A straightforward synthesis method is essential for achieving LPL. In the elevated temperature solid state method (1000 ~ 1500℃) for afterglow phosphors40, the high-temperature melting process towards OLPL materials heavily depends on the melting/boiling point similarity of each component to prevent bond breakage and reorientation41. Conventional solution chemistry methods have been employed for short-lived afterglow of organic-inorganic halides and crystalline/polymeric organic materials22–23,42. In this work, single crystals of CsCdCl3 can be grown using a modified hydrothermal reaction (details in Methods, Fig. 1f), and its crystal lattice adapts a space group P63/mmc (CCDC No. 2313854). The 3D asymmetric unit, as shown in Fig. 1a and Supplementary Fig. 1, is constructed with [CdCl6]4− octahedrons. Two of these share a triangular face to form [Cd2Cl9]5− in C3v symmetry, which then connected with six additional [CdCl6]4− octahedra to achieve corner-shared D3d symmetry. This unique packing arrangement offers numerous coordination sites for diverse halides and divalent metal cations, allowing for the arbitrarily anchoring of Br− or Sn2+ ions at the Cl− or Cd2+ ion sites, potentially leading to distinct optical properties. Upon alternation by Br− or Sn2+ ions, the powdered X-ray diffraction (PXRD) patterns of CsCdCl3:x%Br and CsCdCl3:x%Sn closely agree with the pattern in the PDF#18–0337, confirming the single-phase purity of the synthesized Br− or Sn2+-doped CsCdCl3 (Fig. 1, b to c and Supplementary Figs. 2 and 3). This leads to the expansion of the host lattice, manifested by the shifting of Bragg positions at [104] and [110] to lower angles (Supplementary Figs. 2 and 3). The ion radius values of Br− (r = 1.96 Å) and Sn2+ (r = 1.02 Å, CN = 6) are larger than those of Cl− (r = 1.81 Å) and Cd2+ (r = 0.95 Å, CN = 6), respectively, contributing to the observed lattice expansion. X-ray photoelectron spectroscopy (XPS) profiles describe Br−- and Sn2+-doped samples, with the characteristic peaks of Br− 3d and Sn ion 3d3/2 and 3d5/2 becoming more pronounced with increasing guest-doped concentration (Supplementary Figs. 4 and 5). Particularly, the peaks centered at 3d3/2 = 496.02 eV and 3d5/2 = 487.02 eV correspond to Sn2 + 43,44, suggesting that a tiny amount of Sn2+ doping can preserve its stability (Supplementary Fig. 5b, 5d).
Photophysical Properties
The optical characteristics of CsCdCl3 single crystals were initially investigated. As depicted in Supplementary Fig. 8a to 8c, the optimal excitation wavelength for the photoluminescence excitation (PLE) center of pure CsCdCl3 is 254 nm, inducing a broad emission peak at 595 nm with a full width at half-maximum (FWHM) of 88 nm in both prompt and delayed spectra (collected after 1 ms of excitation). CsCdCl3 displays robust excitonic absorption, aligning well with the PLE spectrum (Supplementary Figs. 7a, 8c). Given the substantial Stokes shift (341 nm) and wide FWHM, the observed orange emission is attributed to the self-trapping excitons (STEs) emission, consistent with the prior report46. Nevertheless, the low emission intensity (PLQY∼25.47%) significantly restricts its applicability (Fig. 2i, Supplementary Fig. 30a).
