A “Si‐Locked” Phosphine Oxide Host with Suppressed Structural Relaxation for Highly Efficient Deep‐Blue TADF Diodes

Abstract

Thermally activated delayed fluorescence (TADF) diodes with deep-blue emission and a state-of-the-art external quantum efficiency are realized by using a novel phosphine oxide host with a high triplet energy of 2.98 eV and a “Si-locked” planar rigid structure. These results manifest the effectiveness of our “insulating lock” strategy in suppressing structural-relaxation induced nonradiative energy losses. Organic semiconductors have great potential and advantages for constructing various energy converters, such as organic light-emitting diodes (OLEDs) and organic solar cells (OSCs).1 In these devices the electrical process optimization as well as the utilization ratio of the input energy are key factors for practical applications.2 However, it is well known that some molecules lose their excited energy during energy-transfer processes and intermolecular interactions.3 Therefore, it is of vital importance to clarify the dominant nonradiative energy-loss channels and thereby develop feasible strategies for achieving high-energy conversion efficiencies.4 Recently, thermally activated delayed fluorescent (TADF) dyes have been employed in OLEDs to realize extremely high electroluminescent (EL) efficiencies, for instance external quantum efficiencies (EQEs) of more than 10% have been achieved by harvesting triplet excitons through reverse intersystem crossing (RISC).5 For most TADF devices the emitters are dispersed in a host matrix to restrain emitter–emitter interaction-induced triplet exciton quenching.6 In this case, because of the similarity between the TADF host and the dopant (both being pure organic compounds), the host molecules, which act as the main emitting layer (EML), are deeply involved in the energy transfer and host–host and host–dopant interactions, making them the predominant energy loss factor .It is noteworthy that the progress of blue TADF diodes lags behind other color congeners, as the photon energy of deep-blue light (as high as 2.70 eV for light of 460 nm) puts an extremely high demand not only on the first triplet energy (T1) of the host material (>2.8 eV) but also on the reduction of the nonradiative energy loss.7 Consequently, up to now, there have only been a few reports on deep-blue TADF diodes with Commission Internationale de l'Eclairage (CIE) coordinates of y < 0.2 and x + y < 0.35, as well as high EQE values above 15%.8 To achieve a high T1 energy, a widely used approach is the interruption of the π-conjugation by incorporating saturated atoms, such as C, Si, and P.9, 10 However, successive σ bonds increase the molecular flexibility, facilitating the structural variation during energy transfer and intermolecular collisions.11 Actually, structural relaxation is one of the main nonradiative de-excitation channels. For a doping system, a flexible host molecule generally worsens the nonradiative energy loss through: i) more remarkable vibrational relaxation (VR) of its singlet and triplet excited states (Scheme 1a);12 ii) deactivation of dopant-localized excitons by structural variation and relaxation of surrounding host molecules during and after collision (Scheme 1b).13 In this sense, structural relaxation of the host material is one of the main channels for nonradiative energy loss. Deeply understanding this effect on the EL performance and developing feasible molecular design strategies has become imperative. In this contribution, saturated Si atoms were used as insulating linkage and the flexible chromophore diphenylether (DPE) in 1-(diphenylphosphinoyl)-2-phenoxy-benzene (DPESPO)14 was cyclized by a so-called “Si lock” diphenylsilanylene (DPSi) to form 4-diphenylphosphoryl-10,10-diphenyl-dibenzoxasiline (DBOSSPO) (Figure 1a), whose chromophore turned into a planar and rigid dibenzoxasiline (DBOS). As expected, in contrast to DPESPO, DBOSSPO revealed a suppressed structural relaxation with a negligible Stokes shift of 159 cm−1 and a T1 energy as high as 2.98 eV. This was accompanied by a reduced relaxation effect on the blue TADF dye bis[4-(9,9-dimethyl-9,10-dihydroacridine)phenyl]sulfone (DMAC-DPS) resulting in a higher emission color purity, better photoluminescence quantum yield (PLQY), and longer lifetime. Consequently, although their electrical properties are comparable, DBOSSPO endowed this DMAC-DPS based device with the deepest blue emission reported so far with CIE coordinates of (0.16, 0.15) and a state-of-the-art EQE of 19%, which was more than three times that of DPESPO-based control devices. This work successfully demonstrates an effective strategy named “insulating lock” for developing high-energy-gap rigid host materials for deep-blue TADF diodes. DBOSSPO was readily prepared through a two-step procedure of silanization and phosphorylation with a moderate total yield of around 25% (Figure 1a and Supporting Information). Its chemical structure was fully characterized using NMR spectroscopy, mass spectrometry, and elemental analysis. The planar configuration of the DBOS ring was further confirmed by single-crystal X-ray diffraction (XRD) analysis of DBOSSPO, in which the dihedral angle between the two phenyl groups of the DBOS ring was only 4.3o (Figure 1b). In accord with the improved thermal properties of DBOSSPO (Figure S1, Supporting Information, and Table 1), a moderate P = O…H intermolecular interaction was observed with a distance of 2.666 Å between the diphenylphosphine oxide (DPPO) and DBOS groups between adjacent molecules, which also suggests the complementing properties of these two groups during charge transfer. According to the Frank–Condon principle, electronic transitions occur before molecular configuration variations. In this case, the difference between the absorption and emission spectra directly reflects the variation between the molecular configuration in the ground state and that of the excited states. It is interesting that in dilute solutions (10−6 m in CH2Cl2), all of the absorption peaks for DPESPO are included in the spectrum of DBOSSPO (Figure 2a and Table 1). Their combined peaks between 250 and 300 nm were identical in wavelength and profile and can be attributed to the n→π and π→π transitions of their DPE segments, whereas the intensity of the peaks around 225 nm was directly proportional to the number of their phenyl groups in the DPSi and DPPO groups. However, the cyclic structure of DBOSSPO generated an additional absorption peak at 309 nm due to the facilitated n→ π transition of the coplanar DBOS ring. Consequently, estimated from the absorption edge, the singlet excited energy (S1) of DBOSSPO was dramatically reduced to 3.86 eV, which is 0.27 eV lower than that of DPESPO. On the contrary, the difference between the fluorescence (FL) peak wavelength of DPESPO and DBOSSPO in dilute solutions was only 5 nm. In this case, the excited-state structural relaxation of DBOSSPO is effectively suppressed by its rigid structure, giving rise to a negligible Stokes shift of only 159 cm−1 and a narrow FL emission with a full width at half maximum (FWHM) as small as 29 nm. However, because of its flexible structure, the Stokes shift of DPESPO is doubled, and its serious excited-state relaxation further renders a long tail in its FL spectrum from 350 to 500 nm. Looking at the triplet characteristics we found the opposite, namely, despite of its planar structure, the 0→0 transition of DBOSSPO shifted hypsochromic to 416 nm, corresponding to a T1 value of 2.98 eV, which is 0.14 eV higher than that of DPESPO. Density functional theory (DFT) calculations showed that the T1 states of DPESPO and DBOSSPO were localized on their DPE and DBOS units, respectively, suggesting a correspondence between the chromophore rigidity and the T1 energy for these two molecules (Figure S3, Supporting Information). As a result, in contrast to that of DPESPO (1.29 eV), the singlet–triplet splitting (ΔEST) of DBOSSPO was dramatically decreased to 0.88 eV, reflecting its reduced energy loss during ISC and triplet VR processes. It is inspiring that two coplanar phenyls locked by DPSi can successfully introduce Si and O linkages and interrupt the π conjugation, thereby enhancing the molecular rigidity without T1 energy reduction and successfully manifesting the effectiveness of the “insulating lock” strategy for constructing high-energy-gap rigid molecules. The molecular configurations in the ground and excited states were further investigated with DFT and time-dependent DFT (TDDFT) simulations (Figure 2b). DPESPO showed a remarkable structural change from the ground state (S0) to S1, whereby the dihedral angle between the two phenyl groups of its DPE unit was reduced from 98.4° to 84.3°, which further decreased to 57.1° in the T1 state. In contrast, the molecular configuration of DBOSSPO was stable whereby the dihedral angles between the two phenyl groups of its DBOS unit for the S0 and S1 states were very similar; although, this