Humidity versus photo-stability of metal halide perovskite films in a polymer matrix

Despite the high efficiency of over 21% reported for emerging thin film perovskite solar cells, one of the key issues prior to their commercial deployment is to attain their long term stability under ambient and outdoor conditions. The instability in perovskite is widely conceived to be humidity induced due to the water solubility of its initial precursors, which leads to decomposition of the perovskite crystal structure; however, we note that humidity alone is not the major degradation factor and it is rather the photon dose in combination with humidity exposure that triggers the instability. In our experiment, which is designed to decouple the effect of humidity and light on perovskite degradation, we investigate the shelf-lifetime of CH3NH3PbI3 films in the dark and under illumination under high humidity conditions (Rel. H. > 70%). We note minor degradation in perovskite films stored in a humid dark environment whereas upon exposure to light, the films undergo drastic degradation, primarily owing to the reactive TiO2/perovskite interface and also the surface defects of TiO2. To enhance its air-stability, we incorporate CH3NH3PbI3 perovskite in a polymer (poly-vinylpyrrolidone, PVP) matrix which retained its optical and structural characteristics in the dark for ∼2000 h and ∼800 h in room light soaking, significantly higher than a pristine perovskite film, which degraded completely in 600 h in the dark and in less than 100 h when exposed to light. We attribute the superior stability of PVP incorporated perovskite films to the improved structural stability of CH3NH3PbI3 and also to the improved TiO2/perovskite interface upon incorporating a polymer matrix. Charge injection from the polymer embedded perovskite films has also been confirmed by fabricating solar cells using them, thereby providing a promising future research pathway for stable and efficient perovskite solar cells.


