Regulation of the luminescence mechanism of two-dimensional tin halide perovskites

Two-dimensional (2D) Sn-based perovskites are a kind of non-toxic environment-friendly luminescent material. However, the research on the luminescence mechanism of this type of perovskite is still very controversial, which greatly limits the further improvement and application of the luminescence performance. At present, the focus of controversy is defects and phonon scattering rates. In this work, we combine the organic cation control engineering with temperature-dependent transient absorption spectroscopy to systematically study the interband exciton relaxation pathways in layered A2SnI4 (A = PEA+, BA+, HA+, and OA+) structures. It is revealed that exciton-phonon scattering and exciton-defect scattering have different effects on exciton relaxation. Our study further confirms that the deformation potential scattering by charged defects, not by the non-polar optical phonons, dominates the excitons interband relaxation, which is largely different from the Pb-based perovskites. These results enhance the understanding of the origin of the non-radiative pathway in Sn-based perovskite materials.

"PEA + cations help the sample to form a better surface morphology without pinholes and clear crystal grains" is corrected to ( lines 154-157) "the PEA + cation helps to form large size grains with obvious grain boundaries without pinholes. For BA + , large and discontinuous perovskite islands were formed. For HA + , small size grains with pinholes were observed. For OA + , the top surface became blurry." 2) The interlayer distance is another key factor in controlling the bandgaps of 2D perovskites. As given in Supplementary Table 1, the interlayer distances are ~16.6 Å for PEA, ~13.8 Å for BA, ~16.4 Å for HA, and ~ 18.8 Å for OA. In other words, BA has the smallest interlayer distance and is expected to have a smaller bandgap. In this case, the detailed density functional theory (DFT) calculations are needed to confirm the trend of bandgaps among these four perovskites. This will also support the descriptions given in lines 258-264.
Reply: For two-dimensional perovskites, the different cations can influence the octahedral tilting angle in the inorganic layer, the length of Sn-I bond, and thus modulate the band gap of the material [J. Phys. Chem. Lett. 2018, 9, 3416−3424, ACS Nano 2018, 12, 3321−3332, and J. Phys. Chem. Lett. 2020, 11, 2955−2964]. It turns out in the series of alkylammonium lead iodide monolayers having from 4 up to 10 (and from 10 up to 18) carbon atoms in the alkyl chain, the increased size of the organic barrier does hardly alter band-edge states and band gaps of the electronic band structure computed for the experimental room temperature structures. The band gap jump between the two sets of chain length can be traced back to a phase transition that induces larger out-of-plane octahedral tilting. The influence of cations on the band gap can be explained more comprehensively by theoretical calculations. Through density functional theory calculations ( Supplementary Fig.7) [ACS Nano 2018, 12, 3321−3332 and J. Phys. Chem. Lett. 2020, 11, 2955−2964], we find that the band gap of (PEA) 2 SnI 4 is the smallest and (BA) 2 SnI 4 , (HA) 2 SnI 4 , (OA) 2 SnI 4 are increasing in order, which is consistent with the results we obtained by steady-state spectroscopy. For the alkyl chain group samples, the increase in the length of the alkyl chain increases slightly the electron band gap while the exciton binding energy remains similar. In the meantime, we have revised the relevant contents of the manuscript. In the original manuscript, "Hence, the difference of the organic cations in the layered A 2 SnI 4 (A = PEA + , BA + , HA + , and OA + ) structures is the main factor influencing the composition-dependent bandgap energies. The first is rooted in the difference in the dielectric constant (ɛ) of the organic cationic layer. The relationship between the optical bandgap and the dielectric limiting effect can be described by the following equation where E 0 denotes the energy bandgap without considering the dielectric limitation, whereas ɛ w and ɛ b are the dielectric constants of the inorganic layer and the organic layer, respectively. Since the dielectric constant of PEA + (ɛ PEA = 3.3) is smaller than that of BA + (ɛ BA = 4.3), the optical bandgap of (PEA) 2 SnI 4 is smaller than that of (BA) 2 SnI 4 . For the alkyl chain group samples, the increase in the length of the alkyl chain increases slightly the electron band gap while the exciton binding energy remains similar. Based on studies such as the aforementioned, we can determine that the optical bandgap of the four perovskites ((PEA) 2 SnI 4 , (BA) 2 SnI 4 , (HA) 2 SnI 4 , and (OA) 2 SnI 4 )) increases gradually." is corrected to "For two-dimensional perovskites, the different cations can influence the octahedral tilting angle in the inorganic layer, the length of Sn-I bond, and thus modulate the band gap of the material. The influence of cations on the band gap can be explained more comprehensively by theoretical calculations. Through density functional theory calculations ( Supplementary Fig.7) [ACS Nano 2018, 12, 3321−3332 and J. Phys. Chem. Lett. 2020, 11, 2955−2964], we find that the band gap of (PEA) 2 SnI 4 is the smallest and (BA) 2 SnI 4 , (HA) 2 SnI 4 , (OA) 2 SnI 4 are increasing in order, which is consistent with the results we obtained by steady-state spectroscopy. For the alkyl chain group samples, the increase in the length of the alkyl chain increases slightly the electron band gap while the exciton binding energy remains similar." (lines 213-222) However, for the A 2 SnI 4 structure material, the VBM mainly comprised the 5s orbital of Sn and the 5p orbital of I, and the conduction band minimum mainly comprised the empty 5p orbital of Sn. A decrease in the lattice constant strengthens the interaction between the 5s orbitals of Sn and the 5p orbitals of I, which results in an increase in the VB width and an increase in the VB energy. 3) The authors assigned second absorption peak at ~520 nm to the charge transfer (CT) transition between the organic spacer and the inorganic layers. It is weird to see the CT state located above the band edge; if it is true, the authors need to further analyze their TA spectra to prove the charge transfer state and clarify whether this CT state becomes nonradiative channel for decreasing the PLQYs of four 2D perovskites. I also think that this feature at 520 nm could be attributed to intraband transitions.
Reply: Many thanks to the reviewers for their comments, which made us realize that we incorrectly attributed the absorption peak at ~520 nm to the charge transfer (CT) transition between the organic spacer and the inorganic layers, and we attributed the absorption peak at ~520 nm to the intraband transition of SnI 4 inorganic layers by combining literature study with the results of pump energy-dependent transient absorption (TA) spectroscopy. Since A 2 SnI 4 perovskite (A = PEA + , BA + , HA + , and OA + ) have similar linear absorption and TA spectra characteristics, we selected (PEA) 2 SnI 4 as a representative for an explanation. The details are shown below: In two-dimensional layered halide organic perovskites (LHOPs), the perovskite layer is the dominant component of band-edge absorption. Therefore, a way to demonstrate energy transfer would be the observation of triplet emission from the organic spacer layer. ] When the lowest excitation energy (E) of the exciton in the perovskite layer aligns with the first triplet (T 1 ) excitation energy of the exciton in the organic layer, charge transfer from the perovskite to the organic layer may occur. After transfer, the T1 excitation energy in the organic layer relaxes to a lower T 1 * energy due to enhanced short-range atomic deformation, thus reaching optimal triplet molecular geometry [Nano Lett. 2019, 19, 8732−8740]. Through theoretical calculations, Neukirch et al. systematically studied organic spacer and perovskite layer pairings for possible transfer of the Wannier excitons from the inorganic perovskite lattice to spin-triplet Frenkel excitons located on the organic cations and successfully identify ten organic spacer candidates for possible pairing with perovskite layers of specific halide composition to achieve triplet light emission across the visible energy range. From their calculations, it is clear that the T1 energy of PEA + remains in a narrow range between 4.43 and 4.46 eV, which is greater than the lowest optical excitation peak energy (2.5 eV) of perovskite layer in the (PEA) 2 PbI 4 . The absorption spectrum shows that the lowest optical excitation peak energy of SnI 4 in (PEA) 2 SnI 4 is smaller than that of PbI 4 in (PEA) 2 PbI 4 (2.02 eV vs 2.5 eV), so we can conclude that the exciton in SnI 4 cannot be transferred to organic cation PEA + . For the alkylammonium chains, it can not be excited by a photon in the visible region [Solid State Commun. 1989, 69, 933-936], so we can also conclude that the exciton in SnI 4 cannot be transferred to organic cation the alkylammonium chains.
In the pump energy-dependent TA spectroscopy experiments, where we used the wavelength of a pump at 400 nm, the wavelength of the pump at 613 nm with resonant band-edge absorption, and the wavelength of the pump at 630 nm below the band gap, we found the existence of photobleaching peaks at ~523 nm (PB1) and ~613 nm (PB2) in the TA spectra of these three different pump energies ( Supplementary Fig. 2.). The relaxation kinetics of A 2 SnI 4 (A = PEA + , BA + , HA + , and OA + ) obtain by low pump fluence have been fitted globally with three components ( Figure 6). We find that PB1 has the same relaxation characteristics and lifetime with PB2, and hence it is not consistent with the occurrence of the CT transition. In Supplementary Fig. 2, the two PB peaks are generated almost simultaneously when excited at 400 nm. However, the PB1 peak first reaches the maximum and then decreases, and the PB2 peak reaches a maximum with a delay of about 0.1 ps compared to the PB1 peak, which is more consistent with the intraband transitions [Science 2013, 342, 344-347.].
