Accelerated Measurement and Mechanism Based Simulation of Hydrogen Cracking

Transcription

1 Accelerated Measurement and Mechanism Based Simulation of Hydrogen Cracking Richard P. Gangloff Center for Electrochemical Science and Engineering Department of Materials Science and Engineering University of Virginia Charlottesville, VA Hydrogen Embrittlement: Multi scale Modeling and Measurement National Physical Laboratory Teddington, UK October, 2014

2 Sponsors Office of Naval Research DoD Corrosion Policy and Oversight Office Naval Air Development Center Questek Innovations Collaborators John R. Scully James T. Burns Sean R. Agnew J. Keith Donald Matthew R. Begley Yongwon Lee Gregor Pioszak Sami Al-Ghamdi ~ Emilio Martinez-Paneda Beth A. Kehler (University of Oviedo, Spain)

3 Hydrogen Environment Assisted Cracking Water vapor H 2 Aqueous electrolytes (Cl - H 2 S) The challenge is to translate the (scientific) insights gained (over the past 50 years) into (quantitative) outcomes of value to the engineering community Risk exists

4 Key Points TOOLS High resolution accelerated measurement of crack growth kinetics provides quantitative input to probabilistic fracture mechanics simulation of H-cracking Damage tolerant approach to SCC-HEAC is justified

5 Precision dcpd measurement of small crack propagation provides accelerated-quantitative characterization of da/dt vs K elastic-plastic for HEAC science and simulation 100 µm day experiment at 1 MPam/h 7 replicate tests Time (hrs) AerMet TM % NaCl 3.5% Test NaCl ID Potential -550 mv AM-23 SCE -0.55V SCE Fe-13Co-11Ni-3Cr-1Mo-0.2C Crack length (in) m Crack length (m) ys = 1725 MPa K IC = 130 MPam Time (seconds)

6 SCCrack Similitude Equal da/dt for equal K (Wei, Williams, Brown; 1967) Monte Carlo selection from distributions of input variables Material da/dt vs K Applied Potential Crack Geometry Applied Loading Residual Stress Initial Defect Size

7 Output of SCCrack Method quantifies effect of surface damage distribution (corrosion, machining, or wear topography) on HEAC performance; as function of ultrahigh strength steel composition and coating 300 mmeandepth 0.18 CoV NaCl 300M OCP ( 700 mv SCE ) AerMet TM mv SCE Ferrium TM M mv SCE 700 mv SCE AerMet TM mv SCE

8 Output of SCCrack Method quantifies effect of surface damage distribution (corrosion, machining, or wear topography) on HEAC performance; as function of ultrahigh strength steel composition and coating 300 m mean depth 0.18 CoV 300M OCP ( 700 mv SCE ) H-cracking control based on propagation is justified by long lives for modern alloys in inhibited environments AerMet TM mv SCE Low Co steel 550 mv SCE 700 mv SCE AerMet TM mv SCE

9 Key Points Tools High resolution-accelerated measurement of crack growth kinetics provides quantitative input to probabilistic fracture mechanics simulation of H- cracking Damage tolerant approach to SCC-HEAC is justified APPROXIMATIONS enable quantitative transition from the decohesion mechanism of H damage, through crack tip modeling of alloy-h cracking kinetics, to component life simulation MODEL PREDICT Fe-13Co-11Ni-3Cr-1Mo-0.2C Turnbull framed the complexity of this HEAC problem K TH da/dt II

10 DECOHESION-INSPIRED MODELS of HEAC threshold stress intensity and H-diffusion limited Stage II da/dt are validated for range of alloys, using single-adjustable parameter that accounts for model uncertainty K TH 1 β' (k exp IG αc α"σ YS Hσ ) 2 da dt II 4D χ H-EFF CRIT erf 1 C Hσ C C Hσ CRIT Steel Low pressure H 2 YS = 1725 MPa High Strength Alloys NaCl solution Fastest da/dt α fit D EFF independent C H > 6C CRIT X CRIT = 900 nm 1st rise permeation D EFF ~

