High efficiency and low roll-off green OLEDs with simple structure by utilizing thermally activated delayed fluorescence material as the universal host

3.1 Fl-OLEDs

The device structure of Fl-OLEDs based on C545T is as follows: ITO/MoO₃ (3 nm)/mCP (25 nm)/DMAC-DPS: x% C545T (15 nm)/PO-T2T (45 nm)/LiF (1 nm)/Al, where x=1.0, 1.5, and 2.0, respectively. We can see that the OLEDs are a simple three-layer structure. The mCP (m-bis(N-carbazolyl)benzene) and a phosphine-oxide-based material of PO-T2T act as the hole transport layer (HTL) and electron transport layer (ETL), respectively. Concentrations of 1.0%, 1.5%, and 2.0% with C545T doped into DMAC-DPS are used for the EML. Amazingly, all the devices obtained an excellent device performance of low turn-on voltage (< 2.5 V) and high maximum luminance (~17,000 cd/m²), which are shown in Figure S1. Under a concentration of 1.5%, the Fl-OLEDs display the best performance; the maximum current efficiency, power efficiency, and EQE are 32.2 cd/A, 26.3 lm/W, and 9.35%, respectively. Even at a luminance of 1000 cd/m², the efficiencies are also as high as 31.8 cd/A, 25.0 lm/W, and 9.26%, respectively. This is almost the highest efficiency based on C545T at the luminance of 1000 cd/m² to date. Figure 1 shows the EL performance of the Fl-OLEDs. Moreover, at a high luminance of 10,000 cd/m², the EQE is also higher than the 5% fluorescent EQE upper limit and reaches 6.9%. We also fabricated the Fl-OLEDs with the same device structure except for the host, which adopted the traditional fluorescent host of Alq and DPVBi. But the corresponding maximum EQEs are merely 2.68% and 3.36%, respectively. Figure S2 and Table S1 display the EL performance of Fl-OLEDs with Alq and DPVBi as the host, respectively. The low efficiency of Fl-OLEDs with a traditional host is due to the waste of triplet excitons, which cannot up-convert to a singlet excited state because of the large ΔE_S-T. With the TADF material as the host, the triplet excitons could up-convert to its singlet excited state and transfer to the dopant through the Förster energy transfer process, which increases the amount of dopant singlet excitons and improves the device efficiency further. The relevant EL spectra of DMAC-DPS as the host are shown in Figure 1B. Under the concentrations of 1.5% and 2.0%, the devices exhibit intrinsic C545T emission with a peak of 513 nm. But at the low concentration of 1.0%, a weak emission peak at ~460 nm appears, which is inferred to the host emission of DMAC-DPS. The appearance of host emission peak at 1.0% concentration and disappearance at 1.5% and 2.0% concentrations indicate the energy transfer process from DMAC-DPS to C545T.

Figure 1:

The EL performance of Fl-OLEDs with different C545T concentrations. (A) The current efficiency-luminance-EQE curves. (B) The EL spectra.

Table 1 summarizes the EL performance and roll-off ratio of the Fl-OLEDs in this paper with representative devices based on C545T. Except for the optimal concentration of 1.5%, the devices also harvest considerable performance at 1.0% and 2.0% with the simple three-layer structure. The maximum EQE reached 7.73% and 6.83%, respectively. Even at a luminance of 1000 cd/m², the efficiency was almost the same, with 7.62% and 6.82%, respectively. Lee et al. utilized TADF exciplex host to sensitize C545T and achieved Fl-OLEDs with a maximum EQE of 14.5% [15], but the roll-off ratio from maximum EQE to EQE at 1000 cd/m² (EQE_Max.~EQE₁₀₀₀) was as high as 40.1%. Kido et al. achieved ~7% EQE through the optimizing charge balance, but the EQE_Max.~EQE₁₀₀₀ still reached ~10% [16], whereas the EQE_Max.~EQE₁₀₀₀ of all the Fl-OLEDs in this work are only ~1%. The reason for the low roll-off will be discussed in detail in the section on Ph-OLEDs.

In order to explain the high efficiency of Fl-OLEDs with the simple device structure, we draw the PL spectrum of DMAC-DPS and absorption spectrum of C545T, which are shown in Figure 2. The rather large spectral overlap between the PL spectrum of the host and the absorption spectrum of the dopant indicates that the energy transfer from the singlet excited state of DMAC-DPS to the singlet excited state of C545T by Förster mechanism can take place efficiently.

Figure 2:

The PL spectrum of DMAC-DPS and absorption spectrum of C545T.

