Effect of addition of alumina and rare-earth oxide particles on the corrosion resistance and mechanism of low carbon low alloy steel

The corrosion mechanism of the low carbon low alloy steel with Al2O3 particles and rare Earth (RE) oxide particles was compared in a simulated marine environment. It is shown that when the Al2O3-containing particles are introduced, the number density of nonmetallic particles of the steel increases twice, and the average particle size decreases from approximately 2.4 μm to 1.4 μm. With the introduction of Al2O3-containing particles, the amount of pitting corrosion increases. Furthermore, pitting corrosion occurs more uniformly owing to the fineness of the Al2O3 particles, thereby leading to smaller, shallower pits after the Al2O3 particles are shed. Hence, the corrosion performance of the steel with Al2O3 particles is significantly improved than that of the steel without Al2O3 particles. By adding RE oxide particles into steel, the nonmetallic particles in steel are refined but not as effectively as that achieved by adding the Al2O3-containing particles. Different from Al2O3 particles, Cu is obviously enriched in the location of RE oxide particles at the initial corrosion stage, which makes the steel exhibit the best corrosion resistance. Cu enrichment is attributed to the mobile Cu present in the rust layer and to the micro acid region formed around the RE oxide particles.


Introduction
Since the 20th century, with the rapid development of coastal cities, there has been wide demand for high-quality steel in marine environments. Owing to the high salinity and humidity in marine environments, corrosion is the primary cause of the loss of ordinary structural steel. The formation of a stable rust layer from enriched alloy elements [1] results in a higher atmospheric corrosion resistance of weathering steel compared to other steels [2]. In a marine atmosphere, a rust layer can effectively isolate the steel substrate from corrosive media such as O 2 , H 2 O, and NaCl particles [3][4][5]. The most common method for increasing the content of alloying elements in the rust layer is to directly add Cu, Cr, Ni, and P to steel during the smelting process [6,7]. The higher the alloying element content in the steel matrix, the higher the content of alloying elements in the rust layer. However, the corrosion resistance resulting from the addition of alloying elements is limited, and Yan [8] demonstrated that the corrosion resistance of weathering steel cannot be improved when the Cu content exceeds 0.5%. Meanwhile, the addition of P results in the cold brittleness of steel [9]. Therefore, researchers have proposed various methods, such as grain refinement and the construction of a homogeneous microstructure [10][11][12][13], to improve the corrosion resistance of materials by improving the size and distribution of inclusions [14][15][16][17][18] and adjusting the ratio of the alloying elements [19,20]. Among these methods, RE treatment has significant advantages and has been widely investigated by many researchers [21,22].
RE elements not only improve the corrosion resistance during the pitting corrosion process [23,24] but also significantly promote the enrichment of alloying elements in the rust layer [25]. Hou [26] demonstrated that the Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
addition of La to low-alloy high-strength steel increased the concentrations of Cr and Mo in the rust layer without changing the phase composition. Wang [22] found that the RE content of low-alloy steel was enriched at the interface between the internal rust layer and the steel matrix, while Si 4+ , P 5+ , and Cu + ions were enriched in the inner rust layer. Although the enrichment of alloying elements in the rust layer is attributed to the addition of RE element, the driving force for the enrichment of alloying elements in the rust layer is unclear.
To improve the mechanical properties, oxide dispersion-strengthened steels have been investigated in the fields of metallurgy and welding [27,28]. Adding large amounts of fine, regularly shaped oxides to steel can simultaneously modify its strength and plasticity. In previous studies, the mechanical properties of 45 steels were significantly improved by adding a large number of fine and evenly distributed Al 2 O 3 particles [29,30]. However, second-phase particles in the matrix adversely affect the corrosion performance owing to local corrosion, especially Al 2 O 3 or MnS particles, caused by incomplete deoxidation and desulfurization during smelting [10,[31][32][33][34][35][36]. It is believed that uniformly distributed second-phase particles, achieved by modifying their size and number, are beneficial for the corrosion resistance of weathering steel [4,37,38]. Therefore, further investigation is required to understand whether the introduction of oxide particles into weathering steel can also improve corrosion resistance requires further study.
