Dual Damascene Etch; · M. Hussein-- Plasma Etch Users Group, 5/13/04 Effect of CO 2 Addition 0 20 40 60 80 100 120 140 0 100 200 300 CO2 Flow Rate (sccm) E t c h R a te (n m / m

History95.1

®®

Dual Damascene Etch; Dual Damascene Etch; Challenges at the 65nm Challenges at the 65nm

Node and BeyondNode and Beyond

Makarem HusseinMakarem HusseinPortland Technology DevelopmentPortland Technology Development

Intel CorporationIntel Corporationemail: [email protected]

History95.2

®® 2M. Hussein-- Plasma Etch Users Group, 5/13/04

OutlineOutline

Integration Scheme Integration Scheme About the About the ““environmentenvironment””Selectivity during etch stop layer etchSelectivity during etch stop layer etchConclusions

History95.3


OutlineOutline

Integration Scheme Integration Scheme About the About the About the “““environmentenvironmentenvironment”””Selectivity during etch stop layer etchSelectivity during etch stop layer etchSelectivity during etch stop layer etchConclusionsConclusionsConclusions

History95.4


ViaVia--First PatterningFirst Patterning

2. Trench Patterning

SLAM

ILD

PR

CU

1. Via Patterning 3. Etch-stop layeretch

History95.5


Post Trench EtchPost Trench Etch

SLAM fills the vias and protect the ESL during trench SLAM fills the vias and protect the ESL during trench etchetch No ES selectivity needsNo ES selectivity needsSLAM etch characteristics SLAM etch characteristics Control via profileControl via profile

ILDSLAM resist

SLAM

ILD

History95.6


Role of EtchRole of Etch--StopStop

Mx

Vx

Mx+1

ILD

Landed Via

ES

Un-Landed Via

Copper diffusion barrierCopper diffusion barrierProtect copper during processingProtect copper during processingA layer on which via etch terminatesA layer on which via etch terminates

History95.7


Proven TechnologyProven Technology

M7M7

66

5544

3322

11

Intel’s 90nm TechnologyIntel’s 130nm Technology

66

55

4433

2211

History95.8


OutlineOutline

Integration SchemeIntegration SchemeIntegration SchemeAbout the About the ““environmentenvironment””Selectivity during etch stop layer etchSelectivity during etch stop layer etchSelectivity during etch stop layer etchConclusionsConclusionsConclusions

History95.9


Scaling at Work!Scaling at Work!

Forces ofForces ofScalingScaling

Tighter pitch& smaller CD

Better RC delay

•Advanced Litho•Advanced etch•Advanced cleans

•Lower-k ILD•Advanced etch stop•Advanced package

Processingvs.

ImpactOn

k-value

Scaling MUST PrevailScaling MUST Prevail

History95.10


Always ~30% SmallerAlways ~30% Smaller

050

100150200250300350400450

130 90 65 45 32

Technology Node (nm)

M2

Pitc

h(nm

)

193nm Litho248nm Litho

History95.11


Quest for Shorter RCQuest for Shorter RC--DelayDelay

S. Thompson et.al, IEDM’02

1mm wire1mm wire

Shorter delay Shorter delay (1) Low resistivity metal(1) Low resistivity metal(2) Low(2) Low--k ILDk ILD

History95.12


Conductor? Anyone?!Conductor? Anyone?!Copper becomes more resistive as Copper becomes more resistive as metal line width decreases below metal line width decreases below 200nm200nmLiner is occupying a Liner is occupying a ““significantsignificant””portion of the metal wire areaportion of the metal wire area

History95.13


What Does LowWhat Does Low--k Mean? k Mean?

0

10

20

30

40

50

60

70

80

0

5

10

15

20

2 2.5 3 3.5 4

Silicon oxide based dielectric materials

Mod

ulus

(GPa

) Hardness (G

Pa)

Dielectric constant, k0

10

20

30

40

50

60

70

80

0

5

10

15

20

2 2.5 3 3.5 4

Silicon oxide based dielectric materials

Mod

ulus

(GPa

) Hardness (G

Pa)

Dielectric constant, k

LowLow--k ILDs possess a lower dielectric constant k ILDs possess a lower dielectric constant andandweak mechanical propertiesweak mechanical propertiesWeak mechanical propertiesWeak mechanical properties

Negatively impact etch and cleansNegatively impact etch and cleansLimit packaging choicesLimit packaging choices

History95.14


Processing Vs. kProcessing Vs. k-- Value Value DegradationDegradation

LowLow--k ILD has a k ILD has a ““memorymemory””!!AshingAshing post via and trench etch:post via and trench etch:

