J Solid State Electrochem (2017) 21:1373–1384

DOI 10.1007/s10008-016-3497-8

ORIGINAL PAPER

Investigation of carbon steel anodic dissolution in ammonium

chloride solutions using electrochemical impedance spectroscopy
Prince Kumar Baranwal 1 & R. Prasanna Venkatesh 1

Received: 6 September 2016 / Revised: 4 December 2016 / Accepted: 26 December 2016 / Published online: 7 January 2017
# Springer-Verlag Berlin Heidelberg 2017

Abstract The anodic dissolution of carbon steel in ammoni- Introduction

um chloride (NH4Cl) solutions (5, 10, and 20 wt%) is inves-
tigated via various electrochemical techniques and other com- Corrosion by means of wet ammonium chloride salts or aque-
plementary techniques. The polarization measurements ous solutions of ammonium chloride (NH4Cl) has plagued the
reveals that the carbon steel is susceptible to general corrosion. oil refining and petrochemical industries for many years. With
The impedance data taken at various overpotentials shows processes and streams where nitrogen and chlorides or ammo-
multiple loops, corresponding to capacitance, inductance, nia (NH3) and hydrochloric acid (HCl) are present, the forma-
and negative capacitance, and the number of time constants tion of ammonium chloride is feasible. The ammonium chlo-
observed is also not the same for various NH4Cl concentra- ride deposits leads to severe failure or corrosion as reported in
tions. From reaction mechanism analysis, a multi-step reaction the literature [1, 2]. The corrosivity of the NH4Cl stream de-
mechanism with three adsorbed intermediates and three pends on various factors such as concentration of NH4Cl,
dissolution paths (one chemical path and two electrochemical hydrodynamic conditions, temperature, etc.
paths) is proposed to describe the observed patterns in imped- Toba et al. [3] studied the effect of relative humidity
ance measurements. The surface coverage of intermediate and concentration of NH4Cl on carbon steel corrosion
species and the contribution of chemical reaction and electro- via weight loss and water absorption tests. The corro-
chemical reaction to the overall corrosion rate are also estimat- sion of various alloys in ammonium chloride solution
ed from the proposed model. The results obtained from field was investigated, and the results show that the nature
emission scanning electron microscopy and Raman spectros- of corrosion (general corrosion or pitting corrosion) and
copy measurements are also reported. the corrosion rate strongly depends upon the alloy com-
position [4]. The corrosion rate is also correlated to
pitting resistant equivalent number (PREN) number
Keywords Carbon steel . Ammonium chloride . which is estimated from alloy composition [5]. The ef-
Electrochemical impedance spectroscopy . Reaction fect of temperature and NH4Cl concentration was also
mechanism characterized using polarization measurements, and the
results show that both the parameters had a pronounced
effect on the corrosion rate [4]. The composition of the
corrosion product analyzed via thermogravimetric analy-
Electronic supplementary material The online version of this article sis (TGA) and spectroscopic studies reveals that the
(doi:10.1007/s10008-016-3497-8) contains supplementary material,
corrosion product contains a mixture of oxides and
which is available to authorized users.
oxy-hydroxides [6, 7].
* R. Prasanna Venkatesh
The reaction mechanism of carbon steel anodic dis-
[email protected] solution in various media is extensively studied. Some
of the important models which are reported in literature
1
Department of Chemical Engineering, Indian Institute of Technology are summarized here: Heusler [8] proposed a two-step
Guwahati, Guwahati 781039, India catalytic dissolution for acidic medium in which
1374 J Solid State Electrochem (2017) 21:1373–1384

