The synthesis and characterisation of p-1urea, p-2urea, and m-2urea are described in the Supporting Information. In addition, single crystals of p-2urea were obtained by vapour diffusion of diethyl ether into a DMSO solution. Single-crystal X-ray diffraction analysis demonstrated that p-2urea forms a hydrogen-bonded solvate complex with DMSO. Oxygen atoms in DMSO interact with two NH groups in urea, forming bifurcated hydrogen bonds (Supplementary Fig. 5).
Combined density functional theory and time-dependent density functional theory (DFT/TD-DFT) calculations for the three compounds and the IHBC of the urea groups and anions were carried out. For the TD-DFT calculations, the geometries of the three urea derivatives without acetate anions were first optimised at the B3LYP/6-31G (d, p) level of theory. Then, using the optimised geometries, excitation energies were calculated at the CAM-B3LYP/6-31G+(d) level of theory33, 34. All calculations were performed by using the Gaussian 16 package35 and a common practice with DFT calculations is to replace tert-butyl groups with methyl groups to simplify the calculations and decrease the computational time. The geometries of the IHBC were optimised before excitation energy calculations were carried out at the CAM-B3LYP/6-31G+(d) level of theory. The results of these calculations are shown in Supplementary Figure 8. p-2Urea and the IHBC of p-2urea and acetate are expected to possess different degrees of conjugation and different electronic structures. The computed highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) of p-2urea without AcO– were found to be delocalised over the entire π-system. In contrast, the computed HOMO of the p-2urea–acetate IHBC was localised on the phenyl urea group substituent of the naphthyl system, and the LUMO of the complex was localised on the naphthyl rings. In addition, the HOMO energy level of the IHBC is increased by the ionic hydrogen bonding between the urea groups and acetate anions. A red shift of the UV absorption spectrum was predicted by TD-DFT calculations due to the charge-transfer character of the anionic p-2urea–acetate complex (Supplementary Fig. 9 and Supplementary Table 2). Therefore, control of the electronic states of fluorescent molecules by non-covalent bonding and the realisation of AIEE characteristics and viscosity-sensitive fluorescence by switching between PICT and TICT fluorescence are expected for the IHBC.
Investigation of hydrogen bonding interactions by 1H NMR and IR spectroscopies
In order to examine the ionic hydrogen bonding between the synthesised compound and acetate anions in solution and in the solid state, the evolution of the solution-phase 1H NMR spectrum upon the addition of tetrabutylammonium acetate (TBAAc) was observed, and an IR spectrum of the solid-state complex was acquired.
First, the shifting of the p-2urea N–H proton peaks in the 1H NMR spectrum was confirmed upon adding TBAAc to p-2urea in DMSO-d6. The N–H proton peaks were shifted from approximately 8.3 ppm to approximately 11.3 ppm and became saturated (Supplementary Fig. 10). From these observations, it was deduced that the electron densities of the N–H protons in the urea group were reduced by the ionic hydrogen bonding with acetate in DMSO. It was previously established that urea compounds form strong ionic hydrogen bonds with acetate, even in highly polar solvents such as DMSO that are thought to inhibit weak hydrogen bonding36.
In addition, FT-IR spectra were measured to confirm that the acetate and urea groups formed ionic hydrogen bonds, even in the solid state. As a result, the stretching vibration of the carbonyl of the urea group was shifted from 1651 cm–1 to 1699 cm–1 upon the addition of 2 equiv of TBAAc (Supplementary Fig. 11). This result demonstrates that the urea groups of p-2urea form intermolecular hydrogen bonds, and the C=O bond in the urea group is weakened in the solid-state without TBAAc. In addition, ionic hydrogen bonding between the N–H protons of the urea groups and acetate ions occurs with the addition of TBAAc, and the weaker intermolecular hydrogen bonds between different urea groups are eliminated with the C=O bond in the urea groups becoming stronger37. From these results, it can be concluded that p-2urea forms an ionic hydrogen bond with acetate ions, both in DMSO and in the solid state.
