Marcus P. Corrosion mechanisms in theory and practice

Подождите немного. Документ загружается.

The first idea to be proposed to account for the effect of soluble sulfides on

pitting initiation was that MnS dissolution provokes locally the formation of a virgin

metal surface. This microarea is exposed to an acidified and sulfur species–enriched

environment produced by the sulfide dissolution. When the solution near the

microarea has reached a certain composition, the contacting metal can no longer

repassivate and the metal starts to dissolve [6]. Following other workers [8], complete

MnS dissolution is not needed, as the pits preferentially initiate at the metallic matrix-

MnS interface. This model keeps open the questions of (a) the MnS dissolution

mechanisms, (b) the nature and the effect of the dissolved sulfur species which form

during this dissolution and (c) the role played by the passive film, either close to or

possibly on the inclusion. From this last viewpoint, the surface condition of the steel

is of major importance. The effect of MnS is probably not the same on freshly

mechanically polished surfaces or after various surface treatments, including

pickling, bright annealing, or electrochemical treatments.

There is some experimental evidence that Cl

–

ions cluster either on the

nonmetallic inclusions [10b] or at their boundaries with the metallic matrix [12b].

When the MnS is directly exposed to the electrolyte, Cl

–

can preferentially adsorb on

the inclusion due to its higher electronic conductivity, giving stronger electrostatic

image forces than on the surrounding oxide film [8]. In our opinion, Cl

–

adsorption

is followed by the potential-assisted formation of the adsorbed complex MnCl

which then dissolves in the aqueous solution according to (MnCl)

ads

+ Cl

–

→ Mn

+ 2Cl

–

. Because their is some evidence that pitting initiation on MnS-containing

steels is strongly dependent on the solution pH (see below), it is suggested that Cl

–

and OH

–

adsorption compete on Mn sulfide surfaces, resulting in pH-dependent

dissolution kinetics.

The MnS dissolution provokes the formation of an Mn

- and Cl

–

- enriched

local electrolyte [10c] (the Cl

–

ions being attracted by the local excess in positive

changes in the solution). As long as the concentration remains below a critical

level, the pit walls can repassivate, but beyond this critical level an MnCl

salt layer

is formed which may prevent the repassivation. The question of what happens at

the metal–salt layer interface has been discussed [16a] and it was suggested that an

FeCr oxychloride could form, whose properties, and also possible remnance once

the salt layer has dissolved, could play a determining role in repassivation mecha-

nisms. Following another idea, MnS would be covered with a defective passive

film [lc] on which the chloride ions first adsorb and then penetrate, leading to the

formation of a nonprotective salt chloride layer. The hydrolysis of this salt layer

would then increase the local acidity, resulting in the dissolution of the defective

passive film. Whatever the proposed mechanisms, chloride adsorption seems to be

the first stage of the sulfide dissolution.

Several models have been proposed for the dissolution of sulfur from MnS

inclusions. The first idea [6,8] is that, in acidified media, MnS dissolves according to

MnS + 2H

→ H

S + Mn

or [17]

MnS + 2H

→ S + Mn

+ H

These reactions can be split into the anodic reaction [6,8,10b]

320 Baroux

MnS → S + Mn

+ 2e

–

whose potential is V(volts) = –0.12 + .0.03 log [Mn

], and the cathodic ones,

+ e

–

↔ 1/2H

or [6] 2H

+ 2e

–

+ S ↔ H

with V = 0.14 – 0.06pH – 0.03 log [H

S], and possibly [5c]

S ↔ H

+ HS

–

, then H

+ 2e

–

+ S ↔ HS

–

Direct action of water to form sulfates, sulfites or thiosulfates should also be

considered [6,8,10b,15b,17].

MnS + 4H

O ↔ Mn

+ SO

2–

+ 8H

+ 8E

–

with V = 0.23 – 0.06pH – .0074 log[Mn

][SO

2–

MnS + 3H

O ↔ Mn

+ HSO

–

+ 5H

+ 6e

–

then

HSO

–

+ 5H

+ 4e

–

→ S + 3H

2MnS + 3H

O ↔ 2Mn

+ S

2–

+ 6H

+ 8e

–

2–

+ H

O → S

ads

+ SO

2–

+ 2H

+ 2e

–

All these reactions are potential and/or pH dependent, so they can occur or not

according to the actual conditions in, or close to, the pit embryo. This makes rather

complex the interpretation of the pitting potential pH dependence when sulfides act

as pitting sites (see below). Furthermore, they do not explicitly take into account the

effect of chloride ions for the dissolution of Mn

; they should then be slightly

modified for this purpose, introducing the solution chloride content as the third

critical parameter for the MnS dissolution kinetics.

