Molecular design and theoretical calculation
Our molecular design strategy and associated theoretical prediction results at the B3LYP-B3(DJ)/6-31G(d,p) level are illustrated in Fig. 1. The DPA-B2, bearing the same double-borylated π-skeleton as B2 reported in ref37, was selected as the parent emitter. With reference to the renowned emitter ν-DABNA 23, DPA-B2 featuring one less nitrogen atom displayed an enlarged energy gap and hypsochromically shifted emission, which is advantageous for preserving the emission peak in the deep-blue region after π-extension. Additionally, the more distorted geometry with the presence of a highly twisted B,N,B-doped [4]helicene, caused by mismatched bond lengths between C-N and C-B bonds, had a strongly positive effect on SOC. This was evident from the substantial improvements in the SOC matrix elements as assessed using PySOC (< S1|ĤSOC|T1 > = 0.15 and 0.01 cm− 1, <S1|ĤSOC|T2 > = 1.12 and 0.21 cm− 1, <S1|ĤSOC|T3 > = 0.28 and 0.19 cm− 1 for DPA-B2 and ν-DABNA, respectively). The elevated SOC values in DPA-B2 arose from the out-of-plane transitions due to the associated changes in the angular momentum. This observation was consistent with state-of-the-art theoretical and experimental studies that underscored the significance of twists/non-planarity in augmenting SOC in graphenoid structures38–42. The above characteristics render DPA-B2 an ideal candidate for further structural modifications.
Building on DPA-B2, we designed triple- and quadruple-borylated emitters, DPA-B3 and DPA-B4 by incorporating one and two rigid boron/nitrogen/oxygen-embedded triangulene units into the peripheral benzenes, where the boron atoms were linearly aligned with nitrogen/oxygen at their ortho-/para-positions. The inclusion of oxygen aimed to minimize the red-shift, given its lesser contribution to π-conjugation extension compared to nitrogen, due to its lower atomic orbital energy43–46. Time-dependent density functional theory (TD-DFT) predictions revealed only a marginal decrease in S1 energy from DPA-B2 (3.12 eV) to DPA-B3 (2.97 eV) and DPA-B4 (2.88 eV), despite the gradually enlarged π-systems. This indicated the success of our molecular design in preserving deep-blue emission. Notably, while linear π-extension enhances the overall structural planarity, the highly distorted geometry of DPA-B2 center remains intact, and the number of twisted regions also rises (Supplementary Fig. 30). With this unique structure, we observed both reduced ΔEST due to promoted SR-CT within the conjugation plane (0.32 eV and 0.26 eV for DPA-B3 and DPA-B4, respectively, compared with 0.41 eV for DPA-B2), and retained large SOC induced by the embedded highly twisted helicene subunits (for example, <S1|ĤSOC|T1 > = 0.02 cm− 1, <S1|ĤSOC|T2 > = 0.63 cm− 1 and < S1|ĤSOC|T3 > = 0.66 cm− 1 for DPA-B4, where both T2 and T3 lie between S1 and T1). To account for electron correlation in double excitations, we also conducted higher-level RI-SCS-CC2 calculations for excited-state energy estimations22,47. The ΔEST values of these compounds decreased progressively with the extension of the π-skeleton, from 0.10 eV for DPA-B2 to 0.04 eV for DPA-B3 and 0.02 eV for DPA-B4, aligning well with TD-DFT calculations and photophysical outcomes (vide infra). The close-to-resonant S1 and T1 states and considerable SOC matrix elements in DPA-B4 are expected to translate into a high kRISC value based on the following relationship: kRISC ∝ <S1|ĤSOC|Tn>²/ΔEST, which was crucial for triplet exciton utilization in the EL device.
