CsPbX3 (X = Cl, Br, I) QDs are emerging semiconductor materials and have attracted increasing attention due to their high defect tolerances, charge carrier mobilities, tunable bandgaps, and long carrier diffusion lengths1,2. Numerous recent reports have demonstrated the advantages of CsPbX3 QDs over conventional semiconducting materials3. In particular, their high photoluminescence quantum yields (PLQYs), narrow emission widths, and wide color range make them competitive candidates for lighting, backlight display, and anti-counterfeiting applications3. As photoluminescent layers, CsPbX3 QDs have been used in WLEDs and anti-counterfeiting coatings exhibiting high performance4. To accelerate practical use of CsPbX3 QDs, a great deal of research on crystal structures and fabrication strategies has been carried out, and remarkable progress has been achieved5. However, it is still challenging to use CsPbX3 QDs as commercial materials because of the drawbacks of prevalent synthetic methods, which usually involve large amounts of organic solvents, ligands, inert gas protection, and tedious cleaning and purification processes, thereby rendering them unsuitable for environmentally friendly and scalable production6,7. In addition, the stabilities of colloid QDs prepared by using solvents are generally mediated by large organic ligands, but unfortunately, the highly dynamic surfaces of QDs might cause desorption of ligands, resulting in structural damage and performance degradation of the ionic CsPbX3 QDs due to moisture or thermal attack8,9.

Great effort has been expended to overcome the instability issues, and various passivation strategies have been proposed, such as surface engineering and matrix encapsulation10. Mohammed et al. found that ligands with –NH3+ groups strongly bound to Br ions on specific surfaces of CsPbBr3 (110) and enhanced the PL intensity and stability11. Compared with ligand modification methods, the inert encapsulation technique showed greater potential for preparing ultrastable perovskite composites. Oxides (SiO212, SiO2/AlOx13, molecular sieves14,15,16), semiconductors (ZnS17 and Pb4S3Br218), polymers (polydimethylsiloxane19 and polystyrene20), and metal-organic frameworks (PCN-333(Fe)21 and UiO-66(NH2)22) have served as effective matrix materials. Y Lin et al. reported the preparation of CsPbX3/SiOx by high-temperature sintering synthesis, and the humidity and heat resistance were significantly improved12. Wang et al. successfully synthesized an ultrastable perovskite composite based on the aluminophosphate AIPO-5, which provided confinement for growth of the nanocrystals, defect passivation, and robust barrier surroundings15. In addition to the enhanced stability, the absolute PL intensities of CsPbX3-agZIF-62 composites are often at least two orders of magnitude higher than those of the corresponding pure CsPbX323. In view of the significant breakthroughs in the stabilities of CsPbX3 QDs, the problems with scalable preparation should be considered because complicated postprocessing steps are inevitable in the strategies mentioned above24.

The solvent-free chemical vapor method is commonly applied in scalable syntheses of functional films, nanotubes, and catalysts due to its relatively low reaction temperatures, flexible product compositions, broad application range, and extraordinary diffusion ability25,26. Molecular sieves are typical active porous materials, and they allow gases and small molecules to pass through their unique channel structures; these are ideal accessory ingredients and support materials in various synthetic reactions27,28. Herein, a novel chemical vapor method is designed and demonstrated for scalable production of CsPbX3 composites with superior optical properties and ultrahigh stabilities. By employing ZSM-5 as the porous template and taking advantage of special reactions between PbBr2 vapor and the Si–O network of ZSM-5, in situ growth of CsPbX3 QDs confined in the nanometer-scale space is achieved without organic solvents and ligands. Additionally, the acquired encapsulation structure has the channels needed for halogen exchange to regulate the halide ratios of the CsPbX3-ZSM-5 composites. Consequently, the CsPbX3-ZSM-5 composites can be mass-produced with high PLQYs and narrow emission FWHM. In addition, owing to the protection and isolation provided by inert ZSM-5, the CsPbX3 QDs are prevented from agglomerating and regrowing, and the composites exhibit exceptional stability under harsh conditions, including heat, water, polar solvents, and UV light. To demonstrate practical viability, the composites are applied in WLEDs with a large color range and multicolor-coded anti-counterfeiting inks.

