Optically Pure Aziridin-2-yl Methanols as Readily Available 1H NMR Sensors for Enantiodiscrimination of α-Racemic Carboxylic Acids Containing Tertiary or Quaternary Stereogenic Centers

Enantiopure aziridin-2-yl methanols 3–7 are used as highly effective sensors for enantiodiscrimination of α-racemic carboxylic acids containing tertiary or quaternary stereogenic centers. A linear correlation between theoretical and observed % ee values for CSA-3 and enantiomerically enriched samples of mandelic acid has been observed, indicating the possible application of these compounds in the ee determination. The free NH and OH groups in 3–7 ensure good recognition.


■ INTRODUCTION
The detection of enantiomeric purity is an important part of synthetic chemistry, pharmacology, biology, food industry, and materials science. 1 Among the methods used to measure the optical purity of chiral compounds, such as HPLC, 2 GC, 3 CD,4 capillary electrophoresis (CE), 5 UV, 6 IR, 7 mass spectrometry, 8 electrophoresis, 9 or fluorescence spectroscopy, 10 NMR spectroscopy proved to be a fast, readily accessible and easy to use an attractive method to study the enantiomeric purity. 11 Socalled chiral solvating agents (CSAs), associating with the racemic sample through noncovalent driving forces such as ion-pairing, hydrogen-bonding, π−π or dipole−dipole interaction, form diastereoisomeric complexes showing differences in the chemical shifts of some signals. The study of the recognition of chiral carboxylic acids and their derivatives are of interest to many research groups due to the fact that such molecules are basic building blocks of many natural products and drug molecules. 12 In the past decades, various CSAs such as chiral and prochiral amines, 13 "calixarene-like" chiral amine systems 14 and other macrocyclic amines and amides, 15 amino alcohols, 16 salene derivatives, 17 crown or aza-crown ethers, 18 Lproline derivatives, 19 BINOL and their derivatives, 20 chiral shift reagents derived from squaramide and indanol, 21 1,2diaminocyclohexane derivatives, 22 and chiral bisthioureas 23 have been reported particularly for mandelic acid, its derivatives, and other α-hydroxy acids. Although variously modified amine systems have been successfully used as CSAs, just one example of an optically active aziridine-derived receptor for the enantiodiscrimination of α-racemic carboxylic acids can be found in the literature. Chiral imines prepared from 1-(2-aminoalkyl)aziridines proved to be effective CSAs for recognition of mandelic acid and its derivatives and Nprotected amino acid. 24 Considering our results 24 and those described by Tan and Lei,19e regarding the use of diphenylprolinol as CSA for enantiodiscrimination of carboxylic acids and based on our experience in the field of the synthesis and catalytic activity in the asymmetric synthesis of chiral aziridines, 25 we decided to prepare a series of chiral aziridin-2-yl methanols to check their action as CSAs toward αracemic carboxylic acids containing tertiary or quaternary stereogenic centers.

■ RESULTS AND DISCUSSION
The chiral aziridines 1−7 were synthesized in a good yield from L-serine, according to the literature ( Figure 1). 26 In order to explore the enantiomeric discrimination ability, the aziridines 1, 2, and 3 were subjected to 1 H NMR analysis with DL-mandelic acid. The NMR experiments were performed with stoichiometric amounts of rac-mandelic acid and CSA (1:1) in CDCl 3 at room temperature. Table 1 shows the values of chemical shift (Δδ) on the C α H proton of mandelic acid after the addition of 1−3, as well as nonequivalences signals corresponding to each enantiomer of the acid (ΔΔδ). The obtained results showed that 1 with an alkyl substituent at the C-2 atom of the aziridine ring indicated very poor recognition (Table 1, entry 1). The N-Tr derivative 2 was completely inactive, whereas (S)-3 with NH and OH groups gave very good recognition, ΔΔδ = 0.094 ppm (Table  1, entries 2 and 3). Considering the above results, it can be assumed that (i) the presence of free NH and OH groups is sufficient to form multiple intermolecular hydrogen bonds between these groups with tested acid, which provide good recognition and that (ii) the upfield change (Δδ < 0) in the position of the signals from the acid suggests deprotonation of the carboxylic group additionally. 13c The above conclusions have been drawn on the basis of previous literature reports. Tan and Lei observed that (S)-diphenyl(pyrrolidin-2-yl)methanol with an unprotected NH group showed more than twice the recognition ability of mandelic acid than its N-benzyl derivative (ΔΔδ = 0.062 ppm for NH and 0.028 ppm for N-Bn derivatives, respectively). 19e We assumed that increasing the steric hindrance at the nitrogen atom should further reduce the possibility of recognition. Having in hands the N-Tr derivative, which is an intermediate in the synthesis of aziridinyl alcohols, we decided to check this thesis. Indeed a trityl derivative 2 having great steric hindrance could not form an effective hydrogen bond with mandelic acid, and the recognition effect was not effective. For this reason, we decided to synthesize NH derivatives 3−7 for our research. As for the hydroxyl group, Tan and Lei have already shown that diphenylprolinol, i.e., a compound containing the OH group, has the ability of enantiodiscrimination of racemic mandelic acid at the level of ΔΔδ = 0.062 ppm, whereas literature data prove that (S)-2-(diphenylmethyl)pyrrolidine without the OH group has a much lower recognition capacity for this acid (ΔΔδ = 0.028). 19a It seems that the presence of the hydroxyl group will increase the possibility of multipoint interactions between aziridinyl alcohols and a carboxylic acid and will promote magnetic anisotropy, thereby improving their chiral recognition ability. We also assume that the formation of multipoint interactions will be favored by nonpolar solvents, while polar solvents will break down the formed agglomerates connected by hydrogen bonds, and thereby reduce recognition. The ΔΔδ values for diastereomeric complexes between racemic mandelic acid and CSA-3 in various solvents are summarized in Table 1. The obtained results confirmed that only nonpolar solvents provide good chiral recognition.
