Composition control of tin-zinc deposits using experimental strategies

Article abstract of DOI:10.1016/j.electacta.2005.10.025

The robust electroplating settings of a direct-current (dc) plating mode for the co-deposition of Sn-Zn deposits with their composition close to the eutectic point (i.e., Sn-9Zn) from the chloride solutions were achieved and investigated by using experimental strategies, including the fractional factorial design (FFD) and central composite design (CCD) coupled with the response surface methodology (RSM). The temperature of the plating bath, pH, and the metallic ion ratio (i.e., Sn⁴⁺/Zn²⁺ ratio) were found to be the key factors affecting the composition of Sn-Zn deposits in the FFD study. The effects of pH and temperature of the plating solution on the composition of Sn-Zn deposits were examined using a regression model in the CCD study. This model, represented as contour plots, showed that pH 5.0 and temperature = 78 °C were the robust electroplating settings for the co-deposition of the eutectic Sn-Zn alloys, which was independent of the substrates. In addition, based on the robust plating settings, the composition of Sn-Zn alloys could be precisely controlled and predicted by adjusting the composition of the plating baths. From the morphologies and crystalline information, the binary Sn-Zn deposits prepared in this work should belong to heterogeneous alloys.

Full text of DOI:10.1016/j.electacta.2005.10.025

Electrochimica Acta 51 (2006) 3692–3698
Composition control of tin–zinc deposits using experimental strategies
Chi-Chang Hu∗,1, Chun-Kou Wang, Gen-Lan Lee
Department of Chemical Engineering, National Chung Cheng University, Chia-Yi 621, Taiwan
Received 21 July 2005; received in revised form 17 October 2005; accepted 18 October 2005
Available online 28 November 2005
Abstract
The robust electroplating settings of a direct-current (dc) plating mode for the co-deposition of Sn–Zn deposits with their composition close to the
eutectic point (i.e., Sn–9Zn) from the chloride solutions were achieved and investigated by using experimental strategies, including the fractional
factorial design (FFD) and central composite design (CCD) coupled with the response surface methodology (RSM). The temperature of the plating
bath, pH, and the metallic ion ratio (i.e., Sn4+/Zn2+ ratio) were found to be the key factors affecting the composition of Sn–Zn deposits in the FFD
study. The effects of pH and temperature of the plating solution on the composition of Sn–Zn deposits were examined using a regression model
in the CCD study. This model, represented as contour plots, showed that pH 5.0 and temperature = 78 ◦C were the robust electroplating settings
for the co-deposition of the eutectic Sn–Zn alloys, which was independent of the substrates. In addition, based on the robust plating settings, the
composition of Sn–Zn alloys could be precisely controlled and predicted by adjusting the composition of the plating baths. From the morphologies
and crystalline information, the binary Sn–Zn deposits prepared in this work should belong to heterogeneous alloys.
© 2005 Elsevier Ltd. All rights reserved.
Keywords: Electroplating; Sn–Zn deposit; Eutectic point; Fractional factorial design; Central composite design
1. Introduction
especially the electroplating process [4–6], in order to increase
the density of interconnections. Recently, lead-free processes
for electronic devices and components are required to address
the environmental concerns and the alpha radiation of impuri-
ties of Pb [7–9]. In addition, the demand of Pb-free bumps rather
than Sn–Pb solder becomes an urgent problem in the electronic
assembly industry. Therefore, different types of Pb-free sol-
der bump materials have been investigated, which are generally
the tin-rich, eutectic alloys (e.g., Sn–Ag, Sn–Cu, Sn–Ag–Cu,
Sn–Bi, Sn–Zn, etc.)[7–10]. AmongtheseeutecticSn-richalloys,
Sn–9Zn generally shows the advantages of a low reflowing tem-
perature (an eutectic point of 198 ◦C), low cost, high wetting
capability, good malleability, and a high anti-corrosive property.
Accordingly, this work focuses on the electroplating and com-
position control of the Sn–Zn deposits with their composition
close to the eutectic point.
Electroplating is recognized to be suitable for making the
fine pitch bumps with high-speed deposition and high reliability
[7,8,11,12]. In addition, this deposition technique attracts indus-
trial considerations because of the low cost, the high throughput,
and the deposition capability on almost any geometry [13].
Moreover, electroplating is a simple, one-step process for the
fabrication of Sn–Pb solders for printed circuit boards, solder
bumps, and lead-frame packages in the electronic applications.
