Anda di halaman 1dari 20

CorrosionScience,Vol. 39, No. l&l 1, pp. 1771-1790, 1997 0 1997Elsevier Science Ltd Printed in Great Britain.

All rights reserved 001&938X/97 317.00+0.00 PII: soolo-938x(97)

LOCALISED

DISSOLUTION KINETICS, SALT FILMS AND PITTING POTENTIALS


N. J. LAYCOCK and R. C. NEWMAN

UMIST, Corrosion and Protection Centre, PO Box 88, Manchester M60 IQD, U.K Abstract-Dissolution kinetics within artificial pit electrodes have been studied and related to the transition in real pits from metastability to stability. For 302 and 316 stainless steel, two regimes of growth were identified; mixed activation/ohmic control was found at lower potentials, but at higher potentials growth was under diffusion control with a metal salt film present on the electrode surface. The transition potential, E-r, between these two regimes was found to increase linearly with the log of the limiting current density, ilim. A mode1 for the anodic dissolution kinetics is proposed which accounts for this observation and predicts that real pits will also grow in either active or salt filmed states depending on their current density. Pitting potentials, Epi,, and transition potentials, ET, were measured as a function of chloride concentration. The differences in Epit between 302 and 316, or between different chloride concentrations for one alloy, were equal to the corresponding differences in ET measured at current densities in the range l-5 A cmm2, typical of real pits. This is consistent with the idea that E,i, is the potential above which pits are able to progress from metastable to stable growth, and therefore depends on the kinetics of pit dissolution rather than the breakdown resistance of the passive film. Metastable pitting occurs at potentials well below Epi,. The variation of E,,, with chloride concentration and molybdenum alloying is completely explained by the effects of these variables on kinetics within the pit environment as manifested in the measurement of ET. On an absolute scale, ET (at l-5 A cmm2) falls within the range of Epit values measured for surface finishes from 120 to 1200 grit. 0 1997 Elsevier Science Ltd
Keywords: A. stainless steel, B. potentiostatic,

C. pitting corrosion.

INTRODUCTION The initiation and propagation of pits can be considered separately, with the initiation stage involving breakdown of the passive film and the development of an aggressive local chemistry at the corroding site, whilst propagation involves the continued stable growth of the corroding area. Perhaps the most commonly measured quantity in pitting studies is the pitting potential, Epit, above which stable pitting is discernible for a given alloy and environment.-3 It was at first thought that this was the lowest potential at which pits could initiate, but the later observation of short lived pitting events at lower potentials,7 now commonly referred to as metastable pitting,4 shows that Epit is really the potential above which pits can stably propagate. The propagation of a pit involves sustained, high rate anodic dissolution within an occluded cavity at potentials where the majority of the metal surface is passive. Even at high anodic potentials, local potential and pH conditions mean that cathodic hydrogen evolution does occur within the pit, but less than 5% of the anodic pit current is supplied by cathodic reactions within the pit879and pitting models often consider the pit entirely as an anode. In

Manuscript received 1I October 1996; in amended form 4 November 1996; accepted 29 March 1997. 1771

1772

N. J. Laycock and R. C. Newman

near-neutral chloride solutions7 anodic polarisation of stainless steels shows no significant active loop, but does show pitting processes, so a local chemistry significantly different from the bulk solution is clearly required within an incipient pit; this is coupled with a variation in the electrode potential between the pit bottom and the passive surface. Localized acidification within a pit cavity was first suggested by Hoar, as early as 1937, as a mechanism for the pitting of tin in near neutral solutions. Later, direct measurements of the pH within artificial pit cavities of stainless steels in near neutral bulk solutions, yielded pH values as low as - 0.13. I3 Pickering and Frankenthal4. used a microprobe to measure the potential drop within growing pits on iron and stainless steel and found that potential differences between the pit mouth and the pit bottom could be greater than 1 V. They proposed a model for pit propagation based on the idea that, even at high applied potentials, the IR drop within the growing pit ensured that the potential of the metal at the pit bottom was always in the region of active metal dissolution, but no consideration was given to the necessary development of an aggressive local chemistry within the pit. Galvele16 used a one-dimensional pit nucleus geometry to represent an imperfection in a passive film and assumed that metal dissolution at the pit bottom was followed by hydrolysis reactions leading to acidification. By considering the transport of ionic species in and out of the pit, he was able to calculate values of a critical quantity ix, where i is the pit current density and x is the pit depth, above which the pH at the pit bottom was low enough to maintain the metal in the active state and therefore allow continued pit propagation. Mankowski and Smialowska7 simulated the local pit environment using acidified bulk dissolved 304 and 3 16 SS in 10 N HCl to solutions of FeCl2, whilst Hakkarainen18.19 produce saturated, or near saturated solutions. It was found that solutions greater than about 80% saturated in metal ions were necessary to prevent passivation and allow a direct transition from active dissolution to diffusion controlled dissolution. Using artificial pit (lead in pencil) electrodes, Isaacs and Newman2 and then Gaudet et ~l.,~ obtained similar results showing the existence of steady states for active dissolution on salt free surfaces, but only for solutions more than 60% saturated in dissolution products and above a critical potential. Real pits on stainless steel have often been found to grow, at least initially, with remnants of the undermined passive film providing a partial cover over the pit cavity.7,22.2 Indeed, it has been suggested that in this case, pitting can be considered a special case of crevice corrosion,24 although other authors prefer to think of crevice corrosion as a special form of pitting.25 A further common observation is that many pits, particularly at relatively low potentials, are crystallographic in nature with flat wails and etched interior surfaces,4.24,26 but pits at higher potentials tend to be approximately hemispherical and possess polished interiors.8,27.28 For iron-base metals, Sato2 distinguished between etching pits and polishing (or brightening state)pits. Etching pits formed at less noble potentials were crystallographic pits and Sato believed that a critical pH must be reached in the pit cavity for this type of dissolution to occur; a combination of this low pH and the IR drop in the pit kept the metal in the active state. Polishing pits had bright (polished) internal surfaces, formed at more noble potentials and required the maintenance of a critical aggressive anion concentration within the pit. Sato noted that the critical local chloride concentration for the transition from etching to brightening pits was close to the critical concentration for stable pitting of stainless steel. Later, Sato2s showed that pits initiated above Epit, grew as hemispherical, polishing state pits, but if the potential was then reduced, the pits either

