Unveiling the additive-assisted oriented growth of perovskite crystallite for high performance light-emitting diodes

Solution-processed metal halide perovskites have been recognized as one of the most promising semiconductors, with applications in light-emitting diodes (LEDs), solar cells and lasers. Various additives have been widely used in perovskite precursor solutions, aiming to improve the formed perovskite film quality through passivating defects and controlling the crystallinity. The additive’s role of defect passivation has been intensively investigated, while a deep understanding of how additives influence the crystallization process of perovskites is lacking. Here, we reveal a general additive-assisted crystal formation pathway for FAPbI3 perovskite with vertical orientation, by tracking the chemical interaction in the precursor solution and crystallographic evolution during the film formation process. The resulting understanding motivates us to use a new additive with multi-functional groups, 2-(2-(2-Aminoethoxy)ethoxy)acetic acid, which can facilitate the orientated growth of perovskite and passivate defects, leading to perovskite layer with high crystallinity and low defect density and thereby record-high performance NIR perovskite LEDs (~800 nm emission peak, a peak external quantum efficiency of 22.2% with enhanced stability).

In this manuscript, Cao et al. delivered a record-high performance near-infrared PeLED with EQE of 22.2%. The authors systematically investigated the different additive-assisted chemical interactions in the perovskite precursor solutions, and thoroughly compared the different additiveassisted crystallization mechanism for FAPbI3 perovskites. Based on these, they applied a new additive with multi-functional groups to facilitate the oriented growth of perovskite and passivate defects, and thus reported a perovskite film with high crystallinity and low defect density. The work provides a reasonable additive-assisted crystal formation pathway for FAPbI3 perovskites, which will be useful to researchers in related fields. Overall, the manuscript is well conceived and well-written, the characterizations can support the conclusion. Accordingly, the reviewer recommends the manuscript published in Nature Communications after the following issues have been addressed 1. The authors claimed that additives with multi-functional groups can facilitate the vertical orientation of perovskite and passivate defects, resulting in highly efficient PeLEDs. However, there is no clear evidence to confirm that the vertical orientation of 3D FAPbI3 perovskite is a key factor in achieving high-efficiency PeLEDs. Therefore, the author should provide more details to explain why the vertical orientation is highly related with device performance. 2. In Line 64, the authors should attach related reference and add more explanation to demonstrate the relationship between the preferentially perpendicular orientation to substrate and the device performance. 3. In Line 171-172, considering the functions of -NH2 and -COOH groups, the amino acid additives, such as 5-AVA (used in the authors' previous work, Nature 562, 249-253 (2018)), will be supposed to be beneficial for obtaining high-efficiency device. Why the authors selected AEAA? Do the additional two oxygen (O) atoms have some extra effect on the crystallization or passivation? The authors should provide more detailed measurement to demonstrate this issue. 4. Xu et al. reported that oxygen (O) atoms in the amino-functionalized additives (EDEA) would reduce the hydrogen-bonding ability (Nat. Photonics 13, 418-424 (2019)). The structure of AEAA is quite similar with the EDEA, and the only difference is that the one of the end groups (-NH2) was replaced by carboxyl (-COOH). Thus, it is important to emphasize the distinctiveness of AEAA.
Reviewer #5 (Remarks to the Author): The present manuscript unveils the perovskite crystallization process and the additive role in achieving the perpendicular nanocrystals growth in favor of developing efficient Light emitting diodes (LEDs). Amine, carboxylic acid and amine + carboxylic acid functional group additives influence on 3D perovskite crystallization and passivation were optimized and characterized to form efficient The quality of the manuscript is good, despite it possess some research gaps and lack of scientific evidences with respect to author's claims. Major revisions needed along with considerable scientific evidences and discussions. Followings have to be addressed, 1. Why emission of designed LEDs is limited to 800nm. Could authors employ this crystallization directing additives in achieving blue LEDs? 2. The achieved EQE of 19.4% is good, how such achieved EQE can be related with crystallization and passivation alone? Is there any other possible scientific reasons behind the EQE achievement? 3. Is vertical growth of perovskite 3D crystals merely related to the H-bond interaction alone? If so, why the authors couldn't achieve the directional growth with other amine additives? 4. EQE values are still confusing, in abstract it is stated as 19.4% whereas in introduction it is mentioned as 22.2%. Authors should finalize their EQE value and discuss with plausible explanation. 5. I suggest plotting luminance vs voltage along with radiance, so that it would be better to contrast the LED performance with other literary works. 