1. Presented by Chean Lee (Balee) General Engineering Research Institute Electronic and Ultrasonic Engineering Supervisors Prof. Dave Harvey Dr. Guangming Zhang 25 March 2011 1

2. Project Objectives Introduction : Acoustic Microscopy Introduction : Simulation Introduction : Validation Simulation Setup Harmonic Response Analysis Governing Equations Conclusion Further work 2

3. Clarify defect detection mechanism Limited published literature regarding subject Primary focus on new generation 3D IC packages Understand acoustic performance within 3D IC packages Balance optimum resolution vs. penetration Analyze defect detection mechanism of engineered faults 3

4. 4 Longitudinal wave which consists of compression and rarefaction Human Hearing 20Hz20kHz100kHz Animal Navigation & Communication SeismologyMedical Diagnostics. Destructive & Non Destructive tools. Audible Infrasound Ultrasound Destructive Ultrasound (>10 W/cm 2 ) Sonochemistry Welding Cleaning Cell Disruption Kidney Stone Removal Non-Destructive Ultrasound (0.1 – 0.5 W/cm 2 ) Flaw detection Medical Diagnosis Sonar Chemical Analysis

5. Non-Destructive technique Sensitive to voids, delaminations and cracks Detects flaws down to sub-micron Image non-transparent solids or biological materials Study microstructures of specimen X-Ray AMI Unreflowed Solder Bump, AMI presents better contrast of defect 5

6. 6 Increasing frequency largely lowers depth penetration Dispersion and attenuation Lower frequency reduces resolution Exacerbated by frequency downshift 230MHz 50MHz

7. 7 Couplant or medium is required Usually deionized water Reflection occurs at the interface between two mediums Air has low acoustic Impedance (Z) Z = ρV = density * sound velocity of medium Water to Steel ratio ~ 20:1 Air to Steel ratio ~ 100,000:1 (near 100% energy reflected) Change in Impedance (Interface) Change in Impedance (Interface)

8. 8 Electronic packages are shrinking and/or stacking Technique is approaching resolution limits Image processing techniques not broadly reliable Transducers have fixed operational frequencies Optimal frequency difficult to determine

9. 9 Provide practical feedback when designing real world systems Diminish cost of system building Rapid Prototyping Simulate design decisions before construction phase Permit the system study of various level of abstraction Allow for Hierarchical Decomposition (top-down building technique) of complex systems

10. 10 ANSYS Parametric Design Language Scripting and automate task in ANSYS Automate complex and repeated task Virtually all ANSYS commands can be used in APDL No compilation. Modifications are immediately realized Resultant macro files are small and easy to share ANSYS Workbench Significantly better Graphic User Interface bi-directional association with CAD Advance contact pre-processing capabilities Advance meshing and defeaturing tools HOWEVER, Ansys Workbench does NOT support Acoustic Simulation

11. 11 Simulation provides an expectation to aid work Experimental and simulated results cannot usually be directly compared Simplifying assumptions or heuristics are applied to reduce computational resources May significantly affect simulation accuracy High abstraction has a tendency to have over simplified or omitted lower level details AND MOST IMPORTANTLY Large number of parameters involved Rubbish in, rubbish out

12. 12 Example 1 : Wide spectrum comparison between simulated and actual frequency response or a microwave cavity Two examples of experimental vs simulated results

13. 13 Example 2 : Comparison of prediction and monitoring result of flip chip solder joint failure Source: Ryan Yang, Solder Joint Reliability Conclusion: Experimental results alone is inadequate to validate a simulation

14. 14 Ansys contains a large library of elements. Each with its unique abilities and method of use.

15. 15 SOLID Elements PLANE Elements FLUID Elements Not all elements are suitable for a specific job

16. 16 Models usually tested with multiple mesh resolution/configuration Ensure circular areas are adequately smooth/circular Mesh resolution and type fits geometry size Generally look right, results largely determined by mesh quality Fine Mesh Coarse Mesh Fine Mesh Coarse Mesh Tetrahedral (triangular) Hexahedral (square)

17. Triangular vs. Quadrilateral (square) mesh Uniform, organized mesh Chaotic, Complex Mesh (inaccurate and/or extra computational resource) Noise and unpredictable propagation patterns were observed 17

18. 18 Element per wave (EPW) defines the element resolution of the simulation according to the highest frequency present in the simulation. The number of elements WILL affect the quality of the result. Rule of thumb; 5 EPW = Draft 10 EPW = Low Resolution 20 EPW = High Resolution Anything above 20EPW (of the highest frequency) is deemed to be a waste of resources. However, any complex simulations above 10EPW is already difficult to manage without cluster computing. Following two slides are comparisons

19. 19 5 Element Per Wave Low Sampling Frequency (Illustrated by the triangle) Rough and blocky wave integration

20. 20 10 Element Per Wave Smooth wave integration

21. 21 Infinite Boundary (radius line) Material layers Transducer Lens (Excited Directly) Fluid Medium (segmented to manage computational resource) Meshing application has a 32bit limitation

22. 22 Discontinuous meshing. Two zones will not interact in the simulation. Therefore geometry has to be merged. This may cause problems with the meshing engine as geometry becomes larger.

