Introduction

Global economic losses due to metal corrosion annually equate to 2–4% of GDP, with microbiologically influenced corrosion (MIC) accounting for a high proportion1,2. MIC is a corrosion process affecting industrial systems and infrastructure, caused directly by microorganisms or through their metabolic processes3,4,5. It is categorized into two types: metabolite-mediated MIC (M-MIC), driven by corrosive microbial byproducts, and extracellular electron transfer MIC (EET-MIC), where microbes accelerate corrosion by directly acquiring electrons via iron oxidation to sustain metabolism6,7.

Sulfate-reducing bacteria (SRB), which are ubiquitous in soil environments, emerge as particularly aggressive contributors to MIC due to their exceptional metabolic activity8. In 2000, a long-distance X65 pipeline of the Korea National Oil Corporation was forced to undergo a complete shutdown for inspection owing to MIC. Subsequent experimental investigation validated the causal role of SRB in driving the corrosion mechanism9. Meanwhile, Liu et al.10 discovered that SRB proliferate in soil and formed biofilms on the X52 steel surface, directly accelerating corrosion rates, thereby enriching the understanding that SRB participate in corrosion processes. Dong et al.11 further clarified the stimulative role of SRB in the localized corrosion behavior of carbon steel and their accelerating effect on corrosion processes, providing important evidence of corrosion morphology. Wu et al.12 performed electrochemical analysis on X80 steel with stripped coatings in simulated soil containing SRB, confirming that SRB significantly enhanced the localized corrosion of X80 steel, highlighting the severity of the corrosion caused by SRB. These important prior studies unequivocally establish the conclusion that SRB is a key factor in exacerbating pipeline corrosion and threatening the structural integrity and safety of pipeline networks.

However, existing studies have shown that M-MIC is not the primary cause of pipeline steel failure due to SRB13,14. Although there is still some carbon steel corrosion induced by biological H2S from SRB metabolism, known as M-MIC, it is negligible compared to EET-MIC15. EET-MIC comprises two distinct mechanisms: direct electron transfer (DET) enabled by microbial surface conductive components (e.g., c-type cytochromes or conductive nanowires)16,17, and mediated electron transfer (MET) that utilizes mobile redox shuttles such as riboflavin or flavin adenine dinucleotide (FAD)18,19. Riboflavin acts as a redox mediator in the MIC and has been shown to promote corrosion via EET20,21. Jin et al.22 demonstrated that riboflavin mediates electron shuttling from iron to microorganisms, leading to a notable escalation in corrosion rate. Zhang et al.18 found that SRB induced more severe pitting corrosion of stainless steel in a medium containing 10 ppm (w/w) riboflavin. Guan et al.23 have obtained similar results through a riboflavin experiment, confirming that riboflavin can quicken SRB corrosion of X70 steel. To summarize, riboflavin expedited the corrosion process by modulating the metabolic activity and electron transfer processes of SRB corrosion24.

X80 steel has been used extensively in China’s West-East Gas Transmission Project, particularly attributable to its excellent mechanical properties25. As a critical process in pipeline fabrication, welding creates these regions with heterogeneous microstructures and properties that become vulnerable points in pipeline integrity26. Three distinct regions compose a welded joint: the base metal zone (BM), heat affected zone (HAZ), and weld zone (WZ)27. Considerable researches have focused on analyzing the corrosion behavior of welded joint. Li et al.28 simulated the corrosion of X80 steel welded joint in CO2 saturated NACE solution (5.0% NaCl + 0.5% HAc) and revealed that the corrosion current density exhibited the order: HAZ > BM > WZ. Wang et al.29 observed that SRB preferentially adhered to the BM and WZ surfaces, leading to higher localized corrosion susceptibility. Further studies indicated that riboflavin can enhance selective SRB corrosion, underscoring EET as a pivotal mechanism in MIC15,24. Such findings challenge conventional corrosion paradigms, emphasizing the need for exploration into multifactorial interactions (e.g., microorganisms and electron carriers) in welded joint corrosion.

Currently, large quantities of X80 steel are buried in acidic soil across China. The soil environment in these areas creates favorable conditions for the survival and proliferation of SRB, inevitably leading to corrosion issues in the pipeline steel30. Our earlier investigations further validated that acidic soil solution accelerate the corrosion of welded joint, with electrochemical polarization serving as the dominant mechanism in the corrosion process27. These aggressive factors of acidic soil (lower pH, resistivity > 100 Ω·m, limited ionic/nutrient) imposed severe physiological stress on SRB, driving their adaptation to environmental challenges through cluster formation. This physiological shift clearly facilitated the corrosion of welded joint in acidic environment. Nevertheless, based on current knowledge, the role of electron carriers in SRB-induced corrosion of welded joint in acidic soil has not yet been clarified. In view of this, this paper aims to study how exogenous riboflavin in acidic soil affects SRB corrosion of pipeline steel welded joint. The findings indicated that riboflavin had the capability to accelerate the MIC process of welded joint in acidic soil, consequently intensifying the corrosion. These mechanistic insights enhance our understanding of MIC in buried pipelines and inform targeted corrosion mitigation strategies.

