Bioelectrochemistry 97 (2014) 15–22

Contents lists available at ScienceDirect

Bioelectrochemistry

j ourna l homepage: www.e lsev ie r .com/ locate /b ioe lechem

Multi-technique approach to assess the effects of microbial biofilmsinvolved in copper plumbing corrosion

Ignacio T. Vargas a,b,⁎, Marco A. Alsina b, Juan P. Pavissich b, Gustavo A. Jeria b, Pablo A. Pastén a,b,Magdalena Walczak c, Gonzalo E. Pizarro a,b

a Centro de Desarrollo Urbano Sustentable (CEDEUS), Pontificia Universidad Católica de Chile, Santiago, Chileb Departamento de Ingeniería Hidráulica y Ambiental, Pontificia Universidad Católica de Chile, Santiago, Chilec Departamento de Ingeniería Mecánica y Metalúrgica, Pontificia Universidad Católica de Chile, Santiago, Chile

⁎ Corresponding author at: Av. VicuñaMackenna 4860,Tel.: +56 2 26864218; fax: +56 2 2354 5876.

E-mail address: itvargas@ing.puc.cl (I.T. Vargas).

1567-5394/$ – see front matter © 2013 Elsevier B.V. All rihttp://dx.doi.org/10.1016/j.bioelechem.2013.11.005

a b s t r a c t

a r t i c l e i n f o

Article history:Received 15 December 2012Received in revised form 19 November 2013Accepted 26 November 2013Available online 4 December 2013

Keywords:CopperMICBiofilmElectron microscopyGI-XRD

Microbially influenced corrosion (MIC) is recognized as anunusual and severe type of corrosion that causes costlyfailures around the world. A microbial biofilm could enhance the copper release from copper plumbing into thewater by forming a reactive interface. The biofilm increases the corrosion rate, the mobility of labile copper fromits matrix and the detachment of particles enriched with copper under variable shear stress due to flow condi-tions. MIC is currently considered as a series of interdependent processes occurring at the metal–liquid interface.The presence of a biofilm results in the following effects: (a) the formation of localized microenvironments withdistinct pH, dissolved oxygen concentrations, and redox conditions; (b) sorption and desorption of labile copperbonded to organic compounds under changing water chemistry conditions; (c) change in morphology bydeposition of solid corrosion by-products; (d) diffusive transport of reactive chemical species from or towardsthemetal surface; and (e) detachment of scale particles underflowconditions. Using amulti-technique approachthat combines pipe and coupon experiments this paper reviews the effects of microbial biofilms on the corrosionof copper plumbing systems, and proposes an integrated conceptual model for this phenomenon supported bynew experimental data.

© 2013 Elsevier B.V. All rights reserved.

1. Introduction

During the last decades, corrosion of metallic piping systems fordrinking water has been considered relevant because of health risksrelated to the release of metals into drinking water, general failure ofthe piping system, and the costs associatedwith repairing infrastructuredamaged by corrosion [1]. Due to its relatively low price and durability,copper is the plumbing material of choice for household drinkingwatersystems [2]. Corrosion of copper pipes proceeds as an electrochemicalreaction, where metallic copper is oxidized to copper ions (anodichalf-reaction), while an oxidizing agent is reduced, typically dissolvedoxygen (DO) (cathodic half-reaction) [3]. The water chemistry andoperational conditions of the piping system control the formation/dissolution of scales and the release of copper into the water [2].Microbially influenced corrosion (MIC) has been identified as an unusu-al and severe type of corrosion that appears in the form of pitting(known as Type 1½), and occurs mainly in soft water supplies in thepresence ofmicroorganisms [1,4].Microbial colonization and the subse-quent biofilm development induce drastic changes in the vicinity of theinorganic environment formed on themetallic surface [1,5]. This type of

7820436Macul, Santiago, Chile.

ghts reserved.

corrosion has caused failures in institutional buildings in several coun-tries around the world, including the United Kingdom, Germany,Saudi Arabia, the USA and Chile [1]. The cost of replacing the damagedpiping can be as high as 100 million, as estimated for a Scottish hospitalwhere the entire piping system collapsed [1].

Relevant advances in the understanding of MIC were accomplishedduring the 1980s as a response to strong industrial demand to mitigaterelated failures. Several research groups worked on unraveling MICmechanisms, including the effect of microorganisms on the corrosionprocesses. By the end of the 1990s and the beginning of the 21st centu-ry, the development of new techniques for surface analysis and electro-chemical characterization have helped to elucidate the function ofbiofilms as reactive interfaces that control the mobility of metallicspecies between themetal and the liquid [6]. Recent advances inmolec-ular biology, including culture-independent techniques constitute apowerful analytical tool and a next step for understanding MIC. Theseadvances allow the determination of microbial communities, and theformulation of a conceptualmodel that includesmetal–microbe interac-tions [6–9].