To discern the relationship between structure and emission properties, we calculated and analyzed the reorganization energy and Huang-Rhys factors (S) using the Molecular Materials Property Prediction Package (MOMAP) software. Our analysis revealed that the DPA-B2 parent molecule underwent substantial structural relaxation during the excitation-emission process, owing to its distorted geometry. This was reflected by the apparent increase in theoretically predicted total reorganization energy when compared to that of ν-DABNA (Fig. 1d and Supplementary Fig. 31). However, this adverse effect was mitigated in DPA-B3 and DPA-B4, which exhibited progressively decreasing total reorganization energy due to their enlarged fused-ring systems. The relationships of reorganization energy and S with normal vibration modes for the S1→S0 transition are further illustrated in Fig. 1e and Supplementary Figs. 33-S34, respectively. Obviously, the net effect of linear π-extension and non-bonding frontier molecular orbitals diminished the vibronic coupling strength of both low-frequency modes (< 500 cm− 1, dominated by the rotational vibrations of trimethylbenzene and the twisting vibration of the π-skeleton) and high-frequency modes (> 500 cm− 1, the swing of C-H at the peripheral benzene rings and the stretching vibration of C-C bonds in the backbone). Specifically, DPA-B4 demonstrated very tiny S values below 0.25 for all vibration modes, which benefitted the formation of both a narrowed FWHM and weakened shoulder peak in the spectrum.
From the above discussion, it was clear that i) linearly π-extended MR scaffold, ii) reasonable element combination, iii) fully non-bonding character, and iv) partial twisted geometry were four key elements in our molecular design principle towards high-performance deep-blue emitters. To further elucidate the importance of the latter two factors, we designed a control emitter, Cz-B4, through intramolecular cyclization of the terminal phenyl units in the B,N,B-doped [4]helicene from DPA-B4. This minor structural difference did not induce notable changes in S1 energy, but it triggered considerably larger ΔEST (0.32 eV from TD-DFT, and 0.12 eV from RI-SCS-CC2 calculations) due to the introduction of π-bonding/antibonding character at the biphenyl site, as well as an overall decrease in SOC values (similar < S1|ĤSOC|T1 > = 0.05 cm− 1, <S1|ĤSOC|T2 > = 0.61 cm− 1 but significantly reduced < S1|ĤSOC|T3 > = 0.08 cm− 1) due to the more planarized geometry. On the other hand, although the intramolecular ring fusion induced higher rigidity in Cz-B4, as evidenced by smaller root-mean-square displacement (RMSD) values between S0 and S1 geometries (0.026 Å for Cz-B4 compared to 0.042 Å for DPA-B4, Supplementary Fig. 32), we observed a 33% increase in its total reorganization energy. This increase is attributed to intensified high-frequency modes with relatively larger S. Of note, the dominant stretching mode at 1654.9 cm− 1 is closely associated with the stretching vibration of the central biphenyl unit, highlighting the adverse effects of π-bonding/antibonding molecular orbitals on emission broadening.
Synthesis, characterization and photophysical properties
With the established design strategy in place, we conducted the synthesis of the emitters using a one-shot multiple borylation process with moderate yields (the detailed synthetic procedure and characterization are given in the Supplementary Information) 37. These four compounds exhibited remarkable thermal stability, as evidenced by high decomposition temperatures (Td, corresponding to a 5% weight loss) of 376°C, 424°C, 497°C, and 513°C for DPA-B2, DPA-B3, DPA-B4, and Cz-B4, respectively (Supplementary Fig. 36), underscoring their potential suitability for OLED fabrication via vacuum thermal deposition. Single crystals of DPA-B3 and DPA-B4 were successfully cultivated through the slow evaporation of an ethanol/dichloromethane mixture at room temperature (Supplementary Fig. 38, Supplementary Tables 1–2). These crystals revealed quasi-planar skeletons with length-to-width ratios of approximately 2.40:1.87 and 2.93:1.77 times their height, respectively. Moreover, the molecular edges exhibited significant twisting due to the incorporation of multiple helicene moieties and perpendicularly aligned phenyl and mesitylene units. This quasi-planar architecture with twisted molecular edges offers several additional advantages, including a high horizontal ratio of emitting dipole orientation (Θ//)48,49 and the mitigation of aggregation-induced quenching50,51. Consequently, these structural characteristics were expected to significantly enhance the efficiency of OLEDs.
