Journal of Electroanalytical Chemistry 720-721 (2014) 121–127

Contents lists available at ScienceDirect

Journal of Electroanalytical Chemistry

journal homepage: www.elsevier .com/locate / je lechem

Local flux of hydrogen from magnesium alloy corrosion investigatedby scanning electrochemical microscopy

http://dx.doi.org/10.1016/j.jelechem.2014.03.0021572-6657/� 2014 Elsevier B.V. All rights reserved.

⇑ Corresponding author. Tel.: +1 5143983898.E-mail address: Janine.mauzeroll@mcgill.ca (J. Mauzeroll).

Ushula Mengesha Tefashe a, Michael Edward Snowden a, Philippe Dauphin Ducharme a, Mohsen Danaie b,Gianluigi A. Botton b, Janine Mauzeroll a,⇑a Laboratory for Electrochemical Reactive Imaging and Detection of Biological Systems, McGill University, 801 Sherbrooke Street West, Montreal, QC H3A 0B8, Canadab Department of Materials Science and Engineering, Brockhouse Institute for Materials Research and Canadian Centre for Electron Microscopy, McMaster University, Hamilton,Ontario L8S 4M1, Canada

a r t i c l e i n f o

Article history:Received 13 January 2014Received in revised form 26 February 2014Accepted 3 March 2014Available online 28 March 2014

Keywords:Magnesium alloyCorrosionScanning electrochemical microscopySubstrate-generation/tip-collectionHydrogen detection

a b s t r a c t

Herein, we report the successful electrochemical detection and quantification of H2 fluxes produced froma corroding magnesium alloy using the substrate-generation/tip-collection mode of scanning electro-chemical microscopy (SECM). Using a platinum microelectrode, the variation in H2 fluxes was imagedrevealing the time-dependent corrosion reaction. Our results demonstrate that through careful controlof the corroding media and immersion time, quantitative SECM approach curves, devoid of convectiveeffects, were acquired. Comparison to an idealized numerical model enabled the quantification of thelocal H2 flux for a given corroding area. These data demonstrated that the active site size increasesthroughout the reaction, whereas the flux of H2 generated at an active site increased for the first hourof immersion followed by decrease in the flux of H2 for times greater than 1 h.

� 2014 Elsevier B.V. All rights reserved.

1. Introduction There is evidence to suggest that Mg alloy corrosion is initiated

Despite the high corrosion rates exhibited by magnesium alloysin aqueous solutions, the automotive industry is currently imple-menting magnesium alloy components into high-end vehicles toreduce the overall weight of the vehicle, thus reducing fuel con-sumption [1,2]. In trying to expand the use of magnesium alloycomponents across all vehicles, automotive industries need toremove expensive separators currently used to limit galvanic cor-rosion occurring between the Mg alloys and other nobler metalcar parts (bolts, sheets). Unavoidably, the improvement in the cor-rosion resistance of the targeted Mg alloys and choice of protectivelayers will require detailed a priori understanding and careful con-trol over the Mg alloy microstructure and distribution, which canbe achieved through the use of a robust high-resolution analyticalmethodology to investigate the corrosion reaction in situ.

Microscopic real time investigations involving scanning electro-chemical microscopy (SECM) will provide a much needed under-standing of how heterogeneities within the Mg alloy or coatinginfluences the local and overall corrosion reaction. Already severalfundamental studies, monitoring macroscopic corrosion propertiesof new Mg alloys [2–4], pointed to an ill-defined time-dependentchange of the surface topography and rate of corrosion.

at heterogeneous sites in the microstructure, e.g. at grain bound-aries [5]. The heterogeneities are mainly due to localized variationsin the chemical composition of the alloy formed during the castingprocess (e.g. die-cast, sand-cast or graphite-cast) [6]. The die-cast-ing process is commonly used in Mg alloy component manufactur-ing because it allows high production rate, reproducibility of thecast component and lower cost compared to sand-cast or graph-ite-cast production [7]. In this work, the industrial AM50 die-castalloy has been used since it represents one of the most corrosionresistant Mg alloys and is thus of great interest to the automotiveindustry. The AM50 alloy has a microstructure typically consistingof primary a-Mg grains along with a network of partially or fullydivorced mixture of intermetallic b-Mg17Al12 and eutectic Mg. Var-ious intermetallic phases of the Al–Mn system are also present inthe microstructure [8,9].

