Advancement of Supercritical Carbon Dioxide Technology through

Round Robin Testing and Fundamental Modeling

Reactor Concepts Research Development and Demonstration

Julie TuckerOregon State University

CollaboratorsUniversity of Wisconsin-Madison

Melissa Bates, Federal POCMatthew Walker, Technical POC

Project No. 15-8495

1 

 

Project Title: Advancement of Supercritical Carbon Dioxide Technology through Round Robin Testing and Fundamental Modeling Reporting Frequency: Final Report, March 2019 Recipient: Oregon State University Award number CFA-15-8495 Awarding Agency: U.S. DOE, NEUP Working Partners: Oregon State University University of Wisconsin, Madison Oak Ridge National Laboratory

National Energy Technology Laboratory Carleton University, Canada Korea Advanced Institute of Science and Technology Electric Power Research Institute

Principal Investigator: Julie Tucker, School of Mechanical, Industrial and Manufacturing Engineering, Oregon State University Title: Assistant Professor Phone: 737-541-5840 Email: Julie.Tucker@oregonstate.edu

Collaborators: Líney Árnadóttir - Oregon State University

Mark Anderson - University of Wisconsin-Madison Bruce Pint - Oak Ridge National Laboratory Ömer Doğan - National Energy Technology Laboratory Henry Saari - Carleton University, Canada Changheui Jang - Korea Advanced Institute of Science and Technology, Korea Steven Kung - Electric Power Research Institute

2 

 

Table of Contents

1 Project Overview ..................................................................................................................... 4 1.1 Proposal Abstract ............................................................................................................. 4 1.2 Milestone deliverables outlined by DOE Work Package ................................................. 4 1.3 Collaboration roles and responsibilities ........................................................................... 5

2 Project Status ........................................................................................................................... 6 2.1 Executive summary of achievements ............................................................................... 6 2.2 Budget status .................................................................................................................... 6 2.3 Communication and reporting status ................................................................................ 6

2.3.1 Peer-reviewed publications ....................................................................................... 6 2.3.2 Presentations ............................................................................................................. 7 2.3.3 Student theses............................................................................................................ 9

3 Technical Review .................................................................................................................. 10 3.1 Introduction .................................................................................................................... 10 3.2 sCO2 Materials Working Group ..................................................................................... 10 3.3 Round Robin Test Program ............................................................................................ 11

3.3.1 Introduction ............................................................................................................. 11 3.3.2 Overview ................................................................................................................. 11 3.3.3 Methods................................................................................................................... 11 3.3.4 Results and Discussion ........................................................................................... 13 3.3.1 Summary ................................................................................................................. 32

3.4 Joint Testing in sCO2...................................................................................................... 33 3.4.1 Oxidation behavior of P91-347H welded structures ............................................... 33

3.5 Model Alloy Testing ...................................................................................................... 34 3.5.1 Experimental Procedure .......................................................................................... 35 3.5.2 Results and Discussion ........................................................................................... 37 3.5.3 Summary ................................................................................................................. 43

3.6 Model Alloy Simulations ............................................................................................... 44 3.6.1 Introduction ............................................................................................................. 45 3.6.2 Computational Details ............................................................................................ 46 3.6.3 Surface Models ....................................................................................................... 47

3 

 

3.6.4 Results and Discussion ........................................................................................... 49 3.6.5 Conclusions ............................................................................................................. 68 3.6.6 Further computational details and supporting information. .................................... 69

4 References ............................................................................................................................. 80

   

4 

 

1 PROJECT OVERVIEW 1.1 PROPOSAL ABSTRACT Growing interest in supercritical carbon dioxide (sCO2) cycles for advanced reactors is driving the need for corrosion data on candidate plant materials. The sCO2 Brayton cycle is being considered for power conversion systems for a number of nuclear reactor concepts including the sodium fast reactor, fluoride salt-cooled high temperature reactor, high temperature gas reactor, and several types of small modular reactors. Multiple organizations have developed test facilities to address the corrosion data knowledge gap in high temperature, high pressure sCO2 environments but, to date, there has been no formal test program among these organizations to validate the consistency of the data they produce. A demonstration of comparable and reproducible results enables a coordinated effort to explore the sCO2 parameter space relevant to advanced reactor technology. This proposal establishes a round robin test plan for sCO2 corrosion testing and the organization of a sCO2 Materials Group to guide future materials testing directions. Furthermore, this proposal outlines efforts to elucidate the mechanisms of sCO2 corrosion by performing identical tests in supercritical steam, testing model alloys with various composition and environmental conditions, accompanied by modeling efforts to study rate controlling mechanisms of corrosion and to help understanding the effects of impurities.

1.2 MILESTONE DELIVERABLES OUTLINED BY DOE WORK PACKAGE This project has been designed to address some of the most critical technological barriers hindering the implementation of sCO2 cycles in advanced reactors. The proposed work has near-term benefits with the testing of commercial alloy base metals and joints, medium-term benefits with the validation of different sCO2 test facilities for future testing, and long-term benefits with fundamental computational and experimental studies on model alloys to provide insight into the mechanisms and rate controlling processes of sCO2 corrosion. This insight can direct future alloy development for more corrosion resistant alloys. The project entails international collaboration between multiple PIs to advance the current state of understanding in the area of supercritical carbon dioxide (sCO2) corrosion. This project has four main objectives:

1. Develop a sCO2 Materials Working Group 2. Perform Round Robin Testing of sCO2 3. Test Corrosion and Mechanical Properties of Joints Exposed to sCO2 4. Perform Corrosion Testing and Corrosion Simulations of Model Alloys in sCO2

The major and minor milestones for the project are:

1. Final report 2. Kick off meeting report 3. Status update meeting report 4. Presentation of final data meeting report 5. Development of testing procedures 6. Acquisition and dispersal of test samples 7. Develop post-test characterization methodology

5 

 

8. Report of corrosion sample comparison 9. Determination of joined and model alloys to be tested 10. Acquisition and dispersal of joined and model alloy samples 11. Develop methodology for atomistic and empirical modeling 12. Report out of result on joined and model alloys 13. Comparison of model predictions with test data results

1.3 COLLABORATION ROLES AND RESPONSIBILITIES  Dr. Julie Tucker (Lead-PI) is an Assistant Professor in the School of Mechanical, Industrial and Manufacturing Engineering at Oregon State University. She was the administrative lead on the project (coordinating meetings, workshops, and reports). Additionally, her group performed sCO2 testing and characterization of round robin alloys in research grade CO2. She worked with NETL on the model alloy and joint testing. She assisted in the atomistic and empirical modeling as well. Dr. Líney Árnadóttir is an Assistant Professor in the School of Chemical, Biological and Environmental Engineering at Oregon State University. Her group performed the DFT calculations and KMC simulations of the model alloys and their reactions with sCO2 and impurities. Dr. Mark Anderson is a Professor in the Department of Mechanical Engineering and Director of the University of Wisconsin's Thermal Hydraulic Laboratory. His research group performed sCO2 testing and characterization of round robin alloys in research grade CO2. Dr. Bruce Pint is the Group Leader of the Corrosion Science & Technology Group in the Materials Science & Technology Division at ORNL. ORNL performed sCO2 testing and characterization of round robin alloys in conjunction with current funding from DOE Fossil Energy. Dr. Omer Dogan is a materials scientist at DOE’s National Energy Technology Laboratory. NETL performed testing and characterization of round robin alloys and joined alloys. They also produced, processed and tested the model alloys for sCO2 testing. Dr. Changheui Jang is an associate professor at the department of Nuclear and Quantum Engineering at Korea Advanced Institute of Science and Technology (KAIST), Rep. of Korea. His research group performed sCO2 testing and characterization of round robin alloys in research grade CO2.

Dr. Henry Saari is an Associate Professor in the Department of Mechanical and Aerospace Engineering at Carleton University, Ottawa, Ontario, Canada. His research group performed sCO2 testing and characterization of round robin alloys in research grade CO2. Dr. Steven Kung is a Technical Executive at Electric Power Research Institute (EPRI). EPRI provided round robin samples to all participants and served in an advisory role at meetings and workshops.

6 

 

2 PROJECT STATUS 2.1 EXECUTIVE SUMMARY OF ACHIEVEMENTS This report summarizes the achievements of the DOE NEUP project titled “Advancement of Supercritical Carbon Dioxide Technology through Round Robin Testing and Fundamental Modeling”. Through this project we established a sCO2 Materials Working group to help guide the future direction of this field. We also perform a round robin test program to compare results of sCO2 exposures at five different facilities. Exposure of joints and model alloys were also conducted to provide more fundamental understanding of the corrosion processes and their impact on welds. Lastly, a detailed study using first principles simulations was conducted to assess the role of alloying on carbon and oxygen binding on Ni-alloy surfaces. This works can be used to guide future alloy design efforts to minimize corrosion in sCO2 environments.

2.2 BUDGET STATUS All budgeting items were kept valid.

2.3 COMMUNICATION AND REPORTING STATUS Communication of the research that evolved within this project was a critical component for advancing materials degradation understanding. It allowed for development and understanding of experiment techniques, and opened discussion for the impact and direction of these materials. The primary methods for dissemination under this project included journal publications and conference proceedings.

2.3.1   Peer‐reviewed publications  1) L. Sprowl, J.D. Tucker, B. Adam, L. Árnadóttir, First-Principles Study of the Products of CO2

Dissociation on Nickel-Based Alloys: Trends in Energetics with Alloying Element, Surface Science, 667 (2018), pp. 219-231. https://doi.org/10.1016/j.susc.2018.06.011

2) B. Adam, L. Teeter; J. Mahaffey, M. Anderson, L. Árnadóttir, J.D. Tucker, Effects of corrosion in supercritical CO2 on the microstructural evolution in 800H alloy, Oxidation of Metals, (2018), pp. 1-16. https://doi.org/10.1007/s11085-018-9852-7.

3) L. Teeter, B. Adam, T. Wood, S. Teysseyre, J.D. Tucker. In-Situ Environmentally Induced

Cracking in sCO2, In The 6th International Symposium-Supercritical CO2 Power Cycles. (2018).

