Excited-state intramolecular proton transfer luminogens regulated by competing dynamic covalent bonds and hydrogen bonds

Abstract

Excited-state intramolecular proton transfer (ESIPT) plays a critical role in many chemical processes, and it is essential to develop diverse regulating mechanisms of ESIPT molecules. Herein we report control of ESIPT luminogens with the combination of competing dynamic covalent bonds and non-covalent hydrogen bonds. The incorporation of amino-containing ESIPT fluorophores into 2-formylbenzenesulfonamide offers a versatile platform for dictating reactivity and fluorescence. The competition between intramolecular hydrogen bonding of sulfonamide NH and dynamic ring-fusion of NH with nearby aldehyde site in ring-chain tautomers enables diverse emission of cyclic hemiaminal, ketoimine upon ESIPT, and anionic intramolecular charge transfer (ICT) band in solution. Dynamic covalent reactions with amines give luminogen-dependent ketoimine or ICT emission, with hydrogen bonding from imine nitrogen affecting proton transfer of sulfonamide. In the solid state of one luminogen the engagement of adjacent aldehyde for intramolecular hydrogen bonding affords ICT emission while higher energy excitation induced ketoimine emission. Mechanical grinding and acid fuming treatment allows structural conversion and cyclic hemiaminal/ketoimine emission, further achieving switch of multistate emission colors. Finally, dynamic covalent reactions with amino acids provide a facile way for tuning distinct emission in the solution and solid state. The results should find broad utility in molecular switches, chemical sensing, and intelligent materials.

It is desirable, but challenging, to develop methods to control excited state intramolecular proton transfer. Here, the authors report the use of competing dynamic covalent bonds and non-covalent hydrogen bonds for control over such luminogens.

Introduction

As one important class of functional organic molecules and materials excited-state intramolecular proton transfer (ESIPT) luminogens are gaining extensive attention due to their wide applications in chemosensors, imaging, lighting and displays, and organic lasers^1–6. With a four-level photochemical process (E → E* → K* → K → E), hydrogen bond-mediated enol-to-keto tautomerization in the excited state leads to a large Stokes shift for the ketone emission and high sensitivity to surrounding environment^7–9. Through tuning intramolecular hydrogen bonding donors XH (X = O, S, NR) and acceptors (Y) and mechanisms of proton transfer, there is continuous interest in developing diverse skeletons of ESIPT fluorophores, including salicylic acid and N-salicylideneaniline, 2-(2’-hydroxyphenyl)benzimidazole/benzoxazole/benzothiazole, flavone and 3-thiolflavone derivatives, etc^10–15. The excited-state long-range proton transfer process-based emitters were also reported^16–18. The interplay between different excited-state proton transfer units was achieved by leveraging the relay^19,20 or the competition^21,22 between hydrogen bonding sites. Moreover, the incorporation of ESIPT emitters into metal-organic coordination polymers has given rise to emerging attributes, such as thermally activated fluorescence and long persistent luminescence²³. Force-regulated ESIPT was realized via the change of dihedral angles in the solid state^24,25 or within mechno-sensitive organic polymers^26,27. As a result, the development of distinct mechanisms and versatile strategies for regulating ESIPT molecules is of significance.

With non-covalent interactions, including hydrogen bonding, as the cornerstone of supramolecular chemistry, the cooperativity and competition between dynamic bonding forces hold a notable position in controlling the structures, reactivity, and properties^28–30. Through rendering the covalent bonds reversible, dynamic covalent chemistry (DCC) sets up a bridge between the realm of covalent synthesis and supramolecular chemistry, finding broad utility in the creation and regulation of functional molecules, assemblies, and materials^31–36. The integration of multiple types of dynamic bonding interactions into one system would further build up structural diversity and complexity, providing potential for emergent properties^37–41. DCC has contributed greatly to the research of sensing, switching, and labeling, as the adaptivity and responsiveness from the reversible formation, scission, and reorganization of covalent architectures afford versatile avenues for modulating signaling outputs^42–47. In particular, the bond connection/disconnection and thus structural transformation of dynamic covalent fluorophores in response to diverse external stimuli would lead to multistate switchable and multicolor luminescent materials^48,49. Nevertheless, the use of DCC for controlling ESIPT scaffolds has virtually escaped from the attention. With intramolecular hydrogen bonding already in place within ESIPT molecules (Fig. 1a), our focus is to introduce diverse competing dynamic bonding sites as a general way for dictating hydrogen bonds and fluorescence (Fig. 1b).

Fig. 1. Modulation of ESIPT luminogens.

The illustration of excited-state intramolecular proton transfer (ESIPT) luminogens (a) and the control with competing dynamic covalent ring-fusion site and hydrogen bonding site as well as dynamic covalent reactions with amines/amino acids in the current work (b). The hydrogen bonds are marked in red.

Considering the hydrogen bonding donor of XH (X = O, S, NR) in ESIPT molecules (Fig. 1a) being nucleophilic, its reversible capture by a nearby electrophilic unit would potentially affect ESIPT via competition for XH (Fig. 1b). By leveraging reversible covalent bond restrained fluorophores with adjacent ring-fusion sites^50,51, the attachment of amino-containing ESIPT fluorophores to 2-formylbenzenesulfonamide would offer a potent platform for the manipulation of reactivity and luminescence (Fig. 1b). With NH…Y intramolecular hydrogen bond turning on and off in open and ring tautomers, the emission of enamine (E) and ketoimine (K) forms would thus be impacted. The acidity of sulfonamide NH, along with the electronic character of the fluorophores, would play a role in determining emitting mechanisms and signaling modes. Dynamic covalent reactions (DCRs) with amines/amino acids to create imine adducts would afford a versatile handle for the modulation. Furthermore, the presence of nearby hydrogen bonding acceptors of aldehyde oxygen/imine nitrogen could influence the system via the competition for hydrogen bonding with sulfonamide NH, offering an additional layer of control. In essence, the interplay between diverse non-covalent hydrogen bonds and dynamic covalent bonds would be pursued for the regulation of proton transfer and emission of ESIPT scaffolds. In the current work, DCC-controlled ESIPT luminogens and associated structural and emission features in both solution and solid state were reported, affording multiresponsive and multistate luminescence and enriching the toolbox of this important class of functional molecules.

