Effect of Humidity of Air on Hydrogen Absorption into Fe with Rust Layer Containing MgCl₂ during Atmospheric Corrosion

Abstract

An Fe plate, whose one side was electro-polished and the other covered with the rust layer containing 25.7 g·m⁻² MgCl₂, was used as the specimen to investigate the effect of humidity on the hydrogen absorption of the plate. The specimen was subjected to an electrochemical hydrogen-absorption test during which the rusted surface was exposed to the air with controlled relative humidity (RH) and atmospheric corrosion occurred on it. When the rusted surface was subjected to dry (RH 0%)–wet (RH 27%) repeated cycle tests for 10.8 ks each, the anodic current density corresponding to the hydrogen-absorption rate was measured on the hydrogen detection surface. The maximum current density was almost independent of the cycle during the first 10 cycles, after which it decreased with an increase in the cycle, reaching almost a steady-state after about 40 cycles. After 55 cycles of the dry–wet repeated cycle test, the specimen was subjected to an electrochemical hydrogen-absorption test to obtain the relationship between the steady-state hydrogen-absorption rate and RH. Hydrogen absorption was observed at RH of about 15%, and the absorption rate increased rapidly with an increase in RH, reached a maximum at RH of about 30%, and then decreased rapidly. When RH increased beyond 40%, the absorption rate increased again, reached a maximum value at RH of 80%, and then decreased gradually. The specimen with the rust layer containing 39.8 g·m⁻² MgCl₂ also showed two peaks in the hydrogen-absorption rate versus RH plot.

1. Introduction

Recently, steels with extremely high strength have been developed and used where energy-saving, environmental protection, and so on are desired. The strength of the recently developed steel has exceeded 2000 MPa. However, some of the high-strength steels suffer fracture when employed in humid conditions for a relatively long time, so called delayed fracture.¹⁾ Therefore, there is an urgent need to clarify the cause of such failure and find possible solutions. Relevant studies have proven that the delayed fracture is caused by hydrogen embrittlement (HE),²⁾ in which atmospheric corrosion induces the generation of hydrogen atoms on the steel, which are absorbed and spread the steel, resulting in HE,. Steel whose 0.2% yield strength is higher than 1200 MPa is susceptible to HE,³⁾ whereas a smaller concentration of hydrogen causes HE in steel with higher strength.⁴⁾

Most of the studies on HE so far was focused on the control of the microstructure of the steel⁵⁾ and the diffusion kinetics of hydrogen in the steel⁶⁾ to develop HE-resistant steels with higher hydrogen concentration. However, only a few approaches have considered the hydrogen absorption of steel. Nevertheless, the Iron and Steel Institute of Japan (ISIJ) has established a research group named “Comprehensive Understanding of Hydrogen-Passive Surface on Steels for Prevention of Hydrogen Embrittlement” in 2013, and their researches have effectively provided adequate fundamental understanding of hydrogen adsorption in steel.⁷⁾ For example, the hydrogen concentration of steel at the surface was fixed in a quite short time after the hydrogen-absorption condition suddenly changed. According to their report, the logarithm of the hydrogen concentration increases linearly with an increase in the hardness of steel and an increase in an inverse of the diffusion coefficient of hydrogen in the steel. Also, the hydrogen concentration decreases with a rise in the cathodic potential, but it slightly increases with an increase in the anodic potential.^8,9) Most of the studies by the group employed the electrochemical hydrogen-absorption technique to steel specimens immersed in the test solutions. They investigated the mechanism of the hydrogen absorption in the steel from the perspective of the electrochemical factors such as the cathodic reaction rate and solution conditions. However, it is necessary to clarify the influence of atmospheric environment on the hydrogen absorption of steel to understand the principle of HE in high-strength steels under normal atmospheric conditions. Akiyama et al.¹⁰⁾ conducted electrochemical hydrogen permeation tests on Fe specimens exposed to air in Beijing and Chongqing, China, as well as in Okinawa, Japan. They discussed the relationship between the maximum hydrogen-absorption rates and the air pollutants. Tsuru et al.¹¹⁾ prepared an Fe plate with a droplet of Na₂SO₃ solution, successfully investigated the hydrogen-absorption rate during atmospheric corrosion by drying the droplet. Their results revealed that Na₂SO₃ decreased the pH and corrosion potential, thus, it enhanced hydrogen absorption. Tada and Miura,¹²⁾ as well as Sakairi and Takagi¹³⁾ investigated the hydrogen absorption in Zn-plated steel during a dry–wet cyclic corrosion test after dropping NaCl and Na₂SO₄ solutions on the steel surfaces. They observed hydrogen absorption when Zn-plate was locally destroyed. Most of the studies on hydrogen absorption were focused on the hydrogen absorption on bare steel surfaces with sufficient polishing. However, it is very important to understand the hydrogen-absorption behavior on rusted steel surfaces as HE occurs on such steel surfaces during atmospheric corrosion. Accordingly, our research group has developed a system that monitors the hydrogen-absorption rate during atmospheric corrosion of specimens with rust layers exposed to humidity-controlled air. We have also investigated the relationship between the hydrogen-absorption rate and the relative humidity (RH) of air in Fe specimens with rust layers containing NaCl during atmospheric corrosion.¹⁴⁾ From the investigation, the following three findings were obtained. First, hydrogen absorption was observed at RH between 42% and 95%. Secondly, the maximum absorption rate was obtained at RH of about 75%, and finally, the pH value in the rust layers during the corrosion in the tested RH range was estimated to be 4.2 and 4.3,^15,16) which implies slightly acidic. This study was conducted as an extension of the previous research,¹⁴⁾ which studied the hydrogen absorption of Fe specimen reacted with MgCl₂ solution to produce rust layers on the surface. The hydrogen-absorption rate was monitored during atmospheric corrosion in humidity-controlled air, and the effect of RH on the hydrogen-absorption rate was investigated.

