Structure and Mechanical Properties of Oxidation Resistant Alloys

Transcription

1 Structure and Mechanical Properties of Oxidation Resistant Alloys Kevin Hemker Johns Hopkins

2 The development of ever more efficient gas turbines has always been paced by the results of research and development in the concurrent fields of design and materials technology. - G.W Goward Surface and Coatings Tech, (1998) 73-79

3 Improved structural design and cooling technology applied to higher strengthat-temperature alloys cast by increasingly complex methods, and coated with steadily improved coating systems, have lead to remarkably efficient turbine engines. - G.W Goward Top coat TGO Bond coat Substrate A modern TBS Surface and Coatings Tech, (1998) 73-79

4 Drivers for bond coat design Improve oxidation resistance and slow TGO formation. Increase bond coat strength. Alloy to avoid phase transformations and CTE mismatch with substrate.

5 Paradigm for materials development Performance - reduced weight - increased temperature Microstructure - crystal structure - grain size - defects - phases Processing - vapor deposition - electrodeposition -cast - wrought Properties - modulus, strength, CTE - temperature resistance

6 Outline A brief history of oxidation resistant alloys and coatings Diffusion aluminide bond coats Microstructure Properties Factors influencing ratcheting Overlay bond coats Microstructure Properties Factors influencing delamination

7 Delhi Iron Pillar The famous iron pillar in Delhi is a metallurgical wonder. This huge wrought iron pillar, 24 feet in height 16.4 inches in diameter at the bottom, and 6 1/2 tons in weight has stood exposed to tropical sun and rain for 1,600 years without showing any visible signs of rusting or corrosion. High concentrations of P catalyze the formation Why? of δ-feooh (iron oxyhydroxide), which is amorphous in nature and forms as an adherent compact layer next to the metal-scale How would interface, you Upon protect formation, the corrosion resistance is significantly enhanced this because iron pillor? δ-feooh forms a barrier between the rust and the metal. - Balasubramaniam (IIT)

8 Paths to oxidation resistance Alloying With select oxide formers Self healing Requires considerable quantities of solute Must not degrade other properties (e.g. creep) Coatings Must be applied Must retard oxidation Must be adherent Cannot degrade properties of the substrate

9 Select oxide formers Preferential oxidation Must form oxide or protective layer before base metal oxide is formed Protective Must form a dense, slow growing, protective film E.g. Cr 2 O 3, Al 2 O 3, SiO 2 E for Oxide Formation (kj/mol)

10 Alloying of steels Low carbon steels Cheap, easily formed, and mechanically strong, but rust at low T and oxidizes at high T. Stainless steels Add significant quantities of Cr (~18%) which forms a very protective oxide film Cuts down the rate of attack by C Affects microstructure and mechanical properties Other elements (e.g. Al, Si, P) also beneficial

11 Influence of Cr in superalloys Metal loss (a la Ashby) Ni -> NiO Ni loss ~0.1mm in 600h at 0.7T m ~950 C Cr -> Cr 2 O 3 Cr loss ~0.1mm in 1600h at 0.7T m ~1,231 C Cr loss ~0.1mm in 1,000,000h at 950 C Ni20% Cr -> Cr 2 O 3 Ni20%Cr loss ~0.1mm in 6,000h at 950 C Foreign elements increase diffusion rates Commercial superalloys limited to <10%Cr Not sufficient to protect alloy from oxidation!

12 Alloying of Ni-base superalloys Modern superalloys have been engineered over decades and contain ~10 elements. Composition of Single-crystalline René N5 Element: Ni Co Cr Ta Al W Re Mo Hf Ti C Wt% At% The composition of most elements is restricted by processing windows and mechanical property requirements.