The galvanic corrosion of Mg in aqueous solutions producemagnesium hydroxide, Mg(OH)2, and molecular hydrogen, H2, assummarized by the following reactions [2].

Anode reaction Mg!Mg2þ þ 2e� ð1Þ

Cathode reaction 2H2Oþ 2e��H2 þ 2OH� ð2Þ

Overall reaction Mgþ 2H2O�MgðOHÞ2 þH2 ð3Þ

122 U.M. Tefashe et al. / Journal of Electroanalytical Chemistry 720-721 (2014) 121–127

where Mg2+/Mg has a standard electrode potential of �2.37 V [10].In dilute chloride solutions, which accelerate the corrosion reactionrate of Mg alloys, the observed corrosion potential of Mg is ca.�1.7 V vs. SCE [2]. Therefore, Mg is readily oxidized when galvani-cally coupled with other metals, e.g. Aluminum. Alloy AM50 under-goes spontaneous corrosion when immersed in aqueous solutiondue to the formation of microgalvanic couples between the bulkMg (noted a in Fig. 1) and the network of eutectic mixture of theintermetallic b-Mg17Al12 and Mg, Al–Mn intermetallic phases orFe-containing inclusions that we simplistically refer to as b inFig. 1. Within these couples and during the initial stages of corro-sion, the a-Mg matrix typically acts as the anodic sites (Eq. (1))thereby releasing Mg2+, while H2 is produced at the b cathodic sites(Eq. (2)) [11,12].

To monitor and quantify the in situ release of H2 produced at theb region and intermetallic cathodic sites of AM50, a platinummicroelectrode (ME) was used (Eq. (4)).

H2�2Hþ þ 2e� ð4Þ

SECM is a well-established scanning probe technique, where aME [13] is positioned close to a substrate surface for localizeddetection of electrochemically active species (e.g. H2) [14–16]and its interaction with the substrate surface [17–21]. Typically,the current signal recorded at the ME relies on the diffusional masstransport of the redox-active species and the local surface reactiv-ity of the substrate. Although a handful of SECM corrosion studiesmapped the generation of specific metal cations at the anodic sitesof the corroding surfaces [22–26] or amperometrically monitoredthe dissolved oxygen evolution in the electrolyte solution at thecathodic areas of the corroding sample [27–29], direct monitoringof H2 or H+ fluxes are limited [15,30,31]. The ability to performquantitative approach curves is particularly challenging duringMg alloy corrosion because there is a limited experimental windowin terms of corroding media and time that is devoid of significantconvection effects originating from gas evolution and severe topo-graphical deviations. Herein, we have successfully identified exper-imental conditions (0.6 M NaCl aqueous solution and immersiontimes smaller than 60 min) where substrate-generation/tip-collec-tion mode of SECM can be used to monitor the local flux of H2 pro-duction from a Mg alloy surface. Eq. (3)shows that one mole ofcorroding Mg is accompanied by the production of one mole ofH2, hence a direct comparison of the detected H2 at the ME tothe rate of corrosion can be made. SECM mapping and multipleSECM approach curves were performed to observe the time-depen-dent change of the die-cast AM50 Mg alloy surface. Quantitativeapproach curves were compared to a numerical model to evaluatethe magnitude of the H2 flux as the active size of H2 producing fea-tures increased.

Fig. 1. Schematic representation of the process considered in SECM investigation ofthe aqueous H2 detection during Mg alloy corrosion for an active surface.

2. Experimental section

2.1. Materials and sample preparation

The aqueous NaCl (ACP, Montreal, QC) solution was made withanalytical grade reagents and deionized water (Millipore MiliQwater 18.2 MO). The samples for this study were industrial AM50die-cast magnesium alloys received from General Motors India Pri-vate Limited. The chemical composition of the alloy is predomi-nately Mg with the following minor components (in wt.%): 4.9Al, 0.45 Mn, 0.2 Zn, <0.05 Si, <0.01 Ni, <0.008 Cu, <0.004 Fe, and0.001 Be [32]. The AM50 samples were successively polished with800 and 1200 silicon carbide grinding papers using water as alubricant prior to cold mounting in epoxy resin (Struers Epofix,Ontario, Canada). The mounted samples with the exposed surfacearea of 1 cm2 were then finely polished to achieve a mirror finishusing an established protocol [33]. The microstructure of the die-cast AM50 alloy was analyzed using a light microscope (Zeiss Axi-oplan 2 Imaging microscope, with a Hal 100 halogen lamp) and athigher magnification with a dual-beam scanning electron micro-scope/focused ion beam (SEM/FIB) (a Zeiss NVision 40).

