surface forces in nanomechanical systems: living on the edge j provine stanford university...

Surface Forces in Nanomechanical Systems: Living on the Edge J Provine Stanford University 2012-01-11 Fermilab Colloquium

Outline Scaling in the micro/nanometer range Introduction to several surface effects Nanoelectromechanical Switches Application As a nanoprobe Device design for probing surface forces Conclusion

A few quick words on scaling We live in the m-cm world (10 0 to 10 -2 m) MicroElectroMechanical Systems (MEMS) and CMOS electronics circa 1990 1m (10 -6 m) Current CMOS, thin film optical coatings, NEMS 10nm (10 -8 m) Carbon Nanotubes, atomic layer deposition coatings, self assembled monolayers 1nm (10 -9 m) Lattice constant of Si 5.4A (10 -10 m) Fermilab

The Dominance of Surface Effects Volume 4/3 r 3 Surface Area 6 r 2 Surface Area:Volume 1/r As the size of an object shrinks, the surface affects become more dominant because the object is becoming all surface.

Some surface effects in nanodevices Photonics effects Adhesion (geckos) Nourredine smith wear/friction Casimir Force

1. Make any material a good optical material 2. Get at the unique optical properties of specific materials Various unique optical material properties can be explored and exploited now because of great materials understanding. Polariton Modes Kerr Effect Birefringence Photoelectric transduction Surface Effects in Photonics New ways to get excellent optical performance from a wide range of materials. Photonic Crystal and Subwavelength Grating design for allows a very wide range of materials to provide desired performance.

PCs come in many flavors Excellent test bed for some deep physics experiments (QED, surface physics, etc.) Telecom and Photonic circuits. Slow light. Lin, et al, 2003 Kuchinsky, et al, 2002

Broadband Reflector Applications M.C.Y. Huang, Y. Zhou, and C. Chang-Hasnain, Feb. 2007 I. Jung, S. Kim, O. Solgaard, Trans. 2007 High temperature, high power handling. CMOS compatible and integrable processing.

11 Monolithic Si Photonic Crystal Slab Dielectric stack (DBR) Slab photonic crystal Monolithic photonic crystal

Materials for PC 20nm (2%) Increase Air gap thickness change Refractive index change 0.2 (5%) Increase Hole radius 10nm (3%) Increase Polysilicon thickness change 20nm (5%) Increase Extensive testing has been done for particular materials (Si, poly-Si, SiN, SiO 2 ) But the key is ANY DIELECTRIC can be used to design PCs. Strong wavelength dependent guided or reflected modes can be created in materials to suit specific applications.

PC Fiber Tip Sensor Applications Biological, chemical, and mechanical sensors (such as accelerometers) at the end of an optical fiber can be useful for control and security applications The small size (125 m diameter) enables them to penetrate tissue or veins for medical applications PCs at the tip of fibers can be used both for free-space and inline applications as a reflector, polarizer and filter

Fiber Tip Assembly Pt weld of PCDirect weld of PC Utilize direct weld of PC with ion beam as opposed to Pt weld to study impact of weld technique.

Index Sensing Experiment 3dB couplerPower meter Optical spectrum analyzer Fiber PC Water/Solvent Broadband source Index Sensing ExperimentExperimental data

Refractive index calculated from volume concentration Refractive Index Sensing Responsivity = R.I./ = 0.04768 [nm -1 ] Sensitivity 4.8 x10 -5 [pm -1 ] Using an optical system (tunable laser, OSA) with picometer resolution Comparable to FBG refractive index sensors [W. Liang, A. Yariv et al, APL 2005] Isopropanol concentration increase in increments of 30ml in DI Water of 150ml IEEE Nanophotonics 2009.

