Spectra stable deep-blue light-emitting diodes based on cryolite-like cerium(III) halides with nanosecond d-f emission

Abstract

Next-generation wide color gamut displays require the development of efficient and toxic-free light-emitting materials meeting the crucial Rec. 2020 standard. With the rapid progress of green and red perovskite light-emitting diodes (PeLEDs), blue PeLEDs remain a central challenge because of the undesirable color coordinates and poor spectra stability. Here, we report Cs₃CeBr_xI_6−x (x = 0 to 6) with the cryolite-like structure and stable and tunable color coordinates from (0.17, 0.02) to (0.15, 0.04). Further encouraged by the short exciton lifetime (26.1 ns) and high photoluminescence quantum yield (~76%), we construct Cs₃CeBr_xI_6−x-based rare-earth LEDs via thermal evaporation. A seed layer strategy is conducted to improve the device’s performance. The optimal Cs₃CeI₆ device achieves a maximum external quantum efficiency of 3.5% and a luminance of 470 cd m⁻² with stable deep-blue color coordinates of (0.15, 0.04). Our work opens another avenue to achieving efficient and spectrally stable deep-blue LEDs.

Cs₃CeX₆ exhibits high PLQY, fast transition rate, and stable deep-blue emission, holding promises for high-performance blue LEDs.

INTRODUCTION

Wide color gamut displays motivate the development of next-generation light emitters with high color purity and precisely defined color coordinates to meet the Rec. 2020 standard (1, 2). Recently, metal halide perovskites have emerged as promising candidates for the light source of next-generation displays because of their low-cost, high color purity, and defect-tolerant properties (3). The state-of-the-art red and green perovskite light-emitting diodes (PeLEDs) have achieved high external quantum efficiencies (EQEs) of 23.0 and 25.6% with precise color coordinates that meet the crucial Rec. 2020 standard (4–6). However, as one indispensable component for full-color display, efficient blue PeLEDs with suitable color coordinates remain a substantial challenge (7). In general, two strategies have been developed to implement blue PeLEDs. The first one relies on composition engineering, the emission peaks could be gradually shifted from green to blue (e.g., from 520 to 420 nm) by partially replacing Br⁻ with Cl⁻, forming the mixed halogen perovskites (7). However, phase segregation under the electric field or light excitation leads to poor spectral stability. Consequently, mixed halogen-based PeLEDs demonstrate severe electroluminescence (EL) emission peak shift from blue to green under continuous device operation (8–10). Another feasible method to achieve blue emission is to use dimensional engineering. For instance, Dong et al. (11) have demonstrated that CsPbBr₃ treated by phenylbutylamine and ethylamine could achieve a deep-blue emission of 455 nm with a photoluminescence quantum yield (PLQY) of up to 57% and an EQE of 4.62%. However, the core component of the heavy metal lead causes severe toxicity and environmental issues, potentially preventing PeLEDs from further commercialization (1). Therefore, it is imperative to develop novel blue light-emitting materials, with high efficiency, standard color coordinates, excellent spectra stability, and environment-friendly properties.

The lanthanide ions have been established as indispensable components in lighting because of their intrinsically high PLQY, excellent color purity, superior stability, and abundant emission lines covering the entire visible light range (12). However, most lanthanide ions suffer from a long excited-state lifetime due to parity-forbidden f-f transitions (table S1), which may cause severe carrier quenching and lead to inferior device performance for LEDs (13). Parity-allowed d-f transition offers one possible solution. Among these lanthanide ions, Ce³⁺ having one single electron (4f¹ configuration) shows the shortest exciton lifetime compared to that of other lanthanide ions (table S1), which holds the potential for realizing high-efficient EL (14). Furthermore, because the 5d orbital of Ce³⁺ is easily affected by the crystal field environment, cerium halides demonstrated tunable emission wavelength in the blue-violet region (15). Besides, cerium is nontoxic and inexpensive with a high abundance of 0.006 weight % (wt %) in Earth’s crust, which is even higher than that of copper (0.005 wt %) (16).

Despite holding great potential, explorations of cerium(III) halides for rare-earth LEDs (RELEDs) are very rare, and their advantages have not been fully demonstrated. The traditional EL based on cerium-based dopant compounds was once obtained through alternating current–driven thin-film electroluminescence (ACTFEL) technology in the last century. They usually use a back-to-back Zener diode structure (17). The tunneling charge injection across the insulators and luminescence from the ionization of activators was assisted by a large electric field (1 to 2 MeV cm⁻¹), which resulted in poor device performance with high turn-on voltage (~100 V), low brightness (~10 cd m⁻²), and low power efficiency [~0.02 lumen (lm) W⁻¹] (17). Because of the poor performance, this technology has been discarded. To improve device performance, we here propose to follow the principles established from PeLED and construct RELED by directly injecting charges into rare earth compounds with d-f transition. This has been demonstrated to be workable in CsEuBr₃ despite the relatively lower luminance (18).

