Highly Emissive Blue Quantum Dots with Superior Thermal Stability via In Situ Surface Reconstruction of Mixed CsPbBr3–Cs4PbBr6 Nanocrystals

Abstract Although metal halide perovskites are candidate high‐performance light‐emitting diode (LED) materials, blue perovskite LEDs are problematic: mixed‐halide materials are susceptible to phase segregation and bromide‐based perovskite quantum dots (QDs) have low stability. Herein, a novel strategy for highly efficient, stable cesium lead bromide (CsPbBr3) QDs via in situ surface reconstruction of CsPbBr3–Cs4PbBr6 nanocrystals (NCs) is reported. By controlling precursor reactivity, the ratio of CsPbBr3 to Cs4PbBr6 NCs is successfully modulated. A high photoluminescence quantum yield (PLQY) of >90% at 470 nm is obtained because octahedron CsPbBr3 QD surface defects are removed by the Cs4PbBr6 NCs. The defect‐engineered QDs exhibit high colloidal stability, retaining >90% of their initial PLQY after >120 days of ambient storage. Furthermore, thermal stability is demonstrated by a lack of heat‐induced aggregation at 120 °C. Blue LEDs fabricated from CsPbBr3 QDs with reconstructed surfaces exhibit a maximum external quantum efficiency of 4.65% at 480 nm and excellent spectral stability.


Introduction
Metal halide perovskite materials have been recognized as promising candidates for next-generation color displays because their high photoluminescence quantum yields (PLQYs), narrow DOI: 10.1002/advs.202104660 full widths at half maximum (FWHM), ease of bandgap tuning, and solution processability fulfil the ITU-R Recommendation BT.2020 (Rec. 2020) of the International Telecommunication Union (ITU). [1][2][3] Recently, significant progress has been achieved in the development of near-infrared, red, and green perovskite light-emitting diodes (LEDs), with external quantum efficiencies (EQEs) reaching over 20%. [4][5][6][7][8] However, the efficiency of blue perovskite materials has lagged far behind, with EQEs of 12.3% in the sky blue region of the spectrum (475-490 nm) and 8.8% in the blue region (460-475 nm) having been reported. [9,10] Furthermore, Joule heating during LED operation is inevitable; thus, the development of perovskite materials with superior thermal stability, in which thermal quenching is minimized, is an important issue for the practical application of perovskite LEDs. Several strategies for obtaining blue-emitting perovskites nanocrystals (NCs) are available. One method involves mixed halides that include both Br and Cl anions. [11][12][13][14][15][16][17] Although this is a convenient method for bandgap engineering, easy formation of Cl − vacancies poses a limitation as it results in a deep trap state within the bandgap. [18][19][20] These defect sites cause perovskite layer degradation and ion migration, resulting in phase segregation in response to the application of an electric field during device operation. [21,22] A second method is to use Br-based 2D perovskite nanoplatelets and take advantage of the exciton quantum confinement effect. [23][24][25][26][27] In inorganic cesium lead bromide (CsPbBr 3 ) nanoplatelets, the emission can be controlled according to the number of [PbBr 6 ] 4− layers; however, strong exciton-phonon coupling and a randomly oriented distribution of nanoplatelets result in low-performance LEDs. [28] A third strategy for achieving blue-emitting perovskites NCs is to reduce the crystal size of a perovskite material such that it is within the quantum confinement regime. [9,[29][30][31][32] CsPbBr 3 NCs with sizes in the quantum confinement regime (denoted by quantum dots, QDs) usually suffer from low PLQYs and stability because they are strongly affected by surface defects when the surface-to-volume ratio is high. [33] In particular, small-sized QDs are easily degraded and undergo aggregation because of their high surface energy, leading to broad emission spectra and poor spectral stability such www.advancedsciencenews.com