Accepted Manuscript
Title: The role of surface deformation in the oxidation
response of Type 304 SS in high temperature deaerated water Authors: K.B. Fisher, B.D. Miller, E.C. Johns, E.A. Marquis
PII: S0010-938X(17)32290-4
DOI: https://doi.org/10.1016/j.corsci.2018.06.033
Reference: CS 7589
To appear in:
Received date: 28-12-2017 Revised date: 7-6-2018 Accepted date: 29-6-2018
Please cite this article as: Fisher KB, Miller BD, Johns EC, Marquis EA, The role of surface deformation in the oxidation response of Type 304 SS in high temperature deaerated water, Corrosion Science (2018), https://doi.org/10.1016/j.corsci.2018.06.033
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
The role of surface deformation in the oxidation response of Type 304 SS in high temperature deaerated water
K.B. Fishera, B. D. Millerb, E. C. Johnsb, and E. A. Marquisa*
a University of Michigan, 2300 Hayward St, Ann Arbor, MI 48109
b Bechtel Marine Propulsion Corporation, 814 Pittsburgh-McKeesport Blvd, West Mifflin, PA 15122
* Corresponding Author: [email protected], 734-764-8717
Graphical Abstract
Highlights:
The microstructure and microchemistry of the oxide depend on the surface condition.
Oxidation mechanisms leading to the different oxide properties are proposed.
On deformed surfaces, a nanostructured oxide develops due to selective oxidation.
On electropolished surfaces, an equilibrium oxide phase develops uniformly.
Abstract
The oxide products found on austenitic stainless steels after high temperature exposures in nuclear reactor environments are well documented, but the mechanisms for the oxide formation are still ambiguous. One issue of practical importance is the role of surface deformation on the oxidation response. To address this question, short high-temperature water exposures were conducted on electropolished and ground surfaces of Type 304 stainless steel. Characterization of the developing oxide scales using transmission electron microscopy and atom probe tomography revealed significantly different responses from the two surface finishes. In the absence of sub-surface deformation, the oxide scale grows as an equilibrium phase of uniform composition with accumulation of Cu to the metal/oxide interface. In the presence of significant sub-surface deformation, a non-equilibrium nanostructured oxide scale develops as a result of selective oxidation along fast diffusion paths.
ACCEPTED MANUSCRIPT
Keywords: stainless steel, atom probe tomography, Raman spectroscopy, SEM, STEM, reactor conditions
1. Introduction
Type 304 and 316 stainless steel (SS) alloys have been widely incorporated into components of pressurized water nuclear reactors (PWRs) and boiling water nuclear reactors (BWRs), and therefore are subject to high temperature aqueous corrosion. Thus, understanding the corrosion processes of 300 series SS in high temperature water has been an area of practical and scientific pursuit for several decades.
In high temperature aerated or deaerated water, a dual layer oxide forms on the surface [1-10]. The outer layer is composed of discrete oxide particles. In oxygenated water, these particles are a mix of magnetite (Fe3O4) and hematite (Fe2O3)[5, 9], while in deaerated water, Fe3O4 ornickel ferrite(NiFe2O4) phases are favored [1, 3, 4, 9, 11]. The reported composition of the inner oxide layer varied widely, which is likely a result of the wide variety of techniques used to measure it, along with differences associated with the test environment. However, despite these differences, prior results typically concluded that the inner oxide layer is a nanocrystalline mixed spinel [1, 4, 8-10] though it is sometimes considered to have an M2O3 component [7, 9, 11, 12].
Despite a significant amount of study, the mechanisms by which this dual layer oxide forms are ambiguous.
In an aqueous environment dissolution/precipitation mechanisms and solid state reactions have been discussed [13]. In the case of Type 304 SS, the inner layer is universally considered to grow by a solid state reaction requiring the diffusion of oxygen to the oxide/metal interface [1, 3, 5-8, 14-16]. The outer oxide was reported by many to occur by a dissolution/precipitation reaction [5, 7, 8, 10], while some concluded it occurs by solid state outward growth at the solution/oxide interface [3, 16]. Evidence for the former includes particles growing on top of one another [5] and a lack of particles in unsaturated water [8].
Evidence for the latter includes the fact that metallic cations rejected from the inner oxide appear to be incorporated into the outer oxide film [3]. Both mechanisms may operate [4], with relative contributions depending on additional environmental factors, such as water flow rate [8] or reactor vessel material. For instance, Fe and Ni autoclaves have been shown to result in incorporation of dissolution products into the oxide, while experiments in Ti autoclaves show little Ti pickup [17, 18]. While the water chemistry determines the types of oxides that form, there is no clear evidence that it influences the mechanisms by which the dual-layer structure forms.
