Transforming colloidal Cs4PbBr6 nanocrystals with poly(maleic anhydride-alt-1-octadecene) into stable CsPbBr3 perovskite emitters through intermediate heterostructures

The challenge of making strongly emissive CsPbBr3 perovskite nanocrystals with a robust surface passivation is solved via Cs4PbBr6 → CsPbBr3 transformation triggered by a reaction of oleylamine ligand with poly(maleic anhydride-1-alt-octadecene).

Procedure. The synthesis of Cs4PbBr6 NCs was adapted from the work of Akkerman et al. 1 with two modifications: 1) reaction was performed in air, 2) 72 mg of PbBr2 were used instead of 36 mg. The stock solution of the cesium oleate precursor was prepared by dissolving 400 mg of Cs2CO3 in 8 ml of oleic acid at 100 o C under stirring on air. Upon cooling to room temperature, the cesium oleate precursor remained clear and transparent.
For the NC synthesis, room temperature cesium oleate was injected into the reaction mixture containing PbBr2 dissolved in a mixture of ligands. For a single batch of NCs, 72±2 mg of PbBr2 were combined with 5 ml of octadecene-1, 1.5 ml of oleylamine, and 0.2 ml of oleic acid in a 20 ml glass vial. The mixture was heated to ~120-130 o C under stirring on top of the hotplate until all visible solid PbBr2 got dissolved, forming a clear colorless solution. The PbBr2 dissolution typically took less than 5 minutes. Next, the vial was transferred into a machined aluminum block pre-heated to 80-82 o C, and the temperature of the reaction mixture was let to cool down and stabilize at ~80 o C. At that point, 0.5 ml of the cesium oleate precursor were swiftly injected. The injection of cesium oleate generally resulted in the formation of a cloudy mixture, which was removed from the heating and allowed to cool in the air under stirring. Two different outcomes of the cesium oleate injection have been observed over the course of the work (about ~40 separate syntheses). 1) The reaction mixture gradually turns cloudy within the first 1-2 minutes after the injection of cesium oleate. In that case, the reaction was cooled down after a total of 2 minutes of elapsed time after the cesium oleate injection. 2) The reaction mixture remains clear and colorless after the injection of cesium oleate precursor. In that case, the reaction mixture was kept at ~80 o C for 10 minutes, after which it was allowed to cool to room temperature turning cloudy white over the course of the cooling. Regardless of the scenario, the isolated NCs synthesized by this procedure had a narrow size distribution and an average size in the 10-16 nm range.

NC Isolation.
Once the reaction mixture cooled down to the room temperature, it was split equally into four 4 ml vials using a Pasteur pipette. The vials were centrifuged for 3 minutes at 4000 rpm, yielding white to light grey precipitate at the bottom and a clear supernatant. The supernatant was discarded. To further clean the precipitate from residual supernatant, the vials were centrifuged again (1 min at 4000 rpm) to collect the residual non-decanted supernatant. For that centrifugation step, the vials were oriented in the centrifuge such that the precipitate was pointing outwards, so the remaining liquid could be collected in the lower and opposite (with respect to precipitate) part of the vial. After the centrifugation, the liquid was removed by a cotton tip, and the centrifugation step repeated to collect all of the visible residual supernatant from the vial. Such centrifugation allows isolation of Cs4PbBr6 NCs without using antisolvents. NCs were dissolved in anhydrous toluene, hexane, tetrachloroethylene, or deuterated solvents for subsequent experiments. NCs were used within a week after the synthesis, as storing them longer in air results in the loss of solubility, possibly due to coalescence. Attempts to wash Cs4PbBr6 NCs by precipitation with antisolvents resulted in either immediate (acetone) or gradual (ethyl acetate) and irreversible aggregation of NCs.
Procedure. An optically clear stock solution of PMAO in toluene with a polymer concentration of ~85 mg/ml was prepared by dissolving 170 mg of PMAO in 2 ml of anhydrous toluene under shaking at room temperature (~21 o C). Mild heating with a heat gun could be applied to facilitate the PMAO dissolution. Once all PMAO dissolved (by visual inspection), the resulting solution was filtered through a hydrophobic 0.45 micron PTFE filter. For the transformation reactions, 1/4 of the NC batch (a full amount of NC solid from one of the four vials, as described above) was dissolved in 2 ml of toluene to form a Cs4PbBr6 NC stock solution.
Typically, the reactions were performed between small aliquots of the NC stock solution (50-100 μL) and PMAO (5-80 μL). After the mixing of the desired amount of NCs and PMAO, the clear and colorless reaction mixture begins to acquire bright green color within tens of seconds or a few minutes, depending on the NC batch. The change in color indicates the formation of CsPbBr3. The PMAO amounts necessary for a complete or partial transformation have been empirically determined for a given batch of Cs4PbBr6 NCs by mixing a fixed volume of NC solution with various volumes of the PMAO stock solution, followed up by checking UV-Vis absorbance spectra (disappearance of the ~314 nm peak of Cs4PbBr6 NCs over time indicates the reaction progress). The described reaction conditions are typical, but exact amounts of reagents may vary from batch to batch, dryness of the solvents, and other parameters that were not controlled.
The full batch of Cs4PbBr6 NCs can be fully-converted to CsPbBr3 by dissolving the solid of isolated Cs4PbBr6 NCs in 1-2 ml of toluene and mixing it with 6 ml of 85 mg/ml solution of PMAO in toluene and leaving the reaction mixture on a shaker overnight at room temperature.