4) J. D. Tucker, L. Teeter, M. Anderson, J. Mahaffey, B. Pint, Ö. Doğan, G. R. Holcomb, C. S.

Carney, H. Saari, C. Jang, J. Shingledecker and S. Kung, Supercritical CO2 Round Robin Test Program, In The 6th International Symposium-Supercritical CO2 Power Cycles, (2018).

7 

 

5) R.P. Oleksak, M. Kapoor, D.E. Perea, G.R. Holcomb, Ö.N. Doğan, The Role of Metal Vacancies During High-Temperature Oxidation of Alloys, npj Materials Degradation (2018) https://doi.org/10.1038/s41529-018-0046-1.

6) R.P. Oleksak, J.H. Tylczak, C.S. Carney, G.R. Holcomb, Ö.N. Doğan, High-temperature oxidation

of commercial alloys in supercritical CO2 and related power cycle environments, JOM (2018) 70 1527-1534. https://doi.org/10.1007/s11837-018-2952-7.

7) R.P. Oleksak, J.P. Baltrus, L. Teeter, M. Ziomek-Moroz, Ö.N. Doğan, Characterization of corrosion films formed on austenitic stainless steel in supercritical CO2 direct power cycle environments, Corrosion (2018) 74 (10) 1047-1053.

8) R.P. Oleksak, C.S. Carney, G.R. Holcomb, Ö.N. Doğan, Structural Evolution of a Ni Alloy Surface

During High-Temperature Oxidation, Oxidation of Metals (2018) 90 (1-2) 27-42. https://doi.org /10.1007/s11085-017-9821-6.

9) R.P. Oleksak, J.P. Baltrus, J. Nakano, A. Nakano, G.R. Holcomb, Ö.N. Doğan, Mechanistic

insights into the oxidation behavior of Ni alloys in high-temperature CO2, Corrosion Science (2017), http://dx.doi.org/10.1016/j.corsci.2017.06.005.

10) G.R. Holcomb, C. Carney, and Ö.N. Doğan, Oxidation of Alloys for Energy Applications in

Supercritical CO2 and H2O, Corrosion Science, 109 (2016) 22-35:

2.3.2   Presentations  

1) “Trends in Adsorbate Interactions with Bimetal Surfaces,” Líney Árnadóttir, Gordon Research Conference on Reactions on Surfaces, Ventura, CA 2019. [Poster]

2) “Materials for Supercritical CO2 Applications,”, G. Subbaraman, S. Kung, H. Saari, Tutorial GT2018-77462 at ASME Turbo Expo 2018, Lillestrom, Norway, June 11-15, (2018).

3) “Computational studies of reactions on surfaces and corrosion mechanisms,” University of

Virginia, Charlottesville, VA, department of material science, Líney Árnadóttir, April 31, 2018. [Invited]

4) “Evaluation of Corrosion Behavior of Steels for Direct Supercritical CO2 Power Cycle

Applications,” R. Repukaiti, L. Teeter, M. Ziomek-Moroz, Ö.N. Doğan, R. Thomas, N. Huerta, J.D. Tucker, Corrosion 2018 Conference and Expo, April 15-19, 2018, Phoenix, AZ.

5) “Corrosion behavior of Fe and Ni commercial alloys in direct-fired supercritical CO2 power

cycle environments,” J. H. Tylczak, R. P. Oleksak, G. R. Holcomb, Ö. N. Doğan, The 6th International Symposium-Supercritical CO2 Power Cycles. Pittsburgh, PA, March (2018).

6) “Comparison of Grade 91 and 347H Corrosion Resistance in the Low-Temperature Components of Direct Supercritical CO2 Power Cycle Systems,” R. Repukaiti, L. Teeter, M.

8 

 

Ziomek-Moroz, Ö.N. Doğan, R. Thomas, N. Huerta, J.D. Tucker, The 6th International Symposium-Supercritical CO2 Power Cycles. Pittsburgh, PA, March (2018).

 7) “Oxidation behavior of Fe- and Ni-base alloys in supercritical CO2 and related environments,”

G. R. Holcomb, C. Carney, R. P. Oleksak, J.H. Tylczak, and Ö. N. Doğan, The 6th International Symposium-Supercritical CO2 Power Cycles. Pittsburgh, PA, March (2018).

8) “The fatigue response of nickel superalloys with prior exposure to supercritical CO2,” K.

Rozman, J.A. Hawk, O.N. Dogan, The 6th International Symposium-Supercritical CO2 Power Cycles. Pittsburgh, PA, March (2018).

9) “Supercritical CO2 Round Robin Test Program,” J. D. Tucker, L. Teeter, M. Anderson, J.

Mahaffey, B. Pint, Ö. Doğan, G. R. Holcomb, C. S. Carney, H. Saari, C. Jang, J. Shingledecker and S. Kung, The 6th International Symposium-Supercritical CO2 Power Cycles. Pittsburgh, PA, March (2018).

10) “Materials for Supercritical CO2 Applications,” G. Subbaraman, S. Kung, H. Saari, Tutorial at

6th International Symposium-Supercritical CO2 Power Cycles. Pittsburgh, PA, March (2018).

11) “Trends in Adsorbate Interactions with Bimetal Surfaces” Líney Árnadóttir, AVS-65 international symposium Long Beach CA. October (2018). [Invited]

12) “In-Situ Environmentally Induced Cracking in sCO2,” L. Teeter, B. Adam, T. Wood, S.

Teysseyre, J.D. Tucker, The 6th International Symposium-Supercritical CO2 Power Cycles. Pittsburgh, PA, March (2018).

13) “Trends in Adsorbate Interactions with Bimetal Surfaces,” Líney Árnadóttir, Gordon Research

Conference on Catalysis, Colby-Sawyer College, New London, NH 2018. [Poster]

14) “CO Dissociation on Nickel-Based Alloy Surfaces for Use in Supercritical CO2 Applications,” L.H. Sprowl, Northwest Regional Meeting of the American Chemical Society, Oregon State University, Corvallis OR, June (2017). 

 15) “Density functional theory study of the interactions of C, O, and CO with nickel surface alloys,”

L.H. Sprowl, 253rd ACS San Francisco, CA, March (2017).

16) “Materials and Manufacturing Challenges for Compact Heat Exchangers of Supercritical CO2 Power Cycles,” Ö.N. Doğan, M. Kapoor, R.P. Oleksak, C.S. Carney, R. Saranam, P.S. McNeff, B.K. Paul, The Proceedings of the 42nd International Technical Conference on Clean Energy, June 11-15, Clearwater, FL (2017).

17) “Corrosion Behavior of Steels in Supercritical CO2 for Power Cycle Applications,” R.

Repukaiti, L. Teeter, M. Ziomek-Moroz, Ö.N. Doğan, and J. D. Tucker, Electrochemical Society Transactions, 77 (11) 799-808 (2017).

9 

 

18) “Materials and Manufacturing Challenges for Compact Heat Exchangers,” Ö.N. Doğan, M. Kapoor, R. Saranam, B.K. Paul, The 1st European Seminar on Supercritical CO2 Power Systems, Vienna, Austria, September 29-30, (2016).

19) “Corrosion Behavior of Austenitic Stainless Steel in Supercritical CO2 containing O2 and

H2O,” L. Teeter, N. Huerta, Ö. Doğan, M. Ziomek-Moroz, R. Oleksak, D. Oryshchyn, C. Disenhof, and J. Baltrus, and J. Tucker, Electrochemical Society Transactions, 72 (17) 137-148 (2016).

20) “Materials Performance in Supercritical CO2 in Comparison with Atmospheric Pressure CO2

and Supercritical H2O,” G.R. Holcomb, C. Carney, Ö.N. Doğan, K. Rozman, J.A. Hawk, M.H. Anderson, Proceedings of the 5th International Symposium on Supercritical CO2 Power Cycles, San Antonio, TX, March 28-31, (2016).

21) “High-Temperature Corrosion of Diffusion Bonded Ni-Based Superalloys in CO2,” Ö.N.

Doğan, C. Carney, R. Oleksak, C. Disenhof, G.R. Holcomb, Proceedings of the 5th International Symposium on Supercritical CO2 Power Cycles, San Antonio, TX, March 28-31, (2016).

22) “Materials for Supercritical CO2 Power Cycles,”, G. Subbaraman, J. Shingledecker, H. Saari,

Tutorial at 5th International Symposium on Supercritical CO2 Power Cycles, San Antonio, TX, March 28, (2016).

23) “Supercritical CO2 Round Robin Test Program Update,” J.D. Tucker, MS&T, Salt Lake City,

October 24 (2016). 24) “Corrosion Behavior of Iron-Nickel-Chrome Alloy in Supercritical CO2,” L. Teeter, B. Adam,

J. Mahaffey, M. Anderson, and J. D. Tucker, MS&T16 Technical Meeting and Exhibition, October (2016).

25) “Supercritical-CO2 Round Robin Overview”, J.D. Tucker, sCO2 Heat Exchanger Workshop,

San Diego, CA, October 15 (2015). [Invited]  2.3.3 Student theses 

1) Presented in part in “Analyzing the Kinetics and Thermodynamics of Surface Reactions using Density Functional Theory” Lynza H. Sprowl for the degree of Doctor of Philosophy in Chemical Engineering presented on May 9, 2018. Lynza H. Sprowl was awarded Schulein Outstanding Graduate Student award from CBEE, for her Ph.D. thesis work, June 2018.

     

10 

 

3 TECHNICAL REVIEW 

3.1 INTRODUCTION   This section of the report covers the technical achievements of the project. It is organized to align with our project objectives. Section 3.2 sCO2 discusses the outcomes of sCO2 materials working group. Section 3.3 summarizes the results from the Round Robin test program. Section 3.4 give results joints exposed to sCO2. Section 3.5 cover model alloy exposure testing and section 3.6 covers model alloy simulations.

