Molecular Engineering of Emitters Toward Synergistic Hot‐Exciton Harvesting and Light Outcoupling for High‐Efficiency OLEDs

ABSTRACT

Simultaneously realizing high internal quantum efficiency (IQE) and light outcoupling efficiency (LOE) in OLEDs solely depending on molecular engineering has never been achieved. Here, we design three OLED emitters (DCzPDO, DPhCzPDO, and DPAPDO), all comprising 3‐hydroxypropenone and β‐diketone tautomers. One prominent feature of them is hot‐exciton emission enabled by high‐lying reverse intersystem crossing (hRISC), hence significantly improving material electroluminescence. Interestingly, in DPAPDO, both isomers exhibit hot‐exciton fluorescence, while in the other two materials, only the β‐diketone tautomer demonstrates this feature. On the other hand, these materials possess different degrees of preferential transition‐dipole‐moment alignment, promoting light extraction in OLEDs. Furthermore, their molecules tend to self‐assemble along 1D directions, inducing a quasi‐parallel V‐shaped nanogroove, significantly enhancing light outcoupling in devices. Among them, DPAPDO achieves the highest device performance, reaching a maximum external quantum efficiency (EQE) of 23.7%. DPAPDO comprises the highest concentration of β‐diketone isomer, with both isomers facilitating effective hRISC transitions, leading to a near unity IQE of luminescence. Additionally, DPAPDO displays the highest‐quality film nanostructure and moderate horizontal dipole ratio, thereby achieving a high LOE of 57%. All these factors account for its high OLED performance. The synergistic interplay between electroluminescence and molecular/supramolecular structures paves the way for cost‐effective, scalable, high‐efficiency OLEDs.

Molecular engineering enables a unique two‐by‐two strategy for achieving near unity (∼100%) internal quantum efficiency and high light outcoupling efficiency (57%) in 3‐hydroxypropenone based OLEDs, achieved via synergistic control of molecular photophysics, isomeric equilibrium, dipole orientation, and film nanotopology. The molecular structure governs exciton harvesting while supramolecular assembly optimizes light extraction, paving the way for cost‐effective and scalable high‐efficiency OLEDs.

1. Introduction

Organic light‐emitting diodes (OLEDs) are attracting growing attention for display and lighting applications, owing to many advantages such as low cost, wide viewing angles, light weight, thin/flexible designs, and potential for large‐area processability [1, 2]. However, in traditional fluorophore‐based OLEDs, the internal quantum efficiency (IQE) is capped at 25% because electron‐hole recombination produces singlet and triplet excitons at a 25:75 ratio. Moreover, non‐radiative optical losses in the device, such as total internal reflection (TIR), waveguide (WG) mode confinement, and surface plasmon polariton (SPP) coupling, significantly limit photon extraction in OLED devices, typically with only ∼20% of the generated photons successfully emitted [3, 4, 5, 6]. Therefore, conventional OLEDs generally exhibit external quantum efficiencies (EQEs) not exceeding 5%. To address these challenges and realize high‐efficiency electroluminescence (EL), extensive efforts have been devoted to enhancing exciton harvesting in emitters while reducing optical losses within devices.

To enhance light extraction and reduce optical losses in devices, various strategies have been proposed through elaborate design of device architecture [4] and/or by increasing the proportion of horizontally oriented emitting dipoles via optimized intermolecular stacking [7, 8]. The strategies of device structural design for enhancing photon extraction mainly involve the use of micro‐ or nano‐structured layers, such as microlens arrays [9, 10], scattering films [11], photonic crystal patterns [12], and periodic gratings [13], positioned at the substrate or electrode interface, to redirect trapped light into the escape cone [1, 2, 3, 4, 5, 6, 14]. Other methods include incorporating low‐refractive‐index interlayers or scattering centers to disrupt waveguide modes [15, 16], as well as employing high‐refractive‐index substrates or layers [17, 18] to reduce index mismatch and thereby suppress TIR losses. Additionally, reflective electrodes with tailored optical properties have been utilized to mitigate SPP losses [19]. Collectively, these advancements have enabled outcoupling efficiencies exceeding 40% in optimized OLED architectures, bringing device EQEs close to their theoretical limits [14, 15, 16]. However, the intricate fabrication processes involved impede the production scalability and cost‐effectiveness required for large‐scale manufacturing.

In parallel, to realize high‐efficiency exciton harvesting, the strategies mainly focus on molecule design of emitters, by introducing room‐temperature phosphorescence (RTP) [20, 21, 22, 23], triplet‐triplet annihilation (TTA)‐based upconverted fluorescence (TTA‐UF) [24, 25, 26], thermally activated delayed fluorescence (TADF) [27, 28, 29, 30, 31, 32, 33, 34], or hot exciton fluorescence [35, 36, 37, 38, 39]. Among these, TTA‐UF allows for the utilization of up to 62.5% of excitons, whereas RTP, TADF, and hot exciton mechanisms can theoretically achieve at most 100% exciton harvesting. Nevertheless, RTP and TDAF devices often experience significant efficiency roll‐off due to triplet‐related annihilation processes such as TTA, singlet‐triplet annihilation (STA), and triplet‐polaron annihilation (TPA), largely stemming from the long lifetime of the lowest triplet state (T1). In contrast, hot exciton materials, which exploit rapid reverse intersystem crossing (RISC) from high‐lying triplet states (T_n, n ≥ 2), have emerged as promising alternatives to enable a favorable balance between high exciton utilization and improved operational stability.

Recent advancements have highlighted the potential of isomer‐ and conformer‐based OLED emitters that enable TADF or hot exciton character [40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50]. These materials offer high device efficiency with broad color tunability via simplified fabrication processes. By leveraging intrinsic self‐doping characteristics, they can exhibit either a single dominant emission or dual emissions via tuning doping concentrations. Conformer‐based emitters are predominantly engineered to adopt planar‐orthogonal and axis‐equatorial conformations, with one conformation exhibiting TADF character, thereby substantially enhancing device EQEs [40, 41, 42, 43, 44]. Isomer‐based emitters primarily comprise excited‐state intramolecular proton transfer (ESIPT) materials that possess either TADF or hot exciton behavior. Adachi group demonstrated green OLEDs with EQE of 14% using planar fused acridones‐based TADF ESIPT emitter [46]. Yang and co‐workers developed high‐performance β‐diketone‐based green and yellow ESIPT TADF fluorophores reaching EQEs of 23.9% and 18.8%, respectively, demonstrating that the ESIPT process facilitates efficient RISC [47]. As demonstrated by our previous work and that of the You and Zhou groups, in azole‐based ESIPT hot exciton emitters, emission can be tuned by modulating the ESIPT process; suppressing it results in high‐energy (blue) emission from the enol tautomer, whereas promoting it leads to low‐energy (green/yellow/orange) emission from the keto tautomer [45, 48, 49, 50, 51, 52, 53]. Through precise molecular design of ESIPT fluorophores, RISC can be selectively regulated within one tautomer to produce dominant emission, or simultaneously activated in both isomers to achieve dual emissions. However, most ESIPT materials deliver relatively low device efficiency due to large energy dissipation caused by configuration relaxation in the excited state. Furthermore, achieving efficient exciton harvesting in both conformer/isomer forms remains challenging. Consequently, high‐performance OLED emitters leveraging such isomers or conformers are still scarce.