In this work, we have developed a series of cerium halides Cs₃CeBr_xI_6−x (x = 0 to 6) with tunable emission spectra ranging from blue to violet region. Among these cerium halides, Cs₃CeI₆ exhibits a Commission International de l’Éclairage (CIE) color coordinate of (0.15, 0.04), which matches well with the crucial Rec. 2020 standard for blue emitters (1). Further density functional theory (DFT) calculation reveals the parity-allowed Ce-5d to Ce-4f transition of Cs₃CeI₆, and the resulting Cs₃CeI₆ achieves a short excited-state lifetime of 26.1 ns and a PLQY of 76.2%. Encouraged by its excellent luminescent properties, we further explored its application for RELEDs. The Cs₃CeI₆ luminescent films are fabricated by dual-source coevaporation, and a seed layer strategy is used to improve the film compactness and crystallinity of the Cs₃CeI₆ films. As a result, the deep-blue LED based on Cs₃CeI₆ with a CIE value of (0.15, 0.04) shows the optimal performance, with a maximum brightness of 470 cd m⁻² and an EQE of 3.5%. Moreover, Cs₃CeBr_xI_6−x-based LEDs with CIE values from (0.17, 0.02) to (0.15, 0.04) were successfully fabricated by tuning the anion composition. These LEDs show spectra stability under a continuous electric field (60 min at 5 V), which implies great phase stability and limited ion migrations. Benefitting from the high manufacturability of the thermal evaporation (TE) method, we further realized a pixelized thin film with a pixel size of 100 μm as well as patterned and large-area (100 mm²) RELEDs.

RESULTS

The Cs₃CeI₆ powder was synthesized by a simple solid-state reaction (see details in Materials and Methods) (19). The cesium iodide–cerium(III) iodide (CsI-CeI₃) system has only one congruent melting point near ~950 K at a molar ratio of 3:1, corresponding to the Cs₃CeI₆ phase (20). The calculated enthalpy of mixing CsI-CeI₃ liquid also shows the minimum value around a 3:1 molar ratio (fig. S1); thus, the pure Cs₃CeI₆ phase can be easily synthesized. Figure S2 depicts the detailed crystal structure based on Rietveld’s refinement of powder x-ray diffraction patterns. Cs₃CeI₆ belongs to the tetragonal space group of P42/m (unit cell: a = b = 10.56319 Å, c = 8.72570 Å, and α = β = γ = 90°). It resembles the cryolite-like structure of Na₃AlF₆, which is related to the double-perovskite structure with reduced symmetry (21). As shown in Fig. 1A, each Ce³⁺ is bound to six I⁻ ions, forming the [CeI₆]³⁻ octahedron, which resembles Pb²⁺ in [PbI₆]⁴⁻ octahedron. These [CeI₆]³⁻ octahedra are completely isolated from each other by Cs⁺, giving rise to the zero-dimensional (0D) structure. The 0D nature of Cs₃CeI₆ enables localized band edges and fast radiative recombination. All Ce³⁺ ions share the same chemical environment, which contributes to narrow emission linewidths suitable for wide color gamut displays.

Fig. 1. Structural and emission characterization of Cs₃CeBr_xI_6−x.

(A) Crystal structure of Cs₃CeI₆. (B) Calculated electronic band structure of Cs₃CeI₆. (C) Schematic diagram of Cs₃CeI₆ Ce-5d → Ce-4f parity-allowed emission. (D) The pseudo-color maps of integrated photoluminescence (PL) intensity curves with different temperatures from 100 to 220 K. (E) PL spectra of Cs₃CeBr₆, Cs₃CeBr₂I₄, and Cs₃CeI₆, with the inset showing the corresponding photographs under ultraviolet (UV) light radiation. (F) Spectra stability measurements of Cs₃CeBr₆ (left), Cs₃CeBr₂I₄ (middle), and Cs₃CeI₆ (right), under continuous UV light radiation. DOS, density of states; a.u., arbitrary units.

