Volume 7 Preprint 15

Keywords:

ISSN 1466-8858 Volume 7 Preprint 15 Research Opportunities in Corrosion Science for Long-Term Prediction of Materials Performance A Report of the DOE Workshop on "Corrosion Issues of Relevance to the Yucca Mountain Waste Repository" sponsored by United States Department of Energy Division of Basic Energy Sciences and Division of Office of Civilian Radioactive Waste Management Report prepared by: Workshop Co-Chairpersons Joe H. Payer, Case Western Reserve University John R. Scully, University of Virginia Contributors and participants: John Beavers, CC Technologies; Jeff Braithwaite, Sandia National Laboratories; Rudy Buchheit, Ohio State University; Clive Clayton, State University of New York; Thomas M. Devine, Jr., University of California; Steve Dexter, University of Delaware; Joe Farmer, Lawrence Livermore National Lab.; Jerry Frankel, Ohio State University; Greg Gdowski, Lawrence Livermore National Lab.; Andrew Gewirth, University of Illinois; Woods Halley, University of Minnesota; Hugh Isaacs, Brookhaven National Laboratory; Russ Jones, Pacific Northwest National Lab.; Larry Kaufman, Massachusetts Inst. of Tech.; Rob Kelly, University of Virginia, Digby Macdonald, Pennsylvania State University; Nancy Missert, Sandia National Laboratories; Roger Newman, UMIST; Chris Orme, Lawrence Livermore National Lab.; Joe H. Payer, Case Western Reserve; Peter Scott, Framatome ANP; John R. Scully, University of Virginia; David Shoesmith, University of Western Ontario; David J. Wesolowski, Oak Ridge National Laboratory; Frank Wong, Lawrence Livermore National Lab. Workshop held on July 29-30, 2003 in Bethesda, MD. December, 2004 This is a preprint of a paper that has been submitted for publication in the Journal of Corrosion Science and Engineering. It will be reviewed and, subject to the reviewers' comments, be published online at http://www.jcse.org in due course. Until such time as it has been fully published it should not normally be referenced in published work. © UMIST 2004. TABLE OF CONTENTS COVER SHEET........................................................................................................................1 TABLE OF CONTENTS ..........................................................................................................2 LEGAL NOTICE AND DISCLAIMER...................................................................................3 EXECUTIVE SUMMARY .......................................................................................................4 INTRODUCTION.....................................................................................................................7 Benefit and Impact of Advances in Corrosion Science and Technology...................................7 CORROSION PERSPECTIVES AT YUCCA MOUNTAIN................................................10 Introduction...........................................................................................................................10 Natural System at Yucca Mountain .......................................................................................10 Repository Conditions Relevant to Waste Package Performance ...........................................11 Corrosion Performance of Highly Corrosion Resistant Materials ...........................................12 Time Evolution of Environment ............................................................................................14 Other Considerations.............................................................................................................15 SELECTED RESEARCH OPPORTUNITIES......................................................................28 Life Prediction and Evolution of Corrosion Damage..............................................................28 Evolution of Corrosive Environments....................................................................................29 Localized Corrosion ..............................................................................................................30 Passivity................................................................................................................................31 Stress Corrosion Cracking .....................................................................................................32 Materials Stability and Aging ................................................................................................33 Fabrication and Advanced Materials......................................................................................33 Methods and Tools ................................................................................................................34 TOPIC AREA GOALS AND OPPORTUNITIES .................................................................35 Predictive Modeling of Life and Materials Performance ........................................................36 Evolution of Corrosive Environments....................................................................................45 Localized Corrosion ..............................................................................................................52 Passivity................................................................................................................................64 Stress Corrosion Cracking .....................................................................................................75 Effects on Aging and Phase Stability.....................................................................................81 Methods and Tools ................................................................................................................85 ACKNOWLEDGEMENTS ....................................................................................................93 APPENDIX..............................................................................................................................93 Appendix A-Corrosion Workshop Participants ......................................................................94 2 LEGAL NOTICE AND DISCLAIMER This DOE Corrosion Workshop was a scientific workshop that is not intended to provide or serve as a work plan or directive to the U.S. Government or any of its agencies. This report represents the opinions and views of the workshop participants that met on 29 and 30 July, 2003 in Bethesda, MD, USA. The views and opinions expressed in the scientific report and the report to be published in an archival scientific journal will not necessarily state or reflect views, opinions or policy of the U.S. Government or any of its agencies. The fact that any particular issue or opportunity is not mentioned in this report does not necessarily imply that it is not important. As such, the report does not intend to be all-inclusive. The Yucca Mountain Project believes that it has a sufficiently strong safety case to proceed with a License Application to the USNRC. However, advances in a number of areas may be feasible which, if realized, could help the Project either through better understanding of the long-term performance of the waste packages, or through better performance, or perhaps through cost savings that do not compromise performance. 3 EXECUTIVE SUMMARY The report summarizes the findings of a U.S. Department of Energy workshop on "Corrosion Issues of Relevance to the Yucca Mountain Waste Repository". The workshop held on July 29-30, 2003 in Bethesda, MD and was co-sponsored by the Office of Basic Energy Sciences and Office of Civilian Radioactive Waste Management. The workshop focus was corrosion science relevant to long-term prediction of materials performance in hostile environments and with special focus on relevance to the permanent disposal of nuclear waste at the Yucca Mountain Repository. The culmination of the workshop is this report that identifies both generic and Yucca-Mountain-Projectspecific research opportunities in basic and applied topic areas. The research opportunities would be realized well after the U.S. Nuclear Regulatory Commission's initial construction-authorization licensing process. At the workshop, twenty-three invited scientists deliberated on basic and applied science opportunities in corrosion science relevant to long-term prediction of damage accumulation by corrosive processes that affect materials performance. Benefit and Impact of Advances in Corrosion Science and Technology In 1999 the U.S. Congress mandated the determination of corrosion costs, and a breakthrough 2-year study, "Corrosion Costs and Preventive Strategies in the United States", was issued in 2002. The study determined that the total annual estimated direct cost of corrosion in the U.S. is a staggering $276 billion--approximately 3.1 percent of the nation's Gross Domestic Product (GDP). The total costs (direct and indirect effects) to society can be twice that or greater. The nation's infrastructure is essential to the quality of life, industrial productivity, international competitiveness, and security. Corrosion is a primary cause of the degradation and a principal threat to the nation's infrastructure. The infrastructure replacement cost is a major driver on the economic impact of corrosion and can still be greater if corrosion preventive strategies are not properly employed. The opportunities for savings by improved corrosion control are significant in every industrial and government sector. Opportunities in Corrosion Science for Long-Term Prediction of Materials Performance Scientific advances for life prediction and performance assessment can result in an increase in reliability and safety, reduction in costs, and conservation of materials and energy. The prediction of long-term performance for corrosion is the most important and challenging matter for corrosion science and engineering. The determination of current status and future performance as a function of materials, design and service conditions are important to guide rationale, reliable and cost effective decisions by policy-makers, managers and engineers. The long-term corrosion performance of waste packages at Yucca Mountain Repository provides a crucial application and a special challenge due to the exceptionally long times (10,000 years and beyond) required. Several categories of research opportunities were identified in this workshop: Opportunities to advance corrosion science as applied to highly corrosion resistant alloys, e.g. Ni-Cr-Mo-W alloys and titanium alloys. This is accomplished through the development of new theories tailored to these alloys and through the extension of corrosion theories developed for Fe-based materials. Opportunities to extend corrosion science for thin film of electrolytes formed by deliquescence and condensation processes. For waste package performance, these conditions pertain for episodic dripping and electrolytes formed by combination of ionic species, water and dusty-particulate layers. This entails extensions and novel advances to the treatments of corrosion behavior in bulk electrolytes and under fully immersed conditions. 4 Opportunities for models for localized corrosion that include propagation, arrest and stifling of corrosion phenomena. Corrosion science for corrosion initiation processes is rich and substantial; however, the treatments of propagation, stifling and arrest phenomena are quite limited. On the other hand, corrosion processes that initiate can become kinetically limited under many realistic conditions, and this retardation should be factored into models and analysis. There is major impact on determination of damage evolution, life prediction, and performance acceptance when full immunity to corrosion is difficult to confirm or is uncertain. Opportunities to advance predictive capability via enhanced computational capability and advanced modeling and prediction protocols. There was a consensus of the potential for large gains and opportunities in this area. Recognition was also given opportunities for advances to verification of long term predictions of material performance via accelerated testing. A path to progress was identified through the use of more susceptibility analogs and extensions to extremely corrosion resistant alloys. Opportunities for advances in computational and experimental methods. The former is represented by application of advanced thermodynamics and computational materials science to metallurgical aging issues and the corresponding corrosion behavior. For the latter, there is a plethora of new techniques with unprecedented ability to interrogate materials with high resolution, improved data acquisition, and higher analytical efficiency and throughput. Opportunities to adapt and apply state of the art developments in other fields to corrosion science. A specific opportunity was the advancement and application of combinatorial methods after those in the pharmaceutical industry. Another was for advanced sensors and diagnostic methods after those in the medical community. The opportunities identified at the Corrosion Workshop provide a special opportunity for advances in the state of corrosion science to benefit not only analysis of nuclear waste disposal systems but a wide-range of applications. Advances in the analysis of performance and life prediction will contribute to advanced solutions for our nation's aging infrastructure, advanced equipment and systems for energy production, transportation and a range of other industrial sectors. Organization and Outputs of the DOE Corrosion Workshop This Corrosion Workshop developed from the leadership of the Office of Basic Energy Sciences (BES) and Office of Civilian Radioactive Waste Management (OCRWM). There was recognition of major opportunities for synergy and collaboration between scientists and engineers working on (a) advances in corrosion science and (b) the analysis of a crucial application of nuclear waste disposal. The workshop technical topic areas were divided into fundamental scientific topics in corrosion regarding forms of corrosion relevant to nuclear waste containers. It also included cross-cutting issues relevant to each form of corrosion. The former were Localized Corrosion Passivity Stress Corrosion Cracking The later were Life Prediction and Evolution of Corrosion Damage Evolution of Environments on Metal Surface Material Stability and Aging Advanced Materials and Fabrication Methods and Tools 5 For each topic area, workshop participants discussed its relevance to corrosion science and to the Yucca Mountain Repository. Current status of understanding and relevant background information was presented and discussed in work groups and plenary sessions. The outputs were research opportunities for critical, unresolved scientific and/or technological issues. Corrosion perspectives at Yucca Mountain were presented as background on important issues in corrosion science in the context of long-term prediction of materials performance. The long-term strategy for safe disposal of nuclear waste at the Yucca Mountain Repository is to completely isolate the radionuclides for long times (1000's of years and longer). Long-lived waste packages are essential to safe performance. The prediction of materials performance is important and common to a broad range of industries and systems; however, the extraordinarily long time period of interest for Yucca Mountain presents a special challenge. Corrosion is clearly the primary determinant of waste package performance and will control the delay time for radionuclide transport from the waste package. Crucial scientific issues remain for a better understanding of corrosion processes and the time evolution of corrosion damage. The application of corrosion science to performance assessment and life prediction is an evolving discipline. The opportunities for advancements in corrosion science for life prediction and performance assessment are important and have great potential benefit. Yucca Mountain provides an application where the public benefits of increased confidence in performance are clear. Progress and advancements in life prediction at Yucca Mountain would also have major impact and extend to condition assessment and life extension for plants and systems for energy production and transportation as well as other industries. For each of the workshop topic areas, an overarching goal is presented along with selected research opportunities identified from the DOE Corrosion Workshop. Comments and findings from the workshop are presented in separate sections for each topic area. A list of participants and information on the organization and logistics of the workshop are presented in appendices. 