Perspective on “An Electrochemical Approach to Predicting Long-Term Localized Corrosion of Corrosion-Resistant High-Level Waste Container Materials,” D.S. Dunn, G.A. Cragnolino, and N. Sridhar, Corrosion 56, 1 (2000): p. 90-104 Open Access

Lifetime predictions of containers used for the disposal of high-level radioactive waste (HLW) are necessary to determine the overall performance of a HLW repository. Failure of the containers by localized corrosion and release of radionuclides into the accessible environment may occur if chloride-containing groundwater infiltrates the repository and interacts with the containers. To assess the repository performance, it is necessary to develop a methodology for predicting the long-term susceptibility of container materials to localized corrosion. Laboratory tests have demonstrated that pitting and crevice corrosion are initiated when the corrosion potential of the material exceeds the repassivation potential for localized corrosion. Once corrosion is initiated, the propagation of the localized attack will result in rapid penetration of the waste packages. However, no localized corrosion can be initiated, and all active corrosion sites repassivate when the corrosion potential of the material is lower than the repassivation potential. Results of this investigation demonstrated that the corrosion potential and repassivation potential are useful parameters to predict the long-term localized corrosion susceptibility of corrosion- resistant candidate container materials such as type 316L (UNS S31603) stainless steel and alloy 825 (UNS N08825).

Editor’s Note: Originally published in Corrosion 56, 1 (2000): p. 90-104. When citing “An Electrochemical Approach to Predicting Long-Term Localized Corrosion of Corrosion-Resistant High-Level Waste Container Materials,” please cite the original version.

KEY WORDS: alloy 825, chloride, high-level nuclear waste containers, localized corrosion, nickel-based alloys, radionuclides, repassivation potential, stainless steel

The primary goals of the U.S. Department of Energy (DOE) updated waste containment and isolation strategy for the proposed repository at the Yucca Mountain (YM) site are the near-complete containment of radionuclides within waste packages (WP) for several thousand years and acceptably low annual doses to a member of the public living near the site.¹  The WP consists of containers, fillers, basket materials, and waste forms (spent fuel [SF] or vitrified waste). One of the most important system attributes recognized in this strategy is container lifetime. To predict the container lifetime confidently, the adequacy of methodologies for extrapolating short-term laboratory data to long-term performance needs to be demonstrated.

Several WP designs have been evaluated by the DOE over the history of the YM repository program. The canistered fuel design is considered most likely to be used for high-level radioactive waste (HLW) disposal.²  In this design, either 21 pressurized water reactors (PWR) or 40 boiling water reactors (BWR) assemblies are contained in a type 316L (UNS S31603)⁽¹⁾ stainless steel (SS) canister (3.5 cm wall thickness) surrounded by an inner overpack made of a corrosion-resistant alloy (2.0 cm wall thickness), which is contained in an outer overpack made of a corrosion allowance material (10 cm wall thickness). The total surface area of such a WP is ∼ 37 m² and the loaded weight is ∼ 65,000 kg. Alternate designs include WP for uncanistered fuel, small-canistered fuel (12 PWR or 24 BWR SF assemblies), and vitrified waste. Nickel-based alloys, such as alloys 825, 625, and C-22 (UNS N08825, N06625, N06022, respectively) presently are being evaluated for the inner corrosion-resistant overpack, and a carbon steel (such as A516 grade 55) is proposed for the outer corrosion allowance overpack.³  If groundwater contacts the container, it is expected by the DOE that uniform corrosion of the carbon steel overpack will proceed at a slow and predictable rate. Evaporative effects and interactions with concrete in the repository may result in an alkaline groundwater that, in the presence of chloride, can lead to localized corrosion of the carbon steel surface.⁴  If the carbon steel overpack is perforated, then the inner overpack will be exposed to the repository environment. Corrosion of the inner overpack may be mitigated by the use of a corrosion-resistant alloy and galvanic coupling of the inner overpack to the carbon steel overpack.^5-6 

The most common approach to long-term corrosion prediction involves obtaining the corrosion rates of candidate materials exposed to environmental conditions that are similar to, or preferably more severe than, the anticipated environmental conditions of the particular application. These rates then are integrated over time to predict service life. The corrosion rates are obtained through the use of controlled laboratory tests, coupon exposures in the field, or performance evaluation of similar materials in analogous applications. An example of this approach is found in the DOE Total System Performance Assessment (TSPA) conducted in 1995.⁷  This analysis uses atmospheric and natural water corrosion data to estimate the corrosion rate of outer overpacks in the repository. For the corrosion-resistant inner overpack, degradation was assumed to occur only by localized corrosion that was described by a pit growth rate that varies with temperature. The general corrosion of the corrosion-resistant container is determined by the passive dissolution rate of the material. This approach, referred to here as the empirical corrosion rate approach, has many drawbacks when applied to corrosion estimates for the repository time scales (eg., 10³ years to 10⁴ years). These limitations include:

First, the relationship between the laboratory or field service environments in which corrosion rates are obtained and the repository environment seldom can be established clearly. Inferring analogous behavior between short-term experiments and the repository setting is especially problematic in the case of the hydrologically unsaturated zone at the YM site. In the proposed repository, the fluid in the environment surrounding the container (typically referred to as the near-field or very near-field environment) may evolve in time and may vary in ionic strength from that of a dilute groundwater to that of a saturated electrolyte because of a complex combination of processes including evaporation, interaction with minerals and man-made materials, and radiolysis.

