Sand impingements on cathodic protection of marine carbon steel in natural sea water

ABSTRACT In this work, the influence of sand impingements on cathodic protection (CP) of marine carbon steel in natural sea water was studied in comparison with the CP performances in static and flowing sea water without sand particles. Results show that the propagation of local active anodic dissolution at the steel inclusions could be totally inhibited by providing sufficient cathodic current in the sea water free of sand particles. However, the local ‘CP shielding’, potential fluctuation and sand impingements would enhance the anodic dissolution at inclusions, thus leading to the occurrence of erosion-corrosion under CP. The negative shift of the CP potential could not retard the propagation of erosion-corrosion pits under sand impingements. The synergy of erosion and corrosion might result in long-term failure of local CP.


Introduction
Carbon steels are widely used to construct marine structures and equipment, such as offshore platforms, subsea pipelines, ships and underwater vehicles [1][2][3]. The marine carbon steel would suffer serious degradation in corrosive sea water without proper protection [4]. Cathodic protection (CP) is one of the most effective and useful methods to protect the marine steel from corrosion [5,6]. The main purpose of CP is to supply additional electrons which could enhance the interfacial cathodic reaction and minimise the anodic reaction [7,8]. It is normally thought when the applied potential is negative than −800 mV (vs. Ag/AgCl) for mild steel or low alloy steel, the corrosion could be mostly eliminated in aerobic sea water [8]. However, the employed cathodic potential should not be more negative than −1150 mV (vs. Ag/ AgCl) to avoid the significant hydrogen evolution [7].
Although the general corrosion of carbon steels could be successfully inhibited using CP techniques in most cases, the pitting corrosion induced by the local CP failure is still a tough issue in practical engineering applications [9]. It is reported that the fluctuations of the applied CP potential or current induced by instable power supply or stray currents could result in pitting initiation around the inclusions and local defects [10][11][12]. On the other hand, the formation of local crevices induced by coating disbandment, deposit covering or contacting of steel parts also could lead to the local CP failure which is normally called 'cathodic shielding' [13][14][15][16]. The 'cathodic shielding' indicates that cathodic current is hard to enter the crevice due to the high solution resistance, thus sometimes resulting in local anodic dissolution [15]. It is acknowledged that stress corrosion cracks and hydrogen embrittlement normally initiate from the corrosion pits [7,10,17]. The pitting damage induced by the local CP failure becomes an important factor for the early failure of marine structures.
The sand particles are always entraining in sea water, especially at the areas around estuaries and seabeds [18,19]. The relative movement between the corrosive slurry and steel surface could induce the occurrence of erosion-corrosion [20][21][22]. It is reported that the erosion steel loss of carbon steel is mostly induced by corrosion-enhanced erosion when the flow velocity of sea water is lower than 5 m s −1 [18]. The surface hardness degradation caused by anodic dissolution is the basic requirement for the initiation of obvious erosion in this case [23,24]. It is normally believed that CP could effectively inhibit both corrosion and erosion of carbon steel in sea water of a relatively low flow velocity [25,26]. In addition, the application of CP is one of the most common methods to obtain the pure erosion component in erosioncorrosion studies [27]. However, it is found that the local sand impingements could lead to a short surface disturbance when the sand particle enters the interfacial boundary layer between the solid steel surface and bulk solution [28][29][30][31]. The disturbance might lead to the fluctuation of CP potential and current in a short period [28,32]. Once the anodic reaction occurs during the CP disturbance, the following sand impacts might induce local plastic deformation or steel flaking [18,33]. As a result, the potential erosion-corrosion risk under CP still exists in theory. However, the duration of the sand particle in the interfacial boundary layer is short [28]. No previous researches thought that erosion could be a factor leading to the local CP failure. Therefore, it is still unclear whether the sand impingements could result in the localised corrosion under CP. In order to figure out this issue, the influences of sand impingements on CP of EH 36 marine carbon steel were studied in this work. The CP performances in the static sea water and flowing sea water without sand particles are also compared to facilitate the understanding of the effect of sand impingements on CP. This research aims to investigate the potential erosion-corrosion risks of carbon steel under CP.

