
- SJ_Zhang
- Jun. 15, 2025
Abstract
Energy level alignment of frontier molecular orbital (FMO) is essential for controlling charge carrier and exciton dynamics in organic light-emitting diodes (OLEDs). However, multiple resonance (MR) emitters with exceptional narrowband luminescence typically suffer from inadequate FMO levels. Herein, a conventional blue MR prototype with a shallow highest occupied molecular orbital (HOMO) level of −5.32 eV is initially employed to reveal the charge carrier and exciton dynamics. Severe hole trapping by its shallow HOMO significantly hinders its transport. More importantly, trapped carriers induce direct exciton formation and recombination at MR emitters in a hyperfluorescent system, leading to triplet accumulation in MR emitters. To resolve these issues, a proof-of-concept wavefunction perturbation strategy is proposed by incorporating cyano motifs at peripheral sites of MR backbone to adjust the energy levels. This approach significantly shifts HOMOs of 0.36 and 0.51 eV without compromising colour purity. The derivative substituting meta-boron position (mCNDB) exhibits a pure-blue emission peaking at 459 nm with a narrow bandwidth of 13 nm. The detrimental carrier trapping effect is eliminated, enhancing external quantum efficiency to exceeding 23%, maintaining around 20% at 1000 cd m−2, and improving the device stability.
Introduction
Comprehending and manipulating the energy level alignments of functional molecules is essential for determining the behaviours of charge carriers and excitons in optoelectronic devices. This is particularly crucial for blue organic light-emitting diodes (OLEDs). Simultaneously achieving high efficiency, stability, and colour purity has been the most challenging topic since the first two-layer OLED was reported by C. W. Tang and S. A. VanSlyke in 19871. Regarding efficiency, the fundamental breakthrough of the charge-transfer (CT) state has allowed the creation of numerous efficient emitters, induced by metal-ligand CT and thermally activated delayed fluorescence (TADF) molecules2,3,4. By versatilely utilizing spin-forbidden triplets for radiation, a unity internal quantum efficiency, as well as a high external quantum efficiency (EQE), has been realized with adequate device architectures5,6,7,8. Furthermore, by leveraging the multiple resonance (MR) effect, an extremely narrow spectral width of less than 20 nm can be achieved by MR-TADF molecules by suppressing the intramolecular vibrations and relaxation in their excited states9,10,11,12.
Nevertheless, device durability under continuous operation, which is highly dependent on charge carrier and exciton dynamics, remains the primary obstacle. Substantial efforts have been devoted to extending its longevity by optimizing the intrinsic properties of each component material. These efforts include reinforcing molecular rigidity to enhance photostability13; differentiating orbital angular momentum or incorporating heavy atoms to accelerate the triplet upconversion rate14,15,16; deuterium substitution to suppress high-frequency vibrations17,18; and constructing a sterically shielded structure to reduce detrimental Dexter energy transfer19. Unlike direct photon absorption under photoexcitation, the performance of electrically-driven devices is heavily associated with complicated charge carrier and exciton behaviours, e.g., single carrier injection20, transport21, trapping22,23, recombination24,25, Förster/Dexter energy transfer26, as well as multiple channels of exciton annihilations27,28,29. In blue OLEDs with wide energy gaps (Eg > 2.6 eV), the alignment of frontier molecular orbitals (FMOs) energy levels between adjacent layers plays a dominant role in determining the carrier dynamics and recombination pathways. For the representative blue MR prototypes like DABNA-2 and ν-DABNA, featuring single- and binary-boron units, respectively, their highest occupied molecular orbital (HOMO) levels are localized at around −5.4 eV9,30. These energy levels are much shallower than those of traditional blue phosphorescent or TADF emitters, as well as those of host matrices or assistant dopants (around −6.0 eV)31, leading to significant carrier traps under electrical excitation. To gain a comprehensive view, the HOMO levels of the reported blue boron-containing MR emitters are summarized in Fig. 1a and Supplementary Table 1. It is evident that the HOMOs of most MR emitters are shallower than −5.6 eV, highlighting the dilemma of inadequate FMOs energy levels. Although given a few examples of blue MR emitters with HOMOs deeper than −5.6 eV, these studies primarily focused on structural modifications aimed at modulating photophysical properties, like reverse intersystem crossing process, emission wavelength, and spectral bandwidth32,33. Charge carrier trapping or blocking by inadequate energy levels has been found to impact device performance, while the underlying mechanism and effective strategies for adjusting the energy levels have not received sufficient attention34,35.