(FeOH)ad is an intermediate adsorbate but not consumed carbon steel in NH4Cl solution is not exploited yet to the best
in the reaction. of author’s knowledge. Thus, proposing a kinetic model to
describe the carbon steel anodic dissolution behavior in
Fe þ OH− ⇄ðFeOHÞad þ e−
− ð1Þ NH4Cl concentrations is of great interest in the present work.
Fe þ ðFeOHÞad þ OH− →ðFeOHÞad þ ðFeOHÞþ
sol þ 2e Electrochemical impedance spectroscopy (EIS) is one of
the powerful tools available for investigation of electrochem-
The research group of Bockris [9–11] suggested a mecha- ical and corrosion reaction mechanisms [14–18]. Various pro-
nism in which the dissolution of (FeOH)ad is considered to be cesses such as faradaic and non-faradaic reactions including
slow. adsorption–desorption, diffusion of chemical compounds/
ions, and passivity of surfaces can be easily detected by means
Fe þ OH− ⇄ðFeOHÞad
ð2Þ of EIS technique. With application of a small sinusoidal per-
ðFeOHÞad →ðFeOHÞþ
sol
turbation, electrode properties and other physical properties
In both of the above models, the dissolution of Fe in the can be determined without interfering the system.
presence of only OH− ions was considered. Besides, if the The reaction mechanism of metal dissolution in various
impedance measurements show more than one time constant medium is successfully investigated by EIS measurements
in addition to electrical double layer (EDL), then it is difficult [12, 19–21]. Keddam et al. [12], in particular, employed an
to suggest the above models as it includes only one interme- approach called reaction mechanism analysis (RMA) to un-
diate adsorbed species Fe(OH)ad. The effects of various ions derstand the anodic dissolution of metal in the solution
in the electrolyte media on Fe dissolution behavior were also through impedance data. Proposing/eliminating a mechanism
considered later. Keddam et al. [12] showed multi-step mech- based on the patterns observed in EIS measurements is possi-
anism for Fe dissolution behavior in acidic media in the pres- ble. For example, a direct dissolution shows only single time
ence of sulfate ions with three adsorbed species (Feþ þ*
ad , Fead ,
constant in the impedance data while the dissolution through
2þ*
Fead ) as shown here: one intermediate species shows either single or double time
constants depending on the parameter values. The various
k1 k2 patterns that are possible for the given mechanism are sum-
Fe → Feþ
ad → Fesol
2þ
k3 marized in the literature [21].
k4
Feþ þ*
ad ⇄Fead → Fesol
2þ Hence, in this present work, EIS measurements were imple-
k‐3
k5 mented besides polarization measurements, field emission
k6 ð3Þ
Feþ
ad ⇄Fead → Fesol
2þ* 2þ scanning electron microscopy (FESEM) measurements, and
k‐5 Raman spectroscopy measurements to investigate the carbon
k4
Fead þ Fe → Feþ*
þ*
ad þ Fesol þ 2e
2þ −
steel anodic dissolution behavior in 5, 10, and 20 wt% NH4Cl
k − solutions at room temperature. The RMA approach is
ad þ Fe → Fead þ Fesol þ 2e
Fe2þ*
6 2þ* 2þ
employed to analyze the data obtained from EIS measurements.
They also added the following passivation step in the above
mechanism to describe the pre-passivation behavior observed
in polarization measurements. Experimental
k7
Materials
2þ*
Fead ⇄Fe2þ
ad ð4Þ
k−7
Carbon steel [C (0.22–0.26)%; Si (0.11–0.14)%; Mn (1.03–
Similarly, the corrosion of carbon steel in chloride media is also 1.06)%; P (0.04 % maximum); S (0.03 % maximum); Cr
extensively investigated. The most widely accepted mechanism is (0.03 % maximum); and rest Fe] was used for all the measure-
proposed by Li and his coworkers [13] which is given below. ments in this present work. NH4Cl (HiMedia) and Millipore
FeðsÞ þ 2Cl− ðaqÞ⇄FeCl2 ðsÞ þ 2e− water were used to prepare solutions of various concentrations
FeCl2 ðsÞ⇄FeCl2 ðinterfaceÞ→FeCl2 ðaqÞ (5, 10, and 20 wt%). All the experiments were carried out at
ð5Þ
FeCl2 ðsÞ þ Cl− ðaqÞ⇄FeCl3 ðsÞ þ e− natural pH of the solutions. The pH values of the solutions
FeCl3 ðsÞ⇄FeCl3 ðinterfaceÞ→FeCl3 ðaqÞ observed were 4.84, 4.69, and 4.22, respectively, for 5, 10, and
20 wt% NH4Cl solutions.
Although the carbon steel dissolution is extensively reported
for various other chloride media, viz. NaCl and HCl, only few Electrochemical measurements
works were reported in the literature in respect to the anodic
dissolution of carbon steel in ammonium chloride solutions. Electrochemical experiments were conducted using a
Especially, the mechanistic reaction pathway of corrosion of potentiostat [Metrohm Autolab, PGSTAT 204], and a standard
J Solid State Electrochem (2017) 21:1373–1384 1375

three electrode system, at (25 ± 1) °C. The working electrode dried in vacuum, and characterized using Raman spectroscopy
fabricated with carbon steel of dia. 9 mm was implanted inside with an excitation wavelength of 633 nm. The Raman spectra
a Teflon tube, such that only the cross section interacts with were recorded over a range of 100 to 2000 cm−1.
the electrolytic solution. The reference electrode was made of
Ag/AgCl (3 M KCl), and a Pt wire was used as the counter
electrode. The corrosion cell was retained within a Faraday Results and discussion
shield during all the experiments, in order to reduce the exter-
nal perturbations from the surroundings. Prior to each exper- OCP and potentiodynamic polarization measurements
iment, the working electrode was smoothed sequentially with
180, 320, and 600 grades of emery paper, followed by 1.0 and Prior to potentiodynamic polarization measurements, OCP
0.3 μm abrasive powder (alumina). The electrode was then measurements were carried out for a period of 300 s. The
washed with Millipore water and further ultrasonicated to OCP values obtained for 5, 10, and 20 wt% NH4Cl solutions
eliminate any adhered particles. The polarization and imped- are −0.710, −0.706, and −0.686 V [vs. Ag/AgCl (3 M KCl)],
ance measurements were carried out only after the open circuit respectively. As OCP reaches the stable value within 300 s,
potential (OCP) reached the stable value. The polarization the same is set as equilibration time for further electrochemical
measurements were performed by sweeping the potential from experiments. Figure 1 shows the anodic polarization curves of
OCP to +500 mV (w.r.t. OCP) to obtain the anodic branch, at a carbon steel in 5, 10, and 20 wt% NH4Cl solutions respective-
scan rate of 1 mV s−1. The impedance measurements were ly. The result shows that with an increase in concentration, the
performed by applying an amplitude of 10 mV (rms) for var- corrosion potential (Ecorr) shifts towards more positive values,
ious dc potential (+0.05, +0.15, and +0.25 V w.r.t. OCP). The though not significantly. Similar trend is also reported in the
frequency range employed was from 100 kHz to 1 mHz, with literature [4]. Also, the anodic current increases with an in-
six frequencies per decade. All the experiments were carried crease in NH4Cl concentration as the anodic branches move
out at least twice, and only the repeatable results are presented towards the right. This reveals that the dissolution rate in-
in this work. All the measurements were conducted at static creases with an increase in ammonium chloride concentration.
conditions using naturally aerated electrolyte solutions. The other observation in the polarization curves is
the appearance of kink in the anodic branch which is
Field emission scanning electron microscopy more evident especially at higher concentrations of
measurements NH4Cl (10 and 20 wt%). It could be expected that it
might be due to the occurrence of pitting corrosion on
The surface morphology images of carbon steel were ob- the carbon steel surface, and the potential below which
served using field emission scanning electron microscope passivation occurs is referred as breakdown or pitting
(Zeiss, Sigma), in order to characterize the type of corrosion. potential (Eb). However, the following features, which
The carbon steel sample was submerged in nitric acid, follow- are usually observed in polarization curve when pitting
ed by rinsing with water and polishing with emery paper and corrosion occurs, are not seen here [22–24]: (1) constant
abrasive alumina powder. After polishing, the sample was current with an increase in potential (formation of
rinsed with Millipore water, ultrasonicated, and air dried. strong passive layer), (2) the sharp increase in current
The sample was then submerged in the respective solutions with a slight increase in the potential (breakdown of
for a particular time period. At the end of 12 h, the sample was
removed out of the system and rinsed with Millipore water.
The washed sample was then air dried, and finally, the images
were taken.

