Trifluoroacetate induced small-grained CsPbBr3 perovskite films result in efficient and stable light-emitting devices

Quantum efficiencies of organic-inorganic hybrid lead halide perovskite light-emitting devices (LEDs) have increased significantly, but poor device operational stability still impedes their further development and application. All-inorganic perovskites show better stability than the hybrid counterparts, but the performance of their respective films used in LEDs is limited by the large perovskite grain sizes, which lowers the radiative recombination probability and results in grain boundary related trap states. We realize smooth and pinhole-free, small-grained inorganic perovskite films with improved photoluminescence quantum yield by introducing trifluoroacetate anions to effectively passivate surface defects and control the crystal growth. As a result, efficient green LEDs based on inorganic perovskite films achieve a high current efficiency of 32.0 cd A−1 corresponding to an external quantum efficiency of 10.5%. More importantly, our all-inorganic perovskite LEDs demonstrate a record operational lifetime, with a half-lifetime of over 250 h at an initial luminance of 100 cd m−2.

formation. TFA-anions do not effectively inhibit the ion migration, or the reaction of formation of the perovskite crystal is so easy. If the TFA-anions can decrease the gain size, the PL and absorbance should exhibit blue shift after the increase of the content TFA. But the PL in Fig S8a exhibit an opposite tendency. Thus, the TFA can decrease the gain size and inhibit the ion migration is not so convincing. In the MS, the description of "which have enhanced the EQE values of PeLEDs from less than 1% to 5%" is so obsolete and not proper, now the EQE is over 16%, even 20%.
Reviewer #3 (Remarks to the Author): In this manuscript, the authors used Cs salt (CsTFA) not the traditional CsBr as the Cs source for synthesizing CsPbBr3 film and fabricating light-emitting diodes (LEDs). The results showed that the new Cs salt has delived higher device efficiency (10.5% EQE), and better operational stability (>250 hours). Generally, I think the efficiency shown in this manuscript is not attractive, while the stability achieved is very meaningful because the stability of PeLED is the main concern of this area and the published results just showed several hours stability. The paper could be further considered after addressing the following issues.
1) For traditional CsPbBr3 formation, it is the mixture of CsBr and PbBr2, the formation equation could be CsBr+PbBr2-->CsPbBr3, no matter how the CsBr excess shown in this manuscript. While for the CsTFA proposed in this study, CsTFA+PbBr2-->CsPbBr3 ??? The CsTFA cannot provide any Br source, how the chemical reaction could lead to CsPbBr3 formation?
2) The authors claimed the TFA-existed in grain boundary based on the density functional theory (DFT) calculations. Is there any experimental proof to confirm it? And the authors argued that the device using CsTFA precursor could suppress the ion migration in the perovskite layer, the authors should provide some experimental results. Figure 7f, the devices using CsTFA showed much better operational stability. While from the Figure S11 b and d, these two devices showed almost the same drop off, even the CsTFA based devices showed more serious drop off. As we known, the drop off could reflect the stability of the devices. On this condition, why the CsTFA showed better operational stability?

Reply to Reviewer #1
In the manuscript, the authors have provided new kind of anion molecules for efficient and ultra-stable all-inorganic perovskite light-emitting diodes (PeLEDs). The authors showed astonishingly long operation lifetime of PeLEDs with significantly enhanced EL efficiency. Also, the paper is well written and organized, so we recommend the publication of this paper after minor revision with a few comments as follows.

Reply:
Many thanks for the reviewer's positive comments.

Comment 1#. The device performance and analysis of chemical and structural
properties with CsTFA were well provided. But still, the reason for choosing the specific molecule is unclear. Can the authors strengthen the explanation about molecular structure and chemical property of it?

