Abstract
This study presents the results on the investigation of the corrosion behavior of carbon steel in model alkaline medium in the presence of very low concentration of polymeric nanoaggregates [
polyethylene oxide 
micelles]. The steel electrodes were investigated in chloride-free and chloride-containing cement extracts. The electrochemical measurements (electrochemical impedance spectroscopy and potentiodynamic polarization) indicate that the presence of micelles alters the composition of the surface layers (i.e., micelles were indeed absorbed to the steel surface) and influences the electrochemical behavior of the steel, i.e., the micelles lead to an initially increased corrosion resistance of the steel whereas no significant improvement was observed within longer immersion periods. Surface analysis, performed by environmental scanning electronic microscopy, energy-dispersive x-ray analysis, and x-ray photoelectron spectroscopy, supports and elucidates the corrosion performance. The product layers in the micelles-containing specimens are more homogenous and compact, presenting protective 
and/or 
, whereas the product layers in the micelles-free specimens exhibit mainly FeOOH, FeO, and 
, which are prone to chloride attack. Therefore, the increased "barrier effects" along with the layers composition and altered surface morphology denote for the initially increased corrosion resistance of the steel in chloride-containing alkaline medium in the presence of micelles.
Export citation and abstract BibTeX RIS
A well-known fact is that corrosion of steel reinforcement can cause reinforced concrete deterioration and thus affect civil structures durability.1, 2 Logically, corrosion-related issues result in a significant economic loss. For example, in the United States, the annual direct cost of bridge infrastructure corrosion was estimated to be $8.3 billion, and the indirect cost was reported to be many times higher.3
In normal conditions, the reinforcing steel embedded in cement-based materials (i.e., cement paste, mortar, and concrete) is in a passive state because of the high alkalinity 
of the pore solution and the cement paste, respectively.4, 5 However, corrosion can be initiated due to carbonation or chloride contamination.6 Carbonation can result in a 
drop (to about 8–9)7, 8 in the pore solution of the bulk cementitious matrix, leading to a general corrosion of the reinforcing steel. In the event of chloride penetration and chloride arrival at the steel/cement paste interface, localized corrosion is initiated; the local 
can drop to even below 6, resulting in fast corrosion propagation. This will cause accelerated damage of the reinforcing steel and possible rapid failure of reinforced concrete.9
To minimize the corrosion processes, various methods and techniques are used. Among them, the polymer-based organic corrosion inhibitors are widely applied because they can provide an easy handling, cost-effective corrosion prevention, delaying corrosion initiation.10–14 The organic inhibitors are typically based on mixtures of alkanolamines and amines or amino acids or alternately on organic acids.15–22 They can work either on initiation period of time (increasing chloride threshold value or reducing chloride penetration rate) or on propagation period, reducing corrosion rate.10, 23 The application of organic inhibitors were widely investigated both in concrete24, 25 and in simulated pore solution.26, 27 However, there are still conflicting opinions about the effectiveness of organic inhibitors on reinforcement corrosion protection;28–31 different mechanisms are also reported: Some investigations illustrate that organic inhibitors are able to form an adsorbed layer on the steel surface, hindering steel dissolution.32 It is also suggested that organic inhibitors can block the anodic and cathodic reactions.21, 33, 34 Other authors report that organic inhibitors decrease the chloride content and chloride diffusion in concrete.34–36
Aiming at establishing novel approaches to corrosion control, this work presents initial tests, performed within a nontraditional investigation, involving the application of polymeric nanoaggregates in reinforced concrete. The present study is part of a 
research project on self-healing of corrosion damages in reinforced concrete, by tailoring the material properties of both steel and concrete via initially admixed micelles and/or vesicles. The nanoaggregates are not meant to have inhibiting effects on the corrosion initiation in the sense that organic inhibitors work, for example, but to release certain compounds, i.e., the so called "active substance," which would modify the pore solution and the steel/cement paste interface in a way, so that restoration of the original environment takes place (e.g., release of alkaline compound that would restore 
). The release of "active substance" is foreseen to be triggered by changes in 
. Preliminary research has been conducted with several types of nanoaggregates—micelles, vesicles, and hybrid aggregates (in model solutions, in cement pastes, and in mortar).37–40 The nanoaggregates themselves are not subject to extensive presentation and discussion. However, because the presence of nanoaggregates in a reinforced cement-based system would inevitably affect the bulk matrix (which would be prior to or simultaneously with affecting the embedded steel), the influence of the incorporated nanoaggregates on the bulk properties of cement-based materials is briefly presented below. Moreover, shortly presenting the results from these initial tests clarifies the aims and need for the present investigation.