Results and discussion

Synthesis and structures

To demonstrate the versatility, two types of ESIPT molecules were chosen: 2-(2’-aminophenyl)benzothiazole⁵² and N-alkyl-7-aminophthalimide⁵³. The reversible bond constrained fluorophores (A1 and A2, Fig. 2) were synthesized from 2-formylbenzenesulfonyl chloride and corresponding amines in a one-step reaction (Supplementary Figs. 1–4). Notably, A2 was prepared via manual grinding in the solid phase due to the poor reactivity of N-(1’-propyl)−7-aminophthalimide.

Fig. 2. Solution and solid-state structures.

a, b Structure of A1/A2 and the ring-chain equilibrium in CDCl₃, CD₃CN, and DMSO-d₆. c Crystal structures of A1_O, A1_R, A2_O, I1, and I2, with the distance of non-covalent contacts, including hydrogen bonds, listed. d, e ¹H NMR spectra of A1/A2 (10 mM) in CDCl₃ and the reaction with 1-butylamine (3.0 equiv.) to create I1/I2.

The ring-chain tautomerization equilibrium of A1 and A2 is highly sensitive to solvent polarity. Taking A1 as an example, the open aldehyde (A1_O) and cyclic hemiaminal (A1_R) forms are dominant in CDCl₃ and DMSO-d₆, respectively, giving A1_O/A1_R ratios of 91:9 and 5:95 (Fig. 2a and Supplementary Fig. 20). The result in CD₃CN falls in the middle (67:33). A similar trend was observed for A2, with ratios of A2_O/A2_R to be 99:1, 58:42, and 9:91 in CDCl₃, CD₃CN, and DMSO-d₆, respectively (Fig. 2b and Supplementary Fig. 21). The sequence is consistent with solvent effects: potential intramolecular hydrogen bonding in open tautomers would be preferred in solvents of low polarity, while competitive solvents, such as DMSO-d₆, would have an opposite effect⁵⁴.

Crystal analysis shed light on structural insights for open/closed tautomers and associated hydrogen bonding. For A1_O the sulfonamide NH resides coplanarly with 2-phenylbenzothiazole unit, and the alignment of an intramolecular NH…N hydrogen bond (1.96 Å) would accommodate potential ESIPT (Fig. 2c). The lack of conjugation between the sulfonamide nitrogen and the arene plane was found in A1_R, with sulfonamide nitrogen engaging in a close contact with the sulfur atom of thiazole (N…S 2.81 Å). X-ray analysis of A2_O also revealed the presence of an intramolecular NH…O hydrogen bond (2.52 Å), though NH rotates out of the arene plane (dihedral angle 19°) (Fig. 2c). Moreover, an additional hydrogen bond between NH and aldehyde oxygen (NH…O 2.29 Å) was apparent in A2_O. It is notable the absence of such intramolecular NH…O (CHO) hydrogen bonding in the solid-state structure of A1_O.

Reactions of aldehydes with primary amines were next run. The reaction of A1 with 1-butylamine (3.0 equiv.) in CDCl₃ gave open imine I1 after 70 min, as evidenced by the methine CH around 9.05 ppm (Fig. 2d and Supplementary Figs. 22 and 23). The open imine was also attained for the reaction in CD₃CN (Supplementary Figs. 24 and 25). The movement of imine methine peak to 9.58 ppm in DMSO-d₆ indicates the formation of the anionic open conjugate base form (I1_OCB) with excess amine as a base in high polar solvent (Supplementary Figs. 26 and 27). The crystal structure of I1 from benzylamine supported the existence of intramolecular NH…N hydrogen bonding (2.06 Å), with NH coplanar with the attached arene and imine unit swinging away from the fluorophore (Fig. 2c). Open imine I2 was also created from the reactions of A2 and 1-butylamine (3.0 equiv.) in three solvents tested (CDCl₃, CD₃CN, and DMSO-d₆) (Fig. 2e and Supplementary Figs. 28–32). The exchange and reversibility of imine bonds were further confirmed (Supplementary Figs. 33–36). The crystal structure of benzylamine derived I2 showed imine nitrogen engaging in hydrogen bonding with nearby sulfonamide NH (NH…N 2.17 Å), echoing that of A2 (NH…O 2.29 Å) but significantly shorter (Fig. 2c). This difference can be attributed to the higher basicity of imine nitrogen over aldehyde oxygen. The varied hydrogen bonding patterns demonstrate structural tunability of DCC-restrained ESIPT molecules, setting the scene for controlling luminescence.

