Introduction

Corrosion has been a great concern in the oil and natural gas (O&G) industry because it adversely affects the structural infrastructure in exploration, production, processing, and transport of oil and natural gas and can cause significant economic costs and safety issues.1,2,–3 Internal corrosion can occur in the natural gas transmission pipelines when aqueous electrolytes are present. Inside the pipelines, electrochemical corrosion takes place in the water phase condensed from wet gas or from liquid water because of upstream plant upsets. The inherently existing acidic gases such as CO2 and H2S can dissolve in the water phase forming corrosive electrolytes.4 Over the last 30 years, corrosion has caused ~ 25% of the incidents in natural gas transmission and gathering pipelines, and 61% of corrosion caused incidents were due to internal corrosion, according to the Pipeline and Hazardous Materials Safety Administration (PHMSA) database.5,6 Natural gas constitutes 32% of the energy consumption in the US in 2019 according to the US Energy Information Administration (EIA).7 The natural gas delivery system includes 528,000 km (328,000 miles) of natural gas transmission and gathering pipelines.8 It has been challenging to monitor internal corrosion effectively as the inside of the pipeline is not readily accessible during regular maintenance and inspection. Moreover, corrosion can occur at some random locations inside and outside pipelines over thousands of miles. Therefore, it is important to locate and monitor corrosion events and corrosive conditions along the long-distance pipeline to effectively mitigate corrosion and ensure safe, secure, and reliable energy transportation.9

The ability to monitor internal corrosion in natural gas transmission pipelines before the structural integrity gets compromised could have a significant impact on preventing methane leaks as well as catastrophic events resulting from corrosion. The fiber optics sensing platform provides many advantages and has been studied for sensing a broad range of chemical and physical parameters, including corrosion monitoring.10,11,12,13,14,15,16,17,18,–19 Compared with other sensing platforms, the fiber optics platform has favorable features such as distributed sensing, non-destructive monitoring, in situ real-time measurements, small size, flexibility, light weight, inherent immunity to electromagnetic interference (EMI), compatibility to optical fiber telecommunication, and improved safety in the presence of flammable gas compared to electrical-based sensors. Spatially distributed monitoring makes the fiber optics sensing particularly suitable for long-distance infrastructures (e.g., pipelines) in the O&G industry, enabling continuous real-time monitoring over the whole structure with location information. To take advantage of fiber optics sensing platform, corrosion proxy sensing materials are a key enabling component.

A recent concept for early corrosion onset detection involves the use of proxy materials integrated with the fiber optics sensing platform, which provides insight into the corrosive conditions in the pipelines. Successful realization of this class of sensors requires a detailed understanding of the corrosion behaviors of the proxy materials. Fe or alloy thin films have been used in corrosion monitoring studies through measurements of electrical resistance or light reflection/transmission of the thin films in NaCl solutions.11,20,21,–22 However, there have been few studies on thin film coated optics in CO2 containing brine solutions, which are more representative of corrosive conditions in natural gas pipelines and other systems in the O&G industry.6,23 Therefore, in this work, Fe thin films of different thicknesses (25 nm, 50 nm, 100 nm) were tested in CO2-saturated 3.5 wt.% NaCl solutions at 30°C. The effects of CO2, pH, and film thickness were studied using optical transmission measurements and in situ electrochemical methods.

Experimental Procedure

Fe Thin Film Preparation

Three thicknesses of Fe thin films were tested (25 nm, 50 nm, and 100 nm), and a 400-nm-thick film was only used in some electrochemical tests. Fe thin films were prepared by sputter deposition onto one side of the fused quartz disk substrates, Ø 25.4 mm × 1.6 mm (Ø 1 in. × 1/16 in.). The substrates were cleaned with isopropanol and dried before deposition. The deposition was completed with a throw distance of ~ 5.2 cm in a sputter-down configuration. The radio frequency (RF) sputtering was carried out with 200 W power and 10 mTorr Ar pressure. The Fe target was pre-sputtered for 10 min before deposition to remove any undesirable contamination.

Fe Thin Film Characterization

An atomic force microscope (AFM, NT-MDT Solver Next) was used to image the as-received Fe thin films on the substrates. The sample was mounted to steel spacers using adhesive to adjust the height of the sample on the AFM stage. Tap300Al-G tips (BudgetSensors) with a 30-nm aluminum reflex coating and 300-kHz resonance frequency were used in a semi-contact mode to minimize surface wear during scans. Scans were completed at 1 Hz for a 1 µm × 1 µm area.

