Chiral and non-conjugated fluorescent salen ligands: AIE, anion probes, chiral recognition of unprotected amino acids, and cell imaging applications

Natural products are usually non-conjugated and chiral, but organic luminescent materials are commonly polycyclic aromatic molecules with extended p-conjugation. In the present work, we combine with the advantages of non-conjugation and chirality to prepare a series of novel and simple salen ligands (41 samples), which have a non-conjugated and chiral (S,S) and (R,R) cyclohexane or 1,2-diphenylethane bridge but display strong blue, green, and red aggregation-induced emission (AIE) with large Stokes shifts (up to 186 nm) and high fluorescence quantum yields (up to 0.35). Through hydrogen and halogen bonds, these flexible salen ligands can be used as universal anion probes and chiral receptors of unprotected amino acids (enantiomeric selectivity up to 0.11) with fluorescence quantum yields up to 0.29 and 0.27, respectively. Moreover, the effects of different chiral bridges on the molecule arrangement, AIE, and anion and chiral recognition properties are also explored, which provide unequivocal insights for the design of non-conjugated chiral and soft fluorescent materials.


Introduction
Chirality is a universal phenomenon throughout nature. 1 Not only many biologically active molecules, such as the naturally occurring amino acids and sugars, are chiral, but also many key chemical and biological processes are profoundly inuenced by chiral structures. Therefore, chiral recognition 2 through uorescence method have attracted increasing attention due to its unique advantages, including high sensitivity, good selectivity, short response time, real-time and parallel monitoring, cell imaging, and in situ and non-invasive measurement. Up to date, a lot of different uorescent chiral receptors have been designed and prepared for enantioselective recognition of chiral amines, acids, or modied amino acids. In the literature, however, there are only a few 1,1 0 -bi-2-naphthol 3 (BINOL)-and tetraphenylethylene 4 (TPE)-based uorescent receptors for chiral recognition of unprotected amino acids, because zwitterionic amino acids have not only a poor solubility in organic solvents but also little interaction with organic receptors in water. 5 Organic luminescent (uorescent or phosphorescent) materials 6 are commonly polycyclic aromatic molecules with extended p-electron conjugation, but these rigid materials might suffer some disadvantages, such as the synthesis difficulty, poor solubility, and uorescence aggregation-caused quenching 7 (ACQ). On the other hand, non-conjugated so materials that might have a better solubility, lower cost, higher exibility, lower cytotoxicity, better biocompatibility, and naturally occurring, 8 are usually taken for granted as nonemissive materials. Until recently, some non-conjugated materials, such as 1,1,2,2-tetraphenylethane, 9 poly[(maleic anhydride)-alt-(vinyl acetate)], 10 polyisobutene succinic anhydrides and imides, 11 and steroidal saponin digitonin, 12 were reported to show amazing ultraviolet (UV) or blue aggregation-induced emission (AIE). 9,13 Our research work continuously focus on the synthesis, optical properties, and sensing applications of N,N 0 -bis(salicylidene)ethylenediamine (salen), a particular class of salicylaldehyde-based bis-Schiff base (Fig. 1), owing to its facile preparation, good stability, and rich optical properties. 7b,14 Our recently work demonstrated that step-like salen ligands 14g (R-C n ) and tripod-like tri-Schiff bases 15 (TSBs, Fig. 1a) linking with non-conjugated long alkyl bridges display strong red-green-blue (RGB) AIE. If chiral (S,S) and (R,R) cyclohexane or 1,2-diphenylethane bridges are employed, it is easy to prepare chiral salen ligands (Fig. 1b). These non-conjugated chiral ligands are widely used as asymmetric catalysts, 16 and they are deemed to be non-emissive and used as turn-on uorescence probes for detecting metal ions. 17 Moreover, non-conjugated so materials might be used as anion probes, because they are exible and might have multiple and strong intermolecular interactions with anions through hydrogen bonds. 15 Particularly, the effects of different chiral bridges on the molecule arrangement, AIE, and chiral recognition properties are still unexplored. Herein, we demonstrate a novel and simple class of colorful chiral salen ligands (41 samples, Fig. 1b) that are linked by a non-conjugated cyclohexane (Cy) or 1,2-diphenylethane (diPh) bridge but have strong AIE and potential applications in cell imaging and recognition of anion and chiral unprotected amino acids.