11 Modeling Goal Predict.. K TH based on H decohesion da/dt II based on H diffusion rate limitation Inputs Occluded crack tip H overpotential H solubility vs H for crack solution Crack tip FPZ dimension and stress distribution Trap-sensitive H diffusivity Mechanism-informed micromechanical cracking criterion

12 Key Points Tools High resolution-accelerated measurement of crack growth kinetics provides quantitative input to probabilistic fracture mechanics simulation of H- cracking Damage tolerant approach to SCC-HEAC justified Approximations enable quantitative transition from decohesion mechanism of H damage, through crack tip modeling of alloy-h cracking properties, leading to component life simulation Example Mechanism-based models predict critical level of reduced-cathodic polarization that eliminates IGSCC in Monel K-500 in neutral NaCl; Scaled to simulate component life Simplified Very low lattice-h diffusivity in Ni, with 2 nd -order trapping at nano-scale precipitates, and nil contribution from bold-surface H Gangloff, Ha, Burns and Scully; Metall. Mater. Trans. (2014) Ai, Ha, Gangloff and Scully; Acta Mater. (2013)

13 Severe IG-SCC in Monel K-500 eliminated by reduced cathodic polarization above E CRIT (-750 to -800 mv SCE )for well-controlled mechanics 4-6 day experiment 1 MPam/h 66Ni-29Cu-3Al-0.5Ti OCP ~ -250 mv SCE 0.3 MPam/h

14 Severe IGSCC in Monel K-500 eliminated by reduced cathodic polarization above ECRIT (-750 to -800 mvsce) mvsce OCP ~ -250 mvsce -700 mvsce 750 mvsce

15 High resolution SEM-FIB-TEM of Ni 201 supports HEAC due to interaction of HELP and HEDE Monel K-500 Ni FIB SEM Hydrogen Enhanced Local Plasticity Crack tip dislocation cell structure stimulated by H TEM Cell structure favors H trapping at walls Local strength elevation by LEDS H decohesion enabled Robertson, et al (2012)

16 Crack tip H solubility in Monel K-500 K Initial = 20 MPam x 2 /G = 20 cm K Final = 80 MPam x 2 /G = 12 cm bold surface crack tip C H-Diff (wppm) = E applied (mv SCE ) C H-Diff = 0 wppm at E applied > -764 mv SCE Scully et al.

17 Experimental calibration ( ) of decohesion model prediction of cathodic polarization dependence of IG HEAC in Monel K Threshold Stress Intensity MPam Monel K % NaCl TDS H Uptake (TDS ATI Allvac) = 0.56 MPam (at frac H) 1 k IG = MPam ATI Allvac Model Predicted Low alpha (0.37) High alpha (0.83) K TH 1 β' (k exp IG αc α"σ YS Hσ C H,diff 0 at -764 mv SCE ) Applied Potential (mv SCE ) α MPa m β' 0.26 (MPa m) -1 (from fit to IN903, IN718, Fe-Si, UHSS) σ H 9.3 GPa (12σ YS ) k IG MPam (no IG HEAC at K of 100 MPam) Crack tip H = 12 YS = 0.05E = 0.56 MPam (at frac H) -1

18 Experimental calibration ( ) of decohesion model prediction of cathodic polarization dependence of IG HEAC in Monel K-500 Threshold Stress Intensity MPam Monel K % NaCl TDS H Uptake (TDS ATI Allvac) = 0.56 MPam (at frac H) 1 k IG = MPam ATI Allvac Special Metals Model Predicted Model Predicted Low alpha (0.37) Low alpha (0.37) High alpha (0.83) K TH 1 β' (k exp IG αc α"σ YS Hσ C H,diff 0 at -764 mv SCE ) Applied Potential (mv SCE ) α MPa m β' 0.26 (MPa m) -1 (from fit to IN903, IN718, Fe-Si, UHSS) σ H 9.3 GPa (12σ YS ) k IG MPam (no IG HEAC at K of 100 MPam) Crack tip H = 12 YS = 0.05E = 0.56 MPam (at frac H) -1