Raman spectroscopy measurements

The corrosion products were analyzed using micro-Raman

spectroscope (Horiba Jobin Vyon, LabRam HR) in order to
identify the type of iron oxide (corrosion product) formed on
the carbon steel surface. The carbon steel sample was sub-
merged in nitric acid, rinsed with water, followed by polishing
with emery paper, and rinsed with Millipore water. The
washed sample was then air dried and dipped in the respective
solution. After 72 h, the sample was removed out of the sys- Fig. 1 Anodic polarization curves of carbon steel corroded in 5, 10, and
tem. The corrosion product formed was then scrubbed out, 20 wt% NH4Cl solutions. Scan rate = 1 mV s−1
1376 J Solid State Electrochem (2017) 21:1373–1384

passive layer and formation of stable pits), and (3) a The results are shown in Fig. 2a–d. There were no corrosion
decrease of Eb and Ecorr with an increase in Cl− ion marks as expected on untreated carbon steel sample. The car-
concentration. In the present case, constant current re- bon steel in various NH4Cl concentrations (5, 10, and 20 wt%)
gime is observed only for the few millivolt ranges and shows that corrosion is more of uniform type as shown in
the increase in current with potential after the Eb is Fig. 2b–d. It supports our presumption made from potentio-
more or less linear. This suggests that the corrosion dynamic polarization studies that general corrosion occurs
observed is of general type although carbon steel is when carbon steel of given composition reacts with NH4Cl
susceptible to pitting corrosion in chloride medium solution.
[25–27]. Especially, when the rate of general corrosion
via hydroxyl groups is more compared to that of the
rate of pitting corrosion via chloride ions, it is attributed Electrochemical impedance spectroscopy measurements
to the thinning of oxide/hydroxide layer [28]. When this
layer is thin and non-protective, one may expect the To understand the anodic dissolution behavior of carbon steel
corrosion is of only general type. As the anodic polar- in different concentrations of NH4Cl, EIS measurements were
ization curve does not show any strong passivation, the conducted at various dc potentials, i.e., +0.05, +0.15, and
corrosion may be of general type for the present system. +0.25 V w.r.t. OCP, and the results are presented in Fig. 3a–
It is also observed that corrosion is of either general i. The absolute potential indicated in these figures referred to
type or pitting type depending on the composition of the value of reversible hydrogen electrode (RHE). The EIS
the alloy in the chloride media [4, 6]. In particular, it measurements were performed after the system reaches its
is reported that carbon steel undergoes general corrosion stable OCP value.
in ammonium chloride solutions [4]. Following observations were made based on the EIS
data: (i) for 5 wt% NH4Cl solution, the data shows three
time constants for all the overpotentials investigated; (ii)
FESEM measurements for both 10 and 20 wt% NH4Cl solutions, three time con-
stants were observed at 0.05 and 0.15 V above OCP, while
To confirm the type of corrosion, the surface morphology four time constants were observed at 0.25 V above OCP;
images of carbon steel were obtained using FESEM, with (iii) the overall impedance decreases with overpotential for
and without treatment in various concentrations of NH4Cl. all the three concentrations, indicating that dissolution rate

Fig. 2 FESEM images of a

untreated carbon steel, b carbon
steel after immersed in 5 wt%
NH4Cl solution for 12 h, c carbon
steel after immersed in 10 wt%
NH4Cl solution for 12 h, and d
carbon steel after immersed in
20 wt% NH4Cl solution for 12 h
J Solid State Electrochem (2017) 21:1373–1384 1377

Fig. 3 EIS measurement of carbon steel at various overpotentials in a–c 5 wt% NH4Cl solution, d–f 10 wt% NH4Cl solution, and g–i 20 wt% NH4Cl
solution. Frequency given is in hertz. The absolute potentials referred in the figures are w.r.t. RHE

increases with overpotential; and (iv) the overall imped- to four especially for 10 and 20 wt% NH4Cl concentration.
ance at any overpotential decreases with an increase in For 5 wt% NH4Cl, three capacitance loops were observed
the concentration of NH4Cl, indicating that dissolution rate at 0.05 V above OCP, while the low frequency capacitive
increases with concentration as observed in polarization loop is transformed into inductive loop for higher
measurements. The loops at higher frequency are attributed overpotentials. For 10 and 20 wt% NH4Cl solutions, two
to the EDL at metal–solution interface, while those at the capacitance loops in the higher and mid frequency regime
mid and low frequencies would arise from faradaic and and one inductance loop in low frequency regime are ob-
non-faradaic processes. It is also to be noted that with an served at 0.15 V above OCP. However, at 0.25 V above
increase in overpotential, the patterns keep on changing OCP, the number of time constants increased to four loops
with an increase in the number of time constants from three (capacitance–inductance–capacitance–inductance). At
1378 J Solid State Electrochem (2017) 21:1373–1384