6 INTRODUCTION This report summarizes the findings of a U.S. Department of Energy workshop on "Corrosion Issues of Relevance to the Yucca Mountain Waste Repository". The workshop was Co-Sponsored by the Office of Basic Energy Sciences and Office of Civilian Radioactive Waste Management. The workshop was held on July 28-30, 2003 in Bethesda, MD. This workshop identified opportunities in corrosion science including the improved prediction of long-term corrosion behavior. It had particular relevance to the post-USNRC license approval state for the permanent disposal of nuclear waste at the Yucca Mountain Repository and generic research opportunities for both basic and applied corrosion research. The culmination of the workshop is this report that identifies research opportunities in basic and applied topic areas. The research opportunities would be realized well after the US Nuclear Regulatory Commission's initial construction-authorization licensing process. This research, if conducted, may (a) improve the technical basis of Yucca Mountain Repository safety case, and/or (b) improve the cost or schedule aspects of the design for the underground waste packages that are an important part of the overall repository engineering design. Twenty-three scientists accepted invitations to participate in the workshop. The Yucca Mountain Project believes that it has a sufficiently strong safety case to proceed with a License Application to the USNRC. However, advances in a number of areas may be feasible which, if realized, could help the Project either through better understanding of the long-term behavior of the waste packages, through better performance, or perhaps through cost savings that do not compromise performance. Benefit and Impact of Advances in Corrosion Science and Technology Cost of Corrosion to the United States Three landmark studies have documented the remarkable costs of corrosion to the United States. In 1949, a study was led by H.H. Uhlig [H.H. Uhlig, "The Cost of Corrosion to the United States," Chemical Engineering News, Vol. 27, p 2764, 1949; or Corrosion, Vol. 6, p 29, 1950.] The annual cost of corrosion was estimated by summing the materials and protection method costs related to corrosion control. Corrosion costs were estimated to be $5.5 billion or 2.1 percent of the 1949 Gross National Product (GNP). In 1978, in response to a Congressional Directive, the National Bureau of Standards [now the National Institute of Standards and Technology] and Battelle determined the cost of metallic corrosion in the United States. The Battelle-NBS study was the first to combine the expertise of corrosion and economics experts to determine the economic impact of corrosion on the U.S. economy [J.H. Payer, W.K. Boyd, D.G. Lippold, and W.H. Fisher, "NBS-Battelle Cost of Corrosion Study ($70 Billion!)," Part 1-7, Materials Performance, May-November 1980]. The results of the Battelle-NBS study for the base year of 1975 were: The total U.S. cost of metallic corrosion per year was estimated to be $70 billion, which comprised 4.2 percent of the GNP in 1975, and 15 percent or $10 billion was estimated to be avoidable by the use of the most economically effective, presently available corrosion technology. A number of studies for other nations have been carried out, and similar large corrosion costs were determined. In 1999, the U.S. Congress mandated the determination of corrosion costs, and a breakthrough 2-year study on the direct costs associated with metallic corrosion was issued in 2002. The study entitled `Corrosion Costs and Preventive Strategies in the United States [www.nace.org/nace/content/publicaffairs/cocorrindex.asp]. Results of the study show that the total annual estimated direct cost of corrosion in the U.S. is a staggering $276 billion-- approximately 3.1 percent of the nation's Gross Domestic Product (GDP). It reveals that, although corrosion management has improved over the past several decades, the U.S. must find more and better ways to encourage, 7 support, and implement optimal corrosion control practices. An important feature of this latter report is that preventive strategies were developed for savings and reduction of corrosion costs. Protection and Life Extension for the Nation's Infrastructure The nation's infrastructure is essential to the quality of life, industrial productivity, international competitiveness, and security. Our society depends upon a functional, reliable, and safe infrastructure system, including food, water and energy needs, transportation for work, education and recreation, the production and delivery of goods and services, communications, and the treatment and long term disposal of wastes. Each component of the nation's infrastructure, such as highways, airports, water supply, waste treatment, energy supply, and power generation, represents a complex system and large investments. Corrosion is damage that results from the interaction of structures and materials with its environment. In some cases, corrosion damage is tolerable and perhaps only leads to somewhat higher maintenance costs and minimal losses; however, corrosion can result in catastrophic failures with loss of life and disruption of essential services. In fact, corrosion is a primary cause of the degradation and a principal threat to the nation's infrastructure. As documented above, the direct costs of corrosion represent 3.2 percent of the U.S. GDP. The total costs (direct and indirect effects) to society can be twice that or greater. The infrastructure replacement cost is a major driver on the economic impact of corrosion and can still be greater if corrosion preventive strategies are not properly employed. The opportunities for savings by improved corrosion control are significant in every industrial and government sector. Opportunities in Corrosion Science for Long-Term Prediction of Materials Performance Scientific advances for life prediction and performance assessment can result in an increase in reliability and safety, reduction in costs, and conservation of materials and energy. The prediction of long-term performance for corrosion is the most important and challenging matter for corrosion science and engineering. The determination of current status and future performance as a function of materials, design and service conditions are vital to guide rationale, reliable and cost effective decisions by policy-makers, managers and engineers. Opportunities for scientific advances in corrosion are clearly linked to progress in materials science and electrochemistry. However, corrosion is truly multi-disciplinary and advances can stem from progress in the sciences of physical chemistry, electrolytes, surface science, geochemical science, microbiology and others. Because of complex and coupled processes during the long-term degradation of materials, the formation of multi-investigator teams to collaborate on important aspects of these complex systems can create synergies that overcome barriers that stifle scientific advances and enhance the opportunity for major advances. Nuclear energy provides several crucial and important opportunities for major impact of advanced corrosion science and technology. Useful perspectives are available from the "Basic Research Needs to Assure a Secure Energy Future", a report from the Basic Energy Sciences Advisory Committee (BESAC), February 2003. "It is clear from the analysis presented in this (BESAC) report that there are a number of opportunities. Essentially all of these are interdisciplinary in character." "The magnitude of the energy challenge should not be underestimated. With major scientific discoveries and development of the underlying knowledge base, we must enable vast technological changes in the largest industry in the world (energy), and we must do it quickly." "Nuclear energy provides somewhat more than 20 percent of electricity in the U.S., 35 percent in the European Union, 24 percent in all OCED countries, and about 16 percent worldwide. The importance of this power source may increase due to concerns of greenhouse gas emissions from fossil-fired plants." "Public acceptance of nuclear power as an important energy source depends upon developing acceptable solutions to the back end of the nuclear fuel cycle. Scientifically based and carefully engineered solutions are more likely to gain public acceptance." The importance of understanding corrosion processes as they affect nuclear waste packages at the Yucca Mountain Repository is noted in the BESCA report. A host of other groups have also identified the analysis of long- 8 term performance of waste packages as crucial: NRC, ACNW, NWTRB, DOE Peer Panels and International Peer Panels. The long-term corrosion performance of waste packages at Yucca Mountain Repository provides a crucial application and a special challenge due to the exceptionally long times (10,000 years and beyond) required. Therefore, analysis for Yucca Mountain provides a special opportunity for advances in the state of science to benefit not only nuclear waste disposal, but advance performance and life prediction for a wide-range of applications. Advances will benefit solutions for our nation's aging infrastructure, advanced equipment and systems for energy production, transportation and a range of other industrial sectors. Origin and Basis for DOE Corrosion Workshop This Corrosion Workshop developed from the leadership of the Office of Basic Energy Sciences (BES) and Office of Civilian Radioactive Waste Management (OCRWM). There was recognition of major opportunities for synergy and collaboration between scientists and engineers working on (a) advances in corrosion science and (b) the analysis of a crucial application of nuclear waste disposal. In order to identify basic and applied research opportunities in corrosion that are relevant to the Yucca Mountain repository, a group of active researchers was gathered from BES, OCRWM and other institutions. The workshop technical topic areas were divided into fundamental scientific topics in corrosion regarding forms of corrosion relevant to nuclear waste containers. It also included cross-cutting issues relevant to each form of corrosion. The former were Localized Corrosion Passivity Stress Corrosion Cracking The later were Life Prediction and Evolution of Corrosion Damage Evolution of Environments on Metal Surface Material Stability and Aging Advanced Materials and Fabrication Methods and Tools For each topic area, workshop participants discussed its relevance to corrosion science and to the Yucca Mountain Repository. Current status of understanding and relevant background information was presented and discussed in work groups and plenary sessions. The outputs were research opportunities for critical, unresolved scientific and/or technological issues. 9 CORROSION PERSPECTIVES AT YUCCA MOUNTAIN Joe H. Payer Case Western Reserve University Introduction Opportunities for advances in corrosion science for life prediction and performance assessment provide the greatest potential benefit The long-term strategy for safe disposal of nuclear waste at the Yucca Mountain Repository is first to completely isolate the radionuclides in the waste packages for long times (1000's of years and longer) and secondly, to greatly retard the egress and transport of radionuclides from penetrated packages. Therefore, long-lived waste packages are essential to safe performance. Further, a high-quality analysis to predict waste package performance and the strength of the technical basis to support that analysis will contribute greatly to a high level of confidence for the repository. The Yucca Mountain Repository presents a common materials performance application that is encountered in energy, transportation and other industries. The widely accepted approach to dealing with materials performance is to identify the performance requirements, to determine the operating conditions that materials will be exposed to and to select materials of construction that perform well in those conditions. The special feature of Yucca Mountain is the extremely long time frame of interest. There is interest in performance behavior for periods of 10,000 years and longer. Thus, the time evolution of the environment in contact with waste package surfaces and the time evolution of corrosion damage that may result are of primary interest in the determination of expected performance. Corrosion is clearly the primary determinant of waste package performance and will control the delay time for radionuclide transport from the waste package. Corrosion is the most probable and most likely degradation process that will determine when packages will be penetrated and the shape size and distribution of those penetrations. The general issues in corrosion science, materials science and electrochemistry are well defined, and the knowledge base is substantial for understanding corrosion processes. However, crucial scientific issues remain for a better understanding of corrosion processes and the time evolution of corrosion damage. The application of corrosion science to performance assessment and life prediction is an evolving discipline. Advancements in corrosion science for life prediction and performance assessment provide the greatest opportunities and have the greatest potential benefit. Yucca Mountain provides an application where the public benefits of increased confidence in performance are clear. Progress and advancements in life prediction at Yucca Mountain would also have major impact and extend to condition assessment and life extension for plants and systems for energy production and transportation as well as other industries. Natural System at Yucca Mountain It is important to recognize that conditions at the repository evolve with time, and conditions are quite different after periods of 500 years, 5,000 years and 50,000 years. A consequence of this is that various processes rise and fall in importance with time. The combination of natural conditions at the mountain and the engineered features of the repository determine the corrosion issues at Yucca Mountain. The natural conditions include the amount of water moving through the mountain to the water table below the repository and the mineralogical, chemical and mechanical properties of the rock. The engineered features include the design, materials of construction and operation of the repository. For example, the time-temperature response for the waste packages and surrounding rock will depend upon the thermal load from spent fuel within packages and the spacing between packages and between adjacent storage tunnels (drifts). 10 It is important to recognize that conditions at the repository evolve with time, and conditions are quite different after say 500 years, 5,000 years and 50,000 years. A consequence of this is that various processes rise and fall in importance with time. Radiation fluxes at the waste package surface are highest initially and decrease to small levels over several hundred years. Heat flux from the waste packages decreases even more slowly over time, and results in a temperature increase over 10-20 years when the repository is closed, and this heat-up period is followed by a long, slow cooling back to ambient temperature over several 1000's of years. The natural system at Yucca Mountain is multiple layers of geologic formations as shown in Figure 1. The repository is in an unsaturated zone about 300 meters below the ground surface and about 300 meters above the water table. In the unsaturated zone, the pores in the rock matrix and fractures in the rock are only partially filled with water. Water moves downward through the unsaturated zone to the water table (saturated rock). An important aspect is that the repository is at near atmospheric pressure at all times, and there are no processes by which pressure can build-up to higher levels. The water flow through the mountain is a critical issue. The climatology and the amount of infiltration will determine how much water enters the unsaturated zone above the repository. At the repository level, water can move through the rock between disposal drifts and/or interact with the drifts and waste packages. Water then continues its movement through unsaturated rock beneath the repository and into the saturated zone at the water table. Thermal-hydrological-chemical effects determine the environment, i.e. the chemical composition and properties of moisture and gases in contact with metal surfaces. Both the environment and the materials of construction determine the corrosion behavior. The dimensional scale of interest will vary over many orders of magnitude depending upon the process being considered. Mountain scale processes are macro-scale (10's to 100's meters). Drift and waste package processes are meso-scale (cm's to meters). Particles and droplets are micro-scale (µm-mm). Passive films, moisture layers and sorption processes can be described at the molecular level or nanoscale (nm-µm). Repository Conditions Relevant to Waste Package Performance The overall repository time frame can be divided into relevant time periods for analysis. The critical issues for long term performance are determination of the presence of moisture on metal surfaces, the realistic corrosive properties of the moisture and the corrosion resistance of waste package materials exposed to these realistic environments. In the repository, the waste packages are placed on support pallets and sit in air at atmospheric pressure. The relative humidity, depending upon the time/temperature, can range from very low values up to full saturation at 100%. There is no feasible scenario that will lead to waste packages being fully immersed in water; rather metal surfaces will be exposed to thin layers of moisture and moist particulate or deposits. There is a limited amount of water moving through the rock, and there is a limited amount of salts and minerals available to deposit on the packages. The ambient waters in the mountain are dilute; however, those ambient waters can be modified and become concentrated by the thermal hydrological chemical processes. Waste packages are exposed to one long, slow cycle. There are no start-up/shut-down cycles. There are no moving parts, and the static exposure does not subject waste packages to cyclic mechanical loads. When the waste packages are emplaced in the repository, and the repository is closed after some 50 years, there is a heat-up and then a very long, slow cool-down. Waste package surfaces heat to a maximum temperature in the range of 120 to 180 C depending upon the particular package. The heat-up period covers 10-20 years after repository closure. This is followed by cooling to ambient temperatures over several hundreds to thousands of years. Compared to industrial equipment, e.g. heat exchangers and steam generators, the heat fluxes and thermal gradients for waste packages are modest. 11 One of the major challenges for analysis of the Yucca Mountain Repository is the extremely long time frame of interest. The prediction of behavior of materials over thousands of years and 10,000's years is unprecedented for engineering applications. While information can be gathered from the behavior of natural analogues and from metal archaeological artifacts that have remained stable for thousands of years, there are opportunities to extend current practices for materials performance and life prediction. The overall repository time frame can be divided into relevant time periods for analysis. Five relevant time periods for a medium temperature waste package are shown in Figure 2. The first period, "I-Preclosure", spans the time of waste package emplacement for 50 years. During this period, the repository is ventilated and temperatures remain low. At 50 years, the repository is closed, and the temperature starts to rise. The second period, "II-Heat-up", extends over 10 to 15 years, i.e. from years 50 to 65. During this period, a thermal barrier results from the drift wall heating to above the boiling point of water, and no drips or seepage onto waste packages will occur. The third period, "III-Thermal Barrier", is from years 65 to 750. At 750 years, the drift wall has cooled to the boiling point, and drips or seepage is possible. The fourth period, "IV-Cool down post-thermal barrier", starts when the drift wall cools to boiling and extends until the waste package surface cools to a temperature below a critical corrosion temperature for corrosion in realistic repository environments. For the case shown here, the critical corrosion temperature was 90 C, and Period IV extends from years 750 to 1375. "Period V-Packages below the critical corrosion temperature" is all time from year 1375 and beyond. This analytical framework defines critical issues for analysis of long-term materials performance: When will moisture be present on waste package surfaces? What are the realistic corrosive characteristics of the moisture? What is the corrosion resistance of waste package materials for these conditions? What is a realistic critical temperature below which no significant corrosion will occur? The critical issues for long term performance are determination of the presence of moisture on metal surfaces, the realistic corrosive properties of the moisture and the corrosion resistance of waste package materials in these realistic environments. In the repository, moisture can form on the metal surfaces by deliquescence and condensation processes as the waste packages cool from high temperatures after the heat-up period. Another source of moisture on the metal surfaces is from drips or seepage from the rock formation onto waste packages. Deliquescence/condensation is of concern during Period III and IV, and drips/seepage is of concern in Period IV. After the waste packages have cooled to the critical corrosion temperature (Period V) corrosion is not likely. Corrosion Performance of Highly Corrosion Resistant Materials Understanding the durability of passive films under realistic conditions for the repository is crucial. Localized corrosion processes (pitting, crevice corrosion, stress corrosion cracking) are the most likely degradation modes to be considered in the repository environments. Determination of damage evolution