19 Predict crack tip H diffusion-rate limited IG HEAC growth in Monel K-500 One adjustable parameter C C = 1100 wppm da dt II 4D X C H erf 1 1 C H, Diff CC σhv exp RT H 2 Stage II Crack Growth Rate (m/s) 1.E E E E E E E E E 07 TDS H Uptake (ATI Allvac) C HCRIT = 1100 wppm H YS X CRIT = 1 m YS = 773 MPa D H EFF = 1 x cm 2 /s Model Predicted Monel K 500 ATI Allvac 3.5% NaCl TDS-measured egress D H of 1x10-10 cm 2 /s controls maximum da/dt II with X C = 1 m erf -1 and C C control steep drop in da/dt II amplifying C H,Diff reduction with rising E applied E Applied Electrode Potential (mv SCE ) Crack tip H = 12 YS

20 Predict crack tip H diffusion-rate limited IG HEAC growth in Monel K-500 One adjustable parameter C C = 1100 wppm da dt II 4D X C H erf 1 1 C H, Diff CC σhv exp RT H 2 Stage II Crack Growth Rate (m/s) 1.E E E E E E E E E 07 TDS H Uptake (ATI Allvac) C HCRIT = 1100 wppm H YS X CRIT = 1 m YS = 773 MPa D H EFF = 1 x cm 2 /s Model Predicted ATI Allvac Model Measured Predicted Special Metals Measured Monel K 500 ATI Allvac 3.5% NaCl TDS-measured egress D H of 1x10-10 cm 2 /s controls maximum da/dt II with X C = 1 m erf -1 and C C control steep drop in da/dt II amplifying C H,Diff reduction with rising E applied E Applied Electrode Potential (mv SCE ) Crack tip H = 12 YS

21 Predict crack tip H diffusion-rate limited IG HEAC growth in Monel K-500 One adjustable parameter C C = 1100 wppm da dt II 4D X C H erf 1 1 C H, Diff CC σhv exp RT H 2 Stage II Crack Growth Rate (m/s) 1.E E E E E E E E E 07 TDS H Uptake (ATI Allvac) C HCRIT = 1100 wppm H YS X CRIT = 1 m YS = 773 MPa D H EFF = 1 x cm 2 /s Model Predicted ATI Allvac Model Measured Predicted Special Metals Measured Monel K 500 ATI Allvac 3.5% NaCl Concentration dependent D EFF for mild-reversible H trapping at 100 nm-spaced? E Applied Electrode Potential (mv SCE )

22 SCCrack simulation of effect of cathodic potential distribution on IGSCC time-to-failure for fastener thread-root crack in Monel K-500 in NaCl solution a INITIAL = 1 mm APPLIED = 40 to 80% YS E MEAN = -830 mv SCE -946 < E APPLIED < -714 mv SCE (+/- 95 pct bounds) 1.2 years 550 simulations

23 Key Points Tools High resolution-accelerated measurement of crack growth kinetics provides quantitative input to probabilistic fracture mechanics simulation of H- cracking Damage tolerant approach to SCC-HEAC justified Approximations enable quantitative transition from decohesion mechanism of H damage, through crack tip modeling of alloy-h cracking properties, to component life simulation Example Mechanism-based models predict critical level of reducedcathodic polarization that eliminates IGSCC in Monel K-500; Scaled to predict component life Recent advances further validate engineering-relevant K TH and da/dt II models, and reduce necessary-adjustable parameters

24 Model consistency and adjustable parameter reduction K TH 1 β' (k exp IG αc α"σ YS Hσ ) 2 da dt II 4D χ H-EFF CRIT erf 1 C Hσ C C Hσ CRIT 2 " Stage II Crack Growth Rate (m/s) 1E E E E E E E E E TDS H Uptake (ATI Allvac) C HCRIT = 610 wppm H = 12 YS X CRIT = 1 m YS = 773 MPa D H EFF = 1 x cm 2 /s Calculated Ccrit ATI Allvac Measured Special Metals Measured Monel K 500 ATI Allvac 3.5% NaCl 1E Applied Electrode Potential (mv SCE ) MONEL K-500 Calculate C CRIT from -fit plus average K TH and da/dt II at to mv SCE C CRIT = 610 wppm H Crack tip stress is only debatable parameter at H = 12 YS

