Optical analysis of CH3NH3SnxPb1–xI3 absorbers: a roadmap for perovskite-on-perovskite tandem solar cells† †Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ta04840d Click here for additional data file.

We propose a novel tandem architecture design in which both top and bottom cells contain perovskite absorbers.

In order to characterize the samples outside a glovebox, CH 3 NH 3 Sn x Pb 1-x I 3 films were encapsulated using a PMMA protecting layer or a cover glass combined with an epoxy sealing. Figure S1 and S2 present XRD and light absorption measurements at various aging states of films encapsulated with the different procedures. The glass cover allows maintaining a N 2 atmosphere around the film, avoiding moisture contact with the perovskite material, resulting in samples that do not degrade over months.

FIGURE S1
XRD patterns of CH 3 NH 3 SnI 3 films aged for 2 weeks with different encapsulation procedures: standard PMMA protecting the perovskite layer (top diagram), and double glass sealed with epoxy (bottom diagram). While the perovskite structure is lost when the first method is employed, for the latter no degradation of the film is observed.
The crystal structures of compositions varying from 0 to 100% Sn were analyzed by XRD ( Figure 1g). As shown by Hao et al., 1 the two peaks between 2θ 22 and 25˚ which clearly indicate the tetragonal structure of the 0 and 15% Sn samples, disappear with Sn concentrations of 50% or above, where the structure moves towards the cubic. For all compositions two peaks are dominant in the pattern. One is located at just above 14˚ which can be indexed as (002) and (110) for the tetragonal I4cm space group and those samples with less than 50% Sn, and as (001) for the cubic P4mm for samples above 50% Sn. The second prominent peak is located at 28.5˚, which can be indexed as (004) and (220) for the tetragonal (0 and 15% Sn) and (002) for the cubic (50, 85 and 100% Sn). 2 A gradual shift in both of the prominent peaks is detected as seen in Figure 1h and i, confirming what has been shown in similar studies. 1,3

FIGURE S2
Experimental total absorptance spectra of CH 3 NH 3 Sn 0.5 Pb 0.5 I 3 films aged for 4 weeks with the two encapsulation procedures: standard PMMA (left panel), and double glass sealed with epoxy (right panel). Black solid lines represent fresh films whose optical response was measured immediately after deposition. Grey dashed lines correspond to fimls expose to air for 4 weeks. While the optical response expected for this kind of perovskite structure is lost with time when the first encapsulation method is employed, for the latter no degradation of the film is taken place. Sn (green line), 85% Sn (orange line) and 100% Sn (red line).
The band gap of the pure lead perovskite is found at Eg=1.60 eV. As the fraction of Sn in the precursor solution increases, the band edge shifts to longer wavelengths, reaching a minimum value in energy, Eg = 1.17 eV, when the Sn content is 85%. After this monotonous redshift, the band gap energy increases, reaching a value of Eg = 1.21 eV for the pure Sn-based perovskite (see Figure 2b). Such band gap evolution, already discussed elsewhere, originates from a difference in the spin-orbit coupling and a crystal phase change from tetragonal to cubic as XRD analysis presented in Figure 1g indicates.  We prepared devices of CH 3 NH 3 PbI 3 and CH 3 NH 3 Sn 0.15 Pb 0.85 I 3 , as shown in Figure S8. The unoptimized device architecture of FTO/compact-TiO2/mesoporous-TiO2/Perovskite/HTM/gold was used for this study, as seen in the cross-sectional SEM in Figure S8a (Photographs of films and devices are displayed in Figure S9). Remarkably, it was found that careful selection of the hole selective layers used (and their dopants), played an important role in the performance, and, presumably, in the degradation mechanism associated with this interface. Figure S8b shows the performance of a CH 3 NH 3 Sn 0.15 Pb 0.85 I 3 device using spiro-OMeTAD with dopants compared to polytriarylamine polymer (PTAA) without any additives. Devices using doped spiro-OMeTAD showed a severe decrease in the photocurrent respect to the undoped PTAA analogue, i.e.
~4 mA·cm -2 vs 25 mA·cm -2 . Since spiro-OMeTAD is well known to require additives (or alternatively oxygen, causing oxidation of the molecule) for improving its mobility, 9,10 PTAA was used instead, which has been shown to perform rather well without additives. 11 Indeed, devices with PTAA and no additives performed much better than those with Spiro-OMeTAD and no additives. We attribute this effect to the oxidization of the Sn 2+ to Sn 4+ by the Co, 4-tert-Butylpyridine (TBP), and Li additives used to obtain the highest photovoltaic performance. PTAA with no additives was therefore used as the HTM in order to show the potential of the current of the devices. Figure S8b shows that devices using CH 3 NH 3 Sn 0.15 Pb 0.85 I 3 as the absorber yields substantially higher photocurrents (25 mA·cm -2 ) than those of the pure Pb analogues (20 mA·cm -2 ). The Sn-Pb alloy also displays relatively high voltages comparable to those reported by others. 12 FF for devices containing both compounds show low values, however, we note that this is proof-of-concept experiment to justify our trends in the currents predicted. The blue curve shows the JV characteristics of a CH 3 NH 3 PbI 3 in which PTAA acts as HTM.  (5)