The properties of the bulk cement-based matrix, containing polyethylene oxide (PEO-block-polystyrene (PS) 
core-shell micelles with concentration of 
mixing water (which is 
per mortar weight), were previously investigated in plain (not reinforced) and in reinforced mortar specimens.37–39 It was found out that even a very low concentration of the micelles induces a significant reduction of porosity (about two times at early hydration stages) and water permeability K (m/s) (approximately 6 orders of magnitude) of the matrix. The recorded coefficient of NaCl permeability, however, was quite different for both micelles-containing and micelles-free groups. For the former (micelles-containing) group, permeability increases in the presence of NaCl 
, while it decreases for the latter (micelles-free) group 
. For the micelles-free mixture, NaCl permeability is reduced, compared to water permeability, due to well-known chloride binding mechanisms, morphological, and structural modifications of the bulk matrix.41, 42 For the micelles-containing specimens, although similar binding mechanisms and structural alterations should be taking place in the presence of NaCl, permeability increases as a result of "shrinkage" of the polymer shell (PEO) of the micelles.43 In the latter case, although NaCl permeability was lower, compared to mixtures without micelles, the remaining question is whether corrosion resistance of the steel reinforcement in the modified matrix will be increased, compared to the micelles-free specimens. The previously reported results39, 40 show a decreased corrosion initiation and propagation in the micelles-modified matrix. Further, tests on carbon steel in model alkaline solutions, where hybrid aggregates were added in a concentration of 
(hybrid formations, synthesized through a layer-by-layer adsorption of oppositely charged ployelectrolytes {PDADMAC [poly(diallyldimethyl ammonium chloride)]/PAA [poly(acrylic acid)]/PDADMAC on CaO crystals} also resulted in an improved corrosion resistance of the steel electrodes.44
These preliminary investigations did not fully reveal whether the nanoaggregates were indeed present at the steel/cement paste interface and/or on the steel surface (considering the nanosize of the formations and the variety of cement hydration/corrosion products on the steel surface). Although steel corrosion in cement-based materials was substantially studied and results were reported (including some of the present authors),45–47 the remaining fact is that the system reinforced mortar (and concrete, respectively) is highly heterogeneous and complicated by itself. Therefore, the presence of nanoaggregates (as hereby discussed) leads to even more complexity in deriving electrochemical response for the steel reinforcement or within the steel surface analysis.
To this end and in order to minimize the influence of the surrounding cement matrix, the present investigation focuses on the effect of polymeric nanoaggregates on the steel corrosion performance in model solutions, simulating the pore solution of the bulk cementitious matrix. The paper comprises two main parts: electrochemical behavior of the steel electrodes in the relevant medium and surface analysis of the steel electrodes after treatment. The results of this study will act as preliminary data for the ongoing investigation in reinforced mortar in the presence of micelles and vesicles with tailored properties.
Materials and Methods
Materials
Steel electrodes
Low-carbon steel (St37) electrodes with a surface area of 
were used; all electrodes were equally treated prior to investigation in the model medium, i.e., they were grinded with no. 500 to no. 4000 grinding papers and polished; further, just before immersion in the relevant solutions, they were cleaned with acetone.
Cement extract
The cement extract (simulated cement pore solution) was prepared from ordinary portland cement OPC CEM I 42.5N and tap water by mixing in the weight ratio of 1:1; the suspension was filtrated after 
rotation, and thus a simulated pore solution (CE) was obtained. The 
of CE is 12.6–12.9. The chemical composition (wt %) of OPC CEM I 42.5N (ENCI, NL) is as follows: CaO 63.9%; 
20.6%; 
5.01%; 
3.25%; 
2.68%; 
0.65%; and 
0.3%. The chemical composition of the CE [derived chemically by inductively coupled plasma (ICP) analysis] is as follows: Ca, 
; K, 
; Na, 
; Al, 
; and Fe, 
. The originally received CE (modified and prepared) was the environmental medium for testing the electrochemical behavior of the steel electrodes. The CE modification was in terms of adding micelles (details specified below) and/or NaCl 
, thus ending up with four solution types: CE only (as control case); 
; 
; and 
. The so received solutions determine the samples groups and designation, which is provided further below.
Polymeric nanoaggregates
The nanoaggregates employed in this study were prepared from 
diblock copolymer. The copolymer was synthesized by atom transfer radical polymerization (ATRP), employing the macroinitiator technique.48 
micelles in aqueous media were obtained by dialysis method. The aqueous solution of 
micelles (micelles concentration of 
), was added to a part of the cement extracts (specimens CN and CNC below), thus 
of the micelles were present in chloride-free and chloride-containing cement extracts. The 
micelles are amphiphilic formations, presenting a hydrophilic PEO shell and a hydrophobic PS core. Dynamic light scattering (DLS) measurement was performed for the as-received micelles solution in order to determine their hydrodynamic radius and stability. Figure 1 presents the DLS measurement results at different time intervals (
, 
and 
), showing an apparent average hydrodynamic radius of 
at all time intervals.
Figure 1. DLS of the micelles solution as received.
Sample designation
As aforementioned, there were totally four testing solutions in this study. The samples designations are control groups "C" (without micelles) and "CN" (with micelles) and corroding groups "CC" (without micelles) and "CNC" (with micelles) (for the "corroding" cases, NaCl was added to the CE in concentration of 
).
Methods
Electrochemical measurements
The hereby employed electrochemical methods are: electrochemical impedance spectroscopy (EIS) and potentiodynamic polarization (PDP). A common three-electrode electrochemical cell [saturated calomel electrode (SCE) as reference] was used, and the measurements were performed after open-circuit potential (OCP) stabilization for all cells (the electrochemical measurements (as well as OCP readings) were performed on at least three replicates per sample type per age). The PDP measurement was performed in the range of 
vs OCP at scan rate 
. The EIS measurements were carried out in the frequency range of 
by superimposing an ac voltage of 
. The used equipment was EcoChemie Autolab-Potentiostat PGSTAT30, combined with FRA2 module, using GPES and FRA interface.