by localized corrosion over a long service period would be enhanced by improved understanding of localized corrosion processes to include initiation, propagation, stifling and arrest processes Highly corrosion resistant materials are selected for the waste packages and drip shields for the repository: Alloy 22, a nickel-chromium-molybdenum alloy and titanium, respectively. Both Alloy 22 and titanium have high corrosion resistant in oxidizing environments that are of interest for the Yucca Mountain Repository. In any corrosion application, the corrosion behavior is determined by the combination of the corrosion resistance of the material and the corrosively of the environment. There are environments that will attack any material. Since no 12 material is immune to all environmental conditions, the challenge is to specify a material with proper corrosion resistance in realistic environments. Alloy 22 belongs to a family of corrosion resistance alloys that have nickel-chrome-molybdenum as the primary alloying constituents. These alloys depend upon the formation and the tenacity of a passive film, i.e. a thin, oxide on the surface, for their corrosion resistance. Measured corrosion rates for passive metals are on the order of 0.1 to 0.01 microns per year. At these rates, it takes 10,000 to 100,000 years to penetrate 1 millimeter of metal. So for passive metals that remain passive, extremely long lives are not unrealistic. The challenge is to determine the boundaries of passive behavior and to compare these boundaries to the realistic environments that pertain to the repository. Localized corrosion processes (pitting, crevice corrosion, stress corrosion cracking) are the most likely degradation modes to be considered in the repository environments and with materials that are being selected. So basing materials selection and design on high localized corrosion resistance is both prudent and a well-accepted corrosion engineering practice. A montage of figures for localized corrosion is shown in Figure 3. There is a rich literature on the chemical, electrochemical, metallurgical processes that pertain for localized corrosion. The figure in the upper left shows the corrosion behavior of a series of alloys in an oxidizing-acid environment in laboratory tests. In this environment, Alloy 825 did not fare well and exhibited severe localized corrosion. In the same environment, Alloy 22 and titanium showed no corrosion. The schematic polarization curves at the bottom of the figure provide a well-accepted method for determination of susceptibility to localized corrosion under laboratory conditions. Comparison of the corrosion potential and the repassivation potential provides a criterion for initiation of crevice corrosion. The rationale is that if the passive film is disturbed at a potential above the repassivation potential, then the crevice can remain active for extended periods of time and significant corrosion damage could occur. The figures in the upper right show conditions within a crevice before and after initiation of crevice corrosion. Figure 4 describes the "susceptible zone" for localized corrosion related to the corrosion resistance of the metal and the corrosivity of the environment. For localized corrosion to occur, the conditions must be in the region of overlap. For realistic conditions, if there is no overlap between the corrosive environments and the susceptible material, localized corrosion will not occur. If the fields overlap, then it is only those environments that overlap with the corrosion resistance of the material that could cause localized corrosion. Where an overlap (susceptible zone) exists there are additional requirements for crevice corrosion to occur. Water of the aggressive composition has to form and persist on the waste package surface, and these conditions must to persist over a long enough time to cause damage. Thus, a crucial issue is the time evolution of the corrosive environment on metal surfaces under realistic conditions for the repository. Figure 5 shows many of the processes that pertain to localized corrosion. These processes pertain to the initiation, propagation, stifling, and arrest stages of localized corrosion. The process is affected by the crevice geometry and properties of the crevice former. For waste packages, metal surfaces can be covered with dust. Particulate, scale and deposits can form from dust, minerals from waters, and corrosion products. An issue is how effective particulate layers and deposits are as crevice formers compared to metal/metal crevices and polymer/metal crevices used in laboratory tests. Chemical, electrochemical and metallurgical factors control the formation and evolution of the crevice chemistry. For crevice corrosion to persist, the critical crevice chemistry must be formed and maintained. Important questions for determination of damage evolution with time are do conditions exist to support localized corrosion, will the corrosion initiate, will it persist, and what damage might result? A localized corrosion fault tree approach identifies key factors to determine the evolution of corrosion damage over long exposure periods: Does a "susceptible zone" exist where corrosive waters that can initiate localized corrosion exist on the waste package surface? Is the corrosion potential under natural conditions more positive than the repassivation potential? Is there a sufficient crevice former on the waste packager surface to support crevice corrosion? If the above criteria are met, then localized corrosion can initiate. 13 What is the initial corrosion penetration rate within the crevice, and how does the rate change with time? For localized corrosion, a power law dependence often pertains and the penetration rate decreases with time. For determination of waste package performance, the above would be applied over the five relevant time periods in the repository. . Time Evolution of Environment A crucial issue is the time evolution of the corrosive environment on metal surfaces under realistic conditions for the repository. Thin layers of electrolyte, particulate and deposits are the conditions of interest. A methodology is available to determine the restricted time period when a waste package is susceptible to localized corrosion and information on the solution chemistry during that period. The time evolution of the movement of water into the tunnels and the composition of the waters on waste package surfaces are primary concerns. The flux of water in the repository is described similarly to annual rainfall and at the repository level is approximately 5 mm per year. When the temperature of the tunnel walls are above boiling, a thermal barrier exists that vaporizes water and no drips or seepage into the tunnels will occur. Strong capillary forces in the rock also hinder dripping and cause significant amounts of water to remain in the rock and divert around the tunnels. Nevertheless, some dripping and seepage into the tunnels can occur. Water that seeps into the drifts and drips onto waste package surfaces will create an aqueous environment along with any condensation or deliquescence on waste package surfaces. Dilute waters dripping onto hot surfaces can evolve to more concentrated solutions through evaporation. It is well accepted that dry metals, without the presence of an aqueous phase, do not corrode at an appreciable rate in the repository environment. Furthermore, full immersion in waters will not occur under any realistic scenario. Corrosion in thin layers of electrolyte, particulate and deposits are the conditions of interest. These thin layers of moisture are a sufficient aqueous environment to support electrochemical dissolution. Anodes, cathodes and the electrochemical corrosion cell can operate in a thin moisture layer. The temperature and relative humidity over time are determined by thermal-hydrological analysis, and the temperature-relative humidity of a medium temperature waste package was shown above. Based on these data and the critical corrosion temperature, the five relevant time periods were described. Other aqueous solution properties of interest include the acidity/alkalinity (pH), oxidizing potential (Eh) and chemical composition of waters. The pH and Eh are determining factors for stability of passive films and localized corrosion processes. The mountain "breathes" and the tunnels are at atmospheric pressure. The waste packages sit upon supports in air which results in an oxidizing environment. The ambient waters at Yucca Mountain are non-corrosive and benign. The ambient waters are near-neutral with low amounts (parts per million) of dissolved solids and mixed salts. There are a number of anions, cations, salts and minerals available. Chemical divide methodology is used to determine the aqueous solution types resulting from the concentration of dilute waters. The types of waters are shown in Figure 6. A controlling factor is the relative ratio of calcium and carbonate ions in the water. On concentration by evaporation, calcium carbonate precipitates and the predominant ion determines the water type branch that is followed at the divide. This methodology can be used to analyze concentration processes such as evaporation of dilute solutions. The corrosion behavior of waste package and drip shield materials can be determined in the various types of solutions, and the results feed into performance and life prediction. 14 Figure 7 is a montage to show several aqueous solution chemistry principles that pertain to the determination of repository environments. The figure in the upper left is a plot of relative humidity and temperature and indicates conditions (to left and above the line) that are inaccessible because the repository is at atmospheric pressure. For any condition in the inaccessible region, water would vaporize, and there can be no aqueous solution. The figure in the upper right shows the deliquescence curves for a number of salts. These identify the minimum relative humidity necessary to form an aqueous solution of the salt as a function of temperature. For conditions below the curve, a dry salt would be present. The behavior of a salt mixture is shown in lower left. The mixture forms an aqueous solution at relative humidity lower than either of the pure salts. The figure in the lower right combines this information and shows the behavior of sodium nitrate and sodium chloride mixtures as a function of temperature and relative humidity. At 100 C, the mixed salts are dry for relative humidity below 45 percent. Between 45 percent and 70 percent relative humidity, an aqueous solution with nitrate:chloride ratio greater than 0.5 will be present. Above 70 percent relative humidity, a solution with nitrate:chloride ratio less than 0.5 will be present. The temperature-relative humidity behavior for a waste package and the solution chemistry behavior are combined in Figure 8 for the time evolution of the environments on the waste package surface. The times and conditions that pertain for Period IV are determined. From the T-RH behavior, the waste package at year 750 is at 101 C, and the relative humidity is 65 percent. That point is shown on the solution chemistry plot. By matching the chemistry constraints to the behavior of the waste package, and the trajectory of possible waters throughout Period IV are determined. For this waste package, the period starts at year 750. As it cools, it follows the chemistry trajectory until it cools to 90 C at year 1375. The nitrate:chloride ratio of 0.5 is shown because at 100 C, a ratio greater than 0.5 was found to inhibit localized corrosion. This methodology gives insight into the times when waste packages are susceptible to localized corrosion and information on the solution chemistry during that period. The analysis also identifies time periods when localized corrosion will not occur under realistic repository conditions. Other Considerations Other considerations for corrosion performance at Yucca Mountain include Stress corrosion cracking Long term stability of alloys Design and fabrication processes Montages for each of the topics are presented. Figure 9 is for stress corrosion cracking. Important factors include residual stresses and specific cracking environments. Alloy stability and aging effects can affect stress corrosion cracking behavior. The microstructural and chemical heterogeneities associated with welds can be a factor. A challenge is the measurement of cracking rates at the very low rates of interest. Advances in both models and experimental methods for SCC can be beneficial. Figure 10 is for long term stability of alloys. The long time periods of interest for Yucca Mountain require consideration of solid-state processes at lower temperatures than pertain for most engineering applications. Extension of the combined experimental and computational approaches being utilized to analyze aging processes for Yucca Mountain conditions can lead to significant benefits to life prediction for other applications as well as for the repository. Figure 11 shows design and fabrication considerations for waste packages and the tunnels (drifts) for the repository. For corrosion, important factors are materials of construction, metallurgical structure and residual stresses. Fabrication processes, particularly welds, are of interest. 15 For more information This section was based upon a presentation the writer made at the Corrosion Workshop and recent updates from a presentation on "Corrosion Resistance of Alloy 22 at the May 18-19, 2004 Spring Board meeting of the United States Nuclear Waste Technical Review Board. Other presentations on corrosion and environments were also made and are available from the NWTRB website http://www.nwtrb.gov. 16 Figure 1-The Yucca Mountain Repository is ~300 meters below surface and ~300 meters above water table. Water transport at Yucca Mountain depends upon external climate, infiltration, movement in unsaturated rock above repository, interaction with drifts and waste packages at repository level, and movement in unsaturated rock below repository to water table. 17 Figure 2-Relevant time periods for corrosion based upon the temperature-relative humidity behavior of a waste package. Relevant Time Periods Regards Corrosion Figure 3-Montage of localized corrosion behavior and processes. For Conditions Below Temp-Relative Humidity behavior as shown Waste Package at 101°C when Drift Wall cooled to 96°C Critical Corrosion Temp 90°C Start to Year 50 Year 50 to ~65 Year ~65 to 750 Year 750 to 1375 Year 1375 and beyond I-Preclosure: II-Heat up: III-Thermal barrier: IV-Cool down post-thermal barrier V-Packages below critical corrosion temperature 18 Figure 3-Montage of localized behavior and processes. Localized Corrosion Processes After Fontana Corrosion Engineering 1.000 0.800 Po ten tial (V ) vs E ref 0.600 0.400 0.200 0.000 -0.200 -10.0 -9.0 -8.0 -7.0 -6.0 -5.0 -4.0 -3.0 -2.0 Log Current Density (A/cm2) 19 Figure 4-Susceptible zone for localized corrosion determined by corrosivity of the environment and corrosion resistance of the metal. 20 Figure 5-Factors affecting localized corrosion processes 21 Figure 6-Application of chemical divide methodology to determine the aqueous solution types resulting from the concentration of dilute waters. 22 Figure 7-Aqueous solution chemistry principles applied that pertain to repository environments on metal surfaces. Solution Chemistry Principles 100 90 80 Relative Humidity (%) 70 60 50 40 30 20 10 0 0 50 100 150 200 Inaccessible region 100 90 K2SO4 KNO3 K2SO4 (BP) Na2SO4 (BP) KCl NaCl NaNO3 KCl (BP) NaCl (BP) KNO3 (BP) NaNO3 (BP) Max RH Boiling Point 100°C MgCl2 KF Max RH Boiling Point 96°C Ca(NO3)2 (BP) CaCl2 (BP) Maximum Relative Humidity Allowable pressure constraints the Relative Humidity above the local boiling point of water Relative Humidity (%) 80 70 60 50 40 30 20 10 0 0 50 KOH MgNO3 K2CO3 Temperature (C) NaOH 100 150 80 Temperature = 100°C Temperature (°C) 75 % Relative Humidity 70 65 60 55 50 45 Aqueous Phase Deliquescence point of pure NaCl Two Phase Region Solid + Aqueous Deliquescence point of pure NaNO3 Deliquescence point of binary mixture Aqueous Phase 40 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 Mole Fraction (NaNO3 / (NaCl+NaNO3) 23 Figure 8-Combination of the temperature-relative humidity behavior and solution chemistry behavior for the time evolution of environments on waste package surfaces. 24 Figure 9- Montage of stress corrosion cracking factors. Stress Corrosion Cracking Notched Keno Specimen Geometry 2.25 (+.00, -.05) 0.8125 (+/ - .01) 0.625 (+/- .02) 0.200 (+/- .02) 0.200 (+/- .02) 0.5 R Typical Important Factors C L Residual stresses Corrosive environment Thread 1/2 - 20 Both Ends D = Gage Section Diameter U-shape Notch Details h r r = 0.010 h = 0.038 D = 0.318 d = 0.242 Alloy stability: aging d D Welds of particular interest Through-Wall Stress in Laser Peened Areas Laser Peening Concept Laser Input at ~ 120-180 J/cm with ~20 ns pulse duration Light-absorbing film on top surface 2 40 Radial Stress, Sx (ksi) 20 0 ~20% Yield -20 -40 -60 ~20% Yield Stress 80 60 Inner ~half of wall has compressive stresses Tamping Material Waste Package Lid Weld Area -80 -100 0 0 0.2 5 Stress less than threshold for initiation of SCC at depth of 2 mm 0.4 10 0.6 15 0.8 20 1 25 Distance from outer surface (mm ) 25 Figure 10- Montage of long term aging of alloys factors. Long Term Stability of Alloys 900 800 Temperature (°C) 700 600 500 400 300 200 100 0 0.00001 0.0001 Projected Repository Peak Temperature 0.001 0.01 0.1 1 10 100 1000 10000 50% GBs Covered Important Factors Aging and long range ordering can affect corrosion resistance Time-temperature-composition Bulk alloy and welds As-welded Alloy 22 (Gas-TungstenArc-Weld, GTAW) Time (Year) AMR (GB Covered) P phase (2%) - Theory 15 Vol. % TCP P phase (15%) - Theory Weld 5 Vol. % 900 800 Tem perature, °C 700 600 500 No Precipitation LRO Partial Coverage of GBs Full Coverage of GBs TCP Precipitation Within Grains and on GBs 400 300 200 1 Temp, °C (no Temperature, Temperature, Temperature, Temperature, GB ppt'n) °C (partial) °C (full) °C (Bulk) °C (LRO) 10 100 1000 10 4 10 5 Time, h 26 Figure 11-Montage of design and fabrication factors. Design and Fabrication Processes Some Important Factors Materials of construction structure Metallurgical Residual stresses Welds of particular interest 27 SELECTED RESEARCH OPPORTUNITIES This report identifies research opportunities in both basic and applied areas. For each of the workshop topic areas, an overarching goal is presented along with selected research opportunities identified from the DOE Corrosion Workshop. Comments and findings are presented from the workshop for each topic areas in separate sections. The topic areas include: Life Prediction and Evolution of Corrosion Damage Evolution of Corrosive Environments Localized Corrosion Passivity Stress Corrosion Cracking Materials Stability and Aging Advanced Materials and Fabrication Methods and Tools Life Prediction and Evolution of Corrosion Damage Goal Quantum advances in state-of-the-art methodologies for life prediction and cumulative damage for highly corrosion resistant materials. This is a primary theme expressed throughout the corrosion workshop and other gatherings of corrosion/materials experts. This is a special time for major advances, and the Yucca Mountain Repository provides an excellent platform for progress that can greatly affect a