Statistical analysis of the repeatability of the crevice corrosion repassivation potential

Martín A. Rodríguez

doi:10.1515/corrrev-2022-0074

Article Publicly Available

Statistical analysis of the repeatability of the crevice corrosion repassivation potential

Martín A. Rodríguez

Published/Copyright: March 20, 2023

Published by

Become an author with De Gruyter Brill

Submit Manuscript Author Information Explore this Subject

From the journal Corrosion Reviews Volume 41 Issue 3

Abstract

An extensive database of crevice corrosion repassivation potentials (E_R,CREV) of corrosion-resistant and high-temperature alloys was analysed with statistical tools. Repeatability of results, which considers multiple tests performed by the same team with the same experimental setup, was assessed. Evaluated variables include those that allegedly affect the outcome of testing techniques used to determine E_R,CREV such as crevice formers material, applied torque, the extent of corrosion propagation before repassivation and electrochemical procedures; and environmental and metallurgical conditions that may change the corrosion susceptibility of materials such as thermal ageing, alloy composition, temperature, chloride concentration and inhibitors. Guidelines to decide on the significance of changes in crevice corrosion repassivation potentials are proposed. Analysis of collected data suggests that a sample size of 5 is appropriate for assessing any change in the environmental or metallurgical conditions on the repassivation potential.

Keywords: chlorides; crevice corrosion; localised corrosion; nickel alloys; passivity

1 Introduction

Crevice corrosion is a type of localised corrosion affecting metals in contact with stagnant solutions within restricted geometries. In the presence of aggressive anions, such as chloride, a concentrated acidic solution may develop in crevices, promoting the local dissolution of the metallic material. Passive metals and alloys are prone to crevice corrosion above a critical temperature and a critical potential, depending on each material/environment system. These critical parameters are experimentally determined (Szklarska-Smialowska 2005). The critical crevice temperature (CCT) is generally determined by method D of ASTM G48 standard (ASTM G48 2020). The critical potential is associated with the crevice corrosion repassivation potential (E_R,CREV) that various testing techniques may obtain (Giordano et al. 2011). E_R,CREV is a critical parameter associated with a certain metallic material resistance (or susceptibility) to crevice corrosion in particular environmental and metallurgical conditions. Alloy chemical composition, microstructure, solution composition and temperature are the main variables that affect E_R,CREV (Dunn et al. 2005a; Hornus et al. 2015; Kehler et al. 2001; Mishra and Frankel 2008; Rodríguez et al. 2010). This parameter is determined by standardised or non-standardised electrochemical tests in the conditions of interest. The selected testing technique, the type and the material of the artificial crevice formers used, and the extent of corrosion propagation before repassivation, among other variables, may affect the value of E_R,CREV that is determined experimentally (Giordano et al. 2011; Rincón Ortíz et al. 2010; Shan and Payer 2010). The desirable features of any technique for determining E_R,CREV are a high reproducibility, a simple experimental set-up and a short testing time. Testing techniques generally applied include cyclic potentiodynamic polarisation, Tsujikawa–Hisamatsu electrochemical and potentiodynamic–galvanostatic–potentiodynamic methods (Evans et al. 2005; Mishra and Frankel 2008; Tsujikawa and Hisamatsu 1980). For a thorough description of these techniques refer to Giordano et al. (2011).

The higher E_R,CREV the better the alloy resistance to crevice corrosion. E_R,CREV has been used to rank alloy performances (Sosa Haudet et al. 2012; Zadorozne et al. 2012), to assess the detrimental/beneficial effects of thermal ageing treatments (Carranza et al. 2008a, 2008b; Dunn et al. 2006; Sridhar et al. 2009), the addition of inhibitors (Carranza et al. 2007; Dunn et al. 2006; Kehler et al. 2001; Miyagusuku et al. 2015; Rincón Ortíz et al. 2013; Rodríguez 2012), temperature changes (Dunn et al. 2005a; Hornus et al. 2015), solution pH (Rodríguez et al. 2010), etc. E_R,CREV was proposed as a critical parameter for assessing the likelihood of localised corrosion occurrence of engineered barriers in nuclear repositories (Carranza 2008; Carranza and Rodríguez 2017; Dunn et al. 2000; Sridhar and Cragnolino 1993).

The objective of this paper is to apply statistical tools (1) to assess the effect of variables that allegedly affect the outcome of testing techniques (electrochemical procedure, the type and the material of the artificial crevice formers used, and the extent of corrosion propagation before repassivation) on the experimental value of E_R,CREV, and (2) to establish a criterion for deciding whether a change in the value of E_R,CREV is significant or not when an environmental or metallurgical variable is modified.

Our research team has collected a large amount of E_R,CREV data for nickel-based alloys and stainless steels in varying testing conditions, over the past 18 years (Carranza et al. 2007, 2008a, 2008b; Giordano et al. 2011; Hornus et al. 2015; Maristany et al. 2016, 2018; Rincón Ortíz et al. 2013; Sosa Haudet et al. 2012; Zadorozne et al. 2012). Part of this database was used to perform the statistical analysis herein. Regarding precision, note that repeatability refers to results of multiple tests performed by the same team with the same experimental setup, replicability refers to results of multiple tests performed by different teams with the same experimental setup, and reproducibility refers to results of multiple tests performed by different teams with different experimental setups (McArthur 2019; Plesser 2018). The issue of reproducibility has raised a recent debate in all scientific fields (Fanelli 2018). In the field of Material Science, ASTM E691 standard sets the definitions of repeatability and reproducibility. Repeatability is the precision of test results from tests conducted on identical material by the same test method in a single laboratory. Reproducibility is the precision of test results from tests conducted on identical material by the same test method in different laboratories (at least six laboratories must be involved) (ASTM E691 2021). Tests considered in this work were performed by various operators, from 2003 to 2021, using similar experimental setups, at the laboratories of the Corrosion group of the National Commission of Atomic Energy (CNEA) of Argentina. Thus variation in this work is associated with variability within the lab (repeatability).

2 Materials and methods

2.1 Cyclic potentiodynamic polarisation

The cyclic potentiodynamic polarisation (CPP) is a common technique used to determine the repassivation potential either for pitting or crevice corrosion (ASTM G61 2018). Anodic polarisation is usually started a few millivolts below the open circuit potential up to reaching a pre-established current density (i_REV, usually from 1 to 10 mA/cm²), at a sufficiently low scan rate (0.167 mV/s). When i_REV is attained, the potential scan is reversed from anodic to cathodic. The test ends when the current density drops below a certain value or becomes cathodic. The repassivation potential may be defined in different ways. The most common definition, and the one used herein, is the crossover potential between the forward and reverse scans. Figure 1a shows a CPP test for mill-annealed alloy 22 in 1 mol/L NaCl, at 90 °C, indicating i_REV and E_R,CREV.

Figure 1:

Determination of E_R,CREV for mill-annealed alloy 22 in 1 mol/L NaCl, at 90 °C by (a) CPP, (b) THE, and (c) PD–GS–PD tests.

2.2 Tsujikawa–Hisamatsu electrochemical method

The Tsujikawa–Hisamatsu electrochemical (THE) method is a testing technique specially designed to determine E_R,CREV (Tsujikawa and Hisamatsu 1980; ASTM G192 2020). ASTM G192 standard suggests CPP method as a first fast screening of E_R,CREV and THE method for fine-tuning E_R,CREV, especially when the environment is not highly aggressive and/or the material is very resistant to localised corrosion. Step 1 of this method consists of an anodic polarisation from 100 mV below the open circuit potential, at a scan rate of 0.167 mV/s, up to reaching a pre-established anodic current density (i_GS, usually from 2 to 20 μA/cm²). Step 2 consists of a galvanostatic polarisation at i_GS for a fixed period, generally t_GS = 2 h, while the potential is registered. Step 2 produces a controlled propagation of localised corrosion (if any develops) at a potential that is intended to be below that of oxidative dissolution of chromium-rich passive films. Step 3 shifts to potentiostatic mode. Successive potentiostatic treatments of 2 h each are applied, each time at 10 mV lower than the previous treatment, while the current is monitored. E_R,CREV is defined as the highest potential step where a decreasing current over time is recorded provided that the following steps show the same decreasing trend. Figure 1b shows a plot of a THE test for mill-annealed alloy 22 in 1 mol/L NaCl, at 90 °C. In this example repassivation occurred after only 3 potentiostatic steps, but further steps are needed to confirm the current-versus-time decreasing trend. 10 or even 15 potentiostatic treatments may be required to finish Step 3 of this method. THE method is significantly more time consuming than CPP (ASTM G192 2020).

2.3 Potentiodynamic–galvanostatic–potentiodynamic method

The potentiodynamic–galvanostatic–potentiodynamic (PD–GS–PD) method was also proposed for fine-tuning E_R,CREV but replacing the time-consuming potentiostatic treatments with a potentiodynamic scan (Mishra and Frankel 2008). Steps 1 and 2 are identical to those of THE method. Step 3 is a cathodic potentiodynamic scan at 0.167 mV/s. E_R,CREV is defined as the crossover potential between the forward and reverse scans. Figure 1c shows a PD–GS–PD test for mill-annealed alloy 22 in 1 mol/L NaCl, at 90 °C.

2.4 Artificial crevice formers and experimental setup

The crevice formers used in the tests described above are generally serrated washers called multiple crevice assembly (MCA) (ASTM G192 2020; ASTM G78 2020). Since several crevice initiation sites are available, localised corrosion may start in many locations. Various materials have been used to fabricate the crevice formers. ASTM G192 specifies that the MCA crevice formers should be fabricated using a hard non-conductive ceramic material (e.g. alumina or mullite) and it should be covered with a 70 µm-thick PTFE tape (standard military grade MIL-T-27730A). Fasteners used to secure the two crevice formers on each side of the test specimen should not suffer localized corrosion. Titanium alloy bolts, nuts and washers are commonly used. The standard pressure on the MCA crevice formers should be 3.4 N m or higher to form a tight crevice. A calibrated torque wrench is used to apply the torque. Electrical contact between the bolt and the test electrode is avoided by insulating the bolt with a non-metallic sleeve or by wrapping the bolt with PTFE tape (ASTM G192 2020; Giordano et al. 2011). These specifications have been met in all tests herein unless otherwise stated.

A finish wet grinding of 600 grit silicon carbide paper was performed to the alloy specimen 1 h before testing. Tested alloys were in the mill-annealed (MA) condition unless otherwise stated. This metallurgical state generally corresponds to full solubilisation of carbides and/or secondary phases. The test solution was purged with high-purity nitrogen at least 1 h before each test and throughout the tests. Reagent grade chemicals were used in all tests. A saturated calomel electrode (SCE) or saturated silver chloride electrode (SSC) were used as reference electrodes. All the potential values reported in this paper are in Volts with respect to the saturated calomel electrode (V_SCE) which has a potential of +0.244 V versus the standard hydrogen electrode (SHE). SSC has a potential of +0.197 V versus the SHE.