We further carried out the first-principles DFT calculation to understand its electronic structure. A Hubbard (+U) correction with a simplified version of Cococcioni and de Gironcoli was used to describe the strongly correlated f-electrons. As shown in Fig. 1 (B and C), Cs₃CeI₆ exhibits a direct bandgap at the G point with an optical bandgap of 2.95 eV. The valence band maximum (VBM) of Cs₃CeI₆ is mainly contributed by Ce-4f orbitals, and the conduction band minimum (CBM) consists primarily of Ce-5d orbitals and a small portion of I-5p and Ce-4f. Therefore, electrons and holes could be directly injected into the orbitals of Ce-5d and Ce-4f to achieve radiative recombination, permitting a relatively higher carrier injection efficiency than the traditional ACTFEL (14, 18). Moreover, the contribution of halogen to the CBM provides the possibility of luminescence regulation through halogen adjustment. Cs⁺ ions do not contribute to VBM or CBM, which could confine the charge carriers inside the [CeI₆]³⁻, as confirmed by the electron and hole charge densities (ρ = 〈ψψ〉) distribution in fig. S3. A high radiative recombination rate is thus guaranteed according to the Fermi golden rule (22). Because of the localized characteristic of Ce-4f orbitals and the 0D structure of Cs₃CeI₆, the band dispersion is much smaller (especially for VBM) than those of 3D materials such as CsPbBr₃, leading to the larger excitons binding energy (23). However, the localized Ce-4f orbitals also lead to the difficulty of hole injection.

Benefiting from spin-allowed and parity-allowed Ce-5d to Ce-4f transition, the Cs₃CeI₆ crystal emits bright deep-blue light under ultraviolet (UV) light excitation with a nanosecond exciton lifetime (~26.1 ns), as shown in fig. S4. The PL spectra of Cs₃CeI₆ are shown in Fig. 1 (D and E), with the main emission peak at 430 nm and a small shoulder at 470 nm. The two characteristic emission peaks result from the transition from the lowest excited states of Ce-5d to the splitting Ce-4f ground state (²F_5/2 and ²F_7/2) (24). The energy separation of the doublet emission peak is calculated to be 1978 cm⁻¹, matching well with the values of 2000 cm⁻¹ for the Ce³⁺ emission reported in the previous studies (16, 24). As shown in Fig. 1D, temperature-dependent PL measurements reveal a high exciton binding energy of ~224.7 meV, which is much higher than that of bulk CsPbBr₃ (~20 meV) (25). The high exciton binding energy together with its short exciton lifetime characteristics enable the efficient light emission from Cs₃CeI₆. The PLQY of the Cs₃CeI₆ powder was measured to be ~76.2% at room temperature (fig. S5). In comparison, bulk CsPbBr₃ shows poor PLQY because of the dispersed energy band edges contributed by Pb-6p and Pb-6s orbitals (23). Strategies such as quantum well confinement have to be used to increase the exciton binding energy and PL efficiency (3).

The calculated CIE color coordinate of Cs₃CeI₆ is (0.15, 0.04), approaching the Rec. 2020 blue CIE standard (0.131, 0.046), as shown in fig. S6. Moreover, the spectra of the Cs₃CeBr_xI_6−x family could also be tuned. As shown in Fig. 1E, their spectra shifted gradually from the deep-blue region to the violet region by improving the content of Br. Cs₃CeBr₆, Cs₃CeBr₂I₄, and Cs₃CeI₆ exhibit very good spectra stability under continuous UV light radiation (30 mW cm⁻²) for 60 min, as shown in Fig. 1F. Such spectra stability is superior compared to that in other mixed halogen Pb-based perovskites (26), which may be attributed to the higher covalency character of the Ce─X (X = Br and I) bond (15, 27). The stable and tunable spectra suggest the great potential for practical RELED applications of the Cs₃CeBr_xI_6−x series.

Cs₃CeI₆ exhibits efficient and stable deep-blue emission with CIE color coordinate meeting the crucial Rec. 2020 standard, which inspires us to further explore its RELED applications. Here, Cs₃CeI₆ films were prepared by a TE process, because it allows intrinsically clean, high-vacuum environments free of pollution (28). TE has intrinsic advantages in manufacturing scalable, large-area, and pixelized LEDs for practical display applications, which has been demonstrated in commercial organic LEDs (OLEDs) (25).

As shown in Fig. 2A, Cs₃CeI₆ films were fabricated by dual-source coevaporation, with the calibrated evaporation rate ratio of CsI and CeI₃ set to be 3:1. As described above, Cs₃CeI₆ is the only stable phase with a congruent melting point in the CsI-CeI₃ binary system and the lowest calculated mixing enthalpy (fig. S1). Therefore, pure-phase Cs₃CeI₆ films can be obtained with good homogeneity by accurately controlling the evaporation ratio of raw materials, as confirmed by the energy-dispersive spectrometry (EDS) mapping results (fig. S7). As shown in the inset of Fig. 2B, the film prepared by TE emits bright deep-blue light under a 365-nm wavelength UV light excitation, which agrees well with that of the Cs₃CeI₆ powder. Benefiting from the oxygen-free environment of the TE process, the +3 valence state of cerium can be well preserved (fig. S8).

