本論文主要是要探討如何有效的去節省在模擬半導體時所需的記憶體空間，我們開發了四種不同的方法去模擬半導體元件。第一種方法是聯合法，，此種方法所需的記憶體空間在所有的方法裡面是最大的。第二種方法是部分分離法，此種方法所需的記憶體空間會比聯合法小。最後我們開發了元件分割法，此種方法可區分為二個不同的模式，分別為部分重疊和非部分重疊，元件分割法的優點是我們執行程式時將不會再受限於記憶體的空間。我們將會用上述的這幾種方法去模擬n型通道的金氧半場效電晶體，並比較它們模擬的結果，最後我們將使用這些方法去模擬互補式金氧半導體並且比較它們模擬的結果。 In this thesis, we focus on how to effectively save required memory space when simulate a semiconductor device. We develop four different methods to simulate semiconductor devices. First method is coupled method(CM). It requires biggest memory space in all of the methods. Second method is partial decoupled method(PDM). The memory space with PDM is less than CM. Last, we develop device-partition method. It can be divided into two different modes. One is overlapped mode, the other is unoverlapped mode. The advantage of device-partition method is that our program can be compiled without limitation by memory space. We will use these methods to simulate an n-channel MOSFET and compare simulation result. Finally, we will use these methods to simulate a CMOS circuit and compare simulation result also.
