You are on page 1of 10

CORROSION ENGINEERING SECTION

Electrochemical Behavior
of Steel-Reinforced Concrete
During Accelerated Corrosion Testing

S.A. Austin,‡,* R. Lyons,* and M.J. Ing*

ABSTRACT INTRODUCTION
Corrosion of reinforcing steel presents a major durability Research Significance
issue worldwide and is the focus of much research activity. The focus of this paper is to examine and review
The long time periods involved in replicating reinforcement
the impressed current method of accelerating chlo-
corrosion within laboratories has resulted in a number of
ride-induced corrosion. While this method is com-
accelerated test methods being developed. The basis of
the research presented in this paper was to examine the
mon, little published work has been found describing
impressed current technique often used to induce reinforce- its suitability as a technique to model reinforcement
ment corrosion. The suitability of the technique to model chlo- corrosion and the theory behind the practice. In par-
ride-induced corrosion was investigated by examining the ticular, the electrochemical implications of applying
electrochemical nature of the test method. Corrosion was an external current are examined and compared
induced in prisms of differing characteristic strengths and against the natural electrochemistry of corrosion of
cover thicknesses by applying a current between 3 days and steel reinforcement. The experimental program inves-
17 days. The acoustic emission (AE) technique detected the tigated the applicability of applying Faraday’s law for
onset of corrosion. The gravimetrical and theoretical mass the calculation of mass loss under the test conditions
losses are compared and a modified expression based on
set out below, and the paper discusses the suitability
Faraday’s law relating the electrical current to the mass loss
of the technique as a model for accelerating chloride-
is also proposed, which accounts for the localized nature
of chloride-induced corrosion. This improved method is induced reinforcement corrosion within the labora-
dependent upon a reliable method of detecting the onset of tory. Corrosion induced by an impressed current was
corrosion, which can be achieved by AE. On balance, the detected with a novel corrosion detection technique
impressed current technique appears suitable to simulate based on acoustic emission.
reinforcement corrosion.
Corrosion Mechanisms
KEY WORDS: acceleration, acoustic emission, anodic polar-
The damaging role that corrosion of reinforced
ization, electrochemical properties, impressed current, steel-
reinforced concrete
concrete has on the durability of concrete structures
is well documented.1-2 To improve understanding of
reinforcement corrosion and its effects, accelerated
Submitted for publication November 2002; in revised form, June modeling of corrosion in the laboratory is often re-
2003. quired. The reliability and accuracy of the results will

Corresponding author.
* Centre for Innovative Construction Engineering, Loughborough depend upon the closeness of the laboratory simula-
University, Leicestershire, LE11 3TU, United Kingdom. tion to corrosion in real-life situations. Good quality
0010-9312/04/000019/$5.00+$0.50/0
CORROSION—Vol. 60, No. 2 © 2004, NACE International 203
CORROSION ENGINEERING SECTION

