Nano Structure of the Rust Formed on an Iron-based Shape Memory Alloy (Fe Mn Si Cr) in a High Chloride Environment

Transcription

1 , pp Nano Structure of the Rust Formed on an Iron-based Shape Memory Alloy (Fe Mn Si Cr) in a High Chloride Environment Toshiyasu NISHIMURA* National Institute for Materials Science (NIMS), Sengen, Tsukuba-city, Ibaraki, Japan. (Received on April 15, 2014; accepted on May 18, 2014) The corrosion resistance of an Iron-based shape memory alloy (28 Mn: Fe Mn Si Cr) was estimated by using wet and dry corrosion test. The structure of the rust formed on the alloy was examined via EELS (Electron Energy Loss Spectroscopy) using TEM (Transmission Electron Microscopy) analysis. The electrochemical behavior of the 28 Mn alloy was investigated via EIS (Electrochemical Impedance Spectroscopy). The 28 Mn alloy showed a much higher corrosion resistance than SM (carbon steel) in wet and dry corrosion test. In EIS measurement following the corrosion tests, the 28 Mn alloy exhibited a much larger value of corrosion resistance (Z 1mHz) as compared to SM, which can be attributed to rust formation. TEM analysis indicated that the inner rust was composed of Mn-rich, Cr-Si rich and interface layer. The corrosion resistance was primarily dependent on the Cr-Si rich layer, and further supported by the Mn rich layer. It was found that the 28 Mn alloy could maintain a protective rust layer composed of effective elements in the wet and dry environment. KEY WORDS: shape memory alloy; corrosion; rust; transmission electron microscopy; electrochemical impedance spectroscopy. 1. Introduction The Shape memory effect has been observed in several alloy systems, including Au Cd, Ti Ni, Cu Zn Al. 1 3) Ironbased alloys have attracted attention, especially because of reduced material cost. Recently, an Fe Mn Si based shape memory alloy has been reported by Sato et al., which possessed the excellent mechanical properties. 4,5) However, there are very few papers concerning the corrosion resistance of this type of iron-based shape memory alloys. 6) Fukai et al. investigated the oxidation resistance of an Fe Mn Si Cr based shape memory alloy via EPMA and XPS, and found a large effect of Cr on the oxidation resistance. 7) In order to identify the atmospheric corrosion resistance of this alloy, it is important to conduct the wet and dry corrosion test containing chloride solution. 8) Generally, this type of alloy produces rust in an atmospheric environment. There are many reports for weathering steels, 9 13) however, there are few papers investigating the structure of the rust formed on shape memory alloys. In this study, as the rust formed as nano-sized oxides, the nano structure of the rust was investigated using TEM. The inner rust formed in the wet and dry test was cut using FIB (Focused Ion Beam), and the chemical state of each element was investigated via EELS. 14) EIS of samples was employed following the wet and dry test in order to investigate the electrochemical behavior of the Fe Mn Si Cr based shape memory alloy. 8) * Corresponding author: NISHIMURA.Toshiyasu@nims.go.jp DOI: 2. Experimental 2.1. Samples, and Wet and Dry Cyclic Corrosion Tests The sample material (28 Mn alloy) was an Fe-28%Mn- 5%Cr-5%Si (mass%) shape memory alloy. Carbon steel (SM) was used for comparison. The corrosion test consisted of 1) wetting the sample surfaces with 0.4 L/m 2 of a 0.5 mass% NaCl solution, and 2) drying the sample in a chamber maintained at 25 C, and 60% RH for 12 hours. 