Salt films and pitting potentials

1773

repassivated or propagated as active state (etching) pits. These active state pits often grew as deep, non-hemispherical holes or into crevice like geometries. According to Vetter and Strehblow,2g a thin resistive layer of metal salt was supposed to exist at a corroding pit surface, although the pit solution was not saturated with this salt. This layer was postulated since their calculations suggested that pH changes and ohmic drops could not be significant enough to explain the sharp active/passive transition occurring at the pit edges, although the presence of an overhanging passive film was not considered and could equally explain this observation. Larger pits are often found to be dish-shaped rather than perfectly hemispherical, because without a pit cover, the pit edges have lower associated solution or diffusional resistances and dissolve more rapidly. In contrast to Vetter and Strehblow, Beck and Alkire3 calculated that pit solutions could indeed become saturated with metal salts within 10-g-10-4 seconds of nucleation. The salt layers then precipitated within the pit were presumed to play a significant role in pit stability. Reviewing the literature, Beck3 concluded that precipitation of a non-porous barrier salt layer was essential during the early growth of a pit. Significantly, he noticed that Vermilyea32 had shown that the pitting potential for several metals, including iron, nickel and magnesium, was close to the standard potential of formation for the relevant metal chloride. In common with Vetter and Strehblow,2g Beck thought that crystallographic etch pits could grow with a salt layer present, provided the film was thin enough (approximately 10 nm), but at higher applied potentials, thicker films would form and result in hemispherical, electropolished pits as have frequently been observed. It is generally accepted that the growth of a metustable pit is essentially the same as the early growth of a stable pit.33 However, the lifetime of a metastable pit is cut short by repassivation when the pit fails some critical test of its stability. Boehni and co-workers carried out a great deal of work on the metastable pitting of austenitic stainless stee1.3b36 With a similar idea to that described by Beck,3 they believed in a quasi-thermodynamic salt film formation potential, Esr, below which metastable pitting was impossible. Above &, at specific weak spots, the passive film was believed to be less stable than the salt film, but the precipitation of the salt film was a kinetically hindered process so that early pit growth occurred under mixed ohmic/charge transfer control until the salt precipitated and diffusion control was established. This mode1336 gave an important role to the pit cover in controlling pit stability. According to Vetter and Strehblow,2g a high resistance layer was required between the passive surface and the active pit bottom. Franke136 suggested that this high resistance was initially provided not by a salt layer, but by the porous pit cover with the pores acting as resistors in parallel. It was proposed that only pits which survived long enough to precipitate a salt film became stable, whilst other pits repassivated when their cover ruptured and the ohmic barrier was lost. If a salt film is present when the cover breaks, the film is presumed to thicken and accommodate the extra potential so that pit stability is maintained. Pistorius and Burstein also believed in an important role for the pit cover, but as a diffusion barrier rather than a resistive one. As established earlier by Galvele16 the critical quantity in pit growth is the product, ix, which Pistorius and Burstein termed the pit stability product. For 304 SS, taking a minimum metal ion concentration for stable growth of 3 M and a maximum of 6 M (allowing for supersaturation), equation (1) was obtained showing the possible range of values of iu for stable pit growth. 0.3 A m- < ia -c 0.6 A m- (1)

1774

N. J. Laycock

and R. C. Newman

Incidentally, Frankel et a/.34 calculated a critical pit stability product of 0.4 A m- , and a value of 0.6 A m ~ can be inferred from the work of Williams et ~1.~~ However, Pistorius and Burstein found that all metastable pits, even those which went on to become stable, initially grew with pit stability products less than 0.3 A m-l. They believed this was possible because the diffusion barrier provided by the pit cover during metastable growth enabled the concentrated local chemistry to be maintained. Since metastable pits in their experiments grew with an essentially constant current density, ia increased as the pit grew and if it reached 0.3 A m- then the pit was able to survive the complete loss of its cover and become stable. In contrast to the model of Frankel et a1.3436 Pistorius and Burstein believed that all pits grew under diffusion control, with a salt film on their surfaces, even before the cover was lost. By growing single, real pits on microelectrodes and scanning the potential downwards during growth, they showed that pits at high potentials (700 mV (SCE)), were indeed under diffusion control. This work concentrated on the kinetics of pit growth in order to establish whether active and salt filmed dissolution were both possible in growing pits under different conditions, or not. In addition, the transition from metastable to stable pit growth was studied to determine the critical factors that control pit stability and so gain a more complete understanding of the pitting potential and the factors which control it.

EXPERIMENTAL

METHOD

All electrochemical experiments used a potentiostat made by ACM Research together with a sweep generator made by Thompson Instruments. Data were recorded digitally using a 386 DX computer fitted with a Keithley DAS 8 A/O data acquisition card used as an analog to digital converter in conjunction with Keithley Easyest LX software. In metastable pitting tests, experiments were designed such that currents were below 2 uA and could be measured using a Keithley model 614 electrometer giving low background noise levels. A saturated calomel electrode (SCE) was always used as the reference electrode and all potentials quoted here refer to this scale. An approximately 10 mm length of 1 mm diameter platinum wire was used as the counter electrode. All test solutions were made from analytical grade chemicals and de-ionized water. Artificial pit electrodes were made by mounting narrow steel wires in Araldite Rapid epoxy resin so that one end of the wire was exposed as the electrode surface. AISI 302 and 316 stainless steel (SS) wires of 10, 50 and 500 urn diameters were used, and the electrode surface faced vertically upwards in all experiments. To begin a series of experiments, an electrode was wet abraded with coarse Sic paper and immersed immediately in the test solution. The applied potential was then stepped to + 750 mV (SCE) so that stable pitting occurred; coalescence of these pits led to general dissolution of the surface and the formation of a one-dimensional artificial pit cavity. Experiments were carried out in 1 M, 0.3 M, 0.1 M and 0.01 M NaCl solutions at room temperature (20 i 2C) with the cell open to the air. At regular intervals during the lifetime of a given pit, the potential was decreased at 10 mV s- whilst the current and potential were recorded. During each of these scans the current remained potential independent for some time before showing a small peak and a decrease indicating loss of the salt film present during diffusion controlled dissolution. The kinetics of active dissolution were recorded down to a current value about 80% of the limiting current, then the scan direction was reversed and the salt was precipitated again. Further pit growth was carried out at 250 or 350 mV before repeating the backscan