Combined with the above analysis, the PB1 peak is not the charge transfer transition between the organic spacer and the inorganic layers, but an intraband transitions process in the perovskite layer in the (PEA) 2 SnI 4 . Therefore, we revised the original incorrect description in the manuscript as follows: In the original manuscript, "the second peak at 520 nm (2.38 eV) was assigned to the charge transfer transition between the organic spacer cations and the inorganic layers" is corrected to "The second peak at 520 nm (2.38 eV) was assigned to the intraband transitions process in perovskite layer rather than the charge transfer transition between the organic spacer cations and the inorganic layers in the (PEA) 2 Fig. 2), we find that the photobleaching peaks at ~523 nm (PB1) and ~613 nm (PB2) have the same characteristics and lifetime of the relaxation decay process, and no new bleaching peaks appear during the decay relaxation of PB1. Despite the PB2 peak reaches a maximum with a delay of about 0.1 ps compared to the PB1 peak excited at 400 nm in Supplementary Fig. 3, which is more consistent with the intraband transitions while not consistent with the occurrence of the CT transition. Reply: Thanks for the reviewer's comment to help us improve the reliability of our UPS experimental results. We optimized the UPS measurement conditions to improve the signal-to-noise ratio of valence band edges and increase the signal intensity to make the UPS data for valence band edges more reliable as shown in Supplementary  Figure 8. At the same time, we revised the information in the manuscript as follows: In the original manuscript, "the energy differences between the top of the VBs and the Fermi levels (EF) are 1.62, 0.9 0.95, and 0.87 eV for (PEA) 2 SnI 4 , (BA) 2 SnI 4 , (HA) 2 SnI 4 , and (OA) 2 SnI 4 samples, respectively." is corrected to "the energy differences between the top of the VBs and the Fermi levels (EF) are 1.01, 0.66, 0.77, and 0.53 eV for (PEA) 2 SnI 4 , (BA) 2 SnI 4 , (HA) 2 SnI 4 , and (OA) 2 SnI 4 samples, respectively." (lines 258-259) 5) Although the authors emphasized that PEA cations have a stronger ability to protect Sn 2+ from oxidation than the organic alkyl chain spacers, the mechanism behind it is completely missing. Moreover, it is not clear whether the oxidation process occurs during the film fabrication or during the measurements exposed to air?

Supplementary
Reply: Thanks for the reviewer's comment. Oxygen and water molecules in the environment are the main sources for the decomposition of perovskite crystals. The insertion of large molecules of ammonium organic ions, such as alkyl-ammonium chain (BA + ) and phenylethylammonium (PEA + ), into perovskite improves the stability of perovskite by virtue of its own hydrophobicity. The main reason is that the large molecules of ammonium organic ions are beneficial for the formation of the compact pinhole-free films and block moisture ingress at the boundaries of perovskite nanolayers [J. Am. Chem. Soc. 2017, 139, 6693−6699]. The PEA + also has higher intrinsic thermodynamic stability for Sn perovskites with respect to the oxidation disproportionation channel. Angelis et al. investigated the beneficial effects of large cations (BA + and PEA + ) on the tin stability at the surface and shown that large cation dipoles of the 2D perovskites modulate tin oxidation potential by hindering the formation of tin vacancies and the degradation of the material. BA + ions increased the defect formation energy of Sn 4+ by 0.33 eV, while PEA + could increase the defect formation energy of Sn 4+ by 0.6 eV [J. Phys. Chem. C 2021, 125, 10901-10908]. Therefore, compared to BA + , PEA + effectively hinders the formation of tin vacancies and tin oxidation. Although the samples are prepared in a glove box, which is almost a nitrogen environment, there is still a trace amount of oxygen that allows the Sn 2+ oxidize to Sn 4+ . Therefore, we have added to the manuscript to further explain that PEA + cations have a stronger ability to protect Sn 2+ from oxidation than the organic alkyl chain spacers, as follows. "The main reason is that the large molecules of ammonium organic ions are beneficial for the formation of the compact pinhole-free films and block moisture ingress at the boundaries of perovskite nanolayers [J. Am. Chem. Soc. 2017, 139, 6693−6699]. BA + ions increased the defect formation energy of Sn 4+ by 0.33 eV, while PEA + could increase the defect formation energy of Sn 4+ by 0.6 eV [J. Phys. Chem. C 2021, 125, 10901-10908]. Therefore, compared to BA + , PEA + effectively hinders the formation of tin vacancies and tin oxidation." (lines 248-253) 6) The low temperature XRD measurements of HA + and OA + are also needed to confirm the phase transition around 200 K.
Reply: Many thanks to the reviewer for their critical and high-quality comments. In order to answer the origin of this trap state, we have re-done systematically the temperature-dependent, excitation-fluence-dependent, and excitation-energy-dependent transient absorption experiments and temperature-dependent PL experiments, combined with a more thorough study of the literature. The following conclusions have been obtained as shown: The first fast process I of transient absorption measurement is attributed to the combination of the defects trapping excitons process and the band-gap renormalization process induced by hot excitons, i.e., the defects trapping excitons process play a leading role at low excitation fluence and as the excitation fluence increases, the bandgap renormalization process induced by hot excitons dominates in the process I. The main reasons are as follows.
(1) This trap state in the process I almost is the chemical defect state in the material and not the intrinsic self-trapped exciton (STE) state caused by the exciton-phonon coupling. In the previous section of the manuscript, we demonstrated that excitonic contribution to the PL is dominant in the materials by excitation fluence-dependent integral PL in (PEA) 2 SnI 4 , (BA) 2 SnI 4 , (HA) 2 SnI 4 , and (OA) 2 SnI 4 materials. Due to the transient and linear absorption spectra of the (PEA) 2 SnI 4 , (BA) 2 SnI 4 , (HA) 2 SnI 4 , and (OA) 2 SnI 4 samples have similar characteristics, and (PEA) 2 SnI 4 has no phase transition in the temperature range of 300K-77K, the treatment of the relaxation process of (PEA) 2 SnI 4 is used as a representation obtain a more detailed elucidation. In the temperature-dependent PL experiment, using the multi-peak fitting methods mentioned earlier in the manuscript, the ratio of the PL percentage of the free excitons (P FE ) to that of the trap state excitons (P TE ) below the band gap decrease with decreasing temperature ( Supplementary Fig. 16a), which is obviously opposite to the feature of the intrinsic STE states emission. More precisely, it may be the emission of the extrinsic STE states [J. Phys. Chem. Lett. 2016, 7, 2258−2263 and Nat. Commun. 2020, 11, 2344]. The PL of the four materials((PEA) 2 SnI 4 , (BA) 2 SnI 4 , (HA) 2 SnI 4 , and (OA) 2 SnI 4 ) shows asymmetry at 77 K ( Supplementary Fig. 11), where PL peaks under the band gap exist in a wide range of trailing feature. This trailing spectral width is more and more serious in (BA) 2 SnI 4 , (HA) 2 SnI 4 , and (OA) 2 SnI 4 , especially in (HA) 2 SnI 4 , and (OA) 2 SnI 4 at 77 K when the phase transition causes PL blue shift with PL peak still existing at 640-750 nm, which is more obvious in (OA) 2 SnI 4 . This broad-spectrum PL trailing phenomenon can be explained by the extrinsic STE effect. Deschler et al. indicated that the broad emission below the optical gap seen at low temperatures in <001> oriented 2D-perovskite materials was due to the light-induced formation of localized trap states, associated with interstitial iodide and iodide Frenkel defects that act as color centers in the crystal [J. Am. Chem. Soc. 2017, 139, 18632-18639]. Besides, Loi et al also highlighted that the extrinsic origin of their broad band emission in-gap states in the crystal bulk was responsible for the broad emission [Nat. Commun. 2020, 11, 2344]. There is a broad (640-750 nm) and weak bleach feature below the optical gap, consistent with the defect state being filled in the transient absorption spectrum for the four materials (Fig. 5). The process of filling bleaching of these defect states is the same synchronous step as the first process I of band-edge exciton relaxation ( Fig. 6b) so that the band-edge excitons are transferred to the in-gap defect states. The electron-phonon coupling effect with defect state trapping could lead to the extrinsic STE in the four materials. The exciton-phonon coupling of (OA) 2 SnI 4 is the strongest among the four materials, so there is a relatively wide PL trailing below the band gap, and this PL trailing is more pronounced at 77 K ( Supplementary Fig. 11).
