Solution-processed light-emitting diodes (LEDs) are emerging as a promising technology for advanced solid-state lighting (SSL) and display industries, owing to their energy-efficient, low-cost, and straightforward manufacturing processes1-3. In this realm, lead-halide based perovskites4, organic semiconductors5, and colloidal core–shell quantum dots (QDs)6 have gained recognition as promising candidates. They have been utilized as emitting layers via solution processing and achieved impressive performance in green and lower energy LEDs. However, developing efficient blue, and particularly deep-blue, LEDs that align with the Rec. 2020 primary blue for full-color displays poses a significant challenge for these materials7. Conventional blue organic LEDs (OLEDs) and perovskite LEDs (PeLEDs) demonstrate compromised structural and spectral stability under elevated operational biases8,9. Additionally, lead-based perovskites10 and Cd-based quantum dots11 are subject to concerns over environmental issues due to their toxicity, while the deployment of low-toxic phosphorescence organic emitters12, InP- and ZnSe-based QDs13 is constrained by their high cost in complex design and synthetic process. These challenges limit the practical application of these materials and raise difficulties to cover the full color gamut. Consequently, developing cost-effective, eco-friendly, and efficient emitters for solution-processed deep-blue LEDs is vital for the advancement of comprehensive SSL and display technologies.
Copper halide-based emitters have recently attracted significant attention due to their earth-abundant nature, low toxicity, air/moisture stability, optical tunability, and high luminescent efficiency14-17. Efforts to employ these materials as the emissive layer (EML) in LEDs, encompassing ligand design for improved solution processability18-22 and optimization of host-dopant structures23, have resulted in notable advancements in device efficiency and stability. Specifically, in the domain of deep-blue LEDs, 0D-Cs 3 Cu 2 I 5 has attracted substantial research interest due to its optimal emission wavelength (445 nm), high photoluminescence quantum yield (PLQY) of 87-95%, and high heat/moisture resistance24. Nevertheless, the maximum external quantum efficiency (EQE max ) of 0D-Cs 3 Cu 2 I 5 -based blue LEDs remains very low, 1.02%25, a constraint primarily attributed to inefficient carrier injection stemming from its molecular (cluster) structure as a common drawback among cluster-based copper-iodide hybrids. Beyond the host-guest approach, overcoming this limitation could be achieved by developing/designing new emitters with higher dimensionality on the material side and advanced interfacial engineering at the device level.
Herein, we report a new one-dimensional (1D) hybrid copper iodide, synthesized using a bifunctional ionic ligand (Hdabco)I (Hdabco = 1,4-diazabicyclo-[2.2.2]octane-1-ium). The compound is composed of 1D-Cu 4 I 8 4- anionic chains coordinated to cationic ligand Hdabco+ with a chemical formular 1D-Cu 4 I 8 (Hdabco) 4 [also referred to as CuI(Hda)]. It exhibits strong photoluminescence (PL) in the deep-blue region (449 nm) with nearly unity (~100%) PLQY. We fabricated high-quality polycrystalline thin films of CuI(Hda) via a solution-process, which exhibits a highly preferred orientation and a PLQY identical to that of single-crystal samples. Temperature-dependent time-resolved emission study, alongside femtosecond/nanosecond transient absorption (TA) spectroscopic analysis, suggests a complex emission mechanism involving fluorescence, thermally activated delayed fluorescence (TADF), and phosphorescence, consistent with the results from density functional theory (DFT) calculations. We further investigated the electrical properties and electronic structure of CuI(Hda), including its anisotropic charge carrier dynamics and band structure using time-resolved microwave conductivity (TRMC) techniques and photoelectron spectroscopy, respectively.