Salt films and pitting potentials

1775

procedure. The transition potential, ET, between diffusion controlled and activation/ohmic controlled growth was defined for a particular limiting current density, irimas shown in Fig. 1. The pit anode resistance was taken to be the reciprocal of the slope of the current vs potential graph (e.g. Fig. I), measured at ET on the increasing potential part of the scan. The pitting potentials of 302 and 316 SS wires were determined as a function of bulk solution chloride concentration at room temperature. The working electrodes were lengths of 500 pm diameter wire positioned so that a surface area of 0.5 cm2 was immersed in the solution, and a waterline was present on the sample. Wires were longitudinally abraded to either a 120 or 1200 grit finish and rinsed with de-ionized water immediately before they were placed in the cell. The test solution volume was approximately 200 cm3 and the solution was thoroughly de-aerated using nitrogen prior to the samples being introduced into the cell. Once the samples were introduced, the cell was sealed, except for nitrogen inlet and outlet lines, and the flow of nitrogen was reduced to a low level to avoid any effects of flow on the pitting characteristics. Each experiment was performed on a freshly prepared electrode, and each length of wire was discarded after 3 or 4 experiments. Once in the solution, the sample potential was held at - 500 mV and then swept anodically at 1 mV s- . Polarisation was continued until the anodic current exceeded 1 mA cmU2 and the pitting potential, Epit, was defined at a sustained current density of 10 uA cmp2 For each test condition (alloy, chloride concentration, surface finish) three measurements were carried out. Some potentiostatic experiments were also carried out on the metastable pitting behaviour of 302 SS in 1 M NaCl at room temperature. So that low noise data could be obtained, and individual pitting transients could be easily resolved over a wide range of potential, a small electrode surface area was used. For experiments at up to 80 mV, a 2 cm

100

200

300

E (mV vs SCE)
Fig. I. Current density vs potential during potential sweep experiment for a 50 pm diameter, 302 SS artificial pit in 0.1 M NaCI, showing definitions of ia,,, and ET.

1776

l\i. J. Laycock

and R. C. Newman

length of 50 urn diameter wire was immersed as the working electrode, but at higher potentials a 1 cm length of the same wire was used so that separate pitting transients could still be distinguished. Each experiment was carried out with a new piece of wire, freshly longitudinally abraded with 120 grit paper and rinsed with de-ionized water. The test cell for these experiments contained approximately 200 cm3 of solution and was open to the air the potential of the sample was stepped from its throughout. After immersion, instantaneous rest value to a chosen test potential between -60 and - 120 mV. The current was recorded for the next 600 s at 12 Hz. The collected data were then analysed by plotting the digital data in 50 s intervals and counting the metastable pitting transients in each interval. Background noise in these experiments was k 5 nA and so only events > 10 nA peak magnitude could be discerned. A number of transients did not consist of a smooth current rise followed by a sharp fall to the background current level, rather the current rise was interrupted by apparently incomplete repassivation events. Either two (or more) spatially separated events occurred almost simultaneously, or one individual pit suffered partial repassivation(s) during growth. For pit counting purposes, the sharp fall of a repassivation event was easiest to count and all fast current drops > 10 nA were counted as separate events (fast means appearing as vertical lines on the display scale used).

EXPERIMENTAL
A model ojpit dissolution

RESULTS

AND

DISCUSSION

kinetics Figure 2 shows the pit depth as a function of reciprocal limiting current density for 50 urn diameter artificial pits of both 302 and 316 SS grown in the diffusion controlled regime with a salt film present on the surface. The pit depth was calculated by integration of

10

1 I i,i, (Ascm2)
Fig. 2. Pit depth plotted against the reciprocal of the limiting current density for 302 and 316 SS 50 pm diameter artificial pits grown under diffusion control in 1 M NaCI. Dashes represent 95% confidence limits for each line and each line is composed of data from two experiments.

Salt films and pitting potentials

1111

the current vs time curve and application of Faradays Second Law, assuming stoichiometric dissolution of Fe, Cr and Ni with valencies of 2, 3 and 2, respectively. The characteristic time, z, for diffusion out of the pit is approximately L2/D, where L is the pit depth and D is the ionic diffusivity, and for a 100 urn deep pit this would give r = 10 s. At this depth, the metal is dissolving at approximately 0.3 urn s- (Fig. 3) and so the depth would increase by 3% during t. For a 1 mm deep pit, z is 1000 s and the dissolution rate is 0.03 urn s- . which again gives a 3% depth increase during r. This justifies the use of steady state calculations such as Ficks First Law to describe the growth of these pits, as was implied by the good linear fit shown in Fig. 2. It is also interesting that for any given pit depth, the limiting current density was slightly higher for 316 than for 302 SS (Fig. 2). Current oscillations due to periodic passivation and reactivation beneath the salt film3 were more pronounced in the 3 16 pits which consequently spent more time in the supersaturated state and therefore had slightly higher average dissolution rates. The transition potential, ET, between activation/ohmic controlled growth and diffusion controlled growth can be plotted against the log of the limiting current density, iiim,as in Fig. 3 for 302 SS in 1 M NaCl solution. Results are shown for 10, 50 and 500 urn diameter artificial pits. A good linear fit is obtained for both the 10 and 50 urn plots, although closer examination of the 50 urn results reveals that the data are beginning to curve upwards at the higher current densities. The small number of tests with 500 urn diameter wire show a more extreme version of this effect so that a linear fit for the data was not appropriate. A simple model of the kinetics within the artificial pit is proposed. At the point where ET was measured in these tests, the pit was in the active (salt-free) state, but exactly at the limiting current density. The overpotential at this point includes contributions from both activation effects, AE,,,, and solution resistance effects, qrn. If the corrosion potential in the saturated pit environment is EC,,,, then equation (2) applies.

200

&,,,(A cm -)
Fig. 3. The transition potential, ET, as a function of limiting current density for IO,50 and 500 pm diameter artificial pits of 302 SS in 1 M NaCI.

