j ourna l ho me page: www.elsev ier .com/ locate /e lec tac ta
etection of microbiologically influenced corrosion bylectrochemical noise transients
.M. Homborga, C.F. Leon Moralesb, T. Tingac, J.H.W. de Witd, J.M.C. Mold,∗
Royal Netherlands Navy, Naval Maintenance and Sustainment Agency, P.O. Box 10000, 1780CA Den Helder, The NetherlandsEndures BV, P.O. Box 505, 1780AM Den Helder, The NetherlandsNetherlands Defence Academy, P.O. Box 10000, 1780CA Den Helder, The NetherlandsDelft University of Technology, Department of Materials Science and Engineering, Mekelweg 2, 2628CD Delft, The Netherlands
r t i c l e i n f o
rticle history:eceived 2 April 2014eceived in revised form 19 May 2014ccepted 19 May 2014vailable online 27 May 2014
eywords:
a b s t r a c t
This work investigates the electrochemical processes involved in pitting corrosion induced by micro-biologically influenced corrosion by using time-resolved instantaneous frequency information ofelectrochemical current noise (ECN) transients obtained from Hilbert spectra. In addition to the time-frequency analyses, also the open corrosion potential is investigated and microscopic examinations ofthe specimens are performed after the tests. Hilbert spectra of the ECN signals indicated the develop-ment of transients in one of the two electrochemical cells containing sulphate-reducing bacteria with
lectrochemical noiseilbert spectralectrochemical current noise transienticrobiologically influenced corrosion
itting corrosion
a different instantaneous frequency decomposition as compared to the background ECN signal, whichresulted from the anaerobic general corrosion process. After day 13, the transients in the ECN signalsdeveloped towards consistent instantaneous frequency decompositions in the Hilbert spectra that aretypical for relatively fast pitting corrosion processes. Post-exposure microscopic observations confirmedthe existence of pits underneath the attached biofilms at the working electrodes.
Microbiologically influenced corrosion (MIC) can be describeds (the acceleration and/or alteration of) corrosion processesesulting from the presence and activities of microorganisms1], generating a biofilm at the metal surface [2,3]. It has beenocumented for metals exposed to seawater, but also to e.g.roundwater and industrial waters [4–6]. MIC therefore is a pro-ess that affects systems operating in maritime environments andany other sectors of industry. The process occurs in environ-ents where corrosion with potentially exceptionally high reaction
ates would otherwise not be expected, e.g. under anaerobic or lowhloride conditions [1,4]. Therefore, MIC can lead to unexpectedailure of systems. The process does not produce a unique type
f corrosion, but it is usually localized, inducing e.g. pitting cor-osion [1,5]. Microorganisms can accelerate the mechanisms of the
corrosion processes, for which they require water, nutrients andelectron acceptors [4,7,8].
1.1. Biofilm
Bacterial biofilms are most recognized for their influence on cor-rosion. For example, it was observed already quite some time agothat in the presence of steel, the amount of sulphate reduced to sul-phide by sulphate-reducing bacteria (SRB) increases [9]. Bacteriacan either exist individually or form colonies [1]. They grow, repro-duce and produce extracellular polymers forming a biofilm [1,3].The morphology of this biofilm depends on the surface materialand roughness [10]. Bacteria either perform aerobic or anaero-bic respiration [1,4]. These processes are symbiotic in such a waythat conditions for the existence of each species within the biofilmare facilitated [1,4]. The bacterial adhesion pattern and its extentdepend on many factors, including bacterial characteristics (e.g.their mobility in the electrolyte), substrate properties, availablenutrients, temperature and influences of electrolyte flow (which
also affects mobility of the bacteria) [1]. Moreover, the (extentof) formation of biofilms is generally not uniform and difficultto predict [1,11]. The presence of a biofilm at a metal substratecan shift the corrosion potential in the noble direction [1,11]. The
Table 1Chemical composition of the carbon steel working electrodes (wt.%).
Element Carbon steel
C ≤0.17Si –Mn ≤1.40P ≤0.045
culture of Desulfovibrio Indonesiensis (further denoted as SRB) was
24 A.M. Homborg et al. / Electro
echanism for this ennoblement is still under discussion, how-ver acceleration of the cathodic oxygen reduction reaction due toicrobial activity is a generally accepted cause [1].
.2. Corrosion mechanism
Although typical biofilm formation involves co-operationetween aerobic and anaerobic bacteria, in this work the primaryocus is on anaerobic bacteria (i.e. SRB). Under anaerobic conditionsn a neutral electrolyte, carbon steel is expected to exhibit a verylow corrosion rate due to the relatively slow cathodic reductioneaction [4]:
H2O + 2e− → H2 + 2OH− (1)
However, in the presence of SRB, corrosion is enhanced becauseulphate acts as terminal electron acceptor due to its reduction byRB [12–17]:
O2−4 + 8H+ + 8e− → HS− + OH− + 3H2O (2)
On carbon steel, the sulphide typically reacts with the iron ionsade available by the anodic reaction and subsequently the anodic
ite acidifies by the formation of iron sulphide [1,4,12,18]:
e2+ + HS− → FeS + H+ (3)
The conductive iron sulphide precipitates at the metal surface,hus facilitating electron flow from the substrate to the biofilm17,19]. Indeed, pitting corrosion attack on carbon steel in the pres-nce of SRB is reported to occur under the biofilm [20].
