Compression Deformation Response of 95.5Sn-3.9Ag-0.6Cu Solder* P. Vianco J. Rejent Sandia National Laboratories Albuquerque, NM *Sanida is a multiprogram laboratory operated by Sandia Corporation, a Lockheed-Martin Company for the United States Dept. of Energy, under Contract DE-AC04-94AL85000. UCLA Pb-Free Workshop 090502 1 Hot Cold Failure Probability Empirical, accelerated aging programs are fast becoming impractical for developing long-term reliability databases Number of Cycles Weibull Distribution/Plot •A vast array of currently-used electronic packages •Rapid development of new electronic packaging technologies •Shift from OEM “circuit board technology” … … to CMS “circuit board assembly” • ...and now -- Pb-free solders UCLA Pb-Free Workshop 090502 2 Computational modeling will be heavily relied upon to predict the reliability of Pb-free solder interconnects Compile materials properties data for model input parameters. Material Material Properties Properties Develop the computational model: model •Constitutive equation •Finite element code (mesh) •Optimization routines Computational Computational Model Model Model validation using limited accelerated aging experiments. Model Model Validation Validation UCLA Pb-Free Workshop 090502 3 A Unified Creep-Plasticity (UCP) constitutive equation provides the basis for a computational model [ ] dε11/dt = fo exp(-Q/RT) sinhp |σ11 - B11| sgn (σ11-B11) βD “A Visoplastic Theory for Braze Alloys,” … Neilsen, Burchett, Stone, and Stephens (1996) dε11/dt ….. the inelastic strain rate (creep + plasticity) σ11 ...……... applied stress T …………. temperature B11 ……….. back stress D …………. isotropic strength (plasticity) β …………. constant (plasticity) fo …………. constant (creep) Q …………. apparent activation energy (creep) p …………. “sinh law” exponent (creep) UCLA Pb-Free Workshop 090502 4 Finite element analysis provides the spatial “locator” of stress and strain rate within the solder joint geometry Finite Finite element element mesh mesh (x, (x, y, y, z) z) ddε ε11 /dt = ƒ {σ(x, y, z); T; t} 11/dt = ƒ {σ(x, y, z); T; t} (x, y, z) UCLA Pb-Free Workshop 090502 5 Objective Develop a unified creep-plasticity constitutive model to predict the reliability of 95.5Sn-3.9Ag-0.6Cu soldered interconnects. •Step 1 … Materials properties database UCLA Pb-Free Workshop 090502 6 Experimental procedures Compression testing (ASTM E9-89A): 95.5Sn-3.9Ag-0.6Cu 10 mm Chill-cast (modified bullet mold) Machined 19 mm 10 mm Specimen Specimen geometry geometry UCLA Pb-Free Workshop 090502 7 Experimental procedures Test Temperatures: Upper platen Upper temperature control platen -25°C, 25°C, 75°C, 125°C, 160°C Strain rates (stress-strain): Temperature control lines 4.2 x 10-5 s-1, 8.3 x 10-4 s-1 Sample Lower temperature control platen Lower platen Mechanical Testing Creep stress (percent of σy): Upper temperature control platen 20%, 40%, 60%, 80% (2.7 - 35 MPa) Sample Samples test conditions: •as-fabricated •post - 125°C, 24 hour heat treat Lower temperature control platen UCLA Pb-Free Workshop 090502 ((-25°C) -25°C) 8 Experimental procedures •Dynamic elastic modulus meausurements: • -50°C to 200°C • Temperature dependence of density, ρ: Rx Sample Tx Furnace 25°C + 4.0 x 10-4 g/cm3-C° - 4.0 x 10-4 g/cm3-C° 7.30 g/cm3 •Thermal expansion coefficient measurements: • • -50°C to 200°C LVDT Convert the expansion data to a coefficent of thermal expansion (CTE). Furnace UCLA Pb-Free Workshop 090502 Sample 9 Microstructure of the as-cast