1778

N. J. Laycock

and R. C. Newman

ET = -&,,r + A&c, + QIR

(2)

The activation effect contribution is given by equation (3) where b, and icorr are the anodic Tafel slope and corrosion current density within the pit environment. The total solution resistance, R,, is given by equation (4) where Rint is the solution resistance within of the pit the pit cavity and R,,, is the resistance of the external solution. The composition solution varies with depth, but if the resistivity of this solution can be represented by an average resistivity, p, then a pit of radius r and length L has an internal resistance given by equation (5). The external solution resistance can be calculated by equation (6) which gives the resistance for semi-infinite current flow from a disc to a counter electrode at infinity in a solution of resistivity p.

Ks = Rint + Rext
Riot = pL/nr'

(4) (5)

(6)
For an artificial pit electrode, the active surface area (m2) is constant and the total IR drop at the limiting current, /ii,,,, is therefore given by equation (7). For bulk solutions such as the 0. l-l M NaCl used in these experiments, p is low and measurements were made with the pit depth much greater than the pit radius (L > r). For example, assuming a 4.2 M saturation concentration of metal ions, stoichiometric dissolution with an average metal ion charge of 2.2, and an effective diffusivity39 of 0.86 x 10K5 cm2 s-, then, using Ficks to a pit depth of approximately 80 urn. 1 M First Law, an iii, of 1 A cm- corresponds NaCl has a resistivity of 13.9 52 cm40 (at 25C) and therefore R,,, for a 10 urn diameter pit growing at 1 A cm 2 is 7 kR. Similarly, the concentrated pit solution resistivity can be estimated as 5 R cm, giving an Rint of 50 kR. In this case R,,, is less than 15% of Rint and the total IR drop is approximated by equation (8). /I,, R, = ii,, p L + iii, pm/4 (7)

hm R, = h, PL

(8)

Applying Ficks First Law to the 302 SS data at a depth of 100 urn, iii, can be used to calculate the saturation metal ion concentration as in equation (9) with the bulk concentration of the dissolving ions taken to be zero. This gives a value for Csat, of 3.95 M which is slightly lower than the literature result of 4.2 M.39 It could be that the presumed diffusion coefficien? of 0.86 x lo- cm2 s- IS responsible for this discrepancy, or simply that the 1 M Cl- present in the supporting electrolyte has slightly reduced the solubility of the metal ions. Cqa, = i,,,L/nFD (9)

It has been established that the IR drop during these tests is given by equation (8) which suggests that ZlimRs should be independent of pit depth. Diffusion is governed by Ficks First Law and SO ili, z l/L but R, is directly proportional to L and therefore IlimR, is a constant

Salt films and pitting potentials

1779

for all pit depths. This result was verified experimentally by measurement of the total pit resistance, R,,, at various iiimvalues. Figure 4 shows the results of such an experiment for the conditions where the above assumptions are most likely to be correct, i.e. for the narrowest wire in the most conducting solution. Rtot was measured as the gradient of the Zvs E curve at ET on the increasing potential part of the potential scan (Fig. 1 shows i vs E). At this point, Rtot is made up of the charge transfer resistance, R,, Sz cm2 and the solution resistance, R,Cl as in equation (10). R,, can be defined as dE/di for the anodic line assuming Tafel kinetics, as shown in equations (11)-(12), and SO Rctxl/iiimSIn Fig. 4, the linear plot of R,,, against l/ irim shows that R, must also be inversely proportional to iiim verifying that ZlimR, is a constant. Z-Got = R, + tZWrr2)
E = a + b,logi

(10) (11)

Ret = bJ(2.3)i

(12)

Equation (2) can now be rewritten more explicitly as equation (13) and since ZlimR, is a constant, this equation explains the linear dependence of ET on log ilim as shown in Fig. 3. The displacement of ET to higher potentials for the 50 pm pit compared to the 10 l.trnpit is caused by the relatively increased contribution of Z& to R,as described by equation (7). At the higher current densities (smaller L) this effect is more pronounced and accounts for the slight upwards curve of the data. For 500 pm diameter pits, the contribution of Z& means that Z,imR, is no longer constant and a linear plot is no longer obtained, although equation 700 600

1 I ilim (A-cm*)
Fig. 4. Pit anode resistance, R,,,, as a function of l/iii,,, for a 10 p diameter artificial pit of 302 SS in 1 M NaCl. Measurements made in the salt-free state at ET.

I780

N. J. Laycock

and R. C. Newman

(13) is still valid.


ET =

Ecorr f balOg

(13)

The contribution of the IR drop outside the pit, IiimRext,can be calculated for any given current density if the bulk solution resistivity is assumed to be constant and unaffected by dissolution products leaking from the pit. The resistivities of 1 M, 0.1 M and 0.01 M NaCl are 13.9, 93.7 and 844 !Acm, respectively4 (at 25C). Subtraction of calculated R,,, values from the data of Fig. 3 produces a better linear fit for the 50 and 500 urn pit lines and brings the absolute values closer to those measured for the 10 urn pit (Fig. 5). Equation (13) shows that the slope of ET vslog iiimplots should be equal to b,, the anodic Tafel slope in the pit environment. The calculated values for the gradients of linear plots in Figs 2 and 5 are shown below in Table 1. The IR,,, correction for the 500 urn data is probably an over-correction since the solution near the pit mouth would have been significantly more conductive than the nominal bulk conductivity due to metal ions transported out of the pit. The average value for 6, from the IR,,, corrected data for the two other datasets is 110 mV/decade. Gaudet et ~1.~ reported a value of 54 mV/decade for 304 SS, whilst Newman and Isaacs2 measured 60 mV/decade for an Fe-19Cr-1ONi alloy with a change to 75 mV/decade at potentials below - 300 mV. For 304L SS, the effect of impurities precluded measurement at potentials below - 250 mV, but above this a Tafel slope of 60 mV/decade was recorded. These results suggest that the average value obtained from Table 1 is a little high, although a possible

0.1

IO

&,,,(A cm-)
Fig. 5. The transition potential, ET, as a function of limiting current density for IO,50 and 500 )tm diameter artificial pits of 302 SS in 1 M NaCI, with the contribution of the potential drop outside the pit subtracted from the experimentally measured values.