ACCEPTED MANUSCRIPT
Even with some degree of ambiguity regarding the formation mechanisms, the oxide products that form on 300 series SS in high temperature aqueous environments have been characterized for a variety of temperature and water chemistry conditions. However, the role of material processing factors, are less well understood despite their practical relevance. Prior deformation, in the form of high dislocation density and deformation bands, is often present in the bulk, or near the surface, due to component fabrication processes involving cold working and surface machining. Bulk deformation at levels as low as 5% cold work (CW) has been linked to increased susceptibility to stress corrosion cracking [19]. However, the role of material deformation in the oxidation process itself is still unclear. Several authors reported a marked decrease in corrosion rate on specimens electropolished or hand polished to a mirror finish compared to ground or machined surfaces [4, 16, 20, 21], which was associated with the formation of a more uniform, non-porous, protective layer, while deformation has been associated with enhanced diffusion rate of oxygen into the metal causing an increased oxidation rate [14]. On the other hand, others concluded that deformation reduces the corrosion rate [2, 10, 22]. This has been attributed to increased rate of Cr diffusion along deformed regions, resulting in a more Cr-rich protective oxide film, and a commensurate inhibition of Fe and Ni diffusion through the Cr-rich oxide [10, 22]. Deformation and surface roughness were also implicated in the outer oxide formation, as mechanically polished (deformed but smooth) surfaces appeared to exhibit larger particles than both electropolished (undeformed and smooth) or ground (deformed and rough) surfaces [3, 4, 10]. It is apparent that the effect of sub-surface deformation is not understood.
Therefore, the current work aimed to clarify the role of surface deformation in the early stages of high temperature aqueous corrosion processes of Type 304 SS and elaborate on the mechanisms of scale formation, by comparing the initial stages of formation of the oxide scale on electropolished surfaces to those formed on surfaces deformed by mechanical grinding. Detailed characterization of the oxide scale using a combination of complementary techniques including atom probe tomography (APT), Raman spectroscopy, and transmission electron microscopy (TEM) illuminated key differences in the development of the oxide scales. From these observations, mechanisms of oxidation relating to the underlying microstructure were proposed.
2. Experimental
Corrosion coupons with nominal dimensions of 12.7 x 12.7 x 3.18 mm were fabricated from a heat of annealed Type 304 SS, with composition shown in Table 1. One side was ground by hand with 220 and 320 grit silicon carbide sand paper, resulting in a rough surface and underlying surface deformation. While direct measurement of the level of deformation induced on the surface was not obtained, the extent of the
ACCEPTED MANUSCRIPT
deformed layer could be identified in TEM images and was on the order of several hundred nanometers, and in some places exceed 1 µm. The opposite side was ground by hand with sequential steps down to 1200 grit sand paper, and then a circular region 1 cm in diameter was electropolished at room temperature in a perchloric acid/acetic acid mixture. The electropolished surface was smooth and adequate material was dissolved to remove the deformation layer induced by grinding. By having both surface finishes on the same specimen, it was assured that the exposure conditions were equivalent for each exposure time.
Exposures were conducted in an environment meant to simulate normal pressurized water reactor operation, without consideration for the role of irradiation exposure or chemistry deviances observed during startup or shutdown procedures, such as increased oxygen level. An Inconel 690 autoclave was used for exposures in deaerated water (35 cc H2/ kg H2O) pH adjusted to a pH of 6.6 at 288 °C and a pressure of 1200 psi.
Corrosion coupons were exposed for times ranging from ~10 minutes to 72 hours. To accomplish such short exposure times while avoiding environmental exposure during heat up and cool down, the following procedure was used. The autoclave was purged with nitrogen gas, and then a calculated amount of room temperature deaerated water was added. This initial water was heated up such that upon thermal expansion it remained just below the test coupon. Water was then added to submerge the coupon while actively heating the autoclave to maintain the desired test temperature. Upon reaching the desired exposure time, the water was cooled as rapidly as possible, causing it to contract below the level of the specimen, and drained to minimize further exposure. Level limit indicators were used to monitor the level of the water compared to the sample coupons to ensure that the desired exposure conditions were met. The test apparatus and procedure was evaluated on one coupon that was never fully submerged to verify oxide growth from condensation on the unsubmerged specimen was minimal. Subsequent exposures were conducted for 0.17, 0.5, 1, 4, 56, and 72 hours. An additional sample with a machined surface (significant subsurface deformation) was oxidized for 2200 hours to understand the oxide growth after longer time periods and in the presence of greater and deeper sub-surface deformation. The findings are discussed in relation to the ground specimens.
Raman spectra were obtained using a Renishaw inVia microscope with a 532 nm laser, a laser power of 10
%, and a minimum of 20 scans on each surface to reduce noise. APT and TEM specimens were fabricated using a site-specific liftout technique [23] on either an FEI Helios 650 Nanolab or FEI Nova 200 Nanolab dual beam scanning electron microscope (SEM)/focused ion beam (FIB) equipped with an Omniprobe micromanipulator. Scanning TEM (STEM) images were obtained on a Hitachi HD 2300A STEM operated at 200 kV and used to measure the average oxide thickness on each specimen and examine the morphology of the inner oxide layer. Oxide thickness was measured every 50 nm across the TEM specimen using
ACCEPTED MANUSCRIPT
imageJ [24], resulting in at least 130 measurements per specimen. APT was performed on a Cameca LEAP 4000X HR in laser mode using a pulse energy of 50 pJ, a detection rate of 5 atoms per 1000 pulses, a base temperature of 50 K, and a pulse rate of 200 kHz. Duplicate samples from most oxidation conditions were obtained and used to determine measurement error by averaging compositions across comparable regions of multiple specimens and finding the standard deviation.
