Cmos design

transcript

CMOS Design

Introduction

Integrated circuits: many transistors on one chip.Very Large Scale Integration (VLSI): very many

Metal Oxide Semiconductor (MOS) transistorFast, cheap, low-power transistorsComplementary: mixture of n- and p-type leads to less

power How to build your own simple CMOS chip

CMOS transistorsBuilding logic gates from transistorsTransistor layout and fabrication

Silicon Lattice

Transistors are built on a silicon substrateSilicon is a Group IV materialForms crystal lattice with bonds to four neighbors

Si SiSi

Dopants

Silicon is a semiconductorPure silicon has no free carriers and conducts poorlyAdding dopants increases the conductivityGroup V: extra electron (n-type)Group III: missing electron, called hole (p-type)

As SiSi

Si SiSi

B SiSi

Si SiSi

p-n Junctions

A junction between p-type and n-type semiconductor forms a diode.

Current flows only in one direction

p-type n-type

anode cathode

NMOS Transistor

Four terminals: gate, source, drain, bodyGate – oxide – body stack looks like a capacitor

Gate and body are conductorsSiO2 (oxide) is a very good insulatorCalled metal – oxide – semiconductor (MOS) capacitorEven though gate is

no longer made of metal

GateSource Drain

bulk Si

Polysilicon

NMOS Operation

Body is commonly tied to ground (0 V)When the gate is at a low voltage:

P-type body is at low voltageSource-body and drain-body diodes are OFFNo current flows, transistor is OFF

GateSource Drain

bulk Si

Polysilicon

Body is commonly tied to ground (0 V)When the gate is at a low voltage:

P-type body is at low voltageSource-body and drain-body diodes are OFFNo current flows, transistor is OFF

GateSource Drain

bulk Si

Polysilicon

NMOS Operation

PMOS Transistor

Similar, but doping and voltages reversedBody tied to high voltage (VDD)Gate low: transistor ONGate high: transistor OFFBubble indicates inverted behavior

GateSource Drain

bulk Si

Polysilicon

Power Supply Voltage

GND = 0 VIn 1980’s, VDD = 5V

VDD has decreased in modern processes

High VDD would damage modern tiny transistors

Lower VDD saves power

VDD = 3.3, 2.5, 1.8, 1.5, 1.2, 1.0, …

Transistors as Switches

We can view MOS transistors as electrically controlled switches

Voltage at gate controls path from source to drain

OFF ON

ON OFF

CMOS Inverter

GNDA Y

A=1 Y=0

CMOS Inverter

A=0 Y=1

CMOS Inverter

CMOS NAND Gate

OFF OFF

CMOS NOR Gate

3-input NAND Gate

Y pulls low if ALL inputs are 1Y pulls high if ANY input is 0

CMOS Transistor Theory

Outline

IntroductionMOS CapacitornMOS I-V CharacteristicspMOS I-V CharacteristicsGate and Diffusion Capacitance

Introduction

So far, we have treated transistors as ideal switchesAn ON transistor passes a finite amount of current

Depends on terminal voltagesDerive current-voltage (I-V) relationships

Transistor gate, source, drain all have capacitanceI = C (DV/Dt) -> Dt = (C/I) DVCapacitance and current determine speed

Also explore what a “degraded level” really means

MOS Capacitor

Gate and body form MOS capacitorOperating modes

AccumulationDepletionInversion

polysilicon gate

silicon dioxide insulator

p-type body+-

Vg < 0

0 < Vg < Vt

depletion region

Vg > Vt

depletion regioninversion region

Terminal Voltages

Mode of operation depends on Vg, Vd, Vs

Vgs = Vg – Vs

Vgd = Vg – Vd

Vds = Vd – Vs = Vgs - Vgd

Source and drain are symmetric diffusion terminalsBy convention, source is terminal at lower voltageHence Vds 0

