Molecule-induced ripening control in perovskite quantum dots for efficient and stable light-emitting diodes

Abstract

Perovskite quantum dots (QDs) show an excellent application perspective in semiconductor optoelectronic devices. However, problems of ligand loss during the growth, purification, film formation, and storage process always induce the aggregation and ripening of QDs, adversely affecting QDs’ and QD-based devices’ performance. Here, we use a bidentate molecule to control ripening toward a notable performance boost in CsPbI₃ QDs. The strong interaction between QDs and the bidentate molecules maintains stable surface states of QDs, inhibiting QDs’ undesirable ripening and generation of defects. We fabricate QD-based light-emitting diodes (LEDs) with a maximum external quantum efficiency (EQE) of 26.0% at 686 nm and an operating half-life of 10,587 hours at an initial radiance of 190 mW sr⁻¹ m⁻² (equivalent to a luminance of 100 cd m⁻² for green perovskite LEDs). Benefiting from the high storability of the target QDs, the as-fabricated devices based on the QD solution storing for 1 month show a maximum EQE of 21.7% (20.3% for 3 months).

A molecule-induced ripening control strategy enables perovskite QLEDs with a maximum peak EQE of 26.0% and a T₅₀ of 10,587 hours.

INTRODUCTION

Perovskite quantum dots (QDs) show the virtue of solution processability, narrow emission spectra, and strong quantum confinement, thus holding the immense potential to be promoted for low-cost, high–color purity, and high-efficiency light-emitting diodes (LEDs) (1–5). Perovskite quantum dot LEDs (QLEDs) have been booming in recent years, and their external quantum efficiencies (EQEs) have been markedly improved to more than 20%, reaching the commercial application threshold (2, 3, 6–9). However, compared with the conventional counterparts of II-VI and III-V QDs, perovskite QDs are more likely to suffer from chemical and optical degradation, leading to an operating lifetime of only tens or hundreds of hours, which is regarded as uncertainty and risk for further commercialization (10–13).

QDs have quite a large surface-to-volume ratio due to their small particle size, resulting in considerable surface energy and high activity (14). Thus, QDs need ligands binding to the surface to retain the colloidal stability. However, different from the conventional II-VI and III-V QDs (11, 15), the ionic interaction between ligands and the surface of perovskite QDs makes the highly dynamic and labile ligands easily fall out during the QDs’ growth, purification, film formation, and storage process, resulting in insufficient surface atom coordination and increased unsaturated and dangling bonds (16–18). Then, the uncoordinated atoms on the surface of QDs are prone to bind to other atoms, leading to the aggregation or Oswald ripening of QDs and generation of various defects, further affecting the luminescent properties and stability of QDs (19). To solve this problem, researchers have been seeking approaches to maintaining the stability of QDs’ surface and to inhibiting the aggregation and ripening of QDs, thereby developing high-performance perovskite QDs and further unleashing the potential of perovskite QLEDs in practical applications (7, 20).

In this study, we design a molecule-induced ripening control strategy by using a bidentate molecule 2-(1H-pyrazol-1-yl)pyridine (PZPY) to enhance the photoelectronic properties and stability of QDs. PZPY shows the small size and molecular flexibility originating from the C–N bond of pyrazole, so it could avoid space steric hindrance and easily attach to the surface of QD by twisting the structure, interacting with uncoordinated Pb²⁺. Thus, the surface energy of perovskite QDs is decreased, and QDs are discontinued from further ripening or growing up. The aberration-corrected scanning transmission electron microscopy (STEM) images present that target QDs have better morphology, enhanced stability, and fewer defects. Chemical composition and optical property information evidence that a strong interaction between QDs surface and PZPY molecules enables QDs with enhanced photoluminescence quantum yields (PLQYs) of 94% and extended photoluminescence (PL) lifetime. As a result, the high-quality, high-efficiency, and stable QD layers contribute to QLEDs with a maximum EQE of 26.0% and an operating half-life (T₅₀, extrapolated) of over 10⁵ hours at an initial radiance of 190 mW sr⁻¹ m⁻², representing one of the best performances of perovskite LEDs. Benefiting from the high storability of target QDs, the as-fabricated QLEDs based on QD solution after being stored for 1 month show a maximum EQE of 21.7%, and the ones based on QD solution after being stored for 3 months show a maximum EQE of 20.3%. These results demonstrate that the proposed ripening control strategy by introducing bidentate molecules for highly stable and efficient QLEDs has a specific practical value and application prospect in future high-definition displays and biomedical treatments.