25 STRESS STATE Several mechanisms produce very high crack tip H and C H Elastic singularity (Oriani 1972) Blunt crack, large FEA (McMeeking, Sofronis 1980) 25 YS 2-5 YS X CRIT m Tip dislocation shielding (Rice, Thompson 1974) (Gerberich, Oriani 1991) GPa (Needleman 2000) 6-8 YS nm Strain gradient plasticity (Gao, Huang, Nix, Hutchison 2000) (Hutchison, Fleck 2001) (Begley, Gangloff, Agnew 2008) (Martinez-Paneda, Betegon 2014) Crack tip H-cell hardening (Robertson et al. 2012) 3-40 YS 100 nm-30 m 100 nm-2 m

26 NEW DEVELOPMENT Large deformation MSGP predicts substantial elevation of crack tip stress, compared to length-independent plasticity theory; justifying H = 12 YS over 800 nm Monel K-500 N = 0.05 l MSGP = 28 µm K = 40 MPam MSGP Martinez-Paneda and Betegon (2014)

27 ISSUE Upper Bound Sharp crack dislocation shielding and small-deformation gradient plasticity predict high, BUT.. Over small distance < nm Dislocation structure for stress model not established Lower Bound Blunt crack large-deformation gradient plasticity predicts elevated stresses up to 30 µm, BUT.. Relevant length scale not established GND hardening description in plastic flow rule must be selected from Phenomenological or Taylor-Mechanism model FEA must capture physical detail of crack tip opening shape

28 ISSUE Independent determination of alloy sensitive l SGP Monel K-500 N = 0.05 K = 40 MPam l MSGP = 28 µm MSGP Large Estimate A 5.4 m for Monel K 500 Estimate B 5.6 m for Ni w/o Evans and Stolken; Soboyejo et al. 5 from simulation comparisons for fixed gradient Wei, Qui and Hwang m for Monel K 500 Martinez-Paneda and Betegon (2014)

29 Is a large deformation MSGP continuum model of a blunting crack tip physically justified for a H crack? COD m Monel K-500 N = 0.05 l MSGP = µm K = 40 MPam Distance m AerMet 100 N = 0.03 l MSGP = µm K = 20 MPam COD m Martinez-Paneda and Betegon (2014) Distance m

30 Key Points Tools High resolution-accelerated measurement of crack growth kinetics provides quantitative input to probabilistic fracture mechanics simulation of H- cracking Damage tolerant approach to SCC-HEAC justified Approximations enable quantitative transition from decohesion mechanism of H damage, through crack tip modeling of alloy-h cracking properties, to component life simulation Example Mechanism-based models predict critical level of reducedcathodic polarization that eliminates IGSCC in Monel K-500; Scaled to predict component life Recent advances validate engineering-relevant K TH and da/dt II models, and reduce necessary-adjustable parameters Example Mechanism-based models predict benefits of H diffusivity and mild cathodic polarization to minimize TG- HEAC in ultra-high strength martensitic steel Kehler and Scully; Corrosion (2008) Lee and Gangloff; Metall. Mater. Trans. (2007) Thomas, Li, Gangloff and Scully; Metall. Mater. Trans. (2002)

31 Crack Tip Diffusible H Concentration (wppm) AerMet M NaCl OCP Applied Potential (V-sce) Fe-Co-Ni-Cr-Mo-C 0.6 M NaCl 20 nm (Mo,Cr) 2 C strengthened Crack tip C H-Diff vs E applied from scaled crevice and H solubility measurements After Kehler and Scully Trap-sensitive D H for H egress; TDS Thomas, Li, Gangloff, Scully 1 x 10-9 cm 2 /s from permeation rise Figueroa and Robinson

32 Small Deformation PSGP K = 40 MPam N = 0.20 L 1-PSGP = 120 nm PSGP H = 12 YS over 24 nm 100 nm Komaragiri, Agnew, Gangloff and Begley (2014) Large Deformation MSGP AerMet 100 N = 0.03 l MSGP = 25 µm K = 20 MPam MSGP H = 4.2 YS over 900 nm Martinez-Paneda and Betegon (2014)

