Escolar Documentos
Profissional Documentos
Cultura Documentos
1 MOSFET Model
- User’s Manual
Copyright © 2001
BSIM4.2.0 Developers:
• Professor Chenming Hu (project director), UC Berkeley
• Weidong Liu, UC Berkeley
• Xiaodong Jin, UC Berkeley
• Kanyu M. Cao, UC Berkeley
• Jeff J. Ou, UC Berkeley
• Xuemei Xi, UC Berkeley
BSIM4.1.0 Developers:
• Professor Chenming Hu, UC Berkeley
• Weidong Liu, UC Berkeley
• Xiaodong Jin, UC Berkeley
• Kanyu M. Cao, UC Berkeley
• Jeff J. Ou, UC Berkeley
BSIM4.0.0 Developers:
• Professor Chenming Hu, UC Berkeley
• Weidong Liu, UC Berkeley
• Xiaodong Jin, UC Berkeley
• Kanyu M. Cao, UC Berkeley
• Jeff J. Ou, UC Berkeley
Web Sites:
http://www-device.eecs.berkeley.edu/bsim3/~bsim4.html
The continuous scaling of minimum feature size brought challenges to compact modeling in two
ways: One is that to push the barriers in making transistors with shorter gate length, advanced
process technologies are used such as non-uniform substrate doping. The second is its
opportunities to RF applications.
To meet these challenges, BSIM4 has the following major improvements and additions over
BSIM3v3:
(1) An accurate new model of the intrinsic input resistance for both RF, high-frequency
analog and high-speed digital applications;
(2) An accurate gate direct tunneling model;
(3) A gate-induced drain/source leakage (GIDL/ GISL) current model, available in BSIM
for the first time;
(4) Flexible substrate resistance network for RF modeling;
Other improvements include: 5) A new accurate channel thermal noise model and a noise
partition model for the induced gate noise; 6)A non-quasi-static (NQS) model that is
consistent with the Rg-based RF model and a consistent AC model that accounts for the
NQS effect in both transconductances and capacitances. 7)A comprehensive and versatile
geometry-dependent parasitics model for various source/drain connections and multi-finger
devices; 8)Improved model for steep vertical retrograde doping profiles; 9)Better model for
pocket-implanted devices in Vth, bulk charge effect model, and Rout; 10)Asymmetrical and
bias-dependent source/ drain resistance, either internal or external to the intrinsic MOSFET
at the user's discretion; 11)Acceptance of either the electrical or physical gate oxide
thickness as the model input at the user's choice in a physically accurate manner; 12)The
quantum mechanical charge-layer-thickness model for both IV and CV; 13) A more
accurate mobility model for predictive modeling; 14)An improved unified flicker (1/f) noise
model, which is smooth over all bias regions and considers the bulk charge effect;
15)Different diode IV and CV characteristics for source and drain junctions; 16)Junction
diode breakdown with or without current limiting; 17)Dielectric constant of the gate
dielectric as a model parameter.
Table of Contents
High-k gate dielectric can be modeled as SiO 2 (relative permittivity: 3.9) with an
equivalent SiO 2 thickness. For example, 3nm gate dielectric with a dielectric
constant of 7.8 would have an equivalent oxide thickness of 1.5nm.
BSIM4 also allows the user to specify a gate dielectric constant (EPSROX)
different from 3.9 (SiO 2) as an alternative approach to modeling high-k dielectrics.
Figure 1-1 illustrates the algorithm and options for specifying the gate dielectric
thickness and calculation of the gate dielectric capacitance for BSIM4 model
evaluation.
No No No
TOXE and TOXP TOXE given? TOXP given?
both given?
• C oxe = EPSROX ⋅ ε 0
TOXE , Coxe is used to calculate Vth , subthreshold swing, Vgsteff, Abulk ,
mobiliy, Vdsat, K1ox, K2ox, capMod = 0 and 1, etc
EPSROX ⋅ ε 0
• C oxp = TOXP , Co xp is used to calculate Coxeff for drain current and capMod =
2 through the charge-layer thickness model:
1 .9 ×10 −9 cm
X DC =
V gsteff + 4(VTH 0 − VFB − Φ s )
0. 7
1 +
2TOXP
• If DTOX is not given, its default value will be used.
doping concentration in the substrate is NSUB. The depletion width in the poly
gate is Xp. The depletion width in the substrate is Xd. The positive charge near the
interface of the poly-silicon gate and the gate oxide is distributed over a finite
depletion region with thickness Xp. In the presence of the depletion region, the
voltage drop across the gate oxide and the substrate will be reduced, because part
of the gate voltage will be dropped across the depletion region in the gate. That
means the effective gate voltage will be reduced.
NGATE
Figure 1-2. Charge distribution in a MOSFET with the poly gate depletion effect.
The device is in the strong inversion region.
The effective gate voltage can be calculated in the following manner. Assume the
doping concentration in the poly gate is uniform. The voltage drop in the poly gate
Vpoly can be calculated as
(1.2.1)
qNGATE ⋅ X
2
V poly = 0.5 X poly E poly =
poly
2ε si
where Epoly is the maximum electrical field in the poly gate. The boundary
condition at the interface of poly gate and the gate oxide is
(1.2.2)
EPSROX ⋅ Eox = ε si E poly = 2 qε si NGATE ⋅ Vpoly
where Eox is the electric field in the gate oxide. The gate voltage satisfies
(1.2.3)
Vgs − VFB − Φ s = V poly + Vox
where Vox is the voltage drop across the gate oxide and satisfies Vox = Eox TOXE.
(1.2.4)
where
(1.2.5)
EPSROX 2
a=
2 qε si NGATE ⋅ TOXE 2
By solving (1.2.4), we get the effective gate voltage Vgse which is equal to
(1.2.6)
Vgse = VFB + Φ s + 1+ −1
EPSROX 2 qε si NGATE ⋅ TOXE 2
(1.3.1)
Leff = Ldrawn+ XL− 2dL
(1.3.2a)
Wdrawn
Weff = + XW − 2dW
NF
(1.3.2b)
Wdrawn
Weff ' = + XW − 2dW'
NF
The difference between (1.3.2a) and (1.3.2b) is that the former includes bias
dependencies. NF is the number of device fingers. dW and dL are modeled by
(1.3.3)
(
dW = dW '+ DWG ⋅ Vgsteff + DWB Φ s − Vbseff − Φ s )
WL WW WWL
dW ' = WINT + WLN
+ W W N + WLN W W N
L W L W
(1.3.4)
LL LW LWL
dL = LINT + LLN
+ LWN + LLN LWN
L W L W
WINT represents the traditional manner from which "delta W" is extracted (from
the intercept of straight lines on a 1/Rds~Wdrawn plot). The parameters DWG and
DWB are used to account for the contribution of both gate and substrate bias
effects. For dL, LINT represents the traditional manner from which "delta L" is
extracted from the intercept of lines on a Rds~Ldrawn plot).
The remaining terms in dW and dL are provided for the convenience of the user.
They are meant to allow the user to model each parameter as a function of Wdrawn ,
Ldrawn and their product term. By default, the above geometrical dependencies for
dW and dL are turned off.
MOSFET capacitances can be divided into intrinsic and extrinsic components. The
intrinsic capacitance is associated with the region between the metallurgical source
and drain junction, which is defined by the effective length (Lactive ) and width
(W active) when the gate to source/drain regions are under flat-band condition. Lactive
and Wactive are defined as
(1.3.5)
Lactive= Ldrawn+ XL−2dL
(1.3.6)
Wdrawn
Wactive= + XW− 2dW
NF
(1.3.7)
LLC LWC LWLC
dL = DLC + LLN
+ LWN + LLN LWN
L W L W
(1.3.8)
WLC WWC WWLC
dW = DWC + WLN
+ W W N + WLN W W N
L W L W
The meanings of DWC and DLC are different from those of WINT and LINT in the
I-V model. Unlike the case of I-V, we assume that these dimensions are bias-
dependent. The parameter δLeff is equal to the source/drain to gate overlap length
plus the difference between drawn and actual POLY CD due to processing (gate
patterning, etching and oxidation) on one side.
The effective channel length Leff for the I-V model does not necessarily carry a
physical meaning. It is just a parameter used in the I-V formulation. This Leff is
therefore very sensitive to the I-V equations and also to the conduction
characteristics of the LDD region relative to the channel region. A device with a
large Leff and a small parasitic resistance can have a similar current drive as
another with a smaller Leff but larger Rds.
capacitnace models, and DWC, DLC, LLC, LWC, LWLC, WLC, WWC and WWLC
will be set to the values of their DC counterparts.
BSIM4 uses the effective source/drain diffusion width Weffcj for modeling
parasitics, such as source/drain resistance, gate electrode resistance, and gate-
induced drain leakage (GIDL) current. W effcj is defined as
(1.3.9)
Wdrawn WLC WWC WWLC
Weffcj = − 2 ⋅ DWJ + WLN + W W N + WLN W W N
NF L W L W
Note: Any compact models have their valided design region, so does BSIM4. The
warning limits for effective channel length, channel width, and gate oxide
thickness are substantially decreased to avoid a large number of warnings when
BSIM4 is used beyond its design region. For meaningful results, it is
recommended to keep these variables/parameters within the BSIM4 design region,
while decreasing warning limits will allow users to extend the model beyond with
fewer warnings. The table below table shows the design limit and warning limit.
For users’ reference, the fatal limitation in BSIM4 is also shown.
(2.1.1)
(2.1.2)
2qε si N substrate
γ =
Coxe
Equation (2.1.1) assumes that the channel doping is constant and the channel
length and width are large enough. Modifications have to be made when the
substrate doping concentration is not constant and/or when the channel is short, or
narrow.
(2.2.1)
qD0 qD
Vth = Vth, NDEP + + K1NDEP ϕ s − Vbs − 1 − ϕ s − Vbs
Coxe ε si
(2.2.2)
(
Vth, NDEP = VTH 0 + K1NDEP ϕ s − Vbs − ϕ s )
with a definition of
(2.2.3)
k BT NDEP
ϕ s = 0.4 + ln
q ni
(2.2.4)
(2.2.5)
0 dep 0
(2.2.6)
(
Vth = VTH 0 + K1 Φ s − Vbs − Φ s − K 2 ⋅Vbs )
(2.2.7)
k BT NDEP
Φ s = 0.4 + ln + PHIN
q ni
where
PHIN = − qD10 ε si
VTH0, K1, K2, and PHIN are implemented as model parameters for model
flexibility. Appendix A lists the model selectors and parameters.
