Introduction

Hybrid organo-lead halide perovskites have received significant attention for use as absorber materials in solar cells, because of their high absorption coefficient1, tunable band gap2,3, high carrier mobility4, and simple solution-processing synthesis method5,6. Since Kojima et al. reported methylammonium lead iodide (CH3NH3PbI3)-based perovskite solar cells (PSCs) in 20097, various synthesis methods and device architectures have been studied8,9,10,11, and the highest power conversion efficiency (PCE), 24.2%, was recently achieved using a double-layered halide structure12,13. However, the inherent instability of CH3NH3PbI3 remains a major concern in perovskite research14,15,16. For example, Conings et al. reported a significant structural change with a 75% drop in the PCE due to degraded CH3NH3PbI3 in ambient conditions17, and Leijtens et al. also observed a 50% decay of the initial performance of unpassivated PSCs within 5 h18. Several attempts have been made to improve the stability of PSCs by modifying the crystal structure19 or replacing the hole-transporting layer (HTL) with carbon nanotube–polymer composites20. More recently, Cheacharoen et al. reported that ethylene vinyl acetate encapsulation retained 90% of the initial device performance after 200 temperature cycles between −40 °C and 85 °C21. However, despite the use of complex chemical compositions22 or various passivation layers21,23,24, ~10–20% drops in PCE were still observed, and the current research interests have focused more on the thermal stability of perovskite. Therefore, to overcome the remaining concerns, it is necessary to find an alternative encapsulating material.

In this work, we suggest an effective poly(p-chloro-xylylene) (Parylene-C) encapsulation method for CH3NH3PbI3 (MAPbI3) solar cells. The passivation was verified by monitoring the time-dependent optical properties of a MAPbI3 perovskite film, including the UV–Vis absorption and photoluminescence (PL) spectra. As a result, Parylene-C successfully passivated a MAPbI3 perovskite film from reaction with water, owing to its hydrophobic behavior. The photovoltaic performance of Parylene-C-coated PSCs was investigated at various air-exposure times, and the Parylene-C-coated PSCs showed improved stability, maintaining almost the initial PCE values (15.5 ± 0.3%) for 196 h.

Results and Discussion

The device configuration of the Parylene-C-coated MAPbI3 solar cell is shown in Fig. 1a. MAPbI3 perovskite was fabricated using the two-step spin-casting method25, with the TiO2 and Spiro-OMeTAD used for the electron and hole transporting materials, respectively. The 100-nm-thick gold electrodes and Parylene-C polymer were sequentially deposited on the top of the MAPbI3 solar cells, and the active solar cell area was 0.15 cm2 (see Supplementary Information for details). The chemical structure of Parylene-C - phenyl rings, with one chlorine and two methylene groups, is shown at the top of Fig. 1a. We confirmed the chemical structure of Parylene-C via Fourier transform infrared (FTIR) spectroscopy (Fig. S1). Parylene-C showed the absorption peaks at 1045 cm−1 and 1492 cm−1, corresponding to the vibration energy of chlorine and phenyl groups, respectively26. The two C-H stretching modes related to methyl groups were also observed at 2857 cm−1 and 2927 cm−1 27. Figure 1b shows a cross-sectional scanning electron microscopy (SEM) image of the Parylene-C-coated MAPbI3 solar cell, and we confirmed that the thickness of the Parylene-C coating was about 700 nm. The rough surface of the Parylene-C layer was generated by the focused ion beam milling process (Fig. S2). Generally, Parylene-C shows a low surface roughness value of 4.1 ± 0.4 nm because of using a vapor-deposition technique28.

Figure 1
figure 1

Structure of the Parylene-C-coated MAPbI3 PSC. (a) Schematic of the MAPbI3 PSC structure with Parylene-C encapsulation and the chemical structure of Parylene-C. (b) Cross-sectional SEM image of the Parylene-C-coated MAPbI3 PSC.

Full size image

First, we measured the time-dependent absorption of bare and Parylene-C-coated MAPbI3 in order to trace the decomposition reaction in ambient conditions; we also took photographs simultaneously (Fig. S3). The perovskite films were exposed to a room temperature (26.1 °C ± 2 °C) environment with 40–50% relative humidity for 196 h. As shown in Fig. 2a, we observed that the bare MAPbI3 changed from dark brown to yellow, and the absorption peak at 762 nm gradually decreased. We also found that a new absorption peak at 510 nm became dominant, as the entire film of bare MAPbI3 became yellow (dashed circle in Fig. 2a). In contrast, the Parylene-C-coated MAPbI3 film exhibited no color changes, and the absorption also remained unchanged after 196 h in air (Fig. 2b).