To clarify the energy transfer process in Fl-OLEDs further, we fabricated the DMAC-DPS: x% C545T films with several concentrations. Figure 3 shows the PL spectra and transient PL decay curves under different concentrations. From Figure 3A, we can see that the emission of the host also appears under low concentration, which is consistent with the EL spectra. And the emission of the DMAC-DPS host reduces gradually and disappears with the increase in doped concentrations, verifying again the energy transfer process between DMAC-DPS and C545T. Figure 3B exhibits the transient PL decay curves of DMAC-DPS: x% C545T films with various doped concentrations. Monitoring the PL emission peak of 510 nm, to our surprise, all of the curves exhibit double-exponential decay and present two lifetimes of prompt and delay through the suitable fitting. In general, as a traditional fluorescent material, C545T only has a prompt intrinsic transient lifetime of several nanoseconds. But in this work, C545T displays a delayed lifetime of several hundred nanoseconds, even reaching orders of microseconds. DMAC-DPS has been confirmed as a highly efficient TADF material, which exhibited two lifetimes of prompt and delay due to the efficient RISC with a small ΔE_S-T between its singlet and triplet excited states [9]. Therefore, we suggest that the longer transient lifetime of C545T in this paper is derived from the RISC and energy transfer of DMAC-DPS triplet excitons. The detailed prompt and delayed lifetimes of C545T with different doped concentrations have been added to Figure 3B. We can see that the delayed lifetimes have a large range when C545T with different concentrations is doped into DMAC-DPS. The delayed lifetime decreases gradually as the concentration of C545T increases. Under optical excitation, the triplet excitons of DMAC-DPS are produced by the efficient intersystem crossing (ISC) process, and the delayed component of C545T results from the energy transfer of up-converted DMAC-DPS triplet excitons. As the concentration increases, the ISC efficiency of DMAC-DPS reduces and the RISC efficiency also reduces, which result in a decrease in delayed lifetime. So under electric excitation, the 25% singlet excitons produced on the DMAC-DPS host would transfer to the singlet state of C545T by the Förster process and the 75% triplet excitons would up-convert to a singlet excited state first through efficient RISC, then the new singlet excitons formed through RISC would also transfer to the singlet excited state of C545T. Finally, a highly efficient C545T emission could be achieved. Hence, based on the discussion above, the almost consistent transient decay behavior of C545T and DMAC-DPS reported previous evidence that the emission of C545T is derived from the energy transfer of DMAC-DPS singlet excitons and up-converted triplet excitons.

Figure 3:

The PL spectra and transient PL decay curves with the film of DMAC-DPS: x% C545T under different concentrations. (A) The PL spectra. (B) The transient PL decay curves.

Next, we discuss the energy transfer efficiency of the host-guest system though an equation. The TADF material of DMAC-DPS is the energy transfer host and C545T is the energy transfer guest, so the energy transfer efficiency (∅_ET) of singlet and triplet could be expressed as follows [17]:

∅ET=ketkr+knr+ket=ket1/τ+ket

In the equation, k_et is the energy transfer rate from host to dopant; k_r is the radiative transition rate of host; k_nr is the nonradiative transition rate of host; and τ is the emission lifetime of the DMAC-DPS host. As we can see from the equation, in order to achieve a high ∅_ET, a higher k_et and longer τ should be required. From Figure 2, we know that there is a rather large overlap between the PL spectrum of the DMAC-DPS host and the absorption spectrum of C545T, which have demonstrated a highly efficient energy transfer rate from the DMAC-DPS host to C545T (that is high k_et). On the other hand, the TADF host of DMAC-DPS has a long delayed fluorescent lifetime (τ) about 3.0 μs at 300 K by transient PL decay measurement [9], [18], which also illustrates a high RISC efficiency. Therefore, a high energy transfer efficiency (∅_ET) could be realized in this system.

