1 Introduction The methods and results of crevice corrosion testing - mainly focused on the amlication of electrochemical methods -h ave been reviewed bv several authors in 1 1 the past [14].M ore recently some additional review papers have been published [5-8]. This paper is an update of the review that was published in 1980 on behalf of the Working Party on Physico-Chemical Methods of Corrosion Testing of the European Federation of Corrosion [l].T he main purpose is to review developments and data as found in the literature since 1979. However, as it was also intended to produce a document that could be read independently some duplication was inevitable. In addition, the opportunity was taken to broaden the scope of the contents somewhat, so protective methods against crevice corrosion and crevice corrosion monitoring have been included in separate chapters. The literature review is based on an online search in the files of the following databases: INSPEC file 2 NTIS file 6 COMPENDEX file 8 Metadex file 32 CA SEARCH file 399. These databases were accessible via the DATALOG DATABANK (the file numbers given above refer to this databank). The descriptor used was 'crevice corrosion'. A first search was performed for the period 1979-1992, obtaining 1161 titles: Metadex 903 titles CA SEARCH 219 titles NTIS 39 titles In the search, the Metadex file was addressed first, while precautions were taken not to include titles from the other files, which were already found in Metadex. Next from the 1161 titles about 750 titles were selected for closer examination, the majority of which are included in this review. To further update the review the same procedure was repeated several times for periods extending from the beginning of 1993 to the end of 1998, adding about 450 titles. In addition, a number of references originating from 1999 were added. Apart from presenting an overview of the data found on crevice corrosion testing and its results the review is intended to provide a base for quick reference to the bulk of the relevant literature. For this reason, many references, in particular on the application side, are very brief. However, as far as these are available the corresponding file numbers of the DIALOG DATABANK have been given in the 2 Survey of Literature oii Crevice Corrosioti - 1979-1998 references to provide interested readers with a quick and easy route to more detailed information than was possible to include in this review. This project was initiated by the EFC Working Party on Physico-Chemical Methods of Corrosion Testing. However, as the WP on Marine Corrosion also expressed an active interest in the results it was decided to finalise it as a common project of both Working Parties. To accommodate the interest of the WP on Marine Corrosion in many relevant cases the marine aspects of the subject matter have been discussed separately. During the long time span of this literature survey the UNS system for indicating specific alloys was introduced. In the references, the alloys are indicated by the codes as used by the authors. UNS numbers together with their common names can be found in a NACE publication.* *NACE International Corrosion Engineers Reference Handbook, Ed. R.S. Treseder, co-ed R. Baboian and C.C. Munger. Publ. NACE International, Houston, Tx, USA. 2nd edn, 1991. 2 Theory and Modelling of Mechanisms 2.1. General Sharland has reviewed the literature on the theoretical models that have been developed to describe and predict the various stages of localised corrosion with particular reference to pitting and crevice attack. In particular, initiation and propagation models are discussed. The accuracy of the predictions of the models and the validity of the various approximations are discussed and both the achievements and the weak areas of the current state of corrosion modelling are demonstrated [9]. The same author presented a finite-element model of the propagation of corrosion in crevices and pits [lo]. Walton described a generalised model capable of modelling mass transport, interphase mass transfer and chemical reactions in crevices and pits. The model was applied to experimental crevice corrosion data on iron in acetate and sulfuric acid, obtaining excellent agreement between model predictions and experimental results [ 111. Alkire reviewed the fundamental principles of transport phenomena with particular emphasis on mass transfer and on current and potential distribution phenomena [12]. Turnbull has presented an analytical solution for the evaluation of the time-dependent distribution of the oxygen concentration along the length and across the width of a parallel- sided crevice or crack in a metal in contact with an aqueous solution [13]. The same author reviewed the extent of information on the solution composition and electrode potential in pits, crevices and cracks [14,15]. For each system, the potential in the cavity is compared with the external potential and the magnitude of the potential drop is discussed in terms of the polarisation