Marcus P. Corrosion mechanisms in theory and practice

Подождите немного. Документ загружается.

However, different attempts have been made to take into account the deviation

from ideality of the crevice solutions:

Walton et al. [98] wrote the transport equation (11) in a form using the activity

instead of the concentrations.

Crevice Corrosion of Metallic Materials 381

However, this formulation assumes the validity of the Nernst-Einstein equation for

any ionic strength.

Gartland [19] made an attempt to use corrected ionic mobility coefficients: he used

the diffusion coefficients collected by Bernhardsson [4] regardless of the

solution concentrations but, to calculate the ion mobility coefficient, he

introduced in the Nernst-Einstein equation an ionic strength–dependent

correction factor based on a limited set of experimental data.

Walton [102] also included a porosity tortuosity factor in the diffusion coefficient

to take into account the presence of porous solids and/or gas bubbles in the

crevice gap.

The ionic fluxes are generally used to derive the current flowing in the crevice:

One common assumption of most, if not all, the available models is that the

transport is purely unidimensional; i.e., the crevice is assumed to be narrow

enough for the transverse concentration and potential gradients to be negligible. In

fact, this assumption is required in order to simplify the numerical solutions.

Mass Balance

The mass balance equation:

includes the production of species by chemical reactions in the liquid phase R

. These

reactions are the complexation and hydrolysis reactions as well as dissolution

reactions on the crevice walls because models assume unidimensional transport fluxes.

Electrical Charge Balance

The electrical charge balance inside (and in one instance outside) the crevice is

modeled in two ways:

Many models assume the electrical neutrality in any point [4,93,95,96]:

The potential distribution is obtained by using the Ohm law applied to the

currents calculated in Eq. (14) and solution resistivity calculated from mobility

coefficients.

Others [2,79,97,99] solve the Poisson equation, which also gives the potential

distribution along the crevice gap:

382 Combrade

The solution of the Poisson equation dramatically increases the complexity and

duration of the numerical simulations and requires numerical shortcuts. In addi-

tion, Watson and Postlethwaite [2] mentioned that the time required for electrical

neutrality to be reached appears to be less than 10

–10

s. This is very short com-

pared with the time required for chemical equilibrium, and it is not clear whether

the simple assumption of electrical neutrality could significantly change the

results of such models.

Crevice Initiation Criteria

All criteria rely on a critical environment for passivity breakdown:

Critical pH [93]

Critical crevice solution [2,19,46,94] taking into account the pH and the chloride

content but not the effect of metal cations

Critical dissolved Al content [95]

Despite the experimental observations, no attempt has been made to model crevice

initiation by local pitting.

Results of the Models

As expected, the models predict the drop of local pH and the increase of chloride and

metal ion content in the crevice gap. The predictions of several models were

compared, with variable success, with the environment modifications observed in

artificial crevices. Figure 31 shows a typical example of results indicating that the

transport process is probably well enough modeled as indicated by the fair prediction

of the chloride content but that the pH calculations are far less accurate. In Figs. 32

and 33, the set of data of Alavi and Cottis [26] has been compared with several

models: the pH drops are generally underestimated (Fig. 32a–e), the potential

gradient inside the crevice is not well described (Fig. 32f), and no model is able to

predict the pH increase observed near the crevice tip (Fig. 3a–c).

Gartland [19] showed (Fig. 33) that the use of a correction factor to evaluate the

activity is necessary to approach the actual pH, regardless of the fact that the

equilibria involving solubility and hydrolysis of metal cations are taken into account.

This may be one of the major causes of the underestimation of pH drop observed in

the preceeding models.

Beside the quantitative results, which are impaired by the lack of basic

knowledge of the crevice chemistry, the models provide interesting parametric results:

All models confirm that the crevice geometry is of major importance. As

previously mentioned, Bernhardsson [4] introduced an L

/h geometric

factor (in fact, a severity factor S = i

/h, Fig. 34) that appears to control the

environment evolution. The trends shown in Figure 34 are one of the main

reasons to study in more detail the effect of crevice depth/width ratio on the

crevice repassivation (see Crevice Repassivation earlier).

More precisely, the results of the Watson model clearly indicate (Fig. 35) that

corrosion can start only if the crevice is tight and deep enough and support

the fact that a critical crevice size may exist. This is consistent with one

conclusion of White et al. [99] which indicates that their model is not able to

predict pH low enough for crevice initiation in gaps of the order of several

tens of micrometers. These results, as well as results of Gartland [19], also

show that efficient crevice gaps are of the order of micrometers or less, in

accordance with experimental results of Oldfield [3]. This supports the

assumption that microcrevices due to asperity may play a major role in

crevice initiation, as suggested by Sridhar and Dunn [60].