broad range of industry, government and public. The directions and approach for progress are clear, and a concerted-coordinated effort can realize great gains. The goal of the Predictive Modeling portion of the workshop was to identify long-term research and technology development topics that could dramatically improve methodologies for predicting long-term materials performance and ultimately service life. Three general classes of reliability techniques exist. Two are "statistical" based and are often termed (1) probabilistic risk assessment (commonly used with nuclear power plants), and (2) traditional (or experimental) assessments. The third class of techniques is referred to as computational reliability, and is based on physical models and probability theory. The computational-reliability methodology is the focus of the topic area section on Life Prediction presented later in this report. The common elements that an analytical toolset based on the computational reliability techniques must contain are identified. The context is a presentation of goals to achieve a fully developed predictive capability using this technique. There are opportunities for major advances in life prediction, and a number of significant scientific issues still exist that were the subject of the other workshop topical sessions. Specific analytical tools have been identified whose advancement could produce much more effective and widely accepted predictive capability. Selected research opportunities from the DOE Corrosion Workshop include: Improved (more robust) deterministic models for predicting the accumulation of corrosion damage Precise definition of the evolving environment and changes in material properties and characteristics Improved methods to validate the evolving process models; possibly using suitable less corrosion resistant analogs 28 Develop and apply novel-experimental methods/equipment to support combinatorial and large designed experiments; large-array micro potentiostats; in order to interrogate and define multi-species environmental space at elevated temperatures, and to examine a wide compositional space for Ni-Cr-Mo and Ti alloys. Determination and validation of model input parameters Identify and develop suitable analogs for ALLOY-22 in long-term performance upon which corrosion phenomena can be effectively studied in laboratory time scales Methodologies to achieve more efficient computation times; transition to more computationally efficient mathematical frameworks and models (e.g., fully parallel) Application of cognitive techniques (e. g., neural networks) to identify relationships within databases (data mining) Opportunity Develop a corrosion life performance model for passivity and localized corrosion of highly resistant alloys (Alloy 22 and Ti) capable of predicting damage evolution. Desirable elements within the damage evolution model would include intact film; single pit and crevice, local breakdown/repair; multiple, interacting events; crevice; local deposit; layer deposit (porous, ionic conductive, mixed ionic and electronic; and transition from localized corrosion to stress corrosion crack. Evolution of Corrosive Environments Goal Develop an extensive knowledge base to include the relevant experimental data that would enable development of a robust model capable of predicting the evolution of aqueous solutions on metal surfaces. The evolution of corrosion damage is determined by the combination of the corrosivity of the environment (evolving and potentially changing over time) and the corrosion resistance of the metals of construction. Advances to robust computational and experimental methodologies that enable the determination of the evolution of corrosive environments on metal surfaces over time are required in order to realize major advances to the determination of performance in real systems and over long times. Understanding the evolution and trajectory of water compositions on metal surfaces is central to the determination of corrosion performance over the long-term. The work should focus on concentrated, multi-species solutions in the 80-150 C range. The directions and approach for this research are well defined, and a concertedcoordinated effort can realize great gains. Selected research opportunities from DOE Corrosion Workshop: Enhanced thermodynamic databases for concentrated and high temperature solutions: activity coefficients, vapor pressures, gas solubility, ion/ gas mobility Silicate and brine interactions: kinetics of heterogeneous reactions, silicates/ brines at elevated temperatures Formation and stability of aqueous solutions on metal surface: condensation, vaporization, sorption, gasliquid-solid interactions on metal surface Processes that lead to formation and control the environments in corrosion scale, dust, deposits, crevices Behavior in thin films and porous products/deposits Anodic stifling, e.g. formation of solid Mo and W compounds with time. MIC: determine corrosion response compared to abiotic behavior; effect of organic species; and then effect or behavior in thin layer electrolytes 29 Opportunity Combine expertise of geochemistry/hydrology with electrochemistry and corrosion experts. Develop an extensive experimental base and a robust model for the evolution of aqueous solutions on metal surfaces under relevant conditions. Coordinate with research on passivity and localized corrosion. Localized Corrosion Goal Extend the knowledge base for highly corrosion resistant alloys such as Alloy 22 and titanium in thin electrolyte films. Enhance the theoretical and experimental foundations to enable a robust model for the time evolution of damage from localized corrosion. Include the processed that control corrosion stifling and arrest by repassivation. Localized corrosion remains a most important damage process to be understood, predicted and controlled. Highly corrosion resistant metals are susceptible to damage in extremely harsh environments. There is a sound, scientific framework for understanding localized corrosion; however, there are also great opportunities to advance the state of knowledge. A more comprehensive knowledge of the relationship between alloy composition and complex environments is obtainable through the combination of emerging experimental, analytical and modeling methods. An electrochemical approach to localized corrosion susceptibility based upon potential difference (E), i.e. the difference between the corrosion potential (Ecorr) and the repassivation potential (Erepass) is well founded in corrosion science. Where localized corrosion cannot be ruled out, further development of the understanding of propagation, stifling and arrest, and repassivation mechanisms can have a great impact on determination of the evolution of damage. Combined applications of Fault Tree-Analysis methodology and Damage Evolution procedures provide a promising framework for progress. Research gaps and opportunities generally lie in the categories of improved understandings and conceptual knowledge in the areas of propagation and either repassivation or stifling of local corrosion. Other opportunities were for advanced tools to measure such phenomena, especially in thin electrolyte films, and models or predictive capabilities to predict behavior over time. Selected research gaps and opportunities identified at the DOE Corrosion Workshop include: Dissolution kinetics, propagation laws; and their extension to thin films Stifling, arrest, and repassivation mechanisms Examine propagation and stifling due to cathode limitations, anode/cathode separation, cathode areas and anode limitations, e.g. competition between neighboring local corrosion sites Identify the metallurgical "culprits" in local corrosion/aging Identify role of solid solution alloying Characterize alloy/film compositions in pits governed by kinetics Address specific environmental issues: effects of extrinsic (nitrate, Pb 2+ , F- ) and intrinsic (S) species on local corrosion; organic acids Combinatorial and/or designed experimental approach to characterize a broad range of concentrated, multispecies solutions Understand cooperative interactions during local corrosion that may play a key role in stabilizing local corrosion 30 Evolution of the corrosion potential (Ecorr) and repassivation potential for crevice corrosion (Ecrev) with time and as a function of alloy and environment Opportunity Develop an extensive knowledge base for corrosion resistant alloys in thin electrolyte films that will enable development of a robust model for the time evolution of damage from localized corrosion. Extend experimental data base for corrosion resistant alloys for a wide range of environments as functions of T, Cl-, NO3-, and other relevant species. Also extend to a range of alloy chemistry, microstructure and surface conditions. Include electrochemical measurements and analysis of mixed electrode reaction kinetics. Coordinate with research on passivity and environment. Develop a robust stifling and arrest model. Passivity Goal Develop an extensive scientific foundation of passive film formation, growth and breakdown in corrosion resistant alloys to enable robust modeling of the formation, growth, stability, breakdown and repassivation of passive films including transpassive dissolution. Stable passive films are the means of corrosion resistance of stainless steels, nickel alloys, titanium, zirconium and other passive metals. Progress in the understanding and long-term prediction of passive metal performance requires a multi-disciplinary set of expertise. Recent progress and advances in analysis of thin films, transport behavior in complex oxides, electrochemistry and computational methods provide an excellent foundation for major advances. Selected research opportunities from the DOE Corrosion Workshop include: Coordinate framework to connect modeling efforts with experimental measurements. Create standardized databases on passive film properties to support modeling, data mining. Develop multiple length scale modeling efforts from atomistic to performance assessment. Develop computationally fast models to integrate over complex evolution path. Define reaction pathway and predict multi-layer surface films for passive corrosion in thin aqueous layers Link thermodynamics predictions to kinetic observations to characterize the composition, structure of films and in anticipated evolution with time Determine the role of pre-oxidized state (thermal treatments, dry/humid oxidation) on passive corrosion Create a link between the structure, composition and defect state of surface oxides subsequent electrochemical behavior; understand the influence of specific alloying elements and their distribution; influence of temperature and salinity on these processes Define the relationship between the open circuit potential and film stability. Identify the nature and source of heterogeneities in the alloy and oxide structure; grain boundary impurities and second phases, inter-metallic precipitates, and other metallurgical features Determine failure/repair processes and determine the temporal features of oxide repair events Elucidate processes at the oxide/alloy interface: segregation of impurities and other minor constituents; defect structures (vacancies, concentrations of impurities and alloying elements, integrated stresses); chemical evolution of the alloy at the alloy/oxide interface Define the chemistry in a thin film or a deliquescent layer in-deposit; chemistry versus bulk environment chemistry and incorporate this knowledge into the understanding of passivity 31 Understand cation/anion selective absorption/transport processes Quantify the kinetics of oxidant/reduction processes on passive films: influence of passive film properties on oxygen reduction kinetics, kinetics of nitrate reduction Opportunity Develop the fundamental foundations to understand passivity of corrosion resistant alloys in thin electrolyte films in order to enable corrosion life performance models for the passivity of highly resistant alloys (Alloy 22 and Ti). Coordinate with research on localized corrosion and evolution of the environment. Link thermodynamic predictions to kinetic observations to characterize the composition, structure of films and the anticipated evolution with time. Ultimately, link the findings to life prediction. Stress Corrosion Cracking Goal Improve the fundamentals associated with stress corrosion cracking by resolving certain critical issues. Extend fundamental understanding towards highly corrosion resistant alloys to enable development of a robust model that is capable of describing the time evolution of damage from stress corrosion cracking. Extend the capabilities to measure and predict at the low stress corrosion cracking growth rates that are relevant to slow growth over long periods of time. Advance the SCC theory and test methodology for treating slow, growth rate regimes. The control of residual stresses and avoidance of more susceptible microstructures are of key importance to SCC, and these issues are integrally linked with the manufacturing and fabrication processes. This effort requires the integrated efforts of design/fabrication engineers to define material and mechanical conditions and materials/corrosion engineers/scientists to advance modern SCC theory. Selected research opportunities from the DOE Corrosion Workshop include: Improved mechanistic understanding to validate the film rupture model including crack tip strain rates and advancement in alternative models of SCC Evaluate the possibility of chemical reaction induced crack growth Elucidate environmental degradation mechanisms by unique nanoscale characterization of buried corrosion interfaces and stress corrosion crack tips: crack tip and oxide film Examine crack initiation processes, e.g. localized corrosion to crack transition, Develop more sensitive tools for utilization in accelerated corrosion that correlates with long term nonaccelerated attack; high resolution crack growth measurement in cases where crack growth rate is not accelerated Advance tools for determination of local crack tip chemistry and local crack tip hydrogen concentrations Develop advanced "damage tolerant approaches" for assessment of stress corrosion cracking Opportunities Improve theories of SCC, test methodologies and extend the same towards highly corrosion resistant alloys. Extend the common approaches of materials selection, design, and fabrication for stress corrosion resistance to include a damage tolerant approach for the evolution of stress corrosion damage by cracks should they occur. 32 Materials Stability and Aging Goal Extend current capabilities in computational materials science to more complex alloy systems and applications. Major gains have been made in this area through the linkage of advanced computational modeling in solid state materials science, controlled exposures (e.g. long­term aging), coupled with careful experimental characterization of structure and composition. Selected research opportunities from DOE Corrosion Workshop include: Extend the modeling of phase evolution to higher component systems, e.g. from Ni- Cr- Mo to Ni- Cr- MoW- Fe Enhance the kinetic database to understand time-temperature-transformation behavior Include geometrical effects on diffusion processes to account for microstructure characteristics Extend treatments to welds Incorporate effects of cold-work, increased dislocation densities on aging phenomena Opportunity Major gains in computational modeling, extension of data base properties, controlled experimental exposures and careful characterization. Fabrication and Advanced Materials Goal Achieve state of the art advances in the structure property paradigm such that the influence of fabrication on structure and the resulting materials behavior can be predicted. Pursue opportunities in advanced materials and material processing that might bring about radical improvements in materials reliability against environmental degradation processes. Fabrication is a crosscutting issue that affects the corrosion, stress corrosion cracking and mechanical properties; and it has a major effect on the costs. Welds, cold work and forming processes, and surface condition are important considerations. Selected research opportunities from DOE Corrosion Workshop include: Determine the effects of welds on metallurgical stability and performance Determine the effects of cold work on corrosion and stress corrosion behavior Multiple fabrication steps pertain in many applications, and the cumulative effects of the individual processing steps, govern the performance and cost. Advance methods to account for variability from fabrication on performance Advance methods to account for an as-fabricated condition (i.e., residual stress distribution, microstructure, and compositional inhomogeneity) on SCC, mechanical failure, and corrosion Develop a Damage Evolution approach that is capable of capturing and evaluating the effects of surface condition and fabrication processes on damage evolution. 33 Opportunity Promote integrated efforts of design/fabrication engineers to determine the effects of fabrication on structure and composition and relate these to materials performance and corrosion. Multiple fabrication steps pertain, and the cumulative effects of the individual processing steps govern the performance and cost. Develop relationships to understand how these factors affect properties. Methods and Tools Goal Extend and adapt advances in the state of the art surface science, electrochemical and metallurgical approaches to advance the understanding of the corrosion behavior of corrosion resistant alloys and to advance the modeling of damage accumulation. In the areas specified above, many of the required goals are presently difficult to achieve without further developments in experimental methodologies and computational approaches. Close collaboration with experts outside the area of corrosion science is required for advances in this area. Selected research opportunities from DOE Corrosion Workshop include: Develop in-situ techniques to examine oxide film in aqueous environment, high temperatures (Raman, polarized ellipsometry, various scanning probe tools). Develop high-resolution spectroscopic techniques. Develop an oxygen and ion mobility database Extend electrochemical measurements in "marginal electrolytes"-- in situ techniques. Develop imaging methods for high info density measurements, e.g. scanning electrochemical probes. Examine canary alloys (metals with less corrosion resistance) and apply combinatorial methods to improve understanding of alloying effects. Conduct model testing-studies for model verification, studies aimed at parameter generation. Apply and extend In-situ IR spectroscopy, In-situ synchrotron, Local (AFM based) chemical probes, In-situ 3- D neutron imaging, Synchrotron x-ray microtomography, speckle interferometry, electrochemical imaging, sum frequency generation spectroscopy Examine thin film micro-fabricated structures to increase sensitivity and localize crack growth Develop mechanisms­alloy discovery through the use combinatorial techniques Opportunity Advance techniques to increase fundamental knowledge with emphasis on methods that will support the development and validation of predictive models that can forecast damage evolution. Co-opt expertise from areas such as chemistry and physics. 