Perovskite solar cells (PSCs) have recently demonstrated the power conversion efficiencies (PCEs) larger than most organic solar cells and comparable to those of commercial silicon solar cells. In this work, we have thoroughly studied the performance of mixed cation PSCs fabricated inside the N₂-filled glove box using anti-solvent spin-coating method. In these experiments, the perovskite solution consists of PbI₂, PbBr₂, MABr, FAI, and CsI with a final formula of Cs_0.07MA_0.1581FA_0.7719Pb₁I_2.49Br_0.51. We fabricated PSCs with efficiencies ~17% using the glove box.

We investigated the effect of DI water in the cleaning process on the performance of device. After sonication of samples with DI water, ethanol, and IPA, rinsing with DI water produces some black spots on the surface of FTO glasses which are visible under the microscope. These spots cannot be cleaned by O₂ plasma etching for even 15 min. Thus, the c-TiO₂ layers deposited by spray pyrolysis have many defects remained on the surface of FTO substrates, which decrease the fill factor, short circuit current and the overall performance of device.

We have also investigated the effect of adding hydrochloric acid 37% (HCl) acid to the c-TiO₂ solution in spray pyrolysis method. We found that the devices with HCl additive in their c-TiO₂ solution show lower open circuit voltage compared to the control devices. The normal V_oc is 1.05-1.10 V for the control cells, however it decreases to 0.7-0.85 V for the samples contain HCl.

We measured the resistance of c-TiO₂ layer by using 4-probe measurement system. The high resistance of c-TiO₂ layer means a low number of defects and pinholes in it. When we do a comparison between samples with different resistances of c-TiO₂ layers, we found that the fill factor is higher in samples with the high-resistance c-TiO₂ layer. The resistance of the ETL layer was easily controlled by the spray time on considering that all other parameters such as solution volume, density, annealing time, distance, and temperature were fixed. Generally, samples with longer spraying time showed higher resistivity compared to the samples with shorter spray time.

Dryness of glove box and spin-coater environment plays an important role in getting a high quality perovskite layer. All researchers are using the two-step spin-coating method with 1000 rpm for 10 s and 4000 rpm for 20 s for deposition of perovskite layer, however it doesn’t work in our fabrication setup. Our spin-coater doesn’t have a cap, so the samples will get dry quickly during the first spin step. Thus, the 10 s spinning at low speed doesn’t provide a shiny, black, and smooth perovskite layer. In this case, we use 1000 rpm for 2-3 s and 4000 rpm for 20 s. Following this strategy, we obtained shiny, black, and thick enough perovskite layers with very good perovskite peak intensity according to the XRD results.

We used the thermal evaporation for deposition of the gold layer. The surface of gold layer under microscope has lots of cracks in the size of 300-500 nm. These cracks may deteriorate the performance of the final cells in many ways. First, the gold layer may not collect the holes effectively from the HTM layer due to its poor adhesion to the HTM layer. Secondly, the cracks inside the gold layer reduce the active area of the device and thus underestimate the real performance. In addition, cracks inside the gold layer can also provide a pathway for the moisture to find a way to contact the HTM layer and lower the stability performance of the device.

Using SEM cross-sectional imaging of our samples, we found that there are always some vapor particles and solvents inside the glove box that distribute on the surface of our samples before we finish the spin-coating of c-TiO₂, mp-TiO₂, and perovskite layer. Two tests are performed to deposit c-TiO₂ layer on the clean FTO substrates. Series A were deposited right after cleaning the glove box, and series B were deposited after a month of using the glove box. We couldn’t find any defects between FTO and c-TiO₂ in the SEM cross-sectional imaging of the A samples, but there are many defects and black spots in SEM images at the interface of FTO and c-TiO₂ for the series B. These defects prevent the strong adhesion between FTO and c-TiO₂ layer, so the overall performance of the device deteriorated.