Enhancement of exogenous riboflavin on microbiologically influenced corrosion of nickel by electroactive Desulfovibrio vulgaris biofilm

Riboflavin, a soluble redox mediator for electron transfer, is involved in various redox processes in biogeochemical systems. This work gives some insights into the MIC mechanism of pure nickel (Ni) caused by Desulfovibrio vulgaris. Compared with pure copper (Cu) metabolite (M)-MIC, Ni MIC by D. vulgaris exhibits distinctly different corrosion behavior. 20 ppm (w/w) riboflavin dramatically enhances the corrosion rate of Ni (59% increase in weight loss), while it has no impact on Cu MIC. Furthermore, headspace H2 detection reveals that neither proton nor H2S corrosion occurs in Ni MIC (Cu MIC caused by biogenic H2S produces large amounts of H2, whereas Ni MIC does not). The thermodynamic analysis and experimental results indicate that Ni D. vulgaris MIC is caused by trapping extracellular electrons from Ni oxidation for energy, which belongs to extracellular electron transfer (EET)-MIC.

Sulfate is a ubiquitous electron acceptor in anaerobic environments 16 , and SRB can take advantage of the existence of this electron acceptor (Reaction 1) for respiration 17 . Based on the BCSR theory proposed by Gu et al. 18 , when the environment is sufficiently rich in CSs, SRB biofilms preferentially use organic CSs as electron donors (such as lactate, Reaction 2). However, when the environment is deficient in CSs, SRB biofilms are in the "starvation" state and use energetic metals (such as elemental Fe, Reaction 3) as electron donors for sulfate reduction 16 .
Fe ! Fe 2þ þ 2e À ðE o ¼ À447mVÞ (3) in which E o0 in bioelectrochemistry is defined as the reduction potential (vs. standard hydrogen electrode, SHE) at 25°C, pH 7 (indicated by the apostrophe in E) and 1 M solutes (or 1 bar partial pressure of gases) excluding H + . Under anaerobic environments, the MIC mechanisms can be divided into two categories: extracellular electron transfer-MIC (EET-MIC) and metabolite-MIC (M-MIC) 18,19 . M-MIC is a process in which corrosive metabolites (such as organic acids, and H 2 S) secreted by microorganisms corrode metals. Copper (Cu) MIC by SRB is a typical M-MIC case 6 . In addition, EET-MIC refers to the process by which microorganisms directly acquire electrons from the metal oxidation 19 . There are two major pathways for EET: One is direct electron transfer (DET), which relies on electron shuttling proteins on the outer cell membrane cytochromes (OMCs), such as c-type cytochromes 20,21 , or conductive microbial appendages like type IV proteinaceous pili [22][23][24] , such as Geobacter sulfurreducens can transport electrons along the pili 23 . OMCs are responsible for electron storage and transfer, which serve as the key electrical connections between SRB biofilms and metal electrodes 25,26 . The other is mediated electron transfer (MET), which involves soluble electron transfer mediators (ETMs), such as riboflavin, flavin adenine dinucleotide (FAD), flavin mononucleotide (FMN) and phenazine, etc. 8,[27][28][29] .
Electron transfer in MIC, like in microbial fuel cells (MFC), is a rate-limiting step. In MIC research, ETMs have been investigated for their capacity to facilitate EET [28][29][30][31] . The soluble flavin and riboflavin secreted by Shewanella oneidensis can transfer electrons cyclically between microbial cells and metal matrix [30][31][32][33][34][35] . Huang et al. 29 confirmed genetically that EET is the bottleneck of MIC, the endogenous phenazine-1 carboxamide produced by manipulating the phzH gene encoding in Pseudomonas aeruginosa, aggravating the corrosion of 2205 duplex stainless steel (DSS). Krantz et al. 36 discovered that endogenous flavin-like molecules mediated EET in Desulfovibrio alaskensis G20 biofilm is associated with more severe C1018 carbon steel MIC. Microorganisms in nature coexist in a coordinated community, and Jin et al. 37 found that riboflavin secreted by S. oneidensis can be utilized by Bacillus licheniformis to accelerate the 316 L SS MIC.
