Sensitivity of Nitrogen K-Edge X-ray Absorption to Halide Substitution and Thermal Fluctuations in Methylammonium Lead-Halide Perovskites

The performance of hybrid perovskite materials in solar cells crucially depends on their electronic properties, and it is important to investigate contributions to the total electronic structure from specific components in the material. In a combined theoretical and experimental study of CH3NH3PbI3—methylammonium lead triiodide (MAPI)—and its bromide cousin CH3NH3PbBr3 (MAPB), we analyze nitrogen K-edge (N 1s-to-2p*) X-ray absorption (XA) spectra measured in MAPI and MAPB single crystals. This permits comparison of spectral features to the local character of unoccupied molecular orbitals on the CH3NH3+ (MA+) counterions and allows us to investigate how thermal fluctuations, hydrogen bonding, and halide-ion substitution influence the XA spectra as a measure of the local electronic structure. In agreement with the experiment, the simulated spectra for MAPI and MAPB show close similarity, except that the MAPB spectral features are blue-shifted by +0.31 eV. The shift is shown to arise from the intrinsic difference in the electronic structure of the two halide atoms rather than from structural differences between the materials. In addition, from the spectral sampling analysis of molecular dynamics simulations, clear correlations between geometric descriptors (N–C, N–H, and H···I/Br distances) and spectral features are identified and used to explain the spectral shapes.

: The XPS signal from a) N 1s, b) C 1s, c) Pb 4f , d) Br 3d, and e) I 4d core levels of MAPbBr 3 (blue) and MAPbI 3 (orange) measured using a photon energy of 535 eV. Intensity normalised and energy calibrated against Pb 4f at 138.5 eV.
S-2 0 5 10 Relative photon energy (eV) Intensity (arb. u.) Figure S2: Three independent N K-edge XAS measurements recorded from MAPI single crystals. Measurements were shifted to place the main edge intensity at 0 eV. Blue: MAPI after subtraction of PbI 2 signal. Orange: MAPI partial electron yield. Green: MAPI total electron yield. Experimental details on the crystal synthesis are given in the text.
In Figure S2 we show three different spectra from measurements of N K-edge (1s-to-2p * ) X-ray absorption measurements of single crystals. The spectra are aligned to the MAPI total yield spectrum reported in the main manuscript ( Figure 2, solid orange). This is done to show that independent measurements recorded from different crystals give similar results.
Also shown is the spectrum obtained after removal the signal recorded for the PbI 2 sample over the same region measured on a similar crystal (see below) measured using the same setup at the PM4 beamline. This shows that the feature 10 eV above N K-edge in the MAPI, as well as the MAPB shown in the main manuscript, is linked to a lead absorption feature (Pb 4d-to-6p * ).
Moreover, the feature 10 eV above the N K edge is also lacking in a partial electron yield spectrum measured at the FlexPES beamline at the synchrotron facility MAX-IV in Sweden. These measurements were performed in partial yield detection mode using 350 S-3 V retarding voltage at an NEXAFS MCP detector to reject the lead NNV auger electrons (kinetic energy of about 250 eV). The samples measured at the FlexPES beamline were prepared and mounted in the same way as described in the main manuscript.
The MAPI spectrum shown in blue was measured on crystals synthesized with a slightly different method to those presented in the main manuscript by modification of a reported procedure by Zhang et al. S1 Briefly, lead iodide (PbI 2 , from Sigma-Aldrich) and methylammonium iodide (MAI, from Dyesol) were dissolved into γ-butyrolactone (GBL) to form a solution where the concentration of PbI 2 is 0.5 M and that of MAI is 1.5 M. Chloroform (≈680 µL/mL) was added to the aforementioned solution. The solution was filtered into a clean, 4 mL glass vial using a 0.45 µm filter. Chloroform was placed into another 4 mL glass vial. Both vials were placed inside a bigger vial, which was then capped and placed into an oil bath kept at 45°C. This arrangement allowed the chloroform to slowly diffuse into the GBL solution. After two days of incubation, millimeter-scale MAPI crystals had formed. Energy (eV) Intensity (arb. u.) Figure S5: Comparison of 50 ps N K-edge spectra for MAPI and MAPB geometries, using the PBE (orange and blue) and BLYP (red and green) functionals. Pseudopotentials match the functional except for lead where the PBE functional was used for both. For PBE, peak intensities occur at 404.23 and 404.52 eV, respectively, while for BLYP they occur at 404.84 and 405.04 eV, respectively. This data shows that the relative differences between the two systems are similar with both functionals, in terms of the pre-edge, peak intensity differences, and post-edge shoulder.

404 407 410 413
Energy (eV) Intensity (arb. u.) N-C average 1.40 -1.45 Å 1.45 -1.50 Å 1.50 -1.55 Å 1.55 -1.60 Å Figure S6: The block-averaged MAPB N 1s XA spectra ordered by N-C distance in 0.05 Å blocks from 1.40 to 1.60 Å. Explanation of the trends is given in the main text, and the block-averaging procedure is given in Figure 5.  Figure S7: The block-averaged MAPB N 1s XAS spectra ordered by average N-H distance in 0.02 Å blocks from 1.00 to 1.12 Å. Explanation of the trends is given in the main text, and the block-averaging procedure is given in Figure 5.  Figure S9: The block-averaged MAPB N 1s XAS spectra ordered by difference in shortest and longest nitrogen H· · · Br distance in 0.25 Å blocks from 0 to 1 Å. Explanation of the trends is given in the main text, and the block-averaging procedure is given in Figure 5. Energy (eV) Intensity (arb. u.) Figure S14: N 1s XAS spectrum for the 50 ps sample of tetragonal MAPI (orange) and cubic MAPB (blue) compared to the tetragonal MAPB sample after an additional 10 ps simulation at 200 K (green). The tetragonal MAPB reproduces a spectrum very similar to the cubic MAPB as opposed to the tetragonal MAPI, indicating the shift is an electronic effect rather than a geometric one.

S-9
402 404 406 408 410 412 Energy (eV) Intensity (arb. u.) Figure S15: N 1s XAS spectrum for the 50 ps sample of tetragonal MAPI and cubic MAPB compared to the tetragonal MAPB sample after an additional 10 ps simulation at 300 K. The tetragonal MAPB reproduces a spectrum very similar to the cubic MAPB as opposed to the tetragonal MAPI, indicating the shift is an electronic effect rather than a geometric one. S-10