APT reconstructions and analyses were performed using Cameca IVAS version 3.6.12. Volumetric reconstructions were created using the voltage evolution method to determine the instantaneous radius [25].
Due to the increase in evaporation field experienced when the specimen reaches the oxide/metal interface, the reconstructions obtained using this method appear bell-shaped. While these are not necessarily accurate representations of the analyzed volumes, in the absence of additional information on the evaporation characteristics of the oxides it was deemed acceptable for the purpose of the current study. However, the scale bars provided in with the reconstructions should be taken with caution and scaling of the oxide and metal phases may differ. During mass spectrum assignment overlapping peaks were checked to determine which element(s) provided the greatest contribution, and labeled as such. Peak deconvolution was performed on select reconstructions to check the validity of this method, and the difference in measured composition between full deconvolution and the method used was never more than 1 at% for a given element. This was deemed acceptable since the specimen to specimen variation in measured composition was typically larger than 1 at% for each major element. Measured phase compositions included a background correction. It should be noted that the measured oxygen content was always below the expected stoichiometric value at the laser energy employed. This phenomenon has been noted previously for laser pulsing APT of insulators and semiconductors [26, 27]. Nevertheless, the measured oxygen content is reported for completeness.
3. Results
3.1 Oxide scale morphology
The oxide scale that developed on the electropolished and ground surfaces consisted of two layers: a continuous inner oxide and an outer oxide formed of discrete particles. On the ground surfaces, outer oxide particles grew in both size and number density with increasing exposure time (Figure 1). The outer particle sizes and number densities observed on the electropolished surfaces were significantly less than those observed on the ground surface through the initial stages of oxidation (Figure 2).
On the ground surfaces, inner oxide penetration varied significantly in the early stages of oxidation, as shown in the phase contrast cross sectional STEM images in Figure 1, resulting in wide standard deviations
ACCEPTED MANUSCRIPT
in the measured thickness. On the other hand, electropolished samples exhibited much more uniform inner oxide thickness (Figure 2), with no additional oxide penetration observed, even at grain boundaries (Figure 2c). After 2200 hours of oxidation, the oxide thickness on machined surface showed much less local variation in thickness than the short term exposure ground specimens (Figure 3).
The average inner oxide thicknesses measured from STEM images are shown as a function of exposure time in Figure 4. Electropolished surfaces exhibited oxide growth too thin to accurately measure for exposure times less than 1 hour. In almost all cases the electropolished surfaces had thinner inner oxide scales than ground surfaces exposed for the same time, while both exhibited approximately cubic kinetics.
The relatively low oxide thickness on the 72 hour ground specimen could be attributed to fewer observed areas of deep penetration in the cross section analyzed compared to the other ground specimens. The 56 hour electropolished specimen also exhibited lower oxide growth than expected based on the trends observed otherwise.
3.2 Oxide scale chemistry
The oxide phases and chemical development were identified using Raman spectroscopy and APT.
Representative Raman spectra from each oxidation condition are shown in Figure 5, and represents signals from both inner and outer oxide layers, since complete surface coverage by the outer layer was not observed in the 72 hour timeframe of the experiments. On the ground samples, the peaks at 226, 292, and 409 cm-1 are characteristic of hematite (Fe2O3) [28]. The characteristic peaks for other likely oxide species FeCr2O4
(683 cm-1), NiFe2O4 (486 and 697 cm-1), and Fe3O4 (665 cm-1) all share the region 660-700 cm-1 [28]. On ground specimens, this peak broadened with increasing exposure time, indicating the simultaneous presence of multiple of these phases. On electropolished specimens, the main peak shifted from left to right, indicating that the chemistry of the spinel phase shifted from initially magnetite-rich towards a nickel ferrite phase.
APT was performed for select immersion times for both the ground and electropolished surfaces with an emphasis placed on capturing the inner oxide layer, and the inner oxide/metal interface. The oxide formed on electropolished specimens was locally uniform in composition for each exposure time, but evolved chemically over the time period studied. From volumetric reconstructions of specimens obtained after 1, 4, and 72 hour exposures (Figure 6), one-dimensional depth concentration profiles showed that the inner oxide films exhibited little to no chemical fluctuation throughout the analyzed depth. The measured average composition of the oxide film for each exposure time is reported in Table 2. The oxide was deficient in Fe, and to a lesser extent Ni, compared to the original matrix composition, but the deficiency appeared to
ACCEPTED MANUSCRIPT
decrease with exposure time. The oxide/metal interface was distinctly enriched in Cu with the Cu concentration and thickness increasing with exposure time (Figure 6). Meanwhile, little to no Ni enrichment was observed at the interface.
In contrast to the electropolished specimens, the composition of the inner oxide formed on ground surfaces was modulated at the nanoscale. In Figure 7, two separate oxide regions of distinct chemistry were identified using a CrO1+ isoconcentration surface. The fluctuations were observed at the earliest oxidation times as well as the longer oxidation times. One oxide region was Cr-rich, while the other was richer in Fe compared to Cr, and had more Ni incorporated, and is called a ‘mixed Fe/Cr’ oxide. The respective compositions of both regions of the inner oxide, along with the total average composition of the entire inner oxide layer, are reported in Table 2. In addition, specimens from both 1 and 72 hours each preserved a portion of the outer oxide particle at the specimen apex (Figure 7). These particles were rich in Fe with small Ni and Cr contributions (Table 2).