nMOS body is grounded. First assume source is 0 .Three regions of operation

CutoffLinearSaturation

nMOS Cutoff

No channelIds = 0

Vgs = 0

p-type body

nMOS Linear

Channel formsCurrent flows from d to s

e- from s to d

Ids increases with Vds

Similar to linear resistor

Vgs > Vt

Vgd = Vgs

Vgs > Vt

Vgs > Vgd > Vt

Vds = 0

0 < Vds < Vgs-Vt

p-type body

s dIds

nMOS Saturation

Channel pinches offIds independent of Vds

We say current saturatesSimilar to current source

Vgs > Vt

Vgd < Vt

Vds > Vgs-Vt

p-type body

s d Ids

I-V Characteristics

In Linear region, Ids depends onHow much charge is in the channel?How fast is the charge moving?

Channel Charge

MOS structure looks like parallel plate capacitor while operating in inversionGate – oxide – channel

Qchannel =

p-type body

source

-drain

channel-

p-type body

SiO2 gate oxide(good insulator, ox = 3.9)

polysilicongate

Channel Charge

MOS structure looks like parallel plate capacitor while operating in inversionGate – oxide – channel

Qchannel = CV

C = Cg = eoxWL/tox = CoxWL

V = Vgc – Vt = (Vgs – Vds/2) – Vt

p-type body

source

-drain

channel-

p-type body

polysilicongate

Carrier velocity

Charge is carried by e-Carrier velocity v proportional to lateral E-field

between source and drainv = mE m called mobilityE = Vds/L

Time for carrier to cross channel:t = L / v

nMOS Linear I-V

Now we knowHow much charge Qchannel is in the channelHow much time t each carrier takes to cross

channel

dsgs t ds

tW VC V V VL

VV V V

ox = W

nMOS Saturation I-V

If Vgd < Vt, channel pinches off near drainWhen Vds > Vdsat = Vgs – Vt

Now drain voltage no longer increases current

dsatds gs t dsat

VI V V V

nMOS I-V Summary

Shockley 1st order transistor models

cutoff

linear

saturatio

dsds gs t ds ds dsat

gs t ds dsat

VI V V V V V

V V V V

pMOS I-V

All dopings and voltages are inverted for pMOSMobility mp is determined by holes

Typically 2-3x lower than that of electrons mn

120 cm2/V*s in AMI 0.6 mm processThus pMOS must be wider to provide same current

In this class, assume mn / mp = 2

Capacitance

Any two conductors separated by an insulator have capacitance

Gate to channel capacitor is very importantCreates channel charge necessary for operation

Source and drain have capacitance to bodyAcross reverse-biased diodesCalled diffusion capacitance because it is associated with

source/drain diffusion

Gate Capacitance

Approximate channel as connected to sourceCgs = eoxWL/tox = CoxWL = CpermicronW

Cpermicron is typically about 2 fF/mm

p-type body

polysilicongate

Diffusion Capacitance

Csb, Cdb

Undesirable, called parasitic capacitanceCapacitance depends on area and perimeter

Use small diffusion nodesComparable to Cg

for contacted diff½ Cg for uncontactedVaries with process

CMOS Fabrication

Overview of CMOSCMOS Fabrication Process OverviewCMOS Fabrication ProcessProblems with Current CMOS FabricationFuture Changes in CMOS Fabrication

Outline

Complementary metal–oxide–semiconductor (CMOS) Has many different uses:

Integrated Circuits Data converters Integrated transceivers Image sensors Logic circuits

What is CMOS?

NAND Circuit

What is CMOS?

CMOS Fabrication Process Overview

1. Create a pattern.2. Oxidize small layer,

about 1µm thick.3. Place photoresist on

top of SiO2

4. Place mask(pattern) above photoresist and expose it to UV light.

CMOS Fabrication Process

1. Etch away SiO2 using HF acid or plasma.

2. Remove remaining photoresist with acids.

To create a n well: Diffusion

Heat wafer in Arsenic gas chamber until diffusion occurs. Ion Implantation

Arsenic or phosphorous are implanted in window.

n well

A thin layer of oxide is deposited.

