Phase Segregation in Potassium-Doped Lead Halide Perovskites from 39 K solid-state NMR at 21.1 T

Transcription

1 Phase Segregation in Potassium-Doped Lead Halide Perovskites from 39 K solid-state NMR at 21.1 T Dominik J. Kubicki, a,b Daniel Prochowicz, b,c Albert Hofstetter, a Shaik M. Zakeeruddin, b Michael Grätzel,* b Lyndon Emsley* a a Laboratory of Magnetic Resonance, Institute of Chemical Sciences and Engineering, Ecole Polytechnique Fédérale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland b Laboratory of Photonics and Interfaces, Institute of Chemical Sciences and Engineering, Ecole Polytechnique Fédérale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland c Institute of Physical Chemistry, Polish Academy of Sciences, Kasprzaka 44/52, Warsaw, Poland Table of contents List of figures and tables 1 Perovskite synthesis 4 Mixed-cation lead iodide perovskites 4 Mixed-cation and mixed-halide lead perovskites 4 Potassium and cesium lead iodides and bromides 5 XRD patterns 6 Details of NMR measurements 16 Details of DFT calculations 18 Cluster generation 18 EFG tensor calculation 19 References 22 List of figures and tables Figure S1. XRD patterns of the materials reported in the main text. Simulated patterns: (a) α-fapbi 3 (black 3D perovskite) (b) γ-fapbi 3, (yellow, hexagonal non-perovskite phase). Experimental patterns of mechanochemical perovskite preparations: (c) K 0.10FA 0.90PbI 3, (d) K 0.05Cs 0.10FA 0.90PbI 3, (e) K 0.05MA 0.10FA 0.85PbI 3, (f) KMAFA(I,Br) 3.,, and indicate the main perovskite phase, PbI 2 and KI peaks, respectively. For K 0.10FA 0.90PbI 3 there is no measurable shift of the main perovskite peaks with respect to the undoped α-fapbi 3. 6 Figure S2. Experimental XRD pattern of the (a) mechanochemical annealed KI:PbI 2 (1:1 mol/mol) material ( KPbI 3 ), (b) KI. ICDD database reference patterns (the numbers are ICDD database reference codes): (c) - S1 -