To further demonstrate the fact that excitons are trapped by chemical defects, we applied the stoichiometry engineering of the cations where the PEAI:SnI 2 ratio is 2.6:1 in (PEA) 2 SnI 4 (PEAI-rich) [Nat. Commun. 2020, 11, 2344, and Adv. Funct. Mater. 2020, 30, 1907505]. We reduce the defect density in PEAI-rich sample to improve the PLQY of the PEAI-rich sample to 2.2% ( Supplementary  Fig. 11b). We found that the occupancy ratio and relaxation rate of the first relaxation process I of the band-edge exciton relaxation processes in PEAI-rich were significantly lower than those of (PEA) 2 SnI 4 sample (Fig. 6c). In addition, at high excitation intensities, the bleaching signal intensity at 676 nm is weaker and the relaxation rate is slower for PEA-rich samples compared to (PEA) 2 SnI 4 . (Supplementary Fig. 16c) These results from the transient absorption spectrum provide more evidence that the exciton capture process of the first process I is the defect trapping process [J. Am. Chem. Soc. 2015, 137, 2089−2096]. Therefore, combining the above results, we attribute the first relaxation process mainly to the defect trapping process.
(2) In the transient absorption spectrum (Fig. 5), the band-edge exciton (~614 nm) shows a photobleaching signal attributed to state filling, i.e., the presence of band gap excitons generated by the pump pulse blocking of the optical transition induced by the probe pulse. Within 1 ps, we observe a redshift of the photobleaching peak of the band-edge exciton leading to photoinduced absorption, which is attributed to the bandgap renormalization caused by the hot excitons, which can also cause process I. Assuming an exciton Bohr radius of 1 nm [Nat.
Commun. 2020, 11, 664.], the exciton saturation density is simply estimated to be 10 14 cm -2 . In the previous manuscript (Fig. 6d), the saturation of process I occur at high excitation fluence of 40 μJ/cm 2 (1.1×10 14 /cm 2 ). This excitation fluence makes it possible for the exciton to fission and for the material to degenerate at high excitation intensities. Therefore, the previous conclusion is incorrect. Therefore, we retested the excitation intensity-dependent transient absorption spectrum experiment and found that process I does not be a little saturated until the excitation intensity is near the saturation density. To further clearly distinguish which plays a role in the first relaxation process between the defect states trap exciton process and bandgap renormalization process induced by the hot exciton, we further research the fluence-dependent transient absorption spectra of the PEAI-rich and (PEA) 2 SnI 4 . (3) In the excited intensity-dependent transient absorption spectra of the PEAI-rich and (PEA) 2 SnI 4 , We find that the first relaxation process is more significantly affected by the defect density of states at weak excitation fluence, approximately no more than 10 μJ/cm 2 (Fig. 6c). With further increase of the excitation fluence ( Supplementary  Fig. 16d), the first process appears to be the defect state density-independent, the excitation intensity-independent, temperature-independent relaxation process, so that the bandgap renormalization process dominates the process I.
In summary, the first fast process I of transient absorption measurement is attributed to the combination of the defects trapping excitons process and the band-gap renormalization process induced by hot excitons, i.e., the defects trapping excitons process play a leading role at low excitation fluence and as the excitation fluence increases, the bandgap renormalization process induced by hot excitons dominates in the process I. As a result, we revised the manuscript content as follows. In the original manuscript: "When the pump fluence is lower than 15 µJ cm −2 , the proportion and relaxation rate of the I process remain basically unchanged. However, with the further increase in the pump fluence, the relaxation rate and the proportion of the I process decrease, as shown in Fig. 7(c). This may occur because the photogenerated excitons fill the shallow trap state in different degrees and reduce the trapping rate. For the trapping process, it should involve the transfer of photogenerated carriers between energy levels, indicated by the redshift of the bleaching peak in the spectra. To prove this, the relaxation kinetics of different wavelengths in Fig. 7(d) show that the completion of the band edge ground state bleaching (614.8 nm) relaxation process during the delay time range of 0.48-1.14 ps is synchronized with the relaxation process of the PIA signal (624.1 nm), changing from positive PIA to negative bleaching and reaching the maximum. This indicates that the photogenerated excitons undergo a transfer from the band edge energy level to the shallow trap energy level, where the radiative recombination of excitons subsequently occurs, inducing a change in the relaxation kinetics of 624.1 nm from positive to negative absorption. Thus, combining all the above characteristics, component I is attributed to the process of exciton trapping by the shallow trap state, and the trap rate is independent of the temperature but related to the carrier density. However, the bandgap renormalization effect is the result of the competition between the reduction of the electron bandgap, caused by the exchange-correlation potential, which leads to the redshift of the bleaching peak, and the reduction of the exciton binding energy, caused by the dielectric screening effect, which leads to the blueshift of the bleaching peak, of the photogenerated carriers. Hence, this effect should show that the bleaching peak undergoes a redshift-blueshift transition, after which the band filling effect increases the bandgap blueshift gradually. Additionally, the redshift caused by this physical effect should increase as the excitation intensity increases; however, the results are not consistent with this law as they indicate that the redshift initially remains unchanged in the low excitation intensity range below 15 µJ cm −2 and subsequently decreases until disappearing as the excitation intensity further increases to greater than 15 µJ cm −2 , as shown in Supplementary Fig. 10. The change in the band edge bleaching peak position at a delay time of 1.25 ps when the redshift process ends with the excitation intensity is observed ( Fig. 8(a)). When the excitation intensity is lower than 15 µJ cm −2 , the bleaching peak position does not change, and the blueshift occurs and becomes more noticeable as the excitation intensity increases because of the band filling effect (Supplementary Note 3). The blueshift is proportional to n 2/3 , as shown in Fig. 8(a). Moreover, if the component I is caused by the bandgap renormalization process, its maximum value of the PIA should be proportional to n 1/2 . As shown in Fig. 7(e), however, it does not conform to this law. Thus, the bandgap renormalization has little effect on the (I) process, which may play a role in the process of bleaching peak generation. This is mainly because the laser pump light pulse width of 350 fs is compared with the generation time of the bleaching peak, i.e., the processes of photogenerated carrier thermalization and exciton formation. Additionally, the optical Stark effect can be excluded because the redshift mainly increases with the excitation intensity. Since the trap state in 2D perovskite materials is induced by electron-phonon coupling and has weak optical transition intensities, it can be seen from the above that the density of trapped states is the lowest in the (PEA) 2 SnI 4 sample, so the trapping rate of excitons and the proportion in relaxation dynamics are the lowest (Fig. 6 and Fig. 7(f)), which is consistent with our experimental data." was revised to "For the trap states in two-dimensional perovskites, we need to define the nature of the trap states, i.e., whether they are intrinsic self-trapped exciton (STE) state, defect trapping state, or the extrinsic STE state. In the temperature-dependent PL experiment, using the multi-peak fitting methods mentioned in the previous content, the ratio of the PL percentage of the free excitons (P FE ) to that of the trap state excitons (P TE ) below the band gap decreases with decreasing temperature ( Supplementary Fig. 16a), which is obviously opposite to the feature of the intrinsic STE states emission, in which the stronger luminescence from the intrinsic STE state and the band edge exciton luminescence intensity decreases with the temperature decreases, mainly because the thermal activation of the detrapping process can not meet requirements of the detrapping barrier and the self-  Fig. 16b). We found that the occupancy ratio and relaxation rate of the first relaxation process I of the band-edge exciton relaxation processes in PEAI-rich were significantly lower than those of (PEA) 2 SnI 4 sample (Fig.  6c). In addition, at high excitation intensities, the bleaching signal intensity at 676 nm is weaker and the relaxation rate is slower for PEA-rich compared to (PEA) 2 SnI 4 ( Supplementary Fig. 16c). In summary, the process I contains the process of defect states trapping exciton. In particular, the trailing spectral width of PL is more and more serious in (BA) 2 SnI 4 , (HA) 2 SnI 4 , and (OA) 2 SnI 4 , especially in (HA) 2 SnI 4 , and (OA) 2 SnI 4 at 77 K when the phase transition causes blueshift of PL with existing PL trailing at 640-700 nm, and (OA) 2 SnI 4 is more obvious (Supplementary Fig. 11). This broad-spectrum PL trailing phenomenon could be explained by the extrinsic STE effect. Deschler et al. indicated that the broad emission below the optical gap seen at low temperatures in <001> oriented 2D-perovskite materials was due to the light-induced formation of localized trap states, associated with interstitial iodide and iodide Frenkel defects that act as color centers in the crystal [J. Am. Chem. Soc. 2017, 139, 18632-18639]. Besides, Loi et al also highlighted the extrinsic origin of their broad band emission in-gap states in the crystal bulk are responsible for the broad emission [Nat. Commun. 