Effect of trace rare earth elements (Ce) on corrosion resistance of high strength steel used for offshore platform

In this study, a high-strength, offshore platform steel, EH420, was used as the original steel. By observing the original austenite morphology, the corrosion resistance of three experimental steels with different rare Earth contents in a simulated marine splash zone was compared and studied using cyclic immersion corrosion and electrochemical corrosion tests. The macro- and micromorphologies of three experimental steels in different cycles of corrosion, as well as the composition of rust layers, were observed and studied by employing an optical microscope, scanning electron microscope (SEM) and X-ray diffraction (XRD). The crystal microstructure of the three experimental of steels was analyzed by EBSD, and the electrochemical behavior of the three experimental of steels after corrosion was analyzed by electrochemical polarization curve and electrochemical impedance spectroscopy. The results show that the trace additions of Ce ( less than 20ppm ) in the steel can refine the grain size, reduce the grain boundary energy, promote the formation of the protective phase α-FeOOH and Fe3O4 of the rust layer, improve the compactness of the rust layer, and reduce the corrosion rate. In addition, the charge transfer resistance ( Rct ) and open circuit potential are increased, and the corrosion current density is reduced. Moreover, the higher the content of Ce is added in the steel, the more obvious its effect.Therefore, the trace additions of Ce improves the corrosion resistance of high strength offshore platform steel.


Introduction
A marine corrosion environment can be vertically divided into five different corrosion areas: the atmospheric zone, splash zone, tidal zone, full immersion zone and marine mud zone. The splash zone undergoes the most serious corrosion out of the five areas, and the corrosion process that occurs here is very complex due to the long-term effects of external factors such as seawater splash erosion, dry-wet alternation, corrosive components in the atmospheric environment, and oxygen, wind and solar radiation [1,2]. Therefore, research into the improvement of the corrosion performance of offshore platform steel in the marine splash zone is highly important.
A large number of experimental studies have shown that adding a certain amount of rare Earth elements in steel can effectively improve the corrosion resistance of steel [3][4][5][6][7][8][9][10][11]. In recent years, Cao Yuxin et al [12,13], through the combination of first-principle calculation and experiments, used mechanisms to determine that the work functions of rare Earth inclusions such as CeAlO 3 and Ce 2 O 2 S were lower than those of the steel matrix. In the early stage of corrosion, the rare Earth inclusions themselves were shown to corrode before the steel matrix to protect the matrix from corrosion. In addition, Ce 3+ can react with oxygen to generate CeO 2 , which consumes oxygen, reduces the concentration of oxygen and further hinders the reaction of steel substrate with oxygen.Most of these results are based on laboratory studies, and the content of rare Earth elements was high (30∼1200 ppm) in these studies. Although rare Earth elements can improve the corrosion resistance of steel, they make the generation of a large number of large-particle inclusions easier, which deteriorates the impact performance of steel and hinders the industrial application of rare Earth elements in steel [14]. Li Hua et al [15] compared the corrosion resistance of EH36 steel with 0.0009 wt% Ce and EH690 steel with 0.0012 wt% Ce. Through salt spray corrosion experiments with different corrosion cycles, it was found that the corrosion rates of the two groups of experimental steels with trace rare Earth elements (less than 20 ppm) were significantly lower than those of the prototype steel without the addition of rare Earth elements. Hereby, on the premise of ensuring the comprehensive mechanical properties of high strength offshore platform steel, adding trace rare Earth elements in steel to improve its corrosion resistance has practical application value,In recent years, because researchers in improving the corrosion resistance of steel by adding a large number of rare Earth elements, but ignore the trace rare Earth elements on the corrosion resistance of steel also plays a very important role, therefore, for the trace rare Earth elements added in the steel is due to inclusions affect the corrosion or rare Earth elements segregated in the grain boundary affect the corrosion mechanism is not perfect, secondly, trace rare Earth elements on the corrosion of steel rust layer structure or have an impact on the matrix research evidence is not comprehensive. Therefore, in this paper, a high-strength, offshore platform steel, EH420, was used as the prototype steel, the seawater corrosion resistance of three experimental steels with different trace Ce contents (less than 20 ppm) under a simulated marine splash zone environment was studied using periodic immersion corrosion and electrochemical corrosion tests. The best level of Ce, which caused the best corrosion resistance in steel, was found, and the influence of trace Ce content on the corrosion resistance of steel was revealed, which has significance in theoretically guiding the promotion of the industrial application of rare Earth in steel.

