Pergamon Corrosion Science, Vol. 38, No. 1, pp. 73-95, 1996 Copyright 0 1995 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0010-938X/96$15.00+0.00 001&938X(96)00105-0 THE INFLUENCE OF SURFACE FEATURES ON BACTERIAL COLONIZATION AND SUBSEQUENT SUBSTRATUM CHEMICAL CHANGES OF 316L STAINLESS STEEL G. G. GEESEY,’ R. J. GILLIS,’ R. AVCI,* D. DALY,3 M. HAMILTON,3 P. SHOPE4 and G. HARKIN ’ Department of Microbiology, 2 Department of Physics, 3 Department of Mathematical Sciences, 4 Department of Computer Science, Center for Biofilm Engineering, Montana State University, Bozeman, MT 59717, U.S.A. Abstract-Biofilm-forming bacteria were found to selectively colonize specific surface features of unpolished 316L stainless teel exposed to flowing aqueous media. Depending on the types of bacteria present, selective colonization resulted in significant depletion of Cr and Fe relative to Ni in the surface film at these features. No such depletion was observed on uncolonized surfaces exposed to sterile flowing aqueous medium. The results demonstrate that non-random, initial colonization of 3 16L stainless teel surfaces by these bacteria leads to changes in alloy elemental composition in the surface film that are enhanced with time. These chemical changes may be a critical step that weakens the oxide film at specific locations, allowing halides such as Cl- ions greater access to the underlying bulk alloy, and thereby facilitates localized attack and pit formation and propagation. Keywords: A. stainless teel, C. microbiological corrosion, C. pitting corrosion. INTRODUCTION The involvement of micro-organisms in localized attack of stainless steels (SS) has been proposed by a number of investigators.‘” Although the mechanisms of microbially influenced corrosion (MIC) are largely based on speculation at the present, progress has been made on identification of metallurgical features associated with MIC.728 In the case of stainless teels, weld and heat-affected zones are most susceptible to MIC-associated pitting corrosion.9,‘0 Base metal attack is less common, although it has been observed. The majority of chemical evidence attributing certain types of localized corrosion to micro- organisms on metal surfaces has been based on data obtained from surface deposits which have accumulated long after the corrosion reaction was initiated.” Unfortunately, the chemistry of the deposits does not necessarily offer insight to the critical, initial microbially- mediated reactions that compromise the protective surface oxide film in the presence of a developing biofilm. Segregation of elements occurs between grains of alloys such as stainless steels.9”2 Chromium, which is the alloying element hat makes steel stainless, can be depleted in these regions when the alloy has been sensitized and lead to preferential anodic dissolution in these alloys.‘2,13 Little et aZ.14 presented evidence for microbially-assisted depletion of nickel in 90/10 and 70/30 copper-nickel in seawater. In this paper we describe the conditions under Manuscript received 4 November 1994; in amended form 20 March 1995. 73 74 G. G. Geesey et cd which surface-associated microbes promote selective depletion of elements such as chromium at grain boundaries of the surface oxide film on unpolished 3 16L stainless steel. Detection of near-surface chemical changes mediated by film-forming bacteria requires surface sensitive analytical techniques. X-ray photo-electron spectrometry (XPS) and Auger electron spectrometry (AES) provide information on the elemental composition and chemistry of the top l-6 nm of the metal surface. Bruemmer” used AES to characterize grain boundary composition. To date, these types of analyses have only been included in a few MIC studiesI To our knowledge, no studies on chemical changes in the surface oxide film during early stages of microbial surface colonization have been conducted with stainless steels. Subtle changes in surface chemistry at this stage of biofilm development should help clarify the role of biofilm microbes in localized pitting corrosion of stainless steels. The colonization and growth of bacteria on a metal surface results in a biofilm which displays non-uniform coverage over the surface.lh. Until recently, little was known about what controls the colonization of bacteria on metal surfaces in contact with flowing fluids. Mueller et al.” reported that attachment of cells of Pseudomonus oeruginosrr and Pseudomonas fluorescens to hand polished 3 16 SS in a laminar flow fluid field was positively correlated with surface free energy, surface roughness and hydrophobicity. Walsh ef al.” concluded, on the basis of microscopic observations, that initial bacterial attachment was random and that bacterial proliferation, biofilm development and consortia formation was determined by metallurgical features that promote the production of high concentrations of microbial metabolites. In some cases, intergranular attack and pitting are believed to be associated with the presence of bacteria.