We then fabricated high-quality deep-blue LEDs using the CuI(Hda) thin film as the sole active EML based on a dual interfacial hydrogen-bond passivation (DIHP) strategy. Our approach adopted a functionalized carbazole-phosphonic acid-based self-assembled monolayer (SAM)26,27 and an ultrathin PMMA capping layer28 as H-bond donor on the HTL/EML and EML/ETL (HTL: hole-transport layer; ETL: electron-transport layer) interfaces, respectively. The coverage/thickness and interfacial interactions were analyzed by angular-resolved X-ray photoelectron spectroscopy (ARXPS). We reasoned that the observed enhancement in device performance is attributable to defect passivation and optimized charge injection, facilitated by the H-bond coupled heterojunctions. The DIHP-CuI(Hda) deep-blue LED devices outperform the pristine and other control devices with non-H-bonded SAMs, achieving a record-high EQE max of 12.57% among metal halide-based blue LEDs and a high maximum luminance (L max ) of 3970.30 cd/m², with color coordinates (0.147, 0.091), closely approaching the Rec. 2020 standard. Furthermore, these devices exhibit exceptional operational stability, with a half-lifetime (T 50 ) of 204 hours under ambient conditions. We subsequently demonstrated a large-area DIHP-CuI(Hda) device (2 cm × 2 cm) that maintains a high EQE max of 7.87%, illustrating the reliability of the solution processability of hybrid copper iodide and our surface modification technique.
Design, thin film fabrication, and characterization of CuI(Hda)
We designed a mono-protonated aliphatic ligand, (Hdabco)I, from dabco, with an active monodentate coordination site and a high LUMO level, targeting the solution-processibility16 and an optimum band gap of the resultant hybrid material for deep-blue emission. High-quality single crystals of CuI(Hda) (Fig. 1a) were grown from a precursor solution using a facile recrystallization method. The precise crystal structure of 1D-Cu 4 I 8 (Hdabco) 4 was determined using the single crystal X-ray diffraction (SCXRD) technique (Fig. 1a, Supplementary Figs. S1 and S2 and Table S1). The refined crystal structure reveals that the compound comprises of anionic 1D-(Cu 4 I 8 )4- inorganic motif and organic cationic ligand (Hdabco)+, connected through both coordinate and ionic bonds. Each Cu(I) atom is coordinated to three iodine atoms and one nitrogen atom from (Hdabco)+, forming a distorted tetrahedron of (CuI 3 N). Each tetrahedron shares edges and corners with its adjacent tetrahedra, resulting in an infinite 1D chain. The organic cations form a H-bonded dimer, (Hdabco) 2 2+, as a result of the unique structure of (Hdabco)+, where one nitrogen atom is protonated and the other serves as a free binding site, enabling the formation of both the Cu-N coordination bond and intermolecular H-bond (Fig. 1a).
We performed first-principles DFT calculations using the Vienna ab initio Simulation Package (VASP) to understand the electronic structure of CuI(Hda). The result shows that it is a direct bandgap semiconductor (Fig. 1b). The band structure shows that the valence band (VB) includes a series of flat bands with relatively small dispersion, while the conduction band (CB) edge is noticeably dispersive, indicating a smaller electron effective mass and more favorable electron transport properties. To correctly estimated the DFT band gap, we used the screened hybrid functional of Heyd, Scuseria and Ernzerhof (HSE). The HSE band gap of CuI(Hda) was calculated to be 3.8 eV. The projected density of states (PDOS) analysis (Fig 1b) indicates that the atomic contributions to its valence band maximum (VBM) are primarily from the inorganic components, specifically Cu 3d and I 5p atomic orbitals. On the other hand, the conduction band minimum (CBM) is populated by atomic orbitals from both the inorganic motif and the organic ligand, specifically Cu 4s, I 5s, as well as C and N 2p atomic orbitals. This differs significantly from previously reported hybrid copper-halides4,15,19,29,30 or all-inorganic31,32 copper halides. The electron transfer process in CuI(Hda) is a combination of (metal/halide)-to-ligand charge transfer [(M/X)LCT] and all-inorganic-based charge transfer typically found in hybrid perovskites (VBM: Pb 6s and I 5p; CBM: Pb 6p and I 5s). This results in a distinct and intricate emission mechanism for CuI(Hda).