Experimental materials
Three kinds of test steels were prepared in a 100 kg vacuum induction furnace, hot-rolled into 25 mm thick plates, air-cooled to room temperature, quenched at 900 ℃, and tempered at 600 ℃. The chemical composition of the three groups of test steels is shown in table 1. Among them, 13 ppm and 19 ppm of cerium (Ce)was added to test steel 2# and 3#, respectively.

Experimental method
The three experimental steel specimens with dimensions of 20 mm×30 mm×25 mm were subjected to heating at 900 ℃ for 40 min, and then, water quenching was applied to observe the original austenite morphology. Original austenite grain boundary is etched by picric acid solution ( dissolving 2.5 g of picric acid and 1.0 g of seagull shampoo in 30 ml of distilled water and adding 7 drops of hydrogen peroxide).The metallography was observed using a Zeiss metallographic microscope, and the size of grain was calculated using Image Pro Plus software.
For the cycle immersion test, in accordance with the GB / T19746-2005 standard metal and alloy corrosion salt solution cycle immersion test, in which its test chamber model used was EN-08, the corrosion period was 96 h, 192 h, 288 h, 384 h, 480 h and 576 h, and the corrosive liquid was 3.5% sodium chloride solution. The samples were immersed in the solution for 15 min, baked in a drying oven for 45 min and cycled every 60 min. Furthermore, 100∼200 ml of deionized water was added to the solution every 24 h, and the corrosion solution was changed every 4 days. The corrosion samples were processed into two sets with dimensions of 20 mm×30 mm×3 mm and 10 mm×10 mm×3 mm. The former set was used to calculate the corrosion rate, and the latter was used in electrochemical tests, rust layer morphology analysis and rust layer structure composition analysis. In all of the samples, 2 mm holes were drilled 1.5 mm away from the edge. Before the experiment, the samples were polished to 2000 mesh with sandpaper, cleaned ultrasonically and dried with cold air. The samples which were used to calculate the corrosion rate were weighed after ultrasonic cleaning. At the end of the corrosion process, macroscopic photos of the samples were taken, and the rust layer was gently scraped using wood chips. The rust layer was removed with the corrosive medium (which was obtained by dissolving 500 ml of hydrochloric acid and 5 g of hexamethylenetetramine solution in 500 ml of distilled water), cleaned ultrasonically and dried with cold air. Then, the corrosion rates of the samples were calculated, the surface morphologies were observed using SEM and the composition of the rust layer was analyzed using XRD with Cu target, 10°∼80°of 2θ and a 4°min −1 scanning rate. The microstructure and crystallographic analysis of the three experimental steels were combined using EBSD technology to analyze the effect of adding different contents of Ce on corrosion resistance. The electrolytic polishing solution of the three groups consisted of 20% HCl and 80% CH 3 OH. In addition, the voltage was 18 V and the time taken was 5 s ∼ 8 s. Following electrolysis, the samples were observed using an FEI QUANTA400 scanning electron microscope with an EBSD lens, and Channel5 software was used for later data processing and analysis.
The electrochemical test was carried out according to GB/T 24196-2009. The rust samples taken in each corrosion cycle were tested using a Zennium electrochemical workstation. The system was a three-electrode system. Among them, a Pt electrode was used as the auxiliary electrode, a saturated calomel electrode was used as the reference electrode, and the test sample was used as the working electrode. This test is capable of determining the dynamic potential polarization curve and electrochemical impedance spectroscopy of electrochemical samples. The polarization potential range was −0.25 V ∼ 0.25 V (relative to open circuit potential), and the scanning speed was 1.0 mV s −1 . The EIS AC amplitude was 10 mV, and the frequency was 10 −2 Hz ∼ 10 5 Hz. Once the open circuit voltage was stable, we began to test the AC impedance and polarization curve.