Fig. 2. Diagrams of Cs₃CeI₆ film fabrication, crystallization processes, and their morphologies.

(A) The schematic diagram of the Cs₃CeI₆ film by dual-source coevaporation. (B) The PL spectra of the Cs₃CeI₆ film with different fabrication techniques. The inset shows a photograph of the Cs₃CeI₆ film under UV light excitation. (C) The top-view scanning electron microscopy (SEM) image of the Cs₃CeI₆ film fabricated at a high temperature. (D) Total Gibbs free energy (G_tot) as the function of particle radius (r*), where G_tot is composed of a volume term (G_v) and a surface term (G_s). (E) G_tot evolution for different crystallization processes. (F) The top-view SEM image of the Cs₃CeI₆ film fabricated with a seed layer. Distribution histograms of grain sizes are shown in the inset of the SEM images.

High-temperature in situ annealing would promote the complete reaction of CsI and CeI₃, which is crucial for the high-quality Cs₃CeI₆ film with high crystallinity and PLQY (Fig. 2, B and C, and fig. S9). However, the reduced nucleation sites at higher temperature results in poor morphology with obvious pinholes (Fig. 2C and fig. S9), which would cause a massive leakage current for the practical LEDs (29). To overcome this problem, a CsI seed layer (fig. S9) strategy was used to provide sufficient nucleation sites at the initial stage of film growth (30, 31).

As shown in Fig. 2F, the achieved Cs₃CeI₆ film shows not only better morphology but also further improved crystallinity (the insets of Fig. 2, C and F, show the statistical distribution of grain size). This crystallization process could be described by the Gibbs free energy (Fig. 2D), which is composed of a volume term (G_v) and a surface term (G_s) (30, 31). When the nucleus is larger than the critical size (r*), it will grow spontaneously; otherwise, it will disintegrate. Without the assistance of the seed layer, the crystallization of Cs₃CeI₆ has to overcome the critical free energy G* (region I in Fig. 2E). After adopting a seed layer, the crystal growth would start immediately and preferentially with a faster rate (region II in Fig. 2E) (32). As a consequence, a high-quality film with good crystallization, dense morphology, and high PLQY can be achieved simultaneously (Fig. 2F). This strategy may provide a general strategy for the preparation of dense and highly crystalline films.

To further explore the photophysical properties of Cs₃CeI₆ films, steady-state and transient spectral measurements were performed. We note that the films are very sensitive to moisture; thus, samples were well encapsulated by quartz before measurements. As shown in Fig. 3A, the Cs₃CeI₆ film exhibits the same PL spectra as the powder, with the main peak at 430 nm and the shoulder peak at 470 nm, as confirmed by the fitting results (fig. S10), indicating the two-exciton emission mechanism. The PL excitation (PLE) spectrum shows two obvious peaks at 283 and 393 nm, which may be attributed to the splitting of Ce-5d levels (33). On the other hand, both the PL and PLE spectra maintain the same shapes and features under different excitation and emission wavelengths, and no obvious energy transfer process could be observed in the time-resolved PL spectra decay (fig. S11) and transient absorption (TA) spectroscopy spectra (fig. S12). Therefore, they could be appropriately regarded as two coupled excitons sharing the same excited states (see details in fig. S13). The time-resolved PL decay curves of the Cs₃CeI₆ film are shown in Fig. 3B. Curves monitored at 430 and 470 nm show a similar trend, which could be well fitted by a single exponential function (14). The exciton lifetime was fitted to be ~26.3 ns, representing one of the fastest radiative d-f transitions among the lanthanide-based luminescent materials (e.g., 150 ns for CsEuBr₃), and also much faster than most of the lanthanide-based f-f transitions, as summarized in table S1 (18).

Fig. 3. Optical characterizations of Cs₃CeI₆ films.

(A) The PL excitation (PLE) spectra measured at different emission wavelengths from 410 to 480 nm and the PL spectra of Cs₃CeI₆ excited by different light from 270 to 400 nm. (B) Time-resolved PL decay curves of Cs₃CeI₆ films at room temperature. The decay curves can be well fitted by a single exponential function. (C) The PL quantum yield (PLQY) spectra of the Cs₃CeI₆ film at the excitation intensity of 0.15 mW cm⁻². (D) Dependence of PL effective lifetime (τ_e) and I₀ (the initial PL intensity of decay curves) on carrier density.

