RESULTS AND DISCUSSION

Synthesis and characterization

Supporting Information S1: Scheme S1 outlines the straightforward synthetic route to D3CZ4F, which was efficiently prepared via a one‐step palladium‐mediated Suzuki coupling reaction using low‐priced 9‐phenylcarbazole‐3‐boronic acid and 1,4‐dibromotetrafluorobenzene raw materials. After purification by column chromatography and vacuum gradient sublimation, the molecular structure was unambiguously confirmed by 1H NMR, 13C NMR, and high‐resolution mass spectrometry (See Supporting Information S1: Figure S11). Thermogravimetric analysis measurements revealed that D3CZ4F exhibited a decomposition temperature (T d, 5% weight loss) of 429°C, while a distinct endothermic signal was observed during the second‐heating curve of differential scanning calorimetry measurements, corresponding to the melting point of 296°C (Supporting Information S1: Figure S1). According to the onset of the oxidation wave from cyclic voltammetry (CV) measurement (Supporting Information S1: Figure S2), the highest occupied molecular orbital (HOMO) energy level of D3CZ4F was estimated to be −5.27 eV. The lowest unoccupied molecular orbital (LUMO) energy level was calculated to be −1.80 eV by the energy difference between the HOMO and optical bandgap (E g).

Theoretical calculations

To elucidate the geometric structure and electronic properties of D3CZ4F, density functional theory (DFT) and time‐dependent DFT (TD‐DFT) calculations were carried out at the B3LYP/6‐31G(d, p) level. Natural transition orbital (NTO) analyses were performed using Multiwfn 3.8. [39] In the ground state (S 0), D3CZ4F adopts a moderately twisted conformation, with dihedral angles of 40.69° and 40.96° between the tetra‐fluorobenzene and carbazole units (Figure 2a). The calculated C–H⋯F distances of 2.384–2.402 Å suggest the presence of intramolecular C–H⋯F hydrogen‐bonding interactions. [6] Upon photoexcitation, the optimized S 1 geometry shows a notable planarization tendency where the dihedral angles between tetrafluorobenzene and carbazole reduce to 26.19° and 26.91°, and the C–H⋯F distances shorten to 2.146–2.169 Å (Figure 2a), which is conducive to rigidifying the molecular framework, promoting electron delocalization and reducing intramolecular vibrational modes, thereby facilitating high fluorescence efficiency for D3CZ4F. Reduced density gradient analysis of the optimized S 0 further confirms non‐covalent C–H⋯F interactions are enhanced in the S 1 geometry where the green region shows the presence of obvious attractive interactions (Figure 2b). As illustrated in Figure 2c, the HOMO is delocalized across the entire π system, whereas the LUMO is primarily localized on the tetrafluorobenzene acceptor and its directly linked benzene rings, resulting in substantial orbital overlap in tetrafluorobenzene and carbazole.

FIGURE 2: (a) The optimized S 0 and S 1 geometries of D3CZ4F. (b) RDG Scattering diagrams based on optimized S 0 and S 1 geometries of D3CZ4F. (c) FMO distributions and calculated energy levels of D3CZ4F. FMO, Frontier molecular orbital; RDG, reduced density gradient.

In the S 1 state, there exists the remarkable orbital overlap and some separation between hole and particle distributions (Figure 3), illustrating that D3CZ4F exhibits the typical HLCT excited state characteristic where the proportions of LE and CT components are 65.97% and 34.03%, respectively. Such a weak CT component endows D3CZ4F with a high oscillator strength (f = 1.3989), which is favorable to obtain high PLQY. As shown in Figure 3 and Supporting Information S1: Table S1, D3CZ4F exhibits a large energy gap (ΔES1T1: 0.776 eV) and a small spin–orbit coupling (SOC) matrix element (0.051 cm−1) between S 1 and T 1, suggesting that the TADF mechanism involving the RISC process from T 1 to S 1 can be effectively ruled out. Notably, as evidenced by the hole and particle distributions, the high‐energy triplet states T 6–T 10 predominantly exhibit HLCT state character, which have minimal energy gaps (ranging from −0.020 to 0.194 eV) with S 1 state. In addition, the SOC matrix elements between S 1 and these triplets are large, reaching 0.166, 0.213, 0.194, 0.401, and 0.194 cm−1, respectively. These conditions collectively demonstrate the potential for facilitating the hRISC processes in the hot exciton channels from T 6–T 10 to S 1.

