- Email: [email protected]
Corrosion initiation of stainless steel in HCl solution studied using electrochemical noise and in-situ atomic force microscope
Electrochimica Acta 54 (2009) 7134–7140
Contents lists available at ScienceDirect
Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta
Corrosion initiation of stainless steel in HCl solution studied using electrochemical noise and in-situ atomic force microscope Yan Li, Ronggang Hu, Jingrun Wang, Yongxia Huang, Chang-Jian Lin ∗,1 State Key Laboratory of Physical Chemistry of Solid Surfaces, Department of Chemistry, College of Chemistry and Chemical Engineering, Xiamen University, Xiamen, 361005, China
a r t i c l e
i n f o
Article history: Received 30 March 2009 Received in revised form 15 July 2009 Accepted 15 July 2009 Available online 23 July 2009 Keywords: Corrosion initiation Stainless steel In-situ AFM ECN
a b s t r a c t An in-situ atomic force microscope (AFM), optical microscope and electrochemical noise (ECN) techniques were applied to the investigation of corrosion initiations in an early stage of 1Cr18Ni9Ti stainless steel immersed in 0.5 M HCl solution. The electrochemical current noise data has been analyzed using discrete wavelet transform (DWT). For the first time, the origin of wavelet coefficients is discussed based on the correlation between the evolution of the energy distribution plot (EDP) of wavelet coefficients and topographic changes. It is found that the occurrence of initiation of metastable pitting at susceptive sites is resulted from the reductive breakdown of passive film of stainless steel in the diluted HCL solution. The coefficients d4 –d6 are originated from metastable pitting, d7 represents the formation and growth of stable pitting while d8 corresponds to the general corrosion. © 2009 Elsevier Ltd. All rights reserved.
1. Introduction Stainless steel is a versatile material that is frequently utilized in various corrosive environments. The corrosion initiation process of stainless steel has aroused a great deal of attention in the last few decades because the corrosion initiation is very complicated and it may cause various types of localized corrosions. Stewart and Williams conducted a detailed analysis of the influence of MnS on pitting initiation process, and confirmed that the inclusions dominated as pitting nucleation sites [1]. Renner et al. employed AFM to observe the topographical characteristics of initial corrosion on the atomic scale under different potentials [2]. Wang and Li used electronic speckle pattern interferometry (ESPI) to study the initial corrosion of stainless steel [3]. Their previous studies mainly focused on the mechanism of initial pitting. However, little work about the corrosion initiation of stainless steel in a dilute HCl solution can be found in the literature. The electrochemical noise (ECN) technique has been developed as a powerful technique in corrosion studies. It has been extensively used for the in-situ detection of spontaneous changes in localized corrosion processes, in particular pitting, cavitation attack, stress corrosion cracking or corrosion under coatings. This is made possible by measuring the fluctuations of the current and potential simultaneously, which are generated during the corrosion process
∗ Corresponding author. Tel.: +86 592 2189354; fax: +86 592 2189354. E-mail address: [email protected] (C.-J. Lin). 1 ISE member. 0013-4686/$ – see front matter © 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2009.07.042
[4–8]. Electrochemical noise analysis has also been widely used for the in-situ characterization and monitoring of corrosion processes of various systems, and it offers some advantages over the common electrochemical techniques. They are [9]: (1) very fast measurements enable the user to register and monitor the corrosion process instantaneously; (2) sensitivity to the early initiation process of localized corrosion, unlike conventional techniques that provide little information [10]; (3) possibility to be used without disturbing the system under investigation [11]; (4) a potential in-situ tool to monitor the corrosion stages of constructional engineering. The sensitivity of ECN measurement to localized corrosion is much higher than that of the traditional techniques. However, the data analysis of ECN remains difficult. Many methods have been developed to analyze the ECN data, including statistical analysis [12–15] and spectral analysis [16–20]. The most commonly used ECN analysis methods (statistical and spectral) are devised for stationary signals that do not show distinctive transients. The main disadvantage of those methods is that they analyze signals by averaging the features across the whole time record. Transient information may be lost in this analysis process. Wavelet transform (WT), especially discrete wavelet transform (DWT), has been proposed as an alternative tool to overcome the limitations of traditional methods in the analysis of non-stationary signals of ECN data [21–23]. WT or DWT can provide information about transients in both time and frequency domain. Moreover, it is possible to work with non-stationary signals, so it has been used to differentiate corrosion types and to study corrosion mechanism. Using the wavelet transform method, the ECN data can be decomposed to different wavelet coefficients with corresponding time
Y. Li et al. / Electrochimica Acta 54 (2009) 7134–7140
7135
Table 1 The component of 1Cr18Ni9Ti stainless steel. C (≤(%))
Si (≤(%))
Mn (≤(%))
P (≤(%))
S (≤(%))
Ti (≤(%))
Cr
Ni
Fe
0.12
1.00
2.00
0.035
0.30
0.98
17.00–19.00
8.00–11.00
Balance
scales [20–22], which are associated with different electrochemical transient events. Aballe et al. introduced the energy distribution plot (EDP) to monitor the fluctuation of transient events [21,22]. Qiao and Ou and Zhao et al., respectively employed EDP to analyze the ECN of Q235 carbon steel in cement mortar [23,24]. However, the corresponding corrosion event of each wavelet coefficient is still unclear. In this work, the corrosion initiation of 1Cr18Ni9Ti stainless steel in dilute HCl solution was studied using an in-situ AFM, optical microscope and electrochemical noise. An attempt was made to verify the origin of electrochemical current noise using the correlation between the corrosion morphologies and the wavelet coefficients.