www.advancedscience.com that the emission color is susceptible to changing to green at high temperature. To overcome these problems, various approaches have been attempted, including amorphous CsPbBr x shelling, [31] the addition of excess Br using ZnBr 2 as the source, [29] Sb 3+ doping, [32] and acid etching-driven ligand exchange. [34] However, few studies have been conducted on the application of blue LEDs incorporating highly stable CsPbBr 3 QDs. [9,34] A quite different approach involves CsPbBr 3 NCs being embedded in a Cs 4 PbBr 6 matrix with a crystal size of several hundred nanometers or more, resulting in improved PLQY and stability. [35][36][37][38] However, the literature reports in this area have been focused only on green emission, and Cs 4 PbBr 6 /CsPbBr 3 embedded structures are not suitable for electroluminescent devices because a 0D Cs 4 PbBr 6 phase has a wide bandgap of 3.95 eV and is insulating. [39,40] In this article, we report a unique method to enhance the PLQY and stability of blue-emitting CsPbBr 3 QDs via simultaneous generation of mixed CsPbBr 3 QDs and Cs 4 PbBr 6 NCs. By controlling the reactivity of the precursors, the size of the CsPbBr 3 QDs was controlled. As a result, an emission wavelength of 470 nm with a high PLQY of >90% was achieved. We investigated the effects of the Cs 4 PbBr 6 NCs on the photophysical properties and thermal stability of the CsPbBr 3 QDs by observing changes in their morphology and optoelectronic properties. Octahedron defect sites on the surface of CsPbBr 3 QDs were etched by the Cs 4 PbBr 6 NCs, resulting in defect removal. This CsPbBr 3 QD surface reconstruction decreases the defect density and eliminates nonradiative recombination pathways, leading to high efficiency and stability for the CsPbBr 3 QDs. The mixed NC solution retained 90% of its initial PLQY value over 120 days of storage under ambient conditions, with little change in the emission peak position and FWHM. Thermally induced aggregation and fusion were suppressed during 60 min of heating at 120°C. Spectrally stable and efficient blue LEDs, having an EQE of 4.65% at 480 nm, based on the mixed NCs were achieved.

Preparation of In Situ Generated Blue-Emitting NCs
We synthesized a mixed solution of CsPbBr 3 and Cs 4 PbBr 6 NCs by modifying a previously reported synthetic method (details provided in the Experimental Section). [41] In a typical synthesis, cesium carbonate (Cs 2 CO 3 ), lead oxide (PbO), oleic acid (OA), and 1-octadecene (ODE) were added to a three-necked round-bottom flask, and metal oleate complexes were formed by heating at 120°C. Oleylammonium bromide (OLAM-Br) was prepared separately by reacting oleylamine and hydrobromic acid (HBr), and then this was injected at low temperature under Ar into the metal oleate complexes. In this step, the reaction temperature and the Cs to Pb precursor ratio, as important factors for obtaining highquality NCs with blue emission in the range of 460-480 nm, were carefully controlled. The CsPbBr 3 to Cs 4 PbBr 6 NC formation ratio was modulated by varying the Cs to Pb precursor feed ratio, as described in a previous report. [42] When the amount of Pb precursor exceeded the amount of Cs precursor, the formation of CsPbBr 3 NCs in the orthorhombic phase was favored ( Figure  S1, Supporting Information). When the Cs and Pb precursor ratio was fixed at 1:1 and the reaction temperature was reduced, the growth of NCs was suppressed, and small CsPbBr 3 NCs and Cs 4 PbBr 6 NCs were simultaneously co-synthesized. The emission wavelength of the QDs was 470 nm, and a high PLQY of above 90% was achieved. This phenomenon occurred because the depletion of the Pb precursor was slower than that of the Cs precursor. [42,43] At low temperature, the difference between the reactivities of the metal precursors was maximized, hence formation of Cs 4 PbBr 6 NCs was promoted and the size of the CsPbBr 3 NCs was reduced (Figure 1).