.3. Investigation of MIC
To investigate MIC effectively, the phenomenon should bereated from a multidisciplinary point of view. This means that elec-rochemical techniques should be combined with other (surfacenalysis) techniques [21]. The possibility to distinguish character-stic MIC signatures using electrochemical noise measurementsENM) is recognized [22,23]. It was found that differences in theype of electrochemical potential noise (EPN) signal could enableifferentiation between biological and non-biological corrosion24]. Investigation of the electrochemical noise (EN) time signalan provide useful information on the type of corrosion processnduced by MIC [23,25]. For example, parameters like characteris-ic charge and frequency of events proved valuable for this purpose23]. Analysis of the variance of the electrochemical current noiseECN) and EPN signal has been shown to reveal the moment ofransition between general and localized corrosion [26]. In therequency domain, application of fractional Fourier transform haseen reported to give satisfactory results in distinguishing MICechanisms, where conventional fast Fourier transform failed [27].
n the time-frequency domain, analysis of EPN transients by usingavelet transform can reveal signal features typical for localized
orrosion associated with the presence of a biofilm [22].In the present work the application of time-frequency analysis of
N data is proposed as an innovative way to detect and characterizeitting corrosion on carbon steel induced by SRB. The identificationf ECN transients generated by pitting corrosion through MIC is aifficult task. The specific signal characteristics of the transientshould be separated from the background ECN generated by thenaerobic general corrosion process of the carbon steel workinglectrodes. But once the transients have been identified, the abilityo analyse only the contribution of the individual pitting processes,nd to omit any instantaneous frequency information that is asso-
iated with the anaerobic general corrosion of the carbon steel, willrove to be very useful.
The approach followed in this work is as follows. The princi-le of transient analysis as introduced in earlier work [28,29] is
S ≤0.045
applied to characterize the pitting corrosion process. To promotethe metabolism of the SRB [30], the initial experimental condi-tions comprise a sterile, anaerobic, nutrient rich condition underelevated temperature, with a bare carbon steel substrate, initiallyin the absence of corrosion product. This is considered importantin the investigation of MIC induced by SRB, since any corrosionproduct present at the metal surface acts as a diffusion layer [30].For proper investigation of the development of EN characteristicsdue to the presence and activity of a bacterial biofilm, in the initialstage the working electrode surfaces should be freely accessible forspecies from the electrolyte.
In the next section, the experimental setup will be describedin more detail, as well as the microscopic investigation and theprocedure to analyse the obtained EN signal. Section 3 then dis-cusses the results, where the visual and microscopic observationsare related to the transients in the EN signal. Finally, in section 4some conclusions are drawn.
2. Experimental
2.1. Electrochemical cell
The four electrochemical cells consist of butyl rubber-stopperedcontainers with a volume of 500 ml. The measurements wereperformed in a conventional three-electrode configuration underopen-circuit conditions, requiring two nominally identical car-bon steel working electrodes and one platinum electrode, actingas reference electrode. The working electrodes consist of roundbars, protruding through the butyl rubber sealing at the top toenable electrical connection while maintaining sterility. The chem-ical composition of the carbon steel working electrodes is providedin Table 1.
The working electrodes were partly coated with araldite glue,which is resistant to the dry sterilization temperature of 190 ◦C, andacts as a corrosion protective coating. Only a well-defined area of19,6 mm2 (corresponding with a diameter of 5 mm) of each work-ing electrode was exposed to the electrolyte. The area of the squareplatinum mesh used as reference electrode was approximately 100mm2. The working electrodes were wet ground using up to 4000-grit SiC paper. The medium used was Postgate C [31], made fromdemineralised water and analytical grade reagent. NaCl was addedto the medium to make the concentration 2,5 wt.% NaCl. After flush-ing the medium with N2 for 1 hour, the medium was sterilized byautoclaving. Subsequently, the medium was allowed to cool downin a glovebox under controlled N2 atmosphere. The electrochem-ical cells were mounted together and dry sterilized at 190 ◦C for2 hours, to prevent the formation of corrosion product at the work-ing electrodes due to autoclaving. After immediate transfer to theglovebox and cooling down, 500 ml of the medium was added toeach electrochemical cell. Subsequently, 5 ml of a freshly grown
added to two of the four electrochemical cells. The other two cellsserved as sterile controls. The electrochemical cells were stored inan incubator at 28 ◦C. No additional nutrients were added duringthe experimental series.
