Ductile-Regime Machining of Silicon Carbide and Quartz

byby

Deepak Ravindra

Thesis Defense (Mechanical Engineering)

Thesis Advisor: Dr. John A. Patten

Presentation Overview

• Introduction

• Research Background

• 4 projects– Ductile to Brittle Transition (DBT) of 4H-SiC– Single point diamond turning (SPDT) of CVD-SiC– Single point diamond turning (SPDT) of CVD-SiC– Hybrid laser-SPDT process for CVD-SiC– SPDT of Quartz

• Future/Upcoming work

• Acknowledgements

Ductile Regime Machining

•Plastic flow of material in the form of severely sheared machining chips occur

•Possible due to High Pressure Phase Transformation (HPPT) or direct amorphization

•Plastic deformation caused from highly localized •Plastic deformation caused from highly localized contact pressure and shear stresses.

•High pressure (metallic) phase could be used to improve manufacturing processes and ductile response during machining.

Ductile Regime Machining of Ceramics

• Model proposed by Blake and Scattergood

• DBT is calculated based on material properties

• Depth exceeding critical • Depth exceeding critical depth will result in brittle machining

• Micro-cracks / surface damage depth, yc should not extend beyond the cut surface plane

Critical Depth of Cut (dc)

• Griffith fracture propagation criteria:dc ~ 0.15 . (E / H) . (Kc / H)2

where: 0.15 = estimated constant of proportionality E = elastic modulusH = hardnessKc = fracture toughness

Why Use Silicon Carbide?•Extreme hardness (~27GPa for Polycrystalline CVD coated)

•High wear resistance

•High thermal conductivity (3.4 W/cm.K)*

•High Temperature Operation

•Wide energy bandgap (3.26eV)*•Wide energy bandgap (3.26eV)*

•High electric field breakdown strength (2.22 x 106 V/cm)*

•High maximum current density

•High saturated electron drift velocity (2 x 107 cm/sec)*

*Values for 4H single-crystal

Project I –Ductile to Brittle Transition (DBT) of a 4H Single Crystal SiC Wafer

by Performing Nanometric Machining

Applications of 4H SiC

•High-Frequency Power Devices–RF & Microwave Amplifiers/Transmitters

•High-Temperature Devices–High temperature electronics & power devices

•Optoelectronic Devices–Laser diodes & photodiodes

•III-V Nitride Deposition–Light emitting devices

Experimental Plan

• 3 depth of cuts (100nm, 500nm & 1000nm) were planned to be carried out using the Nanocut II

• Predicted range of Ductile to Brittle Transition (DBT) is between 100nm – 1000nm (based on calculations and between 100nm – 1000nm (based on calculations and previous experiments)

• A depth greater than 1000nm will only be carried out if the DBT is not identified with the first set of experiments.

• Cuts will be imaged and analyzed using the AFM

Experimental Method-Nanocut II

Diamond ToolZ-axis

Force SensorPZT Tube &

Cap Gage

Figure 1

X-axis

Force Sensor Sample Mount

3”(76.2mm) SiC wafer (4H-Single Crystal)

•The primary flat is the {1010} plane with the flat face parallel to the <1120> direction.

•The primary flat is oriented such that the chord is parallel with a specified low index crystal plane.

•The cutting direction is along the <1010> direction

Figure 2

Nanocut matrix of cuts

•The cuts were done in array pattern to help with imaging

Figure 3

Results – Actual Cuts on SiC

•Cuts image under an optical microscope at 400x magnification

•Dimensions are approximately 10-20µm in length and 120µm in width

•Cutting direction from right to left

•The 10mm tool nose radius makes the cuts wider than they are longer

–The maximum depth of cut is at the middle of the cut

Figure 4

Results- Force Data

• Cutting and thrust forces increase as the depth of cut increases • The cutting forces are consistently greater than the thrust forces for all

depths• The force data does not indicate direct evidence of a DBT

– Beyond the DBT, the force data may indicate brittle fracture• Less energy/force required to fracture material compared to ductile response.

• In this case, the brittle fracture was not so severe as to measurably affect the resultant force trends

Figure 5

Results- Height Profile (Ductile)

•An AFM scanned section of a (1000nm programmed depth) cut.–All the 100nm & 500nm depth of cuts were completely ductile

•The measured depth is 816nm.

•The “V” shape of the ductile cut represents the imprint of the tool

Cutting DirectionFigure 6

Results- Height Profile (DBT)Cutting Direction

•An AFM scanned section of a cut where the brittle characteristic of the material is visible.

•The measured depth is 836nm.

Figure 7

Results- Height Profile (Brittle)Cutting Direction

• The maximum measured depth of cut (1170nm) is more than the programmed depth of cut (1000nm).