Investigation of changes in the UV–visible absorption and fluorescence spectra upon the addition of TBAAc in solution
Changes in the ultraviolet–visible (UV–Vis) absorption spectrum of p-2urea in DMSO (150 µM) upon addition of TBAAc were investigated. The UV–Vis spectrum of p-2urea in DMSO shows maxima at 270 and 340 nm. Upon addition of up to 250 equiv of TBAAc, a red shift in the absorption of p-2urea and isosbestic points at 280 and 310 nm were observed (Fig. 2 a). The red shift may be attributed to an increase in energy of the HOMO due to the ionic hydrogen bonding between p-2urea and acetate ions. Upon further addition of excess TBAAc, an increase in the absorbance of p-2urea over a wide wavelength range was observed, and the isosbestic points were lost. The absorption changes were presumed to be caused by changes in the properties of the solution caused by the addition of excess amounts of TBAAc.
In contrast, the UV–Vis spectrum of m-2urea in DMSO includes a single maxima, at 264 nm. Upon addition of TBAAc (up to 250 equiv), the red shift of the longer absorption region that was seen for p-2urea was not observed, and only one isosbestic point, at 270 nm, was apparent (Fig. 2 b).
The molecular-orbital calculation results support the hypothesis that the difference in the change in absorption wavelength between the positional isomers depends on the ease of charge transfer along the long-axis direction of the molecule, from the urea group to the naphthyl moiety. Comparing compounds that do not form hydrogen bond with acetate, the spatial distribution of the HOMO in m-2urea is mostly localised on the phenyl-group substituents of the naphthalene ring system, and the LUMO orbitals are distributed on the naphthyl rings; the molecular orbitals are not distributed over the urea group (Supplementary Fig. 12). However, in p-2urea, the HOMO and LUMO molecular orbitals are distributed around the naphthalene ring system, and it can be seen that the LUMO molecular orbitals are distributed in the urea group. Thus, p-2urea can be assumed to have greater overlap between its HOMO and LUMO orbitals than m-2urea, having greater oscillator strength on the long wavelength side of the UV–Vis spectrum (Supplementary Fig. 13 and Supplementary Table 4).
Compared with the molecular orbitals of the hydrogen-bonded acetate complexes, the same tendency is seen for m-2urea, and the molecular orbitals in the hydrogen-bonded complex of m-2urea with acetate are strongly localised to specific substituents (Supplementary Fig. 14). For the HOMO of the hydrogen-bonded complex of m-2urea, no electron density was observed on the naphthalene ring, which suggests that there is weaker absorption in the long wavelength region for this complex because there is little overlap between the HOMO and LUMO orbitals (Supplementary Fig. 15 and Supplementary Table 5). These calculation results are in agreement with the difference in the shape of the experimental absorption spectra for the two positional isomers on the long-wavelength side.
The emission spectrum of p-2urea in DMSO (150 µM) is characterised by an unstructured broad band centred at 436 nm (λex = 359 nm), while the emission maxima of p-1urea and m-2urea in DMSO (150 µM) occur at 399 and 423 nm, respectively. These results indicate that the extent of the π-conjugation in p-2urea is greater than those in p-1urea and m-2urea. The addition of TBAAc (up to 250 equiv) induced a substantial quenching of the p-2urea emission,30 while a new emission band developed at lower energies (Fig. 3). It was suggested that the lower-energy band originates from a charge-transfer transition involving a tautomeric form of the FL urea–acetate IHBC in which the N–H proton of the urea groups is transferred from urea to acetate in the excited state38. Upon the addition of a large excess of acetate (>250 equiv), a new emission band developed at 450 nm, the intensity of which did not reach a limiting value even at 20000 equiv. It is worth mentioning that the intensity value of the new emission band at 20000 equiv is higher than that of the p-2urea emission band before the addition of TBAAc. Changes in the emission spectra of p-1urea and m-2urea upon adding excess TBAAc were also investigated, to examine the generality and characteristics of the nonlinear fluorescence intensity change phenomenon. The nonlinear phenomenon was also observed for p-1urea and m-2urea, although the fluorescence intensity increase for p-1urea was minimal, and the value for m-2urea was even smaller compared to that of p-2urea (Figs. 3 and S17). These experimental results indicate that the nonlinear phenomenon upon adding of TBAAc is a general phenomenon for fluorescent urea compounds with a phenylnaphthalene skeleton, and π-conjugation from the hydrogen-bonded urea groups to the naphthyl groups is probably an important factor because of the significant difference between the results for p-2urea and m-2urea.
