In Situ Evaluation of Carbon Steel Corrosion under Salt Spray Test by Electrochemical Impedance Spectroscopy

The corrosion monitoring was performed under SST conditions by using an EIS setup with a two-electrode cell, as shown in Fig. 1a. Carbon steel (SM490) plates of four different sizes, 10 mm × X_w (0.2, 0.5, 1, and 10 mm), were used as specimens. The chemical composition is shown in Table I. For the cell, a pair of identical carbon steel plates were embedded in epoxy resin with a gap of X_g = 50 μm. The cell surface was polished with emery papers up to #600 grade and then ultrasonically cleaned in ethanol.

Zoom In Zoom Out Reset image size

The corrosion test was conducted by using an SST instrument (Suga Test Instruments Co., Ltd model CASSER-IIR-ISO-3 type). The electrochemical cells for EIS measurements and the coupon specimens for mass loss measurements were placed in the SST chamber. A 5% NaCl solution was sprayed on the cell and the specimen surfaces with 0.093 MPa of atomizing over pressure at 35 °C.

The EIS measurements were taken in situ during the SST using the cell depicted in Fig. 1 with an electrochemical unit (Solartron 1280B), at an AC voltage amplitude of 10 mV and a frequency range of 10 kHz to 10 mHz. The obtained EIS was analyzed with a software (Scribner Associates, Inc. ZView version3.5e). In EIS measurements under a thin solution film, the current distribution over the electrode should also be considered,¹¹ and the width of the electrode (X_w in Fig. 1) is an important factor in determining this current distribution. First, to determine the optimum electrode size, EIS measurements were obtained from the SST chamber using four electrochemical cells, each containing an electrode of different width (X_w = 0.2, 0.5, 1, and 10 mm), with the cells placed at an angle of 75°. Later, using measurements obtained from the cell with the optimum electrode size, the corrosion behavior of carbon steel was investigated by using EIS. Special attention was paid to the angles at which the cells were fixed: 0° (horizontal), 30°, 45°, 60°, 75°, and 90° (vertical). The corrosion test duration was 24 h.

Using the same carbon steel type as that used in the EIS measurements, the average CR of a plate (50 mm × 50 mm × 2 mm) was determined by measuring the mass loss after a 24 h exposure to salt spray. For this test, the specimen surface was first ground to #600 grit finish using SiC papers. The specimens were then mounted on 70 mm × 150 mm × 2 mm polyvinyl chloride boards. The exposure angles of the specimens were 0°, 30°, 45°, 60°, 75°, and 90° to the horizontal, same as those used in the EIS measurements. The test duration was 24 h. The rust was removed by the immersion in solution of 10% diammonium hydrogen citrate C₆H₁₄N₂O₇ + 0.3% inhibitor (Asahi Chemical Co., Ltd, IBIT No. 30AR).

The average solution film thickness X_f at each exposure angle was roughly estimated by measuring the solution resistance R_sol between the gap distance of the two electrodes by EIS, assuming that a uniform liquid film was maintained on the electrode surface under SST. For R_sol measurement, a two-electrode cell shown in Fig. 1b was employed, with electrodes of 10 mm × 0.5 mm (X_w = 0.5 mm) dimensions, and with a gap of 10 mm between the two electrodes. To accurately measure the R_sol value, we employed electrodes with a smaller X_w and a larger gap than in those used in the corrosion monitoring cell (Fig. 1a). The surface of the electrode cell was ground to #600 grit finish using SiC papers.

The relationship between the R_sol and X_f values of the cell of Fig. 1b was obtained in advance as follows. The 5% NaCl solution films of various thicknesses, with the same concentrations as used in the SST, were laid on the horizontally fixed cell surface in a chamber at 35 °C. R_sol was obtained from the impedance at 10 kHz. The X_f value was calculated from the volume of the solution film placed on the cell surface, assuming that the shape of the solution film can be approximated by a disk with a diameter of 30 mm (=diameter of epoxy resin), since X_f (100 μm ∼ 2 mm) of the solution film employed was sufficiently smaller than the diameter. Using the obtained relation between the R_sol vs X_f, the X_f value for each angle of the electrode in the SST was roughly estimated from the monitored R_sol during SST.

