You are on page 1of 11

Journal of The Electrochemical Society, 164 (13) C717-C727 (2017) C717

Kinetic Model of Aluminum Behavior in Cement-Based Matrices


Analyzed by Impedance Spectroscopy
Sylvie Delpech, a,z Céline Cannes,a Nicole Barré,a Quan Thuan Tran,a Clément Sanchez,a
Hugo Lahalle,b David Lambertin,b Sandrine Gauffinet,c and Céline Cau Dit Coumesb
a Institut de Physique Nucléaire, CNRS, Univ. Paris-Sud 11, 91406 Orsay Cedex, France
b CEA, DEN, DTCD, SPDE, F-30207 Bagnols-sur-Cèze cedex, France
c UMR6303 Laboratoire Interdisciplinaire Carnot de Bourgogne, Université de Bourgogne Dijon, Faculté des Sciences
Mirande, 21078 Dijon cedex, France

Aluminum reactivity in cement-based matrices, generally used for conditioning low- and intermediate-level radioactive wastes, is
a problem due to dihydrogen produced by the corrosion reaction which depends on the pH of the pore solution. Electrochemical
impedance spectroscopy has been used to propose a mechanism for aluminum corrosion in cementitious matrices based on ordinary
Portland cement (OPC, pH ≈ 13) or magnesium phosphate cement (MKP, pH ranging between 4 and 9) containing or not LiNO3 as
a corrosion inhibitor. The fit of impedance diagrams recorded on aluminum electrode as a function of time has been realized using
electrical parameters to model the cement and kinetic constants for the faradaic impedance. The so determined kinetic constants
are used to calculate the corrosion current and the dihydrogen production as a function of time. Comparison of these results with
experimental measurements of dihydrogen release obtained by gas chromatography shows a very good agreement except in the MKP
matrix containing LiNO3 . In the OPC matrix, dihydrogen production is between 500 and 1000 times higher than in the MKP matrix,
which put in evidence the significant benefit in using MKP cements for aluminum encapsulation.
© The Author(s) 2017. Published by ECS. This is an open access article distributed under the terms of the Creative Commons
Attribution 4.0 License (CC BY, http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse of the work in any
medium, provided the original work is properly cited. [DOI: 10.1149/2.0211713jes] All rights reserved.

Manuscript submitted May 31, 2017; revised manuscript received September 13, 2017. Published September 23, 2017.

It has long been common practice to stabilize and solidify low- corresponding to the aluminum passivation domain.15,16 Our results
and intermediate-level radioactive wastes using cement before sending have also shown that the corrosion rate can still be reduced by
them to disposal.1–3 The wastes are confined with a cement or a mortar adding LiNO3 into the MKP matrix. The inhibition efficiency of
in a steel container which is closed. Ordinary Portland Cement (OPC) LiNO3 against aluminum corrosion has already been demonstrated
is one of the most widely used materials for conditioning radioactive in OPC material17 and explained by the formation of an insulating
wastes. Indeed, it is inexpensive, it can be easily supplied and, after layer deposited onto the aluminum surface. The authors, Matsuo
hydration, it can exhibit high stability over time.4 et al., have shown by X-ray diffraction analysis that the protective
However, wastes produced by nuclear activities are very diverse layer comprises LiH(AlO2 )2 .5H2 O, Al2 O3 .3H2 O and Al2 O3 phases.
and, under certain circumstances, can chemically react with cement- The objective of this work is to determine by electrochemical tech-
based phases or aqueous interstitial solution, thus reducing the quality niques the aluminum redox mechanism and to measure the corrosion
of the conditioning material. For instance, the dismantling of old nu- rate as a function of the mortar composition. For that purpose, we have
clear reactors generates a large volume of low and intermediate-level selected from our previous work the worst and the best matrices, re-
wastes, some of them containing metallic aluminum. Aluminum is a spectively based on Ordinary Portland Cement and magnesium phos-
reactive amphoteric metal, forming a protective oxide layer on contact phate cement with addition of lithium nitrate.14 The electrochemical
with air or water. This layer is generally stable in the pH range 4–9.5–7 analysis consists in open circuit potential (OCP) and electrochemical
However, in a strongly alkaline medium, such as that encountered impedance spectroscopy (EIS) measurements at OCP. At Al electrode,
in conventional cementitious materials based on OPC, this layer is because a corrosion reaction occurs, OCP corresponds to the corro-
soluble, resulting in continued corrosion with associated liberation sion potential. This electroanalysis gives qualitative indications on the
of dihydrogen. Then, the use of Portland cement to encapsulate large Al corrosion as a function of the cementitious material. Quantitative
amounts of aluminum is prohibited. The formation of dihydrogen con- results can also be obtained by fitting EIS diagrams using kinetic pa-
tributes to increase the pressure in the container which is able to crack. rameters which allow the calculation of corrosion rates. Finally, H2
It is the reason why the main criterion for metallic waste management volumes are calculated and compared to the volumes measured by gas
is based on the volume of dihydrogen production. chromatography in Ref. 14.
In this context, our work concerns the aluminum reactivity in
cement-based matrices for conditioning radioactive wastes. To over-
come the Al corrosion, the strategy proposed consists in (i) replacing
OPC by an alternative mortar leading to a pore solution with an ap- Experimental
propriate pH value, and (ii) adding a corrosion inhibitor to the mortar Reagents and mortars preparation.—Portland cement (CEM I
formulation. Indeed, several inorganic or organic compounds, such as 52.5 PM-ES grade according to European standard EN 197-1) was
LiNO3 , have already been proved to inhibit efficiently the aluminum mixed with demineralized water to produce a cement paste with a
corrosion in alkaline solution.8–13 water/cement ratio of 0.4. The characteristics of this cement is reported
Recently, we have confirmed that aluminum exhibits a very high in the Reference 14.
corrosion rate in an OPC matrix (released dihydrogen rate reach- Magnesium phosphate cement was prepared in a Turbula blender
ing up to 40 L.day−1 .m−2 of Al which is equivalent to about by mixing the following powders for 20 minutes: 12.64 g of MgO
4 mm.year−1 of aluminum corrosion) while a very low corro- (M.A.F. Magnesite); 42.66 g of KH2 PO4 (VWR, purity ≥ 98%); 55.29
sion rate has been measured by gas chromatography in a mag- g of fly ash (EDF, Chemical composition (wt%): CaO (5.1%), SiO2
nesium phosphate cement-based (MKP) mortar (less than 0.026 (51.5%), Al2 O3 (25.2%), Fe2 O3 (5.8%), MgO (1.8%), Na2 O (0.4%),
L.year−1 .m−2 of Al, equivalent to a corrosion of 7 nm.year−1 of K2 O (1.3%), TiO2 (1.3%), P2 O5 (0.98%), SO3 (0.66%)); 55.29 g of
aluminum).14 This remarkable decrease in reactivity can be related sand (SIBELCO MIOS 0.1–1.2); 1.66 g of H3 BO3 (VWR, purity ≥
to the nearly neutral pH values (4 – 9) of MKP pore solution, 99.5%); 1.66 g of Na5 P3 O10 (VWR, purity ≥ 99%). For MKP-LiNO3
mortar, 1.11 g of LiNO3 (Sigma, purity ≥ 99%) was also mixed with
z
E-mail: delpech@ipno.in2p3.fr the powders.