Thus, we concluded that the nonlinear phenomenon caused by the addition of acetate was a result of the ionic hydrogen bonds between acetate ions and urea groups, which increased the electron-donor character of the urea group and caused intramolecular charge-transfer fluorescence and TICT fluorescence. In general, fluorophores exhibiting TICT fluorescence often display fluorescence intensity increases with increasing solvent viscosity and aggregation-induced emission enhancement23. In addition, solutions with an excess of TBAAc are considered to exhibit high viscosity because the acetate anion and tetrabutylammonium cation of TBAAc experience a Coulomb force39. Therefore, we investigated the relationship between the increase in fluorescence intensity and the solvent viscosity change when adding an excess of TBAAc.
Viscosity dependence of fluorescence intensity
First, viscosity increases occurring upon adding TBAAc to DMSO were measured using a viscometer, revealing that the solvent viscosity was increased from 2.3 to 4.7 cP as the amount TBAAc was increased from 250 to 20000 equiv. The Förster–Hoffmann equation40, 41 describes the relationship between the fluorescence quantum yield and viscosity. To understand this relationship in our system, we plotted the ratio of the FL intensities at 452 nm of p-2urea before and after the addition of ≥250 equiv of TBAAc (I/I250eq) versus viscosity (Fig. 4). A linear relationship with a slope of 3.7 (R2 = 0.999) was obtained in the viscosity range from 2.3 to 4.7 cP at 25 °C, as expressed by the following equation (Eq. 1):
log(I/I250eq) = C + xlog(η /cP), (1)
where η is the viscosity and C is a proportionality constant that depends on concentration and temperature. The slope x is the most important parameter for evaluating the viscosity-sensing capabilities of probe materials because the value is probe-dependent and reflects the degree of change in I with respect to η. Values of x = 3.7 and C = –1.4 were determined for this molecular system. The torsional motions of the phenyl ring substituents of naphthalene should be inhibited as the viscosity of the media increases. Thus, nonradiative emission decay caused by this motion the phenyl rings should be effectively suppressed in more viscous media. It should be noted that x = 3.7 for the probe-dependent constant is a remarkably high value, compared, for example, with x = 0.59 for 9-(dicyanovinyl)julolidine (DCVJ) in a mixture of glycerol and ethylene glycol, a well-known highly viscosity-sensitive fluorescent probe23, 42.
Next, to examine the effects of ionic hydrogen bonding between urea groups and acetate anions on viscosity sensitivity, FL measurements of p-2urea in a mixture of glycerol and DMSO, without adding TBAAc, were performed as a control experiment. The relationship between the fluorescence intensity at 436 nm and viscosity was examined with a fixed concentration of p-2urea (1.5 μM) in mixed DMSO–glycerol solvents of various DMSO: glycerol ratios to modulate viscosity [DMSO/glycerol (v/v) = 10/0, 9/1, 8/2, 7/3, 6/4, 5/5, 4/6, 3/7, 2/8, and 1/9] (Supplementary Fig. 18). For the mixed solvents with glycerol amounts up to 80%, the emission intensity at an excitation wavelength of 359 nm increased with viscosity. Indeed, the emission intensity of p-2urea in 80% glycerol was significantly increased, being 1.7-fold greater than that without glycerol. However, in the solvent with a 90% glycerol content, the FL intensity decreased with increasing concentration because of the aggregation of p-2urea, which is insoluble in glycerol. The sensitivity constant of the dye, x in equation (1), was found to be 0.12 (R2 = 0.971). Therefore, p-2urea without ionic hydrogen bonding with acetate displayed low viscosity sensitivity. The viscosity sensitivities of IHBCs of m-2urea and p-1urea with acetate ions were also investigated in the presence of excess amounts of TBAAc, and lower viscosity sensitivities than that of p-2urea were obtained (Supplementary Figs. S19 and S20). These results indicate that the FL intensity of the p-2urea–acetate ion IHBC strongly depends on fluid viscosity.
