You are on page 1of 14

Journal of Solid State Electrochemistry (2020) 24:1291–1304

https://doi.org/10.1007/s10008-020-04613-2

ORIGINAL PAPER

Mechanistic analysis of anodic dissolution of cobalt in alkaline


glycine solution
Twinkle Paul 1 & Ramanathan Srinivasan 1

Received: 14 December 2019 / Revised: 17 April 2020 / Accepted: 21 April 2020 / Published online: 11 May 2020
# Springer-Verlag GmbH Germany, part of Springer Nature 2020

Abstract
The kinetics of dissolution of Co in alkaline glycine solution under anodic conditions was investigated. Potentiodynamic
polarization and electrochemical impedance spectroscopic (EIS) data of Co dissolving in 0.1 M glycine solution at pH 10 were
acquired. At low overpotentials, in the active region, the polarization current increases with potential, whereas at high
overpotentials, in the passive region, it decreases with potential. Co surface exposed to the alkaline glycine solution was
characterized using scanning electron microscopy and x-ray photoelectron spectroscopy, and based on the binding energies of
the deconvoluted peaks, oxides, hydroxides, and Co-glycine complexes were identified as the likely species present on the
surface. Impedance data was subjected to electrical equivalent circuit analysis, and a circuit with three Maxwell elements was
found to model the faradaic impedance. Polarization and EIS data were also analyzed in the framework of mechanistic analysis,
and a mechanism with four adsorbed intermediates is proposed. The proposed model captures the salient features of the
polarization and EIS results.

Keywords Cobalt anodic dissolution . Complexing agent glycine . Chemical mechanical planarization . Potentiodynamic
polarization . Electrochemical impedance spectroscopy . Reaction mechanism

Introduction industry, Co CMP has been extensively studied, with slurries


containing various organic chemicals such as glycine [1, 3, 7],
Due to its low resistivity as well as super adhesive and electro- citric acid, oxalic acid, tartaric acid [8], lysine, proline, argi-
migration resistance properties, Co has emerged as a promis- nine [9], acetate [10], and an EDTA derivative [11], while
ing replacement for adhesion or barrier layer for Cu metalli- ECMP of Co in KOH [12] and H2O2-based systems [13] has
zation and Cu interconnect metal in integrated circuits [1–5]. been reported.
Chemical mechanical planarization (CMP) of Co is one of the Glycine is a simple amino acid which can complex with Co
important processes in the semiconductor industry, and it uti- [1]. Co CMP with glycine is usually conducted in alkaline
lizes synergic effects of chemical dissolution along with me- environment [3, 9, 10, 14, 15] although acidic slurries have
chanical abrasion in a slurry to remove excess Co. also been evaluated [1]. Lu et al. [1] evaluated Co CMP in
Electrochemical mechanical polishing (ECMP) is a process glycine-based slurries at pH 3 and proposed that Co2+ ions are
in which the removal rate is enhanced by applying anodic chelated by glycine. Jiang et al. [3] studied Co CMP in slurries
potential to the metal, and ECMP offers better lower with H2O2 and 0.1 M glycine at pH 8 and suggested that
defectivity [6]. By virtue of its importance in semiconductor glycine forms a water-soluble Co(III) glycine complex during
CMP. He et al. [16] evaluated effect of inhibitors in glycine
solution for Co CMP application, at pH 8.5. Based on electro-
Electronic supplementary material The online version of this article
(https://doi.org/10.1007/s10008-020-04613-2) contains supplementary chemical quartz crystal microbalance studies, they reported
material, which is available to authorized users. that glycine enhances Co dissolution and the dissolution rate
can be reduced by 1,2,4, triazole. In addition to Co CMP
* Ramanathan Srinivasan applications, glycine has been employed as a complexing
srinivar@iitm.ac.in agent in post-CMP cleaning of Co [2] and in Co electrodepo-
1
sition [17]. While ECMP of Co in KOH [12] and H2O2-based
Department of Chemical Engineering, Indian Institute of
Technology-Madras, Chennai 600036, India
systems [13] has been evaluated, ECMP of Co in glycine has
1292 J Solid State Electrochem (2020) 24:1291–1304

not been reported. A clear understanding of anodic dissolution electrodes such as Ag/AgCl or saturated calomel electrode
of Co in glycine can help in development of Co ECMP pro- can lead to chloride contamination of the solution, which
cess in glycine-based systems. can lead to pitting corrosion of Co [23] and hence SSE was
Detailed mechanistic studies of Co anodic dissolution in employed as reference electrode in this work. The electrolyte
alkaline environment has been reported in literature, but main- contained 0.1 M glycine, with 1 M NaClO4 (99.9%, Loba
ly in bicarbonate/carbonate solutions [18–20]. Real et al. stud- Chem) as supporting electrolyte. Analytical grade chemicals
ied Co dissolution in carbonate/bicarbonate solution in the pH from Merck were used in Milli-Q water of 18.2 MΩ cm resis-
range of 8.9–10.5, using polarization and EIS techniques, and tivity. The solution pH (10) was adjusted using 1 M NaOH.
proposed a mechanism involving Co(OH)ad as an intermedi- Co electrode surface was polished using grit paper of increas-
ate species and [Co(CO3)2]2−as the soluble product. Calderon ing fineness and final abrasion was performed with 3000 grit
et al. [19, 20] investigated Co electrodissolution in carbonate/ paper and 0.3 μm alumina powder. Then, the electrode was
bicarbonate media and proposed a reaction model involving ultrasonicated in ethanol and rinsed in Milli-Q water. Before
Co(HCO3)ads and CoO as intermediates, with [Co(CO3)2]2−as each experiment, N2 was bubbled in the electrolyte for 10 min,
the soluble product. Gallant et al. [21] analyzed cyclic volt- and the electrode was polarized at negative potential for 1 min
ammogram of Co in bicarbonate/carbonate solution and to reduce the native oxide.
employed a deconvolution algorithm to assign the observed PARSTAT 2263 (AMETEK, USA) electrochemical work-
peaks to various oxidation reactions. Based on the analysis, station was used to conduct the electrochemical experiments.
formation of Co(CO3) and its subsequent dissolution as The working electrode was rotated at 900 rpm using Pine
[Co(CO3)2]2−was proposed as the reaction corresponding to Instruments MSR since rotations at higher rpm yielded essen-
the first peak; formation of CoO and its subsequent dissolution tially the same results. Initially, the open circuit potential
as Co2+ was assigned to the second peak, and the oxidation of (OCP) of Co in the electrolyte was monitored as a function
CoO to Co3O4 was assigned to the last peak. Ismail and of time and all the other experiments were performed only
Badawy reported polarization, impedance, and XPS studies after stabilization of OCP. Polarization experiments were per-
of Co in KOH solution and proposed that surface would be formed by starting at OCP and scanning the potential at
passivated by a film containing oxides and hydroxides [22]. 1 mV s−1 in the anodic direction. EIS experiments were per-
According to their analysis, the inner layer of the film would formed by superimposing an ac perturbation of 10 mV ampli-
contain a mixture of Co(OH)2 and CoO, while the outer layer tude on a given dc bias. The perturbation frequency was var-
would be made of Co(III) in the form of CoO(OH). Although ied from 20 kHz to 0.1 Hz at 7 frequencies per decade, which
mechanistic analysis of Co in bicarbonate/carbonate alkaline were logarithmically spaced. At lower frequencies, the data
solutions and in KOH is available in literature, to the best of was noisy and not reproducible, and hence are not shown. All
our knowledge, there is no published report of detailed mech- the EIS data presented in this work were validated using linear
anism of Co dissolution in glycine solution, and in particular KKT software [24] .
in alkaline conditions. The surface morphology of the electrode was characterized
In this work, we investigate Co dissolution in alkaline so- using SEM (Hitachi S4800) and the surface oxidation state
lution in presence of glycine, using potentiodynamic polariza- was investigated using PHI Versaprobe III using Al Kα x-
tion and electrochemical impedance spectroscopy (EIS). The ray source at 1486.7 eV. The data was corrected using the
surface was characterized using scanning electron microscopy value of C1s peak at a binding energy of 284.8 eV and
(SEM) and x-ray photoelectron spectroscopy (XPS). A multi- Shirley approximation was used for background removal.
step reaction mechanism with four adsorbed intermediates XPS Peak 4.1 software (R.W.M. Kwok) was used to de-
was proposed. Based on the proposed model, polarization convolute the XPS peaks. Potential-pH diagram of Co in gly-
and EIS data were simulated to match with the experimental cine was prepared using the freely available software
results. From the model and best fit kinetic parameters, the Medusa® [17, 25, 26]. To generate the potential-pH diagram,
fractional surface coverage of the intermediate species at an- the concentration of Co2+ species in solution was fixed at
odic potential was estimated. 10−5 M and the concentration of total glycine was fixed at
0.1 M. A database containing various complexes was gener-
ated using the Hydra® component of the software, and from
Experimental the database, the potential pH diagram was generated using
the Medusa interface. Medusa software has been used in lit-
A standard three electrode cell, with a 5 mm diameter Co as erature to create potential-pH diagrams of Cu in phosphate-
working electrode (Alfa Aesar, 99.95% purity), saturated sul- buffered saline [27], to generate chemical equilibrium species
fate electrode (Hg/Hg2SO4 with saturated K2SO4, SSE) as of Co in glycine solution [28], Cu-NH3-citrate bath [29], and
reference electrode and Pt mesh (Alfa Aesar, 99.99% purity) Ti in alkaline solution [26] and to identify passive region in Zn
as counter electrode was used. Commonly used reference in alkaline solutions [30].
J Solid State Electrochem (2020) 24:1291–1304 1293