One strategy to modulate PL properties is introducing different halide cations with tunable band gaps, as observed in lead-based perovskites to achieve nanosecond luminescent lifetimes47. Doping a series of Br− ions produces noticeable changes in the corresponding optical spectra. As shown in Fig. 2a and Supplementary Fig. 8a, under 254 nm excitation, CsCdCl3:x%Br (x = 0.2–15) crystals exhibit a blue-shifted and progressively stronger broad emission peak at 482 nm compared to the pristine CsCdCl3 crystals. The delayed spectra reveal that intensities peaking at 482 and 595 nm are both enhanced with increasing Br− doping concentration (Fig. 2a, Supplementary Fig. 8b), resulting in a merging into a single large peak spanning from 350 to 800 nm, akin to the behavior observed in the prompt spectra. CsCdCl3 possesses both C3v and D3d symmetries48, with the a1→e transition allowed in C3v symmetry and the a1g→eg transition in D3d symmetry undergoing an S-T-splitting route49. The radiative transition in both symmetries originates from the triplet exciton, with the energy gap of D3d tending to be larger than that of C3v49. Previous studies have indicated that D3d symmetry's PL is in the UV region at low temperatures due to constrained molecular vibrations accelerating the S-T splitting process49,50. However, CsCdCl3:x%Br (x = 0.2–15) crystals show robust emission centered at 482 nm without the need for low temperatures. To comprehend this behavior, we analyze the structure-luminescence relationship: replacing Cl− ions with Br− ions distorts the [CdCl6]4− octahedron into [CdCl6 − nBrn]4− due to distinct Cd–Cl (2.66 Å) and Cd–Br (2.71 Å) bond lengths51. The emission peak at 482 nm, with a large Stokes shift (228 nm) and wide FWHM, is attributed to STEs, consistent with previous reports52–54. All the PLE spectra of CsCdCl3:x%Br show gradually red-shifting and broadening peaks beyond 300 nm, aligning well with the absorption spectrum (Fig. 2a, Supplementary Figs. 7a, 8c to 8d). This further suggests that transitions involving [e + a1] →a1 and [e + a1] →e, a1 in C3v, as well as [eu + a2u] →a1g in D3d are all activated. Therefore, CsCdCl3:x%Br samples exhibit a significantly enhanced PLQY up to 84.47% without relying on rare-earth metals (Fig. 2i, Supplementary Fig. 30), signifying a resource-saving approach for high-efficiency luminescence.
To understand the PL mechanism, temperature-dependent PL spectra for the representative of CsCdCl3:0.8%Br and CsCdCl3:10%Br were conducted. As illustrated in Supplementary Fig. 9a, the PL peak (band 3) emerges at low temperatures, followed by band 2 as temperature increases to room level, and then band 3 becomes more prominent at higher temperatures. Obviously, because of the low doping concentration of Br− ions, both D3d and C3v exist in [CdCl6]4− and [CdCl6 − nBrn]4− forms, with band 3 assigned to pure [CdCl6]4− in D3d symmetry at low temperature49, and band 2 with broad emission corresponding to the Br-doped of [CdCl6 − nBrn]4− in both D3d and C3v symmetry, while band 1 represents the undoped [CdCl6]4− unit in C3v symmetry52,53. For the delayed spectra (Supplementary Fig. 9b), bands 3 and 2 have maintained long-lived photoemission owing to the triplet exciton arising from D3d and C3v symmetry to cause phosphorescence. The increased FWHM by photon-phonon coupling in a specific temperature range of 217 to 277 K (Supplementary Fig. 9a, 9b), along with the large Stokes-shift, provides direct evidence that band 2 emission is associated with STEs51,54–55. Upon 10% Br− ion doping, the prompt spectra show that band 3 blends into band 2 at low temperatures due to the transformation of [CdCl6]4− into [CdCl6 − nBrn]4− in D3d (Supplementary Fig. 9c), again confirming the band 3 originates from D3d symmetry. The corresponding luminescence mechanism is depicted in Fig. 5e. It is noteworthy that band 1 in both samples becomes stronger with increasing temperature up to 377 K (Supplementary Fig. 9), illustrating its anti-thermal quenching ability, which will be further discussed below. Intriguingly, both the prompt and delayed spectra of CsCdCl3:0.8%Br and 10%Br exhibit excellent temperature-dependent luminescent properties (Fig. 2e and 2f, Supplementary Fig. 10). Of that, CsCdCl3:0.8%Br displays remarkable color variation, ranging from blue to cyan, then across yellow-green and finally to orange-red, observable with naked eyes (Fig. 4d), in good agreement with CIE coordination (Fig. 4b, 4c). Such a wide range of full-color tunable luminescence and the anti-thermal quenching properties are still rare, particularly in state-of-the-art LPL materials (Supplementary Table 4).