concrete will normally offer excellent chemical pro- duration of the wetting and drying periods.6 Further-
tection for the steel reinforcement against corrosion3 more, the subsequent rate of corrosion is not acceler-
due to the high alkalinity and the low permeability of ated and is dependent upon a sufficient supply of
the matrix.4 At a pH of 13.5, the interaction between oxygen and water. The first two methods are either
the steel and the hydroxyl ions present in the pore combined or separate, depending upon the reason
solution results in the formation of an insoluble behind the study.
γ–Fe2O3 layer, rendering the underlying metal pas- Where an impressed current is used to drive
sive. The low permeability of the concrete represents corrosion, the amount of mass loss is related to the
the ability of the concrete matrix to resist diffusion electrical energy consumed once passivity has been
of aggressive species, such as the chloride ion, into compromised, and can be modeled using Faraday’s
the concrete.4 law, Equation (1):
Neither the high alkalinity of the pore solution or
the low permeability of the cover can guarantee that Mit
the steel will resist corrosion, especially in aggressive Mass loss = (1)
zF
environments such as marine structures. Chloride
ions may enter the concrete during mixing, as admix- where M is the molar mass (55.847 g/mol for iron);
tures, or after curing from external sources such as i is the current (A); t is the time (s); z is the number
seawater and deicing salts. Once chlorides have of electrons transferred; and F is Faraday’s constant
reached bar level, they depassivate the embedded (96,487 Coulombs/mole).
steel by locally breaking down the protective layer.5-6
The time chloride ions take to reach the level of the Electrochemistry of Corrosion
reinforcement depends on the mechanism of intru- Corrosion is an electrochemical reaction involv-
sion, the external concentration of the chlorides, ing the transfer of charge from one species to an-
and the internal microstructure of the concrete.7 In other. It consists of two half-cell reactions, an anodic
water-saturated concrete, chlorides are considered to reaction and cathodic reaction, connected together
penetrate the cover by a diffusion mechanism with by an electrolyte. To maintain a charge balance,
the driving force being differential concentration gra- the rate of reduction (anodic) and rate of oxidation
dients.7 In this circumstance, Fick’s second law of (cathodic) must be equal.3
diffusion can be applied to model the rate of chloride Steel, when in the presence of water and oxygen,
penetration. In partially saturated concrete, chlorides will undergo a reduction reaction at the anode where
enter by adsorption and capillary forces7-8 and the the exact nature of this reaction is dependent upon
period of intrusion is generally known as the initia- the reaction thermodynamics. For example, iron can
tion period. This is followed by propagation of the dissolve to form Fe++, but it also can form solid oxides
corrosion, the duration being controlled by factors that in some instances may protect the underlying
such as temperature, oxygen concentration, and steel (such as the passive layer). All the possible oxi-
moisture content. At low corrosion rates, of the order dation reactions between iron and water have been
of 1 µ A/cm2, it is estimated that the propagation pe- measured and calculated by Pourbaix18 and pre-
riod may continue for 20 years before cracking of the sented in a potential-pH diagram, which illustrates
cover occurs.9 the thermodynamically stable phases of the possible
To study the effects of corrosion within a realis- reactions as a function of electrode potential, and
tic time scale, it is sometimes necessary to accelerate pH, giving rise to various zones of corrosion, passiv-
the initiation period and occasionally control the rate ity, or immunity.19
of corrosion during the propagation stage. The meth- At a pH of 13.5, two regions generally apply for
ods used to do this will vary upon the nature of cor- iron: passivity or immunity, as shown in Figure 1.
rosion under investigation. For example, to accelerate Immunity is only obtainable by lowering the potential
carbonation-induced corrosion, the concrete can be below the equilibrium potential of hydrogen (Line A),
placed in a carbon dioxide (CO2) chamber at 50% achieved by applying external energy to the reaction
relative humidity (RH) to rapidly increase neutraliza- (cathodic protection). At potentials more positive
tion of the pore water.10-11 than –1,125 mV vs standard calomel electrode (SCE),
Accelerated chloride-induced corrosion tech- the steel is covered by a stable film of magnetite
niques are used frequently to reduce the time taken (Fe3O4) or ferric oxide (Fe2O3), which reduces corro-
for a critical level of chlorides to reach the reinforce- sion of the active underlying metal to a negligible rate
ment bar.12-17 Three methods are common: admixed (0.1 µ A/cm2). Consequently, the potential of passive
chlorides, impressed voltage/current, and wet/drying steel can vary between –1,125 mVSCE and 175 mVSCE
techniques. The latter usually requires several (the equilibrium potential of O2 at pH 13.5), governed
months before sufficient levels of chloride ions have by the pH of the pore solution and the oxygen con-
permeated the cover to cause depassivation as the tent at the rebars, as shown in the cathodic Nernst
extent of the movement of salt will depend on the Equations (2) and (3):20