Downloaded on 2017-11-09 to IP 190.111.30.254 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
C718 Journal of The Electrochemical Society, 164 (13) C717-C727 (2017)

The fly ash reacts, but to a very small extent. The water-to-
cementitious materials ratio, defined as water/(MgO + KH2 PO4 ), is
equal to 0.56. This is slightly above the chemical water demand of the
binder (the minimum amount of water needed for complete hydration
of the cement) (w/c = 0.51).
Magnesium phosphate mortars were prepared with an anchor stir-
rer according to the following protocol:

1) start with demineralized water (31.18 g),


2) add premixed powders while stirring at low speed (≈100 rpm)
and
3) mix for 5 minutes at faster speed (≈200 rpm).

The mixture is immediately poured in the electrochemical cell, no


pre-conditioning of the device is done before measurements.

Electrodes, electrochemical set-up and apparatus.—A three-


electrode cell was used for the electrochemical measurements: a work-
ing electrode (aluminum foil, S = 0.6 cm2 Goodfellow purity 99%
or platinum wire, S = 0.3 cm2 Goodfellow purity 99.95%), a counter
electrode (platinum wire, S = 0.3 cm2 ) and a quasi-reference electrode
(platinum wire, S = 0.3 cm2 ). The aluminum foils were cleaned with
ethanol in an ultrasonic bath during 5 minutes and dried in air while
the platinum foils were heated until bright-red color with a blowtorch.
The measurement setup consisted of a potentiostat (Princeton Ap- Figure 1. Calculation of Tafel curves in the case of metallic corrosion in
plied Research model 263A) connected to a frequency response an- media with various electrolyte resistances (A- 0 ; B- 1 ; C- 100 ).
alyzer (Solartron model SI 1255) and driven with Solartron ZPLOT
software. EIS spectra were recorded at the open circuit potential with
a 10 mV amplitude. Frequency ranged between 105 Hz to 0.1 Hz, with
10 frequencies values per logarithmic decade. The analysis and the trix as an electrolyte is that the contribution of the cement to the elec-
fit of the diagrams were performed using a homemade software de- trochemical impedance cannot be attributed only to the electrolyte
veloped in the laboratory operating with Kaleidagraph (version 4.03) resistance and the double layer capacitance. The physico-chemical
interface. and electrical properties of the cement have an influence on the elec-
EIS measurements have been done only in steady-state condi- trochemical impedance.
tions in order to be able to do the mathematical analysis. Recently To estimate and to model the contribution of the cement on the
Dynamic Electrochemical Impedance Spectroscopy (DEIS) has been electrochemical impedance, EIS experiments were realized at OCP
done because this technic enables continuous impedance monitor- using a platinum (inert) electrode in water at pH 13, in OPC and in
ing of non-stationary processes.18 For our experiments we consider MKP paste(s) (Figure 2a). By comparison with diagrams recorded
that the stationarity of the process is reached during the whole mea- in aqueous solution at pH 13, the diagrams obtained in the cement-
surement period (about 20 minutes). Nevertheless, an evolution of based matrices present supplementary impedance at high frequencies
impedance can be observed during the setting phase of the mortar due which represents the influence on the impedance of the conduction
to a strong modification of the ionic conduction in the “electrolyte”. It through the cement-based matrices. At low frequencies, the shape
is the reason why no impedance diagrams have been analyzed during of the diagrams is characteristic of the response of an inert working
the first 24 hours. electrode with a capacitive behavior because the faradaic impedance
Tafel representations are generally used to study corrosion pro- Zf tends to infinity.
cesses and to determine by extrapolation of Tafel slopes the corrosion The impedance related to the cement paste has already been de-
current. In the present study, we prefer to use only EIS technic be- scribed by several authors19–27 and electrical models have been pro-
cause (i) Tafel extrapolation requires the scan of a wide potential posed. Some authors propose electrical circuits with one or two re-
range around OCP which modifies the electrode/mortar interface and sistance//capacitance circuits in series or nested.19–24 In our study,
(ii) errors can be done when the electrolyte resistance is large which to simulate the mortar contribution, we have retained the electrical
is observed when mortars are used as electrolytes. Figure 1 shows model proposed by Song,25 Cabeza26 and Cruz27 based on three par-
the calculation of the corrosion current from Tafel extrapolations with allel branches (Figure 2b) characteristic of the three kinds of conduc-
electrolyte resistances equal to 0 (curve A) 1 (curve B) and 100  tive paths described in a mortar.25 One leg of the circuit is constituted
(curve C) respectively. The plateau observed in the anodic branch of a resistance (RCEM1 ) modelling the impedance of the continuous
leads to determine the corrosion current when the electrolyte resis- conductive paths (continuously connected micro-pores in the mortar),
tance is low. We observe that the cathodic branch is modified with the second one is a capacitance (CPECEM1 ) modelling the capacitance
the ohmic drop. When the electrolyte resistance increases, it becomes across the mortar, and the third one, constituted of a resistance and a ca-
impossible to measure the plateau and then impossible to measure the pacitance in series (RCEM2 and CPECEM2 ), is modelling the impedance
corrosion current. In the conditions of curve C, despite a large scan of the discontinuous conductive paths (discontinuously connected
of potential, it is not possible to measure the corrosion current. In this micro-pores). As cementitious matrices are complex porous mate-
case, we observe that the corrosion current determined by extrapola- rial, constant phase elements (CPE), as described by MacDonald28
tion of the Tafel slopes (Iexp corr ) is strongly different of the theoretical and Brug et al.,29 have always been used instead of pure capacitance
one (Icorr ). Acquisition of Tafel curves in a mortar which presents a to take into account the heterogeneity of the mortar/metal interface.
high electrolyte resistance is not suitable. The fit of these diagrams has been realized (lines of Figure 2a)
using the electrical circuit given Figure 2b and the values reported in
Table I. The conduction through the matrix is then modelled by the
Results and Discussion
three parallel legs,25–27 the electrolyte conductivity and the charge
Contribution of cement to electrochemical impedance studied transfer by the electrolyte resistance Re and the double layer ca-
with a platinum electrode.—The particularity of using a cement ma- pacitance in parallel with the charge transfer resistance (which is