2020, 11, 2344]. The electron-phonon coupling effect with defect states trapping in the four materials may give rise to the extrinsic STE. The exciton-phonon coupling of (OA) 2 SnI 4 is the strongest among the four materials, so there is a relatively wide PL trailing below the band gap, and the PL trailing is more pronounced at 77 K. Despite the relaxation rate and the proportion of the I process decrease at the high pump fluence as shown in Fig. 6d, this feature does not mean that the defect state is filled with the process, and it needs to be considered whether the excited exciton density reaches Mott density and thus allows exciton fission or the material is degenerate by strong light excitation. Assuming an exciton Bohr radius of 1 nm [Nat. Commun. 2020, 11, 664], the exciton saturation density is simply estimated to be 10 14 cm -2 . The saturation of process I only a little occurs at high excitation fluence more than 40 μJ/cm 2 (1.1×10 14 cm -2 ), so the excitation intensity exceeding the saturation density causing the saturation of process I cannot be characterized as a defect state trapping process. In the TA spectrum within 1 ps ( So this result also further confirms the presence of the bandgap recombination process in the process I. To further clearly distinguish which plays a role in the first relaxation process between the defect states trap exciton process and bandgap renormalization process induced by the hot exciton, we further research the fluence-dependent transient absorption spectra of the PEAI-rich and (PEA) 2 SnI 4 . We find that the first relaxation process is more significantly affected by the defect density of states at weak excitation fluence, approximately no more than 10 μJ/cm 2 (Fig. 6c). With further increase of the excitation fluence ( Supplementary Fig. 16d), the first process appears the defect state density-independent, the excitation intensity-independent, temperature-independent relaxation process, so that the bandgap renormalization process dominates process I The change of the band edge bleaching peak position at a delay time of 1.25 ps with the excitation intensity is observed, when the redshift process ends (Fig. 7a). As the excitation intensity is lower than 15 µJ cm −2 , the bleaching peak position does not change, and the blueshift occurs and becomes more noticeable as the excitation intensity increases because of the band filling effect ( In addition, we add content discussion to further confirm the validity of our conclusions. In addition, the relaxation rate of the second component (II) in PEAI-rich with smaller defect density states is slower than that of (PEA) 2 SnI 4 and PEAI-rich has higher PLQY ( Supplementary Fig. S16b), indicating that the effect of deformation potential scattering by charged defects on exciton interband recombination can be weakened by reducing the density of defect states, and improve the PLQY. (lines 639-644) The third component (III) is attributed to defect-assisted excitons recombination, which involves the radiative recombination induced by relatively shallow defect states, where the ratio of the PL the defect states (center at 660 nm) to that of the free exciton(center at 614nm) is about 30% at room temperature ( Supplementary Fig. 16a), and the non-radiative recombination processes induced by deep defects which make the PLQY very low. Besides, the emission range of relatively shallow defect states increases with the enhancement of exciton-phonon scattering interaction among all the four materials (Fig.3, Fig.4, and Supplementary Fig. 11) Supplementary Fig. 16b), which makes the integrated area of the bleaching peak relaxation process of trap states center at 676 nm of PEA-rich after 1.2 ps smaller than that of PEA. (Supplementary Fig. 16c).
In the original manuscript: "Conversely, the third long relaxation component (III) with a nanosecond lifetime is derived from the trapped state exciton radiative recombination process" was revised to "Conversely, the third long relaxation component (III) with a nanosecond lifetime is attributed to defect-assisted excitons recombination, which involves the radiative recombination induced by relatively shallow defect states, where the ratio of the PL the defect states (center at 660 nm) to that of the free exciton (center at 614nm) is about 30% at room temperature ( Supplementary Fig. 16a), and the non-radiative recombination processes induced by deep defects which make the PLQY very low. Besides, the emission range of relatively shallow defect states increases with the enhancement of exciton-phonon scattering interaction among all the four materials (Fig.3, Fig.4, and Supplementary Fig. 11) In addition, the answer to the question of whether the bleaching peak at 520 nm is a CT or an intraband transition is given in detail in Question 1.
We thank you very much again for your valuable review comments and sincerely hope that our responses will be approved by you.
Reviewer #2 (Remarks to the Author): The authors have studied the emission properties of the 2D tin halide perovskites with different organic cations, based on photoluminescence and transient absorption spectroscopy as function of temperature and power excitation. The understanding of the luminescent mechanisms in lead-free 2 hybrid perovskites is important for the applications. Deciphering the influence of the organic cation on the optical properties is particularly important to guide the optimization of the material. The results are interesting and original. However, the clarity of the discussion and the interpretation of these results could be improved. I have the following remarks: We are very grateful to reviewer for spending valuable time making constructive and useful comments. We will address the reviewer s specific points on the following pages. The comments are underlined in black and the content of our response is highlighted in both blue and red, with those highlighted in red representing additions and corrections in the revised manuscript and Supporting Information. The comments are responded to in a point-by-point manner.
1. Line 161 in the manuscript, the authors claims that the absorption peak at 520 nm has been assigned to charge transfer transition between the organic spacer and the inorganic layers. However, the assumptions on which are based this assignation are not clear. Interestingly, the authors observe a bleaching in transient absorption at this wavelength in addition to the bleaching of the free exciton absorption. I don't think that a similar observation has been made for lead halide 2D perovskites. The authors could provide some comment on that point. Could they resolved the dynamics of the supposed charge transfer transition? Is there any difference between the TA of the charge transfer exciton and the one of the free exciton? It seems surprising that this charge transfer is independent of the choice of the organic cation.
Reply: Many thanks to the reviewers for their comments, which made us realize that we incorrectly attributed the absorption peak at ~520 nm to the charge transfer (CT) transition between the organic spacer and the inorganic layers, and we attributed the absorption peak at ~520 nm to the intraband transition of SnI 4 inorganic layers by combining literature study with the results of pump energy-dependent transient absorption (TA) spectroscopy. Since A 2 SnI 4 perovskite (A = PEA + , BA + , HA + , and OA + ) have similar linear absorption and TA spectra characteristics, we selected (PEA) 2 SnI 4 as a representative for an explanation. The details are shown below: In two-dimensional layered halide organic perovskites (LHOPs), the perovskite layer is the dominant component of band-edge absorption. Therefore, a way to demonstrate energy transfer would be the observation of triplet emission from the organic spacer layer. ] When the lowest excitation energy (E) of the exciton in the perovskite layer aligns with the first triplet (T 1 ) excitation energy of the exciton in the organic layer, charge transfer from the perovskite to the organic layer may occur. After transfer, the T1 excitation energy in the organic layer relaxes to a lower T 1 * energy due to enhanced short-range atomic deformation, thus reaching optimal triplet molecular geometry [Nano Lett. 2019, 19, 8732−8740]. Through theoretical calculations, Neukirch et al. systematically studied organic spacer and perovskite layer pairings for possible transfer of the Wannier excitons from the inorganic perovskite lattice to spin-triplet Frenkel excitons located on the organic cations and successfully identify ten organic spacer candidates for possible pairing with perovskite layers of specific halide composition to achieve triplet light emission across the visible energy range. From their calculations, it is clear that the T1 energy of PEA + remains in a narrow range between 4.43 and 4.46 eV, which is greater than the lowest optical excitation peak energy (2.5 eV) of perovskite layer in the (PEA) 2 PbI 4 . The absorption spectrum shows that the lowest optical excitation peak energy of SnI 4 in (PEA) 2 SnI 4 is smaller than that of PbI 4 in (PEA) 2 PbI 4 (2.02 eV vs 2.5 eV), so we can conclude that the exciton in SnI 4 cannot be transferred to organic cation PEA + . For the alkylammonium chains, it can not be excited by a photon in the visible region [Solid State Commun. 1989, 69, 933-936], so we can also conclude that the exciton in SnI 4 cannot be transferred to organic cation the alkylammonium chains.
In the pump energy-dependent TA spectroscopy experiments, where we used the wavelength of the pump at 400 nm, the wavelength of the pump at 613 nm with resonant band-edge absorption, and the wavelength of the pump at 630 nm below the band gap, we found the existence of photobleaching peaks at ~523 nm (PB1) and ~613 nm (PB2) in the TA spectra of these three different pump energies ( Supplementary Fig. 2.). The relaxation kinetics of A 2 SnI 4 (A = PEA + , BA + , HA + , and OA + ) obtain by low pump fluence have been fitted globally with three components (Supplementary Fig. 14). We find that PB1 has the same relaxation characteristics and lifetime with PB2, and hence it is not consistent with the occurrence of the CT transition. In Supplementary Fig. 2, the two PB peaks are generated almost simultaneously when excited at 400 nm. However, the PB1 peak first reaches the maximum and then decreases, and the PB2 peak reaches a maximum with a delay of about 0.1 ps compared to the PB1 peak, which is more consistent with the intraband transitions [Science 2013, 342, 344-347.].
