Efficient multicolor X-ray excited persistent luminescence enabled by Gd-mediated trap clusters

Abstract

Persistent luminescence materials are promising for night-vision displays, background-free medical diagnostics, and high-resolution radiography, yet achieving efficient violet, yellow, and red emission within a single robust and scalable host remains a longstanding challenging. Here, we overcame this limitation by constructing Gd³⁺-mediated cluster traps within alkaline-earth fluorochlorides to minimize energy loss during electron migration. These clusters serve as both intrinsic emitters and efficient energy transfer platforms for various activators, including Eu²⁺, Sm²⁺, Tb³⁺, and Mn²⁺, enabling bright and spectrally tunable multicolor persistent luminescence upon X-ray irradiation. The persistent luminescence intensity of Eu²⁺ is enhanced by up to 32.7-fold upon Gd³⁺ codoping. Moreover, violet persistent luminescence from Eu²⁺ is employed to excite perovskite quantum dots for full-color time-domain dynamic displays, while Sm²⁺ emission facilitates low-dose, high-resolution delayed X-ray imaging. These findings establish a generalizable strategy for designing efficient multicolor persistent materials for advanced multifunctional optical technologies.

Achieving efficient violet, yellow, and red emission in a single robust, scalable host is challenging. The authors achieve this by constructing Gd³⁺-mediated cluster traps in alkaline-earth fluorochlorides to reduce energy loss during electron migration.

Introduction

Persistent luminescence (PersL) is a fascinating optical phenomenon wherein materials continue to emit photons following the cessation of excitation^1–4. This time-delayed photon emission holds great promise for a wide range of applications, including optical data storage^5–7, night-vision displays⁸, background-free medical diagnostics⁹ and high-resolution radiography^10,11. Despite being discovered in the early 17th century, efficient inorganic PersL materials remain largely confined to blue- (e.g., CaAl₂O₄:Eu/Nd)^12,13 and green-emitting (e.g., SrAl₂O₄:Eu/Dy and Sr₂MgSi₂O₇:Eu/Dy)^14,15 systems. Longer-wavelength emissions, such as yellow and red, are typically achieved in sulfide-based materials^16,17, which suffer from poor stability and low brightness, while efficient violet PersL yet to be demonstrated. Extending PersL into the violet, yellow, and red spectral regions with high efficiency, especially using environmentally robust and scalable hosts remains a formidable challenge, hindering progress toward full-color displays^18–20, multiplexed bioimaging²¹, and multichannel information encryption^22–24. More critically, realizing multicolor PersL within a single host is exceptionally difficult, yet essential for minimizing compatibility issues and simplifying integration into multi-functional optical devices.

Room-temperature PersL typically arises from the spontaneous release of charge carriers trapped in shallow energy states, followed by their migration through the host lattice and subsequent radiative recombination at luminescent centers²⁵. However, long-range carrier migration is often accompanied by substantial non-radiative losses, including multi-phonon relaxation²⁶, defect-induced quenching²⁷, and inefficient energy transfer^28,29, which significantly compromises overall luminescence efficiency. In upconversion luminescence, spatially confining sensitizers and activators into clusters has proven effective for suppressing energy loss by shortening energy transfer distances³⁰. Drawing inspiration from this strategy, we propose that co-localizing traps and activators into confined clusters can significantly enhance PersL by minimizing migration-associated energy dissipation, which in turn provides a powerful approach to realize bright multicolor PersL within a single host lattice through doping with various activators.

As a proof-of-concept, we introduce a Gd³⁺-mediated cluster engineering strategy in alkaline-earth fluorochloride hosts to confine trap carrier migration (Fig. 1). Substituting divalent alkaline-earth with trivalent Gd³⁺ induces the formation of [-Gd-F-]_n clusters. Upon X-ray irradiation, fluoride ions coordinated to Gd³⁺ preferentially form Frenkel defects, resulting in Gd³⁺-associated trap clusters that produce intense PersL from Gd³⁺ itself. More importantly, these clusters serve as efficient energy transfer platforms to various activators, including Eu²⁺, Sm²⁺, Tb³⁺, and Mn²⁺, enabling bright and spectrally tunable multicolor PersL. Notably, emissions from Eu²⁺ and Sm²⁺, despite their narrower full widths at half maximum (FWHM), exhibit much higher brightness than the benchmark green-emitting SrAl₂O₄:Eu/Dy. Moreover, these fluorochloride-based systems demonstrate superior air stability compared to commercial aluminates. We also demonstrate that the violet PersL from Eu²⁺ can serve as an excitation source for perovskite quantum dots, enabling full-color dynamic displays. Meanwhile, the red PersL from Sm²⁺ facilitates low-dose, high-resolution delayed X-ray imaging.

Fig. 1. Design of Gd³⁺-mediated trap clusters for enhanced PersL.

