Direct Crystallization Resolution of Racemates Enhanced by Chiral Nanorods: Experimental, Statistical, and Quantum Mechanics/Molecular Dynamics Simulation Studies.

Three chiral nanorods of C14-l-Thea, C14-l-Phe, and C14-d-Phe were first synthesized and utilized as heterogeneous nucleants to enhance the resolution of racemic Asp via direct crystallization. Through the statistical analysis from 320 batches of nucleation experiments, we found that the apparent appearance diversity of two enantiomeric crystals of Asp existed in 80 homogeneous experiments without chiral nanorods. However, in 240 heterogeneous experiments with 4.0 wt % chiral nanorods of solute mass added, the appearance of those nuclei with the same chirality as the nanorods was apparently promoted, and that with the opposite chirality was totally inhibited. Under a supersaturation level of 1.08, the maximum ee of the initial nuclei was as high as 23.51%. When the cooling rate was 0.025 K/min, the ee of the product was up to 76.85% with a yield of 14.41%. Furthermore, the simulation results from quantum mechanics (QM) and molecular dynamics (MD) revealed that the higher chiral recognition ability of C14-l-Thea compared to C14-l-Phe that originated from the interaction difference between C14-l-Thea and Asp enantiomers was larger than that between C14-l-Phe and Asp enantiomers. Moreover, the constructed nanorods exhibited good stability and recyclability.

Introduction

Chirality exists universally with diverse forms at all levels.[1] In an achiral environment, mutual enantiomers have the same chemical and physical characteristics but often possess different performances in a chiral environment. Therefore, in many fields, particularly the pharmaceutical industry, it is critical to manufacture pure enantiomeric compounds.[2,3] At present, practical technologies to prepare pure enantiomers include chiral separation[4] and asymmetric synthesis.[5] Although the asymmetric synthesis is the main method for the preparation of enantiomers, it generally suffers from the disadvantage that the product frequently has a lower enantiomeric excess (ee) and accordingly requires further enrichment.[6,7] For some chiral systems, the preparation of enantiomers through the separation of racemates has obvious advantages, particularly including that the multifarious selection of expensive and suitable catalysts is unnecessary.[8] The emerged chiral separation techniques involve membrane separation,[9,10] chromatographic separation,[11] enzymatic kinetic resolution,[12] crystallization,[13,14] liquid–liquid extraction,[15] etc.

Specifically, crystallization is widely utilized to resolve racemates as well as to purify enantiomers in a large scale because its operation is convenient, economical, and widely applicable.[16] Nowadays, the frequently used crystallization resolution methods include preferential crystallization,[17] diastereomer crystallization,[18] spontaneous crystallization,[19] and that using tailor-made additives[20] or nucleants.[21] Among them, the diastereomeric approach is the most verbose to produce optically pure enantiomers by means of sequential steps. On the contrary, direct crystallization for racemates promoted by crystal seeds, additives, or nucleants has been emerging as a promising alternative.[22]

In a crystallization process, nucleation is undisputedly the most key factor, which determines the structure, morphology, and size distribution of final products. Nucleants often play a significant role in the enantioselective nucleation through different mechanisms, for example, selectively adsorbing solute molecules from racemic solutions or melts, being adsorbed on nuclei and crystals, and inducing heterogeneous nucleation of specific species.[23] There is no doubt that suitable nucleants can intensify the direct crystallization resolution of racemates.[21] To date, nucleants such as insoluble chiral polymers as well as their composites,[24] self-assembly surfaces, etc. have been developed to effectively improve enantioselective crystallization.[25−28] In this work, three optically active nanorods of C14-l-theanine (C14-l-Thea), C14-l-phenylalanine (C14-l-Phe), and C14-d-phenylalanine (C14-d-Phe) were first synthesized by amidation reaction and adequately characterized. Selecting aspartic acid (Asp) as a model chiral system, solid-state characterization as well as solubility measurements were then conducted for d-, l-, and dl-Asp crystals. Meanwhile, the metastable zone widths (MSZWs) as well as the induction periods (tind) of three species were compared in the absence/presence of three nanorods as nucleants. The heterogeneous nucleation rates were accordingly obtained by classical heterogeneous nucleation theory (CHNT). After that, the appearance probabilities of enantiomeric and racemic nuclei of Asp from racemate solutions were statistically studied using four supersaturation levels in the absence/presence of chiral nanorods at a nucleation temperature of 303.15 K. Furthermore, under different cooling rates with the addition of 4.0 wt % chiral nanorods of solute mass, the ee values and yields of products from the same racemate solutions were monitored over time. At the same time, the chiral recognition mechanism of the synthesized nanorods to d- and l-Asp was analyzed through quantum mechanics (QM) calculation and molecular dynamics (MD) simulation. Finally, the stability and reusability of three chiral nanorods as heterogeneous nucleants were verified. The direct crystallization resolution of dl-Asp by chiral nucleants in this work was schematically illustrated in Figure .

Figure 1

Flowchart of direct crystallization resolution of dl-Asp by chiral nucleants.