Fig. 1: Chemical structures, theoretical calculation, and photophysical properties of studied molecules.
a Summarized HOMO energy levels of reported blue boron-containing MR emitters. b Molecular design strategy and chemical structures. c Fluorescence spectra. d HOMO/LUMO distributions. e Experimental FMOs energy level alignments. f Transient PL decay curves of CzDB, mCNDB, and pCNDB in solution.
In this work, we found that the energy levels of MR emitters at a low doping concentration (i.e., 1 wt%) can significantly influence the charge carrier and exciton dynamics in devices. The shallow HOMO energy level of the blue MR prototype, CzDB (EHOMO = −5.32 eV), induces severe hole trapping, which substantially hinders the charge carrier mobility, thus resulting in unideal current density?voltage (J?V) characteristics. Investigations into the transient delayed electroluminescent (EL) characteristics reveal that trapped charge carriers would affect the localization of the recombination zone. More importantly, these trapped carriers enable the determination of the exciton generation pathway, indicating that excitons are prone to be directly formed at MR emitters rather than at assistant dopants or hosts in hyperfluorescent (HF) devices26. Consequently, the inadequate, shallow energy level of blue MR emitters accelerates device degradation. To address this issue, an effective strategy is proposed to modulate the FMO energy levels through wavefunction perturbation of the MR backbone. In specific, the electron-accepting cyano-groups are substituted at the peripheral positions of the MR backbone (CzDB). Due to the limited size of the cyano fragment, the wavefunction distribution in the molecular skeleton is not significantly reorganized, thus maintaining the dominance of the MR effect. This approach offers the advantages of (i) maintaining a narrow spectral bandwidth with high colour purity, and (ii) localizing the spectral wavelength to guarantee blue emission. Notably, although the photophysical properties of the modified molecules remain consistent with those of the pristine CzDB, the energy levels of the meta- and para-substituents, mCNDB (EHOMO = −5.68 eV) and pCNDB (EHOMO = −5.83 eV), are much deeper than that of CzDB (EHOMO = −5.32 eV). Consequently, the remarkable FMO energy level adjustment (?EHOMO of 0.36 and 0.51 eV, respectively) leads to higher EQEs, reduced efficiency rolloff, and optimized operational stability in blue OLEDs.
Results
Molecular design and energy level tuning
As illustrated in Figs. 1b and 1c, double-boron-embedded CzDB is selected as the MR prototype because of its blue narrowband emission (λpeak = 471 nm, FWHM = 17 nm)36. The structural and electronic properties of CzDB were initially investigated using density functional theory (DFT) and time-dependent DFT (TD-DFT) calculations at the PEB0/6-31 G(d,p) level (Fig. 1d). As expected, HOMO and the lowest unoccupied molecular orbital (LUMO) wavefunctions are mainly distributed at the discrete nitrogen and boron atoms, resulting in short-distance MR effect within the robust skeleton, i.e., narrowband emission. Moreover, the atomic HOMO/LUMO separation induces a small experimental singlet-triplet splitting (ΔEST = 0.11 eV), triggering the TADF property (Supplementary Fig. 1). However, the simulated low-lying HOMO level of CzDB is localized at −5.12 eV, which is consistent with traditional MR molecules (summarized in Fig. 1a and Supplementary Table 1). To resolve this dilemma, electron-accepting cyano motifs are employed to substitute at the meta- and para-positions of the central boron atoms in the periphery BN backbone. As depicted in Fig. 1d and Supplementary Fig. 2, the simulation suggests that the limited size of the cyano fragments would barely influence the molecular rigidity and the wavefunction distribution at the BN core. Thus, the MR effect still governs the electronic properties of both mCNDB and pCNDB. These molecules exhibit an extremely small root