Aluminum deposition and nucleation on nitrogen-incorporated tetrahedral amorphous carbon electrodes in ambient temperature chloroaluminate melts

Article abstract of DOI:10.1149/1.1393908

The electrodeposition of aluminum on the atomically smooth nitrogen-incorporated tetrahedral amorphous carbon (taC:N) electrode in ambient temperature AlCl₃/EMIC chloroaluminate melts has been interpreted using a prior model of three-dimensional diffusion controlled nucleation and growth. Aluminum requires an unusually high overpotential for nucleation on taC:N because of the low density of intrinsic active sites, which act as critical nuclei during the initial stage of deposition. The current-time characteristics of nucleation on taC:N show a strong dependency on overpotential. Generation of additional, overpotential-induced active sites imposes a partial progressive nature on the overall nucleation process, resulting in a slight deviation from the limiting behavior of an ideal instantaneous nucleation model.

Full text of DOI:10.1149/1.1393908

3
370
Journal of The Electrochemical Society, 147 (9) 3370-3376 (2000)
S0013-4651(99)11-076-0 CCC: $7.00 © The Electrochemical Society, Inc.
Aluminum Deposition and Nucleation on Nitrogen-Incorporated
Tetrahedral Amorphous Carbon Electrodes in Ambient
Temperature Chloroaluminate Melts
a,
a,
b
c
d
Jae-Joon Lee, * Barry Miller, ** Xu Shi, Rafi Kalish, and Kraig A. Wheeler
a
b
c
Department of Chemistry, Case Western Reserve University, Cleveland, Ohio 44106-7078, USA
School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore 639798
Solid State Institute and Department of Physics, Technion Institute of Technology, Haifa 32000, Israel
d
Department of Chemistry, Delaware State University, Dover, Delaware 19901, USA
The electrodeposition of aluminum on the atomically smooth nitrogen-incorporated tetrahedral amorphous carbon (taC:N) elec-
trode in ambient temperature AlCl /EMIC chloroaluminate melts has been interpreted using a prior model of three-dimensional
3
diffusion controlled nucleation and growth. Aluminum requires an unusually high overpotential for nucleation on taC:N because
of the low density of intrinsic active sites, which act as critical nuclei during the initial stage of deposition. The current-time char-
acteristics of nucleation on taC:N show a strong dependency on overpotential. Generation of additional, overpotential-induced
active sites imposes a partial progressive nature on the overall nucleation process, resulting in a slight deviation from the limiting
behavior of an ideal instantaneous nucleation model.
©
2000 The Electrochemical Society. S0013-4651(99)11-076-0. All rights reserved.
Manuscript submitted November 19, 1999; revised manuscript received May 31, 2000.
4Al Cl ϩ 3eϪ } Al ϩ 7AlClϪ
Ϫ
[2]
Nitrogen-incorporated tetrahedral amorphous carbon (taC:N) de-
2
7
4
3
posited at ambient temperature forms a predominantly sp bonded
semiconductor with a bandgap of approximately 2.0-2.5 eV in uni-
form films of near atomic smoothness and glass-like structure which
The AlCl /EMIC ambient temperature melts have a number of
attractive features, including very low vapor pressure, high intrinsic
electric conductivity, thermal stability, and good solubility for many
Because of these features they have been
employed as electrolytes for electroplating of metals and recharge-
Numerous studies relevant to these applications
have been reported concerning the fundamentals of electrodeposi-
tion of metals at various substrates in these ionic melts.
3
1
has not yet been fully characterized. This material has recently been
5
,10
transition metal ions.
examined electrochemically in aqueous media and shows an extra-
ordinary combination of the stability associated with boron-doped
diamond (BDD) and enhanced electroanalytical properties and sur-
1
1
1
1-13
able batteries.
2
face activity. Notably, electrodes of this material have an extremely
1
4-23
Most
wide potential window between hydrogen and oxygen evolution in
aqueous media, stability to anodic processes such as chlorine evolu-
tion, and an absence of underpotential or alloying interactions with
metals. The resistivity (ϳ10 ⍀ cm) of these 50 nm thick films is also
not limiting. Here, we extend the study of these films to aprotic,
ambient temperature chloroaluminate molten salts and compare their
Al deposition and nucleation features to those of W and glassy car-
bon (GC) electrodes.
previous electrodeposition studies employed common metallic and
nonmetallic substrates, such as W and GC, where the initial stage of
the deposition process is controlled by hemispherical diffusion lim-
1
4-23
ited nucleation kinetics.
There are no studies yet reported using taC:N as a substrate for
nonaqueous electrochemical metal deposition. To understand the
electrochemical metal deposition properties of this apparently ex-
tremely inert electrode with near atomic smoothness, we focus here
on nucleation at the initial stage of Al electrodeposition in this apro-
tic, ambient temperature chloroaluminate melt as electrolyte.
