Tough transparent glass ceramics for multi-mode programmable dynamic tunable persistent luminescence via phase engineering

Abstract

Realizing multi-mode programmable dynamic tunable persistent luminescence within a single solid is promising for multi-dimensional information storage and encryption applications. However, coupling luminescent centers with differentiated defect states remains a challenge. Here we report transparent glass ceramics that exhibit photo/thermally dynamic tunable afterglow. A lithium-ion doping-assisted phase separation principle is developed to control the precipitation of defective non-stoichiometric nanocrystals (Zn_1.7SiO₄: Li) in glass matrix, constructing a biphasic microenvironment with differentiated defect states. The persistent luminescence color can be manipulated by simultaneously engineering the distributions of Mn²⁺ activators in the amorphous glass matrix and nanocrystals, and the mechanism is discussed. The developed transparent composites exhibit excellent hardness of up to 10 gigapascal and high thermal stability, which is applicable for harsh conditions. This work pioneers a strategy for modulating dynamic afterglow in a single solid and inspires more potential applications in muti-dimensional information storage and encryption.

Realizing tunable persistent luminescence in a single solid is promising for applications, but challenging. Here, the authors report tough transparent glass ceramics for muti-mode programmable dynamic tunable persistent luminescence via phase engineering.

Introduction

Persistent luminescence (PersL) is an intriguing photophysical phenomenon with the gradual charge recombination after photoexcitation¹. With the ability to storage energy for periods of time, applications of PersL materials have spread from civil cases, i.e., emergency lights, decoration displays, etc., to advanced scientific applications including information storage, optical encryption, medical delivery identification and bioimaging^2–8. However, conventional PersL materials, encompassing inorganic, organic, and hybrid materials, generally exhibit static monochromatic emission, limiting their ability to address the multifunctional requirements in advanced optoelectronic devices.

Currently, great efforts have been made to develop dynamic PersL materials, which can flexibly respond to diverse external stimuli such as excitation light, temperature, humidity, electric field or mechanical stress, and translate these responses into detectable afterglow outputs with changeable properties (intensity, color or lifetime)^9–12. Typically, the host-guest doping strategy and molecular engineering have been proved as two effective methods for dynamic tunable PersL in organic materials^13–15. However, these materials inevitably suffer from both unstable triplet excitons that restrict phosphorescence to seconds and poor environmental tolerance (e.g., heat/light) that results in degraded luminescence efficiency during long-term use. In another case, doping ions to engineer trap states and modifying energy transfer process between multi-emission centers are employed to realize dynamic tunable PersL in inorganic materials and hybrid materials^16–20. For organic-inorganic metal halides perovskites, low thermal stability still persists as a critical limitation, causing a narrow operation temperature range^17,18. Besides, most reported micro-sized PersL phosphors exhibit unavoidable opacity and strong self-absorption, resulting in low afterglow brightness. What’s worse, for practical printing applications, they must be dispersed into solutions or polymer matrices, which could degrade their optical performance by inducing undesirable surface states^16,19,20. Therefore, achieving stable dynamic tunable PersL in a monolithic system with synchronous responses to multiple external stimuli remains a key scientific challenge.

Transparent glass, with excellent physicochemical stability (including corrosion/heat resistance) and mechanical property, has long been established as an exceptional medium for three-dimensional (3D) information storage and anti-counterfeiting applications^21–23. Especially, controllable nanocrystal precipitation and flexible ion doping in the rigid glass matrix provide feasible pathways to tune the optical properties and functionalities within a monolithic glass/glass ceramic^24,25. Engineering trap states among different emission centers in the glassy phase or crystal phase to modify the trapping/de-trapping properties of photo-generated carriers promises to realize the dynamic color-tunable PersL in these transparent solids, potentially enabling applications in high-dimensional information storage and encryption²⁶. However, current crystal-engineered PersL glass ceramic usually yields static PersL color due to the similar environments and homogeneous defect states among emission centers in crystal domains. Consequently, no reliable spatiotemporally dynamic tunable PersL has been explored in a single-component glass-ceramic system.

Herein, we report a type of transparent glass-ceramic characterized by the precipitation of non-stoichiometric nanocrystals. Using a phase-separation strategy, Mn²⁺ activators are spatially confined within the glassy phase and the non-stoichiometric nanocrystalline phase, which grants them distinct coordination environments and defect-state properties to enable the multi-mode programmable dynamic tunable PersL within a single glass-ceramic solid. The obtained PersL color can be flexibly manipulated by altering the excitation wavelengths and ambient temperatures. Furthermore, the transparent composites show excellent mechanical properties and thermal stability, enabling great potential for applications under harsh conditions.