A thin layer of polysilicon is deposited using Chemical Vapor Deposition (CVD) .

p substraten well

Remove oxide layer using acid.

Dope open area using Ion implantation or diffusion.

p substraten well

n+n+ n+p+p+p+

p substrate

Thick field oxide

n well

n+n+ n+p+p+p+

Problems with Current CMOS Fabrication

Optical lithography is limited by the light frequency.

Material limitationsYield limitationsSpace limitations

Future Changes in CMOS Fabrication

Material changes like using high-k materials.

Design changesSOI(Silicon On Insulator)Double Gate (Finfet)Twin-Tub Process

DC Characteristics of a CMOS Inveter

DC Characteristics of a CMOS Inverter

A complementary CMOS inverter consists of a p-type and an n-type device connected in series.

The DC transfer characteristics of the inverter are a function of the output voltage (Vout) with respect to the input voltage (Vin).

The MOS device first order Shockley equations describing the transistors in cut-off, linear and saturation modes can be used to generate the transfer characteristics of a CMOS inverter.

Plotting these equations for both the n- and p-type devices produces the traces below.

IV Curves for nMOS

PMOS IV Curves

The DC transfer characteristic curve is determined by plotting the common points of Vgs intersection after taking the absolute value of the p-device IV curves, reflecting them about the x-axis and superimposing them on the n-device IV curves.

We basically solve for Vin(n-type) = Vin(p-type) and Ids(n-type)=Ids(p-type)

The desired switching point must be designed to be 50 % of magnitude of the supply voltage i.e. VDD/2.

Analysis of the superimposed n-type and p-type IV curves results in five regions in which the inverter operates.

Region A occurs when 0 leqVin leq Vt(n-type).

The n-device is in cut-off (Idsn =0).p-device is in linear region,Idsn = 0 therefore -Idsp = 0Vdsp = Vout – VDD, but Vdsp =0 leading to an output of Vout = VDD.

Region B occurs when the condition Vtn leq Vin le VDD/2 is met.

Here p-device is in its non-saturated region Vds neq 0.n-device is in saturation

Saturation current Idsn is obtained by setting Vgs = Vin resulting in the equation:

CMOS Inverter DC Characteristics

CMOS Inverter Transfer Characteristics

In region B Idsp is governed by voltages Vgs and Vds described by:

: thatRecall

DDoutDDouttpDDinptnin

dspdsn

DDoutDDouttpDDinpdsp

VVVVVVVVV

VVVVVVVI

DDoutdsDDings VVVVVV and

Region C has that both n- and p-devices are in saturation.

Saturation currents for the two devices are:

tnintninn

DDtpintpDDinp

VVVVVVI

Region D is defined by the inequality

p-device is in saturation while n-device is in its non-saturation region.

Equating the drain currents allows us to solve for Vout. (See supplemental notes for algebraic manipulations).

tpDDinDD VVV

tninout

outtninndsn

DDtpintpDDinp

VVVVVVI

CMOS Inverter Static Charateristics

In Region E the input condition satisfies:

The p-type device is in cut-off: Idsp=0

The n-type device is in linear modeVgsp = Vin –VDD and this is a more positive value compared

to Vtp.

