Visualization of enantioselective recognition and separation of chiral acids by aggregation‐induced emission chiral diamine

Enantioselective recognition and separation are the most important issues in the fields of chemistry, pharmacy, agrochemical, and food science. Here, we developed two optically active diamines showing aggregation‐induced emission (AIE) that can discriminate 5 kinds of chiral acids with high enantioselectivity. Especially, a very high fluorescence intensity ratio (Il/Id) of 281 for (±)‐Dibenzoyl‐d/l‐tartaric acid was obtained through the collection of fluorescence change after interaction with chiral AIE‐active diamine. By virtue of AIE property and intermolecular acid‐base interaction, enantioselective separation was facilely realized by simple filtration of the precipitates formed by chiral AIE luminogen (AIEgen) and one enantiomer in the racemic solution. The chiral HPLC data indicated that the precipitates of AIEgen/chiral acid possessed 82% l‐analyte (the enantiomeric excess value was assessed to be 64% ee). Therefore, this method can serve as a simple, convenient, and low‐cost tool for chiral detection and separation.

existed sensors are often applied to chiral detection but rarely used for enantioseparation. [31] Over the past decades, tremendous efforts have been made to develop optically pure molecules mainly through asymmetric synthesis and enantioselective separation. Although asymmetric synthesis has made a significant progress, [32,33] most of the commercially chiral compounds are still obtained by enantioseparation due to its advantages of easy operation, inexpensive, and high reliability for mass production. Now, several enantioseparation tactics were developed, for example, stereoselective crystallization, [34][35][36] enzyme controlled separation, [37] and enantiomer-selective-magnetization, [38] selective aggregation, [31] chromatography. [39][40][41] Undoubtedly, selective crystallization is one of the most inexpensive and convenient technique to realize mass production. [31,38] However, the separation efficiency always needs to be analyzed by HPLC. Here, if we use the fluorescence technique to visualize the resolution process by selective complexation with one enantiomer in the racemate, it will greatly simplify the operation process, reduce cost and increase the separation efficiency. Unfortunately, traditional fluorophores with planar structures often emitted strong fluorescence in the molecularly dissolved state, but faint emission in the condensed phase, [42][43][44] which dramatically inhibited their real world application.
Here, we reported two TPE-based chiral probes bearing optically pure (1R, 2R)-or (1S, 2S)-diaminocyclohexane named as (1R, 2R)-TM or (1S, 2S)-TM, which presented excellent enantiomeric discrimination for 5 pairs of chiral acids. Especially, (1R, 2R)-TM can enantioselectively aggregate with dibenzoyl-L-tartaric acid from a pair of enantiomers to emit bright fluorescence accompanied by a PL intensity ratio (I L /I D ) up to 281. Such big differences in PL and morphologies can be used to visualize the chiral separation process. Chiral HPLC analysis demonstrated that the precipitates were composed of 82% dibenzoyl-L-tartaric acid. Moreover, their sensing mechanism was investigated by NMR titration and 2D NOESY spectrum. Therefore, it is a promising strategy to enantioselectively separate optically pure chemicals by chiral AIEgens.

RESULTS AND DISCUSSION
Generally, chiral fluorescent probes were composed of fluorophore and chiral source. TPE was wildly utilized to build chemo/bio-sensors. [1,46] It is anticipated that AIEgens decorated with (1R, 2R)/(1S, 2S)-(±)-cyclohexanediamine will endow the probes with the ability of enantioselective recognition of chiral acids through acid-base interaction.

Photophysical properties
The absorption spectra of (1R, 2R)/(1S, 2S)-TM were measured in tetrahydrofuran (THF). The as-prepared chiral AIEgens showed a main absorption peak at 311 and 241 nm ( Figure S17A), which come from intramolecular charge transfer and π-π transition, respectively. CD spectra were also recorded in THF. As Figure S17B indicated, (1S, 2S)-TM presented three positive CD peaks at ∼317, ∼284, and ∼231 nm, while (1R, 2R)-TM exhibited negative ones with identical CD intensity. Their AIE behaviors were validated in the THF/water solution by PL spectroscopy. (1R, 2R)-TM showed weak fluorescence in THF. When adding water to THF, the fluorescence intensity had no obvious change from 0 to 80% water fraction ( Figure 1). Further increasing the water fraction, a sky-blue fluorescence was emerged at 477 nm. When raising the water fraction to 95%, the PL intensity was enhanced by 211-fold in comparison with the THF solution.

