The interaction of bacteria and metal surfaces

Transcription

1 Electrochimica Acta 52 (2007) Review article The interaction of bacteria and metal surfaces Florian Mansfeld Corrosion and Environmental Effects Laboratory (CEEL), The Mork Family Department of Chemical Engineering and Materials Science, University of Southern California, Los Angeles, CA , USA Received 9 August 2006; received in revised form 7 May 2007; accepted 7 May 2007 Available online 13 May 2007 Abstract This review discusses different examples for the interaction of bacteria and metal surfaces based on work reported previously by various authors and work performed by the author with colleagues at other institutions and with his graduate students at CEEL. Traditionally it has been assumed that the interaction of bacteria with metal surfaces always causes increased corrosion rates ( microbiologically influenced corrosion (MIC)). However, more recently it has been observed that many bacteria can reduce corrosion rates of different metals and alloys in many corrosive environments. For example, it has been found that certain strains of Shewanella can prevent pitting of Al 2024 in artificial seawater, tarnishing of brass and rusting of mild steel. It has been observed that corrosion started again when the biofilm was killed by adding antibiotics. The mechanism of corrosion protection seems to be different for different bacteria since it has been found that the corrosion potential E corr became more negative in the presence of Shewanella ana and algae, but more positive in the presence of Bacillus subtilis. These findings have been used in an initial study of the bacterial battery in which Shewanella oneidensis MR-1 was added to a cell containing Al 2024 and Cu in a growth medium. It was found that the power output of this cell continuously increased with time. In the microbial fuel cell (MFC) bacteria oxidize the fuel and transfer electrons directly to the anode. In initial studies EIS has been used to characterize the anode, cathode and membrane properties for different operating conditions of a MFC that contained Shewanella oneidensis MR-1. Cell voltage (V) current density (i) curves were obtained using potentiodynamic sweeps. The current output of a MFC has been monitored for different experimental conditions Published by Elsevier Ltd. Keywords: Microbiologically influenced corrosion (MIC); Corrosion inhibition; Bacteria; Bacterial battery; Microbial fuel cell Contents 1. Introduction Examples for the interaction of bacteria and metal surfaces Microbiologically influenced corrosion (MIC) Microbiologically influenced corrosion inhibition (MICI) The bacterial battery The microbial fuel cell (MFC) Summary and conclusions Acknowledgements References Introduction Traditionally it has been assumed that the interaction of bacteria and metal surfaces always results in increased corrosion address: mansfeld@usc.edu. activity. The term microbiologically influenced corrosion (MIC) is usually interpreted as to indicate an increase in corrosion rates due to the presence of bacteria that accelerate the rates of the anodic and/or cathodic corrosion reaction, while leaving the corrosion mechanism more or less unchanged. The possibility of corrosion inhibition caused by microorganisms has rarely been considered. Videla [1] has presented a short summary of the /$ see front matter 2007 Published by Elsevier Ltd. doi: /j.electacta

2 F. Mansfeld / Electrochimica Acta 52 (2007) literature concerning corrosion inhibition by bacteria. In this review different examples for the interaction of bacteria and metal surfaces will be presented that are based on work reported previously by various authors and work performed by the author with colleagues at other institutions and with his graduate students at CEEL. Researchers at the University of Connecticut, University of Southern California and the University of California at Irvine have evaluated the concept of corrosion control using regenerative biofilms (CCURB) for a variety of materials such as Al 2024, mild steel and cartridge brass in laboratory tests [2 8] as well as field tests [9]. Recent results have shown that two strains of Shewanella produced corrosion inhibition of Al 2024, brass and mild steel in artificial seawater (AS) [10]. Al 2024 is very susceptible to pitting corrosion in seawater, however, it has been found that a number of microorganisms are able to prevent pitting of Al 2024 in AS. These results suggest that microbiologically induced corrosion inhibition (MICI) is a more common phenomenon than was previously assumed. It has been shown that MICI occurs only in the presence of a live biofilm. The concept of the bacterial battery has been evaluated recently and it has been found that the power output of this battery increased continuously for several weeks [11]. Finally, various types of microbial fuel cells (MFC) have been proposed in which bacteria are assumed to oxidize the fuel and transport electrons to the fuel cell anode [12]. 