mosfet modeling in spicevada.skku.ac.kr/classinfo/vlsicad/lecture-ohp/mos3bsi… ·  ·...

UC Berkeley EE241 Andrei Vladimirescu

MOSFET Modeling in SPICE

Andrei Vladimirescu

UC Berkeley EE241 Andrei Vladimirescu

MOSFET Device

Mname nd ng ns nb Modname <<L=>L> <<W=>W>+ <AD=AD> <AS=AS> <PD=PD> <PS=PS>+ <NRD=NRD> <NRS=NRS>

<OFF><IC=vgs0,vds0,vbs0>

UC Berkeley EE241 Andrei Vladimirescu

MOS Level=1 DC

l Parameters: VTO, KP, GAMMA, PHI, LAMBDA

IDS =

0 for VGS ≤VTH

KP2

WLe ff

VGS − VTH( )2 1 + LAM BDA ⋅VDS( ) for 0 < VGS − VTH ≤VDS (3.29 )

KP2

WLe ff

VDS 2 VGS − VTH( )− VDS( )1+ LAM BDA ⋅VDS( ) for 0 < VDS < VGS − VTH

VTH = VTO + GAM M A PH I − VBS − PH I( )

UC Berkeley EE241 Andrei Vladimirescu

MOS1 IDS Characteristics

UC Berkeley EE241 Andrei Vladimirescu

Large-Signal Model

UC Berkeley EE241 Andrei Vladimirescu

Dynamic Model

l Gate-Oxide Charges: TOX, CGSO, CGDO, CGBO

ê If TOX specified CGS, CGD and CGB represent Qch=f(V) below G- CGSO, CGDO and CGBO model just overlap of G over diff/bulk

l D and S Junction: CBD, CBS, P, MJ

Cox = εoxε0

TOXCGDO = CGSO = 1

2 CoxLCGBO = CoxW

CBD = CBD1 − VBD PB( )M J

CBS = CBS1 − VBS PB( )M J

LEVEL=1 w/o TOX

UC Berkeley EE241 Andrei Vladimirescu

Sidewall Junction Capacitance

UC Berkeley EE241 Andrei Vladimirescu

Dynamic CG-V Model

UC Berkeley EE241 Andrei Vladimirescu

Small-Signal Model

gds = 1rds

= dIDS

dVDS

gm = dIDS

dVGS

gm bs = dIDS

dVBS

UC Berkeley EE241 Andrei Vladimirescu

.MODEL Parameters MOS1

l .MODEL Modname NMOS/PMOS <VTO=VTO...>

UC Berkeley EE241 Andrei Vladimirescu

Second-Order Effects in MOS3

l Bulk-Charge Contributionl Small-Size Effectsl Subthreshold Conductionl Limited Carrier Velocity Saturation

UC Berkeley EE241 Andrei Vladimirescu

MOS Level=3 DC

IDS = β VGS − VTH − 1 + FB

2VDS

VDS

β = µ e ffCox

WLe ff

µ s = UO1 + TH ETA VGS − VTH( )

µ e ff = µ s

1 + µ s

VM AX ⋅Le ff

VDS

VTH = VFB + PH I − σVDS + γFS PH I − VBS + FN PH I − VBS( )

σ = ETA8.15⋅10 − 22

CoxLe ff3

UC Berkeley EE241 Andrei Vladimirescu

MOS3 Saturation

l Velocity Saturation: VMAX, KAPPA

VDSAT = VGS − VTH

1 + FB

+VMAX ⋅Le ff

µs

− VGS − VTH

1 + FB

2

+VMAX ⋅Le ff

µ s

2

∆L= X d

EPX d

2

2

+ KAPPA VDS − VDSAT( )− EPXd

2

EP = IDSAT

GDSATLe ff

UC Berkeley EE241 Andrei Vladimirescu

MOS3 Subthreshold

l NFS

VON = VTH + nk Tq

n = 1 + Cfs

Cox

+ Cd

Cox

Cfs = q ⋅NFS

Cd = ∂QB

∂VBS

= − γSd

dVBS

PH I − VBS − ∂γS

∂VBS

PH I − VBS + DELTA πεsi

4CoxW

Cox

UC Berkeley EE241 Andrei Vladimirescu

MOS3 SubVT Characteristics

UC Berkeley EE241 Andrei Vladimirescu

Temperature Model

UC Berkeley EE241 Andrei Vladimirescu

Noise Model

l Resistive Channel and Flicker: TOX, KF, AF

ids2 = 8k Tgm

3∆f + KF⋅IDS

AF

fCoxLe ff2 ∆f

UC Berkeley EE241 Andrei Vladimirescu

Advanced MOSFET Modelsfor ICs

Andrei Vladimirescu

UC Berkeley EE241 Andrei Vladimirescu

Physical Effects in Level 2,3

l Short and Narrow-channel effectsl Mobility reduction due to electrical fieldl Bulk-charge effectl Channel-length modulationl Subthreshold Conductionl Carrier velocity saturationl Parasitic Drain and Source resistance

