of Supersaturated Aqueous Glycine Solutions. Phys. Rev. Lett. 2002, 89 (17), 175501. (18) He, G.; Bhamidi, V.; Wilson, S. R.; Tan, R. B. H.; Kenis, P. J. A.; Zukoski, C. F. Direct Growth of γ-Glycine from Neutral Aqueous Solutions by Slow, Evaporation-Driven Crystallization. Cryst. Growth Des. 2006, 6 (8), 1746−1749. (19) Xu, M.; Harris, K. D. M. Alteration of Polymorphic Selectivity through Different Crystallization Mechanisms Occurring in the Same Crystallization Solution. J. Phys. Chem. B 2007, 111 (30), 8705−8707. (20) Aber, J. E.; Arnold, S.; Garetz, B. A.; Myerson, A. S. Strong dc Electric Field Applied to Supersaturated Aqueous Glycine Solution Induces Nucleation of the I3̂ Polymorph. Phys. Rev. Lett. 2005, 94 (14), 145503. (21) Weissbuch, I.; Addadi, L.; Berkovitch-Yellin, Z.; Gati, E.; Weinstein, S.; Lahav, M.; Leiserowitz, L. Centrosymmetric Crystals for the Direct Assignment of the Absolute Configuration of Chiral Molecules. Application to the α-Amino Acids by Their Effect on Glycine Crystals. J. Am. Chem. Soc. 1983, 105 (22), 6615−6621. (22) Poornachary, S. K.; Chow, P. S.; Tan, R. B. H. Influence of Solution Speciation of Impurities on Polymorphic Nucleation in Glycine. Cryst. Growth Des. 2007, 8 (1), 179−185. (23) Chew, J. W.; Black, S. N.; Chow, P. S.; Tan, R. B. H.; Carpenter, K. J. Stable Polymorphs: Difficult to Make and Difficult to Predict. CrystEngComm 2007, 9 (2), 128−130. (24) Weissbuch, I.; Lahav, M.; Leiserowitz, L. Toward Stereochemical Control, Monitoring, and Understanding of Crystal Nucleation. Cryst. Growth Des. 2003, 3 (2), 125−150. (25) Torbeev, V. Y.; Shavit, E.; Weissbuch, I.; Leiserowitz, L.; Lahav, M. Control of Crystal Polymorphism by Tuning the Structure of Auxiliary Molecules as Nucleation Inhibitors. The β-Polymorph of Glycine Grown in Aqueous Solutions. Cryst. Growth Des. 2005, 5 (6), 2190−2196. (26) Chattopadhyay, S.; Erdemir, D.; Evans, J. M. B.; Ilavsky, J.; Amenitsch, H.; Segre, C. U.; Myerson, A. S. SAXS Study of the Nucleation of Glycine Crystals from a Supersaturated Solution. Cryst. Growth Des. 2005, 5 (2), 523−527. (27) Erdemir, D.; Chattopadhyay, S.; Guo, L.; Ilavsky, J.; Amenitsch, H.; Segre, C. U.; Myerson, A. S. Relationship between Self-Association of Glycine Molecules in Supersaturated Solutions and Solid State Outcome. Phys. Rev. Lett. 2007, 99 (11), 115702. (28) Lee, I. S.; Kim, K. T.; Lee, A. Y.; Myerson, A. S. Concomitant Crystallization of Glycine on Patterned Substrates: The Effect of pH H

dx.doi.org/10.1021/cg300452n | Cryst. Growth Des. XXXX, XXX, XXX−XXX