The Fe thin films were also characterized using x-ray photoelectron spectroscopy (XPS, PHI 5600ci) with a monochromatic Al Kα x-ray source. The XPS survey spectra were obtained using the pass energy of 187.85 eV and step size of 0.8 eV. Binding energies were corrected for charging by referencing to the C 1s peak at 284.6 eV. Atomic concentrations were calculated from the XPS spectra using manufacturer-provided sensitivity factors. Ar ion gun sputtering was used to remove adventitious carbon on the top surface and to obtain depth profiles of the films.

Optical and Electrochemical Tests

Figure 1 shows the experimental system assembled in a photo-electrochemical (PEC) flow cell (ZAHNER-elektrik). The Fe-coated quartz disk was mounted as the rear window and sealed with an O-ring. The Fe film facing inward was the working electrode. A piece of copper tape (5 mm × 5 mm) was attached to the Fe film edge without touching the solution. The copper tape was in contact with a patch of conductive cloth on the conducting plate, which was electrically connected to the working electrode terminal through a soldered wire. The counter electrode was a Pt coil pre-installed in the cell and was connected to the counter electrode terminal. The Ag/AgCl reference electrode in 3 M NaCl solution was inserted in the cell as the reference electrode. The measured electrode potential was converted to the standard hydrogen electrode (SHE) scale. This setup allows simultaneous optical and electrochemical measurements.

Fig. 1
figure 1

Adapted from Ref. 6 (Color figure online).

Experimental setup of the PEC flow cell. Blue arrows represent the solution flow direction at 30°C. White arrows represent the collimated light rays.

Full size image

The flow-cell test solution contained 3.5 wt.% NaCl at 30°C at ambient pressure, which was pumped through the cell at a flow rate of 113 mL/min. Three different solution (Soln.) compositions are listed in Table I. Soln. #1 was the N2-saturated solution, Soln. #2 was the CO2-saturated solution, and Soln. #3 was the CO2-saturated solution with 0.03 molal NaHCO3 (around pH 6). pH was measured using a Mettler Toledo SevenExcellence pH meter.

Table I Modeled solution speciation, pH, and corrosion rates in the test solutions at 30°C5
Full size table

The optical transmission through the PEC cell was measured using a spectrophotometer (Ocean Optics Jaz) and an optical power meter (Thorlabs PM100D) simultaneously via a bifurcated cable. The light transmittance (T%) was normalized to a value of 100% when the Fe thin film was totally dissolved. T% was calculated following Eq. 1.

$$ T\% = \left( {I_{\text{meas}} - I_{0\% } } \right)/\left( {I_{100\% } - I_{0\% } } \right) $$
(1)

where Imeas is the measured transmitted light intensity through the PEC flow cell, I100% (light reference) equals Imeas when the Fe thin film is totally dissolved, and I0% (dark reference) equals Imeas when the light source is turned off.

The electrochemical tests were performed using a Gamry Reference 600+ potentiostat, including linear polarization resistance (LPR), electrochemical impedance spectroscopy (EIS), and linear sweep voltammetry (LSV). LPR was measured between ± 10 mV versus the corrosion potential (Ecorr) with a sweep rate of 0.125 mV/s. EIS was performed from 100 kHz to 10 mHz with an amplitude of 10 mV versus Ecorr. LSV was scanned within ± 300 mV versus Ecorr with a sweep rate of 1 mV/s. The 400-nm-thick Fe film was used for LSV because thinner films dissolved too quickly to complete the sweep before the film was gone, especially for anodic sweeps.

In a three-electrode system, when corrosion is under charge transfer control and the working electrode is polarized within a sufficiently small range around Ecorr, the corrosion current density, jcorr, can be calculated according to the Stern-Geary equation, Eq. 2.24,25,26,27

$$ j_{\text{corr}} = \left( {b_{\text{c}} b_{\text{a}} } \right)/\left[ {2.303R_{\text{pol}} \left( {b_{\text{c}} - b_{\text{a}} } \right)} \right] $$
(2)

where Rpol is the area-specific polarization resistance, and bc (negative) and ba (positive) are cathodic and anodic Tafel slopes, respectively. Rpol was obtained from LPR and corrected for the solution resistance (Rsol) measured from EIS. Tafel slopes were calculated following Eqs. 3 and 4, assuming the symmetry coefficient β to be 0.5.

