Introduction

Organic luminescence materials have played an indispensable role in a wide variety of applications, including organic light emitting diodes (OLED)1,2,3, chemical/biological sensors4,5,6,7, bioimaging probes8,9,10,11, and anti-counterfeiting technique12,13. After the maturity of mechanisms for thermally activated delayed fluorescence (TADF) and the improvement of design strategy for TADF systems, a high-efficiency electroluminescent material was realized14,15,16,17. Traditionally, the TADF is originated from the reversed intersystem crossing (RISC) from the lowest triplet state (T1) to the lowest singlet state (S1). However, there exists a TADF utilizing the exciton generated from a higher triplet state (Tn) via high-level reversed intersystem crossing (hRISC). The hot exciton material can effectively utilize the triplet nonradiative exciton and realize near 100% internal quantum efficiency (IQE) in theory18,19,20,21,22,23,24. Hence, this type of material has attracted increasing interest recently. To implement hot exciton TADF, one of the key factors, the rate constant of ISC (kISC) between the singlet state and the triplet state, needs to be mentioned. Based on the equation to calculate kISC8,25:

$${K}_{{{\rm{ISC}}}}({{{\rm{S}}}}_{{{\rm{n}}}}-{{{\rm{T}}}}_{{{\rm{n}}}})=\frac{2\pi }{\hslash }{{{\rm{SOCME}}}}^{2}\frac{1}{\sqrt{4\pi {k}_{{{\rm{B}}}}T}}\exp \left(-\frac{\Delta {{E}_{{{\rm{ST}}}}}^{2}}{4\lambda {k}_{{{\rm{B}}}}T}\right)$$
(1)

From the Eq. (1), we can figure out that two parameters will have a significant impact on the kISC, the energy level between the singlet state and triplet states (ΔEST) and the spin-orbital coupling matrix element (SOCME). The most common strategy to gain a large kISC is constructing the electron donor-acceptor system to create a charge transfer (CT) state which will significantly lower the ΔEST value, especially the through space charge transfer (TSCT)26,27,28,29,30,31. Compared to the through bond charge transfer (TBCT), TSCT can realize a larger extent of charge separation and thus significantly narrow ΔEST31. Moreover, a large enough gap between Tn and T1 is another key factor to ensure effective hRISC, because it is highly related to the rate constant of internal conversion (IC) which is a competitive process for hRSIC32.

Apart from the efficiency enhancement, the regulation of the emission color of the organic luminescence material is also an essential task33,34,35. According to some previous reports, controlling the aggregation state of the solid is an effective method to realize the regulation, taking advantage of different intermolecular interactions or molecule conformation in polymorphs36,37. The formation of different polymorphs is determined by various factors, including solvent type, aggregation rate, temperature, humidity, pressure, mechanical force, and so on37. Thereinto, utilizing the adjustment of temperature, pressure, and mechanical force, the organic luminescence molecules have been fabricated into thermochromic material38,39, piezochromic material40,41,42, and mechanochromic material43,44,45,46.

However, even though many polymorph TADF materials have been reported, the detailed mechanism and ultrafast processes involved for the emission in the solid state were rarely studied, especially excited-state photodynamic variation in different aggregation modes. In this study, fluorene which exhibits a hRISC property was selected as the electron donor23,47, and anthraquinone as the electron acceptor because it presents the potential of forming different polymorphs when bonding with another moiety37. The 2-(9,9-dimethyl-9H-fluoren-2-yl)anthracene-9,10-dione (F-AQ) was designed, synthesized, and fabricated into different solid phases (Fig. 1a), amorphism-G (amo-G), amorphism-Y (amo-Y), crystal-YG (cry-YG), and crystal-O (cry-O), with the 1 wt% doped PMMA film and glass state for reference. The steady-state UV-vis absorption, photoluminescence (PL) spectroscopy and transient PL decay spectroscopy was used to study the ground-state and excited-state properties of these samples, revealing the cry-O will have the absoprtion peak with the longest wavelength and lifetime of hot exciton TADF. Moreover, the femtosecond/nanosecond transient absorption (fs/ns-TA) spectroscopy48,49,50 were conducted to investigate the photodynamics and the transient species involved in the luminescence of TADF from hRISC, and a clear TSCT signal is observed for cry-O. This intriguing observation and characterization for solid-state photodynamics provides insight into the design of luminescence materials with the consideration of polymorphism-dependent behavior.

