Low-cost bumping Single chip Ni/Au UBM depositions were tested
Download
Report
Transcript Low-cost bumping Single chip Ni/Au UBM depositions were tested
WP6
INTERCONNECT TECHNOLOGY PART
Slides prepared by Sami Vaehaenen
Presented by M. Campbell
WP6 – 3D packaging part
• Manpower
– Sami Vaehaenen
– Timo Tick
– Michael Campbell
CERN Fellow 100%
ACEOLE PhD Student 100%
CERN Staff 20%
Recent Actions / Ongoing Work
• Low-cost bumping
–
–
–
–
Single chip Ni/Au UBM depositions were tested for Medipix2 chips at PacTech.
Looking forward to test PacTech’s solder ball placement systems for rapid solder deposition.
There have been negotiations with a Swedish company called “Smoltek” about collaborating
with development of state-of-the-art anisotropically conductive films (ACF) for radiation pixel
detectors.
Visit at S.E.T. in St Jeoire, the place where the state-of-the-art flip chip bonders are made.
• 3D test vehicle
–
–
3d test vehicle has been designed at CERN and it can be used in the development and
evaluation of all the three key technologies: through silicon vias (TSV), low cost bump
bonding and large-area BGA solder ball interconnections on different substrates.
Test vehicle mask design finished. Waiting for TSV or low-cost bumping technique
characterizations (daisy chain, Kelvin via and/or bump measurements.
• Through silicon via (TSV) project with VTT (funded by Medipix2
Collaboration)
–
–
TSV process development project with VTT has been started and masks are under
purchasing process.
Sami and Timo visited VTT to finalize the process flow for dummy wafers.
• Substrate technologies
–
Timo is searching suitable substrate candidates for 3d integrated structures (TSV’s). LTCC
and CCL are the most interesting candidates at the moment.
Low-cost Bumping Work at PacTech
•
•
•
•
•
•
•
•
PacTech has done some single chip Ni/Au UBM depositions for a handful of Medipix2 chips.
Prototyping process was quickly set up for the chips. There were some issues with plating
quality mainly because single chips are more difficult to plate than whole wafers.
It was shown that electroless Ni/Au can be grown uniformly even in very fine-pitch bump layout.
Encouraging results (pictures) were obtained for the use of electroless technology for MPX2.
Test batch of wafers is under preparation for characterization of yields with electroless Ni/Au
UBM’s. Electroless plated UBM’s could be used with solder or ACF’s.
Ni/Pd/Au metallization would be most reliable electroless solution Ni is strongly passivated
and it will remain solderable even if exposed for elevated temperatures for a long time.
Gang ball placement (GBP) technology is advancing and 40 µm sized solder balls have been tried
in process prototyping (animation).
40 µm sized solder spheres would be suitable for many of CERN’s present day detectors if they
could be transferred with high yields (> 99%). PacTech aims at 100% bumping yield with GBP
technology because of the visual inspection routine before bonding the solder spheres.
Reflow
Bonding
Flux deposition
Cleaning
Smoltek ACF Technology
•
Smoltek has a very interesting solid carbon nanotube growth process, which can be used in
fabrification of sub-micron resolution ACF’s.
•
Estimated fibre diameters are ≈ 30 nm. Fibres are grown in clusters and the pitch of
patterned clusters ≥ 100 nm.
•
Fibres can be metallized with nickel.
•
We are looking forward to collaborate with Smoltek. Smoltek would provides the ACF on
wafers and CERN would provide the data about the flip chip yields.
•
ACF
–
–
–
–
–
•
Nanofibres can be grown on many materials (Ni, Cu, Al, glass, etc…)
•
Restriction: only 4” wafers or wafer pieces can be deposited in the chamber at the moment.
fabrication process consist of:
Spinning of liquid catalyte.
Patterning by lithography.
CVD growth of carbon nanotubes at 400 ̊C (≈ 50 nm/min) on patterned areas.
Spinning of polymer on fibres.
Removal of the top surface of the polymer to expose the carbon fibre ends.
3D Test Vehicle
•
•
•
•
3D test vehicle has been designed for characterization of bump bonding, through
silicon via and BGA assembly processes.
Aim is to provide feedback for collaborating companies and to acquire reliability
data at CERN.
Measurements can be done for daisy chains on three layers, Kelvin bumps at
sensor– ROC interface and Kelvin TSV’s on readout chips. Possibility for very
demanding through package testing from 4 outputs on carrier.
The layout enables advanced packaging technologies, such as chip-to-wafer and
wafer-to-wafer bonding, for future interconnection development.
Probing pads can be diced off for 2D
tiling of structures.
Benefits
−Easily measureable test platform.
−Flip chip bonding yield and low-cost bumping
technology evaluation.
−Via yield and reliability testing.
−BGA interface optimization and evaluation.
−Development of assembly procedures.
−Parallel development of vias, low-cost
bumping, re-routing layers and BGA interface.
−Create reference comparison between
foundries.
Potential Substrate Technologies for 3D
•
Glass epoxy laminates (FR4): Traditional circuit board
–
–
•
Cheap and very mature technology
High density wiring (HDI built-up technology)
Low Temperature Cofired Ceramics (LTCC): Multilayer ceramic
circuit board
–
–
•
Mature technology
High density wiring
Carbon Composite Laminates (CCL): FR4 board with carbon
composite sheets
–
–
Compatible with traditional process -> cheap
Compatible with HDI built-up technology
Silicon
CTE [ppm/K]
Thermal Conductivity [W/mK]
2.6
13
Low Temperature Cofired
Ceramics (LTCC)
6-7
2
Density [g/cm2]
Relative Permittivity
Young’s modulus [GPa]
Via size [um]
Through substrate via size [um]
Line width
Line pitch
Substrate thickness
2.3
11.7
130
<1
10-50
<1
<1
100-300
3
5-8
350
100
100
100
200
500-5000
Glass Epoxy
Laminates (FR-4)
16
0.25
Carbon Composite
Laminate (CCL)
1-10
80
1.9
4-5
15
50
100
50
100
500-5000
1.55
NA
170
50
100
50
100
500-5000
TSV Development Project with VTT
•
Medipix2 collaboration has started a TSV development project with VTT.
The aim is at developing a Cu TSV process, which could be later used in
production of detector assemblies.
•
In the beginning, process is developed with blank wafers. After
establishing a TSV process, we will proceed with the real process wafers.
•
The complete 3D integration process is complex and many process entities
have to be at good level to build an operational device in the end.
•
Via filling plays the most demanding role in 3D integration process
•
3D test vehicle for electrical measurements will be created by CERN, which
is used for characterization of processes.
•
Process flow for dummy wafers was finalized at VTT in the end of May.
Actions to Come
•
Batch of test vehicle wafers has to be prepared for the low cost bumping
work.
–
–
–
–
UBM depositions will be continued at PacTech.
Prototyping with gang ball placement technology at PacTech.
Test state-of-the-art anisotropically conductive adhesives for flip chip.
Smart Equipment Technology (S.E.T.) should provide some simulation results
for advanced flip chip bonding systems.
•
32 pin probe card has to be designed and ordered for the 3D test vehicle.
•
Looking forward to first TSV wafers processed by VTT.
•
Sami and Timo are in Electronics Components Technology Conference in
San Diego, U.S.