c) Steady-state optical spectroscopy on liquid samples
Samples of NCs, PMAO, and their mixtures were prepared as dilute toluene solutions in 4 mm by 10 mm quartz cuvettes (Hellma Analytics, 114F) and were used for the measurements in UV-Vis spectral region. Samples of NCs and 1-octadecene, oleylamine, and cesium oleate stock solution were prepared as dilute solutions in tetrachloroethylene in 10 mm by 10 mm quartz cuvettes (Firefly Sci, 41FL-UV-10) for absorbance measurements in near-infrared (NIR) spectral region.
Optical absorbance spectra in UV-Vis spectral region were collected using Cary 500 spectrophotometer, and Cary 5000 spectrophotometer was used to collect the spectra in NIR spectral region. In each case, the spectra were corrected for the absorbance of the solvent blank by subtracting the absorbance spectra of (cuvette + solvent) from the sample spectra. Photoluminescence (PL) spectra in 400-600 nm wavelength region were collected using Cary Eclipse spectrofluorimeter with excitation wavelength set at 350 nm, excitation and emission slits were typically set at 2.5 nm each.
The phase purity of CsBr and PbBr2 commercial powders was confirmed by XRD. For XRD, the samples of bulk CsBr and PbBr2 powders were prepared by mixing finely ground powders with a small amount of high vacuum silicone grease (Dow Corning®) and compacting it on top of a zero-diffraction silicon wafer.

EDS in Scanning TEM (EDS-STEM).
EDS spectra in STEM were collected using the JEOL-JEM1400 electron microscope operating at 120 kV acceleration voltage. Samples of bulk CsBr and PbBr2 powders were prepared by drop-casting toluene suspensions of the finely ground powders onto amorphous carbon-covered Cu grids. Grids prepared in that way contained submicron particles suitable for EDS-STEM analysis of "bulk" materials. The Cs4PbBr6 NC samples were prepared by drop-casting dilute solutions onto amorphous carbon-covered Cu grids. The EDS data were acquired in STEM mode, sample tilt 6 o , spot size was adjusted between S2 or S3 to optimize the signal counts. STEM images and corresponding EDS spectra were collected from several areas of the grid.

Quantification of elemental composition in EDS-STEM.
Collected EDS spectra were analyzed using Analysis Station software ver. 3.8.0.59 (JEOL Engineering Co., Ltd), JED Series AnalysisProgram ver. 3.8.0.37. Cs, Br, and Pb were chosen as the only elements for quantification (other detected elements were C, O, Cu, and Crthey originate from the TEM grid). Cs L, Br K, and Pb L lines were used for the quantification. Quantification of the EDS spectra was performed using a standardless Ratio quantification method built into the software.

e) Fourier transform infrared spectroscopy (FTIR)
A solid of as-synthesized NCs was dispersed in toluene, and a drop of the toluene solution was mixed with 100 mg of KBr powder. After that, the mixture was dried under vacuum at 40°C and compressed into a pellet for FTIR measurements. A similar procedure was employed to record the FTIR spectrum of the cesium oleate precursor. The measurements were carried out using a Bruker Vertex 70 spectrometer in transmission mode from 4000 to 500 cm -1 with 64 scans.