13 Diffusion aluminide coatings Pack cementation aluminizing Parts packed in Al powder, sal ammoniac (NH 4 Cl), graphite and alumina filler and heated to 450 C for 2h. First used for Fe wire or ribbon heating elements, Cu condenser tubes, etc. First patent Van Aller, Allison, Hawkins (GE) 1911 Way to calorize metals to render then inoxidizable Attributed to selective formation of alumina scales

14 Schematic of the pack cementation process

15 Pack cementation reactions Pack Components Source (Cr, Al, Si or their alloys) Activator (NaCl, NH 4 Cl or other halide) Inert Filler (often alumina) Reactions between Source and Activator (Aluminizing) 3NaCl(g) + Al(l) = AlCl 3 (g) + 3Na(g) 2AlCl 3 (g) + Al(l) = 3AlCl 2 (g) AlCl 2 (g) + Al(l) = 2AlCl(g) Deposition on Substrate (Aluminizing) Disproportionation 2AlCl(g) = Al + AlCl 2 (g) 3AlCl 2 (g) = Al + 2AlCl 3 (g) Displacement AlCl 2 (g) + Ni = Al + NiCl 2 (g) Direct Reduction AlCl 3 (g) = Al + 3/2Cl 2 (g) And, for activators which contain hydrogen e.g. NH 4 Cl, Hydrogen Reduction AlCl 3 (g) + 3/2H 2 (g) = Al + 3HCl(g) (The underlined symbols refer to species in the solid substrate.)

16 Pack cementation coatings High-activity diffusion coating on a Ni-base superalloy - Al diffuses in - Low-activity diffusion coating on a Ni-base superalloy - Ni diffuses out -

17 Schematic of CVD processing

18 Aero history of diffusion aluminides 1950 s Allison and Curtiss Wright hot dipping of Ni blades 1960 s P&W aluminizing of Ni blades by slurry processing 1970 s Most vane and blade coatings applied by pack cementation and more recently by CVD 1990 s Recognized as useful TBC bond coats Forms an adherent α-al 2 O 3 TGO

19 Making a diffusion aluminide coating NiAl µm Ni substrate Ni and Al diffuse together and form an intermetallic coating

20 Ni-Al binary phase diagram:

21 Ni-Al binary phase diagram: γ FCC

22 Ni-Al binary phase diagram: γ L1 2

23 Ni-Al binary phase diagram: β B2

24 Commercial Pt aluminide coatings Pt (Pt,Ni)Al + X i µm Superalloy substrate Bond coat chemistry and microstructure effected by interdiffusion with substrate reactive elements appear to be good!!

25 TEM observations of diffusion alumined coating As-deposited Thermal cycled SADP SADP {110} T 100 nm Micro-beam DP B2 L1 0 Martensite

26 Compositional changes: Concentration (at. %) BC IDZ (a) As-TBC'd Ni 20 0 Al Pt 60 (b) 28% FCT life Ni Al 0 Pt 60 (c) 100% FCT life Ni Al 0 Pt Distance from surface of bond coat (µm) β β M M +γ + M +γ + γ +γ γ γ +γ As-TBC d 28% FCT life γ +γ 100% FCT life Sample Ni Al Pt Cr Co Ta Re W 0% % % Ni diffuses outward

27 Bond coat evolution Martensite transformation has been reported in this composition range. For example: Smileck, et al,, 1973, Metall.. Trans. Khadkikar, et al,, 1993, Metall.. Trans.

28 Failure by ratcheting and spallation Key observations: Isothermal exposure is not the same as thermal cycling. TGO thickens, lengthens and roughens as a result of thermal cycling. Bond coat must creep to allow TGO to ratchet. Top coat cannot creep; it cracks instead Mumm, Evans et al. (2001)

29 Advanced model of TGO ratcheting: Balint-Hutchinson model A special purpose multilayer code that takes input from micromechanical parameters and predicts the progressive development of undulations of the TGO layer upon thermal cycling. Key microstructural parameters: E sub, α sub, E bc, α bc, YS bc, (creep rate) bc, t bc, ε mart, M s, A s, α tgo, E tgo, (growth rate) tgo, (creep rate) tgo, YS tgo, t tgo, (shape) tgo, E tbc, α tbc, (creep rate) tbc, YS tbc, Note: property measurements and microstructural observations of bond coat are crucial to model development. Balint and Hutchinson, Acta 2003, JMPS 2005

30 Measuring the mechanical properties of diffusion aluminide bond coats

31 Bond coat microsample preparation: Bond coat Substrate 600 µm slice is scalped from 1 inch diameter button using a wire EDM. A lapping machine is employed to grind the superalloy side of the slice to get a uniform thickness of ~150µm. 500 µm A sinking EDM is then used to punch microsamples from each slice. Faces of the microsample are polished to a mirror finish and a final thickness of 30~50 µm with TEM dimpler. 100µm Pt lines are deposited on both sides of the microsample using a Focused Ion Beam (FIB).