RESULTS

Ligand engineering and ripening control of QDs

Perovskite CsPbI₃ QDs capping with oleylamine (OAm) and oleic acid (OA) ligands were synthesized by the hot-injection method, referencing our previously reported work (5, 21). We added the PZPY molecules directly to the colloidal solution to obtain the PZPY-treated CsPbI₃ QDs (referred to as target QDs). PZPY has good chemical and thermal stability and will not decompose during the preparation of QDs (fig. S1).

The use of PZPY could effectively inhibit QDs’ ripening and eliminate undesired secondary growth, thus notably enhancing QDs’ luminescent properties and stability (Fig. 1). During the initiation of the QD synthesis, most of the small-sized nuclei proliferate toward the thermodynamic equilibrated critical radius (22). In this process, the ligands offer colloidal stability, terminate the growth of QDs, and prevent undesired ripening (23). During the purification, film formation, and storage process, QDs with high surface energy usually suffer secondary growth and Oswald ripening after losing ligands, negatively affecting the luminescent properties and stability of QDs (16). Hence, we introduced PZPY with a bidentate molecular structure to offer strong interaction between the uncoordinated lead cations (Pb²⁺) of CsPbI₃ QDs and the coordinate bond of PZPY. We used density functional theory calculations to compute the molecular electrostatic potential (ESP) of PZPY, revealing that the electron density of the aromatic ring shifts toward the N atom in ESP, making it tend to coordinate with Pb²⁺ using its lone pair electrons that are not involved in conjugation (inset in Fig. 1). Because of the small size and C–N bond of pyrazole providing good molecular flexibility, PZPY could attach to the surface of QD by twisting its structure to coordinate with Pb²⁺ (24). When ligands fall off, QDs could possibly combine with other falling-off ligands but more likely aggregate for decreasing surface energy. However, when PZPY exists in this system, PZPY preferentially binds to the QD because of its strong nucleophilicity, thus avoiding QD aggregation or Ostwald ripening. The surface energy of QDs with ligand loss problems is thereby reduced so that good monodispersing and luminescent properties are successfully maintained. Besides, owing to the weak conjugation of the pyrazole ring, PZPY exhibits a large band gap, effectively preventing exciton harvesting from the perovskites.

Fig. 1. CsPbI₃ QD synthesis and ripening control strategy.

Schematic illustration of synthesized pristine and PZPY-treated CsPbI₃ QDs. During the purification, film formation, and storage process, PZPY could effectively inhibit QDs’ ripening phenomena and enhance their luminescent properties and stability. The inset at the bottom left shows the molecular formula, structure, and electrostatic potential (ESP) of PZPY molecules.

Microstructure and morphology

To verify this hypothesis, we first investigated the microstructures of the as-prepared pristine QDs and target QDs (Fig. 2) via STEM. The pristine QDs suffer adhesion between QDs (Fig. 2A), while the target QDs retain cubic morphology (Fig. 2D). To explore the stability of QDs, such as-prepared samples were exposed to the air for 3 days before the STEM test. We found that the pristine QDs tended to aggregate, and their boundaries became more ambiguous with irregular shapes after exposure (Fig. 2B). However, the target QDs still maintained their initial morphologies (Fig. 2E) and decent crystallinity (fig. S2). The digestive ripening process will promote the shape transition of QDs from cubes to spheres, originating from the dissolution of vertices or some edges (25). Besides, the target QDs presented evident size reduction and no lattice spacing change (6.5 Å) (Fig. 2, C and F), which also proves that PZPY inhibits Oswald ripening and secondary growth of QDs. Then, further detailed microstructures were analyzed to explore the root of the poor performance of the pristine QDs. The ripening and aggregation process create different kinds of defects, including interface fusion (Fig. 2G), low-angle boundary (Fig. 2H), high-angle boundary (Fig. 2I), antiphase boundary (Fig. 2, J and K), and dislocation (Fig. 2L). First, some surface defects act as charge recombination centers and thus degrade QDs (26). Second, ripening and interface fusion weaken the quantum confinement effect of QDs and adversely affect their photoelectronic properties (23). Third, because of interface fusion and boundary defects, morphology and carrier transport properties of the QD film are inevitably affected (27, 28). In short, from STEM results, we can conclude that PZPY could inhibit Oswald ripening and secondary growth of QDs, further suppressing the production of different kinds of defects.

