Background and Overarching Goals of the PRISM Center Jayathi Murthy Purdue University. Annual Review October 25 and 26, 2010

Transcription

1 Background and Overarching Goals of the PRISM Center Jayathi Murthy Purdue University Annual Review October 25 and 26, 2010

2 Outline Overview of overarching application Predictive simulation roadmap V&V and UQ overview Progress during last year Organization and management 2

3 PRISM Mission Accelerate substantially the development of MEMS technologies for civilian and defense applications Significantly improve understanding of long-term reliability of MEMS and survivability in harsh environments Achieve this goal by simulating rigorously with quantified uncertainty, the physics of failure coupled electrical, mechanical, thermal and materials behavior from atoms to devices verification and validation 3

4 Dielectric (SiO2/Si 3 N 4 ~200 nm ) Target PRISM Device Ni membrane (~1-3 µm thick) Ti layer (~250 A) Anchor Contacting capacitive RF MEMS switch Used for contact actuators and capacitive switches Metal membrane makes periodic contact with thin dielectric layer Pull-down electrode ( Au ~0.5 µm thick Ti layer ~ 250 A) Membrane Length ~ 400 µm Width ~ 100 µm Thickness ~ 1-3 µm N 2 or air environment (~ 1 atm) Actuation voltage ~ V Hold down voltage ~ 5-15V Switching frequency 100 Hz-10 khz Response time 3-10 microseconds 4

5 RF-MEMS Operation OFF ON Off state: Low capacitance On state: high capacitance 5

6 0.0 Volts y 0 y 0V V Volts time Pull-In and Pullout Voltage 2 d y m = FE k( y 0 y) b 2 dt F F E =1/2ε 0 V 2 /y 2 y 0 F E ε ε AV 2 ( yd + εyr ) ( y + ε y) dy dt yd yd 0 r rav ε r AV 2 d + r d r 0 x εε = 2( y) 2 ρ ( rt, ) drdx Charge trapping in dielectric decreases both F s =k(y pull-in 0 -y) and pull-out voltage. Device fails due to stiction. F s,f E V 6

7 Role of Creep and Fluid Damping 0V V Volts 2 d y m = FE k( y 0 y) b 2 dt dy dt 0.0 Volts time Creep changes effective restoring force Challenges: Identification of creep mechanisms Microstructure, grain size distribution Residual stress effects New mechanism: Plastic strain recovery Fluid damping retards motion affects velocity and acceleration at impact Challenges: Regime changes periodically from slip to rarefied and back Near-impact damping poorly understood because of interaction with roughness scales 7

8 Role of Dynamics, Contact and Impact Displacement (µm) Time (s) Velocity (m/s) Time Scales Closing time of switch ~ 5 µs Damping time (Q factor~10) ~ 100 µs Bounce time at impact ~ 100 ns Stress wave travel ~ 0.1 µs Voltage pulse time (100 khz) = 10 µs Switch dynamics matter at high frequencies Contact and impact matter if contact area changes or defects are created during contact. Stages of Contact Long range van der Waals interactions Impact possible plastic deformation, solid bridging and defect generation Pull-out van der Waals interactions; need to overcome strength of solid bridges 8

9 Goal I: Lifetime Prediction with Periodic Contact under Accelerated Testing Voltage (V) V failure V pull-in V pull-out OFF Volts Accuracy goal: Predicted mean lifetime to within an order of magnitude of experimental mean Voltage ON 0.0 Goldsmith et al., time Cycles 9

10 Goal II: Pull-in and Pull-out under Sustained Contact Voltage (V) V pull-in V pull-out Accuracy goal: Predicted pull-in and pull-out voltage after sustained contact to within 20% of experimental mean Voltage Pulse Duration 10

11 Goal III: Gap versus Voltage Predict the change of gap with respect to time at fixed voltage, V hold Combination of creep and dielectric charging Our goal is to predict mean of gap vs. time curve to with 20% with respect to experimental mean. 11

