Effects of nickel and palladium activations on the adhesion and I-V characteristics of as-plated electroless nickel deposits on polished crystalline silicon

Article abstract of DOI:10.1149/1.1773582

We present a study of the adhesion and current-voltage characteristics of as-plated electroless nickel deposits on polished crystalline silicon activated using nickel and palladium. A highlight of this study is the derivation of practically significant trends by collating the results of a large number of simple experiments as a function of substrate doping polarity, doping level, plating area, and plating duration. The study reveals that palladium activation is most effective on P⁺ substrates while nickel activation is most effective on N⁺ substrates, due to the requirement of substrate holes in the former activation and electrons in the latter. An activation process always improves adhesion, but, in some cases, degrades the electrical properties of the plating-silicon interface, because it introduces an intermediate suicide layer between nickel and silicon. Electroless nickel layer adheres better to nickel activated silicon, than to palladium activated silicon. However, the rectifying nature of the electroless nickel contacts on palladium activated silicon is superior to those on nickel activated silicon. Further, palladium suicide forms at 200°C, which is much lower than the temperature of 400°C required for nickel suicide formation.

Full text of DOI:10.1149/1.1773582

C554
Journal of The Electrochemical Society, 151 ͑9͒ C554-C558 ͑2004͒
0013-4651/2004/151͑9͒/C554/5/$7.00 © The Electrochemical Society, Inc.
Effects of Nickel and Palladium Activations on the Adhesion
and I-V Characteristics of As-Plated Electroless Nickel
Deposits on Polished Crystalline Silicon
Shreepad Karmalkarz and V. Pradeep Kumar
Department of Electrical Engineering, Indian Institute of Technology, Madras 600 036, India
We present a study of the adhesion and current-voltage characteristics of as-plated electroless nickel deposits on polished crys-
talline silicon activated using nickel and palladium. A highlight of this study is the derivation of practically significant trends by
collating the results of a large number of simple experiments as a function of substrate doping polarity, doping level, plating area,
and plating duration. The study reveals that palladium activation is most effective on Pϩ substrates while nickel activation is most
effective on Nϩ substrates, due to the requirement of substrate holes in the former activation and electrons in the latter. An
activation process always improves adhesion, but, in some cases, degrades the electrical properties of the plating-silicon interface,
because it introduces an intermediate silicide layer between nickel and silicon. Electroless nickel layer adheres better to nickel
activated silicon, than to palladium activated silicon. However, the rectifying nature of the electroless nickel contacts on palladium
activated silicon is superior to those on nickel activated silicon. Further, palladium silicide forms at 200°C, which is much lower
than the temperature of 400°C required for nickel silicide formation.
© 2004 The Electrochemical Society. ͓DOI: 10.1149/1.1773582͔ All rights reserved.
Manuscript submitted July 21, 2003; revised manuscript received February 9, 2004. Available electronically August 11, 2004.
Since its proposal four decades ago,1 electroless nickel plating
protecting the opposite ͑rough͒ surface using a photoresist. The
substrates meant for I-V measurements were patterned to create
0.5 ϫ 0.5 mm windows into the oxide layer of the polished surface,
for metal contact. During this process, the oxide on the rough sur-
face of the substrate was removed.
has found increasing application in the silicon industry, for contact
metallization,2,3 via-hole filling of integrated circuits ͑ICs͒ to inter-
connect the different metallization layers,4 and fabrication of metal
microstructures in microelectromechanical systems ͑MEMS͒.5 Elec-
troless nickel plating is an autocatalytic reaction, which requires a
catalytic substrate for nucleation and growth. Amorphous, porous, or
heavily etched silicon substrates are sufficiently catalytic for this
purpose,6,7 but not polished crystalline silicon substrates, whose
catalytic nature also depends upon substrate doping. Insufficiently
catalytic surfaces either do not plate, or give poor plating adherence.
To make them catalytic, an activation step is usually introduced
prior to electroless plating.
We report a comprehensive study of the adhesion and current-
voltage characteristics of as-plated electroless nickel deposits on
polished crystalline silicon activated using nickel and palladium.
The characteristics have been measured for different doping levels,
and, in addition, adhesion has been measured as a function of area
and plating duration. This study reveals several interesting trends
not reported in literature,1-5,8 and should be useful, because polished
͑on at least one side͒ and crystalline form of silicon substrates are
the most widely used in the semiconductor industry.