$$ b_{\text{a}} = \left( {2.303RT} \right)/\left[ {\left( {1 \, - \beta } \right)nF} \right] $$
(3)
$$ b_{\text{c}} = \left( {2.303RT} \right)/\left[ {\left( { - \beta } \right)nF} \right] $$
(4)

where R represents the gas constant, T is the absolute temperature, F is the Faraday constant, and n is the electron number transferred during corrosion. The corrosion rate (CR) was calculated from jcorr following Eq. 5 assuming uniform corrosion.

$$ {\text{CR}} = \left( {j_{\text{corr}} M} \right)/\left( {nF\rho } \right) $$
(5)

where M and ρ are the molecular mass and density of Fe, respectively. n was taken as 2 for oxidation from Fe to Fe2+.

Results and Discussion

CO2 Corrosion Modeling

The solution speciation, corrosion rates, and Pourbaix diagram were modeled using OLI Analyzer Studio V9.5 based on the mixed-solvent electrolyte model.28,29,–30 CR was modeled assuming a rotating disk electrode (RDE) system at 300 rpm. The modeling results in Table I show that addition of CO2 lowers the pH compared to the N2-saturated solution, and CO2(aq) is the dominant carbonate species while HCO3−(aq) increases at higher pH in Soln. #3. The modeled pH values were consistent with the measured pH.

The dissolved CO2 lowers the pH through dissociation of H2CO3(aq) and promotes electrochemical corrosion reactions. For cathodic reactions in deaerated conditions, H2CO3(aq) could provide an additional hydrogen source for the hydrogen evolution reaction through two levels of dissociation. Direct reduction of non-dissociated H2CO3(aq) has also been proposed by some researchers.31,32,33,34,–35 For anodic reactions, CO2(aq) was found to participate in oxidation of Fe through formation of an intermediate.5,36 The anodic reaction mechanism has been frequently assumed to follow the Bockris mechanism.34,37,38,–39

The model predicts that CR of Fe increases by more than two orders of magnitude in the CO2-saturated solution (Soln. #2) compared to Soln. #1. The higher pH in Soln. #3 results in a lower CR. The Pourbaix diagram in supplementary Fig. S1 shows that the thermodynamically stable products from CO2 corrosion are soluble Fe2+(aq) at pH 4 and FeHCO3+(aq) at pH 6, respectively. This indicates active corrosion, and Fe thin films would dissolve in the CO2-saturated brine solutions.

Fe Thin Film Characterization Results

AFM Results

Figure 2 shows the AFM images of Fe thin films with different thicknesses. The grain size for each sample was determined using the concentric circles method (ASTM E112).40 The surface roughness was obtained using Gwyddion software built-in roughness analysis. The average grain size and roughness are listed in Table II, and they increased with the increasing film thickness from 25 nm to 100 nm.

Fig. 2
figure 2

AFM topography and 3D images for Fe thin films to illustrate the grain sizes and roughness.

Full size image
Table II Properties of Fe thin films with different thicknesses
Full size table

The sheet resistance (Rs) was calculated from the Fe bulk resistivity (Rb, 9.71 × 10−8 Ω·cm) and the film thickness (t) following Eq. 6,41 and the values are listed in Table II. Apparently, Rs decreases as the film gets thicker.

$$ R_{\text{s}} = R_b/t $$
(6)

XPS Results

The XPS depth profiles of Fe thin films are depicted in supplementary Fig. S2. Adventitious carbon was sputtered away within 0.1–0.2 nm of the top surface. An oxidized top layer (1.3–1.7 nm thick) was observed because of exposure to air during the sample transport (Table II). The boundary between the oxidized layer and the Fe film bulk was taken at half of the maximum oxygen content. The oxidized top layer can be protective enough to hinder corrosion of the underlying Fe, especially at higher pH, so precathodization of − 300 mV versus Ecorr was applied for < 1 min to remove this layer in some electrochemical tests. As the Fe film was sputtered away, the SiO2 substrate was revealed. Figure 3 shows the XPS spectrum of the underlying bulk of an Fe thin film, confirming metallic Fe was successfully deposited onto the substrate.

Fig. 3
figure 3

XPS result for 100 nm Fe thin film after being sputtered to 30 nm depth.