(2.2.8)
K1 = γ 2 − 2K 2 Φ s − VBM
(2.2.9)
(γ 1 − γ 2 )( Φ s − VBX − Φ s )
K2 =
2 Φs ( )
Φ s − VBM − Φ s + VBM
where γ1 and γ2 are the body bias coefficients when the substrate doping
concentration are equal to NDEP and NSUB, respectively:
(2.2.10)
2qε si NDEP
γ1 =
Coxe
(2.2.11)
2 qε si NSUB
γ2 =
Coxe
VBX is the body bias when the depletion width is equal to XT, and is
determined by
(2.2.12)
qNDEP ⋅ XT 2
= Φ s − VBX
2ε si
(2.3.1)
( )
Vth = VTH 0 + K 1 Φ s − Vbs − Φ s ⋅ 1 +
LPEB
Leff
− K 2 ⋅ Vbs
LPE 0
+ K1 1 + −1 Φs
L
eff
(2.3.2)
∆Vth (DITS ) = −nvt ⋅ ln
(1 − e −Vds / vt
)⋅ L eff
Leff + DVTP0 ⋅ 1 + e ( − DVTP1⋅V ds
)
(2.3.3)
Leff
∆Vth (DITS ) = −nvt ⋅ ln
eff (
L + DVTP0 ⋅ 1 + e − DVTP1⋅Vds )
(2.4.1)
∆Vth (SCE, DIBL ) = −θ th (Leff )⋅ [2(Vbi − Φ s ) + Vds ]
(2.4.2)
k BT NDEP ⋅ NSD
Vbi = ln 2
q n i
(2.4.3)
θ th (Leff ) =
0 .5
cosh ( )− 1
Leff
lt
(2.4.4)
ε si ⋅ TOXE ⋅ X dep
lt =
EPSROX ⋅η
(2.4.5)
2ε si (Φ s − Vbs )
X dep =
qNDEP
Xdep is larger near the drain due to the drain voltage. Xdep / η represents the
average depletion width along the channel.
Note that in BSIM3v3 and [4], θ th(Leff ) is approximated with the form of
(2.4.6)
Leff L
θ th (Leff ) = exp − + 2 exp − eff
2l t lt
which results in a phantom second Vth roll-up when Leff becomes very
small (e.g. Leff < LMIN). In BSIM4, the function form of (2.4.3) is
implemented with no approximation.
(2.4.7)
0.5 ⋅ DVT 0
θ th (SCE ) =
(
cosh DVT 1⋅
Leff
lt
)− 1
(2.4.8)
∆Vth (SCE) = −θ th (SCE) ⋅ (Vbi − Φ s )
with l t changed to
(2.4.9)
ε si ⋅ TOXE ⋅ X dep
lt = ⋅ (1 + DVT 2 ⋅Vbs )
EPSROX
(2.4.10)
θ th (DIBL ) =
0 .5
(
cosh DSUB ⋅
Leff
lt 0
) −1
(2.4.11)
∆Vth (DIBL ) = −θ th (DIBL )⋅ ( ETA0 + ETAB ⋅Vbs )⋅ Vds
and l t0 is calculated by
(2.4.12)
ε si ⋅ TOXE ⋅ X dep0
lt 0 =
EPSROX
with
(2.4.13)
2ε siΦ s
X dep 0 =
qNDEP
DVT1 is basically equal to 1/(η)1/2. DVT2 and ETAB account for substrate
bias effects on SCE and DIBL, respectively.
(2.5.1)
This formulation includes but is not limited to the inverse of channel width
due to the fact that the overall narrow width effect is dependent on process
(i.e. isolation technology). Vth change is given by
(2.5.2)
In addition, we must consider the narrow width effect for small channel
lengths. To do this we introduce the following
(2.5.3)
0.5 ⋅ DVT 0W
∆Vth (Narrow- width2 ) = −
(
cosh DVT1W ⋅
Leff Weff '
l tw )− 1⋅ (V
bi −Φs )
with l tw given by
(2.5.4)
ε si ⋅ TOXE ⋅ X dep
ltw = ⋅ (1 + DVT 2W ⋅ Vbs )
EPSROX
(2.5.5)
(
Vth = VTH 0 + K 1ox ⋅ Φ s − Vbseff − K 1⋅ Φ s 1 + ) LPEB
Leff
− K 2oxVbseff
LPE0
− 1 Φ s + (K 3 + K 3B ⋅Vbseff )
TOXE
+ K1ox 1 + Φs
Leff Weff '+W 0
(Vbi − Φ s )
DVT 0W DVT 0
− 0 .5 ⋅ +
( ) (
cosh DVT 1W effltweff −1 cosh DVT 1 lefft −1
L W ' L
)
−
0 .5
(ETA0 + ETAB ⋅Vbseff )⋅ Vds
(
cosh DSUB lteff0 −1
L
)
where TOXE dependence is introduced in model parameters K1 and K2 to
improve the scalibility of Vth model over TOXE as
(2.5.6)
TOXE
K1ox = K1⋅
TOXM
and
(2.5.7)
TOXE
K 2ox = K 2 ⋅
TOXM
Note that all Vbs terms are substituted with a Vbseff expression as shown in
(2.5.8). This is needed in order to set an upper bound for the body bias
during simulations since unreasonable values can occur during SPICE
iterations if this expression is not introduced.
(2.5.8)
where δ1 = 0.001V, and Vbc is the maximum allowable Vbs and found from
dVth /dVbs = 0 to be
(2.5.9)
K12
Vbc = 0.9 Φ s −
4 K 22
(3.1.1)
where
(3.1.1a)
VOFFL
Voff ' = VOFF +
Leff
(3.1.2)
Qchs0 = Coxe ⋅ (Vgse − Vth )
A unified charge density model considering the charge layer thickness effect is
derived for both subthreshold and inversion regions as
(3.1.3)
Qch0 = Coxeff ⋅Vgsteff
(3.1.4)
Coxe ⋅ Ccen ε si
Coxeff = with Ccen =
Coxe + Ccen X DC
(3.1.5)
1.9 × 10−9 cm
X DC =
+ 4(VTH 0 − VFB − Φ s )
0 .7
V
1 + gsteff
2TOXP
In the above equations, Vgsteff, the effective (Vgse-Vth) used to describe the channel
charge densities from subthreshold to strong inversion, is modeled by
(3.1.6a)
m∗ (Vgse −Vth )
nvt ln 1 + exp
nvt
=
( )
1 − m (Vgse −Vth ) −Voff '
Vgsteff ∗
2Φs
m∗ + nCoxe ⋅ exp−
qNDEPε si nvt
where
(3.1.6b)
arctan (MINV )
m ∗ = 0 .5 +
π
MINV is introduced to improve the accuracy of Gm, Gm/Id and Gm2/Id in the
moderate inversion region.
To account for the drain bias effect, The y dependence has to be included in
(3.1.3). Consider first the case of strong inversion
(3.1.7)
Qchs ( y ) = Coxeff ⋅ (Vgse − Vth − AbulkVF ( y ))
VF (y) stands for the quasi-Fermi potential at any given point y along the channel
with respect to the source. (3.1.7) can also be written as
(3.1.8)
Qchs ( y ) = Qchs0 + ∆Qchs ( y )
The term ∆Qchs(y) = -Coxeff Abulk VF (y) is the incremental charge density introduced
by the drain voltage at y.
In subthreshold region, the channel charge density along the channel from source
to drain can be written as
(3.1.9)
A V ( y)
Qchsubs( y ) = Qchsubs0 ⋅ exp − bulk F
nvt
Taylor expansion of (3.1.9) yields the following (keeping the first two terms)
(3.1.10)
A V ( y)
Qchsubs ( y ) = Qchsubs0 1 − bulk F
nvt
(3.1.11)
Qchsubs( y ) = Qchsubs0 + ∆Qchsubs( y )
where ∆Qchsubs(y) is the incremental channel charge density induced by the drain
voltage in the subthreshold region. It is written as
(3.1.12)
AbulkVF ( y )
∆Qchsubs ( y ) = −Qchsubs0 ⋅
nvt
To obtain a unified expression for the incremental channel charge density ∆Qch(y)
induced by Vds, we assume ∆Qch (y) to be
(3.1.13)
∆Q chs ( y ) ⋅ ∆Qchsubs( y )
∆Qch ( y ) =
∆Qchs ( y ) + ∆Qchsubs ( y )
(3.1.14)
VF ( y )
∆Qch ( y ) = − Qch0
Vb
(3.1.15)
Vgsteff + 2vt
Vb =
Abulk
A unified expression for Qch (y) from subthreshold to strong inversion regions is
(3.1.16)
V ( y)
Qch ( y ) = Coxeff ⋅ Vgsteff ⋅ 1 − F
Vb
(3.2.1)
where
(3.2.2)
W qε si NDEP 2
I0 = µ vt
L 2Φ s
vt is the thermal voltage and equal to kB T/q. Voff’ = VOFF + VOFFL / Leff is the
offset voltage, which determines the channel current at Vgs = 0. In (3.2.1), n is the
(3.2.3)
Cdep Cdsc −Term + CIT
n = 1 + NFACTOR ⋅ +
Coxe Coxe
As the gate oxide thickness is scaled down to 3nm and below, gate leakage current
due to carrier direct tunneling becomes important. This tunneling happens between
the gate and silicon beneath the gate oxide. The tunneling carriers can be either
electrons or holes, or both, either from the conduction band or valence band,
depending on (the type of the gate and) the bias regime.
In BSIM4, the gate tunneling current components include the tunneling current
between gate and substrate (Igb), and the current between gate and channel (Igc),
which is partitioned between the source and drain terminals by Igc = Igcs + Igcd.
The third component happens between gate and source/drain diffusion regions (Igs
and Igd). Figure 4-1 shows the schematic gate tunneling current flows.
Igs Igd
Igcs Igcd
Igb
Figure 4-1. Shematic gate current components flowing between NMOST terminals
in version.
