822 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 15, NO.

4, AUGUST 2006

Air-Damped Microresonators With

Enhanced Quality Factor
Lijie Li, Gordon Brown, and Deepak Uttamchandani, Senior Member, IEEE

Abstract—It is known that the dissipative damping force due and the underlying substrate. More recently, a numerical anal-
to the air film trapped between the bottom of surface microma- ysis has been undertaken on surface micromachined resonators
chined resonators and the substrate on which they are fabricated where the three-dimensional (3-D) Stokes equations have been
decreases in magnitude as the separation between the two in-
solved. Results from the analysis [10] show that for laterally os-
creases. The practical outcome of this is that microresonators
located close to a substrate will have higher damping and a lower cillating microresonators, the 1-D model overestimates values
quality factor . In order to further investigate this effect and of by around 74% in comparison to the values of ob-
compare experimental findings with theory, a new test device tained from the more comprehensive 3-D model. In [9] and
that enables modulation of the damping interaction between a [10], laterally oscillating microresonators were fabricated and
surface micromachined resonator and the substrate has been tested to compare the experimentally measured values of with
fabricated. The device consists of a surface micromachined the theoretical results obtained. In both sets of experiments,
polysilicon microresonator, which is self-elevated out of the plane
of the substrate using a bimorph beam. A second, identical mi- however, the microresonators investigated were fabricated to lie
croresonator lying parallel to the plane of the substrate has also parallel to the substrate, with an approximately 2- m-thick air
been fabricated. Both devices have been fabricated using the film between the resonator and the substrate. The fabrication
polysilicon multiuser microelectromechanical systems (MEMS) in both cases involved surface micromachining of polysilicon
processes (polyMUMPs). The resonator-to-substrate separation resonators using the polysilicon multiuser microelectromechan-
of the elevated resonator is varied by changing the temperature ical systems (MEMS) processes (polyMUMPs) [11]. The super-
of the bimorph beam, and the factors for different separations
have been measured. Experimental results show that the elevated strate in both cases was an “infinite” thickness of air above the
microresonators have values which are 65% higher than the microresonators. In order to further investigate the effect of the
in-plane microresonators. These experimental findings show good substrate on damping and compare experimentally measured
agreement with the theoretical model of damping used. [1497] factors with theoretical values, a new test device that enables
Index Terms—Damping analysis, elevated microresonators, modulation of the damping interaction between a surface micro-
microresonators, quality factor analysis. machined resonator and the substrate is described in this paper.
To achieve this, surface micromachined resonators have been
fabricated that are lifted out of the plane of the substrate. El-
I. INTRODUCTION evation of the microresonator is achieved using a purpose-de-
signed bimorph structure at whose end the microresonator is

A IR-DAMPED microresonators have been extensively used

in a variety of applications such as resonant accelerome-
ters [1] and gyroscopes [2]; other resonant sensors for chem-
attached. The self-lifting process avoids the need for external
microassembly tools. Moreover, by varying the temperature of
the bimorph, the separation between the microresonator and the
ical sensing [3], biological sensing [4], pressure sensing [5], substrate can be varied, thereby allowing to be measured for
and temperature sensing [6]. Microresonators have also been different values of air-gap thickness. The for such elevated
used in resonant filters for signal processing [7] and commu- microresonators is compared to the of identical in-plane de-
nications [8]. In all of these applications, the quality factor vices which have been fabricated on the same substrate, and
of the microresonators, defined as the ratio of the strain en- whose separation from the substrate is constant and has a value
ergy stored to the energy dissipated per cycle of oscillation, is of approximately 2 m. At the maximum 65- m value of sepa-
a parameter which designers aim to maximize. The main ex- ration achieved between the lowest point of the microresonator
ternal effect responsible for energy dissipation from the res- and the substrate, it is experimentally shown that the microres-
onator is viscous damping due to the ambient air. This effect onator oscillates with a which is 65% higher than the of
has been investigated in microresonators which were fabricated the in-plane resonator.
parallel to a silicon substrate and separated from the substrate This paper is organized as follows. Section II describes the de-
by a distance of around 2 m [9]. The damping was modelled sign and fabrication process followed to achieve a novel elevated
by applying a one-dimensional (1-D) Stokes damper model for microresonator using the polyMUMPs process. Section III de-
the air lying between the laterally oscillating microresonator scribes the experimental procedures, and presents the values of
measured from the in-plane and elevated microresonators.
Manuscript received January 4, 2005; revised December 8, 2005. Subject Ed- Section IV presents the theoretical model of the resonator vi-
itor G. K. Fedder. brating in a fluid, and uses the model to estimate the impact of
The authors are with the Microsystems Group, Department of Electronic
and Electrical Engineering, University of Strathclyde, Glasgow G1 1XW, U.K.
the resonator-to-substrate separation on the value of for com-
(e-mail: [email protected]). parison with the experiments. Section V presents the discussion
Digital Object Identifier 10.1109/JMEMS.2006.879375 and conclusions.
1057-7157/$20.00 © 2006 IEEE
LI et al.: AIR-DAMPED MICRORESONATORS WITH ENHANCED QUALITY FACTOR 823

