Micromechanical Testing of Thin Films

WarrenOliverMTS Nano Instruments

Oak Ridge, Tennessee

Nano Indenter® G200

Precise mechanical testing in the micro to “sub-nano” range of loads and displacements

Testing instrumentation: Nano Indenter ® XP

Coil/magnet assembly

Leaf spring

Capacitance gauge

Indenter

Sample

Nickel

0

5

10

15

20

25

30

0 100 200 300 400 500

Load

(m

N)

Displacement (nm)

0

5

10

15

20

25

30

0 500 1000 1500 2000

Lo

ad

(m

N)

Load-displacement behavior

Aluminum, typical of soft metallic behavior, shows very little displacement recovery upon unloading

Fused silica, typical of ceramic behavior, shows large elastic recovery upon unloading

Aluminum

Fused silica

Read Heads

The problem…

Indentation results for 1-m low-dielectric-constant (low-k) film on silicon

“skin” effect

substrate effect

All data are affected, to some extent, by either “skin” or substrate. So what is the modulus of this film? Problem is not too bad for 1-micron films, and hardness is less sensitive than modulus. But our microelectronics customers tell us they really want to test 200nm films!

The goal

The goal of this work is to develop an empirical model that:

• Is appropriate for a realistic range of low-k materials• Correctly models the influence of the silicon substrate• Requires no a-priori knowledge of film properties beyond

thickness• Can be incorporated into Testworks• Is relatively independent of diamond tip radius

Developing the model

There is much to be learned from the process of developing the model.

1. Survey experimental results. Select properties that bound the range of interest in terms of E and H.

2. Perform preliminary simulations to get y = f(E,H). Select

properties that bound the range of interest in terms of E and y.

3. Perform simulations for “boundary” samples.4. Calculate errors in modulus and hardness, relative to expected

properties for bulk materials. 5. Plot error as a function of parameters that are relevant, knowable,

and dimensionless. Derive model for error by curve fitting.6. Test model with more simulations (on materials inside boundaries)7. Test the model experimentally

Virtual IndenterTM Features

• Real area functions, spheres, flat punches• Bulk materials• Up to three stacked films• Particle/fiber/disk in a matrix• Range parameters easily• A variety of constitutive models• Automated Excel output

Uncorrected modulus from simulation

0

2

4

6

8

10

12

14

16

18

20

0 100 200 300 400 500

Displacement Into Surface [nm]

E* O

P [

GP

a]

100nm film200nm film300nm film400nm film500nm film600nm film700nm film800nm film900nm film1000nm filmBulk

Uncorrected modulus vs. normalized contact radius

0

2

4

6

8

10

12

14

16

18

20

0% 20% 40% 60% 80% 100% 120% 140% 160%

Displacement Into Surface/Film Thickness [%]

E* O

P [G

Pa]

100nm film200nm film300nm film400nm film500nm film600nm film700nm film800nm film900nm film1000nm filmBulk

Corrected modulus vs. normalized contact radius

0

2

4

6

8

10

12

14

16

18

20

0% 20% 40% 60% 80% 100% 120% 140% 160%

Displacement Into Surface/Film Thickness [%]

E* O

P [G

Pa]

100nm film200nm film300nm film400nm film500nm film600nm film700nm film800nm film900nm film1000nm filmBulk

Applying the model to experimental data

Wafers supplied by SEMATECH:

• 4 wafers of nominally the same film, different thicknesses• 250nm, 488nm, 747nm, 1156nm• k ~ 2.3• Technology targeted for use beyond 45nm node• Deposited using “porogen” and then UV cured to cause

residual pores. Cure times varied with thickness. UV cannot penetrate past 750nm.

Calculating modulus old way and new way

0

2

4

6

8

10

12

0 10 20 30 40 50 60

Contact Radius / Film Thickness [%]

EO

P [

GP

a]

250nm

488nm

747nm

1156nm

0

2

4

6

8

10

12

0 50 100 150 200 250 300

Displacement Into Surface [nm]

E* O

P [

GP

a]

250nm

488nm

747nm

1156nm

Old way: take minimum

New way: take data for 30% < a/t < 35%

Calculating modulus old way and new way

0

2

4

6

8

10

12

250 488 747 1156

Film Thickness [nm]

Mo

du

lus

[G

Pa]

Min of uncorrected data, E*_OP

Corrected E_OP over range 30% < a/t < 35%

10nm

- 20

nm

25nm

- 35

nm

40nm

- 50

nm

90nm

- 10

0nm

29nm

- 34

nm

63nm

- 74

nm

103n

m -

120n

m

168n

m -

195n

m

Moduli calculated by old way are too high by 30%, because data at minima are significantly affected by substrate.