Schematic illustration of structural evolution in MFCl (M = Ca, Sr, Ba) upon Gd³⁺ doping and the corresponding charge transfer (CT) and energy transfer (ET) pathways. For clarity, Cl^- ions are omitted from the crystal structures. Two Gd³⁺ ions substitute for three Ba²⁺ ions, generating one Ba²⁺ vacancy to maintain charge neutrality. The Gd³⁺ dopants tend to aggregate within the crystal lattice, promoting the formation of X-ray-induced Frenkel defects in Gd-rich clusters. Upon Gd³⁺ codoping, the detrapping process predominantly occurs within these clusters, where released electrons migrate toward Gd³⁺ ions and recombine with captured holes, populating the excited states of Gd³⁺. Subsequently, ET from Gd³⁺ to the activator ions gives rise to strong multicolor PersL. In contrast, the absence of such clusters leads to long-range carrier migration and increased non-radiative losses, yielding faint PersL.

Results

Characterization and benchmarking of lanthanide-doped BaFCl phosphors

Lanthanide-doped BaFCl phosphors were scalable synthesized via a conventional scalable solid-state reaction, but under significantly milder conditions (1073 K for 1 h, Supplementary Fig. 1) compared to those typically required for commercial SrAl₂O₄:Eu/Dy (>1573 K for over 4 h)³¹. X-ray diffraction (XRD) patterns of the products with and without Gd³⁺ doping matched well with the phase-pure tetragonal BaFCl structure (Supplementary Fig. 2), featuring an optical bandgap of 5.73 eV (Supplementary Fig. 3). Atomic-resolution integrated differential phase contrast scanning transmission electron microscopy (iDPC-STEM, Fig. 2a) and high-resolution TEM (Supplementary Fig. 4) revealed a highly crystalline microstructure. Elemental mapping (Fig. 2b) and energy-dispersive X-ray (EDX) spectroscopy (Supplementary Fig. 5) demonstrated the presence of F, Cl, Ba, Gd and Eu elements. Inductively coupled plasma optical emission spectrometry (ICP-OES) further confirmed the Eu and Gd doping levels in BaFCl:2Eu/1Gd to be ~1.51 and 0.62 mol%, respectively, closely agreement with the nominal composition. X-ray photoelectron spectroscopy (XPS) of the Eu 3d states revealed the coexistence of Eu²⁺ (1125, 1155 eV) and Eu³⁺ (1135, 1165 eV)³², with the Eu³⁺/Eu²⁺ ratio remaining unaffected by Gd³⁺ codoping (Supplementary Fig. 6).

Fig. 2. Structural characterization and PersL performance of lanthanide-doped BaFCl.

a iDPC-STEM image of BaFCl:2Eu/1Gd. b Elemental mapping of Ba, F, Cl, Eu, and Gd. c PersL spectra of BaFCl:2Eu/Gd with varying Gd concentrations; inset shows normalized integrated PersL intensities. d PersL decay profiles of BaFCl:2Eu/Gd with different Gd concentrations. e PersL spectra of BaFCl:2Eu/1Gd under different X-ray tube voltages (tube current, 200 μA); inset shows corresponding normalized integrated intensities. Comparative PersL spectra (f) and decay curves (g) of BaFCl:2Eu/1Gd and representative commercial persistent phosphors. h Long-term PersL stability under repeated irradiation at one-day intervals for BaFCl:2Eu/1Gd, SrAl₂O₄:Eu/Dy and ZnS:Cu/Co. The XEOL intensity of commercial Gd₂O₂S:Tb is used as an internal reference to correct for instrumental fluctuations. i PersL photographs of BaFCl:2Eu and BaFCl:2Eu/1Gd at different delay times after X-ray irradiation. X-rays, 50 kV. Source data are provided as a Source Data file.

Upon cessation of X-ray irradiation, the BaFCl:Eu exhibited weak PersL, featuring a sharp peak at 365 nm (Eu²⁺: ⁶P_7/2 → ⁸S_7/2) and a broad emission centered at 387 nm (Eu²⁺: 4f⁶5d¹ → ⁸S_7/2) (Supplementary Fig. 7)^33,34. The optimal Eu doping concentration was determined to be ~2 mol% (Fig. S7b, c). Notably, characteristic Eu³⁺ emission was absent, likely due to its extremely low probability of capturing trapped carriers¹⁰. As shown in Fig. 2c, the PersL intensity of BaFCl:Eu is markedly enhanced upon Gd³⁺ codoping. At an optimal Gd³⁺ concentration of 1 mol%, the integrated PersL intensity increases by up to 32.7-fold. The PersL decay curves reveal that both the peak intensity and afterglow duration are substantially improved with Gd³⁺ incorporation (Fig. 2d). The BaFCl:Eu/Gd exhibits rapid charging kinetics, reaching saturation within ~30 s, and maintains detectable PersL for over 100 h at room temperature (Supplementary Fig. 8). Time-lapse photographic images further corroborate the enhanced and long-lasting brightness across various decay intervals (Fig. 2i). Elevating the X-ray tube voltage from 10 to 50 KV resulted in an additional 65.8-fold enhancement in PersL intensity (Fig. 2e), attributed to the increased secondary electron generation and more efficient trap filling at higher excitation energies.