characteristics of the metal, the ease of mass transport, etc. The pH values in the cavities are interpreted based on the nature of the electrode reactions in the cavity and the hydrolysis of dissolved metal cations. In addition, the aspects of chemistry within localised corrosion cavities were discussed [ 161. Markworth presented a semi-quantitative description of the spatiotemporal distribution of the potential in a long, narrow crevice, based on an analysis of experimental data [ 17,181. Page et nl. characterised crack tip electrochemistry using a simulated crack [19]. Simulated cracks were used by a number of other authors, including Charles et nl. [20], applying several electrochemical techniques and Parkins and Craig who performed current and potential measurements [21]. Tsujikawa et al. presented an analysis of mass transfer in a crevice region for the concept of the repassivation potential as a characteristic for crevice corrosion [22]. Reingeverts et al. studied the distribution of the corrosion process within a crevice in a unidimensional approximation [23,24]. Moran et al. reviewed the mechanisms of crevice corrosion of several passivating and non-passivating alloys in chlorinated environments [25]. Turnbull presented a review of modelling of pit propagation kinetics [26]. Nystrom et nl. published an approach for estimating anodic current distributions in crevices 4 Siirziey of literature OIZ Crevice Corrosioiz - 1979-1998 crevice wall using a double numerical differentiation method [27]. Polyanchukov discussed the mathematical simulation of crevice corrosion in acid medium, aimed at the mechanism of inhibition and protection [28]. Kurov et al. discussed the similarity of corrosion conditions in a narrow crevice and in a crack under stress [29]. Parkins presented results showing the significance of crevices, pits and cracks in environment-sensitive crack growth [30]. Zuo and Tin studied chemical and electrochemical changes within corrosion cracks, in particular the interrelation between currents, chloride migration and pH values 1311. Attia published thermal modelling of crevice corrosion and its interaction with fretting in confined geometries [32]. Walton presented a mathematical and numerical model for evaluation of crevice and pitting corrosion in radioactive waste containers [33]. Mazza discussed the crevice corrosion process as a form of corrosion that spontaneously occurs when specific thermodynamic, geometric and kinetic conditions combine at the metal surface, leading to the formation of an occluded cell. The whole process of crevice corrosion has been delineated in a 'phenomenological tree' where the prominent single processes appear linked up with the phenomena occurring at the external and internal surfaces of the crevice and inside the solution [34]. Pickering et al. developed a general model for crevice attack which can be used in conjunction with active as well as passivable alloys; in this model the influence of IR drop between the outside surface where the cathodic reaction predominates and the anodically reacting crevice walls is duly recognised (Figs 1 and 2) [8].X u and Pickering proposed a new index - the critical distance into the crevice - for characterising the crevice corrosion resistance of a material under specified conditions. Its advantages are: 1. It may be obtained through experiment and may be estimated through computational approaches; 2. It has a distinct and straightforward physical meaning; 3. It may be employed in the engineering design, and 4. It is a single parameter, which can reflect the integrated influence of several factors known to affect the crevice corrosion resistance [35]. Pickering and Ateya investigated the dynamics of interaction between the IR potential drop and compositional changes during the activation and propagation of crevice corrosion [36]. Pickering discussed some consequences of IX voltage in electrode processes occurring within pores and cavities [37]. Evitts proposed modelling of crevice corrosion, applying physicochemical simulation [38]. Engelhardt et al. proposed a simplified method for estimating corrosion cavity growth rates. The method is based on the assumption that if the rate of an electrode reaction depends (in an explicit form) only on the potential, the pit growth rate depends only on the concentrations of those species that determine the potential distribution near the metal within the cavity [39]. 