In this study, Al 2 O 3 and RE oxide particles were introduced into low-alloy carbon steels. The effect of nonmetallic particles on the incipient corrosion of steel was studied. The mechanism of Cu enrichment in the rust layer after the introduction of RE oxide particles was also discussed. The corrosion mechanisms of the experimental steels in a marine environment were compared. We hope that this study can provide an in-depth understanding of the corrosion resistance mechanism of RE oxide particles and broaden the applications of oxide particles in weathering steel.

Experimental procedure
The experimental steel was Q500NH steel. As Cu can significantly enhance corrosion resistance, the Cu content in this study was reduced to 0.05% to facilitate the analysis of the corrosion resistance mechanism of steel after the introduction of oxide particles. Experimental steel ingots (Φ170 mm × 410 mm) were prepared in a vacuuminduction-melting furnace by Anshan Iron and Steel Group Co., Ltd The ingots were named NH, NHAl, and NHRE. No additional oxide particles were added to NH ingots. Dispersed Al 2 O 3 -containing particles were introduced into the NHAl ingot using the master alloying method. Approximately 90 wt% of Al 2 O 3 particles with a size of 0.1-10 μm were purchased from commercial refractory materials. The Al 2 O 3 -containing particles and 20 steel particles with a size of 2 mm were mixed in the mass ratio of 1:9, smelted in a medium frequency induction furnace, and then poured into an iron mold. This ingot was used as master alloy 20 wt% master alloy nuggets and pure metal pieces were prepared, melted in the vacuum induction furnace, and cast into the NHAl steel ingot. More details about the experiment could be found in [29,30,39]. The RE oxide particles were introduced into the NHRE ingot by directly adding pure RE pieces (La and Ce) to the molten steel during the final smelting stage. The compositions of the three ingots are listed in table 1. C and S were measured using a carbon-sulfur analyzer, and O was measured using an oxygen-nitrogen analyzer. The Al content was measured using an inductively coupled plasma optical emission spectrometer (ICP-OES). The total Al and O contents in NHAl are higher than those in NH, and this implies that the Al 2 O 3 particles in NHAl steel are higher than those in NH. Other alloy elements were measured using direct-reading spectroscopy.
Samples were taken from half the radius of the ingot, ground, and polished. A three-electrode system was used for the electrochemical tests. The sample, Pt sheet, and saturated calomel electrode served as the working, counter, and reference electrodes, respectively. The test area of each sample was 1 cm 2 . The kinetic potential polarization curves and electrochemical impedance spectra (EIS) were obtained using a Princeton VersaStat 4 electrochemical workstation. Each set of samples was tested thrice to ensure the accuracy and reproducibility of the data.
Nonmetallic particles were observed through a field-emission scanning electron microscopy (FE-SEM) using a JSM-7100F. Elemental analyses of the particles and rust layer were performed using a JXA 8530F electron probe microanalyzer (EPMA). A 3.5 wt% NaCl solution was used for the corrosion and electrochemical experiments. The temperature was maintained at 25°C during the corrosion tests. The entire process of corrosion testing was sealed to avoid impurities. The corrosion weight loss experiment was conducted by immersing the experimental steels in a 3.5% NaCl solution for 20 d and recording their daily mass loss.