The steady-state photophysical properties of the four emitters were next investigated in dilute toluene solution (1 × 10− 5 M) at 300 K, with the results provided in Fig. 1a and Table 1. The photoluminescence (PL) spectra of the four compounds revealed intense emission bands in the deep-blue area, with emission maxima at 440, 448, 458, 455 nm, Stokes shift values of 16, 10, 10, 14 nm, FWHM values of 32, 14, 12, 24 nm and CIE coordinates of (0.150, 0.037), (0.150, 0.035), (0.140, 0.050), (0.139, 0.072) for DPA-B2, DPA-B3, DPA-B4 and Cz-B4, respectively. The observed trends in the Stokes shift and FWHM values in the DPA-Bn series indicated a positive effect of linear π-extension in suppressing S1-S0 structural shifts, thereby compensating for the negative spectral effects caused by the twisted geometry. In contrast, the increase in Stokes shift and FWHM values in Cz-B4 was in agreement with its enlarged reorganization energy, emphasizing the importance of minimizing π-bonding/antibonding interactions in designing ultranarrow-band deep-blue emitters. Notably, the FWHM value of DPA-B4 sets a record for organic blue emitters, and despite its relatively low energy emission onset/maxima, it exhibits excellent CIEy of 0.050, closely approaching the BT.2020 blue standards. This impressive performance can be attributed to its narrow, near-Gaussian shaped spectrum and minimal vibrational shoulder peak, as a result of the well suppressed vibration modes in DPA-B4. The PL spectra of these emitters in different solvents revealed a positive solvatochromic effect with slight bathochromic shifts and marginally increased FWHMs from hexane to acetonitrile (Supplementary Fig. 39), confirming the SR-CT nature of the emission and maintained high color purity upon raising the environmental polarities.
Table 1
The physical data of DPA-B2, DPA-B3, DPA-B4 and Cz-B4.
Emitter | λabsa [nm] | λema [nm] | FWHMa [nm] | EHOMOb [eV] | ELUMOc [eV] | ES1d [eV] | ET1d [eV] | ∆ESTe [eV] | τpf [ns] | τdf [µs] |
DPA-B3 | 424 | 440 | 32 | -5.21 | -2.38 | 2.92 | 2.72 | 0.20 | 2.4 | 45.6 |
DPA-B3 | 438 | 448 | 14 | -5.14 | -2.38 | 2.83 | 2.69 | 0.14 | 1.9 | 12.5 |
DPA-B4 | 448 | 458 | 12 | -5.14 | -2.43 | 2.78 | 2.76 | 0.02 | 1.8 | 2.5 |
Cz-B4 | 441 | 455 | 24 | -5.19 | -2.47 | 2.79 | 2.64 | 0.15 | 3.8 | 17.4 |
aPeak of absorption (λabs), fluorescence (FL) spectra, full-width at half-maximum (FWHM) of fluorescence measured in toluene (1 × 10− 5 M). bHOMO energy determined in dichloromethane. cEstimated from the HOMO and Eg. dS1, and T1 energy value measured in diluted toluene solution at 77 K. eS1-T1 energy gap. fPrompt fluorescence lifetime (τPF), delayed fluorescence lifetime (τDF).
The photophysical properties of four emitters in film states, specifically doped in 7-((2'-methyl-[1,1'-biphenyl]-4-yl)oxy)-3,11-di-o-tolyl-5,9-dioxa-13b-boranaphtho[3,2,1-de]anthracene (DOBNA-OAr)23,52, were subjected to further analysis. The doping concentration was set at 1% to prevent interchromophore quenching. As shown in Supplementary Fig. 42, the emission spectra bore a resemblance to those observed in the solution state. The slightly broadened FWHMs of 40, 20, 16, and 29 nm could be attributed to the enhanced polarity of the surrounding components and intermolecular interactions. Subsequently, the absolute ФPLs of these emitters in the corresponding deposited film were evaluated, yielding values of 82%, 93%, 99%, and 93% for DPA-B2, DPA-B3, DPA-B4 and Cz-B4, respectively. The ФPL value of the DPA-B4-doped film approached unity, owing to suppressed nonradiative transitions stemming from the extended rigid skeleton with minimized bonding/anti-bonding character. Furthermore, the energy levels of S1/T1 were calculated to be 2.92/2.72 eV, 2.83/2.69 eV, 2.78/2.76 eV, and 2.79/2.64 eV from the onset of fluorescence and phosphorescence spectra at 77 K (Supplementary Fig. 40), resulting in ΔESTs of 0.20, 0.14, 0.02 eV, and 0.15 eV, respectively. These values exhibited very good consistency with the above theoretical results.