Reply:
Our rational for choosing this particular molecule can be summarized as follows.
The CsTFA molecule (CF 3 COOCs) can be considered as consisting of two parts: the Cs + cations and the TFAanions (CF 3 COO -). The molecular structure of TFAis given in Figure R1. The strong electronegativity of TFAanions makes it easy for CsTFA to get ionized in polar solvents, which guarantees a high solubility of CsTFA in DMF/DMSO, thus providing abundant Cs + cations for perovskite formation. The solubility of CsBr is on the other hand rather limited, making it challenging to obtain high-quality CsPbBr 3 films. The size of TFAanion (2.38 Å) is larger than that of Br -(1.96 Å), making it difficult to dope TFAinto the CsPbBr 3 crystal structure; at the same time, the O=C-O-groups of TFAcan easily bound to Pb (Adv. Mater. 2018, 30, 1800764), which is helpful for the formation of perovskite films consisting of CsPbBr 3 crystals whose surfaces are passivated by TFAanions, as we demonstrate in this work. The strong electron pulling ability of F results in the uniform distribution of electrons within the TFAanions, which is helpful for the overall stability of this molecular structure, resulting in stable CsPbBr 3 films with an enhanced device performance. These are the factors allowing us to realize dense, smooth, and pinhole-free CsPbBr 3 perovskite films with high thermal stability, whose grain boundaries are well passivated in order to achieve not only a substantial LED performance enhancement but a greatly improved stability. Related discussion has been added to the revised manuscript on Pages 5 and 6. Figure R1. Molecular structure of TFAanion. Elements color coding: O (red), C (grey), and F (purple). Fig 5,

Reply:
We characterized the surface potential of the TFA-derived perovskite film by kelvin probe force microscopy (KPFM), and found that the test results consist well with our suggested energy scheme. The perovskite film surface is directly accessible by the AFM probe to measure the contact potential difference. Figure R2a shows the topography of the film surface, and Figure R2b shows the contact potential difference. Some individual grains are clearly distinguishable, and grain boundaries have higher surface potential than that of the particle interior, which pushes charge carriers away from the grain boundaries and drifts them into particle interior. The topography, contact potential, and their overlapped 3D maps of a 3×3 μm 2 area ( Figure R3a, b, and c) and a 1×1 μm 2 area ( Figure R3d, e, and f) provide information of the contact potential difference between the grain boundaries and the crystals. We note that our contact potential data are different from those previously reported of the pure perovskites, whose grain boundaries typically have lower contact potential values than that within the grains (J. Phys. Chem. Lett. 2015, 6, 875−880). This is because the TFAanions are abundant at the grain boundaries and can push the charges into the grains. Related discussion has been added to the revised manuscript on Pages 16, 41, and 42.

Q3:
Can the authors provide the temperature-dependent photoluminescence (PL) result to compare the exciton binding energy of the perovskite films with or without CsTFA?

Reply:
The CsBr(1.7) and the CsTFA(1.7) samples were chosen for the temperature-dependent photoluminescence (PL) measurement to compare the exciton binding energy difference, as shown in Figure R4. We determined the exciton binding energy (E B ) using the following equation: where I 0 is the emission intensity at 0 K, A is a scaling factor, and k B is the Boltzmann constant. The calculated E B values are 65.5 meV for the CsTFA-derived film and 50.7 meV for the CsBr-derived film. The higher E B for the CsTFA-derived film is mainly due to the smaller crystal sizes, and the formation of larger bandgap/smaller bandgap (grain boundary/grain) structures. The carriers within the CsTFA-derived films are easier to bound together and form excitons, which greatly enhances the PL QY and the device EL efficiency. Related discussion has been added to the revised manuscript

Reply:
Just as the reviewer mentioned, the introduction of CsTFA to perovskite films suppresses the ion migration, which plays an important role in improving the device stability performance. There was already a strong evidence in our manuscript on this point, which is the reduced hysteresis of the TFA-derived CsPbBr 3 LEDs ( Figure R5).
We do not have access to the equipment which would allow us to perform temperature-dependent conductivity measurements, but following the reviewer's suggestion, we have developed another set of experiments to further proof this particular point of the suppressed ion migration, such as described in the following.
Since the bandgap changes over the halogen ratios for mixed-halide perovskites, we can determine the components based on the emission color. Thus, to provide a direct observation of ion migration, the mixed-halide CsPb(Br/I) 3 perovskites were taken, and their emission was traced in the LED structure of ITO/PEDOT:PSS/CsPb(Br/I) 3 /TPBI/LiF/Al. First, the PL evolution over time under a constant voltage (2V, below the turn-on voltage) was investigated. As shown in Figure R6, the PL spectra experience 15 and 6 nm red-shifts occurring within 3 min for CsBr-and CsTFA-derived CsPb(Br/I) 3 devices, respectively. The stronger PL shift of the CsBr-derived device indicates more intense ion migration within the film. In addition, the PL completely disappeared for the CsBr-derived device in 9 min, most probably due to destroyment of perovskite structure or the formation of defects related to the halogen deficiency. In contrast to that, the CsTFA-derived CsPb(Br/I) 3 device was still shining brightly, because the ion migration was suppressed here. Besides, when the applied bias increased from 5 to 8 V, the EL peaks experienced 12 and 4 nm red-shifts for CsBr-and TFA-derived CsPb(Br/I) 3 perovskite LEDs, respectively ( Figure R7), which further demonstrates that the ion migration has been suppressed with the help of TFA ions. Related discussion has been added to the revised manuscript on Pages 19, 44, and 45.  devices under a constant bias of 2 V. PL spectra were recorded from 0 min to 3 min. Figure R7. Normalized EL spectra of (a) the CsBr-and (b) CsTFA-derived CsPb(Br/I) 3 devices. EL spectra were recorded from 5V to 8V.

