E240_MEMSReliability - MEMS Reliability Issues Alissa M....

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: MEMS Reliability Issues Alissa M. Fitzgerald, Ph.D. A. M. Fitzgerald & Associates, LLC 655 Skyway Rd. Suite 118 San Carlos, CA 94070 www.amfitzgerald.com (650) 592-6100 Stanford E240 6 March 2007 A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Company Introduction • A.M. Fitzgerald & Associates provides technical consulting and contract R&D services – MEMS/Microsystems prototyping: “feasibility to first silicon” – Modeling and failure analysis – Technical due diligence and patent analysis • Founded in 2003 – Over 30 clients served – Start-ups to Fortune 500 companies, universities A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 Outline An Overview of Reliability Issues in MEMS Devices • Motivation • Electrical Failure Modes • Mechanical Failure Modes • A Detailed Look at Fracture • Summary A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 Motivation • Many MEMS devices are developed for applications that demand proven reliability: – Automotive: Airbag sensors, stability control systems – Medical: surgical devices, implantable sensors, drug-delivery devices – Telecom: optical switches, multiplexers – Consumer electronics: displays, RF components – Aerospace: navigation, safe & arm – Failure rates < 1 in 106, Lifetimes > 10 years • 15 um Texas Instruments DLP McAllister, Georgia Tech Reliability is a major hurdle to commercialization ~ 0.5 mm Sandia Laboratories Safe & Arm A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 Physics of Failure • • • Review of common electrical failure modes Well understood in IC industry Catastrophic: – Dielectric breakdown – Electromigration • Performance degradation: – Joule heating – Parasitics – Electrostatic charging A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 Dielectric breakdown electrodes • All insulators have a dielectric strength – Maximum applied electric field – E = V/d • • When dielectric strength is exceeded, electrons avalanche from cathode and destroy insulator Thin insulator structures will break down at low voltages A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com V d insulator Dielectric Strength Air 30 kV/cm Thermal Oxide 10 MV/cm PECVD Oxide 8 MV/cm Alissa M. Fitzgerald Stanford E240 3/6/07 Electromigration • • • High current density in narrow conductor lines; current-crowding Mass of electrode moves with time, can create voids or hillocks Progressive device degradation leading to failure http://www.dwpg.com/content.php?contid=2&artid=68 A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 Parasitics • Joule Heating – – – – • I2R heating of a conductor carrying current Affects resistance of conductor Performance shifts Will accelerate dielectric breakdown and electromigration Parasitic Capacitance – Overlapping or side-by-side interconnects – Can be controlled with clever design – Affects performance and measurement of devices with small capacitance A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 Charging • • • • • Malformed bonds at interface between metals and insulators can trap charge Charge can also become trapped in body of insulator PECVD dielectrics susceptible Charge rate = f(E field, Temp) Electrostatic devices will be affected, insulator will degrade with time electrode V oxide --------------silicon MOS transistor electrodes V -- - - - - - - - - - - - - -- - - - - - - - - - - - - Electrostatic MEMS A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 Physics of Failure • Mechanical Failure Modes – – – – Stiction, Wear Corrosion Curling, Buckling Fracture A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 Stiction • Stiction often a process issue – Sacrificial etching in wet chemical causes beams and membranes to adhere to surface (usu. permanently) – Critical point drying usually solves problem – Can occur in operation of liquid MEMS devices Primaxx Inc. A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 Stiction from contact • Sliding friction: surface properties important – Hardness – Roughness • Many environmental causes: – – – – • Van der Waals Electrostatic charging Capillary forces Contamination Solutions: vapor-phase deposition of low-friction materials – Parylene – Surface assembled monolayers (SAMs) A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 Wear • Depends on complex interaction of several variables: – – – – – Surface roughness Surface hardness Environment: humidity Temperature Pressure A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 Wear as a function of humidity 612,000 cycles, 31% RH 600,000 cycles, 1.8% RH 10 μm Dr. D. Tanner, Sandia Nat’l Labs A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 Stress Corrosion • Most MEMS devices need to interact with an environment – Moist air, liquids, chemicals, human body, etc. • • • Stress plus environmental factors can cause accelerated material degradation, i.e. “stress corrosion” Typically a “static” fatigue effect: loads are monotonic, not cyclic Phenomena is well studied in metals, less so with dielectrics A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 Stress-Corrosion Crack Growth in Glass (SiO2) Crack tip Water molecule Strained crack tip bonds Chemical Reaction Fracture Courtesy of Michael Lane Applied Stress A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 Curling and Buckling • Residual stresses in films – Thermal: Caused by mismatched