Long-Term Evolution of Corrosion Potential of Carbon Steel in Alkaline Radioactive Waste Environments

Aging structures exist throughout our modern world, including highway infrastructure, water distribution pipelines, nuclear power plants, oil and gas production and transportation systems, and power transmission grids. The failure modes of these structures are influenced by their historical and current operations. Unanticipated failures may occur in these systems with the passage of time due to changes in environmental and material conditions. The management of aging structures beyond their original design life involves an evaluation of conditions that may lead to these failure modes and identifying effective mitigation methods.

At the Hanford, Washington site, approximately 200,000 cubic meters (55 million gallons) of radioactive wastes are stored in 177 carbon steel tanks, of which 28 tanks are double-shell tanks (DSTs) and the others are single-shell tanks (SSTs).¹  At the Savannah River Site (SRS), there are 43 active tanks classified into four types and containing about 135,000 cubic meters (37 million gallons) of waste as of 2013.²  Long-term stewardship of the tanks and associated equipment is needed to store, extract, and stabilize these wastes. The tanks contain layers of solids, sludge, and supernatant liquids, which are alkaline solutions containing nitrate (⁠⁠), nitrite (⁠⁠), phosphate (⁠⁠), carbonate (⁠⁠), sulfate (⁠⁠), chloride (Cl⁻), fluoride (F⁻), and, in some tanks, organics. Various heavy metals are also present in these layers. The carbon steel in some of these tanks has experienced localized corrosion and stress corrosion cracking (SCC). Decades of laboratory studies have yielded much engineering information on how to control localized corrosion and SCC by controlling the waste chemistry.^3-4  While chemistry control is the most practical approach to managing the tank integrity, it has also been demonstrated that electrochemical potential has an impact on the failure modes.^4-5  The waste chemistry control specifications are largely derived from relatively short-term, accelerated tests, in which the open-circuit potential (OCP) is relatively low (around −300 mV_SCE saturated calomel electrode, SCE).³  An understanding of the long-term evolution of the OCP of carbon steel under different waste chemistries is important in predicting potential failure modes and mitigating them. Corrosion monitoring programs, including corrosion potential measurements, of the tanks have been conducted in previous years.^5-6  The laboratory studies help in understanding the in-tank measurements.⁷  More importantly, mechanistic studies aid in extrapolating laboratory and field measurements into the future.

The occurrence of corrosion failure modes is determined by the value of corrosion potential, often with respect to a critical potential (Figure 1). The critical potential is a function of material and environmental factors. Typically, critical potentials exist for materials exhibiting passive behavior; whereas, for actively corroding materials, the change in corrosion rate is more gradual with an increase in corrosion potential. A large body of research exists that defines the critical potentials of different alloys in different environments.^8-17  Multiple regions of critical potentials, e.g., in the active to passive and passive to transpassive transition regions, can exist. In Figure 1, the critical potential is indicated schematically by a broad region. The OCP, also called the corrosion potential, can take different trajectories with respect to time, as indicated by Curves (a) to (d) in Figure 1. In the case of Curve (a), the rise in OCP is rather steep leading to the occurrence of localized corrosion or SCC rather rapidly. In the case of Curve (b), the increase in OCP occurs over a longer time, creating conditions for failure due to aging. Curve (c) shows a gradual increase in OCP that does not intersect with the critical potential region and therefore is not of as much concern from the point of view of localized failure modes. Finally, there may be cases where the OCP decreases with time as indicated by Curve (d). The evolution of the OCP along these trajectories is determined by the partial anodic and cathodic polarization curves. This is shown in Figure 1(b) in the form of the well-known Evans diagram plotting the potential against the logarithm of absolute current density of anodic and cathodic polarization curves. The ohmic resistance drop is ignored for simplicity. For a passive metal (indicated by the polarization curves P₁ and P₂), an increase in OCP occurs either due to the reduction of passive current density with time (P₁ to P₂) or an increase in cathodic current density (C₁ to C₂), or a combination of the two. This results in the shift in OCP from 4 to 1. An opposite shift in OCP occurs if the polarization curves vary with time from P₂ to P₁ or C₂ to C₁. It can also be seen that the shift in OCP for an active metal (indicated by anodic curves A₁ and A₂) is much less dramatic than that for the passive metal.

