Maekawa et al.: High-Speed Laser Plating for Wire-Bonding Pad Formation (1/7)
[Technical Paper]
High-Speed Laser Plating for Wire-Bonding Pad Formation Katsuhiro Maekawa*, Kazuhiko Yamasaki*, Tomotake Niizeki*, Mamoru Mita**, Yorishige Matsuba***, Nobuto Terada***, and Hiroshi Saito*** *Ibaraki University, 4-12-1 Nakanarusawa, Hitachi 316-8511, Japan **Mita Engineering Office, 3-16-14 Tajiri, Hitachi 319-1416, Japan ***Harima Chemicals, Inc., 5-9-3 Tokodai, Tsukuba 300-2635, Japan
(Received July 24, 2010; accepted September 22, 2010)
Abstract The present paper proposes high-speed laser plating for forming wire-bonding pads on a Cu leadframe using Ag nanoparticles. The novelty of the process lies in the implementation of drop-on-demand laser plating on the specially designed leadframe. Various aspects of the proposed method are investigated, including experimental set-up, multistep ink-jet printing, laser-plating parameters, quality of the sintered film, and wire bondability. It is found that both the quality of the sintered Ag pad and wire bondability are comparable to those of an electroplated Ag film when the near-infrared CW laser irradiates the pad for a short time of milliseconds. The superiority of the high-speed laser plating is confirmed from the viewpoints of material consumption, the necessity of pre- and post-processing, thermal damage to the pad and substrate, and environmental protection.
Keywords: Laser Sintering, Metal Nanoparticles, Metallization, Plating, Ink-jet Printing, Patterning, Wire Bonding, Leadframe
1.
Introduction
However, thermal damage and adhesion to electronic
Conventional fabrication of functional films for elec-
substrates are problems to be solved. Recently, a process
tronic wiring and electrode formation relies on wet pro-
called laser sintering has been developed for gold or silver
cesses such as liquid cleaning, chemical etching, and elec-
nanoparticles.[2–4] Densification of metal nanoparticles
troplating, which need plenty of energy and resources. For
consists of such sequences as evaporation of solvents,
example, electroplating includes pre- and post-processing
decomposition of dispersant, necking of adjacent particles
procedures such as alkali degreasing, acid pickling, elec-
and grain growth. Near-infrared lasers with little absorp-
trolytic cleaning, water washing and drying. Besides, these
tion in the paste heat the substrate first, and develop metal-
conventional technologies are not compatible with the
lization up to the paste surface. As a result, easy evapora-
need for low-cost production and less environmentally
tion makes the sintered part denser, and interdiffusion
harmful emissions. As an alternative to these wet pro-
between the substrate and sintered part yields firm adhe-
cesses, ink-jet printing with metal nanoparticles together
sion.[5]
with an additional metallization process are attracting
In the present paper, the laser sintering method with Ag
much attention.[1] This printed electronics technology
nanoparticles is proposed as a tool for the formation of
enables us to make conductive patterns by applying a
wire-bonding pads on a copper leadframe. The 5-nm-particle
small amount of metal nanoparticles only to the part where
silver paste is uniformly coated with ink-jet (IJ) printing as
the functional film is required.
large as around φ 100 μ m on the leads. Then, metallization
The conductive pattern is mainly obtained by a process
of the paste is completed with laser irradiation of a milli-
of thermal curing, in which a large depression of melting
second order. We name this functional-film formation pro-
point can be utilized; when the particle diameter is smaller
cess high-speed “laser plating” as an alternative to electro-
than 5 nm, metallization takes place at a low temperature
plating. The laser-plated film is observed and analyzed with
of below 250°C with a holding time of around 60 min.
FIB-SIM, TEM, XPS and laser scanning microscope
7
Transactions of The Japan Institute of Electronics Packaging
Vol. 3, No. 1, 2010
(LSM), and its thickness and flatness as well as its metal-
ing.[5] The Ag microstructure sintered by visible lasers,
lographic structure are discussed. Then, wire bondability
488 nm and 532 nm in wavelength, was more porous than
between the Ag pad and an Au wire is examined by a pull
that sintered by near-infrared ones, 980 nm and 1064 nm
test. Finally, these experimental results are compared with
in wavelength. The specific resistivity of the Ag film sin-
those of furnace curing and electroplating.
tered by the near-infrared laser was about 5 μ Ω •cm, which is smaller than that produced by the visible one. The rapid
2.