Figure 2 shows the EIS measurements after 24 h exposure in the SST chamber using cells with the four different X_w values exposed at 75°. Note that the absolute value of impedance Z is not normalized by the electrode surface area, as it may be necessary to consider non-uniform current distribution. The electrode and solution film interface can be expressed by a one-time constant equivalent circuit comprising the solution resistance in series with a parallel circuit of charge transfer resistance R_ct and double layer capacitance. In this equivalent circuit, R_ct is given by the impedance at the low frequency limit Z_Low and the reciprocal of R_ct is proportional to the corrosion current i_corr.²⁶ Accordingly, for all the electrodes illustrated in Fig. 2, the converted value of Z_Low (Ω cm²) value per unit surface area must be equal. Indeed, it can be observed that the converted Z_Low (Ω cm²) values at 10 mHz for X_w = 0.2, 0.5, or 1 mm are almost equal. However, the converted Z_Low (Ω cm²) value for X_w = 10 mm is approximately four times larger, which means that at the low frequency limit, the current was evenly distributed for X_w values of 0.2, 0.5, or 1 mm, whereas it was unevenly distributed for X_w = 10 mm. The current distribution can be evaluated from the phase shift θ because it strongly depends on the frequency.^11,12,15 As the frequency increases, the current tends to concentrate near the edges between the two electrodes (Fig. 1), and as the frequency decreases, the current becomes more evenly distributed over the electrodes. In the previous papers,^11,12 it was reported that when the θ exceeds 45° at a certain frequency f_crit, the current distribution changes from non-uniform to uniform at the f_crit. In a lumped constant equivalent circuit with uniform current distribution, the impedance Z can be expressed as Z ∝ exp (jθ). In this case, θ varies from 0 to 90 °. On the other hand, Z is given by Z ∝ exp (jθ/2) in a 1D distributed constant equivalent circuit with non-uniform current distribution. In this case, θ varies from 0 to 45 °. Therefore, if θ exceeds 45 °, the current will be evenly distributed over the electrodes. It can be observed from Fig. 2 that θ exceeds 45° except for X_w = 10 mm. This indicates that the current distributions for X_w = 0.2, 0.5, and 1 mm became uniform in the lower frequency region than the f_crit and an accurate R_ct can be obtained from Z_Low. Conversely, in case of X_w = 10 mm, it indicates that the current was unevenly distributed at the low frequency limit because the θ value was smaller than 45° in the whole frequency region. Thus, in case of X_w = 10 mm, i_corr will be underestimated if Z_Low is assumed as R_ct.

Zoom In Zoom Out Reset image size

In case the solution film thickness over the electrodes is very less, as in the SST environment, a TML equivalent circuit that takes into account the current line distribution should be used to determine R_ct.^11,15 The TML equivalent circuit is shown in Fig. 3, where, X_w: width of the electrode in Fig. 1; R_ct^*: charge transfer resistance per unit length; CPE^*: constant phase element per unit length; R_s^*: solution resistance per unit length; and R_sol: solution resistance between the two electrodes. The results of curve-fitting the EIS data to the TML are represented by the lines in Fig. 2. It is evident that the fitting curves are in good agreement with the experimental data (symbols). The obtained parameters from the curve-fitting are summarized in Table II. The obtained R_ct value for X_w = 10 mm almost accorded with that for X_w = 0.2, 0.5, or 1 mm.

Zoom In Zoom Out Reset image size

As mentioned above, an accurate R_ct can be obtained by curve-fitting the obtained EIS data to the TML equivalent circuit, even if large electrodes (large X_w) are employed. However, the curve-fitting becomes extremely difficult if the current distribution is significantly concentrated on the near edges between the two electrodes, and if the electrode surfaces are covered with thick rust layers. To save time, the impedance observed at 10 mHz was adopted as R_ct without curve-fitting and by using a smaller electrode.

