Zincate-free, electroless nickel deposition on aluminum bond pads

Article abstract of DOI:10.1149/1.1836131

A zincate-free electroless nickel deposition on aluminum bond pads is investigated. A three-step, etch, rinse, and electroless plating, is demonstrated for deposition on aluminum bond pads patterned with polyimide. The chemicals used are compatible with this dielectric material. The deposition has been achieved with good selectivity, uniformity, and deposition rate at 40 × 40 μm aluminum bond pads. The adhesion and contact resistance were also determined and improved through anneals in the range 200-400°C for 1 h. The optimized condition for adhesion and contact resistance was an anneal at 400°C. The combination of a nickel hypophosphite reducing agent and the additives used leads to an active plating bath in the early stages of deposition, by comparison with commercial solutions, and hence, good coverage of the aluminum bond pad using the simplified process.

Full text of DOI:10.1149/1.1836131

C32
Journal of The Electrochemical Society, 152 ͑1͒ C32-C35 ͑2005͒
0
013-4651/2004/152͑1͒/C32/4/$7.00 © The Electrochemical Society, Inc.
Zincate-Free, Electroless Nickel Deposition on Aluminum Bond
Pads
,z
a
b
James F. Rohan,* Patricia A. Murphy, and John Barrett
NMRC, National University of Ireland, Cork, Lee Maltings, Prospect Row, Cork, Ireland
A zincate-free electroless nickel deposition on aluminum bond pads is investigated. A three-step, etch, rinse, and electroless
plating, is demonstrated for deposition on aluminum bond pads patterned with polyimide. The chemicals used are compatible with
this dielectric material. The deposition has been achieved with good selectivity, uniformity, and deposition rate at 40 ϫ 40 ␮m
aluminum bond pads. The adhesion and contact resistance were also determined and improved through anneals in the range
200-400°C for 1 h. The optimized condition for adhesion and contact resistance was an anneal at 400°C. The combination of a
nickel hypophosphite reducing agent and the additives used leads to an active plating bath in the early stages of deposition, by
comparison with commercial solutions, and hence, good coverage of the aluminum bond pad using the simplified process.
©
2004 The Electrochemical Society. ͓DOI: 10.1149/1.1836131͔ All rights reserved.
Manuscript submitted March 15, 2004; revised manuscript received June 16, 2004. Available electronically December 2, 2004.
Electroless nickel deposition on aluminum usually requires a se-
reflow over ͑SRO 702͒ by ATV Technology, Inc. The adhesion of 40
␮m high nickel bumps was measured using a Dage BT 24 shear
tester from Dage Precision Industries, Inc. ͑CA͒. In these tests
nickel bumps were sheared from the aluminum substrate using a 100
␮m tool head which has a lead edge perpendicular to the substrate
and travels parallel to the substrate at a rate of 500 ␮m/s at a height
of 5-10 ␮m. The force to shear the bump recorded is the average of
15 tests for each substrate.
ries of pretreatment steps to ensure good quality deposits. The pre-
treatments are necessary to remove the native surface oxide film on
the aluminum and to prevent reoxidation of the etched aluminum.
1
The most common pretreatment employed in industry and more
recently in chip-bump processing2 is the zincate treatment which
involves immersion of the aluminum substrate in a concentrated
sodium hydroxide-based solution containing zinc ions which re-
moves the oxide and deposits a protective zinc layer. It is usually
necessary to repeat the process to obtain the best quality coverage.
Upon immersion in the electroless nickel plating solution, the zinc
dissolves and is replaced by nickel ions, which initiates the autocata-
lytic deposition of nickel directly at the aluminum surface. In mi-
croelectronics applications this process has some potential draw-
backs, including process complexity and the exposure of the
substrate to undesirable metal ion contaminants such as zinc. For
these reasons alternative activations of the aluminum substrate for
electroless nickel deposition have been investigated, such as HF
-6
An alkaline etch composed of 50 mL each of H O and NH OH
2
2
4
of the concentrates ͑30% semiconductor grade, supplied by Ashland
Chemicals, OH͒ in 1 L of deionized water was used to clean the
aluminum surface and remove the oxide layer. An optimized etch
time of 30 s removed 0.3 ␮m of aluminum; the native aluminum
oxide was removed and the surface was slightly roughened. The use
of an alkaline etch step in the plating process was critical to the
activation of the aluminum bond pads to electroless nickel plating.