Enantioselective recognition
Enantioselective recognition of chiral acids by (1R, 2R)-TM proceeded in the mixture of THF and water. As shown in Figure 2, Figure S19, and Table S1 Figure 2D). These results implied that (1R, 2R)-TM may be served as chiral selectors for enantioselective separation just through a simple filtration.
As a control experiment, Di-p-toluoyl tartaric acid and tartaric acid were chosen as chiral analytes to study their enantioselective recognition behaviors by (1R, 2R)-TM. As illustrated in Figure 2B, Figure S19D, and Table S1, (1R, 2R)-TM exhibited good discrimination for Di-p-toluoyl tartaric acid. After complexation with L-Di-p-toluoyl tartaric acid, tiny aggregates were formed with brightly sky-blue fluorescence. In comparison, the mixture of (1R, 2R)-TM and D-Di-p-toluoyl tartaric acid only exhibited weak fluorescence in the solution. The fluorescence intensity ratio (I L /I D ) was up to 23. While, for tartaric acid, there is no obvious fluorescence change upon interaction with (1R, 2R)-TM neither in THF solution nor in the mixed solvents, indicating that phenyl rings in Dibenzoyltartaric acid and Di-p-toluoyl tartaric acid played vital role in enantioselective recognition. Additionally, (1R, 2R)-TM can also respond to other chiral acids with good enantioselectivity, such as mandelic acid (I R /I S = 2.7), 2-chloromandelic acid (I R /I S = 2.1), malic acid (I L /I D = 2.0) ( Figure S19 and Table S1). Here, it is worth noting that the fluorescence of (1R, 2R)-TM was significantly enhanced in the presence of (−)-Dibenzoyl-L-tartaric acid or L-Di-p-toluoyl tartaric acid. However, for mandelic acid, chloromandelic acid, malic acid, and tartaric acid, the fluorescence intensity obviously quenched compared to the free (1R, 2R)-TM after interaction. It is inferred that (1R, 2R)-TM could closely combine with (−)-Dibenzoyl-L-tartaric acid or L-Di-p-toluoyl tartaric acid with suitable configuration and molecular size to form compact complex, which will restrict the free motion of AIEgens to activate the radiative transition, thus emitting bright fluorescence. For other chiral acids in Figure S19, we think the small stereospatial structure cannot well match with (1R, 2R)-TM, thus forming a loose complex with weak emission. Moreover, for those acids, there may exist charge transfer between the AIEgen and acids to result in decreased emission. From our experience, the solvents also played important role in enantioselective recognition. Because different chiral analytes have various polarity and solubility, to achieve optimal enantioselectivity for analytes, it should be measured case by case to study the solvent proportions and concentration. Of course, the recognition could be tested in same solvent condition, but the fluorescence intensity ratio can't always reach the maximum in a single solvent system for all chiral analytes. Therefore, the solvent proportions and concentrations are different for corresponding analytes.