8) Immediately before wetting the sample with solution, the sample was washed in distilled water and dried to prevent progressive salt accumulation Physical Analysis of the Rust on 28 Mn Shape Memory Alloy SEM (Scanning Electron Microscopy) and TEM were employed for the rust analysis. After the sample was mounted in resin, it was mirror-polished using emery paper and diamond paste. Carbon was then evaporated onto the sample in order to compensate for charging effects. A cross section of the rusted sample was examined via EDXS (Energy Dispersive X-ray Spectroscopy) analysis using SEM. The distribution of Mn, Cr and Si in the rust was measured to identify the location and existence of these elements. In order to investigate the rust at the nano level, TEM was employed. The rust of 28 Mn alloy was cut by using FIB from above the sample using SEM. A cross section of the rusted sample was examined via EDXS-TEM analysis with a beam width of 0.2 nm over an area of 5 8 nm. The EELS-TEM analysis was carried out in order to examine the chemical state of the Fe, Mn, Cr, Si and Oxygen in the rust ISIJ

2 of the 28 Mn alloy EIS Measurement After the corrosion test, electrodes were fabricated out of the 28 Mn alloy. Then, EIS measurements were performed in 0.1 M Na 2SO 4 solution using a 2-electrode system with a measurement frequency range of 20 khz to 1.0 mhz and an applied voltage of 10 mv. 3. Results and Discussion 3.1. Corrosion Resistance of 28 Mn Shape Memory Alloy The corrosion resistance of the Fe Mn Si Cr shape memory alloy was estimated by the wet and dry corrosion test for 4 weeks. Figure 1 shows the corrosion test results for the 28 Mn alloy and SM. The extent of corrosion was Fig. 1. Corrosion of 28 Mn alloy and carbon steel (SM) after wet and dry corrosion test using 0.5 mass% NaCl. determined by the weight loss of samples following removal of the rust. The amount of corrosion of SM increases with time, and is not saturated at 28 days. On the other hand, the amount of corrosion of the 28 Mn alloy is much less than that of SM after 28 days. Thus, the 28 Mn alloy exhibits excellent corrosion resistance as compared to SM in the corrosion tests. Figure 2 shows the results of EIS measurement for the 28 Mn alloy after 28 days of the corrosion test. Two resistance and one capacitance components can be recognised in the spectrum. The impedance in the higher frequency region (Z h) is considered to be related to the impedance of the rust (Z rust). As the value of Z rust is low compared to the oxides, Z rust is assumed to be the resistance of the solution in the pores in the rust. On the other hand, the impedance in the lower frequency region (Z l) is considered to be the corrosion reaction impedance (Z ct). Figure 3 shows the results of EIS measurement for the samples after the corrosion test, showing Z rust and Z ct against the test time (in days). Here, Z rust and Z ct were measured from the impedance at 1 khz and 1 mhz, respectively, in Fig. 2. The 28 Mn alloy shows a higher value of Z rust than SM over the entire testing period. Since Z rust is assumed to be the resistance of the solution in the pores in the rust, this fact shows that the rust of 28 Mn alloy has a finer structure than that of SM. Moreover, the 28 Mn alloy shows much higher value of Z ct than SM over the entire testing period. Since Z ct is assumed to be the corrosion resistance, this fact shows that the 28 Mn alloy has a higher corrosion resistance than SM. Especially, after 28 days, the 28 Mn alloy has much higher Z rust and Z ct, values, and thus, much higher corrosion resistance than SM. Fig. 2. Impedance spectra of 28 Mn alloys after corrosion test. Fig. 3. Impedance values of Z rust at 1 khz and Z ct at 1 mhz for 28 Mn alloy and Carbon steel (SM) after corrosion test ISIJ 1914