The flip–chip bonding is one of various surface-mounting
techniques in the integrated circuit (IC) assembly industry,
which was developed to provide high-density interconnections
between the chip and the substrate [1]. The flip–chip bonding
process entails (1) disposing a plurality of solder bumps on the
upper-surface of the die that is mounted directly to the substrate,
(2) flipping the die and mating the solder bumps with the cor-
responding bonding Al pads located on the substrate, and (3)
heating the die and the substrate in order to reflow the solder
bumps. Each reflowed bump forms a bond between the die and
the substrate, which functions as an electrical and physical con-
tact.
In general, tin–lead (Sn–Pb) solder was widely used as the
bump material for the flip–chip packaging applications [2–4],
which was originally deposited by means of a vacuum evapo-
ration technique developed by IBM [1]. However, a variety of
new methods/processes were developed to form solder bumps,
∗
Corresponding author. Tel.: +886 5 2720411x33411; fax: +886 5 2721206.
E-mail address: chmhcc@ccu.edu.tw (C.-C. Hu).
ISE member.
1
0013-4686/$ – see front matter © 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.electacta.2005.10.025

C.-C. Hu et al. / Electrochimica Acta 51 (2006) 3692–3698
3693
Table 1
Accordingly, it is worthy being paid attention on the production
of lead-free bump interconnections by means of the electroplat-
ing technologies [7,8,10–12].
According to Brenner’s definition [14], the co-deposition of
Sn–Zn alloys from the cyanide-based baths belongs to the alloy
electroplating of the irregular type since the effects of the plating
temperature and current density on the composition of Sn–Zn
are irregular. In this work, however, cyanide is not employed and
thus, the reaction mechanism is generally proposed as following:
Composition and variables for the electroplating of a nickel film from a watts
nickel bath
Compounds/variables
Concentration/conditions
NiSO4·6H2O (mol/dm3)
NiCl2·6H2O (mol/dm3)
H3BO3 (mol/dm3)
Temperature (◦C)
pHa
1
0.2
0.5
55
3.8
60
Current density (mA/cm2)
SnO32− + 3H2O + 4e− → Sn + 6OH−
ZnO22− + 2H2O + 2e− → Zn + 4OH−
(1)
(2)
a
pH was adjusted by 1 M HCl or NaOH.
an ultrasonic bath for 5 min, acid-cleaned with 5 M H2SO4 for
1 min. Finally, the Cu substrates were rinsed with pure water and
vibratedinanultrasonicbathfor5 min. ForthesubstrateswithNi
or Fe thin films (denoted as Ni/Cu and Fe/Cu, respectively), the
pre-cleaned Cu substrate was vertically placed in a 500-ml jacket
cell with a watts nickel bath (see Table 1) or a solution containing
0.5 M FeSO4·7H2O + 0.1 M H3BO4 with pH 2.0. The Cu sub-
stratewassurroundedwithananodeofplatinum-coatedstainless
steel mesh. After the electroplating of Ni or Fe at 60 mA/cm2 for
5 min under 55 ◦C, the substrates were rinsed with pure water.
These substrates were vertically placed in another 500-ml jacket
cell surrounded with an anode of platinum-coated stainless steel
mesh and electroplated with Sn–Zn deposits at 80 mA/cm2 for
20 min. After the Sn–Zn deposition, these electrodes were rinsed
with pure water and vibrated in an ultrasonic bath for 5 min.
The plating bath mainly consisted of SnCl4·nH2O, ZnCl2,
and Ca(C6H11O7)2·H2O (0.1 M, complex agent) with pH being
adjusted with 1 M HCl and 1 M NaOH. All solutions were pre-
paredwithpurewaterproducedbyareagentwatersystem(Milli-
Q SP, Japan) at 18 Mꢀ cm and all reagents were Merck, GR.
Solution temperature was maintained at the specified tempera-
tures with an accuracy of 0.1 ◦C by means of a water thermostat
(Haake DC3 and K20).
Based on the fact that Sn(II) is easily oxidized to Sn(IV) under air
agitation in high-temperature plating solutions, a chloride-based
Sn(IV) bath is employed in this work.
The experimental strategy is a sequential procedure to reach
the optimal operation conditions of interest [15–20]. Based on
this strategy, the response variable(s) (e.g., the Sn content in the
Sn–Zn alloys) is a function of the quantitative control variables.
Since the starting point of experimental conditions is usually
remote from the optimum meanwhile there may be several key
factors influencing the response(s) of interest, the fractional fac-
torial design (FFD) is usually employed to efficiently find these
key variables [15–17]. From the FFD, the response variable(s)
against the key factors near the starting point in the system is
simply fitted as a first-order model. After the FFD experiment,
the methodology of the steepest ascent path is often used to
approach the vicinity of the optimal conditions if the starting
point of experimental conditions is really far from the optimum.