Results

Bacterial growth and corrosion rate

Figure 1a illustrates the growth curves of planktonic D. vulgaris cells in acidic soil solution without and with riboflavin. In both environments, D. vulgaris cells initially underwent rapid growth and proliferation, followed by a gradual decline toward the experimental endpoint. This trend was primarily attributed to nutrient depletion driven by the metabolic activity of D. vulgaris, which directly correlated with the reduction in the density of SRB. Notably, riboflavin promoted the proliferation of D. vulgaris cells, implying that riboflavin may facilitate the growth of planktonic D. vulgaris cells31. The corrosion rate of BM specimen is shown in Fig. 1b. The corrosion rate of the BM specimen exhibited a marked increase from 32.5 μm·y⁻¹ to 51.6 μm·y⁻¹ (an increase of 58.8%) upon the introduction of exogenous riboflavin. This acceleration in MIC is likely attributable to riboflavin’s function as a redox-active mediator, which facilitates the corrosion of BM specimens by D. vulgaris.

Fig. 1: D. vulgaris cells count and corrosion rate.
figure 1

a The number of planktonic D. vulgaris cells in acidic soil solution without and with 10 ppm riboflavin; b corrosion rate of BM of the welded joint in acidic soil solution without and with 10 ppm riboflavin.

Full size image

Corrosion product morphology

Figure. 2 displays the corrosion products on the specimen surfaces in acidic soil solution without and with riboflavin. In both environments, the microorganisms on the steel surfaces grew well, leading to the formation of a mature biofilm. Among them, distinct stratification was observed in the corrosion products formed on the BM and WZ surfaces, with the outer layer being loose and the inner layer dense. Meanwhile, numerous rod-shaped D. vulgaris cells were readily observable on the specimen surface. In contrast, a more uniform and flattened distribution of corrosion products was observed on the HAZ surface, with its corrosion products significantly lower than those on the BM and WZ surfaces. In particular, exogenous riboflavin triggered substantial aggregation of corrosion products on specimen surfaces. These findings collectively indicated that riboflavin facilitates the SRB-induced corrosion and subsequently increases the formation of corrosion products.

Fig. 2: The morphologies of corrosion products on welded joint.
figure 2

a, b the morphologies of corrosion products on the BM surface in acidic soil solution without riboflavin; c, d the morphologies of corrosion products on the HAZ surface in acidic soil solution without riboflavin; e, f the morphologies of corrosion products on the WZ surface in acidic soil solution without riboflavin; a′, b′ the morphologies of corrosion products on the BM surface in acidic soil solution with 10 ppm riboflavin; c′, d′ the morphologies of corrosion products on the HAZ surface in acidic soil solution with 10 ppm riboflavin; e′, f′ the morphologies of corrosion products on the WZ surface in acidic soil solution with 10 ppm riboflavin.

Full size image
Table 1 EDS results of corrosion products in different regions of welded joint (at%)
Full size table

The EDS results of the corrosion products on specimen surfaces after 7 days of incubation in acidic soil solution without and with riboflavin are presented in Table 1. To clarify the compositional differences of corrosion products, the EDS point scanning was performed at two distinct locations on the specimen surfaces in both environments. The main elements of the corrosion products on the specimen surfaces in both environments were Fe, O, S and C. The element S was detected on the specimens surfaces, which was mainly related to the metabolism of SRB32. After the addition of riboflavin, the S content on the specimen surfaces increased, primarily attributable to riboflavin-mediated acceleration of SRB metabolic activity24.

Corrosion morphology

The corrosion morphology of the specimens in acidic soil solution without and with riboflavin is depicted in Fig. 3. Apparently, numerous hemispherical corrosion pits were noted across all specimen surfaces in both environments, indicating localized corrosion characteristics of the specimens in these environments. In the identical environment, localized corrosion pits predominantly occurred on the BM and WZ surfaces, whereas the SRB-induced localized corrosion on the HAZ surface was less severe than that on the BM and WZ surfaces. Furthermore, the density and dimension of corrosion pits greatly increased with the addition of riboflavin. These observations collectively demonstrated that riboflavin promoted localized corrosion of specimens in acidic soil solution.

Fig. 3: The corrosion morphologies on welded joint.
figure 3

a, b The corrosion morphologies on the BM surface in acidic soil solution without riboflavin; c, d the corrosion morphologies on the HAZ surface in acidic soil solution without riboflavin; e, f the corrosion morphologies on the WZ surface in acidic soil solution without riboflavin; a′, b′ the corrosion morphologies on the BM surface in acidic soil solution with 10 ppm riboflavin; c′, d′ the corrosion morphologies on the HAZ surface in acidic soil solution with 10 ppm riboflavin; e′, f′ the corrosion morphologies on the WZ surface in acidic soil solution with 10 ppm riboflavin.

Full size image

Figure 4 displays the 3D morphology of the corrosion pits with maximum depth on specimen surfaces in acidic soil solution without and with riboflavin. In the same environment, the maximum depth of corrosion pits of the HAZ surface was the smallest and its localized corrosion sensitivity was also the lowest. In the acidic soil solution without riboflavin, the maximum depth of corrosion pits of the BM and WZ surfaces were close to each other, about 12.27 μm and 11.23 μm, respectively, which were about three times the maximum depth of corrosion pits of the HAZ surface (4.48 μm). With the addition of riboflavin, the maximum depth of corrosion pits of all regions was increased. To be specific, the WZ and BM surfaces displayed elevated the maximum depth to 15.37 μm and 13.92 μm, respectively, both by approximately 25%. Whereas, the maximum depth of corrosion pits of the HAZ surface were always the smallest, only increasing by about 14% to roughly 5.13 μm. This suggested that the exogenous riboflavin can further promote localized corrosion of all specimens.