High-speed Laser Plating
metallization starting from the paste surface with the visi-
Laser plating is defined as metallization of nanoparticles
ble laser makes the removal of solvent and dispersant dif-
with laser irradiation for the purpose of forming functional
ficult, resulting in an insufficient sintering with large pores.
films. Basically, the process consists of the following:
Near-infrared lasers with little absorption in the paste are
(1)
Metal nanoparticles with dispersant and solvents
more effective than visible ones in obtaining a dense metal
are pasted on the substrate by various methods
structure.
including IJ printing and spin coating. (2)
A short preheating is necessary to remove organic substances in the paste.
3.
Experimental Figures 2 and 3 show the experimental apparatus used
The paste is metallized by a millisecond-order irra-
for laser plating. The IJ printer consists of IJ head, XY
diation of a laser beam under atmospheric condi-
stage, ink reservoir, and controller, having 128 micro noz-
tions.
zles, an ink discharge volume of 11 pL and a resolution of
Not only metallization by sintering but also inter-
1200 dpi. As soon as printing finishes, the sample is heated
diffusion or fusion takes places at the coating-
on a hot plate to remove solvents in the paste. Then, it is
substrate interface, leading to firm adhesion there.
placed on a stage in the Nd:YAG laser equipment
Figure 1 schematically shows the laser plating method
(SOS8956QSS, LASER SOS Ltd.) and sintered by moving
with IJ printing. With Ag nanoparticles, we have succeeded
the XY stage under a laser wavelength of 1064 nm; the
in wiring on polyimide.[5] Padding on a Cu leadframe is
beam diameter is around 0.2 mm. A continuous-wave
the subject of the present study. In comparison with elec-
mode is preferable to a pulsed output of beam power.[4]
(3)
(4)
troplating, no special attention is paid to pretreatment before the laser processing, whereas chemical cleaning and degreasing, and thorough rinsing of the substrate prior to electroplating are essential. Previous findings show that sintering is largely affected by paste composition prior to laser irradiation; especially, the content of dispersant and solvents.[4] Bulk growth is boosted with less organic substances. Otherwise, insufficient sintering occurs or a porous structure is formed. An appropriate preheating condition was set at 100°C for 1 min on a hot plate when the silver NanoPaste® (NPS-J, Harima Chemicals) was used.[6]
8
Laser wavelength is another factor that influences sinter-
Fig. 2 Ink-jet printing as part of laser plating.
Fig. 1
Fig. 3
Schematic of laser plating using Ag nanoparticles.
Laser sintering as part of laser plating.
Maekawa et al.: High-Speed Laser Plating for Wire-Bonding Pad Formation (3/7)
The stage was scanned once at 4 mm/s in an argon atmo-
Round pad patterns of around φ 100 μ m have to be
sphere with a flow rate of 3 mL/min. A stainless-steel jig
printed on the leads, and a flat pad surface with a thickness
was placed on the XY table to fix the leadframe.
over 2 μ m is required for wire bonding. However, it is well
Metal nanoparticles prepared by a gas evaporation pro-
known that the “coffee stain phenomenon” takes place as
cess have many advantages such as freedom from contami-
a droplet of a nanoparticle colloidal solution dries.[8, 9] If
nation, narrow size distribution, and broad range of met-
the contact angle of the droplet is less than 90°, and the
als.[1, 7] The silver NanoPaste® (NPS-J, Harima Chemicals)
ambient conditions encourage droplet drying, the droplet
was used in the experiment. The nanoparticles, being cov-
has a maximum evaporation rate at the boundary. Due to
ered by a protective compound, or an amine-type disper-
temperature and hence surface tension gradients, there
sant, are very stable. The TEM image revealed that the
results an effective flow of nanoparticles to the boundary.
size of nanoparticles was quite uniform with an average
When the droplet completely dries out, we are left with a
diameter of 5 nm. Neither aggregation nor precipitation
ring-like stain of nanoparticles which decreases in concen-
leading to a broad size distribution was observed. Table 1
tration from the periphery inwards.
summarizes the properties of the paste used in the exper-
In order to form a thick, flat pad, we can make use of this
iment. It has a metal content of around 65 mass%, and a vis-
effect by means of controlling discharge rate and substrate
cosity as low as about 9
mPa•s.