Because electroless plating is known to be very sensitive to sur-
face conditions, the surface resistivity or surface doping is more
relevant than the bulk or sheet resistivity, and also, substrate needs
thorough cleaning. The substrate was cleaned as follows. It was
immersed, under mild agitation, in trichloroethylene, which was
heated to the boiling point, to remove the organic impurities. To
remove the remnants of trichloroethylene, the substrate was im-
mersed in hot acetone under agitation. It was then rinsed in deion-
ized ͑DI͒ water, and boiled in nitric acid to remove metallic impu-
rities. After another rinse in DI water, it was subjected to one of the
surface treatments, palladium activation and nickel activation, for
the purpose of improving its catalytic nature. These treatments are
discussed below.
Activation
Palladium activation was achieved using a solution of PdCl2 and
NH4F in dilute HF, prepared as follows.9 A palladium chloride stock
solution was prepared by dissolving 0.05 gm of palladium chloride
in 1.5 mL of HCl and 10.5 mL of DI water. An NH4F stock solution
is prepared by dissolving 11.2 gm of NH4F in 88 mL of DI water.
Before activation, 0.8 mL of palladium chloride stock solution and
16 mL NH4F stock solution were mixed with 7.3 mL of 10% HF
solution and 48 mL of DI water. Thorough mixing was ensured by
vigorous agitation and allowing a settling time of 15 min. 5 mL of
this activator solution was used for activating each substrate. The
substrate was dipped in 10% HF for 5 s to remove any native oxide,
and immersed in the activator solution for 2 min under agitation, at
room temperature. Then, it was thoroughly cleaned in DI water,
dried, and annealed in nitrogen ambient at 200°C for 30 min to
convert the deposited palladium into palladium silicide. On removal
from the furnace, the substrate was swabbed gently with wet cotton
to remove any nonadherent palladium deposit, and thoroughly
rinsed in DI water. The entire process consisting of dilute HF dip,
immersion in palladium activator solution, annealing, and swabbing,
and DI water rinse was repeated, to improve adhesion.
Substrate Preparation
Polished substrates with a mirror finish and ͗111͘ orientation
were employed, which were cut into areas of 0.5 ϫ 1 cm, 1 ϫ 1
cm, and 1 ϫ 2 cm. The lightly doped substrates had a resistivity of
0.5-1 ⍀ cm, which amounts to a doping level of 4-8 ϫ 1015/cm3
for phosphorus doped ͑N-type͒ substrates and 1-2 ϫ 1016/cm3 for
boron-doped ͑P-type͒ substrates. Heavily doped surfaces were ob-
tained by diffusing boron and phosphorus into lightly doped sub-
strates of opposite polarity. The sheet resistivity and junction depth
were measured to be 17.4 ⍀/ᮀ and 1.8 ␮m for boron, and 3.9 ⍀/ᮀ
and 5.15 ␮m for phosphorus. These amount to Pϩ and Nϩ surfaces
with doping of ϳ1020/cm3 estimated from Irvin curves which show
surface doping as a function of the sheet ␳ϫ junction depth product.
Care was taken to remove the glass formed during boron diffusion.
An oxide layer of thickness Ͼ0.3 ␮m was grown on all substrates
͑on heavily doped substrates this was done during drive-in of the
diffusion͒.
On the substrates meant for adhesion measurements, the oxide
layer on the polished surface of the substrate was etched away, while
The nickel activation process was similar to the palladium
activation described above. After a 5 s 10% HF dip, the substrate
was immersed for 2 min in an alkaline electroless nickel bath
having the composition: nickel chloride, NiCl2•6H2O ͑30 g/L͒,
sodium hypophosphite, Na2H2PO2 ͑10 g/L͒, sodium citrate,
z E-mail: skarmalkar@hotmail.com
Downloaded on 2015-04-07 to IP 141.217.58.200 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, 151 ͑9͒ C554-C558 ͑2004͒
C555
Figure 1. Measurement setups for ͑a͒ adhesion and ͑b͒ current-voltage char-
acteristics.