Second, several corrosion phenomena, especially localized corrosion and stress corrosion cracking (SCC), have relatively long initiation times that may be longer than the time elapsed in short-term tests, which are typically < 7 days.⁸  Once corrosion initiates, however, the rates can be quite high. The initiation time may be a function of the material and test environment. Thus, short-term corrosion rate data may be more useful in eliminating materials that are not suitable for a given environment than for predicting the long-term performance of materials.

Third, it is difficult to evaluate effects of design features in corrosion processes such as galvanic coupling. Hence, a large number of long-term tests (i.e., several weeks to years), using various possible design configurations, have to be conducted.

Finally, proof of the performance of a material in a repository cannot be obtained easily because of the lack of long-term exposure data for the candidate corrosion-resistant container materials. If feasible, models developed to simulate the long-term corrosion behavior of a material should be constructed from a mechanistic understanding of uniform and localized corrosion processes. The predictive capability of the models can be verified using a diverse set of corrosion data. However, the empirical corrosion rate approach is not based on a mechanistic understanding of the processes involved. As a result, long-term estimations made using this approach are more uncertain. The usefulness of the empirical corrosion rate approach is constrained further when most of the field and laboratory measurements of corrosion rates do not include measurement of the corrosion potential of the metal in the test environment. As a result, data do not provide a suitable means to predict the performance of the material when the duration of the exposure is increased. Because the corrosion potential of the test specimens is not always measured, valuable information necessary to predict the onset of localized corrosion in long-term exposures often is not available.

An electrochemical approach can overcome many of the limitations of the empirical corrosion rate approach by explicitly calculating corrosion rates in terms of constituent electrochemical reactions^9-10  or by using critical electrochemical parameters for determining the onset of various corrosion processes.^5,11  An electrochemical approach has several advantages.

First, the occurrence of various corrosion modes such as pitting, crevice corrosion, and SCC can be accelerated in short-term tests using a range of environments and applied potentials. The critical potentials for the occurrence of various corrosion modes then can be determined. The extrapolation from these short-term data is made then by calculating the evolution of the natural potential of the metal (corrosion potential) in the repository environment and comparing these to calculations derived from the measured critical potentials. The calculations of corrosion potentials are supported by fundamental electrochemical models based on the kinetics of the corrosion reactions. Thus, a clearer and more defensible relationship can be established between the test environments and the anticipated repository environment.

Second, the initiation time for localized corrosion and SCC explicitly can be considered in terms of its relationship to the corrosion potential. In some cases, kinetics of the corrosion modes can be measured independently and supported by electrochemical models.

Finally, the corrosion performance of a given material in diverse applications can be assembled in a single map of critical electrochemical parameters. This approach, similar to the one used by Staehle,¹²  could provide greater confidence in the performance prediction.

Electrochemical models do not dispense with the need to measure corrosion rates; rather, they assign different corrosion rates to different regimes of corrosion, with the delineation of the regimes of corrosion and the rates of corrosion within each regime supportable by fundamental, mechanistic reasoning. It also must be emphasized that electrochemical models, just as other models, are only as good as the quality of the underlying data. Thus, electrochemical models offer an additional advantage over empirical corrosion rate models because they can be used to evaluate critical data needs.

A number of investigators have proposed the use of the repassivation potential (E_rp) as a parameter to predict the long-term initiation of localized corro- sion^13-16  and SCC.^11,17  If the corrosion potential (E_corr) of the metal exceeds E_rp, then localized corrosion can be initiated. Prior to initiation of localized corrosion, the metal is expected to corrode at a slow rate corresponding to the passive current density. After the initiation of localized corrosion, the pit would propagate through the metal container at a rate that is orders of magnitude greater than would be expected during passive dissolution. If the E_corr of the inner barrier decreases below E_rp, active localized corrosion would cease and the metal will repassivate. The concept of repassivation is not new. Pourbaix, et al., proposed the use of the E_rp (also called the protection potential) below which localized corrosion would not be initiated.¹⁸  However, later investigations by Wilde,¹⁹  and Rosenfeld, et al.,²⁰  indicated that the E_rp decreased with increasing extent of prior corrosion, leading to the conclusion that this parameter cannot be used as a conservative criterion. Hence, to be a conservative parameter for predicting long-term localized corrosion, it must be demonstrated that a lower bound value for E_rp can be established.

In the present study, data on type 316L SS and alloy 825 generated by the Center for Nuclear Waste Regulatory Analyses (CNWRA) program during the past few years was reviewed along with data from the literature to demonstrate the conservatism of E_rp. Additionally, the applicability of E_rp for prediction of the long-term occurrence of localized corrosion was examined using results of ongoing long-term tests. Finally, the current mechanistic understanding of localized corrosion initiation and repassivation processes was assessed.

A variety of corrosion-resistant and corrosion allowance materials are being tested in the CNWRA program. However, this study focused mainly on alloy 825 and type 316L SS because its objective was to establish the predictive methodology for corrosion- resistant materials. The compositions of type 316L SS and alloy 825 are given in Table 1.

Table 1.