34 TOPIC AREA GOALS AND OPPORTUNITIES Comments and findings from the workshop topic areas are presented in the following sections. Each group provided input in somewhat different formats; however, the common intent was to identify research opportunities for major advances in corrosion science and the long-term prediction of materials performance. The following topic areas are included: Life Prediction Evolution of Environments Localized Corrosion Passivity Stress Corrosion Cracking Material Stability and Aging Fabrication and Advanced Materials Methods and Tools 35 Predictive Modeling of Life and Materials Performance Digby Macdonald and Jeff Braithwaite with John Beavers, Joe Farmer, Russ Jones, John Scully, J. Woods Halley and David Shoesmith Introduction and Background The goal of the Predictive Modeling portion of the workshop was to identify long-term research and technology development topics that could dramatically improve methodologies for predicting long-term materials performance, damage accumulation and ultimately service life. To better understand the issues associated with this topic, some background information on how the time-based reliability of complex systems is mathematically calculated is instructive. Three general classes of reliability techniques exist. The first two are "statistical" based and are often termed (1) probabilistic risk assessment (commonly used with nuclear power plants), and (2) traditional (or experimental) assessments. Both of these are typically based on the collection of a large number of samples or observations involving failure data from both field (historical) and accelerated (stress) testing. Metrics include failure rate and mean-time-to-failure. Their prime deficiency is their empirical nature, which means that the resultant findings apply to the existing condition of the system and provide a snapshot of its reliability characteristics at the time of sampling. Extrapolation to other possible environmental conditions and clearly to greatly extended service lifetimes is dubious. Importantly, physics-based analyses are difficult to include in these statistical methods. The third class of techniques is referred to as (3) computational reliability, and is based on physical models and probability theory. The techniques utilize certain random or statistically distributed variables to describe uncertainty in physical models. The prime deficiency is that extensive resources are required to develop the assessment toolset. The computational-reliability methodology is the focus of this section, and the following discussion is in the context of goals to achieve a fully developed predictive capability using this technique. There are opportunities for major advances in life prediction in corrosive processes, and a number of significant scientific issues still exist that were the subject of the other workshop topical sessions. Specific analytical tools have been identified whose advancement could produce much more effective and widely accepted predictive capability. Some of these developments are quite formidable and long-term in nature. To permit the prime development issues to be understood, a brief philosophical-type discussion is first presented of the common elements that an analytical toolset based on the computational reliability techniques must contain. Elements of the desired predictive toolset A number of important technical, philosophical and practical issues exist when attempting to predict the future behavior of complex physcio-chemical systems such as waste isolation containers at Yucca Mountain. These issues are not just of academic interest, but can have profound impact on public perception and acceptance. From a strictly technical standpoint, the ultimate solution for materials performance and life prediction would be to develop a completely rigorous and robust analytical toolset that mechanistically accounts for every possible corrosion process and environmental condition that could occur over time in each and all components. The basic mathematical approach that would lead to this highly desirable "bottom-up" capability is relatively straightforward and would consist of the integration of the three computational elements: corrosion process models, a mathematical framework that permits uncertainty/variability to be addressed, and an overarching model for performance assessment that allows system-level failure/reliability to be quantified. The concurrent mathematical integration of these elements would ensure that the myriad of possible degradation and environmental interdependencies are all explicitly addressed. In the first element, the set of deterministic corrosion process models is required to describe every form of materials degradation (e.g., uniform corrosion, localized corrosion, and environmentally assisted cracking) that would affect a change in the ability to perform the desired isolation function. The degradation models can operate alone or in concert with others to degrade performance, accumulate damage and ultimately cause system failure. Importantly, for this capability to be truly predictive, these materials models must have a physical, mechanistic basis, preferably using first principles.. The important feature of determinism is that the solution to the constitutive 36 equations is constrained to that which is physically meaningful by invocation of the natural laws, which are presumed to hold over the desired time period of interest. Ultimately, predictive description of the microscopic processes leading to corrosion must be based on detailed deductions from the fundamental quantum mechanical laws of nature. Though in corrosion processes this is a formidable task, some very substantial progress in this direction has been made in recent years. For example, by linking continuum finite element calculations with atomistic molecular dynamics calculations and quantum mechanical density functional calculations, a group at the University of Southern California recently simulated an entire stress corrosion cracking process in silicon. Long term success in this direction depends on identification of the rate limiting atomic and electronic events (because the entire system cannot be simulated at the electronic and atomic levels), on refinement of the algorithms which link the calculations at different scales and on improvement of algorithms which scale linearly with the numbers of atoms and electrons (to speed the electronic calculations). The explicit inclusion of the second toolset element (algorithms to address the uncertainty and variability inherent in many material properties and the operating environment) is also a mandatory toolset element. Even under the most desirable situation, uncertainty and variability will exist (e.g., temperature, alloy composition). To properly illustrate this need, consider the simple case of a containment vessel that is susceptible to three types of localized corrosion. The damaged area increases gradually with time as corrosion proceeds. At some time, a critical amount of damage is reached, and the container will no longer provide its isolating function as desired. A straightforward deterministic calculation would yield a service-life prediction based normally on average properties. However, if the real-world uncertainty is included in the analysis, a distribution of predicted damage sites with specific values (i.e., the "damage function") can be calculated at each time. The tail of each distribution that is above the critical voided area could represent the probability of failure at that time, which typically occurs at times much shorter than calculated using average values. In high-reliability systems, such as Yucca Mountain, failure may become unacceptable when only a small tail of the distribution exceeds a small critical value. Another very important point to make is that the failure criteria for each area on a container will, in general, be different for each specific location (because of potential differences in the environment, the underlying metal structure, etc.). Therefore, the output from the integrated corrosion model can really only consist of time-based distributions of changes in some metric (e.g., damage area) and not a failure probability, unless the property distributions across the surface can be effectively described. The key enabler of this needed capability involves the development and use of very efficient algorithms that permit parameter uncertainty and variability to be wrapped around the deterministic corrosion process models. Two needs exist that require the integration of the output from the deterministic/uncertainty analysis with a system-level performance assessment model (3rd element). Firstly, because corrosion is the environmentally induced degradation of metals, a precise spatial and time-based definition of the environment is critical. The bane of corrosion engineering is that often such definitions are not available. The performance assessment model must therefore have a highly sophisticated ability to predict the localized and time varying environmental conditions. The corrosion evolutionary path (CEP) is a related term that involves the identification of the variation of environmental parameters that impact the corrosion rate and hence the accumulated damage as a function of time. Included are such parameters as temperature, pH, [Cl-], potential, and ionic strength (that affects solubility of oxygen and ion activity coefficients). Contact of the electrolyte film with the corroding surface as a function of time must also be incorporated. The effective definition of the corrosion evolutionary path will require the integration of sophisticated water chemistry/geochemical codes into the performance assessment model for both the adsorbed water films and other relevant environments. An additional need is a comprehensive database for the thermodynamic properties of ionic species in concentrated electrolyte solutions. Secondly, system failure cannot be determined solely based on waste package corrosion behavior. That is, system failure (and thus decreased reliability) only occurs within the context of the system performance itself. That is, specifically for Yucca Mountain, the radionuclide release and transport is affected by many processes in addition to corrosion (e.g., species transport). Besides the environmental model, the performance assessment model must therefore include fully deterministic, age-aware models of all the other relevant processes. As such, a mathematical linkage between critical aspects dealing with scientific understanding for a wide range of potentially relevant degradation modes and complex environments is identified and maintained all the way up to the system level. 37 Critical unresolved scientific and/or technological opportunities Two general types of significant technical issues are barriers to progress in life prediction modeling and the development of a more rigorous toolset for prediction: (1) properly formulated deterministic process models and (2) effective integration of the three elements of a comprehensive toolset at length scales that reduce the need for wholesale model abstraction, but yet can be computationally solved. To address both of these issues, a process to use the evolving predictive toolset and supporting experimental studies to guide the toolset development process itself is needed. That is, an improved capability is needed to perform effective sensitivity studies to dramatically improve the utilization of available knowledge and resources. Each of these general deficiencies is discussed separately in this subsection. Formulation of proper deterministic models for all relevant corrosion and environmental processes There are a large number of directly coupled environmental, species transport, and materials degradation processes that can occur within the corrosion evolutionary path that need to be characterized at some level (e.g., fundamental, phenomenological, empirical). As examples, consider the following list, each of which can impact component service life: General environment as f(time) o temperature o amount of water and moisture o water chemistry Localized "micro" environment as f(time) o All the above Water and species transport Water adsorption Corrosion processes o general o pitting o crevice o SCC o MIC Metallurgical effects The goal is for our understanding to be sufficient to completely formulate robust, validated, deterministic process models that have a physical basis. The other topical sessions in this workshop addressed how to gain this dramatically improved understanding of the majority of these processes and their findings demonstrate the associated significant technical challenges. Code complexity including length-scale discrepancy The length scales between the important physical processes and the dimensions of the engineered components vary widely. As an example, consider general atmospheric corrosion where electrochemical processes are occurring in adsorbed water layers that are < 10 nm thick. However, the engineered components have dimensions of interest on the m, cm and mm scale. Clearly, corrosion processes cannot all be modeled at the nm level because millions of finite elements would be required to cover non-uniform conditions that can occur on large or complex engineered structures. But, even when these "sub-grid" nm processes are rolled up in constitutive equations at, for example, a m level, convergence of the numerical solution can be achieved only when the time steps are sufficiently small. This situation will dramatically impact computational requirements. Clearly, the entire corroding system cannot be 38 modeled with atomic resolution at the sub-nanometer level because systems on a cm scale involve numbers of atoms of the order of 1023 . The key to microscopic prediction must thus be the identification of rate limiting processes which depend on atomic and electronic dynamics and which can be modeled precisely from first principles. The current practical limits for atomic calculations are about 106 atoms and for electronic calculations, about 103 electrons. By identifying rate limiting steps, the atomic and electronic calculations can be limited to calculations of these feasible orders of magnitude (which are increasing yearly) if they are correctly linked to finite element calculations on larger scales. There is currently a great deal of work in several fields on such multi-scalar calculations and the corrosion community should be able to profit from the advances. Recently Vashista, Kalia and coworkers put together a code and carried out such a multi-scalar simulation of a stress corrosion cracking event in silicon using linked computers in California and Japan and implementing finite element (FE) and molecular dynamics (MD) codes developed at USC with DFT codes developed by a Japanese team [Rachid, et al., 2002]. Such calculations, involving thousands of parallel processing links, represent the future potential for truly predictive corrosion science as well as studies using more phenomenological and more easily implemented models which are currently available and in use. The computational demands are then truly compounded by the coupling with the performance assessment model code that is required for proper environmental definition, species transport behavior, and transfer of corrosion damage results. For the desired code execution, the performance assessment model calculations must be performed as a function of time with the corrosion voiding properties of each component also changing as a function of time, as specified by the integrated corrosion model. Finally, there is the computational impact of including uncertainty around the deterministic calculations An incentive related to computational speed exists for developing composite numerical/analytical algorithms; in which as much as possible of numerical computational functions are replaced by approximate analytical solutions. The decision that has to be made is whether the loss in predictive precision is justified by the gain in speed. A good example of where this has been done is in describing the potential distribution down a crevice that is coupled to processes occurring on the external surfaces. The numerical solution of Poisson's equation is a very lengthy process, because of the small steps required. However, high speed, approximate analytical solutions have been developed that yield results that are imperceptibly different from those given by the lengthy numerical codes. As a second example, the deterministic Coupled Environment Fracture Model (CEFM) was originally developed as a numerically-driven code for calculating crack growth rate for Inter Granular Stress Corrosion Cracking (IGSCC) in Boiling Water (nuclear) Reactor (BWR) primary coolant circuits [Macdonald, Lu, UrquidiMacdonald and Yeh, 1996]. Later, an approximate analytical solution was developed to the problem, resulting in a decrease in the computational time by a factor of 105 to 106 , without significant loss of accuracy. [Engelhardt, Macdonald and Urquidi-Macdonald, 1999.] The issue of the level of detail that is required in a physical model for each type of relevant corrosion process that is being included is often overlooked. The simple answer to this question is that sufficient detail should be included to capture the physics of the system at the relevant length scale. An alternative would be the use of somewhat empirical "abstraction" models, but then the ability to include process interdependencies (i.e. environment/ corrosion/ transport) is lost and, in some cases, so is the mandatory physical basis. It is clear, then, that modeling and in particular deterministic modeling is always a compromise between what is desired and what is possible. However, work that is ongoing will ensure that computational techniques become faster and more efficient leading to the ability to incorporate greater detail into models and hence to make more precise prediction, provided that the theoretical basis remains valid. Sensitivity studies driven by system-level requirements The question of how to most efficiently deploy limited resources is a third major challenge that must be addressed. In general, a system-requirement driven (top-down) approach is the most effective. However, some degree of incompatibility exists between this and the needed bottom-up, rigorous approach desired for the toolset itself. Some aspects of the conflicting needs associated with ongoing decisions related to resource allocation are presented below. The cost of acquiring reliable data typically increases with component complexity. That is, it is cheaper to purchase, test and analyze discrete engineered components in the laboratory than the entire system in an operating, service environment. At the system level alone, confidence is limited, because the population often cannot be properly sampled. Unfortunately, testing cannot only be performed at the materials or component level, because of the increasing level of uncertainty associated with unknown effects (e.g., environmental) and interdependencies. 