The characteristics of the oxide/metal interface on ground specimens also differed from the electropolished specimens, in that Ni accumulated at the interface, while Cu segregation in this region, while present, was at significantly lower levels. As seen in Figure 7, the interface was largely lost in the 1 hr and 3 month specimens, but regions below the actual interface still exhibited Ni and Cu enrichment. The interface was captured in the 72 hour specimen and shows that Cu and Ni were both present, albeit discontinuously, at the oxide/metal interface. The Cu enrichment for this specimen only reached levels of ~ 3 at% compared to 40 at% in the electropolished 72 hour specimen, while Ni enrichment reached levels of ~50 at% in some areas.
4. Discussion
The results presented herein clearly show that subsurface deformation plays a significant role in the resulting oxide scale morphology and chemistry. The differences observed between the electropolished and ground surface provided significant clues as to the required diffusion processes that resulted in these unique oxide films.
The outer particle-like oxide layer is typically considered to be a byproduct of the surface oxidation process and does not confer any corrosion protection due to incomplete coverage by discrete particles. It was not explicitly studied in this work. Nonetheless, the composition of the outer oxide was measured from a few particles observed on some of the APT specimens from the ground surface, and was found consistent with
ACCEPTED MANUSCRIPT
magnetite with some Ni incorporation, in agreement with prior results obtained for exposures in similar environments [1, 3, 4]. Dissolution/precipitation and solid state outward growth have both been discussed as possible mechanisms for the formation of the outer oxide layer. In either case, the chemistry of the outer layer is governed by the metal cation saturation of the water, which is in part a result of the diffusion kinetics of the inner layer. It has previously been reported by Allen et al. [29] that the diffusion rates of metal cations through a spinel lattice follows the order: DFe > DNi >> DCr, which is consistent with the outer oxide being primarily Fe, with some Ni, and very little Cr. Additionally, while small hematite peaks were identified in the Raman spectra of most of the ground specimens, and one of the electropolished specimens, it was not a significant component of the outer oxide layer. Since hematite is the primary outer oxide product in elevated temperature, oxygenated water [5, 6], some small amount of oxygen may have remained in the autoclave during the exposures.
More interestingly, the inner oxide layer exhibited significant differences between ground and electropolished surfaces, which can be related to the underlying deformation of the material and its influence on the mechanisms of oxide layer growth. The inner oxide layer formed on each electropolished specimen was very uniform in both thickness and local chemistry (Figures 2 and 6). Furthermore, the Raman spectra for electropolished specimens exhibited a peak shift with increasing exposure time, but maintained a sharp peak indicative of a mixed spinel. The measured composition matched closely with that predicted for a thermodynamically stable mixed spinel [3]. Thus Fe, followed by Ni, was the most depleted metal species in the inner oxide. This is consistent with Fe being the fastest diffusing element through the inner oxide. Furthermore, no diffusion profile for any of the chemical species was observed in the matrix beneath the oxide layer, suggesting diffusion processes in the matrix were limited in the absence of deformation pathways. These observations are consistent with a steady-state inward growth mechanism primarily limited by the diffusion rate of oxygen through the oxide film. In other words, the metal cations not involved in the formation of the equilibrium oxide phase are able to diffuse out through the inner oxide at a faster rate than the oxygen ingress and reaction at the inner oxide/metal interface when no sub-surface deformation is present. The process is shown schematically in Figure 8. It should also be noted that no additional oxide penetration was observed at grain boundaries of electropolished specimens, despite the fact that increased lattice disorder would be expected in the grain boundary region. This suggests that the degree of disorder is much less than that imparted by mechanical deformation, as discussed next.
The oxide film developed on ground surfaces was more complex than that on electropolished surfaces due to the underlying deformation in the material microstructure, and growth appeared to occur in two concurrent steps. First, a Cr-rich oxide film developed primarily along regions affected by deformation (ie.
ACCEPTED MANUSCRIPT
dislocations), that allowed for faster diffusion pathways of oxygen and selective oxidation of Cr, as evidenced by oxidized Cr-rich ‘fingers’ extending into the matrix, seen in Figure 7a. This is consistent with previous APT studies that highlighted the oxidation of dislocation networks in highly cold worked materials [30]. As the Cr-rich inner oxide formed, Fe and Ni were rejected either to the adjacent unoxidized regions, or through the oxide to the solution interface. The matrix regions adjacent to the fast forming Cr- rich oxide eventually started oxidizing, resulting in an oxide region that has both Fe and Cr contributions.
