Electrochemical and AFM study of cobalt nucleation mechanisms on glassy carbon from ammonium sulfate solutions

Article abstract of DOI:10.1016/j.electacta.2004.05.028

Nucleation mechanisms of cobalt on a glassy carbon electrode (gce) from aqueous ammonium sulfate solutions were investigated through the electrochemical techniques of cyclic voltammetry (cv) and chronoamperometry (ca), coupled with atomic force microscopy (AFM) studies. The studied parameters were pH, cobalt concentration, temperature, scanning rate, and deposition potential. It was found that scanning in the cathodic direction produced two peaks, corresponding to cobalt and hydrogen reduction, respectively. Scanning in the anodic direction was characterized by cobalt dissolution, which was interrupted by formation of cobalt hydroxide, causing a second anodic peak. The amperometric study found progressive nucleation mechanisms, in contrast to the instantaneous nucleation mechanisms determined by the AFM study. An explanation for the contradictory nucleation mechanisms shown in the two studies is provided.

Full text of DOI:10.1016/j.electacta.2004.05.028

Electrochimica Acta 49 (2004) 4719–4732
www.elsevier.com/locate/electacta
Electrochemical and AFM study of cobalt nucleation mechanisms
on glassy carbon from ammonium sulfate solutions
*
Darko Grujicic, Batric Pesic
Department of Materials Science and Engineering, College of Engineering, University of Idaho,
McClure Hall, Moscow, ID 83844-3024, USA
Received 29 January 2004; accepted 21 May 2004
Available online 26 June 2004
Abstract
Nucleation mechanisms of cobalt on a glassy carbon electrode (gce) from aqueous ammonium sulfate solutions were investigated through
the electrochemical techniques of cyclic voltammetry (cv) and chronoamperometry (ca), coupled with atomic force microscopy (AFM)
studies. The studied parameters were pH, cobalt concentration, temperature, scanning rate, and deposition potential. It was found that
scanning in the cathodic direction produced two peaks, corresponding to cobalt and hydrogen reduction, respectively. Scanning in the anodic
direction was characterized by cobalt dissolution, which was interrupted by formation of cobalt hydroxide, causing a second anodic peak. The
amperometric study found progressive nucleation mechanisms, in contrast to the instantaneous nucleation mechanisms determined by the
AFM study. An explanation for the contradictory nucleation mechanisms shown in the two studies is provided.
ß 2004 Elsevier Ltd. All rights reserved.
Keywords: Cobalt nucleation; Glassy carbon electrode; Ammonia; Cyclic voltammetry; Chronoamperometry; Atomic force microscopy
1. Introduction
reaction mechanisms, it has been found that cobalt electro-
deposition initiates predominantly via progressive nuclea-
tion, which can change to instantaneous nucleation
mechanisms by enhancing mass transfer, by using ultrasound
[7], or by increasing cobalt concentration [10], for example.
It is surprising that the electrodeposition of cobalt from
ammoniacal solutions has not received greater attention
considering that the complexing property of ammonia can
be utilized to modulate the properties of deposited cobalt via
solution speciation. Soto et al. [9] and Palomar-Pardave et al.
[11] investigated cobalt electrodeposition onto glassy carbon
from ammonium chloride solutions. They found that cobalt
nucleation on glassy carbon follows either a progressive
nucleation mechanism or a combinationof severalnucleation
mechanisms. However, their results were not accompanied
by a morphological study, so the proposed nucleation models
could not be verified. In another study using ammonium
chloride solutions Mendoza-Huizar et al. [12] found evi-
dence of underpotential deposition of cobalt on gold.
This paper presents a study of cobalt nucleation mechan-
isms from ammonium sulfate solutions on glassy carbon
According to recent advances in the theory and produc-
tion of nanomagnets [1,2] as memory storage units, the
morphology of deposited magnetic materials greatly influ-
ences the magnetic properties of thin films. Cobalt (and its
alloys) is one of the important ferromagnetic components of
such magnetic thin film materials. Electrochemical deposi-
tion is the most often used technique for its incorporation
into the structures of, for example, GMR heads and spin
valve devices [3–6] (and references therein).
Cobalt electrodeposition studies can be categorized
according to the working electrode and the type of solution
systems used. Most studies have been performed with glassy
carbon electrodes [4,7–11], others being gold [10,12], nickel
[13] and copper [4,11]. Chloride solutions have been the
preferred systems for studying the electrochemistry of cobalt
[8–13], rather than sulfate systems [7,14]. Regarding the
* Corresponding author. Tel.: +1 208 885 6569; fax: +1 208 885 2855.
E-mail address: pesic@uidaho.edu (B. Pesic).
0013-4686/$ – see front matter ß 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.electacta.2004.05.028

