Uniformity control of Ni thin-film microstructures deposited by through-mask plating

Article abstract of DOI:10.1149/1.1833320

The thickness uniformity within a specimen and the cross-sectional profiles of electroplated individual Ni microstructures have been investigated as a function of the electroplating conditions. It was found that the uniformity and profiles of microstructures could be controlled by varying the process conditions. A uniform thickness distribution and microstructures with flat profiles could be obtained at optimal plating conditions of 8 mA/cm² and 60°C. Above this optimal plating current density, the microstructure has a rabbit-ears profile, and the thickness of a narrow microstructure is thicker than that of wide ones. Below this current density, the microstructure has a cap-like cross-sectional profile, and a narrow structure is thinner than wide ones. Increasing the plating temperature enhances the nonuniformity, whereas other process parameters have insignificant effects on it. The current crowding observed in patterned specimens is responsible for the rabbit-ears profile of individual microstructures, while a combination of the fluidic friction on the sidewall of the photoresist and the electrophoresis of the ions in the solution are believed to be responsible for the abnormal cap-like profile of individual microstructures and the thinning effect on narrow microstructures.

Full text of DOI:10.1149/1.1833320

C36
Journal of The Electrochemical Society, 152 ͑1͒ C36-C41 ͑2005͒
0013-4651/2004/152͑1͒/C36/6/$7.00 © The Electrochemical Society, Inc.
Uniformity Control of Ni Thin-Film Microstructures Deposited
by Through-Mask Plating
J. K. Luo,a,z D. P. Chu,a,b A. J. Flewitt,a S. M. Spearing,c N. A. Fleck,a
and W. I. Milnea
aDepartment of Engineering, University of Cambridge, Cambridge CB2 1PZ, United Kingdom
bCambridge Research Laboratory of Epson, Cambridge CB4 0FE, United Kingdom
cDepartment of Aeronautics and Astronautics, Massachusetts Institute of Technology,
Cambridge, Massachusetts 02139, USA
The thickness uniformity within a specimen and the cross-sectional profiles of electroplated individual Ni microstructures have
been investigated as a function of the electroplating conditions. It was found that the uniformity and profiles of microstructures
could be controlled by varying the process conditions. A uniform thickness distribution and microstructures with flat profiles could
be obtained at optimal plating conditions of 8 mA/cm2 and 60°C. Above this optimal plating current density, the microstructure
has a rabbit-ears profile, and the thickness of a narrow microstructure is thicker than that of wide ones. Below this current density,
the microstructure has a cap-like cross-sectional profile, and a narrow structure is thinner than wide ones. Increasing the plating
temperature enhances the nonuniformity, whereas other process parameters have insignificant effects on it. The current crowding
observed in patterned specimens is responsible for the rabbit-ears profile of individual microstructures, while a combination of the
fluidic friction on the sidewall of the photoresist and the electrophoresis of the ions in the solution are believed to be responsible
for the abnormal cap-like profile of individual microstructures and the thinning effect on narrow microstructures.
© 2004 The Electrochemical Society. ͓DOI: 10.1149/1.1833320͔ All rights reserved.
Manuscript submitted March 3, 2004; revised manuscript received June 3, 2004. Available electronically December 7, 2004.
Electrodeposition ͑or electroplating͒ has become one of the most
Experimental
important technologies for microelectronics and microelectrome-
chanical systems ͑MEMS͒, as it produces high-quality metal films in
a simple way at a low cost. Various metal and alloy films such as
Ni, In, Cu, Au and NiFe, NiP, CoFe have been formed by
electrodeposition.1-3 A combination of photolithography technology
and electrodeposition ͑so-called through-mask plating4͒ makes it
possible to fabricate microstructures suitable for microsensors and
microactuators.5-7 However, one of the problems associated with
electrodeposition is the uniformity control of the electroplated films
both in thickness and composition.8,9 This affects the material prop-
erties and resulting performance of microcomponents.