Results and discussion

Polarization results

The OCP of Co in 0.1 M glycine solution at pH 10, as a


function of time is shown in Fig. 1. It is clear that the OCP
is very stable and is − 1054 mV vs. SSE. All the experiments
were conducted after the working electrode was held at OCP
for 5 min. Figure 2 shows the potentiodynamic polarization of
Co in glycine solution. The experimental results, shown as
continuous line, were reproducible within 2%. Mechanistic
analysis predictions, explained later, are shown as dashed
lines. The experimental results show that the current increases
with potential initially, and after reaching a maximum of ~
5.2 mA cm−2 at − 844 mV vs. SSE, the current decreases with
potential. The region where current increases with potential is
the active dissolution region, and the region where it decreases
with potential can be considered as passive region. It is to be
noted that the term passive is not used to indicate that the
Fig. 2 Potentiodynamic polarization curve of Co, rotated at 900 rpm, in
surface is protected from dissolution; here, it only implies that
0.1 M glycine solution at pH 10, with 1 M NaClO4 as supporting
the current decreases with potential in this region, while in electrolyte. The potential of the working electrode was scanned at
fact, dissolution continues to occur. In bicarbonate/carbonate 1 mV s−1 . Experimental data are shown as continuous line and
solutions at pH 8.9, Co is reported to exhibit a similar trend of mechanistic model results are shown as dashed line. The dc potentials,
where EIS data were acquired, are marked as open diamonds
initial increase in current followed by a decrease in current
with potential, due to a passivation process [19].
Beyond − 660 mV vs. SSE, the current increases with po- dissolution requires modeling of the 3D film, along with the
tential, and in this region, transpassive dissolution occurs. In movement of the vacancies and ions through the film [33].
this region, the surface would normally be covered by a three- The investigations in this paper are limited to analyze the
dimensional film, and metal dissolution would occur by the processes in the potential region before transpassive dissolu-
movement of ions through the film, under a large electric field tion occurs.
[31]. Electrochemical machining of Co is in fact conducted in
transpassive dissolution region [32]. Analysis of trans passive
Potential pH diagram

Glycine is known to form complexes with Co in acidic neutral


and alkaline conditions and enhance the dissolution rate [1,
17, 34]. Potential pH diagrams, also known as Pourbaix dia-
grams, are valuable in identifying solubility and stability of
metal species in the solution of interest [35]. The potential pH
diagrams of Co in water [36] and ammonia [32] solutions are
available in literature. The potential pH diagram for Co in
glycine solution was constructed using Medusa® software
and is presented in Fig. 3. The concentration of glycine used
was 0.1 M and the activity of Co2+ was chosen to be 10−5. In
water at alkaline pH, Co would be passivated by oxides or
hydroxides [36], but glycine reduces the stability region of
these passivating films (Fig. 3), and soluble complexes can
be formed at pH 10. Specifically, [Co(gly)3]− complex is the
likely form, in the potential region where the EIS data were
acquired (open diamonds in Fig. 3). The potential pH diagram
facilitates identification of the most stable forms of the com-
Fig. 1 Open circuit potential vs. time of Co held in solution with 0.1 M plexes, although the detailed mechanism of formation and
glycine and 1 M NaClO4 supporting electrolyte at pH 10 kinetics cannot be obtained from it.
1294 J Solid State Electrochem (2020) 24:1291–1304

(a)

(b)

Fig. 3 Potential-pH diagram of Co-glycine-water system, for a dissolved


Co activity of 10−5 M and glycinate species activity of 0.1 M. The poten-
tials at which EIS data (shown in Fig. 6) were acquired are marked as
open diamonds

