Engineering MEMS Resonators with

Low Thermoelastic Damping
Temperature-dependent
internal friction in silicon
nanoelectromechanical systems
AdviserŘProf. Li, Wang Long
SpeakersŘHwang, Chih-Jay Q26971037
Chen, Po-Wei Q26974093
Li, Wen Rong Q26971061
Nguyen, Huu Nghia Q28977013
Introduction Introduction
To identify the thermal modes that contribute most to damping,
and illustrates how this information may be used to design devices
with higher quality factors.
We calculate damping in typical micromechanical resonator
structures using Comsol Multiphysics
We compare the results with experimental data reported in
literature for these devices..
Micromechanical resonators are used in a wide variety of
applications, including inertial sensing, chemical and biological
sensing, acoustic sensing, and microwave transceivers.
The resonator¶s Quality factor(Q), which describes the mechanical
energy damping.
In all applications, it is important to have design control over this
parameter, and in most cases, it is invaluable to minimize the
damping.
Zener developed general expressions for thermoelastic damping in
vibrating structures, with the specific case study of a beam in its
fundamental flexural mode.
Zener calculated the thermoelastic Q of an isotropic homogenous
resonator to be:
Zener made in assuming only thermal gradients in one direction
along the beam were significant does not capture the most
important thermal mode, even for a simple beam.
In addition, past efforts to estimate Q without explicitly
calculating the weighting functions have greatly overestimated the
damping behavior of real systems.
We describe a method for using full numerical solutions to
evaluate Q using Zener¶s approach.
The imaginary component represents the mechanical vibration
frequency, while the real part provides the rate of decay for an
unforced vibration due to the thermal coupling.
The quality factor of the resonator is defined as
The eigenvalues, Ȧ
i
, are complex.
In some cases, the experimental data appears to have achieved the
thermoelastic limit. For these devices, it is clear that structural
modifications that can engineer a higher thermoelastic limit are
warranted.
In devices where the measured Q value is less than half the
thermoelastic limit, investigation into and minimization of other
damping mechanisms is warranted.
This remarkable correlation between simulation results and
experiments suggests that the flexural beam Q is limited by
thermoelastic damping.
Experiment
Resonator Parameters
Resonator Cases
Results and conclusion Results and conclusion
We find that the TED Q is two orders higher than the measured Q.
This suggests that thermoelastic damping, for the fundamental
longitudinal mode, is not a significant contributor to the overall
energy loss in this resonator.
A paddle resonator operating in its torsional resonance was
simulated. The simulated resonant frequency was about 20% lower
than the measured torsional frequency.
The simulated result is consistent with the physical understanding
that torsional deformations produce little or no volumetric expansion
and should therefore have negligible thermoelastic damping.
By studying the damping contributions of individual thermal
modes, their mode shapes, and their frequencies, it is possible to
engineer MEMS resonators for higher Q.
In addition, by reviewing the fundamental coupled thermodynamic
energy expressions, we achieve a greater insight into the energy
loss mechanismitself.
The simulations were conducted in Comsol Multiphysics. This
software can parameterize the material parameters and geometry,
so that detailed optimization studies are enabled.
uoveining
Equation
Generalized Hooke·s law for
linear isotropic elastic solids
‡ ı
x
= Ȝ ( e
x
+ e
y
+ e
z
)+2ȝe
x
‡ ı
x
= Ȝ ( e
x
+ e
y
+ e
z
)+2ȝe
y
‡ ı
x
= Ȝ ( e
x
+ e
y
+ e
z
)+2ȝe
z
‡ Ȣ
xy
= 2ȝe
xy
‡ Ȣ
yz
= 2ȝe
yz
‡ Ȣ
zx
= 2ȝe
zx
‡ Ȝ : Lame`¶ constant
‡ ȝ : shear modulus
Thermal strain
‡ İ
T
= Į(¨T)
‡ Į : cofficient of thermal expansion
Force balance
3-D equation of motion
Fourier¶s law
Entropy
Combine differential fourier¶s law and entropy
Reference Reference
Amy Duwel, Tob N. Candler, Thomas W. Kenny, and Mathew
Varghese, ³ Engineering MEMS Resonators With Low
Thermoelastic Damping´, Journal of Miroelectromechanical
Systems, Vol. 15, NO. 6, Dec 2006.
Introduction Introduction
The understanding and control of composition, nanostructure, and
interface properties are important for the development of
nanostructured materials.
High-frequency mechanical resonators presenting high quality
factors are of interest for the development of sensitive force
detecting devices.
Quality factor of resonant micromechanical devices decreases
steadily with device dimension.
Defect motion is governed by an activation energy that
will induce Debye relaxation peaks in the temperature
dependence of internal friction.
Debye relaxation is the dielectric relaxation response of
an ideal, non-interacting population of dipoles to an
alternating external electric field. It is usually expressed
in the complex permittivity of a medium as a function of
the field's frequency Ȧ:
Fabrication and electrostatic operation of