Surface analysis
For the relevant morphological aspects, qualification-quantification, and semiquantification analyses of the product layers (a mixture of iron oxides/hydroxides, micelles, and oxides/hydroxides of alkaline metals from the cement extract), formed on the steel surface in each medium, the steel electrodes were examined via scanning electronic microscopy (SEM), using environmental SEM (ESEM Philips XL30), coupled with energy dispersive-ray analysis (EDX), operating at an accelerating voltage of 
. The EDX analysis was conducted in a local area of 
at a magnification of 500 times). The EDX results were the average value of three locations per each sample. Further, surface analysis was performed by X-ray photoelectron spectroscopy (XPS). The XPS measurements were carried out in the UHV chamber of an electron spectrometer ESCALAB-MkII (VG Scientific) with a base pressure of about 
(during the measurement 
). The photoelectron spectra were obtained using unmonochromatized Al 
radiation. Passing through 
slit (entrance/exit) of a hemispherical analyzer, electrons with energy 
are detected by a channeltron. The instrumental resolution measured as the full width at a half-maximum (fwhm) of the 
, photoelectron peak is about 
. The energy scale is corrected to the C1s—peak maximum at 
for electrostatic sample charging. The fitting of the recorded XPS spectra was performed, using a symmetrical Gauss-Lorentzian curve fitting after Shirley-type subtraction of the background.
Results and Discussion
Electrochemical response
OCP mapping
In general, OCP mapping determines the time to corrosion initiation. In chloride-containing medium, corrosion initiation is due to the passive layer breakdown on the steel surface, i.e., localized corrosion. For the reinforced concrete (and model pore solution, respectively), the steel surface is considered passive if OCP is equal or more anodic than 
.49–51
Figure 2 shows the OCP evolution for the steel electrodes in the testing solutions (readings are the average value of three samples per condition; deviation also depicted in the plot). As can be observed, the micelles did not significantly influence the OCP of the steel for the control groups (specimens C and CN); both specimens showed OCP values more anodic than 
, i.e., a stable passive layer formed on the steel surface in the chloride-free cement extracts (both with and without micelles). For the corroding groups (specimens CC and CNC), a positive effect was observed for the steel immersed in the micelles-containing cement extract (specimen CNC) but only at earlier stages (1 and 
), evidenced by more anodic OCP values. Within continued treatment, there was no obvious difference in OCP values. The result indicates that the presence of micelles in the chloride-containing cement extract causes a delay in the corrosion initiation.
Figure 2. OCP readings of steel electrodes for all studied solutions.
EIS
EIS is a widely applied electrochemical technique for studying the corrosion performance of steel in both reinforced concrete and simulated pore solution.52–55 Figures 3 and 4 depict an overlay of the EIS responses of the steel electrodes with the time of immersion in cement extract (micelle-free and micelle-containing solutions). Figure 3 presents the response for control cells C (Fig. 3a) and CN (Fig. 3b) from 
; similarly, Fig. 4 depicts the response for the corroding cells CC and CNC (Figs. 4a and 4b), respectively.
Figure 3. EIS response in Nyquist and Bode format for control specimens.
Figure 4. EIS response in Nyquist and Bode format for corroding specimens.
The experimental EIS response was fitted using an equivalent circuit, consisting of two time constants in series with the electrolyte resistance, 
. The summarized values for all cells and time intervals are given in Table I. In general, different equivalent electrical circuits [as well as both pure capacitances and/or constant phase elements (Q)] can be used for interpretation and data fitting of the experimental EIS response. Bearing in mind the duration of the tests, the addition of more time constants would not add more clarity and accuracy in the results. Therefore, the hereby used circuit is considered to be sufficiently describing the behavior of the steel electrodes; moreover, a clear physical meaning can be thus ascribed to each parameter involved. The first time constant deals with the charge transfer resistance and pseudodouble layer capacitance (
and 
, respectively); the second time constant is related to redox transformations, mainly 
, in the surface layers (
and 
, respectively). The replacement of pure capacitance (C) with constant phase element (CPE or Q) in the equivalent circuits is widely accepted for systems as in this study,56–59 being denoted to inhomogeneities at different levels, hereby being mainly relevant to the heterogeneity of hydration/corrosion products that form on the steel surface, in addition to the participation of micelles in the product layer formation. The CPE is an empirical mathematical description of the observed impedance response and is defined as52: 
, being further quantified by the parameters 
and 
(CPE constant and CPE factor, respectively), where 
is a parameter with units 
and 
. When 
a CPE simplifies an ideal capacitor (when 
CPE, simplifies a pure resistor); when 
, characteristic is a nonideal capacitive response. The similarity between the impedance relations of a capacitor (C) and CPE makes it tempting to approximate Q as a capacitance when 
approaches 1 (as in this study, 
values for C and CN are in the range of 0.86–0.99), but this could be an inappropriate approximation;60 Q does not have units of capacitance, and small deviations of the 
values from 1 can lead to large computational errors of capacitance.60–62 Therefore, using CPE (or Q), instead of pure C, was considered in the hereby used equivalent circuits (a thorough description of CPE behavior is provided, for example, in Refs. 63–67). In addition, in order to consider previously reported investigation for similar systems as the hereby investigated ones,45, 46, 54 the derived pseudocapacitance values were recalculated as pure capacitance [based on known relations of CPE and C, e.g., for a given frequency ω, the following relation between the imaginary part of the impedance of the CPE 
and the impedance of the fitted capacitance 
is valid: 
].68 A comparison of 
and 
for the investigated cases in relation to OCP values is given in Fig. 5.