Vout = 0

tpDDin VVV

nMOS & pMOS Operating points

Vtp Vtn

0VDD/2 VDD+Vtp VDD

Both in satnMOS in satpMOS in sat

Vout =Vin-Vtp

Vout =Vin-Vtn

Circuits & Layout

Outline

CMOS Gate DesignPass TransistorsCMOS Latches & Flip-FlopsStandard Cell LayoutsStick Diagrams

CMOS Gate Design

Activity:Sketch a 4-input CMOS NOR gate

Complementary CMOS

Complementary CMOS logic gatesnMOS pull-down networkpMOS pull-up networka.k.a. static CMOS

pMOSpull-upnetwork

outputinputs

nMOSpull-downnetwork

Pull-up OFF Pull-up ON

Pull-down OFF Z (float) 1

Pull-down ON 0 X (crowbar)

Series and Parallel

nMOS: 1 = ONpMOS: 0 = ONSeries: both must be ONParallel: either can be ON

OFF OFF OFF ON

ON OFF OFF OFF

g1 g2 0 0

OFF ON ON ON

(d) ON ON ON OFF

11 0 1

Conduction Complement

Complementary CMOS gates always produce 0 or 1Ex: NAND gate

Series nMOS: Y=0 when both inputs are 1Thus Y=1 when either input is 0Requires parallel pMOS

Rule of Conduction ComplementsPull-up network is complement of pull-downParallel -> series, series -> parallel

Compound Gates

Compound gates can do any inverting functionEx:

A B C DA B

Example: O3AI

Y = ((A+B+C).D)’

Signal Strength

Strength of signalHow close it approximates ideal voltage source

VDD and GND rails are strongest 1 and 0nMOS pass strong 0

But degraded or weak 1pMOS pass strong 1

But degraded or weak 0Thus nMOS are best for pull-down network

Pass Transistors

Transistors can be used as switches

Pass Transistors

0 strong 0

Input Output

1 degraded 1

0 degraded 0

Input Output

strong 1

Transistors can be used as switches

Transmission Gates

Pass transistors produce degraded outputsTransmission gates pass both 0 and 1 well

g = 0, gb = 1

g = 1, gb = 0

0 strong 0

Input Output

1 strong 1

g = 1, gb = 0

Tristates

EN A Y

Tristate buffer produces Z when not enabled

Nonrestoring Tristate

Transmission gate acts as tristate bufferOnly two transistorsBut nonrestoring

Noise on A is passed on to Y

Tristate Inverter

Tristate inverter produces restored outputViolates conduction complement ruleBecause we want a Z output

EN = 0Y = 'Z'

EN = 1Y = A

Tristate inverter produces restored outputViolates conduction complement ruleBecause we want a Z output

Tristate Inverter

Multiplexers

2:1 multiplexer chooses between two inputs

S D1 D0 Y

2:1 multiplexer chooses between two inputs

S D1 D0 Y

0 X 0 0

0 X 1 1

1 0 X 0

1 1 X 1

Multiplexers

Gate-Level Mux Design

How many transistors are needed? 20

1 0 (too many transistors)Y SD SD

Transmission Gate Mux

Nonrestoring mux uses two transmission gatesOnly 4 transistors

Inverting Mux

Inverting multiplexerUse compound AOI22Or pair of tristate invertersEssentially the same thing

Non inverting multiplexer adds an inverter

4:1 Multiplexer

4:1 mux chooses one of 4 inputs using two selectsTwo levels of 2:1 muxesOr four tristates

S1S0 S1S0 S1S0 S1S0

D Latch

When CLK = 1, latch is transparentD flows through to Q like a buffer

When CLK = 0, the latch is opaqueQ holds its old value independent of D

a.k.a. transparent latch or level-sensitive latch

Multiplexer chooses D or old Q

CLKCLK

D Latch

D Latch Design

Circuits and LayoutSlide 90

Multiplexer chooses D or old Q

CLKCLK

D Latch Operation

CLK = 1

CLK = 0

D Flip-flop

When CLK rises, D is copied to QAt all other times, Q holds its valuea.k.a. positive edge-triggered flip-flop, master-slave

flip-flop

D Flip-flop Design

Built from master and slave D latches

CLKCLK

D Flip-flop Operation

CLK = 1

CLK = 0

Race Condition

Back-to-back flops can malfunction from clock skewSecond flip-flop fires lateSees first flip-flop change and captures its resultCalled hold-time failure or race condition

Non overlapping Clocks

Non overlapping clocks can prevent racesAs long as non overlap exceeds clock skew

We will use them in this class for safe designIndustry manages skew more carefully instead

Stick Diagram & Scalable Stick Diagram & Scalable Design RulesDesign Rules

IC Layout Concept

Stick Diagram Design Rules Layout Verification

Basic Concept

Based on the view point of IC layout, the stick diagram can help us understand the circuit function and its geometrical location relative to other circuit blocks.