'i Passive current Cathodic reaction density at outer surface + 02+2H20 4e 40H- + rJ I hr L I1 le E Crefice Anodic reaction M+ Mn+ +ne- (Crevicing) + logi Fig. 1 Sckeiiintic of n metnl zilitk n crezyice (left) mid its rrintckiiig polnrisnfiori ciirue (rigkf), 7diicI1e xists iii tke xdirectioii on the creziice zcmll nnd illustrntes tke operntioii oftke IR ineckniiistn diiriii<gc orrosion. Tkefollozoing explanation is derivedfrom tke snm reference. As the oxidnnt iii the crezlice is coiisiinzed, lorig cirrreiit pntks finin iiistend ofriin) cniz deuelop betzwen nnodic sites i/i tke creuice niid cntkodic sites on tke exteriinl surfnce fleff). Tkis current (I), flozoiiig tkrozigk n iiiediirin ofresistniice R such ns tke nyieous solution iri the creziice, produces nn IR ~oltnged rop. Tkis results in nn electrode poteiztinl 077 the creuice zunll, E(x), tknt is less noble tknri tkepoferitinl0 77 tke crezlice operiirig 017 fke outer sriqnce x = 0, E(s), by! tlie rnngnitude oftke IR drop befztlcen t h el ocnfions: E(x) = E(s) - IR. Herice, the rnngrzitude qftke potential shift is deperideiit mi flie currentflorcling out ofthe creoice nizd the resistmice oftke currerit path, zokick is njiiiction oftlie creuice apeiiing dimerisioii, electrolyte coizductiuity, niid the possible preserice of riorr-coiidiictiiig (solid niid/or gaseous) corrosion products. For metnllelectrolyfe systems coiztniniizg nn actizle/pnssiz.e trniisitioiz the IR drop zuifkiri the creuice cnii be lnrge erioiigk to lower tke poteritinl E(x) into tke nctizle regioii witkont tke iieedfor accoinpanyitzg pH cknrzges. Tliits, Efx) < Epassb eyoiid the xpassl ocntioiz on the creuice r~nllE, pas,being the poteiitinl of the nctizle to pnssizv trnnsitioii i7i the polarisntioii curue niid xpaStsk e distniice iuto the crevice where Epassexisfs OH tke creztice rcinll. The Figure also illirstrntes tknf beyorid the xpassdistniice in n siijficiently deep creuice the creaice zrnll contailis nll poteiitinls in the nctizv region oftke niiodic polarisntion cwue. According to the IR mechanism,for crevice corrosioii to occiir tke poterztinl skiff rnust be grenter thnii tke difference betzoeen E(si mid Epass( denoted ns A@*). Therefore the criteriorzfor creuice corrosioii is IR > A@". Mieii a metnl/electrolyte system does not contnin mi nctiue region creziiciiig cnii not occiir by potential skffn lone. Hoztwer, iffhes olution coinpositioii ckniiges ouer time imide tke creziice (e.g. by ncid$cntion nnd/or chloride ion nccuinulnfion), n sigiiificaiit penk con result. Aiz exnriiple of tkis is the cnse ofirori in riiost nlknline solutions. No actiue penk existsfor tke bulk solution, nltkougk, after ncidificatioii niidior chloride ion build-up /ins occurred iri tke creztice, penk curreiits inside the crezJicec niz be sezlernl orders of mngiiitude lnrger tkniz tke pnssizle current. In tke context qf tke IR riiechniiisiri tkese nggressizle solutioii comfitions promote creuice corrosioii by decrensing A@* nizd iircrensiiig IR (see Fig. 2). After Pickeriiig et al. 181 6 Survey of Literature oii Crevice Corrosioiz - 1979-1998 [ ,I! /R : Chloride : : Chloride-free Fig. 2 Sckeinatic aiiodic polarisntioii curves slioziiiiig the roles of acidjficntion and chloride ion ncciimulnfion in increming the size of the active peak, zclhich in tztrii decreases A@* aizd iiicrenses IR, Wereby proinotiiig the crezlice condition, IR > A@*. Crevice nt the top inatdies the E-pH ciirzv for the pH 4, cliloride-cof?tniiiirzg solution. These cliniiges iri A@* arid IR resultfrom the iiicrense in size oftke nctive penk that occurs for a decrense in pH and/or increase in chloride ion concenfmtion. Hozi7ezw, the orderofrnngiritude increase in I cnused by the decrense in pH is qfset som~zi~hhnfj the decrense in R, zuhich is caused mainly by the increase iii the hydrogetr ioii conce~zfrationT. he Figure nlso illzisfrntes that pitting corrosioti cnii he iiiifinted on the outer sirrfnce nizd port ofthe zvny into the crevice, zvheii clrloride iotis ore present in the birlk soliitiori nrid the oirter siiTface potelifid, E,, is nbosv the pitting poferitial Epit. This phenortieiion hns beeii ohsenled diwing the crezGce nttnck qf iron snmples. Tlie appearmice of pitted, pnssive mid active regions at iiicreasing distmice x on tlie crezlice zonll coilfirm that the poterz tinl distribii tior1 oil the crmlice rid1 is flint qftlie aiiodic polnrisnfioiz cirrzr, zrlhicli is nlso s1iowi by direct potential ttiensuretim ts. Accordiizg to Pickering et al. [8]. It is well known there is a strong relation between pitting and crevice corrosion mechanisms. However, to include the theory and further details of pitting in this review was thought to be outside the scope of this work. Nevertheless, without going into detail a number of papers on pitting have been included on behalf of the interested reader [40-431. Flis discussed the interdependence between pitting, crevice corrosion and stress corrosion cracking (441. 