Table I. Best fit parameters from experimental EIS results in cement extract solutions; equivalent electrical circuit: 
; 
.
| Time interval |
|
 , 
|
|
|
 , 
|
|
|
|
|---|---|---|---|---|---|---|---|---|
| Specimen C (control, micelles free) | ||||||||
| 1.90 | 4.02 | 0.8273 | 157.30 | 1.78 | 0.9441 | 159.20 | 186 |
| 2.89 | 3.42 | 0.8956 | 84.50 | 2.19 | 0.9663 | 87.39 | 158 |
| 0.76 | 10.1 | 0.9879 | 100.10 | 1.51 | 0.9119 | 100.86 | 122 |
| 0.43 | 8.76 | 0.9887 | 118.95 | 1.36 | 0.9069 | 119.38 | 128 |
| 0.59 | 2.58 | 0.8664 | 116.35 | 2.18 | 0.9114 | 116.94 | 182 |
| Specimen CC (corroding, micelles free) | ||||||||
| 0.52 | 7.00 | 0.8567 | 43.55 | 5.74 | 0.8822 | 44.07 | 58 |
| 0.65 | 8.82 | 0.7934 | 5.20 | 4.62 | 0.9206 | 5.85 | 10 |
| 0.13 | 6.83 | 0.8755 | 1.60 | 8.25 | 0.8944 | 1.73 | 5 |
| 0.07 | 6.92 | 0.8018 | 1.95 | 2.62 | 0.7308 | 2.02 | 5 |
| 0.13 | 66.9 | 0.9871 | 3.64 | 1.01 | 0.7604 | 3.77 | 5 |
| Specimen CN (control, micelles containing) | ||||||||
| 1.82 | 3.30 | 0.8782 | 130.65 | 2.56 | 0.9607 | 132.47 | 143 |
| 1.95 | 2.51 | 0.8667 | 68.90 | 2.73 | 0.9409 | 70.85 | 114 |
| 0.46 | 1.78 | 0.9989 | 88.15 | 1.48 | 0.9235 | 88.61 | 174 |
| 0.33 | 7.64 | 0.9899 | 99.45 | 1.17 | 0.8688 | 99.78 | 94 |
| 0.39 | 5.61 | 0.9919 | 76.70 | 1.16 | 0.8654 | 77.09 | 101 |
| Specimen CNC (corroding, micelles containing) | ||||||||
| 1.24 | 2.76 | 0.8349 | 104.00 | 3.82 | 0.9989 | 105.24 | 114 |
| 0.83 | 4.29 | 0.8846 | 36.40 | 1.84 | 0.8823 | 37.23 | 33 |
| 0.15 | 4.37 | 0.8876 | 1.30 | 1.77 | 0.7761 | 1.45 | 4 |
| 0.26 | 6.64 | 0.9891 | 3.90 | 1.53 | 0.8096 | 4.16 | 5 |
| 0.15 | 9.69 | 0.9998 | 4.16 | 0.55 | 0.7131 | 4.31 | 5 |
Figure 5. Evolution of 
and 
for (a) control cases C and CN and (b) corroding cases CC and CNC.
The shape of the experimental curves for all specimens is reflecting typical response of steel in alkaline medium, simulating cement paste and in conditions of chloride present at the steel/paste interface.69–71 At low frequencies 
, a close to capacitive behavior was observed for specimens C and CN (control cases). In contrast, inclined to the real axis, semicircles are characteristic for specimens CC and CNC (corroding cases). The response for specimens CC and CNC reflects the evolution of corrosion with time, also evidenced by the more significant phase angle drop (from 80 to 
and lower) for these specimens, compared to the control ones C and CN (phase angle of 76–85°). The EIS results indicate that the micelles apparently alter the surface layer composition and result in an initial corrosion resistance improvement. Table I depicts the calculated global polarization resistance 
values derived by EIS (based on a well known, simplified calculation of 
when the overall reaction rate and 
, respectively, is also related to product layer transformations, including both oxidation and reduction, etc.72).
Figure 5 depicts a comparison of capacitance values (
and 
) for all cells at all time intervals. As can be observed, the values of 
and 
for the control cells C and CN tend to be similar and at lower scale (OCP values between 
and 
vs SCE at the end of the test), compared to these for the corroding specimens CC and CNC (Fig. 5). The recorded evolution of 
and 
at the corresponding OCP values accounts for the low rate of oxidation processes for specimens C and CN (indicating passivity).54 The derived values and trends of capacitance change with treatment, also related to the OCP values, are consistent with the range of such, previously derived for model solutions46, 54 and correspond well to the derived high resistance values (
and 
) in the range of 
, denoting for the passive state of the steel electrodes (specimens C and CN).
In contrast, for specimens CC and CNC, significantly higher values for both 
and 
were recorded (Fig. 5b). For the former case (specimen CC), the higher capacitance 
values correspond to a lower resistance [
in the range of 
(
response) to 
(
response)]; the resistance 
values for CNC are higher (well in line with lower 
), being in the range of 
(
response) to 
(
response).
The evolution of 
values derived from EIS present a similar trend to the 
values derived from PDP tests (although PDP presents the behavior of the steel electrodes with external polarization and therefore exactly the same 
value from both methods is not possible and consequently not recorded). The summarized 
values from PDP measurements are given in Table I, and the recorded behavior with external polarization is summarized in what follows.