Full size image

Optical Studies of Fe Thin Film Corrosion

Light Transmittance Calculation

The light transmittance was calculated using standard Fresnel-based models of transmission through a thin film system. For a light-absorbing Fe thin film with a uniform thickness (d1) on a quartz substrate, the complex refractive index of Fe thin film is written as n1 − ik1, where n1 is the refractive index and k1 is the extinction coefficient. At normal incidence of a parallel beam of light, the transmittance (T%) of the film can be calculated following Eq. 7.42n0 and n2 are refractive indices of the solution and the quartz substrate, respectively.

$$ T\% = \frac{{n_{2} }}{{n_{0} }}\frac{{\left[ {\left( {1 + g_{1} } \right)^{2} + h_{1}^{2} } \right]\left[ {\left( {1 + g_{2} } \right)^{2} + h_{2}^{2} } \right]}}{{e^{{2\alpha_{1} }} + \left( {g_{1}^{2} + h_{1}^{2} } \right)\left( {g_{2}^{2} + h_{2}^{2} } \right)e^{{ - 2\alpha_{1} }} + C \cos 2\gamma_{1} + D \sin 2\gamma_{1} }} $$
(7)

where \( g_{1} = \frac{{n_{0}^{2} - n_{1}^{2} - k_{1}^{2} }}{{\left( {n_{0} + n_{1} } \right)^{2} + k_{1}^{2} }} \), \( g_{2} = \frac{{n_{1}^{2} - n_{2}^{2} + k_{1}^{2} }}{{\left( {n_{1} + n_{2} } \right)^{2} + k_{1}^{2} }} \), \( h_{1} = \frac{{2n_{0} k_{1} }}{{\left( {n_{0} + n_{1} } \right)^{2} + k_{1}^{2} }} \), \( h_{2} = \frac{{ - 2n_{2} k_{1} }}{{\left( {n_{1} + n_{2} } \right)^{2} + k_{1}^{2} }} \), \( \alpha_{1} = \frac{{2\pi k_{1} d_{1} }}{\lambda } \), \( \gamma_{1} = \frac{{2\pi n_{1} d_{1} }}{\lambda } \), \( C = 2\left( {g_{1} g_{2} - h_{1} h_{2} } \right) \), \( D = 2\left( {g_{1} h_{2} + g_{2} h_{1} } \right) \).

To demonstrate T% is a function of the Fe film thickness, one example is given in Fig. 4 at the wavelength of 600 nm. n0 and n2 were taken as 1.332 (water) and 1.458 (fused quartz).43,44n1 and k1 were assumed to be the same as bulk Fe for different thicknesses, n1 = 2.328 and k1 = 3.259.45T% decreases with increasing Fe film thickness as expected, with the strongest dependence occurring for film thicknesses < ~ 60 nm as thicker films are substantially opaque. Therefore, T% through the Fe thin films can serve as a measure of film thickness, which enables an optical sensor to monitor corrosion/mass loss and quantify the CR with Fe thin films as a corrosion proxy. In cases where solid corrosion products (e.g., FeCO3) may be present on the film, it will lead to a different, albeit still effective, sensing mechanism in terms of light transmission, because the corrosion products have a different complex refractive index than Fe.

Fig. 4
figure 4

Adapted from Ref. 6.

Calculated transmittance of an Fe thin film as a function of film thickness at the wavelength of 600 nm.

Full size image

Light Transmittance Measurements During Corrosion

The measured light transmittance increased as the Fe thin films became thinner during corrosion. Figure 5 shows the transmittance spectra of a 25-nm Fe film exposed to the CO2-saturated 3.5 wt.% NaCl solution at 30°C. The spectra showed a broadband T% response to corrosion between the wavelength of 400 nm and 900 nm, and T% increased over time until the Fe thin film totally dissolved in the solution. It is representative of T% measurements in all the tests in this work. For the film thickness of 25 nm, the measured T% at the wavelength of 600 nm before corrosion (at 0 min) was ~ 15%, close to the calculated T% of ~ 13% at 25 nm thickness in Fig. 4.

Fig. 5
figure 5

Optical transmittance spectra of 25 nm Fe thin film over time in CO2-saturated 3.5 wt.% NaCl solution at 30°C.

Full size image

Besides the optical spectra, transmitted light power also increased corresponding to corrosion of Fe thin films. Figure 6 depicts the plots of T% versus time measured with the optical power meter for different film thicknesses in the three test solutions. All the plots showed first a slow increase (a small slope) and then a sharp increase (a large slope) before the thin films totally dissolved. This followed the pattern of the calculated T% versus thickness in Fig. 4 with the dependence of transmittance on film thickness growing stronger when the transparency of such films increases.