Synthesis and characterization
All salen ligands were reasonably easy to synthesize by the condensation of primary diamine with 2 equivalents of salicylaldehyde precursor in ethanol under reuxing condition. 14 Most salen ligands have a bad solubility in petroleum ether, hexane, and water but a good solubility in CH 3 CN, CH 2 Cl 2 , DMF, and DMSO. All salen ligands either in solution or in solid state are stable within several months under air. For most salen ligands, good-quality single crystals can be obtained by the method of slow solvent diffusion/evaporation (CH 2 Cl 2 /hexane). For the purpose of comparison, the reference substance C 2 (ref. 14g) (Fig. 1a) was prepared as well.

Photophysical and AIE properties
The UV/visible absorption and uorescence data of all synthesized Cys and diPhs at room-temperature are listed Tables S1, S2, Fig. S1, and S2 (in ESI †). Since the photophysical and AIE properties of (S,S), (R,R), and racemic salen ligands are similar, only racemic salen ligands are discussed in this section. As we expected, all salen ligands exhibit weak uorescence in pure organic solvent of MeCN or DMSO (Tables S1-S3 †). For example, racemic 3,5-Cl-Cy shows very weak blue uorescence (V ¼ 0.010) in the dilute MeCN, because the dynamic intramolecular rotations (IRs) of central cyclohexane bridge in 3,5-Cl-Cy molecules provide a possible way to non-radiatively annihilate its excited states and result in the absence of uorescence consequently. However, if water is added into MeCN, the green uorescence of 3,5-Cl-Cy is strengthened (Fig. 2), because 3,5-Cl-Cy molecules cannot be dissolved in water, which causes 3,5-Cl-Cy molecules to precipitate and aggregate. At the same time, adding water would red shi its absorption spectrum ( Fig. S3 †), which further conrms this aggregation. Its green uorescence reaches the maximum (l em ¼ 511 nm, V ¼ 0.064) with a large Stokes shi of 117 nm, when the volume fraction (f) of water is increased to 90%. Furthermore, solid 3,5-Cl-Cy displays strong green-yellow uorescence (l em ¼ 532 nm, V ¼ 0.32) under 360 nm UV lamp ( Fig. 2 and 3). All these ndings reveal the fact of its AIE nature.
All uorescence quantum yields (V) (Tables S1 and S2 †) of the aggregated salen ligands in water and solid state were measured by the optical dilute method of Demas and Crosby 18 with a standard of quinine sulfate (V r ¼ 0.55, quinine in 0.05 mol dm À3 sulfuric acid) and an integrating sphere, respectively. Although the salen ligands with a non-conjugated linker have a small p-conjugated system, the uorescence quantum yields of some salen ligands are unexpected high (up to 0.35 and 0.22 for Cys and diPhs, respectively, Tables S1 and S2 †). In general, the introduction of -F (3-F-Cy, V ¼ 0.060; 3-F-diPh, V ¼ 0.12), -Cl (3-Cl-Cy, V ¼ 0.14; 3,5-Cl-Cy, V ¼ 0.32; 3-Cl-diPh, V ¼ 0.16; 3,5-Cl-diPh, V ¼ 0.20) substituents to Cy (V ¼ 0.018) or diPh (V ¼ 0.10) would improve its uorescence quantum yield, but the presence of other substituents would bring positive or negative effects on uorescence quantum yield. Combining with the previous results, 14g,15 we can draw a conclusion that chlorination or uorination is an much more efficient way to improve the uorescence quantum yields of non-conjugated materials than those of p-conjugated materials.