2.2. SECM apparatus and procedure

The SECM experiments were carried out using a HEKA scanningelectrochemical microscope ELPro scan 1 integrated with shearforce unit (HEKA Electronik, Germany). The cell was made fromTeflon with a small opening in the middle into which the epoxymounted alloy sample was tightly fitted. A platinum microelec-trode (ME) was utilized as the working electrode, a chloridizedsilver wire as a quasi-reference electrode (Ag/AgCl QRE) and a0.5 mm diameter platinum wire as a counter electrode (GoodfellowCambridge Limited, Huntingdon, England). MEs were fabricatedin-house by sealing Pt wires (25 lm diameter, Delta ScientificLaboratory Products Ltd., Canada) into borosilicate glass capillarieswith outer diameter of 1.5 mm, inner diameter of 0.7 mm (SutterInstrument, USA) and sharpening the probe end to a ratio RG = rglass/rT � 10, where rglass is the radius of the insulating sheath and rT isthe radius of the active ME. The resulting Pt MEs were polishedon a microcloth pad (Struers MD Chem cloth) using a series ofwater-based alumina slurries (1 lm, 0.3 lm and 0.05 lm). Electro-chemical cleaning was performed in H2SO4 (0.1 M) cycling between1 V and �0.5 V vs. Ag/AgCl until well-defined typical features ofplatinum were recognized, including H2 adsorption/desorption,and platinum oxidation/reduction [34]. All electrochemical mea-surements for the local detection of dissolved H2 were performedin freely aerated 0.6 M aqueous NaCl solution. We have chosen thisconcentration, which is standard for Mg alloy corrosion tests byGeneral Motors, to obtain industrially relevant data. The approachcurves were measured over the same location on the alloy surfaceat different immersion times. Line scan experiments were per-formed using SECM in shear force distance control mode to decou-ple the effect of surface topography from the reactivity on MEcurrent. In this mode, the Pt ME is scanned at a constant distanceabove the sample enabling exclusive surface reactivity measure-ment. The amplitude controlled shear force scan parameters usedin this experiment were; stimulation amplitude = 120 mV, stimula-tion frequency = 317 kHz, scan speed = 0.2 lm s�1, and piezo stepsize = 3 nm. Time-dependent 3D SECM maps were acquired inconstant height mode over the corroding alloy surface at atranslation speed of 20 lm s�1. The ME was biased at ET = -0.05 Vvs. Ag/AgCl in order to fully oxidize H2.

Numerical simulations were performed using the finite elementmethod (FEM) modeling package, Comsol Multiphysics 4.3a (Com-sol AB, Sweden) with the Livelink for MatLab R2012b (MathWorks,

U.M. Tefashe et al. / Journal of Electroanalytical Chemistry 720-721 (2014) 121–127 123

USA). Details on the parameters for the simulations are given inSupporting Information.

3. Results and discussion

3.1. AM50 alloy characterization

When imaged by optical microscopy (Fig. 2a), the AM50 die-cast Mg alloy shows a region of a phases within a network of sec-ondary phases (darker regions) distributed across the substratesurface. Importantly, the optical micrograph showed a high qualitysubstrate surface, clean and free from macroscopic polishingdefects. Further details of the microstructure were observed athigher magnification in the SEM micrographs (Fig. 2b). The precip-itates showing higher intensity are the Al–Mn intermetallics. This

Fig. 2. (a) Optical microscopy image of the polished die-cast AM50 alloy showingthe general microstructure. SEM micrographs showing the microstructure at ahigher magnification, (b) before corrosion and (c) after 1 h corrosion tests in 0.6 MNaCl(aq) solution. Both micrographs in (b) and (c) were acquired using secondaryelectrons signal.

was confirmed using X-ray energy-dispersive spectroscopy.Fig. 2c is an SEM micrograph of a typical AM50 die-cast region fol-lowing 1 h of immersion in 0.6 M NaCl. Corroded AM50 die-castsurfaces clearly demonstrated the presence of depressed regionsresulting from Mg2+ dissolution and the presence of corrosionproducts was observed across the surface of the AM50 alloy. Bothobservations were consistent with the time-dependent corrosionprocess proposed in this work and corroborated by previous stud-ies [7].

