Highly Efficient Green and Red Narrowband Emissive Organic Light‐Emitting Diodes Employing Multi‐Resonant Thermally Activated Delayed Fluorescence Emitters

Abstract Herein, we demonstrate how judicious selection of the donor decorating a central multi‐resonant thermally activated delayed fluorescence (MR‐TADF) core based on DiKTa can lead to very high‐performance OLEDs. By decorating the DiKTa core with triphenylamine (TPA) and diphenylamine (DPA), 3TPA‐DiKTa and 3DPA‐DiKTa exhibit bright, narrowband green and red emission in doped films, respectively. The OLEDs based on these emitters showed record‐high performance for this family of emitters with maximum external quantum efficiencies (EQEmax) of 30.8 % for 3TPA‐DiKTa at λEL of 551 nm and 16.7 % for 3DPA‐DiKTa at λEL=613 nm. The efficiency roll‐off in the OLEDs was improved significantly by using 4CzIPN as an assistant dopant in hyperfluorescence (HF) devices. The outstanding device performance has been attributed to preferential horizontal orientation of the transition dipole moments of 3TPA‐DiKTa and 3DPA‐DiKTa.

S-4 by sparging with DCM-saturated nitrogen gas for 5 minutes prior to measurements. All measurements were performed using 0.1 M DCM solution of tetra-n-butylammonium hexafluorophosphate, [nBu4N]PF6]. An Ag/Ag + electrode was used as the reference electrode while a platinum electrode and a platinum wire were used as the working electrode and counter electrode, respectively. The redox potentials are reported relative to a saturated calomel electrode (SCE) with a ferrocenium/ferrocene (Fc/Fc + ) redox couple as the internal standard (0.46V vs SCE). [1] Photophysical measurements. Optically dilute solutions of concentrations on the order of 10 -5 or 10 -6 M were prepared in spectroscopic or HPLC grade solvents for absorption and emission analysis. Absorption spectra were recorded at room temperature on a Shimadzu UV-2600 double beam spectrophotometer with a 1 cm quartz cuvette. Molar For emission studies, steady-state emission, excitation spectra and time-resolved emission spectra were recorded at room temperature using an Edinburgh Instruments FLS980 fluorimeter. Samples were excited at 340 nm for steady-state measurements.
Photoluminescence quantum yields for solutions were determined using the optically dilute method, [2] in which four sample solutions with absorbances of ca. 0.10, 0.075, 0.050 and 0.025 at 342 nm were used. The Beer-Lambert law was found to remain linear at the concentrations of the solutions. For each sample, linearity between absorption and emission intensity was verified through linear regression analysis with the Pearson regression factor (R 2 ) for the linear fit of the data set surpassing 0.9.
Individual relative quantum yield values were calculated for each solution and the values reported represent the slope obtained from the linear fit of these results. The S-5 quantum yield of the sample, FPL, can be determined by the equation [2] where A stands for the absorbance at the excitation wavelength (λexc = 342 nm), I is the integrated area under the corrected emission curve and n is the refractive index of the solvent with the subscripts "s" representing sample and "r" representing reference. Fr is the absolute quantum yield of the external reference Rhodamine 6G (Fr = 95% in ethanol). [3] An integrating sphere (Hamamatsu, C9920-02) was employed for the photoluminescence quantum yield measurements of thin film samples. [4] The ΦPL of the films were then measured in air and N2 environment by purging the integrating sphere with N2 gas flow. The photophysical properties of the film samples were measured using an Edinburgh Instruments FS980 fluorimeter. Time-resolved PL measurements of the thin films were carried out using the multi-channel scaling (MCS) technique. The samples were excited at 379 nm by a pulsed laser diode (PicoQuant, LDH-D-C-375, FWHM < 40 ps, pulse energy = 58.5 ± 1.2 pJ, peak power = 1.5 ± 0.3 W, laser spot diameter = 0.4 ± 0.1 mm, power density = 11.6 ± 3.7 mW/cm 2 ) and were kept in a vacuum of < 8 × 10 −4 mbar. The singlet and triplet state energies were determined from the onset values of the prompt fluorescence and phosphorescence spectra at 77 K. The singlet-triplet energy gap (∆EST) was estimated from the difference in energy of the prompt fluorescence and phosphorescence spectra. Phosphorescence spectra were measured from 1 ms after photoexcitation with an iCCD exposure time of 8.5 ms.
Prompt fluorescence spectra were measured from 1 ns after photoexcitation with an iCCD exposure time of 100 ns. The films were excited by a femtosecond laser emitting at 343 nm (Orpheus-N, model: SP-06-200-PP). Emission from the samples was focused onto a spectrograph (Chromex imaging, 250is spectrograph) and detected on a sensitive S-6 gated iCCD camera (Stanford Computer Optics, 4Picos) having sub-nanosecond resolution.