2. Experimental

2.1. Specimen Preparation

The material studied herein was a 2-mm thick Fe sheet (Nilaco Co.) with purity of 99.5 mass%. The sheet was cut into square specimens of dimensions 40 mm × 40 mm. The specimens were fully annealed at 1073 K for 1.8 ks in air followed by furnace cooling to remove the residual stresses induced by the sheet fabrication processes. After this heat treatment, the specimen showed almost equiaxial grains with an average grain size of about 0.69 mm. The surfaces of the specimens were mechanically polished using SiC papers to #6/0 (corresponding to #800) to remove the oxide films. Thereafter, they were electro-polished to remove the deformation layers on the surfaces introduced by the mechanical polishing. The electro-polishing was performed using a potentiostat (Toho Tech. Res. Co. Ltd., PS-07) with a Pt counter electrode and an Ag/AgCl (sat. KCl, room temp.) reference electrode. The specimens were immersed in a solution of H₃PO₄ (concentration: 85 mass%) and H₂SO₄ (concentration: 95 mass%) with volume percent of 75 and 25 vol%, respectively, at 298 K. Then, a potential of 1.5 V_Ag/AgCl was applied to the specimen for 84.6 ks. After the polishing, the surface of the specimen was removed to about 50 μm. Thereafter, the specimen was electroplated with Ni, as described below. Watts bath (NiSO₄·6H₂O:250 kg·m⁻³, NiCl·6H₂O:45 kg·m⁻³, H₃BO₃:40 kg·m⁻³) was prepared at 333 K. One surface of the specimen was fully covered with a polytetrafluoroethylene tape and a current density of −10 A·m⁻² was applied to the other surface for 420 s in the bath using the potentiostat. A Ni layer of about 15 nm thick was then deposited on the surface according the procedure reported by Yoshizawa et al.^17,18)

2.2. Formation of Rust Layer on the Specimen

The specimen surface without the Ni plate was fully covered with the rust layer by the following procedure. A 0.1 kmol·m⁻³ solution of MgCl₂ was prepared. An O-ring (inner diameter of 31 mm) was properly fixed to the surface and 2.0 × 10⁻⁶ m³ of the MgCl₂ solution was poured within the ring. Thereafter, the specimen was dried for 86.4 ks in a sealed container with silica gel. Subsequently, this step was repeated but with pure water to produce a uniform rust layer on the surface. The amounts of MgCl₂ contained in the rust was estimated to be 25.7 g·m⁻². The specimen with the rust layer containing 39.8 g·m⁻² MgCl₂ was also prepared in the similar way using 3.0 × 10⁻⁶ m³ of 0.1 kmol·m⁻³ MgCl₂ solution.

