OP: COLLABORATIVE RESEARCH: Integrated Simulation of Non-homogeneous Thin-film Photovoltaic Devices

太阳能电池作为既不使用化石燃料也不产生温室气体的能源是理想的。它们必须被设计成有效地吸收阳光并将其转化为电能。太阳能电池需要足够厚以吸收整个太阳光谱的光。为了减少这种厚度，从而降低制造成本，使用了几层材料：首先，帮助光线穿透太阳能电池，然后将其捕获在内部。这些层中的一些是半导体，其中产生电力，而其他层（例如，周期性波纹金属背层）可能通过捕获其表面附近的光来帮助吸收。然而，太阳能电池产生的电量不仅取决于吸收，还取决于太阳能电池层内电子的传输。如果电子密度在传输过程中下降，从而超过了阳光吸收的任何增益，那么所选择的光管理策略将不会有成效。该项目小组将开发一对集成的计算机模型，利用数值分析的现代技术，同时预测太阳能电池对阳光的吸收和随之而来的电气性能。这些代码将扩展当前的模拟技术，以允许半导体层的特性因地而异，并允许器件的完全三维模型。使用他们的代码，PI将优化设备设计，以获得最佳发电效果。将提供有关薄膜光伏太阳能电池的预测，从而为廉价和可持续的电力生产提供重大进展。这些程序将提供给其他光伏研究人员。薄膜光伏太阳能电池的整体模型将有一个光子子模型和一个电学子模型。在光子模型中，准周期麦克斯韦方程组将使用棱边有限元求解。为了提高灵活性，PI将分析和实施非标准砂浆技术，以照顾准周期性。兼容的电气模型将进行分析和实施的六面体元素的非线性对流和扩散问题的控制漂移和扩散的电荷载流子使用Hybridizable不连续伽辽金（HDG）方法。HDG方案将需要一个新的分析，以了解这个非线性对流扩散问题。首先将探讨线性对流扩散问题的稳定性和收敛性，然后将该方法扩展到漂移扩散系统的隐式-显式时间步进格式。除了完整的3D模型外，PI还将开发2D模型，假设光伏器件在一个横向方向上具有平移不变性。 本研究的第二步是利用新的模拟能力，通过差分进化算法设计最佳的非均匀薄膜光伏太阳能电池。这将优化最大的光伏发电效率。此外，域和系数导数将被表征和实现，以允许计算灵敏度和使用基于梯度的优化。详细的光子和电气建模与双周期背反射器和非均匀的光吸收层将允许一个主要的扩展太阳能电池的设计方法，除了产生最大的光伏发电效率的最佳设计。

Solar cells are desirable as energy sources that neither use fossil fuels nor produce greenhouse gases. They must be designed to efficiently absorb sunlight and convert it to electricity. The solar cell needs to be sufficiently thick to absorb light across the solar spectrum. To reduce this thickness, and so reduce manufacturing cost, several layers of materials are used: first, to help light penetrate the solar cell, and then to trap it inside. Some of these layers are semiconductors in which electricity is generated, while others (for example, a periodically corrugated metallic back layer) may help absorption by trapping light near their surface. However, the amount of electricity generated by a solar cell does not just depend on absorption, but also on the transport of electrons within the layers of the solar cell. If the density of electrons decreases during transport, thereby trumping any gain in sunlight absorption, then a chosen light-management strategy will not be fruitful. The project team will develop an integrated pair of computer models that simultaneously predict the absorption of sunlight and the consequent electrical performance of the solar cell using modern techniques from numerical analysis. The codes will extend current simulation technology to allow for semiconductor layers with properties that vary from place to place and allow fully three-dimensional models of the device. Using their codes, the PIs will optimize device designs for best electricity generation. Definitive predictions will be provided about thin-film photovoltaic solar cells, thereby providing significant progress towards inexpensive and sustainable production of electricity. These codes will be made available to other photovoltaic researchers.The overall model of the thin-film photovoltaic solar cell will have a photonic submodel and an electrical submodel. In the photonic model, the quasi-periodic Maxwell's equations will be solved using edge finite elements. To improve flexibility, the PIs will analyze and implement a non-standard mortaring technique to take care of quasi-periodicity. Compatible electrical models will be analyzed and implemented using the Hybridizable Discontinuous Galerkin (HDG) method on hexahedral elements for the non-linear convection and diffusion problem governing drift and diffusion of electrical charge carriers. The HDG scheme will require a novel analysis to understand this non-linear convection-diffusion problem. Stability and convergence will be explored first for linear convection-diffusion problems, and the methodology will then be extended to an implicit-explicit time-stepping scheme for the drift-diffusion system. Besides a full 3D model, the PIs will develop a 2D model assuming translation invariance of the photovoltaic device in one transverse direction. The second step of the proposed research is to use the new simulation capability to design optimal nonhomogeneous thin-film photovoltaic solar cells via the Differential Evolution Algorithm. This will optimize for maximal photovoltaic electricity-generation efficiency. Additionally, domain and coefficient derivatives will be characterized and implemented to allow the computation of sensitivities and the use of gradient-based optimization. Detailed photonic-and-electrical modeling with doubly periodic back-reflectors and non-homogeneous light-absorbing layers will permit a major expansion of solar-cell design methodologies, besides yielding optimal designs for maximal photovoltaic electricity-generation efficiency.