(4.2.1a)
Voxacc = V fbzb − VFBeff
(4.2.1b)
(4.2.2)
and
(4.2.3)
VFBeff = V fbzb − 0.5 (V fbzb − Vgb − 0.02) + (V − Vgb − 0.02)2 + 0.08V fbzb
fbzb
(4.3.1)
Igbacc = Weff Leff ⋅ A ⋅ ToxRatio ⋅ Vgb ⋅Vaux
⋅ exp [− B ⋅ TOXE ( AIGBACC − BIGBACC ⋅Voxacc ) ⋅ (1 + CIGBACC ⋅Voxacc )]
NTOX
TOXREF 1
ToxRatio = ⋅
TOXE TOXE 2
Vgb −V fbzb
Vaux = NIGBACC ⋅ vt ⋅ log 1 + exp −
NIGBACC ⋅ vt
(4.3.2)
Igbinv = Weff Leff ⋅ A ⋅ ToxRatio ⋅Vgb ⋅ Vaux
[
⋅ exp − B ⋅ TOXE ( AIGBINV − BIGBINV ⋅Voxdepinv )⋅ (1 + CIGBINV ⋅ Voxdepinv ) ]
Voxdepinv − EIGBINV
Vaux = NIGBINV ⋅ v t ⋅ log1 + exp
⋅
NIGBINV v t
Igc, determined by ECB for NMOS and HVB (Hole tunneling from
Valence Band) for PMOS, is formulated as
(4.3.3)
Igc = Weff Leff ⋅ A ⋅ ToxRatio ⋅ Vgse ⋅Vaux
[
⋅ exp − B ⋅ TOXE (AIGC − BIGC ⋅Voxdepinv )⋅ (1 + CIGC ⋅Voxdepinv ) ]
where A = 4.97232 A/V2 for NMOS and 3.42537 A/V2 for PMOS, B =
7.45669e11 (g/F-s2) 0.5 for NMOS and 1.16645e12 (g/F-s 2)0.5 for PMOS,
and
Vgse − VTH 0
Vaux = NIGC ⋅ vt ⋅ log 1 + exp
NIGC ⋅ vt
Igs and Igd -- Igs represents the gate tunneling current between the gate
and the source diffusion region, while Igd represents the gate tunneling
current between the gate and the drain diffusion region. Igs and Igd are
determined by ECB for NMOS and HVB for PMOS, respectively.
(4.3.4)
(4.3.5)
[ (
⋅ exp − B ⋅ TOXE ⋅ POXEDGE ⋅ AIGSD − BIGSD ⋅Vgd ⋅ 1 + CIGSD ⋅ Vgd
'
)( '
)]
where A = 4.97232 A/V 2 for NMOS and 3.42537 A/V2 for PMOS, B =
7.45669e11 (g/F-s2) 0.5 for NMOS and 1.16645e12 (g/F-s 2)0.5 for PMOS,
and
NTOX
TOXREF 1
ToxRatioEdge = ⋅
TOXE ⋅ POXEDGE (TOXE ⋅ POXEDGE )2
Vgd =
'
(V gd − V fbsd ) + 1.0e − 4
2
Vfbsd is the flat-band voltage between gate and S/D diffusions calculated as
To consider the drain bias effect, Igc is split into two components, Igcs and
Igcd, that is Igc = Igcs + Igcd, and
(4.3.6)
PIGCD ⋅Vds + exp (− PIGCD ⋅Vds ) − 1 + 1.0e − 4
Igcs = Igc ⋅
PIGCD2 ⋅Vds2 + 2.0e − 4
and
(4.3.7)
1 − (PIGCD ⋅ Vds + 1) ⋅ exp (− PIGCD ⋅Vds ) + 1.0e − 4
Igcd = Igc ⋅
PIGCD2 ⋅ Vds + 2.0e − 4
2
(4.3.8)
B ⋅ TOXE V
PIGCD = 1 − dseff
Vgsteff 2 ⋅ Vgsteff
2
BSIM4 uses Abulk to model the bulk charge effect. Several model parameters are
introduced to account for the channel length and width dependences and bias
effects. Abulk is formulated by
(5.1.1)
A0 ⋅ Leff
⋅
Leff + 2 XJ ⋅ X dep
1
Abulk = 1 + F− doping ⋅ 2
⋅
− ⋅ Leff + B0 1 + KETA⋅Vbseff
Leff + 2 XJ ⋅ X dep Weff '+B1
1 AGS Vgsteff
where the second term on the RHS is used to model the effect of non-uniform
doping profiles
(5.1.2)
Note that Abulk is close to unity if the channel length is small and increases as the
channel length increases.
(5.2.1)
Q + ( Qn 2 )
Eeff = B
ε si
The physical meaning of Eeff can be interpreted as the average electric field
experienced by the carriers in the inversion layer. The unified formulation of
mobility is then given by
(5.2.2)
µ0
µ eff =
1 + ( Eeff E0 ) ν
For an NMOS transistor with n-type poly-silicon gate, (5.2.1) can be rewritten in a
more useful form that explicitly relates Eeff to the device parameters
(5.2.3)
Vgs + Vth
Eeff ≈
6TOXE
BSIM4 provides three different models of the effective mobility. The mobMod = 0
and 1 models are from BSIM3v3.2.2; the new mobMod = 2, a universal mobility
model, is more accurate and suitable for predictive modeling.
• mobMod = 0
(5.2.4)
U0
µ eff = 2
V + 2Vth V + 2Vth
1 + (UA + UCVbseff ) gsteff + UB gsteff
TOXE TOXE
• mobMod = 1
(5.2.5)
U0
µ eff =
Vgsteff + 2Vth Vgsteff + 2Vth
2
1 + UA + UB TOXE (1 + UC ⋅Vbseff )
TOXE
• mobMod = 2
(5.2.6)
U0
µ eff =
+ C0 ⋅ (VTHO − VFB − Φ s )
EU
V
1 + (UA + UC ⋅ Vbseff ) gsteff
TOXE
RDSWMIN + RDSW ⋅
Rds (V ) =
( ) 1 (1e6 ⋅ W ) WR
gd fbsd
(5.3.3)
RSWMIN + RSW ⋅
Rs (V ) = 1
[(1e6 ⋅W ) WR
⋅ NF ]
− PRWB ⋅ Vbs + 1 + PRWG ⋅ (V − V )
effcj
gs
fbsd
Vfbsd is the calculated flat-band voltage between gate and source/drain as given in
Section 4.3.2.
The diffusion source/drain resistance Rsdiff and Rddiff models are given in the
chapter of layout-dependence models.
(5.4.1)
dVF ( y )
I ds ( y ) = WQ ch ( y )µ ne ( y )
dy
(5.4.2)
µeff
µne( y) =
Ey
1+
Esat
(5.4.3)
V ( y ) µ eff dVF ( y )
I ds ( y ) = WQ ch0 1 − F
Vb E dy
1+ y
Esat
(5.4.3) is integrated from source to drain to get the expression for linear
drain current. This expression is valid from the subthreshold regime to the
strong inversion regime
(5.4.4)
V
Wµ eff Qch0Vds 1 − ds
I ds 0 = 2Vb
V
L1 + ds
Esat L
(5.4.5)
Idso
Ids =
RdsIdso
1+
Vds
(5.5.1)
µeff E
v= E < Esat
1 + E Esat
= VSAT E ≥ Esat
where Esat corresponds to the critical electrical field at which the carrier velocity
becomes saturated. In order to have a continuous velocity model at E = Esat , Esat
must satisfy
(5.5.2)
2VSAT
Esat =
µ eff
In this case, the LDD source/drain resistances are either zero or non zero
but not modeled inside the intrinsic channel region. It is easy to obtain Vdsat
as [7]
(5.6.1)
EsatL(Vgsteff + 2vt)
Vdsat =
AbulkEsatL + Vgsteff + 2vt
(5.6.2a)
− b − b 2 − 4 ac
Vdsat =
2a
where
(5.6.2b)
1
a = Abulk Weff VSATCoxe Rds + Abulk − 1
2
λ
(5.6.2c)
(Vgsteff + 2vt ) λ −1 + Abulk EsatLeff
2
b = −
+ 3 Abulk(Vgsteff + 2vt )Weff VSATCoxe Rds
(5.6.2d)
(5.6.2e)
λ = A1Vgsteff + A2
(5.6.3)
The first region is the triode (or linear) region in which carrier velocity is not
saturated. The output resistance is very small because the drain current has a strong
dependence on the drain voltage. The other three regions belong to the saturation
region. As will be discussed later, there are several physical mechanisms which
affect the output resistance in the saturation region: channel length modulation
(CLM), drain-induced barrier lowering (DIBL), and the substrate current induced
body effect (SCBE). These mechanisms all affect the output resistance in the
saturation range, but each of them dominates in a specific region. It will be shown
next that CLM dominates in the second region, DIBL in the third region, and
SCBE in the fourth region.
3.0 14
10
Rout (KOhms)
2.0
8
Ids (mA)
1.5
6
1.0
4
0.5
2
0.0 0
0 1 2 3 4
Vds (V)
The channel current is a function of the gate and drain voltage. But the current
depends on the drain voltage weakly in the saturation region. In the following, the
Early voltage is introduced for the analysis of the output resistance in the
saturation region:
(5.7.1)
∂I ds (Vgs ,Vds )
I ds (Vgs ,Vds ) = I dsat (Vgs ,Vdsat ) + ∫
Vds
⋅ dVd
Vd s a t ∂Vd
= I dsat (Vgs ,Vdsat )⋅ 1 + ∫
Vds 1
⋅ dVd
Vd s a t VA
(5.7.2)
∂I (V ,V )
−1
VA = I dsat ⋅ ds gs ds
∂Vd
We assume in the following analysis that the contributions to the Early voltage
from all mechanisms are independent and can be calculated separately.