Results and discussion
3.1. Electrochemical tests and corrosion weight loss experiment 3.1.1. Dynamic potential polarization curve Figure 1(a) shows the dynamic potential polarization curves of the three steels. Table 2 presents the corrosion potential (E corr ) and corrosion-current density (I corr ), fitted in the linear anodic branch of the polarization curve. The corrosion-current density reflected the corrosion rate of the experimental steel in the solution. The higher the corrosion-current density, the higher the corrosion rate of the experimental steel. Additionally, the corrosion potential reflects the thermodynamic characteristics of the system. A lower corrosion potential indicates weaker corrosion resistance [21]. Polarization resistance (Rp) represents the oxidation resistance of the metal under the applied potential. The higher the Rp, the better the corrosion resistance. As shown in table 2, the corrosion potential of the NHRE steel is significantly higher than that of the other steels, and its corrosioncurrent density is the lowest. The corrosion potential of NHAl steel is slightly higher than that of NH steel, while  the corrosion-current density of NHAl steel is significantly lower than that of NH steel. Hence, when RE elements are introduced, the corrosion potential increases significantly, and the corrosion current and corrosion rate decrease. No changes occurred in the corrosion potential of the NHAl steel, and the corrosion-current density is significantly lower than that of the NH steel. Figure 1(b) shows the impedance map of the experimental steels and figure 1(d) shows the EIS equivalent circuit diagram. The fitted data are listed in table 3, where R s represents the resistance of the electrolyte, C dl is a double layer capacitance, and R ct is the charge transfer resistance. Generally, the larger the value of R ct , the more difficult the charge transfer and the better the corrosion resistance of the material. As can be seen from table 2, the NHRE steel has the largest R ct value and the best corrosion resistance. The NH steel has the smallest R ct value and worst corrosion resistance [21]. The corrosion resistance of low-alloy steels can be improved by introducing Al 2 O 3 or RE oxide particles. respectively. The corrosion weight loss of the NHRE steel is the smallest, indicating its best corrosion resistance. Figure 2 shows the SEM images of nonmetallic particles in the experimental steels. The number density of these particles in the NH steel is approximately 23 per square millimeter, which accounts for approximately 1/3 of the nonmetallic particles in the NHAl steel. Moreover, the NHAl steel has a larger number of nonmetallic particles, and its distribution is more uniform, as shown in figures 2(a) and (b). The average diameter of these particles in NHAl steel is approximately 1.4 μm, which is smaller than that of NH steel (about 2.4 μm). Furthermore, the shape of the nonmetallic particles in the NH steel is irregular ( figure 2(d)), and they not only are harmful to the corrosion resistance but also significantly reduce the mechanical properties of the steel. In comparison, the shape of the particles in the NHAl steel is spherical ( figure 2(e)). In the NHRE steel (figures 2(c) and (f)), the  morphology of the nonmetallic particles is similar to that of NHAl, and the number density is greater than that of the NH steel. The average diameter is about 1.7 μm, which is moderate among the three steels. Figure 3 shows the typical morphologies of the nonmetallic particles and their EDS spectra. The nonmetallic particles in the NH and NHAl steels are rich in O and Al only, whereas the atomic ratio of O to Al is proportionally 3:2; therefore, the nonmetallic particles in these two experimental steels are considered to be Al 2 O 3 (figures 3(a) and (b)). After extensive observation, it can be found that a small amount of MnS is attached next to some Al 2 O 3 .For the NH steel, these Al 2 O 3 particles are from the deoxidation products in the smelting process, while for the NHAl steel, they are mostly from the introduction of the Al 2 O 3 -containing master alloy. With the introduction of RE elements, the nonmetallic particles observed in the NHRE steels are enriched in RE, O, and S but no longer enriched in Al ( figure 3(c)). In this study, the mixed RE, O, and S particles are labeled as RE oxide particles. Figure 4 shows histogram of the size distribution of the nonmetallic particles in the steels. It can be seen that the number of nonmetallic particles less than 3 μm accounts for ∼ 80% of the total particles. The number of particles larger than 3 μm decreases significantly in NHAl and NHRE. The number of small-sized nonmetallic particles smaller than 1 μm in NH, NHAl, and NHRE steels accounts for 10%, 35%, and 22%, respectively. This means that during the melting process, heterogeneously shaped nuclei not only refine the organization but also effectively prevent the aggregation of small-sized nonmetallic particles to form large-sized nonmetallic particles, resulting in a great refinement of nonmetallic particles.