Q5:
The format for reference citation is not consistent. Please make sure the reference list set in the same format.

Reply:
All reference citations have been corrected to become fully consistent.

Reply to Reviewer #2
In this paper, the author reports efficient and stable inorganic perovskite-based

Reply:
We respectfully disagree with this referee's analysis, and we provide our grounds in the reply below, based on the following point-to-point comparison with the mentioned published work. The referee raised several points in his/her comments, including: 1) whether the device EQE is high enough; 2) whether the device stability is sufficiently enhanced than previously reported results, 3) whether the TFA-derived CsPbBr 3 films should have larger crystal sizes, and 4) whether the ion migration in TFA-derived CsPbBr 3 films is indeed suppressed. Six papers have been mentioned to support some of these points, which we are listing below as follows: Ref  (Figure R8). This is an encouraging development and continuation of the work reported in our original manuscript, which also nicely emphasizes the overall generality of our reported TFA-related approach.  shown excellent stability performance, namely "The authors showed astonishingly long operation lifetime of PeLEDs" and "the stability achieved is very meaningful because the stability of PeLED is the main concern of this area".  Mater. 2014, 26, 6503), and 2) via doping to increase the crystal formation energy and thus to decrease the crystal growth rate (ACS Appl. Mater. Interfaces. 2017, 9, 2403). This is because when decreasing the perovskite crystal growth rate, the nucleation process results in formation of fewer seed crystals, which would consume more precursors, resulting in larger sized crystals. In comparison, if one increases the crystal growth rate, the nucleation process becomes [Redacted] fiercer and forms more seed crystals, which would then consume less precursors, resulting in smaller sized crystals. This has also been reported while introducing an antisolvent to increase the crystal growth rate during the perovskite formation, which has also decreased the crystal size (Science 2015, 350, 1222). Based on these observations, the conclusion is that TFA can indeed lead to fiercer reaction, but the smaller crystals can also be simultaneously obtained in this case which is beneficial to the increased amount of seed crystals.
1. The fourth comment was that introducing TFA ions cannot suppress the ion migration. We have already provided a reply on this very same question 4 of the reviewer 1, which appeared above. Besides, the referee pointed out that the PL peak doesn't keep blue-shifting when increasing the content of the TFA anions.
This is because that even with TFA ions, our perovskite crystal sizes are still much larger than the Bohr diameters of the respective bulk perovskites (Chem. Mater. 2017, 29, 3644), and thus no quantum confinement effects are expected.
Lastly, the description "which have enhanced the EQE values of PeLEDs from less than 1% to 5%" has been revised to outline the true improvement observed.

Q2: The authors claimed the TFA-existed in grain boundary based on the density functional theory (DFT) calculations. Is there any experimental proof to confirm it?
And the authors argued that the device using CsTFA precursor could suppress the ion migration in the perovskite layer, the authors should provide some experimental results.

Reply:
Since the O=C-O-groups can easily bond to the Pb 2+ ions of perovskites (Adv. Figure R9 and Figure R10, and the optical bandgap doesn't become narrower for the TFA-derived films (doping larger sized TFAinto the crystal lattice would narrow the bandgap), we conclude that the TFAions sit at grain boundaries. For the TFA derived films, the TFApassivated grain boundary regions form a larger bandgap than that of the particle interior, which pushes both electrons and holes away from the grain boundaries and let them drift into the grains. Thus, the contact potential difference between the grain boundaries and the grains can help to determine the exact location of TFAions. The surface potential of the TFA-derived perovskite film was characterized the by kelvin probe force microscopy (KPFM). The film surface is directly accessible by the AFM probe to measure the contact potential difference. Figure R2a shows the topography of the film surface, and Figure R2b shows the contact potential difference. Some individual grains are clearly distinguishable and grain boundaries have higher surface potential than that of the particle interior, demonstrating that the TFAdoes exist in grain boundaries. All related discussions have been added to the manuscript on Pages 16, 41 and 42.  About the ion migration suppression, we addressed the same question of the Reviewer 1 in our reply to his/her question #4, see above. The related discussion has been added to the revised manuscript on Pages 19, 44, and 45. Figure 7f, the devices using CsTFA showed much better operational stability. While from the Figure S11 b and d, these two devices showed almost the same drop off, even the CsTFA based devices showed more serious drop off.

