Interplay between Perovskite Magic-Sized Clusters and Amino Lead Halide Molecular Clusters

Recent progress has been made on the synthesis and characterization of metal halide perovskite magic-sized clusters (PMSCs) with ABX3 composition (A = CH3NH3+ or Cs+, B = Pb2+, and X = Cl−, Br−, or I−). However, their mechanism of growth and structure is still not well understood. In our effort to understand their structure and growth, we discovered that a new species can be formed without the CH3NH3+ component, which we name as molecular clusters (MCs). Specifically, CH3NH3PbBr3 PMSCs, with a characteristic absorption peak at 424 nm, are synthesized using PbBr2 and CH3NH3Br as precursors and butylamine (BTYA) and valeric acid (VA) as ligands, while MCs, with an absorption peak at 402 nm, are synthesized using solely PbBr2 and BTYA, without CH3NH3Br. Interestingly, PMSCs are converted spontaneously overtime into MCs. An isosbestic point in their electronic absorption spectra indicates a direct interplay between the PMSCs and MCs. Therefore, we suggest that the MCs are precursors to the PMSCs. From spectroscopic and extended X-ray absorption fine structure (EXAFS) results, we propose some tentative structural models for the MCs. The discovery of the MCs is critical to understanding the growth of PMSCs as well as larger perovskite quantum dots (PQDs) or nanocrystals (PNCs).

Perovskite magic-sized clusters (PMSCs) are particles with a single size or narrow size distribution that are smaller than PNCs but still contain the same perovskite composition [16][17][18][19][20][21]. In addition, PMSCs possess unique optical properties compared to PNCs such as bluer and sharper absorption and emission bands [22,23]. They are also important for understanding the underlying mechanisms of perovskite crystal growth [24][25][26][27][28][29][30][31][32][33][34][35]. However, one major challenge in the study of PMSCs is to determine their structure. Given the ultrasmall size of PMSCs, PMSCs do not have a long enough ordered structure to be characterized by X-ray diffraction and cannot be detected by imaging techniques, such as transmission electron microscopy (TEM). Moreover, due to their instability in polar solvents, PMSCs cannot be measured by a mass spectrometry. Thus far, the only evidence of their presence is their characteristic optical absorption and emission bands, which are sharp, well-defined, and significantly blueshifted with respect to their bulk form [16][17][18][19][20][21].
In our effort to better understand the structure and growth process of PMSCs using CH 3 NH 3 PbBr 3 as an example, we made an important discovery in one of our control experiments that a new species could be generated, without the CH 3 NH 3 + or methylammonium (MA + ), that absorbs bluer (402 nm) than the PMSCs (424 nm) with BTYA and VA as ligands. This species clearly does not have the full perovskite composition since MA + is not present. Therefore, we attribute this new species to be a MC that contains the components Pb 2+ , Br -, and BTYA. We carried out a more detailed study to help shed light on these MCs and their relation to PMSCs. In addition, the MC synthesis was demonstrated not only using PbBr 2 but also was successful with PbCl 2 and PbI 2 precursors. Furthermore, extended X-ray absorption fine structure (EXAFS) studies were conducted to help gain some understanding of their structural compositions.

