UniSolar：太阳电池多维工艺、光学和电学仿真的统一框架

马法军^1,2,3, 卢毅轩¹, 徐嘉玉², 张志园², 秦云², 张武², 刘群², 林佳继²,

林文杰³, Bram Hoex³, 周浪¹

¹南昌大学共青城光氢储技术研究院

² 拉普拉斯新能源科技股份有限公司

³School of Photovoltaic and Renewable Energy Engineering, University of New South Wales, Australia

E-mail: 87222473@qq.com

报告摘要

UniSolar 是一个旨在简化多种太阳电池技术多维仿真的统一框架，该框架将典型的光学、电学和工艺仿真流程系统性地归纳总结为十一个通用步骤，并通过精心设计的十一组变量全面定义仿真参数，显著降低了使用 Sentaurus 等商业仿真软件的复杂度。用户无需深入掌握特定仿真软件的操作细节，仅通过编辑纯文本文件中的变量即可高效完成从工艺制备到光电性能分析的复杂仿真任务。

该框架的核心创新在于其高度结构化的变量设计、极简的参数输入方法和简单易用的语法规则单元。每个变量被定义为编程语言中常用的列表数据结构，利用列表的可嵌套性可容纳从简单参数到复杂描述的多类参数信息，既保证了一/二维仿真输入的简洁性，又兼顾了三维仿真的复杂需求；变量都有默认设定值，生成的结构以及相应坐标都可以在其它变量中引用，使结构的变化自动同步到其它变量之中，变量已经输入的内容可以被复用，多种实验条件组合轻松完成；前面两个创新都依赖语法的支持，通过组合一个又一个语法规则单元形成严谨的语法检查系统，UniSolar 确保了变量输入参数的准确性和一致性，有效规避和分析出用户的设置错误。此外，框架支持显式和隐式两种方法生成复杂几何形状，能够准确模拟随机金字塔纹理、三维发射极分布等非理想结构，并集成先进的物理模型（如场效应钝化、陷阱辅助复合等），从而在光学损耗、载流子传输和界面复合等关键物理过程中实现高精度仿真。

在应用层面，UniSolar 已成功用于TOPCon和HJT等高效太阳电池的工艺-光学-电学联合仿真。通过生成具有随机分布的金字塔纹理和三维掺杂 profiles，并结合光 trapping 模型和自由载流子吸收损失计算，仿真结果在量子效率和电流-电压特性方面与实验测量高度吻合。同时，框架支持批量参数调整和快捷拟合功能，极大提高了仿真校准和优化效率。UniSolar 不仅为光伏研究者提供了一个易于使用且功能强大的仿真工具，其标准化、结构化的设计也有利于促进全球范围内的科研协作与数据共享，该框架有望为人工智能提供大规模、低成本的光伏仿真数据集，进一步推动太阳电池技术的创新发展。

关键词：太阳电池；多维仿真； UniSolar

Unified framework for multidimensional process, optical and electrical simulations of solar cells

Fa-Jun Ma^1,2,3, Yixuan Lu¹, Jiayu Xu², Zhiyuan Zhang², Yun Qin², Wu Zhang², Qun Liu², Jiaji Lin²,

Wenjie Lin³, Bram Hoex³, Lang Zhou¹

Abstract

UniSolar is a unified framework designed to streamline multidimensional simulations of various solar cell technologies. It systematically consolidates typical optical, electrical, and process simulation workflows into eleven universal steps and comprehensively defines simulation parameters through eleven meticulously designed variable sets, significantly reducing the complexity associated with using commercial simulation software such as Sentaurus. Users can efficiently complete complex simulation tasks—from process preparation to optoelectronic performance analysis—simply by editing variables in a plain text file, without needing an in-depth understanding of specific simulator operations.

The core innovations of the framework lie in its highly structured variable design, minimalist parameter input method, and user-friendly syntax rule units. Each variable is defined as a list, which is a common data structure in programming. Leveraging the nestable nature of lists, it can accommodate a wide range of parameter information—from simple values to complex descriptions—ensuring both concise input for one-/two-dimensional simulations and meeting the complex demands of three-dimensional simulations. Variables come with default values, and the generated structures and corresponding coordinates can be referenced in other variables, allowing structural changes to automatically synchronize across the system. Previously entered variable content can be reused, enabling easy combination of multiple experimental conditions. The first two innovations rely on syntactic support by combining one syntax rule unit after another, UniSolar forms a rigorous syntax checking system that ensures the accuracy and consistency of variable input parameters while effectively identifying and analyzing user input errors. Additionally, the framework supports both explicit and implicit methods for generating complex geometric shapes, accurately simulating non-ideal structures such as random pyramid textures and three-dimensional emitter distributions. It also integrates advanced physical models (e.g., field-effect passivation, trap-assisted recombination), enabling high-precision simulations of key physical processes including optical loss, carrier transport, and interface recombination.

Practically UniSolar has been successfully used for coupled process-optical-electrical simulations of high-efficiency solar cells such as TOPCon and HJT. By generating randomly distributed pyramid textures and three-dimensional doping profiles, and incorporating light-trapping models and free-carrier absorption loss calculations, the simulation results show strong agreement with experimental results in terms of quantum efficiency and current-voltage characteristics. Meanwhile, the framework supports batch parameter adjustment and quick fitting functions, greatly enhancing simulation calibration and optimization efficiency. UniSolar not only provides photovoltaic researchers with an easy-to-use and powerful simulation tool but also promotes global scientific collaboration and data sharing through its standardized, structured design. The framework holds promise for supplying large-scale, low-cost photovoltaic simulation datasets to artificial intelligence, further advancing innovation in solar cell technology.