nanomechanical beams as thin as 30 nm and frequencies
as high as 380 MHz.
Dynamical modeling and characterization of paddle
oscillators operating in the 1±10 MHz range.
Reporting the temperature dependent behavior of these
paddle oscillators and observing Debye internal friction
peaks in the T=160ņ190K range.
Experimental
Approach
Using electron beam lithography on silicon-on-insulator (SOI)
wafers consisting of a 400-nm-thick oxide buried underneath 200
nm of single crystal <100> silicon.
pumped down to the 10
-5
Torr range.
The cold finger allows temperature access and control over the T=4±300K
range.
The quality factor (Q) is closely approximated from the width of
the resonance peaks using the relation
f
0
is the center of the resonance response, and ¨f
FWHM
is its full
width half maximum.
Nanofabricated paddle oscillator
d=5.5 mm, w=2 mm, L=2.5 mm, b=175 nm, a=200 nm, h=400 nm
Identifying two modes of oscillation attributed to the
flexural and torsional motion of the supporting beams.
These modes are sufficiently decoupled to allow their
independent excitation by the application of the
appropriate actuation frequency.
Temperature dependence of the two resonant
frequencies of a metallized device
The frequency steadily
increases as the temperature
decreases to T=80K, at which
point an inflection of the slope
is observed. Overall increases
in resonant frequency of 6.5%,
and 1.5%are observed at the
lowest temperature for the
flexural and torsional modes.
Temperature dependence of the internal friction for the two
modes of motion of a metallized and nonmetallized device
Within the precision of our measurement, all four sets of data
show a peak structure centered at T=160±180 K.
The existence of this peak in both metallized and nometallized
devices suggests that the metal overlayer is not responsible for this
loss.
The reduction of the sloped dissipation background in the
nonmetallized device suggests that metal film monotonically
contributes to the total internal friction in that temperature range.
This contribution could possibly peak at much higher temperatures,
as expected from bulk polycrystalline metals.
A similar peak has been observed at T=135K in larger kilohertz
range microcantilevers, and has been attributed to surface or near-
surface related phenomena such as damage or presence of oxide.
The peaks observed in our megahertz-range devices could
potentially be related to similar phenomena, as a shift from
T=120± 140K at 2±10 kHz to T=160± 180K at 5±7 MHz would be
consistent with a Debye relaxation behavior dictated by an
activation energy of E
a
=0.25± 0.5 eV.
Results and conclusion Results and conclusion
The characterization of both modes of motion of these single-stage
paddles consistently suggested a material 50% softer than
expected from bulk silicon.
A temperature dependent frequency shift has been observed.
Low-temperature studies of internal friction at 5±7 MHz have also
revealed a double peak centered in the T=160± 180K range that
would be consistent with the activation energies expected from
near-surface phenomena previously reported in larger devices.
A thorough understanding of the various extrinsic, intrinsic, and
fundamental processes leading to internal losses at such scales. It
will enhance the quality of such RF structures.
Previous description allows the development of high-quality
resonators for technological applications, and provide access to
fundamental studies of surface effects and mesoscopic internal
friction.
•imulate
Select multi-physics modes
Import object (.sat)
Import constant
Subdomain setting ± Solid, Stress-strain
boundary setting ± Solid, Stress-strain
Subdomain setting ± Heat Transfer by Conduction
Boundary setting ± Heat Transfer by Conduction
Solver Parameters setting
Solver
Calculate Q (Quality factor)
Quality factor (Q)
Final Conclusion Final Conclusion
Designing the micromechanical resonators we need to calculate
quality factor Q that describes the mechanical energy damping and
play an impotant role in the structure.
By using Zener formular (reference 1) and experimence the
dependence between temperature, resonant frequency and inertial
loss (reference 2) we can easily get Q value and compare the value
between calculation and experimence.
The result for the case Torsional 3D we solved with the value
Q = 2e8 simulating by Comsol Multiphysics that is fixed with
measured value at simulated frequency 4.4 MHz.
So we can solve the fully coupled thermoelastic dynamcs to obtain
exact expressions for Q in an arbitrary resonator with Comsol
Multiphysic.
With this reason, designing a micromechanical resonator is more
simple by simulating and calculating for exact results.
Final Conclusion Final Conclusion
Thanks for Thanks for
your attention your attention

Introduction

 To identify the thermal modes that contribute most to damping, and illustrates how this information may be used to design devices with higher quality factors.  We calculate damping in typical micromechanical resonator structures using Comsol Multiphysics  We compare the results with experimental data reported in literature for these devices..

