Effect of grain size on pitting corrosion of 304L austenitic stainless steel

Corrosion Science 94 (2015) 368–376

Contents lists available at ScienceDirect

Corrosion Science

journal homepage: www.elsevier .com/locate /corsc i

Effect of grain size on pitting corrosion of 304L austenitic stainless steel

http://dx.doi.org/10.1016/j.corsci.2015.02.0240010-938X/� 2015 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel./fax: +98 51 38763305.E-mail address: [email protected] (M.H. Moayed).

A. Abbasi Aghuy, M. Zakeri, M.H. Moayed ⇑, M. MazinaniMetallurgical and Materials Engineering Department, Faculty of Engineering, Ferdowsi University of Mashhad, Mashhad 91775-1111, Iran

a r t i c l e i n f o a b s t r a c t

Article history:Received 18 July 2014Accepted 13 February 2015Available online 25 February 2015

Keywords:A. Stainless steelB. PolarisationC. Pitting corrosion

The influence of grain size on pitting corrosion of 304L stainless steel in 3.5 wt.% NaCl solution was inves-tigated by electrochemical methods employing the statistical approach. The potentiodynamic resultsrevealed that grain refinement has no effect on the pitting potential. To understand the reason of ineffec-tiveness of grain size on pitting potential of 304L stainless steel, potentiostatic tests were conducted.Evaluation of current transients observed in potentiostatic tests revealed that the frequency of meta-stable pit initiation decreases with grain refinement. However, grain size reduction increases the kineticsof metastable pits dissolution and the probability of transition from metastable pitting to stability.

� 2015 Elsevier Ltd. All rights reserved.

1. Introduction

It is known that pitting corrosion is characterised by two conse-quent stages comprising pit initiation and stable pit growth [1].Since pit initiation and transition to stability are considered to bethe key factors in pitting corrosion phenomenon, evaluation ofmetastable pitting gives valuable information about pittingcorrosion [2,3]. Investigations of metastable pitting also allow a sta-tistical assessment of corrosion data necessary to study stochasticmodels of pitting [3]. It is known that the environment, the chemi-cal composition, heat treatment, and the microstructure affect thepitting potential [4]. Grain refinement may change the electro-chemical behaviour of metal as a consequence of variation in grainboundary density. The effect of grain size on corrosion properties ofseveral materials has been reviewed by Ralston and Birbilis [5].They suggested that the impact of grain refinement on corrosionresistance is dependent on the ability of surface to be passivated.For the active conditions, a reduction in the alloy grain size dete-riorates the corrosion resistance whereas in an environment atwhich passivity could be established, grain refinement causes thecorrosion resistance to improve. In conditions at which metalshows the active–passive behaviour, grain refinement enhancesthe uniform corrosion rate and lowers the localised corrosion rate[5].

There is not a general agreement in the literature as to the effectof grain size on corrosion resistance of ferrous alloys. Grain sizevariation could be achieved using a number of different processingincluding rolling, extrusion, and electrodeposition. However, onthe basis that thermomechanical processing eventuates some

microstructural changes leading to affect the corrosion resistance[6], recrystallisation annealing was used to eliminate the effect ofgrain size on other microstructural alternations. It has been report-ed that microstructural defects namely dislocations, twin bound-aries, and stacking-faults in 304L austenitic stainless steel wouldbe entirely eliminated after recrystallisation annealing at tem-peratures above 900 �C for 1 h [7].

In author’s knowledge, only a few fundamental studies havebeen available on the effect of grain size on pitting corrosion resis-tance of austenitic stainless steels. Pit occurrence is often consid-ered to be a random phenomenon [8] and probabilisticmeasurement of pitting characteristics is necessary in evolutionof pitting susceptibility [3]. In the present work, the effect of grainsize on pitting corrosion of 304L stainless steel was investigated byevaluation of pitting probability using a series of potentiodynamicexperiments. Statistical measurement of the characteristics ofmetastable pits was performed by potentiostatic method, in orderto provide valuable information about the initiation and the tran-sition from metastable towards stable pits.