2 , KPbI 3 (ICDD quality mark: low precision), (d) , KPbI 3 (ICDD quality mark: prototype), , K 2PbI 4 (e) (ICDD quality mark: low precision). 7 Figure S3. Experimental XRD patterns of (a) mechanochemical MAPbI 3, 2 (b) mechanochemical, annealed KI:PbI 2 (1:1 mol/mol) material ( KPbI 3 ), (c) KI, (d) mechanochemical ( M ) K 0.10MA 0.90PbI 3, (e) K 0.10MA 0.90PbI 3 prepared by the solution ( S ) route, (f) PbI 2. and indicate the main perovskite phase and PbI 2 peaks, respectively. For K 0.10MA 0.90PbI 3 there is no measurable shift of the main perovskite peaks with respect to the undoped MAPbI 3. 7 Figure S4. (a) Simulated XRD pattern of α-fapbi 3 (black 3D perovskite). Experimental XRD patterns of (b) PbI 2, (c) KI, (d) mechanochemical ( M ) K 0.05Cs 0.10FA 0.90PbI 3, (e) K 0.05Cs 0.10FA 0.90PbI 3 prepared by the solution ( S ) route, (f) mechanochemical K 0.50Cs 0.50PbI 3.,, and κ indicate the main perovskite phase, PbI 2, KI and a K-rich non-perovskite phase peaks, respectively. 8 Figure S5. ICDD database reference patterns (the numbers are ICDD database reference codes): (a) CsPbBr 3, (b) CsPbI 3 (black, high-temperature phase), (c) CsPbI 3 (yellow, room-temperature phase). Experimental XRD patterns of mechanochemical perovskite preparations (d) CsPbBrI 2Br, (e) K 0.075Cs 0.975PbBrI 2Br, (f) KPbI 3, (g) K 0.50Cs 0.50PbI 3. and κ indicate the main perovskite phase and a K-rich non-perovskite phase peaks, respectively. In the case of K 0.075Cs 0.925PbI 2Br there is no appreciable shift with respect to CsPbI 2Br either (beside a shift of θ for the 29.2 peak after K + doping which can be caused by (a) a change in I/Br ratio, (b) specimen displacement errors, as discussed in the manuscript). 9 Figure S6. Experimental XRD patterns of mechanochemical (a) CsPbI 3 (yellow, room-temperature phase), (b) KPbI 3, (c) K 0.10Cs 0.90PbI 3, (d) K 0.50Cs 0.50PbI 3, (e) K 0.90Cs 0.10PbI 3, (f) KI. 10 Figure S7. (a) Experimental XRD pattern of mechanochemical KPbI 3 (red) and its best fit (blue). The fit residual is given in (b). The weighted-profile R-factor R wp = Figure S8. CT-selective 39 K MAS NMR spectrum of TiO 2-KI, at 11.7 T at 20 khz MAS and 300 K (recycle delay: 10 s, acquisition time: 117 h). TiO 2 was activated at 300 C and mechanochemically ground with anhydrous KI in a 10:1 molar ratio. Asterisks indicate spinning sidebands. The symbol indicates a quadrature detection artefact. 13 Figure S9. CT-selective spectra for compositions in fig. 2c, e and g in the main text. Acquisition parameters: (a) number of scans: 4096, recycle delay: 12 s, acquisition time: ~13.7 h (b) number of scans: 10240, recycle delay: 1 s, acquisition time: ~3 h, (c) number of scans: 23604, recycle delay: 3 s, acquisition time: ~20 h. 14 Figure S Pb MAS NMR spectra at 11.7 T, 298 K and 20 khz MAS of (a) MAPbI 3, number of scans: 27280, recycle delay: 0.1 s, acquisition time: 45 min., processed with 1 khz Lorentzian apodization, (b) δ-cspbi 3, number of scans: 30364, recycle delay: 0.1 s, acquisition time: 51 min., processed with 10 khz Lorentzian apodization, (c) KPbI 3, number of scans: , recycle delay: 0.02 s, acquisition time: 34 min, processed with 10 khz Lorentzian apodization. 207 Pb chemical shifts were referenced to Pb(CH 3) 4 using solid Pb(NO 3) 2 as a secondary reference (-2961 ± 1) ppm Figure S11. 1 H MAS NMR spectra at 21.1 T, 298 K and 20 khz MAS of (a) mechanochemical K 0.05Cs 0.10FA 0.85PbI 3 (proton-containing impurities originating from the supplied PbI 2 and amounting to ~1% of protons in the sample are indicated). K 0.05Cs 0.10FA 0.85PbI 3 prepared by the solution route: (b) after initial annealing in air, (c) after 12 h of vacuum drying. 15 Figure S12. Quantitative 39 K MAS NMR spectrum at 21.1 T, 298 K and 20 khz MAS of the mechanochemical KPbI 3 preparation. The spectrum was acquired using a recycle delay of 45 s, hence it is quantitative with respect to KI (T 1=9 s) and confirms there is no unreacted KI (δ=59.3 ppm) in the KPbI 3 phase. Number of scans: 32. Acquisition time: 24 minutes. 15 Figure S13. Scatterplot showing 39 K experimental shifts against DFT calculated chemical shielding. The solid lines show the different linear models used to convert chemical shielding to shifts (green: fixed slope (b=1), red: variable slope (b=1.248)). 20 Table S1. Numerical results of the fit in fig. S7a S2 -

3 Table S2. Nuclear 39 K T 1 values measured using a saturation-recovery sequence and fitted using a monoexponential function (unless otherwise stated). The uncertainties of fits given are one standard deviation Table S3. Acquisition and processing parameters used for the spectra in fig. 2, 3 and 5 of the main text Table S4. Source and modifications of the cluster structures used in the DFT calculations Table S5. DFT calculated and experimental chemical shieldings, chemical shifts and EFG tensor parameters S3 -