Riboflavin is the precursor to FAD and FMN. Riboflavin has a complex isoalloxazine ring structure ( Supplementary Fig. 1a, supporting information), which follows the proton-coupled electron transfer (PCET) mechanism 38 . PCET is a non-traditional redox reaction characterized by the unique interdependence in electron and proton transfer processes. Riboflavin typically has three oxidation states 39 , each oxidation state contains protonated 1 and deprotonated forms ( Supplementary Fig. 1b, supporting information) [38][39][40] . The charge distribution and electron-receiving ability of the various substituents in the isoalloxazine ring are affected by the active sites (N 3 and N 5 ) 28 . Each riboflavin receives electrons from electron donors, and reduced riboflavin can transport the electrons to OMCs 41,42 , accelerating metal corrosion. The exact mechanism of riboflavin shuttles extracellular/intracellular electrons remains to be determined by biological scientists.
Nickel (Ni) is an element to improve the corrosion resistance of alloys. Ni-based alloys are indispensable in the applications of geothermal pipes, hydroelectric turbines, etc [43][44][45][46] . Ni is a typical passivated metal 47 , and a few nanometers thick passive film is sufficient to protect the matrix 48,49 . The Ni-based alloys usually outperform conventional stainless steel 50,51 , but alloys are also susceptible to MIC 52 55,56 . Although the protection of passive films is usually effective against non-biological corrosion, SRB biofilm can destroy the passive film, resulting in MIC. Thus, it is worthwhile to investigate the corrosion of pure Ni caused by SRB.
The E o of Ni 2+ /Ni (E o = −250 mV) is slightly more positive than that of Zn 2+ /Zn (E o = −763 mV) and Fe 2+ /Fe (E o = −447 mV), but Ni is also prone to lose electrons in the following oxidation reaction 57 , At 25°C, pH 7 and 1 M solutes (or 1 bar partial pressure of gases), the reaction coupling Ni oxidation (Reaction 4) and sulfate reduction (Reaction 1) has a positive cell potential of E cell = +33 mV > 0, this positive E cell indicates that elemental Ni corrosion with sulfate as the electron acceptor is thermodynamically favorable.
About 20 ppm (w/w) riboflavin has been proven to be an effective tool for identifying EET-MIC from M-MIC 58 . If the MIC mechanism of a metal belongs to or exists EET-MIC, the corrosion rate will be accelerated when CSs are limited or exogenous ETMs are provided 16 . Inspired by this, this work compared with Cu D. vulgaris MIC to investigate the hypothesis that exogenous riboflavin plays a significant role in accelerating Ni D. vulgaris MIC.   Figure 3 shows the headspace H 2 concentration of the 75 mL headspace in anaerobic vials (Cu + D. vulgaris, Ni + D. vulgaris, Cu + Abiotic, Ni + Abiotic, and D. vulgaris only) during the 7-d incubation. The abiotic headspace H 2 curves in Fig. 3 varied linearly during the 7-d incubation period, and the H 2 concentration in the vial with Cu and Ni reached 50 ± 5 ppm (v/v), 60 ± 8 ppm (v/v), respectively. The biotic H 2 evolution curves were all inverted V shape. The H 2 evolution curve in D. vulgaris vial containing Ni coupons was very similar to that of D. vulgaris media alone, with a peak H 2 concentration occurring on the 3rd day, about (7.0 ± 0.5) × 10 2 ppm (v/v). In contrast, a significant rise in H 2 concentration was detected in D. vulgaris vial containing Cu coupons, with a peak headspace H 2 concentration of (2.1 ± 0.2) × 10 3 ppm (v/v).  Figure 4 shows the X-ray diffraction (XRD) (D8 advance, Bruker, Karlsruhe, Germany) patterns of corrosion product films on Cu and Ni coupons after the 7-d incubation in D. vulgaris media without and with riboflavin. The corrosion products of Cu were Cu 2 S. However, the Ni XRD patterns only identified the typical Ni peaks, To study the compositions of corrosion products, X-ray photoelectron spectroscopy (XPS) (Kratos Axis UltraDLD, Shimadzu, Kyoto, Japan) measurements were conducted on Cu and Ni coupons after the 7-d incubation in D. vulgaris media without and with riboflavin, as displayed in Fig. 5. The XPS spectral lines of Cu 2p, Ni 2p, and S 2p were fitted by Gaussian fitting and the corresponding binding energy (BE) are listed in Supplementary Table 1 (supporting information). As shown in Fig. 5a and b, the Cu 2p 1/2 peak was located at 952.4 eV, the Cu 2p 3/2 peak was located at 932.4 eV, and the S 2p 3/2 peak was located at 161.5 eV, corresponding to the BE of Cu + and S 2+ in the standard data of Cu 2 S. In Fig. 5c, the two typical peaks with BE of 874.4 and 856.4 eV were ascribed to Ni 2p 1/2 and Ni 2p 3/2 , respectively, which were characteristic of NiS. In addition, there were also two satellite peaks at 861.5 and 879.8 eV (Fig. 5c). The two peaks at 162.7 eV (S 2p 1/2 ) and 161.5 eV (S 2p 1/2 ) were ascribed to the presence of NiS in Fig. 5d.