Results

Li⁺-doping assisted nonstoichiometric phase transformation

The phase-separation has been demonstrated as an effective principle for tuning optical properties of materials^27,28. Here, a type of transparent glass ceramic with non-stoichiometric crystals in the glass matrix was created based on a phase separation method, and distinct crystal-field and defect-state environments were generated for activator ions through ion doping or thermal process adjustment. The flexible network structure of non-stoichiometric crystals offers far greater flexibility in defect-state design, thus having more potential for designing dynamic tunable PersL than their rigid stoichiometric counterparts. The detailed procedure is schematically displayed in Fig. 1. Through heat-treatment-induced phase separation, defective non-stoichiometric nanocrystals are deliberately designed into the glass matrix, with activators being selectively incorporated into the biphasic system featuring differentiated defect states. Therefore, activators at different doping sites will exhibit distinct carrier dynamics (trapping and de-trapping), thereby enabling dynamic tunable PersL. Where Mn²⁺ ions were chosen as the activators to modulate the luminescence properties through controlled phase separation and defect engineering, whose emissions are highly dependent on the crystal field strength^29,30.

Fig. 1. Schematic of designing biphasic heterogeneous defect states through non-stoichiometric crystallization in glass matrix.

(PLE: photoluminescence excitation, ES/IS: extended states and localized states in glassy phase^61,62, CB: conduction band in crystal phase, ɛ₁/ɛ₂: trap depth).

To validate our proposed scheme, exploratory experiments based on Mn²⁺-doped (0.5 mol%) zinc silicate glass material were conducted. Based on the radius similarity principle (Li⁺: 0.59 Å, Zn²⁺: 0.60 Å, coordination number CN = 4; or Li⁺: 0.76 Å, Zn²⁺: 0.74 Å, CN = 6) and inspired by previous successful doping precedents^31,32, varying amounts of Li₂CO₃ dopants (0–4 mol%) were introduced into the system to induce the lattice distortion that modifies the Zn²⁺ coordination symmetry, thereby facilitating the formation of non-stoichiometric nanocrystals by disrupting the bond periodicity of Zn–O. All glass ceramics were prepared by using a melt quenching method followed by a secondary heat treatment (700 °C, 2 h) at a reducing atmosphere, which were then tagged as KLI0-KLI4 according to Li₂CO₃ doping content.

First, X-ray diffraction (XRD) measurements and high-resolution transmission electron microscopy (HRTEM) characterizations were carried out to validate the effectiveness of Li⁺ doping in promoting the in-situ precipitation of non-stoichiometric crystals in the glass matrix. With increasing Li⁺ doping, the evolution of the diffraction peaks (Fig. 2a) and measured lattice fringe spacings (Supplementary Fig. 1) demonstrate a precipitated crystalline phase change from β-Zn₂SiO₄ (JCPDS No. 14-0653) to Zn_1.7SiO₄ (JCPDS No. 24-1466), accompanied by a clear grain morphology transformation from regular spheres to irregular spheres in nano-scale (Supplementary Fig. 2). As a result, nanocrystalline structure is anticipated to evolve from an orthorhombic framework where zinc atoms synchronously occupy both four-coordinated and six-coordinated sites, to a new configuration where Zn atoms just occupy the tetrahedral sites^33,34. This results in a change in the coordination environment of Mn²⁺ from an octahedral to a tetrahedral geometry as shown in Fig. 2b. Further, to investigate the role of the introduced Li⁺ ions, XRD tests were then performed on Mn-free samples (Supplementary Fig. 3b) to eliminate the lattice expansion effect caused by the ionic radius difference from Mn²⁺ doping (Supplementary Fig. 3a, Mn²⁺: 0.66 Å vs Zn²⁺: 0.60 Å, CN = 4). Obviously, the diffraction peaks shift to lower angles with increasing Li⁺ doping. This lattice expansion behavior suggests the increased interstitial occupation at high Li⁺ doping concentration, which matches well with the blue-shifted photoluminescence (PL) emissions (Supplementary Fig. 5b)^35,36, confirming that Li⁺ ions do contribute directly to this non-stoichiometric transformation by entering the crystal lattice. Accordingly, the chemical formula Zn_1.7SiO₄: Li is used in the following discussions. Consequently, the negatively charged zinc vacancies (V''_Zn) or lithium substitutional defects (Li'_Zn) can be charge-compensated by positively charged oxygen vacancies (V_ö) or interstitial Li⁺ (Li_i) to ensure the overall lattice charge neutrality as shown in Fig. 2b. Moreover, the enhanced diffraction peak intensity (Fig. 2a) and larger grain radius (Supplementary Fig. 2) indicate the additional influence of Li⁺ ions in improving crystallinity. The promoted crystallization kinetics can be confirmed by differential scanning calorimetry (DSC) measurements (Supplementary Fig. 4). Clearly, higher Li⁺ doping not only lowers the glass transition temperature (T_g: 628 → 583 °C) and crystallization temperature (T_c: 715 → 681 °C) but also increases the crystallization enthalpy (ΔH_c: 1.37 → 7.73 mJ/mg). This is reasonable, as Li⁺ doping can weaken the glass network by introducing non-bridging oxygens, thereby facilitating the flexible atomic arrangement during heat treatment^37,38. These results presented above lead us to the conclusion that Li⁺ doping concentration regulation promotes the expected non-stoichiometric phase crystallization by expanding the glass network channel to promote ion migration and entering the crystal lattice to create distortions.

