Manufacturing and Reliability Modelling Silicon Chip C Bailey University of Greenwich London, England Printed Circuit Board Airflow Temperature Stress at end of Reflow Stress Product Performance in-service
Contents Why use Modelling Manufacturing Models Solder paste printing Melting of solder paste Predicting solder joint shape Predicting stress after reflow process. Reliability Models Data requirements Microstructure effects Example comparing SnPb and SnAg
Why use Modelling. Cost of Quality warranty liabilities due to field failures; redesign; rework; and scrap costs. Lateness of product to market First two manufactures to market lock up 80% of business. Source: Harvard Business School (Business Week)
Trends in Packaging Technology 1980 1990 1995 2000 PIN GRID ARRAY QUAD FLAT PACK BALL GRID ARRAY FLIP CHIP ON BOARD PGA QFP BGA FCHIP 100% 86% 39% 5% (BASED ON 320 PIN APPLICATION)
Trends in PackagingTechnolgy Increasing on chip and off chip operating frequencies Higher I/O counts, larger chips /packages Array technology becoming preferred choice for high I/O components Uptake of naked die (flip chip) for high performance modules Pitch miniaturisation of all component package styles Lots of Challenges for Manufacturers - Modeling can help
Modelling of Solder Interconnects Macro Level (Joint + Component) Printing Solder Paste Solder Solidification, Underfill Reliability Melting Shape Residual Stress. Micro Level Voidage Intermetallics Ageing/Damage Large amount of activity Some modelling activity. Very little, if any.
University of Greenwich - Our Skills High-Performance computer-based modelling Centre for Numerical Modelling and Process Analysis 11 Academic staff; 20+ Post-Docs; 40+ PhD students. Provide Computational Modelling Technology: Numerical Analysis Mathematical Modelling Software Engineering. Parallel Processing Optimisation
Modelling Technology - PHYSICA (http://physica.gre.ac.uk) - Multiphysics Solidification - Parallel Flow Stress Integrated Solution Procedures Coupled algorithms for flow, heat transfer, stress, Electromagnetic radiation. Takes advantage of parallel architectures.
Modelling Technology - Surface Evolver (http://www.geom.unm.edu/software/evolver) Predicts liquid shapes governed by: Surface Tension Gravity
Modelling Technology Surface Evolver Surface Mesh Equilibrium shapes FEMGEN PHYSICA Volume Mesh Temperature, Stress PATRAN Stress Creep Strain
Flip-Chip Modelling Identify process route for low cost, high volume sub 100µ pitch flip-chip assembly Solder Paste Printing Solder Bump Solidification Solder Fatigue Void movement Die Warpage Furnace Solder Solidifies Apply Underfill Final Product
The Solder Paste Stencil Printing Process Why Model the Printing Process Understand dynamics of print flow Help design stencils Ensure good print quality Identify optimal process conditions
Fluid Dynamics Modelling for Solder Paste Modelling Requirements Rheology Data Non-Newtonian η η η η 0 Flux Properties Density, Viscosity Size Distributions = 1+ 1 K γ m Continuum Modeling of Paste Roll Solder Particle Dynamics - Filling and Aperture
Predicting Solder Joint Shapes Why Model Solder Joint Shapes Stand-off heights Alignment forces Good joint shape for reliability Aid in identifying suitable pad dimensions
Predicting Solder Joint Shapes Modelling Requirements Surface Tension Design Interface now used at Celestica
Solder Paste Melting Why Model Solder Melting Gain greater understanding of joint formation Help understand generation of defects such as: Voids Flux entrapment Modelling Requirements Flux Viscosity Flux-Oxide reaction rates. Flux Viscosity Surface Tension. Time Necking of Pellets
Heat Transfer + Solidification Why Model Solder Solidification Predict solidification rates Microstructure Formation Dissolution of materials Intermetallic formation SOLID MUSHY ZONE LIQUID
Predicting Solidification of Solder Joints Modelling Requirements Thermal Conductivity Specific Heat Latent Heat Liquidus, Solidus, Dissolution rates. Solidification Fronts Stress in Bumps
Fillet Lifting Lead Free Marangoni Flow TIME Solidification Stress
THERMAL CYCLING Reliability Modelling requirements Creep constitutive law. Lifetime model Materials Data Young s Modulus Poisons Ratio Encapsulant Solder bumps PCB CHIP Coefficient of Thermal Expansion ISSUES Microstructure changes Voids Materials data from bulk or joint tests Underfill Die Solder Cu pad Substrate
Thermal Cycling Flip Chip Creep Strain rate (Darveaux) [ ] n Q = A sinh( ασ ) exp( ) cr eff RT ε Lifetime Model (Coffin Manson). N f 3 2 1.96 = 0.38 ( ε cr sum ) s σ ij eff Temperature Stress Creep Strain
Underfill Material Properties Parametric Study Models can help to identify optimal underfill 100000 Number of cycles to failure 90000 80000 70000 60000 50000 40000 30000 20000 10000 0 1 8 15 Coef Thermal Expansion (ppm) 26 34 1.5 4.5 9 15 Youngs Modulus (GPa)
Modelling Microstructure effects : OOF s (NIST) Shear stress micrograph Adaptive mesh FE analysis
Microstructure Modelling: examples Case 1 Case 1 Case 2 Maximum creep strains are dependent on grain distributions. Case 2
Modelling Lead Free Materials OBJECTIVES Build 3-D Model of R2512 - geometry data supplied by National Physics Laboratory (UK) Predict stress-strain profiles and damage in PbSn solder for Different temperature cycle times Build model for Lead Free Solder (Data obtained from literature) Investigate response to different temperature cycles
Creep Law for solder γ cr = A sinh n ( ασ ) exp( Q RT ) Sn-Pb A=9.60E4/s, n=3.3, alpha=0.087/mpa, Q=8058.37R Sn-Ag A=9E5 (1/s) n = 5.5 a = 0.06527/MPa Q = 8690R reference: R. Darveaux and K. Banerji, Ball Grid Array Technology,1995,McGraw-Hill,pp379-442
Damage prediction Lifetime models only available for SnPb. No lifetime data for Lead-Free Accumulated strain energy density per thermal cycle is used as damage indicator V2 V1
Temperature profiles - Supplied by NPL Temperature(degrees) 140 120 100 80 60 40 20 0-20 -40-60 D 0 10 20 30 40 50 time(minutes) C A B Damage 9 8 7 6 5 4 3 2 1 0 Modelling Results for SnPb D 0 10 20 30 40 50 t(minutes) C A B Thermal cycles: Profile A, B and C are measured oven temperature Accumulated strain energy density (Damage) Profile D is hypothetical
NPL Project: Damage in Solder Creep Strain Energy Density used as damage indicator. V2 V1 Symmetry Plane Effective stress when thermal load is applied. Strain energy density at the end of a thermal cycle
Damage in SnAg and SnPb Joints Damage 9 8 7 6 5 4 3 2 1 0 D 0 10 20 30 40 50 t(minutes) C B Damage(MPa) Per Cycle Solder Area A B C D SnAg SnPb Modelling Results for SnAg A V2 V1 10% More damage in SnAg V1 7.55 4.51 6.22 6.789 V2 0.6 0.364 0.498 0.557 V1 7.19 4.27 5.89 7.6 V2 0.548 0.338 0.455 0.544
Conclusions Definite lack of data for lead-free solders. Key model requirements Thermal data Mechanical data All temperature dependent. Need set of experiments for creep data and lifetime relationships Effect of microstructure needs to be addressed. Link between manufacturing and reliability models required.