TABLE I
DESIGN PARAMETERS

Fig. 1. Schematic of the microresonators.

II. MICRORESONATOR DESIGN AND FABRICATION

In this section, the design and fabrication of both in-plane and
elevated microresonators is described. The elevated microres-
onators are self-assembled using a metal/polysilicon bimorph
beam.

A. Fabrication of Microresonators
The microresonators were fabricated using polyMUMPs
process offered by the commercial foundry MEMSCAP Inc.,
Durham, NC, [11]. The fabrication process is described next,
noting that all dimensions given are the nominal dimensions
stated for the process. The polyMUMPs process starts with a
100-mm n-type silicon substrate on which a 600-nm low-stress
silicon nitride is deposited. After this, a 500-nm polysilicon
film (poly0) is deposited and patterned. A 2- m phosphosili- Before detailing the beam design, a brief overview of tech-
cate glass layer is next deposited and patterned, and niques that have been applied in lifting MEMS devices out
constitutes the first sacrificial layer. After this step, the first of the plane of the substrate are described. Several assembly
polysilicon structural layer (poly1) is deposited to a thickness methods can be used to lift MEMS devices out of plane. The
of 2 m. The (poly1) layer was patterned to form the comb simplest involves manual assembly using micromanipulators.
fingers, mass, and suspension springs of the microresonator. The technique is time-consuming and there is the potential
The second oxide layer follows, and this can be of devices being damaged by the micromanipulators or not
patterned and etched using two mask layers (poly1_poly2_via) being accessible to the micromanipulator probes. Another
and (anchor2), with different outcomes. The second polysilicon method is to use dedicated microactuators for 3-D lifting and
structural layer (poly2) with nominal thickness of 1.5 m is assembly. Approaches using electrically active thermal [12] or
then deposited and patterned. A 0.5- m gold layer (metal) electrostatic actuators [13], [14] have been reported. The use of
is the final layer deposited in the polyMUMPs process. The surface tension forces generated from a range of materials is an
(poly2) and (metal) layers combined to produce a bimorph alternative technique for 3-D self-assembly. Meltable pads of
lifting beam which has stress-induced curvature. The microres- thick photoresist have been used to self-assemble micromirrors
onator is fabricated at the free end of this curved beam, which by utilizing surface tension forces [15]. Molten Pb/Sn solder
is how it becomes elevated. More details on the design of this spheres [16] and thermally shrunk polyimide joints [17] are
lifting beam are given in Section III. In the devices described other materials that have also been used for 3-D assembly.
here, the (poly1_poly2_via) mask level was used to achieve Magnetic forces [18] including plastic deformation magnetic
mechanical and electrical connection between the (poly1) and assembly [19] have also been used to assemble microstructures,
the metal/polysilicon bimorph beam (poly2 metal) in the and this requires the incorporation of a magnetic material on
elevated microresonator device. Meanwhile, the (anchor1) the sections of the structures requiring assembly. Direct elec-
layer was used to fix the in-plane microresonator parallel to trostatic forces have also been applied to lift microstructures
the silicon surface. The configuration of both the in-plane and [20] and electrical potential differences of between 35–40 V
elevated microresonators are shown in Fig. 1, while the device have been shown to be sufficient to produce lifting forces for
dimensions are presented in Table I. With the exception of the hinged micromirrors.
(anchor1) layer, both microresonators are identical in design. In surface micromachining processes, when a bimorph can-
The devices were fabricated adjacent to each other in order to tilever is formed from two different films having different values
minimize the effect of process variations which might occur of internal stress, the cantilever will bend to a concave or convex
across the wafer surface. shape upon release depending on the values and type of stress
(i.e., whether tensile or compressive). Polysilicon and metal are
B. Stress-Induced Beam Design materials often used in surface micromachining technology and
This section considers the design of the bimorph beam whose can be used for forming such self-bending beams because of
stress-induced curvature is responsible for lifting the microres- the different internal stresses present in these materials. Bent
onator out of the plane of the silicon substrate. bimorphs of this kind have several interesting applications. In
824 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 15, NO. 4, AUGUST 2006