Fig. 2. Dynamic tunable PersL in Zn_1.7SiO₄: Li precipitated glass ceramic.

a XRD pattern of glass-ceramics with varying Li₂CO₃ content. b Zn coordination environment change (from CN = 6 → 4) before and after non-stoichiometric phase transformation, and the schematic diagram describes the defect generation through Li⁺ doping. c Tunable PersL spectra (535-640 nm) dependent on Li⁺ doping amount by 254 nm excitation, and the inset shows the recorded afterglow pictures. d Normalized excitation-wavelength-dependent afterglow spectra of KLI3. e Images of the charging-temperature-dependent afterglow recorded at a fixed delay of 2 s after 365 nm excitation for 5 min. f CIE coordinate change for the charging-temperature-dependent PersL spectra (20 s exposure). g Time-dependent afterglow recorded immediately after 365 nm excitation for 3 s. ‘Source data are provided as a Source Data file.

Interestingly, the phase transformation modifies the coordination environments of Mn²⁺, leading to Li⁺-doping dependent tunable photoluminescence (PL) (Supplementary Fig. 5a) and PersL (Fig. 2c). After 254 nm excitation, the room temperature PersL spectra exhibit tunable emissions peaking across a broad wavelength range of 535–640 nm. Deconvolution analyses of PersL spectra (Supplementary Fig. 6) have resolved three distinct emission bands centered at 534–537 nm, 580–585 nm, and 593–647 nm, assigned to Mn²⁺ in tetrahedral sites in Zn_1.7SiO₄: Li, octahedral sites in β-Zn₂SiO₄ and in glass matrix, respectively.

Dynamic tunable PersL properties in KLI3

As anticipated, obvious dynamic tunable PersL is realized in this Mn²⁺-activated defective Zn_1.7SiO₄: Li nanocrystal-precipitated glass-ceramic, confirming the effectiveness of the non-stoichiometric strategy based on phase engineering. The programmable trapping of charge carriers at the heterogeneous defect states within the dual-phase structure (glass and Zn_1.7SiO₄: Li nanocrystals) can be achieved by manipulating either the excitation wavelengths to alter the carrier transition pathways or the charging temperatures to control the critical depth for stable trap filling. By which, the carriers’ de-trapping properties (population and duration) after excitation can be modified. This in turn alters the intensity ratio of the two emission bands, thereby enabling continuous tuning of the afterglow chromaticity. Concretely, as shown in Fig. 2d, a reversible PersL color from green to orange was detected by changing the excitation wavelength from 254 nm to 365 nm. And, as the charging temperature increased, afterglow color pre-excited by 365 nm for 5 min shifted from orange to green (Fig. 2e), with the relevant CIE color coordinates changing (Fig. 2f) from (0.483, 0.489) to (0.410, 0.551). The corresponding normalized integrated temperature-dependent PersL spectra (Supplementary Fig. 7) revealed an enhanced dominance in the green emission band at elevated temperatures. Furthermore, the afterglow color is also time-dependent. Across tested temperatures (298 K, 373 K and 423 K), the sample exhibited a progressive afterglow color shift to yellow/green within the first 5 s of decay time (Fig. 2g, Supplementary Movies 1–3) after being excited by 365 nm for 3 s. These observations demonstrate that the heterogeneous defect states are continuously distributed in a wide energy level (trap depths) range to enable the interesting spatio-temporal dynamic tunable PersL. Here, it should be noted that during the temperature-dependent measurement, the temperature was fixed for each testing during and after excitation. No higher temperature was necessary to induce PersL. There was also no additional light irradiation to induce de-trapping after cessation of excitation. Consequently, the PersL in the current work is characterized by persistent luminescence but not thermally stimulated luminescence or photostimulated luminescence.