f) Nuclear magnetic resonance (NMR)
The NC only, NC-PMAO, PMAO, and model mixture solutions of ligands were prepared in toluene-d8 and transferred into 5 mm diameter tubes for the NMR experiments. The destructive ligand composition analysis was performed by dissolving the solid of NCs (as isolated after the synthesis) in 500-600 μL of DMSO-d6 with the addition of 5-7 μL of trifluoroacetic acid (TFA) and performing the NMR experiment in a 3 mm or a 5 mm diameter tube.
1 H NMR spectra (Figures S7, S8, S9, S20, S21, S22, S23a-c, S24-S26) were acquired on a Bruker Avance III 400 MHz spectrometer equipped with a Broad Band Inverse probe (BBI) at 300.0 K. 32-128 transients (depending on the sample concentration), and 64k of digit points were accumulated after applying a 90-pulse, with a relaxation delay of 30 s, over a spectral width of 20.55 ppm and with the offset centered at 6.18 ppm.  Figure S6) were conducted on a Bruker Avance III 600 MHz spectrometer, equipped with 5 mm QCI cryoprobe, using 24 scans, 2048 data points, and 648 increments, over a spectral width of 15.00 ppm for 1 H and 200.0 ppm for 13 C, with the offsets at 6.00 and 110.0 ppm respectively.

g) Transmission electron microscopy (TEM)
The as-synthesized Cs4PbBr6 NCs and CsPbBr3 NCs derived from them after PMAOinduced transformation were imaged using JEM 1400-Plus JEOL electron microscope operating at an acceleration voltage of 120 kV. The samples for TEM experiments were prepared by dropcasting sample solutions onto carbon-coated copper grids with a pipette and letting the solvent evaporate.
The images were collected using DigitalMicrograph software (version 3.11.1048.0) and processed using ImageJ software (version 1.51j8) 2 and a different version of DigitalMicrograph (1.71.38) equipped with PASAD plugin for Fast Fourier Transform (FFT) image analysis (PASAD-tools, v1.0). 3 The thresholding analysis was performed on the images of Cs4PbBr6 NCs for their size determination (Section S13), 4, 5 while FFT analysis was performed on the images of CsPbBr3 NCs for their size determination (Section S13). 6

h) X-ray diffraction (XRD)
XRD patterns were collected using a PANalytical Empyrean X-ray diffractometer equipped with a Cu Kα cathode (λ=1.5406 Å) operating at 45 kV and 40 mA. The NC and PMAO-NC samples were prepared for XRD by drop-casting from toluene solutions onto a zerodiffraction silicon wafer and letting the solvent evaporate. The XRD data processing and analysis were performed in HighScore ver. 4.7 software using "Search Peaks…" and "Search Match…" options.

i) PL quantum yield (QY) of NC samples in solution
The PLQYs of partially-and fully-transformed Cs4PbBr6-CsPbBr3 NC samples were measured on diluted samples using FLS920 Edinburgh Instruments spectrofluorimeter equipped with an integrating sphere. The samples for PLQY were prepared in the atmosphere of air or a nitrogen-filled glovebox by diluting concentrated PMAO-NC solutions with 1 ml of toluene (anhydrous toluene was used for the NC sample prepared in the glovebox) in 4 mm x 10 mm quartz cuvettes capped with white PTFE stoppers (Hellma-Analytics, part number 114F-10-40). Dilutions were performed shortly before the measurements.
The samples were excited at 400 nm using the xenon lamp (Xe900) with an excitation slit width set at 10 nm. The emission slit width was set at 0.20-0.25 nm. The cuvettes were oriented inside the sphere such that the excitation was through the 4 mm path length. The photon number spectra for the samples and toluene solvent reference were collected over 375-625 nm spectral range (375-600 nm range is plotted in Figures S13-S15) with a step size of 1 nm, and a dwell time of 0.2 seconds per step. Five consecutive scans of each measurement (excitation and scatter spectra) were added together to obtain the data for PLQY calculations. Corrections for the background, PMT detector sensitivity, and the lamp reference detector were applied automatically during the data collection by the software. For PLQY calculations, the photon number spectra were integrated in the range of 385-415 nm in order to calculate the total number of scattered photons (sc photons), and in the ranges 465-575 nm (partially-and fully-transformed NCs in the air) or 435-575 nm (fully-transformed NCs in the glovebox) to determine the total number of emitted photons (em photons). The values of PLQY were calculated using the following formula: , % = 100 * ( , and reported without correction for self-absorption.

j) High-resolution TEM (HRTEM)
The partially-converted Cs4PbBr6-CsPbBr3 NC heterostructures were prepared by mixing Cs4PbBr6 NCs with an amount of PMAO pre-determined to produce partially converted samples. Within minutes after mixing, already green-looking NCs-PMAO solution was drop-casted onto an ultrathin carbon/holey carbon-coated 400 mesh copper grids and placed into a JEOL JEM-2200FS microscope for HRTEM investigation. The microscope operates at 200 kV, and it is equipped with a CEOS spherical aberration corrector for the objective lens and an in-column image filter (Ω-type).