4 Perovskite synthesis Mixed-cation lead iodide perovskites K0.10MA0.90PbI g of KI (0.10 mmol), g of MA ΗΙ (0.90 mmol) and g of PbI2 (1.00 mmol) were mixed to prepare the K0.10MA0.90PbI3 powder. K0.10FA0.90PbI g of KI (0.10 mmol), g of FA ΗΙ (0.90 mmol) and g of PbI2 (1.00 mmol) were mixed to prepare the K0.10FA0.90PbI3 powder. K0.05MA0.1FA0.85PbI g of KI (0.05 mmol), g of MA ΗΙ (0.10 mmol), g of FAI (0.85 mmol) and g of PbI2 (1.00 mmol) were mixed to prepare the K0.05MA0.1FA0.85PbI3 powder. K0.05Cs0.1FA0.85PbI g of KI (0.05 mmol), g of CsI (0.10 mmol), g of FA ΗΙ (0.85 mmol) and g (1.00 mmol) of PbI2 were mixed to prepare the K0.05Cs0.1FA0.85PbI3 powder. K0.1Cs0.9PbI g of KI (0.10 mmol), g of CsI (0.90 mmol) and g of PbI2 (1.00 mmol) were mixed to prepare the K0.1Cs0.9PbI3 alloy. K0.5Cs0.5PbI g of KI (0.50 mmol), g of CsI (0.50 mmol) and g of PbI2 (1.00 mmol) were mixed to prepare the K0.5Cs0.5PbI3 alloy. K0.9Cs0.1PbI g of KI (0.90 mmol), g of CsI (0.10 mmol) and g of PbI2 (1.00 mmol) were mixed to prepare the K0.9Cs0.1PbI3 alloy. Mixed-cation and mixed-halide lead perovskites CsPbI2Br g of CsBr (0.50 mmol) and 0.23 g of PbI2 (0.50 mmol) were mixed to prepare the CsPbI2Br powder. K0.075Cs0.925PbI2Br g of KI (0.075 mmol), g of CsI (0.925 mmol), g of PbBr2 (0.50 mmol) and 0.23 g of PbI2 (0.50 mmol) were mixed to prepare the K0.075Cs0.925PbI2Br powder - S4 -

5 KMAFA(Br,I) The triple cation perovskite was fabricated according to the previously published recipe using KI instead of CsI g of FA ΗΙ (1 mmol), g of PbI2 (1.1 mmol), g of MA ΗBr (0.2 mmol), g of PbBr2 (0.22 mmol) and g of KI (0.055 mmol) were milled to prepare the KMAFA black powder. Potassium and cesium lead iodides and bromides KPbI3 perovskite g of KI (0.50 mmol) and 0.23 g (0.50 mmol) of PbI2 were mixed to prepare the KPbI3 powder. KPb2Br g of KBr (0.50 mmol) and g of PbBr2 (1.00 mmol) were mixed to prepare the KPb2Br5 powder. KPbI2Br g of KI (1.00 mmol), 0.23 g of PbI2 (0.50 mmol) and g of PbBr2 (0.50 mmol) were mixed to prepare the KPbI2Br powder. CsPbI g of CsI (1.00 mmol) and g of PbI2 (1.00 mmol) were mixed to prepare CsPbI3. Perovskites prepared by the solution route Polycrystalline powder of the K10MA90PbI3 compositions was synthesized as follows g of KI (0.10 mmol), g of MA ΗΙ (0.90 mmol) and g of PbI2 (1.00 mmol) were dissolved in 1 ml of DMF/DMSO mixture (4:1, v:v), and then drop-cast on a petri dish and heated at 120 C in air. The resulting powder was scratched from the glass and dried under vacuum at 120 C for 12 h. For the synthesis of K0.05Cs0.1FA0.85PbI3 powder, the same procedure was followed using g of KI (0.05 mmol), g of CsI (0.10 mmol), g of FAI (0.85 mmol) and g (1.00 mmol) of PbI2. - S5 -