Compared with reported dynamic PersL materials (Table 1), this photo/thermal programmable glass ceramic demonstrates exceptional performance metrics including^{10,11,16–18,39–42}: (i) an elevated maximum operating temperature of 503 K (higher than listed reported systems, ≤423 K), (ii) an expansive operating temperature range of ~200 K (more suitable for practical applications), and (iii) a prolonged time-dependent decay detectable for over 5 s (a fivefold improvement over most existing materials).

Table 1.

Comparison of dynamic PersL properties of different reported materials

Material type	TimD-PersL	TemD-PersL	ExD-PersL	Ref.
Polymers	~1 s	×	254–365 nm	39
Polymers	×	×	254–365 nm	10
Perovskite	×	200–300 K	×	16
Perovskite	~30 s	97–377 K	×	11
Hybrid Metal Halides	~3 s	270–333 K	×	18
Hybrid Metal Halides	~1 s	×	×	17
Carbon dots	~1 s	×	×	40
Carbon dots	2 s	193–423 K	×	41
Supramolecular Glass	~1 s	×	295–395 nm	42
Glass ceramics	> 5 s	298–503 K	254–365 nm	This work

TimD, Time-dependent; TemD, Temperature-dependent; ExD, Excitation-dependent.

Annealing temperature control

Then, controlled thermal treatment processes from 620 °C to 700 °C (samples tagged as GC620 to GC700) were performed on the amorphous precursor base glass (BG) of typical KLI3 glass ceramic with 3 mol% Li₂CO₃ doping according to the measured glass transition temperature (T_g = 586 °C, Supplementary Fig. 4), aiming at analyzing how thermal history to influence the phase composition and defect structures. All prepared glass ceramics are highly transparent, as confirmed by their optical images (Supplementary Fig. 8), which provide an excellent matrix for optical functionalities. Further, it should be noted that the glass/glass ceramic exhibits outstanding mechanical property, with hardness values (9.3–10.7 gigapascal, Supplementary Fig. 9) significantly higher than conventional transparent glass ceramics (Supplementary Table 1). This mechanical enhancement may originate from the incorporation of Al³⁺ into the silicate network as [AlO₄] tetrahedra, a process promoted by the presence of alkali metal cations in the glass matrix⁴³. The larger ionic radius of Al³⁺ (0.53 Å) compared to Si⁴⁺ (0.40 Å) results in a more compact glass network due to the expanded [AlO₄] units, equipping the material with an exceptional hardness value potential for applications in harsh conditions. All results cooperatively demonstrate the comprehensive performance advantages of this non-stoichiometric crystallized glass ceramics.

The consistent XRD peak positions for samples annealed at elevated temperatures confirm the stable precipitation of Zn_1.7SiO₄: Li non-stoichiometric nanocrystals with negligible β-Zn₂SiO₄ crystals due to the absence of typical diffraction peaks and the absence of the characteristic ~580 nm emissions (Fig. 3b, Supplementary Fig. 13)^44,45. Besides, the negligible intensity from the undoped glass ceramic (GC700-0Mn) in both PL and 2D-thermoluminescence (2D-TL) test (Supplementary Fig. 10a, b) further rules out the contribution of the intrinsic self-trapping exciton emission of the host to PersL. These results collectively attribute the dual-peak emissions to the combined contributions from Mn²⁺ in tetrahedral sites of the Zn_1.7SiO₄: Li crystals and Mn²⁺ in octahedral sites of the glass matrix. The confirmation of Mn ions’ presence in both non-stoichiometric nanocrystals and glassy domains according to the element mapping image of Mn in GC700 (Supplementary Fig. 11) provides direct evidence to support the above discussions. Moreover, Fourier transform infrared spectroscopy (FTIR, Supplementary Fig. 12) tests reveal the obviously enhanced vibrational modes at 575 cm⁻¹ and 617 cm⁻¹ assigned to [ZnO₄], and 457 cm⁻¹, 472 cm⁻¹ and 904 cm⁻¹ assigned to [SiO₄], indicating the intensified tetrahedral coordinated environment around Mn^{2+ 46,47}. This results in the crystallinity-dependent tunable PL by redistributing Mn²⁺ ions from the glassy phase to the tetrahedral sites in the crystal phase. Optical PL images (Fig. 3a, line 1) and spectra (Supplementary Fig. 13) under 254 nm and 365 nm excitation reveal the obvious chromaticity changes from red to green and the enhanced dominance in the green emission band. Here, a slight blue shift (<15 nm) in the green emission band and a more pronounced one (>40 nm) in the orange-red emission band can be detected in Gaussian deconvolution analyses for PL spectra (Supplementary Fig. 14). The former is due to the weakened crystal field caused by lattice expansion. While the latter can be ascribed to the inhibited Mn–Mn coupling interaction since crystallization dilutes the distribution of Mn²⁺ ions within the residual glassy phase, as illustrated in the diagram (Supplementary Fig. 15)^48,49.

