Efficient solution-processed light-emitting diodes based on organic-inorganic hybrid antimony halides

Abstract

Organic-inorganic hybrid antimony halides are emerging emitters for solution-processed light-emitting diodes. However, achieving high-efficiency electroluminescence remains challenging resulting from the non-radiative recombination within emitters and inferior charge transport within device. Here, we develop an organic cation engineering to design a carbazole-functionalized triphenyl(9-ethyl-9H-carbazol-3-yl) phosphonium (TPPEtCz⁺), which enables a (TPPEtCz)₂Sb₂Br₈ film with good luminescence and achieves an improved charge transport within device. The TPPEtCz⁺ facilitates strong hydrogen bonding with the [Sb₂Br₈]²⁻ species and dichloromethane solvent, resulting in a more complete crystal restructuring, thus improving the quality of films. Moreover, non-covalent π–π interactions between carbazole moieties of (TPPEtCz)₂Sb₂Br₈ and benzimidazole moieties of electron-transport TPBi modify the interfacial contact that promotes electron transport and injection. Consequently, our light-emitting diodes reach a peak external quantum efficiency of 19.4% and half-lifetime of 10,190 min at 100 cd m⁻². These discoveries provide critical insights into the cation design of hybrid devices that are promising for practical applications.

Ma et al. report an organic cation engineering strategy to improve the luminescent properties and charge transport in organic-inorganic antimony halide films, enabling red LEDs with efficiency up to 19.4% and half-lifetime of 10,190 min. Large-area (12.15 cm²) LEDs show efficiency of 14.2%.

Introduction

Solution-processed light-emitting diodes (LEDs) are highly promising next-generation light source technology owing to their simple and low-cost fabricating processes^1–5. Many types of solution-processed materials and device configurations have been developed so far, including organic LEDs⁶, colloidal core-shell quantum dot LEDs⁷, perovskite LEDs⁸, and organic-inorganic hybrid LEDs⁹. Among them, organic-inorganic hybrid metal halides are attractive because they can integrate the semiconductor properties of inorganic units and the structural diversity of organic components, which is unavailable in classical all-inorganic materials and organic chromophores^10–12. Particularly, zero-dimensional (0D) hybrid antimony (Sb) halides, as a new kind of emitter in organic-inorganic material systems, are recognized for their superior photoluminescence quantum efficiency (PLQY) and stability, by virtue of their quantum-confined structure at the molecular level^13–16. With strategies on designing the host-guest structures and functionalizing the organic components, the achievement of Sb halide materials for LEDs is increasingly demonstrated. Nevertheless, the highest reported external quantum efficiency (EQE) based on hybrid Sb halides is only 5.12%¹⁶, even though the corresponding PLQY of their emitters reaches an impressive 86.1%. This pronounced discrepancy between photoluminescent and electroluminescent efficiency reveals that the primary limitation lies not in the luminescent properties of the emitters, but in the inferior charge transport and injection within the device¹⁷. An additional critical factor is the serious non-radiative recombination in the Sb halide emitter due to the presence of high-density defects caused by the fast crystallization process^4,15. Enhancing electroluminescence efficiency beyond current limitations is feasible via the coordinated efforts in both material and device structure design, probably through the creation of low-defect emitters and the development of advanced interfacial engineering.

In this work, we design a carbazole-functionalized triphenyl(9-ethylcarbazole) phosphonium (TPPEtCz⁺) cation, which not only enables the formation of (TPPEtCz)₂Sb₂Br₈ films with good luminescence and few defects but also facilitates π–π stacking interactions with the electron-transport TPBi to modify the interfacial contact. The TPPEtCz⁺ is conducive to form strong hydrogen bonding with the [Sb₂Br₈]²⁻ species and dichloromethane solvent, which results in a slow-release crystallization of (TPPEtCz)₂Sb₂Br₈ films, resulting in improved surface morphology and radiative recombination. The π–π stacking interaction between the carbazole moieties of (TPPEtCz)₂Sb₂Br₈ and the benzimidazole moieties of electron-transport TPBi also promotes the transport and injection of electrons within the device. These synergistic combinations lead to efficient Sb halide LEDs with a maximum EQE of 19.4% and a half-lifetime of 10,190 min at 100 cd m⁻², representing a significant breakthrough in organic-inorganic hybrid LEDs systems.

Results

Organic cation design

We designed two similar organic cations—that is, ethyltriphenyl phosphonium (TPPEt⁺) and triphenyl(9-ethylcarbazole) phosphonium (TPPEtCz⁺), for fabricating Sb halide LEDs (Fig. 1a, b). Both cations share an identical triphenylphosphine core, but TPPEtCz⁺ features an additional ethylcarbazole unit within the side chain. We synthesized the carbazole-functionalized TPPEtCzBr using a substitution reaction, and their structure was validated by ¹H and ¹³C nuclear magnetic resonance (NMR) and high-resolution mass spectral data (Supplementary Fig. 1)^18–20. Based on TPPEtCzBr and TPPEtBr organic salts, we synthesized the (TPPEtCz)₂Sb₂Br₈ (referred to as target sample) and (TPPEt)₂SbBr₅ (referred to as control sample) single crystals (SCs) by antisolvent vapor diffusion method²¹, and their crystal structures were verified using single-crystal X-ray diffraction (SCXRD) (Supplementary Fig. 2 and Supplementary Tables 1, 2). For the (TPPEt)₂SbBr₅, each Sb atom is bound to five Br atoms forming a square [SbBr₅]²⁻ pyramid. While two square pyramidal structures in (TPPEtCz)₂Sb₂Br₈ are linked by edge sharing to form a [Sb₂Br₈]²⁻ dimer. Differential scanning calorimetry analysis (Supplementary Fig. 3) shows that (TPPEtCz)₂Sb₂Br₈ exhibited a melting point of 167.1 °C, emphasizing its superior thermal stability compared to (TPPEt)₂SbBr₅ with a melting point of 149.7 °C. The XRD patterns of (TPPEtCz)₂Sb₂Br₈ and (TPPEt)₂SbBr₅ powders match well with the simulated results from SCXRD data (Supplementary Fig. 4), indicating the high phase purity of both samples.