2. Experimental procedure

The investigated material was AISI 304L stainless steel platewith the composition shown in Table 1. Samples were initiallysolution annealed at 1050 �C for 1 h followed by water quenching.Subsequently, all samples were cold rolled to 70% of thickness. Tostudy the influence of grain size on pitting corrosion resistance,recrystallisation annealing was performed at 1050 �C for 3, 15and 30 min. After submission in recrystallisation annealing tem-perature, all samples were water quenched. Flat samples with0.5 cm2 exposed surface area were prepared for electrochemicalevaluations. In order to avoid crevice corrosion in metal/epoxy

Table 1Chemical composition of experimented 304L SS (wt.%).

Element Fe C Si Mn P S Cr Ni

Wt.% 71.8 0.03 0.31 1.26 0.004 0.009 17.75 8.33

Mo Al Cu Co Ti Nb V WWt.% 0.06 0.005 0.11 0.05 0.008 0.02 0.075 <0.007

Fig. 1. Sample optical micrographs of 304L SS after recrystallisation annealing at1050 �C for: (a) 3 min, (b) 15 min, and (c) 30 min.

A. Abbasi Aghuy et al. / Corrosion Science 94 (2015) 368–376 369

interface during electrochemical tests, samples were first prepassi-vated at anodic potential of 850 mV (SCE) for 30 min in 0.1 MNa2SO4 solution and then mounted in an epoxy resin [9,10]. Toreveal the austenitic grain boundaries, polished specimens wereelectro-etched in 60% HNO3 solution at 1.5 V for 1 min. To studythe correlation between the grain size and pitting corrosionresistance, quantitative measurements were performed usingMicrosoft Image Processing Software (MIP™).

Electrochemical measurements were carried out in 3.5 wt.%NaCl solutions using a Gill AC automated potentiostat (ACM Instru-ments). The solutions were prepared with deionised water andanalytical grade chemicals. The electrochemical cell was comprisedof three electrodes. A platinum foil and a saturated calomel elec-trode (SCE) were used as the counter and the reference electrodes,respectively. All electrochemical evaluations were performed atambient temperatures and all potentials quoted in this work referto SCE. Prior to each test, electrodes were prepared by wet grindingup to 1200 grit SiC paper and then washed with deionised waterand next dried with warm air. To yield a steady-state condition,the open circuit potential was recorded for 30 min.

Potentiodynamic polarisation measurements were conductedby sweeping the potential from 50 mV below the rest potential,at a given scan rate (30 mV min�1) until the current densityexceeded 300 lA cm�2 and continued to increase. The potentialat which current density exceeded 100 lA cm�2 and then progres-sively increased, was considered as the pitting potential (Epit). Thesame specimen was used for all 15 experiments and each potentio-dynamic polarisation test was repeated for 15 times under identi-cal condition to provide a statistical basis for the results.

Samples were potentiostatically polarised at 0, 50, and 100 mV(SCE) for 600 s and the current responses were recorded at fre-quency of 50 Hz. Each potentiostatic test was repeated for 3 timeson the same specimen and under the identical conditions. The val-ue of 15 nA for current peak value was considered as the thresholdfor analyses of metastable events. In the cases that overlappingwas observed, each current transient at which the current dropsto a value lower than the half of the peak current size was consid-ered as an individual event.

For scanning electron microscopy, grinding was performedusing silicon carbide papers to 1200 grit size followed by an addi-tional grinding stage with 0.05 lm alumina slurry to create mirrorface and then low scan rate potentiodynamic polarisation (at thescan rate of 3 mV min�1) was performed. Samples’ surfaces wereelectro-etched in 60% HNO3 and then subjected to an ultrasonicbath to remove the cover formed on the pit surface. The formedpits were examined by Scanning Electron Microscopy (SEM) andEnergy Dispersive X-ray Spectroscopy (EDS) techniques.

3. Experimental results

3.1. Microstructural observations

Example microstructures of alloy annealed for various durationtimes are displayed in Fig. 1 showing a microstructure consisting of100% austenite. The growth of recrystallised austenite grains withannealing time is evident from the optical micrographs. The graindiameters of samples recrystallisation annealed at various times

were obtained from quantitative analysis. Fig. 2 shows the averagegrain diameters versus recrystallisation annealing time. Five mea-surements were performed, and the error bars represents 95% con-fidence limits. It is shown that increasing the recrystallisation timefrom 3 min to 30 min increases the average of austenite grain sizefrom 5 lm to 28 lm.

