Atomic-scale decoration for improving the pitting corrosion resistance of austenitic stainless steels

Stainless steels are susceptible to the localized pitting corrosion that leads to a huge loss to our society. Studies in the past decades confirmed that the pitting events generally originate from the local dissolution in MnS inclusions which are more or less ubiquitous in stainless steels. Although a recent study indicated that endogenous MnCr2O4 nano-octahedra within the MnS medium give rise to local nano-galvanic cells which are responsible for the preferential dissolution of MnS, effective solutions of restraining the cells from viewpoint of electrochemistry are being tantalizingly searched. Here we report such a galvanic corrosion can be greatly resisted via bathing the steels in Cu2+-containing solutions. This chemical bath generates Cu2−δS layers on the surfaces of MnS inclusions, invalidating the nano-galvanic cells. Our study provides a low-cost approach via an atomic scale decoration to improve the pitting corrosion resistance of stainless steels in a volume-treated manner.

Stainless steels are susceptible to the localized pitting corrosion that leads to a huge loss to our society. Studies in the past decades confirmed that the pitting events generally originate from the local dissolution in MnS inclusions which are more or less ubiquitous in stainless steels. Although a recent study indicated that endogenous MnCr 2 O 4 nano-octahedra within the MnS medium give rise to local nano-galvanic cells which are responsible for the preferential dissolution of MnS, effective solutions of restraining the cells from viewpoint of electrochemistry are being tantalizingly searched. Here we report such a galvanic corrosion can be greatly resisted via bathing the steels in Cu 21 -containing solutions. This chemical bath generates Cu 22d S layers on the surfaces of MnS inclusions, invalidating the nano-galvanic cells. Our study provides a low-cost approach via an atomic scale decoration to improve the pitting corrosion resistance of stainless steels in a volume-treated manner.
S tainless steels, widely used in industry and in our daily life, are in fact not ''stainless''. In the presence of aggressive anionic species, they are susceptible to the localized pitting corrosion that is one of the major causes of the materials' failure, leading to a huge loss to our society. Therefore, to clarify the origin of pitting process and to improve the pitting resistance in stainless steels has been one of the major research focuses in several disciplines such as materials science, chemistry, and metallurgy 1-5 . Sulphur, particularly in austenitic stainless steels, plays critical roles in tuning the material's properties. For example, manganese sulphide (MnS) as a major existence of sulphur in stainless steels taking a role of lubricant is needed for machining and extending cutting tool life 6 . However, high sulphur content, and thus high volume of MnS in stainless steels, results in poor pitting resistance, since it is generally believed that the pitting events originate from the local dissolution of MnS [7][8][9][10][11] . Our recent studies 12,13 based on high-resolution transmission electron microscopy (HRTEM) indicated that the MnS inclusion is not only compositionally inhomogeneous but also structurally dissimilar. We found that novel MnCr 2 O 4 nano-octahedra embedded in the MnS medium catalyzed the local dissolution of MnS by the local nano-galvanic cells of MnCr 2 O 4 /MnS. Since then, effective solutions of restraining the cells from electrochemically working have been being tantalizingly searched.
However, invalidating the cells in a low-cost and high-efficiency way has been unattainable with current treatments performed both in industry and laboratories. Usually, final annealing and the subsequent pickling performed on the finished products are the two most common processes to enhance the localized corrosion resistance of stainless steels. During the former processes, the improvement of corrosion resistance is ascribed to the oxidation of steel surface 14 . While the principle beneficial factor of the latter process is the removal of MnS inclusions 15 . Nevertheless, the crevices and residual sulphide after removal of MnS were still identified as the sensitive sites for pitting initiation 16 . Other particular approaches, such as eliminating the MnS at the steel surface by laser or high current pulsed electron beam surface treatment 10,17,18 which have been tried in laboratories and to some extend enhanced the pitting resistance, are of high-cost and low-efficiency. Their applications are believed to be rather limited in volume-treated industrialized processes.