The TADF properties were subsequently evidenced by the transient decay curves of the doped films (Fig. 2b). At 300 K, all the emitters exhibited bi-exponential decays with nanosecond-scale and microsecond-scale components, corresponding to prompt fluorescence and TADF, respectively. DPA-B4 displayed a shortened prompt lifetime (τPF = 1.8 ns) and delayed lifetime (τDF = 2.5 µs), compared to DPA-B2 (2.4 ns / 45.6 µs), DPA-B3 (1.9 ns / 12.5 µs), and Cz-B4 (3.8 ns / 17.4 µs). To provide a quantitative understanding of the kinetic processes, key rate constants such as singlet radiative decay (kr), non-radiative decay (knr), and kRISC were determined (Supplementary Table 3, refer to Supplementary Information for calculation details). Consistent with its highest ФPL, the DPA-B4-doped film exhibited the highest kr of 8.47 × 107 s-1 and the lowest knr of 0.86 × 107 s-1. The kRISCs were determined to be 0.10 × 106, 0.44 × 106, 2.60 × 106 and 0.11 × 106 s-1 for DPA-B2, DPA-B3, DPA-B4 and Cz-B4, respectively. Notably, DPA-B4 displayed the fastest RISC process, which was 26-fold higher than that of DPA-B2. These values surpass those reported for any previously known heavy-element free MR-TADF molecules, even with more extended MR frameworks25, and are on par with those of selenium MR-TADF molecules31. This is a clear consequence of its near-zero ΔEST and large SOC values, origined from the twisted molecular skeleton. The superiority of DPA-B4 in both singlet and triplet exciton harvesting would significantly enhance EL efficiency and mitigate roll-off issue at high brightness in devices.
The distinctive molecular configurations of the four compounds imparted a noteworthy degree of anisotropy in the doped films, resulting in horizontal molecular orientation ratios (Θ//) of 91%, 94%, 97%, and 97% for DPA-B2, DPA-B3, DPA-B4, and Cz-B4, respectively, as elucidated by the angle-dependent p-polarized photoluminescence spectra (Supplementary Fig. 44). The general upward trend in these ratios demonstrated that the presence of larger fused ring systems promoted a horizontal dipole orientation due to the expansion of the molecular plane, wherein the transition dipole moment resides. These high Θ// values were expected to positively impact the optical out-coupling factor (ηout) when compared to common molecules with randomly oriented dipoles, consequently enhancing the EQEs of the devices.
Electroluminescent performance
Motivated by the excellent photophysical parameters demonstrated by these deep-blue emitters, their potential electroluminescence (EL) performance was further assessed in devices. The OLEDs were fabricated with the following optimized configuration: indium tin oxide ITO/dipyrazino[2,3-f:2′,3′-h]quinoxaline-2,3,6,7,10,11-hexacarbonitrile (HAT-CN, 5 nm)/1,1-bis((di-4-tolylamino)phenyl)-cyclohexane (TAPC, 30 nm)/tris(4-carbazolyl-9-ylphenyl)amine (TCTA, 15 nm)/1,3-di(9H-carbazol-9-yl)benzene (mCBP, 10 nm)/99 wt% DOBNA-OAr: 1 wt% emitter (20 nm)/2,8-bis(diphenylphosphoryl)dibenzo[b,d]furan (PPF, 15 nm)/1-(4-(10-([1,1′-biphenyl]-4-yl)anthracen-9-yl)phenyl)-2-ethyl-1H-benzo[d]-imidazole (ANT-BIZ, 30 nm)/8-hydroxyquinolinolato-lithium (Liq, 2 nm)/aluminum (Al, 100 nm). HAT-CN was used as a hole injection layer, TAPC and TCTA functioned as the hole transporting layers, and ANT-BIZ were employed as electron-transport layers, respectively. mCBP and PPF served as exciton-blocking layers, and Liq and Al were acted as the electron injection layer and cathode, respectively. The chemical structures of the organic materials, the energy levels diagram and the EL performance are depicted in Fig. 3, Supplementary Fig. 45 and detailed parameters were provided in Table 2.