Fluorescence studies in solution

UV-vis absorption spectra of A1 were recorded in organic solvents with varied polarity, showing a strong absorption band around 338 nm (Supplementary Fig. 37). Relatively weak absorption was found between 370 and 450 nm in solvents with proton-accepting abilities, such as DMSO, DMF, and CH₃CN. UV-vis titration of A1 with the base 1,8-diazabicyclo[5.4.0]undec-7-ene (DBU) in CDCl₃ afforded a red shift from 330 to 360 nm, indicating base-triggered deprotonation of sulfonamide NH and formation of the anionic form (i.e., open conjugate base A1_OCB) of A1_O. A larger amount of DBU was needed to reach the plateau in CDCl₃ than CH₃CN and DMSO, and such a trend is consistent with the higher acidity of sulfonamide NH in more polar solvents (Supplementary Figs. 38-40). The UV-vis spectra of imine I1 showed similar trends towards solvents and base (Supplementary Figs. 41–43). With 1-butylamine as the base, the absorption of anionic I1_OCB was obtained in DMSO (Supplementary Fig. 43). Analogous results were obtained with A2/I2, albeit with higher acidity over corresponding A1/I1 (Supplementary Figs. 44–50).

The fluorescence of A1 were then investigated. Upon excitation of a solution of A1 (100 μM) in CHCl₃ at 350 nm, the yellow emission was found, with a maximum around 560 nm (Fig. 3a and Supplementary Fig. 51). A large Stokes shift of 12446 cm⁻¹ falls in line with the ESIPT mechanism (i.e., ketoimine emission) of open aldehyde (A1_O). In contrast, A1 in DMSO afforded blue-purple emission at 370 nm when excited at 330 nm (Supplementary Fig. 51). Without the conjugation between sulfonamide nitrogen and fluorophore in the cyclic hemiaminal (A1_R), a blue shift would be expected. Two emission bands around 370 and 560 nm (λ_ex = 335 nm) were detected in CH₃CN (Supplementary Fig. 51). These findings match the emission of open/closed isomers (A1_O and A1_R) in solution, with the extent of ring-chain equilibrium affecting the band intensity.

Fig. 3. Fluorescence and proposed mechanism in solution.

a, e Fluorescence spectra of A1 (100 μM)/A2 (50 μM) in CHCl₃ by using varied excitation wavelength. b, f Fluorescence spectra of A1 (100 μM)/A2 (50 μM) and the reaction with 1-butylamine (100 equiv.) in CHCl₃. The inset: fluorescence spectra of I1 (100 μM)/I2 (50 μM) in CHCl₃ upon titrating wtih DBU. c, g Explanation of the emission pattern of A1_O/A2_O in CHCl₃ and the regulation with imine formation and the deprotonation by base. d, h Proposed mechanism of the diverse emitting patterns of A1 and A2.

Encouraged by varied emitting species in solution, excitation-dependent luminescent performance was further explored. When dispersed in a series of solvents and irradiated with excitation wavelengths tuned across a suitable range, A1 yielded three emission bands centered at 370, 450, and 560 nm, respectively (Fig. 3a and Supplementary Figs. 52–69). The band at 370 nm is attributed to the cyclic hemiaminal (A1_R), and the band at 560 nm is assigned to the ketoimine form of the open aldehyde (A1_O), matching results earlier. We interpreted that the signal at 450 nm originates from intramolecular charge transfer (ICT) band of the anionic form (A1_OCB), which is in acid-base equilibrium with A1_O^55–58. After adding DBU into A1 in acetonitrile to generate A1_OCB, a single emission band at 450 nm was found, confirming this band from excited anionic state (Supplementary Fig. 70). Base titration of A1 also verified the change to the anionic state emission (Supplementary Figs. 71–73). These results collectively show the excitation-dependent emission behaviors of A1, stemming from the population of multiple coexisting structural states under applied excitation conditions, which contribute to distinctive emission features.

The fluorescence of I1 was then measured after the dilution of pre-formed imines. An emission band at 560 nm was found for I1 (100 μM) in CHCl₃ or CH₃CN (λ_ex = 350 nm), echoing ESIPT-based ketoimine band (Supplementary Figs. 74 and 75) and crystal structure, as imine group resides away from the ESIPT luminogen. However, for the open imine I1 in DMSO, a fluorescence signal appeared around 440 nm (λ_ex = 320 nm) (Supplementary Fig. 76). This is likely due to the weak ICT emission band from anionic conjugate base of open imine (I1_OCB), as revealed in UV-vis spectrum (Supplementary Fig. 43).

When the reaction of A1 (100 μM) with 1-butylamine was tracked in CHCl₃, the emission intensity of I1 at 560 nm was significantly increased in relative to A1 (λ_ex = 350 nm) (Fig. 3b and Supplementary Fig. 77). The fluorescence at 560 nm was switched down upon the titration of DBU into I1, with a minor ICT band around 440 nm (Fig. 3b and Supplementary Fig. 78). Analogous results of base-triggered fluorescence modulation of open imine I1 were found in acetonitrile, with I1 again more emissive than A1 at 560 nm (Supplementary Figs. 79 and 80). We attribute the enhancement of ketoimine band for I1 over A1 to less electron-withdrawing ability of imine than aldehyde (Fig. 3c). The measurement of quantum yield (Ф) further supported the explanation, with values of 1.17% and 3.94% for A1 and I1 in CHCl₃. Upon the deprotonation of sulfonamide NH by DBU to create the conjugate base ESIPT unit is sabotaged, turning down ketoimine fluorescence (Fig. 3c).