mean square displacement (RMSD) of around 0.030 Å between the optimized singlet excited state (S1) and ground state (S0) configurations, indicating negligible structural relaxation under the excited state with cyano substituents. Furthermore, Huang?Rhys (HR) factors for the S1 → S0 transition are calculated. The dominant vibration modes of these molecules are primarily from the out-of-plane distortion of the CzDB core at low frequency, whereas cyano units barely contribute to the structural vibrations (Supplementary Fig. 3). As a result, the spectral widths of both mCNDB and pCNDB are as narrow as 13 and 15 nm, respectively, preserving the high colour purity in pure blue (459 nm) to sky blue (483 nm) emission (Supplementary Fig. 4). Gaining deeper insight into the function of cyano-units, the varying substitution positions lead to different effects on their electronic characteristics. In mCNDB, cyano is localized at the meta-position of boron, as well as the para-position of nitrogen, primarily contributing to the HOMO distribution. Thus, this weakens the electron-donating ability of the skeleton, resulting in a deeper HOMO level (−5.67 eV) than that of the pristine CzDB (−5.12 eV). In contrast, incorporating the cyano unit in pCNDB enhances the conjugation coupling with the para-B atoms, thus mainly participating in the LUMO distribution. The reinforced electron-accepting strength induces a much deeper LUMO level (−2.09 eV) than that of CzDB (−1.44 eV), along with a deeper HOMO level (−5.64 eV), due to the electronic effect in the highly conjugated molecular system37. Noteworthy, despite the significant shifts in their HOMO/LUMO levels, wide bandgaps of 3.68, 3.76, and 3.55 eV are reserved for CzDB, mCNDB, and pCNDB, respectively, with narrowband blue luminescence. To validate the accuracy of the simulated HOMO/LUMO energy levels, cyclic voltammetry (CV) and ultraviolet photoelectron spectroscopy were used to measure the energy levels in the solution and the solid state, respectively (Supplementary Figs. 5?7). The HOMO levels of CzDB, mCNDB, and pCNDB are determined to be −5.32, −5.68, and −5.83 eV, respectively, which are consistent with the prediction of the DFT calculation.
The photophysical characteristics of CzDB, mCNDB, and pCNDB in solid-state films are successively investigated. The blend films are composed of 1.5 wt% emitters in a 3,3’ -bis(carbazol-9-yl)biphenyl (mCBP) host matrix. The photoluminescent quantum yields (PLQYs) of the mCNDB- and pCNDB-based films are 86% and 89%, respectively, which are slightly higher than that of the CzDB-based film (85%). Furthermore, distinct delayed components are observed in these blend films from the transient PL decay profiles, confirming the TADF nature of the three emitters (Supplementary Figs. 8 and 9). In specific, the larger ΔEST values of mCNDB (0.16 eV) and pCNDB (0.15 eV) than that of CzDB (0.10 eV) in the solid-state films (Supplementary Fig. 10), which is consistent with the singlet-triplet energy difference measured in toluene. It can be partially explained by a larger wavefunction overlap integral (0.53, 0.57, and 0.60 for CzDB, mCNDB, and pCNDB, respectively) after incorporating cyano-units. Therefore, the reverse intersystem crossing (RISC) rates of the mCNDB- and pCNDB-based films (2.7 × 103 and 3.3 × 103 s−1, respectively) are slightly slower than that of the CzDB-based film (6.4 × 103 s−1). Additionally, the photostability of the three molecules in a degassed toluene solution was evaluated. The comparable PL intensity degradation under a constant excitation power of 5 mW cm−2 indicates that the incorporation of cyano units barely affects the molecular stability (Supplementary Fig. 11). The detailed photophysical data of the three molecules are summarized in Supplementary Table 2. Although the photophysical characteristics of mCNDB and pCNDB are comparable to those of CzDB, their energy level variation would sufficiently influence the electrical properties.