Fig. 1: Photophysical properties of F-AQ and its different solid states.
Fig. 1: Photophysical properties of F-AQ and its different solid states.
Full size image

a Molecular structure and four different morphs of F-AQ. b Steady-state UV-vis absorption spectra of F-AQ solution of ACN, and steady-state PL spectra of F-AQ measured in the solvents of cyclohexane (λmax = 450 nm), toluene (λmax = 501 nm), dichloromethane (λmax = 565 nm), tetrahydrofuran (λmax = 533 nm), acetonitrile (λmax = 599 nm), and dimethylsulfoxide (λmax = 607 nm). c Steady-state PL spectra of F-AQ@PMMA film (λmax = 549 nm), glass state (λmax = 563 nm), amo-G (λmax = 532 nm), amo-Y (λmax = 538 nm), cry-YG (λmax = 543 nm), and cry-O (λmax = 575 nm). d PL decay curve of the emission maxima for F-AQ@PMMA film, glass state, amo-G, amo-Y, cry-YG, and cry-O. e The temperature-dependent PL decay curves of F-AQ@PMMA film from 80 K to 230 K. f CIE color space diagram of F-AQ@PMMA film, glass state, amo-G, amo-Y, cry-YG, and cry-O. g Comparison of fluorescence peak and phosphorescence peak of F-AQ@PMMA film. h Transient emission spectra of F-AQ@PMMA film from 1 ns to 2 s with a gate width of 200 ms.

Results and discussion

The investigation of F-AQ was initiated from its spectroscopic analysis in the solution phase. From the steady-state UV-vis absorption spectrum, there appears three peaks located at 390 nm, 315 nm, and 250 nm. The emission peaks of F-AQ show evident solvent effects from 450 nm to 600 nm with the increasing of the solvent polarity, indicating the existence of a CT species (Fig. 1b). The solids of F-AQ with a different aggregation state associated with F-AQ doped PMMA film were also studied by the same spectroscopy methods. As a result, the F-AQ@PMMA film shows similar features to those of the solution phase, while the aggregation solid of F-AQ shows the growth of an absorption band in the UVA and visible region in varying degrees due to intermolecular interactions (Fig. S1). The emission spectra of each solid also presents obvious distinguishing spectra with variations in color of the different morphisms. As shown in Fig. 1c and Fig. 1d, the emission maxima were located at 549 nm, 563 nm, 532 nm, 538 nm, 543 nm, and 575 nm with the CIE coordinates of (0.39, 0.56), (0.44, 0.53), (0.27, 0.59), (0.39, 0.57), (0.33, 0.58), and (0.51, 0.49) for the F-AQ@PMMA film, glass state, amo-G, amo-Y, cry-YG, and cry-O, respectively. The PL quantum yields for these samples were also measured, 0.13, 0.04, 0.06, 0.46, 0.05, and 0.38, respectively, which suggests the aggregation mode of F-AQ will vary the luminescence efficiency in different directions. The packing mode of glass state, amo-G or cry-YG will cause aggregation-induced quenching (ACQ), while the stacking of amo-Y or cry-O can enhance the intensity of emission.

