Professional Documents
Culture Documents
Scully2000 Polarization Resistance Method For Determination of Instantaneous Corrosion Rates
Scully2000 Polarization Resistance Method For Determination of Instantaneous Corrosion Rates
J.R. Scully*
ABSTRACT INTRODUCTION
The polarization resistance method for the determination of A variety of methods such as electrical resistance,
instantaneous corrosion rates of metals were reviewed. The gravimetric-based mass loss, quartz crystal microbal-
assumptions in electrode kinetics that govern the connection ance-based mass loss, electrochemical, and solution
between the slope of steady-state applied electrochemical analysis methods enable the determination of corro-
potential-applied current density (Eapp-iapp) measurements
sion rates of metals. The polarization resistance
(i.e., the polarization resistance) and the corrosion rate were
method, based on electrochemical concepts, enables
restated. Electrochemical impedance, as well as statistical
and spectral electrochemical noise methods for obtaining determination of instantaneous interfacial reaction
polarization resistance, also were discussed. Traditional rates such as corrosion rates and exchange current
sources of error such as high excitation voltage amplitudes, densities from a single experiment. There are a vari-
insufficiently slow voltage ramp rates, high alternating cur- ety of methods capable of experimentally determining
rent (AC) frequencies, inadequate polarization hold periods, instantaneous polarization resistances such as
high solution resistance, presence of parallel reduction- potential step or sweep, current step or sweep,
oxidation reactions, and nonuniform current and potential impedance spectroscopy, as well as statistical and
distributions were examined with the goal of defining some spectral noise methods. All of these methods utilize
of the conditions and circumstances where these complicat-
either two-, three-, or four-electrode electrochemical
ing factors are important. Other complicating factors such as
cells. Instantaneous corrosion rate information can
diffusion-controlled reactions, potential-dependent film cover-
ages, and adsorption pseudo-capacitance that affect
be obtained from such relatively rapid electrochemi-
electrochemical reaction rates were discussed. Their influ- cal measurements in short time periods. Other
ence on measured interfacial resistance values and methods such as the electrical resistance change of
subsequent determination of corrosion rates was discussed. the solid electronic conducting phase, gravimetric
and quartz crystal microbalance measurements of
KEY WORDS: current-step method, impedance method,
mass loss, and solution analysis to detect metallic
instantaneous corrosion rate, linear polarization resistance,
cations released into the ionic conducting phase all
noise resistance, polarization resistance, potential-step
method, potentiodynamic method, power spectral density,
provide historical or integrated information on the in-
solution resistance, spectral noise resistance stantaneous corrosion rates that have occurred over
some period of time. Therefore, the derivative of mul-
tiple measurements over exposure time must be used
Submitted for publication February 1998. Presented as paper no. to obtain rate information. Therefore, instantaneous
304 at CORROSION/98, March 1998, San Diego, CA. rates can not be determined by a single measure-
* Center for Electrochemical Sciences and Engineering, Department
of Materials Science and Engineering, Thornton Hall, University of ment of these types and electrochemical methods, for
Virginia, Charlottesville, VA 22903-2442. instantaneous rate determination remain extremely
0010-9312/00/000039/$5.00+$0.50/0
CORROSION–Vol. 56, No. 2 © 2000, NACE International 199
CRITICAL REVIEW OF CORROSION SCIENCE AND ENGINEERING
Eω
Zω = (5)
iω
Rp jωCR2p
Z ω = Rs + _ (7)
1 + ω2 R2p C2dl 1 + ω2 R2p C2dl
(a)
where = 2πf is the frequency of the applied signal
(rad/s), f is the frequency of the applied signal (Hz),
and Cdl is the interfacial capacitance (F/cm2). The
complex plane, Bode magnitude, and phase plots
resulting from a circuit such as shown in Figure 3(a)
and described by Equation 7 are shown for three
different values of Rp in Figure 4. At very low
frequencies:
Zω→0 ω = Rs + Rp (8)
Rp = Rct + RD (11)
(a)
A very low frequency or scan rate may be required
to obtain Rp defined by Equation (11) under certain
circumstances (Figure 6). Here, a Ieff of 0.1 cm and
D = 10–5 cm2/s requires that a frequency < 0.1 mHz
be implemented to obtain Rp from |Z()| at the zero
frequency limit. Hence, a common experimental
problem for diffusion-controlled electrochemical reac-
tions is that extremely low frequency (or scan rates)
are required to complete the measurement of Rp.
When Rp is defined by Equation (11), the Stern
approximation of Equations (2) and (3) must be
modified to account for the Tafel slope for the anodic
or cathodic reactions under diffusion-controlled con-
ditions (i.e., →∞). In fact, Equation (1) becomes
invalid. Moreover, it has been argued that the use of
Rp in the Stern equation breaks down in the case of
mixed or partial diffusion control.