We fabricated CuI(Hda)thin films via single-step spin-coating of the precursor solution on ITO and larger area quartz substrates (Fig. 1c), followed by antisolvent dripping to initiate the crystallization and annealing at room temperature for a relatively long time. Pin-hole free thin-film samples were obtained with a thickness of 90 nm and R a of 0.177 nm, as measured by atomic force microscopy (AFM, Fig. 1c and Supplementary Fig. S3). Top-view scanning electron microscopy (SEM) and energy dispersive X-ray spectroscopy (EDS) images (Supplementary Figs. S4-S8) demonstrate the good compactness, surface smoothness, and even elemental distribution of the as-fabricated CuI(Hda)thin films. We assessed the crystallinity and phase purity of the as-made thin films by comparing the powder X-ray diffraction (PXRD) patterns of a CuI(Hda) thin film sample and a sample of CuI(Hda) powders scratched from multiple thin films with the simulated pattern from single crystal data (Fig. 1d). The thin film sample exhibits only one prominent diffraction peak at 2θ = 12.9°, which corresponds to the (200) plane of the structure, while the pattern of the scratched powders matches well with the simulated one, confirming the formation of phase pure CuI(Hda). The GIWAXS patterns reveal the polycrystalline nature of the CuI(Hda) thin film and confirm the preferred orientation of (200) plane (perpendicular to the 1D-(Cu 4 I 8 )4- chain) in out-of-plane scattering profile (Fig. 1e and Supplementary Figs. S9 and S10). The combination of low surface roughness, high crystallinity, and an identical PLQY (99.4%, Supplementary Fig. S11) to that of single-crystal samples affirms the high-quality of the thin film samples, which is attributed to the excellent solution processability of this type of materials.
Photophysical properties of CuI(Hda)
Steady-state optical absorption and diffuse reflectance spectra of polycrystalline thin film (90 nm) and powder samples of CuI(Hda) were collected and analyzed using UV-vis spectroscopy at room temperature. The nearly identical absorption edges of the two types of samples suggest the direct band gap nature of this compound (Fig. 2a and inset). The estimated optical band gap is ~3.7 eV, aligned well with the calculated HSE band gap. Room temperature PL of the thin film sample reveals a strong single-band emission peak centered at 449 nm with a full width at half maximum (FWHM) of 97 nm and color coordinates (0.147, 0.091) (Fig. 2a). CuI(Hda) features a large Stokes shift and relatively broad emission band, similar to those observed in some pure inorganic copper halides and most hybrid copper halides. Both the single crystals and thin film samples of CuI(Hda)exhibit near-unity PLQY values of 99.4 ± 0.4% when excited at 285 nm, marking the highest value for blue emitters33.
Insight into the emission mechanism of CuI(Hda)comes from the temperature dependent TRPL decay profiles carried out on a parallel thin film sample (Fig. 2b). The intensity-weighted average lifetime values show a small inverse temperature-dependence, decreasing from 5.87 μs at 78 K to 3.26 μs at 298 K. The tri-exponential fitting results of TRPL decay curves are summarized (Table S2), suggesting an additional nanosecond-scale decay (τ 1 ) path compared to the previously reported copper iodide hybrids. Specifically, the nanosecond timescale and temperature-independent contribution ofτ 1 to the decay profile suggest a fluorescence decay path. The sub-microsecond lifetime τ 2 , showing a positive temperature-dependent weight to the decay, can be attributed to the temperature-activated delayed fluorescence (TADF). The longest microsecond lifetime τ 3 is recognized as phosphorescence. The density of states (DOS) plots obtained from DFT calculations show that the 4s orbitals of Cu and 5s orbitals of I are also contributing to the CBM of CuI(Hda) (Fig. 1b) , enabling an additional inorganic-based decay path similar to other perovskite metal halides. Total radiative rate (k r ) and non-radiative decay rate (k nr ) at 298K were calculated to be 3.05 × 105 s-1 and 1.84 × 103 s-1, respectively, quantitatively manifesting the efficient radiative decay of the deep blue emitter.
