Copper Tube Corrosion in Potable Water
Content
CORROSION SCIENCE SECTION
Microbiologically Influenced Corrosion of Copper in Potable Water Systems—pH Effects
B.J. Webster,* S.E. Werner, D.B. Wells,** and P.J. Bremer***
ABSTRACT
Copper tubing used in potable water plumbing systems occasionally experiences reactions that lead to the release of copper corrosion by-products into the water. The main factor controlling microbiologically influenced corrosion (MIC) of copper has been identified as a decrease in pH, which in conjunction with the incorporation of bacterially produced extracellular polymeric substances (EPS) in the copper oxide film, decreases the protective nature of the film. The biofilm (bacteria and EPS) is believed to have a secondary role to the nature of the oxide film in controlling the rate of corrosion. MIC was produced in laboratory reactors containing copper electrodes exposed to simulated potable water in the presence of a biofilm composed of microorganisms isolated from a field site. In the presence of the biofilm, small but significant reductions in pH occurred—from an initial value of 7.5 to between 6.5 and 6.9. Using electrochemical impedance spectroscopy (EIS), it was shown that the presence of a biofilm caused instances of higher corrosion rates similar to those measured in inorganic tests at pH 6.8. Modeling the oxide film as a thin barrier layer covered by a porous layer, EIS data revealed similarities between the oxide structure on samples experiencing MIC to samples exposed in sterile tests at pH 6.8. KEY WORDS: blue water, copper, electrochemical impedance spectroscopy, microbiologically influenced corrosion, pH
INTRODUCTION
Copper tubing has been used widely in potable water distribution systems because it is durable and easy to install. However, increasing numbers of reports on the release of copper by-products from the inside wall of copper tubes resulting in the appearance of blue- or green-colored water have caused concern. Manifestations of “blue water” or copper by-product release have been reported for locations in New Zealand, Australia, the United States, Japan, and Europe.1-9 In these instances, the first-drawn water can contain between 5 ppm and 300 ppm copper (as Cu2+) as fine-particulate precipitates. This copper byproduct release occurs in soft waters where the water is neutral or has a neutral-alkaline pH (as distinct from corrosion of copper under conditions of low pH). Corrosion associated with the blue water problem does not significantly compromise the tube integrity but rather results in contamination of the water. Several anomalous cases of blue water have been reported worldwide, consistent with an microbiologically influenced corrosion (MIC) mechanism.2,7,10-11 In these instances, the water was found to have a highoxygen demand, frequently was nonchlorinated, and was held stagnant in pipes for long periods of time. Corrosion damage was pitting (described as pepperpot pitting)3 and more generalized sites of corrosion that contained small pits (causing blue water).12 Investigations into MIC of copper have shown that certain species of bacteria can adhere and grow within a biofilm on copper surfaces, dispelling the earlier belief that copper ions are toxic to all bacteria.4
Submitted for publication March 1999; in revised form, April 2000. * Industrial Research Limited, PO Box 2225, Lower Hutt, New Zealand. ** Industrial Research Limited, PO Box 31-310, Auckland, New Zealand. *** Crop and Food Research, University of Otago, PO Box 56, Dunedin, New Zealand.
942
0010-9312/00/000181/$5.00+$0.50/0 © 2000, NACE International
CORROSION—SEPTEMBER 2000
CORROSION SCIENCE SECTION
Two models have been proposed to describe MIC of copper. One model suggested that the extracellular polymeric substances (EPS) generated during microbial activity created preferential cathodic sites through the cation-selective nature of the EPS.3,13 The other model described MIC of copper as involving the development of a copper ion concentration cell caused by the binding of copper by microbiallyproduced EPS and the generation of a weakly acidic environment.14 It is believed that the impact of preferential cathodes on pitting corrosion induced by MIC would be limited. Rather, for MIC to cause pitting corrosion, it would seem that any influence of microbial activity would have to be on stabilizing or accelerating the anodic reaction in much the same way as has been suggested by the alternative model of MIC of copper.14 It is possible that the data on which the preferential cathode model was suggested was influenced strongly by the development of the copper oxide film on the copper and/or inhibitory properties of the EPS. The present study has added to the model proposed by Geesey, et al., for MIC of copper by determining the relative contributions of the copper ion concentration cell/pH effect and other nonbiological parameters known to control copper corrosion (i.e., the nature of the copper oxide film).14 Others have shown that in simulated potable water systems, the structure of the copper oxide film formed at pH 7.6 is more compact and protective than that produced at pH 5.15 In this work, controlled laboratory studies using bacterial isolates from a field site that produced blue water were compared with studies in inorganic solutions of pH 6.8 and pH 8. This paper primarily covers data from a single test, although five similar tests were conducted and some of the results are reported elsewhere.16-17 In this work, electrochemical measurements and techniques were used to characterize MIC of copper.
TABLE 1 Electrolyte Composition (ppm)(A)
NaHCO3 NaCl Na2SO4 Glucose Yeast extract(C)
(A) (B)
98(B) 33(B) 29(B) 1,250 —
(C)
pH adjusted to 7.5 using dilute H2SO4. Components used in inorganic salt solutions (for tests at pH 6.8 and pH 8 and RDE analyses). Yeast extract was 200 ppm on Day 0 and thereafter 50 ppm.
EXPERIMENTAL PROCEDURES
Microbiological Tests
Epoxy-mounted copper disk electrodes (area = 0.38 cm2) and copper rotating disk electrodes (RDE) (area = 0.51 cm2) were prepared from 99.97% (phosphorus deoxidized) copper sheet and 99.99% copper rod, respectively. Electrodes were ground with wet silicon carbide (SiC) abrasive papers to a 1,000-grit finish, rinsed with deionized water, and dried in warm air before being sterilized. Electrochemical experiments were carried out in duplicate electrochemical cells, which, in addition to six copper electrodes, contained a platinum foil electrode (2 cm2) and a silver-silver chloride (Ag-AgCl)/saturated (sat.) potassium chloride (KCl)
†
Trade name.
reference electrode in a Luggin capillary filled with 0.1 M KCl (all potentials are referenced to the saturated calomel electrode [SCE]). One cell was used as a sterile control and the other cell was inoculated with bacteria. Cells and associated tubing were sterilized in an autoclave (20 mm at 121°C). Copper disks, platinum electrodes, and the empty Luggin capillary were sterilized by soaking in 70% ethanol (C2H5OH) overnight. Syringe-filter sterilized 0.1 M KCl was used to fill the Luggin capillary. Electrochemical cells were assembled and solutions added using aseptic techniques in a downflow cabinet. Each electrochemical cell was filled with ~ 800 mL of an electrolyte, which simulated the anionic composition of potable water and contained glucose and yeast extract to ensure biofilm establishment in a reasonable time-frame (Table 1). The medium was replenished by draining 90% of the spent medium and pumping fresh medium from a 20-L reservoir into the cell; the effluent medium was analyzed for microbiological composition and pH. The first medium exchange occurred on Day 15 and subsequent medium changes generally were conducted every 7 days to 10 days. Electrochemical cells were equilibrated to the atmosphere through a microbiological air filter but were kept in the dark to eliminate photoelectrochemical effects.18 The test cell was inoculated aseptically with a 2-mL subculture of microorganisms at Day 0. The subculture contained five isolates from a field site that was experiencing blue water. All of the isolates were capable of growth on copper surfaces. Further details about the microbes are given elsewhere.16 Planktonic bacterial numbers were measured from fluid removed at each medium exchange. Serial dilutions (0.1% peptone) were plated out in triplicate onto either standard plate count (SPCA, Oxoid†) or R2A (Difco†) agar plates and incubated at 25°C for 7 days. Isolates were differentiated on the basis of colonial morphology, counted, and recorded. At the end of the test, electrodes were stained with a solution of acridine orange and examined for biofilm formation using an epifluorescent microscope. pH was measured on fluid removed from the reactor during media exchanges. The pH probe was calibrated using pH 4.00 ± 0.02 and pH 7.00 ± 0.02
CORROSION—Vol. 56, No. 9
943
CORROSION SCIENCE SECTION
refreshed over the duration of the test. Corrosion potential and EIS measurements were conducted over the duration of the tests. These tests were not conducted aseptically but were conducted in electrolyte free of glucose or yeast extract. (a)
RESULTS
Microbiological Data
Planktonic bacterial numbers in the inoculated cell were ≈ 1 × 107 colony forming units (CFU) per mL for the duration of the test to confirm microbiological activity. The control cell remained sterile to Day 87, which is the period that the majority of the information reported here was recorded. Examination of the surface of the control electrodes by epifluorescence microscopy at the end of the test (Day 130) revealed the absence of bacteria, despite contamination of the bulk fluid. Examination of electrodes from the test cell revealed a surface covered with a continuous layer of cells that were typical of the blue water isolates. The biofilm had a honeycomb structure that contained pores, and bacteria were evident within sites of corrosion.