(b) Another relevant situation to corrosion rate de-
FIGURE 5. (a) Bode magnitude and (b) Bode phase angle plots for termination occurs when an adsorbed intermediate
hypothetical corroding interfaces with RP = 100 Ω, C dl = 100 µF, and participates in the overall reaction process (Figure
RS = 1, 10, or 100 Ω using the electrical equivalent circuit model of 3[c]).22 Here, Rct is a faradaic resistance proportional
Figure 3(a). to a charge-transfer process that occurs instanta-
neously given a certain adsorbed intermediate
coverage. Rs and Cdl have their usual meanings. Cf
is a new capacitance term that depends upon the
Either the anodic or cathodic half-cell reaction
change in the charge stored in the adsorbed interme-
can become mass-transport limited and restrict the
diate film resulting from a gradual change in the film
rate of corrosion at Ecorr. The presence of diffusion-
coverage with potential.2 The adsorbed film has a
controlled corrosion processes does not invalid the
charge associated with it in this case because it is
electrochemical impedance spectroscopy (EIS)
formed by a charge-transfer process. Recall that the
method but does require extra precaution and a
source of any fixed capacitance is a charge that is
modification to the circuit model of Figure 3(a). In
proportional to potential. Hence, the rate of change of
this case, the finite diffusional impedance is added in
surface charge is proportional to the rate of change
series with the usual charge-transfer parallel resis-
in interfacial potential. The capacitance (C) is the
tance shown in Figure 3(b). The transfer function for
proportionality factor that is independent of the rate
the frequency dependent finite diffusional impedance
of change in charge or potential assuming a fixed
(ZD) has been described by:21
capacitance:2
tanh jωs
ZD ω = RD (10) dq dt dqT
jωs C= = (12)
dE dE dt
dqF dθ dt dθ
Cφ = = qM = qM (13)
dE dE dE dt
The reference electrode noise can be defined sepa- momentarily, which induces a further difference in
rately as the electrochemical voltage noise between anodic and cathodic half-cell currents on the first
two nominally identical reference electrodes.31 If the electrode. Current now flows between the first and
reference electrode noise is low, then the √2 correc- second electrodes such that the sum of anodic and
tion factor is not needed. In either case, the third cathodic half-cell currents over all patches on both
electrode (reference electrode) is connected to the electrodes is equal. When the interfacial resistances
first two via a high-impedance voltmeter. These ar- return to normal values over all patches, the
rangements enable simultaneous recording of the potential difference between the two electrodes is
galvanic current with time and the galvanic couple eliminated and so is the measurable current between
potential versus time. The standard deviation of the the two electrodes. Bertocci argued that the external
voltage noise divided by the standard deviation of the current fluctuation measured between two identical
current noise has been proposed to yield a statistical electrodes is identical to the fluctuation in one elec-
parameter called the noise resistance (Rn).23-30,32 Fur- trode.28-29 Others have argued using concepts of
ther analysis of simulated noise data has led to the mixed potential theory that, at worst, the current
conclusion that the ratio of the standard deviations sampled is only one half of the total for equal-sized
of the current and voltage noises measured between electrodes.
two identical electrodes can be normalized by surface Theoretical relationships establishing the con-
area by multiplying by √AvAi:29 nection between Rn and Rp have been sought by
several researchers,29,33-36 but their validity has been
σV(meas) questioned. A great concern has been that the largest
Rn = A νA I (14) current peaks would occur during the most rapid
σI(meas)
voltage fluctuations since the electrode interface con-
tains a capacitance through which current can be
where V(meas) is the standard deviation of the voltage shorted.28-29 Thus, when voltage fluctuations are
noise, I(meas) is the standard deviation of the current rapid, the measured noise current will be shorted
noise, and Av and AI are the surface areas of the elec- through the interfacial capacitance assuming a
trodes used for voltage and current measurement, simple electrical equivalent circuit model consisting
respectively. Correlations between this parameter of a two-parallel resistor-capacitor network describ-
and conventionally determined Rp and mass- ing the interface for each electrode connected in
loss-based corrosion rates have been obtained.24,30 series through Rs. This situation would lead to the
Unfortunately, experimental confirmation of the area lowest impedance between the two electrodes during
normalization factor has not been performed exten- the most rapid voltage fluctuations that, in turn,
sively. Recall that in the case of a Rp determined from produce the greatest current fluctuations. The
E-iapp data or EIS data at the zero frequency limit, theoretical maximum measured current would be
measured resistance can be multiplied by electrode given by the voltage fluctuation divided by Rs. The
area and will yield the same area normalized Rp over outcome would be a statistical Rn parameter that is