Benefiting from both the high-quality film and the fast radiative transition rate, the Cs₃CeI₆ film exhibits a high PLQY of ~71.4% with a negligible influence of excitation power (Fig. 3C and fig. S14). To fully understand the emission mechanism of the Cs₃CeI₆ film, we further explored its carrier dynamics by power-dependent transient PL decay measurements, as shown in fig. S15. We plotted τ_e (effective decay lifetime, defined as the time it takes when the PL intensity decay to the 1/e of its initial value) and I₀ (the initial PL intensity of decay curves) as a function of carrier density in Fig. 3D. We note that I₀ is the intensity of emission immediately after excitation (time zero), which is only determined by the radiative recombination process (34). The obvious linear relationship between I₀ and carrier density confirms the monomolecular recombination characteristic of Cs₃CeI₆ excitons, which is due to the strong confinement of Ce orbitals and the resulting large exciton binding energy (~224.7 meV, as confirmed above) (35). On the basis of the exciton emission model, k₁ (exciton recombination constant), k₂ (trap-mediated recombination constant), and k₃ (Auger recombination constant) are estimated to be 2.68 × 10⁷, 1.07 × 10⁷, and 4.61 × 10⁻⁹ s⁻¹ cm³, respectively, through the excitation power-dependent PL decay measurements (see details in the Supplementary Materials). We note that k₃ may be underestimated here because of the nanosecond time scale resolution of the PL decay measurements. The higher k₁ and lower k₂ and k₃ lay the foundation for the efficient and bright emission from Cs₃CeI₆, promising the potential for fabricating high-performance LEDs.

On the basis of the optimized Cs₃CeI₆ films with desirable morphology and PL properties, we continue to explore their potential applications for RELEDs. The optimized seed layer technique was used for the fabrication of a high-quality Cs₃CeI₆ emitting layer. To accurately determine the energy band structure of Cs₃CeI₆, we conducted UV photoelectron spectra (UPS) and absorption measurements as shown in Fig. 4A and fig. S17. The used device structure is indium tin oxide (ITO)/ZnO:polyethyleneimine ethoxylated (PEIE) (30 nm)/Si₃N₄ (5 nm)/Cs₃CeI₆(50 nm)/cyclohexylidenebis[N,N′-bis(p-tolyl)aniline] (TAPC; 10 nm)/TAPC:1,4,5,8,9,11-hexaazatriphenylene hexacarbonitrile (HAT-CN; 50 nm, 30%)/HAT-CN (5 nm)/Al. The overall flat band energy level diagram of the resulting Cs₃CeI₆-RELED is shown in Fig. 4B. The turn-on voltage was measured to be 4.5 V at a brightness of 1 cd m⁻², which is attributed to the large bandgap and injection barrier of Cs₃CeI₆. The brightness of the devices increases continuously as the applied voltage becomes higher, achieving a maximum luminance (L_max) and EQE of 250 cd m⁻² and 0.9%, respectively.

Fig. 4. Device architecture and performance of Cs₃CeBr_xI_6−xRELEDs.

– (A) UV photoelectron spectra (UPS) data of the Cs₃CeI₆ film. (B) Flat band energy level diagram of Cs₃CeI₆–rare-earth light-emitting diode (RELED). Voltage-current density curves (C), voltage-luminance curves (D), and EQE-luminance curves (E) of Cs₃CeI₆-RELEDs with different hole injection layers [1,3-bis(N-carbazolyl)benzene (MCP) and cyclohexylidenebis[N,N′-bis(p-tolyl)aniline] (TAPC)] and different electron blocking layers (Si₃N₄ and Al₂O₃). HAT-CN, 1,4,5,8,9,11-hexaazatriphenylene hexacarbonitrile; ITO, indium tin oxide. (F) The electroluminescence (EL) spectra of the device operated at various voltages. The inset plot shows the normalized EL spectra. (G) The Commission International de l’Éclairage (CIE) coordinates of Cs₃CeI₆ at various luminance. (H) The CIE comparison of Cs₃CeI₆-RELEDs with Rec. 2020 standard for the blue emitter and some representative deep-blue LEDs reported so far. The CIE values of CsPbBr_3−xCl_x (45) and Cs_xFA_1−xPbBr₃ nanoparticles (46), organic LED (OLED) (47), and PEA₂(Cs_1−xEA_xPbBr₃)PbBr₄ (48) are reported by previous literature. (I) Spectra stability of Cs₃CeBr₂I₄-RELED under continuous working conditions. Note that all the spectra are normalized.