(b)
FIGURE 1. Equivalent circuit used to model EIS data of the copper electrodes: (a) circuit that includes diffusion of copper ions through the oxide film and (b) a circuit, which includes components that describe the film as a porous oxide.
buffers. Accuracy of the measurements were limited by buffer solutions and was ± 0.02. Electrochemical measurements made over the duration of the test were corrosion potential, redox potential, and electrochemical impedance spectroscopy (EIS). A PARC EG&G 273A† potentiostat, M5210† lock-in amplifier, and M388† electrochemical software were used for all electrochemical measurements.19 EIS measurements were conducted at the corrosion potentials over a frequency range from 10 kHz to 5 mHz, using an alternating current (AC) amplitude of ± 10 mV. EIS data were modeled using EQUIVCRT† to the equivalent circuits given in Figure 1.20 In Figure 1(a), RS is the solution resistance, Rct is charge-transfer resistance for copper dissolution, Cdl is double-layer capacitance, and Z is a diffusion component representing diffusion of copper ions through the copper oxide film and/or biofilm.15 In Figure 1(b), Rpo is the pore resistance and Cfilm is the oxide film capacitance. Polarization resistance (Rp) values were estimated through extrapolation of the data to zero frequency. RDE electrodes were removed from the cells and placed in an inorganic salt solution (Table 1) for EIS measurements after the potential had equilibrated. Rotation speeds examined were 0, 300, 1,000, and 3,000 rpm.
Potential-Time Data
Potential vs time data for the copper and platinum electrodes are shown in Figures 2 and 3 for the sterile and inoculated cells, respectively. Copper corrosion potentials in the control cell (Figure 2) and the two inorganic tests at pH 6.8 and pH 8.0 were between –20 mV and 80 mV. Values obtained were consistent with values expected for copper and platinum in naturally aerated potable water. In the test cell (Figure 3), the platinum electrode showed decreases in potential in response to changes in the bulk fluid redox conditions. Significant microbiological consumption of oxygen and medium refreshments caused the platinum potential to vary from –30 mV to 150 mV. The lowest platinum potentials generally were measured prior to a medium exchange (indicated on the x-axes of Figures 2 and 3). The subsequent slow rise in potential suggests that the platinum electrode was covered with a biofilm. Copper electrode potentials in the inoculated cell (Figure 3) either varied markedly between –170 mV and 70 mV (e.g., Cu1) or to a somewhat lesser degree, between –50 mV and 50 mV (e.g., Cu2). The Cu1 electrode was found to be more passive (having Rp values > 100 kΩ-cm2) and was responding to the redox conditions to a greater extent, while Cu2 experienced MIC (having Rp values < 100 kΩ-cm2 and down to 30 kΩ-cm2). Rp values are summarized in Figure 4.
Inorganic Tests
Copper disk electrodes (99.97% Cu, A = 0.38 cm2) were exposed to salt solution (Table 1) at nominal pH values of 6.8 and 8.0 for 36 days. The pH of the solution was checked periodically and, in the test conducted at pH 6.8, adjusted through the addition of 0.05 M sulfuric acid (H2SO4; equating to an 2– additional 19 ppm SO4 ). The solution was not
pH Data
The pH data shown in Figures 2 and 3 are for the effluent medium; the fresh medium had a pH of
944
CORROSION—SEPTEMBER 2000
CORROSION SCIENCE SECTION
7.50. The medium in the control cell (Figure 2) increased to ~ 8.0, while that in the test cell (Figure 3) decreased to between 6.6 and 6.9. The pH in the inorganic tests had average values of 6.8 and 8.0 (with standard deviations of 0.3 and 0.1, respectively). The difference in pH between the two tests was small but sufficient to produce differences in corrosion behavior as reflected in the Rp values given in Figure 4 and Table 2. Rp values of samples at pH 6.8 were significantly lower than at pH 8.0 (Figure 3), indicating the former had higher corrosion rates. Moreover, with time, the corrosion rate decreased at pH 8.0 and increased at pH 6.8 (Table 2). Electrodes removed from the two tests were examined using epifluorescent spectroscopy, and bacteria were found to be absent. The addition 2– of an extra 19 ppm SO4 from H2SO4 in the pH 6.8 2– test increased the SO4 concentration to 38 ppm, 2– the total SO4 concentration still being within the range expected for potable water. Others used 2– 40 ppm SO4 at pH 7.6 and produced corrosion rate values equivalent to the data reported here at pH 8.0.15
FIGURE 2. Potential and pH data vs time for replicate copper (Cu1, Cu2, and Cu3) and platinum electrodes exposed to sterile conditions.
EIS Data
The range of EIS data measured for the control, test, pH 6.8, and pH 8.0 tests are shown in Figures 5 through 8, respectively. Two relaxations were obvious in the EIS data from samples exposed to the inorganic solutions at pH 6.8 and pH 8.0 (Figures 7 and 8, respectively) and those that experienced MIC in the inoculated cell (Figure 6, Day 46). In these instances, the two relaxations became more distinct with time. In the control cell (Figure 5) and prior to the onset of MIC in the test cell, distinct relaxations were not obvious. Corrosion rate information was estimated from the EIS data of all tests by extrapolating to zero frequency to determine the Rp. Figure 4 shows the average values of Rp and one standard deviation for each environment. In the presence of a biofilm, Rp values varied significantly, giving values that encompassed the range of data measured for electrodes exposed to sterile conditions and to fluid of pH 6.8 and pH 8.0 within one standard deviation (Figure 4). In the presence of a biofilm, Rp values were distributed between 40 kΩ-cm2 to 220 kΩ-cm2, rather than being localized at the high and low extremes, indicating that not all electrodes exposed to the microbiological culture were subject to MIC. Rp values measured under sterile conditions and at pH 8.0 (> 85 kΩ-cm2) correlated with those measured by other workers.15 At pH 7.6, others found that Rp values increased from ~ 110 kΩ-cm2 after 2 days immersion to 230 kΩ-cm2 after 21 days. Based on Rp values at the lower extreme of the range of the data presented in Figure 4, in the presence of a biofilm, the corrosion rates of copper
FIGURE 3. Potential and pH data vs time for replicate copper (Cu1, Cu2, and Cu3) and platinum electrodes exposed in the inoculated cell.
FIGURE 4. Summary of the range of Rp values measured in the environments examined. Data shows the average R p and one standard deviation for each environment.
would be ~ 0.5 µ A/cm2 to 1 µ A/cm2 (i.e., 30 kΩ-cm2 to 60 kΩ-cm2 and assuming a B value of 20 mV). This current range does not represent significant metal wastage; however, it is sufficient to produce a copper concentration of 40 mg/L in a small diameter copper tube within a few hours. A mass balance shows that the cathodic process of oxygen reduction
CORROSION—Vol. 56, No. 9
945
CORROSION SCIENCE SECTION
TABLE 2 Values for Equivalent Circuit Components
Test and Time Sterile Day 11 Day 38 Inoculated Day 6 Day 46 Inorganic Test pH 6.8 Day 1 Day 15 Inorganic Test pH 8.0 Day 1 Day 20
(A) (B)
Ecorr (mVSCE) 18 4
Rs (KΩ-cm2)
Cfilm (µF/cm2)
Rpo (KΩ-cm2)
Rct (KΩ-cm2)
Q x 10–6 Normalized for Area
n
Rp Σ(Rpo + Rct) 241 199
1 1
10.8 9.1
120 125
121 74
28.3 43.9
0.6 0.6
–69 72
1 1
4.8 1.7
64 12
250(A)
(B)
6.3 98.5
0.4 0.4
250(A) 70(A)
45 55
1 1
7.9 3.9
7 3
163 40
29.1 69.4
0.6 0.6
170 43
59 37
1 1
16.7 6.3
10 43
73 249
35.6 29
0.6 0.7
83 292
Value estimated visually from Figure. Insufficient data available to extrapolate the data using EQUIVCRT.
