Marcus P. Corrosion mechanisms in theory and practice

Подождите немного. Документ загружается.

On the basis of these ideas, Macdonald and co-workers [13,14] developed

their model of passivity and its breakdown involving the action of vacancies within

the passive layer. It is assumed that cation vacancies migrate from the oxide-

electrolyte to the metal-oxide interface, which is equivalent to the transport of

cations in the opposite direction. If these vacancies penetrate into the metal phase

at a slower rate than their transport through the oxide, they accumulate at the

metal-oxide interface and finally lead to a local concentration. The related voids

lead to stresses within the passive film and its final breakdown. The inward diffusion

or migration of cation vacancies is affected by the incorporation of Cl ions at the

oxide-electrolyte interface according to the following mechanism: The concentration

c of metal ion V

M+

, and O

2–

vacancies V

2–

are determined by the equilibrium of

the Schottky pair formation at the oxide-electrolyte interface [Eq. (3)], which

causes an inverse dependence of their concentrations [Eq. (4)].

Mechanisms of Pitting Corrosion 249

In the presence of Cl

–

, its incorporation in O

2–

vacancies occurs according to the

equilibrium of Eq. (5), which is affected by the potential drop E

2,3

at the interface

according to Eq. (6) with the concentration of oxygen vacancies c

v,O

2–

at the surface,

the activity of chloride a

–

within the solution, and the potential-independent part

of the equilibrium constant K.

According to Eq. (6), c

2–

depends on a

–

and the potential drop E

2,3

; c

2–

in turn

affects c

v,M

, which is the driving force of the diffusion of the cation vacancies to

the metal-oxide interface. Thus the interdependence of the concentrations of

cation and anion vacancies within the oxide and the incorporation of Cl

–

determine

the concentration gradient of cation vacancies and their transport through the

oxide layer that will cause a critical concentration for breakdown at the pitting

potential. According to this outline, the further discussion yields a semilogarithmic

dependence of the pitting potential E

and the chloride activity a

–

within the

electrolyte similar to Eq. (1) [Eq. (7)]. The constant B contains the diffusivity

constant of the cation vacancies within the oxide layer.

Objections to the point defect model are that in its original form it assumes linear

transport equations with diffusion of the different species within the passive layer,

whereas migration with an exponential dependence on the high electrical field

strength of some 10

V/cm should be dominating as a driving force. It also

concentrates the changes in electrode potential ΔE to the potential drop E

2,3

at the

oxide-electrolyte interface, so that it fully enters the equilibrium of Cl

–

incorporation

of Eqs. (5) and (6). However, large parts of ΔE contribute to Δφ, i.e., are located within

the barrier part of the passive layer (Fig. 2), and E

2,3

changes only with the

composition of the oxide surface, if at all, and the pH of the solution but not with

the applied electrode potential, at least for stationary conditions. An overpotential η

2,3

for the potential drop E

2,3

is mainly expected for nonstationary conditions of the passive

layer that will directly influence the adsorption equilibrium of Cl

–

. The specific role of

chloride and other halides in breakdown of passivity is still not sufficiently understood

in light of this theory. Further refinements might improve this interesting view.

Surface analytical methods unfortunately do not always give a clear answer

about the penetration of aggressive anions. Some authors found chloride within the

film with XPS, Auger electron spectroscopy (AES) [15,16], and secondary ion mass

spectroscopy (SIMS) [17]; others could not find it within the film [18–22]. The

contradictory results may be explained in terms of sample preparation and the

sensitivity of the methods. Very careful XPS studies with Fe-Cr alloys show

incorporation within the outer hydroxide part of the duplex passivating film [23,24].

The inner oxide layer remains free of Cl

–

if prepared within chloride-free electrolytes

before exposure to the aggressive anions. Similarly, incorporation of Cl

–

within the

outer hydroxide layer was found for pure Ni [25]. It was found in the inner oxide part

only when the passive layer was formed in solutions already containing Cl

–

. If the

electrode potential was above the critical value for breakdown, Cl

–

penetrated

into the preexisting oxide with possible lateral fluctuations of its concentration,

leading finally to the formation of pits. According to these studies, the accumulation

within the hydroxide overlayer serves as an accumulation of a sufficiently large

amount to cause breakdown in the following step. Further discussions of the

surface analytical investigations of passive layers that have been exposed to

chloride-containing solutions may be found in the chapter on passivity in this book.