Impacts the out come of the dual Impacts the out come of the dual damascene patterningdamascene patterningMay affect kMay affect k--valuevalue

No impact to No impact to ILDILD’’ss kk--value adds to etch complexityvalue adds to etch complexity

History95.15


HarderHarder Patterning with ArF Patterning with ArF LithoLitho

Topof trench

Bottomof trench

via

space

ArF resist

ArF resist is needed for ArF resist is needed for λλ=193nm lithography=193nm lithographyArF resist ArF resist ““agglomeratesagglomerates”” under plasma conditionsunder plasma conditionsEtch performance of ArF resist is poorEtch performance of ArF resist is poor

Before etchBefore etch Post trench etch & cleansPost trench etch & cleans

History95.16


HarderHarder Patterning with ArF Patterning with ArF LithoLitho

ArF Patterning KrF Patterning

ILD ILD

SAME Via etch process. SAME cleans.

Etching under ArF resist is more susceptible Etching under ArF resist is more susceptible to generating deep striations in the ILDto generating deep striations in the ILDBalancing selectivity to etchBalancing selectivity to etch--stop and stop and maintaining profile and CD is getting hardermaintaining profile and CD is getting harder

History95.17


More Challenging Cleans More Challenging Cleans

Cleans are influenced by:Cleans are influenced by:Nature of the ILDNature of the ILDDry etch chemistry / strategyDry etch chemistry / strategy

Dealing with lowDealing with low--k ILD and ArF resist k ILD and ArF resist significantly complicates the cleans significantly complicates the cleans

History95.18


OutlineOutline

Integration SchemeIntegration SchemeIntegration SchemeAbout the About the About the “““environmentenvironmentenvironment”””Selectivity during etch stop layer etchSelectivity during etch stop layer etchConclusionsConclusionsConclusions

History95.19


Have Enough Selectivity?Have Enough Selectivity?

((--)193nm)193nm resist etch resistanceresist etch resistance(-) Reduction of resist thicknessReduction of resist thickness

(-) thkns does NOT scale ARC

resist

ILD

Copper(-) thickness

reduction

Etch-sensitive

History95.20


More Selectivity DemandsMore Selectivity Demands

ESLPrior Metal

ILDESLPrior Metal

ILD

EtchEtch--stop layer (ESL) etch is performed with stop layer (ESL) etch is performed with dual damascene pattern as MASKdual damascene pattern as MASKRequires etch rate of ESL >> Etch rate of ILDRequires etch rate of ESL >> Etch rate of ILDMinimum to erosion of exposed edgesMinimum to erosion of exposed edges

History95.21


Experimental SetExperimental Set--upup

ILD

ESL

ILD:ILD: LowLow--k, carbonk, carbon--doped oxidedoped oxideESL:ESL: Silicon carbideSilicon carbide

Substrate: 300mm WaferSubstrate: 300mm Wafer

400 KHzRF

µ-wave 2.45 GHz

Plasma

α = UndercutSiC etched

History95.22


Experimental DesignExperimental Design

Design:Design: Three factors, central Three factors, central composite with oncomposite with on--face axial face axial pointspointsResponse:Response: ILD ER, ESL ER, ILD ER, ESL ER, underunder--cut in ESLcut in ESL

-- ++CF4 (CF4 (sccmsccm)) 5050 200200CO2 (CO2 (sccmsccm)) 00 200200Pressure (Pa)Pressure (Pa) 11 33

One die

isolated

nested

larger

trench

ILD ER: OpticalILD ER: OpticalESL ER: XESL ER: X--sectionsection

History95.23


Selectivity: InteractionsSelectivity: InteractionsSe

lect

ivity

Sele

ctiv

ityPressure

CO

2 FlowC

F4 Flow

High pressure and High COHigh pressure and High CO22 flow enhance selectivityflow enhance selectivity

History95.24


Effect of COEffect of CO22 AdditionAddition

020

4060

80100

120140

0 100 200 300CO2 Flow Rate (sccm)

Etch

Rat

e (n

m/m

in)

0.0

0.5

1.0

1.5

2.0

2.5

3.0

Sele

ctiv

ity (S

iC:C

DO

)

SiCSiC

CDOCDO

SelSel

SiC ER is more responsive to oxygen addition than CDOSiC ER is more responsive to oxygen addition than CDO