Materials and methods

Materials and chemicals

PbBr2 (lead bromide, 99%, Shanghai Aladdin Biochemical Technology Co., Ltd., China), CsBr (cesium bromide, 99.5%, Aladdin), ZSM-5 molecular sieves (SiO2/Al2O3, molar ratio ~ 40–50, Aladdin), tetrabutylammonium chloride (TBAC, 97%, Aladdin), and tetrabutylammonium iodide (TBAI, 99%, Aladdin) were used as received without further purification.

Synthesis of CsPbX3-ZSM-5 composites

The green-emission composite was synthesized via a facile one-step vapor diffusion method. CsBr and PbBr2 were weighed to give a 1:1 stoichiometric ratio, and an appropriate amount of ZSM-5 was added (designed mass ratio of (CsBr+PbBr2): ZSM-5 = 1:2). The mixture was calcined at 650 °C for 300 min with a heating rate of 10 °C min−1 and then cooled to 30 °C in a muffle furnace in air. The processes used to prepare materials with different proportions were similar. CsPbBrxCl3−x-ZSM-5 and CsPbBrxI3−x-ZSM-5 with different halogen compositions were synthesized via ion exchange by mixing and grinding the CsPbBr3-ZSM-5 powder with a certain amount of tetrabutylammonium iodide or tetrabutylammonium chloride and then calcining at 250 °C for 300 min to obtain CsPbBrxCl3−x-ZSM-5 and CsPbBrxI3−x-ZSM-5.

Syntheses of ZSM-650, CsBr-ZSM, PbBr2-ZSM

ZSM-650 was prepared by annealing ZSM-5 at 650 °C. CsBr-ZSM was prepared by mixing CsBr with ZSM-5 (mass ratios from 1:2 to 2:1) and then annealing at 650 °C. Similarly, PbBr2-ZSM was prepared by mixing PbBr2 with ZSM-5 (mass ratios from 1:4 to 1:1) and then annealing at 650 °C.


The morphologies and microstructures of the CsPbX3-ZSM-5 composites were examined by scanning electron microscopy (SEM, Zeiss GeminiSEM 300) and transmission electron microscopy (TEM, FEI Talos F200X) at 200 kV. The compositions of the composites were determined by energy-dispersive X-ray spectroscopy (EDS) using an accessory manufactured by Oxford Instruments. To characterize structures, powder X-ray diffraction (XRD) studies were performed with a Rigaku Smartlab 3 kW X-ray diffractometer with Cu Kα radiation (λ = 1.54056 Å, 40 kV, 30 mA, 10° min−1 from 5 to 80°). X-ray photoelectron spectroscopy (XPS) was carried out on a Thermo ESCALAB 250Xi system. The structures of the composites were analyzed by Fourier transform infrared spectroscopy (FTIR, Thermo Scientific, Nicolet iS60) with the KBr pellet method. PL properties were investigated on a fluorescence spectrophotometer (Hitachi F-4600) with an excitation wavelength of 365 nm, and UV–Vis spectra were recorded on an ultraviolet spectrometer (AOE Instruments, UV-A390). Thermal properties were determined by simultaneous thermal analyses (ZCT-B TG/DTA) run at a heating rate of 10 °C min−1 in air, and micropore analyses, including surface areas and pore volumes, were conducted with a Micro Active ASAP 2460 system.

PLQY measurements

The composite powders were dispersed in glycol to prepare samples. Samples were photoexcited at an excitation wavelength of 365 nm, and a spectrophotometer (FLS 1000, Edinburgh, UK) was used for PLQY analysis.

Fabrication of WLEDs

WLEDs were fabricated by combining synthesized green-emission and red-emission perovskite composites with blue LED chips. The composites were mixed with epoxy resin by vigorous stirring, dropped onto the chip and cured at 80 °C. The devices were characterized using a Keithley 2450 system as a current source, and the output spectra were recorded with an HP-8000 LED fast-scan micro-spectrophotocolorimeter.