Fig. 2. Electrochemical cells after 16 days: (a) Con
a Acta 136 (2014) 223–232 225
2.2. Bacterial density, activity and dissolved oxygen
The bacterial density of the media containing SRB was estimatedby the plate count technique. Sample media were extracted fromthe electrochemical cells under sterile conditions, using needlesprotruding the butyl rubber sealing. Dilutions were made and bac-terial counts per ml were calculated by multiplying the average(manually counted) number of colonies per plate by the reciprocalof the dilution factor used. In the sterile media, the transparency (i.e.absence of turbidity) was considered as indicative for the absenceof microbial contamination. This was verified afterwards by micro-scopic investigation of all working electrode areas.
Bacterial activity was monitored after each ENM by measur-ing the amount of dissolved sulphides, according to the procedureas introduced by Cord-Ruwisch [32]. This method photometri-cally quantifies precipitation of CuS by adding a fixed amount ofthe medium to a copper reagent (consisting of 50 mM HCl and5 mM CuSO4). Furthermore, microscopic investigation of samplesshowed a mixture of small, single SRB with high mobility andincreasingly long strings of interconnected SRB that are much lessmobile. Monitoring the mobility and development of strings ofSRB was performed regularly and considered as a basic additionalverification of the culture’s condition. Finally, the developmentof ECN transients, together with ennoblement of the correspond-ing working electrodes and the existence of pits underneath abiofilm was considered proof of the results of this microbial activ-ity.
To ensure that anaerobic conditions were retained throughoutthe measurement series, the presence of dissolved oxygen in theelectrolyte was detected afterwards. This was done by immersion ofoxygen indicator papers in each electrochemical cell, after openingthem in a glovebox under controlled N2 atmosphere. The presenceof dissolved oxygen was also checked by continuous inspection of
the ECN signals, as it can be associated to a low corrosion rate of theworking electrodes. Additionally, visual investigation of the work-ing electrodes afterwards was performed to check the formation ofcorrosion product.
trol 1, (b) Control 2, (c) SRB 1 and (d) SRB 2.
2 chimica Acta 136 (2014) 223–232
2
iAwaseaws
iso
2
aEnprmtwtmaws
ewR
tc
vstwsaoclmctsmdanbtcs
sic
neous frequencies present in these areas are averaged, in order toobtain the frequency behaviour of the pitting corrosion processes inthe ECN signal. In this way, good discrimination between different
26 A.M. Homborg et al. / Electro
.3. Microscopic investigation
Bacteria attached to the working electrode surface and form-ng a biofilm were investigated using epifluorescence microscopy.fter finishing the measurement series, the working electrodesere gently washed with sterile, anaerobic water (to retain only the
ttached biofilm) and their surface was wetted using a DNA-specifictain (SYTO® 9, �ex,max. 485 nm, �emm,max. 498 nm). After 5 minutes,xcess stain was removed and the samples were observed undern epifluorescence microscope. When SYTO® 9 comes into contactith DNA, it shows green fluorescence caused by excitation at the
pecified wavelength.Afterwards, the biofilm was removed by rubbing and using ster-
le, anaerobic water and the samples were dried. Subsequently, theamples were microscopically inspected using a Reichert MEF4 Mptical microscope with maximum magnification of 1000x.
.4. Electrochemical noise
The electrochemical cells were placed in a Faradaic cage tovoid electromagnetic disturbance from external sources. ECN andPN signals were recorded using a Compactstat from Ivium Tech-ologies working as zero resistance ammeter (ZRA in Fig. 1) andotentiometer (E in Fig. 1), controlled by a Windows-based PCunning dedicated software. The sampling frequency used for theeasurements described in this work was 20 Hz. A low-pass fil-
er of 10 Hz (which is the Nyquist frequency at this sampling rate)as applied during data recording. It was verified that instrumen-
al noise generated by the measuring equipment did not affect theeasurements. This is described in detail in an earlier paper by the
uthors [33]. The maximum range of the zero resistance ammeteras set at 100 nA. The maximum range of the potentiometer was
et at 1 V.The data was processed using Matlab from MathWorks. The
mpirical mode decomposition and the Hilbert-Huang transformere calculated using a publicly available Matlab procedure fromilling et al. [34,35].
The measurement series was performed in triplicate and at leasthree (consecutive) measurements were performed at each SRBell.