Study of enantioselective recognition mechanism
To reveal the possible sensing mechanism, scanning electron microscopy (SEM), fluorescence microscope, and confocal microscope were used to study their morphologies change ( Figure 4 and Figure S22). SEM images of (1R, 2R)-TM and the mixture of (1R, 2R)-TM and (+)-Dibenzoyl-D-tartaric acid were first collected in THF/water mixture. As shown in Figure S22, both (1R, 2R)-TM and the complex of (1R, 2R)-TM and (+)-Dibenzoyl-D-tartaric acid formed amorphous nanoparticles in the THF/water mixture (f w = 65.8%). While, upon complexation with (−)-Dibenzoyl-L-tartaric acid, the aggregates gave micron, even millimeter grade floccule suspension ( Figure 4A,B). Inspired by the morphology difference, the concentration and standing time were further increased, a gel was facilely acquired with blue fluorescence ( Figure 2D). From SEM images we can see that abundant blocky structures of various sizes were formed ( Figure 4C,D). Based on above results, it is hypothesized that (−)-Dibenzoyl-L-tartaric acid can well match with (1R, 2R)-TM to form ordered assembles to further emit strong fluorescence due to the suitable spatial configuration and AIE effect.
Then, 1 H NMR titrations were carried out to disclose the chiral sensing mechanism in THF-d 8 . As illustrated in Figure 5, the resonance of protons of (1R,  (Table S2 and Figure S23). While, the proton H e , H h , H j , and H k of (1R, 2R)-TM showed an evident downfield shift of δ 0.131, 0.285, 0.665, 0.052, and 0.099, respectively. Chemical shifts of (1R, 2R)-TM and (+)-Dibenzoyl-D-tartaric acid had a similar tendency as the solution of (1R, 2R)-TM and (−)-Dibenzoyl-L-tartaric acid (Table S3 and Figure S24). According to the NMR titration results, it was inferred that the amino groups of (1R, 2R)-TM can easily interact with carboxyl groups of (−)-Dibenzoyl-L-tartaric acid to generate corresponding ionic compounds through acid-base interaction. The obtained −NH 2 + group can efficiently decrease the electron density of neighboring protons to cause an obvious downfield shift in NMR spectra. Inversely, an upfield shift of NMR was recorded because of the enhanced shielding effect of electron-rich acid ions (−COO − ). Besides, after interaction of (1R, 2R)-TM and (−)-Dibenzoyl-L-tartaric acid, the proton H h and H k exhibited much more upfield shift than that of (+)-Dibenzoyl-D-tartaric acid. It is implied that hydrogen bonds were formed between the benzoyl groups of (−)-Dibenzoyl-L-tartaric acid, and H h of (1R, 2R)-TM. 1 H NMR titration also suggested that (1R, 2R)-TM complexed with (−)-Dibenzoyl-L-tartaric acid in a 1:1 molar ratio as well ( Figure S23), this was consistent with the PL titration ( Figure S20). Moreover, 2D NOESY NMR spectrum of the mixture of (1R, 2R)-TM and (−)-Dibenzoyl-L-tartaric acid was collected in THF-d 8 (Figures S25 and S26). On account of the suitable spatial configuration and acid-base interaction between amino and carboxyl groups, intermolecular NOE signals of H a −H i , H a −H k between (1R, 2R)-TM and (−)-Dibenzoyl-L-tartaric acid were observed, which indicated that carboxylic groups are close to amino groups of (1R, 2R)-TM. While, for the mixture of (1R, 2R)-TM and (+)-Dibenzoyl-D-tartaric acid, intermolecular NOE signals cannot be detected, which implied a weak interaction between (1R, 2R)-TM and (+)-Dibenzoyl-D-tartaric acid ( Figures S27  and S28). To evaluate the effect of pH on chiral recognition, fluorescence spectra were collected with different equivalents of hydrochloric acid (0.5 to 1000 equiv.) under the same test conditions. As illustrated in Figure S29, (1R, 2R)-TM emitted weak fluorescence in THF/water mixture (f w = 65.8%). When adding the hydrochloric acid to the solution for 0.5 h or overnight, the fluorescence intensity showed a minor change in contrast to Figure S29A. Thus, we think the pH has a minor effect on the emission of (1R, 2R)-TM.
Therefore, we speculated the mechanism for enantioselective recognition is that amino groups of (1R, 2R)/(1S, 2S)-TM can afford efficiently charge-aided hydrogen bonding interaction with the carboxyl group of chiral acid. Due to the opposite configuration of enantiomers, (1R, 2R)-TM could closely combine with one enantiomer with suitable configuration to form compact complex, which will restrict the free motion of AIEgens to activate the radiative transition, thus resulting in bright emission. For the other acidic enantiomers, it possessed oppositely stereospatial structure and showed low spatial matching to (1R, 2R)-TM, and forming loose complex with weak or no emission. To better understand their interaction mode, a simulation of the interaction of (1R, 2R)-TM and (−)-Dibenzoyl-L-tartaric acid was conducted ( Figure S30).

Enantioselective separation
Enantioselective separation based on fluorescence technique has increased large interest owing to its high selectivity and sensitivity for chiral chemicals. Considering the conspicuous enantioselectivity for L-and D-dibenzoyl tartaric acid, (1R,  Figure S31B). The solution was mainly consisting of (+)-Dibenzoyl-D-tartaric acid with an enantiomer content of 85% ( Figure S31B). Although the separation efficiency cannot compare with the chiral HPLC, this promising strategy showed great potential in separation of chiral drugs and reagents.

CONCLUSION
In this work, two optically active AIEgens were facilely prepared, namely (1R, 2R)-TM and (1R, 2R)-TM, which can discriminate 5 kinds of chiral acids. Especially, the fluorescence intensity ratio of two complexes was up to 281 for (−)-Dibenzoyl-L-tartaric acid and (+)-Dibenzoyl-D-tartaric acid. Besides, these chiral acids are able to induce obvious CD signal change upon interaction with (1R, 2R)-TM. The results of 1 H NMR and 2D NOESY NMR suggested that the acid-base interaction of amino and carboxyl groups and suitable stereoselectivity were responsible for the enantioselective recognition. Then, the enantioselective resolution of D/L-dibenzoyl tartaric acid was realized by using the chiral AIEgens. The HPLC results revealed that the enantiomer separation efficiency was achieved to be 82% in the sediments (enantiomeric excess value was assessed to be 64% ee). Compared with the chiral HPLC separation, it is anticipated that this strategy will afford a simple and convenient method for enantioselective recognition and separation.

Materials
All reagents and solvents were chemically pure (CP) grade or analytical reagent (AR) grade and were used as received unless otherwise indicated.