High‐Performance Deep‐Blue OLEDs Harnessing Triplet–Triplet Annihilation Under Low Dopant Concentration

Organic blue emitters capable of proceeding triplet–triplet annihilation (TTA) are of great importance for high‐efficiency blue organic light‐emitting diodes (OLEDs). Herein, two deep‐blue emitters PAPE and PAPF with two diphenylanthracene (DPA) moieties linked by fluorene and ether, respectively, together with a single DPA model emitter PAPES are synthesized and characterized. Theoretical calculations indicate that the TTA mechanism of these materials is energetically favorable, which is evidenced by platinum octaethylporphyrin‐sensitized TTA‐upconversion (UC) study. The OLED devices employing o‐DiCbzBz host doped with a low concentration of (1%) PAPE and (3%) PAPF furnish superior maximum external quantum efficiency (EQEmax) up to 7.3% and 7.2% and realize deep‐blue emissions with Commission International de L’Eclairge (CIE) coordinates of (0.15, 0.05) and (0.15, 0.04), respectively, that outperform the model device doped with (1%) PAPES with EQEmax 4.12%. The delayed emission lifetimes from TTA of PAPE‐ and PAPF‐doped devices observed by the time‐resolved electroluminescence (EL) analyses are rather short. Fast TTA is observed with magneto‐electroluminescence, supporting the possibility of intramolecular TTA operation in the devices. This work manifests the potential of multichromophoric strategy to design fluorescent emitters that is able to boost the efficiency of deep‐blue OLEDs even with a low doping concentration.


Introduction
The technologies based on organic lightemitting diodes (OLEDs) have been applied in displays of smartphones and TV, [1] lighting panels, [2] organic lasers, [3] and light sources for biomedical applications. [4] In OLEDs, the recombination of electron and hole generates 25% of singlet excitons and 75% of triplet excitons according to spin statistics. [5] Accordingly, the maximum external quantum efficiency (EQE max ) of OLEDs is dependent on the nature of emitters. Fluorescent emitters can only utilize the singlet excitons due to the selection rule, whereas heavy transition metal-based phosphorescent emitters can harvest both singlet and triplet excitons by the aid of strong spin-orbit coupling that accelerates the rate of intersystem crossing (ISC) process and renders the triplet excitons emission. [6] The high quantum efficiency achieved by phosphorescence green and red emitters makes the commercialization of OLED display feasible. However, issues regarding cost and low Organic blue emitters capable of proceeding triplet-triplet annihilation (TTA) are of great importance for high-efficiency blue organic light-emitting diodes (OLEDs). Herein, two deep-blue emitters PAPE and PAPF with two diphenylanthracene (DPA) moieties linked by fluorene and ether, respectively, together with a single DPA model emitter PAPES are synthesized and characterized. Theoretical calculations indicate that the TTA mechanism of these materials is energetically favorable, which is evidenced by platinum octaethylporphyrinsensitized TTA-upconversion (UC) study. The OLED devices employing o-DiCbzBz host doped with a low concentration of (1%) PAPE and (3%) PAPF furnish superior maximum external quantum efficiency (EQE max ) up to 7.3% and 7.2% and realize deep-blue emissions with Commission International de L'Eclairge (CIE) coordinates of (0.15, 0.05) and (0.15, 0.04), respectively, that outperform the model device doped with (1%) PAPES with EQE max 4.12%. The delayed emission lifetimes from TTA of PAPE-and PAPF-doped devices observed by the time-resolved electroluminescence (EL) analyses are rather short. Fast TTA is observed with magneto-electroluminescence, supporting the possibility of intramolecular TTA operation in the devices. This work manifests the potential of multichromophoric strategy to design fluorescent emitters that is able to boost the efficiency of deep-blue OLEDs even with a low doping concentration.