Remarkably, CsCdCl3:x%Br samples exhibit substantial LPL with distinctive time-dependent afterglow alterations. As illustrated in Fig. 4a, the pristine CsCdCl3 host manifests an orange-red afterglow extinguishing promptly upon cessation of the 254 nm UV lamp. CsCdCl3:x%Br, conversely, hold a bright blue-green color when exposed to 254 nm UV light. Subsequent to excitation cessation, CsCdCl3:x%Br exhibit color-variable LPL from blue-green to orange-red, effectively covering the entire visible spectrum. Notably, unlike conventional time-dependent luminescence, herein the time-valve in the color change can be regulated based on Br− ion concentration. To elucidate this phenomenon, steady-state luminescent decay curves were monitored at emission centers of 482 and 595 nm. As shown in Fig. 2g, the afterglow intensity at 595 nm rapidly diminishes in the first 500 s, followed by a gradual decline extending up to 2000 s to discern from the background. The decay lifetime of 482 nm in the initial 60 s window is extended with increasing the Br− ion concentration to 10% (Fig. 2h). As observed in Supplementary Movie 1, the afterglow persists for 1800 s to naked eyes. Analysis of time-resolved PL mapping in Fig. 2b and Supplementary Fig. 11 indicate that the emission band at 595 nm remains consistently strong, while the band at 482 nm gradually intensifies with Br− ion concentration doping to 5%. The time-dependent afterglow spectrum exhibits that the emission bands at 595 and 482 nm initially merge into a broad acromion (Fig. 2c, Supplementary Fig. 12), with the intensity at 595 and 482 nm linked to both time evolution and Br− ion doping concentrations. Hence, the color-variable LPL can be modulated by different decay lifetimes of emissions at 482 nm and 595 nm, as well as Br− ion concentrations in CsCdCl3:x%Br.
Charge trap state analysis through TL measurements is an effective method for elucidating LPL. Firstly, CsCdCl3:x%Br show good thermal stability (Supplementary Fig. 6a). No TL signal is detected for the pristine CsCdCl3, implying the absence of LPL nature (Fig. 2d). For CsCdCl3:x%Br, four different cases arise: ⅰ) x = 0.2 ~ 0.5, the TL single peak center appears at 370 K; ⅱ) x = 0.8, three peaks of 366, 392 and 465 K are observed; ⅲ) x = 1 ~ 5, the TL peaks lie in 380 and 445 K; ⅳ) x = 10 ~ 15, the TL peaks occur at 371 and 355 K. The trap energy level can be estimated by Urbach’s empirical formula Etrap =Tm/500 (Tm is the temperature of TL peak)56. Accordingly, the aforementioned TL peaks are determined to be 0.74 (370 K), 0.73 (366 K), 0.78 (392 K), 0.93 (465 K), 0.76 (380 K), 0.89 (445 K), 0.74 (371 K), and 0.71 eV (355 K), respectively. These CsCdCl3:x%Br samples all possess shallow traps ranging from 0.67 to 0.76 eV, preferable for creating an ideal depth for LPL57. In addition, the trap depths in the range of 0.8–1.6 eV are categorized as deep traps, typically resulting in low LPL at room temperature due to the activation energy barrier58. However, as the temperature increases, the charge carriers in the deep traps overcome the activation energy barrier and migrate to the emission center, leading to an anti-thermal quenching property.
An alternative avenue to achieve benefits involves selecting appropriate metal cations as activators in metal halides. Consequently, we opt for the Sn2+ ion as the doped emissary for CsCdCl3 to elucidate the luminous properties. As illustrated in Fig. 3a and Supplementary Fig. 13b,13c, under the excitation of 254 nm UV light, a predominant emission peak at 595 nm is significantly enhanced in both prompt and delayed spectra with increasing Sn2+ ion dopant concentration from 1–10%. However, further doping results in a slightly reduced peak intensity. CsCdCl3:10%Sn displays the highest PLQY up to 65.71% (Fig. 3f, Supplementary Fig. 31). Subsequently, the PLQY experiences a marginal decline with a further increase of Sn2+ ion concentration, attributed to intensified Sn2+-Sn2+ dipole interactions causing nonradiative energy transition. The prompt and delayed PLE spectra exhibit a primary peak centered at 254 nm (Supplementary Fig. 13a, 13c), with an apparent shoulder peak at 282 nm becoming more pronounced as Sn2+ ion concentration increases, consistent well with the absorption spectra (Supplementary Fig. 7b).