3.2. Analysis of the location of metastable pits

The morphology of metastable pits after the electrochemicaltest were studied by Scanning Electron Microscopy technique. AnSEM image of metastable pits in sample with 28 m grain size isshown in Fig. 3 for instance. Clearly, pits are generated in bothgrain interiors and grain boundaries. The pit nucleation sites wereanalysed by SEM microscope equipped by EDS. The energy disper-sive spectroscopy revealed the composition (wt.%) as follows: Mn

Fig. 2. Grain size versus recrystallisation annealing time for alloy 304L SS. Fig. 4. Scanning Electron Microscopy (SEM) of a sulphide inclusion as a pitinitiation site. The EDS analysis revealed the composition of the inclusion asfollows: Mn (27.3 wt.%), S (15.1 wt.%), Cr (11.5 wt.%), Ni (4.4 wt.%), O (1.2 wt.%) andFe (bal.).

370 A. Abbasi Aghuy et al. / Corrosion Science 94 (2015) 368–376

(27.3), S (15.1), Cr (11.5), Ni (4.4), O (1.2) and Fe (bal.). High con-tent of the manganese and sulphur indicate that pits nucleate onthe MnS inclusions sites (Fig. 4). The dissolution of the sulphideinclusions has preferably taken place close to the metal matrix[9,11,12]. As expected, Fig. 5 shows the distribution of inclusionswas not changed by recrystallisation annealing time.

3.3. Electrochemical evaluations

3.3.1. Pitting potential probabilityTypical potentiodynamic curves of 304L SS with various grain

sizes are illustrated in Fig. 6. The abrupt increase in current densi-ty, which is observed following the metastable pitting activity inthe passivity region of potentiodynamic tests, is an indication ofstable pitting corrosion in all specimens. As shown in Fig. 6, therewas no distinct difference in pitting potential and passivity currentof various samples.

Because of the probabilistic phenomenon of pitting potential[3], a statistic attitude was sought to investigate the effect of gainsize on pitting potential of 316L SS. The cumulative probability ofpitting potential distribution of 304 stainless steel with differentgrain sizes obtained from potentiodynamic polarisation conductedin 3.5 wt.% NaCl is depicted in Fig. 7. Each data point represents thepitting potential of an individual test on a freshly prepared sample.

Fig. 3. Morphology and location of generated metastable pits on alloy with averagegrain size of 28 lm after the low scan rate potentiodynamic polarisation test. Scanrate was 3 mV min�1. Samples surfaces were electrochemically etched with 60%HNO3 and then subjected to an ultrasonic bath to remove the cover formed on thepit surface.

The probability function (P(E)) is according to Eq. (1), where n isthe number of sample and N is the total number of samples tested[2,13]. (In this study, N was equal to 15).

PðEÞ ¼ nN þ 1

ð1Þ

Comparison of the median of the probability distribution, i.e.equal to P(E) = 0.5, is a way to compare the pitting distribution[14,15]. There was only a negligible change (ca. 20 mV) in themedian value of Epit of samples with various grain sizes. The medi-an of pitting potentials was between 270 mV and 290 mV (SCE).Besides, no distinct trend of Epit with recrystallisation time wasobserved. The maximum and minimum values of pitting corrosion(Epit) for samples with finer grain size structure were between themaximum and minimum values of Epit which were obtained forother samples. Therefore, we can conclude that grain size variationhas no important effect on the pitting potential of 304L SS.

3.3.2. Metastable pittingThe early development of stable pits is identical to that of meta-

stable pits [16], and the probability of stable pitting is directlyrelated to the metastable pitting characteristics [17]. From meta-stable pitting studies, many aspects of stable pitting can be under-stood. In an effort to provide fundamental information about theinfluence of grain size on the metastable pitting initiation and onthe transition from metastability to stability, potentiostatic testswere conducted at three constant potentials.

Typical current versus time curves obtained from the potentio-static polarisation tests of 304L SS with various grain sizes areshown in Fig. 8. Experiments were repeated there times and meta-stable pitting parameters were drawn to be evaluated.

Typical current transients are shown in Fig. 9. The observed cur-rent instabilities are associated with metastable events. The shapeof current transients reflects the initiation, growth and repassiva-tion of metastable pits [18].

3.3.2.1. Metastable pit initiation. Metastable pit frequencies, i.e. thenumber of metastable pit events per surface area divided bythe time period which they are counted (100 s) obtained fromthe potentiostatic polarisation of 304L stainless steel samples withdifferent grain sizes are plotted versus time at various potentials(see Fig. 10). Generally, an increase in the grain size causes adecrease in the frequency of metastable pit events in each given

Fig. 5. SEM micrographs of sulphide inclusions, distribution of MnS inclusions in (a)cold worked sample, (b) sample recrystallised at 1050 �C for 3 min, and (c) samplerecrystallised at 1050 �C for 30 min.

