| |
| Comparing Flip-Chip and
Wire-Bond Interconnection Technologies |
TECHNICAL
BENEFITS
Technology
Thresholds
Current leading-edge flip-chip designs are in
production with as many as 5,908 bumps/chip at 200-µm pitch and
1,500 bumps/chip at 170-µm pitch in area array configurations.
Figure 3 shows a high I/O-flip-chip device at
178-µm pitch full area array. Future production area array pitches
will continue to reduce to 150-µm in the next two years and eventually
to the 100-125-µm range as the substrate technology becomes cost
effective. The rate of pitch reduction is likely to moderate as
design tools become available to take full advantage of the area
array capabilities of flip-chip technology.
 |
Figure
3:HIGH
I/O |
Current mass production for wire bonding is in the
60-µm range, and a production capability of 50-µm is believed to
represent the leading edge in wire bonding.
 |
Figure 4 shows development results for very
fine 35-µm pitch wire bonds. Wire bonding technology in this
range is feasible, and equipment that will allow manufacturers
to reach this threshold is under development.
|
Figure
4:
FINE
- PITCH WIRE BONDS |
Getting
the Lead Out
An inherent advantage of thermosonic/ ultrasonic bonding
is that there is a true weld, joining the two metals through an
inter-metallic phase.
Conversely, all flip-chip devices use solder
interconnects, formulated in two, three- and four-part (binary,
ternary and quaternary) alloys. Solder is a metallic alloy with
a low melting point.
The solder melts (reflows) and the liquid phase
wets the two metals that are joined. When the solder solidifies,
it provides a mechanically strong joint with low electrical resistance.
The best solder alloys for flip-chip applications are the Sn/Ag/Cu
alloy systems, with the possible inclusion of an additional element.
A solder bumping technology, such as solder paste, is preferred
over plating to achieve the control of the elements in the alloy
as bumped.
Typical electrical solders contain lead-this
is expected to change as lead-free alloys are introduced. Lead-free
telecommunications devices will be on the market this year.
Green marketing is the primary driver for the
development and commercialization of lead-free solder technology.
Once the first device marketed and advertised with lead-free logos
and designated as environmentally friendly enters the marketplace,
it will be a primary qualifier for an OEM in doing business.
Ternary and quaternary lead-free solder alloys
will be the primary choices for these products.
An additional benefit of lead-free alloy systems
is that they have the potential to minimize the soft errors produced
by lead alloys. These errors are caused by alpha particle emissions
from radioactive isotopes common in Pb-based alloys.
In Japan, many OEMs will specify lead-free to
meet upcoming legal requirements. In Europe and the U.S., the telecom
market will demand this feature as a basis for conducting business.
Reliability
Flip-chip processes include fluxing, placement and reflow,
as well as the underfill material and dispensing processes. Under-fill
is normally required for flip-chip devices on laminate substrates
to eliminate solder fatigue.
This fatigue is caused by cyclic strain resulting
from a TCE mismatch between the chip and the substrate during thermal
cycling. The underfill must be stiff enough to restrain the large
expansion/ contraction of the laminate and reduce the strain on
the solder bump. A well-characterized flip-chip assembly process
can achieve high yields and reliability.
Fine-pitch wire bonding is also a complicated
process. Molding fine-pitch wire bonds requires additional process
capabilities to avoid wire sweep and deflection.
Finer diameter wires, required to achieve fine
pitch, are not as strong or as stiff as larger diameter wires. To
address these constraints, manufacturers are developing higher strength
wire alloys, copper-wire bonding and encapsulant formulations to
minimize sweep at fine pitches and with long wires.
Through process integration, additional supplier
support, and improved equipment and materials capabilities, fine-pitch
wire bonding also can achieve high yields and reliability.
Cost
Analysis
Cost comparisons and analyses are highly dependent on modeling
conditions and, in particular, are sensitive to the choice of substrate
technology. Optimum substrate choices often change through the life
of a product, depend/ing on production volume, substrate layers,
microvias5 vs. conventional multilayer, among other considerations.