FIGURE 3: (a) The NTO analysis for S 0 → S 1 of D3CZ4F. (b) Proposed mechanism for the multi hRISC channels, NTO distributions, energy level diagram of excited states, and the SOC matrix elements between S 1 and T 6–T 10 states. hRISC, high‐lying reverse intersystem crossing; NTO, natural transition orbital; SOC, spin–orbit coupling.

Photophysical properties

As illustrated in Figure 4a, the UV‐Visible (UV‐vis) absorption and room‐temperature photoluminescence (PL) spectra in dilute tetrahydrofuran (THF, 1.0 × 10−5 M) of D3CZ4F were examined. Absorption bands centered at 294 and ∼320 nm are assigned to π–π* and n–π* transitions of the phenyl‐carbazole moiety, respectively. [40] The E g deduced from the absorption onset can be determined to be 3.47 eV. The PL spectrum shows a UV emission peak at 388 nm with an exceptionally high PLQY of 98%. As shown in Supporting Information S1: Figure S3, compared to the PL spectrum in air atmosphere, the PL intensity in nitrogen atmosphere is significantly enhanced, representing that the triplet excitons participate in the radiative decay process. Low‐temperature fluorescence and phosphorescence spectra allow the determination of the S 1 and T 1 energy levels as 3.33 and 2.56 eV, respectively, yielding a large ΔES1T1 of 0.77 eV, which implies that the TADF mechanism is difficult to realize. To probe the nature of the excited state, solvatochromic studies were conducted (Figure 4b and Supporting Information S1: Figure S4). With increasing solvent polarity, there is no significant red shift in the absorption peak, but the fluorescence peak redshifts from 367 nm with weak vibronic structures in n‐hexane solution to 398 nm with featureless emission profiles in acetonitrile solution, confirming a substantial CT component in the excited state.[ 41 , 42 ] Applying the Lippert‐Mataga equation to the experimental data (Figure 4c; Supporting Information S1: Table S2) reveals a single linear correlation across the entire polarity range, giving an excited state dipole moment (μ e) of 15.00 D, which is smaller than that of typical CT material 4‐(N,N‐dimethylamino)benzonitrile (≈23 D),[ 43 , 44 ] illustrating a quasi‐equivalent hybridization between LE and CT configurations.[ 31 , 45 , 46 , 47 , 48 ] Moreover, fluorescence decay profiles in both low and high polarity solvents display single‐exponential nanosecond‐level life‐times without any delayed components (Figure 4d), affirming that emission originates from a single HLCT state rather than mixed states between LE and CT components.

FIGURE 4: (a) UV–vis absorption and PL spectra at room temperature, and low‐temperature fluorescent and phosphorescent spectra at 77 K of D3CZ4F in THF solution (1.0 × 10−5 M). (b) Solvatochromic PL spectra. (c) Solvatochromic Lippert–Mataga model of D3CZ4F. (d) Transient PL decay spectra of D3CZ4F in different solvents (1.0 × 10−5 M). (e) PL spectra of D3CZ4F in non‐doped and doped films. (f) Transient PL decay spectra of D3CZ4F in non‐doped and doped films in vacuum. PL, photoluminescence; THF, tetrahydrofuran; UV, ultraviolet.