Methods
The test setup used in this work is schematically presented in Figure 1(a). The rotation disc system (Yunchi, China) is same as the one used in previous studies [23,34]. The diameter and the thickness of the rotation disc are 60 and 20 mm, respectively. Independent working electrodes (WEs) were mounted in the cylinder test cell of 80 mm in diameter. The WEs employed in this work are made of EH 36 marine carbon steel which chemical compositions are (mass%) C 0.06, Si 0.22, Mn 1.48, P 0.009, S 0.002, Nb 0.03, Al 0.04, Cu 0.03, Cr 0.14 and Fe balance. The working surface of the WE was machined in the size of 7 mm × 7 mm (Figure 1(b)). The WE was gradually polished from 400 to 1200 grit papers before the test. A titanium mesh and an Ag/AgCl electrode were used as counter electrodes (CE) and reference electrodes (RE), respectively. The CP was applied using a Reference 600 + electrochemical workstation (Gamry, US) which acted as a potentiostat in this work. The natural sea water taken from Dalian Sea area was employed as the test solution, which main ion contents (g/L) are Cl − 17.09, Na + 9.45, SO 4 2− 2.20, Mg 2+ 1.06, Ca 2+ 0.03, K + 0.26, HCO 3 − 0.13, Br − 0.03. The sea water was maintained at 30°C using a water bath. The setting temperature of the solution aims to restore the marine environment at South China Sea where the highest sea water temperature could reach 32°C in recent years. Silica sand particles with an average diameter of 430 μm (Figure 1 (c)) were employed in this work.
Three group of experiments were conducted under different stirring conditions: in the static sea water, in the sea water under rotation speed of 800 rev min −1 and in the sea water under rotation speed of 800 rev min −1 with 10% sand added (by weight). At each flow condition, the WE was tested at four CP potentials of −800, −900, −1000 and −1100 mV (vs. Ag/AgCl), respectively. The test duration of each potential was 24 h. The three groups of tests were compared to understand the influence of sea water flow and sand impacts on CP effect. After the 24 h of test at each flow condition, the WE was immediately photographed by an EOS digital camera (Canon, Japan). Then, the surface morphology was characterised by an EM-20AX Plus scanning electron microscopy (SEM, Coxem, Korean) and energy-dispersive spectrometry (EDS) analysis. Thereafter, the deposited caliche or corrosion products on the steel surface were cleaned by a brush in conjunction with the solution suggested in ASTM G1-03. The steel surfaces with the removal of deposits and rusts were observed by SEM and EDS as well. The local 3D profiles were further examined by an OLS 5000 infinite microscope (Olympus, Japan). Each group of tests was repeated twice to ensure the repeatability.

Results
The CP performance in static sea water The dynamic changes of the cathodic current in static sea water at different CP potentials are plotted in Figure 2. The measured cathodic current of the repeated test is plotted in the Supplementary Materials. It is seen that the initial current shifts negatively along with the applied potential changing from −800 to −1100 mV. The magnitudes of the CP current show a significant decrease to around 4 × 10 −3 mA after 24 h of test at −800 to −1000 mV. The current of the steel at −1100 mV has an obvious negative shift from −0.078 to −0.133 mA in the first 4 h of immersion. Then, the current shifts positively to −0.057 mA at the end of the test. The higher cathodic current at −1100 mV might be induced by the hydrogen evolution.