After etching, the wafer was immediately rinsed in propan-2-ol. This
solvent rinse was used, rather than an aqueous rinse, to help prevent
the reoxidation of the fresh aluminum surface. The sample was
transferred directly from the solvent to the plating solution. The
plating bath operates at a temperature of 87-90°C and the boiling
point of propan-2-ol is 84.2°C; thus, any propan-2-ol remaining on
the aluminum pads after the solvent rinse evaporates on immersion
of the samples into the plating bath. An electroless nickel bath for-
mulated from its constituents ͑Table I͒, and three commercial baths
were examined. Details for the commercial solutions have been de-
7
etching, deaerated rinsing solutions, and displacement nickel depo-
8
sition or activation by dimethyl amine borane ͑DMAB͒ solution,
and laser irradiation of the aluminum to remove anodically formed
9
oxide. An additional process step, the deposition of a thin Ni-B
base layer from alkaline solutions, has also been required in some
4
,8
cases before deposition from the more common hypophosphite
could take place. This paper describes a simple etch, rinse, and
plating process for the selective deposition of electroless Ni-P
bumps on aluminum bond pads for use in flip-chip assembly.
1
0
scribed earlier. All baths were based on a hypophosphite reducing
agent and operated at a pH of 4.0-5.0. The bath in Table I was also
designed to be sodium ion-free, utilizing nickel hypophosphite as
the reducing agent. This electroless nickel bath afforded selective
uniform plating on the aluminum bond pads. The commercial elec-
troless nickel baths were not sufficiently active to initiate nickel
plating on the aluminum surface using the simplified alkaline etch
and solvent rinse process.
Experimental
All experiments were performed in glass beakers with magnetic
stirring at 100 rpm. The plating solution was heated on a hot plate
with a Teflon-coated temperature probe to maintain a constant tem-
perature. The pH was adjusted using ammonium hydroxide to the
required value at room temperature and then heated to the operating
temperature. Teflon substrate holders were used. The substrates were
patterned aluminum ͑1.5 ␮m thick͒ bond pads (40 ϫ 40 ␮m) de-
posited using an Endura AMAT 5500 physical deposition tool. The
patterning material was 3 ␮m thick polyimide ͑PI͒ The deposit se-
lectivity and morphology were analyzed using a Hitachi S-400 field
effect scanning electron microscope ͑SEM͒ with a PGT IMIX
energy-dispersive X-ray ͑EDX͒ system with intensity correction for
elemental analysis of deposit composition and uniformity. Deposit
height and uniformity was determined using a Tencor Alpha-Step
Results and Discussion
Plating process.—Nickel bumps were selectively deposited us-
ing a simple process that did not require the use of an additional
metal deposition step such as occurs in the standard zincate process.
Table I. The composition of the electroless nickel bath.
Chemical
Molarity
2
00 surface profilometer and correlated with data recorded on the
SEM. Samples were annealed under a nitrogen ambient in a solder
Nickel sulfate
Ammonium sulfate
Nickel hypophosphite
Lactic acid
Calcium succinate
pH
0.05
0.25
0.13
0.22
0.01
*
Electrochemical Society Active Member.
a
Present address: Intel Ireland Ltd., Leixlip, County Kildare, Ireland.
Present address: Cork Institute of Technology, Bishopstown, Cork, Ireland.
E-mail: james.rohan@nmrc.ie
b
z
4.0-5.0
85-95°C
Temperature
Downloaded on 2015-04-13 to IP 128.122.253.212 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 152 ͑1͒ C32-C35 ͑2005͒
C33
Figure 2. Growth lines for electroless and electrolytic deposits on aluminum
bond pads.
grows both vertically and laterally at the same rate as shown in the
cross-sectional SEM image of Fig. 2. The lateral growth causes the
bump to spread onto the passivation surrounding the aluminum bond
pad. The adhesion of this lateral plating to the passivation, in this
case PI, may be low. In electrolytic metallization of bond pads, a
metal seed layer is deposited which covers both the aluminum bond
pad and the passivation. Photolithography is used to constrain bump
growth to the required areas over the bond pads and the bump ma-
terial is deposited only over metal which provides adequate adhe-
sion. The nickel-to-aluminum adhesion required from electrolessly
plated deposits therefore must be greater than the comparable bump
plated by electrolytic means. To improve the bump-wafer adhesion
with electroless plating, either the nickel to passivation or the nickel
to aluminum adhesion, or both, must be increased.
