Effective Stabilization of Perovskite Cesium Lead Bromide Nanocrystals through Facile Surface Modification by Perfluorocarbon Acid

CsPbX3 (X = Cl, Br, I) perovskite nanocrystals (NCs) have attracted much attention as promising materials for next-generation optoelectronic applications. However, improvement of their low stabilities against heating and humidity is needed for practical use. In this work, we focused on perfluorodecanoic acid (PFDA) as a surface ligand and investigated the thermal and chemical stabilities of the photoluminescence (PL) properties of CsPbBr3 NCs. Oleic acid (OA) adsorbed on the NCs was exchanged for decanoic acid (DA) and PFDA. OA-modified and DA-modified NCs exhibited drastic fluorescence quenching to 12.9 and 21.1% of their initial PL intensities, respectively, after heating at 100 °C for 4 h. In contrast, the PFDA-modified NCs maintained 92.1% of their PL intensity after the same heating. Furthermore, the polar solvent resistance was also improved by PFDA modification. These improvements can be attributed to the strong adsorptivity and high chemical stability of the PFDA ligand.


INTRODUCTION
Lead halide perovskites have been intensely researched as materials for next-generation optoelectronic applications, such as solar cells, 1−5 light-emitting diodes, 6−11 wide color gamut displays, 12−14 photodetectors, 10,15−17 and lasers. 18−21 Lead halide perovskites are classified as organic−inorganic and allinorganic materials based on their elemental composition. 12 Organic−inorganic halide perovskites such as CH 3 NH 3 PbX 3 (X = Cl, Br, I) have been developed in solar applications. 22 Their critical problem is their low stability, preventing practical applications. On the other hand, all-inorganic CsPbX 3 perovskites exhibit higher durability both in the air and under heating. 23,24 Recently, CsPbX 3 nanocrystals (NCs) have attracted much attention as fluorescent materials. 25,26 Their outstanding characteristics, including their precisely tunable band gap and photoluminescence (PL) wavelength, very narrow PL peak width, excellent absolute PL quantum yields (PLQYs), and short PL lifetimes, were first reported by Kovalenko's group. 12 CsPbX 3 NCs are therefore very fascinating fluorescent materials; however, improvements in their stabilities against light irradiation, heating, humidity, and polar solvents are needed for practical use. There have been many reports on stabilization of CsPbX 3 NCs through core/ shell structuring, 27−29 Janus structuring, 30 hybridization with polymer, 31 cation exchange of Pb 2+ to Mn 2+ , 32 and surface passivation using CsX solutions. 33 On the other hand, surface ligands also have a significant influence on the stability of nanometer-sized materials, including CsPbX 3 NCs, because of their large specific surface areas. Oleic acid (OA) and oleylamine (OLA) are used as surface ligands in the synthesis of CsPbX 3 NCs by the hot injection method, which has been frequently used by many researchers following Kovalenko's group. 12 OA adsorbs on the surface of the NCs by carboxylate coordination with Pb, while the ammonium group of protonated OLA interacts with Br through hydrogen bonding. 25 Degradation trigger mechanisms of CsPbBr 3 NCs are thought to include desorption of surface ligands, e.g., codesorption of OA and OLA through proton transfer between the carboxylate and ammonium groups in adsorbed ligands 34 and desorption of a pair of ammonium ligands and coordinated Br − . 35 Accordingly, the halide vacancies formed by desorption act as trap levels that cause nonradiative recombination, leading to PL quenching; 36,37 therefore, ligands that cannot desorb from the surface have been exploited to suppress the deterioration of CsPbBr 3 NCs. 38, 39 Liu et al. reported that the PLQY of CsPbI 3 NCs with OA and OLA adsorbed as surface ligands decreased from 86 to 60% after room-temperature storage for 30 days, while the use of trioctylphosphine (TOP) instead of OA in the synthesis of the NCs successfully improved the PLQY stability. 