investigation of the structural resistance of silicon membranes for microfluidic applications in...

Post on 15-Dec-2015

215 Views

Category:

Documents

2 Downloads

Preview:

Click to see full reader

TRANSCRIPT

1

Investigation of the structural resistance of Silicon membranes

for microfluidic applications in High

Energy PhysicsC. Gabry

A. Mapelli, G. Romagnoli, D. Alvarez, P. Renaud,

2

Summary1 – Objectives of the project

2 – Literature review

3 – Single-Edge V-Notch Beam (SEVNB) tests 3.1 – SEVNB specimens 3.2 – Experiments 3.3 – Results 3.4 – Discussion

4 – Channel geometry beam specimens under bending stress 4.1 – Specimens 4.2 – Experiments 4.3 – Results 4.4 – Discussion

5 – Conclusion

3

ContextAim : reduce material budget of SCSi micro-

devices

Minimize of thickness of SCSi membranes

Insure sufficient resistance to internal pressures=> Need of fracture prediction tool to optimize

structures

4

1 – Objectives

Main goal : provide fracture data for the implementation of a fracture prediction tool

Requirements : simple situation to allow CERN’s Engineering Office to make reliable models

5

1 – Objectives2 testing rounds :

Fracture toughness test => to set parameters of FEA model

Test with slightly more complex geometry to validate those parameters

6

2 – Literature review

7

Fracture ToughnessDefinition : Fracture toughness (Kc) is the ability

of a material to resist against propagation of a pre-existing crack.

Fracture toughness = critical stress intensity

Unit : Pa√m

High fracture toughness : Ductile fracture

Low fracture toughness : Brittle fracture

8

Fracture modes3 different fracture modes :

tensile (mode I), -typically the weakest- shear (mode II), tear (mode III)

9

Fracture toughness testTo perform a fracture toughness determination

test, one need :

Sample with pre-existing crack of known geometry

Tool able to apply and measure load & displacement

The load at fracture to allow for fracture toughness calculation

10

3 – Single-Edge V-Notched Beam (SEVNB) in three point bending

tests

11

3.1 – SEVNB Specimens

Easy to manufactureExisting standards

(ASTM C1421-10 for ex.)Possible to perform FEAEasy to testNo tensile stress (except at crack tip)

Reasons of the choice

12

3.1 – SEVNB Specimens

1st step : Alignment to crystalline structure

2nd step : Photolithography

3rd step : Notch etching (KOH)

4th step : Dicing

Fabrication process

13

3.1 – SEVNB Specimens

Mask for SEVNB specimens fabrication

Fabrication process

14

3.1 – SEVNB SpecimensObtained specimens

Obtained notches :

Picture of full specimen ?

Mask ?

15

3.1 – SEVNB SpecimensDimensions of specimens

Standard sizes are too large (e.g. W=3mm)

ASTM standards gives two main ratios to respect :W/B = 0.75 0.35 < a/W < 0.6 Fabricated specimens :

16

3.1 – SEVNB SpecimensAnalytical formula for fracture toughness

FEA Comparison between Sharp and V-Shaped crack

17

3.1 – SEVNB SpecimensAnalytical formula for fracture toughness

Comparison between various analytical formulas for fracture toughness

18

3.2 – SEVNB ExperimentsExperimental setup – Overview

1 actuator

1 load cell

Mechanical support to link & align parts

Removable chuck to adapt to every test

Transportable & adaptable system

Taken from “In-situ MEMS Testing”, A. Schifferle, A. Dommann, A. Neels and E. Mazza

19

3.2 – SEVNB ExperimentsExperimental setup – Chucks

Vertical pins (1mm spacing)

Available chucks :S0=5mm

S0=10mm

20

3.2 – SEVNB ExperimentsCompliance calibration

Extremely stiff sample (1mm diameter steel rod) in place of specimen

21

3.2 – SEVNB ExperimentsData analysis

Initial slack correction by linear fit

Tool compliance correction on measured deformation

δb : Deformation of the beam P : Load δt (=P*Ct) : Deformation of the toolδm : Measured deformation Ct : Compliance of the tool,

22

3.3 – SEVNB ResultsCompliance

Analytical compliance : 1.55*10-6 m.N-1

Mean experimental compliance : 1.27*10-6

± 0.39*10-6 m.N-1

23

3.3 – SEVNB ResultsFracture Load

Fracture toughness from literature (for (110) plane)* :1.23±0.18 MPa.√m-1

Experimental fracture toughness : 1.50±0.09 MPa.√m-1

*T. Ando, X. Li, S. Nakao, T. Kasai, M. Shikida and K. Sato, “Effect of crystal orientation on fracture strength and fracture toughness of single crystal silicon”, Micro Electro Mechanical Systems, 17th IEEE International Conference on MEMS, pp. 177-180 (2004)

24

3.3 – SEVNB ResultsScatter in results

No correlation between fracture load and compliance

Higher scatter for compliance than fracture toughness

25

3.4 – Sources of scatterSharpness of notch

Notch sharpness :

22.89nm radius for this measurement

26

3.4 – Sources of scatterDepth of notch (a)

Notch depth conform to what was planned

(143um for the measured sample, 2um bigger than expected)

Make more measurements to see standard deviation

27

3.4 – Sources of scatterMisalignment of sample

500um misalignment=> 5.8% compliance variation

28

3.4 – Sources of scatterThickness of sample (B)

10um variation of B=> 1.1% variation of compliance

B fixed by dicing=> low variability expected

Actual measurements of B with standard deviation !!