Introduction
The remarkable improvement in the photoconversion efficiency (PCE) of organometallic halide perovskite solar cells (PSCs) from an initial 3.8% in 2009 1 to 421% in 2016 2 has taken the research community by surprise.The increase in the PCE is due to optimization of typically employed perovskite (CH 3 NH 3 PbX 3 , where X = I, Cl, Br) by adding formamidinium (FA) and/or cesium (Cs) into a methylammonium cation (MA) [2][3][4] and enabling better charge extraction at electron and hole selective contacts. 5The global interest in this emerging class of solar cells is due to their high absorption coefficient (1.5 Â 10 4 cm À1 at 550 nm), 6 ambipolar charge transport, 7,8 direct band gap (B1.55 eV for CH 3 NH 3 PbI 3 ) 9 and high charge carrier mobility (B66 cm 2 V À1 s À1 for CH 3 NH 3 PbI 3 ) 10 which enables a diffusion length up to 175 mm 11 in single crystals.Despite these excellent properties usually only found in inorganic semiconductors, high efficiency perovskite films are solution processible under ambient conditions.Although their device design was initially inspired from that of dye-sensitized solar cells, where an absorber layer is anchored to a mesoporous layer of an n-type metal oxide semiconductor such as TiO 2 , 12,13 subsequent research on perovskite solar cells (PSCs) demonstrated that the mesoporous layer giving high PCEs of 18-19% is not needed and scaffold-free (planar) PSCs achieve the same high efficiencies. 14,15imilarly, we find reports of high efficiency (10-13%) in large area modules, [16][17][18][19] in flexible PSCs (B15.6%), 20and even the first flexible perovskite solar modules. 21PSCs have proven to work in dim-light, 22 and the possibility of semi-transparency [23][24][25] has strengthened their case for a deployable PV technology e.g. for building integration.Despite the remarkable progress shown to improve the PV performance of PSCs, the key challenges prior to their commercial deployment are their toxicity (due to presence of lead), 26,27 scalability and long-term operational stability. 28or a new PV technology to reach the market, it has to offer four key parameters: (i) high efficiency (comparable to thin film and silicon solar cells on the market), (ii) low cost o$ 0.05 kW h, 29 to reach grid parity, (iii) long lifetime (420 years), and (iv) added functionality (flexibility, transparency and aesthetics). 30PSCs need to be considerably better in at least one of these key parameters and have to match the others to be commercially interesting compared to conventional solar technologies.After the past five years of efficiency driven research on PSCs, the focus now is shifting to understanding the underlying processes governing the high efficiency and also to obtaining long-term stable devices.2][33] Whereas humidity induced instability in these devices arises from the fact that the initial precursors employed to synthesize perovskites, i.e., CH 3 NH 3 PbI 3 , are water soluble, the reasons for photodegradation, particularly at low light, are not yet clear despite the initial reports. 34,35Due to the humidity sensitive nature of these perovskites, their thin films are often prepared in situ which may hinder their mass production.However, when exposed to outdoor conditions, perovskite films undergo structural and morphological changes which affect their optical properties and consequently results in the decrease of their PV performance over time.For example, a significant decrease in device performance is reported in the presence of moisture when a CH 3 NH 3 PbI 3 perovskite film undergoes rapid decomposition to PbI 2 (as evident from the change of color from dark brown to yellow). 36,37Yang et al. observed that, besides humidity, molecular oxygen and light can also lead to decomposition of CH 3 NH 3 PbI 3 . 38A report from Aristidou et al. also affirmed decomposition of CH 3 NH 3 PbI 3 to PbI 2 and I 2 , and methylamine under ambient conditions. 39t is important to note that although humidity is reported to affect the long-term stability of PSCs, controlled moisture is one of the prerequisites to obtain high photoluminescence and thereby a high PCE in perovskite films. 40Controlled moisture reduces trap states due to the partial solvation of the MA ion and allow ''self-healing'' of the perovskite crystal. 41However, excess humidity is shown to decompose perovskite crystals due to their water solubility. 42,43Various strategies such as device encapsulation to avoid moisture ingress, incorporation of UV-filters to avoid UV-induced instability or device fabrication and packaging under inert conditions prior to their exposure to ambient conditions are proposed to overcome humidity instability. 44,45However, such strategies lead to the increase of device fabrication cost and also add difficulties in device fabrication.In order to fully obtain the key features of easy processability and roll-to-roll compatible mass production, not only do perovskite thin films need to be synthesized under ambient conditions but also, more importantly, their structural, optical and morphological characteristics need to be maintained during their long-term operation under outdoor conditions.
Besides external factors such as humidity, oxygen, temperature and light (including UV) irradiation, the perovskite layer is also known to degrade internally by reacting with the TiO 2 nanoparticle (NP) layer underneath, as also suggested in our previous reports. 18,46Such a reactive interface might lead to ion migration across the interface.The surface defects on TiO 2 nanocrystals have also been shown to act as a humidity trap leading to decomposition of perovskite crystals which is overcome by post treatment of TiO 2 layers with CdS, 47 and Sb 2 S 3 48 to avoid surface degradation.In a similar study, Li et al. 49 used caesium bromide (CsBr) to modify the TiO 2 interface to improve the photocatalytic activity of the device and also reduced interfacial defect sites.It is thereby evident that optimization of the TiO 2 / CH 3 NH 3 PbI 3 interface is crucial to improve the photocatalytic properties of the device, not only to protect it from external stress factors but also to avoid a possible reaction of perovskite with the selective contact underneath.The research on exploring the role of interfacial degradation becomes more intriguing after the report that perovskite films, by themselves, can withstand humidity and it is rather the photon dose that leads to device degradation. 50 systematic experiment to decouple the humidity and light stability of perovskite films is therefore crucial in order to understand the origin of such a degradation.Herein, we report a systematic experiment to investigate the stability of a CH 3 NH 3 PbI 3 thin film in light as well as in the dark under high humidity conditions (Rel.H. 70-80%) to investigate the origin of instability.We then synthesize a CH 3 NH 3 PbI 3 -poly-(vinylpyrrolidone) (PVP) matrix to protect it from humidity, oxygen and light and to further investigate the role of interfacial degradation.A systematic investigation of polymer embedded perovskite (PeP) films at different concentrations (0 wt%, 5 wt%, 10 wt% & 20 wt%) demonstrated many fold enhanced air stability of CH 3 NH 3 PbI 3 films.The morphological, structural and optical investigation of PeP thin films under such high humidity conditions showed that they were stable even after 2000 h of shelf-life testing in the dark and B800 h in room light soaking, significantly higher than a pristine perovskite film (stable for o600 h in the dark and in o100 h in light).It is noteworthy that the first demonstration of perovskite films in a polymer (PEG) was recently reported with a PCE of B16%; however the focus of the study was towards humidity stability only. 51We report a detailed analysis of perovskite films in order to understand the reasons for the stable performance of PeP films and the different degradation mechanisms in humidity only and 'humidity coupled with light'.

Processing of perovskite films at high humidity
Glass plates that are used as a substrate to deposit perovskite films were sequentially cleaned with detergent, acetone and ethanol under sonication for 15 min each.Mesoporous TiO 2 (mp-TiO 2 ) paste was diluted in absolute ethanol at a volumetric ratio (1 : 2) and was spin coated onto the cleaned glass slides at 2000 rpm for 40 s to form a mesoporous scaffold layer.The films were dried on a hotplate at 100 1C for 15 min prior to annealing at 450 1C for 30 min in air to evaporate the organic binder in the TiO 2 paste.The CH 3 NH 3 PbI 3 precursor solution was synthesized by dissolving CH 3 NH 3 I and PbI 2 (1 : 1 molar ratio) in 40 wt% solution of anhydrous DMF and stirring at 80 1C for 3 h.This solution was spin casted onto mp-TiO 2 /glass films at 2000 rpm for 40 s and sintered (100 1C for 1 h and 125 1C for 15 min) to form CH 3 NH 3 PbI 3 crystals.To improve the surface coverage of perovskite, the TiO 2 coated glass substrates were preheated at 100 1C.The deposition was carried out at room temperature (25-30 1C) and high humidity (70-80%).

