Dicing of Sensitive MEMS Devices:

Challenges and Solutions

F. Lewis*, P. Wright*, D. Martin**, S. Michel* and L. Ouellet*
* TELEDYNE DALSA Semiconductor, 18 boul. de lAroport, Bromont, J2L1S7, Canada
** Disco Hi-Tec America, 3270 Scott Blvd., Santa Clara, CA 95054-3011 U.S.A
Teledyne DALSA Confidential and Proprietary
Outline
Dicing Overview
MEMS vs Dicing
Stealth dicing
How does it works?
Process flow
Critical parameters
Case study
Optical MEMS
Blade dicing
Stealth dicing
Conclusion
Further improvements
Teledyne DALSA Confidential and Proprietary
A brief MEMS primer: what they are, why the
are interesting?
MEMS
one product one process
moveable 3D structures
electrical and mechanical
performance is key
CMOS
one process
planar processing
linewidth name of the game
Teledyne DALSA Confidential and Proprietary
A brief MEMS primer: what they are, why the
are interesting?
Challenge: singulate released MEMS devices
Teledyne DALSA Confidential and Proprietary
MEMS vs Dicing
Blade dicing
Ablation laser Microjet
Scribe and break
http://www.csi-sensor.com.tw/eng/laserdicing.aspx
http://www.dynatex.com
http://www.synova.ch/english/laser-cutting-machine/laser-microjet-machines.html
Stealth dicing
http://www.disco.co.jp/eg/solution/library/stealth.html
Teledyne DALSA Confidential and Proprietary
MEMS requirements
Particle-free process
No water
No cleaning
Stealth dicing
Blanket Si
Teledyne DALSA Confidential and Proprietary
Stealth dicing - How does it works?
Patented by Hamamatsu Photonics (http://www.hamamatsu.com/)
The high temperature and high compressive pressure induce
Dislocation
Partial recrystallization
Microcrack
Lambda: NIR
http://jp.hamamatsu.com/resources/products/etd/pdf/SD_tech_TLAS9004E01.pdf
Partial transmission
Teledyne DALSA Confidential and Proprietary
Stealth dicing - How does it works?
Peak power density > 1 x 10
8
W/cm
2
(US patent 6,992,026)
Pulse width of 1 us or less
Localized temperature > 1000K in a volume of 10 um
3
for
nanoseconds
Dicing from the bottom to the top
Cross-section (after singulation)
Teledyne DALSA Confidential and Proprietary
Stealth dicing
Process flow
3 steps process
Released structures
Silicon wafer
Non-contact
Wafer Mounting Frame
Stealth dicing
Singulation by
Tape expand
UV dicing tape
Expanded
dicing tape
Hoop
Teledyne DALSA Confidential and Proprietary
Stealth dicing
Vendor summary
Company Wafers up to 8 Wafers up to 12
Disco DFL7340 DFL7360
Accretech ML200 ML300
Laser head
Thickness SDE01 SDE03
100 m 1 pass 1 pass
200 m 4 pass 2 pass
300 m 7 pass 5 pass
400 m 12 pass 5 pass
625 m 19 pass 8 pass
Data provided by Disco. Blank Si wafers with 1 Ohm.cm.
* Blue number indicates that it includes a mapping pass for Z axis adjustment
Teledyne DALSA Confidential and Proprietary
Stealth dicing - Critical parameters
Resistivity > 0.01 Ohm.cm
Highly doped wafers induce too much absorption
Kumagai et al. - IEEE Semiconductor Manufacturing (2007) p. 1-4
38 . 7
393 .