29

3.4 – Sources of scatterMisalignment of tool

Multiple types of possible misalignments

(In plane, out of plane…)

Not possible to estimate without changing boundary conditions of FEA model

Source of global error ? (same misalignment for a batch)

30

3.4 – Sources of scatterCompliance of tool too high

Tool compliance higher than sample compliance… Not ideal !

But in theory possible to correct

31

3.4 – Sources of scatterConclusion

Most sources did not have a big enough impact to explain all the scatter

Misalignment inside the tool was not estimated, but is probably the main source of mismatch (not scatter)

Brittle materials typically have a random behaviour

32

4 – Channel geometry beam under bending

stress

33

4.1 – Introduction & goals

Sample with channel-like notch

Goals :1 – validate simulation

parameters establishedduring SEVNB tests

2 – study influence of manufacturing method on overall strength of sample

34

4.2 – Specimens

1st step : Alignment to crystalline structure

2nd step : Photolithography

3rd step : Notch etching (KOH or DRIE)

4th step : Dicing

Fabrication process

35

4.2 – SpecimensObtained specimens

4 types of specimen :

DRIE90° KOH

DRIE80° DRIE60°

36

4.2 – SpecimensDimensions of specimens

Optical measurements performed on 5 samples for each type

37

4.3 – ExperimentsExperimental process

3 point bending :

Chuck just under channel

Precise alignment

38

4.3 – ExperimentsExperimental process

Tests both in three– and four point bending

39

4.4 – Results3 point bending

DRIE 60° DRIE80°

DRIE 90° KOH

40

4.4 – Results4 point bending

DRIE 60° DRIE80°

DRIE 90° KOH

41

4.4 – ResultsFracture loads

42

4.4 – ResultsCompliance

Much higher mismatch for 3 point bending than for 4 !

43

4.4 – ResultsCompliance – Summary

Much higher mismatch for 3 point bending than for 4 !

44

4.4 – Parametric analysis

FEA estimation of influence of notch depth on compliance

Estimation of measurement errors through FEA model

45

4.4 – Parametric analysis

Measurement errors have more impact for 3 point bending tests than for 4 point ones

Estimation of measurement errors through FEA model

46

5 – ConclusionGathering of fracture data with SEVNB specimens to

allow setting parameters of fracture prediction tool

Validation of parameters through second testing batch

Essays on tensile-test machine for micro-scale specimens

Observation of sensitive aspects in fracture tests (scatter sources)

Observation of impact of manufacturing methods on fracture strength

47

48

Backup slides

49

Energy based approach

Elastic energy released = surface energy created

Energy release rate (G = πσ2a/E) : speed at which the energy is released by growth of the crack

GIc= KIc2*(1-ν2)/E[hkl]

(E : Young’s modulus, a : half crack length, σ : stress ; ν : Poisson’s coefficient ; [hkl] : Miller indexes)

50

Possible test methodsA – Micro-indentation

B – Double Cantilever Beam

C – Compact Tests

D – Compression Loaded Double Cantilever Beam

E – Three/four point bending

F – Other (cantilever bending, On-chip tensile test device)

51

A – Micro-indentationEasy to implement

Possible to test several crystalline orientations

Machinery quite common

Measuring the crack length

Residual stress after indentation

Hard to simulate with FEA

52

B – Double Cantilever Beam

Theoretically possible to measure propagation values

Plastic deformation in arms

Direction of crack growth

Tensile load applied

53

C – Compact TestsLess plastic deformation in thick arms

Short arms => crack growth direction more controllable

Plastic deformation & parasitic crack growth at load pins

Only initiation

Tensile load applied

D – Compression-Loaded Double Cantilever Beam

No tensile stress applied

Side groove to help crack grow in the intended direction

Stable crack growth

Crack growth monitored with thin film resistance

Wrong direction of crack growth

Hard to test such specimens (need a frame to hold them)

54

55

E – Three/four point bending of notched

specimenASTM standard procedures available

Test under bending conditions

Easy modeling and manufacturing

Standard not adapted to our needs (scale)

No propagation measurements

F – Other

On chip tensile test device :

Complicated analysis

Sharpness of notch

Complicated fabrication process

Easy to apply the load

56

57

Wet etching for vertical sidewalls

45° orientation of specimens relative to primary flat can lead to vertical sidewalls with KOH etch

%wt of KOH and temperature have an influence on the obtained slope of the walls

Ex : 60%wt & 60°C

58

Alignment to crystalline planes

Anisotropic KOH etching step results in squares under the circle openings

Squares which are the most alligned give the cristalline orientation

59

Support of the beams

Beam should be free to rotate around the rollers

Diameter : 0.5 – 1 mm

Supports considered :Razor bladesSteel wires

60

3 points bendingmanufactured wafer

61

3.2 – Compression Loaded Double

Cantilever Beam

62

Reasons of the choiceCL-DCB has been chosen

as second specimen :

Easy to manufacture Possible to perform

FEA No tensile stress

(except at crack tip) Eventually possible to

measure both initiation & propagation values

Stable crack growth Easy to monitor crack growth (metal

gage)

63

Dimensions of specimensb defined by wafer

thickness

Other dimensions set to keep ratios of previous experiments with this specimen

64

Fabrication process

1st step : Alignment to crystalline structure

2nd step : DRIE with specimen shape

3rd step : Smoothing (RCA cleaning, SiO2 oxidation, etching of SiO2 layer)

4th step : Releasing of specimens by grinding of back side

65

Required machinery

Possible to estimate fracture load P and displacement by assuming KIc = 1Pa√m

For B=1mm, W=525μm, L=10mm, a=300μm : Fc = 450 mN

yc = 23 μm

=>Resolution required : ≈ 10mN & 1μm

(Tensile test machine at CERN has 8mN and 1μm resolution)

66

Tools available

top related