Temperature dependent (from 78 to 298 K, Fig. 2c) and power-dependent (Supplementary Figs. S12 and S13) steady-state PL spectra were measured on a thin film sample of CuI(Hda)on quartz substrate. No obvious shift of the emission peak and FWHM was observed at multiple temperatures and pump powers, which implies a low electron−phonon coupling due to the higher structural rigidity of the 1D-(Cu 4 I 8 )4- chain compared to other molecular (0D) copper halides, beneficial to the more efficient radiative decay process. An exciton binding energy (E b ) of 107 meV was fitted from the Arrhenius plots of the PL intensity with temperature (Supplementary Fig. S14). The large E b value is considered to originate from the 1D nature of the compound as well as the strong interaction between the cationic ligands and the anionic inorganic motif, thereby the exciton’s localization was strengthened and the excitonic nature of CuI(Hda) was revealed.
We employed femtosecond (fs-TA) and nanosecond (ns-TA) transient absorption spectroscopy to investigate the excited-state dynamics of CuI(Hda) films at room temperature. The fs-TA spectra (Fig. 2d and 2e and Supplementary Fig. S15) disclose a ground state bleaching band at 381 nm and an excited-state absorption band that shifts from 601 nm to 624 nm as the delay time increases following optical pumping. These bands were analyzed using quadric- and tri-exponential decay models, respectively, consistent with a picosecond vibration relaxation process and the tri-radiative decay observed in TRPL studies (Supplementary Figs. S16 and S17 and Tables S3 and S4). Complementary ns-TA experiments extended our analysis beyond the 4.5 ns delay limit of fs-TA, confirming the bleaching band and yielding decay times that quantitatively match TRPL findings (Supplementary Figs. S18 and S19 and Table S5). Integrating time-resolved data from both TRPL and TA measurements, we propose a kinetic model for the excited-state decay process of CuI(Hda) (Fig. 2f).
Electrical properties and electronic structure of CuI(Hda)
We explored the anisotropic charge carrier dynamics, mobility, equilibrium doping density and dielectric constant in CuI(Hda) single crystals and thin films by TRMC and dark microwave conductivity (DMC) techniques. A bi-exponential fit of the TRMC transient for a polycrystalline thin film of CuI(Hda) on quartz (Fig. 3a) reveals two lifetime components: a primary one of 4.6 ns, and a secondary one of 250 ns, with a maximum yield-mobility product of 1.5×10-4 cm2V-1s-1.
We performed additional TRMC measurements on a CuI(Hda) single crystal (1×1×4 mm) to investigate potential birefringence and/or mobility anisotropy attributable to its markedly anisotropic crystal structure (Fig. 3b). Analysis of the power reflectance curves for two different orientations of a CuI(Hda) single crystal relative to the microwave electric field (Fig. 3c) indicates nearly isotropic conductivity at 5.5×10-5 S/cm and a dielectric constant of 4.75 ± 0.5. Although the crystal does not exhibit any pronounced birefringence, the carrier mobility appears slightly anisotropic. TRMC transients for the CuI(Hda) single crystal, excited by both parallel and perpendicular pump laser polarizations relative to each crystal orientation with the microwave electric field, were examined (Fig. 3d). Charge carrier mobility (at 9 GHz) measured 4.4×10-4 cm2V-1s-1 perpendicular to the a-axis, approximately twice that measured parallel to it, substantiating the anisotropic charge carrier properties.