Fig. 2. SEM of microresonators. (a) SEM image showing two microresonators fabricated adjacent to each other. (b) Close-up SEM image of elevated microres-
onator. (c) Close-up SEM image of in-plane microresonator.

micro-optics, they have been used in the realization of curled layer thickness is , and Young’s modulus of the gold is .
cantilever optical switches [21]. They have also been used for It can be shown that the radius of curvature is [24], as shown
elevating flat micromirrors parallel to the plane of the silicon in (1) at the bottom of the page.
substrate in order to obtain a higher angular deflection from the For the polyMUMPs process, the value for the stress in the
micromirrors [22]. In RF applications, they have been used to deposited polysilicon is 11 MPa (compressive) [11], which is
form elevated inductors [23]. This section describes how this obtained from the polyMUMPs run 60 data; likewise the stress
principle of a curved bimorph cantilever induced by differen- of the deposited gold layer, obtained from the polyMUMPs run
tial stress has been used to realize a microresonator with a large 60 data, is 24 MPa (tensile) [11]. , the Young’s modulus of
gap between resonator plate and substrate, i.e., an elevated mi- polysilicon, is taken to be 169 GPa [25], while , the Young’s
croresonator. modulus of gold, is taken to be 70 GPa [26]. For the poly-
For a bimorph beam composed of two layers—polysilicon at MUMPs process, the thickness of polysilicon equals to 1.5
the bottom and a metal (gold) layer on the top—the differential m, and the thickness of gold equals to 0.5 m. For a bi-
stress generated in the two layers during deposition produces morph cantilever, the displacement of the tip is related to the
curvature of the beam after it is released by etching. The radius radius of curvature as
of curvature is denoted as . Assume that initial stress in the
polysilicon layer is , the layer thickness is , Young’s mod-
ulus of polysilicon is , the initial stress in the gold is , the (2)

(1)
LI et al.: AIR-DAMPED MICRORESONATORS WITH ENHANCED QUALITY FACTOR 825

Fig. 3. SEM images of resonators operating in SEM vacuum. (a) Elevated resonator. (b) In-plane resonator. The motion blur is clearly seen in both images.

Fig. 4. Frequency-response measurement for in-plane resonator and elevated resonator in the SEM vacuum. The Q factor of in-plane resonator is measured to be
Q
6400, and factor of elevated resonator is measured to be 11 600.

where is the beam length. In our design, a 900- m length and the SEM chamber during the testing was Pa. SEM
300- m width bimorph beam was sufficient to obtain a large images were recorded showing the motion blur of the laterally
separation between the elevated microresonator tip and the sub- oscillating microresonators. A computer program was written
strate. Fig. 2 shows the scanning electron microscope (SEM) to analyze the images and translate the blur into displacement
picture of both the in-plane and elevated microresonators that amplitude. Examples of the images are shown in Fig. 3, while
were fabricated. The angle of the elevated resonator was mea- the normalized frequency response of both the in-plane and
sured to be 22 1.5 , while the height of the lowest point of elevated microresonators are shown in Fig. 4. Table II shows
the elevated resonator above the substrate was measured to be the key results obtained.
65 m. At very low pressures, the effects of internal friction and sup-
port losses become the dominant mechanisms of energy dissipa-
III. TESTING
tion. The critical pressure where these effects become dom-
A. Testing in Vacuum inant is given in [27] as
The in-plane and elevated microresonators were initially
tested by being introduced into the vacuum chamber of an
SEM. The SEM was also equipped with electrical feed-through (3)
connectors using which the devices could be electrically pow-
ered. The voltages applied to the fixed and movable comb of where the is universal molar gas constant (8.31 J/K/mol), is
the microresonators were selected to minimize the nonlinearity the ambient temperature taken to be 300 K, is the molar mass
associated with the quadratic relationship between potential of air ( kg/mol). For the elevated resonator in the SEM
differences and electrostatic forces. For the in-plane resonator, vacuum chamber, is measured to be 11 600, and is 24.78
the fixed comb was set to 0 V and the movable comb set to 3.2 kHz, and so the critical pressure is calculated to be 5.5 Pa. The
V with a 0.8-V amplitude sinusoid. For the elevated resonator, value of critical pressure for the in-plane resonator is calculated
the same driving waveform was applied. The air pressure in to be 9.95 Pa. The SEM chamber pressure of Pa is
826 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 15, NO. 4, AUGUST 2006