Table 1 shows the identification (ID) and the experimental conditions of each sample considered in the present study. Table 2 shows the nominal chemical composition of studied alloys. They include corrosion-resistant (CRA) and high temperature (HTA) alloys. CRAs are mainly used for low temperature aqueous or condensed systems. HTAs are designed to withstand high temperature and dry or gaseous corrosion (Rebak 2000). Table 2 also shows the Pitting Resistance Equivalent (PRE) which is a parameter indicative of the localised corrosion resistance of chromium-containing alloys. PRE is obtained by Equation (1), as a function of the contents in weight % of Cr, Mo and W (x_Cr, x_Mo, x_W), respectively (Szklarska-Smialowska 2005).

(1)PRE=xCr+3.3(xMo+0.5xW)

Table 1:

Identification (ID) and testing conditions for each sample.

Sample ID	Test conditions
Sample#01	PD–GS–PD tests on MA alloy 22 in deaerated 1 mol/L NaCl at 90 °C (i_GS = 2 μA/cm², t_GS = 2 h) with ceramic crevice formers wrapped with 70 μm-thick PTFE tape, torqued to 5.0–8.0 N m
Sample#02	PD–GS–PD tests on MA alloy 22 in deaerated 1 mol/L NaCl at 90 °C (i_GS = 2 μA/cm², t_GS = 2 h) with solid PTFE crevice formers torqued to 1.0 N m
Sample#03	PD–GS–PD tests on MA alloy 22 in deaerated 1 mol/L NaCl at 90 °C (i_GS = 20 μA/cm², t_GS = 2 h) with ceramic crevice formers wrapped with 70 μm-thick PTFE tape, torqued to 5.0–8.0 N m
Sample#04	THE tests on MA alloy 22 in deaerated 1 mol/L NaCl at 90 °C (i_GS = 2 μA/cm², t_GS = 2 h) with ceramic crevice formers wrapped with 70 μm-thick PTFE tape, torqued to 5.0–8.0 N m
Sample#05	CPP tests on MA alloy 22 in deaerated 1 mol/L NaCl at 90 °C (i_REV = 1 mA/cm²) with ceramic crevice formers wrapped with 70 μm-thick PTFE tape, torqued to 5.0–8.0 N m
Sample#06	THE tests on MA alloy 22 in deaerated 1 mol/L NaCl at 90 °C (i_GS = 2 μA/cm², t_GS = 2 h) with ceramic crevice formers wrapped with 70 μm-thick PTFE tape, torqued to 8.0 N m
Sample#07	THE tests on MA alloy 22 in deaerated 1 mol/L NaCl at 90 °C (i_GS = 2 μA/cm², t_GS = 2 h) with ceramic crevice formers wrapped with 70 μm-thick PTFE tape, torqued to 5.0 N m
Sample#08	Alloy 22 aged for 10 h at 760 °C and quenched in water. Tests in 1 mol/L NaCl at 90 °C.^a
Sample#09	Alloy 22 aged for 5 min at 870 °C and quenched in water. Tests in 1 mol/L NaCl at 90 °C.^a
Sample#10	Alloy 22 aged for 30 min at 870 °C and quenched in water. Tests in 1 mol/L NaCl at 90 °C.^a
Sample#11	Alloy 22 aged for 1000 h at 538 °C and air-cooled. Tests in 1 mol/L NaCl at 90 °C.^a
Sample#12	MA alloy C-22HS. Tests in 1 mol/L NaCl at 90 °C.^a
Sample#13	Age-hardened alloy C-22HS. Tests in 1 mol/L NaCl at 90 °C.^a
Sample#14	Alloy 600 solution-annealed at 1100 °C for 25 min followed by water quenching. Tests in 0.01 mol/L NaCl at 90 °C.^a
Sample#15	Alloy 600 solution-annealed at 1100 °C for 25 min followed by water quenching and then heat-treated for 10 h at 716 °C followed by water quenching. Tests in 0.01 mol/L NaCl at 90 °C.^a
Sample#16	Alloy 690 solution-annealed at 1100 °C for 25 min followed by water quenching. Tests in 0.01 mol/L NaCl at 90 °C.^a
Sample#17	Alloy 690 solution-annealed at 1100 °C for 25 min followed by water quenching and then heat-treated for 10 h at 716 °C followed by water quenching. Tests in 0.01 mol/L NaCl at 90 °C.^a
Sample#18	MA alloy 600. Tests in 1 mol/L NaCl at 60 °C.^a
Sample#19	MA alloy 690. Tests in 1 mol/L NaCl at 60 °C.^a
Sample#20	MA alloy 800. Tests in 1 mol/L NaCl at 60 °C.^a
Sample#21	MA alloy G-35. Tests in 1 mol/L NaCl at 60 °C.^a
Sample#22	MA alloy G-30. Tests in 1 mol/L NaCl at 60 °C.^a
Sample#23	MA alloy 625. Tests in 1 mol/L NaCl at 60 °C.^a
Sample#24	MA alloy 22. Tests in 1 mol/L NaCl at 60 °C.^a
Sample#25	MA alloy C-22HS. Tests in 1 mol/L NaCl at 60 °C.^a
Sample#26	MA alloy 600. Tests in 0.1 mol/L NaCl at 90 °C.^a
Sample#27	MA alloy 690. Tests in 0.1 mol/L NaCl at 90 °C.^a
Sample#28	MA alloy 800. Tests in 0.1 mol/L NaCl at 90 °C.^a
Sample#29	MA alloy 625. Tests in 0.1 mol/L NaCl at 90 °C.^a
Sample#30	MA alloy 22. Tests in 0.1 mol/L NaCl at 90 °C.^a
Sample#31	MA alloy C-22HS. Tests in 0.1 mol/L NaCl at 90 °C.^a
Sample#32	MA alloy HYBRID-BC1. Tests in 0.1 mol/L NaCl at 90 °C.^a
Sample#33	MA alloy 625. Tests in 0.1 mol/L NaCl at 20 °C.^a
Sample#34	MA alloy 625. Tests in 0.1 mol/L NaCl at 30 °C.^a
Sample#35	MA alloy 625. Tests in 0.1 mol/L NaCl at 40 °C.^a
Sample#36	MA alloy 625. Tests in 0.1 mol/L NaCl at 50 °C.^a
Sample#37	MA alloy 625. Tests in 0.1 mol/L NaCl at 60 °C.^a
Sample#38	MA alloy 625. Tests in 0.1 mol/L NaCl at 70 °C.^a
Sample#39	MA alloy 625. Tests in 0.1 mol/L NaCl at 80 °C.^a
Sample#40	MA alloy 625. Tests in 0.1 mol/L NaCl at 90 °C.^a
Sample#41	MA alloy C-22HS. Tests in CaCl₂ 5 mol/L at 40 °C.^a
Sample#42	MA alloy C-22HS. Tests in CaCl₂ 5 mol/L at 50 °C.^a
Sample#43	MA alloy C-22HS. Tests in CaCl₂ 5 mol/L at 60 °C.^a
Sample#44	MA alloy C-22HS. Tests in CaCl₂ 5 mol/L at 70 °C.^a
Sample#45	MA alloy C-22HS. Tests in CaCl₂ 5 mol/L at 80 °C.^a
Sample#46	MA alloy C-22HS. Tests in CaCl₂ 5 mol/L at 90 °C.^a
Sample#47	MA alloy C-22HS. Tests in CaCl₂ 5 mol/L at 100 °C.^a
Sample#48	MA alloy C-22HS. Tests in CaCl₂ 5 mol/L at 110 °C.^a
Sample#49	MA alloy C-22HS. Tests in CaCl₂ 5 mol/L at 117 °C.^a
Sample#50	MA alloy 690. Tests in 0.01 mol/L NaCl at 60 °C.^a
Sample#51	MA alloy 690. Tests in 0.1 mol/L NaCl at 60 °C.^a
Sample#52	MA alloy 690. Tests in 1 mol/L NaCl at 60 °C.^a
Sample#53	MA alloy 690. Tests in 5 mol/L NaCl at 60 °C.^a
Sample#54	MA alloy 800. Tests in 0.01 mol/L NaCl at 90 °C.^a
Sample#55	MA alloy 800. Tests in 0.1 mol/L NaCl at 90 °C.^a
Sample#56	MA alloy 800. Tests in 1 mol/L NaCl at 90 °C.^a
Sample#57	MA alloy 800. Tests in 5 mol/L NaCl at 90 °C.^a
Sample#58	MA alloy HYBRID-BC1. Tests in 0.1 mol/L NaCl at 90 °C.^a
Sample#59	MA alloy HYBRID-BC1. Tests in 1 mol/L NaCl at 90 °C.^a
Sample#60	MA alloy HYBRID-BC1. Tests in 5 mol/L CaCl₂ at 90 °C.^a
Sample#61	MA alloy 22. Tests in 1 mol/L NaCl + 0.01 mol/L NaNO₃ at 90 °C.^a
Sample#62	MA alloy 22. Tests in 1 mol/L NaCl + 0.02 mol/L NaNO₃ at 90 °C.^a
Sample#63	MA alloy 22. Tests in 1 mol/L NaCl + 0.05 mol/L NaNO₃ at 90 °C.^a
Sample#64	MA alloy 22. Tests in 1 mol/L NaCl + 0.1 mol/L NaNO₃ at 90 °C.^a
Sample#65	MA alloy 22. Tests in 0.1 mol/L NaCl.^a
Sample#66	MA alloy 22. Tests in 0.1 mol/L NaCl + 0.01 mol/L Na₂SO₄.^a
Sample#67	MA alloy 22. Tests in 0.1 mol/L NaCl + 0.02 mol/L Na₂SO₄.^a
Sample#68	MA alloy 22. Tests in 0.1 mol/L NaCl + 0.05 mol/L Na₂SO₄.^a
Sample#69	MA alloy 22. Tests in 0.1 mol/L NaCl + 0.05 mol/L Na₂WO₄.^a
Sample#70	MA alloy 22. Tests in 0.1 mol/L NaCl + 0.1 mol/L Na₂WO₄.^a
Sample#71	MA alloy 22. Tests in 0.1 mol/L NaCl + 0.2 mol/L Na₂WO₄.^a

^aPD–GS–PD tests (i_GS = 2 μA/cm², t_GS = 2 h) in deaerated solutions using ceramic crevice formers wrapped with 70 μm-thick PTFE tape, torqued to 5.0 N m.

Table 2:

Nominal chemical composition (main elements) and PRE of studied alloys.

Alloy	Chemical composition (weight %)	PRE
22 (UNS N06022)^a	Ni-22%Cr-13%Mo-3%W	66
C-22HS (UNS N07022)^a	Ni-21%Cr-17%Mo	72
600 (UNS N06600)^b	Ni-16%Cr-8%Fe	16
690 (UNS N06690)^b	Ni-29%Cr-9%Fe	29
800 (UNS N08800)^b	Fe-32%Ni-21%Cr	21
G-35 (UNS N06035)^a	Ni-33%Cr-8%Mo	57
G-30 (UNS N06030)^a	Ni-30%Cr-15%Fe-5.5%Mo-2.5%W-2%Cu	50
625 (UNS N06625)^a,b	Ni-22%Cr-9%Mo-3.5%(Nb + Ta)	49
HYBRID-BC1 (UNS N10362)^a	Ni-22%Mo-15%Cr	81

^aCRA; ^bHTA.