It should be noted that photoelectrochemical investigations [15b] suggest

that the last two reactions, involving the formation and then dismutation of S

2–

and resulting in the formation of adsorbed sulfur, are operating in some cases.

Photoelectrochemical microscopy shows the deposition around the inclusions

of a ring of material deduced to be sulfur, which is consistent with the results of

several other studies [6,10b]. A detailed analysis of the fundamental aspects of the

sulfur-assisted corrosion mechanisms is presented in Chapter 9.

AN EXAMPLE OF THE EFFECT OF NONMETALLIC INCLUSIONS

Studied Steels and Their Nonmetallic Inclusions

In the following, the steels under consideration are mainly some annealed AISI

430-type FeCr ferritic alloys containing also either Nb (steels A and A') or Ti

(steels B and B') additions. Also, some solution-treated 304 and 321 AlSI-type

steels were considered (C and D), in order to separate the matrix and inclusion

effects. These two steels are austenitic and Ni bearing but the latter contains Ti and

is MnS free, whereas the former contains MnS inclusions.

Pitting Corrosion of Stainless Steels 321

Except for steel A, some Al additions are used as deoxidizing agents during the

melting process, which leads to Al ~ 0.030% and induces the presence of some Al

inclusions. Note the low sulfur content, which is easily attained with modern

steelmaking techniques, and the presence of some stabilizing elements (Ti or Nb),

which trap carbon and avoid the formation of chromium carbides. For steels A and

A', sulfur is trapped under the form of manganese sulfide. The difference between

steels B and B' is their Cr content; both contain Ti, which also traps nitrogen in the

form of titanium nitrides and sulfur in the form of Ti sulfides; this trapping occurs

during the steelmaking process before the steel solidification. Concerning steel A',

some Mn sulfides are found around aluminum oxides (Fig. 5). Because these oxides

have a poorer ductility than the metallic matrix, the cold-rolling process provokes

some microdecohesions around the inclusion, where some manganese sulfides are

often located, leading to the possible formation of noxious microcrevices. This

phenomenon is not observed on Al-free steels, where oxides are mainly malleable

silicates. However, the art of the steelmaker consists in avoiding the formation of Cr

oxides, which could produce a similar but worse effect than Al oxides. In every case,

some Mn sulfides are also found closely stuck to Nb carbonitrides (Fig. 6a). Because

these carbonitrides precipitate at ~1200°C, this shows that (for the sulfur content

considered), the high-temperature sulfur solubility is sufficient for MnS precipitation

to occur in the solid steel (lower than 1200°C) while for higher sulfur contents MnS

is generally considered to precipitate at the end of the steel solidification. Note that

very few isolated MnS precipitates are found, probably because Nb(C,N) act

as precipitation sites. This situation contrasts with the behavior of steel C (304),

for which (a) no carbides may nucleate the MnS precipitate and (b) sulfides can

precipitate in austenite at the delta→gamma transformation temperature.

Figures 5b and 6b show the location of Ti sulfides on steel B, i.e., around the

titanium nitrides, embedded in a Ti carbide belt. Detailed examination suggests that

al high temperature a homogeneous Ti carbosulfidc belt precipitates around the Ti

nitrides (which formed in the liquid steel). Then, lowering the temperature, titanium

sulfides and carbides separate at some points as shown in Figure 5b, producing the Ti

carbide TiC and a Ti sulfide, identified in some cases (using electron diffraction) as

the hexagonal phase Ti

S. As another result, note that Al oxides have been identified

in the core of the Ti nitrides (which is not known in the figures), playing the role of

nuclei for TiN precipitation.

322 Baroux

Table 1 Steel Composition in wt % or (ppm)

pH Effects and Pitting Sites

The pitting potentials were measured for all these steels using the potentiokinetic

method in NaCl aqueous media whose pH was varied from 3 to 6.6. The results for

steels A' and B in NaCl (0.02 M) are shown in Figure 7a. No pH dependence is

observed for steel B. In contrast, for steel A', a sharp pitting potential decrease is

evident when the pH is lowered below a critical value pH

ranging between 4.5 and 5.

Since the main difference between the two steels is the presence or absence of MnS,

and Ti sulfides are known to have better stability in aqueous electrolytes than MnS,

Pitting Corrosion of Stainless Steels 323

Figure 5 Nonmetallic inclusions (before pitting). Scanning electron microscopy, ×3000.