Legend:

contactmetal 2metal 1

polyndiffpdiff

Although the stick diagram is an abstract presentation of real layout, it can use graphical symbols or legend to allocate the circuit to 2-diomensional plane and reach the aim same as the physical layout does.

The stick diagram is similar to a backbone of the real layout but without the real size and aspect ratio of the devices, it still can reflect the real condition to layout of the silicon chip.

Basic Concept

Notations of the stick diagram

Intermediate representation between the transistor level and the mask (layout) level.

Gives topological information (identifies different layers and their relationship)

Assumes that wires have no width. It is possible

to translate stick diagram automatically to layout with correct design rules.

Stick Diagram

When the same material (on the same layer) touch or cross, they are connected and belong to the same electrical node.

When polysilicon crosses N or P diffusion, an N or P transistor is formed. Polysilicon is drawn on top of diffusion.Diffusion must be drawn connecting the source and the

drain.Gate is automatically self-aligned during fabrication.

When a metal line needs to be connected to one of the other three conductors, a contact cut (via) is required.

Stick Diagram

Manhattan geometrical rule: When we use only vertical and horizontal lines In orthogonal to describe circuitry.

Boston geometrical rule: The stick diagram also allows curves to describe circuitry.

In order to describe N/PMOS more completely, to add n-well 、P+ select 、 well contact and substrate contact are optional for 4-terminal notation.

Stick Diagram

Conclusion

Stick diagram is a draft of real layout, it serves as an abstract view between the schematic and layout.

Stick diagram uses different lines, colors and geometrical shapes to present circuit nodes, devices, and their relative location.

Stick diagram doesn’t include information about the accurate coordinates and sizes of device, the length and width of conductors and the real size of well region.

CMOS Inverter Stick Diagrams

Basic layout

․ More area efficient layout

CMOS inverter described in other way.

CMOS Inverter Stick Diagrams

The transmission gate Circuit schematic Stick diagram

CMOS Transmission Gate

CMOS Stick DiagramsNAND/NOR

CMOS Stick Diagrams NAND

< Exercise 1 >

To draw the following circuitry by using a stick diagram

CMOS Stick Diagrams NOR

Lambda-based Design Rules

Lambda design rules are based on a reference metric λthat has units of um.

All widths, spacing and distances are written in the form Value = m λ

Where m is scaling multiplier.

<e.g.> λ= 1um w = 2 λ=2um

s = 3λ=3um

Lambda-based Design RulesLambda based design: half of technology since 1985. As technologychanges with smaller dimensions, a simple change in the value of can be used to produce a new mask set.

All device mask dimensions are based on multiples of , e.g., polysilicon minimum width = 2. Minimum metal to metal spacing = 3

Active Contact and Surround Rule

Example of surrounded design rule

Potential Problem - Misalignment

Misalignment –induced defect

Potential Problem – Short between Source and Drain

Example of an extended (gate over hang)design rule

Design Rule (0)

Due to the photo resolution, concentration, temperature and reaction time of the chemical reagents, the layout should tolerate some errors caused by process environment.

In order to avoid the influence from process variation, the layout of the circuit schematics should follow the design Rule。

The purpose of design rules

Interface between designer and process engineerGuidelines for constructing process masksUnit dimension: Minimum line width

scalable design rules: lambda parameterabsolute dimensions (micron rules)

Design Rule (1)

Layout rules are used for preparing the masks for fabrication.Fabrication processes have inherent limitations in accuracy.Design rules specify geometry of masks to optimize yield and

reliability (trade-offs: area, yield, reliability).Three major rules:

Wire width: Minimum dimension associated with a given feature.

