Coordination-mediated cold quenching for the synthesis of strongly quantum-confined FAPbI₃ quantum dots

Abstract

Strong quantum confinement can fundamentally reshape excitonic processes and radiative recombination in perovskite quantum dots (QDs), yet accessing this regime remains challenging in organic-inorganic hybrid FAPbI₃ nanocrystals because of their low crystallization barrier and rapid growth. Here, we report a coordination-mediated cold quenching strategy for synthesizing strongly quantum-confined FAPbI₃ QDs with tunable emission from 696 to 759 nm. The key step involves the injection of a precooled ethyl methacrylate (EMA, –70 °C), which rapidly lowers the reaction temperature while simultaneously retarding precursor conversion and suppressing QD regrowth. Such a strategy enables small QDs with an average size of 6.4 nm and a near-unity photoluminescence quantum yield. Representative light-emitting diodes based on these QDs deliver electroluminescence at 715 nm with a peak external quantum efficiency of 16.3%. This work provides a viable strategy for accessing ultrasmall hybrid perovskite QDs and advances the rational control of quantum confinement in soft ionic semiconductors.

Strong quantum confinement could improve hybrid perovskite nanocrystals, but producing ultrasmall, uniform particles is difficult. You, Wang et al. used coordination-assisted cold quenching to create efficient emitters for light-emitting diodes and optical communication.

Introduction

Precisely tailoring the particle size of quantum dots (QDs) into the strong-confinement regime is crucial for modulating their electronic structure, excitonic interactions, and carrier relaxation/recombination dynamics^1–3. To date, considerable progress has been made in the synthesis of all-inorganic perovskite QDs, particularly CsPbX₃ (X = Cl, Br, I), with well-controlled particle sizes and pronounced quantum-confinement effects^4–6. These advancements have enabled the development of efficient perovskite light-emitting diodes^7–10 (PeLEDs) and provided model systems for probing unique photophysical phenomena, such as multiexciton generation and phonon-bottleneck-mediated retardation of hot-carrier cooling^11,12.

These advances motivate the extension of precise size control to organic-inorganic hybrid perovskites, particularly formamidinium (FA)-based systems, in which the molecular A-site cations introduce stronger dielectric screening, richer low-frequency phonon modes, more pronounced phonon-bottleneck and polaronic effects^13–15. These characteristics provide additional opportunities for investigating confinement-dependent photophysics and developing optoelectronic devices¹⁶. So far, strongly confined FAPbBr₃ QDs have been demonstrated¹⁷, establishing that strong quantum confinement is accessible in FA-based perovskites. However, achieving comparable size control in FAPbI₃, with its narrower bandgap and near-infrared emission, remains very challenging because of its distinct crystallization characteristics.

This challenge originates from two coupled aspects of FAPbI₃ crystallization. First, compared with FAPbBr₃ and the more extensively studied CsPbX₃ systems (X = Cl, Br, I), FAPbI₃ exhibits a much lower formation-energy barrier and faster crystallization kinetics. Nucleation is therefore rapidly followed by crystal growth upon precursor mixing, leaving a narrow temporal window for controlling the final particle size and size distribution^18,19. Second, once ultrasmall FAPbI₃ QDs are formed, their high surface energy, arising from the enlarged surface-to-volume ratio, promotes continued growth, aggregation, and Ostwald ripening during purification and storage^20,21. Although ligand engineering and regulation of precursor-conversion kinetics have been explored^19,22, simultaneously suppressing rapid early-stage growth and stabilizing the high-energy surfaces of ultrasmall FAPbI₃ QDs remains difficult. Consequently, reported FAPbI₃ QDs predominantly exhibit PL emission peaking at 750–800 nm, with only modest blueshifts relative to bulk FAPbI₃^19,23. The strong-confinement regime, where quantum confinement can fundamentally reshape the electronic structure and excitonic properties, remains largely unexplored in FAPbI₃ QDs.

In this study, we develop a coordination-mediated cold quenching strategy that enables the synthesis of ultrasmall FAPbI₃ QDs in the strong-confinement regime. The strategy relies on the direct injection of a precooled ethyl methacrylate (EMA) solution (–70 °C) into the reaction mixture, which effectively decouples nucleation from subsequent crystal growth. Meanwhile, the carbonyl groups of EMA strongly coordinate with undercoordinated Pb²⁺ sites, suppressing nanocrystal regrowth and providing efficient surface passivation. This approach affords precise size control from 6.4 to 10.1 nm, with the smallest QDs exhibiting a strongly blue-shifted emission at 696 nm, indicative of pronounced quantum confinement. Leveraging these ultrasmall QDs, we further demonstrate PeLEDs and a real-time optical communication prototype with fast response. More importantly, the realization of ultrasmall hybrid perovskite QDs in the strong-confinement regime creates a model platform for uncovering size-dependent photophysical phenomena and for advancing high-performance quantum photonic devices.

Results and discussion

Growth kinetics insights for FAPbI₃ quantum dots

As a reference, we first synthesized FAPbI₃ QDs via the conventional hot-injection method and quenched the reaction by a standard ice-water bath (0 °C, Supplementary Figs. 1a and 2). We employed in situ PL measurements to monitor the nucleation and growth dynamics of FAPbI₃ QDs (Supplementary Fig. 3), with the evolution of the PL spectra and emission peaks shown in Fig. 1a and d, respectively. Upon injecting the FA-oleate (FA-OA) precursor into the 1-octadecene solution containing PbI₂, oleic acid (OA), and oleylamine (OAm), the reaction mixture rapidly turned red within 1 s, indicating the formation of FAPbI₃ nanocrystals. Despite immediate transfer to an ice-water bath, the PL emission continuously redshifted from 646 to 750 nm over approximately 15 s (Fig. 1a and d), revealing substantial crystal growth during the cooling process. Furthermore, control experiments showed that FA-OA reacts readily with the PbI₂-OA-OAm complex even at room temperature (Supplementary Figs. 4 and 5), confirming the intrinsically low formation energy of FAPbI₃ and difficulties in preserving small particles at room temperature during purification and storage. These results demonstrate that conventional cooling cannot effectively terminate nanocrystal growth, allowing continuous growth and ripening during both cooling and post-treatment processes. Therefore, accessing the strong-confinement regime in FAPbI₃ QDs requires an ultrafast quenching strategy capable of arresting crystal growth immediately after nucleation.