Experimental results and analysis
3.1. Original austenite morphology Figure 1 shows the original austenite morphology and size distribution of three experimental steels at 1/4 full thickness. It can be seen from the figure that the average grain sizes of the three experimental steels were 7.47 μm, 5.92 μm and 5.68 μm, respectively. The grains between 3.4 and 8.5 μm of steel 1# made up about 38.75% of the steel, and large grains above 13.6 μm made up about 6%. The proportion of grains between 3.3 and 7.2 μm in steel 2# was about 72.07%, and the large grains above 13.6 μm made up about 0.3% of the steel. Grains between 3.2 and 7.6 μm in steel 3# accounted for about 73.49% of the steel, and no large grains above 13.6 μm were present. Among the three experimental steels, the grain size in steel 3# was the most uniform. The addition of trace Ce to offshore platform steel can increase the degree of supercooling, improve the austenite nucleation rate and refine the grains [16]. Therefore, the grains in specimen 2# and 3# were smaller than those in sample 1#.
After adding Ce, the original austenite grain size in experimental steels 2# and 3# was significantly smaller than that in experimental steel 1#, in which additions of trace Ce were not made. Grain refinement has previously been shown to reduce the ratio of the grain size area of cathodic reduction corrosion to the total corrosion area, thus hindering the occurrence of pitting corrosion and promoting homogeneous corrosion [17]. In addition, the segregation of rare Earth elements at grain boundaries has been shown to reduce the interface energy and avoid the occurrence of local corrosion. Wel G et al [18,19] showed that the rare Earth atoms with larger atomic radii dissolve in the lattice of iron atoms with smaller radii, which inevitably causes lattice distortion. In order to reduce the lattice distortion energy and reduce the free energy of a system, rare Earth atoms with large radii preferentially aggregate at grain boundaries and near defects. The segregation of rare Earth atoms purifies the grain boundary, thereby reducing the induced source of pitting corrosion, improving the selfcorrosion potential of steel, promoting the transformation of active components in corrosion products to stable α-FeOOH and enhancing the stability of the rust layer. Therefore, in the corrosion resistance results shown below,the corrosion resistance of experimental steels 2# and 3# was better than that of experimental steel 1#.

Corrosion rate
The average corrosion rate is calculated using the following formula (1): In the formula (1), CR-average corrosion rate (mm/y); 87600-a constant measured in mm units of the annual corrosion rate; Δm-matrix mass change before and after corrosion of the sample (g); S-total surface area of the sample (cm 2 ); ρ-sample density (g/cm 3 ); t-corrosion time (h).
Three groups of parallel samples were prepared for each group of test steel, and the average corrosion rate of three groups of test steel was calculated using Formula (1), as shown in figure 2. Among them, the discreteness of each group of parallel samples of three groups of test steel under different corrosion cycles was shown to be small. It can be seen from the figure that the curves were divided into three stages: the corrosion rates in the first stage increased during erosion periods from 96 h to 288 h. In the early stage of corrosion, the corrosion solution easily made contact with the substrate, and the oxygen in the corrosion solution was sufficient, which accelerated the corrosion of the experimental steels. Between 192 h and 288 h, the outer rust layer of the test steels gradually became uniform, so the corrosion rate began to slow down. The corrosion rates in the second stage from 288 h to 480 h declined, because an abundance of corrosion products, such as α-FeOOH and Fe 3 O 4 , were generated to protect the substrate of the steels as the corrosion time increased. Meanwhile, the oxygen content in the corrosion solution decreased, which effectively hindered the electrochemical reaction. In the third stage, from 480 h to 576 h, the corrosion rates tended to be stable, because continuous, dense, internal and external rust layers formed on the surface of the specimens, which effectively cut off contact between the matrix and corrosive medium, meaning the corrosion rates slowed down. In addition, the corrosion rates of both steel 2# and 3# with Ce were lower than that of steel 1#, and the corrosion rate of steel 3# was the lowest with the largest content of Ce. It was because Ce adding into the steel promoted the transformation of the unstable phases into stable phases, such as α-FeOOH, Fe 3 O 4 and other protective products in scales, thus slowing down the erosion rate of both steel 2# and 3#.

