New Frontiers in Casimir Force Control Sept.

27-29, 2009, Santa Fe, New Mexico

Surface Forces in MEMS – Adhesion and Friction Experiments
Maarten P. de Boer
MEMS Technologies Dept. Sandia National Laboratories Albuquerque, NM, 87185

Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company,
 for the United States Department of Energy under contract DE-AC04-94AL85000.

MEMS – surface micromachining implementation
A series of structural and sacrificial layers are deposited Ground plane layer (Poly 0) 4 structural levels (Poly 1 - Poly 4) Chemical Mechanical Planarization (CMP) 1 µm design rule Create freestanding thin film structures by “release” process
Sniegowski & de Boer, Annu. Rev. Mater. Sci. (2000)
Poly 4 Poly 3 Poly 2 Poly 1 Poly 0

Design A A’


A’ Crosssection drawing

FIB crosssection 5 µm

slide 2

With polysilicon MEMS we can reliably accomplish electromechanical and optical functions
- thousands of devices simultaneously - no assembly required - hundreds of device concepts explored
Integrated inertial sensor

High performance comb drive with mechanical amplifier Polychromator : programmable diffraction grating

slide 3

friction and wear become the most important failure mechanisms of contacting MEMS slide 4 .Allowing contact between MEMS surfaces significantly broadens the design space Complex Mechanical Logic Gears Pop-up Mirrors Pin-in-maze guides linear racks hinges but … static friction can dominate the forces required dynamic friction can dominate energy loss adhesion.

Adhesion (“stiction”) is a big problem in micromachining Initially free beam. but still in water Drying leads to “stiction” s slide 5 .

We can use cantilevers to quantify the adhesion. Your t s Capillary adhesion can be avoided by critical point drying or by applying monolayer coatings (drying from water) (de Boer and Michalske. Journal of Applied Physics. Γ h The image cannot be displayed. 1999) slide 6 .

Microcantilever process and test flow a Deposit landing pad polysilicon on insulating substrate L t Deposit sacrificial oxide and structural polysilicon h Release in HF acid. CPD and coat with molecular monolayer s d Apply voltage Vpad slide 7 .

Oxidize the Poly 0 Surface to change surface roughness Nanotexturing of the lower layer or polysilicon (P0) was accomplished via thermal oxidation in dry O2 at 900° C for increasing times.4 nm rms 1 µm slide 8 1 µm 300 Å oxidation.3 nm rms . 10. 5. 4.6 nm rms 1 µm 100 Å oxidation.6 4. t (min) 0 20 136 400 tox (Å) -100 300 600 rms (nm) 2.3 1 µm No oxidation.4 5.6 10. 2.6 nm rms 600 Å oxidation.

F H .C .C . If the Si slide 9 .C .H H .H H .H H .F F .2.F F .H Si Si Si Si O O O O Native SiO2 The image cannot be displayed.C .2-tetrahydrodecyltris(dimethylamino)silane ) vapor deposition 8 carbon chain van der Waals forces not strong enough to self assemble (tangled) contact angle ~ 110° FOTAS 8-carbon fluorinated chain (disordered. tangled) F .C . Restart your computer. and then open the file again.H H .MEMS monolayer coupling agent FOTAS (tridecafluoro-1. or the image may have been corrupted.1.C .H H .F F .H H . Your computer may not have enough memory to open the image.C .F F F F .C .C .C .F F .H H .C .C .C .C .

3 nm slide 10 .6 nm 100 µm rms roughness = 10.Interferograms show qualitative relationship between surface roughness and crack length Vpad = 50 V s d 100 µm rms roughness = 2.4 nm 100 µm rms roughness = 5.6 nm 100 µm rms roughness = 4.

Adhesion measurement with applied voltage Finite element analysis (ABAQUS) and user subroutines were used to find beam profiles with surface adhesion. slide 11 (Knapp & de Boer. electrostatic loading and initial stress gradient. The only free parameter in the models is the adhesion Γ. JMEMS. A least squares fit between the model and experiment was used to determine the value at each voltage. 2002) .

Contour map of gap separation between the two surfaces All separations 0 nm Small separations 0 nm y (µm) y (µm) 45 nm 5 nm x (µm) x (µm) slide 12 .The surfaces separation is everywhere less than 100 nm.

  surface heights entered into force displacement routine Numerical Force-Displacement Routine Import AFM height data Separate surfaces by initial displacement Calculate separation for each pixel 4.  512 x 512 matrix with 3.AFM topography data is analyzed using a numerical force-displacement routine AFM Images 1.  7.  2.  Calculate force for each pixel Find total force (sum) Move surfaces towards each other Repeat steps 3-6 to create attractive loaddisplacement curve 1 µm 1 µm Anandarajah and Chen 1995 slide 13 .  6.  5.

Calculate the total force-displacement curve using the AFM analysis and Hertzian mechanics Attractive forcedisplacement curve based on AFM analysis Repulsive forcedisplacement curve based on Hertzian mechanics DMT Adhesion Model Calculate adhesion energy by evaluating the area under the total forcedisplacement curve from the equilibrium displacement to infinity. slide 14 .

Predicted values of adhesion with AFM data We placed the surfaces together in the following combinations for each roughness: •  Poly 0 and Poly 0 •  Poly 0 and Poly 2 The average surface separation Dave is calculated for each AFM pair according to DelRio.. Nature Materials (2005) slide 15 . de Boer et al.