Fig. 4: The 3D morphologies of the corrosion pits with maximum depth of welded joint.
figure 4

a The 3D morphology of the corrosion pits with maximum depth on BM surface in acidic soil solution without riboflavin; b the 3D morphology of the corrosion pits with maximum depth on HAZ surface in acidic soil solution without riboflavin; c the 3D morphology of the corrosion pits with maximum depth on WZ surface in acidic soil solution without riboflavin; a′ the 3D morphology of the corrosion pits with maximum depth on BM surface in acidic soil solution with 10 ppm riboflavin; b′ the 3D morphology of the corrosion pits with maximum depth on HAZ surface in acidic soil solution with 10 ppm riboflavin; c′ the 3D morphology of the corrosion pits with maximum depth on WZ surface in acidic soil solution with 10 ppm riboflavin.

Full size image

To illustrate the corrosion pattern of the specimens, the depth and diameter of 10 corrosion pits were measured randomly, as shown in Fig. 5a. In both environments, the average pit depth and diameter were consistently the largest for the BM surface, followed by the WZ surface, with the HAZ surface displayed the smallest values. This divergent corrosion behavior was predominantly linked to preferential SRB colonization patterns within the welded joint, whereby SRB heightened localized corrosion degree of the BM and WZ while diminishing it in the HAZ29. The riboflavin notably promoted localized corrosion of all specimens, with the most pronounced increases observed in the BM and WZ. Moreover, we calculated p values for the different regions. Based on corrosion pit depth measurements, the calculated p-values for BM, HAZ, and WZ specimens were 0.0002, 0.025, and 0.0001, respectively. For corrosion pit diameters, the p-values were 1.7 × 10−6, 0.002, and 1.1 × 10−6 for BM, HAZ, and WZ specimens. All p-values were below the significance threshold of 0.05, statistically confirming significant differences in pit depth and diameter across all regions in both environments, especially the BM and WZ specimens. This indicated that riboflavin had a greater promoting effect on BM and WZ specimens.

Fig. 5: The pitting growth probability of welded joint in acidic soil solution without and with 10 ppm riboflavin.
figure 5

a The depth and diameter of corrosion pits; b the cumulative probability; c the reduced variation; d the probability of various pit.

Full size image

The cumulative probability distribution exhibited a positive correlation with pit depth, demonstrating elevated likelihood of severe localized corrosion across all specimens (Fig. 5b). Additionally, the pitting depth distributions on the BM (8.3–15.1 μm) and WZ (8.8–13.8 μm) surfaces consistently exhibited broader ranges than those on the HAZ (3.3–7.8 μm) surface, statistically confirming that the HAZ surface experienced the least severe localized corrosion. Riboflavin remarkably broadened the pitting distribution across different regions, signifying its role in enhancing SRB-induced corrosion, which in turn elevated the likelihood of pit initiation on specimen surfaces.

Upon riboflavin supplementation, the reduced variants (Y) of specimens increased, showing that riboflavin promoted pitting (Fig. 5c). Linear fitting analysis showed that only a single linear region was present on the HAZ surface in the riboflavin-free environment, with the maximum pit depth approximately was 5.7 μm, suggesting that only metastable pits may exist on the HAZ surface. Nevertheless, in the remaining specimens, two linear regions could be observed, with the emergence of these regions potentially attributed to the coexistence of metastable and stable pits33. Moreover, the increase of pitting depth and Y was mainly attributed to riboflavin, which may have facilitated the transition pitting on the specimen surfaces from metastable pits to stable pits.

Meanwhile, the pitting probability on BM and WZ surfaces persisted at obviously higher levels than on the HAZ surface, indicating that the localized corrosion sensitivity of the HAZ surface was the lowest (Fig. 5d). The pitting probability of all specimens was sharply increased after the addition of riboflavin, suggesting that riboflavin promoted the development of pitting.

OCP of welded joint

Figure 6 depicts the curves of OCP with time for different regions in acidic soil solution. In acidic soil solution without and with riboflavin, the trends of OCP in different regions were essentially the identical. The all specimens were positively shifted by about 10 mV at the beginning of the experiment, followed by a gentler positive shift. This outcome can be attributed to a synergistic interaction between the formation of SRB biofilm and its metabolic activity34. Specifically, bacteria can rapidly adhere to metal surfaces via extracellular polymers, forming biofilm. The invasive factors of biofilm and the metabolic activity of sessile bacteria can alter electrochemical processes, accelerate iron dissolution, and thereby enhance the corrosion tendency of the specimen24. Compared to the environment without riboflavin, the OCP values all shifted positively after adding riboflavin. This indicated that riboflavin can promote SRB corrosion of the welded joint. It is mainly related to the fact that riboflavin can increase the number of planktonic SRB cells (Fig. 1a), which in turn promotes the formation of SRB biofilm on the specimen surface35.

Fig. 6
figure 6

The OCP values of welded joint in acidic soil solution without and with 10 ppm riboflavin.