204 CORROSION—FEBRUARY 2004


CORROSION ENGINEERING SECTION

RT
Ec = Eo + ln α O2 ( dis ) (2)
4F

0.059
E c = +1.22 – pH (3)
4

where Ec is cathodic potential (mV); Eo is open poten-


tial (mV); R is Molar gas constant; T is absolute tem-
perature (K); and a is activity.
The mixed potential (Ecorr) measured in any cor-
roding solution is a combination of the anodic and
cathodic reactions, governed by the mixed potential
theory. Both reactions are polarized toward each
other, as shown schematically in Figure 2(a). This
corresponds to a potentiostatic curve for steel in an
alkaline solution showing three regions of interest:
general corrosion, passivity, and oxygen evolution.
Ecorr is equal to the intersection of the anodic and
cathodic polarization curves. Each value of Ecorr has
a corresponding current value (Icorr). However, this FIGURE 1. Potential-pH diagram for iron in water.
value is only apparent, as no net current exists due
to the balance of charge.
In noncarbonated, chloride-free concrete, the pH 32 MPa with a range of 27 MPa to 36 MPa and an
is usually fairly constant over time, and variations in average standard deviation coefficient of 11%. The
the potential are usually a result of the changing average strength of the C35 mixes were 54 MPa with
oxygen content at the steel-concrete interface. At a range of 47 MPa to 60 MPa and an average stand-
very low oxygen levels, the potential (Ecorr) can fall to ard deviation coefficient of 6% (3.6 to 10). The C50
very negative values (Figure 2[b]), where the passive mix had a mean strength of 68 MPa (range: 60 to 75)
layer is no longer stable and may be reduced leaving with a standard deviation coefficient of 9%.
the steel surface in an active state. However, the The prisms contained one ordinary Grade 460
corrosion rate in these circumstances is reported to deformed steel rebar having a nominal diameter
be not much greater than when passive.3 of either 12, 16, or 20 mm placed so that the cover
was equal on three sides. The thickness of the cover
EXPERIMENTAL PROCEDURES (c) was intentionally varied between 16, 25, and
40 mm (±1 mm). The dimensions and properties of
Experimental Details individual prisms are given in Table 2. The last
The experimental methodology was as follows. 20 mm of each bar end was wrapped in electrical
Corrosion of the rebar was induced by applying a insulation tape prior to casting to eliminate edge
fixed anodic current of 100 µ A/cm2 to the prisms. effects. Before casting, the reinforcing bars were
The electrochemical potential of the steel was moni- wire-brushed to remove any scale and then weighed
tored periodically using a saturated calomel reference to a precision of two decimal places. The prisms were
electrode placed on the concrete surface. Due to the water-cured for 28 days, then stored in laboratory
effect of the applied voltage on the potential, corro- conditions of 20°C ± 5°C for a minimum of 28 days
sion activity was monitored continuously using the before testing.
novel acoustic emission technique. Upon completion In any one test, three prisms, each of the same
of the test, the prisms were examined visually for type, were placed in a corrosion cell as shown in Fig-
surface cracking and a visual inspection of the cor- ure 3. A 24-h period passed before the current was
roded area was undertaken. The gravimetric mass passed through the cell. Acoustic emission (AE) read-
change was recorded after the rebar was removed ings were taken during this initial 24-h period to es-
from the prism and cleaned. tablish background levels of emission for each prism.
Singly reinforced concrete prisms were cast with A 30 VDC/2A power supply was set to constant cur-
designed characteristic strengths of 20, 35, and rent mode supplying an anodic current of 100 µA/cm2
50 MPa. All mixes contained uncrushed river gravel to each rebar. This current density is reported as be-
with a maximum aggregate size of 10 mm and had a ing the maximum corrosion current found in real
free-water content of 180 kg/m3. The mix proportions conditions9 and has been used by several research-
for each strength are shown in Table 1, together with ers.9,21-22 Electrical continuity was provided by a wet
the target strengths. All mixes used Type 1 Portland wick saturated in a 5% sodium chloride (NaCl) solu-
cement. The average strength of the C20 mix was tion. The counter electrode was an AISI 304 (UNS

CORROSION—Vol. 60, No. 2 205


CORROSION ENGINEERING SECTION

S30400)(1) stainless steel plate immersed in the elec-


trolyte. The half-cell of the concrete was recorded
initially and then periodically throughout the test
duration. Termination of the test occurred between
1 day and 3 days after the initiation of corrosion ac-
tivity was detected by the AE. This gave a range to
the extent of corrosion. In a few instances the test
was terminated ~1 day before it was expected AE
would start, to verify that corrosion was not occur-
ring prior to the onset of characteristic emissions.
A fourth control prism from the same batch,
without an impressed current, was used as a control
and monitored continuously for the duration of the
test to establish the background noise levels associ-
ated with the testing environment and the concrete.
Upon completion of the test, the control beam was
opened to check that there was no visible corrosion
of the rebar and hence confirm that in the absence
of any significant AE corrosion was not occurring.
On completion of the test, the prisms were pho-
tographed and examined for any evidence of external
cracking. The prisms were then sliced in half on ei-
(a)
ther side of the rebar with a diamond saw. The rebar
was removed and the total area of corrosion products
was calculated after measuring with a rule marked in
0.5-mm divisions. Finally, the bars were wire-
brushed to remove any corrosion products and re-
weighed. Mass loss also was calculated theoretically
and compared to the gravimetric value.

Detection of Corrosion Activity


A technique was required that would not inter-
fere with or be affected by the applied current, and
could be used continuously so that a reasonably ac-
curate estimate of corrosion initiation could be made.
Application of AE to the detection of corrosion-in-
duced concrete damage has been reported as being a
suitable laboratory tool.23 Furthermore, it is reported
to be sufficiently sensitive to detect corrosion-related
deterioration processes at an early stage.24 The use of
this technique is not widely practiced and relatively
little information has been published of its use as a
laboratory tool. However, previous work has found
that this nondestructive technique was able to detect
active corrosion before traditional methods such as
(b)
half-cell and galvanic current measurements.25-26 The
FIGURE 2. (a) Schematic potentiostatic curve for steel in concrete,
(1)
showing the mixed potential. (b) Effect of a decreasing oxygen UNS numbers are listed in Metals and Alloys in the Unified Num-
concentration on the corrosion rate. bering System, published by the Society of Automotive Engineers
(SAE International) and cosponsored by ASTM International.