Q3: As the authors shown in
As we known, the drop off could reflect the stability of the devices. On this condition, why the CsTFA showed better operational stability?

Reply:
For the perovskite LEDs, it is true that more pronounced efficiency drop-off means more unbalanced charge injection, resulting in both charge accumulation and the efficiency decrease. However, in our case, we need to analyze the efficiency drop-off and the stability performance from two aspects, namely the emitting layer quality and the charge injection barrier. On one hand, as we see from the UPS data (Table R3), the TFA-derived film has a deeper VBM (-5.95 eV) than that of CsBr (-5.87 eV), indicating that the hole injection barrier becomes higher for the TFA-derived LEDs, and thus induces more pronounced efficiency drop-off. On the other hand, as show in Figure R11a, the pin-holes in the CsBr-derived films lead to electrical shunting paths and lower the radiative recombination of the emissive layer, which greatly decreases the device operational stability. In comparison, the TFA-derived films are smooth, compact, and pin-hole free, which greatly enhances their stability ( Figure R11b). That's why even so the TFA-derived LEDs show more obvious efficiency drop-off, they still have better stability performance. It also means that through the proper charge injection optimization or interface engineering, we can further improve the EQE and stability of TFA-derived LEDs.

Reviewers' comments:
Reviewer #1 (Remarks to the Author): In the manuscript, the concerns of the reviewers were adequately addressed in respect of molecular origin of the new precursor's benefit on device performance and luminescent property, and definite evidence on the energy level alignment at the crystal surface after the revision. Also, additional experiment using mixed-halide system clearly showed the suppressed halide ion's migration with CsTFA. We think the manuscript is largely strengthened after the revision and can be accepted by Nature Communications now. With best regards, Tae-Woo Lee Reviewer #2 (Remarks to the Author): The manuscript reports that the trifluoroacetate can significantly improve the performance of all inorganic perovskite based LEDs (a max. EQE of 10.5%, a half-lifetime of over 250 h at an initial luminance of 100 cd m-2). This work presented is interesting and important to some extent, but I think it does not reach the criterion of Nature Communication based on the comments as follows. It lacks significant detail and needs major additions. As for the previous comments, a. The author think the result is high in the all inorganic based CsPbBr3, but the current paper (EQE of >20% , Nature 562, 245, (2018)) shows that the result in the Yang's draft is not so high . In the recent paper (Nature 562, 245, (2018)), Wei et al think the emitter is all inorganic based CsPbBr3, the MABr is only on the surface of the emitters. Meanwhile, there are plenty of organic TFA in this work, similar to the paper in Nature with organic MABr. However, the above published result is 2-fold higher than this paper. b. As for the stability, the authors think the result is compatible the EQE with the stability. But the author still misses the result (EQE of 12.98% and the operating lifetimes T50 with an initial brightness of 1000 cd/m2 is 173 hours. Thus, T50 @100cd/m2 is exceeding 1000 hours in an operating conditions) in the paper (Materials Today Chemistry 10 (2018) 104-111). In addition, some new comments as follows, 1. I do not think that TFA-ions are located at the grain boundaries instead of being doped into the crystal structure of the perovskite. According to the authors' speculation, EDX mapping of the TFAperovskite film should accumulate more proportion of F elements at the boundaries of the films, which should not be homogeneous distribution as shown in Figure S7. And the crystal and lattice structure of the film need be further probed. This is particularly important when considering the characterization methods, such as the TEM. 2. Although the author gives some explanation about the stability of the device enhanced by the TFA, the expression is not so convinced. As we all known, the TFA is a short organic materials. Its own stability is not very good, compared to Cs. So the evidence of the enhancement of the stability should be strengthened in the MS.
3. The authenticity of the cross-sectional view is heavily masked by the added color, the contact surface between PEDOT: PSS and perovskite are so smooth. These would misunderstand the readers. The author should revise it. 4. The introduction of FA and MA in perovskite films increased the performance of corresponding devices, while if it can reduce the stability as both MA and FA are extremely moisture-sensitive. Further, please analysis the reason for the development of device performance with FA and MA, especially the beneficial effect of CsTFA. 5. Author says that the charge carriers within the CsTFA-derived films are easier to bound together and form excitons, but author do not prove that the excitons are effective for radiative recombination, as excitons may have many behaviors, such as Auger recombination. 6. Whether such a thick perovskite emitter ~200 nm will adversely affect the electrical transport properties of the device. 7. The authors claimed that the PL and EL shifts of mixed-halide CsPb(Br/I)3 device caused by different time and voltage are attributed to ion migration, and how the authors determined that this phenomenon was not caused by phase separation. Other characterization technique such as TOF need be further probed. 8. This manuscript needs to be revised in more detail, there are still some errors. For example: i: In the sentence "The strong interaction between DMSO with Pb2+ ions and Cs+ ions is evidenced from the Fourier transform infrared (FTIR) spectroscopy (Fig. 1b) The authors have almost addressed my concerns, but the explanation for improving the device stability is still not very convincing, especially the response for the Q3 mentioned by the Reviewer 3#. The paper could be accpeted for publication in present form, but it could be much better if further improvement could be done.