Sn-Ag-Cu solder Dendritic microstructure was prevalent near the cylinder walls. D C B A E Untested 100 µm UCLA Pb-Free Workshop 090502 10 Microstructure of the as-cast Sn-Ag-Cu solder Equiaxed microstructure was observed near the cylinder interior. D C B A E Untested 100 µm UCLA Pb-Free Workshop 090502 11 Microstructure of the as-cast Sn-Ag-Cu solder Effect of the 125°C, 24 hour aging treatment on the Sn-Ag-Cu solder microstructure: The dendritic morphology near the cyclinder walls became slightly more equiaxed in appearance. There was no noticeable change to the microstructure that was interior to the cylinder sample geometry. D C B A E UCLA Pb-Free Workshop 090502 12 Stress/strain response of the Sn-Ag-Cu solder 60 -25°C True Stress (MPa) 50 40 25°C 30 20 75°C 10 0 5 s--1 4.2 4.2 xx 10 10--5 s1 0 0.02 0.04 0.06 0.08 0.1 0.12 True Strain • “Roll-up” to linear-elastic deformation for tests at -25°C and 75°C. • Transition from linear-elastic to plastic deformation for tests at 25°C. UCLA Pb-Free Workshop 090502 13 Stress/strain response of the Sn-Ag-Cu solder 20 Aged ... 8.3 x 10 -4 s Aged ... 4.2 x 10 -5 s -1 -1 -5 As-cast ... 4.2 x 10 15 True Stress (MPa) -4 As-cast ... 8.3 x 10 -1 s -1 s 10 5 125°C 125°C 0 0 0.02 0.04 0.06 0.08 0.1 0.12 True Strain Plastic deformation appears to reflect two simultaneous processes: work hardening ? dynamic recovery. recovery UCLA Pb-Free Workshop 090502 14 Stress/strain response of the Sn-Ag-Cu solder Plastic deformation at 125°C: work hardening ? dynamic recovery Thermal aging: WORK HARDENING > dynamic recovery. Faster strain rate: WORK HARDENING > dynamic recovery. UCLA Pb-Free Workshop 090502 15 Stress/strain response of the Sn-Ag-Cu solder 16 -4 As-cast ... 8.3 x 10 14 Aged ... 8.3 x 10 -4 s Aged ... 4.2 x 10 -5 s True Stress (MPa) 12 10 As-cast ... 4.2 x 10 s -1 -1 -1 -5 s -1 8 6 4 160°C 160°C 2 0 0 0.02 0.04 0.06 0.08 0.1 0.12 True Strain Plastic deformation appears to reflect two simultaneous processes: work hardening ? dynamic recrystallization UCLA Pb-Free Workshop 090502 16 Stress/strain response of the Sn-Ag-Cu solder Plastic deformation at 160°C: work hardening ? dynamic recrystallization Thermal aging: WORK HARDENING > dynamic recrystallization. Faster strain rate: WORK HARDENING > dynamic recrystallization. Work hardening, dynamic recovery and dynamic recrystallization are not adequately understood to develop quantitative state variables. UCLA Pb-Free Workshop 090502 17 Deformation microstructure of Sn-Ag-Cu solder D C B A E As-cast 160°C 8.3 x 10-4 s-1 100 µm UCLA Pb-Free Workshop 090502 18 Yield stress versus temperature (ASTM E9-89) 50 As-cast condition Aged: 125°C, 24 hours Yield Stress (MPa) 40 5 s--1 4.2 4.2 xx 10 10--5 s1 30 20 10 0 -50 0 50 100 150 200 Test Temperature (°C) UCLA Pb-Free Workshop 090502 19 Yield stress versus temperature (ASTM E9-89) 50 As-cast condition Aged: 125°C, 24 hours 40 Yield Stress (MPa) 4 s--1 8.3 8.3 xx 10 10--4 s1 30 20 10 0 -50 0 50 100 150 200 Test Temperature (°C) UCLA Pb-Free Workshop 090502 20 Static elastic modulus versus temperature (ASTM E111-82) 6000 Elastic Modulus (MPa) 5000 4000 3000 2000 As-cast condition 1000 Aged: 125°C, 24 hours 5 s--1 4.2 4.2 xx 10 10--5 s1 0 -50 0 50 100 150 200 Test Temperature (°C) UCLA Pb-Free Workshop 090502 21 Dynamic (acoustic) elastic modulus versus temperature 60 As-cast sample #1 As-cast sample #2 Aged sample #1 Elastic (Young's) Modulus (GPa) 55 50 45 40 35 30 -50 0 50 100 150 200 Test Temperature (°C) UCLA