Edge stabilization in reduced-dimensional perovskites

Reduced-dimensional perovskites are attractive light-emitting materials due to their efficient luminescence, color purity, tunable bandgap, and structural diversity. A major limitation in perovskite light-emitting diodes is their limited operational stability. Here we demonstrate that rapid photodegradation arises from edge-initiated photooxidation, wherein oxidative attack is powered by photogenerated and electrically-injected carriers that diffuse to the nanoplatelet edges and produce superoxide. We report an edge-stabilization strategy wherein phosphine oxides passivate unsaturated lead sites during perovskite crystallization. With this approach, we synthesize reduced-dimensional perovskites that exhibit 97 ± 3% photoluminescence quantum yields and stabilities that exceed 300 h upon continuous illumination in an air ambient. We achieve green-emitting devices with a peak external quantum efficiency (EQE) of 14% at 1000 cd m−2; their maximum luminance is 4.5 × 104 cd m−2 (corresponding to an EQE of 5%); and, at 4000 cd m−2, they achieve an operational half-lifetime of 3.5 h.


Fabrication of perovskite crystals
The perovskite crystals were prepared by combining PbBr 2 , CsBr and PEA in appropriate molar ratios in a HBr solvent mixture following a previous reported method 1 .

Photoluminescence decay mapping
As synthesized perovskite crystals were mechanically exfoliated to generate micron-sized crystals, which were then dispersed into toluene (spectroscopy grade, Sigma-Aldrich). The

AFM measurements
Sample morphology was characterized by atomic force microscopy (AFM). Topographical and phase images were obtained with an Asylum Research Cypher AFM operated in AC mode in air.
Imaging was done using ASYELEC-02 silicon probes with titanium-iridium coatings from Asylum Research. The probes had a typical spring constant of 42 N/m. Smoothness and morphology of perovskite films were measured by atomic force microscopy (AFM). Edgestabilized perovskite films shows a smaller grain size (<50nm) and lower root mean square (r.m.s) roughness (~2nm) than the perovskite film (r.m.s ~10nm) (Supplementary Fig. 29).

XRD measurements
XRD measurements of the oriented films were conducted using a Rigaku MiniFlex 600 diffractometer (Bragg-Brentano geometry) equipped with a NaI scintillation counter detector and a monochromatized Cu Kα radiation source (λ = 1.5406 Å) operating at a voltage of 40kV and current of 15mA. The diffraction patterns of PEA 2 Cs 2.4 MA 0.6 Pb 4 Br 13 perovskite film revealed distinctive to compare with 3D CsPbBr 3 and 2D PEA 2 PbBr 4 perovskites, due to the composition of different n grains with a preferential orientation along the substrate. We observed two additional XRD peaks in these films, at 2θ=10.11° and 20.22°, which correspond to the diffraction from (TPPO) 2 PbBr 2 complexes 2 (Supplementary Fig. 17 for the XRD of TPPOprecursor reference).

XPS measurements
XPS measurements were carried out with the Thermo Scientific K-Alpha XPS system. An Al Kα source with a 400µm spot size was used for measurements to detect photo-electrons at specific energy ranges to determine the presence of specific elements.

FTIR measurements
The Thermo Scientific Nicolet iS50 ATR-FTIR was used to obtain the FTIR spectra. The prepared films were placed on top of the ATR crystal. Spectra were obtained using 16 scans with a resolution of 4cm -1 , and the collection range was between 550 cm -1 to 4000 cm -1 .

Raman spectroscopy measurements
Raman spectra of TPPO, PbBr 2 , TPPO-PbBr 2 (named as TPPO-precursor), perovskite, and TPPO-perovskite deposited on glass slides were collected using a 561-nm continuous-wave diode laser (Cobolt). Scattered light was collected by an aspheric collection lens in a back scattering geometry. The Rayleigh line was attenuated by a 561-nm notch filter (Semrock, Stopline). A single monochromator with a 1200 gr/mm diffraction grating (Princeton Instruments, TriVista used as single) dispersed the back-scattered light onto a CCD camera (Princeton Instruments, Pixis 400). The entrance slit-width was kept at 100-microns. A total of 10 spectra were averaged for each sample where each spectrum was collected using a 10 second exposure time for 10 accumulations (i.e. each spectrum represents a 1000-s average). The power was maintained at 80-mW. Glass background spectra were obtained by advancing the focal plane from the slides front face into its center where only glass signal was observable. A small baseline was then drawn to correct for the residual from the glass subtraction. Spectra were collected within three Raman windows to capture low, mid, and high-frequency regions. Spectra were smoothed using a 4-point binomial smoothing algorithm. The Raman shift axis for each of these regions was calibrated to a cyclohexane standard, establishing peak frequency accuracy to +/-5cm -1 (based on variation within and between windows of known peak positions).
Raman spectra of TPPO and TPPO mixed with MABr were collected using a 785 nm continuous-wave diode laser (Renishaw) and an upright inVia Raman Microscope setup (Renishaw). A 50x objective was used to focus and collect the light to and from the sample, respectively. The Rayleigh line was attenuated with an edge filter set. A 1200 gr/mm diffraction grating was used to disperse the light onto a CCD. The power was kept at 200-mW. Spectra were taken, using a 10 second exposure time, for several Raman windows, and a baseline was subtracted from the spectra.
The TPPO Raman spectrum changes significantly upon addition to the PbBr 2 precursor or to the perovskite (Supplementary Fig. 14 and 15). The most notable difference between the TPPO and TPPO-precursor spectra are the intensities of the 997, 1003, 1030, 1156, and 1165 cm -1 peaks visible in the mid-frequency spectrum. The relative intensities of the features at 997 and 1003 cm -1 change, while the 1030, 1156 and 1165 cm -1 peaks are reduced in intensity.
Supplementary Fig. 14b contains the low-frequency window spectra of TPPO and TPPOprecursor. To further confirm that the intensity change indicates the complex formed via TPPO-Pb, we also measured Raman spectrum with a mixture of TPPO and MABr ( Supplementary   Fig. 15) using a 785 nm Raman system. Since MABr contains no Pb, we expect that they should not form a complex. The Raman spectra show no intensity change compared to TPPO only, suggesting no complexation.

UPS measurements
UPS spectra of the perovskite films were measured on Au coated substrate. Photoelectron spectroscopy was performed in a PHI5500 Multi-Technique system using non-monochromatized He-I radiation (UPS) (hv=21.22ev). All work function and valence-band measurement were performed at a takeoff angle of 88º, with chamber pressure near 10 -9 Torr.

Superoxide probe test
To characterize the possibility of superoxide generation from the reaction of photogenerated electrons with oxygen, we measured fluorescence emission of a hydroethidine (HE) dye solution where the perovskite film located in probe solution, exposed to UV light and blow dry air during the illumination. The rate of increase of dye emission at 610 nm is faster in perovskite, while in