 Micromechanical resonators are used in a wide variety of applications, including inertial sensing, chemical and biological sensing, acoustic sensing, and microwave transceivers.  The resonator¶s Quality factor(Q), which describes the mechanical energy damping.  In all applications, it is important to have design control over this parameter, and in most cases, it is invaluable to minimize the damping.

 Zener developed general expressions for thermoelastic damping in vibrating structures, with the specific case study of a beam in its fundamental flexural mode.  Zener calculated the thermoelastic Q of an isotropic homogenous resonator to be:

even for a simple beam. past efforts to estimate Q without explicitly calculating the weighting functions have greatly overestimated the damping behavior of real systems.  We describe a method for using full numerical solutions to evaluate Q using Zener¶s approach. .  In addition. Zener made in assuming only thermal gradients in one direction along the beam were significant does not capture the most important thermal mode.

 The imaginary component represents the mechanical vibration frequency. i. are complex. .  The quality factor of the resonator is defined as  The eigenvalues. while the real part provides the rate of decay for an unforced vibration due to the thermal coupling.

it is clear that structural modifications that can engineer a higher thermoelastic limit are warranted. For these devices. the experimental data appears to have achieved the thermoelastic limit. .  In devices where the measured Q value is less than half the thermoelastic limit.  This remarkable correlation between simulation results and experiments suggests that the flexural beam Q is limited by thermoelastic damping. In some cases. investigation into and minimization of other damping mechanisms is warranted.

Experiment .

 Resonator Parameters  Resonator Cases .

Results and conclusion .

 We find that the TED Q is two orders higher than the measured Q. is not a significant contributor to the overall energy loss in this resonator. . The simulated resonant frequency was about 20% lower than the measured torsional frequency. for the fundamental longitudinal mode.  A paddle resonator operating in its torsional resonance was simulated. This suggests that thermoelastic damping.  The simulated result is consistent with the physical understanding that torsional deformations produce little or no volumetric expansion and should therefore have negligible thermoelastic damping.

so that detailed optimization studies are enabled.  In addition. we achieve a greater insight into the energy loss mechanism itself. By studying the damping contributions of individual thermal modes.  The simulations were conducted in Comsol Multiphysics. their mode shapes. and their frequencies. This software can parameterize the material parameters and geometry. by reviewing the fundamental coupled thermodynamic energy expressions. . it is possible to engineer MEMS resonators for higher Q.

‘˜‡”‹‰ “—ƒ–‹‘ .

Generalized Hooke·s law for linear isotropic elastic solids ‡ ‡ ‡ ‡ ‡ ‡ ‡ ‡ = ( ex + ey + ez )+2 ex ( ex + ey + ez )+2 ey x = ( ex + ey + ez )+2 ez x = xy = 2 exy yz = 2 eyz zx = 2 ezx : Lame`¶ constant : shear modulus x .

Thermal strain ‡ ‡ T= (¨T) : cofficient of thermal expansion .

.

Force balance .

3-D equation of motion .

.

Fourier¶s law Entropy Combine differential fourier¶s law and entropy .

.

Reference .

NO. Kenny. Dec 2006. Candler. 6. ³ Engineering MEMS Resonators With Low Thermoelastic Damping´. . Journal of Miroelectromechanical Systems. Amy Duwel. Thomas W. Tob N. and Mathew Varghese. 15. Vol.

Introduction .

and interface properties are important for the development of nanostructured materials. nanostructure. The understanding and control of composition. .  Quality factor of resonant micromechanical devices decreases steadily with device dimension.  High-frequency mechanical resonators presenting high quality factors are of interest for the development of sensitive force detecting devices.

non-interacting population of dipoles to an alternating external electric field. Defect motion is governed by an activation energy that will induce Debye relaxation peaks in the temperature dependence of internal friction. It is usually expressed in the complex permittivity of a medium as a function of the field's frequency : .  Debye relaxation is the dielectric relaxation response of an ideal.