2.5 Statistical analyses

Figure 2 shows a flowchart of the procedure followed herein for comparing two samples of E_R,CREV from previously collected data. It is important to highlight that these data collection was not originally intended for the comparison made in this work. A confidence interval (CI) was constructed for each sample according to Equation (2), where x is the E_R,CREV sample mean, t* is the Student’s-t value that corresponds to the selected confidence level (95% in this case) and degrees of freedom of the sample, s is the sample standard deviation and n is the sample size (Illowsky and Dean 2018).

(2)x±t*(s/n)

Figure 2:

Procedure applied to compare two samples of E_R,CREV data.

If the CIs of comparing samples did not overlap, E_R,CREV of samples were assumed to be significantly different. Otherwise, a hypothesis test was used to decide whether there were significant differences between samples or not. Hypothesis tests assume that the null hypothesis (H₀) holds and both samples belong to the same population. Student’s-t test was used for samples of equal variances and Welch’s-t test was used when variances were different (Illowsky and Dean 2018). Levene’s hypothesis test was applied to decide on homogeneity of variances (Levene 1960). A significance level of 95% (α = 0.05) was considered for hypothesis tests.

If there were significant differences among samples, either because CIs overlapped or H₀ was rejected in the hypothesis test, the effect size was calculated and reported. To this purpose, Cohen’s d parameter, which is meant to quantify the magnitude of the difference between means, was calculated according to Equation (3), where x₁ and x₂ are the means of the two considered samples, and s₁₂ is the pooled standard deviation (Cohen 2013).

(3)d=(x1−x2)/s12

Table 3 shows rules of thumb values (Cohen 2013; Sawilowsky 2009) for the classification of the effect size depending on the value of parameter d. Besides this guide, a classification of the effect size tailored to E_R,CREV will be proposed later.

Table 3:

Effect size according to (Cohen 2013; Sawilowsky 2009) and suggested values tailored for the comparison of E_R,CREV.

Effect size	Cohen’s d parameter	ΔE_R,CREV, V
Very small	0.01	0.010
Small	0.20	0.030
Moderate	0.50	0.050
Large	0.80	0.100
Very large	1.20	0.300
Huge	2.0	0.600

If H₀ failed to be rejected, and thus no significant changes were found between the samples, the statistical power (SP) of the hypothesis test was calculated and reported. The SP of a hypothesis test is the probability that it correctly rejects H₀. A customary value for SP is 0.80 (80%). Thus, if the condition SP ≥ 0.80 is not met, the sample size should be increased. The sample size for SP = 0.80, α = 0.05 and the corresponding effect size (d) were calculated.

Statistics, confidence intervals, hypothesis tests, power analyses and sample sizes were obtained by codes in the programming language Python (https://www.python.org/) using the libraries NumPy (https://numpy.org/) and SciPy (https://scipy.org/).

3 Results and discussion

The 71 samples indicated in Table 1 were selected among hundreds of data sets for present statistical analyses. Since the obtained values of E_R,CREV may be affected by the selected testing technique and the corresponding experimental setup, the first part of the analyses is devoted to state the significance of variables that are thought to affect E_R,CREV. The second part of the analyses focuses on the effects of metallurgical and environmental variables on E_R,CREV. Finally, significance of E_R,CREV changes are analysed and a guideline for a planned comparison of E_R,CREV is proposed.

3.1 Variables allegedly affecting the outcome of testing techniques

This part of the analyses assesses the effect of variables that allegedly affect the outcome of testing techniques on the experimental value of E_R,CREV. These variables include those associated with the testing techniques and experimental setups, namely, the electrochemical procedure, the type and the material of the artificial crevice formers used, and the extent of corrosion propagation before repassivation. Comparisons were limited to available data sets (samples) for each testing condition within the laboratories of the Corrosion group (CNEA, Argentina).

Table 4 shows statistics, and Figure 3 shows a box and whisker plot of the considered samples. The lower and upper values of each box correspond to the first (Q₁) and third quartile (Q₃), respectively, and the median (Q₂) is the horizontal line within each box. Whiskers extend up to 1.5 times the interquartile range (Q₃–Q₁) or at the farthest data point within that interval. Outliers, if there are any, are represented by circles.

Table 4:

Sample size, mean value, standard deviation and 95% confidence interval for E_R,CREV.

Sample ID	Sample size	Mean value (V_SCE)	Standard deviation (V)	95% confidence interval
Sample#01	13	−0.181	0.010	(−0.187, −0.175)
Sample#02	9	−0.136	0.048	(−0.172, −0.099)
Sample#03	4	−0.187	0.020	(−0.219, −0.155)
Sample#04	21	−0.151	0.013	(−0.157, −0.145)
Sample#05	20	−0.108	0.036	(−0.125, −0.091)
Sample#06	13	−0.141	0.020	(−0.153, −0.129)
Sample#07	3	−0.140	0.010	(−0.164, −0.116)
Sample#08	2	−0.184	0.019	(−0.355, −0.012)
Sample#09	3	−0.169	0.006	(−0.184, −0.155)
Sample#10	3	−0.181	0.006	(−0.195, −0.168)
Sample#11	3	−0.198	0.002	(−0.204, −0.193)
Sample#12	3	−0.138	0.010	(−0.163, −0.113)
Sample#13	3	−0.189	0.005	(−0.200, −0.177)
Sample#14	3	−0.215	0.034	(−0.300, −0.131)
Sample#15	3	−0.222	0.017	(−0.264, −0.180)
Sample#16	3	−0.271	0.018	(−0.316, −0.227)
Sample#17	3	−0.243	0.023	(−0.300, −0.186)

Figure 3:

Box and whiskers plot of E_R,CREV for Samples #01 to #07. Tests on MA alloy 22 in deaerated 1 mol/L NaCl, at 90 °C.

3.1.1 Effect of the crevice formers material

Crevice formers material is reported to affect the values of E_R,CREV in some extent (Shan and Payer 2010, Giordano et al. 2011). The effect of the crevice formers material on E_R,CREV was studied for alloy 22 (UNS N06022) in deaerated 1 mol/L NaCl at 90 °C. The PD–GS–PD method was applied with i_GS = 2 μA/cm² and t_GS = 2 h. The comparison was performed between ceramic crevice formers wrapped with 70 μm-thick PTFE tape and torqued from 5.0 to 8.0 N m (Sample#01) versus solid PTFE crevice formers torqued to 1.0 N m (Sample#02). Solid PTFE crevice formers cannot be torqued to 5.0 N m or above since they suffer severe plastic deformation. Ceramic crevice formers are usually torqued from 5.0 to 8.0 N m (the effect of applied torque is assessed later in the text). Consequently, the comparison was performed in the relevant torque conditions for each type of material.

Sample#01 has a mean of −0.181 V_SCE and a standard deviation of 0.010 V, while Sample#02 has a mean of −0.136 V_SCE and a standard deviation of 0.048 V. Ceramic crevice formers led to a more conservative and a less scattered repassivation potential with respect to solid PTFE crevice formers. CI for Sample#01 is (−0.187, −0.175) V_SCE while CI for Sample#02 is (−0.172, −0.099) V_SCE. Since CIs do not overlap, crevice formers material had a significant effect on E_R,CREV. The effect size was d = 1.465, which is very large from the statistics viewpoint (Table 3). The absolute difference between sample means was 0.045 V, which is significant but not very large from the viewpoint of corrosion resistance. As a conclusion, ceramic crevice formers wrapped with 70 μm-thick PTFE tape and torqued from 5.0 to 8.0 N m are recommended for determining E_R,CREV since they produced a more conservative and less scattered E_R,CREV than solid PTFE crevice formers.

Comparison of results of E_R,CREV values from the literature obtained by using solid PTFE crevice formers (Dunn et al. 2005a, 2005b, 2006) and ceramic crevice formers wrapped with PTFE tape (Carranza et al. 2007; Hornus et al. 2015) indicate that the former lead to higher values of E_R,CREV than the latter.

It is important that ceramic crevice formers are wrapped with thick PTFE tape (Shan and Payer 2010). Otherwise, crevice corrosion may not occur. Shan and Payer (2010) report surface roughness of the crevice former and the specimen affects the tightness of the crevice gap and contributes to the severity of the crevice. Covering the ceramic crevice former with the PTFE tape helps to reduce the crevice gap and results in faster initiation and propagation of crevice corrosion since the tape conforms to the roughness and cavities. Giordano et al. (2011) report results of PD–GS–PD tests on alloy 22 specimens with finished grindings of abrasive papers numbers 220 and 600, and 1 μm diamond paste in 1 mol/L NaCl at 90 °C. Increasing the surface roughness results in a slight increase of the repassivation potential and a moderate increase in the statistical dispersion.

3.1.2 Effect of the extent of corrosion propagation before repassivation

The total anodic charge or the maximum anodic current applied affects the extent of crevice corrosion propagation before repassivation, which in turn may affect the crevice gap and hence E_R,CREV. Pitting corrosion studies have shown that E_R,CREV decreases as the anodic charge increases until a stable and low value is attained, which is independent on the anodic charge (Sridhar and Cragnolino 1993; Thompson and Syrett 1992). In crevice corrosion studies, excessive propagation of the attack may increase the crevice gap, which in turn may produce stifling and arrest of localised corrosion (Rincón Ortíz et al. 2010). The effect of varying anodic charge was studied for alloy 22 in deaerated 1 mol/L NaCl at 90 °C by the PD–GS–PD method with ceramic crevice formers wrapped with 70 μm-thick PTFE tape and torqued from 5.0 to 8.0 N m. Comparison was performed between tests with i_GS = 2 μA/cm² (Sample#01) and i_GS = 20 μA/cm² (Sample#03, Table 4 and Figure 3). t_GS = 2 h was used in both cases. Since corrosion propagation occurred mainly in the GS stage, circulated charges were 14.4 mC/cm² and 144 mC/cm²_, respectively. The applied i_GS values are below those associated with transpassive dissolution for alloy 22 in the testing conditions (Mishra and Frankel 2008; Rodríguez et al. 2010).

Sample#01 has a mean of −0.181 V_SCE and a standard deviation of 0.010 V, while Sample#03 has a mean of −0.187 V_SCE and a standard deviation of 0.020 V. CIs are (−0.187, −0.175) V_SCE for Sample#01 and (−0.219, −0.155) V_SCE for Sample#03. The CI for Sample#03 is large and contains that of Sample#01. Since sample means are close to each other (difference of 6 mV) differences among samples are not expected to be significant. To reinforce this conclusion, Welch’s-t test was applied resulting in a p-value of 0.62. H₀ failed to be rejected, reinforcing the previous conclusion. SP of this test was 0.12 and sample sizes of 76 would be needed for SP = 0.80 and ɑ = 0.05. From a practical viewpoint, this effect is too small to be of any interest. A 6-mV difference between sample means of E_R,CREV is irrelevant. Even though E_R,CREV did not depend on i_GS, it is worth noting that i_GS = 2 μA/cm² (Sample#01) led to a narrower CI than i_GS = 20 μA/cm² (Sample#03). Consequently, PD–GS–PD tests with i_GS = 2 μA/cm² are recommended.