3
-5
Ambient temperature molten salts are mixtures of aluminum
chloride and an organic chloride donor (RCl), such as 1-ethyl-3-
methylimidazolium chloride (EMIC), having an adjustable Lewis
acidity and ionic composition dependent on AlCl /RCl molar
3
Experimental
6
,7
ratios. In brief, melts with the ratio greater than 1 are Lewis acidic;
less than 1, Lewis basic; and neutral if the ratio is unity. The anion
composition is governed by the following equilibrium which greatly
All electrochemical studies were performed in a nitrogen-filled
glove box system (VAC), equipped with moisture and oxygen ana-
lyzers. Experiments were at ambient temperature except for some at
0 or 75 (Ϯ3ЊC) which was maintained in the cell with a hot plate.
The glove box system and the methods used to prepare the
AlCl /EMIC melts were identical to those described earlier. All
melts used in these studies for Al deposition and nucleation experi-
ments had an acidic AlCl :EMIC molar ratio of 1.1:1 with calculated
Al Cl ] of 0.445 M, unless stated otherwise. The neutral melt was
prepared stoichiometrically followed by minor adjustment until a
clean, neutral electrochemical window was obtained with W and Pt.
The electrodeposition and nucleation of Al were investigated at
three different electrodes. Tungsten and GC rods were sealed in
Ϫ 5-7
favors the formation of AlCl4
6
Ϫ
2
AlCl4Ϫ } Al Cl ϩ ClϪ
[1]
2
7
24
Ϫ
3
In acidic melts, the dominant anionic species are AlCl and
4
Ϫ
Ϫ
Al Cl and the potential limits are fixed by AlCl oxidation and
Al Cl reduction. In basic melts, AlCl4 and Cl dominate and,
2
7
4
Ϫ
7
5-9
Ϫ
Ϫ
3
2
Ϫ
[
2 7
correspondingly, the limits are set by chloride oxidation and reduc-
ϩ
tion of the organic cation (R ). Neutral melts, in which only major
Ϫ
anionic species is AlCl , display the widest electrochemical win-
4
Ϫ
Ϫ
7
dow, up to 4.5 V, because they essentially have no free Cl or Al Cl
2
to shorten the anodic and cathodic limits, respectively. AlCl /EMIC
2
3
epoxy, exposing circular surface areas ca. 0.08 and 0.07 cm , respec-
binary molten salts are advantageous in terms of their electrochemi-
cal windows of stability and the width of the composition range over
tively, for working electrodes. Prior to the experiments, these elec-
trodes were polished with alumina compounds down to 0.05 ␮m. The
preparation of ϳ50 nm thick films with ϳ12% N content of taC:N on
5
,8-10
which AlCl /EMIC is liquid at ambient temperature.
Aluminum
3
Ϫ
deposition in acidic melts proceeds by reduction of Al Cl , the only
ϩϩ
1,2
2
7
p
Si and their characteristics were described earlier. An epoxy
reducible, aluminum-containing species within the electrochemical
2
sealed square or rectangular area of taC:N (ca. 0.05 cm ) was used
directly for cyclic voltammetry (CV) and chronoamperometry. The
reference electrode for CV was an Al wire (Alfa/AESAR, 99.999%)
6
-10
window of this kind of ambient temperature molten salt
*
*
* Electrochemical Society Student Member.
* Electrochemical Society Fellow.
immersed in an acidic composition of AlCl /EMIC melt (1.5:1) iso-
3
Downloaded on 2015-03-07 to IP 130.15.241.167 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, 147 (9) 3370-3376 (2000)
3371
S0013-4651(99)11-076-0 CCC: $7.00 © The Electrochemical Society, Inc.
lated from the main cell compartment by porous Vycor glass. In case
of potential-step experiments for nucleation study, aluminum wires
dipped directly into the melts were used as reference electrode to
monitor overpotentials. A coiled Al wire dipped in the melt was used
as the counter electrode.
Potential control was achieved with a Pine RDE3 potentiostat
coupled with a PARC 175 universal programmer. Data were ac-
quired using a digital recorder (HP7090A) interfaced with an IBM
compatible computer through an IEEE-488 bus for all experiments
other than potential-step experiments for nucleation study. For the
latter, a Pine AFCBP1 bipotentiostat with Pinechem 2.5 software
was used for potential control and data.
Results and Discussion
A series of CVs for taC:N electrodes was acquired in neutral,
basic, and acidic compositions of AlCl /EMIC room temperature
3
melts in parallel with those for W and GC electrodes. Generally,
taC:N exhibits a wider electrochemical window in different compo-
sitions of melt than those of W and GC. Figure1 clearly shows that
both anodic and cathodic limits of the neutral melt are nearly identi-
cal at W and GC, but are more extended with taC:N.
The wider electrochemical window is even more evident in a
basic (0.9:1) melt, Fig. 2, where the cathodic limit is the reduction of
organic cation and the anodic limit of the window is chlorine evolu-
tion. In particular, Fig. 2b shows greater overpotentials for both chlo-
rine evolution and its corresponding reduction at taC:N than those at
W and GC. This behavior of taC:N is less irreversible than that of
BDD electrodes, paralleling their recently described contrast in
Figure 2. (a) CVs for W, GC, and taC:N in basic melt (0.9:1) (b) Expanded
chlorine evolution and reduction in the same melt. v ϭ 100 mV/s.