Fitting of time-resolved luminescence measurements:
Time-resolved PL measurements were fitted to a sum of exponentials decay model, with chi-squared (χ 2 ) values between 1 and 2, using the EI FLS980 Each component of the decay is assigned with a weight, (wi), which is the contribution of the emission from each component to the total emission.
The average lifetime was then calculated using the following expressions: [5] 1. Two exponential decay model: where A1 and A2 are the preexponential-factors of each component.

2.
Three exponential decay model: with weights defined as ) = where A1, A2 and A3 are the preexponential-factors of each component.

OLED Fabrication and Characterization:
The OLED devices were fabricated in a bottom-emitting structure via thermal evaporation in a high vacuum at a base pressure of <5×10 -7 mbar. A pre-patterned glass substrate coated with indium doped tin oxide (ITO) was cleaned sequentially by ultrasonication in acetone, and isopropanol for 15 minutes. The temperature of ultrasonication bath was set at 60-70 o C. The cleaned substrate was exposed to oxygen plasma for 3 min to remove all dust and organics on the ITO surface and to increase the work function of ITO anode for better hole injection from the anode to organic layer. The substrate was loaded in the thermal evaporator.
Organic layers were deposited at a rate of 0.3-1.0 Å/s, monitored using a quartz crystal.
The electron injection layer, LiF, was deposited at a rate of 0.05 Å/s, while the Al cathode was deposited initially with a rate of 0.5 Å/s to obtain 10 nm thickness and after that the rate of Al cathode was increased to 3 Å/s. Two custom-made shadow masks were used to define the area of the evaporations. The organic layers and LiF were evaporated with a same shadow mask but Al was evaporated with the other mask.
The active area of the OLED was 2 mm 2 , determined by the spatial overlap of the anode and cathode electrodes. All the devices were encapsulated with glass lids and UV epoxy resin inside a N2 filled globe box. The luminance-current-voltage characteristics were measured in an ambient environment using a Keithley 2400 source meter and a homemade photodiode circuit connected to a Keithley 2000 multimeter for the voltage reading. The external quantum efficiency was calculated assuming Lambertian emission pattern for the OLEDs. The electroluminescence spectra were recorded by an Andor DV420-BV CCD spectrometer.

Determination of emitter dipole orientation by angle-resolved PL measurement:
Dipole orientation of emitter molecules was determined by angle-resolved PL measurements of thin films doped with each emitter. [6] The doping concertation of the films are same as the ones used for the OLEDs, i.e., 2 wt%, while the thicknesses of these films are around 50 nm. To quantify this, an anisotropy factor (a) was used, which is defined by the ratio of emitted power by vertical dipoles to total emitted power by all dipoles. [7] We note that for perfectly horizontal dipole orientation (i.e. parallel to the S-8 substrate surface), a = 0, for isotropic orientation, a = 1/3, and for perfectly vertical orientation, a = 1. The details of our set-up and calculation can be found in reference. [8] We obtained the optical constants and thickness of each organic layer using a variable Figure S1. Refractive index spectra of each organic layer fabricated on glass substrate.

Calculation of out-coupling efficiency of OLEDs:
The out-coupling simulation of the OLEDs is based on emission dipole as forced damped harmonic oscillator and S-9 embedded in thin film stacks. [9] The details of the calculation can be found in reference. [8] In the optical calculation, it was assumed that emitter dipole is localized at the mCP and TmPyPB interface. This is reasonable because the hole conduction is dominant in the host, mCP. The actual dipole position within the emission layer was not determined.
Errors of 10% in maximum for the 3TPA-DiKTa and around 30% for 3DPA-DiKTa were estimated (See Figure S2). Functional Theory (DFT) within Gaussian 16 [10] as well as the second order algebraic diagrammatic construction Spin-Component Scaling (ADC (2)-SCS) method using the Turbomole/7.5 package. For the DFT calculation, the ground state was optimized with PBE0 [11] functional and the 6-31G(d,p) basis set, [12] and excited state calculations have been performed using Time-Dependent DFT within the Tamm-Dancoff approximation S-10 (TDA-DFT) [13] with the same functional and basis set as for the ground state geometry optimization in gas phase. The molecular orbitals were visualized with Gaussview 5.0 software. For the ADC(2) calculation, the ground states was optimized with ADC (2)-SCS method and cc-pVDZ basis set in gas phase based on the geometry calculated by DFT. [14] Vertical excited states were performed on the ground state optimized structure using ADC(2)-SCS method. Different density plots were used to visualize change in electronic density between the ground and excited state and were visualized using the VESTA package. [15] S-11 Literature Study Figure S3. Molecular structures discussed in the introduction. S-12