Stability evaluation

Water and polar solvent stability tests were performed by dispersing the powders directly in the solvents. The samples for stability tests involving heating, aging, and continuous radiation were prepared by putting solutions of CsPbX3-ZSM-5 composites onto quartz plates to form films. The films were exposed to a heating plate under aging conditions of 80 °C and 80% relative humidity in a constant temperature and humidity incubator as well as continuous UV light illumination (365 nm) separately. Photoluminescence spectra were acquired after different testing conditions using a fluorescence spectrometer.

Results and discussion

Chemical vapor syntheses of CsPbX3-ZSM-5 composites

A novel chemical vapor technique with advantages such as stability in an air atmosphere, operation without organic solvents, and scalability was developed to produce commercially available CsPbX3 composites. As illustrated in Fig. 1a, CsBr and PbBr2 were placed separately at the bottom of a ceramic crucible and covered by a layer of ZSM-5 particles and then annealed in an air atmosphere with a programmed temperature. After cooling the furnace to room temperature, a CsPbBr3-ZSM-5 composite exhibiting bright green emission was obtained and named Green-Z, The SEM image and elemental maps in Fig. 1b and Fig. S1 in the Supplementary Information suggest the formation of CsPbBr3. The detailed SEM images of pristine ZSM-5 (Fig. 1c) and Green-Z (Fig. 1d) clearly showed the great morphological changes of ZSM-5 particles from a regular square shapes to a smooth and round surfaces. According to the transmission electron microscopy (TEM) image (Fig. 1e), the isolated CsPbBr3 QDs were distributed uniformly in the matrix. A high-angle annular dark-field (HAADF) image and elemental maps are presented in Fig. 1f. It is known that TEM contrast is commonly limited by the thicknesses of samples or matrices with high density, and incomplete information is sometimes provided by TEM images. Unlike TEM, HAADF image intensities are approximately proportional to the square of the atomic number of the substance, so HAADF images are less affected by the surrounding medium29. The HAADF image contrast was consistent for the different atoms in ZSM-5 and CsPbBr3 QDs, and the average size of the CsPbBr3 QDs was 6.40 nm. It should be noted that when the (CsBr+PbBr2) proportion increased, the signal from CsPbBr3 increased, and the color of the composite became deeper (Fig. S2, Supplementary Information); however, when the CsPbBr3 was overloaded, it grew on the surfaces of ZSM, which was investigated by immersing the composites in water. Samples 6# and 7# showed deeper colors and certain amounts of perovskite on their surfaces before immersion in water and became whiter powders with porous structures after immersion in water, while the other samples showed little change (Fig. S3, Supplementary Information).

Fig. 1: Synthesis of the CsPbBr3-ZSM-5 composite.
figure 1

a Schematic illustration of the chemical vapor synthesis procedure for the CsPbBr3-ZSM-5 composite. b SEM image of Green-Z and corresponding elemental maps for Br, Pb, and Cs. SEM images: c Pristine ZSM-5 and d Green-Z. e TEM image of Green-Z. f HAADF image of Green-Z and corresponding elemental maps for Br, Pb, and Cs (the inset shows the particle size distribution histogram).

CsPbX3-ZSM-5 samples with different halide ratios and emission colors were prepared by vapor halogen exchange, as Fig. 2a illustrates. The synthesized Green-Z composite acted as a raw material and was heated to 250 °C together with tetrabutylammonium halide. The PL emission of CsPbX3-ZSM-5 was tunable with the halide ratio, and three typical samples exhibiting bright green, blue and red emission under UV light are shown in Fig. 2a. The elemental maps in Fig. 2b and Figs. S4 and S5 in the Supplementary Information also indicated successful halogen exchange. Despite the different emission colors, different CsPbX3-ZSM-5 samples, designated Red-Z, Orange-Z, Yellow-Z, Turquoise-Z, Blue-Z, and Purple-Z, had similar morphologies (Fig. 2c), demonstrating that halogen exchange did not change the morphology of Green-Z. Moreover, the structures of the composites were characterized by XRD (Fig. 2d). The peaks from the Green-Z composite were indexed to CsPbBr3 (PDF#18-0364) and ZSM-5 (PDF#44-0003). Slight shifts in the perovskite diffraction peaks from the other six samples were also observed, which showed that halogen exchange in the vapor phase was effective. For example, the shifts of peaks from Red-Z to the left reflected an increase in interplanar spacing caused by the larger radii of I- ions, while Purple-Z showed the opposite effect. The diffraction signals from CsPbX3 were weaker than those from ZSM-5 due to the large size differences between the nanoscale CsPbX3 dispersed in the pore channels and the micron-scale ZSM-5 matrix.

