Chaulker Or H, Turkulets Yury, Shalish Ilan
School of Electrical and Computer Engineering, Ben-Gurion University, 8410501, Beer Sheva, Israel.
Sci Rep. 2025 May 29;15(1):18773. doi: 10.1038/s41598-025-97446-w.
GaN is taking over from silicon in power electronics, but its density of interface states has yet to be adequately controlled. The major step turning GaN into a technological semiconductor was its p-type doping. Mg is currently the only p-type dopant in technological use in GaN. Its incorporation into the lattice is difficult, requiring a thermal treatment that only partially activates the Mg. To achieve moderate p-type doping requires high doses of Mg that mostly remain inactive - a potential for defects. These defects have mostly been studied by photoluminescence that cannot differentiate bulk from surface states. Here, we used an absorption-based spectroscopy, originally invented by the inventors of the transistor - surface photovoltage spectroscopy. The results seem to transform the picture radically as far as our understanding of the famous Mg-associated blue luminescence. We observe an optical transition from the valence band into a deep trap around 0.49 eV above the valence band, along with what appears to be a complimentary transition from the same trap into the conduction band peaking at 2.84 eV (precisely coinciding with the "blue luminescence" energy). The similar shape of the spectra, their complimentary energies within the bandgap, and their opposite nature (hole vs. electron trap), appear to be more than a coincidence suggesting an Mg-related surface state. We suggest that small amounts of surface-segregated Mg oxidize during the post-growth Mg-activation heat treatment forming a surface state. HCl etch is observed to affect the photovoltage at the "blue luminescence"-related energy. A surface treatment is unlikely to affect the bulk. The only case that could support existence of these blue-emitting centers deeper than the surface is when they decorate extended defects - a special case of surface states. Finally, we show that reported luminescence from pure MgO produces the same blue luminescence even at the absence of GaN.
氮化镓正在电力电子领域取代硅，但它的界面态密度尚未得到充分控制。将氮化镓转变为技术半导体的主要一步是其p型掺杂。镁是目前氮化镓技术应用中唯一的p型掺杂剂。将其掺入晶格很困难，需要进行热处理，而这种热处理只能部分激活镁。要实现适度的p型掺杂需要高剂量的镁，而这些镁大多保持非活性状态——这可能会产生缺陷。这些缺陷大多是通过光致发光进行研究的，而光致发光无法区分体相态和表面态。在此，我们使用了一种基于吸收的光谱学方法，它最初是由晶体管的发明者发明的——表面光电压光谱学。就我们对著名的与镁相关的蓝色发光的理解而言，这些结果似乎从根本上改变了情况。我们观察到一个从价带跃迁到价带上方约0.49电子伏特处的深陷阱的光学跃迁，以及一个从同一陷阱跃迁到导带的互补跃迁，其峰值出现在2.84电子伏特处（恰好与“蓝色发光”能量一致）。光谱的相似形状、它们在带隙内的互补能量以及它们相反的性质（空穴陷阱与电子陷阱），似乎并非巧合，这表明存在与镁相关的表面态。我们认为，在生长后镁激活热处理过程中，少量表面偏析的镁会氧化，形成一种表面态。观察到盐酸蚀刻会影响与“蓝色发光”相关能量处的光电压。表面处理不太可能影响体相。唯一能支持这些比表面更深的蓝色发光中心存在的情况是，当它们修饰扩展缺陷时——这是表面态的一种特殊情况。最后，我们表明，即使在没有氮化镓的情况下，报道的纯氧化镁发出的光也会产生相同的蓝色发光。