It is known that the composition of sulphide inclusions plays a critical role in the initiation of pitting corrosion of stainless steels 19,20 . For instance, Cu-containing MnS inclusions are found to be electrochemically inactive in initiating pitting corrosion 8 . Besides, alloying Cu could eliminate the harmful sulphur species in corrosive solution by producing insoluble Cu 2 S 21,22 . In the present study, we synthesized, in virtue of the cation exchange reaction 23-26 , a thin film of Cu 22d S on the exposure layer of MnS. Such chemical decoration leads to the inactivation of the anodic MnS inclusions and avoiding the nanogalvanic cells from electrochemically working, by which the pitting resistance of austenite stainless steels is significantly enhanced.

Results
MnCr 2 O 4 nano-octahedron catalyzing local dissolution of MnS in stainless steels. A commercial hot-rolled 316F stainless steel with high sulphur content was chosen since it provides a large number of MnS inclusions for analysis. The mechanical rolling made MnS inclusions needle-shaped and parallel to the rolling direction. Dimensions of the MnS inclusions are several tens of micrometers in length and a few hundred nanometers in diameter. MnS has a NaCl-type structure with lattice parameter of 0.52 nm. TEM observations were performed under the newly developed highangle-angular-dark-field (HAADF) technique, which provides strong contrast associated with the local variety of chemical composition (atomic number) and/or thickness contribution by detecting high-angle scattering electrons 27 . As reported in the earlier study 12 , the dissolution of MnS is strongly localized in salt water: the pit which results from the local dissolution of MnS always features a nano-scale undissolved core whose structure and composition are different from that of MnS. Namely, the initial site of MnS dissolution is at the periphery of a nano-particle which was embedded in MnS medium (Fig. 1a).
Based on a series of EDPs and energy dispersive X-ray spectroscopy (EDS), the nanoparticles are identified to be spinel MnCr 2 O 4 , or Mn(Cr,Ti) 2 O 4 if small amount of titanium is taken into account. By means of large-angle tilting experiments and 3D tomography in the TEM, we found that each of the MnCr 2 O 4 nano-particles has a specific geometric shape, which was identified basically to be an octahedron enclosed by eight triangles. The 3D morphology of an octahedron is displayed in Fig. 1b where each of the {111} planes is alternately exhibited. It is worthwhile to mention that not all the MnCr 2 O 4 nano-particles feature a regular octahedron-shaped, as that shown in our previous study. In some cases, they are octahedron but with remarkable distortion, shown in Fig. 1b. This indicates that the detailed atomic configurations at the octahedron surface are quite complicated.
The spinel MnCr 2 O 4 particles shown here, although they are as fine as nano-meter scale, are present in almost all the MnS inclusions. So far, little attention has been paid to the ultra-fine inclusions in stainless steels, which is based on the cognition that they do not undermine the mechanical properties of the steels. However, the nano-sized oxide particles shown above are found to play a critical chemical role in catalyzing MnS dissolution and pitting corrosion of the steels.
Chemical modification improving the pitting resistance. The chemical modification of MnS inclusions was performed by immersing the polished stainless steel sheets in a dilute CuSO 4 solution at room temperature. The electron-probe X-ray microanalysis (EPMA) experiment clearly reveals that the cation exchange reaction has taken place during the immersion, namely, Cu ions have selectively substituted some Mn 21 in MnS inclusions after 1 hour immersion (Fig. 2). Since the depth of characteristic X-rays produced by electrons is about one or two micrometers for the elements with medium atomic mass 28 , the whole MnS with diameter of several hundred nanometers is in the detection range of EPMA. Therefore, the co-existence of Mn and Cu signals in Fig. 2b indicates that the elements replacement should happened on the skin layer of MnS. Thanks to the passive film of stainless steels, iron has no obvious replacement reaction with copper ions since we did not find any morphology or composition change on the steel matrix. The elements substitution and the thickness of reaction layer are determined by the TEM cross-sectional observation of the treated sample which provides direct information on morphology and chemical compositions after the reaction. After the bulk sample was bathed in 0.05 mol/L CuSO 4 solution for 1 h, a HAADF image and EDS elemental mapping of the selected area are shown in Fig. 2c. It is seen that the surface layer of MnS almost all transformed to copper sulphide. The thickness of copper sulphide layer was nearly 200 nm. A sharp interface cannot be distinguished between manganese sulphide and copper sulphide since the reaction process was controlled by cation diffusion.