As shown in Figure 4e, D3CZ4F still maintains desirable UV emission with a maximum peak at 393 nm in a neat film, showing a slight bathochromic‐shift of 5 nm compared to that in THF, which may be ascribed to the suppressed intermolecular aggregation effect induced by the large torsion angles within phenylcarbazole segments. In the doped film with a concentration of 5 wt%, D3CZ4F exhibits a blue shifted UV emission peak at 379 nm with a FWHM of 47 nm, accompanied by a considerable PLQY of 86%. Transient PL decays of both neat and doped films under vacuum were performed in the nanosecond‐ and microsecond‐range detection window (Figure 4f, Supporting Information S1: Figure S5). There shows no microsecond scale delayed fluorescence; instead, they display biexponential nanosecond decays with the prompt and delay lifetimes of 1.10 and 4.48 ns for neat film, and 1.07 and 5.23 ns for doped film), respectively. The former can be attributed to direct singlet emission while the letter is attributed to the radiative transition of singlet excitons formed from the hRISC process. The radiative decay rates (k r) were calculated as 1.85 × 108 and 6.96 × 108 s−1, and the corresponding hRISC rates (k hRISC) reach 2.7 × 107 and 2.2 × 107 s−1, for neat and doped films, respectively (Table 1, Supporting Information S1: Table S3). These rapid k r and k hRISC can effectively reduce the accumulation of excitons in the light‐emitting layer, thereby reducing the efficiency roll off in the OLEDs.[ 49 , 50 , 51 ] Notably, compared to 3,3′‐(2,5‐difluoro‐1,4‐phenylene)bis(9‐phenyl‐9_H_‐carbazole) (D3CZ2F) lacking the two fluorine atoms, [52] the k r and k hRISC for the D3CZ4F doped film were significantly improved, leading to a higher PLQY, which further verified the effect of multiple C‐H···F interactions in D3CZ4F.

Compound | Solutiona | Neatb/dopedc film | HOMO/LUMO/E g d [eV] | ES1/ET1/ΔES1T1 e [eV]
λ Abs [nm] | λ Fluo [nm] | PLQY [%] | λ Fluo [nm] | PLQY [%] | τ [ns]
D3CZ4F | 294 | 388 | 98 | 393/379 | 25/86 | 1.73/1.63 | −5.27/−1.80/3.47 | 3.33/2.56/0.77

EL performance

Considering the high thermal stability, excellent PLQY, large k r, and efficient hRISC of D3CZ4F, we fabricated non‐doped and doped bottom‐emitting devices (N1 and D1) to characterize its EL characteristics with the following architecture: indium tin oxide/1,4,5,8,9,11‐hexaazatriphenylenehexacarbonitrile (ITO/HATCN) (20 nm)/di‐(4‐(N,N‐dimethyl‐4‐amino)‐phenyl)cyclohexane (TAPC) (45 nm)/tri(4‐carbazolyl‐9‐ylphenyl)amine (TCTA) (10 nm)/emissive layer (20 nm)/1,3,5‐tris(1‐phenyl‐1_H_‐benzimidazol‐2‐yl)benzene (TPBi) (30 nm)/lithium fluoride (LiF) (1 nm)/Al (100 nm). Among them, ITO and aluminum (Al) served as anode and cathode; HATCN and LiF were used as hole and electron injection layers; TAPC and TPBi were functioned as hole and electron transport layers; TCTA acted as an exciton blocking layer. The emissive layer of the doped device D1 comprises non‐polar 4,4'‐bis(9‐carbazolyl)‐1,1'‐biphenyl (CBP), as the host matrix with a low doping concentration of 5 wt%, which can further alleviate the exciton quenching induced by intermolecular host‐guest dipole‐dipole interactions. The corresponding energy‐level alignment and layer‐by‐layer layout are depicted in Figure 5 and Supporting Information S1: Figure S6. The EL performances are shown in Figure 5, Supporting Information S1: Figure S7 and summarized in Table 2.