Mechanism of AIE
The molecular structures and arrangements play a key role in AIE. For most AIE-active materials, they are p-conjugated molecules and stack closely with not only a short interplanar distance (d, for plane molecules 14f ) or intermolecular aromatic H Ar/p hydrogen bonds (for non-plane molecules, such as silole 19 or TPE 20 ) but also weak intermolecular face-to-face p-p interactions. The former can ensure to eliminate the molecular rotation, the later would prevent the formation of excimer. For non-conjugated AIE-active R-C n (ref. 14g) and TSBs, 15     S5-S18. † Unlike step-like R-C n (ref. 14g) and tripod-like TSBs 15 molecules, these Cys and diPhs molecules are V-and propellertype, respectively.
As expected, two intramolecular N/H hydrogen bonds (1.6776 and 1.6348Å) between H (-OH) and N (C]N) atoms are found in Cy (Fig. 5a), which would eliminate free rotations of phenol rings. The dihedral angles between two panels of pconjugated phenol rings are about 56.6 , resulting in its V-type molecular conguration and no intramolecular face-to-face p-p interactions in one Cy molecule. In dilute organic solution, it is obvious that the rotatable C-N and C-C single bonds in the central cyclohexane bridge afford a possible way to non-radiatively annihilate its excited states and quench the uorescence consequently. In the solid state, however, Cy molecules exhibit a head (cyclohexane)-to-tail (phenol ring) orientation (Fig. 5b). No intermolecular face-to-face p-p interactions ( Fig. 5a (Fig. S5 †). These above intramolecular and intermolecular interactions would help to restrain the IRs and lead to the presence of AIE consequently. Owing to the existence of chiral carbon atoms, two (S,S) and (R,R) enantiomers (1 : 1) are found in the single crystals of racemic Cy ( Fig. 5a and b). The R/S-chirality-induced interactions between two enantiomers (Fig. S5 †) would help Cy molecules arrange in a tight head-to-tail orientation packing.
Similar two (S,S) and (R,R) enantiomers (1 : 1) are also found in the single crystals of racemic 3,5-Cl-Cy (Fig. 5c, S6, and S7 †), but 3,5-Cl-Cy molecules exhibit a different tail-to-tail orientation from Cy molecules (Fig. 5c). The R/S-chirality of two enantiomers induce their phenol rings to pack together with a d of 3.46Å ( Cl/Cl interactions (halogen bond 24 ) that is one possible reason why chlorination or uorination can improve uorescence quantum yield (see later discussion). All these data are consistent with the fact that solid 3,5-Cl-Cy and its enantiomers have the highest V values (0.32-0.35).
Our previous work demonstrated that the introduction of strong electron-accepting -NO 2 substituents would be an efficient way to red shi uorescence bands of non-conjugated AIEactive materials. 14g, 15 In this work, 3-NO 2 -Cy also shows a red-shied uorescence band up to 565 nm (Fig. 4). We failed to grow good-quality single crystals for X-ray analysis in the previous and present work. It is fortunate that the X-ray single crystal structure of 3-NO 2 -(R,R)Cy is obtained in the literature. 23 Unlike the above results, there are strong intermolecular faceto-face p-p interactions in 3-NO 2 -(R,R)Cy (d ¼ 3.31Å, Fig. S16 †), according with the fact that of 3-NO 2 -Cy has a red-shied uorescence band with a not-so-high V value (0.051).
As shown in Fig. 7, two phenol rings and two benzene rings in 3-F-diPh and 3-Cl-diPh molecules are cis form rather than trans form, due to the existence of two chiral (R,R) or (S,S) carbon atoms. Two phenol rings and two benzene rings are separated each other to form their propeller structures, just like another well-known AIE-active material of TPE. 9,13,20 Therefore, 3-F-diPh (2.4888-2.8510Å) and 3-Cl-diPh (2.3507-2.8279Å) molecules can tightly packed without any face-to-face p-p interactions in their single crystals ( Fig. S17 and S18 †). It should be noted that strong intermolecular F/F (3.2922Å) and F/O (3.1760Å) interactions are found in 3-F-diPh and 3-Cl-diPh molecules, respectively. All these factors enable solid 3-F-diPh and 3-Cl-diPh have a high V value of 0.12 and 0.16, respectively.
Having small p-conjugated systems, these pure organic salen ligands emit strong solid-state emission along with large Stokes shis (up to 186 nm). This would enable us to doubt that their emission originates from phosphorescence 25 rather than uorescence. Time-resolved emission decay spectra (Fig. S19 †) reveal that the emission decay lifetime of 3-F-Cy, 3-F-(S,S)Cy, 3-F-(R,R)Cy, 3-F-diPh, 3-F-(S,S)diPh, and 3-F-(R,R)diPh powders is 0.70, 0.66, 0.66, 1.32, 1.72, and 0.93 ns, respectively. These data are in the range of ns, indicating that their emission is not phosphorescence but uorescence. The large Stokes shis might be contributed to the intramolecular hydrogen bonds which would lead to keto and enol tautomers through an ultrafast excited state intramolecularproton transfer (ESIPT) process. 26 With the same R substituent, R-Cy and R-diPh have the similar uorescence properties, because the low-energy transition is mainly be assigned to the p / p* transition involving molecular orbitals essentially localized on the iminomethylphenol units with little contribution from the non-conjugated cyclohexane and 1,2-diphenylethane bridges (see the later discussion). Moreover, our previous work revealed that both the multi-Schiff base structure and -OH groups are of the greatest importance for AIE. 14g