40 On the other hand, according to previous work by Wu et al., CsPbBr 3 NCs with adsorbed OA and OLA dispersed in cyclohexane exhibited a decrease in the PL intensity to 14% of the initial intensity in 20 min after the addition of ethanol, whereas the PL intensity was maintained at 95% using trioctylphosphine oxide for surface modification. 41 Furthermore, the PL properties of the NC dispersion were readily improved by adding adequate organic molecules, such as difluoroacetic acid and tributylphosphine, through ligand exchange with the adsorbed OA. 36,42 We previously reported that the photostability of OA-adsorbed CsPbBr 3 NCs in toluene was improved by adding a suitable amount of OA in the dispersion to facilitate readsorption of OA on the exposed surface after the photoinduced desorption of OA. 43 These previous works revealed that surface ligands have an important role in the improvement and stabilization of the PL properties of CsPbX 3 NCs.
To the best of our knowledge, there are fewer works on improving the thermal stability of CsPbBr 3 NC dispersions than the storage stability under ambient conditions and photostability under excitation irradiation. In previous work by Wang et al., 44 the PL intensities of a hexane dispersion of CsPbBr 1.2 I 1.8 NCs with adsorbed OA and OLA decreased to 16% of the initial intensity after heating from 20 to 90°C, followed by recovery of up to 39% after cooling down to 20°C . This result indicates irreversible thermal damage and thermal quenching. In contrast, when TOP was added to the NC dispersion, the PL intensity at 90°C was 43% of the initial intensity at 20°C, followed by recovery to 93% after cooling to 20°C, revealing that the TOP modification suppressed the thermal damage of the NCs. Luminescent materials used in optoelectronic devices are inevitably heated; therefore, improvement of the thermal stability of CsPbX 3 NCs is an important issue. 44−46 We focused on the adsorption of a carboxylic acid as a surface modifier to improve the stability of CsPbBr 3 NCs. Carboxylic acid in a deprotonated state adsorbs on the NC surface. The ease of deprotonation can be evaluated from the acid dissociation constant, K a . Therefore, a carboxylic acid with a low pK a (=−log K a ) works as a strongly adsorbing ligand on the NC surface. Fluorocarboxylic acids have lower pK a values than OA because they have highly electron-withdrawing fluorine atoms. Their deprotonated state can be more stable; therefore, they frequently adsorb on the NC surface through coordination bonds and are not expected to desorb through proton transfer from the surrounding oleylammonium ligands, which are simultaneously adsorbed on the surface. In previous studies, improvements in the PL intensity were reported for a CsPbBr 3 thin film passivated by trifluoroacetate ions and for a CsPbX 3 NC dispersion modified with tetrafluoroborate ions. 47,48 However, the influence of the surface ligand of fluorocarboxylic acid on the stability of colloidal CsPbX 3 NCs has not been evaluated.
In this work, perfluorodecanoic acid (PFDA) was chosen to investigate the stabilization of CsPbBr 3 NCs. We prepared colloidal CsPbBr 3 NCs modified by deprotonated OA (oleate) and tetraoctylammonium ligands. The oleate ligands on the surface of CsPbBr 3 NCs exchanged for deprotonated PFDA through repeated adsorption and desorption processes. In a control experiment, decanoic acid (DA) instead of PFDA was also examined (chemical structures of DA and PFDA are exhibited in Figure S1). Changes in the particle morphology and optical properties under heating by ligand exchange of OA for PFDA and DA were evaluated. Moreover, another significant problem of CsPbBr 3 NCs is their remarkably low stability against polar solvents. 41 We assumed that surface modification with fluorocarbon acids will protect CsPbBr 3 NCs from serious damage by polar solvents; therefore, the effect of PFDA modification on the polar solvent resistance of the NCs was also evaluated.

