Ladder-like energy-relaying exciplex enables 100% internal quantum efficiency of white TADF-based diodes in a single emissive layer

Development of white organic light-emitting diodes based on purely thermally activated delayed fluorescence with a single-emissive-layer configuration has been a formidable challenge. Here, we report the rational design of a donor-acceptor energy-relaying exciplex and its utility in fabricating single-emissive-layer, thermally activated delayed fluorescence-based white organic light-emitting diodes that exhibit 100% internal quantum efficiency, 108.2 lm W−1 power efficiency, and 32.7% external quantum efficiency. This strategy enables thin-film fabrication of an 8 cm × 8 cm thermally activated delayed fluorescence white organic light-emitting diodes (10 inch2) prototype with 82.7 lm W−1 power efficiency and 25.0% external quantum efficiency. Introduction of a phosphine oxide-based acceptor with a steric group to the exciplex limits donor-acceptor triplet coupling, providing dual levels of high-lying and low-lying triplet energy. Transient spectroscopic characterizations confirm that a ladder-like energy relaying occurs from the high-lying triplet level of the exciplex to a blue emitter, then to the low-lying triplet level of the phosphine oxide acceptor, and ultimately to the yellow emitter. Our results demonstrate the broad applicability of energy relaying in multicomponent systems for exciton harvesting, providing opportunities for the development of third-generation white organic light-emitting diode light sources.

Supplementary Fig. 2 DFT simulations of molecular structures of mCP, pDPBITPO and DpPBITPO and associated contours and energy levels of frontier molecular orbitals and spin-density distributions of T1 states (TSDD). All molecules show overlapped HOMOs and LUMOs, indicating their unipolar characteristics. The relatively centralized LUMOs of PO acceptors on the phenyls linked with DPPO moieties reflect the electron-withdrawing effect of the P=O group on electron injection to the molecules. The HOMO energy gaps between mCP and PO acceptors are ~0.8 eV, comparable to their LUMO gaps of ~1.0 eV. The HOMO-LUMO gap of mCP is ~4.7 eV, similar to ~4.5 eV of PO acceptors. Comparable charge trap depths and HOMO-LUMO gaps of these donor and acceptor molecules can facilitate electronic coupling at singlet states. The triplet configuration of mCP with a high triplet energy level of 3.4 eV is similar to its ground state. In contrast, the phenylbenzimidazole groups in pDPBITPO and DpPBITPO are coplanar at their triplet states. Such structural relaxation induces remarkably lower triplet energy levels. The large triplet energy gaps prevent triplet electronic coupling between mCP and PO acceptors. Supplementary Fig. 3 Electrical properties of pDPBITPO, DpPBITPO, mCP:pDPBITPO and mCP:DpPBITPO. a, Cyclic voltammograms (CV) of pDPBITPO and DpPBITPO, measured at room temperature in a nitrogen atmosphere with 100 mV s -1 scanning rate. b, IV characteristics of pDPBITPO and DpPBITPO-based, electron-only devices with a configuration of ITO|LiF (1 nm)|PO (100 nm)|LiF (1 nm)|Al (100 nm). c, IV characteristics of mCP:pDPBITPO and mCP:DpPBITPO-based, single-carrier transporting devices with configurations of ITO|MoO3 (6 nm)|Exciplex (100 nm)|MoO3 (6 nm)|Al (100 nm) for hole-only and ITO|LiF (1 nm)|Exciplex (100 nm)|LiF (1 nm)|Al (100 nm) for electron-only, respectively. According to onset voltages of the redox peaks in CV curves, the HOMO and LUMO energy levels of pDPBITPO and DpPBITPO are -6.5 and -3.1 eV, respectively, for effective electron injection and hole blocking (Table S1). By fitting IV curves of the nominal electron-only devices with a space charge-limited current (SCLC) model, electron mobilities (e) of pDPBITPO and DpPBITPO are estimated at the level of 10 -4 cm 2 V -1 s -1 , making them competent as electron-transporting materials (Table S1). In contrast to their PO acceptors, e of mCP:pDPBITPO and mCP:DpPBITPO is roughly halved, helping confine electrons in emissive layers for recombination (Table S2). Meanwhile, mCP:DpPBITPO reveals a more balanced carrier transporting capacity with a hole mobility (h) of ~3 x 10 -5 cm 2 V -1 s -1 , which is one order of magnitude larger than that of mCP:pDPBITPO.