To benchmark performance, we compared the BaFCl:Gd/Eu with several commercial persistent phosphors, including CaAl₂O₄:Eu/Nd, SrAl₂O₄:Eu/Dy, Sr₁₃Al₂₂Si₁₀O₆₆:Eu, ZnS:Cu/Co, and CaS:Eu. Remarkably, the BaFCl:Gd/Eu outperformed all references (Fig. 2f), exhibiting significantly stronger PersL across a wide range of X-ray tube voltages (Supplementary Fig. 9). Notably, it shows a much narrower FWHM (~30 nm) compared to 72 nm for SrAl₂O₄:Eu/Dy, while its integrated PersL intensity is ~3.5 times higher under identical conditions (Supplementary Fig. 10). More importantly, the PersL intensity of BaFCl:Gd/Eu decreased by only ~5% after 60 days of air exposure with repeated irradiation and daily measurements, markedly outperforming SrAl₂O₄:Eu/Dy and ZnS:Cu/Co (Fig. 2h, and Supplementary Fig. 11), highlighting its superior long-term durability. As shown in Supplementary Fig. 12, the PersL intensity and spectral profile remained nearly unchanged during 60 continuous irradiation cycles (1 min each), confirming that the decrease in PersL intensity of SrAl₂O₄:Eu/Dy is not associated with X-ray irradiation damage. Despite the extremely low sensitivity of the human eye to violet light, the violet PersL luminance from BaFCl:Gd/Eu still reaches about 0.16 cd m⁻² at 2 s after ceasing X-ray irradiation. This outstanding performance is attributed to the high X-ray absorption coefficient and the effective trapping of charge carriers (Supplementary Fig. 13). Additionally, the BaFCl:Gd/Eu achieves a photoluminescence quantum yield (PLQY) of 92.7%, far exceeding that of SrAl₂O₄: Eu/Dy (54.7%) (Supplementary Fig. 14). It should be noted that the comparison of reported quantum yields may be affected by the fact that SrAl₂O₄: Eu/Dy exhibits trapping and detrapping under UV excitation, which can reduce the overall efficiency, whereas BaFCl:Eu does not.

Mechanistic investigation

To uncover the origin of the enhanced PersL, we investigated how heterovalent Gd³⁺ doping modifies the local structure, trap states, and energy migration dynamics. Extended X-ray absorption fine structure (EXAFS) spectroscopy was employed to probe the coordination environment of Gd³⁺ in the BaFCl lattice. As shown in Fig. 3a, Supplementary Fig. 15, EXAFS fitting reveals that Gd³⁺ is coordinated by an average of ~2.1 F⁻ and 2.1 Cl⁻ ions, notably lower than the corresponding values for native Ba²⁺ sites (4.2 for F⁻ and 5.1 for Cl⁻, Fig. 3b). In addition, the Gd–F bond length is ~1.88 Å, much shorter than the Ba–F bond length of 2.54 Å (Supplementary Table 1), indicating evident local lattice contraction surrounding the Gd³⁺ dopants. Despite the Gd³⁺ doping level being below 1 mol%, a distinct Gd–Gd scattering signal is observed in the Fourier-transformed EXAFS spectrum (Fig. 3c). This unexpected signal, combined with the reduced coordination numbers and shortened bond lengths, manifesting the formation of locally aggregated [-Gd-F-]_n clusters³⁵. In addition, Raman spectroscopic confirms the appearance of an additional Gd–F vibrational mode^36,37, in addition to the characteristic phonon modes of the BaFCl host lattice (Supplementary Fig. 16).

Fig. 3. Mechanistic investigations of the improved PersL.

Experimental FT-EXAFS spectra and corresponding fitting curves for the Ba L-edge (a) and Gd L-edge (b). c Wavelet transform of the Gd L-edge EXAFS signals. d Schematic illustration of the generation of three types of F⁻ Frenkel defects and their corresponding E_f values in the constructed BaFCl crystal model containing [-Gd-F-]_n clusters. F⁻ surface, F⁻ ion located on the surface of the cluster. F⁻ inner, F⁻ ion inside the cluster. F⁻ bulk, F⁻ ion bonded to Ba. e Temperature-dependent X-ray active TL glow contour mapping recorded at a heating rate of 60 K min⁻¹. f PersL spectra of BaFCl:1Gd and BaFCl:2Eu/1Gd. Monte Carlo simulations of energy migration dynamics from trap to activators in two configurations, defects and activators randomly distributed (g), and defects and activators aggregated into clusters (h). Source data are provided as a Source Data file.

To further elucidate the structure of the clusters, density functional theory (DFT) simulations were performed to model the Gd L-edge X-ray absorption near edge structure (XANES) spectra for different Gd³⁺ substitution scenarios. Among the constructed crystal models, the configuration in which six Gd³⁺ ions replace nine Ba²⁺ ions and create three Ba²⁺ vacancies to maintain charge neutrality, produced a calculated XANES spectrum that closely matches the experimental data (Supplementary Fig. 17a, d). In contrast, models with either a single Gd³⁺ substituting one Ba²⁺, or two Gd³⁺ ions replacing three Ba²⁺ ions yielded simulated XANES profiles that deviate significantly from the experimental spectrum (Supplementary Fig. 17b, c, e, f). These results indicate a strong tendency for Gd³⁺ ions to aggregate within the BaFCl lattice, forming Gd-rich clusters.

