self forming barrier layers from cux thin films

0 0.5 1 1.5 2 2.5 3 3.5 40

5

10

15

20

25

30

35

40

45

50

Affects of Pre-anneal Air exposure on Resistivity

Pure Cu 325nmCuTiCuTi (no air exposure)Power (CuTi (no air exposure))

Anneal Time (hours)

Resis

tivity

(μΩ

-cm

)

0 0.5 1 1.5 2 2.5 3 3.5 40

5

10

15

20

25

30

35

Affects of Pre-anneal Air exposure on Resistivity

Pure Cu 325nmCuMnCuMn (no air exposure)

Anneal Time (hours)

Resis

tivity

(μΩ

-cm

)

Self Forming Barrier Layers from CuX Thin Films Shamon Walker, Erick Nefcy, Samir Mehio

Dr. Milo Koretsky, Eric Gunderson, Kurt LangworthySponsors: Intel, Oregon Metals Initiative , ONAMI

A1

B1

C1

TEM Sample – B1

CuTi (5.6 at. % Ti) Barrier Layer Reaction Study

Figure 1. Scanning Electron Microscope (SEM) image of a CuTi (5.6 at.% Ti) alloy annealed for 2 hrs at 500oC in O2.

Fig 3. SEM image of a CuTi (5.6 at.% Ti) alloy annealed for 2 hrs at 500oC in UHV w/ pre-anneal air exposure of 30 days.

Figure 2. SEM image of a CuTi (5.6 at.% Ti) alloy annealed for 2 hrs at 500oC in UHV with NO pre-anneal air exposure.

• Anneal ambient significantly affected CuTi structure.• A CuTi phase change was observed for B1 but not for C1.• The only difference between B1 and C1 was pre-anneal air exposure.• Water molecules from the air adsorbing to the films surface could be the reason

why no CuTi phase change was observed in C1.

200 300 400 500 600 700 800 900 1000 1100-400

-300

-200

-100

0

100

200

300

400

500

Gibbs Energy of Reaction for M + SiO2 → MxOy + Si

Al2O3

Ti3O5

Ti2O3

MgO

TiO

TiO2

MnO

NiO

GeO2

MnO2

Mn2O3

Mn3O4

Temperature (K)

Gib

bs E

nerg

y of

Rea

ctio

n (k

J/m

ol)

CuX Resistivity Study (X= Ge, Ni, Mn, Ti)

Mathematical Modeling: Diffusion of X through Cu

f f fg h T s

,1

n

rxn i f ii

g g

2

D,

1

4 1C z,t sin for 1,3,5,...XCun tLX o

Xn

C n z e nn L

2

2

( )X XXCu

C C tD z t

CuTi

SiO2

CuTi

SiO2

Cu7Ti2

CuTi

SiO2

SiO2

Pt

CuTi

Barrier Layers

CuTi

SiO2

TixSi TiyO2

Possible OxideBarrier Layer

Possible OxideBarrier Layer

Possible OxideBarrier Layer

2 2x yM SiO M Si M O

• A simple model for diffusion of X through Cu was formulated using symmetrical boundary conditions. The model was created by combining a material balance and Fick’s 1st law of Diffusion for a thin slab. The governing equation can be seen below:

• Interdiffusion of metals into SiO2 is often observed upon annealing metal overlayers supported on thin SiO2 films (Dallaporta et al.). It is believed that a large chemical potential gradient facilitates Ti diffusion to the interface where it reduces SiO2 and forms a titanium oxide compound.

z 0 z L

• Pretorius et al. found that Ti, Zr, Hf, V, and Nb react with SiO2 to produce oxides and silicides at the interfaces of metal/SiO2/Si films. They suggested a direct reaction between a silica film and metal overlayer (M) as follows:

• The solution to the model can be found below:

Project Background• Ultra thin diffusion barriers (between Cu and SiO2) are required for

super computing.

•The current deposition orientation for laying interconnect material on an integrated circuit uses a tantalum nitride (TaN)/tantalum barrier to keep SiO2 separate from Cu:

Si/SiO2/TaN/Ta/Cu.

