KGD Probing of TSVs at 40 um Array Pitch

1
KGD Probing of TSVs at 40 um Array Pitch
Ken Smith, Peter Hanaway, Mike Jolley, Reed
Gleason, Chris Fournier, and Eric Strid

• 3D-TSV Probe Technology Goals
• MEMS probe tip evolution
• Contact performance
• TSV pad damage (or lack thereof)
• Conclusions

2
3D-TSV Probe Technology Development Goals

• Scale array pitch to 40 um
• Reduce pad damage to allow prebond probe
• Decrease cost of test
• Simplified, high yield process
• Fundamental understanding and accurate models of
  contact performance

3
Pyramid Probe Technology

• RF filters, switches
• Process monitors (including M1 copper)
• RFSOC Multi-DUT

4
3D Probing Requires a New Cost Structure
DRAM Flash
Logic/SoC
Vertical probe cost increases with density
4
2
3D Requires constant cost per chip
COGS/ pin () in 2012
1
Printed probe nearly constant cost per area
0.50
0.25
0.12
Array Pitch (um)
0.06
400
800
6
3
1600
200
100
50
25
12

• Technology must be printed, repairable, scalable,
  compliant

5
Scaling a Probe Card

• Decrease XYZ dimensions by K
• Same materials
• Decrease Z motions by K
• Force per tip decreases by K2 tip pressure
  constant

100 um pitch 10 gm/tip
35 um pitch 1 gm/tip
6
3D TSV Probe Card Architecture

• Pyramid Probe ST Pads on membrane
• Routing limitation 3-4 rows deep from DUT pad
  perimeter
• Replaceable contact layer

PCB
PCB
Plunger
7
Replaceable Contact Layer

• Tips are 5 um square and 20 um tall
• 35 um pitch array
• 24 x 48 tips

8
Contact resistance versus probing force

• Single 12 um square tip
• Sn plated wafer 5 um thick

9
Contact resistance versus probing force

• 6 um tip
• Force required is similar to 12 um tip

10
Force (gmf ) vs. Deflection (um)

• 1gmf /um tip design
• High durometer elastomer

11
Force (gmf ) vs. Deflection (um)

• 0.1 gmf tip design
• Low durometer elastomer

12
Pyramid Probe ST Routing

• Unique fine-pitch routing
• High-frequency performance similar to Pyramid
  Probes
• Example is memory array
• 50 um x 40 um pad pitch
• 40 x 6 pad array

13
Fully routed 6x40 array with 40-50 um pitch
14
Optical photograph of probe mark array

• Marks are exceptionally uniform
• 1 gram / contact for low pad damage

15
Profilometer scan of probe mark array

• Maximum depth 100 nm
• Maximum berm 500 nm

16
Probe marks on ENIG TSV pad

• Exaggerated conditions 10 TDs at 2.5 gf
• Navigation grid (50 x 40 um) shows 3 probe marks
  on the 100 um diameter pad

17
Probe mark depth less than surface roughness
(200 nm)
18
Probe mark on ENIG pad

• 3 x 7 um
• Exposed Ni 50
• Depends on surface grains

19
Probe mark uniformity Profilometer scans

• Depth Mean 68, Stdev 11
• Berm Mean 363, Stdev 76

20
TDR traces on open and short

• lt40 ps rise / fall times (100 ps / div)
• Limited by routing density in ST

21
Conclusions

• Practical probe cards are capable of 40 um pitch
  and tip forces below 1 gm
• Pad damage at these low forces is extremely small
  with scrub marks less than 100 nm deep
• Lithographically printed probe cards enable a
  scalability path to lower cost and finer pitches
• Probing the TSVs is not out of the question