Surface characterization

Figure 4a shows the SEM image of Co dissolving in 0.1 M


glycine solution in the active region (held at − 914 mV vs.
SSE for 30 min). The surface appears rough, due to enhanced
dissolution of the metal in the active region. On the other
Fig. 4 Scanning electron micrographs of Co held in a solution with 0.1 M
hand, the SEM image of Co dissolving in the same solution glycine and 1 M NaClO4 for 30 min in a active region (− 914 mV vs.
in the passive region (− 704 mV vs. SSE) exhibits relatively SSE), b passive region (− 704 mV vs. SSE)
less roughness, as seen in Fig. 4b. This is likely because the
surface is protected by a film in the passive region. The sample
surface was also characterized using XPS and the results of O 779.5 eV [40], and 779.6 eV [41, 42]. In addition, organic
1s and Co 2p are presented in Fig. 5. Markers are the exper- complexes of Co also show peak in this binding energy range.
imental results, the continuous lines are the model fits, and the To the best of our knowledge, there is no report on XPS peaks
colored areas are the individual, de-convoluted peaks. XPS is of Co-glycine complex. Escard et al. analyzed the XPS peaks
an ex situ technique, and the sample may be oxidized during of certain organometallic compounds and reported that
transfer from the electrochemical cell to the XPS chamber. [(CH3)4N]4Co(NCSe)6 exhibits a peak at 779.0 eV [43].
Since the sample was held under passivating conditions, the Barber et al. analyzed (π C 2 H 5 ) 2 Co and the peak at
sample would be covered with a passive film and further sur- 779.3 eV was assigned to Co 2p3/2 [44]. It is possible that
face modifications are less likely. In Fig. 5a, between a bind- the peak observed at 779.1 eV could arise from Co-glycine
ing energy of 775 eV and 790 eV, four peaks at 778.3, 779.1, complex or Co3O4 or from both, since if the peaks from Co-
780.7, and 786.7 eV corresponding to Co 2p3/2 levels were glycine complex and Co3O4 are very close, then they cannot
resolved. The first peak can be attributed to metallic Co be separated easily.
(2p3/2), and the last peak to Co2+ (2p3/2Sat), while the two at Haber and Ungier analyzed Co, CoO, and Co3O4 and
779.1 and 780.7 eV could arise from Co2+ or Co3+ (2p3/2) or a assigned the peak at 780.4 eV to Co2+ 2p3/2 in CoO [39].
mixture of them. Another study reported that Co2+ adsorbed on α-FeOOH at
The XPS peak of metallic Co 2p3/2 is reported to occur at pH 10 exhibited a peak at 780.6 eV, while Co(OH)2 exhibited
778.3 eV in one study [37] and at 778.5 eV in another study a peak at 781 eV [45]. The peak observed at 780.7 eV in this
[38], and thus, the peak observed at 778.3 eV in this report can study can be due to cobalt oxides or hydroxides. Literature
be assigned to metallic Co 2p3/2. The binding energy of Co indicates that a satellite peak at 786.4 eV was observed for
2p3/2 in Co3O4 was reported to occur at 779.3 eV [39], Co(OH)2 and a satellite peak at 786.1 eV was observed for
J Solid State Electrochem (2020) 24:1291–1304 1295

Fig. 5 XPS data of Co held in


0.1 M glycine and 1 M NaClO4 (a)
for 30 min in passive region (−
0.704 V vs. SSE) a Co 2p, b O1s.
Experimental results are shown as
points, and de-convoluted com-
ponents are shown as dashed lines
with colored area

(b)

Co2O3 [46]. The peak seen at 786.7 eV in the present study is arise from Co2O3 or Co3O4. The peak at 531.9 eV is likely to
the satellite peak of either Co(OH)2 or Co2O3. arise from cobalt glycine complex, since oxygen in another Co
From a binding energy of 790 eV to 805 eV, three more organic complex (Co with 3-Br-2,4-pentanedione) is reported
peaks at 793.4, 794.8, and 797.4 eV corresponding to Co 2p1/2 to exhibit a peak at 531.9 eV [50, 51]. Based on the XPS
levels are resolved. The energy difference of the doublet peaks analysis of the Co held at − 704 mV vs. SSE in glycine solu-
of Co (2p3/2) and Co (2p1/2) was 15.1 eV, which matches well tion, one can conclude that the Co surface analyzed contains a
with that reported in literature [46]. Alstrup et al. reported that mixture of Co oxides, hydroxides, and organo-metallic
metallic Co 2p1/2 peak occurs at 793.6 eV, and the peak at complexes.
793.3 eV in this work is assigned to metallic Co. The binding
energy of Co 2p1/2 in Co3O4 is reported to occur at 794.5 eV Impedance results
[42], or 794.8 eV [47] and 795.2 eV [48]. Hence, the peak at
794.8 eV in Fig. 5a can be assigned to Co3O4 (2p1/2). The peak Electrochemical impedance spectroscopy is an established
at 797.4 eV in Fig. 5a of this study is also close to reported technique and is widely used to study a variety of electro-
value of Co3O4 (2p1/2) peak (797.8 eV) in the literature [48]. chemical phenomena [52, 53]. EIS data were acquired by
The O 1s spectrum (Fig. 5b) shows a peak 530.1 eV and superimposing an ac perturbation on certain dc bias values
another at 531.9 eV. Wagner et al. analyzed several metal which are denoted by diamond markers in Fig. 2. The results
oxides using XPS and Auger and reported that the O1s peak are shown as complex plane plots in Fig. 6. The markers
of Co3O4 occurred at 530 eV. In another study, the O1s peak of represent the experimental data while the lines represent
Co3O4 was reported to occur at 529.7 eV and that of Co2O3 at EEC model predictions, explained later. Figure 6a shows that
529.9 eV [49]. The peak seen in this work at 530.1 eV would at − 984, − 914, and − 864 mV vs. SSE, in the active region, at
1296 J Solid State Electrochem (2020) 24:1291–1304

Fig. 6 Complex plane plots of


impedance spectra acquired in a (a)
the active region and b passive
region. Experimental results are
shown as markers and electrical
equivalent circuit (EEC) model
results using the circuit given in
Fig. 7 are shown as lines

(b)

least two loops are visible to the naked eye. The high- would show negative resistance. It is well known that in the
frequency loop impedance would arise from the parallel ele- passive region, low-frequency negative resistance would be
ments of double layer and the charge transfer resistance, and observed in the impedance spectra [54–56]. As the polariza-
the mid- and low-frequency loops would correspond to the tion curve exhibits a decrease in current with an increase in
faradaic process influenced by the relaxation of adsorbed potential, the sinusoidal current response to low-frequency ac
intermediates. potential would exhibit a phase difference of 180 °, which
Figure 6a also shows that when the dc bias is increased would manifest as negative resistance.
from − 984 to − 914 mV vs. SSE, the diameter of the high-
frequency loop diminishes, indicating that the charge transfer Electrical equivalent circuit analysis
resistance decreases with potential. There is no clear and sig-
nificant difference in the high-frequency loops observed at dc The impedance data presented in Fig. 6 were modeled using
bias values of − 914 mV and − 864 mV vs. SSE. When the dc circuit shown in Fig. 7, where Rsol is the solution resistance, Rp
bias is changed from − 984 to − 914 mV vs. SSE, the low- is the polarization resistance, and (Ri,Ci) (i = 1, 2, and 3) are
frequency limit of impedance decreases, but then, the dc bias the Maxwell pairs. It is possible to use Voigt or ladder circuits
is decreased further to − 864 mV vs. SSE, and the low- to represent the faradaic impedance equally well since they are
frequency limit of impedance increases significantly. degenerate [57]. The best fit parameters are listed in Table 1
Figure 6b shows the results in the passive region, and it is and the values in parenthesis are the corresponding standard
seen that the low-frequency impedance values tend toward the errors. The standard error is relatively large for Maxwell ele-
negative real axis. At high frequencies, a capacitive loop cor- ments and small for the CPE parameters, Rsol and Rp. The
responding to the double layer in parallel with the charge high-frequency loop is actually a depressed semi-circle and
transfer resistance is observed. Beyond this point, at mid fre- was modeled using a constant phase element rather than a
quencies, the real part of impedance increases with a decrease simple capacitor. Initially, a circuit with one or two Maxwell
in frequency, suggesting a capacitive behavior. This is marked elements was evaluated, but the fit was poor for most of the
by the arrow in Fig. 6b. At lower frequencies, the data points spectra. Only the data at − 984 mV vs. SSE could be modeled
move toward the second quadrant in the complex plane with a with two Maxwell elements and the remaining data were
decrease in frequency, indicating that lower frequency values modeled with three Maxwell elements. The CPE exponent n
J Solid State Electrochem (2020) 24:1291–1304 1297

