MAKING A “MODEL” - University of Kentucky College …elias/lectures/00_Modeling_Peddenpohl.pdf · WHAT IS A BSIM SPICE MODEL? ... MEAS. VTH IDS METAL THICK ILD THICK SILICON QUAL

MAKING A “MODEL”

BOB PEDDENPOHL MODELING MANAGERCYPRESS MODELING CENTERLEXINGTON, KY

2 Cypress Confidential

OUTLINE

WHAT DOES CY KENTUCKY DO?WHAT IS A BSIM SPICE MODEL?HOW TO MAKE A MOS SPICE MODEL?


DESIGN KIT MAKES MONEY

PRODUCT ($)DESIGN KIT (CAD, R&D)

MARKET NEEDS

PRODUCT SPECS

CIRCUIT SCHEMATIC

LAYOUT

CIRCUIT DESIGN

SPICERCX

MODELS

MEAS.VTHIDS

METAL THICKILD THICK

SILICON QUAL

SCHEMATICSLAYOUTS

DRCLVS

E-TEST MODULESTEST CHIP TAPEOUT

PRODUCT PLANS

PRE-SILICON WORK


DESIGN KIT MAKES MONEY

PRODUCT ($)DESIGN KIT (CAD, R&D)

MARKET NEEDS

PRODUCT SPECS

CIRCUIT SCHEMATIC

LAYOUT

CIRCUIT DESIGN

SPICERCX

MODELS

MEAS.VTHIDS

METAL THICKILD THICK

SILICON QUAL

SCHEMATICSLAYOUTS

DRCLVS

E-TEST MODULESTEST CHIP TAPEOUT

PRODUCT PLANS

PRE-SILICON WORK


OUTLINE



INTRODUCTION: MODELS

GENERIC DEFINITIONMAN MADE EXPRESSIONS TO REPRESENT MOTHER NATURE

VLSI DESIGN DEFINITIONMODELS = DESIGNERS PERCEPTION OF TECHNOLOGY

ENGINEERING DEFINITIONMODELS = PHYSICAL EQUATIONS + PARAMETERS

Ids = BETA (Vgs-VT)^2 where VT = 0.6 BETA = w/l*COX*MOBILITY = 1E-6


INTRODUCTION:TYPES OF MODELS

SIMULATION MODELS

TABLE LOOKUP

SIMULATORS ACCESS MEASURED DC/AC DATA IN A TABULAR FORM

ANALYTICAL (OR COMPACT)

ANALYTICAL OR COMPACT DEVICE MODELS BASED PRIMARILY ON DEVICE PHYSICS. FITTING PARAMETERS INTRODUCED TO IMPROVE ACCURACY

NUMERICAL

NUMERICAL SOLUTION OF DEVICE CHARACTERISTIC


SCHEMATICS USE BSIM COMPACT MODELS


INTRODUCTION: MODELS LIMITATIONS

IDEAL VS REALITYIDEAL DESIGN SIMULATIONS EXACTLY EQUAL SILICON MEASUREMENTS

REALITYMODEL NOT PERFECTMODEL HAS ACCURACY LIMITATIONSGOOD DESIGNER UNDERSTANDS MODEL LIMITATIONSNEED TO MODEL PROCESS VARIATIONSNEED MODELS QUICKLY TO ENABLE DESIGNERS


OUTLINE



WHAT MODELS USED AT UK?WHAT CY TECHNOLOGY DID YOU USE?

RAM7: Wmin/Lmin = 0.42/0.20um, Vcc=1.8V, Idrive = 9.99 mA

WHEN WAS TECHNOLOGY QUALIFIED?MODEL FROZEN Q302

WHAT TYPE OF MOSFETS?LV MOS (NSHORT/PSHORT), LVT PMOS (PLOWVT)CELL FETS (NPASS, NPD, PPU)

WHAT’S NSHORT ELECTRICAL TOX? JUNCTION DEPTH?TOX= 41 A, XJ = 0.1um


MODEL DEVELOPMENT PROCESS

MEASUREMENT (DC, AC, TRAN)

EXTRACT WAFER CASE MODEL

RO MEAS = RO SIMS

CENTER TO EDR NOMINAL (TT)

SELECT “GOLDEN” WAFER

SKEW MODELS (FF, SS, FS, SF)