Chemical and physical modification of the PI prior to plating was
studied as a method to increase the Ni-PI adhesion. The factors
examined in order to increase the Ni-Al adhesion included improv-
ing the plating process and annealing the samples after plating. Ini-
tial bump shear strength tests indicated that the bump-to-wafer ad-
hesion strength was 35 g per pad for the 40 ϫ 40 ␮m metal
connection, when the nickel was deposited at pH 4.2 and a plating
bath temperature of 85°C. The results of the bath operating condi-
tions and substrate treatments which lead to a large increase in
nickel-to-wafer adhesion are given in Table III.
Figure 1. Selective electroless nickel bumps deposited in 3 h plating on
aluminum 40 ϫ 40 ␮m bond pads: ͑a͒ SEM image of nickel bump array and
͑b͒ cross section of a nickel deposit.
The physical treatments to roughen the PI surface included an
oxygen plasma ash and abrasion of the surface using silicon carbide
paper. The SEM image in Fig. 3 illustrates the surface of the PI
following abrasion using silicon carbide paper. All treatments that
roughen the PI surface lead to an increase in adhesion shear
strength. Oxygen plasma etching of the PI prior to plating afforded
the largest shear strength increase; a value of 130 g/pad was mea-
sured which compares very favorably with shear strength values
The plating process utilized consisted of an etch step and a solvent
rinse prior to plating. This metal deposition method has a number of
advantages over other aluminum activation processes such as zinca-
tion or palladium activations. First, the sample is not exposed to
additional species such as zinc, sodium, or palladium, and second,
the process is selective and the rear of the wafer is not plated.
Initial plating experiments using the acidic nickel bath operating
at a pH of 4.2 and operating temperature of 85°C afforded nickel
plating at a typical rate of 8 ␮m/h. Increasing the operating tempera-
ture to 90°C and solution pH to 4.6 increased the deposition rate to
5
quoted for electroless nickel bumps. Chemical treatments included
alkaline and acid solutions and an n-methyl-pyrrolidinone ͑NMP͒
dip. NMP is used to strip PI at elevated temperatures and softens the
material at lower temperatures. Chemical treatment of the samples
prior to plating did not, however, lead to an increase in the bump-
wafer adhesion.
To further improve the overall adhesion, the effect of annealing
on the plated die was also examined. Samples were annealed at a
range of temperatures between 200 and 400°C. Annealing was found
20 ␮m/h. Figure 1 illustrates a selectively plated array of 58 ␮m
high nickel bumps on a wafer that was plated in the acidic nickel
bath for 3 h. The plating uniformity of the electrolessly deposited
bumps is indicated in Table II, which lists the mean and standard
deviation for 30 sites on the substrate. The data indicates that the
deposit uniformity is maintained over extended deposition times.
Table III. Adhesion data for nickel deposits on substrates
sujected to listed treatment.
Adhesion.—In electroless plating, deposition commences auto-
catalytically only on the patterned aluminum of the bond pad. Ad-
ditional photolithographic patterning is not required. This is a sig-
nificant advantage of the electroless process. However, the deposit
Standard
Deposit height Shear strength
deviation
Treatment
͑␮m͒
͑g per pad͒
͑g per pad͒
Untreated sample
Oxygen plasma etch
Abrasion 4000 grit SiC
Abrasion 2500 grit SiC
Abrasion 1200 grit SiC
Anneal 200°C
40
45
30
30
30
40
40
40
66.0
130.1
84.8
89.6
95.7
61.8
95.2
173.2
11.1
8.3
4.2
5.1
4.0
7.2
21.1
30.7
Table II. Nickel bump height uniformity.
Number of
Mean bump
Standard deviation
Sample
measurement sites
height ͑␮m͒
͑␮m͒
1
2
30
30
52.52
51.90
1.61
2.40
Anneal 300°C
Anneal 400°C
Downloaded on 2015-04-13 to IP 128.122.253.212 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C34
Journal of The Electrochemical Society, 152 ͑1͒ C32-C35 ͑2005͒
Figure 3. SEM image is of an aluminum bond pad and polyimide after
abrasion using 4000 grit silicon carbide paper.
to increase the nickel-to-aluminum adhesion, with higher anneal
temperatures affording the greatest increase in adhesion. Shear
strengths of 170 g/pad were measured when samples were annealed
at 400°C in an oxygen-free environment for 1 h, again indicating
good adhesion for the small pad size.