Polymer-perovskite matrix for air stable films
In order to improve the air-stability of perovskite films, CH 3 NH 3 PbI 3 was embedded in a PVP matrix (PPM).A given amount of PVP powder was added to perovskite precursors during solution synthesis to form PPM in the following weight ratio of perovskite : PVP: (i) 100 : 0, (ii) 95 : 5, (iii) 90 : 10, and (iv) 80 : 20.The samples will be denoted as 0%, 5%, 10% and 20% throughout the article and the label assigned corresponds to the ratio of PVP to perovskite in a particular PeP film.To synthesize CH 3 NH 3 PbI 3 -PVP matrix solution and to deposit it on the mp-TiO 2 coated glass substrate, the procedure as described before has been repeated.

Shelf-life testing to investigate air stability of perovskite films
3][54] The pure and polymer coated perovskite films were stored in (i) dark (in a desiccator B25-30 1C, 55-70% Rel.H.) and in room light (light intensity B5 mW cm À2 , 25-30 1C, 70-80% Rel.H.). The shelf life is monitored until degradation is tracked in each sample which is confirmed from their (i) change in color, (ii) relative changes in the absorbance and photoluminescence, and (iii) XRD spectra.The lifetime of each sample with different polymer concentrations kept in light or in the dark was noted in order to measure the lifetime of each film.In order to probe the degradation of the films and the failure mechanism, we carried out their surface characterization using field emission scanning electron microscopy (FESEM, JEOL), optical characterization techniques such as UV-visible absorption (UV-2600, Shimadzu) and photoluminescence (PL, NIR 300/2 Edinburgh), and chemical and structural characterization using X-ray diffraction (XRD) and Fourier transform infrared spectroscopy (FTIR).

Fabrication of perovskite solar cells and characterization
PSCs employing pure perovskite and their polymer incorporating counterparts were fabricated in an inert atmosphere.The details of fabrication are given in the ESI.† The photovoltaic performance of the PSCs is measured using an AM 1.5G solar simulator in an inert atmosphere at a light intensity of B85 mW cm À2 and using a Fraunhofer ISE certified Si reference diode with an attached KG5 filter.A shadow mask defined the active area of B0.125 cm 2 for the PSCs.