39 Similarly, there has been an increasing reliance on computer simulations to augment physical experimentation. However, in addition to trying to characterize the uncertainty in relevant parameters, such as environment and material properties, the analysts must also account for the uncertainties introduced by the mathematical abstraction inherent in the simplified process models. As such, proper attention to verification and validation of the analytical models becomes critical. A balance must be obtained between the information gained from testing and modeling at various levels of system indenture, the resources required for performing testing and developing models, and the uncertainties introduced in predicting system performance. The ability to logically combine information from these various areas as well as the organization, characterization, and quantification of this myriad of uncertainties are critical aspects that the chosen model development approach must include. Importantly, although the ultimate objective is to obtain accurate predictions, real merit exists for simply obtaining more robust solutions than are presently possible (i.e. just improving our confidence in the estimates of useful life). The advantages and limitations of using a top-down analytical approach for determining optimum allocation of resources can be illustrated by considering how Bayesian statistical techniques can be incorporated into the system-level analytical framework. The prime advantage of the Bayesian methodology is its demonstrated ability to improve the efficiency of the development process [Gelman, Carlin, Stern and Rubin, 2000; Robinson, 2001]. This gain is possible, because of its required system perspective, the effective use of sensitivity studies and engineering judgment, and the direct mathematical incorporation of information from many different types and levels of sources. Specifically, hierarchical Bayesian methods are particularly suited to this latter capability, in that relationships between test articles can be explicitly included in the analysis. Classical Bayesian methods assume that the articles under test are not related in any manner, even though the articles may be identical. In particular, the use of the modern hierarchical methodology will permit relevant expert opinion, materials aging, field reliability data, and laboratory failure data to be codified and merged. For example, if laboratory experiments and detailed mathematical simulations are too expensive, empirical models based on data from field measurements and/or component and subsystem testing can drive model refinement. In addition, this methodology has the capability to address both failures due to aging effects and random failures due to unaccounted for latent defects. An example of this approach is the development of Artificial Neural Networks to calibrate/validate deterministic models [Macdonald, and Urquidi-Macdonald, 1995]. ANNs are ideal for this purpose, because they do not contain any preconceived physical or mathematical (functional) model and, when operating in the pattern recognition mode with supervised learning, are capable of defining highly non-linear, coupled relationships between the dependent and independent variables in the database for a complex system. The great advantage of this method, besides not being constrained by preconceived dependent/independent variable relationships, is that the process is not dependent on the results of a single study or even whether that single study measured the dependent variable as a function of all of the independent variable, provided that the database is sufficiently large. Finally, it is important to distinguish an ANN from an Expert System; the latter contains preconceived rules (the "opinions" of experts), which may or may not be true (unlike the natural laws that are used in a deterministic model, which are universally true). Thus, the danger with an expert system is that dependent/independent variable relationships may be obscured by the preconceived rules. In developing a modern hierarchical methodology, initial estimates of how corrosion affects service life are formulated primarily using expert opinion and engineering judgment. This paper-type study effectively replaces the first stage of detailed model-development activities. The resulting and typically empirical equations obtained by this method are input into some level of performance assessment model. The apparent sensitivity of the system service life to specific materials degradation processes can then be determined. This type of system-level analyses provides more quantitative understanding about which types of materials aging processes probably contribute most strongly to system failure. Additionally, the sensitivity study results are coupled with additional expert opinion to improve the initial model formulations. This information is then analyzed to identify aspects of materials aging where an enhanced definition/quantification would provide optimal benefit at minimal cost. To aid the model refinement process, results from accelerated-aging studies at the component level can be used in conjunction with traditional lab and field degradation data. Critically, resources must be focused into very selected and specifically identified areas to develop the deterministic "physics of failure" basis required for a truly predictive capability. The information produced by the 40 deterministic models is combined with data from a broad range of related sources, including (1) experimental results from the literature and previous investigations, (2) laboratory materials experiments, (3) laboratory "component level" testing, and (4) historical field-aging data from relevant subsystems with similar designs. This collective knowledge base is then used to refine and improve the initial "skeleton" model and this process is iteratively continued. Often, the translation from the materials physics model to the improved skeleton model requires higherlevel performance assessment model simulations and/or the inclusion of expert opinion. Importantly, if this type of iterative refinement loses focus, the process will tend back towards the simple bottom-up rigorous solution and thus the development process may become unacceptably costly and time consuming. To avoid this situation, significant attention must be paid to keep the scope and detail of the material model refinement to a minimum by performing frequent iterations with the system-level performance assessment model. That is, the system-level model will be used to reveal the changes in (system) sensitivity after the new (skeleton) models are employed. Although the identification of a high-level process to develop the desired analytical capability is conceptually straightforward, achieving success certainly is not. The formidable nature of this activity is the result of several factors including the "soft" reliance on expert opinion (not receptive to rigorous analysis), the existence of a wide range of diverse requirements that must be incorporated, and the need for the toolset to have a very computationally efficient framework. Summary position Recognizing its formidable nature, a "bottom-up" rigorous corrosion model should still be pursued. However, it is recognized that, even with the composite analytical/numerical codes that result from the needed "topdown" system-driven analyses, greatly enhanced computational power will be required eventually to capture all of the significant aspects of relevant corrosion phenomena. Research opportunities The primary opportunities for long-term research are associated with (1) formulation of proper deterministic models and associated quantification of needed modeling parameters, and (2) the identification, development and validation of a new analytical strategy with an improved computational framework for predicting long-term service life of engineered structures. The specific aspects of each of these topical areas that were discussed at the workshop are described in this section. Other topical sessions in this workshop were structured to address the mechanistic underpinnings, identify novel techniques to better characterize the corrosion mechanisms, and identify information required in a proper environmental definition. Importantly, the primary output from all of these related activities flows directly into the enhanced lifetime prediction models. Given this, the intent of this section is to focus primarily on those aspects that specifically relate to predictive modeling. Nevertheless, before the specific gaps in modeling are discussed, a few specific needs that were identified in the modeling session and involve the other topical areas are worth noting: The phase space of susceptibility to local corrosion based on fundamental models. Induction time until initiation of localized corrosion given actual time-based environmental changes, in addition to oxide aging and micro-structural evolution. Prediction of corrosion propagation that includes the corrosion evolutionary path with transient nature of local chemistry and ohmic potential drop along with stifling (delayed repassivation) and arrest of localized corrosion. Further refinement of theories to enable prediction of initiation and propagation as a function of environment and alloy and that can span length scales 41 Corrosion-process models and model parameters The core of the desired predictive toolset consists of deterministic, physics-based models of the relevant corrosion and environmental processes. Of note, the computational efficiency and complexity considerations of the overall mathematical framework must be taken into account in the formulation of these models. Factors include the large number of physical and chemical processes that are involved in corrosion reactions, the differences between component and process length scales, the large number of required computations that often will be multidimensional, and the need to sometimes include moving boundaries. Taken together, these factors will probably force the ultimate use of continuum-level models in which the real mechanistic processes occur at a length scale much smaller than that associated with any finite elements used. Nevertheless, the ultimate goal is to develop deterministic models that capture the mechanistic essence of the damaging processes, that require minimal calibration, and that can be used (possibly within the continuum models) to predict the evolution of corrosion damage over the requisite time to within the required engineering accuracy. In addition to the opportunities associated with specific aspects of deterministic model development, two nontraditional types of activities were identified that should be considered in more detail: Create the opportunity and resource environment that will encourage the development of new deterministic models for predicting the accumulation of corrosion damage, and Identify how to properly validate these process models, possibly by using suitable, less corrosion-resistant analogs for which corrosion phenomena can be effectively studied in laboratory time scales. Specific deterministic process models General corrosion: An example of a deterministic model for general corrosion is the General Corrosion Model. This model is based on the Mixed Potential Model that was previously developed for calculating corrosion potentials and general corrosion rates of Type 304 SS in Boiling Water Reactor primary coolant circuits [Macdonald, 1992]. Under simplified aggressive conditions, it yields reasonable values for the accumulated damage and a corresponding average corrosion rate. There are a number of ways in which this model can be significantly improved. First, the passive film is assumed to comprise only of the barrier layer and no account is taken of the formation of the outer layer via the precipitation of oxides, hydroxides, and oxyhydroxides at the barrier layer/solution interface. The outer layer may also contain alien species (e.g., mineral salts) that are incorporated as it forms on the surface. In any event, the formation of the outer layer may result in the corrosion rate being considerably reduced, particularly at long exposure times, by restricting access of the cathodic depolarizer (e.g., O2 ) to the surface, or by inhibiting the dissolution of the barrier layer, or both, thereby leading to a reduction of the accumulated damage. This is a possible explanation of the steady decrease with exposure time in the corrosion rate of passive metals in many applications. A second aspect that requires attention involves the role of the barrier layer in the kinetics of the cathodic reactions. Thus, current charge transfer theory posits that the transfer of charge carriers across a passive film occurs by direct or indirect quantum mechanical tunneling of electrons or electron holes, the probability of which is a very sensitive function of the potential profile and barrier layer thickness. Thus, any realistic calculation of the tunneling probability, and hence of the exchange current density, for a redox reaction (e.g., O2 reduction), will require significant revision of current charge transfer theory. Localized corrosion: The theory for passivity breakdown, in the form of the Point Defect Model [Macdonald, 1999] and the prediction of accumulated localized corrosion damage in the form of Damage Function Analysis [Macdonald, Engelhardt, Jayaweera, Priyantha and Davydov, 2003] have been developed and tested. However, the current theory addresses only constant environmental conditions and hence is not ideally suited for calculating the nucleation rate of pits along a corrosion evolutionary path where the conditions change with time. The road to generalizing the models to accommodate variable conditions has been examined and defined, and no major impediments to modifying the current theory to accomplish that task have been identified. Furthermore, methods have been developed recently for examining the role of aggressive species, such as Cl-, in great detail and, when applied to passivity breakdown on nickel in chloride-containing solutions, confirmed that chloride ion catalyses the ejection of cations from the barrier layer with the concomitant generation of cation vacancies. Because Alloy-22 is a nickel-based alloy, albeit one whose passive film at low voltages is n-type in electronic character, rather than p-type, as is that on nickel, it is expected that the mechanism of passivity breakdown will be the same, 42 but this needs to be demonstrated experimentally. It is important to note that the passive film on Alloy-22 becomes p-type at voltages above about 0.6 Vshe, a voltage that appears to be inaccessible under open circuit conditions. Also, while the current, "coupled environment" models for pit (and crack) growth are highly deterministic and are well founded in electrochemical, dilute solution, and mass transport theory [Engelhardt, Macdonald and UrquidiMacdonald, 1999], they need to be rendered more sophisticated by the inclusion of concentrated solution theory, for example. Perhaps the greatest needs are the development of firm theoretical bases for prompt repassivation (meta stable pitting) and, in particular, delayed repassivation ("stifling"). That delayed repassivation, which describes the death of stable pits, has a profound impact on the development of localized corrosion damage is clearly evident from simulated damage functions [Macdonald, Engelhardt, Jayaweera, Priyantha and Davydov, 2003]. There is also the need to incorporate and cope with initiation of corrosion at microstructural heterogeneity's. Environmentally Assisted Cracking: The key aspects relative to a predictive capability for environmentally assisted cracking are covered in the Stress Corrosion Cracking section of this report. No additional relevant information on this subject was identified in the modeling sessions. Modeling parameters: kinetic, physical, thermodynamic and environmental A critical, but sometimes overlooked, requirement is that all the parameters associated with the final predictive models must be precisely characterized, including those that contain uncertainty. Often, this process is much more difficult and resource intensive than the formulation of the models themselves. All of the deterministic models require that values for various model parameters be determined by independent experiment or calculation, if possible. If that is not possible for any given parameter, then the value may be determined by calibration of the model against other dependent variables and independent variable combinations. For example, all but one of the parameters contained within the Point Defect Model, which yields the metastable pit nucleation rate, may be determined using electrochemical impedance spectroscopy. In addition to the obvious kinetic, physical property, and environmental parameters, thermodynamic properties for many species may also have to be re-determined experimentally, particularly those involved in complex equilibria involving mineralogical species at elevated temperatures and at high concentration. The specific opportunities that have to be explored in concert with the model development process involve how to properly quantify these model input parameters. One potentially attractive option involves the application of cognitive techniques (e.g., neural networks) to identify dependent-independent variable relationships within existing and future empirical databases (data mining). The use of Artificial Neural Networks operating in the pattern recognition mode and trained under supervised learning protocols could lead to the identification of mechanistic information, that may otherwise be missed because of the multivariate nature of the problem, and complex, nonlinear relationships between the dependent and independent variables. These relationships need to be carefully defined by using the most advanced cognitive techniques. Once defined, the relationships need to be used to assess various existing, deterministic models for predicting the dependent variable and to guide the modification and development of more advanced models. The artificial neural networks may be used themselves to predict damage; their advantage in the empirical sense being that they do not contain any preconceived mathematical or physical model. Finally, artificial neural networks are powerful tools for assessing the relative importance of various independent variables and hence for recognizing where emphasis should be placed in experimental programs and model development. System-driven analysis strategy and improved computational framework With the clear recognition that a rigorous "do it all" predictive toolset that includes all process models and needed parameters will be difficult to fully attain, an alternative code development approach must be defined and implemented that will allow the necessary deterministic process models to be identified, efficiently formulated, and then subsequently incorporated into a mathematical framework that can be solved using modern computing resources (i.e. a much more computationally efficient, fully parallel code). (1) The first driver involves prioritization of the needed work using an "importance analysis." With this approach, decisions are based on the probability and potential severity of the related consequence at the accessible environment. That is, compromise of 43 selected barriers must be allowed to permit a consequence to occur. (2) The capability must exist to incorporate relevant information from all available sources (e.g., scientific discovery, natural analogs, field observations). The final system-level consideration is the need for a more computationally efficient mathematical solution framework. Solving this deficiency is the