0.05 V above OCP, three capacitance loops are observed suggested the following mechanisms with Fe2+ as dis-
for 10 wt% NH4Cl solution, while a negative capacitance solution species:
is observed in the low frequency regime for 20 wt% NH4Cl
k1 k2
solution. Most of the systems exhibit only a change in k3
Fe ⇄Feþ
ad ⇄Fead → Fesol
2þ 2þ
impedance with the patterns remaining the same when the k‐1 k‐2
overpotential is varied [19, 20, 29]. However, the change k4
Feþ
ad → Fesol
2þ
in observed patterns with a change in overpotential is also k5 ð6Þ
2þ* k6
reported for some systems [13, 30]. Feþ
ad⇄Fead → Fe2þ
sol
k‐5
The EIS data obtained here is validated with k6 −
Kramers-Kronig transform (KKT) to ensure the linearity, Fead þ Fe → Fead
2þ* 2þ*
þ Fe2þ
sol þ 2e
stability, and causality of the system [31–33]. The trans-
formed data maps well with the experimental results where Feþ 2þ
ad and Fead correspond to adsorbed metal
(not shown here). As the EIS data shows different pat- 2þ
species, while Fesol corresponds to metal ion dissolved
terns at different overpotentials, as well as the number in solution. Fead 2þ*
represents the intermediate adsorbed
of time constants are increased to four at higher species but also acts as a catalyst for the step k6.
overpotentials, it is difficult to analyze all the experi- Though this mechanism is more similar to Keddam et al.
mental observations via electrical equivalent circuit model [12], here we considered only three adsorbed species
(EEC) modeling approach which is commonly employed
Feþ 2þ 2þ* þ
ad , Fead , and Fead instead of four adsorbed species Fead ,
by many researchers [13, 34–36]. Thus, in this present
work, a RMA approach is employed to describe the Feþ* 2þ 2þ* 2þ
ad , Fead , and Fead . Also, the pre-passivation due to Fead

process occurring at metal–solution interface for all the is neglected as polarization measurements did not show any
concentrations through the obtained EIS data. pre-passivation behavior in the investigated potential range.
Three dissolution paths via k3, k4, and k6 are considered in
Reaction mechanism analysis this mechanism. Feþ ad may be present as unstable metal ion
complex with oxidation state +1, and Fe2þ ad may exist as oxide
In RMA, a reaction mechanism is propounded based on and hydroxide of Fe with oxidation state +2 on the carbon
the experimental observations. The impedance data is steel surface. However, the film is non-protective in nature
then simulated for the considered mechanism and corre- and does not completely passivate the surface. Fe2þ ad may also
lated with the experimental impedance data, especially exist in the form of ferrous chloride as chloride ions are pres-
with respect to the patterns obtained [15, 19–21]. For ent in the solution. The existence of oxides/oxy-hydroxides
the systems being studied, a maximum of three time with Cl atoms inclusion is also possible.
constants are observed besides a capacitive loop, which The following assumptions were incorporated in the
corresponds to EDL. Thus, a mechanism with at least development of impedance equations for the proposed
three i ntermediate adsorbed species should be mechanism: (i) linear perturbation is applied using ac
propounded to capture the experimental observations. voltage signal of small amplitude, i.e., 10 mv (rms),
Keddam et al. [12] proposed a kinetic model with four so that only linear terms are considered in the equa-
intermediate adsorbed species to capture the four time tions; (ii) the kinetic parameters are exponentially pro-
constants. Recently, Fasmin et al. [37] suggested a portional to the voltage, and the forward reaction has a
mechanism for Ti dissolution in HF medium with two positive exponent and the reverse equation has a nega-
intermediate species to capture the three time constants. tive exponent; and (iii) surface coverage is restrained to
Thus, mechanisms such as direct dissolution and those unity, employing Langmuir adsorption isotherm model
involving one or two adsorbed intermediate species to the adsorption of charged species on the surface. It
were discarded in the present work. is to be noted that consideration of other adsorption
Samide et al. [6] reported from M ssbauer spectrom- models would not lead to any additional time constant
etry studies that the main corrosion product of carbon but could match the experimental data quantitatively
steel dissolution in 0.1 M NH 4 Cl is a mixture of better [12].
ferri(III)hydrite and Fe(III)oxide-hydroxide. Also, X- In the proposed mechanism, the k3 step is a chemical reac-
ray photoelectron spectroscopy (XPS) and thermogravi- tion and is independent of voltage while the rest of steps are
metric analysis (TGA) of corroded products of carbon electrochemical reactions. The details regarding the derivation
steel in 1 M NH4Cl solution show that the dissolution of impedance equation for the given reaction scheme have
species is more likely of Fe3+ [7]. However, Fe2+ is been presented well in the literature [15, 21, 37–39], and only
more thermodynamically stable in the existing pH val- the important steps for the proposed mechanism are summa-
ue of the given system. Thus, in the present work, we rized below.
J Solid State Electrochem (2017) 21:1373–1384 1379