temperature during IJ printing, together with varying pre-
The standard method for connecting a bare die to a
heating conditions. Figure 5 illustrates the process of mul-
board is the chip-and-wire technique. Wire bonding is a
tistep IJ printing to produce discrete pads on the lead-
technique for the production of discrete electrical connec-
frame. An appropriate stage temperature is necessary for
tions, generally from a chip on a substrate. Wire bonding
controlling the spread of the droplet. The bank formed by
the connections must have suitable contact areas, or so-
the first droplet prevents the second one from overrunning
called pads. The pad or thin film serves to increase adhe-
it, and the third one fills the interior space. The first step
sive strength and reliability. Copper leadframes are often
plays the important role of controlling the wettability of the
used as a substrate, and the pad is formed at the top of the
second droplet as well as making a bank. Three steps are
lead. Figure 4 shows the Cu leadframe specially designed
sufficient for making a φ 100 μ m flat pad. After each step,
for laser plating. The leadframe consists of Cu/99.28
heating is required for reducing the solvents present in the
mass%, Cr/0.27 mass%, Sn/0.25 mass% and Zn/0.2 mass%,
paste as well as maintaining the pad profile.
having a thickness of 100 μ m, a lead width of 300 μ m, and
The Ag functional film thus obtained can be used as a
a line surface roughness of 0.06–0.07 μ m in Ra and 0.6–0.9
wire-bonding pad. A thin gold-nickel or silver electroplated
μ m in Rz. The leadframe was used as it was without any
pad is commonly used to increase bond strength. A ball-
chemical cleaning or degreasing, or thorough rinsing prior
and-wedge semiautomatic wire-bonder (HB10, TPT) was
to the IJ printing.
employed to bond an Au wire of φ 25 μ m on the sintered Ag film: ball bonding for the first bond and wedge bonding for
Table 1 Properties of Ag-nanoparticle paste before/after furnace curing.[7]
Before
After
Appearance
Dark blue
Particle diameter
3–7 nm
Metal content
62–67 mass%
Solvent
Tetradecane
Viscosity
7–11 mPa•s
Specific gravity
1.8–2.2
Curing temperature
220°C
Curing time
60 min
Appearance
Silver gray
Electric resistivity
3 μ Ω•cm
Metal content
99 mass%
Fig. 4 Cu leadframe for laser plating.
Fig. 5 Schematic of multistep IJ printing.
9
Transactions of The Japan Institute of Electronics Packaging
Vol. 3, No. 1, 2010
the second one. Figure 6(a) shows the semiautomatic
are visible; a lattice spacing of 0.2 nm can be seen in the
wire-bonder used in the experiment. To measure bond
TEM image. Taking the beam scan speed of 4 mm/s and
strength, the wire bonded on the sintered Ag film was
the scan distance of 100 μ m into account, Ag nanoparticles
pulled by a hook with a load meter, as shown in Fig. 6(b).
have been crystallized in the short time of around 75 ms.
Surface observation of the sintered Ag films was carried
As can be seen in Fig. 9, an XPS analysis revealed that
out by an LSM (VK-8700, Keyence Corporation), and
the top surface consists of almost all Ag and the interface
cross-sectional observation of the sintered layers was car-
between the film and the copper substrate contains some
ried out by FIB (FB-2100, Hitachi High Technologies).
oxygen. We have not yet identified the source of this oxy-
The sintered film was ion-milled stepwise and tilted by 30°
gen. Some diffusion takes place at the interface to form a
for SIM observation. The sintered film was also analyzed
thin diffused layer, which probably causes firm adhesion to
with TEM to confirm the formation of the Ag crystalline
the substrate used without any pre-treatments.
lattice and lattice spacing in the sintered Ag layer, and with
4.2 Multistep printing of 5-nm-particle Ag paste
XPS to measure chemical compositions.
Using the multistep IJ printing as illustrated in Fig. 5, we can overcome the coffee-stain effect and achieve a thin, flat
4.
Results and Discussion
4.1 Single-step printing of 5-nm-particle Ag paste Figure 7 shows the appearances of single-step IJ-printed and laser-sintered patterns: (a) an IJ-printed droplet after preheating at 100°C for 1 min, and (b) a laser-sintered one. The coffee-stain effect takes place: Ag concentrates at the periphery, forming a slight bank of around 0.5 μ m in height; the thickness is 0.2 μ m at the center. Figure 8 shows the cross-sectional image near the pad center: (a) FIBed SIM and (b) TEM images. In these fig(a) FIBed SIM image
ures, the Ag bulk structure and the Ag crystalline lattice
(a) Semiautomatic wire-bonder
(b) Pull test of bonded wire and fracture mode
Fig. 6 Wire bonding and bondability test.
Fig. 7
(a) IJ-printed (b) Laser-sintered Single-step IJ-printed and laser-sintered patterns on
Cu leads.
10
(b) TEM image Fig. 8 Cross-sections of sintered φ 5-nm-particle paste.
Fig. 9
Atomic percent profile of cross-section of sample
in Fig. 8.