The expected Al underpotential deposition (UPD) on W amount-
6
,15,24,31
ing to 1/3-1/2 of a monolayer was seen.
Like GC,taC:N
shows no indication of UPD, which is evident at W in the gradual
increase in cathodic current before bulk deposition and a corre-
sponding underpotential stripping after the bulk peak upon switch-
ing the scan direction. However, the stripping of Al at taC:N is
unusual, being very slow with poor efficiency and a variable broad
anodic region. This wide stripping potential region narrows after
repeated cycling, but the peak remains poorly defined. The overall
coulombic efficiency is obtained from Fig. 3b, showing the total
charge attributable to Al dissolution (Q ) vs. that of deposition (Q )
2
aqueous chloride. BDD showed more than 1.0 V greater anodic
overpotential for Cl evolution than the processes at Pt and graphite
2
5
in saturated CaCl (>11 M chloride). Noise and residual or capac-
2
itive current at these nearly atomically smooth taC:N deposits are
comparatively very low in all these figures on a geometric area basis.
Figure 3a shows Al deposition and stripping from acidic
AlCl /EMIC chloroaluminate melts on W, GC, and taC:N elec-
3
a
c
trodes. A nucleation overpotential is clearly shown for all three elec-
trodes, largest at taC:N. These voltammograms exhibit the current
loops that are typical of deposition processes requiring nucleation
at taC:N. Both the initial normal stripping peak and the further broad
anodic current up to 2.0 V were totaled to yield Q . The slope of the
a
6
,27-30
plot, 0.97, suggests complete anodic recovery, though most of the
dissolution process is delayed and needs more positive potential.
This behavior is similar in many respects to film stripping observed
overpotential.
Generally, these current loops occur because the
deposition of metal onto a different substrate during cathodic scan
requires a considerable overpotential in order to initiate the nucle-
ation and subsequent growth of a metal deposit. When the scan
direction is reversed, the reduction current continues to flow because
the deposition of metal now takes place on a less hostile surface of
2
5,26
for other metals at BDD,
but what deposit characteristics lead to
this peculiar anodic behavior remain to be clarified.
The nucleation process was further investigated with potential-
step experiments in unstirred acidic chloroaluminate melts from
6
,15
deposited metal.
Figure 1. Series of cyclic voltammograms (CVs) for W, GC, and taC:N in
Figure 3. (a) Al deposition and stripping on W, GC, and taC:N in acidic melt
(1.5:1). v ϭ 100mV/s. (b) The overall coulombic efficiency for Al deposition
and stripping on taC:N.
neutral AlCl /EMIC melt. For all figures, the reference electrode is an Al
3
wire immersed in an isolated acidic melt (1.5:1). v ϭ 100 mV/s.
Downloaded on 2015-03-07 to IP 130.15.241.167 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

3
1
372
Journal of The Electrochemical Society, 147 (9) 3370-3376 (2000)
S0013-4651(99)11-076-0 CCC: $7.00 © The Electrochemical Society, Inc.
.0 V, where no Al deposition takes place, to a value sufficiently neg-
with the observation of highest nucleation overpotential at taC:N in
Fig. 3.
Several models have been developed for the nucleation and
growth phenomena taking place during bulk metal deposition.
The models involving the two limiting cases of three-dimensional
nucleation processes, instantaneous and progressive, in conjunction
with the hemispherical diffusion-controlled growth of nuclei have
been found to be appropriate to analyze metal deposition in most
cases, including Al deposition on various substrates in ambient tem-
ative to initiate the nucleation and growth process after a short time
delay. Examples of current-time profiles resulting from such experi-
ments are shown in Fig. 4 for W and taC:N, where the overpotentials
for each transient are expressed against the Al wire dipped in the
melt. The data are obtained at 60ЊC for W and 75ЊC for GC and
taC:N electrodes. The shape of these chronoamperograms is typical
of a diffusion-limited reaction for the nucleation and growth of a
metal deposit on a foreign substrate.27-30
First, they exhibit a sharp current spike due to charging of the
electrode double layer.27 This capacitive transient is followed by a
rise in the current due to an increase in effective electrode area,
either as each independent nucleus forms and grows in size and/or
2
7-30
2
7-30,32
perature chloroaluminate molten salts.
ation refers to fast nucleation on a small number of active sites, with
all nucleation sites activated instantly at t ϭ 0.
Instantaneous nucle-
1
4,16,20,21,30
Progres-
sive nucleation is slow nucleation on virtually an infinite number of
active sites which are gradually activated as the time of the applied
potential step increases.
2
7-30
the number of nuclei increase.
At this stage of growth, the nuclei
1
6,19,22,30
develop hemispherical diffusion zones initially and then, as these
zones overlap, hemispherical mass transfer gives way to a semi-infi-
nite linear diffusion effectively to a planar electrode. The current
However, nucleation processes are often intermediate between
these two extremes. In such cases, values of the rate of nucleation
reaches a maximum (i ) as the diffusion zones of the Al nuclei begin
per active site (A) and the number density of active nucleation sites
m
Ϫ1/2
16,18,21,32
to overlap, then decreases to show the usual i proportional to t
(N ) can be obtained from the experimental data.