PDP
The PDP curves for all investigated cases at the time intervals of 1, 3, 6, and 
are shown in Fig. 6 (6 and 
response being almost identical, therefore 
data are not presented).
Figure 6. Potentiodynamic polarization of steel in CE at different time intervals.
As seen from Fig. 6, the specimens immersed in the chloride-free solutions (specimen C and CN) depict similar behavior with external polarization (as also recorded by EIS): the corrosion potential is more anodic than 
, and the corrosion current density is low, meaning that the steel electrodes from both specimens C and CN are in passive state. For the specimens immersed in the chloride-containing solutions (specimen CC and CNC), the data showed a pronounced positive effect of the micelles at early stages (Figs. 6a 
and 6b 
), evidenced by a more anodic corrosion potential, lower corrosion, and anodic current densities for the micelles-containing specimen CNC (actually similar to the control specimens at 
time interval). Within continued treatment, there was no significant influence of the micelles (Fig. 6c 
and 6d 
). However, for the micelles-containing specimen CNC, the corrosion potential remained slightly more anodic and the anodic current density slightly lower, compared to the micelles-free specimen CC.
As mentioned in the introduction part, the micelles used in this study do not belong to organic corrosion inhibitors (because the previously synthesized and stabilized micelles are added to the model solutions, rather than adding organic compounds known to form micelles); however, some similar effects with regard to steel corrosion behavior can be observed. The PDP curves (Figs. 6a and 6b) show that, for the corroding specimens (before corrosion initiation), the presence of micelles reduced the anodic current by approximately 1 order of magnitude; on the other hand, their effect on the cathodic current was not obvious. The results indicate that the micelles mainly hinder the anodic reaction on the steel surface (due to the micelles absorption, which can impede chloride diffusion through the surface layer and delay steel corrosion), which is an effect (although for different environmental conditions) similar to the action of reported organic inhibitors of different composition and structure.21, 30, 32 The summarized 
values, derived from potentiodynamic polarization, are shown in Table I. For the control specimens, the 
was quite high (in the range of 
), which indicates passivity. For the corroding specimens, the same trend as within EIS measurements was observed: the 
value of the steel immersed in the micelles-containing solution was higher than that in the micelles-free solution at early stages, such as 
(
for the former, compared to 
for the latter case) and 
(
for specimen CNC versus 
for specimen CC). Further, similar 
values for both specimens CC and CNC were recorded at later stages. Based on the electrochemical measurements, it can be stated that the presence of micelles in the model solutions is able to increase the corrosion resistance of the steel electrodes; an obvious improvement, however, only appears at early stages, which is to a certain extent an expected outcome, considering the very low concentration of micelles in this study (of only 
).
Steel surface analysis
Morphological observation and EDX analysis on the steel surface
Figures 7 and 8 shows the morphology of the steel surface after treatment in cement extracts for 1 and 
(no significant difference between 
and earlier intervals as 3, 6, 12, and 
was observed), respectively. The results from EDX analysis (most relevant elements) at 
time interval are summarized in Table II.
Table II. Summarized data from EDX analysis of steel surface.
| Elements | C (wt %) | CN (wt %) | CC (wt %) | CNC (wt %) |
|---|---|---|---|---|
| Na | 2.27 | 2.26 | 8.76 | 8.23 |
| Al | 0.25 | 0.35 | 0.29 | 0.23 |
| Si | 0.21 | 0.41 | 0.33 | 0.22 |
| S | 0.31 | 0.58 | 0.44 | 0.42 |
| Cl | 0.37 | 0.37 | 3.13 | 2.11 |
| K | 1.15 | 1.60 | 1.47 | 1.06 |
| Ca | 0.32 | 0.53 | 0.33 | 0.31 |
| Fe | 83.22 | 78.81 | 76.41 | 75.97 |
At early immersion stage 
, for the control specimens (Figs. 7a and 7b), similar morphology and composition (Table II) were recorded. For the corroding specimens (Figs. 7c and 7d), in addition to the well pronounced "pitting" corrosion (both for specimens CC and CNC), micelles absorbed to the steel surface in specimen CNC can be observed at a higher magnification (Fig. 7d). The presence of micelles on the steel surface would be expected to improve at least the barrier properties of the formed layer; this effect, however, appeared to be relevant only at early immersion intervals as evidenced by the electrochemical response (Figs. 3, 4, 5 and 6). Within longer immersion times, there is no obvious difference in the appearance of the product layers as observed by ESEM (Fig. 8) except that the product layer for the micelles-containing specimen was visually more homogeneous and compact, which also corresponds to the similar corrosion behavior of the specimens at these stages.
Figure 7. Morphology of steel surface in CE solutions at 
obtained by ESEM.
Figure 8. Morphology of steel surface in CE solutions at 
obtained by ESEM.
The micelles, present in the solutions (both control and corroding cases CN and CNC), exerted alterations in the product layer formation and composition on the steel surface. Although EDX analysis is a qualitative and semiquantitative technique, it can serve for comparative purposes and additional evidence: for example, based on the EDX results at 
time interval, more carbon was detected for the micelles-containing specimens (for control specimens CN, the carbon content was 
, compared to 
for the micelles-free specimen C), indicating that the micelles were absorbed on the steel surface. In addition, the chloride concentration was lower for the micelles-containing corroding specimen CNC 
, compared to specimen CC 
, which would result in a layer with a better protective ability for specimen CNC. Further, steel surface analysis was performed by XPS, which will be discussed in what follows.