Salt films and pitting potentials Table 1. Slopes of the linear regression lines for the & vs ilimplots shown in Figs 2 and 5 Pit diameter (pm) Experimental slope (Fig. 3) mV/decade 113 148 ZkX1,corrected slope (Fig. 5) mV/decade 105 115 88

1781

10 50 500

explanation is that these data were measured in an approximately 100% saturated pit solution and the actual dissolution rate of stainless steels has been shown to decrease slightly in the range from 85-100%.2 Equally possible is that the ZR correction used by other authors202 was overestimated, or that at the higher current densities used in this work there is a genuinely increased Tafel slope.
Comparison with real pits

Equation (13) can be applied equally to real pits, so that the results of Fig. 3 should resemble those obtained if the same experiment could be performed with real pits, although current densities would be slightly higher during metastable growth (e.g. l-10 A cme2). In addition, the form of the calculations for R, is different for hemispherical cavities than for linear artificial ones. Slight differences exist between literature expressions for the ohmic resistance of an open pit,8,3',34 but in all cases R,w l/v and according to Pistorius and Burstein8 R, is given by equation (14), where p is the solution resistivity. This equation was derived and experimentally verified for a constant solution resistivity, so for pits in relatively conducting bulk solutions, it is reasonable to use the bulk solution resistivity and ignore the effects of the concentrated pit solution,241 although this should be recognised as a simplification. Similarly, the diffusion limited current from an open, hemispherical pit is given by equation (15) Combination of these two results shows that, as for artificial pits, ZlimRs is a constant for all pit sizes.
R, = p/3r Zlim = nF3rDCSat ZlimRs= nFpD&

(14)
(151 (16)

Obviously, the presence of the pit cover during metastable growth would complicate the calculations. However, the effects of the cover would be to both increase R, and to decrease the effective diffusion coefficient, D. These two effects would to some extent cancel each other out with regard to the value of the transition potential, ET.
Ohmic control or diffusion control?

It is commonly accepted that for 18Cr-8Ni stainless steels, such as 302 and 316, pit propagation can only occur with the local environment > 75% saturated in metal ions i.e. approximately 3 M.8,9,21 Supersaturation of the pit solution, possibly to a factor of 1.5, is required in order to precipitate a salt film, but as a first approximation it is reasonable to

1782

N. J. Laycock

and R. C. Newman

assume that pit growth could occur in the active state with solutions between 75 and 100% saturated, or with a salt film present and the concentration fixed at 100% ofsaturation. In this case, any given pit growing at an applied potential above the ET value for that pit will be salt filmed and under diffusion control. At lower potentials, the pit would grow in the active state. Pit growth in the active state is believed responsible for the etch pits found at low observed polished and potentials by many authors,0,25*26 whilst the commonly hemispherical pits are more likely to have grown with a salt film present on their surface. Sato26.27 made a precise distinction between these two forms of pitting and believed that the polished pits could only form when a critical concentration of chloride ions was maintained within the pit. Other authorsR*29.42 have stressed the importance of salt films in maintaining pit stability. Frankel et u1.34-36 suggested that metastable pit growth is stabilised by the porous pit cover, but that the transition to stable pit growth only occurs for pits which can precipitate a salt film. In this model then, the criterion for stable pitting is the ability to maintain a 100% saturated pit chemistry in an uncovered, hemispherical pit. Using this concept, it is possible to use equation (13) or Fig. 3, to predict the potential above which stable pitting is possible. It is proposed that the defining characteristic of any pit is its current density. Typical current densities measured in the first seconds of growth for pits on stainless steel are in the range from l- IO A cm _ *. and most studies of metastable pits below the pitting potential find values at the lower end of this scale.34,43 Since all pits must be growing within a narrow range of local chemistries (between 75 and 150% saturated), these current densities approximately represent the range of limiting current densities possible in pits. According to equation (13) the il,, value for a given pit determines the value of ET. Figure 6 shows a typical recording of metastable pitting transients from 302 SS at an applied potentials of 0 mV. It was found that the number of observable metastable pits increases with potential as illustrated in Fig. 7, but the qualifying comment that these are

100

200

300

time (s)
Fig. 6. Current decay following a potential step to 0 mV showing 1 M NaCl with a 120 grit finish. metastable pitting on 302 SS in

Salt films and pitting potentials

1783

-100

-50

50

100

150

E (mV vs SCE)
Fig. 7. The number of metastable pits with > 5 nA peak current as a function of potential for 302 SS in 1 M NaCI. Three results are averaged for each data point and the error bars represent the standard deviation. Also shown are selected ET values from the 10 pm data in Fig. 3.

only the observable pits is important. The number of pit nucleation events may not be affected by potentiaJM but the stability of metastable growth increases with potential. The number of events counted and plotted in Fig. 7 is therefore only the number of events which grew to a peak current higher than the detection limit (approximately 5 nA). Current densities calculated for these metastable pits were in the range 0.5-2.5 A cmw2, which corresponds to ET values from 0 to 120 mV. Similar results to Fig. 7 have been obtained by other authors,45,46 and have often been considered to fit a smoothly increasing curve. An alternative suggestion is that a sudden increase in pit stability occurs as most pits become salt filmed, resulting in an apparent increase in the number of observable pits above approximately 40 mV. It is important to note that many metastable pits will grow with salt films present underneath the cover, but a pit can only become stable if it maintains the salt film when the cover ruptures. In fact, since ET is proportional to log (iI& any stable pit must already have been salt filmed in its metastable stage. In practice, it may only be necessary for the pit to maintain a 75% saturated solution when the cover ruptures, although a salt film could actually help the pit to survive the locally violent event. When the cover breaks, increased mass transport from the pit acts to dilute the solution. A salt layer at this time could act as a reservoir of fresh solution; saturation would be maintained until all the salt had dissolved. As long as some salt remains, the solution must remain saturated and the fastest a salt layer could dissolve would be if the pit repassivated beneath it so that virtually no flux of metal ions was produced by metal dissolution. Ignoring the effects of migration and convection, the time for the salt to dissolve, t,, is given by equation (17), where d is the initial salt film thickness, m, is the molar volume of the salt, and iii,,., is the limiting current density for the open pit produced when the cover ruptures.