Results and Discussion
MAPbBr 3 PMSCs with BTYA and VA passivating ligands were synthesized using the reprecipitation method, as described in our previous report [20]. Detail of the MAPbBr 3 -VA-BTYA PMSC synthesis is described in the supporting information (SI). The UV-vis absorption and PL spectra of MAPbBr 3 -VA-BTYA PMSC solution are shown in Figure 1(a), with the major absorption peak at 424 nm and PL peak at 434 nm, as we reported previously [20]. Interestingly, these optical peaks decrease in intensity, and simultaneously, a new absorption peak at 402 nm and PL peak at 410 nm appear with intensity increasing over time. Figure 1(b) shows the change of the absorption spectra overtime. The absorption peak initially increases in intensity and red shifts from 424 nm to 432 nm, indicating an initial growth of the PMSCs. After~14 minutes, the 432 nm peak decreases in intensity while a new peak at 402 nm appears and increases in intensity. After~4.3 h, the 432 nm peak completely disappears, leaving behind the 402 nm peak with intensity similar to that of the original 424 nm peak, and no change is observed thereafter.
Similar observations are made in the PL spectra, as shown in Figure 1(c). The 434 nm PL peak increases in intensity in the first~14 minutes, and then begins to decrease. As the 434 nm PL peak's intensity decreases, a new 410 nm PL peak appears and increases in intensity until the complete disappearance of the 434 nm PL peak. The changes in both the absorption and PL spectra are observed with an isosbestic point at 419 and 425 nm, respectively. This suggests that there is an equilibrium between the two species, and that they are mechanistically related. This phenomenon has been previously reported for CdSe nanoparticles [36]. When an excess amount of BTYA was added to a~1.6 nm CdSe nanoparticle solution, a broad absorption band centered at 445 nm was converted into a very narrow absorption band which centered at 414 nm. In this previous report, Cd-Se bonds are broken until a thermodynamically stable size or structure configurates. Similarly, in the present work, Pb-Br bonds in the PMSCs may break following similar principles as the CdSe nanoparticles. In our previous study, we suggested that the 402 nm absorption peak and 410 nm PL peak are derived from smaller PMSCs that fragmented from a larger PMSC. However, in the present study, we found through control experiments that the 402 nm peak is not due to PMSCs but can be attributed to molecular clusters (MCs), as explained next.
In order to elucidate the structure and growth mechanism of this new species absorbing at 402 nm, labeled as MC402, we designed and conducted a set of control experiments. First, we investigated the effect of performing the synthesis without the A component (MA + ) in the ABX 3 perovskite. We performed the same reprecipitation reaction using only PbBr 2 , VA, and BTYA precursor components. Detail of the PbBr 2 -VA-BTYA synthesis is described in the SI. As shown in Figure 2, a sharp and strong absorption band peaked at 402 nm is observed, along with a PL peak at 410 nm. The PLQY of the PbBr 2 -VA-BTYA sample is determined to be 23 ± 5% (vs quinine sulfate, 58%), which is slightly higher than the previously reported MAPbBr 3 -VA-BTYA PMSCs (19 ± 3%). The FWHM of the absorption band is 11 nm, while that of the PL band is 10 nm. Compared to MAPbBr 3 -VA-BTYA PMSCs at their initial optical spectra, these bandwidths are significantly narrower. However, their peak positions as well as their bandwidths are very similar or the same as that of the MAPbBr 3 -VA-BTYA PMSCs after they are transformed into the new species absorbing at 402 nm with PL at 410 nm. This indicates that the new species transformed from the original MAPbBr 3 -VA-BTYA PMSCs is the same as the product synthesized using PbBr 2 -VA-BTYA without the MA. At this point, because VA protonates BTYA to produce BTYA + , we assumed that BTYA + could be playing the role of the A cation in the ABX 3 perovskite structure. However, further experiments show that this is not the case. The bluer absorption and PL peaks of the MCs compared to PMSCs are largely due to their smaller size or more localized electronic wave functions, essentially quantum confinement.
As another important control experiment, only PbBr 2 and BTYA components were added to the precursor solution and reprecipitated, following the same method of synthesis. In this case, without the VA component, BTYA cannot be protonated, and therefore, there is no A cation component present in the precursor solution. Interestingly and surprisingly, the PbBr 2 -BTYA sample exhibits the same characteristic optical spectra as PbBr 2 -VA-BTYA or the final transformed product of MAPbBr 3 -VA-BTYA PMSCs. In addition, we extended the synthesis from PbBr 2 to PbCl 2 and PbI 2 as the precursor to mix with BTYA. Their UV-vis absorption and PL spectra of the synthesized products are shown in Figures 3(a) and 3(b). The absorption peaks are 335, 402, and 515 nm, while the PL bands are centered at 344, 410, and 519 nm for PbCl 2 -BTYA, PbBr 2 -BTYA, and PbI 2 -BTYA samples, respectively. The dependence of the peak position on the halide is evidence that the halide is part of the structure of the MCs. The FWHM of their absorption bands is 18, 15, and 13 nm, respectively, while the FWHM of their PL bands is all around 10 nm. These significantly blueshifted peak positions, with respect to their perovskite bulk form or NCs, and very narrow optical bandwidths indicate the presence of ultrasmall and well-defined clusters for each PbX 2 -BTYA sample. For PbCl 2 -BTYA, in addition to the sharp absorption band peaked at 335 nm, there is a broad background that is likely due to larger aggregated structures. While the PLQY of PbCl 2 -BTYA is very low (>0.1%), indicating a high density of trap states, the PLQY for PbBr 2 -BTYA and PbI 2 -BTYA is much higher at 26 ± 6% and 11 ± 3%, respectively.
