Silicon solar cells have dominated the photovoltaic industry for the past 25 years; however it is becoming increasingly challenging to increase single-junction silicon cell efficiency any further. Perovskite solar cell technology has drawn attention because of how quickly its efficiency has increased from 3.8% to 22.7%. Perovskites are inexpensive and high quality light absorbers. Multijunction solar cells offer the pathway to increase efficiency beyond the thermodynamic limits imposed by the physics of single-junction solar cells by incorporation of different bandgap semiconductors in tandem while maintaining the current matching attribute. The bandgap of CH₃NH₃PbI₃ perovskite is 1.57 eV, which is near ideal for a two junction tandem design having silicon on the bottom with a bandgap of 1.12 eV.  Perovskite-silicon multi-junction cells now boast efficiencies exceeding perovskite only cells. The latest reported perovskite-silicon efficiency is 23.6%. The three specific research objectives of this project were: (1) Provide transfer-matrix based optimization of optical absorption, reflection and internal quantum efficiency vs. wavelength of each layer in the tandem stack comprising Perovskite/Si heterojunction, Perovskite/GaAs/Si heterojunction and Perovskite/GaAs/GaInAs solar cells. Light trapping mechanism was incorporated to further increase the tandem efficiency. (2) Provide current density vs. voltage simulations for each subcell layer in the tandem and then optimize those parameters to match the current and (3) Provide understanding of carrier transport mechanisms related to drift and diffusion currents through solar cell structure for further optimization. In this research we used two optical modeling techniques: the transfer matrix method and finite element analysis (FEA) of Helmholtz’s equation. The transfer matrix is limited to planer structures and thus cannot account for textured surfaces (unless artificially modified). FEA can do both planar and textured surfaces but takes more computation time. We started with simulating complex refractive indices vs. wavelength for each layers. Complex refractive index is a very important property that determines the propagation of light through and absorption in a particular layer. We implemented evolutionary optimization algorithms and used parameters such as material layer, thickness, composition ratio, doping, carrier lifetime etc. We used transfer matrix method along with evolutionary optimization algorithms to simulate power conversion efficiency and losses vs. wavelength. Different types of losses were considered such as: reflection loss, parasitic loss (light absorbed in antireflective coating layers, window layers and tunnel junctions), thermal loss, recombination loss and spatial relaxation loss. Using the junction characteristics from the single layer simulations, we stacked a perovskite cell on top of a silicon cell. We allowed the optimizer to vary the bandgap of the perovskite as well as the depth of the perovskite and some silicon layers. The resulting efficiency was 26.79%. The iodide to bromine mixture ratio was 22% bromide to 88% iodide. This implied that the bandgap of perovskite was 1.67 eV. We have matched the current for the perovskite/Si heterojunction and found the power conversion efficiency and losses. They are shown in the figures attached.  A triple junction solar cell comprising Perovskite/GaAs/Si was optimized and we varied the bandgap of the perovskite as well as the depth of the perovskite, gallium arsenide, and silicon layers. The resulting efficiency was 28.77%. The iodide to bromine mixture ratio was 43% bromide to 57% iodide. This implied that the bandgap of perovskite was 1.80 eV. The current was matched to achieve highest efficiency for the combination. Power conversion efficiency, losses, current to voltage characteristics before and after current matching are shown in the figures attached. Using the junction characteristics from the single layer simulations, we simulated a stacked Perovskite/GaAs/GaInAs triple junction solar cell. We allowed the optimizer to vary the bandgap of the perovskite and GaInAs as well as the depth of all layers. The resulting efficiency was 33.99%. The iodide to bromine mixture ratio was 38% bromide to 62% iodide. This implied that the bandgap of perovskite was 1.76 eV. On the bottom, the gallium to indium ratio was 64% gallium to 36% indium. The resulting bandgap of GaInAs was 0.95 eV. Power conversion efficiency, losses, current to voltage characteristics before and after current matching are shown in the figures attached. We then performed FEA analysis of non-textured and textured surfaces and reduced the reflection loss and improved the efficiency by using textured surfaces. After adding texture to the perovskite/silicon tandem design in the front middle and back of the cell, the power conversion efficiency increased to 27.36%. We noticed that the reflection loss was reduced to 20.89% from the 24.53% obtained in the non-textured configuration. The plots are shown in the attached figure. We also performed carrier transport analysis and found the generation rate using FEA analysis and used drift-diffusion model to calculate current out of the terminals. We organized an outreach activity and taught middle school students how to build solar houses, simple batteries from lemons and fruits and how to generate electricity from piezoelectric devices.

Last Modified: 09/04/2018
Modified by: Indranil Bhattacharya