The analysis of EN was performed by investigation of average DCalues of the last 100 s of the EPN signals and by analysing Hilbertpectra of the ECN signals. From the DC values of the EPN signalhe evolution of the open corrosion potentials (OCPs) is obtained,hereas the Hilbert spectra are used for a time-frequency analy-
is. Hilbert spectra are produced by the Hilbert-Huang transforms was first proposed by Huang et al. [36]. This transform is basedn the assumption that any nonlinear and non-stationary signalonsists of multiple characteristic scales, or intrinsic modes of oscil-ation, each superpositioned on another. These so-called intrinsic
ode functions are based on the local properties of the signal andan be identified empirically by their characteristic time scaleshrough empirical mode decomposition. The basis of the decompo-ition is derived directly from the data itself, making the empiricalode decomposition flexible and adaptive [36–41]. A detailed
escription of the application of the empirical mode decompositionnd Hilbert-Huang transform procedure for the analysis of EN sig-als under open-circuit conditions in corrosion studies is reportedy the authors in a prior work [33]. It was shown that Hilbert spec-ra enable a detailed determination of the instantaneous frequencyomposition of individual corrosion phenomena observed in the ENignal at any given moment in time.
To identify the pitting corrosion induced by SRB, the corre-ponding transients in the ECN signals are located and decomposednto their instantaneous frequencies using Hilbert spectra. The pro-ess of transient analysis is identical to the procedure as proposed
Fig. 3. Composite (a) micrograph and (b) epifluorescence micrograph of a pit at thesurface of working electrode 1 in cell SRB 1. Arrows indicate the pit boundary.
earlier [29] and consists of two steps. First the areas in a Hilbertspectrum corresponding to the occurrence of individual transientsare detected. Subsequently, only the amplitudes of the instanta-
Fig. 4. Micrograph of the same pit at the surface of working electrode 1 in cell SRB1 as visible in Fig. 3, after removal of the biofilm. Arrows indicate the pit boundary.
ig. 5. Micrograph of a pit with a depth of 5 �m at the surface of working electrode in cell SRB 1, after removal of the biofilm. Arrows indicate the pit boundary.
orrosion processes can be obtained. Moreover, any instantaneousrequency information present outside the areas of interest (associ-ted here with the anaerobic general corrosion of the carbon steel)s neglected.
. Results and discussion
In this section, first the visual and microscopic observations arereated. After that, the ECN time signals and their decompositionn instantaneous frequencies in Hilbert spectra, together with thenalysis of transients present in these ECN signals, are discussed.
.1. Visual observations
Fig. 2 shows the intact electrochemical cells after 16 days. Theontrol media (Fig. 2a and b) were still clear and no visual corro-ion product was present at the working electrodes. The two cellsontaining SRB (Fig. 2c and d) had become turbid and the coloniesf SRB were clearly visible. Also in these cells, no visual corrosionroduct was present at the working electrodes. This indicates thathe corrosion rate is low, which is due to the anaerobic conditionuring the series of experiments.
.2. Microscopic observations
Fig. 3a shows a composite micrograph (constructed by stitching separate micrographs) of a pit at the surface of working electrode in cell SRB 1, with a depth of 10 �m, determined by the opticalicroscope. Fig. 3b shows a composite epifluorescence micrograph
f the same area.The presence of SRB can be observed from the epifluorescence
icrograph, since they appear as green regions/stains. Fig. 3b there-ore shows that an area with a diameter of over 300 �m, including
Fig. 6. Example (a) micrograph and (b) epifluorescence microg
Fig. 7. Micrograph of the surface of working electrode 1 in cell SRB 2 after removal ofthe biofilm, showing specific areas of corrosion attack. Arrows indicate the bound-aries of these areas.
the pit itself, was covered with a biofilm. The thickness of theareas of biofilm observed at this working electrode varied between1-10 �m. It should be noted here that biofilm thickness is not homo-geneous. The biofilms are patchy and in many cases localized. Fig. 4shows a micrograph of the same pit at larger magnification, afterremoval of the biofilm.
The entire working electrode was covered with different areasof biofilm. The smallest areas of biofilm, i.e. those with a diameterin the order of 10 �m and consisting of a small number of indi-vidually identifiable strings of SRB, did not show pitting corrosionattack. Pitting attack (with pit depths between 3 and 10 �m) couldbe observed under larger biofilms, with diameters in the order ofseveral hundreds �m.
At working electrode 2 of cell SRB 1, pits were less developed ascompared to working electrode 1: smaller and shallower (between1 and 6 �m) pits were observed. In Fig. 5 a pit with a depth of 5 �mat the surface of this working electrode is shown. Epifluorescencemicroscopy indicated that the biofilms had diameters still in theorder of 10-20 �m and consisted of individually identifiable stringsof SRB, i.e. they had not yet developed the thickness and size as wasthe case for biofilms at working electrode 1. A micrograph contain-ing small areas of attached bacteria with diameters in the orderof 10-20 �m is shown in Fig. 6a, together with the correspondingepifluorescence micrograph (Fig. 6b).
Working electrode 1 in cell SRB 2 showed specific areas wheremore severe corrosion attack was concentrated, whereas outsidethese areas only relatively mild general corrosion attack, compara-ble to that visible on the working electrodes in the control cells, wasobserved. Fig. 7 shows a micrograph of the surface of this working
electrode, after removal of the biofilm. The corrosion attack withinthese areas was superficial, i.e. without measurable depths.
