Marcus P. Corrosion mechanisms in theory and practice

Подождите немного. Документ загружается.

The type I crack population evolves in three steps following a first hardening

of samples, which needs 5 to 10% of the reduced lifetime (N/N

). The first step

consists of numerous nucleations of new cracks and leads to a surface density of

approximately 50 type I cracks per mm

in less than 25% of the reduced lifetime.

Crossing of the grain boundary corresponds to a second step during which the new

crack nucleation rate decreases by two thirds. A few type II cracks form. This step

extends up to 60 to 70% of the reduced lifetime N/N

. The type I crack density

reaches about 70 cracks per mm

. This is the highest type I crack density during

the test. Stage III of the damage process corresponds to the remaining lifetime.

There is a speeding up of damage accumulation. Numerous type II cracks form

while the type I crack population decreases by a half (final density is about 40

cracks per mm

). Their interactions and their surface propagation lead to generation

of type III cracks. Two to three type IV cracks are eventually formed for the total

specimen surface and propagate into the bulk, leading to rupture.

High short crack density and competitive growth of the multiple surface

short cracks during more than 90% of lifetimes suggest a random process to form

a fatal crack. High surface crack densities have been reported on other materials

submitted to fatigue tests [33–35]. Randomness must occur during both surface

crack nucleation and propagation. Consequently, crack interactions are likely to be

unpredictable one by one. As fatigue damage results from a combination of these

mechanisms (i.e., nucleation, propagation, and interactions), the usual approach of

fatigue through a single crack growth relation is unsatisfactory. As a matter of fact,

this approach does not take into account the effect of crack densities and hence the

statistical aspect of fatal crack formation.

A result [33,34] has been achieved emphasizing both the importance of the

statistical aspect of the fatigue damage accumulation and the role of type I surface

short cracks. An accumulation of elementary damage events, the cracking of a single

grain, was assumed and was shown to be successful in deducing the Coffin-Manson

relationship using the statistical physics of disordered systems.

A few numerical models have been developed to take into account the

developments and the interactions of surface short crack populations [34,36]. The

next paragraph briefly presents results of such modeling.

Numerical Modeling

The simulation has been developed according to the statistical aspect of the physical

damage accumulation process. Surface short cracks nucleate at random. Their

subsequent propagation is not continuous from a time viewpoint. Hence, a random

treatment was chosen to deal with both nucleation and propagation mechanisms.

Moreover, only surface data are taken into account because surface fatigue damage

dominates over 80% of the fatigue lifetime in low-cycle fatigue. The modeling

therefore belongs to a 2D Monte Carlo type [36].

Figure 22 shows predictions of crack density evolutions with statistical scatter.

Moreover, the simulations help to plot major crack developments. Figure 23a

illustrates the fatal crack evolutions. Interestingly, for Δε

/2 = ±4 × 10

–3

, the curve

clearly exhibits multiple coalescence even for crack length of the order of two grain

sizes (i.e.. 100 μm). The coalescences lead to significant increases of surface lengths.

Note the absence of those sharp increases in Figure 23b for Δε

/2 = ±4 × 10

–4

472 Magnin

More results are detailed elsewhere [36]. For example, Table 1 compares a few

experimental lifetimes with the corresponding simulation results.

Furthermore, the three short crack type densities correlate well. Also, the

screen display of the simulated specimen surface shows similar patterns of crack

spatial distribution (Figs. 23 and 24).

Finally, the relevance of the mesoscopic scale considered in the present study

(i.e., the grain size) can be emphazised when predicting the corrosion fatigue

behavior of 316L alloys in a 3.5% NaCl solution under free potential. In the

corrosive solution the fatigue lifetime decreases by a factor of 2 for an applied

plastic strain Δε

/2 = ±4 × 10

–3

[11]. In this case the localized anodic dissolution at

the slip band emergence increases mainly the number of nucleation sites, the surface

micro-propagation threshold (particularly to cross the grain boundaries), the kinetics

of surface propagation, and then the formation of the fatal crack. The experimental

evolution of the short crack densities is shown in Figure 25a. It must be noted that

the evolution of type I, II, and III crack populations is homologous to its evolution

Corrosion Fatigue Mechanisms 473

Figure 22 In-air test simulations. Surface short crack density vs. N/N

at (a) a

plastic strain amplitude of 4 × 10

–3

, N

= 6800 ± 800 cycles and (b) a plastic strain

amplitude of 4 × 10

–4

, N

= 250,000 ± 35,000 cycles. Scatter is calculated from

several computations.