2.3. Electrochemical Hydrogen-absorption Test and Other Tests

In the hydrogen-absorption test, a modified Devanathan-Stachursky type cell¹⁹⁾ was employed.¹⁴⁾ The schematic illustration of the system is shown in Fig. 1. The cell was set in a box filled with dry air of an RH less than 5%. The specimen was set between two cells. The cell in contact with the rusted surface was the hydrogen-absorption cell, into which air with a controlled RH was introduced. The other cell in contact with the Ni-plated surface was the hydrogen detection cell, into which a 0.1 kmol·m⁻³ NaOH solution was introduced. The cell was equipped with an Ag/AgCl reference electrode and a Pt counter electrode, which were connected to the potentiostat.

Fig. 1.

Schematic illustration of a system for electrochemical hydrogen absorption test in atmospheric corrosion under controlled relative humidity (RH) and temperature. (Online version in color.)

The procedure for the test is described as follows: The specimen was set between the two cells, and the NaOH solution was introduced into the hydrogen detection cell. After the Ni-plated surface was fully covered by the solution, a passivation potential of 0 V_Ag/AgCl was applied to the surface and dry air of 0% RH was introduced into the hydrogen-absorption cell. When the passivation current density reached 0.2 mA·m⁻² and remained sufficiently stable, the dry–wet repeated cycle test was conducted on the rusted surface as follows: For the rust layer containing 25.8 g·m⁻², the dry air and the wet air with a controlled RH of about 27% were alternately introduced to the hydrogen-absorption cell for 10.8 ks each to induce and terminate atmospheric corrosion, respectively. During the test, RH and the temperature of the hydrogen-absorption cell, as well as the anodic current density on the hydrogen detection cell, were monitored simultaneously. While, the specimen with the rust layer containing 39.8 g·m⁻² MgCl₂ was subjected to the dry (RH 0%)–wet (33%) repeated cycle test in 100 cycles.

After the dry–wet repeated cycle test, the specimen was subjected to the test for determining steady-state hydrogen absorption rate depending on RH. First, the rusted surface was exposed to the dry air for 50 to 100 ks until almost passive current density of Ni was obtained. Thereafter it was exposed to the wet air with a controlled RH to obtain steady-state anodic current density, and then to the dry air again for 50 to 100 ks. From the test, the steady-state hydrogen-absorption rate, i_H, was defined as the difference between the steady-state anodic current densities during the atmospheric corrosion in the wet air and that in the dry air.

An X-Ray diffraction (XRD) system (Rigaku Co. Ltd., RINT-2550), a charge-coupled device (CCD) camera system (Moritex Co. Ltd, MS-804), a 3D-measurement system (KEYENCE Co. Ltd., VR-3200), and a Scanning Electron Microscope (SEM) (JEOL Co. Ltd., JSM-6060LV) were used for the characterization of the rust layer. The XRD system was equipped with a Cu-Kα radiation source (0.1541 nm) working at 40 kV/300 mA. The SEM was operated at an accelerated voltage of 15 kV.

3. Results

3.1. Hydrogen-absorption Rate during the Dry–wet Repeated Cycle Test

Figure 2 shows the change in the anodic current density on the hydrogen detection side (i.e., Ni-plated surface) during the dry (RH 0%, 10.8 ks)–wet (RH 27%, 10.8 ks) repeated cycle test. Each time the wet air was introduced into the hydrogen-absorption cell under the dry condition, a rapid increase in RH was recorded, which reached a steady-state in about 3 ks. Also, the anodic current density started to increase about 2 ks after the increase in RH and reached a steady-state in about 4 ks. On the contrary, a rapid decrease in RH was recorded each time the dry air was introduced into the wet cell, dropping to 0% in 0.6 ks. Similarly, the anodic current density decreased with a decrease in RH, reached 20% of the maximum value in 0.6 ks, and then decreased gradually until 10.8 ks.

Fig. 2.

Changes in RH and anodic current density on hydrogen detection side during the dry−wet (RH c.a. 27%) repeated cycle test.

The maximum anodic current density in each dry−wet cycle was recorded and plotted against the cycle, as shown in Fig. 3. As shown in the figure, the current density was almost independent of the number of cycle until about 10 cycles when it decreased almost linearly. It remained almost constant after about 40 cycles. It can also be observed that the current density more or less fluctuated during the repeated cycle till about 10 cycles. The fluctuation is considered to correspond to that of the temperature within about ±2 K. Ootsuka et al.²⁰⁾ developed a sensor for detecting hydrogen-absorption rate during an atmospheric corrosion, during which they discovered that the hydrogen-absorption rate varies greatly with a slight temperature change. The decrease in the current density recorded herein during the dry–wet repeated test up to 40 cycles is quite similar to that in the hydrogen-absorption rate recorded in a previous as the cumulative day of exposure for a steel sample in open air increased.²¹⁾

Fig. 3.