Synthesis and shelf-life analysis of perovskite films under high humidity conditions
Fig. 1a shows a cross sectional view of the CH 3 NH 3 PbI 3 thin film (500-600 nm) coated over a mesoporous TiO 2 layer (B500 nm).Clearly, the high humidity (70-80%) favoured growth of a dense film with large CH 3 NH 3 PbI 3 crystals (size up to 200-250 nm), contrary to the perovskite films synthesized in an inert atmosphere in our previous reports (Fig. SI-1, ESI †) with a smaller crystal size and large grain boundary density.This is due to the fact that humidity may induce grain creeps resulting in merging of neighbouring perovskite grains. 55The larger CH 3 NH 3 PbI 3 crystals obtained in the present work with less grain boundaries may result in efficient charge transport across the perovskite film.7][58] The perovskite films absorb most of the visible light o600 nm which gradually decreases in the 600-780 nm range (Fig. 2c).In order to validate the presence of methyl amine (MAI) in PbI 2 , i.e., formation of CH 3 NH 3 PbI 3 , we carried out FTIR measurements.The FTIR transmittance spectrum is recorded in the range 500-4000 cm À1 (data from 1000-4000 cm À1 are only shown in Fig. 1c).The peaks recorded for symmetrical and asymmetrical vibrations in MAI are at 3459, 2959, 1560, 1489, and 1404 cm À1 which correspond to N-H stretching, C-H stretching, N-H bending, C-N stretching and C-H bending, respectively whereas in the case of CH 3 NH 3 PbI 3 , the corresponding peaks are at 3452, 2925, 1627, 1460, 1382 cm À1 .The shift and peak broadening observed in the case of perovskite is due to the interaction of MAI with PbI 2 that results in a shift in the corresponding symmetrical and asymmetrical vibrations.
As the prime focus of this article is to investigate air stability of perovskite thin films, we carried out shelf-life testing of our samples kept in (i) the dark and (ii) room light in order to understand the effect of humidity, oxygen and also light exposure.The room temperature was 25-30 1C and the humidity varied between 70 and 80%.The samples were tested unless a complete decay in their optical properties (absorbance and PL) and also decomposition of perovskite crystals is confirmed via XRD.Fig. 2a and b shows FESEM micrographs of the degraded samples after 600 h and 100 h in the dark and room light, respectively.Although the total film thickness (TiO 2 + CH 3 NH 3 PbI 3 ) remained almost the same (900-1000 nm), a distinction between TiO 2 and perovskite layers is not possible anymore and the perovskite crystals seem to be decomposed and diffused into the mesoporous TiO 2 .Surprisingly the degradation pattern of the two samples (dark and light) is notably different.The films kept in the dark, upon degradation, did not show any presence of PbI 2 , which is contrary to the common understanding that CH 3 NH 3 PbI 3 decomposes to PbI 2 in the presence of humidity.This could be verified from the photographs of the aged samples (Fig. 2c, inset, labelled as dark degraded) which still shows a dark brown color associated with perovskite crystals.Although the UV-spectrum of these films showed a considerable drop in relative absorbance, the absorbance spectrum still shows characteristics of perovskite.The presence of perovskite is also confirmed from the emission peak related to perovskite at B770 nm although PL spectra showed B20 times reduction in the relative intensity.On the other hand, the samples kept in room light showed the presence of PbI 2 as a color change is observed from dark brown to yellow (Fig. 2c, inset) and also no corresponding perovskite absorbance or PL peaks are visible.The absorbance features of these films correspond to PbI 2 which also matches with its absorbance characteristics reported in the literature (o350 nm). 59n order to validate the degradation pattern in these films, we carried out their XRD analysis (Fig. 2d) and compared it with a reference sample prepared in a glove box. 18All representative peaks of the fresh film can be indexed to CH 3 NH 3 PbI 3 (labelled as P) and TiO 2 (labelled as T).It is important to note that the XRD spectra of the perovskite films prepared under high humidity conditions show broadened peaks related to a much larger size of perovskite crystal compared to those prepared under inert conditions as in our previous report where a small perovskite crystal size showed sharp intensity peaks. 18Interestingly, even after 4600 h of shelf-life testing, the perovskite films kept in the dark did not show contents of PbI  The inset shows a photograph of the same, and (d) compares the XRD spectra of the three with a reference perovskite film prepared in a glove box.The reference XRD is taken from our previous report. 18hich affirms the presence of PbI 2 as from their change in color (Fig. 2c, inset).Surprisingly, no peaks for perovskite can be found in the XRD spectrum of the degraded sample (Fig. 2d, pink line) which explains why emission and absorbance features of perovskite are not present in its respective spectrum.
Contrary to the widely established understanding that perovskite films are extremely sensitive to humidity, our results suggest that this is not the case.The CH 3 NH 3 PbI 3 films showed little dependence on humidity and major degradation only occurs when the samples are exposed to photon dose.In fact a similar observation was also made by Matsumoto et al. 50that a perovskite layer itself is stable when exposed to humidity; however, the authors noted that the nature of the TiO 2 /perovskite interface plays a dominant role in degradation, as also observed previously. 60,61The small degradation of our samples in the dark (even after 600 h) and rapid degradation in merely B100 h in light suggests that the nature of the TiO 2 /perovskite interface is highly sensitive to light.Perhaps, upon exposure to light, the TiO 2 thin layer, which is known to be reactive with CH 3 NH 3 PbI 3 , 18,48,49 tends to react much rapidly leading to a surface degradation of perovskite.This could also explain the presence of a few unknown peaks in the XRD spectrum (marked by a *), which may be related to intermediate phases formed by TiO 2 /perovskite.Nonetheless, it is noteworthy that the degradation of CH 3 NH 3 PbI 3 films in light took place B12 times faster compared to the samples placed in the dark suggesting that the TiO 2 surface defects -which are known as a humidity site -are triggered upon exposure to light leading to a drastic decomposition of perovskite films.