SEM analysis was also used to examine the aluminum bond pads
of plated samples where the nickel bumps had been removed during
shear testing. Where samples had been annealed the surface of the
aluminum was rougher than samples that had not been heat-treated.
No nickel was detected by EDX at the bond pad following shear
test. This suggests that annealing the samples causes a localized
interaction between the nickel and the aluminum bond pads. This
interaction probably accounts for the increase in adhesion upon an-
nealing. Figure 4a and b shows SEM images of aluminum bond pads
for plated samples where the nickel bumps have been removed by
shear testing. The sample in Fig. 4a was not annealed while the
sample in Fig. 4b has been annealed at 400°C prior to shear testing.
Figure 4. SEM images of aluminum bond pads of plated samples where the
nickel bumps had been removed: ͑a͒ no heat-treatment and ͑b͒ annealed at
400°C prior to shear testing. No nickel was detected at the remaining alumi-
Auger analysis.—Auger depth profiling of a series of samples as
plated and after anneals at 200, 300, and 400°C did not show sig-
nificant interdiffusion at the Ni/Al interface. The profiled samples
were plated with a thin nickel deposit (ϳ5␮ m thick͒. Auger spectra
were acquired affording detailed information on the composition of
the nickel/aluminum interface. Comparing the spectra of the sample
as-plated and annealed at 400°C showed no detectable variation in
peak shape or shift at the interface areas. This suggests that the
oxidation states of nickel and aluminum do not change around the
interface region, and thus intermetallic compounds are not being
formed. Figure 5a and b illustrates the depth profiles of atomic com-
position of an as-plated sample and a sample annealed at 400°C. In
the spectra the Al is observed at the right side of the spectra, the
nickel and phosphorous are seen on the left side, and the oxygen
signal is at the bottom of the spectra. These results were supported
by EDX line scans which did not indicate a significant concentration
change at the interface following an anneal at 200-400°C.
num of the bond pad.
The resistance values measured on a series of samples are illus-
trated in Fig. 7. The resistance measured vs. the number of Ni/Al
interfaces is plotted for a series of samples that were measured as-
plated and annealed at 200, 300, and 400°C. All the resistance val-
ues measured indicate a low Ni/Al contact resistance per bond pad.
Annealing the samples affords a reduction in resistance values at the
Ni/Al interface. In Fig. 7 a dramatic decrease in resistance was
observed when samples were annealed at 400°C. Based on previous
11
analysis, the crystallization of the electroless nickel deposit may
be the factor contributing most to this result.
Based on the plating observations and deposit analysis, it is pro-
posed that the peroxide/ammonia etchant removes the oxide at the
surface of the aluminum, as indicated by the 0.3 ␮m decrease in
aluminum pad height following this treatment. The etched aluminum
surface is protected from reoxidation by immersion in the nonaque-
ous rinse solvent, which is similar to the protection afforded by zinc
in the standard zincation process. This solvent also protects the alu-
minum surface during transfer to the nickel plating bath and during
the initial seconds of immersion in that bath. The zincate process
affords a more durable protection and is better suited to the full
range of substrate dimensions. The solvent protection described in
this work protects the relatively small pad area adequately during
the time scale of removal from the solvent rinse and immersion in
the plating bath. Once in the plating solution the exposed aluminum
surface is most likely involved in a displacement reaction with the
nickel ions, again in a mechanism similar to that which occurs fol-
Resistance measurements.—The contact resistance between the
deposited electroless nickel and the underlying aluminum of the
wafer was investigated using two-point probe resistance measure-
ments. These measurements were carried out on bumped wafers that
had not been annealed and on samples that were annealed at tem-
peratures of 200, 300, and 400°C. Test structures ͑Fig. 6͒ measured
on the substrate involved multiple nickel-aluminum daisy chained
contacts. Using this structure, the resistance of many Ni/Al inter-
faces and Al links can be measured at one time with a two-probe
system. Measurements at this test structure provide a better repre-
sentation of the resistance of the system, as the interfaces are not
influenced by the pressure applied to the deposit during probing.
Downloaded on 2015-04-13 to IP 128.122.253.212 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

Journal of The Electrochemical Society, 152 ͑1͒ C32-C35 ͑2005͒
C35
Figure 7. Resistance measurements vs. Ni/Al interfaces. Samples as-plated
and after annealing at 200, 300, and 400°C.
lytic deposition that occurs in the early stages ͑0.3 ␮m/min expected
by extrapolating from the long term deposition experiments͒ does so
in a uniform manner across the bond pads at a high rate before the
aluminum reoxidizes. In addition to the increased deposition rate
under the optimized plating conditions, shear test analysis indicated
that the nickel has better adhesion to the aluminum when the nickel
is deposited at a higher rate. This most likely results from direct
nickel-to-aluminum contact, which occurs at a substantially oxide-
free surface.