Effect of nonmetallic particles on short-term pitting corrosion 3.3.1. Microstructure evolution during pitting corrosion
The initial corrosion process mostly starts in the active part of the steel substrate, particularly in the nonmetallic particles. In the initial corrosion stage, the boundary between nonmetallic particles and steel substrate becomes blurred owing to the generation of corrosion products such as iron oxide, as shown in figure 5(a). After a period of corrosion, the particles are completely wrapped in the corrosion product. The surrounding of these particles is loose and further fell off, thus forming a pit of 3 ∼ 10 μm in length at the original location of the steel substrate as shown in figure 5(b). As can be seen from the elemental maps in figure 5(b), Al, Mn, and S are not enriched in the pit owing to detachment, while O is still enriched around the pit. Figure 6(a) shows the EPMA images of extension process of localized corrosion in NH steel. The rougher and darker areas on the left are stable corrosion-product areas, whereas the smoother and lighter areas on the right are unstable corrosion-product areas. The boundaries of these two areas are clear, as indicated by the red dotted curve. Figure 6(b) shows a pit with a length of approximately 8 μm, where the nonmetallic particle has fallen off. A stable corrosion product area is generated after long-term corrosion. The corrosion product around the pit becomes loose. Figure 6(c) shows the elemental maps near the pit in the stable corrosion product area. O element is enriched near the pit and diffused into the steel substrate around the pit. Compared with the enrichment of O in the unstable corrosion product area as seen in figure 5(b), the corrosion of the steel substrate is expanded around the pit, and the corrosion products covers this area in the stable corrosion product area. Figure 7 compares the stable corrosion product area of three steels after soaking in 3.5% NaCl solution for 0.5 h. For NH steel, the stable corrosion product area starts from several pit positions where the nonmetallic particles exist, then expands, merges, and eventually covers the entire surface of the steel substrate. NH steel with  secondary particles with extremely irregular shapes and a significantly large volume is prone to pitting corrosion. Thus, the pits where the original particles detach after long-term corrosion are irregular, large, and deep, as shown in figures 5(b) and 6(b). This results in the rapid expansion of the stable corrosion product area in the NH steel, and only a few pits are observed in this area.
No large or deep pits are observed in the NHAl and NHRE steels (figures 7(c) and (d)). Instead, there are many protrusions in the NHAl and NHRE steels formed through the accumulation of corrosion products due to pitting corrosion. Because the nonmetallic particles in NHAl steel are greater in number, smaller in size, and distributed more uniformly (figures 2(a), (b), and 4), pitting corrosion occurs more uniformly on the steel substrate. Thus, the small pitting corrosion areas rapidly contact each other forming a large corrosion area and achieving further corrosion expansion. Regardless of the area of unstable or stable corrosion products, all pits are smaller in size, shallower in depth, and closer to spherical (figures 8(a) and (b)). During the pitting corrosion  stage, these pits effectively reduce the corrosion rate. The pitting corrosion occurs more evenly on the steel substrate, forming a more uniform and flat rust layer. Because a stable rust layer can be tightly attached to the steel substrate, large cracks are less likely to appear in the rust layer.
After 0.5 h of corrosion, no large or deep pits are found in the NHRE steel; however, protrusions made of corrosion products are numerous and evenly distributed ( figure 7(d)). This indicates that pitting corrosion can occur at more sites. During short-term corrosion, the RE oxide particles protect the steel substrate from corrosion through self-consumption. During the long-term corrosion process, the presence of RE promotes the generation of protective rust layers. A more detailed mechanism of pitting corrosion induced by RE oxide particles can be found in literature [23,24].

Mechanism of the distribution of Cu element in the steel and corrosion product
As can be seen from figures 2 and 7, only a small amount of nonmetallic particles underwent pitting corrosion. The following conclusions can be drawn. After contacting with the corrosive medium, similar particles cannot occur pitting corrosion simultaneously because of their various sizes and shapes. As shown in figure 7, the pits remaining after a limited amount of pitting corrosion can be observed in a relatively large corrosion area. Therefore, most particles have not yet undergone pitting corrosion when covered with corrosion products. Instead, most particles are gradually covered by corrosion products during the expansion process of corrosion area and enter the stable corrosion product area. In this process, the Al 2 O 3 and RE oxide particles have different effects on the distribution of Cu. Figure 9(a) shows elemental distribution maps of the RE oxide particles before corrosion obtained from EPMA. Cu is uniformly distributed on the substrate of the NHRE steel. After 1 h of corrosion, in the unstable corrosion product area, La, Ce, and S gradually diffuse, whereas Cu gathers at the RE oxide particles ( figure 9(b)). This also shows that both the RE oxide particles that participate in the initial pitting corrosion and the RE oxide particles that are gradually covered by corrosion products will gradually decompose. As the corrosion area extends, the area enriched in Cu is gradually covered by the corrosion products owing to the increase in the corrosion product thickness and the end stage of RE oxide particle decomposition (figure 9(c)). Similar experiments were conducted on the NHAl steel, and the distribution of Cu is shown in figure 10. It can be seen that regardless of the corrosion stage (beginning of corrosion or after long-time corrosion), the distribution of Cu element appears unchanged, and it is uniformly distributed in the three states of the steel substrate.