u
) exp( d I I
o

Teledyne DALSA Confidential and Proprietary
Stealth dicing - Critical parameters
Thickness vs Throughput (provided by Disco)
Wafer Thickness SDE01
UPH
SDE03
UPH
Blade Dicing UPH
100um 25 25 5 @ 25mm/sec
200um 9 16 9 @ 50mm/sec
300um 5 7 15 @ 100mm/sec
400um 3 7 15 @ 100mm/sec
625um 2 5 12 @ 75mm/sec
Based on 8 wafer, 5mm x 5mm die size
Blade dicing based on single pass dicing process
All numbers are estimates
Note:
Blade dicing feed speed changes depending on wafer thickness
SD feed speed remains the same for all thicknesses but number of passes increases
Teledyne DALSA Confidential and Proprietary
Stealth dicing - Critical parameters
Silicon vs SOI wafers
Blanket Silicon wafers SOI wafers
BOX
Teledyne DALSA Confidential and Proprietary
Stealth dicing - Critical parameters
SOI wafers
BOX
Teledyne DALSA Confidential and Proprietary
Stealth dicing - Critical parameters
Design consideration - exclusion zone
Around 40% of the wafer thickness (No metal, void, polymer residue,)
Die size difference between stealth dicing (No kerf width) and blade dicing
Stealth
dicing
Blade
dicing
Teledyne DALSA Confidential and Proprietary
Case study
Released MEMS Optical MEMS
Comparison between blade dicing with stealth dicing
Si resistivity : 0.01 Ohm.cm (near the limit)
6 wafers - SOI
Total thickness: ~500 um
Exclusion zone: 200 um
Die size : 3.5 mm x 8.5 mm (BD), 3.6 x 8.6 mm (SD)
Teledyne DALSA Confidential and Proprietary
Equipments
Blade dicer
ADT 7200 - Fully Automatic Dicing Systems
Blade: S3050 KnS
Dicing tape: Ultron 1042R
Stealth dicer
Stealth dicer: DAL-7360 with SDE03 laser head
Tape expander: Ultron UH130
Dicing tape: Ultron 1042R
Dicing recipe: 13 passes
ADT 7200
DFL 7340
Teledyne DALSA Confidential and Proprietary
MEMS vs Blade dicing
Extensive usage of water is incompatible with sensitive
MEMS structures
Teledyne DALSA Confidential and Proprietary
MEMS vs Blade dicing
Process flow
Process time: 70 min
1 Cap wafer
2 Wafer encapsulation
3 Wafer mounting
4 Blade dicing
5 UV cure
6 Detaping
7 Wafer mounting
8 UV cure
9 Decapsulation
Parameters to optimize
Encapsulation parameters
Applied force
Vacuum level
Dicing parameters
Spindle speed
Feed rate
Number of passes
Blade
Diamond size
Diamond concentration
Bond hardness
Decapsulation
UV curing time
Decapsulation force
Glass wafer
UV patterned tape
Teledyne DALSA Confidential and Proprietary
Blade dicing - Issues
Chipping (back chipping)
Slag, chip, cracks
Teledyne DALSA Confidential and Proprietary
Blade dicing - Issues
Contamination
Slag due to adhesion problems
Particles
contamination near
the die center
Chip
Teledyne DALSA Confidential and Proprietary
Stealth dicing
Process flow
Process time: 15 minutes
Parameters to optimize
Non-contact vacuum mounting
Air bubble
Stealth dicing
Laser power (W)
Laser passes position
Feed rate (mm/s)
Tape expand
Dicing tape
Push-up amount
Push-up speed
1 Non-contact wafer mounting
2 Stealth dicing
3 Tape expand
Teledyne DALSA Confidential and Proprietary
Stealth dicing
Yield Die rejection
Pass
Fail Optical tests
Fail Visual inspection
Blade dicing
45 / 416 - optical tests
104 / 416 - visual inspection
Stealth dicing
37 / 416 - optical tests
19 / 416 - visual inspection
Fail Process defect
Teledyne DALSA Confidential and Proprietary
Stealth dicing - Issues
Process dependant - Voids
Bonded
layers
Teledyne DALSA Confidential and Proprietary
Stealth dicing - Issues
Die spacing
Expansion level
Percentage of singulation
Tape tearing
Probing, Pick and Place (rubbing, alignment)
Teledyne DALSA Confidential and Proprietary
Stealth dicing - Minor Issues
Meandering
Meandering ~ 3-5 um
Teledyne DALSA Confidential and Proprietary
Stealth dicing - Minor issues
Zipper - Esthetic defect caused by reflected light on the
BOX
Teledyne DALSA Confidential and Proprietary
Conclusion
Stealth dicing Blade dicing
Contamination Low High
Process time Low High
Cleaning Not required Required
Yield High Low
Design limitation Medium Low
Material limitation Medium Low
Tape limitation Expandable None
Die spacing Process dependent Uniform
Stealth dicing can provide great cut quality
Must control the MEMS design and fabrication
For released MEMS
Teledyne DALSA Confidential and Proprietary
Further improvements
Blade dicing
Dust cap for frontside process
Lower chipping
Challenge: Fabricate dust cap and temporary bond on wafer
Stealth dicing
Back side process
Lower the exclusion zone
Challenge: Hardware modification to the stealth dicer and transparent dicing
tape
Teledyne DALSA Confidential and Proprietary
Acknowledgements
Teledyne DALSA:
Daniel Bourgoin
Robert Antaki
Manon Plante
Pierre-Louis Fortin
Disco Hi-tech America
Kevin Wade
Eric Brown
Disco Japan
Zhao Jinyan