Combined with the above analysis, the PB1 peak is not the charge transfer transition between the organic spacer and the inorganic layers, but an intraband transitions process in the perovskite layer in the (PEA) 2 SnI 4 . Therefore, we revised the original incorrect description in the manuscript as follows: In the original manuscript, "the second peak at 520 nm (2.38 eV) was assigned to the charge transfer transition between the organic spacer cations and the inorganic layers" is corrected to "the second peak at 520 nm (2.38 eV) was assigned to the intraband transitions process in perovskite layer rather than the charge transfer transition between the organic spacer cations and the inorganic layers in the (PEA) 2 Supplementary Fig. 2), we find that the photobleaching peaks at ~523 nm (PB1) and ~613 nm (PB1) have the same characteristics and lifetime of the relaxation decay process, and no new bleaching peaks appear during the decay relaxation of PB1. Despite the PB2 peak reaches a maximum with a delay of about 0.1 ps compared to the PB1 peak excited at 400 nm in Supplementary Fig. 3, which is more consistent with the intraband transitions while not consistent with the occurrence of the CT transition. Reply: Many thanks to the reviewer for the comment. For the 2D perovskites, the surrounding organic layer with a low dielectric constant is less polarizable and hence decreases the screening of the hole-electron Coulomb interaction. It results in an increase of the exciton binding energy. The discrepancy of dielectric constants ε between the inorganic framework (semiconductor) and the organic layers (surrounding) should be anticipated to give rise to the dielectric confinement on the exciton, which can be modulated by changing the composition of the organic cations [Nat. Commun. 2018, 9, 2254]. The decrease in ε b and/or the increase in ε w would lead to an increase in the 2D exciton binding energy. The PLQY of a layered tin perovskite may thus be further improved by enlarging the dielectric contrast between the tin halide layer and the intercalating ammonium cation. For the (A) 2 SnI 4 (A: PEA + , BA + , HA + , and OA + ), ε A is larger than ε w (w:SnI 4 ), which leads to an enhancement of the Coulomb interaction between the electron and hole pair composing each exciton, which is a consequence of the reduced dielectric screening of the exciton electric field partially located outside the quantum well [Nat Commun. 2018, 9, 2254 and J. Am. Chem. Soc. 2019, 141, 10324−10330]. And ε BA /ε PEA is larger than 1, which makes the exciton binding energy of (BA) 2 SnI 4 is larger than that of (PEA) 2 SnI 4 . In the meantime, we obtained the exciton binding energy by fitting the steady-state absorption spectrum using a more rigorous Elliott theory, in which exciton binding energy of (PEA) 2 SnI 4 is 213±2 meV smaller than that of (PEA) 2 SnI 4 (245 meV) (Supplementary Figure 6)). Plochocka et al. revealed that one of the biggest challenges in perovskites is in determining accurately the exciton binding energy in the material by experiments. The main contradiction lies in the fact that the polar nature of perovskites and the associated polariton effect are neglected [Adv. Energy Mater. 2020, 10, 1903659]. However, such a simple comparison of the dielectric constants of organic cations allows qualitative conclusions to be drawn. The fluorescence quantum yield is not only related to the exciton binding energy, but also the defect density caused by the difference of organic cations. At the same time, we add the relevant parts of the manuscript, as follows. For the (A) 2 SnI 4 (A: PEA + , BA + , HA + , and OA + ), ε A is larger than ε w (w:SnI4), which leads to an enhancement of the Coulomb interaction between the electron and hole pair composing the exciton. It is a consequence of the reduced dielectric screening of the exciton electric field [Nat Commun. 2018, 9, 2254 and J. Am. Chem. Soc. 2019, 141, 10324−10330]. And ε BA (2.2)/ε PEA (3.3) is larger than 1, which makes the exciton binding energy of (BA) 2 SnI 4 is larger than that of (PEA) 2 SnI 4 . In the meantime, we obtained the exciton binding energy by fitting the steady-state absorption spectrum using a more rigorous Elliott theory, in which exciton binding energy of (PEA)2SnI4 is 213±2 meV smaller than that of (PEA) 2 SnI 4 (245 meV) (Supplementary Figure 6)). However, such a simple comparison of the dielectric constants of organic cations can provide qualitative conclusions due to that one of the biggest challenges in perovskites is in determining accurately the exciton binding  Table 1). The values are given without measurement uncertainty and with a precision which seems really overestimated. In particular, for the compounds based on the cation BA, HA and OA, the authors could not present measurements at temperature below 200 K. For the range of temperature fitted here, the equation 2 in the Supplementary note (which needs a correction, for the Bose-Einstein term, kbT is missing) gives an almost linear relation as observed on the figure 4b, c and d above 200K. It seems impossible to separate ΓL0 from EL0 properly in this situation. If we compare the figure 4b and 4d, we observe that the data are very similar. However, the results from the fit are very different, with a factor 2 on ΓL0 between OA and BA.
Reply: Many thanks to the reviewer for the comments. In order to improve the reliability and accuracy of the experimental data, we have performed several temperature-dependent PL experiments under the same experimental conditions, from high temperature down to low temperature, in steps of 25 K. And each temperature was kept for 20 minutes to ensure the temperature stabilization. Also, in order to improve the realism and credibility of the fitting results, we refer to the optical phonon energies obtained from the steady-state Raman experiments (Supplementary Figure 12). And we did not discard the last term of the model representing the inhomogeneous broadening caused by ionized impurities to improve the reasonableness of the exciton-phonon coupling model (Supplementary Note 2) fitting experimental data for the two-dimensional Sn-based properties system we studied. The results of the model fit are shown in Figure 4 in the manuscript, and the fitted parameters were shown in Table 1. Compared with the alkyl chain samples, the (PEA) 2 SnI 4 sample has a relatively smaller Fröhlich coupling intensity (Γ LO ). The results indicate that the (PEA) 2 SnI 4 sample is more ordered, and the non-radiative energy loss of PL is smaller than those of the alkyl chain samples. The main reason is that PEA + cations having the CH-π stacking characteristics that alkyl chain cations lack limit their thermal movement between the inorganic layers and induce weak dynamic changes in the SnI 4 structure. For the samples with an alkyl chain, (HA) 2 SnI 4 (254.9±5 meV) has the lowest Γ LO , followed by (BA) 2 SnI 4 (272.8±4 meV), and finally (OA) 2 SnI 4 (320.2±7 meV), which means that the longer alkyl-ammonium chain tends to enhance the intensity of exciton-phonon scattering. To investigate the relationship between the material structure and electron−phonon interactions, we also used atomic displacement parameters Ueq extracted from single-crystal X-ray diffraction (SCXRD) data corresponding to the four materials ( Supplementary  Information). The results revealed that the atomic displacements of the different atoms of A 2 SnI 4 (A:BA + , HA + , and OA + ) were distinctly larger than those of (PEA) 2 SnI 4 , as shown in Figure 4e and f. Thus, (PEA) 2 SnI 4 has a more rigid structure compared to (BA) 2 SnI 4 , (HA) 2 SnI 4 , and (OA) 2 SnI 4 , which is consistent with the results of the exciton-phonon coupling model and theoretical calculations [J. Phys. Chem. Lett. 2020, 11, 2955−2964]. At the same time, we modified the relevant parts of the manuscript, as follows.
In the original manuscript: "Compared with the alkyl chain group samples, the (PEA) 2 SnI 4 sample not only has a smaller Γ 0 but also has a relatively smaller Fröhlich coupling intensity (Γ LO ), although they have similar optical phonon energies (E LO ). The results indicate that the (PEA) 2 SnI 4 sample is more ordered, and the FWHM and non-radiative energy loss of PL caused by exciton-optical phonon scattering are smaller than those of the alkyl chain samples. The main reason is that PEA + cations having the CH-π stacking characteristics that alkyl chain cations lack limit their thermal movement between the inorganic layers and induce weak dynamic changes in the SnI 4 structure. For the samples with an alkyl chain, (HA) 2 SnI 4 (232.6 meV) has the lowest Γ LO , followed by (BA) 2 SnI 4 (257.94 meV), and finally (OA) 2 SnI 4 (500.58 meV), which means that the longer alkyl-ammonium chain tends to enhance the intensity of exciton-phonon scattering. Notably, the exciton-phonon Fröhlich interaction is over one order of magnitude larger than that reported for lead perovskites (Supplementary Table 4).

The giant carrier phonon coupling can lead to an increase in the FWHM of the PL.