4720
D. Grujicic, B. Pesic / Electrochimica Acta 49 (2004) 4719–4732
electrode which combines the electrochemical methods of
cyclic voltammetry and chronoamperometry with atomic
force microscopy.
2. Experimental
Reagent grade CoSO4ꢀ7H2O was used to prepare 0.005,
0.01 and 0.05 M Co(II) solutions. All solutions (Type I
purity water, Barnstead NANOpure II) contained 1 M
reagent grade (NH4)2SO4 that served as a buffering agent
and a supporting electrolyte, unless stated otherwise in the
text. Solution pH was adjusted by adding dilute NH4OH (pH
6) or concentrated NH4OH (pH 9).
The electrochemical setup was a standard three-electrode
cell with glassy carbon as a working electrode, platinum foil
as a counter electrode, and silver–silver chloride as a
reference electrode (Cypress Systems microelectrode,
E0h = +0.222 V). The glassy carbon working electrode
(0.442 cm2) was a non-porous disk (Sigri), RMS about
1nm upon polishing. Electrochemical experiments were
controlled with a potentiostat/galvanostat (Ametek, PAR
283) under computerized control (PAR, M270 Software).
Surface morphology was characterized by atomic force
microscopy (VEECO, Digital Instruments, Model Nano-
scope IIIa-MultiMode) by using the taping mode technique
under fluid. A Digital Instruments computer program was
also used for the cross-section and nuclei population density
analyses of AFM images.
Fig. 1. EAg/AgCl–pH diagram for the cobalt–ammonia solution containing
0.01 M Co2+ and 1.0 M (NH4)2SO4.
tials, the predominant cobalt species was cobaltous ion,
which according to [16] is actually in the form of
2+
Co(H2O)6 aquo complex. The pH 7–11 region was char-
acterized by cobalt–ammonia complexes in which the num-
ber of ammonia ligands in hydration shell successively
increased from one to five as pH increased. The third region,
pH above 11, contained cobalt hydroxide as the predominant
species. Metallic cobalt was stable in the entire pH range at
lower solution potentials.
Based on the information provided by the Eh–pH dia-
gram, pH 6 and 9 were selected for the electrochemical
studies. The pH value of 6 was selected because electro-
deposition of cobalt occurred from hydrated cobaltous ion.
To study the role of cobalt complexation by ammines during
electrodeposition, pH 9 solutions were used. The third
region, pH above 11, was not considered in this study.
Distribution–Eh diagrams for cobalt-ammonia system at
pH 6 and 9 are presented in Fig. 2a and b, respectively.
At pH 6, the predominant cobaltous species at more
positive solution potentials was Co2+. The concentration
of free ammonia in the 1 M (NH4)2SO4 solution at pH 6 was
only 0.001 M, which explains the limited presence of
Thermodynamic calculations were performed by thermo-
chemical software (Stabcal) [15].
3. Results and discussion
3.1. Cobalt–ammonia chemistry
In aqueous solutions, cobalt is present in the form of
divalent cobaltous ion with octahedral coordination, having
four equatorial and two axial coordination sites [16], each
occupied by water molecules. When ammonia is added,
water molecules are displaced from octahedral sites, first
from equatorial and then from axial positions, because of the
preferred coordination of cobalt with nitrogen. Depending
on ammonia concentration and pH, ammonia molecules can
successively replace water molecules in cobalt’s coordina-
tion octahedron, forming cobaltous ammines that may have
between one and six ammonia molecules.
2+
CoNH3 and Co(NH3)22+. According to Fig. 2a, cobalt
reduction occurred at potentials more negative than ꢁ0.6
to ꢁ0.7 V. At pH 9, cobalt – ammine complexes were stable
in the fairly narrow potential regions from ꢁ0.7 to ꢁ0.8 V
and ꢁ0.2 to ꢁ0.3 V. Cobalt ammines were not stable at
higher potentials because of the formation of cobalt hydro-
xide. Reduction to metallic cobalt began at ꢁ0.7 to ꢁ0.8 V.
Because of numerous cobalt species that can exist in
aqueous ammonia solutions, it was necessary to determine
cobalt speciation in the entire pH range. For that purpose,
Eh–pH and distribution–Eh diagrams were constructed for
the cobalt–ammonia systems under investigation. Fig. 1
shows the Eh–pH diagram for the system with 0.005 M
cobaltous sulfate and 1 M of ammonium sulfate.
3.2. Cyclic voltammetry
3.2.1. Cathodic branch
Cyclic voltammetry was studied as a function of scanning
rate, number of cycles, concentration of reacting species and
solution pH. All cv experiments were performed in the
potential range from an initial potential of 0 mV to a vertex
potential of ꢁ1300 mV.
Several characteristic pH regions of cobalt speciation
were present. In the pH 0–7 region at more positive poten-

D. Grujicic, B. Pesic / Electrochimica Acta 49 (2004) 4719–4732
4721
from the crystallographic misfit between cobalt (hexagonal
closely packed lattice) and glassy carbon (interlocked,
twisted and entangled microfibrils [18]). The peak that
corresponded to reaction (1) at pH 6 and reaction (2) at
pH 9 was denoted Ic on the cvs. Further increase of negative
potential resulted in peak IIc, which corresponded to the
hydrogen ion reduction reaction. At pH 6, peak Ic was
observed only as a change of slope of the cathodic trace
because of the significant contribution of current from
hydrogen ion reduction during electroreduction of cobalt.
As the scan was reversed at ꢁ1300 mV, the current
decayed but remained cathodic up to around ꢁ700 mV,
indicating further reduction of cobalt species. As the poten-
tial became more positive, the current crossed over the
forward current trace, became anodic, increased and formed
peak Ia, which was immediately followed by peak IIa. After
peak IIa, the current decayed to zero and remained zero, or
slightly anodic, until the end of the scan.
3.2.1.2. Effect of cycling
In order to investigate the reversibility of reactions in the
systems under study, three cycle cvs in 0.05 M Co2+ solu-
tions at pH 6 and 9 were run, Fig. 4. At pH 6, the current at
the end of the first cycle was anodic, indicating that an
oxidation reaction was still present. At the beginning of the
second cycle, the current remained anodic up to ꢁ620 mV.
Peak Ic in the second cycle was smaller and at a more
positive potential than peak Ic in the first cycle, indicating a
reduced concentration of reacting cobaltous species. Anodic
peaks Ia and IIa were also smaller than the corresponding
peaks in the first cycle. The same trend continued for the
third cv cycle, in which cathodic peak Ic as well as anodic
peaks Ia and IIa were the smallest peaks of all three cycles.
At pH 9, the current was also anodic at the beginning of
the second cv cycle. As the potential sweep proceeded in the
negative direction in the second cycle, the current decayed
and became zero at ꢁ670 mV. With further increase in
negative potential, the current became cathodic and formed
one additional cathodic peak, before peak Ic. This peak
corresponded to the reduction of new species formed during
the first cycle, residual Co(OH)2. It should be noted that the
peak that corresponded to cobalt hydroxide reduction was
not present for lower cobalt concentrations, 0.005 and
0.01 M. As the potential swept toward the positive direction,
the same two anodic peaks appeared as in the first cycle.
Unlike the results at pH 6, the magnitudes of anodic peaks
remained the same for all three cycles, while the cathodic
peaks and their magnitudes stayed the same for the second
and third cycle.
Fig. 2. (a and b) Distribution–EAg/AgCl diagrams for cobalt–ammonia
systems at (a) pH 6 and (b) pH 9, each containing 0.01 M Co2+ and 1 M
(NH4)2SO4.
3.2.1.1. Effect of scanning rate
Fig. 3a and b shows the effect of scanning rate on cvs at
pH 6 and 9, respectively. The presence of zero current from
the beginning of the scan to about ꢁ1100 mV indicated no
reaction in this potential range. Below ꢁ1100 mV (the exact
potential depends on the pH and scanning rate) the cathodic
current began to increase, resulting in the formation of the
first cathodic peak, Ic, which corresponds to cobalt reduction
according to reaction (1) at pH 6, and reaction (2) at pH (9):
2þ
CoðH2OÞ6 þ 2e ¼ Co0 þ 6H2O;
E0h ¼ ꢁ0:277 V ðEA0 g=AgCl ¼ ꢁ0:499 VÞ (1)
2þ
CoðNH3Þ5 þ 2e ¼ Co0 þ 5NH3;
3.2.1.3. Effect of cobalt concentration
Fig. 5a and b presents the effect of Co(II) concentration
on cvs at pH 6 and 9, respectively.
E0 ¼ ꢁ0:440 V ðEA0 g=AgCl ¼ ꢁ0:662 VÞ (2)
The difference between the equilibrium cobalt redox
potential and the onset of cobalt deposition was due to
the nucleation overpotential on glassy carbon [17] resulting
The magnitude of the current peaks increased propor-
tionally to the increase of Co(II) concentration, as expected.
The concentration of cobalt ions in solution affected the