TABLE II
MEASUREMENTS TAKEN IN VACUUM

TABLE III
MEASUREMENTS FROM TESTING IN AIR

far less than both the calculated critical pressures, therefore veri- with the high-resolution images to display and measure the mo-
fying that the resonators are operating in a regime where the sup- tion of MEMS devices in three dimensions [28].
port and internal friction losses are dominant. The factor of The lateral oscillation of three in-plane and three elevated
the elevated microresonator is measured to be around 1.8 times microresonators were characterized using the Veeco analyzer.
higher than the of the in-plane microresonator. The internal The data obtained was fitted to a normalized frequency-response
friction losses of both devices would be reasonably expected to function
be of the same order of magnitude, as they have the same dimen-
sions and are fabricated in the same material. The difference in
the values of measured in vacuum indicates that the support (4)
losses for the in-plane microresonator are higher, although the
exact source of energy loss due to the supports is not understood
for our structures. with representing displacement amplitude measured at fre-
quency . The measured data and fitted frequency responses are
shown in Fig. 5. The fit minimizes the sum of the mean square
B. Testing in Air
difference between the frequency response of (4) and the mea-
factor testing in air was undertaken using a Veeco NT1100 sured data set, and yields values of and the resonant frequency
DMEMS analyzer. The NT1100 is essentially an optical sur- . The measurements obtained are shown in Table III.
face profiler that is also capable of carrying out measurements The measurements obtained show that in air the average reso-
on moving microstructures. The profiler uses white light in- nant frequency of the in-plane resonator is 4.8% lower than the
terferometry to obtain full-field profiles of surface height with resonant frequency of the elevated resonator. When both res-
nanometric resolution. It also uses stroboscopic illumination to onators are operated in the vacuum of the SEM, it is once again
capture interferograms and high-resolution images that are syn- observed that the resonant frequency of the in-plane resonator is
chronized to the driving voltage applied to the device under test. lower than the resonant frequency of the elevated resonator, this
The profiler software combines the demodulated interferograms time by around 3.3%. The frequency difference between the two
LI et al.: AIR-DAMPED MICRORESONATORS WITH ENHANCED QUALITY FACTOR 827

TABLE IV
MEASUREMENTS OBTAINED BY CHANGING THE AIR GAP

where and are real constants [29] and

is the resonant frequency. Taking into account the notation for
, the damping force can be written as the sum of two terms with
real coefficients, and can be expressed as
where is the first derivative of the velocity. The part of the
damping force proportional to is the dissipative part and is
responsible for energy loss. The other part of the damping force
, written as is called the inertial part—this part
has the effect of increasing the mass of the vibrating system
and lowering its resonant frequency. Just as the dissipative part
of the damping force ceases to be a factor in vacuum, so the
inertial part of the damping force will also cease to be a factor
in vacuum. The resonant frequency in vacuum will, therefore,
be expected to be higher than the resonant frequency in air, and
this is observed from the measurements. A theoretical analysis
of this inertial damping force is beyond the scope of this work.