.  Reporting the temperature dependent behavior of these paddle oscillators and observing Debye internal friction peaks in the T=160 190K range. Fabrication and electrostatic operation of nanomechanical beams as thin as 30 nm and frequencies as high as 380 MHz.  Dynamical modeling and characterization of paddle oscillators operating in the 1±10 MHz range.

Experimental Approach .

and ¨f FWHM is its full width half maximum. Using electron beam lithography on silicon-on-insulator (SOI) wafers consisting of a 400-nm-thick oxide buried underneath 200 nm of single crystal <100> silicon. .  pumped down to the 10-5 Torr range.  The cold finger allows temperature access and control over the T=4±300K range.  The quality factor (Q) is closely approximated from the width of the resonance peaks using the relation f0 is the center of the resonance response.

h=400 nm . L=2. w=2 mm. a=200 nm.5 mm.5 mm. b=175 nm.Nanofabricated paddle oscillator d=5.

.  These modes are sufficiently decoupled to allow their independent excitation by the application of the appropriate actuation frequency. Identifying two modes of oscillation attributed to the flexural and torsional motion of the supporting beams.

. at which point an inflection of the slope is observed.5%. and 1.5% are observed at the lowest temperature for the flexural and torsional modes.Temperature dependence of the two resonant frequencies of a metallized device The frequency steadily increases as the temperature decreases to T=80K. Overall increases in resonant frequency of 6.

Temperature dependence of the internal friction for the two modes of motion of a metallized and nonmetallized device .

 The reduction of the sloped dissipation background in the nonmetallized device suggests that metal film monotonically contributes to the total internal friction in that temperature range. . all four sets of data show a peak structure centered at T=160±180 K.  The existence of this peak in both metallized and nometallized devices suggests that the metal overlayer is not responsible for this loss.  This contribution could possibly peak at much higher temperatures. Within the precision of our measurement. as expected from bulk polycrystalline metals.

and has been attributed to surface or nearsurface related phenomena such as damage or presence of oxide. A similar peak has been observed at T=135K in larger kilohertz range microcantilevers. as a shift from T=120± 140K at 2±10 kHz to T=160± 180K at 5±7 MHz would be consistent with a Debye relaxation behavior dictated by an activation energy of Ea=0.  The peaks observed in our megahertz-range devices could potentially be related to similar phenomena. .25± 0.5 eV.

Results and conclusion .

.  Low-temperature studies of internal friction at 5±7 MHz have also revealed a double peak centered in the T=160± 180K range that would be consistent with the activation energies expected from near-surface phenomena previously reported in larger devices.  A temperature dependent frequency shift has been observed. The characterization of both modes of motion of these single-stage paddles consistently suggested a material 50% softer than expected from bulk silicon.

. It will enhance the quality of such RF structures. intrinsic.  Previous description allows the development of high-quality resonators for technological applications. A thorough understanding of the various extrinsic. and fundamental processes leading to internal losses at such scales. and provide access to fundamental studies of surface effects and mesoscopic internal friction.

•‹—Žƒ–‡ .

Select multi-physics modes .

Import object (.sat) .

.

Import constant .

Subdomain setting ± Solid. Stress-strain .

boundary setting ± Solid. Stress-strain .

.

.

Subdomain setting ± Heat Transfer by Conduction .

Boundary setting ± Heat Transfer by Conduction .

Solver Parameters setting .

.

Solver .

Calculate Q (Quality factor) .

.

Quality factor (Q) .

The result for the case Torsional 3D we solved with the value Q = 2e8 simulating by Comsol Multiphysics that is fixed with measured value at simulated frequency 4.Final Conclusion Designing the micromechanical resonators we need to calculate quality factor Q that describes the mechanical energy damping and play an impotant role in the structure. . By using Zener formular (reference 1) and experimence the dependence between temperature. resonant frequency and inertial loss (reference 2) we can easily get Q value and compare the value between calculation and experimence.4 MHz.

. designing a micromechanical resonator is more simple by simulating and calculating for exact results.Final Conclusion So we can solve the fully coupled thermoelastic dynamcs to obtain exact expressions for Q in an arbitrary resonator with Comsol Multiphysic. With this reason.

Thanks for your attention .

Master your semester with Scribd & The New York Times

Special offer for students: Only $4.99/month.

Master your semester with Scribd & The New York Times

Cancel anytime.