Amicrobial biofilm is basically composed ofmicroorganisms (mainlybacteria) embedded in a heterogeneous matrix of extracellular poly-meric substances (EPS) that facilitate the attachment of themicrobial com-munity growing on the aggressive environment existing at the metal–liquid interface. EPS are a complex mixture of various macromolecules

16 I.T. Vargas et al. / Bioelectrochemistry 97 (2014) 15–22

(i.e. proteins, polysaccharides, lipids, nucleic acids and amphiphilicpolymers), whose nature depends on the specific microbial species inthe biofilm, and relevant environmental factors such as nutrient avail-ability and environmental stress [10]. The reactivity of the biofilm tocopper is determined by the capability of EPS and the bacterial cellsitself to bind copper ions (sorption) and retain copper particles pro-duced as a consequence of the precipitation of corrosion by-products[11,12]. This patchwork of bacteria, EPS, and solid corrosion by-products form a complex reactive surface that induces metallic copperoxidation, resulting in the release of ions and/or particulate corrosionby-products into the water during the operational flow–stagnationcycle.

The study of MIC requires the integration of several analytical andmodeling techniques contributed by a number of study fields, includingmicrobiology, surface and interface science, electrochemistry, waterchemistry, and fluid mechanics. The complexity of the processes in-volved in MIC poses considerable difficulty for a robust simulationand, consequently, the existing models of MIC are not yet complete tothoroughly describe the interactions between the biofilm and the corro-sion by-products under flow–stagnation cycles.

Current models consider MIC as a series of simultaneous or succes-sive processes. The proposed mechanisms include the following effectsof the biofilm matrix: (a) formation of localized microenvironmentswith different pH, DO concentrations, and redox conditions [6,13,14],(b) sorption and desorption of labile metal bonded to organic com-pounds under changing water chemistry conditions [13,15], (c) changeinmorphology of the solid corrosion by-products [6], (d) diffusive trans-port of chemical species at the metal surface [6], and (e) detachment ofscale particles from the pipe surface underflow conditions [6,13].Whileabundant information related to the first two mechanisms (a, b) isavailable in the literature [6,13–15], we consider that the experimentalevidence to support the remainingmechanisms (c–e) is still insufficient.

The study of the biofilm effects should consider the local propertiesof the material where the biofilm is growing. Copper is a scale-formingmaterial that, under abiotic conditions, shows moderate corrosionrates [2]. Dissolved copper initially released from the metallic surfaceleads to precipitation of corrosion by-products on the inner surface ofthe pipe, thus passivating the surface and decreasing the concentrationof dissolved copper in about an order of magnitude compared to theinitial dissolved copper released from the metallic surface [2]. In thepresence of a biofilm, passivation is prevented and EPS can accumulatethe released copper. Moreover, the biofilm heterogeneity could alsoinduce particle detachment that, in case of a scaling forming material,may highly increase the mass of copper released into the bulk water[13]. Under certain conditions, particle detachment increases the cop-per concentration of the bulkwater to concentration up to 20 mg/L, giv-ing the water a blue–green coloration [16]. Such episodes are knownas “blue water” which is associated with microbial activity [5,16–18].Non-scale forming materials (i.e. stainless steel) also form passivefilms, but in those cases corrosion is so slow that high concentrationsof metallic ions never accumulate and the presence of a biofilm mayresult in different effects. Due to its antimicrobial properties, copperhas currently been used to replace stainless steel, aluminum and plas-tic touch surfaces designed for human contact (e.g. door handles,bathroom fixtures, or bed rails) and reduce hospital microbial contam-ination [19]. Thus, a conceptual model of MIC in copper should includethe interaction between corrosion by-products and the biofilm, the EPSmatrix reactivity, and the effect of copper toxicity on the microbialecology.

In this paper, we review theMIC mechanisms related to copper by acombination of traditional tests (e.g. electrochemistry) and new ap-proaches used by our group (e.g. in-situ DO measurements) to accountfor the effect of microorganisms on corrosion. Here, we present newexperimental evidence based on results obtained in laboratory tests(e.g. pipes and metal coupons) and field campaigns performed in amodel household drinking water distribution system affected by MIC.

We finally propose a new conceptualization of MIC that may be usedas a more thorough framework for the development of mathematicalmodels of copper corrosion.

2. Materials and methods

2.1. Pipe experiments

Pipe experiments were conducted on copper pipes with an internaldiameter of 1.95 cm (3/4″) (type L, manufactured by MADECO-Chileunder ASTM-B88 standard). The pipeswere extracted from a householdsystem affected by MIC [13] and tested under laboratory conditionsusing tap water from the same location. Additionally, abiotic experi-ments were prepared using synthetic MilliQ water, sodium bicarbonate(ACS grade, 99.7% pure, Merck KGaA, Darmstadt, Germany) and nitricacid (p.a. EMSURE® ISO, 65% pure, Merck KGaA, Darmstadt, Germany)to adjust the pH (tap water conditions).

2.1.1. DO consumption experimentsDO consumption experiments were carried out under biotic

(pH = 6.0, HCO3− = 1.6 mM) and abiotic conditions (pH adjusted to

6.0 ± 0.1, [HCO3−] of 0, 1.6, 3.0, and 7 mM). DO concentration was

continuously measured during stagnation according to the methodolo-gy developed in our previous works using both a 2 mm diameter FiberOptic Oxygen Sensor (FOXY,OceanOptics, Inc.) [20] and a LuminescenceDissolved Oxygen (LDO) sensor connected to a HQ40d multimeter(HACH Company, Loveland, CO, USA) [21]. Fluorescence sensors wereinserted in the bulkwater at one end of the tested pipe, while the oppo-site end was closed with a rubber stopper.