2. Examples for the interaction of bacteria and metal surfaces 2.1. Microbiologically influenced corrosion (MIC) The interaction of bacteria and metal surfaces results in the formation of biofilms in a process known as biofouling. MIC has been reported in the chemical processing, oil and gas, and power generation industries in a wide variety of environments. Acid producing bacteria have been found to be the main cause of MIC of carbon steels. One of the first studies of MIC involved sulfate-reducing bacteria (SRB) that thrive only under anaerobic conditions and are found widespread in many waters and soils. SRBs easily reduce inorganic sulfates to sulfides in the presence of hydrogen or organic matter and are aided in the process by the presence of an iron surface. von Wolzogen Kuehr [13] in 1923 proposed the so-called cathodic depolarization mechanism which assumes that the SRBs remove atomic hydrogen from the iron surface which causes accelerated corrosion of iron. The validity of the mechanism seems questionable since the concentration of atomic hydrogen is extremely small in the ph range of Little et al. [14] have given examples for MIC and also have described the use of different experimental techniques for the evaluation of biofilm formation and MIC. No other phenomenon has probably fascinated those studying MIC more than ennoblement, i.e. the increase of the open-circuit potential E corr due to the formation of a biofilm. Ennoblement has been mainly observed for stainless steels (SS) exposed to natural seawater (NS). Mansfeld and Little [15] have reviewed experimental results and various attempts to explain the observed ennoblement phenomena. Early explanations used thermodynamic arguments that suggested that the reversible potential E 0 of the oxygen electrode increased in the presence of biofilms either due to an increase in the partial pressure of oxygen or a decrease of the surface ph. Since E 0 increases only very slightly with an increase of oxygen pressure and since it is unlikely that acidification would increase passivity, these explanations need to be rejected. Formation of H 2 O 2 has also been suggested as causing an increase of E 0 [16]. The possibility that biofilms cause an increase of the exchange current density for oxygen reduction has apparently not been considered. Johnsen and Bardal suggested that ennoblement was due to a change of the cathodic properties of the stainless steels as a result of microbiological activity on the surface [17]. The very interesting explanation of ennoblement observed for SSs in river water by Lewandowski et al. [18,19] involving formation of MnO 2, which has an E 0 close to the observed ennobled E corr of SSs, deals with a special case of ennoblement. Linhardt [20,21] found large amounts of Mn minerals (mainly MnOOH and MnO 2 ) on severely pitted stainless steel turbine blade runners in a hydroelectric plant and suggested that pitting was due to ennoblement caused by biomineralized Mn oxide. The possibility that formation of a biofilm can decrease the passive current density which would also lead to ennoblement has not been considered [22]. One of the few exceptions is the suggestion by Eashwar et al. [23] that ennoblement of stainless steels in seawater is due to the production of inhibitors by bacteria that are retained in the biofilm matrix. An important observation, which has not been explained, is the fact that E corr for SSs exposed to NS can exceed the pitting potential E pit measured in sterile seawater without pitting to occur. This result could be due to formation of inhibiting species by the biofilm as suggested by Eashwar et al. [23] or to a reduction of the chloride concentration at the SS surface that is covered by the biofilm. Little and Mansfeld [22] have discussed the general topic of passivity of stainless steel (SS) in natural seawater with emphasis on the various attempts to formulate a mechanism for the observed phenomenon of ennoblement. They suggested that the Mo content of stainless steel may play a role since in previous studies [17] SSs with higher Mo content showed more positive values of E corr. Little and Mansfeld observed that E corr for 254 SMO stainless steel with 6% Mo was about 300 mv more positive than that of Ti grade 2 after exposure the Pacific Ocean water for 1 year [24]. They observed that the impedance for these two materials did not change with time despite formation of a biofilm and suggested that the porous and water-like structure of the biofilm did not produce the characteristic changes in the impedance spectra that result from the application of protective polymer coatings [25]. Mansfeld [26] exposed samples of SS 316L in NS at Key West, FL for periods up to 6 months. Samples were exposed at three different times of the year in the time period between May 1995 and August 1996 to determine whether seasonal variations affect the degree of ennoblement. As shown in Fig. 1 E corr reached values between 300 and 400 mv versus Ag/AgCl in about 2 months. Despite the ennoblement the impedance spectra did not change with time with more or less constant