Polymer incorporating perovskite films and enhanced air stability
Our results suggest that it is rather the photon dose coupled to humidity that induces instability in perovskite films instead of humidity alone.Towards this, the TiO 2 /CH 3 NH 3 PbI 3 interface, which is known to react with perovskite owing to the surface defects in TiO 2 18,48 plays a major role.To overcome this, we synthesised the CH 3 NH 3 PbI 3 -PVP matrix at different concentrations of PVP (5 wt%, 10 wt% & 20 wt%) and investigated variations in their surface properties as a function of time.It is important to note that for each type of device tested in the dark and in light, the aging time varies.This is because the devices were tested until a notable degradation is observed in order to track their air stability.Fig. 3a-l shows three types of surface images of each device (0, 5, 10, and 20 wt% PVP).Each wt% represents the concentration of PVP which is added to the perovskite solution.The pristine CH 3 NH 3 PbI 3 films with 0 wt% (Fig. 3a-c) degraded within B600 h in the dark and o100 h when exposed to light showing decomposition in perovskite crystals.Whereas the fresh CH 3 NH 3 PbI 3 films show crystals with a size of B100-250 nm and perovskite and TiO 2 layers are distinguishable, the sample after shelf-life testing in light and in the dark shows that the perovskite film is diffused in mesoporous TiO 2 .Not only are the two layers not distinguishable anymore after This journal is © the Owner Societies 2016 degradation, but a reduction of B150-250 nm is also observed in the total film thickness (employing CH 3 NH 3 PbI 3 and TiO 2 ) which changed from B900-1100 nm to B750-950 nm.Interestingly, no color change is observed for the CH 3 NH 3 PbI 3 films kept in the dark suggesting (inset Fig. 3b) no decomposition to PbI 2 , whereas when exposed to light the presence of PbI 2 is evidenced from a change in the color of the perovskite film from dark brown to yellow (inset Fig. 3c).Furthermore, the films exposed to light seem to suffer from more intense degradation despite their B7 times shorter shelf-life (B100 h) compared to the film in the dark (B600 h) suggesting photon-induced degradation to be more dominant than that by humidity.Upon incorporation of perovskite into PVP, a two-fold change is noticed in the resultant PeP films: not only had the crystallite size reduced significantly from B200-250 nm to o100 nm but also, a clear distinction between CH 3 NH 3 PbI 3 and TiO 2 is hard to be made.The thickness of the perovskite layer systematically increased from B500-600 nm to up to 41 mm in 20 wt% PVP concentration, which is attributed to the higher solution viscosity of the final polymer-perovskite solution.It is interesting to note that, at higher PVP concentrations (10 wt% and 20 wt%), the perovskite crystals started to agglomerate and the films are very densely packed with a pinhole free texture.Towards their shelf-life testing, the PeP film survived surface degradation even when exposed to dual stress factors; i.e., humidity and light (Fig. 3e, f, h, i, k and l).While the polymer layer seems to retard the humidityinduced instability, it also avoided interfacial degradation of CH 3 NH 3 PbI 3 by making the perovskite/TiO 2 interface less reactive.Also, unlike the pristine CH 3 NH 3 PbI 3 film, there is no notable variation in the surface properties of the perovskite film and their thickness is observed in the dark as well as when exposed to light.For the PeP films exposed to humidity and kept in the dark, the presence of PbI 2 is not observed (a color change to yellow) although the slight color change from dark to light brown is observed (inset of Fig. 3e, h and k) suggesting structural variations in CH 3 NH 3 PbI 3 .However, when exposed to light, the 5, 10 and 20 wt% PeP films showed resistance to degradation, especially in the region where perovskite is coated over the TiO 2 Electron transport layer (ETL) (inset of Fig. 3f, i and l), unlike the pristine CH 3 NH 3 PbI 3 film which is completely converted to PbI 2 , as evidenced from their color change (Fig. 3c).
In order to understand the optical properties of PeP films, we carried out absorbance and photoluminescence (PL) measurements (Fig. 4a and b).The PeP films showed a systematic increase in absorbance throughout the visible spectrum and a steeper absorption onset is observed in PeP films (10 and 20 wt%).We attribute such an extended absorption onset to the light scattering phenomena in the PeP films owing to the formation of polymer-perovskite or perovskite agglomerates.The increase in the absorption coefficient throughout the visible spectrum range can be attributed to the higher film thickness of the PeP films, which increases systematically from a pure CH 3 NH 3 PbI 3 film with increasing PVP concentration.The PL spectra (Fig. 4b) showed a systematic blue shift in the emission peak upon increasing the PVP concentration; the emission peaks observed for pure CH 3 NH 3 PbI 3 (0 wt%), 5, 10, and 20 wt% samples were at 77, 757, 760 and 755 nm.The blue shift observed in the PeP films is because the CH 3 NH 3 PbI 3 is grown in a PVP matrix with physical constraints to the growth of perovskite crystals, which is known to induce a blue shift in the PL emission peak. 62The PeP films show many orders of magnitude higher PL intensity despite nearly the same absorbance indicating the removal of surface traps due to polymer passivation leading to improved emission properties.This is due to the fact that the PL intensity is in fact a measure of the photogenerated charges in semiconductors which depends on the electron/hole pairs generated by incident light. 63,64Furthermore, the higher PL intensity in the PeP films can also be attributed to relaxation of perovskite crystals upon incorporation of the polymer.
The recombination properties of light excited electrons and holes in semiconductors can be explored using PL spectra.incorporating film as the conductivity of PVP is several orders of magnitude lower than that of perovskite.No notable peak shift is observed in PeP upon interfacing with TiO 2 ; however, the 20 wt% PeP film showed two emission peaks (732 and 785 nm), the origin of which is not clear at the moment.
We carried out absorbance and PL measurements of the pure and PVP incorporated perovskite films as a function of time stored in the dark and also in room light to investigate their air-stability (ESI, † Fig. SI-3-6).For the samples kept in the dark, the absorbance of pure CH 3 NH 3 PbI 3 and 5 wt% PVP-CH 3 NH 3 PbI 3 films showed a notable drop while maintaining the absorbance features of perovskite (Fig. SI-3a-d, ESI †).This suggests that although a reduction in relative absorbance is noted, the films have not decomposed to their initial precursors (MAI and PbI 2 ), at least notably.The significant drop in absorbance could be attributed to structural decomposition of the perovskite crystal from 3D to 2D.The absorbance in glass/ TiO 2 /CH 3 NH 3 PbI 3 films dropped down by one third of its initial value in B600 h whereas in glass/TiO 2 /CH 3 NH 3 PbI 3 -5 wt%-PVP films it dropped to half of its initial value in B800 h (Fig. 5a).On the other hand, at higher concentrations of PVP, no notable degradation is observed even after air exposure for B2000 h.Interestingly, upon exposure to light, drastic degradation is observed even in much shorter shelf-lifetime.Whereas the absorbance in pure glass/TiO 2 /CH 3 NH 3 PbI 3 films dropped nearly completely in merely 100 h and the resultant UV spectra suggesting complete degradation, a notable reduction in PeP films is noted only after 300, 600, and 800 h only for 5, 10 and 20 wt% PVP concentrations, respectively (Fig. 5b).It is important to note that the UV spectra of these samples after degradation show the presence of PbI 2 , the extent of which varies inversely with increasing PVP concentration suggesting that the PeP films resisted decomposition of CH 3 NH 3 PbI 3 .This is also visible from the photographs of the devices shown in the insets of Fig. 3a-l.
