Product Documentation
Spectre Circuit Simulator Components and Device Models Reference
Product Version 23.1, June 2023

Core Model

Lateral Gradient Factor

The lateral field gradient in SPMOS is reduced with surface potential through the following semi-empirical formula:

where

Effective Drain-Source Voltage

The saturation voltage is given by

where

The effective drain-source voltage is given by

Surface Potential

Surface Potential at Source End of Channel

The approximate analytical solution for surface potential is

where the normalized imref splitting is

In the process of computing surface potential, the following are computed as well

where

and

The evaluation of , , is carefully ordered to avoid over/underflow problems.

After evaluating surface potential , you can compute the normalized inversion charge at the source

where

Series resistance is given by:

Series resistance factor is given by:

Effective vertical field is given by:

where for n-channel and for p-channel MOSFETs.

Effective mobility at the source end of the channel is given by:

The variable is given by

where the term introduces non-universality. The denominator assures that does not exceed 5 for extreme (and unphysical) .

and are temporary variables. Eventually these will be changed to assure the symmetry of the model. Also, if external model of series resistance is used.

Surface Potential at Drain End of Channel

Surface potential at the drain end of the channel is

where

The above equation is used when

where

For , it is more efficient to:

  1. Compute
  2. Determine normalized drain-source surface potential difference
  3. Compute .

While computing , the following variables are computed as well:

where

Mid-Point Surface Potential

The following variables are used:

Normalized inversion charge is calculated by:

Linearization coefficient is

Series resistance:

Series resistance factor:

Effective vertical field is

where for n-channel and for p-channel MOSFETs.

Effective mobility

Quantum Mechanical Corrections

In SPMOS quantum-mechanical (QM) corrections are considered in the most common case which is of interest for the charge-sheet models.

QM corrections are directly used for and .

This is preferable to correcting and , especially in the case when is a small difference of two large variables.

In the following equations, superscript 0 refers to variables uncorrected for QM effects.

For (i.e. for )

where

For ,

There is no correction for . This form is introduced to eliminate the singularity or unphysical behavior near . Coefficients 0.04 and 3 are not affected by model parameters and are fixed.

In addition to correcting and , QM effects are introduced into

and variables , which are given by the above expressions but with corrected for QM effects.

Polysilicon Depletion

In SPMOS polysilicon depletion equations are conditioned to provide smooth device characteristics for a wide voltage range but at present the poly effects are only included for .

The normalized poly surface potential at midpoint

where

In this section the superscript 0 indicates that the variable is not corrected for poly depletion effect.

Poly corrections are introduced into and rather than into and directly.

The corrected midpoint surface potential is

where

The correction to normalized surface potential difference is as follows

where is given in section “Quantum Mechanical Corrections” with superscript 0 indicating that the variable is not corrected for poly depletion effect

In addition to changing the surface potentials, poly correction affects the linearization of inversion charge and intrinsic charges. The expressions in sections “Drain Current” and “Intrinsic Charges” include these corrections.

The case of no poly effect can be recovered by setting . While physically this corresponds to , in SPMOS eliminating poly effects is formally prescribed by setting in the parameter file.

Drain Current

This ensures that and during SPICE convergence when can be unphysically high.

The channel length modulation factor is calculated by the following equation:

Drain current is calculated by

where the inversion charge linearization (including polysilicon depletion effect) is:

The reduced channel length is

Intrinsic Charges

All charges are normalized to .

Gate charge

where

Inversion layer charge

Drain charge (computed using Ward-Dutton partition)

Source charge

Bulk charge

Bias-Dependent Body Factor

If NSLP>10-3,

Otherwise

Normalized body factor is


Return to top
 ⠀
X