Na3C6H5O7•2H2O ͑84 g/L͒, and ammonium chloride, NH4Cl ͑50
g/L͒. A bath pH of 8-10 and temperature of 90-95°C was maintained
during immersion. For each substrate, 20 mL of the solution was
used. Note that, a larger volume of solution than in the palladium
activation described above is used, due to elevated temperature of
the process, during which the pH change may be significant due to
the loss of ammonia, if smaller volume of the solution is used. On
removal from the solution, the substrate was annealed in nitrogen
ambient for 30 min to convert the deposited nickel into nickel sili-
cide; an annealing temperature of 400°C was employed, because
200°C was found to be inadequate for improving adhesion. The
annealed substrate was boiled in nitric acid for 5 min to remove any
nonadherent remnants and rinsed thoroughly in DI water.
Electroless Nickel Plating
The activated substrate was immersed in 20 mL of the alkaline
electroless nickel bath described above, for 2 min. The plated sub-
strate was rinsed thoroughly in DI water and dried with air gun. This
process gave a plating thickness of ϳ0.2 ␮m measured as follows.
The plating was done over a window opened into a wet oxidized
wafer having a known ͑measured͒ oxide thickness of 0.3 ␮m, and
the difference between the thickness of the plating and the oxide was
obtained using a surface profiler. The plating thickness is a function
of bath composition, temperature, pH, and plating duration, which
were controlled by taking a fresh solution for every experiment. The
thickness depends on surface conditions also. The conclusions of
our study are not affected by this fact.
Figure 2. Adhesion as a function of area for different surface doping. ͑a͒ No
activation, ͑b͒ palladium activation, and ͑c͒ nickel activation. The numbers
indicate the substrate area in cm2.
Adhesion and Current-Voltage Measurement Procedure
Adhesion strength was measured by a pull test ͑see Fig. 1a͒. The
substrate was sandwiched between two metal blocks using ex-
tremely strong, rapid cure cyanoacrylate adhesive ͑ANABOND 202,
Anabond Pvt. Ltd., Chennai, India͒. Prior to this, the metal blocks
were cleaned in acetone and isopropyl alcohol, and heated in an
oven along with the substrates at 120°C for 15 min to remove ad-
sorbed moisture in the metal block. The blocks were pulled apart @
1 mm/min using a universal testing machine ͑UTM, model-
INSTRON, no. 4301͒ until they separated. The force required for
separation was measured.
The plated substrates used in current-voltage measurements were
heated in nitrogen ambient at 150°C for 5 min, prior to measure-
ment, to remove any moisture, which is a cause of instability in
characteristics. The measurement set up is shown in Fig. 1b, along
with a cross-section of the substrate and an equivalent circuit giving
the elements-RB, RC, and RS, which contribute to the measured
characteristics of the electroless nickel—polished silicon contact.
The back contact resistance RB is much less than the contact resis-
tance, RC, of the nickel-silicon contact, on account of the surface
roughness and much larger area associated with the former. For the
substrate doping and geometry described above, the substrate resis-
tance RS was estimated to be several times less than the measured
resistance.10 So, the measured characteristics reflect the properties of
the plated metal-silicon contact.
Measured Results
First, we present the measured adhesion trends in Fig. 2 and 3.
The end points of a vertical segment in each of these figures indi-
cates the two adhesion values obtained during two separate but iden-
tical experiments; the length of the segment represents the disper-
sion arising out of factors beyond experimental control. The doping
level of substrates in Fig. 3 corresponds to those used in MEMS
applications, where thick plating may be necessary. The figures
show that activated substrates provide higher adhesion than unacti-
vated ones. The other trends are discussed below.
For palladium activation, the improvement in adhesion strength
is maximum in the case of heavily doped P-type substrate, and de-
creases progressively as the substrate becomes less P-type or more
N-type. Interestingly, on P-type substrates, whether lightly or
heavily doped, adhesion strength decreases with substrate area;
Downloaded on 2015-04-07 to IP 141.217.58.200 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C556
Journal of The Electrochemical Society, 151 ͑9͒ C554-C558 ͑2004͒
Figure 4. Current-voltage characteristics as a function of the surface doping.
Pd-palladium activation, Ni-nickel activation, and None-no activation.
and ohmic for heavily doped substrates. The forward voltage for 1
mA current extracted from this data is shown in Fig. 5. The forward
voltage for nickel activation is higher than that for palladium acti-
vation, for lightly doped N or P-type and Pϩ substrates, even though
nickel activation provides better adhesion than palladium activation
for these substrates. Similarly, on Nϩ substrates, the forward voltage
without any activation is lower than that with either nickel or palla-
dium activation, which provide as good or better adhesion. These
facts show that the means of improving mechanical properties of the
nickel-silicon contact system do not always improve electrical prop-
Figure 3. Adhesion as a function of plating duration for lightly doped N-
and P-type substrates ͑a͒ No activation, ͑b͒ palladium activation, and ͑c͒
nickel activation. The numbers indicate the plating duration in minutes;
plating area ϭ 1 cm2.
however, on N-type substrates, no particular trend is observed in this
regard. Adhesion decreases as the plating duration increases.