Supplementary Fig. 4 PL spectra of dichloromethane solutions of mCP and PO acceptors (a, pDPBITPO and b, DpPBITPO) with different ratios. The solution concentration of PO acceptors is 0.1 mol L -1 . The mCP solution (0.1 mol L -1 ) is gradually added to the PO acceptor solutions. Increasing mCP concentration generates a new emission peak around 450 nm and decreases the emission intensity of PO acceptors. Compared to the emission of mCP and PO acceptors, new emission bands correspond to exciplex emission, with a large red-shift of ~70 nm. Therefore, charge transfer between the mCP donor and PO acceptors induces the formation of mCP:DpPBITPO and mCP:pDPBITPO exciplexes. Furthermore, For all ratios, the emission from acceptors can be observed, reflecting limited donor-acceptor interactions.
Supplementary Fig. 5 PL spectra of vacuum-evaporated mCP:DpPBITPO films with different donor/acceptor molar ratios. Even at a low DpPBITPO ratio of 0.01, new emission peaks are generated at wavelengths beyond 400 nm. Upon increasing the DpPBITPO ratio to 0.1, deep-blue peaks around 400 nm become predominant, accompanied by weakened emissions of mCP at 350 nm. The fine structure of newly generated peaks is due to diverse donor-acceptor interactions at low DpPBITPO concentrations. Further increasing the mCP/DpPBITPO ratio to 1:1 gives rise to smooth, broad, structureless emissions centered at 413 nm, accompanied by the disappearance of mCP peaks. The emission bathochromic shift and broadening upon increasing the concentration of DpPBITPO indicate the gradual formation of the mCP:DpPBITPO exciplex.
Supplementary Fig. 7 Electroluminescence characterizations of TADF WOLEDs with emissive layers of mCP:DpPBITPO:35% DMAC-DPS:x% 4CzTPNBu (x = 0.2, 0.3, 0.4, 0.5 and 1.0). a, Doping concentration-dependent EL variation of as-fabricated devices. b, Variation tendency of CIE coordinates and correlated color temperature (CCT) with increasing doping concentrations. Values of standard illuminants D65, D50, and A are marked for comparison. c, Voltage-J-luminance curves of the devices. d, Luminance vs. efficiency correlation of the devices. Upon increasing the doping concentration, the blue component intensity gradually decreased and disappeared at x = 1.0, due to enhanced energy transfer from DMAC-DPS to 4CzTPNBu. The gradually changed EL spectra also reflect that exciton allocation in the emissive layers depends mainly on energy transfer rather than direct charge/exciton capture. It shows that with increasing concentration from 0.2% to 0.5%, the CIE coordinates of the devices moved along the Planckian locus in the chromaticity diagram, corresponding to emission color change from pure white to warm white (Fig. 3b). CIE coordinates and CCT values of these WOLEDs with x = 0.2, 0.3 and 0.5 are (0.31, 0.35) and 6582 K, (0.35, 0.42) and 5136 K, and (0.44, 0.47) and 3474 K, which are close to the Illuminants D65 ((0.31, 0.33)/6499 K), D50 ((0.35, 0.36)/5003 K) and A ((0.45, 0.41)/2856 K), corresponding to average daylight, midday sunlight, and incandescent light, respectively. Significantly, all these WOLEDs display state-of-the-art performance, including turn-on voltages less than 3.0 V, maximum ηPE and ηEQE values beyond 70 lm W -1 and 25%, and reduced roll-offs (Supplementary Table 3). Based on the same single-EML structure, mCP:DpPBITPO enables pure white and warm white OLEDs with high-quality white light, large luminance, high ηPE, and the IQE of ~100%, which are prerequisites for daily lighting applications.
