Modeling and Design Optimization of High Frequency Micro and Nano Structures Exploiting Adjoint Sensitivity Analysis

在过去的几年里出现了许多高频技术，它们提供了有趣的功能。例如，超材料引入了具有自然界中不存在的特性的材料。它们被用来制造在不同频率范围内工作的微型天线、传感器和滤波器。等离子体器件利用金属-电介质界面来引导低于衍射极限的光。这允许实现尺寸小得多的光子器件，从而有望实现更多的光子处理单元的集成。用于高频能量采集和视距通信的纳米天线被提出。由于最近太赫兹源和探测器的出现，太赫兹(THz)的应用也出现了。所有这些技术使用的材料的特性可能依赖于源值(非线性)、依赖于频率(色散)或依赖于激励方向(各向异性)。这些特性的不同组合可能存在于这样的高频结构中。这些复杂的属性使得对这些设备的准确建模非常耗时和内存。这些结构的优化通常需要估计目标响应相对于所有参数的灵敏度。估计这些灵敏度的经典方法需要对这些时间密集型结构进行重复模拟，这可能是令人望而却步的。

A number of high frequency technologies emerged over the past few years that offer interesting functionalities. Metamaterials, for example, introduced materials with properties that do not exist in nature. They are used in building miniaturized antennas, sensors, and filters that operate in different frequency ranges. Plasmonic devices utilize metal-dielectric interfaces to guide light below the diffraction limit. This allows the implementation of photonic devices with much smaller dimensions thus promising more integration of photonic processing units. Nano antennas (nantennas) were presented for high frequency energy harvesting and line-of-sight communication. Terahertz (THz) applications also emerged thanks to the recent availability of THz sources and detectors. All these technologies utilize materials whose properties may be dependent on source values (nonlinear), dependent on frequency (dispersive), or dependent on the direction of excitation (anisotropic). Different combinations of these properties may exist in such high frequency structures. These complex properties make the accurate modeling of these devices time- and memory-intensive. Optimization of these structures often requires estimating sensitivities of the target response with respect to all parameters. The classical approach for estimating these sensitivities requires repeated simulations of these time-intensive structures which may be prohibitive.