The surface morphologies of the WEs after 24 h of CP test in static sea water are shown in Figure 3. It is seen from Figure 3(a) that the steel surface is covered by a thin white layer with the mixture of brown rusts at −800 mV. The white layer gradually becomes thicker and covers the whole steel surface at −900 to −1100 mV ( Figure 3(b-d)). It is seen from the local SEM image that obvious precipitation of needle-like crystals is found on the thin white layer at −800 mV. The precipitation of the needle-like crystals becomes more significant at −900 to −1000 mV. The whole steel surface is covered by the crystals at −1100 mV and some grass-like precipitations are also found. It is further seen from the local EDS results of Points 1 (Figure 3(a-d)) that the components of the needle-like crystals are Ca, C and O for all the CP potentials, indicating that the composition of the needle-like crystal is CaCO 3 . It is seen from the EDS results of local Points 2 ( Figure 3(a,b)) that only Fe element is found, suggesting that the bottom white layer is too thin to be identified by EDS at −800 and −900 mV. Along with the CP potential reaching −1000 mV, the Ca element in the bottom layer could be sensed by EDS ( Figure  3(c)). It is deduced from the EDS results of Point 2 ( Figure 3 (d)) that the grass like deposits generated at −1100 mV is Mg (OH) 2 . It is seen from the test results that the stable cathodic currents were almost the same at −800 to −1000 mV after 24 h of CP. The decrease of the cathodic current is induced by the surface alkalisation due to the generation of hydroxyl under CP, which could result in the precipitation of a thick calcite layer. Although a more compact calcite and brucite layer forms on the steel surface at −1100 mV in static sea water, the hydrogen evolution would lead to a higher stable cathodic current in this case.
Since the whole steel surface is covered by a layer of caliche after 24 h of CP in static sea water, the steel degradation beneath the calcium layer could not be directly observed. Accordingly, the depositions on the steel surface are cleaned and the steel morphologies with the removal of depositions are presented in Figure 4. It is seen from the photos ( Figure  4(a-d)) that the steel surfaces are shining and the polish scratches could be clearly seen, indicating that general corrosion is well inhibited by CP. However, some small pits are found from the local SEM observations. It is seen from the local EDS results that the local steel dissolutions are initiated from the steel inclusions where Ca, Si and Mn are probed. The initiation of the small pits from the steel inclusions under CP is also found in previous studies [11]. It is noted that the formation of small pits under CP is different from the traditional definition of pitting corrosion at open circuit potential (OCP), which is generally induced by the local acidification and the enrichment of chloride ions. The small pits are most possibly induced by the local active dissolution where more negative OCP of the inclusions are registered. The geometrical parameters of the small pits could be further seen from the typical 3D profiles. It is observed that the diameter of the pit has an obvious decrease from 7.5 to 4.3 μm along with the negative shift of the CP potential from −800 to −900 mV. The diameters of the pits are in the range of 3.0 to 4.5 μm at −900 to −1100 mV. The depth of the pit decreases from 5.6 to 0.9 μm along with the CP potential changing from −800 to −1100 mV, indicating that a more negative CP potential could inhibit the longitudinal growth of the pits in static sea water.
The CP performance in flowing sea water The dynamic changes of the cathodic current in flowing sea water at different CP potentials are plotted in Figure 5. The measured cathodic current of the repeated test is plotted in the Supplementary Materials. It is seen that the variation trends of the current at different CP potentials are similar, which show an obvious negative shift at the initial immersion stage and then immediately climb to high positive levels in the latter stage. The magnitude of the initial cathodic current at −800 mV is 0.184 mA, which is the lowest among all the CP potentials. The current of −800 mV finally becomes stable around −0.102 mA after 24 h of test, which is the most negative value of all the cathodic currents. The initial cathodic currents of −900 to −1100 mV are close, which are −0.389, −0.482 and −0.453 mA, respectively. The cathodic currents of −900 to −1100 mV gradually become stable after the significant positive shifts of the current. The cathodic current of −1100 mV has a slight decrease at the end of the test. The final cathodic currents of −900 to −1100 mV are −0.034, −0.034 and −0.058 mA, respectively.