Figure 2
figure 2

Parylene-C passivation effect in ambient conditions. Time-series photographs and linear absorption spectra of (a) bare MAPbI3 and (b) Parylene-C-coated MAPbI3. (c) XRD spectra of as-prepared bare MAPbI3 (black), aged bare MAPbI3 (blue), and Parylene-C-coated MAPbI3 (red) after 26 d.

Full size image

To understand these absorption changes, X-ray diffraction (XRD) analysis of the samples exposed to air for 26 d was performed (Fig. 2c). The predominant XRD peaks observed in the as-prepared bare MAPbI3 (black, top) significantly decreased in the aged bare MAPbI3 (blue, middle) and a new peak appeared at 12.64°. According to a previous report, this peak corresponds to the (001) diffraction of PbI2, which is a well-known by-product of the degradation of MAPbI3 perovskite29. We confirmed that the absorption change below 550 nm (2.25 eV) in Fig. 2a was due to the formation of PbI2 because the band gap energy of PbI2 is about 2.30 eV. In contrast, the diffraction patterns of the Parylene-C-coated MAPbI3 film were almost identical to those of the as-prepared MAPbI3, even after 26 d (red, bottom). This result indicates that the Parylene-C passivation layer effectively slowed the decomposition reaction in the MAPbI3 perovskite.

We further confirmed the passivation effect of Parylene-C using time-dependent PL as a function of air-exposure time. In Fig. 3a, the PL intensity of the bare MAPbI3 decreased dramatically over time, and the PL peak position was slightly blue-shifted (black dashed line). According to a previous report, the blue-shifted PL peak of aged MAPbI3 was related to PbI2 formation17, and we observed an increased PL intensity at 510 nm from the aged MAPbI3 (Fig. S4), which is consistent with the band gap of PbI2. In contrast, Parylene-C-coated MAPbI3 maintained its initial PL intensity and peak position for 196 h (Fig. 3b), showing the effective passivation effect of Parylene-C on MAPbI3.

Figure 3
figure 3

Optical properties of MAPbI3 with and without Parylene-C deposition. Time-dependent PL spectra of (a) bare and (b) Parylene-C-coated MAPbI3. TRPL decay profiles of (c) bare and (d) Parylene-C-coated MAPbI3 with the instrument response function (dashed gray line). (e) Integrated PL intensity as a function of air-exposure time for bare (black) and Parylene-C-coated MAPbI3 (red) (inset: photographs of each sample after 196 h).

Full size image

Time-resolved photoluminescence (TRPL) measurements were conducted simultaneously on both bare and Parylene-C-coated MAPbI3 in Fig. 3c,d. The integrated PL intensity of the bare MAPbI3 was zero after 120 h; thus, we compared the TRPL spectra taken for 120 h (excluding data taken after 196 h). As shown in Fig. 3c, the PL decay curves of the bare MAPbI3 decreased gradually with increasing exposure time, whereas those of the Parylene-C-coated MAPbI3 remained unchanged for 120 h (Fig. 3d). We calculated the average lifetime from two decay components fitted to a bi-exponential function and plotted the changes as a function of air-exposure time (Fig. S5). The carrier lifetime in the bare MAPbI3 decreased remarkably from the initial value of 8.67 ns to 0.99 ns because of the decomposition reaction. In contrast, Parylene-C-coated MAPbI3 initially exhibited a longer carrier lifetime than the bare film, i.e., 9.19 ns, maintained for 120 h.

Finally, we investigated the photovoltaic performance of Parylene-C-coated MAPbI3 solar cells in terms of the effect of Parylene-C encapsulation. The solar cell performance was measured under the AM 1.5 G illumination with a power density of 100 mW cm−2 (see Supplementary Information for details). In Fig. S6, the comparison of current density–voltage (J–V) curves before and after Parylene-C deposition revealed that the MAPbI3 PSC was not damaged during the vapor-deposition process. Figure 4 shows time-dependent photovoltaic characteristics for the bare, polymethyl methacrylate (PMMA)-coated, and Parylene-C-coated MAPbI3 PSCs. We selected PMMA for comparison, as it is the most commonly used polymer for PSC encapsulation. Interestingly, the PMMA-coated PSCs showed a 30% drop from their initial efficiency (Fig. 4b) because chlorobenzene, a pre-dissolution of PMMA and Spiro-OMeTAD, can damage Spiro-OMeTAD during PMMA coating30. This result indicates that the use of PMMA limits the selection of hole-transport materials in solar cells, although it is among the most widely used of various passivation polymers. On the contrary, Parylene-C can be used in combination with various hole-transport materials without affecting the initial device performance. In terms of device stability, the solar cell performance of the bare and PMMA-coated PSCs decreased dramatically after exposure to ambient conditions, and these cells were completely degraded within 196 h (Fig. 4a,b). In contrast, the PCE of the Parylene-C-coated PSC exhibited no notable change over a period of 196 h (Fig. 4c).