Supplementary Fig. 8 Chromatic stability of WOLEDs with emissive layers of mCP:PO acceptor: 35% DMAC-DPS:0.5% 4CzTPNBu. EL spectra and CIE coordinates of WOLEDs based on (a) mCP:pDPBITPO and (b) mCP:DpPBITPO in the luminance range of 1000 to 10000 nits. With increasing voltage, the ratio variations of blue and yellow components are almost negligible, rendering unchanged CIE coordinates. This can be attributed to stable and balanced exciton allocation for blue and yellow emissions at high exciton concentrations. Therefore, optimal, stepwise energy transfer in the emissive layers is responsible for the chromatic stability of these devices in indoor and outdoor lighting applications.
Supplementary Fig. 9 Electroluminescence characterizations of WOLEDs with emissive layers of mCP:acceptor:35% DMAC-DPS:0.5% 4CzTPNBu. a, EL spectra of the devices. b, Voltage-J-luminance curves of the devices. The EL spectra of mCP:TPBI-based W3 overlapped with mCP: pDPBITPO-based W1, but the blue component in the EL spectra of mCP:BPhen-based W4 was weaker, indicating more exciton allocation to 4CzTPNBu. The electron mobility of Bphen and TPBI is higher than that of pDPBITPO and DpPBITPO. Therefore, the current densities of these WOLEDs are directly proportional to the mobility of the acceptors. Nevertheless, turn-on voltages of W1 and W2 were much lower than those of W3 and W4, due to suppressed exciton quenching in W1 and W2. Since the triplet states of pDPBITPO, DpPBITPO, TPBI, and Bphen can serve as intermediate energy levels for triplet energy transfer to 4CzTPNBu, the triplet states of PO acceptors, embedded by the peripheral DPPO groups, are protected from collisional quenching. This superiority of the PO acceptors is crucial for alleviating triplet quenching during the stepwise energy transfer process, giving rise to the dramatically improved device efficiencies of W1 and W2 (Fig. 3d).
Supplementary Fig. 12 Electro-stability of DMAC-DPS and PO acceptors. a, Measured oxidation cyclic voltammogram of DMAC-DPS in dichloromethane (10 mg mL -1 ). b, Reduction cyclic voltammograms of DMAC-DPS, pDPBITPO and DpPBITPO measured in tetrahydrofuran (10 mg mL -1 ). Curves were measured in a nitrogen atmosphere with a scan rate of 100 mV s -1 . Redox peaks of DMAC-DPS shifted in large ranges during 20 cycles. DMAC-DPS is fragile during the electrochemical reaction, which induces decomposition and generates oligomer from fragment radicals. In contrast, reduction curves of PO acceptors remain almost unchanged during 20 cycles, reflecting the high electro-stability of PO acceptors during the electron gain-loss process. where Ir is the relative PL intensity, I is the measured PL intensity, I0 is the initial PL intensity, APO is the 365-nm absorbance of phosphine oxide acceptor, and A is the 365-nm absorbance of the material. UV exposure induces a sharp decrease in the PL intensity of DMAC-DPS film. After 720 min exposure, PL of DMAC-DPS film nearly completely vanished, with a Ir/I0 ratio of as small as 38%. In contrast, after 720-min exposure, PL intensity of DpPBITPO film keeps 63% of I0, and the Ir/I0 ratio of 4CzTPNBu (87%) is the biggest. Furthermore, for mCP:DpPBITPO:35%DMAC-DPS:0.5%4CzTPNBu film, before 30-min exposure, its Ir/I0 ratio is similar to that of neat DpPBITPO film. However, from 30 to 720 min, its Ir/I0 variation becomes similar to that of DMAC-DPS. It means before 30-min exposure, most of the UV light was absorbed by the matrix. After 30 min, the matrix absorption saturates, and DMAC-DPS becomes involved in absorption, inducing sharply decreased Ir/I0. Luminance vs. efficiency correlation of the devices. EL spectrum includes two peaks from 2CzPN and 4CzTPNBu, respectively. The turn-on voltage of this device is 3.15 V at 1 cd m -2 . The maximum efficiencies reached 68.9 cd A -1 , 68.7 lm W -1 and 24.2%, respectively.