C. Adjusting the Air Gap of the Elevated Resonator

Adjustment of the air-gap thickness between the elevated
resonator and the substrate can be achieved by changing the
temperature of the structure. Differential thermal expansion be-
tween the gold layer and polysilicon layer which constitute the
lifting bimorph will change its overall height—increasing the
temperature lowers the height of the bimorph because the upper
gold layer of the bimorph beam has greater expansion than the
lower polysilicon layer. A Peltier element was used to increase
the temperature of the die containing the microresonators. Mea-
surements of the lateral oscillations of the in-plane structure
Fig. 5. Frequency-response measurement for in-plane resonator and elevated were made to determine the effect of the increased temperature
resonator using Veeco DMEMS unit. (a) Chip 1. From the second-order fre-
Q
quency-response fitting, the factor and f of in-plane resonator is measured on the resonant frequency and . The die was heated to 90
Q f
to be 28 and 23.65 kHz, respectively, and factor and of elevated resonator is and the resonance frequency and of the in-plane resonator
measured to be 42 and 24.5 kHz, respectively. (b) Chip 2. From the second-order
frequency-response fitting, the Q factor and f of in-plane resonator is mea-
were measured. The values obtained are 23.68 and 25 kHz,
Q
sured to be 24.7 and 23.14 kHz, respectively, and factor and f of elevated respectively. Compared to the values at room temperature, a
resonator is measured to be 42.4 and 24.58 kHz, respectively. (c) Chip 3. From change of 0.2% occurs in the resonant frequency and a change
Q f
the second-order frequency-response fitting, the factor and of in-plane res-
Q
onator is measured to be 25.5 and 23.73 kHz, respectively, and factor and f of 2% occurs in when the in-plane device is heated to 90 .
Measurements were then undertaken on the elevated microres-
of elevated resonator is measured to be 45 and 24.8 kHz, respectively.
onator during which it was noted that raising the temperature
of the die to 90 lowered the height of the elevated bimorph
microresonators is attributed to small structural, dimensional, or from 65 m to 6 m, with intermediate temperatures producing
material property differences arising during microfabrication. intermediate values of resonator-to-substrate separation. Data
The damping force on a vibrating structure with a velocity was captured using the Veeco dynamic analyzer, and the
expressed in complex notation as can be written as factors were determined from the measured data using curve
828 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 15, NO. 4, AUGUST 2006

Fig. 6. Experiment and theory of Q factor for different minimum air-gap thicknesses.

Fig. 7. Factors affecting damping and Q.

fitting, as described in Section III-B. The measurements ob- of a microresonator, thereby determining the overall of the
tained are given in Table IV. The values of measured for the microresonator. In the following, the expressions given for the
different resonator-to-substrate separations are plotted in Fig. 6 values of have been taken from [30].
along with the results of the theoretical analysis of Section IV. 1) Damping Between the Resonant Structure and Substrate
: The air damping between the resonant structure and the
D. Characterization of the Bimorph Beam substrate leads to a factor which can be calculated from the
The bimorph supporting the elevated microresonator was also following:
analyzed for its oscillation characteristics. Using the ANSYS
modeling tool, the fundamental resonant frequency of the bi-
(5)
morph beam was determined to be 1.46 kHz while higher order
modes of the bimorph had frequencies of 9.5, 12.7, 27.3, and
39.5 kHz. The elevated microresonators were driven at their res- where is the viscosity of the fluid (air), which is
onant frequencies, and at the same time the Veeco dynamic pro- N.s/m , and is the effective plate area
filer was focused on the bimorph beam to detect any motion of which includes the area of the plates, fingers, and beams of
the beam in the , , and directions. No motion was detected the microresonator. This effective plate area is given by the
confirming that energy from the laterally oscillating microres- expression where is the area
onator was not being coupled to the bimorph beam. of the plate, which includes the two connecting bars and the
comb fingers, is the area of the truss, is the area of
the beam, is the resonant frequency, and is the distance
IV. MODELING OF THE FACTOR between the lowest point of the inclined resonant structure
Modeling of the factor of microresonators and the effect of and the substrate. The term is the effective mass, where
air damping has been reported in [9], [10], [27], and [30]. Fig. 7 and is the density of
shows the four main factors which contribute to the damping the polysilicon, which is 2300 kg/m [31]. The thickness of
LI et al.: AIR-DAMPED MICRORESONATORS WITH ENHANCED QUALITY FACTOR 829

TABLE V
PARAMETERS FOR THE CALCULATIONS OF Q

the structure is taken as 2 m. For the microresonators used in where is the characteristic dimension of the moving structure,
this paper, is calculated to be kg. and is taken to be half the length of the beam shown in Fig. 1,
2) Damping From the Top : The air damping between the and has a value of 240 m.
resonant structure and the air above the device leads to a factor The overall quality factor obtained from (5), (6), (7), and (8)
which can be calculated from the following: can be written as