Composition of Alloy 825 and Type 316L Stainless Steel (wt%)

View large

View Large

Tests were conducted at 95°C±2°C in an ASTM G-31²¹ -type electrochemical cell with ≈ 1,500 mL solution containing 1,000 ppm Cl⁻, 85 ppm ⁠, 20 ppm ⁠, 10 ppm ⁠, and 2 ppm F⁻, all as sodium salts. The solution composition reflected the expected composition of groundwater that may be expected in a HLW repository with an increased chloride concentration. The effect of chloride concentration was studied by adding chloride as sodium chloride (NaCl) to the above solution. However, because of the solubility limitation, solutions with chloride concentration > ∼ 5.5 M were achieved by the use of lithium chloride (LiCl) or magnesium chloride (MgCl₂) solutions, without the addition of other anionic and cationic species. The pH of the concentrated MgCl₂ solutions were measured to be in the range from 2.0 to 2.7. In tests conducted under air- saturated conditions, a high-purity mixture of 79% N₂ + 21% O₂ (carbon dioxide [CO₂] free air or zero air) gas was used to purge the solution.

Cylindrical specimens of alloy 825 and type 316L SS measuring 6.2 mm in diameter and 48.6 mm long with a 600-grit finish (Figure 1[a]) were used in cyclic potentiodynamic polarization (CPP) tests. These tests were conducted in an ASTM G-5²² -type five-neck flask equipped with a calibrated thermometer, a platinum counter electrode, and a salt bridge with a porous silica tip filled with test solution. The salt bridge was connected to a saturated calomel electrode (SCE) that was maintained at room temperature and served as the reference electrode. Test solutions were deaerated thoroughly with high- purity N₂ (99.999 %) through a glass frit bubbler. Prior to the start of each test, the E_corr of the specimen was measured vs the reference electrode using a high-impedance (100 MΩ) electrometer. The potential of the test specimens were increased at a rate of 0.167 mV/s using a computer-controlled potentiostat. Potential scans were started at the E_corr. The current density of the specimen was monitored continuously. At a current density of 5 mA/cm², the direction of the potential scan was reversed and the potential of the specimen was decreased using the same scan rate. The potential scan was stopped upon reaching the initial value of E_corr. Following each scan, the specimen was examined for localized attack. The purpose of the CPP tests was to establish initiation potentials for pitting (E_p) and crevice corrosion (E_crev), as well as E_rp over a wide range of environmental conditions (10⁻⁴ M Cl⁻ to 9 M Cl⁻, 25° C to 95°C). An SCE maintained at room temperature was used as a reference in these tests.

Simulated, single-pit propagation tests were conducted on alloy 825 using lead-in-pencil specimens (Figure 1[b]). The advantage of this geometry is that the location of pitting is known and the geometry is amenable to theoretical analyses using relatively simple analytical models. In addition, the repassivation of very deep simulated pits can be investigated with this specimen geometry. The disadvantages of the single-pit specimen are that the morphologies and growth rates of the simulated pit specimens are different than those of naturally formed pits. As a result, the growth rates obtained in single-pit tests cannot be used to predict the penetration of naturally formed pits. Prior to the start of the tests, specimens were weighed to ±0.03 mg. The specimens then were covered with a polyolefin tubing that allowed only one end of the specimen to be exposed to the solution. Electrical contact to the specimen was made outside the test cell. The specimen was oriented so that the opening of the unidirectional pit faced upward. The pit had an area of 0.041 cm². Corrosion was initiated at potentials of 500 mV_SCE to 600 mV_SCE. After ≈ 5 h, the anodic current density was constant, indicating that stabilization of the actively growing pit was achieved and the potential was decreased to values from −300 mV_SCE to 100 mV_SCE at a rate of 5 mV/s. The current and potential were recorded allowing real time calculation of pit depth and propagation rate. This calculation was made using the theoretical weight loss of 2.64 × 10⁻⁴ g/coul assuming congruent dissolution of Fe, Cr, and Ni in alloy 825 as Fe²⁺, Cr³⁺, and Ni²⁺, respectively.²³  At the conclusion of the tests, specimens were weighed again to confirm the accuracy of the propagation rate calculations. The purpose of these tests was to measure the growth rate and repassivation of single pits and compare them to measurements made on samples where an uncontrollable number of multiple pits formed.

Potentiostatic polarization tests were conducted using cubic specimens (Figure 1[c]). These specimens had a total surface area of 15 cm², including polished and Cr-depleted mill-finished surfaces, as described in previous publications.^15,23  Potentiostatic and open- circuit corrosion tests also were conducted using specimens with a controlled crevice (Figure 1[d]).

Potentiostatic pitting and crevice corrosion initiation times were measured by polarizing cylindrical and creviced specimens to potentials above and below the E_rp measured in CPP tests. Specimens were connected to a computer-controlled multichannel potentiostat. A constant potential was applied for test intervals of 28 days. Between test periods, specimens were examined for visible signs of corrosion and weight loss using an analytical balance with ±0.00005 g reproducibility (standard deviation) while the test solutions were changed. No further exposures were conducted on specimens after localized corrosion occurred. Open-circuit tests were conducted with a setup and solutions similar to those used for the potentiostatic polarization tests. However, instead of controlling the potential of the specimen with a potentiostat, the corrosion potential of alloy 825 was monitored continuously in air-saturated, 1,000 ppm Cl⁻ solutions at 95°C. Periodically, the test was interrupted and the specimen was examined for signs of localized corrosion. These methods allowed the validity of using E_rp from short-term CPP tests as a parameter for determining long-term performance to be tested.