• Due to microcracks that could extend deeper than

the depth of cut below the machined surface

Figure 9

Results – Brittle Cut

Cutting Direction

•Optical image (200x) of 1000nm cut showing brittle fracture

• Jagged edges are due to crack propagation and uncontrolled material removal

in the brittle regime.

Figure 10

Results – Brittle Cut

Cutting Direction

Figure 11Ra

•White light interferometric microscope (WYKO) image of 1000nm brittle cut

• The actual depth varies from zero at the ends (top and bottom of the cuts, outside the field of view) to a maximum in the middle.

Conclusion•The DBT depth for a 4H-SiC wafer was determined to be between 820nm-830nm in the <1010> cutting direction.

•Cutting forces and thrust forces increase as the depth of cut increases

•Beyond the DBT depth, the cut produced becomes brittle

•The fracture from brittle cutting then leads to pitting and microcracks, this results in significant and uncontrolled subsurface damage

• In order to machine a semiconductor or ceramic in the ductile-regime, it is crucial to know its DBT depth

Project Goals (common for the three SPDT experiments)

• Improve surface finish (surface roughness) via ductile mode machining

• Increase material removal rate (MRR) by altering:– Feed– Feed– Depth of Cut– Cutting Speed

• Minimize diamond tool wear

*Establish machining parameters to meet all three criteria's (project goals)

Concept of improving surface roughness1

2

3

Project II - Improving the Surface Roughness of a

CVD-SiC by Performing Ductile Regime SPDT

• 6” disk from POCO Graphite Inc. was used

• Mirror finish surface required

• To be used as optic mirrors in an Airborne Laser (ABL) device

• CVD coated SiC is preferred because:

– High Purity (>99.9995%)

– High Density (99.9%)

– Homogeneity

– Chemical & Oxidation Resistance

– Good Cleanability & Polishability

– Good Thermal & Dimensional Stability

Experimental Procedure

• Initial test matrix is design (with varying depths

of cuts and feeds)

• Preparatory machining on 2” polished CVD-SiC

• Preliminary machining on 6” CVD-SiC (as-• Preliminary machining on 6” CVD-SiC (as-

received)

• Final experimental matrix is designed based on

preparatory and preliminary machining results

• Final machining on 6” CVD-SiC

Experimental Setup for SPDT

Results (surface roughness & feed correlation)

Surface Roughness (Ra) vs. Feed

Pass #

16

Pass #

15

Pass #

14

Pass #

13

Pass #

12

Pass #

11

Pass #

10

Pass #

9

Pass #

8

Pass #

7

Pass #

6

Pass #

5

Pass #

4

Pass #

3

Pass #

2Pass #

1

400

600

800

1000

1200

1400R

a(n

m)

• Data is arranged from highest to lowest feed

• Feed is more dominant than depth of cut when improving surface roughness

• Feed is reduced when surface roughness does not improve

Pass #

16

Pass #

15

Pass #

14

Pass #

13

Pass #

12

Pass #

11

0

200

As

Received

30

2µm

30

2µm

30

2µm

30

2µm

30

2µm

5

2µm

5

2µm

5

2µm

5

0.5µm

5

0.25µm

1

1µm

1

0.5µm

1

0.5µm

1

0.5µm

1

0.5µm

1

0.25µm

Feed (µm/rev)

Results (Rz for all machining passes)

Peak-to-valley (Rz) vs. Pass Numbers

2

3

4

5

6

7

8

9

Rz (

µm

)

• The Rz value was used to determine the

required depth of cut

• The feed and/or depth of cut is reduced when

the Rz value does not improve.

0

1

As

Recd.

1

2µm

2

2µm

3

2µm

4

2µm

5

2µm

6

2µm

7

2µm

8

2µm

9

0.5µm

10

0.25µm

11

1µm

12

0.5µm

13

0.5µm

14

0.5µm

15

0.5µm

16

0.25µm

Pass Numbers

Results (Cutting force vs. depth of cut)

Cutting forces vs. depth of cut and the corresponding feed

Pa

ss

#1

0

Pa

ss

#1

6

Pa

ss

#9

Pa

ss

#1

5

Pa

ss

#1

4

Pa

ss

#1

3

Pa

ss

#1

2Pa

ss

#1

1

Pa

ss

#8

Pa

ss

#7

Pa

ss

#6

Pa

ss

#5

Pa

ss

#4

Pa

ss

#3

Pa

ss

#2

Pa

ss

#1

100

200

300

400

500

600C

utt

ing

Fo

rce

s (

mN

)