TABLE 1
Mix Proportions

Strength (MPa) Target Strength (MPa) Cement Water Sand Aggregate

20 28 1 0.57 2.06 3.8


35 43 1 0.45 1.57 2.9
50 58 1 0.33 0.6 2.4

206 CORROSION—FEBRUARY 2004


CORROSION ENGINEERING SECTION

TABLE 2
Series Description
Rebar Diameter Characteristic Strength Cover Thickness Actual Strength Dimensions (mm)
(mm) (MPa) (mm) (MPa) w×h×L

16 35 16 53 48 × 70 × 250
16 20 16 32 48 × 70 × 250
16 50 16 68 48 × 70 × 250
12 35 16 57 44 × 70 × 250
20 35 16 54 52 × 75 × 250
16 35 25 51 66 × 82 × 250
16 35 40 50 96 × 100 × 250

ability of this technique to detect early age corrosion


without perturbation of the corrosion cell is a clear
advantage over other methods. Acoustic emissions
are generated as a consequence of microcrack devel-
opment and their propagation in the concrete during
the migration of corrosion products.27
The AE DiSP 4-channel board was used with
piezoelectric transducers connected in series to an
external pre-amp. Both hit- and time-driven data
were recorded onto a hard drive for post-processing.
The transducer was attached directly to the top of
the prism, using adhesive grease as a couplant. FIGURE 3. Accelerated corrosion cell.

RESULTS
obtained from Figure 5. At low potentials, Line AB,
Half-Cell Potentials general corrosion of the steel occurs with increasing
Half-cell potentials taken during the initial 24-h current density until the primary passive potential
period ranged between –90 mVSCE and 120 mVSCE, (Epp) is reached. This potential corresponds to a sud-
placing the bars within the passive region. This was den drop in current density, which can be several
confirmed by the absence of significant acoustic orders of magnitude, corresponding to the formation
emissions. This is the usual state for noncarbonated, of a very thin passive film (Line CD).19,28 This oxide
chloride-free concrete. protects the underlying steel even at very positive po-
When the anodic current was applied to the rebar, tentials. At potentials greater than the transpassive
the half-cell reading value shifted to extremely posi- potential (Etp), the oxidation reaction that occurs
tive values in the range of 550 mVSCE to 1,455 mVSCE. corresponds to the oxidation of the pore water to oxy-
This was not accompanied by any noticeable increase gen, where Etp is the equilibrium potential between
in AE. The rate of AE and half-cell potentials re- water and oxygen.29
mained reasonably constant between 2 days and The passive oxide present on the steel is a con-
17 days, dependent upon the cover thickness, after ductor of charge; therefore, it is only able to maintain
which time a sudden increase in acoustic emissions a small potential difference (~1 V)19 across the pas-
occurred and in some instances a small drop in the sive film, significantly less than the potential applied
electrochemical potential. to achieve the desired current density. Consequently,
Breaking open a prism ~1 day before an increase impressing a current raises the potential on the pore
in AE revealed that the rebar showed no evidence of water side of the film to a more noble value within
corrosion on the metal surface, despite the very posi- the oxygen evolution potential range.19 Figure 5
tive half-cell potential. However, opening the sample shows the shift from equilibrium values (Ecorr and Icorr)
1 day to 2 days after the onset of emission revealed necessary to obtain the desired current density. The
corrosion of the rebar surface in all cases. This con- potential difference (Eapplied) must be supplied by an
firms first that the AE technique is a good method for external source and can be expressed as shown:
detection of corrosion in reinforced concrete, and
second, steel remains passive at high potentials in
the absence of any aggressive specie. A typical ex- E applied = (E ′a – E corr ) + ( E corr – E ′c ) (4)
ample of the increase in AE is shown in Figure 4.
An understanding of the electrochemical influ- where Ea′ is the imposed anodic potential and Ec′ is
ence involved in impressing an anodic current can be the imposed cathodic potential.