Fig. 6. Potentiodynamic polarisation curves of type 304L SS with various grain sizesin 3.5 wt.% sodium chloride solution. Scan rate was 30 mV min�1.

A. Abbasi Aghuy et al. / Corrosion Science 94 (2015) 368–376 371

potential. There are some deviations in the results of the test per-formed at 50 mV (SCE). However, the overall behaviour should beconsidered in the statistical cases. Besides, the nonconformity isnot observed in the initial time of the test which the major partsof metastable pitting activities are observed. The major parts ofmetastable pitting activities are observed in the initial times oftests, which are mostly decreased with test-duration time [19].As can be seen, the frequency in all samples decreased expeditious-ly with time which is believed to be associated with decrease of thetotal number of surface sites available for metastable pitting [19].In the first 300 s which the higher percentage of events were

observed, an increase in the metastable pit frequency with grainsize is evident.

By comparing the metastable pit initiation frequency at poten-tials of 0, 50 and 100 mV (SCE), which are shown in Fig. 10a–c, it isevident that in all samples, the metastable pit frequency generallyincreases with applied potential.

3.3.2.2. Current transient characteristics.3.3.2.2.1. Metastable life time measurement. The total time betweeninitiation and repassivation was calculated for each individualmetastable pit. Cumulative distribution of metastable pit life timeat various potentials for samples with different grain sizes are pre-sented in Fig. 11. Comparison of median life of metastable pitstransients shows that there is a general decrease in life time withgrain coarseness.3.3.2.2.2. Metastable peak current measurement. Peak current ofmetastable pits (Ipeak) were determined by subtracting the back-ground current (Ibase) from the maximum current of each indi-vidual pit. Fig. 12a shows the cumulative distribution of peakcurrent of metastable transients of alloy with various grain sizesat 0 mV (SCE). An overall decrease in the median value of peak cur-rent with grain coarsening are observed. Alloy with average grainsize of 5, 11 and 28 lm has the median value of peak current equalto 45.6, 45.6, and 75.5 nA. The values of cumulative distribution ofIpeak obtained at 50 mV (SCE) and 100 mV (SCE) are presented inFig. 12b and c. Similar to what observed at 0 mV, the dissolutionrate of the metastable pits generated at these two potentialsincreases with grain refinement. Furthermore, comparison ofresults obtained at different potentials, shows that increasing thepotential generally intensify the identified peak current of meta-stable pits. For instance, in alloy with average grain size of28 lm, increase in potential from 0 to 50 mV (SCE) causes themedian peak current to increase from 45.6 nA to 66 nA. Whenpotential further increased to 100 mV (SCE), the median value ofIpeak increased to 99 nA.3.3.2.2.3. Metastable pit radius measurement. Following Pistoriusand Burstein [1], it is assumed that metastable pits are hemi-spherical. The metastable pit radius was determined via Faraday’ssecond law using the amount of charge passed during the meta-stable pit transient (the time between the initial increase of current(ti) and the time that metastable peak current is reached (tf))according to Eq. (2):

rpit ¼3Z

2pnFq

� �Z tf

ti

Ipeak � Iorg� �

dt

" #13

ð2Þ

Fig. 7. Probability distribution of pitting potential obtained from potentiodynamicpotential measurements for 304L stainless steel recrystallised for various time, in3.5 wt.% sodium chloride solution.

372 A. Abbasi Aghuy et al. / Corrosion Science 94 (2015) 368–376

where Z (mean molecular weight) is 55.2 mg mol�1, q (mean alloydensity) is 7.9 mg cm�3, and F is Faraday’s constant. Forexperimented 304L alloy it was assumed that the alloying elementsare oxidised to Fe2+, Cr3+, and Ni2+ during dissolution. Thus, theaverage oxidation state of cations (n) was considered to be 2.19.Calculated radius of metastable pits for 304L SS with various grainsizes plotted as the cumulative distribution using the probabilityfunction (Eq. (1)) are shown in Fig. 13. Comparing the pit radius dis-tributions obtained in the same potentials, it can be seen that radiusof metastable pits initiated in samples with finer grained structuresare higher. The median value of metastable pits radius at 0 mV(SCE) was 1 lm in samples with 5 lm grain size and drops to0.7 lm in alloy with both 11 and 28 lm grain size. Similarly, theresults obtained at higher potentials show that the median of meta-stable pit radius increases with grain refinement. Comparison ofplots obtained at various potentials indicates that an increase inapplied potential increases the radius of metastable pits.

