Promising operational stability of high-efficiency organic light-emitting diodes based on thermally activated delayed fluorescence

Organic light-emitting diodes (OLEDs) are attractive for next-generation displays and lighting applications because of their potential for high electroluminescence (EL) efficiency, flexibility and low-cost manufacture. Although phosphorescent emitters containing rare metals such as iridium or platinum produce devices with high EL efficiency, these metals are expensive and their blue emission remains unreliable for practical applications. Recently, a new route to high EL efficiency using materials that emit through thermally activated delayed fluorescence (TADF) was demonstrated. However, it is unclear whether devices that emit through TADF, which originates from the contributions of triplet excitons, are reliable. Here we demonstrate highly efficient, stable OLEDs that emit via TADF by controlling the position of the carrier recombination zone, resulting in projected lifetimes comparable to those of tris(2-phenylpyridinato)iridium(III)-based reference OLEDs. Our results indicate that TADF is intrinsically stable under electrical excitation and optimization of the surrounding materials will enhance device reliability.

In this work, to investigate the possibility of TADF-based OLEDs with high operational stability, we carefully designed a device architecture that included exciton-blocking layers (EBLs) at the interfaces of the emission layer (EML) and optimized the concentration of the emitter. Here we show that expanding the carrier recombination zone to enhance the electron carrier injection efficiency significantly affects the operational stability of the device. We realize lifetimes of more than 2,500 h at an initial luminance of 1,000 cd/m 2 in TADFbased OLEDs while maintaining a high g EQE that is comparable to those of OLEDs containing iridium complexes.

Results
Design of OLED configuration. To maximize the TADF efficiency, we prepared a guest-host system that escapes concentration quenching, similar to conventional fluorescence and phosphorescence based OLEDs. Also, because TADF is based on an up-conversion process of triplet excitons into a singlet state, improper protection of the triplet excitons induces Dexter energy transfer into the surrounding molecules during the lifetime of the guest's triplet state, which resulted in exciton quenching. It is therefore necessary to introduce host and carrier transport molecules with higher triplet energies than that of a guest TADF molecule.
We derived the theoretical maximum g EQE for the 10 wt.%-doped OLED using the following equation 8 :  where g r,S is the proportion of singlet excitons (25%) and g r,T is the proportion of triplet excitons (75%) produced under electrical excitation. From transient PL analysis (Fig. S2), the contributions from the prompt component (W F ) and the TADF component (W TADF ) were estimated to be 27% and 51%, respectively, resulting in a reverse ISC (RISC) efficiency (W RISC ) of 69%, which was estimated from the relation (1 2 W F ) W RISC 5 W TADF . The theoretical maximum g EQE is thus estimated to be 14.2%, assuming a light outcoupling efficiency of 20%. Therefore, near-complete carrier recombination and exciton confinement in the EML were realized in the 10 wt.%-doped device. Then, we changed the hole blocking layer (HBL) from T2T to CzTRZ and BAlq 2 to examine the effect of the recombination zone position on the device performance. The OLEDs with CzTRZ and BAlq 2 HBLs had inferior characteristics when compared with those of the devices with T2T HBLs (Fig. 2(b)). In particular, an emission originating from CzTRZ was observed ( Fig. 2(b), inset), indicating both the inefficient carrier confinement at the EML/EBL interface because of the shallow HOMO of the CzTRZ layer (26.1 eV) and the location of the carrier recombination zone near the EML/HBL interface.
Device operational stability. Figure 2(c) shows the normalized luminance of the 4CzIPN-based OLEDs as a function of operational time at an initial luminance L 0 of 1,000 cd/m 2 and Table 1 summarizes the OLED properties. We observed a significant dependence of the operational lifetime, defined as operation at 90% of the initial luminance (LT90), on the 4CzIPN doping concentration. At low doping concentrations, LT90 was only 40 and 65 h for the 3 and 6 wt.%-doped devices, respectively. In contrast, LT90 values of 190 and 253 h were observed for the 10 and 15 wt.%-doped devices, respectively. These results show that the 4CzIPN concentration strongly affects the operational stability of these devices.
To predict LT50 for the devices containing 10 and 15 wt.% 4CzIPN at L 0 5 1,000 cd/m 2 , we estimated an acceleration factor of 1.92 for each device from the lifetime measurements at L 0 5 2,000, 5,000 and 10,000 cd/m 2 using the following well-known equation 17 : where LT is the operational lifetime and n is an acceleration factor. Based on Eq. (2), LT50 is predicted to be 1,900 h for the 10 wt.%doped device and 2,800 h for the 15 wt.%-doped device, which is comparable to that predicted for a 6 wt.% Ir(ppy) 3 -doped device (4,500 h), as summarized in Table 1.