c.t.e. between film and silicon wafer – Intrinsic: Caused by trapped gases, microstructure – Released structures are susceptible: beams, membranes – Metals typically tensile – Can be used advantageously • parc’s StressedMetal AFM tips Buckling – Due to compressive residual stresses – Typical source of fractures in films – Oxides always compressive Fraunhofer IOF A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 Fracture • • • Why Strength of Materials Doesn’t Apply Review of Fracture Mechanics Fracture Initiation and Surface Flaws A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 Why Strength of Materials Analysis Doesn’t Scale 4mm MIT Micro Turbine Macro Materials Surface Roughness: Structure Dimensions Micro Ductile: Aluminum, Steel, Titanium Brittle: Silicon, Silicon Nitride, Oxide 1:105-106 Insignificant surface flaw size 1:102-104 Significant flaw size in certain processes A.M. Fitzgerald, Ph.D. Thesis, Stanford University, 2000 A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 Material Behavior Ductile σ Brittle Crack tip blunting Good flaw tolerance σ Crack tip blunting, if any Poor flaw tolerance A.M. Fitzgerald, Ph.D. Thesis, Stanford University, 2000 A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 Structural Analysis • • Ductile Flaws can safely be ignored in analysis Strength of materials analysis σ SafetyFactor = σ σy • σ σ ε Failure Probability = f(σ, flaw size, flaw distribution) σ Plastic Elastic A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com σy • Brittle In c-Si, process-derived flaws must be characterized Fracture mechanics and statistics A.M. Fitzgerald, Ph.D. Thesis, Stanford University, 2000 ? Rupture Elastic ε Alissa M. Fitzgerald Stanford E240 3/6/07 Stress Intensity, K σ ij = σij r θ • K f ij (θ ) 2 πr A sharp crack distorts the stress field – K is the amplitude, or stress intensity factor [MPa-m1/2] – Fracture occurs when K > KIc, the fracture toughness Courtesy of Kathy Flores Mode I A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Mode II Mode III Alissa M. Fitzgerald Stanford E240 3/6/07 Fracture Toughness is a Material Property Material KIc Fracture Toughness MPa-m1/2 Glass 0.7-0.8 Silicon 0.6-1.2 Concrete 2-2.3 Aluminum 20-40 Steel 50-110 A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com A.M. Fitzgerald, Ph.D. Thesis, Stanford University, 2000 Alissa M. Fitzgerald Stanford E240 3/6/07 Stress-Flaw Size Relation 3.5 3 • Mode I tensile crack Stress, GPa 2.5 2 σ 1.5 K 1 a K I = 1.12σ πa IC 0.5 0 0 0.2 0.4 0.6 0.8 1 A.M. Fitzgerald, Ph.D. Thesis, Stanford University, 2000 Flaw size, um A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 Surface Effect • Scatter in strength of brittle materials is typically attributed to volume flaws – Silicon is crystalline and high purity, few voids and defects – MEMS have very large surface area/volume ratios • Look to surface flaws for source of scatter A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 Surface Conditions: Plasma Etch (DRIE) Plasma reflection from oxide etch stop Striations left by alternating etch/passivation cycles 30 um Aaron Partridge, photo by JPL A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com R.M.S. Feature size = 0.2 μm A.M. Fitzgerald, Ph.D. Thesis, Stanford University, 2000 Alissa M. Fitzgerald Stanford E240 3/6/07 Surface Conditions: Wet Etch Anisotropic etch leaves intersecting (111) planes with sharp corners R.M.S. Feature size = 24 Å A.M. Fitzgerald, Ph.D. Thesis, Stanford University, 2000 JPL Micro Gyroscope A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 Probability of Failure Probability of Failure 99.99 99.9 Sharp crack 99 95 90 80 70 50 30 20 10 5 1 .1 .01 0.6 Plasma Wet 0.8 1 1.2 1.4 1.6 1.8 1/2 Applied Stress Intensity, MPa-m A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com A.M. Fitzgerald, Ph.D. Thesis, Stanford University, 2000 2 Alissa M. Fitzgerald Stanford E240 3/6/07 Silicon Microstructures Failure Analysis • Device level analysis: – Characterize surface flaws induced by processing • • • • • Will vary according to process methods and tools Lithography variations Center-to-edge of wafer etch variations Wafer-to-wafer etch variations Process stability over time – Model probability of fracture initiation based on test data • Depending on desired confidence level, may require hundreds of samples – Model applied stress • Linear elastic model – Combine all results to establish a failure distribution function • Risk tolerance will drive final design specifications A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 Conclusions • • • • • • • MEMS for critical applications must be designed for reliability Large body of data on microscale electrical failure phenomena thanks to IC chip industry Microscale mechanical processes not as well understood, particular fretting, wear, cyclic fatigue of thin films Fracture mechanics approach is required to characterize failure probability in c-Si MEMS devices Application, device, and process-specific data must be gathered to make a meaningful reliability analyses More information: http://www.amfitzgerald.com Contact: [email protected] (650) 592-6100 A. M. Fitzgerald & Associates, LLC www.amfitzgerald.com Alissa M. Fitzgerald Stanford E240 3/6/07 ...
View Full Document

This note was uploaded on 11/10/2010 for the course AEROSPACE AE 1202 taught by Professor Dr.adib during the Spring '10 term at Sharif University of Technology.

Ask a homework question - tutors are online