A study of the variation of the OCP with time must be accompanied by measurements of polarization behavior and other electrochemical parameters with time to understand the fundamental mechanisms behind such changes and develop a rational method to extrapolate into the future. This paper focuses on the long-term OCP of carbon steel in alkaline solutions containing predominantly and ⁠. A previous paper reviewed the critical potentials for localized corrosion and SCC of carbon steel in these environments.⁷ 

American Association of Railways Tank Car (AAR TC) 128 Grade B carbon steel (TC-128) was used in the experimental work as representative material to simulate the legacy steel, ASTM A515 Grade 60 carbon steel, that was used in the construction of some of the Hanford double-shell tanks, including 241-AY-102 (AY-102). The chemical composition of the TC-128 steel used and the specifications for the legacy ASTM A515 Grade 60 and AAR TC-128 Grade B tank car steels are shown in Table 1.

The nonradioactive simulants used in the research were based on the waste associated with AY-102 due to a leak that was discovered inside the annular region of this tank.⁶  The possible chemistries of the waste in the annular space were simulated using various scenarios including partial evaporation of the in-tank waste and equilibration with atmospheric CO₂. Some of these scenarios included information from the analysis of materials retrieved from the annular space. Information about the selected simulants along with their chemical compositions is shown in Table 2. The various simulants and their details are arranged in Columns 2 through 13 of the table. For each simulant, the following information is included: (1) type of waste simulated (supernatant, interstitial liquid, or a 50% mixture); (2) state of atmospheric CO₂ equilibration of the waste (none, partial, or complete); (3) partial drying of the waste (no or yes); (4) molar concentrations of the analytes present; (5) total organic and inorganic carbon contents; (6)  molar ratio; and (7) pH measured at about 50°C.

Most of the tests were conducted under quiescent air exposure. In a few tests, the effect of deaeration on the electrochemical behavior of carbon steel was explored. Deaeration was achieved by bubbling high purity nitrogen through the solution for at least 4 h prior to introducing the specimen and then bubbling throughout the test. Except for Sim 4 and 5, the pH change during the test was not significant (over 2 y, the decrease in pH ranged from 1% to 2% for most solutions, 6% for Sim 7, and about 13% for Sim 4 and 5, most of this change occurring in the first year). Measurements of OCP were also conducted on creviced and noncreviced specimens of steel in an actual tank waste to compare to laboratory simulants. The supernatant liquid from a tank, AP-102 was obtained. Measurements were made in hot cell facilities located in 222-S laboratories at Hanford at ambient temperature of approximately 28°C. An Ag/AgCl/Saturated KCl was used as the reference electrode. The reference electrode was immersed in the solution only when a measurement was taken.

For the electrochemical tests, cylindrical test specimens with a rounded nose (3.175 cm length and 0.478 cm diameter) were fabricated from plates of the steel. The specimens were polished to 600-grit finish, degreased in acetone, rinsed in deionized water, and dried prior to testing. Long-term OCP tests were conducted in 1-L PFA (perfluoroalkoxy fluoropoylmer) cells for time periods ranging from about 200 to about 650 d. The PFA cells were maintained at a constant test temperature of either 50°C or 77°C. Each cell was equipped with ports and compression fittings for insertion of replicate test specimens. Each test specimen was mounted on a threaded stainless steel rod sheathed in a PTFE (polytetrafluoroethylene) holder and completely immersed in the solution during the tests. A SCE was used as the reference electrode. The reference electrode was inserted into the test cell only when collecting OCP measurements and conducting electrochemical impedance spectroscopy (EIS) scans on the specimens. A Luggin capillary containing the test solution was used for maintaining the reference electrode near room temperature. The specimen was fully immersed in the simulant to create an exposed surface area of approximately 5 cm². Duplicate or triplicate specimens were tested for each environment. One or more specimens were removed from the solution after intermediate test periods and subjected to electrochemical polarization and Mott-Schottky (M-S) tests.

After transferring from the OCP measurement cells to another cell with a fresh simulant, the specimens for polarization and M-S tests were given 24 h to develop a stable OCP prior to conducting the electrochemical tests. Anodic and cathodic polarization tests were conducted at a scan rate of 0.17 mV/s. Anodic polarization tests were started at a potential of −50 mV_OCP and scanned in the positive potential direction until reaching a maximum anodic current density of 1 mA/cm². Cathodic polarization scans were started at a potential of +50 mV with respect to OCP and then scanned in the negative potential direction until achieving a maximum cathodic current density of 1 mA/cm². A Gamry Reference 600^† potentiostat and a conventional three electrode configuration were used for performing electrochemical measurements.