The unsteady-state mass balance equation for the proposed k 1 k 5 ðk 3 þ k −2 Þ

θ3ss ¼ ð17Þ
mechanism, corresponding to the adsorbed species, gives, D

dθ1 where,
τ ¼ k 1 ð1−θ1 −θ2 −θ3 Þ−k 2 θ1 −k 4 θ1 −k 5 θ1 −k −1 θ1
dt
D ¼ k −5 ðk 3 þ k −2 Þðk 1 þ k −1 þ k 2 þ k 4
þ k −2 θ2 þ k −5 θ3 ð7Þ
þ k 5 Þ−k 2 k −5 ðk −2 −k 1 Þ−k 5 ðk 3 þ k −2 Þðk −5 −k 1 Þ ð18Þ
dθ2
τ ¼ k 2 θ1 −k 3 θ2 −k −2 θ2 ð8Þ
dt The unsteady-state current density is given by,
dθ3 J ¼ nF ½k 1 ð1−θ1 −θ2 −θ3 Þ þ k 2 θ1 þ k 4 θ1 þ k 5 θ1
τ ¼ k 5 θ1 −k −5 θ3 ð9Þ
dt
þ 2k 6 θ3 ð1−θ1 −θ2 −θ3 Þ−k −1 θ1 −k −2 θ2 −k −5 θ3  ð19Þ
where τ refers to the total number of sites available per unit area;
θ1, θ2, and θ3 refer respectively the surface coverage of adsorbed Under steady-state conditions,
species Feþad , Fead , and Fead ; and (1 − θ1 − θ2 − θ3) refers the
2þ 2þ*
J ss ¼ nF ½2k 3 θ2ss þ 2k 4 θ1ss þ 2k 6 θ3ss ð1−θ1ss −θ2ss −θ3ss Þ
vacant sites available. k1,k−1, k2, k−2, k3, k4, k5, k−5, and k6 are the
rate constants for the corresponding steps, given generally as, ð20Þ

k i ¼ k i0 ebi V ð10Þ The faradaic impedance is determined by differentiating

Eq. (19) with respect to the voltage, given as,
where i = (1 to 6) for forward reactions, and i = (−1 to − 5) for 0 1−1
backward reactions. V is the overpotential. dJ @ . A
¼ Z
Here, dV F;m s

αnF
bi ¼  ð11Þ 2   3
RT dk 1 dθ1 dθ2 dθ3 dk 2 dθ1
ð1−θ1ss −θ2ss −θ3ss Þ þ k 1 − − − þ θ1ss þ k 2
6 dV dV dV dV dV dV 7
6 7
6 dk 4 dθ dk dθ dk dθ dk 7
where α is referred as transfer coefficient, having values with- 6 þ θ1ss þ k 4 1 þ 5 θ1ss þ k 5 1 − −1 θ1ss −k −1 1 − −2 θ2ss 7
6 dV dV dV dV dV dV dV 7
¼ nF 6 7
in 0 and 1; n is the number of electrons involved in a rate 6 dθ2 dk −5 dθ3 dk 6 7
6 −k −2 − θ3ss −k −5 þ 2 θ3ss ð1−θ1ss −θ2ss −θ3ss Þ 7
6 
dV dV dV dV  7
determining step; and F, R, and T are respectively the faraday 4 dθ3 dθ1 dθ3 dθ2 dθ3 dθ3 5
þ2k 6 −θ3ss −θ1ss −θ3ss −θ2ss −2θ3ss
constant, the ideal gas constant, and the temperature. Value of dV dV dV dV dV dV
b for forward reaction is considered as positive, while for ð21Þ
reverse reaction, it is considered as non-positive. Under steady
state conditions, we get, On rearrangement, we get,
0 1−1
k 1 ð1−θ1ss −θ2ss −θ3ss Þ þ k −2 θ2ss þ k −5 θ3ss @Z .A
F;m s
¼ k 2 θ1ss þ k 4 θ1ss þ k 5 θ1ss þ k −1 θ1ss ð12Þ 2 3
dθ1 dθ2
6 ðk 1 þ k −1 −k 2 −k 4 −k 5 þ 2k 6 θ3ss Þ þ ðk 1 þ k −2 þ 2k 6 θ3ss Þ 7
¼ R−1
t −nF 4
dV dV 5
k 2 θ1ss ¼ k 3 θ2ss þ k −2 θ2ss ð13Þ þ
dθ3
ðk 1 −2k 6 þ 2k 6 θ1ss þ 2k 6 θ2ss þ 4k 6 θ3ss þ k −5 Þ
dV
k 5 θ1ss ¼ k −5 θ3ss ð14Þ ð22Þ

Here, all the rate constants are estimated at the dc potential where,
(the potential w.r.t. OCP) where upon the impedance was 2 3
ð1−θ1ss −θ2ss −θ3ss Þðk 1 b1 þ 2k 6 b6 θ3ss Þ
measured. The subscript ss indicates the steady state. On solv- R−1 4 5
t ¼ nF
ing the above three equations, the respective fractional steady- þ ðk 2 b2 þ k 4 b4 þ k 5 b5 −k −1 b−1 Þθ1ss −k −2 b−2 θ2ss −k −5 b−5 θ3ss
state surface coverage are found to be as, ð23Þ
k 1 k −5 ðk 3 þ k −2 Þ
θ1ss ¼ ð15Þ Here, Rt indicates the charge transfer resistance.
D
and dθ
dθ1 dθ2
dV , dV , dV values were attained by expanding mass
3

k 1 k 2 k −5 balance equations using Taylor series and by neglecting

θ2ss ¼ ð16Þ
D higher order terms:
1380 J Solid State Electrochem (2017) 21:1373–1384

Table 1 Best fit RMA

parameters attained for carbon Parameter Values for various NH4Cl concentrations Units
steel dissolution in different
systems 5 wt% 10 wt% 20 wt%