3 7672 F. Mansfeld / Electrochimica Acta 52 (2007) Fig. 1. Time dependence of E corr for SS 316L exposed to NS at Key West, FL. values of the polarization resistance R p and electrode capacitance C dl (Fig. 2). These results can be explained by assuming that while E corr increased, it remained in the passive region with a constant value of the passive c.d. i pass. Ennoblement was considered to be due to an increase of the rate of oxygen reduction by a yet unknown mechanism. Chemical analysis with various surface analytical techniques did not show any evidence of the deposition of manganese oxide compounds. Little and Mansfeld [27] exposed polymer coated steel samples with intentional defects to NS and observed that a large number of bacteria were present on the rust layers of samples that were exposed at their natural E corr, while no bacteria were found for samples that were cathodically protected by connection to a piece of zinc. The effect of applied potential on biofilm formation is being evaluated in more detail at present because of its potential significance in the prevention of MIC as well as corrosion of metallic implants and resulting infection in the human body Microbiologically influenced corrosion inhibition (MICI) Corrosion inhibition due to the formation of biofilms has been observed for different materials exposed to corrosive environments in the presence of different bacteria [2 8]. In the following the concept of MICI will be illustrated for Al 2024, mild steel and brass exposed to artificial seawater (AS) [28]. Al 2024-T3 (UNS A92024) was exposed to AS prepared as Vätäänen nine salts solution ((VNSS), ph 7.5) with and without a growth medium that was a mixture of peptone, starch, glucose and yeast extract. Electrochemical impedance spectroscopy (EIS) data were used to evaluate the corrosion behavior of Al 2024 exposed to AS with and without bacteria. Impedance measurements were made with a model PCI4/300 system (Gamry Instruments) using a three-electrode cell. For tests with S. algae or S. ana the electrochemical cell consisted of two identical electrodes with an exposed area of 4.0 cm 2 for each electrode Fig. 2. Time dependence of R p (a) and C dl (b) for SS 316L exposed to NS at Key West, FL. and a saturated calomel electrode (SCE) as a reference electrode. A sinusoidal voltage signal of 10 mv was applied in a frequency range of 10 5 to 10 3 Hz. For tests with S. algae anodic and cathodic polarization curves were obtained at a scan rate of 0.6 V/h after recording of EIS data for 7 days. The two electrodes used for collecting EIS data were used separately to obtain the anodic and cathodic polarization curves, respectively. The tests in sterile AS and in the presence of Bacillus subtilis were performed by Prof. T. K. Wood and co-workers at the University of Connecticut. Their experimental approach has been described elsewhere [2 8]. For tests in abiotic AS the impedance data for Al 2024 had the typical shape observed previously when pitting occurred [2 5]. Fig. 3 shows experimental impedance spectra obtained for Al 2024 during exposure to AS for 30 days. Only four of the spectra collected during this time are shown in the Bode plots of Fig. 3. The spectra suggest that pitting occurred during the entire test period as evidenced by the typical low-frequency minimum of the phase angle which is partially masked by the scatter of the data points below 0.01 Hz. Nevertheless, the spectra in Fig. 3 are in agreement with the pitting

4 F. Mansfeld / Electrochimica Acta 52 (2007) Fig. 3. Bode plots for Al 2024 exposed to AS for 30 days. model proposed by Mansfeld et al. [29,30]. Qualitatively it can be observed that the polarization resistance of active pits R pit, which is close to the impedance value Z at the frequency minimum at low frequencies, increased with increasing exposure time as the pit growth rate decreased. S. algae and S. ana are classified under the group known as iron reducing bacteria. Both strains are able to grow in seawater under aerobic and anaerobic conditions. S. algae or ana were added to VNSS containing growth medium and EIS data were collected for 1 week. The spectra in Fig. 4a and b which are typical for those found for passive metals demonstrate that both strains provided excellent corrosion protection for Al 2024 in AS. After 1 week exposure time 200 g/ml kanamycin was added to the solution containing S. ana to kill the bacteria and EIS data were obtained for another week or longer. After addition of kanamycin the impedance spectra did not change immediately (Fig. 5). Significant changes in the spectra indicating that pitting had been initiated were observed in the low-frequency region only after total exposure for 15 days (Fig. 5). Fig. 5. Bode plots for Al 2024 exposed to AS containing S. ana; addition of kanamycin after 7 days. Fig. 4. Bode plots for Al 2024 exposed to AS containing Shewanella algae (a) and Shewanella ana (b).