Since at open circuit, no net current flows in the system, the corrosion kinetics cannot be measured using electrochemical techniques on a single specimen. Thus, a series of open-circuit corrosion tests was conducted with a creviced alloy 825 specimen connected to a large noncreviced alloy 825 plate that served as a cathode through a zero-resistance ammeter (ZRA). If crevice corrosion occurred, the potential of the crevice specimen decreased, resulting in an anodic current in the external circuit. Tests were conducted in solutions containing 1,000 ppm Cl⁻ at 95°C±2°C. The redox potential of the test solution was controlled by using a redox couple such as Cu⁺/Cu²⁺ added as chloride salts with hydrochloric acid (HCl) or NaCl additions to achieve the desired total chloride concentration. The area ratio of the alloy 825 cathode plate to the creviced area of the test specimen was at least 10:1. During the tests, the current was measured with a resolution of 1 nA. The potential of the specimens also was recorded with a resolution of 0.1 mV_SCE maintained at room temperature. This test method allowed the measurement of corrosion rates under natural exposure conditions at various redox potentials.

Pit repassivation tests also were conducted on alloy 825 specimens (Figure 1[a]). Tests were conducted by initiating pits at 600 mV_SCE. After initiation, pits were grown at 400 mV_SCE. Following the accumulation of a specified charge density, the applied potential of the specimens was decreased to the desired value. The repassivation time was measured as the time required for the current density to decrease < 50 μA/cm² after the potential of the specimen was decreased rapidly from 400 mV_SCE to the desired potential. Details of this technique have been published previously.^11,16  These tests allowed the validity of E_rp to be tested on specimens with multiple pits.

Another notable observation (Figure 3) was the relative independence of E_rp on pH between pH 1 and pH 8.2. This observation was consistent with the observations in the literature, as well as the current understanding of pitting process. The pH inside the pit was dictated by the hydrolysis of cations and became independent of the external pH.

Previous work has established the dependence of E_rp on other environmental factors such as the concentrations of ⁠, ⁠, and F⁻.²⁴  As shown in Figures 2 and 3, thiosulfate (⁠⁠) decreased the E_rp significantly for both alloys. The effect of observed in this study was similar to that found by Nakayama, et al.²⁵ 

Although the presence of oxygen was not expected to alter the values of E_p and E_rp, the occurrence of the oxygen reduction reaction determined the corrosion potential of the container materials in a repository environment. The corrosion potential data measured in air-saturated solutions also are plotted in Figures 2 and 3. In addition, the corrosion potential of alloy 825 in Cl^- solutions with the addition of 5 mM hydrogen peroxide (H₂O₂) is shown in Figure 3. As mentioned before, the occurrence of localized corrosion in a given environment can be gauged by comparing the E_corr under anticipated redox conditions and E_rp in the same environment, but without the redox species. While redox species such as oxygen and H₂O₂ were not expected to alter the E_rp significantly, it generally is convenient to obtain E_rp values in the absence of these redox species to expand the range of potentials studied. From Figure 2, it was anticipated that type 316L SS would exhibit localized corrosion in air- saturated solution with chloride concentrations >10^-3 M, as a result of the E_corr being more noble than the E_rp. In contrast, the data plotted in Figure 3 suggests that alloy 825 did not suffer localized corrosion in air-saturated solutions even at chloride concentrations as high as 4 M since the E_corr was less noble than the E_rp. However, predictions based on the data shown in Figures 2 and 3 are valid only if the values of E_rp and E_corr are not time-dependent. Long-term tests, conducted as part of this investigation to determine the effect of exposure time on the value of these parameters, indicated the E_corr increased with time while E_rp was not time-dependent.

Long-term E_corr data for alloy 825 in a 1,000-ppm Cl^- solution is shown in Figure 4. Despite the fluctuations, a trend of an increasing E_corr was readily apparent. Over the course of almost 600 days of testing, E_corr increased from an initial value of -270 mV_SCE to a high of 160 mV_SCE. E_rp values for alloy 825 in ppm Cl^-, ranging from 50 mV_SCE to 100 mV_SCE, are indicated in this figure as reported previously by Dunn, et al.²⁶  After ≈ 350 days, the E_corr exceeded the upper value of E_rp (100 mV_SCE) and crevice corrosion was observed. Although the depth of attack in each case was shallow and the weight loss minimal, crevice corrosion was observed visually each time E_corr exceeded 100 mV_SCE.

The effect of penetration depth on the E_rp for pitting and crevice corrosion is shown in Figure 5. These tests were conducted by initiating pitting or crevice corrosion at potentials of 500 mV_SCE to 600 mV_SCE. After initiation, the localized corrosion was allowed to propagate at 400 mV_SCE.¹¹  Specimens then were repassivated by reducing the potential at a rate of 5 mV/s until the current density was continuously < 50 μA/cm². After repassivation, the depth of the pits and crevice-corroded regions were measured. The relationship between depth of attack and E_rp can be determined by varying the time during which localized corrosion was allowed to propagate. From the results plotted in Figure 5, there was a strong dependence of E_rp on pit depth for shallow penetrations. For deep pits, the E_rp appeared to approach a lower limit. For creviced specimens with shallow penetrations, there was considerable scatter in the data. This could have been related to the location of the localized attack within the crevice region. Because of the stochastic nature of localized corrosion, attack did not always occur at the same location inside the crevice. When depths of attack were shallow (i.e., low charge density), the diffusion distance for attack initiated near the center of the crevice area was greater than for attack initiated near the edge of the crevice- forming washer. As a result, the E_rp for shallow attack initiated near the center of the crevice was expected to be lower than the E_rp for attack located near the edge of the crevice. Variation in the measured E_rp values for shallow penetrations ranging from −80 mV_SCE to 20 mV_SCE was attributed to variation of the initiation site location within the crevice area. The E_rp ranged from −80 mV_SCE to 20 mV_SCE for a scan rate of 5 mV/s. When crevice attack was allowed to propagate to depths > 1.0 mm, however, the E_rp approached a lower bound value of −80 mV_SCE.