2.1.2. Flushing testsFlushing experimentswere carried out in accordance to themethod-

ology presented elsewhere [13,22]. The laboratory testswere conductedon 1 m long copper pipes of 0.3 L volume. Pipes under biotic and abioticconditionswere previously aged throughwater changes three times perweek (Monday,Wednesday and Friday) [23] using syntheticwaterwith3.2 mM [HCO3

−], and water obtained from the field [13], respectively.After 4 weeks of aging and 8 h of stagnation, pipes were flushed at aflow rate of 0.24 L/min at 25 °C, to simulate the behavior of domesticpipe systems under laminar conditions (Re = 280.44). During flushing,samples of 15 mL for the first 0.6 L, and 100 mL for the remaining vol-ume, were taken to determine copper concentration until approximate-ly 6 L of water was flushed through the pipe. An ICP/OES spectrometer(Thermo Electron iCAP 6300) was used to measure the concentrationof copper released into the bulk water. The mass of copper releasedwas estimated by integrating the copper concentration versus the vol-ume of water extracted from the pipe.

2.1.3. Surface analysisScanning electron microscopy (SEM, LEO 1420VP) with elemental

analysis (EDS, Oxford 7424 solid-state detector) was used to study themorphology and composition of the biofilm matrix and the corrosionscales formed on the inner surface of the pipes. Several 1 × 1 cm cou-pons were cut from (i) new pipes tested in the laboratory under bioticand abiotic conditions, and (ii) used pipes extracted from the modelhousehold system affected by MIC. The samples were kept hydratedwith the field water before preparation for microscopic examination.The coupons were treated with critical point drying [24] and coatedwith a thin gold film. Transmission electron microscopy (TEM, PhillipsTecnai 12 Bio Twin) was used to determine the presence of colloidalmaterial (b0.45 μm). The particles were captured through a three stepprocedure, adapted from the methodology developed by Lienemannet al. [25] for the examination of colloids. First, the water sample(50 mL) was centrifuged at 500 rpm (RCF = 27 ×g) for 5 min toexclude particles above the colloidal size. After this preliminary centri-fugation, the supernatant was carefully withdrawn and 5 mL of the

Table 1Composition of water extracted from the studied location affected bymicrobial influencedcorrosion [13].

Parameter Unit Concentration

Copper mg Cu/L 0.07Iron mg Fe/L 0.045Chloride mg Cl2/L 15.6Total alkalinity mg CaCO3/L 80Total hardness mg CaCO3/L 160Sulfate mg SO4

−2/L 30.2Nitrate mg NO3

−–N/L 3.2Phosphate mg PO4

−3/L 1.66pH – 6Dissolved oxygen mg O2/L 8.11Conductivity μS/cm 235Dissolved inorganic carbon (DOC) mg/L 1.5Magnesium mg Mg/L 6.44

17I.T. Vargas et al. / Bioelectrochemistry 97 (2014) 15–22

sample was extracted from the bottom of the centrifugation tube, resus-pended with MilliQ water, and centrifuged at 6000 rpm (RCF = 3823 ×g)for 30 min. After centrifugation, the sample was filtered using a0.45 μm pore-size acetate cellulose membrane. Finally, an aliquot ofthe filtered water was deposited over a hydrophilic resin (Nanoplast).

2.2. Coupon experiments

Electrochemical experiments to estimate corrosion rates under biot-ic and abiotic conditions were conducted using pretreated coppercoupons (99.99% purity) of 2.5 cm2 surface area. Pretreatment includeda 10 minute immersion in a 1:4 HCl/distilledwater solution followed byseveral washes with distilled water until reaching circumneutral pH.After pretreatment, coupons were aged immediately in 100 mL glassbottles filled with water extracted from the study location (Table 1).Bottles were incubated in triplicate for 150 days in a shaking incubatorat 25 °C and 25 rpm. The abiotic controls were aged using the samewater after autoclaving.

Grazing incidence X-ray diffraction (GI-XRD) [26] was used to ana-lyze the solid phase formed on the tested copper coupons. GI-XRDmea-surements were performed using a Bruker D2 Phaser diffractometerwith a Lynxeye detector and a Cu Kα radiation. The scans were mea-sured from 10° to 80° with a step size of 0.02°.

Electrochemicalmeasurementswere conducted using a potentiostat(Gamry, Reference 600). Corrosion rates of copper coupons were esti-mated by linear polarization using the Echem Analyst Software(Gamry, Philadelphia, PA, USA). Experiments were performed using a50 mL homemade electrochemical cell. A platinum wire (CHI115, CHInstruments Inc.) was used as counter electrode and Ag/AgCl (CHI111,

Fig. 1. SEMmicrographs of a copper pipe surface extracted from a 2 year old system affected bymatrix. Large (a) and small (b) pits (2500× of magnification). (c) Bacteria and minerals near t

CH Instruments Inc.) as reference electrode. Autoclavedwater extractedfrom the field location was used as electrolyte.

Epifluorescencemicroscopy (Olympus CX31)was used to detect thepresence of a biofilm on the tested coupons. Micrographs were takenusing an attached digital camera. Sample preparation consisted in cellfixation with ethanol and subsequent staining of nucleic acids with 1%v/v Acridine Orange for 20 min.