To verify the rationality of the parameters obtained by our fitting, the four samples were subjected to steady-state Raman spectroscopy experiments (Fig. 4(e)). A Raman peak is located at 454.3 cm −1 (56.3 meV) for (PEA) 2 SnI 4 and 472.1 cm −1 (58.5 meV) for the alkyl chain group samples, respectively, which is close to those of the optical phonons obtained by the fitting. The interpretation of these Raman peaks requires strict theoretical calculations, which is beyond the scope of this study." was revised to "In order to improve the realism and credibility of the fitting results, we refer to the optical phonon energies obtained from the steady-state Raman experiments (Supplementary Figure 12). A Raman peak is located at 454.3 cm −1 (56.3 meV) for (PEA) 2 SnI 4 and 472.1 cm −1 (58.5 meV) for the alkyl chain group samples, respectively. The interpretation of these Raman peaks requires strict theoretical calculations, which is beyond the scope of this study." Compared with the alkyl chain samples, the (PEA) 2 SnI 4 sample has a relatively smaller Fröhlich coupling intensity (Γ LO ). The results indicate that the (PEA) 2 SnI 4 sample is more ordered, and the non-radiative energy loss of PL is smaller than those of the alkyl chain samples. The main reason is that PEA + cations having the CH-π stacking characteristics that alkyl chain cations lack limit their thermal movement between the inorganic layers and induce weak dynamic changes in the SnI 4 structure [J. Phys. Chem. Lett. 2020, 11, 2955−2964]. For the samples with an alkyl chain, (HA) 2 SnI 4 (254.9±5 meV) has the lowest Γ LO , followed by (BA) 2 SnI 4 (272.8±4 meV), and finally (OA) 2 SnI 4 (320.2±7 meV), which means that the longer alkyl-ammonium chain tends to enhance the intensity of exciton-phonon scattering. Notably, the exciton-phonon Fröhlich interaction of the 2D Sn-based properties system we studied. is over one order of magnitude larger than that reported for lead perovskites (Supplementary Table 4). The giant exciton -phonon coupling can lead to an increase in the FWHM of the PL and the possibility form self-trapped exciton (STE) states, which further might couple with defective states [Acc. Chem. Res. 2018, 51, 3, 619-627, J. Phys. Chem. Lett. 2016, 7, 2258-2263, and Nat. Commun. 2020, 11, 2344.]. To investigate the relationship between the material structure and electron−phonon interactions, We also studied the atomic displacement parameters Ueq extracted from single-crystal X-ray diffraction (SCXRD) data corresponding to the four materials ( Supplementary  Information). The results revealed that the atomic displacements of the different atoms of A 2 SnI 4 (A: BA + , HA + , and OA + ) were distinctly larger than those of (PEA) 2 SnI 4 , as shown in Figure 4e   There are many methods to determine the exciton binding energy, such as absorption spectrum methods, temperature-dependent PL methods, Magneto-Optical Investigation, etc. However, these methods have insurmountable drawbacks of their own that hinder the experimental simplicity and convenience of indeed being able to excite the true binding energy. In the present work, our accurate determination of the exciton binding energy is not the focus of our study; we need the nature of the optical transition in the material at room temperature to be predominantly the exciton. This is because our intensity-dependent PL experimental results indicate that the nature of the optical transitions in the material is a single-particle radiative recombination feature. Also, we determine the exciton binding energy higher than K B T at room temperature by temperature-dependent PL, such that the result generally concludes that excitons can be stable at room temperature. Determination of exciton binding energy by temperature-dependent PL is mainly using the quenching of the integrated photoluminescence (PL) intensity with the temperature with assuming that the rate of nonradiative recombination is related solely to the thermally activated exciton dissociation [Energy Environ. Sci. 2014, 7, 399 and Nano Lett. 2013, 13, 4505]. The PL intensity was fitted using an Arrhenius formula. However, there are some shallow trap states affecting the fitting accuracy of the fitting method of the Arrhenius formula. In order to reduce the influence of this factor, we performed a split-peak fitting of the PL at each temperature and extracted the sub-PL peaks of the radiation recombination from edge-free excitons. And this sub-PL peak was fitted using the Arrhenius formula and the binding energy of the free exciton was obtained in (PEA) 2 SnI 4 as 43.86± 4 meV (Supplementary Figure 10.). Second, although we use continuous light excitation fluorescence with low excitation intensity, which excludes higher-order intermittent compound processes, it is also difficult to exclude the influence of other compound processes. In addition, we obtained exciton binding energies of about 213 meV for (PEA) 2 SnI 4 using a linear steady-state absorption spectroscopy method in combination with the Tau plot [Nat.
Commun. 2018, 9, 2254] (Supplementary Figure 9), which is bigger than 43.86± 4 meV obtained by the Arrhenius formula. The exciton binding energy obtained by absorption spectroscopy is more trustworthy, mainly because of the distinct exciton absorption peak in the steady-state absorption spectrum and the more pronounced separation from the band edge of the continuum. Although the difference between these two results is relatively large, it still illustrates the optical leap characteristics in the exciton dominant material at room temperature for n=1 2D perovskites [Nat.
Commun. 2018, 9, 2254]. So we made the corresponding changes in the manuscript. In the original manuscript: "Furthermore, the exciton binding energy of (PEA) 2 SnI 4 , obtained by fitting of A peak using Arrhenius relation, is 30.6 meV ( Supplementary Fig. 6(g)), which is greater than the thermal energy (KBT = 25 meV at 300 K) and the same order of magnitude as that for (PEA) 2 SnI 4 (44.9 meV) reported in the literature. Thus, it is a large part of the unseparated excitons that affect the optical properties of the material, despite that only a few excitons split into free carriers, which is consistent with the results of the THz study." was revised to "There are many methods to determine the exciton binding energy, such as absorption spectrum methods, temperature-dependent PL methods, Magneto-Optical Investigation, etc [Nat Commun. 2018, 9, 2254, Adv. Energy Mater. 2020. But the polar nature of perovskites and the associated polariton effect are neglected, which makes the exciton binding energy obtained by different methods under different experimental conditions highly discrepant. [Adv. Energy Mater. 2020, 10, 1903659]. The accuracy of the exciton binding energy obtained by fitting the temperature-dependent PL intensity using the Arrhenius formula is severely affected by recombination processes such as shallow captured-state emission and Auger recombination, so we obtained the exciton binding energy by fitting the steady-state absorption spectrum using a more rigorous Elliott theory [Nat. Commun. 2020 11, 850]. (Supplementary Figure 6.) The exciton binding energy of (PEA) 2 SnI 4 is 213±2 meV greater than the thermal energy (K B T = 25 meV at 300 K), which reveals excitons dominate the nature of optical transitions at room temperature [Nat. Commun. 2018Commun. , 9, 2254." (lines 448-460) The femtosecond optical pumping THz transient absorption spectra can prove the exciton generation and decay in materials on a time scale. Our technology to study THz is not mature yet, but we have been working to optimize the technology of the THz research platform, and we will focus on using THz to study the photogenerated carrier dynamics of two-dimensional tin-based perovskites in our next work. Reply: Many thanks to the reviewer for the critical and high quality comments. In order to answer the origin of this trap state, we have re-done systematically the temperature-dependent, excitation-fluence-dependent, and excitation-energy-dependent transient absorption experiments and temperature-dependent PL experiments, combined with a more thorough study of the literature. The following conclusions have been obtained as shown: The first fast process I of transient absorption measurement is attributed to the combination of the defects trapping excitons process and the band-gap renormalization process induced by hot excitons, i.e., the defects trapping excitons process play a leading role at low excitation fluence and as the excitation fluence increases, the bandgap renormalization process induced by hot excitons dominates in process I. The main reasons are as follows.
(4) This trap state in the process I almost is the chemical defect state in the material and not the intrinsic self-trapped exciton (STE) state caused by the exciton-phonon coupling. has no phase transition in the temperature range of 300K-77K, the treatment of the relaxation process of (PEA) 2 SnI 4 is used as a representation obtain a more detailed elucidation. In the temperature-dependent PL experiment, using the multi-peak fitting methods mentioned earlier in the manuscript, the ratio of the PL percentage of the free excitons (P FE ) to that of the trap state excitons (P TE ) below the bandgap decrease with decreasing temperature ( Supplementary Fig. 16a), which is opposite to the feature of the intrinsic STE states emission. More precisely, it may be the emission of the extrinsic STE states [J. Phys. Chem. Lett. 2016, 7, 2258−2263 and Nat. Commun. 2020, 11, 2344]. The PL of the four materials((PEA) 2 SnI 4 , (BA) 2 SnI 4 , (HA) 2 SnI 4 , and (OA) 2 SnI 4 ) shows asymmetry at 77 K ( Supplementary Fig.11), where PL peaks under the bandgap exist in a wide range of trailing feature. This trailing spectral width is more and more serious in (BA) 2 SnI 4 , (HA) 2 SnI 4 , and (OA) 2 SnI 4 , especially in (HA) 2 SnI 4 , and (OA) 2 SnI 4 at 77 K when the phase transition causes PL blue shift with PL peak still existing at 640-750 nm, which is more obvious in (OA) 2 SnI 4 . This broad-spectrum PL trailing phenomenon can be explained by the extrinsic STE effect. Deschler et al. indicated that the broad emission below the optical gap seen at low temperatures in <001> oriented 2D-perovskite materials was due to the light-induced formation of localized trap states, associated with interstitial iodide and iodide Frenkel defects that act as color centers in the crystal [J. Am. Chem. Soc. 2017, 139, 18632-18639]. Besides Loi et al also highlighted the extrinsic origin of their broadband emission in-gap states in the crystal bulk are responsible for the broad emission [Nat. Commun. 2020, 11, 2344]. There is a broad (640-750 nm) and weak bleach feature below the optical gap consistent with the defect state being filled in the transient absorption spectrum for the four materials (Fig. 5). The process of filling bleaching of these defect states is the same synchronous step as the first process I of band-edge exciton relaxation (Fig.  6b) so that the band-edge excitons are transferred to the in-gap defect states. The electron-phonon coupling effect with defect state trapping could lead to the extrinsic STE in the four materials. The exciton-phonon coupling of (OA) 2 SnI 4 is the strongest among the four materials, so there is a relatively wide PL trailing below the bandgap, and this PL trailing is more pronounced at 77 K ( Supplementary Fig.11).