A detailed bilayer or even multilayer structure is observed for passive films on

many metals and alloys [26,27]. The outer part is usually a hydroxide, whereas the

main inner part is an oxide [23,25–27]. The hydroxide structure may well act as an

ion exchanger or at least absorb anions, as has been proved for some systems.

Although the access of aggressive anions leads to changes of the passive layer

detected by ellipsometry [28] and reflection spectroscopy [29], it is still unclear what

conclusion may be drawn from these observations. If the penetration of aggressive

anions leads to weak channels where intense dissolution may start, it is unclear why

the film does not re-form specifically at this site and why these defects do not

repassivate. The self-healing mechanism of the passive layer is essential for its

excellent protecting property. The specific role of the aggressive anions is missing

in this mechanism of breakdown. Any explanation should involve the characteristic

chemical properties of anions such as the halides.

Experiments with well-prepassivated specimens show that the formation of a

corrosion pit may be an extremely fast process, in the time range of <1 s or even

<1 ms [4,30]. Figure 3 depicts as an example the increase of the current density within

less than 1 ms as a consequence of growing pits for an Ni specimen prepassivated

for 1 s before a potential shift above the critical pitting potential. A simple

comparison of the small stationary passive current densities in the range of μA/cm

with these short times leads to contradictions with the penetration mechanism

[30,31]. If anions migrate inward as cations migrate outward during stationary

dissolution, these fast nucleation times cannot be understood. Furthermore, it seems

questionable that large anions such as SO

2–

and ClO

–

migrate sufficiently fast in the

250 Strehblow

electrical field through the passive layer to cause nucleation in times t < 1 s

[30,31]. These anions may also cause pitting of iron at the lower potential part of

the passivity range and ClO

also at potentials close to the transpassive range

[2,3]. However, the inward migration of aggressive anions and the outward

migration of cations may be facilitated at local defects within the passive layer.

Even for those situations, one needs the special chemical properties of the

anions to understand why the defects do not repassivate but develop to a corro-

sion pit. Any mechanism of nucleation and growth of corrosion pits has to

include the specific role of the aggressive anions that causes the formation of a

pit instead of repassivating the defect site. These details will be discussed in one

of the following sections.

In this regard, one experiment with well-passivated Fe electrodes should be

mentioned [6]. If the penetration of aggressive anions through the passive layer was

the rate-determining step, a high electrical field strength should favor their migration

and accelerate pit nucleation. For nonstationary conditions, for passive Fe, exactly

the contrary has been observed (Fig. 4). After passivation for 1 h at 1.18 V (SHE) in

phthalate buffer pH 5.0 plus 0.1 M K

, the nucleation rate was severely increased

when the potential was stepped to 0.78 V immediately before Cl

–

addition to 0.01 M.

This pretreatment resulted in a much faster increase of the current density with time

and a slope of 3 for a double logarithmic current-time plot was found, which is an

indication of an increased nucleation rate that is constant with time. These

Mechanisms of Pitting Corrosion 251

Figure 3 Pit nucleation of Ni (A = 0.02cm

) at E = 1.30 V (SHE) in phthalate buffer, pH

5.0 + 0.1 M KCl after 1 s prepassivation at E = 0.5 V. (From Ref. 30.)

observations were confirmed by direct microscopic examination of the specimen’s

surfaces. If one waits at the lower potential before chloride addition, the nucleation

is reduced again with a related change of the current increase from a slope 3 to 2

for the double logarithmic plot, which indicates a variation from a constant

nucleation rate to a constant number of pits, i.e., no further nucleation. Apparently

the nucleation is severely increased for nonstationary conditions and is reduced

again when approaching a new stationary state. An explanation of this effect with

the penetration mechanism leads to contradictions. When the potential is stepped

to lower values the electrical field strength is decreased and one expects reduced

rather than increased migration of the aggressive anions through the passive layer

(Fig. 2). Remaining at the lower potential before Cl

–

addition should decrease the

thickness of the passive layer because of an excess of film dissolution over its

formation. The approach to the new stationary state with time at a lower potential

252 Strehblow

Figure 4 Double logarithmic plot of the increase of the geometric current density with time

for electropolished iron during pitting corrosion, 1 h prepassivated at 1.18 V and potential drop

to 0.78 V (solid line), Δt = time between potential change and chloride addition, 1 h prepassivated

at 0.78 V (dashed line), phthalate buffer, pH 4.9, 0.1 M SO

2–

, 0.01 M Cl

–

. (From Ref. 6.)