a broad range of electrode areas. proportional to, or heavily influenced by higher fre-
Moreover, the correlation has lacked a rigorous quency data. Indeed, Rn is found to equal an absolute
fundamental foundation for correlating Rn with corro- impedance at some frequency that depends on the
sion rate, despite the intuitive connection between frequency of the voltage fluctuations and the resis-
sV and sI given by the proportionality factor Rn. It is tive-capacitive (RC) time constant of the electrode
clear that the surface of one freely corroding elec- interface in one study of simulated noise.28 Unfortu-
trode could be divided into areas that experience nately, a Rn value obtained at high frequency would
fluctuations in interfacial resistance that produce be smaller in magnitude than the Rp obtained at the
changes in anodic and cathodic half-cell reaction zero frequency limit. Hence, it would not represent
rates in any one patch. The electrode potential must the desired zero frequency limit interfacial resistance,
change then in each patch to drive the half-cell reac- Rp. Indeed, such underestimations in the true value
tions such that the sum of all the anodic half-cell of Rp have been observed experimentally.30,32
currents from all patches equals the sum of all Recently, a more rigorous theoretical and experi-
cathodic half-cell currents, regardless of whether the mental analysis has been made comparing the
source of cathodic half-cell current is from capacitive spectral noise resistance (Rsn[v]) obtained at each fre-
discharge or electrochemical reaction.28 Some global quency with the Rp obtained from the zero frequency
change in potential also occurs on the electrode. If limit of impedance data |Z(v = 0)|, as well as the
the first electrode is now connected to a second elec- frequency dependent impedance of two electrodes.37-41
trode whose interfacial properties and global The spectral noise resistance (Rsn[v]) is determined by
electrode potential do not change on their own at the taking the square root of power spectral density of
same instant in time and by the same degree as on the voltage noise, VPSD, (V2/Hz)1/2 and dividing it by
the first electrode, then a galvanic cell is created the square root of power spectral density of the cur-
VPSD
Rsn ω = (15)
IPSD
low-frequency plateau can be approximated by: In separate but parallel efforts, Gabrielli,53
Macdonald,52 and Townley54 each discussed the
1 appropriate choice of voltage scan rates for accurate
fmax < fbp ≅ (17) Rp determination from small amplitude cyclic
2πC Rs + Rp
voltammograms. Each investigator considered a
simple equivalent circuit (Figure 3[a]). Macdonald
where fbp is an approximation of the lower breakpoint and Townley derived the current response of the
frequency (Hz) and fmax is the maximum test fre- standard three-element electrical equivalent circuit
quency (Hz). Thus, it can be seen from Equation (17) to describe an electrochemical interface (Figure 3[a])
that increasing values of C, Rs, or Rp dictate that a to a small amplitude triangular voltage excitation
lower fmax is required to accurately obtain Rp + Rs on (e.g., a potential sweep at fixed scan rate with a
the low-frequency plateau as Z→0 (Figure 4). Capaci- reverse sweep at some peak potential). The E-iapp
tances may become larger than that expected for the response of this circuit to the triangular voltage exci-
Cdl when adsorption C are present. Such a C may tation was shown to be a complex function of circuit
be caused by an adsorbed intermediate with some elements, ∆Epp, and the voltage sweep rate. The rela-
fractional monolayer coverage. Corrosion of steel in tionships between the various ∆E/∆iapp slopes (R) and
concrete can complicate LPR measurements owing to scan rate were determined. Macdonald characterized
very large apparent capacitances that actually may the E-iapp response by the diagonal resistance associ-
represent the finite diffusional impedance associated ated with iapp at the peak potentials, Rd. The apparent
with mass transport control of the oxygen reduction resistance (Rapp) determined from the tangent to the
reaction.57 One way that Rp may be increased is by E-iapp curve at the end of the upward sweep before
passivity. Another is by diffusion-controlled corrosion the downward sweep (and vice versa), Rapp, and the
such that Rp = Rct + RD. A 1-mHz frequency typically is hysteresis current. Townley obtained Rmeas, which is
chosen as a reasonable initial choice of fmax, but it is similar to the tangent resistance, Rapp:
obvious from Equation (17) that either lower or
higher frequencies may be required depending upon
1 1 2R2pC e a∆E pp/vb – 1
the exact values of Rs, Rp, and C. Since the magni- = + (20)
tudes of C, Rs, and Rp are not known explicitly, a Rd Rs + Rp ∆E Rs + Rp e a∆E pp/vb + 1
priori prudence dictates that fmax be selected as one
tenth of the estimated fbp. Mansfeld and Kendig pro-
posed that fmax determined from the impedance
1 1 2Rp 1
method defines the maximum voltage scan rate (vmax) = + (21)
for the potentiodynamic method.43 This derivation is Rapp Rs + Rp Rs Rs + Rp a∆E pp/vb
1+e
based on the assumption that the vmax should not
exceed the maximum rate of change of voltage with
time during the sinusoidal voltage excitation at fmax.