k) NC-PMAO transformation in the film
Film preparation. An 85 mg/mL solution was obtained by dissolving 170 mg of PMAO in 2 mL of toluene. After mixing by vortex until the complete dissolution of the polymer powder, the so-obtained solution was filtered through a 0.45 m PTFE syringe filter. 80 L of PMAO solution were mixed with 50 L of Cs4PbBr6 NCs dispersed in toluene, in a glass vial. After a brief shake for a few seconds, 20 L drop was cast on top of a quartz substrate (1×1 cm 2 ). The samples were dried under a gentle nitrogen flow in order to obtain solid-state films for further spectroscopic analysis.
Photoluminescence. Photoluminescence (PL) spectra were measured by exciting the samples with a continuous wave (CW) diode laser (emission wavelength of 405 nm, spot size about 2 mm and excitation power 0.7 mW), and collecting the spectra with a fiber-coupled spectrometer (Flame, Ocean Optics). Samples were mounted in an integrating sphere, which allows for measuring concomitantly the film absorbance and the photoluminescence quantum yield (QY). 7 In order to investigate the reaction evolution, these measurements were performed at regular time intervals over a period of 2 hours starting a few minutes after the mixing of the Cs4PbBr6 NCs and PMAO solution, and the film preparation.
Fluorescence confocal microscopy was carried out by using an inverted microscope (Olympus), equipped with a confocal laser scanning head. The samples were excited through a 10× objective (Numerical aperture: 0.4) by a 405 nm CW laser, and collecting the emitted photons through the same objective in a back-scattering configuration.
Time-resolved photoluminescence spectra were measured by utilizing a streak camera (Hamamatsu), coupled to a spectrometer (Princeton). The sample was optically pumped by a pulsed laser (Legend, Coherent) with an emission wavelength of 400 nm and a pulse duration of about 100 femtoseconds.
Raman. Micro-Raman measurements were performed with a Renishaw InVia spectrometer equipped with a confocal optical microscope. Samples were excited by a 785 nm laser, using a 100× objective and 1 μm excitation spot size. The 785 nm excitation laser allows avoiding PL signals from the studied samples. Typical acquisition times were of the order of 10 s, which allowed for collecting Raman spectra at time intervals of 1-5 minutes and for investigating the evolution of the reaction in the films.

S2. EDS-STEM
The EDS-STEM analysis of NC monolayer yielded ~Cs4.9Pb1Br5.5 stoichiometry, which is Cs-rich and can be explained by the presence of cesium oleate on the NC surface. Atomic compositions below are reported as an average of the results from three different areas ± standard deviation unless otherwise noted.

S3. 1 H and 1 H-13 C NMR investigation of surface passivation of Cs4PbBr6 NCs
An NMR characterization of the Cs4PbBr6 NC solutions was performed to identify and quantify organics in the samples. 1 H NMR spectra contained broadened resonances attributable to oleylamine/oleylammonium and oleate moieties, and sharp resonances belonging to the residual octadecene-1 and non-deuterated toluene (Figures S6-S8). The broadened peaks in the 1 H NMR spectra indicate that all of the organic species, with the exception of solvents, were either bound or in dynamic interaction with NC surface. [8][9][10][11] In order to quantify the relative amine/acid ratio in the NC samples, we employed a destructive method by dissolving NCs in DMSO-d6 and adding a small amount of trifluoroacetic acid. Trifluoroacetic acid facilitates the NC dissolution and additionally acts as a protonating agent for organics, which helps to separate the α-CH2 signals of oleylamine from that of oleic acid in the 1 H NMR spectrum, thus circumventing peak overlap. The ratio between the integrated peak intensities of α-CH2 resonances yielded amine/acid ratios of ~3:2 ( Figure S9). The presence of oleylamine and oleic acid on the NC surface was further confirmed by two-dimensional 1 H-13 C heteronuclear single quantum coherence (HSQC) measurements in toluene-d8 in which α-CH2 resonances from both ligands ( Figure S6) were observed, with the chemical shifts ( 1 H 2.86 ppm, 13 C 41.40 ppm) for oleylamine, and (2.54, 38.81) for oleic acid. The 2.86 ppm shift of α-CH2 of oleylamine is similar to that in an equimolar mixture of oleylamine and oleic acid from a prior study (Figure 4c in ref. 12 ). That similarity points to the NCs containing a mixture of neutral and protonated oleylamine. The protonated oleylamine is most probably in the form of oleylammonium oleate, as is the case of CsPbBr3 NCs synthesized in a similar reaction environment. 10 Overall, the structure of Cs4PbBr6 NCs is that of an inorganic crystalline core surrounded by a ligand shell consisting of cesium oleate (to account for the Cs-rich elemental composition), and oleylamine/oleylammonium oleate species.   . Annotated 1 H NMR spectra of Cs4PbBr6 NCs dispersed in toluene-d8 (top spectrum) and dissolved in DMSO-d6 (bottom spectrum, DMSO signal is marked with asterisks). The signal from α-CH2 in oleate (at ~2 ppm) is overlapped with the signal from allylic hydrogens, "TOL" and "ODE" mark signals from residual toluene and 1-octadecene. The addition of trifluoroacetic acid ( Figure S9) allows us to clearly separate α-CH2 signals from two ligands.