A2 (50 μM) exhibited excitation-dependent emission in CHCl₃, with two emission peaks at 450 nm and 600 nm (Fig. 3e and Supplementary Figs. 81–83). The excitation spectra obtained at the two emission wavelengths are different, indicating two distinct ground state substances. The emission at 600 nm is attributed to the ketoimine band of A2_O from ESIPT, while that at 450 nm corresponds to cyclic hemiaminal (A2_R). The quantum yields of cyclic hemiaminal and open ketoimine emission were 11.7% (λ_ex = 380 nm) and 1.94% (λ_ex = 340 nm), respectively. The fluorescence of the cyclic structure was supported by the emission of the control compound in CHCl₃ (Supplementary Fig. 84). Excitation wavelength-dependent fluorescence behaviors were also apparent for A2 in CH₃CN (Supplementary Figs. 85–87), though there was a red-shift for cyclic hemiaminal (A2_R) emission (around 475 nm). One rationalization comes from partial deprotonation of sulfonamide NH in more polar solvent (acetonitrile). Such a hypothesis was supported by the movement of cyclic hamiaminal fluorescence to 450 nm in the presence of methanesulfonic acid (MA, 2.0 equiv.) (Supplementary Figs. 88). Notably, a shoulder around 410 nm was apparent and assigned to the enamine band in acetonitrile, which shares the ground state with its ketoimine tautomer band upon ESIPT (600 nm) and was validated by excitation spectra. Furthermore, the emission around 520 nm (λ_ex = 410 nm) was found for A2 in DMSO (Supplementary Fig. 89). This signal was assigned to the ICT band arising from the anionic form A2_OCB in high-polarity DMSO. The addition of MA into A2 in DMSO afforded emission band around 445 nm (λ_ex = 375 nm), in consistence with the fluorescence of cyclic hemiaminal (A2_R) (Supplementary Fig. 90). Analogous phenomenon as the data in CHCl₃ and CD₃CN was observed for other solvents tested (Supplementary Figs. 91–100). The spectral overlap between the emission bands of A2_R and A2_OCB (450–520 nm) was found, with the ketoimine band around 600 nm. Moreover, the addition of DBU into A2 resulted in turning on the fluorescence around 520 nm for three solvents tested, further validated the appearance of ICT emission (Supplementary Figs. 101–103). The difference of A2 from A1 lies in the higher acidity of sulfonamide NH in the former as well as the varied electron-pulling ability of the fluorophores in relative to 2-formylbenzenesulfonyl group (Fig. 3c, g).

The dilution of imine samples gave fluorescence band around 520 nm for I2 (50 μM) (Supplementary Figs. 104–106). Upon titration of 1-butylamine into a solution of A2 (50 μM) in CHCl₃, the fluorescence at 520 nm increased significantly with excess amounts of amine serving as a base for I2 (Fig. 3f and Supplementary Fig. 107). Furthermore, the emission intensity of I2 was dramatically raised after adding DBU (Fig. 3f and Supplementary Fig. 108), also supported by a base-induced 20-fold increase of quantum yield from 1.97% to 39.4%. In addition, the turning up of fluorescence was detected for I2 with base present in CH₃CN (Supplementary Figs. 109 and 110). The strong emission of I2 as the anionic form was found in DMSO (Supplementary Fig. 111). We interpreted these results as following: instead of ketoimine band from ESIPT the ICT mechanism would account for the emission of I2, and the deprotonation of sulfonamide NH to create anionic conjugate base I1_OCB would further favor the ICT (Fig. 3g). With 2-formylbenzenesulfonyl being more electron-withdrawing than corresponding imine and serving as an electron sink pulling electron density away from the fluorophore, the emission of A2 with base present would be suppressed⁵⁹. Such an explanation of ICT emission aligns with crystal structure and was further backed with the emission of isolated imines in solution (Supplementary Figs. 112-117).

The effects of external media, including different pH values and solvent viscosities, on luminescent properties were further investigated. Upon raising pH of a solution of A1 in 4:6 DMSO/phosphate buffer, the amount of cyclic hemiaminal decreased, and its fluorescence intensity at 378 nm gradually declined (Supplementary Figs. 118 and 119). When the pH was above 6, the ICT mission band (486 nm) of anionic form A1_OCB emerged (Supplementary Figs. 120 and 121). Compound A2 also showed the appearance of an anionic emission band (554 nm) at pH 6.83, which was further enhanced until reaching pH 10 (Supplementary Figs. 122 and 123). The results fall in line with equilibrating emitting species in solution. We also studied the fluorescence responses of the compounds in ethyl acetate/liquid paraffin mixed solvents with different volume ratios (Supplementary Figs. 124–127). For both A1 and A2, the cyclic hemiaminal emission bands gradually intensified with increasing viscosity. These findings demonstrate the impact of solvent viscosity on the fluorescence.

To verify the occurrence of ESIPT, transient absorption spectra of A1 and A2 in CHCl₃ following photoexcitation with a 330 nm laser were measured. The spectra of A1 in CHCl₃ revealed two distinct bands: a positive excited-state absorption (ESA) band between 430 and 530 nm, and a negative stimulated emission (SE) band from 540 to 640 nm, centered around 560 nm (Supplementary Figs. 128 and 129). To further examine ESIPT process, the temporal evolution of the ESA signal at 451 nm was analyzed. Based on the fitting results, the respective lifetimes were 541 fs and 270 ps (Supplementary Fig. 130). The short-lived component is consistent with the typical time scale of ESIPT, whereas the longer-lived component corresponds to the decay of the ESIPT species to the ground state¹¹. For A2 its ESA band locates between 460 and 620 nm, and SE band resides from 630 to 730 nm (Supplementary Figs. 131-145). The lifetimes fitted at 484 nm were 56 ps and 470 ps. The difference of time scale of ESIPT between A1 (fs) and A2 (ps) can be attributed to stronger versus weaker hydrogen bonding^9,11,14,15. A stronger hydrogen bond leads to a faster proton transfer rate because the energy barrier to proton transfer is reduced. Tying it all together, the distinct mechanisms and diverse emitting patterns in response to solvent, amine, and base for multistate fluorescence were accomplished. The potential excitation-dependent mechanisms of A1 and A2 are illustrated in Fig. 3d, h.