Reply to Reviewer #1
In the manuscript, the concerns of the reviewers were adequately addressed in respect of molecular origin of the new precursor's benefit on device performance and luminescent property, and definite evidence on the energy level alignment at the crystal surface after the revision. Also, additional experiment using mixed-halide system clearly showed the suppressed halide ion's migration with CsTFA. We think the manuscript is largely strengthened after the revision and can be accepted by Nature Communications now.
With best regards,

Reply:
We highly appreciate Prof. Lee's positive comments; it was a pleasure to have you among our reviewers.

Reply to Reviewer #2
The manuscript reports that the trifluoroacetate can significantly improve the performance of all inorganic perovskite based LEDs (a max. EQE of 10.5%, a half-lifetime of over 250 h at an initial luminance of 100 cd m -2 ). This work presented is interesting and important to some extent, but I think it does not reach the criterion of Nature Communication based on the comments as follows. It lacks significant detail and needs major additions.

Reply:
We appreciate these additional comments of the reviewer which have helped us to further improve this manuscript. In the following, we are addressing all these comments and we hope that the reviewer would appreciate our efforts.
a. The author think the result is high in the all inorganic based CsPbBr 3 , but the current paper (EQE of >20%, Nature 562, 245, (2018)) shows that the result in the Yang's draft is not so high. In the recent paper (Nature 562, 245, (2018)

Reply:
The emitter in Wei's devices was indeed inorganic CsPbBr 3 , but of a rather different kind, namely a CsPbBr 3 /MABr quasi-core/shell structure (please see the description in their paper, Figure R1). The MABr is an ionic compound, which is not very stable. For our case, although the organic TFA is used to improve the quality of inorganic perovskite films and suppress the ion migration, the small molecule TFA with a covalent bond is more stable than the ionic compound MABr. Note that organic LEDs (OLEDs) with small-molecule functional materials have been already commercialized (Nature communications 9, 3207 (2018).). Therefore, our devices show more than 2-fold better stability compared with Wei's devices (250h vs 105h), and even longer device stability could be obtained by further improving the device performance. Moreover, the efficiency of Wei's devices is not twofold higher than in our work, as we also obtained very similar efficiency by adjusting the perovskite composition (20.3%-EQE for Wei's devices, 10.5%-EQE for our TFA-derived CsPbBr 3 devices and 17.0%-EQE for our TFA-derived FA 0.11 MA 0.10 Cs 0.79 PbBr 3 devices). Figure R1 from the Wei's Nature paper. [Redacted]