• Depth of cut, feed and surface roughness

influence the cutting forces

• Depth of cut is the dominant parameter

Pa

ss

#1

0

Pa

ss

#1

6

Pa

ss

#9

Pa

ss

#1

5

Pa

ss

#1

4

Pa

ss

#1

3

Pa

ss

#1

2

0

100

2 (3

0µm

/rev)

2 (3

0µm

/rev)

2 (3

0µm

/rev)

2 (3

0µm

/rev)

2 (3

0µm

/rev)

2 (5

µm/re

v)2

(5µm

/rev)

2 (5

µm/re

v)1

(1µm

/rev)

0.5

(1µm

/rev)

0.5

(1µm

/rev)

0.5

(1µm

/rev)

0.5

(1µm

/rev)

0.5

(5µm

/rev)

0.25

(1µm

/rev)

0.25

(5µm

/rev)

Depth of cut (µm) with corresponding feed

Results (Surface image of 6 passes)

• Image was taken at 50x magnification

• Surface continuously improved after each pass

• Band between pass 4 & 5 was due to tool

chatter

Results (Surface image comparison)

• Images comparing the surface before (left) and after (right) SPDT

• Images were taken at a 1000x magnification

• The sharp/uneven peaks on the surface disappeared after the SPDT operation

Results (SEM images of tool wear)

• Tool was used for pass 1 (2µm depth & 30µm/rev feed)

• SEM images are used to measure tool wear

– Wear length across cutting edge

– Rake wear

– Flank wear

Results (Cutting force vs. tool wear across

cutting edge) Cuting Force vs. Measured Tool Wear Across Cutting Radius

Pass #

1(2µ

m)

Pass #

2 (

2µ

m)

Pass #

3 (

2µ

m)

Pass #

4 (

2µ

m)

Pass #

5 (

2µ

m)

Pass #

6 (

2µ

m)

Pass #

7 (

2µ

m)

Pass #

8 (

2µ

m)

Pass #

11 (

1µ

m)

200

250

300

350

400

450

500

Cu

ttin

g F

orc

e (m

N)

• Wear across cutting edge radius is a function of depth of cut

• Wear length and cutting forces increase as depth of cut increases

Pass #

10 (

0.2

5µ

m)

Pass #

16 (

0.2

5µ

m) Pass #

11 (

1

Pass #

12 (

0.5µ

m)

Pass #

14 (

0.5µ

m)

Pass #

13 (

0.5µ

m)

Pass #

15 (

0.5µ

m)

Pass #

9 (

0.5

m)

0

50

100

150

200

488360355355346345340340268217208214205147145122

Measured Wear (µm)

Cu

ttin

g F

orc

e (m

N)

Results (6” disk before & after SPDT)

Conclusion

• Recommended machining parameters for commercial manufacturing of CVD-SiC

• Best surface finish is achieved from the lowest feed

• Tool wear can be minimized by using cutting fluids

Project III-Developing a Hybrid Laser-SPDT Machining Process for Smoothing Ceramics

• To develop a more efficient manufacturing process of SiC

• Laser machining done by Mound Laser & Photonics Center (MLPC) using a pico-second pulsed laser

SPDT of SiC

Images of CVD-SiC Samples

• All plateaus were laser machined using different ablation parameters

• These plateaus were then diamond turned at a 1µm depth with a 1µm/rev feed.

Results (Surface roughness)

Surface roughness (Ra) after 1µm/rev feed pass

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

5

Ra (

µm

)

As Received

After Pass 1

• Data is arranged from the roughest to the smoothest

as-received plateau

• The Ra values dropped by almost 50% after just a

single pass

0

Square (6

0-1)

#7 (C

ntrl A

vg)

#2 (1

60-1

)

#4 (5

00-5

)

#1 (8

0-1)

#5 (8

0-5)

#6 (5

00-1

)

Plateau Details

Results (SEM images of tool wear)

Results (tool wear vs. track length)

• The longer the track length, the more the measured

tool wear across the cutting edge radius

Results (tool wear vs. track length)

• The longer the track length, the more the measured

rake & flank wear

• Flank wear > Rake Wear

Results (surface roughness comparing 1µm/rev

& 5µm/rev feeds)Surface roughness, Ra (As Received vs. Pass 1) for a 1µm & 5µm feed

rate

2

2.5

3

3.5

4

4.5

5

Ra

(µ

m) As Received

Pass 1 (1µm feed)

Pass 1 (5µm feed)

• A second set of experiments were carried out to extend the tool life

• Changing the feed did not affect the surface roughness data significantly

0

0.5

1

1.5

#4 (500-5) #5 (80-5) #6 (500-1)

Plateau Details

Results (tool wear data comparing 1µm/rev &

5µm/rev feeds)

Comparison of tool track length for different feed rates

15

20

25

Tra

ck

Le

ng

th (

m)

1µm feed

• Life of tool was extended by increasing the feed

(due to shorter track length)

0

5

10

#6 (500-1) #4 (500-5) #5 (80-5)

Plateau Details

Tra

ck

Le

ng

th (

m)

1µm feed

5µm feed

Conclusion

• The 60-1 laser ablated sample proved to be the best combination for SPDT

• This combination yielded in low surface roughness, low cutting force and longer track length.