Fluorescence studies in solid state

The attention was then turn to luminescence behaviors in the solid phase. A1 exhibited yellow emission (550 nm) in the pristine state with quantum yield (Ф) of 15.9% (λ_ex = 400 nm, Fig. 4a and Supplementary Figs. 146–148). When A1 was subjected to manual grinding, the color remained nearly intact (λ_ex = 380 nm, λ_em = 550 nm, Fig. 4a and Supplementary Figs. 149–151). We attributed the solid-state fluorescence of A1 to ketoimine tautomer band from ESIPT. Notably, compound A1 shows no excitation-dependent behavior, either before or after grinding.

Fig. 4. Fluorescence regulation in solid state.

a The solid-state fluorescence spectra of the pristine (λ_ex = 400 nm) and ground (λ_ex = 380 nm) samples of A1. b The solid-state fluorescence spectra of the pristine sample of A2. c The solid-state fluorescence spectra of the ground sample of A2. d The solid-state fluorescence spectra of the ground sample of A2 after exposure to HCl vapor. e The responses of the solid sample of A2 under varying conditions, with key structures shown. The photographs of samples were taken under a UV lamp.

The pristine solid state of A2 afforded excitation-wavelength-dependent emission (Fig. 4b and Supplementary Figs. 152–156). When the excitation wavelength was in the range of 390-420 nm, the emission peak at 520 nm dominated. Upon lowering the excitation wavelength (330-350 nm) a new peak at 600 nm emerged accompanied by the concomitant decrease of the peak at 520 nm. The 520 nm peak is attributed to the ICT band arising from an intramolecular hydrogen bond formed between NH and the aldehyde group (i.e., formyl oxygen as an internal base), while the 600 nm peak comes from the ketoimine emission. Due to the partial overlapping of two bands, a yellow color was apparent under a 365 nm UV lamp (Fig. 4e). The emission around 480 nm was obtained upon grinding of A2 with the mechanical force, with the band at 600 nm maintained (Fig. 4c and Supplementary Figs. 157–160). One rationalization comes from the fact that grinding leads to an increase in the surface to volume ratio of A2, thereby affecting ring-chain tautomerization equilibrium and resulting in the formation of cyclic hemiaminal structure (A2_R)^60,61. Nevertheless, aldehyde-mediated partial proton transfer still occurred. Mechanofluorochromic (MFC) materials are generating significant attention in sensing, lighting, and anti-counterfeiting, and the discovery of distinctive structures and mechanisms holds broad appeal^62–65.

Furthermore, the fuming of the ground sample of A2 with HCl vapor led to dual-mode cyclic hemiaminal/ketoimine emission at 450 and 600 nm (Fig. 4d and Supplementary Figs. 161 and 162). With acid facilitating protonation of sulfonamide nitrogen, the effect of residual ICT band would be removed. The blue and orange colors under 395 and 365 nm UV lamps confirmed spectral changes (Fig. 4e). Differently, the same HCl fuming treatment of a pristine sample of A2 afforded broad emission from 450 to 510 nm, with ketoimine emission remaining at 600 nm (Supplementary Figs. 163 and 164). Moreover, the exposure of A2 to HCl and Et₃N vapors enabled reversible interconversion between orange and green emission under a 365 nm UV lamp (Fig. 4e and Supplementary Figs. 165–174). The re-dissolution and evaporation of ground sample of A2 restored the dual emission at 520/600 nm (Fig. 4e and Supplementary Figs. 175 and 176). For MFC transition (pristine/ground), recovery (dissolution/evaporation), and acid/base fuming of A2, switching after multiple cycles showed fatigue-resistance with stable performance (Supplementary Figs. 177 and 178). As a result, the multistate emission with diverse colors (blue color 450-480 nm, green color at 520 nm, and orange color at 600 nm) can be reversibly switched with different stimuli and conditions.

Mechanistic insights from DFT studies

In order to elucidate the underlying mechanisms of varying fluorescence modes, DFT calculations were conducted. Firstly, conformation distribution regarding formyl-attached bond rotation offered a barometer for the orientation of the aldehyde group and, accordingly, its influence on the intramolecular hydrogen bonding of the sulfonamide NH (Fig. 5a and Supplementary Fig. 179). For A1 rotamers A1_O−2 and A1_O−3 with formyl oxygen pointing away from the fluorophore are thermodynamically preferred over A1_O−1 (Fig. 5b). The XRD structure of A1 closely resembles A1_O−2, with the distance of 1.97 Å for NH…N hydrogen bonding. Despite the energy gap, those rotamers can interconvert in solution and are also in equilibrium with the cyclic hemiaminal tautomer, contributing to complex mixtures and diverse emitting patterns. However, in solid state the stable rotamers (A1_O−2 and A1_O−3) dominate and lead to ketoimine band from ESIPT, as the nearby aldehyde unit doesn’t interfere with the emission.

Fig. 5. Calculated structures and hydrogen bonding features.

a The illustration of conformational rotamers of A1/A2. b Calculated structures of rotamers of A1, with free energy difference and hydrogen bonding distances listed. c Calculated structures of rotamers of A2, with free energy difference and hydrogen bonding distances listed.

