- If a library says it is for 45nm process node, shouldn't the length and width of MOS be specified in lib itself ?
It means that transistors below a gate length of 45nm are not realizable. If you choose to make larger devices, the model will be chosen automatically (called binning).
- Although I can set the required values from schematic itself, How will I know if I am entering practical values, like drain and source
area and perimeters. I believe if I enter W and L in nano meter scale
and leave drain perimeter and area as it is, then I will get some
impractical results.
That is called circuit design. Correct dimensions are chosen by the designer not the simulator.(Addendum-point-2. The circuit designer chooses a W/L to realize specifications for each device. The layout designer (who is often, but not always, the same person) does the layout for the devices. Then a post-layout extraction is performed and a simulation is run with actual drain/source areas as well as other parasitics included in the netlist. The results from schematic simulation should match the results from the post-layout simulation.)
- Is there any way to get default values of these parameters ? and do I need to know about all the parameters written in library ? as most
of them don't make sense to me.
An analog designer has to worry about aspect ratio (W/L) of each of the transistor in the circuit. Indeed it is the only thing any designer has control over.
A digital designer often uses libraries that provide digital primitives and are made available with the design kit. The designer of such libraries are analog designers themselves.
- I am unable to find values of gamma and lambda in downloaded spice model
Trying to relate level 1,2,3 models with BSIM4 models is a non-trivial exercise. At this point, it seems that you are not conversant with design flows for your kit's ecosystem.
(EDIT-2-Addendum on pt4) Spice modeling of on-chip devices is an area of research in itself. In the 70s when VLSI was in its infancy and SPICE had just been introduced, the process nodes were at 10um and the device modeling was done with few parameters. Each of the parameters had a well-understood physical meaning. Fast forward half a century, a BSIM model can have upwards of 200 parameters depending on the level and these are basically a result of fitting copious amounts of measured data using a computer. The complex higher order effects can't be described using any physical model. In other words, not all these parameters have a physical meaning anymore. You can find more information from publications at this Berkeley web page.