As mentioned above, efficient hole injection is regarded as one significant challenge for RELEDs due to the localized 4f orbitals (14, 18). To solve this problem, 1,3-bis(N-carbazolyl)benzene (MCP) with a deeper highest occupied molecular orbital energy level (−5.9 eV) is rationally selected to enhance the hole injection (36). As shown (Fig. 4, C to E), the device based on MCP exhibits enhanced performance (L_max = 350 cd m⁻², EQE_max = 1.7%) because of the improved hole current (fig. S18). However, the electron current is still much larger than the hole current. We sought to optimize the charge balance by using a denser electron blocking layer of Al₂O₃ by atomic layer deposition (fig. S18) (37, 38). As a result, the optimized device realized an L_max = 470 cd m⁻² and a EQE_max = 3.5%. The operational stability of the Cs₃CeI₆-RELEDs was measured at a constant voltage of 5 V (fig. S19). The half-lifetime of the optimized device was determined to be ~45 min. To evaluate its reproducibility, we measured 21 individual devices and obtained an average EQE of 3.32%, as shown in the histogram in fig. S20, indicating the good reproducibility of our devices fabricated by TE. As summarized in table S2, there is room for the further performance improvement of our Cs₃CeI₆-RELEDs. A better understanding of the EL mechanism and optimization of device structure will promote the performance.

Notably, Cs₃CeI₆-RELED emits a similar deep-blue spectrum to the PL of the powder, with a CIE of (0.15, 0.04). The inset of Fig. 4E displays a vivid photograph of a Cs₃CeI₆-RELED under forwarding voltage, which emits a bright deep-blue light. Our Cs₃CeI₆-RELEDs also show decent spectra stability. We recorded the spectra under different forward working voltages from 5 to 7 V in Fig. 4F. The normalized EL spectra (the inset of Fig. 4F) show negligible change under different voltages, which agrees well with the above photophysical analysis. As a result, the CIE values can be kept constant under different brightness levels (Fig. 4G) or under continuous working conditions (5 V for 60 min in fig. S21), which is necessary for a high-quality display. Moreover, such a CIE value (0.15, 0.04) matches well with Rec. 2020 standard for blue emitters (Fig. 4H).

We have also fabricated mixed halide RELEDs with blue-violet emission peaks at 411 nm by introducing CsBr during TE, verifying the spectral tunability of the Cs₃CeBr_xI_6−x (x = 0 to 6) family. As shown in fig. S22, the mixed halide RELEDs exhibit good EL performance with a maximum brightness of 190 cd m⁻². This mixed halide RELED also shows stable EL with negligible spectra shift under continuous working conditions for about 30 min, as shown in Fig. 4I, which may be due to the higher covalency character of the Ce─X bond. We summarized the EL performance based on Cs₃CeBr_xI_6−x (x = 0 to 6) in table S3.

Considering the high PLQY of Cs₃CeI₆, there is still much room for improvement in its EL device performance. We believe that the main reason for the low performance is the difficulty of carrier injection and transport. As mentioned above, the energy band of Cs₃CeI₆ is mainly contributed by the Ce-5d and Ce-4f orbitals. The highly localized Ce-4f orbits results in hole injection difficulty, causing unbalanced electron and hole current and hence limiting device performance. One way to solve this problem is to select more appropriate hole injection materials (e.g., MoO₃ and CuI) and interface modifications (e.g., polyethylene oxide postprocessing) to boost hole current (39). Another way is to adopt the host-dopant strategy, a mature technique used in OLEDs and has also been proven effective in low-dimensional CsCu₂I₃-LEDs (28, 40).

The advantages of the TE technique could promise the Cs₃CeI₆-RELED for large-scale, patterned, high-resolution display applications. We thus present in Fig. 5 the large-area Cs₃CeI₆-RELEDs and microscale pixel array. First, we have realized a large-scale Cs₃CeI₆-RELED with an effective area of 100 mm², as shown in Fig. 5A. Under forward working voltage, the homogeneous and bright deep-blue emission was achieved. Furthermore, by using the mask process, we realized a patterned RELED of the “HUST” logo, with a scale of 10 mm by 20 mm, as shown in Fig. 5B. At present, the pixel size for commercial display is around 50 μm, which is generally realized by the fine metal mask (FMM) process in the OLED display industry. Learning from the mature pixel technology of OLED, we also obtained the microscale pixel array of Cs₃CeI₆. The fabrication procedures were schematically illustrated in fig. S23. The resulting Cs₃CeI₆ pixel array emits homogeneous and bright deep-blue emissions under UV light excitation, perfectly preserving the good optoelectronic properties of Cs₃CeI₆, as shown in Fig. 5C. Besides, each pixel shows a good surface profile and a clear and sharp edge, precisely reproducing the FMM pixel size.