Fig. 2: Syntheses of CsPbX3-ZSM-5 composites.
figure 2

a Schematic illustration of the vapor halogen exchange and photographs of Green-Z, Blue-Z, and Red-Z under UV excitation at 365 nm. b Elemental maps for typical Blue-Z and Red-Z samples. c SEM images and d XRD patterns for multicolor CsPbX3-ZSM-5 composites.

Mechanism of chemical vapor synthesis

Experiments were designed to elucidate the mechanism for the formation of the CsPbX3-ZSM-5 composites. First, pristine ZSM-5, a mixture of CsBr and ZSM-5 and a mixture of PbBr2 and ZSM-5 were annealed at 650 °C and named ZSM-650, CsBr-ZSM (1:2), and PbBr2-ZSM (1:2), respectively. SEM images of the three samples are shown in Fig. 3a. ZSM-650 and CsBr-ZSM showed no significant difference compared with pristine ZSM-5 (Fig. 1c); however, PbBr2-ZSM was similar to Green-Z (Fig. 1d) with obvious etching tracks. Then, other lead compounds, such as PbO, Pb(CH3COO)2, and Pb(NO3)2, were also annealed at 650 °C together with ZSM-5, but a similar etching phenomenon was not observed (Fig. 3a and Fig. S6 in the Supplementary Information), indicating the indispensable synergistic role of Pb2+ and Br in the structural evolution. In addition to the morphological evidence, XRD patterns (Fig. 3b, Figs. S7 and S8 in the Supplementary Information) further confirmed the unique effect of PbBr2 that prompts the crystal structure of ZSM-5 to collapse and become amorphous, and the more PbBr2 there was, the more obvious the collapse. It should be noted that with further increases in the mass ratio of CsBr to ZSM (from 1:2 to 2:1), the morphology and crystallinity of ZSM were still maintained (Figs. S9 and S10 in the Supplementary Information).

Fig. 3: Morphological and structural evolution of the ZSM-5 matrix annealed with other substances.
figure 3

SEM images of ZSM-650, CsBr-ZSM, PbBr2-ZSM and Pb(NO3)2-ZSM. b XRD patterns, c FTIR spectra, and d DTA curves of ZSM-5 (gray line), CsBr-ZSM (pink line), and PbBr2-ZSM (blue line).

To study the reaction process, Fourier transform infrared (FTIR) spectroscopy (Fig. 3c and Fig. S11 in the Supplementary Information) was carried out. The vibrational peaks for Si–O–Si systems appeared at 543 cm−1 (double five-ring vibrations in the ZSM-5 framework) and 795 cm−1 (internal symmetric stretching vibrations, which are also observed for silica)30,31, and the peak at 620 cm−1 (double tetrahedral rings in the ZSM-5 framework)30 for ZSM-5 as well as the weak broad band from 3648 to 3655 cm−1 for –OH of Si–OH moieties were also observed. The pristine ZSM-5 and CsBr-ZSM samples exhibited similar spectra. However, the spectrum for PbBr2-ZSM differed in that the peak at 620 cm1 and the bands for Si–OH were extremely weak. In particular, the peak at 543 cm−1 for vibrations of double five-rings in the ZSM-5 framework disappeared, indicating that the structure of ZSM-5 had changed. Small changes in the peak at 795 cm−1 were ascribed to the internal symmetric stretching vibrations of silica, which resisted structural modification31. Based on these results, it is proposed that PbBr2 might react with the Si–OH species in ZSM-5 and contribute to breakage of the matrix structure.