Gartland [19] also predicted, in agreement with observation (Fig. 26), that initial

crevice damages are mainly located near the crevice mouth (Fig. 36), where

the IR drop is low, which makes possible higher dissolution rates.

Finally, these models also show that the initial pH drop may be fast but that,

rapidly, the rate of evolution becomes very slow (Figs. 34 and 37).

In summary, these models supply interesting information but still rely on

experimental fitting to predict the initiation times correctly. Among their

weaknesses, there is lack of precise data on the local chemistry of concentrated

solutions and lack of prediction of the effect of the potential of the free surfaces. The

use of a unidimensional transport equation and the assumption of instantaneous

equilibrium of the hydrolysis and solubility reactions are also questionable.

Crevice Corrosion of Metallic Materials 383

Figure 31 Comparison of the results of Zuo et al. [37] with the predictions of the model

of Sharland [96]: note that the chloride content is fairly well predicted, whereas the pH

drop is underestimated.

Finally, whatever their accuracy, the use of crevice initiation models will always

meet with some limitation because of the uncertainty in the actual crevice geometry.

However, a very useful model to be developed would be a propagation model,

particularly in an attempt to obtain theoretical support to describe the repassivation

potentials and, if possible, to predict self-arrest of crevice corrosion.

384 Combrade

Figure 32 Comparison of the experimental results of Alavi and Cottis [26] with the

predictions of the models of White et al. [99], Sharland [96], and Walton et al. [98] assuming

various current densities.

EXPERIMENTAL CHARACTERIZATION OF THE ALLOYS VERSUS

CREVICE CORROSION

In order to select alloys for safe use in aerated chloride environments, a series of

tests and criteria has been designed. They can be divided in several classes:

Exposure tests in environments representative of the service environment

Exposure tests in conventional environments

Electrochemical tests, which can be divided into two classes: (a) those intended to

measure critical potentials for crevice initiation or protection and (b) those

used to derive initiation criteria from the behavior of materials in

conventional solutions supposed to simulate the crevice environment

For a more detailed description of crevice testing, the reader may consult the

papers of Oldfield [103] and Ijsseling [104]

Crevice Former Devices

All the exposure and electrochemical tests that involve the use of crevice former

devices encounter the problem of the geometry of the crevice. Many different designs

of crevice former have been used, some examples of which are given in Figure 38a.

The ASTM standard G78 defines a crevice former (Fig. 38b) that involves 20 crevice

zones and gives a series of recommendations, for example, for the setup of the

specimens in the test vessels. This device allows more reproducible results because

of the statistical effect of the large number of crevice zones. However, the standard

does not specify important points such as the condition of the contact surface of the

crevice former.

Crevice Corrosion of Metallic Materials 385

Figure 32

386 Combrade

Figure 33 Effect of H

activity correction on the pH calculated by the Gartland model.

Figure 34 Effect of severity factor on pH drop calculated by the Bernhardsson model.

Figure 35 Effect of crevice geometry on crevice initiation time calculated by the Watson

model.

Exposure Tests in Representative Environments

These tests are designed to rank the behavior of different alloys, to get orders of

magnitude of initiation times in service, and/or to select alloys that should guarantee

long-term resistance to crevice initiation. Thus, they can be very long (months or

even years) with carefully selected crevice former devices, materials, and

surface conditions. A problem is the test monitoring to detect crevice initiation

without any perturbation of the local processes.

Apart from the selection of the crevice former device, it is important to have in

mind that the severity of the test may depend on the availability of cathodic current.

Crevice Corrosion of Metallic Materials 387

Figure 36 Prediction of the initial corrosion damage near the crevice mouth by the

Gartland model (see Fig. 26 to compare with experimental data).

Figure 37 Drop of pH versus time as predicted by the Oldfield model.

Thus, it is important to use specimens with large free surface area, particularly if

the environment has high electric conductivity that allows long-distance galvanic

coupling (seawater, for example).

There are different monitoring possibilities for exposure tests:

Periodic examination of the specimens is the simplest technique but it may

dramatically modify the incubation period if the crevice zone becomes more

or less dried during the examination. In particular, opening the crevice

former to observe the inner surfaces may modify completely the local

environment and cause the effective test time to go back to almost zero.

Monitoring of the corrosion potential allows one to detect any increase of the

corrosion rate due to the polarization of the cathodic reaction on the free

surfaces. Thus, the initiation of active corrosion in the crevice may be

detected. In addition, electrochemical noise (i.e., potential fluctuations)

indicating some dissolution transient may be interpreted as precursor events

of passivity breakdown or at least of a low degree of passivity. The analysis

of potential noise in terms of stochastic versus chaotic features has been

shown [106] to allow early detection of pit initiation. This type of analysis

should be checked for crevice corrosion.