Upon X-ray irradiation, high-energy photons interact with heavy atoms primarily through the photoelectric effect, generating hot electrons and deep holes³⁸. These primary carriers undergo multiple scattering and Auger processes, producing abundant low-energy secondary electrons. Concurrently, the F⁻ related Frenkel defects are formed, acting as carrier traps and generating PersL after the cessation of X-rays^39–43. As shown in Supplementary Fig. 18, the current gradually increases upon X-ray irradiation and decays slowly with a delay exceeding 7 s after the X-ray is turned off, while the peak current rises at higher tube voltages. These observations indicate that the measured conductivity predominantly arises from defect-related ionic transport under X-ray irradiation. DFT calculations reveal the formation energy (E_f) of F⁻ Frenkel defects within the [-Gd-F-]_n clusters, is substantially lower than that of F⁻ bound to Ba²⁺, indicating the traps preferentially form within the clusters (Fig. 3d). Especially, for surface F⁻ ions within the clusters, the E_f for displacement toward the cluster is significantly lower than that for displacement away from the cluster. In this scenario, both the trapped carriers and the associated recombination energy remain spatially confined within the clusters. To probe the role of Ln³⁺ doping on trap states while avoiding interference from Gd³⁺-mediated energy transfer, the optically inert Lu³⁺ ion with an ionic radius similar to that of Gd³⁺ was employed as a control dopant. X-ray active thermoluminescence (TL) analysis reveals that Lu³⁺ doping results in similar shallow traps, with no significant changes in trap depth (~0.76 eV) or density (Supplementary Fig. 19)⁴⁴. In contrast, codoping with Gd³⁺ leads to a substantial enhancement in TL intensity below 400 K (Fig. 3e, and Supplementary Fig. 20), which is attributed to efficient charge transfer from the trapped carriers to Gd³⁺ ions within the clusters, followed by subsequent transfer to Eu²⁺.

Unlike X-ray excited PersL, which arises from the thermal release and migration of trapped electrons, X-ray excited optical luminescence (XEOL) originates from the recombination of free-electrons in the conduction-band with holes⁴⁵. As shown in Supplementary Fig. 21, in the absence of Eu²⁺, strong characteristic Gd³⁺: (⁶P_5/2 → ⁸S_7/2 at 311 nm and ⁶P_{7/ 2} → ⁸S_7/2 at 316 nm) emissions are observed in the XEOL spectrum⁴⁶. With increasing temperature or excitation power, the I_316nm/I_311nm intensity ratio increases, indicating enhanced electron–phonon coupling and elevated nonradiative relaxation via the ⁶P_5/2 → ⁶P_7/2 transition (Supplementary Fig. 22). Upon Eu²⁺ codoping, the XEOL intensity of Gd³⁺ is nearly fully quenched (Supplementary Fig. 23), suggesting a near-unity energy transfer efficiency from Gd³⁺ to Eu²⁺ (Supplementary Fig. 24). A similar suppression is observed in the PersL intensity (Fig. 3f, and Supplementary Fig. 25) and decay curves (Supplementary Fig. 26) of Gd³⁺, further supporting this efficient energy transfer pathway. This efficiency arises from both the mean Gd–Eu distance of ~0.81 nm (Supplementary Fig. 27a, b), which is far below the typical critical distance of 2 to 6 nm for Förster resonance energy transfer^47,48, and the strong overlap between the Eu³⁺ excitation spectrum and the Gd³⁺ emission spectrum (Supplementary Fig. 27c). Notably, the PersL from Gd³⁺ in BaFCl: Gd lasts for more than 168 h (Supplementary Fig. 28). However, the XEOL intensity and pulsed X-ray excited life-time of Eu²⁺ in the BaFCl: Eu/Gd are similar to those in BaFCl: Eu and BaFCl: Eu/Lu (Supplementary Fig. 29), indicating that Gd³⁺ codoping does not enhance the excitation efficiency of the Eu²⁺: 5d states via free-electron recombination. This behavior is consistent with the spectral overlap between the broad Eu²⁺ excitation band and the Gd³⁺ emission region, where Gd³⁺ acts as an alternate excitation route rather than enhancing the excitation probability. This interpretation is further supported by the comparable photoluminescence intensity of Eu²⁺ under various UV excitation wavelengths, irrespective of Gd³⁺ codoping (Supplementary Fig. 30). These results reveal that if trapped electrons are thermally released as free-electrons before captured by activators, Gd³⁺ exerts minimal influence on the PersL intensity of Eu²⁺. Therefore, the observed enhancement in PersL upon Gd³⁺ doping is attributed to the formation of Gd³⁺-mediated trap clusters, which localize the trapped electrons and enable their efficient transfer to nearby Eu²⁺ ions. Monte Carlo simulations³⁰ further reveal that spatial clustering of traps and activators within the lattice significantly increases the activation probability of luminescent centers compared to random distribution (Fig. 3g, h), confirming that energy clustering effectively minimizes energy loss during long-range carrier migration.