Finally, the central composite design (CCD) is used to model
the apparent curvature of response variable(s) against the key
factors at the vicinity of optimum. Since mathematic models
of the response variable(s) against the control variables can be
represented as contour lines or a plane in graphs, this method is
called a response surface methodology (RSM).
The purpose of this work is to identify the suitable electroplat-
ing conditions to produce Sn–Zn deposits with the composition
close to the eutectic point (i.e., Sn–9Zn). The key variables
affecting the composition of Sn (and Zn) in the Sn–Zn deposits
were screened out by the FFD [15–17]. These variables were
subjected to the CCD coupled with the RSM [15,16] to exam-
ine the relationship between the composition of Sn–Zn alloys
and the plating variables around the optimal plating conditions
of Sn–9Zn deposits. Finally, the proposed plating settings were
applied to deposit Sn–9Zn alloys onto various substrates (i.e.,
Fe, Ni, and Cu) to demonstrate the robust property.
The average composition of all deposits was measured using
an energy-dispersive X-ray (EDX) spectroscope with standards
at five points coupled with a scanning electron microscope
(SEM, JEOL JSM35). The mean error of this EDX analysis
is ca. 1.5 atomic percents (at%).
3. Results and discussion
3.1. Fractional factorial design
To efficiency find the key variables affecting the Sn (and Zn)
content in the Sn–Zn deposits, the fractional factorial design
is introduced to screen out these key variables. This experiment
design can observe the influences of each preparation variable at
avarietyofothervariablelevelsaswellastheinteractionsamong
these variables on the composition of Sn (and Zn). Based on the
literature review, the effects of the following electroplating vari-
ables were investigated in the FFD study: (A) pH of the plating
solution, (B) total concentration of metal ions (M), (C) temper-
ature of the plating solution (◦C), (D) [Sn4+]/[Zn2+] molar ratio,
and (E) agitation rate (rpm). Note that in our preliminary study,
the effect of the current density on the composition of Sn–Zn
2. Experimental details
Tin–zinc deposits were electroplated onto commercially pure
(99.5%) 1 cm × 2 cm copper plates or the Cu plates deposited
with a thin film of either Fe or Ni. The copper plates were
first cleaned with trichloroethylene, rinsed with pure water, and
then, anodized at 30 mA/cm2 in a 0.1 M NaOH solution for
5 min. After anodizing, the plates were cathodically polarized at
75 mA/cm2 in another 0.1 M NaOH solution for 10 s, vibrated in

3694
C.-C. Hu et al. / Electrochimica Acta 51 (2006) 3692–3698
Table 2
Table 4
Factors and levels for 25−1 fractional factorial design
Analysis of variance for the average Sn content in the Sn–Zn deposits from the
25−1 fractional factorial design
Factor
Level
Source
SS
d.f.
MS
F*
−
+
A
C
D
E
AD
CE
Error
175.6
8920.8
1660.6
82.8
75.7
78.3
1
1
1
1
1
1
9
175.6
19.7
A. pH
6.0
0.1
20
80/20
10
7.0
0.2
70
90/10
20
8920.8
1660.6
82.8
75.7
78.3
1001.5
186.4
9.3
B. Total concentration (M)
C. Temperature (◦C)
D. [Sn4+]/[Zn2+] (atomic ratio)
E. Agitation rate (rpm)
8.5
8.8
80.2
8.9
Total
11073.9
15
deposits is insignificant. In addition, the effect of the current
density on the composition of Sn–Zn deposits was found to be
obviously minor, even in the cyanide-based baths [14]. Accord-
ingly, the current density of electroplating is set at 80 mA/cm2
to achieve high-speed deposition. The fixed levels of these five
variables are listed in Table 2 meanwhile the 25−1 design matrix
with the experimental data were given in Table 3.
Remark: F0.01(1, 9) = 10.56; F0.05(1, 9) = 5.12; R2 = 1 − (SSE/SST) = 0.993.
tion), and D ([Sn4+]/[Zn2+] ratio). Thus, the effects of factor E
and ABCD interaction are said to be confounded [15,16]. From
the principle of the sparsity of effects [16], a system is likely
to be driven primarily by the effects of certain main factors and
low-orderinteractions. Thus, effectsofthehigh-order(e.g., three
and higher order) interactions are assumed to be negligible and
hence, the effect of factor E can be isolated from the confounded
effects by this FFD experiment (resolution = V) [16].