Fig. 1. Design concept and EL performance of Sb halide LEDs.

Schematic diagram of device structure based on control (TPPEt)₂SbBr₅ (a) and target (TPPEtCz)₂Sb₂Br₈ (b) emitters, highlighting the chemical structures of organic cation. c Electrostatic surface potential of (TPPEt)₂SbBr₅, TPBi, and (TPPEtCz)₂Sb₂Br₈, highlighting the interfacial interaction between emitters and TPBi. Current density−voltage−luminance curves (d) and EQE−current density curves (e) of (TPPEt)₂SbBr₅ and (TPPEtCz)₂Sb₂Br₈-based LEDs. f Maximum EQE of state-of-the-art lead-free metal halide LEDs (Supplementary Table 3 for details). g Statistical maximum EQE histograms of (TPPEt)₂SbBr₅ and (TPPEtCz)₂Sb₂Br₈-based LEDs. h Half-lifetime measurement of two LEDs.

Two Sb halide emissive layers were prepared by a spin-coating method using the same precursor solution as the preparation of SCs. The device architecture is composed of indium tin oxide (ITO)/poly(3,4ethylene-dioxythiophene)-poly(styrenesulfonate) (PEDOT:PSS)/Sb halide/1,3,5-tri(phenyl-2-benzimi-dazolyl)-benzene (TPBi)/lithium fluoride (LiF)/aluminium (Al) (Supplementary Fig. 5). At the molecular level, both (TPPEtCz)₂Sb₂Br₈ and TPBi contain planar π−conjugated aromatic rings of carbazole and benzimidazole moieties, respectively. Electrostatic surface potential (ESP) analysis reveals that the central regions of these aromatic rings form electron-rich (negative) centers, while the hydrogen atoms at the peripheral positions exhibit electron-deficient (positive) characteristics (Fig. 1c). These π−conjugated aromatic rings with complementary charge distribution create favorable conditions for intermolecular π − π stacking between (TPPEtCz)₂Sb₂Br₈ and TPBi, which provides an effective pathway for enhanced charge transport¹⁸. In contrast, the ethyl-terminated surface of control (TPPEt)₂SbBr₅ lacks extended conjugated planes, making effective π − π interaction with TPBi highly improbable. Furthermore, the bulky TPPEtCz⁺ cation introduces significant steric hindrance that is expected to modulate crystallization kinetics, thereby facilitating the formation of highly uniform emissive films with improved morphological properties¹⁹.

LEDs performance

Figure 1d shows the current density−voltage−luminance curves of two LEDs. The (TPPEtCz)₂Sb₂Br₈-based LEDs exhibit a lower turn-on voltage and higher current densities as well as luminance compared with the control device, which could be ascribed to the improved charge transport and injection. A maximum luminance of 3669 cd m⁻² and a peak EQE of 19.4% were realized for (TPPEtCz)₂Sb₂Br₈-based LEDs. By contrast, the control (TPPEt)₂SbBr₅-based devices had a maximum luminance of 34.0 cd m⁻² and a peak EQE of only 1.1% (Fig. 1e). The high EQE for (TPPEtCz)₂Sb₂Br₈-based LEDs represents one of the most efficient lead-free metal halide LEDs achieved so far (Fig. 1f and Supplementary Table 3). The histogram of the peak EQE for 30 devices shows an average EQE of 18.5% (Fig. 1g), indicating a reasonable reproducibility associated with the use of TPPEtCz⁺ cations. Supplementary Fig. 6 shows the EL spectra of two devices under various voltages, which exhibit a monotonic increase in EL intensity with a superior spectral stability. A slight redshift was detected in the EL spectra compared to the PL spectra, which might be caused by the Stark effect resulting from the electric field¹⁰.

Operational stability of two devices was evaluated at a constant current density (Fig. 1h). The device based on (TPPEtCz)₂Sb₂Br₈ exhibited a half-lifetime (T₅₀) of 10,190 min from an initial luminance of 100 cd m⁻² (0.41 mA cm⁻²). This represents a substantial improvement over the control device, which showed a T₅₀ of only 11 min at an initial luminance of 30 cd m⁻² (4.32 mA cm⁻²). Note that this is the highest T₅₀ reported to date for all lead-free metal halide LEDs (Supplementary Table 3), which is also significantly better than the most of thriving red-emitting perovskite LEDs (Supplementary Table 4). Furthermore, the ultraviolet-cured epoxy-encapsulated (TPPEtCz)₂Sb₂Br₈ devices showed good storage stability, with luminance and EQE remaining stable for 30 days under ambient storage conditions (25 °C, 40−60% humidity) (Supplementary Fig. 7).

Electronic structure properties

Density functional theory calculations^1,22,23 were conducted to investigate the role of organic cations in the electronic structure of the Sb-based halides. The projected density of states (pDOS) for (TPPEt)₂SbBr₅ indicate that the hybridized Sb 5s and Br 4p orbitals constitute the valence band maximum (VBM), whereas the conduction band minimum (CBM) originates largely from the TPPEt⁺ cations (Fig. 2a and Supplementary Fig. 8a). This electronic configuration results in a type-II band alignment between TPPEt⁺ cations and [SbBr₅]²⁻ polyhedrons (Fig. 2b), wherein the photoexcited electrons in [SbBr₅]²⁻ transfer to TPPEt⁺ energy levels, undergoing the non-radiative recombination via thermal vibrations²⁴. In contrast, both VBM and CBM of (TPPEtCz)₂Sb₂Br₈ are mainly composed of Sb and Br orbitals, while the large organic TPPEtCz⁺ plays a separator to isolate the inorganic [Sb₂Br₈]²⁻ parts (Fig. 2c and Supplementary Fig. 8b). These results indicate that the excitons are confined on the inorganic [Sb₂Br₈]²⁻ polyhedrons, which constitutes a type-I band alignment with strong exciton localization that readily favors the efficient radiative recombination²⁵ (Fig. 2d). Electronic charge density distributions also suggest that the electron density is mostly located within [Sb₂Br₈]²⁻ polyhedrons for (TPPEtCz)₂Sb₂Br₈, while for (TPPEt)₂SbBr₅, part of the electron density is distributed in the organic fragment (Supplementary Fig. 9).