QA & RELEASE TO DESIGN



IDEAL: MODELING SILICON CLOSE TO NOMINALREALITY: ~400+ PARAMETERS, ONLY MOST IMPORTANT ON TARGET

NOMINALMIN MAX

WAFER





RO MEAS = RO SIMS






MEASUREMENTS: HARDWARE & SOFTWARE


MEASUREMENTS: COMPLETE MOS

FET DC (VTH0, RDSW)FET AC (CGDO,DLC)DIODE DC (JS,JSW)DIODE AC (CJ, CJSW)


MEASUREMENTS: FET DC


MEASUREMENTS: FET DC

MODEL NEEDS SCALE WITHIN ALL GEOMETRY, TEMP


Threshold Voltage vs Length

0.3

0.4

0.5

0.6

0.1 1 10 100

Length (L) in microns

VTH Vth

MEASUREMENTS: DC FET QA, VTH VS. L

MODEL ACCURACY <=> MEASUREMENT ACCURACYCONDENSED DATA TRENDS

Strong Halo , L dependenceStrong Halo , L dependence

Halo with SCEHalo with SCE

Normal SCENormal SCE


MEASUREMENTS: DC FET QA, VTH VS. W

MODEL ACCURACY <=> MEASUREMENT ACCURACYCONDENSED DATA TRENDS

Threshold Voltage vs Width

0.3

0.4

0.5

0.6

0.1 1 10 100

Width (W) in microns

VTH Vth

LOCOS (+k3)

STI (-k3)


MEASUREMENTS: FET AC

-2.0 -1.2 -0.4 0.4 1.2 2.0Vgate (V)

11.0

14.14

17.28

20.42

23.56

26.7

Cg

g (

pF

)

NCGG_GC.CV W/L=16800.00/0.15 T=25C

OXTC Α∝ε

( ))(vXTC

DEPOX+Α

∝ε


0.0 0.4 0.8 1.2 1.6 2.0Vbias (V)

4.0

5.02

6.04

7.06

8.08

9.1

Max.Err%= Rms Err%=0.22

MEASUREMENTS: DIODE DC/AC

-5.0 -3.8 -2.6 -1.4 -0.2 1.0V (V)

1.0e-11

1.0e-10

1.0e-9

1.0e-8

1.0e-7

1.0e-6

1.0e-5

1.0e-4

1.0e-3

1.0e-2


REVERSE BIAS DC CHARACTERISTIC

I_FORWARD ~mA

I_Reverse ~ pA

REVERSE BIAS AC CHAR.= f(CJA, CJP, EX,)


MEASUREMENTS: TRANSIENT

RING OSCILLATOR VALIDATION OF MODEL

ININO/PO/P

( )CTINTERCONNEFETCTINTERCONNE

d

CCRWHERE

NDELAYRO

⊕×∝

××=

τ

τ2_

C9R10

TECH DESCRIPTION DELAY (PS/STAGE) SPECC9 143 stages WP/WN=4/2 FanOut=1 55.7 55.0

R10 143 stages WP/WN=4/2 FanOut=1 31.0 31.0





RO MEAS = RO SIMS






WAFER CASE: DC MOS EXTRACTION

MODEL = EQUATIONS + PARAMETERS

EQUATIONS (BSIM3V3) + MODEL PARAMETERS = WAFER CASE MODEL

Threshold Model

Mobility ModelDrive Current

Channel Length

Modulation

Short Channel Effects


WAFER CASE: MOS MODEL BINNING

Long/Wide Long/Wide Constant Constant VtVt

Narrow Width Effects Narrow Width Effects (STI/LOCOS)(STI/LOCOS)

Sho

rt C

hann

el E

ffect

s (H

ALO

/DIB

L)


WAFER CASE: AC FET + DIODE

MODEL EXTRACTION

MODEL = EQUATIONS + PARAMETERS

EQUATIONS (BSIM3V3) + PARAMETERS (EXTRACTED FROM MEASUREMENTS) = MODEL (WAFER CASE)

Accumulation

Inversion

Intrinsic Cap for Analog Design

BSIM3 Limitation

0.0 0.4 0.8 1.2 1.6 2.0Vbias (V)

4.0

5.02

6.04

7.06

8.08

9.1


-5.0 -3.8 -2.6 -1.4 -0.2 1.0V (V)