3.2. Detection of H2 at the Pt ME

Fig. 3a shows a typical CV recorded at 100 mV s�1 at a 25 lmdiameter Pt ME positioned 5 lm above a glass substrate (red line)and the Mg alloy substrate (black line) in 0.6 M NaCl(aq) solution.When positioned over the glass substrate, the CV response showsno significant signal due to H2. When positioned over the Mg alloysubstrate, a steady-state signal of 45 nA is observed during theanodic sweep. From Fig. 3a, we conclude that the levels of H2 areundetectable in the electrolyte in the absence of the corrosionreaction and that a potential of �0.05 V is suitable to detect theoxidation of H2 at the Pt ME during SECM experiments. The oxygenreduction current is partially suppressed by hindered diffusionwhich makes it possible to set the ME potential at �0.05 V todetect H2.

Fig. 3. (a) Cyclic voltammetry at a 25 lm diameter Pt ME positioned 5 lm abovethe substrate surface performed at a scan rate of 100 mV s�1 in 0.6 M NaCl solutionmeasured over a glass substrate (red) and over a die-cast AM50 Mg alloy (black). (b)SECM-shear force line scans for simultaneous measurement of topography (redline) and ME current for the detection of H2 (black line) evolved from a corrodingdie-cast AM50 sample surface after 1 h of immersion in 0.6 M NaCl solution. The MEwas held at �0.05 V vs. Ag/AgCl to a steady state oxidation of H2 at the electrodesurface. The shear force scan parameters: stimulation frequency 317 kHz, stimu-lation amplitude = 120 mV, lateral translation speed = 0.2 lm s�1, and z-piezo stepsize = 3 nm. (For interpretation of the references to colour in this figure legend, thereader is referred to the web version of this article.)

124 U.M. Tefashe et al. / Journal of Electroanalytical Chemistry 720-721 (2014) 121–127

3.3. SECM line scans

Macroscopic studies of Mg corrosion suggest a time-dependentevolution of the surface due to the formation of a corrosion productfilm or material loss during corrosion, which was also observed inFig. 2c following long immersion times (1 h). Significant variationsin topography could influence the electrochemical signal recordedat the ME. Therefore, the generation of H2 was further studied bylateral scanning of the ME over a corroding AM50 surface, whilstsimultaneously measuring the topography of the surface by shearforce microscopy. SECM with shear force distance control modehas been demonstrated to decouple the surface topography fromsurface reactivity [35–37]. Fig. 3b presents the SECM line scan per-formed over a die-cast AM50 Mg alloy substrate where the topog-raphy (red line) and the electrochemical current (black line), due toH2 evolved from the AM50 alloy, was measured simultaneously.Variations in topography were in the 100 nm range, which leadsto negligible ME current contributions given the size of the ME(rT = 12.5 lm, RG = 10). Therefore the significant changes observedin the electrochemical current (black line in Fig. 3b) are due to H2

generated from the AM50 alloy. Based on the small variation in thetopography, it is also reasonable to use constant height (fixed zposition) SECM scan to adequately track the AM50 alloy for immer-sion times less than 1 h.

3.4. SECM 3D images of corroding Mg alloy surface

From the SECM images (Fig. 4) obtained with the ME positionedover a die-cast AM50 surface in a 0.6 M NaCl(aq) solution a time-dependent corrosion process was observed. The ME was poisedat ET = �0.05 V vs. Ag/AgCl in order to fully oxidize H2 during thetranslation of the ME at 20 lm s�1. We have also recorded linescans at different scan speeds and observed no significantdifference between the line scans (Fig. SI-4). Fig. 4a is a SECM

Fig. 4. SECM images of H2 fluxes from the die-cast AM50 alloy after (a) 5 min immersionthe current distribution extracted from the entire maps in a and b, respectively. The Mtranslation of the ME at 20 lm s�1. An overall scan time of 13 min was required to acqu

map recorded after 5 min of immersion which was indicative ofinitial corrosion events and shows current in the range of0.2–0.6 nA. Fig. 4a and c shows a predominantly uniform currentresponse (0.5 ± 0.1 nA) with four regions of significantly lower cur-rent (0.2 nA). Based on corrosion mechanisms previously reportedfor Mg alloys [38,39], these sites of lower current may be due toinitial locations for pit formation on Mg alloys. Based on micro-scopical characterization of the surface, illustrated in Fig. 2a–c, itis possible to observe relative size of anodic/cathodic sites of500 nm–1 lm. In the electrochemical images, much larger activesite sizes of ca. 77 lm diameter were observed. Due to the compar-atively large ME size (25 lm diameter) and the high diffusioncoefficient of H2 (D = 5 � 10�5 cm2 s�1), deducing the exact micro-structure from the SECM scans is non-trivial because of diffusionaloverlap [40].