Fig. 7 Circuit employed to model


experimental EIS data in Fig. 6.
The data at 0.07 V vs. OCP were
modeled using only two Maxwell
elements, while all other data
were modeled using three
Maxwell elements. The
polarization resistance Rp was
allowed to hold negative values to
model the data in the passive
region

is less than 1 indicating that the surface is rough. CPE behav- circuit model cannot be used to predict the potentiodynamic
ior could arise from a variety of causes including non- polarization curve (Fig. 2). To understand the details of reac-
uniformity in chemical or structure, variation along the surface tions better, the polarization and EIS data were analyzed in the
or perpendicular to the surface (e.g., heterogeneous film), framework of mechanistic analysis, as described below.
electrode porosity etc. [58, 59], and in case of Co dissolving
in glycine solutions, it is likely caused by the surface
roughness. Reaction mechanism analysis
The value of Rsol remains more or less a constant (2.2 ±
0.1 Ω cm2) for all the EIS data reported here. The low- Initially, a three-step mechanism with two adsorbed interme-
frequency limit of faradaic impedance is the polarization re- diates viz. Co2þ 3þ
ads and Coads , described in Eqs. (1)–(3) was
sistance, and Table 1 shows that the variation of Rp with the dc evaluated. Here, ki (where i = A, -A, B, -B, C, D, and E) are
bias is non-monotonic. In the passive region, since the polar- the rate constants of each step, with negative sign correspond-
ization current decreases with potential (Fig. 2), the polariza- ing to the reverse reaction. In the dissolution steps, the reverse
tion resistance is negative, which is along the expected lines reactions are neglected since the dissolved species is carried
[55, 60]. The standard errors in the Maxwell element values away from the interface. The rate constants of electrochemical
are relatively larger, and more importantly, it is challenging to steps are assumed to have exponential relationship with po-
obtain physical insights from these values. In addition, the tential, while those of chemical steps are independent of

Table 1 Electrical equivalent circuit parameters, for Co dissolution in 0.1 M glycine solution at pH 10. The EIS data in Fig. 6 were modeled using the
circuit shown in Fig. 7. The values in parenthesis are the standard errors

DC bias (E) /mV vs. SSE − 984 − 914 − 864 − 784 − 704
Rsol/Ω cm2 2.3 (1%) 2.4 (1%) 2.4 (1%) 2.4 (2%) 2.4 (1%)
Q-Y0/× 10−4 Ω−1 Sn cm−2 3.9 × 10−4 (5%) 3.2 × 10−4 (2%) 3.1 × 10−4 (2%) 3.4 × 10−4 (11%) 3.0 × 10−4 (7%)
n 0.79 (1%) 0.80 (1%) 0.72 (4%) 0.77 (2%) 0.80 (2%)
Rp/Ω cm2 25.3 (1%) 18.6 (1%) 28.3 (2%) − 45.1 (8%) − 36.0 (4%)
C1/mF cm−2 0.3 (13%) 1.3 (20%) 2.1 (6%) 80.0 (6%) 65.3 (4%)
R1/Ω cm2 85.3 (3%) 45.2 (9%) 36.5 (5%) 15.4 (3%) 17.5 (6%)
C2/mF cm−2 0.1 (13%) 0.5 (9%) 16.1 (6%) 7.5 (15%) 10.0 (18%)
R2/Ω cm2 138.7 (12%) 66.2 (7%) 38.8 (5%) 24.7 (7%) 26.4 (8%)
C3/mF cm−2 – 4.7 (7%) 0.1 (16%) 47.8 (21%) 0.1 (20%)
R3/Ω cm2 – 72.5 (7%) 49.8 (7%) 37.6 (10%) 51.2 (8%)
χ2 9.7 × 10−5 9.3 × 10−5 2.0 × 10−4 4. 6 × 10−4 2.4 × 10−4
1298 J Solid State Electrochem (2020) 24:1291–1304

potential [60–62]. For the electrochemical reactions, the rate evaluated. Here, ki (where i = 1, − 1, 2, − 2, 3, 4, 5, − 5, 6, − 6,
constant can be expressed as k i ¼ k i0 ebi E where ki0 is the pre- 7, and 8) are the rate constants of each step. In addition, the
exponential factor of rate constant. The exponent is given by reverse reactions in the steps 5–8 are neglected. A pictorial rep-
bi ¼ mα iF
RT where m is the number of electrons transferred in the
resentation of the same is given in Fig. 8b. This mechanism has
step, αi is the transfer coefficient (0 ≤ αi ≤ 1), F is the faraday been employed earlier to describe the dissolution of Zr in acidic
constant, R is the universal gas constant, and T is the fluoride media [68]. The polarization trends and impedance spec-
temperature. tra of Zr dissolution process were qualitatively different from
those exhibited by Co dissolution in alkaline glycine solution.
The stable oxidation states of Zr are + 3 and + 4, whereas they are
kA
‐ + 2 and + 3 for Co. Nevertheless, the dissolution of each metal-
Co ⇄Co2þ
ads þ 2e ð1Þ solution combination needs not to proceed via a unique and
k ‐A

kB
distinct mechanism, and hence, this model was evaluated.
‐ Most likely, in the mechanism described in Fig. 8, the ox-
ads ⇄Coads þ e
Co2þ ð2Þ

k ‐B ides, hydroxides, or glycinates of Co(II) are described, respec-
kC tively, by Co2þ 2þ
ads−1 and Coads−2 , and those of Co(III) are de-
ads → Cosol
Co3þ ð3Þ

scribed by Coads−1 and Co3þ

ads−2 , although the exact identities
kD ‐
ads → Cosol þ e
Co2þ ð4Þ
3þ of these species are not known. The fractional surface cover-
kE age of the intermediates species Co2þ 2þ 3þ
ads‐1 ,Coads‐2 ,Coads‐1 , and

ads þ Co → Coads þ Cosol þ 3e
Co3þ ð5Þ
3þ 3þ
Co3þads‐2 is given by θ1, θ2, θ3, and θ4, respectively. For the

