Control of strain in perovskite crystals has been considered as an effective strategy to ensure the phase stability of perovskite films where a compressive strain is particularly preferred over a tensile strain due to a lowered Gibbs free energy by the unit cell contraction effect. Here we adapt the strategy of strain control into perovskite solar cells in which the compressive strain is applied by utilizing a thermal expansion difference between the perovskite film and an adjacent layer. Poly­(4-butylphenyldiphenylamine), with a higher thermal expansion coefficient compared to that of perovskite, is employed as a substrate for perovskite crystal growth at 100 °C, followed by cooling to room temperature. The applied compressive strain at the interface, as a result of a greater contraction of the polymer compared to the perovskite film, is confirmed by grazing incidence X-ray diffraction showing a red peak shift with increasing secondary angle. The compressive strain-induced perovskite film shows relatively constant absorbance spectra as a function of time. In the meantime, the absorbance spectra of a film without strain control exhibit a gradual decay with developing an Urbach tail. Importantly, the effect of strain engineering is remarkably prominent in the long-term photovoltaic performance. The photocurrent drops by 41% over 911 h without controlling strain, which is significantly improved by employing compressive strain, showing only a 6% drop in photocurrent from a shelf-stability test without encapsulation. It is also noted that an S-shaped kink appears in the current–voltage curves since 579-h-long storage for the device without strain control, leading to unreliable and overestimated fill factor and conversion efficiency. On the other hand, a 16% increase in fill factor with a stable performance is derived over 911 h from the compressive strain-induced device.

Novel TiO2 nanorods (NRs) with various lengths of 70–200 nm and uniform widths of 46–48 nm are selectively synthesized by a solvothermal reaction under a basic environment. The length of TiO2 NRs is reproducibly tuned by varying the concentration of tetramethylammonium hydroxide (TMAH), while the NRs in the pure anatase phase are grown in the [001] direction, caused by the preferential binding affinity of TMAH to the TiO2 (101) facet. TiO2 NRs of various lengths are then applied to form the electron transporting layer (ETL) of mesoscopic perovskite solar cells (PSCs). We found that PSC devices with NRs exhibit superior photovoltaic (PV) performance to those with conventional 46 nm-sized TiO2 nanoparticles (NP46). Particularly, the PSC with TiO2 NRs of 110 nm length (NR110) exhibits the optimum PV conversion efficiency (PCE): the average PCE is 22.64% with a V OC of 1.137 V, a J SC of 24.60 mA·cm–2, and a FF of 80.96%, while the champion PCE is 23.18%. In addition, the PSC with NR110 (PSC-NR110) reveals significantly improved long-term stability in air with a relative humidity of 40–50%. In 1000 h, its PCE is reduced by only 9% whereas that of PSC with NP46 decreases by 25%. The PSC properties analyzed by impedance spectroscopy and J–V curve measurements under dark conditions and at various light intensities provide evidence that PSC-NR110 has fewer defects and shows significantly reduced charge recombination. We discuss the advantages of NR structures in preparing the ETL of PSC devices and also explain why the charge recombination is suppressed.