Moreover, the kinetics of the emission from F-AQ and its different solid phases were investigated by time-correlated single photon counting (TCSPC) technique excited by a 375 nm picosecond diode laser. The kinetics of F-AQ in exhibits a single-exponential decay, with the time constants of 1.53 ns, 3.62 ns, and 3.62 ns in cyclohexane, DCM, and ACN, respectively.(Fig. S3). For the situation in toluene, two time constants τ1 = 0.3 ns and τ2 = 1.2 ns was found, which may originate from the LE state emission and CT state emission. From the decay curve of each solid in Fig. 1e, the amo-G, and cry-YG exhibit the shortest lifetimes, and the amo-Y, glass state, PMMA film, and the cry-O will decay slower. Based on the double-exponential fitting results, we found that the PL decay curves of all solids displayed a two-step decay. The first lifetime can be assigned to prompt fluorescence. The temperature-dependent decay curve exhibited the prolonged lifetime and the enlarged component of the second species for each solid with the increasing temperature, (Fig. 1f and Fig. S5). Moreover, considering the delayed component of different solid phases shows much shorter lifetime values compared to the traditional one (ns scale vs. ms scale), the longest-lived feature can be attribute to the TADF generated from the hot triplet exciton via hRISC18,19,20,21,22,23,24. Hence, the lifetimes of the prompt fluorescence (PF) and hot exciton delayed fluorescence can be concluded, the τPF = 0.98 ns, 2.32 ns, 0.62 ns, 0.69 ns, 0.68 ns, and 2.14 ns, τDF = 5.20 ns, 7.92 ns, 4.00 ns, 9.23 ns, 5.46 ns, and 9.37 ns for F-AQ@PMMA film, glass state, amo-G, amo-Y, cry-YG, and cry-O, respectively (Fig. S4 and Table 1).

Table 1 Photophysical properties of F-AQ in solution, F-AQ@PMMA film, glass state, amo-G, amo-Y, cry-YG, and cry-O
Full size table

To further verify that assumption of hot exciton delayed fluorescence, theoretical calculation were carried out to study the excited-state energy levels and SOCME for F-AQ in single molecule and the crystal phases (Figs. S14–S16)51,52. For the energy diagram of the geometry of single F-AQ molecule in the S1 state, the exciton from the lowest singlet excited state can transition to T1 and T2 state with a SOCME of 0.11 cm–1, and 4.14 cm–1. Since the energy gap between the S1 and T1 state is too larger for RISC (-0.73 eV), it may originate from the higher T2 state. For the minimal-energy geometry in the T2 state, the RISC becomes allowed as the energy gap narrows to 0.01 eV and the SOCME = 6.04 cm–1 is also satisfied for the flipping of the electron spin. Furthermore, the energy gap between T2 and T1 is large enough (0.49 eV) to avoid “hot” exciton decay before RISC.

The comparison of the fluorescence and phosphorescence spectra of the F-AQ@PMMA film are in good agreement with the calculation results. From Fig. 1g, the phosphorescence peak located on the left side of fluorescence peak, which means the anti-Kasha phosphorescence was generated from high level triplet state. The low-temperature transient emission (TRE) spectra of F-AQ@PMMA from 1 ns to 100 ns confirmed that the PF of it located at around 549 nm (Fig. S6), and the emission peak presents a blueshift to 532 nm from 1 ns to 2 s (Fig. 1h and S7), unveiling the conversion from hot exciton TADF to the anti-Kasha phosphorescence. Similar comparisons were also carried out for different solid states, the fluorescence and phosphorescence peaks show a close value to each other (Fig. S8), which provides a satisfactory energy level for efficient hRISC. The high-lying triplet state shows a higher energy level than singlet state for F-AQ@PMMA film and amo-G, and it will make their delayed fluorescence exhibited low sensitivity to temperature (Tables S3, S4). Similar characteristics were found in amo-Y and cry-O, which may result from the small energy splitting between S1 and T2 (Tables S5–S7). As for the cry-YG, it will be affected by temperature to a larger extent because the energy gap of it is much larger than the other solids (Table S6).