Fig. 4b was measured after 1 h of immersion and showscurrents of 85–135 nA. The rise in current was nearly three ordersof magnitude over 1 h of immersion time. We attributed this rise incurrent to the increased rate of H2 generation at the Mg alloysurface. Due to the stoichiometric relationship between H2 releaseand Mg dissolution (Eqs. (1) and (2)), these maps provide strongevidence for the increasing rate of Mg corrosion within the firsthour of exposure to an aqueous solution of 0.6 M NaCl. Fig. 4bshows a SECM map with distinct current regions. Within each ofthese regions a homogeneous current (�110 nA) is observed witha variation of ±5% (Fig. 4d). The different current regions couldbe attributed to the combined effects of the heterogeneous alloymicrostructure and the presence of deposited Mg(OH)2 (Fig. 2c).To further investigate this effect by SECM, the resolution must beincreased by employing nanoelectrode probes.

Overall the 3D SECM maps allow us to conclude that there is atime-dependent change of the alloy surface where an overallincrease in the corrosion rate was observed for the first hour anda change in the corrosion microstructure from small pitting sites

, (b) 1 h immersion in a 0.6 M NaCl solution. (c) and (d) Represent the histograms ofE was poised at ET = �0.05 V vs. Ag/AgCl in order to fully oxidize H2 during the

ire the SECM maps.

U.M. Tefashe et al. / Journal of Electroanalytical Chemistry 720-721 (2014) 121–127 125

to the entire surface [41]. Thus, the proposed mechanism of thetime-dependent corrosion process on Mg alloy during quantitativeapproach curve initially considers reaction sites grow from smallindividual active areas. At intermediate times the surface pittingfeatures expand until the reaction sites cover the entire surface.This final stage could be due to expansion of existing areas of activ-ity upon the surface, or by the formation of a thin film. The sche-matic illustration is provided as the insets of Figs. 5c, SI-5c and d.

3.5. Time-dependent approach curves

Time-dependent approach curves were preferred over x–y scansfor quantification because of their inherent increased time resolu-tion and improved sensitivity to the rate and location of H2 evolu-tion (vide infra). Fig. 5a shows approach curves obtained at thesame x–y position on the Mg alloy surface. From these approachcurves we can observe three distinct types of behavior: (i) in theearly stage of corrosion (t 6 1 h), a small ME current far from thesurface (dh > 4rT) is measured and then rises rapidly as the ME-sub-strate distance decreases; (ii) for 1 h < t < 3 h, an increase in MEcurrent far from the surface, which increases as the ME to substrateseparation decreases until dh < 0.5rT, where the current decreasessharply; and (iii) for times greater than 3 h a decrease in currentwith time was observed, and for t P 5 h a steady-state current isobserved upon approach to dh = rT where the current decreasesrapidly.

From the SECM maps presented in Fig. 4, we propose that thearea of the active site and the flux from an active site vary withtime. To understand the significance of both these factors, FEM cal-

Fig. 5. (a) Experimental SECM approach curves on a corroding die-cast AM50 Mg alloimmersion times shown as inset). The curves were acquired using a 25 lm diameter PtSimulated approach curves for a 25 lm diameter Pt ME over active sites of differentcorrespond to active site diameters of 500, 250, 175, 125, 75, 25, 17.5, 10, and 2.5 lm reimmersion and (d) 1 h immersion time to simulated approach curves (solid lines) over actthe lowest flux values corresponds to the active site H2 fluxes of (c) 7, 5 and 3 mmol m�2

of the proposed time-dependent corrosion process on the Mg alloy samples.

culations were performed. Details about the FEM are appended insupporting information. For this study, we make the assumptionthat we are probing an isolated active site, which is located cen-trally beneath the ME. Due to the very high diffusion coefficientof H2, and the density of active sites, these assumptions will under-estimate the ME current response when the electrode is far fromthe substrate surface [42,43]. However, these assumptions are rea-sonable to quantify a range of H2 flux for the corrosion process,which is consistent with the trends in the reported experimentalapproach curves (Fig. 5).