3. Results and discussion

Results below are organized to discuss the biofilm effects on thecorrosion processes (mechanisms (a) through (e)) of drinking watercopper pipes.

3.1. Microenvironments

SEM micrographs in Fig. 1 show the inner surface of a copper pipeaffected byMIC. The presence of a biofilm togetherwith corrosion scalesforms a heterogeneous matrix that may result in the formation of mi-croenvironments on the surface due to spatial chemical gradients.Fig. 1a–b shows botryoidal scales, observed by light microscopy asgreenish solid phases and identified through X-ray absorption spectros-copy as mainly being malachite (Cu2CO3(OH)2) [13] a carbonate hy-droxide corrosion by-product with passivating properties. Within themalachite, well defined pits were observed. The presence of pitssurrounding the corrosion by-products strongly suggests the formationof anodic and cathodic zones where the metallic copper oxidation andoxygen reduction take place, respectively. The diffusional resistanceand metabolic activity of a mature biofilm matrix (microorganisms,EPS, and inorganic solid phases) limit the diffusion of oxygen to cathodicareas and diffusion of ions from anodic areas [6]. This kind of patchworkarrangement was observed by Geesey et al. [4], and described as an un-usual type of pitting corrosion named corrosion type ½. In a studyassessing the interactions between microorganisms andminerals, Littleet al. [26] suggested that the presence of a biofilm maintains diversechemical microenvironments at the metal–liquid interface, which maybe completely different from the bulk liquid in terms of pH, DO, andorganic and inorganic species [27]. Therefore, thermodynamic calcu-lations based on the bulk chemistry may not be adequate to predictthe precipitation and solid phase replacement reactions (e.g., cuprichydroxide (Cu(OH)2) to tenorite (CuO) or malachite) that occur at themetal–liquid interface [27]. It is likely then that a complex matrix of bi-ological and inorganic phases, like the one observed in Fig. 1c, increasesthe physical heterogeneity and chemical heterogeneity of the surfaceproducing microenvironments that facilitate the formation of

MIC. Images reveal physical heterogeneity of the surface attributed to the biofilm–mineralhe pit (10,000× of magnification).

Fig. 2. SEM micrographs of the biofilm growing on a 2 year old copper pipe affected byMIC. Summary of EDS analysis for the EPS structures (a): C = 63 wt.%, O = 16 wt.%,Cu = 21 wt.%; the bacteria (b): C = 65 wt.%, O = 6 wt.%, Cu = 30 wt.%; and beneaththe biofilm (c): C = 27 wt.%, O = 2 wt.%, Cu = 71 wt.%.

18 I.T. Vargas et al. / Bioelectrochemistry 97 (2014) 15–22

electrochemical cells, enhancing copper oxidation and the consequentincrease of copper released into the bulk water.

3.2. Sorption of soluble copper

As a reactive barrier, biofilms can sequester and accumulate labilecopper. This process has the potential to disturb the chemical equilibri-umand to induce high copper concentrations into thebulkwater duringflushing events. In general, the biofilm matrix accumulates copper bytwomechanisms: copper ions are bound to EPS and cells during stagna-tion periods (sorption), and copper phases are enriched in the biofilmmatrix through the formation of corrosion by-products. Furthermore,corrosion by-products may also sorb copper ions, although this couldbe more difficult to observe and advanced synchrotron spectroscopictechniques would be needed to quantify this process. Fig. 2 shows abiofilm formed on the inner surface of a 2 year old copper pipe. EDSanalysis suggested that both microorganisms and EPS matrix accumu-late soluble copper by mineralization and sorption of labile copper.Analysis of the EPS structures indicated an elemental composition of63% C, 16% O and 21% Cu, while on the bacterial cells showed a compo-sition of 65% C, 6% O and 30% Cu. The surface beneath the biofilmseemed to be formed basically by corrosion by-products and metalliccopper (27% C, 2% O, 71% Cu).

In a previous study [7] we performed a bacterial community analysisof the biofilm growing on copper pipes affected by MIC and concludedthat themicrobial ability, linked to biofilm growth, is a key factor relatedto the observed increase in surface heterogeneity, corrosion and theoccurrence ofmicroenvironments and bacterial survival. Using bacterial

Fig. 3. SEM micrographs (10,000× of magnification) of malachite structures forme

isolates obtained in this previous work [7], we performed solublecopper sorption tests. Preliminary results have shown sorption capaci-ties of as much as 18 mg Cu/mg of dry biomass (unpublished data).These initial results suggest that biofilms in copper pipes may have avery high copper accumulation capacity.

3.3. Morphology of scales

Malachite mineralization produced during corrosion events com-monly displays a botryoidal morphology; although observed texturefeatures seem to vary according to the formation conditions, includingthe presence of a biofilm. Malachite formed under abiotic conditionstends to show a plumose texture, resembling “spines” [21,28]. Mala-chite formed under biofilm presence tends to show an acicular texture,with “hair-like” filaments [27]. SEMmicrographs of copper pipes affect-ed by biotic corrosion and abiotic corrosion (Fig. 3) indicate that thepresence of the microbial–EPS matrix alters the morphology of mala-chite precipitates [13,20]. This is consistent with our findings in coppercorroded pipes, while “hair-like” filaments were observed under bioticconditions (Fig. 3a), the formation ofmorphologies with columnar crys-tal habits was observed under abiotic conditions (Fig. 3b). Interestingly,under biotic conditions malachite maintains a botryoidal configuration(Fig. 3c) found also for abiotic conditions. Thus, this “hair-like” mor-phology of malachite together with the observed patchwork betweenthe biofilm and precipitates form a film that, under the shear stressassociated with the pipe flow, could increase the release of particlesinto the bulk water.