6 XRD patterns Diffractograms were recorded on an X Pert MPD PRO (Panalytical) diffractometer equipped with a ceramic tube (Cu anode, λ = Å), a secondary graphite (002) monochromator and an RTMS X Celerator (Panalytical) in an angle range of 2θ = 5 to 40, by step scanning with a step of 0.02 degree. Figure S1. XRD patterns of the materials reported in the main text. Simulated patterns: (a) α- FAPbI3 (black 3D perovskite) (b) γ-fapbi3, (yellow, hexagonal non-perovskite phase). Experimental patterns of mechanochemical perovskite preparations: (c) K0.10FA0.90PbI3, (d) K0.05Cs0.10FA0.90PbI3, (e) K0.05MA0.10FA0.85PbI3, (f) KMAFA(I,Br)3.,, and indicate the main perovskite phase, PbI2 and KI peaks, respectively. For K0.10FA0.90PbI3 there is no measurable shift of the main perovskite peaks with respect to the undoped α-fapbi3. - S6 -

7 Figure S2. Experimental XRD pattern of the (a) mechanochemical annealed KI:PbI2 (1:1 mol/mol) material ( KPbI3 ), (b) KI. ICDD database reference patterns (the numbers are ICDD database reference codes): (c) , KPbI3 (ICDD quality mark: low precision), (d) , KPbI3 (ICDD quality mark: prototype), , K2PbI4 (e) (ICDD quality mark: low precision). Figure S3. Experimental XRD patterns of (a) mechanochemical MAPbI3, 2 (b) mechanochemical, annealed KI:PbI2 (1:1 mol/mol) material ( KPbI3 ), (c) KI, (d) - S7 -

8 mechanochemical ( M ) K0.10MA0.90PbI3, (e) K0.10MA0.90PbI3 prepared by the solution ( S ) route, (f) PbI2. and indicate the main perovskite phase and PbI2 peaks, respectively. For K 0.10MA 0.90PbI 3 there is no measurable shift of the main perovskite peaks with respect to the undoped MAPbI 3. Figure S4. (a) Simulated XRD pattern of α-fapbi3 (black 3D perovskite). Experimental XRD patterns of (b) PbI2, (c) KI, (d) mechanochemical ( M ) K0.05Cs0.10FA0.90PbI3, (e) K0.05Cs0.10FA0.90PbI3 prepared by the solution ( S ) route, (f) mechanochemical K0.50Cs0.50PbI3.,, and κ indicate the main perovskite phase, PbI2, KI and a K-rich non-perovskite phase peaks, respectively. - S8 -

9 Figure S5. ICDD database reference patterns (the numbers are ICDD database reference codes): (a) CsPbBr3, (b) CsPbI3 (black, high-temperature phase), (c) CsPbI3 (yellow, room-temperature phase). Experimental XRD patterns of mechanochemical perovskite preparations (d) CsPbBrI2Br, (e) K0.075Cs0.975PbBrI2Br, (f) KPbI3, (g) K0.50Cs0.50PbI3. and κ indicate the main perovskite phase and a K-rich non-perovskite phase peaks, respectively. In the case of K0.075Cs0.925PbI2Br there is no appreciable shift with respect to CsPbI2Br either (beside a shift of θ for the 29.2 peak after K + doping which can be caused by (a) a change in I/Br ratio, (b) specimen displacement errors, as discussed in the manuscript). - S9 -

10 Figure S6. Experimental XRD patterns of mechanochemical (a) CsPbI3 (yellow, roomtemperature phase), (b) KPbI3, (c) K0.10Cs0.90PbI3, (d) K0.50Cs0.50PbI3, (e) K0.90Cs0.10PbI3, (f) KI. Figure S7. (a) Experimental XRD pattern of mechanochemical KPbI3 (red) and its best fit (blue). The fit residual is given in (b). The weighted-profile R-factor Rwp = Table S1. Numerical results of the fit in fig. S7a. No. Pos. [ 2θ] d-spacing [Å] Height [cts] (8) (6) S10 -