a digital microscope (KEYENCE, VHX-600E), which achieved a field depth at least 20 times greater than that of traditional optical microscopes. The scanning electron microscope (SEM) and the wavelength dispersive spectrometer (WDS) were executed by the electron probe micro-analyzer (Model JXA-8100, JEOL, Japan). 3. Results and discussion Fig. 1 shows the topographical changes of 1Cr18Ni9Ti stainless steel in 0.5 M HCl solution over different immersion times. Due to the fatigue effect of the Piezo scanner, these continuous in-situ AFM images shift slightly. To reduce the influence of the shift, part of an image, measuring 10 m × 10 m within the 20 m × 20 m real time images, is presented to show the same area of the sample. From Fig. 1(a) and (b), rapid dissolution can be observed in the first 15 min immersion. Because this topographical change is reversible when the sample is removed from the solution and exposed to air, it is attributed to the dissolution of air-formed oxide film. The chemical dissolution of oxide film occurred evenly over the whole surface. As a result, polishing scratches became clearer in Fig. 1(b). The dissolution of the oxide film also led to a direct exposure of the metallic substrate on the aggressive medium. Five minutes later, as shown in Fig. 1(c), some defects or inclusions appeared on the surface. Consequently, localized dissolution and micro pitting were initiated, as marked by the black arrow in Fig. 1(c). Between 20 min and 2 h, few changes in the pitting hole can be observed. In Fig. 2, the profiles of the pitting site at different immersion times are plotted. The position of each profile on the surface is the same, and is marked by the green line in Fig. 1(f). It can be clearly observed that the pitting was initiated after 15 min of immersion. Its depth and size increased rapidly, as shown in Fig. 2(d); then it kept stable as the immersion continued (Fig. 2(e and f)). This region can be characterized as a metastable pitting nucleus in its early stage. The metastable pitting nucleus forms quickly after a uniform dissolution of the oxide film on stainless steel in a reductive acid of dilute HCl solution. This is because the corrosion susceptive sites are directly exposed to the aggressive medium. However, not all metastable pitting can develop into a macro pitting. Many complicated influences are closely related to this process, including repassivation, the nature of surface defects, dissolution of inclusion, and the local environment at the metastable pitting sites [25]. As can be observed in Fig. 1, during the whole immersion period, some corrosion active inclusions eventually dissolved and disappeared, and some chemically stable particles remained unchanged on the surface. SEM and EDS experiments confirmed that these were TiN particles. Since the 0.5 M HCl solution is also aggressive to the scanner of AFM instrument, in-situ AFM measurement cannot last in it for a very long time. The optical microscope was used to observe in-situ topographical changes in the corrosion process for longer periods. Fig. 3 shows in-situ digital optical microscope images of 1Cr18Ni9Ti stainless steel over the different immersion times in the 0.5 M HCl solution. Fig. 3(a) shows the image of the sample before immersion. It is evident that, except for a couple of cubic TiN particles, the sur-
2. Experimental The chemical composition of 1Cr18Ni9Ti stainless steel used in this work is listed in Table 1. The samples were prepared and embedded in epoxy resin. The working surface was polished with 400, 800, 1200 and 1500 grit silicon carbide paper and 1 m alumina powder. The polished samples were ultrasonically cleaned in acetone and ethanol. The ECN was measured using EN500 electrochemical noise and galvanic corrosion monitor (Wuhan Corrtest Instrument Co., Ltd.). Two identical stainless steel electrodes were used as working electrodes to measure electrochemical current noise. Both of them were buried in epoxy with two square faces of 1 mm × 1 mm exposed as the working surface. The polished electrode was mounted to a cell, which was designed to execute the ECN measurement and in-situ optical microscope detector synchronously. The electrochemical noise data recording started immediately after the solution was added to the cell. The sampling time of each ECN measurement was set to 1024 s with a frequency of 2 Hz. In order to record the current noise in 9 h of immersing time, the ECN measurement was executed continuously 30 times. DWT was employed to analyze the ECN data in this work; Aballe et al. had described its background [21,22]. By using DWT, the electrochemical current noise data was decomposed into two kinds of coefficients: (1) the smooth coefficient, s8 , which contains information about the general trend of the signal; (2) the detail coefficients, dj = (d1 , d2 , . . ., d8 ), which contains information about the local fluctuations in the signal. Each set of coefficients d1 , d2 , . . ., d8 and s8 is called a crystal. Transients with different time scales are attributed to the detail coefficients, d1 –d8 , whose time scales are shown in Table 2. The slowest processes represented by s8 are the principal part of the dc trend. They are attributed to general process, or the asymmetry between two working electrodes, which usually contains most of the signal’s energy. In order to eliminate the dc trend influence, the energy of s8 will not be considered in this paper when plotting the EDP of wavelet coefficients. The in-situ AFM measurement was performed using UltraObjective Nanostation III (SIS GmbH). Topographic images were recorded in non-contact mode. The time duration for obtaining one image composed of 256 lines was about 4.27 min at a scanning frequency of 1 Hz. The in-situ optical microscope images were collected using Table 2 Time scale of crystal d1 –d8 . Crystal
d1
d2
d3
d4
d5
d6
d7
d8
Time scale(s)
0.5–1
1–2
2–4
4–8
8–16
16–32
32–64
64–128
7136
Y. Li et al. / Electrochimica Acta 54 (2009) 7134–7140
Fig. 1. In-situ AFM images (10 m × 10 m) of 1Cr18Ni9Ti stainless steel in the air (a), immersing in 0.5 M HCl for 15 min (b), 20 min (c), 25 min (d), 1 h (e), 2 h (f).