In Table 3, the low and high levels of A, B, C, D, and E were
denoted as “−” and “+”, respectively. The contrast coefficients
of factor E (i.e., column 6 in Table 3) is generated from the
contrast coefficients of factors A, B, C, and D through Eq. (3).
From an examination of the results in Table 3, the composition
of Sn–Zn deposits is ranged from 16.5 to 92.9 at%, indicating
that certain factors and/or interactions should show significant
effect on the Sn content in the Sn–Zn deposits. Accordingly,
analysis of variance (ANOVA) was carried out on the basis of
the data shown in Table 3 and the result of the statistical analysis
is summarized in Table 4. The analysis of variance (ANOVA) is
derived from partitioning the total variability (SST) into its com-
ponent parts (i.e., sums of square for model and error, SSmodel
and SSE, respectively), which can be calculated on the basis of
the following equations [16]:
A 2k−1 fractional factorial design matrix can be constructed
by writing down a basic design matrix consisting of a full 2k−1
factorial design and then adding the kth factor by identifying
its plus and minus levels with the plus and minus signs of the
highest order interaction ABC . . . (k−1). For example, the 25−1
fractional factorial design with the defining relation, I = ABCDE
(introduced by Box et al. [17]), shows the following property:
E = E · I = E · ABCDE = ABCDE2 = ABCD
(3)
Therefore, the 25−1 fractional factorial design is obtained by
writing down the full 24 factorial as the basic design and then
equating factor E to the ABCD interaction [16]. According to
Eq. (3), the orthogonal contrast coefficients of factor E are equal
to that of the ABCD interaction. In addition, the combination
of observations used to estimate the effect of factor E (agitation
rate) is identical to that used to estimate the four-factor interac-
tion effect of the aliases A (pH of the plating solution), B (total
concentration of metal ions), C (temperature of the plating solu-
Table 3
k−1
SST = 2 (yi − y¯)2
(4)
The design matrix and experimental data of the average Sn content in the Sn–Zn
depositsfor25−1 fractionalfactorialdesignwiththedefiningrelationI = ABCDE
ꢀ
i=1
Run
Factors
A
Sn content (wt%)
(Ci)2
2k−1
B
C
D
E
SSi =
(5)
(6)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
−
+
−
−
+
−
−
−
−
+
+
+
+
−
−
−
−
+
+
+
+
−
−
−
−
−
−
−
−
+
+
+
+
+
+
23.4
17.2
16.5
28.2
68.8
70.1
71.4
73.3
30.3
53.4
43.9
49.0
85.7
92.8
84.3
92.9
−
−
+
SSE = SST − SSmodel
−
+
+
where yi and y¯ are indicative of the ith response and the grand
average of all the observations, respectively. Note that Ci is
indicative of the contrast of factor (or interaction) i, which is the
sum of multiplying the observations (i.e., yi) with the appropriate
contrast coefficients (i.e., the plus–minus signs in the appropri-
ate column of the design matrix). In addition, SSi indicates the
sum of square corresponding to factor (or interaction) i. The sum
−
+
−
−
+
−
+
−
+
+
+
−
−
+
−
+
−
−
+
−
+
+
+
−
−
+
−
−
+
+
of SSi with the statistical significance is defined as the SSmodel
.
SSi
d.f.i
SSE
error
+
+
+
−
−
+
The quantities MSi =
and MSE =
are defined as the
d.f.
−
mean squares of factor (or interaction) i and the mean square of
+
+
error, respectively. The d.f.i and d.f.error indicate the degree of

C.-C. Hu et al. / Electrochimica Acta 51 (2006) 3692–3698
3695
freedom for factor (or interaction) i and error, respectively. In
Table4, theteststatistics, F*, definedasMSi/MSE, areemployed
to test the statistical significance of each factor and the two-
factor interactions. If the calculated value of F* is greater than
that in the F table at a specific probability level (e.g., α = 0.01),
a statistically significant factor or interaction is obtained. After
the test, only factors A, C, and D exhibit statistically significant
effects on the composition of Sn–Zn deposits. Actually, factor
E as well as interactions AD and CE are considered as the fac-
tor/interactions with marginal significance (i.e., their effects are
not as important as the effects of factors A, C, and D) since
their calculated F values are somewhat lower than the critical
value (F0.01(1, 9) = 10.56). The multiple correlation coefficient,
R2 = 1 − (SSE/SST), is the proportion of sum of squares of total
variances (SST) explained by the fitted equation. A R2 value
close to 1 means a good fit to the experiment data. In this work,
R2 = 0.993, an excellent fitting, when factors A, C, D, and E as
well as interactions AD and CE are considered to be significant
and the other effects (including factor B and other two-factor
interactions) can be pulled into the error.