5 7674 F. Mansfeld / Electrochimica Acta 52 (2007) Fig. 6. Time dependence of E corr (a), R p and R pit (b) for Al 2024 exposed to AS containing S. algae or S. ana. Fig. 6 shows the time dependence of E corr, the polarization resistance R p of passive surfaces and the polarization resistance R pit of growing pits for Al 2024 in AS and AS containing S. algae or ana. Also shown are results obtained after addition of kanamycin which is an antibiotic. In the presence of the bacterial strains E corr was more negative than in the abiotic solution (Fig. 6a). After addition of kanamycin an immediate increase of E corr was observed and the values of E corr observed in the abiotic solution were reached after about 10 days exposure time. Due to passivation in the presence of the bacteria R p was much larger than R pit (Fig. 6). Despite the immediate increase of E corr (Fig. 6a) R p remained at the levels observed before the addition of kanamycin for several days (Fig. 6b). R pit reached similar values as those determined in abiotic AS only after 1 week following the addition of kanamycin, indicating that active pits had started to grow again (Fig. 6b). As can be seen from Fig. 6 the sharp drop of R p occurred at the time when E corr reached or exceeded the values observed in the sterile solution, where E corr equals the pitting potential E pit. Anodic and cathodic polarization curves were determined using the two electrodes for which the impedance spectra in AS containing S. algae were obtained [10]. As shown in Fig. 7 the rate of oxygen reduction was drastically reduced, while no significant changes in the rate of the anodic reaction were found. These findings indicate that the observed prevention of pitting of Al 2024 in the presence of Shewanella was due to the decrease of Fig. 7. Potentiodynamic polarization curves for Al 2024 after 7 days of exposure to AS containing S. algae or S. ana. E corr below E pit as a result of the decrease of the rate of oxygen reduction. Similar tests were performed for cartridge brass and mild steel [10]. As shown in Fig. 8 S. ana produced a larger decrease of E corr (Fig. 8a) and a much larger increase of R p (Fig. 8b) than S. algae for brass. After addition of kanamycin E corr increased rapidly (Fig. 8a), while R p remained unchanged for at least another 3 days before it dropped to values similar to those found in abiotic AS (Fig. 8b). Contrary to the results obtained for Al 2024 and brass E corr for mild steel was more positive in AS containing the bacteria than in sterile AS (Fig. 9a). No significant changes of E corr for mild steel were observed after the addition of kanamycin to AS containing S. ana. R p remained unchanged for at least 3 days and then decreased to values comparable to those found in sterile AS (Fig. 9b). The polarization curves for brass were similar to those obtained for Al 2024 (Fig. 7), however, for mild steel the rate of the metal dissolution was decreased and only a small decrease of the rate of the cathodic reaction was found [10]. As also discussed by Dubiel et al. [31] the decrease of the corrosion rate of mild steel in the presence of Shewanella is due to a reduction of the rate of both the anodic and the cathodic reaction. In the presence of B. subtilis pitting occurred in the first 2 days of exposure, however, after 3 days the spectra agreed with those for a passive surface, i.e. a simple one-time-constant

6 F. Mansfeld / Electrochimica Acta 52 (2007) Fig. 8. Time dependence of E corr (a) and R p (b) for brass in AS containing S. algae or S. ana. model in which R p is in parallel with the electrode capacitance C (Fig. 10). The fairly high values of R p, which approached the M cm 2 range, suggest that pits formed in the initial stages of exposure have become passivated. Very similar results were obtained in the presence of B. subtilis producing polyglutamate or polyaspartate [2 5]. The increased R p values suggest that the inhibitors produced by the bacteria provided additional corrosion protection. Fig. 11 illustrates the time dependence of the relative corrosion rates expressed as 1/R o pit for the tests in the absence of bacteria (test #45) and 1/R o p for the tests in the presence of B. subtilis (tests #42 44). These values have been obtained by normalizing the experimental R p values with the total exposed area and the R pit values with the time dependent values of the pitted area A pit determined by analysis of the impedance spectra as explained elsewhere [32]. For the tests in the absence of bacteria R p could not be determined due to the lack of sufficient low-frequency data (Fig. 10). Fig. 11 clearly demonstrates the inhibition of pitting corrosion in the presence of B. subtilis. The lowest corrosion rates were observed for the biofilm producing polyaspartate (test #44). The inhibition of pitting in the presence of B. subtilis could be due to exclusion of oxygen from the metal surface which would reduce the rate of the cathodic reduction reaction resulting in a decrease of E corr below E pit as observed for tests in the presence Fig. 9. Time dependence of E corr (a) and R p (b) for mild steel in AS containing S. algae or S. ana. of Shewanella (Fig. 6). However, the experimental values of E corr had the lowest values in the absence of bacteria, while a certain degree of ennoblement was observed in the presence of bacteria (Fig. 12) [6]. This beneficial effect is apparent even when the biofilm contains bacteria that were not engineered to produce inhibitors. Indeed, the observation that pitting occurred in all cases in the first 2 days of exposure clearly suggests that formation of a stable biofilm is needed to stop growth of active pits. The importance of live biofilms has been evaluated in more detail for Al 2024 exposed to sterile AS or sterile AS con- Fig. 10. Bode plots for Al 2024 exposed to AS containing Bacillus subtilis.