(5.7.3)
∂I (V ,V ) ∂L
−1
= I dsat ⋅ ds gs ds ⋅
∂Vd
VACLM
∂L
(5.7.4)
VACLM = Cclm ⋅ (Vds − Vdsat )
where
(5.7.5)
1 Vgsteff R ⋅ I V 1
Cclm = ⋅ F ⋅ 1 + PVAG 1 + ds dso Leff + dsat ⋅
PCLM
Esat Leff
Vdseff Esat litl
and the F factor to account for the impact of pocket implant technology is
(5.7.6)
1
F=
Leff
1 + FPROUT ⋅
Vgsteff + 2vt
(5.7.7)
ε siTOXE ⋅ XJ
litl =
EPSROX
(5.7.8)
VADIBL = I dsat ⋅ ⋅
∂Vth ∂Vd
(5.7.9)
where θrout has a similar dependence on the channel length as the DIBL
effect in Vth, but a separate set of parameters are used:
(5.7.10)
PDIBLC1
θ rout =
2 cosh (
DROUT⋅ Leff
lt 0 )− 2 + PDIBLC 2
Parameters PDIBLC1, PDIBLC2, PDIBLCB and DROUT are introduced to
correct the DIBL effect in the strong inversion region. The reason why
DVT0 is not equal to PDIBLC1 and DVT1 is not equal to DROUT is
because the gate voltage modulates the DIBL effect. When the threshold
voltage is determined, the gate voltage is equal to the threshold voltage.
But in the saturation region where the output resistance is modeled, the
gate voltage is much larger than the threshold voltage. Drain induced
barrier lowering may not be the same at different gate bias. PDIBLC2 is
usually very small. If PDIBLC2 is put into the threshold voltage model, it
will not cause any significant change. However it is an important parameter
in VADIBLC for long channel devices, because PDIBLC2 will be dominant if
the channel is long.
When the electrical field near the drain is very large (> 0.1MV/cm), some
electrons coming from the source (in the case of NMOSFETs) will be
energetic (hot) enough to cause impact ionization. This will generate
electron-hole pairs when these energetic electrons collide with silicon
atoms. The substrate current I sub thus created during impact ionization will
increase exponentially with the drain voltage. A well known Isub model [8]
is
(5.7.11)
B ⋅ litl
I ds (Vds − Vdsat ) exp − i
Ai
I sub =
Bi V −
ds dsat
V
(5.7.12)
V −V
I ds = I ds− w / o− Isub + I sub = I ds− w / o −Isub ⋅ 1 + Bi ds Bdsat
(i ⋅litl
Ai exp Vds −Vd s a t )
The Early voltage due to the substrate current VASCBE can therefore be
calculated by
(5.7.13)
Bi B ⋅ litl
VASCBE = exp i
Ai ds
V − V dsat
(5.7.14)
It has been shown that a long-channel device with pocket implant has a
smaller Rout than that of uniformly-doped device [3]. The Rout degradation
factor F is given in (5.7.6). In addition, the pocket implant introduces a
potential barrier at the drain end of the channel. This barrier can be lowered
by the drain bias even in long-channel devices. The Early voltage due to
DITS is modeled by
(5.7.15)
VADITS =
1
[
⋅ F ⋅ 1 + (1 + PDITSL ⋅ Leff )exp ( PDITSD ⋅ Vds ) ]
PDITS
(5.8.1)
(5.8.2)
VA is written as
(5.8.3)
VA = VAsat + V ACLM
where VAsat is
(5.8.4)
VAsat =
[
Esat Leff +Vdsat + 2RdsvsatCoxeWeff Vgsteff ⋅ 1− 2(Vgsteff
A
bulk dsatV
+ 2vt ) ]
RdsvsatCoxeWeff Abulk −1+ λ2
VAsat is the Early voltage at Vds = Vdsat. VAsat is needed to have continuous drain
current and output resistance expressions at the transition point between linear and
saturation regions.
In addition to the junction diode current and gate-to-body tunneling current, the substrate
terminal current consists of the substrate current due to impact ionization (Iii), and gate-
induced drain leakage current (IGIDL ).
(6.1.1)
(6.1.2)
(6.2.1)
where AGIDL, BGIDL, CGIDL, and EGIDL are model parameters and explained
in Appendix A. CGIDL accounts for the body-bias dependence of IGIDL and IGISL.
WeffCJ and Nf are the effective width of the source/drain diffusions and the number
of fingers. Further explanation of WeffCJ and Nf can be found in the chapter of the
layout-dependence model.
Accurate modeling of MOSFET capacitance plays equally important role as that of the
DC model. This chapter describes the methodology and device physics considered in both
intrinsic and extrinsic capacitance modeling in BSIM4.0.0. Complete model parameters
can be found in Appendix A.
• Separate effective channel length and width are used for capacitance models.
• capMod = 0 uses piece-wise equations. capMod = 1 and 2 are smooth and single
equation models; therefore both charge and capacitance are continous and smooth
over all regions.
• Threshold voltage is consistent with DC part except for capMod = 0, where a long-
channel V th is used. Therefore, those effects such as body bias, short/narrow channel
and DIBL effects are explicitly considered in capMod = 1 and 2.
• Overlap capacitance comprises two parts: (1) a bias-independent component which
models the effective overlap capacitance between the gate and the heavily doped
source/drain; (2) a gate-bias dependent component between the gate and the lightly
doped source/drain region.
• Bias-independent fringing capacitances are added between the gate and source as well
as the gate and drain.
The accumulation charge and the substrate charge are associated with the
substrate while the channel charge comes from the source and drain
terminals
(7.2.1)
Qg = −( Qsub + Qinv + Qacc )
Qb = Qacc + Qsub
Q = Q + Q
inv s d
The substrate charge can be divided into two components: the substrate
charge at zero source-drain bias (Qsub0 ), which is a function of gate to
substrate bias, and the additional non-uniform substrate charge in the
presence of a drain bias (δQsub). Qg now becomes
(7.2.2)
Q g = −(Qinv + Qacc + Q sub0 + δQsub )
The total charge is computed by integrating the charge along the channel.
The threshold voltage along the channel is modified due to the non-
uniform substrate charge by
(7.2.3)
Vth ( y ) = Vth (0) + ( Abulk −1)Vy
(7.2.4)
Lactive Lactive
where Ey is expressed in
(7.2.5)
Wactiveµ eff Coxe
Vgt − bulk Vds Vds = Wactive µ eff Coxe (Vgt − AbulkVy )E y
A
I ds =
Lactive 2
(7.2.6)
∂Qi
Cij =
∂V j
∑ C = ∑C
i
ij
j
ij =0
(7.2.7)
Vgsteff ,CV
Vdsat, IV < Vdsat, CV < Vdsat, IV Lactive→ ∞ =
Abulk
(7.2.8)
Vgsteff ,CV
Vdsat, CV =
CLC CLE
Abulk ⋅ 1 +
Lactive
(7.2.9)
V gse − Vth −VOFFCV
Vgsteff, CV = NOFF ⋅ nvt ⋅ ln 1 + exp
NOFF ⋅ nv t
Model parameters CLC and CLE are introduced to consider the effect of
channel-length modulation. Abulk for the capacitance model is modeled by
(7.2.10)
A0 ⋅ Leff B0 1
Abulk = 1 + F− doping ⋅ ⋅+ ⋅
Weff '+ B1 1 + KETA⋅ Vbseff
Leff + 2 XJ ⋅ X dep
where
zero at threshold voltage. The BSIM4 charge and capacitance models are
formulated by substituting Vgst with Vgsteff,CV as
(7.2.11)
Q(Vgst ) = Q(Vgsteff , CV )
(7.2.12)
∂Vgsteff ,CV
C (Vgst ) = C(Vgsteff , CV )
Vg , d , s , b
(7.2.13)
VFBeff = V fbzb − 0.5(V fbzb − Vgb − 0.02 ) + (V − Vgb − 0.02 )2 + 0.08V fbzb
fbzb
where
(7.2.14)
V fbzb = Vth zeroV bs and Vds − Φ s − K1 Φ s
An effective Vds , Vcveff, is used to smooth out the transition between linear
and saturation regions.
(7.2.15)
{
Vcveff = Vdsat,CV − 0.5 V4 + V4 + 4δ 4Vdsat, CV
2
} whereV = V
4 dsat,CV − Vds − δ 4; δ 4 = 0.02V
The inversion charges are partitioned into Qinv = Qs + Qd. The ratio of Qd to
Qs is the charge partitioning ratio. Existing charge partitioning schemes are
0/100, 50/50 and 40/60 (XPART = 1, 0.5 and 0).
This is the most physical model of the three partitioning schemes in which
the channel charges are allocated to the source and drain terminals by
assuming a linear dependence on channel position y.
(7.2.16)
Lactive
y
Qs = Wactive ∫ q c 1 − dy
L
0 active
Lactive
Q = W y
active ∫ q c dy
d 0
Lactive
Cgg (µF/cm )
2
B D
0.6
D
0.4 C
0.2
0.10
C -4 -3 -2 -1 0 1 2 3
Vgs (V)
A Tox=30A
0.05 Nsub=5e17cm
-3
0.00
0 20 40 60
Depth (A)
Figure 7-1. Charge distribution from numerical quantum simulations show significant
charge thickness at various bias conditions shown in the inset.
CTM is a charge-based model and therefore starts with the DC charge thicknss,
XDC . The charge thicknss introduces a capacitance in series with Cox as illustrated
in Figure 7-2, resulting in an effective Coxeff. Based on numerical self-consistent
solution of Shrodinger, Poisson and Fermi-Dirac equations, universal and
analytical XDC models have been developed. Coxeff can be expressed as
(7.3.1)
Coxe ⋅ Ccen
Coxeff =
Coxe + Ccen
where
Ccen = ε si X
DC
Vgs
Figure 7-2. Charge-thickness capacitance concept in CTM. Vgse accounts for the poly
depletion effect.
