Jaber
Saghaei
a,
Manikandan
Koodalingam
a,
Paul L.
Burn
*a,
Ian R.
Gentle
a,
Almantas
Pivrikas
b and
Paul E.
Shaw
a
aCentre for Organic Photonics & Electronics (COPE), School of Chemistry and Molecular Biosciences, The University of Queensland, Brisbane, Queensland 4072, Australia. E-mail: p.burn2@uq.edu.au
bSchool of Engineering and Information Technology, Murdoch University, Perth, Western Australia 6150, Australia
First published on 3rd December 2021
In this study, efficient solution-processed organic light-emitting diodes (OLEDs) based on a light-emitting layer composed of a blue emissive phosphorescent dendrimer and exciplex host were investigated. Employing poly(styrene sulfonic acid) doped poly(ethylenedioxythiophene):poly(styrenesulfonate) (m-PEDOT:PSS) as the hole injection layer resulted in devices with external quantum efficiencies (EQEs) over 20% for luminances of up to 100 cd m−2. A feature of these devices was a relatively slow electroluminescence turn-on for the initial voltage scan, which was absent when m-PEDOT:PSS was replaced with molybdenum oxide as the hole injection layer. The initial turn-on rate was found to decrease with increasing PSSH content, with the results suggesting that proton migration from (m-)PEDOT:PSS is the cause of the change in the device performance. The overall device performance was found to be scan dependent, with sequential voltage scans leading to a decrease in EQE, which is ascribed at least in part to electromer formation. We also demonstrate solution-processed OLEDs with the same exciplex host and bis[2-(4,6-difluorophenyl)pyridinato-C2,N](picolinato)iridium(III) (FIrpic) as the emitter, that have maximum EQE and power efficiencies of 17.0% and 25.3 lm W−1, respectively.
While the creation of exciplex host blends for green and red guest emitters is relatively straightforward, it is more difficult to create donor–acceptor combinations that form suitable hosts for blue emissive materials,14 and in particular, solution-processed blue phosphorescent emitters. The emission energy of the exciplex host is governed by the ionisation potential of the donor and the electron affinity of the acceptor and hence there needs to be a sufficiently large energy offset such that neither component of the exciplex host nor the formed exciplex quench the triplet state of the blue emitter. In the context of solution processed blue-phosphorescent emitters, there have been a number of reports of exciplex hosts composed of 4,4′,4′′-tris(carbazol-9-yl)triphenylamine (TCTA) (donor) (and derivatives) and phosphine oxide-based acceptors with bis[2-(4,6-difluorophenyl)pyridinato-C2,N](picolinato)iridium (III) (FIrpic) as the emitter.15–19 In fact, most reports describing solution-processed bulk exciplex hosts for blue emissive guests include the use of phosphine oxide derivatives as the acceptor in the blend, which is due to their relatively small electron affinities and high triplet energies.15 The best performing of these thus far had a device architecture of ITO/PEDOT:PSS/Ph–O–TCTA:PhPO:FIrpic/TPBI/Cs2CO3/Al [PEDOT:PSS = poly(ethylenedioxythiophene):poly(styrenesulfonate), Ph–O–TCTA = a tris–TCTA based macromolecule, PhPO = tris[4-(diphenylphosphoryl)phenyl]benzene, TPBI = 2,2′,2′′-[1,3,5-benzinetriyl]-tris[1-phenyl-1-H-benzimidazole], Cs2CO3 = caesium carbonate, Al = aluminium] giving an external quantum efficiency of 16.5%, and power (PE) and current (CE) efficiencies of 20.6 lm W−1 and 33.6 cd A−1, respectively.19 However, one of the issues of developing solution-processed blue phosphorescent-exciplex host films is that many of the available materials have been developed for evaporative processing and do not always form good quality films from solution. For example, solution processed films of FIrpic generally contain aggregates.20–23 In early work on solution processed phosphorescent OLEDs it was shown that light-emitting dendrimers could form good quality films when blended with host materials that had poor solubility and propensity to aggregate when deposited from solution.24 In this paper we show that a blue phosphorescent light-emitting dendrimer, Ir(DTri)3, blended with a TCTA:POPH exciplex host (see Fig. 1 for structures) can be solution processed to form films of good quality. The choice of a TCTA:POPH blend as the exciplex host was based on the fact that it had been successfully used for FIrpic and that Ir(DTri)3 was also a sky blue emitter. Furthermore, we report and discuss the photophysical, electronic and device properties of the blend emissive films and compare the device performance with films using FIrpic as the emissive material.