Graphical
o
2
7
behavior (Cottrell condition). The time to the current maximum
tm), depends on the overpotential (␩), and decreases as the applied
analysis of the ascending portion of each current-time transient, be-
fore the overlap of diffusion zones of individual nuclei, might be the
(
2
7-30
potential is made more negative.
simplest way of determining the type of nucleation.
taneous nucleation model is applicable if the current growth increas-
The instan-
From the transients shown in Fig. 4, it is readily seen that taC:N
needs a greater overpotential to obtain a similar current-time profile
because nucleation at taC:N is more difficult than at W and GC. The
potential range for a similar current-time profile at GC, not shown
here, lies between those of W and taC:N. The results are consistent
1
/2
es linearly with t , whereas a progressive model implies the current
growth increases linearly with t . Figure 5a shows the result of ana-
3
/2
lyzing the rising portions of the current-transients at taC:N, obtained
Figure 5. (a) Current (i) vs. t1/2 plots for chronoamperometric data obtained
at a taC:N electrode in acidic melt (1.1:1) at 75ЊC (same data in Fig. 4). (b)
Comparison of the dimensionless experimental data derived from the current-
time transients shown in Fig. 4 with the theoretical models for instantaneous
and progressive nucleation.
Figure 4. A series of current-time transients for Al deposition on taC:N from
Ϫ
1
1
7
.1:1 AlCl /EMIC ([Al Cl ] ϭ 0.445 M). The potential was stepped from
.0 V to the values of interest. (a) W electrode at 60ЊC. (b) taC:N electrode at
5ЊC.
3
2
7
Downloaded on 2015-03-07 to IP 130.15.241.167 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, 147 (9) 3370-3376 (2000)
3373
[5]
S0013-4651(99)11-076-0 CCC: $7.00 © The Electrochemical Society, Inc.
from Fig. 4b. It can be seen from the good linearity in i vs. t1/2 for
i tЈ ϭ 0.1629(nFAC) D
2
2
m m
various overpotentials that instantaneous three-dimensional nucle-
where n is taken as 0.75 and C is 0.445 M for 1.1:1 AlCl :EMIC
3
ation of Al dominates on taC:N electrodes. Analysis of the data for
W and GC, not shown here, also shows good linearity of the t1/2 plot.
melts. Equation 5 refers to the number of electrons transferred per
3
ϩ
each Al deposited and C is the concentration of Al in melt that
deposits via Eq. 2. However, the applied overpotential should be
However, this graphical method for discriminating between dif-
ferent nucleation mechanisms is not completely free of ambiguity
because only the initial segments of the ascending portion of the data
are used. A superior method for differentiating the two nucleation
models is to compare the complete dimensionless experimental cur-
Ϫ
7
negative enough to reduce the surface concentration of Al Cl vir-
2
tually to zero in order to use this relation for the diffusion coeffi-
2
7
2
cient. Then, the product i tЈ is nearly invariant with the overpo-
m m
tential applied, at a given bulk concentration of electroactive species,
and the growth process of nuclei is diffusion limited.
rent-time transients to the dimensionless theoretical transients for
each nucleation mechanism.27-30 The theoretical transients for limit-
2
The experimental values of i tЈ and diffusion coefficient ob-
m m
ing cases of instantaneous (Eq. 3) and progressive (Eq. 4) nucleation
are represented, respectively, by
tained from each transient are also summarized in Table I. The dif-
Ϫ
fusion coefficients of Al Cl measured from nucleation data, 6.6 ϫ
2
7
Ϫ7
Ϫ6
Ϫ6
2
2
10 (W, 60ЊC), 1.4 ϫ 10 (GC, 75ЊC), and 1.5 ϫ 10 (taC:N,
(
i/i ) ϭ 1.9542(t /t){1 Ϫ exp[Ϫ1.2564(t/t )]}
[3]
[4]
m
m
m
2
7
5ЊC) cm /s, are in good agreement with those obtained from the
plot of the decaying portion of high overpoten-
tial transients by means of the Cottrell equation, 6.8 ϫ 10 (W,
60ЊC), 1.6 ϫ 10 (GC, 75ЊC), and 1.3 ϫ 10 (taC:N, 75ЊC) cm /s.
These diffusion coefficients of Al Cl are also in good agreement
2
2
Ϫ1/2
(
i/i ) ϭ 1.2254(t /t){1 Ϫ exp[Ϫ2.3367(t/t )]}
slopes of i vs. t
m
m
m
Ϫ7
These expressions correlate the dimensionless current (i/i ) to the
Ϫ6
Ϫ6
2
m
dimensionless time (t/t ) for each type of nucleation. It is necessary
Ϫ
7
m
2
to correct the experimental data for the generally small delay time, t ,
o
with previous measurements considering the temperature varia-
29
preceding the onset of nucleation. Thus, we redefine the time axis
as tЈ ϭ t Ϫ t and the time of the maximum current as tЈ ϭ t Ϫ to.
