1

Growth and Properties of Ultrathin Metal Films for ULSI Interconnects

John G. EkerdtDepartment of Chemical Engineering

The University of Texas at AustinAustin, Texas

Collaborators: Yangming Sun and J. M. White

Students: Darren Gay, Jinhong Shin, Qi Wang

Sponsored by: Welch Foundation, SRC and Sematech Advanced Materials Research Center.

2005 ULSI Metrology Conference

Cu-Diffusion Barriers for ULSI Interconnect

2

Diffusion and electromigration resistanceGood adhesionLow electrical resistivity (~100 µΩ•cm)Good step coverage

A barrier layer (generally refractory metals and metal nitrides) must be employed to separate Cu from physical contact with other interconnect materials.

ULSI Interconnect

Metal Via Barrier Requirements

Production year

2010 2013 2016

MPU 1/2 pitch (nm)

45 35 22

Barrier thickness (nm)

5 3.5 2.5

ManufacturableSolutions:

[1] International Technology Roadmap for Semiconductors 2003 Update, http://pubic.itrs.net/, (2003).Known

NOT Known[1]

Motivation for Ru Barriers

3

• Challenges on future diffusion barriers– Prevent Cu diffusion at a thickness of only 3~5 nm– Adhere well to ILD layer and to Cu– Seedless Cu plating in high aspect ratio of via holes

• A composite barrier structure is under study.– A Ta film serving as the primary diffusion barrier– A Ru film in contact with Cu

• Ru is a noble metal and Cu is insoluble in it

• Ru oxide is also conductive and has a favorable reduction potential

• Ru is expected to improve wetting and adhesion properties between Ta and Cu

• Ru could enable direct copper plating on barrier surface without first coating a Cu seed layer.

ULK

Cu Ta

Ru

In-situ Film Deposition and Characterization System

4

Ultra-thin film deposition systems

Physical vapor deposition (PVD)

Chemical vapor deposition (CVD)

Atomic layer deposition (ALD)

Barrier characterization systems

Real time XPS and ISS analysis with CO2 laser annealing

Electrical measurement

Annealing facilities

CO2 infrared laser

Halogen lamp

In-situ sample transfer system

CVDE-test

ISSXPS

ALD

L/L

CO2 Laser

PVDL/L

Chemical Vapor Deposition of Ruthenium Using Ruthenium Carbonyl

5

Ruthenium carbonyl [Ru3(CO)12] is a solid precursor.It is stable in air and moisture at room temperature.It begins to evaporate at ~80 oC.It decomposes at 150 oC.

Ru

RuRu

OC

OC

OC

CO

CO

CO

OC

OC

OC

OC

OC

CO

Why ruthenium carbonyl?• A pure Ru film with minimal carbon and oxygen residues can be deposited!• The compound can be decomposed as low as 150 oC.• No reactive gas is needed! (Substrate can be protected!)

But wait – the carbonyl gives poor step coverage so it is not feasible formanufacturing. It does show Ru’s potential.

Q. Wang, et al., APL 84 (2004) 1380.

Low Temperature Thermal CVD Ru on Ta

6

• A 150 nm Ta film was deposited on the Si substrate using PVD.• A 6 nm Ru film was deposited on Ta surface without any reactive gas at

the temperature as low as 150 oC.• XP spectra indicated a pure Ru film with low C and O contents (<1 %)

was deposited at this low temperature.

70x103

60

50

40

30

20

10

Inte

nsity

(a. u

.)

1000 800 600 400 200 0Binding Energy (eV)

Ta 4fTa 4d

O 1s

Ru 3dRu 3p

As-deposited Ta on Si

~6 nm Ru deposited on Ta

25

20

15

10

5

x103

295 290 285 280

Ru 3d5/2

Ru 3d3/2 or C 1s

545 540 535 530 525

O 1s

Binding Energy (eV)

Si(100)150 nm Ta6 nm Ru

Q. Wang, et al., APL 84 (2004) 1380.

Ultrathin Ru Film Roughness on SiO2

7

Si(100)15 nm SiO2

3 nm Ru

• A 3 nm Ru film was deposited on the SiO2 substrate.• The Ru film roughness was ~1.4 nm, measured by AFM.• The SiO2 substrate roughness was 0.2 nm.

8

Ru Film Properties –Microstructure of 30 min film

Top-down view Tilted view

• Polycrystalline and columnar structure• Average grain size was ~ 20nm.• Grains are granular Reveals limitations of Ru-carbonyl precursor; need a different precursor to increase the nucleation density.• XRD shows crystalline & hexagonal structure of the film.