XPS
The XPS analysis was performed on the steel surface of all investigated cases after treatment of 
(the time interval "
" was chosen for XPS, because the difference in electrochemical behavior were most apparent at this stage). Table III presents the surface atomic concentration (atom %) of the most relevant elements for all specimens.
Table III. Surface atomic concentrations (atom %).
| C1s (%) | O1s (%) | K2p (%) | Ca2p (%) | Cl2p (%) | Na1s (%) | Fe2p (%) | |
|---|---|---|---|---|---|---|---|
| C3h | 49 | 28 | 10 | 2 | 4 | 6 | 1 |
| CC3h | 47 | 27 | 3 | 3 | 7 | 12 | 1 |
| CN3h | 55 | 26 | 8 | 1 | 3 | 6 | 1 |
| CNC3h | 56 | 22 | 3 | 3 | 5 | 10 | 1 |
XPS carbon (C1s) ionizations
Table IV presents the total carbon content on the surface of the investigated layers in comparison to a "control" case of the nanoaggregates (
-b-
micelles) only. The XPS measurement for micelles only was performed on a droplet of the as-received micelles solution on aerosil 
. Figure 9 depicts the detailed C1s spectrum for the micelles. As seen from Tables III and IV, specimens CN and CNC present higher amounts of carbon (also evident from the higher intensity of the C1s 
peaks, Fig. 10), compared to specimens C and CC. The fitting of the recorded XPS spectra was performed, using a symmetrical Gauss-Lorentzian curve fitting after Shirley-type subtraction of the background and considering reported procedures.73, 74 The relevant binding energy (BE) for the characteristic peaks after curve fitting (Figs. 9 and 10) and the corresponding bonds (including the percentage of the relevant peaks from the total carbon content) are given in Table IV [to be noted is that the C1s spectra (Fig. 10) include the relevant peaks for K, at binding energies higher than 
, which will be further separately discussed].
Table IV. Surface atomic concentrations C1s (atom %).
| BE | CC3h | C3h | CN3h | CNC3h | PEO-b-PS (control) |
|---|---|---|---|---|---|
| 40 | 42 | 47 | 48 | 57 |
 (C–O) in C–H–O (PEO) | — | — | 6 | 4 | 5 |
| 7 | 7 | 2 | 4 | — |
 in C–H–O (PS) | — | — | — | — | 3 |
| Total % C1s | 47 | 49 | 55 | 56 | 65 |
Figure 9. Detailed C1s spectra for the "control" micelle only sample.
Figure 10. C1s spectra for (a) C and CC and (b) CN and CNC.
The C1s spectra for the control (micelles only) sample and for specimens CN and CNC were decomposed into three components (Figs. 9 and 10b, Table IV): at 
, corresponding to carbon only, bound to carbon and hydrogen [C–(C–H)]; near 
, corresponding to a single bond carbon-oxygen (C–O); and at 
, due to a double bond carbon-oxygen 
, which is denoted to 
containing compounds. Similar curve fit of the C1s spectrum was applied for the "control" (micelle only sample), except that there is no peak between 289.2 and 
, but the characteristic peak (shake up) for PS (the "core" of the micelles) appears at 
(Ref. 75) (Fig. 9, Table IV). The peak at about 
for the "control," micelle sample, including the shape of the C1s spectrum as a whole (Fig. 9), denotes for the PEO "part" (the "shell") of the micelles.76–78 In other words, the shape of the C1s spectrum for specimens CN and CNC, including the characteristic binding energies (Table IV) and surface and shape of the peaks at about 
(Figs. 9 and 10), proves the presence of micelles (6% for specimen CN and 4% for specimen CNC) in the layer, formed on the steel surface, whereas these features of the C1s spectra in specimens C and CC are not present (Fig. 10a).
XPS potassium (K in C1s) ionizations
For the K2p spectra
The shape of the peak in specimen C differs from the rest of the specimens; it is curve fitted (not hereby presented) into two peaks with binding energies at 293.6 and 
, denoted to 
at the higher energy (BE 
in the relevant S2p spectra) and KCl at the lower energy (BE 
in the Cl2p spectrum, S2p and Cl2p spectrum not presented). The K2p spectrum for specimen CN denotes for KCl at 
(BE 
in the relevant Cl2p spectrum); whereas in specimens CC and CNC, 
is only relevant at 
[characteristic peaks with BE of 
in the S2p spectrum (not presented) and 
in the K2p spectrum].
XPS calcium (Ca2p) ionizations
Considering the Ca2p spectra (not presented) and the already discussed C1s spectra (Fig. 10), including the O1s ones (Fig. 11b), the following compounds are relevant: for specimen C: 
, corresponding to binding energy 
(75% of the total Ca content), and 
, corresponding to binding energy 
(25% of the Ca content). For specimen CN: 
at binding energy 
(76% from the Ca content) and 
at 
(at 24% from the total Ca content). For the specimens CC and CNC (corroding surfaces), only 
is relevant, corresponding to 
in the Ca2p spectra and 289.5 and 
in the C1s spectra, respectively.
Figure 11. (a) Fe2p and (b) O1s spectra for all specimens.