6. It is vital to frame the comparison table for the present study with recently published literatures in terms of luminance, EQE, current efficiency, and stability to pronounce the importance. 7. Various amino group passivating agents were presented in the supporting figure, among which why do authors fix the certain agent on basis of what criteria? From XRD plots, it certainly shows the crystallization differences. Authors should compare the crystallization of the mentioned passivation agents with various doping ratios because one particular doping condition is insufficient in understanding their role in crystallization. 8. Along with the various amino passivating agent doping ratios, crystallite size monitoring with XRD peaks is demanding to reveal the crystallization effects. 9. By the way, the as-compared carboxylic acid agent effects on the morphology and crystalline features also lags with various doping ratios. The data obtained with various passivating agents remains unsatisfactory and it remains complex with no clear outputs. 10. Why authors considering only the end group effects while choosing passivating agents. What will be the effects of the other mid chain groups on perovskite crystallization? 11. The GIWAXS data recorded for PAM and PAC time varies for 7,12s and in particular, why the time variation observed with PAM and PAC agents? 12. Why particularly stronger hydrogen bond exists between PAM and FAI? If so other amine additives failed to exhibits such characters, why it occurs predominant in this case? 13. Why titration procedure for FAI equivalent is fixed to 6 and why PbI2 equivalent is fixed to 2.2 in observing the NMR studies? 14. How the result do suggests the chemical interaction extent for different amine additives? How to scale the interaction factor by the additives? According to author's discussion, the interaction remains intact, irrespective of chain length aromatic and aliphatic nature of the additives? Kindly recheck the FTIR peaks and discuss with more scientific backgrounds for audience better understanding. 15. Does only the chemical interaction plays role in deciding the out of plane deformation. It is strange that there is no structural and geometric conformational contribution from the added components. 16. The scheme didn't work satisfactory in clearing the selective vertical growth of perovskite. It is highly recommended that addition of chemical equations in achieving the molecular ordering stage 1, 2, and 3. How do authors affirm the release of FA+ in stage 3. The total Figure 4 looks like a mystery as it didn't give more scientific insight with the process involved along with the characterization evidences. 17. Authors claims that Carboxylic acid group influence in crystallization is little. The interaction between carboxyl and lead is evidenced with XPS peaks. Why such interaction didn't influence the crystallization, whereas PAM interaction influenced the crystallization significantly. Moreover, I'm curious on the selective passivation on the iodine site instead of other sites present in the perovskite matrices. 18. The device stability results are compromising with glass-epoxy encapsulation. What would be the stability of LED device without any encapsulation, such figures can add value to the research carried out. 19. The light outcoupling efficiency attainment elevates the efficiency. How such outcoupling contributions were studied. The methodology wasn't provided and the characterization regarding those remains vague. 20. Lack of studies such as PL, lifetime, EL data and how the optimized color purity achieved with respect to crystallization control. Those data along with narrative flow and collective recent references can enhance the quality considerably. 21. How does the simple traditional 3D perovskite blending methodology worked efficiently, as there are numerous existing methodologies (2D, quasi 2D, nanowires, nanorods and quantum dots) works influential in achieving highly efficient and stable LEDs. How do authors will employ this strategy in establishing their future works? It is mandatory to describe the bottlenecks associated and how it can be solved in the upcoming works?
After careful consideration and reviewing process, I suggest major revisions on the above for better understanding and clarifications. All the best for the revision.

Comment #1: In this manuscript, Cao et al. delivered a record-high performance near-infrared PeLED
with EQE of 22.2%. The authors systematically investigated the different additive-assisted chemical interactions in the perovskite precursor solutions, and thoroughly compared the different additiveassisted crystallization mechanism for FAPbI3 perovskites. Based on these, they applied a new additive with multi-functional groups to facilitate the oriented growth of perovskite and passivate defects, and thus reported a perovskite film with high crystallinity and low defect density. The work provides a reasonable additive-assisted crystal formation pathway for FAPbI3 perovskites, which will be useful to researchers in related fields. Overall, the manuscript is well conceived and well-written, the characterizations can support the conclusion. Accordingly, the reviewer recommends the manuscript published in Nature Communications after the following issues have been addressed.

