Supplementary Figure S1. Verifying the CH 3 NH 3 PbI 3-x Cl x sensitized TiO 2 coating UV-vis spectrum of the solution obtained by dissolving the spiro-ometad from a perovskite-filled mesoporous TiO 2 layer in cholorobenzene. The same preparation conditions were used as for the perovskite sensitized devices presented in the main text and depicted in Figure 1a. The arrow points to the extinction coefficient of spiro-ometad, while the table gives the relevant numbers used to calculate the fraction of the total porosity filled by spiro-ometad, and hence not by the perovskite.
Supplementary Figure S2. Aging of dye sensitized solar cells Current density vs voltage curves (JV curves ) for C106 (a) and D102 (b) sensitized solid state dye-sensitized solar cells taken under vacuum conditions after different periods of exposure to simulated sun light.. a) b)
Figure S3. Aging of dye sensitized solar cells with a UV filter Current density vs voltage curves for C106 (a) and D102 (b) sensitized solid state dye-sensitized solar cells taken under vacuum conditions after several solar simulated light exposure times, but now a 435 nm cutoff UV filter is placed between the light source and the solar cells. a) b)
Figure S4. Monitoring the absorption spectrum of aged perovskite films The UV-Vis spectra of complete solar cells without the gold electrodes, with (solid lines) and without (dashed lines) encapsulation in a nitrogen filled glovebox are presented before (grey) and after (blue) 3 hours aging under full spectrum solar simulated light (100 mw cm -2 ). The offset for encapsulated samples is due to the extra reflection of the glass encapsulation.
PCE (%) Supplementry Figure S5. Aging of TiO 2 -based solar cells with a perovskite capping layer (a) Normalized power conversion efficiency (PCE) of a TiO 2 based perovskite solar cell with the pores completely filled with perovskite. (b) SEM image and description of the solar cell architecture used for the measurement in (a). 1.00 0.75 0.50 0.25 0.00 0 1 2 3 4 Time (hrs) Silver electrode Spiro overlayer Perovskite overlayer TiO 2 particles + perovskite FTO Glass a) b)
Supplementary Figure S6. Aging devices prepared in a nitrogen filled glovebox Current density-voltage curves for a representative D102 sensitized solid state dye sensitized solar cell fabricated (every step after the TiCl 4 treatment) in a nitrogen filled glovebox. The black open squares depict the JV curve for the solar cell measured in a nitrogen atmosphere, without any air exposure. The open red circles depict the JV curve for the same solar cell after 48 hours of air exposure. The open blue triangles depict the JV curve of the cell after another 48 hours in a nitrogen filled glovebox, and tested in a nitrogen environment after 5 minutes of unfiltered light exposure. These cells were measured in a small testing chamber at 70 mw cm -2 solar simulated light intensity.
RecombinationLT (s) Transport LT (s) Figure S7. Small perturbation photovoltage and photocurrent decay measurements Short circuit transport and recombination lifetimes for encapsulated perovskite sensitized TiO 2 solar cells before (black squares and circles) and after (blue squares and circles) an hour of exposure to full spectrum solar simulated sunlight at 100 mw cm -2. 10-3 Non-Encapsulated 1 Non-Encapsulated 2 Encapsulated UV Aged 1 Encapsulated UV Aged 2 10-4 10-3 10-4 0.01 0.1 1 Light Intensity (Suns)
Supplementary Figure S8. Change in background conductivity of TiO 2 with UV aging Time resolved conductivity of dye (C106) sensitized mesoporous TiO 2 films before (blue) and after (grey) UV light (40 mw cm -2 at 365 nm) exposure. The film is photoexcited by a 550 nm laser pulse.
Figure S9. Photoinduced conductivity changes in spiro-ometad Time resolved conductivity of spiro OMeTAD infiltrated into C106 sensitized mesoporous TiO 2 before (grey) and after (blue) UV light exposure (same as in Supplementary Figure S8). A probe wavelength of 510 nm is chosen to monitor the concentration of oxidized spiro-ometad molecules.