Rsol 1.5 1.3 1.1 Ω cm2

k10 5.4 × 10−10 6.4 × 10−10 1.5 × 10−9 mol s−1 cm−2
b1 17 18.5 19 V−1
k−10 2.4 × 10−6 6 × 10−6 6 × 10−6 mol s−1 cm−2
b−1 0 0 0 V−1
k20 6.4 × 10−7 1.3 × 10−6 1.4 × 10−6 mol s−1 cm−2
b2 13 13 13 V−1
k−20 8 × 10−6 1 × 10−5 1 × 10−5 mol s−1 cm−2
b−2 -6 -6 -6 V−1
k30 4 × 10−6 6.5 × 10−6 6.5 × 10−6 mol s−1 cm−2
b3 0 0 0 V−1
k40 4 × 10−8 5 × 10−8 5 × 10−8 mol s−1 cm−2
b4 21 21 21 V−1
k50 2.4 × 10−7 1 × 10−6 2 × 10−6 mol s−1 cm−2
b5 3 1 1 V−1
k−50 8 × 10−8 1 × 10−7 1 × 10−7 mol s−1 cm−2
b−5 -5 -5 -5 V−1
k60 4 × 10−9 2 × 10−9 2 × 10−9 mol s−1 cm−2
b6 22 23 23 V−1
Y0 At OCP + 0.05 V 6 × 10−5 6 × 10−5 6.2 × 10−5 Ω−1 cm−2 sn
n1 At OCP + 0.05 V 0.9 0.91 0.91
Y0 At OCP + 0.15 V 6.9 × 10−5 6.9 × 10−5 7.8 × 10−5 Ω−1 cm−2 sn
n1 At OCP + 0.15 V 0.92 0.93 0.95
Y0 At OCP + 0.25 V 7.9 × 10−5 7.9 × 10−5 8.1 × 10−5 Ω−1 cm−2 sn
n1 At OCP + 0.25 V 0.95 0.97 0.98
τ 2 × 10−6 7 × 10−6 8 × 10−6 mol cm−2

dθ1 AF 1 J 1 þ BIG þ CGJ 1 G ¼ k 3 þ k −2 þ jωτ

¼ ð24Þ
dV D1 GJ 1 −AEJ 1 −BHG H ¼ k5
 
dθ1
E þ F1 I ¼ k 5 ðb5 −b−5 Þθ1ss
dθ2 dV
¼ ð25Þ
dV G J 1 ¼ k −5 þ jωτ
 
dθ1 The overall impedance of the system finally is given as,
H þI
dθ3 dV
¼ ð26Þ 1
dV J1 Z total ¼ Rsol þ 0 1−1 ð27Þ
@Z .A þ jωðY 0 Þn1
In the above equations, F;m s

n¼1
Table 2 Rt and CPE values estimated from RMA approach
A ¼ k −2 −k 1
NH4Cl conc. (wt%) dc overpotential (V) RMA
B ¼ k −5 −k 1 Rt Cdl
Ω cm2 F cm−2
C ¼ k 1 b1 ð1−θ1ss −θ2ss −θ3ss Þ−ðk 2 b2 þ k 4 b4 þ k 5 b5 þ k −1 b−1 Þθ1ss
5 +0.05 312.5 2.1 × 10−5
þ k −2 b−2 θ2ss þ k −5 b−5 θ3ss +0.15 36.9 3.1 × 10−5
+0.25 3.6 4.8 × 10−5
10 +0.05 250 2.3 × 10−5
D1 ¼ k 1 þ k 2 þ k 4 þ k 5 þ k −1 þ jωτ +0.15 27.6 3.4 × 10−5
+0.25 2.6 5.8 × 10−5
E ¼ k2 20 +0.05 97.1 2.4 × 10−5
+0.15 10.2 4.7 × 10−5
+0.25 1.1 6.6 × 10−5
F 1 ¼ k 3 ðb2 −b3 Þθ2ss þ k −2 ðb2 −b−2 Þθ2ss
J Solid State Electrochem (2017) 21:1373–1384 1381

Fig. 4 Simulated impedance values from RMA at various overpotentials for a–c 5 wt% NH4Cl solution, d–f 10 wt% NH4Cl solution, and g–i 20 wt%
NH4Cl solution. Frequency given is in hertz. The absolute potentials referred in the figures are w.r.t. RHE

Here, Y0, n1, and Rsol respectively indicates the constant [40, 41], we considered the following model with the assump-
phase element (CPE) parameter, the exponent of CPE, and tion that system exhibits CPE behavior due to the variation of
the solution resistance. The CPE parameters have been incor- properties along the metal surface rather than on normal to the
porated in the equation, as the EIS data shows a suppressed surface.
semicircle in the higher frequency region. The CPE values are "  n−1 #1 =n
analyzed with following Brug formula, in order to verify that 1 1
the electrical double layer capacitance (Cdl) values obtained C eff ¼ Y 0 þ ð28Þ
Rsol Rct
are consistent with the literature and the estimated values are
given in Table 2. Although different models were reported in An optimization technique, sequential quadratic program-
the literature to calculate Cdl values from CPE components ming (SQP), was used to obtain the RMA parameters, where
1382 J Solid State Electrochem (2017) 21:1373–1384