Plating bath and solution.—The plating solution was commer-
cially purchased from Celtic Chemicals, Ltd., and consisted of
nickel sulfamate ͑300 g/L͒, nickel chloride ͑10 g/L͒, and boric acid
͑40 g/L͒. The pH value of the as-made plating solution was 3.8,
which gradually increased to around 5.0 after plating up to approxi-
mately 100 times. A Ni plate with an area of ϳ4 cm2 was used as the
anode. A magnetic pellet was used to stir the solution to keep the
concentration uniform and to prevent the formation of hydrogen
bubbles on the sample surface. The plating temperature was varied
in the range 20-80°C with Ϯ2°C accuracy, and the current density
was varied in the range 1-30 mA/cm2.
The thickness uniformity of electroplated metal films was inten-
sively studied during the 1990s, especially for the microelectronics
industry as Cu became the standard for interconnects in integrated
circuits and also for the MEMS society as plated metal microstruc-
tures became popular microcomponents for sensors and actuators.
The thickness variation of a Cu film on a Si wafer typically has a
‘‘rabbit-ears’’ shape distribution ͑it is also called a nodule shape͒,
i.e., it is thinner in the center of the wafer and becomes thicker
toward the edge.9,10 The ‘‘resistive seed layer’’ model has been de-
veloped to explain the thickness variation on the wafer which is due
to the voltage drop from the edge terminal contact to the center of a
wafer.9,10 For patterned wafer plating, the uniformity has an effect
not only on the thickness variation across a wafer but also on the
thickness of the individual microstructures, which develop rabbit-
ears ͑or nodule͒ cross-sectional profiles.11 The ‘‘active-area density’’
model has been developed to explain this behavior, which it ascribes
to current crowding at the edge of photoresist patterns.11 Little at-
tention has been paid to the effects of the process conditions on the
profiles of microstructures and their uniformity across a wafer. In
this paper we report on the optimization of electroplated Ni micro-
structure profiles and thickness uniformity as a function of plating
conditions. A previously unreported abnormal cap-like cross-
sectional profile for microstructures is found when they are plated at
low current density, and the thickness of a narrow structure is much
less than that of a wide one for the same plating conditions, which is
significantly different from those structures with rabbit-ears shape
profiles.
Sample preparation.—The through-mask plating technique was
used to fabricate the Ni microstructures. The process flow to fabri-
cate the photoresist micromolds was as follows: A Si substrate was
used as the plating base. It was first cleaned using a solvent in an
ultrasonic bath and was then rinsed in deionized water. A thin seed
layer was deposited on a Si-wafer substrate by sputtering ͑CCR
sputtering system͒. The seed layer consisted of a combination of
ϳ5 nm Cr and ϳ50 nm Cu. The thin Cr layer is to improve the
adhesion between the Cu seed layer and the Si substrate. The seed
layer was then coated with multiple layers of photoresist ͑AZ5214͒
to obtain a thick photoresist micromold structure. Each coating was
baked at 100°C for 10 min. This was followed by exposure using an
EV620 aligner. Developing and rinsing steps followed. The final
photoresist micromolds were soft baked at 110°C for ϳ10 min to
improve the adhesion between the photoresist and the seed layer.
Micromold structures of bars with various widths from 2 to 200 ␮m
were used. The typical photoresist thickness of micromold was
ϳ4 ␮m. After electroplating, the photoresist was removed by im-
mersing in acetone. The seed layer outside the plated Ni microstruc-
tures remained on the substrates, as the acid used to remove the seed
layer also attacks the Ni microstructures and this would modify the
profiles of the microstructures. More details of the process can be
found in Ref. 12.
Measurements.—The microstructure profiles and the thickness
were measured using a DekTak 8 profiler, which has a resolution of
1 nm in thickness. A Digital Instruments atomic force microscope
DI3100 was used to measure the roughness of the Ni microstruc-
tures.
z E-mail: jkl22@cam.ac.uk
Downloaded on 2015-06-30 to IP 130.209.6.61 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 C36-C41 ͑2005͒
C37
Figure 2. The nonuniformity of microstructures as a function of the plating
current density at plating temperatures of 20 and 60°C. The profile of a
microstructure has a cap-like shape ͑negative͒ and gradually shifts to a
rabbit-ears shape ͑positive͒ as J is increased. The nonuniformity is more
pronounced when plated at high temperature than those plated at low tem-
perature.