This second region is more similar in composition to the oxide formed on electropolished specimens, suggesting a formation mechanism comparable to that on electropolished specimens. Additionally, the Raman spectra from ground specimens exhibited significant peak broadening with increasing oxidation time, consistent with the simultaneous presence of Fe3O4, NiFe2O4 and FeCr2O4. In other work under similar oxidizing conditions, local electrical resistivity measurements of the oxide layer had also alluded to such variations in oxide microchemistry [12]. Once formed, the dual oxide structure was preserved within the inner oxide, even after long exposure times (as demonstrated by the 2200 hrs exposure). The suggested oxidation process is shown schematically in Figure 9. The proposed mechanism accounts for these features based on direct observation of the resulting microstructure within the inner oxide layer.
The oxide/metal interface differed radically between the electropolished and the ground specimens. The electropolished specimens consistently exhibited a smooth and sharp interface that was highly enriched in Cu (Figure 6). While copper at the oxide/metal interface was observed previously in similar materials [31, 32], its relevance had not been noted. Here, significant effort was placed on ensuring the Cu was not contamination due to the electropolishing procedure, or an artifact of the measurement technique. Auger and X-ray photoelectron spectroscopy were used to verify that no Cu was present on the electropolished surface prior to oxidation. Additionally, energy dispersive x-ray spectroscopy mapping (not shown) was utilized on TEM cross sections to verify the presence of a Cu layer at the oxide/metal interfaces of electropolished specimens, and its absence at the interface of ground specimens. Finally, the water chemistry was sampled and there was no indication of abnormal anion or cation contamination of any species, including Cu which was below detectable limits of 10 ppb. Based on these observations, it was determined that the Cu originated from the material itself, and was high localized at the interface due to its high electrochemical nobility resulting in rejection from the growing oxide layer (Figure 8). The APT measured Cu composition was used to estimate the volume of Type 304 SS material that would originally contain that amount of Cu based on the bulk composition (0.15 at%), assuming full rejection and accumulation of Cu at the interface. The estimated necessary volume was consistently larger than the oxidized depth could account for, suggesting uniform dissolution of the metal (Fe, Cr, and Ni) to expose the matrix Cu was a component of the oxidation process. From this volume discrepancy, an estimated
ACCEPTED MANUSCRIPT
dissolution rate, averaged over the duration of the exposure, was calculated and is presented in Figure 10.
This proposed dissolution process further supports the formation of an inner oxide with a different Fe/Cr and Fe/Ni ratios than the matrix without any matrix concentration gradient. The decreasing rate with exposure time is rationalized in terms of cation diffusion through a thickening oxide layer.
The oxide/metal interface of ground specimens exhibited relatively less Cu enrichment, but did exhibit pockets of Ni enrichment (Figure 7). Ni enrichment has previously been observed at the oxide/metal interface of similar materials oxidized in similar environments, and was attributed to rejection due to high electrochemical nobility [30], or due to the faster diffusion and selective oxidation of Cr [2, 33], leaving Ni rich regions behind. While the precise mechanism cannot be determined based on the current results, it is clear that deformation is a necessary condition for Ni enrichment, as no such enrichment was observed on electropolished specimens. Based on the proposed mechanism of Cu enrichment for electropolished specimen, this suggests that less dissolution was occurring on ground specimens despite the increased prevalence and size of outer oxide particles, which points to a solid-state growth mechanism rather than a dissolution-precipitation mechanism for the outer oxide. Conversely, it could be said that deformation inhibits the localization or enrichment of Cu, although this explanation seems less likely given the increased prevalence of Ni accumulation.
The distributions of elements in the inner and outer oxide films were consistent with the relative diffusion kinetics of the different metal cation species through a spinel oxide lattice. The overall kinetics, measured based on inner oxide thickness, were approximately cubic for both surface roughness conditions, with the rate constant for ground specimens being greater than for electropolished specimens (Figure 3). Previous results, based on bulk measurements as opposed to localized ones as used here, had typically identified the rate law for oxide growth in similar conditions to be parabolic [3, 17, 21], as proposed by Wagner based on the solid state diffusion kinetics of metal cation species [34]. However, slower cubic kinetics have been discussed by some authors as a result of charge build up in the oxide film slowing the diffusion rates [8, 35]. More work will be needed to rationalize the exact kinetics of oxide growth taking into account alloy and oxide scale microstructures such as described here.
To conclude, the present results suggest that surface finish plays and important role in the oxidation mechanisms and resulting oxide structure that can be leveraged to reduce the risk of harmful corrosion processes. By removing surface deformation, the overall oxidation rate is decreased. The formation of a Cu layer on deformation-free surfaces potentially forms a corrosion barrier. Furthermore, the uniformity of the inner oxide layer on electropolished specimens that lacks areas of enhanced penetration may also decrease
ACCEPTED MANUSCRIPT
the risk of crack nucleation and propagation. However, to truly verify the benefits of electropolishing with regards to corrosion and SCC, testing should be performed with an applied stress.
5. Conclusions
A combination of the atom probe tomography and transmission electron microscopy techniques were used to understand the development of the inner oxide film formed on ground and electropolished surfaces during exposure to high temperature deaerated water. The following conclusions were reached:
Irrespective of surface finish, a dual layer oxide structure forms in high temperature deaerated water.
The outer oxide was rich in Fe and composed of discrete particles, while the inner oxide layer was compact and richer in Cr.