For nickel activation, the results are shown in terms of absolute
values of adhesion, unlike in the case of palladium activation, where
the adhesion is reported per unit area. This is because, for nickel
activation, during pull-test associated with adhesion measurement,
the failure occurred in silicon rather than at the nickel-silicon inter-
face, where the failure occurred for palladium activation. Hence, the
reading shown on the adhesion axis corresponds to the fracture in
silicon rather than to the nickel-silicon adhesion, which is higher
than the reading shown. Thus, the trends in adhesion of nickel to
silicon as a function of doping, area and duration become irrelevant,
and what is significant is that nickel activation provides a much
higher adhesion than palladium activation, independent of doping,
area and duration.
Next, we present the current-voltage characteristics in Fig. 4. The
characteristics were fairly uniform over the substrate area. The char-
acteristics are seen to be Schottky type for lightly doped substrates
Figure 5. Forward voltage at 1 mA current for contacts on unactivated and
activated substrates.
Downloaded on 2015-04-07 to IP 141.217.58.200 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, 151 ͑9͒ C554-C558 ͑2004͒
C557
neous reactions of metal-substrate and metal-metal bonding taking
place during plating. Inverted pyramids are formed since the metal-
metal bond formation ͑in the vertical direction͒ has a higher rate
than the metal-substrate bond formation ͑in the lateral direction͒.
The pyramids coalesce, after an initial growth, to provide a continu-
ous surface. The various observed trends in adhesion and I-V char-
acteristics can be interpreted based on the intermediate layer of
nickel or palladium silicide, present on activated silicon substrates,
between the plating and silicon.
The higher adhesion due to nickel activation implies that silicon-
nickel silicide-nickel bond is mechanically much stronger than the
silicon-palladium silicide-nickel bond. Because nickel activation in-
volves immersion in an alkaline solution at elevated temperature, it
is necessary to investigate the contribution of roughening of the
silicon surface due to etching during this immersion, to the adhesion
improvement. Alkaline silicon etchants used in MEMS,17 e.g., KOH
and TMAH, etch silicon @ ϳ1 ␮m/min. In KOH etching,17 the
temperature used is comparable to that used in our nickel plating
bath, but the pH 13 used therein is much higher than pH 9 used in
our experiments. Thus, the effective molar concentration of the
etchant species used in our experiment is ϳ10,000 times lower than
that in KOH etching, implying an etching of a few A° only, during
the 2 min immersion associated with the activation process. Further,
we studied the effect of mild etching of silicon, using an acidic
HNO3 :HF:H2O etchant ͑200:1:75͒ which etches silicon @ ϳ125
A0/min. The plating adhesion due to a 2 min etch instead of nickel
activation, prior to electroless plating, was only ϳ20% of the value
obtained using nickel activation. Both heavily doped and lightly
doped, N-type as well as P-type, substrates of 1 cm2 area were
covered in this study. These observations establish that the improved
adhesion from nickel activation is mainly due to the strength of the
silicon-nickel silicide-nickel bond and not due to surface roughening
during activation.
Figure 6. Model of the electroless nickel-polished silicon interface.
͑a͒ Top view showing nucleation sites, ͑b͒ cross-section showing unactivated
and activated silicon. The diagram is not to scale ͑nickel thickness Ӷ silicon
thickness͒.
erties, and that activation, which always improves adhesion, does
not necessarily reduce contact resistance. The significant increase in
forward voltage on Nϩ substrates due to palladium activation points
out the disadvantage of using this activation on Nϩ substrates. On
the other hand, the low adhesion and high forward voltage, observed
for Pϩ substrates without nickel or palladium activation, establish
the necessity of activation for these substrates.