operation stability of the devices incorporating blue transition metal complexes remain challenging. [7] Alternatively, the emergence of organic emitters with thermally activated delayed fluorescence (TADF) provides a possible solution to conquer the cost-ineffective issue. A number of TADF-based OLEDs with EQE max exceeding 35% have been reported recently. [8] Nevertheless, the blue TADF emitters that can offer a promising device operation lifetime are still pending. One of the solutions to overcome the "blue catastrophe" is to endow the emitter with the capability of utilizing the triplet-triplet annihilation (TTA) mechanism. When the population of triplet exciton ( 3 A*) is sufficient, the collision between two 3 A* can generate a spin-correlated complex j [ 3 A*··· 3 A*] ( j ¼ 1, 3, or 5), [9] and thus initiate intermolecular TTA (inter-TTA). As a result, under the conservation of the total spin, one of 3 A* in 1 [ 3 A*··· 3 A*] collisional complex receives a sum of two triplet-state energy and then populates at the first singlet excited state ( 1 A*), whereas the other 3 A* returns to its ground state ( 1 A). Some high-efficiency OLEDs employing TTA-enabling emitters have been successfully demonstrated. [10] Of particular interest are the anthracene derivatives [11] with strong TTA behavior. These organic emitters can achieve highly efficient deep-blue emission with Commission International de L'Eclairge (CIE) coordinates y < 0.1, which is otherwise rather difficult to attain using phosphorescent and TADF emitters. To realize an efficient inter-TTA mechanism, an effective collision between triplet excitons is necessary. This causes the demand of high doping concentration, making the risk of concentration quenching (CQ) that limits the device efficiency and perhaps reduces the electroluminescence (EL) color purity due to the undesired intermolecular interactions. [10c,12] Also, most of the reported organic materials possessing TTA characteristics are polycyclic aromatic hydrocarbon compounds, [13] giving less flexibility for modulating the frontier orbital energy levels. Nevertheless, attempts for further improving the charge-transporting characteristics and photoluminescence quantum yield (PLQY) of the anthracene-centered emitters have been made by judiciously selected functional groups. These inevitably lead to changes of the molecular packing and hence additional intramolecular and/or intermolecular π-π interactions that would potentially form quenching states such as excimer or aggregates, restraining their practical applications as blue emitters of OLEDs. [14] In addition to the inter-TTA mechanism, to date, several works have pointed out the possibility of exploiting intramolecular TTA (intra-TTA) mechanism with tailor-made bichromophore molecules. In 2016, Albinsson et al. reported the possibility of strengthening TTA-upconversion (UC) performance by covalently connecting multi-diphenylanthracene (DPA) molecular systems in a rigid-polymer matrix. [15] Furthermore, Ikeda et al. clearly showed that the critical factor (Ith) for judging the efficiency of TTA-UC was lower for the σ-spacer linked DPA dimer with a sufficiently close interchromophore distance as compared to that of the DPA monomer and concluded that intra-TTA was operative. [16] Fundamentally, the distance between annihilators is not the only factor that affects intra-TTA behavior. The relative configuration between chromophores is of great importance as well. Wong et al. studied the phenylene-bridged DPA derivatives and found that the meta-linked DPA dimer showed the shortest triplet lifetime, the highest UC quantum yield, and the largest magnetic field effect, giving the highest probability of intra-TTA. [17] Congreve et al. in 2019 reported that πÀbridged tetracene dimers have greater TTA-UC yield and less sensitivity to the concentration, and a smaller power threshold for excitation than the corresponding monomer. [18] In 2021, Albinsson et al. deeply investigated the π-bridged DPA dimer by time-resolved UC emission measurements. These, together with computer simulations, let them firmly ascertain the "double sensitization" model possibly incorporating the intra-TTA mechanism. [19] Till now, most of the intra-TTA mechanisms were explored in solution. To our best knowledge, OLED devices with intra-TTA emitter have not been demonstrated yet, which is a goal pursued by the researchers.