Classical Heterogeneous Nucleation Theory (CHNT)

Based on CHNT, the heterogeneous nucleation rate (Jhet) in the presence of foreign particles is expressed by[29]where B refers to kinetic constant; T is in Kelvin; κ is Boltzmann constant; and ΔGhetcrit means the free energy change during the critical nuclei formation in heterogeneous solutions which can be calculated by[30,31]Here, ϕ stands for the coefficient, and ΔGhomcrit means the free energy change during the critical nuclei formation in clear solutions. They are acquired bywhere θ refers to the contact angle of solution on the foreign particles; S is the supersaturation level in practice represented by the ratio of molar fraction concentration (x) to the solubility (x*) at nucleating temperature; and γhom refers to the solution–nucleus interfacial energy which can be calculated by[32]Here, Vs is the molecular volume, and l is the slope of ln(tind) ∼ 1/ln2S.

In practice, in the presence of foreign particles, the effective interfacial energy (γhet) can be derived from γhom for a pure homogeneous process as[31]Furthermore, the radius of critical nuclei (rhetcrit) as well as the number of molecules in them (nhetcrit) are calculated by[33,34]The Jhet can be finally expressed as[29,35]Here, m refers to the molecular mass of solute; meanwhile, ρ stands for the number of solute molecules in one volume of nucleating solution.

Experimental Section

Materials

d-Aspartic acid (d-Asp, with ee ≥ 99.5%), l-aspartic acid (l-Asp, with ee ≥ 99.5%), dl-aspartic acid (dl-Asp), d-phenylalanine (d-Phe, with ee ≥ 99.5%), l-phenylalanine (l-Phe, with ee ≥ 99.5%), l-theanine acid (l-Thea, with ee ≥ 99.5%), petroleum ether, myristoyl chloride (C14H27ClO), acetone, dimethyl sulfoxide (DMSO), NaOH, and HCl were purchased from Adamas-Beta (Shanghai, China). The ee of the received enantiomers and racemate was double checked by HPLC with chiral columns.

Preparation of Nanorods

Amounts of 4.37 g of l-Thea and 1.00 g of solid-state NaOH were first added into a mixture containing 60 mL of ice water and 40 mL of acetone. Subsequently, excessive myristoyl chloride (7.40 g) was dropped into the mixture at 273.15 K over 1 h under stirring. The stirring was further continued for 2 h after dropping was finished. At the same time, the pH value of the reaction system was kept above 12.0 by the NaOH aqueous solution (1.0 M). The mixture was then acidified to pH = 1.0 with concentrated HCl to precipitate crystalline C14-l-Thea. Finally, the crude crystals were washed using water and petroleum ether, respectively, and dried in a vacuum. C14-l-Phe and C14-d-Phe were also synthesized using the same method.[36]

Characterization of Nanorods

1H NMR spectroscopy was performed on an AS600 spectrometer (Q. One Instruments, Wuhan, China) at room temperature. Reported free-induction decay signals were the average of 32 scans. The recycle delay between scans was set as 10 s.[37] IR was carried out using a 650 spectrometer (Gangdong, Tianjin, China). The scanning wavenumber ranged from 500 to 4000 cm–1.[38] PXRD was operated on a TD-3700 diffractometer with Cu Kα radiation (Tongda Instruments, Dandong, China). The step length from 3 to 80° was 0.05°, and the time per step was 5 s.[39] The thermal analyses by DSC and TGA were conducted on an HSC-4 differential scanning calorimeter (Henjiu Experimental Equipment, Bejing, China) and a WRT-12 thermogravimetric analyzer (Beiguang Hongyuan Instruments, Beijing, China), respectively. The heating rate was 5.0 K per minute.[38] DLS was employed for the measurement of particle size distribution using a Zen 3690 nanosizer with an APD detector (Malvern Panalytical, Malvern, UK). A 4 mW helium–neon gas laser was used as the light source (633 nm). The scattering angle range was set from 30 to 90°.[40] The morphologies of the synthesized nanorods were viewed using a Gemini 300 SEM system (Zeiss, Oberkochen, Germany). Before being observed, the samples were coated by gold in advance.[41] The optical rotations of nanorods were measured on an SGW 532 polarimeter (Wuguang Instruments, Shanghai, China) at 589 nm with an accuracy of ±0.001°. The nanorods were dissolved in DMSO before measurements.[42] The contact angles on nanorods of l-, d-, and dl-Asp saturated aqueous solutions at 298.15 K were measured using an SDC-200S contact angle goniometer (Sindin, Dongguan, China) with the measuring angle range and the accuracy of 0 to 180° and ±0.1°, respectively. During measurements, the droplet volume was 1.0 μL, and the temperature was 296.15 K.[43]

Characterization of Crystalline Enantiomers and Racemates

The solid-state characterization of three species of Asp was carried out by use of IR and PXRD, respectively. The details were described as above.