3.3.2.2.4. Estimation of pit stability product. The pit stability product(i � a), is a diffusion-based criterion proposed by Pistorius and Bur-stein [1] which is the product of pit dissolution current density (i)and the pit radius (a). A metastable pit which survived from nucle-ation stage grows unstably in diffusion-controlled regime. In thisstage, an effective barrier to diffusion is provided by lacy coverover the pit mouth. If the cover rupture promotes the anolyte dilu-tion below the critical value, repassivation occurs and the growthwould be stopped. Above the critical condition, i.e. pit stability pro-duct, maintenance of an effective barrier to diffusion is provided bythe depth of pit. In this work, the interior surface area of meta-stable pits were determined assuming hemispherical geometry,and pit stability product at different potentials was determinedfor metastable pits generated in alloy with various grain sizes.Shown in Fig. 14a–c, which represent the median of i � a values, itis evident that fine grained samples have higher pit stability prod-uct. Furthermore, comparing i � a values obtained at various poten-tials reveals that not only the median value of pit stability product,but also the minimum value of i � a increases with appliedpotential.

Fig. 8. Typical current–time curves obtained from potentiostatic polarisation of304L SS with various grain sizes conducted in 3.5 wt.% sodium chloride solution at(a) 0 mV (SCE), (b) 50 mV (SCE), and (c) 100 mV (SCE).

4. Discussion

4.1. Microstructural observations

It has been reported that heat treatment at elevatedtemperatures leads to metal recovery followed by recrystallisation.

Newly nucleated grains grow and replace the high-dislocation-density cold-worked grains and the stored energy in the cold-worked metal is released [9]. At extended annealing time, somegrains grow at expense of others. Therefore, the fine grainedrecrystallised structure is replaced by coarser grained material. Itwas apparent in the microstructure of experimented alloy thatprolonged annealing at 1050 �C leads to coarser grain structurein single phase 304L SS. It is reported that recrystallisation anneal-ing at temperatures above 900 �C for 1 h completely removes themicrostructural defects in 304L austenitic stainless steel [7].

Fig. 9. Typical shape of a current transient obtained from potentiostatic polarisa-tion of 304L SS in 3.5 wt.% sodium chloride solution at 0 mV (SCE).

Fig. 10. Metastable pit frequency as a function of time in 3.5 wt.% sodium chloridesolution for 304L SS with various grain sizes at potential of: (a) 0 mV (SCE), (b)50 mV (SCE), and (c) 100 mV (SCE).

A. Abbasi Aghuy et al. / Corrosion Science 94 (2015) 368–376 373

4.2. Electrochemical evaluations

Evaluation of pitting probability of 304L SS showed that pittingpotential is not affected by grain refinement. This is not consistentwith the observation presented by Schino and Kenny [20], whoshowed that decreasing the grain size improves pitting potentialresistance of 304 SS.

As will be discussed below, the statistical analysis of currenttransients gives key information to study the pitting corrosionresistance based on the stochastic pitting models.

4.2.1. Metastable pit initiationIt is clear from the presented data that grain refinement

decreases the number of metastable pits that can be generated.Improved passive film with grain refinement has been reportedin large range of metals [21–24]. Relatively lower frequency ofmetastable pit formation could be explained based on the abilityof grain boundaries to form a more strong passive film. It has beenreported that in the environments that alloy is in passive state,grain refinement leads to more stable passive film [5]. Passive filmin Fe alloys is termed as being more electrochemically stable with ahigher diffusion rate than coarser grained microstructures [21–23].The enhanced diffusion of Cr into the passive film causes a chromi-um rich protective film to build up. Finer grain 304L has a highdensity of grain boundaries for forming passive layers having morechromium. In other words, since fine grained structures have high-er grain boundaries density, they have more fraction of surfacearea which the passive film formed over is more stable. Besides,the distribution of sulphide inclusions as susceptible sites of pitnucleation [25] does not change with grain refinement, thus it isprobable that in finer grained structure alloy, more nucleation siteswould be located at grain boundaries which are capable to formmore stable passive film. Hence, the frequency of metastable pitinitiation is reduced by grain refinement.