3.2 SCO2 MATERIALS WORKING GROUP    The goal of the sCO2 working group is to direct and organize materials testing in the sCO2 community. To date there has been no effort to coordinate sCO2 corrosion data collection or storage. The sCO2 working group is being brought together to help prioritize data collection in order to best support the larger sCO2 community. The initial meeting of the sCO2 Round Robin Test Group was held in San Diego, CA on Tuesday Oct. 13th, 2015. This initial meeting of the round robin test group also formed the primary membership of the sCO2 materials working group. The meeting was held in conjunction with the EPRI conference “Corrosion in Power Plants” because all of the round robin test participants were in attendance as well as several key stakeholders from across the sCO2 community. The objective of this meeting was to introduce round robin team members, exchange contact information, provide an overview of the project, discuss team member capabilities, select alloys and discuss testing parameters. The second sCO2 Materials Working Group meeting was held in conjunction with the Supercritical CO2 Symposium held in San Antonio, TX during March 28-31, 2016. The objective of this meeting was to finalize test procedures, sample preparation and characterization procedures. The group also discussed the need for a central sCO2 corrosion database. Another meeting was held in conjunction with the MS&T conference on October 24, 2016 in Salt Lake City, UT. There was a symposium organized by working group members titled “Materials Degradation in sCO2 Environments” and a number of working group members were in attendance at the conference to participate in the symposium. Dr. Tucker presented an update on the round robin testing program to the sCO2 community as part of the symposium. During the meeting, a tour of Ceramatec was organized for the working group members for the following day. The most recent sCO2 working group meeting was held at the sCO2 Symposium in Pittsburgh, PA March 28, 2018. An update on the round robin test was given and an upcoming sCO2 symposium at MS&T 2018 was discussed. Member from the working group had organized this symposium and were encouraging others to submit abstracts. Through these meetings, led by the PI of this NEUP project, the sCO2 materials community has connected and are leading this field. We are organizing symposiums, discussing the need for a centralized database and have created a community for collaboration that was missing before. The working group was a mechanism to bring us together and it will be impactful for years to come.

11 

 

3.3 ROUND ROBIN TEST PROGRAM   

3.3.1 Introduction  Growing interest in supercritical carbon dioxide (sCO2) cycles is driving the need for corrosion data on candidate plant materials. The sCO2 Brayton cycle [1, 2] is being considered for power conversion systems including solar, fossil and nuclear heat sources. Multiple organizations have developed test facilities to address the knowledge gap in corrosion data in high temperature, high pressure sCO2 environments [3-6]. In the past, there has been no formal test program among multiple organizations to validate the consistency of data or to make use of the different facilities to develop a consistent collaborative data generation plan. This type of testing is essential for the development of the sCO2 power cycles. A demonstration of comparable and reproducible results enables a coordinated effort to explore the sCO2 parameter space relevant to advanced reactor technology. This paper presents preliminary results from the round robin test plan for research grade, low flow rate sCO2 corrosion testing among international organizations.

3.3.2 Overview    There are 7 organizations participating in the round robin test program. The organization involved in the project are: Oregon State University (OSU), University of Wisconsin (UW), Oak Ridge National Laboratory (ORNL), National Energy & Technology Laboratory (NETL), Carleton University (CU), Korea Advanced Institute of Science and Technology (KAIST) and the Electric Power Research Institute (EPRI). Six organization have sCO2 testing capabilities and one organization (ERPI) is providing coupons for testing. A description of the testing capabilities at each organization are shown in Table 1.

Table 1. Summary of sCO2 testing capabilities for round robin participants

Organization Maximum Temperature

Maximum Pressure

Chamber Volume

Flow rate (mL/min)

Autoclave Material

OSU 800°C 26 MPa 1235 cm3 0-24 Haynes 230 UW

(2 systems) 750°C 760°C

25 MPa 38 MPa

900 cm3 (combined)

0-24 0-24

Inconel 625 Haynes 282

ORNL 850°C 30 MPa 1400 cm3 0-24 Haynes 282 NETL 800°C 28 MPa 1040 cm3 0-24 Haynes 230

Carleton 750°C 25 MPa 1150 cm3 0-250 Inconel 625 KAIST

(2 systems) 700°C 25 MPa 1077 cm3

(each) 0-24 Inconel 625

3.3.3 Methods    The round robin team has tested five alloys with varying corrosion rates and temperature capabilities: 740H, 625, 316, HR120 and Grade 91. The chemical compositions of these alloys is provided in Table 2. Most alloys were tested at two temperatures, 550°C and 700°C. However, alloy 740H was only tested at 700°C because it is not expected to corrode significantly at 550°C.

12 

 

Grade 91 was only tested at 550°C because it is not considered to be suitable for 700°C applications. All samples were tested at 20 MPa in a target environment of 99.999% pure CO2 (research grade). The CO2 flow rates were set to refresh the test chamber at a minimum of every two hours. Each team was given six specimens of each alloy for each temperature. Samples were exposed for a total of 1500 h in 500 h increments. After each increment, all samples were removed for mass change measurements, at least one sample from each alloy was kept for additional characterization and the remainder were returned to an autoclave for additional exposure. Table 3 summarizes the exposure test matrix.

Table 2. Alloy compositions (wt.%) Alloy Fe Cr Ni Co Al Mn Mo Nb Cu Ti Si V W Gr 91 89.27 8.23 0.13 0.018 0.010 0.45 0.93 0.063 0.091 0.003 0.279 0.196 0.141 316L 68.29 16.84 9.93 0.214

13 

 

microscopy (TEM), Auger electron spectroscopy (AES), x-ray photoelectron spectroscopy (XPS), were performed as needed. Table 4 summarizes the characterization test plan.

Table 4. Round robin characterization matrix

Characterization As Received 500h 1000h 1500h Weight X x X x

SEM-EDS x x X x XRD x As needed As needed As needed TEM As needed As needed As needed As needed

AES/XPS As needed As needed As needed As needed  

3.3.4 Results and Discussion   

3.3.4.1 As‐Received Characterization   

Characterization techniques performed on the Round Robin as-received samples included Wide Angle XRD, imaging using SE and BSE techniques and grain size analysis via images and maps acquired in SE and EBSD modes. Wide Angle X-ray diffraction results are presented in Figure 1 in staggered manner to compare all peaks better. It is apparent from this graph that all austenitic alloys, that is 740, 625, 316 and 120, show peaks corresponding with austenite peak angles and ratios. However, upon comparing exact peak locations and heights in Figure 2, it appears that IN625 has a slight deviation in location and increased intensity than the other austenitic alloys, but better aligned with the standard peaks for austenite. This is likely due to the increased Ni levels, maximizing the austenite stability and approximating the lattice parameter for pure austenite, i.e. pure Ni for example. The ferritic grade 91 shows good agreement with the standard peaks for α-ferrite, as expected for its mainly ferritic matrix.

14 

 

Figure 1. Wide angle X-ray spectra staggered for all five Round Robin test alloys in as-received

condition, step size 0.1° and dwell time of 0.1s  

Figure 2. Wide angle X-ray spectra for all five Round Robin test alloys in as-received condition, superimposed with peak markers for γ-austenite and α-ferrite, step size 0.1° and dwell time 0.1s

Since the change in microstructure of the base metal and its local chemistry are of primary concern for subsequent exposure tests, imaging of the cross sections of the as-received samples was performed to establish a reference point, seen in Figure 3. Note that, alloy Gr 91 has a much finer microstructure and was thus imaged at a higher magnification in order to adequately showcase the grain structure. Additionally, in Figure 4 X-ray spectra from the same bulk regions for all alloys

15 

 

are presented. From initial microstructural characterization it can be concluded, that the chemistries of the alloys are in good agreement with the literature values for these alloys.

Figure 3. SEM images of bulk microstructures of all five alloys (Gr 91, 316L, HR 120, 625 and

740H) in the as-received condition

Figure 4. X-ray spectra collected on bulk microstructure surface seen in Figure 3 for all five

Round Robin test alloys in as-received condition

Further characterization for the as-received test alloys involved grain size analysis, for which electron back scatter diffraction (EBSD) was chosen to allow for large area scans and accurate determination of grain boundary types and registration. For convenience, all maps are displayed along the x-axis as inverse pole figure (IPF) images in Figure 5, with magnifications suitable for meaningful grain size analysis in terms of statistical number of grains in the field of view. All maps were acquired at with a Zeiss FE-SEM at an accelerating voltage of 15 kV at a 120 μm

16 

 

aperture and an Oxford Nordlys Nano EBSD detector. The acquisition results are shown in Table 5, with the normalized hit rate for the assumed body-centered cubic and face-centered cubic solutions. The hit rate value indicates where successful indexing was performed. Here, MAD means the mean angle deviation between fit and solution and the grain size numbers. It is worth noting that the large standard deviations for the grain areas are due to stark size difference between twins, for all alloys except Gr 91. In the case of Gr 91, the large standard deviations are due to a generally large range of grain sizes, which is also expected base on its as-received heat treatment.

 

Figure 5. EBSD IPF-X maps for alloy Gr 91, 316L, HR-120, 625 and 740H

 

17 

 

Table 5. Acquisition results for EBSD maps on bulk microstructures of all five as-received round robin test alloys, with hit rate, mean area deviation (MAD), ASTM grain size number, equivalent

circle diameter (ECD) and standard deviation (SD) for grain areas and ECD

Samples Norm. hit rate [%]

Mean MAD

ASTM no.

Mean ECD [μm]

SD ECD [μm]

Mean area [μm2]

SD area [μm2]

625 100 0.47 10.3 9.56 5.58 96.23 120.95

316L 100 0.52 9.9 11.41 6.47 135.08 160.72

HR-120 100 0.55 5.9 43.63 27.73 2094 2499

740H 99.44 0.51 4.8 62.93 42.43 4516 5711

Gr 91 99.88 0.44 14.8 1.98 1.3 4.41 7.47

3.3.4.2 Post Exposure Characterization    All round robin test alloys were exposed to sCO2 at 550°C and/or 700°C in 500 h increments. After each exposure, the samples were weighed and compared to their original weight. Figure 6 and Figure 7 summarize the mass change for each alloy with exposure by institution. Note that, some round robin exposure tests are still underway.