S7. Reactivity of bulk Cs4PbBr6 powder with PMAO
A small portion (~7 mg) of green-emitting Cs4PbBr6 powder containing an impurity of CsBr (its preparation and characterization are described in work by Ray et al. 15 ) was mixed with 1 ml of ~43mg/ml toluene solution of PMAO. The amounts were chosen to test the reactivity under conditions of a large excess of PMAO (molar ratio between Cs4PbBr6 and PMAO estimated ~1:22). The starting Cs4PbBr6 powder had the appearance of white powder with a light green tint. No change in its appearance was observed within the first hour after mixing with PMAO. Next, the sample was heated to 80 o C for 5 hours, briefly sonicated, and put on an automatic shaker overnight (~11 hrs). In the end, there was a color change from light green to light yellow (Figure S16). Trace amounts of CsPbBr3 were detected by XRD in the final sample ( Figure S17).
For the XRD, the starting sample of Cs4PbBr6 powder was mixed with a small amount of high vacuum silicone grease and compacted onto a zero diffraction Si wafer ( Figure S16, left photo). After the reaction, the Cs4PbBr6-PMAO sample was centrifuged to separate solid phase from liquid, and a precipitated slurry was drop cast onto zero diffraction silicon wafer ( Figure  S16, right photo), in the latter case the residual PMAO holds the fine powder together.

S8. FTIR and NIR absorbance spectra
Infrared spectra (Figure S18) of as-synthesized Cs4PbBr6 NCs evidenced the presence of oleate (peaks at ~1410 cm -1 and 1555 cm -1 ) and oleylamine/oleylammonium species (a broad peak at ~3440 cm -1 ), and the absence of free oleic acid (no ν(C=O) stretch at ~1711 cm -1 ). Optical absorption spectra of concentrated NC solutions in the near-infrared ( Figure S19) were consistent with these findings. No evidence of water or Pb-OH moieties [17][18][19] in NC samples was found by FTIR or optical absorption, despite performing the synthesis and handling samples in air. Figure S18. FTIR spectra of cesium oleate precursor (top trace) and three replica batches of Cs4PbBr6 NCs. The spectra were normalized to the amplitude of the CH3 bend (~1464 cm -1 ). The carbonyl stretch (C=O, ~1711 cm -1 ) of free oleic acid is clearly observable in the cesium oleate precursor due to the excess of oleic acid but is absent in NC samples. A very broad asymmetric peak at ~3400-3500 cm -1 is interpreted as an oleylamine/oleylammonium mixture. The assignment of the peaks is based on prior works. 20 Figure S19. NIR (800-3300 nm) absorbance spectra of tetrachloroethylene solutions of Cs4PbBr6 NCs, cesium oleate precursor, oleylamine (oleylamine, tech. grade 70%, Aldrich), and 1-octadecene. Vertical dashed lines indicate the correspondence of the absorbance peaks in NCs with the peaks from the samples of organics.