Mishra and Frankel (2008) obtained a mean value of E_R,CREV = −0.165 V_SCE for i_GS = 2.13 μA/cm² and E_R,CREV = −0.173 V_SCE for i_GS = 30 μA/cm² for MA alloy 22 in identical testing conditions. However, these values correspond to E_R,CREV obtained by the THE method (ASTM G192 2020) but using the definition of a crossover repassivation potential as done with the PD–GS–PD method. The reported values are slightly higher but consistent with those of Sample#01 and Sample#03 (Table 4).

3.1.3 Effect of the electrochemical procedure

Various electrochemical procedures have been applied to determine E_R,CREV of passive alloys. The effect of the electrochemical procedures PD–GS–PD, THE and CPP on E_R,CREV was assessed for alloy 22 in deaerated 1 mol/L NaCl at 90 °C, using ceramic crevice formers wrapped with 70 μm-thick PTFE tape and torqued from 5.0 to 8.0 N m. Potential staircase methods are not considered herein and their description may be found elsewhere (Anderko et al. 2008; Tormoen et al. 2010).

A comparison was performed between PD–GS–PD tests (Sample#01) and THE tests (Sample#04, Table 4 and Figure 3) with i_GS = 2 μA/cm² and t_GS = 2 h. Sample#01 has a mean of −0.181 V_SCE and a standard deviation of 0.010 V, while Sample#04 has a mean of −0.151 V_SCE and a standard deviation of 0.013 V. CIs are (−0.187, −0.175) V_SCE for Sample#01 and (−0.157, −0.145) V_SCE for Sample#04. CIs do not overlap. PD–GS–PD tests yielded more conservative results than THE tests with identical variances. The difference between sample means of 0.030 V was significant but not large from an electrochemical viewpoint, although d = 2.544 indicates a huge effect size (Table 3). Mishra and Frankel (2008) reported that E_R,CREV for MA alloy 22 in 1 mol/L NaCl at 90 °C obtained by the PD–GS–PD method is similar to that obtained by the THE method. E_R,CREV values are relatively independent of scan rate in the range from 0.167 mV/s to 0.0167 mV/s.

Another comparison was performed between PD–GS–PD tests (Sample#01) with i_GS = 2 μA/cm² and t_GS = 2 h and CPP tests (Sample#05, Table 4 and Figure 3) with i_REV = 1 mA/cm². Sample#01 has the mean, standard deviation and CI shown above. Sample#05 has a mean of −0.108 V_SCE, a standard deviation of 0.036 V, and its CI is (−0.125, −0.091) V_SCE. CIs do not overlap. The difference between sample means of 0.073 V was significant and d = 2.528 indicates a huge effect size (Table 3).

Finally, if we compare THE tests (Sample#04) with i_GS = 2 μA/cm² and t_GS = 2 h and CPP tests with i_REV = 1 mA/cm² (Sample#05), CIs do not overlap. The difference between sample means of 0.043 V was significant and d = 1.594 indicates a very large effect size (Table 3).

In the selected testing conditions, PD–GS–PD and THE methods produced more repeatable E_R,CREV than CPP. Comparison among E_R,CREV obtained by PD–GS–PD, THE and CPP methods should be performed with due care, since there is a bias among E_R,CREV determined by these methods, at least in the present testing conditions. In these conditions, the bias was 30 mV between PD–GS–PD and THE methods, 43 mV between THE and CPP methods, and 73 mV between PD–GS–PD and CPP methods. PD–GS–PD and CPP methods produced the most and the least conservative results, respectively.

In the literature, E_R,CREV determined by CPP method is reported to have an error of ±40 mV (Anderko et al. 2008; Sridhar and Cragnolino 1993). This result agrees with the standard deviation of 36 mV obtained for Sample#05 (Table 4). Anderko et al. (2008) indicate the potential staircase methods have a lower associated dispersion of ±10 to ±20 mV depending on the chloride concentration.

3.1.3.1 THE method from ASTM G192 round robin

ASTM G192 standard reports results from a round robin which involved 5 laboratories worldwide that provided from 3 to 6 results each in testing conditions similar to those studied herein (Sample#04, Table 4 and Figure 3). Reported reproducibility statistics are as follows, a mean value of E_R,CREV = −0.154 V_SCE, standard deviation of 0.011 V, and 95% CI of (−0.184, −0.124) V_SCE (ASTM G192 2020). These results align with Sample#04, which involved a larger sample size but a single operator (Table 4).

3.1.4 Effect of the applied torque to ceramic crevice formers wrapped with PTFE tape

MCA crevice formers specified in ASTM G192 standard are used along with fasteners to secure the two crevice formers on each side of the test specimen. The torque applied to these fasteners affects the geometry of the crevice. The effect of applied torque was studied by THE tests on alloy 22 in deaerated 1 mol/L NaCl at 90 °C (i_GS = 2 μA/cm², t_GS = 2 h) with ceramic crevice formers wrapped with 70 μm-thick PTFE tape. Specimens were torqued to 8.0 and 5.0 N m (Sample#06 and Sample#07, respectively, Table 4 and Figure 3). This comparison is relevant since reducing applied torque from 8.0 to 5.0 N m was observed to diminish the fracture occurrence of ceramic crevice formers.

Sample#06 has a mean of −0.141 V_SCE and a standard deviation of 0.020 V, while Sample#07 has a mean of −0.140 V_SCE and a standard deviation of 0.010 V. CIs are (−0.153, −0.129) V_SCE for Sample#06 and (−0.164, −0.116) V_SCE for Sample#07. Sample means are almost identical and CIs overlap. To reinforce the conclusion that both samples come from the same population, Student’s-t test was performed, which produced p-value = 0.93. Although SP was as low as 0.051, the effect size of d = 0.060 was also very small. Such a small effect is deemed irrelevant given that the means and standard deviation of both samples are practically the same. As a conclusion, reducing the applied torque to ceramic crevice formers from 8.0 to 5.0 N m did not affect the value of E_R,CREV. However, there is evidence that applied torque values below 2.0 N m produce an increase of E_R,CREV (Giordano et al. 2011).

3.2 Metallurgical and environmental variables

Variation of metallurgical (or internal) and environmental (or external) conditions may change the crevice corrosion susceptibility of material as given by E_R,CREV. Concerning metallurgical variables, the effects of (1) various heat treatments performed on alloys 22, C-22HS, 600 and 690, and (2) small and large changes in the chemical composition of alloys are shown. The study of environmental variables includes the effects of (1) temperature, (2) chloride concentration and (3) inhibitor concentration. Generally, the considered samples had a size of 2 or 3 since duplicates or triplicates are the usual standard for determining E_R,CREV. Consequently, the statistical power of any hypothesis test was low unless the size effect (i.e. the magnitude of the change of E_R,CREV) was significantly large. Comparisons herein were made among collected samples that were not originally meant for this purpose.

3.2.1 Effect of thermal ageing

Thermal ageing performed to nickel alloys may be detrimental to their crevice corrosion resistance. Depending on the particular alloy composition, temperature and time of exposure secondary phases may precipitate which in turn may affect E_R,CREV in the corrosive environment (Rebak 2000). Precipitation of secondary phases may occur due to welding processes, heat treatments, non-desired temperature excursions, etc. Collected data of alloys with varying heat treatments include alloy 22 under five different thermal ageing treatments, alloy C-22HS in mill-annealed and age-hardened conditions, and alloys 600 and 690 in mill-annealed condition and a thermal ageing for 10 h at 716 °C plus water-quenched.

Table 4 shows statistics and Figure 4 shows a box and whisker plot of the considered samples. Samples #08 to #11 correspond to different thermal ageing treatments of alloy 22. Sample#12 and Sample#13 correspond to MA and age-hardened alloy C-22HS (UNS N07022), respectively (Table 1). Both alloys belong to the Ni-Cr-Mo family. Thermal ageing of Ni-Cr-Mo alloys leads to various microstructure changes depending on the temperature range and exposure time at temperature. A long-range ordering (LRO) reaction occurs from 350 to 600 °C, producing an ordered Ni₂(Cr,Mo) phase (Tawancy et al. 1983; Rebak et al. 2000). Tetrahedral close-packed (TCP) phases (like μ, σ and P) precipitate, starting at grain boundaries, in the range from 600 to 1100 °C. Since TCP phases are rich in Mo and Cr, zones adjacent to the precipitates may suffer Cr and/or Mo depletion (Heubner et al. 1989; Tawancy 1996; Tawancy et al. 1983).

Figure 4:

Box and whiskers plot of E_R,CREV for Sample#01 and #08 to #13. Data from PD–GS–PD tests in deaerated 1 mol/L NaCl, at 90 °C.

Thermal ageing treatments of Sample#08, #09 and #10 produced various degrees of TCP precipitation. Thermal ageing treatment of Sample#11 produced LRO transformation. All these samples are compared with Sample#01, which corresponds to a fully solubilized alloy 22, all other variables being the same.

Comparison of Sample#8 with Sample#01 indicated that mean values are very close (−0.184 vs. −0.181 V_SCE), standard deviations are equivalent according to Levene’s test, and CIs overlap. However, CI of Sample#08 is very large, and the sample size is very small (only 2 data). This comparison did not find differences on E_R,CREV of MA versus aged material for 10 h at 760 °C since a statistical power of 0.059 is very low. However, the small effect size (d = 0.229) suggested this thermal ageing was not detrimental for the corrosion resistance of alloy 22.

Sample#09 shows a mean value of −0.169 V_SCE, a standard deviation of 0.006 V and CI of (−0.184, −0.155). CIs of Sample#01 and #09 overlap and Student’s-t test resulted in p-value = 0.063 which failed to reject H₀. Consequently, there are no differences among samples. The SP of the test was 0.47, the effect size was very large (d = 1.29) and at least 11 measurements would be needed to increase it up to 0.80. The aged material showed a small increase of E_R,CREV, which means a better resistance. There is no further interest in increasing the sample size since the effect was not detrimental. Sample#10 shows a mean value of −0.181 V_SCE, a standard deviation of 0.006 V and CI of (−0.195, −0.168). CIs of Sample#01 and #10 overlap and Student’s-t test failed to reject H₀ (p-value = 0.97). The SP of the test was very low (0.050) but the effect size was small (d = 0.028) and a huge sample size would be needed for SP = 0.80. Consequently, thermal ageing treatments of 5 and 30 min at 870 °C were not detrimental for the corrosion resistance of alloy 22.

Sample#11 shows a mean value of −0.198 V_SCE, a standard deviation of 0.002 V and CI of (−0.204, −0.193). CIs of Sample#01 and #11 do not overlap and thus thermal ageing of 1000 h at 538 °C produced a decrease of E_R,CREV of alloy 22. Although the effect size of d = 1.95 is very large for statistics, a 17 mV decrease is small for E_R,CREV from the viewpoint of corrosion.

Overall, none of the considered thermal ageing treatments performed to alloy 22 significantly changed its crevice corrosion resistance, as given by E_R,CREV. This observation agrees with results reported for MA, as-welded and thermally aged alloy 22 (Evans et al. 2005; Mishra and Frankel 2008). However, precipitation of TCP phases in alloy 22 is reported to increase the crevice corrosion susceptibility of alloy 22 (Dunn et al. 2005a, 2006). Differences may be ascribed to the different methods, especially crevice former materials, used in the various studies. Evans et al. (2005), Mishra and Frankel (2008) and the present study used PTFE-wrapped ceramic crevice formers while Dunn et al. (2005a, 2005b, 2006 used solid PTFE crevice formers. As stated above, results obtained using different crevice former materials are not comparable.