should yield the original field strength with an increase of penetration. Again, the

opposite effect was found [6]. Even the adsorption mechanism cannot explain

these observations. If the potential is decreased, the voltage drop E

2,3

at the oxide-

electrolyte interface will decrease intermediately with a negative overvoltage η

2,3

for O

2–

formation from the water, i.e., the dissolution of the oxide, until a new

stationary state with E

2,3

= E

2,3,s

is achieved (Fig. 2). This situation should disfavor

the adsorption of Cl

–

at this interface, which again should slow down the penetration

instead of increasing it.

Film Breaking Mechanism

The occurrence of fissures within the passive layer is a possible explanation for the

observations mentioned last, especially for an nonstationary state of the passive

layer. A sudden change of the electrode potential even in a negative direction will

cause stresses within the film. Chemical changes [32] or electrostriction [7,8] is a

reasonable explanation. Chemical changes, i.e., a reduction of Fe(III) to Fe(II),

have been detected with XPS when negative potentials were applied to electrodes

passivated at positive potentials well within the passive range [32]. There also

exists direct evidence of film breaking events for these nonstationary conditions

according to observations with a rotating Fe-disk-Pt-ring electrode [33]. Even in

the absence of aggressive anions, a sudden potential decrease from 1.3 to 0.7 V

causes release of Fe

ions after a few seconds into the electrolyte, as detected by

a temporary peak of the analytical ring current detecting Fe

ions (Fig. 5). This is

a consequence of film breaking and repair events. Any crack within the passive

layer will lead to direct contact of small parts of the metal surface with the electrolyte

and thus to dissolution of Fe as Fe

with current densities that refer to the relatively

positive potentials, but without any protecting oxide at these defect sites. The

self-healing mechanism of the passive layer causes only temporary Fe

formation

of a few seconds duration. These potential changes should cause numerous defects

of nanometer dimensions within the passive layer in order to get a sufficient

amount of unprotected metal surface so that the temporary release of Fe

becomes

measurable. In the presence of aggressive anions, their direct access to the metal

surface will prevent repassivation and pits can form. The serious increase of the

nucleation rate (Fig. 4) after these potential changes is further evidence for the

suggested formation of a large number of defects in the nanometer range.

As already mentioned, pit nucleation is a very fast process. For well-passivated

Fe specimens, the time for the occurrence of a first pit is less than a second for

potentials well above the critical value E

, as determined by the increase of the

current density after the addition of chloride to the solution and direct microscopic

observation with a camera [4]. Most of the time is used to get the aggressive anions

to the specimen surface by convection. If chloride is already present in the electrolyte,

a short passivation below the pitting potential for approximately 1 s and a subsequent

potential increase cause pit formation within less than a millisecond [4]. Results

obtained with these electrochemical pulse techniques suggest that nonstationary

conditions of the passive layer favor the film breaking mechanism. The penetration

of aggressive anions would require much longer times than observed during these

pulse experiments. If the experimental conditions are not in favor of pitting, i.e., for

Mechanisms of Pitting Corrosion 253

small concentrations of aggressive anions (<10

–3

M) or low potentials in the vicinity

of E

, pit nucleation requires times longer than these extremely short times.

Long passivation times suggest that film breaking is a good explanation for

the pit nucleation for nonstationary conditions too. The current increase caused by

pitting slows down when the passivation time is increased up to more than 10 h

[6,31]. The passive layer is formed during the first seconds and does not grow

much after that time. This is a direct consequence of the barrier character of these

films, which leads to logarithmic or inverse logarithmic growth with time. Further

decrease of the passive current density and especially of the pit nucleation after

hours may easily be explained by constantly occurring film breaking and repair

events. The related stresses will heal with passivation time so that these events

become rare with time. Figure 6 shows an example of a decrease of pitting with

passivation time and the related reduction of the current increase [31].

254 Strehblow

Figure 5 Disk current I

and ring currents I

for Fe

analysis and I

for Fe

analysis

for a potential variation from E

= 0.7 V to 1.30 V and back. E

= 0.10 V, E

= 1.40 V,

0.5 M H

, pH 0.3, f = 1000 min

–1

, disk area A = 0.09cm

. (From Ref. 33.)