The interesting question is how the different samples, MAPbBr 3 -VA-BTYA, PbBr 2 -VA-BTYA, and PbBr 2 -BTYA, all show the same 402 nm absorption and 410 nm PL bands? Without the A cation component for PbBr 2 -VA-BTYA, and without ammonium cation for PbBr 2 -BTYA, we can only attribute the species responsible for these optical bands to molecular clusters (MCs) rather than PMSCs since they clearly do not have the full ABX 3 composition. Because their optical properties only depend on and originate from the Pb 2+ , Br -, and BTYA components, the PbBr 2 -BTYA MCs are thermodynamically more stable since they can form even when methylammonium ions are present to compete to form PMSCs.
To confirm the presence and role of the BTYA ligand, FTIR spectra were collected for PbCl 2 -BTYA, PbBr 2 -BTYA, and PbI 2 -BTYA MCs as well as free unbound BTYA ligands. As shown in Figure S1, there is a shift in N-H 2 bending frequency from free ligand at 1606 cm -1 to 1586, 1579, and 1571 cm -1 for PbCl 2 -BTYA, PbBr 2 -BTYA, and PbI 2 -BTYA MCs, respectively. This is consistent with shift to lower frequency when ligand is bound to heavier atoms [37,38]. The N-H 2 bending peak for PbCl 2 -BTYA, PbBr 2 -BTYA,  To elucidate the structure of PbX 2 -BTYA MCs, X-ray absorption spectroscopy measurement was conducted on the Pb L(II) edge, which corresponds to the electron transition from 2p 1/2 to unoccupied d states [39][40][41].     Table 1 summarizes the coordination number (CN) and bond distance (R) of Pb-N and Pb-X bonding from the two-shell fitting results. For each sample, the highest intensity peak corresponds to the Pb-N bonding, providing evidence for the BTYA coordination to Pb 2+ via the lone pair electrons on the amine functional group. The higher Pb-N coordination than Pb-X provides further evidence that the samples represent MCs rather than an ordered perovskite structure. In addition, due to the low Pb-X coordination, the MCs must be positively charged, as the simplest possible structure is 3+ and becomes increasingly positive as the proposed structure increases in size. The simplest proposed model and most closely matched bonding coordination number from an EXAFS two shell fitting for PbCl 2 -BTYA and PbBr 2 -BTYA MCs is shown in Figures 4(a) and 4(b). Two Pb 2+ atoms are bridged by a halogen anion, while the Pb 2+ atoms are also coordinated with 3 BTYA molecules via nitrogen's lone electron pair. Given the relatively large uncertainty in coordination number and the numerous potential coordination numbers of Pb 2+ , an exact model is not possible at this point. However, the real structure should be similar or close to what we have proposed. For PbI 2 -BTYA MCs, the Pb-I coordination is slightly higher at 1:6 ± 0:8 compared to PbCl 2 -BTYA and PbBr 2 -BTYA MCs at 0:9 ± 0:5 and 1:0 ± 1:0, respectively. Therefore, a different model for PbI 2 -BTYA MCs is proposed to match the coordination more directly from the EXAFS fitting, as shown in Figure 4(c). We propose that a PbI 6 is present in the sample. In addition, the Ianion bridges an additional Pb 2+ atom that is terminated by 3 BTYA molecules.
From these structural studies, we illustrate in Figures 5(a) and 5(b) the relationship among the MAPbBr 3 -VA-BTYA PMSCs, PbBr 2 -BTYA MCs, and the initial precursor ions that are solvated or complexed with ligands in solution. In Figure 5(a), as indicated by the red shift in the absorption spectra from 424 to 432 nm, the MAPbBr 3 -VA-BTYA PMSCs initially grow in size. The PMSC structure likely rearranges itself to accommodate for BTYA complexation. The restructured intermediate with an absorption peak at 432 nm finally transforms into the MCs. As shown in Figure 5(b), our results suggest two pathways to forming MCs. The first is direct formation from the precursor ions, with no A component or ammonium ions. The second pathway is to start with precursor ions, form MSC424, convert into MSC432, and finally form MCs. Under the experimental conditions we have explored so far, we observed clear evidence for transformation from PMSCs to MCs, but not from MCs to PMSCs, indicating that MCs are more stable than PMSCs under the current conditions.
To investigate reversibility from MCs back to PMSCs, we added additional MABr to push the equilibrium back to PMSCs. As shown in Figure S2, the UV-vis absorption spectra were measured after the PMSCs were converted to MCs and with the addition of MABr. As indicated by the absorption spectra, after the addition of MABr, MCs immediately began to convert back to PMSCs. Moreover, as the reaction temperature was decreased from 100°C to 15°C, the equilibrium increasingly favors the formation of PMSCs. This demonstration of reversibility also indicates that these MCs are smaller and more stable than the related PMSCs.

Conclusion
In summary, we have made an interesting finding on the transformation from MAPbBr 3 -VA-BTYA PMSCs to PbBr 2 -BTYA MCs. This is confirmed by monitoring the MAPbBr 3 -VA-BTYA PMSCs absorption spectra overtime. When the absorption band centered at 424 nm progressively transformed to 402 nm, there is an isosbestic point, suggesting that an equilibrium exist between the two absorption bands and species. After probing the synthesis and transformation through a series of control experiments, we conclude that the absorption band centered at 402 nm derived from only the Pb 2+ , Br -, and BTYA components is from MCs rather than PMSCs, because MA + is absent. In addition, we demonstrated that this synthesis can be extended also to PbCl 2 and PbI 2 precursors. EXAFS studies confirm that these new species are MCs rather than PMSCs and help to established structural models. The results lead us to conclude that the Pb-Br bridging bonds in the PMSC structure are broken when smaller and thermodynamically more stable MCs form. This interplay between the PMSC and MCs indicates that the MCs are a possible precursor of the PMSC. The findings from this work are important for understanding the mechanism of growth of both bulk and nanostructured perovskites including PMSCs, PNCs, or PQDs that have great potential for many emerging applications.

Conflicts of Interest
The authors declare no competing financial interest.