11 (9) (9) (2) (4) (9) (1) (2) (3) (2) (2) (2) (3) (4) (3) (2) (1) (4) (3) (2) (3) (1) (8) (3) (1) (2) (4) (5) (4) (5) (3) (3) (1) (5) (5) (3) (1) S11 -

12 (2) (5) (5) (3) (7) (3) (3) (4) (5) (2) (4) (2) (4) (4) (3) (3) (8) (5) (4) (9) (3) (3) (5) (1) (5) (9) (2) (1) (2) (6) (2) (7) (4) (1) (2) (3) S12 -

13 (5) (5) (6) (6) (3) (2) Figure S8. CT-selective 39 K MAS NMR spectrum of TiO2-KI, at 11.7 T at 20 khz MAS and 300 K (recycle delay: 10 s, acquisition time: 117 h). TiO2 was activated at 300 C and mechanochemically ground with anhydrous KI in a 10:1 molar ratio. Asterisks indicate spinning sidebands. The symbol indicates a quadrature detection artefact. - S13 -

14 Figure S9. CT-selective spectra for compositions in fig. 2c, e and g in the main text. Acquisition parameters: (a) number of scans: 4096, recycle delay: 12 s, acquisition time: ~13.7 h (b) number of scans: 10240, recycle delay: 1 s, acquisition time: ~3 h, (c) number of scans: 23604, recycle delay: 3 s, acquisition time: ~20 h. Figure S Pb MAS NMR spectra at 11.7 T, 298 K and 20 khz MAS of (a) MAPbI3, number of scans: 27280, recycle delay: 0.1 s, acquisition time: 45 min., processed with 1 khz Lorentzian apodization, (b) δ-cspbi3, number of scans: 30364, recycle delay: 0.1 s, acquisition time: 51 min., processed with 10 khz Lorentzian apodization, (c) KPbI3, number of scans: , recycle delay: 0.02 s, acquisition time: 34 min, processed with 10 khz Lorentzian apodization. 207 Pb chemical shifts were referenced to Pb(CH3)4 using solid Pb(NO3)2 as a secondary reference (-2961 ± 1) ppm. 3 - S14 -

15 Figure S11. 1 H MAS NMR spectra at 21.1 T, 298 K and 20 khz MAS of (a) mechanochemical K0.05Cs0.10FA0.85PbI3 (proton-containing impurities originating from the supplied PbI2 and amounting to ~1% of protons in the sample are indicated). K0.05Cs0.10FA0.85PbI3 prepared by the solution route: (b) after initial annealing in air, (c) after 12 h of vacuum drying. Figure S12. Quantitative 39 K MAS NMR spectrum at 21.1 T, 298 K and 20 khz MAS of the mechanochemical KPbI3 preparation. The spectrum was acquired using a recycle delay of 45 s, hence it is quantitative with respect to KI (T1=9 s) and confirms there is no unreacted KI (δ=59.3 ppm) in the KPbI3 phase. Number of scans: 32. Acquisition time: 24 minutes. - S15 -

16 Details of NMR measurements Table S2. Nuclear 39 K T1 values measured using a saturation-recovery sequence and fitted using a monoexponential function (unless otherwise stated). The uncertainties of fits given are one standard deviation. compound 39 K T1 [s] KI 9.03 ± 0.02 KBr 7.73 ± 0.06 KPbI ± 0.03 KHCO ± ± (biexponential) - S16 -

17 Table S3. Acquisition and processing parameters used for the spectra in fig. 2, 3 and 5 of the main text. composition recycle delay [s] 39 K spectra number of scans acquisition time [h] Lorentzian apodization [Hz] KI KPbI K0.10MA0.90PbI K0.10FA0.90PbI K0.05MA0.10FA0.85PbI KBr KMAFAPb(I,Br) KBr + 2PbBr KBr + PbI K0.05Cs0.10FA0.85PbI3 (CT selective) K0.05Cs0.10FA0.85PbI3 (nonselective) K0.10Cs0.90PbI K0.50Cs0.50PbI K0.90Cs0.10PbI K0.075Cs0.925PbI2Br composition recycle delay [s] 133 Cs spectra number of scans acquisition time [h] Lorentzian apodization [Hz] CsPbI K0.10Cs0.90PbI K0.50Cs0.50PbI K0.90Cs0.10PbI K0.05Cs0.10FA0.85PbI3 (mechanosynthesis) K0.05Cs0.10FA0.85PbI3 (solution synthesis) Cs0.10FA0.90PbI K0.075Cs0.925PbI2Br CsPbI2Br S17 -

