Marcus P. Corrosion mechanisms in theory and practice

Подождите немного. Документ загружается.

a moiré fringe technique to observe directly the metal dissolution in a crevice,

Shinohara et al. [65] confirmed the localized character of the early dissolution events

and their lateral extension and coalescence.

It is worth noting that the crevice may favor pit initiation in two ways: (a) the

pitting potential of a stainless alloys decreases with increasing chloride content

(Fig. 21a); (b) the restricted transport conditions may contribute to stabilize pit

embryos at potentials lower than on the free surfaces [66]. This last point is supported

by the experimental and modeling work of Laycock et al. [62]. Indeed, the deleterious

effect of anodic polarization on the initiation time is obviously consistent with a

pitting process.

Role of Sulfide Inclusions on Stainless Steel

Besides their possible role in pit initiation (see earlier), the dissolution of the MnS

inclusions can generate thiosulfates [47, 67] or more likely sulfide [51, 52] ions,

which are known to favor passivity breakdown. Indeed, Crolet et al. [54] observed

higher depassivation pH on stainless steels with high sulfur content. According to

Brossia and Kelly, the initiation involves a critical [Cl

–

]/[HS

–

] ratio. This effect of

MnS is unlikely in modern stainless alloys with very low sulfur contents.

Passivity Breakdown Due to High (Metal) Chloride Concentration

On pure aluminum, Hebert and Alkire [68] proposed that passivity breakdown

occurred when a critical concentration of aluminum chloride is attained in the

crevice environment. Indeed, they showed convincingly (Fig. 25) that the presence of

aluminum cations in the solution caused passivity breakdown for a pH and chloride

concentration at which solutions containing only sodium cations did not.

Based on an experimental work on alloy 7475 crevices and propagating cracks,

Holroyd et al. [30] agree with this assumption. The critical aluminum concentration

required decreases as chloride content increases [68] but low pH and high chloride

content are not a prerequisite for passivity breakdown.

The possibility of the existence of a critical concentration of chloride rather than

a critical pH has been proposed for stainless steels, for example, by Zakipour and

Leygraf [69] in relation to a critical composition of the passive film (chromium

content of about 48%).

Crevice Corrosion of Metallic Materials 371

Figure 24 Crevice initiation by micropitting observed by Oldfield and Sutton [46] on

316 SS in 1 M NaCl, pH 6 solution.

On stainless steels, several authors [70–73] showed that a minimum

concentration of metallic chloride is required for stable propagation to occur. As

mentioned previously, the external layer of the passive films on stainless steels

contains hydroxyl groups and chloride. Thus, it may be possible that the presence in

the solution of metal complexes modifies the external layers of the passive films and

contributes to lowering their stability.

The Role of IR Drop in the Initiation of the Crevice Corrosion

The role of IR drop in crevice initiation is not clear. Different authors [58,60,61]

observed crevice initiation on stainless steels at a very low IR drop level. It is clear

that initiation processes can be separated into two classes: (a) those that operate at

relatively high potentials (pitting in the crevice gap), which cannot be enhanced by

large ohmic drops, and (b) those that occur at low potential (general passivity

breakdown), which are favored by large IR drops. However, on stainless steels and

nickel-base alloys, there is at present no direct evidence to support the last type of

process, mainly because high free surface potential always enhances crevice

initiation of passive alloys.

On titanium alloys, however, the metal still has an active-passive transition in

the crevice environment and the surface potential must be low enough for the active

corrosion to occur. Thus, IR drop is required to stabilize the active dissolution [9,74],

particularly when large cathodic currents are available.

Propagation of Crevice Corrosion

Propagation of crevice corrosion occurs by active dissolution in an occluded cell

and it is generally considered analogous to the propagation stage of pitting corrosion.

372 Combrade

Figure 25 Effect of dissolved Al on the activation of pure Al in pH 3.6 solution [68].

Test under galvanostatic conditions.

At least in a first stage, corrosion may not affect the whole crevice gap: it has

been shown that on austenitic stainless steels, dissolution is mainly located near the

mouth of the crevice [3,19] (Fig. 26), whereas on ferritic stainless steels the

maximum of dissolution rate is located deeper in the crevice [3].

As a consequence of rapid corrosion inside the crevice, the rate of evolution of

the environment toward more acidic and more concentrated solutions increases, and

indeed several experiments confirmed this point. However, the dissolution rate and

the environment evolution may be limited by the following factors:

Immediately after passivity breakdown, the dramatic increase of the dissolution

current has two consequences that contribute to limit the dissolution rate by

decreasing the corrosion potential of the active surfaces in the crevice:

1. The cathodic reaction kinetics increase on the free surfaces to balance the

increased anodic dissolution and this induces a drop of their corrosion

potential. Potential drops of several hundreds of mV can be observed on the

free surfaces due to the initiation of crevice corrosion (see Fig. 5) Large

cathodic areas, high solution conductivity, and cathodic reaction depolarizers

(such as carbon deposits) can increase significantly the amount of available

cathodic current and thus the crevice propagation rate.