The EIS data were collected under potential control using a sine wave excitation of 10 mV (root mean square [RMS]) vs. OCP. The starting frequency of the EIS scans was 100 kHz. The final frequency of the EIS scans was 0.001 Hz except for measurements that were associated with short aging times (less than 48 h). For short aging periods, the EIS scans were limited to a final frequency of 0.01 Hz due to the lengthy measurement times required for lower frequencies. Most of the EIS measurements were conducted in the same test vessel that was used for the long-term OCP monitoring (aging) of the specimens. The validity of the impedance data was checked using Kramers-Kronig (K-K) transformation for a steel specimen immersed in Sim 9 solution. The imaginary part of the impedance was calculated from the real part of the measured impedance data using Equation (1):

where Z′ is the imaginary impedance, Z is the real impedance,  ω is the frequency at which Z′ is calculated from Z, and x is any frequency in the spectrum other than the frequency at which Z′ is calculated. The value of Z′ was calculated for all of the experimental frequencies and compared to the experimentally measured Z′. Because the spectra are collected using finite frequency intervals in the range of 1 mHz to 100 KHz, there will be some integration error between calculated values and Equation (1), especially at the ends of the frequency spectrum. The agreement between the measured and calculated Z′ was within 5% over most of the frequencies. In addition, the linearity and stability of the impedance data were evaluated by conducting the EIS tests using 5 mV, 10 mV, and 20 mV amplitudes and scanning frequencies from low to high and backward. The measured impedance data were in good agreement, with the highest difference in imaginary impedance at 100 mHz of about 16%. These tests indicate that the measured EIS values reflected the corrosion phenomena accurately (there were no spurious signals) and the analysis of EIS data can confidently assume linearity and stability.

The M-S tests were performed sequentially on the same specimen to collect two data sets based on the DC potential step direction: the first M-S test stepped the DC potential in the negative direction starting from the OCP and ending at −600 mV_SCE; the second M-S test was performed immediately afterward by stepping the DC potential in the positive direction beginning at −600 mV_SCE and ending at +400 mV_SCE. The potential was stepped in 25 mV divisions for each sweep direction. At each potential step, a sinusoidal voltage of 5 mV (RMS) was applied. In the preliminary M-S testing, three different frequencies were examined: 5 kHz, 3 kHz, and 1 kHz. The frequency of 1 kHz was ultimately selected based on the better degree of linearity that was found in the C⁻² vs. E curves as well as the widespread use of this frequency in the literature.

The SCC susceptibility of the steel was evaluated in three categories of conditions: (1) steel test specimens exposed fresh to a waste simulant at an applied potential close to the OCP after 500 d to 600 d; (2) steel test specimens exposed to a waste simulant for 200 d at OCP and then tested at OCP; and (3) steel test specimens that had been exposed to a waste simulant for 200 d and then tested at an applied potential corresponding to OCP after 500 d. SCC susceptibility was evaluated using slow strain rate (SSR) tests per ASTM G129-00¹⁸  using cylindrical tensile specimens at a constant extension rate of 2.54 × 10⁻⁶ cm/s, unless otherwise noted. For the 2.54 cm (1-in) gage length specimen, this corresponds to an initial strain rate of 10⁻⁶ s⁻¹. The exposed area of the SSR test specimen was approximately 60 cm². The specimens exposed to a waste simulant for long times are called “aged” in this paper. In the second case, an unstressed sample 35 cm² in area was also exposed to the simulants for the same time interval and then coupled to the SSR test specimen to stabilize the potential. For each test, the specimen was placed in a Teflon^† (PTFE) test cell and a load was applied using clevices that entered the cell through PTFE fittings. This assembly was then inserted into a load frame, after which the solution of interest was introduced and heated to the desired test temperature (e.g., 50°C). Tests were conducted at OCP or at an applied potential with respect to a SCE reference, which was maintained at room temperature using a Luggin probe/salt bridge filled with the test solution. A platinized niobium wire loop was used as a counter electrode for the tests at applied potential. All tests were performed under quiescent conditions or deaerated with high purity nitrogen.

All test specimens were pulled to failure. Examination of the specimens after failure consisted of stereographic optical examination at 10× to 63× magnification and imaging in a scanning electron microscope (SEM). The fracture surface of every test sample was examined using the SEM to identify regions of intergranular fracture, indicative of SCC.