1 1 Rp
= + e –t/τ (22)
Rmeas Rs + Rp Rs Rs + Rp
π∆Eppfmax
vmax = (18)
10
In these equations, a = Rs + Rp, b = RsRpC, t is time,
Here, ∆Epp is the peak-to-peak voltage excitation. v is sweep rate, and τ = 1/RsC + 1/RpC. Time (t) in
Therefore, vmax is 0.16 mV/s assuming a 10-mV Equation (22) can be equated to the sweep rate since
peak-to-peak amplitude for the equivalent circuit t = ∆Epp/v. Therefore, all three equations include a
shown in Figure 3 with Rp = 1,000 Ω, Rs = 10 Ω, and term that is independent of the voltage scan rate and
C = 1,000 µF. If ∆E is ±5 mV, then the time required a second transient term that depends on voltage scan
for one complete cyclic voltammogram at such a rate or time relative to the RC time constant of the
potential sweep rate is 125 s. However, for a slowly equivalent circuit representing the electrochemical
corroding electrode with the Rs + Rp = 10,010 Ω, vmax interface. The scan rate dependent term becomes
becomes 0.0157 mV/s and the time required is negligible at low scan rates. These analyses predict
1,250 s. Similarly, a frequency above the high- an increase in current hysteresis and deviation of Rd
frequency breakpoint (fhf) must be applied to obtain and Rmeas from Rp + Rs at high scan rates, as shown in
Rs. Typically, fapp must be in the KHz range to deter- the theoretical cyclic E-iapp data of Figure 9(a). Figure
mine accurately Rs: 9(b) shows actual experimental results from a Cu-Ni
alloy in flowing seawater. A plot of experimental 1/Rd
1 and 1/Rapp data vs scan rate is shown in Figure 9(c)
fhf ≥ (19) for the Cu-Ni system. Macdonald and Townley de-
2πCRs
duced (Equations [21] and [22]) that it is theoretically
possible in the ideal case to determine Rp + Rs from transport in corrosion may not be accounted for
the tangent to the E-iapp curve at the end of the for- properly at very fast scan rates because the surface
ward and reverse sweeps at much higher scan rates coverage and diffusional boundary layer does not
than predicted by Equations (17) and (18). The ideal have time to react fully and adjust to a very fast
scan rate maxima can be seen in Figures 10(a) and change in voltage. Rct may be determined instead of
(b) for the same circuit parameters as examined Rp at the DC limit if the scan rate is so fast so as to
above (i.e., Rs = 10 Ω, Rp = 1,000 Ω, and C = 10–3 F). freeze in the coverage (Figure 3[c]). Unfortunately, a
Specifically, it is shown that 1/Rmeas and 1/Rapp ap- sufficiently fast scan rate that “freezes-in” an inter-
proach the values 1/(Rs + Rp) at scan rates as high as mediate coverage also may result in large capacitive
100 mV/s when the slope of the E-iapp plot is care- currents that could result in Rd values far below Rp.
fully taken at the very end of the forward or reverse Moreover, as the scan rate or frequency is lowered,
voltage scan just before reversal. However, accurate the effects of an adsorbed intermediate that gives rise
determination of 1/(Rs + Rp) from experimental 1/Rd to an adsorption C and dependency of Rp on the
values still requires a scan rate < 1 mV/s, which is in sum of Rct and the parallel impedance of C||RF be-
agreement with the predictions from Equations (17) come factors in the resistance measured (Figure 3[c]).