the thickness of a wide structure in the center. It is apparent that the
nature of the uniformity varies with current density, and that the
width of the features affects the uniformity. The term a ‘‘wide mi-
crostructure’’ used hereafter means that a structure has a typical
width of у100 ␮m; similarly, a ‘‘narrow microstructure’’ means a
typical width of р10 ␮m. Those with a width between them typi-
cally have transient structures, and we do not discuss them here in
detail. At a plating current density J ϭ 30 mA/cm2, the profile of a
wide microstructure shows a typical rabbit-ears shape. The thickness
of the plated microstructure at the middle is thinner than that at the
edge. The typical variation in thickness across a wide structure is
around 10-20%. As the width of a given microstructure becomes
smaller, the rabbit-ears profile gradually disappears, and the micro-
structure becomes thicker. However, when the specimen was plated
at a current density of 8 mA/cm2, the wide microstructure becomes
flatter with a uniform thickness from the edge to the center of the
microstructure. The thickness of a narrow microstructure is close to
that of a wide microstructure. Further reducing the plating current
density to 4 mA/cm2 leads to a change in the cross-sectional profile
and the thickness ratio. The profile of a wide microstructure be-
comes cap-like such that the middle of a wide microstructure is
thickest and becomes thinner toward the edge. Meanwhile, the
thickness of a narrow microstructure becomes much less than that of
the wide ones with a value of typically ϳ80% of the wide micro-
structures. It is therefore clear that the uniformity of microstructure
profiles and the thickness can be altered by varying the plating cur-
rent density.
Figure 1. Normalized microstructure profiles plated at J ϭ (a) 30, ͑b͒ 8,
and ͑c͒ 4 mA/cm2 at a fixed temperature of 60°C. The wide microstructure
has a rabbit-ears shape cross-sectional profile when it is plated at a high J,
and the film of a narrow structure is thicker than that of a wide one. The
profile and the thickness variation are reversed when it is plated at low J.
The wide structure has a cap-like profile, and the thickness of a narrow
structure is thinner than that of a wide one.
In order to quantify the variation of the nonuniformity ␩ upon
the process conditions, we define the nonuniformity of a microstruc-
ture as follows
hi Ϫ hmid
␩ ϭ
100%
͓1͔
*
Results
hmid
Variation of microstructure profiles
Here hmid is the thickness of a wide structure in the center, and hi is
the thickness at the edge of a wide microstructure or the thickness of
a narrow microstructure. Figure 2 shows the nonuniformity of plated
Ni microstructures as a function of the plating current density at
T ϭ 20 and 60°C, respectively. A negative nonuniformity means
that the thickness of a narrow microstructure or the edge of a wide
microstructure is less than the thickness of a wide microstructure in
the middle, while a positive nonuniformity means the opposite. At
low plating current density, the nonuniformity is negative where the
Current density effect.—It is well known that electroplated micro-
structures tend to have rabbit-ears cross-sectional profiles. For the
Ni microstructures fabricated here, similar rabbit-ears profiles were
observed when they were plated at
a
current density of
2
Ͼ
J
12 mA/cm . However, the profiles of microstructures can be
controlled by varying the current density. Figure 1a-c shows the
profile variation of microstructures as a function of current density.
The thickness h of a microstructure has been normalized to hmid
,
Downloaded on 2015-06-30 to IP 130.209.6.61 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C38
Journal of The Electrochemical Society, 152 ͑1͒ C36-C41 ͑2005͒
Figure 4. Asymmetric profile of a microstructure plated using an aged so-
lution with a pH value Ͼ5. The film on the side facing the on-coming flow is
thicker than that away from the on-coming flow.
of plated microstructures have also been investigated. It was found
that the effects from these process parameters are not as significant
as the effects of plating current density and temperature when the
pH value is lower than 5. For a newly made solution, there was no
visible difference in the microstructure profiles for plating with or
without magnetic pellet stirring, when other parameters were fixed.