‘* Walsh et al.” concluded that pits formed at locations adjacent to bacterial accumulations on surfaces and only in rare cases was pitting observed directly beneath a biofilm at early stages of growth. Similar observations were made with respect to the locations of pits and accumulations of biofilm bacteria on fully hydrated copper coupons using atomic force microscopy.” Preferential corrosion attack is not uncommon at grain boundaries, triple points and large inclusions adjacent to microcolonies of bacteria.” In this paper the selective colonization of surface features of unpolished 316L SS by defined pure and mixed cultures of biofilm-forming bacteria are described together with the changes in the elemental composition at these surface features that accompany the colonization process. A rigorous quantitative and statistical approach has been used to verify the relationships between surface features, surface film chemistry and microbial colonization. Based on these results, a mechanism is proposed for the role of biofilm micro- organisms in the localized attack of unpolished forms of this alloy. EXPERIMENTAL Coupon preparation Unpolished, mill-run coupons (20 x 2 cm) made of 20 gauge (0.9144 mm) 3 16L stainless steel with a 2B finish (as received) were obtained from Metal Goods Services, Spokane, WA. The as-received coupons were cut from a coil of product that had been hot rolled (2300”F), pickled and passivated, cold-rolled, solution annealed at 2050°F. rapidly cooled and subjected to a final cold role to flatten and shape. The as-received product exhibited an ASTM #7 grain size. The 2B finish that was applied to the product was based on the procedure described by Lula.20 As-received coupons to be used in reactors for microbioiogical studies were cleaned and repassivated according to ASTM Standard Bacterial colonization and chemical changes of 316L stainless steel 75 Practice A-380-78 code F.21 Passivation involved treatment with 50% nitric acid at 71°C for 30 min. A polishing step was not included in this study in order to simulate surface conditions of material that is placed in service without this treatment. For bulk grain structure analysis, as-received coupons were mounted in phenol- formaldehyde medium, subjected to wet grinding on silicon carbide papers, 120-600 grits, then polished with 12 and 3 pm diamond lap. The polished coupons were etched in 60/40 HNOs/HzO for 45 s according to standard procedures.22 After etching, photographs of the surface were taken using a camera mounted on a metallographic microscope. Grain area was estimated by cutting out images of 134 individual grains on photographic paper, weighing the paper and comparing with the weight of paper representing standardized areas. Bacteria The facultative anaerobic bacterium, Citrobacter freundii, and the sulfate reducing bacterium, Desdfovtbrio gigas, were used in these studies. C. freundii, identified by an Analytical Profile Index (API) 20E assay (Analytical Products Division, Sherwood Medical, Plainview, NY), was an environmental isolate from an active tubercle from a Tennessee Valley Authority pipeline obtained from M.W. Mittelman, University of Tennessee at Knoxville. D. gigas (ATCC #19364) was obtained from the American Type Culture Collection (ATCC) Rockville, MD. It was selected for the study because this species of sulfate reducing bacterium had also been isolated from the corroding stainless teel pipeline. Its large cell size (5 x 40-50 pm) and spirilloid shape permitted differentiation from C. freundii (short rod) by microscopic examination in coupon colonization and suspended culture studies. Culture medium The culture medium (EPRI) used for the flow-through reactor studies is described in Tables l-3. The medium was steam sterilized for 15 min at 121 “C, then incubated at 22°C for 7 days to assure sterility. If no turbidity (indicating microbial contamination) was observed in the medium after 7 days, it was diluted 1: 10 in de-ionized (Dl) water in a carboy, sterilized by autoclaving and amended with filter sterilized solutions of 10 ml sodium sulfite (0.13 g ml-‘), 3.3 ml of 10 mM ferric chloride and 10 ml of modified Hutner’s salts solution2 The final pH of the culture medium was 7.2. The total organic carbon content of the culture medium was approximately 45 ppm. Table 1. Composition of EPRI medium for culturing bacteria in flow-through reactor Sodium lactate Sodium succmate Ammonium nitrate Sodium sulfate Yeast extract Monobasic potassium phosphate Dibasic potassium phosphate Modified Hutner’s salts solution 500 mg 500 mg 500 mg 500 mg 500 mg 1.9 g 6.3 g 10ml Quantities specified per 1 liter deionized water. 