Calculation of the estimates for factors and two-factor inter-
actions is followed the procedure recommended by Box et al.
[17], which is equal to Ci/2k−2. The estimates of all factors A–E
and interactions with marginal significance are shown in Fig. 1.
From Table 4 and Fig. 1, it is very clear that the sequence of
factors/interactions with respect to decreasing the influence is:
C > D > A > E > CE > AD although the last three factors show
the marginal statistical significance from Table 4. Based on the
results of Table 4 and Fig. 1a, an increase in pH will enhance
the tin content in the Sn–Zn deposits. This effect, however, is
conflict to the results obtained from the cyanide-based solu-
tions [14,21,22], presumably due to the complicated interactions
among metal ions, OH−, and cyanide. Since factor B shows no
significant effect, the variation in the total concentration of metal
ions has no influence on the composition of Sn–Zn deposits.
This result may be indicative of that the co-deposition of Sn–Zn
alloys from the non-cyanide baths used in this work belongs to
the alloy co-deposition of a regular type [14]. This proposal is
also supported by the insignificant effect of the current density of
electroplating on the composition of Sn–Zn deposits since the
regular co-deposition process is defined as the electroplating
under the control of diffusion phenomena [14]. Factor C shows
the largest effect, indicating that the increase in the plating tem-
perature from 20 to 70 ◦C largely promotes the Sn content from
ca. 33 to 80 at%. In general, effects of the electroplating temper-
ature on the Sn–Zn deposition include an increase in the current
efficiencies of both Sn and Zn and an increase in the Sn content
[22,23]. Furthermore, an additional merit of the non-cyanide
baths is the absence of a rapid decomposition of cyanide at tem-
peratures ≥65 ◦C [14]. The [Sn4+]/[Zn2+] ratio, i.e., factor D,
shows a positive effect on the composition of the Sn deposits.
This also supports that the co-deposition of Sn–Zn alloys from
the non-cyanide baths belongs to the alloy co-deposition of a
regular type. Since factor E shows a marginal effect, varying the
agitation rate should change the thickness of the diffusion layer,
which renders a minor variation in the relative current efficiency
for the deposition of Sn and Zn.
Fig. 1. Effects of (a) factors (A) pH, (B) total concentration of metal ions (M),
(C) temperature of the plating bath (◦C), (D) [Sn4+]/[Zn2+] ratio, and (E) agi-
tation rate (rpm) and (b) interactions AD and CE on the average Sn content of
Sn–Zn deposits, where (+) and (−) indicate the high and low levels of factors,
respectively.
The interaction effects with marginal significance (i.e., AD
and CE) are shown in Fig. 1b. In this figure, for the plating
solution with a high [Sn4+]/[Zn2+] ratio (i.e., 90/10), increasing
pH of the plating solution leads an obvious increase in the Sn
content in the Sn–Zn deposits. However, when factor D is under
the low level, pH has not effect on the Sn content. The above
phenomenarevealtheexistenceofaninteractionbetweenfactors
A and D. Another CE interaction effect indicates that the effect
of C is more obvious when factor E is under the low level in
comparison with that as factor E is under the high level. From
all the above results and discussion, the composition of Sn–Zn
deposits is not only dominated by the plating temperature but
also by pH and the [Sn4+]/[Zn2+] ratio in the plating solution.
From the analysis of variance and regression analysis of the
results shown in Table 3, a fitted polynomial model (including
the marginal effects) can be generated. This model, quantita-
tively elucidating the effects of all plating variables with statis-
tical significance, is expressed in the following:
y = 56.4 + 3.3xA + 23.7xC + 10.2xD + 2.3xE
+ 2.2xAxD − 2.2xCxE
(7)

3696
C.-C. Hu et al. / Electrochimica Acta 51 (2006) 3692–3698
Table 5
where xi are the coded variables for factor i (i.e., A, C, and D).