7 7676 F. Mansfeld / Electrochimica Acta 52 (2007) Fig. 11. Time dependence of R o pit and R o p for Al 2024 exposed to sterile AS (test #45) and to AS containing B. subtilis (test #42), B. subtilis producing polyglutamate (test #43) or B. subtilis producing polyaspartate (test #44). taining B. subtilis [33]. The spectra obtained in the sterile AS indicated that pitting occurred as shown in Fig. 3, while passive behavior was observed in the presence of the bacteria as shown in Fig. 10. In another experiment, the live biofilm was killed by adding 2.5 mg/ml penicillin G and 1 mg/ml neomycin simultaneously after exposure of the Al 2024 sample to AS containing the bacteria for 90.5 h. The effect of adding the antibiotics is shown in Fig. 13. The impedance spectrum recorded 0.5 h after addition of the antibiotics indicated that Al 2024 was still passive, however, the spectrum obtained 7 h later showed indication of pitting (Fig. 13). The spectra recorded after a total exposure time of h and 124 h were unstable at the lowest frequencies, however, the following five tests resulted in stable spectra similar to the ones recorded in sterile AS (Fig. 3). The impedance spectra obtained in the sterile solution did not change when the antibiotics were added after 90.5 h [33]. Fig. 12. Time dependence of E corr for Al 2024 exposed to sterile AS (test #45) and AS containing B. subtilis (test #42), B. subtilis producing polyglutamate (test #43) or B. subtilis producing polyaspartate (test #44). Fig. 13. Bode plots for Al 2024 exposed to AS containing B. subtilis, addition of antibiotics after 90.5 h The bacterial battery As discussed above, a number of different bacteria are able to reduce corrosion rates of different materials in several corrosive media. The difference between the mechanism of corrosion inhibition of brass and Al 2024 produced by Shewanella (S. ana and algae) and B. subtilis was that the corrosion potential E corr became more negative in the presence of Shewanella, but became more positive in the presence of B. subtilis. The observation that one type of bacteria can shift E corr of one metal in the positive direction, while another type can shift E corr of some other metal in the negative direction suggests that it might be possible to construct a bacterial battery that has a larger cell voltage than the same battery that does not contain bacteria. Some preliminary results are discussed below including a new observation concerning the formation of the biofilm on copper [11]. A galvanic cell with pure copper and Al 2024 and an electrolyte containing S. oneidensis MR-1 in Luria Bertani (LB) medium has been prepared. The changes in electrode surface properties have been monitored by means of EIS and measurement of E corr of both electrodes. A control cell, which did not contain the bacteria, has also been tested. Potentiodynamic polarization experiments were used to determine current voltage (I V) curves as a function of time. From the I V curves current power (I P = I V) curves were calculated.

8 F. Mansfeld / Electrochimica Acta 52 (2007) Fig. 14. i V and i P curves for the cell containing sterile LB medium (a) and LB + MR-1 (b). The current density i V and i P curves for the cell without bacteria shown in Fig. 14a demonstrates that the maximum power output was obtained in the first day and dropped drastically in the following days. For the cell with MR-1 the power values increased with time (Fig. 14b). Although the change was slow, it was continuous for about 100 days and it remained at similar values for up to 200 days. Impedance spectra were recorded for each electrode before each i V measurement. In sterile LB medium there was a large increase in the total capacitance of the Al 2024 with exposure time, which suggests that active pitting occurred. Pitting was accelerated by the deposition of metallic Cu resulting from the copper corrosion reaction. Visual observation at the end of the test in LB without MR-1 showed that the Al 2024 surface was pitted and covered by metallic Cu. The spectra for the Cu electrode in sterile LB medium showed the typical behavior for an actively corroding metal. On the other hand, the impedance of Al 2024 in the LB medium containing MR-1 remained at high levels in the first week of exposure and then decreased continuously. Visual observation showed that the Al 2024 surface remained shiny and unattacked. In addition, epifluorescence microscopy of the Al 2024 and Cu plates immersed in the solution containing MR-1 revealed the presence of a patchy biofilm. Important changes have been observed in the impedance spectra of the Cu electrode in the LB + MR-1 cell. Although the impedance for Cu in the first days of exposure was typical for that of a corroding metal (similar to the spectra for Cu in LB (Fig. 15a)), after 4 days of exposure a significant change in impedance was observed and a second time constant appeared (Fig. 15b). The impedance increased continuously and the second time constant became more significant with time. Since these spectra were very similar to those usually observed for metals covered by polymer films, they can be attributed to biofilm formation on the Cu surface. Epifluorescence microscopy of the copper plate immersed in the solution containing MR-1 revealed the presence of a dense biofilm covering the surface. The changes of the impedance were accompanied by an enno- Fig. 15. Impedance spectra for copper in sterile LB medium (a) and LB + MR-1 (b).

