Fig. 1: Schematic Diagram Representing Average Reflection and Parasitic Absorption Losses in Perovskite/silicon Tandem Cells. the Average Reflection Was Calculated by Fresnel's Law (without Accounting for Multiple Reflections Here). Optical Approaches to Improving Perovskite/si Tandem Cells

Recently, metal-halide perovskites have demonstrated an extraordinarily rapid advance in single junction cell efficiency to over 20%, while still offering potentially low costs. Since the bandgap is larger than the ideal single-junction value, perovskite-based tandem cells can theoretically offer even higher efficiencies. Instead, however, the record tandem cell performance in experiments to date has come in slightly below that of record single junctions, although slightly higher than the same single junctions. In this work, we consider both how this disconnect can be explained quantitatively, and then devise experimentally feasible, variance-aware approaches to address them. The first stage of our approach is based on reconfiguring dielectric front coatings to help reduce net reflected power and balance junction currents by reshaping the reflection peaks. This method could be applied to post-fabrication stage of perovskite/c-Si tandem cells, and also applicable to cell and module level structures. In the second stage of our approach, we can almost entirely eliminate Fresnel reflection by applying a conformal periodic light trapping structure. In the best case, a short circuit current (Jsc) of 18.0 mA/cm 2 was achieved, after accounting for 4.8 mA/cm of parasitic loss and 1.6 mA/cm reflection loss. Further improvements may require a change in the baseline materials used in perovskite cells. INTRODUCTION Metal-halide perovskites have gained a great deal of attention for their extraordinarily rapid increase in single junction efficiencies, rising from below 1% to now exceeding 20 % [1,2]. Furthermore, perovskites are solution-processable materials that appear resilient in the presence of defects [1,2]; thus, they have a potential to serve as ultra-low cost solar cells. Although longterm stability and reliability is a major potential challenge, recent work has suggested these problems could be addressed successfully [3]. Furthermore, metal-halide perovskites have a large and tunable bandgap, which raises the prospects of even higher efficiencies as a top cell in a tandem solar cell. Recently, perovskite-crystalline silicon tandem cells were proposed, with theoretical efficiencies expected to exceed 30 % [4–6]. However, measured efficiencies more typically fall between 13% and 18 %, which is below that [5,7–9]. Reaching this theoretical limits is still challenging for a variety of reasons. First, the hole transport layers (HTL) for perovskites are not transparent enough for visible and infrared wavelengths [10]. For example, Spiro-OMeTAD shows excellent hole-collecting efficiency, but suffers nearly 10% parasitic losses at the most commonly employed thicknesses of around several hundred nanometers. In single junction perovskites, Spiro-OMeTAD parasitic loss is less important, because it is usually deposited at the back side. In a tandem configuration, it may absorb many more photons when it is placed at the intermediate layer between top and bottom junctions, or top window layer. Recently, PEDOT:PSS [11,12] was suggested as a window layer of perovskites, but its charge collection efficiency was not comparable to Spiro-OMeTAD. The other challenge in this tandem configuration is mismatched bandgaps. Bottom junctions made of crystalline silicon require top junctions with bandgaps of 1.7-1.8 eV for highest performance. In other words, 1.55 eV metalhalide perovskites may suffer slight open-circuit voltage loss, as well as short-circuit current mismatch. Thus, many optical studies have been performed to predict the optimum perovskite thickness for matched junction current [13–15]. However, there has been no clear agreement among these studies. It is natural, because perovskites are typically unstable as photovoltaic materials [16], which cause significant performance variations both between and within samples made through the same manufacturing process [17]. Finally, current perovskite/c-Si tandem designs may suffer non-trivial reflection losses to large refractive index contrast among the layers. The average reflectivity among the layers are shown in Fig. 1. Even though perovskite/cSi tandem cells may suffer substantial reflection losses, conventional light trapping may not help, as it may increase the HTL parasitic loss simultaneously. In this study, we propose front coatings which potentially can solve both problems, shortcircuit mismatch and non-trivial amount of total reflection occurring in the experimental cells. Next, light trapping approach will be applied for a further performance enhancement with considering HTL parasitic loss.