In this work, we report the strategic design of organic emitters possessing two DPA units and demonstrate their high-efficiency deep-blue OLEDs under a low doping concentration. Three anthracene-based deep-blue emitters, namely PAPE, PAPF, and PAPES (Scheme 1) were synthesized and characterized. To examine the relationship between molecular configurations and the associated photophysical properties, high molecular rigidity fluorene and lower molecular rigidity ether was adopted as the linkage to bridge two DPA groups for molecules PAPF and PAPE, respectively. As compared with the model molecule PAPES with a single DPA, the possibilities of PAPF and PAPE to perform intra-TTA in solution and pristine films, as well as doped films, were explored. Upon excitation with a green laser (532 nm) to the toluene solution comprising a platinum-based sensitizer and the respective new molecule, a blue emission from TTA-UC can be clearly observed. The intensity of upconverted blue emission significantly reduces when the solution is aerated, suggesting the triplet-involved mechanisms. The possibility of TTA-UC facilitated by intramolecular dual DPA in PAPE (cf. PAPES) was supported by a sensitization experiment. Finally, three doped OLED devices with an ultralow doping concentration (1 vol%) of emitter show deep-blue emissions and excellent EL performance. In particular, the devices using PAPE and PAPF as emitters furnish noticeably high EQE max of 7.3% and 7.2%, surpassing the theoretical limit of traditional fluorescent emitters, and realize deep-blue emissions with CIE coordinates of (0.15, 0.04) and (0.15, 0.05), respectively. In comparison, the standard deep-blue device based on PAPES can only obtain a significantly lower but regular EQE max of 4.12%. The time-resolved EL analyses reveal that the delayed emission lifetimes (<0.2 μs) of devices with PAPE and PAPF emissions from TTA are much faster as compared to those of the counter devices employing PAPES and the reported anthracene derivatives. [20] The fast TTA can also be observed under the magnetic field, verifying the possibility of performing intra-TTA mechanism in the devices. The device characteristics achieved in this work demonstrate a new strategy of using bisanthracene emitter capable of performing intra-TTA process for high-performance deep-blue OLEDs with a low dopant concentration.

Electrochemical Properties
The electrochemical behaviors of PAPE, PAPF, and PAPES were probed by cyclic voltammetry (CV). The cyclic voltammograms are depicted in Figure 1, and the data are summarized in Table 1. An oxidation potential was observed at about 1.30 V (vs Ag/AgCl) for PAPE, PAPF, and PAPES, which can be assigned to the oxidation of anthracene moiety according to other anthracene derivatives reported previously. [22] PAPE and PAPF exhibit a similar reversible reduction at À1.70 and À1.71 V (vs Ag/AgCl), respectively, whereas PAPES shows a reversible  Calculated from the oxidation potential of CV spectra. HOMO ¼ À4.8 (eV) À (E oxi À Ferrocene oxi ). b) Calculated from the absorption onset at 10 À5 M solution in toluene. c) LUMO ¼ HOMO þ E g . d) Estimated by 5% weight loss. e) Not detectable.
www.advancedsciencenews.com www.adpr-journal.com reduction potential at À1.66 V. The reduction potentials of PAPE and PAPF are almost identical to that reported to DPA. [23] In contrast, PAPES displays a slightly lower reduction potential (0.04 V) as compared to those of PAPE and PAPF. Based on the obtained CV results, the corresponding highest occupied molecular orbital (HOMO) levels and lowest unoccupied molecular orbital (LUMO) levels are then calculated by taking ferrocene/ ferrocenium redox as the standard, which exhibits a HOMO energy level of À4.8 eV in vacuum. [24] The calculated HOMO energy levels for PAPE, PAPF, and PAPES are À5.55, À5.58, and À5.55 eV, respectively. The LUMO energy levels are deduced by the equation LUMO ¼ HOMO þ E g , which are À2.56, À2.58, and À2.54 eV for PAPE, PAPF, and PAPES are, respectively.

Photophysical Properties
The steady-state UV-Vis absorption and PL spectra of PAPE, PAPF, and PAPES in 10 À5 M toluene were measured at room temperature and 77 K shown in Figure 2a. Pertinent data are summarized in Table 2. PAPE, PAPF, and PAPES exhibit identical vibronic profiles in absorption spectra with absorption peaks located at 358, 376, and 397 nm, which can be assigned to the π-π* transition originating from the DPA moiety. [25] Independent of the number of DPA being anchored, the fluorescence of PAPE, PAPF, and PAPES in room temperature toluene exhibits similar spectral features, showing λ max centered at 417 nm together with vibronic progression at 434 and 466 nm ( Figure 2a). Therefore, there is no π-conjugation extension of DPA as two DPA groups are bridged by the C9 of fluorene and oxygen atom. In comparison to the room temperature spectra, the emission λ max of PAPE, PAPF, and PAPES measured in the 77 K toluene matrix is slightly blue-shifted at 410 nm, accompanied by the vibronic progression at 433 and 458 nm that is associated with the transition of DPA moiety. [25] Notably, the vibronic structure of PAPF is more obvious than that of PAPE     at 77 K (Figure 2a), manifesting the enhancement of molecular rigidity via the fluorene bridge in PAPF, whereas two DPAs connecting to an oxygen atom allow PAPE to possess more single bond rotation flexibility. Figure 2b depicts the absorption and emission spectra of the vacuum-deposited film (50 nm thickness) of PAPE, PAPF, and PAPES under nitrogen environment. All pertinent data are summarized in Table 2. The absorption spectra in film show a similar pattern as compared to those obtained in the solution, whereas the emission is a redshift of %10 nm resulting from the intermolecular interactions of DPA moieties. In addition to the main emission peak, PAPF exhibits a strong red-shifted emission peaked at 506 and 553 nm, which are reasonably attributed to the DPA excimer emission. [26] This result indicates that the fluorene bridge provides molecular rigidity to restrict the free rotations of DPA branches, leading to a higher propensity of intermolecular interactions in the film upon vapor deposition.
