Supporting Information Two Boron Atoms versus One: High-performance Deep-blue Multi-resonance Thermally Activated Delayed Fluorescence Emitters

Two new deep-blue narrowband multi-resonant emitters, 1B-DTACrs and 2B-DTACrs, one of which shows thermally activated delayed fluorescence (TADF), based on boron, nitrogen, and oxygen doped nanographenes are reported. Devices based on 2B-DTACrs showed an EQEmax of 14.8% and CIE coordinates of (0.150, 0.044), which are very close to the BT.2020 requirement for blue pixels.

Thermally activated delayed fluorescence (TADF) materials can achieve 100% internal quantum efficiency (IQE) by harvesting both singlet and triplet excitons. In 2012, Adachi and coworkers first reported organic light-emitting diodes (OLEDs) that showed IQEs of 100% based on organic donor-acceptor (D-A) type TADF emitters. 1 Successively, a wide variety of D-A type TADF emitters have been developed, resulting in devices showing high external quantum efficiencies (EQEs) of over 20% for blue, green, and red emission colors. [2][3][4][5][6] Nonetheless, D-A type TADF emitters usually possess broad emission bands that result in devices with poor color purity pixels. To achieve the BT.2020 standard for high-definition displays, narrowband electroluminescence (EL) in devices that show high EQE is highly desired.
Beyond the D-A type emitters, in 2016, Hatakeyama and coworkers first reported pure-blue boron-containing multi-resonance TADF (MR-TADF) emitters that feature narrowband emissions and high photoluminescence quantum yields (F PL , PLQYs). 7 The narrowband emission behaviour of MR-TADF emitters is the most suitable for high-definition display with a wide colour gamut, i.e., BT.2020 standard. Since then, the development of MR-TADF emitters has attracted much attention. [8][9][10] To date, there have been numerous reports on high-efficiency OLEDs based on MR-TADF emitters with blue, green, and red emissions. [8][9][10][11][12][13] Nonetheless, there are a limited number of reports on high-efficiency MR-TADF emitters that can fulfil the requirement of the colour gamut for the blue pixel, i.e., Commission Internationale de lÉclairage coordinates (CIE x,y ) = (0.131, 0.046). 14 As a result, it is necessary to develop high-efficiency deep-blue MR-TADF emitters that satisfy the colour gamut requirement of the BT.2020 standard, i.e., CIE y r 0.05.
Here, we report two boron-based deep-blue emitters containing a mixed O/N donor atom system, one of which is MR-TADF. 1B-DTACrs and 2B-DTACrs were synthesized in five steps starting from 3,5-difluorobromobenzene. The key borylation reaction proceeded in 62% and 58% yield, respectively, for 1B-DTACrs and 2B-DTACrs from the common intermediate 4.
Crystals for 1B-DTACrs were obtained by slow diffusion of hexane into a saturated solution of the emitter in THF. Crystals of 2B-DTACrs were obtained directly from a temperature gradient sublimation method (Fig. 1). The ditolylamine (DTA) unit adopts a propeller structure. In the case of 1B-DTACrs, where there is only one boron in the compound, the MR-TADF core remains nearly planar. The o-tolyl unit is approximately perpendicular to the aromatic core in both emitters. Yasuda and co-workers reported BSBS-N1, 15 a structurally analogous sulfur-containing compound to 2B-DTACrs, which was described as having a highly distorted structure, whereas 2B-DTACrs possesses a nearly planar geometry at the B,Osubstituted pentacene core with an out-of-plane alignment of two aryl rings of the DTA unit. 15 There are no significant intermolecular p-p interactions in the solid state for either compound, which may be a consequence of the orthogonally disposed o-tolyl unit.