6,9,15,31-34
tion.
o
m
m
Once it is established that the nucleation is instantaneous, the nu-
Estimation of t can be accomplished better and more conveniently
o
cleation number densities (N) at different overpotentials can be cal-
2
from the linear regression analysis of the linear plot of (i/i ) vs. t
27-30
m
culated by Eq. 6 and 7
1
/2 14-21,30
rather than that of i vs. t
Table I for each electrode.
The dimensionless experimental plots, (i/i ) vs. tЈ/tЈ along with
.
The resulting values are shown in
im ϭ 0.6382nFCD(kN)1/2
[6]
[7]
m 2
m
k ϭ (8␲CM/␳)1/2
theoretical plots resulting from Eq. 3 and 4, are shown in Fig. 5b, for
taC:N, and Fig. 6, for W and GC. All these plots suggest that the
deposition of Al on all three electrodes agrees well with the limiting
model of instantaneous nucleation, although there is some small
deviation for taC:N we treat below. Particularly for the W electrode,
where M and ␳ are the molecular weight and density of deposited
material, Al. The value of k was calculated to be 0.167 from the given
Ϫ
concentration of Al Cl and known values of M and ␳ for Al (26.98 g
2
7
3
and 2.7 g/cm , respectively). Average diffusion coefficients, 6.7 ϫ
6
,33
Ϫ7
Ϫ6
2
most previous studies in both ambient
and high temperature
10 for 60ЊC (W) and 1.5 ϫ 10 cm /s for 75ЊC (GC and taC:N)
are used for calculation of the number density of nucleation (N).
The values obtained also appear in Table I and clearly show po-
tential dependency, with ranges 10 -10 for W and 10 -10 /cm for
GC and taC:N, within the overpotentials applied to each electrode.
The nucleation densities at these potential regimes correspond to less
than one in a million surface atoms, from the density of atomic sites
9
,14
(
AlCl /NaCl) molten salts
showed a good instantaneous limiting
3
behavior in the nucleation of Al on this substrate, though progressive
nucleation was reported by Carlin et al. from their nucleation and
morphology studies of Al on a small (250 ␮m diam) W disk elec-
6
7
5
6
2
1
5
trode in this AlCl /EMIC melt.
3
It is also possible to get information about the diffusion coeffi-
Ϫ
15
2
14,32
cient of the electroactive species, Al Cl , from the maximum of cur-
(ϳ10 /cm ) for a normal substrate, such as W and GC.
results have been observed for other systems following the three-
Similar
2
7
2
7-30
rent-time transients in the case of instantaneous nucleation
via
Ϫ
7
Table I. Analysis of Al nucleation process on W, GC, and taC:N using chronoamperometric data and diffusion coefficient of Al Cl . Percent
2
Ϫ7
Ϫ6
2
averaged diffusion coefficients 6.7 ϫ 10 for W and 1.5 ϫ 10 cm /s for GC and taC:N were used.
6
2
6
a
10Ϫ6
N
(1/cm )
6
b
Ϫ␩
mV)
103 im
(A/cm )
tm
(s)
to
(s)
10 i tЈ
10 D
10 D
m
m
N
C
2
2
4
2
2
2
(
(A s/cm )
(cm /s)
(cm /s)
W (60ЊC)
1
1
35
50
18.00
20.30
0.33
0.29
0.005
0.002
106
111
0.63
0.69
0.66
11.60
14.70
Average D:
0.68
GC (75ЊC)
1
2
2
2
75
00
25
50
15.50
17.40
21.40
24.10
0.97
0.81
0.56
0.41
0.061
0.047
0.035
0.029
218
230
240
226
1.30
1.40
1.40
1.30
1.40
01.72
02.17
03.29
04.17
Average D:
1.60
1.30
taC:N (75ЊC)
3
3
4
4
5
00
50
00
50
00
09.00
11.50
14.80
18.00
22.70
3.38
2.10
1.28
0.82
0.57
0.240
0.170
0.095
0.067
0.065
255
254
260
246
262
1.50
1.50
1.50
1.40
1.60
1.50
00.58
00.95
01.58
02.34
03.71
Average D:
a
2
D : Diffusion coefficient obtained from nucleation data i tЈ .
N
m m
b
Ϫ1/2
D : Diffusion coefficient obtained from the linearity in i vs. t
(Cottrell relation).
C
Downloaded on 2015-03-07 to IP 130.15.241.167 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

3
374
Journal of The Electrochemical Society, 147 (9) 3370-3376 (2000)
S0013-4651(99)11-076-0 CCC: $7.00 © The Electrochemical Society, Inc.
2
{
1 Ϫ exp(Ϫx[t/t ) ϩ ␣(1 Ϫ exp(Ϫxt/␣t ))]}
2
m m
(
i/i ) ϭ (tm /t)
m
1 Ϫ exp[Ϫx ϩ ␣(1 Ϫ exp(Ϫx/␣))]}2
{
[8]
where ␣ and x are adjustable parameters that contain information
about the number density of active sites (N ) for nucleation on the
o
electrode surface, and the potential dependent nucleation rate (A) per
2
7-30,32
active site defined in Eq. 9 and 10
N ϭ x/[␲kDt ]
[9]
o
m
A ϭ x/␣tm
[10]
For the limiting cases of instantaneous and progressive nucleation, ␣
3
2
approaches 0 with x Ϸ 1.2564 and ϱ with x Ϸ 2.3367, respectively.