Downloaded on 2017-11-09 to IP 190.111.30.254 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
Journal of The Electrochemical Society, 164 (13) C717-C727 (2017) C719

Figure 2. (a) EIS diagrams obtained on Pt working electrode at OCP in aqueous solution at pH 13, OPC and MKP (t = 52 days) cements. Points: experimental
results. Lines: calculated impedances. (b) Electrical equivalent circuit: Re , electrolyte resistance, CPEcem1 and Rcem1 first constant phase element and resistance
related to the mortar, CPEcem2 and Rcem2 second constant phase element and resistance related to the mortar, CPEcd constant phase element related to the double
layer capacitance, Rt charge transfer resistance (considering equal to faradaic impedance in this case).

equivalent to the faradaic impedance measured at OCP on an inert reaction. Two modes of corrosion can be observed: aqueous corrosion
electrode).30 The calculation of the diagrams using this electrical cir- and corrosion by dioxygen dissolved in water. The measurement
cuit shows a good agreement with the experimental measurements. of OCP leads to propose a corrosion mode. For instance, in the
Then, we have chosen the three parallel branches to model the con- case of aqueous corrosion of Al, the corrosion potential is ranging
duction in the cement-based matrix in the subsequent stage of this between the standard potentials of the redox systems Al(III)/Al
study. and H2 O/H2 (g) or close to the cathodic limit. The cathodic limit
The values of CPEcem1 and CPEcem2 determined by fitting the EIS of cement-based matrices, corresponding to the reduction of water,
diagrams have the same order of magnitude because, whether by the has been experimentally determined by cyclic voltammetry on
two paths, they characterize the same mortar. The values of pcem1 and a Pt working electrode (Figure 3). Considering the determined
pcem2 being close to 1, these components are close to pure capacitances. electroactivity limits in the cement-based matrices, we can define a
high corrosion zone which corresponds to OCP values lower than
Electrochemical measurements on Al electrode.—OCP −0.9 V/Pt. Indeed, in this case, the formation of dihydrogen is
measurements.—OCP measured on working electrode depends characterized by gas evolution. Between the cathodic limit and the
on the reactions involving at the electrode. When an electrode is middle of the electroactivity range, a corrosion reaction occurs with a
inert, OCP is ranging between the two electroactivity limits of the lower kinetic, the vapor pressure of dihydrogen decreasing when the
electrolyte which is mainly water in our case. Between these two potential increases.
limits the electrolyte is stable under atmospheric pressure at room Figure 4 presents OCP measured on Al electrode in the 3 cements
temperature and in equilibrium with H2 and O2 which pressures are as a function of time. In all the mortars considered for this study, we
lower than 1 atm. Beyond these limits, water is no more stable but observe that OCP of aluminum electrode is close to the cathodic limit
oxidized to dioxygen in the anodic side and reduced to dihydrogen of water which indicates that the main corrosion of aluminum is an
in the cathodic side and the vapor pressure of these gases tends to aqueous corrosion with dihydrogen production. Anyway, aluminum
1 atm.5 When a corrosion reaction occurs, OCP corresponds to the shows specific redox reactivity depending on the matrix. Indeed, in
corrosion potential and it is ranging between the two redox potentials the Portland cement, the OCP measured on aluminum working elec-
characteristic of the two half-reactions governing the corrosion trode increases with time until an almost constant value of −1.3 V/Pt

Table I. Set of electrical values introduced for the simulation of impedance diagrams obtained on Pt working electrode.

CPEcem1 CPEcem2 CPEcd

Re () Kcem1 (F.s(p-1) ) pcem1 Rcem1 () Kcem2 (F.s(p-1) ) pcem2 Rcem2 () Kcd (F.s(p-1) ) pcd Zf ()
Aqueous solution (pH = 13) 8 - - - - - - 5.2e-5 0.95 3e5
OPC (52 days) 2000 3.5e-6 0.95 1500 4e-5 0.95 1200 3.7e-5 0.91 2.6e5
MKP (52 days) 1100 6e-6 0.9 1600 1.1e-5 0.9 2000 5e-5 0.92 8e5

Downloaded on 2017-11-09 to IP 190.111.30.254 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
C720 Journal of The Electrochemical Society, 164 (13) C717-C727 (2017)