Morphology of corrosion samples 3.3.1. Macroscopic corrosion morphology
The macroscopic corrosion morphology of the three experimental steels under different corrosion cycles is shown in figure 3, which shows that the rust layers were distributed uniformly on the substrate. In the early stage of corrosion, the color of the rust layer was light yellow. With the extension of the corrosion period, the color of the rust layer first changed to yellowish-brown and later turned brown, and black blisters even appeared in some parts. The variation in the color of the rust layer was due to the transformation of free iron ions into ferrous hydroxide, ferrous oxide, iron hydroxyl oxide, etc Thus, stable protective products such as Fe 3 O 4 and α-FeOOH were formed in the later stage of corrosion, meaning that the color of the rust layer hardly changed.
When the corrosion time was 96 h, partial substrates on the surface of the three experimental steels were observed to have not corroded. As the corrosion progressed, after samples were etched for 288 h, the color of the rust layer turned yellow-brown. There was a mass of bubbles on the surface of experimental steel 1#, whereas the rust layers of steel 2# and 3# were denser. In the late stage of corrosion, there were black bubbles on the substrates of the three experimental steels, among which, the rust layer on the surface of experimental steel 1# was spalling. Compared with steel 1#, steel 2# and 3# had fewer and smaller blisters. Experimental steel 3# (with the highest Ce content) had denser rust layers than steel 2#. The formation of black bubbles was because the experimental steels were subjected to alternate dry-wet environments. Some parts of a steel substrate undergo corrosion stresses, which promote the swelling of rust layers, meaning corrosive ions make contact with the matrix, thus accelerating the corrosion reaction [20]. Therefore, it can be seen from the macroscopic corrosion morphology that with the trace additions of Ce elements, the rust layers of experimental steel 2# and 3# were more continuous, there were significantly fewer and smaller blisters and the seawater corrosion resistance was better.

Micro-corrosion morphology
The microscopic morphology of the three experimental steels after rust removal and at the end of the immersion corrosion experiment is shown in figure 4. From the figure, it can be seen that the three experimental steels corroded uniformly. When the corrosion time was 96 h, there were some uncorroded areas on the surfaces of the specimens; furthermore, the corrosion pits were shallow and small. After 288 h of corrosion, deeper corrosion pits on the surfaces of the samples were formed. As the corrosion time was prolonged to 576 h, most of the steel matrixes were dissolved, meaning the surrounding pits joined together to form a larger and deeper pit.
Comparing the micromorphology of the three group samples, the corrosion in sample 1# was worst, in which the corrosion pits were the largest and the deepest. Measured using Nano Measure software, the maximum diameter of corrosion pits in steel 1#, 2# and 3# were 409.23 μm, 308.39 μm and 283.48 μm, respectively. Therefore, it was shown that the corrosion resistance of steel 3# was the best, followed by experimental steel 2#.
Studies have shown that [17] initial corrosion occurs around inclusions to form microarea corrosion because the potential of inclusions differs greatly from that of the steel matrix; meanwhile, corrosive pits are formed on the surface of steel. With the development of electrochemical corrosion, steel substrate is activated continuously as an anode, which accelerates the corrosion process so that pitting corrosion gradually transforms into uniform corrosion on the steel surface. Long inclusions such as MnS and irregularly shaped Al 2 O 3 with sharp corners transform into spherical or ellipsoidal rare Earth sulfur oxides and rare Earth aluminum oxides because of trace additions of Ce. In addition, the size of the inclusions is also significantly reduced, which decreases the activation area for corrosion and prevents corrosion [21]. Compared with steel 1#, the corrosion pit areas of test steel 2# and 3# were reduced, which indicates that the trace additions of Ce reduced the probability of pitting corrosion, promoted the occurrence of uniform corrosion and thus improved the seawater corrosion resistance of steel.