The coded variables, xi, are defined in the standardized form as
following [15]:
Design matrix and experimental data for the average Sn content in the Sn–Zn
deposits for the central composite design with a quadratic form fit
Run
Factor
Sn content (wt%)
(Xi,HIGH − Xi,MEAN
)
A (pH)
C (T, ◦C)
xi,HIGH
=
(= +1)
(8)
(9)
Si
1
2
3
4
5
6
7
8
9
+1 (7.0)
+1 (78.0)
0 (70.0)
92.5
92.6
95.3
90.2
90.3
85.8
82.9
88.8
86.0
86.5
86.9
√
(Xi,LOW − Xi,MEAN
)
+
2 (7.4)
xi,LOW
=
(= −1)
+1 (7.0)
0 (6.0)
−1 (62.0)
Si
√
−
2 (58.8)
(Xi,HIGH + Xi,LOW
)
−1 (5.0)
−1 (62.0)
√
Xi,MEAN
=
(10)
(11)
−
2 (4.6)
0 (70.0)
2
−1 (5.0)
0 (6.0)
0 (6.0)
0 (6.0)
0 (6.0)
+1 (78.0)
√
(Xi,HIGH − Xi,LOW
)
+
2 (81.2)
Si =
0 (70.0)
0 (70.0)
0 (70.0)
2
10
11
where Xi,HIGH and Xi,LOW are the high and low levels of factor i
in the natural unit, respectively. Note that when xA, xC, and xD
are equal to +1 (i.e., factors A, C, and D are in the high level), the
predicted value of y (i.e., the Sn content) is equal to 93.6 that is
higher than the eutectic point of Sn–Zn alloys (i.e., 91 at% Sn).
Accordingly, there is no need to carry out the steepest ascent
study from the results and discussion of the FFD study. More-
over, from an examination of Table 3, the experimental settings
of runs 13/15 or runs 14/16 should be very close to the opti-
mal conditions for the electroplating of eutectic Sn–9Zn deposit.
Accordingly, the experimental settings of run 15 are considered
as the center point in the central composite design study coupled
with theresponse surfacemethod [15,16]. It is worthy noting that
the Sn contents of runs 13–16 shown in Table 3 are above 84 at%
that is very close to the composition of metal ions in the plating
baths. Therefore, the response variable (i.e., the Sn content in
the Sn–Zn deposits) is considered to be a smooth function of the
plating variables, which is consistent with the prediction of Eq.
(7).
The design matrix with the corresponding results in the CCD
study is shown in Table 5. The step size of pH and the plat-
ing temperature is 1 and 8 ◦C, respectively. The experiments on
the original (central) point are repeated three times in order to
evaluate the pure error between each experiment. The regression
analysis and ANOVA for the data shown in Table 5 were carried
out andtheresultantsecond-ordermodelrepresentingthedepen-
dence of the Sn content on factors A (pH) and C (temperature
of the plating solution) was generated as follow:
y = 86.5 + 3.0xA − 1.5xC + 1.6xA2 + 1.7xC2 + 1.2xAxC (12)
where y, xA and xC are indicative of the Sn content, pH, and
temperature of the plating bath, respectively. This fitting has an
R2 value of 0.90, indicating that Eq. (12) suitably describes the
dependence of the Sn content on factors A and C.
The contour plots for the dependence of the Sn content in the
Sn–Zn deposits on factors A and C were constructed by using the
regression model (i.e., Eq. (12)) and a typical contour diagram
is shown in Fig. 2. Note that a minimum (ca. 84 at%) of the
Sn content in the deposits is located around the conditions at
pH 5.0 and 78 ◦C. In addition, the Sn content increases slowly
when pH and temperature of the plating solutions are increased
and decreased simultaneously. Note that a plateau with very
weak dependence of the Sn content on the pH and temperature
of the plating solutions is clearly found around the minimum.
This indicates that the Sn content around the minimum is very
stable, which is good for production (i.e., the robust conditions).
Unfortunately, the minimal Sn content is not close to the eutectic
point (i.e., Sn–9Zn) and thus, an additional confirmation test has
to be performed.
3.2. Central composite design
The purpose of the central composite design is to provide
enough tests to fit the second-order model correlating the key
electroplating variables and the composition of Sn–Zn deposits
close to the eutectic point. Based on the results and discussion in
previous section, the experimental settings of run 15 in Table 3
are used as the central (original) point in the central composite
design (CCD) study. Note that factor D is not considered in this
CCD study because the deposits with composition very close to
the metal ion ratio in the plating bath were preferred to prepare
[19]. Thus, the [Sn4+]/[Zn2+] ratio of the plating bath was kept
constant ([Sn4+]/[Zn2+] = 90/10) in the CCD study.