S9. 1 H NMR investigation of Cs4PbBr6-PMAO reactivity
The reaction between oleylamine and PMAO leads to the broadening of the vinyl hydrogens resonance of oleylamine in the 1 H NMR spectrum (Figures S20-S22, sharp multiplet at 5.44 ppm of free oleylamine broadens and shifts to ~5.33 ppm in PMAO-oleylamine adduct), as discussed in the main text. The addition of oleylamine to cyclic anhydride yields an amic acid derivative, as was confirmed in a control reaction between oleylamine and succinic anhydride in DMSO-d6 at room temperature (see figure Figure S23 for 1 H and 1 H-13 C HSQC NMR spectra). The DMSO-d6 was chosen as a solvent due to poor solubility of succinic anhydride in toluene-d8. The succinic anhydride was used as a small molecule proxy of the reactive functional group of PMAO to avoid line broadening in NMR caused by the macromolecules. Therefore, it is assumed that PMAO-oleylamine adduct is a polysuccinamic acid in the rest of the discussion.
Note: commercial samples of PMAO contain a likely impurity of residual 1-octadecene monomer ( Figure S24). Purification of the starting PMAO reagent was not attempted, but the final CsPbBr3/PMAO NCs could be purified (washed) from 1-octadecene by several precipitation/redispersion cycles with ethyl acetate/toluene antisolvent/solvent pair ( Figure S25).
Next, the reaction between PMAO and Cs4PbBr6 NCs was investigated by inspecting the ~5.3-5.9 ppm region of vinylic hydrogens in 1 H NMR spectra of two mixtures: transformed CsPbBr3 NCs in a crude reaction mixture and after one washing cycle. The spectra were compared with 1 H NMR spectra of the starting Cs4PbBr6 NCs (Figure S20, see Figures S21, S22 for full spectra). The region of vinylic hydrogens was chosen because it allows us to distinguish between bound and free states of the oleylamine and oleate ligands. The starting Cs4PbBr6 NCs show a single broad peak at ~5.49 ppm (Figure S20, "OLAM+OLAC bound") assigned to the overlapping resonances of vinyl hydrogens of oleylamine (OLAM) and oleate (OLAC) from the ligand shell of the NCs. After the addition of PMAO and a complete Cs4PbBr6 → CsPbBr3 transformation (confirmed by the absence of 314 nm peak in UV-Vis absorption spectra), the 1 H NMR spectrum of the crude reaction mixture revealed that a single broad peak of OLAM/OLAC ligands had been split in two: an even broader peak at ~5.55 ppm and a sharp multiplet at 5.47 ppm [ Figure S20, "CsPbBr3/PMAO (transformed, not washed)"]. The broad peak at ~5.55 ppm was assigned to the PMAO-oleylamine adduct based on control discussed above, while a multiplet at 5.47 ppm was assigned to the unbound ligands displaced from the NC surface. The ratio of the areas under the broad (~5.55 pm) and the sharp (5.47 ppm) peaks was estimated to be ~1.8 (via the "Line Fitting" tool in MestReNova, software ver. 12.0.0-20080). This ratio is very similar to ~3:2 molar ratio of (oleylamine+oleylammonium):oleate in the starting Cs4PbBr6 NCs. That similarity led us to the interpretation that a sharp multiplet at 5.47 ppm most likely belongs to oleate species displaced from the NC surface by a polysuccinamic acid formed as a result of oleylamine addition to PMAO.
The removal of oleate from the NC surface was further supported by two experiments: i) a comparison between the 1 H NMR spectra of the CsPbBr3/PMAO NCs before and after washing with ethyl acetate. The washing purified the sample from unbound species (excess of polymer and displaced ligands) as evidence by a single broad peak at ~5.57 ppm corresponding to the PMAO-oleylamine adduct [ Figure S20, "CsPbBr3/PMAO (transformed, washed)"]; ii) dissolution of washed CsPbBr3/PMAO in DMSO-d6/TFA mixture. 1 H NMR spectrum of the dissolved CsPbBr3/PMAO NCs only contained signatures of oleylammonium, and oleic acid was absent ( Figure S26).      22 We hypothesize that this impurity is either a residual 1-octadecene monomer left after PMAO manufacturing or an unidentified side product/unsaturated hydrocarbon. Luckily, 1-octadecene is inert to Cs4PbBr6 and CsPbBr3 at room temperature and does not appear to play a role in the chemical transformation.
The sample of CsPbBr3/PMAO NCs washed once still contains a residual 1-octadecene (e.g. Figures S20, bottom spectrum). The residual 1-octadecene can be removed by repeated precipitation/redispersion of CsPbBr3/PMAO NCs with ethyl acetate (for precipitation) and toluene (for redispersion). For example, four consecutive purifications reduced the amount of 1octadecene in the sample to the amounts below the detection limit of the 1 H NMR experiment ( Figure S25). The CsPbBr3/PMAO NC sample washed four times is sufficiently stable to perform an NMR experiment. Figure S25. Comparison of the 1 H NMR spectra of CsPbBr3/PMAO NCs after a single cycle of precipitation/redispersion with ethyl acetate and toluene (top spectrum) and after four cycles (bottom spectrum). The consecutive purification reduces the amounts of residual 1-octadecene to below the detection limit. Figure S26. Comparison of 1 H NMR spectrum of the starting Cs4PbBr6 NCs dissolved in DMSO-d6/TFA (green curve, data reproduced from Figure S9) with a spectrum of washed (x4 times) CsPbBr3/PMAO NCs dissolved in DMSO-d6/TFA (maroon curve). The spectrum of the dissolved Cs4PbBr6 NCs shows both oleylammonium and oleic acid species (α-CH2 multiplets at ~2.72 ppm, and ~2.14 ppm, respectively). The spectrum of the dissolved CsPbBr3/PMAO NCs shows oleylammonium species due to the acid hydrolysis of PMAO-oleylamine adduct by TFA. However, the characteristic signal from oleic acid, α-CH2 peak at ~2.14 ppm, is absent in the spectrum of the dissolved CsPbBr3/PMAO sample. That further confirms that oleic acid was removed from the final washed NCs.