To evaluate the enhancement of pitting resistance after chemical treatment, the potentiodynamic polarization tests on untreated and treated samples were performed in 0.5 mol/L NaCl electrolyte. Some typical potentiodynamic polarization curves are shown in Fig. 3a. It can be seen that all the treated samples feature a noble shift of the open circuit potentials (E ocp ) and pitting potential (E pit ) and an obvious decrease of passive current density, which indicates the chemical treatment results in an improved pitting resistance. Generally, the pitting potential data of stainless steel is scattered. To evaluate the beneficial effect of the chemical treatment in a statistic analysis, we tested many samples, including untreated and treated samples under different treating conditions. The distributions of E pit values for the samples before and after different treatment were plotted as illuminated in Figs. 3b and 3c. Cumulative probability shown in the vertical was calculated by a mean rank method: P cum 5 i/(N 1 1), where P cum is the cumulative probability of measured pitting potential (E pit ), i is the order in the total number N (i 5 1, 2, 3, …, N). In Fig. 3b, it is found that the onset of stable pitting on the untreated samples occurs at near 230 mV, while for the samples treated in CuSO 4 solution with duration from 10 min to 10 h, all E pit shift to noble values by 100 mV on average. The treating duration longer than half an hour seems not to have more positive effect. It is probably because the cation exchange reaction reaches the equilibrium state in 30 minutes. We have carried out systemic electrochemical tests on the samples experienced bathing in CuSO 4 solution with variant concentrations, like 0.001 mol/L, 0.005 mol/L, 0.01 mol/L, 0.05 mol/L, and 0.5 mol/ L. The pitting potential of the samples treated in solution with different concentrations is shown in Fig. 3c. It is seen that the E pit values For comparison with other steels, the pitting potential of a commercial 316L stainless steel sample was determined which is about 320 mV (as shown in Fig. 3d). Obviously, the chemical treatment makes 316F stainless steel win an equivalent level of pitting resistance with that of 316L stainless steel, which is generally believed to have an excellent corrosion resistance.
To confirm the universality of the chemical modification, we treated the 316L stainless steel in the Cu-containing solution and also found the treatment improved the pitting resistance as that in the 316F stainless steel; for instance, the E pit has a shift of nearly 100 mV after one hour treatment. As a comparison, small pieces of 316L sample with and without visible MnS inclusions, respectively, were selected for electrochemically testing. As seen from the potentiodynamic polarization curves (Fig. 3d), the chemically modified 316L sample with MnS inclusions is very close to that of its MnSfree counterpart, which indicates that such a chemical modification make MnS inclusions inactive although the spinel nano-oxides are still there.
In addition to the CuSO 4 solution, dilute CuCl 2 solutions with various concentrations are also applied for the chemical modification of MnS, and almost the same enhancement is achieved as that in CuSO 4 (Fig. 3e).
Chemical modification substantially lowering the pitting nucleation rate. The samples undergoing different treatments were potentiodynamically polarized to the potential above E pit , and the tests were terminated when the anodic current density reached the value of 1 mA/cm 2 . Then all the samples were unloaded and examined by scanning electron microscopy (SEM). Figure 4 shows three typical SEM images of the samples suffered electrochemical tests. In the untreated sample, each MnS dissolves severely and a large number of pits appear in the vicinity (arrowed, in Fig. 4a). In contrast, in the sample treated in the Cu-containing solution for 10 minutes, the number of the pits decreases dramatically (arrowed, Fig. 4b). Onehour-treatment results in a further decrease of the number of pits (Fig. 4c). Moreover, no obvious dissolution of the MnS inclusions was observed on the surfaces of treated samples even after the stable pits appeared, which indicates that the nano-galvanic cells of MnCr 2 O 4 /MnS have been invalided by the chemical decoration.