Wire separation: Allowable separation.Contact: overlap rules.

Two major approaches:“Micron” rules: stated at micron resolution. rules: simplified micron rules with limited scaling

attributes. may be viewed as the size of minimum feature.Design rules represents a tolerance which insures very high

probability of correct fabrication (not a hard boundary between correct and incorrect fabrication).

Design rules are determined by experience.

Design Rule (2)

Terminology & Definition

Min. Width : The min. width of the line (layer) <Example> Wpoly(min.) = 0.5um

Min. Space : The min. spacing between lines with same material

<Example> Spoly-poly(min.) = 0.5um

<Min. Extension : The min. extension over different layers

<Example> Poly-gate extension over diffusion area = 0.55um

Min. Overlap : The overlap between different layers

<Example> Poly1 overlap Poly2 min. = 0.7um

Max. area of the specific region. <Example> Bonding Pad Area, max. =

100um x 100um

Conventional Layer Definition

Polysilicon

Metal1

Metal2

Contact To Poly

Contact To Diffusion

Well (p,n)

Active Area (n+,p+)

Color Representation

Yellow

Magenta

Select (p+,n+) Green

Polysilicon

Metal1

Metal2

Contact To Poly

Well (p,n)

Active Area (n+,p+)

Yellow

Magenta

Polysilicon

Metal1

Metal2

Contact To Poly

Well (p,n)

Active Area (n+,p+)

Yellow

Magenta

IntraIntra--Layer Design RulesLayer Design Rules

Metal24

Active3

Polysilicon

Different PotentialSame Potential

Metal13

Contactor Via

Select

SCMOS Design Rules

Metal toPoly ContactMetal to

Active Contact

SCMOS Design Rules

WellSubstrate

Select3

SCMOS Design Rules

MOSIS Layout Design Rules

3 basic design rules: Wire widthWire separationContact rule

Layout Verification

DRC – Design Rule CheckERC – Electrical Rule CheckLVS – Layout Versus SchematicLPE – Layout Parameter Extraction

Layout Verification

DRC(Design Rule Check)： => To check the min. line width and spacing

based on the design rules.

ERC(Electrical Rule Check)： => To check the short circuit between Power

and Ground, or check the floating node or devices.

LVS(Layout versus Schematic)： => To verify the consistency between Schematic and

Layout. For example ： to check the amount of transistor numbers, sizes of W/L.

LPE or PEX(Layout Parameter Extraction)： => From the database of layout, to extract the devices

with parasitics including effective W/L, parasitic capacitances and series resistance. The extracted file is in SPICE format and can be used for Post-Layout Simulation。

Layout Verification

Simulations

Pre-Layout Simulation - before layout workPost-Layout Simulation – after layout work, post layout simulation will reflect more realistic circuit performance.

Layout Verification

The complete design environment of Fill-Custom Design Design database – Cadence Design Framework IICircuit Editor – Text editor/Schematic editor (S-edit, Composer)Circuit Simulator – SPICE,TSPICE, HSPICELayout Editor – Cadence Virtuoso, Laker, L-editLayout Verification Diva, Dracula, Calibre, Hercules

Layout Verification

Concluding Remarks

Milestones technology in silicon eraTransistor Integrated Circuits CMOS

TechnologyKey weapons in SOC era

Design AutomationDesign Reuse

Breakthrough techniques in design automationSimulation (e.g., SPICE, Verilog-XL, etc.)Automatic Placement and Routing (APR)Logic Synthesis (e.g., Design Compiler)Formal VerificationTest Pattern Generation

It is EDA that pushes the IC design technology forward !

Cmos design

Documents