(7.3.2)
−0 .25
1 NDEP V − V − VFBeff
X DC = Ldebye exp ACDE ⋅ ⋅ gse bseff
2 ×10
16
3 TOXE
where Ldebye is Debye length, and XDC is in the unit of cm and (Vgse - Vbseff - VFBeff)
/ TOXE is in units of MV/cm. For numerical statbility, (7.3.2) is replaced by (7.3.3)
(7.3.3)
X DC = X max −
1
2
(X0 + X 02 + 4δ x X max )
where
X 0 = X max − X DC − δ x
(7.3.4)
1.9 × 10−9 cm
X DC =
+ 4(VTH 0 − VFB − Φ s )
0 .7
V
1 + gsteff
2TOXP
In inversion region, the body charge thickness effect is modeled by including the
deviation of the surface potential Φs (bias-dependence) from 2 Φ B [2]
(7.3.5)
V ⋅ (VgsteffCV + 2 K1ox 2Φ B
ϕδ = Φ s − 2Φ B = ν t ln gsteffCV
MOIN ⋅ K1ox 2ν t
(7.3.6)
qinv = −Coxeff ⋅ (Vgsteff ,CV − ϕ δ )
7.4.1 capMod = 0
Accumulation region
Qg = Wactive LactiveCoxe (Vgs − Vbs − VFBCV )
Qsub = −Q g
Qinv = 0
Subthreshold region
Qg = −Qsub0
Qinv = 0
CLC CLE
Abulk ' = Abulk 1 +
L
eff
Linear region
2
Qg = CoxeWactive Lactive Vgs − VFBCV − Φ s −
Vds
+
Abulk 'Vds
2 A ' V
12 Vgs − Vth − bulk ds
2
Qb = CoxeWactive Lactive VFBCV − Vth − Φ s +
(1 − Abulk ')Vds
−
(1 − Abulk ') Abulk 'Vds
2
2 A 'V
12Vgs − Vth − bulk ds
2
50/50 partitioning:
2 2
Qinv
= −CoxeWactive Lactive Vgs − Vth − Φ s −
Abulk 'Vds
+
Abulk ' Vds
2 Abulk 'Vds
12Vgs − Vth −
2
Qs = Qd = 0.5Qinv
40/60 partitioning:
Qd = −CoxeWactiveLactive
V −V
gs th
− +
[
(V −V )2 A 'V (V −V ) ( A 'V )2
Abulk 'Vds Abulk 'Vds gs 6 th − bulk ds8 gs th + bulk40 ds ]
2 2 (
12 Vgs − Vth − Abulk2'Vds
2
)
Qs = −(Q g + Qb + Q d )
0/100 partitioning:
Qs = −(Q g + Qb + Q d )
Saturation region
V
Q g = CoxeWactiveLactive Vgs − VFBCV − Φ s − dsat
3
Qb = −CoxeWactive Lactive VFBCV + Φ s − Vth +
(1 − Abulk ')Vdsat
3
50/50 partitioning:
40/60 partitioning:
Qs = −(Q g + Qb + Q d )
0/100 partitioning:
Qd = 0
Qs = −(Q g + Qb )
7.4.2 capMod = 1
Qinv = Qs + Qd
Vgsteffcv
Vdsat, cv =
Abulk '
1
2 2
Abulk ' Vcveff
Qinv = − WactiveLactiveCoxe ⋅ Vgsteff ,cv − Abulk 'Vcveff +
2 Abulk 'Vcveff
12 ⋅ Vgsteff ,cv −
2
1 − A ' (1 − Abulk ') ⋅ Abulk 'Vcveff
2
Wactive LactiveCoxe
QS = −
(A 'V ) − ( Abulk 'Vcveff )
2
A 'V + 2 V 2 2 3
2 Vgsteff ,cv − bulk cveff
2 3 gsteff , cv bulk cveff 15
5
−
3
V gsteff , cv Vgsteff , cv 2 Abulk 'Vcveff
WactiveLactiveCoxe 3
QD = − 2
A 'V
2 Vgsteff , cv − bulk cveff
+ V ( A 'V ) − (Abulk'Vcveff )
2 gsteff , cv bulk cveff
2 1 3
5
Wactive LactiveCoxe 3 A ' 2
V 2
QD = −
⋅ Vgsteff ,cv − Abulk'Vcveff +
bulk cveff
2 2 Abulk 'Vdveff
4 ⋅ Vgsteff, cv −
2
7.4.3 capMod = 2
Vgbacc =
1
2
[
⋅ V0 + V02 + 0.08V fbzb ]
V0 = V fbzb + Vbseff − Vgs − 0.02
1
(
Vcveff = Vdsat − ⋅ V1 + V12 + 0.08Vdsat
2
)
V1 = Vdsat − Vds − 0.02
Vgsteff , cv − ϕδ
Vdsat =
Abulk '
V ⋅ (VgsteffCV + 2 K1ox 2Φ B
ϕδ = Φ s − 2Φ B = ν t ln gsteffCV
MOIN ⋅ K1ox 2ν t
1
2 2
Abulk ' Vcveff
Qinv = − WactiveLactiveCoxeff ⋅ Vgsteff , cv − ϕδ − Abulk 'Vcveff +
2 Abulk 'Vcveff
12 ⋅ Vgsteff, cv − ϕδ −
2
1 − A ' (1 − Abulk ') ⋅ Abulk'Vcveff
2
δQsub = Wactive LactiveCoxeff ⋅ bulk
Vcveff −
2 Abulk 'Vcveff
12 ⋅ Vgsteff , cv − ϕδ −
2
50/50 partitioning:
WactiveLactiveCoxeff 1 Abulk '2 Vcveff
2
QS = QD = − Vgsteff, cv − ϕδ − Abulk 'Vcveff +
2 2 A 'V
12 ⋅ Vgsteff ,cv − ϕδ − bulk cveff 2
40/60 partitioning:
WactiveLactiveCoxeff
( Vgsteff,cv − ϕδ )3 − 5 (Vgsteff,cv − ϕδ )2 Abulk'Vcveff
3
QD = − 2
A 'V + (V − ϕ δ )(Abulk 'Vcveff ) − ( Abulk 'Vcveff )
2 1 3
2 Vgsteff, cv − ϕδ − bulk cveff
2 gsteff ,cv 5
0/100 partitioning:
WactiveLactiveCoxeff 1 A ' 2
V 2
QS = −
⋅ Vgsteff, cv − ϕδ + Abulk 'Vcveff −
bulk cveff
2 2 Abulk'Vcveff
12 ⋅ Vgsteff ,cv − ϕ δ −
2
Wactive LactiveCoxeff 3 A ' 2
V 2
QD = −
⋅ Vgsteff , cv − ϕδ − Abulk 'Vcveff +
bulk cveff
2 2 Abulk 'Vdveff
4 ⋅ Vgsteff , cv − ϕδ −
2
(7.5.1)
2 ⋅ EPSROX ⋅ ε
CF = ----------------------------------------0 ⋅ log 1 + 4.0e – 7
-------------------
π TOXE
(7.5.2)
(7.5.3)
2
(7.5.4)
(7.5.5)
V gd , overlap =
1
V gd + δ 1 −
2
(V gd + δ 1 ) + 4δ 1 , δ 1 = 0. 02V
2
(7.5.6)
Qoverlap, g = − (Qoverlap,d + Qoverlap, s + (CGBO ⋅ Lactive ) ⋅ Vgb )
If CGSO and CGDO (the overlap capacitances between the gate and the
heavily doped source/drain regions, respectively) are not given, they will
be calculated. Appendix A gives the information on how CGSO, CGDO
and CGBO are calculated.
As circuit speed and operating frequency rise, the need for accurate prediction of circuit
performance near cut-off frequency or under very rapid transient operation becomes
critical. BSIM4.0.0 provides a set of accurate and efficient high-speed/RF (radio
frequency) models which consist of three modules: charge-deficit non-quasi-static (NQS)
model, intrinsic-input resistance (IIR) model (bias-dependent gate resistance model), and
substrate resistance network model. The charge-deficit NQS model comes from
BSIM3v3.2 NQS model [11] but many improvements are added in BSIM4. The IIR model
considers the effect of channel-reflected gate resistance and therefore accounts for the
first-order NQS effect [12]. Thus, the charge-deficit NQS model and the IIR model should
not be turned on simultaneously. These two models both work with multi-finger
configuration. The substrate resistance model does not include any geometry dependence.
Figure 8-2 shows the RC sub-circuit of charge deficit NQS model for
transient simulation [13]. An internal node, Qdef(t) , is created to keep track
of the amount of deficit/surplus channel charge necessary to reach
equilibrium. The resistance R is determined from the RC time constant (τ).
The current source icheq(t) represents the equilibrium channel charging
effect. The capacitor C is to be the value of Cfact (with a typical value of
–9
1 × 10 Farad [11]) to improve simulation accuracy. Qdef now becomes
(8.1.1)
Qdef (t ) = Vdef × C fact
Qdef
icheq(t) Vdef R
C
Considering both the transport and charging component, the total current
related to the terminals D, G and S can be written as
(8.1.2)
∂Qd , g, s (t )
iD,G ,S (t ) = I D ,G,S (DC) +
∂t
(8.1.3)
Qdef (t ) = Qcheq(t ) − Qch (t )
and
(8.1.4a)
∂Qdef (t ) ∂Qcheq (t ) Qdef (t )
= −
∂t ∂t τ
(8.1.4b)
∂Qd , g ,s (t ) Q def (t )
= D , G, S xpart
∂t τ
The transit time τ is equal to the product of Rii and WeffLeffCoxe, where Rii
is the intrinsic-input resistance [12] given by
(8.1.5)
where Coxeff is the effective gate dielectric capacitance calculated from the
DC model. Note that Rii in (8.1.5) considers both the drift and diffusion
componets of the channel conduction, each of which dominates in
inversion and subthreshold regions, respectively.
(8.1.6)
∆Qcheq (t )
∆Qch (t ) =
1 + jωτ
where ω is the angular frequency. Based on (8.1.6), it can be shown that the
transcapacitances Cgi , Csi, and Cdi (i stands for any of the G, D, S and B
terminals of the device) and the channel transconductances Gm, Gds, and
Gmbs all become complex quantities. For example, now Gm have the form
of
(8.1.7)
G m0 G ⋅ ωτ
Gm = + j − m0 2 2
1+ω τ2 2
1+ ω τ
and
(8.1.8)
Cdg0 Cdg 0 ⋅ ωτ
Cdg = + j −
1 + ω 2τ 2 1+ω τ
2 2
Those quantities with sub “0” in the above two equations are known from
OP (operating point) analysis.