Fig. 2. Microstructures of CsPbI₃ QDs.

(A) STEM image of the pristine QDs. (B) STEM image of the pristine QDs after leaving the QDs dispersed on the Cu grid in the air for 3 days, indicating that the pristine QDs tend to aggregate. (C) High-resolution STEM image of the pristine QDs. (D) STEM image of the target QDs. (E) STEM images of the target QDs after leaving the QDs dispersed on the Cu grid in the air for 3 days, indicating that the target QDs show decent stability. (F) High-resolution STEM images of the target QDs. (G to L) Different types of defects in the pristine QDs stored for 3 days. (G) Interface fusion. (H) Low-angle boundary. (I) High-angle boundary. (J) Antiphase boundary. (K) Dislocation. (L) Low-angle boundary, antiphase boundary, dislocation, and hole.

We then investigated the QD films’ micromorphology by using scanning electron microscopy (SEM) and atomic force microscopy (AFM) measurements. As the SEM images presented in fig. S3, the pristine QDs tended to aggregate because of the increased surface energy originating from surface ligands loss, leading to large-sized islands and affecting the coverage above the substrate. Nevertheless, a smooth, dense, and refined surface was observed for the target QD films. AFM images of the pristine and target QD films showed dense and uniform morphologies (fig. S4) with root-mean-square roughness values of 5.48 and 4.02 nm, respectively. The thickness of the pristine QD films is about 26.46 nm, while that of the target QD film is about 23.76 nm (fig. S5). In brief, benefitting from improved morphologies and ripening control, the target QDs have better film formation characteristics.

Chemical composition and optical properties

To gain an in-depth understanding of the detailed molecular interaction between the PZPY molecules and QDs, we evaluated the chemical composition and optical properties of QDs. Fourier transform infrared and x-ray photoelectron spectroscopy (XPS) measurements were applied to study the surface chemical states of QDs. As shown in Fig. 3A, C═N stretching modes (1500 to 1650 cm⁻¹) indicated the existence of PZPY in target QDs. Besides, C–H stretching modes (2700 to 3000 cm⁻¹) were weakened for target QDs, suggesting fewer OAm/OA ligands in the target QD solution (fig. S6) (29). After using PZPY, some molecules partially connected to uncoordinated Pb²⁺ so that the problems of ligand falling off and QD aggregation could be largely mitigated in the subsequent process. Figure 3B and fig. S7 presented the results of XPS measurements. Compared to the pristine QDs, target QDs showed peak shifts for the I 3d and Pb 4f signals, while no peak shift was observed for the Cs 3d signal. These results indicate that the surface or environmental composition of target QDs has changed, and these changes mainly originate from Pb–I bonds. The peaks at 137.85 and 142.70 eV for the pristine QDs could be assigned to the Pb 4f 5/2 and 4f 7/2 signals, respectively (30). For target QDs, the peaks of Pb 4f 5/2 and 4f 7/2 signals both show broadening (Fig. 3B), while N 1s signals show multiple peaks (fig. S7), which can be attributed to a richer chemical environment on the target QD surface. By peak fitting, Pb atoms in the QDs were supposed to exist in two chemical environments (Fig. 3B). The higher and lower binding energy regions belong to Pb-PZPY and Pb–I, respectively, because of the stronger electron-withdrawing ability of PZPY. Besides, the I/Pb content ratio increased, while the O/Pb content ratio (originating from OA/OAm ligands) decreased after PZPY treatment as measured by quantitative XPS analysis (Fig. 3C), suggesting that PZPY could bond on the QD’s surface and fill the halide vacancies originating from ligand loss. We performed nuclear magnetic resonance (NMR) to confirm further the interaction between QDs and PZPY (Fig. 3D). The PZPY molecule and PZPY-QDs showed similar ¹H NMR signals, suggesting that PZPY could attach at the surface of QDs and interact with them. Besides, an additional peak was observed at δ 7.57, which originated from the –NH₂ group of OAm. The above results suggest that the use of PZPY could offer strong interaction with the Pb²⁺ of perovskite QDs, which could inhibit the aggregation of QDs to maintain their good dispersion.