Universal enhancement of PersL across diverse activators and hosts

To validate the universality of the Gd³⁺-mediated trap cluster enhancement mechanism and expand its applicability to multi-color PersL, we investigated additional activators, including divalent Sm²⁺, trivalent Tb³⁺ and transition metal Mn²⁺, as well as alternative fluorochlorides hosts such as CaFCl and SrFCl. As shown in Fig. 4a–e and Supplementary Fig. 31, all three dopant systems in the BaFCl host exhibited pronounced increases in PersL intensity upon Gd³⁺ codoping. The enhancements reached up to 150.3-fold for Sm²⁺, 22.9-fold for Tb³⁺, and 20.2-fold for Mn²⁺ (Fig. 4f), attributed to efficient cascade energy transfer from the trap cluster to Gd³⁺, and subsequently to the activator (Supplementary Fig. 32). The corresponding decay profiles further confirm that Gd³⁺ incorporation significantly improves both the PersL intensity and duration (Fig. 4g–i). Time-lapse photographic images intuitively demonstrate enhanced brightness across various decay intervals (Fig. 4j–l). In particular, in the absence of Gd³⁺, the PresL from Sm²⁺ and Mn²⁺ was invisible to the naked eye within 1 s post irradiation, but display clear visible emission even after 60 s with Gd³⁺ codoping. The initial PersL luminance values of the Sm²⁺, Tb³⁺, and Mn²⁺-doped systems reached 3.28, 0.87, and 0.56 cd m⁻², respectively. The BaFCl:Gd/Sm achieves a high PLQY of 84.2% (Supplementary Fig. 33), and its integrated red PresL intensity was ~2.2 times higher than that of green SrAl₂O₄:Eu/Dy and ~47.8 times greater than that of red CaS:Eu (Supplementary Fig. 34). Furthermore, the Gd³⁺ codoping strategy was extended to other hosts, including CaFCl:Eu and SrFCl:Eu, yielding enhancements in PersL intensity by 101.5 and 107.8 times, respectively (Fig. 4d–f). These results highlight the broad compatibility of the Gd³⁺-mediated trap cluster engineering approach across different activator ions and host lattices. Notably, the XEOL intensity of these systems does not increase upon Gd³⁺ codoping (Supplementary Fig. 35), further confirming that the enhanced PersL originates from the formation of Gd³⁺-mediated trap clusters.

Fig. 4. PersL characterization across diverse activators and hosts.

Comparison of PersL spectra with and without Gd³⁺ incorporation in: a BaFCl:2Sm and BaFCl:2Sm/1Gd, b BaFCl:2Tb and BaFCl:2Tb/1Gd, c BaFCl:1Mn and BaFCl:1Mn/1Gd, d CaFCl:2Eu and CaFCl:2Eu/1Gd, e SrFCl:2Eu and SrFCl:2Eu/1Gd. f Integrated PersL intensities of the corresponding samples, normalized to those without Gd³⁺. PersL decay curves of BaFCl:2Sm (g), BaFCl:2Tb (h), and BaFCl:1Mn (i) with and without Gd³⁺ codoping. j–l PersL photographs of the samples in g–i, captured under identical conditions. X-rays, 50 kV. A long-pass filter (λ > 420 nm) was used for capturing Mn PersL photographs. Source data are provided as a Source Data file.

Full color PersL dynamic displays and low-dose delayed X-ray imaging

The superior violet-range PersL performance of BaFCl:Eu/Gd enables its use as an excitation source for halide perovskite (CsPbX₃, X = Cl, Br, I) quantum dots (QDs) (Supplementary Fig. 36), thereby achieving full-spectrum PersL across the visible range. The bandgap and corresponding emission wavelength of the QDs were tuned by varying the halide composition⁴⁹. A flexible BaFCl:Eu/Gd-integrated flexible PDMS film was placed above dispersions of different QDs, irradiated with X-rays for 30 s, and the resulting PersL photographs were captured using a smartphone after the cessation of X-rays (Supplementary Fig. 37). As shown in Fig. 5a and Supplementary Fig. 38, varying the QD composition from CsPbCl₂Br₁ to CsPbBr₃ and CsPbI₃ shifts the persistent emission peaks from 433 to 508 and 686 nm, respectively, covering a continuous color range from blue through green to red. Notably, the QD dispersions alone exhibited no observable XEOL or PersL under X-ray excitation (Supplementary Fig. 39), indicating the full-color PersL arises from radiative energy transfer from the BaFCl:Eu/Gd PersL to the QDs via reabsorption⁵⁰. To further demonstrate dynamic time-domain multicolor display capability, we designed a composite pattern consisting of BaFCl doped with Eu/Gd, Sm/Gd, and Mn/Gd, along with four selected QDs (CsPbCl₁Br₂, CsPbCl_0.5Br_2.5, CsPbBr_1.3I_1.7 and CsPbBr₂I₁) placed above the BaFCl:Eu/Gd (Fig. 5b). Upon X-ray irradiation, each region displayed distinct PersL colors corresponding to their respective emitters. As Eu²⁺ PersL decayed over time, the emission from the QDs faded sequentially: yellow (~10 s), followed by red (~60 s). After 120 s, the Sm²⁺ PersL became invisible due to its faster decay compared to Eu²⁺ and Mn²⁺, while the emission from CsPbCl_0.5Br_2.5 shifted from blue-green to blue.

