The Effect of Elastic Compliance of PYSZ EB-PVD Thermal Barrier Coatings on Cyclic Test Lives Byron Williams – BAE Systems David. S. Rickerby – Rolls-Royce Surface Engineering Robert E. Jones – Rolls-Royce Surface Engineering December 2005 Rolls-Royce - data private
Background (i) Turbine Surface Technologies Ltd manufacture thermal barrier coatings (TBC) for Rolls-Royce Process steps involve; Production of Ni based super alloy turbine blades at Rolls-Royce Platinum coat applied by electroplating Heated to produce Ni-Pt-Al alloy bond coat December 2005 Rolls-Royce - data private
SEM image showing cross-section through bond coat December 2005 Rolls-Royce - data private
Background (i) Turbine Surface Technologies Ltd manufacture thermal barrier coatings (TBC) for Rolls-Royce Process steps involve; Production of Ni based super alloy turbine blades at Rolls-Royce Platinum coat applied by electroplating Heated to produce Ni-Pt-Al alloy bond coat PYSZ TBC deposited by electron beam vapour deposition December 2005 Rolls-Royce - data private
SEM Image showing columnar microstructure of PYSZ TBC Coating December 2005 Rolls-Royce - data private
Background (i) Turbine Surface Technologies Ltd manufacture thermal barrier coatings (TBC) for Rolls-Royce Process steps involve; Production of Ni based super alloy turbine blades at Rolls-Royce Platinum coat applied by electroplating Heated to produce Ni-Pt-Al alloy bond coat Y-Zr TBC deposited by electron beam plasma One blade from every batch (~30) sectioned for microscopy, chemical analysis and cyclic testing December 2005 Rolls-Royce - data private
Background (ii) Cyclic testing of blade sections gives very variable results December 2005 Rolls-Royce - data private
Variability of cyclic test data for standard production blades Data for average minimum cyclic life – error bars = +/- 1 standard deviation December 2005 Rolls-Royce - data private
Analysis of residual stress in TBC Background (ii) Cyclic testing of blade sections gives very variable results Work program initiated to elucidate factors giving rise to variability which was structured as; Detailed study of bond coat chemistry Characterisation of thermally grown oxide (Al2O3) Analysis of residual stress in TBC December 2005 Rolls-Royce - data private
asubstrate > apYSZ > aTGO Background (iii) EB-vapour deposition process carried out at ~ 1100oC Coefficient of thermal expansion asubstrate > apYSZ > aTGO On cooling the blade to RT the thermal mismatch generates residual stress in TBC coating Can this be measured – or inferred? Does this vary from sample to sample? Measure the cross-sectional hardness’ of representative samples December 2005 Rolls-Royce - data private
TBC Characterisation Microindentation (Vickers Hardness – Buehler Micromet indentation tester) Indenting at 30 points along centre of TBC CX & CC (Analogous procedure used for nanoindentation) Convex Concave December 2005 Rolls-Royce - data private
Vickers Hardness Testing Indenting at 30 points along middle of TBC coating Position of indents 50mm December 2005 Rolls-Royce - data private
Typical Microindentation Results December 2005 Rolls-Royce - data private
Microindentation Results: Summary Hardness data very scattered. Mean hardness for concave surface always greater than convex ~5GPa cx, ~6.2 GPa cc. Sample to sample variation within analysed population. Good cyclic life – narrower distribution Poor cyclic life – harder regions in coating Mean hardness does not vary markedly across population but sample-to-sample hardness distributions show marked variability December 2005 Rolls-Royce - data private
Correlation with Cyclic Life Data December 2005 Rolls-Royce - data private
Correlation with Cyclic Life Data December 2005 Rolls-Royce - data private
Weibull plot of all TBC Hardness Data December 2005 Rolls-Royce - data private
Correlation Results No trend apparent for convex surface, but concave data shows weak trend of increasing lifetime correlating with lower hardness Data sufficiently encouraging to justify further study – use nanoindentation to investigate variability in modulus December 2005 Rolls-Royce - data private
Nanoindentation Samples analysed at Cranfield University Analysed at 50 mN & 200mN / Triangular diamond indenter Two samples evaluated based on extremes from correlation data December 2005 Rolls-Royce - data private
Nanoindentation sample selection December 2005 Rolls-Royce - data private
Nanoindentation Samples analysed at Cranfield University Analysed at 50 mN & 200mN / Triangular diamond indenter Two samples evaluated based on extremes from correlation data Literature data used to calculate Modulus Weibull statistics used to present distribution December 2005 Rolls-Royce - data private
Nanoindentation: Data Treatment (i) Poisson ratio for PSYZ taken from Rolls-Royce data base – 0.15 Literature data used for Poisson ratio and hardness for diamond; Hardness = 1141GPa Poisson’s ratio = 0.07 Thus obtain value for modulus from nanoindentation data using the following . . . . . . . . December 2005 Rolls-Royce - data private
Nanoindentation: Data Treatment (ii) Nanoindentation analysis returns a value for the reduced modulus viz, Rearranging and substituting values; December 2005 Rolls-Royce - data private
Nanoindentation Data (i) – Convex surfaces E values, 50mN load. December 2005 Rolls-Royce - data private
Nanoindentation Data (ii) – Concave Surfaces E values, 50mN load. December 2005 Rolls-Royce - data private
Nanoindentation Data (iii) – Convex Surfaces (again) Data for modulus, E. 200mN load. December 2005 Rolls-Royce - data private
Nanoindentation Data (iv) – Concave Surfaces (again) Data for modulus, E. Note b values & high end data for 11225 . 200mN load. December 2005 Rolls-Royce - data private
Nanoindentation 50 vs 200mN Marked change in response going from 50 to 200 mN indentation load Distribution in modulus data larger for lower load (CX & CC) Lower indentation loading, obtain material (i.e. column) properties Higher indentation loading – interrogating coating properties December 2005 Rolls-Royce - data private
Conclusions Microindentation testing of TBC’s shows variability in sample-to-sample hardness Correlation of hardness data with cyclic test data suggests trend of high measured hardness values correlating with low cyclic test life Nanoindentation analysis of samples showing large differences in cyclic test life shows marked differences in modulus Poor performing sample more rigid cf sample showing increased test life Data suggest variation in mechanical properties contributes significantly to variability observed for cyclic test performance December 2005 Rolls-Royce - data private
Acknowledgements Andy Payne, Prof. John Nicholls – Cranfield University Paul Morrell – Rolls-Royce Surface Engineering BAE Systems / Rolls-Royce December 2005 Rolls-Royce - data private