Molar Fraction Solubility Measurements

The solubilities of three species of Asp in water from 288.15 to 323.15 K were determined, referring to ref (39). Each solubility measurement was performed using 50 mL of double-jacketed crystallizers. First, excessive solid solutes were transferred into the crystallizers, in which 30 mL of water was preadded. The suspensions were at desired temperatures and regulated by an HH-S11.6 water bath (Jingda, Changzhou, China) with magnetic stirring for 48 h to attain liquid–solid equilibrium. After that, the suspensions stood for 24 h without stirring. Amounts of 3.0 mL of supernatants were then collected and filtered, and 1.0 mL of filtrate was moved into vials preweighed by an AUW 120D balance (Shimadzu, Shanghai, China) with an accuracy of ±10 μg. At last, the filtrates as well as the vials were weighed and dried in an DHG-9145A oven (Huitai, Shanghai, China) at a vacuum atmosphere and 318.15 K until the mass stayed unchanged. All measurements were duplicated five times, and the reported solubilities were the average values. To check whether the crystalline forms changed or not during the solubility measurement process, PXRD was utilized to analyze the remaining solids.

Metastable Zone Width Measurement (MSZW)

The MSZWs of three species of Asp in water with or without 4.0 wt % of nanorods to solute were measured using a laser method. A series of saturated solutions were first made up to 30 mL at various temperatures under a stirring of 200 rpm. Then they were cooled using an F25-ME circulator (Julabo, Seelbach, Germany) at 0.20 K/min. Finally the temperatures were recorded as the intensity of the transmitted laser decreased steeply. Each measurement was duplicated four times, and thus the reported MSZWs were their averages.

Induction Period Measurement

For the purpose of further investigating the influence of newly constructed nanorods upon the nucleation of enantiomeric and racemic crystals, the tind values in water with or without 4.0 wt % of nanorods to solute were measured using the above-mentioned laser method at 303.15 K and seven supersaturation levels. First, 20 mL of saturated solutions of three species were prepared at seven temperatures in the presence and absence of 4.0 wt % of nanorods to solute. Second, the solutions were rapidly moved to 30 mL of crystallizers of which the temperature was kept at 303.15 K accompanied by the start of stirring of 200 rpm. The moment the stirring was started, the time should be recorded as 0. The time was recorded again as the intensity of the transmitted laser decreased steeply. Each measurement was duplicated four times, and thus the reported tind values were their averages.

Statistical Analysis of Nucleation Experiments

In our previous work, the competitive nucleation mechanism of the two enantiomers in racemic solutions was elementarily investigated, showing that an actual nucleation environment might allow one enantiomer to nucleate preferentially, and the breakage of chiral symmetry was more pronounced when supersaturation was low.[44] For further verification, 320 batches of nucleation experiments were conducted using dl-Asp aqueous solution with or without 4.0 wt % of nanorods to solute at 303.15 K and four supersaturation levels. As soon as sufficient amounts of nuclei emerged in the solutions, the suspensions were filtered, immediately followed by the fact that the ee’s of cakes were measured by using HPLC. Using pure racemic Asp as the standard sample, the uncertainty of HPLC in the actual measurement was found to be 0.98%. The appearance probability (P) of different types of nuclei was statistically calculated bywhere F was the total batches of isolated experiments, and F* was counted into the following categories in light of the ee of nuclei: (i) the batch was counted into the case of l-Asp being nucleated if it was larger than 0.98%; (ii) the batch was counted into the case of d-Asp being nucleated if it was less than −0.98%; and (iii) the batch was counted into the case only if dl-Asp was nucleated if it was equal to −0.98% to 0.98%.

Direct Crystallization Resolution

First, the solutions with a concentration of 1.29 × 10–2 g of dl-Asp/g of H2O were prepared, filtered, and moved to 500 mL crystallizers, for which the temperature was preset at 318.15 K. After being stirred at 200 rpm for 30 min, the solutions were cooled in the absence and presence of 4.0 wt % of nanorods to solute at 0.025, 0.050, 0.10, and 0.20 K/min, respectively. During the cooling process, the suspensions with or without nanorods were sampled at an interval of 0.5 or 1.0 K; meanwhile, the time as well as the temperature were recorded. Subsequently, the sample suspensions were centrifuged, and the cakes were dried in the oven. The dried solids were dissolved in water, and the mixtures were centrifuged. The cakes were collected, and the ee values of supernatants were measured by use of HPLC. After that, the above experiments were repeated again. Particularly, in the duplicate batches, all suspensions were centrifuged to collect all solids at the times which were determined in the previous batches as the moments the solid maximum ee value emerged, and finally the product yields were calculated by the weight ratio of the precipitated Asp to the feeding Asp.[45]

Recyclability Examination

After each crystallization resolution experiment, the solid was collected by centrifugation and then dissolved in excess water. The obtained suspension was centrifuged again, and the cake was washed with water and centrifuged twice. The collected nanorods were dried under vacuum and stored in a desiccator for the next crystallization resolution. Meanwhile, the recovered nanorods were recharacterized using the method described in section .