Fig. 3. Effect of PZPY on the optical, structural, and electric properties of CsPbI₃ QDs.

(A) C=N stretching vibration peak of Fourier transform infrared spectra. (B) High-resolution x-ray photoelectron spectroscopy (XPS) spectra of Pb 4f. (C) Quantitative XPS analysis. (D) ¹H NMR signals. (E) X-ray diffraction (XRD) patterns. ppm, parts per million. (F) PL spectra and schematic illustration of enhanced emission for target QDs. (G) PLQYs and the photographs of CsPbI₃ QD solution under ultraviolet lamps. (H) Time-resolved PL curves. (I) Current-voltage characteristics and device structure of the hole-only devices. a.u., arbitrary units.

We then applied x-ray diffraction (XRD) measurements to investigate the crystal structures of QD films (Fig. 3E). In both the pristine QD film and target QD film, the main diffraction peaks are located at 14.1°, 20.0°, and 28.6°, corresponding to (100), (110), and (200) planes of the cubic phase structure, respectively (5). For the target QDs, no additional diffraction peaks appeared, and the main diffraction peaks were not shifted, indicating that the crystalline structure did not change. The broader typical diffraction peak (from 0.51° to 0.61°) signified the size reduction of target QDs (31).

On the other hand, PZPY could improve the optical properties of QDs. The target QD solution exhibited an enhanced PL emission with a 3-nm blue shift compared to the pristine QD solution (Fig. 3F), suggesting the decreased particle size of target QDs. The target QD solution presented saturated and brighter deep-red light under ultraviolet light with a higher absolute PLQY of 94% than that of the pristine QD solution (46%), as shown in Fig. 3G. The enhanced deep-red PL emission and PLQY can be ascribed to the reduced defects and suppressed nonradiative recombination paths, which is in good agreement with the conclusions drawn from the STEM results in Fig. 2 (32).

We next applied time-resolved PL measurements on QDs to analyze carriers’ behaviors (Fig. 3H). We fit the decay curves by applying a bi-exponential equation (Eq. 1) (see table S1 for fitting results). The longer decay lifetime indicates slower recombination in the target QDs with lower defect densities. The average carrier lifetimes (τ_avg) of the pristine QDs and target QD solution were calculated to be 14.52 and 19.65 ns (Eq. 2), respectively. Besides, after introducing PZPY, the fast decay component (corresponding to nonradiative recombination) was reduced from 0.51 to 0.35, while the slow decay component (corresponding to radiative recombination) was enhanced from 0.41 to 0.59. All the above results confirm that the use of PZPY could adequately inhibit nonradiative recombination and enhance the excitonic recombination of QDs (33)

$$I=A_1\text{exp}(-\frac{t}{{\mathrm{\tau }}_1})+A_2\text{exp}(-\frac{t}{{\mathrm{\tau }}_2})$$ (1)

$${\mathrm{\tau }}_{\text{avg}}=\frac{A_1{{\mathrm{\tau }}_1}^2+A_2{{\mathrm{\tau }}_2}^2}{A_1{\mathrm{\tau }}_1+A_2{\mathrm{\tau }}_2}$$ (2)