A similar trend is observed for the emission peaks of the perovskite films in the dark and light (Fig. 5c and d).Whereas a significant drop in the PL emission intensity is observed after 600 h in the dark, degradation occurs 6 times faster when exposed to light.The PeP films, on the other hand, showed a relatively stable performance in the dark although a notable reduction in the PL intensity is noticed when exposed to room light.Also importantly, the PL spectra of the corresponding perovskite films kept in the dark retain their characteristics with little variation (5-10) nm observed in the PL peak intensity; however, the samples exposed to light undergo a drastic variation (Fig. SI-6a-d shows little variation whereas for the sample exposed to light, the Stokes shift cannot be calculated.This is due to the fact that the absorption onset at B790 nm related to perovskite bandedge absorption is not visible anymore and the new onset appears in the range 550-650 nm.The Stokes shift calculated from the new onset for degraded films in light is 3-4 times higher than a corresponding fresh film.Whereas a small Stokes This journal is © the Owner Societies 2016 shift (samples in dark) is related to band-to-band radiative recombination within CH 3 NH 3 PbI 3 , the large Stokes shift (samples in light) can be attributed to surface defects that are created by isolated PbI 2 octahedra.l. 65Although the optical investigations affirm our initial findings from surface analysis (SEM images) and the change in color of the samples that humidity alone is not the major degradation factor and it is rather the nature of the interface (surface defects of TiO 2 ) that is triggered under photon dose, chemical analysis is inevitable to find its origin.
In order to note any variation in the chemical structure we recorded the XRD spectra of the perovksite films kept in the dark and exposed to light (both in high humidity 470%) as a function of time.The corresponding peaks observed in fresh samples can be indexed to TiO 2 and CH 3 NH 3 PbI 3 with perovskite major peaks (110), (220), and (330) are visible at 2y B 14.8, 28.8 and 43.41, respectively, suggesting no structural variation upon incorporation of PVP (Fig. 6a). 66The presence of amorphous PVP that typically shows corresponding peaks at 2y B 221 is not observed in the PeP films which could be attributed to its low concentration.The XRD spectra of the pure and PVP incorporated CH 3 NH 3 PbI 3 films kept in the dark and exposed to light measured after a notable degradation is tracked via their optical measurements are shown in Fig. 6b and c.Although a structural deformation is observed for the pure CH 3 NH 3 PbI 3 film in the dark after B600 h as the peak intensity of (110) at 2y B 14.81 is significantly dropped, no corresponding peaks are observed for PbI 2 or I 2 , which typically show decomposition of perovskite in the presence of humidity.The XRD spectra of PeP films stored in the dark (measured after 800, 1800, 2000 h for 5, 10, and 20 wt%, respectively) show no structural variations and explain why perovskite films in the dark did not change color (Fig. 3b, e, h  and k).This provides the most compelling evidence that humidity alone does not decompose perovskite films, as least notably, contrary to the widespread understanding in the literature.
On the other hand, the perovskite films when exposed to light undergo drastic structural decomposition, particularly, for pure CH 3 NH 3 PbI 3 samples without any polymer matrix (Fig. 6c).The appearance of peaks corresponding to PbI 2 and I 2 at 2y B 12.81 and 2y B 38.81 shows vanishing of the major characteristics peaks of CH 3 NH 3 PbI 3 and evidences decomposition of the perovskite crystals.This structural deformation is more intense for pure CH 3 NH 3 PbI 3 and 5 wt% PeP films that degrade within 100 and 300 h and systematically reduced upon further increase in PVP concentration.The 10 wt% PeP films however showed that PbI 2 major perovskite peaks at 2y B 14.8, 28.8 and 43.41 are also visible even after 600 h.Alternatively, no traces of PbI 2 are found for 20 wt% PeP films and dominant perovskite peaks are also clear although a small peak corresponding to I 2 is noted after 800 h.This explains the color of perovskite films observed in their corresponding photographs (Fig. 3c, f, i, and l).
To further validate the relative change in the crystal structure of perovskite in the dark and light, we recorded the FTIR spectra of the four representative thin films after a notable degradation is tracked using optical characterization techniques.The fresh FTIR and also the corresponding spectra of the devices after degradation in the dark are given in the ESI † (Fig. SI-7 and 8).Fig. 7 shows the FTIR spectra of perovskite films exposed to light.The corresponding peaks related to symmetrical and assymetrical vibrations of N-H stretching, C-H stretching, N-H bending, C-N stretching and C-H bending at 3453, 2707, 1638, 1464, and 1383 cm À1 retained their peak position in the dark showing no structural variations; however, for the samples exposed to light the corresponding peaks related to C-N stretching and C-H bending disappeared for the 0 and 5 wt% PVP incorporated films.Also, a shift in the peak position is noted for C-H stretching from 2707 to 2844 cm À1 and for C-H bend from 1638 to 1652 cm À1 .The disappearance of peaks and the relvant peak shift evidences structural variations of perovskite crystals that could be attributed to the deformation of CH 3 NH 3 PbI 3 .
Fig. 8 shows a track of the lifetime (time taken until a notable degradation in the perovskite films is observed) of the perovskite films as a function of the PVP concentration.The polymer matrix enhanced the air stability of the PeP films by an order of magnitude when exposed to extreme stress factors; i.e., Rel.H. 4 70% and room light (Rel.H. 4 70% and light intensity B 5 mW cm À2 ).The observation that perovskite films do not degrade significantly when exposed to humidity alone and it is rather the nature of the ETL/perovskite interface and also the surface defects of TiO 2 that are triggered in the presence of light (as observed from optical and surface investigations and confirmed from FTIR and XRD) is also hinted in a few previous reports. 35,50,67Various possible reasons could lead to structural decomposition when a dual stress (humidity and light) is applied to perovskite films.Ion migration across the TiO 2 /perovskite interface is one of such possibilities as also suggested in our previous reports. 18,46iven the highly volatile nature of the CH 3 NH 3 I + cation in the CH 3 NH 3 PbI 3 crystal 68 such an interfacial migration will result in the formation of new Ti-CH 3 NH 3 I + complexes which may be the unidentified peaks in the XRD spectra (Fig. 6b and c, marked by a *).Furthermore, as TiO 2 is a photosensitive material, well known for its surface defects, it appears that these surface traps are activated in the presence of light, and act as decomposition centres and films degrade due to a possible photochemical reaction.Upon incorporating CH 3 NH 3 PbI 3 in a PVP matrix such an interfacial degradation could be controlled.
Finally, we also employ PeP films to fabricate a PSC and compare its performance to a reference device made using pure perovskite films (see Table 1).The details of device fabrication are in the ESI.† The unoptimized PSCs resulted in a best PCE of B 4.91% for a reference device whereas the PCE dropped by B27% when a 5 wt% PeP film is employed.The primary objective of device fabrication in this article is to examine whether or not charge injection from the PeP occurs such that the PSCs show J SC and V OC .This test is particularly important given that PVP, embedded in the perovskite crystals, is an insulator.Promisingly, the devices using PeP showed acceptable J SC and V OC values (Table 1); these values are similar to the average J SC and V OC routinely reported in DSSCs and OSCs.Notable drawbacks of the present devices are poor fill factor (FF o 40%) and large hysteresis (B50%) -overcoming them would result in highly efficient and stable PSCs.It is important to note that a high efficiency of B16% for perovskite films embedded in a polymer (PEG) has been recently reported with high stability. 51Our future research will include optimizing the thickness of the PeP films (currently B1-1.3 mm) as well as determining the factors contributing to the lower FF and large hysteresis to improve their photovoltaic performance.