To further demonstrate the fact that excitons are trapped by chemical defects, we applied the stoichiometry engineering of the cations where the PEAI:SnI 2 ratio is 2.6:1 in (PEA) 2 SnI 4 (PEAI-rich) [Nat. Commun. 2020, 11, 2344, and Adv. Funct. Mater. 2020, we reduce the defect density in PEAI-rich to improve the PLQY of the PEAI-rich to 2.2% ( Supplementary Fig. 11b). We found that the occupancy ratio and relaxation rate of the first relaxation process I of the band-edge exciton relaxation processes in PEAI-rich were significantly lower than those of (PEA) 2 SnI 4 sample (Fig. 6c). In addition, at high excitation intensities, the bleaching signal intensity at 676 nm is weaker and the relaxation rate is slower for PEA-rich compared to (PEA) 2 SnI 4 . (Supplementary Fig.16c) These results from the transient absorption spectrum provide more evidence that the exciton capture process of the first process I is the defect trapping process [J. Am. Chem. Soc. 2015Soc. , 137, 2089Soc. −2096.
Therefore, combining the above results, we attribute the first relaxation process mainly to the defect trapping process. (5) In the transient absorption spectrum (Figure 5), the band-edge exciton (~614 nm) shows a photobleaching signal attributed to state filling, i.e., the presence of bandgap excitons generated by the pump pulse blocking of the optical transition induced by the probe pulse. Within 1 ps, we observe a redshift of the photobleaching peak of the band-edge exciton leading to photoinduced absorption, which is attributed to the bandgap renormalization caused by the hot excitons, which can also cause process I. Assuming an exciton Bohr radius of 1 nm [Nat.
Commun. 2020, 11, 664.], the exciton saturation density is simply estimated to be 10 14 cm -2 . In the previous manuscript (Fig. 6d), the saturation of process I occur at high excitation fluence of 40 μJ/cm 2 (1.1×10 14 /cm 2 ). This excitation fluence makes it possible for the exciton to fission and for the material to degenerate at high excitation intensities. Therefore, the previous conclusion is incorrect. Therefore, we retested the excitation intensity-dependent transient absorption spectrum experiment and found that process I do not be a little saturated until the excitation intensity is near the saturation density. To further clearly distinguish which plays a role in the first relaxation process between the defect states trap exciton process and bandgap renormalization process induced by the hot exciton, we further research the fluence-dependent transient absorption spectra of the PEAI-rich and (PEA) 2 SnI 4 . (6) In the excited intensity-dependent transient absorption spectra of the PEAI-rich and (PEA) 2 SnI 4 , We find that the first relaxation process is more significantly affected by the defect density of states at weak excitation fluence, approximately no more than 10 μJ/cm 2 (Figure 6c). With further increase of the excitation fluence ( Supplementary  Fig.16d), the first process appears the defect state density-independent, the excitation intensity-independent, temperature-independent relaxation process, so that the bandgap renormalization process dominates the process I.
In summary, the first fast process I of transient absorption measurement is attributed to the combination of the defects trapping excitons process and the band-gap renormalization process induced by hot excitons, i.e., the defects trapping excitons process play a leading role at low excitation fluence and as the excitation fluence increases, the bandgap renormalization process induced by hot excitons dominates in the process I. As a result, we revised the manuscript content as follows. In the original manuscript: "When the pump fluence is lower than 15 µJ cm −2 , the proportion and relaxation rate of the I process remain basically unchanged. However, with the further increase in the pump fluence, the relaxation rate and the proportion of the I process decrease, as shown in Fig. 7(c). This may occur because the photogenerated excitons fill the shallow trap state in different degrees and reduce the trapping rate. For the trapping process, it should involve the transfer of photogenerated carriers between energy levels, indicated by the redshift of the bleaching peak in the spectra. To prove this, the relaxation kinetics of different wavelengths in Fig. 7(d) show that the completion of the band edge ground state bleaching (614.8 nm) relaxation process during the delay time range of 0.48-1.14 ps is synchronized with the relaxation process of the PIA signal (624.1 nm), changing from positive PIA to negative bleaching and reaching the maximum. This indicates that the photogenerated excitons undergo a transfer from the band edge energy level to the shallow trap energy level, where the radiative recombination of excitons subsequently occurs, inducing a change in the relaxation kinetics of 624.1 nm from positive to negative absorption. Thus, combining all the above characteristics, component I is attributed to the process of exciton trapping by the shallow trap state, and the trap rate is independent of the temperature but related to the carrier density. However, the bandgap renormalization effect is the result of the competition between the reduction of the electron bandgap, caused by the exchange-correlation potential, which leads to the redshift of the bleaching peak, and the reduction of the exciton binding energy, caused by the dielectric screening effect, which leads to the blueshift of the bleaching peak, of the photogenerated carriers. Hence, this effect should show that the bleaching peak undergoes a redshift-blueshift transition, after which the band filling effect increases the bandgap blueshift gradually. Additionally, the redshift caused by this physical effect should increase as the excitation intensity increases; however, the results are not consistent with this law as they indicate that the redshift initially remains unchanged in the low excitation intensity range below 15 µJ cm −2 and subsequently decreases until disappearing as the excitation intensity further increases to greater than 15 µJ cm −2 , as shown in Supplementary Fig. 10. The change in the band edge bleaching peak position at a delay time of 1.25 ps when the redshift process ends with the excitation intensity is observed (Fig. 8(a)). When the excitation intensity is lower than 15 µJ cm −2 , the bleaching peak position does not change, and the blueshift occurs and becomes more noticeable as the excitation intensity increases because of the band filling effect (Supplementary Note 3). The blueshift is proportional to n 2/3 , as shown in Fig. 8(a). Moreover, if the component I is caused by the bandgap renormalization process, its maximum value of the PIA should be proportional to n 1/2 . As shown in Fig. 7(e), however, it does not conform to this law. Thus, the bandgap renormalization has little effect on the (I) process, which may play a role in the process of bleaching peak generation. This is mainly because the laser pump light pulse width of 350 fs is compared with the generation time of the bleaching peak, i.e., the processes of photogenerated carrier thermalization and exciton formation. Additionally, the optical Stark effect can be excluded because the redshift mainly increases with the excitation intensity. Since the trap state in 2D perovskite materials is induced by electron-phonon coupling and has weak optical transition intensities, it can be seen from the above that the density of trapped states is the lowest in the (PEA) 2 SnI 4 sample, so the trapping rate of excitons and the proportion in relaxation dynamics are the lowest (Fig. 6 and Fig. 7(f)), which is consistent with our experimental data." was revised to "For the trap states in two-dimensional perovskites, we need to define the nature of the trap states, i.e., whether they are intrinsic self-trapped exciton (STE) state, defect trapping state, or the extrinsic STE state. In the temperature-dependent PL experiment, using the multi-peak fitting methods mentioned in the previous content, the ratio of the PL percentage of the free excitons (P FE ) to that of the trap state excitons (P TE ) below the bandgap decreases with decreasing temperature (Supplementary Fig. 16a), which is obviously opposite to the feature of the intrinsic STE states emission, in which the stronger luminescence from the intrinsic STE state and the band edge exciton luminescence intensity decreases with the temperature decreases, mainly because the thermal activation of the detrapping process can not meet requirements of the detrapping barrier and the self- Res. 2018, 51, 3, 619-627] The relaxation kinetics of different wavelengths show that the process of filling bleaching of these defect states (663.4 nm) and the relaxation process of the PIA signal (624.1 nm of PL center) changing from positive PIA to negative bleaching and reaching the maximum are the same synchronous step as the process I of the band-edge exciton (614.8 nm) relaxation (Fig. 6b), revealing that the band-edge excitons are trapped to the in-gap defect states. To further demonstrate the fact that excitons are trapped by chemical defects, we applied the stoichiometry engineering of the cations where the PEAI:SnI 2 ratio is 2.6:1 in (PEA) 2 SnI 4 (PEAI-rich) [Nat. Commun. 