HPLC Analysis

The ee values of Asp products from nucleation and crystallization resolution experiments were measured by a LC-20A HPLC system (Shimadzu, Suzhou, China) combined by Crownpak CR(+) series columns (Daicel Chiral Technologies, Shanghai, China) with an uncertainty of 0.98%. An aqueous solution (pH = 1.5) containing 0.27 wt % of perchloric acid was utilized as the mobile phase. The temperature of the column was set at 280.15 K. An UV detector was employed, and it worked at 200 nm.[44]

Simulation Methodology

Quantum Mechanics Calculation

The four structures of d-Asp, l-Asp, C14-l-Thea, and C14-l-Phe were geometrically optimized in water using the Gaussian package at a B3LYP-D3/6-311G** basis set until self-consistency, and the simulation details and results were shown in the Supporting Information (Figures S1 and S2).[46]

Molecular Dynamics Simulation

In order to further reveal the exact chiral recognition mechanism of chiral nanorods as nucleants, MD simulations were adopted to obtain the structures of the heterogeneous nucleation systems as well as the interactions between C14-l-Thea and two Asp enantiomers. First, eight C14-l-Thea molecules were randomly put into a cylinder box at a length to diameter ratio of 10 in light of the SEM image using the PackMol program to obtain a nanorod model. Then, the model nanorods as well as 100 l-Asp, 100 d-Asp, and 50 000 water molecules were randomly mixed in an 8 × 8 × 8 nm3 box using Gromacs (ver. 2016.1).[47−51] The number of l-Asp, d-Asp, and water molecules was chosen on the basis of the molar solubility of dl-Asp in water at 318.15 K. Subsequently, the MD simulation was run by energy minimization to remove false interactions and overlaps under periodic boundary conditions. An Amber force field was employed because it could describe small organic compounds made of P, N, O, H, C, as well as halogen.[52] The water model was described by TIP3P. The precise atomic charge in all species was obtained by the Multiwfn program through electrostatic potential fitting. Before quenching, the system was simulated by NVT for 5 ns to equilibrate with 318.15 K and 1.00 atm. During the simulation for the equilibrium, the position of all species except water was restricted. After that, the nucleation state was established by quenching the system to 303.15 K. The quenching time of NPT was set as 15 ns at a step of 2 fs. The system was energetically minimized to build a new equilibrium. Finally, the structure of the nucleating system after new equilibrium was obtained and visually output by use of VMD 1.9. Meanwhile, the interactions between the model nanorod, namely, C14-l-Thea, with l-Asp and d-Asp, respectively, were analyzed through an average independent gradient model (aIGM) also performed on the Multiwfn 3.8 package, and the aIGM isosurface diagrams were displayed by the VMD 1.9. The cutoff distance for such intermolecular interactions of electrostatic and Lennard-Jones was pregiven as 12 Å. The van der Waals force was rectified by the EnerPres option. The long-range electrostatic interaction was obtained through summation using the particle-mesh Ewald (PME) method. The pressure and temperature were coupled successively by V-rescale and Berendsen algorithms.[53]

Results and Discussion

Characterization of C14-l-Thea, C14-l-Phe, and C14-d-Phe Nanorods

The 1H NMR spectra of three nanorods are shown in Figure S3 (Supporting Information). The hydrogen protons in the spectrograms can be paired with those of the desired products, indicating that the three nanorods have been successfully synthesized.[37] Meanwhile, the yields of C14-l-Thea, C14-l-Phe, and C14-d-Phe are 66.51%, 60.50%, and 64.45%, respectively.

The IR spectra of d-Phe, l-Phe, l-Thea, C14-d-Phe, C14-l-Phe, and C14-l-Thea and myristoyl chloride are presented in Figure , respectively. In Figure a, the peak around 720 cm–1 intimately connected with Cl–C stretching in the acyl halide group of myristoyl chloride is not found in the spectrum of C14-l-Thea, which confirms that Cl has been replaced or removed during the reaction between myristoyl chloride and l-Thea. Besides, the peak around 3329 cm–1 that was intimately connected with H–N stretching in the amino group of l-Thea bathochromically shifts to ca. 3309 cm–1 in the spectrum of C14-l-Thea, implying C14-l-Thea has been synthesized. As shown in Figure b, the peaks around 3451 cm–1 as well as 1645 cm–1 intimately connected with H–N and C=O stretching, respectively, in the amide group arise in the spectrum of C14-l-Phe, and the peak around 720 cm–1 intimately connected with Cl–C stretching in the acyl halide group of myristoyl chloride is not found in the spectrum of C14-l-Phe, which strongly proves that the C14-l-Phe has been successfully prepared. Similarly, as shown in Figure c, C14-d-Phe has been synthesized successfully, which has an identical IR spectrum to that of C14-l-Phe.[54]

Figure 2

Infrared spectra of l-Thea, myristoyl chloride, and C14-l-Thea (a), l-Phe, myristoyl chloride, and C14-l-Phe (b), and d-Phe, myristoyl chloride, and C14-d-Phe (c).

The PXRD diffractograms of three nanorods from 10° to 35° (2θ) are illustrated in Figure . The nanorod of C14-l-Thea possesses such characteristic peaks at ca. 8.6°, 23.7°, 25.2°, etc. The nanorods of C14-d-Phe and C14-l-Phe have identical diffractograms with characteristic peaks at ca. 7.4°, 21.6°, 22.2°, 24.2°, etc. All synthesized nanorods possess a complete crystal structure. Besides, the thermal analyses of nanorods show that the melting points of C14-l-Thea, C14-d-Phe, and C14-l-Phe are 390.41, 340.13, and 338.75 K, respectively (Figure S4, Supporting Information).