Fig. 2. Electronic structure, optical, and morphological characteristics of Sb halide films.

a–d pDOS diagrams of (TPPEt)₂SbBr₅ (a) and (TPPEtCz)₂Sb₂Br₈ (c) with emphasizing the charge density plots for VBM and CBM, respectively. Illustration of the energy level alignments of (TPPEt)₂SbBr₅ (b) and (TPPEtCz)₂Sb₂Br₈ (d), including the evolution of exciton formation, transfer, and radiative recombination processes. UV-visible absorption and PL spectra (e), PLQY (f), pseudo color mapping of time-resolved PL spectra (g) of two films. In-situ PL spectra (h) of two films during the spin-coating process, with the PL intensity evolution (i) extracted from the in-situ PL spectra. j AFM images of two films. Scale bars: 1 µm. k GIWAXS patterns of two films.

Photophysical properties

We investigated the influence of organic cation on the optical properties of two films. Compared with the control (TPPEt)₂SbBr₅ films, the absorption edge and PL peak of (TPPEtCz)₂Sb₂Br₈ films show a slight blue shift (Fig. 2e), possibly due to the structural distortion effect of [Sb₂Br₈]²⁻ dimers²⁶. Two samples are both characterized by broadband red emission, suggesting that their luminescence originates from the self-trapped excitons (STEs)-related recombination as demonstrated in previous work^27–31. We also found that the (TPPEtCz)₂Sb₂Br₈ sample features a narrower full-width at half-maximum (FWHM) and a smaller Stokes shift than the (TPPEt)₂SbBr₅ (Supplementary Fig. 10), which is attributed to the weaker electron–phonon coupling and reduced lattice distortion of the (TPPEtCz)₂Sb₂Br₈ (Supplementary Figs. 11, 12, Supplementary Tables 5, 6, and Supplementary Note 1). Note that the absolute PLQY (87.6%) of (TPPEtCz)₂Sb₂Br₈ films was significantly improved compared to the control (TPPEt)₂SbBr₅ films (20.2%) (Fig. 2f and Supplementary Fig. 13), implying that the TPPEtCz⁺ cation enables fewer non-radiative recombination sites. Figure 2g shows the time-resolved PL spectra of two films, which can be fitted using a biexponential decay function¹⁰ (Supplementary Note 2). The fitted average lifetime of (TPPEtCz)₂Sb₂Br₈ films is 3.57 μs, longer than that of the control counterpart (1.23 μs), which is consistent with the increased PLQY. Compared with the control films, the (TPPEtCz)₂Sb₂Br₈ films have a smaller fraction of fast decay and a larger fraction of slow decay, evidently denoting more radiative recombination component and fewer defects using TPPEtCz⁺ cation. Moreover, the radiative recombination rate of (TPPEtCz)₂Sb₂Br₈ films was estimated to be 2.45 × 10⁷s⁻¹ (Supplementary Fig. 14 and Supplementary Note 3), nearly 1.5 times higher than that of the (TPPEt)₂SbBr₅ films (1.64 × 10⁷s⁻¹), and the non-radiative recombination rate markedly decreases from 6.49 × 10⁷s⁻¹ (control) to 0.35 × 10⁷s⁻¹ (target) simultaneously.

Origins of luminescence enhancement

To investigate how the TPPEtCz⁺ cation affects the crystallization process of (TPPEtCz)₂Sb₂Br₈ films, in-situ PL and UV-vis absorption measurements were performed during the spin-coating and annealing process (Supplementary Fig. 15)³². The initial PL peak position (<420 nm) is correlated with the organic cations. As shown in Fig. 2h, for the (TPPEt)₂SbBr₅, the PL intensity of TPPEt⁺ cations gradually decreases from the beginning of the spin-coating process, and the PL signal peaked at 680 nm from (TPPEt)₂SbBr₅ appears at about 3.86 s, which corresponds to the nucleation stage of [SbBr₅]²⁻ clusters (Phase I). Following closely behind is a rapid increase of PL intensity, indicating that after the crystal nucleus concentration reaches a critical value, the nuclei undergo explosive growth, corresponding to the crystal nuclei growth stage (Phase II, 0.53 s). Finally, the PL intensity shows a slow upward trend until the end of spin-coating, reflecting an enhanced crystallization via the Ostwald ripening process (Phase III, 0.67 s). In contrast, the utilization of the TPPEtCz⁺ cation notably delayed all three stages in the (TPPEtCz)₂Sb₂Br₈ films, including nucleation (Phase I, 4.23 s), crystal growth (Phase II, 1.82 s), and Ostwald ripening process (Phase III, 2.63 s). This delayed crystallization process can also be validated by the in-situ UV-vis absorption spectra in Supplementary Fig. 16. During the annealing process, the in-situ PL intensities of two films decreased slightly over time without any peak shift (Supplementary Fig. 17), indicating that crystal nucleation and growth were essentially completed during the spin-coating phase, which is attributed to the rapid solvent evaporation of dichloromethane used in our experiment with a low boiling point. We employed the Avrami model to investigate the crystallization rate of (TPPEt)₂SbBr₅ and (TPPEtCz)₂Sb₂Br₈ films based on the PL intensity evolution during the spin-coating process. The model is defined as kt = (−ln(1−α))^1/n, where k is the effective reaction rate constant (crystallization rate constant herein), n is the Avrami exponent, t is time, and α is the conversion fraction characterized by the PL intensity of the films³². From the PL intensity-time curves (Fig. 2i), two rate constants k₁ and k₂ can be fitted, which k₁ corresponds to the growth of the crystal nucleus, while k₂ represents the Ostwald ripening process. In comparison, k₁ (0.187 s⁻¹) and k₂ (0.018 s⁻¹) values of the (TPPEtCz)₂Sb₂Br₈ are smaller than those of the control (TPPEt)₂SbBr₅ (k₁ = 0.246 s⁻¹, k₂ = 0.096 s⁻¹). These specifical results show that TPPEtCz⁺ cations effectively retard the crystallization rate of the crystal, yielding compact films with reduced trap density and superior luminescent properties^33,34. The space-charge-limited-current (SCLC) measurements^35,36 (Supplementary Fig. 18 and Supplementary Note 4) show that the trap state density (N_t)^37–39 of (TPPEtCz)₂Sb₂Br₈ films is much lower than that of the control films (Supplementary Fig. 19), manifesting that the defects were effectively suppressed within (TPPEtCz)₂Sb₂Br₈ films, in line with the enhanced PL performance.