3 3.2. Surface Analysis of the Rust on 28 Mn Shape Memory Alloy In order to identify the corrosion preventing mechanism of the 28 Mn alloy, surface analysis was undertaken. Figure 4 shows the distribution of the elements for Fe, Mn, Si and Cr in the rust obtained by SEM-EDXS after 28 days of the test. It is clear that Mn, Si and Cr are enriched in the inner rust. In particular, Si is strongly enriched at the interface between the rust and the alloy. These elements are believed to form a complex oxide with iron during the corrosion test. Thus, in order to examine it in more detail, the inner rust was cut by using an FIB at the position as shown in the SEM image in Fig. 4. A TEM observation was conducted to investigate the nanostructure of the inner rust. The inner rust was cut by using FIB as shown in Fig. 5. From EDXS analysis, the inner rust contains Mn, Si, Cr and Fe. In fact, the inner rust consists primarily of two layers: an Mn-rich and a Cr-Si rich Fig. 4. SEM-EDXS analysis for the rust of 28 Mn alloy after corrosion test for 4 weeks. Fig. 5. TEM-EDXS analysis for the rust of 28 Mn alloy ISIJ

4 layer. The rust at the interface of the rust and the alloy is the Cr-Si rich layer, covered by the Mn rich layer upper on this layer. A crack is visible in the Cr-Si rich layer, which was produced during the preparation of the sample. The chemical state in each element in the rust was identified by using EELS. The EELS observation positions are shown in Fig. 6. The Mn rich and Cr-Si rich layers are again visible. The Cr-Si rich layer appears to be separated into two layers. At the interface between the rust and the alloy, the rust is particulate in shape. Therefore, the rust at the interface was examined separately from the rust of the Cr-Si rich layer. In addition, at position 1 and 2 (P1 and P2), only Mn is enriched in the layer. However, it is recognized in Fig. 5 that at P8 and P9, the Mn-rich layer contains a little Cr and Si. Thus, P8 and P9 are additionally estimated in this study. Metal EELS spectra for 28 Mn alloy are shown in Fig. 7. These EELS spectra are classified into 3 groups: P (1, 2) in a Mn-rich, P (3, 4) in a Cr-Si rich, and P (5, 6) in the interface layer. In the case of P7, which is in the alloy, the spectrum has a peak at 58 ev, which is indicative of Fe. The spectra of P (1, 2) in the Mn-rich layer have a sharp peak at 52 ev, indicative of Mn. Similarly, the spectra of P (3, 4) in the Cr-Si rich and P (5, 6) in the interface layer have 2 peaks at 52 and 58 ev, for Mn and Fe. Additionally, they have peaks at 45 and 54 ev for Cr and Si. In the case of P (8, 9) which contains Si and Cr in the Mn-rich layer, the spectrum is different. The spectrum of P8 has 2 strong peaks at 52 and 58 ev, showing the existence of both Fe and Mn. On the other hand, the spectrum of P9 has a single strong peak at 58 ev, for Fe. Thus, the spectrum of P8 and P9 differ with respect to identity of the primary element. Oxygen EELS spectra for the rust of 28 Mn alloy are shown in Fig. 8. These EELS spectra are also classified into 3 groups: P (1, 2) in a Mn-rich, P (3, 4) in a Cr-Si rich, and P (5, 6) in the interface layer. The spectra of P (1, 2) in the Mn-rich layer have peaks at 534 and 545 ev, which are thought to show the existence of primarily Mn. Similarly, the spectra of P (3, 4) in the Cr-Si rich layer and P (5, 6) in the interface layer have 3 peaks at 531, 538 and 578 ev, which are thought to show the existence of Cr and Si. In the case of P (8, 9) which contains Si and Cr in the Mn-rich lay- Fig. 6. TEM-EELS observation points for the rust of 28 Mn alloy. Fig. 8. TEM-EELS spectra of Oxygen for the rust of 28 Mn alloy. Fig. 7. TEM-EELS spectra of metals for the rust of 28 Mn alloy. Fig. 9. TEM-EELS spectra of Mn-L for the rust of 28 Mn alloy ISIJ 1916