Figure 3

PXRD patterns of nanorods.

The SEM images of C14-l-Thea, C14-d-Phe, and C14-l-Phe nanorods in Figure S5 (Supporting Information) show that they are rod-like, and their size distribution curves are illustrated in Figure . Each nanorod with a certain aspect ratio exhibits two peaks with average hydrodynamic diameters of 252.46–294.66 nm and 2.36–2.89 μm, respectively. The former and latter peaks might be associated with the diameter and length of the nanorods, respectively, as their agglomeration can be ignored. Meanwhile, the polydispersity indexes (PDIs) derived from two peaks of three nanorods are 0.10–0.13 and 0.02–0.04, respectively, implying that the three as-prepared nanorods possess uniform sizes.[40]

Figure 4

Size distribution curves of C14-l-Thea, C14-l-Phe, and C14-d-Phe.

The specific optical rotations of C14-l-Thea, C14-d-Phe, and C14-l-Phe nanorods in DMSO at 293.15 K are listed in Table . The concentration of the solution used to calculate the specific rotation is 4 mg/mL. The average specific optical rotations of the three nanorods are −2.65, −18.13, and 18.40, respectively.[42]

Table 1

Specific Optical Rotation Values of Nanorods in DMSO at 293.15 K

	[α]D20
nanorods	1	2	3	4	5	6	average	Ra
C14-l-Thea	–2.64	–2.64	–2.64	–2.64	–2.67	–2.67	–2.65	0.015
C14-l-Phe	–18.79	–18.66	–18.39	–17.97	–17.66	–17.28	–18.13	0.594
C14-d-Phe	18.50	18.50	18.20	18.50	18.50	18.20	18.40	0.154

R is the standard deviation.

The contact angles of l-, d-, and dl-Asp saturated aqueous solutions at 298.15 K on the nanorods are listed in Table , which are acquired from the snapshots supplemented in the Supporting Information as Figure S6. All three nanorods exhibit hydrophobicity in which C14-d-Phe is the most hydrophobic with maximum l-Asp, d-Asp, and dl-Asp contact angles of 124.51°, 120.96°, and 118.94°, respectively, whereas C14-l-Thea is the least hydrophobic with those of 101.98°, 102.67°, and 103.18°, respectively. In theory, C14-l-Phe and C14-d-Phe are enantiomers, and the contact angles of dl-Asp solutions on them should be the same. In practice, the contact angles should also be related to the morphology and size of particles, which leads to the deviation. Furthermore, the difference in hydrophobicity of three nanorods is certainly related to the fact that different functional groups exist on their surfaces.[43]

Table 2

Contact Angles between Nanorods and Solutions

	contact angles (deg)
nanorods	l-Asp	d-Asp	dl-Asp
C14-l-Thea	101.98	102.67	103.18
C14-l-Phe	123.52	121.45	117.53
C14-d-Phe	124.51	120.96	118.94

Chiral Nature of Studied Model Systems

The IR and PXRD spectrograms of enantiomeric and racemic Asp crystals are presented in Figure S7 (Supporting Information), which are consistent with those reported by Lee et al.[55] and Pinto et al.[56]

The solubilities of three species in water are shown in Table , which is in accordance with the work of Wu et al.[57] All three substances possess positive solubility characteristics, but in general under the same temperature the solubility of dl-Asp is about 1.5 times larger than those of l-Asp and d-Asp, which are the same. In addition, by comparing the PXRD patterns of solid residuals after equilibrium with those of raw materials (Figure S8, Supporting Information), there is no crystalline form change in the solubility measurement.

Table 3

Solubilities of Three Asp Species in Water

	103x*
T (K)	l-Asp	d-Asp	dl-Asp
288.15	0.436	0.435	0.667
293.15	0.528	0.525	0.807
298.15	0.636	0.635	0.972
303.15	0.768	0.768	1.155
308.15	0.910	0.909	1.376
313.15	1.091	1.102	1.649
318.15	1.320	1.334	1.931
323.15	1.509	1.523	2.282

The early identification of its racemic nature of a racemate is extremely vital for the selection and design of its practical crystallization resolution. The IR spectra, PXRD diffractograms, and molar fraction solubilities of three species of Asp illustrate that the chiral nature of dl-Asp belongs to a racemic compound, which is in accordance with the work of Lee and Lin who have reported the ee value at one of the eutectic points of the racemic system to be ca. 67%.[55]

Metastable Zone Widths

Figure depicts a series of MSZWs of three species of Asp in water with or without 4.0 wt % of nanorods to solute. In the pure water, when being cooled from the same saturation temperature and at the same cooling rate, the average MSZWs are 17.56 K for l-Asp, 16.14 K for d-Asp, and 13.99 K for dl-Asp, respectively (Figure a). Correspondingly, they decrease to 8.67, 9.34, and 9.46 K in the presence of C14-l-Thea (Figure b) and to 10.26, 10.76, and 11.12 K in the presence of C14-l-Phe (Figure c), respectively. Similarly, as shown in Figure d, when C14-d-Phe is added, they decrease to 10.73, 10.36, and 11.58 K, respectively.