Polarized PL spectroscopy experiment was conducted to corroborate the anisotropic charge carrier properties of CuI(Hda) single crystals (Extended Data Fig. 1). When using a non-polarized pump on (100) and (01) planes, there is a negligible difference in the PL intensity when detected in either linear polarization direction (Extended Data Fig. 1a-d). However, for the (011) face, the PL intensity in the polarization direction parallel (0°) to the a-axis of the unit cell is considerably higher (Extended Data Fig. 1e-f). This suggests a charge separation in the perpendicular direction (90°). The phenomenon can be explained by the isosurface plots of the VBM and CBM, which display greater dispersion in the perpendicular direction. The same trend was observed in the PL spectra when both the pump and detection are linearly polarized. Specifically, a perpendicular (90°) polarized pump on (100) and (011) is more effective than any of the parallel (0°) scenarios. This increased efficiency might be attributed to an anticipated higher absorption coefficient for band-edge absorption at the Γ point. Consequently, the combined effects of anisotropic absorption and photo-generated carrier dispersion result in the distinct polarized PL intensity.
As a complement to the TRMC mobility, space charge-limited current (SCLC) measurements were also performed on hole-only and electron-only devices of CuI(Hda). We extracted a hole-mobility of µ h = 5.9 × 10−4 cm2 V−1 s−1, an electron-mobility of µ e = 8.8 × 10−4 cm2 V−1 s−1 alongside a trap state density (η trap ) of 8.2 × 1016 cm−3 (Fig. 3e), using the dielectric constant determined from the microwave cavity resonance in TRMC measurements. The noted higher electron mobility correlates well with the DFT-calculated band structure. The observed SCLC mobilities surpass those from TRMC, aligning with the expected lower free charge yield upon photoexcitation in TRMC assays. Notably, the charge carrier mobilities in CuI(Hda) thin films are comparable to those of high-quality perovskite films for analogous applications4 and are approximately an order of magnitude higher than those reported for metallic copper halides6.
We conducted X-ray photoelectron spectroscopy (XPS, Supplementary Figs. S20-S24, Table S6) and reflected electron energy loss spectroscopy (REELS, Supplementary Fig. S25) experiments on the thin film samples of CuI(Hda), confirming the correct composition and indicating an absorption onset 3.6 eV below the elastic peak, in good agreement with the optical band gap value estimated from UV-vis spectroscopy and HSE band gap from DFT calculations. To determine the distance between the valence band edge and the vacuum level (VL), the secondary electron cutoff (SECO) and valence band (VB) were measured by ultraviolet photoelectron spectroscopy (UPS) (Supplementary Fig. S26). The sharp VB edge, found 0.7 eV below the Fermi level, is compatible with the calculated electronic structure and is mostly due to Cu 3d and I 5p atomic states, suggesting the p-type semiconductor nature of the compound. The energy difference between VB and VL is estimated to be 5.8 eV. The measured band structure of CuI(Hda) is summarized in Supplementary Fig. S27.
Interfacial H-bonds on CuI(Hda) heterojunctions
We selected NiO x as HTL and Ca(acac) 2 (calcium acetylacetonate) as the ETL34 for CuI(Hda)deep-blue LEDs based on the energy diagram derived from photoemission spectroscopy for optimized electron-hole injection. We sought to address interfacial trap-assist carrier recombination and unbalanced charge injection using the DIHP approach on both heterojunctions of the CuI(Hda) EML.
On the HTL/EML interface, we designed a new tri-functional 2PACz-based SAM (Ac2PACz) for HTL functionalization, and compared its performance with three reported bi-functional SAMs26,35 from the same class (Fig. 4a, Supplementary Figs. 28-35). 2PACz-based SAMs are known for their dual functionalities of surface passivation through chemical adsorption of phosphonic acid and enhanced hole mobility via carbazole group. The new SAM, featuring the additional acetyl group (Ac) as a strong electron withdrawing group, showing both a high electrostatic potential (ESP) on the C=O group as effective H-bond donor and a suitable HOMO level for band alignment (Fig. 4a).