It is worth noting that E_R,CREV does not reflect any change in the cathodic kinetics that might be affected by the various heat treatments. Changes in the cathodic kinetics may affect the open circuit potential which should remain below E_R,CREV for avoiding crevice corrosion occurrence (Carranza et al. 2008a, 2008b). External cathodic reactions may limit crevice corrosion propagation (Cui et al. 2005). Enhancing the internal cathodic reaction of hydrogen ion reduction may result in the increase of pH within the crevice, which in turn leads to alloy repassivation (Turnbull 1998). The oxide film that forms on Ni-Cr-Mo alloys at high temperature also affects the open circuit potential producing an ennoblement (Dunn et al. 2005a; Rebak et al. 2006; Zadorozne et al. 2010).

Alloy C-22HS is age-hardened by a two-step heat treatment of 16 h at 705 °C followed by furnace cool to 605 °C, then kept for 32 h at 605 °C, and finally, it is air-cooled. The age-hardening treatment produces LRO transformation but also may lead to precipitation of TCP phases (Lu et al. 2007). Sample#12 shows a mean value of −0.138 V_SCE, a standard deviation of 0.010 V and CI of (−0.163, −0.113), while Sample#13 (age-hardened material) shows a mean value of −0.189 V_SCE, a standard deviation of 0.005 V and CI of (−0.200, −0.177). CIs do not overlap. The age hardening of alloy C-22HS produced a decrease in its E_R,CREV, which is a huge effect size (d = 6.5) and a significant drop in the potential scale (51 mV). Age-hardened alloy C-22HS was less corrosion resistant than MA alloy C-22HS, but as corrosion resistant as MA alloy 22 (Figure 6).

Table 4 shows statistics of samples #14 to #17. Samples #14 and #16 correspond to alloys 600 (UNS N06600) and 690 (UNS N06690), respectively, in the solution-annealed condition. Samples #15 and #17 correspond to alloys 600 and 690, respectively, thermally-aged for 10 h at 716 °C and then water-quenched. This latter metallurgical condition leads to the highest degree of sensitization of alloy 690 and it is the specified condition for the steam generator tubes of nuclear power plants (Gonzalez et al. 2018). Comparisons between samples #14 and #15, and between samples #16 and #17 allow to evaluate the effect of thermal ageing on the crevice corrosion resistances of alloys 600 and 690, respectively.

Sample#14 has a mean value of −0.215 V_SCE, a standard deviation of 0.034 V and CI of (−0.300, −0.131) and Sample#15 has a mean value of −0.222 V_SCE, a standard deviation of 0.017 V and CI of (−0.264, −0.180). CIs overlap and Student-t test failed to reject H₀ with p-value = 0.78. SP = 0.057 was very low but the effect size was small to moderate (d = 0.25). A huge sample size would be needed for having an SP ≥ 0.80. The comparison clearly indicated that thermal ageing did not produce a significant reduction of the crevice corrosion resistance of alloy 600 in the tested conditions. Tormoen et al. (2010) reported a 100-to-150 mV decrease of E_R,CREV of alloy 600 for initial aging times at 700 °C for alloy 600. Healing was observed within 24 h of thermal ageing. However, these tests were performed by the CPP or potential staircase methods using solid PTFE crevice formers torqued at 0.14 N m. Consequently, E_R,CREV values are not comparable with those from present work.

Sample#16 has a mean value of −0.271 V_SCE, a standard deviation of 0.018 V and CI of (−0.316, −0.227) and Sample#17 has a mean value of −0.243 V_SCE, a standard deviation of 0.023 V and CI of (−0.300, −0.186). The mean value of E_R,CREV for the thermally-aged alloy was higher than that of the solution-annealed material. CIs overlap and Student’s-t test failed to reject H₀ with p-value = 0.16. SP = 0.26 was obtained and the effect size was d = 1.39. Increasing the sample size to 10 would lead to an SP ≥ 0.80. Corrosion resistance of alloy 690 was not affected by the thermal ageing treatment in the tested conditions. This result was expected since this thermal ageing only leads to a 7 wt.%-reduction of the chromium content at the grain boundaries, from 29 to 22 wt% (Gonzalez et al. 2018).

3.2.2 Effect of chemical composition of the alloy

Tested alloys are nickel or iron based which develop chromium-rich passive films. CRAs contain alloying elements such as molybdenum, tungsten, copper, tantalum and niobium that enhances their resistance to localised corrosion (Rebak 2000). E_R,CREV of various alloys was determined in 1 mol/L NaCl at 60 °C (Figure 5). Alloys 600, 690 and 800 (UNS N08800) are HTAs. As expected, they showed lower E_R,CREV with regard to the CRAs, which contains beneficial alloying elements (Table 2). Although HTAs and CRAs are meant for different applications PRE is used for both types of alloys. HTAs may suffer crevice corrosion during downtime of equipment if humidity condensates at low temperatures. Alloy 690 showed the lowest resistance, while alloy 800 showed the highest resistance to crevice corrosion among HTAs (disregarding alloy 625, which is both HTA and CRA). E_R,CREV of alloy 800 showed CIs that overlap with some of the CRAs in the tested conditions (Figure 7). Student’s-t test resulted in differences among E_R,CREV of alloy 800 and alloys G-30 (UNS N06030) and 625 (UNS N06625), but no differences between E_R,CREV of alloy 800 and that of alloy G-35 (UNS N06035).

Figure 5:

Mean values and 95% confidence intervals of E_R,CREV as a function of PRE obtained from PD–GS–PD tests in deaerated 1 mol/L NaCl, at 60 °C from this work and results from single tests of Mishra and Shoesmith (2014).

Figure 6:

Mean values and 95% confidence intervals of E_R,CREV as a function of temperature. Data from PD–GS–PD tests in deaerated chloride solutions. Some overlapping data are slightly displaced to the right for better visualization.

Figure 7:

Mean values and 95% confidence intervals of E_R,CREV as a function of chloride concentration. Data from PD–GS–PD tests in deaerated solutions.

Comparison of E_R,CREV led to cases where alloys showed significantly different E_R,CREV (CIs do not overlap) and their corrosion resistances could be distinguished. However, in other cases, alloys showed similar E_R,CREV (CIs overlap) even when chemical compositions were very different (e.g., alloy 800 vs. alloy G-35).

In general, for CRAs a higher E_R,CREV was correlated with a higher PRE (Table 2, Equation (1)). However, E_R,CREV did not increase with PRE for HTAs (Figure 5). Analyses of the correlation of E_R,CREV versus PRE can be found elsewhere (Mishra and Shoesmith 2014; Zadorozne et al. 2012). PRE is determined by the contents of chromium, molybdenum and tungsten. Chromium improves the passive film stability, while molybdenum improves the film stability against localised breakdown and promotes alloy repassivation. Tungsten has an effect similar to that of molybdenum (Haugan et al. 2017; Mishra and Shoesmith 2014). HTAs contain chromium but neither molybdenum nor tungsten, with the exception of alloy 625 which contains 9 wt% Mo. CRAs contain chromium, molybdenum and, some of them, tungsten.

Figure 5 shows data from Mishra and Shoesmith (2014) obtained by PD–GS–PD tests with t_GS = 40 h in step 2 (instead of t_GS = 2 h as in the present work) with a variety of Ni-Cr-Mo-(W) alloys in 1 mol/L NaCl, at 60 °C. They reported E_R,CREV for alloys 625, 22, C-4 (UNS N06455), C-276 (UNS N10276), 59 (UNS N06059) and C-2000 (UNS N06200). Their results show lower E_R,CREV values than those from this work in nearly identical testing conditions. The longer holding period for step 2 may have resulted in more conservative E_R,CREV values.

3.2.3 Effect of temperature

E _R,CREV of nickel alloys is reported to decrease with increasing temperatures in a linear fashion with slopes from 0.002 to 0.009 V/°C (Hornus et al. 2015). The absolute value of the E_R,CREV versus temperature slope decreases as the chloride solution becomes more concentrated. In some cases, E_R,CREV saturates at high temperatures. The lower bound for E_R,CREV is the corrosion potential of the alloy in the concentrated acidic solution developed within the crevice (Hornus et al. 2015). If the considered temperature range is sufficiently large, the decreasing trend of E_R,CREV is easily observed. It is worth studying when a temperature change is large enough to produce a detectable change in E_R,CREV.

Figure 6 shows mean values and 95% confidence intervals of E_R,CREV as a function of temperature for alloy 625 in 0.1 mol/L NaCl and for alloy C-22HS in 5 mol/L CaCl₂. The confidence interval for Sample#34 is not shown since sample size of 2 led to an excessively large interval. All the other considered samples had a size of 3. Note that the CIs in the more concentrated chloride solution were generally smaller than the CIs in the dilute chloride solution for the same temperatures. E_R,CREV of alloy 625 in 0.1 mol/L NaCl was determined in the range from 20 °C to 90 °C, every 10 °C. In the range from 40 to 70 °C, hypothesis tests indicated that a 10 °C drop in temperature produced a significant change of E_R,CREV (Figure 6). The reported value of the slope of E_R,CREV versus temperature is 0.004 V/°C. A detailed statistical analysis of this linear correlation for Ni-Cr-Mo alloys can be found elsewhere (Hornus et al. 2015).

E _R,CREV of alloy C-22HS in 5 mol/L CaCl₂ was determined in the range from 40 °C to 117 °C. In this case, the reported slope of E_R,CREV versus temperature is 0.002 V/°C (Hornus et al. 2015). With a few exceptions, a 10 °C shift in temperature did not produce a considerable change in E_R,CREV. CIs of E_R,CREV for consecutive temperatures overlap (Figure 6). Hypothesis tests only helped to distinguish between Sample#42 (50 °C) and Sample#43 (60 °C), and between Sample#44 (70 °C) and Sample#45 (80 °C). A 20 °C shift in temperature led to significant changes of E_R,CREV in most cases.

3.2.4 Effect of chloride concentration of the environment

E _R,CREV decreases linearly as the logarithm of chloride concentration ([Cl⁻]) of the environment increases. The slope of this linear relationship depends on the considered alloy and temperature. Free chlorides affect E_R,CREV while chloride complexes do not have a significant effect. There is evidence that at very low chloride concentrations the slope of E_R,CREV versus log[Cl⁻] is steeper (Anderko et al. 2004, 2008). For the CRAs alloys 22 and C-22HS, the linear dependence is as high as 0.160 V/log([Cl⁻]) at 40 °C, but it becomes nil at 90 °C (Hornus et al. 2015). On the other hand, for the HTA alloy 800 the slope remains in the narrow range from 0.105 to 0.117 V/log([Cl⁻]) at 30, 60 and 90 °C (Maristany et al. 2016). A detailed statistical analysis of these linear correlations can be found elsewhere (Hornus et al. 2015; Maristany et al. 2016).