RESULTS AND DISCUSSION
2.1. Characterization of As-Synthesized CsPbBr 3 NCs. Cubic CsPbBr 3 was confirmed for the as-synthesized sample by X-ray diffraction (XRD) analysis (see Figure S2). The elemental composition was determined by X-ray fluorescence (XRF) analysis. The compositional ratio of Cs/Pb/Br was 1.6:2.4:6.1, which corresponded to the stoichiometric ratio of CsPbBr 3 within the range of experimental error. The band gap (E g ) of the as-synthesized CsPbBr 3 NCs dispersed in toluene was determined from a Tauc plot calculated from the UV−vis spectra (see Figure S3a,b). The Tauc plot was calculated by using eq 1 49 where a is the absorbance, h is the Plank constant, A is a constant, and v is the frequency. The value of n was 0.5 because CsPbBr 3 is a direct transition-type semiconductor. 50 The E g of the as-synthesized NCs was 2.47 eV, which is larger than the E g of bulky cubic CsPbBr 3 (∼2.34 eV); 12 therefore, the NCs should show the quantum size effect. These NCs exhibited green luminescence under 400 nm excitation ( Figure  S3a). The PL peak position, full width at half maximum, and

ACS Omega
Article PLQY were 498.9, 28.9 nm, and 72.2%, respectively. The PL decay curves were measured (see Figure S3c) and fitted according to eq 2 where f(t) is a fitting function, t is the time, A i is each amplitude, and τ i is each PL lifetime. The average PL lifetime τ ave was calculated using eq 3.
The analysis results for the PL decay curves are summarized in Table S1. τ ave was 4.6 ns. The fitting curve could be decomposed into two components of τ 1 = 2.7 ns and τ 2 = 8.3 ns. The short component (τ 1 ) is attributed to exciton radiative recombination, while the long component (τ 2 ) might be associated with charge-trapping. 51 2.2. Influence of PFDA Modification on the Thermal Stability of CsPbBr 3 NCs. PFDA and DA were added to OA-NCs at 0.06 mmol L −1 to obtain PFDA-NCs and DA-NCs, respectively. Figure 1A shows the changes in the PL spectra upon the addition of PFDA and DA ligands to the OA-NCs. The addition of PFDA enhanced the PL intensity by 1.48 times. Correspondingly, the PLQY increased from 72.2% of the OA-NCs to 90.1% of the PFDA-NCs, whereas that of the DA-NCs was almost kept at 73.6%. This increase can be explained by a decrease in the number of bromide vacancies or low-coordinate lead atoms through surface passivation by PFDA, which adsorbed on the surface more frequently and rigidly than OA. The improved PLQY should be realized by a decrease in the number of surface traps. 42 Bromide vacancy on the NC surface makes a defect level, which causes nonradiative relaxation under the lowest conduction band energy ( Figure  1B). 36 Passivation of the surface defect by surface ligand leads to a decrease in the possibility of the nonradiative relaxation, resulting in enhanced PLQY. The red shift of the PL peak by ∼7 nm may reveal that the strong interaction of PFDA has a non-negligible influence on the band structure. The conduction band, which is composed of 6p orbitals of Pb, 12 should be affected by the PFDA coordination. The coordination by fluorocarbon compound might reduce the conduction band minimum, 52 leading to a decrease in E g and red shift of the PL peak. On the other hand, the DA addition did not affect the PL properties, implying that there was no apparent difference in the passivation effects of DA and OA.
To evaluate the thermal stability, OA-NCs, DA-NCs, and PFDA-NCs were heated at 100°C for 4 h. Herein, the heated samples were cooled to room temperature before measurements of the optical properties. Figure 2 shows the changes in the appearance of the dispersions, which contain NCs in toluene at 0.8 g L −1 , under white light and UV light with heating time. Initially, the three samples were transparent, greenish-yellow dispersions under white light. Yellow sediment was observed for the OA-NCs and DA-NCs during heating, whereas the PFDA-NCs maintained a clear solution and color. The sedimentation resulted from strong aggregation of the NCs, which was attributed to the significant promotion of surface ligand desorption during heating. The color change from greenish-yellow to yellow is explained by PL quenching of the NCs, which absorb UV and blue light under white light and show a green emission. Figure 3a−c shows the changes in the UV−vis absorption spectra for the OA-NCs, DA-NCs, and PFDA-NCs. An increase in the absorbance in the region of 500−800 nm was observed for the OA-NCs and DA-NCs during heating. This increase is explained by the increased light scattering intensity from the aggregated NCs. On the other hand, the change in the UV−vis absorption spectrum of the PFDA-NCs was smaller. The PFDA-modified NCs therefore had better dispersibility than the OA-modified and DA-modified NCs. Herein, the absorption peak at ∼488 nm, which was attributed to the interband transition of the CsPbBr 3 NCs, was maintained during heating, indicating that the NCs did not dissolve in toluene. The red shift of the absorption edge clearly observed for the OA-NCs and DA-NCs indicates a decrease in E g . The E g was determined from the Tauc plots (shown in Figure S4) calculated from the above UV−vis absorption spectra. Figure 3d shows the changes in E g during heating. The E g of the OA-NCs and DA-NCs decreased monotonically, while the E g of the PFDA-NCs was nearly constant, even after the same heating duration. Therefore, we validated the obvious stabilization of the optical absorption properties of the CsPbBr 3 NCs by PFDA.
As shown in the photograph in Figure 2, gradual PL quenching was observed for the OA-NCs and DA-NCs during heating, while the PFDA-NCs maintained bright luminescence under UV light. Figure 4a,b shows the changes in the PL spectra for the OA-NCs and DA-NCs. Their PL peaks red-