We used NMR spectroscopy to probe the interactions between inorganic and organic components in the precursor solution to investigate the reasons for the modulation of crystallization kinetics by organic cations. The ¹H NMR signals of pure TPPEtBr exhibit a downfield shift compared to (TPPEt)₂SbBr₅, indicating a hydrogen bond interaction between TPPEt⁺ and [SbBr₅]²⁻ clusters (Supplementary Fig. 20a). Importantly, the ¹H NMR peaks of pure TPPEtCzBr in (TPPEtCz)₂Sb₂Br₈ also experience a downfield shift, but with a significantly larger magnitude than that of the (TPPEt)₂SbBr₅ (Supplementary Fig. 20b), indicating that TPPEtCz⁺ has a stronger hydrogen bond interaction with [Sb₂Br₈]²⁻. This stronger hydrogen-bonding interaction slows down the crystal growth, enabling the acquisition of high-quality (TPPEtCz)₂Sb₂Br₈ films with few defects³⁵. 2D Hirshfeld surfaces and fingerprint plots (Supplementary Fig. 21) of two Sb halide structures show that H···Br hydrogen bonds and H···Sb interactions account for 91.5 and 3.9% of the total intermolecular interactions in (TPPEtCz)₂Sb₂Br₈, respectively, markedly higher than those in the control (TPPEt)₂SbBr₅ (H···Br = 85.4%, H···Sb = 2%). Moreover, we examined the solvent-ligand interactions between dichloromethane (DCM) and organic components in the precursor solution. The ¹H NMR results show that, compared with TPPEt⁺, the stronger hydrogen interaction between the carbazole moiety in TPPEtCz⁺ and DCM likely enhances the solvation effect (Supplementary Fig. 22), which moderates the crystallization process, working in synergy with the hydrogen-bonding effects between the organic and [Sb₂Br₈]²⁻ to modulate the crystallization kinetics.

Scanning electron microscopy (SEM) images show that the control (TPPEt)₂SbBr₅ films exhibit a poor microscopic morphology with a heterogeneous appearance due to its rapid crystallization, while the (TPPEtCz)₂Sb₂Br₈ films are highly dense and uniform (Supplementary Fig. 23). Atomic force microscopy (AFM) images show that the (TPPEtCz)₂Sb₂Br₈ films present a smoother surface with a root-mean-square roughness of ~0.7 nm, significantly lower than that of the control films (~6.2 nm) (Fig. 2j). From the confocal PL mapping images (Supplementary Fig. 24a, b), the (TPPEtCz)₂Sb₂Br₈ films display much stronger PL intensity with consistent fluorescence, which is related to better films quality and lower surface defect density that diminish the non-radiative recombination within films³². The PL lifetime imaging results also show a prolonged exciton lifetime with homogeneous distribution for (TPPEtCz)₂Sb₂Br₈ films (Supplementary Fig. 24c, d). Grazing-incidence wide-angle X-ray scattering (GIWAXS) analysis uncovered the crystal planes alignments in the films (Fig. 2k). The appearance of Debye-Scherrer rings in the (TPPEt)₂SbBr₅ films indicated a random crystallographic orientation, consistent with a disordered crystalline packing. In contrast, (TPPEtCz)₂Sb₂Br₈ films revealed a focused diffraction spot, evidencing a preferential crystalline orientation. These results suggest that the TPPEtCz⁺ cation is beneficial for crystal nucleation, growth, and crystallization processes, potentially improving its optoelectronic performances, which is illustrated schematically in Supplementary Fig. 25.

Carrier transport properties

As mentioned earlier, the presence of planar π–conjugated aromatic rings in both (TPPEtCz)₂Sb₂Br₈ and TPBi creates favorable conditions for intermolecular π–π stacking, providing an effective pathway for improved charge transport. We performed Kelvin probe force microscopy (KPFM) and ultraviolet photoelectron spectroscopy (UPS) measurements^40,41 to characterize the charge transport properties of two samples. A thin layer of TPBi (2 nm) was deposited on (TPPEtCz)₂Sb₂Br₈ to probe their interfacial interaction. As shown in Fig. 3a, the average contact potential difference (V_CPD) of the pure (TPPEtCz)₂Sb₂Br₈ films was 302.6 mV, which increased significantly to 330.4 mV for TPBi-treated (TPPEtCz)₂Sb₂Br₈ films, indicating a reduction in the work function (W_F) after TPBi treatment that promotes more efficient electron injection from TPBi into the emitter layer. According to the cutoff region of the UPS spectra, the W_F of TPBi-treated (TPPEtCz)₂Sb₂Br₈ and (TPPEtCz)₂Sb₂Br₈ films is estimated to be 4.83 and 4.88 eV, respectively (Supplementary Fig. 26b). The reduction in W_F arising from the TPBi/(TPPEtCz)₂Sb₂Br₈-induced π − π stacking facilitates the electron transport within the emitter^36,39. Additionally, this interaction is associated with an upward shift in the emitter layer’s energy levels, where the VBM rises from −5.82 to −5.66 eV and the CBM from –3.26 to −3.10 eV. This collective energy level upshift optimizes the band alignment, reducing charge injection barriers and alleviating interfacial charge accumulation to enable more efficient hole and electron injection (Supplementary Fig. 26c). On the contrary, the V_CPD, W_F, and energy level structure of the control (TPPEt)₂SbBr₅ films and TPBi-treated (TPPEt)₂SbBr₅ films remained almost unchanged (Supplementary Fig. 27) due to the lack of π − π stacking interaction between the (TPPEt)₂SbBr₅ and TPBi.