200

150

100

50

0

605550454035

2θ (degree)

Ru (100)

Ru (101)Ru (002)

XRD

Si(100)SiO2

500 nm Ru

Ru Film Properties – ISS/XPS

9

• LEISS Surface coverage • Substrate Si peak is attenuated by overlaying Ru film. The Si peak intensity is function of film thickness and mean free path. I/I0 = exp (-d/λcosθ)• Calculated minimum thickness of continuous Ru film ~ 3nm

ISS

EHe+

E/E0 = f(m2/m1)

E0

Si(100)1 5 nm SiO2

Ru

Si Peak (Substrate)Ru Peak

XPSλ = 16.71 Å

X-ray e-

I / I0 = f( λ / t )

Si(100)15 nm SiO2

10

Film thickness measurement

XPS Attenuation

22.5nm

• Min. thickness of continuous Ru film was ~ 3nm.• However, SEM/TEM shows a much thicker film; ~25nm for the 30-min sample.• The error seems to be caused by film roughness or film discontinuity. ISS overestimates surface coverage due to a shadowing effect, and XPS can detect the substrate Si peak due to surface roughness or film discontinuity.

TEM

xSEM

Ru Film Properties – Electrical Test

11

Barrier electrical test structure

Cu/Ru/Ta MOS capacitor was built, and the flat band voltage shift of a C-V curve (∆VFB) was used to characterize barrier effectiveness against Cu diffusion.

Samples need to be prepared carefully to obtain reproducible C-V curves.– Annealing for 90 min at 350 oC in high vacuum to neutralize interface trapped charges.– In situ deposition of Cu dots on the top of Ru dots with shadow mask.– To minimize device damages caused by sputter deposition, two-step Cu deposition was applied.– A 20 nm Cu film was first deposited using 10 W DC power.– A 200 nm Cu film was then deposited using 50 W DC power.–Subsequent anneal for 60 min at 350 ºC in 110 mTorr H2/N2 forming gas (after ambient exposure) and test ex situ

Ta

Aln-Si

SiO2 ~15 nm

Cu

Ru

C/I

1.0

0.8

0.6

0.4

0.2

-3 -2 -1 0 1 2 3

As-deposited2.5 nm Ru after annealing4 nm Ru after annealing

2.5nm / 4nm Ru Barrier(XPS Thickness Basis)

Challenges of Characterizing Ultra-thin Films

12

Immediate Challenges:• Description of the thin films

and the true barrier• Ion and electron

spectroscopies best suited for flat surfaces

E1/E0=f(m1/m0)

m1

He+E0 E1m0

Analyz

erIon Gun

E1/E0=f(m1/m0)

m1

He+E0 E1m0

Analyz

erIon Gun

Shadowing effect of ISS- Effect of ion gun & analyzer angles

Shadowing Effect - ISS

13

δθ

ab

c

g=α·d

d

Sub.Ru

θ: Ion gun angleδ: Detector angled: Pitcht: grain heightg: Grain sizeα: Surface coverage

• Assumption: Ru grains are equally sized and spaced.• Fraction of sub. peak : Isub/Itot = b/(a+b+c) • Isub ∝ b, since Itot /(a+b+c) = const.• b = (1- α)·d – (1/tanθ+1/tanδ)·t · sinθ• Minimum α that sub. peak disappears: 1/αmin = 1 + (1/tanθ+1/tanδ)·(t/g)

t

Shadowing Effect - ISS

14

• Grain height & size ratio was assumed to be one, d=g.• As angles decrease, αmin decrease.

Sub. peak disappears with lower surface coverage. • Shadowing effect will be gone with θ=δ=90º.• With θ=δ=60º and d=g, substrate peak disappears with only 47% of surface coverage.

Shadowing in XPS

15

δθ

ab

c

dSub.

Ru t

ℓg

θ: x-ray incident angleδ: Detector angle (takeoff angle)t: Heightg: Grain sized: Space between grainsℓ: Area that blocks the photoelectron

• Assumption: X-rays can penetrate Ru and reach the substrate.Ru film is discontinuous and thick enough to completely attenuate Si peak under Ru.

• dmin = ℓ: Minimum space that detector can see Si peak.• With 60º of takeoff angle and 20 nm of Ru film, dmin = 11.5 nm.