Fig. 1. Mechanism of FAPbI₃ QD synthesis using different quenching strategies.

In situ photoluminescence (PL) spectra of FAPbI₃ QDs synthesized under different quenching conditions a ice‑water bath quenching (0 °C), b –70 °C ethyl methacrylate (EMA), and c –70 °C toluene cold quenching. Corresponding wavelength evolution profiles extracted from the in situ PL spectra d ice‑water bath quenching (0 °C), e –70 °C EMA, and f –70 °C toluene cold quenching. g Schematic illustration showing the key differences between the cold quenching and ice-water bath quenching methods.

To rapidly terminate the reaction and suppress subsequent regrowth, we developed a cold quenching method for synthesizing ultrasmall FAPbI₃ QDs. Under otherwise identical reaction conditions, the conventional ice-water bath quenching step was replaced by direct injection of a precooled ethyl methacrylate (EMA) solution at –70 °C after about 1 s of reaction (Supplementary Figs. 1b and 2). As shown in Fig. 1b and e, the PL emission of FAPbI₃ was fixed at 646 nm upon EMA injection, and no further redshift of the PL emission was observed thereafter, suggesting the successful preparation of ultrasmall FAPbI₃ QDs.

To decouple the effects of rapid cooling and the potential role of EMA in inhibiting QD regrowth, we employed precooled toluene ( −70 °C), a nonpolar and non-coordinating solvent, as a control quenching medium. Although cold toluene quenched the reaction more effectively than the conventional ice-water bath, the PL peak still gradually redshifted within 10 s (Fig. 1c and f). Moreover, the cold-toluene-quenched products degraded rapidly after isolation from the original solution. Therefore, these observations indicate that rapid cooling suppresses early-stage growth but is insufficient to stabilize the resulting QDs, whereas EMA additionally restricts further growth and enhances colloidal stability.

To investigate potential interaction between EMA and QDs, we first performed Fourier-transform infrared (FTIR) spectroscopy to characterize the prepared FAPbI₃ QDs, using neat EMA and control QDs as references (Supplementary Fig. 6). The vibration signatures associated with EMA, including the C=O stretching vibration of the ester group (1740 cm⁻¹), the C=C stretching vibration of the terminal vinyl group (1642 cm⁻¹), the C–O stretching vibration of the ethoxy moiety (1324 cm⁻¹), and the asymmetric C–O–C stretching vibration of the ester linkage (1169 cm⁻¹), remained detectable in the QDs after cycles of precipitation and redispersion²⁴.

We further performed ¹H nuclear magnetic resonance (¹H NMR) spectroscopy to examine the ligand environment (Supplementary Fig. 7). In addition to the signals of OA/OAm²⁵, the EMA-quenched FAPbI₃ QDs exhibited resonances at δ = 6.05, 5.16, and 3.92 ppm, assigned to the two nonequivalent vinyl protons and the ethoxy methylene protons of EMA, respectively. These EMA-related resonances are shifted relative to those of free EMA, consistent with their association with the QDs. Analysis of the integrated ¹H NMR signals gives an estimated molar ratio of EMA to the native OA/OAm ligands of approximately 1.5:1. This result indicates that EMA is a major component of the ligand shell, together with OA and OAm.

To further elucidate the interaction between EMA and the perovskite precursors, we investigated the evolution of the UV-Vis absorption spectra of iodoplumbate species in DMF by systematically varying the FAI:PbI₂ molar ratio from 10:1 to 100:1. Increasing the FAI content increases the concentration of iodide anions in the precursor solution, thereby promoting the formation of iodoplumbate complexes, including [PbI₃]⁻ and [PbI₄]²⁻, which exhibit characteristic absorption bands at approximately 370 and 423 nm, respectively. As shown in Supplementary Fig. 8, the intensities of both absorption bands gradually increased with increasing FAI:PbI₂ ratio, irrespective of the presence of EMA. However, at a given FAI:PbI₂ ratio, the absorption signals of the iodoplumbate species were consistently weaker in the EMA-containing solutions than in the corresponding EMA-free solutions. A substantially higher iodide content was required in the presence of EMA to achieve absorption intensities comparable to those observed without EMA. These results indicate a competitive coordination process, in which the carbonyl groups of EMA and iodide compete for coordination with Pb²⁺, thereby suppressing the formation of highly coordinated iodoplumbate species. This result is consistent with our FTIR measurements, in which the C=O stretching vibration of the ester group (1740 cm⁻¹) shifted to a lower wavenumber, confirming the coordination of the carbonyl groups with Pb²⁺.

Based on these results, we propose the growth mechanisms of FAPbI₃ QDs under different quenching conditions, which are illustrated in Fig. 1g. Upon the injection of precooled EMA ( −70 °C), the reaction temperature decreases rapidly, markedly slowing crystal growth. Meanwhile, the carbonyl groups of EMA coordinate with Pb-containing species in the reaction medium, thereby reducing precursor reactivity and retarding further QD growth. EMA molecules also remain adsorbed on the QD surface after purification, blocking potential growth sites and suppressing subsequent Ostwald ripening. These combined effects enable the formation of ultrasmall FAPbI₃ QDs. By contrast, rapid cooling alone is insufficient to fully arrest QD growth, as continuous PL redshifts are still observed in the cold-toluene-quenched system, although to a lesser extent than in the conventional ice-water-bath-quenched system.

