側向擴散金氧半電晶體之多諧波失真模型及表面聲波氣體感測器之設計

Abstract

本論文主要針對主動元件在不同佈局結構、非線性模型與感測領域進行分析與應用。一般而言，半導體裡的主動元件，例如金氧半場效電晶體(MOSFET)或雙極性接面電晶體(BJT)，為無線通訊系統或感測系統中最重要的元件之一。其元件特性足以影響應用系統裡的整體表現、價格與穩定性。而在一般無線通訊領域或基地台等遠距離的發射器，大都以側向擴散金氧半電晶體(laterally-diffused MOS transistor)作為主要之放大元件。在本研究中，首先提出了側向擴散金氧半電晶體之新型佈局樣式，利用圓形佈局樣式以達到低開啟通道電阻(On-resistance)與小面積的元件佈局。依據實驗的量測結果，圓形的佈局樣式因有較小的寄生電容與較高之轉導增益，所以有較高的截止頻率(fT)與最大震盪頻率(fmax)。除了小訊號分析外，論文中也進行大訊號特性分析比較。而在研究的量測結果，與傳統佈局樣式比較之下，圓形佈局樣式亦有較佳之大訊號特性表現。 除了功率元件佈局設計外，主動元件之非線性特性模型在設計與應用上也相形重要。在研究中，我們分析了多諧波失真模型，並且利用此模型模擬出主動元件的非線性行為。經由晶圓級非線性向量網路分析儀所萃取出之多諧波失真模型，在1.9GHz的操作頻率點，射頻側向擴散金氧半電晶體的大訊號特性模擬結果與傳統功率量測的結果一致，並且無需另外進行電腦最佳化與曲線近似處理。 論文中的另一部分，我們利用了半導體中的主動元件與表面聲波(Surface acoustic wave, SAW)延遲線元件進行感測器之設計與分析。研究中針對了元件的特性，進行此感測器之電性測試與氣體感測結果分析。在感測的實驗結果中，表面聲波感測器在50x103ppm的酒精氣體濃度裡，其中心訊號有近10kHz的訊號漂移。研究的最後並針對感測元件與系統，提出改進及可能的發展方向。

This dissertation presents the layout design, nonlinear modeling and sensing application in terms of active device. The active device, such as MOSFET or BJT in semiconductor, is one of the most important components of a wireless communication or sensing system generally. It plays a significant part in determining the overall performance, cost, and reliability of these application systems. In the world of RF wireless communications, the base-stations and long range transmitters use silicon laterally-diffused MOS (LDMOS) high power transistors almost exclusively. To achieve lower on-resistance and a more compact device size, this study adopted an annular structure in the layout design. According to the measurement results, the smaller drain parasitic resistance in the annular structure could be the key factor for improving ft and fmax. In additional to the small-signal analysis, the large-signal characteristics, such as power gain and power added efficiency, were also improved compared to the transitional structure of LDMOS. In addition to high power device design, the behavior model of the nonlinear characteristics for active device is also crucial. In this study, we analyze the polyharmonic distortion model (PHD) and use this model to predict the nonlinear behavior of active device. By way of the PHD model extracted using on-wafer nonlinear vector network analyzer (NVNA), the large-signal validation of this model also shows a good match with measurements at 1.9 GHz without optimization and curve fitting. In another part of this thesis, we discussed and analyzed the sensor design completely using CMOS active device and SAW delay-line device. Their electrical characteristics are evaluated as well as vapor sensing results. The sensing experimental results show that the maximum oscillation frequency shift between gas on and off is approximately 10 kHz with 50 x 103 ppm alcohol vapor concentration. Conclusions on the sensing device and system, and recommendations concerning potential improvements of these components are discussed, finally.