Change in the maximum anodic current density on hydrogen detection side by 1 cycle during the dry−wet (RH c.a. 27%) repeated cycle test. Amount of MgCl₂: 25.7 g·m⁻².

The XRD profiles of the rusted surface on the specimen after 10 and 55 cycles of the dry–wet repeated cycle test are shown in Fig. 4. It is confirmed from Fig. 4(a) that α-FeOOH was the main phase and MgCl₂·6H₂O was also present in the rust layer after 10 cycles. Fe was not detected. Figure 4(b) also reveals that β-FeOOH, γ-FeOOH, and α-FeOOH were abundant in the rust layer after 55 cycles, and the amount of MgCl₂·6H₂O decreased relatively.

Fig. 4.

XRD profiles of the rusts containing 25.7 g·m⁻² MgCl₂ on the specimens after (a) 10 and (b) 55 cycles of the dry−wet (RH c.a. 27%) repeated cycle test.

The surfaces of the specimen after 10 and 55 cycles of the test were observed by the CCD camera system and their shapes were measured by the 3D measurement system. The results are shown in Fig. 5. After 50 cycles, the rust layer was dark brown and almost flat-shaped with an average thickness of about 0.05 mm. After 55 cycles, the rust layer remained dark brown but became rough, and its average thickness increased to about 0.18 mm.

Fig. 5.

Surfaces of the specimens covered with the rusts containing 25.7 g·m⁻² MgCl₂. observed after (a)(c) 10 and (b)(d) 55 cycles of the dry–wet repeated cycle tests. The images were obtained by (a)(b) the CCD camera system and (c)(d) the 3D-measurement system.

The results of the SEM observation for the central parts of the rust layers in Figs. 5(a) and 5(b) are shown in Fig. 6. It is observed that the surface of the rust layer after 10 cycles of the test seemed to be flat, and it was composed of a number of small blocks at low magnification as shown in Fig. 6(a). However, at high magnification (Fig. 6(e)), the blocks were shown to be tiny particles. When the dry–wet repeated cycle test reached 55 cycles, the rust layer was relatively smooth and flat in comparison with that after 10 cycles at low magnification (Fig. 6(b)), and the surface after 55 cycles was composed of the particles whose average diameter was about 3 μm, larger than that after 10 cycles, at high magnification (Fig. 6(f)).

Fig. 6.

SEM images for surfaces of the specimens covered with the rust containing 25.7 g·m⁻² MgCl₂. (a)(c)(e) 10 cycles and (b)(d)(f) 55 cycles. Magnification: (a)(b)<(c)(d)<(e)(f).

3.2. Effect of RH on the Hydrogen-absorption Rate for the Specimen After the Dry–wet Repeated Cycle Test Beyond 50 Cycles

As described above, the specimen with the rust layer containing 25.7 g·m⁻² MgCl₂ was subjected to the dry–wet repeated cycle test, and the anodic current density on the hydrogen detection side was monitored simultaneously. It was found that the anodic current density (corresponding to the hydrogen-absorption rate) decreased with an increase in the test cycle, reaching a steady-state after about 40 cycles. Therefore, the specimen subjected to 55 cycles of the test was used for the hydrogen-absorption test to obtain the effect of RH on the steady-state hydrogen-absorption rate almost independent of the dry−wet repeated cycle.

Figure 7 shows the change in RH with the anode current density during the hydrogen-absorption test. The dry air (RH 0%) was introduced into the hydrogen-absorption cell at the beginning of the test and the anode current density was kept at a constant value of 0.040 × 10⁻⁴ A·m⁻² as the steady-state passive current density of Ni. Thereafter, the wet air was introduced into the cell and the RH increased rapidly, reaching a stable value of about 72%. At the same period, the anode current density gradually increased and reached a stable value of 1.67 × 10⁻⁴ A·m⁻². When the dry air was introduced again, the RH rapidly dropped to 0% and the anode current density decreased gradually, reaching the initial value of 0.040 × 10⁻⁴ A·m⁻². The hydrogen-absorption rate (i_H) at RH 72% was obtained to be 1.63 × 10⁻⁴ A·m⁻².

Fig. 7.