76 G. G. Geesey er al. Table 2. Composition of Hutner’s salt solution Nitrilotriacetic acid MgS04 CaCl>. 2Hz0 (NH&Mo7Ox 4HzO FeS04 Hz0 Metals ‘44’ solution log 14.45g 3.335g 9.25 mg 99.0 mg 50 ml Quantities specified per 1 liter deionized water. Table 3. Metals ‘44’ solution Ethylenediaminetetraacetic acid ZnS04 7Hz0 FeS04 7Hz0 MnS04 H20 CuSO4 5HzO Co(NOa)2 6H10 Na2B40, 1 OH20 250 mg 1095 mg 500 mg 154mg 39.2 mg 24.8 mg 17.7 mg Quantities specified per 100 ml Reactor preparation and inoculation The reactor consisted of several silicone tubes, each containing a stainless steel coupon, through which sterile culture medium, diluted 1:lO with Dl water, was fed by a peristaltic pump from a common reservoir (Fig. 1). The reactor was assembled, connected to the reservoir containing Dl water and leak-tested before adding the EPRI medium to the reservoir and steam sterilizing the entire system for 15 min at 12 1 “C. After sterilization, the reactor was placed in an anaerobic chamber. Sterile anaerobic Air Filter Silicone tubing Medium break tube / \ 10 L Carboy Pump wInoculation port flask Fig. 1. Schematic diagram of flow-through anaerobic reactor used to establish microbial biofilms on surfaces of 3 16L SS coupons. Bacterial colonization and chemical changes of 316L stainless steel 71 EPRI medium, diluted 1: 10 with sterile Dl water, was pumped from the reservoir through each set of silicone tubing. The flow rate of the medium through each set of tubing was 1.6 x lOA cm s-‘. This flow rate was selected to mimic flow conditions where corrosion problems in stainless teel piping systems have been observed (Susan Borenstein, Structural Integrity Associates, Inc., San Jose, CA; personal communication). The volume of each set of silicone tubing was approximately 60 ml, providing a culture medium residence time of approximately 3 h. The effluent from each set of tubing was sampled and tested for contamination by the AODC method before inoculation of selected tubing. Each silicone tube containing an SS coupon was inoculated with a different combination of bacteria (D. gigas only; C. freundii only; D. gigas and C. freundii together), except for one in each experiment which was maintained as a sterile control, The tube inoculated with C. fieundii received a one time inoculum of 1 x 10’ cells, as verified by AODC and enumeration of colony forming units (CFU) after plating on MacConkey Agar as described below. The C. freundii inoculum was obtained from a 24 h anaerobic batch culture in EPRI medium maintained at 22°C. The tube inoculated with only D. gigas received an initial inoculum of 1 x IO5 cells, as verified by AODC. The inoculum was obtained from a 1Cday anaerobic culture in EPRI medium maintained at 22°C. The tube receiving inocula of both types of bacteria was inoculated with a similar number of cells of each species as described above. The inoculum of D. gigas was introduced after the concentration of C. freundiiin the reactor effluent (derived from the biofilm growing on the coupon and reactor surface) had achieved a steady level. The concentration of D. gigas in the reactor effluent was determined aily by AODC while the concentrations of C. freundii were determined by AODC and plating on MacConkey’s Agar until the experiment was terminated, at which time the coupons were recovered for microbiological and surface chemical characterization. Coupon sampling At the termination of the flow-through reactor experiments, each silicone tube was cut open using a sterile razor blade, the culture media poured into a sterile beaker, and the coupon removed with sterile forceps. Each coupon was quartered into 2 x 5 cm sections using sterile tin snips. Three sections were placed directly into liquid nitrogen for subsequent analysis by Auger electron spectroscopy (AES), while the remaining section was prepared for enumeration of attached bacteria using the following procedure. The bacteria were scraped off one side of the section of coupon with a sterile scalpel blade and transferred to a vial containing sterile phosphate buffer (PB; 0.19 g l- ’ KH2P04 and 0.63 g 1-l KzHP04). The cell suspension was vortexed to disperse the