Guided by this mechanism, we further investigated whether other ester-based molecules could serve as coordination-mediated cold-quenching media. In addition to EMA, we examined five structurally related solvents: methyl methacrylate, butyl methacrylate, ethyl isobutyrate, isopropyl propionate, and ethyl acetate (Supplementary Fig. 9). Because effective cold quenching requires both rapid heat removal and suitable chemical interactions with Pb²⁺ species, the candidate solvents were evaluated according to their melting points, low-temperature fluidity, molecular structures, polarity, miscibility with the reaction medium, and compatibility with the resulting QD colloids. In particular, the solvent must remain sufficiently fluid at the quenching temperature to allow rapid and homogeneous mixing. Most of the tested solvents produced distinctly blue-shifted, relatively symmetric PL emission bands relative to QDs prepared by conventional ice-water-bath quenching under otherwise identical conditions (Supplementary Fig. 9), indicating that the growth-suppression effect is not unique to EMA. However, the PL peak position, spectral symmetry, and colloidal stability varied substantially among the resulting QDs, reflecting the combined effects of cold quenching, carbonyl-Pb²⁺ interactions, solvent polarity, and molecular structure. Notably, ethyl acetate yielded a broad PL spectrum with multiple emission features. This behavior likely arises from its relatively high polarity, which may destabilize a fraction of the QDs, particularly the smaller ones. Among the tested solvents, EMA gives rise to the smallest QDs. More broadly, these results suggest that coordination-mediated quenching media can be tailored to the crystallization kinetics and surface chemistry of different perovskite compositions. We believe further molecular design and optimization may enable stronger quantum confinement, improved size uniformity and colloidal stability, as well as enhanced optoelectronic properties for a broader range of applications.

Broad size tunability and photophysical properties

Based on the coordination-mediated cold quenching method with EMA, we next synthesized a series of FAPbI₃ QDs with different sizes by varying the reaction temperature and the FA:Pb precursor ratio. Transmission electron microscopy (TEM) images reveal that the QD diameter can be tuned from 6.4 to 10.1 nm (Fig. 2a−l). These values are smaller than the reported exciton Bohr diameters of FAPbI₃ (approximately 12.3−13.4 nm)^3,26–28, indicating appreciable quantum confinement across the QD series, with progressively stronger confinement as the QD size decreases. Statistical analysis of the TEM images confirms a relatively low particle-size standard deviation, indicating excellent monodispersity and narrow particle size distribution compared with the QDs prepared by the conventional ice-water bath (Supplementary Figs. 10 and 11). These results align with the evolution of peak broadening from X-ray diffraction (XRD) patterns, further confirming the reduced crystal size²⁰ (Supplementary Fig. 12).

Fig. 2. Structural and optical characterizations of FAPbI₃ QDs via EMA cold quenching at different reaction temperatures.

Transmission electron microscopy (TEM) images of FAPbI₃ QDs synthesized at a 60 °C, b 65 °C, c 70 °C, d 75 °C, e 80 °C, f 90 °C. Scale bars: 100 nm. g–l Corresponding particle-size distribution histograms for the QDs shown in (a–f), with average diameters of 6.4 ± 1.0 nm (g), 7.0 ± 0.9 nm (h), 7.5 ± 1.2 nm (i), 8.5 ± 1.6 nm (j), 9.5 ± 1.9 nm (k), 10.1 ± 1.4 nm (l). m Normalized PL spectra. n Normalized absorbance spectra. o Size-dependent exciton binding energy (E_b) of the FAPbI₃ QDs, fitted using the Elliott formula. p Band gap (E_g) as a function of diameter for FAPbI₃ QDs; the gray solid line represents the theoretical fitting curve^33,34(using a simplified effective-mass-approximation model: $E_g=E_{g,bulk}+\frac{{2\hslash }^2{\pi }^2}{m{d}^2}$), and the red dotted line marks the exciton Bohr diameter of FAPbI₃ QDs^3,26–28.

Having achieved the size tunability of FAPbI₃ QDs, we proceeded to investigate their size-dependent photophysical properties. As shown in Fig. 2m and Supplementary Figs. 13 and 14, with decreasing QD size, the PL emission peak exhibits a continuous blueshift from 759 nm to 696 nm, while the band gap (E_g) widens from 1.63 eV to 1.80 eV, as determined by the corresponding absorption onset of UV-Vis absorption spectra (Fig. 2n). Notably, the purified QDs exhibit redshifted PL emission relative to the in situ PL recorded immediately after EMA injection (Fig. 1e). This difference may arise from changes in the dielectric environment upon solvent exchange and/or limited Ostwald ripening during the purification process^29,30. Based on the absorption features, the exciton binding energies (E_b) were extracted by fitting the near-band-edge absorption spectra using the Elliott model³¹ (see methods). As shown in Fig. 2o, smaller QDs exhibit larger E_b (up to 144 meV), which gradually decreases with increasing particle size. This trend reflects the enhanced electron-hole Coulomb interaction under stronger spatial confinement, highlighting the size-dependent nature of the quantum confinement effect³².