From the viewpoint of current distribution, a smaller electrode is better. However, if X_w is smaller than the solution film thickness X_f, the oxygen diffusion through the solution film changes from one-dimensional to two-dimensional, which may accelerate the CR. In addition, smaller electrodes increase the external noise during EIS measurement. Thus, X_w = 0.5 mm was selected as the optimum electrode size.

Based on the above discussed examination, electrochemical cells with X_w = 0.5 mm and X_g = 0.5 mm were fabricated, and EIS measurements were obtained at cell exposure angles of 0° (horizontal), 30°, 45°, 60°, 75°, and 90° (vertical) in the SST chamber. Figure 4 shows the EIS measurements after 24 h of exposure. Comparing the high-frequency impedance at 10 kHz (Z_10kHz), the horizontal installation (0°) displayed the smallest value probably because the solution film covering the surface was the thickest. For the cell angle range of 0°–60°, Z_10kHz increased with the angle because the solution film became thinner and the value decreased even more with a further increase in the angle from 75° to 90°. Z_10kHz increased in the order Z(0) < Z(30) = Z(75) < Z(90) < Z(45) < Z(60), but there were no significant differences. The details are discussed in later sections.

Zoom In Zoom Out Reset image size

The current distribution on the electrode was evaluated with θ. As θ at 0°, 75°, and 90° exposure angles exceeded 45°, it can be concluded that the current distribution is uniform at the low frequency limit. On the other hand, θ at 30°, 45°, and 60° exposure angles did not exceed 45°. This is not due to the onset of uneven current distribution, but rather to the formation of a thick rust. If the CPE parameter α is unity, the critical phase shift θ_crit required for judging whether the current is even or uneven is 45°. In other words, if θ exceeds 45°, the current distribution will be uniform at the low frequency limit, otherwise it will be considered non-uniform. However, if α is smaller than 1, the θ_crit value decreases as α is smaller.²⁷ In the EIS values shown in Fig. 4, as the electrode surface was covered with a thick rust, the value of α should be significantly small and θ_crit is expected to be significantly lower than 45°. As will be shown later, the CRs estimated from the impedance at 10 mHz (≈ R_ct) and from corrosion mass loss values agree very well. Therefore, it can be concluded that the current distribution in the low frequency limit was uniform at all angles.

In this study, the impedance at 10 mHz was taken as the charge transfer resistance R_ct. The corrosion current density i_corr (A cm⁻²) at each angle was estimated with the Stern-Geary equation (Eq. 1) by using R_ct (Ω cm²) and a proportional constant k of 20 mV,²⁸

$$\begin{eqnarray}&&{i}_{{\rm{corr}}}={\rm{k}}/{R}_{{\rm{ct}}}\end{eqnarray} \tag{ 1 }$$

Then, the i_corr value was converted into the CR (mm y⁻¹). The CR observed after 24 h of exposure was 0.36 mm y⁻¹ at 0°, 2.53 mm y⁻¹ at 30°, 2.45 mm y⁻¹ at 45°, 1.77 mm y⁻¹ at 60°, 1.83 mm y⁻¹ at 75°, and 1.12 mm y⁻¹ at 90°. The CR of horizontally fixed carbon steel was the lowest, the CRs for specimens at 75° and 90° angles were the next lowest, and the CRs for 30°, 45°, and 60° were similar. The CR in the SST chamber seemed to be significantly higher than that in the bulk NaCl solution under static conditions. The details of the corrosion mechanism are discussed in a later section.

To measure the corrosion mass loss at the same cell angles as those used for the earlier tests, coupon specimens created from the same carbon steel as that constituting the electrodes were fixed on polyvinyl chloride boards in the SST chamber. After 24 h exposure, the coupons were removed from the chamber, the photos of the surface appearance were taken, and then the corrosion mass loss was measured after removing the corrosion products.