32 Microsample testing Microsamples: 3.5 mm x 1.2 mm x 25 µm (overall) 250 µm x 500 µm x 25 µm (in gage) Fringe Detector Laser Fringe Detector Linear Air Bearing Load Cell Piezoelectric actuator Grips Manual turnbuckle Temperature range of C can be achieved by resistive heating. ISDG provides a strain resolution of better than 5 µstrain.

33 Measuring bond coat CTE (T): Engineering Strain (%) y = x R 2 = As-bond-coated Temperature ( o C) σ = ~2 MPa CTE of as-coated bond coat has been determined to be 15.5 ppm o C -1 between 400 o C and 850 o C. CTE measurement above 850 o C was hindered by creep.

34 Optical CTE measurements Rene N5 dε xx ( T ) CTE = dt = d dt u x

35 Optical CTE measurements?

36 Measuring bond coat E(T): Stress (MPa) T = 25 o C x E Strain (%) T = 650 o C Young's Modulus (GPa) BC-1, as-coated BC-2, as-coated BC-1, cycled NiAl literature values Temperature ( o C)

37 Microsample tests of Rene N5 : 1200 Engineering Stress (MPa) E = 129 GPa E = 126 GPa E = 126 GPa T = 25 o C Engineering Strain Modulus of [100] oriented single-crystalline Rene N5 microsamples measured to be 126 ± 3 GPa (8 measurements). GE reported modulus to be E = 129 GPa.

38 Texture effects on E: Young's Modulus The Elastic Constants of NiAl (Ref. Rusovic et al, 1977) C 11 C 12 C GPa 137 GPa 116 GPa In-plane Angle (Degree) (001) plane (011) plane (111) plane The Average E in Certain Texture of NiAl E(GPa) (001) plane (011) plane (111) plane Voigt Reuss Each angle corresponds to one certain orientation in that plane. The measured value of E is comparable to that of NiAl with a [001] texture. X-ray texture determination did not evidence any obvious out-of-plane texture of bond coats.

39 Measuring bond coat σ yield (T): Engineering Stress (MPa) E T = 650 o C T = 890 o C T = 980 o C Yielding Strength (MPa) As-BC'd -1 As-BC'd -2 TC Engineering Strain(%) Temperature ( o C)

40 Effect of thermal cycling 600 BC1,as-coated Yield Strength (MPa) BC1, cycled BC2, as-coated BC2, cycled BC3, as-coated BC3, cycled Deng Pan et al., Acta Temperature (C)

41 Effect of thermal cycling 600 BC1,as-coated 500 BC1, cycled BC2, as-coated Yield Strength (MPa) L1 0 Martensite BC2, cycled BC3, as-coated BC3, cycled 100 nm B2 Deng Pan et al., Acta Temperature (C)

42 High T deformation

43 Stress relaxation behavior: Stress (MPa) BC-1, as-coated 400 o C -experimental 650 o C -experimental 900 o C -experimental 1150 o C -experimental Time (s) Strain rate*exp(q/rt) (1/s) n = n = o C ε pl =A(σ/E sample ) n exp(-q/rt) 1150 o C Q = 250kJ/mol 125kJ/mol σ/e 650 o C Empirical fit suggests that Q creep is only half of Q D for NiAl and n = 4.

44 Stress relaxation behavior as-deposited Stress (MPa) o C -experimental 650 o C -experimental 900 o C -experimental 1150 o C -experimental Power law 50 Empirical fit: Time (s) ε pl =-σ/e machine = 1.2 x (σ/e sample ) 4 exp(-125(kj/mol)/rt)

45 Characterizing the microstructural changes in diffusion aluminide bond coats

46 In-situ TEM observations: 25 C -> 900 C L1 0 B2 Transformation from martensitic L1 0 to B2 observed at approximately 800 C.