CORROSION—Vol. 60, No. 2 207


CORROSION ENGINEERING SECTION

and the water cement ratios associated with the dif-


ferent strengths. The source of the chloride ion was
the electrolyte.
In concretes where chlorides are present, a
fourth region is introduced into the potential-pH
diagram (Figure 6),28 denoting the range of potentials
and pH over which pitting corrosion is possible. This
introduces a rupture potential, Er, on the anodic po-
larization curve at which passivity breaks down and
localized corrosion can initiate on the rebar surface.
At low concentrations of chlorides, Er may exist at
high potential values; thus, corrosion of depassivated
areas might occur concurrently with oxygen evolu-
tion where passivity is intact. In Figure 7, the
half-cell values have been plotted together with cu-
mulative AE energy over the duration of the test for
a representative beam. AE energy is measurement in
atto Joules of the energy contained in each hit. Fig-
FIGURE 4. Increase in acoustic emission signifying onset of
ure 7 clearly shows the change in potential imposed
corrosion.
by applying the current to the rebar at the start of
Day 1. It is also evident that the potential remains
fairly constant until the end of the test where the
value falls, corresponding with a sudden increase in
AE energy. The imposed potential is always greater
than Line A, corresponding to Etp for steel in a solu-
tion of pH 13.5, suggesting that oxygen evolution is
probably occurring in passive areas of the beam.
Figure 8 illustrates the effect of cover thickness
on the time taken for corrosion to initiate at an ap-
plied current density of 100 µ A/cm2. It is apparent
that the cover thickness has a significant influence
on the time until corrosion initiates, which may be
related to the diffusion of the chloride ion into the
cover. It was assumed that initiation of corrosion
would be immediate in the case of zero cover, where
access to chlorides is unrestricted. The applied volt-
ages were in the range of 2.98 V to 7.25 V, with an
average of 4.7 V. The variation in voltage is likely to
be due to a combination of factors; hence, no direct
correlation between voltage, strength, rebar diameter,
or cover thickness was observed.
FIGURE 5. Change in potential required achieving the desired current
density. Corrosion
Localized breakdown of passivity combined with
high imposed anodic potentials can result in severe
Polarization localized corrosion of the rebar, confirmed by the ex-
Polarization of the rebar also induces a second perimental work. The percentage of surface corrosion
reaction, which under the test conditions eventually ranged between 0.2% and 39.8%, with a median
results in the corrosion of the rebars. It is well re- value of 4.9%. Corrosion products were typically
ported that chlorides move quicker through concrete located on the underside of the rebar, supporting the
when an electrical field is applied between an area of argument that corrosion was initiated due to the
low chloride concentration and an area of high con- presence of chlorides as this was the closest point to
centration.7,29-30 The thickness of the concrete and the the proceeding chloride front, assumed to be migrat-
applied voltage influences the depth of penetration of ing through the cover under the influence of the
chlorides over a period of time.31 Within the scope of applied electric field.
the experiment, the thickness of the concrete varied With the exception of three bars that had a cor-
between 16, 25, and 40 mm, and the applied voltage roded area of >16%, the extent of corrosion was gen-
was a function of both the resistivity of the concrete erally low as a result of terminating the test within a

208 CORROSION—FEBRUARY 2004


CORROSION ENGINEERING SECTION

FIGURE 7. Typical half-cell potential and AE results.

FIGURE 6. Potential-pH diagram for iron and water in the presence


of chlorides.

short time after initiation was indicated by the AE.


Consequently, potentials in all cases remained con-
sistently greater than Etp (Figure 7), implying that
oxygen evolution was probably occurring where pas-
sivity was still intact and anodic dissolution was oc-
curring where passivity had been compromised. This
suggested a split in the applied current used for the
oxidation of water and steel.
The split anode reactions will influence the theo-
retical mass loss calculated using Equation (1). A FIGURE 8. Time to corrosion initiation for various cover thickness
more reliable estimate of steel loss requires a modifi- and strength combinations.
cation to the terminology used in Faraday’s law. The
variable time, measured in seconds, is not the total
time duration of the applied current, but the dura- reported by Auyeung, et al.14 Similarly, other work-
tion (dc) of the corrosion activity from the start of ers32 observed underestimations when the impressed
depassivation, identified by the onset of AE, to the current was continued for a significantly longer pe-
termination of the test. Secondly, the value used for riod (70 days to 221 days) and the concrete contained
the current must be proportional to the area of the admixed chlorides promoting general corrosion. An
steel on the bar that has been depassivated (ac) and element of scatter is still apparent in Figure 9 due to
not the whole surface area (Ae). This assumes that the physical errors inherent in obtaining both mass
equal charge is used in the reduction of oxygen as in loss values. The mass loss as a result of passivity
the dissolution of steel and that the area covered in was considered to be negligible during the relatively
corrosion products has depassivated. The modified short experimental period and has been discounted.
version of Faraday’s law, developed for this work, is
expressed in Equation (5):
DISCUSSION
Midc a c Suitability of the Impressed Current Technique
Theoretical mass loss = (5)
A e zF Impressing an external current to accelerate the
corrosion of steel in concrete has been shown to
The ratio of theoretical and gravimetric mass losses reliably induce corrosion of the reinforcement.
obtained from Equations (1) and (5) has been plotted The analysis presented has been based upon the
in Figures 9 and 10. Good agreement (i.e., an ap- Pourbaix diagram and Pourbaix’s theoretical
proximate ratio of unity) is evident in Figure 9 where potentiostatic curves for steel in a solution at pH
passivity has been taken into account, whereas in 13.5.28 While the analyses provide a sound thermody-
Figure 10 the mass loss has been overestimated by a namic basis for understanding the corrosion reac-
factor between 4 and 200. Overestimations have been tions, their interpretation is limited to a rough guide