Fig. 3. Influence of π − π interaction on charge transport.

a KPFM images of (TPPEtCz)₂Sb₂Br₈ and TPBi-treated (TPPEtCz)₂Sb₂Br₈ films. Scale bars: 1 µm. b EIS curves of devices based on (TPPEt)₂SbBr₅ and (TPPEtCz)₂Sb₂Br₈ films. c Low-frequency (f = 5 kHz) C–V curves of two LEDs. N 1s fine spectra (d), FTIR spectra (e), and ¹H NMR spectra (f) of pure TPBi, TPBi-treated (TPPEtCz)₂Sb₂Br₈, and pure (TPPEtCz)₂Sb₂Br₈. g Three adsorption model structures of TPBi and (TPPEtCz)₂Sb₂Br₈. h Comparison of adsorption energies (E_ads) of the three adsorption models.

Figure 3b displays the electrochemical impedance spectroscopy (EIS) of two devices. According to the equivalent circuit model, we estimated the series resistor (R_s), recombination resistor (R_rec), and capacitance (C_d) of them. The (TPPEtCz)₂Sb₂Br₈-based devices with π − π interaction exhibit a reduction of R_s from 34 to 29 Ω, which is beneficial to the electronic charge injection and transport. Moreover, R_rec was greatly decreased from 47 to 1.4 kΩ, which facilitates the electron-hole recombination within devices. The capacitance-voltage (C-V) curves^42,43 of both devices were also measured (Supplementary Note 5). The (TPPEtCz)₂Sb₂Br₈-based LEDs have a smaller peak capacitance accompanied by a lower peak voltage compared to the control devices (Fig. 3c), revealing less traps in the emissive films. In addition, the slope of (TPPEtCz)₂Sb₂Br₈-based LEDs in the recombination region is steeper than that of the control devices, signifying a rapid radiative recombination rate. The carrier dynamic processes of LEDs were studied by the transient EL test (Supplementary Fig. 28). (TPPEtCz)₂Sb₂Br₈-based LEDs show a steeper upward trend at the beginning compared to the case of control devices, which manifests a rapid establishment of steady-state EL⁴⁴. When the pulse voltage was deactivated, a fast EL decay was detected for (TPPEtCz)₂Sb₂Br₈-based LEDs, which is attributed to the fewer cumulative carriers caused by balanced injection and rapid recombination⁴⁵ of carriers through interfacial π − π interaction.

To gain theoretical insight into carrier dynamics, we conducted semiconductor device simulations⁴¹ of two devices. The carrier concentration profiles (Supplementary Fig. 29a, b) yield hole and electron concentrations of 1.08 × 10⁸cm⁻³ and 2.60 × 10²²cm⁻³, respectively, for the (TPPEt)₂SbBr₅-based devices. In contrast, the (TPPEtCz)₂Sb₂Br₈-based devices show significantly higher values of 2.48 × 10¹¹cm⁻³ and 3.62 × 10²³cm⁻³. This indicates a more efficient and balanced charge transport in the latter system (Supplementary Fig. 29c), which benefits from the π − π interactions. Supplementary Figs. 29 d, e display the simulated energy band distribution of two devices in equilibrium state, where the energy barrier of the (TPPEtCz)₂Sb₂Br₈-based devices (φ_h = 0.96 eV; φ_e = 0.51 eV) was lower than that of the control ones (φ_h = 1.23 eV; φ_e = 0.79 eV). In addition, the calculated exciton concentration and exciton recombination rate of (TPPEtCz)₂Sb₂Br₈-based devices were both higher than those of the control (TPPEt)₂SbBr₅-based devices (Supplementary Fig. 29 f). These changes improve the charge carrier transport and exciton radiative recombination in the device with interfacial π − π interaction, promoting the electro-optical energy conversion, resulting in enhanced EL performance of (TPPEtCz)₂Sb₂Br₈-based LEDs.

π−π interaction between (TPPEtCz)₂Sb₂Br₈ and TPBi

We conducted X-ray photoelectron spectroscopy (XPS), Fourier transform infrared spectroscopy (FTIR), and NMR measurements to reveal the mechanisms of π − π interactions between (TPPEtCz)₂Sb₂Br₈ and TPBi. In the high-resolution XPS spectrum of N 1s (Fig. 3d), the pure (TPPEtCz)₂Sb₂Br₈ films exhibit a single C−N peak (400.11 eV) from the carbazole moiety, while the pure TPBi films not only show C−N peak (400.60 eV) and C=N peak (398.48 eV) from benzimidazole moiety, but also display a characteristic π − π satellite peak (405.82 eV) of N-containing aromatic compounds. Importantly, in the TPBi-treated (TPPEtCz)₂Sb₂Br₈ films, all relevant N 1s peaks, including the C−N peak of carbazole, the C−N peak and C=N peak of benzimidazole, undergo discernible shifts compared with their positions in the pure TPBi and (TPPEtCz)₂Sb₂Br₈ films, indicating an electronic interaction between TPBi and (TPPEtCz)₂Sb₂Br₈. Moreover, the π − π satellite peak is significantly enhanced and broadened in the TPBi-treated (TPPEtCz)₂Sb₂Br₈ films, confirming an existence of a parallel stacking structure dominated by π − π interactions. The FTIR spectra show that the C−N and C=N stretching vibrations of TPBi-treated (TPPEtCz)₂Sb₂Br₈ films shifted to higher wavenumbers relative to those in pure TPBi films (Fig. 3e), consistent with enhanced bond polarization resulting from π − π stacking between the conjugated benzimidazole and carbazole aromatic rings. Further confirmation of this π − π interactions is provided by proton ¹H NMR spectra (Fig. 3f and Supplementary Fig. 30). The chemical shift of benzimidazole moieties in the pure TPBi was located at 7.45–7.09 ppm (labeled as H₁₋₉), while these signals shift downfield to 7.42−7.02 ppm in the TPBi-treated (TPPEtCz)₂Sb₂Br₈ sample. This shift arises because the protons on the planar conjugated rings involved in π − π stacking enter the shielding region generated by another aromatic ring, indicating that the induced magnetic field from the ring current of the aromatic ring opposes the external magnetic field, offsetting part of the external field and requiring a higher external magnetic field to induce proton resonance. In contrast, the protons of the carbazole moieties in (TPPEtCz)₂Sb₂Br₈ (labeled as H_10–13) show the opposite trend, with their chemical shifts move upfield from 8.17–7.85 ppm to 8.23–7.90 ppm, indicating the formation of a shielding region generated by the carbazole-conjugated moieties. Collectively, these ¹H NMR results confirm that π − π interactions form via the stacking between benzimidazole (from TPBi) and carbazole (from (TPPEtCz)₂Sb₂Br₈) moieties.