We calculated the favorable surface adsorption modes of Ac2PACz on the NiO (111) surface (Supplementary Figs. 36 and 37) , then assessed the surface interaction between the conformation-optimized SAM and CuI(Hda) (200) plane (Fig. 4b). A H-bond energy of 0.76 eV (73 kJ/mol) was obtained, demonstrating strong interfacial H-bond coupling (> 40 kJ/mol) between adsorbed Ac2PACz and CuI(Hda). The H-bonded heterojunction facilitates charge density relocation to the (Hdabco) 2 2+ ligand on HTL-EML surface, as validated by charge density difference calculation.
We conducted Kelvin probe force microscopy (KPFM) study which reveals that the Ac2PACz-NiO x shows a surface potential of -0.34 V, suggesting an increase of work function of 0.3 eV compared to pristine NiO x , while a uniform morphology and homogeneous surface potential distribution was retained (Fig. 4c). Surface coverage of SAMs on the NiO x surface was quantitatively assessed by ARXPS (Fig. 4c, Supplementary Figs. 38-45), illustrating a near-optimal monolayer adsorption. We confirmed optimized band alignments (Supplementary Fig. 46, Table S7) between the functionalized HTL/EML heterojunctions and established a low surface-energy (Supplementary Fig. 47) and hole-selective interface (Supplementary Figs. 48-51). Specifically, Ac2PACz-functionalized HTL shows a marked reduction in contact angle, the lowest trap density, and largest enhancement in hole mobility compared to other SAMs without H-bond coupling, demonstrating the effectiveness of H-bonded interface.
To balance the higher electron mobility in CuI(Hda) and reduce the surface defect level that introduced by direct solution-processing of Ca(acac) 2 on the EML, we utilized an ultrathin PMMA layer (< 3 nm) as an electron-blocking buffer. As a known insulating polymer, the PMMA effectively modulates electron injection into the EML by varying the layer thickness, while its high transparency minimized the impact to the light output. Additionally, PMMA's carbonyl groups contribute to surface passivation through hydrogen bonding interactions with surface (Hdabco) 2 2+ (Fig. 4d). ARXPS (Extended Data Fig. 2 and Supplementary Figs. 52-63) was employed to verify the PMMA overlayer thickness across samples prepared with varying spin rates, enabling precise control over interfacial properties. We validated the interfacial H-bonding on the PMMA-EML interface through the observed inverse dependency between the PMMA layer thickness and the binding energy shift in O1s (C=O) (Fig, 4d), indicative of the carbonyl group acting as a H-bond donor which leads to a reduction in electron density. Moreover, ratios of integrated C1s (C=O) /C1s (C-C, C-H) peak intensity at normal emission angle (0°) remain a constant of 1:3 in all PMMA thin films, precisely reflects the respective chemical environment of the polymer (Extended Data Fig. 2). On the contrary, a more rapid decreasing of C1s (C=O) /C1s (C-C, C-H) ratio was observed in thinner PMMA overlayers as the incident angle increased in ARXPS, elucidating a higher concentration of the C=O group on the CuI(Hda)-PMMA surface resulting from the interfacial H-bonding. This analysis, alongside the findings on Ac2PACz-CuI(Hda) interface, concluded a successful introduction of the DIHP approach for both heterojunctions of the HTL/EML/ETL structure, achieving synergistic surface passivation and charge injection enables by strong interfacial H-bonds.