Full size image

EIS of welded joint

The EIS results of different regions in acidic soil solution without and with riboflavin are displayed in Figs. 7 and 8. In the low-frequency region of the Nyquist plot, the linear correlation between the real and imaginary parts, it indicated that the corrosion behavior was controlled via diffusion. Moreover, it was commonly proposed that the radius of the semicircle arc was inversely correlated with its general corrosion rate. The radius of the specimens shrinks over time and radius of the HAZ specimen was the smallest. This suggested that the SRB can promote the general corrosion, with the degree of general corrosion being relatively greater for the HAZ specimen. Meanwhile, the addition of riboflavin led to a prominent decrease in the radius of all specimens. This was mainly because riboflavin accelerated the general corrosion of the specimens36. The analysis of the Bode plot revealed that throughout the experiment, the phase angle peaks of the specimens progressively shifted toward the low-frequency region over time. This behavior indicated a gradual weakening of the protective capacity of biofilms or corrosion products on the specimen surfaces37. Furthermore, Warburg impedance was detected in low-frequency region in both environments, arising from the formation of relatively dense corrosion products38. Consistent with the Biocatalyzed Cathodic Sulfate Reduction (BCSR) theory29, SRB accelerated cathodic sulfate reduction within the biofilm. Nevertheless, extracellular dissolution of the steel matrix occurred, the transfer of the sulfate in the reaction (1) could be the rate controlling step:

$${{\rm{SO}}}_{4}^{2-}+{9{\rm{H}}}^{+}+{8{\rm{e}}}^{-}\to {{\rm{HS}}}^{-}+{4{\rm{H}}}_{2}{\rm{O}}$$
(1)
Fig. 7: The Nyquist and Bode plots of welded joint in acidic soil solution without riboflavin.
figure 7

a, b The Nyquist and Bode plots of the BM; c, d the Nyquist and Bode plots of the HAZ; e, f the Nyquist and Bode plots of the WZ.

Full size image
Fig. 8: The Nyquist and Bode plots of welded joint in acidic soil solution with 10 ppm riboflavin.
figure 8

a, b The Nyquist and Bode plots of the BM; c, d the Nyquist and Bode plots of the HAZ; e, f the Nyquist and Bode plots of the WZ.

Full size image

The EIS results were fitted using the equivalent circuit shown in Fig. 7a. In the circuit, Rs, Rf, Rbf and Rct represent the resistance of the solution, corrosion product, biofilm and charge transfer, respectively. Qf, Qbf and Qdl represent constant phase angle elements of the corrosion product, biofilm and double layer, respectively. W represents the Warburg impedance. Table 2 presents the fitting results. Plots of Rct values of all specimens in acidic soil solution without and with riboflavin are shown in Fig. 9a. In both environments, Rct values of the BM (192.1–1580.3 Ω·cm2) and WZ (695.3–1525.7 Ω·cm2) specimens were greater than that of the HAZ (281.8–513.7 Ω·cm2) specimen, and Rct underwent a gradual decline over time. In addition, Rct values of all specimens in the environment with riboflavin was consistently less than Rct values in the environment without riboflavin. It was defined here R-1 = (Rbf + Rf + Rct)−1, and the R−1 values were positively correlated with general corrosion rate. R−1 values of the HAZ specimen consistently exceeded those of the BM and WZ specimens under identical environment, indicating the highest general corrosion rate in the HAZ (Fig. 9b). In the riboflavin-free acidic soil solution, the R-1 values for BM and WZ specimens exhibited a distribution of 6.5 × 10-4–1.3 × 10−3 Ω−1·cm−2, while the HAZ specimen displayed a notably broader range of 1.9 × 10−3–3.4 × 10−3 Ω−1·cm−2. Upon riboflavin addition, R−1 values of all specimens increased obviously, indicating that riboflavin promoted the general corrosion of the specimens. Specifically, the BM and WZ specimens manifested R−1 values ranging from 4.4 × 10−3–6.4 × 10−3 Ω−1·cm−2, while the HAZ specimen recorded a more extensive range of 1.5 × 10−2–1.6 × 10−2 Ω−1·cm−2. These results highlighted that the general corrosion rate of the HAZ in acidic soil solution far exceeded that of the BM and WZ specimens. Meanwhile, the exogenous riboflavin promoted SRB corrosion in all regions.

Fig. 9: The fitting results in the acidic soil solution without and with 10 ppm riboflavin.
figure 9

a Rct values; b R−1 values; c Cdl values; d δ values.

Full size image
Table 2 Fitting results of EIS data of welded joint in the acidic soil solution without/with riboflavin
Full size table

The double electric layer of Cdl of different regions was calculated using the EIS fitting results by the following formula39:

$${C}_{\text{dl}}={{Y}_{\text{dl}}\tfrac{1}{{n}_{\text{dl}}}\left(\frac{{R}_{\text{s}}{R}_{\text{ct}}}{{R}_{\text{s}}+{R}_{\text{ct}}}\right)}^{\frac{1-{n}_{\text{dl}}}{{n}_{\text{dl}}}}$$
(2)

where Ydl and ndl are the two CPE parameters of the double layer capacitor. Throughout the experiment, the Cdl values of all specimens slowly climbed (Fig. 9c). The Cdl values of the HAZ specimen were found to be greatly elevated compared to those of the BM and WZ specimens in the same environment. Furthermore, the Cdl values of all specimens were reduced by the addition of riboflavin. This suggested that in the acidic soil solution without riboflavin, there was a higher concentration of electrons at the specimen/membrane interface.