The results of the SSR tests are shown in Figure 2. The specimens tested at the short-term OCP values (i.e., labeled fresh solutions) did not show any indication of SCC. This was evident both in the high ratios of time to failure and the absence of secondary cracking. The fracture surfaces revealed an essentially ductile failure mode, as shown in Figure 3(a). Almost all of the specimens that were polarized in the fresh solutions shown in Table 2 or that were aged in the same solutions for over 200 d showed SCC as indicated by the fracture surface shown in Figure 3(b) and by optical observation of secondary cracks on the gage section. In most cases, there was a significant decrease in the time to failure compared to that in air. The severity of SCC was greatest for specimens polarized to the long-term OCP in fresh solutions, lesser for specimens exposed to solution for long times and polarized to the longer-term OCP, and least for the long-term exposed specimens tested at OCP. The specimen tested in a fresh solution of Sim 8 at applied potential did not show SCC. This simulant has a high molar ratio of 24.3 (Table 2). This is consistent with previously reported results⁷  in simulants of ratios above two.

The time evolution of OCP of the steel in a waste simulant (Sim 8) is shown in Figure 4 for 77°C. Replicate specimens were tested in all of the simulants and some were retrieved for evaluation before the OCP had completely stabilized. The measured dispersion in OCP was small compared to the time-based changes in OCP (Figure 4). The OCP increased sharply in the first month of exposure after which there was a much smaller increase over longer time. When a steel specimen exposed for a long time was reintroduced into a freshly prepared solution, the OCP attained the same value as observed in the beginning of the long-term exposure tests (Figure 4). In contrast, when a freshly prepared specimen was introduced into a solution in which long-term exposure studies were conducted, the OCP remained at −446 mV_SCE to −300 mV_SCE for an hour (Figure 4). These findings confirm that the increase in OCP was not related to changes in the solution, but changes in the steel surface. The changes in OCP over long periods of time are shown in a logarithmic plot for all of the simulants (Figure 5) and follow the general relationship as shown in Equation (2):

Equations (3) and (4) suggest that the rate of increase in OCP with time decreases with an increase in pH, at least within the test duration. As indicated in a previous paper,⁷  the OCP at the end of the experimental duration for each simulant shows a strong dependence on pH (Figure 6). No significant dependence on ratio was found. The results shown in Figure 6 include previous findings of Gui, et al.¹⁹  The OCP was also not significantly affected by deaeration with nitrogen, as shown in Figure 7. The change in OCP of steel in an actual radioactive tank waste over a 9-month period is shown in Figure 8. The increase in OCP in this supernatant is similar to that found in various simulants (Figure 5). This suggests that radiation did not play a significant role in the evolution of OCP.

The presence of high-temperature oxide scale (referred to as mill scale) can increase the OCP. Gui, et al.,¹⁹  assembled coupons on a steel threaded rod and the rod was hung in air in a furnace using two alumina crucibles to avoid contacting the furnace interior wall. The specimens were then heated to 700°C for 20 min and furnace cooled. The OCP values of the artificial mill-scale surface are plotted as a function of OCP values of machined surfaces for various solution chemistries in Figure 9, with the 45° line representing equal values of the two surface conditions. The mill-scaled samples exhibited higher OCP values than the machined samples. The OCP was also measured in solutions of and containing various metal ions, anions, and organics (Table 3). It can be seen from Figure 9 that the addition of metal ions resulted in an increase in OCP, whereas the addition of anionic species and organics resulted in a decrease in OCP. The addition of gluconate decreased the OCP much more than that of the other organics, but the combination of gluconate and citrate resulted in the largest decrease in OCP.

The effect of exposure time at OCP on anodic polarization behavior of steel in Sim 9 is shown in Figure 10. The forward and reverse scans are part of the same cyclic potentiodynamic polarization test, but are exhibited separately for clarity. Several features are worthy of note: (1) there is an increase in the potential at zero current (transition between anodic and cathodic curves) with time of exposure; (2) the forward anodic current density at potentials ranging from −200 mV_SCE to +500 mV_SCE decreases with an increase in exposure time at OCP; (3) the reverse anodic current density is smaller than the forward anodic current density (negative hysteresis); and (4) there is a cathodic peak at 0 mV_SCE on the reverse anodic curve (especially visible for the 28 d and 63 d curves) that increases with exposure time. The forward anodic current densities at 100 mV_SCE are plotted for two different simulants as a function of exposure time in Figure 11. A power-law dependence of anodic current density in the passive region with exposure time at OCP is observed, with the time exponent ranging from −0.7 to −0.5.