16

Why is there a Si XPS Signal for the 30-min film?

dmin = 11.5nm

g = 20nm d = 40nm 30nm 10nm 2nm5nm 1nm20nm

No Si peakSi peak from substrate

SiRu

Si 2p predicts ~ 3 nm for 30 min

30 min22.5nm

TEM

SEM

Avg grain size ~ 20 nm

CVD Film Continuity versus Thickness

17

XPSISS

X-ray e-

I / I0 = f( λ / t )

Si(100)15 nm SiO2

30

25

20

15

10

5

0

Inte

nsity

(a.u

.)

1.00.80.60.40.2

O SiRu

0 nm

2.0 nm

2.5 nm

1.7 nm

E/E0

115 110 105 100 95Binding Energy (eV)

Si 2p peak attenuation

0 nm

1.7 nm

2.0 nm

2.5 nmE

He+

E/E0 = f(m2/m1)

E0

Si(100)1 5 nm SiO2

Ru

(a) 0 min

(b) 10 min

(c) 20 min

(d) 30 min

Comparison of PVD and CVD Film Structures

18

SEM Images of Ru

~3.5 nm PVD (quartz balance and XPS) Ru ~3.5 nm CVD - XPS0.105 nm rms

1.443 nm rms∆0.53 nm (line scan)∆5.3 and 7.9 nm (line scan)

XPS and ISS measurements for PVD Ru filmThe thickness for fully covering Si substrate is between 1.2 to 2.2 nm

19

80x103

60

40

20

0

Inte

nsity

1200 1000 800 600 400 200 0Binding Energy (eV)

15 sec Ru deposition

30 sec Ru deposition

45 sec Ru deposition

C

C

C

C

O

O

O

O

Si

Si

Si

Si

Ru

Ru

Ru

-800

-700

-600

-500

-400

-300

-200

Inte

nsity

1.0 0.8 0.6 0.4 0.2E/E0

O

O

O

O

Si

Si

Si

Si

Ru

Ru

Ru

0s

15s

30s

45s

1.2nm Ru

2.2nm Ru

3.2nm Ru

20

Determining Film Growth Mechanism by Combined Use of ISS and XPS

(XPS)

Shape suggests island growthShape

suggests layer growth

Jiménez et al., Appl. Surf. Sci. 141 (1999) 186.

Yubero et al., Surf. Sci. 457 (2000) 24.

A oSi

A iSiA oRu

A iRu

A oRu

A iRu

i

,,

,,

,,

+

=Θ

21

Energy Dispersive Spectroscopy (EDS) Approach

The detection depth of EDS is much larger than XPS, so it is notsensitive to the surface morphology

Si

RuVolume (more than

100 nanometer deep) analyzed

E-beam (10 kV)

X-ray

22

EDS Measurement for PVD and CVD Ru Films

Ru Film EDS counts

3.5 nm PVD Film 81.6

12 nm PVD Film 284.4

3.5 nm CVD Film 201.6

CVD Film thickness = 7.8 to 8.6 nm

XPS Depth Profile for CVD Ru film (~ 4 nm thick)

23

100

80

60

40

20

0

Atom

ic C

once

ntra

tion

160140120100806040200Sputtering Time, sec

SiRu

CVDSi CVDRu

XPS Depth Profile for 3.5 nm PVD Ru Film

24

100

80

60

40

20

0

Ato

mic

Con

cent

ratio

n

100806040200Sputtering Time, sec.

25

Comparison of Depth Profile for CVD and PVD Films

100

80

60

40

20

0

A.C

160140120100806040200 Sputtering Time, sec

SiRu

CVDSi CVDRu PVDRu PVDSi

If we assume after the line is the interface, then the thickness of CVD film is 7.8 nm, calculated from from the sputtering time

26

Comparison of PVD and CVD Film StructuresEquivalent Thickness is ~(2-3 ×) or ~ (8 ×)

~3.5 nm PVD (quartz balance and XPS) Ru

~3.5 nm CVD - XPS

0.105 nm rms

1.443 nm rms

∆0.53 nm (line scan)

∆5.3 and 7.9 nm (line scan)

EDS of 3.5 and 11 nm PVD films and the “3.5” nm CVD film suggests a 7.8 to 8.6 nm CVD film

XPS depth profiling suggests the “3.5”nm CVD film is 7.8 nm

Understanding the True Barrier

27

• Barrier likely to be much thinner than the physical thickness of the metal film and just when this continuous layer is formed remains an experimental challenge.

• More extensive studies with 2D (PVD) films needed to establish the barrier properties and interfaces that form with the substrate layer.

• New precursors and growth processes (ALD) under study are leading to smoother films than the Ru3(CO)12 system presented.

• Island nucleation is the key to thinner films.• As film roughness increases we need a way to relate

in situ measures to film thickness – the equivalent film thickness.