I784

N. J. Laycock

and R. C. Newman

t, = nFd/(m,i,i,)

(17)

For FeCl,, the density is 1.93 g cm 3.40 giving an m, of 103 cm3 mall and an open pit of 5 pm radius would have an illm of around 7 A cm --. Salt film thicknesses have been found to be less than 1 um4 and for a very thin salt layer of only 10 nm, equation (17) gives a dissolution time of 0.3 s. This time would be shortened by transiently increased convection caused by collapse of the cover, but the delay before dilution of the pit environment may enable the pit to survive until the stability of the pit in the new steady state conditions can be tested. Furthermore, when the cover breaks, the effective R, would decrease but the salt film could thicken to accommodate this extra applied potential in the self regulating manner described by Isaacs4 This view is in good agreement with those of Frankel et uI.,~~ Sato2627 and Beck3 who all suggested that pit growth could occur in the active state, even if only as a transient stage prior to salt precipitation. This is, however, in complete contrast with the opinions of Vetter and Strehblow29 or Pistorius and Burstein.8 These authors proposed models in which all pits are necessarily covered with a salt film. Vetter and Strehblow believed a highly resistive salt layer to be necessary in order to account for the potential difference between active and passive areas at the edge of pits. Pistorius and Burstein measured single pit polarisation curves and found these pits to be under diffusion control, and Burstein44 went on to propose a pit nucleation mechanism in which the salt film existed before the pit itself. However, Pistorius and Burstein only presented single pit polarisation curves for pits on 304 SS grown at high potentials (700 mV). with current densities of approximately 8 A cm-*. Predicted ET values for these pits are around 150 mV and so growth would be expected to be diffusion controlled. Pits on this type of steel have been detected at potentials as low as - 200 mV45 and it is the growth of such pits which would be occurring in the active state as predicted by Fig. 3.

PITTING

POTENTIALS

All pits initially grow metastably with a porous cover to protect the local environment, but this cover eventually ruptures to produce an open, hemispherical pit. In order to propagate from now on, the pit must be salt filmed. ET for a given pit represents the minimum potential at which the salt film can be maintained and therefore is the minimum potential at which this pit can become stable. In general, for a macroscopic stainless steel sample, E,,, will be the lowest ET value from the distribution of pits on the sample at the instant they lose their cover. In some cases, lower pitting potentials will be measured if the pit cover is exceptionally strong, or if the inclusion at which the pit initiated was particularly large, so that the open pit can remain stable without a salt layer due to the size of the cavity alone. Figure 8 shows EPlt values measured on 302 SS as a function of [Cll] and for two different surface finishes, 120 and 1200 grit. Also included on this graph is the variation of ET for an iiim value of 2 A cmW2. (This value was chosen as being typical of the fastest growing metastable pits seen in this work.) These Er values were determined using 50 urn diameter artificial pits to produce ET vs log itim plots (Fig. 9) from which the data in Fig. 8 were then taken. The difference in Epit values between 120 and 1200 grit finishes reflects the nature of the sites available for pit initiation. Pits initiate at specific sites on the surface (probably MnS

Salt

filmsand piningpotentials
slop0 = 40 mV

1785

,
400 -

0 Epr (I= SW A E,
(12OOgrit)

slopa = -93 mV slop = -100 mV

* 2 g 1 E Q, 3 Q

300

200 -

IOO-

O-

-100

---I

0.001

o.ojo

0.100

1.000

10

[cl-l
Fig. 8. Pitting potentials (I&) for 302 SS as a function of log [Cl-] for 120 and 1200 grit surface finishes, compared with ET measured using a 50 pm diameter artificial pit for an ilim value of 2 A cn-.

inclusions) and rougher surfaces generally provide more occluded geometries around these sites. Metastable growth is easier to maintain at these occluded sites and consequently more metastable pits are seen on rougher surfaces. However, these pits grow at relatively low current densities because both ohmic and diffusional resistances are higher for these sites

-100

!
0.1

1 ilim(A cme2)

10

Fig. 9.

ET vs ilimfor 50 pm diameter artificial pits of 302 SS in various chloride concentrations.

1786

N. J. Laycock and R. C. Newman

is why they are favourable). Using a probabilistic argument, it is now easy to see why E,i, is lower for the coarser surface finish; there are simply many more metastable pits43.47 and the chances of one becoming stable are therefore increased. The initiation of pitting occurs on length scales significantly less than the surface roughness in this work, and therefore metastable pitting was stabilised by the extra occlusion provided by the surface geometry. However, in terms of ET and its relationship to stable pitting, the relevant ET value is not for metastable growth, but for the open cavity produced when the cover collapses. High current densities are generally required for these open pits to propagate. For instance, using equation (14) and dividing by the surface area of a hemispherical pit (2nr2), ifi, for an open pit of radius 5 urn would be approximately 7 A cm-2. From Fig. 3, the best estimate of Er is obtained from the 10 urn diameter pit data, 120 mV. Comparison with &ii for the and for an ii,,,, of 7 A cm -*, E-r is approximately 1200 grit surface finish and 1 M [Cl _ ] (Fig. 8) shows that EJ-is indeed a reasonable estimate of Epit for this surface condition. For a lower current density metastable pit to survive, it must either be additionally protected by the surface roughness, or grow to a relatively large size before losing its cover, but Isaacs and Kisselz3 have shown that coarse surface finishes reduce the lifetime of metastable pits by weakening the pit cover. The size of repassivated metastable pits is usually in the region of l-10 urn, which can be compared with grit sizes of approximately 100 urn for a 120 grit finish. Therefore. it is not unreasonable to expect extra protection for pits on coarsely finished surfaces, and the lower Epi, values are consistent with the lower ET values for lower current density pits. In this case, the measured Epi, of approximately 70 mV for I M NaCl at the 120 grit finish is consistent with an iii, of approximately 1 A cme2, which is near the lower end of the current density distribution measured in these tests.
(this

The effect qf chloride concentration


The influence of chloride concentration on the pitting potential is shown in Fig. 8 for 302 SS and chloride concentrations from 0.01-l M. These data are in agreement with literature results648 showing that Er,it follows the relationship of equation (18). Galvele16 accounted for this behaviour by considering the effect of [Cl-] on the ZR drop within the pit. In this model, the value of B should be 59 mV; a value which was found experimentally for pitting of ironI but for stainless steel B has been measured as 90 mV.16*48 Newman et a1.49 confirmed that the IR,,, drop in artificial pits followed a relationship identical to that of equation (18) and found B to be 60 mV for iron and 90 mV for stainless steel. Eplt = A - B log [Cl-] (18)