cells, diluted with PB and C. freundii enumerated by AODC and by plating on MacConkey Agar while D. gigas was enumerated by AODC and the three-tube Most Probable Number (MPN) method as described below. Enumeration of bacteria CFU and AODC were used to determine concentrations of C. freundii in reactor effluents. For enumeration using the culture method, effluents were serially diluted in PB, plated on MacConkey Agar and incubated at 22°C for 24 h under aerobic conditions before enumerating colony forming units (CFU). Preliminary studies indicated that no significant difference in recoverable CFU was detected when anaerobically-grown cells of C. freundii were incubated on MacConkey Agar under aerobic or anaerobic conditions. The three-tube Most Probable Number (MPN) was carried out in anaerobic EPRI medium to enumerate 78 G. G. Geesey et al D. gigas. MPN tubes were considered positive for the SRB if they contained black iron sulfide precipitate after incubation at 25°C for 2 weeks. Densities of C. freundii and D. gigas in tubing effluents were determined by AODC using conventional epifluorescence microscopy. Aqueous suspensions of bacterial cells ranging in volume from 0.1 to 1 .O ml were mixed with 0.1 ml of a 1 pg ml-’ aqueous solution of acridine orange and the final volume adjusted to 2 ml by addition of PB. After a 5 min staining period, the cell suspensions were filtered through blackened polycarbonate membranes (25 mm dia., 0.22 pm pore size) mounted on glass filtration assemblies. The membrane containing the trapped cells was transferred to a glass microscope slide, a drop of immersion oil applied to the membrane surface and a coverslip applied before examining with an Olympus BH2 microscope (Olympus Optical Co. Ltd., Japan) equipped with an ocular counting grid and an Olympus BHZ-RFL-T2 mercury lamp. An Olympus BP490 filter (excitation 490 nm, barrier 515 nm) was used for viewing acridine orange stained preparations. Cells of C. freundii were enumerated at 1000 x magnification, while cells of D. gigas were enumerated at 400 x and distinguished from C. freundii on the basis of size. Bacterial densities were based on counts from at least 30 fields. A segment of each coupon over which the biofilm had not been disturbed was submerged for 5 min in a sterile ll*g ml-’ aqueous solution of the fluorochrome, 4,6 diamidino-2- phenylindole (DAPI), rinsed with PB, air dried and then viewed under the Olympus microscope described above using a UGl filter combination (excitation 340 nm, barrier filter 420 nm) to determine sterility of the coupon in the uninoculated silicone tube as well as cell densities of D. gigas attached to the undisturbed coupon surface. Cells displaying the size characteristic of D. gigas in at least 30 fields were counted and used to estimate their density on the coupon surface. Coupon segments used for determining surface-associated bacterial distribution and density evaluation by confocal scanning laser microscopy (CSLM) were prepared in a manner similar to that described above except that a 1 pg ml-’ aqueous solution of acridine orange was substituted for DAPI because the laser energy was outside the excitation range for the latter fluorochrome. CSLM images of segments of the coupon surface were collected using a Bio-Radsg, Model MRC-600 confocal scanning laser microscope equipped with a krypton/argon laser (Bio-Rad Microscience Division, Cambridge, MA) using a 488 nm excitation and 505 nm barrier filter. The CSLM was used in conjunction with an Olympus microscope (Model BH2, Olympus Optical Co. Ltd., Japan) equipped with an Olympus D Plan Apo UV series 16Oj1.30 oil immersion objective. Ten randomly chosen fields were examined per coupon segment at 2000 x magnification. For each field, two images were saved as digitized images (pit files) using the Bio-Radc@ Comos software. One image was a reflected white light image of the stainless steel surface, showing surface features of the alloy. The second image was an epifluorescent image of acridine orange-stained, surface- associated bacteria. Image processing and analysis Quantitative summaries of the images were computed using in-house software called MARK. Details of the procedure are presented in the Appendix. Energy dispersive analysis by X-ruys Sections of coupons as-received from the supplier as well as sections recovered after cleaning and repassivation were evaluated by energy dispersive X-ray analysis/scanning Bacterial colonization and chemical changes of 3 16L stainless teel 19 electron microscopy (EDX/SEM) to verify bulk elemental abundances reported by the coupon