Fig. 6
figure 6

T% of different thicknesses of Fe thin films as a function of time at 30°C. (a) Soln. #1: N2 deaerated NaCl solution, pH 7; (b) Soln. #2: CO2-saturated NaCl solution, pH 4; (c) Soln. #3: CO2-saturated NaCl solution, pH 6.

Full size image

Comparing the three test solutions, Fe thin films corroded faster in the CO2-saturated NaCl solution (Soln. #2) than in the N2 deaerated solution (Soln. #1) and relatively slower in Soln. #3 with higher pH than in Soln. #2. This is consistent with the CR modeling in Table I. The T% plots in Soln. #1 were not as smooth as those in Soln. #2 and #3. Some small pieces of undissolved Fe film were observed in the circulating solution deaerated with N2, suggesting a tendency for film delamination in addition to thickness reduction due to mass loss in Soln. #1. The CR was much slower in Soln. #1; however, Cl−(aq) could promote localized corrosion, weakening the structural integrity of the films.46,47 Thus, the Fe thin films became susceptible to delamination under the flow, which caused sudden increases of T% as small pieces of Fe film fell off. Localized corrosion is not as predictable as the general (non-localized) active corrosion. This may explain that the plot slopes in Soln. #1 had a larger difference among different thicknesses than those in the other two solutions.

Naturally, the thicker films took a longer time to get totally dissolved in each solution, but it is important to note that the initial CR was actually higher for the thicker films as discussed in more detail in the subsequent sections. The total dissolution times of Fe thin films when yielding 100% transmittance are summarized in Table III for comparison.

Table III Ecorr, Rpol, CR, and total dissolution time of Fe thin films compared with bulk carbon steel (CS)5
Full size table

Electrochemical Studies of Fe Thin Film Corrosion

Table III lists Ecorr, Rpol, and CR of Fe thin films from the electrochemical measurements. In each solution, Ecorr grew slightly more negative as the film thickness increased from 25 nm to 100 nm, indicating the thicker film was more prone to corrosion. From Soln. #1 to #3, Ecorr of the Fe thin films generally agreed with Ecorr of the bulk X65 carbon steel studied in an RDE system.5 API 5L X65 carbon steel is commonly used for transmission pipelines, and it is compared with the Fe thin films in Table III.

The electrochemical results showed that CO2 increased the CR of Fe thin films by more than one order of magnitude, comparing Soln. #2 with Soln. #1. The higher pH in Soln. #3 led to a decrease in CR compared with Soln. #2. This trend is in reasonable agreement with the modeled CR in Table I for the three solutions. The CR increase was caused by the lower pH and the promoted corrosion reactions in the presence of CO2.34,36,48 The higher pH in Soln. #3 had an opposite effect. Detailed corrosion mechanisms have been discussed in Ref. 5. Rpol and CR of Fe thin films in the three solutions resembled the corrosion behaviors of the bulk X65 carbon steel. This resemblance supports that Fe thin films can serve as a corrosion proxy to indicate the corrosion onset and rate of bulk pipeline steel. Additionally, as the films are much thinner than the corresponding bulk materials, they will yield a significant change in optical transmission long before significant amounts of pipeline material have corroded, capable of monitoring early corrosion onset.

The electrochemical CR was observed to increase with increasing thickness in each solution (Table III). From 25 nm to 100 nm Fe, CR increased by a factor of 6.5, 1.9, and 2.3 in Soln. #1, #2, and #3, respectively. A similar correlation between CR and film thickness was also reported in an aerated NaCl solution previously.49 As shown in Table II, the morphology and Rs of Fe films varied with thickness. Although a smaller grain size in the thinner film may lead to a higher CR,50 the thicker film has greater surface roughness and a lower Rs, which can accelerate corrosion.51 In general, CR of Fe thin film was close to that of the bulk carbon steel in each solution, and the slight discrepancy could be explained by the difference in experimental configurations between the PEC flow cell and the RDE system.

The total dissolution time was calculated from the electrochemical CR, assuming it was the average CR throughout the process, and is compared with the dissolution time obtained from light power measurements in Fig. 6. In Soln. #2 and #3, the dissolution times were in good agreement between these two types of measurements. In Soln. #1, the electrochemical CR indicated a longer corrosion process than the optical measurements. The reason could be that general corrosion was assumed for electrochemical measurements while localized/pitting corrosion and erosion probably occurred in Soln. #1, causing mechanical delamination of the Fe films that could not be fully measured electrochemically.