The cathodic polarization curves as a function of time of exposure at OCP are shown in Figure 12. In the potential range of about +50 mV_SCE to −300 mV_SCE, the cathodic current density was dependent on both time at OCP as well as nitrogen deaeration. The current density was lower for the nitrogen deaerated solution than for the air-exposed solution for similar exposure times. For the air-exposed solutions, the cathodic current density in this potential regime increased slightly with exposure time. In this potential regime, the polarization behavior may correspond to a combination of the oxygen reduction reaction and reduction of a surface film (perhaps Fe(III) to Fe(II) valence state within the oxide film). The former is dependent on dissolved oxygen and the latter is dependent on the time of exposure at OCP. The cathodic curves in the potential region from −400 mV_SCE to about −800 mV_SCE were independent of the time of exposure at OCP and deaeration. This potential region may be related to nitrate or reduction reaction and the rate limiting step may be related to nitrogen diffusion away from the surface.

The EIS data were collected on the long-term OCP test specimens in the same solution at various time periods. The EIS data were analyzed using an equivalent circuit consisting of a solution resistance (R_s), coupled in series to a parallel circuit consisting of a polarization resistance (R_p), and a constant phase element (CPE). The use of a CPE instead of a capacitor to model the interfacial properties of the film can be rationalized by the heterogeneity of surface passive films in the directions parallel and perpendicular to the specimen surface as well as irregularities in surfaces, which leads to a distribution of time constants.^20-22  Furthermore, the microstructure of the steel, combining ferrite and iron carbide phases contributes to the heterogeneity of the surface. As can be seen from Figure 13, the phase angle is relatively constant over a large range, suggesting that the use of a CPE is appropriate for this system in modeling the EIS response. However, the relationship between the interfacial capacitance from the CPE constant depends on assumptions regarding the distribution in the decay time constant of the electrochemical interface.²²  In this paper, the relationship proposed by Brug, et al.,²³  is used, with the modification that the CPE constant, Q, used by Brug, et al.,²³  is the inverse of the Q used²⁰  in Equation (5):

The value of α that provided the best fit to the data was between 0.85 and 0.96. The absolute value of C_int depends on the form of Equation (5), but the trends in the calculated C_int are similar. The effect of exposure time on the polarization resistance (R_p), and capacitance (C_Int) of steel in two simulants is shown in Figures 14 and 15. Generally, the R_p increases initially with time; whereas, C_Int decreases initially with time.

Investigation into the semi-conducting properties of the aged steel passive films was performed using the Mott-Schottky technique. The M-S technique assumes that the space charge capacitance is the dominant term in the overall measured capacitance, the passive film is at steady-state during the measurement, and that the capacitance depends on the applied DC potential according to Equation (6):

where C is the capacitance, ε is the dielectric constant of the film (assumed to be a constant value of 12), ε₀ is the permittivity of vacuum, N_D is the donor density, E is the applied potential, E_FB is the flat band potential, k is Boltzmann’s constant, q is the electron charge, and T is temperature in K. From the C⁻² vs. E plot, the value of N_D can be obtained based on the magnitude of the slope and assuming a value for the passive film dielectric constant. The ± symbol in Equation (3) indicates that the slope can be positive or negative depending on whether the film behaves as an n-type or a p-type semiconductor, respectively. Figure 16 shows the M-S results that were obtained on TC-128 steel after various aging times in Sim 9 at 77°C. The plot on the left in Figure 16 shows the C⁻² vs. E curves that were obtained in the negative potential step direction while the plot on the right displays the C⁻² vs. E curves that were collected in the positive stepping direction. The positive slopes that were observed in the M-S results indicate n-type semiconducting behavior of the passive films over a wide potential range. The magnitude of the slopes appeared to increase over short aging periods (7 d to 28 d) but then showed a substantial decrease over a longer aging period (162 d). The estimated donor density corresponding to the 28-d aging time was around 4 × 10²¹ cm⁻³. In comparison, N_D was approximately 4 × 10²² cm⁻³ after 162 d of aging time. These values are roughly of the same order of magnitude as those reported in the literature for iron.^24-25  It must be noted that the absolute values of donor densities are not as important as the trends, because many assumptions are made in deriving these values.

Most of the studies reported in the literature have not examined the evolution of OCP over the length of time reported in this paper, despite claims of attaining “steady-state.” As seen in Figures 5 and 9, the OCP can continue to increase over several months. It is obvious that the OCP cannot rise indefinitely. Therefore, the central question in life prediction of steel in this system is: “what is the true steady-state value of the OCP?” The radioactive waste tanks have been exposed to the solutions over a much longer period than the specimens in the long-term laboratory tests. Therefore, to compare the laboratory results with tank corrosion potential monitoring results, a better understanding of the effects of microstructure, surface condition of the steel, and environmental chemistry is necessary. Some of these factors are discussed in this paper.