The profile and the thickness variation were determined principally
by the current density used, but when it was plated in an aged
solution, which had a pH value of Ͼ5, the profile of the microstruc-
ture was typically asymmetrical as shown in Fig. 4. The film on the
side facing the on-coming flow was typically thicker than that away
from the on-coming flow, and the average plating rate was lower
than those when a solution with a low pH value was used. Stirring
the solution increased the diffusion of Ni ions near the cathode,
leading to a higher plating rate on the on-coming side. In an aging
solution, Ni-ion concentration is lower as they are consumed gradu-
ally, though there is supplier from anode, while hydrogen inclusion
in the cathode electrode gradually consumes acidic ions, leading to
an increase in pH value ͑approaching the neutral point pH ϳ7).
These factors may be responsible for the slow down in plating rate.
The uniformity of the microstructures as a function of film thick-
ness has also been investigated in the range 0.5-5 ␮m. The unifor-
mity does not show a clear dependence on film thickness.
Figure 3. The profiles of individual microstructures plated at J ϭ 28 and
2 mA/cm2 and 20°C. Comparing with those plated at 60°C with similar
current density shown in Fig. 1, it is clear that low-temperature plating pro-
duces microstructures with better uniformity.
film is thinnest for a narrow microstructure, and it becomes positive
as the current density is increased. A near-zero nonuniformity re-
gime ͑a uniform thickness͒ for those plated at 60°C exists at a cur-
rent density of ϳ8 Ϯ 2 mA/cm2. The nonuniformity over the mea-
sured current density range is Ϯ20% of the thickness.
Temperature effect.—It was found that the uniformity and profiles of
microstructures are strongly affected by plating temperature. Figure
3a and b shows the profiles of microstructures plated at J ϭ 28 and
2 mA/cm2 at 20°C, respectively. In comparing these profiles with
those plated at 60°C and similar current densities as shown in Fig.
1a and c ͑30 and 4 mA/cm2), it is clear that low-temperature plating
leads to a more uniform profile. The thickness of a narrow structure
plated at 20°C is only ϳ10% less than that of a wide one when
plated at J ϭ 2 mA/cm2; this difference is much smaller than those
structures plated at 60°C. Similar results were also obtained for
high-current-density plating. Figure 2 also summarized the nonuni-
formity of microstructures plated at 20°C at different current densi-
ties. Although the nonuniformity follows the same trend with the
current density at different temperatures, i.e., it changes from a
negative to a positive nonuniformity, non-uniformity becomes stron-
ger when plated at a higher temperature. The thickness uniformity
and the profile of a microstructure plated at 20°C is generally better
than those plated at 60°C over the whole range of current densities
used here. A higher temperature corresponds to a higher diffusion
rate of Ni ions in the solution. This may enhance the current crowd-
ing effect in patterned microstructure plating, thus enhancing the
nonuniformity of plated microstructures as was observed.
Thickness distribution within a specimen.—The distribution of
the thickness across a plated wafer had been studied by many
groups.8-10 The thickness of plated microstructures on a wafer has
been seen to have a distribution characterized by a rabbit-ears shape.
It is thinner in the middle of a wafer, and becomes thicker toward
the edge. We have conducted similar experiments on specimens with
much reduced areas of 2 ϫ 2 cm2. For a specimen plated at high
current density, a similar rabbit-ears thickness distribution was ob-
served. However, when it was plated at low current density (J
ϭ 2-4 mA/cm2), the thickness variation across a specimen was
reversed, and it was thicker in the middle of a specimen and became
thinner toward the edge. Figure 5 shows an example of the thickness
distribution within a specimen plated at 4 mA/cm2. It is obvious that
the thickness distribution is cap-like, significantly different from the
rabbit-ears thickness profile. Similarly, an optimal condition should
exist for a uniform thickness distribution within a wafer. The result
also indicates that the cross-sectional profile of an individual micro-
structure is strongly correlated to the shape of the thickness distri-
bution within a wafer. When a thickness distribution has a rabbit-
ears shape within a wafer, the individual microstructure also has a
rabbit-ears profile. Similarly a cap-like shape in thickness distribu-
tion corresponds to a cap-like profile for individual microstructures.