To further reveal the intrinsic factors that affect the excited-state dynamics of F-AQ in their different aggregation-states, the crystal structures of cry-YG and cry-O were investigated by single-crystal X-ray diffraction. As the results show, the crystalline phase of cry-YG is monoclinic with space groups of C2/c, while the cry-O exhibits a triclinic phase and P-1 space groups. The selected dimers extracted from the full crystal structure of cry-YG and cry-O (Figs. S10 and S11) are shown in Fig. 2a, and the intermolecular interaction is clearly displayed for two crystals. For cry-YG, there exists an apparent π-π stacking between the acceptor anthraquinone group of two molecules with a distance of about 3.62 Å. However, the π-π stacking is observed between the acceptor anthraquinone group and the donor fluorene group in cry-O, the distance between the hexatomic ring of fluorene and anthraquinone, pentatomic ring of fluorene and anthraquinone are 3.58 Å and 3.90 Å, respectively. Moreover, the Hirshfeld surface was analyzed based on the crystal structure of cry-YG and cry-O (Fig. 2b)53,54. Distinct differences are presented on the surfaces of the two crystals, the interaction facing the conjugated plane of anthraquinone and fluorene in cry-O shows apparently large intensity compared to cry-YG, which is representative for the π-π interaction. Thus, from the above results, we can conclude that the cry-O can have a large possibility to proceed with a TSCT process26,27, while the charge transfer of cry-YG may only exist intramolecularly. Moreover, the proportion of intermolecular C⋯C interaction is 10.0% in cry-YG, whereas this value is 16.9% cry-O (Fig. 2b), suggesting the π-π interaction is much stronger in cry-O compared to cry-YG. The electron-hole distribution of the simulated dimer of these two molecules in an excited state also supports this hypothesis (Fig. 2c). The electron (yellow) and hole (pink) are mainly distributed on the anthraquinone and fluorene in the bottom F-AQ molecule in the picture, whereas the electron is on the anthraquinone part of the upper molecule, and the hole is located on the fluorene group of the bottom F-AQ.

Fig. 2: Crystal structure and analysis of cry-YG and cry-O.
Fig. 2: Crystal structure and analysis of cry-YG and cry-O.
Full size image

a Molecular interaction and overlaps of the F-AQ dimers extracted cry-YG and cry-O (Hydrogen atoms were hidden for clarity). b Hirshfeld surface analysis and the proportions of intermolecular H⋯H, C⋯H, O⋯H, C⋯C, C⋯O, and O⋯O and other interactions to the total intermolecular interactions based on their crystal structures. c Electron-hole distribution of cry-YG and cry-O (The yellow isosurface represents for electron and pink isosurface represents for hole, isovalue = 0.003).

To obtain deeper insight into the deactivation pathway and photodynamics of the F-AQ, the fs/ns-TA spectroscopy experiments were first conducted to investigate its excited-state behaviors in the solutions with different polarity. As the results show in Fig.3, the excited state behaviors of F-AQ will be significantly affected by the polarity of environment. Based on the global analysis utilizing the sequential kinetics scheme, two evolution-associated difference spectra (EADS) species can be separated from the fs-TA spectra measured in low-polarity toluene (Fig. 3a, b), with time constants of 131 ps, and > 3.0 ns, whereas three EADS species can be obtained for the spectra measured in strongly polar ACN (Fig. 3c, d), with time constants of 0.39 ps, 397 ps, and > 3.0 ns. The first EADS species in ACN displays a broad band across 600 to 800 nm can be assigned to the local excited state (1LE) species as it shows similar features to the signal observed in fs-TA spectra in nonpolar cyclohexane (Fig. S18). The second species with excited state absorption (EAS) peaked at around 550 nm is originated from the generation of charge transfer state (1CT) species, as it appears simultaneously with the negative stimulated emission (SE) signal emitted from CT state. Finally, the long-lived third species at ~700 nm can be attributed to triplet state species. For the situation in toluene, the ESA signal of first species is parallel to the mix of 1LE ( ~ 625 nm) state and 1CT state ( ~ 550 nm), indicating an incomplete CT process in a low polar environment, and the second one is the long-lived triplet state species. Moreover, the later period excited-state dynamics is further studied by ns-TA spectroscopy (Fig. 3e–h). In ns-TA spectra, it is noted that a strong signal located at 450 – 500 nm appears within 10 ns, which may be shielded by ground state bleaching (GSB) in the fs-TA spectra. This signal is possible to be generated by the high-lying triplet state (3LE), as the quick IC process from 3LE to the lowest triplet state (3CT) is captured in the fs-TA spectra of F-AQ measured in cyclohexane (Fig. S18b). Hence, in ns-TA spectra, the synchronous decay of 1CT ( ~ 550 nm) and 3LE ( ~ 470 nm) state species can be observed, revealing the equilibrium between the ISC and hRISC processes. Apart from the transition back to singlet state, the 3LE state species will also decay to the lowest triplet state 3CT ( ~ 700 nm), the ratio of these two parallel processes is greatly impacted by polarity of the environment (Figs. 3e–g).