9 7678 F. Mansfeld / Electrochimica Acta 52 (2007) Fig. 16. Time dependence of cell voltage in sterile LB medium and LB + MR-1. blement of Cu in LB medium containing MR-1. While the E corr values of Cu in sterile LB medium remained between 0.3 and 0.2 V versus SCE, they exceeded 0.1 V after about 300 h in LB containing MR-1. For Al 2024 E corr remained at around 0.55 V in both media. The cell voltage reflected these trends (Fig. 16) The microbial fuel cell (MFC) It has been suggested that in a microbial fuel cell (MFC) the chemical energy of the fuel is converted into electrical energy by the catalytic actions of microorganisms [12]. Many factors may affect the overall performance of these MFCs including the type of the utilized microorganism and the fuel as well as the exposed areas of the anode and cathode and the nature of the ion exchange membrane. Other factors that may also play a significant role include the microorganism growth conditions, the anolyte/catholyte substances and their respective ph values. The present studies examine these factors and their effects on the power density generation in a MFC as well as how the electrode surface properties are affected by the presence of a microorganism. Some preliminary results are discussed below. A dual compartment MFC was assembled using a pretreated Nafion 117 ion exchange membrane and graphite electrodes (1 cm 2 surface area). The electrodes were constructed from graphite felt bonded to platinum wire with carbon conductive cement adhesive. The assembled MFC was sterilized in an autoclave for 10 min at 121 C. Two sets of measurements were obtained using the sterile MFC. The first set of measurements was recorded without microorganisms added to the MFC. The anode compartment was injected with 8 ml of sterile Shewanella Federation Minimal Media (un-inoculated) which contained lactate as the fuel. The cathode compartment was injected with 6 ml of sterile Minimal Media plus 2 ml of potassium ferricyanide (40 mm). Impedance spectra for the anode and cathode were recorded in a frequency range of 100 KHz to 5 mhz by applying an ac amplitude of 10 mv at E corr. Ag/AgCl reference electrodes were placed in the anode or cathode compartment. Polarization curves were recorded following the impedance measurements. Polarization started at the open-circuit potential of the MFC at which the current I = 0 and ended at the short-circuit potential at which the cell voltage V = 0 and I = I max. A scan rate of 0.1 mv/s was used. From these measurements I V and I P curves were con- Fig. 17. I V and I P curves for the un-inoculated (1 5, 2 29 h) and inoculated (3 53, 4 77 h, h) MFC. structed. The first impedance and polarization measurements for the un-inoculated MFC were performed after 5 h and repeated after a 29 h exposure period. The measurements in the presence of bacteria were obtained with the same MFC after the anode compartment was injected with S. oneidensis MR-1. Two milliliters of anolyte were removed from the MFC and replaced by 2 ml of dense culture MR-1 in fresh Minimal Media. The catholyte remained unchanged. The inoculated MFC was exposed for 24 h before any measurements were conducted. Impedance spectra and polarization curves were then recorded for the inoculated MFC every 24 h over a period of 4 days. Preliminary results for I V and I P curves are presented in Fig. 17 for the un-inoculated and the inoculated MFC. The generated power increased slightly between the two measurements performed in the un-inoculated solution, but dramatically decreased upon the initial addition of MR-1. Each subsequent I P curve for the inoculated MFC showed a substantial increase from the initial power generation and also an increase relative to the un-inoculated case. Ultimately, the power generated by the inoculated MFC was greater than that generated by the un-inoculated MFC. E corr of the inoculated anode became more negative with time. The initial value was 104 mv versus Ag/AgCl and the final E corr value after 92 h of exposure was 11 mv. Impedance spectra for the anode and cathode are given in Fig. 18a and b, respectively. The spectra for the anode followed a one-time-constant model and did not change significantly upon addition of MR-1 (Fig. 18a). The spectra for the cathode seem to be influenced by a diffusion effect as indicated by the frequency dependence of the low-frequency impedance (Fig. 18b). Another set of experiments was conducted using different dual compartment MFC s. Each MFC was assembled using Nafion 424 proton exchange membranes and graphite felt electrodes (18 cm 2 surface area) bonded to Pt wire with carbon conductive epoxy. The cathode electrodes were coated with Pt to catalyze the oxygen reduction reaction, while the anode electrodes remained bare graphite. Each MFC compartment held 30 ml of electrolyte (50 mm sodium phosphate, 100 mm sodium chloride) and nitrogen was passed through the anode compartment to maintain anaerobic conditions. Additionally, air was