Maekawa et al.: High-Speed Laser Plating for Wire-Bonding Pad Formation (5/7)
pad on the lead. Figure 10 shows the shape-controlled
A to E, in the figure indicate where breakage takes place
pads positioned at the Cu lead tips. The SEM image also
in the course of testing.
shows that no silver adheres to the side of the lead.
Table 2 summarizes the pull strength when the IJ-
Besides, no thermal damage such as oxidation to the lead
printed pads were prepared by preheating at 100°C for 10
takes place after the single-path laser irradiation through
min just before laser irradiation. The number of test sam-
the center of the pads from left to right.
ples was approximately 100. Very few wires separated from
A more detailed profile of a lead and pad is shown in Fig.
the pads; almost all broke at B or C. In the case of the mul-
11. The LSM images make clear that the coffee-stain effect
tistep-printed pad, the average pull strength is 8.6 cN, and
is almost resolved to yield a flat pad with a thickness of
the minimum one is 7.0 cN.
around 3 μ m. The plateau is as large as φ 100 μ m, which is
In comparison, the electroplated pad has average bond
large enough for a φ 10–25 μ m wire to be bonded. How-
strengths of 8.4 and 8.5 cN for the pad thickness of 0.2 and
ever, a few undesired cracks and voids can be seen on the
2.0 μ m, respectively, being close to the results of the laser-
pad.
plated film with the φ 5-nm-particle paste. In addition,
Figure 12 shows the SIM image of the FIBed cross-
breakage mode in the electroplated pad was more stable;
section. Using the multistep IJ printing method, we can
every wire breakage occurred at the middle of the wire, or
increase pad thickness from 0.2 μ m to 3 μ m. Note that the
at B or C.
cross-section has been tilted by 30° for observation.
Regarding the reliability of the wire-bonded leadframe,
Although full crystallization is not achieved and a porous
the specimen exposed in an atmospheric electric furnace
structure appears in the sintered Ag portion, it seems that
at a temperature of 150°C with a holding time of 1000 h
firm adhesion is obtained at the sintered Ag and Cu sub-
showed no changes in pull strength and fracture modes:
strate.
i.e., the average pull strength of 8.5 cN was maintained,
4.3 Wire bondability of laser-plated Ag pads
and separation did not take place at the pad; in all cases of
Wire bondability was examined between adjacent leads in the manner illustrated in Fig. 6, in which the symbols,
breakage only the wire broke. 4.4
Comparison with furnace curing and electro-
plating The IJ-printed Ag-nanoparticle paste can be metallized
(a) IJ printing (b) Laser sintering Fig. 10 Laser-plated Ag pads on Cu leads.
Fig. 12
FIBed cross-section of laser-plated Ag pad and Cu
lead.
Table 2
Pull strength of φ 25 μ m Au wire bonded to Ag pads.
Pad formation
Laser plating Fig. 11
LSM image of surface profile of laser-plated Ag pads
on Cu lead tip.
Electroplating
Thickness, μm
Strength, cN Max.
Min.
Avg.
0.2
10.3
6.0
8.2
3.0–3.2
10.5
7.0
8.6
0.2
9.9
7.0
8.4
2.0
10.1
7.6
8.5
11
Transactions of The Japan Institute of Electronics Packaging
Vol. 3, No. 1, 2010
5.
Summary High-speed laser plating for forming wire-bonding pads
on a Cu leadframe using Ag nanoparticles has been proposed. Its novelty lies in the implementation of drop-ondemand laser plating on the specially designed leadframe. Various aspects of the proposed method have been investigated, including experimental set-up, multistep printing, laser-plating parameters, quality of the sintered film with Fig. 13 FIBed cross-sections of furnace-sintered and electro-
FIB-SIM, TEM, XPS and LSM, and wire bondability
plated pads on Cu leadframe.
between the Ag pad and Au wire. Experimental results with Ag nanoparticles have been compared with those of
by furnace curing. The leadframe was heated under atmo-
furnace curing and electroplating.
spheric conditions in an electric oven at 220°C for 60 min.
It was found that the structural quality of the sintered Ag
Figure 13(a) shows the FIBed cross-section. Compared
pad was almost the same as that of an electroplated Ag
with Fig. 12, the structure does not become fully dense but
film, so that no difference in wire bondability was obtained
is more porous under the recommended curing condi-
when the near-infrared CW laser was irradiated for a short
tions. Curing is more developed at the paste surface, so
time: a millisecond order per lead. In comparison with fur-
that a dense layer is formed there, which prevents the
nace curing and electroplating, the superiority of the high-
evaporation of solvents. As a result, the interior of the pad
speed laser plating was confirmed from the viewpoints of
is more porous, as can be seen in the cross-sectional
material consumption (picoliter order), necessity of pre-
image.