To quantitatively assess the quantum confinement effect, we employed a simplified effective-mass approximation to estimate the size-dependent bandgap^33,34:

$$E_g=E_{g,bulk}+\frac{{2\hslash }^2{\pi }^2}{m{d}^2}$$ 1

where $E_{g,bulk}$ is the bulk bandgap of FAPbI₃, d represents the QD diameter, ħ is the reduced Planck constant, and m* is the exciton reduced mass. The optical band gaps extracted from the UV-Vis absorption spectra (Fig. 2n) for FAPbI₃ QDs of different sizes are plotted against QD size in Fig. 2p. The E_g evolution of FAPbI₃ QDs agrees well with the theoretical predictions, reflecting the expected weakening of quantum confinement and the accompanying convergence of the band gap toward the bulk value as the QD size increases³³. Additionally, the experimentally determined E_g values are slightly smaller than the theoretical predictions, which may arise from dielectric screening in the perovskite lattice and sub-bandgap states that are not explicitly included in the model.

We next investigated the size-dependent photophysical properties of the FAPbI₃ QDs using time-resolved photoluminescence (TRPL) and fluence-dependent photoluminescence quantum yield (PLQY) measurements. Overall, the PL lifetime of FAPbI₃ QDs decreased from 50.6 to 18.6 ns as the size decreased from 10.1 to 6.4 nm (Fig. 3a–f and Supplementary Fig. 15). The 10.1 nm QDs exhibit a biexponential decay profile, whereas the decay becomes monoexponential as the QD size decreases to 9.5 nm and below. This transition is consistent with progressively stronger quantum confinement and the increasing dominance of monomolecular excitonic recombination³⁵ (Fig. 3a–f and Supplementary Tables 1 and 2). Concurrently, both solutions and films retained a high PLQY exceeding 80% even at low excitation densities (10 mW cm⁻²) (Fig. 3g and Supplementary Fig. 17). These results are consistent with enhanced exciton binding and exciton localization under strong quantum confinement.

Fig. 3. Optical properties of FAPbI₃ QDs synthesized by the EMA cold quenching method.

Time resolved photoluminescence (TRPL) decay of FAPbI₃ QDs synthesized at different temperatures, a 60 °C (6.4 ± 1.0 nm), b 65 °C (7.0 ± 0.9 nm), c 70 °C (7.5 ± 1.2 nm), d 75 °C (8.5 ± 1.6 nm), e 80 °C (9.5 ± 1.9 nm), f 90 °C (10.1 ± 1.4 nm). g Excitation fluence-dependent photoluminescence quantum yield (PLQY) of solution and film samples with PL emission at 712 nm. h Urbach energy (E_U) of FAPbI₃ QDs with various sizes. i PL full width at half maximum (FWHM) as a function of emission wavelength for FAPbI₃ QDs in this work and reported articles^19,42–48. Corresponding details were listed in Supplementary Table 3.

To further evaluate the band-edge characteristics of the strongly quantum-confined QDs, we extracted the Urbach energy (E_U) from the steady-state absorption spectra (Fig. 2n). As shown in Fig. 3h and Supplementary Fig. 16, E_U increases approximately linearly with decreasing particle size. Given that QDs of different sizes exhibit comparable PLQYs (Supplementary Figs. 17–18), the increase in E_U for smaller QDs is unlikely to originate primarily from sub-bandgap states. Instead, it may arise from concomitant size and surface-dependent effects, including ligand-induced structural distortion at surfaces, greater heterogeneity of the local dielectric environment, and size-distribution-induced inhomogeneous electronic states^36,37. As summarized in Fig. 3i, the smallest FAPbI₃ QDs with emission peak wavelengths at 696 nm show a full width at half maximum (FWHM) of ~56 nm. Despite their ultrasmall particle size and relatively high E_U, the emission linewidth remains comparable to those of previously reported FAPbI₃ QDs, indicating that the degree of inhomogeneous broadening remains within a reasonable range.

To evaluate the generality of our coordination-mediated cold-quenching strategy, we first extended it to methylammonium lead triiodide (MAPbI₃), another organic–inorganic hybrid lead iodide perovskite. By combining precooled EMA injection with adjustment of the reaction temperature, the PL emission of MAPbI₃ QDs was continuously tuned from 752 to 717 nm (Supplementary Fig. 19a–c). TEM characterization confirms an overall reduction in QD size under the cold-quenching conditions (Supplementary Fig. 19d–g), although increased size heterogeneity is observed for the smaller MAPbI₃ QDs. We then examined the compositional applicability of the strategy to all-inorganic CsPbI₃ and bromide-based FAPbBr₃ QDs. The shifts in their absorption and PL spectra with EMA cold quenching are consistent with suppressed QD growth, suggesting that the strategy can be extended across different A-site cations and halide compositions (Supplementary Fig. 20). Further optimization of the ligands, precursor composition, and reaction temperature is expected to improve size uniformity and overall nanocrystal quality while enabling stronger quantum confinement. These results demonstrate the applicability of the strategy to a broader range of lead halide perovskite QDs.

Device performance of strongly quantum-confined FAPbI₃ QDs

Having established the strong quantum-confinement characteristics and high photoluminescence quality of the FAPbI₃ QDs, we next evaluated their optoelectronic quality and application potential by fabricating PeLEDs. As illustrated in Fig. 4a, the devices adopt a multilayer architecture of indium tin oxide (ITO)/ poly(3,4-ethylenedioxythiophene):poly(styrene sulfonate) (PEDOT:PSS)/ poly[bis(4-phenyl) (2,4,6-trimethylphenyl) amine] (PTAA)/FAPbI₃ QDs/1,3,5-tris(1-phenyl-1H-benzimidazol-2-yl)benzene (TPBi)/ 2,4,6-tris[3-(diphenylphosphinyl)phenyl]−1,3,5-triazine (POT2T)/ lithium fluoride (LiF)/ aluminum (Al). A representative PeLED with electroluminescence (EL) centered at 715 nm was selected for detailed characterization. As shown in Fig. 4b, the optimized PeLED exhibits typical current density-voltage-radiance (J-V-R) characteristics, with a turn-on voltage of 2.1 V and a peak radiance of 24.8 W sr⁻¹ m⁻². No EL spectral shift is visible during the voltage sweep from 3.0 to 8.0 V (Supplementary Fig. 21). The corresponding current density-external quantum efficiency (J-EQE) curve yields a maximum external quantum efficiency (EQE) of 16.3% (Fig. 4c). Statistical analysis of 30 devices demonstrates excellent performance consistency, with an average EQE of 14.3% (Fig. 4d), confirming the high reproducibility of the fabrication process. Furthermore, operational lifetime measurements conducted under a constant current density of 1.0 mA cm⁻² in a N₂-filled glovebox (Supplementary Fig. 22) revealed a half-life (T₅₀) of approximately 120 min.