Crucially, achieving synergistic enhancement of both internal exciton utilization and light extraction entirely through integrated molecular engineering—without relying on complex optical outcoupling architectures—remains largely underexplored, which is critical for the developing cost‐effective and high‐efficiency OLEDs with simple architectures and high scalability. It is anticipated that this goal is achievable if all the key improvements, including efficient exciton harvesting in tautomeric systems while suppressing energy dissipation during excited‐state tautomerization, enhanced light‐outcoupling through self‐assembled nanostructures, and optimized high horizontal dipole orientation ratio via tailored supramolecular stacking, can be attained solely through molecular engineering.

In this study, we have developed a series of 3‐hydroxypropenone‐based high‐efficiency hot exciton emitters that stabilize minor β‐diketone tautomeric form via ground state intramolecular proton transfer (GSIPT), enabling a distinctive two‐by‐two synergistic optimization strategy exclusively through synergistic molecular engineering of molecular and supramolecular structures, via independent dual‐channel enhancement of both the exciton harvesting and light outcoupling.

2. Results and Discussion

2.1. Synthesis and Tautomerism Investigation

Three donor–π–acceptor–π–donor (D–π–A–π–D) compounds—DCzPDO, DPhCzPDO, and DPAPDO—featuring a diphenyl 3‐hydroxypropenone core linked to various electron‐donating side moieties have been synthesized via Buchwald‐Hartwig coupling amination reactions of 1,3‐bis(4‐bromophenyl)‐3‐hydroxypropenone (DBrPDO) with carbazole, diphenylcarbazole, or diphenylamine, respectively (Scheme 1). DBrPDO was prepared via Claisen condensation of methyl 4‐bromobenzoate with 4‐bromoacetophenone. In the resulting D–π–A compounds, the hydroxypropenone moiety serves as the electron‐accepting core (A), the peripheral carbazole or diphenylamine groups act as electron‐donating terminals (D), and the two phenyl spacers between D and A moieties function as π‐conjugated bridges. The molecular structures and purities of the synthesized compounds were confirmed by 1H NMR and high‐resolution mass spectrometry (HRMS) (Figures S1–S11).

SCHEME 1.

Synthetic routes of DPAPDO, DPhCzPDO, and DCzPDO. Synthetic conditions: a) NaH, THF, 80°C; b) [(t‐Bu)₃PH]BF₄, Pd(OAc)₂, NaOtBu, xylene, 130°C.

Some of the hydroxypropenone derivatives are reported to exhibit β‐diketone isomers, caused by the enol–keto tautomerization (Figure 1a), likely due to their small energy differences [47, 54]. In this work, the S₀→S1 excitation energies in the hydroxypropenone isomers for DCzPDO, DPhCzPDO, and DPAPDO are calculated to be only 0.13, 0.11, and 0.26 eV smaller than those of the corresponding β‐diketone forms, respectively, using the time‐dependent density functional theory (TDDFT). Consistently, in the 1H NMR spectra, the presence of strong downfield hydroxyl proton signal (δ 17.00–17.19 ppm; integral 0.91−0.95) and a strong alkene proton signal (δ 6.60–7.10 ppm; integral 0.89−0.94), alongside a weak methylene signal (δ 4.46–4.83 ppm; integral 0.04−0.21), indicates minor β‐diketone tautomer coexisting with the dominant hydroxypropenone form (Figure 1b, Figures S4, S7 and S10). Based on the integration ratio of the methylene to alkene protons, the molar fractions of β‐diketone isomers were determined as 2.6% for DCzPDO, 2.2% for DPhCzPDO, and 10.6% for DPAPDO. The predominance of the hydroxypropenone form has also been demonstrated by the single‐crystal data, which clearly reveal the precise molecular configuration of the hydroxypropenone moiety in DCzPDO and DPAPDO (Figure 1c). Their crystallographic data have been provided in Table S1.

FIGURE 1.

a) The isomerization between the 3‐hydroxypropenone and β‐diketone tautomers. b) 1H NMR spectra of hydroxypropenones, with weak methylene absorption peaks at 4.46–4.83 ppm corresponding to a small proportion of β‐diketone forms. c) The crystal structures of DCzPDO and DPAPDO, showing from front and top views. d) FTIR spectra of DCzPDO, DPhCzPDO and DPAPDO.

As seen in the single crystal structures, the hydroxypropenone core is completely planar (Figure 1c). This is mainly attributed to the strong intramolecular hydrogen bonding between the hydroxyl and carbonyl groups, consistent with negligible shift of the hydroxyl proton resonances around 17 ppm in ¹H NMR spectra with increasing concentration (Figure S13). Moreover, the hydroxypropenone core plane remains nearly coplanar with the adjacent π‐bridged phenyl spaces, showing dihedral angles of only 1.92°/−8.97° for DCzPDO and 3.56°/−10.52° for DPAPDO (Figure 1c). However, the overall molecular architectures are moderately twisted. Specifically, the torsion angles, between the diphenyl hydroxypropenone π–A–π quasi‐plane and the two side D groups of carbazole or phenylamine, are measured as 47.9°/55.94° for DCzPDO and 39.17°/38.13° for DPAPDO, respectively. This nonplanarity plays a key role in mitigating concentration‐induced quenching, resulting in moderate photoluminescence quantum yields (PLQYs) of 23.4%, 51.3%, and 41.6% for DCzPDO, DPhCzPDO and DPAPDO, respectively, which are essential for achieving high‐performance OLEDs. Additionally, these three materials exhibit high degradation temperatures of 335, 439, and 364°C for DCzPDO, DPhCzPDO, and DPAPDO, respectively (Figure S14a). The glass transition temperatures of DCzPDO and DPAPDO were determined as 90°C and 83°C, respectively, whereas no distinct glass transition signal was observed for DPhCzPDO (Figure S14b). These good thermal stabilities are advantageous for high film‐forming capabilities in device fabrication.