12 Predictive Simulation Roadmap 12

13 Predictive Simulation Roadmap Year 1 13

14 Mode 1 Year 1 Damping ratio within 10% of experiments 14

15 Predictive Simulation Roadmap Year 2 15

16 Year 2 Dielectric charging in MIM capacitor PRISM Si 3 N 4 surface topology (AFM) Atomistics-based contact model 16

17 Predictive Simulation Roadmap Year 3 17

18 Predictive Simulation Roadmap Years 4 & 5 18

19 Software Integration C/C++ building blocks combined with Python Modularity, maintenance, extensibility Flexible orchestration of solver suite 19

20 UQ Approach Tier-Wise Validation Approach Coarse-Grained System Model System Physics Tier 3 Uncertainty Propagation through Generalized Polynomial Chaos (gpc) Coupled Physics Single Physics Tier 2 Tier 1 Bayes Networks for System Level UQ b e a c d f g 20

21 Software Development Progress 21

22 Software Development Progress 22

23 Verification Progress 23

24 Verification Progress 24

25 Verification Progress 25

26 Verification Progress 26

27 Validation Progress 27

28 Validation Progress 28

29 Organization and Management 29

30 Organization and Management Prof. Sankaran Mahadevan of Vanderbilt University joined PRISM in August 2010 Expertise: Probabilistic Computational Mechanics Reliability and Risk Assessment Structural Durability, Fatigue and Fracture Model Verification and Validation Under Uncertainty Reliability-Based Design Optimization Haldar, A., and S. Mahadevan, "Probability, Reliability, and Statistical Methods in Engineering Design," John Wiley & Sons, Haldar, A., and S. Mahadevan, "Reliability Assessment Using Stochastic Finite Element Analysis," John Wiley & Sons, Mahadevan, S., and Rebba, R., Validation of Reliability Computational Models Using Bayes Networks, Reliability Engineering and System Safety, Vol. 87, No. 1, pp , Rebba, R., and S. Mahadevan, "Validation and Error Estimation of Computational Models," Reliability Engineering and System Safety, accepted. 30

31 Organization and Management Name Affiliation Expertise Prof. Andreas Cangellaris External Advisory Board ECE, UIUC Dr. James Allen Sandia MEMS Dr. M.S. Anand Rolls Royce CFD Electromagnetics simulation, MEMS Prof. Bharat Bhushan OSU Nanotribology Dr. Marco Brunelli Siemens Probabilistic design Dr. Barry Farmer AFRL Material Science Dr. Chuck Goldsmith MEMtronics MEMS Prof. Yogesh Jaluria ME, Rutgers Heat Transfer, Numerics Dr. Ravi Mahajan Intel Microelectronics 31

32 Summary Significant progress during the last year: Uncertainty quantification of PRISM device operation using MEMOSA Pull-in behavior Damping Dielectric charging Coarse-grained model with uncertainty quantification New Bayes network approach to system level UQ New physical models for dielectric charging, contact, creep Extensive new experimental data for damping, pull-in and creep Multiscale material model from atomistic to continuum scales New UQ methods and software Improvements to numerical algorithms and scaling 32

33 Science and Engineering Contributions Fundamental understanding of the behavior of complex microsystems Understanding of the role of atomistic phenomena in mesoscale behavior Understanding of the role of mesoscale phenomena in microsystems Fundamental understanding of dielectric charging mechanisms Fundamental understanding of contact physics and connection to device performance Understanding of the role of micromechanics in device behavior Understanding of the role of sub-continuum fluid damping effects in device behavior 33

34 Science and Engineering Contributions New frameworks for simulation and uncertainty quantification in complex microsystems Simulation algorithms for strongly interacting multiphysics multiscale systems Uncertainty quantification across physics and scales Immersed boundary framework for new classes of physics Collocation UQ based on sparse and adaptive grids for in complex microsystems Highly scalable solution algorithms: molecular dynamics algorithms for complex force fields; SPIKE Public-domain software deployed on memshub One-of-a-kind database of experimental data to support UQ in microsystems 34