Conclusions
In conclusion, we compared the air stability of pure perovskite (CH 3 NH 3 PbI 3 ) thin films with those employing a polymer (polyvinylpyrrolidone, PVP)-perovskite matrix stored in the dark and in light at a high humidity of 470%.We note that even a prolonged exposure of CH 3 NH 3 PbI 3 films results in minor decomposition of perovskite to its initial constituting precursors when exposed to humidity in the dark and this degradation could be completely overcome in polymer-perovskite (PeP) films.Alternatively, when exposed to light, the films showed B10 times faster degradation in the presence of humidity, the extent of which was inversely proportional to increasing polymer concentration, suggesting photodegradation to be the dominant stress factor.Based on detailed surface, optical and chemical characterization of the pure perovskite films and their PeP counterparts in the dark and light, we conclude that degradation is caused due to native surface defects of TiO 2 , which are activated in light and result in an imbalance of Fig. 7 FTIR spectra of the four perovskite thin films exposed to light at high humidty.The spectra are recorded after the degradation is tracked using their optical properties.charges in the perovskite crystals.Also, the surface traps act as humidity sites resulting in the decomposition of CH 3 NH 3 PbI 3 to CH 3 NH 3 I, PbI 2 and I 2 .Furthermore, we control the reactivity of the TiO 2 /perovskite interface via incorporating a polymer matrix which avoided ion migration across the interface.Our results can support ambient processing of high quality and air-stable perovskite films without the need for an inert environment and can be applied to mass production compatible fabrication of perovskite solar cells.The perovskite solar cells fabricated using the polymer embedded perovskite films confirm that charge injection occurs from this composite and results in an appreciable current (B11 mA cm À2 ) and voltage (0.80 V).However, the fill factor of the devices was o40% and showed a hysteresis of B50% -eliminating these drawbacks would result in stable and efficient perovskite solar cells.

