Initially, we started from 3-substituted benzaldehyde (1a) and dioxazolone (2a) to explore the Rh-catalyzed enantioselective intramolecular C–H cyclization and sequential C–H amidation to construct a new donor-bridge-acceptor motif. Chiral amine (R)-(+)-1-(1-naphthyl)ethylamine (CA2) was employed as the CTDG and 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP) as the solvent. To our delight, 41% yield with 70% ee of desired product was obtained at the initial attempt. A mixed solvent of HFIP and nonpolar solvents (1,2-dichloroethane, toluene and chlorobenzene) improved the ee to 87%, while the yield decreased slightly (Table 1, entries 2–4). Since acidic conditions were crucial to the formation and dissociation of imine compound, we screened the amount of KH2PO4 to tune the acidity of the system. A 45% yield of desired product was obtained when we added 1.0 equiv. of KH2PO4. Increasing the amount of KH2PO4 to 1.5 equiv. improved the yield to 53%, whereas further increasing to 2.0 equiv. led to drop down of both yield and enantioselectivity (Table 2, entries 5–7). As carboxylic acids accelerated the C–H cleavage step in many cases, we tried to add carboxylic acid as an additive. Though pivalic acid increased the product to 60% yield, the selectivity was dramatically dropped to 70% (Table 1, entry 8). Interestingly, mesitylene carboxylic acid (MesCO2H) turned out to be effective, and the product was isolated in 59% yield with 87% ee (Table 1, entry 9). Then, a lot of carboxylic acids were tested, including protected amino acids (Table 1, entries 10–18). When phthalic-protected L-phenylalanine (A8) was employed in this reaction, the product was obtained in 89% yield with 91% ee (Table 1, entry 15). Further enlarging the size of carboxylic acid by using A9 did not improve the selectivity. Other class of protected amino acids (A10 and A11) also failed to provide better yield or ee (Table 1, entries 17 and 18). Notably, the sense of the chiral induction was completely determined by the chiral amine, as evidenced by almost the same yield and ee with CA2 and phthalic-protected D-phenylalanine (R-A8) as the CTDG and additive, respectively (Table 1, entry 19). When (R)-(+)-1-(1-phenyl)ethylamine (CA1) was employed instead of CA2, both the yield and ee decreased dramatically (Table 1, entry 20). Increasing the steric hinderance at the alkyl group of the chiral amine (CA3) provided the desired product in 58% yield with 91% ee (Table 1, entry 21). When chiral amine CA4 was employed as the CTDG, the bulky arene immediately shut down the reactivity and we did not get significant amount of desired product (Table 1, entry 22). Deceasing the amount of chiral amine to 30 mol % dropped the yield to 57% with 91% ee, while further decreasing to 20 mol % only obtained the desired product in 31% yield (Table 1, entries 23 and 24). Catalyst screening was also conducted, for instance, [Cp*IrCl2]2, Cp*Co(CO)I2, [Ru(p-cymene)Cl2]2 and Mn(CO)5Br, while no desired product was detected for any of the forementioned catalysts (Table 1, entries 25–28).Therefore, the optimized conditions of Rh-catalyzed twofold and unsymmetrical functionalization of enantioselective C–H activation via transient directing group strategy were as follows: [Cp*RhCl2]2 (5 mol %), AgSbF6 (20 mol %), CA2 (50 mol %), A8 (0.3 equiv.) and KH2PO4 (1.5 equiv.) in a mixed solvent of DCE/HFIP (1 : 1) at 60 °C for 36 h.