Charge carrier dynamics in non-sensitized OLEDs
To evaluate the impact of varied energy levels on their EL properties, OLEDs incorporating a emitting layer (EML) with the host-guest system are fabricated as the following architecture (Supplementary Fig. 12): indium tin oxide (ITO), 1,4,5,8,9,11-hexaazatriphenylene hexacarbonitrile (HAT-CN, 10 nm), N4,N4’-di(naphthalen-1-yl)-N4,N4’-diphenyl-[1,1’-biphenyl]−4,4’-diamine (NPD, 20 nm), 9-phenyl-3,6-bis(9-phenyl-9H-carbazol-3-yl)−9H-carbazole (Tris-PCz, 10 nm), mCBP (5 nm), 1.5 wt% emitters: mCBP (30 nm), SF3-TRZ (10 nm), 30 wt% 8-quinolinolato lithium (Liq): SF3-TRZ (30 nm), Liq (2 nm), and aluminium (Al, 100 nm). Specifically, mCBP is employed as the electron-blocking layer and host matrices, which governs the dynamic behaviours of charge carriers. Moreover, the adjacent SF3-TRZ serves as the hole-blocking layer, which dominates electron injection. The HOMO and LUMO levels of the compounds in the charge carrier blocking layers and EML are illustrated in Fig. 2a. The characteristics of current density-driving voltage-luminance (J-V-L) are exhibited in Fig. 2d. In contrast of mCNDB- and pCNDB-based OLEDs, the CzDB-based device demands a much larger driving voltage. Noticeably, HOMO/LUMO levels (−5.32/−2.74 eV) of CzDB are much shallower than those of either mCNDB (−5.68 and −3.07 eV) or pCNDB (−5.83 and −3.34 eV), which are responsible for the electrical properties. To clarify the mechanism of different J-V characteristics, electron-only and hole-only devices (EOD and HOD) are respectively designed to independently reveal the single charge carrier dynamics under electrical excitation without the hole-electron recombination process (Fig. 2b, c, and Methods). At the electron side, LUMO levels of SF3-TRZ and the adjacent mCBP matrix were reported to be −3.10 and −2.74 eV, respectively, leading to a distinct energy gap (0.36 eV) for insufficient electron injection21. Consequently, EOD based on the mCBP neat film exhibits a large electron injection voltage (Vei, ~5.5 V) and poor electron transport, as shown in Fig. 2e. After doping 1.5 wt% CzDB in mCBP matrix, the Vei (~5.0 V) is slightly reduced, and the electron transport capability is enhanced. In contrast, mCNDB- and pCNDB-based EODs show the opposite characteristic of a reduced Vei (~3.0 V), while a much poorer electron flow upon increasing the driving voltage. The variation in EODs can be ascribed to the LUMO alignment of the CzDB derivatives, where the LUMO levels of mCNDB (−3.07 eV) and pCNDB (−3.34 eV) are equivalent to that of SF3-TRZ ( −3.10 eV), leading to a non-barrier electron injection. Conversely, CzDB, with a shallow LUMO (−2.74 eV), is close to mCBP, maintaining a relatively large energy gap with SF3-TRZ. In addition, HOD undergoes different dynamic behaviours (Fig. 2c and f). Either mCNDB- or pCNDB-based HODs have an identical J-V property to that of mCBP-based HOD, suggesting the hole transport is governed by mCBP matrices. However, the CzDB-based HOD shows much-suppressed hole transport, as shown in Fig. 2f. Notably, CzDB has the shallowest LUMO level of −5.32 eV, and the substantial energy gap (?EHOMO = 0.76 eV) with mCBP makes it act as the deep wells to trap the mobile holes. Upon applying a reverse bias voltage (−10 V) after switching off the pulsed EL excitation (pulse width of 100 μs), an intense EL spike appears in the CzDB-based OLED, indicating the hole trapping effect in CzDB molecules with shallow energy levels (Fig. 2g)38,39.
Fig. 2: Device performance and charge carrier dynamic of non-sensitized OLEDs.
