面向3D激光纳米打印的可制造性感知纳米光子器件逆向设计
Fabrication-Aware Inverse Design of Nanophotonic Devices for 3D Laser-Nanoprinting
AI总结：
本研究提出一种逆向设计方法，明确建模3D激光纳米打印的直接激光写入过程，在实验可用设计空间内优化，可预补偿制造特定效应并利用其提升纳米光子器件功能。
AI中文摘要：
3D激光纳米打印技术的进展使我们能够按需制造出具备广泛功能的3D纳米光子器件。通过利用全部三个空间维度，这类纳米光子器件拥有了极为庞大的设计空间。然而，仅靠直觉无法高效探索如此庞大的设计空间，尤其是在设计自由形态纳米光子器件时更是如此。基于密度的拓扑优化为3D纳米光子设计提供了一种天然工具，可高效设计拥有数百万自由度的器件。传统基于密度的拓扑优化依赖启发式措施来考虑制造方法带来的限制，实际上，制造方法很少被纳入设计流程的正向模型中。在本研究中，我们提出了一种逆向设计方法，该方法明确对3D纳米打印所用的直接激光写入过程进行建模。将直接激光写入模型的可微公式纳入考量，使我们能够在实验可用的设计空间内设计3D纳米光子器件，并针对制造特定效应进行预补偿。在实验可用的设计空间内进行优化，确保我们施加给优化的约束由制造方法的参数化方式决定，而非由启发式方法决定，后者可能会对优化问题施加过度或不足的约束。此外，对3D激光纳米打印过程进行明确建模，不仅能考虑邻近效应等制造特定效应，还能让优化过程主动利用这些制造特定效应，以提升器件的功能。
英文摘要：
Advances in 3D laser-nanoprinting enable us to fabricate 3D nanophotonic devices with a wide range of functionalities on demand. By exploiting all three spatial dimensions, an enormous design space becomes available for these nanophotonic devices. However, such an immense design space is impossible to explore efficiently by intuition alone, especially when designing free-form nanophotonic devices. Density- based topology optimization offers a natural tool for 3D nanophotonic design by allowing the efficient design of devices with millions of degrees of freedom. Traditional density-based topology optimization relies on heuristic measures to account for limitations imposed by the fabrication method. Indeed, the fabrication method is rarely considered as part of the forward model in the design pipeline. In this work, we introduce an inverse design method that explicitly models the direct-laser-writing process used in 3D nanoprinting. Incorporating a differentiable formulation of the direct-laser-writing model allows us to design 3D nanophotonic devices within the experimentally available design space and to precompensate for fabrication-specific effects. Optimizing inside the experimentally available design space ensures that the constraints we put on the optimization are given by our parametrization of the fabrication method and not by heuristic methods, which might over- or underconstrain the optimization problem. Furthermore, modeling the 3D laser-nanoprinting process explicitly allows us to not only take fabrication-specific effects, such as the proximity effect, into account but also enables the optimization to actively make use of these fabrication-specific effects to increase the functionality of the device.