The growth rate of the inner oxide film was higher on ground specimens than electropolished specimens. Qualitatively, the outer oxide particles were also larger and more numerous on ground specimens than on electropolished specimens.
On electropolished specimens, a thermodynamically stable mixed spinel oxide formed that was uniform in composition and thickness, distinguishing it from ground specimens. The absence of Fe, Cr, or Ni enrichment at the oxide/metal interface suggested significant diffusion of metal cations through the oxide layer. Cu enrichment at the oxide/metal interface was speculated to be the result of enhanced dissolution during the initial stages of exposure, resulting in Cu availability, coupled with its high electrochemical nobility resulting in rejection to the interface.
On ground surfaces, selective oxidation of Cr occurred first along short circuit diffusion pathways, resulting in complex Cr-oxide network developing ahead of the inner oxide interface. Subsequent oxide growth led to a mixed Fe/Cr oxide filling in the adjacent unoxidized regions. The process resulted in an oxide layer whose composition is modulated at the nanoscale. The deformed microstructure promoted rejection of Ni by the fast growing oxide, or increased diffusion of Cr for selective oxidation, resulting in Ni enrichment at the oxide/metal interface. Cu was not observed to significantly enrich at the oxide/metal interface, further suggesting a qualitatively different mechanism for oxide growth rationalized here as an absence of enhanced dissolution of the deformed matrix material.
Acknowledgements
The authors would like to acknowledge the engineering and testing support provided by Jared Malys and Chuck Austin at the Naval Nuclear Lab, as well as the Michigan Center for Materials Characterization for use of the instruments and staff assistance.
ACCEPTED MANUSCRIPT
Funding
This research was performed under appointment (K.B. Fisher) to the Rickover Fellowship Program in Nuclear Engineering sponsored by the Naval Reactors Division of the U.S. Department of Energy.
References
1. Terachi, T., et al., Corrosion Behavior of Stainless Steels in Simulated PWR Primary Water—Effect of Chromium Content in Alloys and Dissolved Hydrogen—. Journal of Nuclear Science and Technology, 2008. 45(10): p. 975-984.
2. Cissé, S., et al., Effect of surface preparation on the corrosion of austenitic stainless steel 304L in high temperature steam and simulated PWR primary water. Corrosion Science, 2012. 56: p. 209- 216.
3. Ziemniak, S.E. and M. Hanson, Corrosion behavior of 304 stainless steel in high temperature, hydrogenated water. Corrosion Science, 2002. 44(10): p. 2209-2230.
4. Ziemniak, S.E., M. Hanson, and P.C. Sander, Electropolishing effects on corrosion behavior of 304 stainless steel in high temperature, hydrogenated water. Corrosion Science, 2008. 50(9): p. 2465- 2477.
5. Kuang, W., et al., Microstructural characteristics of the oxide scale formed on 304 stainless steel in oxygenated high temperature water. Corrosion Science, 2010. 52(11): p. 3654-3660.
6. Kuang, W., X. Wu, and E.-H. Han, The oxidation behaviour of 304 stainless steel in oxygenated high temperature water. Corrosion Science, 2010. 52(12): p. 4081-4087.
7. Tapping, R.L., et al., The composition and morphology of oxide films formed on type 304 stainless steel in lithiated high temperature water. Corrosion Science, 1986. 26(8): p. 563-576.
8. Lister, D.H., R.D. Davidson, and E. McAlpine, The mechanism and kinetics of corrosion product release from stainless steel in lithiated high temperature water. Corrosion Science, 1987. 27(2): p.
113-140.
9. Kim, Y.-J., Characterization of the oxide film formed on type 316 stainless steel in 288 C water in cyclic normal and hydrogen water chemistries. Corrosion, 1995. 51(11): p. 849-860.
10. Ming, H.L., et al., Short time oxidation behavior of 308L weld metal and 316L stainless steel with different surface state in simulated primary water with 0.1 mg/L dissolved oxygen. Materials and Corrosion, 2015. 66(9): p. 869-881.
11. Xu, J., X.Q. Wu, and E.H. Han, The evolution of electrochemical behaviour and oxide film properties of 304 stainless steel in high temperature aqueous environment. Electrochimica acta, 2012. 71: p. 219-226.
12. Massoud, T., et al., Nanostructure and local properties of oxide layers grown on stainless steel in simulated pressurized water reactor environment. Corrosion Science, 2014. 84(1): p. 198-203.
13. Stellwag, B., The mechanism of oxide film formation on austenitic stainless steels in high temperature water. Corrosion Science, 1998. 40(2–3): p. 337-370.
14. Lozano-Perez, S., et al., Atom-probe tomography characterization of the oxidation of stainless steel. Scripta Materialia, 2010. 62(11): p. 855-858.
15. Lozano-Perez, S., et al., The role of cold work and applied stress on surface oxidation of 304 stainless steel. Corrosion Science, 2012. 56(0): p. 78-85.
16. Robertson, J., The mechanism of high temperature aqueous corrosion of stainless steels. Corrosion Science, 1991. 32(4): p. 443-465.
17. Kuang, W., et al., Effect of nickel ion from autoclave material on oxidation behaviour of 304 stainless steel in oxygenated high temperature water. Corrosion Science, 2011. 53(3): p. 1107- 1114.