Y. Li et al. / Electrochimica Acta 54 (2009) 7134–7140
7137
Fig. 2. Topographical profiles of the pitting site with different immersion time (a) in the air, (b) 15 min, (c) 20 min, (d) 25 min, (e) 1 h, (f) 2 h.
face is smooth and featureless. The whole surface becomes rougher continuously during the 9-h immersion. Moreover, some localized topographical changes are noticeable. After being immersed in a 0.5 M HCl solution for 15 min, some tiny pitting appears rapidly on the surface in Fig. 3(b). Fig. 3(c) and (d) shows that the number of pitting increases (refer to Fig. 3(b)), and many new pitting sites appear after immersion for 1 h. In Fig. 3(d) the number of pitting increases significantly, but their size looks unchanged after immersion for 1 h. As shown in Fig. 3(e) and (f), the topographical character of pitting changes obviously. Most of the pitting sites which formed in Fig. 3(d) turned to passive; only a few of them remained active. Pitting corrosion concentrated on these sites, which were marked in Fig. 3(f) by a red arrow, with the size increased. It was found, from Fig. 3(g) and (h), that the corrosion morphology changed remarkably in this period. The main corrosion type turned out to be the growth of stable pitting. The size of pitting marked in Fig. 3(f) increases distinctly. Fig. 4 shows typical SEM images of 1Cr18Ni9Ti stainless steel before and after immersing in 0.5 M HCl solution for 9 h. It is evident that the surface appears smooth and featureless except for a couple of white particles before immersion. A significant pitting corrosion is observed in Fig. 4(b), and two kinds of particles on the surface can be differentiated. One is some rectangular particles with size around 10 m, the other is some small round-shape particles with size of 1 m. The wavelength dispersive spectrometer (WDS) was employed to confirm their composition. These two kinds of particles were numbered in Fig. 4(b), and Table 3 gives the results of their atomic ratio. It is revealed that the main elements of such particles in 1Cr18Ni9Ti stainless steel are Ti and N. The atomic ratio of Ti and N is close to 1:1 except for tiny C element. It is found from Fig. 4 that these white particles remain unchanged in the whole corrosion processes, and the pitting tends to concentrate on the sites closed to
Table 3 Atomic ratio of intermetallic particles numbered in Fig. 4(b). Ti (%)
N (%)
Total (%)
Rectangular particles
No. 1 2 3
C (%) 1.9326 1.0897 2.7355
50.2021 46.9671 47.3699
47.8653 51.9432 49.8946
100.0 100.0 100.0
Round-shape particles
4 5 6
10.6317 9.8754 10.0584
45.6060 46.6521 45.7642
43.7623 43.4725 44.1774
100.0 100.0 100.0
these white particles, and the same results have also been observed in Figs. 1 and 3. The boundary of these TiN particles becomes pitting corrosion susceptive sites. The detailed mechanism of the corrosion initiation of TiN particles in 1Cr18Ni9Ti will be discussed in another paper. From both the in-situ AFM images and in-situ optical microscope images, it is found that, during the first 15 min immersion, the major event is the uniform dissolution of the oxide film formed on the stainless steel. Consequently, the corrosion susceptive sites on the metal surface were directly exposed to the aggressive medium, and resulted in metastable pitting nuclei. From 15 min to 3 h, competitive processes such as the initiation, growth and repassivation of metastable pitting occurred on the surface. Some of the pitting sites occurred on the surface, but most of them repassivated. Only a few of them remained active and formed stable pitting, as marked in Fig. 3(f). The initiation and repassivation of the metastable pitting were caused by main corrosion events in this process. After being immersed for 3 h, no more metastable pitting nuclei formed, and the growth of existing stable pitting became dominant. When the 1Cr18Ni9Ti stainless steel electrode was immersed in the 0.5 M HCl solution, the ECN measurements were executed continuously 30 times in order to collect the current noise during 9 h. The DWT was employed to decompose these 30 groups of ECN data, and then the energy of each set of wavelet coefficients (d1 –d8 ) at different times was calculated. The energy distribution plot (EDP) over time is shown in Fig. 5. The energy of the wavelet coefficients from d1 to d3 is weak during the whole process, which means that corrosion events with small time scales ranging from 0.5 to 4 s are negligible in this process. According to the variation of the dominant components in EDP, the process can be divided into three stages. The first stage lasted from 0 to 17 min. The detail coefficient d8 is dominant in this stage. The second stage, lasting from 17 min to 3 h, exhibits a competition between two sets of crystals: d4 –d6 and d7 , d8 . In the early part of this stage, detail coefficients d4 , d5 and d6 are the most active components. However, the relative energy of d4 , d5 and d6 gradually fades away, whereas that of d7 and d8 increases with time and become dominant by the end of this stage. The third stage lasted from 3 to 9 h, when the energy concentrated on d7 and d8 , the energy of the other detail coefficients is negligible. EDP can provide information about the fluctuation of electrochemical transients in both time and frequency domain. However, the physical meaning of the wavelet coefficients and the corre-
7138
Y. Li et al. / Electrochimica Acta 54 (2009) 7134–7140
Fig. 3. In-situ optical microscope images of 1Cr18Ni9Ti stainless steel with different immersing time in 0.5 M HCl, (a) 0 min, (b) 15 min, (c) 30 min, (d) 1 h, (e) 3 h, (f) 5 h, (g) 7 h, (h) 9 h.
sponding transient events are unclear without evidence derived from other techniques. Hudson employed ellipso-microscopy for surface imaging (EMSI) and specially adapted high-resolution contrast-enhanced optical microscopy to study the onset of pitting corrosion on stainless steel. He proved that the initiation and
repassivation of metastable pitting resulted in current spikes lasting between 4 and 10 s [26]. In this study, an attempt was made to correlate the alteration of dominant events in electrochemical noise currents with topographical changes, by using an in-situ AFM and optical microscope.