Other effects.—The effects created by other process parameters such
as stirring speed, pH value, and aging of solution on the uniformity
Downloaded on 2015-06-30 to IP 130.209.6.61 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 C36-C41 ͑2005͒
C39
Figure 7. The roughness of Ni films as a function of the plating temperature
at a fixed current density of 2 mA/cm2. The roughness decreases with in-
creasing temperature up to 60°C; then it increases at a temperature of 75°C.
A high plating rate is observed due to electrolysis.
Figure 5. The thickness distribution within a specimen with an area of
2 ϫ 2 cm2 plated at J ϭ 4 mA/cm2 and T ϭ 60°C. The thickness distribu-
tion has a cap-like shape, significantly different from those with rabbit-ears
shape reported by others.
plated Ni films. Figure 6 shows the topology of Ni films plated
under various conditions. The surface of a Ni film plated at low T is
very rough with a root-mean-square ͑rms͒ of 20 nm, but it becomes
smoother with rms of 11 nm as the temperature is increased to 60°C.
Further increasing the temperature to 75°C leads to a dramatic in-
crease of the roughness to an rms value of 25 nm. Figure 7 shows
the roughness variation as a function of temperature at a fixed cur-
The correlation between the profile of microstructures and the thick-
ness distribution in a specimen therefore may originate from the
same source.
Surface morphology and roughness.—Atomic force microscopy
͑AFM͒ has been used to investigate the surface morphology of
Figure 6. AFM topologies of Ni films plated at various current densities and temperatures: ͑a͒ 2 mA/cm2 and 20°C, ͑b͒ 2 mA/cm2 and 60°C, ͑c͒ 15 mA/cm2
and 60°C, and ͑d͒ 31 mA/cm2 and 60°C. All pictures have the same scale, x ϭ 1 ␮m/div, and z ϭ 200 nm/div.
Downloaded on 2015-06-30 to IP 130.209.6.61 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C40
Journal of The Electrochemical Society, 152 ͑1͒ C36-C41 ͑2005͒
Figure 8. Roughness as a function of the plating current density at a fixed
temperature of 60°C. Roughness increases with current density first, and then
decreases with further increasing current density due to the growth of fine-
grained structure.
Figure 10. The fluidic friction and electrophoresis effects model to explain
the slow plating rate ͑left͒ at the edge of a wide microstructure and ͑right͒ in
a narrow structure. This boundary layer leads to insufficient supply of Ni
ions, thus a slow plating rate, and thinner film at the edge and in a narrow
structure.
rent density of J ϭ 2 mA/cm2. The increase in the roughness at
75°C is due to the high plating rate induced by the electrolysis
process at this temperature.13 Figure 8 shows the dependence of the
roughness on the plating current density at a fixed temperature of
60°C. The surface roughness of Ni films increases with increasing
plating current, up to J Ϸ 15 mA/cm2; further increase of the cur-
rent density causes it to decrease. The initial increase in the surface
roughness with current density is due to the increased plating rate in
this small current range, where grain sizes grow bigger as current
density increases. Further increasing the current density leads to a
rapid increase in the plating rate, so there is insufficient time for
grains to grow larger. Thus the film becomes a fine-grained structure
with a smooth surface.14 Similar effect was observed for those
plated at different temperatures, with the peak current shift to the
low side of current density when plated at a lower temperature.
Similar surface roughness effects have also been reported by other
researchers.15
plating is proportional to the current density, a thicker film is depos-
ited on the edge of a wide structure due to the increased current
density, whereas the middle of the structure is thinner. As the mi-
crostructure becomes narrower, the current crowding becomes more
severe across the whole width of the patterned microstructure, and
the effective current density becomes much higher than that in the
wide structure. This leads to a thicker film for a narrow microstruc-
ture as shown on the right side of Fig. 9. As to the thickness varia-
tion within a small specimen, the voltage drop due to resistive seed
layer from the edge to the center is negligible. The current crowding
at the edge of the small specimen is believed to be the major cause
of the thickness increases at the edge when it is plated at high
current density.
The active area density is a measure of the physical layout of the
photoresist patterns, so it is not associated with the potential strength
applied. Current crowding should always exist for patterned wafer
plating. Clearly this model is unable to explain the cap-like micro-
structure profiles and the thickness distribution within a small speci-
men, so there must be other mechanisms that are responsible for the
slower plating on the edge of microstructures.