Supplementary Table S1. Performance parameters of TiO 2 based solar cells before and after encapsulation We depict the initial performance parameters of the solar cells used for Figure 2 in the main text, before and after encapsulation. The high performances are clearly indicative of well-functioning solar cells of this type. The quotation marks are used to depict which solar cells we refer to in relation to those in Figure 2 of the main text. Jsc PCE (%) Voc FF "Non Encapsulated" 12.7 5.9 0.73 0.62 "UV Filtered" Before Seal 15.17 6.2 0.73 0.57 "UV Filtered" After Seal 12.9 5.7 0.93 0.47 "Sealed" Before Seal 13.8 7.1 0.92 0.55 "Sealed After Seal 15.2 6.2 0.73 0.56
Supplementary Note 1. Verifying the CH 3 NH 3 PbI 3-x Cl x sensitized TiO 2 coating To verify the solar cells structure depicted in Figure 1a of the main text, we perform a spiro-ometad pore filling analysis similar to that previously presented by Ding et al. 53 The results are presented in Supplementary Figure S9. From Supplementary Figure S9, we can see that the total mass of spiro-ometad in the sample is significantly larger than the mass of spiro-ometad in the capping layer (assuming a density of 1 g cm -3 ). 54 Moreover, when we go on to calculate the pore filling fraction (using the remaining mass of spiro-ometad after subtraction of the capping layer and a porosity of 50 %), 53, 54 we estimate a spiro-ometad pore filling fraction of 0.82. This means that the light absorber is taking up negligible space inside the pores, and hence must be acting as just a sensitizer. Supplementary Note 2. Aging of dye sensitized solar cells Here, the dye sensitized solar cells were tested in vacuum under 70 mw cm -2 AM 1.5 solar simulated irradiation (this is the intensity that is transmitted through evacuated sample holders widow when placed under the solar simulator). Both organic and ruthenium complex dyes display rapid degradation under these conditions, similar to that observed with the perovskite absorber on mesoporous TiO 2, as presented in Figure 1 in the main text. After performing the degradation measurements shown in Supplementary Figure S2, we allowed these devices to rest in air for 24 hours after which point they fully recover in performance when tested in air. This fits well with our proposed mechanism of adsorbed oxygen pacifying surface traps since re-adsorption over time reverts the solar cell to its well operating condition. After recovery, we placed the same cells back in the testing box and flushed with nitrogen and evacuated for the tests with a 435 nm UV filter present. There is an immediate drop in photocurrent because all the light below 435 nm is simply cut out of the spectrum, and there is an additional 10% reflection across the whole spectrum due to the extra air/glass/air interface. Despite this however, it is apparent that the cells display greatly improved stability compared to the same cells without the UV filter (Supplementary Figure S3). Supplementary Note 3. Aging TiO 2 based perovskite solar cells with a perovskite capping layer The presence of this capping layer is evidence that the pores of the TiO 2 are completely (or close to completely) filled by the perovskite; as the solvent evaporates during the spin coating procedure and precursor diffuses from the wet reservoir into the pores, the precursor concentration inside the pores will increase until it is saturated, and only then will a capping layer form. This has been already described for the infiltration of spiro- OMeTAD. 53, 55 Hence, the presence of a capping layer in Supplementary Figure S4 is evidence that the pores are completely, or close to completely, filled with perovskite. The hole transporter is spin coated on top of the perovskite capping layer, to give a sandwich-like device structure (TiO 2 perovskite / perovskite capping layer / spiro-ometad capping layer) as depicted in the SEM image. These devices were then encapsulated and
measured in the same manner as that described in the main text experimental section. We see in Supplementary Figure S4, that under unfiltered solar simulated light, these devices also suffer from a rapid deterioration in performance, dropping to 50 % of the initial performance within 5 hours. Supplementary Note 4. Solid state DSSCs prepared in a nitrogen-filled glovebox For Supplementary Figure S6, we perform all of the steps for making the solid state dye (D102) sensitized solar cells after the TiCl 4 treatment in a nitrogen-filled glovebox. After two days in air, all of the device parameters 25, 26, 28 are improved, as might be expected from the increased oxygen-induced doping of the hole transporter. After two days back in the nitrogen filled glovebox, and several minutes under the solar simulator, the solar cells have again dropped dramatically in photocurrent. Notably, the slopes of the JV curve near open circuit are very similar to those seen in air, suggesting that the series resistance through the hole transporter has not been significantly altered by placing the devices back in nitrogen. This suggests that after oxygen-induced p-doping of the hole transporter, which is stable even after the removal of oxygen, 26 the solar cells suffer from a decreased photocurrent when tested in an oxygen-free atmosphere, which is unlikely to be related to a change in conductivity of the hole transporter. This is also demonstrated in Supplementary Figure S9. Hence, it is very likely that the degradation mechanism is taking place at the TiO 2 surface, which we investigate in the main text. Supplementary Note 5. Small perturbation photovoltage and photocurrent decay measurements Supplementary Figure S7 shows the transport and the short circuit recombination lifetimes for both nonencapsulated (without any aging) and encapsulated (after 5 hrs aging under unfiltered solar simulated light) perovskite sensitized TiO 2 solar cells. Surprisingly, we show that that a given light intensity, the transport lifetimes are unaffected, while the recombination rate seems somewhat slowed after aging. The solar cells only produce approximately 20% of the initial photocurrent after the aging, yet the transport rate does not seem slowed. The implications of this finding are presented in the main text. Supplementary Note 6. Change in background conductivity of TiO 2 with UV aging In Supplementary Figure S6, we show that the background conductivity (t < 0) increases from 10-5 to 10-4 S cm - 1 after 5 minutes of UV aging (40 mw at cm -2 365 nm). This is due to the increase in long-lived free electron concentration that are photogenerated by UV absorption in the TiO 2 in the absence of oxygen as a scavenger. 30, 56, 57 This results in an increase in overall conductivity. However, the change in the conductivity due to electron injection from the excited (t = 0) dye sensitizer is less. We discuss the implications of this change in the main text.