Fig. 5. Multicolor PersL displays and delayed X-ray imaging.

a Schematic of full-color PersL realization using BaFCl:Eu/Gd as the excitation source, along with corresponding PersL spectra and photographs. PersL spectra of QDs from left to right are CsPbCl₂Br₁, CsPbCl₁Br₂, CsPbCl_0.5Br_2.5, CsPbBr₃, CsPbBr₁I₂, CsPbBr_1.3I_1.7, CsPbBr₂I₁ and CsPbI₃. Scale bar, 10 mm. X-rays, 50 kV. b Time-delayed PersL photographs of a composite pattern comprising BaFCl doped with various activators (Eu/Gd, Sm/Gd, Mn/Gd) and four types of QDs: (1) CsPbCl₁Br₂, (2) CsPbCl_0.5Br_2.5, (3) CsPbBr_1.3I_1.7, and (4) CsPbBr₂I₁. c MTF curve of delayed X-ray image recorded from the BaFCl:Gd/Sm film. LP, line pairs. d Photographs of two commercial circuit boards and their corresponding delayed X-ray images. Source data are provided as a Source Data file.

Benefiting from its rapid charging capability, the BaFCl:Sm/Gd-integrated transparent PMMA film enables low-dose PersL imaging. As shown in Supplementary Fig. 40, after just 5 s of X-ray irradiation at a dose of 1.6 μGy, a time-delayed X-ray image of the standard Pb line-pair template demonstrates a resolution of up to 20 lp mm⁻¹, as validated by gray-value contrast analysis. Further analysis of the modulation transfer function (MTF) (Fig. 5c) confirms a spatial resolution of ~25.7 lp mm⁻¹ at an MTF value of 0.2⁵¹. The dose rate (0.32 Gy_air/s) is an order of magnitude lower than that used in medical diagnostics (5.5  μGy_air/s)^52,53, while the resolution outperforms most previously reported PersL systems (Supplementary Table 2) and conventional flat-panel X-ray detectors (typically lower than 5 lp  mm⁻¹)⁵⁴. Moreover, time-delayed images reveal embedded structures in commercial circuit boards that are otherwise invisible (Fig. 5d), highlighting the potential of this system for ultralow-dose, high-resolution, and background-free radiographic imaging.

Discussion

In conclusion, we demonstrate efficient multicolor PersL, spanning violet, yellow, and red spectral regions, within scalable and stable alkaline-earth fluorochloride systems. By constructing Gd³⁺-mediated trap clusters, we establish a general platform that facilitates efficient energy transfer to diverse activators, including Eu²⁺, Sm²⁺, Tb³⁺, and Mn²⁺. This trap cluster design significantly suppresses long-range electron migration losses, resulting in highly efficient PersL. The red PersL from Eu²⁺ is remarkably enhanced, by up to 32.7-fold, through Gd³⁺ codoping. Notably, BaFCl:Gd/Eu exhibits a narrow emission bandwidth but an integrated PersL intensity 3.5 times higher than that of the benchmark SrAl₂O₄:Eu/Dy. In addition, the developed PersL materials exhibit excellent air stability. We further demonstrate the utility of bright violet Eu²⁺ PersL as an excitation source for perovskite quantum dots to enable full-color dynamic displays, and employ Sm²⁺ PersL for low-dose, high-resolution delayed X-ray imaging. These findings provide a generalizable design strategy for the development of efficient, multicolor, and environmentally robust PersL materials for advanced optical technologies.

Methods

Chemicals and reagents

BaF₂ (99.9%), BaCl₂·2H₂O (99.99%), EuCl₃·6H₂O (99.9%), GdCl₃·6H₂O (99.99%), SmCl₃·6H₂O (99.9%), MnCl₂·4H₂O (99.9%), TbCl₃·6H₂O (99.99%), CaF₂ (99.5%), CaCl₂ (99.9%), SrF₂ (99.9%), SrCl₂·6H₂O (99.99%), Cs₂CO₃ (99.9%), C₃H₉ClSi (99%), C₃H₉BrSi (98%), C₃H₉SiI (97%), Pb(C₂H₃O₂)₂·3H₂O (99.99%), cyclohexane (99.9%), N,N-dimethylformamide (DMF, ≥98%), Poly (methyl methacrylate) (PMMA) were all purchased from Shanghai Aladdin Reagent Co. 1-Octadecene (ODE, 90%), Oleic acid (OA, 90%) and Oleylamine (OAm, ≥98%) were supplied by Sigma Aldrich Co. Commercial SrAl₂O₄:Eu/Dy, CaAl₂O₄:Eu/Nd, Sr₁₃Al₂₂Si₁₀O₆₆:Eu, ZnS:Cu/Co, and CaS:Eu phosphors were purchased from Shenzhen Looking Long Technology Co. Gd₂O₂S:Tb was purchased from Felles Photonic Instruments (China) Co. Polydimethylsiloxane (PDMS) was purchased from Shenzhen Song Shine Technology Co. All chemicals employed in the experiment were of analytical grade and used directly without further treatment.