We used hyperspectral wide-field microscopy and spatially resolved confocal fluorescence–lifetime imaging microscopy to investigate the microscopic optical properties, carriers’ lifetime, and behaviors of QD films. Hyperspectral wide-field microscopy spatially mapped the PL uniformity, as shown in fig. S8. The target QD film showed a more uniform PL emission at the microscopic scale, while the pristine QD film presented many nonluminescent points, displaying that using PZPY could improve the PL homogeneity of the QD films. Carrier lifetime maps also indicated more uniform morphology and longer lifetime for the target QD film, as shown in fig. S9. In short, together with the SEM and AFM images above, hyperspectral wide-field microscopy and spatially resolved confocal fluorescence–lifetime imaging microscopy results all display that the target QD film is of great unity and micro-homogeneity, which is a good base for the following device construction.

We operated space-charge-limited current measurements to further verify trap sites and densities in QD film under an electrical field. Such a hole-only device with a structure of indium tin oxide (ITO)/poly(ethylene dioxythiophene) (PEDOT):polystyrene sulfonate (PSS):perfluorinated ionomer (PFI)/poly[bis(4-phenyl)(2,4,6-trimethylphenyl)amine] (PTAA)/QDs/MoO_x/Al was fabricated (Fig. 3I). The density of trap states is proportional to the trap-filled limit voltage (V_TEL). The V_TEL values of the control and target device are 0.74 and 0.64 V, respectively, suggesting lower trap density and fewer defect sites in the target QD film.

Efficient and stable perovskite QLEDs

We lastly fabricated and evaluated perovskite QLEDs. Figure 4 shows the structure and performance of QLEDs based on the pristine QDs and target QDs. We use a device configuration consisting of ITO, PEDOT:PSS:PFI (160 nm), PTAA (40 nm), perovskite QDs (20 to 30 nm), 1,3,5-tris(1-phenyl-1H-benzimidazol-2-yl)benzene (TPBi; 40 nm), LiF (1 nm), and Al (100 nm), as shown in Fig. 4 (A and B) and fig. S10. Current density and radiance curves as a function of the voltage of the QLEDs are presented in Fig. 4 (C and D), respectively. Because of the introduction of conjugated ligands enhancing the conductivity of QDs, the target device showed higher current densities. Besides, the target device presented a much higher radiance, indicating superior carrier transport and injection. The maximum radiance of the target device approached 43,691 mW sr⁻¹ m⁻² at 6.0 V, much higher than that of the control device (9752 mW sr⁻¹ m⁻² at 5.8 V). By optimizing concentrations of PZPY (fig. S11), a high peak EQE of 26.0% (Fig. 4E) was achieved for the champion device, comparable with the best batch of reported perovskite LEDs (fig. S12 and table S2). In comparison, the control device showed a peak EQE of 7.4%. The grossly enhanced EQE also suggests that the carrier transport and recombination in the target QD films have been improved, and the nonradiative decay rate originating from defects has been suppressed. A histogram of EQEs for 30 target devices shows an average EQE of 23.8% (Fig. 4F), indicating that the device performance is highly reproducible. Besides, various PZPY-derived molecules were applied to treat QDs, including 2,2′-bithiazole, 2-(pyridin-2-yl)thiazole, and 2-(pyridin-2-yl)oxazole (see fig. S13 for chemical structure and ESP information), suggesting that such a molecule-induced ripening control strategy is generalized and feasible for enhancing the performance of perovskite QLEDs (fig. S14).

Fig. 4. Device structure and performance of CsPbI₃ QLEDs.

(A) Illustration of the device structure and EL. (B) Energy diagram of QLEDs. (C) Current density-voltage curves of QLEDs. (D) Radiance-voltage curves of QLEDs. (E) EQE curves of QLEDs. (F) The statistics of the distribution of maximum EQE of 40 devices. (G) EL spectra of QLEDs. (H) CIE diagram of QLEDs.

Figure 4G presented typical electroluminance (EL) spectra of the control and target devices, with peaks at 689 and 686 nm, respectively, narrow emission, and high color purity. The corresponding Commission Internationale de l’Eclairage (CIE) chromaticity coordinates of the control and target devices are (0.73, 0.27) and (0.71, 0.27), as indicated in Fig. 4H. The shift in EL and CIE chromaticity coordinates indirectly confirmed the size reduction of QDs, consistent with the results of STEM, XRD, and PL.