Figure 8 shows experimentally determined values of B to be 90 and 100 mV for 120 and 1200 grit surface finishes, respectively, and that the value of ET also decreases with log [Cl-] with a slope of 93 mV. This similarity between the dependence of &it and ET on [Cl-] is demonstrated again in Fig. 10 which shows ET vs log iiim plots for various chloride concentrations; as [Cl-] decreases from 1 M to 0.1 M, ET is displaced to higher potentials by approximately 90 mV/decade. As suggested by Galvele,16 this can be explained by an increase in the Iiim R, term in equation (13) due to the decrease in average conductivity of the pit solution as [Cl-] decreases. For the 0.01 M NaCl solution the conductivity of the bulk solution is low and the IR drop outside the pit cavity becomes significant at high current densities, so that the curved plot in Fig. 9 is produced. A correction for this effect (similar to

Salt films and pitting potentials

1787

-100 0.001

0.010

0.100

1.ooo

10

r Cl1 Fig. 10. The pitting potential, &if, measured for 302 and 316 SS in various NaCl solutions with a 120 grit surface finish.

that carried out for the 50 and 500 pm diameter pits in Fig. 5) was not successful, because enrichment of the solution outside the pit mouth by dissolved ions significantly affected the solution conductivity in this initially dilute solution, but to an unknown degree. Alloying with molybdenum Figure 10 shows the well known increase in Epi, caused by the addition of Mo to an alloy. 316 SS is nominally identical to 302 SS except for the addition of 2.5% MO, and its pitting potential is 7&100 mV more noble for all chloride concentrations. Many possible reasons for this effect have been proposed and can be split into two basic categories; those which suggest MO improves the properties of the passive film, and those which consider the effect of MO on the dissolution kinetics within a pit. The model of pit dissolution described by equation (13) obviously takes no account of the passive film and this approach has already been shown to explain the effect of [Cl-] on Epi,. In Fig. 11 the variation of ET with iii,,,for 50 urn diameter artificial pits of both 302 and 3 16 SS is plotted and, at the current densities relevant to pitting, the 3 16 data are displaced to higher potentials by - 100 mV (measured at iii,,,= 2 A cm-*). Given the definition of Epi, stated earlier, this effect on pit propagation accounts completely for the increase in Epit for 316 relative to 302, and is consistent with the work of Newmano95 who showed that MO alloying caused the anodic Tafel line in the pit environment to be shifted to higher potentials. In terms of ET, as given by equation (13), MO acts to increase EC,, and decrease I,,,, in the pit environment by inhibiting the anodic reaction. In addition, the increased slope of the 3 16 data in Fig. 11 suggests that there is also a slight increase in b,. Ezuber43 found that the distribution of metastable pit current densities was shifted to higher potentials for 3 16 in relation to 304. That is, the modal pit current density increased with potential in the same way for both steels, but a given value was reached at a potential about 100 mV more positive for 316. This is also consistent with the idea that MO simply

I788

N. J. Laycock

and R. C. Newman

-100
0.1 1

10

&,.,, (A

cmm2)
pits of 302 and 3 16 SS in

F1g.I I

ET vs i,,,, for 50 pm diameter

artificial

I M NaCl

hinders anodic dissolution within the pit such that higher potentials are required same current density. Therefore, all other things being equal, higher potentials for pits on 3 16 SS to precipitate a salt film and become stable, as shown by argument will apply for all MO additions that do not raise the critical pitting above the ambient temperature. At some MO content, stable pitting at ambient ceases altogether due to passivation within the saturated pit environment.52-55

to reach the are required Fig. 11. This temperature temperature

CONCLUSIONS Artificial pit electrodes of 302 and 316 SS have been used to measure the transition potential, ET, between active and salt covered dissolution, and this potential can be described by the simple equation shown below; ET = EC,,, + b,log
(

2
>

him&

(19)

Metastable pit growth is initially stabilised by the presence of a pit cover which helps to maintain the aggressive local chemistry within the pit. This pit cover will eventually collapse and a pit must survive this event in order to become stable. It is proposed that, in normal circumstances, only pits which can maintain a salt film on their surfaces in the absence of a cover will become stable. In effect, the ET value for pits at the instant they lose their cover defines the pitting potential, EpIt. Increasing solution chloride concentration has been shown to decrease Epi, by between 90 and 100 mV/decade for 302 SS. The value of ET was also found to decrease by 90 mV/ decade with increasing log[Cl-] which is consistent with the suggestion that ET defines E,,,,.

Salt films and pitting potentials

1789

This effect is due to a decrease in the Zt&2, term in equation (19) as reported by Galvele32 and Newman et aLI 316 SS is essentially identical to 302 SS but for the addition of 2.5% MO. The effect of this MO is to raise the Z&i, of 316 SS by 70- 100 mV relative to 302 SS at all chloride concentrations. At the current densities typical of metastable pitting in these steels (1-5 A cmm2) ET was also found to be approximately 100 mV higher for 316 SS than for 302 SS. This is due to inhibition of anodic dissolution in the pit environment by Mo,~ such that EC,,, is raised and i,,,, is lowered (equation (19)). Therefore, the beneficial effect of MO on pitting resistance is completely accounted for by its effect on the anodic kinetics within the pit.
Acknowledgements-Financial support from the EPSRC and from Unilever Research is gratefully acknowledged.