CORROSION—Vol. 60, No. 2 209


CORROSION ENGINEERING SECTION

sis results may not be directly comparable to non-


impressed tests; hence, these were not recorded.
The impact of the localized changes in pH on the
steel will depend upon the mobility of the hydrogen
ions, which can migrate from the rebar surface under
the action of the electric field, differential concentra-
tion gradients,33 or be diluted by the OH– reserves in
the matrix. The stability of the passive film in the ab-
sence of chlorides is dependent upon the pH of the
adjacent pore water. If significant electrolysis occurs,
the local reduction in pH induced by hydrolysis
may be sufficiently large to fall below the values re-
quired to maintain passivity of the steel.35 To investi-
gate this possibility, a phenolphthalein solution was
sprayed onto an area of concrete, which had been in
immediate contact with the rebar at a location where
FIGURE 9. Mass loss ratio, accounting for passivity. corrosion had not occurred. The results obtained
from the several beams tested indicated that at ter-
mination of the test the pH of the concrete in imme-
diate contact with the steel remained sufficiently
alkaline to support passivity. Hence, any change in
the pH of the pore solution due to electrolysis was
not considered to be significant to the durability of
the passive film over the relatively short test dura-
tions and within the applied voltage range. This is
also supported by Grimes, et al.,36 who found that
the pH of the concrete adjacent to noncorroded steel,
having undergone impressed current corrosion tests,
was 11.9 to 12.1.
It is generally accepted that the acidity produced
during chloride-induced corrosion results in the for-
mation of soluble oxides, which diffuse away from
the anode to areas of lower pH within the cement
matrix. The corrosion product found on the rebars
FIGURE 10. Mass loss ratio, not accounting for passivity. upon opening was examined visually and can be de-
scribed as being typically a ‘wet’ green/black oxide
paste surrounded occasionally by a brown oxide on
because the basis of the graphs originate from the periphery. On exposure to the air, the green oxide
pure metals in pure, stirred solutions. Thus, they do turned into a ‘rust red’ color. The green/black color
not consider the possible effects of restricted mass and phase of the oxide would suggest the oxide was a
transfer; therefore, inherent differences exist between type of iron chloride complex, formed as an interme-
the two. diate stage of chloride-induced corrosion.6,37 An ap-
The primary objective of any accelerated tech- proximate measure of the pH of the oxide indicated
nique should be to resemble naturally induced corro- that it was only just acidic (pH = 6), which is sup-
sion in terms of the type of corrosion products and ported by other work where it is reported to be be-
method of attack. However, the similarities and de- tween 5.5 and 5.7.36 Oxides also were present at the
viations between the two mechanisms also must be concrete/steel interface, exhibiting the start of oxide
considered to enable a decision regarding its effec- migration into the cover. In a few cases, the longer
tiveness as a technique to be made. duration of applied current enabled oxides to diffuse
When an external electric field is applied to pas- further into the bulk matrix, typically around aggre-
sive steel embedded in concrete, electrolysis of the gates as opposed to through the paste, where they
pore water can occur at both electrodes.7 At the an- had precipitated into a solid orange oxide, probably
ode, the reaction previously discussed, involving the due to the alkaline environment. It is suggested that
evolution of oxygen, also generates hydrogen ions this action replicates the behavior of oxides in real
that diffuse into the pore water adjacent to the rebar structures where chloride-induced corrosion is char-
lowering the local pH.33 A reduction in the pH can acterized by localized rust stains on the surface,
lower the critical chloride concentration required to which often can be observed on deteriorating struc-
induce corrosion.34 Therefore, chloride content analy- tures. It is the formation of these higher oxides states

