Operational lifetimes of organic light-emitting diodes dominated by Förster resonance energy transfer

Organic light-emitting diodes are a key technology for next-generation information displays because of their low power consumption and potentially long operational lifetimes. Although devices with internal quantum efficiencies of approximately 100% have been achieved using phosphorescent or thermally activated delayed fluorescent emitters, a systematic understanding of materials suitable for operationally stable devices is lacking. Here we demonstrate that the operational stability of phosphorescent devices is nearly proportional to the Förster resonance energy transfer rate from the host to the emitter when thermally activated delayed fluorescence molecules are used as the hosts. We find that a small molecular size is a requirement for thermally activated delayed fluorescence molecules employed as phosphorescent hosts; in contrast, an extremely small energy gap between the singlet and triplet excited states, which is essential for an efficient thermally activated delayed fluorescent emitter, is unnecessary in the phosphorescent host.


2
Supplementary Section 1 Fig. S1 (a) Transient PL curve of the 6 wt%-host-material:Ad-Cz [S1] films. (b) Fluorescence and phosphorescence spectra of 6 wt%-host-material:Ad-Cz films. Red and blue lines represent fluorescence spectra at 77 K and phosphorescence spectra at 77 K, respectively. The black dotted line represents the supporting line used to determine S1 and T1 energy. (c) HOMO/LUMO molecular orbital distribution obtained from the Gaussian09 program with B3LYP/6-31G(d,p) basis sets.
As can be seen in Fig. S1(a), a clear delayed fluorescence is observed from 2c-Ph, whereas Cz-Ph-TRZ does not show a clear delayed fluorescence. That's because, the EST of Cz-Ph-TRZ is relatively large as compared to other TADF materials such as 2a, 2b, 2c and 2c-Ph [see Fig. S1(b) and Table 1 in the manuscript]. The orbital overlap between HOMO and LUMO in Cz-Ph-TRZ is also relatively large as compared to other TADF materials as shown in Fig. S1(c).  The optimal dopant concentration was carefully determined by fabricating PHOLEDs with different dopant concentrations as shown in Table S2. The table shows that the optimal dopant concentration in the PHOLEDs in the s-Czs host family was 3 wt%.
Therefore, in these experiments, we fabricated PHOLEDs with a dopant concentration of 3 wt%.
As the molecular weight strongly depends on the host material, the mol% of the dopant in the emitting layer also depends on the host material. This causes the observed difference in the average distance of the guest [S2]. However, Table S2 shows that differences in mol% have little influence on the operational stability.

Supplementary Section 3
Comparison of device characteristics between PHOLEDs and host-only devices. To evaluate the stability of TADF host on its own, we measured the stability of the hostonly device as shown in Fig. S3d. The configurations of host-only devices are almost the same as those of PHOLEDs, except for the existence of the dopant.
As the efficiency of host-only devices is much lower than that of PHOLEDs, the stability was compared by setting the current density to approximately 1 mA cm −2 , which corresponds to the current density in a PHOLED at approximately L0 = 1,000 cd m −2 .
Supplementary Section 4 6 Summary of PHOLED characteristics using PIC-TRZ2, DIC-TRZ and DIC-TRZ-Ph as the hosts.  The black dotted line represents the supporting line used to determine S1 and T1 energy. The J-V characteristics of PHOLEDs using the host family comprising ICz, in which the thickness of the EML is 35 nm and the thickness of the TPBi is 25 nm, are more dependent on the host than those of PHOLEDs using the host family comprising s-Czs (see Fig. S4-1a and Fig. 1c). This difference was caused by differences in the carrier transportability of the hosts. When we compare the device characteristics of these PHOLEDs, the operational lifetime of these PHOLEDs can be affected by differences in the carrier balance compared with the PHOLEDs using the host family comprising s-Czs. However, we used all the data for analysis as we needed as much data as possible to systematically understand the characteristics of host-dependent PHOLEDs.
The fluorescence and phosphorescence spectrum of DIC-TRZ and DIC-TRZ-Ph are also shown in Fig. S4-2. 8

Supplementary Section 5
The accuracy of the lifetime values (LT50s). Almost all PHOLEDs are fabricated at least twice on different days, and the accuracy of the LT50 is checked. As shown in Fig. S5, the variation in estimated LT50 is quite small even in the most long-lived PHOLED using 2c as a host. Thus, the accuracy of the estimated LT50 is extremely high.

Supplementary Section 6
Summary of PHOLED characteristics using Cz-Ph-TRZ, CBP as the hosts. Photophysical properties of the host that are related to kFRET such as PL and PL were measured in a 6-wt%-TADF-host:Ad-Cz film. Although both prompt and delayed 11 components were observed, we used the values related to the prompt components in the analysis. This is because the fluorescence process from S1 to S0 in a TADF host determines kFRET.

Supplementary Section 8
Calculation of the maximum molecular radius (Rmax) of each molecule.  The distance between the centre of mass and the farthest hydrogen atom was calculated in the optimized structure, and the van der Waals' radius of the hydrogen atom was added to it. The coordinates of the optimized structure were obtained from the Gaussian09 program with B3LYP/6-31G(d,p) basis sets. One example of the results we calculated is shown in Fig. S8; the distance between the centre of mass and the farthest hydrogen was calculated in 2a. As the original point of the coordinates corresponds to the centre of mass in this calculation, the distance between the centre of mass and the farthest hydrogen atom can be calculated as shown in Fig. S8.  The data used for analysing the lifetime with respect to kFRET is also shown ( Table S10).