tion of time. These spectra present the same shape: an impedance at


high frequencies which is attributed to the mortar contribution (as ob-
served previously on Pt electrodes), followed by a second impedance
at intermediate and low frequencies which is attributed to the process
of corrosion. At low frequencies in OPC paste, a straight line with
a phase angle of 45◦ is observed, which is characteristic of diffusion
impedance. Based on the OCP analysis, we have shown that corrosion
by dioxygen is not the main phenomenon involving in the mortar.
Then, the only compound involved in the aqueous corrosion process
and able to diffuse in the cement-based matrix is H2 O. We recall that
the property of cement is to be hydrated, then the quantity of free
water in the porosity can be very low and time-varying. This behavior
clearly indicates a faradaic process at the working electrode character-
istic of the corrosion reaction, which is in agreement with the previous
observations of OCP evolution. The impedance diagrams observed in
OPC can be attributed to a corrosion reaction without passivation
process.
Figure 3. Cyclic voltammograms recorded on Pt working electrode at 100 Figures 5b and 5c present the diagrams obtained on the aluminum
mV/s in OPC; MKP and MKP+LiNO3 mortars during the first hour of contact. electrode respectively in MKP and MKP-LiNO3 . The magnitude of
Grey zone = “high corrosion zone”. the impedance starts to decrease during the first days before increases
again. Respectively 42 and 12 days are necessary to reach the same
of impedance magnitude as initial time. From that time, the spectra
tend to a vertical line corresponding to a capacitive behavior. These
after 30 days of encapsulation in the mortar. This potential is com- results are in good agreement with the OCP measurements: a corro-
prised within the potential range of high corrosion. Aluminum is then sion of aluminum occurs during the first days of immersion of Al in
continuously corroded at every age of the cement and dihydrogen is MKP containing or not LiNO3 and the passivation appears later. The
always produced. In the magnesium phosphate cement paste with and pore solution pH of the MKP mortar increases from 4 to 9 with ongo-
without LiNO3 (MKP + LiNO3 and MKP), OCP is lower than −0.9 V ing hydration.14 We assume that in this environment, the oxide layer
/ Pt at the beginning of immersion and then increases to reach values which initially covers the aluminum electrode is firstly dissolved,
higher than −0.9 V / Pt after about 15 days. After this time, aluminum which explains the decrease in the impedance observed during the
is protected against corrosion and the production of dihydrogen is first days of immersion in the two MKP matrices. In MKP without
strongly reduced. Initially, aluminum is covered by an aluminum ox- LiNO3 , when pH reaches 6–8, the layer of aluminum oxide is again
ide layer. The solubility of this layer depends on pH.5 The lowest stabilized which limits the corrosion. In MKP with LiNO3 , the alu-
solubility is observed in the pH range of MKP. Thus, MKP is intrin- minum oxide layer may be gradually replaced by a layer of lithium
sically favorable to stabilize a protective coated aluminum layer and aluminate hydrate, as previously observed by Matsuo et al.17 We note
limit the corrosion of aluminum. Furthermore, we observe a delay be- that no diffusion impedance appears at the low frequencies in MKP
tween MKP and MKP-LiNO3 which could be attributed to a different matrices.
nature of the protective layer. Consequently, the kinetic formation of
the layers is different. Corrosion mechanism and rate measurement.—Reaction
mechanism.—The qualitative electroanalysis has put in evidence an
EIS measurements.—Figure 5a presents the Nyquist diagrams aqueous corrosion reaction. The mechanism of this redox reaction
recorded on Al electrode in the ordinary Portland cement as a func- would include a diffusion step and the formation of a passivating
layer. The faradaic impedance is then complex and cannot be likened
only to the charge transfer resistance as we did previously in the case
of Pt electrode. The determination of the faradaic impedance requires
to propose a reaction mechanism and to analyze the associated kinetic
relations.
The different steps of the faradaic process deduced from the sev-
eral observations are a diffusion process (especially observed in the
OPC), the formation of a protective layer (its nature depends on the
presence or not of LiNO3 and its stability depends on the nature of the
matrix) and the two half-reactions involving in the corrosion process
corresponding to two charge transfer steps (oxidation of aluminum
and reduction of water). To model the kinetic of formation of the
protective layer, an adsorption/desorption model is chosen. Finally,
the following general mechanism for the corrosion of aluminum in
cement-based matrices (OPC and MKP) is proposed:
1- Charge transfer:
- oxidation of aluminum
ka1
Al + s → Al (I I I ) , s + 3e
(1 − θ) θ
- reduction of water
kc2
3H2 O + 3e → 3/2H2 + 3O H −
Figure 4. Variation of OCP measured on aluminum electrode as a function of
time in the three cementitious matrices. Grey zone = “high corrosion zone”. Cel

Downloaded on 2017-11-09 to IP 190.111.30.254 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
Journal of The Electrochemical Society, 164 (13) C717-C727 (2017) C721

Figure 5. Evolution as a function of time of impedance diagrams obtained at OCP on Al electrode embedded in (a) OPC, (b) MKP, (c) MKP-LiNO3 cement-based
matrices.

Downloaded on 2017-11-09 to IP 190.111.30.254 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
C722 Journal of The Electrochemical Society, 164 (13) C717-C727 (2017)

2- Diffusion of water The concentration depends both on time and distance. Thus, we
D have to use the Fick’s second law:
H2 O, matri x → H2 O, el
∂C(x) ∂ 2C
Cs Cel =D 2 [7]
∂t ∂x
3- Desorption step
With the initial and boundary conditions given by:
K1
Al (I I I ) , s → Al (I I I ) + s x=0 C = Cel
θ  (1 − θ)

In these equations: x = δ C = Cs
ka1 (s−1 ) and kc2 (cm.s−1 ) are the kinetic constants of charge transfer
relative to Al oxidation and H2 O reduction respectively. The definition where δ (cm) is the diffusion layer thickness.
of these constants is given by Buttler-Volmer relations (given below). 3- Desorption
K1 (s−1 ) is the kinetic constant of desorption. The kinetic equation for θ variation with time is given by:
D (cm2 .s−1 ) represents the diffusion coefficient of H2 O. dθ
Cs and Cel (mol.cm−3 ) respectively stand for the concentration of = ka1 (1 − θ) − K 1 θ [8]
dt
free water in the mortar and at the electrode.
 (mol.cm−2 ) is the number of active sites (s) at the aluminum Linearization of kinetic equations and Laplace transforms
surface. The value of , estimated considering the crystalline lattice The principle of the calculation of Zf consists in using the first-
(fcc) of aluminum and the lattice parameter (a = 4.04 Å), is close to order Taylor series approximation of the functions (if , Cel , θ) for their
2.10−9 mol.cm−2 . θ (0 < θ < 1) is the proportion of sites occupied by linearization and next Laplace transformations.30,36
Al(III). By difference, (1-θ) is the number of free sites. 1- Current
In this case, the backward electrochemical reactions are not Linearization and Laplace transform of relation 5 leads to:
considered.
i f = 3F Sba1 ka1 (1 − θ)E − 3F Sbc2 kc2 Cel E
Analytical expression of the impedance.—The calculation of the
−3F Ska1 θ − 3F Skc2 Cel [9]
faradaic impedance is realized using a method previously described
in several papers.30–35 It requires considering the kinetic equations of 2- Diffusion
all the steps of the mechanism. The resolution of Eqs. 6 and 7 has been explained by Bockris.37
Kinetic equations of each step of the corrosion mechanism The expression of Cel is given by:
1-Charge transfer
The faradaic current if1 and if2 corresponding to the two electro- Cel = −T (bc2 kc2 Cel E + kc2 Cel ) [10]
chemical steps (Al oxidation and H2 O reduction) are given as follows:
  
i f 1 = 3F Ska1 (1 − θ) [1] tanh δ jω
D
with T = √ [11]
jωD
i f 2 = −F Skc2 Cel [2] in√which ω is the pulsation and ω = 2πf (f being the frequency)
and j = −1.
 