Figure 5

MSZWs of three Asp chiral species in water (a), with C14-l-Thea (b), C14-l-Phe (c), and C14-d-Phe (d).

In general, the average MSZWs of three species become narrow when the nanorods are added as nucleants.[58] The nucleation barrier of Asp species is reduced, and the solute molecules are easier to nucleate in the presence of nanorods. In particular, the average MSZWs of l-Asp become narrower than those of d-Asp when using l-nanorods as nucleants, and vice versa. That is, the chiral nanorods can mostly promote the nucleation of those enantiomeric crystals having an identical chirality to the rods’, which shall provide more opportunities to acquire pure enantiomeric crystals when using them as nucleation promoters. Furthermore, the promotion degree by C14-l-Thea to the nucleation of l-Asp is larger than that by C14-l-Phe, which contributes to a smaller contact angle of l-Asp solution on the former surface, leading to a smaller nucleation barrier than C14-l-Phe.

Induction Period

As shown in Figure and Table S1 (Supporting Information), the tind values of three chiral species generally decrease with increasing S, and in the absence of nanorods, under the same S, the tind values of enantiomeric crystals are almost the same but larger than that of racemic crystals; that is, dl-Asp has the narrowest MSZW and the shortest tind.

Figure 6

Influence of S on the tind of Asp in water (a), with C14-l-Thea (b), C14-l-Phe (c), and C14-d-Phe (d) at 303.15 K.

In addition, in the presence of nanorods, the tind values of all species are shortened. At each S, the tind order is d- > l- > dl-Asp when C14-l-Thea and C14-l-Phe are presented, suggesting that l-nanorods are more promotive to the nucleation of l-crystals than that of d-crystals. Likewise, when C14-d-Phe is presented, the order of tind changes to l- > d- > dl-Asp, indicating that d-nanorods are more promotive on the nucleation of d-crystals than that of l-crystals. The results are in accordance with those of MSZW measurements.[59]

Heterogeneous Nucleation Rates

To further study the effect of three as-prepared chiral nanorods on the nucleation of various Asp crystals, their primary nucleation parameters in the absence/presence of the chiral nanorods have been obtained using CHNT. The scatter plots of ln(tind) over 1/ln2S in the absence/presence of nanorods are illustrated in Figure . It is obvious that when tind is measured in the pure water with high supersaturation, the scatter plots of ln(tind) over 1/ln2S can be well fitted into a straight line to obtain the slope. Nevertheless, when the supersaturation is at low levels or in the presence of nanorods, the scatter plots of ln(tind) over 1/ln2S are incapable of being fitted into straight lines because in this case the measured tind values are generally inaccurate. In this work, the heterogeneous nucleation rates in the presence of nanorods have been calculated using eqs –9. Table lists the heterogeneous nucleation parameters at 303.15 K and different S values.

Figure 7

Plots of tind versus S in the absence/presence of nanorods: l-Asp (a), d-Asp (b), and dl-Asp (c).

Table 4

Heterogeneous Nucleation Parameters at 303.15 K by Using the Nanorods

nucleant	crystal	ϕ	S	ΔGhomcrit (J)	ΔGhetcrit (J)	nhetcrit	rhetcrit (nm)	γhet (J/m2)	jhet (#/(m3·s))
C14-l-Thea	l-Asp	0.653	1.99	7.72 × 10–21	5.04 × 10–21	5.09	0.543	0.00593	1.77 × 1023
			2.26	6.52 × 10–21	4.26 × 10–21	3.06	0.458		3.01 × 1023
			2.62	5.52 × 10–21	3.60 × 10–21	1.86	0.388		4.78 × 1023
	d-Asp	0.662	1.99	7.72 × 10–21	5.11 × 10–21	5.16	0.546	0.00596	1.74 × 1023
			2.26	6.52 × 10–21	4.31 × 10–21	3.10	0.460		2.97 × 1023
			2.62	5.52 × 10–21	3.65 × 10–21	1.88	0.390		4.73 × 1023
	dl-Asp	0.668	1.98	7.71 × 10–21	5.15 × 10–21	5.28	0.553	0.00589	3.92 × 1023
			2.28	6.39 × 10–21	4.27 × 10–21	3.00	0.458		6.48 × 1023
			2.64	5.43 × 10–21	3.62 × 10–21	1.84	0.389		1.01 × 1024
C14-l-Phe	l-Asp	0.872	1.99	7.72 × 10–21	6.73 × 10–21	6.78	0.598	0.00653	1.24 × 1023
			2.26	6.52 × 10–21	5.68 × 10–21	4.09	0.505		2.25 × 1023
			2.62	5.52 × 10–21	4.81 × 10–21	2.48	0.427		3.76 × 1023
	d-Asp	0.856	1.99	7.72 × 10–21	6.61 × 10–21	6.67	0.594	0.00649	1.27 × 1023
			2.26	6.52 × 10–21	5.58 × 10–21	4.01	0.502		2.29 × 1023
			2.62	5.52 × 10–21	4.72 × 10–21	2.43	0.425		3.82 × 1023
	dl-Asp	0.897	1.98	7.71 × 10–21	6.92 × 10–21	7.09	0.610	0.00649	2.70 × 1023
			2.28	6.39 × 10–21	5.73 × 10–21	4.03	0.506		4.80 × 1023
			2.64	5.43 × 10–21	4.87 × 10–21	2.47	0.429		7.91 × 1023
C14-d-Phe	l-Asp	0.879	1.99	7.72 × 10–21	6.79 × 10–21	6.85	0.600	0.00655	1.22 × 1023
			2.26	6.52 × 10–21	5.73 × 10–21	4.12	0.506		2.22 × 1023
			2.62	5.52 × 10–21	4.85 × 10–21	2.50	0.428		3.73 × 1023
	d-Asp	0.852	1.99	7.72 × 10–21	6.58 × 10–21	6.64	0.594	0.00648	1.28 × 1023
			2.26	6.52 × 10–21	5.55 × 10–21	3.99	0.501		2.31 × 1023
			2.62	5.52 × 10–21	4.70 × 10–21	2.42	0.424		3.84 × 1023
	dl-Asp	0.835	1.98	7.71 × 10–21	6.44 × 10–21	6.59	0.596	0.00634	3.00 × 1023
			2.28	6.39 × 10–21	5.33 × 10–21	3.75	0.494		5.21 × 1023
			2.64	5.43 × 10–21	4.53 × 10–21	2.30	0.419		8.47 × 1023