Orazen et al.40 established the relationship between the effective thickness of corrosion product (δ) and its CPE (expressed in terms of Yf and nf) through numerical modeling:

$$\delta =\frac{{\left(\varepsilon \cdot {\varepsilon }_{0}\right)}^{{n}_{{\rm{f}}}}}{{{Y}_{{{\rm{f}}}^{{g\rho }_{\delta }}}}^{\left({1-n}_{{\rm{f}}}\right)}}$$
(3)
$$g\,\text{=}\,1\,\text{+}\,2.88{(1-{n}_{\text{f}})}^{2.375}$$
(4)

where, Yf and nf are the admittance and dispersion coefficients of the corrosion products capacitor, respectively. ε (20) and ε0 (8.85 × 10−14 F·cm−1) are the relative dielectric constants of the medium and vacuum. ρδ (1980.8 Ω·cm) is the resistivity at oxide/electrolyte interface. g is the function related to capacitance dispersion coefficient of corrosion products. Figure 9d displays the thickness of the corrosion product film obtained using Eqs. (3) and (4). For all specimens in both environments, the δ values increased over time, which indicated that the corrosion products were accumulating on the specimen surfaces. Concurrently, the δ values of the HAZ specimen remained lower than those of the BM and WZ specimens, a phenomenon predominantly linked to SRB-induced selective corrosion. In contrast, δ values for all specimens in riboflavin-supplemented environment were remarkably, higher than those of riboflavin-free.

PDP curve of welded joint

Figure 10 presents the PDP curves and corrosion current density (icorr) of different regions after 7 days of immersion in the acidic soil solution with and without riboflavin, and Table 3 depicts the fitting results. The PDP curves of all specimens exhibited analogous shapes, suggesting fundamental similarities in their corrosion processes. Upon riboflavin addition, the PDP curves of different regions shifted toward the anodic direction. From the icorr values obtained from the fitting, the icorr values of the HAZ (47.5–214.6 μA·cm−2) specimen were consistently greater than that of the BM (21.8–113.5 μA·cm−2) and WZ (26.9–159.0 μA·cm−2) specimens. This revealed that the HAZ specimen was more prone to corrosion, which was consistent with the EIS results. Meanwhile, in the acidic soil solution without riboflavin, the icorr values of all specimens were smaller than those in the acidic soil solution with riboflavin. This designated that the exogenous riboflavin promoted the corrosion rate of different regions.

Fig. 10: The PDP results of welded joint.
figure 10

a The PDP curves of welded joint in the acidic soil solution without riboflavin; b the PDP curves of welded joint in the acidic soil solution with 10 ppm riboflavin; c the corrosion current density in the acidic soil solution without and with 10 ppm riboflavin.

Full size image
Table 3 Fitting results of the Tafel polarization curve
Full size table

Discussion

In this study, it was found that the accumulation of corrosion products on the BM and WZ surfaces consistently exceeded that on the HAZ surface in the acidic soil solutions without/with riboflavin. This was mainly related to the selective adsorption of SRB. Numerous studies have demonstrated that SRB exhibit a preferential attachment to grain boundaries in carbon steel. Notably, the grain boundary density in the BM and WZ was significantly higher than that in the HAZ. Furthermore, EDS analysis revealed the presence of trace elemental Mn on the HAZ specimen, which preferentially formed toxic MnS29. This phenomenon inhibited SRB colonization of the HAZ surface.

Riboflavin functioned as an electron shuttle, facilitating electron transfer from the metal surface to microorganisms and thereby accelerating anodized iron corrosion. Collectively, these results demonstrate that riboflavin induced a 58.5% increase in weight loss of the BM specimen, directly correlating with enhanced general corrosion rates. The electrochemical results also showed a remarkable decrease in the Rct values with the addition of riboflavin. By the experiment’s conclusion, Rct values of the BM and WZ specimens decreased to below 200 Ω·cm2 and that of the HAZ specimen to below 100 Ω·cm2, which was a reduction of 74.8%, 75.0%, and 77.2% for the BM, WZ, and the HAZ specimens, respectively, compared to those without riboflavin. This suggested that the exogenous riboflavin accelerates the MIC of welded joint. This was mainly because riboflavin reduces the resistance to charge transfer between the electrode-solution interface, allowing electrons to be transferred more rapidly between the metal surface and the SRB. Moreover, corrosion product thickness derived from fitting results indicated a notable increase in all specimens following riboflavin addition. To further investigate the distribution of corrosion products in different regions of the welded joint, we used Image-Pro Plus software to statistically analyze the area fraction of corrosion products, with specific data shown in Fig. 11a. It can be observed that after the addition of riboflavin, the corrosion products on the specimen surface increased significantly. This may be attributed to the fact that during the SRB corrosion of carbon steel, the transfer of electrons from the iron matrix to the bacterial cytoplasm is the rate-limiting step of the entire EET process41. However, riboflavin can promote transmembrane electron transfer and enhance sulfate reduction, thereby accelerating the SRB-induced corrosion of carbon steel and leading to the formation of more corrosion products.

Fig. 11: The area fraction of corrosion products and the pitting acceleration factor (Kpit).
figure 11

a The area fraction of corrosion products of welded joint in the acidic soil solution without and with 10 ppm riboflavin; b Kpit of welded joint in the acidic soil solution without and with 10 ppm riboflavin.

Full size image

If the corrosion products covered the specimen surfaces, it will form an obstacle for the SRB to obtain external nutrients, resulting in the SRB living in a carbon-starved environment, and then attacked the iron substrate as a way to barely maintain the survival needs of SRB19. Wang et al.42 revealed that starved SRB enhance microbial corrosion in both the BM and WZ specimens, thereby facilitating selective SRB-induced corrosion of the welded joint. The results of this paper were consistent with these results.