It is worth noting that only forward reaction constants are adsorbed intermediate species, Langmuir isotherm model
considered in steps given in Eqs. (3)–(5) for the sake of sim- was invoked and mass transfer is assumed to be rapid. The
plicity since the concentration of dissolved Co species in the mass balance equations for a similar mechanism have been
solution is negligible. In a rigorous definition, the term mech- described earlier for Zr dissolving in HF [68] and are summa-
anism would consist of elementary steps only, which means rized below.
that multi-electron transfer steps, or any step with an order
other than 1 or 2, cannot be part of a mechanism [61, 63, dθ1
64]. On the other hand, literature has also employed such Γ ¼ k 1 θV −k −1 θ1 −k 3 θ1 þ k −3 θ3 −k 7 θ1 ð6Þ
dt
pseudo-elementary steps as part of detailed mechanistic de-
dθ2
scription of an overall reaction [65, 66], and the kinetics of Γ ¼ k 2 θV −k ‐2 θ2 −k 4 θ2 þ k ‐4 θ4 −k 8 θ2 ð7Þ
dt
such steps can still be described by an equation similar to
Butler-Volmer equation [67]. In this work, the more lenient dθ3
Γ ¼ k 3 θ1 −k −3 θ3 −k 5 θ3 ð8Þ
definition of mechanism is employed, and pseudo elementary dt
steps are included in the mechanism. dθ4
Γ ¼ k 4 θ2 −k −4 θ4 −k 6 θ4 ð9Þ
While the polarization data in the active region could be dt
modeled using Eqs. (1)–(3), the decrease in current with po-
Here, θv is the fraction of surface with exposed bare metal,
tential in the passive region could not be captured by the
given by θv = 1- θ1- θ2- θ3- θ4, ki and k‐i are the rate constants
model, and hence, it was rejected. Next, the mechanism de-
of forward reaction and reverse reaction and those dependent
scribed in Eqs. (1)–(4) was evaluated and it could model the
with potential for electrochemical steps. Γ(gamma) is the
active and passive regions of the polarization curve, but the
available sites per unit area. The faradaic current is given by
EIS data could not be modeled satisfactorily with this mech-
 
anism (Supplementary Figs. S1–S3). Subsequently, another ð2  fk 1 θV −k −1 θ1 þ k 2 θV −k −2 θ2 gÞþ
iF ¼ F ð10Þ
four-step mechanism given by Eqs. (1)–(3) and (5) was also ðk 3 θ1 −k −3 θ3 Þ þ ðk 4 θ2 −k −4 θ4 Þ þ k 7 θ1 þ k 8 θ2
evaluated. The step is Eq. (5) is known as catalytic step, and
this mechanism could match the polarization data well, but the The steady-state fractional surface coverage values (θi,ss)
match between the model and experimental data of the imped- can be obtained by setting Eqs. ( 6)–(9) to zero
ance spectra was poor. (Supplementary S4). The steady-state faradaic current (iF,ss)
Since three Maxwell elements were required to model the EIS can be calculated from Eq. (11).
results, any mechanism chosen to describe the Co dissolution in
i F;ss ¼ 3 F½ðk 5 þ k 7dc Þθ1ss þ ðk 6 þ k 8dc Þθ2ss  ð11Þ
glycine should employ three or more intermediate species.
Therefore, the mechanism given in Fig. 8a, with four intermedi- The subscript ‘dc’ in the rate constants indicates that they
ate species viz. Co2þ 3þ 2þ 3þ
ads‐1 ,Coads‐1 ,Coads‐2 , and Coads‐2 , was are evaluated at the dc potential Edc. The equations for the rate
J Solid State Electrochem (2020) 24:1291–1304 1299

Fig. 8 Proposed reaction (a) 2 e- Co


mechanism. a Equation. b 2 e-
k1
Pictorial representation k2
k-1
k-2
2+ 2+
Co ads-1 Co ads-2
k7 k8
e- e- e- e-
3+
Co sol
k3 k-3 k4 k-4

k5 k6

Co3+
ads-1 Co3+
ads-2

(b) Co3+
sol

Co3+
ads-2

Co3+
ads-1 2+
Co ads-2
2+
Co ads-1

constants and fractional surface coverage values can be ex- rearrangement of Eqs (6)–(9) and (10), the charge transfer
panded in Taylor series and truncated after the first two terms, resistance Rct can be written as
to linearize the equations. After linearization and

 
2½b1 k 1dc θVss −b−1 k −1dc θ1ss þ b2 k 2dc θVss −b−2 k −2dc θ2ss þ
R−1 ¼F ð12Þ
ct ðb3 k 3dc θ1ss −b−3 k −3dc θ3ss Þ þ ðb4 k 4dc θ2ss −b−4 k −4dc θ4ss Þ þ b7 k 7dc θ1ss þ b8 k 8dc θ2ss

and the Faradaic impedance (ZF) can be written as


2 3
dθ1
6 ð2k 1dc þ 2k −1dc þ 2k 2dc −k 3dc −k 7dc Þ þ 7
6 dE 7
1 Δi F −1 6 dθ 7
≃ ¼ Rct −F 6 ð2k 1dc þ 2k 2dc þ 2k −2dc −k 4dc −k 8dc Þ
2
þ 7 ð13Þ
Z F ΔE 6 dE 7
4 dθ3 dθ4 5
ð2k 1dc þ 2k 2dc þ k −3dc Þ þ ð2k 1dc þ 2k 2dc þ k −4dc Þ
dE dE
1300 J Solid State Electrochem (2020) 24:1291–1304

The total impedance ZTotal can be written as results are captured by the simulations. At − 984 mV vs.
Z Total ¼ Rsol þ QðjωÞ1n þZ −1 . SSE, the model predicts a capacitive loop at high frequen-
F

The expressions for the derivatives of fractional surface cies, another capacitive loop at mid frequencies, and a
coverage with potential are given in Supplementary S5. The third capacitive loop at low frequencies, as seen in Fig.
Matlab® programs employed to calculate the polarization and 9a. The match between simulated and experimental results
electrochemical impedance data can be obtained by contacting is good in the high- and low-frequency regions, and in the
the corresponding author. The best fit kinetic parameters are mid-frequency region, the match is relatively poorer. An
listed in Table 2. The dashed lines in Fig. 2 are the model examination of Fig. 9b and c shows that there again, the
predictions of the polarization curve, and it is seen that the match is good in the high- and low-frequency regions and
simulated results are in fairly good agreement with the exper- is poorer at the mid-frequency region.
imental results in the active and passive regions. Figure 10 shows the comparison of experimental imped-
Table 2 shows that the pre-exponents of the forward reac- ance data and reaction mechanism model results in the passive
tions are smaller than the pre-exponents of the corresponding region. At − 784 mV vs. SSE, the model matches the data in
reverse reactions, but the actual values of the rate constants the high frequency and low frequency ends well and in par-
depend on the potential and the rates at any given potential ticular, the tendency of the real part of impedance value to
depend on the corresponding rate constant as well as the frac- move towards negative values at low frequencies is modeled
tional surface coverage of the relevant species. Under steady- correctly. In the mid-frequency region, the capacitive behav-
state conditions, the reaction rates of the forward steps (ri for ior, as indicated by the arrow in Fig. 10a, is also captured,
the ith step) are always larger than the rates of the correspond- although the match between experimental and simulation re-
ing reverse steps. sults is not quantitative. Figure 10b also shows that the mech-
The mechanistic model predictions of impedance data, anistic simulation results match the salient features of the ex-
in the active region, are shown in Fig. 9. Although the perimental results and that the match is good in the high- and
mechanistic simulation results are not identical to the ex- low-frequency ends, while it is relatively poor in the mid-
perimental data, the major features of the experimental frequency region. The assumption of Langmuir isotherm
model may not be accurate and models such as Frumkin or
Temkin isotherm may offer a better description of the process,
but will also increase the complexity of the model. It may be
Table 2 Best fit
parameter values used to Parameter Units Value possible to obtain better fit at mid frequencies by replacing the
simulate the mechanistic Langmuir model with other isotherms and by increasing the
model results in Figs. 2, k10 mol cm−2 s−1 1.8 × 10−10 number of intermediates and the reaction steps, but the confi-
9, and 10, using the k20 5.7 × 10−09 dence in the results would be poorer and hence they were not
reaction mechanism k30 5.0 × 10−10
shown in Fig. 8 attempted [53].
k40 1.4 × 10−08 It should be noted that the match in the mechanistic analy-
k50 7.2 × 10−14 sis (Figs. 9 and 10) is not as good as that observed in EEC fit
k60 4.4 × 10−11 (Fig. 6), but in case of EEC, each spectrum is fitted indepen-
k70 1.1 × 10−15 dently, and the polarization data cannot be predicted by EEC.
k80 1.2 × 10−08 The solution resistance values and the double-layer capacitor,
k−10 4.8 × 10−09 modeled by CPE, were obtained from EEC. In addition, the
k−20 1.9 × 10−07 minimum number of adsorbed intermediates present in the
k−30 9.9 × 10−06 chosen mechanism was guided by the number of Maxwell
k−40 4.4 × 10−06 elements required to describe the faradaic impedance in EEC
b1 V−1 3.7 [62]. Although it is possible to extract the kinetic paramters
b2 10.7 from EEC element values, the process is more tedious than
b3 15.1 fitting the mechanistic model results to the experimental data,
b4 15.6 and hence, direct model optimization was employed in this
b7 32.3 work.
b8 21.3 In case of RMA, if we were to analyze only one impedance
b−1 − 14.4 spectrum, it could be matched very well (results not shown),
b−2 − 15.8 but the confidence in the model will be poor and it is not
b−3 − 15.2 possible to obtain physical insights reliably in both active
b−4 − 21.5 and passive regions. Therefore, the complete data set of im-
Г mol cm−2 4.2 × 10−8 pedance and the polarization data were fitted simultaneously
using the proposed mechanism.
J Solid State Electrochem (2020) 24:1291–1304 1301