Noted that, in contrast with large discrepancy in the bond lengths between the hydroxyl proton and the two adjacent oxygen atoms within the six‐membered planar hydroxypropenone core for DPAPDO (1.656 Å for H∙∙∙O═C hydrogen bond and 0.840 Å for H─O bond in hydroxyl group), the two bond lengths for DCzPDO are very close, measured as 1.217 Å for H∙∙∙O═C hydrogen bond and 1.286 Å for H─O─C bond, respectively (Figure 1c). This observation suggests the coexistence of an isomeric equilibrium between two mirror‐image enantiomers of 3‐hydroxypropenones in DCzPDO, presumably driven by a fast 1,5‐hydrogen migration between the two oxygen atoms mediated by intense intramolecular hydrogen bonding (Scheme S1). This induces strong resonance within the C═C─C unit inside the hydroxypropenone core, resulting in bond length equalization between the double (C═C) and single (C─C) bonds, evidenced by their nearly identical bond lengths of 1.388 and 1.390 Å, respectively (Figure 1c). Simultaneously, this resonance stabilization minimizes vibrational frequency differences of the out‐of‐plane C−H bending peaks in the range of 650−800 nm, enabling closer proximity of the two peaks at 746 and 723 cm⁻¹ (Figure 1d). Consistently, this bond length averaging between C═C and C─C bonds redshifts the ═C─H stretching vibration peak to lower frequencies of 2980–2800 cm⁻¹, with significantly greater intensity than observed in the other two materials. The resonance equilibrium between two enantiomers results in significant energy dissipation, accounting for its lowest PLQY (23.4%) among the three emitters.

For DPAPDO, the O─H bond length (0.840 Å) in the hydroxyl group is significantly shorter than the H···O═C intramolecular hydrogen bonding (1.656 Å), indicating the greater stability of the enol isomer exhibiting almost no 1,5‐hydrogen migration between hydroxypropenone enantiomers (Figure 1c). This enhanced stability favors for the formation of the β‐diketone tautomer induced by 1,3‐hydrogen migration, consistent with its significantly higher β‐diketone proportion (10.6% for DPAPDO) than DCzPDO (2.6%).

Although the β‐diketone isomer cannot be isolated via crystal growth due to its low abundance, its existence in the solid state can still be evidenced by Fourier transform infrared (FTIR) spectroscopy (Figure 1d). Thanks to the highest concentration of the β‐diketone tautomer in DPAPDO (10.6%) among the three samples, the C═O stretching band at a higher wavenumber of 1606 cm⁻¹ corresponds to the β‐diketone minor form is observed, in addition to the C═O stretching band at 1587 cm⁻¹ from the hydroxypropenone major form. The higher frequency of C═O stretching peak for β‐diketone is mainly caused by the interruption of molecular conjugation via the central methylene group.

2.2. Spectroscopic Properties

The films of all three materials (in PMMA matrix) exhibit two main absorption bands, peaking at 290/392, 293/401, and 293/433 nm for DCzPDO, DPhCzPDO, and DPAPDO, respectively (Figure 2a). The long‐wavelength absorption peak, ranging from 392 to 433 nm, originates from the lowest excited state (S1) and is primarily attributed to the low‐lying HLCT state (Tables S2–S7). The short‐wavelength absorption band peaking at 290–293 nm is assigned to the high‐lying HLCT state, exhibiting different charge distributions across the three materials.

FIGURE 2.

Normalized a) absorption and b) emission spectra of the films, and c) solvent‐dependent emission spectra and d) solvatochromic Lippert‐Mataga models (inset: natural transition orbitals of S₀ → S₁) of DCzPDO, DPhCzPDO, and DPAPDO.

For the photoluminescence (PL) spectra, the films of DCzPDO, DPhCzPDO, and DPAPDO exhibit featureless emission peaks at 465, 494, and 496 nm, respectively (Figure 2b). The longer‐wavelength emissions of DPhCzPDO and DPAPDO than DCzPDO correlate with the larger electron‐donating strength of the diphenylcarbazole (in DPhCzPDO) and diphenylamine donors (in DPAPDO) than the carbazole one (in DCzPDO). All their emissions are assigned to a CT‐dominated HLCT character, as supported by their solvent‐dependent PL spectra and the charge distributions in the natural transition orbitals (NTOs) (Figure 2c,d). In the nonpolar solvent n‐hexane, all the compounds exhibited fine‐structured emissions with three vibronic peaks at 412/434/460 nm, 424/449/473 nm, and 439/465/494 nm for DCzPDO, DPhCzPDO, and DPAPDO, respectively, which are characteristic of locally excited (LE) features. As the solvent polarity increased, the emission maximum wavelength gradually redshifted, indicating typical CT characteristics, with the transition dipole moments (µ_e) calculated to be 18.65, 25.23, and 15.02 D, respectively, according to the Lippert‐Mataga model.

Consistently, the charge distributions observed in the NTOs of the S₀ → S₁ transitions clearly demonstrate the CT‐dominated HLCT character of the excited state (Figure 2d inset, Table S2–S7). For all three compounds, the hole is delocalized across the entire molecule, excluding the keto and hydroxyl groups within the hydroxypropenone core, while the electron density is primarily confined to the hydroxypropenone core and the π‐bridged phenyl units. The key difference lies in hole localization: DPAPDO exhibits greater charge density distribution across the phenyl π–bridges and the hydroxypropenone acceptor core than DCzPDO and DPhCzPDO. This results in a lower CT ratio in DPAPDO than in DCzPDO and DPhCzPDO, which is consistent with the lower µ_e for DPAPDO (15.02 D) than those for DCzPDO (18.65 D) and DPhCzPDO (25.23 D). Moreover, all these emitters exhibit transition dipole moments preferential aligned along their molecular symmetric axes, showing near‐parallel orientation relative to the central 3‐hydroxypropenone plane (Figure S15). These CT‐dominated HLCT characteristics are promising for providing possible efficient hRISC channels for the high‐efficiency harvesting of triplet excitons in OLEDs [37, 38, 55].