2. The migration current from the crevice gap increases dramatically and this also

increases the ohmic drop: a role of the ohmic drop in the control of the

propagation rate is consistent with a maximum of corrosion rate located near

the crevice mouth as observed by several author [3,19]. One must notice that

ohmic drop is not limited to the solution inside the crevice. Significant ohmic

drops may also occur in the bulk solution near the crevice mouth, particularly

in dilute environments.

Later, several other events may also occur to limit the environment evolution

and the dissolution rate:

The solution inside the crevice reaches saturation and a salt film is formed on the

active surface. The corrosion rate is then limited by mass transport trough the

salt layer. Recent results [34,35,73,76–78] confirm this possibility, which was

not as clearly established as it is for pit propagation.

Crevice Corrosion of Metallic Materials 373

Figure 26 Preferential propagation near the crevice mouth [75].

The pH and the potential in the crevice may become low enough for local water

reduction to be possible. This stifles further pH decrease as the local cathodic

reaction is able to balance the production of cations by anodic dissolution.

Balance between production and outward transport of cations and dilution due to

geometric changes of the crevice [79] are also invoked as possible limiting

processes. However, on the simple basis of the difference of density between

the dissolved metal and the saturated crevice solution, Hakkarainen [80]

pointed out that at least 97% of the dissolved cations must be evacuated out

of the active crevice. This weakens the preceding point. But, on the contrary,

the possible limitation by the diffusion of water through a saturated salt layer

has not been considered.

Thus, even though the corrosion occurs by active dissolution, its rate is not

controlled by an activation process and, generally, it remains limited to current

densities of the order of tens of mA/cm

in the saturated solutions, which prevail

on well-developed crevices [19,33,57,70,71]. Figure 27 shows how the dissolution

rate decreases when saturation of the solution is approached.

When discussing the propagation of crevice corrosion, it is worth noting that the

analogy with the propagation of pits may have limitations. Most of the experimental

work and many simplified calculations refer to a geometry of the corroding crevice

with an aspect ratio L/h close to 1. This may have consequences for conclusions

regarding the distribution of the dissolution current inside the crevice gap and the

rate-controlling processes because an almost uniform corrosion potential on the

corroding surface and an almost uniform transport path for the diffusion and

migration processes can be assumed in pits. In very deep crevices, the potential and

the transport path vary along the crevice profile; thus, the evolution of the local

environment may vary. Consequently, the saturation and the formation of salt films

leading to a dissolution process controlled by diffusion may not occur simultaneously

on the whole crevice surfaces. In addition, the increasing IR drop near the tip of the

crevice gap may allow the reduction of water to occur, changing the local pH

evolution before any saturation has been reached.

374 Combrade

Figure 27 Anodic dissolution of 304 SS in simulated crevice solutions [71]. The numbers

indicate the degree of saturation of the solution.

Crevice Repassivation

Spontaneous Crevice Arrest

Observations (see Propagation of Crevice Corrosion earlier) suggest that, under

constant bulk environment conditions, crevice corrosion may spontaneously cease.

This has been taken into account in a model developed by Gartland [19]. Gartland

attributed this behavior to an environment dilution caused by the geometric change

of the crevice, but we have seen that this argument is not always likely. It also seems

appropriate to invoke salt precipitation and/or local dryout due to the slow diffusion

of water through a salt film of increasing thickness. This suggests that mass transport

could not be efficient enough to eliminate such a large amount of corrosion products

and this could lead to the precipitation of solid products that would fill the crevice

gap. Another cause of self-arrest, or at least of kinetic limitation, could be lack of

sufficient water in the crevice gap. Further studies are required to establish the

conditions and the mechanisms involved in this spontaneous arrest, in order to take

advantage of it in practical situations.

Repassivation Potential

The propagation of crevice corrosion can also be arrested by decreasing the potential

of the outside surfaces below a critical value (see earlier). The existence of a

“repassivation” or “protection” potential was recognized very early, in particular by

Pourbaix et al. [81] for pitting corrosion. From a practical point of view, the existence

of a protection potential below which no crevice corrosion is possible is of major

importance because it guarantees the immunity of passivated alloys in near-neutral

chloride solutions in the absence of oxidizing species and because it makes possible

the cathodic protection of structures.

However, the significance and the intrinsic nature of this potential are still under

discussion. There are at least two possible causes for active corrosion inside crevices

to stop below a critical potential:

First, the critical potential may be a “deactivation potential”, i.e., a potential that

corresponds to a cancellation of the overpotential required for dissolution in

the crevice. According to Starr et al. [82], if active corrosion stops by

deactivation with decreasing potential, a further increase of the corrosion

potential would cause an immediate reactivation of the crevice. Indeed, Starr

et al. [82] and Dunn and Sridhar [83] observed such a reactivation on

low-grade stainless steels in acidic environments.