Fig. 9 Mechanistic model results


and experimental EIS data, (a)
displayed as complex plane plots
in the active region a − 984 mV
vs. SSE, b − 914 mV vs. SSE, and
c − 864 mV vs. SSE.
Experimental results are shown as
open circles and mechanistic
model results are shown as filled
squares

(b)

(c)

Figure 11 shows the fractional surface coverage values of of Fig. 12 refer to the steps shown in Fig. 8. The rate of each
Co2þ 3þ 2þ 3þ
ads‐1 ,Coads‐1 ,Coads‐2 , and Coads‐2 as predicted by the model.
electrochemical step is much larger than that of the corre-
Initially, near OCP, the surface is mostly bare metal, since sponding chemical step. Note that the ordinate is in logarith-
glycine would form soluble complex with Co. With an in- mic scale. The electrochemcial dissolution rate (step 8) in-
crease in overpotential, the fractional surface coverage of creases with potential, reaches a maximum at − 865 mV vs.
Co(II) species, as given by the sum of θ1 and θ2, increases, SSE, and then decreases with a further increase in potential.
reaches a maximum near − 805 mV vs. SSE, and then de- This is essentially the product of the relevant rate constants,
creases to negligible values at − 650 mV vs. SSE. On the other which depend exponentially on potential, and fractional sur-
hand, the fractional surface coverage of Co(III) species is very face coverage values of Co2þ ads‐2 , which exhibits a maximum
low near OCP and continues to increase, reaching almost full (Fig. 11). At large overpotenials, the increase in the values of
coverage at − 650 mV vs. SSE. These predictions are in line the rate constants of electrochemcial steps does not match the
with the low current values observed at large anodic potentials decrease in fractional surface coverage of Co2þ ads‐2 with poten-
in Fig. 2. The fractional surface coverage values of the species tial, and the net result is a decrease in electrochemcial disso-
at a given potential are determined by the balance between lution rate beyond − 865 mV vs. SSE. On the other hand, the
formation rate and the dissolution rate, and at large anodic chemical reaction rate constants are independent of potential,
potentials, the oxide and hydroxide formation rate would be and the chemical dissolution rate (step 6) follows the variation
higher, leading to passivation. of Co3þads‐2 species with potential faithfully and shows a con-
The rate of chemical and electrochemical dissolution of Co tinuous increase till − 650 mV vs. SSE.
in 0.1 M glycine solution at pH 10, predicted by the mecha- The rate of other electrochemical dissolution (step 7) is
nism in Fig. 8, are presented in Fig. 12. The steps in the legend significantly smaller than step 8, and the rate of the
1302 J Solid State Electrochem (2020) 24:1291–1304

(a)

Fig. 11 Fractional surface coverage of bare Co and various adsorbed


intermediates of the proposed mechanism (Fig. 8), as a function of poten-
tial in 0.1 M glycine solution at pH 10

spectra are influenced by even minor reactions whose contri-


bution to the polarization data may be small. The proposed
mechanism describes the observed polarization and imped-
ance data satisfactorily and offers insights into the variation
of the fractional surface coverage of the intermediate species
and the dissolution rates as a function of potential. Impedance
alone cannot be used to identify the species present on the
surface, but as a versatile in situ technique, it can offer valu-
able insights into the physicochemcial processes occuring at
the electrode-electroltye interface. In conjuction with
potential-pH diagram and XPS results, mechanistic analysis
of polarization data and impedance spectra can be used to
identify the mechanism and the possible intermediate species.

(b)

Fig. 10 Mechanistic model results and experimental EIS data, displayed


as complex plane plots in the passive region; a − 784 mV vs. SSE and b −
704 mV vs. SSE. Experimental results are shown as open circles and
mechanistic model results are shown as filled squares

corresponding chemcial dissolution (step 5) is even smaller


Fig. 12 Chemical and electrochemical Co dissolution rates vs. potential
than that. The contribution of step 5 and step 7 to the polari- in 0.1 M glycine solutions at pH 10, as predicted by the reaction
zation data is minimal. The results also show that impedance mechanism model
J Solid State Electrochem (2020) 24:1291–1304 1303