2.3. OLED Performance

The highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) energy levels of DCzPDO, DPhCzPDO, and DPAPDO are determined to be −5.6/−2.9, −5.0/−2.5 and −4.5/−2.1 eV, respectively (Figure S16). To explore the applications of these emitters in OLEDs, a series of multilayer OLEDs were fabricated using the following configuration: indium tin oxide (ITO)/dipyrazino[2,3‐f:2′,3′‐h]quinoxaline‐2,3,6,10,11‐hexacarbonitrile (HAT‐CN)/molybdenum trioxide (MoO₃)/1,1‐bis [(di‐4‐tolylamino)phenyl]cyclohexane (TAPC)/4,4′,4″‐tri(N‐carbazolyl)triphenylamine (TCTA)/9,9′‐(1,3‐phenylene)bis‐9H‐carbazole (mCP)/mCP:TPBi:emitters/2,2′,2″‐(1,3,5‐Benzinetriyl)‐tris(1‐phenyl‐1‐H‐benzimidazole) (TPBi)/1,3,5‐tri(m‐pyrid‐3‐ylphenyl)benzene (TmPyPB)/lithium fluoride (LiF)/aluminum (Al) (see the device structure in Figure S17). Here, TAPC and TmPyPB served as the hole‐ and electron‐transport layers, respectively, whereas TCTA and TPBi were used as the hole‐ and electron‐blocking layers, respectively. HAT‐CN/MoO₃ and LiF were applied as the hole and electron‐injecting layers, respectively. The mixing ratios of co‐hosts (mCP: TPBi) in the emitting layers (EMLs) were 1:3 for DCzPDO based device, and 1:1 for the DPhCzPDO and DPAPDO–based devices, respectively. The optimized thicknesses of each layer and the emitter concentrations are summarized in Table S8.

The OLED devices incorporating DCzPDO, DPhCzPDO, and DPAPDO emitters produced blue, sky‐blue, and green EL, respectively, with measured Commission Internationale de l'Éclairage (CIE) coordinates of (0.143 ± 0.007, 0.217 ± 0.021), (0.183 ± 0.011, 0.443 ± 0.046), and (0.200 ± 0.008, 0.581 ± 0.018) (Figure 3 insets, Figure S18, Table 1). Moreover, OLEDs employing DCzPDO, DPhCzPDO, and DPAPDO show EL spectra with full widths at half maximum (FWHMs) of 64, 72, and 64 nm, respectively. The broadest emission from DPhCzPDO‐based devices is possibly attributed to the most pronounced CT character in DPhCzPDO, as indicated by its largest transition dipole moment (25.23 Debye). Notably, the DPhCzPDO‐based devices also exhibited a slight blue shift in the EL peak, moving from 500 to 492 nm as the applied voltage increased from 4 to 14 V, with corresponding CIE coordinates shifting from (0.195, 0.490) to (0.176, 0.410). By contrast, devices based on DCzPDO and DPAPDO showed stable spectral profiles, maintaining EL peaks at 472 and 504 nm, respectively, regardless of voltage changes. Consistently, the CIE coordinates show very small variation, from (0.151, 0.276) to (0.141, 0.235) for the DCzPDO‐based device, and from (0.208, 0.598) to (0.195, 0.562) for DPAPDO‐based device, with the operating voltage increasing from 4 to 10 and 14 V, respectively.

FIGURE 3.

a) EQE, b) CE, and c) PE versus luminance curves; insets: EL spectra, PL spectra pf the EML film, and devices operating photographs for a) DPAPDO, b) DPhCzPDO, and c) DPAPDO; (d) Current density‐voltage‐luminance curves. The rough plots in electrical characterization at low current densities originate from current fluctuation during J–V–L measurement.

TABLE 1.

EL performance of the OLEDs.

Emitter	V on [V]	CEmax [cd A−1]	PEmax [lm W−1]	EQEmax [%]	L max [cd m−2]	λPL/FWHM a [nm]	CIE a
DCzPDO	4.37	8.95	6.42	5.3	3513	472/64	(0.143, 0.217)
DPhCzPDO	3.41	31.13	25.23	12.0	16472	496/72	(0.185, 0.448)
DPAPDO	2.98	72.04	75.72	23.7	11499	504/64	(0.198, 0.579)

^a At 1000 cd m⁻².

All the devices demonstrated low turn‐on voltages of 4.4, 3.4, and 3.0 V, and reached maximum luminance of 3513, 16472, and 11499 cd m⁻² for DCzPDO, DPhCzPDO, and DPAPDO‐based devices, respectively. The DCzPDO‐based device delivered the lowest device performance, with a peak current efficiency (CE) of 8.95 cd A⁻¹, power efficiency (PE) of 6.42 lm W^−1, and EQE of 5.3%. By comparison, the DPhCzPDO‐based device showed significantly improved performance, exhibiting a maximum CE and EQE of 31.13 cd A⁻¹ and 12.0%, respectively. The DPAPDO‐based device exhibited the highest device efficiency, with a maximum CE, PE, and EQE of 72.04 cd A⁻¹, 75.72 lm W^−1, and 23.7%, respectively. These emitters demonstrate hRISC from upper‐level triplets (T_n, n ≥ 2), which will be discussed in the following section, confirming their hot exciton character. The device performance of the sky‐blue and green OLEDs based on DPhCzPDO and DPAPDO emitters is among the highest performances reported for hot exciton emitters [56, 57, 58, 59, 60, 61]. Moreover, the device performance of the green OLED is even comparable to that of its TADF counterparts based on the β‐diketone core [47]. All the EL performance data have been summarized in Table 1. Notably, all these OLEDs show high reproducibility, indicating good potential for scaling up high‐efficiency devices (Figure S19).

It is worth noting that although the lowest performance of the DCzPDO device (EQE of 5.3%) aligns with its lowest PLQY (23.4%) among the three materials, the discrepancy in PLQYs for DPhCzPDO (51.3%) and DPAPDO (41.6%) cannot explain the nearly two‐fold difference in their device EQEs (12.0% for DPhCzPDO and 23.7% for DPAPDO). Therefore, deep investigations of the underlying mechanism for their device efficiency are highly demanded.

