Aggregation‐induced conversion from TADF to phosphorescence of gold(I) complexes with millisecond lifetimes

Long‐lived luminescent materials possess wide application prospects in various areas, but their constructions still face a huge challenge, especially the achievement in solution. Manipulating aggregate is an effective method to attain intriguing luminescence, thus it is expected to build long‐lived materials in solution. In this work, a series of new gold(I) complexes were developed by employing unique carbazole and phosphine as the ligands, and these resulting systems exhibited rare long lifetimes of milliseconds in the aggregate state. It was well unveiled that these complexes emitted blue thermally activated delayed fluorescence (TADF) with the lifetimes of several microseconds in dilute solution; while orange‐red phosphorescence with much longer lifetimes of several milliseconds were achieved in the aggregate state. To our knowledge, this is the first implementation of aggregation‐induced conversion from TADF to phosphorescence. Based on their excellent luminescent properties, we successfully applied these complexes in cell imaging and inhibition of cancer cells.


INTRODUCTION
Long-lived luminescent materials have attracted considerable attention from researchers due to their long lifetimes and extensive applications in display, bioimaging, biotherapy, information storage, anti-counterfeiting, and so forth. [1][2][3][4][5] Currently, most of the existing long-lived materials are implemented via the strategy of promoting the intersystem conversion (ISC) process and suppressing the nonradiative process. The spin-orbit coupling (SOC) and ISC channel can be enhanced through molecular modification by introducing heavy atoms or groups such as heteroatoms, metal atoms, and carbonyl groups. [6,7] On the other hand, nonradiative decay can be inhibited via providing a rigid environment by crystallization, multiple interactions, H-aggregates, or self-assembly. [8][9][10] However, it is very difficult to exert effective manipulation of these internal or external factors of molecules due to the complex modification of molecular structures for constructing long-lived luminescent materials. Therefore, how to develop an efficient and simple way to construct long-lived materials still remains a great challenge.
Room temperature phosphorescent (RTP) materials, as one of the main long-lived materials, have been the star materials in the field of solid luminescent materials. [11] Such kind of materials are mainly obtained by constructing molecular crystals. In crystals, molecules all arrange neatly and pack tightly, making their luminescence difficult to be annihilated by external water or air. [12] And the molecular conformations are fixed to the maximum extent, which facilitates the ISC process and thus enhances the lifetime. [13] For example, Xie and coworkers reported a series of pure organic molecular crystals to achieve RTP and disclosed that the compact faceto-face packing arrangement was in favor of the long RTP lifetime. [14] Li's group developed RTP crystals with noncentrosymmetric structure and proposed that building hydrogen bonds and H-aggregates were beneficial to immobilize the molecular conformation and extend the phosphorescence lifetime. [15] However, the strict building conditions of molecular crystals usually result in their poor reproducibility and low stability and seriously limit their application in different circumstances, such as biological environment.
Compared with long-lived crystal materials in the solid state, the luminescent materials that can maintain long-lived properties in the solution state doubtlessly will have much broader application. Typically, in biological field, long-lived phosphorescence can eliminate the fluorescence interference of background. [16] But it is very difficult to achieve long lifetime in the solution state, because molecules can move more freely in solution and the manipulation of their motions and interactions are more difficult. Fortunately, a unique phenomenon, aggregation-induced emission (AIE) coined by Prof. Tang et al., [17] provides infinite opportunity for the effective molecular regulation in the solution state. In AIE systems, the changes of luminescence intensity, emission color, and even lifetimes all could be realized through aggregation regulation. [18][19][20] Based on the literature research, aggregation-induced delay fluorescence (AIDF) systems with lifetime of millisecond were reported, which showed obvious delayed components in the aggregate. [19h] Additionally, a few sporadic systems with aggregation-induced phosphorescent emission (AIPE) properties were also developed recently, where long lifetimes of several microseconds were achieved in the aggregate state. [21,22] Although AIE has shown huge potential in the construction of long-lived aggregated materials, such systems with long phosphorescence are still very rare until recently. Therefore, there is an urgent need to seek new design strategy to develop abundant AIE systems that can achieve long-lifetime luminescence in solution.
Gold, as an attractive heavy metal atom with large relativistic effect, can effectively enhance the ISC process and promote longevity. [23] In another aspect, carbazole and phosphine groups have been widely explored as triplet chromophores for both TADF and RTP. [24] Hence, in this work, we proposed a new strategy via integrating these three components to construct a new class of gold(I) complexes. Further utilizing aggregation manipulation, the resulting systems were expected to exhibit long-lived luminescence in aggregates. As envisaged, these newly designed complexes emitted blue TADF with the lifetime of several microseconds in diluted solution. More impressively, after aggregation, these complexes emitted bright orange-red phosphorescence with much longer lifetimes of several milliseconds. As far as we know, such amazing aggregation-induced conversion from TADF to phosphorescence has not been reported so far (Scheme 1). Furthermore, these complexes were successfully applied in bioimaging and inhibition of cancer cells.