As depicted in Fig. 3a and Supplementary Fig. 13e, h, an obvious emission peak at 565 nm is enhanced upon increasing the Sn2+ ions under the excitation wavelength of 282 nm. To further investigate this uncommon phenomenon, a series of excitation-wavelength dependent PL spectra and 2D excitation maps were performed for these CsCdCl3:x% Sn. As shown in Fig. 3i and Supplementary Fig. 14, 18, as the excitation wavelength red-shifts from 250 to 285 nm, the emission band center at 595 nm undergoes a significant blue shift to 565 nm with reduced intensity in both prompt and delayed spectra. Continuous shifting of the excitation to 310 nm results in a faint peak at 480 nm, which is negligible in the 2D excitation map (Fig. 3i, bottom, Supplementary Figs. 14 to 18, right). Notably, such a large contrast in the blue-shifted emission based on continuous red-shifted excitation at room temperature has not been reported in halide perovskites, even rarely among various current luminous materials (Supplementary Table 4).
To elucidate this intriguing phenomenon, we conducted temperature-dependent PL spectra for representative CsCdCl3:3%Sn and CsCdCl3:10%Sn. As shown in Supplementary Fig. 19, 20, under 254 nm irradiation at low temperatures, the initial observation is the emergence of prompt PL band ⅲ, faintly mirrored in the delayed spectrum, alongside the prompt PLE peak at 254 nm (Supplementary Fig. 21). These findings suggest that the band ⅲ originates from unaltered [CdCl6]4− in D3d symmetry. As the temperature rises to 177 K, a broad prompt PL band ⅱ becomes prominent, replacing the weakened band ⅲ, a change similarly in the delayed spectra, indicating their common origin in triplet excitons (Supplementary Figs. 19b, 20b). Considering Sn2+ ions with a 5s2 electron configuration and 1P1/3P0,1,2 energy levels59. transitions such as 1S0 →3P0 and 1S0 →3P2 are deemed forbidden, while allowed transitions include 1S0→3P1 and 1S0→1P1 due to spin-orbit coupling60,61. The broad emission band ii is primarily attributed to the 3P1 →1S0 transition of Sn2+ ions situated in D3d [SnCl6]4− symmetry, partially involving the breaking of the forbidden transition 3P2→1S0 at low temperatures62. Upon further temperature elevation, the emission band ⅰ gradually intensifies, with its FWHM increasing due to photon-phonon coupling in both prompt and delayed spectra (Supplementary Figs. 19, 20). Given the large Stokes shift is similar to the pristine CsCdCl3, band ⅰ can also be identified as STEs originating from Cd2+ ions in conjunction with the distortion of [SnCdCl9]5− moieties in C3v symmetry.
To further substantiate this hypothesis, we adjusted the excitation wavelength to 282 nm. As illustrated in Supplementary Figs. 22, 23, the PL band ⅲ diminishes, while band ⅱ remains robust at low temperature, affirming their origin from [CdCl6]4− and [SnCl6]4− in D3d symmetry, respectively. With increasing temperature, a newly emerging band ⅳ with an emission center at 565 nm gains strength progressively. This observation is influenced by two main factors: a) Lower excitation energy effectively populates the exciton to the lower-energy 3P1 state of the Sn2+ ion in distorted [SnCdCl9]5−, where the thermodynamically favored 3P1 exciton serves as a trapped state, enhancing anti-thermal quenching ability. 60–61,63 b) The wide FWHM and large Stokes shift crucially support its STEs64. Further, a series of excitation-dependent PL experiments were conducted on CsCdCl3:3%Sn and CsCdCl3:10%Sn at 97 K. As shown in Fig. 3h, Supplementary Figs. S24a, 25a, varying the excitation wavelength from 250 to 300 nm results in a red-shift of the prompt emission center from 440 to 565 nm, corresponding well with the above-mentioned band ⅲ, band ⅱ and band ⅳ. In the delayed spectra (Supplementary Figs. 24b, 25b), the emission band at 595 nm (band ⅰ) decreases, and the broad emission band centered at 525 nm (band ⅱ) intensifies, confirming that optical properties are influenced by the geometric symmetry and energy states of the metal centers63–65. The corresponding luminescence mechanism is depicted in Fig. 5f. The distinctive feature as Janus-type luminescence – forward excitation-dependence at low temperature and the reverse excitation-dependence at room temperature – has not been reported previously (Supplementary Table 4), holding promise for potential applications in information safety and temperature recognition. After 46 days of storage at room temperature, all samples exhibit no significant spectral changes (Supplementary Figs. 28, 29), underscoring the excellent optical stability of CsCdCl3:x% Sn.