and (18). Indeed, Macdonald observed that an accurate value
The equivalence of Rapp to (Rs + Rp) at fast scan of Rs + Rp was, in fact, not obtained from Rapp or Rd
rate must be treated with precaution in real corrod- determined at fast scan rates for a corroding Cu-Ni
ing systems. Such an equivalence is only true for alloy in flowing seawater (Figures 9[b] and [c]).52 The
ideal, fast charge-transfer-controlled corrosion pro- experimental case of Cu-Ni in seawater shows that a
cesses and ideal finite Cdl (e.g., when Rp = Rct). The much lower scan rate is required despite the more
presence of mass transport control or an adsorption optimistic predictions of Equations (20) through (22)
C caused by formation of an adsorbed intermediate (Figures 10[a] and [b]). In contrast, a model like that
that participates in the corrosion process may com- of Figure 3(a) for an ideal RpC interface, where Rp =
plicate the results and the ability to use faster scan Rct, suggests that 4 mV/s to 20 mV/s scan rates
rates. Roles of adsorbed intermediates and mass would be more than slow enough to obtain an Rapp
Thus, Rct = Rp in the simple cases where ideally fast right-hand second term by conducting experiments
charge transfer occurs and the coverage is fixed or at fast scan rate to freeze-in or fix the coverages.66
when there is no coverage dependency. However, the lower impedance of the Cdl and
A more complicated result is obtained at slow adsorption C at high frequency may complicate
scan rate or low AC frequency when the coverage can measurements by switching capacitive currents into
change with potential. Again, the linear approxima- the iapp measurement. Furthermore, it also must be
tion is enabled as ∆E becomes small: recognized that the Tafel slopes presented in the first
right-hand term of Equation (32) still would need to
be the “iso-coverage” Tafel slopes, ideally obtained by
∂Iapp 1 1 1 experiments where the coverages were fixed at their
= = 2.3Icorr + +
∂E Rp β aθ(a)corr β cθ(c)corr values present at Ecorr. Unfortunately, a and c rarely
(32) are obtained under iso-coverage conditions in practi-
1 ∂θ a 1 ∂θ c cal corrosion situations.67-71 Iso-coverage Tafel
Icorr –
θ a ∂E θ c ∂E parameters have been produced by correction of HER
reaction rate data by exploiting hydrogen absorption
and permeation through iron foils to control cover-
The first right-hand term is 1/Rct (e.g., the instanta- age.67 Here, foil thickness decreased as current
neous portion related to the faradaic current) and density is increased to change permeation flux such
the whole right-hand expression gives 1/Rp.60-61 The that surface coverage is held constant over a range of
1/(∂/dE) terms are typical nonzero contributions at reaction rates. Consequently, determination of corro-
Ecorr when there are potential dependent coverages sion rates in the presence of adsorbed intermediates
that are nonzero near Ecorr. There is absolutely no is not a simple matter of obtaining Rct at high fre-
reason that the coverages should be zero at Ecorr quency or scan rate, where ∂a/∂E equals zero
since the corrosion potential is an “open-circuit” po- because might be frozen during such rapid poten-
tential established at a specific potential somewhere tial changes. As shown here, another source of errors
between the respective reversible electrode potentials is the values of a and c obtained in slow scans at
of the cathodic and anodic half-cell reactions. Spe- high overpotential because this mixes ∆E-iapp data
cifically, Ecorr is established at the potential where iaa where the coverage is frozen-in (e.g., Rct) with E-
A = icc A. Therefore, it is clear that Rp may be > or log(iapp) Tafel data (e.g., a and c) where potential
< Rct, depending whether the partial derivatives dependent coverages may differ from those at Ecorr.
∂a/∂E and ∂c/∂E are positive or negative quantities An additional problem is that there is now another
as the applied potential is swept in forward and re- reason why ∂i/∂E should not be linear even as ∆E
verse directions. This can account for the common approaches zero if the coverage changes significantly
observation that Rp > Rct in the case of so-called with potential over the same potential range. More-
pseudo-capacitive behavior,22 as shown in the circuit over, there is no reason why E-log (iapp) data obtained
model of Figure 3(c) containing an adsorption over a large potential range should be linear on a
pseudo-capacitance. Moreover, it is possible that semi-log plot when the coverage changes with poten-
Rp < Rct,64 as in the case of so-called pseudo-inductive tial. Nonstraight Tafel data often are observed.68,71
behavior where an inductance is observed at low fre- Theoretical E-log iapp data with changing intermediate
quency (Figure 6[b]63). The condition, Rp > Rct, has coverage are shown in Figure 11. The data is pre-
been seen by subtracting the Cdl from the faraday im- sented for charge-transfer-controlled oxidation
pedance data.65 Thus, Equation (32) supports the kinetics over a fraction of the surface area given by
view that Rp at the zero frequency limit can be < or coverage a. The E-log(iapp) curve was created under
> Rct, depending on how the coverages that factor the following conditions. Anodic coverage (a) is com-
into the anodic and cathodic reaction rates change monly an exponential function of potential as
with potential. The differences between Rp and Rct are governed by the following expression assuming a
observed sometime to decrease when the magnitude Langmuir isotherm. This expression yields a nonzero
of the voltage perturbation is reduced such that the coverage at E = Ecorr:
coverage is negligibly altered.64 An alternative view of
the effect of voltage perturbation in the case of in-
hibitors is that the inhibitor is desorbed irreversibly θa F E – Erev
by potential scanning above Ecorr such that the corro- = kexp (33)
1 + θa RT
sion rate is increased.64 However, such phenomena
also might be understood using Equation (32) since
Rp would be expected to be similar to Rct when the As an example, a hypothetical E-iapp relationship is
voltage perturbation is smaller. Also, physically unre- considered for corrosion involving an overall oxida-
alistic parameters such as the electrical inductor do tion half-cell reaction (M + OH– → MOH+ + 2e–) that
not have to be invoked. It is tempting to eliminate the depends on the coverage a as described in Equation
other situations can be envisioned that have not The Wagner parameter (W) is the ratio of the
been covered here, such as inhibiting species desorb- kinetic resistance to the ohmic resistance. The W is
ing as a function of potential, coverage governed by the ratio of the true polarization slope given by the
Frumkin or Temkin isotherms, reaction mechanisms partial derivative, ∂Etrue/∂iapp, (evaluated at the
that change near Ecorr, or multiple-step, single- overpotential of interest at constant pressure, tem-
electron-transfer reactions that have other coverage perature, and concentration) divided by the
dependencies, etc. The purpose of the present ex- characteristic length and the Rs.2,74
ample was only to illustrate the role of changing
coverage in creating a difference between Rct and Rp
κ ∂E
and in adding complexities to the determination of W= (35)
L ∂i
Tafel slopes.
