Abstract
This thesis is composed of two parts: the first part deals with the performance optimization of perovskite (PVK) solar cells exhibiting p-i-n architecture, and second part reports an electron spectroscopy investigation of the so-called light soaking effect in solar cells. The p-i-n device was chosen because it possesses some advantages that include easier fabrication, lower cost and wider application, as compared with n-i-p device. Light soaking effect is related to an interesting observation that the solar cell parameters like Voc and FF increase with a prolonged exposure to AM 1.5 sunlight, leading to an overall enhancement of power conversion efficiency. In the case of PVK solar cell, the present understanding tends to believe that the migration of halide ions under the solar illumination, a common occurrence, is primarily responsible for the light soaking effect. A previous study by Li et al,1 has presented a clear evidence showing iodide ions and positively-charged iodine vacancies, assisted by the applied electric field, will migrate toward the respective electrode and accumulate in the region of PVK that borders either hole transport layer (HTL) or electron transport layer (ETL). As a result, the band bendings at two interfaces of PVK/ETL and PVK/HTM take place, resulting in a change of Voc. The change in chemical makeup and the band bending at the interface can be conveniently studied by XPS and UPS, respectively, which forms the basis of the second part of this thesis. The architecture of our p-i-n device is ITO/PEDOT:PSS/PVK/C60/Ag. The thicknesses of both PEDOT:PSS (HTL) and C60 (ETL) layers are to be optimized first. Frist, the thickness of PEDOT:PSS film is changed by varying spin-coating condition. The optimal PEDOT:PSS film that supports the fabrication of “best” solar cell was formed at a spin-coating speed of 3000 rpm for 60 s. Second, the properties of peroskites based on a different combination of solvent/antisolvent pairs were evaluated by using SEM, AFM, XRD, UV-Vis, and UPS with the derived information covering surface morphology, crystallinity, and electronic property. A recipe of using DMF as a solvent and toluene as an antisolvent results in smoother and more crystalline PVK films. Third, the optimal thickness of vacuum-deposited C60 film (an electron transport material) was concluded to be 20 nm by examining the thickness dependence of device performance. Putting all things together, we produced p-i-n structural devices exhibiting a power conversion efficiency of 9.0 %. To investigate the light soaking effect with XPS and UPS while the device is in operation, a special solar cell exhibiting two different cross section profiles but electrically connected was fabricated. The large thickness region was fabricated with the thickness of the relevant layers conformed to the real solar cell requirement. The small thickness region is topped with a very thin Ag film (< 1nm) as an electrode layer and, underneath this layer, either a C60 ETL or a CuPc HTL of the same small thickness (< 1 nm) is formed on top of 300 nm thick PVK. The reason for making such a small thickness region is that the thickness of transport layer for real devices is much larger than the probing depth of XPS and UPS that are essentially surface sensitive. Without thinning down both electrode and ETL (or HTL), the PVK/ETL and PVK/HTL interfaces cannot be directly probed. UPS data show that as the light soaking is in progress, valence band maximum (VBM) of PVK/ C60 interface shifts to lower binding energy (b. e.) while the VBM of PVK/CuPc shifts to higher b. e.. The direction of b. e. shift is in accord with the prediction based on the accumulation of iodide and iodine vacancy at PVK/CuPc and PVK/C60 interfaces, respectively. The accumulation of negatively-charged iodide ions generates a downward band bending at interface. Whereas an upward band bending is observed for PVK/C60 interface. Elemental composition analysis revealed by XPS data shows the formation of a new I 3d component at both interfaces. This new component is believed to be a complex formed between C60 (or CuPC) and iodide. This assignment is supported by the result published by Kobayashi et al2, in which iodide ions adsorb onto C60 with a modest adsorption energy of 49.0 kJ mol-1. Iodide anions not only migrate to the PVK/HTL interface under the influence of build-in potential, but are also capable of adsorbing onto C60 and CuPc.