Typical example of changes in RH and anodic current density on hydrogen detection side with time. Amount of MgCl₂:25.7 g·m⁻². Average RH in the steady state: 72%.

With the method described in Fig. 7, i_H was recorded against RH, as summarized in Fig. 8. As shown in the figure, hydrogen absorption was observed at RH above 15%, i_H increased rapidly with an increase in RH, reached the maximum value at RH of about 30%, and then decreased rapidly afterward. When the RH increased above 40%, i_H increased again, reached the maximum value at RH of about 80%, and then decreased gradually. Hereafter, the two peaks of i_H observed at the lower and higher RH are denoted as the 1st and 2nd peaks, respectively. Furthermore, a specimen with the rust layer containing 39.8 g·m⁻² of MgCl₂ was prepared and characterized. During the dry–wet repeated cycle test, it was obtained that the maximum anode current density during each wet condition decreased with an increase in the cycle until it reached a steady-state. After the repeated cycle test in 100 cycles, i_H was recorded against RH, and the results are depicted in Fig. 8. It is observed that, similar to the former specimen, the latter specimen also showed two peaks in the plot of i_H against RH. However, the latter specimen, which contain higher amount of MgCl₂, exhibited higher i_H at the 1st and 2nd peaks, as well as smaller RH at the 2nd peak.

Fig. 8.

Effect of RH during wet period on hydrogen absorption rate as a function of amount of MgCl₂ in the rust.

4. Discussion

4.1. Effect of RH on Hydrogen-absorption Rate under Steady-state

The specimens with the rust layers containing 25.7 and 39.8 g·m⁻² MgCl₂ were prepared and subjected to the dry–wet repeated cycle tests beyond 40 cycles. Thereafter, the steady-state hydrogen-absorption rate (i_H) was measured as a function of RH, and the change in the rate with RH was obtained, as shown in Fig. 8. As shown in the figure, hydrogen absorption was first detected in the RH range from about 15%, and i_H increased rapidly with an increase in RH, reached a maximum value at RH of about 30%, and then decreased rapidly. When the RH increased above 40%, i_H increased again, reached a maximum value at RH between 65 and 80%, and then decreased gradually. The change in i_H with RH is discussed as follows.

In general, MgCl₂ solid deliquesces into a saturated solution in the air with RH of 33% (at 298 K). Below this humidity, water molecules only adsorb on the surface of the MgCl₂ solid. This implies that the Fe plate suffers almost no atmospheric corrosion with a MgCl₂ solution in the air at RH below 33%. Some researchers have investigated the atmospheric corrosion of steel with MgCl₂ solution in the RH range from 33%.^22,23) However, Fig. 8 reveals that hydrogen absorption induced by atmospheric corrosion was obtained at RH of 15%. In a previous report,¹⁴⁾ the hydrogen-absorption behavior of a specimen with a rust layer containing NaCl was investigated in the air with the controlled RH, and hydrogen absorption was obtained in the RH range of over 40%, which is below the deliquescence humidity of NaCl (RH 75% at 298 K). In addition, the phenomenon was explained on the basis of capillary condensation of the saturated NaCl solution in the rust layer with a lot of tiny pores.²⁴⁾ Since the result in Fig. 8 is similar to that in the previous report, the mechanism is considered on the same basis as the previous one.¹⁴⁾ It is known that RH of air has a correlation with the radius of concave curvature of liquid water surface (r) as shown by Kelvin’s equation,²⁵⁾   

$$\mathrm{l}\mathrm{n}\mathrm{\hspace{0.17em} }\mathrm{R}\mathrm{H}=-2\gamma V_{\mathrm{m}}/(rRT),$$(1)

where γ is the surface tension of water, V_m the molar volume of water, R the gas constant, and T the absolute temperature. It is noted from the equation that RH decreases with a decrease in r. This relationship may be applicable to not only water but also aqueous solutions. This implies that saturated aqueous solution of MgCl₂ with flat surface can be stable in the air with RH of 33% and the solution with concave surface stably exists in an air with RH below 33%, according to the Eq. (1). This specific RH may be defined as the deliquescence humidity of MgCl₂, below which MgCl₂ solid can transform to a saturated solution with concave surface of the specific radius. Because the valid values of some variables in Eq. (1) are not obtained, the quantitative relationship between r and RH cannot be calculated. However, as shown in Fig. 8, hydrogen absorption was recorded at RH from 15%. It is inferred that the MgCl₂ solid deliquesced into saturated solution with concave surface in the rust layer. Then, the solution came in contact with the Fe plate and atmospheric corrosion was induced, resulting in hydrogen absorption. The cathodic reaction at this point is considered to be the reduction reaction of oxygen. Since the concentration of dissolved oxygen in the saturated solution is very small, oxygen mainly diffuses from the peripheral zone of the solution particles having contact with the Fe plate and reduces in the zone according to Eq. (2).   

