Bagnall Kevin R, Wang Evelyn N
Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts 02139, USA.
Rev Sci Instrum. 2016 Jun;87(6):061501. doi: 10.1063/1.4954203.
Micro-Raman thermography is one of the most popular techniques for measuring local temperature rise in gallium nitride (GaN) high electron mobility transistors with high spatial and temporal resolution. However, accurate temperature measurements based on changes in the Stokes peak positions of the GaN epitaxial layers require properly accounting for the stress and/or strain induced by the inverse piezoelectric effect. It is common practice to use the pinched OFF state as the unpowered reference for temperature measurements because the vertical electric field in the GaN buffer that induces inverse piezoelectric stress/strain is relatively independent of the gate bias. Although this approach has yielded temperature measurements that agree with those derived from the Stokes/anti-Stokes ratio and thermal models, there has been significant difficulty in quantifying the mechanical state of the GaN buffer in the pinched OFF state from changes in the Raman spectra. In this paper, we review the experimental technique of micro-Raman thermography and derive expressions for the detailed dependence of the Raman peak positions on strain, stress, and electric field components in wurtzite GaN. We also use a combination of semiconductor device modeling and electro-mechanical modeling to predict the stress and strain induced by the inverse piezoelectric effect. Based on the insights gained from our electro-mechanical model and the best values of material properties in the literature, we analyze changes in the E2 high and A1 (LO) Raman peaks and demonstrate that there are major quantitative discrepancies between measured and modeled values of inverse piezoelectric stress and strain. We examine many of the hypotheses offered in the literature for these discrepancies but conclude that none of them satisfactorily resolves these discrepancies. Further research is needed to determine whether the electric field components could be affecting the phonon frequencies apart from the inverse piezoelectric effect in wurtzite GaN, which has been predicted theoretically in zinc blende gallium arsenide (GaAs).
显微拉曼热成像技术是用于测量氮化镓(GaN)高电子迁移率晶体管中局部温度升高的最常用技术之一,具有高空间和时间分辨率。然而,基于GaN外延层斯托克斯峰位置变化进行准确的温度测量需要适当考虑逆压电效应引起的应力和/或应变。通常将夹断状态用作温度测量的无功率参考,因为在GaN缓冲层中引起逆压电应力/应变的垂直电场相对独立于栅极偏置。尽管这种方法得到的温度测量结果与从斯托克斯/反斯托克斯比率和热模型得出的结果一致,但从拉曼光谱的变化中量化夹断状态下GaN缓冲层的机械状态存在很大困难。在本文中,我们回顾了显微拉曼热成像的实验技术,并推导了纤锌矿GaN中拉曼峰位置对应变、应力和电场分量的详细依赖关系的表达式。我们还结合半导体器件建模和机电建模来预测逆压电效应引起的应力和应变。基于从我们的机电模型获得的见解以及文献中材料特性的最佳值,我们分析了E2高和A1(LO)拉曼峰的变化,并证明在逆压电应力和应变的测量值和模型值之间存在重大的定量差异。我们研究了文献中针对这些差异提出的许多假设,但得出结论,没有一个能令人满意地解决这些差异。需要进一步研究以确定电场分量是否可能在纤锌矿GaN中除了逆压电效应之外还影响声子频率,这在理论上已在闪锌矿砷化镓(GaAs)中得到预测。