Meanwhile, the same characterizations were conducted on control (TPPEt)₂SbBr₅ and TPBi-treated (TPPEt)₂SbBr₅ samples using XPS, FTIR, and NMR measurements. As summarized in Supplementary Fig. 31, all characteristic spectral signatures of the benzimidazole unit remain unchanged in both position and linewidth across all samples, indicating the absence of any detectable interaction between ethyl moieties of (TPPEt)₂SbBr₅ and benzimidazole moieties of TPBi. These results further indicate that the interfacial enhancement observed in the (TPPEtCz)₂Sb₂Br₈/TPBi system specifically arises from the π − π stacking interaction between the planar conjugated units, rather than non-specific surface contacts.

To further elucidate the π − π interactions, we theoretically studied the interaction between the TPBi layer and (TPPEtCz)₂Sb₂Br₈. We calculated the adsorption energy (E_ads) of TPBi on the (TPPEtCz)₂Sb₂Br₈ (001) surface to identify the thermodynamically favorable stacking configuration (Fig. 3g). Specifically, three adsorption models were established, including a vertical stacking configuration with the TPBi positioned directly above the (TPPEtCz)₂Sb₂Br₈, an in-plane parallel stacking configuration from intermolecular hydrogen bonding, and a parallel stacking configuration with the overlapping of aromatic rings of two molecules. The calculated E_ads is −0.28, −0.78, and −1.61 eV (Fig. 3h), respectively, indicating that the parallel stacking configuration with overlapping aromatic rings is the thermodynamically stable arrangement. It is widely acknowledged that this parallel stacking configuration of overlapped aromatic rings stems from the intermolecular π − π interactions³¹, indicating an existence of π − π interactions between (TPPEtCz)₂Sb₂Br₈ and TPBi.

We performed Independent Gradient Model (IGM) analysis to visualize the interaction pattern between (TPPEtCz)₂Sb₂Br₈ and TPBi molecule, which can enable 3D visualization of weak non-covalent interactions. Supplementary Fig. 32a shows the sign(λ₂)ρ colored isosurface maps of two systems derived from IGM analyses. We found that a broad green isosurface appears between carbazole and benzimidazole, corresponding to the sign(λ₂)ρ values of –0.02 ~0 a.u. (Supplementary Fig. 32b), which confirms a π − π stacking interaction between them.

Demonstration of large-area LEDs

Taking advantage of the favorable surface morphology with good luminescence of (TPPEtCz)₂Sb₂Br₈ films and improved charge transport caused by the (TPPEtCz)₂Sb₂Br₈/TPBi π − π interaction, we further fabricated large-area red LEDs. The resulting large-area (4.5 × 4.5 cm²) (TPPEtCz)₂Sb₂Br₈ films emit uniform and bright red emission under UV light irradiation (Fig. 4a). The PL distribution and surface roughness test demonstrates a homogeneous feature with small fluctuations throughout the entire films (Fig. 4b and Supplementary Fig. 33), illustrating an excellent uniformity. Further, large-area LEDs with an emitting area of 12.25 cm² were achieved, exhibiting a bright red light (Fig. 4c) and outstanding electroluminescent uniformity (Fig. 4d) under forward bias. The large-area devices can reach high EQE of 17.2% and luminance of 3641 cd m⁻², as well as outstanding repeatability (Fig. 4e, f).

Fig. 4. Demonstration of large-area devices.

a PL photograph of large-area (TPPEtCz)₂Sb₂Br₈ films with an emitting area of 4.5 × 4.5 cm² under UV light irradiation. b PL intensity distribution of large-area films. c Photograph of a large-area LED with an emitting area of 3.5 × 3.5 cm² under a bias voltage of 8.0 V. d Luminance uniformity statistics of the large-area LEDs. Current density−voltage−luminance (e) and EQE−current density (f) curves of large-area LEDs.

Universal characterization

To demonstrate the universality of our proposed carbazole-functionalized organic cation strategy, other carbazole derivatives, including triphenyl(9-methyl-9H-carbazol-3-yl)phosphonium bromide (TPPMeCzBr), triphenyl(9-phenyl-9H-carbazol-3-yl)phosphonium bromide (TPPPhCzBr), and triphenyl(9-(naphthalen-2-yl)−9H-carbazol-3-yl)phosphonium bromide (TPPNaCzBr), were designed and employed to fabricate the LEDs. The structures of these carbazole derivatives are shown in Supplementary Fig. 34a. Notably, the TPPPhCz⁺ organic cation generates a lower delayed crystallization rate compared to the TPPMeCz⁺ and TPPNaCz⁺ systems, which contributes to a lower defect density and a higher PLQY of TPPPhCz-based Sb halide films (Supplementary Figs. 35, 36, and Supplementary Note 6). Moreover, the electron cloud distribution and ESP of the three carbazole-functionalized organic cations reveal that the TPPNaCz-based Sb halide films possess lower electrical conductivity (Supplementary Figs. 37, 38, and Supplementary Note 7). As a result, the resulting devices based on the three carbazole-functionalized organic cations exhibit slight but discernible performance differences. Specifically, the maximum luminance and EQE are 3488 cd m⁻² and 17.2% for the TPPMeCz-based device, 3421 cd m⁻² and 16.5% for the TPPPhCz-based device, and 3312 cd m⁻² and 15.1% for the TPPNaCz-based device, respectively (Supplementary Fig. 34c−e). Despite these differences, all three carbazole derivatives-based devices significantly outperform the control TPPEt-based device, confirming the general applicability and effectiveness of the proposed strategy.