3.4. Transport of chemical species

Passivation of the metallic surface by accumulation of corrosionby-products controls the transport of ions to the bulk water, limitingthe rate of copper oxidation [29]. However, microbial local acidificationinduces chemical weathering and dissolution of solid phases such astenorite or malachite, thus forcing mineral replacement reactions that,in turn, enhance copper release into the bulk [27].

GI-XRD analysis of coupons exposed for 20 days to water extractedfrom the study field suggested solid phase replacement due tomicrobialactivity at the metal surface. X-ray diffraction data identifies cuprite(Cu2O) and tenorite on the surface in the biotic experiments, but onlytenorite in the abiotic ones (data not shown). This result suggestedchanges in the local chemistry at the solid–liquid interface due to thepresence of the biofilm. Presence and absence of bacteria were checkedusing epifluorescence microscopy. In a previous work we detected cu-pric hydroxide, a non-expectedmetastable amorphous phase, coexistingwith malachite in an old piping system affected by MIC [13]. Usingellipsometric measurements on Al-brass and Ti surface tested underbiotic and abiotic conditions Busalmen et al. [30] concluded that

d under (a) biotic conditions, (b) abiotic conditions, and (c) larger field of (a).

Fig. 4. SEM micrographs of a copper pipe surface under stagnation. (a) Biotic conditions (t = 3 months; 700× of magnification), (b) zoom of a (5000× of magnification), (c) abioticconditions (t = 1 week; 700× of magnification), and (d) biotic conditions (t = 8 h; 5000× of magnification). Circles indicate individual cells.

19I.T. Vargas et al. / Bioelectrochemistry 97 (2014) 15–22

bacteria inhibit the kinetics of oxide growth resulting in decreasedoxide film thickness. The thinning occurs probably due to a local dropof pH, along with restricted oxygen diffusion through the biofilm.

SEM micrographs of the inner surface of the pipes tested under lab-oratory conditions showed the development of a biofilm after 3 monthsof stagnation (Fig. 4a–b), where it is possible to observe that thebacteria–EPS–scales matrix compromises film homogeneity and thusinhibits the passivation of copper. The presence of a biofilm increasedthe heterogeneity and overall surface roughness compared with a sur-face tested under abiotic conditions (Fig. 4c). According to our previousstudies [20,21] abiotic conditions at longer aging times lead to passiv-ation of the metallic surface by precipitation of malachite. Fig. 4cshows that, in the absence of a biological matrix, corrosion by-productsare formed using the surface features forming the pipe manufacturingprocess as template. That is, parallel to the direction of the flow andconsistent with the direction of the pipe extrusion. The aggregationpattern observed in Fig. 4a was not observed in the biotic experiments(Fig. 4a), probably due to the heterogeneity imposed by the biofilm. Inaddition, Fig. 4d shows initial microbial colonization (open circles) on

Fig. 5. Dissolved oxygen measurements during water stagnation on copper pipes underbiotic (pH = 6.0, [HCO3

−] = 1.6 mM) and abiotic conditions (pH adjusted to 6.0 ± 0.1,[HCO3

−] of 0, 1.6, 3.0, and 7 mM, respectively).

a new copper pipe after 8 h of stagnation. The evidence of microorgan-isms attached to initial corrosion films without a supportive organicmatrix suggested that microbes and EPS are intimately entrained withthe inorganic compounds, preventing homogeneous passivation of thesurface. Previous analysis of bacterial isolates extracted from the fieldsite revealed the presence of copper resistance genes and high coppertolerance [31] in some of the most abundant bacterial groups. Thesefindings are consistent with the observation of microbial colonizationwithout a supportive organic matrix (Fig. 4d).

In-situDOmeasurements under biotic and abiotic conditions (Fig. 5)suggested that a biofilm in the inner surface of the copper pipe inducesmetallic copper oxidation. In previousworks [20,21]we used in-situ DOmeasurements in drinkingwater copper pipes to estimate rates ofmetaloxidation during stagnation in water under or without passivatingconditions due to precipitation of malachite. Notorious differenceshave been observed in DO consumption kinetics between passivatedand non-passivated surfaces where electron transfer and subsequentcorrosion take place [20,21]. Normalized DO measurements testsshow that DO consumption rates are similar in biotic and abiotic testswith negligible dissolved carbonate concentration, in which no surfacepassivation is expected. Indeed in abiotic tests with 0 mM HCO3

− andthe biotic test with 1.6 mM HCO3

−, where a biofilm–mineral matrix isformed (Fig. 2) instead of passivation, DO is depleted between 40 and50 h, following a zero-order kinetic rate law [20]. On the other hand,in abiotic tests with higher carbonate concentrations (where a passivefilm is expected to form), DO is depleted after 70 h following a first-order kinetic rate law [20]. According to our estimations the consump-tion of DO due to microbial respiration is negligible in biotic tests with1.5 mg/L of dissolved organic carbon (DOC). Hence, the physical hetero-geneity and increase in surface roughness that a biofilm imposes on theinner surface of the pipe may prevent passivation (Fig. 4a–b). We pro-pose that the measurement of DO consumption is a helpful non-destructive test to evaluate changes on surface passivation due to theestablishment of a microbial biofilm in copper pipes. This test couldbe used as an alternative or complementary tool to electrochemical ex-periments in drinking water piping systems.