A CCD study generally consists of a 2m factorial design with
2m axial runs and nC center runs. Hence, the total number of
experimental runs in a CCD study is equal to 2m + 2m + nC. For
instance, factors A and C are considered in this CCD study. The
total number of experimental runs in this CCD study is equal
In order to demonstrate the robust property of the electro-
plating settings for the co-deposition of the eutectic Sn–Zn
alloys, the dependence of the Sn content for various Sn–Zn
deposits plated at pH 5.0 and temperature = 78 ◦C on the
ratio of [Sn4+]/[Zn2+] in the plating solutions was investigated
(see Table 6). From this table, the Sn content in the Sn–Zn
deposits is gradually increased from 83.4 to 97.8 wt% when the
[Sn4+]/[Zn2+] ratio in the plating solutions is steadily increased
from 90/10 to 97/3. Note that the composition of a Sn–Zn deposit
plated from the solution with the [Sn4+]/[Zn2+] ratio of 93/7 is
to 22 + 2 × 2 + 3 = 11. Since the distance from the experimental
√
points to the central points is constant ( 2 in this case), this
design will have a constant variance of the response variable
at all experimental points when the design is rotated about the
origin [16].

C.-C. Hu et al. / Electrochimica Acta 51 (2006) 3692–3698
3697
Fig. 2. Contour plots for the constant Sn contents in the Sn–Zn deposits against
pH and temperature of the plating solutions.
approximately equal to the eutectic point (i.e., run 4 in Table 6).
In addition, when the substrate was changed to be Ni/Cu or Cu,
the composition of the Sn–Zn deposits prepared under the con-
ditions of run 4 in Table 6 is still very close to the eutectic point
(i.e., 90.8 and 90.9 wt% of Sn for the Ni/Cu and Cu substrates).
The above results indicate that pH of 5.0 and temperature of
78 ◦C are the robust conditions for the co-deposition of Sn–Zn
alloys with their composition close to the eutectic point, which
is weakly dependent on the temperature and pH of the plating
solutions.
3.3. Morphology and crystalline structure of Sn–Zn
deposits
The morphologies of Sn–Zn deposits with the Sn content
of ca. 83, 90, and 97 wt% are shown in Fig. 3. In general,
these binary Sn–Zn deposits show polyhedral crystallites. In
addition, the surface of these binary Sn–Zn deposits becomes
smoother and compacter with increasing the Sn content although
all deposits are micro-rough. Moreover, the Sn–Zn deposits with
Table 6
The dependence of the Sn content in the Sn–Zn deposits on the ratio of
[Sn4+]/[Zn2+] in the plating baths
Fig. 3. The morphologies of Sn–Zn deposits with the Sn content of (a) 83, (b)
90, and (c) 97 wt%.
Run
[Sn4+]/[Zn2+
]
Sn content (wt%)
1
2
3
4
5
6
7
8
90/10
91/9
92/8
93/7
94/6
95/5
96/4
97/3
83.4
86.9
89.8
91.2
93.0
95.2
96.7
97.8
90 and 97 wt% Sn mainly consist of spherical grains. Since the
morphology of these deposits shows the intimate mixing and
uniting of Sn and Zn, the binary Sn–Zn deposits should belong
to the heterogeneous alloys [14,24].
Fig. 4 shows the XRD patterns of Sn–Zn deposits with the
Sn content of ca. 83, 90, and 97 wt%. On curve 1, there are
several diffraction peaks corresponding to the crystalline faces
of ␤-Sn, Zn, and Cu on this binary deposit, indicating that the
Remark: pH 5.0; temperature, 78 ◦C; total concentration of metal ions, 0.2 M;
agitation rate, 10 rpm; current density, 80 mA/cm2. All Sn–Zn deposits were
electroplated at 80 mA/cm2 under pH of 5.0 and 78 ◦C for 20 min.

3698
C.-C. Hu et al. / Electrochimica Acta 51 (2006) 3692–3698
trolled by adjusting the composition of the plating baths. From
the SEM and XRD results, the binary Sn–Zn deposits prepared
in this work should belong to heterogeneous alloys.
Acknowledgment
The financial support of this work, by the National Science
Council of the Republic of China, is gratefully acknowledged.
References
[1] E.M. Davis, W.E. Harding, R.S. Schwartz, J.J. Corning, IBM J. Res.
Dev. (1964) 102.
[2] M. Carano, Plat. Surf. Finish. 91 (2004) 36.
[3] C. VanHorn, R. Oberle, G. Shawhan, Packaging, Production
Conference—Proceedings of the Technical Program (West and East),
vol. 1, 1997, p. 336.
[4] P. Bratin, E. Shalyt, M. Pavlov, J. Berkmans, Proceedings of
the IEEE/CPMT International Electronics Manufacturing Technology
(IEMT) Symposium, 2003, p. 395.
Fig. 4. The XRD patterns of Sn–Zn deposits with the Sn content of (1) 83, (2)
90, and (3) 97 wt%; where o, z, and ꢁ indicate the diffraction faces of Sn, Zn,
and Cu, respectively.