the code was written in MATLAB. The best fit RMA param- possible reason for general corrosion. If the surface is covered
eters were retrieved by reducing the subsequent residue term: with continuous oxide layer, then Cl− ions might penetrate
through the layer and induce pitting corrosion which is not
Residue the case here.
h  2  2 i The dissolution rate via two dissolutions paths (chemical
¼ ∑ ωRe Z Reexperimental −Z Rebestfit þ ωIm Z Imexperimental −Z Imbestfit path and electrochemical path) is also estimated as a function
of overpotential for various NH4Cl concentrations, and the
ð29Þ
results are shown in Fig. 6. For all NH4Cl concentrations,
the dissolution via electrochemical step is predominant at
where ωRe and ωIm are the weighing functions, considered as higher overpotential while dissolution via chemical step is
unity for RMA simulations. The best fit RMA parameters are predominant at lower overpotential. It would be attributed to
given in Table 1 and the RMA simulated impedance patterns the increase in θ3ss value with overpotential for all NH4Cl
in Fig. 4a–i. Although our simulations show that the simulated 2þ*
concentrations as the presence of Fead species facilitates
impedance are more close to the experimental data at these
the electrochemical dissolution via k6 step.
particular parameter sets, the possibility of showing a similar
The values of Cdl estimated from Yo and n1 and Rt values
match at other parameter values could not be discarded.
estimated using Eq. 23 are shown in Table 2. The exponent
Although quantitative difference appears in impedance
value of CPE (n1) is less than 1 in all the cases which might
values between the experimental and the modeled data, the
arise from the heterogeneities of the corroded surface [42–46].
proposed kinetic model reproduces the observed patterns in
The Rt values clearly show that the anodic dissolution rate
the EIS data. Only the negative capacitance observed at
increases with concentration of NH 4 Cl at any given
0.05 V w.r.t OCP for 20 wt% NH4Cl concentration could
overpotential. It would be attributed to the increase of k1, k2,
not be captured by the proposed model. The remaining fea-
and τ values with concentration as the remaining rate con-
tures such as decrease in impedance with overpotential for the
stants are not changing at higher concentrations. The increase
given NH4Cl concentration, decrease in impedance with
of dc current with an increase in dc potential as observed in
NH4Cl concentration at a given overpotential, and the number
polarization curves is also qualitatively captured by the sug-
of time constants for the given overpotential and concentration
gested model. Consideration of additional steps in the kinetic
are also well captured.
model may require explaining the EIS data quantitatively bet-
The surface coverage is estimated from these RMA param-
ter. However, the main aim of this present work is to propose a
eters for the three adsorbed species. The results show that the
kinetic model with minimum number of parameters that could
surface coverage of both Feþ 2þ
ad and Fead is negligible though explain the carbon steel corrosion behavior adequately well.
they are increasing with concentration as well as with
overpotential. It suggests that the formation and dissolution
of these adsorbed species are more dynamic in nature. At any
given overpotential and concentration, the surface is covered
with mostly Fe2þ*ad species. The maximum surface coverage
obtained is 0.26 for 20 wt% NH4Cl concentration at 0.25 V
above OCP as shown in Fig. 5. It suggests that the surface is
not covered with thin film of oxide which would be the

*
Fig. 5 Surface coverage of Fe2þ ad species as a function of dc Fig. 7 Raman spectra of the corrosion products obtained from carbon
overpotential for various concentrations of NH4Cl steel dissolution in 5, 10, and 20 wt% NH4Cl solutions
J Solid State Electrochem (2017) 21:1373–1384 1383

Fig. 6 Contribution of chemical step and electrochemical step to dissolution rate at various overpotentials for a 5 wt%, b 10 wt%, and c 20 wt% NH4Cl
solutions

Raman spectroscopy measurements present in air during sample preparation. Hence, in order to
understand the reactions occurring at the carbon steel–electro-
The corrosion products formed on the surface were analyzed lyte solution interface better and to support our observations
using Raman spectroscopy measurements, and the recorded made from our measurements, other relevant in situ methods
Raman spectra are shown in Fig. 7 for various concentrations such as electrochemical tunneling microscopy could be
of NH4Cl. The spectra exhibit peaks at 223, 289, 404, 492, employed.
and 609 cm−1 in addition to one shoulder peak at 245 cm−1 for
all NH4Cl concentrations. All these peaks correspond to he-
matite (α-Fe2O3) [47, 48]. The peak observed at 1317 cm−1 is Conclusion
shifted to lower wavenumber (1300 cm−1) when the NH4Cl
concentration increases from 5 to 20 wt% which may be due The anodic dissolution of carbon steel in various NH4Cl con-
to change in crystallinity or grain size of the corrosion product centrations is investigated. The dissolution rate increases with
[49]. The existence of γ-FeOOH (lepidocrocite) in corrosion an increase in NH4Cl concentration within the range investi-
products could not be discarded as the Raman spectrum of the gated. The polarization measurements reveal that general cor-
same exhibits peak at 1307 cm−1. Although the dissolution rosion occurs on the surface. The surface morphology images
species is Fe2+, it would be oxidized into Fe3+ due to the of the carbon steel surface corroded with various NH4Cl con-
presence of dissolved oxygen in electrolyte solution. Other centrations also confirms the same. EIS measurements were
possible reason is oxidation of Fe2+ species due to the oxygen carried out at different overpotentials, and the results show a
1384 J Solid State Electrochem (2017) 21:1373–1384