To quantify the role of riboflavin in accelerating SRB-induced corrosion, we defined the pit corrosion acceleration factor (Kpit) using the average pit depth of the specimens42, as follow:

$${K}_{\text{pit}}=\frac{l}{{l}_{0}}$$
(5)

where, l and l0 represent the average pits depth on the specimen surfaces in the acidic soil solution with and without riboflavin, respectively. Obviously, Kpit values of any specimens in acidic solution without riboflavin was equal to 1. As shown in Fig. 11b, Kpit values of all specimens was greater than 1 after the addition riboflavin, indicating that riboflavin sped corrosion by enhancing the EET of SRB, promoting localized corrosion of all specimens. Meanwhile, the BM and WZ specimens may be more susceptible to riboflavin, leading to accelerated pitting.

Riboflavin, as a redox-active mediator, played a crucial role in MIC, and it can effectively promote the EET process between SRB and iron surfaces. In particular, it was noteworthy that in the molecular structure of riboflavin, the isoalloxazine ring was the critical structural part that exerts the redox effect. The chemical structure of the isoalloxazine ring contains multiple double bonds as well as nitrogen atoms, and it was these unique structural features that allow it to conduct reversible redox reactions. This allowed riboflavin to function as an electron transfer mediator between SRB and metal surfaces, establishing optimal conditions for EET43. Figure 12 provides a detailed structural schematic of the interconverted forms of riboflavin.

Fig. 12
figure 12

Structure of the interconversion of oxidized riboflavin and reduced riboflavin.

Full size image

When riboflavin accepted electrons, it transitions to a reduced state. In this configuration, riboflavin gains the capability to shuttle electrons to the metal surface, the process accelerated electron transfer rate of the metal interface. This enhanced electron flux renders metal atoms more susceptible to electron loss, ultimately driving the MIC of welded joint in acidic soil solution.

The mechanism of SRB corrosion in welded joint is depicted in Fig. 13. During corrosion, SRB primarily derived energy through sulfate reduction metabolism, converting SO2-4 to HS⁻. This process consisted of two main phases: the activation phase and the reduction phase. In the activation phase, ATP sulfurylase (ATPS) catalyzed the formation of adenosine sulfate (APS, C10H15N5O10PS) and pyrophosphate (PPi, P2O-7) from adenosine triphosphate (ATP, C10H16N5O13P3) and SO2-444. In the reducing phase, APS accepted electrons and generated SO2-3 and adenosine monophosphate (AMP, C₁₀H₁₄N₅O₇P) in the presence of APS reductase (APSR)45. Finally, sulfite reductase (Dsr) catalyzed the further reduction of SO2-3 to the highly corrosive HS-, thus promoting the corrosion of specimens46.

Fig. 13
figure 13

Mechanism of D. vulgaris corrosion of welded joint in acidic soil solution.

Full size image

Riboflavin acted as an electron shuttle that captured electrons released from iron oxidation and converts them to reduced riboflavin. This electron shuttle mechanism facilitated electron transfer to c-type cytochromes of the SRB cells membrane, thereby enhancing electron flux through the APSR and Dsr pathways. The amplified electron transport capacity elevated rate of sulfate reduction, ultimately intensifying the localized corrosion15,24.

(6)
(7)
$${\text{SO}}_{3}^{2-}+7{\text{H}}^{+}+{6\text{e}}^{-}\to \text{H}{\text{S}}^{-}+{3\text{H}}_{2}\text{O}$$
(8)
$${{Riboflavin}}_{{\rm{ox}}}+{{\rm{Fe}}}^{0}\longrightarrow {{Riboflavin}}_{{\rm{red}}}+{{\rm{Fe}}}^{2+}$$
(9)
$${{Riboflavin}}_{{\rm{red}}}+c-{\rm{type\; cyto}}{{\rm{chrome}}}_{{\rm{ox}}}\longrightarrow {{Riboflavin}}_{{\rm{ox}}}+c-{\rm{type\; cyto}}{{\rm{chrome}}}_{{\rm{red}}}$$
(10)

Furthermore, Fe may be more susceptible to dissolution in acidic soil solution, providing more electron donors to SRB and thus indirectly affecting corrosion process of welded joint.

Indeed, the key findings of this study reveal not only that riboflavin accelerated corrosion across different regions of the welded joint, but also that it exacerbated selective SRB corrosion of the welded joint. The SEM micrographs clearly showed a significantly higher density of corrosion products and a larger number of pitting corrosion sites on the BM and WZ surfaces compared to the HAZ surface. This directly reflected that riboflavin-mediated EET does not occur uniformly across the different regions of the welded joint. It exhibited a stronger facilitating effect on the BM and WZ surfaces, leading to more severe localized corrosion.

This selectivity was rooted in the differential adhesion of SRB cells to various regions of the welded joint. The SRB cells preferentially attached to the BM and WZ surfaces, showing lower affinity for the HAZ surface. Hence, the lower densities of SRB cells on the HAZ surface results in reduced pit density, diameter, and depth compared to the BM and WZ. This preferential attachment primarily stemmed from the tendency of SRB cells to colonize grain boundary locations. Crucially, the BM and WZ possessed a higher grain boundary density than the HAZ15. This inherent microstructural difference led to a larger available surface area for bacterial attachment in the BM and WZ. Consequently, the majority of SRB cells colonize these two areas, driving the selective SRB corrosion of the welded joint. Furthermore, in the solution at the metal/biofilm interface, the concentration of Fe²⁺ increased beneath microbial clusters47. This created an Fe²⁺ concentration gradient between the different regions of the welded joint. Locally elevated Fe²⁺ concentrations led to the detoxification of H₂S, which in turn promoted the growth of adjacent SRB cells. Therefore, the Fe2+ enrichment increased SRB cells attachment and enhanced MIC in the BM and WZ of welded joint.