The PLQYs of PAPE, PAPF, and PAPES in solution and solid film were measured; the corresponding data are summarized in Table 2. In the degassed toluene solution all three derivatives, within the experimental error, show PLQYs of nearly 100%, which is close to the reported PLQY of DPA in toluene. [27] In sharp contrast, the film PLQYs are significantly reduced for all three studied compounds (Table 2), reaffirming the above proposed intermolecular interactions that lead to CQ in the solid film. Particularly, PAPF exhibits the lowest PLQY (15%) owing to the excimer formation. Attempts to acquire phosphorescence spectra recorded by 1 μs delay after excitation failed to resolve meaningful signal in both 77 K toluene matrix and solid film (very small signal/noise perhaps due to the scattering, see Figure S1, Supporting Information). This result is not surprising because all PAPE, PAPF, and PAPES have nearly %100% fluorescence yield in the solution, indicating the negligible efficiency of ISC. In contrast, the reduction of fluorescence yield in the solid film for the three studied compounds is due to the aggregation quenching (or excimer formation) not the ISC. Therefore, direct excitation of PAPE, PAPF, and PAPES results in virtually no T 1 population.

Computational Approach
To shed light on the structure-optical property relationship as well as gains more insight into the TTA property, we then applied density functional theory (DFT) and time-dependent DFT calculations for PAPE, PAPF, and PAPES. The approaches were conducted at B3LYP/6-311 þ G** level using Gaussian 16 program where solvent effects were considered using the polarizable continuum model with toluene as the solvent (dielectric constant ε ¼ 2.38). [28] As a result, the ground state structures of PAPE, PAPF, and PAPES exhibit a highly twisted conformation due to the sp 3 -hybridized C9 bridge of fluorene and the oxygen linkage. The calculated dihedral angles between two DPA groups in the optimized PAPE, PAPF, and PAPES are around 70°as shown in Figure 3 and Table S1, Supporting Information. This molecular conformation corresponds to a distance between the two phenyl groups around 8.3 Å and a distance between the C9 and C9 0 position of the DPA groups is 9.9 and 10.2 Å for PAPF and PAPE ( Figure S2, Supporting Information), respectively. As reported by Ikeda et al., [16a] the molecules exhibiting different emission behaviors from the DPA chromophore such as reducing threshold intensity and increasing TTA rate with the distance of two DPA groups ranging from 8.3 to 12.3 Å. Note that TTA requiring the Dexter type energy transfer, which is efficient when the distance between the two triplet species 3 DPA* is smaller than 10 Å. [29] Thus, it is reasonable to expect the possibility of a through-space intramolecular interaction, Dexter energy transfer more specifically, for proceeding the subsequent  intra-TTA in PAPE and PAPF. The calculated frontier molecular orbital distributions are shown in Figure 3 and S3, Supporting Information. Table S2, Supporting Information, lists the calculated optical excitations and their compositions for the optimized PAPE, PAPF, and PAPES. Singlet and triplet energy levels at S 0 optimized structures of PAPE, PAPF, and PAPES are depicted in Figure 4. The calculated S 1 state energy of PAPE, PAPF, and PAPES is 405.4 nm (3.06 eV), 406.4 nm (3.05 eV), and 406.5 nm (3.05 eV), respectively, which is consistent with the experimental values (vide supra). The T 1 energy is estimated to 1.76 eV for all three compounds, twice of which is higher in energy than the S 1 state, making the TTA process energetically favorable. It is worth noting that the calculated energies of S 1 and T 1 are close to those of the DPA monomer in the reported literature. [30] The NTO analyses of S 1 and T 1 transitions for all three systems are shown in Figure S4, Supporting Information. In the optimized ground state structures of PAPE, PAPF, and PAPES, both S 1 and the T 1 states are dominated by the π ! π* transitions, exhibiting local excitation character. Only slight variation is found in molecular conformations between ground and excited states, indicating small reorganization energy and hence a slow rate of internal conversion. This, in combination with the rather slow rate of S 1 (ππ*) ! T 1 (ππ*) ISC according to the El-Sayed rule, [31] rationalizes the high PLQY for all studied DPA derivatives measured in solution (vide supra).