BSIM4 provides four options for modeling gate electrode resistance (bias-
independent) and intrinsic-input resistance (IIR, bias-dependent). The IIR
model considers the relaxation-time effect due to the distributive RC nature
of the channel region, and therefore describes the first-order non-quasi-
static effect. Thus, the IIR model should not be used together with the
charge-deficit NQS model at the same time. The model selector rgateMod
is used to choose different options.
rgateMod = 0 (zero-resistance):
rgateMod = 1 (constant-resistance):
Rgeltd
Rgeltd =
(
RSHG ⋅ XGW + 3⋅ NGCON
W
effcj
)
NGCON ⋅ ( Ldrawn − XGL) ⋅ NF
Refer to Chapter 7 for the layout parameters in the above equation.
Rgeltd+R ii
In this case, the gate resistance is the sum of the electrode gate resistance
(8.1.9) and the intrinsic-input resistance R ii as given by (8.1.5). An inter-
nal gate node will be generated. trnqsMod = 0 (default) and acnqsMod =
0 (default) should be selected for this case.
Rgeltd
Rii
Cgso Cgdo
In this case, the gate electrode resistance given by (8.1.9) is in series with
the intrinsic-input resistance R ii as given by (8.1.5) through two internal
gate nodes, so that the overlap capacitance current will not pass through
the intrinsic-input resistance. trnqsMod = 0 (default) and acnqsMod = 0
(default) should be selected for this case.
The model selector rbodyMod can be used to turn on or turn off the
resistance network.
rbodyMod = 0 (Off):
rbodyMod = 1 (On):
IDS
Iii + IGIDL
RBPS RBPD
sbNode dbNode
bNodePrime
RBPB
RBSB RBDB
bNode
Figure 8-3. Topology with the substrate resistance network turned on.
The following noise sources in MOSFETs are modeled in BSIM4 for SPICE noise
ananlysis: flicker noise (also known as 1/f noise), channel thermal noise and induced gate
noise and their correlation, thermal noise due to physical resistances such as the source/
drain, gate electrode, and substrate resistances, and shot noise due to the gate dielectric
tunneling current. A complete list of the noise model parameters and explanations are
given in Appendix A.
BSIM4 provides two flicker noise models. When the model selector
fnoiMod is set to 0, a simple flicker noise model which is convenient for
hand calculations is invoked. A unified physical flicker noise model, which
is the default model, will be used if fnoiMod = 1. These two modes come
from BSIM3v3, but the unified model has many improvements. For
instance, it is now smooth over all bias regions and considers the bulk
charge effect.
9.1.2 Equations
(9.1.1)
KF ⋅ I ds AF
Sid ( f ) = 2
Coxe Leff f EF
(9.1.2)
kBTq2µeffIds
(
NOIA⋅ log N0 + N + NOIB⋅ (N0 − Nl ) + NOICN02 − Nl 2 )
*
Sid,inv( f ) = 10 *
CoxeLeff Abulkf ⋅10
2 ef
Nl + N 2
k TI ∆L NOIA+ NOIB⋅ Nl + NOIC⋅ Nl
2 2
+ B ds
⋅ clm
W ⋅ L f ⋅10
eff eff
2 ef 10
Nl + N*
2
( )
where µeff is the effective mobility at the given bias condition, and Leff and
Weff are the effective channel length and width, respectively. The parameter
N0 is the charge density at the source side given by
(9.1.3)
N 0 = Coxe ⋅ Vgsteff q
(9.1.4)
AbulkVdseff
N l = Coxe ⋅Vgsteff ⋅ 1 − q
V
gsteff + 2ν t
N* is given by
(9.1.5)
k BT ⋅ (Coxe + Cd + CIT )
N* =
q2
∆Lclm is the channel length reduction due to channel length modulation and
given by
(9.1.6)
V ds − V dseff
+ EM
∆L clm = Litl ⋅ log Litl
E sat
2VSAT
E sat =
µ eff
(9.1.7)
NOIA ⋅ k BT ⋅ I ds2
Sid , subVt( f ) = 2
Weff Leff f EF
N * ⋅ 1010
(9.1.8)
S id ,inv ( f ) × Sid , subvt ( f )
Sid ( f ) =
Sid , subvt ( f ) + Sid ,inv ( f )
(9.2.1)
4 kBT∆f
id2 = ⋅ NTNOI
Rds (V ) +
Leff 2
µeff Qinv
(9.2.2)
AbulkVdseff Abulk2Vdseff 2
Qinv = WactiveLactiveCoxeff ⋅ NF ⋅ Vgsteff − +
2 (
12 ⋅ Vgsteff − bulk2 dseff
A V
)
v d2
i d2 i d2
Source side
• tnoiMod = 1 (holistic)
In this thermal noise model, all the short-channel effects and velocity
saturation effect incorporated in the IV model are automatically included,
hency the name “holistic thermal noise model”. In addition, the
amplification of the channel thermal noise through Gm and Gmbs as well as
the induced-gate noise with partial correlation to the channel thermal noise
are all captured in the new “noise partition” model. Figure 9-1b shows
schematically that part of the channel thermal noise source is partitioned to
the source side.
(9.2.3)
Vdseff ∆f
vd 2 = 4k BT ⋅θ tnoi2 ⋅
I ds
and the noise current source put in the channel region with gate and body
amplication is given by
(9.2.4)
Vdseff ∆f
id = 4 kBT
2
[Gds + β tnoi ⋅ (Gm + Gmbs )]2
I ds
− vd ⋅ (Gm + Gds + G mbs )
2 2
where
(9.2.5)
V
2
θ tnoi = 0.37 ⋅ 1 + TNOIB ⋅ Leff ⋅ gsteff
Esat Leff
and
(9.2.6)
V
2
β tnoi = 0.577 ⋅ 1 + TNOIA ⋅ Leff ⋅ gsteff
Esat Leff
In the following, the equations for the source-side diode are given. The
model parameters are shown in Appendix A.
• dioMod = 0 (resistance-free)
(10.1.1)
qVbs
I bs = I sbs exp − 1 ⋅ f breakdown + Vbs ⋅ Gmin
NJS ⋅ k BTNOM
(10.1.2)
I sbs = Aseff J ss (T ) + Pseff J ssws(T ) + Weffcj ⋅ NF ⋅ J sswgs(T )
(10.1.3)
q ⋅ ( BVS + Vbs )
f breakdown = 1 + XJBVS ⋅ exp −
NJS ⋅ k BTNOM
• dioMod = 1 (breakdown-free)
No breakdown is modeled. The exponential IV term in (10.1.4) is
linearized at the limiting current IJTHSFWD in the forward-bias model
only.
(10.1.4)
qVbs
I bs = I sbs exp − 1 + Vbs ⋅ G min
NJS ⋅ k BTNOM
• dioMod = 2 (resistance-and-breakdown):
Diode breakdown is always modeled. The exponential term (10.1.5) is
linearized at both the limiting current IJTHSFWD in the forward-bias mode
and the limiting current IJTHSREV in the reverse-bias mode.
(10.1.5)
qVbs
I bs = I sbs exp − 1 ⋅ f breakdown + Vbs ⋅ Gmin
NJS ⋅ k BTNOM
The drain-side diode has the same system of equations as those for the
source-side diode, but with a separate set of model parameters as explained
in detail in Appendix A.
• dioMod = 0 (resistance-free)
(10.1.6)
qVbd
I bd = I sbd exp −1 ⋅ f breakdown + Vbd ⋅ G min
NJD ⋅ k BTNOM
(10.1.7)
I sbd = Adeff J sd (T ) + Pdeff J sswd (T ) + Weffcj ⋅ NF ⋅ J sswgd (T )
(10.1.8)
q ⋅ (BVD + Vbd )
f breakdown = 1 + XJBVD ⋅ exp −
NJD ⋅ k BTNOM
• dioMod = 1 (breakdown-free)
No breakdown is modeled. The exponential IV term in (10.1.9) is
linearized at the limiting current IJTHSFWD in the forward-bias model
only.
(10.1.9)
qVbd
I bd = I sbd exp − 1 + Vbd ⋅ Gmin
NJD ⋅ k BTNOM
• dioMod = 2 (resistance-and-breakdown):
Diode breakdown is always modeled. The exponential term (10.1.10) is
linearized at both the limiting current IJTHSFWD in the forward-bias mode
and the limiting current IJTHSREV in the reverse-bias mode.
(10.1.10)
qVbd
I bd = I sbd exp −1 ⋅ f breakdown + Vbd ⋅ G min
NJD ⋅ k BTNOM
(10.2.1)
Cbs = Aseff C jbs + Pseff C jbssw + Weffcj ⋅ NF ⋅ C jbsswg
where Cjbs is the unit-area bottom S/B junciton capacitance, Cjbssw is the
unit-length S/B junction sidewall capacitance along the isolation edge, and
Cjbsswg is the unit-length S/B junction sidewall capacitance along the gate
edge. The effective area and perimeters in (10.2.1) are given in Chapter 11.
Cjbs is calculated by
if Vbs < 0
(10.2.2)
− MJS
Vbs
C jbs = CJS (T ) ⋅ 1 −
PBS (T )
otherwise
(10.2.3)
Vbs
C jbs = CJS (T ) ⋅ 1 + MJS ⋅
PBS (T )
Cjbssw is calculated by
if Vb s < 0
(10.2.4)
− MJSWS
C jbssw = CJSWS (T ) ⋅ 1 −
Vbs
PBSWS (T )
otherwise
(10.2.5)
C jbssw = CJSWS (T ) ⋅ 1 + MJSWS ⋅
Vbs
PBSWS (T )
Cjbsswg is calculated by
if Vb s < 0
(10.2.6)
− MJSWGS
= CJSWGS (T ) ⋅ 1 −
Vbs
C jbsswg
PBSWGS (T )
otherwise
(10.2.7)
− MJSWGS
= CJSWGS (T ) ⋅ 1 −
Vbs
C jbsswg
PBSWGS (T )
(10.2.8)
Cbd = Adeff C jbd + Pdeff C jbdsw + Weffcj ⋅ NF ⋅ C jbdswg
where Cjbd is the unit-area bottom D/B junciton capacitance, Cjbdsw is the
unit-length D/B junction sidewall capacitance along the isolation edge, and
Cjbdswg is the unit-length D/B junction sidewall capacitance along the gate
edge. The effective area and perimeters in (10.2.8) are given in Chapter 11.