The effect of applied potential as well as corrosion potentials under various redox conditions on the time to initiate localized corrosion is shown in Figure 6. Results showed short-term tests and long-term tests conducted by holding test specimens at constant applied potentials using a potentiostat. The corrosion current density of the specimens were recorded throughout the duration of the tests. When localized corrosion was initiated, the current density increased from a passive current density of ≈ 10⁻⁷ A/cm² to values > 10⁻⁴ A/cm². In addition, corrosion products were observed on these specimens and localized corrosion was verified by optical examination at the conclusion of the tests. At potentials near the E_p measured in CPP tests (e.g., 600 mV_SCE), the initiation time was on the order of 200 s. Decreasing the potential to 500 mV_SCE resulted in initiation times of 200,000 s. In contrast, relatively short initiation times for crevice corrosion were observed even at applied potentials several hundred millivolts below the E_p. Figure 6 also shows pitting and crevice corrosion initiation time for specimens at open-circuit potentials under various redox conditions (1,000-ppm Cl⁻ solutions with either 1:1 or 1:7 Cu²⁺/Cu⁺ added as a redox couple). The error bands in these data points denote the variation in the corrosion potentials for the same redox couple used. No pitting corrosion was observed on the boldly exposed surfaces (i.e., the surface of the specimen not covered with a crevice forming device) of any of the creviced specimens. In addition, no pitting corrosion was observed on a boldly exposed specimen in which the open-circuit potential was in the range from 300 mV_SCE to 440 mV_SCE. It was clear that the initiation time for crevice corrosion under natural exposure conditions was consistent with that under applied potential conditions.

Repassivation times are not shown in Figure 6. However, these experiments demonstrated that the repassivation time increased with an increase in po- tential.¹⁶  The E_rp for alloy 825 measured in CPP tests with a 1,000-ppm Cl⁻ solution also is shown in Figure 6. It was evident that the initiation time for localized corrosion increased as the potential was decreased and approached E_rp. At potentials < E_rp, no localized corrosion was initiated. This was indicated by the results of several creviced specimens continuously maintained at applied potentials < E_rp (Figure 6). The observation that no crevice corrosion had been initiated on long-term test specimens maintained below the E_rp measured in short-term tests suggests that the E_rp was a stable parameter, independent of time. Many of the long-term tests shown in Figure 6 were still ongoing at various applied and natural potentials on alloy 825.

Figure 8 shows current density vs time for creviced alloy 825 specimens maintained at potentials between 300 mV_SCE and 500 mV_SCE. At the higher potentials, the current density quickly increased, indicating that crevice corrosion had been initiated rapidly. In contrast, at lower potentials, significant increases in the current density were observed only after some incubation time. At the conclusion of the tests, all specimens were examined. For specimens tested at 400 mV_SCE and above, crevice corrosion was initiated on at least 14 of the 24 crevices formed by the serrated polytetrafluoroethylene (PTFE) crevice washer. For specimens tested at potentials of 350 mV_SCE or less, only three or four crevices actively corroded. For all specimens, the current density was calculated using only the area of the corroded regions. Each individual crevice had a surface area of 0.06 cm² yielding a total creviced area of 1.44 cm² using two serrated crevice washers with 12 teeth per washer.

These results have been used in the Engineered Barrier System Performance Assessment Code (EBSPAC) to predict HLW container lifetimes.⁵  In this code, the E_rp is used as a threshold parameter and the E_corr is used as an enabling parameter that triggers localized corrosion and SCC when it exceeds E_rp. For deep pits and crevices (i.e., >1 mm), E_rp is mainly dependent on the material, Cl⁻ concentration of the bulk environment, and temperature. Hence, for a given material/environment combination, E_rp is a stable parameter that is independent of time. Certain anionic species, such as (Figures 2 and 3) may decrease E_rp while others such as NO3⁻ may act as inhibitors and increase E_rp. In the specific case of a HLW repository, the environment surrounding the WP will vary with time. Modeling calculations have predicted that the oxygen concentration,²⁷  pH, temperature, and Cl⁻ concentration will change significantly after emplacement of the HLW.²⁸  As shown in Figures 2 and 3, E_rp can be measured rather rapidly using the CPP technique and can be expressed in terms of a well-established logarithmic relationship to the Cl⁻ concentration.²⁹  The reduction in E_rp as the Cl⁻ concentration and temperature are increased indicates that, in environments with low Cl⁻ concentrations (and/or low temperatures), a high E_con. is needed to initiate localized corrosion. However, as the aggressiveness of the environment is increased, localized corrosion can be initiated at lower values of E_corr.