Upward trend of metastable initiation frequency with potentialhas been reported by other researchers [26–28]. Increasing the pitinitiation frequency with increase in potential could be explainedbased on some major influences of potential. The current densityin pre-existence flaws increases with applied potential and conse-quently, active corrosion is initiated at the bottom of the flaws [9].Pistorius and Burstein [29] have suggested that more open sites,which allowed faster diffusion rate, are only susceptible to ini-tiation at higher potentials because the dilution of pit electrolytecould be prevented by higher active dissolution rate. Furthermore,by increasing the potential, surface is becoming more positive andattraction of negative ions to the surface would be facilitated. It

seems that the majority of metastable pits are growing underactivation-ohmic controlled and their response to increase inoverpotential is reflected by increasing in pit current.

4.2.2. Current transient characteristicsPeak current and life time are two important features in the

growth of metastable pits [19]. An increase in the value of theseparameters by grain refinement through recrystallisation impliesthe changes in the kinetic of metastable pit growth [15,28]. Meta-stable pits developed in finer grained alloy have higher peak cur-rent as well as larger life time and subsequently have increasedanodic current during alloy dissolution inside the pit cavity. A

Fig. 11. Cumulative distribution of metastable pit life time for 304L SS with variousgrain sizes at potential of: (a) 0 mV (SCE), (b) 50 mV (SCE), and (c) 100 mV (SCE).

Fig. 12. Cumulative distribution of metastable pit peak current for 304L SS withvarious grain sizes at potential of: (a) 0 mV (SCE), (b) 50 mV (SCE), and (c) 100 mV(SCE).

374 A. Abbasi Aghuy et al. / Corrosion Science 94 (2015) 368–376

direct consequent of the observed increase in anodic charge is theenlargement of pit size which is confirmed by the comparison ofcalculated pit radii. Based on the fact that metastable pit currentis a measure of its growth rate and likewise the dissolution rateof alloy, higher current peak and higher life time of finer grainedalloy demonstrate the higher dissolution rate in this recrystallisa-tion annealing condition. Grain refinement increases the densityof grain boundaries which are commonly etched more rapidly thanthe rest of the grain due to the higher reactivity of the disarrayedmetal [15]. The increased reactivity of grain boundaries could beused to support the notion that grain refinement leads to a higherpeak current probability. A grain boundary, which is a planardefect in a crystal and has a higher energy than the surrounding

crystal [9], corrodes at higher rate. In the environments which elic-it an active passive kind of response, grain refinement wouldenhances the rate of uniform corrosion. Although the grain bound-aries are unaffected in the environments that alloy demonstratespassive behaviour, the preferential dissolution of grain boundariesin active conditions has been proposed [30]. If the passivity break-down occurs at the grain boundaries, the dissolution rate of alloy inacidified solution of nucleated metastable pit would be moreenhanced compared to the cases that the metastable pit are locatedat the grains interior. Thus, the peak current of events generated inthe alloy with finer grain size would be increased.

Fig. 13. Cumulative distribution of metastable pit radius for 304L SS with variousgrain sizes at potential of: (a) 0 mV (SCE), (b) 50 mV (SCE), and (c) 100 mV (SCE).

Fig. 14. Cumulative distribution of metastable pit stability product for 304L SS withvarious grain sizes at potential of: (a) 0 mV (SCE), (b) 50 mV (SCE), and (c) 100 mV(SCE).

A. Abbasi Aghuy et al. / Corrosion Science 94 (2015) 368–376 375

Comparing the peak currents of all specimens demonstratedthat the peak current size is a function of potential. This is consis-tent with the observation of Pistorius and Burstein [1] who showedthat current of metastable pits developed at more noble potentialis higher in compared to those initiate at lower potentials.

4.2.3. Metastable pitting transition to stabilityIt was observed that the values of pit stability product comput-

ed for alloy in different grain sizes are below the critical valuedetermined by Pistorius and Burstein [1]. Surviving a metastablepit is more probable for metastable pits generated in finer grainedstructures, since the possibility of exceeding the pit stability pro-duct from the critical value is higher in alloy with lower grain size.