10 F. Mansfeld / Electrochimica Acta 52 (2007) followed by a more gradual increase. After reaching a maximum the current decreased again. The fuel concentrations were monitored using a high pressure liquid chromatography technique and these results showed that all of the fuel was consumed when the current reached a minimum. When additional fuel (1.5 mm lactate) was injected, the current showed a larger immediate increase relative to the first fuel injection. This steep rise was followed by a more gradual increase until a second maximum was reached. The current then decreased to values that were observed before the first addition of fuel (Fig. 19). The higher current production after the second fuel injection may be due to a larger amount of MR-1, i.e. a thicker biofilm, being attached to the anode after the 24 exposure. 3. Summary and conclusions Fig. 18. Impedance spectra for the un-inoculated (1 5, 2 29 h) and inoculated (3 53, 4 77 h) MFC (a) anode and (b) cathode. passed through the cathode compartment to enhance the rate of the oxygen reduction reaction. A 10 resistor was placed between the anode and the cathode and the current generated by the MFC was monitored as a function of time for different experimental conditions. Fig. 19 shows results for three MCFs that had been prepared in the same way. When only electrolyte was present in the cathode department, a very small current was recorded. A small initial increase in current was observed when MR-1 was added to the anode compartment (Fig. 19). Upon addition of fuel (1.5 mm lactate) a sharp rise of the current occurred Fig. 19. Current time curves for three MFCs for different experimental conditions. Corrosion inhibition caused by bacteria has been observed for Al 2024, mild steel and brass exposed to AS containing growth medium in the presence of S. algae or S. ana. E corr decreased in the presence of bacteria suggesting that the observed prevention of pitting for Al 2024 was due to the creation of anaerobic conditions on the metal surface as a result of which E corr is maintained at values that are lower than E pit which in abiotic AS is equal to E corr. The increase in the corrosion resistance of brass was also considered to be due to reduction of the oxygen concentration at the electrode surface. Potentiodynamic polarization curves obtained in AS containing S. algae after exposure for 7 days showed indeed a significant reduction of the cathodic currents for Al 2024 and brass. For mild steel an increase in the corrosion resistance was accompanied by an increase of E corr. In this case anodic polarization curves demonstrated a small decrease of the anodic current in AS containing S. algae with little change of the cathodic polarization curve. The observed decrease of corrosion rates accompanied by an increase of E corr could be due to production of an inhibiting species and/or as suggested recently by Dubiel et al. [31] due to microbial respiration involving reduction of Fe 3+ to Fe 2+ accompanied by a reduction of the oxygen concentration at the electrode surface. On the other hand, the results obtained in the CCURB project have consistently shown that corrosion inhibition of Al 2024 and brass in AS by B. subtilis, B. licheniformis or E. coli was always accompanied by an ennoblement of E corr [2 8]. In these cases the main cause of prevention of pitting of Al 2024 and reduction of corrosion rates of brass apparently was the production of inhibitors by living cells in the biofilm. It is important to note that the cases discussed here, where the same bacteria inhibit corrosion of a number of different metals, differ from the cases of corrosion inhibition by chemical compounds, where inhibition normally is observed only for very specific combinations of metal/compound/environment. A notable difference in the effect of the additions of biocides has been observed for Shewanella on the one hand and B. subtilis on the other hand. In the first case E corr of Al 2024 and brass increased slowly, but R p remained more or less constant for several days. In the second case E corr decreased almost immediately and pitting of Al 2024 initiated after a few hours. These findings demonstrate that live biofilms formed by different bacteria can