Response:
We thank the reviewer for recognizing the importance of our work and for the constructive comments.

Comment #2:
The authors claimed that additives with multi-functional groups can facilitate the vertical orientation of perovskite and passivate defects, resulting in highly efficient PeLEDs. However, there is no clear evidence to confirm that the vertical orientation of 3D FAPbI3 perovskite is a key factor in achieving high-efficiency PeLEDs. Therefore, the author should provide more details to explain why the vertical orientation is highly related with device performance.

Response:
We thank the reviewer for this comment. The high device performance can be mainly attributed to the low defect density in perovskite film. It has been reported that the sub-grain homogeneity and less grain boundary can reduce the defect density (Science 367, 1352-1358 (2020)), so the formation of orientated FAPbI3 perovskite with high crystallinity can lead to improved device performance. We have added this in the manuscript (Line 195 to 196, Page 10, highlighted).

Comment #3:
In Line 64, the authors should attach related reference and add more explanation to demonstrate the relationship between the preferentially perpendicular orientation to substrate and the device performance.

Response:
We thank the reviewer for this suggestion. We have added this in the revised manuscript (Line 65 to 66, Page 4, highlighted).

Comment #4:
In Line 171-172, considering the functions of -NH2 and -COOH groups, the amino acid additives, such as 5-AVA (used in the authors' previous work, Nature 562, 249-253 (2018)), will be supposed to be beneficial for obtaining high-efficiency device. Why the authors selected AEAA? Do the additional two oxygen (O) atoms have some extra effect on the crystallization or passivation? The authors should provide more detailed measurement to demonstrate this issue.
Response: Besides the high crystallinity and defect passivation induced by the -NH2 and -COOH groups, the ether group can form a chelate ring with carboxyl group and unsaturation Pb, which better passivate iodine-vacancy defects. We have added this in the manuscript (Line 199 to 204, Page 10, highlighted).

Comment #5: Xu et al. reported that oxygen (O) atoms in the amino-functionalized additives (EDEA)
would reduce the hydrogen-bonding ability (Nat. Photonics 13, 418-424 (2019)). The structure of AEAA is quite similar with the EDEA, and the only difference is that the one of the end groups  was replaced by carboxyl (-COOH). Thus, it is important to emphasize the distinctiveness of AEAA.

Response:
We thank the reviewer for this comment. We have added this in the manuscript (Line 176 to 177, Page 9 and Line 205 to 207, Page 10, highlighted).

Reviewer #5:
Comment #1: The present manuscript unveils the perovskite crystallization process and the additive role in achieving the perpendicular nanocrystals growth in favor of developing efficient Light emitting diodes (LEDs). Amine, carboxylic acid and amine + carboxylic acid functional group additives influence on 3D perovskite crystallization and passivation were optimized and characterized to form efficient.
The quality of the manuscript is good, despite it possess some research gaps and lack of scientific evidences with respect to author's claims. Major revisions needed along with considerable scientific evidences and discussions.

Response:
We thank the reviewer for the positive comments.
Comment #2: Why emission of designed LEDs is limited to 800 nm. Could authors employ this crystallization directing additives in achieving blue LEDs?