To determine the stoichiometry of the forming complex, 1 H NMR titrations were performed by adding incremental amounts of the most effective receptor (S)-3 to the tubes containing a solution of (±)-MA in CDCl 3 ( Figure 2). Upon gradual addition of (S)-3, the 1 H NMR signal of the C α H proton of racemic MA shifted upfield, and the chemical shift difference between the two enantiomers increased gradually, until the addition of stoichiometric quantities of (S)-3 [(S)-3/ (±)-MA = 1:1] to receive the best chiral recognition showing a 0.094 ppm difference. Subsequent addition of (S)-3 only slightly shifts signals upfield but does not increase the chemical shift difference.
Additionally, the stoichiometry was determined according to the Job's method of continuous variation. Figure 3 shows the Job plots of Δδ*X versus the molar fraction X of (R)-and (S)-MA. A maximum was observed when the ratio of (S)-3 to (R)-or (S)-MA was 1:1 (X = 0.5), which indicates that the (S)-3 and the mandelic acid form a 1:1 complex under these conditions.
After determining the stoichiometry of the complex, we tested the ability of enantiodiscrimination of aziridin-2-yl Averaged between signals from both enantiomers.  Carboxylic acids containing tertiary stereogenic centers 8− 14 were subjected to the first tests, and the results are summarized in Table 2. For easy observation, we have marked the ability to chiral recognition using colors. Green was used for very good values of ΔΔδ ≥ 0.1 ppm, orange for good (0.05 < ΔΔδ <0.1 ppm), yellow for average (0.02 < ΔΔδ < 0.05 ppm), and white for weak (ΔΔδ < 0.02 ppm). Generally, all CSAs 3−7 showed a high ability of enantiodiscrimination for racemic mandelic acid 8 and its derivatives 9−12 ( Table 2). The largest ΔΔδ values of 0.111−0.180 ppm exhibited aziridine 4 used as the CSA, while the p-CF 3 substituted aziridine-alcohol 7 showed the lowest ΔΔδ values from 0.028 to 0.052 ppm. It should be noted that chiral discriminations were also observed for the OCH 3 signals of (±)-12. In the presence of (S)-CSAs 3−7, comparable or higher ΔΔδ values were obtained for protons of the methoxy group compared to ΔΔδ values of α-H signals of this acid. Aliphatic α-racbromopropionic acid 13 in the presence of 3−7 gave poor results of enantiodiscrimination, both for the methine proton C α H and for the methyl group protons.
Although the ΔΔδ values of C α H signals were unsatisfactory, the methoxy, and in particular CH 3 protons, can be well recognized with ΔΔδ values up to 0.106 ppm. Considering the obtained results, in particular for mandelic acid 8 and its derivatives 9−12, it can be assumed that the recognition of these acids is based on the formation of the hydrogen bond between CSAs and the carboxyl group of mandelic acid, and the chemical shift difference is caused by the different shielding effect of CSAs on carboxylic acid. It would seem that the electron-donating group (3) helps the amino group to provide electrons to form a stronger hydrogen bond, thus enhancing the recognition effect. On the contrary, the electron-withdrawing group (7) is not conducive to form a hydrogen bond, and the recognition effect becomes poor. However, the lower ΔΔδ values obtained in the presence of 5 containing the stronger electron-donating OCH 3 group suggest a more complex mechanism of enantiodiscrimination of the tested acids by aziridin-2-yl methanols 3−7. Table 2    and Fu 16c provided better recognition of (±)-8 and (±)-12, respectively.