1Pa / 125 CF4 / 10W RF1Pa / 125 CF4 / 10W RF

History95.25


UnderUnder--Cut: InteractionsCut: InteractionsN

orm

aliz

ed U

nder

cut

Nor

mal

ized

Und

ercu

tPressure

CO

2 FlowC

F4 Flow

High pressure and COHigh pressure and CO22 flow enhance undercutflow enhance undercut

History95.26


General Trends: FlowsGeneral Trends: Flows

0

1

2

3

4

0 100 200CF4 Flow (sccm)

Sele

ctiv

ity

-0.15

0.00

0.15

0.30

0.45

Und

ercu

t

UC

Sel

100 CO2100 CO20

1

2

3

4

0 100 200 300

CO2 Flow (sccm)

Sele

ctiv

ity

-0.15

0.00

0.15

0.30

0.45

Und

ercu

t

UC

Sel

125 CF4125 CF4

Addition of CO2 improves selectivity at the expenseAddition of CO2 improves selectivity at the expenseOf undercutOf undercut

History95.27


General Trends: RF & PresGeneral Trends: RF & Pres

0

1

2

3

4

0 10 20 30 40

RF Bias (W)

Sele

ctiv

ity

-0.15

0.00

0.15

0.30

0.45

Und

ercu

t

Sel UC

0

1

2

3

4

1 2 3 4

Pressure (Pa)

Sele

ctiv

ity

-0.15

0.00

0.15

0.30

0.45

Und

ercu

tUCSel

Selectivity improves at higher pressureSelectivity improves at higher pressureUndercut increases with RF biasUndercut increases with RF bias

History95.28


Model: Selectivity & UCModel: Selectivity & UC

High PressureHigh Pressure

Add CO2Add CO2

More radicalsMore radicals

Enhances SiC EREnhances SiC ER

Possibly more free F

Possibly more free F

Suppress CDO ERSuppress CDO ER

Less CFxLess CFx

Less wall passivation Less wall passivation undercutundercut CDO ERCDO ER

History95.29


On The Role Of RFOn The Role Of RF……Observations:Observations:

Undercut increases with RF Undercut increases with RF Selectivity degrades with RF Selectivity degrades with RF ER increases with RF biasER increases with RF biasNo threshold for ER of either No threshold for ER of either SiC or CDOSiC or CDO 0

1

2

3

4

0 100 200 300 400 500RF Bias Vpp(volts)

Sele

ctiv

ity

3Pa / 200 CF4 / 200 CO23Pa / 200 CF4 / 200 CO2

Conclusions:Conclusions:SiC etching exhibits appreciable chemical behaviorSiC etching exhibits appreciable chemical behaviorPassivation, rather than ion bombardment, is the Passivation, rather than ion bombardment, is the source of anisotropysource of anisotropy

History95.30


ConclusionsConclusionsScaling will prevail, making patterning more Scaling will prevail, making patterning more challenging than ever.challenging than ever.In a downstream high density etch system, In a downstream high density etch system, it is difficult to attain high selectivity without it is difficult to attain high selectivity without undercut between CDO ILD and SiC ESL.undercut between CDO ILD and SiC ESL.Selectivity is becoming a serious challenge Selectivity is becoming a serious challenge and basic understanding is much needed:and basic understanding is much needed:

Impact of plasma on ArF resistImpact of plasma on ArF resistSelectivity modulationSelectivity modulation

History95.31


AcknowledgementAcknowledgementData and discussions with the following Data and discussions with the following colleagues significantly contributed to this colleagues significantly contributed to this work:work:

Dry Etch:Dry Etch: Satyarth Satyarth SuriSuri, Max Heckscher, and , Max Heckscher, and James JeongJames JeongWet Etch:Wet Etch: Lourdes Dominguez, Lana Jong, and Lourdes Dominguez, Lana Jong, and Mike NashnerMike NashnerIntegration:Integration: PhucahnPhucahn Nguyen, and Peter MoonNguyen, and Peter MoonQuality & Reliability:Quality & Reliability: Jun He and Barbra MinerJun He and Barbra Miner

History95.32

®®

BackBack--upup

History95.33


020406080

100120140160

0.5 1.5 2.5 3.5Pressure (Pa)

ER (n

m/m

in)

10W RF10W RF

CDO

SiC

History95.34


0

1

2

3

4

1 2 3 4

Pressure (Pa)

Sele

ctiv

ity

-0.15

0.00

0.15

0.30

0.45

Und

ercu

tUCSel

Documents

Dual Damascene Etch; · M. Hussein-- Plasma Etch Users Group, 5/13/04 Effect of CO 2 Addition 0 20 40 60 80 100 120 140 0 100 200 300 CO2 Flow Rate (sccm) E t c h R a te (n m / m