In order to characterize the substantial decrease of the number of pits on the chemically modified samples, the number of pits per unit length of the inclusion was statistically counted, based on over 300 MnS inclusions in each sample (Fig. 5). The minimum number reaches 4, approximately, on the sample experienced one hour immersion.

Discussion
The general phenomenon of MnS dissolution is that it happens at the interface of MnS/MnCr 2 O 4 . The novel MnCr 2 O 4 nano-octahedra embedded in the MnS medium catalyzed the local dissolution of MnS by the local nano-galvanic cells between them 12,13 . Once the dissolution process starts, the whole process goes rapidly. During the chemical modification in Cu 21 -containing solution, the surface layer of MnS inclusions transforms into copper sulphide crystals, which is thought to be electrochemically inactive to initiate pitting corrosion 21 .
To figure out the details of reaction products, the TEM specimens before and after immersion in a CuSO 4 solution for 20 minutes were  Fig. 6a) were performed, as shown in Fig. 6b. It is seen that, initially, Cu is not present in the MnS inclusion. However, the composition of the same MnS inclusion (shown in Fig. 6c) exhibits a remarkable change after the chemical modification (Fig. 6d). Most of the MnS is replaced by copper sulphide, especially in the area where specimen thickness is only several tens of nanometers. The detailed chemical compositions analysis by means of EDS line scanning was performed. Fig. 7 displays one example of EDS analysis, where a MnCr 2 O 4 nano-particle is embedded in MnS medium in the HAADF image (Fig. 7a). The EDS profiles (Fig. 7b-7d) are labeled which are obtained by linescanning along the red line in Fig. 7a. The intensity of Cu-signal inside the inclusion gradually increases which suggests that the exchanged layer becomes thicker, with extending the immersion duration from 0, to 10 minutes and 20 minutes. Both steel matrix and the MnCr 2 O 4 particle inside the MnS inclusion were noticed to remain unchanged during the process of chemical modification.
Determined by HRTEM images along [001] and 110 ½ directions ( Fig. 8a and 8b, respectively), the final products has an isotypic  fluorite structure with lattice constant a 5 0.54 nm. A combination of lattice parameter and chemical compositions concludes that such a compound corresponds to Cu 22d S,namely digenite, of a stoichiometry approaching to Cu 1.8 S 29 . In such crystals, cations occupy the tetrahedral 8(c) sites and trigonal 32(f) sites randomly at room temperature. The crystal cells of cubic MnS and Cu 22d S are displayed in Fig. 8c.
The final product Cu 22d S indicates that both Cu 1 and Cu 21 are involved in the cation exchange process. The Cu 1 might be derived from the redox reaction between Cu 21 in CuSO 4 eletrolyte and steel matrix. Although corrosion of stainless steels in dilute CuSO 4 solution develops rather slowly 30 , it provides enough cuprous ions for the occurrence of cation exchange reaction at the free surfaces of MnS inclusions, since the area of the MnS inclusions is rather smaller compared with that of the whole steel surface. We propose the reaction pathway for the formation of Cu 1 be written as follows: CuzCu 2z <Cu z ð2Þ Since cubic Cu 22d S and MnS share the same sub-lattice of S 22 anions and lattice constant of these two compounds are quite close, this enables the exchange between Cu 1 and Mn 21 through the frame of S 22 anions without rearrangement of S 22 lattices (see Fig. 8c). In this reaction, the large difference of solubility (solubility product K MnS 5 10 213 , K Cu2S 5 10 248 ) provides the thermodynamic driving force for the spontaneous ions replacement 26 . The ideal reaction is demonstrated as follows: In the present study, the increase in pitting potential was attributed to the newly formed Cu 22d S thin film which prevents the MnS from dissolution. In the galvanic system of MnS/MnCr 2 O 4 , the potential difference between MnS dissolution (E MnS ) and oxygen reduction reaction (E ORR ) provides the thermodynamic driving force. Anode MnS tends to dissolve, while ORR takes place on the cathode MnCr 2 O 4 . Once the cation exchange occurs, the exposure layers of MnS are replaced by Cu 22d S which make the MnS be insulated from the solution. According to the E-pH diagram 7,22 , the equilibrium potential of Cu 22d S is much higher than that of MnS in a neutral environment and Cu 22d S can stably preserve in a wide range near E ORR , Therefore, the nano-galvanic cells was effectively invalided and the pitting potential of the alloy was enhanced by such a chemical decoration. Ever since the MnS inclusion was identified as the initiation site for pitting corrosion, several methods have been tried in engineering to remove surface MnS inclusions, such as pickling, laser and high cur-   rent beam modification. These approaches are useful but have considerable deficiencies. In contrast, the chemical modification via cation exchange reaction in the present study exhibits a much more feasible methodology to improve pitting corrosion resistance of stainless steels, since it is easy to operate for variant dimensions of components. Hence, this study provides a low-cost approach via an atomic scale decoration to improve the corrosion resistance of steels in a volumetreated manner, and has a great signification in improving the integrated properties of stainless steels serving in wet environments.