4722
D. Grujicic, B. Pesic / Electrochimica Acta 49 (2004) 4719–4732
Fig. 3. (a and b) Cyclic voltammetry of cobalt as a function of scanning rate at (a) pH 6 and (b) pH 9. Conditions: 0.01 M Co2+, 1 M (NH4)2SO4.
position of peak Ic, which became more positive as the
concentration of Co(II) increased. The shift of peak Ic
potential is probably related to concentration overpotential,
as the higher concentration of reacting species reduced the
concentration overpotential for cobalt reduction [19]. The
significance of the shift of peak Ic is obvious at pH 6. At the
lowest cobalt concentration of 0.005 M peak Ic overlaps
with hydrogen ion reduction, represented by peak IIc. At the
highest cobalt concentration of 0.05 M, peaks Ic and IIc are
separated due to the shift in peak Ic. This implies that at
higher cobalt concentrations, deposition initially proceeds in
the absence of hydrogen ion reduction, an important fact for
the following chronoamperometric study.
3.2.1.4. Morphological study
In order to establish the effect of cv parameters on the
morphology of deposited cobalt, a series of AFM images
was taken at the cv crossover potentials. A crossover poten-
tial, defined as the potential at which the reverse current
trace crosses over the forward current trace, is characteristic
for a cv in which a metal is being deposited and dissolved
[17]. In an ideal case, when there are no background currents
during a cv, the crossover potential should be located on an
x-axis, with no net current; i.e., when cathodic current
changes to anodic.
The results of the morphological AFM study are pre-
sented in Fig. 6a–h. The most characteristic feature of cobalt
nucleation was that each nucleus had a hexagonal plane,
with a multidirectional 3D-growth (best seen in Fig. 6e).
Nakano et al. [14], who performed a study of electrolysis
factors affecting the crystal orientation and morphology of
deposited cobalt, found by X-ray diffraction that cobalt
nucleates in a closely packed hexagonal form with several
crystal orientations. The flat hexagonal plane seen on the
nuclei in this study corresponds to the (0 0 0 2) plane of
densest packing.
According to the present AFM study, the nuclei sizes
were a function of scanning rate and pH. The nuclei sizes
were the largest at the slowest scanning rate, 5 mV sꢁ1
.
Nuclei population density, on the other hand, increased when
the scanning rate was increased from 5 to 100 mV sꢁ1. This
effect was more pronounced at pH 6 than at pH 9.
The difference in the nuclei population densities
observed at pH 6 and 9 was probably due to hydrogen
evolution at pH 6 and its absence at pH 9. The overlap of
peak Ic with peak IIc shown in the cyclic voltammograms in
Fig. 3a suggests that at lower cobalt concentrations hydro-
gen evolution coincided with the formation of cobalt nuclei.
Control cvs of glassy carbon electrode in 1 M ammonium
sulfate solutions at pH 6 and 9 in the absence of cobalt (not
presented) showed that there was no hydrogen evolution on
gce in the entire potential range studied. Therefore, the only
viable substrates for hydrogen gas evolution reaction were
Fig. 4. Effect of cycling on cyclic voltammetry of cobalt at pH 6 and 9.
Conditions: 0.05 M Co2+, 1 M (NH4)2SO4, 50 mV sꢁ1
.

D. Grujicic, B. Pesic / Electrochimica Acta 49 (2004) 4719–4732
4723
Fig. 5. (a and b) Cyclic voltammetry of cobalt as a function of cobalt concentration at (a) pH 6 and (b) pH 9. Scanning rate: 20 mV sꢁ1
.
the newly formed cobalt nuclei. As soon as metallic cobalt
appeared on the surface of the electrode, it served as a
reaction site for hydrogen evolution. The products of the
hydrogen reduction were hydrogen gas bubbles that formed
on the surface of cobalt nuclei, grew, and eventually pre-
vented further cobalt reduction from taking place on the
cobalt nuclei surface. Because cobalt ions could not be
reduced on the surface of metallic cobalt, they had to be
reduced elsewhere on glassy carbon, leading to the forma-
tion of new cobalt nuclei. Overall, the size of nuclei
decreased, but the population density increased.
3.2.2. Anodic branch
The mechanism of cobalt oxidation in neutral and basic
media has been a subject of previous studies [20–23].
Although there are differences in the interpretation of the
overall oxidation mechanism, these studies show that cobal-
tous hydroxide forms either as an intermediate or as a final
product of oxidation.
According to Pourbaix [24], cobalt is passivated in aqu-
eous solutions by forming a hydrated cobalt oxide layer,
Co(OH)2, according to a hydrolytic reaction (3):
Co þ 2H2O ¼ CoðOHÞ2 þ H2
In order to relate the presence of ammonia to two anodic
peaks, an additional set of experiments was performed in
oxygen-free solutions under argon atmosphere. Cobalt films,
(3)
Due to much lower hydrogen ion concentration at pH 9
than at pH 6, a near-absence of hydrogen gas evolution
allowed unhindered growth of cobalt nuclei, resulting in
nuclei larger than those deposited at pH 6.
Fig. 6. (a–h) Effect of scanning rate on nuclei population density of cobalt for solutions at (a) pH 6 and (b) pH 9. The AFM images were taken at the vertex
potential, E = ꢁ1300 mV. Conditions: 0.005 M Co2+, 1 M (NH4)2SO4.