23. 23 Mesh defect occurring at the interface between two materials. In this case, the segmented fluid medium (which has the same material properties) Will cause inconsequential discrepancies with low element resolution.

24. 24 Simulation uses a sustained cyclic load to produce a harmonic response. In other words, a continuous wave which is ideal for studying transducer design. Focus depth, spot size, axial pressure and beam shape are obtained from this method.

25. 25 Real Transducer : Model design based on existing transducer. Virtual Transducer : Design with reduced focal distance (half) for resource saving Note: In certain materials, the focal distance becomes shorter. 141 Gbytes 3.9 Gbytes

26. 26 Focus depth 1mm Focus depth 1.5mm Transducer axial pressure

27. 27 Real Transducer Virtual Transducer Axial cross section of transducer output

28. 28 C = speed of sound = ρ o = mean fluid density K = bulk modulus of fluid P = acoustic pressure t = time This equation neglects viscous dissipation. Therefore represents a lossless wave equation for sound in fluids. For Fluid-Structure Interactions, the transient dynamic equilibrium equation below is considered simultaneously with the above acoustic wave equation [M] = Structural Mass Matrix [C] = Structural Damping Matrix [K ] = Structural Stiffness matrix {ϋ} = nodal acceleration vector {ύ} = nodal velocity vector {U} = nodal displacement vector {F a } = applied load vector The equation is employed using the generalized-α method. This method has been widely accepted to produce better results for Transient analysis (Chung, 1993)

29. 29 Where Z = ρV = density * sound velocity of material Longitudinal wave are normal incidence Therefore α = 0 Reflection Equation is can be simplified into

30. 30 Data is exported from Ansys and analyzed in Matlab Easier User Friendly Incident Pulse Reflection Fast Fourier Transform

31. 31 4 Interfaces First Reflection First Interface Result of Reflection calculation Calculated Result = 5.44 Ansys Result = 5.21

32. 32 The importance of validation is highlighted Broadband transient transducer successfully built into the simulation Most simulation defects and inconsistencies has been addressed Cost saving methods implemented with mixed success Basic calculated results correlate with simulated result

33. 33 1.Review calculations with comparisons between flat and focused transducers 2.Review accuracy of transducer lens (geometry) excitation method 3.Scale up simulation to higher frequencies 4.Find alternative resource saving methods 5.Conclude validation and Implement electronic packages in the simulation 6.Setup non-linear acoustic simulation with narrow band transducer

34. 34 J Chung, GM Hulbert. A time integration algorithm for structural dynamics with improved numerical dissipation: The generalized-α method. Journal of Applied Mechanics, June 1993, Vol 60, Pg 371. Sound Reflection. http://www.sal2000.com/ds/ds3/Acoustics/Wave%20Reflection.htm Computer Aided Engineering Associates Inc. Ansys-Customization and Automation with APDL. 2002. http://www.scribd.com/doc/25083369/ANSYS%C2%AE-Customization-and-Automation-With-APDL Alex Karpelson. Wide-band ultrasound piezotransducers with non-uniform electric field. NDT.net, Aug 2003, Vol 8, No 8. http://www.ndt.net/article/v08n08/karpels/karpels.htm Brian Lempriere. Ultrasound and Elastic Waves: FAQ. Academic Press, 2002. Ansys Reference Material. Theory Reference for the Mechanical APDL and Mechanical Applications. Mario Kupnik, Ira O. Wygant, and Butrus T. Khuri-Yakub. Finite element analysis of stress stiffening effects in CMUTs. IEEE International Ultrasonics Symposium Proceedings, pg 487, 2008. Goksen G. Yaralioglu, Baris Bayram, Amin Nikoozadeh, B.T. Pierre Khuri-Yakub. Finite elemenet modeling of capacitive micromachined ultrasonic transducers. Medical Imaging 2005: Ultrasonic imaging and signal processing, pg77, vol 5750.

35. 35 Thank You Questions Please?