The computation found the stabilization of rotamer A2_O−1a over optimized XRD structure A2_O−1b by 1.05 kcal/mol (Fig. 5c). The conformer (A2_O−2) with the aldehyde oxygen swinging away from NH fell in the middle (0.94 kcal/mol). The rotation about C-S bond gave another conformer A2_O−3 (0.51 kcal/mol), and the formyl group in A2_O−3 resides on the opposite side of NH. The comparison of hydrogen bonding distance offered valuable structural insights (Fig. 5c). While the significantly shorter NH…O (formyl) (2.07 Å) than NH…O (amide) hydrogen bonding (2.47 Å) was revealed in A2_O−1b, pretty close interacting distances were found for A2_O−1a (2.20 and 2.16 Å). The coplanarity of sulfonamide NH with phthalimide facilitates intramolecular NH…O hydrogen bonding (2.25 Å) in A2_O−2. The energetically favorable isomers of A2 are consistent with solution studies. The trend of hydrogen bonding distances of A2_O−1b, A2_O−1a, and A2_O−2 is in accordance with the fluorescence features of open aldehydes from pristine state, ground state, and the ground sample after further acid fuming. For imines I1 and I2 the stable conformers match their corresponding aldehydes A1 and A2 (Supplementary Fig. 180). However, according to natural bond orbital (NBO) analysis the strength of NH…N (imine) hydrogen bond in I2 is significantly enhanced over that of NH…O (formyl) from A2, favoring ICT emission for the former (Supplementary Figs. 181–183). Moreover, NBO data suggested that the intramolecular NH…N hydrogen bonding in A1 is stronger than NH…O hydrogen bonding in A2, consistent with the faster ESIPT process in A1. The ground and excited states for A2 were further calculated to gauge excitation and emission properties (Supplementary Figs. 184–187), and the trend of excited state energies of A2_R, A2_O-K, and A2_OCB is in consistence with the experimental emission bands (Supplementary Fig. 186).

DCC and Fluorescence Studies

Inspired by varied structures and fluorescence responses of A1 and A2, their reactions with amine nucleophiles were further conducted in the solution and solid state toward the creation of responsive luminescent materials. First, reactions of aldehydes with amino acids in buffer solution were studied. There was a sharp increase in the anionic state emission of A1 at 485 nm after adding homocysteine (Hcy), cysteine (Cys), or Histidine (His) at pH 7.4 and equilibration (Fig. 6a and Supplementary Figs. 188–192). Analogous results were obtained at pH 10 (Supplementary Figs. 193–196). Similarly, fluorescence intensity from the anionic form of A2 at 510 nm was significantly enhanced by Cys and Hcy at pH 7.4 (Fig. 6b), and by Cys, Asn, and His at pH 10 (Supplementary Figs. 197–205). The high affinity of aldehyde-containing A1/A2 toward amino acids with bifunctional reactive sites supports the creation of cyclic adducts, such as thiazolidine (C2) for Cys⁴⁶. It is interesting to note that under physiological pH these two probes afforded distinct selectivity profiles. A1 showed the highest response to Hcy, while the superior emission was found for A2 and Cys with low limit of detection (Supplementary Fig. 199).

Fig. 6. Fluorescence regulation with amino acids.

a Normalized fluorescence intensity of A1 (100 μM) toward various amino acids at pH 7.4 (λ_ex = 395 nm). Solvent: DMSO/buffer = 40:60, v/v. b Normalized fluorescence intensity of A2 (50 μM) toward various amino acids at pH 7.4 (λ_ex = 390 nm). Solvent: DMSO/buffer = 10:90, v/v. c ¹H NMR spectra of a ground solid-state sample of A2 and cysteine (1.5 equiv.) in DMSO-d₆. d ¹H NMR spectra of a ground solid-state sample of A2 and lysine (1.5 equiv.) in DMSO-d₆. e The solid-state fluorescence spectra of a sample of A2 and cysteine (1.5 equiv.) upon grinding. f The CIE 1931 chromaticity diagram of a ground sample of A2 and cysteine (1.5 equiv.) under different excitation wavelength. g Calculated structures of C2, with free energy difference and hydrogen bonding distance listed. h The solid-state fluorescence spectra of a sample of A2 and lysine (1.5 equiv.) upon grinding, as well as a sample of A1 and cysteine (1.5 equiv.) or lysine (1.5 equiv.) upon grinding.

After validating the feasibility of the sensing system, we proceeded to the detection of a series of amino acids via pattern recognition protocols⁶⁶. Twenty amino acids were employed as analytes, and differential responses were acquired. To maximize information extraction, we collected the emission data of probes A1 and A2 at 7 wavelengths each under pH 7.4 and pH 10 conditions (A1: 440, 455, 470, 485, 500, 515, and 530 nm; A2: 450, 470, 490, 510, 530, 550, and 570 nm), which were subsequently subjected to linear discriminant analysis (LDA). As a multivariate statistical analysis method, LDA allows for the classification of data and the assignment of unknown analytes to their appropriate classes. The resulting LDA score plot demonstrated the successful differentiation of all 20 amino acids (Supplementary Figs. 206-208 and Supplementary Tables 7-12). With four assay conditions the F1 axis accounted for 78.56% of the variance, the F2 axis contributed 14.21%, and the F3 axis yielded 7.09%, with these three variables collectively achieving a cumulative discrimination degree of 99.86% (Supplementary Fig. 208 and Supplementary Table S11). Jack-knife analysis showed 100% accuracy for the classification (Supplementary Table S12). These results show potential in the labeling and sensing of biological relevant molecules.