5 er, the peaks are the same as those of a Cr-Si rich layer. Thus, the spectra of P8 and P9 are thought to show the presence of Cr and Si. As the EELS spectra of Mn-L in Fig. 9 are somewhat complicated, those of standard Mn oxides were investigated for comparison in Fig. 10. The chemicals are Mn, MnO, Mn 3O 4, Mn 2O 3 and MnO 2. The spectra show 2 peaks of Mn- L 2 and Mn-L 3 which are shifted according to the valence of the Mn. Besides, the peak ratio (Mn-L 3/Mn-L 2) varies. In Fig. 11 showing the parameters, the spectrum of metallic Mn (0 valance) is defined as the standard. S-L 2 and S-L 3 are the chemical shift of Mn-L 2 and L 3 relative to the peaks of metallic Mn. (L 3/L 2) is the peak ratio of Mn-L 3 against L 2. S-L 2 and S-L 3 increase as the valence of Mn in the oxides increases from 0 to 2. Specially, S-L 3 increases suddenly between valence 2 and 3 in Mn oxides. It is typical that (L 3/ L 2) shows a maximum value of more than 4. Laffont reported very similar EELS spectra for Mn oxides. 15) The EELS observation position of 7 (P7) is the result of the metallic alloy. S-L 2 and S-L 3 (ev) at each position are measured as compared to P7. Besides, the peak ratios of (L 3/ L 2) are indicated for the spectra at each position. The S-L 2 at P3 in the Cr-Si rich layer shows high value of 2, and lower value of (L 3/L 2) of 2.9. Thus, the valence of Mn at P4 is thought to be 3, which implies that Mn exists as Mn 3+ oxide state in the Cr-Si rich layer. In the same way, P5 and P6 in the interface show a high value of 2, and a lower value of (L 3/L 2) as 2.5 is observed. Thus, the valence of Mn in the interface is thought to be 3 showing Mn 3+ oxide state. On the other hand, the S-L 2 of P1 and P2 in the Mn-rich layer has a low value of 0.5, and higher value of (L 3/L 2) more than 5. Thus, the valence of Mn at P1 and P2 is thought to be 2. The S-L 2 at P8 and P9 containing Cr and Si in the Mn-rich layer show low value of 0.5, and higher value of (L 3/L 2) more than 4.6. Thus, the valence of Mn at P8 and P9 is also thought to be 2. Thus, the date implies that at P1, P2, P8 and P9, Mn exists as Mn 2+ oxide state in a Mnrich layer. The Cr-L EELS spectra in the rust of the 28 Mn alloy are shown in Fig. 12. There are 2 peaks representing Cr-L 2 and Cr-L 3 in the spectra of Cr-L. They are shifted and show several (Cr-L 3/Cr-L 2) ratios, which change according to the valence of the Cr oxides. Table 2 shows the parameters for Cr-L spectra at P1 P9 for the rust of 28 Mn alloy. The parameters of Cr-L spectra are also indicated for the Fig. 10. TEM-EELS spectra of Mn-L for the standard chemicals of Mn oxides. Table 1. Parameters of Mn-L spectra at S1 S9 for the rust of 28 Mn alloy. 1) S-L 2, S-L 3: Chemical shift (ev) of Mn-L 2 and L 3 against the peak of alloy (0 valence Mn). 2) (L 3/L 2): Peak ratio of Mn-L 3 against L 2. layer Mn-rich Cr-Si rich Interface alloy Position S-L S-L L 3/L Table 2. Parameters of Cr-L spectra at S1 S9 for the rust of 28 Mn alloy. 1) S-L 2, S-L 3: Chemical shift (ev) of Cr-L 2 and L 3 against the peak of metallic alloy (0 vallent Cr). 2) (L 3/ L 2) : Peak ratio of Cr-L 3 against L 2. 3) Chemicals: standard chemical of Cr oxides of 0 4 valence. layer Mn rich Cr-Si rich Interface alloy Position S-L 3 x x S-L 2 x x L 3/L 2 x x Fig. 11. Parameters of Mn-L spectra for the standard chemicals of Mn oxides. S-L 2, S-L 3: Chemical shift of Mn-L 2 and L 3 against the peak of 0 valence Mn. (L 3/L 2) : Peak ratio of Mn-L 3 against L 2. Chemicals Cr(0) Cr(2) Cr(3) Cr(4) S-L S-L L 3/L ISIJ