2.4. Factors Underlying Anomalous High Device Efficiency

2.4.1. Balancing Enol and Ketone Tautomers for Enhanced Exciton Harvesting

2.4.1.1. Enol‐Tautomer‐Enabled hRISC Activation in DPADPO

Notably, the maximum EQEs of DPhCzPDO and DPAPDO based devices (12.0% and 23.7%, respectively) far exceed the theoretical maximum value (5%) for conventional fluorescent OLEDs [2, 62, 63, 64]. This strongly suggests high exciton utilization efficiency with the involvement of triplet excitons in the emission process. To explore this, the oxygen sensitivity of both the steady‐state and transient PL spectra of the three emitters was examined (Figure 4). Upon deaeration, all the steady‐state PL spectra exhibited notable emission enhancements, whereas the transient PL spectra showed slightly prolonged lifetimes. Prior to deoxygenation, the materials displayed biexponential fluorescence decays, with the shorter lifetimes of 0.48, 1.41, and 1.01 ns and the longer lifetimes of 3.13, 5.09, and 2.42 ns for DCzPDO, DPhCzPDO, and DPAPDO, respectively (Figure 8 and Table S9). These findings align with the time‐resolved emission spectra (TRES), which also revealed two distinct fluorescence lifetimes for each material (Figure S20). Following deaeration, the shorter and longer lifetimes of DCzPDO, DPhCzPDO, and DPAPDO shifted to 0.46/3.68, 1.48/5.74, and 1.08/2.60 ns, respectively, with the proportion of the longer‐lifetime component marginally increasing from 13.21%, 10.36%, and 33.55% to 18.54%, 10.67%, and 34.23%, respectively. Consequently, the average fluorescence lifetimes slightly increased from 0.83, 1.79, and 1.48 ns to 1.06, 1.93, and 1.60 ns for DCzPDO, DPhCzPDO, and DPAPDO, respectively. These results provide solid evidence for the involvement of triplet excitons in the emissive processes of these materials.

FIGURE 4.

a, d, g) Steady‐state and b, e, h) transient PL spectra in toluene before and after deaeration (λ_ex = 380 nm), and c, f, i) Transient EL spectra of a, b, c) DCzPDO, d, e, f) DPhCzPDO, and g, h, i) DPAPDO. The transient PL and EL spectra were all detected at their maximum emission peaks.

FIGURE 8.

AFM images of DCzPDO, DPhCzPDO and DPAPDO in doped emitting films captured at the 10 µm × 10 µm scale (top) and 3D height representation (bottom).

The observed nanosecond‐scale emission lifetimes clearly rule out TADF mechanisms, consistent with the large S₁ → T₁ energy gaps (∆E_ST). ∆E_ST were measured as 0.75, 0.57, 0.55 eV for DCzPDO, DPhCzPDO, and DPAPDO, respectively, as indicated from the fluorescence and phosphorescence peaks (Figure S21). These substantial gaps are larger than the singlet‐triplet energy gaps required for thermally accessible T₁→S₁ RISC transitions at room temperature (generally exhibiting ΔE_ST < 0.25 eV). In addition, the EL decay curves show the ratios of TTA‐UF not exceeding 32%, 13%, and 5.1% for DCzPDO, DPhCzPDO, and DPAPDO‐based devices (Figure 4c,f,i).

To further elucidate the underlying molecular photophysics, TDDFT calculations were conducted to examine the energy levels of the singlet and triplet excited states, along with the corresponding spin‐orbit coupling (SOC) matrix elements (Figure 5). We mainly examined the transition probabilities for those hRISC processes originating from high‐lying triplet states with ΔE_ST < 0.25 eV. For clarity, the 3‐hydroxypropenone and β‐diketone tautomers are hereafter termed as the enol and keto forms, respectively.

FIGURE 5.

Energy level diagrams of molecular excited states and hRISC channels in the 3‐hydroxypropenone tautomers (enol form) for a) DCzPDO, b) DPhCzPDO, and c) the DPAPDO. Dashed arrow: internal conversion; red lines: α ≥ 10 cm⁻², blue lines: 1 < α < 10 cm⁻², solid lines: E_S ≤ E_T, and dotted lines: E_S > E_T. Solid line: barrier‐free transitions; Dashed line: thermally activated transitions.

Since the enol tautomer predominates (>89%) in all three emitters, we first investigated potential hRISC pathways in the majority 3‐hydroxypropenone structure (enol forms). For DCzPDO enol form, the energy differences between T₃ and T₂ (ΔE_T3–T2) and between T₂ and T₁ (ΔE_T2–T1) are calculated to be 0.25 and 0.23 eV, respectively, indicating fast internal conversion (IC) which prohibits hRISC from T₂/T₃ to S₁, although the SOC matrix elements of <S₁|Ĥ_soc|T₂> and <S₁|Ĥ_soc|T₂> are as large as 1.05 and 0.44 eV (Figure 5a). Similarly, for DPhCzPDO enol form, the small ΔE_T3–T2 (0.25 eV) effectively prevents any hRISC processes (Figure 5b).

In contrast, in the DPAPDO enol form, ΔE_T3–T2 is relatively large as 0.43 eV. This reduces internal conversion considering the relatively rigid molecular configuration, and enables fast barrierless hRISC from T₃ (0.12 eV higher than S₁) to S₁ with a large SOC element matrix of 1.30 cm⁻¹ (Figure 5c). It is known that the rate of RISC (k_RISC) is proportional to the transition probability of RISC between singlet and triplet states (α), where α is defined as α = <S|Ĥ_soc|T>²/ΔE_ST after simplifying the first‐order perturbation theory of Fermi's golden rule [65, 66]. The α value of T₃ → S₁ transition is calculated to be 14.51 cm⁻² eV⁻¹. In addition, the thermally activated hRISC from T₃ (0.21 eV lower than S₂) to S₂ exhibits a SOC element matrix and α value of 0.60 cm⁻¹ and 1.70 cm⁻² eV⁻¹, respectively. Therefore, among the three emitters, only in DPAPDO can its major enol form undergo effective hRISC channels, which significantly contribute to its high device performance.

2.4.1.2. Ketone‐Tautomer‐Mediated Triplet Harvesting in All Emitters

Noted that, the DPhCzPDO‐based device still exhibits a peak EQE as high as 12.0%, although the major enol form of the emitter exhibits no possible hRISC channels. This indicates that there should be other factors accounting for the high device performance. Thus, we further examined the possible hRISC channels in their β‐diketone tautomers (keto forms).

The DCzPDO keto form exhibits a relatively large ΔE_T3–T2 of 0.42 eV to inhibit the fast IC from T₃ to T₂ (Figure 6a). Therefore, barrierless hRISC channels from energetically higher‐lying T₃ to lower‐lying S₂ and S₁ are enabled, with corresponding SOC matrix elements <S₂|Ĥ_soc|T₃> and <S₁|Ĥ_soc|T₃> of 5.57 and 0.69 cm⁻¹, respectively. Here, T₃ is 0.06 and 0.12 eV higher than S₂ and S₁, respectively. Specifically, the α value for T₃→S₂ transition is extremely large as 487 cm⁻² eV^−1, due to its very small energy gap (0.06 eV) and large SOC matrix element (5.57 cm⁻¹), while that for T₃→S₁ transition is only 3.89 cm⁻² eV⁻¹, respectively.