All the fabricated devices displayed nearly identical turn-on voltages (Von) of 3.6 V and high maximum luminescence (Lmax) over 10000 cd m-2. The EL spectra maintained stable at different voltages (from the turn-on voltage to 9 V) for all devices measured, indicating that the exciton recombination zone was almost consistent at different voltages. The maximum emission peaks were recorded at 442, 448, 457 and 456 nm for DPA-B2, DPA-B3, DPA-B4 and Cz-B4, with the associated FWHM values of 30, 15, 13, and 26 nm, respectively, which were consistent with the trend observed in toluene solution and doped film. The CIE coordinates of the four devices were (0.150, 0.052), (0.150, 0.050), (0.140, 0.050) and (0.139, 0.077), respectively, with the DPA-Bn series approaching the BT.2020 standard for primary blue emission. Of particular note, the device based on DPA-B4 exhibited the narrowest FWHM value of 13 nm, which is not only the smallest value reported for OLEDs to date (Supplementary Fig. 46), but also smaller than that of quantum-dot and perovskite light-emitting diodes53.
Table 2
The EL data of the blue OLEDs.
Device | λELa [nm] | Lmaxb [cd m-2] | CEc [cd A-1] Max (Average) | PEd [lm W-1] Max (Average) | EQEe [%] Max (Average) | EQEf [%] 100/1000 | FWHMg [nm] | CIEh [x, y] |
DPA-B2 | 442 | 11790 | 10.5 (9.6 ± 0.9) | 8.5 (7.4 ± 1.0) | 21.9 (20.2 ± 1.8) | 15.6/10.0 | 30 | (0.150, 0.052) |
DPA-B3 | 448 | 10960 | 15.8 (15.2 ± 0.5) | 12.7 (12.2 ± 0.5) | 30.6 (29.6 ± 0.9) | 25.8/17.2 | 15 | (0.150, 0.050) |
Cz-B4 | 456 | 11843 | 15.3 (14.7 ± 0.8) | 10.9 (10.6 ± 0.9) | 21.3 (20.4 ± 1.1) | 19.1/14.3 | 26 | (0.139, 0.077) |
DPA-B4 | 457 | 10081 | 17.8 (16.2 ± 0.7) | 12.9 (12.7 ± 0.6) | 31.5 (30.6 ± 0.5) | 28.7/23.6 | 13 | (0.140, 0.050) |
DPA-B4 (HF) | 460 | 39329 | 43.3 (41.4 ± 1.6) | 37.7 (34.8 ± 2.7) | 44.7 (42.6 ± 1.9) | 43.9/38.5 | 16 | (0.150, 0.110) |
DPA-B4 (Tandem) | 458 | 56246 | 77.3 (73.9 ± 3.1) | 33.7 (31.7 ± 1.8) | 74.5 (70.4 ± 2.9) | 73.3/65.3 | 17 | (0.142, 0.116) |
aElectroluminescence peak wavelength. bMaximum luminance. cMaximum current efficiency. dMaximum power efficiency. eMaximum external quantum efficiency. fExternal quantum efficiency at 100, 1000 cd m-2. gFull-width at half-maximum. hCommission Internationale de l’Eclairage coordinates. HF represents hyperfluorescent device. Tandem represents the two-unit tandem OLED.