To verify this hypothesis, differential thermal analyses (DTA) were performed, and the curves are shown in Fig. 3d. No peak was observed for ZSM-5 because of its high thermal stability, and the one endothermic peak at 628 °C from CsBr-ZSM corresponded to the melting point of CsBr. For PbBr2-ZSM, the endothermic peak at 370 °C was due to melting of the PbBr2, and the peak at ~525 °C suggested an endothermic process that might be a reaction between PbBr2 and ZSM-5. To simulate the main synthetic processes, the DTA curve for a mixture of CsBr, PbBr2, and ZSM-5 (Fig. S12a in the Supplementary Information) was measured and compared with that for pure CsPbBr3 (Fig. S12b in the Supplementary Information). In Fig. S12a in the Supplementary Information, endothermic peaks for the melting points of PbBr2 (~373 °C) and CsPbBr3 (~563 °C) were identified. The exothermic (~479 °C) and endothermic (~502 °C) peaks might be associated with the reactions that cause the morphological changes in ZSM-5, which is consistent with the result observed from the DTA curve for PbBr2-ZSM.

To further explore the etching reaction, changes in the chemical states of ZSM-5 were monitored by XPS. As shown in Fig. 4a–c, the peaks for Si 2p, O 1s, and Al 2p binding energies with PbBr2-ZSM showed obvious right shifts compared to those for pristine ZSM-5. The Si 2p XPS peak of pristine ZSM-5 was located at 103.6 eV, and that of CsBr-ZSM appeared at 103.5 eV. The Si 2p peak of PbBr2-ZSM showed a 0.8 eV shift from 103.6 to 102.8 eV as a result of silicate formation32. The O 1s XPS peaks of pristine ZSM-5 and CsBr-ZSM appeared at 532.7 and 532.5 eV, respectively, whereas that of PbBr2-ZSM was shifted to 531.9 eV. The negative binding energy shifts can be ascribed to changes in the chemical environments of Si–OH species. These results are consistent with previous observations that Pb2+ can easily replace protons in Si–O–H aluminosilicates due to the high electronegativity (2.33) of Pb2+ and its strong oxygen binding capacity33. The slight shifts of the Si 2p and O 1s peaks to lower energies and the shift of the Al 2p peak from CsBr-ZSM to higher energy might be ascribed to weak interactions of the CsBr with ZSM-5. In addition, as shown in Fig. 4d–f, there were obvious Br 3d and Pb 4f signals from PbBr2-ZSM, and only a faint Cs 3d XPS peak was observed for CsBr-ZSM, indicating high efficiencies for entrapment of Pb and Br under specific conditions. Overall, the XPS results suggested that the Si–OH bonds in ZSM-5 reacted with PbBr2 to form “Si–O–PbBr” bonds and the Si–O–Pb–O-Si network at high temperatures, and the additional HBr might contribute to the structural etching and collapse of ZSM-5.

Fig. 4: Chemical state evolution and postulated mechanism for the chemical vapor method.
figure 4

High-resolution XPS spectra for ZSM-5 (gray line), CsBr-ZSM (pink line), and PbBr2-ZSM (blue line): a Si 2p, b O 1s, c Al 2p, d Br 3d, e Pb 4f, and f Cs 3d binding energies. g Schematic diagram of the postulated formation mechanism.

Brunauer–Emmett–Teller (BET) gas absorptiometry was used to investigate the porous structures. As listed in Table S1 in the Supplementary Information, pristine ZSM-5, with a large surface area of 324.4509 m2 g−1 and a narrow mean pore size of 2.0674 nm, constituted a desirable substrate for loading of functional nanomaterials. After the reaction with PbBr2, the surface area decreased slightly to 266.2026 m2 g1, and the pore size and volume increased slightly from 2.0674 to 2.1025 nm and 0.13997 to 0.144903 cm3 g1, respectively, due to etching. However, the Green-Z composite had a smaller surface area (183.3242 m2 g1), which decreased with increasing loading, indicating that the CsPbBr3 QDs filled the pores and occupied the spaces. Other than the excessive loading for sample #7 (Fig. S2 in the Supplementary Information), some pores remained in the Green-Z composite to provide channels for halogen exchange. The pore sizes were smaller than the average size for CsPbBr3 QDs distributed in ZSM-5 (~6.4 nm), implying the formation of perovskite in the interrupted structure of the substrate9,15.