DelRio. Nature Materials (2005) slide 16 . Roughest Surface Adhesion contribution mainly from contacting asperity (converging to Fuller-Tabor/Maugis model for single asperity).Histogram of adhesion contributions vs. pixel separation Smoothest Surface Adhesion contribution from both contacting asperities and noncontacting areas (combination of two extreme adhesion models). de Boer et al..

Roughness on top and bottom surfaces is correlated! Top of bottom surface Bottom of top surface slide 17 .

Summary .DRY adhesion in MEMS Microcantilevers are used to measure adhesion in MEMS Adhesion is in the µJ/m2 range For low surface roughness. adhesion dominated by retarded van der Waals forces (Casimir forces) For higher surface roughnesses. adhesion dominated by normal van der Waals forces Surface topography correlations between upper and lower surfaces play an important role slide 18 .

a stepper motor with 50 nm steps •  40 nanometer step size •  moves ± 100 µm •  high force actuator •  requires traction (friction) to move Plate Length. Lp Friction clamps k Δ~40 nm Actuation Electrode “standoff” to prevent shorting large tangential force range slide 19 .Nanotractor for on-chip actuation .

vol.Nanotractor implementation Trailing clamp Leading clamp electrodes Actuation plate frictional stop Actuation Plate Cross-section(schematic) 200 um clamp clamp foot 10 µm Displacement gauge Suspension spring (0. DL. p. Maboudian. 2004.1. Corwin. JM Journal of Microelectromechanical Systems. de Boer. Feb.13. Luck. Redmond. R. AD. JA. Walraven. WR.63-74 slide 20 . MP.4 N/m) normal force electrode High-performance surface-micromachined inchworm actuator. Ashurst. no.

Driving the Nanotractor trailing clamp P (a) Δ (b) actuation plate Lp A leading clamp (c) Δ (d) slide 21 (a) Clamp RHS (b) Pull down driver beam Operates up to 5 mm/s (c) Clamp LHS (d) Relax RHS & driver beam .

AD & de Boer. Corwin.4 N/m slide 22 .Friction.5•10-9 kg 0.damped oscillator to measure dynamic friction dynamic friction test at small tensile load (FOTAS monolayer): Fel k Effect of adhesion on dynamic and static friction in surface micromachining. MP Applied Physics Letters (2004) M 2.

There is dynamic friction at zero applied load Measured and modeled fit for zero applied load FOTAS monolayer Dynamic friction over a range of applied loads Effect of adhesion on dynamic and static friction in surface micromachining. slide 23 Corwin. AD & de Boer. MP Applied Physics Letters (2004) .

Static friction testing with the nanotractor a) k Xo (c) Δ (d) (a) FN (b) slip 100 nm time b) slide 24 c) .

J. P. Corwin & M. Microelectromechanical Systems (2009) slide 25 . D.Rich static friction behavior is observed by varying the hold time sliding bifurcation static friction aging A. de Boer.

decreases with increasing hold force …!!! slide 26 . the logarithmic rate of aging.β.

The normal force rampdown rate also affects the static friction value static friction dependence on ramp-down rate other measurements th increasing from 2 to 512 s slide 27 .

“Release time” measurement Ft xo Ff (a) k FN Normal Force Fh Fr Camera trigger Block slides 0 (b) 0 Time th tr slide 28 .

“Release time” is far longer than inertial response time and shows the same qualitative dependencies as static friction (a) (b) (c) slide 29 .

de Boer. PRB (submitted) . Corwin & M. D.All the release time data collapse onto a single curve slide 30 A. P.

b3 equates with the logarithmic rate of “re-aging” after the interface de-ages.The release time equation can be used to directly predict the static friction dependence A single parameter “b3”. has been introduced. slide 31 .

17 (b3=-0. 17 (b3=0) slide 32 . Fh increasing Eq.The release time equation also predicts the suppression of β with increasing hold force.0036) Eq.

underlies the static friction behavior. release time quantitatively predicts static friction aging behavior including aging suppression slide 33 .Summary – Friction effect in MEMS The nanotractor is a friction-based actuator that produces useful work at the µscale The clamps form a controlled interface so that friction measurements can be made and modeled Van der Waals attraction is responsible for dyanmic and static friction in the absence of applied force Static friction aging effects have been observed “Release time”. Introducing a re-aging parameter. much greater than inertial response time.

slide 34 .

Backup slides slide 35 .

de Boer et al..Taking correlation into account makes model/experiment agreement nearly perfect Capillary forces can dominate vdW forces! Model and measurement accounting for surface correlations slide 36 DelRio. Applied Physics Letters (2007) .

slide 37 A significant part of the area is too far apart to contribute to the adhesion.2 nm Israelachvili 1992 The forces across non-contacting portions of the surfaces.Two extreme models for adhesion Smooth Surface Rough Surface Dave Parallel Plate Model Dave Single Asperity Model Dave Anandarajah and Chen 1995 R2 R1 0. only the van der Waals forces near the single point of contact contribute. whose area is far greater than the contacting area at the one asperity. . will dominate the adhesion.

5 nm rms) Rough surface contact mechanics considerations … asperity radius of curvature R ~ 20 to 500 nm (typically ~50 nm) rms roughness 1.5 to 10 nm contact diameter ~10 nm. 8 nm rms) top counterface (bottom of P12.Surface contact is an aggregate of asperities 1 µm bottom counterface (top of P0. pressure ~10 GPa real contact area << 10-3•(apparent contact area) slide 38 .

Sign up to vote on this title
UsefulNot useful