Fig. 3: Fs-TA spectroscopy study on the F-AQ in solution.
Fig. 3: Fs-TA spectroscopy study on the F-AQ in solution.
Full size image

a Contour plots of fs-TA spectra for F-AQ measured in toluene; (b) Normalized EADS obtained from global analysis of fs-TA spectra for F-AQ measured in toluene; (c) Contour plots of fs-TA spectra for F-AQ measured in ACN; (d) Normalized EADS obtained from global analysis of fs-TA spectra for F-AQ measured in ACN; (e) Contour plots of ns-TA spectra for F-AQ measured in toluene; (f) Normalized EADS obtained from global analysis of ns-TA spectra for F-AQ measured in toluene; (g) Contour plots of ns-TA spectra for F-AQ measured in ACN; (h) Normalized EADS obtained from global analysis of ns-TA spectra for F-AQ measured in ACN.

Based on the ultrafast spectroscopic investigation in the solution phase, the solid-state photodynamics can be discussed from the analogous results of fs-TA spectroscopy. From the spectra of the F-AQ@PMMA (Fig. S20), a clear signal of conversion between 1CT state and 3LE state ( ~ 550 nm and ~470 nm) was displayed, revealing the hot exciton mechanism of F-AQ in the single-molecular state. Moreover, the glass state was used to study the general excited-state mechanism of F-AQ in an aggregated state. As a result, similar to F-AQ in toluene solution phase, the CT is not completely proceeded and there exist parallel photophysical processes. Three EADS species can be obtained from the global analysis with time constants of 6.40 ps, 182.2 ps, and >3.0 ns (Figs. 4a and 4d). The first species mainly located at 550 nm and 625 nm, which represents the formation of the 1CT state and 1LE state, and for the second EADS species, the ESA signal at 550 nm and 625 nm slightly decrease and the peak at 470 nm increases, indicating the generation of the 3LE state. Eventually, the long-lived third species can be easily attributed to another parallel process of IC from the 3LE state to the 3CT state. Similarly, the three EADS species originated from the fs-TA spectra of cry-YG shows the same deactivation channel to the glass state, with time constants of 20.6 ps, 297 ps, and > 3.0 ns (Figs. 4b and 4e). The first species shows a broad band from 500 to 650 nm, which results from the signal of the 1CT state and 1LE state. Subsequently, the signal at around 475 nm rises and shows higher intensity than ~550 nm and ~625 nm, suggesting the generation of the 3LE state, and finally the growth peak at 600 nm to 700 nm denotes the formation of the lowest 3CT state. Consequently, the evolution of the ESA signal of cry-O displays an apparent difference to cry-YG. With the decay of the broadband, a new signal centered at 590 nm gradually appears and reaches a maximum at about 2 ns (Fig. 4c). This feature can be attributed to the excited-state AQ anion (Fig. S23)55,56, indicating the intermolecular charge separation has occurred. Considering the signal of the TBCT species locates at 525 nm, we can exclude the possibility of this process. Moreover, the crystal structure analysis indicates there is a satisfactory environment for TSCT, so this slow-generated species can be assigned to the TSCT species. The global analysis revealed that three EADS species with time constants of 0.40 ps, 13.4 ps, and >3.0 ns was involved in the decay pathway of cry-O (Fig. 4f), the first species may be attributed to the generation of the 1CT state and 1LE state, and the second species is assigned to the TSCT process. Considering the third species shows similar features to the second one, it could be caused by the lattice relaxation of the TSCT state. Moreover, the signal ~475 nm which could be the generation of the 3LE state species in the third EADS species, so this lattice relaxation process was accompanied by the ISC process simultaneously. With regard to the amorphous sample, the excited-state behaviors of amo-G and amo-Y is close as they exhibit a similar evolution of signals in their fs-TA spectra (Fig. S21 and S22). Three EADS species were separated by the global analysis, and they can be attributed to the time constants for the generation of the 1CT and 1LE states, the formation of the 3LE state and decay from 3LE state to the lowest 3CT state referred to the analysis result of glass state and solution. Hence, the mechanism of the morphism-dependent TSCT process for F-AQ can be summarized in Fig. 5. The cry-YG, associated with amo-G and amo-Y, first undergo an incomplete TBCT process after excitation, and then both the 1TBCT and 1LE species can transition to the high-lying triplet 3LE state which will transition back to the 1TBCT state to emit TADF (Fig. 5b). As for cry-O, the 1LE state species will generate the TSCT state taking advantage of the face-to-face arrangement between the donor and acceptor. Subsequently, the lattice relaxation of the TSCT state species occurs, followed by the ISC and hRISC to generate hot exciton TADF (Fig. 5c).

