Logo image
在矽基板上之氮化鎵高電子遷移率場效電晶體之基板模型建立與分析
Thesis

在矽基板上之氮化鎵高電子遷移率場效電晶體之基板模型建立與分析

張展溢
Masters, 國立清華大學, 電子工程研究所
2010

Abstract

氮化鎵 基板模型
GaN-based transistors have been developed successfully for high power and microwave RF applications. Compared with Si, GaN has a higher breakdown field and electron drift velocity, which makes GaN-based devices an excellent candidate for such applications. It has been reported that AlGaN/GaN high electron mobility transistors (HEMTs) on SiC or sapphire substrates demonstrated excellent RF performance such as high output power and low high-frequency noise. However, the cost of both SiC and sapphire substrates are relatively higher compared to the silicon substrate. In addition, the poor thermal conductivity of the sapphire substrate also degrades the device performance under high power operation. Compared with the SiC and sapphire substrates, GaN-on-silicon is much more attractive with low cost, large size substrates, and good thermal conductivity. Recent publications have shown that GaN-based HEMTs grown on silicon substrate could reach good DC characteristics, but their RF performance is not as expected mainly due to the parasitics from the silicon substrate. In this work, a novel extraction methodology is proposed, allowing extraction of the small-signal model directly including the parasitic capacitance and resistance of substrate. The parameters of substrate parasitics can be extracted by solving the analytical equations without any assumption. The sub-micron AlGaN/GaN HEMTs are fabricated using the in-house developed technology. With a 0.5-贡m gate length and a width of 100 贡m, the device demonstrated fT= 21 GHz and fmax= 37 GHz on high resistivity silicon substrates. The proposed modeling approach are applied to several types of devices with different thicknesses of the GaN buffer layer. Based on the extracted small-signal equivalent circuit models, the fmax could be improved by 10 to 20 percent if the substrate parasitics are removed.

Metrics

1 Record Views

Details

Logo image