Temperature Sensing Experiment Experimental data

Temperature measurement taken while cooling from 80C to room temperature Temperature Sensing Responsivity = temp/ = 16.0858 [C/nm] Sensitivity 0.016 [C/pm] Using an optical system (tunable laser, OSA) with picometer resolution Almost an order better sensitivity than a FBG temperature sensor [A. D. Kersey et al., Fiber Grating Sensors Invited Paper, JLT 1997] LEOS annual meeting 2009

19 [www.cnconveyorbelt.com][www.tommcmahon.net] Harsh environments High voltage, high power machinery High temperature [blog.mlive.com] [Onur Kilic] Motion/Vibration/Explosion detection Acoustic sensing Gyro/Acceleration [www.af.mil] [www.blueparrotevents.coml] Bio/chemical detection Biological/chemical agents Fluid, Gas sensing [www.gallagher.com] Structural Health monitoring Combustion chambers, Turbines Aircraft, wind turbines, bridges, dams, oil wells, pipelines Smart structures: Integrated fiber-optic sensors (aging, vibrations) [newswhitehouse.com] [www.reuk.co.uk] Impact & Applications

1m1m Si beam SiC coating Accessing a particular optical property in a novel material: SiC Spitzer, et al, Phys Rev., 1959. The optical properties of SiC have also been studied for a long time. Recently the interest has expanded because of the extremely strong mid-IR Phonon Polariton resonance.

Device Fabrication SiO 2 SiC Bulk Si LPCVD SiC @ 800 C LPCVD SiO 2 for hard mask Transfer photolithographic mask through SiO 2 and SiC by RIE RIE of SiC is HBr/HCl Release membrane by XeF 2 etch Remove hard mask with HF dip 80 sidewall

Polariton Gap Hole Array Patterned Film Unpatterned Film Theroretical simulation with FD3D Finite Difference Time Domain code. a d t Extraordinary Transmission t = 4 m a = 10.4 m d = 5.6 m

Extraordinary Transmission Polariton Gap Hole Array Patterned Film Unpatterned Film t = 4 m a = 10.4 m d = 5.6 m Polariton Gap t = 1.5 m Polycrystalline SiC Experimental Data from FTIR Unpatterned Film

d=5.6 m Unpatterned Film Polariton Gap Hole Array Patterned Film Unpatterned Film t = 4 m a = 10.4 m d = 5.6 m Polariton Gap t = 1.5 ma = 10 m Polycrystalline SiC Experimental Data from FTIR Extraordinary Transmission

d=5.6 m d=3.9 m Polariton Gap Hole Array Patterned Film Unpatterned Film t = 4 m a = 10.4 m d = 5.6 m Polariton Gap t = 1.5 ma = 10 m Polycrystalline SiC Experimental Data from FTIR Extraordinary Transmission Unpatterned Film Provine, et al, OMEMS 2007

Reflection Spectra t = 1.5 m a = 10 m Polycrystalline SiC Experimental Data from FTIR d=3.9 m d=5.6 m d=4.8 m d=3.1 m

Ongoing Experiments: A True Meta-Material Selective metal surface coatings. (Catrysse and Fan, Physical Review B, 2007)

Adhesion at the Nanoscale Work between Autumn Lab (Lewis & Clark) & Kenny Lab (Stanford)

Casimir Force in Metals Valid at 0 K and vacuum. Uncharged metals (equipotential) will still attract. Purely a quantum & geometrical effect. Hotly debated and studied because of the relation to the cosmical constant. At the nanoscale starts to have appreciable forces.

Casimir effect in Pt nanobeams Nanobeam constructed from a single sheet of evaporated Pt (equipotential). Slices are made with ion beam and then released from unlying Si with XeF 2. Crystal orientation makes this a challenging study.

Application: a downside of scaling E. J. Nowak, IBM J. Res & Dev. 2002 As modern CMOS electronics scales to smaller and smaller devices, the power consumption rising rapidly. Because of the ubiquitous computing ongoing (and being proposed) the amount of energy going to servers and even personal computing is becoming appreciable.

A Solution: Back to the Future Babbage Analytical Engine c 1877 Mechanical computing can be an answer to this issue because it can deliver zero off-state power consumption. Additional benefits: Radiation hard operation Lower thermal dependence Is this a CMOS killer? NO But it can have many applications and certainly help with energy consumption. (see for instance Chen et al FPGA 2010.)