The surface morphologies of the WEs after 24 h of tests in flowing sea water are shown in Figure 6. It is seen from Figure  6(a) that the steel surface is covered by a thin layer with the mixture of brown rusts at −800 mV. The mixture of the scattered brown rusts in the thin layer still could be found at −900 mV (Figure 6(b)). The whole steel surfaces are covered by obvious white layers at both −1000 and −1100 mV ( Figure  6(c,d)), indicating the formation of compact deposition layer. Unlike the CP performances in static sea water, it is further seen from the SEM images that no obvious CaCO 3 precipitation is found at −800 and −900 mV. The precipitation of the CaCO 3 crystals occurs at −1000 and −1100 mV in flowing sea water. Meanwhile, the density of the CaCO 3 crystals is much less than those formed in static sea water. It suggests that the dynamic flow could reduce the interfacial concentration of generated hydroxyl ions, thus inhibiting the surface alkalisation. Accordingly, the flow of sea water could retard the quick deposition of the calcium layer under the same CP potential in comparison with those in static sea water. In this case, the steel with more negative CP potentials would suffer faster decreases of cathodic current at the initial immersion stage, which is caused by the priority deposition of the calcium layer.  Since the deposition layers formed in flowing sea water are less compact than those formed in static sea water, small pits could be observed without the removal of the deposition layer. The small pits generated at −800 and −900 mV could be directly seen from the local SEM images (Figure 6(a,b)). It is seen from the local EDS results that the pits are filled by the iron oxide in company with the inclusions enriched in Al and S at −800 and −900 mV. Although the pits are covered by the CaCO 3 layer at −1000 and −1100 mV (Figure 6(c,d)), the inclusions enriched in Si, Al and S are still measured by EDS.
The surface morphologies of the WEs after the removal of rusts and depositions are presented in Figure 7. It is seen from the photos that the fine polished scratches could be well seen on the steel surface at all CP potentials, indicating the general corrosion could also be inhibited by CP in flowing sea water. The small pits could be observed from the local SEM images, which are similar to the pitting damage in static sea water. Since the iron oxides are mostly cleaned by the acid wash, the initiation of the small pits from the local steel inclusions enriched in Cu, S, Si, Mn and Al is clearly identified from the EDS analysis. It is further seen from the typical 3D profiles that both the diameter and depth of the pits show a significant decrease along with the CP potential changing from −800 to −900 mV. The diameters of the pits are similar at −900 to −1100 mV which range from 3.0 to 4.5 μm. The depth of the pit only shows a slight decrease with the negative shift of the CP potential from −900 to −1100 mV.

The CP performance in sand entraining sea water
The dynamic changes of the cathodic current in sand entraining flowing sea water at different CP potentials are plotted in Figure 8. The measured cathodic current of the repeated test is plotted in the Supplementary Materials. It is seen that the cathodic current of −800 mV shows an obvious negative shift from −0.325 to −0.398 mA in the first 2 h of immersion. Then, the cathodic current gradually changes to −0.459 mA at the end of the test. The variation trend of the cathodic current at −900 mV is similar to that at −800 mV. The current dramatically shifts from −0.364 to −0.487 mA in the first 2 h of immersion. Thereafter, the magnitude of the cathodic current shows a slight increase and finally fluctuates around −0.578 mA. The cathodic current of −1000 mV is relatively stable which shows a slight fluctuation at the initial 5 h of immersion. The cathodic current gradually becomes stable around −0.600 mA at the latter stage. The cathodic current of −1100 mV has a quick positive shift from −0.591 to −0.492 mA at the beginning of the test. Then, the cathodic current intensely shifts to a more negative value of −0.680 mA in the following several hours. The cathodic current of −1100 mV finally reaches −0.702 mA at the end of the test. It is found that the final cathodic currents are all negative than the initial cathodic currents in the sand entraining flowing sea water, which is totally different from the positive change of the cathodic current in static sea water and flowing sea water without sand particles.
The surface morphologies of the WEs after 24 h of CP test in sand entraining flowing sea water are shown in Figure 9. Obvious pitting damage with the local covering of brown rusts could be directly seen from the photos (Figure 9(a-d)). No obvious deposition of the calcium layer could be seen from the photos and local SEM images. Instead, the porous rusts are observed around the pits at all CP potentials. By comparing the variations of cathodic currents in static and flowing sea water without sand particles, it is found that the steel would suffer much higher cathodic currents without the protective calcium layer. The steel with more negative CP potential would suffer a higher cathodic current due to the enhanced cathodic reaction.  It is further seen from the EDS results of Figure 9(a) that the O content of Point 1 is extremely low, indicating the solid pieces at the pit boundary are the steel matrix. The fracture and peeling of the steel matrix at the pit boundary could be clearly observed. The flaking of the steel debris inside the pits is found in Figure 9(b-d) as well. It suggests that erosion occurs along with the pitting corrosion. The erosion-corrosion could not be totally suppressed by CP in the sand entraining sea water. It is seen from the EDS results of local Points 2 that the iron oxides could accumulate inside the pits. The inclusions enriched in Al and S are also found (Figure 9(a)). The calcium depositions inside of the pit are found at both −1000 and −1100 mV (Figure 9(c,d)).