(a) Steel A': MnS nucleated on an aluminum oxide. (b) Steel B: Ti nitride surrounded by

sulfur compounds.

324 Baroux

Figure 6 Nonmetallic inclusions (before pitting). STEM (×18,000) on thin foils. (a)

Steel A': MnS nucleated on an Nb carbonitride. Intermetallic (Fe, Nb) phases are also

observed, (b) Steel B: Ti nitride surrounded by a Ti carbide, in which some Ti sulfides

are embedded.

one can assume that this decrease is due to the pH-assisted MnS dissolution. Note

that the same phenomenon is observed (Fig. 7b) when comparing the two steels C

and D, which are respectively MnS containing the MnS free, showing that the

observed effect is related to the nature of sulfides present in the steel and not to other

metallurgical factors. Figure 7b also shows the results obtained for steels, A and B',

which confirm the above findings. The slight difference between steels B and B' may

be attributed to the difference in Cr concentration. The strong difference between

Pitting Corrosion of Stainless Steels 325

Figure 7 Effect of the solution pH on the pitting potentials. (a) Steels A' and B in NaCl

(0.02 M). (b) Steels A, B' (FeCr steels) and C, D (FeCrNi steels) in NaCl (0.02 M). (c)

Effect of the solution chloride content dependence for a Ti-containing steel. For low

chloride contents (0.02 and 0.1M), there are no pH effects. For 0.5 M, lowering the pH

decreases the pitting resistance.

steels A and A' may be due to the difference in Cr concentration and sulfur contents

but also to the Al oxides present in steel A' and around which MnS are found. The

effect of the solution chloride content was also investigated between 0.02 and 0.5 M.

Figure 7c shows that for high enough chloride concentrations, a pitting potential pH

dependence is found even for Ti-containing steels, suggesting that Ti sulfides are not

so stable in such electrolytes. The discontinuity of the pitting potential versus pH

variations should then be related to the dissolution of sulfur species, which occurs

easily for MnS-containing steels (whatever the steel matrix composition) but only for

high enough chloride contents for Ti-bearing steels. Note that the critical pH which

is found (4.5<pH<5) is the same for Ti-free and Ti-bearing steels and is close to the

one deduced from the potential-pH equilibrium diagrams in some chloride-

containing aqueous solutions [6,8]. This could support the idea that the critical pH

can be deduced from the sulfur species-pH equilibria, regardless of the nature of

the dissolved cation (manganese or titanium), once the conditions are reached for

the sulfide to dissolve.

Figure 8 shows the typically observed pit initiation sites. For steel A', whatever

the pH or the chloride content, pits initiate either around Al oxides (Fig. 8a), where

some MnS is located, or on MnS inclusions (Fig. 8b) around Nb(C,N) or (seldom)

isolated. This confirms the preceding hypothesis. For steel B in NaCl (0.5 M)

solution, pitting generally occurs at the TiN boundary (Fig. 8c), where Ti sulfides are

present. This shows clearly that for such chloride concentrations, Ti sulfides act as

pitting sites and becomes unstable when the potential increases. For lower chloride

concentration (0.2 M NaCl), the situation is not so clear and pits could initiate

directly on the metallic matrix, with no direct relation to nonmetallic inclusions, or in

some cases on titanium nitrides. In the two situations, however, Ti sulfides do not

seem to act as pitting sites, which is consistent with the absence of any pitting

potential pH dependence.

A practical consequence can be drawn from these results: Ti-stabilized steels

exhibit better pitting resistance than Ti-free ones, provided that the corrosive medium

is not too severe, i.e., the chloride content does not exceed a critical value, depending

on the solution pH, which could cause the Ti sulfides destabilization. In the case of

crevice corrosion, the situation is more complex. Inside a crevice, the pH decreases

and the chloride concentration increases with time. It is an accepted idea that

corrosion occurs when the pH becomes lower than a critical value, which is referred

to as the depassivation pH. However, at least for the less resistant steels, pitting

corrosion may occur in the crevice before this general depassivation. From these

results, it is concluded that MnS-containing steels are much more sensitive to this

pitting-induced crevice corrosion than Ti-bearing ones. For very severe crevices,

however, the chloride content can drastically increase with time Ti-stabilized steels

are no longer different from Ti-free ones.