Hence, the mechanism for in situ syntheses of CsPbX3 composites is postulated in Figs. 4g and S13 in the Supplementary Information, and the main reactions are described in Scheme 1 in the Supplementary Information. First, owing to the relatively low melting point, PbBr2 melts in the heating process and moves upward to the pores of ZSM-5 driven by air convection. The H sites in ZSM-5 react with PbBr2 to form HBr, and the remaining PbBr+ or Pb2+ binds with Si–O, resulting in incorporation of Pb into the framework. Second, HBr attacks the Si–O bonds and etches channels in the ZSM-5. Subsequently, as the temperature is raised, CsBr reacts with the Si–O–PbBr and PbBr2 trapped by the matrix and intensifies etching at the higher temperatures, which further accelerates the structural collapse of ZSM-5 and seals the CsPbBr3 formed in the channels. During cooling, CsPbBr3 condenses and crystallizes to form QDs in the confined spaces. The remaining channels are responsible for further halogen exchange. A dynamic video clip is included in the Supporting Information to illustrate the process used for syntheses of the CsPbX3-ZSM-5 composites.

Photoluminescence properties and stabilities

The photophysical properties of the CsPbX3-ZSM-5 composites were investigated, and all optical parameters are summarized in Table S2 in the Supplementary Information. First, the absorption and PL emission spectra (Figs. 5a, b and S14 in the Supplementary Information) can be tuned almost across the visible light region simply by changing the halide ratios. Typical Stokes shifts (ΔEs) of approximately 40 meV attributed to perovskite QDs have been observed34. Moreover, the narrow emission bands of these composites (with FWHMs from 90 to 143 meV) were comparable to those of solution-synthesized QDs35. It should be noted that Green-Z (QD average lengths ~6.4 nm) showed an emission peak at approximately 523 nm, which was a higher wavelength than that previously reported for monodispersed CsPbBr3 nanocubes with edge lengths of 6.3 nm (the emission peak was at approximately 504 nm)36. The large difference was probably due to the special packaging and collective effects of the QDs and the molecular sieve matrix, which differed from those of monodispersed QDs formed in solvents. Similar size distributions and PL mismatches have been reported for QDs embedded in glass37 (sizes averaging 3.8 nm, PL at 516 nm) and CsPbBr3 NCs@mesoporous silica nanospheres4 (sizes averaging 7.8 nm, PL at 524 nm). With regard to PL efficiency, the PLQY of the Green-Z composite in glycol solution was 92.52% (λex = 365 nm) at room temperature, and those of the Blue-Z and Red-Z composites were approximately 46.24% and 16.91%, respectively (Fig. S15 in the Supplementary Information). Compared to those seen in the glycol dispersions, the PLQYs of Green-Z, Blue-Z, and Red-Z powders showed lower values of 24.11, 5.09, and 4.81%, respectively, which may be due to self-absorption. The time-resolved PL decay spectra of Blue-Z, Green-Z, and Red-Z are shown in Fig. 5c, and the PL decay curves were fitted with tri-exponential decay functions, as shown in Table S3 in the Supplementary Information. The average PL lifetimes of Blue-Z, Green-Z, and Red-Z were 16.44, 22.92, and 39.19 ns, respectively, longer than that of pure perovskite QDs. This could be explained by passivation arising from confined growth in the ZSM-5 structure as well as reduced nonradiative transitions38. These results validate the excellent PL properties of the CsPbX3-ZSM-5 composites.

Fig. 5: Photoluminescence properties and stabilities of the synthesized composites.
figure 5

a Absorption spectra of CsPbX3-ZSM-5 from 400 to 700 nm (the inset shows photographs of CsPbX3-ZSM-5 dispersed in glycol under ambient light and 365 nm UV light). b PL spectra of CsPbX3-ZSM-5 from 400 to 700 nm. c Time-resolved PL decay and fitting curves of the Blue-Z, Green-Z, and Red-Z composites. PL intensity plots: d Blue-Z, e Green-Z, and f Red-Z composites at 20 and 200 °C as a function of cycling number. g Water stability evaluation: time-dependent PL intensity of the Blue-Z, Green-Z, and Red-Z composites soaked in water. h Aging and stability test: time-dependent PL intensities of the Blue-Z, Green-Z, and Red-Z composites at 80 °C and 80% relative humidity. i Photostability test: time-dependent PL intensities of Blue-Z, Green-Z, and Red-Z upon continuous illumination (365 nm).