The surface appearance at different angles is shown in Fig. 5. It can be seen that the corrosion morphology is significantly different for various exposure angles. The specimen surface exposed at 0° (horizontal) angle was completely covered with a relatively thin red rust. It was visible from the amount of corrosion products that the CR was much lower than that for the other angles (30°–75°). The surfaces at 30° and 45° angles were partially covered and almost completely covered with thick red rust at 60° and 75°, with a striped form. Probably, during the SST, the solution film drifted downward on the surfaces at 30–75° angles. On the surface exposed at 90° (vertical) angle, several dots of red rust were observed, where the solution may have remained in the form of droplets. The different morphologies may have resulted from the different states of the solution film on the specimen surfaces.

Zoom In Zoom Out Reset image size

Figure 6 shows the average CRs determined from the corrosion mass loss of specimens exposed at different angles during 24 h of exposure in the SST. The specimen fixed horizontally (0°) had the lowest CR of approximately 1 mm y⁻¹. The CR of the specimen fixed vertically (90°) was the second lowest (approximately 1.5 mm y⁻¹). The CRs for specimens fixed at 30°, 45°, 60°, and 75° were similar (approximately 2 mm y⁻¹). The CRs calculated from the R_ct value are also shown in Fig. 6. The tendency of the angle dependency agreed well between the mass loss and EIS results. The absolute CR values for all angles were similar except for 0°. The CR determined with EIS was less than half of that calculated from the mass loss at 0°. Notably, the CR obtained from the mass loss observations is the average value for 24 h SST, whereas the CR obtained from EIS measurements is the instantaneous rate observed just at the completion of 24 h exposure, which suggests that the CR decreased with exposure time at 0° and did not change with time at the other angles.

Zoom In Zoom Out Reset image size

The R_sol measurements with the cell described in Fig. 1b are shown in Fig. 7. At angles of 0°, 30°, 45°, 60°, and 75°, the R_sol value declined rapidly during the initial 3 ∼ 5 h and then reached a steady state. This decrease is due to the surface wetting by the salt spray. On the other hand, at 90°, R_sol fluctuated greatly during monitoring, indicating that the solution film was unstable in a vertical position. Therefore, vertical exposure was excluded from the analysis of solution film thickness.

Zoom In Zoom Out Reset image size

Figure 8 shows the relation between R_sol and X_f measured for samples with thicknesses ranging from 200 to 2000 μm, using the cell described in Fig. 1b. The dotted line in the figure is the result obtained from the electrical conductivity (к = 0.086 S cm⁻¹)²⁹ of 1 M NaCl (5.9%) solution at 25 °C. For the calculation, the length, width, and height of the solution were 1 cm, 1 cm, and X_f, respectively (Fig. 1b). As shown in Fig. 8, the measured points are on the dotted line when the solution film is thick, but deviate from the dotted line for a thin film. At this point, the deviation cannot be clearly explained, but using this standard curve, the thicknesses of the solution films in SST, estimated roughly from the R_sol values at 6 h in Fig. 7, were obtained as approximately 3000 μm at an exposure angle of 0°, 500–600 μm at 30°, 100 μm at 45°, and 60 μm at 60°. It may be difficult to estimate it at 75° because R_sol (approx. 2 × 10⁴ Ω) at that angle was significantly out of the standard curve range. Assuming that there is a linear relationship between R_sol and X_f on the thinner side of the film, the extrapolation of the straight line provides an estimate of a few tens of a micrometer for X_f.