47 n situ heating X-ray diffraction: Reversible transformation Crystallographic relationship c 25 C, cooled from 1150 C between B 2 and L1 0 b 1150 C L1 B 2 0 (110) L1o (110) L12 (200) L1o (200) L12 (201) L1o (220) L1o (221) L1o (202) L12 (311) L1o (111) L1o Intensity (100) B2 (111) L12 (200) L12 (111) B2 (200) B2 (210) B2 (211) B2 (110) B2 a 25 C (011)B2//(11-1)L θ (degree)

48 Transformation to B2: Intensity (counts) (111)L10 (110)B θ (degree) 450 ºC 500 ºC 550 ºC 600 ºC 625 ºC 650 ºC 675 ºC 685 ºC 700 ºC (200)L10 Volume Ratio(B2/ M) Volume ratio between B2 and Martensite calculated from the intensities of (110)B2 and (111)L Temperature (ºC) As The A s temperature is approximately 620ºC.

49 Martensite transformation temperatures: Heat Flow Endo Down (mw) DTA (Differential Thermal Analyzer) Peak T of B2 to M Cooling at ~25 C/min Peak T of M to B2 Heating at 40C/min The M to B2 transformation occurs at ºC The B2 to M transformation occurs at ºC M s is approximately 40-50ºC less than A s Temperature(C)

50 Thermal and transformation strains: Thermal strain of B Transformation M s Strain (%) strain cooling A s heating Thermal strain of L Temperature ( C)

51 Effect of martensitic transformation Balint and Hutchinson, JMPS 2005

52 Strategies for bond coat design Improve oxidation resistance and slow TGO formation. Increase bond coat strength. Alloy to avoid martensitic and other phase transformations.

53 Overlay coatings for superalloys Developed to allow for greater flexibility in bond coat compositions. Addition of reactive elements known to be beneficial Peg formation, S gettering, increased TGO adhesion??? Can tailor the phases present in the bond coat β + γ, β + γ, γ + γ, etc. Requires deposition Initially PVD More commonly now LPPS

54 Schematic of an EBPVD process

55 EBPVD coatings CoCrAlY coating as-deposited CoCrAlY coating after peening and heat treatment

56 Schematic of a plasma deposition system

57 NiCoCrAlY argon-shrouded plasma sprayed coating NiCoCrAlY coating after peening and heat treatment

58 Aero history of overlay coatings FeCrAlY Alloy merged from GE nuclear programs (1964) P&W applied (EBPVD) as a coating (1970) Limited at high T by formation of a brittle NiAl layer CoCrAlY (1972) Useful oxidation resistance but too brittle NiCrAlY (1973) Limited hot corrosion resistance but ductile NiCoCrAlY (1975; 1986) Good compromise; 40+ patented MCrAlY s

59 Microstructural observations of Pratt & Whitney s two-phase NiCoCrAlY(Hf,Si) bond coat

60 Effect of thermal cycling As-received o C Near-TGO γ T i Average across entire thickness γ Vf [%] β Temperature [C] two-phase bond coat (β-nial + γ-ni)

61 TEM observations of the bond coat β-phase Ni Ka 1250 Counts Co L/ Ni L 500 Al Ka Co Ka Cu Ka Cr Ka Ni Kb Cr Kb Cu Kb Energy (kev) γ-phase Ni Ka 1 µm Counts Cr Ka Co Ka As received NiCoCrAlY Co L/ Ni L Cr L C K Al K Si K Cr Kb Energy (kev) Cu Ka Co Kb Cu Kb

62 TEM observations of the γ grains: As-received BF 1 10 DF nm 200 nm ELTB BF 500 nm DF γ grains filled with very fine γ precipitates modest coarsening suggests the γ dissolves at T.