RESULTS AND DISCUSSION
The synthetic route to the target gold(I) complexes 1-4 is shown in Scheme S1. Carbazole potassium salts were obtained by reacting carbazole with potassium hydroxide. Phosphine ligands and (tht)AuCl were coordinated to obtain gold(I) chloride complexes 1b-4b. Then the target complexes 1-4 were prepared with high yields of 82%-88% using the readymade carbazole potassium salt and the gold(I) chloride complexes. These complexes could be dissolved in organic solvents such as dichloromethane, tetrahydrofuran, acetonitrile, but hardly dissolved in water, n-hexane, and ether. Using MeCN and water as the good solvent and poor solvent, respectively, the emission spectra with different volume fractions of water (f w ) in MeCN/water mixtures (c = 50 µM) were tested. Taking complex 1 as an example, its pure MeCN solution exhibited very weak blue emission with a peak at 400-500 nm. With the addition of poor solvent water, new strong emission peaks at 579 and 625 nm appeared and were gradually enhanced. When f w reached 70%, a 126-fold increase was realized in emission intensity due to the formation of aggregates ( Figure 1A,B). Additionally, the detected quantum yield also increased from 13.68% to 22.29%. More amazingly, there was also an obvious increase in the detected lifetimes during the aggregation. As shown in Figure 2C,D, the lifetime of complex 1 in pure MeCN solution was 31.79 µs, while the lifetime of its aggregate (f w : 70%) reached 2.30 ms. Such huge difference should be unprecedented in AIE systems. [18][19][20][21][22] In addition, decreased and wider UV absorption after aggregation indicated the formation of nano-aggregates after the addition of water ( Figure  S14). According to these results, it could be speculated that both the monomers and aggregates were emissive, while their luminescence might have originated from different excited states. [25] Furthermore, consistent microsecond lifetimes of complex 1 in pure MeCN solution at different concentrations were obtained, which indicated that the long millisecond lifetime of its MeCN-water mixture was due to the formation of aggregates ( Figure S15 and Table S2). To verify the reliability of the above phenomena, we further tested the other three complexes. And the results showed that their MeCN-water mixtures exhibited similar changes in luminescence and corresponding lifetime to those of complex 1, which reinforced that such systems were a new class of AIPE luminogens with long lifetimes ( Figure S13).
Considering the unique properties of the present systems in the aggregate states, we speculated that the restriction of intramolecular motion (RIM) and the enhanced intermolecular interactions might suppress the nonradiative decay channel, leading to the long lifetime after aggregation. To further investigate the luminescence sources of these complexes, TD-DFT calculations for complex 1 were performed in both the solution and aggregate states (Figure 2). The calculated vertical excitation energy was 436 nm (2.84 eV), which was in good agreement with the experimentally observed value of 430 nm. According to the calculated results, when the molecule of solution state was excited from the ground state S 0 to the minimum of the singlet excited state S 1 , a large change of the molecular conformation occurred. Such obvious conformational change might have resulted from the free motions of the molecule in solution. After careful analysis, we found that the molecular conformation of the S 1 excited state (DF, Figure S17) was very close to that of the intersection of S 1 and T 1 states (MECP[T 1 S 1 ], Figure S17) with a small energy difference ( Figure S16), which was an important feature of TADF materials. [26] In addition, the center distance between the HOMO and the LUMO was as large as 12.56 Å, which was also consistent with the orbital-separation characteristics of TADF materials ( Figure S16A). [27] Hence, we reasoned that the molecule could cross back the radiable singlet excitons (S 1 ) through the reverse intersystem crossing (RISC) process and then emitted TADF in solution. In fact, the temperature-dependent photoluminescence spectra of complex 1 were measured in solution ( Figure S18). The results further suggested that the thermal activation process indeed existed at 110 K-180 K. All these results demonstrated that the emission of complex 1 in solution originated from TADF. In contrast, the molecules could not