3.5.1. Effect of active site sizeAccording to the proposed mechanism (vide supra), corrosion is

initiated at discrete active sites [9,39,44], which form due to theexistence of micro-galvanic couples caused by the presence of het-erogeneities within the microstructure of the Mg alloy. To under-stand the influence of the change in active site size, FEMsimulations were performed over a site that produced a flux of0.7 mmol m�2 s�1 (Fig. 5b). As the active site size increased thecurrent increased; however, three distinct trends in approachcurve were observed. (i) For active sites with a diameter of25 lm or less, a low bulk current was observed which increasedas the ME-to-substrate separation decreased, similar to the shortimmersion time (t 6 1 h) experimental approach curves. (ii) Foractive site radii between 12.5 lm and 125 lm, the current initiallyincreased as the ME-to-substrate separation decreased until theME was �10 lm above the substrate, where the current decreasedsharply. The wave shape was similar to that observed experimen-tally for times of 1 h < t < 3 h. (iii) For active site areas with

y sample for the detection of dissolved H2 after different immersion times (theME with an RG of 10 at a z approach rate of 2 lm s�1 in a 0.6 M NaCl solution. (b)sizes producing a H2 flux of 0.7 mmol m�2 s�1. Highest to lowest current valuesspectively. Comparison of experimental approach curves (dashed lines), (c) 30 minive site diameter of (c) 10 lm and (d) 15 lm. For the simulated curves the highest tos�1, (d) 8, 7, 6 and 5 mmol m�2 s�1. The inset in (c) shows the schematics illustration

126 U.M. Tefashe et al. / Journal of Electroanalytical Chemistry 720-721 (2014) 121–127

rS > 125 lm, a gradual increase in current was observed uponapproach until the ME is �20–50 lm above the substrate surface,where a sharp decrease in current was observed. The magnitudeof current increased as the active site size increased; however,the rate of current change during approach decreased, tendingtowards a steady-state response when the ME position was greaterthan 50 lm above the surface. This trend was observed for theexperimental data of times greater than 2 h.

3.5.2. Effect of active site fluxIn Fig. 5c and d both the active site flux and size contribution to

the ME current have been investigated. In order to quantify theactive site flux of H2, there was a limited range of active site sizeswith a corresponding flux that reproduced the experimentalapproach curves (Fig. 5c and d). Furthermore, the solution becomessaturated by dissolved molecular hydrogen creating bubbles of H2

after 30 min of immersion (see Fig. SI-2). However, it is only after1 h of corrosion that small H2 bubbles form on the entire surfacedue to the high corrosion resistance of the die-cast AM50 Mg alloy.

Notwithstanding the simplicity of the FEM, the range of H2

fluxes extracted with our model are valid for short corrosion times(less than 1 h in 0.6 M NaCl(aq) solution). Under this regime weobtained a good agreement when the experimental data was over-laid on the simulated approach curves. Fig. 5c presents simulatedapproach curves at different H2 fluxes for an active site of 10 lmdiameter overlaid with the experimental approach curve recordedat t = 30 min. From the overlay of the two curves, it is possible toshow that the H2 flux should lie within 3–5 mmol m�2 s�1 for afixed active site size of 10 lm. Fig. 5d presents simulated approachcurves at different H2 fluxes overlaid with the experimentalapproach curve measured after t = 1 h of corrosion. The experimen-tal approach curve lies between the calculated data for H2 fluxes of6 and 7 mmol m�2 s�1 generated at a 15 lm active site size.

However, for immersion time t > 1 h, the simulated datadeviates significantly from the experimental approach curves(Fig. SI-5a and b). For the prolonged corrosion time: (i) the H2

gas evolution adds convection to the experimental system makingthe assumption of mass transport by diffusion only no longer valid,(ii) multiple neighbouring active sites appear with time (asdepicted by Fig. SI-2) and will contribute to the overall ME current.These effects are not included in the FEM numerical model andgave rise to discrepancies between the experimental and simulatedapproach curves for t > 1 h.