At each S without nanorods, the nucleation rates of l- and d-crystals are nearly equal, and both are less than that of dl-crystals. In the case that the nanorods are presented, the ΔGhetcrit of the three crystals decreases, resulting in increased nucleation rates which increase with S. However, different nanorods have different degrees of change in the nucleation rates of the three crystals. When C14-l-Phe or C14-d-Phe is added, the nucleation rate order of l-, d-, and dl-Asp crystals from their respective supersaturated solutions is dl- > d- > l-Asp. However, when C14-l-Thea is added, the order becomes dl- > l- > d-Asp. Moreover, when C14-l-Thea is used as the nucleant, the nucleation rates of enantiomeric and racemic crystals are higher than those using the other two nucleants. It might be because the hydrophilicity of C14-l-Thea to l-, d-, and dl-Asp solutions is the highest, leading to it having the most promotion on nucleation.

Statistical Analyses

The statistical results on the homogeneous nucleation in the racemate solutions at different S are shown in Figure and Table S2 (Supporting Information). At S values of 1.67, 1.43, 1.19, and 1.08, the P of the l-enantiomeric nuclei is 0%, 5%, 10%, and 5%, respectively, and that of the d-enantiomeric nuclei is 35%, 50%, 65%, and 70%, respectively. Besides, the P of the racemic nuclei is 65%, 45%, 25%, and 25%, respectively. In the initial precipitates, the amount of dl-Asp is far greater than that of d-Asp or l-Asp, resulting in very low ee values of the collected solids. Generally, the nucleation of enantiomeric crystals accompanied by the appearance of racemic nuclei remains highly random. In addition, by comparing the nucleation probabilities under the four S, it is found that S has some effect on the P, and the influence law is the same as the results of our previous work.[44] That is, the appearance probability of d-enantiomeric nuclei increases with the decline of S, and vice versa.

Figure 8

ee of initial particles collected from homogeneous nucleation experiments at different S: 1.67 (a), 1.43 (b), 1.19 (c), and 1.08 (d).

The statistical results on the heterogeneous nucleation in the racemate solutions with nanorods at different S are shown in Figure and Table S3 (Supporting Information). When nanorods are presented, the diversity vanishes, and P increases to 100% at low S. Specifically, when l-nanorods are added, the nucleation of l-Asp crystals is completely promoted, and the d-nanorod has the same promoted to d-Asp crystals. In addition, the ee value of nucleation products increases significantly with the decline of S. This may be because at low S the smaller the nucleation driving force is, the slower the nucleation rate becomes, the longer the time is for the chiral recognition, and the higher the optical purity of the nuclei achieved. Specifically, at the same S, when C14-l-Phe is added, l-Asp is induced to preferentially nucleate, and d-Asp is also preferentially nucleated when C14-d-Phe is used. By comparing the statistical results of those using C14-l-Thea and C14-l-Phe, the ee value of nuclei induced by C14-l-Thea is generally larger than that by C14-l-Phe. It is probably because C14-l-Thea has an additional amide (CONH) group compared to C14-l-Phe, which provides more hydrogen-bond acceptors and donors to intensify the intermolecular interaction between C14-l-Thea and l-Asp, resulting in a higher chiral recognition ability.[60]

Figure 9

ee of initial particles collected from heterogeneous nucleation experiments at different S: 1.67 (a), 1.43 (b), 1.19 (c), and 1.08 (d).