Synthesis of BaFCl: 2Eu/1Gd and other phosphors

BaFCl:2Eu/1Gd was synthesized via a high-temperature solid-state reaction. A stoichiometric mixture of 10 mmol BaF₂, 9.4 mmol BaCl₂·2H₂O, 0.4 mmol EuCl₃·6H₂O and 0.2 mmol GdCl₃·6H₂O was thoroughly ground in a mortar for 15 min. The resulting powder was heated at a rate of 7 °C/min to 800 °C and sintered for 1 h under a flowing 5% H₂/N₂ atmosphere. After cooling to room temperature, the product was ground for 5 min to yield the BaFCl:2Eu/1Gd phosphor. Other BaFCl-based phosphors, including BaFCl: Eu, BaFCl:Sm/Gd, BaFCl:Mn/Gd, BaFCl:Tb/Gd were synthesized using the same procedure by substituting the corresponding dopant precursors.

Synthesis of flexible BaFCl:Eu/Gd-integrated PDMS film

BaFCl:Eu/Gd phosphor (0.5 g) was dispersed into 1.1 g of PDMS (prepolymer:crosslinker = 10:1 by weight) and stirred for 20 min at room temperature to achieve a homogeneous mixture. The resulting slurry was then cured at 70 °C for 2 h, yielding a flexible BaFCl:Eu/Gd-integrated PDMS film.

Synthesis of cesium-oleate (Cs-OA) precursor

Cs₂CO₃ (0.409 g), ODE (20 mL), and OA (1.25 mL) were added into a 50 ml three-neck flask. The mixture was degassed under N₂ at 120 °C for 1 h, then heated to 150 °C until a clear solution formed. The resulting Cs-oleate solution was transferred to a 10 mL volumetric flask and stored for subsequent use.

Synthesis of CsPbX₃ quantum dots (QDs)

Pb(C₂H₃O₂)₂·3H₂O (0.0759 g) was loaded into a 50 mL three-neck flask, followed by the sequential addition of 5 mL ODE, 2.5 mL OA, 2.5 mL OAm, and 250 µL Cs-OA. Then mixture was heated to 110 °C under a nitrogen atmosphere and maintained for 45 min. Subsequently, the temperature was rapidly increased to 220 °C, at which point 200 µL of a halide precursor solution (Cl, Br, and I in the desired molar ratio) was quickly injected. After 10 s of reaction, the flask was immediately cooled in an ice-water bath to room temperature. The resulting solution was centrifuged at ~5700 × g for 5 min to remove the supernatant. The precipitate was redispersed in 6 mL cyclohexane and centrifuged again at ~4600 × g for 5 min.

Synthesis of composite pattern

BaFCl:Eu/Gd (1 g), BaFCl:Sm/Gd (0.3 g), and BaFCl:Mn/Gd (1 g) phosphors were each dispersed in 1.1 g, 0.33 g, and 1.1 g of PDMS, respectively. The mixtures were then cast into a silicone butterfly mold, with each phosphor occupying designated regions. The mold was then cured at 70 °C for 2 h to form the composite pattern. The wings embedded with BaFCl:Eu/Gd phosphor were further decorated with various halide perovskite QDs.

Synthesis of BaFCl:Sm/Gd-integrated film

PMMA powder (0.65 g) was dissolved in 3 mL of DMF by heating to 70 °C under stirring for 4 h to obtain a transparent solution. A 1 mL aliquot of this solution was then mixed with 50 mg of BaFCl:Sm/Gd and stirred for 2 h until homogeneous. The resulting mixture was ultrasonicated for 5 min, poured into a glass mold, and left to stand for 10 min to eliminate air bubbles. Finally, the film was cured at 70 °C for 2 h to obtain the BaFCl:Sm/Gd-integrated film.

Characterizations

X-ray diffraction (XRD) analysis was collected using a powder diffractometer (Bruker D8 Advance) equipped with a Cu-Kα radiation (λ = 1.5405 Å). Microstructure images and elemental compositions was obtained using a scanning electron microscope (SEM) (Japan Hitachi SU 8010) equipped with an energy dispersive X-ray spectroscopy (EDS, Aztec X-Max 80T). X-ray photoelectron spectroscopy (XPS) analyzes were performed using Thermo Fisher Scientific K-Alpha with a tube voltage of 15 kV and a tube current of 10 mA. Photoluminescence (PL) and photoluminescence excitation (PLE), and photoluminescence quantum yield (PLQY) were carried out using Edinburgh FS-5 spectrofluorometer (Xenon Lamp, 150 W). X-ray excited optical luminescence (XEOL) spectra, X-ray excited persistent luminescence (PersL) spectra, and PersL decay curves were recorded using a spectrometer (OmniFluo 960 X-ray, Zolix, China) coupled with an X-ray tube and Hamamatsu CR131 Photomultiplier Tube. The afterglow curves reported in this work represent photon counts directly obtained from the spectrometer. The PersL spectra presented in Fig. 4a–c, Supplementary Figs. 8a, 22, 31, 34, 35a, b, and 38 were recorded using a fiber-coupled spectrometer (QEpro) linked to the X-ray source, with an acquisition time of ~5 s. Inductively coupled plasma optical emission spectrometry (ICP-OES) was performed using an Agilent 7700 instrument. Thermoluminescence (TL) curve was tested by LTTL-3DS type TL 3D spectrometer. Raman microscope (Gloucestershire, UK) was utilized to acquire Raman spectra. The temperature-dependent XEOL spectra were recorded by TAIZI Godzilla fiber spectrometer (Shanghai, China) with a copper heater. Atomic-resolution integrated differential phase contrast scanning transmission electron microscopy (iDPC-STEM) and high-resolution TEM images were conducted using a Double spherical aberration-corrected transmission electron microscope (Spectra 300, Thermofisher, US). UV-vis spectra were measured using a Craic Technologies micro-spectrophotometer. Pulsed X-ray excited decay curve was measured using an X-ray excitation luminescence lifetime test system (picoX 5084, Shanghai UPU Optoelectronic Technology co.). X-ray imaging experiments were conducted using a custom-built setup including a miniature X-ray tube and an ORCA-Fusion BT camera (C15440 - 20UP).

