金氧半電晶體次臨界電流之統計模擬

Abstract

金氧半電晶體(MOS) 操作在次臨界區域具有許多優點，特別是在對積體類 比電路的設計與實現上。故有必要瞭解金氧半電晶體元件與其構成之電路 操作在次臨界區域的特性。因此首先我們必須根據實驗數據建立一個有效 率的模型，藉由表面反應法(RSM) ，包括實驗計劃法與迴歸分析等步驟， 最後我們可以建立一個二次式函數的統計模型; 根據輸入結構參數即可得 到所要之元件特性參數值。此模型可幫助瞭解金氧半電晶體次臨界電流相 應結構參數之變動影響，因此高性能CMOS類比電路操作於次臨界區域設計 與實現可以獲得。 MOSFET's operating in the subthreshold current region have many advantages in neural network system implementations and applications. In order to in vestigate MOS devices and circuits whose components are operating in the subthreshold current region for large-scale analog circuits, we must first build up an efficient model based on the experimental data. By the response surface methodology based on experiment design and regression analysis, we can establish the quadratic model between device performance response and input structure parameters. This model can help us understand the variations of MOSFET subthreshold current and thus the high-performance CMOS analog curcuits using the subthreshold current characteristics can be achieved.