Working electrode 2 contained small pits, similar in dimensionsto the ones observed at working electrode 2 in cell SRB 1, however
raph of the surface of working electrode 2 in cell SRB 1.
icrograph of the surface of working electrode 1 in cell SRB 2.
f22i1e
fidC
u
3
sa
osta
lcoItiamv
Fg
0.0E+00
5.0E+07
1.0E+08
1.5E+08
2.0E+08
2.5E+08
0 5 10 15 20
Cells
(ml-1
)
SRB 1SRB 2
gous to the previous comment, in particular for the last days ofexposure where the accumulation of SRB becomes visible the totalamount of SRB is not properly reflected by bulk cell counts any-more. Fig. 11 provides information about the total activity of the
Fig. 8. Example (a) micrograph and (b) epifluorescence m
ewer in number in this case. Both working electrodes in cell SRB showed areas of attached biofilm, similar to working electrode in cell SRB 1, i.e. less developed than at the surface of work-
ng electrode 1 in cell SRB 1. A micrograph of working electrode in cell SRB 2 is shown in Fig. 8a, together with the correspondingpifluorescence micrograph (Fig. 8b).
The working electrodes of the two control cells showed super-cial general corrosion attack, in most cases without measurableepth. A typical view on the surface of working electrode 2 of cellontrol 2 is shown in Fig. 9.
At these working electrodes, no biofilm presence was observedsing epifluorescence microscopy.
.3. Bacterial densities and activity
The bacterial densities in the media of cells SRB 1 and 2 arehown in Fig. 10. Measurements were performed at 5, 7, 9, 13, 14nd 15 days after starting the experiment.
Fig. 11 provides the amount of dissolved sulphides in the mediaf cells SRB 1 and 2, which indicates the bacterial activity. The mea-urement of this activity was triggered by transients appearing inhe ECN signals. Therefore these measurements started at day 7,nd were repeated at days 9, 13, 14 and 15.
In Fig. 10, the densities at the last two days were significantlyower than in the preceding period, where a steady increase inell count was observed. A possible explanation is the formationf strings of multiple SRB, which were counted individually earlier.n addition, accumulation of SRB at the working electrodes and athe bottom of the two containers (visible in Fig. 2c and d) increased
n the final days. This means that counting the number of SRB in
diluted sample of the medium may no longer provide a reliableeasure for the bacterial density in the container. In Fig. 11, the
alue for cell SRB 2 at day 13 may be considered to be an outlier,
ig. 9. Micrograph of the surface of working electrode 2 in cell Control 2, showingeneral corrosion attack.
Day
Fig. 10. Bacterial densities in the media of cells SRB 1 and 2.
possibly originating from an error during sample extraction. Theamount of dissolved sulphides in the medium of a similar measure-ment series was about 1,5 mM at day 2. This indicates that, priorto the trigger in electrochemical activity at day 7, the SRB showedsubstantial increase in activity. Fig. 11 shows a larger amount ofdissolved sulphides for cell SRB 1 for all measurements. Analo-
0
2
4
6
8
10
12
14
16
18
0 5 10 15 20
Conc
entr
a�on
(mM
)
Day
SRB 1
SRB 2
Fig. 11. Dissolved sulphide concentration in the media of cells SRB 1 and 2 as afunction of time of exposure.
A.M. Homborg et al. / Electrochimic
Table 2pH values of the media of all four electrochemical cells at day 16.
Cell pH
Control 1 6,00Control 2 5,99
Sabcsc
1ttoc
3
yt
t1
ei5iemoaFctic
SRB 1 6,83SRB 2 6,80
RB, which partly occurs inside the biofilm. Therefore, although themount of SRB in the solution in cell SRB 1 is lower, it is still possi-le that local activity inside the biofilm is higher in cell SRB 1 thanompared to cell SRB 2, resulting in a larger amount of dissolvedulphides. This corresponds with the observed pitting corrosion inell SRB 1.
Table 2 shows the pH values of the media of all four cells at day6. The two cells with SRB showed a higher pH than the two con-rols, which would be expected according to equation (2). Althoughhe anodic sites acidify locally (under the biofilm), the overall pHf the media with SRB increases. The pH values of the two controlells are similar, which is also the case for the two SRB cells.
.4. Electrochemical noise analysis
In this section, the ECN signals are used for time-frequency anal-sis using Hilbert spectra. The EPN signals are used to investigatehe development of the OCPs during the entire measurement series.
Fig. 12 shows the OCPs of all measurements for the four elec-rochemical cells. The OCPs were determined by averaging the last00 s of each EPN signal.