Inhibitive Effect of Sulfate Ions

Sulfate ions are known for inhibiting the pitting corrosion in chloride-containing

media. Figure 9 presents the elementary pitting probability–potential variations

obtained for steels A and B in 0.1 M NaCl. Three points are noticeable: (a) SO

–

additions decrease the pitting probability (for a given potential) and increase the

conventional pitting potential (for a given chloride concentration). This effect is

326 Baroux

Figure 8 Scanning electron microscopy (×3000) on pitted samples. (a) Steel A': pitting

at the boundary of an alumina inclusion. (b) Steel A': pitting on a manganese sulfide. (c)

Steel B: pitting at the boundary of a Ti nitride.

more marked for steel A than for steel B, but complementary tests in 0.5 M NaCl

show that the difference between the two steels regarding the sulfate addition

efficiency becomes smaller when the chloride concentration increases. (b) Sufficient

sulfate additions result in a deviation from the standard exponential ω(V) law. The

sulfate amount necessary to produce a noticeable deviation is much larger for steel A

than for steel B. (c) For high enough sulfate amounts, steel B exhibits a bimodal

behavior: exponential ω(V) laws are found both for high and for low electrode

potentials. Between these two domains a transition behavior is noted. Further work is

needed to interpret these results, but the difference between the MnS-containing and

MnS-free steels is now well established.

328 Baroux

Figure 9 Effect of sulfate ions on the pitting probability in 0.1 M NaCl (pH 6.6). I is the

sodium sulfate molarity and V the electrode potential. (a) Steel A; (b) steel B.

Prepitting Events

Recording the anodic current i(t) during a polarization below the conventional pitting

potential (Fig. 10a) provides two types of information. First, the average anodic

current decreases with time, probably corresponding to the onset of improved

passivity. Second, some oscillations of this anodic current are observed, which,

through the same SEM observations as above, were shown to correspond to some

initiated and then repassivated pits. The problem which arises for analyzing these

oscillations is to separate them from the average anodic current decay (the “anodic

current baseline”), which probably corresponds to some passive film modifications.

The baseline is determined by associating to each point [t, i(t)] the point [t, i

'(t) = min

{i(t

'), t' = t– τ/2 to t + τ/2}] where τ is an arbitrary time constant. The frequency 1/τ

should be smaller than the characteristic frequencies of the prepitting noise. Figure 10b

shows the typical reduced signal obtained for steel A' using this method. For MnS-

containing steels (A or A'), the prepitting events are numerous and produce what one

may call a prepitting noise. In contrast, for MnS-free steels the anodic events are

much less numerous and well seperated, as shown in Figure 10c. It is therefore

logical to relate the so-called prepitting noise to the dissolution of manganese

sulfides. For MnS-containing steels, the current time signature of the individual

prepitting events is found to be a current increase varying approximately as t

followed by a sharp current decrease (Fig. 11a). The peak amplitude at 200mV/SCE

is of the order of 0.1 to 1 μA. For the MnS-free steels, the form of the i(t) transient is

quite different (type II; see Fig. 11b) and the peak intensity is smaller (10 to 100nA

at 200mV/SCE). In neutral NaCl (0.02M) solution, the transient is characterized by

a sharp increase followed by a slow current decay, as expected for local passive film

breakdown followed by passive film healing.

A simple way to analyze the reduced signal corresponding to the prepitting

noise (Fig. 10b) is to build up a “distribution function” N(I) which counts, for a given

time period T(~ 8 min), the number of events for which the anodic intensity i is

larger then I. This function N decreases with I (Fig. 12a). One sees that the prepitting

noise increases with the electrode potential and the solution chloride content.

Moreover, the solution pH acts in the same way as for the pitting potentials, since a

discontinuity of N between pH 4 and 5 is once more evidenced (Fig. 12b). Last, the

prepitting noise decreases with the polarization time (Fig. 12c).

Another way is to calculate the power spectral density (PSD) associated with

the anodic current fluctuations. Let us consider a Poissonian series of birth and

death events (birth frequency λ, death frequency μ), with a parabolic growth law

between birth and death. This situation can be modeled as follows:

i(t)=∑

(t) with i

(t)=

–

(t – t

)

H(t – t

)[1 – H(t – t

– θ

)]

where t

and Θ

are, respectively, the induction time and the lifetime of the event

number n, H is the step function [H(x) = 0 when x < 0 and 1 when x > 0], and α is a con-

stant. The average current 〈i〉 and the PSD Ψ(f ) are found [4a] to be 〈i〉 = αλ/μ

and

Pitting Corrosion of Stainless Steels 329