angle was slightly enlarged to 8.7° in the T1 state, accompanied by a decrease in the excited energy. Therefore, we can conclude that the excited-state relaxation of DBOSSPO was effectively suppressed by its “Si locked” structure. After doping with DMAC-DPS (10 wt%) the main contribution in the absorption spectra of the vacuum-evaporated thin films came from the DPESPO and DBOSSPO hosts (Figure 2c and Table 1). The large spectral overlap between the DPESPO and DBOSSPO emissions and the absorption of DMAC-DPS in the range from 300 to 350 nm supports the efficient host-dopant energy transfer, giving rise to pure DMAC-DPS-originated emissions. Nevertheless, compared to DPESPO, DBOSSPO had a remarkably more narrow profile with a smaller FWHM of 71 nm. According to the time-decay curves, the lifetime of the DMAC-DPS-originated emissions in the DBOSSPO matrix at 7.6 μs was significantly longer than that in the DPESPO matrix (6.2 μs), implying a reduced quenching of dopant-localized excitons in the former (inset of Figure 2c and Table 1). Furthermore, the DMAC-DPS-doped DBOSSPO film achieved a high PLQY approaching 90%, which was three times that of the DPESPO-hosted film. The superiority of the DBOSSPO-based film in terms of emission color purity, lifetime, and PLQY rationally validates the limited relaxation effect of DBOSSPO on the excited DMAC-DPS. This relaxation effect enhanced the triplet exciton stability, which in turn facilitated the RISC process, as indicated by a RISC rate constant that was one order of magnitude larger than that of the DPESPO-based film (see Supporting Information). As expected, the “Si-locked” molecular configuration of DBOSSPO effectively suppressed the relaxation-induced nonradiative energy losses on the host itself as well as the dopant. However, contrary to DPESPO the electrically inert DPSi in DBOSSPO hardly changed its electrical properties despite the involvement of DBOS in both the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) of DBOSSPO (Figure S3, Supporting Information). DPESPO and DBOSSPO have similar HOMO and LUMO energy levels that are around -6.6 and -2.5 eV, respectively, and both show predominantly electron-transporting characteristics with hole and electron mobilities on the order of 10−7 and 10−6 cm2 V−1 s−1, respectively (Figure S4, Supporting Information, and Table 1). In consequence, DPESPO and DBOSSPO provided a favorable platform to selectively investigate the correlation between the structural relaxation of the host materials and the EL performance of their blue TADF diodes. The DMAC-DPS-based blue TADF devices of DPESPO and DBOSSPO were fabricated with the same configuration to be able to investigate the structural relaxation effect of the hosts on their device performance (Figure 3a). A conventional flexible Si-based host p-bis(triphenylsilyly)benzene (UGH2) and the most popular blue TADF host bis[2-(diphenylphosphino) phenyl]ether oxide (DPEPO) were also fabricated to serve as the control devices. The doping concentration was optimized to 10 wt% (Figure 3b,c). According to the frontier molecular orbital (FMO) energy levels of the employed materials, the direct charge and exciton capture on DMAC-DPS should be dominant in the EL process (Figure 3a). Nevertheless, as is the case in the majority of EMLs, the host materials would also be involved in carrier transportation. In this case, the electron-predominant features of DBOSSPO and DPESPO would make the charge-carrier recombination zones shift to the mCP layers (Figure S4, Supporting Information). All of the devices revealed deep-blue emissions with maxima at 460 nm, indicating the confinement of the radiative excitons on the DMAC-DPS (inset in Figure 3d and Table 2). The hyperchromic shift of the EL emission can be attributed to microcavity effects. However, the same results were found for their thin-film PL spectra (Figure 2c), the intensity of the green component in the EL spectrum of DBOSSPO-based devices was weaker than that of DPESPO-based analogues. Consequently, using DBOSSPO as a host successfully realized the deepest blue TADF emission from DMAC-DPS reported so far with outstanding color purity and dramatically improved CIE coordinates of (0.16, 0.15). The I–V characteristics of these devices were in accord with the charge mobility of DPESPO, UGH2, DPEPO, and DBOSSPO (Figure 3d). UGH2-based devices generally suffer from high driving voltages due to the weak electrical performance of UGH2; whereas, DPEPO endowed its devices