Analysis of corrosion products
The phases of rust layers in the three group specimens after periodic wetted corrosion for 596 h were measured using XRD, and their analysis data were mapped using Origin software, as shown in figure 5. The main phases of rust layers in the three experimental steels, such as α-FeOOH, γ-FeOOH and Fe 3 O 4 , were the same, but their contents varied.
In the early stage of corrosion, loose corrosion rust layers were formed, which accelerated the corrosion rate because sufficient corrosive media, such as water and Cl − , easily made contact with the steel matrix. As the corrosion process continued, protective phases began to form a dense and continuous rust layer, thus slowing down the corrosion rate. The corrosion reaction process is shown as follows [22,23]. The formation of α-FeOOH and γ-FeOOH is an irreversible oxidation reaction, and the reaction process is as follows: The recent researches show that the larger the number of the protective coefficient α/γ * value is, the tighter the rust layers form, which means the corrosion resistance is better for carbon steel. The α is means to the content of α-FeOOH in the rust layer, and γ * refers to the sum contents of γ-FeOOH, β-FeOOH and spinel (S) oxide in the rust layer where spinel ( S ) contains Fe 2 O 3 or Fe 3 O 4 [24][25][26][27]. Table 2 shows the semi-quantitative analysis results regarding the rust layers in three experimental steels. The α/γ * values of experimental steels 1#, 2# and 3# were 0.26, 0.353 and 0.414 respectively. The α/γ * value increased with the trace additions of Ce, which indicates that the addition of rare Earth elements promotes the formation of stable α/-FeOOH phase and Fe 3 O 4 phase, thus improving the corrosion resistance of experimental steel 2# and 3#.

Microcrystalline structure
The corrosion resistance of the crystallographic microstructure of the three experimental steels was analyzed using EBSD technology. Figure 6 shows anti-pole diagrams for the three experimental steels. It can be seen from the figure that the three experimental steels had no preferred orientation; this shows that the corrosion type for the three experimental steels was uniform corrosion. Figure 7 shows the grain boundary orientation difference distribution in the three groups of test steels. In the figure, the distribution frequency of the small-angle grain boundaries and large-angle grain boundaries of the three experimental steels is compared. It was found that the distribution frequency of the small-angle grain boundaries of test steel 3# was the highest, and the proportion of the large-angle grain boundaries was the smallest, followed by test steel 2#. Some studies have shown that a lowangle grain boundary (LAGB) has a grain boundary orientation difference of less than 15°, and a high-angle grain boundary (HAGB) has a grain boundary orientation difference of more than 15° [28,29]. Among them, smallangle grain boundaries mainly include sub-grain boundaries, and large-angle grain boundaries mainly include original austenite grain boundaries and granular bainite grain boundaries. They have high energy and irregular atomic arrangements [30,31]. In addition, these large-angle grain boundaries are high-frequency areas of  corrosion. However, small-angle grain boundaries are exactly the opposite, in which the energy of the grain boundary is low and the corrosion resistance is better. The distribution frequency of the small-angle grain boundaries of test steel 1# was the smallest, and the distribution frequency of the large-angle grain boundaries was the largest. Therefore, from this perspective, it can be concluded that the addition of Ce improves the distribution proportion of small-angle grain boundaries in steel and inhibits the corrosion performance of steel, this effect is more obvious for a higher content of Ce. Figure 8 shows the distribution of coincidence site lattice in the three experimental steels, where Σ is the ratio of unit cell of coincidence site lattice to that of actual lattice. As we know, the lower the ratio is, the higher the frequency of lattice points between two intersecting lattices coincides [32][33][34]. Figure 9 shows the statistics for grain boundary distribution of coincidence site lattice of the three experimental steels. Among them, Σ3 is a coherent grain boundary which has low energy, little impurity segregation and is difficult to migrate. These characteristics show that the more Σ3 grain boundaries in steel, the more obvious corrosion resistance can be improved. It is not difficult to see from the diagram that after adding  trace rare Earth elements, the Σ3 of experimental steels 2# and 3# was more obvious than that of sample steel 1#. Therefore, from the angle of lattice grain boundary at the crystal coincidence position, it can also be seen that after adding Ce, the corrosion resistance of experimental steels 2# and 3# is expected than that of test steel 1#, which rare Earth elements were not added to.   Figure 11 shows the polarization curves of the three experimental steels that were immersed for 96 h. It can be seen that the polarization curves of the three experimental steels are the same; that is, the anode was activated and dissolved, but the self-corrosion potential of experimental steel 3# with the highest addition of trace Ce was the highest, followed by experimental steel 2# and 1#. From the thermodynamic point of view, corrosion potential reflects the thermal stability of a material, that is, at the same current density, the corrosion resistance of steel improves with the increase of anode potential [35,36]. It can be seen from table 3 that the self-corrosion potential was close to the open circuit voltage, that is, the self-corrosion potentials of experimental steels 2# and 3# were slightly higher than that of experimental steel 1# by 3.317% and 6.075%, the current densities were  lower by 9.52% and 24.82%, and the corrosion rates were lower by 9.522% and 24.821%, respectively. This shows that the higher the increase in Ce content is in steel, the more effectively the corrosion tendency is inhibited, which hinders the charge transfer between the double layers, reduces the electrochemical reaction rate and improves the corrosion resistance of the steel.