Measurement of the electrochemical current noise has the aim of correlating

the observed current fluctuations with breakdown and repair events that might lead

to the formation of stably growing pits [34,35]. In the view of these theories, the

application of statistical methods to the occurrence of current spikes and the

observed probability of pit formation lead to a stochastic model for pit nucleation.

The evaluation of current spikes in the time and frequency domains yields parameters

such as the intensity of the stochastic process λ and the repassivation rate τ [34].

They depend on parameters such as the potential, state of the passive layer, and

concentration of aggressive anions.

An interesting discussion of current measurements on microdimensional

electrodes of stainless steel wires is given by Mattin and Burstein [36]. Their

analysis of current transients at a very low level in chloride-containing 0.075 M

HClO

leads to the distinction of metastable and stable pits. According to their

discussion, the remaining passive layer protects the pit analyte from being diluted

from the bulk solution. Only when the film breaks off too small pits will repassivate,

whereas a few larger ones are deep enough to keep their local environment

undiluted so that they survive.

Adsorption Mechanism

The passive current density is influenced by the action of anions even for conditions

where pitting does not occur. Passive iron in 0.5 M H

shows stationary current

Mechanisms of Pitting Corrosion 255

Figure 6 Increase of current density for passive Fe with pitting corrosion in phthalate buffer,

pH 5.0 and 0.05 M Cl

–

at E = 1.38 V for different prepassivation times t

. (From Ref. 31.)

densities (7 μA/cm

) that are higher by approximately one order of magnitude

than for 1 M HClO

(<1 μA/cm

) [37]. This observation has been explained by a

catalytic effect of SO

2–

anions on the transfer of Fe

from the oxide to the electrolyte

[38]. This cation transfer is the rate-determining step for the passive corrosion

reaction. In the presence of SO

2–

, an FeSO

complex forms. It is reasonable that

this complex, with only one positive charge, requires less activation energy to be

transferred from its O

2–

ligands within the oxide matrix to the electrolyte than the

highly charged Fe

ion. The complexation of cations by organic reagents cause a

similarly enhanced dissolution in the passive state. The dissolution of Ni

from

passive nickel and nickel base alloys is enhanced by organic acids such as formic

acid, which may lead to the removal of NiO in the passive layer [39]. Similarly,

additions of citrate to the electrolyte cause thinning of the passive layer on stainless

steel and increase of the Cr content within the oxide layer. This is a consequence

of accelerated transfer of Fe

from the film surface to the electrolyte [40].

Apparently the complexation and transfer of Cr

are not enhanced.

It is well established that metals that show passivity within strongly acidic

electrolytes are far from the dissolution equilibrium of the oxide. The barrier

character of the passive layers for these conditions requires slow dissolution kinetics

at the oxide-electrolyte interface. Similarly to the catalytic effect of SO

2–

and

complexing organic agents, halides enhance the transfer of Fe

and Ni

from the

oxide to the electrolyte. The adsorption mechanism for pit nucleation starts with the

formation of surface complexes that are transferred to the electrolyte much faster than

uncomplexed Fe

ions (Fig. 2c). These details have been studied for Fe in Cl

–

- and

–

-containing solutions [21,41–43]. Similarly, Ni in solutions containing F

–

has been

studied [22,42–44]. Fe is a good metal with which to analyze the processes leading

to passivity breakdown because of the characteristic changes of the valence of

the dissolving ions, which may be measured with the rotating-ring-disk (RRD)

technique. Fe

ions are dissolved if the metal surface is still covered completely with

Fe(III) oxide, whereas Fe

is detected when bare metal comes into contact with

the electrolyte at a pit surface or at a defect site, similar to Fe dissolution in the active

state. The release of Fe

ions immediately after Cl

–

has access to a prepassivated Fe

electrode was taken as a measure of locally enhanced dissolution of the oxide film.