Weight loss and pit analysis
The abiotic weight loss for Cu and Ni coupons after the 7-d immersion in the sterile medium with riboflavin was within the range of weighing error (Fig. 6a). Supplementary Fig. 3 (supporting information) shows the surface profile curves of corrosion pits on Cu and Ni coupons after the 7-d immersion in the sterile medium without and with riboflavin. There was no statistically significant difference among the four pit depth data. 20 ppm (w/w) riboflavin did not cause or accelerate the corrosion of Cu and Ni coupons in the sterile medium. Figure 6a also shows the biotic weight loss data of Cu and Ni coupons after the 7-d incubation in D. vulgaris media without and with riboflavin. There were 26.6 ± 3.1, 27.5 ± 2.6, 1.2 ± 0.2, and 1.9 ± 0.3 mg cm −2 corresponding to Cu without riboflavin, Cu with riboflavin, Ni without riboflavin, and Ni with riboflavin, respectively. In the presence of D. vulgaris, the addition of riboflavin resulted in a 59% increase in weight loss after 7-d incubation of corrosion in Ni, whereas the weight loss increase of Cu was insignificant.
The confocal laser scanning microscope (CLSM) (VK-X250 K, Keyence, Osaka, Japan) was used to examine the whole coupon surfaces at low magnification to identify the area with the deepest pit, and then obtain the maximum pit depth data (Figs. 6b, 7) at high magnification. After the addition of riboflavin, the maximum pit depth of Cu increase was statistically insignificant (22.1 ± 5.9 vs. 25.2 ± 2.5 μm), whereas the maximum pit depth of Ni was from 5.4 ± 0.8 to 13.0 ± 2.1 μm, increased by 140%. Unlike Ni MIC, Cu corrosion was severe uniform corrosion accompanied by pitting, and the original polished lines on Cu surfaces have disappeared ( Fig. 7a, b). In fact, pitting is much more serious than uniform corrosion, because surface polishing lines are still visible (Fig. 7c, d), which makes pitting difficult to detect.

Cyclic voltammetry of D. vulgaris biofilms
The steady-state cyclic voltammetry (CV) test was performed in a pH 7.2 phosphate buffered saline (PBS) solution to assess the electroactivity of the D. vulgaris biofilm formed on graphite electrodes (GEs), as shown in Fig. 8. GEs were taken from the D. vulgaris media and evaluated in sterile de-oxygenated PBS to exclude the effect of metabolites. Figure 8a and b show CV curves by various scan rates after the 7-d incubation in D. vulgaris media without and with riboflavin, respectively. The obvious redox peak current signal (red line) was observed in Fig. 8c, suggesting that the electroactivity of the D. vulgaris biofilm was enhanced by adding riboflavin. At 100 mV s −1 scan rate, the redox peak current increased approximately threefold after adding riboflavin, indicating the D. vulgaris biofilm was positively correlated with soluble riboflavin. In addition, the relationship between the current density difference and the scanning rate (Fig. 8d), i.e., the linear slope value, is used to indicate the electroactivity of D. vulgaris biofilms. A steeper slope denotes a more electroactive biofilm. Figure 9 shows the electrochemical results of Cu and Ni coupons after the 7-d immersion in the sterile medium without and with  riboflavin. The abiotic impedance of Cu and Ni coupons slightly increased after the 7-d immersion (Fig. 9a-d). The fitted electrochemical parameters are listed in Tables 1, 2, and 4. Theoretically, the polarization resistance (R p ) is inversely related to the corrosion rate 59 . During the 7-d immersion period, the R p of abiotic Cu and Ni coupons remained around 60 and 25 kΩ cm 2 , respectively, indicating good corrosion resistance. The corrosion current density (i corr ) of abiotic coupons (Fig. 9f) were 1.32 × 10 −7 , 8.96 × 10 −8 , 1.52 × 10 −6 , and 1.08 × 10 −6 A cm −2 corresponding to Cu, Cu with riboflavin, Ni, and Ni with riboflavin, respectively. These results indicated that riboflavin did not affect the electrochemical data of abiotic coupons. Figure 10a shows the open circuit potential (OCP) variation vs. time of Cu and Ni coupons in