Energy level alignment of EBL/EML/HBL in (a) OLED, (b) EOD, and c HOD based on 1.5 wt% CzDB, mCNDB, and pCNDB doped in mCBP blend films. Current density-to-driving voltage curves for (d) OLED, (e) EOD, and (f) HOD based on 1.5 wt% CzDB, mCNDB, and pCNDB doped into mCBP blend films. g Transient EL decay characteristics after applying a reverse bias voltage of −10 V. h Transient EL delayed components after switching off the driving voltage. i EL spectra of the corresponding OLEDs at 1000 cd m−2.
Furthermore, the varied charge carrier dynamics alter exciton generation and recombination. To accurately determine the recombination zone, a deep-red emitter, bis(1-phenylisoquinoline)(acetylacetonate)iridium(III) [Ir(piq)2acac], is used as an indicator for the exciton density at different positions (5, 15, 25, and 30 nm) in the EML. The thickness of Ir(piq)2acac is controlled at 0.3 nm to eliminate the effect on charge carrier transport. As shown in Supplementary Fig. 13, the deep-red spectra from Ir(piq)2acac with the wavelength maximal at 635 nm is proportional to the probability of charge carrier recombination. Therefore, it is found that the recombination zone of mCNDB- and pCNDB-based OLEDs is primarily localized at the EML/SF3-TRZ interface (between positions 4 and 3). Instead, an extensive shift of the recombination zone to the hole injection side occurs in the CzDB-based device. Moreover, it becomes broader from positions 1 to 3, indicating that trapped single carriers can affect the distribution of the recombination zone. Although EQEs exceeding 10% are achieved for both mCNDB- and pCNDB-based OLEDs operating at a low current density (<10−2 mA cm−2), severe efficiency rolloff takes place because of their long triplet lifetime (~500 μs, Fig. 2h) and the associated exciton annihilation when increasing the current density (Supplementary Fig. 14). Particularly, mCNDB- and pCNDB-based OLEDs show narrowband spectra with FWHMs of 17 nm at pure-blue (466 nm) and sky-blue (489 nm) regimes.
Charge carrier dynamics in hyperfluorescent (HF) OLEDs
To further enhance the device performance, a sky-blue TADF assistant HDT-1 is utilized to construct the hyperfluorescent (HF) system40. The doping concentration of HDT-1 is controlled at 20 wt%, which would remarkably affect the charge carrier dynamics in the devices (Fig. 3). HOMO and LUMO levels of HDT-1 are measured at −5.98 and −3.31 eV, respectively (Supplementary Fig. 15). Initially, HOD based on a 20 wt% HDT-1: mCBP blend film is fabricated, and the J?V curve is identical to that of the mCBP-based HOD, indicating that the HDT-1 dopant does not affect the hole injection and transport capabilities (Fig. 3a, d). As expected, since the HOMO level of HDT-1 is as deep as that of the mCBP host matrix, the hole injection and transport characteristics of the corresponding HF devices, after doping 1.5 wt% mCNDB and pCNDB with relatively deeper HOMO levels, are identical to those of the host-guest devices (Fig. 3a, d). Notably, the hole flow in the CzDB-based HF HOD remains hindered, which is attributed to the significant HOMO gap (0.66 eV) between CzDB and HDT-1. In contrast, the electron dynamics undergo substantial changes. As HDT-1 has a much higher doping concentration (20 wt%) and a deeper LUMO level (−3.31 eV) than SF3-TRZ (−3.10 eV, Fig. 4b), electron injection from SF3-TRZ to the EML becomes easier, thus reducing the corresponding Vei to 2.0 V. Particularly, the electron flow density is inversely proportional to the LUMO gap between the emitters and HDT-1. Thus, mCNDB- and pCNDB-based HF EODs with smaller LUMO gaps (?ELUMO = 0.24 and 0.03 eV) exhibit optimized electron transport, whereas CzDB-based HF EOD demonstrates slightly poor electron transport because of its large ?ELUMO of 0.57 eV. Subsequently, the corresponding HF OLEDs are fabricated. As anticipated, the CzDB-based HF device still exhibits a worse J?V?L characteristic due to the hole-trapping effect, which can be confirmed by a transient EL spike after applying the reverse bias (−10 V, Fig. 3g) to the device. In addition, the impedance spectroscopy indicates