Changes in fluorescence properties in solution and the solid state due to the addition of acetate ions
The absolute fluorescence quantum yields (QYs) in solution, in the absence and presence of TBAAc, were measured to confirm whether p-2urea shows AIEE in the presence of TBAAc. Measurement of FL QYs in the solid-state involves using a powder containing microcrystals obtained by slowly distilling off the solvent and an amorphous state after being ground in a mortar for 10 min, because FL QYs often change depending on the morphology of the FL material, i.e., whether it is in a microcrystalline or amorphous state.
The fluorescence QY of DMSO solution of p-2urea was 18%, while the QYs of powdered p-2urea, which contained microcrystalline and amorphous material, were 7% and 25% before and after grinding, respectively (Fig. 5). These unexpected differences between powders before and after grinding may be useful for mechano-stimuli-responsive luminescent materials. The FL QYs of the IHBC of p-2urea and acetate were also measured, in solution and in the solid state. The FL QY of the IHBC DMSO solution to which 200 equiv of TBAAc had been added, in which the ionic hydrogen bonding between urea groups and acetate ions is considered to be saturated, was revealed to be 7%. However, the solid-state p-2urea–acetate IHBC, prepared by drop casting the solution of p-2urea with 5 equiv of TBAAc, produced a QY of 41%, which is more than five times higher than that of the solution-phase IHBC. These results indicate that IHBC exhibits AIEE. The IHBC of p-2urea and acetate ions in the solid-state with 5 equiv of TBAAc exhibits an emission red-shift of approximately 18 nm relative to the emission of p-2urea powder in the absence of TBAAc (Fig. 6). In addition, the effects of the amount of TBAAc on the QY of the mixture of p-2urea and TBAAc were investigated by adding 2, 5, 10, 30, 60, 100, and 1800 equiv of TBAAc to the solid-state sample, resulting in FL QYs of 37, 41, 39, 31, 29, 29, and 29%, respectively (Supplementary Fig. 21). These results indicate that the cause of the increase in QY of the IHBC in the solid-state was neither the suppression of intermolecular interactions of the IHBC nor the suppression of self-absorption in the presence of an excess of TBAAc. Furthermore, the effects of the addition of TBAAc to p-1urea and m-2urea were also investigated. In the case of m-2urea, the QY of the DMSO solution was 17% (Supplementary Fig. 22). However, the powder containing microcrystals and the powder after grinding exhibited QYs of 37 and 34%, respectively. Interestingly, m-2urea shows AIEE even without adding TBAAc, which is an opposite result to that of p-2urea. In addition, the DMSO solution of IHBC exhibited 2% QY and the complex with 5 equiv of TBAAc in the solid-state exhibited a QY of 18%. The emission of the m-2urea–acetate ion IHBC in the solid state is red-shifted by 21 nm relative to the m-2urea powder emission in the absence of TBAAc (Supplementary Fig. 23).
In the case of p-1urea, the DMSO solution without TBAAc shows a high QY of 54%, while the QYs of the powder including microcrystals and the powder after grinding are 44 and 47%, respectively (Supplementary Fig. 24). In addition, the QY of the IHBC of p-1urea and acetate ions was 3% in solution and the QY of the IHBC increased to 50% in the solid-state in the presence of 5 equiv of TBAAc. The p-1urea–acetate ion IHBC in the solid-state exhibits an emission red-shift of approximately 9 nm relative to p-1urea powder in the absence of TBAAc (Supplementary Fig. 25).
These results indicate that IHBC of p-2urea, p-1urea, and m-2urea with acetate ions exhibit much higher QYs in the solid state than in solution. In particular, the IHBC of p-2urea and p-1urea with acetate ions in the solid-state exhibit higher QYs than those in the absence of TBAAc in the solid-state. In addition, the tendency of the IHBC QY to increase in the solid-state corresponds to enhanced fluid viscosity sensitivity for the fluorescence intensity.