Remarkably, all CsCdCl3:x%Sn samples emit orange-red LPL after the cessation of 254 nm UV light (Fig. 4e and Supplementary Movie 1). The steady-state luminescent decay of the emission bands at 595 nm and 565 nm were investigated. Following the termination of 254 nm irradiation, the intensity of the CsCdCl3:3%Sn rapidly decreases in an initial 500 s and persists for up to 2000 s (Fig. 3b). For Fig. 3c, the intensity of the CsCdCl3:10%Sn at 565 nm also decays quickly in the first 400 s, and gradually slows down until reaching 1000 seconds. The characteristics of LPL at 595 nm and 565 nm are demonstrated through time-resolved PL mapping (Fig. 3d, 3e, Supplementary Fig. 26) and the time-dependent afterglow emission spectrum (Supplementary Fig. 27). Moreover, CsCdCl3:x%Sn exhibit good thermal stability (Figure S6 b). Focusing on the TL spectra (Fig. 3g), CsCdCl3:x%Sn exhibit distinct peak centers and trap energy levels (estimated by Etrap = Tm/500) at different x values: x = 1 (355K→0.71eV), 3 (355K→0.71 eV, 395K→0.79 eV), 5 (363K→0.73 eV, 434K→0.87 eV), 10 (363K→0.73 eV, 374K→0.75 eV and 452K→0.90 eV), and 15 (342K→0.68 eV, 436K→0.87 eV), respectively. Given this, it is no surprise that trap states play a crucial role in saving excitons from the excited state, evoking the LPL of these CsCdCl3:x%Sn by releasing excitons from shallow traps and thermally breaking the energy barrier in deep traps to the ground state.
Density Functional Theory (DFT)
To gain deeper insights into the electronic structure and luminescence mechanisms of Br-doped and Sn-doped CsCdCl3, we conducted band structure and total density of states (DOS) calculations. The pristine CsCdCl3 exhibits a direct bandgap, whereas the Br-doped and Sn-doped models exhibit indirect bandgaps (Fig. S33 to S34), thereby mitigating hole-electron recombination and extending exciton lifetimes66–67. In Fig. 5a and Fig. S36 to S40, the projected DOS (PDOS) of D3d-Bri-C3v (i = 1,2,3) and C3v-Brj-C3v (j = 4,5,6) models reveal that the valence band (VB) is mainly composed of Cl 3p and Br 4p orbitals, while the conduction band (CB) consists of Cd 5s/5p, Br 4p, and Cl 3p orbitals. In contrast, Cs orbitals play a negligible role in the band structures of these models. The charge density maps illustrate that the valance band maximum (VBM) is localized at Br− and Cl− ions in [CdCl6 − nBrn]4− moiety of both D3d and C3v symmetry, while the conduction band maximum (CBM) is predominantly formed from Cd2+ (Fig. 5c, Supplementary Figs. 36b to 40b). The small discrepancy in bandgaps between the D3d-Bri-C3v (i = 1,2,3) and C3v-Brj-C3v (j = 4,5,6) models, approximately 12.2 meV, has negligible impact on the PL of [CdCl6 − nBrn]4− in both D3d and C3v symmetries (Supplementary Fig. 33g). This reaffirms that the broad emission band at 482 nm (band 2) in Br-doped CsCdCl3 originates from the combined emission center of [CdCl6 − nBrn]4− in both D3d and C3v symmetries (Fig. 2a, 5e, Supplementary Fig. 9).