The ground-and excited-state properties of 1B-DTACrs and 2B-DTACrs were modelled using DFT and spin-component scaling second order approximate coupled-cluster (SCS-CC2) methods, respectively (Fig. 2). Both compounds possess difference density distributions between the ground and excitedstate reminiscent of other MR-TADF emitters. 10 Addition of the second boron atom in 2B-DTACrs resulted in a small stabilization in the HOMO and LUMO energy levels compared to those of 1B-DTACrs and a corresponding small decrease (0.12 eV) in the band gap ( Fig. 2(a)). The S 1 state is stabilized from 3.38 eV in 1B-DTACrs to 3.27 eV in 2B-DTACrs, while the T 1 states are essentially isoenergetic. Thus, there is a much smaller DE ST value of 0.08 eV for 2B-DTACrs compared to 0.20 eV of 1B-DTACrs. The oscillator strength ( f ) for the S 0 -S 1 transition is higher at 0.36 for 2B-DTACrs compared to 0.23 for 1B-DTACrs likely for the apparent largest delocalization of the S 1 excited state. Both the S 2 and T 2 states of 2B-DTACrs are also stabilized compared to those of 1B-DTACrs. The small energy difference between S 1 and T 2 (DE ST 2 ) is of particular interest as many reports have highlighted that reverse intersystem crossing (RISC) occurs via intermediate triplet states. 16 The DE ST 2 drops from 0.34 eV in 1B-DTACrs to 0.02 eV in 2B-DTACrs. For each compound, spin-orbital coupling (SOC) is modest between S 1 and T 1 , calculated to be 0.07 cm À1 and 0.04 cm À1 for 1B-DTACrs and 2B-DTACrs, respectively. This increases to 0.81 cm À1 and 0.98 cm À1 between S 1 and T 2 for 1B-DTACrs and 2B-DTACrs, respectively (Table S3, ESI †). These results suggest that RISC likely proceeds from T 2 to S 1 , particularly for 2B-DTACr. 17 Tables S1-S3 (ESI †) provide a full summary of the excited state energies, transitions and SOC values.
The electrochemical properties were measured using cyclic voltammetry (CV) and differential pulse voltammetry (DPV) in deaerated dichloromethane (DCM) with 0.1 M tetra-nbutylammonium hexafluorophosphate as the supporting electrolyte (Fig. S21 and Table S4, ESI †). The oxidation potentials, E ox , determined from the peak of the first oxidation wave of the DPV, are 1.07 V and 1.17 V versus a saturated calomel electrode (SCE), respectively, for 1B-DTACrs and 2B-DTACrs. The corresponding reduction potentials, E red , are À2.10 V and À1.89 V, respectively, versus SCE for 1B-DTACrs and 2B-DTACrs. The corresponding HOMO and LUMO levels were calculated to be À5.42 eV and À2.24 eV for 1B-DTACrs. 18 The HOMO and LUMO energy levels are stabilized to À5.51 eV and À2.46 eV, respectively, for 2B-DTACrs. The electrochemical gap is thus reduced from 3.17 eV for 1B-DTACrs to 3.05 eV for 2B-DTACrs, in line with the trend predicted by the gas-phase DFT calculations.
The photophysical properties of 1B-DTACrs and 2B-DTACrs were firstly examined in dilute toluene solutions (10 À5 M). The photophysical data are shown in Fig. 3, and Table 1. Both emitters showed absorption bands below 350 nm, which are assigned as locally excited (LE) p-p* transitions, while the absorption bands beyond 400 nm are assigned as short-range  charge transfer (SRCT) bands. Both compounds possess narrow high-energy bands assigned to transitions to the S 1 state, with molar absorptivity coefficients, e, of 3 and 8 Â 10 5 M À1 cm À1 for 1B-DTACrs and 2B-DTACrs, respectively, in line with the larger computed f associated with the S 0 -S 1 transitions (vide supra).