Mechanochemical synthesis offers a green methodology for dynamic chemistry in a solvent-free fashion^67,68. Upon grinding manually a mixture of A2 and cysteine (Cys, 1.5 equiv.) for 10 min quantitative formation of thiazolidine (C2) as two sets of diastereomers was obtained according to ¹H NMR (Fig. 6c and Supplementary Fig. 209) and mass analysis (Supplementary Fig. 210). The reaction of A2 and Hcy also gave cyclic thioaminals (Supplementary Figs. 211 and 212). Imine I2 was instead obtained when lysine (Lys) or arginine (Arg) was subjected to the similar treatment (Fig. 6d and Supplementary Figs. 213–216). Solid-state reactions of A1 with Cys/Hcy and Lys/Arg were also realized, affording cyclic thioaminal C1 and imine I1, respectively (Supplementary Figs. 217–224). In addition, the grinding a mixture of A1/A2 and 1-hexadecylamine (3.0 equiv.) for 10 min resulted in quantitative formation of imine I1/I2 (conjugate base form) (Supplementary Figs. 225–228).

The solid-state luminescence was further monitored upon reactions with amino acids. Interestingly, white emission was attained for Cys derived C2 under the UV lamp. Fluorescence spectrum with broad peaks around 490 and 590 nm was obtained upon excited at 365 nm, matching the co-existence of ICT and ketoimine emission bands (Fig. 6e and Supplementary Figs. 229–232). The excitation wavelength-dependent luminescence behavior was further visualized using the CIE 1931 chromaticity diagram (Fig. 6f). One plausible explanation comes from the steric hindrance imposed by thiazolidine ring, affecting proton transfer from sulfonamide NH to thiazolidine nitrogen (Fig. 6g). DFT calculation supported the argument, as in the favorable rotamer (C2-2) thiazolidine nitrogen locates away from sulfonamide NH, maintaining intramolecular NH…O hydrogen bonding and hence ESIPT (Fig. 6g). For rotamer (C2-1) a distance (2.15 Å) was found for NH…N (thiazolidine) hydrogen bonding, rendering ICT emission. Furthermore, the fluorescence spectrum of the ground sample of A2 with Cys (C2) upon fuming treatment with Et₃N vapor was centered at 508 nm (Supplementary Figs. 233 and 234), echoing base-induced ICT emission. The dual emission (490 and 590 nm) was also obtained for C2 from Hcy (Supplementary Figs. 235–238). The incorporation of basic amino acids Lys and Arg into I2 gave strong green fluorescence around 520 nm (Supplementary Figs. 239–244). With Lys and Arg the hydrogen bonding between sulfonamide NH and imine nitrogen would invoke the ICT emission. The yellow emission around 550nm (ketoimine tautomer band from ESIPT) was detected for Cys-derived C1, Hcy-derived C1, Lys-derived I1, as well as Arg-derived I1, matching results of A1 itself (Fig. 6h and Supplementary Figs. 245–256). The solid-state fluorescence of 1-hexadecylamine derived imines I1/I2 is consistent with the results from basic amino acids (Supplementary Figs. 257–260). Therefore, fluorescence in the solid state can be facilely tuned by dynamic covalent bonding of amino acids and by distinct hydrogen-bonding patterns in imine derivatives.

In summary, we developed a versatile platform of dynamic covalent bond constrained ESIPT luminogens for the regulation of emission. The competition with intramolecular hydrogen bonding by nearby dynamic covalent aldehyde sites in ring-chain tautomers allowed the realization of multistate fluorescence of cyclic form, ketoimine from after ESIPT, and anionic form in solution. Dynamic covalent reactions with amines to create imines further enabled ketoimine band from ESIPT (A1/I1) or ICT (A2/I2) based emission. While the ketoimine emission was detected in the solid state of A1, the adjacent aldehyde engaged in hydrogen bonding with sulfonamide NH and allowed ICT emission in the solid state of A2, and the higher energy excitation gave enamine emission of A2, leading to excitation-wavelength-dependent characteristics. The change to cyclic hemiaminal/ketoimine dual emission was achieved for A2 by using mechanical force and acid stimuli, with the switching between multistate fluorescence colors further realized. Finally, dynamic covalent reactions of amino acids enabled the binding of amino acids with discrimination ability in buffer and facile color tuning in solid state. The strategy of competing dynamic covalent and non-covalent bonding interactions makes a valuable addition to ESIPT luminogens, and the results shown should be beneficial to future design and applications of functional molecules, assemblies, and materials.

Methods

¹H NMR and ¹³C NMR spectra were recorded on a 400 MHz Bruker Biospin avance III spectrometer. Deuterated reagents for characterization and in situ reactions were purchased from Sigma-Aldrich Chemical Co. and Cambridge Isotope Laboratories, Inc. (purity ≥ 99.9%). The chemical shifts (δ) for ¹H NMR spectra, given in ppm, are referenced to the residual proton signal of the deuterated solvent. High-resolution mass spectra (HRMS) were acquired on the Thermo Scientific Exactive Plus Mass spectrometer equipped with an electrospray ionization (ESI) source. Crystallographic data of A1_O were collected on Hybrid Pixel Array detector equipped with Ga-Kα radiation (λ = 1.3405 Å) at 100.15 K. Crystallographic data of A1_R, A2_O, I1, and I2 were collected on Synergy-R-Cu diffractometer equipped with a graphite-monochromated Cu-Kα radiation source (λ = 1.54184 Å) at 293 K. The structures were solved with direct methods by using OlexSys or SHELXS-97 and refined with the full-matrix least-squares technique based on F2 by using the OlexSys or SHELXL-97. All other reagents were obtained from commercial sources and were used without further purification, unless indicated otherwise.