Using new model also reduces uncertainty

0%

1%

2%

3%

4%

5%

6%

250 488 747 1156

Film Thickness [nm]

Rel

ativ

e u

nce

rtai

nty

[%

]

Min of uncorrected data, E*_OP

Corrected E_OP over range 30% < a/t < 35%

10nm

- 20

nm

25nm

- 35

nm 40nm

- 50

nm

90nm

- 10

0nm

29nm

- 34

nm

63nm

- 74

nm

103n

m -

120n

m

168n

m -

195n

m

Conclusions

A model has been developed to compensate for the influence of the substrate on the indentation properties of thin low-k films.

Model has been incorporated into a Testworks test method.

Model significantly reduces both error and uncertainty, especially for very thin films.

We continue to test the model on more low-k films.

Uniaxial Testing of Free Standing Films

Warren C. Oliver and Erik G. Herbert, MTS CorporationJohnathan Doan, Reflectivity

Nanovision Stage

Travel: 100 m x 100 mResolution/Noise: 2 nmFlatness of travel: 1-2 nmAccuracy: 0.01 %Settling Time: 2 ms-Capacitive feedback control

Automated Indent and Scan

Scan Procedure

Film

Step1) Scan substrate to determine slope of surface

Step 2) Find top of post

Step 3) Scan plane of predetermined slope just below top of post, but above film

Leveled Targeting Scan

8 μm Wide, Doubly Clamped Bridges

0

2

4

6

8

10

12

14

0.0E+00 5.0E-07 1.0E-06 1.5E-06 2.0E-06

Displacement (m)

Lo

ad

on

Sa

mp

le (

μN

)

L = 34 μm L = 66 μm

L = 50 μm

L = 18 μm

Load Displacement Curves

9.95

10

10.05

4.95 5 5.05

Nominal ForceExcitation Force

Lo

ad

(m

N)

Time (seconds)

0

2

4

6

8

10

12

0 10 20 30 40 50 60 70 80

Nominal Force, P/P = Constant

Load (

mN

)

Time (seconds)

.

Continuous Stiffness Measurement Technique (CSM)

CSM - Elastic & Viscoelastic

Elastic Viscoelastic

-2

-1

0

1

2

-1

-0.5

0

0.5

1

0 10 20 30 40 50

Exc

itatio

n fo

rce

(µN

)

Response displacem

ent (nm)

Time (milliseconds)

-2

-1

0

1

2

-1

-0.5

0

0.5

1

0 10 20 30 40 50E

xcita

tion

forc

e (µ

N)

Response displacem

ent (nm)

Time (milliseconds)

= 0°

= 90°

8 μm Wide, Doubly Clamped Bridges

0

10

20

30

40

50

60

70

0.0E+00 5.0E-07 1.0E-06 1.5E-06 2.0E-06

Displacement (m)

Bri

dg

e S

tiff

nes

s (N

/m)

L = 34 μm

L = 66 μm

L = 50 μm

L = 18 μmHarmonic Displacement = 30 nm

Stiffness Displacement Curves

Describing Bridge Tensile Specimens

P

w L+(L-w)/2

h

L

PF

F

sin)(2sin2 rEAFP

For not quite so small angles!3

sin3

3tan

31

The Stiffness Displacement Relationship:

)()(2tansin2

)(

4 1 EwLhA

wL

AEhP r

3

2

)(

)(24

)(

4

wL

hEA

wL

A

dh

dP rr

Testworks: A Complete Solution

Now it gets Interesting

TestWorks and the Nano Indenter G200

Design of MEMS structural experiments was easily done with the flexibility and control offered by the TestWorks software

TestWorks provides a user interface that facilitates the design of new (i.e. MEMS) and novel experiments without the need to have knowledge of C++ programming

The Nano Indenter G200 system can provide this information quickly and reproducibly, offering manufacturers an attractive tool for product development

Thank you!