(6)

(9)
where is the effective distance, , and is the
density of air, which is 1.168 kg/m . For the devices used in this
paper is calculated to be 15 m. is also called the penetration In (9), is a scaling factor which depends on the design of the
depth and is the distance beyond which the fluid in which the plates. Here, we take a value of 2 for [30]. All the parameters
resonator is oscillating may be assumed to be stationary. used for the calculations are listed in Table V.
3) Damping From the Side Wall : The damping from the Using (9) and the values in Table V, the factor for the
side walls of the comb-driven resonant structure leads to a in-plane resonator is calculated to be 21, while the factor
factor which can be calculated from the following: for the elevated resonator for a resonator-to-substrate air-gap
thickness of 15 m which is equal to the effective distance
is calculated to be 44. Changing the minimum air-gap thickness
(7) (i.e., gap measured from the lowest part of the inclined struc-
ture) from 2 m to 50 m produces a variation in the theoretical
value of as shown in Fig. 6.
where is the area of the side walls that parallel with the
direction of motion. For the devices used in this paper, is V. DISCUSSION AND CONCLUSION
m . The term is the distance between the side
In-Plane Microresonator: In-plane microresonators have
walls of the comb fingers, and is 3 m.
been fabricated and evaluated using the polyMUMPs process.
4) Damping From the Comb Finger End [27]: The
Measurements of the value of at room temperature have
damping from the ends of the comb fingers leads to a factor
been obtained for these resonators, and an average value of
which can be calculated from the following:
26.07 has been experimentally obtained. Using the models for
outlined in this paper, the theoretical value of has been
obtained as 21. This is in good agreement with the experi-
(8) mental values given the uncertainty in the value of the air-gap
830 JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 15, NO. 4, AUGUST 2006