ACS Omega
Article shifted from 498.9 and 498.1 to 508.3 and 506.1 nm, respectively. In contrast, as displayed in Figure 4c, the PL peak position of the PFDA-NCs changed slightly from 505.4 to 505.7 nm. Figure 4d plots the changes for each PL intensity during heating. Herein, the PL intensity was normalized to the initial intensity. The OA-NCs and DA-NCs exhibited a drastic decrease in the PL intensity to 12.9 and 21.1%, respectively, after heating for 4 h. On the other hand, the PFDA-NCs maintained 92.1% of their initial PL intensity after the same heating duration. For the PLQYs, the OA-NCs and DA-NCs showed a significant decrease from 72.2 and 73.6 to 22.7 and 30.7%, respectively, while the PFDA-NCs exhibited a smaller

ACS Omega
Article change from 90.1 to 85.8%. From these results, the ligand exchange of OA for PFDA remarkably improved the thermal stability of the PL properties of the CsPbBr 3 NCs. Figure 5 displays the transmission electron microscopy (TEM) images of each sample before and after heating. From the corresponding particle size distributions ( Figure S5), the average particle sizes of the OA-NCs and DA-NCs increased from 6.0 ± 0.8 and 5.4 ± 0.9 to 11.6 ± 2.6 and 8.6 ± 2.0 nm, respectively, revealing that the observed decrease in the E g and red shift of the PL peak can be (Figures 3d and 4a,b) attributed to weakened quantum size effects, which were caused by crystal growth. The smaller NCs were dissolved during heating, and the larger NCs grew through reprecipitation of dissolved ions on their surface. This process is known as Ostwald ripening. On the other hand, the average particle size of the PFDA-NCs remained at 5.0 nm, indicating that the PFDA modification suppressed dissolution and reprecipitation processes under heating. This result explains the constant E g and PL peak position (Figures 3d and 4c).
The PL lifetimes were analyzed from the PL decay curves shown in Figure 4 and are summarized in Table 1. The average PL lifetimes of the OA-NCs and DA-NCs significantly increased from 4.6 and 4.8 to 16.1 and 17.3 ns, respectively. These results were obtained from biexponential fitting curves. In contrast, the PL decay curve of the PFDA-NCs was fitted by a monoexponential curve, and the calculated PL lifetime, 4.2 ns, exhibited a negligible change to 4.1 ns. The prolonged PL lifetimes might be attributed to a decrease in the total nonradiative combination probability through surface trap levels accompanied by a reduction in the specific surface area due to crystal growth, although surface defects were formed and promoted desorption of the surface ligands under heating. It should be noted that a decrease in the nonradiative combination probability through surface trap levels generally enhances the PLQY; however, the PLQY decreased with heating, as described above. This PLQY decrease can be explained by the weakened quantum confinement effect due to crystal growth.