Y. Li et al. / Electrochimica Acta 54 (2009) 7134–7140
7139
and it sharply rises during the continued immersion. At the third stage, the contribution from d4 to d6 becomes negligible. During the same period of time, initiation and repassivation of metastable pitting is observed using an in-situ AFM and optical microscope. One believes that current transients with time scales between 4 and 32 s should correspond to initiation and repassivation of metastable pitting. This finding is in agreement with the previous investigation [26]. The energy of d7 increases at the end of the second stage, and the other detail coefficients with lower time scales overwhelm at the third stage. The growth of stable pitting experiences the same procedure, as observed using in-situ optical microscopy. It can be deduced that d7 originates from the stable pitting. 4. Conclusions The initiation process of corrosion for 1Cr18Ni9Ti stainless steel in a 0.5 M HCl solution was investigated by using an in-situ AFM, an optical microscope, SEM/WDS and electrochemical noise measurements. The DWT was employed to analyze the ECN data, and the EDP of the detail wavelet coefficients was used to evaluate the activity of various current transients with different time scales. The origin of the wavelet coefficients of the ECN data was verified for the first time according to the simultaneous topographical changes from the in-situ AFM and optical microscope images. The conclusions drawn out are as follows:
Fig. 4. SEM image of 1Cr18Ni9Ti stainless steel before (a) and after (b) immersed in 0.5 M HCl solution for 9 h.
Fig. 5 shows that the energy of d8 is higher than that of the other wavelet coefficients in all three stages. Meanwhile, the AFM and optical microscopic images show that general dissolution and general corrosion occur during the 9-h immersion. It is therefore believed that events with larger time scales originate from general dissolution or general corrosion. It is observed that the energy accumulated by wavelet coefficients d4 , d5 and d6 is small at the beginning of the immersion,
1. The corrosion process of 1Cr18Ni9Ti stainless steel in 0.5 M HCl solution is divided into three stages. At the first stage the airformed oxide film is reductively dissolved, which is associated with d8 . And a number of metastable pitting nuclei occur and repassivate at the second stage, which are represented by d4 , d5 and d6 . The growth of stable pitting associated with d7 becomes dominated at the last stage. 2. The DWT offers a powerful tool for analysis of ECN data. The evolution of EDP over time can be used to distinguish the different corrosion stages. Further research in this field, covering a wider variety of corroding systems, is needed in order to popularize the ECN method as a facile in-situ monitor for different corrosion stages of stainless steel under various applied environments. Acknowledgement The financial support for this study was provided by the Natural Science Foundation of China (50731004) and Technology Support Program of China (2007BAB27B04). References
Fig. 5. EDP with immersing time of 1Cr18Ni9Ti stainless steel in 0.5 M HCl.
[1] J. Stewart, D.E. Williams, Corros. Sci. 33 (1992) 457. [2] F.U. Renner, A. Stierle, H. Doshch, D.M. Kolb, T.L. Lee, J. Zegenhagen, Nature 439 (2006) 707. [3] M.F. Wang, X.G. Li, N. Du, Y.Z. Huang, A. Korsunsky, Electrochem. Commun. 10 (2008) 1000. [4] G.T. Burstein, S.P. Mattin, Corros. Sci. 35 (1993) 57. [5] J. Stewart, D.E. Willianms, Corros. Sci. 33 (1992) 457. [6] A.M. Riley, D.B. Wells, Corros. Sci. 32 (1991) 1307. [7] S.T. Pride, J.R. Scully, J. Electrochem. Soc. 141 (1994) 3028. [8] J.G. Yu, J.L. Luo, Langmuir 18 (2003) 6637. [9] A.M. Lafront, E. Ghali, Electrochim. Acta 51 (2005) 4891. [10] J.A. Wharton, B.G. Mellor, Corros. Sci. 45 (2003) 97. [11] E. Slavcheva, G. Schmitt, Mater. Corros. 53 (2002) 647. [12] C. Liu, D.D. Macdonald, J.M. Bueno, Corrosion 50 (1994) 687. [13] F. Mansfeld, Z. Sun, C.H. Hsu, Electrochim. Acta 46 (2001) 3651. [14] A. Nagiub, F. Mansfeld, Corros. Sci. 43 (2001) 2001. [15] F. Mansfeld, H. Xiao, J. Electrochem. Soc. 140 (1993) 2205. [16] A. Legat, C. Zevnik, Corros. Sci. 35 (1993) 1661. [17] X.D. Dai, R.L. Motard, B. Joseph, Ind. Eng. Chem. Res. 39 (2000) 1256. [18] F. Mansfeld, L.T. Han, C.C. Lee, J. Electrochem. Soc. 143 (1996) L286.
7140 [19] [20] [21] [22] [23]
Y. Li et al. / Electrochimica Acta 54 (2009) 7134–7140
U. Bertocci, C. Gabrielli, F. Huet, J. Electrochem. Soc. 144 (1997) 31. N. Zaveri, R. Sun, N. Zufelt, A. Zhou, Y.Q. Chen, Electrochim. Acta 52 (2007) 5795. A. Aballe, M. Bethencourt, F.J. Botana, Electrochim. Acta 44 (1999) 4805. A. Aballe, M. Bethencourt, F.J. Botana, Electrochem. Commun. 1 (1999) 266. G.F. Qiao, J.P. Ou, Electrochim. Acta 52 (2007) 8008.
[24] B. Zhao, J.H. Li, R.G. Hu, R.G. Du, C.J. Lin, Electrochim. Acta 52 (2007) 3976. [25] C.J. Lin, Z.W. Tian, Acta Phys. Chim. Sinica 3 (1987) 479. [26] C. Punckt, M. Bölscher, H.H. Rotermund, A.S. Mikhailov, L. Organ, N. Budiansky, J.R. Scully, J.L. Hudson, Science 305 (2004) 1133.