With optimal conditions in hand, we further investigated the generality and scope of our reaction. Firstly, product 3aa was obtained in 87% yield with 91% ee when we amplified the reaction scale to 0.15 mmol. Furthermore, the S configuration of 3aa was confirmed by X-ray crystallographic analysis. 3-Substituted benzaldehydes 1 containing different types of (E)-styrenyl units were explored, and all these substrates worked well to provide the desired products in 44–87% yields with 73–93% ee. 3-Substituted aldehydes bearing -Me, -F, -Cl and -CF3 groups at the para-position of (E)-styrenyl units reacted smoothly, and the corresponding products were isolated in 74, 72, 77 and 66% yields with 90, 91, 90 and 92% ee, respectively (Table 2, 3ba–3ea). Both electron-donating and electron-withdrawing groups substituted at the meta-position of (E)-styrenyl units were tested in this method, providing the desired products in 61–85% yields with 88–92% ee (Table 2, 3fa–3ja). When ortho-substituted substrates were employed to examine the compatibility of our method, the yields dropped to moderate (44–64%) and the ee also appeared to decrease slightly (3ka, 3ma and 3la). Di-substituted substrate was also explored under the optimized conditions, providing the desired product in 58% yield with 90% ee (Table 2, 3na). Alkyl-substituted substrates were also investigated under the modified conditions. Moderate yield and ee were obtained for 3oa when 1o was tested at 70 oC. Increasing the steric hindrance of C=C bond by using TMS instead of H improved the yield and ee to 40% and 70%, respectively (Table 2, 3pa). Besides that, 1,1-substituted alkene units were also compatible under the modified conditions. When the benzyl alcohols substituted alkenes (1q–1t) were employed as the substrates, 3qa–3ta were obtained in moderate yields with good enantioselectivities (40–64% yields and 86–90% ee, respectively). Interestingly, the thiophene substrate was feasible to provide the desired product (3ua) in 56% yield with 87% ee. More importantly, substrate 1v was also successful under the modified conditions, and the corresponding product 3va was produced smoothly with 84% ee. Furthermore, the modification of the benzene ring of the aldehyde was also conducted to evaluate the generality of our protocol. Product 3wa was isolated in 71% yield with excellent ee under the optimized conditions, while the efficiency and selectivity would decrease slightly when F was replaced by -Cl (3xa). Substrate bearing -OMe at the 5-position of the benzene ring was compatible with our method, producing the desired product in 43% yield with 82% ee (Table 2, 3ya). Substrate 1z was successful to give the corresponding product (3za) in 66% yield with 81% ee. Interestingly, fully substituted aromatic aldehyde 3zaa was also provided in moderate yield with 88% ee under our conditions.
We then turned our attention to the scope of dioxazolones. Dioxazolones (2) bearing not only electron-rich groups (-OMe and -Me) but also electron-deficient groups (-F, -Cl and -CO2Me) at the para-position reacted with 3-substituted aldehyde 1a smoothly to produce the desired products in moderate to good yields (43–84%) as well as commendable enantioselectivities (89–94%) (Table 2, 3ab–3af). Meta-substituted dioxazolones exhibited good reactivities and enantioselectivities to provide the corresponding products in 68–87% yields with 89–91% ee. More than that, Ortho-fluoro dioxazolone was tested in our method, giving the desired product 3ak in 88% yield with 90% ee. 1-Naphthyl dioxazolone 2l was also adaptable to produce the corresponding product in 84% yield with 90% ee. To further evaluating the generality of our method, alkyl dioxazolones were explored. Moderate yields and good enantioselectivities were achieved under the optimized conditions (Table 2, 3am–3ao).
For probing the mechanism of this reaction, we carried out some experiments. Firstly, we started from Int-1 and Int-1' to clarify the sequence of activation of two C–H bonds since either pathway was practicable. To our surprise, both the two intermediates reacted smoothly to provide the final product in moderate to good yields (Scheme 2, a and b). Shortening the reaction time to 2 h from 1a and 2a as the substrates only produced the Int-1 and 3aa with almost none of In-1', which might indicate the reaction started from intramolecular C–H cyclization followed by intermolecular C–H amidation (Scheme 2, c).
To shed light on the observed high enantioselectivity, a reasonable asymmetric induction model was explored. Firstly, a simplified intermediate was synthesized from chiral imine (4) and [Cp*RhCl2]2, and the key intermediate (Int-2) was isolated in 64% yield (Scheme 3). The X-ray structure of Int-2 showed the hydrogen of the chiral amine moiety faced the Cp ring and limited perpendicular distance (3.510 Å) from chiral carbon to the Cp plane, which might restrict the free rotation of C–N bond. As shown in the structure, the distance of C–N bond to the hydrogen of methyl group was measured as 2.072 Å, which implied that even a methyl group would restrict the rotational freedom of the C–N bond. Therefore, the methyl and naphthyl groups would be restricted at two sides of the substrate. The C=C bond would tend to coordinate with metal center from the methyl side instead of naphthyl side, which resulted in good selectivity. Furthermore, the absolute configuration of the desired product was consistent with the asymmetric model.
Based on the results and previously reported literature,62 we proposed a plausible mechanism (Scheme 4). Initially, active catalyst I is formed from [Cp*RhCl2]2 via Cl- extraction by silver salt. Meanwhile, aldehyde 1a reacts with CA2 to form the corresponding chiral imine and followed by C–H activation by I to produce intermediate II. Then, II experiences intramolecular ligand-exchange with alkene unit to provide III, which can undergo C=C insertion to form IV and followed by protonation, and then sequentially activates the second C–H bond to form V. Dioxazolone 2a is a good nitrene precursor, which can react with V to form VI by the extrusion of CO2. After that, VII is formed via Rh=N insertion. A protodemetalation of VII produces I and imine product, which undergoes hydrolysis to give the final product 3aa and simultaneously release the chiral amine CA2.