$${\mathrm{O}}_2+2{\mathrm{H}}_2\mathrm{O}+4{\mathrm{e}}^{-}\to 4\mathrm{O}{\mathrm{H}}^{-}$$(2)

As shown in Fig. 8, i_H increased with an increase in RH from 15%, reached a maximum value at RH of about 30%, and then decreased rapidly. This transient in i_H with RH can be explained as follows. An increase in RH induces an increase in the radius of the concave surface of the solution according to Eq. (1). This increase leads to an increase in size and number of the solution particles, then an area of oxygen reduction reaction, and finally, an overall corrosion rate corresponding to the hydrogen-absorption rate. When RH approaches the deliquescence humidity of MgCl₂ (33%), the radius of the solution rapidly approaches an infinite value (i.e., the solution surface becomes flat) as stated above. In the case where the surface of the Fe plate is almost completely covered with the saturated solution film, oxygen supply to the substrate is suppressed by the film, thus, the corrosion rate, as well as the hydrogen-absorption rate, is suppressed.

On the contrary, Fig. 8 also reveals the following change in i_H with RH beyond 40%. When RH increased above 40%, i_H increased again, reached a maximum value in the RH range between 65% and 80%, and then decreased gradually. This change can be explained as follows. From a thermodynamics perspective, the activity of water in MgCl₂ solutions with a flat surface is equal to that of water in the air, i.e., RH above 33%. This relationship implies that an increase in RH above 33% induces an increase in the concentration of water corresponding to a decrease in the concentration of MgCl₂ in the solution. In addition, the amount of MgCl₂ in the rust layer is maintained even when RH in the air changes during the test. Therefore, an increase in RH induces an increase in the amount of water in the solution, volume of the solution, and thickness of the solution layer. It is assumed that the surface of Fe plate was almost covered with the MgCl₂ solution when RH was over 33%. Under this condition, it is inferred that the diffusion-limited reduction rate of dissolved oxygen determines the corrosion rate (i_corr) and then the following equation is applicable based on Eq. (3),   

$$i_{\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{r}}=4F\mathrm{\hspace{0.17em} }{D}_{\mathrm{O}2}\mathrm{\hspace{0.17em} }{C}_{\mathrm{O}2}{}^0/L,$$(3)

where F is the Faraday constant, D_O2 the diffusion coefficient of dissolved oxygen, C_O2⁰ the dissolved oxygen concentration at the air-side surface of the solution film, and L the thickness of the solution film. Assuming D_O2 is little affected by the MgCl₂ concentration of the solution film depending on RH, then, an increase in RH would give rise to a decrease in the MgCl₂ concentration, an increase in C_O2⁰, and then, an increase in i_corr corresponding to i_H according to Eq. (3). However, an increase in RH also causes an increase in L, and then, a decrease in i_corr corresponding to i_H, according to Eq. (3). The above two trends are used to explain the formation of the 2nd peak in Fig. 8. The increase in i_H as RH increases at the lower RH region is caused mainly by an increase in C_O2⁰ than that in L. Similarly, the decrease in i_H as RH increases is attributed more to the increase in L. If L increases and exceeds the thickness of the diffusion layer (L_D, constant), L in Eq. (3) would be replaced with L_D. Then, i_H would increase again with an increase an RH owing to the continuous increase in C_O2⁰. However, this was not observed in Fig. 8. Therefore, the result obtained herein suggests that L was lower than L_D in the tested RH range. Air with RH of 100% makes the concentration of MgCl₂ solution approach 0 kmol·m⁻³ and the solution film thickness infinite. This behavior speculates that i_corr corresponding to i_H approaches the specific value at RH of 100%, independent of the amount of MgCl₂ in the rust layer.