It is generally assumed that a 10-fold variation of [Cl⁻] leads to significant changes of E_R,CREV. To check this assumption, E_R,CREV of alloy 690 was studied at 60 °C, and E_R,CREV of alloys 800 and HYBRID-BC1 were studied at 90 °C. Figure 7 shows a the mean values and 95% confidence intervals of E_R,CREV as a function of chloride concentration for the considered cases. All the samples had a size of 3. E_R,CREV of alloy 690 decreases at 0.057 V/log([Cl⁻]) for a temperature of 60 °C (Maristany et al. 2016). CIs of E_R,CREV for chloride concentrations of 0.01 mol/L (Sample#50) and 0.1 mol/L (Sample#51) overlap. However, Student’s-t test helped to distinguish between Sample#50 and Sample#51, rejecting H₀ with p-value = 0.020. The same distinction was achieved when comparing Sample#51 and Sample#52 ([Cl⁻] = 1 mol/L), rejecting H₀ with p-value = 0.044. Sample#52 and Sample#53 ([Cl⁻] = 5 mol/L) could not be distinguished by hypothesis test. However, this comparison involved a 5-fold variation of chloride concentration. E_R,CREV of alloy 800 decreases at 0.113 V/log([Cl⁻]) for a temperature of 90 °C (Maristany et al. 2016). This high slope of E_R,CREV versus log([Cl⁻]) allowed the distinction among E_R,CREV at the considered chloride concentrations. E_R,CREV of alloy HYBRID-BC1 (UNS N10362) decreases at 0.068 V/log([Cl⁻]) for 90 °C (Hornus et al. 2015). CIs of E_R,CREV for chloride concentrations of 0.01 mol/L (Sample#58) and 0.1 mol/L (Sample#59) overlap but Student’s-t test helped to distinguish between samples (H₀ was rejected with p-value = 0.010). CIs of Sample#59 and Sample#60 ([Cl⁻] = 10 mol/L) do not overlap. Overall in the analysed cases, a 10-fold variation of the chloride concentration of the environment led to significant changes of E_R,CREV.

3.2.5 Effect of inhibitors

Crevice corrosion inhibitors must produce a significant increase of E_R,CREV of the alloy when present in a chloride solution and/or lead to the complete inhibition of crevice corrosion when added above a threshold concentration (Rodríguez 2012). When a complete inhibition of crevice corrosion occurs, the crossover potential between the forward and reverse scans in the PD–GS–PD method (Figure 1c), if there is any, is not related to crevice corrosion. The observed increase of current density at high potentials is due to oxidative dissolution of the chromium-rich passive film and/or oxygen evolution. It means that E_R,CREV does not exist in conditions where crevice corrosion is absent at any potential. The threshold inhibitor concentration that leads to the complete inhibition of crevice corrosion is generally indicated as a critical inhibitor-to-chloride molar ratio (R_CRIT). It is worth studying whether an inhibitor has any effect below R_CRIT or not. Figure 8 shows data of E_R,CREV for alloy 22 considering various additions of NaNO₃ to 1 mol/L NaCl solutions at 90 °C, and for additions of Na₂SO₄ and Na₂WO₄ to 0.1 mol/L NaCl solutions at 90 °C. Unfortunately, data gathered for these comparisons only involve samples of size 2. Consequently, CIs for E_R,CREV are very large and overlap, and hypothesis tests have a very low statistical power.

Figure 8:

Mean values and 95% confidence intervals of E_R,CREV as a function of the inhibitor-to-chloride concentration ratio. Data from PD–GS–PD tests in deaerated solutions from this work and from Mishra and Frankel (2008). Some overlapping data are slightly displaced to the right for better visualization.

Considering a 1 mol/L NaCl solution (Sample#01), an addition of 0.01 mol/L NaNO₃ (Sample#61) did not produce a significant effect on E_R,CREV. However, additions of 0.02, 0.05 and 0.1 mol/L NaNO₃ (Samples #62-#64) led to a significant increase of E_R,CREV (Figure 8). The nitrate concentration for a complete inhibition of crevice corrosion on alloy 22 in 1 mol/L NaCl is 0.2 mol/L (R_CRIT = 0.2) (Rincón Ortíz et al. 2013). Other researchers reported similar results: R_CRIT = 0.2 (Mishra and Frankel 2008) and R_CRIT = 0.1–0.15 (Dunn et al. 2005b). Figure 8 shows the range of E_R,CREV reported by Mishra and Frankel (2008) for NaNO₃ additions to 1–4 mol/L NaCl solutions, other testing conditions being identical to the present ones. Their results agree with those of present work. For sulphate and tungstate additions to a 0.1 mol/L NaCl solution (Sample#65), comparison was complicated by the small size of samples. Sample#66, corresponding to an addition of 0.01 mol/L Na₂SO₄, produced a significant increase of E_R,CREV with respect to Sample#65. However, further sulphate additions (Sample#67 and Sample#68) resulted in no beneficial effect (Figure 8). The sulphate concentration for a complete inhibition of crevice corrosion on alloy 22 in 0.1 mol/L NaCl is 0.1 mol/L (R_CRIT = 1) (Rincón Ortíz et al. 2013). Mishra and Frankel (2008) reported R_CRIT = 0.8 for sulphate in identical testing conditions. Tungstate additions to a 0.1 mol/L NaCl solution only led to a significant increase of E_R,CREV for the concentration 0.2 mol/L (Sample #71, Figure 8). Additions of Na₂WO₄ to 0.1 mol/L NaCl lead to saturation of the salt without reaching a complete inhibition of crevice corrosion of alloy 22 at 90 °C (Rincón Ortíz et al. 2013).

3.3 Comparison of E_R,CREV: when is a change significant?

In the previous sections, collected E_R,CREV data from tests on CRAs and HTAs in various metallurgical and environmental conditions were analysed to determine whether changes in these variables had a significant effect on E_R,CREV or not. The flowchart in Figure 2 was used as a guideline. When two samples of previously collected values of E_R,CREV are considered, there may be several issues that make comparison difficult. Sample sizes of 2 or 3, customary for E_R,CREV tests, may not be appropriate since they were not meant for a given statistical test. Moreover, sample sizes may be different. Consequently, the statistical power of hypothesis tests are generally low, which means that the effect of the studied variable may not be detected. On the other hand, large sample sizes and/or low data dispersion (small standard deviation) may result in statistically significant differences between samples, even though the absolute difference of the means of E_R,CREV is slight (a few millivolts). Note that a statistically significant difference between samples does not necessarily imply that the difference is significant from the point of view of corrosion resistance. Whenever a statistically significant difference between two samples is found, relevance from the point of view of corrosion resistance has to be verified. The suggested threshold values for classifying effect sizes (Cohen 2013; Sawilowsky 2009) are a good starting point, but they may be misleading if strictly applied to E_R,CREV. The passive range of CRAs is typically 1.0 V wide (Rodríguez et al. 2010). The range of E_R,CREV values considering all the alloys and testing conditions analysed in this work lies from −0.420 to 0.197 V_SCE. Let us define ΔE_R,CREV as the difference between the E_R,CREV mean values of two samples. The range for ΔE_R,CREV lies approximately from 0 to 0.600 V. The lower bound for ΔE_R,CREV may be set at 10 mV since differences of less than 10 mV may be ascribed to the drift of reference electrodes. Table 3 states the proposed classification of the effect size for ΔE_R,CREV from the point of view of corrosion resistance. This classification is not intended to replace that based on Cohen’s d parameter but to be complementary, and it is meant to be applied after the statistical analysis described in Figure 2 is performed.

3.3.1 Suggested guideline for a planned comparison of E_R,CREV

Figure 9 is proposed as a guideline for a planned comparison of E_R,CREV between two samples. The first step is to state clearly the objective of the comparison, that is, which effect on E_R,CREV is to be assessed. The second step is an estimation of the effect size. The effect size of a population (θ) is obtained according to Equation (4), where μ₁ and μ₂ are the means of the two populations and σ is the standard deviation of either or both populations.

(4)θ=(μ1−μ2)/σ

Figure 9:

Suggested guide for a planned comparison of E_R,CREV.

In the present case, μ₁ − μ₂ can be assimilated to the minimum value of ΔE_R,CREV that we consider significant in our comparison from the point of view of corrosion resistance. The value of σ can be estimated from the standard deviation of all the samples collected in this work.

Figure 10 shows a histogram of the samples standard deviation, including PD–GS–PD tests performed by 6 operators in 239 different testing conditions performed at the at the laboratories of the Corrosion group (CNEA). Since the distribution is highly asymmetric, the median (0.015 V) is more representative than the mean (0.021 V). Figure 11 shows a box and whiskers plot of the standard deviation of the sample discriminating the six operators included in this study. Standard deviation was operator-dependent, but the asymmetric distribution, showing the median is lower than the mean, is observed for all the operators. The medians of the standard deviation of the samples for the six operators were in the range from 0.006 to 0.030 V. Some of the outliers in Figure 11 may be due to testing conditions where crevice corrosion occurred in some tests but was absent in other tests. This is a common observation when tests include highly resistant alloys (such as alloy HYBRID-BC1), low temperatures, low chloride concentrations and/or high inhibitor concentrations. Crossover potentials from tests where crevice corrosion does not occur are not true E_R,CREV and must be discarded. However, there is some uncertainty when the localised attack is shallow and/or corroded spots are scarce. Standard deviation is also a function of the testing conditions, being higher for less severe testing conditions.

Figure 10:

Histogram of the standard deviation of the sample in all the PD–GS–PD tests.

Figure 11:

Box and whiskers plot of the standard deviation of the sample for the six operators included in this study.

For the purpose of estimating the size effect, μ₁ − μ₂ = 0.030 V and σ = 0.015 V are considered (Equation (4)). This results in θ = 2, which is a huge effect size based on d, but a small effect size from the point of view of corrosion (Table 3). The next step in the flowchart (Figure 9) is estimating the sample size for a size effect of d = 2, a confidence level of 95% and a statistical power of 80%. Estimation results in a sample size of 5. The next step is performing the PD–GS–PD tests five times and obtaining E_R,CREV for the two testing conditions to be compared. Then CIs are obtained. If they do not overlap, there are significant differences among samples. Otherwise, a hypothesis test has to be applied. Student’s-t test is appropriate since equal variances were considered in the development of this procedure. Rejection of H₀ will indicate that samples are different. Otherwise, no significant differences are obtained. The flowchart in Figure 9 helps to solve the issue of statistical significance. However, whenever statistical differences are found among samples, the criteria established in Table 3 will help to estimate the size effect from the point of view of corrosion resistance.

According to the analysis above, a sample size of 5 will allow a comparison of E_R,CREV between two samples ensuring type I error (rejecting H₀ when it is true) is bound to 5%, and Type II error (failing to reject H₀ when it is not true) is bound to 20%. For sample sizes of 4, 3 and 2 the statistical power diminishes to 65.7%, 43.6% and 21.8%, respectively. If we consider a non-skilful operator or a mildly corrosive testing condition, then σ = 0.030 V applies. In such conditions, a size effect θ = 2 is obtained for μ₁ − μ₂ = 0.060 V. This implies that a sample size of 5 would only allow distinguishing differences of 60 mV or higher between E_R,CREV means with a confidence level of 95% and a statistical power of 80%.