D. vulgaris media without and with riboflavin during the 7-d incubation period. There was no significant difference in OCP values of Cu electrodes without riboflavin and with riboflavin. The daily OCP values of Ni electrodes with riboflavin were lower than that without riboflavin. The OCP values of Cu electrodes (−750 ± 80 mV, vs. saturated calomel electrode, SCE) were more negative than that of Ni electrodes (−600 ± 50 mV), indicating a stronger corrosion tendency for Cu MIC than for Ni MIC. Figure 10b shows the R p curves of Cu and Ni coupons in D. vulgaris media without and with riboflavin during the 7-d incubation period. For the trend of Ni R p values without and with riboflavin, R p values were closer on the 1st-d (6-8 kΩ cm 2 ) and then reduced over time. The Ni R p curves with riboflavin were significantly lower than that without riboflavin, indicating that the addition of riboflavin resulted in a more severe Ni MIC. But the Cu R p curves with and without riboflavin were nearly the same. In addition, the Cu R p values (0.4-0.7 kΩ cm 2 ) were far lower than that of Ni (2-8 kΩ cm 2 ), corresponding to the much larger weight loss data for Cu coupons in Fig. 6. Figure 11 shows the Nyquist and Bode plots of Cu and Ni coupons after the 1-d, 3-d, 5-d, and 7-d incubation in D. vulgaris media without and with riboflavin. The Nyquist plots of Cu and Ni electrodes exhibit distinctly different characteristics. The Nyquist plots of Cu without and with riboflavin were almost similar, indicating that riboflavin did not affect the Cu MIC. Unlike the Cu MIC, the radius of the semicircle for Ni electrodes with riboflavin was smaller than that of Ni electrodes without riboflavin, indicating that Ni MIC was accelerated by riboflavin. In the Bode diagrams, the low-frequency impedances of Ni electrodes with riboflavin were lower than that without riboflavin, corresponding to the R p (Fig. 10b) and weight loss data (Fig. 6). The Bode diagram also shows the constant phase element (CPE) behavior in the low and mid-frequency regions. The time constants in low-frequency and mid-frequency regions correspond to the charge transfer reaction and corrosion product film, respectively 59 .

Electrochemical measurements
The equivalent circuit models are illustrated in Fig. 12a to fit the electrochemical impedance spectroscopy (EIS) data, equivalent circuits I and II for Cu, and equivalent circuit III for Ni. Model I: Rs(Q f R f )(Q d lR ct ) is a series circuit that represents the electron transport between the media solution/film and film/Cu interfaces (for 1-d).
Model II: Rs(Q f (R f (Q d lR ct W))) is a parallel circuit, suggesting the corrosion product deposit on the Cu surface is a poor barrier (for 3-d, 5-d, and 7-d) 60 . Model III: R s (Q f (R f (Q d lR ct ))) is used to describe the response of the media solution/film and film/Ni interfaces (for 1-d, 3d, 5-d, and 7-d). In equivalent circuit models, R s , R f , and R ct stand for solution, film, and charge transfer resistances, respectively. Q f and Q d l represent the film capacitance, and double-layer capacitance, respectively. W is the Warburg impedance, which represents the resistance to mass transfer. The appearance of W indicates the electrochemical reaction process is controlled by diffusion, which may be due to the formation of a relatively dense Cu 2 S film that inhibits ions from transporting to Cu surfaces for reaction. The nonideal capacitive behavior used CPE instead of capacitance, defined as: where Y 0 and n are the parameters of CPE (n = 1, an ideal capacitor; 0.5 < n < 1, a nonideal capacitance; and n = 0 represents a perfect resistor), ω is the angular frequency, and j is the imaginary unit. The selected fitting circuits fit well with EIS data (Fig. 11), and the fitted values are listed in Tables 1 and 2. R p is the total of R f and R ct for the equivalent circuits with two-time constants, i.e., R p = R f + R ct , the sum of resistances is also a key parameter for MIC 37 . Figure 12b shows the variation of R f + R ct values of Cu and Ni electrodes in D.