Supplementary Fig. 15 Electroluminescence characterizations of 10-inch 2 TADF WOLEDs with mCP:PO acceptor:35%DMAC-DPS:0.5%4CzTPNBu. EL spectra of (a) mCP:pDPBITPO-and (b) mCP:DpPBITPO-based devices at 4-9 V. c, Voltage-J-luminance curves. d, Luminance vs. efficiency correlation of the devices. The 10-inch 2 devices were fabricated through vacuum evaporation and directly measured at ambient conditions without encapsulation. Turn-on voltages of the two 10-inch 2 devices were slightly lower than those of the 0.09-cm 2 basic unit. Power efficiencies of the 10-inch 2 devices were much higher than achieved by previously reported TADF WOLEDs (for benchmarking, see Supplementary Table  3). The T1 energy level of DpPBITPO lies between the excited energy levels of DMAC-DPS and the S1 energy level of TBRb. Since the radiative state of TBRb is S1 state, the energy transfer to TBRb is mainly from the S1 state of DMAC-DPS rather than from the T1 state of DpPBITPO. Nevertheless, W5 achieved good EL performance with a maximum ηEQE of 13.7% and a small roll-off. On the contrary, the radiative states of the phosphors are T1 states; therefore, the T1 state of DpPBITPO facilitates triplet energy transfer from FIrpic to PO-01. This case is similar to mCP:DpPBITPO-based white TADF systems. W6 achieved maximum efficiencies of 88.3 lm W -1 and 32.4%, comparable to the best phosphorescent WOLEDs reported so far 3 . It is noted that the blue emissive component of W6 was weaker than that of W2. For mCP:DpPBITPO:35%FIrpic:0.5%PO-01, the T1 state of FIrpic was simultaneously involved in blue emission and blue-to-yellow energy transfer, leading to direct competition between blue phosphorescence and blue-to-yellow energy transfer. On the contrary, for mCP:DpPBITPO:35%DMAC-DPS:0.5%4CzTPNBu, blue emission is from the S1 state of DMAC-DPS rather than the T1 state. Incorporation of the T1 state for DpPBITPO in triplet energy transfer optimizes triplet-exciton allocation to 4CzTPNBu, but hardly affects DMAC-DPS in singlet-exciton utilization. Thus, the exciplex-based energy relaying mechanism is universal for optimization of exciton allocation in single-layer, white-emitting systems.
Supplementary Fig. 17 Electroluminescence characterizations of blue TADF diodes with emissive layers of mCP:PO acceptor:30%DMAC-DPS. a, Energy transfer process in the emissive layer. Singlet electronic coupling between mCP and PO acceptors generates S1 energy levels for the exciplexes, while mCP and PO acceptors contribute to high-lying and low-lying triplet energy levels, respectively. b, Voltage-J-luminance curves and EL spectra (inset) of as-fabricated devices. c, Luminance vs. efficiency correlation of the devices. Excitons are first formed on exciplex molecules and then transferred to DMAC-DPS dopants. However, low-lying triplet energy levels lead to triplet exciton leakage from DMAC-DPS and nonradiative deactivation. Therefore, the blue devices displayed low EL performance, e.g., a luminance within 1000 cd m -2 and a maximum ηEQE < 8%, far below conventional devices (> 10000 cd m -2 and ~20%) 20 .