We investigated the dependence of the photophysical properties of the as-synthesized NCs on the reaction temperature in the range of 60-160°C. These NCs exhibited two peaks in the UV-vis absorption spectra: the first excitonic peak, between 450 and 500 nm, and a sharp peak centered at 313 nm (Figure 2a). The absorption peak at 313 nm originated from optical transitions between localized states of the isolated [PbBr 6 ] 4− octahedron of Cs 4 PbBr 6 NCs. [44][45][46] X-ray powder diffraction (XRD) patterns of the NCs allowed us to confirm that the first exciton peak is related to the bandgap of the luminescent CsPbBr 3 NCs. As shown in Figure 2b, the sample of synthesized NCs included both orthorhombic CsPbBr 3 NCs and rhombohedral Cs 4 PbBr 6 NCs, and the Cs 4 PbBr 6 -to-CsPbBr 3 ratio increased as the reaction temperature decreased. This result is consistent with the increase in the intensity of the absorption peaks at 313 nm in the UV-vis spectra. As the reaction temperature decreased from 160 to 60°C, the first exciton peak was blueshifted, and the emission peak center shifted from the green (504 nm) to the blue (470 nm) wavelength region (Figure 2c). Transmission electron microscopy (TEM) analysis also indicated that CsPbBr 3 and Cs 4 PbBr 6 NCs coexisted in the samples, and lowering the temperature reduced the overall size of the NCs (Figure 2d-g and Figure S2 (Supporting Information)). In addition, when the particle sizes (diameters) were measured by focusing on the luminescent CsPbBr 3 NCs, the average size and relative standard deviations decreased from 9.3 to 3.5 nm and from 38.1% to 9.0%, respectively, as the reaction temperature decreased from 160 to 60°C, because of the slow crystal growth rate at low temperature ( Figure 2h). Because the size of the CsPbBr 3 NCs synthesized at a temperature below 160°C was smaller than the exciton Bohr radius (≈7 nm), [47] the blueshift in the emission peaks resulted from a quantum confinement effect, and the reduction in FWHM was due to the size distribution and defect density decrease. It should be noted that as the reaction temperature decreases, the PLQY increases from 57.7% to 90.1%, contrary to previously reported general trends of increasing crystallinity and PLQY with temperature ( Figure 2i). [41]

Stability of In Situ Generated Perovskite NCs
In order to analyze the stability of the in situ generated CsPbBr 3 -Cs 4 PbBr 6 NCs (denoted by ISNCs), we synthesized small CsPbBr 3 QDs (denoted by C-QD 113 ) as a control group using a conventional method based on hot injection of Cs-oleate at 90°C with control of the OA and oleylamine (OLAM) ligands (details provided in the Figure S3 in the Supporting Information and the Experimental Section). [48] Conventional CsPbBr 3 QDs (denoted by C-QD 113 ) were obtained by hot injection of Cs-oleate at 90°C with control of the OA and OLAM ligands. In general, small CsPbBr 3 QDs have high surface energies and a large number of defects, and hence they can be ripened easily, resulting in a photoluminescence (PL) redshift. Upon examining the colloidal stability of ISNCs that had been stored under an environment with relative humidity of 65% and room temperature of 25°C, little change in emission peak and FWHM was observed over 120 days, and 90% of the initial PLQY was retained (Figure 3a). By contrast, the emission wavelength of the C-QD 113 without Cs 4 PbBr 6 rapidly redshifted, moving from 458 to 475 nm within 30 days, even at room temperature ( Figure 3d). To compare in detail the thermal stability of the prepared QDs, they were dispersed in toluene and incubated for a period of time at 120°C. The C-QD 113 were easily ripened and fused together at high temperatures, and the emission gradually shifted to longer wavelengths, with a green emission at 500 nm observed after 60 min of incubation ( Figure 3e). The XRD data showed that a nonluminescent CsPb 2 Br 5 tetragonal phase was formed simultaneously ( Figure 3f). In the case of the ISNCs, the intensity of the emission centered at ≈510 nm increased slightly with time during the annealing process, but there was no shift in the position of the emission peak maximum and the blue emission was maintained with a slight decrease in PL intensity ( Figure 3b). Furthermore, other crystal structures such as CsPb 2 Br 5 were not generated, and the diffraction patterns indicated that the crystallinity of the ISNCs was improved (Figure 3c). These results clearly demonstrated that the Cs 4 PbBr 6 effectively suppressed heat-induced aggregation and decomposition of small CsPbBr 3 QDs. These significant changes could be attributed to a reduction in surface energy via surface passivation. When comparing the water contact angles of ISNC and C-QD 113 films, the contact angle of the ISNC film was 107.2°, i.e., greater than that of the C-QD 113 film, which was 65.4°( Figure S4, Supporting Information). The increased contact angle indicates a reduction in surface energy due to effective passivation of unsaturated atoms by hydrophobic ligands. [32,34]

Effects of Cs 4 PbBr 6 NCs on CsPbBr 3 QDs Quality
We hypothesized that the cause of the high PLQY and good thermal stability of the ISNCs was related to the presence of the Cs 4 PbBr 6 NCs generated with the CsPbBr 3 QDs. To investigate the role of the Cs 4 PbBr 6 NCs, we conducted a systematic study to monitor the differences in the optical properties, defect levels, and morphology of CsPbBr 3 QDs with different amounts of added Cs 4 PbBr 6 NCs. For this purpose, nonluminescent pure Cs 4 PbBr 6 NCs (denoted by NC 416 ) with diameters of 13.6 nm were prepared separately via a previously reported method (Figure S5, Supporting Information). [49] CsPbBr 3 QDs were separated from a crude solution and then mixed with Cs 4 PbBr 6 NCs in different weight ratios from 0 to 5 in a nonpolar solvent. The separation of the CsPbBr 3 QDs from the crude solution involved extraction via a size selection process using methyl acetate as an antisolvent; a clear emission spectrum peaking at 477 nm was observed for the separated CsPbBr 3 QDs. Because Cs 4 PbBr 6 NCs and relatively large CsPbBr 3 QDs were removed during the separation process, the absorption peak at 313 nm disappeared, and the PL peak was slightly blueshifted. The XRD and TEM results for the CsPbBr 3 QD sample verified that the rhombohedral Cs 4 PbBr 6 phase was not present and that the sample consisted entirely of CsPbBr 3 QDs of 4.3 nm in size ( Figure S6, Supporting Information).