2020, 11, 2344, and Adv. Funct. Mater. 2020, 30, 1907505], we reduce the defect density in PEAI-rich to improve the PLQY of the PEAI-rich to 2.2% (Supplementary Fig. 16b). We found that the occupancy ratio and relaxation rate of the first relaxation process I of the band-edge exciton relaxation processes in PEAI-rich were significantly lower than those of (PEA) 2 SnI 4 sample (Fig.  6c). In addition, at high excitation intensities, the bleaching signal intensity at 676 nm is weaker and the relaxation rate is slower for PEA-rich compared to (PEA) 2 SnI 4 ( Supplementary Fig. 16c). In summary, process I contain the process of defect states trapping exciton. In particular, the trailing spectral width of PL is more and more serious in (BA) 2 SnI 4 , (HA) 2 SnI 4 , and (OA) 2 SnI 4 , especially in (HA) 2 SnI 4 , and (OA) 2 SnI 4 at 77 K when the phase transition causes blueshift of PL with existing PL trailing at 640-700 nm, and (OA) 2 SnI 4 is more obvious (Supplementary Fig. 11). This broad-spectrum PL trailing phenomenon could be explained by the extrinsic STE effect. Deschler et al. indicated that the broad emission below the optical gap seen at low temperatures in <001> oriented 2D-perovskite materials was due to the light-induced formation of localized trap states, associated with interstitial iodide and iodide Frenkel defects that act as color centers in the crystal [J. Am. Chem. Soc. 2017, 139, 18632-18639]. Besides, Loi et al also highlighted the extrinsic origin of their broadband emission in-gap states in the crystal bulk are responsible for the broad emission [Nat. Commun. 2020, 11, 2344]. The electron-phonon coupling effect with defect states trapping in the four materials may give rise to the extrinsic STE. The exciton-phonon coupling of (OA) 2 SnI 4 is the strongest among the four materials, so there is a relatively wide PL trailing below the bandgap, and the PL trailing is more pronounced at 77 K. Despite the relaxation rate and the proportion of the I process decrease at the high pump fluence as shown in Fig. 6d, this feature does not mean that the defect state is filled with the process, and it needs to be considered whether the excited exciton density reaches Mott density and thus allows exciton fission or the material is degenerate by strong light excitation. Assuming an exciton Bohr radius of 1 nm [Nat. Commun. 2020, 11, 664], the exciton saturation density is simply estimated to be 10 14 cm -2 . The saturation of process I only a little occurs at high excitation fluence more than 40 μJ/cm 2 (1.1×10 14 cm -2 ), so the excitation intensity exceeding the saturation density causing the saturation of process I cannot be characterized as a defect state trapping process. In the TA spectrum within 1 ps (Fig.  5), the redshift of the photobleaching peak of the band-edge exciton leading to the PIA center at 626 nm appears, which is attributed to the bandgap renormalization So this result also further confirms the presence of the bandgap recombination process in the process I. To further clearly distinguish which plays a role in the first relaxation process between the defect states trap exciton process and bandgap renormalization process induced by the hot exciton, we further research the fluence-dependent transient absorption spectra of the PEAI-rich and (PEA) 2 SnI 4 . We find that the first relaxation process is more significantly affected by the defect density of states at weak excitation fluence, approximately no more than 10 μJ/cm 2 (Fig. 6c). With further increase of the excitation fluence ( Supplementary Fig. 16d), the first process appears the defect state density-independent, the excitation intensity-independent, temperature-independent relaxation process, so that the bandgap renormalization process dominates process I  Supplementary Fig. 12. The change of the band edge bleaching peak position at a delay time of 1.25 ps with the excitation intensity is observed, when the redshift process ends (Fig. 7a). As the excitation intensity is lower than 15 µJ cm −2 , the bleaching peak position does not change, and the blueshift occurs and becomes more noticeable as the excitation intensity increases because of the band filling effect (Supplementary Note 3) and the blueshift is proportional to n 2/3 [Nat Photonics, 2014, 8, 737-743], as shown in Fig. 7a. Additionally, the optical Stark effect can be excluded because we observe the shift well beyond the time duration of the pump laser pulse [J. Am. Chem. Soc. 2015, 137, 2089−2096 and J. Phys. Chem. C 2015, 119, 14714−14721]. Since it can be seen from the above that the density of defect states is the lowest in the (PEA) 2 SnI 4 sample, so the trapping rate of excitons and the proportion of process I in relaxation dynamics are the lowest, which is consistent with our experimental data ( Supplementary Fig. 14 and  Fig. 6f)." (lines 489-577) In addition, we add content discussion to further confirm the validity of our conclusions.
In addition, the relaxation rate of the second component (II) in PEAI-rich with smaller defect density states is slower than that of (PEA) 2 SnI 4 and PEAI-rich has higher PLQY ( Supplementary Fig. S16), indicating that the effect of deformation potential scattering by charged defects on exciton interband recombination can be weakened by reducing the density of defect states, and improve the PLQY. (lines 639-644) 6. The conclusion of the authors is that for tin based 2D perovskite, "it is different from the Pb-based perovskite characterized by high defect tolerance." Line 562 However, the PL quantum yield reported for 2D lead halide perovskite thin films is very low (<1%) (see Yuan et al. Nature Nanotechnology 11 (10) 872-877 (2016) and Duim et al. Advanced Functional Materials, 30 (5) 1907505 (2019) In reality, the PLQY reported in this manuscript for tin based 2D perovskites is much higher than that. The conclusion is not consistent with the results.

Reply:
We are grateful to the reviewers for their comments, which helped us to correct the wrong expressions and perceptions. For the defect tolerance, it is often for three-dimensional Pb-based perovskites, mainly due to the easy migration of ions in the inner of three-dimensional perovskites, which can reduce the non-radiative composite centers [J. Phys. Chem. Lett. 2017, 8, 2, 489-493]. For two-dimensional (2D) Pb-based perovskites (n=1), the defect density of states is relatively high, which greatly increases the probability of non-radiative recombination [Nature Nanotech. 11, 2016, 872-877 and Phys. Rev. Appl. 2014, 2, 034007.]. However, compared to 2D Pb-based perovskites, the oxidation potential of Sn 2+ /Sn 4+ (−0.15 eV) is considerably lower than that of Pb 2+ /Pb 4+ (−1.8 eV) [ACS Energy Lett., 2017, 2, 1089-1098], 2D Sn-based perovskites have higher Sn 4+ defect states than the 2D Pb-based perovskites, which can decrease the PLQY. In addition, the magnitude of PLQY depends on the sample fabrication method and experimental conditions (excitation intensity, temperature), so the PLQY in different literatures comparing with each other is generally not rigorous enough, but to some extent indicates the general characteristics of the problem. In order to rigorously compare the PLQY magnitudes of 2D Pb-based perovskites and 2D Sn-based perovskites, we used the same sample fabrication method and the same experimental conditions, and we obtained that the PLQY of (BA) 2 PbI 4 (0.54%) is higher than that of (BA) 2 SnI 4 (0.15%) ( Supplementary Fig. 16f). This result is consistent with the theoretical results, in which 2D Sn-based perovskites have higher defect states than the 2D Pb-based perovskites decreasing the PLQY. In the previous manuscript, we tested the fluorescence quantum yield of two-dimensional tin-based chalcogenides inaccurately, mainly because the unsuitable test conditions reduced the signal-to-noise ratio of the experimental data and obtained incorrect data. We optimized the experimental conditions ( Figure R1) to improve the signal-to-noise ratio of the experimental data and obtained reasonably correct data in Fig. 1b.
We corrected the experimental data in the original manuscript (Fig. 1b) and added relevant content discussion in the corresponding position in the manuscript ( Supplementary Fig. 16f), as shown: In the original manuscript: "For the 2D Sn-based perovskites, it has higher defect states than 2D Pb-based perovskites." was revised to "For the 2D Sn-based perovskites, it has higher defect states than 2D Pb-based perovskites, obtained from the PLQY measured under the same experimental conditions ( Supplementary Fig. 16f)." (lines 618-621) In the original manuscript: "So it is different from the Pb-based perovskite characterized by high defect tolerance." was revised to "So it is different from the Pb-based perovskite characterized by fewer defect states, in which the main scattering mechanisms for excitons in the scatterings via deformation potential by acoustic and homopolar optical phonons [ACS Nano 2016, 10, 9992−9998]." (lines 679-681) Figure R1. Optimization of experimental conditions for PLQY of (BA) 2 SnI 4 Figure R2. Correction of PLQY experimental data in Figure 1b of the manuscript.