In summary, riboflavin caused a substantial augmentation in corrosion products and corrosion pits on the specimen surface. Notably, both the BM and WZ exhibited significantly higher quantities of corrosion products and deeper corrosion pits compared to the HAZ, confirming the selectivity of SRB corrosion in welded joint. This was mainly attributed to the stronger ability of riboflavin to accelerate EET on the BM and WZ surfaces. Riboflavin can greatly promote SRB corrosion in acidic soil solution, which emphasizes the role of riboflavin in enhancing Fe0-to-microbial electron transfer for D. vulgaris corrosion, leading to the following conclusions:

  1. (1)

    The metabolic activity of D. vulgaris accelerated the corrosion of the welded joint in acidic soil solution. The riboflavin greatly enhanced the MIC process of the welded joint.

  2. (2)

    The experiments above in the both environments primely reproduced the EET process in real soil environment. The riboflavin intensified both general and localized corrosion of welded joint. Moreover, welded joint always remained the selective SRB corrosion in the acidic soil solution.

  3. (3)

    In acidic soil solution without/with riboflavin, the thickness of corrosion products exhibited a distinct gradient across welded joint, with the HAZ showing clearly thinner deposits than the BM and WZ. The riboflavin facilitated the formation and growth of microbial clusters on the BM and WZ surfaces, while its promoting effect on the HAZ surface was relatively weak.

Methods

Materials, bacteria, solution

The welded joint utilized was obtained from a pipe that had been replaced in a pipeline in China, and its elemental composition is as follows (wt%): 0.0466 C, 0.204 Si, 1.754 Mn, 0.0082 P, 0.0009 S, 0.206 Ni, 0.235 Cr, 0.174 Cu, 0.0524 Nb, 0.0022 V, 0.0142 Ti, 0.125 Mo, 0.0265 Al, 0.0004 B, and Fe balance. Several separate regions (BM, HAZ, WZ) of the welded joint were prepared by wire-cutting method. Prior to the experiment, the backside of the electrochemical specimen was welded to a copper wire, and its non-working surface was sealed with epoxy resin. Metallographic photographs of different regions of welded joint are depicted in Fig. 14. The BM mainly consists of uniformly distributed equiaxed ferrite and bainite, the HAZ as mainly granular Bainite with obvious boundaries of prior Austenite, and the WZ consists of fine ferrite and Widmanstätten structures. In addition, it can be clearly found that the grain size of HAZ is larger than that of WZ and BM, indicating that the number of grain boundaries in HAZ is less. Due to the restricted dimensions of the HAZ and WZ, the weight-loss experiment was only conducted on the BM (40 mm × 20 mm × 4 mm).

Fig. 14: Microstructure of welded joint.
figure 14

a BM; b HAZ; c WZ.

Full size image

All working surfaces were carefully polished using aluminum oxide water-resistant sandpaper and silicon carbide water-resistant sandpaper, followed by washing with distilled water and anhydrous ethanol. Before the experiment, all samples were placed in a closed vacuum glove box and sterilized by ultraviolet light (UVC 10 W, Antione) for more than 30 minutes to avoid the contamination with bacteria.

The SRB strain utilized in this study was Desulfovibrio vulgaris (D. vulgaris). The solution to cultivate the D. vulgaris strain was ATCC 1249 medium, as detailed in Table 4. The pH values of the medium were adjusted to 7.0 ± 0.2 and the dissolved oxygen in the medium was removed with pure N2 (99.99%). Then, the medium were sterilized in an autoclave at 121 °C for 30 min. Prior to the experiment, the SRB strain was activated in ATCC 1249 medium for three generations to obtain a more active and reproducible SRB strain for subsequent experiments12,15,38.

Table 4 Chemical composition of 1 L ATCC 1249 medium
Full size table

The soil was obtained from Xiangtan, Hunan Province (27.8811° N, 112.0936° E). Before preparing the acidic soil solution, the soil underwent a pretreatment process: it was first oven-dried at 105 °C for 12 h, then ground using a grinder, sieved through a 1 mm sieve via shaking, and subsequently mixed with distilled water at a 1:1 weight ratio (soil/water). This mixture was left to stand for 12 h, after which the supernatant was filtered to obtain the acidic soil solution. Table 5 presents the pH and ionic concentrations of this solution.

Table 5 The pH value and ion concentration of the acidic soil solution (mg/L)
Full size table

The experimental solution was a 1:100 (vol%) mixture of ATCC 1249 medium and acidic soil solution, deoxidized with pure N2 for 2 h, sterilized in an autoclave at 121 °C for 30 min, and then cooled naturally to room temperature and stored in a refrigerator at 4 °C.