Supplementary Fig. 18 Electroluminescence characterizations of yellow TADF diodes with emissive layers of mCP:PO acceptor:5% 4CzTPNBu. a, Energy transfer process in the emissive layer. b, Voltage-J-luminance curves and EL spectra (inset) of the as-fabricated devices. c, Luminance vs. efficiency correlation of the devices. Since the large energy gaps between the exciplexes and 4CzTPNBu do not support efficient energy transfer, the low-lying T1 levels of the exciplexes dominate energy transfer to 4CztPNBu. However, the population of the T1 states is limited upon charge recombination. Nevertheless, deeper frontier molecular orbitals of 4CNTPNBu facilitate direct charge-carrier capture and recombination, alleviating inefficient host-dopant energy transfer. The maximum ηEQE of yellow devices was ~15%, two-thirds of the reported values for 4CzTPNBu-based devices 34 . The blue (see Supplementary Fig. 12) or the yellow device alone cannot achieve 100% exciton harvesting.
Supplementary Fig. 19 Comparison of electroluminescent properties for a singly-doped yellow TADF diode, a doubly doped white TADF diode, and a doubly doped yellow TADF diode. a, Energy transfer process of TADF diodes with an emissive layer of [mCP:PO acceptor]:DMAC-DPS:4CzTPNBu. b,c, Voltage-J-luminance curves and EL spectra (inset) (b), and luminance vs. efficiency correlation (c) of yellow and white devices with emissive layers of mCP:DpPBITPO:5%4CzTPNBu (yellow), mCP:DpPBITPO:30%DMAC-DPS:0.5%4CzTPNBu (white) and mCP:DpPBITPO:30%DMAC-DPS:5%4CzTPNBu (yellow). The 4CzTPNBu doping concentration of the yellow device with mCP:DpPBITPO:30%DMAC-DPS:5%4CzTPNBu increased 10-fold to achieve pure yellow emission. The energy transfer process of this doubly doped yellow device is the same as the white device. In contrast to the singly doped yellow devices, the incorporation of DMAC-DPS in the emissive layer improved maximum efficiencies of the doubly doped yellow devices by ~40% (75.9 lm W -1 and 24.5%). In doubly doped systems, triplet excitons can leak from DMAC-DPS to the low-lying T1 level of the hosts before being utilized by 4CzTPNBu for radiation. At the same time, the T1 level of DMAC-DPS serves an intermediate energy level between the exciplexes and 4CzTPNBu to improve energy transfer.
Supplementary Fig. 20 Electroluminescence characterizations of single-and dual-emissive-layer WOLEDs based on mCP:DpPBITPO. a, Device structure and energy level diagram of the WOLEDs with dual emissive layers of mCP:DpPBITPO:5%4CzTPNBu|mCP:DpPBITPO:30%DMAC-DPS. b, Comparison of the energy transfer processes of single-and dual-emissive-layer devices. c, Voltage-J-luminance curves and EL spectra (inset). d, Luminance vs. efficiency correlation. To experimentally verify the role of the T1 levels of the hosts in the electroluminescence process, a control device containing dual emissive layers was fabricated, with DMAC-DPS and 4CzTPNBu spatially separated by mCP:DpPBITPO doping. Since the short-distance Dexter mechanism dominates the triplet energy transfer, the distance between participants in the energy transfer process of DMAC-DPS → PO acceptor → 4CzTPNBu must be short. In the dual-emissive-layer device, the short distance between DMAC-DPS-PO acceptors-4CzTPNBu cannot be simultaneously satisfied to establish a continuous energy transfer process. As a result, the blue emission of the dual-EML device is more intense than that of the single-EML devices, due to inefficient energy transfer of the former. The dual-EML device suffered from triplet leakage in the blue emissive layer and exciton nonradiative transition in the yellow emissive layer, reducing maximum efficiencies to 24.6 cd A -1 , 21.4 lm W -1 and 9.3%. These values are less than half of the maximum efficiencies achieved by single-EML devices. k Zero-field electron and hole mobility evaluated with I-V characteristics of single-carrier transporting devices according to field-dependent space charge-limited current (SCLC) model.