E_rp values shown in Figure 6 span a potential range of 150 mV. This span is a result of the test technique. The 150-mV range of E_rp values would likely be reduced if the material was tested using either a much slower scan rate or a potentiostatic test method. The scan rate used in the present study (0.167 mV/s) may result in some variation in maximum penetration depth that, in turn, alters the measured E_rp values (Figure 5) for shallow penetrations. While these test methods may reduce the range of the measured E_rp values, it is expected that the value of the E_rp of a material in a given environment will be distributed over a range of potential values. The key difference between the distributions of E_p and E_rp is that the distribution of E_p values decreases with time, whereas the distribution of E_rp values are not time-dependent. This is readily apparent from the results shown in Figure 6.

Potentiostatic testing carried out for times ranging from a few days to > 1,000 days (Figure 6) clearly indicates that localized corrosion can be initiated at potentials far less than the E_p measured in CPP tests (Figure 3). Thus, the E_p measured in CPP tests is not a useful parameter to predict the long-term initiation of localized corrosion. At potentials close to E_p, multiple crevice corrosion sites were initiated on the specimens. As the potential was reduced, the number of crevice corrosion sites decreased. For specimens tested at potentials near the E_rp, localized attack usually occurred at only one site. From the results shown in Figure 6, it is apparent that the initiation time for the creviced specimens under applied potentiostatic control compared well with the initiation times measured in open-circuit tests with Cu²⁺/Cu⁺ redox couple added to the solutions to increase the redox potential of the environment. These data provide increased confidence that the concept of repassivation potential can be applied to materials exposed to natural redox conditions. No localized corrosion has been initiated to date at potentials below the E_rp even after 1,000 days of testing, indicating that this parameter, as measured in short-term tests, can be used as the basis for long-term predictions of container performance.

Results obtained with alloy 825 single-pit specimens (Figure 9) indicated that, at potentials below the E_rp, the dissolution rate is approximately the same as the passive dissolution of alloy 825 (< 2 × 10⁻⁷ A/cm² or < 2 × 10⁻³ mm/y). At potentials above the E_rp, the simulated pits dissolved at a very rapid rate (1 × 10⁻² A/cm² or 10² mm/y). However, these pit growth rates were obtained under potentiostatic conditions where there was no cathodic limitation. Additionally, the geometry of the single-pit specimens used inert walls that did not allow the lateral propagation of the localized corrosion front. Thus, the propagation rate cylindrical pits formed using these specimens do not represent pit propagation rates under natural conditions.

For localized corrosion to be initiated in a repository setting, the E_corr of the corrosion-resistant barriers must be more noble than the E_rp. Initial measurements of the E_corr for type 316L SS shown in Figure 2 indicated that localized corrosion in deaerated environments was not likely. However, the E_corr was greater than E_rp in air-saturated environments and the possibility of localized corrosion existed. Although clearly more resistant, especially at low chloride concentrations, alloy 825 also was susceptible to localized corrosion when E_corr exceeded E_rp. Long-term corrosion potential data for alloy 825 in a 1,000-ppm Cl⁻-based solution are shown in Figure 4. The increase in the corrosion potential of this specimen was most likely the result of the dissolution of the mill-finished surface, which previously had been shown to be deficient in Cr compared to the bulk material. As a result, the initial surface layer corroded at a faster rate than the polished surfaces. This was supported by the initial weight loss observed in the early stages of exposure. The Cr-depleted layer yielded a lower E_corr because the alloy had a higher passive current density. After the Cr-depleted layer was removed, the exposed surface contained an increased amount of Cr. Passive films formed on surfaces that were not depleted in Cr had improved corrosion resistance and lower passive dissolution rates.

Since the E_corr was determined by the potential at which the sum of all anodic currents and all cathodic currents equaled zero, a decrease in the passive current density resulted in an increase in the E_corr. Other candidate corrosion-resistant WP materials such as alloys 625 and C-22 also will have a mill-finished surface that is depleted in Cr as a result of the production process.³⁷  A similar increase in E_corr was observed with thermally oxidized specimens. The corrosion potential of alloy 825 in air-saturated 1,000-ppm chloride solutions increased by as much as 200 mV after oxidation in dry air at 100°C for 30 days prior to the exposure to aqueous environ- ments.²³  Thickening and aging of the oxide layer in air-saturated solutions or during thermal exposures to dry air can reduce the anodic dissolution kinetics resulting in an increase in the E_corr.³⁸ 

The E_corr measurement shown in Figure 4 emphasizes the need to monitor the E_corr in long-term laboratory and field immersion tests. Because the parameters used in calculating the E_corr from electrochemical models are uncertain, especially for highly irreversible reactions such as oxygen reduction,⁶  calibration of models against measured open-circuit potentials is necessary when these electrochemical parameters are used to predict the long-term evolution of corrosion potentials. The E_corr can decrease after significant localized corrosion has occurred because of the influence of active pit growth region on the overall E_corr. The extent of decrease would depend on the relative area ratio of active localized corrosion area to the passive corrosion area. The lowering of E_corr after significant localized corrosion growth occurs was not addressed in the present study.