2.2. Iron and Steel Dmytrakh discussed electrochemical conditions in corrosion crevices in steel, stainless steel and aluminium alloys [45]. Leidheiser et al. discussed the factors affecting the Theory nnd Modelling of Mechnizistns 7 pH developed within carbon steel crevices [46]. Siitari performed experimental and theoretical modelling studies of the initiation of crevice corrosion using pure A1 and Fe [47]. McCafferty published thermodynamic aspects of the crevice corrosion of iron in chromatelchloride solution [48]. Ibe et rzl. numerically simulated localised chemical reactions in a crevice of an Fe- specimen immersed in water at elevated temperature (285°C)[ 49].W alton developed a transient mathematical and numerical model of crevice corrosion for active and passive metals. The model is general in format and applicable to a variety of metallic and electrolyte systems. It considers electrode kinetics, including both cathodic and anodic reactions with an active /passive transition. Chemical reactions in solution are generalised to facilitate simulation of a variety of electrolytes. The model is applied to experimental data on crevice corrosion initiation in stainless steel and on active corrosion of iron [50]. Shinohara et al. investigated numerical and experimental simulation of iron dissolution in a crevice with a very dilute bulk solution containing NaCl or Na,SO, [5l].C hang et al. published a modelling study on the propagation stage of crevice corrosion in carbon steels [52]. A microscopy /local probe method was developed by Pickering for studying crevice corrosion and its application to iron. Transparent polyethylene is used as one surface of the crevice, enabling in situ optical microscopical examination of the metal- crevice surface. This new realtime microscopy technique was combined with a potential probe to allow simultaneous potential measurement at the observed crevicing site on the crevice wall, together with measurements of the current flowing out of the crevice and of the pH in the crevice by extraction of the crevice solution. Application of the technique to iron in both acid and alkaline solutions shows that the crevice wall in the depth direction is a plot of the electrode potential of the crevice- electrolyte polarisation curve when the active crevicing region is in the active loop of the plot. The role of both acidification and chloride ion buildup in the crevice is to enlarge the active loop so that a smaller IX drop in the crevice sets up the crevicing action. Thus, crevicing occurs in shallower or more open crevice geometries in the presence of chloride anion or if acidification takes place [53]( see further explanation in next subchapter). Cho and Pickering discussed the IR-induced mechanism of crevice corrosion of iron in various electrolytes, demonstrating the different aspects of the mechanism (541. 2.3. Stainless Steels The crevice corrosion process has been simulated by a number of steady state and transient models. The steady-state models (Turnbull [55], Galvele [56-581, Sharland et al. [59-611 describe the steady-state concentrations of the various ions and compounds in a series of simultaneous equations, which are solved by appropriate methods. However, time being an important variable in crevice corrosion a number of transient models have also been developed. Thus, Sharland also presented a mathematical model of the evolution of the solution chemistry and electrostatic potential tvithin a passive crevice in ferrous alloys with Cr contents ranging from 1 to 25% [62].W atson et al. developed a transient mathematical model for calculation of the incubation time of crevice corrosion. In the model, a new treatment of the 8 Survey of Literature 017 Crevice Corrosion - 2979-2998 transport processes is used, including both ionic migration and diffusion. Transient concentration results demonstrated that the major reason for the pH decrease in the crevice solution is the production of soluble metal hydroxides, particularly Cr(OH)2+. The model was able to predict relatively accurate incubation times for alloys 304 and 316L. However, simulation for alloys 904L and Inconel 625 demonstrated the necessity of further refinement of some factors, including long-term passive currents for alloys, effects of metal chlorides on crevice pH and a more accurate definition of the critical crevice solution pH [63,64]. Another transient model has been proposed by Fu and Chan, in which the potential of the crevice solution is calculated by specific and unique equations [65]. Oldfield and Sutton originally presented a model for crevice attack on stainless steels, based on four consecutive steps [66]: (1) oxygen depletion within the crevice, (2) decrease of pH and increase of C1- concentration in the crevice solution, (3) permanent breakdown of the passive film, and (4) crevice corrosion propagation. In Fig. 3 the many factors which are operative in the mechanism in one way or another are shown schematically. A number of variables have to be supplied to the CREVICE TYPE CREVICE GEOMETRY TOTAL GEOMETRY - metal/metal - exterior to interior crevice - metahon-metal - number of crevices BULK SOLUTION / COMPOSITION ALLOY COMPOSlTlON - - CI- content major constituents - 0, content - minor additions - - impurities PH -pollutants PASSIVE FILM BULK SOLUTION ENVIRONMENT - CHARACTERISTICS temperature CREVICE SOLUTION -passive current - hydrolysis equilibria -film stability - -volume reaction rates - corrosion products ELECTROCHEMICALR EACTIONS MASS TRANSPORT IN AND OUT - metal dissolution OF CREVICE - 0, reduction - migration - H, evolution -diffusion - other reduction reactions -convection - IR drop Fig. 3 Irnporlnrztfnctors iiz crevice carrosioii rriecliniiisrri; after Oldfield orid Siittorr 1661. Theory atid Mollelliizg of Mechnizisms 9 model as input parameters: crevice geometry, bulk concentration of dissolved oxygen, passive current density, diffusion coefficients, stainless steel composition, composition and concentration of the bulk solution, pH, chloride concentration of the critical crevice solution (CCS), i.e. when passivity cannot be maintained and specific constants for the chemical reactions that are apt to occur in the crevice solution, for instance hydrolysis reactions. With these inputs the model can predict the time required to deplete oxygen in the crevice, the pH and the composition of the crevice solution as a function of time and finally the time required to break down the passive film and so initiate crevice corrosion. The main application of the model is to calculate initiation times for stainless steels under different conditions. The authors extended their mathematical model to propagation [67,68]. The basis of this was to consider an electrochemical corrosion cell with the actively dissolving alloy within the crevice as the anode, in conjunction with a cathode usually sustaining oxygen reduction outside the crevice and/or hydrogen evolution within the crevice. The rate of the propagation reaction is determined by the cell voltage, electrolyte resistance, cell geometry and the relevant polarisation characteristics for the anodic and cathodic reactions. During propagation, the overall corrosion cell geometry might be influenced by a large number of factors, Le.: - crevice gap and depth, - exterior surface to interior crevice area ratio, - number of crevice sites, - location of anode(s) with respect to the crevice mouth, - available cell voltage, - electrolyte resistance and pH within the crevice, - electrolyte resistance outside the crevice, - extent of oxygen reduction on the external crevice area, - degree of hydrogen evolution possible within the crevice. As a first step, a preliminary model was presented to calculate the anticipated initial propagation rate. Prior to initiation it is assumed that the passive current flows over the entire crevice area with an equal oxygen reduction current external to the crevice. Since this current is small, it is unlikely to be restricted by IR drop considerations. However, on breakdown of passivity within the crevice, both of these assumptions change: breakdown will occur over a discrete area within the crevice and the current will increase by orders of magnitude so it becomes necessary to consider the IR drop down the crevice. It is recognised that as propagation proceeds, several parameters such as the cell voltage, crevice resistance and crevice geometry 10 Siirvey ofliteroture oil Crevice Corrosioii - 1979-1998 will likely change in some interrelated fashion. Assuming that the breakdown occurs in the last 1 mm of the crevice, furthest from the crevice mouth, it is possible to calculate the resistance from the mouth of the crevice to the active area. In Fig. 4 a schematic illustration is given of the change in current within the crevice as a function of crevice solution pH. After an instantaneous increase due to depassivation, there is a linear relationship between log peak current and pH as determined from polarisation data. The pH of the crevice solution eventually becomes limited by mass transfer considerations, which can occur either before or after the current is limited by the IR-drop along the crevice. By applying Ohm's law, the maximum value of the IR-limited current is determined from the solution resistance inside the crevice and the corrosion cell voltage. So the extension of the model to handling the propagation stage requires two additional parameters: the slope of the log peak current vs pH curve and the corrosion cell voltage. The extended model allows calculations of anticipated initial propagation rates within defined crevice conditions. It is recognised that as propagation proceeds ./ Propagation current - without IR restriction / I / I // 1 / ;-/- Max. IR limited / I propagation current - - ---/ Increase in I I propagation current 1 I I I -- Propagation current I at breakdown I I I Passive cyrent I I\ I I I I I I I I I I I v - I pH of Limiting pH due breakdown to mass transfer Decreasing pH Fig. 4 Scheinntic illzistrntion ofnnodic czirreiit ns nfunction ofcrevice solution pH. The czrrrent is nssirnied to iiicrense iiisfnntmieoirslyf roiii the pnssizv zwlire of typicnlly 0.1 FA cw2 to nil actizle zdue of 10 PAc ~ii-I~t. t hen inctenses nccording to n lirzenv relntionship betziwiz the log peak current nnd the pH ns detenniiiedfrom the polarisntioii dntn. The pH ofthe cvezlice solcrtiori ezientztnlly becomes limited by the IR drop dozi~iit he crevice. The ninxiiiiitni ztnlzre of this IR limited current is determinedfrom the tesistnnce dozivi the crezlice niid the corrosioii cell zloltnge by npplyiiig Oliin's kI711. After Old$eld, Lee niid Kniii 1671.