Figure 23 In-air test simulations. Major crack development vs. the number of cycles. (a)

Δε

/2 = ±4 × 10

–3

and (b) Δε

/2 = ±4 × 10

–4

. Coalescence mechanisms participate strongly

in formation of a fatal crack at Δε

/2 = ±4 × 10

–3

for in-air tests when referring to N/N

. The simulation (Fig. 25b) shows satisfactory

agreement with experiments. A very interesting tool for corrosion fatigue prediction,

based on the effect of the corrosive solution at the mesoscopic scale, is then available.

This is new in this field. Other effects of a corrosive solution (such as pitting) can

now be introduced in the software to simulate other possible corrosion fatigue

mechanisms in metal and alloys.

Finally, such Monte Carlo treatment has been successfully applied to other

fcc single-phase materials [37].

INTERACTIONS WITH STRESS CORROSION CRACKING (SCC)

It is well known that SCC can interfere with corrosion fatigue during crack

propagation (Fig. 13). But this interaction also occurs for crack initiation and

crack propagation near the threshold, as illustrated in Figure 3 for an fcc Fe-Mn-Cr

alloy cyclically deformed at low strain rate in a Cl

–

solution and in Figure 8 for

Al-Li alloys in 30 g/L NaCl. On the fracture surface of the Fe-Cr-Mn alloy, a

crack initiates from a pit (which is favored by fatigue processes) by SCC with a

cleavage-like process that has been described in detail elsewhere [22]. Then

fatigue and corrosion fatigue take place with classical striations. Finally, it must

be noted that the distinction between stress corrosion cracking and fatigue cracking

is not always easy to make because SCC can also be discontinuous (see the

mechanisms in Ref. 22) with crack arrest markings very similar to fatigue ones.

Thus, combination of pure SCC and fatigue mechanisms is often possible during

corrosion fatigue damage.

CF can sometimes be more dangerous than SCC because it occurs whatever

the strain rate is. This is illustrated in the case of a 7020 Al-Zn-Mg alloy for two

different heat treatments, T4 (underaged) and T6DR (peak aged). Figure 26 compares

the SCC behavior through the ratio between the elongation to fracture in 30 g/L NaCl

and the elongation in air for tensile tests at an imposed strain rate with the CF

behavior at a given applied plastic strain Δε

/2 = 10

–3

, for different strain rates.

One can see that, whatever the strain rate, CF always occurs, unlike SCC, which

takes place only below a critical strain rate dependent on the heat treatment.

Moreover, CF is affected by SCC below the critical SCC strain rate, as shown by

the fact that crack initiation is intergranular only when SCC occurs.

474 Magnin

Table 1 Comparison of Experimental and Fatigue Lifetimes (316L Alloy at ε

= 10

–3

–1

)

Nevertheless, cyclic loading is not always deleterious in regard to SCC. The

example of cracking in alloy 600 for a pressurized water reactor (PWR) illustrates

such behavior. Intergranular SCC is known to occur in the alloy 600/PWR system

[38] when tests are performed on CT specimens for constant loading conditions

with a stress intensity factor K

= 30 MPa√

—

m. The effect of cycling loading at a

load ratio R = 0.7 with a triangular wave form at K

max

= 30 MPav√

—

m is shown in

Figure 27 according to the imposed frequency.

An increase in frequency induces a decrease of the crack growth rate and a

transition from intergranular to transgranular cracking mode until air cyclic behavior

becomes predominant.

Corrosion Fatigue Mechanisms 475

Figure 24 Screen display of modeling surface. Δε

/2 = ±4 × 10

–3

This last example clearly shows that cycling can completely change the

crack propagation mode when stress corrosion crack propagation is intergranular.

CONCLUDING REMARKS

The analysis of CF micromechanisms related to anodic dissolution and hydrogen

effects is in progress but needs to be more quantitative through the localized corrosion-

deformation interactions. The trends for future research are mainly related to:

476 Magnin

Figure 25 Influence of the environment on the evolution of type I, II, and III surface

crack densities vs. the number of cycles (thick lines are for in-air test, thin ones for a 3.5%

NaCl solution test). (a) Experimental data and (b) simulation results.