The OCPs of the two control cells steadily increased during thentire measurement series. Initially, the OCPs of the cells contain-ng SRB also increased gradually, with similar values at days 1 and. Subsequently, both cells containing SRB showed an acceleration
n the increase in OCP, which can be explained by the ennoblementffect generated by the acceleration of the cathodic reaction due toicrobial activity [1], as was mentioned in section 1.1. The extent
f ennoblement corresponds with the observed corrosion attack,s cell SRB 1 showed pitting corrosion at both working electrodes.rom day 13 onwards, the OCPs are in a steady state. The two controlells showed an OCP of at least 100 mV lower than the cells con-
aining SRB. Already at day 1, this difference is present. Microscopicnvestigation confirmed that in this case only superficial generalorrosion attack was present at the working electrodes.
0
0.1
0.2
0.3
0.4
0.5
0.6
0 5 10 15 20
E(V
vs S
HE)
Day
Control 1Control 2SRB 1SRB 2
Fig. 12. OCPs of all measurements for the four electrochemical cells.
a Acta 136 (2014) 223–232 229
Fig. 13 shows ECN signals of cell SRB 1 at days 5, 6, 13 and 16.Transients in the negative direction originate from localized cor-rosion processes occurring at the surface of working electrode 1,whereas localized corrosion activity at working electrode 2 resultsin transients in the opposite direction. Differences in the DC levelof the ECN signals are due to the continuous anaerobic general cor-rosion process. This generates continuous changes in net anodic orcathodic behaviour of the two working electrodes.
An increase of the magnitude of the transients relative to their‘base’ signal (as visible in Fig. 13a, in the absence of transients)can be observed in time. The ECN signals of the two control cellswere always similar to the ECN signal shown in Fig. 13a. Togetherwith the microscopic observations of the working electrode sur-faces afterwards (shown in Fig. 9), this type of ECN signal isconsidered to originate from the slow, anaerobic general corrosionprocess occurring at the working electrode surfaces. In the case ofcell SRB 2 this type of ECN signal was also observed throughoutthe entire measurement series, except for a few occasions whereone or two small transients (comparable to the small transientsvisible in the ECN signal of Fig. 13b) were present in either direc-tion. However, the associated localized corrosion activity did notdevelop further within the time frame of the measurement series.This was confirmed by the microscopic observations, showing thepresence of biofilm at both working electrodes (as visible in themicrographs of Fig. 8), together with concentrated areas of cor-rosion attack at working electrode 1 (Fig. 7) and the presence ofsmall pits at working electrode 2 (comparable to the pit visiblein Fig. 5). A similar series of experiments showed pitting corro-sion at only one working electrode after 14 days of exposure.It is therefore difficult to predict pitting corrosion due to MIC,even under otherwise seemingly identical conditions. This indi-cates the value of the analysis of EN transients for the detectionof MIC.
In the case of cell SRB 1, the ECN signals at days 1 and 5 aresimilar to the ECN signal of Fig. 13a. No transients are present inthe signal. From day 6 onwards, transients appeared in most ofthe ECN signals, as visible in Fig. 13b, and developed until day 16with an increase in amplitude by approximately one order of mag-nitude (see Figs. 13c and d). From the direction of the transientspresent in the ECN signals it was anticipated that working elec-trode 1 exhibited more localized corrosion activity than workingelectrode 2. This was confirmed by microscopic observations after-wards, indicating the presence of biofilm and multiple small pits atthe surface of working electrode 1, as well as one large pit coveredwith biofilm (visible in Figs. 3 and 4). The surface of working elec-trode 2 also contained areas of attached biofilm, but only severalsmaller pits, as visible in Fig. 5. In addition, less and smaller areasof biofilm were observed here as compared to working electrode 1.Because of the similar experimental conditions, it is considered thatthe only expected variable here is the heterogeneity of the workingelectrode surface. The presence of small imperfections and inclu-sions may influence the initial attachment of SRB to the electrodesurface.
Fig. 14a shows the Hilbert spectrum of the ECN signal of Fig. 13b.The original ECN signal is displayed at the back of the figure withits relative amplitudes.
In this Hilbert spectrum, the instantaneous frequency contribu-tion of the relatively small transients is observable with respectto the overall instantaneous frequency decomposition of the back-ground signal, in between the transients. The areas of the Hilbertspectrum in between the occurrence of transients represent theanaerobic general corrosion process and no noticeable dominant
instantaneous frequencies are observed there. Figs. 14b and 14cshow the Hilbert spectra of the ECN signals of Figs. 13c and d. Theoriginal ECN signals are again displayed at the back of the figureswith their relative amplitudes.
In these Hilbert spectra the instantaneous frequency contri-ution of the transients is more pronounced than in the Hilbertpectrum shown in Fig. 14a. The transients can now clearly be dis-inguished from the areas in between the transients. Moreover,ince the contribution of the transients is especially in the higherrequency range, this difference confirms the increased localizedorrosion activity at days 13-16 as compared to the earlier period.he biofilm has developed to a greater extent in terms of areand thickness, and increased metabolic activity induces more pro-ounced pitting corrosion.