210 CORROSION—FEBRUARY 2004


CORROSION ENGINEERING SECTION

that have a large increase in volume, which induces against the direction of gravity. This resulted in gen-
the major stresses in the cover, eventually cracking eral corrosion occurring on the depassivated areas.
the concrete.38 In comparison with naturally induced corrosion,
While attack along the length of the beam was the technique differs so much that the potential of
localized, the morphology of the attack on the active the passive rebar during the initial stages is far re-
areas on the rebars resembled general corrosion. moved from the natural potential range. Further-
This may be due to the electro-migration of the chlo- more, the subsequent corrosion rate is not governed
rides to this region forming a sufficient quantity of by the mixed potential theory but by the applied
chlorides to induce general corrosion or, as the current. The applied current compensates for the
authors suspect, due to the early termination of the consequential discrepancy between the anodic and
test while corrosion was at a pre-pitting stage. In cathodic potentials and offers the benefit of reducing
trial experiments, the current was impressed for the sensitivity to changes in oxygen content, mois-
many days resulting in the formation of pits. In work ture content, or temperature as occurs naturally.
undertaken by González, et al.,21 corrosion was con- This enables good control over the corrosion rate for
tinued for a significantly greater duration (1 month), experimental purposes.
under the action of an impressed current where pit-
ting corrosion was reported. They also stated that the CONCLUSIONS
impressed current technique produced R values (de-
fined as the ratio of maximum penetration to average ❖ The scientific justification for accelerating corro-
penetration) similar to, although slightly higher than, sion using an impressed current is strong, dramati-
those obtained under nonimpressed conditions. cally reducing the initiation period required for
The primary electrochemical difference of this depassivation from years to days and fixing the
technique to naturally corroding systems is the rais- desired rate of corrosion without compromising the
ing of the potential to a value greater than Etp, where reality of the corrosion products formed. Based on
the corrosion rate does not correspond with an equi- the experimental work and argument presented, the
librium mixed potential or a potential obtainable following conclusions may be drawn.
under usual service conditions without applying an ❖ The impressed current technique has been con-
external source of energy. Hence, imposing such a firmed to be an effective and quick method of accel-
potential will result in the steel being in an artificially erating chloride-induced corrosion. However,
polarized state. While this had no apparent effect on electrochemistry behind the mechanism differs from
the bars even after a 17-day period, a gradual reduc- naturally induced corrosion.
tion in the pH due to electrolysis might occur to- ❖ Accounting for oxygen evolution and passivity
gether with migration of other ions, which may when applying Faraday’s law was shown to signifi-
accelerate the degradation process39 if continued over cantly improve the correlation between theoretical
an extended time period. and gravimetrical mass loss.
The impressed current method of corrosion ac- ❖ Oxygen is assumed to be evolved from the applied
celeration has many advantages, in addition to the current in the absence of chloride ions.
obvious savings in time and money. One advantage ❖ The acoustic emission technique was effective and
over other accelerated techniques is the ability to reliable at detecting the onset of corrosion. This view
control the rate of corrosion, which usually varies is supported by good prediction of mass loss by
due to changes in the resistivity, oxygen concentra- Faraday’s law taking into account the passive period
tion, and temperature. Any change in one of the vari- indicated by the AE measurement.
ables would be compensated for. For example, a ❖ The thickness of the concrete cover had a greater
change in the resistivity of the concrete as a result of influence than the concrete strength on the time re-
temperature fluctuations or evaporation of the pore quired for chlorides to permeate through the cover.
water can be counterbalanced by supplying a greater ❖ During the first few days of testing, the corrosion
voltage, thereby maintaining the desired corrosion was restricted to a small number of sites on the bar
rate. This removes much of the variation encountered surface, similar to naturally induced corrosion.
in corrosion measurements with time.
While the corrosion rate is fixed, the area of steel ACKNOWLEDGMENTS
depassivated and corroding cannot be controlled due
to the preferential flow path of chlorides between the This work was undertaken at the Centre for
aggregate/cement interface31 and the existence of Innovative Construction Engineering, Loughborough
preferential areas along the bar surface due to geo- University. The authors acknowledge the financial
metrical heterogeneities.40 Thus, the total mass loss assistance from the EPSRC and Balvac Whitley
varied between samples. The geometry of the corro- Moran Ltd., in addition to specialist equipment loan
sion cell was such that depassivation occurred on and technical assistance from Physical Acoustics
the underside of the rebar, closest to the electrolyte, Limited. Atkins Consultants Ltd. also gave assistance.