Discussion

In summary, we have revealed the critical role of carbazole-functionalized organic cations in inhibiting the non-radiative recombination of Sb halide films and enhancing the charge transport of the device. With this in mind, we have synthesized the (TPPEtCz)₂Sb₂Br₈ films by utilizing the carbazole-based TPPEtCz⁺, which demonstrated significantly improved radiative recombination efficiency and films quality. Concurrently, the π–π interaction between the carbazole moieties of (TPPEtCz)₂Sb₂Br₈ and the benzimidazole moieties of TPBi effectively regulated the (TPPEtCz)₂Sb₂Br₈/TPBi interface, promoting the transport and injection of electrons within the device. Finally, we have created efficient Sb halide LEDs featuring a peak EQE of 19.4%, an enlarged emitting area of 12.25 cm², and a long half-lifetime of 10,190 min, and also verified the universality of carbazole-functionalized organic cation strategy. This work offers original thinking for the rational design of organic cations for developing high-performance organic-inorganic hybrid metal halide LEDs.

Methods

Materials

Triphenylphosphine (TPP, 99%), ethyltriphenyl phosphonium bromide (TPPEtBr, 99%), 3-bromo-9-ethyl-9H-carbazole (EtCzBr, 99%), nickel bromide (NiBr₂, 99.9%), PEI (80% ethoxylated solution) dichloromethane (DCM, 99.5%), chlorobenzene (anhydrous, 99.8%), dimethylformamide (DMF, anhydrous), hydrobromic acid (48%), and diethyl ether (Et₂O, anhydrous) were purchased from Sigma-Aldrich. 3-bromo-9-methyl-9H-carbazole (MeCzBr, 99%), 3-bromo-9-phenyl-9H-carbazole (PhCzBr, 99%), 3-bromo-9-(2-naphthalenyl)−9H-carbazole (NaCzBr, 99%), antimony tribromide (SbBr₃, 99.9%) and MoO₃ (99.97%) were purchased from Aladdin. poly(ethylenedioxythiophene):polystyrenesulfonate (PEDOT:PSS, AI4083), ZnO nanoparticles (ethanol, 12%), 4,4′,4″-tris(carbazol-9-yl) triphenylamine (TCTA, 99%), 1,3,5-tris(1-phenyl-1H-benzimidazol-2-yl)benzene (TPBi, 99.9%), and lithium fluoride (LiF, 99.99%) were purchased from Xi’an Polymer Light Technology Corp. All chemicals were used directly without any purification unless otherwise stated.

Synthesis of TPPEtCzBr

The TPPEtCzBr was synthesized according to the substitution reaction. TPP (5.24 g, 0.02 mol), EtCzBr (5.48 g, 0.02 mol) and NiBr₂ (0.5 g, 0.002 mol) were added in 50 mL of ethylene glycol solution with stirring for 12 h at 170 °C. Then the deep green solution was dissolved in water and was then extracted with DCM. Et₂O was used to wash the white powder product out. Yield: 98%. TPPMeCzBr, TPPPhCzBr, and TPPNaCzBr were synthesized by similar methods as mentioned above.

Growth of (TPPEt)₂SbBr₅ and (TPPEtCz)₂Sb₂Br₈ single crystals

Two single crystals were synthesized via antisolvent vapor diffusion. Specifically, 0.1 mmol SbBr₃ and 0.2 mmol TPPEtCzBr (or TPPEt) were mixed with 1:2 molar ratio and dissolved in 1 mL DCM to form a clear precursor solution. Then 2 mL Et₂O was diffused into this precursor solution at room temperature overnight to grow Sb halide single crystals.

Device fabrication

The ITO-coated glass substrate was sequentially cleaned by sonication in acetone, ethanol, and deionized water for 15 min, respectively, and was further treated by ultraviolet ozone illumination for 20 min. Then, the PEDOT:PSS aqueous solution was spin-coating onto the substrate at 4000 rpm for 60 s in an air ambient and annealed at 135 °C for 30 min. After that, the substrate was transferred into an N₂-filled glovebox. The Sb halide precursor solution made from 0.01 mmol of SbBr₃ and 0.02 mmol of TPPEtCzBr (or TPPEtBr) with a molar ratio of 1: 2 in 1 mL DCM was spin-coated on top at 4000 rpm for 60 s (acceleration = 500 rpm s⁻¹). The precursor concentrations of large-area devices and universal devices were 0.008 mmol mL⁻¹ and 0.007 mmol mL⁻¹, respectively. The spin-coated adduct films were annealed at 100 °C for 10 min. Then TPBi (30 nm), LiF (2 nm), and Al (120 nm) layers were deposited onto the luminescent films in turn by a high-vacuum organic/metal evaporation instrument under a high vacuum of ~1 × 10⁻⁴ Pa. The overlapping area of the ITO and Al electrodes was used to define the working area of the LED. The emitting area of the small-area LEDs is 0.07 cm², while that of the large-area LEDs is 12.25 cm². Devices for capacitance, transient EL, and electrochemical impedance study were fabricated with a device structure of ITO/PEDOT:PSS/(TPPEt)₂SbBr₅ or (TPPEtCz)₂Sb₂Br₈/TPBi/LiF/Al. Hole-only and electron-only devices were fabricated using the device structures ITO/PEDOT:PSS/(TPPEt)₂SbBr₅ or (TPPEtCz)₂Sb₂Br₈/TCTA/MoO_x/Ag and ITO/ZnO/PEI/(TPPEt)₂SbBr₅ or (TPPEtCz)₂Sb₂Br₈/TPBi/LiF/Al, respectively.