Williams et al. [31] have argued that the pitting corrosionresistance is directly dependent on the rate of metastable pitsformation. Generated metastable pits which are able to attain thecritical pit stability product would control the possibility ofdeveloping stable pits [1,18]. Likelihood of developing damagingmetastable pits declines by reduction of the number of metastablepits [19]. The present data confirms that grain refinement general-ly decreases the frequency of metastable pit events. Therefore, it isclear that lowering the grain size would lead to improve the corro-sion resistance of 304L SS via lowering the rate of metastable pitsformation. On the other hand, it was revealed in this study that the

376 A. Abbasi Aghuy et al. / Corrosion Science 94 (2015) 368–376

transition from metastability to stability is made easier in finergrained 304L alloy. Overall, the results indicate that the resultantof these two probabilities would lead the pitting potential of304L stainless steel to be constant with altering the grain size.

5. Conclusions

In this study, the effect of grain size on pitting corrosion of 304Laustenitic stainless steel was studied by conducting electro-chemical tests. From the experiments, the following conclusionsmay be drawn:

1. Analysis of the morphology of metastable pits revealed that pitsare generated in both grain interiors and grain boundaries. Itwas clarified that pits nucleate on the inclusion sites. Distribu-tion of inclusions was not different in cold worked specimenand specimens recrystallised for various times.

2. Potentiodynamic measurements conducted in 3.5 wt.% sodiumchloride solution revealed that grain size variation has no con-siderable effect on pitting potential of 304L austenitic stainlesssteel.

3. Obtained results from potentiostatic tests revealed that grainrefinement decreases metastable pit frequency. This could beexplained based on the improved passivity of film on grainboundaries which are the credible sites of pit generation.

4. Metastable pits generated in finer grained structure alloy hadincreased dissolution kinetics of metastable pits. Higher peakcurrent, life time, and larger pit radius result in the increasedprobability of metastable pits transition to stability due to theincrease in the pit stability product.

5. Based on the stochastic model of formation of stable pits frommetastable ones, combination of the decreased number ofmetastable pits and the increased probability of formation ofstable pits from metastable pits would result in the pittingpotential to be unaffected by grain refinement.

Acknowledgments

This work was performed under the sponsorship of the Fer-dowsi University of Mashhad – Iran. Nahamin Pardazan Asia CO.(NPA) is gratefully acknowledged for valuable help in image analy-sis quantification. The authors also would like to appreciate MsMasoumeh Naghizadeh for her precious helps during this paperpreparation.

References

[1] P.C. Pistorius, G.T. Burstein, Metastable pitting corrosion of stainless steel andthe transition to stability, Philos. Trans. R. Soc. Lond., A 341 (1992) 531–559.

[2] A.H. Sherry, T.J. Marrow, 1.04 – mechanical properties and fracture ofmaterials, in: B. Cottis, M. Graham, R. Lindsay, S. Lyon, T. Richardson, D.Scantlebury, H. Stott (Eds.), Shreir’s Corrosion, Elsevier, Oxford, 2010, pp. 77–88.

[3] R.P.M. Procter, 1.03 – outline of structural metallurgy relevant to corrosion, in:B. Cottis, M. Graham, R. Lindsay, S. Lyon, T. Richardson, D. Scantlebury, H.Stott (Eds.), Shreir’s Corrosion, Elsevier, Oxford, 2010, pp. 52–76.

[4] P. Marcus, Corrosion Mechanisms in Theory and Practice, third ed., CRC Press,2011.

[5] K.D. Ralston, N. Birbilis, Effect of grain size on corrosion: a review, Corrosion 66(2010) (075005-075005-075013).

[6] C. op’t Hoog, N. Birbilis, Y. Estrin, Corrosion of pure Mg as a function of grainsize and processing route, Adv. Eng. Mater. 10 (2008) 579–582.

[7] I. Shakhova, V. Dudko, A. Belyakov, K. Tsuzaki, R. Kaibyshev, Effect of largestrain cold rolling and subsequent annealing on microstructure andmechanical properties of an austenitic stainless steel, Mater. Sci. Eng. A 545(2012) 176–186.

[8] T. Shibata, T. Takeyama, Stochastic theory of pitting corrosion, Corrosion 33(1977) 243–251.

[9] E.O. Nunez Moran, MSc Thesis, Evaluation of the Localized CorrosionResistance of 21Cr Stainless Steels, The Ohio State University, 2010.