normally would limit the amount of copper contamination to 38 mg/L for 12.5-mm copper tubing. The majority of EIS data could be modeled using the circuit given in Figure 1(a); however, the data could be described better by incorporating components representing a porous film covering a barrier film—the barrier film separating the electrolyte next to the base of the pores from the metal substrate (Figure 1[b]).21-22 Evaluations of components for the equivalent circuit given in Figure 1(b) for the data shown in Figures 5 through 8 are given in Table 2. A constant (Q) and its fractional power (n) were used to describe frequency-dependent components when analyzing the data, whereby, the impedance of an ideally polarizable electrode is given by Z(w) = R + Q–n. The term Q(jw)–n corresponds to a complex impedance, referred to as a constant phase element (CPE).23 Q is a constant, j = √–1, w is the angular frequency, and n is a number that expresses how the impedance response deviates from a purely capacitive response; n = 1 for a capacitor.23 Using the equivalent circuit model in Figure 1(b), the higher frequency (HF) relaxation had n values of 0.7 to 0.8—values typical of pure capacitive behavior of nonhomogeneous surfaces.23 Hence, values of capacitance have been reported in Table 2. It is proposed that this capacitance (Cfilm) represents that of the copper oxide film. The relaxation at lower frequencies (LF) had an n value between 0.4 and up to 0.7, with 0.4 being more typical of combined contributions from capacitance, porosity, and diffusion. Because of the combined contributions from porosity and diffusion, values of Q and n are reported in Table 2.
EIS of RDE
In the control test, step change decreases in Rp were measured at some higher rotation speeds. These results were similar to those of other workers,15 showing that the increase in corrosion rate with rotation speed was caused by a reduction in the oxide film thickness. Typical EIS data for electrodes (exposed for 130 days) covered with a biofilm are shown in Figure 9 for rotation speeds of 0 rpm and 3,000 rpm. Modeling of the data using the equivalent circuit shown in Figure 1(b) showed that Rpo decreased with rotation and Cfilm increased with rotation (Table 3). These changes generally occurred gradually after the rotation speed was increased. Under stagnant conditions, there was a significant contribution from porosity and diffusion (through the oxide and biofilm) as n = 0.2, such that the slow response of the electrode did not allow values of Rp to be estimated (hence, an assessment of the effect of rotation rate on the corrosion rate was not possible). With rotation, some of the diffusion/porous electrode behavior was removed as the value of n increased (Table 3, n = 0.58 at 3,000 rpm)—as might be expected if a diffusion boundary layer was being reduced in thickness. The decrease in diffusion boundary in this instance included biofilm and oxide film removal (Cfilm decreased). Microscopic examination showed that prior to rotation the biofilm was an almost continuous layer of cells having a honeycomb structure (or layers of cells with pores). Rotation reduced the extent of the biofilm or, in some instances, caused significant removal of the biofilm.
946
CORROSION—SEPTEMBER 2000
CORROSION SCIENCE SECTION
FIGURE 5. EIS data for copper electrodes exposed to sterile conditions.
FIGURE 6. EIS data for copper electrodes that experienced MIC in the inoculated cell.
DISCUSSION
Contribution of pH to MIC
Rp results in Figure 4 show that there is a direct correlation between corrosion rate, pH, and MIC when it is considered that only some of the electrodes experienced MIC. In instances of MIC, Rp values were in the range from 40 kΩ-cm2 to 70 kΩ-cm2 (Figure 4), values similar to those measured in fluid of pH 6.8 (Figure 4). The pH of the fluid in the microbiological culture was 6.5 to 6.9 (Figure 3). In instances when MIC did not occur, Rp values were similar to those measured under sterile conditions and at pH 8.0 (Figure 4). pH is known to have a significant influence on copper corrosion. Inorganic studies in simulated water have shown higher rates of corrosion at pH 5 than at pH 7.6, the difference being a result of the copper oxide being more compact and protective at pH 7.6.15 In this work, a pH decrease of the bulk fluid was linked to MIC (Figure 4). In other works, tests showed that a bulk pH decrease was not always necessary to cause MIC; Rp values of 30 kΩ-cm2 to 60 kΩ-cm2 were measured when the bulk pH was 7.2.16-17 Also, in studies of large copper coupons in small volumes of fluid, visible corrosion has occurred on samples exposed to the microbiological culture without the bulk fluid pH changing. The bulk fluid pH decrease seems to be linked to frequent medium refreshments. The finding that MIC occurred irrespective of a drop in the bulk fluid pH suggests that pH changes are caused by the production of EPS rather than smaller molecular weight organic acids and EPS production may occur in the bulk fluid or at the metal/fluid interface depending on the growth conditions of the microbiological culture. At low nutrient levels, such as under batch conditions, the EPS is more likely to be produced at surfaces.3 Notably, there was no obvious correlation between planktonic or sessile microbial numbers and a decrease in pH.
FIGURE 7. EIS data for copper electrodes exposed to inorganic salt solution at pH 6.8.
FIGURE 8. EIS data for copper electrodes exposed to inorganic salt solution at pH 8.0.
The generation of a reduced pH environment through microbial activity and the onset of MIC is in agreement with the work conducted by Geesey and coworkers.9,14 They proposed a model of MIC that involves the development of a copper ion concentra-
CORROSION—Vol. 56, No. 9
947
CORROSION SCIENCE SECTION
FIGURE 9. EIS of an electrode covered with a biofilm at 0 rpm and 3,000 rpm. Inset is the Nyquist plot for the same data.
tion cell as a result of the binding of copper by microbially produced EPS and the generation of a weakly acidic environment.14 Model circuits (Figure 1) represent the anodic processes involved in copper corrosion. Other work has shown that the diffusion of copper ions through the oxide controls the corrosion at neutral pH.15
Contribution of the Nature of the Oxide Film to MIC
Under inorganic conditions, the nature of the copper oxide film has been shown to have a controlling influence on the corrosion of copper.15 These workers also showed that there was a direct relationship between the nature of the copper oxide film, the mechanism of corrosion, and pH. Therefore, it would be expected that a model for MIC that involves a reduction in pH also must include the nature of the copper oxide film. Information about the nature of the copper oxide film was provided through modeling of the EIS data. EIS data has been modeled using an equivalent circuit for a material covered with a porous oxide film and a barrier oxide film (Figure 1). The capacitance of a dielectric such as a copper oxide film is given by:
C= εε 0 A d
(1)
where ε0 is the permittivity of free space, ε is the dielectric of the film material, A is the surface area of the film, and d is film thickness. Assuming the surface area of the oxide remained the same, an increase in thickness (as is known to occur with time15) was expected to decrease Cfilm. In this work, modeling the EIS data using an equivalent circuit for a bilayer film showed that the parameter Cfilm decreased with time in all environments (Table 2). This, combined with Cfilm values, being more typical of a film (i.e., as low as 0.3 µ F/cm2, Table 3) than a metal/solution interface, confirmed the applicability of the Cfilm component in the equivalent circuit. Based on the decreasing Cfilm values (Table 2), the oxide that was produced in instances of MIC and at pH 6.8 was found to be thicker than films produced under sterile or pH 8.0 conditions; this is in agreement with others’ findings in low pH environments.15 Evaluation of other parameters corresponding to the properties of the oxide film (i.e., Rpo [Table 2]) showed distinct differences between inoculated and pH 6.8 environments compared to sterile or pH 8.0 environments. In instances of MIC and at pH 6.8, there was a progressive decrease in Rpo with time (Table 2). In contrast, under sterile conditions or pH 8.0, Rpo increased with time (Table 2). It was postulated that the decrease in Rpo was caused by increased conductivity of the fluid within the pores as a result of greater copper ion solubility. This result, with the changes in Cfilm, indicated that small changes in pH significantly can alter the nature of the oxide film. It is proposed that production of EPS creates a slightly acidic environment similar to conditions at pH 6.8. Furthermore, it was suggested that the copper oxide film produced by MIC may be influenced by an additional element—incorporation of EPS. In a separate work, the copper oxide composition was depth-profiled using small-spot XPS. Carbon was enhanced significantly within the oxide film in the presence of a biofilm compared to an equivalent sterile control. Therefore, in addition to the biofilm generating a slightly acidic interfacial condition, the biofilm may contribute to additional oxide porosity through the incorporation of bacteria and their EPS into the oxide structure. These aspects
TABLE 3 Values for Equivalent Circuit Components for RDE Studies of an Electrode Covered with a Biofilm
Rotation Speed (rpm) 0 3,000 Rs (kΩ-cm2) 1 1 Cfilm (µF/cm2) 0.20 0.34 Rpo (kΩ-cm2) 23 17 Rct (kΩ-cm2) — — Q x 10–6 Normalized for Area 73 131
n 0.22 0.58
948
CORROSION—SEPTEMBER 2000
CORROSION SCIENCE SECTION
of MIC of copper have been summarized using the schematic given in Figure 10(c). In Figure 10, MIC is compared with pH 6.8 and pH 8.0 environments examined in this work. Figure 10 shows that in instances of MIC the biofilm creates at least an interfacial condition of low pH and creates an oxide that is relatively thick and of high porosity (like that produced at pH 6.8 rather than pH 8), which includes bacteria and EPS. Higher rates of corrosion generally correlate with decreased Rpo and Cfilm (Table 2) in instances of MIC and at pH 6.8, suggesting that the oxide structure is a key factor in the corrosion process in these environments. This relationship would confirm that the corrosion rate is being controlled by diffusion of copper ions through the oxide film, as proposed for pH 7.6—rather than by diffusion of copper ions in the bulk fluid at pH 5.15 With a flow of solution over the electrodes, Rpo decreased and Cfilm increased in instances of MIC (Table 3). The increase in Cfilm indicates that oxide became thinner with rotation and the decrease in Rpo may have been caused by the pores increasing in size as material was removed.