Rp still is often estimated from slow scans ap-
proaching the DC limit by taking a tangent to the where (or 1/) is the specific solution conductivity
nonlinear E-iapp data near Ecorr. The Stern-Geary (Ω-cm)–1 and L is a characteristic length (cm), or the
approximation still is utilized to estimate icorr using dimension of “irregularity.” Its value marks the tran-
an apparent B factor involving apparent Tafel slopes sition from the primary to the secondary current
that are imprecise. Such data still often agrees with distribution. When the Wagner number is much less
gravimetric mass loss and solution analysis data. than 1, the ohmic component dominates and current
Two reasons emerge. Although the terms 1/(∂/dE) and potential distributions are governed primarily by
in Equation (32) are often nonzero at Ecorr, they often cell geometry. When it is much larger than 1, the
are small. Moreover, the weak dependence of kinetic component dominates and the resistance of
corrosion rate on the parameter B in the Stern the interface primarily governs the current flow from
approximation permits imprecise estimates of Tafel counter electrode to working electrode. In practice,
slope to be tolerated in many cases. Of course, Rp the primary current distribution is said to exist when
values obtained at extremely low-frequency or scan W < 0.1, and the secondary current distribution ex-
rates should be used in the modified Stern equation ists if W > 102. The partial derivative, ∂E/∂i, can be
in the case of total mass transport control of either taken from any expression describing the interfacial
the anodic or cathodic reaction (with an infinite Tafel potential (Etrue) as a function of iapp or obtained as a
slope for the mass-transport-controlled reaction) as Rp from experimental E vs iapp data. The partial de-
well as in the case of passive dissolution. Use of the rivative, ∂E/∂i, has the units of Ω-cm2. The Rs
Rct obtained at higher frequency or scan rate yields expressed in the same units is given by (/L). The
the value of corrosion rate that would be pertinent if resulting Wagner number in the low-overpotential
both anodic and cathodic reactions were under in- region where the linear E-i approximation can be
stantaneous charge-transfer control. Rct should be applied is:
used in the Stern approximation if Tafel slopes under
charge-transfer-controlled conditions are utilized.66 Rp
In some cases, effects of inhibitor concentration on W= (36)
Rs
iron corrosion rate were best assessed using Rct as
opposed to Rp.66
Thus, the experimenter interested in estimating the
Effect of Solution Resistance current distribution “regime” pertinent to their polar-
Another frequently encountered complication is ization cell can estimate whether the primary or
the need to correct polarization data for errors that secondary current distribution applies from knowl-
arise from the contribution of Rs. Rs contributes to a edge of Rp and Rs. The corrosion engineer can use
voltage error as well as a scan rate error.2 Since the this information in the following way. If the primary
applied potential is increased by an ohmic voltage current distribution applies (W < 0.1), then current
component, an apparent value of Rp is obtained, distributions are likely to be very nonuniform unless
which overestimates Rp by an amount equal to Rs. an ideal cell geometry leading to a uniform primary
Consequently, the corrosion rate is underestimated. current distribution is used. Otherwise, errors in
polarization resistance and other kinetic parameters
Nonuniform Current and Potential Distributions are then likely because the electrode area actually
to Polarization Resistance Probes undergoing polarization differs from the total area.
A dimensionless parameter known as the Low scan rate data at the zero frequency limit may
Wagner number is useful for qualitatively predicting yield the same apparent value of Rp + Rs, but the area
whether a current distribution will be uniform or required to compute a true Rp value is uncertain.
nonuniform.2,74-75 This parameter helps to answer the Errors in Rp from nonuniform primary or second-
question of which current distribution applies to the ary current distributions have been documented.76
electrochemical cell: primary, secondary, or tertiary? Data presented in the literature (Figure 12) show
17. A.J. Bard, L.R. Faulkner, Electrochemical Methods: Fundamen- 61. I. Epelboin, M. Keddam, H. Takenouti, J. Appl. Electrochem. 2
tals and Applications (New York, NY: John Wiley and Sons, 1980). (1972): p. 71-79.
