Delta 摻雜磷化銦鎵/砷化銦鎵/砷化鎵調變摻雜場效電晶體之研究

Abstract

我們利用Subnikov-de Hass量測方法,在絕對溫度零點六度時,發現 delta摻雜的磷化銦鎵材料具有二維電子雲的存在.次能帶的能位及次能帶 上的波函數是利用自組式計算法解帕松和薛丁格方程式得來的.測得的三 個次能帶上的二維電子雲密度分別為 1.73 x 10^12, 0.87 x 10^12 及 0.086 x 10^12 cm-2.計算的四個次能帶的能位分別為 -39.33, -12.31, -8.00 及 -4.09 meV. 一個閘長2 um, n-型磷化銦鎵/砷化鎵的調變摻雜電晶體展示一個74 mS/ mm的最大異本質互導,一個121 mS/mm 的最大本質互導.這元件具有一個源 極電阻50歐姆,一個最大振盪頻率(fmax) 16GHz,一個截止頻率(ft) 6.2GHz. 第二種元件是一個閘 長1.2 um,Delta-摻雜磷化銦鎵/砷化銦鎵/砷化鎵的調變摻雜場 效電晶 體.展示一個 64 mS/mm的最大異質互導,一個163 mS/mm的最大本質互導. 這元件具有一個源極電阻50歐姆,一個最大振盪頻率(fmax) 13.4 GHz,一 個截止頻率(ft) 5.2 GHz.最大異質互導質不高是由於高源極電阻和低電 子遷移率所致. 我們也利用Medici軟體預 測第二種元件的D.C.特性.模擬結果跟實驗結果十分吻合.這顯示Medici是 一套不錯的軟體工具可用於元件特性預測及最佳化.元件的小訊號等效電 路 利用 Touchstone軟體搭配測量的S-參數得來的.從元件的小訊號等效 電路,我們可歸納出三個影響這些元件特性的原因:(1). 互導受限於高源 極電阻和低電子遷移率(2). ft受限於大的寄生電容 (3). fmax值不高是 因本質互導與輸出互導比值和閘-源極電容與汲-閘 極電容比值皆不高所 致. We have investigated the existence of electron subbands in delta-doped InGaPmaterial with Si using Subnikov-de Haas measurement at 0.6K. The Subbandenergy levels and the electron wave functions of these subbands are calculatedself- consistently by solving the Poisson and the Schrodinger eqautions.The desities of 2DEG are 1.73 x 10^12, 0.87 x 10^12, and 0.086 x 10^12 cm^-2for the 1st, 2nd, and 3rd subband, respectively. The calculated subband energylevels are -39.33, -12.31, -8.00, and -4.09 meV below the Fermi level for thecorresponding four subbands. An n-In0.49Ga0.51P/GaAs MODFET with 2 um gate length demostrates a maximumextrinsic transconductance of 74 mS/mm and a maximum intrinsic transconduc-tance of 121 mS/mm at 300K. The device has a source resistance of 53 ohms, amaximum oscillaton frequency (fmax) of 16 GHz, and a cutoff frequency (ft) of6.2 GHz. The second device is a delta-doped In0.49Ga0.51P/In0.1Ga0.9As/ GaAs MODFETwith 1.2 um gate length. It demonstrates a maximum extrinsic transconductanceof 64 mS/mm at 300K. The lower extrinsic transconductance is due to highersource resistance (95 ohms) and the lower mobility. The measured famx and ftare 13.4 GHz and 5.2 GHz, respectively. We have also employed the software, Medici, to predict the D.C. characteris-tic of the second device. The measured characteristics are reasonable agree-ment with the simulated results. It shows that Medici is a good tool to pre-dict and optimize device performance. An equivalent circuit has been obtainedusing the Touchstone program along with the measured S- parameters. From theequivalent circuit, we conclude that three are three reasons that affect theperformance of these MODFETs: (1) the transconductance is limted by large ser-ies resistance and low mobility, (2) ft is restricted by large parasitic capa-citances, and (3) fmax is poor due to lower ratios of transconductance to out-put transconductance and gate- source capacitance to drain-gate capacitance.