(a)
(b)
Contribution of the Bioflim to MIC
In instances of MIC, the presence of the biofilm contributed to the corrosion process by providing an additional diffusion barrier, reflected by low values of n (Table 3, 0 rpm). Furthermore, the n value progressively decreased with time, presumably in response to the biofilm becoming thicker. By late in the test (> Day 120), the slow response did not permit evaluation of an Rp from the EIS data. Furthermore, removal of part of the bioflim along with part of the oxide film through rotation also did not permit evaluation of Rp (Figure 9). In other work, where it was possible to determine Rp from EIS data (presumably caused by diffusion barriers being of a lesser dimension), it was shown that biofilm and oxide removal by rotation caused only a small increase in corrosion rate.17 As microscopic examination of this electrode showed that twothirds of the copper surface was stripped of biofilm after rotation, it was concluded that the biofilm was not a particularly important barrier with respect to corrosion. It would seem that the influence of the biofilm as a diffusion barrier on the corrosion rate is secondary to the nature of the copper oxide. However, the biofilm has a significant role to play in terms of creating and maintaining a low pH interfacial condition. This has been illustrated in other work where, following a period of forced corrosion, Cfilm values continued to decrease in the presence of a biofilm (i.e., the oxide film became thicker through corrosion) while it increased in the absence of a biofllm—the presence of the biofilm enabling corrosion to be sustained.16
(c)
FIGURE 10. Schematic diagrams indicating the nature of the copper oxide films at (top) pH 8.0, (middle) pH 6.8, and (bottom) in instances of MIC. The oxide is thinner and more compact at pH 8.0 compared to the thicker more porous oxide produced at pH 6.8 and instances of MIC. In instances of MIC, the biofilm maintains an interfacial condition of reduced pH and EPS is incorporated into the oxide.
CONCLUSIONS
❖ A lower pH condition at the metal/solution interface in the presence of a biofilm was proposed to control the MIC of copper. The low pH condition, possibly combined with the incorporation of EPS, changed the morphology of the copper oxide film. The biofilm is believed to be critical in maintaining the low interfacial pH condition and in stabilizing corro-
CORROSION—Vol. 56, No. 9
949
CORROSION SCIENCE SECTION
sion, but is believed to have only a secondary role in controlling the rate of corrosion. This is controlled by the porosity of the oxide film. ❖ Fitting EIS data to an equivalent circuit simulating a surface covered with a bilayer film was justified for the corrosion of copper in simulated potable water. Modeling the bilayer oxide as a porous layer over a thinner barrier layer permitted the key parameter of MIC of copper to be defined (i.e., the nature of copper oxide film).
ACKNOWLEDGMENTS
The authors acknowledge the financial support of the International Copper Association (Project no. 484-96) and the New Zealand Foundation for Research, Science, and Technology (C08516 and C08621). Thanks also to C. Osborne and K. MacLaren for their assistance.
REFERENCES
1. G.G. Page, New Zealand J. Sci. 16 (1973): p. 349. 2. P. Arens, G.J. Tuschewitzki, M. Wollmann, H. Follner, H. Jacobi, Zbl. Hyg. 196 (1995): p. 444. 3. D. Wagner, A.H.L. Chamberlain, W.R. Fischer, J.N. Wardell, C.A.C. Sequeria, Mater. Corros. 48 (1997): p. 311. 4. G.G. Geesey, P.J. Bremer, W.R. Fischer, D. Wagner, C.W. Keevil, P. Angell, “Unusual Types of Pitting Corrosion of Copper Tubes in Potable Water Systems,” in Biofouling and Biocorrosion in Industrial Water Systems, eds. G.G. Geesey, Z. Lewandowski, H.-C. Fleming (Boca Raton, FL: CRC Press, 1994), p. 243. 5. C.M.E. Dutkiewicz, “Cuprosolvency in Adelaide Drinking Water: Assessment of Microbial Involvement” (Bachelor’s thesis, Flinders University of South Australia, 1995). 6. W.R. Fischer, D. Wagner, H. Peinneman, P. Arens, G.J. Tuschewitzki, Proc. Int. Conf. Microbially Influenced Corrosion, held May 1995 (Houston, TX: NACE International, 1996).
7. G.G. Page, MP 2 (1972): p. 53. 8. E.C. Potter, New Zealand Plumbers J. 11 (1970): p. 35. 9. G.G. Geesey, P.J. Bremer, Mar. Technol. Soc. J. 3 (1990): p. 36-40. 10. R.J. Taylor, Interim Report TPT 522/523C (New York, NY: International Copper Association, 1997). 11. D.M.F. Nicholas, The P.F. Thompson Memorial Lecture, “Arsenic and Old Brass,” paper no. 2, Corrosion Protection Conference 34 (1994), p. 1-10. 12. B.J. Webster, K.R. Maclaren, D.B. Wells, K. Jackson, P.J. Bremer, Interim Report Project no. 484-94 (New York, NY: International Copper Association, 1994). 13. H. Seidlarek, D. Wagner, W.R. Fischer, H.H. Paradies, Corros. Sci. 36 (1994): p. 1,751. 14. G.G. Geesey, M.W. Mittleman, T. Iwaoka, P.R. Griffiths, MP 25 (1986): p. 37-40. 15. Y. Feng, W.-K. Teo, K.-S. Siow, K.-L. Tan, A.-K. Hsieh, Corros. Sci. 3 (1996): p. 369. 16. B.J. Webster, D.B. Wells, P.J. Bremer, “Influence of Potable Water on Copper Corrosion,” CORROSION/96, paper no. 294 (Houston, TX: NACE, 1996). 17. B.J. Webster, D.B. Wells, P.J. Bremer, “Potable Water Biofilms, Copper Corrosion, and Copper By-Product Release,” 13th Int. Corros. Cong., paper no. 408, held Nov 25-29 (Clayton, Victoria, Australia: Australasian Corrosion Association [ACA], 1996). 18. R.J. Taylor, P.H. Cannington, “Corrosion of Copper Tubes in Potable Waters—Causes and Control,” Corrosion and Prevention ’93, paper no. 33 (Victoria, Australia: ACA, 1993). 19. EG&G Instruments Manual, Models 378-388, Electrochemical Impedance Systems Instruction Manual (1983-1989). 20. B.A. Boukamp, EQUIVCRT† Software (Enschede, The Netherlands: University of Twente). 21. J.L. Dawson, G.E. Thompson, M.B.H. Ahmadun, ASTM STP 1188, “Evolution of Electrochemical Impedance During Sealing of Porous Anodic Films on Aluminium,” in Electrochemical Impedance: Analysis and Interpretation, eds. J.R. Scully, D.C. Silverman, M.W. Kendig (West Conshohocken, PA: ASTM, 1993), p. 255. 22. F. Mansfeld, Inhibitors and Coatings in Electrochemical and Optical Techniques for the Study and Monitoring of Metallic Corrosion, eds. M.G.S. Ferreira, C.A. Melendres, NATO ASI Series B, vol. 203 (The Netherlands, Kluwer Academic Publishers, 1991), p. 533. 23. E.D. Bidoia, L.O.S. Bulhoes, R.C. Rocha-Filho, Electrochim. Acta 39, 5 (1994): p. 763.