Response:
We thank the reviewer for this comment. We have tried to fabricate 3D FAPb(BrxCl1-x)3 blue perovskites. As the mixed-halide blue perovskites suffer from serious phase segregation and poor morphology, it is difficult to evaluate the effect of additives.
Nevertheless, we have employed this approach to red perovskite LEDs. We have introduced AEAA to the FA0.47Cs0.53Pb (I0.87Br0.13)3 perovskite with emission peak at 693 nm. The XRD data show that the inclusion of AEAA can significantly enhance the crystallinity of red perovskite film (see the below figure). Furthermore, the inclusion of 0.5-ratio AEAA significantly increases the peak EQE of FA0.47Cs0.53Pb (I0.87Br0.13)3 LEDs from 0.1% to 5.1% and the half-lifetime from 1 to 20 min, without further optimization of the device fabrication process.  Response: Firstly, if we assume the variation of device outcoupling efficiency is negligible, the EQE is mainly related to the PLQE of perovskite films. The inclusion of AEAA additive results in a record PLQE of ~80% compared with other 3D perovskite films, which is due to the low trap density close to that of perovskite single crystals (see the time-resolved PL measurements in Supplementary Figure   12, Science 367, 1352-1358 (2020)). Secondly, the defect density of perovskite film is mainly related to the crystallinity and surface passivation effect (Science 367, 1352(Science 367, -1358(Science 367, (2020; Nature 555, 497-501 (2018)). Therefore, we believe that the enhanced EQE of AEAA-based perovskite LED can be attributed to the crystallization and passivation induced by amino, carboxyl, ether groups.

Response:
We thank the reviewer for this comment. We have changed the EQE to 22.2% in abstract.
Comment #6: I suggest plotting luminance vs voltage along with radiance, so that it would be better to contrast the LED performance with other literary works.
Response: For the visible LEDs, luminance is useful for human eyes to define the brightness of LEDs.
However, the EL peak of our LEDs locates in the NIR regime (~800 nm). We are hard to understand the point of plotting luminance vs voltage. Comment #10: By the way, the as-compared carboxylic acid agent effects on the morphology and crystalline features also lags with various doping ratios. The data obtained with various passivating agents remains unsatisfactory and it remains complex with no clear outputs.

Response:
As responded to the Comment #8, here we want to show the general effect for acid additives, which almost has no impact on the crystallinity of FAPbI3. Moreover, the investigation of PAC doping ratio has shown that it will not cause significant change in the morphology and crystalline features of perovskite layers (Supplementary Figure 3).

Comment #11:
Why authors considering only the end group effects while choosing passivating agents.
What will be the effects of the other mid chain groups on perovskite crystallization?
Response: Firstly, the 1 H NMR measurements show that the mid chain group has no interaction with the perovskite precursor. Secondly, we have compared amino/acid additives with different chain lengths and molecular structures, and different end-group additives with the same mid chain (Supplementary Figure 2). It shows that the end group plays the key role on the crystallization process.  and FA, the deficit of free FA + in the precursor solution will suppress the crystallization of perovskite.
In contrast, there is weak interaction between PAC and FA, so the PAC-based perovskite forms earlier than the PAM-based sample. We have added this in the revised manuscript (Line 143, Page 7, highlighted). How to scale the interaction factor by the additives? According to author's discussion, the interaction remains intact, irrespective of chain length aromatic and aliphatic nature of the additives? Kindly recheck the FTIR peaks and discuss with more scientific backgrounds for audience better understanding.

Response:
We thank the reviewer for this suggestion.
Firstly, we have shown that the amine additives have stronger hydrogen bond with FAI than other functional-group additives, irrespective of chain length, aromatic and aliphatic nature of the additives (see the response to Comment #13).
Secondly, as FTIR measurement is not surface-sensitive, and it is difficult to determine the weak signal from trace material in the surface of perovskite films, we carried out more surface-sensitive XPS measurements. We find that the signal of Pb 4f state in various amine additive samples moves to higher binding energy (see the below figure), which is due to the effect of amine group in determining the crystal growth process, leading to perovskite film with fewer defects.
Comment #16: Does only the chemical interaction plays role in deciding the out of plane deformation.
It is strange that there is no structural and geometric conformational contribution from the added components.
Response: As shown in the response to Comments #4 and 13, the change of size and geometry of additives will cause some variation of perovskite layers, but the chemical interaction between the end group of additive and FAI plays the key role in the crystallization of perovskites.

Comment #17:
The scheme didn't work satisfactory in clearing the selective vertical growth of perovskite. It is highly recommended that addition of chemical equations in achieving the molecular ordering stage 1, 2, and 3. How do authors affirm the release of FA + in stage 3. The total Figure 4 looks like a mystery as it didn't give more scientific insight with the process involved along with the characterization evidences.