Anion probe properties
As shown in our previous work, 15 TSBs (Fig. 1a) are tripod-like side-single-opening cages with three exible arms of phenol ring and can be used as anion hosts through hydrogen bond. In this work, the two phenol rings of these salen ligands are cis form with a small dihedral angles, and thus they acting as clamps might react with anion through hydrogen bonds and detect anion consequently.
In addition, 3-F-(S,S)diPh has an excellent ef value of 0.11 for enantioselective recognition of D-and L-Val (Table S8 and Fig. S54 †). 3-F-(R,R)diPh can discriminate D-and L-Arg (ef ¼ 4.89) (Table S9 †). 3-Cl-(S,S)diPh can efficiently discriminate Arg (ef ¼ 5.48) and Gln (ef ¼ 0.21) enantiomers, on the other hand, Arg (ef ¼ 4.43) and Trp (ef ¼ 0.18) enantiomers can be can be distinguished by 3-Cl-(R,R)diPh (Table S9 †). 3-F-Cy (Table S5 †), 3-Cl-Cy (Table S6 †), and 3,5-Cl-Cy (Table S7 †) generally exhibit worse ef values than 3-F-diPh (Table S8 †), 3-Cl-diPh (Table S9 †), and 3,5-Cl-diPh (Table S10 †). However, since it is too difficult to grow good-quality single crystals from the mixture of the dye and amino acid in DMSO solution, we failed to get a clear insight into the chiral discrimination ability (see the later discussion). We just guess that diPhs have a 1,2-diphenylethane bridge but not a circular bridge of cyclohexane, hence, compared with Cys, they would be more exible and consequently might have a better ability to form a cavity 4a for the recognition of a tiny chiral structure difference between D-and L-amino acid.

Mechanism of anion and amino acid probes
In order to investigate the possible interaction mechanism between the dye and anion/amino acid, 1 H nuclear magnetic resonance ( 1 H NMR) analysis was done. As shown in Fig. 11a, the 1 H NMR signal of H 1 atoms in 3,5-Cl-diPh appears a small broad peak at downeld (d ¼ 14.5 ppm) due to the existence of the intramolecular hydrogen bonds (-OH/N, Fig. 5). 15,27 If adding 100 equivalent of OH À (Fig. 11b), F À (Fig. 11c), or L-Arg (Fig. 11d) (in D 2 O), this downeld peak would disappear and reform a new small sharp peak at upeld (d ¼ 10.1-10.2 ppm), which might be assigned to the ArOH without intramolecular hydrogen bonds. 27,28 This upeld shi indicates that the strong intramolecular hydrogen bonds are destroyed by adding anion/ amino acid. At the same time, the reduction (about 1/2-2/3) of peak area in new peak reveals the formation of new intermolecular hydrogen bonds. These intermolecular hydrogen bonds would increase the molecular electron density through-bond effect and lead to upeld shis of other H atoms consequently. 29 Moreover, the 1 H NMR signal of H 3 at chiral carbon atoms also shows an obvious upeld shi from d ¼ 5.2 ppm into d ¼ 4.7, (5.0 and 4.2), and 3.9 ppm upon adding OH À , F À , and L-Arg, respectively, which might indicate that the neighboring nitrogen atoms have strong intermolecular interactions with anion and amino acid through hydrogen and halogen bonds.  Furthermore, the blank experiments ( Fig. S64 and S65 †) revealed that the 1 H NMR peak at 10.1-10.2 was not caused by an impurity.
The mechanisms of the aggregation-and anion/amino acidinduced uorescence enhancement are totally different. For example, the green AIE of 3-F-Cy is originated from the aggregated state (in water or solid) but its blue anion/amino acidinduced uorescence is originated from the monomer (1.0 Â 10 À5 mol dm À3 in DMSO). To gain insight into the nature of the excited states and transitions, gas-phase density functional theory (DFT) and time-dependent-DFT (TD-DFT) were also carried out for 3-F-Cy with the Gaussian 09 program package (B3LYP 6-31G(d,p)). The energy absorption bands of 3-F-Cy (l abs ¼ 316, 253, and 211 nm in MeCN) are reproduced well by the computation, which predict three absorption peaks at 318, 248, and 213 nm (Fig. 12). The lower energy absorption is mainly contributed to the highest occupied molecular orbital (HOMO) / lowest unoccupied molecular orbital (LUMO) (318 nm, oscillator strength, f osc ¼ 0.0814). The energy level and orbital isosurfaces diagrams of 3-F-Cy (Fig. 12) reveal that its HOMO and LUMO are mainly made up of the p-functions of iminomethylphenol units rather than the cyclohexane bridge. Therefore, the lower energy absorption can mainly be assigned to the p / p* transition involving molecular orbitals essentially localized on the iminomethylphenol units with little contribution from the non-conjugated cyclohexane bridge. As mentioned above, F À , OH À , or L-Arg would lead to changing -OH into O À , and thus 3-F-Cy in form of O À is used to evaluate the nature of the anion/amino acid-induced excited states and transitions by calculation. For optimized structure, 3-F-Cy (in form of O À , Fig. 12) has a much bigger dihedral angle (79.0 ) between two phenol rings than 3-F-Cy (in form of OH, 51.5 ). The lower energy absorption band of 3-F-Cy (in form of O À , l abs ¼ 365 and 350 nm for experiment and computation, respectively) are mainly composed of three transitions, including HOMO / LUMO (360 nm, f osc ¼ 0.150), HOMOÀ1 / LUMO+1 (344 nm, f osc ¼ 0.175), and HOMO / LUMO+1 (338 nm, f osc ¼ 0.127). The p-functions of HOMO, HOMOÀ1, LUMO and LUMO+1 of 3-F-Cy in form of OH and O À are similar, which reveals that anion/amino acid would not induce to change of the way of transition (p / p* transition). This is much different from our previous work 15 that anion would induce to change of the way of transition from n / p* into p / p* for TSBs. Moreover, anion/amino acid would red shi the lower energy absorption band with an increment of molar extinction co-efficient ( Fig. 12 and S66 †). This phenomenon can be reasonably explained by the fact that electron-rich ArO À anions have resonance structures of benzoquinone. 28 These resonance structures lead the p-function Frontier molecular orbitals to a uniform distribution in the molecule skeleton, and consequently 3-F-Cy (in form of O À ) has not only a smaller energy gap of transition but also an increment of molar extinction co-efficient and oscillator strength. This also might be contributed to its high uorescence quantum yield in dilute DMSO solution. It is strange that dilute 3-F-Cy (in form of O À ) DMSO solution having a red-shied absorption spectrum shows a blue-shied uorescence spectrum, but AIE of 3-F-Cy (in form of OH, in solid or water) is green-yellow. There is no anymore ESIPT for 3-F-Cy (in form of O À ), which might be contributed to its blue uorescence.