3F   3- Adsorption/desorption step
with ka1 = k ◦1 exp (1 − α1 ) (E − E ◦1 ) = k ◦1 exp ba1 (E − E ◦1 ) Linearization and Laplace transform of relation 8 leads to:
RT
[3]
  jωθ = −K 1 θ + ba1 ka1 (1 − θ)E − ka1 θ [12]
F  
k ◦2
and kc2 = exp (−α2 ) (E − E ◦2 ) = k ◦2 exp bc2 (E − E ◦2 ) Calculation of the faradaic impedance Zf
RT
[4] To solve the equation system composed of Eqs. 9, 10 and 12, all the
in which k◦ 1 (s−1 ) and k◦ 2 (cm.s−1 ) are the charge transfer intrinsic relations are divided by if , so as to extract the faradaic impedance
constants, α1 and α2 the charge transfer coefficients, E◦ 1 (V) and E◦ 2 term Zf = E/if .
(V) the standard potentials related to the charge transfer Reactions 1 The system constituted of the relations 9, 10, 12 is solved by a
and 2, E (V) is the applied potential, R (J.K−1 .mol−1 ) the ideal gas matrix resolution:
⎛ ⎞⎛ ⎞
constant, T (K) the temperature and F (C.mol−1 ) the Faraday constant. 1/Rt −3FSka1  −3FSkc2 E/if
And the total current at the OCP (which is in this case equal to the ⎜ ⎟⎜ ⎟
A = ⎝−ba1 ka1 (1 − θ) jω + ka1 + K1 0 ⎠⎝ θ/if ⎠
corrosion potential) is given by:
bc2 kc2 TCel 0 1 + kc2 T Cel /if
i f = i f 1 + 3i f 2 = 0 [5] ⎛ ⎞
1
2- Diffusion
= ⎝0⎠ [13]
The steady-state diffusion flux of water from the matrix to the
0
electrode is known empirically through Fick’s first law:
∂C In which Rt is the global charge transfer resistance, Rt1 and
−D = J = steady-state flux where D is the diffusion
∂x Rt2 the charge transfer resistances corresponding respectively to if1
coefficient. and if2 . Their expressions are deduced from the definition (Rt =
So
(∂ E/∂ I f )C(0,t),θ(t) )30,35 and given by:
∂C
−D = kc2 Cel [6] 1 1
∂x Rt1 = and Rt2 = [14]
3F Sba1 ka1 (1 − θ) −F Sbc2 kc2 Cel
Downloaded on 2017-11-09 to IP 190.111.30.254 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
Journal of The Electrochemical Society, 164 (13) C717-C727 (2017) C723

From the variation of θ with time:



= ka1 (1 − θ) − K 1 θ = 0 [19]
dt
and from Fick’s law:
∂C ∂ 2C ∂C
= = 0 and − D = kc2 Cel [20]
∂t ∂x2 ∂x
We can deduce the following relations:
D Cs
Cel = δ
[21]
kc2 + D δ
Figure 6. Electrical equivalent circuit characteristic of the Al corrosion mech-
anism in cement-based matrices. and
ka1
1 1 1 1 θ= [22]
Rt = and = +3 ka1 + K 1
3F Sba1 ka1 (1 − θ) − 3F Sbc2 kc2 Cel Rt Rt1 Rt2
[15] Because a corrosion reaction occurs at the electrode, the charge
transfer kinetic of dihydrogen formation is correlated to aluminum
det A1 ( jω + ka1 + K 1 ) (1 + kc2 T ) oxidation. Moreover, it is the oxidation of aluminum which controls
Zf = = the global process through the adsorption step and the passivation of
det A 1
Rt
( jω + K 1 ) + R1t2 ka1 + R1t1 ( jω + K 1 ) kc2 T the electrode by solid phase formation.
[16] Then, at OCP, the global current is equal to 0 and we have:

Electrical equivalent circuit.—The faradaic impedance Zf is a i f 1 = −3i f 2 and then ka1 (1 − θ) = kc2 Cel [23]
combination of several elementary electrical components. We are Combining Equations 14 and 23, the following relation is
able, from Eq. 16, to split it up into a sum of electrical component established:
impedances. Indeed, we can demonstrate that: bc2
Rt1 = −Rt2 [24]
1 1 1 3ba1
= + [17]
Zf Z1 Z2 The Constant Phase Element impedance is defined as:27
with 1
  C P Ex = [25]
ka1 Rt2 K x ( jω) px
Z 1 = Rt1 1+ and Z 2 = [1 + kc2 T ] [18]
jω + K 1 3 The impedance of CPE has unit of . Then, the CPE magnitude
The expression of Z1 is equivalent to an electrical circuit constituted K has unit of F.s(p-1) .
of a resistance Rt1 in serie with a R//C parallel circuit with the ca- Then, all the relations required to fit the impedance diagrams have
pacitance equal to Rt11ka1 (homogeneous with Farad) and the resistance been established.
to Rt1Kk1a1 (homogeneous with Ohm). Z2 is equivalent to an electrical Experimental impedance diagrams obtained in OPC and MKP
cements with and without LiNO3 have been fitted as a function of time
circuit constituted of two elements: a resistance equal to Rt2 /3 and the using the expression of the faradaic impedance established previously
diffusion impedance equal to Rt2 kc2 T/3. This term is not homogeneous and the circuit model of the Figure 2b for the cement contribution.
with an electrical unit. It is equivalent to Warburg impedance30 and Some examples of the fits are given Figure 7 and the determined
presents the term (jω)−0.5 characteristic of a diffusion. data are gathered in Table III. By fitting the diagrams in the case of
Taking into account of the double layer capacitance (characteristic OPC (the ones presenting diffusion impedance), we have observed
of the capacitive current30–36 ) the global electrical circuit defined for that the D and Cs terms were strongly linked to each other and fits
this system is given in Figure 6. The analytical expressions of all the are similar for a given ratio (D∗ Cs ) value, regardless of the respective
electrical parameters involving in the faradaic impedance are gathered values of D and Cs . Then, we have chosen to fix one parameter and
in Table II. This table indicates clearly the connection between all the vary the other one. The term D was set to 6.10−6 cm2 .s−1 . In the
electrical components and the kinetic constants. It shows that the val- case of OPC, both diffusion and desorption steps are limiting steps.
ues of the electrical components cannot be introduced independently We observe a decrease by a factor of 100 of the diffusion parameter
to fit the impedance diagrams. with time. That can be due to a decrease of the volume of interstitial
solution available in the pores or to a decrease of the number of pores.
EIS fitting.—To fit the experimental impedance diagrams, it is Obviously, in parallel, an increase of the resistances, Re , Rcem1 and
necessary to establish the relation of Cel and θ at steady-state, these Rcem2, of the mortar is observed, the electrical conductivity in the
parameters being involved in the faradaic impedance relation. mortar being managed by the free ionic solution in the mortar. The
relative permittivity of the matrix can vary with time but that was not
observed in our experiments. Indeed, no major evolution of CPEcem1
Table II. Analytical expressions of the electrical elements involved and CPEcem2 has been evidenced with time. In some cases, low values
in the faradaic impedance. of p (0.7 or 0.5) can be obtained in porous electrodes or electrodes
with specific structures.38,39 In our case, the porosity involves in the
Analytical expressions of electrical components electrolyte and not in the electrode. The values of p determined are
Rt = 3F Sba1 ka1 (1−θ)−3F
1
Sbc2 kc2 Cel close to 1 which characterizes a capacitance behavior.
Rt1 = 1 The experimental impedance diagrams obtained in MKP mortar
3F Sba1 ka1 (1−θ)
without and with LiNO3 are not characteristic of a limitation by diffu-
Rt2 = −F Sbc21 kc2 Cel sion and their fits show that impedance depends only on the desorption
R1 = Rt1Kka1 step. We observe that the value of KD obtained by fitting impedance
1
diagrams is 50 to 100 times lower in MKP than in the OPC and this
C1 = Rt11ka1
difference increases in presence of LiNO3 . This value is in agreement
Z d = R3t2 kc2 T with the solubility of Al(III) which is 1000 times lower at pH 9 than