Second, the critical potential may be a repassivation potential. It has been shown

in artificial active crevices that lowering the potential of the free surfaces

causes the local environment to become less aggressive (see, for example,

Pourbaix [36]). Thus, at some point, the environment becomes not aggressive

enough for active dissolution to be sustained and the metal surface in

the crevice becomes passive. In this case, a subsequent increase of the

corrosion potential does not produce immediate reactivation inside the

crevice. Starr et al. [82] observed this situation on 12% Cr stainless

steels in near-neutral environments; Dunn and Sridhar [83] observed the

same behavior on alloy 825. However, the repassivation may be

attributed to different environment changes: an increase of local pH in

the crevice [82,84], a destabilization of the salt film that controls the

Crevice Corrosion of Metallic Materials 375

dissolution kinetic in the crevice [77,85,86], or a decrease of the chloride

concentration below a critical value [34,35,71–73,77,78,83]. Recent experi-

mental observations [34,35,78] show that corrosion may be stable even

after dissolution of the salt film.

In fact, studies of Cragnolino et al. [8,11] suggest that, depending on the

conditions of the experimental determination, the “arrest” potential measured by

backscan polarization may be close to either potential: using a relatively high scan

rate, which does not allow time for the local environment to be significantly

modified, the measured arrest potentials are close to the deactivation potentials and

they are significantly lower than those measured under a low potential scan rate. The

latter are more likely to be due to a modification of the local environment toward less

aggressive conditions. We have already seen that the maximum repassivation

potential seems to be almost identical to the minimum initiation potential (Fig. 8).

This observation has been reported for a relatively resistant stainless alloy (alloy 825)

and it would support the idea of Gartland [19,57] (see earlier) that there is a critical

environment for passivity breakdown and that the local environment required for

repassivation is not very different. This could be an argument against the mechanism

of initiation by pitting in the crevice and in favor of a mechanism involving a critical

environment.

A pending question is still to know whether this repassivation potential has

a unique value or depends on corrosion damages and crevice geometry. The following

arguments are in favor of such a unique value:

1. For a given bulk environment and a given crevice configuration, the

repassivation potential becomes independent of the corrosion extent as soon

as some critical damage size is exceeded [8] (Fig. 28).

2. At least for high chloride concentrations, the repassivation potential exhibits

low values which are poorly dependent on the chloride content of the

environment [87] (Fig. 29). Dunn et al. expressed the average values of the

repassivation potential as follows:

376 Combrade

Figure 28 Influence of the corrosion damage on the repassivation potential of an alloy

825 in a 1000 ppm Cl

–

solution at 95°C [8].

However, the Fig. 29 shows that the effect of chloride content is low on the

repassivation potential, particularly when looking at the lower bound of the scatter

b and of the measured value which is almost constant for chloride contents in excess

of 10

–1

M. This is consistent with the presence of a nearly saturated solution and/or

precipitated salt film in well developed crevice corrosion, and indeed such

environments should be almost independent on the bulk solution chemistry. Thus, a

repassivation potential which, according Pourbaix [36], is close to the local potential

of an actively corroding crevice should be poorly or no dependent on the bulk

environment and crevice geometry.

Nevertheless, this point still requires more work, particularly regarding the

exact meaning of the high values of the repassivation potentials measured in low

chloride and/or low temperature environments (left side of the Fig. 10, 11 and 29). In

this case, the local environment inside the crevice is likely to be controlled by mass

transport and thus the repassivation potential could be dependent on the crevice

geometry.

MODELING CREVICE CORROSION

In order to try to predict the behavior of passive materials in chloride environments

and particularly stainless alloys in seawater, many attempts have been made to

model the environment in a crevice, mostly for stainless Fe-Ni-Cr-Mo alloys. Two

kinds of model have been developed:

Steady state model [2,89–92], which try to calculate the environment in a

propagating crevice or, more frequently, a pit, and

Transient models [2,4,19,57,79,93–99], which intend to describe the environment

evolution in a crevice and, in most instances, to predict the passivity

breakdown by using a criterion usually based on a critical environment

resulting from experimental data.

Crevice Corrosion of Metallic Materials 377

Figure 29 Influence of the chloride content on the repassivation potential of an alloy

625 at 95°C [87].

with

Modeling the environment or the environment in a crevice requires solving

a set of equations including:

The kinetics of surface reactions in the crevice, mainly the metal dissolution

because the cathodic reaction is generally assumed to occur only on the

outside surfaces

The kinetics of the reactions within the crevice solution, mainly the formation of

metal chlorides and the hydrolysis of cations, which is the fundamental

cause of pH drop within the crevice

The transport processes including electrolytic migration and chemical diffusion, as

most of the models assume no convection inside the crevice gap

The mass and electrical charges balance in the crevice solution, across the

metal-solution interfaces, and at the mouth of the crevice

Surface Reactions

With few exceptions [19], the passive anodic current of metal dissolution is usually

taken as a constant and potential independent during the initiation stage and/or it is a

parameter of the calculations.