REFERENCES
I. M. Pourbaix, L. Klimzack-Mathieu, Ch. Martens, J. Meunier, Cl. Vanleugenhaghe, L. de Munck, J. Laureys,
L. Nellmans and M. Warzu, Corros. Sci. 3,239 (1963). 2. Z. Szklarska-Smialowska and M. Janik-Czachor, Br. Corros. J. 4, 138 (1969). 3. K.K. Starr, E.D. Verink and M. Pourbaix, Corrosion 32, 47 (1976). 4. L. Stockert, F. Hunkeler and H. Boehni, Corrosion 41, 676 (1985). 5. P.H. Balkwill, C. Westcott and D.E. Williams, Muter. Sci. Forum 44/45, 299 (1989). 6. D.E. Williams, C. Westcott and M. Fleischmann, J. Electrochem. Sot. 132, 1796 (1985). 7. D.E. Williams, C. Westcott and M. Fleischmann, J. Electrochem. Sot. 132, 1804 (1985). 8. P.C. Pistorius and G.T. Burstein, Phil. Trans. R. Sot. Lond. A 341, 531 (1992). 9. R.C. Alkire and K.P. Wong, Corros. Sci. 28, 411 (1988). 10. W. Schwenk, Corrosion 20, 129t (1964). 11. T.P. Hoar, Trans. Faraday. Sot. 33, 1152 (1937). 12. B.E. Wilde and E. Williams, Electrochim. Acta 16, 1971 (1971). 13. T. Suzuki, M. Yanabe and Y. Kitamura, Corrosion 29, 18 (1973). 14. H.W. Pickering and R.P. Frankenthal, J. Electrochem. Sot. 119, 1297 (1972). IS. H.W. Pickering and R.P. Frankenthal, J. Electrochem. Sot. 119, 1304 (1972). 16. J.R. Galvele, J. Electrochem. Sot. 123, 464 (1976). 17. J. Mankowski and Z. Szklarska-Smialowska, Corros. Sci. 15, 493 (1975). 18. T. Hakkarainen, Mater. Sci. Forum 8, 81 (1986). 19. T. Hakkarainen, in Corrosion Chemistry within Pits, Crevices and Cracks, ed. A. Turnbull. HMSO, London, 1987, p. 17. 20. R.C. Newman and H.S. Isaacs, J. Electrochem. Sot. 130, 1621 (1983). 21. G.T. Gaudet, W.T. MO, T.A. Hatton, J.W. Tester, J. Tilly, H.S. Isaacs and R.C. Newman, AZChE. J. 32,949
(1986). 22. W. Schwenk, Corrosion 20, 129t (1964). 23. H.S. Isaacs and G. Kissel, J. Electrochem. Sot. 119, 1628 (1972). 24. I.L. Rosenfeld and I.S. Danilov, Corros. Sci. 7, 129 (1967). 25. L. Stockert and H. Boehni, Mufer. Sci. Forum 44145, 313 (1989). 26. R. Ke and R.C. Alkire, J. Electrochem. Sot. 142, 4056 (1995). 27. N. Sato, in Corrosion and Corrosion Protection, ed. R.P. Frankenthal and F. Mansfeld. The Electrochemical

Society, Pennington NJ, 1981, p. 101. 28. N. Sato, Corros. Sci. 37, 1947 (1995). 29. K.J. Vetter and H.H. Strehblow, in Localized Corrosion, ed. R.W. Staehle et al. NACE, Houston, 1974, p. 240. 30. T.R. Beck and R.C. Alkire, J. Elecfrochem. Sot. 126, 1662 (1979). 31. T.R. Beck, in Advances in Localized Corrosion, ed. H. Isaacs, U. Bertocci, J. Kruger and S. Smialowska. NACE, Houston, 1990, p. 85. 32. D.A. Vermilyea, J. Electrochem. Sot. 118, 529 (1971).

1790 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54.

N. J. Laycock

and R. C. Newman

55.

H.S. Isaacs. Corros. Scl. 29, 313 (1989). G.S. Frankel, L. Stockert. F. Hunkeler and H. Boehni, Corrosion 43, 429 (1987). F. Hunkeler, G.S. Frankel and H. Boehni. Corrosion 43, 189 (1987). G.S. Frankel, in Advances in Localized Corrosion, ed. H. Isaacs. U. Bertocci, J. Kruger and S. Smialowska. NACE, Houston, 1990, p. 137. D.E. Williams, J. Stewart and P.H. Balkwill, in Critical Fucrors in Localized Corrosion, ed. G.S. Frankel and R.C. Newman. The Electrochemical Society, Pennington NJ. 1992, p. 36. N.J. Laycock and R.C. Newman, Mater. Sci. Forum 192-194, 649 (1995). H.C. Kuo and D. Landolt, Electrochim. Acta 20, 393 (1975). D.R. Lide (Ed.), CRC Handbook of Chemistry and Physics. CRC Press, Boca Raton FL, 1974. U. Steinsmo and H.S. Isaacs, J. Electrochem. SOL-.140, 643 (1993). U.F. Franck, in Proceedings qflst Internalional C0ngres.r on Metallic Corrosion. Butterworth, London, 1962, p. 120. H. Ezuber and R.C. Newman, in Critul Facrors in Localized Corrosion, ed. G.S. Frankel and R.C. Newman. The Electrochemical Society, Pennington NJ, 1992. p. 1.20. G.T. Burstein and S.P.Mattin, in CriGcul Factors in Localized Corrosion II, ed. P.M. Natishan, R.G. Kelly, G.S. Frankel and R.C. Newman. The Electrochemical Society. Pennington NJ, 1995, p. I. J. Stewart and D.E. Williams, in Advances in Localized Corrosion, ed H.S. Isaacs, U. Bertocci, J. Kruger and S. Smialowska. NACE, Houston, 1991, p. 131. H.S. Isaacs, J. Electrochem. Sot. 120, 1456 (1973). G.T. Burstein and P.C. Pistorius, Corrosion 51, 380 (1995). H.P. Leckie and H.H. Uhlig, J. Electrochem. Sac. 113, 1262 (1966). R.C. Newman, M.A.A. Ajjawi. H. Ezuber and S. Turgoose. Corros. Sci. 28, 471 (1988). R.C. Newman and A.J. Betts, in Advances in Localized Corrosion, ed. H.S. Isaacs, U. Bertocci, J. Kruger and S. Smialowska. NACE. Houston, 1991, p. 271. R.C. Newman. Corros. Sci. 25. 341 (1985). R.J. Brigham and E.W. Tozer, Corrosion 29, 33 (1973). R. Qvarfort. Corros. Sci. 29, 987 (1989). N.J. Laycock, M.H. Moayed and R.C. Newman, in Crifical Factors in Localized Corrosion II, ed. R.G. Kelly, P. Natishan. G.S. Frankel and R.C. Newman. The Electrochemical Society, Pennington NJ, USA, 1995, p. 68. N.J. Laycock, M.H. Moayed, and R.C. Newman, J. Nectrochem. Sot., in press.

Anda mungkin juga menyukai