Fig. 4. Performance of FAPbI₃ QDs-based devices.

a Device structure of PeLEDs. b Current density-voltage-radiance (J-V-R) curves. c External quantum efficiency-current density curves (EQE-J) and d Statistical distribution of peak EQE of the PeLEDs with electroluminescence centered at 715 nm. e EL spectra of PeLEDs based on FAPbI₃ QDs with different sizes. f Reported peak EQE and EL wavelength of PeLEDs based on colloidal FAPbI₃ QDs^{23,38,41,49–57}. Corresponding performances of devices were listed in Supplementary Table 4. g Transient electroluminescence (trEL) curves of PeLEDs with different EL emission wavelengths under pulsed operation (T = 10 μs, V_on = 4.0 V; device area = 1.0 mm²). h Optical communication responses of the PeLEDs: the upper trace shows the input electrical pulses encoding the character sequence NPUIFE, and the lower trace shows the corresponding output recorded by the photodetector.

We further examined the electroluminescence characteristics of PeLEDs fabricated using FAPbI₃ QDs with different particle sizes (Supplementary Fig. 23). As shown in Fig. 4e, the EL peak progressively redshifts from 697 to 732 nm as the QD size increases (collected at 6.0 V in all cases), consistent with the size-dependent PL evolution of the corresponding QDs. This result confirms that the emission wavelength of the devices can be modulated through controlled adjustment of the QD size. Figure 4f compares the EL wavelengths and maximum EQEs of the present devices with those reported for colloidal FAPbI₃ QD-based PeLEDs. Although several previously reported devices exhibit higher EQEs at longer wavelengths, the present devices extend the accessible EL range toward shorter wavelengths while maintaining competitive efficiencies^23,38. These results demonstrate that strongly quantum-confined FAPbI₃ QDs provide an effective route to wavelength-tunable electroluminescence.

Motivated by the accelerated recombination dynamics of strongly quantum-confined FAPbI₃ QDs and their potential for fast optical communication, we investigated the transient electroluminescence (trEL) kinetics of the corresponding PeLEDs. To obtain stable trEL responses, the devices were driven using continuous square-wave voltage pulses with fixed amplitudes (V_on) and pulse widths (T), while the pulse intervals were varied³⁹ (Supplementary Fig. 24). As shown in Fig. 4g, devices with shorter EL emission wavelengths exhibited shorter T_rise, a trend consistent with TRPL measurements⁴⁰. This correlation indicates that exciton radiative recombination accelerates under strong quantum confinement, leading to faster transient responses.

We then constructed a proof-of-concept optical communication system consisting of a programmable signal generator, a PeLED transmitter, a photodetector, and an oscilloscope (Supplementary Fig. 25). Character information was encoded into 8-bit binary sequences. As shown in Fig. 4h, the upper trace represents the input electrical signal, whereas the lower trace corresponds to the signal recorded by the photodetector. The detected output closely follows the input modulation with clearly distinguishable waveforms, demonstrating the feasibility of using strongly quantum-confined FAPbI₃ QD-based PeLEDs for fast optical communication.

In summary, we have developed a coordination-mediated cold-quenching strategy to overcome the kinetic barriers associated with synthesizing ultrasmall, strongly quantum-confined FAPbI₃ QDs, thereby achieving tunable quantum confinement, precise size control, and optimized photoluminescence. By leveraging the synergistic effects of ultra-low temperature quenching and the carbonyl-group coordination of EMA with Pb-containing species, this approach effectively decouples nucleation from growth and passivates surface defects, thereby retarding Ostwald ripening and enabling the precise fabrication of monodisperse nanocrystals as small as 6.4 nm. The resulting QDs exhibit pronounced quantum confinement, enhanced exciton binding energy, and near-unity photoluminescence quantum yield. As the emissive layer, these QDs yield PeLEDs with a peak EQE of 16.3% and an emission wavelength of 715 nm. Furthermore, the accelerated radiative recombination kinetics inherent to these strongly confined systems facilitate fast optoelectronic responses, demonstrating their capability in proof-of-concept optical communication. This work not only provides a robust and generalizable synthetic route for accessing ultrasmall hybrid perovskite nanocrystals but also establishes a critical link between quantum confinement engineering and dynamic device performance. These findings highlight the potential of strongly confined perovskite QDs for advanced optoelectronic applications requiring fast emission dynamics.