18 Details of DFT calculations Cluster generation We start from the assumption, that K is incorporated into the FAPbI3 lattice, replacing the FA cation, without significantly changing the perovskite lattice formed by the [PbI6] 4- octahedra. Thus, in a first step we replace the FA ions of the cubic (black) FAPbI3 crystal structures by K + cations. For the pure FAPbI3 phase the 1 H positions inside the [PbI6] 4- cage were optimized using a periodic system within the density functional theory (DFT) framework and the generalized gradient approximation (GGA) functional PBE within the Quantum Espresso suite. The DFT optimization includes the Grimme dispersion correction and relativistic effects up to spin-orbit couplings. For every calculation we use a plane-wave maximum cut-off energy of 100 Ry and a 2x2x2 Monkhorst-Pack grid of k-points. The K + cations are either positioned at the geometrical centre of the replaced FA ion, leading to a slightly interstitial K +, or at the centre of the surrounding [PbI6] 4- cage. We assemble the final clusters from the relaxed structures as one central [PbI6] 4- cage with surrounding A + ions as KA19Pb8I36 analogue to the ones used in the previous paper by Kubicki et al. 1, ensuring charge compensation, high symmetry and a direct comparability of results. The effect of the surrounding A + ions is investigated by either using FA + or Cs + as A + ions. The 1 H optimized FAPbI3 crystal structure and the coordinates of the clusters are given in the cluster.zip file. Additionally we investigate the 39 K chemical shift and EFG tensor in the perovskite structures given in a recent by work by Kubicki et al. 1 where we replace the central A + ion by K +. The coordinates of the substituted clusters are given in the cluster.zip file. - S18 -

19 Table S4. Source and modifications of the cluster structures used in the DFT calculations. Structure name Original structure Modifications in periodic system (with Quantum Espresso 4 ) Modifications of cluster (with ADF 5,6 KFA 19Pb 8I 36 (A-site replacement) KFA 19Pb 8I 36 (interstitial K + ) KCs 19Pb 8I 36 (A-site replacement) KCs 19Pb 8I 36 (interstitial K + ) KCs 19Pb 8I 36 (from cubic FAPbI 3 ) Cubic FAPbI 7 3 Optimization of 1 H positions Symmetric Replacement of central FA + to K + Cubic FAPbI 7 3 Optimization of 1 H positions Asymmetric Replacement of central FA + to K + Cubic FAPbI 7 3 Optimization of 1 H positions Symmetric Replacement of central FA + to K + and replacement of surrounding FA ions with Cs ions Cubic FAPbI 7 3 Optimization of 1 H positions Asymmetric Replacement of central FA + to K + and replacement of surrounding FA ions with Cs ions Cubic FAPbI 7 3 Optimization of all atomic Symmetric Replacement of positions after replacement of central Cs + to K + FA ions with Cs ions KCs 19Pb 8I 36 (from tetragonal MAPbI 3 ) Tetragonal MAPbI 3 8 Optimization of Cs position after replacement of MA ions with Cs ions Symmetric Replacement of central Cs + to K + KCs 19Pb 8I 36 (from hexagonal CsPbI 3 ) KFA 19Pb 8I 36 (from hexagonal FAPbI 3 ) Hexagonal CsPbI 8 3 None Symmetric Replacement of central Cs + to K + Hexagonal FAPbI 8 3 Optimization of 1 H positions Symmetric Replacement of central FA + to K + EFG tensor calculation We use the Amsterdam Density Functional (ADF) suite 5,6 to perform the EFG tensor calculation within the DFT framework. For the calculations we employ the GGA BP86 9,10 functional including the Grimme dispersion correction 11 and relativistic effects up to spin-orbit couplings within the ZORA approximation. We use all-electron triple-ζ basis sets with two polarization functions (TZ2P). Both the cluster generation and the EFG tensor calculation are set up analogue to the previous paper by Kubicki et al. 1 and in accordance with recent computational studies of systems including heavy atoms. 3, S19 -

