1. Introduction

This study focuses on the 2D numerical modeling of perovskite-silicon heterojunction (PK-Si) two-terminal (2T) tandem solar cells (see Figure (top)), aiming to enhance the understanding and reliability of characterisation measurements. The approach involves developing a technology computer-assisted design (TCAD) comprehensive model to evaluate the impact of diverse physical phenomena, using Silvaco ©Victory Device. We use the tool to simulate photogeneration and to solve drift-diffusion equations, while the electrochemistry module handles mobile ion transport. The silicon heterojunction (SHJ) model is inspired by previous work on defectivity physics.[1] Among other critical interfaces, modeling the type II heterojunction between the hole transport layer of the top-cell and the n-doped silicon (Si) through transparent conductive oxide is challenging.

We propose an efficient solution and present one of the first 2D large-scale detailed model of PK-Si cells, while former ones were constrained to unidimensional, simplified simulations.[2,3]

2. Novelties and preliminary results

Building the perovskite (PK) top-cell involves numerous, hardly experimentally accessible parameters (e.g., carrier mobilities, ion densities and diffusivities, Shockley-Reed-Hall (SRH) recombination rates, ...): this, and the lack of standard materials, hinder an easy setting of the simulation. Our solution to calibrate the present model is to use time-resolved photoluminescence (TRPL) measurements on the full cell: these measurements are directly linked to the material properties and interfaces of the active PK region. We represent such a measurement result in the Figure (middle) in blue, along with a simulation result after machine learning (ML) calibration (red). By performing numerous 2D-TCAD simulations with varying parameters, we could train a ML regression model to predict the distance between the simulated and experimental TRPL curves. This allowed to converge toward a set of parameters yielding the simulation of the Figure. The parameter values resulting from the fitting procedure are shown in the middle inset.

As stated above, an important innovation of the present work is the 2D engineering of the TCAD model, including an electrode (see Figure, bottom). This opens the way to investigate the impact of inhomogeneities in real-world systems. Indeed, in the process of cell fabrication, defects may occur (device handling, layer growth, ...). Their impact on cell efficiency might be dramatic since locally the junction separating holes and electrons might not be present anymore, creating a ”shunt” in the device. Also, defects in the bulk of the PK absorber might act as deleterious recombination centers. The present demonstration features a defect in the PK with a hundredfold lower SRH lifetime, simulating for example a highly defective PK grain. Its width was swept to yield the bottom inset in the Figure. Interestingly, Voc and Jsc seem not to be affected by these defects, contrarily to the FF that dramatically drops when the width increases from 1 to 300 μm. On the Figure (bottom), the extracted SRH recombination rate proves that losses occur at this defect, rather close to the front surface.

3. Perspectives

As a conclusion, the present 2D-TCAD-based simulation framework helps us improve our understanding of the realistic device under operation. Future work will focus on extending the TCAD model to include additional defects. Subsequently, a sensitivity analysis of the different types of defects will be possible.