The OCP behavior over time is determined by the alloy composition, microstructure, surface condition, and the environmental composition, especially the presence of redox species. In the case of carbon steels, it is known that higher carbon contents in the ferrite lead to increased passive current density in nitrate solutions²⁶  and thus lower OCP. Zapp and Van Zee²⁷  showed that anodic passive current density was lower for pure iron than ASTM A537 carbon steel in containing alkaline solutions and the corresponding OCP were higher for pure iron than carbon steel. They also showed that the presence of and increased the OCP significantly compared to pH buffer only solutions. Sarafian²⁸  examined the role of microstructure on the anodic polarization curves of ASTM A 516, Grade 70 steel in nitrate solutions at various pH levels. At pH values of less than 6, the quenched microstructure (less carbides) showed lower anodic current densities; whereas, at a pH of 13.7, the quenched microstructure showed a higher anodic current density. Unfortunately, the effect of carbon content and steel microstructure on OCP has not been investigated sufficiently. In the present investigation, the steels tested in the lab approximated the carbon content of the steel used in the tanks. However, the effects of welding and other heat treatments on OCP have not been studied.

As shown in Figure 9, the specimens with laboratory produced scale showed higher OCP values in alkaline nitrate environments. This was also observed in the case of steels in soil environments²⁹  and atmospheric wet-dry conditions.³⁰  The role of mill-scale in increasing the potential of steel pipelines to the critical SCC potential region is also recognized in NACE International and other standards.^31-32  It has been shown that under these conditions, the presence of mixed valent oxides on the surface results in redox reactions involving transition from Fe(III) to Fe(II).³⁰  The OCP value for the surface-scaled and machined surfaces attained similar values when metallic cations were added to the solution (Figure 9). The addition of ferric species to the solution at a pH of greater than 13 will form iron oxy-hydroxide precipitates on the surface. Therefore, the higher value of OCP in the presence of metal cations, such as Fe(III), may reflect the presence of redox reactions involving Fe(III)/Fe(II) mixed valent precipitates on the surface. Heat treating the steel to produce thermal scales, without altering the microstructure, may be one method to accelerate exposure effects.

Deaeration with N₂ reduced the OCP only slightly with respect to air-exposed solution (Figure 7). This suggests that the cathodic reaction is supplied by other redox species, in addition to oxygen. The equilibrium potentials of various redox reactions involving and (Figure 17) shown in relation to the measured long-term OCP values suggest that and reduction to N₂, nitrous oxide (N₂O), and Ammonium Hydroxide (NH₄OH) may all be candidates. The reduction to N₂ and N₂O involves 3 and 2 electrons per mole, respectively. In contrast, reduction involves 4, 5, or 8 electrons and, therefore, will not be kinetically favorable. Thomas and Nurse³³  found that reduction contributes about a tenth of the cathodic currents in comparison to oxygen reduction, whereas reduction occurred in the potential region close to −600 mV_SHE (standard hydrogen electrode, SHE) (−842 mV_SCE). Zapp and Van Zee²⁷  showed that 0.1 M alone increased the OCP of steel by about 600 mV, whereas 0.1 M alone increased the OCP by only 200 mV. A mixture of 0.1 M + 0.1 M showed intermediate values of OCP increase. Their study, albeit for only about 8 h, showed that the OCP increase on carbon steel was less than for pure iron. Furthermore, the OCP increase on carbon steel was dependent on concentration, whereas the OCP increase on pure iron was not dependent on concentration. A previous study by Li, et al.,³⁴  showed that, in air-exposed solutions with only and no ⁠, the short-term OCP tended to drift downward, suggesting a passivation role for ⁠. The dual role of in both promoting cathodic reactions and passivity and the role of carbon in steel need further exploration. Burstein, et al.,⁶³  have suggested that nitrite reduction to nitrogen takes place on an austenitic stainless steel. Whether a similar phenomenon can take place on ferritic steel requires further evaluation.

The effects of organic species on OCP depends on the type of organic species present (Figure 9 and Table 4). Organic species such as acetate and formate, mainly affect the pH, and in the highly alkaline environments, are not expected to have any effect on the cathodic or anodic reactions. The chelating agents, such as EDTA, can complex dissolved Fe(II) species and may increase the anodic kinetics. However, the effect of EDTA on the anodic kinetics appears to be small and mainly related to the removal of the outer, porous layer of the passive film.^35-36  The effects of other chelating agents on anodic dissolution of iron have not been studied.