S10. Control reaction between amine-free Cs4PbBr6 NCs and PMAO
Oleylamine-free Cs4PbBr6 NCs were synthesized using the tri-n-octylphosphine oxide (TOPO) instead of oleylamine following the previously described procedure. 23 The resulting NCs are larger in size, so they scatter in the optical spectra, consistent with the original JACS report. 23 PMAO was dissolved in anhydrous toluene to form an 85 mg/ml concentrated solution.      25 in conjunction with a molar extinction coefficient of CsPbBr3 from ref. 26 and scaled with dilution as appropriate (for example, it was impossible to directly record absorption spectrum of ~26 mg/ml solution through the thinnest available cell with a known pathlength [200-micron thick cuvette (d)] due to the very strong light attenuation by the NC solution (optical density at 500 nm, i.e., below the band edge, was already ~1.6), so the concentrated solution was diluted and the initial concentration back-calculated using the known dilution factor).    Averaging the periodicity across 5 images yielded 10.8 nm ± 0.2 nm. That spatial period is interpreted as a sum of NC width and interparticle spacing. 6 The spacing between NCs of ~2.8 nm ± 0.4 nm was estimated from real space images using ImageJ. The difference between the abovementioned values gives an estimated value for the NC edge length of ~8 nm ± 0.4 nm.  Averaging the periodicity across 9 images yielded 14.2 nm ± 1.8 nm. That spatial period is interpreted as a sum of NC width and interparticle spacing. 6 The spacing between NCs of ~2.2 nm ± 0.5 nm was estimated from real space images using ImageJ. The difference between the abovementioned values gives an estimated value for the NC edge length of ~12 nm ± 1.9 nm.

c) Calculations of conversion stoichiometry
The NC shapes are assumed to be a sphere for Cs4PbBr6 NCs and a cube for CsPbBr3 NCs. Another assumption is that the transformation happens one for one, i.e., one NC Cs4PbBr6 transforms into one NC of CsPbBr3.
A sphere of bulk Cs4PbBr6 with a diameter of 10.1 nm has a volume of:  Table S1 summarizes calculated cube edges for CsPbBr3 NC, assuming 1 equiv. Cs4PbBr6 converts to n equiv. of CsPbBr3, and using 3 = 4.86 g/cm 3 (ref. 14 ) and 3 = 579.82 g/mol for a reverse calculation from the number of moles to the NC cube dimensions.