FIGURE 6.

Energy level diagrams of molecular excited states and hRISC channels in the β‐diketone tautomers (keto form) for a) DCzPDO, b) DPhCzPDO, and c) DPAPDO. Dashed arrow: internal conversion; red lines: α ≥ 10 cm⁻², blue lines: 1 < α < 10 cm⁻², solid lines: E_S ≤ E_T, and dotted lines: E_S > E_T. Solid line: barrier‐free transitions; Dashed line: thermally activated transitions.

The DPhCzPDO keto form enables a barrier‐free hRISC pathway from the high‐lying T₃ to the low‐lying S₂ state, while a thermally activated hRISC route from the lower T₁ to the higher S₁ state, as the IC between T₃ and T₂ is suppressed by a substantial energy difference of 0.46 eV (Figure 6b). Here T₃ is 0.18 eV above S₂, while T₁ is 0.23 eV below S₁, with the SOC matrix elements determined as 0.47 cm⁻¹ for <S₂|Ĥ_soc|T₃> and 0.56 cm⁻¹ for <S₁|Ĥ_soc|T₁>, and the corresponding α values of 1.22 and 1.37 cm⁻² eV⁻¹, respectively.

In the DPAPDO keto form, the energy gap between T₃ and T₂ is 0.58 eV, enabling hRISC to single states from T₃ (Figure 6c). With a small energy gap ΔE_T3–S1 of only 0.02 eV and the SOC element matrix of <S₁|Ĥ_soc|T₃> of 0.48 cm⁻¹, the α value is calculated to be 13.71 cm⁻² eV⁻¹. Notably, although ΔE_T3–S2 has a relatively large energy difference of 0.13 eV, the corresponding SOC element matrix of <S₂|Ĥ_soc|T₃> is extremely large as 8.71 cm⁻¹, thus the corresponding α value reaches as high as 592 cm⁻² eV⁻¹.

Consequently, DPAPDO stands out as the exclusive emitter among the three materials that simultaneously enables accessible hRISC channels in its major enol conformation (hydroxypropenone tautomer) and exhibits the highest molar fraction of the minor keto form (β‐diketone tautomer) that provides supplementary efficient hRISC pathways, resulting in its highest efficiency of exciton harvesting. This correlates well with its highest device performance, with a maximum EQE achieving 23.7%.

For DCzPDO and DPhCzPDO, although their enol forms exhibit no available hRISC routes, their keto forms are found to possess effective hRISC transitions. Although the minor keto form of DCzPDO exhibits higher SOC and α values compared to DPhCzPDO, the excitation energy of the major enol form is greatly dissipated through rapid resonance interconversion between mirror‐symmetric hydroxypropenone enantiomers via 1,5‐proton migration, as we have discussed above. This explains the inferior device efficiency of the DCzPDO‐based device (EQE = 5.30%) compared to the DPhCzPDO‐based OLED (EQE_max = 12.0%), demonstrating that the chemical structure of the major enol form critically governs isomerization pathways and ultimately influence device performance.

All the above results imply hot exciton mechanism dominates in these devices. This is well consistent with the low activation energies of 31.65, 20.47, and 7.41 meV in a wide temperature range of 20°C–140°C/220°C/180°C for DCzPDO, DPhCzPDO, and DPAPDO, respectively (Figure S22, S23 and Table S10). Kinetic modeling of key photophysical processes (Table 2) reveals distinct behaviors among the three emitters. Although DCzPDO demonstrates the fastest reverse intersystem crossing (k_hRISC = 4.95 × 10⁸ s⁻¹), this benefit is negated by its exceptionally rapid triplet depletion through internal conversion (k_IC ^Tn = 1.62 × 10⁹ s⁻¹)—three times faster than k_hRISC—combined with its accelerated internal conversion in the lowest singlet state (k_IC ^S = 3.59 × 10⁸ s⁻¹), cumulatively causing substantial triplet loss and compromised fluorescence efficiency. In contrast, DPhCzPDO and DPAPDO demonstrate a more favorable balance between k_hRISC and k_IC ^Tn, enabling efficient triplet harvesting for delayed fluorescence. Notably, DPAPDO exhibits a unique kinetic profile where k_ISC (1.96 × 10⁸ s⁻¹) is substantially lower than k_hRISC (4.42 × 10⁸ s⁻¹), while maintaining a moderately enhanced k_r ^S (2.03 × 10⁸ s⁻¹), collectively promoting triplet harvesting for singlet fluorescence. These findings provide a consistent explanation for the device performance, with DCzPDO‐ and DPAPDO‐based OLEDs showing the lowest and highest efficiencies, respectively.

TABLE 2.

Photophysical properties of the DCzPDO‐, DPhCzPDO‐, and DPAPDO‐doped films (corresponding to EML contents and concentrations).

Emitter	kr S (108 s−1)	KIC S (108 s−1)	kISC (108 s−1)	kPF (108 s−1)	kDF (108 s−1)	khRISC (108 s−1)	kIC Tn (108 s−1)
DCzPDO	1.10	3.59	3.07	7.75	2.99	4.95	16.2
DPhCzPDO	1.10	1.04	1.81	3.95	1.67	3.07	2.92
DPAPDO	2.03	2.85	1.96	6.85	3.15	4.42	6.21

In our work, significant roll‐off was observed in all the devices. This likely stems from the complex interplay of multiple factors, including charge balance, exciton quenching, TTA‐UF lifetimes, and device stability (Figure 4c, f, i, Figure S24, S25 and Table 2). The DPAPDO‐based device showed moderate charge balance, the shortest operational lifetime, and the longest TTA‐UF lifetime, exhibiting the most significant efficiency roll‐off among the three devices. In contrast, the DCzPDO‐based device maintained moderate charge balance, intermediate operational lifetime, and severe exciton quenching, leading to a modest efficiency roll‐off. The DPhCzPDO‐based device outperformed the others in optimal charge balance, the longest operational lifetime, and the shortest TTA‐UF lifetime. These combined characteristics resulted in its smallest efficiency roll‐off.

Notably, in the DPhCzPDO‐based device, the low concentration of hRISC‐active keto tautomer (2.2%), which is energetically higher than the major enol form, proves insufficient to explain its high device performance (EQE_max = 12.0%), implying other contributions from additional mechanisms.