950
CORROSION—SEPTEMBER 2000
Microbiologically Influenced Corrosion of Copper in Potable Water Systems—pH Effects
B.J. Webster,* S.E. Werner, D.B. Wells,** and P.J. Bremer***
ABSTRACT
Copper tubing used in potable water plumbing systems occasionally experiences reactions that lead to the release of copper corrosion by-products into the water. The main factor controlling microbiologically influenced corrosion (MIC) of copper has been identified as a decrease in pH, which in conjunction with the incorporation of bacterially produced extracellular polymeric substances (EPS) in the copper oxide film, decreases the protective nature of the film. The biofilm (bacteria and EPS) is believed to have a secondary role to the nature of the oxide film in controlling the rate of corrosion. MIC was produced in laboratory reactors containing copper electrodes exposed to simulated potable water in the presence of a biofilm composed of microorganisms isolated from a field site. In the presence of the biofilm, small but significant reductions in pH occurred—from an initial value of 7.5 to between 6.5 and 6.9. Using electrochemical impedance spectroscopy (EIS), it was shown that the presence of a biofilm caused instances of higher corrosion rates similar to those measured in inorganic tests at pH 6.8. Modeling the oxide film as a thin barrier layer covered by a porous layer, EIS data revealed similarities between the oxide structure on samples experiencing MIC to samples exposed in sterile tests at pH 6.8. KEY WORDS: blue water, copper, electrochemical impedance spectroscopy, microbiologically influenced corrosion, pH
INTRODUCTION
Copper tubing has been used widely in potable water distribution systems because it is durable and easy to install. However, increasing numbers of reports on the release of copper by-products from the inside wall of copper tubes resulting in the appearance of blue- or green-colored water have caused concern. Manifestations of “blue water” or copper by-product release have been reported for locations in New Zealand, Australia, the United States, Japan, and Europe.1-9 In these instances, the first-drawn water can contain between 5 ppm and 300 ppm copper (as Cu2+) as fine-particulate precipitates. This copper byproduct release occurs in soft waters where the water is neutral or has a neutral-alkaline pH (as distinct from corrosion of copper under conditions of low pH). Corrosion associated with the blue water problem does not significantly compromise the tube integrity but rather results in contamination of the water. Several anomalous cases of blue water have been reported worldwide, consistent with an microbiologically influenced corrosion (MIC) mechanism.2,7,10-11 In these instances, the water was found to have a highoxygen demand, frequently was nonchlorinated, and was held stagnant in pipes for long periods of time. Corrosion damage was pitting (described as pepperpot pitting)3 and more generalized sites of corrosion that contained small pits (causing blue water).12 Investigations into MIC of copper have shown that certain species of bacteria can adhere and grow within a biofilm on copper surfaces, dispelling the earlier belief that copper ions are toxic to all bacteria.4
Submitted for publication March 1999; in revised form, April 2000. * Industrial Research Limited, PO Box 2225, Lower Hutt, New Zealand. ** Industrial Research Limited, PO Box 31-310, Auckland, New Zealand. *** Crop and Food Research, University of Otago, PO Box 56, Dunedin, New Zealand.
942
0010-9312/00/000181/$5.00+$0.50/0 © 2000, NACE International
CORROSION—SEPTEMBER 2000
CORROSION SCIENCE SECTION
Two models have been proposed to describe MIC of copper. One model suggested that the extracellular polymeric substances (EPS) generated during microbial activity created preferential cathodic sites through the cation-selective nature of the EPS.3,13 The other model described MIC of copper as involving the development of a copper ion concentration cell caused by the binding of copper by microbiallyproduced EPS and the generation of a weakly acidic environment.14 It is believed that the impact of preferential cathodes on pitting corrosion induced by MIC would be limited. Rather, for MIC to cause pitting corrosion, it would seem that any influence of microbial activity would have to be on stabilizing or accelerating the anodic reaction in much the same way as has been suggested by the alternative model of MIC of copper.14 It is possible that the data on which the preferential cathode model was suggested was influenced strongly by the development of the copper oxide film on the copper and/or inhibitory properties of the EPS. The present study has added to the model proposed by Geesey, et al., for MIC of copper by determining the relative contributions of the copper ion concentration cell/pH effect and other nonbiological parameters known to control copper corrosion (i.e., the nature of the copper oxide film).14 Others have shown that in simulated potable water systems, the structure of the copper oxide film formed at pH 7.6 is more compact and protective than that produced at pH 5.15 In this work, controlled laboratory studies using bacterial isolates from a field site that produced blue water were compared with studies in inorganic solutions of pH 6.8 and pH 8. This paper primarily covers data from a single test, although five similar tests were conducted and some of the results are reported elsewhere.16-17 In this work, electrochemical measurements and techniques were used to characterize MIC of copper.
TABLE 1 Electrolyte Composition (ppm)(A)
NaHCO3 NaCl Na2SO4 Glucose Yeast extract(C)
(A) (B)
98(B) 33(B) 29(B) 1,250 —
(C)
pH adjusted to 7.5 using dilute H2SO4. Components used in inorganic salt solutions (for tests at pH 6.8 and pH 8 and RDE analyses). Yeast extract was 200 ppm on Day 0 and thereafter 50 ppm.
EXPERIMENTAL PROCEDURES
Microbiological Tests
Epoxy-mounted copper disk electrodes (area = 0.38 cm2) and copper rotating disk electrodes (RDE) (area = 0.51 cm2) were prepared from 99.97% (phosphorus deoxidized) copper sheet and 99.99% copper rod, respectively. Electrodes were ground with wet silicon carbide (SiC) abrasive papers to a 1,000-grit finish, rinsed with deionized water, and dried in warm air before being sterilized. Electrochemical experiments were carried out in duplicate electrochemical cells, which, in addition to six copper electrodes, contained a platinum foil electrode (2 cm2) and a silver-silver chloride (Ag-AgCl)/saturated (sat.) potassium chloride (KCl)
†
Trade name.
reference electrode in a Luggin capillary filled with 0.1 M KCl (all potentials are referenced to the saturated calomel electrode [SCE]). One cell was used as a sterile control and the other cell was inoculated with bacteria. Cells and associated tubing were sterilized in an autoclave (20 mm at 121°C). Copper disks, platinum electrodes, and the empty Luggin capillary were sterilized by soaking in 70% ethanol (C2H5OH) overnight. Syringe-filter sterilized 0.1 M KCl was used to fill the Luggin capillary. Electrochemical cells were assembled and solutions added using aseptic techniques in a downflow cabinet. Each electrochemical cell was filled with ~ 800 mL of an electrolyte, which simulated the anionic composition of potable water and contained glucose and yeast extract to ensure biofilm establishment in a reasonable time-frame (Table 1). The medium was replenished by draining 90% of the spent medium and pumping fresh medium from a 20-L reservoir into the cell; the effluent medium was analyzed for microbiological composition and pH. The first medium exchange occurred on Day 15 and subsequent medium changes generally were conducted every 7 days to 10 days. Electrochemical cells were equilibrated to the atmosphere through a microbiological air filter but were kept in the dark to eliminate photoelectrochemical effects.18 The test cell was inoculated aseptically with a 2-mL subculture of microorganisms at Day 0. The subculture contained five isolates from a field site that was experiencing blue water. All of the isolates were capable of growth on copper surfaces. Further details about the microbes are given elsewhere.16 Planktonic bacterial numbers were measured from fluid removed at each medium exchange. Serial dilutions (0.1% peptone) were plated out in triplicate onto either standard plate count (SPCA, Oxoid†) or R2A (Difco†) agar plates and incubated at 25°C for 7 days. Isolates were differentiated on the basis of colonial morphology, counted, and recorded. At the end of the test, electrodes were stained with a solution of acridine orange and examined for biofilm formation using an epifluorescent microscope. pH was measured on fluid removed from the reactor during media exchanges. The pH probe was calibrated using pH 4.00 ± 0.02 and pH 7.00 ± 0.02
CORROSION—Vol. 56, No. 9
943
CORROSION SCIENCE SECTION
refreshed over the duration of the test. Corrosion potential and EIS measurements were conducted over the duration of the tests. These tests were not conducted aseptically but were conducted in electrolyte free of glucose or yeast extract. (a)
RESULTS
Microbiological Data
Planktonic bacterial numbers in the inoculated cell were ≈ 1 × 107 colony forming units (CFU) per mL for the duration of the test to confirm microbiological activity. The control cell remained sterile to Day 87, which is the period that the majority of the information reported here was recorded. Examination of the surface of the control electrodes by epifluorescence microscopy at the end of the test (Day 130) revealed the absence of bacteria, despite contamination of the bulk fluid. Examination of electrodes from the test cell revealed a surface covered with a continuous layer of cells that were typical of the blue water isolates. The biofilm had a honeycomb structure that contained pores, and bacteria were evident within sites of corrosion.