UC Berkeley EE241 Andrei Vladimirescu

Physical Effects Needed

l Non-uniform channel dopingl Drain-induced barrier loweringl Substrate-current-induced body effectl Temperature effectsl Poly-gate depletion effectl Velocity overshoot

UC Berkeley EE241 Andrei Vladimirescu

Model Requirements

l Analytical» Continuity of IDS=f(VDS,VGS,VBS) and its first derivatives» Parameter/Formulation choice for accuracy, scalability

l Simulation» Computationally efficient (time and #iterations)» Solid convergence» Robust (no singularities)» Charge representation for transient

UC Berkeley EE241 Andrei Vladimirescu

BSIM Models

l Based on Bell Labs CSIM (1981)l BSIM1 (1984) - L>1µ, Tox>150Al BSIM2 (1990) - BSIM for deep-submicronl BSIM3 (1993)

UC Berkeley EE241 Andrei Vladimirescu

BSIM References

l ___, BSIM3v3 Manual (Final Version), Univ. of California, Berkeley, 1995.l S.Liu and L.W.Nagel, Small-Signal MOSFET Models for Analog Circuit Design,

IEEE JSSC, Vol. SC-17, no. 6, pp. 983-998, Dec. 1982.l B.J.Sheu, D.L. Scharfetter and H.C. Poon, Compact Short-Channel IGFET

Model (CSIM), ERL Memo M84/20, Univ. of California, Berkeley, Mar. 1984.l B.J.Sheu, D.L.Scharfetter and P.K.Ko, SPICE2 Implementation of BSIM, ERL

Memo M85/42, Univ. of California, Berkeley, May 1985.l M.C.Jeng, Design and modeling of Deep-Submicron MOSFETs, ERL Memo

M90/90, Univ. of California, Berkeley, Oct 1990.l BTA Technology, Inc., BSIMPro and Presentations on MOSFET Modeling,

Santa Clara, California.

UC Berkeley EE241 Andrei Vladimirescu

Parameter Philosophy

l For each process parameter P there is» length correction PL» width correction PW

l Three Model Parameters for each Effect» P0, PL, PW, e.g., VFB, LVFB, WVFB

P= P0 + PL

Li − DL+ PW

W i − DW

UC Berkeley EE241 Andrei Vladimirescu

BSIM3(v3)

l Single I-V formulation for IDS, Rout» from subthreshold to strong inversion» from saturation to linear

l W dependence of QB and Rds

l Improved scalability» ∆L and ∆W dependency on L and W

l Improved Capacitance model for small-sizel Non-quasi static relaxation model

UC Berkeley EE241 Andrei Vladimirescu

Deep Subµ MOSCharacteristic

UC Berkeley EE241 Andrei Vladimirescu

BSIM3 VTH

l Vertical and Lateral Nonuniform Doping» K1, K2

VTH = VTide al + K1 φs − VBS − φs( )+ K 2VBS + K 1 1 + NLX

Le ff

φs

UC Berkeley EE241 Andrei Vladimirescu

BSIM3 VTH (Cont’d)

l Short, Narrow Channel

VTH = VTide al + ∆VTdoping + (K 3 + K 3BVBS ) tox

W e ff + W 0

φs

− DVTOW e − DVT1w W e ff Le ff 2ltw + 2e − DVT1w W e ffLe ff ltw( )VBI − φs( )−− DVTO e − DVT1Le ff 2lt + 2e − DVT1w Le ff lt( )VBI − φs( )−− e − D sub Le ff 2lt0 + 2e − D sub Le ff lt0( )η0 + ηBVBS( )VDS

UC Berkeley EE241 Andrei Vladimirescu

BSIM3 VTH = f(L)

UC Berkeley EE241 Andrei Vladimirescu

BSIM3 Poly-Gate Effect

UC Berkeley EE241 Andrei Vladimirescu

BSIM3 Poly-Gate Effect on IDS

UC Berkeley EE241 Andrei Vladimirescu

BSIM3 Mobility

UC Berkeley EE241 Andrei Vladimirescu

BSIM3 IDS

UC Berkeley EE241 Andrei Vladimirescu

BSIM3 Unified IDS

UC Berkeley EE241 Andrei Vladimirescu

Vdseff Function

UC Berkeley EE241 Andrei Vladimirescu

BSIM3 NQS

l QS model ignores finite time for channel charge build-upl Elmore equivalent of channel charge retains lowest freq pole

UC Berkeley EE241 Andrei Vladimirescu

BSIM3 Characteristics

UC Berkeley EE241 Andrei Vladimirescu

BSIM3 Summary

l Continuous Model for wide range of W and Ll Major physical mechanisms of subm devicesl New narrow-width modell Introduction of Non-Quasi-Static behaviourl Superior scaling and statistical modelingl Charge conservingl Computationally efficient

UC Berkeley EE241 Andrei Vladimirescu

MOS Modeling Trends

l Single-formulation Models» BSIM3, MOS9, EKV, ...

l Standard Models (Public Domain)l Good fit for large- and small-signal characteristics

(function and derivatives)l Scalability with device dimensionsl Support for statistics