Discussion
There are several possible reasons for the enhancement of LT50 with an increase in the doping concentration. We envisage that the position of the carrier recombination zone strongly affects the operational stability of these devices. In fact, although the EL from only S 1 of 4CzIPN was observed at 1,000 cd/m 2 in all devices, another emission peak around the deep blue region was observed in the 3 and 6 wt.%-doped devices at 20,000 cd/m 2 , suggesting that the carrier recombination zone position changes in these devices, as shown in Fig. 3(a). Also, after the device begins to degrade (LT75 5 164 h), an additional emission signal was observed from the 3 wt.%-doped devices ( Fig. 3(b)). This signal was similar to that observed from the pristine device at 20,000 cd/m 2 , indicating that the carrier recombination zone moves during constant operation. However, no additional emission signal was observed from the 15 wt.%-doped devices (Fig. 3(c)), even after the device began to degrade (LT75 5 820 h).
These results indicate the absence of the carrier recombination zone at the EML/EBL interface in the 15 wt.%-doped devices.
To determine the effect of the doping concentration on the carrier transport properties, electron-only devices (EODs) and hole-only devices (HODs) were fabricated, as shown in the insets of Fig. 4(a) and (b), respectively. Although the 3 wt.%-doped EOD showed a  Fig. 4(a)). In contrast, the HODs showed almost no dependence on the driving voltage as the 4CzIPN doping concentration increased, as shown in Fig. 4(b). This clearly indicates that only the electron injection/transport efficiency is enhanced by an increase in the doping concentration. Therefore, an increase in the doping concentration enhances the efficiency of electron injection from T2T into the EML and subsequent transport in the EML. These results also suggest that the carrier recombination zone shifts from the EML/ETL interface into the bulk of the EML when the doping concentration is as high as 10 wt.%. The LUMO level of 4CzIPN (23.4 eV) is located considerably lower that of mCBP (22.4 eV), while their HOMO levels are similar (25.8 and 26.0 eV for 4CzIPN and mCBP, respectively). Therefore, the 4CzIPN molecules in an mCBP host act as strong electron trapping sites, so the recombination process mainly involves direct carrier injection, transport and recombination at the 4CzIPN molecules. A higher dopant concentration expands the exciton formation sites into the bulk of the EML, which produces highly reliable OLEDs 18,19 . In addition, reduction of the hole accumulation at the interface between the EML and the EBL is another possible reason for the enhancement of the device operational stability 20 . Also, because the undesired degradation products of carbazole derivatives have been identified after device degradation 21 , the reduction in the excited state formation on the mCBP host molecules by enhancement of the direct electron injection and transport from a T2T layer to 4CzIPN, e.g., by direct exciton formation at 4CzIPN, is another possible reason for the enhanced device stability.   Finally, we consider the possibility of further enhancement of the device reliability. In this study, we adopted conventional host and carrier transport materials and obtained comparable degradation lifetimes for 4CzIPN and Ir(ppy) 3 . This is an encouraging sign that the emitter itself is quite stable for redox and oxidation processes under electrical excitation. Thus, because Ir(ppy) 3 derivative-based OLEDs with optimum materials and device architectures have realized very long lifetimes of over 100,000 hrs. 22,23 , we can expect further improvements in the device reliability when TADF is combined with the best possible combination of surrounding materials and device architectures.
In summary, we clarified that the operational lifetime of 4CzIPNbased OLEDs depends strongly on the emitter concentration in the EML and demonstrated highly efficient TADF-based OLEDs (g EQE of 13.9 6 0.5% at 1,000 cd/m 2 ) with excellent operational stability, showing LT50 of 2,800 h at 1,000 cd/m 2 and of over 10,000 h at 500 cd/m 2 . We also found that the 4CzIPN molecules act as strong electron trapping sites in the mCBP EML, and that the position of the recombination zone strongly affects the operational lifetime of these devices. Our results confirm that TADF-based OLEDs show great potential for realization of both high efficiency and operational stability.

Methods
Sample preparation and characterization for photoluminescence. Samples for the optical measurements were fabricated by co-depositing host materials and 6 wt.% 4CzIPN with a thickness of 50 nm on a quartz substrate. The PL quantum efficiency (W PL ) was measured by an absolute PL quantum yield measurement system (C11347-01, Hamamatsu Photonics, Japan) under a nitrogen gas flow with excitation wavelengths of 337 or 275 nm. The low-temperature PL intensity and the emission lifetime were measured using a streak camera (C4334, Hamamatsu Photonics, Japan) and a cryostat (Iwatani Industrial Gases Co., Japan) with a nitrogen gas laser (MNL200, Laser Technik, Germany) as the excitation light source under a pressure of about 3 Pa.
Sample preparation and characterization for electroluminescence. Green TADFbased OLEDs with an effective area of 1 mm 2 were fabricated on 110 nm-thick indium tin oxide (ITO)-coated glass substrates with a 2 mm stripe pattern. Deposition was performed under vacuum at pressures from 5 3 10 24 to 5 3 10 25 Pa. After fabrication, the devices were immediately encapsulated with glass lids using epoxy glue in nitrogen-filled glove boxes (O 2 , 0.1 ppm, H 2 O , 0.1 ppm). A commercial calcium oxide desiccant (Dynic Co., Japan) was included in the encapsulated package. The devices were exposed once to nitrogen gas after the formation of the organic layers because a metal mask was included to define the cathode area. The J-V-luminance characteristics were evaluated using a Keithley 2400 source meter and an absolute g EQE measurement system (C9920-12, Hamamatsu Photonics, Japan). The operational lifetime was measured using a luminance meter (CS-2000, Konica Minolta, Japan) at a constant DC current at room temperature.