Fig. 5. Application of Cs₃CeI₆-RELEDs for large-area and high-resolution display.

(A) Photograph of 100 mm² Cs₃CeI₆-LED under working conditions (B) EL photograph of Cs₃CeI₆-RELED with a HUST pattern (10 mm by 20 mm). (C) The images of fluorescence microscopy characterization for the patterned Cs₃CeI₆ film. The pixel size is 100 μm by 100 μm.

DISCUSSION

In summary, we have synthesized a series of Ce-based emitting materials Cs₃CeBr_xI_6-x, demonstrated their deep-blue RELEDs meeting the Rec. 2020 standard, and primarily explored their display application potentials. The bright deep blue emission was ascribed to the spin- and parity-allowed Ce-5d → Ce-4f transition through DFT theory calculation and optical characterization. The high PLQY (76.2%), short exciton lifetime (26.1 ns), and deep-blue CIE (0.15, 0.04) make it a promising candidate for high-quality blue LEDs for display. We then realized the dense and uniform Cs₃CeI₆ film with high crystallinity through TE, with the assistance of a seed layer strategy and high-temperature in situ crystallization. On the basis of this, the Cs₃CeI₆-RELEDs were successfully fabricated, with bright and stable deep-blue light emission. The maximum EQE and brightness reach 3.5% and 470 cd m⁻², respectively. Moreover, by introducing the Br element, Cs₃CeBr_xI_6−x-RELEDs realized tunable CIE values from (0.15, 0.04) to (0.17, 0.02). Unexpectedly, these RELEDs show excellent spectra stability under a continuous electric field, indicating great phase stability and limited ion migration. Thanks to the mature pixel technology of TE, the large-area (100 mm²) RELEDs and pixel array (100 μm) can be easily realized. We believe that our work represents the beginning of efficient direct EL of lanthanide rare-earth compounds.

MATERIALS AND METHODS

Materials

CeI₃ (99.99%) was purchased from Alfa Aesar Co. Ltd. CsI (99.999%) was purchased from Aladdin Reagent Co. Ltd. MCP, TAPC, and HAT-CN were purchased from Xi’an Polymer Light Technology Co. Ltd. Zinc acetate dihydrate, methoxyethanol, and PEIE (80% ethoxylated solution, 37 wt % in H₂O) were purchased from Sigma-Aldrich. All the materials were used directly without further purification. To prepare ZnO:PEIE precursors, 0.5 wt % PEIE was dissolved in methoxyethanol and stirred for at least 12 hours. Then, 0.075 g of zinc acetate dihydrate was added into 1 ml of 0.5 wt % PEIE solution and stirred until the solution became clear. The Zn-to-N mole ratio was about 15:1 for the prepared ZnO:PEIE precursors.

Synthesis of Cs₃CeI₆ powders

The Cs₃CeI₆ powder was synthesized by a solid-state reaction; 0.3 mmol CsI (0.7794 g) and 0.1 mmol CeI₃ (0.5208 g) powder were ground thoroughly until evenly mixed, and the as-prepared powder was transferred to a cleaned quartz tube and then sealed. All the processes were carried out in a nitrogen-filled glove box to avoid water and oxygen erosion. After that, the sealed tube was heated to 800°C and maintained for 10 hours and then dropped to room temperature at the rate of 100°C per hour.

Deposition of the Cs₃CeI₆ film

The Cs₃CeI₆ films were fabricated by thermal coevaporation of CsI and CeI₃ precursors in the eight-source TE system (OMV-FS300, Fangsheng Optoelectronic Co. Ltd.) integrated into a glove box. Note that a seed layer–assisted crystallization method was used here to improve the quality of the Cs₃CeI₆ film. Specifically, the as-cleaned quartz substrates were transferred into the evaporation chamber and a 5-nm seed layer on the substrate at room temperature. Then, we raised the substrate temperature to 250°C to complete the deposition of the subsequent Cs₃CeI₆ film through coevaporation of CsI and CeI₃ with the evaporation rate of 3 and 1 Å/s respectively. It is noted that the CsI and CeI₃ precursors were placed in two sources with opposite positions, and the evaporation rates were monitored by two separate quartz crystal oscillators near the two sources respectively. When the evaporation rates stabilized, we open the baffle and start to deposit. After deposition, the films were naturally cooled to room temperature and encapsulated in a glove box for subsequent characterizations.