4724
D. Grujicic, B. Pesic / Electrochimica Acta 49 (2004) 4719–4732
Fig. 7. (a and b) Effect of conditioning time on linear sweep voltammetry in the potential range corresponding to anodic dissolution of cobalt for solutions at (a)
pH 6 and (b) pH 9. Conditioning is defined as the time of cobalt film exposure to the solution prior to initiation of linear sweep voltammetry. Inset in (b) is the
optical microscope image of a broken and curled cobalt film after 5 min of conditioning at pH 9.
120-nm-thick, were deposited from 0.05 M CoSO4 + 1 M
(NH4)2SO4 solutions at pH 6 and pH 9 by applying a
constant potential of ꢁ1150 mV for 60 s. The deposited
cobalt was then oxidized in a linear sweep voltammetry
experiment in a positive direction, either immediately after
deposition, or after the film was conditioned in a cobalt-
ammonia solution for some time. The expectation was that
the surface cobalt hydroxide layer would be dissolved by the
ammonia solution, freeing metallic cobalt below for further
reaction. The effect of conditioning time on linear sweep
voltammetry is given in Fig. 7a and b.
3.2.2.1. Effect of anions
The effect of anions on the electrodeposition of cobalt
was examined by running cyclic voltammetry experiments
in two additional ammonium solutions, ammonium carbo-
nate and ammonium chloride. The results (not presented)
were very similar to those with ammonium sulfate.
3.2.2.2. Reaction mechanisms of anodic cobalt oxidation
Based on cyclic voltammetry and linear sweep data,
anodic oxidation occured according to reactions ((4)–(11)):
Linear sweeps at pH 6, Fig. 7a, which started immedi-
ately after the electrode was re-immersed in the electrolyte,
contained both anodic peaks. As the conditioning time
increased, peak Ia diminished, so that after 10 min only
peak IIa was present. At pH 9, Fig. 7b, the anodic reactions
of cobalt were not as strong as at pH 6.0. Peaks Ia and IIa had
much smaller current densities, peak Ia being barely present
even in the first potential sweep.
Anodic peak
Reaction
pH 6
Co þ 2OHꢁ ! CoðOHÞ2 þ 2e
(4)
(5)
(6)
(7)
Ia and IIa
Chemical reaction
CoðOHÞ2 þ ðNH4Þ2SO4
! CoSO4 þ 2NH3 ꢀ H2O
CoSO4 þ 2NH3 ꢀ H2O
Examinations of the electrode after potential sweeps at
pH 6 and 9 revealed that at the end of every sweep at pH 6 the
electrode surface was barren, while after every sweep at pH
9 there was still a film on the electrode. The film left after the
anodic sweep at pH 9 was imaged using a CCD camera, the
inset in Fig. 7b. The deposit observed on the surface of the
electrode was cobalt film covered with the formed layer of
cobalt hydroxide. The image shows breakage and curling of
the film. Dark curls revealed the pristine unoxidized metallic
cobalt that was on a side toward the electrode. These
surfaces were smooth because of their conformation to
the smooth glassy carbon surface, unlike the side of the
film covered with the hydroxide layer.
2þ
! CoðNH3Þ2 þ SO24ꢁ þ 2H2O
Overall chemical
reaction:
CoðOHÞ2 þ ðNH4Þ2SO4
2þ
! CoðNH3Þ2 þ SO24ꢁ þ 2H2O
pH 9
Co þ 2OHꢁ ! CoðOHÞ2 þ 2e
(8)
(9)
Ia and IIa
Chemical reaction
CoðOHÞ2 þ ðNH4Þ2SO4 þ NH3 ꢀ H2O
! CoSO4 þ 3NH3 ꢀ H2O
CoSO4 þ 3NH3 ꢀ H2O
2þ
! CoðNH3Þ3 þ SO24ꢁ þ 3H2O
(10)
(11)
Overall chemical
reaction:
CoðOHÞ2 þ ðNH4Þ2SO4 þ NH3 ꢀ H2O
2þ
! CoðNH3Þ3 þ SO24ꢁ þ 3H2O