Implementing a NEM Swith

Examples of NEM Switches: Metallic Structures Vertically actuated W Colorado, Boulder Laterally actuated Ru Sandia National Labs Vertically actuated Ni KAIST

Examples of NEM Switches Conducting Ceramics Vertically actuated TiN KAIST Laterally actuated TiN Stanford

Examples of NEM Switches: Semiconducting Structures Vertically actuated poly-Si KAIST Vertically actuated W coated SiGe California, Berkeley

Arbitrary NEM Logic Design Methodologies Only 6T relays required for all 3 generations Our lateral 6T elemental logic block New elemental block allows new design methodologies G

Gate 1 Drain Source 1 Source 2 Gate 2 Beam =Mold Layer (eg, Polysilicon) =Insulating Layer (eg, Hafnium Oxide) =Conductive Layer (eg, TiN or Pt) Isolation The Logic Element: 6T Relay

Y-Device Process Flow AA BB (a) Deposit 1um polysilicon on 1.5um oxide. (b) Pattern polysilicon (mask 1). Oxide Substrate

Y-Device Process Flow AA BB (c) Deposit 20nm HfO 2 via ALD. (d) Blanket etch of HfO 2.

Y-Device Process Flow AA BB (e) Deposit 20nm Pt or TiN via ALD. (f) Etch Pt or TiN and pattern pads (mask 2).

Y-Device Process Flow AA BB (g) Pt or TiN wet etch for sidewall isolation (mask 3). (h) Release in 49% HF followed by CPD.

Fabricated Device

Y-Device Switching Properties [S. Lee et al., Transducers 2011] Current Flow (Source to Drain) No Beam Current

Mechanical Delay Measurement 1.2s

Easy, right? Not Always

ALD Platinum Coated Relay Large pull-out variation! Adhesion force variations: asperity deformation Single device, multiple cycles

Other issues Desired improvement in Total Lifetime Uniformity between devices (same chip) Uniformity between devices (different wafers) Concerns Fabrication tolerance Adhesion forces Contact Resistance

Controlling the contact mechanism and apparent contact area: Existing designs Point-surface contact mechanism with limited asperity-asperity contact Flexible Contact Surface Flexible surface- surface contact Before pull-in After pull- in Point contact Overdrive voltage surface-surface contact NEM Relays with improved contact properties

Mechanically robust designs with large overdrive voltage:

NEM Relays with Small Footprint Using the coating as the main structural material: 200nm process and 50nm coating: Electrode length: 5m Beam length : 5m Source-gate gap: 100nm TiN coating:50nm

NEM Relays: 6T Relays 6T relays are sensitive to fabrication tolerances: 500nm process and 20nm coating: Beam length: 21um Gate length: 19um Coating thickness:20nm Beam-gate gap: 560nm Source-drain gap:460nm Source-drain tol.:10nm (2%) FEM simulations: First contact : 23V Second contact:> 37V V = 37V Result: Extensive overdrive often necessary.

NEM Relays: 6T Relays Insensitive to Fabrication Tolerance Relays with flexible source-drain:

NEM Relays: New Designs (6T) Relays with flexible source-drain: FEM Simulations: Displacements (V = 5V)

NEM Relays: New Designs (6T) Relays with flexible source-drain: FEM Simulations: Contact pressure (V = 5V) Result: Overdrive minimized for relay.

NEM Relays: New Designs (6T) Relays with flexible source-drain: FEM Simulations:

Switch vs. Nanoprobe While the switching application is important and interesting, surface affects mean that simultaneously: o We need to understand surface forces more accurately to optimize our switches o The switches can operate as excellent nanoprobes to determine what is happening. o Different materials o Different ambient conditions o Different designs to isolate particular material properties

NEM Relays: Reliability Material and surface characterization: Youngs modulus Structural and air damping

NEM Relays: Reliability Material and surface characterization:

NEM Relays: Reliability Material and surface characterization: Adhesion Stiction to substrate Minimum gap Maximum length Stiction to side walls

Take Home Messages In general, nanofabrication has grown up to the point we can make almost anything. Lots of materials A wide ranges of sizes (mm to A) While this opens up a wide range of new applications, it just as quickly allows (necessitates?) new science to be explored.

Acknowledgements NEMS Logic Team (in particular Kamran Shavazipur) Stanford Roger Howe Group Philip Wong Group Olav Solgaard Group UC Berkeley Roya Maboudian Group Tsu-Jae King Liu Group Center for Interfacial Engineering of MEMS (CIEMS) DARPA and NSF for funding

Thank you. Questions?

surface forces in nanomechanical systems: living on the edge j provine stanford university...

Documents