manufacturer. Coupons were examined using a JEOL model JSM 6100 scanning electron microscope equipped with an energy dispersive X-ray spectrometer and software (Noran Instruments, Middleton, WI) using a 15 kV electron beam. Under these operating conditions, the electron beam penetrated approximately 0.85 pm into the stainless steel based on a material density of 8.03g /cm-3.25 The elements were quantified from each spectrum using standardless analysis routine on Norton software. Spectral data were also gathered from areas corresponding to grains and grain boundaries of the surface oxide film on each coupon. The beam was focused on the feature of interest o allow elemental data to be acquired from each feature independently utilizing point spectroscopy. Auger electron spectroscopy A scanning Auger electron spectrometer (Perkin-Elmer, Physical Electronics Division, Phi Model 595, Eden Prairie, MN) equipped with a scanning Auger microprobe (Physical Electronics Instruments Inc., Eden Prairie, MN) coupled to a computer interface was used to obtain surface elemental information on segments of the as-received coupons as well as segments of coupons recovered at the end of an experiment. Surfaces were sputtered using an argon ion beam to penetrate the region that was heavily contaminated by carbon contributed by the aqueous medium and/or the biofilm. Four sequential sputtering events of 15 s intervals (0.5 PA current, 3 keV beam rastered within a 3 x 3 mm* area) were carried out, each followed by a survey Auger analysis of three fields at 2000 x magnification, each covering an area of approximately 50 x 50 pm to locate the region below the surface of the film where carbon contamination was reduced to a level where this element contributed approximately 10% of the total elemental abundance. Using a sputtering rate of 0.1 nm s-‘, based on pure silicon dioxide, we determined the region where the carbon abundance was reduced to this level to be approximately 6 nm below the surface. Small spot Auger spectroscopic analysis was then employed to gather spectra from three fields (0.79 pm*) within a grain or grain boundary of the oxide film within the sputtered area. The vacuum chamber of the spectrometer was baked out prior to sample introduction in order to eliminate the possibility of chamber-derived carbon contamination of the sample. Relative elemental percentages were calculated for each spectrum by measuring the peak height (PHi) of each respective element (E,), dividing by a known sensitivity factor (Ski) for that element, then normalizing to 100% as shown in equation (1) as described by Davis et a1.26 !z Rel. elemental % for E; = SF; -& % x loo %, (1) where I = number of elements of interest, i = 1, 2,..., I. The elemental percentage from each of the following elements was calculated: silicon, phosphorus, sulfur, chlorine, potassium, carbon, nitrogen, oxygen, chromium, iron, nickel and sodium. Auger electron kinetic energies (KE) were used to determine peak locations on each spectrum for each element and the sensitivity factors (SF) were used to calculate elemental percentages as shown in Table 4. Data on molybdenum, one of the key alloying elements in 316L SS, was not included in this survey because the peak for calibrating this element at 2044 eV is outside the energy window used to analyse the other elements of interest. The lines at 186 and 22 1 eV were too weak to obtain useful data. 80 G. G. Geesey ef al. Table 4. Kinetic energies (KE) and sensitivity factors (SF) used for elemental percentage calculations Si P S Cl K C N 0 Cr Fe Ni Na KE 82 130 153 180 235 252 380 505 525 605 750 980 SF 0.35 0.52 0.80 1.0 0.80 0.20 0.30 0.50 0.35 0.20 0.27 0.22 Chromium to nickel and iron to nickel ratios were calculated for each spectrum collected after the fourth 1.5 s sputtering interval (60 s total sputtering time) in order to obtain alloy elemental abundances below the carbon contaminated surface layer of the oxide film. Although the presence of carbon should not affect the abundance of the alloying metals (Cr, Fe and Ni) with respect to each other, excessive carbon contamination ( > 25% carbon elemental abundance) does reduce the contribution of the alloying elements to levels that compromise the accuracy of the resulting ratios. The region sampled below the layer that is heavily contaminated with carbon will be referred to as the near surface. Atomic force microscopy Atomic force microscopy (AFM) was performed on as-received coupons and repassivated coupons exposed to bacteria to determine the topography of the oxide film. The coupons were imaged in air using a Nanoscope