Downloaded on 2017-11-09 to IP 190.111.30.254 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
C724 Journal of The Electrochemical Society, 164 (13) C717-C727 (2017)

Figure 7. Experimental (points) and calculated (lines) impedance of diagrams recorded on Al working electrode at OCP as a function of time. (a) in OPC t = 23
hours, 50 d, 203 days (b) in MKP t = 10 d, 60 d, 193 days (c) in MKP+LiNO3 t = 14 d, 84 d, 207 days.

at pH 12.5 On the other hand, we do not expect a strong modification Combining the relations 1, 3, 4, 15 and 24, we can establish a rela-
of the diffusion process in the two matrices. tion between the corrosion current and the charge transfer resistance:
On the experimental diagrams no electrical contribution of the
protective layer of lithium aluminate or alumina has been specifically
identified. The major influence of this solid layer is its resistance which RT 1 1
i corr = [27]
increases with time and could be related to an increase of the thick- F Rt (n 1 (1 − α1 ) + n 2 α2 )
ness of the layer with time. Unfortunately, the electrical impedance
related to this layer cannot be separated from the cement electrical This relation is the same as that previously established by Stern
contribution. and Geary40 to be applied to Tafel plots analysis. The slope of Tafel
plots leads to determine the polarization resistance but not the charge
Corrosion rate calculation.—The corrosion current is calculated transfer resistance. In some special cases, without ohmic drop, no
using the kinetic constants determined by the impedance diagrams electrode passivation, and a limitation only due to the charge trans-
fits, the relations 1 and 2 and considering that: fer, the charge transfer resistance and the polarization resistance can
be consider as equal. However, in some papers41,42 the term Rt was
i corr = i f 1 = −3i f 2 [26] replaced by Rpol in the relation 27. Obviously, the calculation of icorr

Downloaded on 2017-11-09 to IP 190.111.30.254 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
Journal of The Electrochemical Society, 164 (13) C717-C727 (2017) C725

Table III. Set of parameters (electrical and kinetic) used to fit EIS experimental diagrams.

Al//OPC 23 hours 10 days 30 days 50 days 101 days 203 days


OCP (V) −1.756 −1.586 −1.334 −1.31 −1.291 −1.257
Re () 460 1810 3150 4400 6400 7100
CPEcem1 Kcem1 (F.s(p-1) ) 4e-6 5e-6 7e-6 1e-5 9e-6 6.5e-6
pcem1 0.9 0.9 0.9 0.9 0.9 0.9
Rcem1 () 60 700 600 1050 1450 1750
CPEcem2 Kcem2 (F.s(p-1) ) 8e-6 7e-6 1e-5 1.4e-5 9e-6 9e-6
pcem2 0.9 0.9 0.9 0.9 0.9 0.9
Rcem2 () 20 200 650 700 1050 1150
CPEcd Kcd (F.s(p-1) ) 1.6e-5 1.55e-5 2.8e-5 2.7e-5 1.65e-5 1.5e-5
pcd 0.94 0.94 0.92 0.92 0.94 0.94
K1 (s−1 ) 0.184 0.0046 0.02 0.0085 0.0072 0.0072
D∗ Cs (mol.s−1 .cm−1 ) 5.1e-12 3.72e-13 1.5e-13 1.212e-13 7.5e-14 7.2e-14
δ (cm) 0.0055 0.0011 0.00195 0.00335 0.00375 0.00385
With E◦ 1 = −2.4 V, E◦ 2 = −1 V, k◦ 1 = 9e-4 s−1 , α1 = 0.79, α2 = 0.86, S = 0.47 cm2 , T = 298 K
Al//MKP 20 hours 10 days 35 days 60 days 102 days 169 days 193 days
OCP (V) −1.164 −0.96 −0.688 −0.524 −0.469 −0.448 −0.44
Re () 22 235 410 500 680 1200 1340
CPEcem1 Kcem1 (F.s(p-1) ) 3.5e-6 4.5e-6 4.8e-6 4.5e-6 4e-6 4e-6 4e-6
pcem1 0.92 0.92 0.92 0.92 0.92 0.92 0.92
Rcem1 () 70 600 750 1100 1200 7000 7500
CPEcem2 Kcem2 (F.s(p-1) ) 4.5e-5 8e-6 5.5e-6 4e-6 5.5e-6 7.8e-6 1e-5
pcem2 0.92 0.92 0.92 0.92 0.92 0.92 0.92
Rcem2 () 10 250 250 400 500 1400 1300
CPEcd Kcd (F.s(p-1) ) 1.25e-5 0.9e-5 0.7e-5 0.67e-5 0.62e-5 0.56e-5 0.57e-5
pcd 0.94 0.95 0.93 0.93 0.92 0.95 0.935
K1 (s−1 ) 0.0006 0.00089 0.00027 0.000105 0.00004 0.000028 0.00001
With E◦ 1 = −2.4 V, E◦ 2 = −1 V, k◦ 1 = 9e−4 s−1 , α1 = 0.86, α2 = 0.85-0.87, D∗ Cs = 6e-13 mol.s−1 .cm−1 , δ = 0.005 cm, S = 0.47 cm2 , T = 298 K
Al//MKP-LiNO3 2 days 14 days 35 days 84 days 100 days 143 days 207 days
OCP (V) −0.955 −0.835 −0.735 −0.789 −0.729 −0.597 −0.476
Re () 23 310 420 800 930 1200 1920
CPEcem1 Kcem1 (F.s(p-1) ) 12e-6 4.2e-6 4e-6 3.3e-6 4e-6 3.7e-6 3.5e-6
pcem1 0.9 0.9 0.92 0.92 0.92 0.92 0.87
Rcem1 () 600 1500 2800 6300 7000 11000 22500
CPEcem2 Kcem2 (F.s(p-1) ) 1.6e-5 6.6e-6 5.9e-6 5e-6 4.2e-6 5e-6 5.8e-6
pcem2 0.9 0.9 0.9 0.9 0.9 0.92 0.87
Rcem2 () 30 300 750 1700 1900 3200 3500
CPEcd Kcd (F.s(p-1) ) 1.2e-5 1.08e-5 0.95e-5 0.75e-5 0.75e-5 0.75e-5 0.65e-5
pcd 0.92 0.92 0.92 0.92 0.91 0.91 0.91
K1 (s−1 ) 0.00017 0.000203 0.00009 0.000035 0.000025 0.000024 0.00002
With E◦ 1 = −2.4 V, E◦ 2 = −1 V, k◦ 1 = 9e-4 s−1 , α1 = 0.865, α2 = 0.87, D∗ Cs = 6e-13 mol.s−1 .cm−1 , δ = 0.005 cm, S = 0.47 cm2 , T = 298 K