Project IV – Machining & Improving the Surface Roughness of Quartz by SPDT

• Advantages of Quartz:– High hardness (~9.8GPa)

– Large supply (most abundant non-metallic mineral on earth)

– Good optical properties (optical range from 180nm to – Good optical properties (optical range from 180nm to 2000nm)

• Goal of the experiment:– Obtain surface roughness (Ra) lesser than 50nm

– Predict tool wear and other machining parameters to be able to diamond turn a 14” quartz

Use of final 14” disk

• To be used as ABL device nose cover

• Mirror finish surface required for the above use

• Image courtesy of Boeing Corporation

Experimental Plan

• Preparatory machining will be conducted to:– To confirm that ductile regime SPDT can be carried

out

– Correlate the trust force and depth of cut – Correlate the trust force and depth of cut

– Adjust sample runout for force/machining stability

– Observe tool wear

• Final machining experiment (if preparatory experiments are successful)

SPDT setup for Spectrosil 2000

Results (preparatory exp.)

• Image of surface shows no signs for cracks/fracture

• Feed marks seen on machined region

• Tool showed no significant wear

Results (preparatory exp. machining chips)

• Continuous plastically deformed chips indicate ductile

mode machining.

• Both images were taken at a 400x magnification

Results (final machining)

Surface Roughness After SPDT

400

600

800

1000

1200

Su

rface R

ou

gh

ness (

nm

)

Ra

Rz

• A total of two passes were carried out to achieve the targeted Ra (achieved Ra~41nm)

• A feed of 1µm/rev was used for both passes

0

200

400

As Received Pass 1 (1µm) Pass 2 (500nm)

Pass Details

Su

rface R

ou

gh

ness (

nm

)

Results (final machining surface image)

Pass 1 Pass 2

• Images comparing the surface after Pass 1 and 2 were carried out

• No signs of brittle machining

• A very smooth surface is seen after Pass 2

Results (tool wear)

• SEM images show slight wear of the tool after Pass 1

• No tool wear was observed after Pass 2

Conclusion

• Table shows the recommended machining parameters for final 14” disk machining

• A low feed (1µm/rev) is recommended as higher feeds will worsen the surface

• From the 6” disk results, it is predicted that a single tool will last for each pass when attempting to machine the 14” disk

Ductile Machining vs. Brittle Machining

Ductile Brittle

Well defined & straight edges

Jagged edges & chipped material

Controlled material removal process

Hard to control as microcracks extend below the machined surfacethe machined surface

Final depth of cut can be predicted below the DBT depth

No direct control of the resultant depth beyond the DBT depth

Good surface finish and mechanical properties

Poor surface finish and could end in a catastrophic failure at times

Future Developments

• Subsurface damage analysis on SiC & quartz using scanning acoustic microscopy & Raman spectroscopy.

• Phase transformation (if any) using Raman spectroscopy.

• X-ray diffraction to identify crystal orientation & preferred machining direction (if any)

• Transmission Electron Microscopy on SiC & quartz • Transmission Electron Microscopy on SiC & quartz machined chips

• Experiment other tool & machining parameters (i.e. rake angle, tool nose radius, depth of cut and feed)

• Attempt laser assisted machining on ceramics

• Attempt to design a more efficient/accurate ductile machining model

Acknowledgements

• Dr.John A. Patten (Thesis Advisor)• Dr.Pnina Ari-Gur & Dr.Muralidhar Ghantasala (Thesis Committee)• Funding for this research:

– National Science Foundation (Grant #: 25 700 4160)– Third Wave Systems (Grant #: 29 700 9090)– Mound Laser & Photonics Center (Grant #: 25 700 9530)

• John Cernius for SEM training and imaging• Dr.Valery Bliznyuk (AFM)• Glenn Hall (Western Michigan University Machine Shop)• Dr. Rob Eversole (Western Michigan University Biological Imaging Center)• Dr. Rob Eversole (Western Michigan University Biological Imaging Center)• Bill Sloan at Centre for Tribology Inc. (Tribometer technical support)• Ramesh Chandra (Tribometer training)

Family and friends who have continuously supported and encouraged me throughout my research work.

Thank you for your patience & attentionattention