For Sn-doped CsCdCl3 (Fig. 5B, Fig. S41 to S43), PDOS shows that VBs consist of Cl 3p and Sn 5s orbitals, while the CBs comprise Cl 3p, Sn 5p and Cd 5s/5p orbitals in D3d-Sni (i = 1,4) and C3v-Snj (j = 2,3) models. The bandgap of D3d-Sni (i = 1,4) models is larger than C3v-Snj (j = 2,3) models, around 30.4–95.7 meV (Supplementary Fig. 34g), impacting the PL characteristics of [SnCl6]4− in both D3d and C3v symmetries due to exceeding the thermal energy of 26 meV68. This observation aligns with the experimental trend of Sn-doped materials exhibiting a reduced bandgap (Supplementary Fig. 7d). Additionally, these results underscore that the luminescence center of Sn2+ ions in C3v [SnCdCl9]5− symmetry requires less matching excitation energy, resulting in the reverse excitation-dependent behavior observed in experiment. From Fig. 5d and Supplementary Figs. 41b to 43b, the charge density maps highlight Sn2+ ion with unique 5s-related energy levels significantly influencing the VBM and contributing to multimode luminescence. Interestingly, when Sn2+ ion doping at C3v [SnCdCl9]5− symmetry (Supplementary Figs. 41b to 42b), the charge density of CBM is primarily located at the Cd2+ ion of C3v symmetry, further supporting the hypothesis that band ⅰ stems from C3v symmetry (Fig. 3a, 5f, Supplementary Figs. 19, 20).
High-level anti-counterfeiting using multifunctional LPL
Programming advanced anti-counterfeiting technology is of considerable practical importance, by leveraging the LPL characteristics of all-inorganic halide perovskites to demonstrate their distinctive spatial-time-resolved, and time-logical color-variable afterglow properties. As depicted in Supplementary Fig. 44, no perceptible alterations are observed under ambient lighting conditions. However, upon exposure to 254 nm UV irradiation (Fig. 6a), the combination of chaotic orange-red and cyan colors could potentially convey misleading messages such as “I ❤ BNU 8888” (BNU: Beijing Normal University) and two other false messages when individually viewed in the orange-red or cyan channel. After ceasing the UV lamp for 1s (Fig. 6b), the subsequent error information is interpreted as "I ❤ BNU 2083", accompanied by two additional error messages, resembling a three-dimensional (3D) encryption featuring spatial-time-dual-resolved patterns. The cyan “❤”transfers to milk white, and some local areas exhibit color changes in 3 s (Fig. 6c). Despite the message still reading as “I ❤ BNU 2083”, it has evolved into the 4D anti-counterfeiting category due to its variable color factor. Ultimately, a cohesive orange-yellow color palette effectively conveys the intended message of “I ❤ BNU 2023” (Fig. 6d and Supplementary Movie 2). The overall process can be regarded as a 5D anti-counterfeiting technology, which is anticipated to surpass conventional afterglow materials due to its additional time-gated color change facilitated by Br-doping engineering.
Furthermore, proof-of-concept experiments were conducted by filling a series of as-synthesized perovskites into the QR code groove. As depicted in Fig. 6e, the security model operates under a mechanism wherein specific attributes are assigned to the QR code of each region: emission loss (marked in red), QR code with afterglow duration every 0.5 s (marked in blue), and modification of the afterglow color (marked in green). These attributes correspond to the binary shifts of one bit (+ 1) in the respective regions. Upon excitation by the 254 nm UV lamp, the QR code displays chaotic orange-red and cyan hues, representing the binary exchange algorithm “000”. At a delay time of 0.5 s, region C undergoes the first transition from cyanogen to cyanogen yellow, introducing binary ciphers for “011”, “110” and “101”. One second later, the emission from region A is nearly extinguished, while region C transforms to orange-yellow, resulting in adjustments of the binary ciphers to “11010”, “10110” and “10101”. After 18 s, the new binary ciphers take form, generating security codes “1161, 1764, 1508, 1481, 1175 and 1179” through binary and decimal conversion (Fig. 6h). The final lock code is determined as “1508”. It is noteworthy that the above binary ciphers are applicable only when i = j = n at the respective time nodes. If the binary digits of all time nodes could be exchanged (Fig. 6f), it would result in an information Big-Bang, achieving the maximal information loading capacity (Fig. 6h). Therefore, these perovskites are anticipated to be highly effective for advanced high-security anti-counterfeiting due to leveraging the advantages of time-sensitive color and spatial-time four-resolved functionality.