Probing the TTA
The nearly unity PLQY for all three studied compounds makes infeasible the direct excitation to probe TTA-induced fluorescence properties. Alternatively, we then performed a photosensitization experiment to populate the T 1 state of the studied systems. The experiment comprised a triplet sensitizer (10 À4 M), platinum octaethylporphyrin (PtOEP), and the respective derivative (10 À4 M) in the degassed toluene solution. The lowest lying triplet state of PtOEP is reported to be around 1.90 eV, [32] which is higher than the calculated T 1 state of 1.76 eV for PAPE, PAPF, and PAPES (vide supra), ensuring the energy transfer to be thermally favorable. As shown in the inset of Figure 5a, TTA-UC in this study could be realized by selective excitation of the sensitizer only at >500 nm, where triplet exciton can be efficiently generated on PtOEP via the Pt enhanced spin-orbit coupled ISC, followed by Dexter energy transfer to the target emitter, giving TTA to emit blue delayed fluorescence. As a result, strong blue emissions (400-500 nm) from PAPE, PAPF, and PAPES solutions are clearly observed, verifying the TTA-UC mechanism occurred in PtOEP-sensitized solution containing PAPE, PAPF, and PAPES. The intensity of blue emission via TTA-UC largely reduces as the solution was aerated, and no blue emission in the absence of PtOEP was detected ( Figure S5, Supporting Information), affirming the origin of the fluorescence from TTA. Under the same excitation energy and PtOEP absorbance as well as the concentration (10 À4 M) prepared for PAPE, PAPF, and PAPES, Figure 5b shows the corresponding intensity of the delayed fluorescence acquired by the intensified charge-coupled detector (ICCD) where the gate is open at a delay time of 1 μs with a width of acquisition widow of 100 μs after exciting PtOEP (532 nm YAG laser with 5 ns pulse duration). Note that the emission spectral feature shown in Figure 5a is distorted from the steady state emission acquired by the photomultiplier tube. This is mainly due to the non-calibrated ICCD that is less sensitive at <430 nm. Clearly, the TTA-triggered delayed fluorescence for PAPE and PAPF is about 3.4 and 1.3 times higher than that of PAPES. In the sensitization experiment, the kinetics of TTA is relevant to the diffusive bimolecular energy transfer and annihilation; hence, the efficiency of generation of delayed fluorescence is concentration dependent and complicated. Nevertheless, the results, in a qualitative manner, manifest the advantage of dual anthracene anchored on PAPE and PAPF, which may be considered as an increase in the anthracene concentration effectively. The smaller enhancement of the delayed fluorescence for PAPF (cf. PAPE) is believed to be www.advancedsciencenews.com www.adpr-journal.com due to the high concentration (>10 À4 M) required for the sensitization experiment, where PAPF is subject to the aggregation and hence the excimer formation (vide supra). The aggregation effect can be reduced in the later OLEDs application because the doping concentration can be further reduced by direct population in the triplet manifold (vide infra).