Fig. 3. Tunable PersL and trap types modified by annealing temperatures.

a Optical images of annealing-temperature dependent PL and PersL after 365 nm (left) and 254 nm (right). b, c Annealing-temperature dependent PersL spectra after (b) 365 nm and c 254 nm excitation. d XPS spectra of Zn 2p of BG and different annealed glass ceramics. e Increased oxygen vacancy defects (g = 1.99) with increased annealing temperatures. f, g 3D-TL (excitation wavelength-temperature-emission intensity) heatmap of GC700 from 300–500 K, pre-excited by (f) 365 nm and g 254 nm for 5 min. ‘Source data are provided as a Source Data file.

The approximate peak positions between PersL spectra (Fig. 3b, c) and PL spectra (Supplementary Fig. 13) indicate that afterglow originates from a delayed fluorescence emission process. Though, the change in the relative intensity of dual emissions together with a slight peak shift gives rise to a color difference between PL and PersL (e.g., green PL color vs orange PersL color in 365 nm-excited GC700 glass ceramic, Fig. 3a). As described in diagram in Supplementary Fig. 16, such dual-peak relative intensity difference demonstrates that the relative number of carriers released from trap energy levels to luminescence centers in dual phases during the PersL process—limited by their distinct trap distributions—is not proportional to that in the PL process, which is governed by excitation efficiency and radiative transition probability. The peak shift between PersL and PL spectra indicates the unavoidable local microenvironment difference (crystal field strength and trap states) of Mn²⁺ ions even within a single phase⁵⁰. It’s reasonable that the obtained PersL spectra are dominated by carriers from those Mn²⁺ ions coupled with shallower trap energy levels due to their faster carrier release and recombination, which might have smaller energy gaps to generate a longer wavelength. In fact, the marked color differences between PL and PersL under fixed-excitation conditions confirm the heterogeneous defect distributions across the two phases and reveal their governing role in the afterglow behavior.

Mechanism investigations of dynamic tunable PersL

The excitation spectra analyses (Supplementary Fig. 17) of dual-emissions reveal that the operated excitation wavelengths (254 nm and 365 nm) separately activate two different kinds of charge transfer transitions: (1) the charge transfer band (CTB) of O²⁻-Mn²⁺ (appearing as a broad band below 330 nm), and (2) the d-d transitions of Mn²⁺ (manifested as a series of discrete bands beyond 330 nm)⁵¹. Comparatively, the remaining Mn²⁺ ions in the glassy phase undergo rapid excited-state energy transfer via phonon assistance, causing the characteristic discrete excitation peaks to merge into broad bands⁵². Due to different charge transition paths, all glass ceramics exhibit excitation-wavelength-dependent afterglow, displaying an orange-to-green color evolution by changing excitation wavelength from 365 nm to 254 nm. Carriers excited by high-energy photons (254 nm) can approach the conduction band directly and receive the rapid and saturated trap filling from shallow to deep levels. The green afterglow dominance in 254 nm-excited three-dimensional thermoluminescence (3D-TL, Fig. 3g) heatmap of GC700 at all temperatures indicates a higher trap concentration across the entire depth range in the crystal phase than that in the glass phase. However, charge carriers photoexcited by low-energy (365 nm) excitation in Fig. 3f tend to induce less efficient trap filling from shallow to deep levels, mediated by conduction band (CB) transport and quantum tunneling (QT) in crystal phase⁵³. The unavoidable continuous de-trapping of carriers from deep traps throughout the entire temperature range also complicates the identification of dominant luminescence centers coupled with deep traps. This leads to the fact that the contribution of green afterglow emission at high temperatures is not as pronounced as that shown in Supplementary Fig. 7.