After mixing the separated CsPbBr 3 QDs (denoted by S-QD 113 ) and NC 416 for 1 h at room temperature in a weight ratio of 1:0 to 1:5, changes in morphology and optical properties were observed in a nonpolar solvent (Figure 4). As the relative NC 416 amount was increased, the absorption intensity at 313 nm increased, while the first exciton peaks of S-QD 113 gradually decreased in intensity and were blueshifted (Figure 4a). The emission peaks were also shifted to shorter wavelength, and the PL intensity was concomitantly improved (Figure 4b and Figure S7 (Supporting Information)). In addition to the optical properties, changes in particle size and morphology were observed. In the 1:2 ratio mixed solution, the size of the S-QD 113 decreased from 4.32 to 3.87 nm, while the NC 416 size increased from 13.38 to 14.51 nm (Figure 4c and Figure S8 (Supporting Information)). The morphology of the NC 416 transformed from hexagonal to truncated diamond and assembled into zigzag shapes ( Figure S9, Supporting Information). The etching of S-QD 113 and the variation in the NC 416 shape are a result of the high surface energy of the small CsPbBr 3 QDs and the structural lability of perovskite as a function of its ligand environment. The CsPbBr 3 phase was successfully converted into the Cs 4 PbBr 6 phase and vice versa using excess OLAM and OA in a previous study. [50] In our system, NC 416 had a relative excess of the OLAM ligand, so small S-QD 113 were easily etched and the NC 416 size was increased ( Figure S10, Supporting Information).
Time resolved photoluminescence (TRPL) measurements were performed to determine the effects of the PLQY enhancement along with NC etching on the optical spectroscopic characteristics (Figure 4d). Decay curves were used to analyze the excited state radiative relaxation dynamics. The S-QD 113 decay curve was fitted to a biexponential function with a 3.89 ns average lifetime. When S-QD 113 was mixed with NC 416 , the decay curve of the resultant mixture fitted a monoexponential function well, and the average lifetime was gradually increased to 5.25 ns. This increment in the average lifetime could be a result of various effects, such as energy transfer between S-QD 113 and NC 416 , a S-QD 113 size effect, and a reduction in trap-state density; hence, we decided to examine each of these possible causes in turn. First, Cs 4 PbBr 6 NCs have a wider bandgap than CsPbBr 3 NCs, so energy transfer is possible when the distance between the two materials is sufficiently close. Xuan et al. reported that the lifetime and PLQY were increased in a perovskite composite, in a study in which CsPbBr 3 NCs were embedded in Cs 4 PbBr 6 NCs. [38] In addition, Chen et al. synthesized CsPbBr 3 -embedded Cs 4 PbBr 6 and analyzed the effect of metal halide interlayer in determining their photoluminescence excitation (PLE) properties. [51] To investigate the energy transfer, we acquired PLE and PL spectra of S-QD 113 and ISNCs for various the excitation wavelengths ( Figure S11, Supporting Information). The S-QD 113 emission intensity gradually decreased as the excitation wavelength increased, in line with the trend observed for the PLE spectrum. The PLE spectrum of the ISNCs included a sharp drop centered at around 313 nm corresponding to the Cs 4 PbBr 6 NC absorption peak. These results demonstrate that the origin of the PL of the ISNCs can be attributed to band edge emission of the CsPbBr 3 QDs rather than energy transfer from the Cs 4 PbBr 6 NCs. In the latter case, the ISNC emission would have improved significantly before the Cs 4 PbBr 6 NC absorption region. Therefore, we excluded the energy transfer effect in ISNCs. Second, for perovskites, lifetimes tend to decrease as bandgaps widen. [52][53][54][55] Since S-QD 113 is in the strong quantum confinement regime, the lifetime is expected to decrease with the decrease in particle size. However, in our study, when NC 416 was mixed with S-QD 113 , the size of the S-QD 113 particles decreased slightly because of surface etching. This result, indicating a longer lifetime for smaller NCs, was contrary to our expectation. Therefore, we conclude that surface passivation effects are likely to be the main reasons for the longer lifetime, as reported previously. [56,57] The Cs 4 PbBr 6 NCs might promote the elimination of defects and radiative recombination in the CsPbBr 3 QDs.
