CDS&E: Coupled Thermal, Piezoelectric, and Hot Carrier Effects in AlGaN/GaN HEMTs: Multiscale Modeling of Time Evolution of Device Degradation

基于氮化物材料体系的高电子迁移率晶体管（HEMT）具有高击穿电压、高输出功率、高效率、宽带宽、低噪声、耐温抗辐射等独特的性能，在无线通信、国土安全、雷达和卫星系统以及新兴的恶劣环境计算、传感、云网络和电力转换电子产品。然而，与这些设备的长期可靠性相关的问题仍然是一个主要问题。可靠性可以定义为系统在特定条件下运行给定时间而不发生故障的概率。对于半导体器件，器件参数的不可恢复的变化（诸如输出电流的劣化）可以被认为是故障。如最近报告的应力或加速测试所证明的，高温度梯度（热效应）、高电场和内置或诱导的机械应力（逆压电效应）以及高电流密度（热载流子效应）都可能在组成材料中诱导损坏或缺陷，并最终导致器械失效。然而，控制缺陷形成的确切物理机制以及这些缺陷的性质和分布仍然不清楚。这不仅对解释实验结果，而且对预测或推断设备寿命提出了重大挑战，这些任务对于例如远程应用中使用的设备至关重要。众所周知，为了研究实验上难以处理的物理过程，数值模拟变得至关重要。该项目旨在开发一个多尺度和多物理模拟框架，用于模拟AlGaN/GaN HEMT退化的时间演化和物理机制。开发的模拟器将使器件设计提高可靠性。模拟器和相关的教学材料将在nanoHUB.org上部署和免费提供，供更广泛的社区在研究和课堂活动中使用。提出的研究的目的是开发一个多尺度，多物理场模拟框架（HEMT 3-D）的AlGaN/GaN HEMT器件退化机制建模。 需要解决的具体基本问题包括：a）金属扩散、极化和感应电荷密度之间的相关性，B）缺陷的起源、空间和时间分布，以及它们如何影响静电、能带结构、栅极泄漏、载流子去限制和俘获-去俘获以及接触电阻，c）晶格加热和热电子注入阻挡材料之间的相关性，d）应变和逆压电效应及其温度依赖性，以及e）通过工程几何形状、材料组成、通道取向以及通过界面特性的微观调节增强势垒-通道和缓冲层-衬底界面处的热传递来优化器件。为了正确处理纳米结构有源区中的原子对称性以及在不同长度和时间尺度上复杂、非线性、高度随机和动态耦合的底层物理过程，模拟器将采用模块化方法，集成第一原理分子动力学、晶格动力学蒙特卡罗和量子校正的电子-声子输运内核。该模拟器的可移植性和运行效率将通过使用开源科学软件、编译器和库以及结合GPGPU平台的优化模型、算法和扩展来实现。在软件开发工作的每个阶段，将根据文献中的实验数据以及通过与研究实验室、学术界和工业界的实验人员合作，考虑对计算结果进行验证。

High electron mobility transistors (HEMTs) based on nitride material systems feature a unique combination of high breakdown voltage, high output power, high efficiency, wide bandwidth, low noise, and temperature and radiation hardness and have great potential in applications such as wireless communication, homeland security, radar and satellite systems, as well as emerging harsh-environment computing, sensing, cloud-networking and power conversion electronics. However, issues related to long-term reliability of these devices still remain a major concern. Reliability can be defined as the probability of operating a system for a given time under specified conditions without failure. For semiconductor devices, unrecoverable change of a device parameter (such as degradation in the output current) may be considered as a failure. As evidenced in recently reported stress or accelerated tests, high temperature gradients (thermal effect), high electric fields and built-in or induced mechanical stresses (inverse piezoelectric effect), and high current densities (hot carrier effect) may all induce damages or defects in the constituent materials and ultimately lead to device failure. Yet, the exact physical mechanisms governing the defect formation as well as the nature and distribution of these defects are still not clearly understood. This poses a significant challenge to not only interpreting the experimental results but also predicting or extrapolating the device lifetime, tasks that are critical, for example, for devices used in remote applications. It is well acknowledged that to study physical processes that are experimentally intractable, numerical modeling becomes essential. This project sets out to develop a multiscale and multiphysics simulation framework for modeling the time evolution of and the physical mechanisms responsible for AlGaN/GaN HEMT degradation. The developed simulator will enable device design for improved reliability. The simulator and the related instructional materials will be deployed and made freely available on nanoHUB.org for the broader community to use in research and classroom activities. The objective of the proposed research is to develop a multiscale, multiphysics simulation framework (HEMT 3-D) for modeling device degradation mechanisms in AlGaN/GaN HEMTs. Specific fundamental issues to be addressed include: a) correlation between metal diffusion, polarization, and induced charge density, b) origin, spatial and temporal distribution of defects, and how they affect electrostatics, band structure, gate leakage, carrier deconfinement and trapping-detrapping and contact resistances, c) correlation between lattice heating and hot-electron injection into the barrier material, d) strain and inverse-piezoelectric effects and their temperature dependence, and e) device optimization through engineering geometry, material composition, channel orientation, and enhancement of heat transfer at barrier-channel and buffer-substrate interfaces via microscopic tuning of the interface characteristics. To properly treat the atomistic symmetry in the nanostructured active region as well as the underlying physical processes that are complex, nonlinear, highly stochastic and dynamically-coupled at different length and time scales, the simulator will employ a modular approach integrating first-principles molecular dynamics, lattice kinetic Monte Carlo, and quantum-corrected electron-phonon transport kernels. Portability and run-time efficiency of the simulator will be achieved through the use of open-source scientific software, compilers and libraries, as well as incorporating optimized models, algorithms and extensions for GPGPU platforms. Verification of the computational results will be considered at every stage of the software development effort against experimental data available in literature as well as through collaboration with experimentalists in research laboratories, academia and industries.