Stability is another vital requirement for commercial applications. The thermal and moisture resistance of the CsPbX3-ZSM-5 composites were first evaluated. An obvious PL decrease and approximately unchanged PL peak positions were observed during heating due to the self-trapping excitons of CsPbX3 (Fig. S16 in the Supplementary Information)38. All of the tested composites (Blue-Z, Green-Z, and Red-Z) showed only approximately 50% PL loss during heating from 20 to 100 °C, and 10% of the PL intensity remained even at 200 °C, indicating that the excitons in the CsPbX3-ZSM-5 composites were more thermally stable than those of pure CsPbX3 QDs39. As shown in Fig. 5d–f, the PL decrease caused by thermal quenching was recoverable, and the PL intensity was nearly equal to the initial value after cooling to room temperature. In addition, reversible PL was observed when heating the samples from 20 to 200 °C for 20 cycles. The outstanding thermal stability can be ascribed to three main factors. First, ZSM-5 has low thermal conductivity and protects the internal CsPbX3 QDs from high temperatures. Second, confinement within the porous structure of ZSM-5 hinders the aggregation of CsPbX3 QDs at a high temperature. Finally, no ligands were used in the syntheses of the composites, thus avoiding fluorescence damage caused by oxidation and delamination of organic ligands14.

In the solvent resistance tests, the Blue-Z, Green-Z, and Red-Z composites were immersed in different solvents to monitor PL evolution. Figure 5g shows that after immersion in water for 30 days, Blue-Z, Green-Z, and Red-Z maintained 84, 87, and 86% of their initial PL intensities, respectively. Stability in other solvents was also assessed, as shown in Figs. S17, S18, and S19 in the Supplementary Information. The samples were immersed in ethanol, isopropanol, glycol, dimethylformamide, dimethyl sulfoxide, and acetylacetone, and all of them maintained strong emission intensities after 180 days. The effects of aging are presented in Fig. 5h. The PL intensity of the CsPbX3-ZSM-5 composite was still higher than 90% of the initial value after 30 days of aging at 80 °C under 80% relative humidity (abbreviated as “80 °C, 80% R.H.”). To further evaluate the stability under other harsh conditions, the samples were exposed to continuous UV radiation at 365 nm, and the PL intensities exhibited nearly no changes after 500 h (Fig. 5i). The excellent resistance of the CsPbX3-ZSM-5 composites to water, polar solvents, aging, and UV radiation was attributed to protection from ZSM-5, which precluded direct exposure of the CsPbX3 QDs to these external factors and inhibited interparticle fusion of the perovskite nanoparticles. These results provided strong evidence showing that the proposed strategy overcame the inherent instability of CsPbX3 QDs.

Applications for WLEDs and anti-counterfeiting

Because of the advantages of the synthetic strategy, scalable preparation (Fig. 6a) can be easily realized and makes commercialization a possibility. In practice, the use of composites commonly involves organic polymers such as polydimethylsiloxane (PDMS) and polyurethane (PU); therefore, the practicability of these perovskite composites was investigated. Good compatibility with PDMS is indicated in Fig. 6b, and the uniform fluorescent disks emitted bright fluorescence under UV light excitation. In addition, a large-area film measuring 35 cm × 25 cm and made of the Green-Z composite and PU was prepared by screen printing (Fig. S20 in the Supplementary Information).

Fig. 6: Applications of the synthesized composites.
figure 6

a Photographs of 1 kg of the CsPbBr3-ZSM-5 composite under ambient and UV illumination. b Photographs of CsPbX3-ZSM-5/PDMS disks with tunable emission under ambient and UV light. c Emission spectrum of the WLED fabricated by depositing the Green-Z and Red-Z composites on blue LED chips (the inset shows a photograph of the device operated at 20 mA). d CIE chromaticity coordinates and color range for the WLED. e Fluorescent images of the patterns prepared by silk printing and demonstration of anti-counterfeiting coding with CsPbX3-ZSM-5-based fluorescent security inks.