Several field experiences can be analyzed to show the applicability of E_rp in predicting the occurrence of localized corrosion. Bernhardsson and Mellstrom reported failure caused by pitting of alloy 825 compressor coolers on a gas lift platform.³⁹  The environment was chlorinated seawater at 42°C, and failure took place after 4 months (minimum penetration rate was estimated to be ∼ 5 mm/y). The E_rp for alloy 825 in 0.5 M chloride (the approximate concentration of chloride in seawater is 0.55 M) at 20°C is 60 mV_SCE.¹¹  Since the E_rp was expected to decrease with an increase in temperature and the E_corr in chlorinated seawater had been measured to be ∼ 200 mV_SCE,⁴⁰  this in-service failure was consistent with the use of E_rp as a determining factor for the initiation of localized corrosion. In stagnant, aerated 3% NaCl solution at 60°C, Bernhardsson and Mellstrom also found that type 316L SS underwent crevice corrosion after 2 months, while alloy 825 did not.³⁹  Although these are still relatively short-term tests, the results are consistent with even shorter-term E_rp values. For example, Okayama, et al., reported a value of −360 mV_SCE for E_rp of type 316 SS and −100 mV for alloy 825 in 3% NaCl at 80°C.⁴¹  The E_corr of these alloys in aerated chloride solution at 95°C has been reported to be —280 mV_SCE and is anticipated to increase slightly with a decrease in temperature as a result of higher solubility of oxy- gen.¹¹  This value of E_corr is higher than the E_rp of type 316L SS and lower than that of alloy 825, thus explaining the localized corrosion of the former.

Protection against corrosion in highly oxidizing pulp and paper bleach plant washer fluids using controlled potential has been reported.⁴²  In this case, the material is typically a type 317L SS (Fe-19% Cr-12% Ni-3.5% Mo [UNS S31703]) and the E_corr in chlorine- containing waters is in the range from 100 mV_SCE to 500 mV_SCE, depending upon residual chlorine. This is far higher than the potential at which crevice corrosion was observed in potentiostatic tests (considered to be roughly equal to the E_rp). For type 317L SS in this environment (typically 1,000 ppm Cl⁻ to 5,000 ppm Cl⁻ and pH 2), E_rp was measured to be ∼ −30 mV_SCE. In these cases, cathodic protection by maintaining the potential at −500 mV_SCE resulted insuccessful performance of the washer drums for > 10 years.

In filtered, natural seawater (mean temperature: 25.2°C), Kain observed crevice corrosion on type 316L SS after 60 days, but no crevice corrosion on alloy 59 (61% Ni-22% Cr-15% Mo [UNS N06059]).⁴³  Slight crevice corrosion was observed on alloy C-276 (Ni-16% Cr-5% Fe-16% Mo-4% W [UNS N10276]). Kain measured a E_corr in the range of 200 mV_SCE to 300 mV_SCE. In natural seawater, microbially enhanced cathodic reduction of oxygen is believed to be responsible for the high E_corr values. Again, the observed crevice corrosion of type 316L SS is consistent with the measured E_rp of 0 mV_SCE at that tempera- ture.⁴¹  Many of Fe-Ni-Cr-Mo alloys were tested in natural, filtered seawater at 30°C for 30 days by Hack.⁴⁴  Specimens were fitted with crevice washers of PTFE. In this limited duration test, alloys C-276 and 625 (Ni-22% Cr-5% Fe-9% Mo-4% Nb) suffered no localized corrosion, while type 316 SS and alloy 825 suffered significant crevice corrosion. The E_rp for alloy 825 in 0.5 M Cl⁻ was −60 mV_SCE, which was lower than the E_corr values reported for these alloys in natural seawater. An evaluation of all the published data on the E_rp values of alloy 825 and type 316/316L SS indicates that, if the chloride concentration becomes higher than = 0.5 M, alloy 825 may not be significantly better than type 316L SS in localized corrosion resistance. The relatively superior localized corrosion resistance of alloys 625 and C-276 compared to alloy 825 is consistent with experimental evidence that the E_rp of alloys 625 and C-276 is higher than that of alloy 825.⁴¹  The ongoing experimental studies at CNWRA on alloys 625 and C-22 also support these field observations.

Pit and crevice corrosion initiation have been thought of as originating from different processes. Pit initiation is well established as a stochastic process, with many metastable pits nucleating and repassivating dynamically on the metal surface below the critical potential for stable pit initiation.^29,45  Crevice corrosion initiation traditionally has been thought of as a deterministic process dictated by the development of a critical solution composition with a high chloride concentration and a low pH deep in the crevice such that the passive film is dissolved⁴⁶  or a critical potential drop caused by ohmic resistance of the crevice electrolyte such that active corrosion is initiated.⁴⁷  However, recent evidence has shown that crevice corrosion in highly corrosion-resistant alloys occurs essentially by the nucleation and growth of pits. Sridhar and Dunn showed that critical solution composition and potential drop did not occur until an increase in crevice current was detected, indicating that pits nucleated inside the crevice first (measured by an increase in current density) followed by growth of these pits to a sufficient spatial extent to affect crevice chemistry.⁴⁸  Shinohara, et al., used the Moire fringe technique to show that pits nucleated inside the crevice and then spread in depth and lateral extent.⁴⁹  Hence, at least for the corrosion- resistant materials, a view of crevice corrosion is emerging where the crevice is thought to provide a long enough diffusion path to stabilize metastable pits.⁵⁰  The observation of the similarity of E_rp for deep pits and crevices in this study reinforced this point of view.