Contents lists available at ScienceDirect
Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta
Corrosion initiation of stainless steel in HCl solution studied using electrochemical noise and in-situ atomic force microscope Yan Li, Ronggang Hu, Jingrun Wang, Yongxia Huang, Chang-Jian Lin ∗,1 State Key Laboratory of Physical Chemistry of Solid Surfaces, Department of Chemistry, College of Chemistry and Chemical Engineering, Xiamen University, Xiamen, 361005, China
a r t i c l e
i n f o
Article history: Received 30 March 2009 Received in revised form 15 July 2009 Accepted 15 July 2009 Available online 23 July 2009 Keywords: Corrosion initiation Stainless steel In-situ AFM ECN
a b s t r a c t An in-situ atomic force microscope (AFM), optical microscope and electrochemical noise (ECN) techniques were applied to the investigation of corrosion initiations in an early stage of 1Cr18Ni9Ti stainless steel immersed in 0.5 M HCl solution. The electrochemical current noise data has been analyzed using discrete wavelet transform (DWT). For the first time, the origin of wavelet coefficients is discussed based on the correlation between the evolution of the energy distribution plot (EDP) of wavelet coefficients and topographic changes. It is found that the occurrence of initiation of metastable pitting at susceptive sites is resulted from the reductive breakdown of passive film of stainless steel in the diluted HCL solution. The coefficients d4 –d6 are originated from metastable pitting, d7 represents the formation and growth of stable pitting while d8 corresponds to the general corrosion. © 2009 Elsevier Ltd. All rights reserved.
1. Introduction Stainless steel is a versatile material that is frequently utilized in various corrosive environments. The corrosion initiation process of stainless steel has aroused a great deal of attention in the last few decades because the corrosion initiation is very complicated and it may cause various types of localized corrosions. Stewart and Williams conducted a detailed analysis of the influence of MnS on pitting initiation process, and confirmed that the inclusions dominated as pitting nucleation sites [1]. Renner et al. employed AFM to observe the topographical characteristics of initial corrosion on the atomic scale under different potentials [2]. Wang and Li used electronic speckle pattern interferometry (ESPI) to study the initial corrosion of stainless steel [3]. Their previous studies mainly focused on the mechanism of initial pitting. However, little work about the corrosion initiation of stainless steel in a dilute HCl solution can be found in the literature. The electrochemical noise (ECN) technique has been developed as a powerful technique in corrosion studies. It has been extensively used for the in-situ detection of spontaneous changes in localized corrosion processes, in particular pitting, cavitation attack, stress corrosion cracking or corrosion under coatings. This is made possible by measuring the fluctuations of the current and potential simultaneously, which are generated during the corrosion process
∗ Corresponding author. Tel.: +86 592 2189354; fax: +86 592 2189354. E-mail address: [email protected] (C.-J. Lin). 1 ISE member. 0013-4686/$ – see front matter © 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2009.07.042
[4–8]. Electrochemical noise analysis has also been widely used for the in-situ characterization and monitoring of corrosion processes of various systems, and it offers some advantages over the common electrochemical techniques. They are [9]: (1) very fast measurements enable the user to register and monitor the corrosion process instantaneously; (2) sensitivity to the early initiation process of localized corrosion, unlike conventional techniques that provide little information [10]; (3) possibility to be used without disturbing the system under investigation [11]; (4) a potential in-situ tool to monitor the corrosion stages of constructional engineering. The sensitivity of ECN measurement to localized corrosion is much higher than that of the traditional techniques. However, the data analysis of ECN remains difficult. Many methods have been developed to analyze the ECN data, including statistical analysis [12–15] and spectral analysis [16–20]. The most commonly used ECN analysis methods (statistical and spectral) are devised for stationary signals that do not show distinctive transients. The main disadvantage of those methods is that they analyze signals by averaging the features across the whole time record. Transient information may be lost in this analysis process. Wavelet transform (WT), especially discrete wavelet transform (DWT), has been proposed as an alternative tool to overcome the limitations of traditional methods in the analysis of non-stationary signals of ECN data [21–23]. WT or DWT can provide information about transients in both time and frequency domain. Moreover, it is possible to work with non-stationary signals, so it has been used to differentiate corrosion types and to study corrosion mechanism. Using the wavelet transform method, the ECN data can be decomposed to different wavelet coefficients with corresponding time
Y. Li et al. / Electrochimica Acta 54 (2009) 7134–7140
7135
Table 1 The component of 1Cr18Ni9Ti stainless steel. C (≤(%))
Si (≤(%))
Mn (≤(%))
P (≤(%))
S (≤(%))
Ti (≤(%))
Cr
Ni
Fe
0.12
1.00
2.00
0.035
0.30
0.98
17.00–19.00
8.00–11.00
Balance
scales [20–22], which are associated with different electrochemical transient events. Aballe et al. introduced the energy distribution plot (EDP) to monitor the fluctuation of transient events [21,22]. Qiao and Ou and Zhao et al., respectively employed EDP to analyze the ECN of Q235 carbon steel in cement mortar [23,24]. However, the corresponding corrosion event of each wavelet coefficient is still unclear. In this work, the corrosion initiation of 1Cr18Ni9Ti stainless steel in dilute HCl solution was studied using an in-situ AFM, optical microscope and electrochemical noise. An attempt was made to verify the origin of electrochemical current noise using the correlation between the corrosion morphologies and the wavelet coefficients.