The target QD solution has excellent storage stability thanks to the introduction of conjugated ligands suppressing the ripening and aggregation of QDs. After as long as 1 month of storage, the target QD solution maintained uniform dispersity in the solvent, and the prepared target device retained good performance (Fig. 5, A to C). In comparison, the control QDs had degraded severely (inserted picture in Fig. 5A). The as-fabricated QLEDs based on the target QD solution after being stored for 1 month presented a maximum radiance of 43,152 mW sr⁻¹ m⁻² at 5.4 V (Fig. 5B) and a maximum EQE of 21.7% (Fig. 5C). Furthermore, QLEDs based on the target QD solution after being stored for 3 months showed a maximum EQE of 20.3%. These results confirm the high storability of the target QDs.

Fig. 5. Stability of CsPbI₃ QLEDs.

(A) The current density-voltage curve of the QLED based on QD solution after being stored for 1 month. The inset shows a photograph of the pristine and target QD solution after being stored for 1 month. (B) Radiance-voltage curves of the QLED based on QD solution after being stored for 1 month. (C) EQE curve of the QLEDs based on fresh QD solution, QD solution after being stored for 1 month and for 3 months. (D) Accelerated ageing tests for the control QLEDs under different initial radiance (R₀). (E) The T₅₀ lifetimes as a function of R₀ for the target QLEDs; the solid line is the fitting of the T₅₀ data to equation R₀ⁿ T₅₀ = constant, where n is the acceleration factor (n = 1.86). (F) Voltage shifts of the control QLED at a constant current density. (G) Accelerated ageing tests for the target QLEDs at different R₀. (H) The T₅₀ lifetimes as a function of R₀ for the target QLEDs; the solid line is the fitting of the T₅₀ data to equation R₀ⁿ T₅₀ = constant, where n is the acceleration factor (n = 1.87). (I) Voltage shifts of the target QLED at a constant current density.

We then evaluated the operating stability of the fresh devices, as shown in Fig. 5 (D to I). The T₅₀ at the initial radiance (R₀) of ~15,400 mW sr⁻¹ m⁻² of the target device was 175 min, indicating a 50-fold improvement compared to the control devices (only 3.3 min at 15,402 mW sr⁻¹ m⁻²). When testing the operating stability of QLEDs under a constant current density, the voltage of the target device presented a small lifting range (0.46 V for 300 min at an initial radiance of 11,301 mW Sr⁻¹ m⁻²), while the voltage of the control device increased rapidly (0.46 V for 50 min at an initial radiance of 3650 mW sr⁻¹ m⁻²), as shown in Fig. 5 (F and I). We also estimated the T₅₀ at 190 mW sr⁻¹ m⁻² through an empirical scaling law R₀ⁿ T₅₀ = constant (where n is the acceleration factor) (34). Here, the luminance in candela per square meter is determined by the photopic luminous efficiency function V(λ), which is used to measure the sensitivity of human vision to visible light radiation of different wavelengths. For instance, the value of V(λ = 525 nm) is 0.79, while V(λ = 686 nm) is 0.01. Thus, to more accurately evaluate the operating stability of deep-red QLEDs, we went for radiance over the luminance commonly used in green or pure red emission. For reference, an efficient perovskite green LED located at 525 nm has a luminance of 100 cd m⁻², corresponding to a radiance of 190 mW sr⁻¹ m⁻² (fig. S15). By testing the T₅₀ of devices at different R₀, we calculated that the acceleration factors (n) are 1.86 and 1.87 for the control and target devices, respectively. The T₅₀ for the target device can be estimated to be 10,587 hours at an initial radiance of 190 mW sr⁻¹ m⁻² (221 hours for the control QLED), indicating a high operating stability for perovskite QLEDs (table S2). These results indicate that using PZPY could simultaneously improve the efficiency and stability of the QLEDs.