that the trapped charge carriers in CzDB dopants in the HF device notably results in a much higher recombination resistance than that of mCNDB- and pCNDB-based HF devices (Supplementary Fig. 16), highlighting the effect by the FMO energy level alignment. The EL spectra of HF OLEDs are demonstrated in Fig. 3h, with emission peaks at 478 nm (CzDB-based), 466 nm (mCNDB-based), and 489 nm (pCNDB-based), ranging from pure-blue to sky-blue. Additionally, narrow FWHMs of around 20 nm are obtained in these devices. Benefiting from the more balanced carriers, mCNDB- and pCNDB-based HF OLEDs exhibit maximum EQEs exceeding 23% while maintaining around 20% at a high luminance of 1000 cd m−2 (Fig. 3i). By comparison, a lower maximum EQE of 20% is observed in the CzDB-based HF OLED, accompanied by more severe efficiency rolloff (24.8%) at 1000 cd m−2, which is larger than those (19.3% and 12.9%) of mCNDB- and pCNDB-based HF devices (Supplementary Table 3).
Fig. 3: Device performance and charge carrier dynamic of HF OLEDs.
Energy level alignment of EBL/EML/HBL in (a) HOD, (b) EOD, and (c) HOD based on CzDB, mCNDB, and pCNDB doped in HF (HDT-1: mCBP) blend films. Current density-to-driving voltage curves for (d) OLED, (e) EOD, and (f) HOD based on CzDB, mCNDB, and pCNDB doped in HF (HDT-1: mCBP) blend films. g Transient EL decay characteristics after applying the reverse bias voltage. h. EL spectra of the corresponding HF OLEDs. i EQE?current density curve for HF devices. The insets are photographs of the operating OLEDs at 1000 cd m−2.
Fig. 4: Exciton dynamic and operational lifetime of HF OLEDs.
Illustration of the exciton generation and energy transfer processes in (a) CzDB-, (b) mCNDB-, and (c) pCNDB-based HF devices (the doping concentrations of emitter and HDT-1 are 1.5 and 20 wt%, respectively). Transient EL delayed emission properties of host-guest and HF OLEDs based on (d) CzDB-, (e) mCNDB-, and (f) pCNDB. (g) Operational EL stability of CzDB-, mCNDB-, and pCNDB-based HF OLEDs. The initial luminance is 1000 cd m−2.
Exciton generation and recombination
As the performance variation of the HF devices is based on CzDB and its derivatives, it is suggested that charge carrier trapping significantly alters the exciton generation and recombination process. The incorporation of HDT-1 results in a significant change in the electron injection and transport properties, which subsequently influences the exciton recombination behaviours. Similar to the host-guest devices, the recombination zone distribution in HF devices is experimentally determined by inserting an ultra-thin layer (0.3 nm) of a deep red indicator [Ir(piq)2acac] at different positions (5, 15, 25, and 30 nm) in the EML of the HF devices. As illustrated in Supplementary Fig. 17, at a low current density (~0.01 mA cm-2), the maximal probability of exciton recombination in all HF OLEDs occurs at the EBL/EML interface, which is consistent with the enhanced electron injection and transporting capability in EODs. However, as the current density increases to 1 mA cm-2, the recombination zone of mCNDB- and pCNDB-based HF OLEDs shifts toward the SF3-TRZ side due to the reduced electron transport. Despite the worse electron transport capability in CzDB-based HF devices than in mCNDB- and pCNDB-based HF devices, the drift of its recombination zone to the electron injection side is slower. In other words, the exciton density at positions 1 and 2 remains dominant. Combined with the fixed recombination zone distribution in the CzDB-based host?guest device, these abnormal characteristics indicate that the shallow HOMO level not only blocks charge carrier migration but also affects the exciton generation pathways. As illustrated in Fig. 4a−c, in the cases of mCNDB and pCNDB with deeper HOMO energy levels, exciton generation in the corresponding HF devices is primarily localized at HDT-1 molecules, followed by sufficient Förster resonant energy transfer (FRET) to the terminal emitters for radiation. In contrast, the trapped holes are expected to induce direct exciton formation at CzDB molecules with a shallow HOMO level