Fig. 4: Fs-TA spectroscopy study on the F-AQ in solid states.
Fig. 4: Fs-TA spectroscopy study on the F-AQ in solid states.
Full size image

a Contour plots of fs-TA spectra for glass state of F-AQ; (b) Contour plots of fs-TA spectra for glass state of Cry-YG (c) Contour plots of fs-TA spectra for Cry-O; (d) Normalized EADS obtained from global analysis of fs-TA spectra for glass state of F-AQ; (e) Normalized EADS obtained from global analysis of fs-TA spectra for Cry-YG; (f) Normalized EADS obtained from global analysis of fs-TA spectra for Cry-O.

Fig. 5: General mechanism of F-AQ in different solid states.
Fig. 5: General mechanism of F-AQ in different solid states.
Full size image

a Schematic illustration of the different excited-state behaivours between cry-YG and cry-O. b Excited-state deactivation pathway of cry-YG. c Excited-state deactivation pathway of cry-O.

Utilizing the variation of the luminescence property in the polymorphism of F-AQ, potential applications in different functional luminescence materials, such as luminescence pigments, thermochromic material, or mechanochromic material can be realized. Utilizing the strong impact of solvent polarity on the formation of aggregated state, the F-AQ can be fabricated into two pigments with yellow and green colors, and the perforate glass plate is acted as canvas (Fig. 6a). The F-AQ solution with different solvents were carefully dropped into specific positions based on the design, and the pattern can be generated after evaporation of the solvents (Fig. 6b). Moreover, the yellow pattern “Loong” displayed in Fig. 6c was fabricated by amo-Y. After 160 °C heating for 10 min, the color of the pattern changed into green. In addition, the yellow green cry-YG can generate green amo-G, whereas orange cry-O can generate yellow amo-Y after grinding (Fig. S24). These two instances present the great potential of F-AQ to be fabricated into the thermochromic material and mechanochromic material.

Fig. 6: Application utilizing the polymorphism properties of F-AQ.
Fig. 6: Application utilizing the polymorphism properties of F-AQ.
Full size image

a General procedures of preparing the pigments with different colors for painting. b Design and product of the pattern drawn in porous glass plates utilizing the F-AQ pigments. c The change in luminescence color of the “Loong” pattern fabricated by cry-O after heating to 160 °C.

In conclusion, we reported a polymorphism hot exciton luminescence material, and the differences in emission properties and the excited-state dynamics were investigated with a variety of spectroscopic methods. Thereinto, the cry-O which has the face-to-face stacking of acceptor anthraquinone and the donor fluorene shows a significantly longer emission lifetimes and higher quantum yields compared to cry-YG with a stacking between the anthraquinones of two F-AQ molecules. Utilizing ultrafast spectroscopy, the photodynamics of these solids were characterized. An increasing signal intensity of the excited-state anthraquinone anion was observed in the spectra of cry-O, indicating the generation of a TSCT species via intermolecular charge transfer between the packing molecules. However, the other three solid state polymorphs will not generate this species and traditionally generate hot exciton TADF similar to the situation in the solution phase. Finally, these molecules show a promising potential in the multicolor pigments, mechanochromic materials, and thermochromic materials by utilizing their polymorphism. In this work, we directly observed the morphism-dependent through-space charge transfer for one molecule and increased our knowledge of polymorphism-dependent properties of solid-state systems.

Methods

Synthesis

In a 250 ml sealed flask, 287.98 mg (1.0 mmol) of 2-bromoanthracene-9,10-dione, 285.79 mg (1.2 mmol) of (4- (9,9-dimethyl-9H-fluoren-2-yl)boronic acid, 529.90 mg(5.0 mmol) of Na2CO3, and 34.69 mg (0.03 mmol) of Pd(PPh3)4 were dissolved in the mixture of 20 ml THF and 2 ml H2O. The mixture was heated to 80 °C and stirred for 12 h under the nitrogen atmosphere. After cooling to room temperature, the mixture was extracted by EA for three times. The organic layer was washed by saturated NaCl three times and dried by anhydrous Na2SO4. After filter and evaporation, the crude product was further purified by silica column chromatography. Orange solid was obtained in 79.8% yield.