4.2. Change in Hydrogen-absorption Rate during the Dry–wet Repeated Cycle Test

The dry–wet repeated cycle test (RH 0%, 10.8 ks and RH 27%, 10.8 ks) was performed on the specimen with the rust layer containing 25.7 g·m⁻² MgCl₂. As shown in Fig. 2, the current density started to increase about 2 ks after the initial increase in RH. Also, it is observed that RH at which the current density started to increase was around 15%. Figure 8 also shows that no i_H was detected at RH lower than 15% when the specimen was exposed to the air with lower RH for a long time. Therefore, it is considered that the time lag was caused by the fact that RH increased gradually through the threshold below which no i_H is generated. When the wet air was converted back to the dry cell, both RH and anode current density decreased almost simultaneously in a short time. Figure 8 reveals that RH of 27% induced sufficient i_H and RH of 0% generated no i_H. This is considered to cause the simultaneous decrease in RH and current density.

As shown in Fig. 3, the maximum value of the anodic current density during each wet condition was almost independent of the cycle until about 10 cycles. But with further increase in cycle, the current density decreased almost linearly and stabilized again beyond 40 cycles. Figures 4, 5, 6, 7 summarize the findings of the appearance of the rust layer. After 10 cycles of the test, the rust layer was dark brown and flat with an average thickness of 0.05 mm. Magnified observation revealed the presence of tiny particles. Abundant α-FeOOH and a tangible amount of MgCl₂·6H₂O were detected from the rust surface. After 55 cycles, the rust layer was dark brown and relatively rough. The small particles were observed to have undergone growth, and the average thickness of the layer was increased to 0.18 mm β-FeOOH, γ-FeOOH, and α-FeOOH were observed to be dominant, but MgCl₂·6H₂O was relatively small. Relating the feature of the rust layer with the hydrogen-absorption behavior, the suppression of the atmospheric corrosion corresponding to hydrogen absorption is considered to be caused by the feature of the latter rust layer, which is a thick rust layer consisting of dense coarse particles. The relatively small amount of MgCl₂·6H₂O observed in the latter rust layer is attributed to the fact that MgCl₂·6H₂O particles were dispersed in the grown rust layer in 0.18 mm thick, and the just surface of the layer by about 0.01 mm was analyzed by the XRD test. In addition, it was found that the latter rust layer, which exhibited corrosion resistance, contained β-FeOOH, γ-FeOOH, and α-FeOOH in abundance. However, it is known that β-FeOOH and γ-FeOOH are formed in the early stage of rusting. Also, they have poor corrosion resistance and gradually change to α-FeOOH during the dry–wet repeated cycle tests to form corrosion-resistant layers.^26,27) The results obtained herein are not consistent with the reported ones. This is considered to be a result of the fact that the XRD analysis was performed on about 0.01 mm from the surface of the rust layer. Therefore, there is a need to investigate the composition of the rust layer deep below its surface in future studies on corrosion resistance.

5. Conclusions

• The dry–wet repeated cycle test (RH 0%, 10.8 ks and RH 27%, 10.8 ks) was performed on an Fe-plate specimen with a rust layer containing 25.7 g·m⁻² MgCl₂. The maximum anodic current density under each wet cycle was obtained to be independent of the dry–wet repeated cycle within the first 10 cycles. Afterwards, it decreased almost linearly with an increase in the cycle and relatively stabilized again beyond about 40 cycles.

• After the first 10 cycles, the rust layer was dark brown and flat with an average thickness of 0.05 mm, and it contained a number of small particles. Also, α-FeOOH was obtained to be dominant in the layer, whereas MgCl₂·6H₂O was present in relatively small amount. After 55 cycles, the rust layer remained dark brown but relatively rough. The growth of the small particles was observed, and the average thickness of the layer was increased to be 0.18 mm. β-FeOOH, γ-FeOOH, and α-FeOOH were observed in abundance, whereas MgCl₂·6H₂O was relatively small.

• The specimens with the rust layers containing 25.7 and 39.8 g·m⁻² MgCl₂ were subjected to the dry–wet repeated cycle test beyond 40 cycles, followed by the electrochemical hydrogen-absorption test to measure the steady-state hydrogen-absorption rate during the atmospheric corrosion in the air with a controlled RH. Hydrogen absorption was observed when RH was more than 15%, and the absorption rate showed two peaks at RH of about 30% and between 65% and 80%.