Response:
The scheme in Figure 4 is based on the 1 H NMR, ESI-TOF MS and in situ GIWAXS Moreover, I'm curious on the selective passivation on the iodine site instead of other sites present in the perovskite matrices.

Response:
We thank the reviewer for this comment.
Firstly, the in situ GIWAXS measurement indicates the crystallinity of the perovskite layers is mainly determined at the very early stage of the spin-coating process (Fig. 2). The 1 H NMR and ESI-TOF MS measurements suggest that the interaction between PAC and perovskite precursor is much weaker than with PAM ( Fig. 3 and Supplementary Fig. 6c), so the PAC has little influence in crystallization, but mainly plays the role of defect passivation through the interaction between carboxyl and unsaturation Pb.
Secondly, the 1 H NMR spectra show that the peak pattern of COOH become sharp in the PAC PbI2 solution, suggesting the weak interaction between PbI2 and PAC (Fig. 3) The author has addressed all my concerns and the quality of the manuscript has been greatly improved. Hence, I recommend the paper publish in Nature Communications as the current version.
Reviewer #5 (Remarks to the Author): The present manuscript unveils the perovskite crystallization process and the additive role in achieving the perpendicular nanocrystals growth in favor of developing efficient Light emitting diodes (LEDs). Amine, carboxylic acid and amine + carboxylic acid functional group additives influence on 3D perovskite crystallization and passivation were optimized and characterized to form efficient LED devices.
The quality of the manuscript is improved with reviewer suggestions. Despite the revision, some of the corrections made wasn't satisfactory and herein, I suggest authors to solve the following before publishing this article, 1. The response provided by authors for scheme correction is not convincing. The scheme didn't work satisfactory in clearing the selective vertical growth of perovskite. It is highly recommended that addition of chemical equations in achieving the molecular ordering stage 1, 2, and 3. 2. Why do the stability peak exhibits initial increments and later it degrades, whereas the control sample device degrades gradually. 3. No PL lifetime values were provided. Authors should contrast the reduction in trap density values with lifetime result values. 4. It is well known from the reported works that chain lengths, aromatic/aliphatic nature and mid chain groups have considerable effects in achieving the betterment in the device performance and stability. In addition to it, DFT results compliance proves the involvement of those effects in achieving good results. In this manuscript, the results and discussion part for the stronger bonding with various additives suggest some contradictory results. It is essential to discuss the effect of those contributions when discussing with the different additives. For authors reference, Nature Photonics, 2021, 15, 148-155. Advanced Materials, 2021, 33, 2007855. Science Advances, 2019, 5. eaax4424. Nature Communications, 2021. Authors claimed that the introduction of AEAA to the FA0.47Cs0.53Pb (I0.87Br0.13)3 perovskite with emission peak at 693 nm. The inclusion of 0.5-ratio AEAA significantly increases the peak EQE of FA0.47Cs0.53Pb (I0.87Br0.13)3 LEDs from 0.1% to 5.1% and the half-lifetime from 1 to 20 min, without further optimization of the device fabrication process. Why the authors failed to achieve the excellence in EQE of the FA0.47Cs0.53Pb (I0.87Br0.13)3 LEDs as achieved in FAPbI3 perovskites LEDs. 6. The point of adding luminance vs voltage curve is to contrast the present study with other reported studies and to understand the turn on voltage characteristics. 7. The crystallite sizes measured using Scherrer's equation of perovskites with various PAM ratios are calculated as 23, 27, 28, 42, 44, 60 nm, respectively. State the Scherrer's equation in the manuscript for the audience clarity. 8. The PLQE values achieved is about 80% and it records high value. I suggest citing some recent references. 9. From the several amine additive studied, authors conclude that induce directional growth of perovskite, exhibiting enhanced crystallinity. The reason for choosing particular PAM additive among other additives still remains vague. 10. To support and strengthen the submitted manuscript, we suggest adding the following recent articles in to your reference list, Nature Photonics, 2018, 12, 681-687. Nature Communications, 2020, 11, 3674. Nature Communications, 2019, 10, 665. Chemical Engineering Journal, 2021