1H NMR (400 MHz, Chloroform-d) δ 8.61 (d, J = 1.9 Hz, 1H), 8.40 (d, J = 8.1 Hz, 1H), 8.36 (dp, J = 5.9, 3.5 Hz, 2H), 8.10 (dd, J = 8.1, 2.0 Hz, 1H), 7.88 – 7.69 (m, 6H), 7.48 (dt, J = 7.0, 3.0 Hz, 1H), 7.43 – 7.32 (m, 2H), 1.58 (s, 6H).

13C NMR (101 MHz, CDCl3) δ 183.42, 182.92, 154.63, 154.07, 147.26, 140.16, 138.36, 137.89, 134.22, 134.06, 133.91, 133.71, 133.64, 132.40, 131.96, 128.09, 127.83, 127.31, 127.24, 127.18, 126.50, 125.47, 122.73, 121.54, 120.64, 120.40, 77.34, 77.02, 76.70, 47.11, 27.19.

HRMS (QTOF-ESI) m/z [M + H]+: calculated 401.1643; found 401.1530.

Preparation of the solid in different morphological state

Glass state

The glass state was prepared by melting the F-AQ in 250 °C on the surface of the quartz plate.

Amo-G

The amo-G was prepared by fast evaporation by a rotary evaporator in a saturated solution of F-AQ in DCM:ACN = 1:1.

Amo-Y

The amo-Y was prepared by fast evaporation by a rotary evaporator in a saturated solution of F-AQ in DCM:hexane = 1:1.

Cry-YG

The bulk cry-YG was prepared by slow evaporation in DCM:ACN = 1:1. The fast preparation was realized by heating the supersaturated solution of F-AQ in DCM:ACN = 1:1 to about 60 °C to saturated, and cooling down to collect the precipitated small crystal.

Cry-O

The bulk cry-O was prepared by slow evaporation in DCM:hexane = 1:1. The fast preparation was realized by heating the supersaturated solution of F-AQ in DCM:hexane = 1:1 to about 60 °C to saturated, and cooling down to collect the precipitated small crystal.

Powder X-ray diffraction

The solid sample was measured by MiniFlex 600benchtop powder X-ray diffractometer (Rigaku, Japan).

Single-crystal X-ray diffraction

Single crystal X-ray data was collected on a Bruker D8 VENTURE with Ga Kα radiation (λ = 1.34139 Å) at 100 K.

Steady-state UV-vis spectroscopy

The F-AQ were prepared into 10–5 M solution of ACN. The UV-vis spectra of the solution and solid sample were recorded by Cary 5000 UV-VIS-NIR spectrophotometer (Agilent, USA).

Steady-state photoluminescence spectroscopy and transient decay spectroscopy

For the solution sample, the F-AQ were prepared into 10–5 M solution of n-hexane, toluene, DCM, THF, ACN, and DMSO. The steady-state PL spectra and emission decay curves of the solution and solid sample were measured by FLS1000 Steady State & Time-Resolved Photoluminescence Spectrometer (Edinburgh, UK).

The steady-state PL spectra, emission decay curves, temperature-dependent PL spectra, and temperature-dependent emission decay curves of samples were also measured by FLS1000 Steady State & Time-Resolved Photoluminescence Spectrometer (Edinburgh, UK).

Phosphorescence spectroscopy

The phosphorescence spectra of the films were measured by LP920 transient absorption/emission spectrometer (Edinburgh, UK) with a 355 nm pump beam was obtained from the third harmonic output of an Nd:YAG laser. The Phosphorescence spectrum was measured with a 1-ms delayed gate.

Photoluminescence quantum yield measurement

The F-AQ were prepared into 10–5 M solution of toluene. The PL quantum yield of the solution and solid sample were measured by Quantaurus-QY C11347-12 absolute PL quantum yield spectrometer (Hamamatsu, Japan).