instead of the HDT-1 assistant dopants. To verify this hypothesis, systematic investigations into the transient EL delayed characteristics of OLEDs based on CzDB, mCNDB, and pCNDB are conducted. As can be seen in Fig. 4d−f, the delayed emissive components of both mCNDB- and pCNDB-based HF devices are much smaller than those of their host-guest devices, indicating that a large proportion of excitons are generated at the HDT-1 molecules, whereas the reduced proportion of triplets at mCNDB and pCNDB is attributed to the Dexter energy transfer from the triplet state of HDT-1. In contrast, the HF OLED based on CzDB with a shallow HOMO energy level exhibits different characteristics. The delayed emissive component of the HF device is identical to that of its host-guest OLED (Fig. 4f), demonstrating direct exciton generation at the terminal emitters rather than the HDT-1 sensitizers. Therefore, the increased density of long-lived triplets at CzDB molecules is likely to cause exciton quenching, which is consistent with its lower EQE and more severe efficiency rolloff compared to mCNDB- and pCNDB-based HF devices. More importantly, direct exciton generation at the terminal emitters is detrimental to the device durability. As a result, although these three molecules possess equivalent photostability and FRET efficiency (5×106 s−1) in HF blend films (Supplementary Fig. 18-20. Table 4), pure-blue mCNDB- and sky-blue pCNDB-based HF OLEDs exhibit much optimized operational EL stability (Fig. 4g) with exceeding two-fold enhancement under a high luminance of 1000 cd m−2, highlighting the importance of precious energy level modulation for achieving highly stable blue OLEDs.
Discussion
Numerous efforts have been made to stabilize blue OLEDs. In particular, in advanced HF systems, excitons are expected to initially generate efficient phosphors or TADF sensitizers to manage triplet excitons. Therefore, optimizing the intrinsic photophysical properties of phosphors, TADF assistant dopants, and terminal emitters, such as increasing their triplet energy levels, reinforcing their molecular stability, and accelerating the triplet upconversion rate, has become mainstream. However, unlike photoexcitation, the pathways for exciton formation under electrical excitation are critical yet often overlooked.
This work demonstrated that charge carrier trapping by terminal emitters significantly impacts the charge carrier mobility and determines the pathways of exciton generation. A blue MR molecule, CzDB, with a shallow energy level (EHOMO = −5.32 eV), is chosen as the prototype because its distinct delayed lifetime difference from that of HDT-1 makes it easier to distinguish the sources of the delayed components. In CzDB-based HF devices, it is concluded that the free charge carriers would directly recombine at the terminal emitters rather than the TADF sensitizers, resulting in unsatisfactory device performance. To resolve this issue, a strategy of wavefunction perturbation in blue MR backbones is proposed. The incorporation of electron-withdrawing cyano units maintains the narrowband blue emission (FWHM = 13 and 15 nm, respectively) while substantially deepening the HOMO/LUMO levels (−5.68 and −5.83 eV, respectively). Consequently, severe hole trapping and direct exciton recombination issues caused by MR emitters are circumvented, leading to a higher EQE exceeding 23%, a suppressed efficiency rolloff, and optimized operational durability in blue OLEDs.
As discussed, the prototype molecule CzDB that we chose is aimed at clarifying the pathways of exciton generation and recombination. However, its intrinsic limitations, like a slow RISC rate (~103 s−1), imped the realization of the high performance. Given that the energy levels of MR molecules are generally shallow (Fig. 1a), charge carrier trapping-induced drawbacks prevalently occur in the corresponding devices. Thus, we believe that the wavefunction perturbation strategy enables us to feasibly modulate the energy levels of other advanced MR platforms, such as ν-DABNA9, V-DABNA41, and ω-DABNA42, to advance state-of-the-art blue OLEDs.