Equation 8 was fit to the data in Fig. 5b by adjusting these two
parameters so that the variance between the experimental and calcu-
lated values of (i/i ) is minimized. The estimated values of ␣ and x
2
m
for each transient are summarized in Table II. The best fits of ␣ and
x, for all transients, belong to the ranges 0.06-0.12 and 1.39-1.53,
respectively, and each parameter increases with overpotential. The
theoretical lines generated with these values, for ␩ ϭ Ϫ300 and
Ϫ500 mV, are shown in Fig. 7 and demonstrate that this general
model provides a reasonable representation of the experimental data.
The relatively small values of ␣ for the transients imply that the
nucleation of Al on taC:N is almost instantaneous on a finite number
of active sites, as predicted from earlier analysis. Similar nucleation
Figure 6. Comparison of the dimensionless experimental data derived from
the current-time transients from the same melts used for Fig. 5 with the the-
oretical models for instantaneous and progressive nucleation. (a) W at 60ЊC
and (b) GC at 75ЊC.
dimensional nucleation and growth model.27-29-30 Generally, these
low nuclei densities can be explained by the fast spreading of “nucle-
ation exclusion zones,” which develop due to depleted concentration
around the growing nuclei, since the active sites remaining inside the
exclusion zones are not available for nucleus formation due to the
reduced supersaturation.28
However, if the overpotential range for each electrode is consid-
ered, there are much greater differences in N between the electrodes
because a much larger overpotential was required to obtain a given
value of N in the order of W < GC < taC:N. For example, at ␩ ϭ
1
50 mV, N for GC is less than 10% of that of W and a much small-
er value is expected at taC:N, ignoring other experimental differ-
ences. It can be inferred that the intrinsic number of active sites (No)
at taC:N is much smaller than those at the other electrodes. This fac-
tor can have serious impact on nucleus formation in the very initial
stage of phase transition, and, subsequently, on plate adhesion and
stripping behavior.
Even though the above analysis suggests an instantaneous limit-
ing behavior of nucleation of Al on taC:N, all the dimensionless
plots of each current-time transients for this substrate, Fig. 5b, show
a slight deviation from the ideal limiting case of instantaneous nucle-
ation, unlike W and GC. Because of this slight deviation, the nucle-
ation on taC:N can be also considered as an intermediate type of
nucleation, involving instantaneous nucleation on a finite number of
Figure 7. Comparison of the dimensionless experimental current-time tran-
sients derived from the data in Fig. 5b with the theoretical models of inter-
mediate type of nucleation with diffusion-controlled growth of nuclei repre-
sented by Eq. 8. (a) ␩ ϭ Ϫ300 mV, ␣ ϭ 0.062, x ϭ 1.39 and (b) ␩ ϭ
Ϫ500 mV, ␣ ϭ 0.118, x ϭ 1.53.
2
8,32
active sites,
but with a progressive contribution on some addi-
tional small number of active sites driven by large overpotentials.
For these cases, the theoretical transients are represented by a gen-
1
8,27-30,32
eral model
Downloaded on 2015-03-07 to IP 130.15.241.167 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, 147 (9) 3370-3376 (2000)
3375
S0013-4651(99)11-076-0 CCC: $7.00 © The Electrochemical Society, Inc.
active sites (N ) is increasing more rapidly with overpotential than
o
Table II. Parameters for fitting each current-time transient to
Eq. 8 and the number density of active sites for nucleation and
the rate of nucleation per active site.
their rate of nucleation (A). Some rationale for this follows.
The amorphous and smooth taC:N electrodes normally have a
much lower density of intrinsic nucleation sites, whose physical ori-
gin has not yet been defined. As a result, a much smaller number of
sites is activated initially compared to that of W and GC, even though
the overpotential applied is much greater. Then, instantaneous nucle-
ation follows on all these initially activated sites. Meanwhile, addi-
tional nucleation active sites are being produced slowly, driven by the
substantial overpotential, although the contribution of these addition-
al sites is relatively small overall. Therefore, the partial progressive
nature of nucleation on this material is considered to be a reflection
of the extra nucleation on these slowly generated late active sites,
since the nucleation at active sites is always fast and instantaneous.
This interpretation is also supported by the zero value of the critical
10Ϫ6 N (1/cm )
2
Ϫ␩ (mV)
␣
x
A (1/s)
o
3
3
4
4
5
00
50
00
50
00
0.062
0.075
0.090
0.103
0.118
1.39
1.42
1.46
1.47
1.53
07.1
09.8
13.7
18.9
25.6
0.60
1.00
1.68
2.64
4.10
patterns were observed for the deposition and nucleation of Hg and
19,21
Sn on glassy carbon in acidic AlCl /EMIC melts.