33 Threshold Stress Intensity MPam Ultra-high Strength Steel 3.5% NaCl = 15 MPam (at frac H) -1 k IG = MPam (K IC ) H = 4.2 YS YS = 1725 MPa upper bound C H Diff (Kehler and Scully) AerMet 100 Ferrium M54 Fe-Co-Ni-Cr-Mo-C 0.6 M NaCl 20 nm (Mo,Cr) 2 C strengthened K TH 1 β' (k exp αc α"σ YS Hσ Crack tip H = 4.2 YS from MSGP = 15 MPam(at frac H) -1 for fit IG ) Applied Potential (V SCE ) da dt II 4D χ H-EFF CRIT erf 1 C Hσ C C Hσ CRIT D EFF from 1 st permeation rise C CRIT = 85 wppm calculated from Single-adjustable parameter needed to model potential dependencies of K TH and da/dt II 2 Stage II Crack Growth Rate (m/s) 1E E E E Ultra high Strength Steel 3.5% NaCl 1E C HCRIT = 85 wppm H = 4.2 YS X CRIT = 1 m YS = 1725 MPa 1E D H EFF = 1 x 10 9 cm 2 /s (Robinson) C H TIP (Kehler and Scully) OCP Upper Bound Solution 1E Applied Electrode Potential (mv SCE ) AerMet 100 Ferrium M54

34 Threshold Stress Intensity MPam Ultra-high Strength Steel 3.5% NaCl = 15 MPam (at frac H) -1 k IG = MPam (K IC ) H = 4.2 YS YS = 1725 MPa upper bound C H Diff (Kehler and Scully) AerMet 100 Ferrium M54 Fe-Co-Ni-Cr-Mo-C 0.6 M NaCl 20 nm (Mo,Cr) 2 C strengthened Is variability in measured K TH due to passive film hindrance of H uptake at potentials just cathodic to OCP? (Kehler and Scully) Applied Potential (V SCE ) Is da/dt over-predicted in low C H regime due to: (a) film-reduced H solubility, or (b) H-concentration dependent reduction in D H-EFF? Stage II Crack Growth Rate (m/s) 1E E E E Ultra high Strength Steel 3.5% NaCl 1E C HCRIT = 85 wppm H = 4.2 YS X CRIT = 1 m YS = 1725 MPa 1E D H EFF = 1 x 10 9 cm 2 /s (Robinson) C H TIP (Kehler and Scully) OCP Upper Bound Solution 1E Applied Electrode Potential (mv SCE ) AerMet 100 Ferrium M54

35 Validation SCCrack accurately simulates 425 day cantilever beam HEAC in AerMet 100 in NaCl with distribution of potential near OCP, typical of long term exposure Cantilever test time 1,420 days Rising K test time 3 days Mean E APPLIED = 560 mv SCE 4000 Monte Carlo selected simulations 633 < E < 518 mv SCE (+/ 95pct)

36 Conclusions Tools High resolution-accelerated measurement of crack growth kinetics provides quantitative input to probabilistic fracture mechanics simulation of component H-cracking Damage tolerant approach to SCC-HEAC justified Approximations enable quantitative transition from decohesion mechanism of H damage, through crack tip modeling of alloy-h cracking properties, to component life simulation Example Mechanism-based models predict critical level of reducedcathodic polarization that eliminates IGSCC in Monel K-500; Scaled to predict component life Example Mechanism-based models predict strong impacts of H diffusivity and mild cathodic polarization to minimize TG-HEAC in ultra-high strength steels; Scaled to predict component life Advances validate engineering-relevant K TH and da/dt II models, and reduce necessary-adjustable parameters

37 Future Research Enhance accelerated test method to measure da/dt < 0.3 nm/s by eliminating dv/dt from plasticity Resolve modeling uncertainties Crack-opening geometry-roughness effect on tip ph and potential Crack stress field, impacted by microstructure and dislocation shielding Failure physics, C H-crit vs local, from decohesion; HELPed and AIDEd H diffusivity and diffusion analysis for trap-rich FPZ at relevant length Couple crack tip stress intensity and H concentration similitude Develop HEAC data and H uptake understanding for occluded crack tip under atmospheric-exposure spectra Measure localized-trapped H concentration