EIS
The electrochemical impedance results regarding the three test steels immersed for 96 h are shown in figures 12 and 13. It can be seen from the Nyquist diagram that the impedance diagrams of three test steels are all composed of a capacitive arc, which indicates that the corrosion mechanism was the same and was controlled by charge   transfer in solution. Among them, the size of the capacitance-resisting arc reflects the ability to hinder charge transfer. The larger its diameter, the better corrosion resistance of steel is [37]. Thus, the order of the corrosion rates of the three experimental steels from fast to slow was determined to be 1# > 2# > 3#. It can be seen from the Bode diagram that the variation trend of the impedance modulus values of the three experimental steels is consistent. In the low-frequency region (10 −2 Hz ∼ 1 Hz), the impedance modulus values of experimental steels 2# and 3# were higher than that of experimental steel 1#, and the impedance modulus value of experimental steel 3# was highest after trace additions of Ce. In the high-frequency region (10 3 Hz ∼ 10 5 Hz), the impedance modulus values of the three experimental steels were similar. Zview was used to simulate the equivalent circuit of the impedance of the three experimental steels immersed for 96 h, as shown in figure 14, and the fitting value of the equivalent circuit is shown in table 4. In the electrochemical experiment, the set scanning rate was slightly larger, so slightly fewer experimental points were scanned, resulting in a high chi-square value (χ). However, in a fitting circuit, the maximum χ value cannot exceed 10%, that is, the fitting equivalent of the three experimental steels immersed for 96 h met the requirements. Rs represents the solution resistance, which is related to the composition and concentration of the solution, so its effect can be ignored in the analysis of impedance [38]. Y0 is the pre-factor of phase angle, and n is the dispersion factor, ranging from 0 to 1. When its value is close to 1, the dispersion effect is stronger, and the more likely it is for the phase angle element to be pure capacitance. In other words, the electrode surface is close to smooth, in which the current density distribution uniform [39]. It can be seen from the table that after soaking for 96 h, the n value of three experimental steels did not change greatly, but the n value of experimental steel 3# was slightly higher than that of experimental steels 1# and 2#. When comparing the charge transfer resistance Rct of the three experimental steels, it can be seen that the charge transfer resistance of experimental steels 2# and 3# was 17.33% and 39.32% higher than that of 1#, respectively. The results in the polarization curves and impedance spectra show that the seawater corrosion resistance of high-strength, offshore platform steel can be effectively improved by trace additions of Ce. As the amount of Ce content added increases, the seawater corrosion resistance of high-strength, offshore platform steel improves.

Conclusions
(1) The addition of trace rare Earth element Ce in steel refines the original austenite grain size, increases the low angle grain boundary ( LAGB ) and Σ3 coherent grain boundary, reduces the grain boundary energy, and effectively hinders the occurrence of pitting corrosion. In addition, the higher the content of Ce is added in the steel, the effect is more obvious.
(2) The addition of trace rare Earth element Ce in steel promotes the formation of rust layer protective phases such as α-FeOOH and Fe 3 O 4 , reduces the corrosion rate, enhances the adhesion and compactness of rust layer, hinders the contact between corrosive ions in solution and steel matrix, and inhibits the corrosion of steel. Moreover, with the addition of rare Earth Ce content, the rust layer is denser and the corrosion resistance is better.  (3) The addition of trace rare Earth elements increased the charge transfer resistance ( Rct ) and open circuit potential into the steel, reduced the corrosion current density and improved its corrosion resistance. Furthermore, the more the content of trace rare Earth Ce added in the steel, the more obvious the charge transfer resistance and open circuit potential increase, and the current density decreases significantly.