Conclusions
A zincate-free route for electroless nickel direct deposition on
aluminum bond pads has been demonstrated. This technique is sig-
nificantly simpler, requiring only three steps, than the standard zin-
cate process and does not require the use of hazardous chemicals,
such as HF, employed in some cases as an alternative to the zincate
treatment. It also affords selective deposition which may not be
achieved using Pd or other such activating solutions to assist the
initial deposition on the aluminum. The deposition has been
achieved with good selectivity, uniformity, and at a relatively high
deposition rate of 20 ␮m/h at 40 ϫ 40 ␮m aluminum bond pads.
For flip-chip applications where the electroless nickel acts as an
under-bump metallurgy, adhesion and contact resistance are signifi-
cant factors. These have been determined and optimized through
anneals in the range 200-400°C for 1 h. Similar selective deposition
was not achievable with commercial plating solutions using this
three-step process. It is suggested that the combination of the active
plating bath with few additives, comprising a nickel hypophosphite
reducing agent, leads to a more active plating bath in the early
stages of deposition and hence, good coverage of the aluminum
bond pad using the simplified process described.
Figure 5. ͑a, top͒ Auger depth profiles of atomic composition of the Ni/Al
interfaces of samples as-plated and ͑b, bottom͒ after annealing at 400°C.
lowing zinc dissolution in the zincate process, given the potential
difference between the aluminum dissolution and nickel-ion reduc-
tion reactions. The plating bath used is more active for nickel depo-
sition at the operating pH of 4.6 and temperature of 90°C ͑20 ␮m/h
achieved͒ than at a pH of 4.2 and temperature of 85°C ͑8 ␮m/h
achieved͒, and more active than commercial solutions which did not
yield deposits using this technique. This may be due to the lower
additive content for the bath described in this work, which can yield
deposits at 20 ␮m/h from the zincate-free process. Therefore, it
would appear that the critical displacement and subsequent autocata-
Acknowledgments
This work was funded in part by Intel and Enterprise, Ireland,
under grant no. HE/98/262.
The National University of Ireland, Cork, assisted in meeting the publi-
cation costs of this article.
References
1
. J. B. Hajdu, Electroless Plating: Fundamentals and Applications, Chap. 7, G. O.
Mallory and J. B. Hajdu, Editors, American Electroplaters and Surface Finishers
Society Publishers, Orlando, FL ͑1990͒.
2
3
4
. K. Wong, K. Chi, and A. Rangappan, Plat. Surf. Finish., 7, 70 ͑1988͒.
. J. Liu, Hybrid Circ., 29, 25 ͑1992͒.
. A. M. T. van der Putten and J. W. G. de Bakker, J. Electrochem. Soc., 140, 2229
͑
1993͒.
. Y.-D. Jeon and K.-W. Paik, IEEE Trans. Compon. Packag. Technol., 25, 169
2002͒.
. A. J. G. Strandjord, S. Popelar, and C. Jauernig, Microelectron. Reliab., 42, 265
2002͒.
. C. H. Ting, M. Paunovic, P. L. Pai, and G. Chiu, J. Electrochem. Soc., 136, 462
1989͒.
5
6
7
͑
͑
͑
8
9
. H. Watanabe and H. Honma, J. Electrochem. Soc., 144, 471 ͑1997͒.
. S. Z. Chu, M. Sakairi, H. Takahasi, and Z. X. Qiu, J. Electrochem. Soc., 146, 537
Figure 6. Schematic of the daisy chain layout with aluminum pads con-
nected along the chain through electroless nickel deposit. Probing for resis-
tance is performed only at the start and end of chain, permitting the deter-
mination of the resistance of a series of nickel/aluminum interfaces.
͑
1999͒.
10. J. F. Rohan, G. O’Riordan, and J. Boardman, Appl. Surf. Sci., 185, 289 ͑2002͒.
11. J. F. Rohan and G. O’Riordan, Microelectron. Eng., 65, 77 ͑2003͒.
Downloaded on 2015-04-13 to IP 128.122.253.212 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).