XPS oxygen (O1s) and iron (Fe2p) ionizations
Figure 11 presents the Fe2p and O1s spectra for all specimens. When complex surface layers are investigated by XPS, the interpretation of the O1s spectra is not always straightforward and is a complicated one. In the hereby discussed case, the surface layers contain a mixture of iron oxides/hydroxides, polymer nanoaggregates, oxides/hydroxides of alkaline metals (from the environmental medium, CE), carbonation, etc. Therefore, a simplified curve fitting, including only the main features of the O1s spectra (and the relevant differences between the investigated samples) is considered more appropriate and reliable, rather then involving exact values and complicated interpretations. Further, the characteristic feature in the O1s spectra are correlated to the Fe2p spectra.
The Fe2p line shape in iron oxides is similarly very complex, and the chemical shift between 
and 
components is too small to be exactly resolved within the energy resolution of the XPS measurements 
. Moreover, the percentage of iron/iron oxides on the surface of the investigated specimens is rather low [meaning accumulation of surface layers during treatment, mainly carbonates for the case of specimens C and CC and additionally, a surface, at least partially "blocked" from micelles (evident form the lower oxygen levels) in the specimens CN and CNC (Table III)]. However, certain features in the Fe2p ionizations allow the identification of Fe-oxidation states and the relative compounds, the most reliable evidence being the satellite structures due to charge transfer screening79, 80 and the overall shape of the spectrum for each investigated condition. Combined with the curve fitted O1s spectra, the difference in composition of the product layers can be clearly observed.
Specimens C and CN (control specimens without and with micelles)
For specimen C, the Fe2p photoelectron peaks at 710, 719, and 
(Fig. 11a) represent binding energies of 2p3/2, shake-up satellite 2p3/2 and 2p1/2, respectively. The characteristic binding energies, the satellite at 
, and the overall shape of the Fe2p pattern for specimen C can be attributed to either FeOOH or 
, which are known to exhibit similar features and peak positions. The O1s pattern for specimen C, Fig. 11b, (as well as for specimen CN) is decomposed into three peaks: at 528.7, 531.6, and 
corresponding to 
, 
, and chemically or physically adsorbed water, respectively. The existence of surface 
groups (5% from the total O1s peak) suggests that the oxidized iron is likely in the state of FeOOH. If the surface structure was 
, a single peak at 
should have been observed.81, 82 Additional evidence for the most probable existence of FeOOH is the broadening of the O1s at lower binding energies (the peak at 
, 1% from the total O1s peak), which actually corresponds to FeOOH.81, 83, 84
For specimen CN, the Fe2p photoelectron peaks at 710.3 and 
represent binding energies of 2p3/2 and 2p1/2, respectively; in contrast to specimen C, the satellite at 
is less pronounced, and a plateau region is rather observed between 715 and 
. In addition, there is a broadening of the 2p1/2 peak toward lower energy (at 
). Consequently, the shape and characteristic features of the Fe2p spectra for specimens CN denote for 
, present in the surface layer.73, 85 Considering the features of the O1s spectrum for specimen CN (Fig. 11b), as an evidence for the presence of 
in specimen CN, the following is relevant: the 
peak (characteristic energy for 
), is 2.4% from the total oxygen on the surface and the significantly lower amount of adsorbed water (less OH, evidenced by the 2.4% for the 
peak from the total O1s). Therefore, for specimen CN, 
is characteristic for the product layer on the steel surface.
Specimens CC and CNC (corroding specimens without and with micelles)
The photoelectron peaks at 
(CC) and 
(CNC) represent Fe 2p3/2; there is a broadening at 
and an additional broadening (Fe 2p1/2) at 
and at 
, respectively. These features and the shape of the Fe2p spectrum for both CC and CNC specimens denote for the presence of 
.86 The existence of the latter is also evidenced by the 
peaks in the C1s spectra (previously discussed, Fig. 10). Further, the shape of the Fe2p for CC, the 2p3/2 satellite at 
, and the relatively high intensity of the 2p1/2 peak at 
(Fig. 11a, compared to specimen CNC) denote for the presence of 
(most likely a mixture of FeO and 
). Evidence is also the lower energy peak in the O1s spectrum (Fig. 11b) at 
(corresponding to 
). The presence of FeOOH is not likely, because the 
peak in the O1s spectrum represents only 1.5% from the total O1s, the higher energy broadening of the O1s peak is not as evident as in specimen C for example.
For specimen CNC, along with the presence of 
, the shape of the Fe2p (Fig. 11a) reveals the presence of 
, evidenced by the following characteristic features: 2p3/2 peak at 
, 2p3/2 shake-up at 
, and a well pronounced 2p1/2 peak at 
; additionally, the Fe2p3/2 fwhm is 
(Ref. 56): all these features in Fe2p denote for the presence of 
in the surface layer of specimen CNC. Additional evidence is the 
peak in the O1s spectrum (about 1% of the total O1s, denoted to 
). Further, compared to all other specimens, the peak at 
(denoted to hydroxides) is almost negligible, contributing with less than 1% in the total O1s.
