Atomic-Scale Modeling of Defect-Mediated Device Degradation

大多数电子器件都是用硅制成的，但硅不适合军事上许多应用所需的高温和高功率应用（例如，其他半导体，特别是氮化镓（GaN）和碳化硅（SiC）更合适，但是使用这些材料制造的器件在工作条件下会退化。原子级缺陷，如杂质或丢失的原子是负责的。这种缺陷通常通过与缺陷结合的氢原子而变得良性。在操作条件下，“热电子”可以将氢从这些缺陷上踢下来。在识别相关缺陷方面已经取得了很大进展，但设备设计工程师需要能够预测设备寿命的建模软件。主要研究者在器械可靠性和降解领域拥有丰富的经验。Pantelides和Zhang最近开发了一种计算热电子激活缺陷速率的理论，而Pantelides和Schrimpf一直致力于通过电学测量和量子力学计算识别缺陷，开发简单的工程级模型。本建议的目的是充分实施新开发的理论热电子缺陷激活率，进行相关的计算，并结合联合收割机，它们与现有的和新的电气数据真实的设备，以开发更复杂的工程级模型的热电子功率器件退化。拟议的研究也与发光二极管（LED）和太阳能电池的退化有关。工程级模型将提供给工程界和软件开发人员。学生和博士后将在物理和电气工程之间的桥梁领域进行培训。每年为高中教师举办一次讲习班，使他们了解新的事态发展及其意义。我们还将参加NSF资助的田纳西州路易斯斯托克斯少数民族参与联盟（TLSAMP），每年夏天资助一名少数民族大学生参加我们的研究计划。基于宽禁带材料（如GaN）的高电子迁移率晶体管（HEMT）在高功率、高温应用方面具有很大的前景，但目前其使用受到可靠性问题的限制。目前，器件可靠性建模主要是通过现象学模型来完成的，该模型主要依赖于器件故障和器件参数之间的相关性。基于物理现象的建模需要了解基本的物理过程。在铅PI和合作者最近的工作中，无参数量子力学计算与电应力数据相结合，以证明III-V族高电子迁移率晶体管（HEMT）的热电子退化是由特定缺陷释放的H或特定杂质的重新配置引起的。开发了一个简单的工程级模型，以拟合应力数据并预测操作条件下的长期降解。该模型依赖于真实的设备中玻尔兹曼方程的复杂的蒙特-卡罗解来获得空间和能量中的电子密度，但对于氢释放的捕获截面使用了粗略的阶跃函数。主要研究者在这些领域有着丰富的经验。Pantelides和Zhang最近开发了一种理论，用于计算缺陷对热电子的非弹性散射截面，而Pantelides和Schrimpf则致力于通过电学测量和量子力学计算来识别缺陷，开发简单的工程级模型。该提案的目的是充分实施新开发的横截面理论，并进行相关计算，可用于模拟器件退化的现有和新的实验数据。目标是开发准确的，有效的，工程级的模型来描述和预测短期和长期的热电子引起的器件退化。这些结果也将有助于理解和模拟缺陷在太阳能电池和发光二极管性能和退化中的作用。

Most electronic devices are made using silicon, but silicon is not suitable for high-temperature and high-power applications that are needed for many applications in the military (e.g., all electric ships), the electrical distribution grid, etc. Other semiconductors, notably gallium nitride (GaN) and silicon carbide (SiC) are more suitable, but devices fabricated using materials degrade under operating conditions. Atomic-scale defects such as impurities or missing atoms are responsible. Such defects are often rendered benign by hydrogen atoms that are bound to the defects. Under operating conditions, "hot electrons" can kick the hydrogens off these defects. Much progress has been achieved in identifying the pertinent defects, but device-design engineers need modeling software that can predict the lifetime of devices. The Principal Investigators have extensive experience in the device reliability and degradation area. Pantelides and Zhang recently developed a theory for the calculation of the rates of defect activation by hot electrons, while Pantelides and Schrimpf have been working on the identification of defects through electrical measurements and quantum mechanical calculations, developing simple engineering-level models. The objective of this proposal is to fully implement the newly-developed theory for the rates of hot-electron defect activation, to perform pertinent calculations, and combine them with available and new electrical data on real devices in order to develop more sophisticated engineering-level models for hot-electron degradation of power devices. The proposed research is also relevant to the degradation of light-emitting diodes (LEDs) and solar cells. Engineering-level models will be made available to the engineering community and software developers. Students and post-docs will be trained on the bridge area between physics and electrical engineering. An annual workshop for high-school teachers will make them aware of the new developments and their significance. We will also participate in the NSF-funded Tennessee Louis Stokes Alliance for Minority Participation (TLSAMP) by funding one minority undergraduate every summer to participate in our research program.High-electron-mobility transistors (HEMTs) based on wide-gap materials such as GaN hold great promise for high-power, high-temperature applications, but their use is currently limited by reliability issues. Device-reliability modeling is currently done by phenomenological models that rely primarily on correlations between device failure and device parameters. Modeling based on physical phenomena re-quires knowledge of the underlying physical processes. In recent work by the lead-PI and collaborators, parameter-free quantum-mechanical calculations were combined with electrical stress data to demonstrate that hot-electron degradation of III-V high electron mobility transistors (HEMTs) is caused by either the release of H from specific defects or by a reconfiguration of specific impurities. A simple engineering-level model was developed to fit stress data and predict long-term degradation under operating conditions. The model relies on sophisticated Monte-Carlo solutions of the Boltzmann equation in the real device to get electron densities in space and energy, but uses a crude step function for the capture cross section for hydrogen release. The Principal Investigators have had extensive experience in these areas. Pantelides and Zhang recently developed a theory for the calculation of inelastic scattering cross sections of hot electrons by defects, while Pantelides and Schrimpf have worked on the identification of defects through electrical measurements and quantum mechanical calculations, developing simple engineering-level models. The objective of this proposal is to fully implement the newly-developed theory of cross sections and perform pertinent calculations that can be used to model available and new experimental data on device degradation. The objective will be to develop accurate, validated, engineering-level models to describe and predict short term and long-term hot-electron-induced device degradation. The results will also be useful for understanding and modeling the role of defects in the performance and degradation of solar cells and light-emitting diodes.