11 7680 F. Mansfeld / Electrochimica Acta 52 (2007) provide significant corrosion protection by different at present not well understood mechanisms. A preliminary evaluation of the bacterial battery has shown that the power output of a cell with Cu and Al 2024 electrodes increased continuously with exposure time. While this result is very promising, it has to be realized that similar to the microbial fuel cell the power output is very low. Present work is concentrating on increasing the power output of both devices by evaluating the effects of different electrode and electrolyte materials as well as the use of different strains of Shewanella. One interesting aspect of the MFC seems to be that contrary to conventional fuel cells a large number of fuels can be used. This possibility will be addressed in future research. Acknowledgements The work on the bacterial battery and on the microbial fuel cell is being carried out by Esra Kus and Orianna Bretschger, respectively, who are Ph.D. students in the Mork Family Department of Chemical Engineering and Materials Science at USC. References [1] H.A. Videla, Manual of Biocorrosion, CRC Press, [2] F. Mansfeld, C.H. Hsu, D. Örnek, T.K. Wood, B.C. Syrett, Electrochem. Soc. Proc (2001) 99. [3] F. Mansfeld, C.H. Hsu, D. Örnek, T.K. Wood, B.C. Syrett, J. Electrochem. Soc. 149 (2002) B130. [4] D. Örnek, A. Jayaraman, T.K. Wood, Z. Sun, C.H. Hsu, F. Mansfeld, Corros. Sci. 43 (2001) [5] D. Örnek, T.K. Wood, C.H. Hsu, Z. Sun, F. Mansfeld, Corrosion 58 (2002) 761. [6] F. Mansfeld, C.H. Hsu, Z. Sun, D. Örnek, T.K. Wood, Corrosion 58 (2002) 187. [7] D. Örnek, A. Jayaraman, B.C. Syrett, C.H. Hsu, F. Mansfeld, T.K. Wood, Appl. Microbiol. Biotechnol. 58 (2002) 651. [8] D. Örnek, T.K. Wood, C.H. Hsu, F. Mansfeld, Corros. Sci. 44 (2002) [9] J.C. Earthman, P. Arps, Z.S. Farhangrazi, K. Trandem, T. Wood, Corrosion/2001 Paper no. 1271, NACE, [10] A. Nagiub, F. Mansfeld, Electrochim. Acta 47 (2002) [11] E. Kus, R. Abboud, R. Popa, K.H. Nealson, F. Mansfeld, Corros. Sci. 47 (2005) [12] H.J. Kim, H.S. Park, M.S. Hyun, I.S. Chang, M. Kim, B.H. Kim, Enzyme Microb. Technol. 30 (2002) 145. [13] C. von Wolzogen Kuehr, Water Gas 7 (26) (1923) 277. [14] B.J. Little, P.A. Wagner, F. Mansfeld, Corrosion Testing Made Easy, vol. 5, Microbiologically influenced corrosion, NACE, [15] F. Mansfeld, B. Little, Corros. Sci. 32 (1991) 247. [16] P. Chandrasekaran, S.C. Dexter, Mechanism of Potential Ennoblement of Passive Metals by Seawater Biofilms, Paper no. 493, Corrosion/93, NACE, [17] R. Johnsen, E. Bardal, Corrosion 41 (1985) 296. [18] W.H. Dickinson, Z. Lewandowski, Biodegradation 9 (1998) 11. [19] B.H. Olesen, R. Avci, Z. Lewandowski, Corros. Sci. 42 (2000) 211. [20] P. Linhardt, Werkstoffe und Korosion 45 (1994) 79. [21] P. Linhardt, Failure of chromium nickel steel in a hydroelectric power plant by manganese oxidizing bacteria, in: E. Heitz, -C.H. Fleming, W. Sand (Eds.), Microbially Influenced Corrosion of Materials, Springer Verlag, 1996, p [22] B. Little, F. Mansfeld, Proceedings of the H.H. Uhlig Symposium on Passivity of Stainless Steels in Natural Seawater, vol , The Electrochem. Soc., 1994, p. 42. [23] M. Eashwar, S. Maruthamutu, S. Sathyanarayanan, K. Balakrishnan, Proceedings of the 12th International Corrosion Congress, vol. 5b, September 1993, Houston, TX, NACE, 1993, p [24] B.J. Little, P.A. Wagner, R.I. Ray, Corrosion/93 Paper no. 308, NACE, [25] F. Mansfeld, J. Appl. Electrochem. 25 (1995) 187. [26] F. Mansfeld (Unpublished results). [27] F. Mansfeld, C.C. Lee, L.T. Han, G. Zhang, B.J. Little, P. Wagner, R. Ray, J. Jones-Meehan, The Impact of Microbiologically Influenced Corrosion on Protective Polymer Coatings, Final Report to the Office of Naval Research, Contract No. N , August [28] F. Mansfeld, A. Nagiub, D. Oernek, T.K. Wood, Electrochem. Soc. Proc. 4 (2002) 522. [29] F. Mansfeld, H. Shih, ASTM STP 1134 (1992) 141. [30] F. Mansfeld, Y. Wang, S.H. Lin, H. Xiao, H. Shih, ASTM STP 1182 (1993) 297. [31] M. Dubiel, C.C. Chien, C.H. Hsu, F. Mansfeld, D.K. Newman, Appl. Environ. Mirobiol. 68 (2002) [32] F. Mansfeld, C.H. Tsai, H. Shih, ASTM STP 1154 (1992) 186. [33] R. Zuo, T.K. Wood, E. Kus, F. Mansfeld, Corros. Sci. 47 (2005) 279.