Discussion
The active-area density model for two-dimensional specimens
has been developed to explain the profile of patterned microstructure
plating.11 The mechanism of this model is illustrated in Fig. 9. Re-
gions that are more densely populated with photoresist tend to at-
tract a high current density, and hence tend to grow a thicker film.
This model can be used to explain the observation of rabbit-ears
profiles of microstructures and the thickness distribution within a
specimen when a high plating current density is used. For a wide
microstructure, significant current crowding occurs at the edge but
not in the center, as shown on the left side of Fig. 9. As the electro-
A combination of fluidic friction and electrophoresis effects16,17
is believed to be responsible for this abnormal profile and thickness
distribution. Figure 10 schematically shows the effects of fluidic
friction and electrophoresis on the profile of microstructures. The
fluidic friction between the solution and the solid surface produces a
thin boundary layer, which does not move freely. Ions in the bound-
ary layer have a very small mobility. The mobility of ions near the
solid surface is almost zero and gradually increases as they move
away from the interface. As a consequence, there is insufficient sup-
ply of ions for plating in the boundary-layer regions. However, ions
outside the boundary layer can move toward the surface of the cath-
ode under an electrical field ͑electrophoresis effect͒, providing a
continuous source of ions for plating. Consequently, for a wide mi-
cromold the existence of the boundary layer only affects the plating
on the edge of a wide microstructure; however, for a narrow micro-
mold the boundary layer thickness is comparable to the dimension
of the narrow micromold. Ions are unable to be transported to the
cathode surface and the plating rate in this narrow structure is there-
fore affected significantly, leading to a much-reduced thickness.
The thickness of the boundary layer can be varied by applying an
external potential and varying the ion concentration.16,17 At a low
current density ͑a low electrical field͒, the friction effect is dominant
and a lightly crowded current density is not sufficient to compete
with the friction effect, and so the plating rate on the edge is still
Figure 9. The ‘‘active area density’’ model to explain the thickness nonuni-
formity of a microstructure. The current crowding occurs at the edge of a
wide microstructure and in the whole area of a narrow structure, leading to
higher plating rates and thicker films in these areas.
Downloaded on 2015-06-30 to IP 130.209.6.61 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 C36-C41 ͑2005͒
C41
small. As the current density increases ͑the field increases͒, the
thickness of the boundary layer reduces, and the electrophoresis
effect becomes stronger even for those ions in the boundary layer. A
highly crowded current density leads to much faster plating at the
edge; hence, a thicker film is formed on the edge. These two effects
are superimposed and eventually form a rabbit-ears profile on the
microstructure. Close examination reveals that a structure with
rabbit-ears profile does not have a 90° sidewall, but it has a slope
that may correspond to the fluidic friction effect. A detailed theoret-
ical study to develop a model is in progress to explain this abnormal
behavior.
So far, only the rabbit-ears profile structure and thickness distri-
bution have been noticed and studied. The cap profiles of individual
microstructures have actually been shown in many publications but
without comment. For example, the cap-like profiles and cap-like
pillars in electroplated microstructures were shown in Ref. 1 and 18,
significantly different from those rabbit-ears profiles reported by
others. The reason for not noticing this abnormal behavior is that
most of the electroplating was conducted at a relatively high current
density, where the rabbit-ears profile dominates the microstructure
profile.
1. The uniformity and profiles of electroplated microstructures
are controlled mainly by plating current density and temperature.
Optimal conditions of 8 mA/cm2 and 60°C were obtained to fabri-
cate flat cross-sectional profile microstructures and a uniform thick-
ness distribution across a specimen.
2. Plating at a higher current density produces microstructures
with rabbit-ears shape profiles, and a narrow microstructure is much
thicker than a wide structure when plated at the same conditions.
3. Lower current density plating creates microstructures with
cap-like shape profiles, and a narrow structure is much thinner than
that of a wide microstructure when plated at the same conditions.
4. Plating temperature has a significant effect on the profiles and
the uniformity of microstructures, whereas other process parameters
have little effect on them.