a digital microscope (KEYENCE, VHX-600E), which achieved a field depth at least 20 times greater than that of traditional optical microscopes. The scanning electron microscope (SEM) and the wavelength dispersive spectrometer (WDS) were executed by the electron probe micro-analyzer (Model JXA-8100, JEOL, Japan). 3. Results and discussion Fig. 1 shows the topographical changes of 1Cr18Ni9Ti stainless steel in 0.5 M HCl solution over different immersion times. Due to the fatigue effect of the Piezo scanner, these continuous in-situ AFM images shift slightly. To reduce the influence of the shift, part of an image, measuring 10 m × 10 m within the 20 m × 20 m real time images, is presented to show the same area of the sample. From Fig. 1(a) and (b), rapid dissolution can be observed in the first 15 min immersion. Because this topographical change is reversible when the sample is removed from the solution and exposed to air, it is attributed to the dissolution of air-formed oxide film. The chemical dissolution of oxide film occurred evenly over the whole surface. As a result, polishing scratches became clearer in Fig. 1(b). The dissolution of the oxide film also led to a direct exposure of the metallic substrate on the aggressive medium. Five minutes later, as shown in Fig. 1(c), some defects or inclusions appeared on the surface. Consequently, localized dissolution and micro pitting were initiated, as marked by the black arrow in Fig. 1(c). Between 20 min and 2 h, few changes in the pitting hole can be observed. In Fig. 2, the profiles of the pitting site at different immersion times are plotted. The position of each profile on the surface is the same, and is marked by the green line in Fig. 1(f). It can be clearly observed that the pitting was initiated after 15 min of immersion. Its depth and size increased rapidly, as shown in Fig. 2(d); then it kept stable as the immersion continued (Fig. 2(e and f)). This region can be characterized as a metastable pitting nucleus in its early stage. The metastable pitting nucleus forms quickly after a uniform dissolution of the oxide film on stainless steel in a reductive acid of dilute HCl solution. This is because the corrosion susceptive sites are directly exposed to the aggressive medium. However, not all metastable pitting can develop into a macro pitting. Many complicated influences are closely related to this process, including repassivation, the nature of surface defects, dissolution of inclusion, and the local environment at the metastable pitting sites [25]. As can be observed in Fig. 1, during the whole immersion period, some corrosion active inclusions eventually dissolved and disappeared, and some chemically stable particles remained unchanged on the surface. SEM and EDS experiments confirmed that these were TiN particles. Since the 0.5 M HCl solution is also aggressive to the scanner of AFM instrument, in-situ AFM measurement cannot last in it for a very long time. The optical microscope was used to observe in-situ topographical changes in the corrosion process for longer periods. Fig. 3 shows in-situ digital optical microscope images of 1Cr18Ni9Ti stainless steel over the different immersion times in the 0.5 M HCl solution. Fig. 3(a) shows the image of the sample before immersion. It is evident that, except for a couple of cubic TiN particles, the sur-
2. Experimental The chemical composition of 1Cr18Ni9Ti stainless steel used in this work is listed in Table 1. The samples were prepared and embedded in epoxy resin. The working surface was polished with 400, 800, 1200 and 1500 grit silicon carbide paper and 1 m alumina powder. The polished samples were ultrasonically cleaned in acetone and ethanol. The ECN was measured using EN500 electrochemical noise and galvanic corrosion monitor (Wuhan Corrtest Instrument Co., Ltd.). Two identical stainless steel electrodes were used as working electrodes to measure electrochemical current noise. Both of them were buried in epoxy with two square faces of 1 mm × 1 mm exposed as the working surface. The polished electrode was mounted to a cell, which was designed to execute the ECN measurement and in-situ optical microscope detector synchronously. The electrochemical noise data recording started immediately after the solution was added to the cell. The sampling time of each ECN measurement was set to 1024 s with a frequency of 2 Hz. In order to record the current noise in 9 h of immersing time, the ECN measurement was executed continuously 30 times. DWT was employed to analyze the ECN data in this work; Aballe et al. had described its background [21,22]. By using DWT, the electrochemical current noise data was decomposed into two kinds of coefficients: (1) the smooth coefficient, s8 , which contains information about the general trend of the signal; (2) the detail coefficients, dj = (d1 , d2 , . . ., d8 ), which contains information about the local fluctuations in the signal. Each set of coefficients d1 , d2 , . . ., d8 and s8 is called a crystal. Transients with different time scales are attributed to the detail coefficients, d1 –d8 , whose time scales are shown in Table 2. The slowest processes represented by s8 are the principal part of the dc trend. They are attributed to general process, or the asymmetry between two working electrodes, which usually contains most of the signal’s energy. In order to eliminate the dc trend influence, the energy of s8 will not be considered in this paper when plotting the EDP of wavelet coefficients. The in-situ AFM measurement was performed using UltraObjective Nanostation III (SIS GmbH). Topographic images were recorded in non-contact mode. The time duration for obtaining one image composed of 256 lines was about 4.27 min at a scanning frequency of 1 Hz. The in-situ optical microscope images were collected using Table 2 Time scale of crystal d1 –d8 . Crystal
d1
d2
d3
d4
d5
d6
d7
d8
Time scale(s)
0.5–1
1–2
2–4
4–8
8–16
16–32
32–64
64–128
7136
Y. Li et al. / Electrochimica Acta 54 (2009) 7134–7140
Fig. 1. In-situ AFM images (10 m × 10 m) of 1Cr18Ni9Ti stainless steel in the air (a), immersing in 0.5 M HCl for 15 min (b), 20 min (c), 25 min (d), 1 h (e), 2 h (f).