vulgaris media without or with riboflavin. For Ni MIC, the R f + R ct curve with riboflavin was much higher than that without riboflavin, indicating that riboflavin accelerated the Ni MIC. The variation of R f + R ct values vs. time was similar to that of R p (Fig. 10b). Figure 13 shows the electrochemical responses of D. vulgaris biofilm of 3-d pre-growth on Cu and Ni coupons to riboflavin. The Ni R p decreased by 45% (from 7.22 to 3.98 kΩ cm 2 ) within 1 h after adding riboflavin, while the Ni R p without riboflavin decreased by 6.2% (from 7.22 to 6.83 kΩ cm 2 ). After the 1-h period, the R p value of Ni with riboflavin was 42% lower than that without riboflavin. The response of D. vulgaris biofilm to riboflavin was quite fast, and the variation of R f + R ct (Fig. 13f and Table 3) is consistent with R p values (Fig. 13e). By comparison, riboflavin did not alter Cu R p values significantly within 1 h, which was almost the same as that without riboflavin (~0.51 kΩ cm 2 ). These results also confirmed that the Cu MIC by D. vulgaris was not EET-MIC 6 . Figure 14 shows the potentiodynamic polarization curves of Cu and Ni coupons after the 7-d incubation in D. vulgaris media without and with riboflavin. The corresponding Tafel parameters are listed in Table 5. The i corr of Ni increased from 5.27 × 10 −6 to 8.39 × 10 −6 A cm −2 (59% increase) with the addition of riboflavin. It was further confirmed that the addition of riboflavin increased the MIC rate of Ni. In contrast, the potentiodynamic polarization curves of Cu without or with riboflavin were almost the same and the i corr values did not differ much, i corr values of Cu had only a slight decrease from 1.05 × 10 −5 to 9.54 × 10 −6 A cm −2 . This result corresponds to the R p (Fig. 10b) and EIS data (Fig. 11).

DISCUSSION
The action mechanism of riboflavin is limited to EET routes. The results in this work proved that the corrosion mechanism of Ni caused by D. vulgaris differed from Cu M-MIC. Figure 15   vulgaris biofilm. 20 ppm (w/w) riboflavin accelerated Ni MIC by D. vulgaris. This is not surprising, given that electron transfer is a key step in EET-MIC. In DET, without the assistance of pilis (conductive nanowires), only single-layer D. vulgaris sessile cells can shuttle electrons in the MIC process, because D. vulgaris cells need to directly contact metal surfaces 28 . In MET, riboflavin is a soluble electron transfer mediator, and multiple-layers sessile cells participate in electron transfer through riboflavin 8 . Therefore, riboflavin can accelerate MIC by accelerating electron transfer from metal surfaces to cells and making more sessile cells participate in the MIC process.
Although Cu ions of very low concentration, are also toxic to microorganisms 61 , such as Cu ions released from 2205-Cu DSS have an antibacterial effect on P. aeruginosa 62 . However, robust D. vulgaris biofilms formed on Cu and Ni coupons (Fig. 2) because D. vulgaris can secrete sulfide to precipitate cytotoxic Cu ions with extremely high pK sp value (47.6 for Cu 2 S at 25°C) 6,[63][64][65] . With the addition of riboflavin, sessile cell counts on Cu coupons (p value = 0.86) and Ni coupons (p value = 0.95) (Fig. 1) are not significantly different statistically, reducing the complexity of this work. After the 7-d incubation, the headspace H 2 S concentration ( Supplementary Fig. 4 A large amount of H 2 is detected in D. vulgaris media with Cu coupons (Fig. 3), which is the result of the Cu M-MIC by H 2 S (Reaction 9) 6 . The potential of 2H + /H 2 is much lower than that of Ni 2+ /Ni (E o = −250 mV), suggesting that Ni oxidation (Reaction 4) coupled with proton reduction (Reaction 10) is thermodynamically unfavorable. The abiotic headspace H 2 concentration for Ni (60 ppm) is far lower than 1010 ppm for Fe 19 , indicating that H 2 generation by proton reduction at pH = 7 is not available in Ni MIC. The headspace H 2 evolution curves in D. vulgaris media with and without Ni coupons are almost similar, this result directly and strongly supports that Ni MIC is EET-MIC rather than M-MIC.