The deterioration of the PL properties with heating can be attributed to the strong aggregation of the NCs, the formation of surface defects, and the weakened quantum confinement effect from crystal growth due to dissolution and reprecipitation on the exposed crystal surfaces. These phenomena resulted from significant desorption of the surface ligands from the NCs. However, PFDA modification through ligand exchange improved the thermal stability of the NCs, revealing that the adsorptivity of PFDA was stronger than that of OA and DA. The difference in adsorptivity can be explained by their acidity. Carboxylic acid ligands modify crystal surfaces through coordination of deprotonated carboxy groups with metal cations; therefore, a more stable deprotonated state results in carboxyl acid ligands with higher adsorptivity. The values of pK a for OA, DA, and PFDA are 6.2, 53 4.9, 54 and 2.58, 55 respectively. Since perfluoroalkyl groups have strong electron-withdrawing properties, the acidity of PFDA is high; 56 therefore, the deprotonated state of PFDA is relatively stable, resulting in good adsorption of surface ligands and suppressed desorption. The effective passivation of the surface defects improved the PLQY and thermal stability of the CsPbBr 3 NCs.
To support the qualitative correlation between the acidity of the added carboxyl acids and the thermal stability of the CsPbBr 3 NCs, varelic acid (VA; pK a = 4.82) 54 and stearic acid (SA; pK a = 6.3) 57 were also examined in the same way. Figure  S6a,b shows the changes in the PL spectra for the VA-and SAadded dispersions (VA-NCs and SA-NCs, respectively). The PL peaks of the VA-NCs and SA-NCs red-shifted from 495.8 and 497.6 to 509.1 and 512.0 nm, respectively. Furthermore, as shown in Figure S6c, their PL intensities decreased to 11.5 and 10.7% of the initial intensities, respectively. These results show that VA and SA, which have lower acidities than PFDA, cannot improve the thermal stability of the NCs due to their adsorptivity nearly equivalent to that of OA and DA.
2.3. Influence of PFDA Modification on the Polar Solvent Resistance. To evaluate ligand protection of the CsPbBr 3 NCs in polar solvents, ethanol was mixed with the OA-NCs, DA-NCs, and PFDA-NCs. Figure 6 shows the changes in the sample appearance under white light and UV light after ethanol addition. Under white light, the color of the OA-NCs and DA-NCs changed from yellowish-green to yellow in 15 min, and then yellow sediment was observed at 30 min. At 120 min, the samples became clear and colorless solutions with yellow sediment. The color of the PFDA-NCs also changed to yellow by adding ethanol; however, its color was maintained even after 120 min without any sediment. Figure 7a−c shows the changes in the UV−vis absorption spectra for the OA-NCs, DA-NCs, and PFDA-NCs. Herein, the data at 0 min are of the as-prepared dispersions before the ethanol addition. The absorbance of the OA-NCs and DA-NCs increased until 30 min because of enhanced light scattering due to aggregated NCs and then decreased. The decrease in absorbance can be explained by the dissolution of