Figure 26 Comparison of SCC and CF (at Δε

/2 = 10

–3

) behavior of a 7020 T4 and

T6DR alloy at free potential in a 30 g/L NaCl solution. Crack initiation can be transgran-

ular (T) or intergranular (I) according to the strain rate ε

1. The modeling of crack tip chemistry

2. The quantitative analysis of corrosion-deformation interactions at the CF

crack tip (scale of 1 μm) according to the electrochemical conditions

3. A comparison between CF and SCC based on a detailed analysis of micro-

mechanisms near the fatigue threshold

4. Development of numerical simulations at mesoscopic scales.

These studies are needed to propose predictive laws for CF damage

increasingly based on physicochemical controlling factors.

REFERENCES

1. C. Laird and D. J. Duquette, Corrosion Fatigue, NACE, New York, 1972, p. 82.

2. C. Patel, Corros. Sci. 21:145 (1977).

3. (a) R. P. Wei and G. W. Simmons, Int. J. Fract. 17:235 (1981); (b) J. Congleton, J. H.

Craig, R. A. Olich, and R. N. Parkins, Corrosion Fatigue, Mechanics, Metallurgy,

Electrochemistry and Engineering, ASTM 801 (T. W. Crooker and B. W. Leis, eds.,

ASTM, Philadelphia, 1983, p. 367.

4. R. P. Gangloff and D. J. Duquette, Chemistry and Physics of Fracture (R. Latanision,

ed.), Nijhoff, Dordrecht, 1987, p. 612.

5. S. P. Lynch, Acta Metall. 36, L10:2639 (1988).

6. R. O. Ritchie and J. Lankford, Small Fatigue Cracks, TMS-ASME, 1986, p. 33.

7. K. J. Miller and E. R. de Los Rios, The Behaviour of Short Fatigue Cracks, EGF1,

Mech. Eng. Pub., London, 1986.

8. M. Mueller, Metall. Trans. 13A:649 (1982).

9. B. D. Yan, G. C. Farrington, and C. Laird, Acta Metall. 33:1593 (1985).

10. T. Magnin and L. Coudreuse, Acta Metall. 35:2105 (1987).

11. T. Magnin and J. M. Lardon, Mater. Sci. Eng. A 104:21 (1988).

Corrosion Fatigue Mechanisms 477

Figure 27 Crack growth rates versus frequency during CF and SCC on CT alloy 600

specimens in a PWR environment [38].

12. P. Scott and R. A. Cottis, Environment Assisted Fatigue, EGF/MEP, London, 1990.

13. G. Henaff and J. Petit, Corrosion Deformation Interactions CDI’92 (T. Magnin ed.),

Les Editions de Physique, Paris, 1993, p. 599.

14. L. Hagn, Mater. Sci. Eng. A 103:193 (1988).

15. R. Akid, Env. Degradation of Eng. Materials, Vol. I, Technical Univ. Gdansk and

Univ. Bordeaux 1, 1999, p. 112.

16. A. Navarro and E. R. de Los Rios, Fat. Fract. Eng. Mater. Struct. 11:383 (1988).

17. W. Zhang and R. Akid, Fat. Fract. Eng. Mater. Struct. 20:167 (1997).

18. P. D. Hobson, M. W. Brown, and E. R. de Los Rios. The Behaviour of Short Fatigue

Cracks (K. J. Miller and E. R. de Los Rios, eds.), EGF1, London, 1986, p. 441.

19. A. Turnbull, Corrosion sous contrainte (D. Desjardins and R. Oltra, ed., Les Editions

de Physique, Paris, 1990, p. 79.

20. C. J. Van der Wekken, Corrosion Deformation Interactions CDI’96 (T. Magnin, ed.),

Institute of Materials, London, 1997, p. 76.

21. C. Fong and D. Tromans, Metall. Trans. 19A:2765 (1988).

22. T. Magnin, Advances in Corrosion Deformation Interactions, Trans Tech, Zurich,

1996 (see p. 126, with reference to Gerberich et al., Parkins Symposium, TSM, 191,

1992).

23. D. Delafosse, J. P. Chateau, and T. Magnin, J. Phys. IV 9:251 (1999).

24. J. P. Hirth and J. Lothe, Theory of Dislocations, Krieger Publications, Miami,

Florida, 1992.