Measurement of PersL luminescence brightness: 100 mg of phosphor powder was evenly spread on a transparent plate and gently pressed with a glass plate to ensure uniform thickness. The prepared sample plate was then placed on the probe of a photometer (Xinbao Instrument, SM218E). The sample was irradiated with X-rays at 50 kV for 1 min, and the initial afterglow luminance was recorded immediately after the X-ray source was turned off.

First-principles DFT calculations were carried out using the Vienna Ab initio Simulation Package (VASP, version 5.4.4) with the projector augmented wave (PAW) method⁵⁵. The exchange–correlation functional was treated within the generalized gradient approximation (GGA) using the Perdew–Burke–Ernzerhof (PBE) functional⁵⁶. A plane-wave cutoff energy of 400 eV was employed. Geometry optimizations were performed until the residual forces on all atoms were less than 0.03 eV Å⁻¹. Brillouin-zone integrations were performed using a 3 × 3 × 3 Monkhorst-Pack k-point mesh for a 4 × 4 × 2 supercell. The self-consistent electronic convergence criterion was set to 10⁻⁴eV. The defect formation energies (E_f) of three types of F⁻ defects, F⁻ located inside the Gd cluster, on the surface of the Gd cluster, and far from the Gd cluster, were calculated according to the following equation:

$$E_f=E_{defect}-E_{perfect}$$ 1

Where E_perfect and E_defect represent the total energies of the Gd-doped supercell without and with F⁻ Frenkel defects, respectively. To elucidate the formation energy and migration tendency of F⁻ Frenkel defects, the F⁻ ions were systematically displaced from their original lattice sites by -0.2 Å, -0.4 Å, -0.6 Å, −0.8 Å, −1.0 Å, −1.5 Å, 0.2 Å, 0.4 Å, 0.6 Å, 0.8 Å , 1.0 Å, and 1.5 Å, respectively.

Gd L-edge XANES simulations were performed using the finite difference method for near-edge structure (FDMNES) code (version 2025.5.5), a DFT-based and fully relativistic program for ab initio X-ray spectroscopy calculations^57,58. To identify the formation of Gd-rich clusters in BaFCl, structural models containing different amounts of Gd³⁺ ions were constructed and relaxed using DFT optimization to minimize manual bias prior to FDMNES simulations. Preliminary calculations of the Gd L_2,3-edge were carried out in the multiple scattering mode (Green’s formalism) under the muffin-tin approximation using the energy-dependent Hedin–Lundqvist exchange potential to rapidly identify the structural configurations consistent with the experimental spectra. Subsequently, accurate simulations were conducted using the finite difference method with a self-consistent field cluster radius of 6 Å. The cluster radius (R) was progressively increased until spectral convergence was achieved, and in this case, R = 6.5 Å was found to be sufficient. To ensure optimal agreement with experimental results, convolution parameters were further refined by comparing the simulated and measured spectra.

Energy migration from traps to Gd³⁺ activators was modeled using Monte Carlo simulations (version 3.12.4)^30,59. Specifically, the spatial distribution of traps and activators was constructed based on the EXAFS-derived structural model (Fig. S17a). In the clustered configuration, the actual Gd³⁺ doping concentration was set to 2%, and the lattice parameters of BaFCl were taken as a = b = 4.418 Å, c = 7.281 Å, and α = β = 90°. A total of 40 clusters were generated, each modeled as a 1 nm-diameter sphere with Gd³⁺ ions randomly distributed inside, and randomly positioned within a 10 nm cubic crystal. Trap states were also randomly distributed within the clusters. To compare the differences in energy transfer between clustered and uniform configurations, a non-clustered model was constructed in which both traps and Gd³⁺ ions were randomly distributed throughout a 10 nm cubic crystal, while maintaining identical structural and compositional parameters. A Förster energy transfer framework with a Förster radius of ~1 nm was employed to describe the charge transfer dynamics between traps and Gd³⁺ activators. Each Monte Carlo simulation was repeated 1000 times to ensure statistically reliable results.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

L.L. and B.Y. initiated and designed the project. B.Y. performed materials synthesis. B.Y. and J.Z. conducted optical measurements and taken photographs. D.L. performed theoretical simulations. L.L. wrote and revised the manuscript. B.Y., D.L., R.D., J.Z., Y.W., S.X and L.L. contributed to the data analyses and discussion.

Peer review

Peer review information

Nature Communications thanks Dongpeng Yan, Ting Wang, 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. 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/s41467-026-68799-1.