To investigate in detail the optical properties of the NCs, the temperature-dependent PL of the S-QD 113 and ISNC samples was measured ( Figure S12, Supporting Information). The PL intensity of the S-QD 113 sample gradually decreased as the temperature increased, in agreement with previous reports. [58][59][60] In metal halide perovskites, because of the low exciton binding energy, excitons are dissociated into free charge carriers by ther-mal energy, promoting nonradiative decay. [60][61][62] However, the IS-NCs displayed a constant PL intensity, for the entire temperature range from 20 to 300 K. As the exciton binding energies of NCs are highly dependent on the size of the NCs, [59,62] we expect that the exciton binding energy of the S-QD 113 (≈4.32 nm) and ISNC (≈4.33 nm) samples would be similar. Therefore, the constant ISNC PL intensity as a function of temperature suggests that nonradiative decay paths are substantially reduced by the presence of Cs 4 PbBr 6 .
In addition, we induced an interaction between the S-QD 113 and NC 416 samples on the substrate to identify the effects of defect passivation in the solid state. First, half of the substrate was covered with Kapton tape (3M #5413) and then NC 416 was spin coated on the exposed part. Subsequently, the experiment proceeded in order with the removal of ligands with methyl acetate, peeling off the tape, and coating S-QD 113 over the entire substrate. As a result, half of the substrate was covered with only S-QD 113 , and the other coated with a double layer, with S-QD 113 and NC 416 in contact at the layer interface ( Figure S13, Supporting Information). For this as-prepared film, initially there was no difference in emission between the two regions (A and B); however, the PL intensity was greatly improved in the double layer (layer B) after 12 h (Figure 4e,f). Although the interaction was relatively slow in the solid state, the S-QD 113 defects were also passivated similar to the solution. We summarized the above-described mechanistic findings in a schematic illustration (Figure 4g). Surface reconstruction takes place at the interfaces between the CsPbBr 3 QDs and Cs 4 PbBr 6 NCs, and imperfect octahedrons on the CsPbBr 3 QD surfaces are reduced during the etching process. In this process, the CsPbBr 3 QD crystal size decreases, resulting in a blueshift of the emission wavelength, and the optical properties and colloidal stability are increased by the addition of Cs 4 PbBr 6 NCs.

Fabrication and Characterization of Blue Perovskite LEDs
Encouraged by the high PLQY and stability of the IS-NCs, we constructed LEDs with glass/indium tin oxide (ITO)/poly (3,4-  One of the most important problems for blue-emitting metal halide perovskites is the spectral instability that results from halide segregation. LED emission spectra measured using various voltage bias values are shown in Figure 5e. Because the IS-NCs exhibited blue emission at 480 nm without halide mixing, the ISNC LEDs exhibited stable electroluminescence (EL) emission spectra over the entire range of device operating voltages.

Conclusion
In summary, we demonstrated a novel synthetic method to simultaneously obtain ultrasmall CsPbBr 3 QDs and Cs 4 PbBr 6 NCs by controlling the ratio of the Cs and Pb precursors as well as the reaction temperature. CsPbBr 3 QDs of 3.5 and 4.3 nm in sizesmaller than the exciton Bohr radius -were synthesized at 60 and 80°C, respectively. These QDs emitted blue light at 470 and 477 nm, respectively, with high PLQYs of more than 90%. The Cs 4 PbBr 6 NCs eliminated the imperfect octahedron defect sites on the surfaces of the CsPbBr 3 QDs, which resulted in suppression of nonradiative recombination; this was confirmed via TRPL and temperature-dependent PL measurements. In addition, the colloidal and thermal stabilities of the ISNCs were significantly enhanced by suppressing particle aggregation and fusion. We realized efficient blue LEDs with a maximum EQE of 4.65% at 480 nm. In particular, the EL spectra were stable across various applied voltages, without exhibiting any peak shifts. This surface defect etching method, which relied on CsPbBr 3 and Cs 4 PbBr 6 phase surface reconstruction, was found to be effective even for solid-state films. Thus, our strategy could expand the development of a wide range of perovskite optoelectronic applications, www.advancedsciencenews.com www.advancedscience.com including solar cells and NC-based LEDs, thanks to the introduction of a new method for controlling the amount of defects.