Several transformations were conducted to assess the value of our products (Scheme 5).63-65 Quinazoline 5 was isolated in 72% yield with NH3·H2O and 2-propanol as the solvent. Ethylcyanoacetate was also feasible to react with 3aa to produce compound 6 in 70% yield. Interestingly, the methylation and deacylation of 3aa provided mono-methylated amino benzaldehyde 7 in 93% yield under the conditions of NaH and MeI. Di-methylated amino benzaldehyde 8 was also obtained selectively when we added 6.0 equiv. of NaH and 10 equiv. of MeI, respectively. Compound 9 was gotten in moderate yield with CoCl2 and tert-butyl hydroperoxide (TBHP) at 80 oC for 8 h. The presentative transformations of our products exhibited good potential in synthetic chemistry.
D-A-D type of SBBFs exhibits great photophysical properties.30-31 The new molecules (3) contain two donor groups (NHCOR and OR’) and one accepter group (CHO). Therefore, this kind of motif could be a new D-A-D type of SBBFs. The photophysical properties of 3aa–3zaa and 3ab–3ao were examined in dichloromethane (DCM) (Supporting Information, Table S1). In general, most of these compounds showed maximal absorption peaks ranged 336–398 nm and the corresponding maximal emission peaks ranged 472–525 nm with tunable fluorescence quantum yields (0.010–0.258). Substituent groups located at either dihydrofuran skeleton or amide unit failed to influence absorption and emission wavelength. However, substrates at the 5-position of the benzene ring of the aldehyde would decrease the fluorescence quantum yields dramatically. In addition, the relatively large Stokes shift also brings potential applications in fluorescent dyes.66
The good photophysical properties inspired us to get insight into the mechanism. We synthesized a series of 2-formyl benzamide derivatives and investigated their photophysical properties. Compound 10a showed maximal absorption and emission peaks at 340 and 418 nm, respectively. Red shifts were detected when donating groups at the ortho- and meta-positions (10b: 355 and 432 nm; 10c: 369 and 456 nm). Particularly, the 2-formyl-3-methyl-4-methoxy benzamide (10d) showed similar absorption and emission wavelength with 3aa (Figure 1a, b and c). These results might indicate that the photophysical properties mainly came from the formyl and amide groups as a D-A unit as well as intramolecular hydrogen bond, and the substitutions at the ortho- and meta-positions were also helpful to tune the absorption and emission wavelength. Interestingly, 10a and 10b also showed a long wavelength shoulder peak at 544 and 556 nm, respectively, in their emission spectra (Figure 1c). Different solvents were then tested and the results suggested the shoulder peak turned to be weaker in polar solvents and stronger in nonpolar solvents, which might due to the excited-state intramolecular proton transfer (ESIPT) (Supporting Information, Figure S4).67-68 Both 10a and 10c exhibited moderate fluorescence quantum yields (0.323 and 0.287), while 10b and 10d showed very low fluorescence quantum yields (0.019 and 0.045). Moreover, 3ta, 3ua and 3va also showed low fluorescence quantum yields (0.010, 0.026 and 0.041). These might indicate that the steric hinderance at the ortho-position would lead to the quenching of the fluorescence. One possible reason of 3aa with moderate fluorescence quantum yield might be the rigidly five-membered ring decreased the influence of steric hinderance by preventing the rotation of alkyl group. To verify our hypothesis, a control compound 10e was synthesized and studied. As we expected, both absorption and emission were similar to 10d, while the fluorescence quantum yield was dramatically improved to 0.232. The radiative transition rate constants (kr) of 10c, 10e and 3aa showed slightly larger values to the weak fluorescence of 10d, while the nonradiative transition rate constant (knr) of 10d was much higher than others. The fluorescence lifetime data also showed 10d was dramatically shorter than 10c, 10e and 3aa (Table 3). These results might indicate that the five-membered ring and six-membered NH⋯O hydrogen bonding provide the rigid structure for the new D-A-D type SBBFs and inhibit the internal rotations of substituted alkyl group as well as the amino group, preventing the quenching of the fluorescence.69 Furthermore, blue-shifts of both absorption and emission maximum were observed when NHMe group was replaced by NMe2 group (Supporting Information, Figure S5, 7 vs 8), which might imply that the intramolecular hydrogen bond played a key role in achieving red-shifts of absorption and emission.27
The emissions of 3aa in different solvents were tested as representative example. Moderate fluorescence quantum yields were obtained in most of solvents, while it dropped dramatically when DMSO was used as the solvent (Supporting Information, Table S2). Therefore, the emission of 3aa was monitored in DMSO/water by changing the water fraction (fw), and slightly fluorescence-quenching was appeared in the range of low water fractions (0–80%). Interstingly, further increasing the ratio of water enhanced the fluorescence intensity up to 4.4 times, and red-shifts were observed at 541 nm. These results indicated that 3aa might be a yellow-emissive AIE-gen (Figure 1d). Furthermore, the photophysical properties of 3aa were also tested in the crystal state. Red-shift and good fluorescence quantum yield (42.5%) were observed in the crystal state (Table 3).