Periodic measurements of polarization resistance may also be used to detect the

onset of active corrosion.

Exposure Tests in Conventional Environments:

The Ferric Chloride Test

This test is defined by the ASTM standard G48 and it is used for stainless steels

and nickel-base alloys. Coupons with a crevice former device are exposed to a 6%

FeCl

solution. Two criteria are used to characterize the tested materials:

The weight loss after a given time of exposure at a given temperature.

388 Combrade

Figure 38 Crevice former devices for crevice corrosion testing. (a) Different types of

crevice former (From Ref. 105). (b) ASTM G 78 crevice former device.

A critical crevice temperature obtained by periodically increasing the temperature

by steps of 2.5 or 5°C until passivity breakdown occurs in the crevice.

However, because the passive film becomes more stable with increasing

exposure time, the critical temperature may depend to some extent on the

temperature selected to start the test.

An experimental correlation shows that this test can be used to rank the behavior of

stainless alloys in seawater. In practice, it is also used as a control test to guarantee

the constant quality of a product or to check the effect of fabrication parameters such

as thermal treatment, welding, and surface condition.

Electrochemical Tests

Determination of Critical Potentials for Initiation and Protection

These tests are the same as those used to determine the pitting resistance except that

a crevice former device is present on the test coupons. The measured parameters are:

The crevice potential and the repassivation potential obtained on polarization

curves (potentiokinetic technique) or, in some instances, by using potential

steps or potentiostatic tests

The critical temperature for the crevice at a given potential

Potentiokinetic Techniques Figure 7 of this chapter represents typical

polarization curves of a passivated alloy in a near-neutral chloride solution with and

without a crevice former device. The shapes of the two curves are identical except

that the crevice potential is definitely lower than the pitting potential. In addition, it

may depend on the crevice geometry if not properly measured.

As already discussed, the main difficulty of this technique, beside the problem

of the crevice former geometry and reproducibility, is that the results are strongly

dependent on the potential scan rate both because of the time-dependent stability of

the passive films and because of the time-dependent evolution of the environment

inside a crevice. In particular, the repassivation potential may be overestimated if

corrosion is not well developed in the crevice and it can be underestimated if the

potential backscan is too fast to allow the evolution of the local environment to be in

“quasi-steady” conditions. It is generally admitted that the scan rate has to be very

low, which causes the two critical potentials to become closer. But the appropriate

scan rate must be determined on each system because it may depend on the alloy and

on the environment.

Potentiostatic Techniques To determine the critical potential of crevice

initiation, coupons in a crevice former device are exposed for a fixed period of time

under potentiostatic control and monitoring of the anodic current is used to detect the

onset of active corrosion. Several experiments are performed at different potentials

and the crevice potential is the threshold potential that corresponds to an infinite

initiation time (see Fig. 8 and 9 at the beginning of this chapter).

To measure the repassivation potential, severe crevice corrosion must be

initiated on a “creviced” specimen and allowed to propagate for a fixed period of

time or, more usually, to a definite amount of anodic charge. Then the potential is

Crevice Corrosion of Metallic Materials 389

stepped to a lower value and current monitoring is used to check the propagation or

repassivation. As shown in Figure 8, the repassivation time increases with increasing

applied potential and the repassivation potential is the potential corresponding to an

infinite repassivation time. Apparently, the repassivation potential may depend on the

development of corrosion before the potential is stepped down only if the amount of

accumulated charge is lower than a critical value.

The potentiostatic techniques should be preferred to the potentiokinetic ones,

but they require more specimens and longer experiments.

A more sophisticated repassivation test has been designed [7,12] to minimize

the number of specimens by using a stepwise backscan potential (Fig. 39). During

each step, the evolution of the current is analyzed and time is allowed to determine

the trend. If, after decreasing, the anodic current increases again, indicating that

propagation is not inhibited, the potential is stepped down again until repassivation

occurs.

Determination of the Material Behavior in Simulated Crevice

Environments

The tests that have been widely used in the past years are intended to determine the

conditions of passivity breakdown in the crevice. Alloy coupons with no crevice area

are tested in solutions that are supposed to simulate the local conditions in crevices.

This is usually done by using acidic solutions with an increased chloride content

compared with the bulk environment. For example, to perform tests to characterize

the behavior in seawater, solutions containing 30 to 150 g/L of NaCl were used, the

pH being lowered by HCl additions. Different measurements can be performed.

390 Combrade

Figure 39 Stepwise potential backscan and current response to determine the repassivation

potential [12].