Pb-Free Workshop 090502 22 Differential expansion versus temperature Differential Expansion (mm) 100 10 -3 75 10 -3 50 10 -3 25 10 -3 As -cast As-cast 0 10 0 -50 -25 0 25 50 75 100 125 150 175 200 Temperature (°C) There was no indication of solid-state phase transitions. UCLA Pb-Free Workshop 090502 23 Coefficient of thermal expansion versus temperature Coefficient of Thermal Expansion (°C -1 ) 30 10 -6 25 10 -6 20 10 -6 As-cast #1 As-cast #2 15 10 -6 Aged #1 Aged #2 10 10 -6 -50 -25 0 25 50 75 100 125 150 175 200 Temperature (°C) UCLA Pb-Free Workshop 090502 24 Analysis of the creep test data dε/dtmin = fo exp(-Q/RT) sinhp (ασ) 5 10 -7 0.03 Stress: Stress: 10.2 10.2 MPa MPa Temp: 75°C Temp: 75°C As-cast As-cast condition condition Hyperbolic Sine Creep Law 0.03 0.02 Strain 3 10 -7 Independent variables: • • • • • • ln(fo ), -Q/R, p 2 10 -7 0.01 ln(dε/dtmin), (1/T), ln[sinh (ασ)] Coefficients: 0.02 1 10 -7 0.01 0 0 2 10 5 4 10 5 6 10 5 8 10 5 0 1 10 6 Time (s) Independent parameter Reported Reported data data will will be be for for the -CAST condition. the AS AS-CAST condition. UCLA Pb-Free Workshop 090502 25 Strain Rate (1/s) Multivariable Multivariable Linear Linear Regression Regression Analysis Analysis 4 10 -7 Analysis of the creep test data 5 10 -7 0.03 Stress: Stress: 10.2 10.2 MPa MPa Temp: 75°C Temp: 75°C As-cast As-cast condition condition 0.02 2 10 -7 0.01 1 10 -7 0.01 0 0 2 10 5 4 10 5 6 10 5 8 10 5 0 1 10 6 Stress: Stress: 10.2 10.2 MPa MPa Temp: 75°C Temp: 75°C As-cast As-cast condition condition 0.1 8 10 -7 0.08 6 10 -7 0.06 4 10 -7 0.04 2 10 -7 0.02 Time (s) 0 0 2 10 5 4 10 5 6 10 5 8 10 5 0 1 10 6 Time (s) A large degree of sample-to-sample variability was observed for specimens tested in the as-cast condition. UCLA Pb-Free Workshop 090502 26 Strain Rate (1/s) Strain 3 10 -7 Strain Rate (1/s) 0.02 1 10 -6 0.12 4 10 -7 Strain 0.03 Analysis of the creep test data 10-4 True Strain Rate (/s) -25C True Strain Rate (/s) 25C True Strain Rate (/s) 75C True Strain Rate (/s) 125C True Strain Rate (/s) 160C True Strain Rate (s -1 ) 10-5 10-6 10-7 10-8 10-9 0 5 10 15 20 25 30 35 True Stress (MPa) UCLA Pb-Free Workshop 090502 27 Analysis of the creep test data -10 -12 75°C ln[(d ε/dt) min ] (dε/dt min -1 , s ) 160°C 125°C 25°C -14 -16 -25°C -18 -20 -5 -4.5 -4 -3.5 -3 -2.5 -2 -1.5 ln[sinh(ασ)] (σ, MPa) dε/dtmin = 34856 exp(-43? 13 (kJ/mol)/RT) sinh4? 3 (0.005σ) UCLA Pb-Free Workshop 090502 28 Summary 1. A Unified Creep Plasticity (UCP) model is being developed to describe inelastic deformation in 95.5Sn-3.9Ag-0.6Cu (wt.%) solder. solder for the conditions: As-fabricated Aged: 125°C … 24 hours 2. Yield stress, elastic (Young’s) modulus, bulk modulus, Poisson’s ratio, and coefficient of thermal expansion were determined for: -25°C to 160°C 3. The creep data as-cast samples were fit to a hyperbolic sine law: dε/dtmin = 34856 exp(-43? 13 (kJ/mol)/RT) sinh4? 3 (0.005σ) UCLA Pb-Free Workshop 090502 29 A final note ….. the back stress, B11 [ ] dε11/dt = fo exp(-Q/RT) sinhp |σ11 - B11| sgn (σ11 - B11) βD •The impact of the back stress, B11, or Bauschinger effect, on the fatigue response of solder is not well defined. A scenario in which B11 = 0 can be hypothesized when recovery processes occur very rapidly after load reversal Stress σy tension Strain •The back stress, B11, is difficult to measure experimentally. Experimental techniques almost certainly require load reversal procedures. UCLA Pb-Free Workshop 090502 σy compression σσyy >> σσyy 30