ACCEPTED MANUSCRIPT
18. Orr, J.T., et al., A Consistent Interpretation of Corrosion Test Results Using the Marcus Film Growth Model, in NACE Corrosion. 2016, NACE International: Vancouver, British Columbia, Canada.
19. Kuniya, J., I. Masaoka, and R. Sasaki, Effect of Cold Work on the Stress Corrosion Cracking of Nonsensitized AISI 304 Stainless Steel in High-Temperature Oxygenated Water. Corrosion, 1988.
44(1): p. 21-28.
20. Maekawa, T., M. Kagawa, and N. Nakajima, Corrosion Behaviors of Stainless Steel in High- Temperature Water and Superheated Steam. Transactions of the Japan Institute of Metals, 1968.
9(2): p. 130-136.
21. Morrison, J., et al. Effect of surface finish on the corrosion behaviour of grade 316L stainless steels in pressurized, high temperature, lithiated water. in 17th International Conference on Environmental Degradation of Materials in Nuclear Power Systems – Water Reactors 2015.
Ottawa, Ontario, Canada.
22. Ghosh, S., M.K. Kumar, and V. Kain, High temperature oxidation behavior of AISI 304L stainless steel—Effect of surface working operations. Applied Surface Science, 2013. 264: p. 312-319.
23. Thompson, K., et al., In situ site-specific specimen preparation for atom probe tomography.
Ultramicroscopy, 2007. 107(2–3): p. 131-139.
24. Rasband, W.S., ImageJ. 1997-2016, U.S. National Institutes of Health: Bethesda, Maryland, USA.
25. Gault, B., et al., Chapter 7: Tomographic Reconstruction, in Atom Probe Microscopy. 2012, Springer: New York. p. 157-209.
26. Kirchhofer, R., M.C. Teague, and B.P. Gorman, Thermal effects on mass and spatial resolution during laser pulse atom probe tomography of cerium oxide. Journal of Nuclear Materials, 2013.
436(1–3): p. 23-28.
27. Devaraj, A., et al., Role of Photoexcitation and Field Ionization in the Measurement of Accurate Oxide Stoichiometry by Laser-Assisted Atom Probe Tomography. The Journal of Physical Chemistry Letters, 2013. 4(6): p. 993-998.
28. Lafuente, B., et al., The power of databases: the RRUFF project. Highlights in Mineralogical Crystallography, 2015: p. 1-30.
29. Allen, G.C., et al., A surface study of the oxidation of type 304L stainless steel at 600 K in air.
Oxidation of Metals, 1988. 29(5-6): p. 391-408.
30. Kruska, K., et al. Atom-probe tomography of surface oxides and oxidized grain boundaries in alloys from nuclear reactors. in MRS Online Proceedings Library. 2013.
31. Lozano-Perez, S., et al., Atom-probe Tomography of Surface Oxides in a 20% Cold Worked Stainless Steel Tested Under PWR Primary Water Conditions. Microscopy and Microanalysis, 2009. 15(Supplement 2): p. 304-305.
32. Huang, Y.Z. and J.M. Titchmarsh, TEM investigation of intergranular stress corrosion cracking for 316 stainless steel in PWR environment. Acta Materialia, 2006. 54(3): p. 635-641.
33. Terachi, T., K. Fujii, and K. Arioka, Microstructural characterization of SCC crack tip and oxide film for SUS 316 stainless steel in simulated PWR primary water at 320°C. Journal of Nuclear Science and Technology, 2005. 42(2): p. 225-232.
34. Wagner, C., Diffusion and high temperature oxidation of metals. Atom movements, 1951: p. 153- 173.
35. Douglas, L. and F.C. Zyzes, The Corrosion of Iron in High-Temperature Water (Part 1—Corrosion Rate Measurements). Corrosion, 1957. 13(6): p. 19-32.
Figures
ACCEPTED MANUSCRIPT
Figure 1: SEM images of the surface of ground specimens after oxidation with corresponding STEM images of the cross section showing oxide penetration into the matrix. (a) 1 hour. (b) 4 hours. (c) 72 hours. Dashed lines in the cross sectional images indicate the Pt/outer oxide, outer oxide/inner oxide, and inner oxide/matrix interfaces.
Figure 2: SEM images of the surface of electropolished specimens after oxidation with corresponding STEM images of the cross section showing oxide penetration into the matrix. (a) 1 hour. (b) 4 hours. (c) 72 hours. Dashed lines in the cross sectional images indicate the Pt/outer oxide, outer oxide/inner oxide, and inner oxide/matrix interfaces. (c) also shows the location of a grain boundary, where no additional oxide penetration was observed.
Figure 3: (a) SEM image of the surface and (b) STEM image of the cross section of a machined sample after oxidation for 2200 hours showing much thicker oxide films and a fairly uniform interface despite the deformed microstructure.
Figure 4: Average inner oxide thickness measured from STEM cross sections from each surface condition. Each data point was obtained from >130 measurements made every 50 nm across a TEM lamella. Error bars represent one standard deviation from the average. Cubic (dots) and parabolic (dashes) power law curves are shown bounding the data.