Deep-blue CuI(Hda) LEDs based on DIHP approach
The pristine device was constructed with a layered structure of ITO (185 nm)/NiO x (40 nm)/CuI(Hda)(90 nm)/Ca(acac) 2 (35 nm)/LiF (1 nm)/Al (60 nm), LiF serves as an electron-injection layer and Al as the cathode (Fig. 5a). Cross-sectional Helium-ion microscope imaging (HeIM) of the complete device shows a pin-hole free CuI(Hda) EML with even thickness and uniform interfaces (Fig. 5b). These pristine devices achieved a EQE max of 3.09% and L max of 1714.21 cd m-2 (Fig. 5c). In contrast, the PMMA-capped LEDs (without SAM functionalization of HTL) showed a 2-fold EQE max , with a 0.8-fold lower current density but a 1.3-fold increase in luminance, suggesting the surface passivation and electron blocking effect of H-bonded PMMA-capping layer. Further verification of such functions was elucidated by a set of reference LED devices with various HTL/ETL selection (Extended Data Fig. 3). In each case, the PMMA capping layer reduced the current density at a given voltage by obstructing electron injection. For devices structured as ITO/NiO x /CuI(Hda)/PMMA/Corannulene/LiF/Al, the differences in the J-V and L-V curves between the pristine and PMMA-capped devices were minimal (Extended Data Fig. 3a-c), as chloroform was used to solution process the ETL, which is a good solvent for PMMA. In contrast, for the ITO/PEDOT:PSS/CuI(Hda)/PMMA/Ca(acac) 2 /LiF/Al configuration (Extended Data Fig. 3d-f), the electron-blocking buffer effect was predominantly observed, especially when paired with an effective ETL and a less effective HTL. As a result, the efficiency of the PMMA-capped device surpassed that of the pristine one. For devices employing TPbi (2,2',2''-(1,3,5-Benzinetriyl)-tris(1-phenyl-1-H-benzimidazole)) as the ETL (Extended Data Fig. 3g-l), the PMMA layer further exacerbated the poor electron injection, attributed to energy misalignment which led to reduced efficiency.
We observed a synergistic effect from the DIHP approachin LED devices with the structure: ITO (185 nm)/Ac2PACz-NiO x (40 nm)/CuI(Hda)(90 nm)/PMMA (2 nm)/Ca(acac) 2 (35 nm)/LiF (1 nm)/Al (60 nm). A peak luminance (L max ) of 3970.30 cd m-2 with CIE color coordinates of (0.147, 0.091) (Figs.5d and e) was achieved. Blue CuI(Hda) LED with dual H-bonded interfaces yielded a EQE max of 12.57% and an average of 10.0 ± 1.5% (Figs. 5f and g), marking a ~4-fold increase over the pristine device and setting a new record among metal halide-based blue LEDs with peak emissions below 470 nm. Compared to devices incorporating other non-H-bonded SAMs (Extended Data Fig. 4), the Ac2PACz-modified heterojunction demonstrates significantly enhanced LED performance across all metrics. A large-area device (2 cm × 2 cm) with DIHP treatments maintained a high EQE max of 7.87% (inset of Fig. 5h). Both results demonstrating the superiority and reliability of the solution-processed hybrid copper iodide and the DIHP approach.
We further assessed the continuous operational stability of the deep-blue LEDs with an initial luminance (L 0 ) of ~100 cd/m-2 under ambient conditions. The durability of both unprotected LEDs (pristine and Ac2PACz-PMMA dual-modified) and their encapsulated counterparts is depicted in Fig. 5H. The stability of dual-modified LEDs surpassed the pristine devices substantially, as a result of suppressed surface defects and balanced charge injection. Remarkable T 50 values of 113 h for unprotected and 204 h for encapsulated LEDs were achieved under ambient conditions, exceeding all the reported blue PeLEDs and are at the same level as the best performing all-inorganic copper halides, yet with an EQE 10 times higher. A comprehensive comparison of both device performance and operational stability among metal halide-based blue LEDs is provided in Supplementary Table S8. CuI(Hda) outperforms all previously reported deep-blue emitting compounds in both EQE and half-lifetime. Furthermore, the thermal stability and long-term durability of CuI(Hda) were evaluated under harsh conditions. The results confirmed its outstanding structural robustness (Supplementary Figs. 64-66).