Preparation of OLAM-Br Precursor: The OLAM-Br salt was synthesized according to a previous method with some modifications. [41] In a typical synthesis, OLAM (10 mL) and HBr solution (1.3 mL) were loaded into a 50 mL three-necked round-bottom flask, and the resulting solution was stirred under an Ar atmosphere at 120°C for 2 h. Then, it was heated to 150°C and left to react for an additional 30 min. After cooling the solution to 100°C, it was vacuum dried for 1 h to remove any residual water. The precursor was collected in an Ar-filled vial and stored in a glove box for further use.
Synthesis of ISNCs: Mixed CsPbBr 3 -Cs 4 PbBr 6 NCs were prepared via an OLAM-Br precursor hot-injection method. Typically, Cs 2 CO 3 (32.6 mg, 0.1 mmol), PbO (44.6 mg, 0.2 mmol), OA (1.0 mL), and ODE (10 mL) were stirred in a 50 mL three-necked round-bottom flask and degassed under vacuum at 120°C for 1 h. After complete solubilization of the reaction mixture, the flask was filled with Ar and heated (or cooled) to obtain the desired temperature (60-160°C). Then, the preheated OLAM-Br solution (0.9 mL) was swiftly injected into the reaction mixture. The reaction was quenched in an ice water bath after 30 min.
Purification of Synthesized NCs: The crude solution was transferred to a 50 mL conical tube and ACN and toluene were then added to the solution in a volume ratio of 1:2:3 (crude mixture:ACN:toluene). The nanocrystals were precipitated in a centrifuge at 7000 rpm for 5 min. The supernatant was discarded, and the precipitate was collected and dissolved in hexane. One more centrifugation (7800 rpm, 5 min) was required to purify the NCs and obtain the final product. The clear supernatant was collected and used for future studies.
To eliminate the Cs 4 PbBr 6 NCs from the crude solution, methyl acetate was used as an antisolvent instead of the mixture of ACN and toluene. Typically, methyl acetate (5 mL) was added to the crude solution (5 mL). This solution was centrifuged at 7800 rpm for 5 min and the precipitate was discarded. An additional methyl acetate (20 mL) was added to the supernatant, and this mixture was centrifuged at 7800 rpm for 5 min. The precipitated CsPbBr 3 QDs (S-QD 113 ) were used in further studies after dispersion in hexane.
Synthesis of Pure NC 416 : Monodisperse Cs 4 PbBr 6 NCs were synthesized according to a previously reported method. [49] The Cs-oleate precursor and NCs were prepared in air. For the Cs-oleate preparation, Cs 2 CO 3 (0.4 g) and OA (8 mL) were loaded in a 20 mL vial and stirred on a hot plate at 150°C for 20 min. In a typical synthesis, PbBr 2 (36.7 mg), OA (0.2 mL), OLAM (1.5 mL), and ODE (5 mL) were stirred at 150°C until the solution became transparent. After cooling the solution to 80°C, preheated Cs-oleate (0.75 mL) was swiftly injected into it. The reaction was quenched in an ice water bath after 3 min. The crude solution was washed via centrifugation (4500 rpm, 10 min), which was followed by redispersion in hexane.
Synthesis of C-QD 113 : Small CsPbBr 3 QDs were synthesized according to a previously reported method with some modifications. [48] The synthetic approach was based on hot injection of the Cs-oleate precursor. In brief, PbBr 2 (69 mg) and ODE (5 mL) were loaded in a 50 mL three-necked round-bottom flask and degassed under vacuum at 120°C for 1 h. Dried OA (0.6 mL) and OLAM (0.3 mL) were injected to the reaction mixture at 120°C under Ar. After complete solubilization of the reaction mixture, preheated Cs-oleate precursor (0.4 mL) was swiftly injected into the reaction mixture at 90°C. The reaction was quenched in an ice water bath after 10 s. For Cs-oleate preparation, Cs 2 CO 3 (0.4 g), OA (1.2 mL), and ODE (15 mL) were stirred in a 50 mL three-necked round-bottom flask and degassed under vacuum at 120°C for 1 h. Then, the solution was heated to 150°C and reacted for an additional 30 min. The Cs-oleate precursor was preheated to 100°C before use. Purification of the obtained QDs was achieved via the method mentioned above.