In order to investigate the instantaneous frequency decom-osition of the transients in the ECN signal in more detail,wo-dimensional representations of the Hilbert spectra are used,s described in section 2.4. Fig. 15 shows the two-dimensionalepresentations of Hilbert spectra from four sets of four ECN sig-als of cell SRB 1 measured at days 13 (a), 14 (b), 15 (c) and 16d), respectively. At each day, the four ECN signals were measuredonsecutively.
The comparison of the average instantaneous frequency decom-osition of the transients present in the ECN signals allows
nvestigation of the development of the underlying localized cor-osion processes. At day 13, the distribution of instantaneous
at days 5 (a), 6 (b), 13 (c) and 16 (d).
frequencies varies between the four different ECN signals. Twoof the spectra show a higher contribution in the frequency rangebelow 10−2 Hz (at the right-hand side of the plot).
In the following days, the instantaneous frequency decompo-sition of the transients in the ECN signals gradually reaches adistribution that was also observed for metastable pitting processeson AISI304 in earlier work [28]. A maximum is observed just above10−1 Hz and the contribution tends towards zero in the frequencyregions above 1 Hz and below 10−2 Hz. In the AISI304 case, pitsinitiated and briefly showed metastable growth, after which theyrepassivated again soon thereafter. Combined with the microscopicobservations in the present work, it is considered that this instanta-neous frequency behaviour results from pitting processes occurringunderneath areas of attached biofilm, shielding the surface fromthe medium. Due to acidification of an anodic site underneath thebiofilm, corrosion attack is induced quickly, locally and only briefly.This can potentially occur repeatedly at the same location, due tolocal microbial activity in the biofilm. In this way, a pit can grow
relatively quickly by the occurrence of many of these individualevents.
The instantaneous frequency decomposition of the transientsin the ECN signals measured at day 16 shows a consistent
ig. 14. (a) Hilbert spectrum of the example ECN signal of cell SRB 1 at day 6 as shohown in Fig. 13c. (c) Hilbert spectrum of the example ECN signal of cell SRB 1 at da
istribution over the four different measurements, which implieshat the corrosion characteristics during these four measurementsere comparable.
The difference in electrochemical behaviour between the two
ells containing SRB observed by transient information from theCN signals and microscopic observations was also confirmed byhe OCPs visible in Fig. 12. Initially, OCPs of cell SRB 1 and 2 are
Fig. 15. Two-dimensional representations of Hilbert spectra from each time four
Fig. 13b. (b) Hilbert spectrum of the example ECN signal of cell SRB 1 at day 13 ass shown in Fig. 13d.
similar, with biofilms developing gradually at certain spots on allfour working electrodes. From day 5 onwards the OCPs of bothcells increase faster, with both cells showing an onset of localizedcorrosion activity in their ECN signals. Additionally, an increasing
difference in OCP is observed, corresponding with increasing tran-sient magnitudes of the ECN signal for cell SRB 1, originating frommore pronounced pitting corrosion.
ECN signals of cell SRB 1 measured at days 13 (a), 14 (b), 15 (c) and 16 (d).
2 chimic
4
iiliHtf
•
•
•
fbyt
R
[[[
[[
[[[
[[
[[[
[
[
[
[
[
[
[
[[
[[
[
[
[
[[38] N.E. Huang, M.C. Wu, S.R. Long, S.S.P. Shen, W. Qu, P. Gloersen, K.L. Fan, Proc. R.
Soc. London 459 (2003) 2317–2345.
32 A.M. Homborg et al. / Electro
. Conclusions
This work investigates the electrochemical processes involvedn pitting corrosion induced by MIC by applying transient analysisn ECN signals. It has been demonstrated that the evolution of theocalized corrosion processes can be monitored by time-resolvednstantaneous frequency information of ECN transients throughilbert spectra of the ECN signals. This was combined with inves-
igation of the OCP and microscopic observations afterwards. Theollowing results were obtained:
The immediate effect of the settlement of SRB at the surface ofthe working electrodes was already visible at day 1 by a (approx-imately 100 mV) difference in OCP between the cells containingSRB and the controls. This difference further increased during themeasurement series.From day 6, Hilbert spectra of the ECN signals indicated theoccurrence of transients with noticeable different instantaneousfrequency decomposition as compared to the background ECNsignal. In addition, the difference in OCP between the cells con-taining SRB and the two controls increased, which is an indicationof ennoblement of the working electrodes generated by the pres-ence of a biofilm.After day 13, the transients in the ECN signals of cell SRB 1developed towards a consistent instantaneous frequency decom-position in the Hilbert spectra that is typical for fast pittingcorrosion processes. Microscopic observations confirmed theexistence of pits underneath the attached biofilms at the workingelectrodes.
It can be concluded that the investigation of the instantaneousrequency decomposition of transients in the ECN signals, com-ined with monitoring of the OCP and microscopic observations,ields improved applicability of ENM for the detection and charac-erization of localized corrosion induced by MIC.
eferences
[1] B.J. Little, J.S. Lee, Microbiologically Influenced Corrosion, John Wiley & Sons,Inc, Hoboken, New Jersey, U.S, 2007.