Methods

Materials

Formamidine acetate (FAAc, 99%), oleic acid (OA, 90%), N,N-dimethylformamide (DMF, 99.8%), chlorobenzene (CB, 99.5%), cesium carbonate (Cs₂CO₃, 99.9%), and lithium fluoride (LiF, 99.99%) were acquired from Sigma-Aldrich. Dimethyl carbonate (DC, 99%) was acquired from Beijing InnoChem Science & Technology Co., Ltd. Isopropanol (IPA, 99.8%) was purchased from Thermo Fisher Scientific. Oleylamine (OAm, 80–90%), ethyl methacrylate (EMA, 99%), octadecene (ODE, 90%), methyl acetate (MeOAc, 99%), n-octane (C₈H₁₈, 99%), anhydrous n-hexane (C₆H₁₄, ≥95%), lead bromide (PbBr₂, 99.0%), toluene-d8 (C₆D₅CD₃, ≥99.6, atom% D), methyl methacrylate (C₅H₈O₂, 99%), butyl methacrylate (C₈H₁₄O₂, ≥97%), ethyl isobutyrate (C₆H₁₂O₂, ≥99%), isopropyl propionate (C₆H₁₂O₂, ≥97%), ethyl acetate (C₄H₈O₂, ≥97%) and lead iodide (PbI₂, 99.99%) were sourced from Aladdin. Toluene (C₇H₈, 99%) was obtained from Sinopharm Chemical Reagent Co., Ltd. Poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS, Clevios P VP Al 4083) and poly[bis(4-phenyl) (2,4,6-trimethylphenyl) amine] (PTAA) were acquired from Xi’an Polymer Light Technology Corp. 1,3,5-Tris(1-phenyl-1H-benzimidazol-2-yl) benzene (TPBi, 98%) and 2,4,6-Tris[3-(diphenylphosphinyl)phenyl]−1,3,5-triazine (POT2T, 99%) were obtained from Luminescence Technology Corp. Aluminum (Al, 99.9%) was sourced from ZhongNuo Advanced Material (Beijing) Technology Co., Ltd. Anhydrous acetonitrile (C₂H₃N, 99.9%) was obtained from Macklin. Methylammonium acetate (MAAc, 97%) was purchased from TCI. Dry ice was purchased from Apinno. All chemicals were used as received without further purification.

Precooled ethyl methacrylate

A low-temperature cooling bath was prepared by mixing an adequate amount of dry ice with ethanol for approximately 10 min. Subsequently, 20 mL of EMA was transferred into a 50 mL centrifuge tube. The centrifuge tube was then immersed in the cooling bath, and the temperature of EMA was monitored in real time using a thermometer. Once the temperature stabilized at –70 °C, the precooled EMA was used for subsequent injections.

Preparation of the FA-OA precursor

FAAc (2.53 mmol, 263.4 mg), oleic acid (OA, 8 mL), and 1-octadecene (ODE, 12 mL) were loaded into a 50 mL three-necked flask and degassed under vacuum at 120 °C for 1 h with continuous stirring. Upon complete dissolution of FAAc, the mixture was cooled to 60 °C under a nitrogen atmosphere for further use.

Synthesis of FAPbI₃ QDs with tunable sizes

In a typical synthesis, PbI₂ (0.133 mmol, 61.1 mg) was loaded into a three-necked flask containing 4.6 mL of ODE. The mixture was degassed under vacuum at 120 °C for approximately 1 h. Subsequently, under a nitrogen atmosphere, 1.0 mL of pre-degassed OA and 0.5 mL of OAm were injected sequentially into the flask. Upon complete dissolution of PbI₂, the mixture was cooled to the predetermined reaction temperature. Then, the FA-OA precursor with a designated FA:Pb molar ratio was rapidly injected into the reaction mixture, and the solution immediately turned from transparent yellow to deep red. After 1 s, the reaction was quenched via different protocols to yield a crude FAPbI₃ QD solution.

With the reaction temperature fixed at 60 °C, the PL emission peaks of the resulting products were 694, 699, 704, 708, and 718 nm at FA:Pb molar ratios of 3.82, 4.30, 4.77, 5.25, and 5.73, respectively.

With the FA:Pb molar ratio fixed at 3.82, the PL emission peaks of the resulting products were determined to be 696, 704, 712, 725, 738, and 759 nm at reaction temperatures of 60, 65, 70, 75, 80, and 90 °C, respectively.

Different quenching methods

For the ice-water bath quenching, the flask was immersed in an ice-water bath (0 °C) after 1 s to yield ice-water bath-quenched FAPbI₃ QDs.

For the cold quenching, 5 mL of pre-cooled EMA (–70 °C) was directly injected into the reaction mixture after 1 s to afford EMA quenched FAPbI₃ QDs.

Purification of FAPbI₃ QDs

The crude solution was mixed with MeOAc at a volume ratio of 3:4 and centrifuged at 15,696 × g (where g is the acceleration due to gravity) for 5 min. After discarding the supernatant, the precipitate was redispersed in 2 mL of toluene, followed by reprecipitation with dimethyl carbonate at a volume ratio of 1:3 and a second centrifugation at 15,696 × g for another 5 min to remove excess ligands and residual solvents. The resulting precipitate was dispersed in 0.2 mL of n-octane, yielding a dispersion with an optical density (OD) of 40 at ~400 nm. The final QD dispersion was filtered through a syringe PTFE filter and stored at 4 °C for further use.

Synthesis of CsPbI₃ QDs

Cs₂CO₃ (0.153 mmol, 50 mg), OA (0.5 mL), and ODE (5 mL) were loaded into a three-necked flask and degassed under vacuum at 120 °C for 1 h. After complete dissolution of Cs₂CO₃, the mixture was cooled to 80 °C under an N₂ atmosphere to obtain the Cs-OA precursor. Separately, PbI₂ (0.217 mmol, 100 mg), OA (1 mL), OAm (1 mL), and ODE (6 mL) were added to another three-necked flask and degassed at 120 °C for 1 h. Then the temperature was raised to 160 °C under an N₂ atmosphere. After the temperature stabilized for 10 min, 2 mL of the Cs-OA precursor was rapidly injected. After approximately 1 s, the reaction was immediately quenched by injecting 5 mL of precooled EMA or by placing the flask in an ice-water bath. The crude solution was mixed with MeOAc at a 1:2 volume ratio and centrifuged at 10,900 × g for 3 min. The supernatant was discarded, and the precipitate was redispersed in 1 mL of n-hexane, followed by centrifugation at 1744 × g for 2 min. The supernatant was collected and stored for further characterization.