Zoom In Zoom Out Reset image size

It is well known that the corrosion of carbon steel in a neutral environment proceeds by following a combination of anodic iron dissolution reaction (Eq. 2) and oxygen reduction reaction (Eq. 3).

$$\begin{eqnarray}&&{\rm{Fe}}\to {{\rm{Fe}}}^{2+}+2{{\rm{e}}}^{-}\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&{{\rm{O}}}_{2}+2{{\rm{H}}}_{2}{\rm{O}}+4{{\rm{e}}}^{-}\to 4{{\rm{OH}}}^{-}\end{eqnarray} \tag{ 3 }$$

The rate-determining step (rds) of the corrosion process is the oxygen diffusion. During the atmospheric corrosion of carbon steel, the rate of oxygen reduction reaction (ORR) is controlled by diffusion through a solution film on the surface. Nishikata et al. have also previously reported on atmospheric corrosion mechanism.¹² When the solution film reaches a thickness of approximately 300 μm or more, the CR is independent of X_f. This is because the thicker solution film is composed of two layers: an inner diffusion layer and an outer convection layer. An increase in X_f only increases the convection layer thickness. On the other hand, a solution film thinner than approximately 300 μm is composed of only a diffusion layer, and a decrease in X_f promotes the diffusion of oxygen. The CR increases almost proportionately to the reciprocal of the solution film thickness (X_f⁻¹).¹² However, the CR shows a maximum value of a few tens of a micrometer, and when the solution film is further thinned, the CR starts to decrease. It is considered that this decrease may be due to the fact that the rate of anodic dissolution reaction (Eq. 2),³⁰ that is, the ionization reaction of Fe, is strongly suppressed. Simultaneously, when X_f is a few tens of a micrometer or less, the rds of the ORR also changes from being a diffusion process in the solution film to being a dissolution process of oxygen molecules at the air/solution film interface. Thus, the ORR rate becomes independent of X_f.¹²

The corrosion mechanism in SST should be considered from the point of view of its dependence on the solution film thickness. It is clear from the R_sol results given in Fig. 7 that X_f changes by changing the exposure angle. Conversely, looking at the dependence of the CR on the exposure angle in Fig. 6, X_f is approximately 3 mm at 0° (horizontal) from Fig. 8, and the CR is the lowest among all the angles (approx. 1 mm y⁻¹). For the other angles, X_f was roughly estimated to be 500–600 μm at 30°, approximately 100 μm at 45°, 60 μm at 60°, and a few tens of a μm at 75°. Although X_f seemed to have changed greatly, the CR exhibited an almost constant value of roughly 2 mm y⁻¹. The dependence of CR on X_f during the SST was different from that under static solution film conditions.¹² This may be attributed to differences in the way oxygen is supplied to the specimen surface; that is, under static conditions, oxygen is supplied to the specimen surface by diffusion through the solution film, whereas in SST, fresh oxygen-containing solution is constantly supplied to the specimen surface. This suggests that in SST, the diffusion process of oxygen through the solution film is not completely rate-determining. In other words, the dependence of the CR on X_f in SST is lower than that in stationary solution film test. This can also be confirmed from the fact that the obtained CR in SST was significantly higher than that under static conditions. For example, the horizontal specimen set had an average CR of 1 mm y⁻¹, which was the lowest among all angles. The horizontal specimen was covered with an approximately 3 mm-thick solution film of 5% NaCl in SST. If a 3 mm-thick solution film is present on the surface in a static condition, the CR should be almost the same as that in a bulk solution under natural convection.²⁹ The CR of carbon steel in 5% NaCl solution at 35 °C for 24 h was determined as 0.19 mm y⁻¹ after mass loss measurement. It is evident that the CR of carbon steel under the employed SST conditions was more than five times higher than that under the stationary system, which means that there is an extremely high supply of oxygen to the specimen surface in SST.

In the above corrosion mechanism, the effect of rust was not considered. The low angular dependence of CR and accelerated corrosion can also be explained by the formation of rust, if FeOOH acts as the main oxidant of steel corrosion. In CCT, which includes a drying step in each cycle, electrochemically active FeOOH is formed in the drying step and promotes corrosion by acting as a strong oxidant in the subsequent wetting step.^31,32 Also, since the main oxidant changes from O₂ to FeOOH, the CR may be independent of the solution film thickness. However, in SST without the drying step, it may be difficult to form active FeOOH because the sample surface is always wet.