20 The NMR chemical shielding calculations set up in a similar manner to the EFG tensor calculations. The chemical shieldings (σ %&' ) are transformed to chemical shifts (δ %&' ) using the linear relation δ *+, = σ.*/ bσ %&', where σ.*/ and b are fit using calculated chemical shieldings from known reference compounds (KI, KBr, KCN and KF). The slope b was either fixed at 1 or used as a fit parameter, resulting in b = To estimate the goodness of the fit parameters, we calculate the root-mean-square-deviation (RMSD) between the experimental and DFT calculated chemical shifts. Note, that in the KCN cluster we have two distinct 39 K sites depending on the relative orientation of the CN ion. If we calculate the average of the two 39 K shifts, assuming motional averaging due to CN rotations, we obtain a RMSD of 1.46 ppm and 3.63 ppm, for either a variable or fixed slope, between the experimental and DFT calculated chemical shifts. The reference chemical shifts and the linear regression models are shown in figure S K experimental shift [ppm] K DFT chemical shielding [ppm] Figure S13. Scatterplot showing 39 K experimental shifts against DFT calculated chemical shielding. The solid lines show the different linear models used to convert chemical shielding to shifts (green: fixed slope (b=1), red: variable slope (b=1.248)). We use KCN, KI, KBr, and KF as reference compounds with the experimental chemical shifts obtained from the work by Moudrakovski and Ripmeester. 20 The crystal structures of KCN, KI, KF and KBr were obtained from the works by Price et al. 21, Van Den Bosch et al. 22, Broch et al. 23 and Ott. 24 The DFT accuracy for 39 K chemical shift calculations has been investigated by Wu et al. 25 and Shimoda et al. 26, where they report an 39 K chemical shift root-mean-square deviation between experiment and calculation of around 4-8 ppm. All the ADF input and output files as well as the coordinates of the used clusters are given in the cluster.zip file. - S20 -

21 Table S5. DFT calculated and experimental chemical shieldings, chemical shifts and EFG tensor parameters. Structure DFT chemical shielding [ppm] Experimental shifts [ppm] DFT shifts (s ref=1220, b=1.0) [ppm] DFT shifts (s ref=1512, b=1.248) [ppm] C Q [MHz] η V zz [Vm -2 ] (a) KF (b) KBr (c) KI (d) KCN (site 1) (e) KCN (site 2) (f) KCN (avg.) (g) KFA 19Pb 8I 36 (A-site replacement) (h) KFA 19Pb 8I 36 (interstitial K + ) (i) KCs 19Pb 8I 36 (A-site replacement) (k) KCs 19Pb 8I 36 (interstitial K + ) (l) KCs 19Pb 8I 36 (from cubic FAPbI 3 ) (m) KCs 19Pb 8I 36 (from tetragonal MAPbI 3) (n) KCs 19Pb 8I 36 (from hexagonal CsPbI 3 ) (o) KFA 19Pb 8I 36 (from hexagonal FAPbI 3 ) S21 -