DISCUSSION

In summary, we use PZPY with a bidentate molecular structure to treat perovskite QDs for high-performance QLEDs. PZPY could interact with Pb²⁺ of QDs strongly and maintain stable surface states of QDs. The surface chemical states and structure information of QDs showed that PZPY could adhere to the surface and inhibit QDs’ ripening and aggregation. Thus, target QDs show highly emissive features and effective electronic transportation properties. The as-fabricated device exhibited excellent performance with a maximum peak EQE of 26.0%. Meanwhile, target QDs presented extraordinary operating stability of about 10,587-hour lifetime and high storage stability. QLEDs based on the QD solution after being stored for 1 month showed a maximum EQE of 21.7%, and the ones based on the QD solution after being stored for 3 months showed a maximum EQE of 20.3%. We believe that the proposed molecule-induced ripening control strategy for highly stable and efficient QDs accelerates the industrialization of QLED technology and its application.

MATERIALS AND METHODS

Materials

PbI₂ (99.99%), OAm (80 to 90%), OA (80 to 90%), 1-octadecene (ODE; >80%), isopropanol, ethyl acetate (99%), and toluene were purchased from Aladdin reagent. Caesium stearate (CsSt; 98%) was purchased from J&K reagent. N-octane (>98%) was purchased from TCI reagent. PEDOT:PSS solutions, PTAA, and TPBi were purchased from Xi’an Yuri Solar Co. Ltd. Nafion perfluorinated resin solution (PFI, tetrafluoroethylene-perfluoro-3, 6-dioxa-4-methyl-7-octenesulfonic acid copolymer, 5 wt % in a mixture of lower aliphatic alcohols and water, containing 45% water) was purchased from Sigma-Aldrich. Dichloromethane was purchased from Shanghai Titan Scientific Co. Ltd. All the chemical materials were directly used without any further purification.

Synthesis and purification of CsPbI₃ QDs

Perovskite QDs were synthesized by using our previously reported method with some modifications (35). The cesium precursor was prepared by stirring CsSt (2.5 g, 6 mmol), 40 ml of ODE, and 2 ml of OA and heated to 140°C until the solution became clear and transparent. For the synthesis of the pristine CsPbI₃ QDs, PbI₂ (0.25 g, 0.6 mmol) was loaded into a 100-ml three-neck flask along with ODE (10 ml), OAm (1.5 ml), and OA (2 ml), and the mixture was degassed at 120°C for 10 min under N₂ flow. The temperature was raised to 180°C to complete the solubilization of the PbI₂ salt. Then, 0.75 ml of CsSt solution was quickly injected into the solution. After a short reaction time of 15 s, the three-necked flask was placed in ice water and cooled to room temperature. Then, the crude solution was loaded into a centrifuge tube, and isopropanol was added with a volume ratio of 1:3. Then, the mixture was centrifuged for 1 min at 8000 rpm. The precipitate was collected and dispersed into a mixture of toluene and ethyl acetate with a volume ratio of 1:1:2. Then, OAmI dissoluted with toluene (0.05 M) was added. After the solution was centrifuged for 1 min at 8000 rpm, the precipitate was collected and redispersed in n-octane. For target QDs, the steps were similar, except that PZPY with a certain molar mass was mixed with PbI₂ to add into a 100-ml three-neck flask.

Synthesis of PZPY

A mixture of 2-bromopyridine (0.471 g, 3 mmol), pyrazole (0.136 g, 2 mmol), Cs₂CO₃ (2.93 g, 9 mmol), CuI (0.038 g, 0.2 mmol), and l-proline (0.046 g, 0.4 mmol) in 3 ml of dimethyl sulfoxide (DMSO) was heated at 90°C for 48 hours under N₂. Then, the crude mixture was diluted by water and extracted by dichloromethane. The combined organic layers were washed with brine, dried over anhydrous sodium sulfate, and then evaporated. The resulting mixture was purified by column chromatography (100 to 200 mesh) to give the product a white solid. The product was recrystallized twice with dichloromethane and ether for further use.