X-ray photoelectron spectroscopy (XPS) and electron paramagnetic resonance (EPR) demonstrate the defect composition change under different annealing temperatures. In Fig. 3d, the Zn 2p in GC660 shows a positive peak shift of ~0.4 eV to a higher binding energy relative to BG or GC620, indicating that the negatively charged acceptor defects (e.g., V''_Zn, Li'_Zn) work as the primary defects to reduce the electron cloud shielding effect of the neighboring Zn atoms at the initial crystallinity enhancement^54,55. And the slightly increased oxygen vacancies (Fig. 3e, according to the EPR signal at g = 1.99) are difficult to balance the phenomenon. In contrast, a higher annealing temperature (700 °C) dramatically boosts the formation of oxygen vacancies via accelerated oxygen ion migration, particularly in a reducing atmosphere. The newly formed oxygen vacancies donate free electrons that strengthen the electron cloud shielding at Zn sites, resulting in a lower Zn 2p binding energy in GC700⁵⁶.

The trap depths modified by annealing temperatures are confirmed by comparatively investigating the 2D-TL spectra of the amorphous glass sample of BG (Supplementary Fig. 18) and crystallized samples GC660 and GC700 (Fig. 4a, b) excited by 365 nm. Regarding the TL curve, the peak position can be used to estimate the trap depth, and the integrated intensity reflects the trap density. According to Urbach’s empirical formula, the trap depth can be approximately estimated by^57–59: $E=T_m/500$. Where $T_m$ represents the Kelvin temperature corresponding to the TL peak, and E is the activation energy of the trap (in eV). All spectra can be well deconvoluted into two peaks: a shallow-trap-induced TL peak near 350 K (trap 1, ~0.7 eV) and a deep-trap-induced one above 400 K (trap 2, ~0.84 eV), confirming the coexistence of shallow and deep traps in both amorphous and crystallized samples. Compared to the BG and low-temperature heat-treated sample GC660, GC700 exhibits an increase in the deep-trap-induced TL peak area, with the peak shift to a higher temperature. This indicates that deep traps were generated in the newly-formed non-stoichiometric Zn_1.7SiO₄: Li crystalline phase rather than the amorphous glassy matrix. The higher proportion of deep traps enables GC700 to have a more efficient trap-filling at elevated temperatures, in which case more stably maintained carriers can be thermally released to luminescence centers, leading to a slower decrease in PersL intensity and the better PersL thermal stability in contrast to GC660, as shown in Fig. 4c, d line chart. It is worthwhile to note that the PL intensity of GC700 exhibits an anti-thermal quenching behavior near 325 K. This is due to more efficient thermally de-trapped charge carriers that compensate for the thermal-induced non-radiative losses⁶⁰. Finally, the remarkable thermal robustness of GC700 is highlighted by its preservation of 96% PL efficiency and over 110% PersL intensity at 100 °C (compared to that at the room temperature), coupled with excellent maintained luminescence stability over five heating-cooling cycles (Supplementary Fig. 19). Temperature-dependent decay curves of the dual emissions after 365 nm of GC700 are shown in Fig. 4e (298 K) and Fig. 4f (453 K). At 298 K, the two PersL emitters show a similar decay process, consisting of a first fast decay (lasts for about 3 s) and a second slow decay (lasts for about 35 s). However, at a higher temperature of 453 K, green PersL emissions (37.06 s) exhibit an obvious longer slow decay process than orange PersL emissions (23.64 s), which further demonstrates that more deep traps are generated in nanocrystalline phase rather than in the amorphous glass matrix and slow down the release rate of the immobilized charge carriers, thereby stabling the green PersL from tetrahedral coordinated Mn²⁺ at elevated temperatures. Therefore, a more obvious temperature-dependent PersL color change from orange to green was monitored in the high-temperature-annealed sample GC700 (Supplementary Fig. 20).

Fig. 4. The change of the trap depths controlled by annealing temperatures and the mechanism for dynamic tunable PersL.

a, b 2D-TL curves for (a) GC660 and b GC700 after 365 nm for 5 min. c, d Peak intensity change of the charging-temperature-dependent PL (bar graph) and PersL (line graph) for (c) GC660 and d GC700. e, f The decay curves within 300 s for GC700 at (e) 298 K and f 453 K after 365 nm for 5 min, all curves can be fitted into a biexponential function $y=C+A_1\exp \left(-x/{\tau }_1\right)+A_2\exp \left(-x/{\tau }_2\right)$, where the two lifetime components (${\tau }_1$, ${\tau }_2$) reflect the contributions of shallow traps and deep traps to the luminescence dynamics. g Mechanism diagram of the dynamic tunable PersL. ‘Source data are provided as a Source Data file.