Fig. 1 Structures of TCTA, POPH, Ir(DTri)3, FIrpic, and the electron transport/hole blocking BP4mPy used in the bilayer devices. |
The PL emission of the 20 wt% Ir(DTri)3 in a 1:1 molar ratio TCTA:POPH film is also shown in Fig. 2. At first glance it appears that at an excitation wavelength of 330 nm only emission from Ir(DTri)3 is observed. Indeed, Fig. S2a (ESI†) shows that the PL spectrum is independent of the excitation wavelength (250–350 nm). However, plotting the PL spectra on a logarithmic scale, Fig. S2b (ESI†), it can be seen that even up to a concentration of 30 wt% Ir(DTri)3 there is still a small component of host emission.
The PL quantum yields (PLQYs) of films with different dendrimer concentrations are summarised in Fig. 3. The exciplex host films (with no guest) had a low PLQY of 3 ± 1%. The PLQY was found to reach a maximum of 67 ± 6% for a dendrimer concentration of 20 wt% with the neat dendrimer film having a PLQY of 16 ± 2%. The solution PLQY of the dendrimer is 73 ± 8%, which was close to that of the 20 wt% dendrimer blend film, indicating that the dendrimers are not strongly interacting with each other in the exciplex host at that concentration.28–30
In the final part of the study on the PL properties we measured the PL decays of the exciplex TCTA:POPH blend and Ir(DTri)3 films, and films containing different concentrations of Ir(DTri)3 (Fig. 3b, with the lifetime fitting parameters for the exciplex host, neat dendrimer and 20 wt% dendrimer blend summarised in Table S1, ESI†). The PL decay of Ir(DTri)3 in solution is also shown for comparison. In solution, Ir(DTri)3 had a mono-exponential decay with a lifetime of 4.1 μs, which is consistent with an absence of interchromophore interactions. The 1:1 molar ratio TCTA:POPH blend film was found to have a multi-exponential decay with a lifetime longer than that of individual films of TCTA and POPH, which again confirms exciplex emission from the host film. The neat Ir(DTri)3 also had a multi-exponential decay, with the largest component having a lifetime shorter than that measured in solution. These characteristics are consistent with there being intermolecular interchromophore interactions that lead to the quenching of the luminescence in the neat Ir(DTri)3 film. The film PL lifetimes were found to decrease with increasing Ir(DTri)3 concentration, with the neat film of Ir(DTri)3 having the shortest. These results are consistent with the PLQY measurements and an increase in intermolecular interchromophore interactions between the emissive Ir(DTri)3 macromolecules with increasing guest concentration in the mixed exciplex host.
Prior to fabricating and testing OLEDs containing the emissive layer materials, we measured their hole and electron mobilities using metal–insulator–semiconductor charge extraction by linearly increasing voltage (MIS-CELIV) and photo-MIS-CELIV, respectively (Fig. S3 shows the transients, ESI†). The neat film of Ir(DTri)3 had a hole mobility of (9.6 ± 0.7) × 10−5 cm2 V−1 s−1 and the exciplex blend film had hole and electron mobilities of (1.1 ± 0.3) × 10−6 and (8.5 ± 1.3) × 10−7 cm2 V−1 s−1, respectively. The similarity of the electron and hole mobilities of the TCTA:POPH blend film demonstrates the ambipolar charge transport characteristics and its suitability as a host. The electron [(8.0 ± 1.2) × 10−8 cm2 V−1 s−1] and hole [(4.9 ± 0.8) × 10−7 cm2 V−1 s−1] mobilities of the emissive layer containing 20 wt% of Ir(DTri)3 were lower than those of the TCTA:POPH blend, which is consistent with the Ir(DTri)3 trapping the charges.31,32
Fig. 4 (a) J–V–L, (b) EQE–L, (c) PE–L and (d) CE–L of OLEDs with the structure of ITO/PEDOT:PSS/TCTA:POPH:Ir(DTri)3/BP4mPy/LiF/Al with different Ir(DTri)3 wt% and a TCTA:POPH molar ratio of 1:1. |
It has been reported that adding poly(4-styrenesulfonic acid) (PSSH) to PEDOT:PSS (m-PEDOT:PSS) can increase the work function from 5.04 eV to 5.59 eV and improve hole injection.35 We therefore used m-PEDOT:PSS in place of PEDOT:PSS in the devices with 20 wt% of Ir(DTri)3 in the emissive layer with the results shown in Fig. 5 and the OLED performance characteristics summarised in Table S2 (ESI†). Addition of the PSSH to PEDOT:PSS led to a dramatic difference in the EQE at low luminance for the first voltage scan, with the EQE reaching more than 20% at a luminance of up to around 20 cd m−2. This performance was repeatable on multiple pixels (see Table S2, ESI†). However, at luminances of ≥100 cd m−2 the EQE for the m-PEDOT:PSS containing OLED was similar to that of devices that used pristine PEDOT:PSS (cf. Fig. S4 and S5, ESI†). To determine whether the high initial EQE was dependent on the trap density arising from the emitter concentration we also prepared a device with a smaller Ir(DTri)3 component. However, a high initial EQE was still observed when the Ir(DTri)3 concentration in the emissive layer was reduced to 2 wt% (Fig. S4, ESI†). It should be noted that even though a high EQE was observed on the first scan, the PLQY values for all the emissive layers were less than 70%. To determine whether this phenomenon was due to the Ir(DTri)3 emitter being dendrimeric we also prepared devices with the same configuration but with 10 wt% of FIrpic as the emissive material (the known optimum guest ratio). We found a similar effect with the maximum EQE and PE reaching to 17.0% and 25.3 lm W−1, respectively, for the first scan, which is the highest reported thus far for solution processed exciplex OLEDs containing FIrpic (Fig. S5, ESI†). As with the Ir(DTri)3 containing devices, at higher luminance the performance of the OLEDs containing either PEDOT:PSS or m-PEDOT:PSS were similar (Fig. S5, ESI†).