18 

 

Figure 6. Mass change for all exposures to date at 550°C (left) and 700°C (right) for alloys G91, 740H and 316L

19 

 

Figure 7. Mass change for all exposures to date at 550°C (left) and 700°C (right) for alloys

HR120 and 625

Generally, mass change measurements across institution are in good agreement under conditions with lower amounts of mass change. As the amount of mass change increases, with higher temperatures or for less corrosion resistant alloy G91, differences between organizations are more noticeable. However, there is no consistent trend in ranking of organizations at 550°C, which suggests the differences in mass change are not systematic. At 700°C, ORNL is often the highest and NETL is the lowest in mass change measurements. In most cases, the slope of the mass change is similar across institutions even if the absolute value is not. The 1500 h average mass change for ORNL was not included in Figure 6. ORNL experienced spallation of alloy 316L after the 1500 h exposure. This indicates a similarity in corrosion rate predictions across institutions. Figure 8 and Figure 9 replot the mass change data, assuming a parabolic growth rate. The parabolic rate constant (Kp) is calculated as half the square of the slope of the line fit to each institution’s data. Table 6 summarizes the parabolic rate constants for each institution and exposure condition from the slopes calculated in Figure 8 and Figure 9.

20 

 

Table 6. Parabolic rate constant Kp(mg2/cm4-s) for all conditions and institutions to date 550°C 700°C Institution G91 316L HR 120 625 316L HR 120 625 740H

KAIST 3.13E‐06 3.92E‐10 4.81E‐12 2.96E‐11 7.20E‐09 5.12E‐10 1.20E‐09 3.28E‐09 NETL 8.45E‐07 4.50E‐10 1.51E‐11 3.78E‐11 6.05E‐09 5.45E‐10 8.82E‐10 3.44E‐09 ORNL -  -  -  -  5.45E‐06 5.78E‐10 1.25E‐09 8.82E‐10

UW 1.28E‐06 1.45E‐10 1.45E‐12 1.30E‐11 8.45E‐07 2.00E‐10 5.45E‐10 8.82E‐10 OSU 7.20E‐07 4.21E‐10 8.45E‐11 3.96E‐11 - - - ‐

Average 1.49E‐06  3.52E‐10 2.65E‐11 3.00E‐11 1.58E‐06 4.59E‐10 9.69E‐10 2.12E‐09 Std. Dev. 1.11E‐06 1.40E‐10 3.91E‐11 1.21E‐11 2.61E‐06 1.75E‐10 3.27E‐10 1.43E‐09

Figure 8. Parabolic rate fit at 550°C (left) and 700°C (right) for alloys G91, 740H and 316L

 

21 

 

Figure 9. Parabolic rate fit at 550°C (left) and 700°C (right) for alloys HR 120 and 625

The largest differences in mass gain and parabolic rate constants are observed for alloy Gr 91 at 550°C and alloy 316L at 700°C. As expected, Gr 91 has the highest mass gain and corrosion rate of all the alloys tested, which allows for more variability between groups to be revealed. Mass changes and corrosion rates for alloy 316L at 550°C are in relatively good agreement among the institutions reporting, however, significant differences emerge at 700°C. The ORNL and UW data indicates breakaway corrosion but the NETL and KAIST data suggests parabolic growth. Since 700°C is at the upper end of 316L application range, the alloy may be more susceptible to breakaway corrosion if there is a small surface imperfections, possibly explaining the different behavior observed. Figure 10 and Figure 11 compare the round robin mass gain data to available literature data using the Larson-Miller parameter (P). Data is reported for exposures in both commercial grade (CG) and research grade (RG) CO2. Some data presented also include impurities (IM) in the CO2. Figure 8a for Gr 91, shows the round robin data is in good agreement with existing literature data, though one group of data is lower. For alloy 91, the RG or CG CO2 purity does not significantly affect the mass gain, however, O2 impurities in the CO2 result in lower mass gains at lower values of P. At higher values of P, pure and impure CO2 mass gain results start to converge. In Figure 10a for alloy 316, the round robin data is in good agreement with literature data. Trends for the role of impurities are not clear based on the available data. There does not seem to be a significant

22 

 

difference in mass gain caused by RG or CG CO2 purity, though data is limited. Figure 8c shows the results for alloy HR 120, where the literature data is more limited. The round robin data has slightly lower mass gain than the existing data set.

Figure 10. Round robin and literature data comparison for mass gain as a function of Larson-

Miller parameter for alloy a) Gr 91 b) 316L c) HR 120

In Figure 11a, the round robin data is in good agreement with literature data for alloy 625. The CG CO2 data generally has lower mass gain than the RG data. In Figure 9b, the round robin data is higher than most of the existing data but the available data on alloy 740H is limited and most

23 

 

data points are taken at lower pressure. There is no obvious effect of impurities on weight gain in 740H.

Figure 11. Round robin and literature data comparison for mass gain as a function of Larson-Miller

parameter for alloy a) 625 and b) 740H XRD for each experimental condition was conducted at both NETL and KAIST. The results from KAIST are presented at 550°C in Figure 12 and at 700°C in Figure 13. There were differences in XRD results and will be discussed.

Figure 12. Results of XRD analysis after corrosion testing in sCO2 at 550°C

24 

 

XRD results from both KAIST and NETL indicate that Gr 91 at 550°C developed initially a Fe2O3 which developed into a mix Fe2O3 and M3O4 oxide. NETL results indicate that the oxide is more likely Cr2FeO4. Both indicate that alloy 316L developed a mixture of Fe2O3 and M3O4 oxide irrespective of time interval. The metal could potentially be a mix of Fe, Mn, and/or Cr. Both indicate that HR 120 developed Cr2O3. NETL indicates some NiO detected at 500 h and chromium carbide detected at 1000 h. For 625 at 550°C, Cr2O3 was the primary oxide structure detected. NETL noted some potential M3O4 at 1500 h.

Figure 13. Results of XRD analysis after corrosion testing in sCO2 at 700°C

XRD results from KAIST and NETL indicate that at 700°C 316L developed primarily a Fe2O3 and M3O4 mix structure oxide across all times. The results from both NETL and KAIST XRD for HR 120 at 700°C shows a Cr2O3 and M3O4 mixture as the primary oxide structure. A Cr2O3 and M3O4 mix structure oxide was prevalent for 625 for all time intervals at 700°C. NETL noted some Fe3C detection at 1000 h. Both institutions detected a Cr2O3 and M3O4 mixture as the primary oxide structure for 740H for all time intervals at 700°C.

3.3.4.2.1 Microscopy  550°C: 500h and 1000h SEM images of the four alloys can be seen after 500 h exposure in Figure 14 and after 1000 h exposure in Figure 15. These images can be compared to the following sections to demonstrate oxide growth present in on the alloys. It is clear at 500 h that 316L, HR 120, and 625 have similar oxide sizes. Alloy 316L is just starting to show the beginning of nodule growth in the cross sections. Gr 91 has already developed a large, dual layer oxide. After 1000 h, HR 120 and 625 still have thin protective oxides. Gr 91 has grown, but doesn’t appear to have altered. Alloy 316L, however, has grown much larger and started to develop nodules growing into the base metal.

25 

 

Figure 14. SEM cross sections for samples exposed at 550°C for 500 h

Figure 15. SEM cross sections for samples exposed at 550°C for 1000 h

Gr 91 550°C 1500 h Cross-sectional SEM analysis was conducted on G91 exposed up to 1500 h. As shown in Figure 16, a double-layered oxide can be observed. EDS line scanning across the two layers showed the outer layer to be rich in Fe and O, while the inner layer was rich in Fe, Cr and O. Such oxide formation has been previously reported for 9Cr FMS in similar temperature ranges in steam and sCO2 environments.

a) Gr 91 b) 316L

c) HR 120 d) 625

a) Gr 91 b) 316L

c) HR 120 d) 625

26 

 

Figure 16. SEM cross sections for Gr 91 exposed at 550°C for 1500 h

316L 550°C 1500 h Characterized by a Cr-rich, thin oxide covering nearly all of the sample surface, with occasional duplex oxide nodule formation. A linescan of the nodule was performed which showed the Fe-Cr spinel (with roughly a 1:1 ratio of Fe:Cr) underneath an Fe-oxide, Figure 17.

27 

 

Figure 17. SEM cross sections for 316L exposed at 550°C for 1500 h

HR 120 550°C 1500 h Results of TEM analyses of HR 120 after corrosion in sCO2 at 550°C (20 MPa) up to 1500 h are shown in Figure 18. Results of TEM analyses of HR 120 exposed at 550°C for 1500 h. It is revealed from TEM line scanning that only the outermost oxide layer is rich in Mn and Cr, while the adjacent oxide layer is rich mainly in Cr. This would be the reason both chromia and Mn/Cr-rich spinel oxides were detected by XRD analysis. Meanwhile, enrichment of C and Si can be observed at the oxide/matrix interface, which have been previously reported by our team for other alloys exposed to sCO2 environments. This region was found to be amorphous in nature by high resolution TEM (HRTEM) as also reported previously. In addition, some Cr-rich regions were found in the underlying matrix, which were also found to be somewhat enriched in C suggesting them to be Cr-rich carbides. These regions were found only near the surface, which indicate their formation to be related to corrosion in sCO2. However, this may require further investigation by

28 

 

comparing corrosion behavior of HR 120 and its underlying matrix after exposure in other high temperature environments.

Figure 18. Results of TEM analyses of HR 120 exposed at 550°C for 1500 h

625 550°C 1500 h TEM analysis, Figure 19. Results of TEM analyses of 625 exposed at 550°C for 1500 h, shows only Cr and O enrichment on Alloy 625 after corrosion in sCO2 at 550 oC, which corroborates with SEM and XRD results. Meanwhile, an amorphous region rich in C could also be found for Alloy 625 at the oxide/matrix interface. Nb and Mo-rich regions were detected at the underlying matrix, as inferred from SEM analysis and would correspond to Ni3(Nb,Mo) phases identified by XRD. These phases were previously detected for Alloy 625 after exposure in high temperature steam environments. Their formation was attributed to chemical composition changes in the matrix at the surface due to Cr-depletion by oxide formation. Changes in chemical activity incited upward diffusion of Nb and Mo, which led to formation of Ni3(Nb,Mo) phases in that region. From this, it can be said that formation of such phases is not exclusive to sCO2 exposure, but rather due to oxide formation in high temperature environment.