using the relation 27 with Rpol values instead of Rt values leads to matrices. It shows that the corrosion current is between 100 and 800
wrong determinations. times higher in OPC than in MKP cements. The influence of LiNO3
The polarization resistance Rpol is the impedance at 0 frequency. It on the corrosion current is particularly observed during the first 50
corresponds notably to the opposite of the intensity-potential curve at days. During this period, the corrosion current is about 5 to 10 times
steady state. This data is usually determined by recording Tafel curves. higher in MKP without lithium nitrate. For longer ageing times, the
Considering the electrical equivalent circuit Figure 6, the relation of currents tend to be similar because the desorption rates become equal.
Rpol is the following one: From this time, the kinetic of the desorption step is no more dependent
on the nature of the aluminum oxide layer which covers the surface.
R pol = Re + Rcem 1 + Z f (ω = 0) [28] The analysis of the curve icorr as a function of time leads to the
Indeed, the electrical circuit of the mortar impedance is equal only determination of the volume of dihydrogen produced with time.
to Rcem1 at 0 frequency. The quantities of Al oxidized and dihydrogen produced are de-
Combining this relation with the relation of Zf (Eq. 16), we obtain: duced from the coulomb law:
  3
(ka1 + K 1 ) 1 + kc2 δ D Q = 3N Al F and N H 2 = N Al [30]
R pol = Re + Rcem1 + K 2
[29] 
1
+ kRa1t2 + R1t1 K 1 kc2 δ D
Rt
and Q = i corr dt [31]
(applying the limit of th(x) → x when x tends to 0).
The variation of Rt and Rpol in the three cements is given Figure By combining the relations 30 and 31 the corrosion rate can be
8. In MKP matrices, the polarization resistance is 5 times higher than evaluated by the volume of dihydrogen produced (lH2 .m−2 Al .s−1 ):
the charge transfer resistance. In OPC matrices this factor tends to 10. 
24.45
Figure 9 presents the variation of the corrosion current calculated v(H2 ) = i corr dt [32]
by using the Equation 27 as a function of time in the 3 cement-based 2F S

Downloaded on 2017-11-09 to IP 190.111.30.254 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
C726 Journal of The Electrochemical Society, 164 (13) C717-C727 (2017)

Figure 8. Variation of the resistances Rt and Rpol as a function of time.


Figure 10. Total volume of dihydrogen produced in liter per square meter of
aluminum calculated from kinetic constants and experimentally measured by
In which S (m2 ) is the surface area of aluminum in contact with gas chromatography as a function of time.
the cement and 24.45 the volume (in liter) of one mole of gas at
298 K.
The integration of the curve given Figure 9 leads to the calculation matography present the same order of magnitude except for the MKP
of the relation 31. This evolution is given Figure 10 for the 3 cementi- mortar with LiNO3 . Indeed, in this case, no dihydrogen is measured
tious matrices. We observe that the production of dihydrogen is very by gas chromatography while impedance measurements do not show
high in the OPC matrix, confirming the reactivity of aluminum in the a strong decrease in the corrosion current as compared with MKP.
pH range of the OPC interstitial solution. On the contrary, low values The difference observed between CPG measurements and calculation
are calculated by EIS in MKP mortar. LiNO3 contributes to decrease from impedance analysis in MKP can be due to a reaction inside the
the dihydrogen production by a factor 3. cement-based matrix or a partial adsorption of dihydrogen in the mor-
The amounts of dihydrogen measured by gas chromatography (as tar. A possible explanation is that dihydrogen produced in this mortar
described in Ref. 14) are also indicated Figure 10. The corrosion rates chemically reacts with nitrate to form nitrite or ammonium according
calculated from impedance measurements and measured by gas chro- to the following reactions:
H2 (g) + NO3 − → H2 O + NO2 − or 4H2 (g) + NO3 −
→ NH4 + + H2 O + 2OH−
These reactions are independent on the aluminum corrosion and
cannot be observed through electrochemical measurements on the
aluminum electrode. We observe that a best agreement is observed
for the short times, during the hydration period of the cement-based
matrices.

Conclusions
The behavior of aluminum has been studied in several cement-
based matrices in order to determine the corrosion kinetic. Based on
the electrochemical study, a mechanism for aluminum corrosion with
four elementary steps has been proposed, reflecting all the experi-
mental observations. By comparing the results obtained on a platinum
working electrode in the cement-based matrices and in aqueous so-
lution, an impedance characteristic of the mortars has been pointed
out.
The establishment of the analytical expression of the faradaic
impedance enables to fit the impedance diagrams using both elec-
trical parameters for the mortar contribution and kinetic constants for
the faradaic impedance. The determination of the kinetic constants
leads to calculate the aluminum corrosion current as a function of
time in the 3 matrices and to determine the production of dihydrogen.
Figure 9. Variation of the corrosion current as a function of time in the three In the case of the Portland cement, a strong dihydrogen production has
cement-based matrices. been observed whereas MKP mortar with or without LiNO3 seems to