X-Ray structure analysis
X-ray diffraction data for 3TPA-DiKTa and 3DPA-DiKTa were collected at 173 K using a Rigaku MM-007HF High Brilliance RA generator/confocal optics with XtaLAB P100 or P200 diffractometer [Cu Kα radiation (λ = 1.54187 Å)]. Intensity data were collected using either both ω and φ steps or solely ω steps, accumulating area detector images spanning at least a hemisphere of reciprocal space. Data for both compounds were collected using CrystalClear [16] and processed (including correction for Lorentz, polarization and absorption) using CrysAlisPro. [17] Structures were solved by dual-space methods (SHELXT) [18] and refined by full-matrix least-squares against F 2 (SHELXL-2018/3) [19] . Non-hydrogen atoms were refined anisotropically, and hydrogen atoms were refined using a riding model. All crystals of 3TPA-DiKTa showed very weak diffraction at higher angles, even with long exposures, often showing no diffraction above 1.10 Å. This likely arises from a combination of regions of diffuse solvent in the structure as well as the extent of disorder present. Despite the weak diffraction from this compound, the structure could still be unambiguously determined. Both structures showed some disorder in peripheral phenyl rings, this was extensive in 3TPA-DiKTa, and extended to one of the phenylene bridges. Disorder modelling included restraints to distances, angles and thermal motion as needed, and several of the disordered peripheral phenyl rings were constrained to an idealised geometry. Thermal ellipsoids in the DiKTa core of 3TPA-DiKTa suggested that the core might be somewhat disordered as well, however this could not be successfully modelled. Both structures showed regions of void space containing diffuse electron density (3TPA-DiKTa: 218 Å 3 , 3DPA-DiKTa: 283 Å 3 ) and the SQUEEZE [20] routine implemented in PLATON [21] was used to remove the contribution to the diffraction pattern of the unordered electron density in the void spaces. Despite treatment with S-13 SQUEEZE, the structure of 3TPA-DiKTa showed higher than anticipated values of R1 and wR2, likely due to the extent of the disorder and inability to successfully model disorder into the DiKTa core of the molecule. All calculations except SQUEEZE were performed using the Olex2 interface. [22] Selected crystallographic data are presented in Table S1. Deposition numbers 2183420 and 2183421 contain the supplementary crystallographic data for this paper. These data are provided free of charge by the joint  Calculations Figure S4. Theoretical calculations for DiKTa. HOMO and LUMO orbitals calculated in the gas phase at the PBE0/6-31G(d,p) level and difference density plots of S1, S2, T1 and T2 excited states calculated in the gas phase at the SCS-ADC(2)/cc-pVDZ level.     Where DCT is the distance between the hole and electron density barycentre.
Br3DiKTa were synthesised in four steps as shown in Scheme S1, including Ullmann coupling, bromination, hydrolysis and Friedel-Craft acylation reaction. The detailed procedure was described in our previously reported protocol. [23] Scheme S2. Synthetic route for 3TPA-DiKTa and 3DPA-DiKTa.

S-27
Optoelectronic characterization      Table S5. Photoluminescence quantum yield screening in different host matrices.

PLQY b
Host a mCP mCBP DPEPO

3TPA-DiKTa
a Thin films were prepared by spin-coating with 2 wt% in each host; b FPL values were determined using an integrating sphere (λexc = 305 nm or 340 nm); degassing was done by N2 purge (value given inside parentheses in the presence of O2). FPL values are within an error limit of ± 2%.     For a TADF system, the main exciton loss channels are either singlet or triplet nonradiative transition processes. Owing to high performance, the singlet nonradiative transition process ( nr S ) can be ignored, therefore the exciton loss can be attributed to the triplet nonradiative transition process ( nr T ). The kinetic parameters were calculated according to the following equations and summarized in Table S7. [26] Where the Φp and Φd are the prompt fluorescent and delayed fluorescent quantum efficiency; kp is the rate constant of prompt fluorescence; kd is the rate constant of delayed fluorescence; kr S is the radiative decay rate constant of S1; knr T is the non-radiative decay rate constant of T1; kISC is the intersystem crossing rate constant; kRISC is the reverse intersystem crossing rate constant.     We also fabricated devices A and C from 3TPA-DiKTa by using TCTA (as a HTL) OLEDs performances are shown in Figure S34 and summarized in Table S8. Devices A and C reached lower luminance and low current density with poorer device performance than the optimized device structure (device B) ( Figure S34 and Table S8).    [34] a Emitter structures are shown in Figure S32.