Synthesis of FAPbBr₃ QDs

FAAc (1.25 mmol, 130 mg) and OA (5 mL) were loaded into a three-necked flask and degassed under vacuum at 100 °C for 30 min. After FAAc was fully dissolved, the solution was cooled to 60 °C under an N₂ atmosphere to obtain the FA-OA precursor. Separately, PbBr₂ (0.207 mmol, 75.9 mg), OA (1.2 mL), OAm (0.6 mL), and ODE (5.0 mL) were added to another three-necked flask and degassed at 120 °C for 1 h until PbBr₂ was fully dissolved. The temperature was then raised to 160 °C under an N₂ atmosphere. After stabilization, 2.8 mL of the FA-OA precursor was rapidly injected. Approximately 1 s later, the reaction was immediately quenched by injecting different volumes of precooled EMA or by placing the flask in an ice-water bath. For purification, the crude QD solution was diluted with 5 mL of toluene, followed by the addition of 2.5 mL of anhydrous acetonitrile. The mixture was centrifuged at 8829 × g for 3 min. The obtained precipitate of FAPbBr₃ QDs was redispersed in 2 mL of toluene and centrifuged again at 1744 × g for 2 min. The supernatant was collected and stored for further characterization.

Synthesis of MAPbI₃ QDs

MAAc (2.61 mmol, 238 mg) and OA (5 mL) were loaded into a three-necked flask and degassed under vacuum at 100 °C for 1 h. After complete dissolution of MAAc, the solution was cooled to 60 °C under an N₂ atmosphere to obtain the MA-OA precursor. Separately, PbI₂ (0.191 mmol, 88 mg), OA (1.55 mL), OAm (0.75 mL), and ODE (5 mL) were added to another three-necked flask and degassed at 100 °C for 1 h. The mixture was then cooled to 80, 60, or 50 °C under an N₂ atmosphere. Subsequently, 2.6 mL of the MA-OA precursor was rapidly injected. After approximately 1 s, the reaction was immediately quenched by injecting 5 mL of precooled EMA or by placing the flask in an ice-water bath. For purification, the crude QD solution was mixed with MeOAc at a 3:4 volume ratio and centrifuged at 15696 × g for 3 min. The supernatant was discarded, and the precipitate was redispersed in 2 mL of toluene. The purified QD dispersion was collected for further characterization.

Cold quenching with other ester-based solvents

A low-temperature cooling bath was prepared by mixing dry ice with ethanol and allowing the mixture to equilibrate for approximately 10 min. For butyl methacrylate, ethyl isobutyrate, isopropyl propionate, and ethyl acetate, 20 mL of each solvent was transferred into a separate 50 mL centrifuge tube. The tubes were immersed in the cooling bath, and the solvent temperature was monitored in real time with a thermometer. Once the temperature stabilized at −70 °C, the precooled solvents were used for subsequent injections. For methyl methacrylate, the same procedure was followed, except that the solvent was cooled to −45 °C before use, owing to its higher freezing point, to prevent solidification.

Fabrication of PeLEDs

Patterned ITO substrates were sequentially subjected to ultrasonic cleaning with isopropanol and ethanol (30 min each), followed by rinsing with deionized water. After blow drying with N₂, the substrates were subjected to a 15 min ultraviolet ozone treatment. Subsequently, 70 μL of a mixture of PEDOT:PSS and IPA (V_IPA:V_PEDOT:PSS = 2:1) was spin-coated onto the pre-treated substrates at 5,000 rpm for 30 s, followed by annealing in ambient atmosphere at 140 °C for 15 min. The substrates were then transferred into an N₂ filled glove box, where 30 μL of PTAA solution (4 mg/mL in chlorobenzene) was spin-coated at 2,000 rpm for 30 s and annealed at 120 °C for 10 min. Next, 30 μL of FAPbI₃ QD solution was spin-coated onto the PTAA film, followed by annealing at 60 °C for 10 min to eliminate residual solvent. Finally, TPBi (5 nm, ∼0.5 Å/s), POT2T (72 nm, 1.5 Å/s), LiF (1 nm, ∼0.1 Å/s), and Al (110 nm, ∼3 Å/s) were sequentially deposited onto the QD film via thermal evaporation under high vacuum. In this work, the active area of the fabricated devices was 7.25 mm².

Material characterizations

Transmission electron microscopy (TEM) images were acquired using a Hitachi HT7800 microscope operated at an accelerating voltage at 100 kV. X-ray diffraction (XRD) measurements were conducted on a Bruker D8 Advanced diffractometer with Cu Kα radiation (λ = 1.5406 Å). The Fourier transform infrared (FTIR) spectrum of the QD surface ligands was recorded on a FTIR spectrometer (Bruker Tensor II). Steady-state absorption spectra were recorded using a MAPADA UV-670 spectrophotometer. Photoluminescence (PL) spectra were measured using a home-built system comprising a QE Pro spectrometer and a 450 nm laser excitation source, while fluence-dependent photoluminescence quantum yield (PLQY) was measured using the same system coupled with an integrating sphere. Transient photoluminescence (TRPL) decay dynamics were investigated via time-correlated single-photon counting (TCSPC) using an Edinburgh FLS1000 spectrometer with an excitation wavelength of 450 nm. The PL decay curves were fitted using monoexponential or biexponential functions. QD samples for nuclear magnetic resonance (NMR) measurements were prepared by redissolving the dried QD powder in deuterated toluene. ¹H NMR spectra were recorded on a Bruker ADVANCE NEO 500 spectrometer at room temperature.