Pit initiation theories can be classified as those that focus on the properties of the passive film and factors leading to the destabilization of the film and those that focus on processes in embryonic pits and factors leading to the stabilization of these pits.^29,45  The present study indicated that the latter theories are more relevant to predicting the long-term initiation of localized corrosion. However, this does not imply that passive film properties are unimportant for the long-term initiation of localized corrosion of a material. Clearly, the passive film properties determine the evolution of corrosion potential that dictates the triggering of localized corrosion. Equally important, if the repassivation process is conceived as a competition between maintaining a salt film and a concentrated chloride solution in nascent pits vs the formation of a passive oxide film, then properties of passive films will play a role.

Burstein and Mattin proposed that the embryonic pits covered by the remains of the original oxide passive film acquire a critical solution chemistry because of metal dissolution inside them and consequent chloride ion migration.⁵¹  This solution may reach saturation with respect to metal chloride depending upon the dissolution rate (current) and volume of pit embryo. The presence of increased chloride concentration and lower pH as a result of hydrolysis results in a critical pit solution that maintains active dissolution of the metal inside the pit and an increase in the metastable pit size. From this point on, the metastable pit can adopt two paths: the oxide film cover can rupture (perhaps caused by osmotic pressures created by the chloride concentration difference between the inside and outside of the pit embryo) and the in-rushing external solution can dilute the pit solution and repassivate the pit; or the pit embryo can attain such a size that the oxide ruptures and in-rushing solution does not have a significant dilution effect so that the pit embryo continues to grow and becomes a stable pit. The path taken will be determined by many factors, including the dissolution current (applied potential), geometry, and external solution concentration. This concept is essentially similar to the pit initiation model originally proposed by Galvele.⁵²  If the applied potential is sufficiently high, continued dissolution of the metal helps to maintain the critical concentration and no repassivation occurs. If a crevice is present or the surface roughness is high, the transport path is lengthened and dilution by external solution is minimized. Thus, the pit is stabilized. Scully, et al., expressed a similar concept in terms of a critical ratio of pit current to pit radius (assuming hemispherical pits).⁵³  The presence of a critical pit chemistry for maintaining active pit growth recently has been verified experimentally by Sridhar and Dunn.⁵⁴ 

If the model proposed by Galvele is valid,⁵²  the E_rp would be expected to decrease continually with the depth of the pit. This decrease of E_rp with pit depth would result from the lengthening of the transport path that would require less current (or potential) to maintain the critical pit chemistry. However, E_rp is an externally measured potential. The potential at the bottom of the pit (the pitting front) is decreased by ohmic resistance in the pit solution, which increases with pit depth.⁵⁵  Hence, these competing factors would ensure that, in deep pits or crevices, the E_rp would become constant above a certain pit depth. This was observed in the present case.

The detrimental effect of on lowering E_rp also can be explained by the previously discussed mechanism. It has been shown that, in the active condition, adsorbs as atomic sulfur on the metal surface.⁴⁵  Further, it has been shown that adsorbed sulfur increases from 5 to 12 times the active dissolution current, depending upon the alloy system.⁴⁵  In such a case, adsorbed sulfur will stabilize the pit embryo and prevent repassivation. Lower metal dissolution rates (lower applied potentials) then would be required to promote repassivation in the presence of ⁠. The detrimental effect of reduced sulfur species may be important because of the possibility of the presence of sulfate-reducing bacteria (SRB) near the HLW containers. Although the end products of SRB activity are sulfides, thiosul- fate can be produced as a transient species in the reduction of sulfate^56-58  have shown that the effect of thiosulfate in accelerating pit initiation is similar to that of sulfide in aqueous chloride solutions. In addition, the dissolution of sulfide inclusions in the metal also can result in the formation of thiosulfate in solution.⁵⁹ 

• Results of this investigation showed that the repassivation potential for alloy 825 and type 316L is dependent upon temperature and composition of the environment, in particular, the oncentration of Cl⁻ anions. Numerous test techniques designed to test the adequacy of the repassivation potential have shown that pitting and crevice corrosion cannot be initiated at potentials lower than the repassivation potential. In addition, after initiation, all localized corrosion ceased to propagate if the potential of the material was reduced below the repassivation potential. After repassivation, the corrosion rate of the material was determined by the passive current density.

• Long-term testing of alloy 825 indicated that the corrosion potential under air-saturated conditions increased over a period of 600 days to values higher than repassivation potential measured in short-term experiments. Crevice corrosion was observed when the corrosion potential was a few millivolts greater than the repassivation potential. Under these conditions, the propagation rate for localized corrosion will determine the lifetime of containers made of these materials. The long-term increase in corrosion potential is believed to be caused by the growth and aging of the passive film. Similar increases in potential were observed on specimens covered with a thermally aged oxide film over shorter periods of time. Many of the parameters that are necessary to calculate the corrosion potential in an air-saturated environment, such as exchange current densities for the anodic and cathodic reactions and transfer coefficients for the reactions, are not known for the substrates of interest. However, the measured values of corrosion potential can be used to calibrate the modeling of the evolution of the corrosion potential under repository conditions.

• To judge the performance of a given alloy or container design over the expected performance period of a HLW container, it is necessary to characterize the environment contacting the container. Some of the key parameters include the time-dependent composition of the aqueous environment, especially chloride concentration, oxygen concentration, and temperature profiles of the WP. Other important parameters are related to galvanic effects between various overpack materials. From these results, it may be concluded that, for a given environment, the repassivation potential is conservative and an effective parameter to predict the long-term performance of corrosion-resistant materials.

The authors acknowledge the helpful reviews and suggestions of V. Jain, B. Sagar, and B. Leslie.