III multimode AFM (Digital Instruments, Santa Barbara, CA). The AFM was equipped with a 450 pm silicon cantilever (Digital Instruments) with an imaging force of approximately lo-* N. Three different AFM sections were analysed in three different areas of each coupon. The average vertical displacement across a grain boundary in the surface oxide film was determined from 24 measurements beginning at the top of one grain and ending at the lowest point in an adjacent grain boundary. An average width of a grain boundary was determined from 12 horizontal distance measurements from the edge of one grain to the edge of another adjacent grain in the oxide film. Statistical analyses Statistical significance tests were conducted to see if the data contradict the null hypothesis that bacteria have no preference for attachment to grains over grain boundaries in the surface oxide film. According to the null hypothesis, the probability of a bacterium attaching in a region (grain or grain boundary) equals the proportion of the area of interest represented by that region. The test statistic, denoted by C, is defined in eqn (2). c = 2 N@i - PiI2 , where iz] Pi(l - Pi) Qj = percentage of bacteria within area of interest that are within a grain boundary, Pi = percentage of area of interest containing grain boundaries, Ni = total bacteria within area of interest, i = image number, and n = number of images. If the null hypothesis were true, the statistic C follows a chi-squared probability distribution with n degrees of freedom. By reference to that distribution, the p-value was found. The p-value is the chance of a C as large as observed if the null hypothesis were true. For each AES measurement (raw elemental percentage, ratio, and adjusted percentage), the null hypothesis that the true mean for grains equals the true mean for grain boundaries Bacterial colonization and chemical changes of 316L stainless steel 81 was tested using a two-tailed, unpaired t-test. The test statistics and associatedp-values were calculated using the InStat computer program (version 1. 13, GraphPAD Software). For the chi-squared and t-tests, a small p-value (less than 0.05, say) suggests that the data discredit the associated null hypothesis. RESULTS Coupons of 316L SS with a 2B finish, when examined in an as-received (unpolished), state using a metallurgical microscope, exhibited a surface oxide film that was discontinuous (Fig. 2(a)). The same surface, when viewed at higher magnification by confocal scanning laser microscopy (CSLM) with reflected white light, exhibited a surface oxide film delineated by grains which appeared as white or lighter shades of gray and the region between grains (grain boundaries) which appeared as black or darker shades of gray (Fig. 2(b)). Grains exhibited a mean surface area of 92 pm’. Coupons prepared from Fig. 2. Surface of as-received 316L stainless steel coupon with 2B finish (unpolished) as imaged by (a) metallurgical microscopy, (b) reflected white light confocal scanning laser microscopy, (c) scanning electron microscopy and (d) atomic force microscopy. Grains (g) and grain boundaries (arrow) in the oxide film can be resolved by each method. A topographic map of the surface obtained by atomic force microscopy is presented in (e). The white line in (d) defines path of the probe from which the topographic map was obtained. The depression in the film at the grain boundaries was determined from the vertical distance between the arrows positioned at the top of a grain and bottom of an adjacent grain boundary. The width of the grain boundary was determined by the distance between the edges of adjacent grains. Bar = 20 Aim in (a) and 10 [rrn in (b),(c) and (d). Axes in (e) are in pm. 82 G. G. Geesey et al other lots of rolled 316L SS sheets with a 2B finish also displayed these surface features. These same features were preserved even after coupon cleaning, repassivation and exposure to flowing aqueous medium for 7 days. These features disappeared, however, after electropolishing. Grains and grain boundaries in the oxide film were also observed on as-received coupons when viewed by scanning electron microscopy (Fig. 2(c)). Atomic force microscopy (AFM) of a cleaned, passivated coupon exposed to flowing aqueous medium revealed the oxide film grain boundaries (dark areas) as depressions between the grains (light areas) (Fig. 2(d)). The mean width and depth of the grain boundaries were 1.6 and 0.6 pm, respectively, based on AFM scans collected across 24 grain boundaries (Fig. 2(e)). To determine whether there was a relationship between the grain structure of the