and post-processing, thermal damage to the pad and sub-
The problem is that the Cu substrate suffers a degree of
strate, and environmental protection.
oxidation during the curing process. An oxygen atmo-
As for the reliability of the wire-bonded leadframe, other
sphere is required for curing the nanoparticles because
environmental and endurance tests in addition to the high-
carbon in the dispersant is removed with oxygen.[7] We
temperature storage test mentioned above should be car-
suggest that Cu may come up to the surface due to diffu-
ried out to confirm the advantages of the high-speed laser
sion, leading to poor bondability. In fact, wire bonding to
plating for wire-bonding pad formation.
the furnace-cured Ag pad was not successful, and most of the wires easily separated from the pad at A or E in Fig. 6. In the case of the laser-plated leads, no such thermal problems took place in the laboratory.
Acknowledgments This research was conducted as Practical Application Research and supported by JST Innovation Satellite Ibaraki.
We carried out wire-bonding to an electroplated pad
The authors would like to thank Director Katsutoshi Goto
using the same apparatus. Figure 13(b) shows the FIBed
and staff for their assistance. Acknowledgments are
cross-section of the electroplated Ag pad and Cu lead. Full
extended to Toshiyuki Asano, Ibaraki Prefectural Industrial
crystallization has been achieved, but the Ag surface is
Technology Center for testing and measurements.
rather rough: 0.27 μ m in Ra and 2.75 μ m in Rz, which is significantly higher than the laser-plated one: 0.09 μ m in
References
Ra and 1.51 μ m in Rz (measured area: 100 μ m × 100 μ m).
[1] H. Saito, M. Ueda, Y. Matsuba, and K. Oyama, “Liquid
Various pre- and post-processing procedures such as
wiring by ink-jet printing using NanoPaste®,” The 4th
alkali degreasing, acid pickling, electrolytic cleaning, water
International Workshop on Microelectronics Assem-
washing and drying, are indispensable in conventional
bling & Packaging, Kitakyushu, pp. 31–32, November
electroplating. In particular, the film plated on the side and
2004.
back surfaces of a leadframe must be removed before wire
[2] N. R. Bieri, J. Chung, D. Poulikakos, and C. P.
bonding. Otherwise, a masking process must be added to
Grigoropoulos, “Manufacturing of nanoscale thick-
the leadframe in the course of electroplating. In the pro-
ness gold lines by laser curing of a discretely depos-
posed method, however, no such additional operations are
ited nanoparticle suspension,” Superlattices and
required; the leadframe was used as it was, and the pad
Microstructures, Vol. 35, pp. 437–444, 2004.
was formed only on the top face of the lead.
12
[3] K.-S. Chou, K.-C. Huang, and H.-H. Lee, “Fabrication
Maekawa et al.: High-Speed Laser Plating for Wire-Bonding Pad Formation (7/7)
and sintering effect on the morphologies and conduc-
Ag Nanopaste Film and Its Application to Bond-pad
tivity of nano Ag particle films by the spin coating
Formation,” Proc. ECTC 2008, pp. 745–1750, 2008.
method,” Nanotechnology, Vol. 16, pp. 779–784, 2005.
[7] Y. Matsuba, “Conductive ink using metal nano parti-
[4] K. Maekawa, M. Mita, K. Yamasaki, T. Niizeki, Y.
cles,” CONVERTECH, Vol. 369, pp. 52–57, 2003.
Matsuba, N. Terada, and H. Saito, “Packaging of Elec-
[8] R. D. Deegan, O. Bakajin, T. F. Dupont, G. Huber, S.
tronic Modules through Completely Dry Process,”
R. Nagel, and T. A. Witten, “Capillary Flow as the
Proc. ECTC 2008, pp. 950–955, 2008.
Cause of Ring Stains from Dried Liquid Drops,”
[5] K. Maekawa, K. Yamasaki, T. Niizeki, M. Mita, Y.
Nature, Vol. 389, pp. 827–829, 1997.
Matsuba, N. Terada, and H. Saito, “Influence of
[9] T. Kajiya, D. Kaneko, and M. Doi, “Dynamical visual-
Wavelength on Laser Sintering Characteristics of Ag
ization of “coffee stain phenomenon” in droplets of
Nanoparticles,” Proc. ECTC 2009, pp. 1579–1584,
polymer solution via fluorescent microscopy,” Lang-
2009.
muir, Vol. 24, No. 21, pp. 12369–12374, November
[6] T. Niizeki, K. Maekawa, M. Mita, K. Yamasaki, Y.
2008.
Matsuba, N. Terada, and H. Saito, “Laser Sintering of
13