29 

 

Figure 19. Results of TEM analyses of 625 exposed at 550°C for 1500 h

316L 700°C 1500 h Produced a duplex oxide which was largely thin and apparently protective, but with large nodules. Nodule density and thickness is significantly higher than after 550°C exposure. A linescan was performed through a nodule, showing the duplex structure, Figure 20. The inner spinel contained Cr-content equivalent to the alloying content, with a Cr-to-Fe ratio of roughly 1:1. Oxygen content did not vary over the two phases of oxide, indicating that the same M3O4 ratio was formed in both cases. The inner Fe-Cr spinel appeared to grow inward, while the outer Fe-oxide grew outward.

30 

 

Figure 20. SEM cross sections for 316L exposed at 700°C for 1500 h

For 316L in Figure 21, as the nodular Fe-rich oxides were too thick, TEM specimen was extracted from oxide of thinner thickness. The outermost oxide on SS 316L was found to be rich in Fe and Mn. Outer chromia was formed below the Fe/Mn-rich oxide, while chromia was also formed as inner oxide. Si-rich oxide was formed at the outer/inner chromia interface. Amorphous C regions were found for corroded SS 316L as well, along with Si-rich oxides.

31 

 

Figure 21. Results of TEM analyses of 316L exposed at 700°C for 1500 h

 

HR 120 700°C 1500 h From TEM analyses, Figure 22, platelet-like oxides and outermost oxide layer formed on HR 120 were found to be rich in Mn and Cr, while chromia was formed under those oxides. Internal oxidation due to Al and Si can be found, possibly due to the higher exposure temperature. While not shown in the results below, amorphous C layer could be observed by HRTEM at the oxide/matrix interface. It should be noted that Cr-rich regions suggested to be Cr carbides found after 550°C exposure could not be observed in this case. The cause is unclear and requires further investigation.

Figure 22. Results of TEM analyses of HR 120 exposed at 700°C for 1500 h

625 700°C 1500 h TEM analysis for Alloy 625 exposed to sCO2 at 700°C (20 MPa) for 1500 h showed the outermost oxide layer to be rich in Mn and Cr, which is in accordance to XRD results, Figure 23. It seems that the higher temperature of 700°C resulted in outward diffusion of Mn, resulting in outermost Mn/Cr-rich oxide layer, as such layer was not observed for the specimen exposed at 550°C. Under the Mn/Cr-rich oxide layer, chromia was formed mainly as outer oxide, while Si-rich oxides were found with amorphous C layer at the oxide/matrix interface. Some internal oxidation due to Al could also be observed. Meanwhile, Ni3(Nb,Mo) phases were formed in the underlying matrix. Compared to the specimen exposed to 550°C, the size of these phases were greater, as the oxide thickness and diffusion are greater at the 700°C.

32 

 

Figure 23. Results of TEM analyses of 625 exposed at 700°C for 1500 h

740H 700°C 1500 h 740H showed a similar oxide structure as the other alloys, as suggested by XRD analysis, Figure 24. An outermost Mn/Cr-rich oxide layer with outer chromia could be observed. Si-rich oxides were found at the oxide/matrix interface along with amorphous C regions. The difference with other alloys is the significant amount of internal oxidation due to Ti and Al. Some Ti was also detected at the outermost oxide layer. Such behavior has been reported for alloys containing Ti and Al contents of about 1–3 wt.% when exposed to oxidizing environments above 700°C, which is the case also for 740H in this study.

Figure 24. Results of TEM analyses of 740H exposed at 700°C for 1500 h

3.3.1 Summary 

In summary, the sCO2 round robin test program has been conducted to compare corrosion results for various autoclaves around the world. Mass change data available to date has shown reasonable agreement among teams and with existing literature data. There are some discrepancies between data sets on alloys with higher corrosion rates (Gr 91 at 550°C and 316L at 700°C). The higher mass change shown by the UW group for 316L is consistent with data generated under similar conditions from Saari in 2014 [7] shown in Figure 10b. Additional data from more round robin teams and future oxide analysis will help to clarify sources of discrepancies. XRD revealed that the type of oxide formed in these environments was dominated by either a mix Fe2O3 and M3O4 or a mix Cr2O3 and M3O4. The Cr2O3 and M3O4 oxide formers at 550°C (HR120 and 625) did not develop different oxides at 700°C. It is interesting to note that 316L formed a mix Fe2O3 and M3O4 oxide at 550°C but formed a mix Cr2O3 and M3O4 oxide at 700°C.

33 

 

Generally, mass change measurements across institution are in good agreement under conditions with lower amounts of mass change. The parabolic kp values further demonstrate that the intuitions had relatively similar experimental results. At 550°C, there is no clear indication on institutional differences and therefore no systemic effects. For the 700°C experiments, ORNL experienced the largest average mass gains while NETL usually experienced the lowest. The two conditions with the most mass gain were not in total agreement: Gr. 91 at 550°C and 316L at 700°C. In both conditions, the alloys were exposed at the upper end of their respective service conditions. It may be that differences in mass gain and kp values could be attributed to the early stages of spallation and breakaway corrosion. SEM analysis of Gr 91 shows a thick dual layer oxide of Fe-oxide on top of a Fe/Cr-oxide. SEM analysis of 316L displays a thin Cr-oxide interspersed with thick nodules of Fe/Cr-oxide. TEM analysis of HR 120, 625, and 740H reveals that the alloys developed thin Cr-oxides. The alloys that formed primarily Cr-oxides are shown to have significantly improved corrosion resistance over the alloys that formed Fe- and Fe/Cr-oxides.

3.4 JOINT TESTING IN SCO2 

3.4.1 Oxidation behavior of P91‐347H welded structures    Initial results suggest that the oxidation behavior can vary considerably between the bulk material and welded regions during exposure to sCO2 environments. For example, Figure 25 shows cross-sectional SEM images of the P91 half of a P91-347H weld.    

34 

 

Figure 25. Cross-sectional SEM analysis of the P91 side of a P91-347H weld after exposure to

sCO2 at 550 °C and 200 bar for 1000 h. (a) low magnification image of the interface of P91 and the HAZ (b) low magnification image of the interface of the HAZ and fusion zone (c-e) higher

magnification images showing the oxide formed in each zone.

The P91 portion of the weld formed a thick (≈50 μm) Fe-rich oxide scale, consistent with what has been observed previously for P91 exposed to high temperature CO2. Alternatively, both the heat affected zone (HAZ) and the fusion zone formed a much thinner (

35 

 

direct cycles, will also lower the environmental impact. Furthermore, compact turbo machinery and simple configurations of the sCO2 cycles could result in lower capital cost. In this communication, two aspects of the oxidation behavior of alloys were compared between several indirect- and direct-cycle related environments. The first was to examine the critical Cr content needed in Ni alloys to achieve a compact and protective chromia scale. A series of Ni-xCr model alloys were made and used for these tests. The second was to examine the effect of surface finish on the oxidation behavior of Grade 91 ferritic-martensitic steel.

3.5.1 Experimental Procedure  A series of model alloys were produced using commercial-type techniques to obtain Ni-xCr alloys where the microstructures and impurity levels were similar to commercial alloys. First, a binary master alloy of NiCr was made from commercial purity feed stocks using vacuum induction melting (VIM) followed by electro-slag reduction (ESR) to reduce O and S levels [10, 11]. The ESR slag was 40% CaF2 -30% CaO - 30% Al2O3 (wt%), and the master alloy ingot was approximately 80 kg. Final alloys were made with VIM combining the master alloys with Ni (or sometimes hot top material from a previous melt). Computationally optimized homogenization heat treatments [12] were done, then the ingots were machined to remove surface defects. Wrought processing involved wrapping the ingots in stainless steel foil and preheating for 3 h prior to upset forging and forged flattening. Each ingot was step forged and squared in multiple operations to 3.2 cm, then hot rolled to approximately 0.25 cm. The Grade 91 alloy was machined from seamless tubing (5.7 cm outside diameter), manufactured by Vallourec & Mannesmann Tubes, which had undergone normalizing at 1060 °C for 20 min and tempered at 780 °C for 60 min. Compositions of each alloy are given in Table 7. Table 7. Alloy compositions. Grade 91 values are from the manufacturer. Model alloy values are

from wavelength dispersive x-ray fluorescence spectroscopy using a Rigaku ZSX Primus II, except for the interstitial elements that are from combustion (C and S using a Leco CS744) and

inert gas fusion (O and N using a Leco ON736) methods.

Alloy Ni wt% Cr wt%

Al wt%

Si wt%

Mn wt%

Co wt%

Ti wt%

Fe wt%

Cu wt%

C ppm

N ppm

O ppm

S ppm

91 0.14 8.46 0.01 .036 0.44 Bal 0.16 1000 545 20 Ni5Cr Bal 5.00 0.02

36 

 

alloy coupons were surface finished first by lapping with 17.5 µm alumina, then with a final finish using 600 grit SiC paper. The coupons were dimensioned, ultrasonicated in isopropyl alcohol, and weighed. The Grade 91 coupons were surface finished three ways. One was a “standard” (or 91S) preparation and was prepared as the model nickel alloys. Another was a preparation that retained more “cold work” (or 91CW). In these samples, no lapping or 600 grit paper steps were used—for as-machined surfaces. The last surface preparation retained the least cold work and was done by lapping followed by a single 2-hour polishing step using “colloidal silica” (or 91CS) until a mirror finish was achieved. The alloys were exposed in six environments: sCO2, atmospheric pressure CO2 (aCO2) [13], two direct-cycle type environments DF4 and DF4S, supercritical H2O (sH2O) [13, 14], and laboratory air [15]. The sCO2 exposures were in a flowing CO2 autoclave constructed from Haynes 230. The autoclave was purged 10 times with Ar by cycles of 1 to 7 to 1 bar, and then heated to 700 °C. Research grade CO2 was then introduced at ~10 g/min. Once the target pressure was reached, the CO2 flow was reduced to 2 g/min. The DF4 and DF4S exposures were done in horizontal tube furnaces. Mass flow meters were used to introduce CO2, O2 and SO2 (mixed prior to entry into the furnace); a syringe pump was used to drip water into a pan within the furnace to flash to steam. Test durations were nominally 500 h, after which the samples were weighed and put back into a subsequent test. Table 8 summarizes the test environments.