D. Grujicic, B. Pesic / Electrochimica Acta 49 (2004) 4719–4732
4725
According to the proposed reaction mechanisms, the first
step at pH 6 was the oxidation of cobalt and formation of
passivating cobalt hydroxide film, reaction (4). This reaction
corresponded to peak Ia. Cobalt hydroxide in the surface
film and ammonium ions from ammonium sulfate reacted to
produce soluble cobalt sulfate and ammonia, reaction (5).
The resulting cobaltous ions and ammonia formed a cobalt-
ammonia complex according to reaction (6). Complexation
of cobaltous ion was the last step in the removal of passivat-
ing hydroxide film, which exposes metallic cobalt to solu-
tion, thus enabling reaction (4) to continue, producing the
second anodic peak, IIa. Not all of the cobalt hydroxide was
removed, resulting in its accumulation in the subsequent
cycles, enhancing the passivation. As a result, the subse-
quent anodic (and cathodic) currents were smaller, as seen in
Fig. 4. The reaction mechanisms at pH 9 were similar to the
mechanism at pH 6, the only difference being the presence
of free ammonia for reactions (9) and (11). At pH 6,
however, the independence from cycling after the first cycle
(see Fig. 4) can be explained by the absence of cobalt
hydroxide film buildup beyond that which was formed in
the first cycle, or by the nature of passive film, which at pH 9
had become porous, thus predefining the available area for
anodic (and cathodic) reactions. The anodic cobalt reactions
definitely require further study toward the determination of
exact mechanisms.
A review of the literature points out that the anodic
branch of cobalt cyclic voltammograms has either one peak
[7,8,10,13] or two peaks [9,11,12], depending on pH, two
peaks being the property of solutions with higher pH. No
explanation has been given for the presence of two anodic
peaks, except the suggestion that the two peaks could be due
to dissolution of cobalt as different ionic species and dif-
ferent cobalt phases [9], or that they represent different
allotropic forms of cobalt [11].
3.3. Chronoamperometry
Chronoamperometry was used as a supporting experi-
mental technique to further elucidate the nucleation mec-
hanisms of cobalt on glassy carbon. By stepping up the
potential from open circuit to a potential at which reduction
of cobaltous ions proceeds under diffusion control, the
cobalt nuclei formed will each have a diffusion zone that
controls further supply of cobaltous ions. The overlap of
these zones results in a peak that is characteristic for
nucleation. After the peak, the current is limited by the
mass transfer of cobaltous ion to the electrode surface, as
described by the Cottrell equation [17]. Within the diffusion
zone, the growth of already established cobalt nuclei can
continue, or additional nucleation can be initiated on other
sites.
Fig. 8. (a–f) Effect of deposition potential on chronoamperometry of cobalt in pH 6 and 9 solutions. The study was performed for three different cobalt
concentrations, 0.005, 0.01 and 0.05 M, all with 1 M (NH4)2SO4 as a supporting electrolyte.

4726
D. Grujicic, B. Pesic / Electrochimica Acta 49 (2004) 4719–4732
Fig. 9. (a–h) Effect of deposition potential on morphology of deposited cobalt at pH 6 and 9. Conditions: 0.005 M Co2+, 1 M (NH4)2SO4.
The parameters of interest that were varied in chronoam-
perometric experiments were deposition potential, concen-
tration of cobaltous ions, solution pH and temperature.
was measured from AFM images by using the bearing
routine in the NanoscopeTM software. When total volume
of scanned nuclei was multiplied by the density of cobalt
(8.92 g cmꢁ3) the weight of deposited cobalt, per scanned
area, was obtained. The electrode area divided by scanned
area was used as a multiplication factor to determine the
total weight of deposited cobalt. This result was then
compared with the weight of cobalt calculated by cathodic
current integration and Faraday’s law, Fig. 11.
3.3.1. Effect of deposition potential
The deposition potentials studied, ꢁ1150, ꢁ1200, ꢁ1250
and ꢁ1300 mV, were selected from the more negative side
of an Ic peak in Fig. 3a–f. The results for the effect of
deposition potential for three studied cobalt concentrations
at pH 6 and 9 are summarized in Fig. 8a–f.
According to Fig. 11, the amount of deposited cobalt
measured by AFM was very close to the calculated amount
for all deposition potentials at pH 9 (open and closed circles
in Fig. 11). At pH 6, there was a significant discrepancy as a
function of deposition potential between measured and
calculated amount of deposited cobalt. While the calculated
Peaks corresponding to the overlap of diffusion zones are
clearly seen in most of the cas, except for the conditions at
ꢁ1150 mV in 0.005 M and 0.01 M Co2+ pH 6 solutions, for
which the durationof thecawas too short forpeakexpression.
RegardingthepHeffect, theimportantdifference is(a) slower
diffusion current decay, and (b) smaller peak currents at pH 6
than at pH 9. Slower decay of diffusion current was the result
of the significant contribution of hydrogen reduction current
to the overall current. Smaller peak currents were the result of
the shielding effect of hydrogen evolved on the surface of
cobalt, reducing the nuclei area available for continuation of
cobalt deposition. Due to hydrogen shielding, reduction of
cobalt ions to ad-atoms [25] had to take place elsewhere on
the electrode surface, resulting in new nucleation sites. The
net effect was a higher nuclei population density and smaller
nuclei at pH 6 than at pH 9, as verified by the AFM images in
Fig. 9a–h, and quantified in Fig. 10.
One way to estimate the contribution of the hydrogen ion
reduction current to the overall current during a ca experi-
ment would be to compare the cathodic (cobalt deposition)
with anodic (cobalt dissolution) charge. The difference
would represent the current due to hydrogen ion reduction.
Because cobalt hydroxide film formation could interfere
with the total anodic current, another method was chosen
instead. Accordingly, the quantity of electrodeposited cobalt
Fig. 10. Effect of deposition potential on cobalt nuclei population density
for two solutions, pH 6 and 9. Counting of nuclei was performed on images
in Fig. 9a–h.