Cjbd is calculated by
if Vbd < 0
(10.2.9)
− MJD
Vbd
C jbd = CJD(T ) ⋅ 1 −
PBD (T )
otherwise
(10.2.10)
Vbd
C jbd = CJD (T ) ⋅ 1 + MJD ⋅
PBD (T )
Cjbdsw is calculated by
if Vbd < 0
(10.2.11)
− MJSWD
C jbdsw = CJSWD(T ) ⋅ 1 −
Vbd
PBSWD (T )
otherwise
(10.2.12)
C jbdsw = CJSWD (T ) ⋅ 1 + MJSWD ⋅
Vbd
PBSWD (T )
Cjbdswg is calculated by
if Vbd < 0
(10.2.13)
− MJSWGD
= CJSWGD(T ) ⋅ 1 −
Vbd
C jbdswg
PBSWGD(T )
otherwise
(10.2.14)
C jbdswg = CJSWGD(T ) ⋅ 1 + MJSWGD ⋅
Vbd
PBSWGD(T )
DMCI of
Point contact, isolated
DMCG of
Wide contact, shared
DMDG
Ldrawn-XGL No contact, merged
XGW
DMCI of
Wide contact
Else
P seff computed from N F, DWJ, geoMod, DMCG, DMCI, DMDG,
DMCGT, and MIN.
In the above, Pseff and Aseff will be used to calculate junction diode IV and
CV. Pseff does not include the gate-edge perimeter.
If (rgeoMod == 0)
Source diffusion resistance Rsdiff is not generated.
Else if (number of source squares NRS is given)
Rsdiff = NRS ⋅ RSH
Else
Rsdiff computed from NF, DWJ, geoMod, DMCG, DMCI, DMDG,
DMCGT, RSH, and MIN.
(11.2.1)
Rgeltd =
(
RSHG ⋅ XGW + 3⋅ NGCON
W
effcj
)
NGCON ⋅ ( Ldrawn − XGL) ⋅ NF
Table 11-1 lists the options for source/drain connections through the model
selector geoMod.
geoMod End source End drain Note
0 isolated isolated NF=Odd
1 isolated shared NF=Odd, Even
2 shared isolated NF=Odd, Even
3 shared shared NF=Odd, Even
4 isolated merged NF=Odd
5 shared merged NF=Odd, Even
6 merged isolated NF=Odd
7 merged shared NF=Odd, Even
8 merged merged NF=Odd
9 sha/iso shared NF=Even
10 shared sha/iso NF=Even
Talbe 11-2 lists the options for source/drain contacts through the model
selector rgeoMod.
(12.1.1)
T
Vth (T ) = Vth (TNOM ) + KT1 +
KT1L
+ KT 2 ⋅Vbseff ⋅ − 1
Leff TNOM
(12.2.1)
U 0(T ) = U 0(TNOM ) ⋅ (T TNOM )
UTE
(12.2.2)
UA(T ) = UA(TNOM ) + UA1⋅ (T TNOM − 1)
(12.2.3)
UB (T ) = UB(TNOM ) + UB1 ⋅ (T TNOM − 1)
and
(12.2.4)
UC(T ) = UC (TNOM ) + UC1⋅ (T TNOM − 1)
(12.3.1)
VSAT (T ) = VSAT (TNOM ) − AT ⋅ (T TNOM − 1)
(12.4.1)
RDSW (T ) = RDSW (TNOM ) + PRT ⋅ (T TNOM − 1)
(12.4.2)
RDSWMIN (T ) = RDSWMIN (TNOM ) + PRT ⋅ (T TNOM − 1)
(12.4.3)
RDW (T ) = RDW (TNOM ) + PRT ⋅ (T TNOM − 1)
(12.4.4)
RDWMIN (T ) = RDWMIN (TNOM ) + PRT ⋅ (T TNOM − 1)
(12.4.5)
RSW (T ) = RSW (TNOM ) + PRT ⋅ (T TNOM − 1)
and
(12.4.6)
RSWMIN (T ) = RSWMIN (TNOM ) + PRT ⋅ (T TNOM − 1)
(12.5.1)
I sbs = Aseff J ss (T ) + Pseff J ssws (T ) + Weffcj ⋅ NF ⋅ J sswgs (T )
where
(12.5.2)
E g (TNOM ) Eg (T ) T
− + XTIS ⋅ ln
v (TNOM ) vt (T ) TNOM
J ss (T ) = JSS (TNOM ) ⋅ exp t
NJS
(12.5.3)
Eg (TNOM ) E g (T ) T
− + XTIS ⋅ ln
v (TNOM ) vt (T ) TNOM
J ssws (T ) = JSSWS (TNOM ) ⋅ exp t
NJS
and
(12.5.4)
Eg (TNOM ) E g (T ) T
− + XTIS ⋅ ln
v (TNOM ) vt (T ) TNOM
J sswgs (T ) = JSSWGS (TNOM ) ⋅ exp t
NJS
• Drain-side diode
The drain-side saturation current is given by
(12.5.5)
I sbd = Adeff J sd (T ) + Pdeff J sswd (T ) + Weffcj ⋅ NF ⋅ J sswgd (T )
where
(12.5.6)
Eg (TNOM ) Eg (T ) T
− + XTID ⋅ ln
v (TNOM ) vt (T ) TNOM
J sd (T ) = JSD(TNOM ) ⋅ exp t
NJD
(12.5.7)
Eg (TNOM ) Eg (T ) T
− + XTID ⋅ ln
v (TNOM ) vt (T ) TNOM
J sswd (T ) = JSSWD (TNOM ) ⋅ exp t
NJD
and
(12.5.8)
E g (TNOM ) Eg (T ) T
− + XTID ⋅ ln
v (TNOM ) vt (T ) TNOM
J sswgd (T ) = JSSWGD(TNOM ) ⋅ exp t
NJD
(12.6.1)
CJS (T ) = CJS (TNOM ) ⋅ [1 + TCJ ⋅ (T − TNOM )]
(12.6.2)
CJSWS (T ) = CJSWS (TNOM ) + TCJSW ⋅ (T − TNOM )
and
(12.6.3)
CJSWGS (T ) = CJSWGS (TNOM ) ⋅ [1 + TCJSWG ⋅ (T − TNOM )]
The temperature dependences of the built-in potentials on the source side are
modeled by
(12.6.4)
PBS (T ) = PBS (TNOM ) − TPB ⋅ (T − TNOM )
(12.6.5)
PBSWS (T ) = PBSWS (TNOM ) − TPBSW ⋅ (T − TNOM )
and
(12.6.6)
PBSWGS (T ) = PBSWGS (TNOM ) − TPBSWG ⋅ (T − TNOM )
• Drain-side diode
(12.6.7)
CJD(T ) = CJD(TNOM ) ⋅ [1 + TCJ ⋅ (T − TNOM )]
(12.6.8)
CJSWD (T ) = CJSWD (TNOM ) + TCJSW ⋅ (T − TNOM )
and
(12.6.9)
CJSWGD(T ) = CJSWGD(TNOM ) ⋅ [1 + TCJSWG ⋅ (T − TNOM )]
The temperature dependences of the built-in potentials on the drain side are
modeled by
(12.6.10)
PBD(T ) = PBD(TNOM ) − TPB ⋅ (T − TNOM )
(12.6.11)
PBSWD (T ) = PBSWD (TNOM ) − TPBSW ⋅ (T − TNOM )
and
(12.6.12)
PBSWGD(T ) = PBSWGD (TNOM ) − TPBSWG ⋅ (T − TNOM )
(12.7.1)
7.02 ×10 −4TNOM 2
Eg (TNOM ) = 1.16 −
TNOM + 1108
and
(12.7.2)
7.02 × 10− 4T 2
Eg (T ) = 1.16 −
T + 1108
(12.7.3)
device performance quite well. Values extracted in this manner will now have
some physical relevance.
It is suggested that BSIM4 use group device extraction strategy. This requires
measured data from devices with different geometries. All devices are measured
under the same bias conditions. The resulting fit might not be absolutely perfect
for any single device but will be better for the group of devices under
consideration. In the following, a general extraction methodology is proposed for
basic BSIM4 model parameters. Thus, it will not cover other model parameters,
such as those of the gate tunneling current model and RF models, etc.
13.3.1Extraction Requirements
One large size device and two sets of smaller-sized devices are needed to
extract parameters, as shown in Figure 13-1.
W Large W and L
Wmin
L
Lmin
threshold voltage VTH0, and the body effect coefficients K1 and K2 which
depend on the vertical doping concentration distribution. The set of devices
with a fixed large channel width but different channel lengths are used to
extract parameters which are related to the short channel effects. Similarly,
the set of devices with a fixed, long channel length but different channel
widths are used to extract parameters which are related to narrow width
effects. Regardless of device geometry, each device will have to be
measured under four, distinct bias conditions.
13.3.2Optimization
(13.3.1)
Initial Guess of
Parameters P i
Model Equations
∆P
i
+ ∆P
(m+1) (m)
P =P
i i i
∆P no
< δ
i
(m)
P
i
yes
STOP
To change (13.3.1) into a form that a linear least-squares fit routine can be
used (i.e. in a form of y = a + bx1 + cx 2), both sides of (13.3.1) are divided
by ∂fsim / ∂P1 . This gives the change in P1, ∆P1(m) , for the next iteration
such that:
(13.3.2)
( m + 1)
Pi = Pi( m ) + ∆ Pi( m )
where i=1, 2, 3 for this example. The (m+1) parameter values for P2 and P3
are obtained in an identical fashion. This process is repeated until the
incremental parameter change in parameter values ∆Pi (m) are smaller than
a pre-determined value. At this point, the parameters P1, P2 , and P3 have
been extracted.
13.3.3Extraction Routine
Step 1
Step 2
Step 3
LINT, Rds(RDSW, W, Vbs) One Set of Devices (Large and Fixed W &
Different L).
Fitting Target Exp. Data: Strong Inver- I ds vs. Vgs @ Vds = 0.05V at Different Vbs
sion region I ds(Vgs, Vbs )
Step 4
Step 5
Step 6
DVT0, DVT1, DVT2, LPE0, LPEB One Set of Devices (Large and Fixed W &
Different L).