¹H NMR (400 MHz, CDCl₃): δ 8.56 (d, J = 2.5 Hz, 1H), 8.39 (dd, J = 5.0, 2.5 Hz, 1H), 7.97 (dt, J = 8.0, 1.0 Hz, 1H), 7.79 to 7.76 (m, 1H), 7.72 (d, J = 1.5 Hz, 1H), 7.17 to 7.13 (m, 1H), and 6.46 to 6.45 (m, 1H); ¹³C NMR (101 MHz, CDCl₃): δ 151.9, 148.5, 142.1, 138.7, 127.3, 121.5, 112.5, and 107.9; High-resolution mass spectrometry mass/charge ratio calculated for C₈H₈N₃ [M+H]⁺ 146.0713, found 146.0715.

Device fabrication

The QLED configuration consists of ITO, PEDOT:PSS:PFI (160 nm), PTAA (40 nm), perovskite QDs (20 to 30 nm), TPBi (40 nm), and LiF/Al (1/100 nm). The emissive area of QLED was 0.03 cm², as defined by the overlapping area of the ITO and Al electrodes. For the hole transport layer, a mixture of PEDOT:PSS (Baytron P VPAl 4083) and PFI solutions (filtered through a 0.22-μm filter) were spin coated onto the ITO-coated glass substrates at 4000 rpm for 45 s, followed by annealing at 140°C for 15 min in the atmosphere. Then, PTAA (5 mg ml⁻¹ in chlorobenzene) was spin coated onto the substrate at 3000 rpm for 45 s, followed by annealing 120°C for 25 min in a nitrogen-filled glove box. QD films (synthesized as described above) were spin coated at 2000 rpm for 60 s without any annealing process. Last, we evaporated the electron transport layer TPBi and LiF/Al electrodes through a shadow mask under a high vacuum of ~2 × 10⁻⁴ Pa.

Characterization and device measurements

One drop of the diluted as-synthesized QD dispersion was placed onto a carbon-coated Cu grid, and the octane evaporated at room temperature. STEM images were acquired by using an aberration-corrected Thermo Fisher Scientific (Themis Z) at 300 kV with a beam current of ~2 pA. The QD solution’s PL spectra were obtained by using a Varian Cary Eclipse spectrometer. The PL decay was measured at room temperature by using a time-correlated single-photon counting (TCSPC) spectrofluorometer (FLS920, Edinburg Instrument, UK). The QD samples for NMR measurement were prepared by redissolving dried QD powder in deuterated DMSO. ¹H NMR and ¹³C NMR spectra were measured by using a Brucker AVANCE III spectrometer at room temperature in deuterated DMSO with tetramethyl silane as the internal standard. XRD samples were prepared by dropping the purified QD inks onto glass substrates. XRD patterns were achieved by using a Bruker D8 Advance x-ray diffractometer with Cu Kα radiation (λ = 1.5406 Å). QD solution was spin coated onto the glass substrates and then analyzed by using an atomic force microscope (Cypher ES, Oxford Instruments Asylum Research Inc.) to evaluate the film roughness and thickness. The QD film imaging process was collected by using a Nikon AXR-NSPARC confocal laser scanning microscope. For fluorescence confocal imaging, QDs were excited under a 633-nm solid-state laser light source and collected emission spectrum with a 650- to 700-nm band-pass filter; 20× objectives were used for PL signal collection in sequential mode. Luminescence lifetime imaging was performed by using the TCSPC section of the Nikon AXR-NSPARC microscope. The picosecond pulsed laser light sources were used for excitation, and the TCSPC counter was produced from Qicoquant (PicoHarp 300). QDs were excited at 600 nm, and filters of 680/50 were used for emission collection. The EL spectra, radiance-current density-voltage characteristics, and EQE were collected by using a Keithley 2400 source, a fiber integration sphere, and a PMA-12 spectrometer for light output measurements in a glove box filled with N₂ at room temperature (Ocean Optics Co. Ltd. designed the measurement equipment).

Supplementary Materials

This PDF file includes:

Figs. S1 to S15

Tables S1 and S2

References