22 References (1) Kubicki, D. J.; Prochowicz, D.; Hofstetter, A.; Zakeeruddin, S. M.; Grätzel, M.; Emsley, L. J. Am. Chem. Soc. 2017, 139, (2) Prochowicz, D.; Franckevicius, M.; Cieslak, A. M.; Zakeeruddin, S. M.; Gratzel, M.; Lewinski, J. J. Mater. Chem. Mater. Energy Sustain. 2015, 3, (3) Dybowski, C.; Smith, M. L.; Hepp, M. A.; Gaffney, E. J.; Neue, G.; Perry, D. L. Appl. Spectrosc. 1998, 52, (4) Paolo Giannozzi and Stefano Baroni and Nicola Bonini and Matteo Calandra and Roberto Car and Carlo Cavazzoni and Davide Ceresoli and Guido L Chiarotti and Matteo Cococcioni and Ismaila Dabo and Andrea Dal Corso and Stefano de Gironcoli and Stefano Fabris and Guido Fratesi and Ralph Gebauer and Uwe Gerstmann and Christos Gougoussis and Anton Kokalj and Michele Lazzeri and Layla Martin-Samos and Nicola Marzari and Francesco Mauri and Riccardo Mazzarello and Stefano Paolini and Alfredo Pasquarello and Lorenzo Paulatto and Carlo Sbraccia and Sandro Scandolo and Gabriele Sclauzero and Ari P Seitsonen and Alexander Smogunov and Paolo Umari and Renata M Wentzcovitch. J. Phys. Condens. Matter 2009, 21, (5) Guerra, C. F.; Snijders, J. G.; Te Velde, G.; Baerends, E. J. Theor. Chem. Acc. 1998, 99, (6) Te Velde, G.; Bickelhaupt, F. M.; Baerends, E. J.; Fonseca Guerra, C.; Van Gisbergen, S. J. A.; Snijders, J. G.; Ziegler, T. J. Comput. Chem. 2001, 22, (7) Weller, M. T.; Weber, O. J.; Frost, J. M.; Walsh, A. J. Phys. Chem. Lett. 2015, 6, (8) Stoumpos, C. C.; Malliakas, C. D.; Kanatzidis, M. G. Inorg. Chem. 2013, 52, (9) Perdew, J. P. Phys. Rev. B 1986, 33, (10) Becke, A. D. Phys. Rev. A 1988, 38, (11) Grimme, S. J. Comput. Chem. 2006, 27, (12) Lenthe, E. van; Baerends, E. J.; Snijders, J. G. J. Chem. Phys. 1993, 99, (13) van Lenthe, E.; Baerends, E. J.; Snijders, J. G. J. Chem. Phys. 1994, 101, (14) van Lenthe, E.; Ehlers, A.; Baerends, E.-J. J. Chem. Phys. 1999, 110, (15) Giorgi, G.; Fujisawa, J.-I.; Segawa, H.; Yamashita, K. J. Phys. Chem. C 2015, 119, (16) Dmitrenko, O.; Bai, S.; Beckmann, P. A.; van Bramer, S.; Vega, A. J.; Dybowski, C. J. Phys. Chem. A 2008, 112, (17) Alkan, F.; Dybowski, C. Phys. Chem. Chem. Phys. 2014, 16, (18) Alkan, F.; Dybowski, C. J. Phys. Chem. A 2016, 120, (19) Even, J.; Pedesseau, L.; Jancu, J.-M.; Katan, C. J. Phys. Chem. Lett. 2013, 4, (20) Moudrakovski, I. L.; Ripmeester, J. A. J. Phys. Chem. B 2007, 111, (21) Price, D. L.; Rowe, J. M.; Rush, J. J.; Prince, E.; Hinks, D. G.; Susman, S. J. Chem. Phys. 1972, 56, (22) Van den Bosch A.; Dresselaers J.; Vansummeren J.; Hovi M. Phys. Status Solidi A 1972, 11, (23) Broch, E.; Oftedal, I.; Pabst, A. Z. Fuer Phys. Chem. Abt. B Chem. Elem. Aufbau Mater. 1929, 3, (24) Ott, H. Z. Krist. 1926, 63, (25) Wu, G.; Gan, Z.; Kwan, I. C. M.; Fettinger, J. C.; Davis, J. T. J. Am. Chem. Soc. 2011, 133, S22 -

23 (26) Shimoda, K.; Yamane, A.; Ichikawa, T.; Kojima, Y. J. Phys. Chem. C 2012, 116, S23 -