Fig. 1
Fig. 1 Cross-sectional FESEM image of a fresh perovskite film (glass/TiO 2 / CH 3 NH 3 PbI 3 ) showing perovskite crystals of 200-300 nm size, (b) absorbance and emission intensity of the same as a function of photon energy used to calculate the bandgap of CH 3 NH 3 PbI 3 (1.55 eV), and (c) FTIR spectra of CH 3 NH 3 I and CH 3 NH 3 PbI 3 .
2 or I 2 (major peaks at 2y B 12.81 and B38.81, respectively) which are widely believed to be present in decomposed films.Although the major peak of perovskite at 2y B 14.81 nearly vanishes, the XRD spectrum of the aged perovskite film in the dark (blue line) can still be indexed to various perovskite phases at 2y B 20.81, 24.61, 28.21, 28.761, and 35.861.On the other hand, the samples exposed to light showed a notable presence of PbI 2 (2y B 12.81) and I 2 (2y B 38.81) even B50-100 h of room light exposure

Fig. 2
Fig. 2 FESEM images of the CH 3 NH 3 PbI 3 film stored under high humidity conditions (Rel.H. 70-80%) and exposed to (a) the dark for 600 h, and (b) light for merely B100 h showing the degree of decomposition of the perovskite crystal structure.Figure (c) compares absorbance (left) and PL (on right) of the three films (fresh, in the dark and in light at high humidity).The inset shows a photograph of the same, and (d) compares the XRD spectra of the three with a reference perovskite film prepared in a glove box.The reference XRD is taken from our previous report.18

Fig. 3
Fig.3Representative cross sectional FESEM images of all the pure and PVP-incorporated perovskite films (fresh, in the dark and light stored at high humidity) and recorded after a notable degradation is tracked in each film; i.e., (a-c) 0 wt% PVP (600 h in the dark, 100 in light), (d-f) 5 wt% PVP (800 h in the dark, 300 in light), (g-i) 10 wt% PVP (1800 h in the dark, 600 in light), and (j-l) 20 wt% PVP (2000 h in the dark, 800 in light).The inset of each figure shows a photograph of the same thin film.
Fig. SI-2 (ESI †) showed the PL spectra of perovskite films interfaced with TiO 2 to quantify their charge transfer characteristics.Whereas the PL intensity of glass/TiO 2 /CH 3 NH 3 PbI 3 /0 wt% PVP is quenched to B80% of its original value showing effective charge transfer properties, the increasing PVP concentration showed a systematic reduction in PL quenching such that the PL intensity of glass/20 wt% PeP o glass/10 wt% PeP o glass/5 wt% PeP o glass/0 wt%.The reduced charge transfer across the TiO 2 /CH 3 NH 3 PbI 3 interface in PeP films could be related to the inferior electronic transport of the polymer

Fig. 4
Fig. 4 (a) Absorbance and (b) PL spectra of the pure and PVPincorporated perovskite films on a glass substrate (fresh).
, ESI †), the intensity of which is inversely related to PVP concentration.Whereas the pure CH 3 NH 3 PbI 3 films did not show a corresponding PL peak within B100 h of light exposure (Fig. SI-6a, ESI †), the peak disappeared in 5 wt% PeP films only after 300 h.The 10 and 20 wt% PeP films still showed a corresponding emission peak evidencing the existence of CH 3 NH 3 PbI 3 (Fig. SI-6b-d, ESI †).A comparison of the Stokes shifts calculated for the fresh sample (50, 46, 32, and 12 nm) and degraded films in the dark (56, 34, 55, and 13 nm)

Fig. 8
Fig. 8 Lifetime of the perovskite film as a function of PVP concentration calculated from our experiments kept in the dark and light (both under high humidity, Rel.H. 4 70%).

Table 1
Photovoltaic parameters of PSCs made using pure perovskite and PVP-perovskite films