[5] M. Carano, Plat. Surf. Finish. 90 (2003) 44.
[6] P.A. Kohl, J. Electrochem. Soc. 129 (1982) 1196.
[7] S. Arai, H. Akatsuka, N. Kaneko, J. Electrochem. Soc. 150 (2003) C730.
[8] S. Joseph, G.J. Phatak, T. Seth, K. Gurunathan, D.P. Amalnerkar,
T.R.N. Kutty, IEEE Region 10 Annual International Conference Pro-
ceedings/TENCON, vol. 4, 2003, p. 1367.
[9] O. Khaselev, I.S. Zavarine, Vysotskaya, C. Fan, Y. Zhang, J. Abys,
Trans. Inst. Met. Finish. 80 (2002) 200.
[10] I. Yanada, D. Gudeczauskas, Proceedings of the AESF Annual Technical
Conference, 1999, p. 457.
[11] E. Jung, R. Aschenbrenner, Ch. Kallmayer, P. Coskina, H. Reichl, Pro-
ceedings of the International Symposium and Exhibition on Advanced
Packaging Materials Processes Properties and Interfaces, 2001,
p. 119.
[12] R. Kiumi, J. Yoshioka, F. Kuriyama, N. Saito, M. Shimoyama,
Proceedings—Electronic Components and Technology Conference,
2002, p. 711.
[13] J. Horkans, J. Electrochem. Soc. 128 (1981) 45.
[14] A. Brenner, Electrodeposition of Alloys, vols. 1 and 2, Academic Press,
New York/London, 1963.
Sn–Zn deposit with 83 wt% Sn is composed of polycrystalline
Sn and Zn metals. In addition, the diffraction peaks correspond-
ing to the crystalline faces of Zn are still visible for the Sn–Zn
deposit with 3 wt% (see curve 3 in Fig. 4). This indicates that the
concentration of Zn atoms dispersed in this deposit is still high
enough to exhibit constructive diffraction although the intensity
of the Zn diffraction peaks is very low for all deposits.
Based on all the above textural results, the Sn–Zn deposits
prepared in this work should belong to heterogeneous alloys
sinceSnandZnaremutuallydissolvedeachother[14,24]. More-
over, the eutectic Sn–9Zn deposit can be easily electroplated
under our reliable plating conditions although some studies
reported the presence of eutectic-type Sn–Zn alloys with ca.
20 or 10 wt% Zn [25,26].
4. Conclusions
[15] J.A. Cornall, How to Apply Response Surface Methodology, vol. 8,
ASQC, Wisconsin, 1990.
[16] D.C. Montgomery, Design and Analysis of Experiments, fifth ed., J.
Wiley & Sons, Singapore, 1997.
[17] G.E.P. Box, W.G. Hunter, J.S. Hunter, Statistics for Experiments, J.
Wiley & Sons, New York, 1978, p. 374.
[18] C.-C. Hu, A. Bai, Surf. Coat. Technol. 137 (2001) 181.
[19] P. Tsay, C.-C. Hu, J. Electrochem. Soc. 149 (2002) C492.
[20] C.-C. Hu, T.-W. Tsou, J. Power Sources 115 (2003) 179.
[21] J.W. Cuthbertson, R.M. Angles, J. Electrochem. Soc. 94 (1948) 73.
[22] A.E. Davies, R.M. Angles, Trans. Inst. Met. Finish. 33 (1956) 277.
[23] K. Aotani, J. Electrochem. Soc. Jpn. 21 (1953) 21.
[24] K. Wang, H.W. Pickering, K.G. Weil, Electrochim. Acta 46 (2001) 3835.
[25] E. Guaus, J. Torrent-Burgues, J. Electroanal. Chem. 549 (2003) 25.
[26] K.-Y. Kim, Y. Yang, Surf. Coat. Technol. 64 (1994) 99.
Using the sequential experiment strategies (i.e., the fractional
factorial design and the central composite design coupled with
response surface methodology), the robust deposition settings
for the electroplating of Sn–Zn deposits with the composition
close to the eutectic point (i.e., Sn–9Zn) were clearly demon-
strated. The experimental settings, pH of 5.0, the total concentra-
tion of 0.2 M, the plating temperature of 78 ◦C, the [Sn4+]/[Zn2+]
ratio of 93/7, and the agitation rate of 10 rpm, for the plating of
Sn–9Zn deposits were very robust to the Fe/Cu, Ni/Cu, and Cu
substrates. Based on the robust plating settings proposed in this
work, the composition of Sn–Zn alloys can be precisely con-