Table 8. Test types and environments. Test Gas composition Gas Notes Alloys T,

°C P,

bar Flow rate at T/P, cm/min

sCO2 CO2 99.999% CO2 Ni-xCr 91

700 550

200 200

0.8 0.7

aCO2 CO2 99.999% CO2 Ni-xCr 91

700 550

1 1

25 25

DF4 CO2+4%H2O+1%O2 Deionized H2O Ni-xCr 750 1 25

DF4S CO2+4%H2O+1%O2+0.1%SO2

Deionized H2O Ni-xCr 750 1 25

sH2O H2O Deaerated deionized H2O

Ni-xCr 700 200 2

Air Laboratory air Some H2O present from relative humidity

Ni-xCr 91

700 550

1 1

0 0

The exposed coupons were weighed using a balance capable of measuring 10-5 grams to determine the mass change. A scanning electron microscope (FEI Inspect) with energy dispersive spectroscopy (EDS) capabilities was used to obtain images and microchemical data on the exposed surfaces and cross-sections.

37 

 

3.5.2 Results and Discussion 

3.5.2.1 Nickel‐Chromium Model Alloys  Mass loss changes with time are shown in Figure 26. Mass change results as a function of time for Ni-xCr model alloys in six environments. The bottom row repeats the middle row, but without the heavily oxidized Ni-5Cr data. for the Ni-xCr model alloys. In sCO2, all of the mass changes were small, but those for Ni-5Cr were markedly higher than for the higher Cr alloys. In aCO2, the mass changes were higher than in sCO2, but still low. In contrast with sCO2, Ni-5Cr had the smallest mass change in aCO2. This was the only environment where this occurred. There was no pattern, with respect to Cr levels, in the mass change data for 12-24 Cr alloys in the pure CO2 environments. In sH2O there was a distinct trend with Cr, with mass change in 5Cr > 12Cr > 14Cr > 16-24Cr. The high mass gain for Ni-5Cr in sH2O clearly indicates that a protective chromia layer was not established and maintained. The mass gains for the DF4 and DF4S environments were similar. In both cases Ni-5Cr had high mass gains and the remaining alloys had quite low mass gains. The mass gains in DF4S were somewhat higher than in DF4—showing an effect with 0.1% SO2. The results in air (for significantly longer exposures than the other environments) also had quite high mass gains for Ni-5Cr. When Ni-5Cr is removed from the chart (lower right in Figure 26), the results show low mass gains followed by mass losses. This could be a result of spalling or chromia evaporation. The laboratory air used had some water from the relative humidity in the room, which would favor reactive evaporation of chromia as CrO2(OH)2(g) [16].

38 

 

Figure 26. Mass change results as a function of time for Ni-xCr model alloys in six environments. The bottom row repeats the middle row, but without the heavily oxidized Ni-

5Cr data.

Mass change data after 1500 h are shown in Figure 27 as a function of Cr for the six environments. There are clear differences in the critical Cr level needed to form and maintain a protective chromia scale. The Ni-5Cr alloy in the pure CO2 environments (sCO2 and aCO2) was protective, while it wasn’t in the other environments. When H2O or O2 was part of the gas phase, a transition to protective kinetics occurred somewhere between 5-12Cr in DF4, DF4S and air, and at 14Cr in sH2O. The oxygen activity in sCO2 and sH2O are similar (1.8 ×10-6 and 2.2 × 10-6 atm, respectively at 700°C and 200 bar), so it is the H2O itself that is more aggressive than CO2 for the Ni-5Cr alloy.

39 

 

Figure 27. Mass change results after 1500 h for Ni-xCr model

alloys in 6 environments.

Figure 28 shows cross section microstructures after 2000 h of exposure in aCO2 at 700 °C. The oxide scale for Ni-5Cr shows a duplex oxide scale. Elemental analysis with EDS showed that the inner oxide scale was Cr oxide (likely Cr2O3) with Ni also present, while the outer oxide scale was Ni oxide (presumably NiO). The microstructure for Ni-12Cr showed a sharp transition between an inner and outer oxides, but both were essentially pure Cr oxide (Cr2O3). The microstructures for Ni-14Cr and Ni-24Cr both showed contrast between inner and outer layers of Cr oxide (Cr2O3), but without the sharp transition seen with Ni-12Cr.

Figure 28. Ni-xCr alloys after 2000 h exposure in aCO2 at 700 °C. The images are backscattered SEM.

40 

 

The oxide scale for Ni-5Cr was thinner than the other three compositions shown in Figure 28, which corresponds to the results shown in Figure 26 for aCO2. It is likely that some amount of NiO spalled during heating, cooling, or post-test handling, which lowered the mass change curve in Figure 26 to lower values than the other (higher Cr) alloys. Figure 29 shows cross section microstructures after 2000 h of exposure in air at 700 °C. The oxide scale for Ni-5Cr is much thicker than the other compositions in Figure 29. Elemental analysis with EDS showed a clear and sharp transition between an inner scale of Cr-Ni oxide (likely a spinel) and an outer scale of Ni oxide (NiO). The cross-sections of the other three compositions in Figure 29 were much thinner and were primarily Cr oxide (Cr2O3). The cross-sections were after 2000 h, which was prior to the mass decreases seen in Figure 26 in air for Cr compositions of 12 and higher.

Figure 29. Ni-xCr alloys after 2000 h exposure in laboratory air at 700 °C. The images are backscattered SEM.

Figure 30 shows a cross-section and elemental maps of Ni-12Cr after 2000 h at 700 °C for a portion of the cross-section with an unusual feature. The feature is a pit-shaped depression, filled with Cr-

41 

 

oxide (Cr2O3), underneath an outer oxide dome of Ni-oxide (NiO). One explanation for the structure is that a chromia scale did not initially form at that location. A less protective scale of NiO formed, which resulted in the pit. A chromia scale eventually did form at the metal-scale interface. As the chromia layer grew, it pushed out the NiO scale; it is the remnant of the initial NiO scale that is now seen at the outer oxide location.

Figure 30. Ni-12Cr after 2000 h exposure in laboratory air at 700 °C showing backscattered

SEM and EDS elemental maps.

Although it should be further verified, the results of exposures of Ni-xCr alloys in sCO2 (indirect cycle) and DF4/DF4S (direct cycle) environments indicate that the high strength Ni-base superalloys (generally 19-25%Cr) may be quite resilient to recovery from foreign object damage. Under normal use, Cr is depleted under the protective chromia scale that forms. If the scale is breeched (such as from an impact), then the newly exposed alloy will have a lower Cr level than the original alloy. The ability to regenerate a new protective chromia scale down to 12%Cr in pure CO2, DF4, and DF4S, and to provide some protection down to 5% Cr pure CO2, should make these superalloys resilient. However, it remains to be seen if the other alloying elements in these superalloys interfere with this low Cr protectiveness.

42 

 

3.5.2.2 Surface Finish Effects in Grade 91steel  It is well-known that shot peening, and similar surface preparations, can greatly reduce scaling in ferritic steel tubing and piping in steam boilers. Shot peening adds cold work to the near surface of the alloy, which can increase the diffusion of Cr to the surface during oxidation, decreasing the rate of oxidation. In this study, cold work effects were examined for grade 91 steel exposed to aCO2 and sCO2. For comparison, the alloy was also exposed to air. In this study, machining added cold work to the surface. Figure 31a shows the machined surface. The 91CW samples have more near-surface cold work than the other 2 conditions. After machining, a lapping step was done (Figure 31b). Then either 600 grit paper (91S, Figure 31c) or a colloidal silica polish (91CS, Figure 31d) was done. The colloidal silica polish was aimed at removing the lapped damage, without adding cold work (which the 600 grit paper adds). The 91CS samples had the least cold work. Figure 31d shows that despite the samples appearing highly reflective at this stage, not all of the lapping damage was removed by the colloidal silica process—some surface voids remain.

Figure 31. Backscattered SEM surface images of alloy 91 with four different surface finishes.

(a), (c), and (d) show the surfaces of samples 91CW, 91S, and 91CS, respectively.

43 

 

Mass change results are shown in Figure 32 after exposures in sCO2, aCO2, and air. The mass gains in air were much lower than in the CO2 environments. There were distinct differences in the surface finish effect in the three environments. The 91CW samples had the least mass gains in each environment. In sCO2 and air, the 91S and 91CS samples were similar (in their respective environment). However, in aCO2, the 91CS samples had mass gains like 91CW. The 91S results were similar to those found by Rouillard [17] with similar surface preparation after 550 °C/1 bar and 550 °C/250 bar exposures.

Figure 32. Mass change results as a function of time for alloy 91 with different surface finishes in 3 environments.

The benefits of cold work are clear. However, in the CO2 environments, the mass changes are still high compared to those found in alloys such as 600, 690, and 800 with mass gains on the order of 0.02 mg/cm2 after 1000 h at 550 °C/100 bar [18]. These are chromia forming alloys whereas Grade 91 forms a thicker duplex scale with an inner Fe-Cr spinel layer and an outer magnetite layer [17]. Such thick scales are not desirable in heat exchanger components, but may be adequate for piping applications.