2Cu þ HS À þ H þ ! Cu 2 S þ H 2ðgÞ ðΔG o0 ¼ À58:3 kJ=molÞ (9) 2H þ þ 2e À ! H 2 ðE o0 ¼ À414 mVÞ (10) Compared with Cu D. vulgaris MIC, Ni D. vulgaris MIC exhibits distinctly different corrosion behavior after adding riboflavin. The weight loss of Ni in D. vulgaris media with riboflavin is 1.6-fold higher than that of Ni without riboflavin (Fig. 6). In contrast, the addition of riboflavin has no significant acceleration in Cu weight loss. The actual weight loss data confirm that electron mediator riboflavin causes more severe pitting of Ni, and CLSM data in Fig. 7 validate this result.
The sessile D. vulgaris cells gain electrons from the energetic metal via EET and obtain energy for respiration. EET is ratelimited because planktonic D. vulgaris cells cannot contact metals. Both sessile and planktonic D. vulgaris cells produce H 2 S for M-MIC. As a result, the weight loss in M-MIC may be significantly larger than the weight loss in EET-MIC. The Cu weight loss is 22-fold larger than that of Ni, as demonstrated in Table 1. Fitted electrochemical parameters from EIS data of Cu.
Cu + sterile medium only Ni + sterile medium only  Fig. 6. About 20 ppm (w/w) riboflavin accelerates Ni MIC but not accelerates Cu MIC, and the Cu weight loss remains 14-fold that of Ni. Uniform corrosion of Cu leads to higher weight loss data, whereas dispersed pitting corrosion corresponds to lower weight loss data for Ni. Soluble riboflavin considerably increases the electroactivity of D. vulgaris biofilm adhered to metal electrodes (Fig. 8). The D. vulgaris biofilm has a positive response to riboflavin through EET. Because riboflavin is a soluble molecule, which can transfer electrons in biofilms composed of multiple-layers sessile cells, resulting in more sessile cells participating in the electron transfer process. In summary, riboflavin transports electrons from Ni oxidation to OMCs on cells, where they are stored or transported to the cytoplasm for D. vulgaris respiration, resulting in a more severe Ni MIC.
The more negative the OCP value, the higher the corrosion tendency of working electrodes to lose electrons. In the 7-d incubation period, the OCP curve of Cu is much lower than that of Ni (Fig. 10). Riboflavin did not significantly change the thermodynamic driving force of the MIC process of the two metals. For Ni MIC, riboflavin significantly decreases the R p values during the whole 7-day incubation period, suggesting that the corrosion rate of Ni MIC is greatly accelerated. The reduced semicircle radius (Fig. 11), the reduced fitted R f + R ct (Fig. 12b), and the increased i corr (Fig. 14) also prove that the Ni MIC is aggravated after adding riboflavin. In contrast, riboflavin has no significant change to the linear polarization resistance (LPR), EIS, and Tafel characteristics of Cu MIC. Here, the kinetic electrochemical data from LPR, EIS, and Tafel are in good accordance with the above weight loss (Fig. 6) and pit data (Fig. 7).
This work also examined how rapidly the pre-grown 3-d mature D. vulgaris biofilm responded to riboflavin. Electrochemical response tests are carried out within a short period of 1 h to exclude the effect of D. vulgaris sessile cells. After adding the riboflavin for 1 h, the Ni R p values reduced by 46%, and the R f + R ct values obtained by the fitted EIS data decreased by 45% (Fig. 13). Surprisingly, the R p and R f + R ct values of Ni decreased by 40% within 20 min after adding the riboflavin, which reflects a relatively fast biological reaction time (within 20 min).