Fig. 5 (a) J–V–L, (b) EQE–L, (c) PE–L and (d) CE–L of OLEDs with PEDOT:PSS or m-PEDOT:PSS as the hole injection layer and (1:1) TCTA:POPH:(20 wt%)Ir(DTri)3 as the emissive layer. |
At first sight the results of our OLEDs with the m-PEDOT:PSS seem to confirm the general improvement in performance previously reported. However, when we undertook several voltage scans, we found that the subsequent EQEs were lower (decreasing after the first scan before becoming relatively stable) than that measured on the first scan (note this effect was not highlighted in the published work).35 It is clear from the data in Fig. 6 that the first and subsequent voltage scans have different electrical and optical characteristics. A further observation we noted is that when the pixel was turned on and held at 4 V there was a delay in reaching uniform emission for devices composed of hole injection layers of either PEDOT:PSS or m-PEDOT:PSS (see Fig. 7), with the latter having a significantly slower turn-on. However, when the same pixel was turned off and then back on at 4.0 V there was no delay in illumination and the pixel had instant uniform emission.
Fig. 6 (a) J–V–L, (b) EQE–L (c) PE–L and (d) CE–L of an OLED with the structure ITO/m-PEDOT:PSS/(1:1)TCTA:POPH:(20 wt%)Ir(DTri)3/BP4mPy/LiF/Al for sequential voltage scans (S1–S6). |
Fig. 7 Photos of an OLED composed of m-PEDOT:PSS and Ir(DTri)3 blended into the exciplex host held at 4.0 V. |
The question therefore arises as to what are the origins of both the slow turn on and high EQEs for the first voltage scan. A high EQE in the first voltage scan could arise from one of the charge carriers filling traps in the emissive layer leading to a higher internal electric field that enhances the injection of the second carrier at low voltage. In subsequent scans, the charge traps would remain filled and thus the internal electric field would be reduced. Such a situation could be affected by the distribution of the components in the emissive layer. For example, vertical phase separation of the components could lead to a different trap and electric field profile compared to the case where the materials are evenly distributed throughout the film. To check whether the components in the emissive layer were uniformly distributed we undertook depth profiling X-ray Photoelectron Spectroscopy measurements on a glass/m-PEDOT:PSS/TCTA:POPH:(20 wt%)Ir(DTri)3 film stack (Fig. S6, ESI†). We found that based on the atomic percentages the ratio of the TCTA, POPH and Ir(DTri)3 was as expected from the ratio of the materials in the processing solution (Fig. S7a, ESI†). In addition, there was no evidence of vertical separation of any of the components (Fig. S7b, ESI†). This is in contrast to the reported accumulation of FIrpic at the PEDOT:PSS/EML interface36 and is evidence of the ability of the dendrimeric Ir(DTri)3 to provide the processability needed to form good quality uniform thin films.