Gluconate is a well-known complexant of metal species as well as a corrosion inhibitor for steel.^37-41  Interestingly, the effect of sodium gluconate appears to be concentration dependent.⁴¹  Below approximately 10⁻² M, sodium gluconate reduces the passive current density and increases the critical pitting potential in a 0.1 M chloride solution at a pH of 8.6. Above 10⁻² M, sodium gluconate appears to increase the passive current density and, therefore, decrease the OCP. Touir, et al.,⁴¹  showed that, in simulated cooling water (460 ppm by weight chloride and pH of 7.3), addition of sodium gluconate in the range of 10⁻⁴ to 0.1 M increased the corrosion potential and reduced the corrosion rate. They also showed that the capacitance decreased significantly with addition of gluconate, suggesting adsorption of gluconate on steel (at this pH, passivity is not expected for steel). Unfortunately, no studies of the effect of gluconate on the electrochemical behavior of steel at the pH of interest to tank waste storage (pH >12) have been reported.

As shown in Figure 12, in the 0 mV_SCE to −400 mV_SCE region, the cathodic current decreased with deaeration, suggesting that oxygen reduction contributed partially to cathodic currents. However, the cathodic current density also increased slightly with time, which indicates that some other surface activated process also contributed to cathodic currents. Because oxygen reduction is typically diffusion limited and its bulk concentration is not expected to change with time, especially in quiescent conditions, the cathodic contribution from oxygen reduction is not expected to be time dependent. It is possible that the cathodic reduction of the outer layers of the passive film contributed to the slight increase in cathodic currents in this regime. As the cathodic current density increased only slightly (by a factor of 1.5) with time in the 0 mV_SCE to −400 mV_SCE region, whereas the passive current decreased by almost two orders of magnitude (Figure 11), the latter is the controlling factor in the increase in OCP over time. Thus, predicting the long-term evolution of OCP involves understanding the changes in the passive film over time.

There is considerable literature on the passivity of iron and other alloys.^24-25,42-60  These studies have focused on: (1) the structure and composition of the passive film; (2) the electronic properties of the film; and (3) the growth rate of the film under the influence of point defect movement, potential, and environment. These factors are interrelated, e.g., the structure and composition of the film determine its electronic properties and the point defect movement. Although the evolution of OCP is integral to the three aspects of the passive film, it has seldom been the focus of the passivity studies.

The passive film on steel is generally proposed as a bilayer structure (with an Fe₃O₄-like inner layer and an Fe₂O₃-like outer layer)^25,46-54  or a graded oxide with an iron-rich structure close to the metal-oxide interface and an oxygen-rich structure close to the oxide-solution interface. There may also be the presence of hydrated oxides (or oxyhydroxides) on the outer surface of the passive film. The oxides are nonstoichiometric with cation vacancies close to the metal-oxide interface to compensate for the excess metal cation and anion vacancies close to the oxide-solution interface to compensate for the excess oxygen anion. Many different models for oxide growth have been proposed,²⁵  involving ion migration through the oxide film under electric field, electron tunneling, and place exchange of metal cations and oxide anions within or at the interfaces. All of these models result in either an inverse or direct logarithmic growth law of the form:

where x is the passive film thickness, t is the time, and A, B, A′, and B′ are constants derived from theory or experiments. In either of the logarithmic laws, the oxide thickness does not attain a steady state, but at extremely long time periods Equation (7) is not tenable because of negative thicknesses. For thick oxides, diffusion of ionic species can be rate determining, leading to a parabolic growth kinetics. Although oxide growth can be important in determining OCP at short durations, it cannot explain the long-term evolution of OCP. For example, the measured capacitance (Figures 14 and 15) decrease initially, but remains relatively constant over long periods of time. As capacitance is inversely proportional to film thickness (assuming a parallel plate capacitor model), the initial increase in OCP may be due to film growth, but not the long-time slower increase. At long time periods, the growth of the oxide film is balanced by the dissolution of the film at the film-solution interface.⁶⁰  Based on the cation and anion vacancy transport under a concentration and potential gradients in the film, Macdonald and Urquidi-Macdonald⁶⁰  derived steady-state diagnostic equations for passive film thickness and current density as a function of applied potential. Although this provides an attractive method for estimating the steady-state anodic current density (and from there a steady-state OCP), the constants involved in their equations must be derived from experiments. As almost all passivity studies reported in the literature are for relatively short periods of time (typically a few hours), the empirically derived constants are not necessarily predictive. Another approach may be to derive some of these parameters from first-principles modeling.⁶¹  However, it is difficult to escape the need to fit the model parameters to experimental data. Therefore, long-term data are still necessary. A possible solution to the estimation of parameters involved in the steady-state passive film models is using exposure tests over several time intervals to determine whether they attain asymptotic values.

