In many of the examples in SILVACO ATLAS, i find most of the structures are modeled without substrate. But i can see a lot of difference in the device performance with n without substrate. So why do we not consider substrate usually in simulation?
If the substrate layer is an active part of the device, it must be taken into consideration in calculating the electrical characteristic of the device. If the substrate is just for the mechanical support and it is electrically idle, then one do not to consider it.
As an example assume the devices built in the silicon over insulator. The silicon substrate under the insulator is gust for mechanical support and the devices are built inside the silicon film above the I-layer. In the n-MOS technology, there is two types of transistors in the chip, the depletion and enhancement transistors. The substrate us a part of them and it has to be considered. Also in the p-well cmos technology the p-well acts as a substrata for the nmos transistors while the n-substrate is a part of the p-mos transistor and must be considered.
In summary, before solving for the device you have to define its layer structure and the function of each layer. Then you determine the bulk and the boundaries. The boundaries are enclosing the active regions controlling the device performance. What remains out side the boundaries does not affect the performance of the device however has another functions like mechanical support. So, one can not say that the substrate effect is always negligible.
If the substrate layer is an active part of the device, it must be taken into consideration in calculating the electrical characteristic of the device. If the substrate is just for the mechanical support and it is electrically idle, then one do not to consider it.
As an example assume the devices built in the silicon over insulator. The silicon substrate under the insulator is gust for mechanical support and the devices are built inside the silicon film above the I-layer. In the n-MOS technology, there is two types of transistors in the chip, the depletion and enhancement transistors. The substrate us a part of them and it has to be considered. Also in the p-well cmos technology the p-well acts as a substrata for the nmos transistors while the n-substrate is a part of the p-mos transistor and must be considered.
In summary, before solving for the device you have to define its layer structure and the function of each layer. Then you determine the bulk and the boundaries. The boundaries are enclosing the active regions controlling the device performance. What remains out side the boundaries does not affect the performance of the device however has another functions like mechanical support. So, one can not say that the substrate effect is always negligible.