[SOLVED] [MOVED] Technology parameter vs (W/L)

[harrym]

Hello,
I am simulating the characteristic of a NMOS transistor with HSPICE, and I am also calculating the technology parameter knowing Id and gm, with the formula
un*Cox=gm^2/2Ids *(L/W) (I am operating in saturation).
Increasing both the W and the L of the transistor of the same ratio (like doubling both L and W), I do not get the same value of k, the difference is almost 50%!
I am using a tcms018 model.

Where am I wrong?

harrym

 

[harrym]

I try to rephrase the question.
I always thought that the technology parameter (un*Cox) was a constant of the circuit (after having doped the semiconductor un is constant, and Cox depends only on the thickness of the insulator), so I cannot understand why I get different values from the simulation. Does un change if I change the transistor W or L?

 

[Dominik Przyborowski]

Current gain factor (µCox) depends to biasing conditions due to high values of vertical and lateral electric fields.
In fact You should use this formula:
\[K = \mu ~C_{ox} = \frac{K_0}{[1+(\frac{Vds_{sat}}{E_c L})^\beta]^{1/\beta}[1+V_{od}(\theta+K_0(R_S + R_D)W/L)]}\]
where: K_0 is current gain factor for low fields, beta is equal 1 for holes and 2 for electrons, Ec=0.8MV/m for electrons and 1.95MV/m for holes is critical electric field, theta is parameters depending to technology and is around 1e-7/tox, Rs and Rd are series source and drain resistances (could be quite high for narrow devices).