A second factor in the evolution of the OCP is the change in the electronic character of the film over time. Azumi, et al.,²⁴  found that, for pure iron in borate buffer solution (pH = 8.4), the film thickens with applied potential and passivation time while remaining an n-type semiconductor, while the donor density of the n-type semiconductor was about 10²⁶ m⁻³ and decreased with time over approximately 24 h. In contrast, for an iron passivated in phosphate buffer (pH = 6.5), the donor density increased with passivation time. They suggested that the reduction in donor density in borate buffer was due to decreasing defect density as the film grew, whereas the increase in donor density in phosphate buffer was probably related to the specific incorporation of phosphate in the passive film. In the present case, the donor density increased with time and may be hypothesized to be due to the incorporation of or in the passive film. Surface-enhanced Raman spectra (SERS) generated on iron passivated in pH 10 solutions of nitrate and bicarbonate showed similar spectra related to the presence of γ-FeOOH and did not indicate the presence of nitrate or carbonate in the film.⁴⁹  However, the time of exposure in the passivating range in this study was extremely short of the order of 15 min. It is also likely that the changing ionic and vacancy populations in the film may be responsible for the changing electronic character of the film and the OCP evolution.⁵⁹  The increase in donor density, assuming that Equation (6) is valid, cannot explain the increase of OCP through a decrease in passive current density. Another possibility is that the character of the passive film changed over time from n-type to p-type semiconductor, which was reflected in the slow decrease of M-S slope from positive to negative. The calculation of donor density using Equation (6) assumes that single dopant state exists in the semiconductor resulting in a linear behavior of 1/C² vs. E. If multiple donor states exist, for example one near the conduction band (called surface state) and the other near the valence band (called deep state), then nonlinearity of the 1/C² vs. E will be observed.⁶²  Although Figure 16 is consistent with a single dopant level at any given time, the change in slope could result from changes in dopant level. The calculation of donor density also assumes that the dielectric constant, ε, in Equation (6) is invariant. However, time-based changes in passive film composition and defect structure may result in changes in dielectric constants that are not considered in the donor density calculations.

A bounding approach to estimating the steady-state OCP may be the estimation of the exchange current densities of various cathodic reactions responsible for the OCP. This is because, the intersection of cathodic and anodic polarization curves cannot go below the exchange current density of the cathodic reaction. Unfortunately, exchange current densities for oxygen and reduction reactions for these systems with a long-term passive film are not readily available. These approaches should be explored in the future.

This paper presents a multiyear study of the evolution of OCP of carbon steel exposed to concentrated containing alkaline solutions involved radioactive waste storage.

• •

  In most alkaline to environments, the OCP continues to increase in a logarithmic fashion with time. The rate of increase and the long-term OCP in these systems are mainly functions of pH.

• •

  SCC susceptibility of steel at the long-term OCP values in solutions without organics is higher than at short-term OCP values.

• •

  The OCP change is also affected by other environmental constituents and the steel surface condition. The presence of certain organics, such as gluconate, significantly reduces the OCP and leads to a decrease in OCP with time. The presence of surface oxide scales on steel leads to an increase in OCP.

• •

  The increase in OCP is predominantly determined by a reduction in passive current density of the steel, which decreases as a logarithm of time. The increase in cathodic current density with time is far less significant.

• •

  Electrochemical impedance measurements indicate that the polarization resistance increases with time, whereas the interfacial capacitance initially decreases, but attains a constant value over time, suggesting that the passive film thickness has also attained a steady-state value.

• •

  Mott-Schottky measurements indicate that the passive film initially exhibits an n-type semiconductor behavior whose slope decreases with time. The corresponding donor density increases with time.

The authors gratefully acknowledge the assistance of K.M. Sherer and J. Gerst in conducting the tests. Discussions with Ramgopal Thodla are gratefully acknowledged. The work reported in this paper was performed under sub-contract Nos. 53247, 58792, and 61541 with Washington River Protection Solutions, LLC in support of the U.S. Department of Energy. The authors also acknowledge the discussions with Ted Venetz, Kayle Boomer, Leon Stock, Bruce Wiersma, Russ Jones, Scott Lillard, and Brenda Garcia-Diaz.