Methods
Bulk sample preparation. Commercial 316F type stainless steel with high-sulphur content (0.16%) was chosen because it can provide a large number of MnS inclusions for analysis. The steel was made by Nippon Steel and Sumikin Stainless Steel Corporation. 316L type stainless steel was made by Baosteel Corporation in Shanghai, China.
The hot-rolled steel rod was cut into sheets with the same size along the rolling direction. The samples were carefully polished with standard SiC grinding papers. The samples were then lightly polished using 1 mm diamond paste until there was no visible scratch on the surfaces. After exposure in air for 12 hours, the samples were immersed in a Cu-containing solution at room temperature for different periods. Those treated samples were cleaned by distilled water and dried for the subsequent electrochemical tests and other structural characterization (SEM, EPMA).
TEM sample preparation. The samples cut from the steel rod were ground using variant grit silicon carbide papers, polished with diamond paste, and finally thinned by ion-milling. After the first-round TEM observation, some of the specimens were plasma-cleaned and then immersed into 1 mol/L NaCl solution for immersioncorrosion test. The TEM specimens after immersion were quickly cleaned in distilled water and methanol, dried, and transferred into the TEM for further investigation.
The plan-view TEM samples for analyzing the structural and chemical information of Cu 22d S were prepared by ion-milling and treated in 0.05 mol/L CuSO 4 solution at room temperature after first-round TEM observation.
The cross-sectional sample, used for determining the thickness of reaction layer, was prepared by the conventional methods. Two pieces of 316F stainless steel samples with polished surfaces were immersed in 0.05 mol/L CuSO 4 solution for 1 hour. The treated surfaces of two samples were bonded face-to-face and then thinned by grinding, dimpling and ion-milling.
TEM, SEM and EPMA characterizations. A Tecnai G 2 F30 transmission electron microscope, equipped with HAADF detector and X-ray dispersive spectrometer systems, was used at 300 kV for diffraction, HAADF imaging, composition analysis, and 3D tomography. Microstructure characterizations of Cu 22d S were performed using a Titan G 2 60-300 TEM equipped with double spherical aberration (Cs) correctors. A SUPRA35 field emission scanning electron microscope was used to investigate the morphology of MnS inclusions. The surface composition of the bulk materials was analyzed using an electron probe X-ray microanalyzer (EPMA: Shimazu Seisakusho EPMA-1610).
Electrochemical test procedures. A traditional three-electrode system was used in electrochemical experiment. The working electrode is stainless steel specimen, Pt counter electrode and SCE (saturated with KCl) reference electrode. AUTOLAB PGSTAT302N electrochemical workstation was used in potentiodynamic polarization curve measurements with a scan rate 1 mV/s. All the tests were stopped when the www.nature.com/scientificreports anodic current density reached the value of 1 mA/cm 2 . The tested solution of 0.5 mol/L NaCl electrolytes was maintained at 28uC with electric-heated thermostatic water bath.
The potentiodynamic polarization curve includes the cathodic and anodic branches. In the anodic branch, there exists a well-defined inflexion, above which anodic current density sharply increase with potential, meaning stable pitting corrosion occurs. The potential value corresponding to that inflexion is identified as pitting potential (E pit ).