3.5.3 Summary  The oxidation responses of Ni-xCr model alloys (where x varied from 5 to 24 wt%) were compared in six high temperature environments. The Ni-5Cr alloy in pure CO2 environments (sCO2 and aCO2) was at least somewhat protective, while it was unprotective in the other environments. When H2O or O2 was part of the gas phase, a transition to protective kinetics occurred somewhere between 5-12Cr in DF4, DF4S and air, and at 14Cr in sH2O. The oxygen activity in sCO2 and sH2O were similar, so H2O was more aggressive than CO2 for the Ni-5Cr alloy. The ability to remain protective at low Cr values indicates that nickel base superalloys may be resilient to damage that exposes near-surface alloy that is depleted Cr—especially in pure CO2. The oxidation responses of ferritic steel Grade 91, with three different surface finishes, were compared in three different environments at 550 °C. The mass gains in air were much lower than in CO2 environments. The benefits of near surface cold work were observed—the samples with the most cold work had the smallest mass gains in all three environments. This indicates that surface enhancements to induce more residual stress, such as shot peening, may be of benefit for 9-12Cr ferritic-martensitic steels in sCO2.

44 

 

3.6 MODEL ALLOY SIMULATIONS  Oxidation and corrosion of nickel and Ni-based alloys are a problem for many industrial applications, such as power plants that use supercritical CO2 as the working fluid. In supercritical CO2 environments, CO2 dissociates on the surface forming adsorbed CO and O, which can oxidize the surface. The adsorbed CO can further breakdown via direct CO dissociation or via the Boudouard reaction to form adsorbed C, which can in turn carburize the surface. Understanding how the adsorbed species interact with different Ni-based alloys can help guide the design of future alloys. The interactions of adsorbed O, C, and CO on the (100) and (111) facets of pure Ni and Ni individually alloyed with Al, Co, Cr, Cu, Fe, Mn, Mo, Nb, Ti, V, and W are examined using density functional theory. We find that the binding of CO is energetically similar across all alloy surfaces and both facets, while O binding varies strongly with different metals added to nickel and C binding varies between the different facets but only slights for different metals alloyed to nickel. The binding of O is weaker on pure Ni and Ni alloyed with Cu, Co, Fe, Al, or Mn and stronger on Ni alloyed with Nb, Cr, Mo, Ti, V, or W, while the binding of C is weaker on the (111) facet than the (100) facet. The difference in the binding energies of the adsorbates across the different alloy surfaces is due mainly to the ensemble effect, rather than the ligand effect. The thermodynamics of CO breakdown are also studied and we find that the breakdown of CO via direct CO dissociation is endothermic on the (111) facet and exothermic on the (100) facet, with the alloy surfaces that bind O strongly having the most exothermic reaction energies. The breakdown of CO via the Boudouard reaction has similar reaction energies across the different alloy surfaces of a single facet and is endothermic on both facets, with the (111) facet being most endothermic. This comprehensive study presents a summary of the current literature as well as a well-rounded view of the products of CO2 breakdown on Ni surfaces alloyed with the most common alloying elements used in industrial applications. Figure 33 summarizes the key findings of the computational study described in more details below.

Figure 33. Summary of the key finding of the nickel alloy modeling findings and the factors that control the surface interactions of the different adsorbates. C binding is structure sensitive and is more strongly bond on more open, (100) surface structure, the alloying elements has little effect

on C binding. O interactions are dominated by the type of alloying element and show little structural effect. CO interactions are similar on both surface faces and all alloying surfaces.

45 

 

3.6.1 Introduction  Nickel-based alloys are commonly used in industrial applications as structural materials and as heterogeneous catalysts. In catalysis, Ni-based alloys are common in applications such as the oxidation of hydrocarbon fuels in solid-oxide fuel cells (SOFC) [19-21] and in the catalytic dry reforming of CH4 with CO2 to make syngas [22, 23]. Ni-based alloys are also widely used as structural materials in industrial applications such as heat exchangers, pipework, combustion cans, and engine blades due to their corrosion-resistance, high-strength, and high-temperature properties [24, 25]. An emerging application for Ni-based alloys is in supercritical carbon dioxide (sCO2) power production cycles. Compared to conventional steam power production cycles, sCO2 cycles offer an improved plant efficiency due to less compressive work because of higher CO2 densities. Other advantages of sCO2 cycles include lower cost, reduced emissions, fewer and smaller energy conversion components, and a simpler cycle layout [26, 27]. A major limitation to implementing this step-changing technology is the identification and development of high-strength, corrosion-resistant materials for high temperature (650-800°C) power plant components. Herein, we use density functional theory to study the surface chemistry of a dozen Ni-based alloys that are promising materials for such power plants, and discover trends in the thermodynamic stabilities of the surface species, which give insights into how to improve the long-term stability of these materials. There are many commercially available Ni-based alloys and superalloys that are currently being considered for sCO2 applications such as 740H, 282, 230, 625, 214, 224, and C276 [27-32]. In these Ni-based alloys, the highest concentration alloying elements are Cr, Fe, Co, Mo, W, Al, Nb, Ti, Mn, Si, Cu, C, and V, which are taken as the starting point for the Ni-based alloy surfaces studied herein. All metallic elements, which excludes Si and C, are included in this study giving twelve different alloy surfaces. The initial step in the degradation of Ni-based alloys in sCO2 conditions is expected to be through the dissociation of CO2 to make adsorbed CO plus O (COads + Oads). Computational results [33] have found CO2 to dissociate on Ni(100), while experimental results [34] have observed CO2 to dissociate on Ni(100), but not on Ni(111). However, we find CO2 dissociation to be quite exothermic on both facets with a reaction energy of -1.38 eV on Ni(100) and -1.07 eV on Ni(111), such that CO and/or O could diffuse to the (111) facet after dissociation on the (100) facet. This is followed by carbon deposition from COads dissociation into Cads + Oads or from the Boudouard reaction where two COads react to form Cads + CO2,gas. Deposition of O on the surface leads to oxidation of the surface and the formation of metal oxides while deposition of C on the surface leads to carburization of the surfaces and the formation of metal carbides, which change the chemistry of the surface and eventually the properties of the bulk metal. We show that alloying Ni with other elements affects the CO breakdown reactions, just as it has been shown to affect other reactions such as water gas shift [35] and methane reforming [36]. Herein we study the adsorbed products of CO2 dissociation (Oads, Cads, and COads) and uncover how their stabilities depend on the nature of the alloying element used to make a Ni-based alloy. We also study the thermodynamics of the CO dissociation and the Boudouard reactions to understand the most favorable CO breakdown mechanism. This gives insight into the thermodynamic driving force for the different reaction paths on 12 different surfaces and 2

46 

 

different surface facets. Previous studies have shown that the Brønsted-Evans-Polanyi relations can be used to relate activation energy barrier heights to the dissociative adsorption energy for a given family of reactions [37-39]. Using these relationships, the results presented herein can be used to predict kinetic barriers, although these are not explicitly calculated herein. The adsorption of O [40-43] and CO [43-48] has been studied extensively on the low-index facets of pure Ni surfaces. More recently, studies have included Ni-based alloys looking at the adsorption of O, C, and CO on the (111) facet [19-25, 49, 50], however the (100) facet [51, 52] is less studied. In order to develop new Ni-based alloys, it is essential to understand the fundamental interactions of the corrosive species with the surface and the thermodynamics of the reactions that lead to the corrosive species. Herein we focus on the Ni(100) and Ni(111) facets with Al, Co, Cr, Cu, Fe, Mn, Mo, Nb, Ti, V, or W alloying elements and examine their interactions with adsorbed O, C, and CO. To better understand the direct interaction between the adsorbate and the alloying element, a low concentration alloy is modeled by replacing a single Ni surface atom with the alloying element, leading to a mole fraction of 1/9 in the surface layer but only 1/63 and 1/54 in the (100) and (111) simulation slabs, respectively. Binding energies and binding sites for all adsorbates, ligand and ensemble effects of the alloying atom, and CO dissociation and Boudouard reaction energies are determined on all facets and surfaces.

3.6.2 Computational Details  All results are calculated using density functional theory (DFT) via the Vienna Ab-initio Simulation Package (VASP) [53-56], with some of the calculations run on the Extreme Science and Engineering Discovery Environment (XSEDE) [57]. Electronic structures are calculated for the adsorption of one C, O, or CO adsorbate on the (100) and (111) facets of Ni-based alloys. A p(3x3) unit cell is used to represent the 12 Å thick surface slabs, containing 7 layers for the (100) facet and 6 layers for the (111) facet, with 20 Å of vacuum between slabs. The top 4 metal layers and the adsorbate are allowed to relax while the bottom 3 layers for the (100) facet and 2 layers for the (111) facet are held fixed. Atomic Simulation Environment (ASE) was used to create the initial simulation cells [58]. The exchange correlation potential and energy is described by the generalized gradient approximation (GGA) as defined by the Perdew-Burke-Ernzerhof (PBE) functional [59, 60], and the projector augmented wave (PAW) method is used to represent the core electrons [61, 62]. Spin polarization and magnetization effects are included. Plane-wave calculations are employed with a kinetic energy cutoff of 400 eV for the Kohn-Sham orbitals and the surface Brillouin zone is sampled using a Monkhorst-Pack grid with 5x5x1 k points. Increasing the energy cutoff to 500 eV changes the binding energies of O, C, and CO on Ni(100) by 3 meV, 10 meV, and 3 meV and on Ni(111) by 8 meV, 8 meV, and 1 meV, respectively. Increasing the number of k points to 7x7x1 changes the O, C, and CO binding energies on Ni(100) by 0.04 eV, 0.02 eV, and 0.002 eV and on Ni(111) by 0.03 eV, 0.02 eV, and 0.01 eV, respectively. All calculations for binding energies are relaxed until the forces are converged below 0.01 eV/Å. The calculated lattice constant for Ni is 3.52 Å, in agreement with the experimental value of 3.52 Å [63]. Density of states calculations are performed using 15x15x1 k points to sample the Brillouin zone. Increasing to 19x19x1 k points changes the d-band center of the pure Ni surface by 0.04 eV for

47 

 

the (100) facet and 0.03 eV for the (111) facet. The electron density is converged until the energy is below 10-6 eV for the clean surfaces. Only the top one surface layer is considered when calculating the d-band center of each Ni-based alloy surface. Bader charge calculations [64-67] are also performed on the clean alloy surfaces with 5x5x1 k points and an energy cutoff of 500 eV.