thickness between the in-plane resonator and the substrate—at [9] Y. H. Cho, A. P. Pisano, and R. T. Howe, “Viscous damping model for
the lowest air-gap thickness, the dependence of on thickness laterally oscillating microstructures,” J. Microelectromech. Syst., vol.
3, no. 2, pp. 81–87, Jun. 1994.
is very high, as Fig. 6 shows. For the theory, the value of the [10] W. Ye, X. Wang, W. Hemmert, D. Freeman, and J. White, “Air
air-gap has been taken as 2 m, which is the nominal oxide damping in lateral oscillating microresonators: a numerical and exper-
thickness of the polyMUMPs process. We have not been able imental study,” J. Microelectromech. Syst., vol. 12, no. 5, pp. 557–566,
to independently verify the value of this oxide thickness. Oct. 2003.
[11] J. Carter, A. Cowen, B. Hardy, R. Mahadevan, M. Stonefield, and S.
Elevated Microresonator: A novel self-elevated microres- Wilcenski, PolyMUMPs Design Handbook. Durham, NC: MEM-
onator has also been fabricated also using the polyMUMPs SCAP Inc. [Online]. Available: hhtp://www.memscap.com
process. A key property of the device is that by varying the [12] J. R. Reid, V. M. Bright, and J. T. Butler, “Automated assembly of
temperature of the bimorph, the separation of the microres- flip-up micromirrors,” Sens. Actuators A, Phys., vol. 66, pp. 292–298,
Apr. 1998.
onator from the underlying substrate can be varied. At room [13] T. Akiyama, D. Collard, and H. Fujita, “Scratch drive actuator with
temperature when the device is at its maximum distance from mechanical links for self-assembly of three-dimensional MEMS,” J.
the substrate (65 m) the average value of experimentally Microelectromech. Syst., vol. 6, no. 1, pp. 10–17, Mar. 1997.
measured is 43.13. The theoretical value obtained for is 44 [14] N. C. Tien, O. Solgaard, M. H. Kiang, M. Daneman, K. Y. Lau, and
R. S. Muller, “Surface micromachined mirrors for laser beam posi-
and this compares very well with the experimental values. The tioning,” Sens. Actuators A, Phys., vol. 52, pp. 76–80, Mar.–Apr. 1996.
value of is expected to remain constant at air-gap values [15] R. R. A. Syms, C. Gormley, and S. Blackstone, “Improved yield, accu-
exceeding 15 m, as this is the penetration depth of the racy and complexity in surface tension self-assembled MOEMS,” Sens.
Actuators A, Phys, vol. 88, pp. 273–283, 2001.
resonator, i.e., the distance from the resonator where the air is
[16] K. F. Harsh, V. M. Bright, and Y. C. Lee, “Solder self-assembly for
no-longer moving. Measurements of have also been made three-dimensional microelectromechanical systems,” Sens. Actuators
at resonator-to-substrate separations less than the value of , A, Phys, vol. A77, pp. 237–244, 1999.
and the measured values agree favorably with the theoretical [17] T. Ebefors, E. Kalvesten, and G. Stemme, “New small radius joints
based on thermal shrinkage of polyimide in V-grooves for robust self-
values, as seen in Fig. 6. assembly 3D microstructures,” J. Micromech. Microeng., vol. 8, pp.
In conclusion, a micromechanical resonator in which the 188–194, 1998.
thickness of the air-gap between the resonator and substrate [18] Y. W. Yi and C. Liu, “Assembly of micro-optical devices using mag-
can be changed has been realized. The device allows the netic actuation,” Sens. Actuators A, Phys, vol. 78, pp. 205–211, 1999.
[19] J. Zou, J. Chen, C. Liu, and J. E. Schutt-Aine, “Plastic deformation
damping interaction between the resonator and substrate to magnetic assembly for elevated microstructures: technoogy and appli-
be varied. The damping effect is quantified experimentally by cations,” J. Microelectromech. Syst., vol. 10, no. 2, pp. 302–309, Jun.
measurement of of the resonator. A good match between the 2001.
experimental and the obtained from the theoretical model [20] R. W. Johnstone and M. Parameswaran, “Self-assembly of surface-mi-
cromachined structures using electrostatic attraction,” in Proc. Int. Soc.
of the damping has been achieved. Opt. Eng. (SPIE), MOEMS and Miniaturized Systems II, , 2001, vol.
4561, pp. 66–76.
[21] R. T. Chen, H. Nguyen, and M. C. Wu, “A high-speed low voltage
ACKNOWLEDGMENT
stress-induced micromachined 2 2 2 optical switch,” IEEE Photon.
The authors would like to thank the U.K. Engineering and Technol. Lett., vol. 11, no. 11, pp. 1396–1398, Nov. 1999.
[22] V. A. Aksyuk, F. Pardo, and D. J. Bishop, “Stress-induced curvature
Physical Sciences Research Council (EPSRC) for support. They
engineering in surface-micromachined devices,” in Proc. Int. Soc.
would also like to thank the referees for their helpful comments. Opt. Eng. (SPIE), Design, Test, and Microfabrication of MEMS and
MOEMS, , 1999, vol. 3680, pp. 984–993.
[23] V. M. Lubecke, B. Barber, E. Chan, D. Lopez, M. E. Gross, and P.
REFERENCES Gammell, “Self-assembling MEMS variable and fixed RF inductors,”
[1] J. W. Weigold, K. Najafi, and S. W. T. Pang, “Design and fabrication of IEEE Trans. Microw. Theory Tech., vol. 49, no. 11, pp. 2093–2098,
submicrometer, single crystal Si accelerometer,” J. Microelectromech. Nov. 2001.
Syst., vol. 10, no. 4, pp. 518–524, Dec. 2001. [24] L. Li, J. Zawadzka, and D. Uttamchandani, “Integrated self-assembling
[2] T. Tsuchiya, Y. Kageyama, H. Funabashi, and J. Sakata, “Polysil- and holding technique applied to a 3-D MEMS variable optical attenu-
icon vibrating gyroscope vacuum-encapsulated in an on-chip micro ator,” J. Microelectromech. Syst., vol. 13, no. 1, pp. 83–90, Feb. 2004.
chamber,” Sens. Actuators A, Phys., vol. 90, pp. 49–55, May 2001. [25] N. D. Mankame and G. K. Ananthasuresh, “Comprehensive thermal
[3] L. Fadel, F. Lochon, I. Dufour, and O. Francais, “Chemical sensing: modelling and characterization of an electro-thermal-compliant mi-
millimeter size resonant microcantilever performance,” J. Micromech. croactuator,” J. Micromech. Microeng., vol. 11, pp. 452–462, 2001.
Microeng., vol. 14, pp. 23–30, Sep. 2004. [26] M. A. Haque and M. T. A. Saif, “Deformation mechanisms in
[4] B. Ilic, D. Czaplewski, H. G. Craighead, P. Neuzil, C. Campagnolo, and free-standing nanoscale thin films: a quantitative in situ transmission
C. Batt, “Mechanical resonant immunospecific biological detector,” electron microscopy study,” Proc. Nat. Acad. Sci. USA, vol. 101, pp.
Appl. Phys. Lett., vol. 77, pp. 450–452, Jul. 2000. 6335–6340, 2004.
[5] M. Habibi, E. Lueder, T. Kallfass, and D. Horst, “A surface microma- [27] M. H. Bao, Micro Mechanical Transducers. Amsterdam, The Nether-
chined capacitive absolute pressure sensor array on a glass substrate,” lands: Dutch Amsterdam Publisher, 2000, pp. 131–132.
Sens. Actuators A, Phys., vol. 46–47, pp. 125–128, 1995. [28] Veeco Instruments 2005 [Online]. Available: http://www.veeco.com/
[6] Y. J. Rao, D. Walsh, D. Uttamchandani, and B. Culshaw, “Tempera- default.asp
ture-dependence of resonant-frequency in all fiber optically addressed [29] F. R. Blom, S. Bouwstra, M. Elwenspoek, and J. H. J. Fluitman, “De-
silicon microresonator sensors,” Electron. Lett., vol. 27, pp. 934–936, pendence of the quality factor of micromachined silicon beam res-
May 1991. onator on pressure and geometry,” J. Vac. Sci. Technol. B, Microelec-
[7] L. Lin, R. T. Howe, and A. P. Pisano, “Microelectromechanical filters tron., vol. 10, pp. 19–26, Jan/Feb 1992.
for signal processing,” J. Microelectromech. Syst., vol. 7, no. 3, pp. [30] X. Zhang and W. C. Tang, “Viscous air damping in laterally driven
286–294, Sep. 1998. micro resonators,” Sens. Mater., vol. 7, pp. 415–430, 1995.
[8] C. T. C. Nguyen, “Frequency-selective MEMS for miniaturized low- [31] R. Hickey, D. Sameoto, T. Hubbard, and M. Kujath, “Time and fre-
power communication devices,” IEEE Trans. Microw. Theory Tech., quency response of two-arm micromachined thermal actuators,” J. Mi-
vol. 47, no. 8, pp. 1486–1503, Aug. 1999. cromech. Microeng., vol. 13, pp. 40–46, 2003.
LI et al.: AIR-DAMPED MICRORESONATORS WITH ENHANCED QUALITY FACTOR 831