The surface morphologies of the WEs after the removal of rusts and depositions are presented in Figure 10. It is seen that the whole steel surface becomes rough in the sand entraining sea water, indicating the plastic deformation of the steel surface induced by sand impingements. The local plastic deformation is further seen from the SEM images and 3D profiles (Figure 10(a-d)). It is seen from the EDS results that the pitting damages under CP are also initiated from the inclusions enriched in Mn, Si, Cu and S. The diameters and depths of the pits formed at different CP potentials are similar which are in the ranges of 12-16 μm and 14-20 μm, respectively. The pits generated in sand entraining sea water are significantly larger and deeper than the pits generated in static sea water and flowing sea water without sand particles.

Discussions
By comparing the three groups of tests, it is clearly seen that the sand impingements could lead to the local CP failure and the occurrence of erosion-corrosion in natural sea water. The depth of some obvious erosion-corrosion pits could reach 19 μm in 24 h of test, indicating a high local degradation rate. However, since sand impingements with high impact energy could result in the formation of deep craters under CP [27], the erosion-corrosion pits formed under CP might be deemed as impingement craters, especially in the slurry of high flow rates. Accordingly, the local CP failure and the synergy of erosion and corrosion under CP are normally neglected in previous studies. As the employed flow velocity is relatively low in this work, the local plastic deformation induced by pure erosion and the pitting damage induced by erosion-corrosion could be well distinguished from the local surface characterisations. The pure erosion under CP could only lead to the roughness of the steel surface when the rotation speed of the stirrer is 800 rev min −1 . The steel degradation induced by local plastic deformation is tiny compared with the other factors. The sand impingements could only work with the local anodic dissolution. It evidently shows that the coupled effect of sand impingement and local anodic dissolution is the major contributor to induce the formation of erosion-corrosion pits.
It is seen from the test results that the formation of pitting damage under CP is almost initiated from the steel inclusions enriched in Mn, S, Si, Al and Cu, where might have more negative OCP than the steel substrate [17,35].  Accordingly, once the applied CP potential is not negative enough, the supplied cathodic current could not totally inhibit the localised corrosion at the areas where more negative OCPs are registered. However, the diameters and the depths of the pits formed in static sea water, flowing sea water and sand entraining sea water are different from each other. In order to further understand the propagation of pitting damage under CP at different flow conditions, the statistical results of the pits formed at various test conditions are plotted in Figure 11. The general diameters and depths of the pits are calculated from randomly selected 30 pits on each steel sample. The error bars represent the standard deviations of the selected 30 pits. It is seen from Figure 11(a,b) that the general pit depth and diameter both show obvious decreases along with the negative shift of the CP potential from −800 mV to −900 mV in both static and flowing sea water without sand particles. It suggests that the CP potential of −800 mV cannot totally inhibit the pitting propagation in natural sea water. When the CP potential reaches −900 mV, it is seen that the general pit depth is 3.6 μm for both static and flow conditions. The pit depth generated at −1000 mV in flowing sea water is 3.4 μm which is close to that generated at −900 mV. It indicates that the pitting propagation around the inclusions would normally stop at 3-4 μm when the cathodic current is sufficient. The quick formation of the thick and compact calcium layer at lower CP potentials (−1000 mV and −1100 mV) could further retard the propagation of pitting damage. The positive shifts of the cathodic currents at both static and flow conditions are also caused by the formation of the calcium deposition layers. Accordingly, although pitting damage could still initiate on the marine carbon steel under CP, the pitting corrosion could not propagate to cause safety issues when the applied CP current is sufficient in the sea water without sand particles.