Overall, the active site size and flux analysis from the SECMstudies demonstrate that the active site size increased while theH2 fluxes decreased as AM50 corrosion time increased. This is ingood agreement with fundamental understanding of the corrosionmechanism of Mg alloys where, as corrosion time increased, H2 isgenerated from the entire surface due to increasing exposure ofcathodic surface as well as the local H2 fluxes decreased due tothe deposition of corrosion products across the surface. UsingSECM in situ methodology in combination with numerical simula-tions it was possible to express the rapid time dependent corrosionbehavior of Mg alloys while being able to divide the corrosionmechanism in three different events for a 0.6 M NaCl(aq) solution.First, the initiation of corrosion occurs for t < 1 h, where the H2

fluxes increased rapidly up to 7 mmol m�2 s�1 from small activesite sizes (ca. 15 lm) whilst under diffusion only mass transport.Secondly, for 1 h < t < 5 h, convective effects due to the formationof H2 bubbles on the surface, along with the increase in size andnumber of the active sites were observed. Similar trends expressedby the numerical simulation showed that active site size increasedwhile H2 fluxes decreased. Finally, deposition of corrosion productsacross the surface hindered further H2 evolution highlighted in theproposed numerical model by a decrease of H2 flux whilst theactive site size increased.

4. Conclusion

We have successfully demonstrated a methodology for probingrapid corrosion processes at a Mg alloy surface on the micro-scaleusing SECM. The presented data provided strong evidence tosupport the time-dependent nature of the corrosion process, wherethe initial cathodic reaction sites grow from small individual activeareas to provide sites for the cathodic reaction across the entiresurface. Initially, the local flux from the active areas increases withtime, however, after 1 h of corrosion in 0.6 M NaCl(aq) solution thelocal rate decreased suggesting the formation of a blocking layer, ora surface limitation for the corrosion reaction. These findingssuggest that the SECM methodology is robust in monitoring thecorrosion progress of Mg alloy and does not interfere with thecorrosion reactions. Experimental approach curves show a varia-tion in both size of the active area and local flux generated at theactive area. Comparison of FEM data to experimental approachcurves provided further evidence to support the theory that thelocal flux and the size of active sites was time-dependent. Toenhance the resolution and to decouple the signal from individualactive sites, nanoelectrode SECM probes are envisioned to be used.

Acknowledgments

This work is supported by Natural Sciences and EngineeringResearch Council of Canada (NSERC) and General Motors Canada.

Appendix A. Supplementary material

Supporting information include details of numerical simulation,photographic images of a corroding die-cast AM50 alloy as a func-tion of immersion time, CV for H2 detection at a large ME-to-sub-strate distance, line scans at different scan speed, plot of thecomparison of experimental approach curves to numerical simula-tions for the immersion times of 2 h and 8 h; and the proposedmechanism of H2 evolution after longer immersion times. Supple-mentary data associated with this article can be found, in theonline version, at http://dx.doi.org/10.1016/j.jelechem.2014.03.002.

References

[1] I.J. Polmear, Light Alloys: Metallurgy of the Light Metals, second ed., E. Arnold,London, England, 1989.

[2] G. Song, A. Atrens, Adv. Eng. Mater. 9 (2007) 177.[3] J. Chen, J. Wang, E. Han, J. Dong, W. Ke, Electrochim. Acta 52 (2007) 3299.[4] R.B. Alvarez, H.J. Martin, M.F. Horstemeyer, M.Q. Chandler, N. Williams, P.T.

Wang, A. Ruiz, Corros. Sci. 52 (2010) 1635.[5] A. Pardo, M.C. Merino, A.E. Coy, F. Viejo, R. Arrabal, S. Feliu, Electrochim. Acta

53 (2008) 7890.[6] H. Ding, K. Hirai, T. Homma, S. Kamado, Comput. Mater. Sci. 47 (2010) 919.[7] D. Sachdeva, Corros. Sci. 60 (2012) 18.[8] R.M. Asmussen, P. Jakupi, M. Danaie, G.A. Botton, D.W. Shoesmith, Corros. Sci.

75 (2013) 114.[9] M. Danaie, R.M. Asmussen, P. Jakupi, D.W. Shoesmith, G.A. Botton, Corros. Sci.

77 (2013) 151.[10] G.E. Coates, J. Chem. Soc. (1945) 478.[11] G.L. Makar, J. Kruger, J. Electrochem. Soc. 137 (1990) 414.[12] G.R. Hoey, M. Cohen, J. Electrochem. Soc. 105 (1958) 245.[13] F.R.F. Fan, C. Demaille, Scanning Electrochemical Microscopy, in: A.J. Bard, M.V.