Conclusions patterned structures. ECS J Solid State Sci Technol 6(5):P276–


P283
9. Peethala BC, Amanapu HP, Lagudu URK, Babu SV (2012) Cobalt
Anodic dissolution of Co in 0.1 M glycine solutions was in- polishing with reduced galvanic corrosion at copper/cobalt interface
vestigated using potentiodynamic polarization and electro- using hydrogen peroxide as an oxidizer in colloidal silica-based
chemical impedance spectroscopy. Well-defined active and slurries. J Electrochem Soc 159(6):H582–H588
passive dissolution regions are seen in potentiodynamic polar- 10. Johnson CA, Wei S, Roy D (2018) An alkaline slurry design for Co-
Cu CMP systems evaluated in the tribo-electrochemical approach.
ization data. Impedance spectra acquired in the active and ECS J Solid State Sci Technol 7(2):P38–P49
passive regions were modeled using electrical equivalent cir- 11. Tian Q, Wang S, Xiao Y, Wang C, Wang Q, Liu F, Zhang J, Wang R
cuit as well as mechanistic analysis. A mechanism with four (2018) Effect of amine based chelating agent and H2O2 on cobalt
adsorbed intermediates, viz. two Co(II) and two Co(III) spe- contact chemical mechanical polishing. ECS J Solid State Sci
Technol 7(8):P416–P422
cies, which could be oxides, hydroxides, and glycine com-
12. Xu W, Ma L, Chen Y, Liang H (2018) In situ study of mechanical-
plexes of Co, was proposed and fitted to the experimental data electrochemical interactions during cobalt ECMP. J Electrochem
to obtain the kinetic parameters. The presence of Co oxides, Soc 165(5):E184–E189
hydroxides, and possible organo-metallic complexes was 13. Xu W, Ma L, Chen Y, Liang H (2018) Mechano-oxidation during
cobalt polishing. Wear 416-417:36–43
identified by the analysis of XPS data. The proposed mecha-
14. Ji J, Pan G, Zhang W, Du Y, He P, Tian Y, Wang C (2017) Role of
nistic model predicts that the chemical dissolution rate would additive in alkaline slurries for Co CMP. ECS J Solid State Sci
increase with potential and saturate at large anodic Technol 6(12):P813–P818
overpotential, whereas the electrochemical dissolution rate 15. Turk MC, Shi X, Gonyer DAJ, Roy D (2015) Chemical and me-
would exhibit a maximum at intermediate overpotential and chanical aspects of a Co-Cu planarization scheme based on an al-
kaline slurry formulation. ECS J Solid State Sci Technol 5:P88–P99
would be small at small and large overpotentials.
16. He P, Yang G, Qu X-P (2017) Study on effect of different
complexing agents and inhibitors on Co corrosion in H2O2 based
Acknowledgments The authors thank the Department of Science and alkaline solution by EQCM. In: international conference on
Technology, India, for providing the SEM facility to the Department of planarization/CMP technology 2017, Leuven
Chemical Engineering, Indian Institute of Technology Madras through 17. Critelli RAJ, Sumodjo PTA, Bertotti M, Torresi RM
the FIST program, and Prof. B. Boukamp, U Twente, for the linear KKT (2018) Influence of glycine on Co electrodeposition: IR spectros-
software. copy and near-surface pH investigations. Electrochim Acta 260:
762–771
18. Real SG, Ribotta SB, Arvia AJ (2008) The electrochemical disso-
lution of cobalt in carbonate–bicarbonate solutions from EIS and
steady polarization data. Corros Sci 50(2):463–472
19. Calderon JA, Mattos OR, Barcia OE ,Cordoba de Torresi SI, Pereira
References da Silva JE (2002) Electrodissolution of cobalt in carbonate/
bicarbonate media. Electrochim Acta 47:4531–4541
1. Lu H-S, Zeng X, Wang J-X, Chen F, Qu X-P (2012) The effect of 20. Calderón JA, Barcia OE, Mattos OR (2008) Reaction model for
glycine and benzotriazole on corrosion and polishing properties of kinetic of cobalt dissolution in carbonate/bicarbonate media.
cobalt in acid slurry. J Electrochem Soc 159(9):C383–C387 Corros Sci 50(7):2101–2109
2. Alety SR, Lagudu URK, Popuri R, Patlolla R, Surisetty CVVS, 21. Gallant D, Pézolet M, Jacques A, Simard S (2006) Analysis of a
Babu SV (2017) Cleaning solutions for ultrathin cobalt barriers complex electrochemical process: The anodic dissolution and pas-
for advanced technology nodes. ECS J Solid State Sci Technol sivation of cobalt in aqueous media near neutral pH. Corros Sci
6(9):P671–P680 48(9):2547–2559
3. Jiang L, He Y, Li Y, Li Y, Luo J (2014) Synergetic effect of H2O2 22. Ismail KA, Badawy WA (2000) Electrochemical and XPS investi-
and glycine on cobalt CMP in weakly alkaline slurry. Microelectron gations of cobalt in KOH solutions. J Appl Electrochem 30(11):
Eng 122:82–86 1303–1311
4. Kaloyeros AE, Pan Y, Goff J, Arkles B (2019) Review-cobalt thin 23. Gallant D, Simard S (2005) A study on the localized corrosion of
films: trends in processing technologies and emerging applications. cobalt in bicarbonate solutions containing halide ions. Corros Sci
ECS J Solid State Sci Technol 8(2):P119–P152 47(7):1810–1838
5. Li X, Pan G, Wang C, Guo X, He P, Li Y (2016) Effect of chelating 24. Boukamp BA (1995) A linear Kronig‐Kramers transform test for
agent on reducing galvanic corrosion between cobalt and copper in immittance data validation. J Electrochem Soc 142(6):1885–1894
alkaline slurry. ECS J Solid State Sci Technol 5(9):P540–P545 25. Chemical Equilibrium Diagrams - Software (2010) KTH Royal
6. Wu D, Kang R, Guo J, Liu Z, Wan C, Jin Z (2019) On the reaction Institute of Technology, Stockholm. https://www.kth.se/che/
mechanism of a hydroxyethylidene diphosphonic acid-based elec- medusa. Accessed 13 Dec 2019
trolyte for electrochemical mechanical polishing of copper. 26. Acevedo-Peña P, Vázquez G, Laverde D, Pedraza-Rosas JE,
Electrochem Commun 103:48–54 González I (2010) Influence of structural transformations over the
7. He P, Wu B, Shao S, Teng T, Wang P, Qu X-P (2019) Characterization electrochemical behavior of Ti anodic films grown in 0.1 M NaOH.
of 1, 2, 4-triazole as corrosion inhibitor for chemical mechanical J Solid State Electrochem 14(5):757–767
polishing of cobalt in H2O2 based acid slurry. ECS J Solid State Sci 27. Toparli C, Hieke SW, Altin A, Kasian O, Scheu C, Erbe A
Technol 8(5):P3075–P3084 (2017) State of the surface of antibacterial copper in phosphate
8. Sagi KV, Teugels LG, van der Veen MH, Struyf H, Alety SR, Babu buffered saline. J Electrochem Soc 164(12):H734–H742
SV (2017) Chemical mechanical polishing of chemical vapor de- 28. Ergeneman O, Sivaraman KM, Pané S, Pellicer E, Teleki A, Hirt
posited Co films with minimal corrosion in the cu/co/mn/sicoh AM, Baró MD, Nelson BJ (2011) Morphology, structure and
1304 J Solid State Electrochem (2020) 24:1291–1304

