针对GaN基高频功率HEMT从材料缺陷到器件性能的深能级陷阱机理研究

GaN-based HEMT is one of the most promising candidates for next generation micro-/millimeter-wave power electronic devices. However, the performance of practical devices is far from the theoretical value. Deep-level trapping effect is the largest challenge that limits the potential performance of GaN-based HEMTs, which is a critical problem that needs to be solved. Most of the reported research work about the characterization of traps is based on the capacitance-mode deep-level transient spectroscopy (C-DLTS) for large devices, but it is very difficult for C-DLTS to identify traps for down-scaled higher-frequency devices. In addition, intentional doping is usually adopted during the epitaxial growth for high-frequency power HEMTs, which is helpful to reduce the leakage current and to improve the break-down voltage and short-channel effects. However, intentional doping simultaneously results in deep-level traps, leading to more sever current collapse. So far there is no a uniform theory to explain the mechanism of trapping effect in GaN-based devices, and further experiments are needed to prove the relationship between the material defects and the trap properties as well as the device performances. In order to study the nature of various traps especially in small-scale devices, we propose the project “Research on the trapping mechanism of deep levels in GaN-based high-frequency power HEMTs: from material defects to device performances.” With the design and growth of epitaxial materials with different structures and various crystal qualities (e.g. different dislocation densities or different doping concentrations), the aim of this research project is to compare different deep-level characterization techniques, to study the relationship between the defects in material and the properties of deep levels (i.e. space distribution, energy level and cross section of traps), to investigate the influence of deep-level traps on the electrical characteristics of devices (e.g. break-down voltage, leakage current, current collapse), and finally to propose an effective method of eliminating or minimizing the trapping effect for improving the reliability of GaN-based high-frequency power HEMTs.

GaN基HEMT是最具发展前景的微波/毫米波功率电子器件之一，然而，实际器件性能却远未达其理论值，深能级陷阱是制约其性能提高的最大障碍，也是亟待解决的关键问题。现有报道的陷阱表征大多采用瞬态电容法，该方法对小尺寸高频器件却不再适用。而且，高频功率HEMT的衬底外延材料通常进行掺杂以减小漏电、提高击穿电压并改善短沟效应，但同时会引入更多深能级陷阱，导致电流崩塌更加严重。目前国际上对GaN基器件中的陷阱机理没有统一的理论模型，材料缺陷与陷阱特征参数以及器件特性之间的关系缺乏更深入的实验验证。为此，我们提出本项目申请，通过设计生长不同结构和不同质量（不同位错密度和掺杂浓度）的外延材料，对比不同的陷阱表征技术，分析GaN材料中的位错和杂质等缺陷与深能级陷阱参数（即陷阱浓度、陷阱能级和截获面积等）之间的内在联系，研究陷阱对器件电特性（击穿电压、漏电、电流崩塌等）的影响，提出有效抑制陷阱效应的措施。

本项目针对GaN基HEMT器件面临的深能级陷阱效应等可靠性问题，设计生长了不同结构和不同质量的GaN外延材料，系统全面地分析了深能级陷阱特征参数（陷阱类型、陷阱能级、陷阱截获面积、陷阱空间位置等）和器件特性之间的内在联系，结合实验数据和数值仿真，最终确定了影响GaN基高频功率器件特性（如电流崩塌、漏电、动态导通电阻等）的主要物理机制，提出了有效抑制或减小陷阱效应的方法措施，为进一步提高GaN基高频功率HEMT器件的性能与可靠性提供了理论指导，全面完成了计划书规定的研究内容和研究目标。取得的标志性成果或主要结论为：（1）GaN器件中深能级陷阱效应的内在物理机制及仿真模型的建立：设计制备了小尺寸AlGaN/GaN金属氧化物半导体高电子迁移率晶体管（MOSHEMT），并基于脉冲I-V测试方法对器件的动态特性进行了深入研究，建立了包含深能级陷阱效应的器件二维数值仿真模型，结合实验数据和数值仿真模拟，分析了深能级陷阱对AlGaN/GaN MOSHEMT器件动态特性的影响以及相关陷阱效应的内在物理机制，提取出了器件电流崩塌与静态偏置电压之间的依赖关系；（2）GaN器件中深能级陷阱效应的有效抑制方法措施：研究发现，器件中的深能级陷阱效应主要是由于栅漏电注入和热电子注入两种陷阱机制共同作用，从而导致器件的电流崩塌随着栅极静态偏置电压的增加呈非单调变化趋势。因此，通过改善栅介质的质量以减小栅漏电或提高外延材料质量以减少缺陷密度等措施可以达到抑制陷阱效应的目的，从而抑制电流崩塌；（3）AlGaN/GaN新器件结构及其陷阱效应研究：采用栅介质自对准工艺制备了AlGaN/GaN Fin-MISHEMT新型器件结构，发现与传统平面MISHEMT相比，由于较强的栅控能力，Fin-MISHEMT的电流崩塌较小，而且电流崩塌主要是由栅源和栅漏之间陷阱效应导致，通过优化器件结构（比如仅在栅下设计鳍形沟道）可以进一步抑制动态导通电阻；（4）GaN器件的可靠性研究：提出了多种优化器件结构、抑制陷阱效应的具体措施，例如，优化材料生长条件提高材料质量、插入高质量栅介质层减小栅漏电、优化表面钝化预处理工艺等，成功制备出高可靠性的InAlN/GaN HFET器件。.本研究成果对GaN基高频功率HEMT器件的设计与制备具有重要的指导意义，在该项目资助下课题组共发表SCI或EI收录相关学术文章11篇。