FIGURE 5: (a) Energy level alignment of the materials used in the OLEDs. (b) Current density voltage‒luminance characteristics of the devices N1 and D1. (c) EQE‐luminance characteristics of the devices N1 and D1. (d) EL spectra of the device N1 at different voltages (Insert: CIE coordinate at 5 V). (e) EL spectra of the device D1 at different voltages (Insert: CIE coordinate at 5 V). (f) Summary of EQEmax and CIEy the representative reported non‐doped UV‐OLEDs with an EL peak ≤ 400 nm. CIE, Commission Internationale de I’Eclairage; EL, electroluminescence; EQE, external quantum efficiency; UV‐OLEDs, ultraviolet organic light‐emitting diodes.

Devices | V on [V] | CEmax [cd A−1] | PEmax [lm W−1] | EQEmax/500/1000 [%] | λ EL [nm] | FWHM [nm] | CIE (x, y) | R max [mW cm−2] | R UV a [mW cm−2]
N1 | 3.4 | 0.55 | 0.51 | 7.33/5.40/2.82 | 390 | 43 | (0.164, 0.027) | 58.47 | 34.50
D1 | 3.6 | 0.54 | 0.24 | 8.97/8.15/6.63 | 388 | 42 | (0.165, 0.025) | 150.81 | 96.52

Owing to the well‐matched energy level alignment, devices N1 and D1 exhibit low turn‐on voltages of 3.4 and 3.6 V, with the maximum luminances of 1015 and 2179 cd m−2, respectively. As shown in Supporting Information S1: Figure S8, the luminance‐current density plots of devices N1 and D1 exhibit a good linear relationship with correlation coefficients exceeding 0.99, illustrating that the contribution of the TTA mechanism to triplet exciton utilization can be ruled out. [53] Without out‐coupling enhancement or encapsulation, the non‐doped device N1 achieved maximum forward‐viewing efficiencies of 7.33% for EQE, 0.55 cd A−1 for current efficiency, and 0.51 lm W−1 for power efficiency at a luminance of 45.96 cd m−2. Remarkably, the EQE value of device N1 ranks among the highest reported for non‐doped UV OLEDs emitting below 400 nm (Figure 5f and Supporting Information S1: Table S4), which can be attributed to a high horizontal dipole orientation (84.0%) in the neat film of D3CZ4F (Supporting Information S1: Figure S9). Device D1 achieves a higher EQE of 8.97% at 323.50 cd m−2. The EQE remained at 8.15% and 6.63% at a high luminance of 500 and 1000 cd m−2, respectively, exhibiting a small roll‐off. Such small roll‐off can be contributed to the high oscillator strength of D3CZ4F, faster radiative transition process in the doped film and suppressed exciton quenching in the host‐guest system. Device N1 delivers stable UV emission with a peak at 390 nm and a narrow FWHM of 43 nm at 5 V, corresponding to CIE coordinates of (0.164, 0.027). The device D1 emits a blue‐shift UV light peaking at 388 nm with an FWHM of 42 nm and excellent color coordinates of (0.165, 0.025) (Supporting Information S1: Figure S10). To further evaluate the purity of UV‐OLEDs, here the parameter (UV400) is defined as the area proportion of the UV emission (λ ≤ 400 nm) in the EL spectrum. The device N1 and D1 exhibited high UV400 values of 59% and 64%, manifesting a high UV purity. The maximum radiant power (R max) values can be evaluated to be 58.47 mW cm−2 for device N1, 150.81 mW cm−2 for device D1, corresponding to the radiant power values from UV emission (R UVs) of 34.50 and 96.52 mW cm−2, respectively. It is worth noting that the high EQE of 8.15% at 500 cd m−2 achieved in the device D1 represents one of the highest efficiencies among reported UV‐OLEDs with EL peaks below 390 nm (Supporting Information S1: Table S5), making UV‐OLEDs potentially suitable for high brightness applications. However, owing to the current experimental constraints, the operational stability of the devices was not systematically evaluated in this work. For UV‐OLEDs, the device lifetime is critically compromised by a multitude of factors, including high‐energy exciton‐annihilation induced bond cleavage, Joule heating, morphological instability, material purity, device architecture, as well as fabrication and encapsulation protocols. Achieving commercially viable UV‐OLEDs ultimately demands a unified effort spanning fundamental research and industrial development.