Fitting Target Exp. Data: Vth(Vbs, L, W) Vth(Vbs, L, W)
Step 7
Step 8
Step 9
Step 10
Step 11
Step 12
Step 13
Step 14
Step 15
Step 16
PCLM, θ(DROUT, PDIBLC1, One Set of Devices (Large and Fixed W &
PDIBLC2, L), PVAG, FPROUT, DITS, Different L).
DITSL, DITSD I ds vs. Vds @ Vbs = 0V at Different Vgs
Fitting Target Exp. Data: Rout(Vgs, Vds )
Step 17
Step 18
Step 19
θ DIBL (ETA0, ETAB, DSUB, DVTP0, One Set of Devices (Large and Fixed W &
DVTP1, L) Different L).
I ds vs. Vgs @ Vds = Vdd at Different Vbs
Fitting Target Exp. Data: Subthreshold
region Ids(Vgs, Vbs)
Step 20
Step 21
Step 22
ALPHA0, ALPHA1, BETA0 One Set of Devices (Large and Fixed W &
Different L).
Fitting Target Exp. Data: Iii (Vgs, Vbs)/W I ds vs. Vds @ Vbs = Vbb at Different Vds
Parameter Default
name Description value Binnable? Note
RBODYMOD Substrate resistance network model 0 NA -
(Also an selector ( network
instance off)
parameter)
TRNQSMOD Transient NQS model selector 0 NA OFF
(Also an
instance
parameter)
ACNQSMOD AC small-signal NQS model selector 0 NA OFF
(Also an
instance
parameter)
FNOIMOD Flicker noise model selector 1 NA -
TNOIMOD Thermal noise model selector 0 NA -
DIOMOD Source/drain junction diode IV model 1 NA -
selector
PERMOD Whether PS/PD (when given) 1 (including NA -
includes the gate-edge perimeter the gate-
edge perime-
ter)
GEOMOD Geometry-dependent parasitics model 0 NA -
(Also an selector - specifying how the end S/D (isolated)
instance diffusions are connected
parameter)
RGEOMOD Source/drain diffusion resistance and 0 (no S/D NA -
(Instance contact model selector - specifying the diffusion
parameter end S/D contact type: point, wide or resistance)
only) merged, and how S/D parasitics resis-
tance is computed
GAMMA1 (γ1 Body-effect coefficient near the sur- calculated V 1/2 Note-1
in equation) face
Parameter Default
name Description value Binnable? Note
Parameter Default
name Description value Binnable? Note
Parameter Default
name Description value Binnable? Note
FPROUT Effect of pocket implant on Rout deg- 0.0V/m 0.5 Yes Not mod-
radation eled if
binned
FPROUT
not posi-
tive
Parameter Default
name Description value Binnable? Note
Parameter Default
name Description value Binnable? Note
NIGC Parameter for Igcs, I gcd ,Igs and Igd 1.0 Yes Fatal error
if binned
value not
positive
POXEDGE Factor for the gate oxide thickness in 1.0 Yes Fatal error
source/drain overlap regions if binned
value not
positive
Parameter Default
name Description value Binnable? Note
PIGCD Vds dependence of Igcs and I gcd 1.0 Yes Fatal error
if binned
value not
positive
TOXREF Nominal gate oxide thickness for gate 3.0e-9m No Fatal error
dielectric tunneling current model if not posi-
only tive
Parameter Default
name Description value Binnable? Note
ACDE Exponential coefficient for charge 1.0m/V Yes -
thickness in CAPMOD=2 for accumu-
lation and depletion regions
MOIN Coefficient for the gate-bias depen- 15.0 Yes -
dent surface potential
IJTHDREV
=IJTHSREV
IJTHSFWD Limiting current in forward bias IJTHSFWD No If not posi-
IJTHDFWD region =0.1A tive, reset
to 0.1A
IJTHDFWD
=IJTHS-
FWD
XJBVS Fitting parameter for diode break- XJBVS=1.0 No Note-8
XJBVD down
XJBVD
=XJBVS
BVS Breakdown voltage BVS=10.0V No If not posi-
BVD tive, reset
to 10.0V
BVD=BVS
JSS Bottom junction reverse saturation JSS= No -
JSD current density 1.0e-4A/m2
JSD=JSS
JSWS Isolation-edge sidewall reverse satura- JSWS No -
JSWD tion current density =0.0A/m
JSWD
=JSWS
Parameter
name
(separate for
source and drain
side as indicated Default
in the names) Description value Binnable? Note
JSWGS Gate-edge sidewall reverse saturation JSWGS No -
JSWGD current density =0.0A/m
JSWGD
=JSWGS
CJS Bottom junction capacitance per unit CJS=5.0e-4 No -
CJD area at zero bias F/m 2
CJD=CJS
MJS Bottom junction capacitance grating MJS=0.5 No -
MJD coefficient MJD=MJS
MJSWS Isolation-edge sidewall junction MJSWS No -
MJSWD capacitance grading coefficient =0.33
MJSWD
=MJSWS
CJSWS Isolation-edge sidewall junction CJSWS= No -
CJSWD capacitance per unit area 5.0e-10
F/m
CJSWD
=CJSWS
CJSWGS Gate-edge sidewall junction capaci- CJSWGS No -
CJSWGD tance per unit length =CJSWS
CJSWGD
=CJSWS
MJSWGS Gate-edge sidewall junction capaci- MJSWGS No -
MJSWGD tance grading coefficient =MJSWS
MJSWGD
=MJSWS
Parameter
name
(separate for
source and drain
side as indicated Default
in the names) Description value Binnable? Note
PB Bottom junction built-in potential PBS=1.0V No -
PBD=PBS
PBSWS Isolation-edge sidewall junction built- PBSWS No -
PBSWD in potential =1.0V
PBSWD
=PBSWS
PBSWGS Gate-edge sidewall junction built-in PBSWGS No -
PBSWGD potential =PBSWS
PBSWGD
=PBSWS
Default
Parameter Description value Binnable? Note
name
Default
Parameter Description value Binnable? Note
name
Parameter Default
name Descrition name Binnable? Note
Parameter Default
name Descrition name Binnable? Note
2qε si NDEP
γ1 =
Coxe
2 qε si NSUB
γ2 =
Coxe
γ 12Coxe 2
NDEP =
2qε si
qNDEP ⋅ XT 2
= Φ s − VBX
2ε si
K1 = γ 2 − 2K 2 Φ s − VBM
(γ 1 − γ 2 )( Φ s − VBX − Φ s )
K2=
2 Φs ( )
Φ s − VBM − Φ s + VBM
2 ⋅ EPSROX ⋅ ε
CF = ----------------------------------------0 ⋅ log 1 + 4.0e – 7
-------------------
π TOXE
Note-8:
[1] Y. C. King, H. Fujioka, S. Kamohara, K. Chen, and Chenming Hu, “DC electrical oxide thickness model
for quantization of the inversion layer in MOSFETs”, Semicond. Sci. Technol., vol. 13, pp. 963-966, 1998.
[2] Weidong Liu, Xiaodong Jin, Yachin King, and Chenming Hu, “An efficient and accurate compact model
for thin-oxide-MOSFET intrinsic capacitance considering the finite charge layer thickness”, IEEE Trans.
Electron Devices, vol. ED-46, May, 1999.
[3] Kanyu M. Cao, Weidong Liu, Xiaodong Jin, Karthik Vasanth, Keith Green, John Krick, Tom Vrotsos,
and Chenming Hu, “Modeling of pocket implanted MOSFETs for anomalous analog behavior,” Tech.
Dig. of IEDM , pp. 171-174, 1999.
[4] Z.H. Liu, C. Hu, J.H. Huang, T.Y. Chan, M.C. Jeng, P.K. Ko, and Y.C. Cheng,“Threshold Voltage
Model For Deep-Submicrometer MOSFETs,” IEEE Tran. Electron Devices, vol. 40, pp. 86-95, 1993.
[5] J.A. Greenfield and R.W. Dutton, “Nonplanar VLSI Device Analysis Using the Solution of Poisson's
Equation,” IEEE Trans. Electron Devices, vol. ED-27, p.1520, 1980.
[6] H. S. Lee, “An Analysis of the Threshold Voltage for Short-Channel IGFET's,” Solid-State Electronics,
vol.16, p.1407, 1973.
[7] Yuhua Cheng and Chenming Hu, “MOSFET Modeling & BSIM3 User’s Guide,” Kluwer Academic
Publishers, 1999.
[8] C. Hu, S. Tam, F.C. Hsu, P.K. Ko, T.Y. Chan and K.W. Kyle, "Hot-Electron Induced MOSFET
Degradation - Model, Monitor, Improvement," IEEE Trans. Electron Devices, vol. 32, pp. 375-385, 1985.
[9] T. Y. Chen, J. Chen, P. K. Ko, C. Hu, “The impact of gate-induced drain leakage current on MOSFET
scaling,” Tech. Digest of IEDM, pp. 718-721, 1987.
[11] Weidong Liu, Xiaodong Jin, J. Chen, M. Jeng, Z. Liu, Y. Cheng, K. Chen, M. Chan, K. Hui, J. Huang,
R. Tu, P. Ko, and Chenming Hu, “BSIM3v3.2 MOSFET Model and Users’ Manual,” http://www-
device.eecs.berkeley.edu/~bsim3.
[12] Xiaodong Jin, J-J Ou, C-H Chen, Weidong Liu, Paul Gray, and Chenming Hu, “An effective gate
resistance model for CMOS RF and noise modeling,” Tech. Dig. of IEDM, pp. 961-964, 1998.
[13] Mansun Chan, K. Hui, R. Neff, C. Hu, P. Ko, “A Relaxation time Approach to Model the Non-Quasi-
Static Transient Effects in MOSFETs,” Tech. Dig. of IEDM , pp. 169-172, 1994.
[14] K.K. Hung, P. Ko, C. Hu, and Y. C. Cheng, “A Physics-Based MOSFET Noise Model for Circuit
Simulators,” IEEE Trans. Electron Devices, vol. 37, no. 5, pp. 1323-1333, 1990.
[15] Weidong Liu, Kanyu M. Cao, Xiaodong Jin, and Chenming Hu, “BSIM4.0.0 Technical Notes,” http://
www-device.eecs.berkeley.edu/~bsim3/bsim4.html.