Mirkin (Eds.), Marcel Dekker, Inc., 2001, p. 25.[14] J.V. Macpherson, P.R. Unwin, Anal. Chem. 69 (1997) 2063.[15] E.J. Dufek, D.A. Buttry, J. Electrochem. Soc. 156 (2009) C322.[16] K.C. Leonard, A.J. Bard, J. Am. Chem. Soc. 135 (2013) 15890.[17] A.J. Bard, F.R.F. Fan, D.T. Pierce, P.R. Unwin, D.O. Wipf, F. Zhou, Science 254

(1991) 68.[18] G. Wittstock, M. Burchardt, S.E. Pust, Y. Shen, C. Zhao, Angew. Chem., Int. Ed. 46

(2007) 1584.[19] M.V. Mirkin, W. Nogala, J. Velmurugan, Y. Wang, Phys. Chem. Chem. Phys. 13

(2011) 21196.[20] H. Shiku, T. Takeda, H. Yamada, T. Matsue, I. Uchida, Anal. Chem. 67 (1995)

312.

U.M. Tefashe et al. / Journal of Electroanalytical Chemistry 720-721 (2014) 121–127 127

[21] C. Cougnon, F. Gohier, D. Belanger, J. Mauzeroll, Angew. Chem., Int. Ed. 48(2009) 4006.

[22] Y. González-Garcı́a, G.T. Burstein, S. González, R.M. Souto, Electrochem.Commun. 6 (2004) 637.

[23] R.M. Souto, Y. González-Garcı́a, S. González, G.T. Burstein, Corros. Sci. 46(2004) 2621.

[24] K. Fushimi, T. Okawa, K. Azumi, M. Seo, J. Electrochem. Soc. 147 (2000)524.

[25] K. Fushimi, M. Seo, Electrochim. Acta 47 (2001) 121.[26] J. Izquierdo, A. Kiss, J.J. Santana, L. Nagy, I. Bitter, H.S. Isaacs, G. Nagy, R.M.

Souto, J. Electrochem. Soc. 160 (2013) C451.[27] J.J. Santana, J. González-Guzmán, L. Fernández-Mérida, S. González, R.M. Souto,

Electrochim. Acta 55 (2010) 4488.[28] A.C. Bastos, A.M. Simões, S. González, Y. González-García, R.M. Souto,

Electrochem. Commun. 6 (2004) 1212.[29] Y.F. Cheng, Green Corrosion Chemistry and Engineering, in: S.K. Sharma (Ed.),

Wiley-VCH Verlag GmbH & Co. KGaA, 2012, p. 71.[30] J.C. Seegmiller, D.S.J.E. Pereira, D.A. Buttry, D.T.S.I. Cordoba, R.M. Torresi, J.

Electrochem. Soc. 152 (2005) B45.

[31] C.-A. McGeouch, M.A. Edwards, M.M. Mbogoro, C. Parkinson, P.R. Unwin, Anal.Chem. 82 (2010) 9322.

[32] ASTM Standard B275, ASTM International, West Conshohocken, PA, 2005.[33] D. Trinh, P.D. Ducharme, U.M. Tefashe, J.R. Kish, J. Mauzeroll, Anal. Chem. 84

(2012) 9899.[34] D. Zhan, J. Velmurugan, M.V. Mirkin, J. Am. Chem. Soc. 131 (2009) 14756.[35] U.M. Tefashe, G. Wittstock, C. R. Chim. 16 (2013) 7.[36] B.B. Katemann, A. Schulte, W. Schuhmann, Electroanalysis 16 (2004) 60.[37] C. Cougnon, K. Bauer-Espindola, D.S. Fabre, J. Mauzeroll, Anal. Chem. 81 (2009)

3654.[38] G. Song, A. Atrens, Adv. Eng. Mater. 1 (1999) 11.[39] G. Williams, N. Birbilis, H.N. McMurray, Electrochem. Commun. 36 (2013) 1.[40] T.J. Davies, R.G. Compton, J. Electroanal. Chem. 585 (2005) 63.[41] G. Williams, H.N. McMurray, J. Electrochem. Soc. 155 (2008) C340.[42] T.J. Davies, S. Ward-Jones, C.E. Banks, C.J. Del, R. Mas, F.X. Munoz, R.G.

Compton, J. Electroanal. Chem. 585 (2005) 51.[43] M.E. Snowden, M.A. Edwards, N.C. Rudd, J.V. Macpherson, P.R. Unwin, Phys.

Chem. Chem. Phys. 15 (2013) 5030.[44] A. Samaniego, I. Llorente, S. Feliu, Corros. Sci. 68 (2013) 66.