[4] D.A. Jones, Principles and Prevention of Corrosion, Second Edition ed., Prentice
Hall, New Jersey, U.S.A, 1996.[5] T.S. Rao, T.N. Sairam, B. Viswanathan, K.V.K. Nair, Corros. Sci. 42 (2000)
1417–1431.[6] R. Melchers, R. Jeffrey, Electrochim. Acta 54 (2008) 80–85.
[[[
a Acta 136 (2014) 223–232
[7] C.-G. Peng, J.K. Park, Wat. Res. 28 (1994) 1681–1692.[8] B. Little, J. Lee, R. Ray, Biofouling The Journal of Bioadhesion and Biofilm
Research 23 (2007) 87–97.[9] R. Cord-Ruwisch, F. Widdel, Appl. Microbiol. Biotechnol. 25 (1986) 169–174.10] A. Richter, R. Smith, R. Ries, Appl. Surf. Sci. 144–145 (1999) 419–424.11] B.J. Little, J.S. Leeb, R.I. Ray, Electrochim. Acta 54 (2008) 2–7.12] B.W.A. Sherar, I.M. Power, P.G. Keech, S. Mitlin, G. Southam, D.W. Shoesmith,
Corros. Sci. 53 (2011) 955–960.13] T. Gu, K. Zhao, S. Nesic, Corrosion (2009) 1–12.14] D. Xu, T. Gu, Bioenergetics explains when and why more severe MIC pitting by
SRB can occur, in: NACE International, 2011.15] T. Gu, D. Xu, Demystifying MIC Mechanisms, in: NACE International, 2010.16] T. Gu, J. Microb. Biochem. Technol. 4 (2012) iii–vi.17] D. Enning, H. Venzlaff, J. Garrelfs, H.T. Dinh, V. Meyer, K. Mayrhofer, A.W. Hassel,
M. Stratmann, F. Widdel, Environ. Microbiol. 14 (2012) 1772–1787.18] W.A. Hamilton, Biofouling 19 (2003) 65–76.19] H. Venzlaff, D. Enning, J. Srinivasan, K.J.J. Mayrhofer, A.W. Hassel, F. Widdel, M.
Stratmann, Corros. Sci. 66 (2013) 88–96.20] L.C. Xu, H.H.P. Fang, K.Y. Chan, J. Electrochem. Soc. 146 (1999) 4455–4460.21] W. Wang, J. Wang, H. Xu, X. Li, Mater. Corros. 57 (2006) 800–804.22] T. Whitham, S. Huizinga, Evaluation of Electrochemical Noise Analysis as an
On-Line Monitoring Tool to Distinguish Between Biofilm-Associated LocalisedCorrosion and Oxygen Corrosion, in: D. Thierry (Ed.) Aspects of MicrobiallyInduced Corrosion, 1997, pp. 89-102.
24] W.P. Iverson, L.F. Heverly, Corrosion Monitoring in Industrial Plants UsingNondestructive Testing and Electrochemical Methods, ASTM STP 908 (1986)459–471.
25] A. Padilla-Viveros, E. Garcia-Ochoa, D. Alazard, Electrochim. Acta 51 (2006)3841–3847.
26] M.J. Hernández Gayosso, G. Zavala Olivares, N. Ruiz Ordaz, C. Juárez Ramirez,R. García Esquivel, A. Padilla Viveros, Electrochim. Acta 49 (2004) 4295–4301.
27] N. Zaveri, R. Sun, N. Zufelt, A. Zhou, Y. Chen, Electrochim. Acta 52 (2007)5795–5807.
28] A.M. Homborg, T. Tinga, X. Zhang, E.P.M. van Westing, P.J. Oonincx, G.M. Ferrari,J.H.W. de Wit, J.M.C. Mol, Electrochim. Acta 104 (2013) 84–93.
H.H. Liu, Proc. R. Soc. London, 454 (1998) 903–995.37] N.E. Huang, Z. Shen, S.R. Long, Annu. Rev. Fluid Mech. 31 (1999) 417–457.
39] Z. Wu, N.E. Huang, Proc. R. Soc. London 460 (2004) 1597–1611.40] M. Feldman, Mechanical Systems and Signal Processing 25 (2011) 735–802.41] P.F. Pai, Journal of Sound and Vibration 321 (2009) 305–318.
![ENTERGY NUCLEAR MANAGEMENT MANUAL ENN-DC-149 · - Erosion YOINm(3) Flow-accelerated corrosion YIT]/Nm(4) Ym/N[I](17) Microbiologically influenced corrosion (MIC) Selective leaching](https://static.documents.pub/doc/80x56/60843f4e021bfb1fe738e495/entergy-nuclear-management-manual-enn-dc-149-erosion-yoinm3-flow-accelerated.jpg)