6 Fig. 12. TEM-EELS spectra of Cr-L for the rust of 28 Mn alloy. Fig. 13. TEM-EELS spectra of Si-L for the rust of 28 Mn alloy. standard Cr oxides (0 4 valence). If the spectrum of metallic Cr is used as the standard, then, the chemical shift of S-L 3 and S-L 2 are 0 in the case of 2 valence Cr-L. On the other hand, the S-L 3 and S-L 2 are 2 and 1, respectively, in the case of 3 valence Cr-L. Thus, a Cr valence of 2 or 3 is clearly distinguished via this comparison. Daulton et al. summarized the EELS spectra of Cr oxides, 16) presenting date which is very similar to the previous. In analyzing the date from the rust, P7 is used as the standard. S-L 2 and S-L 3 are the chemical shift (ev) of Cr-L 2 and L 3 against the metallic alloy at P7. (L 3/L 2) is the peak ratio of Cr-L 3 vs L 2. The Cr-L EELS spectra are similarly classified into 3 groups: P (1, 2) in the Mn-rich layer, P (3, 4) in the Cr-Si rich, and P (5, 6) in the interface layer. First, it is found that there is no peak for Cr-L at P (1, 2). This is because there is no Cr in an Mn-rich layer. In the spectra at P (3, 4) in the Cr-Si rich layer, the S-L 3 and S-L 2 are 2 and 1, which shows that Cr is valence 3. Similarly, in the spectra of P (5, 6) in the interface, the S-L 3 and S-L 2 are 2, which shows Cr is valence 3. Thus, the valence of Cr in both the Cr-Si rich layer and the interface is thought to be 3, which implies that Cr exists as Cr 3+ oxide state. Although there is no peak for Cr-L at P (1, 2), the spectra are shown for P (8, 9) which contain a small amount of Cr and Si with Mn. As the S-L 3 and S-L 2 are less than 1, the valence of Cr in these spots is thought to be 2, which implies that Cr exists as Cr 2+ oxide state. The Si-L EELS spectra of the rust of the 28 Mn alloy are shown in Fig. 13. The Si-L EELS spectra are again classified into 3 groups. Firstly, there is no signal for Si-L at P (1, 2) in the Mn-rich layer because there is no Si in the Mn-rich layer. The spectra at P (3, 4) in the Cr-Si rich layer and at P (5, 6) in the interface have the same shape 2 peaks at 108 and 116 ev. In the previous paper, 14,17) EELS spectra of Si oxides were detected including standard chemicals of Si and SiO 2. The above spectra are similar to that of SiO 2. Thus, Si is thought to exist in Si 4+ oxide state in the Cr-Si rich layer and the interface. At P (8, 9) which contains Si and Cr in the Mn-rich layer, although the spectra are week, they have Fig. 14. TEM-EELS spectra of Fe for the rust of 28 Mn alloy. 2 peaks at 108 and 116 ev. Thus, the Si at P (8, 9) exists as Si 4+ oxide state in the Mn-rich layer. Similarly, the Fe-L EELS spectra for the rust of the 28 Mn alloy are shown in Fig. 14, and are classified into 3 groups. The standard spectrum of the metallic alloy of 0 valence is shown in P7 which has an Fe-L 3 at 708 and L 2 at 721 ev. There is a weak signal of Fe-L at P (1, 2) in the Mn-rich layer. This is because there is less Fe in the Mn-rich layer. All of the Fe-L 3 and L 2 at P (3, 4) in the Cr-Si rich and at P (5, 6) in the interface layer are shifted 1 or 2 ev as compared to those of alloy at P7. Thus, the Fe oxide in these layers is of several valances. Tan et al. examined the EELS spectra of Fe oxides including Fe(II) O and Fe 2O 3. 18) The spectra of the results are similar to those for Fe(II) O and Fe 2O 3. In fact, the spectra of those results show a shape intermediate between Fe(II) O and Fe 2O 3. This is because both Fe 2+ and Fe 3+ exist in one spectrum. Thus, Fe is thought to exist as Fe 2+ and Fe 3+ oxide state in the Cr-Si rich and in the inter ISIJ 1918

7 face layer. For P (8, 9), which contains Si and Cr in a Mnrich layer, there are 2 strong peaks although the spectra at P (1, 2) are week. Thus the Fe at P (8, 9) containing Si and Cr in a Mn-rich layer exist as Fe 2+ and Fe 3+ oxide state Nano Structure of the Rust of 28 Mn Shape Memory Alloy In the wet and dry corrosion test, the 28 Mn alloy showed a much higher corrosion resistance than SM. Moreover, in EIS measurements, the 28 Mn alloy had much larger value of Z 1mHz than SM after corrosion test. In the case of conventional weathering steel (0.6Cr-0.4Cu-Fe), the corrosion ratio relative to SM was 90% after 4 weeks of the test. On the other hand, the 28 Mn alloy showed a corrosion ratio of 25% in Fig. 1. Thus it is possible to say that the 