[10] E.O. Nunez Moran, Evaluation of the Localized Corrosion Resistance of 21CrStainless Steels, The Ohio State University, 2010.

[11] D.E. Williams, M.R. Kilburn, J. Cliff, G.I.N. Waterhouse, Composition changesaround sulphide inclusions in stainless steels, and implications for theinitiation of pitting corrosion, Corros. Sci. 52 (2010) 3702–3716.

[12] A. Szummer, M. Janik-Czachor, S. Hofmann, Discontinuity of the passivatingfilm at nonmetallic inclusions in stainless steels, Mater. Chem. Phys. 34 (1993)181–183.

[13] T. Shibata, Stochastic studies of passivity breakdown, Corros. Sci. 31 (1990)413–423.

[14] D. Nakhaie, M. Zakeri, N. Naghizadeh, M.H. Moayed, Effect of thiosulfate onpitting corrosion of 316SS: II. Metastable pitting and transition to stability, J.Electrochem. Soc. 162 (2015) C1–C7.

[15] J. Soltis, Passivity breakdown, pit initiation and propagation of pits in metallicmaterials – review, Corros. Sci. 90 (2015) 5–22.

[16] G. Daufin, J. Pagetti, J.P. Labbe, F. Michel, Pitting initiation on stainless steels:electrochemical and micrographic aspects, Corrosion 41 (1985) 533–539.

[17] J. Liu, T. Zhang, G. Meng, Y. Shao, F. Wang, Effect of pitting nucleation oncritical pitting temperature of 316L stainless steel by nitric acid passivation,Corros. Sci. 91 (2015) 232–244.

[18] P.C. Pistorius, G.T. Burstein, Growth of corrosion pits on stainless steel inchloride solution containing dilute sulphate, Corros. Sci. 33 (1992) 1885–1897.

[19] G.O. Ilevbare, G.T. Burstein, The inhibition of pitting corrosion of stainlesssteels by chromate and molybdate ions, Corros. Sci. 45 (2003) 1545–1569.

[20] A. Di Schino, J.M. Kenny, Effects of the grain size on the corrosion behavior ofrefined AISI 304 austenitic stainless steels, J. Mater. Sci. Lett. 21 (2002) 1631–1634.

[21] B. Hadzima, M. Janecek, Y. Estrin, H.S. Kim, Microstructure and corrosionproperties of ultrafine-grained interstitial free steel, Mater. Sci. Eng. A 462(2007) 243–247.

[22] X.Y. Wang, D.Y. Li, Mechanical and electrochemical behavior of nanocrystallinesurface of 304 stainless steel, Electrochim. Acta 47 (2002) 3939–3947.

[23] W. Zeiger, M. Schneider, D. Scharnweber, H. Worch, Corrosion behaviour of ananocrystalline FeA18 alloy, Nanostruct. Mater. 6 (1995) 1013–1016.

[24] K.M.S. Youssef, C.C. Koch, P.S. Fedkiw, Improved corrosion behavior ofnanocrystalline zinc produced by pulse-current electrodeposition, Corros.Sci. 46 (2004) 51–64.

[25] G. Wranglen, Pitting and sulphide inclusions in steel, Corros. Sci. 14 (1974)331–349.

[26] D.E. Williams, J. Stewart, P.H. Balkwill, The nucleation, growth and stability ofmicropits in stainless steel, Corros. Sci. 36 (1994) 1213–1235.

[27] M.H. Ras, P.C. Pistorius, Possible mechanisms for the improvement byvanadium of the pitting corrosion resistance of 18% chromium ferriticstainless steel, Corros. Sci. 44 (2002) 2479–2490.

[28] W. Tian, N. Du, S. Li, S. Chen, Q. Wu, Metastable pitting corrosion of 304stainless steel in 3.5% NaCl solution, Corros. Sci. 85 (2014) 372–379.

[29] P.C. Pistorius, G.T. Burstein, Aspects of the effects of electrolyte composition onthe occurrence of metastable pitting on stainless steel, Corros. Sci. 36 (1994)525–538.

[30] B. Yu, P. Woo, U. Erb, Corrosion behaviour of nanocrystalline copper foil insodium hydroxide solution, Scr. Mater. 56 (2007) 353–356.

[31] D.E. Williams, C. Westcott, M. Fleischmann, in: A. Turnbull (Ed.), CorrosionChemistry Within Pits, Crevices and Cracks, HMSO, London, 1987, pp. 61–87.