with the second lowest driving voltages. As expected, DBOSSPO supported its devices with dramatically reduced driving voltages of 3.0 V at the onset and 4.6 and 7.0 V at 100 and 1000 cd m−2, which were 0.6, 2.9, and 4.2 V lower than those of DPESPO-based devices, respectively (Table 2). Furthermore, at the same voltages, the luminance of the DBOSSPO-based device was about five-fold that of the DPESPO-based device. Taking into account the similar I–V characteristics of the DPESPO and DBOSSPO-based devices, the significantly reduced driving voltages and increased luminance of the DBOSSPO-based devices reflected the more effective exciton recombination and radiative transition.10 DBOSSPO-based devices showed state-of-the-art EL efficiencies with maxima as high as 28.5 cd A−1 for the current efficiency (CE), a power efficiency (PE) of 29.8 lm W−1, and an EQE of 19.0%, which were about 20% higher than those of DPEPO-based analogues and among the highest values reported so far for deep-blue TADF devices (Figure 3e and Table S2, Supporting Information). Significantly, the efficiency roll-off of the DBOSSPO-based devices was successfully restricted to values as low as 9 and 31% at 100 and 1000 cd m−2 for the EQE, which are the lowest to date. In contrast, the maximum efficiencies of the DPESPO-based devices were less than one third of those of the DBOSSPO-based devices, accompanied with much more serious roll-offs. The situation for the UGH2-based devices was similar to that although their maximum EQE was better at 11.5%, which was still only half of that of the DBOSSPO-based devices. The flexible structure of UGH2 also made the serious roll-offs comparable to those of DPESPO-based analogues. The reduction in EQE at high J of the DBOSSPO-based devices was basically induced by singlet–triplet annihilation (STA) and triplet–triplet annihilation (TTA) (Figure S5, Supporting Information).15 However, DPESPO rendered the much more serious decrease in efficiency of all devices, indicating the significant additional exciton quenching. The device efficiencies of DPESPO and DBOSSPO were actually in direct proportion to the radiative transition probabilities of their DMAC-DPS-doped films. Furthermore, the efficiency roll-offs of the four devices were consistent with the rigidity of their host materials, which restrained the interaction-induced exciton quenching at high J. In this sense, the superiority of the DBOSSPO-based devices in driving voltages and efficiencies is related to the effectively suppressed nonradiative energy loss by their “Si-locked” rigid host. In summary, an effective strategy named “insulating lock” for constructing deep-blue TADF host materials with high excited energy and structural rigidity was successfully demonstrated by DBOSSPO, whose configuration was fixed with a DPSi bridge serving as a “Si lock”. In comparison to its flexible analogue DPESPO, DBOSSPO is superior in structural relaxation suppression, endowing its DMAC-DPS-doped films with a high PLQY, long emission lifetime, and improved color purity. Owing to its reduced nonradiative energy loss, DMAC-DPS-based DBOSSPO devices showed state-of-the-art performances, including ultralow driving voltages of 3.0 V at the onset, high EQEs up to 19%, and reduced roll-off values as low as 9% at 100 cd m−2, accompanied by the deepest blue emissions from DMAC-DPS reported so far. This work has shown the significance of suppressing the structural relaxation of hosts in simultaneously improving the EL color purity and efficiencies of deep-blue TADF diodes. The effectiveness of the “insulating lock” strategy provides a feasible route for this issue, paving the way for realizing high-quality full-color TADF displays and lighting sources. J.L. and D.D. contributed equally to this work. H.X. is grateful to Prof. Runfeng Chen in Nanjing University of Posts and Telecommunications for assistance with the TDDFT calculations. This project was financially supported by the NSFC (61176020 and 51373050), the New Century Excellent Talents Supporting Program of MOE (NCET-12–0706), Program for Innovative Research Team in University (MOE) (IRT-1237), Science and Technology Bureau of Heilongjiang Province (ZD201402 and JC2015002), Education Bureau of Heilongjiang Province (2014CJHB005), the Fok Ying-Tong Education Foundation for Young Teachers in the Higher Education Institutions of China (141012), and Harbin Science and Technology Bureau (2015RAYXJ008). As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.