Living cell imaging
These non-conjugated AIE-active materials might have a lower cytotoxicity and better biocompatibility, and thus green-lightemitting 3,5-Cl-Cy and red-light-emitting 3-NO 2 -Cy were used in cell imaging applications. The HeLa cells were imaged by the dye using a standard cell-staining protocol. As can be seen from Fig. 13, incubated with 3,5-Cl-Cy or 3-NO 2 -Cy, the HeLa cells grow similar as in the control experiments in the absence of the dyes, indicating that both dyes have little toxicity on the living cells. HeLa cells show negligible background uorescence. However, intense intracellular green and red AIE is observed aer HeLa cells are incubated with 3,5-Cl-Cy or 3-NO 2 -Cy, respectively. Therefore, these AIE-active materials may have potentially applications in probing or monitoring biologically important processes in vitro as well as in vivo.

Conclusion
In the present work, we demonstrate the unique AIE, chirality, anion/amino acid probe, and cell imaging properties a series of non-conjugated and cyclohexane/1,2-diphenylethane-linked  salen ligands. These V-or propeller-type salen ligands with small p-conjugated systems emit strong blue, green, and red AIE through little p-p interactions and strong non-covalent intermolecular interactions, such as O/H, H/H, C/H, F/ H, F/F, Cl/H, Cl/O, and Cl/Cl. Moreover, combine with the advantages of non-conjugation and chirality, these so materials have multiple and strong interactions with anions and amino acids by hydrogen and halogen bonds, and thus they might be used as universal anion probes and chiral receptors of unprotected amino acids with a good enantiomeric selectivity. Therefore, we believe that these simple salen ligands provide a new paradigm in the design of chiral and non-conjugated uorescent materials for developing advanced organic optoelectronic devices, orescent bio-probes, and cell imaging, and so on. Further studies on circularly polarized luminescence 30 of these salen ligands are currently underway in our laboratory.

Conflicts of interest
There are no conicts to declare.