The higher calculated f value and measured e in 2B-DTACrs originates from the extended conjugation in the compound. 19 1B-DTACrs and 2B-DTACrs emit in the deep-blue in toluene at 438 and 443 nm, respectively. The PL spectra of 1B-DTACrs and 2B-DTACrs are narrow with full-width at half-maxima (FWHM) values of 27 and 21 nm, respectively. The narrowband emission coupled with the small Stokes-shifts reflect the rigid nature of these compounds and the small degree of reorganization in the excited state. The F PL in degassed toluene solutions of 1B-DTACrs is 99%, while 2B-DTACrs showed a rather lower F PL of 62%. When doped in a 3,3 0 -di(9H-carbazol-9-yl)-1,1 0biphenyl (mCBP) host at 5 wt% of 1B-DTACrs or 2B-DTACrs, the F PL values remain high at 83 and 74%, respectively. Both doped films display emissions (Fig. 4) solely originated from the corresponding emitter, which indicate an efficient energy transfer from mCBP to the emitter. The transient decay profiles of both doped films reveal that 1B-DTACrs showed no TADF behaviour. By contrast, 2B-DTACrs showed TADF behaviour with a delayed lifetime (t d ) of 13.1 ms. The TADF behaviour of 2B-DTACrs was further confirmed by temperature-dependent transient emission decay measurement (Fig. S22, ESI †). The DE ST values of the mCBP doped films were determined from the difference between the onsets of the fluorescence and phosphorescence spectra at 77 K (Fig. S23, ESI †). The fairly large DE ST in 1B-DTACrs-doped film (0.21 eV) may account for the non-TADF behaviour while that of 2B-DTACrs is smaller at 0.16 eV. The rate constants of doped films of 1B-DTACrs and 2B-DTACrs were calculated, and the data are summarized in Table 1. The reverse intersystem crossing rate (k RISC ) of 2B-DTACrs was found to be 1.3 Â 10 5 s À1 , which is amongst the fastest k RISC for deep-blue MR-TADF emitters with CIE y r 0.10 (Table S5, Table 2 depict all the device characteristics. Devices based on 1B-DTACrs (Device I) and 2B-DTACrs (Device II) produced deep-blue narrowband emission at 440 and 447 nm, respectively, which are consistent to their corresponding PL spectra. The EL spectra clearly confirm the complete exciton confinement in both devices. The FWHM of Device I is 30 nm, which is slightly wider than that of Device II (FWHM = 26 nm). The corresponding CIE x,y are (0.154, 0.049) and (0.150, 0.044), respectively. Although the color gamuts of both devices are very close to the BT.2020 standard for the blue pixel, i.e., (0.131, 0.046), the   efficiencies of two devices varied significantly. Taking advantage of enhanced exciton harvesting ability of 2B-DTACrs, Device II demonstrated an excellent EQE max of 14.8%, which is one of the highest performing deep-blue MR-TADF OLEDs with CIE y r 0.05 (Table S5, ESI †); 14 there is, however, considerable efficiency roll-off beyond 100 cd m À2 . By contrast, Device I only showed a poor EQE max of 1.3%, which is in part due to the fact that only singlet excitons can be utilized. Additionally, the poor carrier transporting properties in device I, in which a high turn-on voltage was obtained, may contribute to the lower EQE than the theoretical value (5%).
In conclusion, we presented two deep-blue p,n-doped nanographene emitters (1B-DTACrs and 2B-DTACrs), in which 2B-DTACrs with two boron atoms is TADF-active while 1B-DTACrs with only one boron atom is TADF inactive. It is noteworthy that the emitter 2B-DTACrs does not show redshifted emission. Devices based on 2B-DTACrs achieved a high EQE max of 14.8% and CIE x,y of (0.150, 0.044), which are one of highest efficiency deep-blue MR-TADF devices reported to date. Importantly, the CIE coordinates effectively meet the colour gamut requirement for the BT.2020 standard for the blue pixel.
The authors acknowledge Ms N. Nakamura and Ms K. Kusuhara for their technical assistance with this research. This work was supported financially by the JSPS Core-to-Core Program (grant number: JPJSCCA20180005) and Kyulux Inc. S. M. S. acknowledges support from the Marie Skłodowska-Curie Individual Fellowship (grant agreement No 838885 Narrow-bandSSL). We would like to thank the Leverhulme Trust (RPG-2016-047) for financial support. E. Z.-C. is a Royal Society Leverhulme Trust Senior Research fellow (SRF\R1\201089). We thank the EPRSC for funding (EP/R035164/1

Conflicts of interest
There are no conflicts to declare.
Notes and references