Project Requirements:

1) 4-10 nm barrier layer2) Film resistivity < 3.0 μΩ-cm3) No detectable interdiffusion

between Cu and SiO2.

Sputtering Process

• The lower the Gibbs energy of reaction…the larger the spontaneity of the reaction. • The data sets with < 0 have a high affinity to reduce SiO2 and

form a metal oxide of their own.• All of the Ti oxidation reactions have < 0.

rxng

Figure 6. A plot of the Gibbs energy of reaction vs. reaction temperature for an assumed redox reaction of M + SiO2 → MxOy + Si.

rxng

• CuTi and CuMn have a minimum resistivity of 6.9 μΩ-cm and 3.02 μΩ-cm, respectively. These values, however, did not quite satisfy the project requirements of < 3.0 μΩ-cm . However, CuMn was very close.

• Prolonged pre-anneal air exposure is believed to be the reason behind the peculiar shape of the curves in both figures. Oxygen and water molecules adsorb to the films surface during exposure and eventually form a surface oxide. These ultra thin oxides are highly resistive, and increase the value of the measured resistivity (seen above).

Figure 7. A plot showing the affect of anneal time on thin film resistivity of CuMn. One group of samples was exposed to air before annealing and the other was not exposed to air before annealing. All films were 7 at. % X and annealed at 500oC in Ar at 30 mTorr..

Figure 8. A plot showing the affect of anneal time on thin film resistivity of CuTi. One group of samples was exposed to air before annealing and the other was not exposed to air before annealing. All films were 7 at. % X and annealed at 500oC in Ar at 30 mTorr..

0

0.00833333333333335

0.0222222222222223

0.194444444444446

0.416666666666668

0.666666666666667 1 23.5

012345

15140 265 390

Time (hr)

[Ti]

(#/n

m3)

z (nm) 0

0.0055555555555...

0.0125

0.0222222222222223

0.138888888888889

0.277777777777779

0.416666666666668

0.583333333333333

0.750000000000003 1 1.5 2.5 3.50

0.51

1.52

2.53

3.54

4.55

Time (hr)

[Ti]

(#/n

m3)

Figure 9. Plot of Ti concentration in Cu as a function of anneal time and distance from each

edge in a 430nm CuTi film.

Figure 10. Plot of Ti concentration in Cu as a function of anneal time and distance from each edge

in a 430nm CuTi film.

Center of Film Center Profile

Edge Profile

Figure 5. TEM image of CuTi/SiO2 interface. The CuTi (5.6 at. % Ti) film was annealed for 2 hrs at 500oC in UHV with NO pre-anneal air exposure.

Figure 4. Transmission Electron Microscope (TEM) image of a CuTi (5.6 at.% Ti) alloy annealed for 2 hrs at 500oC in UHV with NO pre-anneal air exposure.

Current Industrial Method:

Proposed Method:

•Pre-anneal deposition orientation:

Si/SiO2/CuX

•Post-anneal deposition orientation may be:

Si/SiO2/Metal Oxide/Cu

Pre-anneal air exposure

No pre-anneal air exposure

• In sputtering, a film is grown by the ejection of material from a solid surface following the impact of energetic ions.

Modeling of Ti diffusion through Cu• CuTi (5.6 at. % Ti) annealed at 500oC

A1B1C1

•Self forming barrier layers using CuX(where X = Mg, Mn, Ge, Ni, Ti, and Al)

AJA Orion IV Sputtering System

• RF Magnetron 300W dual power supply

• Two mass-flow controllers (Ar and O2)• Maximum substrate temperature:

850oC• Base Pressure ≈ 1E-08 Torr• Substrate Rotation• Three magnetron sputtering guns

References•Qiang Fu, Thomas Wagner, Interaction of nanostructured metal overlayers with oxide surfaces, Surf. Sci. 62 (2007) 431-498•R. Pretorius, J.M. Harris, M.A. Nicolet, Reaction of thin metal films with SiO2 substrates, Solid State Electron. 21 (1978)•H. Dallaporta, M. Liehr, J.E. Lewis, Silicon dioxide defects induced by metal impurities, Phys. Rev. B 41 (1990) 5075

Pre-anneal air exposure

No pre-anneal air exposure