4 Conclusions

An extensive database of crevice corrosion repassivation potentials was analysed applying statistical tools to decide whether the effects of various testing conditions were significant or not. Repeatability, which is the precision of test results from tests conducted on identical material by the same test method in a single laboratory, was assessed. The main findings regarding variables that allegedly affect the outcome of testing techniques are as follows.

Ceramic crevice formers wrapped with 70 μm-thick PTFE tape and torqued from 5.0 to 8.0 N m are recommended since they produced more conservative and less scattered results than solid PTFE crevice formers.
Reducing the applied torque to ceramic crevice formers from 8.0 to 5.0 N m did not affect the outcome of tests, and it helped to diminish the fracture occurrence of ceramic crevice formers.
Anodic current densities of 2 and 20 μA/cm² led to similar results in THE tests, but the lower current density improved test repeatability.
Since there was a bias among repassivation potentials determined by different methods in particular testing conditions, comparison among results by PD–GS–PD, THE and CPP methods should be performed with due care. Results from PD–GS–PD and THE methods were more repeatable than those from CPP method.

The main findings regarding metallurgical and environmental variables are as follows.

Various thermal ageing treatments performed on alloy 22 did not significantly change its crevice corrosion resistance. However, the age-hardening treatment performed to alloy C-22HS was detrimental. HTA alloys 600 and 690 were not negatively affected by the thermal-ageing treatment that is customary for steam generator tubing.
Tested alloys showed similar or different behaviours depending on the testing conditions. Significant variations in the chemical composition of alloys did not always result in different corrosion resistance.
A 10 °C variation of the temperature led to significantly different repassivation potentials for tests in 0.1 mol/L chloride solutions. However, a 20 °C variation of the temperature was needed to observe significant changes in tests in 10 mol/L chloride solutions.
A 10-fold variation of the chloride concentration of the environment led to significant changes in the repassivation potential in most cases.
Inhibitors showed some effect below the molar concentration ratio associated with complete inhibition. However, the extent of such effect is not clear yet since small sample sizes (n = 2) were considered.

Practical applications of this statistical analysis includes guidelines for comparing crevice corrosion repassivation potentials for cases where previously collected data are available and when new data are generated mainly for comparison. Based on the analysis of collected data, a sample size of 5 (or larger) is recommended for a planned comparison of any effect on the repassivation potential.

Corresponding author: Martín A. Rodríguez, Instituto Sabato, UNSAM/CNEA, Av. Gral. Paz 1499, San Martín, Buenos AiresB1650KNA, Argentina; National Commission of Atomic Energy of Argentina, Av. Gral. Paz 1499, San Martín, Buenos AiresB1650KNA, Argentina; and National Scientific and Technical Research Council, Godoy Cruz 2290, Ciudad Autónoma de Buenos AiresC1425FQB, Argentina, E-mail: maalrodr@cnea.gov.ar

Funding source: Consejo Nacional de Investigaciones Científicas y Técnicas

Award Identifier / Grant number: PIP CONICET 2021-23 11220200101057CO

Funding source: Agencia Nacional de Promoción de la Investigación, el Desarrollo Tecnológico y la Innovación

Award Identifier / Grant number: PICT-2020-SERIEA-00149

Acknowledgments

The author is grateful to Dr. Edgar C. Hornus and Dr. Mariano A. Kappes for reading the draft of this article and making valuable comments and suggestions. Electrochemical tests were performed by C. Mabel Giordano, Dr. Natalia S. Zadorozne, Dr. Mauricio Rincón Ortíz, Eng. Santiago Sosa Haudet, Dr. Edgar C. Hornus and MSc. H. Guillermo Maristany who are credited in the corresponding articles. In loving memory of my wife, Erica M. Welti (1975–2022).

Author contributions: All the authors have accepted responsibility for the entire content of this submitted manuscript and approved submission.
Research funding: This work was supported by Agencia Nacional de Promoción de la Investigación, el Desarrollo Tecnológico y la Innovación of Argentina (grant PICT-2020-SERIEA-00149), National Scientific and Technical Research Council of Argentina (grant PIP CONICET 2021-23 11220200101057CO), and National Commission of Atomic Energy of Argentina.
Conflicts of interest: The author declares that he has no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

References

Anderko, A., Sridhar, N., and Dunn, D.S. (2004). A general model for the repassivation potential as a function of multiple aqueous solution species. Corrosion Sci. 46: 1583–1612, https://doi.org/10.1016/j.corsci.2003.10.002.Search in Google Scholar

Anderko, A., Sridhar, N., Jakab, M.A., and Tormoen, G. (2008). A general model for the repassivation potential as a function of multiple aqueous species. 2. Effect of oxyanions on localized corrosion of Fe–Ni–Cr–Mo–W–N alloys. Corrosion Sci. 50: 3629–3647, https://doi.org/10.1016/j.corsci.2008.08.046.Search in Google Scholar

ASTM E691 (2021). ASTM E691-21 standard practice for conducting an interlaboratory study to determine the precision of a test method. In: ASTM book of standards, Vol. 14.01. West Conshohocken, PA, USA: American Society for Testing and Materials.Search in Google Scholar

ASTM G48 (2020). ASTM G48-11(2020) standard test methods for pitting and crevice corrosion resistance of stainless steels and related alloys by use of ferric chloride solution. In: ASTM book of standards, Vol. 03.02. West Conshohocken, PA, USA: American Society for Testing and Materials.Search in Google Scholar

ASTM G61 (2018). ASTM G61-86(2018) standard test method for conducting cyclic potentiodynamic polarization measurements for localized corrosion susceptibility of iron-, nickel-, or cobalt-based alloys. West Conshohocken, PA, USA: American Society for Testing and Materials.Search in Google Scholar

ASTM G78 (2020). ASTM G78-20 standard guide for crevice corrosion testing of iron-base and nickel-base stainless alloys in seawater and other chloride-containing aqueous environments. West Conshohocken, PA, USA: American Society for Testing and Materials.Search in Google Scholar

ASTM G192 (2020). ASTM G192-08(2020) standard test method for determining the crevice repassivation potential of corrosion-resistant alloys using a potentiodynamic-galvanostatic-potentiostatic technique. West Conshohocken, PA, USA: ASTM International.Search in Google Scholar

Carranza, R.M. (2008). The crevice corrosion of alloy 22 in the Yucca Mountain nuclear waste repository. J. Miner. Met. Mater. Soc. 60: 58–65, https://doi.org/10.1007/s11837-008-0009-z.Search in Google Scholar

Carranza, R.M. and Rodríguez, M.A. (2017). Crevice corrosion of nickel-based alloys considered as engineering barriers of geological repositories. npj Mater. Degrad. 1: 9, https://doi.org/10.1038/s41529-017-0010-5.Search in Google Scholar

Carranza, R.M., Rodríguez, M.A., and Rebak, R.B. (2007). Effect of fluoride ions on crevice corrosion and passive behavior of alloy 22 in hot chloride solutions. Corrosion 63: 480–490, https://doi.org/10.5006/1.3278400.Search in Google Scholar

Carranza, R.M., Giordano, C.M., Rodríguez, M.A., and Rebak, R.B. (2008a). Effect of organic acid additions on the general and localized corrosion susceptibility of alloy 22 in chloride solutions. In: Corrosion 2008. Houston, TX, USA: NACE International, Paper no. 08578.10.1557/PROC-1107-511Search in Google Scholar

Carranza, R.M., Rodríguez, M.A., and Rebak, R.B. (2008b). Anodic and cathodic behavior of mill annealed and topologically closed packed alloy 22 in chloride solutions. In: Corrosion 2008. Houston, TX, USA: NACE International, Paper no. 08579.10.5006/C2008-08579Search in Google Scholar

Cohen, J. (2013). Statistical power analysis for the behavioral sciences, 2nd ed. Taylor & Francis, New York.10.4324/9780203771587Search in Google Scholar

Cui, F., Presuel-Moreno, F.J., and Kelly, R.G. (2005). Computational modeling of cathodic limitations on localized corrosion of wetted SS 316L at room temperature. Corrosion Sci. 47: 2987–3005, https://doi.org/10.1016/j.corsci.2005.05.051.Search in Google Scholar

Dunn, D.S., Cragnolino, G.A., and Sridhar, N. (2000). An electrochemical approach to predicting long term localized corrosion of corrosion-resistant high-level waste container materials. Corrosion 56: 90–104, https://doi.org/10.5006/1.3280526.Search in Google Scholar

Dunn, D.S., Pan, Y., Chiang, K.-T., Yang, L., Cragnolino, G.A., and He, X. (2005a). The localized corrosion resistance and mechanical properties of alloy 22 waste package outer containers. J. Miner. Met. Mater. Soc. 57: 49–55, https://doi.org/10.1007/s11837-005-0064-7.Search in Google Scholar

Dunn, D.S., Pan, K.-T., Yang, L., and Cragnolino, G.A. (2005b). Localized corrosion susceptibility of alloy 22 in chloride solutions. Part 1: mill-annealed condition. Corrosion 61: 1078–1085, https://doi.org/10.5006/1.3280624.Search in Google Scholar

Dunn, D.S., Pan, K.-T., Yang, L., and Cragnolino, G.A. (2006). Localized corrosion susceptibility of alloy 22 in chloride solutions. Part 2: effect of fabrication processes. Corrosion 62: 3–12, https://doi.org/10.5006/1.3278250.Search in Google Scholar

Evans, K.J., Yilmaz, A., Day, S.D., Wong, L.L., Estill, J.C., and Rebak, R.B. (2005). Using electrochemical methods to determine alloy 22’s crevice corrosion repassivation potential. J. Miner. Met. Mater. Soc. 57: 56–61, https://doi.org/10.1007/s11837-005-0065-6.Search in Google Scholar

Fanelli, D. (2018). Is science really facing a reproducibility crisis, and do we need it to? Proc. Natl. Acad. Sci. USA 115: 2628–2631, https://doi.org/10.1073/pnas.1708272114.Search in Google Scholar PubMed PubMed Central

Giordano, C.M., Rincón Ortíz, M., Rodríguez, M.A., Carranza, R.M., and Rebak, R.B. (2011). Crevice corrosion testing methods for measuring repassivation potential of alloy 22. Corrosion Eng. Sci. Technol. 46: 129–133, https://doi.org/10.1179/1743278210Y.0000000014.Search in Google Scholar

Gonzalez, M.E., Kappes, M.A., Rodríguez, M.A., Bozzano, P., Carranza, R.M., and Rebak, R.B. (2018). Optimization of the double loop electrochemical potentiokinetic reactivation method for detecting sensitization of nickel Alloy 690. Corrosion 74: 210–224, https://doi.org/10.5006/2562.Search in Google Scholar

Haugan, E.B., Næss, M., Torres Rodriguez, C., Johnsen, R., and Iannuzzi, M. (2017). Effect of tungsten on the pitting and crevice corrosion resistance of type 25Cr super duplex stainless steels. Corrosion 73: 53–67, https://doi.org/10.5006/2185.Search in Google Scholar

Heubner, U.L., Altpeter, E., Rockel, M.B., and Wallis, E. (1989). Electrochemical behavior and its relation to composition and sensitization of NiCrMo alloys in ASTM G-28 aolution. Corrosion 45: 249–259, https://doi.org/10.5006/1.3577851.Search in Google Scholar