Figure 5: Raman spectra obtained from the oxidized surface of each sample. Dashed lines are
electropolished samples and solid lines are ground samples. Vertical lines represent the expected Raman shift for observed oxide species. The oxide on the electropolished specimen shifts from Fe3O4 to NiFe2O4
with increasing exposure time, while on the ground specimen the main peak broadens indicating a mixed oxide phase.
Figure 6: APT reconstructions of specimens obtained from electropolished surfaces after select exposure times along with 1-dimensional concentration profiles down the tip axis. The sequence of reconstructions shows the development of the inner oxide layer with exposure time. For each reconstruction a 10 nm slice from the center is shown with 10% of Fe atoms (pink), 100% of Cu (orange), and 100% of CrO (blue) ionic species, along with an 2% (a) or 8% (b and c) Cu isoconcentration surface (orange). 1-dimensional concentration profiles of major species were obtained using a 10nm diameter cylinder placed in the center of the reconstruction parallel to the specimen axis with a bin width of 0.3nm. (a) 1 hr exposure. (b) 4 hr
exposure. (c) 72 hour exposure.
ACCEPTED MANUSCRIPT
Figure 7: APT reconstructions of specimens obtained from ground surfaces after select exposure times along with proxigram plots obtained from the CrO1+ iso-concentration surface from within each bracketed region (1-4). Iso-concentration surface values were chosen to best represent the respective region while minimizing noise in the surface and are shown above the respective graph. The sequence of
reconstructions and proxigrams shows the development of the inner oxide layer with exposure time. For each reconstruction, a 4 nm slice from the center is shown with 30% of Fe atoms (pink), 100% of Cu (orange), 50% of Ni (green), 100% of FeO (yellow), and 100% of CrO (blue) ionic species, along with a 2% Cu iso-concentration surface (orange), a 22% Ni iso-concentration surface (green), and a CrO iso- concentration surface (blue). Each specimen exhibited a small microfracture, which is left as a gap of unknown width in the reconstruction.
Figure 8: Schematic showing the growth of the inner oxide film as a mixed Fe/Cr oxide on
electropolished specimens. (a) illustrates that diffusion and dissolution are happening simultaneously with the relative rates for each species shown by the length of the respective arrow. (b) shows that once Cu is rejected to the oxide/metal interface and accumulates, it ultimately slows the oxidation kinetics.
Figure 9: Schematic showing the growth of the inner oxide film on ground specimens. (a) illustrates the selective oxidation of Cr along dislocation pathways (dashed lines) resulting in the formation of a network of Cr-rich oxide film (medium grey). (b) shows the subsequent reaction of oxygen at the Cr- depleted matrix/Cr-rich oxide interface, resulting in the formation of a second, Fe-rich oxide (light grey) network interspersed within the inner oxide layer. Rejection of Ni along the dislocation pathways results in Ni enriched regions at the inner oxide/matrix interface.
Figure 10: Average dissolution rate (left axis, open squares) and dissolution depth (right axis, closed squares) necessary to account for the observed level of Cu segregated at the oxide/metal interface assuming Cu came entirely from the matrix, and was fully rejected to the oxide/metal interface.
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Table 1: Composition (in at%) of the 304L SS from the vendor certification
Element Fe Cr Ni Mn Cu Other
(C, N, P, S, Si, Co, B, Mo) Concentration 68.6 19.4 9.0 1.7 0.15 1.1
Table 2: APT-measured compositions (in at%) of the oxides formed on the electropolished and ground specimens after select exposure times. Error bars represent the standard deviation observed across different APT specimens from the same exposure conditions.
Electropolished Ground
Inner Oxide Inner Oxide 1 (Mixed Fe/Cr)
Inner Oxide 2 (Cr- Rich)
Inner Oxide
(Combined) Outer Oxide
1 hr
O 47.4 ± 3.4 41.4 ± 9.4 52.8 ± 1.1 47.7 ± 3.1 42.9 Cr 33.2 ± 3.3 15.9 ± 4.2 34.5 ± 3.1 23.8 ± 0.3 3.3 Fe 11.6 ± 5.0 34.8 ± 10.2 9.1 ± 2.1 21.6 ± 0.6 49.3
Ni 5.4 ± 1.9 5.1 ± 1.6 1.4 ± 0.1 3.5 ± 0.4 3.6
4 hr
O 53.4 ± 1.9 - - - -
Cr 26.3 ± 1.2 - - - -
Fe 13.6 ± 1.2 - - - -
Ni 4.9 ± 0.2 - - - -
72 hr
O 53.1 ± 0.2 51.9 ± 3.7 54.9 ± 0.6 54.1 ± 2.0 44.1 ± 1.1 Cr 22.1 ± 0.7 19.1 ± 2.6 35.1 ± 1.0 29.0 ± 3.9 3.2 ± 0.2 Fe 15.9 ± 0.5 20.8 ± 6.4 6.2 ± 2.0 10.3 ± 3.4 47.6 ± 3.5 Ni 8.4 ± 0.5 7.0 ± 3.8 1.8 ± 0.3 4.6 ± 2.6 4.9 ± 2.2