true key to the required mathematical integration of the detailed corrosion process, species transport, and environmental definition models that, in turn, will permit both the explicit inclusion of interdependencies and a time variant response. By explicitly addressing the significant corrosion process interdependencies, some of the sources of non-conservative assumptions in the current life-prediction models would be eliminated. Otherwise, empirical-type abstraction models will still have to be used. Also of importance, this advancement could reduce the need to constantly invoke the approach of applying some of the very conservative assumptions which would permit much more realistic assessments. Potential Impact of the Research Advancements in life prediction and materials performance modeling can benefit the Yucca Mountain Repository and several other DOE and DoD activities that involve effective predictive lifetime models for complex systems (e.g., aircraft, weapons).The primary benefit of improved life prediction modeling is a dramatically increased confidence in the ability of engineered structures and systems to perform their intended function. The implementation of a new, more effective approach could lead, in the end, to substantially reduced technology development, service analysis and characterization, and component fabrication costs. For example, if the lifetime analyses show that the design is overly conservative, substantial capital costs may be realized, because the use of lower cost materials may be justified. References for Predective Modeling of Life and Materials Performance Engelhardt, G.R., D.D. Macdonald and M. Urquidi-Macdonald, Corrosion Science., 41, 2267, (1999). Engelhardt, G.R., D.D. Macdonald and M. Urquidi-Macdonald, "Development of Fast Algorithms for Estimating Stress Corrosion Crack Growth Rate", Corrosion Science, 41, 2267-2302, (1999). Gelman, A., J. Carlin, H. Stern and D. Rubin, Bayesian Data Analysis, Boca Raton, Chapman and Hall/CRC, (2000). Macdonald, D.D., Corrosion, 48(3), 194, (1992). Macdonald, D.D., G.R. Engelhardt, P. Jayaweera, N. Priyantha and A. Davydov, "The Deterministic Prediction of Localized Corrosion Damage to Alloy C-22 HLNW Canisters", Proc. European Corrosion Fed, in press (2003). Macdonald, D.D., P.C. Lu and M. Urquidi-Macdonald, "The Use of Neural Networks in the Prediction of Damage in Water Cooled Nuclear Reactors", Proc. International Symposium on Plant Aging and Life Prediction of Corrodible Structures/95, 1-7, (1995), NACE International, Houston, TX. Macdonald, D.D., P.C. Lu, M. Urquidi-Macdonald and T. K. Yeh, "Theoretical Estimation of Crack Growth Rates in Type 304 Stainless Steel in BWR Coolant Environments", Corrosion, 52(10), 768-785, (1996). Belkada, Rachid; Ogata, Shuji; Shimojo, Fuyuki; Nakano, Aiichiro; Vashishta, Priya; Kalia, Rajiv K., "Mechanisms of Stress Corrosion Cracking in Si: A hybrid Quantum-Mechanical/Molecular-Dynamics Simulation," Materials Research Society Symposium - Proceedings, 750, 2002, p 531-536, and unpublished reports Robinson, D., "Hierarchical Bayes Approach to System Reliability Analysis," SAND2001-3513, Sandia National Laboratories, Albuquerque, NM, (November 2001). 44 Evolution of Corrosive Environments Rob Kelly with Greg Gdowski, Steve Dexter, Roger Newman, Joe Payer, Peter Scott Introduction The evolution of corrosion damage is determined by the combination of the corrosivity of the environment (evolving and potentially changing over time) and the corrosion resistance of the metals of construction. Advances to robust computational and experimental methodologies for the determination of the evolution of corrosive environments on metal surfaces over time are would lead to the realization of major advances in the determination of performance in real systems and over long times. Background: Corrosive Environments Approach by the Yucca Mountain Project For Yucca Mountain, the approach to determining the possible corrosion environments on the waste packages is to combine results from thermal-hydraulic-chemical modeling for a range of scales (mountain, near field, drift) with thermodynamic calculations of the geochemistry and experimental measurements of solution composition as a function of evaporation. In the limit, this approach would lead to an infinite number of solution compositions to be investigated with regards to corrosivity if each ion was considered an independent variable. To surmount this challenge, the chemical divides concept [Garrels, 1967; Eugster, 1978] can be used to group solution into categories based on chemical similarity described by a small number of major ions. The power of this standard approach is that the evolution of a solution during evaporation can be predicted with knowledge of the initial dilute water composition. The chemical divides approach and the geochemistry of multi-species solutions can be used for evaluating the plausible extremes of the chemical composition of the aqueous solutions that develop on the waste package [Gdowski, 2001]. Testing in environments has focused on the use of concentrated simulated waters. Evaporative concentration effects have been simulated by increasing concentrations of the species. In addition, solutions with substantially lower and higher pH have been studied [Estill, 1998]. Testing is conducted over a range of temperatures. The boiling point of pure water at the elevation of Yucca Mountain is 96°C; however, concentrated salt solutions can exist at considerably higher temperatures. Coordinated evaporative concentration and evaporative drip corrosion tests have been conducted in which simulated waters are dripped onto hot metal specimens to better define plausible extremes of the environments that develop on waste package and drip shield surfaces. Summary comments regarding the environments relevant at Yucca Mountain include: Fully immersed conditions on metal surfaces are not relevant to waste package performance. Waste packages are placed in the unsaturated zone of rock some 300 meters above the water table. Rather than full immersion, the environment comprises thin layers of moisture on metal surfaces and episodic dripping onto some waste packages for some of the time. It is expected that after an initial dry period of tens of years during emplacement of waste packages and following the repository heating to above boiling after closure, the repository cools and relative humidity rises in the disposal drifts. Aqueous solutions can be expected to form on the waste package at relative humidity substantially less than saturation due to the presence of deliquescent salts and/or capillary condensation on dust and particulates. These conditions will lead to formation of aqueous solutions at temperatures above the boiling point of pure water at the elevation of Yucca Mountain (96 C). As the temperature of the surface decreases, the relative humidity increases, more water condenses on the surface, and dilution of the aqueous solution on the waste package occurs. The relative concentrations of the dissolved salts change as less deliquescent salts become increasingly soluble. 45 It is highly unlikely that a low pH (i.e., less than pH 3) aqueous solution can be maintained on the waste package in contact with the gaseous environment surrounding it due to the limits on the system pressure. The acid gases (e.g., HCl, HNO3 ) would have to be maintained at pressures well above those attainable in the repository in order to maintain the pH of the solutions at low values. The metal surfaces can be covered by dust and particulates. Deposits of salts and minerals in the water can form as well as the accumulation of any prior corrosion products. Crevice corrosion effects from scale, rock and metal contact are to be considered. A technically sound approach to the problem of determining the environment on the waste packages is to define the physical/chemical bounds of environments that can be expected. This approach involves considering known physical and chemical processes, inherent variability throughout the repository, and uncertainties in quantitative determination of the coupled processes that affect the environment. A comparison of the corrosion performance bounds to the environment bounds guides the determination of corrosion damage evolution over time. Opportunities for Advances in the Determination of the Evolution of Environment Realistic extremes of environments and interactions with corrosion rates There are interdependencies between the corrosion rate evolution and realistic extremes of environments. Understanding those interactions is the foundation upon which any scientifically valid prediction of damage evolution must be built. Aspects of the interrelationship include: Character of the three likely waste package surface conditions o Moist dust from condensation o Wet mineral deposits from dripping and evaporation o Crevices in which accumulation of species can occur Effects of the evolution of the temperature and relative humidity o Establishment of the true deliquescence points for the three surface conditions o Establishment of the relative contributions of transient chemistries to the steady state, i.e., determination of the role of episodic dripping on waste package environment evolution Plausible extremes of chemistry of thin electrolyte layers under the three surface conditions including the contributions of: o Acid gas release If acids were to form in solution during the corrosion process at Yucca Mountain, it is expected that they would quickly volatilize due to the fact that the mountain "breathes," i.e., it is a constant pressure system. Therefore, an understanding of the fate of volatile acids (HCl, H2 SO4 ) and the sorption of CO2 is of interest with regards to the effects on the surface chemistry. o Microbial influences Microbial effects on corrosion are well known [Stoeker, 2001], including the ability of microbes to thrive under extremes of temperature and salinity. The effects are due to changes in the local chemistry due to microbe metabolism and the regions of physical occlusion their colonies create on surfaces. Determination of the extent to which microbes could affect the plausible extremes of environments is of interest to establish the effects on the overlap with the windows of susceptibility of the materials of construction. o Oxidizers present and effects Passive materials exhibit increasing corrosion potentials with time due to a decrease in the passive current density [Kelly, 2002]. One aspect of material susceptibility is the concept of a critical potential above which stable localized corrosion can be sustained [Dunn, 2000]. What bounds the maximum corrosion potential (Ecorr) is the combination of this passive current density evolution with kinetics of the cathodic reactions [Hoare, 1975]. The latter depend on the nature and concentration of the oxidizers present. Characterization of the surface chemistry in terms of its oxidative power is important to predicting the evolution of corrosion damage. 46 Transport measurement and modeling related to corrosion The ease with which mass and charge can be transported often exerts a controlling effect on the location and rate of electrochemical reactions, and thus on the spatial distribution and severity of the corrosion damage evolution. Measurement of transport in thin electrolyte is notoriously difficult, and modeling of the controlling processes with relevant boundary conditions is in its infancy. Modeling and measurement necessarily interact; accurate boundary conditions are needed for accurate models, and modeling results can serve as intelligent guides for measurements by identifying areas of the relevant parameter space within which large changes occur. In practice, modeling and measurement are iterated to best characterize the corrosion situation. Specific opportunities for advances are: Measurements of conductivity and chemical composition under the three surface conditions are needed to quantify the effects of important system parameters (e.g., temperature, relative humidity). Measurements of electrode kinetics are needed under the appropriate conditions, including surface condition, temperature, relative humidity, and mass transport. The most advanced science-based models of thin electrolyte layers in contact with gaseous atmospheres and subsequent atmospheric corrosion [Graedel, 1996] have necessarily used oversimplified electrochemical boundary conditions and have considered one spatial dimension (that perpendicular to the surface). Progress in such modeling requires progress in the application of realistic electrochemical boundary conditions, including transient effects. In addition, lateral interactions will be critical at Yucca Mountain; during some time periods, wetting will occur on limited surface regions. The ability of these limited-area cathodes to provide support for localized corrosion will likely be rate-controlling. Transport parameters are a strong function of concentration and temperature. Only a few current corrosion mass transport models deal with non-ideal transport parameters [Gartland, 1996; Anderko, 2004]. Although object-oriented design computing has made implementation of non-ideal parameters possible [Stewart, 1999], issues remain with regards to defensible "interpolation" between known points. In addition to the need for computational implementation, data are needed that are applicable to the conditions at Yucca Mountain. The coupling of electrochemical and chemical conditions within thin electrolyte layers to those of crevices is required to fully characterize the corrosion situation for each of the three surface conditions. The thin electrolyte layer outside the crevice represents the source of the cathodic reaction primarily responsible for any corrosion growth within the crevice, but the extent to which that area can be effectively coupled to the crevice will be determined by the transport conditions. During the time period of interest, many of the likely crevice locations will be undergoing heat transfer. The heat transfer processes need to be coupled to the mass transport in order to determine the extent to which the time evolution of the chemistry/corrosion are affected. Although heat transfer modeling is at least as mature as mass transfer modeling, the coupling of the two with highly nonlinear boundary conditions represents a special challenge to computational codes. Thermodynamic and physical properties of concentrated, mixed-ion solutions A robust database of thermodynamic, physical and transport properties for the conditions of high temperature, concentrated and multi-species solutions is required for analysis and modeling. Parameters of interest are activity coefficients, boiling points, density, viscosity, diffusivity of water and ions, and solubility. The temperature dependence of these parameters, as well as any strong environmental dependence, also needs to be characterized. Currently, there is no generally accepted theoretical framework that quantitatively can predict all of these properties over the range of dilute solution to precipitation. This knowledge is particularly important when dealing with heat transfer conditions on both the boldly exposed and creviced surfaces. Experimental measurements of relevant parameters in concentrated, mixed-ion solutions are lacking, both because of the lack of validated experimental tools as well as the lack of a theoretical framework that can rationalize and predict property values. 47 Opportunities to Advance State of Knowledge of Corrosion Environments Improved Understanding of Concentrated, Multi-Species Solutions at Elevated Temperature There is a tremendous opportunity for better understanding of the physical chemistry of concentrated solutions. The impact of such understanding on corrosion science would be tremendous, as in a variety of cases, such conditions exist (such as in cracks, pits, and crevices) [Turnbull, 1983]. Dilute solution theory clearly breaks down in thin electrolyte layers during drying, but no generally accepted theory exists to extend to near saturation conditions to predict thermodynamic and physical properties. In addition, accurate measurements of thermodynamic and physical properties are needed to provide a database for modeling. A subtopic would be the mechanisms by which microbes can alter the plausible extreme environments. The influence of microbes on corrosive environments is an area ripe with opportunity for advances to fundamentally sound science. Fundamental Understanding of "Time-of-Wetness" The temporal dependence of thin film electrolytes is often characterized by "time-of-wetness (TOW)" [Jung, 2002]. Although definitions have been standardized [Morcillo, 2002; G-84, 1999], a strong, fundamental understanding does not exist of TOW with regards to its physical meaning as well as its relation to initiation and propagation of corrosion in a thin layer to rationalize alternate immersion aggressiveness. There is a need to better understand what might be considered a trivial question: "When is a surface sufficiently wet to allow corrosion reactions to occur?" Not only is there a highly limited fundamental framework on which to place measurements, but the measurement of time-of-wetness is poorly defined, although standards exist for measurement of TOW. Time-of-wetness makes intuitive sense; corrosion will occur when the surface is wet. Paradoxically, intuition fails when the definition of "wet" is tackled. Quantification of the influence of relative humidity and surface chemistry on the ability of anodic and cathodic sites on a metal surface is poor. A practical application of such knowledge would be a quantitative understanding of the increased aggressiveness of alternate vs. full immersion. Alternate immersion testing represents a poorly characterized corrosion situation, but one that is often preferred in practice because of its significant acceleration of attack. In fact, in some cases, alternate immersion is required for the attack to occur [Craig, 1987; Buchheit, 1995]. A scientific understanding of "time-of-wetness" would represent a quantum leap in corrosion science. Better accelerated tests could be developed, decreasing overall testing time for new materials or surface treatments. In order to achieve such understanding, especially with regards to the initiation and propagation of corrosion in a thin layer, a number of aspects of TOW must be attacked. Mass transport of gaseous species in the system needs better characterization. Gases of importance include oxygen as well as CO2 , which will strongly influence pH in thin electrolytes, as well as any pollutant gases such as SO2 , NOx, and chlorine-containing species [Leygraf, 2000]. The effects of crevices on the TOW are also of importance, with aspects including the geometry of the crevice (gap, depth), the nature of the crevice (metal-metal, metal-deposit, metal-polymer), and the surface energies of the crevice surfaces. Local Corrosion Initiation and Propagation in a Transient, Thin Electrolyte Film The vast majority of localized corrosion studies have used potentiostats to control the electrochemical conditions on the boldly exposed surface. However, actual localized corrosion rarely has such an "infinite cathode;" cathode limitations are particularly relevant to thin electrolyte layers. There is a great need to clarify the factors that control localized corrosion under such conditions in order to determine to what extent experiments and theories developed for full immersion must be modified for such restricted geometries. Transients in the electrolyte thickness and composition are often due to heat transfer effects. Heat transfer will affect evaporation, precipitation, and chemical dissolution of corrosion products, as well as influencing the chemical properties relevant to mass transport. Incorporating heat transfer effects into the understanding of corrosion in thin electrolytes would have a substantial impact on many areas of technology. 