Downloaded on 2017-11-09 to IP 190.111.30.254 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).
Journal of The Electrochemical Society, 164 (13) C717-C727 (2017) C727

be matrices with intrinsic properties of interest for metallic aluminum 15. A. S. Wagh, S. Y. Jeong, D. Singh, R. Strain, H. No, and J. Wescott, Proc. Waste
encapsulation. The difference in dihydrogen production observed in Manag. 1997 (WM’97), Tucson, USA (1997).
16. A. Covill, N. C. Hyatt, J. Hill, and N. C. Collier, Adv. Appl. Ceram., 110, 151
the two MKP matrices has thus been connected to the early age of (2011).
the aluminum-cement wasteforms. At later age, the production of di- 17. T. Matsuo, T. Nishi, M. Matsuda, and T. Izumida, J. Nucl. Sci. Technol., 32, 912
hydrogen tends to a constant value corresponding to a steady state. (1995).
Therefore, this kinetic model offers the advantage to predict the dihy- 18. J. Wysocka, S. Krakowiak, J. Ryl, and K. Darowicki, J. of Electroanal. Chem., 778,
126 (2016).
drogen volume in-situ generated in the long term. 19. M. F. Montemor, A. M. P Simoes, and M. M. Salta, Cem. Concr. Comp., 22, 175
(2000).
20. C. Andrade, M. Keddam, X. R. Novoa, M. C. Perez, C. M. Rangel, and H. Takenouti,
Acknowledgment Electrochim. Acta, 46, 3905 (2001).
21. G. Trabanelli, C. Monticelli, V. Grassi, and A. Frignani, Cem. Concr. Res., 35, 1804
This work was supported by the interdisciplinary NEEDS project (2005).
funded by ANDRA, CNRS, EDF, AREVA, CEA and IRSN. 22. M. Criado, M. Bastidas, S. Fajardo A. Fernandez-Jimenez, and J. M. Bastidas, Cem.
Concr. Comp., 33, 644 (2011).
23. R. Vedalakshmi, R. Renugha Devi, Bosco Emmanuel, and N. Palaniswamy, Mat. and
ORCID Struct., 41, 1315 (2008).
24. M. Keddam, H. Takenouti, X. R. Novoa, C. Andrade, and C. Alonso, Cem. Concr.
Sylvie Delpech http://orcid.org/0000-0002-2406-2041 Res., 27, 1191 (1997).
25. G. Song, Cem. Concr. Res., 30, 1723 (2000).
26. M. Cabeza, P. Merino, A. Miranda, X. R. Novoa, and I. Sanchez, Cem. Concr. Res.,
References 32, 881 (2002).
27. J. M. Cruz, I. C. Fita, L. Soriano, J. Paya, and M. V. Borrachero, Cem. Concr. Res.,
1. J. Li and J. Wang, J. Hazard. Mat., B 135, 443 (2006). 50, 51 (2013).
2. F. Glasser, Application of inorganic cement to the conditioning and immobilisation 28. J. R. MacDonald, Solid State Ionics, 13, 147 (1984).
of radioactive wastes, in Handbook of advanced radioactive wastes conditioning 29. G. J. Brug, A. L. G. Van Den Eeden, M. Sluyters-rehbach, and J. H. Sluyters, J.
technologies, p. 67–135 (2011). Electroanal. Chem., 176, 275 (1984).
3. N. B. Milestone, Adv. Appl. Ceram., 105, 13 (2006). 30. A. J. Bard and L. R. Faulkner, “Electrochemical methods, fundamentals and appli-
4. M. Atkins and F. B. Glasser, Waste Manag., 12, 105 (1992). cations,”, second edition, John Wiley & Sons Inc., (2001).
5. M. Pourbaix, “Atlas d’équilibres électrochimiques,”, p. 171, Gauthier-Villars & Cie, 31. S. Rouquette, D. Ferry, and G. Picard, J. Electrochem. Soc., 136, 3299 (1989).
Paris (1963). 32. S. Rouquette-Sanchez, P. Cowache, P. Boncorps, and J. Vedel, Electrochim. Acta, 38,
6. E. C. Deltombe, M. Vanleugenhaghe, and M. Pourbaix, Atlas of electrochemical 2043 (1993).
equilibria in aqueous solution, Pergamon Press, Oxford, UK (1966). 33. S. Rouquette-Sanchez and G. Picard, Electrochim. Acta, 41, 2035 (1996).
7. V. Vargel, Corrosion of aluminum, Elsevier, Oxford (2004). 34. S. Sanchez, S. Cassaignon, J. Vedel, and H. Gomez-Meier, Electrochim. Acta, 41,
8. H. B. Shao, J. M. Wang, Z. Zhang, J. Q. Zhang, and C. N. Cao, Mater. Chem. Phys., 1331 (1996).
77, 305 (2002). 35. C. Gabrielli, « Méthodes électrochimiques–Mesures d’impédances », Techniques de
9. N. H. Soliman, Corr. Sci., 53, 2994 (2011). l’ingénieur ref P2210, (1994).
10. M. A. Amin, S. S. Abd EI-Rehim, E. E. F. El-Sherbini, O. A. Hazzazi, and 36. B. Trémillon, « Electrochimie analytique et réactions en solution », Tome 2, Ed.
M. N. Abbas, Corr. Sci., 51, 658 (2009). Masson, Paris (1993).
11. A. M. Adbel-Gaber, E. Khamis, H. Abo-ElGahab, and Sh. Adeel, Mater. Chem. 37. J. O’M Bockris, “Modern electrochemistry,” Vol. 1, p 344, Plenum Press, New-York,
Phys., 109, 297 (2008). (1970).
12. O. K. Abiola, J. O. E. Otaigbe, O. J. Kio, and L. Gossipium Hirsutum, Corr. Sci., 51, 38. C. Hitz and A. Lasia, J. of Electroanal. Chem., 500, 213 (2001).
1879 (2009). 39. J. B. Jorcin, M. E. Orazem, N. Pébère, and B. Tribollet, Electrochim. Acta, 51, 1473
13. S. S. A. Reim, Hamdi H. Hassan, and Mohammed A. Amin, Appl. Surf. Sci., 187, (2006).
279 (2002). 40. M. Stern and A. L. Geary, J. Electrochem. Soc., 104, 56 (1957).
14. C. Cau Dit Coumes, D. Lambertin, H. Lahalle, P. Antonucci, C. Cannes, and 41. J. R. Scully, Corrosion, 56, 199 (2000).
S. Delpech, J. Nucl. Mater., 453, 31 (2014). 42. A. D. King, N. Birbilis, and J. R. Scully, Electrochim. Acta, 121, 394 (2014).

Downloaded on 2017-11-09 to IP 190.111.30.254 address. Redistribution subject to ECS terms of use (see ecsdl.org/site/terms_use) unless CC License in place (see abstract).

You might also like