2.4.2. Preferential Dipole‐Orientation Enhanced Photon Extraction

Further, the singlet exciton yields (η_s) of these OLEDs were calculated using the equation ${\eta }_s=\frac{EQ{E}_{max}}{(\gamma \times {\eta }_{PL}\times {\eta }_{out})}$. Here, EQE_max represents the maximum EQE, γ denotes the electron‐hole recombination probability in the emission layer (assumed ideally to be 100%). η_PL refers to the PLQY, which were measured as 23.4%, 51.3%, and 41.6% for DCzPDO, DPhCzPDO, and DPAPDO, respectively. The light outcoupling efficiency (η_out), typically 20–30% in OLEDs, was initially assumed as 25% in this study. Based on these values, the η_s for the DCzPDO‐, DPhCzPDO‐, and DPAPDO‐based devices were calculated as 90%, 94%, and 228%, respectively.

Critically, the calculated η_s (228%) in the DPAPDO‐based device – derived using the assumed η_out of 25% – far exceeds the physical limit of 100%, necessitating an actual η_out of at least 57%. Here, “LOE ≥ 57%” represents an inferred lowest value rather than an independently determined value. This exceptionally high η_out is attributed to the combination of favorable film morphology and high horizontal dipole orientation ratios, which will be discussed in detail, respectively. The horizontal dipole orientation ratios were determined by angle‐resolved PL spectroscopy to be 84.5%, 89%, and 77% for DCzPDO, DPhCzPDO and DPAPDO, respectively (Figure 7). This indicates highly preferential in‐plane dipole alignment across all these materials, significantly enhancing light outcoupling efficiency.

FIGURE 7.

p‐Polarized angle‐dependent PL radiance of the EMLs based on a) DCzPDO, b) DPhCzPDO, and c) DPAPDO.

However, based solely on the horizontal dipole orientation ratio of DPAPDO (77%), the predicted η_out for its OLED device should theoretically not exceed ∼35%. Remarkably, the experimentally determined η_out surpasses 57%, suggesting substantial contributions from additional device structural factors to light extraction enhancement.

2.4.3. Self‐Assembled Nanostructure‐Driven Light Outcoupling

2.4.3.1. Nanogroove Structures for Enhancing Light Outcoupling

To explore this, the surface morphologies of the emitting layer films doped with these three emitters were examined using atomic force microscopy (AFM). The analysis revealed elongated, approximately parallel surface features resembling V‐shaped nanogrooves, indicative of directional surface texturing (Figure 8, Figure S26). The V‐nanogrooves in DCzPDO exhibited heights of 1.5–4 nm and groove widths ranging from 0.2–2 µm, while those in DPhCzPDO and DPAPDO showed heights of 1–4 nm/1–3.5 nm, and groove widths of 0.2–1.5/0.2–2 µm, respectively. Although nanogroove heights are subwavelength‐scale, these wavelength‐scale aperiodic nanostructures substantially enhance OLED light outcoupling efficiency of OLED devices. Similar enhancement has been demonstrated in ultrathin buckles spontaneously formed with nanoscale heights, as reported by another group [13].

Among the three materials, DCzPDO exhibits the highest surface roughness, with an average root mean square (RMS) value of 4.4 nm, with a number of T₁ energy level typically 5–25 nm in height observed (Figure 8a). DPhCzPDO has a less rough surface with an RMS roughness of 2.8 nm, exhibiting much fewer rod‐like nanopillars with heights mostly of 3–10 nm (very few of 12–23 nm, Figure 8b). DPAPDO exhibits the smoothest surface, with a lowest RMS roughness of 1.9 nm and the fewest vertical nanorod defects (mostly 1.2–5 nm in height) sparsely distributed (Figure 8c).

To investigate the influence of these nanostructures on the light extraction, Lumerical FDTD simulations of the aperiodic nanogroove arrays with sharp‐edged features were performed. The results demonstrate that the non‐uniform groove‐like nanostructures effectively reduce optical losses in both WG and SPP modes, leading to enhanced light outcoupling (Figure 9). The far‐field electric field intensity distributions reveal more focused emission with the nanogroove geometry (Figure 9b,c), confirming the enhanced LOE in our OLEDs. A clear reduction in waveguide light is observed when detecting emissions from both sides of the device, indicating that the nanogroove structures efficiently scatter confined modes and redirect optical energy into extractable far‐field directions (Figure 9d,e). Furthermore, the nanostructures disrupt SPP confinement near the Al/LiF interface and promote downward coupling of energy into the glass substrate, whereas in the absence of the nanogroove structures, a greater portion of the energy remains trapped at the interface (Figure 9f). Similar non‐uniform periodic nanowrinkles and nanocorrugated structures have been reported to increase light extraction efficiency by suppressing TIR losses and SPP modes while simultaneously maintaining wide angular emission and broadening the spectral scattering range [13, 67, 68, 69, 70]. These nanoscale features modulate light propagation within the OLEDs by locally altering the refractive index and introducing structural perturbations. This is possibly because such directional nanotextures can promote Bragg scattering or grazing‐angle diffraction, supplying the in‐plane momentum necessary to couple non‐radiative modes—such as WG and SPP modes—into radiative (free‐space) modes [70, 71, 72, 73]. Consequently, the optical losses arising from WG confinement, TIR, and SPP‐related dissipation are mitigated, thereby improving overall light extraction.

FIGURE 9.

a) Simplified device structure with EML nanogroove geometry; FDTD simulated far‐field electric field intensity distributions of device b) without and c) with EML nanogroove geometry; Electric field intensity distributions without (top) and with (bottom) nanogroove structures in the d) left and e) right yz‐planes of the device, indicating waveguide light at two sides of the device; and in the f) xz‐plane at y = 50 nm, illustrating dipole emission near the metal‐organic interface.

2.4.3.2. Self‐Assembling‐Driven Formation of Nanogroove Structures

The formation of these directional nanogroove structures is regarded to be closely related to the molecular assembling behavior. Structural analysis reveals DCzPDO crystals predominantly grow along two preferential orientations: the (−111) and (01−1) planes (Figure 10a). The molecular assembling along the (01−1) plane is mainly driven by the C─H···π interactions (2.73 Å) between the hydrogen at 2‐positions of the carbazole unit in one molecule and the phenyl ring within carbazole of a neighboring molecule, leading to a significant presence of vertical nano‐pillars (Figure 8a). And the supramolecular self‐organization along the (−111) plane is primarily governed by the C─H···π interactions (2.67 Å) between 3‐hydrogen at the carbazole unit and the phenyl ring inside the carbazole. This directional molecular assembling mainly accounts for the resulting 1D V‐shaped nanogroove formation, which will be discussed later.