63 EFTEM of γ in the γ grains: 200nm Elastic image Thickness map Cr L (574 ev) Co L (779 ev) The γ precipitates are rich in Ni and Al while the matrix is rich in Cr and Co. Ni L (854 ev) Al L (73 ev)

64 TEM observations of the β grains As-received ELTB o C 750 nm g 110 g nm -Mottled- 250 nm -Tweed- -Lath- [001] [001]

65 Do β-grains contain martensite? As-received M s << RT M s << T M s ~ T ELTB M s = T M s < RT M s < T After Reynaud Scripta (1977)

66 Microstructural evolution of NiCoCrAlY Beta Gamma As-received Thermal cycled

67 Precipitates found in an ELTB Cr Counts 4000? 2000 Energy (kev) Co Ka Ni Ka Cr Cr L Co Kb Co Kb Si Ka Ni nm Cr-rich intergranular precipitates widely dispersed in the ELTB specimen but not observed in the as received samples. Composition is 55.5 at% Cr, 32.1 % Co, 11.0 % Ni and 1.4 % Si.

68 Crystal structure of Cr-rich precipitate 110/ / o (10 o ) 10 o (7 o ) B = [-110] B = [-221] B = [-111] 30 o (31 o ) 001/ / B = [-140] The precipitate is a (CrCoNi) σ-phase (P4 2 /mnm space group). Tetraganol unit cell with lattice parameters a = 8.37 Å and c = 4.30 Å.

69 Elemental maps of the σ-precipitates Mendis Backscattered image Al map Cr map Tryon Backscattered image Cr map Si map

70 Potential impact of σ-phase Notoriously brittle but bond coat is not a load carrying structure. Appears to trap Si, which would mitigate its effect. Is observed at the TGO interface and may affect adhesion. Bond coat TBC TGO Elastic image Cr L (574 ev) Note: only observed in ELTB; not as-received or after 100hrs at 1100C!

71 Predicting phase stability Achar, Munoz, Arroyo, Singheiser, Quadakkers, Modeling of phase equilibria in MCrAlY coating systems, Surface and Coating Tech (2004)

72 TGO and PEG Chemistry O/ Cr Al EDX of TGO peg columnar TGO Counts Y Co Co/ Ni Ni Hf Cr Hf/ Cu S Cr Ni Hf Energy (kv) 1 µm Al Counts O EDX of columnar TGO peg Y Energy (kv) The peg contains a large number of elements including sulfur, although the neighboring columnar TGO is nearly pure Al 2 O 3.

73 STEM shows S in pegs but not the TGO Bond coat Al O Ni Co S Hf Y Cr

74 Mechanical behavior of P&W s NiCoCrAlY(Hf,Si) bond coat

75 RT strength of NiCoCrAlY 1600 Tensile test of NiCoCrAlY at RT Eng. Stress [MPa] E=155 GPa as received thermally cycled E=130 GPa 1.4 GPa with RT ductility! 200 E=130 GPa True Strain [%]

76 Effect of thermal cycling Diffusion aluminde NiCoCrAlY - strength increases - strength decreases Yield Strength (MPa) Deng Pan et al., Acta 2003 BC1,as-coated BC1, cycled BC2, as-coated BC2, cycled BC3, as-coated BC3, cycled Eng. Stress [MPa] E=155 GPa E=130 GPa as received thermally cycled E=130 GPa Temperature (C) True Strain [%]

77 T im e [ s ] Temperature [C] Stress [MPa] Strain [%] High temperature deformation:

78 High temperature deformation: Eng. Stress [MPa] Constant strainrate Constant strainrate 225 Cross head fixed Cross head fixed True Strain [%] Time [s] 500ºC

79 Comparison of high T strength: 500 Flow stress [MPa] MCrAlY (Ni,Pt)Al corr (Ni,Pt)Al Temperature [C]

80 Failure of NiCoCrAlY TBC systems e.g. NiCoCrAlY bond coat systems Salient observations: No TGO ratcheting. Cracks occur below the TGO. Crack growth is mode- II and effected by interfacial energy and plasticity in the adjacent layers. Cracks must be initiated before they can grow. Xu, Evans et al., Acta 2004

81 Proposed failure mechanism map: No!!