Supplementary Note 7. Photoinduced conductivity changes in spiro-ometad In Supplementary Figure S9, the bottom panel depicts the associated absorption change, where the spiro- OMeTAD cation is probed at 510 nm. We ensure that we selectively monitor the hole conductivity since there is a 200 nm capping layer of spiro-ometad over the sensitzied mesoporous TiO 2 film, upon which we evaporate our contacts. 28, 56 It appears that upon UV aging, there is little difference between the initial change in conductivity, and the number of holes generated in spiro-ometad is again similar. The rate of recombination however, as determined by both the transient conductivity and TAS decay, is marginally increased. Such an increase in recombination rate has been previously assigned to deep surface traps, and can help to explain the decreased photovoltages observed for aged cells in Figure 2. 42-44 This is different to what we measured by small perturbation photovoltage decay of the perovskite-sensitized solar cells (Fig. S5). The reason for the discrepancy could be that if the trap sites in the TiO 2 are deep and localized in energy, then carriers injected into these sites will contribute very little to the photovoltage of the solar cells, so that their recombination will not show up in the photovoltage decay measurements. It is also possible that the difference lies in varying degrees of surface coverage of the TiO 2 by dye or perovskite. Regardless of the subtelties of charge recombination, we show that there is negligible change to the photoinduced hole-transfer efficiency nor mobility in the spiro-ometad after aging, in contrast to what we observed for the photoinduced electrons in TiO 2.
Supplementary Methods Perovskite pore filling analysis Here, we prepare solar cells of the structure in Figure 1a as described in the methods of the main text, but employing an additive-free spiro-ometad and no gold electrode. After measuring the total layer thickness, the sample is placed in a known volume of chlorobenzene for 10 minutes to allow the spiro-ometad to dissolve (the remainder of the cell is unaffected). Knowing the extinction coefficient of spiro-ometad at 393 nm (74700 M -1 cm -1 ), 53 we take a UV vis spectrum of the resulting solution and calculate the total mass, and using a density of 1 g cm -3 the total volume, 54 of spiro-ometad that was in the sample. Measuring the film thickness after dissolution of the spiro-ometad, we can make an approximate estimation of the pore filling fraction as has been preciously described by Ding et al. 53 Knowing the pore filling fraction of spiro-ometad, the maximum pore filling available for the perovskite absorber is the remaining fraction. Aging of dye sensitized solar cells under vacuum The samples were prepared as discussed in the main methods section, placed in the testing box, which was flushed with nitrogen three times, after which a vacuum was drawn using a roughing pump for 3 hours. The J- V curves were then taken as described in the main methods section. TiO 2 -based solar cells with a perovskite capping layer We have prepared devices with a thin (200 nm) layer of TiO 2 by using a 1:4 dyesol (18NR-T) to ethanol (v / v) ratio. This we fill completely with our mixed halide perovskite by spin coating a concentrated (40 wt % instead of 20 wt %) solution of the precursors (solution composition described in methods section in the main text). Small perturbation photovoltage and photocurrent decay measurements Small perturbation photovoltage and photoccurent decays were carried out at as described in the literature, using red LEDs to perturb a white light LED background light source. 17,29
Supplementary References 53 Ding, I. et al. Pore Filling of Spiro OMeTAD in Solid State Dye Sensitized Solar Cells: Quantification, Mechanism, and Consequences for Device Performance. Advanced Functional Materials 19, 2431-2436 (2009). 54 Docampo, P. et al. Pore Filling of Spiro OMeTAD in Solid State Dye Sensitized Solar Cells Determined Via Optical Reflectometry. Advanced Functional Materials 22, 5010-5019 (2012). 55 Snaith, H. J. et al. Charge collection and pore filling in solid-state dye-sensitized solar cells. Nanotechnology 19, (2008). 56 Nelson, J., Eppler, A. M. & Ballard, I. M. Photoconductivity and charge trapping in porous nanocrystalline titanium dioxide. J. Photochem. Photobiol. A: Chem. 148, 25-31, (2002). 57 Pomoni, K., Vomvas, A. & Trapalis, C. Electrical conductivity and photoconductivity studies of TiO< sub> 2</sub> sol gel thin films and the effect of N-doping. J. Non-Cryst. Solids 354, 4448-4457 (2008).