Figure 4
figure 4

Time-dependent photovoltaic characteristics depending on PSC encapsulation. Time-dependent J-V curves of the (a) bare, (b) PMMA-coated, and (c) Parylene-C-coated MAPbI3 PSCs for 196 h (inset: images of a water drop on PMMA and Parylene-C showing contact angles of 72° and 121°, respectively). Comparison of (d) Voc, (e) Jsc, and (f) PCE of MAPbI3 solar cells with different encapsulating layers before (gray) and after (red) exposure for 196 h.

Full size image

To explain how Parylene-C can efficiently retain the initial performance of MAPbI3 PSC, we further performed FTIR spectroscopy. As shown in Fig. S7, two N–H stretching modes around 3150 cm−1 were remarkably suppressed for aged MAPbI3, and a new absorption peak appeared at 3466 cm−1, corresponding to the vibration energy of hydroxyl groups (−OH). This finding agrees well with a previous report31 on hydrated compounds in aged MAPbI3 resulting from adsorption of water vapor. Note that humid conditions are the main cause of degradation in MAPbI3 perovskite, and therefore, to improve stability of the PSC, it is necessary to prevent water penetration. Fortunately, Parylene-C has been known to have low water permeability of 0.14 cm3 ·mil/(100 in2 24 hr ·atm) at 23 °C32, which correlated well with our contact angle data (the inset of Fig. 4c). The surface of PMMA showed a contact angle of 72°, which were in the hydrophilic range (<90°)33. In contrast, Parylene-C showed hydrophobic behavior, with a considerably higher contact angle of 121°.

We tested the water permeability of these encapsulating materials (Video S1). We prepared Spiro-OMeTAD- and PMMA-coated MAPbI3 as well as bare and Parylene-C-coated MAPbI3. The bare and Spiro-OMeTAD-coated MAPbI3 perovskite films turned yellow on exposure to water drops, revealing that the hole transporting material could not protect against reaction with water. Although the PMMA-coated MAPbI3 initially remained unchanged, the decomposition reaction occurred within 6 min. Yoo et al. reported that a part of PMMA molecular chain is hydrophilic; thus, water molecules can hydrolyze the ester groups in PMMA and break down the PMMA structure30. This is consistent with our results, indicating that PMMA is unable to protect perovskite from water damage. Unlike the other tested materials, Parylene-C effectively passivated the perovskite film against water droplet exposure for 30 min. These results indicate that Parylene-C was able to slow down the degradation of PSC because it could effectively passivate MAPbI3 PSC from reaction with water.

For further analysis, we compared the photovoltaic parameters of the as-prepared (0 h) and aged (after 196 h) PSCs with different encapsulating layers. In Fig. 4e,f, the bare and PMMA-coated PSCs showed remarkable decreases in Jsc and PCE values after 196 h. Interestingly, the Voc values hardly changed after 196 h for both the bare and PMMA-coated solar cells (Fig. 4d). Conings et al. reported similar small decreases of Voc, from aged MAPbI3 PSCs, and revealed that Jsc was influenced more by degradation because perovskite degradation increased resistance and recombination near the interface between the perovskite and carrier transporting layers17. On the basis, we considered that the huge decrease of Jsc in our data was also attributable to perovskite layer degradation. Lastly, the Parylene-C-coated PSC showed no noticeable change to any of the parameters; therefore, we confirmed that Parylene-C could effectively encapsulate the MAPbI3 solar cells for 196 h.

Conclusions

We demonstrated successful encapsulation of MAPbI3 solar cells by Parylene-C deposition. By structural and optical analyses, we systemically investigated the origin of the decomposition reaction in MAPbI3 and confirmed that Parylene-C can efficiently slow down the decomposition reaction in the MAPbI3 films. In particular, Parylene-C can efficiently isolate MAPbI3 perovskite from reaction with water, owing to its hydrophobic character, and as a result, the Parylene-C-coated MAPbI3 solar cells maintained almost the initial PCE values (15.5 ± 0.3%) for 196 h. On the basis of this work, we believe that the (p-xylylene) type polymers have shown the potential to improve the lifetime of organo-lead halide perovskite in future photovoltaic applications.