5. The active area density model can be used to explain the
rabbit-ears profile, while a combination of fluidic friction and
electrophoresis is believed to be responsible for the abnormal cap-
like profile of microstructures and the thinning effect for narrow
structures.
The combination of these two models can also explain the effects
of other process parameters qualitatively. As long as the ion concen-
tration is sufficient to provide the plating reaction, the effect on the
profile of individual microstructures and the thickness distribution is
insignificant, as the profile is determined by the thickness of the
boundary layers and the degree of the current crowding. Other
causes that may be responsible for the abnormal profile include the
potential perturbation by the photoresist sidewall and the impurities
absorbed.
The thickness variation across a wafer and for microstructures
with different widths is undesirable for device applications, as it
would lead to a variation in device performance and characteristics.
For Cu plating used in complementary metal oxide semiconductors,
the problem is less critical as the Cu film on a wafer is planarized by
the chemical mechanical polishing process. However, the thickness
variation across a wafer and for microstructures with different
widths will cause problems for MEMS applications, as electroplat-
ing often is employed to construct active components in MEMS
devices without planarization. A variation in thickness for narrow
components of Ϯ20% can lead to a large variation in MEMS per-
formance. This is particularly important when the active structures
are used as mechanical components where the spring constant has a
strong dependence on the thickness of the structures.
Acknowledgment
This project was sponsored by the Cambridge-MIT Institute un-
der grant no. 059/P.
References
1. T. A. Green, M. J. Liew, and S. Roy, J. Electrochem. Soc., 150, 104 ͑2003͒.
2. J. W. Judy, R. S. Muller, and H. H. Zappe, J. Microelectromech. Syst., 4, 162
͑1995͒.
3. J. C. Sadak and F. K. Sautter, J. Vac. Sci. Technol., 11, 771 ͑1974͒.
4. R. V. Shenoy and M. Datta, J. Electrochem. Soc., 143, 544 ͑1996͒.
5. S. Kawahito, Y. Sasaki, M. Ishida, and T. Nakamura, Dig. Tech. Papers, Transduc-
ers, 93, 888 ͑1993͒.
6. W. H. Teh, J. K. Luo, M. A. Graham, A. Pavlov, and C. G. Smith, J. Micromech.
Microeng., 13, 591 ͑2003͒.
7. T. Matsunaga, K. Kondoh, M. Kumagae, H. Kawata, M. Yasuda, K. Murata, and
M. Yoshitake, Jpn. J. Appl. Phys., Part 1, 39, 7115 ͑2000͒.
8. S. E. Hadian and D. R. Gaba, Surf. Coat. Technol., 122, 118 ͑1999͒.
9. K. M. Takahashi, J. Electrochem. Soc., 147, 1414 ͑2000͒.
10. J. Flake, Y. Solomentsev, J. Cooper, and K. Cooper, J. Electrochem. Soc., 150, 195
͑2003͒.
11. S. Mehdizadeh, J. O. Dukovic, P. C. Andricacos, and L. T. Romankiw, J. Electro-
chem. Soc., 139, 78 ͑1992͒.
12. J. K. Luo, A. J. Flewitt, S. M. Spearing, N. A. Fleck, and W. I. Milne, Proc. SPIE,
5344, 201 ͑2004͒.
13. S. A. Watson, Ni Development Institute Technical Series Reports 10047-10055.
14. W. H. Sufranek, J. Vac. Sci. Technol., 11, 765 ͑1974͒.
15. N. S. Qu, D. Zhu, K. C. Chan, and W. N. Lei, Surf. Coat. Technol., 168, 123
͑2003͒.
16. M. Madou, Fundamentals of Microfabrication, p. 581, CRC Press, Boca Raton FL
͑1997͒.
17. J. W. Jorgenson, J. Chromatogr., 255, 335 ͑1983͒.
18. S. Zappe, M. Baltzer, Th. Kraus, and E. Obermeier, J. Micromech. Microeng., 7,
204 ͑1997͒.
Conclusions
The uniformity of electroplated microstructures has been inves-
tigated as a function of plating conditions. The following conclu-
sions can be drawn:
Downloaded on 2015-06-30 to IP 130.209.6.61 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).