Discussion
The observed variation in the effectiveness of the activation pro-
cesses as a function of doping level reveals the polarity of charge
carriers required for the electrochemical reactions associated with
these processes. The trends in this variation are different for nickel
and palladium activations, indicating that the charges required in
these are of opposite polarity. The increasing effectiveness of palla-
dium activation in improving adhesion and lowering forward-
voltage on substrates with higher hole concentration reveals the re-
quirement of positive charges in palladium deposition. For nickel
activation, what could not be revealed in adhesion tests due to the
plating adhesion being higher than silicon fracture strength is
brought out in current-voltage measurements. The fact that the for-
ward voltage on Nϩ substrates is lower when these are unactivated
rather than activated, using palladium or nickel, points to the re-
quirement of negative charges in nickel deposition. These observa-
tions are consistent with the findings of earlier studies, namely-
electrons are required for the progress of an autocatalytic reaction
such as electroless nickel plating,11 and holes are required for pal-
ladium deposition on silicon by displacement of silicon atoms into
the activator solution consisting of fluoride ions.9 It is mentioned
that, for substrates without any activation, the trends revealed in
current-voltage testing could not be discerned from the adhesion vs.
doping data. This brings out the utility of using current-voltage
testing in combination with adhesion measurements for diagnostic
purposes.
We interpret the remaining observed trends in terms of a model
of the electroless nickel-polished silicon interface proposed in Fig.
6. This model can be constructed logically from information avail-
able in the literature.12-16 In our earlier work,12 we gave an SEM
photo of metal islands on the silicon surface at the end of immersion
in an activator solution. The presence of the metal was confirmed
using EDAX. A conceptual diagram of metal islands acting as nucle-
ation sites is shown in Fig. 6a. On heating to a temperature Ͼ200°C,
the metal islands react with silicon to forming metal-silicide. For-
mation of palladium and nickel silicide in this manner, for tempera-
ture Ͼ200°C, has been discussed in Ref. 13-15. The inverted pyra-
midal growth of nickel from nucleation islands during electroless
plating ͑see Fig. 6b and c͒, is inferred from the 2-D model of metal
plating given in Ref. 16. Such a growth results from the simulta-
The resistance associated with nickel silicide layer is higher than
that associated with palladium silicide layer, as shown by the higher
forward voltage of nickel activated substrates. Palladium activation
improves adhesion on all substrates, and in addition, lowers the
forward voltage except on Nϩ substrates, implying that nickel-
palladium silicide-silicon bond has reasonably high mechanical
strength on all substrates, and, at the same time, a low resistance
except on Nϩ substrates.
The rise in forward voltage on Nϩ substrates due to palladium
activation implies that the resistance of the palladium-silicide layer
on these substrates is high. However, the reason for the reduction in
forward voltage on Pϩ substrates due to nickel activation, in spite of
the requirement of electrons in nickel activation, is not clear. These
observations are significant enough to merit further investigation for
a convincing explanation.
The decrease in plating adhesion with increase in substrate area
observed on palladium activated P-type substrates is also observed
on unactivated P-type substrates, and is not observed on N-type
substrates. So, this observation should be related to the electroless
nickel plating and substrate doping, and not to the activation pro-
cess, and can be explained by the combination of facts, namely,
paucity of electrons in a P-type substrate, and the requirement of
electrons in electroless plating. Paucity of electrons produces a num-
ber of activation sites where the nickel is weakly bonded to
palladium-silicide. Weak bonds decide the measured adhesion. As
substrate area increases, the possibility of weaker bonds increases,
and this reduces the adhesion. Note that, while the palladium-
silcide-silicon bonds at the nucleation sites are uniformly good on
substrates where holes are in a majority, and improve adhesion, the
nickel-palladium silicide bonds at many sites tend to be weak on
these substrates where electrons are in a minority. The simultaneous
presence of these factors accounts for the observation on P-type
substrates, that palladium activation enhances adhesion of nickel
plating to silicon, without eliminating the reduction of adhesion with
increase in substrate area.
Downloaded on 2015-04-07 to IP 141.217.58.200 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

C558
Journal of The Electrochemical Society, 151 ͑9͒ C554-C558 ͑2004͒
ward voltage than even the no activation case, on all but Pϩ sub-
strates. Thus, an improvement in adhesion does not necessarily
translate into reduction in forward voltage, due to the resistance
associated with the intermediate silicide layer. The palladium sili-
cide formation in the palladium activation step requires a tempera-
ture of 200°C as against the 400°C required for nickel silicide for-
mation in nickel activation. Palladium activation is preferable over
nickel activation in applications where schottky type contact is re-
quired. For ohmic contacts on Nϩ substrates, a low contact resis-
tance is obtained without any activation. However, for ohmic con-
tacts on Pϩ substrates, nickel or palladium activation is necessary to
lower the resistance.