(b)
FIGURE 1. Equivalent circuit used to model EIS data of the copper electrodes: (a) circuit that includes diffusion of copper ions through the oxide film and (b) a circuit, which includes components that describe the film as a porous oxide.
buffers. Accuracy of the measurements were limited by buffer solutions and was ± 0.02. Electrochemical measurements made over the duration of the test were corrosion potential, redox potential, and electrochemical impedance spectroscopy (EIS). A PARC EG&G 273A† potentiostat, M5210† lock-in amplifier, and M388† electrochemical software were used for all electrochemical measurements.19 EIS measurements were conducted at the corrosion potentials over a frequency range from 10 kHz to 5 mHz, using an alternating current (AC) amplitude of ± 10 mV. EIS data were modeled using EQUIVCRT† to the equivalent circuits given in Figure 1.20 In Figure 1(a), RS is the solution resistance, Rct is charge-transfer resistance for copper dissolution, Cdl is double-layer capacitance, and Z is a diffusion component representing diffusion of copper ions through the copper oxide film and/or biofilm.15 In Figure 1(b), Rpo is the pore resistance and Cfilm is the oxide film capacitance. Polarization resistance (Rp) values were estimated through extrapolation of the data to zero frequency. RDE electrodes were removed from the cells and placed in an inorganic salt solution (Table 1) for EIS measurements after the potential had equilibrated. Rotation speeds examined were 0, 300, 1,000, and 3,000 rpm.
Potential-Time Data
Potential vs time data for the copper and platinum electrodes are shown in Figures 2 and 3 for the sterile and inoculated cells, respectively. Copper corrosion potentials in the control cell (Figure 2) and the two inorganic tests at pH 6.8 and pH 8.0 were between –20 mV and 80 mV. Values obtained were consistent with values expected for copper and platinum in naturally aerated potable water. In the test cell (Figure 3), the platinum electrode showed decreases in potential in response to changes in the bulk fluid redox conditions. Significant microbiological consumption of oxygen and medium refreshments caused the platinum potential to vary from –30 mV to 150 mV. The lowest platinum potentials generally were measured prior to a medium exchange (indicated on the x-axes of Figures 2 and 3). The subsequent slow rise in potential suggests that the platinum electrode was covered with a biofilm. Copper electrode potentials in the inoculated cell (Figure 3) either varied markedly between –170 mV and 70 mV (e.g., Cu1) or to a somewhat lesser degree, between –50 mV and 50 mV (e.g., Cu2). The Cu1 electrode was found to be more passive (having Rp values > 100 kΩ-cm2) and was responding to the redox conditions to a greater extent, while Cu2 experienced MIC (having Rp values < 100 kΩ-cm2 and down to 30 kΩ-cm2). Rp values are summarized in Figure 4.
Inorganic Tests
Copper disk electrodes (99.97% Cu, A = 0.38 cm2) were exposed to salt solution (Table 1) at nominal pH values of 6.8 and 8.0 for 36 days. The pH of the solution was checked periodically and, in the test conducted at pH 6.8, adjusted through the addition of 0.05 M sulfuric acid (H2SO4; equating to an 2– additional 19 ppm SO4 ). The solution was not
pH Data
The pH data shown in Figures 2 and 3 are for the effluent medium; the fresh medium had a pH of
944
CORROSION—SEPTEMBER 2000
CORROSION SCIENCE SECTION
7.50. The medium in the control cell (Figure 2) increased to ~ 8.0, while that in the test cell (Figure 3) decreased to between 6.6 and 6.9. The pH in the inorganic tests had average values of 6.8 and 8.0 (with standard deviations of 0.3 and 0.1, respectively). The difference in pH between the two tests was small but sufficient to produce differences in corrosion behavior as reflected in the Rp values given in Figure 4 and Table 2. Rp values of samples at pH 6.8 were significantly lower than at pH 8.0 (Figure 3), indicating the former had higher corrosion rates. Moreover, with time, the corrosion rate decreased at pH 8.0 and increased at pH 6.8 (Table 2). Electrodes removed from the two tests were examined using epifluorescent spectroscopy, and bacteria were found to be absent. The addition 2– of an extra 19 ppm SO4 from H2SO4 in the pH 6.8 2– test increased the SO4 concentration to 38 ppm, 2– the total SO4 concentration still being within the range expected for potable water. Others used 2– 40 ppm SO4 at pH 7.6 and produced corrosion rate values equivalent to the data reported here at pH 8.0.15
FIGURE 2. Potential and pH data vs time for replicate copper (Cu1, Cu2, and Cu3) and platinum electrodes exposed to sterile conditions.
EIS Data
The range of EIS data measured for the control, test, pH 6.8, and pH 8.0 tests are shown in Figures 5 through 8, respectively. Two relaxations were obvious in the EIS data from samples exposed to the inorganic solutions at pH 6.8 and pH 8.0 (Figures 7 and 8, respectively) and those that experienced MIC in the inoculated cell (Figure 6, Day 46). In these instances, the two relaxations became more distinct with time. In the control cell (Figure 5) and prior to the onset of MIC in the test cell, distinct relaxations were not obvious. Corrosion rate information was estimated from the EIS data of all tests by extrapolating to zero frequency to determine the Rp. Figure 4 shows the average values of Rp and one standard deviation for each environment. In the presence of a biofilm, Rp values varied significantly, giving values that encompassed the range of data measured for electrodes exposed to sterile conditions and to fluid of pH 6.8 and pH 8.0 within one standard deviation (Figure 4). In the presence of a biofilm, Rp values were distributed between 40 kΩ-cm2 to 220 kΩ-cm2, rather than being localized at the high and low extremes, indicating that not all electrodes exposed to the microbiological culture were subject to MIC. Rp values measured under sterile conditions and at pH 8.0 (> 85 kΩ-cm2) correlated with those measured by other workers.15 At pH 7.6, others found that Rp values increased from ~ 110 kΩ-cm2 after 2 days immersion to 230 kΩ-cm2 after 21 days. Based on Rp values at the lower extreme of the range of the data presented in Figure 4, in the presence of a biofilm, the corrosion rates of copper
FIGURE 3. Potential and pH data vs time for replicate copper (Cu1, Cu2, and Cu3) and platinum electrodes exposed in the inoculated cell.
FIGURE 4. Summary of the range of Rp values measured in the environments examined. Data shows the average R p and one standard deviation for each environment.
would be ~ 0.5 µ A/cm2 to 1 µ A/cm2 (i.e., 30 kΩ-cm2 to 60 kΩ-cm2 and assuming a B value of 20 mV). This current range does not represent significant metal wastage; however, it is sufficient to produce a copper concentration of 40 mg/L in a small diameter copper tube within a few hours. A mass balance shows that the cathodic process of oxygen reduction
CORROSION—Vol. 56, No. 9
945
CORROSION SCIENCE SECTION
TABLE 2 Values for Equivalent Circuit Components
Test and Time Sterile Day 11 Day 38 Inoculated Day 6 Day 46 Inorganic Test pH 6.8 Day 1 Day 15 Inorganic Test pH 8.0 Day 1 Day 20
(A) (B)
Ecorr (mVSCE) 18 4
Rs (KΩ-cm2)
Cfilm (µF/cm2)
Rpo (KΩ-cm2)
Rct (KΩ-cm2)
Q x 10–6 Normalized for Area
n
Rp Σ(Rpo + Rct) 241 199
1 1
10.8 9.1
120 125
121 74
28.3 43.9
0.6 0.6
–69 72
1 1
4.8 1.7
64 12
250(A)
(B)
6.3 98.5
0.4 0.4
250(A) 70(A)
45 55
1 1
7.9 3.9
7 3
163 40
29.1 69.4
0.6 0.6
170 43
59 37
1 1
16.7 6.3
10 43
73 249
35.6 29
0.6 0.7
83 292
Value estimated visually from Figure. Insufficient data available to extrapolate the data using EQUIVCRT.