To clarify the potential mechanism of the dynamic PersL process, a reasonable hypothesis has been proposed in Fig. 4g. Under UV excitation, charge carriers in the crystalline and glassy phases are pumped to high-energy levels near the conduction band (CB)/extended states (ES) via Mn²⁺-mediated d-d transitions (under 365 nm) and localized CTB absorption (under 254 nm), respectively. A portion of these carriers then transfers and relaxes to Mn^{2+ 4}T₁(G) level, making a radiative transition to the ⁶A₁(S) ground state to cause PL emissions. When shifting the excitation wavelength from 365 nm to 254 nm, the much stronger Mn-O CTB absorption in the crystalline phase compared to the glassy phase drives more efficient energy transfer to tetrahedrally coordinated Mn²⁺ ions, decisively enhancing the dominance of green emissions to enable the tunable PL behavior dependent on excitation wavelengths. Meantime, the other portion of charge carriers is captured by localized trap states while migrating through the conduction band and extended states. Under 254 nm excitation (State 1), abundant photogenerated carriers rapidly populate trap energy levels to approach saturation through CB/ES transport in both phases. After the cessation of excitation, sufficient carriers are released from shallow traps and return to Mn²⁺ luminescent centers, generating the afterglow predominant by green emissions due to the higher density of shallow traps in the crystalline phase. In contrast at room temperature, 365 nm excitation (State 2) with lower photon energy limits the number of carriers with sufficient energy to overcome the potential barrier for CB/ES transport. Instead, quantum tunneling to deep levels becomes the dominant trapping route. The resulting non-saturated filling of trap states significantly suppresses the afterglow contribution of carriers released from the lattice shallow traps at tetrahedral coordinated Mn²⁺, leading to an orange afterglow color. Furthermore, elevated charging temperatures (State 3 and State 4) provide sufficient thermal energy for carriers to overcome the potential barrier, promoting the filling of the trap energy levels via CB/ES transport. However, the inherent instability of shallow traps at high temperatures also enhances the probability of carriers being captured by the deep trap energy levels dominantly situated within the crystalline phase and facilitates the dominance of green emissions, ultimately leading to the charging-temperature-dependent tunable afterglow performance. In addition, with the prolonged decay time, afterglow emissions manifest as a gradual transition from shallow to deep traps. Since the lattice Mn²⁺ couples with deeper trap energy levels, it exhibits a longer afterglow duration, thus driving the temporal shift in the afterglow chromaticity.

Applications for information storage and dynamic encryption

Finally, the optical information storage and encryption applications of the glass ceramic can be developed by leveraging its excitation-dependent and temperature-dependent characteristics. The patterning design was readily achieved through a simple photomask process. As demonstrated in Fig. 5a, under 365 nm excitation, the system enabled multicolor information writing and reading from orange to green as the temperature increased from 298 K to 423 K. Additionally, in Fig. 5b, by rapidly switching the excitation wavelength (between 254 nm, 295 nm, 365 nm) within 3 s, reversible writing of different PersL color information from dark green to yellowish-green and then to orange could be realized. Importantly, unlike current materials, which typically require the incorporation of multiple luminescent components to achieve multicolor afterglow display, glass matrix offers potential for the real-time controllable multicolor afterglow owing to its excellent bulk excitation characteristics. As shown in Fig. 5c, distinct thermal fields were separately applied to the petal and stem regions with a soldering iron. The temperature gradient between areas induced differential PersL colors across the entire patterned structure. Meanwhile, when implementing spatially controlled excitation with a 254 nm and a 365 nm laser in Fig. 5d, the eagle-patterned area excited by 254 nm exhibited green afterglow, while the branch-patterned region displayed yellow afterglow after combined 254 nm (1 s) and 365 nm (1 s) excitation or orange afterglow after single 365 nm excitation. Finally, exploiting the intrinsic transparency of the glass matrix enabled the successful realization of 3D PersL patterning (Supplementary Fig. 21) through femtosecond laser direct writing, which demonstrates its unique potential for advanced 3D optical applications. Thus, our developed non-stoichiometric crystallization strategy based on phase separation in glass matrix opens a new avenue for the design of dynamic tunable PersL. The developed material overcomes the limitations of multi-component material, enabling the integration of multicolor PersL within a monolithic system by simply modifying the excitation conditions. Combined with the intrinsic transparency of glass, such material is highly promising for three-dimensional information storage and encryption. And its inherent high thermal stability and mechanical strength address key obstacles for practical applications.