Dynamic covalent reactions in solution and solid state

Dynamic covalent reactions (DCRs) were performed in situ in deuterated solvents (CDCl₃/CD₃CN/DMSO-d₆) at room temperature without isolation and purification. To a stirred solution of an aldehyde (~10 mM, 1.0 equiv.) in deuterated solvents (0.60 mL), was added an amine (3.0 equiv.). The mixture was stirred and characterized by ¹H NMR and ESI mass spectral analysis. DCRs in solid state were performed by manual grinding in a quartz mortar. To a powder of an aldehyde (~10 mg, 1.0 equiv.) in a quartz mortar, was added an amine (1.5 equiv.). The solid mixture was ground and then characterized by ¹H NMR and ESI mass spectral analysis. See specific conditions in figure captions of the main text or supporting information if necessary.

Fluorescence experiments in solution and solid state

The spectrally pure dry solvents were used for the spectral measurements. Stock solutions of components were prepared. Components were mixed in a similar way as NMR studies, and the spectra were recorded when the equilibrium was reached. Alternatively, the solutions for the reactions of A1/A2 (10 mM) with 1-butylamine (3.0 equiv.) to create imines I1/I2 were prepared, and then the spectra were recorded by the dilution method. Fluorescence spectra in solution were recorded using a FLS980 fluorescence spectrometer (Edinburgh Instruments) or a microplate reader (BioTek SYNERGY H4) at a concentration of 100 μM for A1/I1 and 50 μM for A2/I2 in a solvent. Fluorescence titration experiments in solution were recorded on a microplate reader (BioTek SYNERGY H4). Fluorescence spectra in solid state were recorded on a FLS980 fluorescence spectrometer (Edinburgh Instruments). Fluorescence lifetimes were determined on an FS95 fluorescence spectrometer (Edinburgh Instruments) by using the time-correlated single photon counting method under the excitation of a laser (330 nm or 375 nm), where the fluorescence lifetimes removed the effect of the instrument response (IR) on the measurement results. The lifetime was fitted with a double-exponential decay to obtain τ₁ and τ₂, respectively⁶⁹. The ultrafast fs transient absorption measurements were conducted on Helios (Ultrafast systems) spectrometers using a regeneratively amplified femtosecond Ti:sapphire laser system (Spitfire Pro-F1KXP, Spectra-Physics). For fluorescence analysis, please see the specific conditions in the figure captions of the main text or supporting information if necessary.

Absolute quantum yields (ϕ) were determined on a FLS980 fluorescence spectrometer (Edinburgh Instruments) by using a calibrated integrating sphere setup equipped with a xenon high-pressure lamp⁷⁰. A degassed solution of each probe (in chloroform, deaerated by a nitrogen flow) was excited at its maximum wavelength. The quantum yield was calculated as Eq. (1).

$$\varphi =\frac{N_{emission}}{N_{absorption}}=\frac{\int L_{emission}}{\int E_{solvent}-\int E_{sample}}$$ 1

where N_emission is the number of photons emitted from a sample, N_absorption is the number of photons absorbed by a sample, L_emission is the emission intensity of sample, E_solvent and E_sample are the intensities of the excitation light not absorbed by the sample and the solvent, respectively.

DFT Calculations

Geometry optimization and frequency calculations of the ground state were performed by using the Gaussian 09 package (Revision: D.01), with the DFT method and basis set of M06-2X-D3/def2-TZVP. An ultrafine integration grid was also employed during the optimization and frequency analysis. Solvent model of polarizable continuum model (PCM) was included for acetonitrile. By frequency analysis, the number of imaginary frequencies for minima is 0. The conformational search was run with the initial geometry originating from the crystal structures. By the rotation of C-C and C-S bonds, conformers were generated. For simplicity, the N-methyl group was employed for imines. Additionally, TD-DFT calculations were also conducted for the major compounds of A2 at the same theoretical methods to optimize the excited-state structures. The B3LYP functional with 6-311++G(3df,2pd) basis set was used to calculate the vertical and emission excitation wavelengths and oscillator strengths (f) at the S₀ and S₁ geometries.

Author contributions

L.Y. conceived of the idea and directed the research. F.Z. designed the protocols, performed the experiments, and analyzed the data. H.Y. conducted density functional theory calculations. H.Z. and J.H. participated in organic synthesis. L.Y. wrote the manuscript. All authors discussed the results and revised the manuscript. F.Z. and H.Y. contributed equally to this work.

Peer review

Peer review information

Nature Communications thanks G Krishnamoorthy and the other, anonymous, reviewers for their contribution to the peer review of this work. A peer review file is available.

Funding

L.Y. discloses support for the research of this work from the National Natural Science Foundation of China [22471274, 92156010], the Strategic Priority Research Program of the Chinese Academy of Sciences [XDB1170000], and the Fujian Science & Technology Innovation Laboratory for Optoelectronic Information of China [2021ZR112]. H.Y. discloses support for the research of this work from the National Natural Science Foundation of China [22101284] and the Natural Science Foundation of Fujian Province [2022J05085]. H.Z. discloses support for the research of this work from the National Natural Science Foundation of China [22101283]. F.Z. and J.H. declare no relevant funding.

Data availability

The data supporting the findings of this study are included within the main text, the Supplementary Information, and the source data files. All data are available from the corresponding author (L.Y.) upon request. The crystallographic data for the structures discussed in this article also have been submitted to the Cambridge Crystallographic Data Centre. The deposition numbers are CCDC 2413677 [10.5517/ccdc.csd.cc2m0mkv] for A1_O, CCDC 2413678 [10.5517/ccdc.csd.cc2m0mlw] for A1_R, CCDC 2413679 [10.5517/ccdc.csd.cc2m0mmx] for A2_O, CCDC 2413680 [10.5517/ccdc.csd.cc2m0mny] for I1, and CCDC 2413681 [10.5517/ccdc.csd.cc2m0mpz] for I2. Copies of the data can be obtained free of charge via www.ccdc.cam.ac.uk/data_request/cif.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-73473-7.