In situ photoluminescence (PL) measurements

The synthesis process of FAPbI₃ QDs was monitored in real time using a home-built in situ PL measurement system. A continuous-wave 380 nm semiconductor laser (power density: ~500 mW/cm²) was employed as the excitation light source. The output laser beam was first directed through a 400 nm short-pass filter (Thorlabs, FES0400) to eliminate long-wavelength stray light components, thereby ensuring the spectral purity of the excitation light. The purified excitation light was then coupled into a three-neck reaction flask via a fiber collimator to irradiate the reaction solution and excite the PL signal. To effectively suppress signal interference arising from the transmitted excitation light, a side detection configuration was adopted in the optical setup. Specifically, a collection optical fiber was positioned perpendicular to the direction of the incident excitation beam to capture the fluorescence emitted from the solution. The collected fluorescence signal was then delivered to a QE Pro high-sensitivity spectrometer for real-time spectral acquisition throughout the entire synthesis process. The acquired spectral data were simultaneously transmitted to a computer for data storage and subsequent analysis. Throughout the in situ measurement, the reaction flask was maintained under constant heating and stirring. In addition, all ambient light sources in the test room were turned off to minimize the interference of stray light on the spectral signals.

Exciton binding energy (E_b) calculation

The Elliott formula was utilized to fit the absorption spectrum. According to this formula, the relationship between the absorption coefficient and photon energy is described by the following equation³¹ (2):

$$\alpha \left(\hslash \omega \right)\propto \left[\sum _i\frac{{2E}_b}{i^3}sech\left(\frac{\hslash \omega -\left(E_g-\frac{E_b}{i^2}\right)}{\mathrm{\Gamma }}\right)+A\times arctan\left(sinh\left(\frac{\hslash \omega -E_g}{\mathrm{\Gamma }}\right)\right){\left(1-exp\left(-2\pi {\left(\frac{E_b}{∣\hslash \omega -E_g∣}\right)}^{0.5}\right)\right)}^{-1}{\left(1-B\left(\hslash \omega -E_g\right)\right)}^{-1}\right]$$ 2

Where E_b and E_g are defined as the exciton binding energy and bandgap of the FAPbI₃ QDs, respectively. Γ denotes the spectral broadening. A and B are constants. The first term of the equation represents the exciton contribution, while the second term corresponds to the continuum contribution. Due to the distinct dependence of the two terms on E_b, the value of E_b can be determined by fitting the absorption spectrum near the band edge.

Optical bandgap energy (E_g) calculation

The optical bandgap (E_g) of FAPbI₃ was determined from its absorption spectrum using the Tauc plot method⁴¹:

$${\left(\alpha \hslash v\right)}^2=A\left(\hslash v-E_g\right)$$ 3

Where, α is the absorption coefficient. $\hslash$ is Planck’s constant. ν is the frequency of the incident photon, A is a constant.

Electrostatic potential (ESP) calculation

The ESP of EMA was calculated using density functional theory (DFT) as implemented in the DMol³ module of Materials Studio (BIOVIA, Dassault Systèmes). The molecular geometry was fully optimized without symmetry constraints using the generalized gradient approximation (GGA) with the Becke-Lee-Yang-Parr (BLYP) exchange-correlation functional. The double numerical basis set with polarization functions (DNP) was employed, along with a fine integration grid, to ensure numerical accuracy. Van der Waals interactions were accounted for by incorporating Grimme’s dispersion correction (DFT-D2). Following the optimization, the ESP was computed on the van der Waals (vdW) surface and mapped onto an electron density isosurface to visualize the charge distribution. The ESP isosurface was rendered with a color scale ranging from –0.06 to +0.035 a.u., where red denotes electron-rich regions and blue denotes electron-deficient regions. All visualizations were generated using the Visualizer module of Materials Studio.

Device characterizations

The PeLEDs were measured in the dark within an inert gas glovebox. The current density-voltage (J-V) characteristics, electroluminescence (EL) spectra, and external quantum efficiency (EQE) of the devices were systematically monitored utilizing a computer-controlled Keithley 2400 source meter, an Ocean Optics QE-Pro spectrometer, and an integrating sphere. During the testing procedure, the perovskite light-emitting diode (PeLED) devices were positioned within the measurement system, and a bias voltage was applied to elicit electroluminescence. The emitted light was subsequently collected by the integrating sphere. Operational stability measurement was conducted in a N₂-filled glovebox (O₂ < 0.1 ppm, H₂O < 0.1 ppm) at room temperature, using a multi-channel LED lifetime testing system (Litos) under a constant current density of 1.0 mA cm⁻².

Author contributions

H.Y. and X.W. contributed equally to this work. W.X. and G.C. conceived the idea and supervised the project. H.Y., X.W., and G.C. conducted the main experiments under the supervision of W.X. H.Y. and X.W. performed the main characterizations and analyzed the data. H.Y. and D.Z. carried out the transient electroluminescence measurements and constructed the real-time optical communication system under the supervision of W.X. and G.C. The TRPL measurements were performed by G.L. and H.Y., while the excitation-fluence-dependent PLQY measurements were conducted by J.B., X.W. and Y.Z. G.C. and H.Y. wrote the manuscript; W.X., F.Z., Y.Z. and E.D. provided revisions. All authors discussed the results and reviewed the manuscript.

Peer review

Peer review information

Nature Communications thanks Xingliang Dai, Tae-Woo Lee, and other anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was supported by the National Natural Science Foundation of China (Grants 52250060, 62274135, 62288102, 52302167, 62505250), the Fundamental Research Funds for the Central Universities (D5000260051), and the Key Project of Ningbo Natural Science Foundation (Grant 2022J056).

Data availability

The authors declare that the main data supporting the findings of this study are available within the article and its supplementary information files, and all data are available from the corresponding authors upon request. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-77849-7.