In the second part of the research, we decided to test the ability of chiral aziridin-2-yl methanols 3−7 as CSAs for enantiomeric discriminating for α-rac-carboxylic acids containing quaternary stereogenic centers (Figure 4, 15−18). On the basis of the available databases, we can conclude that such studies for racemic α-tetrasubstituted acids have not yet been realized. For α-CH 3 -and α-OH-substituted carboxylic acid 15 and 16, better ΔΔδ values for methyl protons were obtained in the presence of chiral sensors 3−5 containing electrondonating groups in the position para of the aromatic ring (∼0.04 ppm for 15 and ∼0.03 ppm for 16) (Table 3), while the (S)-CSAs 6−7 with electron-withdrawing groups were practically ineffective. The α-Meand α-OCH 3 -substituted acid 17 and 18 showed the biggest ΔΔδ values in the presence of all tested CSAs. In particular, high chiral discrimination was observed for the OCH 3 signals of (±)-17 and (±)-18, ΔΔδ = 0.281 ppm for (±)-17 and chiralsensor (S)-4, or ΔΔδ = 0.212 ppm for (±)-18 and (S)-3. It is noteworthy that enantiodiscrimination was observed in several cases for the para-CH 3 substituent in the aromatic ring of acids 16 and 18.
Finally, we demonstrated the practicality of aziridin-2-yl methanols 3−7 as a CSAs for the determination of enantiomeric excess (% ee) of chiral carboxylic acids. Samples containing different ee's of mandelic acid (8) were prepared, and their 1 H NMR spectra in the presence of (S)-3 were measured ( Figure 5). The excellent linear relationship (R = 0.9999) between the gravimetry-determined values and those NMR-determined % ee values was observed ( Figure 6).
Moreover, an experiment with 3 and an enantiomerically enriched sample of 2-methoxy-2-phenylacetic acid (12) showed that aziridin-2-yl methanols 3−7 allow identifying individual enantiomers of carboxylic acids containing tertiary or quaternary stereogenic centers and determining their ratio based on the proton signals from CH 3 or OCH 3 groups (Supporting Information, Figure S80a

■ CONCLUSION
In conclusion, easy to synthesize enantiopure aziridin-2-yl methanols, 3−7, were proven to be effective CSAs for the easy enantiodiscrimination of α-racemic carboxylic acids containing tertiary stereogenic centers. A linear correlation observed between theoretical and observed % ee values indicates the possible application of these compounds for analysis of enantiomerically enriched samples. All performed experiments showed that the unsubstituted NH and OH groups in CSAs 3−7 are sufficient for good recognition of α-chiral acids. Noteworthy, aziridinyl alcohols 3−7 are also very effective sensors for some carboxylic acids containing quaternary stereogenic centers.    Materials. Racemic carboxylic acids used in this protocol, 8−11 and 13, were purchased from Sigma-Aldrich. Other carboxylic acids, rac-12, (S)-12, rac-14−18, and aziridines 1−7, were synthesized by reported procedures.
Determination of Stoichiometry of the Host−Guest Complex (Job plots). Compound (S)-3, and (S)-and (R)-mandelic acid 8 were separately dissolved in CDCl 3 with a concentration of 0.046 mmol/ mL. These solutions were distributed among nine NMR tubes, with the molar fraction X of 8 in the resulting solutions increasing from 0.1 The Journal of Organic Chemistry pubs.acs.org/joc Article to 1.0, and the total concentration of (S)-3 and (S)-and (R)-8 was 0.046 mmol/mL. The complexation induced shifts (Δδ) were multiplied by X and plotted against X itself to afford a 1:1 (host/ guest) complex under these conditions. Typical Procedure for Enantiodiscrimination of rac-Carboxylic Acids 8−18 Using Chiral Sensors 1−7. Sensors 1−7 (0.023 mmol) and carboxylic acid (0.023 mmol) were mixed in 0.5 mL CDCl 3 . Then 1 H NMR was recorded on a 600 MHz spectra at room temperature.
Determination of Enantiomeric Purity of Mandelic Acid 8. To evaluate the accuracy of our determining method, we prepared eight samples containing mandelic acid with 0, 25, 45, 60, and 80% ee (in favor of the S enantiomer) and 15, 45, 70% ee (in favor of the R enantiomer) and determined their enantiomeric purities in the presence of host 3 by using 1 H NMR method. All samples were prepared by adding 1 equiv of host 3 in the solutions of mandelic acid (0.023 mmol in 0.5 mL of CDCl 3 ). The results, which were calculated based on the integrations of the NMR signals, are shown in Figure 5, and the linear correlation between the theoretical and observed% ee values is shown in Figure 6.
Experimental procedures, copies of NMR spectra, and recorded Job plots data (PDF)