Fs-TA spectroscopy

The fs-TA measurements were conducted using the Helios pump-probe transient absorption spectrometer system (Ultrafast Systems, USA) with the femtosecond laser from the Spitfire Pro regenerative amplified Ti:sapphire laser system (Spectra Physics, USA). The 800-nm laser light with 120-fs pulse width was subsequently split into two beams, one as the pump beam and another one as the probe beam. The pump beam passed through a harmonic resonator to generate the 400-nm pump beam (the second harmonic of the fundamental 800 nm), whereas the probe beam passed through a sapphire crystal and generated a white-light continuum (400–850 nm). The time-delayed probe beam was controlled by the optical delay rail with a maximum temporal delay at 7000 ps. It would pass through the samples and the signals were then collected by the detector. A reference probe beam was also used to optimize signal-to-noise ratio. The sample solutions were prepared with absorbance approximately to 1.0 at 400 nm and measured in a 2-mm path-length quartz cuvette, and the films were directly fixed on the sample holder. The amorphous solid was mixed with anhydrous KBr, grinded, and squashed into thin plate. The crystal sample was evaporated onto the 25 mm×25 mm×1 mm quartz plate in a saturated solution with different solvents. The consistency of the photophysical properties between the film sample and single crystal sample was confirmed by PL spectra and decay curves (Fig. S8). The spectrometric data were recorded in a 3D wavelength-time-absorbance matrix.

The subtraction of background, the subtraction of scattering light, and chirp correction were done for all the data before analysis. The principal components were ensured by singular value decomposition (SVD), and then the global fitting was carried out with the selected principal components and exponential function using the sequential kinetics scheme based on the fs-TA spectra by Surface Xplorer 4.5. The single-wavelength kinetic fitting was carried out utilized Eq. (2) by Surface Xplorer 4.5.

$$S\left(t\right)=\,{e}^{{-\left(\frac{t-{t}_{0}}{{t}_{p}}\right)}^{2}}\,\times {\sum} _{i}\,{A}_{i}{e}^{-\left(\frac{t-{t}_{0}}{{t}_{i}}\right)}$$
(2)

Ns-TA spectroscopy

The ns-TA measurement is based on the EOS pump-probe ns-TA spectrometer system (Ultrafast Systems, USA). The 400-nm pump beam is generated from the femtosecond regenerative amplified Ti:sapphire laser system with similar configuration to fs-TA measurement, while the probe beam is generated from a photonic crystal fiber-based sub-nanosecond pulsed probe light source. It would pass through the samples and the signals were then collected by the detector. A reference probe beam was also used to optimize signal-to-noise ratio. All the sample solutions were tested in a 2 mm path-length cuvette. The spectrometric data were recorded in a 3D wavelength-time-absorbance matrix.

The subtraction of background and scattering light were done for all the data before analysis. The principal components were ensured by singular value decomposition (SVD), and then the global fitting was carried out with the selected principal components and exponential function using the sequential kinetics scheme based on the fs-TA spectra by Surface Xplorer 4.5.

Theoretical calculations

The minimal-energy geometry of single-molecular F-AQ in S0, S1, T1, and T2 state were optimized by Gaussian 16 package based on B3LYP-D3/6-311 G(d,p) level. The minimal-energy geometry of the cry-YG and cry-O for F-AQ in S0, S1, T1, and T2 state were optimized by Gaussian 16 package using the QM/MM method. Clusters constructed from the dimer of F-AQ and the surrounded molecules were extracted from the crystal structure of cry-YG and cry-O. The ONIOM high layer included the central dimer highlighted, was calculated based on B3LYP-D3/6-311 G(d,p) level, while the ONIOM low layer included the molecules surrounded by the central dimer, which was calculated using the UFF (Fig. S13). The vertical absorption energy of the minimal-energy geometries in S0 state, vertical emission energy of the minimal-energy geometries in S1, T1, and T2 state, and the spin-orbital coupling matrix element (SOCME) were calculated by ORCA 5.0.3 based on B3LYP-D3/6-311 G(d,p) level. The electron-hole distribution was calculated by Multifwn 3.8, and the depiction was conducted by VMD. The Hirshfeld surface analysis was performed by Multifwn 3.8, and the depiction was conducted by VMD.