Methods
Material
All reagents were purchased from commercial sources and used without further purification (Supplementary Fig. 21). 1H NMR and13C NMR spectra were recorded on a Bruker 400/101 MHz spectrometer in deuterium reagent at room temperature. MALDI-TOF mass data were recorded on a Bruker ultrafleXtreme instrument. High-resolution mass spectrometry (HRMS) data were recorded on Thermo Scientific QE plus (Supplementary Figs 22–27).
Photophysical measurement
UV?vis absorption spectra were recorded on a Hitachi U-3900 spectrophotometer. Photoluminescence (PL) spectra were recorded on a Hitachi F-4600 fluorescence spectrophotometer. Transient PL decays were measured on an integrated streak camera system (C4334, Hamamatsu Photonics). The absolute PLQY was recorded on a Hamamatsu Quantaurus-QY quantum yield spectrometer (C13534-11). The angle-dependent PL spectra of the doped films were measured via a molecular orientation characteristic measurement system (C14234-11, Hamamatsu Photonics). The photostability in solution was estimated by using a multichannel spectrometer (PMA-12, Hamamatsu Photonics) under 300–400 nm light (5 mW cm−2) irradiation via xenon light (MAX-303, Asahi Spectra) with a UV light intensity feedback control unit. The photostability of the HF blend films was measured in a nitrogen atmosphere by using a multichannel analyser (C14631, Hamamatsu Photonics, Japan) under 300–400 nm light (25 mW cm−2) irradiation from a filtered Xe lamp (MAX-350, Asahi Spectra, Japan).
Energy level measurements
The highest occupied molecular orbital (HOMO) energy levels of the emitters in the films were determined by atmospheric ultraviolet photoelectron spectroscopy using a photoelectron emission spectrometer (Riken Keiki AC-3). Cyclic voltammetry (CV) curves were recorded on a CHI660e electrochemistry station in an oxygen-free dichloromethane (DCM) solution with tetrabutylammonium hexafluorophosphate (Bu4NPF6) at a scan rate of 0.1 V s−1.
Theoretical calculation
All calculations were performed with the Gaussian 16 programme package43. The equilibrium geometry of the ground state was determined by density functional theory (DFT) using the PBE0 density functional method with a basis set of 6-31 G(d,p). The Huang?Rhys (HR) factors for the S1 → S0 transition were determined with the DUSHIN module in MOMAP (Molecular Materials Property Prediction Package)44. All calculations were performed in the gas phase. The molecular structure, frontier molecular orbital, and electron distribution models were rendered by Multiwfn and VMD45,46.
Hole-only and electron-only devices (HODs and EODs)
Hole-only device structure: ITO/HAT-CN (10 nm)/NPD (20 nm)/Tris-PCz (10 nm)/mCBP (5 nm)/EML (50 nm)/mCBP (5 nm)/TrisPCz (10 nm)/NPD (20 nm)/HAT-CN (10 nm)/Al (100 nm). Electron-only device structure: ITO/Liq (2 nm)/SF3-TRZ: Liq (30 wt%, 30 nm)/SF3-TRZ (10 nm)/EML (50 nm)/SF3-TRZ (10 nm)/SF3-TRZ: Liq (30 wt%, 30 nm)/Liq (2 nm)/Al (100 nm).
Device characterization
The electrical characteristics of the devices were measured using a source metre (Keysight B2911A, Keysight Technologies) and a luminance metre (CS-2000, Konica Minolta, Japan). The lifetimes of all devices are measured using a luminance metre (SR-3AR, TOPCON, Japan) under constant current density driving conditions with an initial luminance of 1000 cd m−2.
Transient electroluminescent (EL) radiation measurements
Pulsed voltages were applied by using a function generator, the light signals were detected via a photomultiplier tube (PMT) module (H10721-01, Hamamatsu Photonics, Japan), and the current signals from the PMT were amplified using a current amplifier (DHPCA-100, Femto, Germany).
Journal
Jul. 13, 2025