After that, the chiroptical properties of 3aa, 3ea, 3ac, and their enantiomers were also studied via circular dichroism (CD) and circular polarized luminescence (CPL) spectroscopies. The circular dichroism (CD) spectra were mirror images of each pair of enantiomers, and all these showed obvious peaks at about 390 nm (Figure 1e). Interestingly, the CPL experiments showed that S-3aa/R-3aa were CPL-active and displayed clear mirror images of its enantiomer in the crystal state (Figure 1f). The luminescence dissymmetry factors (glum) were measured as −4.8×10−3 for S-3aa and +6.1×10−3 for R-3aa at 583 nm (Figure 1g). With the good photophysical properties, we are confident that such small size and easily accessible SBBFs will bring great potential in many applications.
Further investigation of computational calculation was also conducted to study the electronic properties of 10e. As shown in Figure 2a, the geometries of 10e were optimized in both the S0 and S1 states by B3LYP functionals with 6-31G(d,p) level. The calculated C1-O2 bond for the ground-state was 1.37 Å, while the Franck–Condon excited-state was 1.34 Å, showed a good conjugation in excited state. Furthermore, the dipole moments calculated for 10e was 4.9812 D and 6.7704 D, respectively, which may give strong solvatochromic shifts of the absorption and emission bands.26 Notably, the carbon atom adjacent to the oxygen atom in the dihydrofuran ring was 0.205 Å above the benzene plane in the S0 state of 10e. The distance of 3aa was measured as 0.338 Å from its X-ray structure, which was consistent with our computational results. While in the S1 state, a planar geometry was found for the dihydrofuran ring with a
distance of only 0.004 Å. The conformation changes in the S0 and S1 states may be the origin of the large Stokes shift and broad emission bands of chiral SBBF.26 Moreover, the optimized S0 and S1 states exhibited the electron transfer from the chiral ring to the benzaldehyde moieties. The conformation changes along with the electron transfer, indicated a possible intramolecular charge transfer (ICT) process in this new SBBFs.70 Notably, the TDDFT calculated absorption was in good agreement with the experimental data and the data also showed that the strong absorption band observed in the visible range should be specified for S0→S1 energy states, which were mainly contributed by the HOMO to LUMO transition. The increased conjugation and planar conformation in S1 may inhibit the intramolecular rotation, providing its good fluorescence quantum yields.
To evaluate the imaging potential of these SBBFs in living cells, 3al and 3ao were employed as the representative probes to explore the cellular investigation. Both 3al and 3ao did not show any cytotoxicity after incubation for 24 h at the concentration of 10 μM as demonstrated by the standard CCK-8 assay, indicating the good biocompatibility of 3al and 3ao (Figure 3g and h). Then, co-localization experiments were exploited with lipid droplet (LD) fluorescent dye Nile Red as the reference. As we expected, both 3al and 3ao emitted green fluorescence in HeLa cells. Similar patterns with red fluorescence of the Nile Red and good overlaps were observed from the merging images of 3al and 3ao with the Nile Red (Figure 3a and b). Furthermore, their Pearson’s correlations of 3al and 3ao were calculated as high as 0.87 and 0.90, respectively (Figure 3e and f). These indicated that 3al and 3ao could selectively target the LDs in the living cells. In particular, the probe 3ao exhibited even better selectivity than the Nile Red since the Nile Red also emitted red fluorescence in cytoplasm. Co-staining with a commercial cell membrane dye DiR did not show significant overlaps from the merging images, indicating that 3al and 3ao were unlikely target the intracellular membranes. The merging images of 3al and 3ao with DIR showed distinct regions of the LDs and the intracellular membranes with different fluorescent dyes, which demonstrated potential applications of 3al and 3ao as multicolor cell-imaging probes (Figure 3c and d).