Preparation of Ex Situ Mixed S-QD 113 and NC 416 : The solution was prepared by adding an amount of NC 416 to S-QD 113 (2 mg) to achieve the desired weight ratio in hexane (500 μL). After the addition of NC 416 , it was observed that the PL intensity increased within a few seconds. To achieve a homogeneity, the mixture was stirred for 1 h at room temperature. It is worth noting that when the QD 113 :NC 416 weight ratio exceeded 1:5, the QD 113 sample was completely etched and the emission was lost. To apply this system to the solid state, glass substrates, Kapton tape, QD 113 (10 mg mL −1 in hexane), and NC 416 (20 mg mL −1 in hexane) were used. The glass substrates were first cleaned by sonification while sequentially immersed in deionized water, acetone, and isopropyl alcohol. Then, half of the glass was covered with Kapton tape, and NC 416 was spin coated at 3000 rpm for 1 min. To avoid dissolution of NC 416 layer, methyl acetate was then spin coated on the slide twice at 3000 rpm for 1 min. After peeling off the Kapton tape, S-QD 113 was spin coated at 3000 rpm for 1 min on the entire substrate.
Thermal Stability Test: After dissolving the NCs in toluene and adjusting the concentration to 10 mg mL −1 , a change in the PL spectrum of the sample in a 120°C oil bath was observed as the time elapsed.
Device Fabrication: ITO-patterned glass substrates were cleaned by sonification while sequentially immersed in deionized water, acetone, and isopropyl alcohol. The PEDOT:PSS layer was spin coated at 5000 rpm for 40 s on the ITO substrates after 30 min of UV treatment. The slide was then transferred into a glove box and annealed at 140°C for 10 min. TFB and PVK (volume ratio 1:1) were blended and dissolved in chlorobenzene such that the concentration of the mixture was 3 mg mL −1 . The TFB/PVK mixture solution was spin coated on the substrates (3000 rpm, 40 s) and then annealed at 130°C for 20 min. Perovskite NCs were then spin coated on the substrates (2000 rpm, 30 s). Finally, the slide was sequentially coated with TPBi (50 nm), LiF (1 nm), and Al (100 nm) by thermal evaporation.
Characterization: Absorption spectra were acquired by a Shimadzu UV-1800 UV-vis spectrometer. PL spectroscopy was carried out and quantum yields were obtained for the NCs via the use of a quantum efficiency measurement system (Otsuka QE-2000). Photoluminescence emission spectra were obtained by using an Agilent fluorescence spectrophotometer. XRD was performed by using a Rigaku Ultimate-IV X-ray diffractometer operated at 40 kV and 200 mA using the Cu K line ( = 1.5418 A). TEM images were acquired by a JEOL JEM-2100 microscope with an acceleration voltage of 200 kV using copper grids (Ted Pella, USA). The particle sizes and distributions were measured using DigitalMicrograph software in TEM images. TRPL spectra were obtained by means of a timecorrelated single-photon counting setup (FluoTime 300, PicoQuant) at room temperature. 1 H nuclear magnetic resonance spectra were acquired using a Bruker AVANCE III HD (400 MHz) spectrometer. The residual proton signal of the deuterated solvent was selected as the reference standard. Temperature-dependent PL measurements were performed in the temperature range of 20-300 K using a liquid helium cooler. PL spectra of the nanocrystal films were obtained using the Agilent fluorescence spectrophotometer. Water contact angles were measured using a drop shape analyzer (DSA-100, Krüss). Cross-sectional SEM images of the device structures were obtained using a Nova Nano230 FEI SEM (accelerating voltage 10 kV). To prevent the occurrence of charging, a 5 nm platinum layer was deposited on the samples via sputter coating (Emitech K575x, Tescan). The device performances of the encapsulated LEDs were measured using a Keithley 2400 sourcemeter and spectroradiometer (CS-2000, Konica Minolta) under ambient conditions.