Except for the superb color purity, the devices based on DPA-B4 also exhibited considerable maximum EQE and current efficiency (EQEmax and CEmax) of 31.5% and 17.8 cd A-1, respectively, outstripping those based on DPA-B2 (21.9%, 10.5 cd A-1), DPA-B3 (30.6%, 15.8 cd A-1) and Cz-B4 (21.3%, 15.3 cd A-1), which was a clear consequence of the superior ФPL and Θ// values of DPA-B4. We highlight that the efficiency of the DPA-4BNO-based device was the highest among the reported MR-TADF based OLEDs with CIEy ≤0.05 (Fig. 3f, Supplementary Table 5). Of note, because the CE also depends on the emission profile, the reduction of high-energy component would lead to a further increase in CE as the brightness is corrected by the human eye sensitivity function. This explains a more drastic increase of CEmax (70%) than EQEmax (44%) of DPA-B4 compared with DPA-B2 due to redshifted emission onset in DPA-B4, reflecting the merit of ultranarrow emission band of DPA-B4 in promoting device efficiency. Moreover, the efficiency roll-off of the DPA-B4-based device (28.7% at 100 cd m-2 and 23.6% at 1000 cd m-2) was found to be smaller than that of DPA-B2 (15.6% at 100 cd m-2 and 10.0% at 1000 cd m-2), DPA-B3 (25.8% at 100 cd m-2 and 17.2% at 1000 cd m-2) and Cz-B4 (19.1%, 14.3 cd A-1). This can be attributed to the very short triplet state lifetime (2.5 µs) and fast RISC process (kRISC ~ 2.60 × 106 s-1) for DPA-B4, which reduce the triplet exciton density and inhibit bimolecular exciton deactivations through triplet-triplet annihilation (TTA) and singlet-triplet annihilation (STA) processes.
To further improve the device performance and suppress the efficiency roll-off, a sensitization strategy was employed for the champion emitter DPA-B4 considering its short prompt emission lifetime. A well-known TADF emitter reported in ref.54, 5-(3,11-dimethyl-5,9-dioxa-13b-boranaphtho[3,2,1-de]anthracen-7-yl)-10,15-diphenyl-10,15-dihydro-5H-diindolo[3,2-a:3',2'-c]carbazole (mMDBA-DI), was selected as a sensitizer by taking advantage of its well-matched PL spectra with the absorption spectra of DPA-B4 (Supplementary Fig. 47), high ΦPL and fast RISC process. As illustrated in Fig. 4, despite of slight bathochromic shifted CIE of (0.150, 0.110) due to increased host polarity, the DPA-4BNO-based device maintained a very small FWHM of 16 nm. Meanwhile, this device displayed an ultrahigh EQEmax of 44.7% and CEmax of 43.3 cd A-1, with mitigated efficiency decline at high brightness: EQE values of 43.9% at 100 cd m-2, and 38.5% at 1000 cd m-2, corresponding to EQE drops of merely 1.8% and 13.8%, respectively. This remarkable improvement in efficiency and reduction in efficiency roll-off can be attributed to more efficient triplet harvesting facilitated by the sensitizer and rapid exciton consumption through a long-range Förster resonance energy transfer (FRET) pathway. While DPA-B4 primarily served as a fluorescence emitter in this hyperfluorescence device, and the triplet upconversion of the sensitizer was typically the rate-determining step in exciton dynamics, triplet population on the terminal emitter remained possible10,55,56. This suggests that an accelerated RISC process for the terminal emitter should be advantageous even in the presence of a sensitizer. For comparison purposes, devices DPA-B3 and Cz-B4, based on the same device structure, exhibited not only inferior EQEmax but also more pronounced roll-off, underscoring the advantage of the higher kRISC of DPA-B4 (Supplementary Figs. 48–49 and Supplementary Table 4).
Inspired by the above results, a two-unit tandem OLED was constructed to fully leverage the narrow FWHM and high efficiency of DPA-B4 (Fig. 4d, 4f)57. The resulting EL spectrum closely mirrored that of the single-unit device, featuring a FWHM of 17 nm. Nonetheless, the maximum brightness was increased by 43%, and the required current density was also reduced at the same brightness. The tandem OLED achieved EQEmax and CEmax values as high as 74.5% and 77.3 cd A-1, respectively, representing an increase of approximately 1.7 and 1.5 times compared to the single EL unit device. Impressively, this tandem device maintained EQEs exceeding 40% across a wide luminance range of 0–15,000 cd m-2, adequately covering the luminance requirements for flat-panel displays and general lighting applications. This exceptional performance can be attributed to the efficient charge generation layer and highly effective TADF processes within the tandem OLED architecture.