25. P. C. Paris and F. Erdogan, J. Basic Eng. Trans. AIME D 85:528, (1963).

26. B. Tomkins, Philos. Mag. 18:104 (1968).

27. K. J. Miller, Fatigue Eng. Mater. Struct. 5:223 (1982).

28. T. Magnin, J. Driver, J. Lepinoux, and L. P. Kubin, Rev. Phys. Appl. 19:467 (1984).

29. H. Mughrabi, F. Ackermann, and K. Herz, Fatigue Mechanisms, ASTM STP 675

(J. T. Fong, ed.), ASTM, Philadelphia, 1979, p. 69.

30. T. Magnin, L. Coudreuse, and J. M. Lardon, Scr. Met. 19:1487 (1985).

31. C. Ramade, thesis, St Etienne, 1990.

32. K. J. Miller, H. J. Mohammed, and E. R. de Los Rios, The Behaviour of Short

Fatigue cracks (K. J. Miller and E. R. de Los Rios, eds.), Mech. Eng. Pub., Sheffield,

1986, p. 491.

33. Y. Brechet, T. Magnin, and D. Sornette, Acta Metall. Mater. 40:2281 (1992).

34. A. Bataille, T. Magnin and K. J. Miller, Short Fatigue Cracks (K. J. Miller, ed.),

Sheffield, Esis 13, Mech. Eng. Pub., 1997.

35. J. Weiss and A. Pineau, Low-Cycle Fatigue and Elasto-Plastic Behaviour of Metals

(K. T. Rie, ed.), Elsevier Applied Science, Amsterdam, 1993, p. 82.

36. A. Bataille and T. Magnin, Acta Metall. Mater. 42:3817 (1995).

37. A. Bataille, thesis, University of Lille, 1992.

38. C. Bosch, M. Puiggalli, and J. M. Olive, Corrosion Science, to be published.

478 Magnin

Corrosion Prevention by Adsorbed

Organic Monolayers and Ultrathin Plasma

Polymer Films

Michael Rohwerder, Guido Grundmeier, and Martin Stratmann

Max-Planck-Institut für Eisenforschung GmbH, Düsseldorf, Germany

INTRODUCTION

Traditionally, corrosion protection of many reactive materials is associated with

organic coatings, which are applied as thick multilayer films with thicknesses

ranging from some 10 μm to some 100 μm. However, increasing interest exists in

thin and ultrathin corrosion protecting films. This is particular true for modern areas

of materials research such as microelectronic devices or micromechanics. In both

cases, any corrosion protecting film must be much less than 0.1 μm in thickness

and frequently even monolayer films are wanted. However, even for conventional

materials such as steel, galvanized steel, or Al-based alloys, thin film coatings are of

considerable interest as frequently such films are generated with unusual properties

by modern surface technology. For such applications the time scale for film

preparation must be in the seconds range and many techniques allow only the

preparation of thin films under those restrictions. Then the thin layer coating may

serve as an inner film only with top coats prepared by conventional techniques. The

possible role for the thin film then lies in corrosion protection as well as in adhesion

promotion of the additional polymeric layers. This chapter will focus on two surface

modification techniques: molecular self-assembly and plasma polymerization.

The corrosion rate of reactive metals can be reduced significantly by a

modification of the metal surface by organic molecules or polymers. A well-known

example is the corrosion protection by lacquers and other organic coatings [1–3],

which is used in most practical applications to protect, e.g., cars against atmospheric

corrosion, pipelines against corrosion in humid soil, and ships against corrosion in

seawater. It has long been believed that the corrosion protection is due to the barrier

properties of the coating, which impedes the penetration of water and oxygen [4] to

the metal/polymer interface. However, many coatings are highly permeable for water

and oxygen, and therefore it is not the barrier effect on the diffusion process that

gives rise to the corrosion stability but the specific electrochemical properties of the

metal/polymer interface, in particular the formation of an extended diffuse double

479

layer [5–7]. In the presence of defects (pores, pinholes, etc.), which may penetrate

through the coating, the diffusion barrier is lowered, and the delamination rate of the

polymer at the defect is determined by the formation of galvanic elements [8–12].