The EIS results confirmed active corrosion of Fe thin films in CO2-saturated NaCl solution (Soln. #2). Figure 7a shows that the Nyquist plots were suppressed semicircles with one time constant, which were fitted with a Randle circuit without diffusion. Rpol obtained from EIS was consistent with that from LPR. In the Bode plots (Fig. 7b), compared with the bulk carbon steel, the time constant of Fe thin films shifted to a lower frequency, and the magnitudes of phase angles were smaller. This may be caused by different experimental configurations between thin films and bulk carbon steel with different current distributions. Comparing the Fe thin films, the magnitude of the maximum phase angle became smaller for thicker films. It suggests that the double layer capacitance at the film/solution interface was less ideal for thicker films, corresponding to higher roughness.52,53

Fig. 7
figure 7

Adapted from Ref. 6.

(a) Nyquist and (b) Bode plots of Fe thin films compared with bulk X65 carbon steel in CO2-saturated NaCl solution at 30°C. Symbol p represents the phase angle.

Full size image

The LSV plots of 100 nm and 400 nm Fe thin films showed similarity to that of the bulk X65 carbon steel (Fig. 8). The cathodic region presented a limiting current for both the Fe thin films and the bulk X65, with almost the same value. The anodic LSV slope of 400 nm Fe was parallel to that of the bulk X65 when the electrode potential was < −0.5 V versus SHE. LSV slopes can reveal the corrosion mechanisms, which were discussed in detail previously.5,54 As the potential became more positive, the anodic current decreased for the Fe thin films, which was different from the bulk X65. This is due to gradual dissolution and depletion of Fe thin films, reducing the material and area on the substrates for corrosion instead of passivation.

Fig. 8
figure 8

LSV plots of 100 nm and 400 nm Fe thin films compared with bulk X65 carbon steel in CO2-saturated NaCl solution at 30°C.

Full size image

Despite the thickness effect, electrochemical corrosion of Fe thin films was in accordance with the bulk X65 in terms of Ecorr, CR, and LSV slopes under the test conditions. This further confirms the feasibility of using Fe thin films as the corrosion proxy material. The Fe thin films were sensitive to the presence of CO2 and pH as indicated by the significant changes in CR.

When implementing the corrosion sensor inside the pipeline, it is advisable to locate the sensor in the most susceptible locations or alternatively to enable a low-cost, distributed sensor configuration such that “at risk” segments can be identified prior to significant pipeline corrosion. Different configurations are available to integrate Fe thin films with the fiber optics to effectively take advantage of the optical fiber based sensors. For example, Fe film can be coated at the end of the optical fiber as a low-cost point sensor placed at high risk locations or can be coated along the entire fiber as a distributive sensor through more advanced sensor fabrication and interrogation methods.9,11,18,55

Conclusion

Corrosion of Fe thin films was studied using optical and electrochemical methods and compared with the bulk carbon steel. Fe thin films were sputter deposited onto the quartz substrates, and the XPS spectrum confirms metallic Fe was successfully deposited. The AFM results showed that the average grain size and roughness increased with increasing film thickness from 25 nm to 100 nm. An oxidized top layer (1.3–1.7 nm thick) was observed because of exposure to air during the sample transport. The calculated Rs decreased as the film thickness increased.

T% of Fe thin films increased as corrosion proceeded because of the film thinning. This agrees with the optical calculation using standard Fresnel-based models of transmission in a thin film system. The T% versus time plots demonstrated a faster increase of T% in the presence of CO2 than without CO2 or at higher pH. Therefore, optical responses from Fe thin films can be used to monitor corrosion onset and quantify the CR.

The electrochemical results showed that CO2 increased the CR of Fe thin films by more than one order of magnitude compared with the N2 deaerated condition. The significant increase in CR demonstrated the high sensitivity of Fe thin films to CO2 corrosion as a sensing material. The higher pH of ~ 6 with NaHCO3 decreased the CR compared with pH ~ 4. Rpol, CR, and Ecorr of the Fe thin films were in good agreement with the bulk X65 carbon steel, verifying that Fe thin films can serve as a representative corrosion proxy when integrated with the optical sensing platform.

The CR increased as the film thickness increased from 25 nm to 100 nm in each solution. Although a smaller grain size in the thinner film may lead to a higher CR, the thicker film has greater surface roughness and a lower Rs, which can accelerate corrosion. The total dissolution times were consistent between the optical and electrochemical measurements.