ACS Omega
Article the NCs and the precipitation of significantly aggregated NCs. On the other hand, the PFDA-NCs almost maintained their absorbance even after 120 min, indicating that there were welldispersed NCs without dissolution and aggregation. It should be noted that the PFDA-NCs also showed changes in the UV− vis absorption spectrum in the early stage, possibly due to a locally higher concentration of ethanol damaging the NCs immediately after the ethanol addition before homogeneous mixing. The red shift of the absorption edge may be caused by the weakened quantum size effect due to crystal growth. Figure 8 shows the TEM images of the OA-NCs, DA-NCs, and PFDA-NCs before and after ethanol addition. The average particle sizes were calculated from the corresponding size distributions ( Figure S7). It should be noted that the difference in initial particle size between Figures 5 and 8 was less than 1 nm, which was attributed to an experimental error. For the OA-NCs and DA-NCs, smaller particles approximately 10 nm in size and larger particles with various shapes were observed simultaneously. The former should be NCs grown through dissolution and reprecipitation processes. The latter might be particles of dissolved ions that precipitated during the drying process for the preparation of the TEM samples. In contrast, such larger particles were not observed for the PFDA-NCs, indicating that the dissolution of the NCs was suppressed; therefore, the solvent resistance of the dispersion was effectively improved by PFDA modification. Growth of the NCs from 5.6 ± 0.7 to 7.9 ± 3.4 nm was observed for the PFDA-NCs, corresponding to the red shift in the absorption edge in Figure 7c.
The PFDA-NCs maintained a brighter green luminescence under UV excitation than the others, as displayed in Figure 6. Changes in the PL spectra after ethanol addition are shown in Figure 9a−c. The spectra at 0 min are of the as-prepared dispersions before the ethanol addition, as noted above. Large deterioration in the PL was observed for all samples, but PFDA suppressed the deterioration. The PL peaks of the OA-NCs and DA-NCs red-shifted from 499.4 and 499.2 to 516.3 and 518.4 nm, respectively, in 120 min. On the other hand, the PFDA-NCs exhibited a smaller redshift in the PL peak from 505.9 to 508.0 nm. The PL red shift of the PFDA-NCs ( Figure  9c) was smaller than those of the OA-NCs and DA-NCs (Figure 9a,b), while the observed red shifts of their optical absorption edge were similar (Figure 7). PL red shift of semiconductor NCs is caused by quantum size effect and enhanced self-absorption. The significant PL red shift observed for the OA-NCs and DA-NCs would be mainly affected by the

ACS Omega
Article enhanced self-absorption due to aggregation of the NCs by the addition of ethanol. This is supported by an obvious increase in absorbance at ∼600 nm (Figure 7a,b), which indicates enhanced light scattering intensity by the NC aggregation. In contrast, such an increase in absorbance was not observed for the PFDA-NCs (Figure 7c). This reveals the suppression of NC aggregation, which explains that the PL peak position almost kept unchanged after the addition of ethanol. Figure 9d shows the changes in the PL intensity normalized to the initial intensity. Drastic decreases in the PL intensity to 9.5 and 13.6% in 120 min were observed for the OA-NCs and DA-NCs, respectively, whereas the PFDA-NCs maintained 36.5%. The PL quenching can be explained by the weakened quantum confinement effect and an increase in surface defects including the bromide vacancies, which makes trap levels causing nonradiative relaxations ( Figure 1B), due to desorption of surface ligands. Rigid PFDA modification suppressed the damage caused by a polar solvent.
The PL lifetimes of the dispersions before and after ethanol addition were calculated from the PL decay curves shown in Figure 9 and are summarized in Table 2. The average PL lifetimes of the OA-NCs and DA-NCs increased from 5.1 and 5.0 to 30.5 and 19.7 ns, respectively. In contrast, the PFDA-NCs showed a smaller change from 3.5 to 3.9 ns. The increase in the PL lifetime for the OA-NCs and DA-NCs is explained by the decrease in the total nonradiative combination probability through surface trap levels accompanied by a reduction in the specific surface area, as discussed in the above thermal stability evaluation. Moreover, significant desorption of surface ligands destabilizes the dispersibility of the NCs, leading to strong aggregation and sedimentation, as already observed in Figure 6. The higher adsorptivity of PFDA should suppress PL deterioration in a polar environment. We also observed stability enhancement against ethyl acetate (see Supporting Information). PFDA modification is therefore expected to improve the stability of CsPbBr 3 NCs against various polar solvents.