Y. Li et al. / Electrochimica Acta 54 (2009) 7134–7140
7137
Fig. 2. Topographical profiles of the pitting site with different immersion time (a) in the air, (b) 15 min, (c) 20 min, (d) 25 min, (e) 1 h, (f) 2 h.
face is smooth and featureless. The whole surface becomes rougher continuously during the 9-h immersion. Moreover, some localized topographical changes are noticeable. After being immersed in a 0.5 M HCl solution for 15 min, some tiny pitting appears rapidly on the surface in Fig. 3(b). Fig. 3(c) and (d) shows that the number of pitting increases (refer to Fig. 3(b)), and many new pitting sites appear after immersion for 1 h. In Fig. 3(d) the number of pitting increases significantly, but their size looks unchanged after immersion for 1 h. As shown in Fig. 3(e) and (f), the topographical character of pitting changes obviously. Most of the pitting sites which formed in Fig. 3(d) turned to passive; only a few of them remained active. Pitting corrosion concentrated on these sites, which were marked in Fig. 3(f) by a red arrow, with the size increased. It was found, from Fig. 3(g) and (h), that the corrosion morphology changed remarkably in this period. The main corrosion type turned out to be the growth of stable pitting. The size of pitting marked in Fig. 3(f) increases distinctly. Fig. 4 shows typical SEM images of 1Cr18Ni9Ti stainless steel before and after immersing in 0.5 M HCl solution for 9 h. It is evident that the surface appears smooth and featureless except for a couple of white particles before immersion. A significant pitting corrosion is observed in Fig. 4(b), and two kinds of particles on the surface can be differentiated. One is some rectangular particles with size around 10 m, the other is some small round-shape particles with size of 1 m. The wavelength dispersive spectrometer (WDS) was employed to confirm their composition. These two kinds of particles were numbered in Fig. 4(b), and Table 3 gives the results of their atomic ratio. It is revealed that the main elements of such particles in 1Cr18Ni9Ti stainless steel are Ti and N. The atomic ratio of Ti and N is close to 1:1 except for tiny C element. It is found from Fig. 4 that these white particles remain unchanged in the whole corrosion processes, and the pitting tends to concentrate on the sites closed to
Table 3 Atomic ratio of intermetallic particles numbered in Fig. 4(b). Ti (%)
N (%)
Total (%)
Rectangular particles
No. 1 2 3
C (%) 1.9326 1.0897 2.7355
50.2021 46.9671 47.3699
47.8653 51.9432 49.8946
100.0 100.0 100.0
Round-shape particles
4 5 6
10.6317 9.8754 10.0584
45.6060 46.6521 45.7642
43.7623 43.4725 44.1774
100.0 100.0 100.0
these white particles, and the same results have also been observed in Figs. 1 and 3. The boundary of these TiN particles becomes pitting corrosion susceptive sites. The detailed mechanism of the corrosion initiation of TiN particles in 1Cr18Ni9Ti will be discussed in another paper. From both the in-situ AFM images and in-situ optical microscope images, it is found that, during the first 15 min immersion, the major event is the uniform dissolution of the oxide film formed on the stainless steel. Consequently, the corrosion susceptive sites on the metal surface were directly exposed to the aggressive medium, and resulted in metastable pitting nuclei. From 15 min to 3 h, competitive processes such as the initiation, growth and repassivation of metastable pitting occurred on the surface. Some of the pitting sites occurred on the surface, but most of them repassivated. Only a few of them remained active and formed stable pitting, as marked in Fig. 3(f). The initiation and repassivation of the metastable pitting were caused by main corrosion events in this process. After being immersed for 3 h, no more metastable pitting nuclei formed, and the growth of existing stable pitting became dominant. When the 1Cr18Ni9Ti stainless steel electrode was immersed in the 0.5 M HCl solution, the ECN measurements were executed continuously 30 times in order to collect the current noise during 9 h. The DWT was employed to decompose these 30 groups of ECN data, and then the energy of each set of wavelet coefficients (d1 –d8 ) at different times was calculated. The energy distribution plot (EDP) over time is shown in Fig. 5. The energy of the wavelet coefficients from d1 to d3 is weak during the whole process, which means that corrosion events with small time scales ranging from 0.5 to 4 s are negligible in this process. According to the variation of the dominant components in EDP, the process can be divided into three stages. The first stage lasted from 0 to 17 min. The detail coefficient d8 is dominant in this stage. The second stage, lasting from 17 min to 3 h, exhibits a competition between two sets of crystals: d4 –d6 and d7 , d8 . In the early part of this stage, detail coefficients d4 , d5 and d6 are the most active components. However, the relative energy of d4 , d5 and d6 gradually fades away, whereas that of d7 and d8 increases with time and become dominant by the end of this stage. The third stage lasted from 3 to 9 h, when the energy concentrated on d7 and d8 , the energy of the other detail coefficients is negligible. EDP can provide information about the fluctuation of electrochemical transients in both time and frequency domain. However, the physical meaning of the wavelet coefficients and the corre-
7138
Y. Li et al. / Electrochimica Acta 54 (2009) 7134–7140
Fig. 3. In-situ optical microscope images of 1Cr18Ni9Ti stainless steel with different immersing time in 0.5 M HCl, (a) 0 min, (b) 15 min, (c) 30 min, (d) 1 h, (e) 3 h, (f) 5 h, (g) 7 h, (h) 9 h.
sponding transient events are unclear without evidence derived from other techniques. Hudson employed ellipso-microscopy for surface imaging (EMSI) and specially adapted high-resolution contrast-enhanced optical microscopy to study the onset of pitting corrosion on stainless steel. He proved that the initiation and
repassivation of metastable pitting resulted in current spikes lasting between 4 and 10 s [26]. In this study, an attempt was made to correlate the alteration of dominant events in electrochemical noise currents with topographical changes, by using an in-situ AFM and optical microscope.