3.6.3 Surface Models  The 3x3 surfaces bind one adsorbed species, corresponding to a 1/9 monolayer adsorbate coverage. For the alloy surfaces, one of the Ni atoms in the top layer is exchanged for the alloying metal atom (Ti, V, Nb, Cr, Mo, W, Mn, Fe, Co, Cu, or Al), setting the alloy surface concentration at 1/9. Figure 34 shows the different adsorption sites on the alloyed (100) and (111) facets. There are three unique hollow sites, three unique top sites, and four unique bridge sites on the (100) facet and three unique hcp hollow sites, three unique fcc hollow sites, three unique top sites, and four unique bridge sites on the (111) facet. Herein we use the site name followed by a number to indicate distance from the alloying atom. In this notation, top-1 refers to a top site on the alloying atom and top-3 refers to a top site farthest away from the alloying atom. The different sites, illustrated with different shapes, and different numbers, increasing moving away from the alloying atom, in Figure 34 represent the unique binding locations. Alloy-rich sites have the adsorbate adjacent to the alloying atom and include the top-1, hollow-1, hcp-1, fcc-1, and bridge-1 sites. Ni-rich sites correspond to sites in which the adsorbate is bound to only Ni atoms and include all remaining sites.

Figure 34. Unique binding sites on the (100) and (111) facets. Light blue circles represent top sites, medium blue rectangles represent bridge sites, dark blue squares represent (100) hollow sites, dark blue up-pointing triangles represent (111) hcp hollow sites, and dark blue down-

pointing triangles represent (111) fcc hollow sites. The numbers increase moving away from the alloying atom and label the unique binding locations for each site.

Binding energies of O, C, and CO at each unique site are calculated to determine the most stable binding location for each adsorbate on every alloy surface. The binding energy is defined as:

48 

 

Equation 1 Ebind = Esurf+ads – Esurf – Eref

where Esurf+ads is the total energy of the surface and adsorbate system, Esurf is the energy of the clean surface, and Eref is the adsorbate reference energy. The reference energy is the energy of 1/2 O2 in vacuum, one graphite atom, or CO in vacuum for the O, C, or CO adsorbate, respectively. With this notation, a negative Ebind value indicates a release of energy upon adsorption and a positive Ebind value represents an increase in energy upon adsorption. Binding energy differences are also reported for adsorbates on the different Ni-based alloy surfaces relative to the pure Ni surface when the adsorbate is in its most stable adsorption site on each surface. The difference in binding energy is calculated as

Equation 2 ΔEbind = Ebindalloy – EbindNi = Esurf+adsalloy – Esurfalloy – Esurf+adsNi + EsurfNi

where the superscripts indicate the alloy surface or the pure Ni surface. In this notation, a negative ΔEbind value indicates that the adsorbate binds stronger to the alloy surface than the pure Ni surface and a positive ΔEbind value signifies that the adsorbate binds weaker to the alloy surface than the pure Ni surface. The advantage of calculating the binding energy difference is that it is independent of the choice of reference species, which is useful when comparing to other literature values. The binding energies of the reactants and the products in the CO dissociation reaction (CO → CO) and the Boudouard reaction (2CO → C CO ) are calculated relative to CO in the gas phase. The binding energy of a single CO reactant follows Equation 1 and is calculated as

Equation 3 Ebind,CO = Esurf+CO – Esurf – ECO

where ECO is the energy of CO in vacuum. The binding energy of the products of CO dissociation, C+O, is calculated as

Equation 4 Ebind,C+O = Esurf+C + Esurf+O – 2Esurf – ECO

making the overall CO dissociation reaction energy

Equation 5 Erxn,COdiss = Ebind,C+O – Ebind,CO = Esurf+C + Esurf+O – Esurf+CO – Esurf.

The binding energy of the Boudouard reaction products, C+CO2, is calculated as:

Equation 6 Ebind,C+CO2 = Esurf+C + ECO2 – Esurf – 2ECO

49 

 

where ECO2 is the energy of CO2 in vacuum, making the overall Boudouard reaction energy

Equation 7 Erxn,Boudouard = Ebind,C+CO2 – 2Ebind,CO = Esurf+C + ECO2 + Esurf – 2Esurf+CO.

A positive reaction energy, Erxn, represents an endothermic reaction, with CO being the most energetically stable species, and a negative reaction energy represents an exothermic reaction, where C+O for CO dissociation or C+CO2 for the Boudouard reaction is energetically more favorable. Herein the adsorbates are assumed to be far apart and non-interacting. The aim of this study is to understand the relative stabilities of CO2 dissociation products on Ni-based alloy surfaces. The binding energies reported below reveal that the dissociation of CO2 gas on these alloy surfaces to produce COads + Oads is quite exothermic, with a reaction energy of -1.38 eV on Ni(100) and -1.07 eV on Ni(111). This is in good agreement with a previous study that found the reaction to be exothermic by -1.00 eV on Ni(100) and -1.33 eV on Ni(111) [68]. Herein we focus on the relative energies of the CO2 dissociation products (COads and Oads) and further CO reaction products (Cads) which can be produced by either CO dissociation, CO → C O, or the Boudouard reaction, 2CO → C CO , and analyze how these energies depend on the nature of the alloying element. The binding energies of O, C, and CO are presented in sections 3.6.4, 3.6.4.2, and 3.6.4.3, respectively, and include binding of the adsorbates to the (100) and (111) facets of pure Ni and eleven Ni-based alloy surfaces. Section 3.6.4.4 focuses on understanding the effect of the alloying element on the adsorbate binding energies by exploring the ligand effect and the ensemble effect. In Section 3.6.4.5 the reaction energies for CO dissociation and the Boudouard reaction are determined and key findings are highlighted and discussed.

3.6.4 Results and Discussion   

3.6.4.1 Oxygen Binding  The binding of O is most stable at a hollow site (-3.20 eV) and less stable at a bridge site (-2.50 eV) or a top site (-1.30 eV) on the Ni(100) facet. On the Ni(111) facet, O binds slightly stronger at an fcc hollow site (-2.91 eV) than an hcp hollow site (-2.80 eV) and weaker at a top site (-1.06 eV). The O binding energy at a bridge site on the (111) facet could not be determined. This is in good agreement with previous DFT studies that also found the hollow site on Ni(100) and the fcc site on Ni(111) to be the most stable binding sites for O [43, 68-70]. Experimentally, O is also found to be most stable at a four-fold hollow site on Ni(100) with a binding energy of -2.85 eV and at three-fold hollow sites on Ni(111) with a binding energy of -2.28 eV [41]. This is in good agreement with the results presented here, however non-hybrid DFT functionals, such as PBE, are well known to overbind chemisorbed systems [71, 72], so slightly stronger calculated binding energies are not surprising. Comparing the two facets, O adsorbs 0.29 eV stronger on the (100) facet than on the (111) facet, in agreement with previous DFT studies that found O adsorption is stronger on Ni(100) than Ni(111) by 0.28 eV [69], 0.31 eV [68, 70], 0.32 eV [43], or 0.44 eV [73].

50 

 

Replacing a Ni surface atom with an alloying atom changes the properties of the surface and influences the way O interacts with the surface, as shown in Table 9 by different binding energies and binding sites for adsorbed O. A complete list of O binding energies for all surfaces at different sites can be found in Table 19 and Table 20 in section 3.6.6. The most stable binding site for O adsorption is adjacent to the alloying atom at the hollow-1 site and fcc-1 site on the (100) and (111) facets, respectively, for all Ni-based alloy surfaces except when the alloying element is Cu, Mo, or W. On the Cu alloy surface, O binds stronger at a Ni-rich hollow site than the hollow site adjacent to the Cu atom by 0.13 eV on the (100) facet and by 0.34 eV on the (111) facet. On both facets of the Mo and W alloy surfaces, O binds strongest at the top-1 site right above the alloying atom. On the Mo alloy surface, O is more stable at the top-1 site by 0.26 eV and 0.15 eV than the hollow-1 site on the (100) facet and the fcc-1 site on the (111) facet, respectively, which are the second most stable sites. O is not stable at the hollow-1 site on the (100) facet of the W alloy surface so a comparison cannot be made, however on the (111) facet O binds 0.21 eV stronger to the top-1 site compared to the fcc-1 site. The energetic favorability of O at top-1 sites can be explained by the stronger oxophilic nature of Mo and W.

Table 9. Effects of substituting an alloying atom for a single Ni surface atom on oxygen adsorption. Binding energy (eV) referenced to 1/2 O2 in vacuum, binding energy difference

relative to pure Ni (eV), and binding site for O adsorption on the (100) and (111) facets of Ni-based alloys at 1/9 ML coverage. For site locations, refer to Figure 34.

(100) facet (111) facet Alloying Atom Ebind ΔEbind site Ebind ΔEbind site

Ni (no alloy) -3.20 0.00 hol -2.91 0.00 fcc Cu -3.20 0.00 hol-2 -2.94 -0.03 fcc-2 Co -3.25 -0.05 hol-1 -2.98 -0.07 fcc-1 Fe -3.31 -0.11 hol-1 -3.05 -0.14 fcc-1 Al -3.33 -0.13 hol-1 -3.12 -0.21 fcc-1 Mn -3.39 -0.19 hol-1 -3.19 -0.28 fcc-1 Nb -3.78 -0.58 hol-1 -3.67 -0.76 fcc-1 Cr -3.81 -0.61 hol-1 -3.75 -0.84 fcc-1 Ti -3.94 -0.74 hol-1 -3.80 -0.89 fcc-1 V -3.97 -0.77 hol-1 -3.89 -0.98 fcc-1

Mo -4.08 -0.88 top-1 -3.89 -0.98 top-1 W -4.27 -1.07 top-1 -4.01 -1.10 top-1

All the alloying elements strengthen the O-surface interactions on both the (100) and (111) facets, shown by the negative ΔEbind values in Table 9, except Ni(100) alloyed with Cu in which the O binding energy is not affected. The range over which an alloying element affects the binding energy of O on the surface is broad, with the largest increase being 1.07 eV and 1.10 eV for Ni alloying with W on the (100) and (111) facets, respectively. The order in which the alloying element increases the binding strength of O on the Ni-based alloy surfaces is similar on both facets, increasing with the alloying