Acknowledgments

One of the authors thanks the research groups, “Comprehensive Understanding of Hydrogen-Passive Surface on Steels for Prevention of Hydrogen Embrittlement” and “Corrosion-induced Hydrogen Absorption to Steels” of the Iron and Steel Institute of Japan for their sufficient contribution in the discussion of our research results as well as their financial support. Additionally, one of the authors thanks the Grant-in-Aid for Scientific Research (C) (18K04784) of JSPS and Kansai University Fund for Domestic and Overseas Research Fund for their financial support.

References

• 1)   S.  Matsuyama: Tetsu-to-Hagané, 80 (1994), 679 (in Japanese).
• 2)   A. R.  Troiano: Hydrogen in Metals, ASM, Metals Park, OH, (1974), 3.
• 3)   T.  Fujita and  Y.  Yamada: Stress Corrosion Cracking and Hydrogen Embrittlement of Iron Base Alloys, NACE, Houston, TX, (1977), 736.
• 4)   S.  Takagi,  T.  Inoue,  K.  Tsuzaki and  F.  Minami: J. Jpn. Inst. Met., 65 (2001), 1073.
• 5)   T.  Omura: ISIJ Int., 52 (2012), 267.
• 6)   T.  Omura,  T.  Kushida,  T.  Kudo,  F.  Nakasato and  S.  Watanabe: Zairyo-to-Kankyo, 54 (2005), 61 (in Japanese).
• 7)   M.  Sakairi: ISIJ Int., 56 (2016), 377.
• 8)   T.  Haruna,  K.  Takeuchi,  T.  Kasahara and  Y.  Nii: ISIJ Int., 56 (2016), 378.
• 9)   T.  Haruna and  Y.  Nii: ISIJ Int., 56 (2016), 392.
• 10)   E.  Akiyama,  S.  Li,  H.  Katayama,  B.  Zhang,  K.  Zhao and  W.  Oshikawa: ISIJ Int., 56 (2016), 436.
• 11)   T.  Tsuru,  Y.  Huang,  M. R.  Ali and  A.  Nishikata: Corros. Sci., 47 (2005), 2431.
• 12)   E.  Tada and  Y.  Miura: ISIJ Int., 56 (2016), 444.
• 13)   M.  Sakairi and  S.  Takagi: ISIJ Int., 56 (2016), 452.
• 14)   T.  Haruna,  Y.  Shoji and  Y.  Hirohata: ISIJ Int., 61 (2021), 1079. https://doi.org/10.2355/isijinternational.ISIJINT-2020-018
• 15)   E.  Akiyama,  K.  Matsukado,  M.  Wang and  K.  Tsuzaki: Corros. Sci., 52 (2010), 2758.
• 16)   E.  Akiyama,  S.  Li,  T.  Shinohara,  Z.  Zhang and  K.  Tsuzaki: Electrochim. Acta, 56 (2011), 1799.
• 17)   S.  Yoshizawa and  K.  Yamakawa: Boshoku Gijutsu, 24 (1975), 365 (in Japanese).
• 18)   S.  Yoshizawa,  T.  Tsuruta and  K.  Yamakawa: Boshoku Gijutsu, 24 (1975), 511 (in Japanese).
• 19)   M. A. V.  Devanathan and  Z.  Stachurski: Proc. R. Soc. A, 270 (1962), 90.
• 20)   S.  Ootsuka,  S.  Fujita,  E.  Tada,  A.  Nishikata and  T.  Tsuru: Corros. Sci., 98 (2015), 430.
• 21)   T.  Kushida and  T.  Omura: Kozozairyo no Kankyo Zeika niokeru Suiso no Kino ni kansuru Kenkyu-5 (Research on Hydrogen Relating Hydrogen Embrittlement to Structural Materials - Part 5), ISIJ, Tokyo, (2003), 101 (in Japanese).
• 22)   Y.  Tsutsumi,  A.  Nishikata and  T.  Tsuru: J. Electrochem. Soc., 152 (2005), B358.
• 23)   B.  Maier and  G. S.  Frankel: J. Electrochem. Soc., 157 (2010), C302.
• 24)   T.  Ishikawa,  T.  Yoshida,  K.  Kandori,  T.  Nakayama and  S.  Hara: Corros. Sci., 49 (2007), 1468.
• 25)   W.  Thomson: Philos. Mag., 42 (1871), 448.
• 26)   J. T.  Keiser,  C. W.  Brown and  R. H.  Heidersbach: J. Electrochem. Soc., 129 (1982), 2686.
• 27)   T.  Misawa: Bull. Iron Steel Inst. Jpn., 6 (2001), 325 (in Japanese).