Based on the above observations, the mechanism of Cu enrichment is shown in figure 11. With the expansion of the corrosion area, the nonmetallic particles that does not participate in the initial pitting corrosion gradually rise above the steel surface and eventually enter the corrosion products. Figures 11(a1) and (b1) show the distribution of Cu around the Al 2 O 3 and RE oxide particles before the immersion corrosion started. Cu is evenly distributed inside the steel matrix and there is no Cu segregation. When the immersion corrosion begins, an unstable corrosion product area is formed, as shown in figures 11(a2) and (b2), which corresponds to the state observed in figures 9(b) and 10(b). Because Al 2 O 3 particles do not affect Cu distribution, Cu is distributed evenly in the corrosion products and may be shed into the corrosion environment at any time with unstable corrosion products. The enrichment of Cu near RE oxide particles reduces the possibility of shedding alloying elements into the environment. With further corrosion, nonmetallic particles are gradually packaged by the corrosion products, as shown in figures 11(a3) and (b3). At the final stage of corrosion (figures 11(a4) and (b4)), Al 2 O 3 particles, as massive solids in the rust layer, are easily detached from the rust layer, leaving pits in the layer. The RE oxide particles adhere tightly to the corrosion products and do not easily fall off to form pits. As the thickness of the corrosion products increases, the interface between the inner rust layer and the substrate enriched in RE and Cu will be covered by the rust layer after a long period of corrosion.
The line-scan analysis of some elements in the NHAl and NHRE steels after 24 h of immersion corrosion is shown in figure 12. It can be seen that corrosion products do not cover the entire steel substrate owing to the very short corrosion time. Therefore, the formation mechanism of corrosion products can only be pit corrosion induced by nonmetallic particles, rather than elemental diffusion and interface segregation caused by long-term corrosion. Obviously, in addition to the O element, Cu elements are enriched in the rust layer of NHRE, as shown in figure 12(b). However, no such enrichment is found in NHAl of figure 12(a). This indicates that the RE oxide particles promoted the enrichment of Cu in the rust layer during the local corrosion process.
Cu accumulates in the RE oxide particles during the corrosion of steel, as shown in figures 9 and 10, which increases the Cu content in the corrosion products. This conclusion is the same as that of Jia et al [25]. The initial localized corrosion on the steel surface can be considered as a tiny primary battery, where the high-activity site acts as the anode and the low-activity site acts as the cathode. At the local corrosion site, metal cations and hydroxide anions combined to form M x (OH) y , leading to acidification beneath the pitting site. This promotes the enrichment of H + below the local corrosion site, forming a micro acid region [35]. Conductive and stable Cu atoms can easily assume a negative charge during local corrosion. When the surrounding steel substrate is corroded and becomes soft, the negatively charged Cu tends to move to the micro acid region, transferring electrons to H + . Hence, to achieve Cu enrichment on the nonmetallic particles, Cu atoms must move freely in the steel substrate during the corrosion process, and a micro acid zone must appear on the nonmetallic particles. For the Al 2 O 3 and RE oxide particles that participate in the initial pitting corrosion, although micro acidic region can be generated [4,35], Cu is not enriched or less enriched due to their small number and immovable Cu atoms, as shown in figures 13(a) and (b). For Al 2 O 3 particles that do not participate in the initial pitting corrosion, a micro acid region cannot be formed around them owing to their high stability [4,33,40]; thus, Cu is not enriched ( figure 13(c)). However, RE oxide particles that do not participate in the initial pitting corrosion can decompose and a micro acid region is formed [35]; therefore, Cu is enriched ( figure 13(d)). Elemental line scans after 24 h of corrosion also show that steel with RE oxide particles does not require months or even years of prolonged corrosion and can significantly contribute to the enrichment of Cu in the rust layer at the initial pitting stage ( figure 12).  (2)After the introduction of RE elements into the steel, the number density of nonmetallic particles in the NHRE steel is refined as compared to that of NH steel, but not as effective as Al 2 O 3 -containing particles. Cu is Figure 11. Corrosion process of (a1)-(a4) Al 2 O 3 and (b1)-(b4) RE oxide particles that do not participate in the initial pitting corrosion.  enriched around the RE oxide particles in NHRE steel during the pitting corrosion process. Mobile Cu with an excess negative charge and the generation of a micro acid region around the RE oxide particles are the two necessary conditions for the Cu enrichment. The NHRE steel exhibits the best corrosion resistance.