D. Grujicic, B. Pesic / Electrochimica Acta 49 (2004) 4719–4732
4727
Fig. 11. Effect of cathodic potential on the mass of electrodeposited cobalt.
The weight of cobalt, calculated from the total volume of cobalt nuclei on an
AFM image, is compared to the weight of cobalt calculated from the
corresponding cathodic charge.
amount of cobalt increased almost linearly with the more
negative potentials (closed squares), the corresponding mea-
sured amount of cobalt was significantly lower (open
squares). The difference represents the contribution of
hydrogen current during electrodeposition of cobalt, which
can be as high as 45%, at ꢁ1300 mV. At pH 9, the fraction of
current spent on hydrogen reduction was between 5 and
10%.
Fig. 12. (a0–c00) Schematic representation of the physical role of hydrogen
gas bubbles on the final destination of cobalt nuclei, for two different
deposition potentials (a0, b0, c0) ꢁ1150 mV, and (a00, b00, c00) ꢁ1250 mV. The
number of nuclei and their size after 1 second (a0 and a00) was compared with
the number and size of nuclei on the electrode surface after 10 s of
electrodeposition. Surface movement of cobalt nuclei was influenced by
the size (a function of deposition potential) of formed and affixed hydrogen
gas bubbles, b0 and b00.
It should be expected that at the more negative deposition
potentials the electrode surface would be characterized by
higher nuclei population density. However, according to Fig.
10, the nuclei population density decreased as the potential
became more negative, especially for pH 6 conditions. The
decrease of nuclei population density at more negative
deposition potentials has also been observed in the study
of copper deposition from ammoniacal solutions [26].
To explain how the nucleation population density is
dependent on the deposition potential, the effect of hydrogen
ion reduction is described via experimental results and
schematic illustrations shown in Fig. 12, in which the left
and right columns represent electrodeposition at ꢁ1150 and
ꢁ1250 mV, respectively, from solutions at pH 6.
ing the existing nuclei to grow, and new ones to be formed.
At ꢁ1250 mV, Fig. 12b00, cobalt nuclei also continued to
grow, but the extent of hydrogen ion reduction was much
larger than at ꢁ1150 mV, resulting in larger hydrogen gas
bubbles. The hydrogen bubbles affixed to cobalt nuclei can
produce two effects, either physically pushing a nucleus
away from its original position until it collides with neigh-
boring nucleus, or providing buoyancy to a nucleus so it can
move more easily across the surface. These two physical
effects would be more pronounced with an increasing
volume of hydrogen gas bubbles, at prolonged depositions.
The AFM results after 10 s of deposition at ꢁ1250 mV, Fig.
12c00, tended to confirm this postulate. Accordingly, the
number of cobalt nuclei decreased to 58 mmꢁ2, from
144 mmꢁ2 (after 1 s). Correspondingly, at ꢁ1150 mV, the
hydrogen bubbles were too small to produce the described
effects, allowing the growth of existing nuclei and the
formation of new ones. Thus, after 10 s, the nuclei popula-
tion density increased from 98 to 140 mmꢁ2, Fig. 12c0.
A decrease in nuclei population with time has also been
observed by Gomez et al. [8], who suggested that surface
diffusion of the smaller toward the larger nuclei increased
the resulting nuclei size and decreased the overall population
At ꢁ1150 mV deposition for 1 s, Fig. 12a0, the glassy
carbon surface was covered with very small cobalt nuclei,
with diameters 20 nm and less. The nuclei population
density measured for this image was 98 mmꢁ2. In the
corresponding image for deposition potential ꢁ1250 mV,
Fig. 12a00, the nuclei population density was 144 mmꢁ2. The
vertical scale of these two AFM images was 20 nm, so that
cobalt nuclei could be better accented.
Fig. 12b0, electrodeposition at ꢁ1150 mV, is a schematic
representation of the electrode surface covered with cobalt
nuclei (dark circles) and also hosting the hydrogen gas
bubbles produced (light circles). Under this deposition
potential, hydrogen gas bubbles were still very small, allow-

4728
D. Grujicic, B. Pesic / Electrochimica Acta 49 (2004) 4719–4732
density. They provided no explanation for the driving force
of diffusion.
more effective. The effect of micromagnetic forces on the
surface diffusion of cobalt atoms and nuclei warrants a
separate study.
Although our study provides evidence that cobalt popu-
lation density is a function of hydrogen ion reduction, this is
not the exclusive mechanism for integration of small cobalt
nuclei. Being ferromagnetic, cobalt nuclei can travel across
the surface under the influence of micromagnetic forces
between neighboring nuclei. Again, the buoyancy provided
to the nuclei by hydrogen gas bubbles can make such forces
3.3.2. Effect of cobalt concentration
The effect of cobalt concentration on the morphology of
cobalt nuclei was studied at pH 6 and 9, by using a deposi-
tion voltage of ꢁ1250 mVand a deposition time of 10 s. The
results are presented in Fig. 13a–f. In general, the increase of
Fig. 13. (a–f) Effect of cobalt concentration on the morphology of cobalt electrodeposited at ꢁ1250 mV from solutions at pH 6 and 9.

D. Grujicic, B. Pesic / Electrochimica Acta 49 (2004) 4719–4732
4729
cobalt concentration increased the surface coverage of
cobalt nuclei. Full coverage was achieved in 0.05 M Co2+
solutions for both pH levels.
where im is the current and tm is the time coordinate of a
chronoamperometric peak. In the following graphs, the
instantaneous and the progressive mechanisms will be repre-
sented by solid and dashed lines, respectively.
3.3.3. Effect of temperature
The experimental data from Fig. 8 were used to obtain
reduced currents versus reduced time plots, according to
Eqs. (12) and (13). The results given in Fig. 14a–f represent
the effect of pH, cobalt concentration, and deposition poten-
tial. The inspection of graphs in Fig. 14 reveals that in the
initial stages cobalt nucleation occurred according to the
progressive nucleation mechanisms, regardless of cobalt
concentration, pH and deposition potential.
The effect of temperature on electrodeposition of cobalt
was studied in the temperature range of 5–55 8C. The
experimental conditions were 0.01 M Co2+ concentration,
deposition potential ꢁ1250 mV, solution pH 6 and 9. No
significant effect was found for either pH.
3.3.4. Nucleation models
Scharifker and Hills [27] used current transients to model
nucleation according to both instantaneous and progressive
mechanisms. Instantaneous nucleation corresponds to a slow
growth of nuclei on a small number of active sites, all
activated in the same time. Progressive nucleation corre-
sponds to fast growth of nuclei on many active sites, all
activated during the course of electroreduction [28].
The models for instantaneous and progressive nucleation
are given by Eqs. (12) and (13), respectively,
At pH 6 for 0.005 M and 0.01 M Co2+ solutions, Fig.
14a–b, the reduced current plots followed the progressive
nucleation model (dashed line) from the beginning to the
maximum on the curve. After the maximum the reduced
current plots started to deviate from both instantaneous and
progressive model plots. It was already stated that in 0.005
and 0.01 M Co2+ solutions the contribution from the hydro-
gen ion reduction to the total current is not negligible and
cobalt nuclei serve as suitable sites for hydrogen ion reduc-
tion. While the total surface area of cobalt nuclei is still
relatively small (before the maximum on the reduced current
plot) the contribution from the hydrogen reduction will be
small. However, once the surface area of cobalt becomes
larger, the contribution from the hydrogen ion reduction to
the overall current will become significant, causing the
deviation observed in Fig. 14a and b.
ꢄ
ꢂ
ꢀ
ꢁꢃꢅ
2
i2 1:9542
t
¼
1ꢁexp ꢁ1:2564
(12)
(13)
im2
t=tm
tm
(
"
#)
2
ꢀ
ꢁ
2
i2 1:2254
t
¼
1ꢁexp ꢁ2:3367
im2
t=tm
tm
Fig. 14. (a–f) Reduced current versus reduced time plots, generated by using the data in Fig. 10a–f, represent the effect of deposition potential, cobalt
concentration and pH on the mechanism of cobalt nucleation. Theoretical models for instantaneous and progressive nucleation mechanisms are shown as solid
and dashed lines, respectively.