normally would limit the amount of copper contamination to 38 mg/L for 12.5-mm copper tubing. The majority of EIS data could be modeled using the circuit given in Figure 1(a); however, the data could be described better by incorporating components representing a porous film covering a barrier film—the barrier film separating the electrolyte next to the base of the pores from the metal substrate (Figure 1[b]).21-22 Evaluations of components for the equivalent circuit given in Figure 1(b) for the data shown in Figures 5 through 8 are given in Table 2. A constant (Q) and its fractional power (n) were used to describe frequency-dependent components when analyzing the data, whereby, the impedance of an ideally polarizable electrode is given by Z(w) = R + Q–n. The term Q(jw)–n corresponds to a complex impedance, referred to as a constant phase element (CPE).23 Q is a constant, j = √–1, w is the angular frequency, and n is a number that expresses how the impedance response deviates from a purely capacitive response; n = 1 for a capacitor.23 Using the equivalent circuit model in Figure 1(b), the higher frequency (HF) relaxation had n values of 0.7 to 0.8—values typical of pure capacitive behavior of nonhomogeneous surfaces.23 Hence, values of capacitance have been reported in Table 2. It is proposed that this capacitance (Cfilm) represents that of the copper oxide film. The relaxation at lower frequencies (LF) had an n value between 0.4 and up to 0.7, with 0.4 being more typical of combined contributions from capacitance, porosity, and diffusion. Because of the combined contributions from porosity and diffusion, values of Q and n are reported in Table 2.
EIS of RDE
In the control test, step change decreases in Rp were measured at some higher rotation speeds. These results were similar to those of other workers,15 showing that the increase in corrosion rate with rotation speed was caused by a reduction in the oxide film thickness. Typical EIS data for electrodes (exposed for 130 days) covered with a biofilm are shown in Figure 9 for rotation speeds of 0 rpm and 3,000 rpm. Modeling of the data using the equivalent circuit shown in Figure 1(b) showed that Rpo decreased with rotation and Cfilm increased with rotation (Table 3). These changes generally occurred gradually after the rotation speed was increased. Under stagnant conditions, there was a significant contribution from porosity and diffusion (through the oxide and biofilm) as n = 0.2, such that the slow response of the electrode did not allow values of Rp to be estimated (hence, an assessment of the effect of rotation rate on the corrosion rate was not possible). With rotation, some of the diffusion/porous electrode behavior was removed as the value of n increased (Table 3, n = 0.58 at 3,000 rpm)—as might be expected if a diffusion boundary layer was being reduced in thickness. The decrease in diffusion boundary in this instance included biofilm and oxide film removal (Cfilm decreased). Microscopic examination showed that prior to rotation the biofilm was an almost continuous layer of cells having a honeycomb structure (or layers of cells with pores). Rotation reduced the extent of the biofilm or, in some instances, caused significant removal of the biofilm.
946
CORROSION—SEPTEMBER 2000
CORROSION SCIENCE SECTION
FIGURE 5. EIS data for copper electrodes exposed to sterile conditions.
FIGURE 6. EIS data for copper electrodes that experienced MIC in the inoculated cell.
DISCUSSION
Contribution of pH to MIC
Rp results in Figure 4 show that there is a direct correlation between corrosion rate, pH, and MIC when it is considered that only some of the electrodes experienced MIC. In instances of MIC, Rp values were in the range from 40 kΩ-cm2 to 70 kΩ-cm2 (Figure 4), values similar to those measured in fluid of pH 6.8 (Figure 4). The pH of the fluid in the microbiological culture was 6.5 to 6.9 (Figure 3). In instances when MIC did not occur, Rp values were similar to those measured under sterile conditions and at pH 8.0 (Figure 4). pH is known to have a significant influence on copper corrosion. Inorganic studies in simulated water have shown higher rates of corrosion at pH 5 than at pH 7.6, the difference being a result of the copper oxide being more compact and protective at pH 7.6.15 In this work, a pH decrease of the bulk fluid was linked to MIC (Figure 4). In other works, tests showed that a bulk pH decrease was not always necessary to cause MIC; Rp values of 30 kΩ-cm2 to 60 kΩ-cm2 were measured when the bulk pH was 7.2.16-17 Also, in studies of large copper coupons in small volumes of fluid, visible corrosion has occurred on samples exposed to the microbiological culture without the bulk fluid pH changing. The bulk fluid pH decrease seems to be linked to frequent medium refreshments. The finding that MIC occurred irrespective of a drop in the bulk fluid pH suggests that pH changes are caused by the production of EPS rather than smaller molecular weight organic acids and EPS production may occur in the bulk fluid or at the metal/fluid interface depending on the growth conditions of the microbiological culture. At low nutrient levels, such as under batch conditions, the EPS is more likely to be produced at surfaces.3 Notably, there was no obvious correlation between planktonic or sessile microbial numbers and a decrease in pH.
FIGURE 7. EIS data for copper electrodes exposed to inorganic salt solution at pH 6.8.
FIGURE 8. EIS data for copper electrodes exposed to inorganic salt solution at pH 8.0.
The generation of a reduced pH environment through microbial activity and the onset of MIC is in agreement with the work conducted by Geesey and coworkers.9,14 They proposed a model of MIC that involves the development of a copper ion concentra-
CORROSION—Vol. 56, No. 9
947
CORROSION SCIENCE SECTION
FIGURE 9. EIS of an electrode covered with a biofilm at 0 rpm and 3,000 rpm. Inset is the Nyquist plot for the same data.
tion cell as a result of the binding of copper by microbially produced EPS and the generation of a weakly acidic environment.14 Model circuits (Figure 1) represent the anodic processes involved in copper corrosion. Other work has shown that the diffusion of copper ions through the oxide controls the corrosion at neutral pH.15
Contribution of the Nature of the Oxide Film to MIC
Under inorganic conditions, the nature of the copper oxide film has been shown to have a controlling influence on the corrosion of copper.15 These workers also showed that there was a direct relationship between the nature of the copper oxide film, the mechanism of corrosion, and pH. Therefore, it would be expected that a model for MIC that involves a reduction in pH also must include the nature of the copper oxide film. Information about the nature of the copper oxide film was provided through modeling of the EIS data. EIS data has been modeled using an equivalent circuit for a material covered with a porous oxide film and a barrier oxide film (Figure 1). The capacitance of a dielectric such as a copper oxide film is given by:
C= εε 0 A d
(1)
where ε0 is the permittivity of free space, ε is the dielectric of the film material, A is the surface area of the film, and d is film thickness. Assuming the surface area of the oxide remained the same, an increase in thickness (as is known to occur with time15) was expected to decrease Cfilm. In this work, modeling the EIS data using an equivalent circuit for a bilayer film showed that the parameter Cfilm decreased with time in all environments (Table 2). This, combined with Cfilm values, being more typical of a film (i.e., as low as 0.3 µ F/cm2, Table 3) than a metal/solution interface, confirmed the applicability of the Cfilm component in the equivalent circuit. Based on the decreasing Cfilm values (Table 2), the oxide that was produced in instances of MIC and at pH 6.8 was found to be thicker than films produced under sterile or pH 8.0 conditions; this is in agreement with others’ findings in low pH environments.15 Evaluation of other parameters corresponding to the properties of the oxide film (i.e., Rpo [Table 2]) showed distinct differences between inoculated and pH 6.8 environments compared to sterile or pH 8.0 environments. In instances of MIC and at pH 6.8, there was a progressive decrease in Rpo with time (Table 2). In contrast, under sterile conditions or pH 8.0, Rpo increased with time (Table 2). It was postulated that the decrease in Rpo was caused by increased conductivity of the fluid within the pores as a result of greater copper ion solubility. This result, with the changes in Cfilm, indicated that small changes in pH significantly can alter the nature of the oxide film. It is proposed that production of EPS creates a slightly acidic environment similar to conditions at pH 6.8. Furthermore, it was suggested that the copper oxide film produced by MIC may be influenced by an additional element—incorporation of EPS. In a separate work, the copper oxide composition was depth-profiled using small-spot XPS. Carbon was enhanced significantly within the oxide film in the presence of a biofilm compared to an equivalent sterile control. Therefore, in addition to the biofilm generating a slightly acidic interfacial condition, the biofilm may contribute to additional oxide porosity through the incorporation of bacteria and their EPS into the oxide structure. These aspects
TABLE 3 Values for Equivalent Circuit Components for RDE Studies of an Electrode Covered with a Biofilm
Rotation Speed (rpm) 0 3,000 Rs (kΩ-cm2) 1 1 Cfilm (µF/cm2) 0.20 0.34 Rpo (kΩ-cm2) 23 17 Rct (kΩ-cm2) — — Q x 10–6 Normalized for Area 73 131
n 0.22 0.58
948
CORROSION—SEPTEMBER 2000
CORROSION SCIENCE SECTION
of MIC of copper have been summarized using the schematic given in Figure 10(c). In Figure 10, MIC is compared with pH 6.8 and pH 8.0 environments examined in this work. Figure 10 shows that in instances of MIC the biofilm creates at least an interfacial condition of low pH and creates an oxide that is relatively thick and of high porosity (like that produced at pH 6.8 rather than pH 8), which includes bacteria and EPS. Higher rates of corrosion generally correlate with decreased Rpo and Cfilm (Table 2) in instances of MIC and at pH 6.8, suggesting that the oxide structure is a key factor in the corrosion process in these environments. This relationship would confirm that the corrosion rate is being controlled by diffusion of copper ions through the oxide film, as proposed for pH 7.6—rather than by diffusion of copper ions in the bulk fluid at pH 5.15 With a flow of solution over the electrodes, Rpo decreased and Cfilm increased in instances of MIC (Table 3). The increase in Cfilm indicates that oxide became thinner with rotation and the decrease in Rpo may have been caused by the pores increasing in size as material was removed.