Direct Crystallization Resolution

The ee values of solids over time crystallized from dl-Asp aqueous solutions with or without three nanorods at four cooling rates are shown in Figure , and the raw data of ee values as well as the yields are listed in Tables S4, S5, S6, S7, and S8 (Supporting Information), respectively. At all used cooling rates, when dl-Asp solutions are cooled without nanorods, spontaneous crystallization can not resolve or even partially resolve the racemic Asp. While dl-Asp solutions are cooled with nanorods, all ee values are significantly increased. In particular, the ee values of products first rise and then fall with time under different cooling rates. In the first stage, the added chiral nanorods act as nucleation centers with enantioselectivity, and the adsorbed enantiomers accordingly become seeds, resulting in the increase in the ee value of the product. When the nucleated dl-crystals are much more than enantiomeric crystals, the ee value starts to decrease and finally is close to zero at the completion of crystallization.[21]

Figure 10

ee of solid products over time at cooling rates of 0.025 (a), 0.05 (b), 0.10 (c), and 0.20 K/min (d).

Besides, the ee value of the product from a direct crystallization resolution with C14-l-Thea is larger than that with C14-l-Phe, which is certainly because the former has a higher enantioselective capability than the latter.[61] Meanwhile, it is also observed that the slower the cooling rate employed, the higher the ee value of the product achieved. This may be because the slower the cooling rate adopted, the smaller the nucleation driving force becomes, which is beneficial to the nucleation of the enantiomeric crystals having an identical chirality to the rods. When the cooling rate is 0.025 K/min and C14-l-Thea is used as the nucleant, a solid product containing mainly l-Asp with a maximum ee of 76.85% and a yield of 14.41% is directly crystallized from dl-Asp solutions.

A snapshot of all molecules with water in the nucleating system in the final state simulated by MD is shown in Figure a, and blue, yellow, and red molecules refer to C14-l-Thea, l-Asp, and d-Asp, respectively. In order to exhibit the aggregation state between C14-l-Thea and l-Asp or d-Asp more intuitively, C14-l-Thea and l-Asp or d-Asp are extracted and shown in Figure b and 11c, respectively. By comparing Figures b with 11c, the C14-l-Thea is slightly closer to l-Asp than to d-Asp, showing that the interactions between C14-l-Thea and l-Asp may be larger than those between C14-l-Thea and d-Asp.

Figure 11

Snapshots in the final state of C14-l-Thea, l-Asp, and d-Asp (a), C14-l-Thea and l-Asp (b), and C14-l-Thea and d-Asp (c) in the nucleating system.

The average weak interactions based on MD simulations can provide much more accurate and smoother isosurfaces of C14-l-Thea with l-Asp or d-Asp than that on QM calculations (Figure S2, Supporting Information).[62] The aIGM isosurface diagrams are shown in Figure in which van der Waals interaction, steric repulsion, as well as hydrogen bonding are in green, blue, and red, respectively. In particular, compared with Figure b, more blue domains emerge in Figure a, diagrammatically presenting that stronger interactions exist between C14-l-Thea and l-Asp than C14-l-Thea and d-Asp.

Figure 12

AIGM isosurface diagrams of C14-l-Thea with l-Asp (a) and C14-l-Thea with d-Asp (b). The isosurface of aIGM = 0.1.

Recyclability Examination

The reusability of the nanorods can be evaluated by their recovery ratios after each direct crystallization resolution experiment and the ee values of products using the recovered nanorods as nucleants. As shown in Figure , the ee values of products declined from 10.9% to 12.0% after six cycles, and the average recovery ratios of nanoroads are around 97.7% per cycle, which confirms that the nanorods have good reusability. In addition, the ee values of products decline after six cycles, which may be due to the adsorption of impurities on the nanorods. Meanwhile, the morphologies and diffractograms of the recovered nanorods after six cycles recharacterized by PXRD and SEM, respectively (Figures S9 and S10, Supporting Information), are consistent with those of the fresh ones, further proving that the nanorods can keep stable during the direct crystallization resolution operation.[63]

Figure 13

Effect of cycle number on the ee values of products using recovered nanorods (a) and their recovery ratios (b).

Conclusions

In this work, three optically active nanorods of C14-l-Thea, C14-l-Phe, and C14-d-Phe have been synthesized and used for the direct crystallization resolution of dl-Asp as nucleants. The nucleation experiments and their statistical analysis show that the synthesized chiral nanorods could affect the nucleation parameters and rates of the three species of Asp to a certain extent. In particular, the chiral nanorods can mostly promote the nucleation of enantiomeric crystals with the same chirality and inhibit that of those with opposite chirality to them. The simulation results reveal that a higher chiral recognition of C14-l-Thea than C14-l-Phe to Asp enantiomers is because the difference in hydrogen bonding between C14-l-Thea and Asp enantiomers is larger than that between C14-l-Phe and Asp enantiomers. When 4.0 wt % of C14-l-Thea to solute is added, the ee value of the product crystallized from the dl-Asp solution can be up to 76.85% with a yield of 14.41% under slow cooling. Moreover, the constructed nanorods exhibit good stability and recyclability. The research demonstrates that the chiral nanorods can enhance the direct crystallization resolution of racemic compounds when they are used as the nucleants.