4730
D. Grujicic, B. Pesic / Electrochimica Acta 49 (2004) 4719–4732
Fig. 15. (A–C) Schematic representation of cobalt nuclei morphology produced according to the orientation of hexagonal structures toward the electrode
surface. The planes for easiest incorporation of cobalt ad-atoms are also identified. The last column outlines the resulting 2D shapes when the morphology is
profiled from the top, as by AFM scanning.
Based on Scharifker–Hills equations, nucleation of cobalt
on glassy carbon follows progressive mechanisms, meaning
that new nuclei are formed during the course of an experi-
ment. In that case, the surface of glassy carbon should be
covered with cobalt nuclei of various sizes. However, the
AFM images in Fig. 9a–h, which represent the status of the
electrode surface at the end of chronoamperograms used for
modeling, show otherwise; that is, approximately the same
size of nuclei. The AFM examination clearly suggests
instantaneous nucleation mechanisms. An explanation for
the obvious contradiction between the theoretical models
and experimental findings is provided below.
Fig. 15, labeled as top view, schematically demonstrates the
resulting morphology for each nucleation type when profil-
ing from the top, as done in atomic force microscopy.
Each of the proposed nuclei types is identified in an actual
1.5 mm ꢂ 1.5 mm AFM image in Fig. 16a. Because the third
dimension, the height of nuclei, is not obvious from a planar
AFM image, a separate analytical routine was used for its
estimation by sectioning across each type of nuclei, as
indicated by black arrows and a corresponding letter.
Cross-sectioning lines are numerically labeled 1-10 for
type A, 2-20 for type B and 3-30 for type C nuclei. The
measured height along the cross-sectioning lines is given in
Fig. 16b. It was found that the nucleus parallel to the surface,
the A-type, was 14.5 nm high. The B-type nucleus was
inclined 458 with a height of 56 nm. The C-type nucleus,
75 nm high, was perpendicular to the electrode surface.
The growth rate of each nucleus will be determined by the
area of less densely packed planes available for cobalt
incorporation. According to Fig. 13c, the C-type nuclei will
prevail at the end of electrodeposition. Obviously, during the
growth of nuclei there will be a distribution of nuclei type,
with consequent distribution of (a) planes and (b) surface
area available for deposition. In the Scharifker and Hills
model [27], which was derived for hemispherical geometry,
the growth of nuclei is controlled by the growing surface
area of idealized hemispheres. For the reasons discussed
above, the nucleation of cobalt is characterized by a surface
area larger than that which would be provided by the same
size hemispheres. A larger surface area for a fixed number of
nuclei has the same effect as if the number of nuclei
increased during electrodeposition. In other words, the
nucleation mechanisms should follow the Scharifker–Hills
model for progressive nucleation, which was indeed shown
in Fig. 14.
A closer examination of AFM images revealed three
types of cobalt nuclei, schematically presented in Fig. 15.
The three types differ by their crystallographic orienta-
tion on the surface; more precisely, by the angle between the
hexagonal plane and the electrode surface. The A-type
nuclei are flat hexagonal nuclei with (0 0 0 2) basal plane,
B-type are pyramidal nuclei with (1 0 1 1) basal plane, and
C-type are nuclei with ð1 1 2 0Þ or ð1 0 1 0Þ basal planes. The
hexagonal plane, or its part, represents the plane of closest
packing (0 0 0 2), which consequently is the plane with the
least favorable conditions for incorporation of cobalt atoms.
Favorable planes for incorporation of atoms are schemati-
cally indicated by the arrows around each nucleus type. In
addition, the A-type nuclei have a small available surface for
incorporation of atoms, resulting in slower nuclei growth.
The surface available for incorporation of atoms is the
largest in type-B nuclei. The C-type nuclei have the smallest
area, but their protrusion toward the bulk of the solution
improves the conditions for reduction and incorporation of
cobalt atoms. Cobalt nuclei of C-type cannot be found as a
single nucleus, but always as several nuclei stacked along
their closely packed sides. The last column on the right of