This leads to a local thinning of the passive layer and finally to its complete

breakdown and the formation of a pit. In the case of F

–

in acidic electrolytes the attack

is more general; i.e., the passive layer is subject to a general attack. As a consequence,

the measured passive current density is increased by orders of magnitude (Fig. 7)

[21,42,43]. Unfortunately, the strong FeF

2–

complex that finally forms in

solution prevents its detection by its reduction to Fe

at the analytical ring. The

current increases during the intermediate stage of attack with the first order of the HF

concentration. Therefore a proposed mechanism should yield an electrochemical

reaction order of one for HF. The following sequence of reactions has been proposed

for the processes that lead to thinning of the passive layer and enhanced passive

dissolution [21,42–44]:

256 Strehblow

The adsorption (8) is assumed to be fast and in a quasi-equilibrium. Reaction (9) is

the rate-determining step, and (10) is a fast following reaction step in solution

leading to the stable fluoro-aquo complex. These assumptions lead to a first-order

process for HF according to the rate equations (15) and (16). The Butler-Volmer

equation (11) will hold for the rate-determining step (9) with a charge transfer

coefficient α and with the potential drop E

2,3,s

at the oxide-electrolyte interface for

stationary conditions. E

2,3,s

refers to a solution of pH 0. According to the equilibrium

of O

2–

formation of Figure 2a and Eq. (13), one obtains Eq. (14) for the pH

dependence of the potential drop E

2,3,s

for stationary conditions of the passive layer:

Mechanisms of Pitting Corrosion 257

Figure 7 Current-time dependence of a rotating Pt-split-ring-Fe-disk electrode in 1 M

HClO

after HF addition to 0.1 M, 2 h prepassivation at 0.1 V in 1 M HClO

, Fe

detection

at ring 1. (From Ref. 21.)

Introducing equilibrium reaction (8) with its constant K according to Eq. (12):

and the equilibrium of O

2–

formation and the related pH dependence of the poten-

tial drop E

2,3,s

yields:

The effect of HF is a model for pitting insofar as it involves the total passive

surface so that the measured effects are much more pronounced and refer to a surface

of known size instead of an unknown actual pit surface that even changes with time.

The later stages of the attack of the passive layer lead to current peaks that go along

with equivalent Fe

formation (Fig. 7). Finally, a general breakdown of the passive

layer is observed with a steep increase of the dissolution current density and Fe

for-

mation [21,42,43]. Apparently, one may observe the different steps of breakdown of

passivity for fluoride directly, which are difficult to follow for the local events in the

case of the other halides. The difference in the action of fluoride and the other halides

may be explained by a comparison of the stability constants of their iron complexes as

presented in Table 1 [30]. The stability constants K

with one anion are given, which

refers to the reaction order one that has usually been found for the disssolution of

passive layers under the influence of the aggressive anions. The reaction of Fe

with

HF to FeF

and H

yields the more realistic value of log K

= 2.28 due to the small

dissociation constant of HF (pK = –log K

= 2.98). Table 1 also contains the constants

for Ni

- and Cr

- halide complexes. Their falling values from fluoride to iodide

and Fe

to Ni

support the decreasing tendency for enhanced dissolution of the

passive layer and localized corrosion. These data can be referred to the situation at

the oxide surface. The fluoro complexes are very stable and form with high concen-

trations at all surface sites. Therefore their much faster transfer to the electrolyte

yields enhanced general dissolution, whereas the attack of the other halides is locally

restricted and much less pronounced. Besides thethermodynamically based values, i.e.,

the stability constants, the kinetics of complex formation and of the complex transfer

to the electrolyte are another decisive factor for the attack of the passive layer. In this

sense, the situation of Cr

is very special and will be discussed separately.

XPS measurements of passivated Fe and Ni electrodes that have been

exposed to aggressive anions (Ni and Fe to F

–

; Fe to Cl

–

, Br

–

, and I

–

) but have

not already formed corrosion pits support this mechanism. The quantitative

evaluation of the data clearly shows a decrease of the oxide thickness with time

of exposure [22,48]. Not only F

–

but also the other halides cause thinning of the

passive layer (Fig. 8) [48]. The catalytically enhanced transfer of cations from

the oxide to the electrolyte leads to a new stationary state of the passive layer.

Its smaller thickness yields an increased electrical field strength for the same

potentiostatically fixed potential drop, which in turn causes faster migration of

the cations through the layer to compensate for the faster passive corrosion

reaction (1) at the oxide-electrolyte interface (Fig. 2a). Statistical local changes

258 Strehblow

Table 1 pK Values of Stability Constants of Metal Ion Complexes with Halides for the

Reactions Me

+ X

–

→ MeX

(z–1)+

including Constants Referring to HF (*) in Acidic

Solutions [45–47]