48 Occluded Site Chemistry Computational Toolsets Further development of computational toolsets is for the treatment of occluded site chemistry with: complex, concentrated electrolytes departing from dilute ideal binary electrolyte behavior with respect to transport and solubility complex (i.e., non-Tafel, non-linear), environment-dependent interfacial kinetics precipitation/solid products within the occluded site sub-crevices that alter transport within a occluded site incomplete separation of anodes and cathodes such as occurs in a crevice that becomes isolated from the boldly exposed surface due to drying of the outer surfaces In-situ, real-time measurements In-situ, real-time measurements of ionic content (including pH), potential, conductivity without affecting crevice gap for gaps < 10 microns. Much of the chemical sensor community is working on miniaturization of chemical sensing of solution composition to make it compatible with microelectronics fabrication [Lin, 2000; Tang, 2002]. One driver for this is the desire to lower the expense of manufacture to make such sensors attractive to the large-scale consumer applications (e.g., biomedical sensing). It should be possible to leverage such work to make sensors that can be placed on surfaces or within tight crevices characteristic of the conditions under which Ni-Cr-Mo alloys corrode. Predictive Capabilities The ability to accurately predict both transient and steady state chemical composition of electrolytes under heat transfer conditions. The opportunity here is to take advantage of the increases in computational power. The conceptual model and descriptive mathematics of transient properties are not intractable. The solutions to the descriptive models using realistic boundary conditions are computationally challenging. The modeling is not amenable to parallel computing because the processes are so tightly coupled. Instead, raw computing power combined with highly computationally-efficient differential equation solvers is required. In addition, there is an opportunity to take advantage of the ability of object-oriented programming to handle inputs of differing degrees of information (e.g., if a great deal of information is known about the transport properties of chloride, but less is known about sulfate) which will certainly be the case. Many of the advances can occur through careful coupling of stateof-the-art codes for the different processes. References for Evolution of Corrosive Environments Anderko, A., N. Sridhar and D.S. Dunn, "A General Model for the Repassivation Potential as a Function of Multiple Aqueous Solution Species", Corrosion Science, 46(7), 1583-1612, (2004). Buchheit, R.G., F.D. Wall, G.E. Stoner and J.P. Moran, "Anodic Dssolution-Based Mechanism for the Rapid Cracking, Preexposure Phenomenon Demonstrated by Aluminum-Lithium-Copper Alloys", Corrosion (Houston), 51(6), 417-28, (1995). Craig, J.G., R.C. Newman, M.R. Jarrett and N.J.H. Holroyd, "Local Chemistry of Stress-Corrosion Cracking in Aluminum-Lithium-Copper-Magnesium Alloys", Journal de Physique, Colloque , (C3), C3-825/C3-833, (1987). Dunn, D.S., G.A. Cragnolino and N. Sridhar, "An Electrochemical Approach to Predicting Long-Term Localized Corrosion of Corrosion-Resistant High-Level Waste Container Materials", Corrosion (Houston), 56(1), 90-104, (2000). 49 Estill, J.C., S. Doughty, G.E. Gdowski, S. Gordon, K. King, R.D. McCright and F. Wang, "Long-Term Corrosion Testing of Candidate Materials for High-Level Radioactive Waste Containment", High-Level Radioactive Waste Management, Proceedings of the International Conference, 8th, Las Vegas, Nev., May 11-14, 1998, 669-673, (1998). Eugster, H. P. and L.A. Hardie, in Lerman, A. (ed.), Lakes: Chemistry, Geology, Physics, Springer-Verlag, New York, (1978). Garrels, R.M. and F.T. Mackenzie, in Gould, R. F., (ed.), Equilibrium Concepts in Natural Water Systems, Ch. 10, Am. Chem. Soc., Advances in Chemistry, (1967). Gartland, P.O., "Modeling of Crevice Processes", Proceedings of Corrosion/96 Research Topical Symposia, Denver, Mar., 1996, 311-339, (1996). Gdowski, G.E. T.J. Wolery and N.D. Rosenberg, "Waste Package Environment for the Yucca Mountain Site Characterization Project", Materials Research Society Symposium Proceedings, 713, (Scientific Basis for Nuclear Waste Management XXV), 29-36, (2002). Graedel, T.E., "GLIDES Model Studies of Aqueous Chemistry. I. Formulation and Potential Applications of the Multi-Regime Model", Corrosion Science, 38(12), 2153-2180, (1996). G84-89(1999)e1, "Standard Practice for Measurement of Time-of-Wetness on Surfaces Exposed to Wetting Conditions as in Atmospheric Corrosion Testing", American Society For Testing and Materials, Philadelphia, (1999). Hailey, P. and G. Gdowski, "Thermogravimetric Thin Aqueous Film Corrosion Studies of Alloy 22; Calcium Chloride Solutions at 150 and Atmospheric Pressure", Materials Research Society Symposium Proceedings (2002), Volume Date 2003, 757(Scientific Basis for Nuclear Waste Management XXVI), 765-770, (2003). Hoare, J. P. (1975), "A Kinetic Study of the Rest Potential on a Platinum|Oxygen Diaphragm Electrode", The Journal of Physical Chemistry, 79(20), 2175-2179, (1975). Jung, Sungwon, Y. Kim, H. Song, S. Lee and Y. Kho, "Atmospheric Corrosion Monitoring with Time-of-Wetness (TOW) Sensor and Thin Film Electric Resistance (TFER) Sensor", Corrosion Science and Technology, 31(5), 383-387, (2002). Kelly, R.G., J.R. Scully, R.G. Buchheit and D.W. Shoesmith, Electrochemical Techniques in Corrosion Engineering, Marcel Dekker, 86, (2002). Leygraf, C. and T. Graedel, Atmospheric Corrosion, John Wiley & Sons, Inc., New York, (2000). Lin, J., "Recent Development and Applications of Optical and Fiber-Optic pH Sensors", TrAC, Trends in Analytical Chemistry, 19(9), 541-552, (200). Morcillo, M., E. Almeida, B. Chico and D. de la Fuente, "Analysis of ISO Standard 9223 (Classification of Corrosivity of Atmospheres) in the Light of Information Obtained in the Ibero-American Micat Project", ASTM Special Technical Publication, STP 1421 (Outdoor Atmospheric Corrosion), 59-72, (2002). Rosenberg, N.D., G.E. Gdowski and K.G. Knauss, "Evaporative Chemical Evolution of Natural Waters at Yucca Mountain, Nevada", Applied Geochemistry, 16(9-10), 1231-1240, (2001). Stewart, K.C., "Intermediate Attack in Crevice Corrosion by Cathodic Focusing", Ph.D. Dissertation, University of Virginia, (1999). Stoecker, J.G., Ed., A Practical Manual on Microbiologically Influenced Corrosion, Vol. 2, NACE, International, Houston, TX, (2001). 50 Tang, T.B., E.A. Johannessen, L. Wang, A. Astaras, M. Ahmadian,, A.F. Murray, J.M. Cooper, S.P. Beaumont, B.W. Flynn and D.R.S. Cumming, "Toward a Miniature Wireless Integrated Multisensor Microsystem for Industrial and Biomedical Applications", IEEE Sensors Journal, 2(6), 628-635, (2002). Turnbull, A., "The Solution Composition and Electrode Potential in Pits, Crevices, and Cracks", Corrosion Science, 23(8), 833-70, (1983). 51 Localized Corrosion John Scully, Hugh Isaacs and Roger Newman with Rudy Bucheit, Rob Kelly, Nancy Missert, Joe Farmer, Steve Dexter and Jerry Frankel The following research opportunities were identified in the area of localized corrosion. Opportunities generally lie in the categories of improved understanding and conceptual knowledge in the areas of crevice corrosion propagation especially in thin electrolyte films (e.g., in the absence of an infinite cathode) as well as both repassivation and stifling of local corrosion. There is also the need for advanced tools to measure such phenomena, especially in thin electrolyte films, and models or predictive capabilities to predict behavior over time. These latter issues are discussed in a separate section regarding techniques and methods. Overview of Localized Corrosion Research Opportunities It is generally recognized that there is a shortage of information concerning the localized corrosion (specifically crevice corrosion) of any corrosion resistant material in thin electrolytes films, especially those environments containing particulate matter or under deposits that might act as reactive, ion selective and semipermiable crevices. In addition, it is clear that the overall topics of crevice repassivation in realistic crevices after long propagation times, and stifling due to corrosion products or cathodic starvation are research gaps. They also provide areas where there is an opportunity to determine if, despite high temperatures and possible global chemistries containing high concentrations of halides, that crevice corrosion just cannot be sustained in very thin electrolyte films or moist dusty layers. Therefore, research opportunities include improved understanding of anodic dissolution kinetics and propagation behavior in pits and crevices with emphasis on understanding the impact of deposits, solution species, surface films, external cathode conditions and alloy composition on stabilization, propagation, repassivation and stifling. Advances in the understanding of the effects metallurgical factors such as alloy composition and intermetallic formation on local corrosion in the Ni-base system are also required. These issues are of particular importance because of concerns over either long term aging and/or fabrication. One of the biggest needs and fertile areas for research is in the area of local corrosion stifling and arrest mechanisms. It is recognized that several aspects of exposure in a repository type situation, e.g. thin electrolytes, non-immersed conditions, and episodic wetting, likely impose additional restrictions to crevice propagation that render stifling and/or repassivation more likely than in full immersion. In addition, evolution of damage depths and morphologies should be understood better in crevices under conditions of cathodic starvation, in the case of resistant alloys, and with highly acidic critical crevice solutions. It was also recognized that there are specific unresolved issues that affect local corrosion behavior such as the effects of extrinsic environmental variables (microbes, nitrate, Pb2+ , etc.) and intrinsic metallurgical (S) species on local corrosion of Ni-base materials. The following topics are presented below: Critical Threshold Conditions for Corrosion (Initiation and) Propagation Anodic Dissolution Kinetics and Propagation Behavior in Pits and Crevices o Solution composition effects and properties of films formed in crevices o Pits and crevices governed by alloy composition Computational Understanding of Critical Crevice Solution Chemistries that Enable Localize Corrosion Propagation Local Corrosion Stifling and Arrest Mechanisms Metallurgical effects in local corrosion/ aging o Role of major solid solution alloying on local corrosion o Role of metallurgical phases Specific environmental species on local corrosion: extrinsic (e.g. nitrate, Pb2+ , F- ) and intrinsic (e.g. S) Effects of Microbiological Activity on Localized Corrosion 52 Critical Threshold Conditions for Corrosion (Initiation and) Propagation The anodic kinetics within corrosion cavities control the propagation, and, according to one view, the `initiation' (as opposed to `nucleation'), of localized corrosion. That is, a small cavity on a metal surface has a certain geometry that sets the initial boundary conditions for mass transport; the anodic reaction in the local environment must be able to sustain sufficient chloride concentration by inward electromigration and acidity (via cation hydrolysis) in the face of outward diffusion of the aggressive environment [J.R. Galvele, 1976]. Sufficient acidity and depassivating anion (i.e., chloride) concentration form a critical crevice chemistry. This simple approach lends itself to a graphical representation defining stable and unstable steady states for localized corrosion [N. J. Laycock and R.C. Newman, 1997]. Another factor to consider is the potential inside the crevice, which will be lower than the external surface potential owing to IR drop from current flow between the active crevice interior and external cathodes. The lower potential can either stabilize the attack by moving the potential into the active range or destabilize the attack by consuming part of the driving force for dissolution. But this is not a static picture: consumption of metal changes the cavity geometry, and may for example undercut the passive surface [P. Ernst, N.J. Laycock, M.H. Moayed and R.C. Newman; N.J. Laycock, S.P. White, J.S. Noh, P.T. Wilson and R.C. Newman, 1998]. Alternatively, corrosion products or precipitates might provide a crevice-like geometry and/or an ion-selective membrane action. A simple view of such complexities is that they alter the effective diffusion length associated with the cavity; the longer this length, the more stable the corrosion and the lower the potential required to stabilize it, other things being equal. For these reasons a more robust critical potential is required rather than relying on the long-term validity of a single short term repassivation potential for localized corrosion Er,crev. In other words, the repassivation potential represents a situation where the critical crevice chemistry is not maintained given the particular details of the crevice and transport conditions. To the extent that the critical local environment requires or implies chloride enrichment and high acidity, that enrichment requires a minimum overpotential referred to the open-circuit potential of the metal in the local environment. Solid-solution alloying enters the above argument via anodic dissolution kinetics and critical solution chemistry for passivation. The difference in behavior between two alloys that are both well above their critical temperatures for localized corrosion (e.g., 304 vs. 316) can be understood by comparing the dissolution kinetics in the worst possible local environment. The difference in local environment dissolution kinetics correlates perfectly with the difference in critical potential, both for chloride solutions where Mo has a large effect, and for bromide solutions where it has little or no effect [Newman, 1985]. A more interesting and complicated situation occurs when passivation intervenes, which can occur even in the most aggressive/concentrated local environment; this has been covered in one model for the CPT and CCT of high-alloy materials [Laycock and Newman, 1998]. Now one should consider the dependence of the generalized critical temperature for localized corrosion on alloy composition and cavity geometry. For a particular temperature, as Mo is added to an alloy, passivation is introduced at more and more concentrated local chemistries. For pit initiation on a smooth metal surface in a bulk electrolyte, this soon renders pit initiation impossible at any potential and the CPT is increased. For initiation of corrosion under a deposit, the propagation rate required to sustain the necessary local chemistry is so low that a near-saturated metal salt solution is no longer required; 10 percent or even 1 percent of saturation may suffice. This domain of crevice stability needs to be mapped as it is critical to the slow rates of propagation that may occur under dense deposits. It has been identified that threshold potentials and temperatures remain technique- and conditiondependent, and resulting metrics such as Er,crev can depend on geometric and mass transport factors. Crevice studies show that the diffusion length formed by the crevice former governs Er,crev instead of the depth of corrosion attack [Kehler, Ilevbare and Scully, 2001]. A universally versatile threshold parameter that is not experimental technique dependent is needed. It is desirable to move beyond a critical potential for repassivation towards a set of readily obtainable parameters derived from lab testing that hold for a variety of geometries and mass transport situations. This would be analogous to a set of Galvele-type criteria instead of one critical potential [Galvele, 1976)]. For these situations, it would be worth mapping the domain of crevice corrosion stability even when the propagation rate required to sustain the necessary local chemistry is low. There is uncertainty over the electrochemistry of lowpropagation rate crevice corrosion and long diffusion/migration lengths. The open circuit potential in the pit electrolyte is very conservative, but it represents the most conservative potential threshold in the limit of extreme mass transport limitations. 53 Anodic Dissolution Kinetics and Propagation Behavior in Pits and Crevices Growth Rates and Propagation Laws The anodic dissolution behavior of highly corrosion resistant materials, like Alloy 22, in crevices is important for several reasons. Foremost, the anodic behavior in conjunction with geometric and chemical factors governs crevice stability. Secondly, these phenomena govern growth rates, propagation laws, repassivation and stifling. For a thick waste canister, considerable propagation is required for failure by penetration. The mere initiation of localized corrosion, especially for an alloy like Alloy 22, will not constitute failure. Consideration of localized corrosion propagation and stifling is required for a full prediction of failure. The anodic behavior of localized corrosion of Alloy 22 at elevated temperatures appears similar to stainless steels and Ni alloys and is critically related to maintaining the critical concentrated environment required for localized corrosion to propagate by active dissolution. The alloy composition, solution concentration, corrosion products, and potential at the dissolving interface determine the rate of dissolution. The fundamental understanding of dissolution kinetics of complex multicomponent alloys is not well developed. [Landolt, Matlosz and Sato, 1999]. When the concentration drops below a critical value, repassivation takes place. When the concentration exceeds the solubility of the lowest solubility component, salt films precipitate. [Gaudet, Mo, Hatton, Tilley, Isaacs and Newman, 1986]. The dissolution and repassivation kinetics of Fe-Cr alloys in pit solutions have been investigated. [Steinsmo and Isaacs, 1993]. Localized corrosion studies on Alloy 22 are rare, but those published appear consistent with this behavior. [Kehler, Ilevbare and Scully, 2001]. Still, a direct method to quantify this model and note its limitations are required. One particular uncertain aspect is the influence of Mo at the 10-12 wt. percent level; a hypothesis is that new mineral-type species (molybdenum containing phases) provide the means for deterministic stifling. When stable crevice or pitting corrosion occurs in stainless steels and other alloys such as Ti and Al, propagation rates have been explained by ohmic, mass, and charge transfer controlled models [Vetter and Streblow, 1974]. Localized corrosion growth kinetics typically conform to a power law, d = Ktn, where d is depth, t is time, 0