maximum of four time constants at higher NH4Cl concentra- 18. Bard AJ, Faulkner LR (2001) Electrochemical methods: fundamen-
tals and applications, 2nd edn. John Wiley & Sons, New York
tions while only three time constants were observed at 5 wt%
19. Prasad YN, Kumar VV, Ramanathan S (2009) J Solid State
NH 4Cl. Also, the patterns observed are changing with Electrochem 13:1351–1359
overpotential. Thus, a detailed reaction mechanism is 20. Venkatesh RP, Ramanathan S (2010) J Appl Electrochem 40:767–
employed to understand the dissolution behavior of carbon 776
steel. The analysis shows that the dissolution occurs via three 21. Maddala J, Krishnaraj S, Kumar VV, Ramanathan S (2010) J
Electroanal Chem 638:183–188
dissolution steps which involve both chemical and electro-
22. Trueba M, Trasatti SP (2010) Mater Chem Phy 121:523–533
chemical paths, and the dissolved species is Fe2+. The contri-
23. Kuo HS, Chang H, Tsai WT (1999) Corros Sci 41:669–684
bution of chemical reaction to the overall dissolution rate is 24. McCafferty E (2010) Introduction to corrosion science, 1st edn.
significant at lower overpotential, while the contribution from Springer, New York
electrochemical steps is significant at higher overpotential. 25. Cáceres L, Vargas T, Herrera L (2009) Corros Sci 51:971–978
Although the suggested model predicts the anodic dissolution 26. Lin B, Hu R, Ye C, Li Y, Lin C (2010) Electrochim Acta 55:6542–
6545
behavior qualitatively well, consideration of additional steps
27. Mao X, Liu X, Revie RW (1994) Corros Sci 50:651–657
may require to describe the EIS data quantitatively as well.
28. Zaid B, Saidi D, Zaid AB, Hadji S (2008) Corros Sci 50:1841–1847
29. Venkatesh RP, Ramanathan S (2010) J Solid State Electrochem 14:
Acknowledgements This work was supported by DST-SERB, India 2057–2064
(SERB/F/1365/2014-15). We acknowledge Central Instruments Facility 30. Strměnik D, Gaberšček M, Pihlar B, Kočar D, Jamnik J (2009) J
of Indian Institute of Technology, Guwahati, India for providing facility Electrochem Soc 156:C222–C229
for FESEM and Raman spectroscopy analysis.
31. Agarwal P, Orazem ME (1995) J Electrochem Soc 142:4159–4168
32. Boukamp BA (1995) J Electrochem Soc 142:1885–1894
33. Macdonald MU, Real S, Macdonald DD (1986) J Electrochem Soc
References 133:2018–2024
34. Lu J, Garland JE, Pettit CM, Babu SV, Roy D (2004) J Electrochem
Soc 151(10):G717–G722
1. Alvisi PP, Lins VDFC (2008) Eng Fail Anal 15:1035–1041
2. Speight J (2015) Fouling in refineries. Elsevier Science and 35. Cho BJ, Venkatesh RP, Kwon TY, Park JG (2013) Intl J
Technology, United Kingdom Electrochem Sci 8:4723–4734
3. Toba K, Suzuki T, Kawano K, Sakai J (2011) NACE Intl 67: 36. Freitas S, Malacarne MM, Romao W, Dalmaschio GP, Castro EVR,
0550051 (1-7) Celante VG, Freitas MBJG (2013) Fuel 104:656–663
4. Toba K, Ueyama M, Kawano K, Sakai J (2012) NACE Intl 68: 37. Fasmin F, Praveen BVS, Ramanathan S (2015) J Electrochem Soc
1049–1056 162(9):H604–H610
5. Forsen O, Aromaa J, Tavi M (1993) Corros Sci 35:297–301 38. Macdonald DD, Real S, Smedley SI (1988) Urquidi-Macdonald. J
6. Samide A, Bibicu I, Rogalski MS, Preda M (2004) J Radioanal Electrochem Soc 135:2410–2414
Nucl Chem 261:593–596 39. Bojinov M (1996) J Electroanal Chem 405:15–22
7. Tutunaru B, Samide A, Negrila C (2013) J Therm Anal Calorim 40. Hirschorn B, Orazem ME, Tribollet B, Vivier V, Frateur I, Musiani
111:1149–1154 M (2010) Electrochim Acta 55:6218–6227
8. Heusler KE (1958) Z Elektrochem 62:582 41. Germain PS, Pell WG, Conway BE (2004) Electrochim Acta 49:
9. Bockris JOM, Drazic D, Despic AR (1961) Electrochim Acta 4: 1775–1788
325–361 42. Sadkowski A (2000) J Electroanal Chem 481:222–226
10. Bockris JOM, Kita H (1961) J Electrochem Soc 108:676–685
43. Sadkowski A (2000) J Electroanal Chem 481:232–236
11. Bockris JOM, Drazic D (1962) Electrochim Acta 7:293–313
44. Pajkossy T (2005) Solid State Ionics 176:1997–2003
12. Keddam M, Mattos OR, Takenouti H (1981) J Electrochem Soc
128:257–266 45. Hirschorn B, Orazem ME, Tribollet B, Vivier V, Frateur I, Musiani
13. Li W, Nobe K, Pearlstein AJ (1990) Corros Sci 31:615–620 M (2010) J Electrochem Soc 157:C452–C457
14. Macdonald JR, Barsoukov E (2005) Impedance spectroscopy: the- 46. Hirschorn B, Orazem ME, Tribollet B, Vivier V, Frateur I, Musiani
ory, experiment and applications, 2nd edn. Wiley, New Jersey M (2010) J Electrochem Soc 157:C458–C463
15. Orazem M, Tribollet B (2008) Electrochemical impedance spec- 47. Oh SJ, Cook DC, Townsend HE (1998) Hyperfine Interact 112:59–
troscopy. John Wiley and Sons, New Jersey 65
16. Macdonald DD (2006) Electrochim Acta 51:1376–1388 48. Adhyapak PV, Mulik UP, Amalnerkar DP, Mulla IS (2013) J Am
17. Zhang XG (1996) Corrosion and electrochemistry of zinc, 1st edn. Ceram Soc 96(3):731–735
Plenum, New York 49. Wang A, Haskin LA, Jolliff BL (1998) Lunar Planet Sci Conf 29