However, from
3
Table II and Fig. 7, a moderately increasing trend of deviation from
ideal instantaneous behavior is evident with higher overpotentials (␩)
from the slight increase of ␣ and x with overpotentials, where ␣ is a
number (n ) of atoms for nucleation because each active site already
acts as a critical nucleus at this material (see below).
k
However, this additional nucleation process does not occur at all
the possible active sites induced by high overpotential because of the
competition with a fast spreading of nucleation exclusion zones.
Further nucleation is not normally available in the exclusion zones
due to the depleted concentration resulting from a local deformation
quantity proportional to N /A. The number density of nucleation
o
active sites (N ) and potential dependent nucleation rate (A), calcu-
o
lated by Eq. 9 and 10, are shown in Table II, where the same averaged
Ϫ
7
Ϫ6
2
diffusion coefficient for Al Cl , 1.4 ϫ 10 cm /s at 75ЊC, was used
2
2
8
throughout. In fact, Fig. 8 shows that both quantities are exponential-
ly dependent on the overpotential in the given range, Ϫ300 to
Ϫ500 mV. As with the N in the instantaneous limiting model, the No
also depends strongly on the overpotential but the values do not dif-
fer much from N for each transient and are, generally, slightly larger.
This is to say that larger overpotentials, as required for taC:N, induce
a small progressive nature to the process, i.e., the number density of
of the concentration gradients. The rate of generation of extra
active sites and the propagation of the nucleation exclusion zones
both depend strongly on the overpotential; generation of a higher
number of extra, overpotential-induced late active sites and faster
spreading of exclusion zones are expected with higher overpotential.
The exact interrelation between these two effects is not simple and
is beyond the scope of this work, but zone spreading is more critical
in determination of net nucleation on these extra sites because addi-
tional nucleation is suppressed as soon as these sites become includ-
ed within nucleation exclusion zones.
On the whole, an instantaneous nucleation on the finite number
of active sites is a close description for the Al deposition on taC:N,
even though there is a slight deviation from ideal behavior at large
overpotentials. Other than the latter, the results parallel those on
glassy carbon or W.
The overpotential dependence of A can be used to determine the
critical number of atoms, n , required for the formation of a stable
k
3
2,35-37
nucleus, according to the atomistic theory of nucleation,
1
4,22,32
through Eq. 11
nk ϭ (2.303 RT/nF)[d log(A)/d␩] Ϫ ␤
[11]
where ␤ is the transfer coefficient. Assuming that ␤ ϭ 0.5, Eq. 11
gives n Ϸ 0. This means that the active site itself, on the taC:N elec-
k
trode surface, plays the role of a critical nucleus during the electro-
2
2,26,35-37
deposition of Al from this acidic AlCl /EMIC melt.
Similar
3
low values of n , ranging from 0 to 1, can be found in earlier reports
k
3
7,38
on GC substrates in many cases in aqueous solution
ent temperature chloroaluminate melts. Recently, the same result,
and ambi-
2
2
n ϭ 0, was also reported from this laboratory for Ag deposition on
k
2
6
a BDD substrate from a dilute silver cyanide plating bath. Thus,
this result of a low value of n for taC:N and BDD also supports the
k
generalization of a small value of n for the electrodeposition of met-
k
als on nonmetallic substrates (such as these carbon-based materials),
2
2
regardless of the solvent/electrolyte combination employed.
Conclusions
Electrochemical properties of the amorphous and atomically
smooth taC:N in different compositions of ambient temperature
AlCl /EMIC melts have been studied in parallel with those of W and
3
GC. Generally, taC:N exhibits a wider electrochemical window, in
all melt compositions, than those of W and GC. High nucleation
overpotential for Al deposition with no UPD features on taC:N and
the wider electrochemical windows in different melts suggest an
inertness of taC:N comparable to BDD. Instantaneous nucleation on
small number of active sites with the hemispherical diffusion con-
trolled growth of nuclei for Al deposition is evident on W and GC
while there is a slight deviation from ideal instantaneous nucleation
Figure 8. Overpotential dependence of (a) the number density of active site
for nucleation (N ) and (b) the rate of nucleation per active site (A).
o
Downloaded on 2015-03-07 to IP 130.15.241.167 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

3
376
Journal of The Electrochemical Society, 147 (9) 3370-3376 (2000)
S0013-4651(99)11-076-0 CCC: $7.00 © The Electrochemical Society, Inc.
behavior on taC:N. This feature can be explained by the intrinsic low
value of number density of nucleation active sites, which act as crit-
ical nuclei during the initial stage of nucleation, and the extra active
sites induced by the large overpotential. The additional, i.e., overpo-
tential induced, delayed active sites impose a partial progressive
nature to the overall Al nucleation scheme on taC:N, resulting in a
slight deviation from the ideal instantaneous limiting model. A
kinetic limitation for nucleus generation, at the initial stage of phase
formation, on taC:N is expected from the moderately increasing
trend of deviation with increasing overpotential.
13. J. Fuller, R. A. Osteryoung, and R. T. Carlin, J. Electrochem. Soc., 142, 3632
1995).