magnetic properties of cobalt–nickel films obtained from acidic 50. Srivastava S, Badrinarayanan S, Mukhedkar AJ (1985) X-ray pho-
electrolytes containing glycine. Electrochim Acta 56(3):1399–1408 toelectron spectra of metal complexes of substituted 2,
29. Kim S (2010) Seedless copper electrodeposition onto tantalum dif- 4_pentanediones. Polyhedron 4(3):409–415
fusion barrier by two-step deposition process. Electrochem Solid- 51. Hoof DL, Tisley DG, Walton RA (1973) Studies on metal carbox-
State Lett 13(11):D83 ylates. Part 111. Pyridine-2,6-dicarboxylates of the lanthanides.
30. Thomas S, Cole IS, Sridhar M, Birbilis N (2013) Revisiting zinc Synthesis and spectral studies and the x-ray photoelectron spectra
passivation in alkaline solutions. Electrochim Acta 97:192–201 of several pyridine carboxylate complexes. J Chem Soc Dalton
31. Chao CY, Lin LF, Macdonald DD (1981) A point defect model for Trans 100:200–204
anodic passive films : film growth kinetics. J Electrochem Soc 52. Barsoukov E, Macdonald JR (2005) Impedance spectroscopy the-
128(6):1187–1194 ory, experiment, and applications. Wiley, New Jersy
32. Schubert N, Schneider M, Michealis A (2013) The mechanism of 53. Macdonald DD (1990) Review of mechanistic analysis by electro-
anodic dissolution of cobalt in neutral and alkaline electrolyte at chemical impedance spectroscopy. Electrochim Acta 35(10):1509–
high current density. Electrochim Acta 113:748–754 1525
33. Tzvetkov B, Bojinov M, Girginov A (2008) Nanoporous oxide 54. Sadkowski A (2004) CNLS fits and Kramers–Kronig validation of
formation by anodic oxidation of Nb in sulphate–fluoride electro- resonant EIS data. J Electroanal Chem 573(2):241–253
lytes. J Solid State Electrochem 13:1215–1226
55. Cattarin S, Musiani M, Tribollet B (2002) Nb electrodissolution in
34. Ibrahim MAM, Al Radadi RM (2015) Noncrystalline cobalt coat-
acid fluoride medium. J Electrochem Soc 149(10):B457
ings on copper substrates by electrodeposition from complexing
acidic glycine baths. Mater Chem Phys 151:222–232 56. Lasia A (2014) Electrochemical impedance spectroscopy and its
35. Tamilmani S, Huang W, Raghavan S, Small R (2002) Potential-pH applications. Springer, New York
diagrams of interest to chemical mechanical planarization of cop- 57. Fletcher S (1994) Tables of degenerate electrical networks for use in
per. J Electrochem Soc 149(12):G638–G642 the equivalent‐circuit analysis of electrochemical systems. J
36. Pourbaix M (1974) Atlas of electrochemical equilibria in aqueous Electrochem Soc 141(7):1823–1826
solution. NACE international, Houston 58. Pajkossy T (1997) Capacitance dispersion on solid electrodes: an-
37. Powell CJ (2012) Recommended Auger parameters for 42 elemen- ion adsorption studies on gold single crystal electrodes. Solid State
tal solids. J Electron Spectrosc Relat Phenom 185(1-2):1–3 Ionics 94(1-4):123–129
38. Mandale AB, Badrinarayanan S, Date SK, Sinha APB 59. Hitz C, Lasia A (2001) Experimental study and modeling of im-
(1984) Photoelectron-spectroscopic study of nickel, manganese pedance of the her on porous Ni electrodes. J Electroanal Chem
and cobalt selenides. J Electron Spectrosc Relat Phenom 33(1): 500(1-2):213–222
61–72 60. Mandula TR, Srinivasan R (2017) Electrochemical impedance
39. Haber J, Ungier L (1977) On chemical shifts of esca and auger lines spectroscopic studies on niobium anodic dissolution in HF. J
in cobalt oxides. J Electron Spectrosc Relat Phenom 12(3):305–312 Solid State Electrochem 21(11):3155–3167
40. McIntyre NS, Johnston DD, Coatsworth LL, Davidson RD, Brown 61. Gregori J, Gimenez-Romero D, Garcia-Jareño JJ, Vicente F
JR (1990) X-ray photoelectron spectroscopic studies of thin film (2006) Calculation of the rate constants of nickel electrodissolution
oxides of Cobalt and Molybdenum. Surf Interface Anal 15:262– in acid medium from EIS. J Solid State Electrochem 10(11):920–
272 928
41. Okamoto Y, Imanaka T, Teranishi S (1980) Surface structure of 62. Baranwal PK, Venkatesh RP (2017) Investigation of carbon steel
COO-MoO,/AlO, catalysts studied by X-Ray photoelectron spec- anodic dissolution in ammonium chloride solutions using electro-
troscopy. J Catal 65(2):448–460 chemical impedance spectroscopy. J Solid State Electrochem 21(5):
42. Chuang TJ, Brundle CR, Rice DW (1976) Interpretation of the x- 1373–1384
ray photoemission spectra of cobalt oxides and cobalt oxide sur- 63. Bard AJ, Faulkner LR (2001) Electrochemical methods.
faces. Surf Sci 59(2):413–429 Fundamentals and applications. Wiley, New York
43. Escard J, Mavel G, Guerchais EJ, Kergoat R (1973) X-Ray photo- 64. Macdonald DD, Real S, Smedley SI, Urquidi-Macdonald M
electron spectroscopy study of some metal(ii) halide and (1988) Evaluation of alloy anodes for aluminum‐air batteries IV.
pseudohalide complexes. Inorg Chem 13:695–701 Electrochemical impedance analysis of pure aluminum in at 25 °C.
44. Barber M, Connor JA, Derrick LMR, Hall MB, Hillier IH J Electrochem Soc 135(10):2410–2414
(1973) High energy photoelectron spectroscopy of transition metal
65. Shao H, Wang J, He W, Zhang J, Cao C (2005) EIS analysis on the
complexes. Part 2. Metallocenes. J Chem Soc Faraday Trans 2(69):
anodic dissolution kinetics of pure iron in a highly alkaline solution.
559–562
Electrochem Commun 7(12):1429–1433
45. Schenck CV, Dillard JG, Murray JW (1983) Surface analysis and
the adsorption of Co(ll) on Goethite. J Colloid Interface Sci 95(2): 66. Vicente F, Gregori J, García-Jareño JJ, Giménez-Romero D
398–405 (2005) Cyclic voltammetric generation and electrochemical quartz
46. Tan BJ, Klabunde KJ, Sherwood PA (1991) XPS studies of solvated crystal microbalance characterization of passive layer of nickel in a
metal atom dispersed catalysts. evidence for layered cobalt- weakly acid medium. J Solid State Electrochem 9(10):684–690
manganese particles on alumina and silica. J Am Chem Soc 67. Fletcher S (2009) Tafel slopes from first principles. J Solid State
113(3):855–861 Electrochem 13(4):537–549
47. Tyuliev G, Angelov S (1988) The nature of excess oxygen in 68. Amrutha M, Mandula TR, Srinivasan R (2018) Mechanistic analy-
Co3O4+ϵ. Appl Surf Sci 32(4):381–391 sis of anodic dissolution of Zr in acidic fluoride media. J
48. Bonnelle JP, Grimblot J, Dhuysser A (1975) Influence de la Electrochem Soc 165(3):C162–C170
polarisation des liaisons sur les spectres esca des oxydes de cobalt.
J Electron Spectrosc Relat Phenom 7(2):151–162 Publisher’s note Springer Nature remains neutral with regard to jurisdic-
49. McIntyre NS, Cook MG (1975) X-Ray photoelectron studies on of tional claims in published maps and institutional affiliations.
cobalt, nickel, and copper some oxides and hydroxides. Anal Chem
47(13):2208–22213

You might also like