S14. PL maps of NC samples at room and cryogenic temperatures
Sample preparation: The samples of Cs4PbBr6 NCs and a freshly made partially converted Cs4PbBr6-CsPbBr3 NCs were drop-casted (10-20 μL drop volume) from toluene solutions on top of a 1"-diameter sapphire disk and dried under ambient atmosphere. After that, the sapphire disk was promptly mounted into the cryostat and placed under vacuum. The time between the preparation (i.e. the moment of mixing of Cs4PbBr6 NCs and PMAO) of a partiallyconverted sample and it's reaching the lowest temperature (~35K) was around 3 hours.
Cryostat and optical setup: The excitation-emission correlation maps (PL maps) of a partially-converted sample of Cs4PbBr6-CsPbBr3 NCs at 292K and 35K (Figure 4a,b in the main text) and the as-synthesized Cs4PbBr6 NCs (Figure S43 below) were collected by using a closedcycle helium cryostat (Advanced Research Systems, Inc., model DE204SI) coupled to FLS920 Edinburgh Instruments spectrofluorimeter via a set of optical fibers. The samples were excited using the output of xenon lamp (Xe900) coupled to a monochromator.
The PL maps of the Cs4PbBr6-CsPbBr3 NC sample at 292K and 35K were corrected by subtracting the PL maps of a "blank" sample (drop-casted PMAO film) collected at the respective temperatures in a separate experiment under identical instrumental settings. The PL map of the Cs4PbBr6 NC sample is shown without correction for a blank. Figure S43. PL map of Cs4PbBr6 NCs at T ~27K. The weaker features at around the main emission peak (λexc ~313 nm, λem ~373 nm) are attributed to the to various electronic transitions in Pb 2+ ion. 27 Figure S44. The emission spectrum of Cs4PbBr6 NCs at 27K collected over 324-1600 nm range (λexc ~314 nm). The weak hump at ~618 nm is an artifact from the background. The different colors of the spectra in the visible (VIS PL, black curve) and near-infrared (NIR PL, red curve) regions are due to the different detectors used for each spectral range. S15. Raman, μ-PL, and TRPL of CsPbBr3/PMAO film Figure S45. (a) Raman spectra of (a) Cs4PbBr6 NCs, (b) PMAO polymer powder and (c) dropcasted film after the complete transformation of Cs4PbBr6 into CsPbBr3. (d) Raman spectra of a drop-casted film measured 5 minutes (continuous red line) and 15 minutes (continuous blue line) after the mixing of NCs and PMAO. The vertical arrows highlight the modes at 70 and 74 cm -1 , attributed to the Cs4PbBr6 and to the CsPbBr3 NCs, respectively. (e) Temporal evolution of the ratio between the intensities of the peak at 70 cm -1 (I70) and the peak at 74 cm -1 (I74). The continuous line is a fit to the data by first-order kinetics. Inset: temporal evolution of the intensities of the peaks at 625 cm -1 (I625) and at 825 cm -1 (I825), attributed to the PMAO polymer. The continuous line is a fit to the data by first-order kinetics. Figure S45a shows the characteristic Raman spectra of the Cs4PbBr6 NCs, the PMAO polymer powder (Figure S45b), and the composite film after full conversion to CsPbBr3 NCs ( Figure S45c). One can observe peaks at 625 cm -1 (ring bend) and 865 cm -1 (C-C stretch), characteristic of the maleic anhydride of the PMAO, 30 and peaks at 70, 82, 123, and 1642 cm -1 of the Cs4PbBr6. After the reaction, new peaks at 74, 87, 103, 261, and 323 cm -1 emerge, which are attributed to the CsPbBr3 NCs. 31 The analysis of the evolution of the Raman spectra during the reaction in the solid-state films evidences the conversion of the Cs4PbBr6 NCs to CsPbBr3 NCs, supported by the decrease of the peak at 70 cm -1 of the Cs4PbBr6 NCs with respect to the peak at 74 cm -1 of the CsPbBr3 NCs ( Figure S45d). This is better highlighted in Figure S45e, which displays the ratio of the intensity of those peaks as a function of the reaction time. In fact, the ratio I70/I74 (where I70 and I74 are the intensities of the peaks at 70 and 74 cm -1 , respectively) decreases with increasing reaction time, following a trend similar to the one observed in photoluminescence measurements. Finally, the inset of Figure S45e shows the temporal evolution of the peak at 625 cm -1 , attributed to the maleic anhydride ring bend, normalized to the peak of the C-C stretch at 865 cm -1 , here used a reference. The decrease of the ratio I625/I865 (where I625 and I865 are the intensities of the peaks at 625 and 865 cm -1 , respectively) is indicative of a weakening of the modes associated to the anhydride ring, suggesting a reaction path, which might involve the opening of the anhydride ring. The evolution of the conversion reaction of the Cs4PbBr6 NCs to the CsPbBr3 was also investigated by in-situ micro-photoluminescence (µ-PL), using a fluorescence confocal microscope. The results are summarized in Figure S46. Figure S46a shows a sequence of PL t=32 min t=28 min t=24 min t=20 min t=16 min t=12 min t=8 min t=4 min spatial maps (size 250×250 µm 2 ), clearly evidencing an increase of the sample emission with increasing reaction time. A few bright spots are also observed, which can be attributed to the formation of aggregates. The µ-PL intensity increases following an exponential trend ( Figure  S46b), with a characteristic time of 13 minutes, comparable to the one found for CW and timeresolved macroscopic PL measurements. Interestingly, such a trend is spatially uniform, as shown in Figure S46c, where the µ-PL intensity collected in three different regions (labeled as A, B, and C) of the sample are compared. Figure S47. Example of a temporal decay profile of the PL intensity of a drop-casted film after the transformation of Cs4PbBr6 into CsPbBr3. Two emitting components are identified, with lifetimes τ1=60 ps and τ2=950 ps, as obtained by fitting the data to the sum of two exponential functions, convoluted with the instrumental response function (continuous line).