Corrosion rate experiments conducted on the coupons suggestedthat the age of the biofilm is related with its effect on corrosion. Inshort-term experiments new coupons (NC) showed a corrosion rate of

(a)

(b)

246

157

433Short-term experiments (20 days)Corosion rate (mpy)

New coupon After 20 days oftreatment

After 20 days oftreatment (sterilized)

New coupon After 150 days oftreatment

After 150 days oftreatment (sterilized)

420

147

29

Long-term experiments (150 days)Corosion rate (mpy)

Fig. 6. Corrosion rates estimated by linear polarization in pretreated copper coupons(99.99% purity, 2.5 cm2 surface area) aged in water extracted from the study location.(a) Short-term experiments (t = 20 days of aging). (b) Long-term experiments(t = 150 days of aging). Experimental condition: agitation = 25 rpm; temperature =25 °C.

20 I.T. Vargas et al. / Bioelectrochemistry 97 (2014) 15–22

246 ± 7 mpy (miles per year), after 20 days of agingwhile under bioticconditions (BC) the corrosion rate decreased to 157 ± 18 mpy, andunder (AC) abiotic conditions (autoclaved water) the corrosion ratereaches 433 ± 73 mpy (Fig. 6a). These results are in agreement withour previous observations of copper release in drinking water copperpipes [7], where, after 3 weeks, passivation seems to control copperrelease. In contrast, the long-term experiments showed a decrease

Fig. 7. Electron microscopy micrographs of solid by-products from a 1 year old copper pipe affby-product particle (10,000× of magnification), (c) TEM of corrosion by-product particles N0.4

on corrosion rates for both tested conditions (NC = 420 ± 20 mpy,BC = 147 ± 47 mpy, AC = 29 ± 0 mpy) as shown in Fig. 6b. Howev-er, differences in corrosion rates after 150 days suggested that thepresence of a microbial biofilm decreased surface passivation. Hence,the age of the biofilm determines its effects on copper corrosion.Observed differences in the initial corrosion rates of NC between shortand long-term experiments could be explained due to atmosphericcorrosion during storage time.

Differences between the biofilm community structures of field pipeswith an operation ofmore than 1 year and pipes tested in the laboratoryfor 8 weeks (using samewater as inoculum and feed) suggestedmicro-bial changes and probably ecological succession over time [7]. Thesemi-crobial community dynamics probably explain the different corrosionrates observed in the coupons. While colonization and early biofilmformation may prevent corrosion, a mature biofilm could actuallyinduce it.

3.5. Particle detachment

Under flow conditions the bulk water velocity field imposes me-chanical stress on the pipe surface, mainly shear, due to the viscosityof the water [13]. When shear forces exceed the mechanical resistanceof the material, detachment of biofilm and corrosion by-products fromthe surface can occur. Sloughing of corrosion scales enhances copper re-lease into the bulk water. Although studies have reported a correlationbetween hydrodynamics and metal release from metallic surfaces, thishas focused mostly on mechanical abrasion under flow conditions [32].Recent abiotic laboratory tests in copper pipes [22] showed the detach-ment of nano and micro-particles of malachite (0.05–0.2 μm), inducedby the flow shear stress, increasing the mass of copper released intothe water.

Under abiotic conditions, themass of copper related to nano-particleswas estimated in 0.93 mg (for 4.5 L of flushed water), explaining thedifference between measurements of total and dissolved copper. Forbiotic conditions, we expect the physical heterogeneity of the biofilmto facilitate the detachment of protective corrosion by-products films.Fig. 7 shows micrographs of a 1 year old pipe affected by MIC (Fig. 7a)[33] and of detached particles released from the pipe surface, includingmicro-particles (N0.45 μm) (Fig. 7b) and nano-particles (Fig. 7c) cap-tured from the bulk water.

Laboratory flushing experiments using biotic and abiotic conditionsshowed that the mass of copper released into the water due to flowingwater is higher when biofilms are present (Fig. 8). Experiments weredone with a flow rate of 0.24 L/min, and a total flushing volume of 6 L.The mass of copper released with the first 600 mL of flushed water is

ected by MIC. (a) SEM of the pipe surface (500× of magnification), (b) SEM of a corrosion5 μm.

Fig. 8.Mass of total copper released from the pipe during flushing. Solid line: biotic exper-iments (field water), dashed line: abiotic experiments (synthetic water). Conditions:aging time = 4 weeks; stagnation time (previous to the flow) = 8 h; flow rate =0.24 L/min; temperature = 25 °C.