During the propagation stage, different models are used, from the Tafel law

[100] to experimentally fitted laws [19]. To take into account the strong current

limitation due to the formation of a saturated salt layer on the dissolving surface,

Gartland [19] introduced a damping coefficient that increases exponentially as the

local concentration approaches saturation.

The possibility of water reduction inside the crevice gap is taken into account

by Gartland [19] as a possible situation when the local pH and potential become low

enough. Only one model [99] includes the cathodic reaction outside the crevice gap.

It is more or less implicitly assumed in most models that this reaction is not a

controlling process for the environment evolution inside the gap.

Solution Chemistry

The description of the crevice solution chemistry is probably the most difficult

step (at least from the corrosion or chemistry standpoint if not the computer

programming) as a result of the lack of reliable data. The main problems are:

The thermodynamic description of the solution inside the crevice, which cannot be

done by the dilute solution models, at least during the propagation stage

The nature of the complex ions formed in the crevice and their thermodynamic

properties, which are required to calculate the hydrolysis reactions governing

the environment evolution

Solution Models—Activity Coefficients

Most models use dilute solution assumptions and former models [93] assumed unit

values [2,19,93,94] of all activity coefficients (except in some cases for H

ions).

Bernhardsson et al. [4] reached a compromise by using two sets of equilibrium

constants, one for dilute solutions and the other for concentrated solutions.

378 Combrade

Other authors used Debye-Hückel, the truncated Davies model [96,98], or the B-dot

Debye-Hückel model [98] to derive the activity coefficients in concentrated

solutions.

A particular case is the activity coefficient of H

ions. It has been shown

experimentally (Fig. 30) that the presence of metal ions strongly decreases the pH of

a chloride solution. Many models [2,19,94,97] include a pH correction derived from

the results of Mankowski and Smialowska [33].

More recently, Brossia et al. [35] introduced an approach based on new

correlations for activity coefficients in concentrated solutions. This approach [101] is

based on the Helgeson-Kirkham-Flowers equation for standard-state properties with

a nonideal solution model based on the activity coefficient expression developed by

Bromley and Pitzer. Using specific software, Brossia et al. were able to predict the

dominant ionic species and the salt precipitation in Ni, Fe, Cr, and 308 stainless steels

crevice solutions. Fair agreement was observed with the in situ analyses of these

crevice solutions by Raman spectroscopy.

Hydrolysis and Complexation Products and Reactions

These species and reactions include the products of the metal cation hydrolysis

and the formation of metallic chlorides.

In one of the first models [93], the formation of metal chlorides was not

taken into account and the cation hydrolysis was assumed to produce solid

hydroxides. In the case of stainless steels, the pH drop was assumed to be driven

by the hydrolysis of chromium cations according to the following reaction:

Crevice Corrosion of Metallic Materials 379

Figure 30 Influence of the ionic strength on the H

activity in chloride solutions [19].

In the case of aluminum alloys, the hydrolysis reaction was assumed to be

380 Combrade

Table 3 Solubility and Hydrolysis equilibria in the Model of Gartland [19]

Actually, the drop of pH is related to more complex reactions and species. Thus, in

more sophisticated models, several hydrolysis reactions and metal chloride formation

are taken into account but the selection of species and reactions is somewhat

different from model to model. Oldfield and Sutton [94] and Watson and

Postlethwaite [2] considered only hydroxides as the product of cation hydrolysis.

Sharland [96] introduced simple metallic chlorides. The most complete set of species

and reactions has been used by Bernhardsson et al. [4], which made available the

thermodynamic data of a large number of species, including several iron, nickel,

chromium, and molybdenum polycations as well as metal chlorides and

hydroxychlorides. Gartland [19] used a more limited set of species (Table 3) selected

among the Bernhardsson data. According to their experimental results, Hebert and

Alkire [95] included Al(OH)

3–n

as the hydrolysis product in their model of the crevice

corrosion of aluminum alloys.

A common feature of all the models is that all the chemical reactions of

complex formation and cation hydrolysis are fast enough for thermodynamic

equilibrium to be assumed. This may also be an incorrect assumption because we

have already mentioned that the pH of synthetic crevice solutions can evolve for

significant periods of time (up to several weeks) before reaching a stable value

[33,40].

Transport Processes

Most models use transport equations for dilute solutions including diffusion,

electrolytic migration, and convection such as:

with