D. Grujicic, B. Pesic / Electrochimica Acta 49 (2004) 4719–4732
4731
Fig. 16. (a and b) The identification and profiling of each type of cobalt nuclei depicted in Fig. 15A–C. On the AFM image, lines 1-10, 2-20 and 3-30 have the
objective to intersect A-, B- and C-type of cobalt nuclei, respectively. The physical measurements of the third dimension, the height of nuclei, are given in (b).
The heights of A-, B- and C-type nuclei, marked by the black arrows, are 14.5, 56 and 75 nm, respectively. The scale of AFM image is 1.5 mm ꢂ 1.5 mm.
The need to verify, through additional experimental
methods, the heterogeneous reaction mechanisms deter-
mined by mathematical models, is demonstrated by this
study, which clearly shows that although the experimental
transients fit the progressive nucleation model, cobalt
nucleation is actually instantaneous.
5. Hydrogen reduction, which takes place on the cobalt
nuclei, and not on the bare glassy carbon surface, plays an
important role during the growth of nuclei. Hydrogen gas
bubbles enable cobalt ad-atoms to move across the sur-
face toward growing nuclei. At higher cathodic poten-
tials, larger hydrogen gas bubbles contribute more to
coarsening of cobalt nuclei than the smaller hydrogen gas
bubbles produced at lower cathodic potentials.
4. Conclusions
6. The Scharifker–Hills models suggested progressive
cobalt nucleation mechanisms; i.e., the production of
nuclei of variable size. However, the AFM study indi-
cated instantaneous nucleation; i.e., production of nuclei
of the same size. The discrepancy between the mechan-
isms can be explained by the presence of three different
types of cobalt nuclei, each with its own rate of growth
and reaction surface area.
Electrochemical deposition of cobalt on glassy carbon
from ammonium sulfate solutions was studied by cyclic
voltammetry and chronoamperometry. The corresponding
electrode morphology was examined by atomic force micro-
scopy. The effect of each parameter was studied at pH 6 and
9:
1. The cyclic voltammograms were characterized by two
cathodic and two anodic peaks. The first cathodic
peak (scanning in the negative direction) represents
the reduction of cobaltous ion to metallic cobalt. The
second cathodic peak is due to a hydrogen evolution
reaction.
Acknowledgement
This work has been sponsored by the US Department of
Defense, Department of the Navy, Office of Naval Research,
Grant Number N000140110829, and The National Science
Foundation RII Award (EPS 0132626).
2. The first anodic peak (scanning on return in the positive
direction) is a result of cobalt metal oxidation to cobalt
hydroxide. Dissolution of cobalt hydroxide with ammo-
nia results in resumption of cobalt oxidation underneath
the removed film.
References
[1] L. Lopez-Diaz, M. Klaui, J. Rothman, J.A.C. Bland, J. Mag. Mag.
Mater. 242–245 (2002) 553.
3. An increase of cv scanning rate resulted in an increase of
nuclei population density at pH 6, but not at pH 9.
4. An increase of cobalt concentration in solutions reduced
the concentration overpotential for cobalt deposition,
shifting it to a more positive potential.
[2] M. Klaui, C.A.F. Vaz, J.A.C. Bland, W. Wernsdorfer, G. Faini, E.
Cambril, Appl. Phys. Lett. 81 (1) (2002) 108.
[3] J. Ebothe, S. Vilain, J. Phys. D: Appl. Phys. 32 (1999) 2342.
[4] E. Gomez, E. Valles, J. Appl. Electrochem. 32 (2002) 693.

4732
D. Grujicic, B. Pesic / Electrochimica Acta 49 (2004) 4719–4732
[5] J.M. Maire, G. Pupon, J. Pagetti, M.P. Gigandet, in: L.T. Romankiw, S.
Krongelb, C.H. Ahn (Eds.), Electrochemical Society Proceedings, PV
98-20, Pennington, NH, 1998, p. 136.
[16] A. Cotton, G. Wilkinson, Advanced Inorganic Chemistry, third ed.,
Wiley, New York, 1972.
[17] Southampton Electrochemistry Group, in: T.J. Kemp (Ed.), Instru-
mental Methods in Electrochemistry, Chichester, UK, 1985.
[18] K. Kinoshita, Carbon, Electrochemical and Physicochemcial
Properties, Wiley, New York, NY, 1988.
[6] I. Tabakovic, S. Riemer, P. Jallen, V. Inturi, A. Thayer, in: L.T.
Romankiw, S. Krongelb, C.H. Ahn (Eds.), Electrochemical Society
Proceedings, PV 98-20, Pennington, NH, 1998, p. 531.
[7] S. Floate, M. Hyde, R.G. Compton, J. Electroanal. Chem. 523 (2002)49.
[8] E. Gomez, M. Marin, F. Sanz, E. Valles, J. Electroanal. Chem. 422
(1997) 139.
[19] E. Gileadi, Electrode Kinetics for Chemists, Chemical Engineers, and
Materials Scientists, VCH Publishers, New York, NY, 1993.
[20] T.R. Jayaraman, V.K. Venkatesan, H.V.K. Udupa, Electrochim. Acta
20 (1974) 209.
[9] A.B. Soto, E.M. Arce, M. Palomar-Pardave, I. Gonzales, Electrochim.
Acta 41 (16) (1996) 2647.
[21] N. Sato, T. Ohtsuka, J. Electrochem. Soc. 125 (11) (1978) 1735.
[22] M. Novoselsky, N.R. Menglisheva, Electrochim. Acta 29 (1) (1984)
21.
[10] N. Myung, K.H. Ryu, P.T.A. Sumodjo, K. Nobe, Electrochemical
Society Proceedings, PV 97-27, p. 136.
[11] M. Palomar-Pardave, I. Gonzales, A.B. Soto, E.M. Arce, J. Electro-
anal. Chem. 443 (1998) 125.
[23] W.A. Badawy, F.M. Al-Kharifi, J.R. Al-Ajmi, J. Appl. Electrochem. 30
(2000) 693.
[12] L.H. Mendoza-Huizar, J. Robles, M. Palomar-Pardave, J. Electroanal.
Chem. 521 (2002) 95.
[24] M. Pourbaix, Atlas of Electrochemical Equilibria in Aqueous
Solutions, second ed., NACE, Houston, TX, 1974 (p. 328).
[25] K.J. Vetter, Electrochemical Kinetics Theoretical and Experimental
Aspects, Academic Press, New York, NY, 1967.
[13] C.Q. Cui, S.P. Jiang, A.C.C. Tseung, J. Electrochem. Soc. 137 (11)
(1990) 3418.
[14] H. Nakano, K. Nakahara, S. Kawano, S. Oue, T. Akiyama, H.
Fukushima, J. Appl. Electrochem. 32 (2002) 43.
[26] D. Grujicic, B. Pesic, in press.
[27] B. Scharifker, G. Hills, Electrochim. Acta 28 (7) (1983) 879.
[28] M.P. Pardave, M.T. Ramirez, I. Gonzales, A. Serruya, B.R. Scharifker,
J. Electrochem. Soc. 143 (5) (1996) 1551.
[15] H.-H. Huang, STABCAL—stability calculation for aqueous systems,
Montana Tech, The University of Montana, Bute, MT.