3.2 Ph-OLEDs

Next, we adopted the same host of DMAC-DPS to fabricate Ph-OLEDs with common green phosphorescent material of Ir(ppy)₃ as the emitter. The device structure is as follows: ITO/MoO₃ (3 nm)/mCP (25 nm)/DMAC-DPS: x% Ir(ppy)₃ (15 nm)/PO-T2T (45 nm)/LiF (1 nm)/Al, where x=2.0, 4.0, and 6.0, respectively. To our surprise, all the Ph-OLEDs also achieved good efficiency under the three concentrations of low turn-on voltage (< 3.0 V) and high maximum luminance (> 20,000 cd/m²), which can be seen in Figure S3. Figure 4 and Table 2 show the EL performance of Ph-OLEDs in this paper. At the optimal concentration of 4.0%, the device obtains a maximum current efficiency, power efficiency, and EQE of 64.3 cd/A, 62.4 lm/W, and 18.5%, respectively. Different from the traditional host-guest Ph-OLEDs, the energy transfer process with TADF material as the host is mainly through the long-range Förster energy transfer rather than short-range Dexter energy transfer or direct carrier trapping recombination by the dopant [12], [18]. The triplet excitons produced on the TADF host are more easily up-converted to singlet excited state and achieve the energy transfer by the Förster process owing to the small ΔE_S-T of the TADF host. On the other hand, with the traditional host-guest Ph-OLEDs, the triplet excitons produced on the host cannot up-convert to singlet excited state because of the large ΔE_S-T. So the triplet excitons transfer to the dopant via a Dexter energy transfer process, which happened between the triplet excited state of the host and the dopant. The relatively low EQE of 15.3% with the concentration of 2.0% could be attributed to the incomplete energy transfer process, which is proved from the weak emission of the host DMAC-DPS shown in Figure 4B. Besides, even more surprising is the efficiency roll-off of the Ph-OLEDs. For example, under the representative luminance of 1000 cd/m², the current efficiency and EQE of the Ph-OLEDs with 4% concentration still reached up to 62.0 cd/A and 17.8%, respectively, and the roll-off ratio is just 3.78%. In Table 2, we summarize the EL performance and roll-off ratio of the Ph-OLEDs. Meanwhile, the roll-off ratio with representative devices based on Ir(ppy)₃ or Ir(ppy)₂(acac) are also given. We can see that all the Ph-OLEDs in this paper obtain a very low roll-off with a large range from low luminance to high luminance of 10,000 cd/m². The EQE_Max.~EQE₁₀₀₀ is only ~3.5%, while the ratio is ~25% from maximum EQE to EQE at 10,000 cd/m². We also sought some representative phosphorescent devices as a reference; the EQE_Max.~EQE₁₀₀₀ is 27.9% in Ref. [19] and EQE_Max.~EQE_10,000 are 21.8% and 52.1% in Refs. [20], [21], respectively. Although a low roll-off was also achieved in Ref. [20], an exciton block layer of CF-X was adopted, which added to the complexity of the device. Nevertheless, we attain highly efficient and low roll-off Ph-OLEDs with a simple three-layer structure in this paper.

Figure 4:

The EL performance of Ph-OLEDs with different Ir(ppy)₃ concentrations. (A) The current efficiency-luminance-EQE curves. (B) The EL spectra.

In order to explore the reason of low roll-off, we fabricated single carrier devices, whose structure is as follows:

• Hole-only device: ITO/MoO₃ (3 nm)/NPB (15 nm)/DMAC-DPS (40 nm)/NPB (15 nm)/Al

• Electron-only device: ITO/TPBi (15 nm)/DMAC-DPS (40 nm)/TPBi (15 nm)/LiF (1 nm)/Al

Figure 5A plots the current density-voltage curves of the two devices. As we see, both of the hole-only and electron-only devices have a high current density with the increased voltage, which indicates the bipolar host of DMAC-DPS. Hence, the electron and hole could transport and recombine evenly in EML; meanwhile, the carrier recombination zone will become wide to the whole EML due to the bipolar host. Except for the EML, carrier injection is another important factor in OLEDs. We also draw an energy-level schematic diagram of Ph-OLEDs, which is shown in Figure 5B. Because of the suitable highest occupied molecular orbital energy level of HTL/EML and lowest unoccupied molecular orbital energy level of EML/ETL, the hole and electron injected from the electrode could be transported barrier-free to the EML. The barrier-free injection, balanced transport, and recombination eliminate the accumulation and trapping of carrier, which suppress the triplet-polaron annihilation (TPA) [22]. The very wide carrier recombination zone dilutes the exciton concentration, which efficiently weakens the effect of TTA [22], [23]. Therefore, the barrier-free injection, balanced transport and recombination, and wide carrier recombination zone are responsible for the low efficiency roll-off at high luminance in our Ph-OLEDs. In addition to the above mentioned, the triplet exciton up-conversion of the DMAC-DPS host due to the small ΔE_S-T may be another reason for the low efficiency roll-off because this process decreases the triplet exciton concentration and achieves a rapid energy transfer from DMAC-DPS to Ir(ppy)₃ by Förster mechanism, which could also contribute to the reduction of the TPA [24]. We believe that the relative low roll-off of the Fl-OLEDs results from the same reason mentioned above.

Figure 5:

(A) The current density-voltage curves of single carrier devices. (B) The energy level schematic diagram of Ph-OLEDs.