(a)
(b)
Contribution of the Bioflim to MIC
In instances of MIC, the presence of the biofilm contributed to the corrosion process by providing an additional diffusion barrier, reflected by low values of n (Table 3, 0 rpm). Furthermore, the n value progressively decreased with time, presumably in response to the biofilm becoming thicker. By late in the test (> Day 120), the slow response did not permit evaluation of an Rp from the EIS data. Furthermore, removal of part of the bioflim along with part of the oxide film through rotation also did not permit evaluation of Rp (Figure 9). In other work, where it was possible to determine Rp from EIS data (presumably caused by diffusion barriers being of a lesser dimension), it was shown that biofilm and oxide removal by rotation caused only a small increase in corrosion rate.17 As microscopic examination of this electrode showed that twothirds of the copper surface was stripped of biofilm after rotation, it was concluded that the biofilm was not a particularly important barrier with respect to corrosion. It would seem that the influence of the biofilm as a diffusion barrier on the corrosion rate is secondary to the nature of the copper oxide. However, the biofilm has a significant role to play in terms of creating and maintaining a low pH interfacial condition. This has been illustrated in other work where, following a period of forced corrosion, Cfilm values continued to decrease in the presence of a biofilm (i.e., the oxide film became thicker through corrosion) while it increased in the absence of a biofllm—the presence of the biofilm enabling corrosion to be sustained.16
(c)
FIGURE 10. Schematic diagrams indicating the nature of the copper oxide films at (top) pH 8.0, (middle) pH 6.8, and (bottom) in instances of MIC. The oxide is thinner and more compact at pH 8.0 compared to the thicker more porous oxide produced at pH 6.8 and instances of MIC. In instances of MIC, the biofilm maintains an interfacial condition of reduced pH and EPS is incorporated into the oxide.
CONCLUSIONS
❖ A lower pH condition at the metal/solution interface in the presence of a biofilm was proposed to control the MIC of copper. The low pH condition, possibly combined with the incorporation of EPS, changed the morphology of the copper oxide film. The biofilm is believed to be critical in maintaining the low interfacial pH condition and in stabilizing corro-
CORROSION—Vol. 56, No. 9
949
CORROSION SCIENCE SECTION
sion, but is believed to have only a secondary role in controlling the rate of corrosion. This is controlled by the porosity of the oxide film. ❖ Fitting EIS data to an equivalent circuit simulating a surface covered with a bilayer film was justified for the corrosion of copper in simulated potable water. Modeling the bilayer oxide as a porous layer over a thinner barrier layer permitted the key parameter of MIC of copper to be defined (i.e., the nature of copper oxide film).
ACKNOWLEDGMENTS
The authors acknowledge the financial support of the International Copper Association (Project no. 484-96) and the New Zealand Foundation for Research, Science, and Technology (C08516 and C08621). Thanks also to C. Osborne and K. MacLaren for their assistance.
REFERENCES
1. G.G. Page, New Zealand J. Sci. 16 (1973): p. 349. 2. P. Arens, G.J. Tuschewitzki, M. Wollmann, H. Follner, H. Jacobi, Zbl. Hyg. 196 (1995): p. 444. 3. D. Wagner, A.H.L. Chamberlain, W.R. Fischer, J.N. Wardell, C.A.C. Sequeria, Mater. Corros. 48 (1997): p. 311. 4. G.G. Geesey, P.J. Bremer, W.R. Fischer, D. Wagner, C.W. Keevil, P. Angell, “Unusual Types of Pitting Corrosion of Copper Tubes in Potable Water Systems,” in Biofouling and Biocorrosion in Industrial Water Systems, eds. G.G. Geesey, Z. Lewandowski, H.-C. Fleming (Boca Raton, FL: CRC Press, 1994), p. 243. 5. C.M.E. Dutkiewicz, “Cuprosolvency in Adelaide Drinking Water: Assessment of Microbial Involvement” (Bachelor’s thesis, Flinders University of South Australia, 1995). 6. W.R. Fischer, D. Wagner, H. Peinneman, P. Arens, G.J. Tuschewitzki, Proc. Int. Conf. Microbially Influenced Corrosion, held May 1995 (Houston, TX: NACE International, 1996).
7. G.G. Page, MP 2 (1972): p. 53. 8. E.C. Potter, New Zealand Plumbers J. 11 (1970): p. 35. 9. G.G. Geesey, P.J. Bremer, Mar. Technol. Soc. J. 3 (1990): p. 36-40. 10. R.J. Taylor, Interim Report TPT 522/523C (New York, NY: International Copper Association, 1997). 11. D.M.F. Nicholas, The P.F. Thompson Memorial Lecture, “Arsenic and Old Brass,” paper no. 2, Corrosion Protection Conference 34 (1994), p. 1-10. 12. B.J. Webster, K.R. Maclaren, D.B. Wells, K. Jackson, P.J. Bremer, Interim Report Project no. 484-94 (New York, NY: International Copper Association, 1994). 13. H. Seidlarek, D. Wagner, W.R. Fischer, H.H. Paradies, Corros. Sci. 36 (1994): p. 1,751. 14. G.G. Geesey, M.W. Mittleman, T. Iwaoka, P.R. Griffiths, MP 25 (1986): p. 37-40. 15. Y. Feng, W.-K. Teo, K.-S. Siow, K.-L. Tan, A.-K. Hsieh, Corros. Sci. 3 (1996): p. 369. 16. B.J. Webster, D.B. Wells, P.J. Bremer, “Influence of Potable Water on Copper Corrosion,” CORROSION/96, paper no. 294 (Houston, TX: NACE, 1996). 17. B.J. Webster, D.B. Wells, P.J. Bremer, “Potable Water Biofilms, Copper Corrosion, and Copper By-Product Release,” 13th Int. Corros. Cong., paper no. 408, held Nov 25-29 (Clayton, Victoria, Australia: Australasian Corrosion Association [ACA], 1996). 18. R.J. Taylor, P.H. Cannington, “Corrosion of Copper Tubes in Potable Waters—Causes and Control,” Corrosion and Prevention ’93, paper no. 33 (Victoria, Australia: ACA, 1993). 19. EG&G Instruments Manual, Models 378-388, Electrochemical Impedance Systems Instruction Manual (1983-1989). 20. B.A. Boukamp, EQUIVCRT† Software (Enschede, The Netherlands: University of Twente). 21. J.L. Dawson, G.E. Thompson, M.B.H. Ahmadun, ASTM STP 1188, “Evolution of Electrochemical Impedance During Sealing of Porous Anodic Films on Aluminium,” in Electrochemical Impedance: Analysis and Interpretation, eds. J.R. Scully, D.C. Silverman, M.W. Kendig (West Conshohocken, PA: ASTM, 1993), p. 255. 22. F. Mansfeld, Inhibitors and Coatings in Electrochemical and Optical Techniques for the Study and Monitoring of Metallic Corrosion, eds. M.G.S. Ferreira, C.A. Melendres, NATO ASI Series B, vol. 203 (The Netherlands, Kluwer Academic Publishers, 1991), p. 533. 23. E.D. Bidoia, L.O.S. Bulhoes, R.C. Rocha-Filho, Electrochim. Acta 39, 5 (1994): p. 763.
950
CORROSION—SEPTEMBER 2000