18. F. Mansfeld, Corrosion 36, 5 (1981): p. 301. 62. J.O’M. Bockris, D. Drazic, A.R. Despic, Electrochim. Acta 4
19. F. Mansfeld, M.W. Kendig, S. Tsai, Corrosion 38 (1982): p. 570. (1961): p. 315.
20. ASTM G 106, vol. 03.02, “Practice for Verification of Algorithm 63. E.J. Kelly, J. Electrochem. Soc. 126, 12 (1979): p. 2,064-2,075.
and Equipment for Electrochemical Impedance Measurements,” 64. W.J. Lorenz, F. Mansfeld, Corros. Sci. 21, 9 (1981): p. 647-672.
Annual Book of ASTM Standards (West Conshohocken, PA: 65. H. Schweikert, W.J. Lorenz, H. Friedburg, J. Electrochem. Soc.
ASTM). 127 (1980): p. 1,693.
21. D.R. Franceschetti, J.R. Macdonald, Electroanal. Chem. 101 66. I. Epelboin, C. Gabrielli, M. Keddam, H. Takenouto, ASTM STP
(1979): p. 307. 727, “Electrochemical Corrosion Testing,” eds. F. Mansfeld, U.
22. S.R. Taylor, E. Gileadi, Corrosion 51 (1995): p. 664. Bertocci, (West Conshohocken, PA: ASTM, 1981), p. 150-166.
23. D.A. Eden, A.N. Rothwell, J.L. Dawson, “Electrochemical 67. R.N. Iyer, H.W. Pickering, J. Electrochem. Soc. 137, 11 (1990):
Noise for Detection of Susceptibility to Stress Corrosion Crack- p. 3,512-3,514.
ing,” CORROSION/91, paper no. 444 (Houston, TX: NACE, 68. C. Cato, H.J. Grabke, B. Egert, G. Panzer, Corros. Sci. 24
1991). (1984): p. 591.
24. J.L. Dawson, ASTM STP 1277, “Electrochemical Noise Measure- 69. E.G. Dafft, K. Bohnenkamp, H.J. Engell, Corros. Sci. 19 (1979):
ment for Corrosion Applications,” eds. J. Kearns, J.R. Scully, p. 591.
P.R. Roberge, D.L. Reichert, J.L. Dawson (West Conshohocken, 70. R.N. Iyer, I. Takekeuchi, M. Zamanzadeh, H.W. Pickering, Cor-
PA: ASTM, 1996), p. 3-35. rosion 46 (1990): p. 460.
25. D.A. Eden, K. Hladky, D.G. John, J.L. Dawson, “Electrochemi- 71. R.S. Lillard, J.R. Scully, Corrosion 52 (1996): p. 125.
cal Noise — Simultaneous Monitoring of Potential and Current 72. K.E. Heusler, Habilitation Thesis, Univ. Of Stuttgart, 1966.
Noise Signals from Corroding Electrodes,” CORROSION/86, 73. K.F. Bonhoeffer, K.E. Heusler, Z. Phys., Chem., N.F., 8 (1956):
paper no. 274 (Houston, TX: NACE, 1986). p. 390.
26. D.A. Eden, A.N. Rothwell, “Electrochemical Noise Data: Analy- 74. J. Newman, Electrochemical Systems (Englewood Cliffs, NJ:
sis Interpretation Presentation,” CORROSION/92, paper no. Prentice Hall, Inc., 1973).
292 (Houston, TX: NACE, 1992). 75. C. Wagner, J. Electrochem. Soc. 101 (1959): p. 225.
27. A.N. Rothwell, D.A. Eden, “Electrochemical Noise Techniques 76. W.C. Ehrhardt, ASTM STP 1056, “The Measurement and Cor-
for Determining Corrosion Rates and Mechanisms,” CORRO- rection of Electrolyte Resistance in Electrochemical Cells,” eds.
SION/92, paper no. 223 (Houston, TX: NACE, 1992). Scribner, Taylor (West Conshohocken, PA: ASTM, 1990), p. 5-
28. U. Bertocci, ASTM STP 1277, “Electrochemical Noise Measure- 26.
ment for Corrosion Applications,” eds. J. Kearns, J.R. Scully, 77. K. Nisancioglu, Corrosion 43 (1987): p. 258.
P.R. Roberge, D.L. Reichert, J.L. Dawson (West Conshohocken, 78. M.A. Pech-Canul, A.A. Sagues, P. Castro, Corrosion 54, 8
PA: ASTM, 1996), p. 39-58. (1998): p. 663.
29. U. Bertocci, F. Huet, Corrosion 51 (1995): p. 131.
30. D.L. Reichert, ASTM STP 1277, “Electrochemical Noise Mea- LIST OF SYMBOLS
surement for Corrosion Applications,” eds. J. Kearns, J.R.