21I.T. Vargas et al. / Bioelectrochemistry 97 (2014) 15–22

similar under biotic (0.062 mg of Cu) and abiotic (0.058 mg of Cu) con-ditions. However, after this initial flushing volume (i.e., from 0.6 L to6 L) there is a significant difference in the mass of copper releasedunder biotic (0.0361 mg of Cu) and abiotic (0.0191 mg of Cu) condi-tions. Although themass of copper released in the flushing experimentscannot be directly linked to particle detachment, it is reasonable to as-sume that during the flushing test (approximately 25 min) the copperrelease due to other mechanisms, such as metallic copper oxidation orscale dissolution, is negligible. Thus, our results indicated that biofilmgrowth on the surface enhances release of copper due to detachmentof corrosion by-products.

3.6. Conceptual model

The integration of the revised mechanisms related with MIC ofcopper pipes and the release of copper into water are presented as aconceptualmodel in Fig. 9. This conceptualmodel includes (i) the devel-opment of microenvironments and the formation of anodes and cath-odes across the corroding pipe inner surface, and (ii) sorption ofcopper ions on the biofilm and/or mineralization on the EPS–microbialmatrix, forming a heterogeneous surface that prevents passivation,and facilitates detachment of copper particles under flow conditions.The presented conceptual model does not include microbial mecha-nisms of direct electron up-take from solid and dissolved metal phases,which have been observed for iron and carbon but not for copper.For example, iron-respiring [34], and iron-oxidizing [35] bacteria

Fig. 9.MIC conceptual model including the five co

directly transfer electrons to and from electrodes respectively. Hence,the mechanisms and the proposed conceptual model presentedhere are required to model MIC in copper drinking plumbing systemsunder flow–stagnation cycles.

4. Conclusions

This study points out that the conventional MIC theoretical conceptof a microorganisms–EPS matrix spatially separated from the corrosionby-products is not adequate to understand the effect of biofilms on thecorrosion processes and release of copper into the bulk water in plumb-ing systems under flow–stagnation cycles. Microscopic observations,and analyses of samples extracted from an actual copper piping systemaffected byMIC suggested that the presence of a biofilm plays an impor-tant role on themorphology and reactivity of the surface (i.e. solid phasereplacement, sorption, transport of chemical species). Our experimentalresults confirmed that biofilms increase the release of copper into thewater, mainly due to corrosion and particle detachment.

The evidence presented in this work emphasizes the idea thatbiofilm reactivity is not only associated with a single mechanism(e.g. sorption and desorption of labile copper). The overall effect of a bio-film involved inMIC events,with the consequent enhancement of copperconcentration in drinking water, is the sum of several mechanisms thatrequire an interdisciplinary view, integrating aspects of analyticalchemistry, water chemistry, electrochemistry, microbiology, hydro-dynamics, and mathematical modeling. Therefore, in order to studysuch complex phenomena and elaborate a model for practical uses,it is necessary to combine a number of experimental techniquesthat allow verifying the individual mechanisms and interaction be-tween them. Some of the techniques are established instrumentsof the above-mentioned disciplines but new techniques are also de-veloped, as is the case of the here presented monitoring of DO con-sumption that may serve as a non-destructive alternative to theconventional electrochemical approach.

Acknowledgments

This research was funded by FONDECYT projects 1110440/2011 and11110112/2011. This paper was presented under project CONICYT/FONDAP/15110020. Special thanks to Katherine Lizama for helpingwith the TEM sample preparation, José Tomás Corthon, Max Felis, andSara Acevedo for laboratory work and conducting electrochemical anal-ysis. The authors wish to acknowledge Dr. Bernardo Gonzalez for hisvaluable advice on microbiological analysis.

rrosion mechanisms discussed in this article.

22 I.T. Vargas et al. / Bioelectrochemistry 97 (2014) 15–22

References

[1] C.W. Keevil, The physico-chemistry of biofilm-mediated pitting corrosion ofcopper pipe supplying potable water, Water Sci. Technol. 49 (2) (2004)91–98.

[2] T.H. Merkel, S.O. Pehkonen, General corrosion of copper in domestic drinking waterinstallations: scientific background and mechanistic understanding, Corros. Eng. Sci.Technol. 41 (1) (2006) 21–37.

[3] D.J. Ives, A.E. Rawson, Copper corrosion III. Electrochemical theory of general corro-sion, J. Electrochem. Soc. 109 (6) (1962) 458–462.

[4] G.G. Geesey, et al., Unusual types of pitting corrosion of copper tubes in potablewater systems, in: G.G. Geesey, Z. Lewandowski, H.C. Flemming (Eds.), Biofoulingand Biocorrosion in Industrial Water Systems, Lewis Publishers, New York, 1994,pp. 243–263.

[5] B.J.Webster, et al., Microbiologically influenced corrosion of copper in potablewatersystems—pH effects, Corrosion 56 (9) (2000) 942–950.

[6] H.A. Videla, L.K. Herrera, Microbiologically influenced corrosion: looking to the fu-ture, Int. Microbiol. 8 (3) (2005) 169–180.

[7] J.P. Pavissich, et al., Culture dependent and independent analyses of bacterial com-munities involved in copper plumbing corrosion, J. Appl. Microbiol. 109 (3)(2010) 771–782.

[8] P.L. Waines, et al., The effect of material choice on biofilm formation in a modelwarm water distribution system, Biofouling 27 (10) (2011) 1161–1174.