Y. Li et al. / Electrochimica Acta 54 (2009) 7134–7140
7139
and it sharply rises during the continued immersion. At the third stage, the contribution from d4 to d6 becomes negligible. During the same period of time, initiation and repassivation of metastable pitting is observed using an in-situ AFM and optical microscope. One believes that current transients with time scales between 4 and 32 s should correspond to initiation and repassivation of metastable pitting. This finding is in agreement with the previous investigation [26]. The energy of d7 increases at the end of the second stage, and the other detail coefficients with lower time scales overwhelm at the third stage. The growth of stable pitting experiences the same procedure, as observed using in-situ optical microscopy. It can be deduced that d7 originates from the stable pitting. 4. Conclusions The initiation process of corrosion for 1Cr18Ni9Ti stainless steel in a 0.5 M HCl solution was investigated by using an in-situ AFM, an optical microscope, SEM/WDS and electrochemical noise measurements. The DWT was employed to analyze the ECN data, and the EDP of the detail wavelet coefficients was used to evaluate the activity of various current transients with different time scales. The origin of the wavelet coefficients of the ECN data was verified for the first time according to the simultaneous topographical changes from the in-situ AFM and optical microscope images. The conclusions drawn out are as follows:
Fig. 4. SEM image of 1Cr18Ni9Ti stainless steel before (a) and after (b) immersed in 0.5 M HCl solution for 9 h.
Fig. 5 shows that the energy of d8 is higher than that of the other wavelet coefficients in all three stages. Meanwhile, the AFM and optical microscopic images show that general dissolution and general corrosion occur during the 9-h immersion. It is therefore believed that events with larger time scales originate from general dissolution or general corrosion. It is observed that the energy accumulated by wavelet coefficients d4 , d5 and d6 is small at the beginning of the immersion,
1. The corrosion process of 1Cr18Ni9Ti stainless steel in 0.5 M HCl solution is divided into three stages. At the first stage the airformed oxide film is reductively dissolved, which is associated with d8 . And a number of metastable pitting nuclei occur and repassivate at the second stage, which are represented by d4 , d5 and d6 . The growth of stable pitting associated with d7 becomes dominated at the last stage. 2. The DWT offers a powerful tool for analysis of ECN data. The evolution of EDP over time can be used to distinguish the different corrosion stages. Further research in this field, covering a wider variety of corroding systems, is needed in order to popularize the ECN method as a facile in-situ monitor for different corrosion stages of stainless steel under various applied environments. Acknowledgement The financial support for this study was provided by the Natural Science Foundation of China (50731004) and Technology Support Program of China (2007BAB27B04). References
Fig. 5. EDP with immersing time of 1Cr18Ni9Ti stainless steel in 0.5 M HCl.
[1] J. Stewart, D.E. Williams, Corros. Sci. 33 (1992) 457. [2] F.U. Renner, A. Stierle, H. Doshch, D.M. Kolb, T.L. Lee, J. Zegenhagen, Nature 439 (2006) 707. [3] M.F. Wang, X.G. Li, N. Du, Y.Z. Huang, A. Korsunsky, Electrochem. Commun. 10 (2008) 1000. [4] G.T. Burstein, S.P. Mattin, Corros. Sci. 35 (1993) 57. [5] J. Stewart, D.E. Willianms, Corros. Sci. 33 (1992) 457. [6] A.M. Riley, D.B. Wells, Corros. Sci. 32 (1991) 1307. [7] S.T. Pride, J.R. Scully, J. Electrochem. Soc. 141 (1994) 3028. [8] J.G. Yu, J.L. Luo, Langmuir 18 (2003) 6637. [9] A.M. Lafront, E. Ghali, Electrochim. Acta 51 (2005) 4891. [10] J.A. Wharton, B.G. Mellor, Corros. Sci. 45 (2003) 97. [11] E. Slavcheva, G. Schmitt, Mater. Corros. 53 (2002) 647. [12] C. Liu, D.D. Macdonald, J.M. Bueno, Corrosion 50 (1994) 687. [13] F. Mansfeld, Z. Sun, C.H. Hsu, Electrochim. Acta 46 (2001) 3651. [14] A. Nagiub, F. Mansfeld, Corros. Sci. 43 (2001) 2001. [15] F. Mansfeld, H. Xiao, J. Electrochem. Soc. 140 (1993) 2205. [16] A. Legat, C. Zevnik, Corros. Sci. 35 (1993) 1661. [17] X.D. Dai, R.L. Motard, B. Joseph, Ind. Eng. Chem. Res. 39 (2000) 1256. [18] F. Mansfeld, L.T. Han, C.C. Lee, J. Electrochem. Soc. 143 (1996) L286.
7140 [19] [20] [21] [22] [23]
Y. Li et al. / Electrochimica Acta 54 (2009) 7134–7140
U. Bertocci, C. Gabrielli, F. Huet, J. Electrochem. Soc. 144 (1997) 31. N. Zaveri, R. Sun, N. Zufelt, A. Zhou, Y.Q. Chen, Electrochim. Acta 52 (2007) 5795. A. Aballe, M. Bethencourt, F.J. Botana, Electrochim. Acta 44 (1999) 4805. A. Aballe, M. Bethencourt, F.J. Botana, Electrochem. Commun. 1 (1999) 266. G.F. Qiao, J.P. Ou, Electrochim. Acta 52 (2007) 8008.
[24] B. Zhao, J.H. Li, R.G. Hu, R.G. Du, C.J. Lin, Electrochim. Acta 52 (2007) 3976. [25] C.J. Lin, Z.W. Tian, Acta Phys. Chim. Sinica 3 (1987) 479. [26] C. Punckt, M. Bölscher, H.H. Rotermund, A.S. Mikhailov, L. Organ, N. Budiansky, J.R. Scully, J.L. Hudson, Science 305 (2004) 1133.