Application in OLEDs
To eliminate the CQ effect of TTA emitters and apply them in OLED applications, a host-guest system was employed with (9,9 0 -(2-(1-phenyl-1 H-benzo[d]imidazol-2-yl)-1,3-phenylene)bis (9H-carbazole) (o-DiCbzBz) as a host material. o-DiCbzBz is a wide bandgap bipolar transporting host material with high singlet (S 1 ¼ 3.6 eV) and triplet energy (T 1 ¼ 3.1 eV) to ensure the energy confinement for TTA emitters. [33] The respective dopant concentration PAPE, PAPF, and PAPES is 1%, 3%, and 1% to match the optimized EML structure in OLED applications. The fluorescence spectra of three doped thin films are shown in Figure 6a. They exhibit similar blue emissions profile at 400-500 nm, which are similar to those observed under solution conditions. Especially, the excimer emission at 500-600 nm of PAPF in the pure solid film is not observable in the mixed film, indicating the suppression of intermolecular interaction, to a certain extent, upon host/guest co-deposition. Nevertheless, the intermolecular interactions could not be completely eliminated, as evidenced by the PLQY of 41% and 39% for PAPF and PAPE doped films, respectively, which is lower than that in the diluted toluene solution (vide supra). Moreover, as shown in Figure 6a   Note that the emission spectral feature is distorted from the steady state emission acquired by a photomultiplier detector in (a). This is mainly due to uncalibrated ICCD which is less sensitive in the blue region.
www.advancedsciencenews.com www.adpr-journal.com The TAPC serves as a hole injecting layer, mCP as holetransporting layers (HTLs), o-DiCbzBz as host of an emitting layer (EML), DPPS as an electron-transporting layer (ETL), and LiF as an electron injection layer. In addition, mCP can act as a triplet blocking layer due to its high triplet energy (T 1 ¼ 2.9 eV) to confine the triplet exciton within the EML. [34,35] The thicknesses (X, Y, and Z ) were varied to achieve the charge balance conditions with X ¼ 50, Y ¼ 3, Z ¼ 20 for PAPF and X ¼ 30, Y ¼ 1, and Z ¼ 25 for PAPE and PAPES.
The thickness of ETL, EML, and HTL and the dopant concentration of EML were well optimized in term of EQE max .
For PAPE and PAPES, the dopant concentration in terms of volume ratio ensures the density of anthracene moieties are identical.
The EL characteristics of deep-blue OLEDs are shown in Figure 7 with the corresponding data, including luminance (L)current density ( J)-voltages (V ), current efficiencies (CEs), power efficiencies (PEs), and EQEs versus current densities and normalized EL spectra at 4 V summarized in Table 3. The turn-on voltages at 1 cd m À2 of three deep-blue OLEDs are around 3.6-3.7 V closing to the optical bandgap of the host material. The driving voltages at J ¼ 1 mA cm À2 are also similar. This  Table S3, Supporting Information. Recorded at luminance L ¼ 1 cd m À2 , current density J ¼ 10 mA cm À2 . b) Recorded at maximum and L ¼ 100 cd m À2 .
www.advancedsciencenews.com www.adpr-journal.com result indicates that the charges are mainly recombined on host material followed by energy transfer to emitters. As displayed in Figure 7b, PAPF-based deep-blue OLED shows the highest maximum CE (CE max ) and maximum PE (PE max ) of 3.2 cd A À1 and 2.8 Lm W À1 , respectively. From the EQE performances shown in Figure 7c, PAPE and PAPF-based OLEDs exhibit impressive results with EQE max of 7.3% and 7.2%, respectively. These outstanding EQEs are almost two times higher as compared to that (EQE max ¼ 4.1%) of the reference deep-blue OLED using PAPES (mono DPA moiety) as an emitter. Notably, the EQE performances of PAPE and PAPF-doped deep-blue OLEDs surpass the theoretical limit (EQE ¼ 5%) of conventional fluorescence OLEDs. This result implies a significant role played by the triplet exciton in these deep-blue OLEDs. However, the devices are suffering from the evident efficiency roll-off, in which the PAPE-based OLEDs is slightly better than the PAPF-doped one. The consecutive (25 cycles) cyclic voltammograms as shown in Figure S7, Supporting Information, show unchanged profiles during multiple redox cycles, ruling out the possibility of device efficiency roll-off due to the electrochemical stability of PAPE. We speculate the efficiency roll-off is mainly related to the energy of emitters. As these emitters generate deep-blue excitons which are easy to interact with high energy polaron, resulting in a hot excited state (>6 eV) that degrades the emitters and thus efficiency roll-off. [36a,b] Nevertheless, the origins of the inferior stability of device efficiency need to be carefully identified, which is out of the scope of this work. From the normalized EL spectra shown in Figure 7c, the EL features of deep-blue OLEDs are from the DPA moieties with the peak wavelength centered at 442 nm. The spectral profiles of EL are similar to the DPA emission behavior observed in toluene. Moreover, no excimer emission from DPA around 500À600 nm can be observed. In addition, there is no residual emission from the host at 380À400 nm, representing the efficient FRET occurring from host to emitters. Three OLEDs exhibited CIE coordinates (Figure 7d) of (0.15, 0.04), (0.15, 0.05), and (0.15, 0.04) with PAPE, PAPF, and PAPES dopant, respectively, which can be regarded as deep-blue emission. To the best of our knowledge, the performances of the current deep-blue OLEDs are outstanding among the recently reported deep-blue OLEDs with CIE y % 0.04-0.05 (Figure 7e  and Table S3, Supporting Information).