LEDs are widely recognized as mainstream devices in next-generation backlit displays20. Due to their compatibility with polymers, light conversion LED chips can be fabricated by packaging CsPbX3-ZSM-5 composites into commercial LED chips. The tolerance of Green-Z and Red-Z packed in the single-wavelength LED chips was evaluated by monitoring the output spectra produced by different driving currents under working conditions. As shown in Fig. S21 in the Supplementary Information, the light intensity had a linear relationship with the current applied. Accordingly, WLEDs were fabricated by combining the Green-Z and Red-Z composites with blue LED chips (430 nm). The emission spectrum of the WLED is presented in Fig. 6c, and the triangle for the CIE color coordinates in the CIE 1931 chromaticity diagram is displayed in Fig. 6d. Three obvious emission peaks were observed at 430, 517, and 687 nm, and the CIE chromaticity coordinates were (0.17, 0.01), (0.13, 0.77), and (0.73, 0.27), respectively. The luminous efficiency of the WLED was 6.8 lm/W at a current of 20 mA. The chromaticity coordinate of the WLED was (0.32, 0.33), and the color temperature (CCT) was 6062 K, close to that of standard white light. The area of the triangle was calculated to be 138% of the National Television System Committee (NTSC) standard and 103% of the ITU-R Recommendation BT.2020 (Rec.2020.), demonstrating an ultrabroad color range.

The CsPbX3-ZSM-5 composites with narrow emission bands and tunable colors also have ample potential for use in security printing technology. Except for the intrinsic UV-excited on/off phenomenon, secondary anti-counterfeiting coding is also quite useful. As mentioned above, the WLEDs exhibited three discrete primary color peaks, thereby providing the possibility of anti-counterfeiting responses coded with the three colors. As shown in Fig. 6e, patterns were printed by blending Blue-Z, Green-Z, and Red-Z with epoxy resin. Similar white emissions with different emission spectra were achieved by mixing the three composites in different ratios, which made it difficult for the naked eye to discern the differences. Therefore, the concept of white light-coded anti-counterfeiting is proposed. The rules for coding were enacted and shown in Fig. 6e. First, blue, green, and red emissions were denoted as “0”, “1”, and “2”, respectively. Second, one of the white light spectra was selected as a reference and encoded as (0 1 2). Finally, the relative increase in a specific peak intensity compared to the reference was denoted as “” on the top of the corresponding number; otherwise, it was denoted as “”. For example, white light with a stronger blue emission peak would be noted as (\(\hat 0\) 1 2). To assess the practicality, a batch of fluorescent inks exhibiting white emission was prepared, and based on the aforementioned coding rules, symbols were printed with the four different inks, as shown in Fig. 6e. Although all parts of the symbols showed white emission, they had different spectra. These results revealed the excellent potential of the composites for use in anti-counterfeiting coding.


A novel chemical vapor method was designed and demonstrated for large-scale syntheses of CsPbX3-ZSM-5 composites without the need for organic solvents, organic ligands or an inert environment. Confined growth of CsPbX3 inside the nanopores of ZSM-5 was observed because of reactions between PbBr2 vapor and the Si–O network of ZSM-5, and the resulting encapsulation structure provided the channels needed for halogen exchange. This method offered scalable production of extremely stable composites that exhibited tunable emissions with high PLQYs, narrow emission FWHMs, heat resistance up to 200 °C and radiation resistance, as demonstrated by continuous UV radiation, for 500 h. After immersion in water for 30 days or polar solvents for 180 days, the PL intensity exhibited almost no change. Furthermore, the composites survived aging (80 °C, 80% R.H.) and retained 90% of the initial PL intensity after 30 days. The narrow emissions, high PLQYs and outstanding stability make the CsPbX3-ZSM-5 composites promising for lighting and display applications, especially WLEDs with large color ranges and multicolor-coded anti-counterfeiting inks. This facile, organic-free and ambient atmosphere process indicates the probability for commercial production of these perovskite composites with robust stability.