CONCLUSIONS
In summary, we investigated the effects of PFDA modification on the thermal stability and polar solvent resistance of CsPbBr 3 NCs. PFDA has a higher adsorptivity than other carboxylic acids, such as OA and DA, because of its lower pK a . The PLQY of a dispersion of the as-prepared NCs was readily enhanced from 72.2 to 90.1% by PFDA addition. This increase could be attributed to a decrease in surface defects by effective surface   58 Cs 2 CO 3 (0.163 g), PbO (0.223 g), and OA (5 mL) were mixed and heated at 160°C and then dehydrated for 30 min at 120°C. After adding toluene (5 mL), the obtained Cs-Pb precursor solution was sealed and stored. One milliliter of this solution was mixed with toluene (15 mL) in a glass vessel with vigorous stirring at room temperature. A Br precursor solution containing tetra-n-octylammonium bromide (0.055 g), OA (5 mL) and toluene (2 mL) was swiftly added to the glass vessel to synthesize the CsPbBr 3 NCs. After 10 s, the NCs were precipitated by adding acetone (50 mL) and then collected by centrifugation at ∼8000g (8500 rpm using a rotor with a diameter of 10 cm) for 5 min, followed by redispersion into toluene under ultrasonication and stirring to prepare a toluene dispersion of the CsPbBr 3 NCs. This dispersion was named as OA-NCs.
4.3. Sample Preparation for Stability Experiments. DA-NCs, PFDA-NCs, VA-NCs, and SA-NCs were prepared by adding DA, PFDA, VA, and SA, respectively, to OA-NCs at 0.06 mmol L −1 . The prepared NC dispersions were sealed and stored under ambient conditions in the dark. To evaluate the thermal stability, the NC dispersions were heated at 100°C for 4 h in an incubator (HB-100, Taitec) with shaking at 60 rpm. To evaluate the stability against polar solvents, 400 μL of ethanol or ethyl acetate as a polar solvent was added to 3.1 mL of the OA-NCs, DA-NCs, and PFDA-NCs.

Characterization.
The XRD profiles were obtained with an X-ray diffractometer (Rint-2200, Rigaku) with a Cu Kα radiation source and monochromator. For the XRD measurements, the centrifuged NCs were vacuum dried overnight. The elemental composition was measured using an XRF analyzer (ZSXmini II, Rigaku). The morphologies were observed by a field emission TEM (Tecnai G 2 , FEI). TEM samples were prepared by vacuum drying a drop of the NC dispersion on carbon-reinforced collodion-coated copper grids (COL-C10, Oken Shoji) overnight. The UV−vis absorption spectra of the NC dispersions were measured using a UV/visible/near-infrared optical absorption spectrometer (V-750, JASCO). Herein, for analysis at the same NC concentration, the net absorbance of the as-prepared samples before heating and adding a polar solvent at 400 nm was adjusted to 0.35, corresponding to 0.8 g L −1 of NC concentration. The absorbance data shown in this work are the net values obtained by subtracting the blank data for the pure solvent without NCs from the sample data. The PL spectra of the NC dispersions were measured using a fluorescence spectrometer (FP-6500, JASCO). Each spectral response was calibrated using an ethylene glycol solution of rhodamine B (5.5 g L −1 ) and a standard light source (ESC-333, JASCO). The absolute PLQYs were measured using a quantum efficiency measurement system (QE-2000 311C, Otsuka Electronics). The PL decay curves were measured using a fluorescence lifetime spectrometer (Quantaurus-Tau C11367, Hamamatsu Photonics) equipped with 405 nm LEDs as the light source.

* S Supporting Information
The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.9b03472.
Chemical structures of DA and PFDA ( Figure S1); XRD profile of as-synthesized CsPbBr 3 NC powder ( Figure  S2); optical properties of the as-prepared OA-NC ( Figure S3); analysis result of the PL decay curve in Figure S3 (Table S1); changes in Tauc plots during heating ( Figure S4); particle size distributions from Figure 5 ( Figure S5); changes in PL spectra of the VA-NCs and SA-NCs during heating ( Figure S6); particle size distributions from Figure 8 ( Figure S7); evaluation of the durability against ethyl acetate; changes in sample appearance after the addition of ethyl acetate ( Figure  S8); changes in UV−vis spectra after the addition of ethyl acetate ( Figure S9); changes in PL properties after the addition of ethyl acetate ( Figure S10); analysis results for the PL decay curves in Figure S10 (

Notes
The authors declare no competing financial interest.

ACS Omega
Article