Fig. 5. Applications for muti-color information storage and dynamic encryption.

a, b Reversible muti-color display potential for dynamic anti-counterfeiting through (a) charging temperature control or (b) excitation wavelength control. c, d Real-time multicolor writing for information storage, achieved by controlling (c) the local temperature and d the excitation wavelength in different pattern regions.

In summary, we have designed a lithium-ion doping-assisted phase separation strategy to achieve a new type of transparent glass ceramic with the in-situ precipitation of non-stoichiometric nanocrystals (Zn_1.7SiO₄: Li). Compared to the amorphous glassy phase, the newly formed non-stoichiometric phase provides a quite different ligand field environment (from octahedron to tetrahedron coordinated) and a flexibly adjustable defect environment (the relative concentration of deep defects in the crystalline phase can be selectively mediated by heat treatment temperatures) for the activators Mn²⁺, which breakthroughs the problem of coupling luminescence centers with different defect states. By adjusting the excitation wavelengths (e.g., 254 nm or 365 nm) and charging temperatures (e.g., 298 K or 498 K), the programmability of the trap states in dual-phase leads to a relative intensity change of the dual emissions, making spatiotemporal tunable PersL of wide chromaticity from orange to green in a single glass solid a reality. In addition, we have also verified the practical application advantages of this glass ceramic by confirming its outstanding mechanical property (hardness, 9.3 ~ 10.7 gigapascal) and high luminescence thermal stability (110% PersL intensity at 100 °C relative to room temperature). Finally, the successful realization of this multi-mode programmable dynamic PersL in a transparent media highlights its potential for multi-dimensional information storage, encryption and 3D optical devices. Furthermore, the current principle could be applicable for tunable luminescence in other multiphase systems.

Methods

Material synthesis

The glass precursor powder, comprising 40ZnO-30SiO₂-10Al₂O₃-10H₃BO₃-xLi₂CO₃-(10-x) K₂CO₃-0.5MnCO₃ (Aladdin, >99.9% purity, x = 0–4 mol%), was uniformly mixed for 10 min in an agate mortar. The mixture was then loaded into a furnace and reacted at 1430 °C for 40 min under a reducing atmosphere. The fully reacted glass melt was cast onto a preheated 300 °C steel plate and immediately transferred to a 400 °C annealing furnace for thermal stress relief. The reducing atmosphere throughout the experiment was maintained by using coconut-shell activated carbon granules (1–2 mm diameter) to prevent the oxidation of Mn²⁺. The obtained glass/glass ceramics were then cut, ground, and polished for the necessary measurements. All glass/glass ceramic samples for optical imaging in this work were 12 mm in diameter.

Material characterizations

DSC tests were carried out with a heating rate of 7 K/min on a differential scanning calorimeter (NETZSCH, DSC404F3). XRD analyses were performed by employing an Empyrean 200895 diffractometer with Cu-Ka radiation. The TEM and HRTEM images were performed with a transmission electron microscope (JEOL, JEM-2100F) at 200 kV accelerating voltage. FTIR spectroscopy analysis was carried out by a Nicolet IS50 Fourier infrared spectrometer to detect the vibrations of chemical bonds. XPS technique was carried out using a Thermo 250XI equipment. Portable hand-held UV lamps (3 W, 254 nm; 3 W, 295 nm; 3 W, 365 nm) were employed to irradiate the glass /glass ceramics while taking PL and PersL photographs. Room-temperature PL/PersL spectra and variable-temperature PL/PersL spectra were gained from a self-built confocal fluorescence spectrophotometer equipped with a 365 nm lamp and a 254 nm lamp as the excitation light. EPR spectroscopy was performed to identify the unpaired electrons associated with vacancy defects based on a Bruker A300 spectrometer equipment. 3D-TL spectra and 2D-TL spectra were gained by using a Guangzhou TOSL-3DS equipment at a heating rate of 5 K/s, with the tested samples being pre-excited by UV excitation for five minutes. The afterglow decay curves of the glass ceramic at different charging temperatures were obtained based on an Edinburgh FLS equipment. The three-dimensional (3D) patterning within the glass ceramic was achieved via a femtosecond laser (365 nm, power 2 W, repetition 11 kHz).

Author contributions

W.Y. performed the experiments, collected the data, and wrote the initial manuscript. L.X., R.C., S.K. and Q.J. discussed the manuscript. T.D. conceived the idea, organized and supervised the project, interpreted the results, and revised the manuscript.

Peer review

Peer review information

Nature Communications thanks Yixi Zhuang, Yuansheng Wang, who co-reviewed with Hang Lin, 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 source data generated in this study are provided in the Supplementary Information/Source Data file. 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/s41467-026-69202-9.