The morphology observation and surface analysis (Figs. 7d and 10b) reveal that similar to organic inhibitors, the micelles were absorbed on the steel surface. The absorbed micelles are expected to increase the "barrier effect" of the layer, delaying the breakdown of the passive film, which is a phenomenon generally known to be responsible for corrosion delay in the presence of organic inhibitors.32 Moreover, the XPS results (Fig. 11) indicate that the micelles can also alter the chemical compositions of the product layers: the product layers in the micelles-containing specimens CN and CNC present mainly protective 
and/or 
, leading to an impeded 
penetration and consequently corrosion delay; in contrast, the product layers in the micelles-free specimen C and CC were more hydrated, exhibiting mainly FeOOH, FeO, and 
, which are prone to chloride attack.87, 88
Correlation product layers composition/EIS parameters
The XPS results (supported by ESEM and EDX) indicate that different types of iron oxide/hydroxide layers were formed on the steel surface after treatment in the test solutions with and without the presence of micelles. Further, a Ca-rich outer layer was formed (Figs. 7 and 8), which can additionally protect the steel. However, it is reported that the Ca-rich layer provides only limited protection and the inner layer of iron oxide/hydroxides is predominant to protect the steel.54
The electrochemical response (and in particular the EIS parameters, Table I, Fig. 5) support the findings for product layers composition and the above discussed peculiarities of these layers, formed in each relevant condition. For the control groups, a lower charge transfer resistance 
and a lower 
were recorded for the micelle-containing specimens (CN), compared to the micelle-free specimens C [both C and CN are passive steel, with certainly different (higher) scale of 
values, compared to CC and CNC, Table I]. The lower 
, 
, and 
, respectively, for specimens CN is hereby denoted to the less oxidized steel surface (Table III, also resulting from at least partial "blocking" of the surface from the micelles), and therefore a stable passive layer is apparently taking more time to form. In addition, considering that the electrical conductivity of FeOOH, 
and 
is decreasing in the order: 
,89 the product layer in specimens C, being predominantly FeOOH is less conductive, i.e., higher diffusion limitations within 
measurements are relevant. In contrast, the product layer in specimens CN is mainly 
, thus a higher electrical and ionic conductivity would be expected, resulting in a lower (compared to specimen C) global 
values for the time interval of this test.
For corroding groups, the charge transfer resistance 
for the specimen CNC is slightly higher, compared to the specimen CC; the 
is significantly higher for 1 and 
and still higher for 12 and 
(Table I). The results indicate that the expected improvement of corrosion resistance in the corroding, micelles-containing specimens CNC is only significant at early time intervals, e.g., 1 and 
, which is related to the compositions of the product layer formed on the steel surface.
As seen from Table III, the product layers in specimens CN and CNC (the micelles-containing specimens) present less oxygen, the lowest oxygen content being in specimen CNC 
. The XPS analysis reveals that micelles were apparently presented in the product layers (considering the features of the C1s spectra), also evidenced by ESEM and EDX (Fig. 8). Therefore, the presence of only 
micelles in CE leads to increased barrier effects (at least on earlier stages) but also results in "blocking" the steel surface and less oxygen content.
Comparing the electrochemical response and surface layer properties of control and corroding specimens, the following is relevant: the product layers for specimens C and CC (both micelles-free) present FeOOH (for specimen C) and a mixture of 
, FeO and 
(for specimen CC). Specimen C is a noncorroding specimen, where apparently the product layer was largely hydrated (because after treatment and within the XPS analysis, water is still present in the film). Specimen CC is a corroding specimen, where similar composition to specimen C would be expected before corrosion initiation. Because specimen CC was treated in Cl-containing cement extract, logically the composition of the film changes (within corrosion initiation and propagation): chlorides are known to change the composition, thickness, and density of the passive films.90–92 The incorporation of water in the product layer will result in a more hydrated film with a decreased protective nature and ability because the water "paths" would facilitate chloride penetration and consequently will ease breakdown of the passive film.87, 88 Therefore, an initially more hydrated film (as in CC) along with an easier Cl penetration, results in faster corrosion initiation and propagation. In contrast, a more dehydrated film, as would be in specimens CN and initially in specimens CNC (micelle containing) would be less prone to chloride attack, which in addition to the partially "blocked" by micelles steel surface (e.g., Figs. 7d and 10b), will result in an impeded 
penetration and consequently corrosion delay. These latter mechanisms are actually evidenced by the lower chloride content in the surface layer of corroding specimens CNC, compared to corroding specimens CC (Tables II and III), which in turn means better properties for CNC from a corrosion view point.
Conclusions
A preliminary study on the corrosion behavior of low carbon steel in cement extract (simulated pore solution) in the presence of polymeric nanoaggregates (
micelles) is reported in this work. The electrochemical measurements indicate that a very low concentration of the micelles 
is able to increase the corrosion resistance of the steel in the chloride-containing cement extract However, the significant effect only appears at early stages, which is to some extent as expected, considering the very low concentration of micelles used in this study.
The morphological observation and surface analysis confirm that the micelles are absorbed on the steel surface and would be expected to result in a more uniform and compact layer, i.e., to increase barrier properties. The XPS analysis reveals that the product layers of the micelles-containing specimens exhibit a more homogeneous and protective 
and/or 
, whereas the product layers of the micelles-free specimens exhibit more hydrated FeOOH, FeO, and 
, which is prone to chloride attack. Therefore, the "barrier effect" along with the composition of the product layers on the steel surface (in addition to the very low concentration of the micelles) denote for the initially increased corrosion resistance of the steel in the chloride-containing cement extract in the presence of micelles.
Delft University of Technology assisted in meeting the publication costs of this article.