In many circumstances, the local anode of this elements is the defect and the local

cathode, at which predominantly oxygen is reduced, is given by the delamination

frontier. In other cases—in particular for Al-based alloys—the galvanic element is

just opposite and the delamination frontier is characterized by the local anode. In

the first case the stability of the metal/polymer interface is determined by inhibition

of the oxygen reduction at this interface, as during the electrochemical reduction of

oxygen very aggressive species are liberated (OH·, OH

–

), which will immediately

attack and destroy chemical bonds within the polymer [13–16]. The corrosion

inhibition of the coating therefore depends more on the composition, structure,

and chemical bonds of the metal/polymer interface than on the thickness of the

coating.

It is therefore desirable to improve the chemical interaction between the first

monolayer of the coating and the substrate such that electrochemical reactions such

as the reduction of oxygen are inhibited and the bonds may withstand the attack of

water and other aggressive species such as OH·. Modern concepts are aimed at the

development of “molecular adhesion promoters” (Fig. 1) [17–19], which will provide

a link between the substrate and the organic coating. This model requires, however,

480 Rohwerder et al.

Figure 1 Principal scheme of the chemical modification of metal surfaces.

that reactive centers are prepared on top of the metal substrate, which may serve

as anchor sites for suitable functional groups of the organic molecules, as shown

in Figure 1. The question is how to select reactive centers, organic molecules, and

reaction conditions [20–25].

In order to do so, it is useful to study the relevant literature, which is concerned

with organic molecules being used as corrosion inhibitors [26–29]. It is well known

that many organic compounds are able to reduce the corrosion rate of metals

significantly [30–34]. Inhibitors reduce the rate of either or both of the partial

reactions of the corrosion process, the anodic metal dissolution and the cathodic

oxygen reduction. Various inhibition mechanisms must be considered, taking into

account the different situations created by changing, on the one hand, the chemical

properties of the inhibitor and, on the other hand, the medium. The organic

compounds may either be dissolved in the electrolyte (conventional inhibitors) or

adsorbed onto the metal surface by condensation from the vapor phase (volatile

corrosion inhibitors). From a mechanistic point of view [35], the inhibitors may

change the composition of the medium (e.g., hydrazinium reduces the oxygen

activity) or they may be enriched at the metal-medium interface. In the latter case

they may enhance the formation of passive films on top of the substrate, such

as benzotriazole on Cu [30,31] or benzoates on Fe (passivators), or they may form

monomolecular adsorption layers and prevent the dissolution of the substrate and the

reduction of oxygen by changing the potential drop across the interface and/or the

reaction mechanism (interface inhibitors) [36]. The interaction between organic ions

or dipoles and electrically charged metal surfaces may be purely electrostatic, due to

the electric field at the outer Helmholtz plane of the electric double layer. A

competitive adsorption between water dipoles and the monomer has to be considered,

and water adsorption always takes over the adsorption of nonpolar monomers if the

electrode potential is shifted far from the potential of zero charge (PZC), as then

the surface is highly charged [37]. The adsorption of the monomer prevails in the

vicinity of the PZC. Cations are adsorbed at potentials negative to PZC, anions at

potentials positive to the PZC. Because of the dependence on the charge of the metal

surface, the adsorption is not specific for certain metals.

The charge of the anions or cations may be constant but also may charge during

adsorption. The number of electrons exchanged per adsorbed molecule is called the

adsorption valence (typically less than 1) [38–40]. Because of the low activation

energy, the adsorption based on pure electrostatic interaction is fast and reversible. If no

electron transfer occurs, this kind of adsorption is frequently called physisorption.

However, during adsorption an electron transfer between the substrate and the adsorbed

molecule may occur with orbital overlap between a single pair of electrons of the

adsorbed molecule (N, S, P) and empty bands of the solid. Strong interaction is then

observed (chemisorption) and the adsorption itself is highly irreversible [41–43].

Chemisorption is slower than physisorption, and it requires a higher activation energy.

In contrast to physisorption, it is specific for certain metals. Electron transfer is typical

for transition metals having vacant low-energy d-electron orbitals. On the other hand,

the inhibitor should have free single electron pairs, or π electrons. The higher the polar-

izability of the involved heteroatom, the stronger is the adsorption (Lewis acid-base

concept). The inhibitor is then the electron donor and the metal is the electron acceptor.

This is in agreement with the principle of soft and hard acids and bases (SHAB).

Corrosion Prevention by Adsorbed Monolayers 481