Reply:
Our work is quite different from the above paper (Materials Today Chemistry 10   (2018) 104-111). We have focused on the improvement in the device stability by enhancing the film quality of inorganic perovskite emitter in a standard organic-inorganic hybrid device structure. On the contrary, the operating lifetime improvement in the mentioned Teridi's study is by using an all-inorganic device structure ( Figure R2). Inorganic charge transport materials are generally more stable than organic materials, so that it is unfair to make such a comparison. In addition, in their work, they have used an organic-inorganic hybrid perovskite as emitter and achieved a 12.98%-EQE, while we have obtained a higher EQE of 17% when the organic-inorganic hybrid perovskite (FA 0.11 MA 0.10 Cs 0.79 PbBr 3 ) is used. Our method so far still possesses distinct merits as compared with Teridi's study. In addition, some new comments as follows, homogeneous distribution as shown in Figure S7. And the crystal and lattice structure of the film need be further probed. This is particularly important when considering the characterization methods, such as the TEM. Reply: Due to the resolution limit of SEM, the distribution for the F element may appear homogeneous on the EDX maps of the TFA-derived perovskite films. It is very well known fact that interpretation of the EDX elemental analysis must be taken with a great care, as this method is far from exact. In addition to this single method, we have carefully confirmed the TFA distribution theoretically and experimentally via the density functional theory (DFT) calculation, the Tauc plots, the XRD results, and the kelvin probe force microscopy (KPFM), with a solid conclusion that the TFAions are indeed accumulated at the grain boundaries, as we summarize once again in details below.
From the DFT calculations ( Figure R3), the band gap of the CsPbBr 3 films should increase with the increase of TFA content if the TFAions are doped into the perovskite crystal structure. However, our experimental data show that the band gap of perovskite films with TFA is almost the same with that of perovskite films without TFA ( Figure R4). Moreover, XRD patterns show that no diffraction peak from TFA-derived films shifts to smaller angles, indicating that TFAions are also not doped into the perovskite crystal structure (R TFA >R Br ) ( Figure R5).   The distribution of TFA has been further confirmed by measuring the contact potential difference using kelvin probe force microscopy (KPFM) as shown in Figure   R6. Some individual grains are clearly distinguishable, and grain boundaries have higher surface potential than that of the particle interior, which pushes charge carriers away from the grain boundaries and drifts them into particle interior. The topography, contact potential, and their overlapped 3D maps of a 3×3 μm 2 area (Figure R7a, b, and c) and a 1×1 μm 2 area (Figure R7d, e, and f) provide information of the contact potential difference between the grain boundaries and the crystals. We note that our contact potential data are different from those previously reported for the pure perovskites, whose grain boundaries typically have lower contact potential values than that within the grains (J. Phys. Chem. Lett. 2015, 6, 875−880). This is because the TFAanions are abundant at the grain boundaries and can push the charges into the grains. Therefore, we can conclude that the TFAions are accumulated at the grain boundaries.  2. Although the author gives some explanation about the stability of the device enhanced by the TFA, the expression is not so convinced. As we all known, the TFA is a short organic materials. Its own stability is not very good, compared to Cs. So the evidence of the enhancement of the stability should be strengthened in the MS.

Reply:
The TFA is a small organic molecule with a covalent bond, which is relatively stable as compared with many ionic compounds constituting some forms of the perovskites, such as MABr we already mentioned above. In this work, we put forward the point that the ion-migration induced phase-separation results in a severe performance deterioration of PeLEDs. As the TFA ions are able to accumulate at the grain boundaries of perovskite films, this suppresses the ion-migration, and thus enhances the device stability. Please refer to Figures R8-10.

Reply:
We agree with this point, and we now have provided original TEM image without using of any artificial colorings. Please see Figure R12 (Fig.6a in the revised manuscript).   Sci. 2016Sci. , 9, 1989Sci. −1997 The reasons for the performance enhancement are as follows: i) Doping with larger A-site cations will make the tolerance factor closer to 1, resulting in a more stable device structure and less lattice distortion related trap states (Scientific Reports. 2016, 6, 23592); and ii) introducing organic component into inorganic perovskites will change their conductivity from n-type to near-ambipolar, resulting in more balanced charge transport and flatter band conditions under operation.
We have also have shown that our strategy of controlling the perovskite grain growth using TFA is effective not only for inorganic perovskite films but also for organic-inorganic hybrid perovskite films. After TFA treatment, the stability of FA 0.11 MA 0.10 Cs 0.79 PbBr 3 LEDs has been obviously improved as compared with FA 0.11 MA 0.10 Cs 0.79 PbBr 3 LEDs without TFA treatment.

5.
Author says that the charge carriers within the CsTFA-derived films are easier to bound together and form excitons, but author do not prove that the excitons are effective for radiative recombination, as excitons may have many behaviors, such as Auger recombination.

Reply:
The flatter energy landscape and better optical properties ( Figure R13) of CsTFA-derived films show that most of the excitons decay over radiative recombination channel. Several previous studies have shown, that perovskite nanocrystals are less prone to the Auger recombination (Nano letters, 2016, 16, 6425), which is not much dependent on the excitation power (Angewandte Chemie, 2015,