Discussion
To gain more insight into the anomalously high efficiency of these deep-blue OLEDs under such a low doping concentration, the light outcoupling efficiency (η out ) of EML was first examined, where the horizontal ratios of the emission transition dipole (Θ) of the doped films were measured with an angle-dependent PL intensity (ADPL) experiment (Figure 8). The host material, o-DiCbzBz, was reported to have the capability of promoting the light outcoupling efficiency of dopants by forming a regular mosaic packing structure along the out-of-plane direction. [37] Under the dopant concentration of 1% and 3% for PAPE and PAPF, respectively, which matches the optimized device structure of OLEDs, the mixed films exhibit Θ of 71% and 76%, respectively. Notably, the Θ of PAPE is slightly lower compared with PAPF owing to its higher degree of rotation freedom. This result indicates that the anisotropic emission character, to a certain extent, is beneficial for enhancing light outcoupling efficiency. However, the observed Θ values are lower than the state-of-the-arts emitter with high efficiency (Θ > 90%). [38] Hence, the light outcoupling efficiency could not be the only reason to rationalize the high-efficiency deep-blue OLEDs.
By using a molecular weight of PAPE, PAPF, and PAPES, and the density of 9,10-diphenylanthracene (ρ ¼ 1.22 g cm À3 ), the average distance (R avg ) between emitters within the host matrix can be approximately estimated. [39] R avg values are 4.64, 3.44, and 3.97 nm for PAPE, PAPF, and PAPES, respectively. If the molecules are uniformly dispersed in the host matrix, these distances can barely allow TTA occurring with triplet hopping by Dexter energy transfer. From the ADPL experimentally observed Θ (also Figure S8, Supporting Information) and PLQYs, the maximum theoretical exciton utilization efficiency (EUE) can be calculated by using the formula EQE max ¼ η PLQY Â η out Â EUE Â γ, where γ (electron-hole recombination) assumed to be 1 since the device structure is well optimized. η PLQY are the PLQY values of TTA emitters in o-DiCbzBz host in solid state (39% for PAPE, 41% for PAPF) and η out are the outcoupling efficiencies calculated from optical simulation in Figure S8, Supporting Information, (0.31 for PAPE and 0.35 for PAPF). The calculated EUE of PAPE-and PAPF-based OLEDs is 0.53 and 0.56, respectively (Table S4, Supporting Information). Apparently, the estimated EUE is higher than the assumption (EUE ¼ 0.25) of a typical fluorescence dopant.  To unveil the EL emission mechanism of three deep-blue OLEDs, time-resolved EL (TrEL) were conducted. Figure 9 (left) depicts the TrEL results. A square wave with J ¼ 1 mA cm À2 was employed to charge the device to a steady state at t < 0 (μs). Then, the square wave was turned off at t > 0 (μs) followed by a reversed bias (À9 V) to eliminate the charge effect. Hence, the decay signal can be regarded as the exciton dynamics of emitters. From the TrEL signal of a PAPES-based device, multiexponential decay was observed within a 10 μs window. This observation can be ascribed to the common behavior of TTA process, with the first component represents as the prompt emission from the singlet state, while the second component as the delayed emission from TTA. [40] The time scale of the second component from TTA usually falls on μs window since it consumes time for the triplet excitons to hop among the emitters. The delayed fluorescence lifetime (τ 2 ) of a PAPES-based device is 1.12 μs with a triplet contribution of less than 0.05 (see Figure 9 (left)). Due to the mono-DPA design, the delayed fluorescence originating from PAPES-based deep-blue OLED can only be recognized by the intermolecular TTA. On the other hand, the TrEL signals of PAPE-and PAPF-based devices show only one exponential decay with a lifetime of %0.2 μs, which is limited by our instrument responses. No other delayed fluorescence longer than, e.g., 1 μs can be resolved in TrEL for PAPE-and PAPF-based devices.