The decrease in adhesion with plating duration implies genera-
tion of stresses in the plated film with increase in plating thickness.
Because this has been observed to occur on palladium activated and
unactivated substrates, and not on nickel activated substrates, it can
be concluded that the structural mismatch at the interface is much
less in the case of nickel activation than in the other cases.
Finally, it is pointed out that the scope of our work is limited to
the characteristics for as-plated contact, without any post-plating
and precharacterization heat-treatment at elevated temperatures ca-
pable of altering the nickel-silicon interface. Further, the sintering
temperature used during activation did not exceed 400°C. Post-
plating heating and/or higher sintering temperatures ͑600-700°C͒
during activation are essential to lower the high values of contact
resistance obtained in our work to practically tolerable limits.3 With-
out such treatments, the resistance of electroless contacts tends to be
higher than those of sputtered contacts, possibly because an oxide
layer grows over the substrate area not covered by nucleation sites,
due to the aquous nature of the electroless plating solution.13 High-
temperature heat-treatment causes the metal and phosphorus present
in the electroless deposit to diffuse into silicon ͑through any oxide
barrier͒, increasing contact area and modifying the surface layer
composition, and thus reducing the contact resistance;3,13 it also
eliminates the effects of variation in plating conditions.3
Acknowledgment
The authors thank Faisal Arab and Ashok Bania for their help in
manuscript preparation.
References
1. M. V. Sullivan and J. H. Eigler, J. Electrochem. Soc., 104, 226 ͑1957͒.
2. K. M. Oppermann, H. Kalicki, R. Aschenbrenner, and H. Reichl, Microelectron.
Reliab., 39, 1389 ͑1999͒.
3. B. K. Singh and R. N. Mitra, J. Electrochem. Soc., 127, 2578 ͑1980͒.
4. V. M. Dubin, J. Electrochem. Soc., 144, 898 ͑1992͒.
5. S. Furukawa and M. Mehergany, Sens. Actuators, 56, 261 ͑1996͒.
6. S. Dhar and S. Chakrabarti, Semicond. Sci. Technol., 11, 1231 ͑1996͒.
7. T. Balaji Suresh, S. Satish Kumar, S. Karmalkar, and Enakshi Bhattacharya, Thin
Solid Films, 252, 78 ͑1994͒.
Conclusion
In this work, the effects of activating polished crystalline silicon
with nickel and palladium, on the adhesion and current-voltage char-
acteristics of as-plated electroless nickel are presented and com-
pared. The useful conclusions from this study are the following.
Nickel activation occurs by autocatalytic reaction, which requires
supply of electrons, and is therefore more effective in reducing for-
ward voltage on N-substrates, as compared to P-type substrates. Pal-
ladium activation, on the other hand, occurs by displacement plating
reaction, which requires supply of holes; it is most effective on Pϩ
substrates and disadvantageous on Nϩ substrates. Nickel activation
gives higher adhesion than palladium activation, but a higher for-
8. V. M. Dubin, J. Electrochem. Soc., 139, 1289 ͑1992͒.
9. S. Karmalkar and J. Banerjee, J. Electrochem. Soc., 146, 580 ͑1999͒.
10. V. Pradeep Kumar. MS Thesis, Indian Institute of Technology, Madras ͑2003͒.
11. F. A. Lowenheim, Modern Electroplating, John Wiley & Sons, New York ͑1974͒.
12. S. Karmalkar, D. Sridhar, and J. Banerjee, J. Electrochem. Soc., 144, 1696 ͑1997͒.
13. M. G. Coleman, R. A. Pryor, and T. G. Sparks, in Proceedings of the 13th IEEE
Photovoltaic Specialists Conference, 597 ͑1978͒.
14. D. Gupta, Metall. Mater. Trans. A, 8A, 1431 ͑1977͒.
15. E. G. Colgan, Thin Solid Films, 279, 193 ͑1996͒.
16. S. L. Maltow and E. L. Ralph, Solid-State Electron., 2, 202 ͑1961͒.
17. C. Moldovan, R. Iosub, D. Dascalu, and G. Nechifor, Sens. Actuators B, 58, 438
͑1999͒.
Downloaded on 2015-04-07 to IP 141.217.58.200 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).