CORROSION—Vol. 60, No. 2 211


CORROSION ENGINEERING SECTION

REFERENCES 20. S.C. Das, “Computer-Aided Quantitative Interpretation of


Electropotential Survey Results for RC Structures,” in Proc. 6th
Int. Conf. on Structural Faults and Repair, ed. M.C. Forde
1. B. Boyard, C. Warren, S. Somayaji, R. Heidersbach, ASTM STP
(Edinburgh, Scotland: Engineering Technics Press, 1995), p. 209.
1065, “Corrosion Rates of Steel in Concrete,” eds. N.S. Berke,
21. J.A. González, C. Andrade, C. Alonso, S. Feliu, Cem. Concr.
V. Chaker (West Conshohocken, PA: ASTM International, 1990),
Res. 25 (1995): p. 257.
p. 174.
22. T.A. El Maaddawy, K.A. Soudki, J. Mater. Civil Eng. (2003): p. 41.
2. RILEM, Mater. Struct. 9, 51 (1976): p. 187.
23. M.S. Weng, S.E. Dunn, W.H. Hartt, R.P. Brown, Corrosion 38, 1
3. C.M. Hansson, Th. Frolund, J.B. Markussen, Cem. Concr. Res.
(1982): p. 9.
15 (1985): p. 65.
24. S.E. Dunn, J.D. Young, W.H. Hartt, R.P. Brown, Corrosion 40,
4. P.S. Mangat, B.T. Molloy, Mater. Struct. 25 (1992): p. 404. 7 (1984): p. 339.
5. C.L. Page, P. Lambert, P.R.W. Vassie, Mater. Struct. 24 (1991): 25. Z. Li, F. Li, A. Zdunek, E. Landis, S.P. Shah, ACI Mater. J. 95
p. 243. (1998): p. 68.
6. A.M. Neville, Mater. Struct. 28 (1995): p. 63. 26. A.D. Zdunek, D. Prine, “Early Detection of Steel Rebar Corro-
7. C. Andrade, Cem. Concr. Res. 23 (1993): p. 724. sion by Acoustic Emission Monitoring,” CORROSION/1995,
8. M.G. Grantham, M.J. Gray, J.P. Broomfield, “Chloride Diffusion paper no. 547 (Houston, TX: NACE International, 1995).
Modelling and Corrosion Rate Data for Lifetime Prediction in 27. H. Idrissi, A. Limam, J. Acoust. Emiss. 18 (2000): p. 307.
Marine, Highway, and Car Parking Structures,” in Proc. 7th Int. 28. M. Pourbaix, Lectures on Electrochemical Corrosion (New York,
Conf. on Structural Faults and Repair, ed. M.C. Forde NY: Plenum Press, 1973), p. 243.
(Edinburgh, Scotland: Engineering Technics Press, 1997), p. 29. R.K. Dhir, M.R. Jones, H.E.H. Ahmed, A.M.G. Seneviratne,
275. Mag. Concr. Res. 42, 152 (1990): p. 177.
9. C. Andrade, C. Alonso, A. Molina, Mater. Struct. 26 (1993): p. 30. C. Andrade, J.M. Diez, A. Alaman, C. Alsono, Cem. Concr. Res.
453. 25, 4 (1995): p. 727.
10. J.A. González, S. Algaba, C. Andrade, Br. Corros. J. 15, 3 31. N.R. Buenfeld, S. El-Belbol, Mag. Concr. Res. 43, 155 (1991):
(1980): p. 135. p. 135.
11. N.I. Fattuhi, Cem. Concr. Res. 18 (1988): p. 426. 32. C. Alonso, C. Andrade, J. Rodriguez, J.M. Diez, Mater. Struct.
12. S. Ahmad, B. Bhattacharjee, R. Watson, ACI Mater. J. 94 31 (1998): p. 435-441.
(1997): p. 311. 33. K.G.C. Berkeley, S. Pathmanaban, Cathodic Protection of Rein-
13. G.J. Al-Sulamimani, M. Kaleemullah, I.A. Basunbul, forcement Steel in Concrete (London, U.K.: Butterworth & Co.,
Rasheeduzzafar, ACI Struct. J. 87 (1990): p. 220. 1990), p. 69.
14. Y. Auyeung, P. Balaguru, L. Chung, ACI Mater. J. 97 (2000): p. 34. P.R. Vassie, “Non-Destructive Evaluation of the Corrosion of
214. Concrete Reinforcement by Chlorides,” in Proc. Symp. Corros.
15. J.G. Cabrera, Cem. Concr. Comp. 18 (1996): p. 47. of Reinforcements in Concrete Construction (London, U.K.:
16. W.H. Hartt, “Laboratory Method for Corrosion Testing of Rein- Society of Chemical Industry, 1979).
forced Concrete Using Impressed Current,” CORROSION/79, 35. A. Bentur, S. Diamond, N.S. Berke, Steel Corrosion in Concrete
paper no. 133 (Houston, TX: NACE International, 1979). (London, U.K.: E&FN Spon, 1997).
17. ASTM G109, “Test Method for Determining the Effects of 36. W.D. Grimes, W.H. Hartt, D.H. Turner, Corrosion 35, 7 (1979):
Chemical Admixtures on the Corrosion of Embedded Steel Rein- p. 309.
forcement in Concrete Exposed to Chloride Environments” 37. RILEM Report, “Corrosion of Steel in Concrete,” ed. P. Schiessl
(West Conshohocken, PA: ASTM International, 1993). (London, U.K.: Chapman and Hall, 1988).
18. M. Pourbaix, Atlas of Electrochemical Equilibrium in Aqueous 38. ACI 222R-96, “Corrosion of Metals in Concrete,” ACI (1997).
Solutions, trans. from French by J.A. Franklin (New York, NY: 39. M. Castellote, I. Llorente, C. Andrade, C. Alonso, Mater. Struct.
Pergamon Press, 1996), p. 409-410. 36 (2003): p. 81.
19. J.C. Scully, The Fundamentals of Corrosion (Oxford: Pergamon 40. J.A. González, S. Feliu, P. Rodriquez, E. Ramirez, C. Alonso, C.
Press, 1990). Andrade, Mater. Struct. 29 (1996): p. 40.

212 CORROSION—FEBRUARY 2004