Technical reasons requiring flip-chip's electrical
performance or form factor are usually the most important criteria
for a change from wire bonding to flip chip.
Wafers designed specifically for area array
flip-chip may have smaller die, because peripheral bond-pad layout
(in pad-limited devices) requires a larger die perimeter to escape
the required I/O. Die size is a dominant factor in the cost of a
chip and represents a significant advantage for flip-chip.
This advantage, however, may be reduced, because
the additional substrate density, required to escape the area array
flip-chip bumps, may be more costly and require additional substrate
layers or microvia technology.
High-Speed
Packages
Three years ago the SIA roadmap6 and other sources
predicted that the speed required for new devices would soon exceed
the capabilities of wire bonding and that flip-chip would be the
interconnection method able to meet the speed requirements of advanced
packaging.
Today, the same prediction can be made, but
the edge of the envelope has been redefined. RAMBUS' RDRAM and board-on-chip
packaging is being produced with wire-bonded interconnections operating
at speeds of 400 MHz, which were not considered feasible only a
few years ago.
Wire-bonded RF packages with speeds of 2.4 GHz
are available today. Future packaging designs, with short wire lengths
and very fine bond pitch, will continue to push out the edge of
the envelope for the foreseeable future.
Flip-chip still has a technology advantage,
but wire bonding remains highly competitive.
The
Future
Both wire bonding and flip-chip will continue to coexist
as growth technologies into the foreseeable future, and the IC packaging
market will remain a continuum of technology choices.
Where flip-chip is advantageous, it will be
the production method. As the flip-chip infrastructure is established,
costs will be reduced and the application space will broaden.
Where applications can be produced by wire bonding,
its cost advantages will determine the interconnect method.
New technologies, such as WLP, WLBI (wafer-level
burn-in), stacked packages and optical interconnects, as well as
incremental advances in current technologies-such as flip-chip and
wire bonding-will continue to redefine the continuum of product
and process solutions for the interconnection of the die to the
external world.
References
- P. Elenius, "Au-Bumped Known Good Die," Proceedings 1994 ITAP
& Flip-Chip Conference, San Jose, February 15-18, pp. 94-97.
- L. Levine and I. Hanoon, "Bonder and Tool Design Choices for
CSPs," Chip Scale Review, July/August 1999, pp. 46-49.
- J. Lau, "Flip-Chip and Wire Bond for CSP", Chip Scale Package:
Design, Materials, Processes, and Applications, McGraw-Hill, New
York, 1999, pp. 1-41.
- N. Gamini, "Chip Scale Packaging for High-Speed RDRAM Memory
Applications," Proc, SEMICON West, San Jose, July 13-14, 1999,
pp. C1-C14.
- Prismark Partners LLC, "Microvia Update," July 1999, Cold Spring
Harbor, N.Y.
- Semiconductor Industry Association Roadmap, [sematech.org/ntrs/rdmpmem.nsf]
|
Mr. Elenius is vice president of technology
and chief technical officer for Flip Chip Technologies, a
joint venture of Delphi-Delco Electronics and Kulicke &
Soffa Industries in Phoenix, Ariz. Prior to joining FCT, he
was flip-chip product manager for Kulicke & Soffa. Earlier,
Mr. Elenius managed flip-chip/MCM equipment and process groups
at IBM Microelectronics. He earned a bachelor's degree in
mechanical engineering and a master's degree in manufacturing
systems from the University of Wisconsin, Madison, Wisconsin.
[pelenius@flipchip.com] |
|
Mr. Levine is principal metallurgical engineer for Kulicke & Soffa's Advanced Bonding Systems Division, Willow Grove, Pennsylvania. Last year, he received the John A. Wagnon Technical Achievement Award from the International Microelectronics and Packaging Society. He has been granted four patents and has numerous published works. Prior to joining Kulicke & Soffa, he was senior development engineer at AMP Inc. and chief metallurgist at Hydrostatics Inc. Mr. Levine received a bachelor's degree in metallurgy and materials science engineering from Lehigh University, Bethlehem, Pennsylvania.
[llevine@kns.com] |
|
|