Density functional theory calculations

The generalized gradient approximation (GGA) is chosen to approximate the exchange-correlation energy with the Perdew–Burke–Ernzerhof (PBE) functional^46,47. We use a cutoff energy of 400 eV and a 3 × 2 × 2 Monkhorst-Pack k-point mesh for all the calculations. The convergence threshold for each ionic step is set to be 1 × 10⁻⁵ eV, and the convergence threshold of maximum force is set to be 0.03 eV Å⁻¹. Spin polarization is excluded in our calculations. Each configuration was optimized with the Broyden–Fletcher–Goldfarb–Shanno (BGFS) algorithm with SCF convergence criteria of 1.0 × 10⁻⁵ a.u. The van der Waals corrections of Grimme’s DFT-D3 models were also adopted. The adsorption energy between the adsorbate and the Sb halides can be calculated using the equation of ΔE_ads = E_{adsorbate@Sb halides} − E_{Sb halides} − E_adsorbate, where the E_{adsorbate@Sb halides} and E_{Sb halides} are the total energies of the Sb halides with and without the adsorption of the adsorbate, respectively, and E_adsorbate is the total energy of the adsorbate. According to this equation, a negative adsorption energy corresponds to a stable adsorption structure. For the ESP calculations, structures of materials were optimized using the B3LYP functional and 6–31 G (d, p) basis set in the Gaussian 16 software package. The ESP images were visualized by using Multiwfn 3.8 and the Visual Molecular Dynamics (VMD) program, respectively.

Materials characterizations

¹H NMR B500 is equipped with a high-resolution 5 mm TXI (H-C/N-D) Zg probe. The SC-XRD data of Sb halide SCs were measured using a Bruker D8 Venture METALJET Ga-Target SC-XRD diffractometer with Mo and Cu microfocus sealed X-ray source. The powder XRD patterns of Sb halide powders were carried out from Panalytical X’Pert Pro X-ray diffractometer equipped with Cu Kα radiation. The morphologies and surface roughness of the Sb halide films were measured by SEM (Jeol-7500F, 15 keV), in which the distribution of the elements was analyzed by element distribution (EDS) measurement. KPFM measurements were carried out using a Bruker AFM Multimode model. Synchrotron-based GIWAXS measurements were conducted at the BL14B1 beamline using X-ray with a grazing incidence angle of 0.2° and a wavelength of 1.24 Å. The diffraction patterns were collected by the Pilatus3s 2 m at an exposure time of 100 s with 234 mm from the sample to the detector. FTIR spectra were tested from the Thermo Fisher Scientific Nicolet iS20 spectrometer. ¹H NMR spectra were acquired on the Bruker Advanced II spectrometer. The steady-state PL and absorption spectra of the samples were tested using a steady-state PL spectrum (Horiba; Fluorolog-3) and a Shimadzu UV-3150 spectrophotometer, respectively. Time-resolved PL spectra were conducted using a fluorescence lifetime measurement system with an LED (Horiba; 280 nm). The absolute PLQY of Sb halide films was tested by using a fluorescence spectrometer (Horiba; FluoroMax-4) with an integrated sphere (Horiba; Quanta-φ). A closed-cycle helium cryostat (Jannis, using CCS-100) was used for PL measurements at different temperatures. The in-situ PL measurements of films during the spin-coating process were performed on the self-built systems (Supplementary Fig. 15) that contain excitation light sources, QE-Pro spectrometer (Ocean Optics), and optical fibers. The fluorescence and lifetime distribution imaging were performed through the fluorescence-lifetime imaging microscopy (FLIM) systems. XPS measurements were conducted using a Thermo Scientific Escalab 250Xi. UPS measurements were performed to analyze the energy band properties of Sb halides, and the UPS radiation was generated by a He-gas discharge lamp (He I = 21.22 eV).

Device characterizations

The current density-voltage-luminance, EQE, and EL spectra of LEDs were tested by using a system combining a Keithley 2400 source meter and a standard integrating sphere (FOIS-1-FL) coupled with an QE-Pro spectrometer (Ocean Optics) in a N₂-filled glovebox. The performance of LEDs was double-checked using a Keithley 2400 source meter assembled with a PR705 Spectra Scan spectrophotometer (Photo Research), as shown in Supplementary Fig. 39. The EL performance of the devices provided in the main text is from integrating sphere measurement systems. The transient EL spectra were measured by the multifunctional carrier characteristic analysis system (Paios). The impedance spectroscopy and capacitance-voltage spectra of devices were tested using the Palos systems. The half-lifetime (T₅₀) of the device was tested under continuous current operation provided by a Keithley 2400 source meter. All measurements of devices were conducted at room temperature under N₂-filled glovebox.

Author contributions

C.S., J.P.W., and Z.S. conceived the idea for detailed experiments. M.Z., W.C., Q.Z., M.Z., J.M.W., X.L., and Z.X. performed the material preparation experiments. M.Z., S.L., M.W., X.J., and J.Z. carried out the X.R.D., S.E.M., and A.F.M. measurements. W.C., D.Z., D.Y., and M.Z. conducted the theoretical calculations. N.W., Y.L., Y.H., and L.L. performed the optical measurements. W.C., Q.P., S.L., M.J., X.C., X.L., D.W., and Z.S. conducted the device measurements and analyzed the data. The paper was cowritten by Z.M., W.C., Q.P., and Z.S. Z.S., C.S., and J.P.W. guided the whole project. All authors read and commented on the manuscript.

Peer review

Peer review information

Nature Communications thanks Hong-Bin Yao and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The data that support the findings of this study are available from the corresponding authors upon request. The crystallographic coordinates for structures reported in this paper have been deposited at the Cambridge Crystallographic Data Centre (CCDC) under deposited number CCDC: 2512568 ((TPPEt)₂SbBr₅) and 2513340 ((TPPEtCz)₂Sb₂Br₈), which can be accessible via this link https://www.ccdc.cam.ac.uk/structures/. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-68597-9.