Hence, there are two hypotheses for the high EUE in our devices: 1) The triplet exciton is harvested from the higher lying triplet excited state to the singlet excited state through reversed ISC (rISC). This proposed mechanism incorporating hybridized local and charge-transfer (HLCT) becomes emergent in fluorescence OLEDs to rationalize the anomalously high EQE while lacking outcoupling. [41] However, except for the energy of T 2 being slightly higher than S 2 , the structures of the studied compounds provide no relevance to the HLCT. Therefore, this proposed mechanism is greatly discounted. 2) Alternatively, we consider the TTA process incorporating intermolecular and intramolecular processes. The former can be solely applied to the PAPES-based device to explain its EQE max of 4.1%, whereas the latter emphasizes the dual anthracene moieties in both PAPE-and PAPF-based devices to account for the anomalously high EQE max of 7.3% and 7.2%, respectively. Intuitively, the dual anthracene moiety in PAPE and PAPF effectively double the concentration with respect to PAPES. More than this, considering that T*-T* annihilation is through the hopping of T*, which should be greatly facilitated via intramolecular dual anthracene as the interim bridge, the generation of delayed fluorescence is more efficient and the associated relaxation dynamics are faster (cf. PAPES), as evidenced by the %0.2 μs decay of the delayed fluorescence for PAPE and PAPF versus 1.12 μs in PAPES.
To verify the anomalously short delayed fluorescence observed in TrEL is indeed from TTA, a magneto-electroluminescence (MEL) experiment was conducted. [42] In this experiment, a parameter MEL (in %), is defined as MEL ¼ ΔEL/ EL ¼ (EL(B) À EL(0))/EL(0), where B is the magnetic field, EL(B) and EL(0) are the detected EL intensity with and without an external magnetic field. [43] Figure 9 (right) depicts the MEL curves of the PAPE-based device measured at different bias voltages (current density). As displayed in Figure 9 (right), at a low magnetic field regime (B ¼ 0À250 Oe), the plot of MEL exhibits positive MEL responses, being increased with the increase of the applied magnetic field. However, at the higher magnetic field (>250 Oe), MEL response decreases and turns to negative magnitudes at 2000 Oe. The observation of positive to negative MEL responses from low and high magnetic field regimes is indicative of the TTA process contributing to EL of the PAPE-based device, the result of which agrees well with the previously published literatures. [43,44] The states of the intermediate triplet pairs, which allow TTA process to give EL, reduce at the relatively high magnetic field, resulting in the negative MEL response. [41a] Moreover, the negative MEL responses increase upon increasing the bias voltages (or current density), for example, À0.6%, À0.8%, and À1.2% at 2000 Oe for the PAPE-based OLED measured at 4 V (%0.31 mA cm À2 ), 5 V (%2.97 mA cm À2 ), and 6 V (%9.28 mA cm À2 ), respectively. The negative MEL response increases with the bias current density at the higher magnetic field regime (2000 Oe). Evidently, the device biased at the high voltage (high current density) creates a higher density of triplet excitons that lead to a greater contribution of TTA-induced fluorescence to EL [43] in the PAPE-based device.
From the TrEL measurement, the rate constant of TTA process of PAPE and PAPF could be much faster than that of PAPES and other traditional anthracene derivatives (several to hundreds of μs). [20] This observation is in line with the enhanced delayed fluorescence intensity for PAPE (or PAPF) (cf. PAPES) in the PtOEP sensitization experiment (see Figure 5b).

Conclusion
In summary, two deep-blue emitters PAPE and PAPF containing two DPA moieties linked by fluorene and ether, respectively, together with a single DPA model emitter PAPES were synthesized and characterized. The theoretical calculations reveal that the TTA mechanism of these compounds is energetically favorable, which were verified by PtOEP-sensitized triplet population that leads to strong blue emission via TTA-UC process. The OLED devices employing o-DiCbzBz as host material with a low doping concentration of emitter exhibit excellent EL performance, in which the (1%) PAPE and (3%) PAPF doped devices

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.