[9] H.J. Jang, Y.J. Choi, J.O. Ka, Effects of diverse water pipe materials on bacterial com-munities and water quality in the annular reactor, J. Microbiol. Biotechnol. 21 (2)(2011) 115–123.

[10] H.C. Flemming, J. Wingender, The biofilm matrix, Nat. Rev. Microbiol. 8 (9) (2010)623–633.

[11] W.B. Beech, J. Sunner, Biocorrosion: towards understanding interactions betweenbiofilms and metals, Curr. Opin. Biotechnol. 15 (3) (2004) 181–186.

[12] M. Edwards, N. Sprague, Organic matter and copper corrosion by-product release: amechanistic study, Corros. Sci. 43 (1) (2001) 1–18.

[13] G.R. Calle, et al., Enhanced copper release from pipes by alternating stagnation andflow events, Environ. Sci. Technol. 41 (21) (2007) 7430–7436.

[14] A. Reyes, et al., Biologically induced corrosion of copper pipes in low-pH water, Int.Biodeterior. Biodegrad. 61 (2) (2008) 135–141.

[15] I.B. Beech, J.A. Sunner, K. Hiraoka, Microbe–surface interactions in biofouling andbiocorrosion processes, Int. Microbiol. 8 (3) (2005) 157–168.

[16] M.M. Critchley, R. Pasetto, R.J. O'Halloran, Microbiological influences in ‘blue water’copper corrosion, J. Appl. Microbiol. 97 (3) (2004) 590–597.

[17] M. Edwards, J.F. Ferguson, S.H. Reiber, The pitting corrosion of copper, J. Am. WaterWorks Assoc. 86 (7) (1994) 74–90.

[18] M. Edwards, S. Jacobs, R.J. Taylor, The blue water phenomenon, J. Am. Water WorksAssoc. 92 (7) (2000) 72–82.

[19] G. Grass, C. Rensing, M. Solioz, Metallic copper as an antimicrobial surface, Appl.Environ. Microbiol. 77 (5) (2011) 1541–1547.

[20] I.T. Vargas, et al., Influence of solid corrosion by-products on the consumption of dis-solved oxygen in copper pipes, Corros. Sci. 51 (5) (2009) 1030–1037.

[21] I.T. Vargas, P.A. Pasten, G.E. Pizarro, Empirical model for dissolved oxygen depletionduring corrosion of drinking water copper pipes, Corros. Sci. 52 (7) (2010)2250–2257.

[22] I.T. Vargas, et al., Increase of the concentrationof dissolved copper indrinkingwater sys-tems due to flow-induced nanoparticle release from surface corrosion by-products,Corros. Sci. 52 (10) (2010) 3492–3503.

[23] N. Boulay, M. Edwards, Role of temperature, chlorine, and organic matter in coppercorrosion by-product release in soft water, Water Res. 35 (3) (2001) 683–690.

[24] S. Schadler, C. Burkhardt, A. Kappler, Evaluation of electron microscopic samplepreparation methods and imaging techniques for characterization of cell–mineralaggregates, Geomicrobiol J. 25 (5) (2008) 228–239.

[25] C.P. Lienemann, et al., Optimal preparation of water samples for the examination ofcolloidal material by transmission electron microscopy, Aquat. Microb. Ecol. 14 (2)(1998) 205–213.

[26] S. Sathiyanarayanan, M. Sahre, W. Kautek, In-situ grazing incidence X-ray diffrac-tometry observation of pitting corrosion of copper in chloride solutions, Corros.Sci. 41 (10) (1999) 1899–1909.

[27] B.J. Little, P.A. Wagner, Z. Lewandowski, Spatial relationships between bacteria andmineral surfaces, in: P.H. Ribbe (Ed.), GEOMICROBIOLOGY: Interactions betweenMicrobes and Minerals, The Mineralogical Society of America, Washington, D.C.,1997, pp. 123–159.

[28] T.H. Merkel, et al., Copper corrosion by-product release in long-term stagnationexperiments, Water Res. 36 (6) (2002) 1547–1555.

[29] Y. Feng, et al., The corrosion behaviour of copper in neutral tap water. Part I: corro-sion, Corros. Sci. 38 (3) (1996) 369.

[30] J.P. Busalmen, S.R. de Sanchez, D.J. Schiffrin, Ellipsometric measurement of bacterialfilms at metal–electrolyte interfaces, Appl. Environ. Microbiol. 64 (10) (1998)3690–3697.

[31] J.P. Pavissich, et al., Bacterial diversity, copper resistance and sorption capacity char-acterization of microbially influenced corrosion in copper pipes, EUROCORR 2009,2009, (Nice, France).

[32] B. Poulson, Advances in understanding hydrodynamic effects on corrosion, Corros.Sci. 35 (1–4) (1993) 655–665.

[33] R.J. Oliphant, Causes of Copper Corrosion in Plumbing Systems, Foundation forWater Research, UK, 2003.

[34] M. Mehanna, et al., Geobacter species enhances pit depth on 304 L stainless steel in amedium lacking with electron donor, Electrochem. Commun. 11 (7) (2009)1476–1481.

[35] I.G. Chamritski, et al., Effect of iron-oxidizing bacteria on pitting of stainless steel,Corrosion 60 (7) (2004) 658–669.