FIGURE 10.

Crystallographic orientation analysis of distinct crystal specimens (top), and single‐crystal molecular packing modes and involved dimer interactions for a) DCzPDO and b) DPAPDO. c) Proposed scheme illustrating growth of nanogroove structures (dots represent host molecules; nanoribbons represent emitter molecular assemblies oriented along the preferential growth direction).

In contrast, DPAPDO possesses only one preferential molecular growth direction along the (100) plane, accounting for the nearly parallel V‐shaped nanogrooves (Figure 10b). This directional molecular assembling is mainly caused by the strong hydrogen bonding (2.67 Å) between terminal phenyl proton of triphenylamine unit and the carbonyl oxygen, as well as the C─H···π interaction (2.83 Å) between the terminal phenyl ring and the diphenyl ring connected to the hydroxylpropenone core, and the C–H···π interactions (2.72 and 2.56 Å) between the terminal phenyl ring of triphenylamine unit and the N‐phenyl ring in carbazole.

Based on the above investigations, a possible mechanism for the formation of V‐shaped nanogrooved film structures has been proposed (Figure 10c). During the final stage of the vacuum deposition for the emitting layer, a controlled rap‐down evaporation speed enables the oriented molecular growth of the emitters preferentially along the (−111) plane for DCzPDO and (100) plane for DPAPDO, respectively (indicated by the cyan nanoribbon). Meanwhile, during the 1‐D molecular assembling, the evaporated host molecules (indicated by the little light‐purple balls) deposit alongside both edges of the nanoribbons. Subsequently, a flat‐lying quasi‐triangular‐prism morphology is formed. Further, many triangular prisms align nearly parallel, forming V‐shaped nanogroove film structures, highly favorable for improving light extraction in OLED devices. For DCzPDO, its two nearly perpendicular molecular growth directions result in higher roughness of the V‐nanogroove film morphology, largely decreasing the device performance in the DCzPDO‐based device (EQE = 5.30%). Control OLEDs without nanogroove geometries were also fabricated by increasing the evaporation speed of EML (Figure S27, S28 and Table S11). The EMLs with slow evaporation speed showed obvious V‐nanogroove film morphology, and the corresponding OLEDs showed higher device efficiencies. This result directly proves the effectiveness of our conception in improving OLED efficiencies. Moreover, different batches of the nanogroove structures in EML can be obtained by controlling the evaporation speed (Figure S29), indicating great reproducibility for scaling up the EML‐nanogrooved devices.

In short, the main reason for the abnormal high performance in the DPAPDO‐based OLED device lies in four facts. First, only in DPAPDO, among the three emitters, can its major enol form enable a feasible hRISC pathway. Second, DPAPDO possesses the highest keto concentration, with the keto form providing supplementary effective hRISC channels. Thirdly, DPAPDO exhibits the highest‐qualified film nanostructures with the fewest defects via unidirectional molecular assembling for enhancing light extraction. Finally, its preferentially horizontally aligned molecular dipole concurrently improves the light outcoupling. Consequently, entirely through molecular engineering, independent dual enhancement of both internal quantum efficiency and light outcoupling efficiency has been realized, achieving up to near unity (∼100%) and 57%, respectively.

3. Conclusion

In summary, we have developed a series of high‐performance 3‐hydroxypropenone‐based hot exciton fluorophores—DCzPDO, DPhCzPDO, and DPAPDO—that emit blue, sky‐blue, and green light, respectively. These materials not only incorporate minor β‐diketone tautomers stabilized in the ground state that establish efficient hRISC channels for effective harvesting of triplet excitons in emitters, but also possess optimal dipole orientation and self‐assembled nanogroove emitting layers for enhancing light extraction in OLEDs. The ultimate device performance is governed by the synergistic interplay of chemical structure, isomerization equilibrium, dipole orientation, and film nanostructures — all controlled exclusively through molecular engineering. Among these emitters, DPAPDO distinguishes itself by delivering an impressive EQE of 23.7% in OLED devices, which rivals the state‐of‐the‐art β‐diketone‐based TADF counterparts. This high device performance is attributed to a unique two‐by‐two strategy via dual boost of both exciton harvesting and light outcoupling in parallel, culminating in a four‐fold synergistic improvement across the overall device efficiency: 1) Only in this material (DPAPDO) does the major 3‐hydroxypropenone tautomer enable efficient hRISC. 2) The minor β‐diketone tautomer in DPAPDO presents maximal concentration while providing supplemental hRISC pathways. 3) Supramolecular assembly of DPAPDO generates optimal high‐quality V‐shaped nanogroove film morphology, significantly enhancing light outcoupling efficiency. 4) Its moderate horizontal dipole orientation ratio further enhances light extraction. Consequently, DPAPDO achieves near‐unity IQE (∼100%) and exceptionally high LOE of 57% via molecular engineering — eliminating the need for externally introduced light outcoupling structures. Our work demonstrates a new pathway to scalable, cost‐effective, high‐efficiency OLEDs simply through synergistic molecular design.

4. Experimental Section

4.1. Materials

The fluorophores were synthesized in our laboratory, and the detailed synthesis procedures are provided in the Supporting Information.

4.2. Instrumental Methods

Absorption and fluorescence spectra were measured on Hitachi UH4150 UV–vis and F‐7000 spectrometers, respectively. The degassing of oxygen was realized by bubbling the solution samples with nitrogen for 15 min. The PL decay spectra were recorded on an Edinburgh FLS‐920 instrument using single‐photon counting (the samples were irradiated with a Picoquant LDH‐D‐C‐375 nm pulsed diode laser). The thermal analyses of the DSC and TGA data were performed on STA2500 and DSC214 instruments, respectively.

4.3. OLED Preparation

The ITO glass substrates were sequentially cleaned by ultrasonic treatment in detergent, ethanol, acetone and deionized water, followed by drying at 120°C in a vacuum oven for over 1 h. After undergoing UV‐ozone treatment, the functional layers were deposited by thermal evaporation under vacuum conditions at a pressure of about 1 × 10⁻³ Pa. The current‐voltage‐luminescence characteristics were tested using a Keithley 2602 source meter and a calibrated silicon photodiode, and the EL spectra were recorded with a spectra‐scan PR655. Capacitance‐voltage (C‐V) measurements were performed using an impedance analyzer (Wayne Kerr 6505B) at a constant frequency of 1 kHz.

Funding

The National Natural Science Foundation of China (62575144, 62175115)

Supporting information

Supporting File: smll73898‐sup‐0001‐SuppMat.docx.

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.