Device fabrication

The ITO glass was cleaned with deionized water, detergent, acetone, and ethanol for 30 min in each sonication bath and then dried by flowing nitrogen gas. The ZnO:PEIE precursors were spin-coated at 4000 rpm for 45 s and annealed at 150°C for 10 min. The thickness of the ZnO film was about 30 nm with smooth morphology as confirmed by the atomic force microscope (AFM) measurements (fig. S16). Then, a 5-nm Si₃N₄ or Al₂O₃ insulating layer was deposited on the top of ZnO. The ITO/ZnO/Si₃N₄ substrate was transferred into the vacuum chamber for the deposition of the Cs₃CeI₆ film. Next, TAPC or MCP was deposited as a hole injection layer. After that, a 50-nm hole transport layer of 30% HAT-CN–doped TAPC through coevaporation and a 5-nm hole injection layer of HAT-CN were deposited. Last, 100 nm of Al was deposited as the anode. For large-scale and patterned RELEDs, special large-scale or patterned masks were used when depositing the electrode.

Material and device characterization

XRD measurements were conducted by a Philips X’Pert Pro diffractometer with Cu (Kα) radiation (λ = 1.54 Å). The PL, PLE, and time-resolved PL decay were performed by the Edinburgh FLS920 system. For temperature-dependent PL measurements, the temperature control system was introduced. The absolute PLQY measurements were collected by the quantum yield measurement system with an integrating sphere (Edinburgh FLS980 spectrofluorometer). Films’ morphology and EDS mapping were carried out by the ZEISS GeminiSEM 300 field-emission scanning electron microscopy (SEM). The absorption spectra were obtained by the UV–visible–near-infrared spectrophotometer (Shimadzu Instruments, SolidSpec-3700). For UPS and x-ray photoelectron spectroscopy spectra collection, the AXIS-ULTRA DLD-600W system was used. The EL properties including voltage-current density-luminance curves, EQE-luminance curves, and EL spectra were all measured by the photoelectric integrated test system (XP-EQE-Adv, Xipu Optoelectronics Technology Co. Ltd.), which consists of a Keithley 2400 source meter, an integrating sphere (Labsphere, GPS-4P-SL), and a spectrometer (Ocean Optics). The measurements were completely finished in the glove box to avoid moisture. ZnO film morphology was measured by the AFM (Cypher-Satomic force microscope). The excitation power-dependent PL lifetime measurement was done by the time-correlated single-photon counter technology, where the excitation beam was a picosecond pulsed diode laser (Light Conversion, Pharos) with a 365-nm output wavelength and a 50-ps pulse width. The power density was controlled by the neutral density filters (Light Conversion, Pharos). Time-resolved PL spectra decay was measured by the streak camera system (5200, Xi’an Institute of Optics and Precision Mechanics). To enhance the absorption, the evaporated Cs₃CeI₆ film with a thickness of 1 μm was used in TA measurements. The TA spectra were measured by the TA spectrometer (Helios, Ultrafast Systems). A regenerative amplified Ti:Sapphire (Legend Duo, Coherent Inc.) was used to generate the femtosecond laser pulses of an 800-nm fundamental beam (5-kHz repetition rate and 35-fs pulse width). The pump pulse (325 nm, 390 μW) was generated by using part of the 800-nm fundamental beam to pump an optical parametric amplifier (TOPAS-PRIME, Light Conversion). The broad probe pulse was generated by focusing another part of the fundamental beam into a CaF₂ crystal. Both the pump and probe pulses were directed into the TA spectrometer.

First-principles DFT calculations

The Vienna Ab initio Simulation Package within the framework of DFT was used to conduct the calculations (41). The generalized gradient approximation exchange-correlation function is described by Perdew-Burke-Ernzerhof, and the projector augmented wave pseudopotential was introduced with a cutoff of 480 eV (41–44). A Hubbard (+U) correction with a simplified version of Cococcioni and de Gironcoli was used to describe the strongly correlated f-electrons. Before the calculation, experiment-determined ionic position and crystal lattice parameters were relaxed, with the thresholds of 0.02/Å and 10⁻⁴ eV for total energy and forces. The primitive cell contains 6 Cs atoms, 2 Ce atoms, and 12 I atoms, and the 3 by 3 by 3 Gamma-centered k-points were used for the Brillouin zone sampling.

Supplementary Materials

This PDF file includes:

Tables S1 to S3

Figs. S1 to S23

References