To test whether the slow turn-on (see Fig. 7) arises from trapped charges we held a device after a positive voltage scan at a negative bias of −5 V for 10 minutes to attempt to depopulate the trap states and potentially restore the device to its original state. However, we found that there was no difference in the performance during the second positive voltage scan (Fig. S8, ESI†). In addition, the fact that the time required for a pixel to reach uniform emission on the first scan is on the order of seconds suggests that an alternative mechanism to charge trapping is giving rise to the effect, e.g., proton migration from the (m-)PEDOT:PSS into the emissive layer. To confirm that the PEDOT:PSS is primarily responsible for the effect we prepared equivalent (1:1) TCTA:POPH:(20 wt%)Ir(DTri)3 devices with molybdenum oxide (MoOx) as the hole injection layer. While the devices had lower overall performance (see Fig. S9, ESI†) there was no significant difference in the device characteristics between the first and subsequent voltage scans. Furthermore, the devices turned on instantaneously. Thus, differences in performance of the (m-)PEDOT:PSS devices between voltage scans arise from the conducting polymer layer itself.
To determine whether the amount of PSSH added to the PEDOT:PSS had an effect on the performance of the devices we varied the PSSH concentration. We found that independent of the amount of PSSH added the first voltage scan always gave a high EQE at low luminance, which was higher than that of the devices with pristine PEDOT:PSS (Fig. S10, ESI†). We also note that the delay in reaching uniform emission at an initial voltage of 4.0 V was dependent on the concentration of the PSSH in the m-PEDOT:PSS, with a greater amount of PSSH leading to a slower response. These results suggest that proton migration from (m-)PEDOT:PSS into the emissive layer is primarily responsible for the first scan performance. To see whether we could block proton transfer we followed a previous report where a poly(9-vinylcarbazole) (PVK) layer was deposited onto the PEDOT:PSS layer before depositing the emissive film from solution.37 The driving voltage was found to increase for the device containing the PVK layer, which indicated that it was present despite it being thin (< 15 nm). However, in our devices the PVK layer did not have a significant effect and a higher EQE was still observed for the first scan (Fig. S11, ESI†).
In addition to the electrical characteristics, we measured the electroluminescence (EL) spectra of the devices (Fig. 8). For the devices containing 20 wt% Ir(DTri)3 there was a small peak at 640 nm (note the log scale) in addition to the blue emission from the dendrimer. The peak at 640 nm was found to increase in intensity with multiple voltage scans. Previous reports of exciplex EL from TCTA:POPH blends have assigned the long wavelength peak to an electromer,16 and interestingly, in our work we did not observe the peak for the devices that had FIrpic as the emitter nor those with a MoOx hole injection layer (Fig. S5d and S9d, ESI†). We also observed the peak at 640 nm when the emissive layer only contained TCTA and Ir(DTri)3 (Fig. S12, ESI†), i.e., with no POPH, indicating the feature is not exciplex in origin and is consistent with a TCTA electromer.16 The formation of an electromer can provide an additional weakly radiative decay pathway, which can in part explain the observed reduction in device efficiency with increasing number of scans. The fact that the electromer emission was not observed for the devices that had MoOx as the hole injection layer suggests that the (m-)PEDOT:PSS plays a role in its formation.
Fig. 8 EL spectra of OLEDs on (a) PEDOT:PSS and (b) m-PEDOT:PSS for different voltage scans. S represents “scan” and the scan number is shown. |
Although it is difficult to elucidate the reason for the electromer formation, previous reports have shown that protons38,39 and even fragments of PSS36 can diffuse into the emissive layer. It is therefore not unreasonable to assume that these components could lead to the chemical changes of materials in the emissive layer that could lead to electromer formation or degradation of the materials. It is important to note that electromer formation has been previously reported to be irreversible and will in part impact device stability.16,40
The BCB solution in mesitylene (Sigma-Aldrich 98%) was diluted 1:10 by volume with mesitylene, and the solution was spin-coated at 5000 rpm in a glove box (O2 and H2O < 0.1 ppm) before annealing at 300 °C for 10 min. The substrates were then transfer to a vacuum evaporation chamber (MBRAUN and 10−6 mbar) for deposition of 50 nm MgF2. The emissive layers were spin-coated from 20 mg mL−1 solutions in distilled degassed dichloromethane at 2000 rpm and then annealed at 100 °C for 30 min to give films with a thickness of ≈150 nm. Finally, the samples were transferred to the vacuum evaporation chamber for deposition of 100 nm Al for the photo-MIS-CELIV experiment or 8 nm MoOx and 100 nm Ag for the MIS-CELIV measurement. A High Efficiency VIOLET LED emitter (LZ1-10UB00-01U7 purchased from OSRAM SYLVANIA Inc.) with 385–410 nm emission and 80 mW power was used for photo-MIS-CELIV experiments. The mobility was calculated from the time for maximum current extraction using:
(1) |
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1ma00729g |
This journal is © The Royal Society of Chemistry 2022 |