Hornus, E.C., Giordano, C.M., Rodríguez, M.A., Carranza, R.M., and Rebak, R.B. (2015). Effect of temperature on crevice corrosion susceptibility of nickel alloys containing chromium and molybdenum. J. Electrochem. Soc. 162: C105–C113, https://doi.org/10.1149/2.0431503jes.Search in Google Scholar

Illowsky, B. and Dean, S. (2018). Introductory statistics. Houston, TX, USA: OpenStax.10.5006/2562Search in Google Scholar

Kehler, B.A., Ilevbare, G.O., and Scully, J.R. (2001). Crevice corrosion stabilization and repassivation behavior of Alloy 625 and alloy 22. Corrosion 57: 1042–1065, https://doi.org/10.5006/1.3281677.Search in Google Scholar

Levene, H. (1960). Robust test on equality of variances. In: Contributions to probability and statistics, 1. Stanford University Press, Palo Alto, pp. 278–292.Search in Google Scholar

Lu, Y.L., Pike, L.M., Brooks, C.R., Liaw, P.K., and Klarstrom, D.L. (2007). Strengthening domains in a Ni-21Cr-17Mo alloy. Scripta Mater. 56: 121–124, https://doi.org/10.1016/j.scriptamat.2006.09.011.Search in Google Scholar

Maristany, H.G., Kappes, M.A., Rodríguez, M.A., Carranza, R.M., and Rebak, R.B. (2016). Crevice corrosion of nickel alloys for steam generator tubing of pressurized water reactors. In: Corrosion 2016. Houston, TX, USA: NACE International, Paper C2016-7166.10.5006/C2016-07166Search in Google Scholar

Maristany, H.G., Kappes, M.A., Rodríguez, M.A., and Carranza, R.M. (2018). Localized corrosion of alloys UNS N06690 and N06600 in steam generator lay-up conditions. In: Corrosion 2018. Houston, TX, USA: NACE International, Paper C2018-11306.10.5006/C2018-11306Search in Google Scholar

McArthur, S.L. (2019). Repeatability, reproducibility, and replicability: tackling the 3R challenge in biointerface science and engineering. Biointerphases 14: 020201, https://doi.org/10.1116/1.5093621.Search in Google Scholar PubMed

Mishra, A.K. and Frankel, G.S. (2008). Crevice corrosion repassivation of alloy 22 in aggressive environments. Corrosion 64: 836–844, https://doi.org/10.5006/1.3279917.Search in Google Scholar

Mishra, A.K. and Shoesmith, D.W. (2014). Effect of alloying elements on crevice corrosion inhibition of nickel-chromium-molybdenum-tungsten alloys under aggressive conditions: an electrochemical study. Corrosion 70: 721–730, https://doi.org/10.5006/1170.Search in Google Scholar

Miyagusuku, M., Carranza, R.M., and Rebak, R.B. (2015). Inhibition mechanism of phosphate ions on chloride-induced crevice corrosion of alloy 22. Corrosion 71: 574–584, https://doi.org/10.5006/1373.Search in Google Scholar

Plesser, H.E. (2018). Reproducibility vs. replicability: a brief history of a confused terminology. Front. Neuroinf. 11: 1–4, https://doi.org/10.3389/fninf.2017.00076.Search in Google Scholar PubMed PubMed Central

Rebak, R.B. (2000). Corrosion of non-ferrous alloys. I. Nickel-, cobalt-, copper-, zirconium-and titanium-based alloys. In: Materials science and technology: a comprehensive treatment. Weinheim, Germany: Wiley-VCH Verlag GmbH, pp. 69–111.10.1002/9783527619306.ch11Search in Google Scholar

Rebak, R.B., Koon, N.E., Dillman, J.R., and Crook, P. (2000). Influence of aging on microstructure, mechanical properties, and corrosion resistance of a Ni-22Cr-13Mo-3W alloy. In: Corrosion 2000. Houston, TX, USA: NACE International, Paper no. 00181.10.5006/C2000-00181Search in Google Scholar

Rebak, R.B., Etien, R.A., Gordon, S.R., and Ilevbare, G.O. (2006). Influence of black annealing oxide scale on the anodic behavior of alloy 22. Corrosion 62: 967–980, https://doi.org/10.5006/1.3278235.Search in Google Scholar

Rincón Ortíz, M., Rodríguez, M.A., Carranza, R.M., and Rebak, R.B. (2010). Determination of the crevice corrosion stabilization and repassivation potentials of a corrosion-resistant alloy. Corrosion 66: 105002-1–105002-12, https://doi.org/10.5006/1.3500830.Search in Google Scholar

Rincón Ortíz, M., Rodríguez, M.A., Carranza, R.M., and Rebak, R.B. (2013). Oxyanions as inhibitors of chloride-induced crevice corrosion of alloy 22. Corrosion Sci. 68: 72–83, https://doi.org/10.1016/j.corsci.2012.10.037.Search in Google Scholar

Rodríguez, M.A. (2012). Inhibition of localized corrosion in chromium containing stainless alloys. Corrosion Rev. 30: 19–32, https://doi.org/10.1515/CORRREV.2011.022.Search in Google Scholar

Rodríguez, M.A., Carranza, R.M., and Rebak, R.B. (2010). Effect of potential on crevice corrosion kinetics of alloy 22. Corrosion 66: 015007-1–015007-14, https://doi.org/10.5006/1.3318286.Search in Google Scholar

Sawilowsky, S.S. (2009). Very large and huge effect sizes. J. Mod. Appl. Stat. Methods 8: 597–599, https://doi.org/10.22237/jmasm/1257035100.Search in Google Scholar

Shan, X. and Payer, J.H. (2010). Effect of polymer and ceramic crevice formers on the crevice corrosion of Ni-Cr-Mo alloy 22. Corrosion 66: 105005–105005-14, https://doi.org/10.5006/1.3500833.Search in Google Scholar

Sosa Haudet, S., Rodríguez, M.A., and Carranza, R.M. (2012). Effect of alloy composition on the localized corrosion of nickel alloys. In: Carranza, R.M., Duffó, G.S., and Rebak, R.B. (Eds.), MRS symposium proceedings. Scientific basis for nuclear waste management XXXV, Vol. 1475. New York, NY, USA: Cambridge University Press, pp. 489–494.10.1557/opl.2012.621Search in Google Scholar

Sridhar, N. and Cragnolino, G.A. (1993). Applicability of repassivation potential for long-term prediction of localized corrosion of alloy 825 and type 316L stainless steel. Corrosion 49: 885–894, https://doi.org/10.5006/1.3316014.Search in Google Scholar

Sridhar, N., Tormoen, G., Hackney, S., and Anderko, A. (2009). Effect of aging treatments on the repassivation potential of duplex stainless steel S32205. Corrosion 65: 650–662, https://doi.org/10.5006/1.3319092.Search in Google Scholar

Szklarska-Smialowska, Z. (2005). Pitting and crevice corrosion. Houston, TX, USA: NACE International.Search in Google Scholar

Tawancy, H.M. (1996). Precipitation characteristics of μ-phase in wrought nickel-base alloys and its effect on their properties. J. Mater. Sci. 31: 3929–3936, https://doi.org/10.1007/BF00352653.Search in Google Scholar

Tawancy, H.M., Herchenroeder, R.B., and Asphahani, A.I. (1983). High-performance Ni-Cr-Mo-W alloys. J. Miner. Met. Mater. Soc. 35: 37–43, https://doi.org/10.1007/BF03338300.Search in Google Scholar

Thompson, N.G. and Syrett, B.C. (1992). Relationship between conventional pitting and protection potentials and a new, unique pitting potential. Corrosion 48: 649–659, https://doi.org/10.5006/1.3315985.Search in Google Scholar

Tormoen, G., Sridhar, N., and Anderko, A. (2010). Localised corrosion of heat treated alloys. Part 1: repassivation potential of alloy 600 as function of solution chemistry and thermal aging. Corrosion Eng. Sci. Technol. 45: 155–162, https://doi.org/10.1179/147842208X320315.Search in Google Scholar

Tsujikawa, S. and Hisamatsu, Y. (1980). On the repassivation potential for crevice corrosion. Corros. Eng. 29: 37–40, https://doi.org/10.3323/jcorr1974.29.1_37.Search in Google Scholar

Turnbull, A. (1998). Prevention of crevice corrosion by coupling to more noble materials? Corrosion Sci. 40: 843–845, https://doi.org/10.1016/S0010-938X(97)00169-8.Search in Google Scholar

Zadorozne, N.S., Rodríguez, M.A., Carranza, R.M., Meck, N.S., and Rebak, R.B. (2010). Corrosion resistance of Ni-Cr-Mo and Ni-Mo-Cr alloys in different metallurgical conditions. In: Corrosion 2010. Houston, TX, USA: NACE International, Paper no. 10236.10.5006/C2010-10236Search in Google Scholar

Zadorozne, N.S., Giordano, C.M., Rodríguez, M.A., Carranza, R.M., and Rebak, R.B. (2012). Crevice corrosion kinetics of nickel alloys bearing chromium and molybdenum. Electrochim. Acta 76: 94–101, https://doi.org/10.1016/j.electacta.2012.04.157.Search in Google Scholar

Supplementary Material

This article contains supplementary material (https://doi.org/10.1515/corrrev-2022-0074).

Received: 2022-08-22

Revised: 2023-02-23

Accepted: 2023-02-25

Published Online: 2023-03-20

Published in Print: 2023-06-27

Supplementary Material

Articles in the same Issue

https://doi.org/10.1515/corrrev-2022-0074

Keywords for this article

chlorides; crevice corrosion; localised corrosion; nickel alloys; passivity

Statistical analysis of the repeatability of the crevice corrosion repassivation potential

Article

Abstract

1 Introduction

2 Materials and methods

2.1 Cyclic potentiodynamic polarisation

2.2 Tsujikawa–Hisamatsu electrochemical method

2.3 Potentiodynamic–galvanostatic–potentiodynamic method

2.4 Artificial crevice formers and experimental setup

2.5 Statistical analyses

3 Results and discussion

3.1 Variables allegedly affecting the outcome of testing techniques

3.1.1 Effect of the crevice formers material

3.1.2 Effect of the extent of corrosion propagation before repassivation

3.1.3 Effect of the electrochemical procedure

3.1.3.1 THE method from ASTM G192 round robin

3.1.4 Effect of the applied torque to ceramic crevice formers wrapped with PTFE tape

3.2 Metallurgical and environmental variables

3.2.1 Effect of thermal ageing

3.2.2 Effect of chemical composition of the alloy

3.2.3 Effect of temperature

3.2.4 Effect of chloride concentration of the environment

3.2.5 Effect of inhibitors

3.3 Comparison of ER,CREV: when is a change significant?

3.3.1 Suggested guideline for a planned comparison of ER,CREV

4 Conclusions

Acknowledgments

References

Supplementary Material

Supplementary Material

Articles in the same Issue

Articles in the same Issue

Articles in the same Issue

3.3 Comparison of E_R,CREV: when is a change significant?

3.3.1 Suggested guideline for a planned comparison of E_R,CREV