It is seen from the statistical results that both the general diameter and the depth of the pits show dramatic increases along with the introduction of sand particles into the flowing sea water. The negative shift of the CP potential only has a tiny influence on the pitting diameter and pitting depths, indicating the growing of pits in sand entraining sea water is not mainly determined by the supplemental ability of electrons. According to the test results, the effects of erosion and corrosion on the local CP failure could be qualitatively analysed. As shown in Figure 12(a), an instant crevice would form between the sand and the steel surface when the sand particle travels in the interfacial boundary layer at the initial stage. The formation of the instant crevice during mechanical wear is also reported in a recent study [36]. Although the local chemical environment could not be significantly influenced by the quick movement of the sand particle, the solution resistance in the crevice is high, leading to the instant 'cathodic shielding' beneath the sand particle. On the other hand, the frequent change of the solution resistance between the sand particle and the steel interface could simultaneously induce the local fluctuations of CP potential beneath the sand particle. The dynamic change of the local CP current might enhance the anodic dissolution at the steel inclusions [10]. Pitting damage would initiate from the inclusions due to the formation of significant local anodes. The following sand impingements at these significant anodes would result in serious local plastic deformation (Figure 12(b)) associated with the local degradation of surface hardness [37]. The local areas with obvious plastic deformation would further cause the decrease of local corrosion potential due to the high strain energy [38,39], thus enhancing the pitting propagation beneath the crevice (Figure 12(c)). The pitting corrosion would keep consciously growing under the repeated sand impingements (Figure 12(d)). The fracture and flaking of steel debris induced by repeated sand  impacts would occur at the pit boundary with the assistance of localised corrosion [18,33]. Along with the longitudinal development of the erosion-corrosion pits and the accumulation of iron oxides inside the pits, the solution resistance in the pitting areas would be high. Meanwhile, the corrosion potentials of the erosion-corrosion pits would be low due to the inclusions and high strain energy. Insufficient CP current could be supplied to inhibit the further propagation of erosion-corrosion pits ( Figure 12 (e)). Local acidification and autocatalysis might occur inside of the erosion-corrosion pits [11], possibly resulting in long-term failure of local CP. Since the calcium deposition could be easily removed by the sand impingements, the cathodic current would shift negatively in company with the surface roughening and pitting propagation.
The local CP failure induced by sand impingements leads to the pitting damage of marine carbon steel in sea water. These erosion-corrosion pits might become the nucleation sites of stress crack damage under mechanical loadings in the marine environment [40]. Although the negative shift of the CP potential could slightly mitigate the growth of erosion-corrosion pits, the hydrogen evolution at negative CP potentials would become another issue which might result in the hydrogen embrittlement at the erosion-corrosion pits, especially for high-strength carbon steels [41]. As a result, the sand impingement is a significant threat for the safety operation of marine structures and equipment in sea water. The local CP failure and pitting damage caused by erosion should be paid more attention in practical ocean engineering.

Conclusions
Although pitting damage could still occur around the steel inclusions in natural sea water free of sand particles, the pitting propagation could be inhibited when sufficient cathodic current is supplied. The steel is well protected at the CP potentials of −900 to −1100 mV (vs. Ag/AgCl) in sea water without sand particles. The formation of the calcium deposition layer could further protect the steel from pitting damage.
The sand impingements could retard the formation of calcium layer and result in local CP failure in sea water. The coupled effect of local 'cathodic shielding', CP fluctuation and plastic deformation induced by sand impingements could enhance the anodic dissolution around the inclusions, leading to the pitting propagation. The pitting growth causes the fracture and flaking of steel debris at the pit boundary after repeated sand impingements. The synergy of erosion and corrosion is the main lead to CP failure. The negative shift of the CP potential could not retard the pitting propagation. The formation of erosion-corrosion pits under CP is a potential threat to the marine structures and equipment.

Disclosure statement
No potential conflict of interest was reported by the author(s).