1.0e-11

1.0e-10

1.0e-9

1.0e-8

1.0e-7

1.0e-6

1.0e-5

1.0e-4

1.0e-3

1.0e-2

Max.Err%= Rms Err%=25.4 4.58

MOS DIODE IV MODEL

MOS DIODE CV MODEL

MOSFET CV MODEL





RO MEAS = RO SIMS






RO CAL: LAYOUT EXTRACTED SIMULATION

VALIDATE CAD EXTRACTION RULES + MOS BSIM MODELS

ININO/PO/P

( )CTINTERCONNEFETCTINTERCONNE

d

CCRWHERE

NDELAYRO

⊕×∝

××=

τ

τ2_

R10

LAYOUT (DESIGN DEP.) LAYOUT MODEL: (ILD, METAL THICK)

C9

CALIBRE RCX

CIRCUIT: FET DELAY + Rinterconnect + Cinterconnect

SPICE MODELS

RO SIMS = RO MEAS





RO MEAS = RO SIMS






CORNER MODELS

WAFER CASE SIMULATIONS = WAFER MEASUREMENTSWHAT ABOUT PROCESS VARIATIONS?WILL MY DESIGN YIELD?

NOMINALMIN MAX

WAFER


CORNER MODELS

REALITYEVERY SITE/WAFER/LOT/SPLIT IS DIFFERENT ( PROCESS VARIATIONS)

WORKING WITH REALITYCORNERS: MODELING SPACE TO COVER ALLPOSSIBILITIES (STATISTICALLY) IN PROCESS

TEAM EFFORT TO GET GOOD YIELDFAB: +/-4 SIGMA E-TEST à 99.99% WAFERS INSIDE MIN/MAX MODELING: MIN/MAX MODELS MATCH FAB LIMITSDESIGN: SIMULATE DESIGN WORKING AT MIN/MAX LIMITSALL 3 GROUPS WORKING = GOOD PRODUCT YIELD

NOMINALMIN MAXtt.cor

ss.cor

wafer.cor

ff.cor


WHY 5 MOS CORNERS?

VTs AT SS & FF = 70% SPEC RANGEVTs AT FS/SF = 100% SPEC RANGE

0.660.670.680.690.7

0.710.720.730.740.750.760.770.780.790.8

0.810.820.830.840.850.860.870.88

vtxn

s15

-1.05 -1.03 -1.01 -0.99 -0.97 -0.95 -0.93 -0.91 -0.89 -0.87 -0.85 -0.83vtxps15

fs

sf

7

8

9

10

11

12

idsn

s15

2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 3 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 4 4.1 4.2 4.3 4.4 4.5 4.6 4.7idsps15

ff

ss

VTXNS15 vs. VTXPS15 (V) (Vth @ W/L=25/0.15um)

IDSNS15 vs. IDSPS15 (mA)Idrive (Vgs=Vds=Vcc)W/L=25/0.15um

ss

ff

fs

sf

tt

tt


WHY CORNER METHODOLOGY IMPORTANT

MODEL MUST MATCH DESIGN/FAB AGREED LIMITS

FAB WANTS WIDE MIN/MAX LIMITSSTATISTICAL PROCESS CONTROL (SPC)HOW GOOD DOES A PROCESS RUN WITHIN IT’S NOM/MIN/MAX

DESIGN WANTS NARROW MIN/MAX LIMITSEASIER TO DESIGNSMALL PROCESS VARIATION à SMALLER SI AREA





RO MEAS = RO SIMS






QA: MODEL DOCUMENTATION

MODEL SUMMARY TABLE

MODEL ACCURACY IN SUB-THRESHOLD, GM ACCURACY


APPENDIX

BOB PEDDENPOHL (PED)CYPRESS MODELING CENTER


Applying the Corner Models

Design

Interconnect R

tres, fres, sres

Interconnect C

tpar, fpar, spar

r+c.mod

Interconnects/Passives

trtc, hrlc, lrhc

FET Corners

tt, ff, ss, sf, fs

CellFET Corners

ttcell, ffcell, sscell

Temp coef of R

C for various line/space

Npass

Nlatch

Platch

Nmos/Pmos

Nthick/Pthick (HV)

Diode

PNP

metal/contact/poly/diff

Sheet resistances

MAKING A “MODEL” - University of Kentucky College …elias/lectures/00_Modeling_Peddenpohl.pdf · WHAT IS A BSIM SPICE MODEL? ... MEAS. VTH IDS METAL THICK ILD THICK SILICON QUAL

Documents