surface oxide film elucidated by the microscopic methods described above and the grain structure of the bulk alloy, as-received coupons were polished and etched and then examined under a metallurgical microscope. Coupons prepared in this manner exhibited a typical austenite grain structure and no evidence of annealing twins (Fig. 3). The mean area of a grain was 211 pm2 (n= 134). The area of grains varied from 15-1783 pm’. There was no evidence of carbide precipitation at the grain boundaries, indicating that the metal had not been sensitized. Although the area of grains in the surface oxide film was not significantly different from the area of grains in the bulk alloy due to the wide range of areas observed, the mean area of oxide grains was half that of the grains in the bulk alloy. Thus, the grains of oxide film do not appear to conform to the underlying bulk grain structures. Fig. 3. Surface of as-received 3 16L SS coupon with 2B finish after polishing and etching as imaged by metallurgical microscopy. g = grain, arrow = grain boundary. Bar = 10 {cm. Bacterial colonization and chemical changes of 316L stainless steel Fig. 4. Epifluorescent image of acridine orange-stained. surface-associated cells of C.,fiewrrlii(light objects) superimposed on reflected white light image of surface features of oxide film on 316L SS obtained by confocal scanning laser microscopy after 16 day colonization period. Gray areas represent grains while darker areas which separate grains represent grain boundaries in oxide film. Bar = IO~m. Coupons recovered from the part of the reactor inoculated with a pure culture of C. freundii exhibited a non-random pattern of bacterial colonization. Images obtained by epi- illumination of acridine orange stained cells, when superimposed on reflected white light images of the coupon surface using a confocal scanning laser microscope, revealed a strong partitioning of these bacteria with oxide film grain boundaries as demonstrated after a 16 day exposure period shown in Fig. 4. The density of C. freundii on the coupon was 5.5 x 10’ cells cme2 based on direct microscopic counts of DAPI-stained bacteria on undisturbed surfaces (Table 5). Although bacterial densities increased on the coupon surface with time, the partitioning of cells with grain boundaries was maintained over a 2-4 week period (Fig. 5). Eventually, the bacterial densities on the coupon became so high that the boundaries between grains in the oxide film were obscured. Evaluation of 63 randomly selected coupon areas from four independent experiments indicated that oxide film grain boundaries contributed an average of 32.4% (range of 84 G. G. Geesey et al. Table 5. Density of bacteria (cells/cm*) in undisturbed biofilm on 316L stainless steel coupon exposed to EPRI medium in anaerobic flow reactor Conditions Duration (days) C. freurldii D. gigas D. gigas only Exp. 1 Exp. 2 Co-culture Exp. 1 Exp. 2 Co-culture Exp. 1 Exp. 2 C. fremdii 9 4.2k2.8 x lo3 6.9kO.9 x 10’ 9 4.3k2.6 x lo4 1.1+0.1 X 104 14 5.1*1.2x 10’ 4.6f0.2~ lo3 4.9*0.2x IO’ 1.1+0.1 X 104 14 5.5&lx7x 10’ 11-47%) of the total surface area. An average of 76.6% (range of 61-90%) of the total number of bacteria present on the coupon surface over the 2-4 week exposure period were located within these grain boundaries (Fig. 5). The low percentage contribution of grain boundaries to the coupon surface area and large standard deviation for the percentage of total surface-associated bacteria present within the grain boundaries in 60 40 20 01 I 1 I I I 0 1 2 3 4 5 Weeks after inoculation Fig. 5. Percentage of the total oxide film surface area (grains + grain boundaries) contributed by oxide film grain boundaries (V) and percentage of the total surface-associated C.frettndii bacterial cells that are located at grain boundaries (V) on coupons at different times after inoculation of bacteria to the reactor. Numerals adjacent to data points identify corresponding data sets. The number of images analysed to obtain each data point was n = 9, ?I = 19, n = 19 and n = 16 for data sets 1,2, 3 and 4, respectively. Bar represents + 1 standard deviation. Bacterial colonization and chemical changes of 316L stainless steel 85 Experiment 1 were due to initial problems with software programming and collection of an insufficient number of images, respectively. If the bacteria were randomly distributed over the surface, the fraction of the total population located within the oxide film grain boundaries should not be significantly different from the fraction of the total area contributed by the grain boundaries. The difference was significant (p