Experimental procedure

The above deoxygenated and sterilized acidic soil solution was cooled to room temperature and then inoculated with 1 vol% D. vulgaris strain, followed by the addition of riboflavin at concentrations of 0 ppm and 10 ppm (w/w), respectively. After deoxygenation by pure nitrogen gas for 2 h, the experimental setup was sealed with silicone. The experimental setup was placed in a thermostatic container for 7 days at 30 ± 2 °C. Three BM、HAZ and WZ specimens of welded joint were collected for parallel experiments.

The planktonic D. vulgaris cells in the acidic soil solution was measured periodically using the most probable number (MPN) method (GB/T 14643.5-2009). During the experiment, 3 mL of experimental solution was withdrawn from the experimental apparatus (1 L experimental solution) on 1, 2, 3, 5, and 7 days, and the SRB cells concentration was detected by sequential dilution in three sets of KBC-SRB simple bacterial test bottles (10 mL). Subsequently, the test bottles were stored in a biochemical incubator for 7 days to obtain the number of planktonic D. vulgaris cells. Sampling methods and subsequent procedures were conducted in accordance with the HJ 494-2009 standard.

Weight loss

The initial mass of the specimen was weighed and recorded using an electronic analytical balance (LICHEN, FA324C). The rust remover (500 mL of 18 wt% hydrochloric acid, 500 mL of deionized water, and 3.5 g of hexamethylenetetramine) was configured to remove corrosion products from the sample surface. After sequential rinsing with distilled water and anhydrous ethanol, the specimens were air-dried with cold air, and their final mass was recorded. The corrosion rate was calculated using the following formula (CR, mm·y−1)48:

$${CR}=\frac{87600\times ({m}_{1}-{m}_{0})}{\rho {tA}}$$
(11)

where, ρ, t, and A represent the density (7.85 g·cm−3), immersion time (h), and exposure area (cm2) of the specimen, respectively. m0 and m1 represent the initial and final masses of the BM specimen. To validate result accuracy and reliability, three BM specimens were placed in the acidic soil solution without and with riboflavin. The specimens should be repeatedly weighed three or more times until the mass error is within ±0.2 mg, with corrosion rate deviations visualized as error bars.

Electrochemical measurements

The electrochemical experiments were conducted using a workstation (Wuhan Corrtest Instruments Corp., Ltd.). The three-electrode system comprised the specimen as the working electrode (WE), a platinum sheet as the counter electrode (CE), and a saturated calomel electrode as the reference electrode (RE). Electrochemical impedance spectroscopy (EIS) was performed at 2, 3, 5, and 7 days, with potentiodynamic polarization (PDP) measured at 7 days. For electrochemical measurements, 10 mV signal amplitude was applied with a sinusoidal voltage sweeping from 1 × 105–5 × 10−3 Hz. The EIS data were analyzed by an appropriate equivalent circuit model and fitting it with Zsimpwin software. Using a scan rate of 0.5 mV·s⁻¹, PDP curves were recorded from ±250 mV relative to the OCP. The acquired data were processed by Origin 8.0. All experiments were repeated at least twice.

Morphology characterization

Following 7 days of immersion, the specimens were retrieved and individually submerged in 5.0 wt% glutaraldehyde solution for 5 h to fix the biofilm. Subsequently, a graded ethanol dehydration series (30%, 50%, 70%, and 100%) was applied, with each immersion lasting 20 min to progressively dehydrate the biofilm. At last, the specimens were dried and stored in a vacuum glove box for subsequent experiments.

A scanning electron microscope (SEM, Zeiss EVO) equipped with an energy dispersive spectrometer (EDS, OXFORD X-MaxN) was employed to characterize the morphology and composition of corrosion products on the specimen surfaces. After that, the corrosion morphology of the specimens was recorded using SEM after removal of the corrosion products, and the size of the corrosion pits was determined using a three-dimensional microscope (VHX-2000).

Stochastic model for pitting growth probability

Pitting corrosion is typical manifestation of SRB-induced corrosion49. Recent findings have revealed a correlation between the initiation and growth stages of multiple corrosion pits. This correlation can be explained by the use of a nonhomogeneous Markov process, which allows for probabilistic predictions of the growth patterns. The Gumbel extreme value distribution is a statistical model that is frequently employed to make probabilistic predictions in this context.

In this work, we performed an extreme value statistics analysis of corrosion pit depth, the cumulative probability F(n) is mathematically defined as follows50:

$$F(n)=\frac{n}{N+1}$$
(12)

where N denotes the total number of all pits, and n represents the rank in ordered extreme value data (n = 1, 2, 3 …… N).

Furthermore, the reduced variation (Yn) serves as a parameter to characterize the distribution shape, and its mathematical expression is given by the following equation51:

$$Y{\text{n}}=-{\text{In}}(-{\text{In}}(F{(}_{{\text{n}}})))$$
(13)

In this paper, we used straight-line fit Yn = a·d + b to distinguish between metastable pits and stable pits. d indicates the pit depths. We defined α = 1/b and μ = -a·α, the Gumbel extreme value distribution was applied to model the probability distribution of maximum pit depth, with its predictive capability enabled by a double-exponential mathematical framework51,52:

$$P=1-\exp \left(-\exp \left(\frac{-(x-\mu )}{\alpha }\right)\right)\sqrt{S}$$
(14)

where μ denotes the central parameter, which characterizes the concentration trend of pit depth; α acts as the scale parameter, governing the distribution width, and S signifies the total surface area of the specimen, which we assumed here to be 1 cm2. A plot of probability P against corrosion pit depth d was generated to visually characterize the corrosion severity on the specimen surface.