Lijie Li received the B.Sc. degree in microelec- Deepak Uttamchandani (M’96–SM’03) received
tronics science from the Lanzhou University, the Ph.D. degree from University College London,
Lanzhou, China, in 1995, and the Ph.D. degree London, U.K., in 1985.
in the design, modeling, and characterization of He is currently the Professor of Microsystems En-
microactuators and optical MEMS devices from gineering, and Director of the Center for Microsys-
Center for Microsystems and Photonics, Strathclyde tems and Photonics which is based at the Department
University, Glasgow, U.K., in 2004. of Electronic and Electrical Engineering, University
He continued optical and general MEMS research of Strathclyde, Glasgow, Scotland, U.K. His research
at Strathclyde University. In 2006, he joined the In- has concentrated on applying photonic and microsys-
stitute of System Level Integration, Livingston, U.K., tems technology to optoelectronic sensors and sys-
as a Senior MEMS Design Engineer. He is author and tems, such as optically excited micromechanical res-
coauthor of over 20 scientific papers and one U.K. patent in the field of MEMS. onator sensors, and in mechanical characterization of MEMS materials. More
recently, his work has expanded to include photonic and RF applications of
MEMS technology.

Gordon Brown received the B.Sc. degree from the

University of Glasgow, Glasgow, U.K. and the M.Sc.
in laser engineering from St. Andrews University,
Fife, U.K., in 1995
He is currently a Research Assistant within the
Center for Microsystems Photonics, University of
Strathclyde, Glasgow, Scotland, U.K. He has carried
out research in microsystems-based optical sensors,
and on the use of ultra-high-speed imaging for the
characterization of microsystems. He now focuses
on the metrology and characterisation of MEMS de-
vices and in the application of excimer laser micromachining to microsystems.