28 Mn alloy possesses the high corrosion resistance as the weathering steel. The primary reason for this is considered that a highly corrosion - resistant rust is formed on the alloy during the wet and dry test, which can maintain the high corrosion resistance. TEM-EELS measurements were conducted to identify the nano structure of the rust on 28 Mn alloy. Figure 15 is a schematic diagram which shows the results of the TEM measurements. The structure of the inner rust is classified into 3 groups: an Mn-rich layer, a Cr-Si rich layer, and an interface layer. The chemical state of each element varies in each layer. The chemical states are supposed to be Cr 3+, Si 4+, and Mn 3+ oxide state in the Cr-Si rich and interface layers. In the Mn rich layer, Mn 2+ in Mn only layer is likely. Additionally, (Cr 2+, Si 4+ and Mn 2+ ) are predicted in the Mn rich layer with Cr and Si. Thus, for the inner layer, the rust is thought to contain mainly Cr 3+ and Si 4+ in Fe oxides. On the other hand, on that Fig. 15. Schematic diagram of the results of TEM-EELS spectra of the rust of 28 Mn alloy. layer, the rust is thought to contain mainly Mn 2+ in Fe oxides. As the corrosion resistance of the 28 Mn alloy is thought to depend on the inner layer of the rust, Cr and Si are identified as the effective elements. Moreover, Mn is thought to support the rust by forming a layer enriched in Mn. In this way, the 28 Mn alloy maintains a protective rust layer composed of effective elements in the wet and dry environment. The corrosion resistance mainly depends on a Cr-Si rich layer supported by a Mn rich layer. Thus, the 28 Mn alloy has high corrosion resistance in an atmosphere containing Chloride ions. 4. Conclusions An iron-based (Fe Mn Si Cr) shape memory alloy (28 Mn alloy) showed much higher corrosion resistance than carbon steel (SM) in the wet and dry corrosion test. In EIS measurement, the 28 Mn alloy had a much larger value of corrosion resistance (Z 1mHz) as compared to SM due to the formation of protective rust after corrosion test. In TEM, the inner rust was composed of an Mn rich layer, a Cr-Si rich layer, and the interface layer. The corrosion resistance was mainly depended on the Cr-Si rich layer, which was supported by an Mn rich layer. It was found that the 28 Mn alloy maintained a protective rust layer composed of effective elements in the wet and dry environment. Acknowledgement The author thanks Mr T. Maruyama and Y. Chiba who prepared the sample and Dr T. Sawaguchi and Dr K. Tsuzaki for advices concerning the shape memory ally used in this study. REFERENCES 1) L. C. Chang and T. A. Read: Trans. Am. Inst. Min. Metall. Pet. Eng., 189 (1951), 47. 2) S. Miyazaki and K. Otsuka: ISIJ Int., 29 (1989), ) K. Otsuka, H. Sakamoto and K. Shimizu: Acta Metall., 27 (1979), ) A. Sato, E. Chishima, K. Soma and T. Mori: Acta Metall., 30 (1982), ) A. Sato and T. Mori: Sci. Eng., A 146 (1991), ) H. Otsuka, H. Yamada, T. Maruyama, H. Tanahashi, S. Matsuda and M. Murakami: ISIJ Int., 30 (1990), ) H. Fukai, S. Suzuki, N. Masashi, S. Hanada, T. Maruyama, H. Kubo and Y. Waseda: Mater. Trans., 46 (2005), ) T. Nishimura, H. Katayama, K. Noda and T. Kodama: Corrosion, 56 (2000), ) T. Misawa, K. Asami, K. Hashimoto and S. Shimodaira: Corros. Sci., 14 (1974), ) I. Suzuki, Y. Hisamatsu and N. Masuko: J. Electrochem. Soc., 127 (1980), ) M. Stratmann, K. Bohnenkamp and T. Ramchandran: Corros. Sci., 27 (1987), ) J. Dunnwald and A. Otto: Corros. Sci., 29 (1989), ) H. E. Townsend: Corrosion, 57 (2001), ) T. Nishimura: Corros. Sci., 52 (2010), ) L. Laffont and P. Gibot: Mater. Charact., 61 (2010), ) T. L. Daulton and B. J. Little: Ultramicroscopy, 106 (2006), ) T. Nishimura: Mater. Trans., 48 (2007), ) H. Tan, J. Verbeeck, A. Abakumov and G. Van Tendeloo: Ultramicroscopy, 116 (2012), ISIJ