reach the lowest energy point of S 1 state after being excited in the aggregate state. This might be due to the fact that molecules could only undergo very small conformational changes owing to RIM in the aggregate state. Then, phosphorescence was generated through ISC from S 1 to T 3 and the internal conversion (IC) process from T 3 to T 1 . As a result, impressive long millisecond lifetime was achieved. Crystals are a kind of highly ordered aggregates and can intuitively reflect aggregation patterns. Fortunately, the single crystal of complex 1 was obtained by solvent diffusion in the dichloromethane/ether mixture. The single crystal structure is shown in Figure 3, and the detailed crystallographic data are collected in Tables S5 and S6. The crystal of complex 1 emitted orange-red light (587 and 635 nm) under the irradiation of 365 nm UV lamp, consistent with the luminescence in its aggregate state. Furthermore, the intermolecular interactions and stacking mode in the crystal were studied in detail. As shown in Figure 3, there were multiple C-H•••π interactions between the molecules with the distances of 2.309, 2.766, 2.777, 2.824, and 2.861 Å. Each molecule was connected to surrounding molecules through multiple C-H•••π interactions, forming a head-to-tail antiparallel arrangement. Such rigid environment provided by intermolecular C-H•••π interactions restricted the intramolecular motions and further promoted the occurrence of long phosphorescence in the aggregate state. This observation further reinforced the hypothesis that the RIM in aggregates was responsible for the long millisecond lifetimes of these complexes.
Furthermore, the photophysical properties of these complexes in the solid state were also studied. At room temperature (Figure 4), the emission spectra of complex 1 showed a maximum peak at 587 nm and a shoulder peak at ∼640 nm, showing bright orange-red luminescence. The measured lifetime of its powder was 2.63 ms, which was very close to the value of its aggregate state (2.30 ms), further proving the importance of aggregation for extending lifetime. To verify the phosphorescence nature of these complexes, the emission spectra at 77 K were tested and complex 1 showed two groups of emission peaks in the range of 400-550 and 550-700 nm, respectively. Longer lifetimes in 77 K declared their phosphorescence nature in the solid state. The luminescence of the powders of complexes 2-4 at room temperature were similar to that of complex 1 showing long lifetimes of 3.15, 3.26, and 3.14 ms, respectively. These results indicated that these complexes were a new kind of RTP materials with long millisecond lifetimes.
Based on the long-lived and bright luminescence of complexes 1-4 in the aggregate state, we tried to apply them in cell imaging. We first assessed the photostability of complexes 1-4 by testing the fluorescence intensity changes of A549 cells stained with different complexes after sequential scans of laser irradiation ( Figure S21). The results showed that complexes 1-4 showed good photostabilities, which were desirable for the following bioimaging. As shown in Figure 5 and Figure S22, after incubation with complexes 1-4 for 2 h, the A549 cancer cells exhibited strong emission, indicating their excellent cellular imaging abilities. According to our previous studies, gold(I) complexes possess selective inhibition capability toward cancer cells. [28] Therefore, we further evaluated the anticancer abilities of complexes 1-4. As demonstrated in Figure 5 and Figure S23, except for complex 1, the other three complexes exhibited remarkable cytotoxicity toward A549 cells. And the cell viability decreased gradually with the increase of their respective concentration, indicative of effective and strong dose-dependent cytotoxicity. Among them, the inhibition effect of complex 4 toward A549 cells was the most obvious, suggesting its more superior anticancer ability. Taken all these data together, it was demonstrated that this series complexes were promising for bioimaging and anticancer applications.

CONCLUSION
In summary, a series of gold(I) complexes with phosphine and carbazole ligands were designed and synthesized.

C O N F L I C T O F I N T E R E S T
The authors declare no competing financial interest.

D ATA AVA I L A B I L I T Y S TAT E M E N T
CCDC 2116194 contains the supporting crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via https:// www.ccdc.cam.ac.uk/data_request/cif/.