Scully, P.R. Roberge, D.L. Reichert, J.L. Dawson (West
Conshohocken, PA: ASTM, 1996), p. 79-89. A = electrode area (cm2)
31. P.C. Searson, J.L. Dawson, J. Electrochem. Soc. 135, 8 (1988): B = proportionality constant used in Stern-
p. 1,908.
32. F. Mansfeld, H. Xiao, ASTM STP 1277, “Electrochemical Noise Geary Equation
Measurement for Corrosion Applications,” eds. J. Kearns, J.R. C = capacitance (farad-cm–2)
Scully, P.R. Roberge, D.L. Reichert, J.L. Dawson (West Cdl = double layer capacitance (farad-cm–2)
Conshohocken, PA: ASTM, 1996), p. 59-78.
33. G.P. Bierwagen, J. Electrochem. Soc. 141 (1994): p. L155. C = adsorption pseudo-capacitance (farad-cm–2)
34. F. Mansfeld, H. Xiao, J. Electrochem. Soc. 141 (1994): p. 1,402. E = electrochemical potential (volts)
35. F. Huet, J. Electrochem. Soc. 142 (1995): p. 2,861.
Ecorr = corrosion potential (volts)
36. J.F. Chen, W.F. Bogaerts, Corros. Sci. 37 (1995): p. 1,839.
37. H. Xiao, F. Mansfeld, J. Electrochem. Soc. 141 (1994): p. 2,332 Erev = reversible electrode potential (volts)
38. F. Mansfeld, H. Xiao, J. Electrochem. Soc. 140 (1993): p. 2,205 F = Faraday’s constant (96,485 coul/equivalent)
39. F. Mansfeld, L.T. Han, C.C. Lee, J. Electrochem. Soc. 143
(1996): p. L286.
f = frequency (s–1)
40. U. Bertocci, C. Gabrielli, F. Huet, M. Keddam, J. Electrochem. fmax = maximum frequency (s–1)
Soc. 144 (1997): p. 31. iapp = applied current density (amp-cm–2)
41. U. Bertocci, C. Gabrielli, F. Huet, M. Keddam, J. Electrochem.
Soc. 144 (1997): p. 37. icorr = corrosion current density (amp-cm–2)
42. A. Bautista, F. Huet, J. Electrochem. Soc. 146 (1999): p. 1,730- Ia = anodic half-cell current (amp)
1,736. Ic = cathodic half-cell current (amp)
43. F. Mansfeld, M. Kendig, Corrosion 37, 9 (1981): p. 556.
44. R. Bandy, D.A. Jones, Corrosion 32 (1976): p. 126. IPSD = power spectral density of current noise
45. M.J. Danielson, Corrosion 36, 4 (1980): p. 174. (amp2 -Hz–1)
46. J.C. Reeve, G. Bech-Nielsen, Corros. Sci. 13 (1973): p. 351.
j = √-1
47. L.M. Callow, J.A. Richardson, J.L. Dawson, Brit. Corros. J. 11
(1976): p. 132. k = pre-exponential term defining Langmuir
48. F. Mansfeld, K.B. Oldham, Corros. Sci. 27 (1971): p. 434. coverage
49. S. Barnartt, Corros. Sci. 9 (1969): p. 148.
50. R.L. Leroy, Corrosion 29 (1973): p. 272.
q = charge (farads)
51. F. Mansfeld, Corrosion 29 (1973): p. 397. qF = charge in adsorbed film (farads)
52. D.D. MacDonald, J. Electrochem. Soc. 125 (1978): p. 1,443. qM = charge associated with one monolayer of
53. C. Gabrielli, M. Keddam, H. Takenouti, V. Kirk, F. Bourelier,
Electrochem. Acta 24 (1979): p. 61. film (farads)
54. D.W. Townley, Corrosion 47 (1991): p. 737. R = ideal gas constant (8.31441 J-mol–1-K–1)
55. D.D. Macdonald, J. Electrochem. Soc. 125 (1979): p. 1,977. Rs = solution resistance (Ω-cm2)
56. F. Mansfeld, M. Kendig, Corrosion 37 (1981): p. 545.
57. K. Videm, R. Myrdal, Corrosion 53 (1997): p. 734. Rct = charge-transfer reaction resistance (Ω-cm2)
58. D.A. Jones, N.D. Greene, Corrosion 22 (1966): p. 198. Reff = measured resistance at DC limit in case of
59. B. Syrett, D.D. MacDonald, Corrosion 35 (1979): p. 505.
nonuniform current distribution (Ω-cm2)
60. I. Epelboin, M. Keddam, J. Electrochem. Soc. 117 (1970): p.
1,052-1,056. RF = additional reaction resistance in the pres-