(
1
4. T. P. Moffat, J. Electrochem. Soc., 141, 3059 (1994).
1
5. R. T. Carlin, W. Crawford, and M. Bersch, J. Electrochem. Soc., 139, 2720 (1992).
16. X.-H. Xu and C. L. Hussey, J. Electrochem. Soc., 140, 1226 (1993).
17. R. T. Carlin, P. C. Trulove, and H. C. Delong, J. Electrochem. Soc., 143, 2747
(
1996).
8. W. R. Pitner, C. L. Hussey, and G. R. Stafford, J. Electrochem. Soc., 143, 130
1996).
1
(
19. X.-H. Xu and C. L. Hussey, J. Electrochem. Soc., 139, 3103 (1992).
20. X.-H. Xu and C. L. Hussey, J. Electrochem. Soc., 139, 1295 (1992).
2
2
1. X.-H. Xu and C. L. Hussey, J. Electrochem. Soc., 140, 618 (1993).
2. W. R. Pitner and C. L. Hussey, J. Electrochem. Soc., 144, 3095 (1997).
Case Western Reserve University assisted in meeting the publication
costs of this article.
23. G. R. Stafford, J. Electrochem. Soc., 141, 945 (1994).
24. J.-J. Lee, I. Bae, D. A. Scherson, B. Miller, and K. A. Wheeler, J. Electrochem.
Soc., 147, 562 (2000).
References
2
5. N. Vinokur, B. Miller, Y. Avyigal, and R. Kalish, J. Electrochem. Soc., 143, L238
1
.
.
X. Shi, D. Flynn, B. K. Tay, S. Prawer, K. W. Nugent, S. R. P. Silva, Y. Lifshitz, and
W. I. Milne, Philos. Mag. B, 76, 351 (1997).
K. Yoo, B. Miller, R. Kalish, and X. Shi, Electrochem. Solid-State Lett., 2, 23
(1996).
26. N. Vinokur, B. Miller, Y. Avyigal, and R. Kalish, J. Electrochem. Soc., 146, 125
(1999).
2
(
1999).
27. B. R. Scharifker and G. Hills, Electrochim. Acta, 28, 879 (1983).
28. B. R. Scharifker, in Electrochemistry in Transition, O. J. Murphy, S. Srinivasan,
and B. E. Conway, Editors, p. 499, Plenum Press, New York (1992).
29. P. M. Rigano, C. Mayer, and T. Chierchie, J. Electroanal. Chem., 248, 219 (1988).
30. B. R. Scharifker and J. Mostany, J. Electroanal. Chem., 177, 13 (1984).
31. P. K. Lai and M. Skyllas-Kazacos, Electrochim. Acta, 32, 1443 (1987).
32. V. Tsakova and A. Milchev, J. Electroanal. Chem., 235, 237 (1987).
33. P. K. Lai and M. Skyllas-Kazacos, J. Electroanal. Chem., Interfacial Electrochem.,
248, 431 (1988).
3
4
.
.
C. L. Hussey, Pure Appl. Chem., 60, 1763 (1988).
J. S. Wilkes, J. A. Levisky, R. A. Wilson, and C. L. Hussey, Inorg. Chem., 21, 1263
(1982).
5
.
C. L. Hussey, in Chemistry of Nonaqueous Solutions: Current Progress,
G. Mamantov and A. I. Popovs, Editors, p. 225, New York (1994).
J. Robinson and R. A. Osteryoung, J. Electrochem. Soc., 127, 122 (1980).
T. J. Melton, J. Joyce, J. T. Maloy, J. A. Boon, and J. S. Wilkes, J. Electrochem.
Soc., 137, 3865 (1990).
6
7
.
.
8
.
S. Scordilis-Kelley, J. Fuller, R. T. Carlin, and J. S. Wilkes, J. Electrochem. Soc.,
34. Y. Chryssoulakis, J.-C. Poignet, and G. Manoli, J. Appl. Electrochem., 17, 857
(1987).
1
39, 694 (1992).
9
.
P. Rolland and G. Mamantov, J. Electrochem. Soc., 123, 1299 (1976).
35. A. Milchev, S. Stoyanov, and R. Kaischev, Thin Solids Films, 22, 255 (1974).
36. A. Milchev, S. Stoyanov, and R. Kaischev, Thin Solids Films, 22, 267 (1974).
37. A. Milchev, E. Vassilleva, and V. Kertov, J. Electroanal. Chem., 107, 323 (1980).
38. G. Gunawardena, G. Hills, and I. Montenegro, J. Electroanal. Chem., 138, 241
(1982).
1
1
1
0. B. J. Welch and R. A. Osteryoung, J. Electroanal. Chem., 118, 455 (1981).
1. T. L. Riechel and J. S. Wilkes, J. Electrochem. Soc., 139, 977 (1992).
2. R. T. Carlin, J. Fuller, W. K. Kuhn, M. J. Lysaght, and P. C. Trulove, J. Appl. Elec-
trochem., 26, 1147 (1996).
Downloaded on 2015-03-07 to IP 130.15.241.167 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).