The experimental data in this work support the hypothesis that electron transfer may be the bottleneck of Ni MIC caused by D. vulgaris. Compared with Cu M-MIC, 20 ppm (w/w) riboflavin significantly accelerates the Ni MIC (did not increase the sessile cell counts), supporting the EET-MIC route for Ni. The addition of riboflavin increases the weight loss of Ni from 1.2 to 1.9 mg cm −2 , and also increases the maximum pit depth from 5.4 to 13.0 μm. Ni MIC is caused by harvesting the extracellular electrons from Ni oxidation for D. vulgaris metabolism, which requires EET. Therefore, electron transfer is the rate-limiting step in Ni MIC by D. vulgaris. Meanwhile, this work also provides influential evidence for additional investigation into the Cu-Ni alloys MIC system.

Materials preparation and microorganism cultivation
Pure Cu and pure Ni (>99.99%, w/w) coupons used in this work were provided by the Yidatong Metal Products Co., Ltd (Tianjin, China), with a design specification of 10 mm × 10 mm × 5 mm. The coupon surfaces were polished with a series of SiC sandpaper, ranging from 180 to 1500 mesh, and cleaned with anhydrous ethanol and distilled water. Except for the working surface (1 cm 2 ), other surfaces were coated with polytetrafluoroethylene. And then, coupons were transferred to a vacuum box for 20 min of ultraviolet (UV) irradiation and drying naturally.
An anaerobic SRB strain, D. vulgaris (ATCC 7757), was provided by Beijing Bio Bo Wei Biotechnology Co., Ltd (Beijing, China). The ATCC 1249 culture medium (Supplementary Table 2, supporting   Table 4. Tafel parameter for Fig. 9f. Parameters from the potentiodynamic polarization curves of Cu and Ni coupons after the 7-d immersion in the sterile medium with and without riboflavin. Table 3. Fitted electrochemical parameters from EIS data in Fig. 13. Headspace H 2 concentration was measured daily. 5 mL gas was taken from the anaerobic vial using an injection syringe and injected into a 100 mL aluminum foil collection bag to dilute the gas with fresh air. About 50 mL of diluted gas was taken from the gas bag using a tubeless syringe and then injected directly into the sample port of the gas alert detector. The gas concentration test must be conducted at least three times, with the average result as the data point. The measurement results of the H 2 sensor were not affected by H2S 59 .

Surface corrosion analysis
The chemical composition of corrosion products was analyzed by XRD and XPS. According to ASTM G102 71 , after incubation, Ni coupons were cleaned by using a 10% (v/v) HCl solution to remove biofilms and corrosion products, and a 20% (v/v) H 2 SO 4 solution was used to clean Cu coupons. The pitting morphologies on coupon surfaces were observed by CLSM, and then, the corresponding 3D data (pit depth/diameter) were statistically analyzed.

Electrochemical analysis
The electrochemical corrosion test was carried out in a threeelectrode system using a Gamry 600 electrochemical analyzer (Reference 600, Gamry, Philadelphia, USA). A platinum mesh (2 cm × 2 cm) and an SCE were used as counter electrode and reference electrode, respectively. The scanning range of LPR was ±10 mV (vs. the OCP) with a 0.167 mV s −1 scanning rate. The frequency range of the EIS test is 10 5 -10 −2 Hz with a test amplitude of 10 mV. ZSimDemo software (version 3.30d, Michigan, USA) was used to fit EIS data. The potentiodynamic polarization curve was measured from −500 to +1000 mV (vs. the OCP) with a 0.167 mV s −1 scanning rate. GEs with dimensions of 10 mm × 10 mm × 3 mm were immersed in the D. vulgaris media for 7-d, and then the CV test was performed with a scanning voltage of −0.8-0 VSCE and a scanning speed of 10-100 mV s −1 .
All the above electrochemical tests were conducted after the OCP stabilized and repeated at least three times to assure the dependability of electrochemical data.     Parameters from the potentiodynamic polarization curves of Cu and Ni coupons after the 7-d incubation in the D. vulgaris media with and without riboflavin.